AIGR Gene Hypothesis Deep Research — Glycine max C6T1A2

Focused evaluation of the ProtNLM2 prediction: "negative regulation of abscisic acid-activated signaling pathway" (GO:0009788)

Hypothesis under review: ProtNLM2 predicts that the Glycine max C2H2-type zinc finger protein C6T1A2 functions in negative regulation of the abscisic acid (ABA)-activated signaling pathway (GO:0009788), by similarity to Arabidopsis ZFP7 (IPR053266). C6T1A2 has no curated GOA annotations.

Focus type: computational_prediction · Term: GO:0009788 · Organism: Glycine max (NCBITaxon:3847)

Summary

C6T1A2 (C6T1A2_SOYBN) is, without ambiguity, a plant Q-type (QALGGH) single-domain C2H2 zinc-finger transcription factor and a genuine member of the InterPro IPR053266 (ZFP7) / ZFP1–7 subfamily. In Arabidopsis, several members of this subfamily (ZFP1, ZFP3, ZFP4, ZFP6, ZFP7) act as negative regulators of ABA-activated signaling when overexpressed. In that sense the ProtNLM2 prediction is directionally plausible — it places the protein in the correct structural family and points at a process the family is genuinely associated with.

The problem is specificity. The exact GO:0009788 role cannot be established from sequence, orthology, or domain evidence alone, and it should not be promoted to a direct annotation. Three independent computational lines undercut the specific claim: (1) C6T1A2 is only ~45% identical to Arabidopsis ZFP7 and is essentially tied with ZFP2 (46.0%), a subfamily member not primarily characterized as an ABA repressor — so no confident 1:1 ortholog can be assigned; (2) the entire Arabidopsis ABA-repressor phenotype rests on redundant overexpression (gain-of-function), with loss-of-function lines near wild-type; and (3) 11 soybean paralogs share the identical IPR053266 / PTHR47593 label and identical single-finger architecture, creating a many-to-many over-annotation risk. There are no functional data whatsoever for the soybean protein itself (UniProt PE2, transcript-level evidence).

Bottom line: the molecular-function and localization annotations are defensible from sequence (DNA-binding transcription factor activity, zinc ion binding GO:0008270, nucleus), but the specific biological-process term GO:0009788 should NOT be entered as a direct/experimental annotation. At most it may be retained as an ISS/ISO lead with an explicit redundancy + overexpression-only caveat, or generalized to a broader, better-supported term.

Verdict: Weakly supported / over-annotated at the biological-process level.

Key Findings

Finding 1 — C6T1A2 is a plant Q-type (QALGGH) single C2H2 zinc-finger transcription factor

The primary sequence and domain architecture of C6T1A2 are entirely consistent with a canonical plant C2H2 zinc-finger transcription factor. UniProt records the protein (C6T1A2_SOYBN) as 210 aa, evidence level PE2 (transcript-level only). It contains a single C2H2-type zinc finger spanning residues 79–106, with the canonical Cys-X2-Cys … His-X3-His metal-coordinating spacing. Critically, it carries the plant-specific invariant QALGGH motif at residue 91 (context: …FSCNFCMRKFYSSQALGGHQNAHK…), the diagnostic signature of the plant Q-type C2H2 zinc-finger family that mediates DNA contact.

The InterPro/domain evidence is unambiguous and multi-source: IPR053266 (Zinc finger protein 7), IPR013087 / IPR036236 (C2H2 zinc finger), Gene3D 3.30.160.60, SUPFAM SSF57667, PROSITE PS00028 / PS50157, and PANTHER PTHR47593 (ZFP4-like). Two disordered/acidic low-complexity regions flank the single finger — an architecture typical of transcription-factor activation/repression modules. The only GO term currently on record is GO:0008270 (zinc ion binding), annotated by keyword (IEA-KW). This finding firmly establishes the molecular-function scaffold (zinc-dependent, DNA-binding TF) but says nothing, by itself, about which biological process the protein regulates.

Finding 2 — C6T1A2 belongs to the ABA-linked ZFP1–7 subfamily but is NOT a high-confidence 1:1 ZFP7 ortholog

Global Needleman–Wunsch pairwise alignment of C6T1A2 against a panel of characterized Arabidopsis reference zinc-finger proteins produced the following identity ranking:

The best hits cluster cleanly within the ZFP2/3/4/7 Q-type single-finger subfamily, well separated from the ZAT clade — confirming subfamily membership. But the crucial detail for curation is that ZFP2 (46.0%) edges out ZFP7 (45.4%): the difference (0.6 percentage points) is within noise, so the sequence provides no basis to single out ZFP7 — or any one member — as the specific ortholog. The prediction's implied ZFP7 provenance is therefore a family-level assignment mis-stated as a specific one.

Supporting the general repressor plausibility, a candidate C-terminal EAR-like repression motif (LxLxL pattern; IDLDL / LDLRL, in context …KKIDLDLRL) is present, consistent with — but not diagnostic of — a transcriptional repressor.

The key literature anchor for the family's ABA link is PMID: 24808098, which established (verified snippet): "regulated overexpression of ZFP3 and the closely related ZFP1, ZFP4, ZFP6, and ZFP7 zinc finger factors confers ABA insensitivity to seed germination." This anchors the ABA-repressor role for the subfamily — but explicitly via regulated overexpression (gain-of-function), not native loss-of-function, a distinction that is decisive for how strongly the term can be propagated.

Finding 3 — A distance tree places C6T1A2 in the ZFP1–7 subfamily without resolving a specific ABA-repressor ortholog

An all-vs-all Needleman–Wunsch identity/distance matrix across the 10-sequence panel, followed by UPGMA clustering, confirms the picture quantitatively. C6T1A2's nearest neighbours are ZFP2 (46.0%), ZFP7 (45.4%), ZFP4 (42.6%), ZFP3 (42.4%), then ZFP1 (37.2%); it sits clearly outside the tight ZAT10–ZAT6 clade (ZAT10/ZAT6 are ~70% distant from C6T1A2). The tree places C6T1A2 roughly equidistant (~54–58% distance) from the ABA-linked ZFP3/4/7 AND the development regulator ZFP2, with no 1:1 partner. This is the central topological result: the protein is a genuine subfamily member but has no resolvable orthologous anchor that would let a curator transfer a specific, member-defined phenotype (like ABA repression) with confidence.

{{figure:zfp_tree.png|caption=All-vs-all Needleman–Wunsch percent-identity distance matrix and UPGMA tree for C6T1A2 against characterized Arabidopsis ZFP1–7 and ZAT6/ZAT10 references. C6T1A2 falls within the Q-type ZFP1–7 subfamily but is roughly equidistant from the ABA-repressor members (ZFP3/4/7) and the developmental regulator ZFP2, with no 1:1 ortholog — illustrating why a specific ABA-repressor role cannot be transferred by orthology.}}

Finding 4 — C6T1A2 is one of ~11 soybean paralogs sharing the ZFP7-family label: a many-to-many over-annotation risk

A UniProt query restricted to Glycine max (taxon 3847) returned 11 soybean entries carrying IPR053266 (ZFP7 family) and 11 carrying PANTHER PTHR47593, out of 490 total soybean C2H2-type (IPR013087) proteins. A regex scan of the finger architecture showed it is uniform across the reference set: C6T1A2 and all Arabidopsis ZFP1–7 each contain exactly one C2H2 finger (C-x2-4-C-x9-14-H-x3-5-H) with exactly one QALGGH motif (C6T1A2 finger span 81–101). Because the domain signature is identical across all 11 soybean paralogs, an automated pipeline like ProtNLM2 has no feature to discriminate which — if any — inherited the specific ABA-repressor role. Propagating GO:0009788 to C6T1A2 on this basis would, by the same logic, propagate it to all 11 paralogs indiscriminately — the textbook definition of paralog over-annotation and frequency bias.

Mechanistic Model / Interpretation

The question decomposes cleanly into what sequence can and cannot establish:

      WHAT SEQUENCE/DOMAIN EVIDENCE ESTABLISHES (defensible)
      ─────────────────────────────────────────────────────
  C6T1A2 (210 aa)
   ├── Single C2H2 zinc finger (79–106)            → GO:0008270 zinc ion binding ✔ (already IEA-KW)
   ├── Invariant QALGGH DNA-contact motif (res 91) → DNA-binding TF activity ✔
   ├── Flanking disordered/acidic LC regions       → nucleus / TF module ✔
   └── C-terminal LxLxL (EAR-like) motif           → possible repressor (suggestive, not diagnostic)

      WHAT SEQUENCE/DOMAIN EVIDENCE CANNOT ESTABLISH (the seed's specific claim)
      ───────────────────────────────────────────────────────────────────────
  "Negative regulation of ABA-activated signaling" (GO:0009788)
   ├── Requires a specific ortholog anchor  → NONE (ZFP2 46.0% ≈ ZFP7 45.4%; no 1:1) ✗
   ├── Reference role is overexpression-only → loss-of-function ≈ WT (redundancy)     ✗
   ├── 11 soybean paralogs share the label   → cannot discriminate which inherits role ✗
   └── Zero functional data for the soybean protein (PE2, transcript-level)           ✗

The mechanistic reality of the reference proteins reinforces caution. In Arabidopsis, ZFP3 and its close relatives modulate ABA and light signaling and vegetative development by binding target promoters and repressing transcription (PMID: 38250442; PMID: 24808098). But the ABA-insensitivity phenotype emerges only under regulated overexpression, and knockdown/knockout lines are essentially wild-type "probably due to functional redundancy" (PMID: 38250442). Thus even for the best-characterized member, "negative regulation of ABA signaling" is a gain-of-function, redundancy-masked activity — not a demonstrated obligatory native function. Transferring such a term across ~45% identity and a species boundary, to a protein with no experimental data, compounds several layers of uncertainty.

The soybean literature adds a cautionary note about within-family mechanistic diversity: GsZFP1 (from Glycine soja), a C2H2 ZFP that notably lacks the QALGGH motif, negatively regulates ABA signaling and reduces ABA sensitivity (PMID: 22705253) — showing that ABA-related roles in soybean C2H2 ZFPs are real but are established experimentally, case-by-case, and do not map neatly onto the QALGGH-containing ZFP7 subfamily. Conversely, other C2H2 ZFPs act in the opposite direction on ABA/germination (e.g., IDD1/ENY promotes germination; PMID: 21571950), underscoring that the C2H2 scaffold alone does not fix the sign of ABA regulation.

The immediate molecular activity being tested is sequence-specific, zinc-dependent DNA binding by a single Q-type C2H2 finger, plausibly coupled to transcriptional repression. GO:0009788 is one to two steps removed from that direct activity: (direct) DNA binding → (proximate) repression of specific target genes → (pathway consequence) altered ABA signaling output → (organismal) ABA-insensitive germination. Curation should keep the direct activity distinct from the pathway-level consequence.

Evidence Base

How the literature bears on the finding: PMID: 24808098 is the origin of the family's ABA-repressor reputation and the most likely intellectual basis of the ProtNLM2 prediction — but it is overexpression-based. PMID: 38250442 tempers it by showing loss-of-function is near-WT and the primary characterized output is development/cell-wall, not ABA. PMID: 22705253 and PMID: 21571950 demonstrate that ABA roles among C2H2 ZFPs are protein-specific and can even reverse sign — reinforcing that family membership is not a reliable predictor of the specific GO:0009788 role.

GO Curation Implications (leads — require curator verification)

The MF/CC annotations are the defensible core. GO:0009788 is the precise but unsupported claim — plausible but not decidable from sequence. If any ABA link is retained, the most honest option is an ISS annotation with a "with/from" field pointing to the family plus an explicit note that the source phenotype is redundant and overexpression-derived. "Protein binding" is not recommended as a fallback; informative supported terms are available.

Conflicts and Alternatives

• Paralog confusion (primary risk). 11 soybean IPR053266 paralogs share identical single-finger/QALGGH architecture; orthology-based transfer of the ABA role is uncontrolled among them.
• No specific ortholog. ZFP2 (46.0%) marginally outscores ZFP7 (45.4%); ZFP2 is a development/floral regulator, so the "ZFP7-derived" provenance is really family-level, not member-level.
• Overexpression-only reference phenotype. ABA insensitivity of the Arabidopsis subfamily is gain-of-function; loss-of-function is near-WT (PMID: 38250442).
• Within-family mechanistic heterogeneity. GsZFP1, a soybean C2H2 ZFP lacking QALGGH, negatively regulates ABA (PMID: 22705253) — soybean ABA roles arise in distinct subtypes, established experimentally.
• Sign of regulation is not scaffold-fixed. IDD1/ENY promotes germination and lowers ABA (PMID: 21571950); the C2H2 fold does not determine the direction of ABA effect.
• Species/organism difference. All functional evidence is Arabidopsis or G. soja expressed in Arabidopsis; none is native G. max C6T1A2.

Limitations and Knowledge Gaps

1. No experimental data for C6T1A2 itself (UniProt PE2, transcript-level). Checked: UniProt evidence level, GOA. Why it matters: the whole hypothesis rests on transferred inference. Resolution: direct assays in soybean (below).
2. Orthology unresolved, not merely weak. Checked: pairwise identity + UPGMA. Why it matters: GO transfer by ISO requires a defensible 1:1 ortholog, which does not exist here. Resolution: phylogenomic tree with syntenic anchors across legumes + Arabidopsis with bootstrap support (OrthoFinder / Ensembl Plants gene tree + reciprocal-best-hit).
3. Repressor activity inferred from a motif, not measured. Checked: LxLxL/EAR-like motif scan. Why it matters: activation vs repression sign is functionally decisive. Resolution: transactivation reporter assay.
4. DNA-binding specificity (PWM) unknown. Checked: QALGGH presence only. Why it matters: target genes determine which pathway is affected. Resolution: DAP-seq / PBM / EMSA against candidate ABA-pathway promoters.
5. Redundancy in soybean uncharacterized. Checked: paralog count (11). Why it matters: single-gene perturbations may be masked, as in Arabidopsis. Resolution: higher-order CRISPR knockouts.

Proposed Follow-up Experiments / Actions (Discriminating Tests)

To distinguish "genuine soybean ABA repressor" from "family-level over-annotation":

1. Syntenic phylogenomics (most decisive in-silico test). OrthoFinder across G. max / A. thaliana / legumes with synteny + reciprocal-best-hit to test whether C6T1A2 is a true ZFP3/7 co-ortholog or a ZFP2-like member. If it does not resolve to ZFP7 (>60% identity / RBH), treat the ABA process as unsupported.
2. ABA germination assay. Inducible overexpression of C6T1A2 in Arabidopsis; score ABA insensitivity in seed germination (mirrors PMID: 24808098). Note a soybean null may be WT due to redundancy.
3. Transactivation / repression reporter. Protoplast dual-luciferase or yeast one-hybrid to test whether C6T1A2 activates or represses (tests the EAR-like motif inference).
4. DAP-seq / EMSA. Define the binding motif and test occupancy at ABA-signaling promoters (ABI-family, PYR/PYL, SnRK2 targets) to establish direct pathway linkage rather than assumed linkage.
5. Expression atlas. ABA/drought induction across the 11 soybean paralogs to identify which (if any) is ABA-responsive and prioritize candidates.
6. Subcellular localization. GFP fusion to confirm nuclear localization (supports the CC lead).

Curation Leads (verify before applying)

• Reference to attach: PMID: 24808098 — snippet to verify: "regulated overexpression of ZFP3 and the closely related ZFP1, ZFP4, ZFP6, and ZFP7 zinc finger factors confers ABA insensitivity to seed germination" (basis; Arabidopsis; overexpression).
• Tempering reference: PMID: 38250442 — snippet: T-DNA/silenced lines "were similar to wild-type plants or had only minor differences… probably due to functional redundancy"; ChIP confirmed ZFP3 promoter binding.
• Competing soybean example: PMID: 22705253 — GsZFP1 (QALGGH-lacking) "may be a promising gene for negative regulating ABA signaling."
• Candidate action change: Do not annotate GO:0009788 directly. Retain/add MF+CC terms (DNA-binding TF activity; zinc ion binding [kept]; nucleus). Keep GO:0009788 only as an ISS lead with caveat, not experimental.
• Suggested curator question: Does the local bioinformatics ortholog analysis resolve C6T1A2 to ZFP7 specifically (>60% identity / RBH)? If not, treat the ABA process as unsupported.

Provenance

UniProt REST (C6T1A2 + AtZFP references), an in-house Needleman–Wunsch all-vs-all identity/distance matrix with a UPGMA tree (artifact zfp_tree.png), C2H2 finger-count + QALGGH regex scans, a soybean paralog census (IPR053266 / PTHR47593 / IPR013087, taxon 3847), and EAR-motif scans were executed during this run (Iterations 1–3) and recorded in the knowledge state. Key computed results: single-finger / single-QALGGH architecture shared across C6T1A2 and AtZFP1–7; no 1:1 Arabidopsis ortholog (ZFP2 46.0% ≈ ZFP7 45.4% nearest neighbours); and 11 soybean paralogs share the IPR053266 ZFP7-family label (of 490 soybean C2H2 proteins), evidencing paralog over-annotation.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist zfp tree