MSX1 (human, P28360) review notes

Project context: projects/NEURAL_CREST_ORIGINS.md, Tier 2 (neural plate border
specifiers). Human MSX1 is used as the proxy because there is no reviewed
X. laevis Msx1 entry (frog msx1.L/msx1.S are TrEMBL: A5D8L7, A0A8J0Q853,
A0A1L8HKV3, Q3B8L1, Q642R5).

2026-10-05 — initial review

Identity and molecular activity

• Msh/Msx homeodomain transcription factor (UniProt: Msh homeobox family;
  InterPro HD, IPR050674 Msh_Homeobox_Regulators). UniProt function text is
  mostly "By similarity" to mouse P13297: transcriptional repressor that binds
  C/TAAT cores.
• Mouse Msx1 is a potent repressor, and repression can be independent of its
  own DNA sites, via contacts with the core machinery
  PMID:7823952
  PMID:7823952.
• Msx1 and Msx2 bind the same consensus and both repress; Msx1 is the stronger
  repressor PMID:8861098.
• Target selectivity in vivo via PIAS1 and nuclear-periphery localisation, MyoD
  CER PMID:16600910.
• Context-dependent activation: Msx1 binds Stra8 regulatory sequences and
  stimulates Stra8 transcription in germ cells PMID:22071108.
• Human MSX1 binds DNA in methyl-SELEX (Yin 2017, IDA row for GO:1990837)
  PMID:28473536.
• Human R31P (homeodomain) allele is inactive and acts by haploinsufficiency
  PMID:9742121.
• Partners: Tbx2/3/5 (Cx43 repression in heart cell lines) PMID:18285513;
  p53 (cancer-cell overexpression study) PMID:15705871;
  RBPMS (HuRI Y2H, PMID:32296183 — no functional content).

Mammalian biology (craniofacial, tooth, nail)

• Mouse null: cleft secondary palate, tooth arrest, nasal/frontal/parietal and
  malleus defects PMID:7914451.
  Note: the abstract reports the malleus in the middle ear; the human rows to
  GO:0048839 inner ear development are transferred from a mouse IMP to this
  paper. Mouse P13297 also carries GO:0042474 middle ear morphogenesis. Left
  UNDECIDED (full text not seen) and raised as a question.
• Tooth: Msx1 needed in dental mesenchyme up to cap stage; BMP4 rescues
  PMID:11023873.
  Later, Msx1 keeps dental mesenchyme proliferating and undifferentiated by
  repressing Bmp2/Bmp4/Lef1 PMID:24028588.
• Human genetics: selective tooth agenesis (R31P) PMID:8696335, tooth agenesis
  plus clefting (S105X) [PMID:10742093, title only in cache], Witkop tooth-and-nail
  syndrome (S202X) PMID:11369996;
  cleft association studies [PMID:11332647, PMID:12651933].
• Mouse GO:0001837 epithelial to mesenchymal transition (source of a human IEA)
  is sourced to PMID:11023873, which is about epithelial–mesenchymal
  interactions (tissue recombination), not EMT. Human IEA REMOVED.

Neural crest / neural plate border biology

• Bmp-gradient target at the border; intermediate Bmp gives msx1; msx1 induces
  snail, slug, foxd3; dominant-negative blocks all crest markers; msx1 is upstream
  of snail/slug PMID:14627721
  PMID:14627721.
• Msx1 acts upstream of Pax3 and Zic, mediates FGF8 (not WNT-only) signals
  PMID:15691759
  PMID:15691759.
• Msx1/Msx2 morphants: both required for Slug, mutually compensating
  PMID:16586351.
• Gbx2 is upstream of Pax3 and Msx1 PMID:19736322.
• Msx1+Pax3 is the second-best pair for crest induction from animal caps, after
  Pax3+Zic1 PMID:23509273.
• Mouse Msx1/Msx2 double mutants: crest forms, but cranial/cardiac crest is
  mispatterned and dies PMID:16221730.
  So in mammals the requirement is later (crest patterning/survival), with
  Msx1/Msx2 redundancy; frog evidence is the basis for the border role.

Network layer. Border specifier, one step above Pax3/Zic1: BMP (intermediate
level) → Msx1 → Pax3 + Zic → snail2/foxd3. It is necessary and in neuralized
ectoderm sufficient for early crest genes, but it works through Pax3/Zic rather
than specifying crest itself; Msx1 is also expressed in the epidermal side of the
border and non-crest tissues. Same layer as pax3-a, zic1, gbx2, hes4-a.

Term choice. GO:0014029 neural crest formation (formation of the border
region of ectoderm), as for the other border genes. Not GO:0014036 (fate
specification, reserved for specifiers). Not adding GO:0014034 commitment
either: unlike Pax3+Zic1, which directly activate the specifier enhancers
PMID:24360906, there is no evidence that Msx1 binds crest specifier enhancers;
it acts upstream, through Pax3/Zic.

Comparator checks (QuickGO, 2026-10-05)

1. Downloaded all QuickGO annotations (any evidence, incl. IEA) to GO:0014029,
   GO:0014034, GO:0014036, GO:0014033, GO:0014032, GO:0001755 with
   descendants (goUsage=descendants, is_a/part_of; downloadLimit=100000;
   rows 1912/458/195/45255/43111/40414). Filtered symbols matching
   ^msx|^msh|^hmx: zero hits. Sanity check: pax3.L, ZIC1, zebrafish tfap2a
   present in the same files.
2. Pulled all annotations for Msx proteins directly: human MSX1 P28360 (65),
   MSX2 P35548 (69), mouse Msx1 P13297 (165), Msx2 Q03358 (125); chicken
   MSX1 P28361, MSX2 P28362; zebrafish msxa/b/c/d (Q03357, Q03356, Q01703,
   Q01704), msx1a/msx1b/msx2a/msx2b/msx3 (TrEMBL); X. laevis msx1.L/.S, msx2.L/.S;
   X. tropicalis msx1/msx2. No crest, neural plate or ectoderm term on any of
   them. Xenopus msx entries carry only IBA rows (GO:0000977, 0000981, 0005634,
   0006357, 0048598); X. tropicalis msx1 has GO:0036268 swimming (IMP).
   Zebrafish msxb/msxc/msx1a/msx3 carry otic placode/fin regeneration terms.
   Chicken MSX1 carries GO:0030509 BMP signaling pathway (IDA, PMID:19850029).
3. Same-layer peers (from the project notes, verified by the earlier Tier 2
   reviews): pax3-a, zic1, hes4-a carry GO:0014029 by IMP.

Interpretation: the Msx absence is family-wide and includes the frog genes where
the border work was done (Tribulo 2003, Monsoro-Burq 2005, Khadka 2006). Since
peers in the same role carry the term, this is an uncurated literature gap, not
a convention (the frog msx genes are TrEMBL and were never curated). NEW
GO:0014029 on the human proxy is by ISS from the frog literature; ideally the
frog entries should carry the IMP rows. Raised as a question.

Evolution / outgroups

• Amphioxus Msx is expressed in epidermal ectoderm and extends into the edge of
  the neural plate; BMP shifts it PMID:18562679.
  The border layer (Msx, Zic, Pax3/7, Dlx) is ancestral to chordates; most crest
  specifiers are absent there.
• Lamprey: msx-A expression rises monotonically into late crest, unlike the
  frog dynamics of border factors PMID:39060477.
• So the Msx border role is ancestral; what vertebrates added is the link
  Msx1 → Pax3/Zic → crest specifiers.

Decisions summary

• Core: Pol II-specific repressor (GO:0001227) in nucleus; tooth development
  (GO:0042475, negative regulation of odontoblast differentiation); palate/face
  morphogenesis; neural plate border (GO:0014029, NEW).
• p53 cluster (PMID:15705871, one overexpression study in cancer cells):
  p53 binding, protein stabilization, apoptotic signaling kept as non-core;
  cell morphogenesis, cell growth, protein localization to nucleus and the IC
  DNA-damage row marked over-annotated.
• protein binding (RBPMS, HuRI): REMOVE (uninformative).
• EMT IEA: REMOVE (source paper is about epithelial–mesenchymal interaction).
• inner ear development: UNDECIDED (abstract says middle ear).