AIGR Deep Research — Wheat A0A3B6RKV1 (JmjC-domain protein): catalytic competence and transfer of Arabidopsis JMJ22 developmental roles

Gene: Triticum aestivum A0A3B6RKV1 (UniProt A0A3B6RKV1, 511 aa; "JmjC domain-containing protein", ProtNLM name) Hypothesis slug: prediction-jmj22-developmental-roles Focus type: computational_prediction Terms in scope: GO:0010099 regulation of photomorphogenesis; GO:0010476 gibberellin-mediated signaling; GO:0010114 response to red light; GO:0010030 positive regulation of seed germination

Executive Judgment

Verdict: Partially supported — split decision.

• Family / subfamily assignment: SUPPORTED (and the seed's own labelling should be corrected). A0A3B6RKV1 is a genuine wheat co-ortholog of Arabidopsis JMJ22 (and its redundant partner JMJ20). Global alignment gives 62.3 % identity to JMJ22 (Q67XX3), with an identical F-box + JmjC(Cupin_8) domain architecture and identical InterPro/Pfam/PANTHER signatures. This clade is the plant JMJD6-type histone arginine demethylase group — not a KDM4/KDM5/JARID histone lysine demethylase, and not a KDM7/"PKDM7" H3K9 demethylase. The seed's "KDM/JARID subfamily … JMJ22/PKDM7D" framing mislabels the family: the correct assignment is JMJD6-related F-box+JmjC arginine demethylase.

• Catalytic competence: SUPPORTED at the residue AND structure level. The wheat protein retains the complete JmjC Fe(II) 2-His-1-carboxylate facial triad. The three UniProt-annotated active-site residues of JMJ22 (His324, Asp326, His407) map by alignment to conserved His330, Asp332, His413 in wheat (HxD…H). In the AlphaFold model (AF-A0A3B6RKV1-F1, v6) these residues are high-confidence (pLDDT 95–98) and their coordinating side-chain atoms converge into a single metal-binding pocket in 3D (cluster radius 2.82 Å; His–His N–N 3.11 Å). It is therefore predicted to be a catalytically competent 2-oxoglutarate/Fe(II) dioxygenase.

• Transfer of the four developmental BP terms: NOT supported / OVER-ANNOTATED. In Arabidopsis these four terms are all IMP (mutant-phenotype) annotations arising from one specific study (Cho et al. 2012, 22483719) in which JMJ20 and JMJ22 act redundantly within a defined module: phytochrome B → (repressor SOMNUS) → JMJ20/JMJ22 → removal of H4R3me2 at GA3ox1/GA3ox2 → GA biosynthesis → germination. These are context-dependent, redundant-pair, dicot seed-biology phenotypes. No wheat functional study exists (targeted PubMed search returned none). Projecting photomorphogenesis / red-light / GA-signaling / seed-germination BP terms onto the wheat protein by ProtNLM orthology alone is unwarranted.

Bottom line for the curator: keep/allow the molecular-function inference (histone demethylase / oxidoreductase / metal-ion binding) at ISS/IBA strength; do not carry over the four Arabidopsis-specific developmental biological-process terms to A0A3B6RKV1 without wheat evidence.

Evidence Matrix

GO Curation Implications (leads — require curator verification)

• Molecular function (retain, at ISS/IBA): histone demethylase activity (GO:0032452) and/or the more specific histone H4R3 demethylase-type activity is a defensible orthology-based inference given 62.3 % identity and an intact facial triad. metal ion binding (GO:0046872) and oxidoreductase activity (GO:0016491) are also supported. Prefer the demethylase term over a bare "protein binding."
• Biological process (do NOT transfer): GO:0010099, GO:0010476, GO:0010114, GO:0010030 should be removed / not asserted for A0A3B6RKV1. These are Arabidopsis IMP phenotypes of a redundant gene pair in a specific PHYB→GA3ox module. At most, mark as "not core / context-specific in Arabidopsis" with no wheat support.
• Cellular component: nucleus (GO:0005634) is a reasonable IBA-level carry-over consistent with a chromatin-modifying enzyme.
• Name correction lead: the ProtNLM/seed framing "KDM/JARID … PKDM7D" is inaccurate; the protein is a JMJD6-type F-box + JmjC arginine demethylase (JMJ20/JMJ22 clade).

Mechanistic Scope

Immediate molecular function under test: a 2-OG/Fe(II)-dependent JmjC dioxygenase acting as a histone arginine demethylase (Arabidopsis substrate H4R3me2). This is the direct gene-product activity and is supported by residue conservation.

Downstream / not-direct: GA biosynthesis (GA3ox1/2 derepression), phytochrome-B/red-light responsiveness, photomorphogenesis, and seed germination are pathway and developmental consequences observed in Arabidopsis loss/redundancy genetics — not properties of the enzyme's catalytic chemistry. They require the specific SOMNUS/PHYB regulatory wiring and the GA3ox target loci, none of which is established in wheat.

Conflicts and Alternatives

• Redundancy confound: the Arabidopsis phenotype requires jmj20 jmj22 double mutants; single-gene function is partly masked. A single wheat ortholog cannot be assumed to reproduce the double-mutant phenotype.
• Substrate ambiguity: the JMJD6/Cupin_8 clade includes arginine demethylases and lysyl hydroxylases (IPR050910). The wheat enzyme's true substrate (H4R3me2 vs. hydroxylation vs. other) is not experimentally fixed.
• Paralog/homeolog complexity in wheat: hexaploid wheat has A/B/D homeologs and likely several JMJ20/22-like paralogs; A0A3B6RKV1 may be one of several, further diluting any 1:1 functional transfer.
• Annotation carry-over: the four BP terms likely propagate from the Arabidopsis record via ProtNLM/orthology — a classic frequency-bias/over-annotation route.

Knowledge Gaps

1. Substrate/activity in vitro — checked residues only; no biochemical assay. Resolve with recombinant enzyme + H4R3me2 (and lysine/hydroxylation) substrates.
2. Wheat expression/phenotype — no wheat data. Resolve with germination-stage expression (light/GA treatments) and CRISPR/TILLING knockouts across homeologs.
3. Regulatory module conservation — is there a wheat SOMNUS/PHYB→JMJ→GA3ox circuit? Resolve by checking wheat SOMNUS and GA3ox orthologs and JMJ chromatin targets.
4. 2-OG binding beyond the triad — the C5-carboxylate-binding Lys/Ser/Thr were not individually validated here; a full active-site model (AlphaFold + comparison to a solved KDM) would strengthen the catalytic call.

Discriminating Tests

• Recombinant A0A3B6RKV1 demethylase assay on H4R3me1/2 (± Fe/2-OG dependence; His→Ala triad mutant as negative control).
• AlphaFold-DB model overlay onto a JmjC catalytic core to confirm triad geometry and 2-OG pocket.
• Wheat homeolog knockout germination phenotyping under R/FR light and GA/paclobutrazol.
• Phylogenetic tree of wheat JMJ + Arabidopsis JMJ11–33 to confirm placement in the JMJ20/22 clade and count wheat co-orthologs.

Curation Leads (verify before applying)

• Reference to attach: 22483719 — snippet: "the histone arginine demethylases, JMJ20 and JMJ22, act redundantly as positive regulators of seed germination" (family/MF support); and "upon PHYB activation, JMJ20/JMJ22 are derepressed, resulting in increased gibberellic acid levels through the removal of repressive histone arginine methylations at GA3ox1/GA3ox2" (shows developmental terms are module-/species-specific).
• Candidate MF term: GO:0032452 histone demethylase activity (ISS from JMJ22), with metal-ion binding / oxidoreductase; avoid "protein binding."
• Action change: reject/withhold the four developmental BP terms (GO:0010099, GO:0010476, GO:0010114, GO:0010030) for the wheat entry — flag as Arabidopsis-specific IMP not transferable.
• Suggested curator question: are there wheat expression or mutant data linking this locus to germination/GA before any BP assertion?

Provenance: UniProt REST records for A0A3B6RKV1 and Q67XX3; Needleman-Wunsch global alignment (62.3 % identity) with active-site residue mapping (His330/Asp332/His413 ↔ JMJ22 His324/Asp326/His407); AlphaFold DB model AF-A0A3B6RKV1-F1 (v6) triad geometry (pLDDT 95–98; coordinating atoms converge within 2.82 Å); InterPro/Pfam/PANTHER domain signatures; 22483719. Analyses executed in this run (Iterations 1–2).