smad3a notes (Danio rerio, SMAD family member 3a; UniProt Q8AY15)

2026-09-28 — session log (DANRE_DUPLICATION batch 3, random sample)

Deep research: not available for this gene (Edison/Falcon returned 402 Payment Required;
the OpenAI key is invalid). Not attempted, per instructions. No -deep-research-*.md file
exists. The literature search below was done by hand (Europe PMC queries "smad3a AND
zebrafish", "smad3b AND zebrafish", "(smad3a OR smad3b) AND (mutant OR morpholino OR
knockout OR crispr) AND zebrafish"; relevant papers cached with fetch-pmid).

Accession: Q8AY15 (TrEMBL, 425 aa) holds all 31 GOA rows, including the ZFIN experimental rows.
Paralog: smad3b (Q8AY16). Human ortholog SMAD3 (P84022).

TGD origin

• PANTHER: TGD_tree, Neopterygii|Teleostei, 1:1, PTHR13703 (SMAD); one gar co-ortholog, two
  medaka co-orthologs (projects/DANRE_DUPLICATION/panther_tgd_pairs.tsv).
• Synteny/phylogeny in several teleosts PMID:27703851
• PMID:27703851
• Note that the synteny paper considers smad3b the copy in the ancestral neighbourhood
  PMID:27703851
  (In a WGD both copies are equally "ancestral"; this is about which one kept more of the flanking gene order.)

Protein (my analysis: smad3a-bioinformatics/RESULTS.md)

• smad3a vs smad3b 94.1%; smad3a vs human SMAD3 96.9%; smad3a vs gar SMAD3 (W5N932) 96.3%.
• All annotated functional residues of human SMAD3 conserved in smad3a (Zn site, K40/K41,
  linker phosphosites, K378, SSVS C-terminus). smad3a is the conservative copy; smad3b carries
  most linker changes and S418N.
• Longer smad3b branch in teleosts PMID:27703851
• Human-zebrafish conservation used to model human variants
  PMID:36926042

Expression

• Maternal and ubiquitous early PMID:28687631
• Stronger in new somites PMID:21159776
• Circadian, unlike smad3b PMID:29940038
• Clock1a controls the smad3a promoter PMID:28687631
• Upregulated in regenerating heart PMID:33816481
• Bgee: highest in presomitic/paraxial mesoderm, somite, caudal fin, retina (RESULTS.md).
• Dick et al. 2000 cloned "smad3" before smad3b was described
  PMID:10767528. The
  abstract does not use the a/b names; I infer (not verified) that this is smad3a, since
  smad3b was described as novel in 2002 (PMID:12112463).

Function

• Dominant-negative forms of each paralog (and Smad2) block mesendoderm induction
  PMID:18025082
  PMID:18025082
• smad3a, not smad3b, activates the myf5 promoter when overexpressed; ChIP shows Smad3a on myf5 SBEs
  PMID:21159776
  PMID:21159776
• smad3a mRNA rescues clock1a morphant mesoderm/scl markers
  PMID:28687631
• Constitutively active Smad3a and Smad3b both raise cardiomyocyte proliferation (70% vs 31%)
  PMID:29196619
• Smad3 reporter responds to both; both knockdowns alter neural differentiation
  PMID:25286120
  PMID:25286120
• Double knockout needed for an aortic phenotype; DKO viable
  PMID:42584512
  PMID:42584512
  PMID:40066353
  Single-mutant phenotypes are not given in the cached text (Vanhooydonck 2026 is abstract only).

GOA experimental rows: notes

• PMID:28887217 (Eaf1/2) is abstract-only. Two IMP rows carry the eaf1 morpholino
  (ZDB-MRPHLNO-090820-4 resolves to "MO1-eaf1" on ZFIN) in WITH. Could not see which smad3a
  manipulation was scored. Deferred to the curator for TGF-beta signalling; cell-fate terms kept as
  non-core.
• PMID:16890162 (Tob1a) abstract: PMID:16890162
  Generic protein binding: removed as uninformative (the interaction is not disputed).
• BMP signaling pathway (ARBA IEA): SMAD3 is a TGF-beta/activin/nodal R-SMAD; marked over-annotated.