Deep research status: Falcon deep research has now completed successfully (ATP6AP1-deep-research-falcon.md, 19 citations); see the synthesis section at the end of this file. The original PN-batch attempt timed out before the deep_research_unified tool bugs were fixed. This review uses that report together with the fetched UniProt and GOA records, cached publications, local Reactome entries, and the PN projection report.
Core function synthesis: ATP6AP1/Ac45 is best reviewed as an accessory/regulatory V0-sector subunit of the V-ATPase that supports complex assembly, targeting, stability, and activity rather than as a catalytic proton pump subunit. Human V-ATPase structure work defines ATP6AP1 as a V0 assembly hub PMID:33065002. The human deficiency study supports an Ac45/Voa1-like assembly-factor role and shows disease mutants fail to restore V-ATPase-dependent growth in yeast PMID:27231034.
Localization/function context: Ac45 localizes mainly to the ER and ERGIC in hepatocytes, not predominantly to the TGN or endosomal system in that experiment PMID:27231034. Mature V-ATPase activity is nevertheless central to endolysosomal and secretory-compartment acidification, and Reactome represents ATP6AP1 as an accessory V0 subunit facilitating acidification [Reactome:R-HSA-5252133 "V-type proton ATPase subunit S1 (ATP6AP1) is thought to function as an accessory subunit of the V0 subcomplex of V-ATPase, facilitating acidification"].
PN projection decision: the PN projection reports GO:0007042 lysosomal lumen acidification as already present in GOA and projects ATP6AP1 to GO:0060590 ATPase regulator activity [file:projects/PROTEOSTASIS/reports/pn_projection/pn_projected_annotations.tsv "ATPase regulator activity is the narrowest GO target that preserves the source mechanism without requiring a speculative complex-specific term."]. I accepted the lysosomal acidification projection as already supported and added GO:0060590 conservatively as a new reviewed MF term, using the same evidence to modify the broad GO:0140677 molecular function activator activity row.
Conservative exclusions: generic GO:0005515 protein binding rows were marked over-annotated because they do not describe the ATP6AP1 molecular function. Direct GO:0031267 small GTPase binding rows were removed because the key mTORC1 paper supports V-ATPase-dependent Rag nucleotide loading but did not detect direct Rag interactions with purified V-ATPase subunits PMID:22053050. HIF/iron and broad trafficking terms were kept as non-core secondary consequences where supported PMID:28296633.
The Falcon report (file:human/ATP6AP1/ATP6AP1-deep-research-falcon.md) is fully
consistent with the assembly-factor / structural-hub framing above and adds
structural and pathway detail. Citations use the report's DOIs except where a
PMID is already established in this file.
Structural confirmation of the V0 hub role (Wang et al. 2020, Mol Cell 80:501,
doi:10.1016/j.molcel.2020.09.029; the "structural hub" finding already cited here
as PMID:33065002). The cryo-EM structure shows ATP6AP1's luminal domain adopts a
β-prism fold homologous to LAMP-family domains (despite no sequence homology),
sits centrally inside the c-ring, and is the most connected V0 subunit
(>7,000 Ų buried surface), contacting ATP6AP2, c'', c(1/2/8/9) and subunit d.
This directly supports the assembly/stability MF over any catalytic role and
explains, at residue level, why disease mutations are loss-of-function: Y313C
destabilizes the luminal hydrophobic core, E346K disrupts a luminal salt bridge,
M428I perturbs TM contacts with the c-ring.
Cooperative V0 assembly with ATP6AP2 (Guida et al. 2018,
doi:10.1091/mbc.e18-04-0234). ATP6AP1 and ATP6AP2/(pro)renin receptor together
act as ER V0 assembly factors replacing the single yeast Voa1 function;
co-expression rescues V-ATPase-defective yeast and restores vacuolar quinacrine
accumulation. Relevant context for the neighboring ATP6AP2 review.
Autophagy linkage (non-core, recent/translational). Fei et al. 2024
(doi:10.1016/... breast-cancer lysosomal-gene study) and Yan et al. 2025 report
that ATP6AP1 supports autophagic flux via lysosomal acidification and, newly,
promotes autophagosome-lysosome fusion by enhancing Rab7-HOPS interaction;
overexpression drives chemoresistance. These are downstream/disease-context roles
and should remain non-core relative to the V0 assembly MF, but the Rab7-HOPS
fusion link is worth flagging as a candidate ALP-branch secondary function if it
is independently corroborated.
Other corroborated points (no change to calls): ER assembly then trafficking
to endolysosomes, Golgi (glycosylation), neuroendocrine secretory vesicles, and
osteoclast ruffled border; tissue-specific proteolytic processing (~40 kDa brain,
62 kDa liver, ~50 kDa B cells); V-ATPase-dependent (structural, not pump-direct)
role in mTORC1 amino-acid sensing. Net: no change to the core call (accessory V0
assembly/structural-hub subunit; GO:0060590 ATPase regulator activity + lysosomal
acidification), with strengthened structural justification.
This assessment supersedes the earlier PN-specific decisions above. It preserves all 54 machine-supplied assertions, their order, qualifiers, references and alternative-product records. Final decisions are 25 ACCEPT, 19 KEEP_AS_NON_CORE, nine UNDECIDED and one MODIFY. One qualified V-ATPase core remains; the legacy NEW ATPase-regulator assertion is withdrawn. No new GO annotation is proposed.
ATP6AP1 contributes to the assembled pump as a persistent V0 accessory component. Human structural contacts support assembly and stability; processed human Ac45 complementation in yeast supports function without isolating direct ATPase activation. Human hepatocyte ER/ERGIC localization identifies an assembly context, not ER-lumen acidification. The mouse osteoclast study PMID:22467241 supports targeting and acidification without demonstrating an assembly defect in that assay. The core therefore uses contributes_to proton-transporting ATPase activity and the general V-type complex GO:0033176, covering the distinct assembly and endolysosomal contexts without adding a location restriction to the complex itself. Existing lysosomal and endosomal acidification assertions remain accepted because this subunit contributes structure to the pump itself.
The existing molecular-function-activator annotation remains UNDECIDED: the inspected donor and assembly evidence does not resolve a specific activation mechanism. No ATPase-regulator replacement is asserted. Earlier rejection of small-GTPase-binding annotations is withdrawn in favor of curatorial deference and source-specific uncertainty where the exact ATP6AP1 assay remains unresolved. PMID:22053050 places V-ATPase upstream of Rag nucleotide loading; inspected purified Ragulator contacts involve d1 and D rather than proving direct ATP6AP1–Rag binding. The ATP6AP1-specific supplemental experiment was not reconstructed. These limits do not invalidate the curated complex-associated TORC1 role.
The ordinary source cache for PMID:38448650 was recovered unchanged. Its official identity is doi:10.1038/s41422-024-00938-z, PMC11061317. Root read the complete normal abstract, Fig. 5/6 Results and text legends, recombinant binding and real-time nucleotide-loading Methods after recovery. Purified Rheb associates directly with a synthetic ATP6AP1 tail. The last 12 residues mediate binding, while a 30-residue tail promotes GTP loading. A tri-aspartate mutant retains interaction but loses the loading effect. Human-cell rescue and signaling results support a separate Rheb/mTORC1 role, also evaluated by the independent primary-source consultation.
The authors interpret this mechanism as unconventional GEF activity. The inspected real-time assay mixes Rheb and tail peptide with either fluorescent GDP or fluorescent GTPγS in separate reactions. It does not explicitly measure release from GDP-preloaded Rheb during a nucleotide chase. The description records the reported signaling function; a dedicated GDP-release experiment is proposed. No NEW GO:0005085 or second catalytic core is asserted. This bounded decision neither refutes the paper nor establishes absence of GEF activity. No figure-pixel or raw-data reanalysis is claimed.
The linked author correction (PMID:40016599; doi:10.1038/s41422-025-01088-6) replaces a duplicated supplementary wound-healing image. The authors state that quantification and conclusions remain unchanged; the correction does not state a change to the nucleotide-loading experiments. That notice was independently read as a correction, not treated as a retraction or independent validation of the original raw data.
Other source boundaries remain recorded per reference. In particular, PMID:33065002's short cache lacks Results/Methods despite its full-text flag; selected original structural text was independently consulted. Screen-derived binding assertions without the exact target assay remain UNDECIDED. Supported generic binding stays non-core rather than being labeled false merely for lacking functional specificity. Neither the original GOA nor UniProt/publication/provider caches were edited.
The signaling-specific TORC1, amino-acid-response and contributes_to GEF-activator assertions are retained as non-core, aligning their classification with the single V0 assembly/acidification core. This preserves the curated complex-associated signaling evidence while keeping the ATP6AP1-specific supplemental assay limitation explicit. The lysosomal is_active_in assertion remains accepted because that location independently matches the pump core. Counts are now 20 ACCEPT, 24 KEEP_AS_NON_CORE, nine UNDECIDED and one MODIFY over the same 54 source assertions. No new GO assertion or additional catalytic core is introduced.
PMID:38448650 now has a machine-readable finding and an exact short source quotation for its reported C-terminal Rheb-loading effect. The separate fluorescent nucleotide-addition assays remain distinct from measuring release of preloaded GDP. PMID:40016599 is now represented explicitly as the externally inspected correction, with its ordinary cache correctly marked as lacking full text. Neither source cache was changed.
Supported generic-binding annotations remain non-core under the user-supplied ActionEnum definitions; this preserves correct peripheral observations without manufacturing a replacement function. The generic-binding skill preference does not override those definitions. Uninspected exosome evidence remains UNDECIDED. The original history record and earlier reasoning above are preserved as session provenance.
The follow-up passed focused validation with the same three advisories: two supported generic-binding rows retained as non-core and the existing provider report not used as direct annotation evidence. Rendering passed. The review remains DRAFT; no new repository-wide validation pass is claimed.
The review of PR #3550 at 6f4aedf04 correctly identifies the repository default: an uninformative generic binding annotation can be excluded even when its reported interaction is real. That informational-exclusion criterion is distinct from concluding that the experimental evidence is false. The earlier notes should not be read as saying that the default policy requires biological falsity or that these touched annotations qualify for a legacy exception.
This task explicitly directs that generic binding not be removed solely for informativeness, while preserving uncertainty when the relevant experiment cannot be adjudicated. The retained decisions are a scoped application of those task instructions. They are not a claim of compliance with the default generic-binding policy, a global policy revision, or a reason to silence its advisory warnings.
The retained rows distinguish viral NSP6 coassociation in human lung cells from ATP6AP2 coassociation and construct mapping. Neither is described as purified binary binding or a new ATP6AP1 enzymatic activity. V0 assembly-complex membership is a cellular-component statement; it is not a more specific molecular-function replacement for an enables protein-binding assertion. The existing pump contribution remains the sole core. The separate Rheb nucleotide-loading evidence still does not establish a GDP-preloaded chase assay, and the unresolved experimental rows remain UNDECIDED.
This addendum changes only the explanation of the decision. It adds no primary-source reading, assay verification, quotation, annotation or core function. All current annotation decisions, products, evidence limitations and source objects remain unchanged.