AGR3

Summary: ER localization is well supported for AGR3 by its signal peptide, QSEL retrieval motif, KDEL-receptor localization study, and airway epithelial cell localization evidence.

Reason: ER residence is the best-supported compartment for AGR3 and is consistent with both phylogenetic transfer and direct experimental localization evidence. ORGANELLES check (2026-11-04): the PAINT IBD for GO:0005783 sits at PTN001034224 (PTHR15337), seeded by AGR2 (human O95994, mouse MGI:1344405), TXNDC12/ERp18 (O95881) and their rodent orthologues and by AGR3 itself (Q8TD06); human AGR3 is inside that clade, and its own experimental ER localization is among the seeds, as expected. The only NOT (IRD at PTN009024331) is on a Xenopus branch and does not reach human AGR3. Qualifier is_active_in: the review does not show any AGR3 molecular activity in the ER (no foldase, isomerase or redox activity has been measured; the PDI-like function is family context), so the qualifier is not substantiated by activity evidence for AGR3 itself; the location claim it carries is right and is kept. The lumenal sub-compartment is added as a separate NEW row (GO:0005788).

Summary: AGR3 binding to alpha-dystroglycan was reported in yeast two-hybrid screens, but the evidence is cancer/extracellular-context interaction evidence rather than the main physiological airway/ER function of AGR3.

Reason: The term is specific enough to retain, and the IBA is consistent with the original AGR2/AGR3 two-hybrid evidence, but it should not be treated as the core function because the paper itself called for additional clinical-context validation.

Summary: ARBA transfer of dystroglycan binding is consistent with direct AGR3/DAG1 yeast two-hybrid evidence, but it remains a non-core cancer/extracellular interaction.

Reason: Keep the specific binding term as non-core: the interaction is reported, but the strongest physiological evidence for AGR3 concerns ER-localized control of airway ciliary beat regulation.

Summary: Automated ER localization is consistent with experimental literature and the UniProt-recognized AGR3 ER retrieval motif.

Reason: ER is the principal supported cellular compartment for AGR3. ORGANELLES check (2026-11-04): GO_REF:0000120 here combines ARBA with the UniProt subcellular location SL-0095, and the UniProt line "Endoplasmic reticulum" is curated from experiments (ECO:0000269, PubMed:18086916 and PubMed:25751668), so the root claim rests on direct evidence even though the ARBA component is a prediction. The row stops at the root only because UniProt records no sub-location; the lumen is added under rule 4 as a NEW row.

Summary: This Stitch-seq interactome annotation reports an AGR3 binary interaction but collapses it to generic protein binding, which is not informative for AGR3 functional curation.

Reason: Generic protein binding should not be retained as a meaningful AGR3 function, especially for high-throughput interactome rows without a mechanistic connection to AGR3 airway or ER biology.

Summary: The proteome-scale interactome map contributes several AGR3 binary interaction rows, but the resulting GO term protein binding is too broad to clarify AGR3 function.

Reason: The annotation should not be treated as core AGR3 biology because it is a broad high-throughput interaction label without a specific biochemical or pathway interpretation.

Summary: HuRI reports many AGR3 binary interaction partners, but generic protein binding obscures rather than explains the protein's biological role.

Reason: This high-throughput interaction evidence is useful as candidate-interactor context but should not be propagated as a core or informative molecular function annotation.

Summary: Immunofluorescence-based ER localization is consistent with direct literature showing AGR3 as an ER-resident ciliated-airway protein.

Reason: The experimental cellular-component annotation matches the best-supported localization for AGR3. ORGANELLES check (2026-11-04): HPA immunofluorescence is a whole-organelle assay that cannot separate lumen from membrane, so the row correctly stays at the root (rule 2); the lumen is added separately as a NEW row.

Summary: Proposed sub-compartment. AGR3 is a soluble protein with a cleaved N-terminal signal peptide and a C-terminal KDEL-like QSEL motif that keeps it in the ER, which places it in the ER lumen; the existing annotations stop at the organelle.

Reason: Added under the ORGANELLES policy (rule 4, 2026-11-04), beside the three root GO:0005783 rows, which are kept. Three independent lines agree. (1) Topology - UniProt records an N-terminal signal peptide (1-21) whose cleavage is experimentally supported (ECO:0000269, PubMed:15340161) and no transmembrane segment or membrane anchor, so mature AGR3 (22-166) is a soluble protein delivered into the ER lumen. (2) Retention - the C-terminal QSEL (163-166) is a KDEL-like motif; in PMID:18086916 Myc-tagged AGR3 was ER-localized, and deleting the motif moved it to the Golgi, the behaviour of a soluble lumenal cargo retrieved by KDEL receptors (UniProt FT MOTIF "Prevents secretion from ER", ECO:0000269). (3) Function and family - AGR3 is a thioredoxin-fold PDI-family member whose relatives AGR2 and ERp18 (the same PAINT node) are ER-lumenal; the review treats its redox activity as unproven, so this line is supportive only. The KDEL-receptor assay is on an overexpressed tagged construct, and no protease-protection or lumenal-fractionation assay has been done on AGR3, hence IC rather than IDA. core_functions is left at the root because no AGR3 activity has been shown in the lumen.

Summary: The protein-binding annotation from PMID:12592373 reflects specific yeast-two-hybrid interactions with C4.4a/LYPD3 and alpha-dystroglycan; dystroglycan binding is the more informative existing GO term.

Reason: Protein binding is too generic. Replace it with the specific supported DAG1 interaction term, while noting that the LYPD3/C4.4a interaction does not currently have an equivalently specific GO binding term in this review.

Proposed replacements: dystroglycan binding

Summary: Direct yeast-two-hybrid evidence supports AGR3 binding to alpha-dystroglycan, but this is not AGR3's main supported physiological function.

Reason: Retain the specific molecular interaction as non-core because the best functional evidence instead points to ER-localized regulation of ciliary beat frequency in airway epithelium.

Summary: NEW annotation. Mouse Agr3 loss reduces airway ciliary beat frequency and mucociliary transport while preserving ciliary ultrastructure, supporting a conserved role in regulating the beat frequency of motile airway cilia rather than ciliogenesis.

Reason: This narrows the prior epithelial-cilium-movement proposal to the observed phenotype. AGR3 is required for normal ciliary beat frequency, but current evidence does not identify the ER client or calcium-handling mechanism; the term is proposed for human AGR3 by sequence/orthology-supported inference from the mouse knockout and airway epithelial evidence.