A0A8C9H4D2 (OLFML2A, Piliocolobus tephrosceles) notes

2026-10-20 — Full re-review, independent of the ProtNLM2 predictions

No ProtNLM2 file or prediction-evidence folder was opened for this review.

Identity: short isoform, not the canonical product

• A0A8C9H4D2 is 439 aa (RefSeq XP_023072933.1, "isoform X2"; Ensembl ENSPTEP00000013988.1, gene ENSPTEG00000015637.1). PE 4, unreviewed.
• The same Ensembl gene also encodes A0A8C9H9A8 (655 aa, XP_023072932.1, "isoform X1", ENSPTEP00000013996.1). A0A8C9H9A8 is the full-length product, with a SignalP signal peptide (1..27) and the olfactomedin domain at 397..655 [file:9PRIM/A0A8C9H4D2/A0A8C9H4D2-bioinformatics/A0A8C9H9A8-uniprot.txt].
• Alignment (A0A8C9H4D2-bioinformatics/RESULTS.md): A0A8C9H4D2 9-439 matches A0A8C9H9A8 225-655 at 99.8% identity. A0A8C9H4D2 1-439 matches human isoform 3 (Q68BL7-3, residues 1-214 missing, 215-223 ATGTGSKAQ -> MSKRDKAGK) at 97.3% over its full length. A0A8C9H4D2 is therefore the counterpart of the human alternative-first-exon isoform 3.
• What is missing compared with the canonical product: the signal peptide; both tandem CXCXCX9C motifs (8 Cys, human positions 47-100); and the coiled coil (human 157-183). The olfactomedin domain (181..439 here) is complete. The UniProt CAUTION (PRU00446, "Lacks conserved residue(s) required for the propagation of feature annotation") also appears on the record.
• The photomedin paper PMID:15836428 describes the full-length protein as secreted as disulfide-bonded dimers. This short product lacks the N-terminal regions that define photomedins. There is no evidence in any species that the short isoform is translated, secreted, dimerised or binds GAGs.

Evolutionary placement

• PANTHER PTHR23192:SF29 (OLFACTOMEDIN-LIKE PROTEIN 2A), the same subfamily as human Q68BL7 and the full-length Piliocolobus A0A8C9H9A8. It is a 1:1 ortholog of human OLFML2A (the same gene, a different isoform).
• The TreeGrafter signal-transduction IEA cites PTN000573914 in the GOA with/from column. The only GO:0007165 IBD in the family PAINT file (interpro/panther/PTHR23192/PTHR23192-paint.tsv) is at PTN000573635 (taxon 33213, Bilateria), seeded only by MYOC (Q99972) and two MGI genes. I did not check the node-to-node path between PTN000573914 and PTN000573635. PAINT placed GO:0005576 and GO:0007165 at this node. It did not propagate ECM binding or self-association. Human OLFML2A IBA signal transduction was left UNDECIDED in genes/human/OLFML2A.
• The family review interpro/panther/PTHR23192/PTHR23192-review.yaml lists A0A8C9H4D2 under SF29. I agree with that placement; there is no subfamily disagreement. That review does not, however, note that A0A8C9H4D2 is the short isoform without a signal peptide.

Literature

• The only direct experimental study is on mouse photomedin-1 (PMID:15836428, abstract only): it is secreted, forms disulfide-linked dimers and binds chondroitin sulfate E and heparin.
• The cancer cell-line studies summarised in the falcon deep research (AP-1, integrin, HGF/NGF) are perturbation and correlation studies on human OLFML2A. They are not evidence of signal transduction participation. I did not use them for GO.

Decisions

• GO:0005576 extracellular region (IEA, ARBA SubCell): MARK_AS_OVER_ANNOTATED. The term is correct for the gene's full-length product, but this accession has no signal peptide.
• GO:0007165 signal transduction (IEA TreeGrafter): UNDECIDED, consistent with the human IBA row. The PAINT node placement is a deliberate judgment that I do not have evidence to overturn. There is no OLFML2A-specific evidence for it, and this short product lacks the secretory entry needed for extracellular signalling.
• Previous core function GO:0005201 extracellular matrix structural constituent: dropped as unsupported. Even the full-length protein has only binding evidence.
• Relative to review next: A0A8C9H9A8, the full-length product of the same gene (same_gene_canonical; no review in the repo). The human OLFML2A review (Q68BL7) is already complete.

2026-10-21 — Characterised comparator (PR #5304 review)

The same-gene relative reviewed in this batch now lives in genes/9PRIM/OLFML2A__A0A8C9H9A8/ (renamed from the bare-symbol
folder so it does not shadow this accession's folder). That relative has no experimental data of its own, so the
experimentally characterised comparator for this gene is human OLFML2A, already reviewed in genes/human/OLFML2A/. It sits in the same
PANTHER subfamily (PTHR23192:SF29) as this target. Its review is still IN_PROGRESS on main, so its experimental findings should be checked there before any transfer.

Correction (2026-10-21): the earlier statement in these notes that the human OLFML2A review "is already complete" is superseded; that review is IN_PROGRESS on main.