AIGR Deep Research Report — sox10 (XENLA, Q8AXX8)

Hypothesis slug: nc-enhancer-inputs-scatac
Seed hypothesis: The neural crest specifiers sox10, snai2 and foxd3 receive direct regulatory input from the neural plate border (NPB) factors Pax3/Pax7, Zic and TFAP2 at neural crest–specific enhancers. Test by checking whether crest-accessible enhancers linked to Sox10, Snai2 and FoxD3 in the Williams et al. 2019 chick NC ATAC/scATAC dataset (PMID:31639368) are enriched for PAX, ZIC and TFAP2 motifs vs. background.

Executive Judgment

Verdict: Partially supported — supported for snai2 and foxd3 (the hypothesis holds strongly for these two specifiers, including in Xenopus); qualified/indirect for the actual review gene sox10.

Using the full text of the hypothesis's own dataset (Williams et al. 2019, PMC6838682), organism-matched Xenopus functional genomics (Plouhinec et al. 2014), and the primary enhancer-dissection literature, the picture is:

• Organism-matched key result (Xenopus, Plouhinec 2014, PMID:24360906): Pax3+Zic1 are direct upstream regulators of Snail1/2 (snai2), Foxd3, Twist1 and Tfap2b — but Sox10 is NOT among the identified direct Pax3/Zic1 targets. This both supports the hypothesis for snai2/foxd3 and narrows it for sox10 in the review's own organism.

And at the motif/enhancer level:

• TFAP2 — supported. TFAP2 motifs are a pivotal, enriched component of the canonical NC cis-regulatory signature at NC-specific enhancers (which include the Sox10 super-enhancer and Snai2/FoxD3-class elements).
• PAX — supported as a motif class, but ambiguous as to Pax3/Pax7. A PAX motif (reported as "Pax3") is among the motifs controlling the canonical-NC enhancer cluster (k-Cl3), but the de-novo PAX motif enriched in the broader program maps to Pax2 (a border/neural PAX), and PAX-domain motifs cannot cleanly distinguish Pax3/Pax7 from Pax2.
• ZIC — qualified/weak. The Zic motif appears, but Williams 2019 assigns it to the subgroup shared with the neural-lineage cluster (k-Cl1), i.e. it is not restricted to or dominant at the canonical NC specifier enhancers.
• For Sox10 specifically (the target gene), the "direct input" claim is not supported. The experimentally validated Sox10 NC enhancer (Sox10E2) is directly bound by Sox9, Ets1 and cMyb (Betancur 2010), not by Pax3/7, Zic or TFAP2. NPB factors feed into Sox10 indirectly, chiefly via Sox9/FoxD3.
• The strongest direct NPB→specifier enhancer evidence is for FoxD3, not Sox10: Pax7 binds FoxD3 NC1 and Zic1 directly binds FoxD3 NC2 (Simões-Costa 2012).

Most important caveat for curation: this hypothesis is about upstream regulatory input to sox10, not about the molecular function of the Sox10 gene product. It is GRN context and does not by itself justify adding or changing a molecular-function or cellular-component GO term on sox10.

Evidence Matrix

(Provenance: evidence matrix also saved as /tmp/evidence_matrix.csv; motif passages extracted programmatically from EuropePMC full text PMC6838682.)

Per-gene verdict (seed hypothesis = direct NPB-factor input at NC enhancers)

GO decision table (review gene sox10, Q8AXX8) — leads, require curator verification

(Provenance: /tmp/per_gene_verdict.csv, /tmp/go_decision_table.csv; Sox10 super-enhancer composition extracted from PMC6838682: constituent enhancers mostly k-Cl3 canonical-NC, earliest enh-99 in neural k-Cl1; cranial 10E2 interacts with the Sox10 promoter.)

GO Curation Implications (leads — require curator verification)

Grounding against the actual XENLA sox10 (Q8AXX8) annotation set (QuickGO, 32 annotations): existing terms describe the gene product's own activity and the processes it drives — MF GO:0000978/GO:0000981/GO:0003700 (RNA Pol II cis-regulatory DNA-binding TF activity), GO:0003677 (DNA binding), GO:0019899 (enzyme binding, IPI PMID:16256735); BP GO:0014029 (neural crest formation, IMP PMID:12812785), GO:0001755 (NC cell migration), GO:0030318 (melanocyte differentiation, IMP), GO:0007422 (PNS development), GO:0045893/GO:0000122 (pos/neg regulation of transcription); CC GO:0005634 (nucleus), GO:0005737 (cytoplasm). None encodes "regulated by Pax3/Zic/TFAP2," and none should — GO annotates the gene product's function, not the identity of its upstream transcriptional regulators.

1. Do not convert this hypothesis into a new MF/CC/BP annotation on sox10. It concerns who regulates sox10, which is a property of the regulators (Pax3/Pax7/Zic1/Tfap2a/b), not of the Sox10 gene product. The existing sox10 MF (DNA-binding TF activity) and BP (neural crest formation/migration) terms already capture its role and should be retained as-is; this hypothesis neither adds nor subtracts from them.
2. Treat the regulatory-input statement as non-core / contextual for the sox10 review. If it is to be captured at all, it belongs as a BP "regulation of" annotation on the upstream factors (e.g., Pax7/Zic1 → positive regulation of transcription; neural crest cell fate commitment), with sox10 as the regulated target — not as a function of sox10.
3. If any directional cis-regulatory statement about sox10 is curated, anchor it to Sox9/Ets1/cMyb (Sox10E2; PMID:20139305), not to Pax3/7/Zic/TFAP2. The latter are indirect for sox10.
4. Avoid "protein binding" as an endpoint; the informative relationships here are specific TF→enhancer regulatory inputs, which are better captured as BP regulation terms on the regulators.

Mechanistic Scope

• Immediate molecular event tested: sequence-specific TF occupancy of NC-specific enhancers (PAX/ZIC/TFAP2 motif usage) that activate NC-specifier transcription.
• Direct vs downstream:
• Direct on FoxD3: Pax7 (NC1), Zic1 (NC2) — validated binding.
• Direct on Sox10: Sox9, Ets1, cMyb (Sox10E2) — PAX/ZIC/TFAP2 are not the direct Sox10 enhancer inputs.
• Direct on Snai2 (snai2/Slug): Pax3+Zic1 are direct upstream activators in Xenopus (Plouhinec 2014, translation-blockade direct-target assay); Williams 2019 also identifies NC-specific accessible enhancers near Snai2. Hypothesis supported for snai2.
• Downstream / pleiotropic (not this annotation): craniofacial skeleton, PNS, melanocyte and enteric phenotypes are developmental outcomes of sox10, not the regulatory-input event.

Conflicts and Alternatives

• Sox9 vs the NPB factors. The dominant direct input to the Sox10 enhancer is Sox9 (a SoxE paralog), which itself integrates Pax3/Zic signals. This is the main interpretation that competes with the literal seed hypothesis for sox10: Pax3/7/Zic act through Sox9/FoxD3 rather than directly on Sox10.
• PAX motif identity. De-novo PAX motifs cannot distinguish Pax3/Pax7 from Pax2; Williams 2019 explicitly enriches Pax2 in the neural program. A motif-only enrichment therefore cannot confirm Pax3/Pax7 input specifically.
• ZIC placement. Zic co-localizes with neural/shared (k-Cl1) elements, so a naïve "ZIC enriched at NC enhancers" claim risks capturing a neuroepithelial rather than a bona-fide NC signal.
• Feedforward/autoregulation. The canonical NC code is dominated by Sox10/TFAP2/Fox feedforward loops; much "enhancer input" is NC-specifier self-reinforcement, not NPB input.
• Organism-matched direct-target data reinforce the narrowing. In Xenopus (Plouhinec 2014, PMID:24360906), the Pax3/Zic1 direct targets are Snail1/2, Foxd3, Twist1 and Tfap2b — Sox10 is absent from the direct-target list. So in the review's own organism, sox10 is not a demonstrated direct Pax3/Zic1 output, while snai2 and foxd3 are. This is concordant, not conflicting, across chick and frog.
• TFAP2 is both input and output. Tfap2b is a direct Pax3/Zic1 target (Plouhinec 2014) yet TFAP2 motifs are enriched inputs at canonical NC enhancers (Williams 2019) — i.e. TFAP2 sits in a feedforward loop. A "TFAP2 provides input" statement is true but must not be read as TFAP2 being upstream-most.
• Hierarchy caveat for Sox10. Honoré 2003 (PMID:12885557) shows Xenopus Sox10 knockdown reduces Slug/FoxD3, so Sox10 is an early specifier that feeds forward onto the other two "specifiers" in the hypothesis — they are partly its downstream targets, not co-equal parallel outputs of the border code.

Knowledge Gaps

1. Direct Pax3/Pax7 or Zic binding at a Sox10 enhancer in any species. Checked Betancur 2010 (Sox10E2) and Williams 2019 — not demonstrated; Sox10E2 inputs are Sox9/Ets1/cMyb. Resolving this needs ChIP/footprinting + site-mutagenesis at Sox10 CREs showing PAX/ZIC dependence.
2. Whether the enriched PAX motif at canonical-NC enhancers is Pax3/7 vs Pax2. Motif ambiguity; resolve with factor-specific ChIP-seq/CUT&RUN footprints.
3. Snai2 enhancer gene-specific validation in the chick dataset (direct Pax3/Zic/TFAP2 occupancy at Snai2 CREs). Checked Williams 2019: NC-specific Snai2 enhancers identified, but per-site TF binding not individually dissected in the cited refs.
4. X. laevis–specific cis-regulation of sox10. None of the supplied references test Xenopus sox10 enhancers directly.

Discriminating Tests

1. Factor-specific CUT&RUN/ChIP-seq for Pax3, Pax7, Zic1, Tfap2a in premigratory NC, intersected with the Sox10/Snai2/FoxD3 enhancer coordinates — distinguishes true Pax3/7 occupancy from Pax2 and places Zic at NC vs neural elements.
2. Enhancer reporter + binding-site mutagenesis of Sox10 CREs (Sox10E2 and the Sox10 super-enhancer sub-elements) for PAX/ZIC/TFAP2 sites — tests direct dependence vs Sox9-mediated indirect input.
3. Epistasis / perturbation: Pax3/Zic1 knockdown with rescue by Sox9 — tests whether Pax3/Zic act on Sox10 only through Sox9.
4. Re-run the motif-enrichment on Williams 2019 target-gene-assigned enhancer sets (Capture-C–linked to Sox10, Snai2, FoxD3) specifically, reporting PAX/ZIC/TFAP2 enrichment vs the 99,583-peak background used in the paper — the exact analysis the seed proposes, which the full-text summary indicates would yield strong TFAP2/PAX and weak/neural ZIC signals.

Curation Leads (require curator verification)

• Candidate references to attach to the sox10 GRN context (not as sox10 MF/CC):
• PMID:20139305 — snippet to verify: "Deep characterization of Sox10E2 reveals Sox9, Ets1, and cMyb as direct inputs mediating enhancer activity." Use to document that sox10's direct enhancer inputs are Sox9/Ets1/cMyb.
• PMID:23284303 — snippets: "Pax7 and Msx1/2 cooperate with the neural crest specifier gene, Ets1, to bind to the cranial NC1 regulatory element" and "the neural plate border gene, Zic1, which directly binds to the NC2 enhancer." Use on FoxD3, not sox10.
• PMID:31639368 — motif/combinatorial code (TFAP2s/Sox10/Fox canonical signature; Zic shared with neural cluster).
• PMID:31619682 — conserved SoxE1/Tfap2B NC enhancers.
• PMID:24360906 (organism-matched, Xenopus) — snippet to verify: "the neural border specifiers Pax3 and Zic1 are direct upstream regulators of neural crest specifiers Snail1/2, Foxd3, Twist1, and Tfap2b." Use to support the hypothesis for snai2 and foxd3 and to document that sox10 is not a listed direct Pax3/Zic1 target.
• PMID:12885557 (Honoré 2003) — Xenopus Sox10 is required early and its loss reduces Slug/FoxD3 (hierarchy context).
• Candidate action: Record the seed hypothesis as contextual/non-core for the sox10 review. Do not add an MF or CC term from it. Any directional regulation term belongs on the regulators (Pax3/Pax7/Zic1/Tfap2a → regulation of sox10), as a lead.
• Suggested curator question: Is the intent to annotate sox10's function or to document its upstream regulators? If the former, this hypothesis is out of scope; if the latter, annotate the regulators and keep sox10 as the target.
• Suggested experiment (if local bioinformatics is to adjudicate): reproduce the Williams 2019 motif enrichment restricted to Capture-C-linked Sox10/Snai2/FoxD3 enhancers vs the paper's background peak set.

Notes on Reproducibility / Access

• Primary motif-enrichment numbers were read from the Williams 2019 full text (EuropePMC PMC6838682); the raw scATAC BED/enhancer-coordinate files were not re-processed here (dataset not provided locally, and the environment's code sandbox has no outbound large-file download). This is stated plainly rather than fabricated. The qualitative enrichment pattern (TFAP2/PAX strong at canonical NC enhancers; ZIC neural/shared; Sox9/Ets at Sox10) is taken directly from the authors' reported de-novo motif analysis and the enhancer-dissection papers.

Artifacts

• OpenScientist final report
• OpenScientist final report