Alpha3 is the catalytic component of one plasma-membrane sodium-potassium pump core. Its own ATP-dependent cycle exports sodium and imports potassium; the two direct transport processes are already represented in the machine annotations. No NEW term or redundant electrochemical-gradient ancestor is added to the core. Beta subunits support alpha3 assembly and stability, and FXYD proteins can modulate the complex. Binding these partners does not make alpha3 an autonomous chaperone.
Human alpha3/beta1/FXYD6 expressed in HEK293F cells produces pump current, reaches the cell surface and supports ATP turnover in purified preparations PMID:36075933(https://pmc.ncbi.nlm.nih.gov/articles/PMC9458724/). The nominal E1•ATP structural state uses AMPPCP according to Methods, so it is described here as ATP-like rather than literal ATP-bound. Separate phosphorylation and ATP-hydrolysis assays use ATP. Results and Methods disagree over E324 versus E327A as the first member of a four-residue mutant; that discrepancy is preserved without guessed renumbering. The selected reading establishes human catalytic function but is not a complete figure-pixel or supplemental-data audit.
Human alpha3 coassembles with human beta1, beta2 and beta3 in Xenopus oocytes, with stabilization demonstrated by pulse-chase and immunoprecipitation PMID:10636900(https://doi.org/10.1074/jbc.275.3.1976). Donor species and assay host are separate facts: these are human proteins in a frog host. The three UniProt alternative products remain unchanged, and the experiments are not assigned invented isoform identifiers or presumed to assay all products.
PMID:11546672(https://doi.org/10.1152/ajpcell.2001.281.4.C1336) provides ligand-binding evidence. Official ChEBI:472805 classifies ouabain as a steroid hormone and cardenolide glycoside. Classification plus binding alone does not establish a response process. Independently inspected original PMID:10636900 Methods and Fig. 6 measure intact-oocyte currents before and after 100 micromolar ouabain for human alpha3/beta1, subtracting uninjected-oocyte currents. This functional endpoint corroborates the two broad response terms as non-core pharmacological context. The original PMID:11546672 NAS provenance remains unchanged, with PMID:10636900 recorded as additional evidence. No endocrine transcriptional pathway or native-human-cell assay is inferred.
Human cardiomyocyte staining places alpha3 at plasma and intracellular membranes, but not at the T-tubules reported for alpha1 and alpha2 PMID:19751721(https://pmc.ncbi.nlm.nih.gov/articles/PMC4948948/). PMID:26373354 uses Xenopus alpha3 constructs in transgenic rods; the human construct in its inspected Methods is alpha4. Its inner-segment result remains valid orthology evidence, not a new human alpha3 IDA experiment. The existing ER/Golgi assertions from PMID:15260953 remain undecided because the full localization panels and wild-type/mutant distinction were not available in the bounded reading.
Original PMID:26224839(https://pmc.ncbi.nlm.nih.gov/articles/PMC4538662/) Results report direct human alpha3 binding to pathological amyloid assemblies by SPR (Kd 28.6 nM), alongside rat/mouse recombinant measurements, human cortical staining and rat-neuron perturbation experiments. Selected supplementary text identifies cell-free wheat-germ/asolectin production. Rat neuronal toxicity or neurite experiments are not relabeled as human assays. These contextual binding/response/maintenance assertions remain outside the transport core. The normal cached record is abstract-only despite separate access to indexed original sections.
GO:0086064 describes current transfer between adjacent cell cytoplasms through intercellular channels. Selected original PMID:19683723 evidence concerns mouse ventricular beta1 deletion, ATPase activity, contractility and calcium transients. It does not establish that particular intercellular coupling step for alpha3. The two existing rows are marked over-annotated for specificity; neither an incorrect gene assignment nor absence of all cardiac alpha3 roles is asserted. Root independently agreed with this narrow interpretation after its own selected source and term-definition reading.
The 17 PMID:32814053 generic interaction rows remain undecided because the exact target-pair assays and construct tables are uninspected, even though their pairs appear in UniProt. The PMID:24769233 extracellular-vesicle hit similarly requires its original target-level proteomics record. Existing broad ancestors and valid PAINT assertions are retained. Self-containing PAINT donor lists are legitimate descendant grounding, not circular evidence. No experimental assertion is removed because of a title, abstract-only access or a different foregrounded paralog.
All 79 machine source objects and three products are preserved: 46 ACCEPT, 11 KEEP_AS_NON_CORE, 20 UNDECIDED and two MARK_AS_OVER_ANNOTATED. One pump core and no NEW annotations are proposed. Root scientific review and the separate three-cache import closure were completed before canonical application. All 13 normal source records used by the draft match the canonical caches byte for byte. Existing caches and source alternatives are never overwritten. No provider report is fabricated; this draft is manual synthesis from normal source records and explicitly bounded external primary reading. The local GO-CAM index had no P13637/ATP1A3 hit in the recorded search; this is not a universal absence claim.
Core evidence limit: Human alpha3/beta1/FXYD6 HEK293F expression, whole-cell current, purified ATP turnover and structural states are directly read. The exact 3Na/2K stoichiometry is curated mechanistic knowledge, not claimed as newly measured by this selected assay reading. No independent protein-kinase, proton-pump, chaperone or amyloid-response core.
Availability flags reflect the normal cache, not the extent of external reading. Focused gene validation, rendering and history are recorded below after canonical application.
On 2026-09-29, just validate human ATP1A3 passed and just render human ATP1A3 succeeded. The review preserves all 79 source annotation objects and three alternative products. The curation history was scaffolded with just new-history; no new repository-wide validation result is claimed.
The Falcon command with Perplexity Lite fallback did not produce a provider report. Its first launch could not write to the default uv tool directory. A retry using writable temporary tool directories reached dependency resolution, where the pinned deep-research-client[cyberian]==0.2.7rc1 was unavailable in the offline cache. This review therefore uses the manual primary-source synthesis documented above.
The requested chemical axiom is explicit in the official ontology. The ChEBI ouabain record, outgoing relations, states CHEBI:472805 is_a CHEBI:26764 (steroid hormone) and separately CHEBI:472805 is_a CHEBI:38092 (cardenolide glycoside). The steroid-hormone parent relationships include has_role CHEBI:24621 (hormone). All were rechecked on 2026-09-29; this is a recorded classification, not an inference from the steroid scaffold. The GO:0071383 definition and parents describe a change in cellular state or activity following a steroid-hormone stimulus. The term does not require a transcriptional endpoint. A pharmacological concentration still does not establish physiological endocrine signaling, and the revised reason says so explicitly.
The steroid-hormone and glycoside reasons are now separate. Both preserve PMID:11546672 as the original NAS reference; the extra functional evidence remains PMID:10636900 with human proteins in Xenopus oocytes. Root reread both complete cached abstracts, while the original Methods/Fig. 6 inspection retains the previously recorded bounded scope. The glycoside row now quotes the cached abstract's isoform-specific K(+)/ouabain result. No cache is upgraded from abstract-only to full text.
The core now also includes a short literal pump-current anchor from PMID:36075933, alongside its ATP-turnover excerpt. Root reread the original cached Results paragraphs at lines 100 and 123: human alpha3/beta1/FXYD6 expression in HEK293F, enhanced whole-cell pump activity, and impaired activity of the cation-binding mutant. This supports the same single transport core; no new function or stoichiometric measurement is claimed. A suggested question records whether a more informative cardiac-glycoside-binding term is appropriate.
The request to automatically remove 17 generic protein-binding annotations remains a policy conflict with the supplied user ActionEnum and evidence rules. Their exact primary pair-assay evidence is still uninspected, so they remain UNDECIDED; corroboration in UniProt is not relabeled as a fresh primary-evidence read. No removal is invented to obtain approval. Action counts remain 46 ACCEPT, 11 KEEP_AS_NON_CORE, 20 UNDECIDED and two MARK_AS_OVER_ANNOTATED, with 79 original annotations, three products, one core and no NEW annotations.
Follow-up validation on 2026-09-29: just validate human ATP1A3 and just render human ATP1A3 passed after the reviewed changes. The new EDIT history was scaffolded with just new-history; no repository-wide validation PASS is claimed.