FZD4 is a seven-pass receptor with two complementary ligand-response roles. Wnt ligands engage pathways whose outputs depend on the ligand, coreceptor and cell system. Norrin is a distinct, non-Wnt ligand that signals through FZD4 with an LRP coreceptor. In retinal endothelium, this signal-reception step contributes to vascular growth and organization; endothelial-specific mouse experiments connect the receptor mechanism to human FEVR-associated signaling defects (PMID:19837032, PMID:12172548, PMID:15035989).
The receptor's intracellular interfaces have different levels of experimental resolution. NHERF1 binding depends on the receptor terminal motif and NHERF1 PDZ2 (PMID:20802536). Positive isolated-domain measurements also show recognition of the human FZD4 terminal peptide by several PDZ domains (PMID:36115835); these measurements do not establish endogenous complexes of every full-length protein pair. Dvl2 recruitment and co-immunoprecipitation in the NHERF1 study are positive association evidence, but do not themselves map the Dvl2 interface. Norrin binding is represented by a ligand-specific term rather than generic protein binding.
FZD4-dependent reporter activation is evidence of positive regulation of transcription through signaling; it does not require the receptor to bind DNA. Likewise, increased FZD4 expression after ATRA in human NT2 cells is an expression response within the definition of cellular response to retinoic acid (PMID:12429992). It is not evidence that FZD4 binds retinoic acid or executes neuronal differentiation. The two WNT7B-derived ISS records need a separate upstream source check.
Evidence from brain and retina is not interchangeable. Selected cerebellar experiments establish endothelial barrier integrity and vascular stability, whereas the retinal experiments directly examine vascular development. The unrecovered older cerebellar morphogenesis experiment remains an explicit uncertainty. Native protein products are also distinct from extracellular domains, short peptides, disease variants and engineered structural constructs.
See the source guide and compact source evidence for source locators, construct limits and preserved original contexts. This follow-up preserves all original annotation objects and source bytes. Local validation passed with eight advisories: seven concern retained generic protein-binding rows, and one records the source-specific beta40/beta42 distinction. Strict GOA validation matches all 134 original contexts, and status remains DRAFT. The native EDIT history and private render/local-link and scoped whitespace checks passed. All 80 source files and the original CREATE history remain unchanged.
# FZD4 evidence notes
## Private integration, 2026-11-01
The proposed review combines 134 individually assessed original annotations: 50 source and process contexts reviewed by ROOT and 84 interaction, location and pathway-model contexts reviewed by W. Both source packets remain immutable. Concise summaries were added to the latter rows; their actions and reasons are retained except removal of an irrelevant HTT-specific clause from the four non-HTT Haenig interactions. The two complementary PMID:25424568 assessments are combined, preserving both neuronal morphogenesis support and unresolved dendrite-localization access. The method assessment for phylogenetic inference records all six verified target-lineage placements.
The synthesis separates Wnt receptor activity and Norrin receptor signaling. Experimental fragments, disease variants, structural constructs, donor species and whole-receptor assays remain distinct. The official record identifies one 537-residue canonical precursor and no declared alternative natural products. Two short excerpts supporting the core functions have been verified in retained source bytes; genuine writer-cache validation remains pending.
See the [source guide](FZD4-source-guide.md) for experimental locators and limits and the [evidence archive](FZD4-review-evidence.json) for all original contexts, sequence, source provenance and literal extension groups. Current judgments, reference assessments and core functions are held in the proposed review rather than repeated in the compact archive.
All 55 original reference IDs are covered within 60 manual reference assessments. Four additional publications have bounded source qualification, bringing the prospective normal-source scope to 80 paths. This is private preparation: no canonical gene, publication cache, family file or native seed has been written. The offline forecast retains all 134 rows without collisions; actual native mappings and future validation outcomes are not claimed.
## 2026-11-01 — genuine source acquisition and native reconciliation
All 134 original contexts now reconcile bijectively to the genuine 16-column export and native seed. The canonical 537-residue sequence is unchanged, and no alternative natural products are declared. The 80 normal sources include 47 unchanged main files and 33 newly acquired files. The six original extension atoms in five contexts are retained with their literal relation and group structure; the native export itself does not encode those atoms. Official display-name enrichment is recorded separately without rewriting the original annotation objects.
All approved judgments, both receptor cores and the 60 manual reference assessments retain their scientific content. The PMID:12429992 acquisition-stage sentence now records the actual source availability. The two approved excerpts match exact normal-cache bytes, and question contexts are preserved within the supported question text. The compact evidence record contains original contexts and source provenance rather than duplicate current judgments. Native history and scoped checks follow after application.
## 2026-11-01 — Local validation
Local validation passed with 24 advisories: 23 concern retained positive generic protein-binding annotations under the standing curation policy; one notes differing amyloid-beta-binding decisions whose reasons distinguish the actually resolved beta40 context from unresolved beta42 evidence. Strict GOA matched all 134 original contexts, and the computed status is DRAFT. Native CREATE history, private HTML/local links, scoped whitespace and exact source/context conservation passed. Source bytes and the accepted biological judgments remain unchanged. Scientific PR review and actual merge-group CI are still required before acceptance.
The earlier expression-response paragraph records the prior interpretation. The original GO:0071300/ISS/GO_REF:0000024 annotation is now UNDECIDED because its WNT7B/P56706 transfer remains unresolved and the reported increase in FZD4 transcript after ATRA does not adjudicate the product role asserted by this row. PMID:12429992 is a review abstract; its positive NT2-cell expression observation and exact excerpt remain retained. This is an evidence-specific uncertainty judgment, not a general exclusion of expression changes from response-to terms. WNT7B co-mention is not proof of the source error mechanism, and the existing WRONG_WITH_FROM/CHECK record retains cause UNKNOWN. All other 133 reviews, both cores, the original contexts, eight excerpts and 80 source files are unchanged.
Local validation passed with eight advisories: seven retained generic-binding rows and one source-specific amyloid-beta distinction. Strict GOA validation matches all 134 original contexts, and status remains DRAFT. Native EDIT history, private rendering/local links, source/context conservation and scoped whitespace checks passed. All 80 source files, the compact original-context archive and both earlier history records remain unchanged.
Aligned the neuronal-differentiation reason and PMID:12429992 reference assessment with the existing UNDECIDED retinoic-acid annotation: the review article reports positive transcript induction, but this does not resolve the particular ISS/product assertion. Removed the generic policy sentence from that annotation reason; the scope discussion remains above. All annotation actions, original contexts, source-defect records, eight quotations, both cores and cached sources are unchanged.