Aldh7a1 (mouse, Q9DBF1) notes

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

No ProtNLM2 output was opened or used.

I ran just fetch-gene mouse Aldh7a1, which created the UniProt, GOA (66 rows, collapsed to 57 seeded
annotations) and stub files.
Deep research failed. just deep-research-falcon mouse Aldh7a1 --fallback perplexity-lite returned
falcon "Exception: Failed to authenticate", and the perplexity fallback returned
HTTP 401 insufficient_quota. No deep-research file exists, so this review uses the cached primary
literature below, the UniProt record and my own sequence analysis.
I re-fetched PMID:20207735 (Brocker 2010), which now has full text from PMC (it was abstract-only).
This paper contains mouse experimental data on location.

Lysine catabolism / AASA dehydrogenase (IMP). In constitutive Aldh7a1-KO mice, the metabolites
upstream of the ALDH7A1 step (P6C, α-AASA, pipecolic acid) build up in brain and liver
PMID:32969477. A second KO study confirms
tissue-specific changes in lysine metabolism [PMID:39661514 abstract only, "of tissues from
Aldh7a1-deficient mice, which reveal tissue-specific differences"]. No purified mouse enzyme assay
was found, so the mouse MF rests on substrate accumulation in vivo plus orthology to human.
Location (mouse tissue). Fractionation of mouse liver and kidney found the protein in nuclei,
mitochondria and cytosol but not in microsomes PMID:20207735. Immunohistochemistry of
mouse heart, liver and kidney showed cytosolic and nuclear staining. Mouse has two transcript classes,
one with and one without the 28-aa mitochondrial presequence (isoform 1 = mALDH7A1_v2, mitochondrial;
isoform 2 = mALDH7A1_v1, non-mitochondrial).
Mitochondrion (HDA). Two large-scale mitochondrial proteomic surveys of mouse tissues
(PMID:14651853, PMID:18614015) detected the protein.

Broad aldehyde dehydrogenase activity on betaine aldehyde, lipid-peroxidation aldehydes and
medium-chain aldehydes; homotetramer; osmoprotection. Recombinant human enzyme was used
(PMID:20207735).
Membrane (Golgi, plasma membrane) NADH generation that inhibits COPI fission via BARS and supports
energy homeostasis (PMID:31492851, human HeLa cells). The same mechanism supports FSP1-mediated
ferroptosis suppression (PMID:40233740, human cells, abstract only in the cache).

Mouse is 88.3% identical to human P49419 across the full 539 aa. Catalytic Glu296 and Cys330 are
conserved.
PANTHER 19.0 places mouse in PTHR43521:SF1 along with rat, cow, chicken, Xenopus, fly, worm, plant
ALDH7B4 and bacterial members. Human alone is in SF5. I treat this as a classification artifact.
PAINT node PTN000192435 contains both proteins, and MGI:108186 is itself a seed for the GO:0004043
and GO:0019477 IBDs.
PTN000192435 carries IBDs for mitochondrion, cytosol, nucleus, Golgi membrane, plasma membrane,
GO:0008802, GO:0004043 and GO:0019477. It has no IBD for ferroptosis, energy homeostasis or the
transport-regulation terms. I read that omission as deliberate.
Key finding: human Ser130 (Ser102 in isoform-2 numbering), the AMPK site that controls membrane
recruitment in PMID:31492851/PMID:40233740, is Arg130 in mouse (and in rat; His in Xenopus). The
human stress-triggered recruitment switch therefore does not transfer as such to mouse. I kept
membrane locations as non-core inferences and did not accept the membrane-regulation processes for
mouse.

Core: GO:0004043 (mouse IMP plus orthology), GO:0019477 (mouse IMP), GO:0004029 (orthology),
mitochondrion/mitochondrial matrix, cytosol.
GO:0009063 → MODIFY to GO:0019477. GO:0016491 → MODIFY to GO:0004043/GO:0004029.
GO:0016655 (quinone oxidoreductase) → REMOVE. This is FSP1's activity. ALDH7A1 only supplies NADH
(see the human review notes and PMID:40233740).
Plain transport processes (Golgi to endosome, endosome to lysosome, plasma membrane to endosome) →
REMOVE. In human, ALDH7A1 inhibits these pathways and does not carry them out.
Negative-regulation transport terms, energy homeostasis and ferroptosis → MARK_AS_OVER_ANNOTATED for
mouse. The reasons are: no mouse data, the AMPK site is not conserved, and PAINT did not propagate
these terms.
GO:1905750: QuickGO and the GOA label call it obsolete, but OLS returns it as active. My action does
not depend on which is right.