pax3-a (Xenopus laevis, Q645N4) — curation notes

Project: projects/NEURAL_CREST_ORIGINS.md, Tier 2 (neural plate border specifiers).
Homeolog: pax3-b (Q0IH87; Xenbase pax3.L/.S). Q645N4 is the pax3.S gene (Xenbase XB-GENE-482744, chromosome 5S).

2026-10-05 — initial review

Identity and molecular activity

• Paired-domain (aa 34-161, PAI + RED subdomains) plus paired-type homeodomain (aa 220-279) transcription factor;
  Pax3/7 subgroup of the paired box family (UniProt; PANTHER PTHR45636:SF17).
• Direct, sequence-specific DNA binding and activation in frog: Pax3 binds a predicted site in the snail2 promoter
  (mutation abolishes, anti-Pax3 supershifts) PMID:24360906.
• Positive autoregulation PMID:24360906.
• Immediate-early direct targets (cycloheximide-resistant) include the crest specifiers PMID:24360906. Independent screen
  PMID:24360908.
• In vivo occupancy (X. laevis, stage 14 Pax3-FLAG ChIP-seq) PMID:38683994.
• All frog evidence is for activation; mammalian Pax3 can also repress (mouse Pax3 carries GO:0000122 IGI), but no frog
  repressor evidence found. MF chosen: GO:0001228 (activator) as NEW. Comparator: human PAX3 (P23760) and mouse Pax3
  (P24610) both carry GO:0045944 by IDA (QuickGO, see below), so the activator interpretation is family-consistent.

Expression (network layer)

• Induced in lateral/posterior neural plate by Wnt from posterior non-axial mesoderm PMID:10433827. This makes Pax3 a Wnt target, relevant to the Wnt-pathway rows.
• Co-expressed with Zic1 at the presumptive crest before foxd3/slug PMID:15843410.
• Also in hatching gland progenitors PMID:17409353.
• Border specifiers are "not always maintained in the neural crest progenitors themselves" PMID:24360906.

Gain of function

• Pax3+Zic1 together induce ectopic crest in ventral ectoderm; either alone only expands within dorsolateral ectoderm PMID:15843410.
• Pax3+Zic1 sufficient for full NC determination: migration and differentiation of derivatives PMID:23509273.
• Pax3 alone: hatching gland PMID:17409353.

Loss of function

• MO knockdown: Pax3 required for NC, acts downstream of Msx1, mediates Wnt/FGF8 PMID:15691759.
• Homeolog note: the GOA experimental rows are duplicated on pax3-a and pax3-b (Q0IH87) — the same PMIDs/ECOs
  (QuickGO, below). The MOs/mRNAs are generic "Pax3"; abstracts do not say which homeolog was targeted. Treat the
  evidence as applying to Xenopus Pax3 activity, shared by both homeologs.

Network layer decision

Pax3 is a neural plate border specifier (upstream of Snail/Foxd3/Sox8/Twist), induced by Wnt/FGF via Msx1, that
with Zic1 directly switches on the crest specifiers. It therefore does work at two levels:
1. Forming the border region — GO:0014029 neural crest formation (definition: "formation of the specialized region of
ectoderm between the neural ectoderm (neural plate) and non-neural ectoderm") — the correct level for a border
specifier under the pending project convention. ACCEPT both experimental rows.
2. Initiating crest fate — Pax3+Zic1 directly activate the specifiers and are sufficient for full determination.
GO:0014034 neural crest cell fate commitment (already on the gene, IMP/IGI PMID:15843410) captures this.
Did NOT add GO:0014036 fate specification: it is a descendant (part_of) of GO:0014034 which the gene already
carries, and the specification proper is executed by the downstream specifiers (sox8/sox9/sox10/snai2 narrowed to
GO:0014036 in Tier 1). Pax3 is the upstream inducer, so the commitment-level term (which includes induction of
the fate) is the better level, consistent with foxd3-a (also GO:0014034).
- Hatching gland: a separate border fate driven by Pax3 alone; KEEP_AS_NON_CORE (real, but frog/fish-specific border
derivative). Myogenesis/melanocyte roles (mammalian Pax3) have no frog annotation here; not added.
- FGF and Wnt signaling pathway rows: Pax3 is a transcriptional target/integrator downstream of the signals (its
expression is Wnt-induced; FGF8 acts via Msx1 and Pax3), not a component of the signal transduction cascade.
MARK_AS_OVER_ANNOTATED.

Comparator check (QuickGO, 2026-10-05)

Query: https://www.ebi.ac.uk/QuickGO/services/annotation/downloadSearch?goId=GO:0014029&goUsage=descendants&goUsageRelationships=is_a,part_of
(1911 rows), filtered for symbols matching pax. Results:
- X. laevis pax3-a (Q645N4) and pax3-b (Q0IH87): GO:0014029 IMP/IGI and GO:0014034 IMP/IGI/IEA (identical sets).
- X. tropicalis pax3 (Q28DP6): GO:0014029 and GO:0014034 by ISS.
- All other Pax3/Pax7 hits (many TrEMBL incl. pax7.L/.S, PAX3/PAX7 of other vertebrates) are GO:0014034 IEA (ARBA).
- Human PAX3 (P23760), mouse Pax3 (P24610), zebrafish pax3a/pax7: no annotation in the GO:0014029 subtree.
Mouse Pax3 instead carries GO:0001755 neural crest cell migration (IMP) and GO:0048066 developmental pigmentation
(IMP), plus GO:0045944 (IDA) and GO:0000122 (IGI). Human PAX3 carries GO:0045944 and GO:0045893 by IDA.
So the NC-formation/commitment terms are frog-specific in experimental annotation; mammalian Pax3 crest annotation is
at the migration level (Splotch). This is a coverage difference, not a convention against border genes carrying NC terms
(Pax3 in mouse does carry an NC-branch term). No NEW NC term proposed.

Evolution / outgroups

• Pax3/7 border expression and BMP responsiveness are conserved in amphioxus PMID:18562679,
  while most crest specifiers are absent from the amphioxus border PMID:18562679.
• So Pax3/7 belongs to the ancestral chordate border layer; what is vertebrate-specific is the downstream wiring
  (Pax3+Zic1 -> snail2/foxd3/sox8/twist1). Lamprey deploys the upstream border factors conservedly PMID:17765683.
• Open: whether amphioxus Pax3/7 can bind/activate vertebrate snail2/foxd3 regulatory elements (protein vs cis change).

GO gaps

• No neural plate border formation/specification term exists (project notes); raised as a suggested question.