AP5Z1 (human O43299, SPG48) is the zeta subunit of AP-5. The strongest functional synthesis is a structural contribution to an endolysosomal cargo-sorting assembly with SPG11 and ZFYVE26/SPG15. The historical repair phenotype is reproducible, but neither it nor the early fold prediction establishes helicase catalysis. Nuclear observations remain part of the evidence rather than being discarded because trafficking is the principal role.
PMID:20613862(https://pubmed.ncbi.nlm.nih.gov/20613862/), the original repair screen, was read at the targeted KIAA0415 Results, Figures 3–6 captions, relevant Discussion, BAC/RNAi-rescue, fractionation, immunoprecipitation and mass-spectrometry Methods. Two independent esiRNAs reduced DR-GFP repair in human HeLa and U2OS cells. A mouse-orthologue BAC rescued the human-cell phenotype. Reciprocal affinity isolations associated KIAA0415 with SPG11, SPG15, C20orf29/AP5S1 and DKFZp761E198/AP5B1. These are coassembly data, not isolated binary binding assays. Near-endogenous tagged protein yielded nuclear and cytoplasmic species; the bands were described as possible isoforms but were not mapped to O43299-1, -2 or -3. The species of every BAC bait was not independently recovered from Table S6. The human host is not used to infer every construct's species. The helicase proposal rested on threading and molecular-dynamics models, not ATPase or helicase activity measurements.
PMID:22022230(https://pubmed.ncbi.nlm.nih.gov/22022230/) was read at the search for additional subunits, AP5Z1 perturbation Results, Figures 8/9 captions, relevant Methods and Discussion. KIAA0415 knockdown perturbed CIMPR/Vps26 endosomal compartments. The direct tagged-localization panels primarily assayed other AP-5 subunits, so they are not relabelled as zeta-specific imaging. The paper identified zeta as a large adaptor subunit, noted the absence of essential helicase motifs, and explicitly accepted the prior repair phenotype as robust while proposing possible indirectness. This distinguishes a real phenotype from an unproven direct molecular mechanism. No clathrin-binding activity is assigned to AP-5.
PMID:23825025(https://pubmed.ncbi.nlm.nih.gov/23825025/) has an unchanged abstract-only canonical cache. A separate, normally generated full-text copy of the same original article (PMC3744948; DOI:10.1091/mbc.E13-03-0170) supplied the targeted body sections. The primary locators are Results “Stable association of SPG11 and SPG15 with AP-5,” Figures 1 and 6, and Methods “Quantitative BAC InteraCtomics (QUBIC).” Targeted QUBIC and stoichiometry Results/Methods, endogenous-zeta localization and specificity, SPG11 N-terminal pull-down, and coat/cargo Discussion were read. Human-cell quantitative isolations associate the four AP-5 subunits with SPG11/SPG15 at approximately equimolar abundance. Endogenous zeta puncta depend on the target and overlap the associated machinery. SPG11 N-terminal GST protein recovered zeta from cell extract; this was not purified binary binding. The paper proposed the cargo-sorting/scaffold arrangement rather than directly measuring cargo recognition. The richer candidate remains separate and does not change the canonical cache's availability flag.
PMID:26085577(https://pubmed.ncbi.nlm.nih.gov/26085577/) was read at patient/control fibroblast localization, antibody specificity, ultrastructure, human HeLa knockdown, corresponding Methods and the lysosomal-homeostasis Discussion. Endogenous AP5Z1 puncta overlapped LAMP1 and were lost in patient cells. A persistent reticular Golgi signal was explicitly identified as antibody cross-reactivity and is not positive AP5Z1 localization. Patient cells and AP5Z1-depleted HeLa cells accumulated abnormal multilamellar endolysosomal structures. These human observations independently support the older broad late-endosome/lysosome annotations without placing zeta inside the lumen. Figure legends and text were read, not image pixels.
PMID:29381698(https://pubmed.ncbi.nlm.nih.gov/29381698/) was read at human AP5Z1 knockout/retrieval Results, cargo pull-down/native co-IP, Figure 8 and S2 legends, the relevant cell/CRISPR/trafficking/biochemical Methods and Discussion. AP5Z1 loss altered retrieval of CIMPR and selected Golgi proteins. SPG15 residues 1–709 recovered CIMPR and sortilin from lysate but did not recover zeta. In contrast, native SPG15 complexes recovered both AP5Z1 and CIMPR: “Both AP-5 ζ and CIMPR are specifically brought down with SPG15-GFP.” These different experiments cannot be collapsed into direct zeta–cargo binding. Cargo-retrieval routes and receptor dependence also remain context dependent; no new narrowly defined transport-process term is introduced.
PMID:40175557(https://pubmed.ncbi.nlm.nih.gov/40175557/) was read as a complete abstract plus the published Figures 3–5 captions in the publisher preview. Full Methods/body were inaccessible. The captions describe zeta–SPG11 contacts, assembled-complex membrane assays and SPG11 interface perturbation in cells. RCSB 8YAH identifies mouse Q3U829 as the zeta chain, alongside mixed-species AP-5 components; this is not an isolated human-zeta biochemical assay. The structural evidence complements the earlier human complex/cargo experiments. No autonomous zeta membrane-remodeling, lipid-binding or cargo-recognition activity is inferred.
PMID:37871017(https://pubmed.ncbi.nlm.nih.gov/37871017/) was read at the targeted AP5Z1/SPG15 recruitment, tubulation, motor-association and neuronal transport Results, relevant Discussion, and construct/cell Methods. Mouse fibroblast and cortical-neuron contexts are explicit. Human SPG11/SPG15 constructs do not establish the species of every AP5Z1 reagent. The discussion permits indirect p150Glued coassociation through AP-5; no isolated human-zeta motor-binding term follows.
The actual HuRI record, PMID:32296183, identifies a human binary-interaction resource, but the relevant AP5Z1–NUP93 pair-level experiment was not adjudicated. The annotation stays UNDECIDED; different predominant compartments or absence from narrative text are not negative binding evidence. The BioPlex record, PMID:33961781, supplies a second screen assertion for SPG11. Its individual pair experiment remains uninspected, but the same partnership has independent positive support from the targeted 2010/2013 studies, explicitly cited separately. No cache or pair table was fabricated.
The official Human Protein Atlas AP5Z1 page was read at its current summary and cell table: supported nucleoplasm in A-431 and nuclear-speckle staining in U-251MG/U2OS using HPA035693. This is metadata-level support; image pixels and new antibody-validation experiments were not inspected. Nuclear localization is retained as secondary and does not independently establish repair chemistry.
All 24 original assertion objects, including partners, ordinary qualifiers and reference identifiers, and all three alternative products are preserved. The prospective decisions are nine ACCEPT, twelve KEEP_AS_NON_CORE, two MODIFY and one UNDECIDED. The two refinements name the established AP-5 complex and endosomal transport. True physical complex associations remain non-core under the user's explicit removal criteria; genericity alone is not evidence that they are false. No source field is rewritten to match later experiments.
One NEW molecular-function assertion proposes contributes_to GO:0140312 cargo adaptor activity, anchored to PMID:29381698, with PMID:23825025 assembly support and bounded 2025 structural context. The official definition requires the connection of cargo-associated membrane machinery and a coat scaffold. The combined native cargo association, stable assembly and scaffold interface support this activity of the complex. They do not establish autonomous zeta cargo binding. The primary anchor is the 2018 cargo study; the 2013 proposal alone would be insufficient. Independent prospective consultation accepted this distinction. There is no NEW biological-process term, isolated zeta helicase activity, or motor-binding assertion.
The core uses contributes_to_molecular_function, records AP-5 complex under complex membership, and retains the existing endosomal transport and lysosome organization processes. The official AP-type complex parent and endosomal-transport relations were checked. No AP5Z1/O43299 model was found in the local GO-CAM index; absence is not used as evidence for a new process.
The deterministic normal seed supplied the raw UniProt/GOA files and 24 PENDING assertions. The configured Falcon deep-research attempt and Perplexity fallback both failed (authorization and DNS failures); no provider report was fabricated. The failed provider attempt is recorded in the curation provenance; no private diagnostic file is used as biological evidence. These manual notes record the actual research instead.
The original six publication caches remain unchanged, including the abstract-only 2013 and 2025 records. The three additional read publications already existed as normal caches; no new fetch or overwrite was needed. Independent repair-source and all-24 consultations checked these biological interpretations and reading limits. Their reviewer identifiers describe collaborating agents, not another gene or another biological source. The scientific evidence is the linked primary literature above; curation sessions carry the review provenance.
All 24 original source assertions and three alternative products are preserved. The review contains nine ACCEPT, twelve KEEP_AS_NON_CORE, two MODIFY, one UNDECIDED and one NEW decision. The new cargo-adaptor assertion uses contributes_to and PMID:29381698 as its primary anchor; no new biological process is asserted. Independent prospective and final semantic consultation supported this scope. Two off-target assembly snippets were removed from location rows after consultation, retaining the original NAS citations and the directly relevant localization evidence. Schema, ontology and best-practice validation passed with seven generic-binding advisories retained under the explicit experimental-removal criteria. Rendering and the scaffolded history are checked on the final bytes.
The initial author-created cargo-adaptor assertion is withdrawn. Current GO:0140312 covers direct coat-scaffold binding together with a bridge to membrane, cargo receptor and deformation machinery; clathrin and COPII are examples, not an exclusive requirement. Thus absence of clathrin alone does not exclude AP-5. The introductory unknown-cargo statement in the 2018 paper also does not erase its positive Results: Figure 8C recovers native AP5Z1 and CIMPR with SPG15-GFP. Conversely, a co-isolation does not resolve the simultaneous molecular bridge. Combining it with the separate mixed-species 2025 structure leaves that precise newly asserted activity uncertain. The previous NEW row is preserved in the published history and is not replaced with a new membrane-bending or biological-process assertion.
The core now uses GO:0005198 structural molecule activity. Six targeted binding assertions involving established AP-5 partners are refined to that role using explicitly separate targeted assembly evidence. The 2013 QUBIC/stoichiometry Results and Figure 1 identify the stable complex; endogenous human-target depletion and Figure 6 support its assembly-dependent localization. Zeta depletion removes sigma5 puncta but does not remove SPG11/SPG15 puncta; this is not evidence that zeta stabilizes the accessory proteins. These do not resolve the species of every older BAC construct. Published 2025 Figures 2–3 corroborate zeta–SPG11 structural contacts, with mouse zeta in the resolved mixed-species assembly. The refinement does not convert every original co-IP or screen into a direct structure experiment, and it does not assign an autonomous cargo interface to human zeta. Replacement assertions should be coalesced while preserving their individual source provenance.
The SPG15 fragment result in 2018 Figure 8B remains an explicit limitation: the isolated N-terminal fragment recovered cargo but not AP5Z1. It is not positive evidence for zeta cargo bridging and does not negate the distinct native-complex result. It remains documented in the reference findings rather than attached as support for the withdrawn assertion. The separately corroborated BioPlex AP5Z1–SPG11 interaction remains non-core because its individual screen experiment has not been adjudicated; it is not given the targeted-assembly refinement. The unresolved HuRI AP5Z1–NUP93 pair remains UNDECIDED. Genericity or usual cellular compartment differences alone are not evidence that an experimental interaction is false.
The two earlier transport and complex refinements explicitly merge into already supported target terms. Public paper links and section locators now carry the body-reading provenance; private workflow paths are not evidence citations. All 24 original source assertion objects and three alternative products remain unchanged, with no cache or raw-source overwrite.
Final follow-up checks passed: full validation retains one generic-binding advisory for the explicitly corroborated BioPlex association; rendering and the new curation history validate. Integrity checks preserve all 24 original assertions, three alternative products, nine normal publication caches, both raw source files and the published history. The final review has nine ACCEPT, eight MODIFY, six KEEP_AS_NON_CORE and one UNDECIDED, with no NEW rows or processes.