GPATCH11 (human) — review notes

UniProt: Q8N954 (GPT11_HUMAN). Also called CCDC75 and "centromere protein Y / CENP-Y"
(the latter from the 2010 Ohta et al. proteomic screen, not the protein's established role).
HGNC:26768. Member of the GPATCH family of G-patch-domain RNA/spliceosome regulators.

Context for the review

The annotation review is prompted by geneontology/go-annotation#6450, which flags the
IBA is_active_in kinetochore annotation (propagated from the 2010 IDA) as likely incorrect.
The upstream curator notes the 2010 image is low-resolution and that the 2024 Nat Commun
patient study explicitly does not observe kinetochore localization, instead reporting
diffuse nucleoplasmic and centrosomal localization plus a clear pre-mRNA splicing role.
A new (2026) Sci Rep paper on the S. pombe ortholog Sap34 corroborates the splicing role.

Established function (from current literature)

• Pre-mRNA splicing regulator via the G-patch / DEAH-helicase axis. GPATCH11 belongs
  to the GPATCH family, whose hallmark glycine-rich G-patch motif activates DEAH-box RNA
  helicases on spliceosomal/RNP substrates PMID:39572588. The fission yeast ortholog Sap34 binds U2
  snRNP and U4/U6·U5 tri-snRNP components and activates Prp43 via its G-patch domain;
  sap34Δ causes a global splicing defect with increased intron retention
  [PMID:42260140 "Sap34 forms a complex with components of the U2 small nuclear
  ribonucleoprotein (snRNP) and the U4/U6 × U5 tri-snRNP", "deletion of sap34 leads to a
  global reduction in splicing efficiency, predominantly associated with increased intron
  retention"]. Human GPATCH11 patient retina shows dysregulated splicing in proteomics
  and transcriptomics PMID:39572588.

• Subcellular localization: nucleoplasm + centrosome. The 2024 patient study reports
  "a diffuse presence in the nucleoplasm, as well as centrosomal localization, suggesting
  potential functions in RNA and cilia metabolism" PMID:39572588. No kinetochore signal
  is reported. This is consistent with the gene's GPATCH/spliceosome identity.

• Disease association. Biallelic loss-of-function variants (recurrent c.328+1G>T splice
  variant that removes the G-patch domain) cause a syndrome of early-onset retinal
  dystrophy with neurological impairment and skeletal abnormalities
  PMID:39572588. OMIM #621183.

Reassessing the kinetochore annotation

The original IDA (PMID:20813266, Ohta et al. 2010) is part of the MCCP large-scale
mitotic chromosome proteomics screen that "identified approximately 4000 polypeptides in
highly purified chromosomes" and confirmed only a subset by GFP tagging. GPATCH11 was named
"CENP-Y" in that paper as a predicted centromere-associated protein based on the screen.
The upstream curator (go-annotation#6450) explicitly states the image supporting the
kinetochore call is "quite low resolution" and that subsequent high-resolution localization
work in the 2024 patient study reports nucleoplasm + centrosome with no kinetochore signal.

CLAUDE.md cautions against overruling experimental IDA from incomplete evidence, but here:
- the GO curator who maintains GO-annotation is themselves recommending removal,
- the 2010 IDA came from a low-resolution screening assay whose authors flagged ~97 of
4,000 hits as "uncharacterized" candidates needing follow-up,
- the 2024 paper directly contradicts the localization with better microscopy in
multiple cell types and a mouse model,
- the centromere-association nickname "CENP-Y" has not been adopted in any downstream
functional literature.

Action: MARK_AS_OVER_ANNOTATED for both the IDA and the IBA kinetochore terms, with
proposed replacement terms GO:0005654 nucleoplasm and GO:0005813 centrosome. Using
MARK_AS_OVER_ANNOTATED rather than REMOVE because the 2010 proteomic evidence is
real (the protein was detected in purified mitotic chromosome fractions) but does not
support a specific kinetochore role; per the CLAUDE.md spirit, we flag rather than
silently delete an experimental annotation while citing the contradicting evidence.

Other annotations

• GO:0003676 nucleic acid binding (IEA from InterPro G-patch IPR000467) — supported in
  spirit; the G-patch motif is an RNA-binding/helicase-cofactor module. Accept as a
  general molecular function; propose a more specific MF capturing the G-patch role
  (GO:0003724 RNA helicase activity would be wrong — GPATCH11 is not a helicase; better:
  the regulator role via Prp43 binding, GO:0008386 → not applicable). Keep as is; flag
  pre-mRNA binding (GO:0036002) as a more informative successor candidate.

• GO:0005515 protein binding (IPI from PMID:25416956 / IntAct, partners O00560 SDCBP and
  O60242 ADGRB3) — these are Y2H/AP-MS partners with no clear functional relevance to
  splicing or cilia. Mark as KEEP_AS_NON_CORE because the binding events themselves are
  empirically supported but "protein binding" is uninformative per project guidance.

Core functions to propose

1. Pre-mRNA splicing regulator within the spliceosomal U2/tri-snRNP environment, via
   G-patch-dependent activation of DEAH helicase Prp43 (paralog: GPATCH1/2). Supported by
   PMID:42260140 (Sap34 ortholog) and PMID:39572588 (human/mouse phenotype + proteomics).

2. Nucleoplasmic + centrosomal localization, consistent with roles in RNA metabolism
   and primary cilia function (PMID:39572588).

Suggested new GO terms / queries

• GO:0000398 mRNA splicing, via spliceosome (BP) — propose as IBA-promotable when the
  Sap34 paper is curated.
• GO:0005654 nucleoplasm (CC) — IDA-supportable from PMID:39572588.
• GO:0005813 centrosome (CC) — IDA-supportable from PMID:39572588.
• GO:0005686 U2 snRNP / GO:0046540 U4/U6 × U5 tri-snRNP (CC) — supported in fission
  yeast (PMID:42260140); human evidence indirect.

Suggested questions for experts

• Does human GPATCH11 bind a specific DEAH helicase (e.g., DHX15, the human Prp43 ortholog)?
• Does the centrosomal localization reflect a primary cilium role or a moonlighting role?

Upstream issue link

geneontology/go-annotation#6450

2026-09-04 — finishing pass (PAINT no-IBA project)

Deep quality pass over the draft review; validated to zero warnings and promoted to COMPLETE.

Changes made:

• Fixed a non-verbatim quote for PMID:25416956 (Rolland 2014 interactome): the draft
  quoted an invented sentence ("GPATCH11 (Q8N954) is included in the IntAct-curated
  interactions..."); replaced with the verbatim abstract text PMID:25416956 and
  noted that GPATCH11 appears only in the supplementary interaction data.
• Softened the Prp43 claim. The Sap34 abstract does not demonstrate activation; it shows
  a G-patch-dependent interaction PMID:42260140. Description and core_functions
  now say "interacts with ... suggesting regulation of Prp43 activity" rather than "activates".
• IEA GO:0003676 nucleic acid binding: KEEP_AS_NON_CORE → MODIFY, proposed replacement
  GO:0003723 RNA binding. The G-patch module's ligand is RNA; GPATCH proteins carry
  "multiple RNA-binding motifs" PMID:39572588; no evidence for DNA binding. GO:0003723
  is now the core-function MF (replacing the over-broad GO:0003676).
• Added structured propagation_review to the IBA kinetochore entry (root_cause
  SOURCE_STALE_OR_MISSING, failure_mode SOURCE_EVIDENCE_WEAK). Key new finding: the current
  PAINT curation of PTHR21032 (snapshot 2026-02-25, interpro/panther/PTHR21032/PTHR21032-paint.tsv)
  no longer asserts kinetochore at PTN000471210 — it now carries IBDs for nucleoplasm
  (GO:0005654, seed Q8N954), centrosome (GO:0005813, seed Q8N954) and RNA splicing
  (GO:0008380, seed mouse Gpatch11 MGI:1858435), plus visual perception (GO:0007601) at the
  Euteleostomi node PTN002720678 and a taxon-aware negative IRD pruning centrosome for fungi
  (PTN001077735, replaced by GO:0005815 MTOC). So the 2017 IBA kinetochore row in GOA is a
  stale propagation: PAINT has already done what go-annotation#6450 asked for (issue verified
  live on GitHub today — reporter ValWood, recommends replacing the kinetochore annotation
  with the splicing role so it can propagate through PAINT).
• Merged the two core_functions into one (RNA binding / mRNA splicing via spliceosome /
  nucleoplasm + centrosome), since the "centrosome-cilium" second entry had no distinct
  supportable MF; the cilium link stays in the description text and suggested_questions.
• MARK_AS_OVER_ANNOTATED retained (not REMOVE) for both kinetochore annotations, per
  CLAUDE.md guidance on experimental annotations; the contradicting evidence and the
  upstream flag are recorded.

Family review written: interpro/panther/PTHR21032/PTHR21032-review.yaml (see there for
node-by-node adjudication of the 2026-02-25 IBDs).