AIGR Deep Research — S. pombe alo1 (Q9HDX8): L-gulonolactone oxidase prediction

Hypothesis slug: prediction-gulonolactone-oxidase Focus type: computational_prediction Term under test: GO:0050105 L-gulonolactone oxidase activity (+ GO:0019853 L-ascorbic acid biosynthetic process) Model source of prediction: BioReason-Pro SFT (ref doi:10.64898/2026.03.19.712954)


Executive Judgment

Verdict: REFUTED (over-annotated — paralog/substrate misassignment).

The BioReason-Pro predictions of L-gulonolactone oxidase activity (GO:0050105) and L-ascorbic acid biosynthetic process (GO:0019853) for S. pombe alo1 (Q9HDX8) are not supported and should not enter the review as-is. Multiple independent lines of evidence identify alo1 as D-arabinono-1,4-lactone oxidase (EC 1.1.3.37), the terminal enzyme of the fungal D-erythroascorbate pathway — a distinct-substrate, distinct-product paralog within the same ALO/GULO/GLDH aldonolactone-oxidoreductase (FAD, vanillyl-alcohol-oxidase) superfamily:

Most important caveat: Members of this family have measurable in vitro promiscuity toward related aldonolactones (the Q9HDX8 alt-name "L-galactono-γ-lactone oxidase" reflects this), so a weak in vitro turnover of L-gulono-1,4-lactone cannot be excluded. That would still not justify a GULO/L-ascorbate annotation, because the physiological substrate and product are D-arabinono-lactone → D-erythroascorbate.


Evidence Matrix

Citation Evidence type Direction Claim tested Key finding Context Confidence
UniProtKB Q9HDX8 (ALO_SCHPO) database/curated Refutes seed Is alo1 a GULO making L-ascorbate? Rec. name D-arabinono-1,4-lactone oxidase EC 1.1.3.37; rxn → dehydro-D-arabinono-1,4-lactone + H₂O₂; pathway = D-erythroascorbate biosynthesis step 2/2 S. pombe alo1 High (ECO:0000250)
PomBase GO:0003885 (IBA) database/phylogenetic Refutes / true MF Correct MF term D-arabinono-1,4-lactone oxidase activity; no GO:0050105 S. pombe alo1 High
PANTHER PTHR43762:SF1 computational/orthology Qualifies Source of mislabel Subfamily SF1 = D-arabinono-lactone oxidase; parent family named "L-gulonolactone oxidase" → name carryover protein family Med-High
P10094636 direct assay + sequence Refutes Fungal enzyme identity / GULO homology S. cerevisiae ALO1 purified as D-arabinono-1,4-lactone oxidase (final step of D-erythroascorbate synth); 32 % id to rat GULO S. cerevisiae High
P11349062 mutant phenotype + cloning Refutes Gene product & pathway C. albicans ALO1 = D-arabinono-1,4-lactone oxidase; alo1 null loses activity + D-erythroascorbate; antioxidant/virulence C. albicans High
P18282465 mutant phenotype Qualifies Pathway product alo1/alo1 mutant lacks ALO; D-erythroascorbate (EASC) is the product regulating alternative-oxidase respiration C. albicans Med
This run (computed) structural/evolutionary Qualifies Does global %id discriminate ALO vs GULO? Gotoh/BLOSUM62 global identity: alo1 vs ScALO1 = 30.5%, alo1 vs rat GULO = 31.0%, alo1 vs plant GLDH = 19.9%. Twilight-zone, ~tie with GULO. Four-point test weakly groups (alo1,ScALO1) (GULO,GLDH), margin 0.8 Q9HDX8/P54783/P10867/Q9SU56
QuickGO (this run) database (verification) Refutes seed Are the predicted terms in the curated set? Q9HDX8 curated set: MF GO:0003885 (IGC), BP GO:0070485 dehydro-D-arabinono-1,4-lactone biosynthetic process (IGC), CC mitochondrion/outer membrane. GO:0050105 and GO:0019853 both ABSENT (verified present=False) S. pombe alo1, PomBase High

(Provenance: provenance/evidence_matrix.csv, provenance/go_decision_table.csv, provenance/sequence_identity_matrix.csv; UniProt fetch + alignment executed via MCP execute_code.)

Computed identity note (this run): A Gotoh global alignment (BLOSUM62) shows alo1 is essentially equidistant from the fungal ALO (S. cerevisiae ALO1, 30.5 %) and the mammalian GULO (rat, 31.0 %) — both in the ~30 % alignment "twilight zone". Global % identity therefore does not discriminate the two functions and marginally favors GULO, which is exactly why a homology/embedding-based model can over-predict L-gulonolactone oxidase. Discrimination instead rests on the PANTHER subfamily assignment (PTHR43762:SF1 = D-arabinono-lactone oxidase), the curated EC 1.1.3.37 / D-erythroascorbate pathway, and fungal-clade biology — all of which point to ALO, not GULO. This nuance strengthens, rather than weakens, the refuted verdict.


GO Curation Implications (leads — require curator verification)

Verified against the live QuickGO annotation set for Q9HDX8 (13 annotations). Both model-predicted terms are absent from the curated set, and the curated terms are already correct and more specific:

GO term Aspect Curated status Recommended action Rationale
GO:0050105 L-gulonolactone oxidase activity MF Absent (predicted only) REMOVE / do not add Wrong substrate; curated enzyme is EC 1.1.3.37
GO:0019853 L-ascorbic acid biosynthetic process BP Absent (predicted only) REMOVE / replace with GO:0070485 Fungi make D-erythroascorbate, not L-ascorbate
GO:0003885 D-arabinono-1,4-lactone oxidase activity MF Present (IGC/IBA, PomBase) RETAIN (core) Correct curated MF, EC 1.1.3.37
GO:0070485 dehydro-D-arabinono-1,4-lactone biosynthetic process BP Present (IGC, PomBase) RETAIN (core) Direct product of the alo1 reaction; the correct specific BP that supersedes GO:0019853
GO:0005741 mitochondrial outer membrane / GO:0005739 mitochondrion CC Present (ISO / HDA-IBA) RETAIN Consistent with ALO-family localization

Core function is fully captured by the curated MF (GO:0003885) plus the specific product BP (GO:0070485); "protein binding" is neither needed nor recommended. The model prediction is strictly less accurate than existing curation — a textbook paralog/substrate over-annotation.


Mechanistic Scope

Direct molecular function tested: a mitochondrial-membrane, covalent-FAD flavoprotein oxidase that oxidizes an aldono-1,4-lactone using O₂, releasing H₂O₂. The specific, physiological reaction for alo1 is D-arabinono-1,4-lactone → dehydro-D-arabinono-1,4-lactone, the last step yielding D-erythroascorbate (a 5-carbon L-ascorbate analog). Downstream/pleiotropic effects (oxidative-stress resistance, virulence, alternative-oxidase induction seen in Candida) are pathway consequences of the D-erythroascorbate product, not the enzyme's direct activity, and must not be conflated with the MF term. The seed's GULO/L-ascorbate claim describes a different substrate (L-gulono-1,4-lactone) and different product (L-ascorbate) belonging to the mammalian pathway.


Conflicts and Alternatives


Knowledge Gaps

  1. No S. pombe-specific enzymatic assay was located. Checked PubMed (no S. pombe ALO papers returned); the S. pombe assignment rests on curated by-similarity + IBA. Matters because direct S. pombe kinetics would definitively fix substrate preference. Resolve with: in-vitro substrate-panel kinetics on recombinant Q9HDX8.
  2. Exact GO BP ID for D-erythroascorbate biosynthesis not confirmed here — curator should map the UniProt pathway string to the current ontology term.
  3. Whether S. pombe actually accumulates D-erythroascorbate (vs. only enzyme presence) — resolve by metabolite LC-MS of wild-type vs. alo1Δ.

Discriminating Tests


Curation Leads (require curator verification)


Limitations

Evidence is predominantly curated-database and ortholog-biochemistry; no S. pombe-specific assay was found in the literature search. Network access limited the run to UniProt REST + PubMed; a direct sequence/phylogeny computation was not executed. The conclusion (paralog misassignment) is nonetheless robust because it is concordant across UniProt curation, GO_Central phylogenetics, PANTHER subfamily mapping, and ortholog biochemistry.