sox2 (Sox2, XSox2; O42569) — Xenopus laevis — review notes

Project: NEURAL_CREST_ORIGINS (Tier 3, blastula programme retained in the crest; SoxB1 to SoxE
transition). Reviewed 2026-10-07. Read alongside genes/XENLA/sox3-a/ (SoxB1 paralog, reviewed in
parallel by another agent) and genes/human/SOX2/.

Identity

• O42569, 311 aa SoxB1 HMG-box transcription factor; HMG box 37-105, C-terminal 9aaTAD
  transactivation motif (by similarity to human SOX2) [file:XENLA/sox2/sox2-uniprot.txt].
• UniProt function: "Functions as a switch in neuronal development, participating in the
  differentiation of embryonic neuroectodermal cells into neural tissues by making the ectodermal
  cells responsive to FGF-neuralizing signals." [file:XENLA/sox2/sox2-uniprot.txt]
• Xenbase sox2.L (XB-GENE-865099). A UniProt search for sox2 in taxon 8355 (2026-10-07) returns
  only O42569 (no sox2.S entry), so homeolog splitting does not arise here.
• Paralogs: Sox1, Sox3 (SoxB1 group). Most frog loss-of-function reagents are dominant-negative
  Sox2 constructs (which can interfere with any SoxB1 protein) or combined Sox2+Sox3 morpholinos, so
  attribution to Sox2 alone is often uncertain [file:XENLA/sox2/sox2-deep-research-falcon.md "Most frog
  studies identify sox2 or Sox2 without establishing that an assay discriminates every closely
  related SoxB1 protein or X. laevis homeolog."].

Expression

• Blastula animal pole (pluripotent cells), with Sox3 PMID:30144418; also co-expressed with crest/pluripotency factors in animal caps
  PMID:39060477.
• Prospective neuroectoderm from the onset of gastrulation, induced by Chordin and suppressed by
  BMP4 PMID:9435279.
• Restricted to the neural plate by late gastrula; overlaps SoxE at the border only transiently
  PMID:30144418; PMID:30144418; PMID:30144418.
• Note: Buitrago-Delgado 2015 PMID:25931449 lists Sox3, not Sox2, among the genes with enhanced
  border expression at late gastrula ("several genes, including Oct60, Sox3, Vent2, Ets1, Zic1, Pax3,
  and Snail1, showing enhanced expression at the neural plate border by late gastrula stages"). Sox2
  is mentioned there only as a blastula-expressed pluripotency factor.
• Later: CNS, eye (retina and lens), retinal ciliary marginal zone, spinal-cord ventricular zone
  [file:XENLA/sox2/sox2-uniprot.txt]; PMID:25797152.

Molecular function

• HMG-box DNA-binding transcriptional activator (family-level; frog-specific evidence is indirect).
  Direct frog target: Rax (Rx1) CNS1 enhancer, bound by endogenous Sox2 together with Otx2, with
  synergistic activation and physical Otx2–Sox2 interaction PMID:18385377; PMID:18385377; PMID:18385377.
  -> The IPI protein binding (with Q91813 otx2-a) is MODIFIED to GO:0140297 DNA-binding transcription
  factor binding (Otx2 is a homeodomain DNA-binding TF; mouse Sox2 P48432 carries GO:0140297 by IPI,
  QuickGO 2026-10-07).
• Repression: SoxB1 gain of function at the border lowers foxd3/snail2 (see below), but no direct
  repressed target is mapped in frog; the IBA repression term is kept as non-core.

Biological roles

1. Neural induction / neuroectoderm competence (core). Sox2 alone does not neuralise but makes
   ectoderm responsive to FGF PMID:9435279;
   PMID:9435279. Dominant-negative Sox2
   blocks neural differentiation in caps and embryos PMID:10648237; PMID:10648237. Combined Sox2/Sox3 MOs block Chordin-induced neural induction, rescued by Sox2
   or Sox3 but not SoxE PMID:30144418.
   Caveat: dnSox2 is "Sox2-class"; the authors themselves generalise to SoxB1.
2. Retinal progenitor competence and timing of differentiation (core, neural progenitor role).
   Frizzled-5/Wnt maintains Sox2 in the optic vesicle; blocking Sox2 mimics loss of Xfz5 (less
   proliferation, no proneural onset, non-neural bias) PMID:15820691. Later Sox2 suppresses neuronal differentiation via Notch and favours Müller glia;
   Sox2 must be switched off for neurons to form PMID:19736324; PMID:19736324.
   -> NEW GO:0045665 negative regulation of neuron differentiation (IMP, PMID:19736324) and NEW
   GO:0060041 retina development in camera-type eye (IMP, PMID:15820691). Comparator: human SOX2 carries
   GO:0045665 (ISS) and mouse Sox2 carries GO:0045665 (IDA, PMID:16631155) (QuickGO 2026-10-07).
3. Spinal-cord regeneration (non-core, not annotated). Sox2 knockdown / dnSox2 disrupt recovery
   PMID:25797152. Left as narrative; no NEW term.
4. Blastula pluripotency (competence; not annotated). Combined Sox2+Sox3 MOs abolish animal-cap
   competence to form mesoderm/endoderm in response to activin, rescued by Sox2, Sox3, Sox9 or Sox10
   PMID:30144418. The
   evidence is for SoxB1 jointly (double knockdown), and the project convention (lin28a, id3-a, myc-a,
   snai1) raises stem cell population maintenance as a question rather than annotating it. Raised as a
   suggested question.

Does Sox2 belong in the neural crest network?

• SoxB1 must be switched OFF for crest. Activating SoxB1 (Sox2-GR or Sox3-GR, dex at stage 10) at
  the border lowers foxd3 and snail2, whereas SoxE raises them PMID:30144418. SoxB1 cannot substitute for SoxE in
  crest induction PMID:30144418. SoxE can substitute for SoxB1 in blastula pluripotency (overexpression), not in
  neural induction.
• Who switches whom off? Buitrago-Delgado attribute the SoxB1/SoxE exclusion "at least in part, to
  the repressive activity of Snail2 on Sox2 expression (Acloque et al., 2011)". The cited paper
  PMID:21920318 is about chick/mouse Sox3 and Snail2 at gastrulation ("this decision to
  internalize is mediated by reciprocal transcriptional repression of Snail2 and Sox3 factors"), not
  Sox2 and not the crest. So the crest-side repressor of Sox2 in frog is not directly shown; relevant to
  the sox3-a review too.
• Participation test (CLAUDE.md). Sox2's contribution to crest is (a) upstream, as part of the
  SoxB1 blastula competence that crest-precursor cells inherit, and (b) its removal. A factor that
  must be switched off does not do any of the work of crest formation; the work (fate specification)
  is done by SoxE, Snail, FoxD3 etc. Sox2 provides no scaffold, cofactor or catalytic contribution to
  the crest process. Only gain-of-function evidence links Sox2 to the crest (ectopic SoxB1 represses
  crest markers); no Sox2-specific loss of function expands crest. So: no GO:0014029, no
  GO:0014036, and no GO:0090301/GO:1905296 negative-regulation term.
• Comparator check (QuickGO, 2026-10-07, descendants of GO:0014029, GO:0014033, GO:0090299,
  GO:1905292). Zero annotations on human SOX2 (P48431), mouse Sox2 (P48432), zebrafish sox2
  (Q6P0E1), X. laevis sox3-a (P55863) and X. laevis sox2 (O42569). SoxB1 proteins carry no crest
  process term (positive or negative) in any species; this is consistent with the reading above, not
  a gap. By contrast mammalian SOX2 carries stem-cell maintenance terms: human GO:0035019 (IDA/IMP),
  GO:0097150 (ISS); mouse GO:0019827 (IMP x3), GO:0097150 (IGI). Zebrafish sox2 and frog sox2/sox3
  carry none.
• Network placement. Sox2 is a neural plate / neural progenitor gene and a blastula pluripotency
  factor, not a member of the module's progenitor_competence_maintenance part (which holds
  Myc, Id3, Hairy2: factors that stay on in border/crest progenitors and whose loss causes progenitor
  arrest). If the module represents SoxB1 at all, it should be as an upstream blastula competence state
  (Sox2/Sox3 as a set) whose hand-off to SoxE precedes crest specification, or as a knowledge-gap note
  on the SoxB1 to SoxE transition — not as an annoton with a crest process term.

Evolution

• The SoxB1 to SoxE switch is conserved in lamprey: soxB1 orthologs are expressed in blastula, gastrula
  ectoderm and border, then downregulated in premigratory crest as soxE comes on PMID:39060477. So the hand-off dates
  to the vertebrate ancestor.
• SoxB1 neural roles are pan-bilaterian (the IBA neuron differentiation node includes Drosophila
  SoxNeuro/Dichaete and C. elegans sox-2 donors). The vertebrate novelty is not a new SoxB1 activity
  but the recruitment of SoxE into blastula-like potency; SoxB1 is the ancestral state being handed
  off PMID:30144418.

Decisions summary

• 25 GOA rows: MF activator/DNA binding/nucleus accepted; IPI protein binding -> GO:0140297; IMP
  nervous system development accepted (core, neural induction); IBA neuron differentiation accepted;
  brain development, repression, cytoplasm, cell differentiation and ARBA anatomical term non-core.
• NEW: GO:0045665 (IMP PMID:19736324), GO:0060041 (IMP PMID:15820691).
• No crest term. Questions: SoxB1 blastula pluripotency term; Snail2 -> Sox2 repression evidence in
  frog; paralog attribution of dnSox2 and combined-MO results.