GPC3 (human, P51654) notes

2026-10-21 — Full re-review, independent of the ProtNLM2 predictions

Context: human GPC3 is the experimentally characterised ortholog (same PANTHER
subfamily PTHR10822:SF4, GLYPICAN-3) of HORSE/GPC3 (A0A3Q2L6D3), and a seed of the
PAINT nodes PTN002585559 (plasma membrane) and PTN000850517 (positive regulation of
canonical Wnt signaling). No ProtNLM2 files were opened.

Deep research

just deep-research-falcon human GPC3 --fallback perplexity-lite failed: Edison
API returned 403 on auth (Failed to authenticate), and the perplexity-lite fallback
also failed. No deep-research file was written. The review is built from cached
publications (full text for PMID:14610063, PMID:39822733, PMID:25653284; abstracts
for the rest) and the UniProt record.

Evidence summary, by species

Established experimentally for the HUMAN protein:
- Cell-surface, GPI-anchored, convertase-processed alpha/beta heterodimer
PMID:14610063
- Binds Wnts and stimulates canonical Wnt signalling in HCC cells
PMID:16024626
- Binds Frizzled via HS chains; Wnt-Fz-GPC3 complex is co-endocytosed
PMID:24496449
- In CHO cells represses canonical and non-canonical Wnt reporters; binds Wnt5A
PMID:14610063
- Binds DPP4/CD26 and inhibits its peptidase activity (single abstract-only study)
PMID:17549790
- Binds AFP directly PMID:39822733
- Shed/serum GPC3 in HCC PMID:12851874

Established in MOUSE (ortholog, same PANTHER subfamily) and transferred by ISS:
- Hedgehog binding, competition with Patched, Hh endocytosis/degradation, negative
regulation of Hh signalling PMID:18477453
(convertase processing requirement: PMID:25653284, which UniProt attaches to the human entry).
- Renal branching/proliferation and coronary development phenotypes of Gpc3-null mice
(UniProt By similarity; primary papers not cached).

Zebrafish: morphant gastrulation movements (epiboly, convergent extension), rescued only
by processable GPC3 PMID:14610063.

PAINT

PTHR10822 PAINT file: GPC3 lineage gets PTN002585559 (PM), PTN000850517 (pos. reg.
canonical Wnt; seeds dally + human GPC3), and deep node PTN002247069 (cell surface,
extracellular region, cell migration, regulation of protein localization to membrane).
No hedgehog-regulation IBD is placed on the GPC3 lineage; I treat hedgehog antagonism
as ortholog-based (ISS), not phylogenetically inferred. The negative/cell-type specific
Wnt effects were not propagated, consistent with stimulation being the conserved trait.

Decisions worth flagging

• GO:0060422 peptidyl-dipeptidase inhibitor activity is mis-specified for DPP4 (an
  N-terminal dipeptidyl-peptidase, GO:0008239); MODIFY to GO:0030414 peptidase
  inhibitor activity. Suggested question raised about a missing GO term.
• Developmental knockout/morphant phenotypes (coronary vasculature, gastrulation
  cell migration, anatomical structure morphogenesis) marked over-annotated; kidney
  growth-factor modulation rows kept as non-core.
• protein binding: DPP4 row -> protease binding; SMYLE-complex row and AFP row removed
  (uninformative; PMID:29162697 abstract does not mention GPC3).
• NEW: frizzled binding (IDA, human), Wnt-protein binding (IDA, human), hedgehog family
  protein binding (ISS, mouse-based).