AGR3 Noncanonical PDI: Redox Activity and Physiological Clients

Research question: Does human AGR3 have bona fide thiol-disulfide oxidoreductase/foldase
activity despite its noncanonical DCYQS thioredoxin-like motif, and are there supported
physiological substrates or receptors?

Approach: Literature-based synthesis (PubMed). Iteration 1. Direct AGR3-specific
experiments are distinguished throughout from inference drawn from the AGR2/PDI family.

1. Summary answer

There is no direct biochemical evidence that human AGR3 is a catalytically active
thiol-disulfide oxidoreductase, isomerase, reductase, or foldase. No AGR3-specific in vitro
enzymatic turnover assay has ever been reported, and the solved AGR3 crystal structure shows
it lacks elements of the canonical PDI active site (it carries a single, noncanonical
CXXS-type cysteine — DCYQS — not a redox-active CXXC). Its one well-supported physiological
role — regulation of airway ciliary beat frequency and mucociliary clearance — is
calcium-dependent and has never been tested for a requirement for catalytic-cysteine
chemistry. Any "oxidoreductase/chaperone" attribution to AGR3 is therefore inference by
homology to AGR2, for which covalent catalytic-cysteine chemistry is directly demonstrated.
Reported partners (alpha-dystroglycan, C4.4a) are unvalidated yeast-two-hybrid, cancer-context
leads. GO curation should not assert a specific catalytic molecular function for AGR3.

2. Key findings and evidence

2.1 No direct AGR3 catalytic evidence; structurally non-canonical

• The crystal structure of human AGR3 (Nguyen et al., 2018, PMID 29969106;
  Acta Cryst F, DOI ~10.1107/S2053230X18008129 — DOI uncached/unverified) states plainly
  that "AGR2 and AGR3 lack elements of the active-site motif found in other family members."
• A 2025 review (Law et al., PMID 40867591; DOI uncached) classifies AGR3 among
  non-canonical CXXS PDIs that act as "sensors or effectors of protein folding quality
  control" rather than classical disulfide-shuffling enzymes, and highlights "a unique role
  for AGR3 in cilia."
• No PubMed record contains an AGR3 reductase/oxidase/isomerase turnover assay, a
  catalytic-dead (Cys→Ser) rescue, or an identified covalent AGR3 client.

2.2 The airway/ciliary phenotype is calcium-linked, mechanistically open

• Bonser et al., 2015 (PMID 25751668; Am J Respir Cell Mol Biol, DOI ~10.1165/rcmb.2014-0464OC
  — DOI uncached): Agr3⁻/⁻ mice are viable with normal-appearing cilia but reduced ciliary
  beat frequency (~20% lower baseline, ~35% lower after ATP) and impaired mucociliary
  clearance. The defect disappears in calcium-free solution, indicating AGR3 is required
  for calcium-mediated regulation of ciliary function.
• AGR3 is ER-resident, restricted to ciliated cells, and — unlike AGR2 — not induced by
  ER stress. This diverges from AGR2's goblet-cell/mucin/UPR biology.
• The study performed no active-site mutagenesis, so whether the phenotype needs cysteine
  chemistry, client binding, extracellular signaling, or a non-enzymatic ER Ca²⁺-handling
  scaffold role is undetermined.

2.3 Catalytic-cysteine chemistry is proven only for AGR2 (the inference basis)

• Park et al., 2009 (PMID 19359471; J Clin Invest): a cysteine in AGR2's thioredoxin-like
  domain "forms mixed disulfide bonds with MUC2," a direct covalent role in mucin processing;
  AGR2 is essential for intestinal MUC2 production.
• Cloots et al., 2024 (PMID 38177501; Nat Commun): AGR2 acts as a mucin chaperone and
  rheostat of IRE1β; mutants "lacking their catalytic cysteine, or displaying...H117Y,
  were no longer able to dampen IRE1β activity."
• Human RIFTD disease variants in the AGR2 CXXS region (e.g., Ser84Arg, Takada et al., 2024,
  PMID 39673647; and Al-Shaibi et al., 2021, PMID 34237462) disrupt monomer–dimer
  equilibrium and mucin binding.
• Even for AGR2 this is best described as substrate-selective covalent chaperoning/sensing
  ("pseudo-PDI") rather than robust classical isomerase turnover. Extending it to AGR3 is an
  inference: AGR3 shares the fold and single cysteine but its clients and covalent chemistry
  are unproven.

2.4 Reported partners are leads, not validated substrates/receptors

• alpha-dystroglycan (DAG1) and C4.4a (LYPD3): sole source is Fletcher et al., 2003
  (PMID 12592373; Br J Cancer), a yeast-two-hybrid screen in ER⁺ breast tumour context,
  explicitly conditional ("which if replicated in clinical oncology would demonstrate a
  potential role..."). No orthogonal biochemical validation as ER-lumen folding substrates or
  signaling receptors; both are extracellular/GPI-anchored, topologically inconsistent with
  AGR3's ER-luminal residence except when secreted.
• Other AGR3 partners (FZD4/Wnt-β-catenin in CRC, Chi et al., 2020, PMID 31526829;
  ESR1/estrogen axis and tamoxifen resistance, Jiang et al., 2021, PMID 34519905) are
  cancer-context and indirect; AGR3 also circulates as a secreted serum biomarker in breast
  cancer (Garczyk et al., 2015, PMID 25875093).

3. Supported vs. refuted hypotheses

4. GO-curation recommendation

• Do not assign a specific catalytic molecular function (e.g., GO:0003756
  protein-disulfide isomerase activity, or generic oxidoreductase) to AGR3 by IDA/IMP — such a
  term would be supportable only by ISS/homology from AGR2, and should be flagged as such if used.
• Retain partner interactions (alpha-dystroglycan, C4.4a, FZD4) only as low-confidence
  "protein binding" (IPI) non-core leads.
• Annotate the well-supported layers: cellular component endoplasmic reticulum lumen;
  biological process regulation of cilium beat frequency / mucociliary clearance and
  calcium-dependent regulation (IMP from the knockout).
• Label the core airway molecular mechanism "unresolved" pending (i) catalytic-cysteine
  (Cys→Ser) rescue experiments, (ii) direct in vitro redox/isomerase assays on recombinant
  AGR3, and (iii) identification of a genuine ER-luminal AGR3 client.

5. Limitations and future directions

• This is a literature synthesis; no primary data were analyzed. AGR3-specific biochemistry is
  genuinely sparse, so absence of evidence is partly absence of study.
• DOIs above are provided from memory and are flagged as uncached/unverified; PMIDs are the
  authoritative identifiers.
• Priority experiments: recombinant AGR3 insulin-reduction/di-eosin-GSSG or scrambled-RNase
  isomerase assays; catalytic-cysteine mutant knock-in mouse airway rescue; ER Ca²⁺ imaging in
  AGR3-null ciliated cells; unbiased proximity-labeling (BioID/APEX) in ciliated airway cells to
  find bona fide ER-luminal clients; validation of C4.4a/alpha-dystroglycan binding by
  co-IP/SPR with topology controls.