ATG3 (donkey, A0A9L0IHI0) notes

2026-10-14 DONKEY40 review

• Record: TrEMBL, 342 aa, "Ubiquitin-like-conjugating enzyme ATG3" (ARBA name), Ensembl model from the donkey genome (PMID:33293529; sequencing only). 8 GOA rows, all IEA: 7 TreeGrafter (GO_REF:0000118, PANTHER PTN000305319) and 1 combined (ubiquitin-like protein transferase activity).
• Sequence: [file:EQUAS/ATG3/ATG3-bioinformatics/RESULTS.md "The donkey sequence A0A9L0IHI0 (342 residues) aligns to human Q9NT62 (314 residues) with 98.41% identity among 314 paired residues."] Catalytic Cys-264 conserved (donkey Cys-292); 28-aa insertion after human 222 not seen in human isoforms; activity of this exact model unverified.
• Human ATG3 is the E2 of the Atg8/LC3 lipidation cascade: [file:EQUAS/ATG3/ATG3-bioinformatics/human-Q9NT62-uniprot.txt "E2 conjugating enzyme that catalyzes the covalent conjugation"]; PMID:37252361; cycles between E1 and E3 PMID:24186333.
• ATG12-ATG3 conjugate role in mitochondrial homeostasis [PMID:20723759 title "ATG12 conjugation to ATG3 regulates mitochondrial homeostasis and cell death."].
• GO term structure (QuickGO): GO:0019776 Atg8-family ligase activity is_a GO:0061659 ubiquitin-like protein ligase activity (E3 class); GO:0141046 Atg8-family conjugating enzyme activity is_a GO:0061650 ubiquitin-like protein conjugating enzyme activity (E2 class). Human ATG3 has IDA for GO:0141046. -> TreeGrafter ligase row MODIFY to GO:0141046.
• Selective autophagy terms (mitophagy, nucleophagy, glycophagy): ATG3-mediated LC3 lipidation is part of the core autophagosome-forming machinery used by these processes; kept as non-core (inherited PAINT node assertions; human has IBA for all three).
• ProtNLM2 GO: cytoplasm (GO:0005737). Donkey GOA has cytosol and phagophore assembly site, both part_of cytoplasm -> LSP. No ProtNLM2 function or location text for this accession.

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

• Deep research: just deep-research-falcon EQUAS ATG3 --fallback perplexity-lite failed (falcon "Failed to authenticate"; perplexity 401 insufficient quota). Re-review done from cached literature, UniProt and PANTHER data.
• Identity: A0A9L0IHI0 is an alternative splice product of donkey ATG3, not the canonical-length product. [file:EQUAS/ATG3/ATG3-bioinformatics/RESULTS-rereview.md "Removing the 28-residue segment VVADITKPCWTDFFLPPLLGVNFPPNNP from A0A9L0IHI0 gives exactly A0A9L0J149, the 314-residue donkey ATG3 entry"]. Both are on Ensembl gene ENSEASG00005011413.2 / GeneID 106829515 (RefSeq XP_014694150.1 long vs XP_014694151.1 short). The horse model A0A3Q2L0A4 (342 aa) carries the identical segment, so the extra exon is predicted in both equid genomes. No domains or catalytic residues missing: catalytic Cys-264 -> Cys-292, ATG12-acceptor Lys-243 -> Lys-271. Activity of the long form untested. Neither donkey entry is reviewed in UniProt; A0A9L0J149 is the canonical-length product.
• Evolutionary grounding: donkey, human (Q9NT62), mouse (Q9CPX6) and yeast (P40344) ATG3 all in PANTHER PTHR12866:SF2 (interpro/panther/PTHR12866/PTHR12866-entries.csv; donkey from the UniProt PANTHER xref). TreeGrafter rows come from PTN000305319. No family-review disagreement (the PTHR12866 folder has only entries/metadata, no curated FamilyReview).
• Human ATG3 GOA (QuickGO, 2026-10-21): PAINT IBA is GO:0141046 Atg8-family conjugating enzyme activity, plus IDA (PMID:24186333, PMID:37252361) and IMP (PMID:33446636, PMID:24191030); GO:0019776 on human is only ISS/IEA. So the donkey TreeGrafter GO:0019776 (E3-class ligase) -> MODIFY to GO:0141046 (E2 class) stands. GO:0019787 (general transferase) changed from ACCEPT to MODIFY -> GO:0141046 (correct-but-general rule). Human also has ISS GO:0019777 Atg12 transferase activity; ATG3 is the E2 for its own ATG12 conjugation PMID:20723759, but PAINT did not propagate this to the subfamily, so not added.
• New support: PMID:33446636 (cached this session, full text) PMID:33446636. PMID:12207896 also cached (abstract only), not cited.
• Other calls unchanged: autophagosome assembly, phagophore assembly site, cytosol ACCEPT; mitophagy, nucleophagy, glycophagy KEEP_AS_NON_CORE (inherited family assertions, human has the same IBAs).
• Relative to review next: the same-gene canonical-length product A0A9L0J149 (314 aa, the form matching human ATG3), which has no review; its folder would be genes/EQUAS/ATG3__A0A9L0J149 because genes/EQUAS/ATG3 is keyed to the long splice form. The closest experimentally characterised ortholog, human ATG3 Q9NT62 (same PANTHER subfamily SF2), also has no review in the repo; worm atg-3 (Q9N369) is reviewed.
• Modules mentioning ATG3: modules/metazoan_autophagosome_biogenesis.yaml (ATG3 E2 part, grounded on worm), modules/mitophagy.yaml, modules/phagophore_organelle_contact_site.yaml, modules/dmel_autophagosome_biogenesis.yaml.