dhlA (P22643, XANAU) — Epoxide Hydrolase Activity (GO:0004301) Hypothesis Review

Focus type: function_assignment
Seed hypothesis: dhlA has epoxide hydrolase activity (GO:0004301).
Decision context: GO:0004301 (epoxide hydrolase activity), Evidence IEA, original reference GO_REF:0000118.
Source: genes/XANAU/dhlA/dhlA-ai-review.yaml → existing_annotations[2].function_hypothesis

Executive Judgment

Verdict: Over-annotated (refuted as a direct function).

The gene product P22643/DHLA_XANAU is the archetypal haloalkane dehalogenase (EC 3.8.1.5)
from Xanthobacter autotrophicus GJ10. Its experimentally established, curated molecular
function is hydrolytic cleavage of carbon–halogen bonds (GO:0018786, haloalkane dehalogenase
activity). The GO:0004301 epoxide hydrolase annotation is an IEA:TreeGrafter assignment
(consistent with the supplied GO_REF:0000118) that arises purely from the shared α/β‑hydrolase
fold and the InterPro "Epoxide hydrolase‑like" signature (IPR000639 / PRINTS EPOXHYDRLASE).
Epoxide hydrolases (EC 3.3.2.x) and haloalkane dehalogenases are homologous, use the same fold and
an Asp–His–Asp/Glu triad with a covalent enzyme intermediate, but are distinct enzymatic
activities — interconverting them requires deliberate active‑site engineering. No primary
literature or curated UniProt catalytic-activity statement reports epoxide hydrolysis by DhlA.

Most important caveat: The annotation is not biologically absurd (the fold genuinely is
epoxide‑hydrolase‑like, and promiscuous low-level activity on some epoxides cannot be formally
excluded without an assay), but it is less precise than existing experimental knowledge and
should be treated as a non‑core, computationally derived term that a curator would normally not
retain for this well-characterized enzyme.

Evidence Matrix

QuickGO Annotation Inventory for P22643 (computed provenance, Iteration 3)

Complete set of GO annotations on P22643 (EBI QuickGO API) — all four are IEA:

• Confirms directly that GO_REF:0000118 = the TreeGrafter pipeline (QuickGO assignedBy field).
• Despite DhlA being one of the most experimentally characterized α/β-hydrolases, no experimental
  (EXP/IDA/IMP) GO annotation exists — every term is electronic. The dehalogenase term is nonetheless
  backed by UniRule + the curated EC 3.8.1.5 reaction and decades of primary enzymology; the epoxide
  hydrolase term is TreeGrafter-only and has no corresponding assay.

Family-Wide Annotation Audit (computed provenance, Iteration 2)

UniProt REST query of all 33 reviewed Swiss-Prot EC 3.8.1.5 haloalkane dehalogenases,
checking presence of each GO term:

• The core dehalogenase term is universal; the epoxide hydrolase term is patchy.
• GO:0004301 is present in DHLA_XANAU/DHLA_XANFL and the DhmA subgroup
  (Mycobacterium/Caulobacter) but absent from the entire DhaA subgroup
  (e.g., DHAA_RHORH/P0A3G2, DHAA_MYCTU/P9WMR9) and from LinB (D4Z2G1).
• Zero of the 33 enzymes carry an experimental (non-IEA) epoxide hydrolase annotation.

Interpretation: a genuinely conserved catalytic activity would annotate uniformly; the
39% patchiness tracks TreeGrafter/PANTHER subtree placement, i.e. classic homology-driven
over-propagation of the sibling fold activity — not evidence of real epoxide hydrolysis.

GO Curation Implications

• Lead (requires curator verification): Treat GO:0004301 (epoxide hydrolase activity, IEA/TreeGrafter) as a
  non-core, over-propagated computational term. Recommended action: remove or demote (do not
  retain as a representative function) for P22643, because a more specific, experimentally grounded
  MF term already exists.
• Retain as the primary MF term: GO:0018786 (haloalkane dehalogenase activity) — matches the
  curated EC 3.8.1.5 and all primary enzymology.
• Retain BP: GO:0019260 (1,2-dichloroethane catabolic process) — pathway context.
• The epoxide-hydrolase term is too broad/incorrect in specificity for this gene: it is an MF
  sibling activity on the same fold, not DhlA's demonstrated activity. It should not be generalized
  or made more specific — it should be flagged as homology carry-over.

Mechanistic Scope

• Immediate molecular activity tested: hydrolysis of an epoxide C–O bond (epoxide hydrolase).
• What DhlA actually does (direct): nucleophilic Asp124 performs SN2 attack on the terminal carbon
  of a haloalkane, displacing halide and forming a covalent alkyl–enzyme ester, which is then
  hydrolyzed by a water activated by the His289–Asp260 pair — yielding a primary alcohol, halide, and H⁺.
• Relationship to epoxide hydrolysis: EH uses the same fold and an analogous Asp–His–acid triad
  but opens an epoxide ring (covalent alkoxy/ester intermediate). The mechanistic similarity is the
  reason for the mis-transfer; it is not evidence that DhlA performs epoxide hydrolysis.
• Epoxide hydrolysis would be a separate catalytic function, not a downstream phenotype — so the
  question is purely whether DhlA has this activity, and the record indicates it does not (as a
  demonstrated function).

Conflicts and Alternatives

• Database carry-over / paralog-family confusion (most likely explanation): TreeGrafter grafts
  the sequence onto a PANTHER family (PTHR42977 "HYDROLASE-RELATED") whose members include epoxide
  hydrolases; the InterPro "Epoxide hydrolase-like" signature (IPR000639) and PRINTS EPOXHYDRLASE
  then license GO:0004301. This is frequency/fold bias, not organism-specific biology.
• Promiscuity caveat (cannot be fully excluded): α/β-hydrolase enzymes are catalytically
  promiscuous; low-level, in-vitro-only epoxide hydrolysis by DhlA has not been reported but also has
  not been exhaustively assayed. Even if present, promiscuous activity would be non-core and should
  not be annotated without an experimental reference.
• No conflicting primary evidence supports epoxide hydrolase activity as a physiological function.

Knowledge Gaps

1. Has DhlA ever been assayed against epoxides (e.g., epichlorohydrin, styrene oxide, 1,2-epoxyalkanes)?
   Checked UniProt catalytic-activity statements and primary enzymology abstracts — none report epoxide
   turnover. Matters because a single positive assay would convert "over-annotation" into
   "non-core promiscuous activity." Resolver: direct spectrophotometric/GC epoxide-hydrolase assay on
   purified DhlA.
2. Exact PANTHER subfamily/tree node driving the TreeGrafter call. Checked UniProt cross-refs
   (PTHR42977:SF3). Matters because knowing the grafted family clarifies whether the whole subfamily is
   mis-annotated. Resolver: inspect the PANTHER tree and GO_REF:0000118 propagation rules.
3. Whether sibling haloalkane dehalogenases (DhaA, LinB, DbjA) carry the same GO:0004301 IEA.
   Not yet queried. Matters for systematic correction across the family. Resolver: batch UniProt/QuickGO query.

Discriminating Tests

• Direct epoxide hydrolase assay on purified recombinant DhlA (e.g., styrene oxide, epichlorohydrin,
  1,2-epoxyhexane) measuring diol formation vs. a known EH positive control and a buffer blank — the
  single most decisive experiment.
• Active-site logic check: DhlA's halide-binding pocket (Trp125/Trp175) is tuned to stabilize a
  departing halide, not an alkoxide/diol; structural comparison with StEH1/human EH active sites
  predicts poor epoxide turnover. A docking/structure comparison is a cheap computational discriminator.
• Family-wide annotation audit: QuickGO/UniProt query of all PTHR42977 members and all EC 3.8.1.5
  enzymes to quantify how broadly GO:0004301 is TreeGrafter-propagated (frequency-bias test).

Curation Leads (require curator verification)

• Candidate action: Do not accept GO:0004301 as a represented molecular function for dhlA;
  mark the IEA/TreeGrafter epoxide hydrolase term as over-annotated / non-core and prefer
  GO:0018786 (haloalkane dehalogenase activity) + GO:0019260 (1,2-dichloroethane catabolic process).
• Candidate reference + snippet to verify (PMID:38828992):
  "Epoxide hydrolase StEH1, from potato, is similar in overall structural fold and catalytic mechanism
  to haloalkane dehalogenase DhlA from" — use to document that fold/mechanism similarity (not a real
  shared activity) explains the term; the two activities had to be engineered apart.
• Candidate reference + snippet (PMID:10485878):
  "this domain is similar to that of haloalkane dehalogenase and shares the alpha/beta hydrolase fold"
  — supports homology-driven mis-transfer.
• Database provenance to cite: UniProt P22643 shows GO:0004301 = IEA:TreeGrafter (= GO_REF:0000118),
  while the curated reaction statements are strictly EC 3.8.1.5 dehalogenase.
• Suggested curator question: Is there any experimental reference showing epoxide turnover by DhlA?
  If none, remove/demote GO:0004301.
• Suggested experiment: Spectrophotometric epoxide hydrolase assay on purified DhlA with EH-positive
  control.

Limitations

• This review relied on literature and the public UniProt/InterPro records; no wet-lab assay was run.
• Absence of a reported epoxide-hydrolase assay is evidence of absence of demonstrated activity, not
  proof that trace promiscuous activity is impossible.
• GO_REF:0000118 ↔ TreeGrafter equivalence was inferred from the UniProt evidence string "IEA:TreeGrafter";
  curators should confirm the exact GO_REF mapping in their pipeline.

Artifacts

• OpenScientist final report
• OpenScientist final report