AIGR Deep Research Report — Q2U1U6 (ASPOR): Glycoside Hydrolase vs Polysaccharide Lyase

Gene: Q2U1U6 (Q2U1U6_ASPOR), Aspergillus oryzae RIB40, ORF AO090138000091 · 134 aa · PE=4 (Predicted)
Hypothesis under test (computational_prediction): ProtNLM2 predicts GO:0004553 "hydrolase activity, hydrolyzing O-glycosyl compounds" (i.e. glycoside hydrolase, GH) for Q2U1U6, whose only signature is IPR008929 / SUPFAM SSF48230 (Chondroitin AC/alginate lyase fold). Is Q2U1U6 a GH (hydrolysis) or a polysaccharide lyase (PL, β-elimination), and is the prediction supported or a class misassignment?

Executive Judgment

Verdict: REFUTED (class misassignment), with the residual signal being at most a weak, fold-level, non-catalytic resemblance.

The ProtNLM2 GO:0004553 (O-glycosyl hydrolase) prediction is not supported and is best read as a CAZy class misassignment:

1. Wrong mechanistic class. The single piece of evidence behind any enzymatic call is a fold-level match to SSF48230 / IPR008929, the (α/α)-toroid scaffold of CAZy polysaccharide lyases (PL5/PL8/PL15/PL35/PL40). These enzymes cleave glycosidic bonds by β-elimination (a lyase, EC 4.2.2.-), producing Δ4,5-unsaturated uronate — mechanistically distinct from hydrolysis (EC 3.2.1.-, GO:0004553). If any activity were implied by the fold, it would be lyase, not hydrolase. Predicting a hydrolase from a lyase fold is a category error.
2. Too small and only a partial match. Q2U1U6 is 134 aa; the SSF48230 match covers only residues 6–97 (92 aa) with a weak SUPFAM E-value (4.03e-06). Quantitatively decisive: all 53 reviewed IPR008929 members are 331–1372 aa (median 682); 0% are shorter than 134 aa, and the smallest members are annotated alginate/polysaccharide lyases. Q2U1U6 is ~2.5× smaller than the smallest family member and ~5× below the median — it cannot reconstitute the (α/α)₆ toroid, its substrate groove, or the His/Tyr(/Arg/Asn) catalytic constellation, nor does it provide the Asp/Glu acid–base pair of a hydrolase (geometry check found no canonical GH di-carboxylate pair; only a weak, incomplete His124–Tyr17–Asn85 cluster).
3. No positive evidence for catalysis of any kind. UniProt has no EC and no GO for this entry; PE=4 (Predicted); name is a placeholder ("DNA, SC138"). No primary literature characterizes this ORF. The AlphaFold model is confidently folded (global pLDDT 92.3) but is a compact all-α domain (~58% helix, no β-sheet, Rg 15.1 Å) — consistent with a fragment of the toroid superfamily, not a functional enzyme.

Most important caveat: "Refuted" applies specifically to the GH / GO:0004553 claim. The data do not license a positive PL annotation either; the honest state is uncharacterized protein with a weak all-α fold resemblance to the chondroitin/alginate lyase superfamily. This should not be annotated with any molecular-function enzyme term without experimental evidence.

Evidence Matrix

GO Curation Implications (leads requiring curator verification)

• GO:0004553 (MF, hydrolase / O-glycosyl): Do NOT add. Not supported; a mechanistic class misassignment (lyase fold → hydrolase term). If it appears in any prior action, recommend removal / rejection of the ProtNLM-derived prediction.
• Do NOT substitute a PL/lyase MF term either (e.g., GO:0016837 carbon-oxygen lyase, or GO:0030246-type). The evidence is fold-level and partial only; there is no experimental or full-domain basis for a positive enzyme annotation.
• Recommended stance: leave molecular function uncharacterized (no enzyme MF term). At most, an internal note that the protein shows a weak, partial structural resemblance to the chondroitin/alginate lyase (α/α-toroid) superfamily — orientation only, not an annotation.
• Avoid "protein binding" as a fallback; it is uninformative and equally unsupported.

Mechanistic Scope

The term tested is a direct molecular-function claim (catalysis of O-glycosidic bond hydrolysis). The analysis addresses that directly: (a) the fold family is catalytically a lyase, not a hydrolase; (b) the protein lacks the size and complete active-site architecture for either mechanism. No downstream phenotype, pathway, or developmental inference is involved — this is purely a molecular-activity assignment, and it fails at the molecular level.

Conflicts and Alternatives

• Paralog/label-propagation bias (now supported): A. oryzae has four IPR008929 paralogs — Q2TXR4 (406 aa), Q2USR3 (453 aa), Q2UB60 (395 aa) and Q2U1U6 (134 aa). Q2U1U6 is the lone truncated outlier; ProtNLM2 (a name/label language model) likely over-generalized a carbohydrate-active label from the larger paralogs, collapsing the mechanistically distinct PL vs GH distinction onto a fragment.
• No GH anywhere in the superfamily: Across all organisms, small (≤200 aa) IPR008929 members are annotated as alginate/mannuronate lyases, dermatan sulfate epimerase (EC 5.1.3.19), or uncharacterized — i.e. β-elimination/epimerase (carbanion) chemistry. None are glycoside hydrolases, confirming GO:0004553 is off-class for this fold.
• Alternative benign interpretation: Q2U1U6 may be a non-catalytic small all-α protein (structural/binding module) or a degenerate/truncated remnant of a lyase-fold ancestor. The confident AlphaFold fold plus small size is consistent with a stable mini-domain (or a truncated gene model) rather than a complete enzyme.
• Database carry-over risk: because the entry is TrEMBL/PE=4 with no GO, any downstream GH annotation would propagate an unverified, off-class prediction.

Knowledge Gaps

1. No biochemical assay for Q2U1U6 — checked: no primary literature on this ORF. Matters because activity (GH vs PL vs none) is undecidable from sequence alone. Resolve with recombinant expression + activity screen (see below).
2. No structural superposition to a characterized PL was run (no Foldseek/DALI available here). Checked fold class computationally (all-α, partial); a structural search against PDB PL/GH representatives would sharpen the "fragment vs domain" call.
3. Catalytic-residue geometry not confirmed in 3D: candidate His7/His51/His124 and multiple Tyr exist in sequence, but whether any form a lyase-like His/Tyr dyad in the model was not measured. Matters for judging even a lyase hypothesis.

Discriminating Tests

• Foldseek/DALI of AF-Q2U1U6-F1 against PDB: does it match a full toroid PL catalytic domain or only an isolated α-hairpin cluster? (Fast, decisive on the "partial fold" claim.)
• Active-site geometry check in the model: measure spatial proximity/orientation of His/Tyr candidates and any Asp/Glu pair; absence of a coherent catalytic center supports "non-catalytic."
• HMMER/CAZy dbCAN scan: does Q2U1U6 hit any PL or GH family HMM above threshold, or only the SUPFAM fold? Lack of a family-level PL/GH hit corroborates misassignment.
• Experimental (definitive): recombinant expression and (i) thiobarbituric-acid / A232 assay for β-eliminative unsaturated-product formation (lyase) vs (ii) reducing-sugar/pNP-glycoside assays (hydrolase), on chondroitin/alginate/ulvan and generic glycosides.

Curation Leads (verify before acting)

• Action: Reject/remove the ProtNLM2 GO:0004553 prediction as a PL→GH class misassignment; do not add any enzyme MF term.
• Candidate snippets to verify (exact quotes): PMID 34392362 — "cleaves the β-1,4 glycosidic linkage of chondroitin sulfate (CS) by β-elimination"; PMID 20507980 — "degraded by alginate lyases through a beta-elimination reaction" and "His(311) and Tyr(365) as the catalytic base and acid."
• Suggested reviewer question: Should the entry carry a fold-level "chondroitin/alginate lyase-like superfamily" note (orientation only) while remaining unannotated for MF?
• Suggested experiment: heterologous expression + paired lyase (A232/TBA) and hydrolase (reducing-sugar/pNP) assays to establish whether Q2U1U6 has any glycan-cleaving activity at all.

Provenance

Analyses executed (code + outputs retained in the iteration log):
- Iteration 1: UniProt text fetch (134 aa, no GO/EC, PE=4); InterPro API (SSF48230 match res 6–97, E=4.03e-06); AlphaFold API + PDB parse (global pLDDT 92.3; region 6–97 pLDDT 94.3; ~58% helix; no β-sheet; Rg 15.1 Å; catalytic-candidate residue inventory).
- Iteration 2: IPR008929 reviewed-member length distribution (n=53: 331–1372 aa, median 682, 0% < 134 aa; smallest members = alginate/PL lyases); AlphaFold active-site geometry test (no canonical GH di-carboxylate acid/base pair; single His124–Tyr17 5.3 Å dyad + Asn85, incomplete lyase-type cluster).
- Iteration 3: UniProt paralog query (4 A. oryzae IPR008929 proteins: 406/453/395/134 aa — Q2U1U6 is the truncated outlier); UniProt query of ≤200-aa IPR008929 members across all organisms (annotated as alginate/mannuronate lyases, dermatan sulfate epimerase, or uncharacterized — no glycoside hydrolases).
- Literature via PubMed (PMIDs 20507980, 34392362, 41532755, 22297996, 18500999).

Artifacts

• OpenScientist final report
• OpenScientist final report