UniProt: Q9UDR5 (AASS_HUMAN); HGNC:17366; gene ID 10157; chromosome 7q31.3.
926 aa precursor; mitochondrial transit peptide (1..27); mature chain 28..926.
AASS = alpha-aminoadipic semialdehyde synthase, mitochondrial. It is a bifunctional
enzyme (alias LKR/SDH) that catalyzes the first two steps of the main (saccharopine)
pathway of L-lysine degradation in the mitochondrial matrix:
NADPH + H(+)"; PhysiologicalDirection=right-to-left (i.e. lysine-forming direction
is left, so physiological flux is lysine + 2-OG → saccharopine). RHEA:19373; EC 1.5.1.8.
GO term: GO:0047130 saccharopine dehydrogenase (NADP+, L-lysine-forming) activity.
Saccharopine dehydrogenase (SDH, EC 1.5.1.9) — C-terminal domain
(region 477..926). Oxidizes L-saccharopine to (S)-2-amino-6-oxohexanoate
(= L-2-aminoadipate-6-semialdehyde / alpha-aminoadipic semialdehyde) + L-glutamate,
reducing NAD+. UniProt: "L-saccharopine + NAD(+) + H2O = (S)-2-amino-6-oxohexanoate +
L-glutamate + NADH + H(+)"; PhysiologicalDirection=left-to-right. RHEA:24520; EC 1.5.1.9.
GO term: GO:0047131 saccharopine dehydrogenase (NAD+, L-glutamate-forming) activity.
Net: L-lysine + 2-OG + NADPH → saccharopine → alpha-aminoadipate-6-semialdehyde +
L-glutamate. This is the committed entry into lysine catabolism (steps 1/6 and 2/6 of
glutaryl-CoA from L-lysine; UniPathway UPA00868 UER00835/UER00836).
Structure: homotetramer (by similarity to mouse Q99K67). NAD-binding residues in SDH
domain resolved by crystallography (PDB 5L76/5L78/5O1N/5O1O/5O1P for SDH domain 455-926;
8DDA/8E8T/8E8U/8E8V for LKR domain). L-saccharopine-binding residues 577-578, 604, 703,
724-726 (by similarity to Q9P4R4).
Tissue: expressed broadly, highest in liver. Induced by starvation (by similarity).
This section supersedes the older annotation judgments above where they differ.
The 23 seeded annotation term/evidence/reference/qualifier tuples were preserved.
All annotations were reassessed; none was added. The two enzymatic reactions in
lysine degradation remain the two core functions. The human gene review is now
complete, with explicit unresolved evidence rather than unreviewed rows.
The old GO:0004754 rationale and proposed GO:0047131 replacement were incorrect.
QuickGO definitions were retrieved on 2026-09-25:
| Term | Cofactor pair | Products of saccharopine cleavage |
|---|---|---|
| GO:0004754 | NAD+/NADH | L-lysine + 2-oxoglutarate |
| GO:0047130 | NADP+/NADPH | L-lysine + 2-oxoglutarate |
| GO:0047131 | NAD+/NADH | L-glutamate + allysine |
PMID:36128717(https://pubmed.ncbi.nlm.nih.gov/36128717/) directly assays human
full-length AASS expressed in HEK293 cells and isolated LOR domains. Both forward
and reverse reactions were measured, with the forward reaction faster. The
paper explicitly distinguishes human NADPH usage from yeast NADH usage. The
physiological catabolic direction does not invalidate a reversible reaction
whose GO name is written in the opposite direction. GO:0004754 therefore becomes
MODIFY to GO:0047130, the NADP-dependent LOR activity, rather than to the different
C-terminal SDH activity GO:0047131.
The fly source was independently traced in QuickGO: Q9VLX0 has GO:0004754 IDA
from PMID:18695041(https://pubmed.ncbi.nlm.nih.gov/18695041/). Its full text
explicitly describes NADH oxidation by recombinant fly protein with lysine.
That source is not declared erroneous; its cofactor-specific transfer to human
is the problem. The donor paper's background statement that mammalian LKR cannot
run in reverse is superseded by the later direct human measurements. It is not
used as evidence against the human reverse reaction.
The full PMID:18695041 paper demonstrates fly dLKR/SDH histone H3/H4 binding,
EcR-B1 recruitment, inhibition of CARMER-mediated H3R17 methylation, promoter
occupancy, transcriptional repression and dynamic cytosolic/nuclear localization.
The donor genetic and cell assays are genuine. Its LKR domain supplies the histone
interaction; its SDH region supplies receptor interaction. Repression is
separable from metabolic catalysis. This is not histone demethylase activity.
The paper reports 51% sequence identity and 71% similarity to mouse and human
LKR/SDH. Human AASS retains both domains, so conservation is plausible. However,
the interaction surfaces were not mapped onto human AASS, the tested recruitment
context is EcR/Usp, and the donor's targeting mechanism is incompletely defined
(no nuclear localization signal was identified; partner-dependent shuttling was
suggested). These are specific unresolved transfer questions, not proof of a
lineage-specific loss. Human mitochondrial targeting does not exclude a conditional
nuclear or cytosolic pool. The old claims that these activities are necessarily
fly-specific or incompatible with mitochondria are withdrawn.
Accordingly, the five ISS annotations for transcriptional repression,
corepressor activity, histone binding, cytosol and nucleus are UNDECIDED with
UNRESOLVED propagation/source status. They are not removed or labeled proven
functional divergence. Their source Q9VLX0 and primary evidence are documented
separately from the human uncertainty.
cache_reactome_pathway function. Both human reaction summaries explicitlyThe full PMID:34800366 manuscript and its publisher supplement were inspected.
The workbook was retrieved from:
https://ars.els-cdn.com/content/image/1-s2.0-S1550413121005295-mmc2.xlsx
This is Supplementary Table S1 (linked as mmc2.xlsx by the PMC article XML).
Downloaded workbook SHA-256:
10690847c50567e11055d474fbe265fe5d699384f38437acd95734f0a87a9183.
It was opened with openpyxl.load_workbook(..., read_only=True, data_only=True);
row numbers below are one-based Excel rows, with headings in row 2.
| Sheet | Row | Protein Group ID(s) | Simplified protein IDs | Gene name | MitoCoP |
|---|---|---|---|---|---|
| (A) All protein groups | 5 | 11286 | Q9UDR5 | AASS | 1 |
| (B) MitoCoP (1,134 genes) | 5 | 11286 | Q9UDR5 | AASS | 1 |
Sheet (A) additionally has Identified in this dataset = 1, Ensembl gene
ENSG00000008311 and NCBI gene 10157. Sheet (B) classifies AASS under amino acid
metabolism. This verifies the specific human HTP target assignment, rather than
using the abstract's description of the overall dataset as if it named AASS.
The dataset establishes mitochondrial association; it does not by itself prove
matrix subcompartment or absence of any other pool. The unmodified 13.9 MB
workbook was inspected locally and is reproducibly accessible from its primary
source; it is not duplicated into the repository.
The annotation-reviewer and core-function-synthesizer procedures were applied.
Deep research and original-publication caching were launched concurrently. A fresh
virtual-environment installation race stopped the first research wrapper before
provider startup; once installation completed, Falcon was started with a
1200-second timeout. Perplexity was not retried because the project had already
observed HTTP 401 insufficient_quota. Provider output, if returned, is preserved
as generated and assessed separately from primary-source verification.
The repository-wide baseline had passed 4,975 gene reviews in the coordinator's
checkout. Targeted validation and a semantic comparison of the 23 original
annotation tuples are performed in this gene branch; shared project artifacts are
owned by the coordinator and are not changed here.
Falcon completed successfully in 546.14 seconds (provider metadata), produced its genuine report and artifact, and did not require fallback. Both files are retained unchanged. The report was read in full; its principal chemistry and animal-versus-human disease distinctions corroborate the independently checked studies. It is marked UNVERIFIED as an evidence source: its unnamed-journal Kopec 2017 report was not resolved to a primary publication; its additional 2000 full-text numerical claims and clinical-trial absence were not independently adopted. It also calls AASA the potentially channeled intermediate between the two AASS active sites, whereas that intermediate is saccharopine. Engineered targeting in worm rescue is not used to prove endogenous human localization. The report is not inserted into annotation supported_by fields merely to suppress the validator advisory.
The independent review requested donor targeting and fractionation checks. The live
UniProt Q9VLX0 record, retrieved
2026-09-26, carries Mitochondrion (ARBA00004173) and the Transit peptide
keyword (ARBA00022946). It has no experimentally mapped transit-peptide cleavage
feature. These are electronic assertions, not direct fly import experiments;
critically, absence of a mapped feature is not absence of mitochondrial targeting.
PMID:18695041 Fig. 1F and the corresponding full-text paragraph compare cytoplasmic
and nuclear fractions, with DRONC/DRICE and HP1-beta controls. They do not separately
probe a mitochondrial fraction or mitochondrial marker. Tagged localization and
promoter recruitment provide additional positive source evidence. Neither that
experiment nor the current record supports treating fly localization as exclusively
cytosolic/nuclear or inferring a targeting divergence from human AASS. The putative
export signal and proposed partner-mediated nuclear recruitment remain mechanistic
questions; mitochondrial targeting does not exclude an additional pool.
The five ISS decisions remain UNDECIDED after these concrete checks. Their source
comments now distinguish receptor-dependent repression, SDH receptor contact,
cytoplasmic fractionation, nuclear recruitment, and LKR histone binding. Conserved
catalytic domains make a transfer plausible but do not establish non-catalytic
contact conservation; absence of a direct human assay is likewise not evidence of
loss. The ActionEnum first permits uncertainty about an annotation, then requires
UNDECIDED for inaccessible publications. The latter is a sufficient reason, not its
only permitted use. source_status: UNRESOLVED records an unclassified transfer,
whereas SUPPORTS_SOURCE_BUT_NOT_TARGET would assert propagation is unsafe. These
are completed evidence assessments with explicit remaining biological questions,
not unreviewed PENDING stubs.
The PMID:463877 mitochondrial EXP row is now ACCEPT by deference to the original
curator and independent human MitoCoP/matrix evidence. This supersedes the earlier
access-based abstention above; the full 1979 localization methods remain unavailable
and are not claimed to have been verified. The same distinction is used for its
biochemical annotations, independently supported by later human work.
The fly NADH assay shows that NADH works under the tested conditions, not that NADPH
cannot work. Human NADPH dependence is independently grounded in PMID:36128717,
including its discussion of mitochondrial NADP deficiency. GO:0004754 MODIFY is an
instruction to reconcile/retire that redundant row into the existing GO:0047130
annotation, not add a duplicate. QuickGO's live children endpoint for GO:0004753
was rechecked: both GO:0047130 and GO:0047131 are is_a children.