AIGR Deep Research — sou1 / SPAC8E11.10 (Q9Y6Z9)

Hypothesis: NADP+-dependent arabitol (polyol) dehydrogenase (GO:0008106 + arabitol metabolic process)

Organism: Schizosaccharomyces pombe 972h- · UniProt: Q9Y6Z9 · Gene: sou1 (SPAC8E11.10)
Focus type: computational_prediction (BioReason-Pro SFT)
Seed reference: doi:10.64898/2026.03.19.712954

Executive Judgment

Verdict: PARTIALLY SUPPORTED at the activity-class level / OVER-ANNOTATED at the substrate level.

The prediction decomposes into three claims that must be judged separately:

1. SDR-family NADP+-dependent CH–OH oxidoreductase → SUPPORTED. Fold/family, the SDR
   glycine coenzyme fingerprint (TGGSGGIG), the catalytic Tyr-x-x-x-Lys tetrad (YHATK), and the
   cofactor-discriminating residues (basic cluster R42/K44/K45, no acidic Asp discriminator) all
   independently indicate a functional NADP-preferring SDR reductase. UniProt keyword NADP and the
   NADPH-dependent characterized ortholog agree.
2. "Alcohol dehydrogenase (NADP+) activity" (GO:0008106) → WEAKLY / PARTIALLY SUPPORTED.
   Defensible as a broad activity class, but it is less informative and less precise than the terms
   PomBase already carries (GO:0016616 CH–OH oxidoreductase; specific sugar/polyol reductase terms),
   and it remains experimentally unverified. Not the best available MF term.
3. Arabitol substrate / "arabitol metabolic process" → REFUTED / UNSUPPORTED and technically
   defective. No assay links SPAC8E11.10 to arabitol; the closest characterized ortholog acts on
   L-sorbose/fructose; and the proposed BP is a GO housekeeping problem: the seed cites GO:0019677,
   which is actually "NAD+ catabolic process," not arabitol, while the genuine "arabitol metabolic
   process" term (GO:0051161) is OBSOLETE. The arabitol call is best read as a paralog/substrate
   misassignment.

Most important caveat: the enzyme is officially "conserved unknown" (uncharacterized). No S. pombe
enzymatic assay exists, so every substrate assignment — sorbose, mannitol, xylulose, or arabitol —
is inference. The prediction is not "correct"; at best it names a plausible-but-unproven pentitol within
the right chemical class, wrapped in an obsolete/mis-cited GO term.

Evidence Matrix

GO Curation Implications (leads — require curator verification)

• GO:0008106 "alcohol dehydrogenase (NADP+) activity" (MF): Do not add as a specific
  experimental call. If any MF is asserted computationally, prefer the more accurate, already-present
  GO:0016616 ("oxidoreductase activity, acting on the CH–OH group of donors, NAD or NADP as
  acceptor") with an ISS/IEA evidence code and a "conserved unknown" caveat. GO:0008106 is an
  acceptable-but-suboptimal generalization; it is neither wrong nor the most informative supported term.
• Arabitol metabolic process (BP): Reject. The cited GO:0019677 is the wrong term (NAD+
  catabolism) and the intended arabitol term (GO:0051161) is obsolete. There is no substrate-level
  evidence. Do not annotate an arabitol BP.
• Sorbose reductase (GO:0032115) / mannitol 2-dehydrogenase NADP+ (GO:0050085): These orthology-
  transferred terms should be retained only as low-confidence predictions with the existing CAUTION;
  the sorbose role is explicitly disfavored at species level.
• Recommended framing: Retain the gene as an uncharacterized NADP-preferring SDR CH–OH
  oxidoreductase (carbohydrate assimilation, mitochondrion). The BioReason arabitol/NADP-ADH output
  should not upgrade the annotation beyond the current predicted state.

Mechanistic Scope

Immediate molecular function under test = a single-domain, cytosolic/mitochondrial NADP(H)-dependent
short-chain oxidoreductase that reversibly interconverts a sugar/polyol CH–OH group with the
corresponding carbonyl (ketose ⇌ polyol), using the Tyr/Lys catalytic couple and an NADP cofactor.
This is a direct enzymatic activity claim. "Arabitol metabolic process" and "carbohydrate
assimilation" are pathway/physiology-level consequences that would follow only if the true substrate
were arabitol/a pentitol — which is not established. No loss-of-function phenotype, localization-driven,
or developmental inference is invoked by the prediction; the CC (mitochondrion) is itself predicted.

Conflicts and Alternatives

• Paralog/ortholog substrate transfer: The "sou1" name and arabitol/sorbose reasoning derive from
  C. albicans SOU1, whose measured activity is sorbose/fructose (not arabitol). Transferring a
  substrate across ~1 billion years of fungal divergence is unreliable, and UniProt already flags this.
• Competing subfamily assignment: PANTHER places the protein closest to L-xylulose reductase /
  DCXR (PMID 11882650, 23661708, 21300042), an NADPH-dependent pentose/pentitol/α-dicarbonyl reductase
  (xylitol⇌L-xylulose). This is a different (though chemically adjacent) substrate space than arabitol,
  and — together with the Candida SOU1 sorbose/fructose data — gives two independent comparators that
  are NADPH-dependent but not arabitol-specific.
• Species physiology: S. pombe does not use L-sorbose, directly conflicting with a sorbose-reductase
  reading and casting doubt on naive polyol-substrate transfers generally.
• Cofactor caveat: The NADP signature is a sequence heuristic; some SDRs with basic βB residues
  retain measurable NAD activity. Without a holoenzyme structure or kinetics, NAD+ cannot be fully
  excluded, though NADP+ is the better-supported call.

Knowledge Gaps

Discriminating Tests

1. Recombinant enzyme substrate panel (most decisive): assay purified Q9Y6Z9 for reductase activity
   on D-/L-arabitol-linked ketoses (D-xylulose, L-xylulose, ribulose), L-sorbose, D-fructose, and
   corresponding polyol oxidation, with NADPH and NADH side-by-side. Directly separates arabitol vs
   sorbose vs xylulose specificity and NAD vs NADP.
2. AlphaFold holo-modeling + docking of NADP vs NAD and candidate polyols into the active site to
   corroborate the cofactor fingerprint and rank substrates.
3. Δsou1 growth phenotyping on arabitol, sorbose, xylitol, mannitol as sole carbon sources.
4. Phylogenetic reconciliation of the fission-yeast SDR clade vs Candida SOU1 vs DCXR/L-xylulose
   reductase to test whether "sou1" orthology actually implies sorbose/arabitol substrate.

Curation Leads (require curator verification)

• Action change: Do not adopt the BioReason arabitol/NADP-ADH prediction as a specific annotation.
  Keep gene as uncharacterized NADP-preferring SDR CH–OH oxidoreductase.
• Candidate MF (if any): GO:0016616 (ISS/IEA) preferred over GO:0008106; both experimentally unproven.
• Reject BP: arabitol metabolic process — wrong/obsolete GO ID (GO:0019677 = NAD+ catabolic;
  GO:0051161 obsolete).
• Candidate references to verify: PMID 16134116 (ortholog assay); PMID 31132130 (species carbon
  physiology, basis of UniProt CAUTION); UniProt Q9Y6Z9; PomBase SPAC8E11.10; PANTHER PTHR43008:SF13.
• Suggested question for curators: Is there any S. pombe experimental evidence (kinetics,
  phenotype, metabolomics) for arabitol metabolism at all? If not, the BP prediction cannot stand.
• Suggested experiment: recombinant substrate/cofactor panel (Discriminating Test 1).

Provenance artifacts: sou1_evidence_matrix.png (computed SDR motif / cofactor-residue analysis and
evidence table) and sou1_GO_decision_table.png / sou1_GO_decision_table.csv (per-term curation
decision table). All computational results above are sequence/database-derived inference, explicitly
distinguished from direct assays (PMID 16134116 on the C. albicans ortholog and PMID 11882650/23661708
on the DCXR/L-xylulose reductase subfamily — none performed on Q9Y6Z9 itself, which remains
enzymatically uncharacterized).

GO Decision Table (leads — curator verification required)

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist sou1 GO decision table
  
• OpenScientist sou1 evidence matrix