POLD2 (donkey, A0A9L0IND0) notes

2026-10-14 DONKEY40 review

• Record: TrEMBL, 487 aa, "DNA polymerase delta subunit 2" (name from ARBA), Ensembl gene model from the donkey genome project (PMID:33293529, genome sequencing only; no functional experiment on donkey POLD2). All 14 GOA rows are electronic (InterPro2GO, ARBA, TreeGrafter GO_REF:0000118, Ensembl Compara GO_REF:0000107 from human P49005, combined GO_REF:0000120, SubCell GO_REF:0000044).
• Sequence: 95.52% identity to human POLD2 P49005 over the full human length; donkey has an 18-aa N-terminal extension only [file:EQUAS/POLD2/POLD2-bioinformatics/RESULTS.md "The donkey sequence A0A9L0IND0 (487 residues) aligns to human P49005 (469 residues) with 95.52% identity among 469 paired residues."]. Function transfer from human is therefore well justified.
• Human POLD2 is the B subunit of Pol delta and shared accessory subunit of Pol zeta:
• [file:EQUAS/POLD2/POLD2-bioinformatics/human-P49005-uniprot.txt "Accessory component of both the DNA polymerase delta complex and the DNA polymerase zeta complex"]
• PMID:31449058
• PMID:24449906
• PMID:22465957
• Pol eta binds POLD2 PMID:25662213
• POLD2 itself has no polymerase active site; it contributes to DNA-directed DNA polymerase activity of the holoenzyme. The PolD2/PolD3 cocrystal shows no obvious DNA-binding surface PMID:24449906 -> InterPro2GO DNA binding is an over-annotation (same call as the human review genes/human/POLD2).
• Repair: Pol delta carries out NER synthesis and BIR [human UniProt "DNA polymerase delta carries out approximately half of nucleotide excision"].
• Enzyme activator activity (Ensembl from human): POLD2/POLD3 increase Pol zeta efficiency, but as integral accessory subunits of the polymerase complex, not as a separate regulator; marked as over-annotation.
• ProtNLM2 GO: DNA replication (LSP, exact term in GOA, less specific than DNA strand elongation involved in DNA replication) and DNA repair (COR; absent from donkey GOA; error-prone TLS is under DNA damage tolerance, not DNA repair, in GO; checked with QuickGO ancestors).

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

• Deep research: just deep-research-falcon EQUAS POLD2 --fallback perplexity-lite failed (falcon: "Failed to authenticate"; perplexity-lite: HTTP 401 insufficient quota). No deep-research file was written. Re-review relied on the cached full-text papers already cited (PMID:24449906, PMID:22465957, PMID:31449058, PMID:25662213) and the reviewed human UniProt record P49005 saved in POLD2-bioinformatics/.
• Identity: A0A9L0IND0 is the full-length donkey POLD2 product, not a fragment or isoform. 487 aa vs 469 aa for human P49005; all 469 human residues are paired (95.52% identity), both domains (OB-fold PF18018, subunit B PF04042) complete. The only difference is an 18-residue N-terminal extension (MGWSIARGYCCLDTSSTA) ahead of the residue aligned to the human initiator Met, most likely an upstream start choice in the Ensembl gene model (POLD2 has no catalytic residues to lose). Donkey has no reviewed Swiss-Prot POLD2 entry, so there is no separate canonical product to review.
• Evolutionary grounding: UniProt places the protein in PANTHER PTHR10416:SF0. That subfamily is the family's only subfamily and also contains archaeal PolD/DNA polymerase II small subunits, so subfamily membership alone does not discriminate. The informative placement is PAINT (interpro/panther/PTHR10416/PTHR10416-paint.tsv): node PTN000045936 (Eukaryota) carries delta DNA polymerase complex (IBD, seeds incl. yeast Pol31, S. pombe Cdc1, fly, mouse, rat, human); node PTN000045935 carries zeta DNA polymerase complex and error-prone translesion synthesis (seeds yeast + human), DNA strand elongation involved in DNA replication (seed S. pombe) and enzyme activator activity (seed human P49005, dated 2026-08-28). The donkey TreeGrafter rows (PTN001311496) descend from these. No PAINT IBD for DNA binding or DNA repair.
• Changes vs 2026-10-14 review:
• GO:0008047 enzyme activator activity: MARK_AS_OVER_ANNOTATED -> KEEP_AS_NON_CORE. A PAINT curator has since placed the term at PTN000045935 from human experimental data (POLD2/POLD3 raise Pol zeta kcat/KM and processivity PMID:24449906). The earlier objection was a modelling preference, not a contradiction, so I defer to the IBD. Note the human review (genes/human/POLD2) has no GO:0008047 row because its GOA snapshot predates the human experimental annotation; that is a curation gap for the human folder, not for this one.
• GO:0042276 error-prone TLS: still ACCEPT; added PAINT node grounding to the reason.
• NEW GO:0006281 DNA repair: kept. Reason expanded: absence from donkey GOA reflects the retired keyword pipeline (the human DNA repair row is GO_REF:0000043), not a PAINT curator decision, since no PAINT seed has an experimental DNA repair annotation. Added the BIR statement from human UniProt as a second quote.
• All other actions re-examined and unchanged (DNA binding over-annotated; nucleus/nucleoplasm/replication/strand elongation/delta and zeta complexes accepted; DNA metabolic process, DNA polymerase complex, DNA biosynthetic process TOO_GENERAL; DNA polymerase binding non-core).
• Relative: human POLD2 (P49005), ortholog in the same PAINT clade; complete review at genes/human/POLD2/POLD2-ai-review.yaml.
• Modules: modules/dmel_dna_polymerase_complexes.yaml (Drosophila Pol delta / Pol zeta, includes PolD2) is the only module covering this complex; no mammalian Pol delta module exists.