AIGR TreeGrafter Function-Inference Stress Test — Final Report

Target: Octopus vulgaris DDO (UniProt A0A7E6FSU6, OXDD_OCTVU)
Seed hypothesis (TreeGrafter/PANTHER, IEA, GO_REF:0000118): DDO has D-amino acid catabolic process (GO:0019478)
Focus: function_assignment · existing_annotations[5].function_hypothesis

Summary

The TreeGrafter/PANTHER pipeline propagated the term GO:0019478 (D-amino acid catabolic process) to Octopus vulgaris DDO (A0A7E6FSU6). Independent, primary-evidence analysis confirms this assignment is correct. The query protein is a genuine, catalytically competent D-aspartate oxidase (EC 1.4.3.1) — a peroxisomal FAD-dependent flavoenzyme that oxidatively deaminates the acidic D-amino acids D-aspartate (and D-glutamate/NMDA) to the corresponding 2-oxo acid, ammonia, and hydrogen peroxide. This reaction is, by definition, a D-amino acid catabolic process, so the propagated biological-process term is accurate.

Critically, none of the three characteristic automated-propagation failure modes apply. (1) On granularity, the protein belongs to the correct subfamily — its closest characterized homolog is human D-aspartate oxidase (DDO, Q99489) at 41.4% identity, not the broad-specificity D-amino acid oxidase (DAAO) at ~36%. (2) On pseudo-enzyme status, the specific catalytic, substrate-selectivity, cofactor-binding, and targeting residues are all present and correctly spaced in both sequence and the AlphaFold model — the enzyme is active, not a degenerate fold. (3) On within-superfamily mis-placement, the sequence sits squarely in the acidic-D-amino-acid-oxidase clade, confirmed by the diagnostic acidic-substrate selectivity arginine R234.

The single most decisive piece of evidence is the intact, physically assembled active site: in the high-confidence AlphaFold model (mean pLDDT 95.3), the catalytic Arg275, selectivity Arg234, and active-site Tyr220 cluster within ~5–8 Å of one another — the geometric signature of a functional acidic-D-amino-acid oxidase — while octopus D-aspartate oxidase is an experimentally established enzyme historically used as a specific analytical reagent to quantify D-aspartate (PMID: 8446003). The recommended curation action is to retain GO:0019478, optionally making it more specific by adding the molecular-function term GO:0008445 (D-aspartate oxidase activity). This case is a clean true-positive for the TreeGrafter pipeline.

Executive Judgment

Verdict: SUPPORTED. The seed hypothesis holds. None of the three failure modes the stress test probes for are present:

Single most decisive piece of evidence: the intact, correctly-spaced catalytic cluster in the AlphaFold model (R275–R234 = 5.2 Å; R275–Y220 = 5.4 Å), combined with the experimental precedent that octopus D-aspartate oxidase is a real, active enzyme (PMID: 8446003).

Independent Family / Function Assignment

Independent of the propagated term, my own analysis assigns:

• Specific molecular function: D-aspartate oxidase activity (EC 1.4.3.1) — FAD-dependent oxidative deamination of acidic D-amino acids (D-aspartate, D-glutamate/NMDA).
• Characterized homolog this is based on: human DDO (Q99489), the nearest characterized neighbor at 41.4% pairwise identity.
• GO term I would assign:
• Molecular function: GO:0008445 (D-aspartate oxidase activity) — more specific than, and the enzymatic basis for, the seed term.
• Biological process: GO:0019478 (D-amino acid catabolic process) — same branch as the seed term; correct as propagated.

Granularity relative to the seed term: the seed BP term is correct (same branch); the optional MF addition (GO:0008445) is more specific. No sibling or different-branch reassignment is warranted.

Key Findings

Finding 1 — A0A7E6FSU6 is a bona-fide D-aspartate oxidase with intact catalytic machinery

UniProt curates A0A7E6FSU6 (OXDD_OCTVU) with the recommended name D-aspartate oxidase, EC 1.4.3.1. Its domain architecture is exactly what this enzyme requires: Pfam PF01266 (DAO), InterPro IPR023209 (DAO) plus IPR006181 (D-amino acid oxidase conserved site) with a positive PROSITE PS00677 match, and PANTHER PTHR11530. The entry also carries the UniProt keyword "Direct protein sequencing" — the protein was physically isolated and sequenced, strong evidence that this is a genuinely expressed gene product and not a pseudogene or prediction artifact.

A pairwise global alignment over the 336-residue protein places it closest to the D-aspartate oxidase subfamily rather than the generic DAAO subfamily:

The specific catalytic and substrate-recognition residues are conserved and correctly positioned (full table in Active-Site / Placement Analysis below). The catalytic arginine (human DDO R278 / pig DAAO R283) maps to octopus R275; the acidic-substrate selectivity arginine (human DDO R237) maps to R234; the active-site tyrosine (pig DAAO Y228) maps to Y220; the N-terminal FAD Rossmann fingerprint (GxGxxG) is present as G8-AGVVGL; and the C-terminal peroxisomal targeting signal (PTS1) is present as SKL. The one non-conserved basic position (human DDO R216 → octopus Y213) is distal to the active site and does not affect catalysis. The principal selectivity arginine (R234) — the residue that distinguishes DDO from DAAO by anchoring the side-chain carboxylate of D-aspartate — is fully intact.

Finding 2 — GO:0019478 is correctly propagated; it could be made more specific

D-aspartate oxidase (EC 1.4.3.1) oxidatively deaminates acidic D-amino acids (D-Asp, D-Glu/NMDA) to the 2-oxo acid + NH₃ + H₂O₂ — by definition a D-amino acid catabolic process (GO:0019478). The octopus is an established natural source of this enzyme: "Octopus D-aspartate oxidase" is used as a specific analytical reagent to quantify D-aspartate (PMID: 8446003), and free D-Asp occurs specifically in cephalopod (Octopus vulgaris, Loligo, Sepia) nervous tissue. There is thus independent biochemical precedent that the octopus genome encodes a catalytically active D-aspartate oxidase, and the sequence/active-site analysis confirms A0A7E6FSU6 is that enzyme. The matching molecular-function term GO:0008445 (D-aspartate oxidase activity) is the specific activity underlying the propagated BP term.

Verbatim support (PMID: 8446003): "D-Asp was discovered using specific methods that employ Octopus D-aspartate oxidase and hog kidney D-amino acid oxidase to measure D-Asp levels."

Finding 3 — AlphaFold geometry confirms an assembled, intact active site

The AlphaFold DB model AF-A0A7E6FSU6-F1 (v6, mean pLDDT 95.3) is high-confidence. The three key residues cluster tightly in three dimensions, forming a compact substrate-binding/catalytic pocket:

Two arginines positioned ~5 Å apart is exactly the arrangement expected for anchoring both carboxylate groups of D-aspartate — the structural signature of the acidic-amino-acid oxidase (DDO), not the broad-specificity DAAO. The single non-conserved position (Y213, equivalent to human DDO R216) sits 16–21 Å from the catalytic cluster, i.e. outside the active site, so its substitution is mechanistically irrelevant. The C-terminal PTS1 (…S334-K335-L336) is on a solvent-exposed, lower-pLDDT (82–88) terminus — the appropriate context for Pex5 recognition and peroxisomal import. The N-terminal FAD Rossmann motif centroid lies ~17 Å from the catalytic arginine, consistent with canonical DAO-fold geometry.

Finding 4 — Subfamily placement across a 9-homolog panel; BP term robust to DDO/DAAO ambiguity

Benchmarking against nine characterized homologs reinforces the DDO assignment:

The single best match is human D-aspartate oxidase, and the mean identity to DDOs (37.1%) exceeds that to DAAOs (33.9%). At invertebrate-to-vertebrate evolutionary distance, global identity alone does not cleanly separate DDO from DAAO (the best DAAO, mouse, reaches 37.8%). The decisive subfamily feature is therefore the conserved acidic-substrate selectivity arginine R234 (= human DDO R237), reinforced by UniProt's curated naming and the literature reagent use of octopus D-aspartate oxidase. Crucially, the propagated BP term (GO:0019478) is robust to the residual DDO/DAAO ambiguity: both DDO and DAAO catabolize D-amino acids, so the seed term is correct either way; the distinction only affects the optional, more-specific MF term (GO:0008445 vs GO:0003884).

Mechanistic Model / Interpretation

The enzyme is a peroxisomal FAD-dependent oxidase operating on acidic D-amino acids:

   D-aspartate  +  O2  +  H2O
  │
  ▼   [ FAD-dependent oxidative deamination ]
   oxaloacetate  +  NH3  +  H2O2

Structural logic of the active site (AlphaFold AF-A0A7E6FSU6-F1):

FAD (Rossmann, G8-AGVVGL)
      │  ~17 Å
      ▼
   ┌───────────────────────────────┐
   │   D-aspartate binding pocket   │
   │                                │
   │   R275 (catalytic)             │
   │     │5.2 Å        │5.4 Å       │
   │   R234 (selectivity)  Y220     │
   │        └──7.6 Å──┘             │
   └───────────────────────────────┘
      ▲
   Y213 (non-conserved, 16–21 Å away — outside pocket, irrelevant)

   C-terminus: …S-K-L (PTS1) → Pex5 → peroxisome

The two arginines ~5 Å apart form a dual-carboxylate clamp that selects for acidic D-amino acids (D-Asp, D-Glu/NMDA), distinguishing this DDO subfamily from the broad-specificity DAAO subfamily, which lacks the second selectivity arginine. The FAD cofactor accepts the hydride during oxidative deamination and is re-oxidized by O₂, generating H₂O₂. The PTS1 (SKL) tripeptide targets the enzyme to the peroxisome, the canonical compartment for this reaction. This is internally consistent biochemistry end to end: a correct fold, a correctly assembled and correctly specific active site, the correct cofactor-binding motif, and the correct subcellular targeting signal all point to a single coherent function.

Biologically, this fits the known neurochemistry of cephalopods, where free D-aspartate accumulates in nervous tissue and a dedicated D-aspartate oxidase provides the catabolic counterpart. The net conclusion is that TreeGrafter got this one right; the case serves as a valuable true-positive control for the AIGR stress test — a propagated IEA term that withstands independent primary-evidence scrutiny.

Evidence Base

How the two literature papers bear on the findings. PMID: 8446003 is the key positive anchor: it establishes that an active octopus D-aspartate oxidase exists and is reliable enough to be used as a specific analytical reagent — directly corroborating that the octopus DDO gene product carries genuine catalytic activity. PMID: 24237903 describes the sibling enzyme human DAAO as a peroxisomal flavoenzyme central to brain D-serine catabolism; it supports the family-level mechanism and compartment while underscoring the DDO-vs-DAAO distinction that the active-site analysis resolves.

Active-Site / Placement Analysis (computed)

Catalytic residues: INTACT. Pairwise global alignment of A0A7E6FSU6 against characterized members, with residues mapped from reference numbering to octopus positions:

Interpretation. Identity: octopus vs human DDO 41.4%, vs human DAAO 36.3%, vs pig DAAO 35.6% (~331–333 aligned columns each). Catalytic residues are intact → not a pseudo-enzyme; subfamily is D-aspartate oxidase → not mis-placed. The structural confirmation from AlphaFold (mean pLDDT 95.3) shows the catalytic/substrate residues clustering into a compact, high-confidence active site (R275–R234 = 5.2 Å, R275–Y220 = 5.4 Å, R234–Y220 = 7.6 Å, all pLDDT 94–97), confirming the pocket is physically assembled. The single non-conserved position (Y213 = human DDO R216) lies 16–21 Å from the catalytic cluster, so the substitution flagged in the sequence analysis is structurally peripheral and functionally irrelevant. The sequence therefore belongs to the acidic-D-amino-acid-oxidase (DDO) subfamily of the DAO flavoprotein fold, with the conserved selectivity arginine R234 as the decisive discriminator.

GO Curation Implications

Recommended curation action (lead requiring curator verification):

• GO:0019478 (D-amino acid catabolic process), BP — RETAIN. The propagated IEA term is correct and well-supported by independent primary evidence. The IEA/GO_REF:0000118 propagation is accurate here and can be upgraded with the sequence- and structure-based evidence above.
• Optionally MAKE-MORE-SPECIFIC by adding the molecular-function term GO:0008445 (D-aspartate oxidase activity) / EC 1.4.3.1, which the sequence and active-site evidence directly support.
• Do not remove, generalize, replace with a sibling term, treat as non-core, or fall back to "protein binding." The annotation is core and accurate.

This is a true-positive TreeGrafter propagation and a useful positive control for the AIGR pipeline.

Limitations and Knowledge Gaps

1. No direct enzymatic assay on this exact accession. Catalytic competence is inferred from sequence conservation, AlphaFold geometry, and the general precedent that octopus D-aspartate oxidase is an active, reagent-grade enzyme (PMID: 8446003). A kinetic assay on recombinant A0A7E6FSU6 would convert high-confidence inference into direct evidence.
2. DDO vs DAAO not cleanly separable by global identity. At invertebrate-to-vertebrate distance, the best DAAO homolog (mouse, 37.8%) is close to the DDO mean (37.1%). The DDO call rests principally on the single diagnostic selectivity arginine (R234) plus curated naming — a narrow structural feature. Residual DAAO-like activity on neutral/basic D-amino acids cannot be excluded without substrate-profiling. Importantly, this ambiguity does not affect the BP term under test.
3. AlphaFold is a model, not an experimental structure. The pocket geometry (R275–R234 = 5.2 Å, etc.) is a high-confidence prediction (pLDDT 95.3), but no crystal or cryo-EM structure of this protein exists to confirm it.
4. Accession-to-literature link is indirect. The historical reagent "octopus D-aspartate oxidase" predates this genome-derived accession; their identity is highly plausible but not formally proven.
5. GO_REF:0000118 / IEA provenance was taken as given; the reference-tree graft itself was not re-run here.

Proposed Follow-up Experiments / Actions

1. Recombinant kinetic assay (most decisive). Express A0A7E6FSU6 in E. coli or yeast and measure activity on D-aspartate and D-glutamate/NMDA (expected DDO substrates) versus neutral/basic D-amino acids (D-serine, D-alanine, D-proline — DAAO substrates). A high D-Asp/D-Ser activity ratio confirms strict DDO identity and would justify locking in GO:0008445.
2. Substrate-selectivity mutagenesis. Mutate R234→Y (mimicking the DAAO-like loss of the second selectivity arginine) and test whether acidic-substrate preference is abolished — a direct functional validation that R234 is the DDO determinant.
3. Peroxisomal localization check. Confirm the PTS1 (SKL) functions by expressing a GFP-fusion in a cephalopod or heterologous cell line and verifying peroxisomal import, closing the loop on the compartment.
4. Experimental structure. A crystal structure with a bound substrate/analog (e.g., D-aspartate or a malate/tartrate mimic) would verify the AlphaFold-predicted dual-arginine clamp.
5. Curatorial action. Flag the annotation as a verified true-positive and, pending or in lieu of assay, add GO:0008445 as an ISS/IEA molecular-function term with the active-site evidence as the justification.

Report generated from a 3-iteration autonomous investigation. Findings F001–F004 and literature PMIDs 8446003 and 24237903 as recorded in the knowledge state.

Artifacts

• OpenScientist final report
• OpenScientist final report