AIGR Deep Research — id3-a (Q91399, Xenopus laevis)

Focus: proposed new GO term "neural crest progenitor maintenance"

Hypothesis slug: nc-progenitor-maintenance-scrna

Executive Judgment

Verdict: PARTIALLY SUPPORTED (biology real; proposed evidence type weak; proposed bespoke GO term not recommended as framed).

Two separable claims are bundled in the seed:

1. Functional claim — id3 maintains the neural-crest (NC) progenitor pool.
   This is well supported by direct loss- and gain-of-function experiments in Xenopus
   (Kee & Bronner-Fraser 2005, PMID:15769946): Id3 knockdown eliminates NC precursors through
   cell-cycle arrest and apoptosis of the progenitor pool "rather than a cell fate switch,"
   and Id3 overexpression expands the NC domain. This is the strongest, most direct evidence
   and it is organism-appropriate.

2. Model claim + proposed test — the competence factors id3, myc, hes4 (hairy2),
   pou5f3 and lin28 are expressed continuously from blastula animal-cap cells into premigratory
   NC (progenitors "maintain" rather than "switch off and re-acquire"), testable by scRNA
   expression-continuity in the Briggs et al. 2018 time course (PMID:29700227).
   This is unresolved and the chosen test is under-powered for the curation question.
   Expression continuity is correlative and cannot by itself establish a maintenance function;
   the retain-vs-reacquire question is explicitly framed as open by the same research community
   (Schock, York & LaBonne 2023 review, PMID:35534333), and cross-species data show the premise
   is not universal (lamprey pou5 is absent from NC; York et al. 2024, PMID:39060477).

Most important caveats for the curator:
- The current annotation state already covers the biology. A programmatic pull of UniProt
Q91399 (ID3A_XENLA) shows id3-a already carries GO:0014029 neural crest formation (IMP) and
GO:0051726 regulation of cell cycle (IMP) — exactly the proliferation/pool-maintenance role from
Kee & Bronner-Fraser 2005 — plus a precise MF set (GO:0043425 bHLH transcription factor binding
(IPI), GO:0046983 protein dimerization activity, GO:0003714 transcription corepressor activity).
A new "neural crest progenitor maintenance" term would therefore be redundant.
- id3's molecular function is a general HLH inhibitor of bHLH/E-proteins (sequestration that
blocks their DNA binding), active across many stem/progenitor contexts (hESC, spermatogonia,
erythroid, osteoblast). "Neural crest progenitor maintenance" is a context-specific
developmental outcome (BP), not the gene product's primary molecular activity.
- A brand-new NC-specific GO term risks over-specificity and conflating a downstream developmental
phenotype with the core function.
- I did not re-run the Briggs scRNA analysis. I verified the dataset is public (GEO GSE113074,
136,966 X. tropicalis single cells over 10 timepoints, including neural-plate-enriched libraries)
but a full reanalysis (download of large count matrices, X. tropicalis ortholog mapping, trajectory
re-derivation) was out of scope and I will not fabricate a result. The judgment rests on primary
functional literature, direct UniProt/GO database evidence, and published synthesis — including
Briggs 2018, which itself "assess[es] conflicting models of neural crest development."

Evidence Matrix

GO Curation Implications (leads — require curator verification)

Recommendation: do NOT mint a bespoke "neural crest progenitor maintenance" term for id3-a. The
decisive point is that the supportable biology is already annotated on Q91399. Full decision table is
saved as the artifact id3a_GO_decision_table.csv. Summary:

On the proposed new term specifically: "neural crest progenitor maintenance" is (a) redundant
— GO:0014029 (neural crest formation, IMP) + GO:0051726 (regulation of cell cycle, IMP) already
encode the exact Kee & Bronner-Fraser 2005 phenotype; (b) too narrow/organism-flavored relative to
id3's pleiotropic bHLH-antagonist MF; and (c) not established by the proposed evidence type (scRNA
co-expression is correlative). If curators still want an explicit "maintenance" concept, prefer adding
existing GO:0019827 stem cell population maintenance with an "occurs in / part_of neural crest"
annotation extension (IMP, PMID:15769946), rather than creating a standalone ontology term. The
pluripotency-retention ("maintain vs re-acquire") framing belongs at review/GRN level, not as a
gene-specific GO term, because it is unresolved and not established by direct assay for id3.

Mechanistic Scope

• Immediate molecular function (direct): id3 is an Inhibitor-of-DNA-binding HLH protein. It
  heterodimerizes with class-I bHLH E-proteins (e.g., TCF3/E2A) through its HLH domain and prevents
  them from binding E-box DNA — i.e., it sequesters/inactivates bHLH transcription factors
  (PMID:39342788, PMID:42733085). It does not itself bind DNA. This was confirmed here at the
  sequence level: Q91399 lacks the arginine/lysine-rich basic region that canonical DNA-binding
  bHLH proteins carry adjacent to the HLH (max 13-aa K/R fraction 0.15 for Id3-a vs 0.54 for MYOD1,
  whose basic region is "DRRKAATMRERRR"). Hence any sequence-specific DNA-binding MF would be
  incorrect; the supported MF is GO:0043425 bHLH transcription factor binding / GO:0046983
  protein dimerization activity.
• Immediate cellular consequence (direct, Xenopus): keeps NC progenitors cycling and alive;
  its loss triggers cell-cycle arrest and apoptosis of the progenitor pool (PMID:15769946).
• Downstream / developmental outcome (inferred, not core function): maintenance of the NC
  progenitor domain, broad multi-germ-layer potential, and ultimately NC derivatives. "Neural crest
  progenitor maintenance" belongs here — it is a BP-level outcome of the molecular activity in a
  specific tissue, not the molecular activity itself.
• Model-level claim (unresolved): continuous competence-factor expression from blastula into NC
  as evidence that NC retains (vs re-acquires) a pluripotency program.

Conflicts and Alternatives

1. Correlation ≠ function. The seed's proposed scRNA continuity/co-expression test, even if
   positive, would show co-expression, not maintenance function. The maintenance function is
   already demonstrated more convincingly by perturbation (PMID:15769946).
2. Species variability undercuts universality. Lamprey pou5 is absent from NC (PMID:39060477),
   so "all five factors continuously expressed blastula→NC" is not a vertebrate-general fact; it is at
   best lineage-specific. This directly weakens one of the five named seed genes (pou5f3/oct25/oct60).
3. Retain vs re-acquire is explicitly open in the primary community's own synthesis
   (PMID:35534333: "either retained … or subsequently reactivated").
4. Function is pleiotropic, not NC-specific. id3/ID3 performs the same bHLH-sequestration and
   survival/pluripotency role in hESC, spermatogonia, erythroid progenitors and osteoblasts
   (PMID:35701409, PMID:39342788, PMID:42733085). A NC-specific GO term implies specificity the
   molecule does not have.
5. Paralog/allele considerations. Q91399 is the X. laevis id3-a homeolog (allotetraploid
   genome has id3-a and id3-b). Most classic functional work (Kee & Bronner 2005) predates explicit
   homeolog resolution and used X. laevis MO; Briggs 2018 is X. tropicalis (diploid, single id3).
   Curators should confirm orthology/homeology before transferring the tropicalis dataset claim to
   the laevis id3-a record.

Knowledge Gaps

Discriminating Tests (most efficient)

1. Trajectory co-expression reanalysis (the seed's own test): On the Briggs 2018 X. tropicalis
   object, order cells along the pluripotent→NC-border→premigratory-NC axis and quantify (a) fraction
   of cells co-expressing all five factors per stage, (b) presence/absence of an expression "gap" for
   each gene. Interpretation caveat: a positive result supports continuity but still does not
   establish a maintenance function.
2. Perturbation-with-rescue (decisive for function): Stage-specific id3-a depletion limited to the
   border (vs blastula) to test whether id3 is required to maintain an already-specified progenitor
   pool vs to induce it — distinguishes maintenance from induction.
3. Decouple proliferation/survival from differentiation block: Combine id3 loss with apoptosis/
   cell-cycle rescue and ask whether NC identity is still lost — tests whether "maintenance" is more
   than keeping cells alive.
4. Cross-species comparison: Repeat continuity analysis in zebrafish/lamprey datasets to test the
   generality challenged by PMID:39060477.

Curation Leads (require curator verification)

• Candidate references to attach to the id3-a review:
• PMID:15769946 — snippet to verify: "Morpholino oligonucleotide-mediated depletion of Id3
  results in the absence of neural crest precursors and a resultant loss of neural crest
  derivatives. This appears to be mediated by cell cycle inhibition followed by cell death of the
  neural crest progenitor pool, rather than a cell fate switch." → supports BP: NC development /
  stem cell population maintenance / regulation of apoptotic process (IMP).
• PMID:39060477 — snippet: "a lamprey pou5 orthologue is expressed in animal pole cells but is
  absent from neural crest" → qualifier against universal continuity.
• PMID:35534333 — snippet: "pluripotency is either retained in the neural crest from blastula
  stages or subsequently reactivated" → flags the model as unresolved.
• PMID:29700227 — the dataset underpinning the proposed analysis.
• Candidate GO actions:
• Do not create/assign a standalone "neural crest progenitor maintenance" term as primary.
• Assign/retain MF = negative regulation of DNA-binding transcription factor activity
  (GO:0043433) and bHLH-sequestration binding (not bare "protein binding").
• Assign BP = neural crest development (GO:0014029) and stem cell population maintenance
  (GO:0019827), ideally with an annotation extension localizing to neural crest, supported by IMP
  from PMID:15769946.
• Suggested curator questions:
• Is the proposed term meant as MF, BP or CC? (It reads as BP but is being argued from expression,
  an MF/co-expression observation.)
• Does AIGR want tissue-specificity encoded as a new term or via annotation extensions on existing
  terms?
• Has id3-a vs id3-b homeolog specificity been confirmed for Q91399?
• Suggested experiments: the four discriminating tests above, prioritizing #2 (stage-specific
  perturbation) as the only one that directly tests maintenance.

Provenance

• Reference identification (PMID:39060477 York 2024 Nat Ecol Evol; PMID:29700227 Briggs 2018 Science)
  fetched programmatically via NCBI E-utilities (efetch) within execute_code; abstracts quoted above
  are verbatim from that fetch.
• Functional/mechanistic literature retrieved via search_pubmed; snippets quoted are verbatim from
  returned abstracts.
• Current GO annotation state fetched programmatically from UniProt REST
  (rest.uniprot.org/uniprotkb/Q91399.json, HTTP 200) within execute_code; the GO list and domain
  annotation (single HLH domain aa32–84; InterPro IPR026052 "DNA-binding protein inhibitor") are
  verbatim from that record. Decision table saved as id3a_GO_decision_table.csv.
• Briggs dataset accessibility confirmed via GEO brief record for GSE113074 (HTTP fetch in
  execute_code): public, 136,966 X. tropicalis cells. No scRNA reanalysis was executed and no
  expression-continuity figure is presented, to avoid implying an analysis that was not run.
• Sequence analysis (Iteration 3): FASTA for Q91399 (Id3-a), Q02535 (human ID3) and P15172 (human
  MYOD1) fetched from UniProt; K/R-fraction window scan computed in execute_code. Result: Id3-a max
  13-aa basic-region K/R fraction 0.15 vs MYOD1 0.54 — Id3 lacks a DNA-binding basic region. This is a
  genuine computed result, not a synthesized figure.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist id3a GO decision table