FTSH12 (Q9SAJ3, At1g79560) notes

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

Scope: reviewed Swiss-Prot canonical product Q9SAJ3 (1008 aa) of At1g79560, PANTHER
PTHR43655:SF19. Folder name FTSH12__Q9SAJ3 because genes/ARATH/FTSH12 is keyed to
the alternative gene model A0A1P8ARD2. No ProtNLM2 output was opened or used.

Deep research: just deep-research-falcon ARATH FTSH12__Q9SAJ3 --fallback perplexity-lite
failed (Edison 403 authentication failure; Perplexity 401 insufficient quota). No
deep-research file was produced. Literature was read directly from cached publications;
the existing falcon report for the same locus (genes/ARATH/FTSH12/FTSH12-deep-research-falcon.md)
was consulted only as a pointer to papers, and every claim was checked against the papers.

Experimentally established in Arabidopsis

• Subunit of the ~2-MDa Ycf2/FtsHi complex (BN-PAGE, complex purification)
  PMID:30309901;
  confirmed by native-promoter Myc-FtsH12 IP PMID:33216923.
• ATP-dependent association with inner-envelope translocation intermediates
  (pFd, pSSU, pL11, pLHCP; not OEP7) PMID:30309901.
• Inner envelope localization, C-terminal AAA+/protease domains facing stroma; absent
  from thylakoids PMID:33216923.
• H769Y zinc-site mutant complements embryo-lethal knockout: protease motif dispensable
  PMID:30309901.
• Knockout embryo lethal (heart stage); knockdown pale cotyledons, deformed chloroplasts,
  altered TIC composition; N-terminome shows normal processing of imported proteins PMID:33216923.
• Cryo-EM: heterohexameric AAA+ motor module (PDB 8XKU/8XKV include FTSH12) [PMID:39197452, abstract only].

Inferred only (sequence)

• Walker A GPPGTGKT (533-540), Walker B FVFVDE (587-592), HEAGH (769-773) + D849
  (FTSH12__Q9SAJ3-bioinformatics/RESULTS.md). ATPase and peptidase activity of FTSH12
  itself have never been measured. Kikuchi 2018 notes FTSH12 lacks the conserved
  hydrophobic pore-loop residues and cross-links weakly to preproteins.

Evolutionary grounding

• Kikuchi 2018 phylogeny: FTSH12 and FtsHi derive from the plastid (cyanobacterial)
  FtsH; FTSH12 is land-plant specific (likely absent in Chlorophyta). The IBA m-AAA
  complex / mitochondrial protein processing rows come from PTN000554085 (mitochondrial
  m-AAA node, eukaryote taxon) and were removed as compartment mismatch. The
  metalloendopeptidase IBA (PTN008683692) was kept as non-core because the catalytic
  site is retained, but no activity is demonstrated.

Decisions

• Core: contributes to GO:0016464 chloroplast protein-transporting ATPase activity;
  GO:0045037; part_of GO:0062091; located in GO:0009706.
• Removed: plasma membrane (ISM), thylakoid membrane (contradicted; UniProt source defect
  recorded), m-AAA complex and mitochondrial protein processing (IBA), bare protein binding x6.
• Mitochondrion HDA marked over-annotated (no FTSH12-specific evidence; likely co-fractionation).
• General plastid/chloroplast/envelope/cytoplasm rows MODIFY -> chloroplast inner membrane.
• For transfer to A0A1P8ARD2: only complex membership, inner-envelope location and import-motor
  role are experimentally established; peptidase and ATP hydrolysis are sequence inferences.

2026-10-21: PR #5308 review follow-up

• The protease rows (GO:0004176 IEA and TAIR ISS, GO:0004222 PAINT IBA and IEA, GO:0006508 IEA) changed from KEEP_AS_NON_CORE to MARK_AS_OVER_ANNOTATED, so that they match the A0A1P8ARD2 gene-model review (genes/ARATH/FTSH12). The reasons are the same in both files. The HEAGH zinc site and Asp849 are intact, but the activity has never been measured, H769Y complements the null, the pore-loop residues are missing and no substrate is known. The rows are kept, not removed, because a minor or conditional activity cannot be excluded. The IBA row now has a propagation_review (root cause UNRESOLVED, failure mode FUNCTIONAL_DIVERGENCE, source PTN008683692).
• GO:0016464 chloroplast protein-transporting ATPase activity (TAIR IDA, PMID:30309901) stays ACCEPT as the core activity. The reason now separates the term from the qualifier. The evidence is at the level of the complex, so contributes_to is the better relation, and core_functions uses it. Following the EIF2B2 precedent, the imported enables is kept and a WRONG_QUALIFIER source defect (TAIR, CHECK, PROBABLE) is recorded rather than changed silently.