AP5M1 encodes mu-5 (also called MuD or MUDENG), the medium subunit of adaptor protein complex 5 (AP-5), the fifth and most recently identified member of the heterotetrameric adaptor protein family. AP-5 is built from AP5Z1 (zeta), AP5B1 (beta-5), AP5M1 (mu-5) and AP5S1 (sigma-5), and in cells the tetramer is part of a larger assembly with spatacsin (SPG11) and spastizin (SPG15/ZFYVE26), which supply the membrane-docking and scaffolding functions of a coat. Like other AP mu subunits, mu-5 binds the beta subunit through its N-terminal domain and carries a C-terminal mu-homology domain, but the residues that in AP-1 and AP-2 form the binding site for YxxPhi sorting signals are not conserved in mu-5, no sorting motif has been identified for AP-5, and the complex neither associates with clathrin nor responds to brefeldin A. Mu-5 cycles between the cytosol and the cytoplasmic face of late endosomal and lysosomal membranes, and its abundance falls when other subunits of the complex are lost, indicating that it is an obligate member of an assembled complex rather than an independent protein. Loss of AP-5 impairs retrieval of the cation-independent mannose 6-phosphate receptor and of Golgi membrane proteins from endosomes back to the Golgi, and fills endolysosomes with undigested multilamellar storage material; the reconstituted AP-5-SPG11-SPG15 assembly binds phosphatidylinositol 3-phosphate, senses membrane curvature and deforms membranes in vitro, and the assembly is required for the initiation of autolysosome tubulation. Bi-allelic loss-of-function variants in AP5M1 cause a recessive macular dystrophy, and variants in the partner genes AP5Z1, SPG11 and SPG15 cause complicated hereditary spastic paraplegia. Under the name MUDENG the same protein has been reported to modulate apoptotic signalling in cultured tumour cells, but those effects are seen on over- or under-expression and their direction differs between cell types.
Definition: The process by which lysosomes are regenerated from autolysosomes after autophagic degradation, comprising the extrusion of tubules from the autolysosomal membrane, their extension, and scission to release proto-lysosomes that mature into functional lysosomes.
Justification: GO has no term for this process. The QuickGO ontology search returns no hits for the word reformation; the children of GO:0007040 lysosome organization are three regulation terms plus GO:0001845 phagolysosome assembly and GO:0097212 lysosomal membrane organization; and the nearest existing terms, GO:0170064 lysosome fission and GO:0097749 membrane tubulation, whose only compartment-specific children are GO:0097750 for endosome membranes and terms for the plasma membrane, each capture one aspect and neither names the process. The result is that the best-characterised cellular defect of the whole AP-5, SPG11 and SPG15 axis has to be annotated to the generic parent GO:0007040, which is how it currently reaches AP5M1. The gap affects at least SPG11, ZFYVE26, AP5Z1, AP5B1, AP5S1 and AP5M1, plus the PI4K2A and mTOR machinery that acts in the same pathway. This is deliberately the same request that genes/human/ZFYVE26/ZFYVE26-ai-review.yaml already makes from the other side of the same complex, under the same name and with the same proposed parent GO:0007040; the two should be merged into one term request rather than filed twice. The definition given here is the fuller of the two - it names the tubulation, extension and scission steps and the proto-lysosome intermediate - and is compatible with, rather than an alternative to, the ZFYVE26 wording "a lysosome organization process in which autolysosomal membranes tubulate and regenerate functional lysosomes after autophagic degradation". A curator acting on either entry should take this one and close the other.
Parent term: lysosome organization
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005764 lysosome | IBA GO_REF:0000033 | ACCEPT | Summary: Correct and core. AP-5 works on the cytoplasmic face of late endosomal and lysosomal membranes, and the phylogenetic assertion that this is ancestral is well founded. Reason: The PAINT slice interpro/panther/PTHR16082/PTHR16082-paint.tsv resolves this row completely. There is an IBD on PTN002699615 in PTHR16082, seeded by UniProtKB:Q9H0R1, placed at taxon:2759, with no IRD or IKR anywhere in the family. The seed is the target's own IDA from PMID:22022230, where GFP-tagged mu-5 shows substantial overlap with LAMP1; the IBA then adds the claim that the compartment is ancestral rather than human-specific, which the same paper's phylogeny supports by finding AP-5 subunits in all five eukaryotic supergroups. Independent corroboration came later from the hexamer work (PMID:29381698) and from antibody staining in a completely different cell type, human retinal pigment epithelium (PMID:40081374). PTHR16082 contains AP-5 mu subunits and nothing else, so no cross-complex donor can have leaked in. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002699615 · PTHR16082 AP-5 complex subunit mu-1 IBD node at taxon:2759 SUPPORTS TRANSFER The node carries five IBDs, of which this is one. AP5M1 is inside the inheriting clade by definition, being the family's only human member and its seed. The family index lists six reviewed members, all AP-5 mu subunits, all in subfamily PTHR16082:SF2. UniProtKB:Q9H0R1 · AP5M1 (human, the target itself) SUPPORTS TRANSFER The target's own IDA from PMID:22022230 is the descendant evidence the PAINT curator used to place the IBD. This is the expected self-referential configuration, not circularity, and it means experimental grounding for the term exists on the target. Supporting Evidence: PMID:22022230 This colocalisation with LAMP1 suggests that C14orf108 may function in the late endocytic pathway. PMID:29381698 The AP-5/SPG11/SPG15 complex localises to late endosomes and lysosomes |
| GO:0005764 lysosome | IDA PMID:22022230 The fifth adaptor protein complex. | ACCEPT | Summary: Direct observation on mu-5 itself. GFP-tagged mu-5 gives punctate labelling that overlaps LAMP1, a lysosomal and late endosomal marker. Reason: This is mu-5-specific evidence, not a complex-level surrogate. The caveat worth recording is that the punctate pattern was only visible after siRNA depletion of the endogenous protein, because the tagged construct otherwise competes unsuccessfully for membrane docking and appears cytosolic. That is a real limitation of the experiment, but it does not undermine the compartment assignment, which the authors quantified and which has since been reproduced with antibodies against the endogenous protein in human retinal pigment epithelium (PMID:40081374). Supporting Evidence: PMID:22022230 there was considerable overlap with LAMP1, a marker for late endosomes and lysosomes PMID:40081374 a punctate pattern of AP5Z1, AP5M1, and AP5B1 staining and co-localization with markers of late endosomes and the Golgi |
| GO:0005764 lysosome | NAS PMID:40175557 Structural basis for membrane remodeling by the AP5-SPG11-SP... | ACCEPT | Summary: ComplexPortal statement taken from the cryo-EM paper. Redundant with the IDA and the IBA but not wrong. Reason: The cited paper is abstract-only in the cache, and its abstract makes a functional rather than a localisation statement, tying the SPG11-SPG15 partner complex to lysosome formation and autolysosome tubulation. As a non-traceable author statement that adds nothing the IDA does not already establish it is harmless, and the compartment itself is not in doubt. It is kept rather than removed because ComplexPortal NAS rows record the complex-level consensus, which here is correct. Supporting Evidence: PMID:40175557 essential for the initiation of autolysosome tubulation PMID:29381698 The AP-5/SPG11/SPG15 complex localises to late endosomes and lysosomes |
| GO:0005765 lysosomal membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Faithful mapping of the UniProt lysosome membrane location onto GO. AP-5 is a peripheral protein on the cytoplasmic face of this membrane, which is where it acts. Reason: The mapping is mechanical and correct. What matters is the strength of what is mapped. The UniProt line carries ECO:0000305 from PubMed:22022230, a curator inference from the LAMP1 colocalisation plus the membrane-versus-cytosol fractionation, rather than a direct observation of the membrane itself. That is the right strength for this claim and the IEA inherits it accurately. The membrane term is more informative than the bare organelle term for a peripheral coat component that cycles on and off, so it is kept as core. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0157 · UniProt subcellular location Lysosome membrane SUPPORTS TRANSFER Resolves to the lysosome membrane location line of the Q9H0R1 record, annotated peripheral membrane protein, cytoplasmic side, at ECO:0000305 from PubMed:22022230. The SubCell to GO mapping neither adds nor loses meaning here. Supporting Evidence: file:human/AP5M1/AP5M1-uniprot.txt {ECO:0000305|PubMed:22022230}. Lysosome membrane file:human/AP5M1/AP5M1-uniprot.txt {ECO:0000305|PubMed:22022230}. Note=May cycle on and off membranes. PMID:22022230 C14orf108 partitions approximately equally between membranes and cytosol, indicating that it cycles on and off the membrane; but unlike clathrin and AP-1, it is not enriched—or even detectable—in the CCV fraction. |
| GO:0005770 late endosome | IBA GO_REF:0000033 | ACCEPT | Summary: Correct and core. The late endosome is the compartment AP-5 was first localised to and the one its sorting step leaves from. Reason: Same node and same seed as the lysosome IBA, and the same reasoning applies. PTN002699615 in PTHR16082, seeded by the target's own IDA, at taxon:2759, in a family containing only AP-5 mu subunits. The compartment assignment is stronger than for a generic organelle term because the later functional work names the same compartment as the origin of the sorting step, so this is where the protein is active and not merely where it can be detected. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002699615 · PTHR16082 AP-5 complex subunit mu-1 IBD node at taxon:2759 SUPPORTS TRANSFER One of five IBDs on this node. The eukaryote-root placement matches PMID:22022230, which finds AP-5 subunits in all five eukaryotic supergroups with repeated coordinate loss; AP5M1 is inside the retaining clade. UniProtKB:Q9H0R1 · AP5M1 (human, the target itself) SUPPORTS TRANSFER Self-referential and expected: the target's own late endosome IDA from PMID:22022230 is one of the descendant evidences behind the IBD. Supporting Evidence: PMID:29381698 The AP-5/SPG11/SPG15 complex localises to late endosomes and lysosomes PMID:40175557 SPG11-SPG15 can cooperate with the fifth adaptor protein complex (AP5) involved in membrane sorting of late endosomes. |
| GO:0005770 late endosome | IDA PMID:22022230 The fifth adaptor protein complex. | ACCEPT | Summary: Direct observation on mu-5, from the same GFP and LAMP1 experiment that supports the lysosome IDA, plus the brefeldin A control that distinguishes this compartment from the AP-1, AP-3, AP-4 and COPI compartments. Reason: Two independent features of the 2011 data support a late endocytic rather than a Golgi or early endosomal assignment: substantial overlap with LAMP1, and complete insensitivity to brefeldin A in cells where AP-3 redistributed to the cytosol in the same experiment. The compartment has since been confirmed for the endogenous protein in a second system, where AP5M1 co-stained with the late endosomal marker Rab7 in iPSC-derived retinal pigment epithelium (PMID:40081374). Supporting Evidence: PMID:22022230 there was considerable overlap with LAMP1, a marker for late endosomes and lysosomes PMID:22022230 The insensitivity of C14orf108 to BFA suggests that its recruitment onto membranes does not depend on ARF GTPases |
| GO:0005770 late endosome | NAS PMID:22022230 The fifth adaptor protein complex. | ACCEPT | Summary: ComplexPortal statement of the complex's compartment, made from the founding paper. It duplicates the IDA from the same reference. Reason: Correct, and the duplication is not a problem. GOA legitimately carries the same term from a database statement and from the underlying experiment. The statement is about AP-5 as a whole, which is the appropriate level for ComplexPortal, and the tetramer plus SPG11 and SPG15 were later shown to colocalise on this compartment as a unit. Supporting Evidence: PMID:22022230 This colocalisation with LAMP1 suggests that C14orf108 may function in the late endocytic pathway. PMID:23825025 In addition, AP-5, SPG11, and SPG15 colocalize on a late endosomal/lysosomal compartment. |
| GO:0005770 late endosome | NAS PMID:40175557 Structural basis for membrane remodeling by the AP5-SPG11-SP... | ACCEPT | Summary: ComplexPortal statement from the cryo-EM paper, whose abstract explicitly places AP-5's sorting role at late endosomes. Reason: Of the three NAS rows sourced to the structure paper this is the one its abstract states most directly. It adds no independent evidence beyond the 2011 IDA, but it is an accurate record of the current complex-level consensus and nothing about it needs changing. Supporting Evidence: PMID:40175557 SPG11-SPG15 can cooperate with the fifth adaptor protein complex (AP5) involved in membrane sorting of late endosomes. |
| GO:0005829 cytosol | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: True but peripheral. About half of mu-5 is soluble at steady state and the node placement is sound, but the cytosolic pool is the off-membrane reservoir of a coat component rather than the place the complex does its work. Reason: The IBD is on the same well-resolved node as the other four, seeded by the target's own IDA, and there is nothing wrong with the propagation. A coat adaptor that cycles on and off membranes genuinely has a cytosolic phase, and that is conserved across the family. The reservation is about centrality rather than correctness. The functional events, recruitment by the SPG11-SPG15 scaffold, cargo sorting and membrane deformation, all happen on the membrane, and the same fractionation experiment that shows the cytosolic pool shows that it represents unassembled or released complex. Recording it as non-core keeps the row without letting it compete with the membrane locations in the core function synthesis. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: PANTHER:PTN002699615 · PTHR16082 AP-5 complex subunit mu-1 IBD node at taxon:2759 SUPPORTS TRANSFER The node placement is not in question. Membrane cycling is a general property of AP complexes and of this family. The demotion is a judgment about which compartment is the functional one, argued in review.reason, not a defect in the transfer. UniProtKB:Q9H0R1 · AP5M1 (human, the target itself) SUPPORTS TRANSFER The target's own cytosol IDA from PMID:22022230 seeds the IBD. Expected self-reference, and the experiment behind it, high-speed fractionation of HeLa homogenates, is the strongest evidence for the cytosolic pool that exists. Supporting Evidence: PMID:22022230 C14orf108 partitions approximately equally between membranes and cytosol, indicating that it cycles on and off the membrane; but unlike clathrin and AP-1, it is not enriched—or even detectable—in the CCV fraction. PMID:23825025 we show that the four AP-5 subunits can be coimmunoprecipitated with SPG11 and SPG15, both from cytosol and from detergent-extracted membranes, with a stoichiometry of ∼1:1:1:1:1:1. |
| GO:0005829 cytosol | IDA PMID:22022230 The fifth adaptor protein complex. | KEEP AS NON CORE | Summary: Directly observed. Mu-5 partitions roughly equally between the high-speed supernatant and the membrane pellet of HeLa homogenates, and GFP-tagged mu-5 is largely cytosolic when expressed over the endogenous protein. Reason: The observation is solid and was made on the endogenous protein with an antibody validated by siRNA knockdown. It is kept, but as non-core for the same reason as the IBA: this is the soluble reservoir of a peripheral membrane coat component. The 2011 authors themselves treat the cytosolic signal as the uninformative fraction, having to deplete endogenous mu-5 before the membrane-bound pool became visible by microscopy. Supporting Evidence: PMID:22022230 C14orf108 partitions approximately equally between membranes and cytosol, indicating that it cycles on and off the membrane; but unlike clathrin and AP-1, it is not enriched—or even detectable—in the CCV fraction. |
| GO:0005829 cytosol | IEA GO_REF:0000044 | KEEP AS NON CORE | Summary: Faithful mapping of the UniProt cytosol location, which is the one subcellular location on this record carrying experimental ECO:0000269 evidence. Reason: The mapping is mechanically correct and, unusually among the three SubCell IEAs, it maps an experimentally supported line rather than a curator inference. It is still demoted to non-core on the same biological grounds as the IBA and IDA rows for this term: for a coat adaptor the cytosolic pool is where the complex waits, not where it acts. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB-SubCell:SL-0091 · UniProt subcellular location Cytosol SUPPORTS TRANSFER Resolves to the cytoplasm, cytosol line of Q9H0R1, which carries ECO:0000269|PubMed:22022230, experimental rather than inferred. The transfer is accurate; the demotion is a judgment about centrality. Supporting Evidence: file:human/AP5M1/AP5M1-uniprot.txt SUBCELLULAR LOCATION: Cytoplasm, cytosol |
| GO:0007040 lysosome organization | NAS PMID:40175557 Structural basis for membrane remodeling by the AP5-SPG11-SP... | ACCEPT | Summary: Correct and core. Loss of AP-5 changes lysosome composition and morphology profoundly, and the complex is required for the tubulation step that regenerates lysosomes from autolysosomes. Reason: This is the one NAS row from the structure paper that adds something the other annotations do not. Two independent lines support it. Loss of AP-5 fills endolysosomes with multilamellar storage material, in patient fibroblasts and in siRNA-treated HeLa cells, to the point that the authors classify AP-5 deficiency as a lysosomal storage disease (PMID:26085577). And the reconstituted assembly is required to start the lysosomal tubulation that regenerates lysosomes (PMID:40175557). GO has no term for autophagic lysosome reformation, so lysosome organization is the closest available parent and is correctly chosen rather than over-specific; the missing term is proposed separately. Supporting Evidence: PMID:40175557 essential for the initiation of autolysosome tubulation PMID:26085577 AP-5 deficiency represents a new type of LSDs |
| GO:0016020 membrane | IDA PMID:22022230 The fifth adaptor protein complex. | MODIFY | Summary: True but the least informative term on the record. The same paper identifies which membrane, so the annotation should be deepened rather than left at the root. Reason: The supporting experiment is the high-speed fractionation showing that about half of mu-5 is in the membrane pellet. Taken alone that says only membrane, which is why the row exists at this depth. But the identity of the membrane is established in the same figure series by the LAMP1 colocalisation and the brefeldin A insensitivity, and UniProt's curators drew exactly that conclusion, annotating late endosome membrane and lysosome membrane, peripheral membrane protein, cytoplasmic side. Both replacement terms already exist on the gene as SubCell IEAs; deepening this row makes the experimental evidence point at them instead of at the root. It is not removed, because the observation is real and is what the membrane association rests on. Proposed replacements: late endosome membrane lysosomal membrane Supporting Evidence: PMID:22022230 C14orf108 partitions approximately equally between membranes and cytosol, indicating that it cycles on and off the membrane; but unlike clathrin and AP-1, it is not enriched—or even detectable—in the CCV fraction. PMID:22022230 there was considerable overlap with LAMP1, a marker for late endosomes and lysosomes |
| GO:0016192 vesicle-mediated transport | NAS PMID:22022230 The fifth adaptor protein complex. | MODIFY | Summary: True parent, but two levels shallower than what the cited paper establishes, and shallower than the endosomal transport term the same gene already carries. Reason: PMID:22022230 does not leave the process at vesicle-mediated transport. It localises mu-5 to a late endocytic compartment, shows that depleting mu-5 mislocalises the CIMPR and Vps26 and swells multivesicular bodies, and concludes for a role in endosomal trafficking. GO:0016197 endosomal transport is the term for that and is already on the gene by IBA and IMP, so this ComplexPortal row should point there rather than at the grandparent. It is deepened rather than removed: the statement is not wrong, only uninformative, and it correctly records that AP-5 belongs to the vesicular trafficking machinery. Proposed replacements: endosomal transport Supporting Evidence: PMID:22022230 Together with the C14orf108 localisation data, these observations point to a role for the C14orf108-DKFZp761E198 complex in endosomal trafficking. |
| GO:0016197 endosomal transport | IBA GO_REF:0000033 | ACCEPT | Summary: Correct, core, and at the right depth for a family-level assertion. The specific step is known only from human cells and only for a handful of cargo, so pushing the node's term deeper would over-claim. Reason: This is the only biological-process IBD on PTN002699615, seeded by the target's own IMP from PMID:22022230, where mu-5 siRNA, with both a SMARTpool and individual oligos, redistributed the CIMPR and Vps26 into enlarged perinuclear puncta and produced swollen multivesicular bodies with emanating tubules. The breadth is deliberate and right. AP-5 has been lost from Drosophila, C. elegans and S. cerevisiae, its cargo are known only in human cells, and the endosome-to-Golgi step was identified only in 2018. A family node spanning all eukaryotes should assert the process, not the step. The deeper step is added separately as a human-specific NEW row rather than by modifying this one. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002699615 · PTHR16082 AP-5 complex subunit mu-1 IBD node at taxon:2759 SUPPORTS TRANSFER The family's only process IBD. Placing it at the eukaryotic root is consistent with AP-5 subunits occurring in all five supergroups, and the term's breadth is appropriate to a clade in which most members have never been assayed. UniProtKB:Q9H0R1 · AP5M1 (human, the target itself) SUPPORTS TRANSFER The target's own IMP from PMID:22022230 is the descendant evidence behind the IBD. Expected self-reference; it also means this IBA is the rare case where a family's process claim rests on an experiment in the target species. Supporting Evidence: PMID:22022230 Together with the C14orf108 localisation data, these observations point to a role for the C14orf108-DKFZp761E198 complex in endosomal trafficking. PMID:29381698 it has been lost from several model organisms, including Drosophila melanogaster, Caenorhabditis elegans, and Saccharomyces cerevisiae PMID:22022230 AP-5 subunits can be found in all five eukaryotic supergroups, but they have been co-ordinately lost in many organisms. |
| GO:0016197 endosomal transport | IMP PMID:22022230 The fifth adaptor protein complex. | ACCEPT | Summary: The strongest annotation on the record: a loss-of-function experiment on mu-5 itself, with both a pooled and individual siRNAs, giving a reproducible endosomal sorting phenotype. Reason: Graded carefully, because the question of whether the experimental rows test mu-5 or only the complex decides how much this record is worth. This one tests mu-5. Depleting C14orf108 redistributed the CIMPR and the retromer subunit Vps26 into fewer, larger, brighter perinuclear puncta, and by electron microscopy produced swollen multivesicular bodies with emanating tubules. The phenotype was reproduced with individual oligos as well as the pool, excluding an off-target effect, and it is phenocopied by knockdown of beta-5, the complex partner. It is kept at endosomal transport rather than deepened, because the direction of the disrupted step was not established until PMID:29381698; that deeper term is added as a separate NEW row. Supporting Evidence: PMID:22022230 When C14orf108 is depleted, both the CIMPR and Vps26 localise to much larger puncta in the perinuclear region of the cell PMID:22022230 Knocking down C14orf108 causes the cells to accumulate swollen multivesicular bodies (MVBs), which have tubules emanating from them PMID:23825025 all six knockdowns cause the cation-independent mannose 6-phosphate receptor to become trapped in clusters of early endosomes. |
| GO:0030119 AP-type membrane coat adaptor complex | IBA GO_REF:0000033 | MODIFY | Summary: Right family of complexes, one level too shallow. GO:0044599 AP-5 adaptor complex exists, is a child of this term, and is exactly what PTHR16082 is the mu subunit family of. Reason: The parent term is not wrong, and in particular it is not a clathrin-requiring term. Its definition covers heterotetramers that link clathrin or another coat-forming molecule to a membrane, so AP-5 belongs under it despite not associating with clathrin. That matters because the analogous review of AP3M1 found a whole-family mu node carrying a clathrin cargo-adaptor term seeded only by AP-1 and AP-2 mu subunits. Nothing of that kind reaches AP5M1: its only PANTHER assignment is PTHR16082, a family of AP-5 mu subunits and nothing else, every IBD on PTN002699615 is seeded by AP5M1 itself, and no clathrin term arrives. The single defect here is granularity. The donors agree, and a strictly more specific term is available and is already on the gene from ComplexPortal: GO:0044599, whose definition is explicitly agnostic about whether AP-5 forms clathrin coats, and whose ancestors include GO:0030119. The node should assert the child. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: GRANULARITY MISMATCH Sources checked: PANTHER:PTN002699615 · PTHR16082 AP-5 complex subunit mu-1 IBD node at taxon:2759 SUPPORTS TRANSFER The node is correctly placed and the target is inside the inheriting clade. The family contains AP-5 mu subunits only, so there is no cross-complex leakage of the kind found in the general AP mu family. What the node asserts is simply one level shallower than the evidence supports, and the child term GO:0044599 was available. UniProtKB:Q9H0R1 · AP5M1 (human, the target itself) SUPPORTS TRANSFER The target's own IDA from PMID:22022230, where immunoprecipitating tagged sigma-5 brought down mu-5 and beta-5, seeds this IBD. The seed evidence is for AP-5 membership specifically, which is why the parent term under-describes it. Proposed replacements: AP-5 adaptor complex Supporting Evidence: PMID:22022230 we suggest that C14orf108 and DKFZp761E198 should be renamed μ5 and β5, respectively, and that the complex that they form should be called AP-5. PMID:22022230 AP-5 does not associate with clathrin and is insensitive to brefeldin A. PMID:29381698 the highly conserved cargo binding sites found in the other AP complexes are absent in AP-5. file:human/AP5M1/AP5M1-bioinformatics/RESULTS.md AP5M1 retains **3 of 14** (K240, I461, R465). The comparators retain 11/14 |
| GO:0030119 AP-type membrane coat adaptor complex | IDA PMID:22022230 The fifth adaptor protein complex. | MODIFY | Summary: The experiment behind this row demonstrates membership of AP-5 specifically, not of the AP family in general, so the annotation should carry the child term. Reason: The direct evidence is a co-immunoprecipitation. Anti-GFP pulled tagged sigma-5 out of HeLa lysates together with mu-5 and beta-5, alongside a yeast two-hybrid interaction between mu-5 and beta-5 residues 16 to 229 and reciprocal destabilisation on knockdown. That establishes which complex, and the paper names it in the same breath. GO:0044599 was created for exactly this complex and is already on the gene from ComplexPortal; pointing the IDA at it removes the only granularity defect in the record. The parent stays true, so this is a deepening rather than a correction. Proposed replacements: AP-5 adaptor complex Supporting Evidence: PMID:22022230 the anti-GFP antibody not only brings down the construct itself, but also C14orf108 and DKFZp761E198. PMID:22022230 C14orf108 is stabilised by binding to DKFZp761E198. PMID:23825025 we show that the four AP-5 subunits can be coimmunoprecipitated with SPG11 and SPG15, both from cytosol and from detergent-extracted membranes, with a stoichiometry of ∼1:1:1:1:1:1. |
| GO:0031902 late endosome membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Faithful mapping of the UniProt late endosome membrane location, and the most informative compartment term on the record for a peripheral coat component. Reason: Mechanically correct, and the underlying UniProt line is at ECO:0000305 from PubMed:22022230, a curator inference combining the LAMP1 colocalisation with the membrane-versus-cytosol fractionation, with the peripheral and cytoplasmic-side qualifications that make biological sense for an adaptor complex. The compartment has since been supported independently by the co-staining of endogenous AP5M1 with Rab7 in iPSC-derived retinal pigment epithelium and by the structural work placing AP-5's sorting role at late endosomes. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0151 · UniProt subcellular location Late endosome membrane SUPPORTS TRANSFER Resolves to the late endosome membrane line of Q9H0R1, annotated peripheral membrane protein, cytoplasmic side, at ECO:0000305 from PubMed:22022230, with the record's own note that the protein may cycle on and off membranes. Mapped without distortion. Supporting Evidence: file:human/AP5M1/AP5M1-uniprot.txt Late endosome membrane {ECO:0000305|PubMed:22022230}; Peripheral PMID:40081374 a punctate pattern of AP5Z1, AP5M1, and AP5B1 staining and co-localization with markers of late endosomes and the Golgi PMID:40175557 SPG11-SPG15 can cooperate with the fifth adaptor protein complex (AP5) involved in membrane sorting of late endosomes. |
| GO:0044599 AP-5 adaptor complex | NAS PMID:22022230 The fifth adaptor protein complex. | ACCEPT | Summary: Core, and the single most informative annotation on the gene. AP5M1 is the mu subunit of AP-5, an obligate assembly whose subunits destabilise one another's absence. Reason: Membership is established three independent ways. Yeast two-hybrid with full-length mu-5 as bait recovered beta-5, mapping to the N-terminal quarter that contacts mu in AP-1 and AP-2; co-immunoprecipitation of tagged sigma-5 brought down mu-5 and beta-5; and the six components co-precipitate at roughly equal stoichiometry with SPG11 and SPG15. Obligacy is shown by protein-level dependence in both directions, mu-5 falling when beta-5 is knocked down and falling again in AP5Z1-nonsense patient fibroblasts, which the authors attribute to the instability of unassembled AP subunits. The term's definition names mu5 explicitly and is deliberately agnostic about clathrin, which matches the evidence that AP-5 is absent from clathrin-coated vesicle fractions. Supporting Evidence: PMID:22022230 we suggest that C14orf108 and DKFZp761E198 should be renamed μ5 and β5, respectively, and that the complex that they form should be called AP-5. PMID:26085577 concomitant reduction in levels of µ5 compared with controls, which is due to protein instability of AP subunits that occurs in the absence of complex assembly PMID:41915273 TurboID-based proximity labeling revealed reproducible interactions with AP5B1 and AP5M1 subunits |
| GO:0042147 retrograde transport, endosome to Golgi | IMP PMID:22022230 The fifth adaptor protein complex. | NEW | Summary: The specific transport step AP-5 acts in, missing from the record. Loss of AP-5 leaves the cation-independent mannose 6-phosphate receptor and the Golgi proteins GOLIM4 and GOLM1 stranded on endosomes instead of returning them to the Golgi. Reason: GOA stops at endosomal transport, two levels above what is now known. The step was identified by subcellular fractionation profiling and quantitative mass spectrometry of AP5Z1-knockout HeLa cells, followed by immunolocalisation of the affected cargo (PMID:29381698). That experiment perturbs zeta rather than mu-5, so it is not on its own grounds for a mu-5 annotation. But mu-5 is destabilised whenever the complex fails to assemble (PMID:26085577), so an AP5Z1 knockout is an AP-5 knockout, and the cargo phenotype it produces is the same one that mu-5's own siRNA produces. That is why this row is coded IMP against PMID:22022230, the paper that knocked down mu-5 itself and first saw the CIMPR mislocalise, with PMID:29381698 supplying the identity and direction of the step and PMID:23825025 confirming that mu-5 is among the six knockdowns giving the phenotype. The broader parent GO:0016197 is retained on its own rows rather than replaced, because the family-level IBA should not inherit a human-specific step. Supporting Evidence: PMID:29381698 impaired retrieval of the cation-independent mannose 6-phosphate receptor (CIMPR), GOLIM4, and GOLM1 from endosomes back to the Golgi region PMID:22022230 When C14orf108 is depleted, both the CIMPR and Vps26 localise to much larger puncta in the perinuclear region of the cell PMID:23825025 all six knockdowns cause the cation-independent mannose 6-phosphate receptor to become trapped in clusters of early endosomes. PMID:29381698 sortilin may act as a link between Golgi proteins and the AP-5/SPG11/SPG15 complex |
Loading supporting content…
Download this section (compressed HTML)Q: For the PAINT curators of PTHR16082: all five IBDs on PTN002699615 are seeded by human AP5M1 alone, and one of them asserts the parent term GO:0030119 rather than GO:0044599 AP-5 adaptor complex, which is a child of it and is what this family is the mu subunit family of. Was the parent chosen deliberately in 2019, or would the node take the child now?
Q: For the AP-5 groups: does the mu-5 MHD bind any peptide motif at all? The tyrosine subsite is gone, but three of the fourteen mu-2 contact positions survive, so the open question is whether the remaining face binds a different class of signal or has been repurposed for a protein-protein contact inside the coat.
Q: Is mu-5 required for AP-5 function, or only for AP-5 assembly? Every phenotype attributed to mu-5 is also produced by losing another subunit, because the subunits destabilise one another, so no published experiment separates a structural requirement from a functional one.
Q: Why does complete loss of AP5M1 cause an isolated macular dystrophy while complete loss of AP5Z1 causes spastic paraplegia, when the two are subunits of the same obligate complex and each destabilises the other?
Q: Does the pathogenic AP5M1 p.Tyr313Cys allele disrupt the complex or the fold? Tyr313 sits inside the MHD and is a tyrosine at the same alignment column in all five human mu subunits and in mouse and Arabidopsis mu-5, and the corresponding AP4M1 residue Tyr284 is mutated in spastic paraplegia type 50.
Q: Is the MUDENG cell-death phenotype a downstream consequence of impaired endolysosomal homeostasis rather than a separate activity? The two literatures on this protein cite each other almost not at all.
Q: Does isoform 2, which replaces residues 241 to 284 and deletes everything C-terminal of 284 and so removes most of the MHD, make a protein, and if so does it assemble with beta-5?
Experiment: Proteomic peptide phage display, or a synthetic peptide array tiling the cytosolic tails of the candidate AP-5 cargo (CIMPR, sortilin, GOLIM4, GOLM1), against the recombinant mu-5 MHD, with the mu-2 MHD and a YxxPhi peptide as the positive control. A negative result should be published: it would be as informative as a positive one, and it is currently the untested half of the only mechanistic statement anyone makes about this subunit.
Hypothesis: Mu-5 has lost peptide-motif recognition entirely, rather than having switched to a different sorting signal.
Type: peptide-binding screen
Experiment: Build separation-of-function alleles from the cryo-EM model. Identify the mu-5 surface that contacts beta-5, make a mu-5 that cannot assemble, and ask whether it still supports CIMPR and GOLIM4 retrieval in AP5M1-null cells. Pair it with an MHD deletion that still assembles, to ask the converse. This is the experiment that would separate the structural from the functional requirement.
Hypothesis: Mu-5's contribution to AP-5 is structural, so an allele that cannot assemble and an allele that assembles but lacks the MHD should give different phenotypes.
Type: structure-guided mutagenesis with rescue
Experiment: Knock the three AP5M1 macular dystrophy alleles, p.Arg33Ter, p.Trp389Ter and p.Tyr313Cys, into iPSC-derived retinal pigment epithelium and measure AP-5 subunit levels by western blot, CIMPR and GOLIM4 distribution by immunofluorescence, and autolysosome tubulation after starvation and refeeding. The two nonsense alleles give the null baseline against which the missense can be graded.
Hypothesis: The AP5M1 p.Tyr313Cys missense allele is a hypomorph rather than a null.
Type: isogenic iPSC-RPE allelic series
Experiment: Subunit dropout in the in vitro membrane-remodelling assay: compare SPG11-SPG15 alone, AP-5 alone, the full assembly, and an assembly lacking mu-5, on giant unilamellar vesicles containing PI3P, scoring tubule formation and curvature preference. Use human subunits throughout, since the deposited structures use mouse mu-5 and zeta.
Hypothesis: Membrane deformation by the AP-5-SPG11-SPG15 assembly is contributed by SPG11 and SPG15, not by AP-5 or mu-5.
Type: reconstituted liposome tubulation assay
Experiment: Repeat the TRAIL sensitisation and overexpression death assays in AP5M1-null cells derived from the macular dystrophy families and in cells expressing an assembly-incompetent mu-5, and in parallel measure lysosomal proteolytic capacity and autophagic flux in the same cells. If the death phenotype tracks with lysosomal failure and not with mu-5 protein per se, the two literatures describe one function.
Hypothesis: The MUDENG cell-death phenotypes are downstream of disrupted endolysosomal homeostasis rather than a second activity of the protein.
Type: genetic epistasis with lysosomal function readout
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: No molecular function has been established for mu-5 itself. It has the mu-homology domain fold but not the tyrosine-signal binding site, and no sorting motif of any kind has been identified for AP-5, so it is not known whether mu-5 recognises cargo at all, and if it does, what it recognises.
OPEN BIOLOGY MF_DARK
What is known: Firmly established: mu-5 carries an MHD over residues 206 to 476 by UniProt, PROSITE PS51072, CDD cd09256 and Pfam PF00928; it binds beta-5 through its N-terminal domain in the canonical mu-beta arrangement; and it is an obligate member of AP-5. Equally firmly established, and measured for this review: of the fourteen mu-2 residues that contact the TGN38 DYQRLN peptide in PDB 1BXX, AP5M1 retains three, and the three closest contacts to the signal tyrosine, mu2 F174, D176 and W421, are serine, serine and alanine in mu-5, while AP4M1 and AP1M1 retain all three. So the fold is present and the tyrosine site is not.
Significance: AP-5 is the only AP complex with no known sorting signal. Whether it recognises a different motif, recognises a lipid or conformational feature instead, or has abandoned cargo recognition entirely and become a purely structural member of a coat, determines what kind of machine the AP-5-SPG11-SPG15 assembly is, and it decides whether any cargo-binding molecular function should ever be annotated to this gene. It is also the reason this review annotates mu-5 as a structural subunit rather than as an adaptor.
What would resolve it: An unbiased search for the motif, for example peptide or proteomic-peptide phage display against the recombinant mu-5 MHD, or crosslinking mass spectrometry of the assembled hexamer on membranes bearing candidate cargo. A negative result would be informative in itself and should be recorded, since the alternative hypothesis, that mu-5 is structural, is testable by asking whether an MHD-deleted mu-5 still supports complex assembly and CIMPR retrieval.
Provenance (the field's own admissions):
Gap: It is not known which subunits of the AP-5-SPG11-SPG15 assembly carry out the membrane deformation the complex performs in vitro, so mu-5's contribution to the one demonstrated molecular activity of the assembly cannot be stated more precisely than complex membership.
NARROWING BIOLOGYCURATION RESIDUAL_SUBGAP
What is known: Established: the reconstituted assembly binds phosphatidylinositol 3-phosphate, senses membrane curvature and remodels membranes in vitro, and is needed to initiate autolysosome tubulation; spastizin supplies a FYVE domain that binds PI3P and was proposed to dock the coat, and spatacsin was proposed to scaffold it. Not established: whether AP-5 itself, or mu-5 in particular, contributes anything to bilayer deformation beyond being part of the assembled particle.
Significance: This is the only molecular activity ever demonstrated for anything containing mu-5, so resolving it would give the gene its first real molecular function rather than a structural placeholder. Until it is resolved, contributes_to is the strongest defensible statement, and the added caution is that the mu-5 and zeta chains in the deposited cryo-EM models are the mouse orthologues rather than the human proteins.
What would resolve it: Subunit dropout or truncation in the in vitro membrane-remodelling assay, reconstituting SPG11-SPG15 without AP-5 and AP-5 without mu-5, would attribute the activity. Reporting the assay on the human complex would remove the species caveat.
Provenance (the field's own admissions):
Gap: Whether the cell-death phenotypes reported for this protein under the name MUDENG reflect anything it does at physiological levels is unresolved, and the published effects point in opposite directions.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: Established: ectopic overexpression kills Jurkat, HeLa and cervical carcinoma cells, and the cervical carcinoma effect requires BAX; depletion in astroglioma and glioblastoma cells sensitises them to TRAIL, placing the protein downstream of Bid; and the protein is cleaved by caspase-3 at D276 and D290, both of which are aspartates inside the annotated MHD. Not established: any mechanism connecting these observations to AP-5, to membrane traffic or to the endolysosomal system, or any phenotype at endogenous expression levels that is not a sensitisation assay.
Significance: Two literatures describe this protein and cite each other barely at all. If the death phenotype is a consequence of disrupted endolysosomal homeostasis, which is plausible given that AP-5 loss produces a lysosomal storage phenotype, then it is downstream of the trafficking function and should not be annotated as a separate process. If it is a genuinely separate activity it would be the second function of the protein. GOA currently carries no apoptosis term for this gene, and this review proposes none.
What would resolve it: Test the death phenotype in cells carrying the AP-5-null alleles that are now available from the macular dystrophy families, and ask whether it requires assembly into AP-5 by testing a mu-5 variant that cannot bind beta-5. Measuring lysosomal function in the MUDENG depletion and overexpression systems would test the downstream hypothesis directly.
Provenance (the field's own admissions):
Gap: GO cannot express the process that AP-5, SPG11 and SPG15 are best characterised for, regenerating lysosomes from autolysosomes by tubulation and scission, so the most specific thing known about this gene's biology is annotated to a generic parent.
OPEN ONTOLOGYCURATION BP_DARK
What is known: Established: the assembly is required to initiate autolysosome tubulation, and its loss produces a lysosomal storage phenotype. Established about the ontology: no GO term names this process; the closest are GO:0170064 lysosome fission and GO:0097749 membrane tubulation, neither of which is specific to the autolysosome or to lysosome regeneration.
Significance: At least six genes in this axis, plus the lipid kinase and mTOR machinery acting with them, currently collapse onto GO:0007040 lysosome organization, which hides both the shared pathway and the specific defect that unites these hereditary spastic paraplegia and macular dystrophy genes.
What would resolve it: Create the term proposed in proposed_new_terms, as a child of GO:0007040, and re-annotate the AP-5, SPG11 and SPG15 gene products to it. Note that the same term is already requested from the other side of the complex in genes/human/ZFYVE26/ZFYVE26-ai-review.yaml; the two requests should be reconciled into one before either is acted on.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)