AIGR TreeGrafter Function-Inference Stress Test — Final Report

Target: ptxD / PP_3376 (UniProt Q88HI1), Pseudomonas putida KT2440 (NCBITaxon:160488)
Seed hypothesis (TreeGrafter/PANTHER, IEA, GO_REF:0000118): ptxD has glyoxylate reductase (NADPH) activity (GO:0030267)
Hypothesis slug: function-hypothesis-go-0030267

Summary

The TreeGrafter/PANTHER graft of GO:0030267 (glyoxylate reductase, NADPH) onto Q88HI1 is substantially correct. By independent global pairwise alignment the 320-aa query is 53.4% identical to the biochemically characterized E. coli GhrB (glyoxylate/hydroxypyruvate reductase B, EC 1.1.1.79/1.1.1.81) — its single nearest characterized neighbor and exactly the PANTHER subfamily (PTHR10996:SF283) it was grafted onto. Its catalytic machinery (His285, Glu266, Arg237 triad; NAD(P)-binding Rossmann motif GMGRIG@152; both diagnostic PROSITE signatures) is fully intact, and the cofactor-specificity region matches the NADPH-using GhrB. Of the three characteristic phylogenetic-propagation failure modes, only #1 (granularity) applies, and only mildly: GhrB-type enzymes are dual glyoxylate and hydroxypyruvate reductases, so the single seed term understates the characterized activity.

The notable twist of this stress test is that the TreeGrafter graft is more trustworthy than the gene name. The record's competing annotation — "ptxD / phosphite dehydrogenase" (GO:0050609, EC 1.20.1.1) — rests only on an automated EMBL gene name and a distant 35.1%-identity homolog (characterized PtxD of P. stutzeri WM88, O69054). The query lacks PtxD's distinctive phosphite-binding residues (Lys76→Val75, Trp314→Ser307), retaining only the generic family triad that all reductases share. This is positive evidence against phosphite dehydrogenase activity. The real over-call in the record is therefore the ptxD symbol-driven phosphite annotation, not the propagated glyoxylate-reductase term.

All conclusions rest on sequence/structure homology; no direct enzymology exists for the P. putida protein. The recommended curation action is to retain GO:0030267, add GO:0016618 (hydroxypyruvate reductase) to capture the dual activity, and flag the phosphite-dehydrogenase (GO:0050609) annotation for curator scrutiny / removal.

Executive Judgment

Verdict: SUPPORTED, partially (too specific) — the graft is essentially correct; the applicable failure mode is #1 (granularity), not #2 (pseudo-enzyme) or #3 (mis-placement).

Most decisive evidence: the query is 53.4% identical to characterized E. coli GhrB — within-subfamily, function-conserving territory — and GhrB is its nearest characterized neighbor and the exact subfamily it was grafted to.

Independent Family / Function Assignment

• Most likely specific function: NADPH-dependent glyoxylate/hydroxypyruvate reductase (GhrB subfamily of the D-isomer-specific 2-hydroxyacid dehydrogenase family).
• Characterized homolog it is based on: E. coli GhrB (UniProt P37666; EC 1.1.1.79 glyoxylate reductase / EC 1.1.1.81 hydroxypyruvate reductase), 53.4% identity; PANTHER PTHR10996:SF283; InterPro IPR050223; Pfam PF00389 + PF02826. Human GRHPR (39.4%) is a second, independent corroborating neighbor.
• GO term I would assign: GO:0030267 (glyoxylate reductase, NADPH) together with GO:0016618 (hydroxypyruvate reductase, NAD(P)H). Relative to the seed term, this is the same branch — the seed is on the correct branch but narrower than the protein's likely dual-substrate activity.

Key Findings

Finding 1 — Closest characterized homolog is GhrB, not phosphite dehydrogenase

Global Needleman–Wunsch alignment of the 320-aa query against a panel of characterized family members produced an unambiguous ranking. The query is 53.4% identical over 320 aligned columns to E. coli GhrB (P37666, glyoxylate/hydroxypyruvate reductase B). All other characterized candidates are substantially more distant: characterized phosphite dehydrogenase PtxD (P. stutzeri WM88, O69054, EC 1.20.1.1) = 35.1%; E. coli SerA (P0A9T0) = 28.8%; E. coli GhrA (P75913) = 23.5%. The query carries the InterPro family signature IPR050223 (D-isomer 2-hydroxyacid dehydrogenase) with Pfam PF00389 + PF02826, and PANTHER places it in PTHR10996:SF283 "glyoxylate/hydroxypyruvate reductase B" — i.e., the GhrB subfamily. The graft target is therefore the genuine nearest relative, directly ruling out failure mode #3.

Finding 2 — Catalytic machinery is fully intact (not a pseudo-enzyme)

The query retains the complete D-2-hydroxyacid dehydrogenase catalytic apparatus, ruling out failure mode #2. The NAD(P)-binding Rossmann glycine-rich motif GMGRIG is present at position 152. The conserved family catalytic base His285 (within the NVVATP-H-IGS motif) and the paired acidic residue Glu266 align 1:1 to the characterized PtxD/GhrB catalytic triad. UniProt reports both PROSITE catalytic signatures PS00065 and PS00671 (D-2-hydroxyacid dehydrogenase) as matched. The protein is full-length (320 aa) with no frameshift or truncation. A conserved fold with a degenerate active site would argue against a catalytic MF term; here the active site is conserved, so the catalytic MF term is warranted.

Finding 3 — Query lacks PtxD's phosphite-binding residues while retaining the generic 2-hydroxyacid reductase triad

Mapping characterized PtxD (O69054) active/binding residues onto the query confirms a clean split. The generic family catalytic triad is conserved (PtxD Arg237 = query Arg237; PtxD Glu266 = query Glu266; PtxD His292 = query His285 in GhrB-like numbering; PtxD Ser295 = query Ser288). But the PtxD-distinctive phosphite-pocket residues are NOT conserved: PtxD Lys76 (phosphite binding) → query Val75; PtxD Trp314 → query Ser307; PtxD Tyr139 → query Phe139. Across a broadened panel, the two nearest characterized neighbors are both glyoxylate/hydroxypyruvate reductases (GhrB 53.4%, human GRHPR 39.4%), with PtxD (35.1%), formate DH (28.8%), SerA (28.8%), and GhrA (23.5%) all more distant. The query's catalytic His position (285) and length (320 aa) match GhrB (H285, 324 aa), not PtxD (H292, 336 aa). This is the decisive argument that the protein is not a phosphite dehydrogenase.

Finding 4 — Cofactor-determinant region matches the NADPH-using GhrB

In the Rossmann cofactor-specificity window (~residues 152–177, immediately after the GxGxxG motif), the query (GMGRIGEALARRAAAGFGMRVLYHSQ) is near-identical to NADPH-dependent GhrB (GMGRIGMALAQRAHFGFNMPILYNAR). Both carry basic Arg residues in the 2′-phosphate recognition region and lack the acidic Asp that H-bonds the adenosine 2′/3′-OH in strict NAD(H)-specific members (that Asp is present in NAD-dependent SerA, "YFYDIE", and in the PtxD region). This supports the NADPH qualifier of the seed term — at fingerprint level, not kinetic level.

Mechanistic Model / Interpretation

Q88HI1 is a soluble, NADPH-dependent D-2-hydroxyacid dehydrogenase of the GhrB type. Enzymes of this subfamily reduce small 2-oxo/2-hydroxy acids using NADPH:

   Glyoxylate      + NADPH + H+  --->  Glycolate     + NADP+     (EC 1.1.1.79; GO:0030267)
   Hydroxypyruvate + NADPH + H+  --->  D-glycerate   + NADP+     (EC 1.1.1.81; GO:0016618)

The chemistry is standard D-2-hydroxyacid dehydrogenase hydride transfer: His285 acts as general acid/base, oriented by Glu266, while Arg237 polarizes the substrate carbonyl; NADPH binds in the Rossmann domain whose 2′-phosphate pocket is lined by basic residues (the GhrB-like Arg-rich fingerprint).

Family placement and failure-mode triage:

Family: D-isomer 2-hydroxyacid dehydrogenases (IPR050223; PF00389+PF02826)
   |
   |-- GhrB-type (glyoxylate/hydroxypyruvate reductase, NADPH) ... 53.4% id  <== Q88HI1 lives here
   |      (E. coli GhrB P37666; human GRHPR 39.4%)
   |-- PtxD-type (phosphite dehydrogenase, EC 1.20.1.1) .......... 35.1% id  (NOT this protein)
   |-- SerA-type (3-P-glycerate dehydrogenase, NAD) ............. 28.8% id
   |-- Formate dehydrogenase .................................... 28.8% id
   |-- GhrA-type ............................................... 23.5% id

The one genuine annotation hazard in this gene record is not the TreeGrafter graft but the gene symbol ptxD, which implies phosphite dehydrogenase (EC 1.20.1.1, GO:0050609). That interpretation is refuted by the residue-level analysis (missing Lys76/Trp314 phosphite-binding residues; 18-point identity gap to true PtxD). The reviewed literature describes authentic P. stutzeri PtxD used as an NAD(H)-regeneration tool — a genuinely different, more distant enzyme — and should not be conflated with Q88HI1.

Active-Site / Placement Analysis (provenance)

Catalytic residues: INTACT. The generic triad, the Rossmann motif, and both PROSITE signatures are present; PtxD's phosphite-pocket residues are absent.

(a) Nearest-characterized-neighbor ranking (global NW, 320-aa query):

The two nearest characterized homologs are both glyoxylate/hydroxypyruvate reductases; phosphite dehydrogenase ranks third.

(b) Active-site discrimination — phosphite pocket vs generic triad (PtxD O69054 numbering):

Conclusion: Catalytic machinery is intact → competent oxidoreductase, not a pseudo-enzyme. The protein is a glyoxylate/hydroxypyruvate reductase, not a phosphite dehydrogenase. A low level of residual phosphite activity cannot be formally excluded from sequence alone, but both the nearest-neighbor ranking and the missing phosphite-pocket residues point away from it.

Provenance note: these computations (global NW alignments, residue mapping, motif identification) were executed during iterations 1–3; numeric outputs are reproduced above. The cofactor-specificity call is fingerprint-level, not kinetic — stated as a limitation.

Evidence Base

How the literature bears on the findings: The three PtxD papers are all about authentic phosphite dehydrogenase (P. stutzeri WM88) used as a cofactor-regeneration tool. They establish what genuine PtxD does — oxidize phosphite to phosphate with NAD(H)/NADPH regeneration — and thereby sharpen, rather than support, the distinction: Q88HI1 lacks the phosphite-binding residues these enzymes require and is 18 identity points more distant from true PtxD than from GhrB. The UniProt records for GhrB (P37666) and PtxD (O69054) anchor the two competing branches, and the homology ranking resolves the query firmly onto the GhrB (reductase) branch.

Limitations and Knowledge Gaps

• No direct enzymology on Q88HI1 itself. All evidence is sequence/structure-based homology; no measured kcat/Km, cofactor preference, or substrate panel exists for the P. putida protein. Confidence is high for "GhrB-type reductase," moderate for the exact preferred substrate (glyoxylate vs hydroxypyruvate) and cofactor (NADPH vs NAD(P)H).
• Cofactor specificity (NADPH vs NADH) is inferred from a sequence fingerprint, not measured. The GO:0030267 NADPH qualifier is plausible but not experimentally pinned for this ortholog.
• Residual phosphite activity cannot be formally excluded from sequence alone, though the missing phosphite-pocket residues make it unlikely.
• Paralog caution: Pseudomonas genomes carry several 2-hydroxyacid dehydrogenase paralogs; the ptxD label is likely database carry-over and may invite the wrong EC.
• Alignment method: crude global Needleman–Wunsch scoring; margins are large and robust, but a profile/HMM or structural superposition would strengthen the placement.

Proposed Follow-up Experiments / Actions

1. Enzyme assay (most decisive, single experiment). Express and purify recombinant Q88HI1 and measure steady-state kinetics for (a) glyoxylate + NADPH, (b) hydroxypyruvate + NADPH, versus (c) phosphite + NAD⁺. This resolves activity vs pseudo-enzyme, glyoxylate vs hydroxypyruvate ranking, NADPH vs NADH, and definitively excludes phosphite dehydrogenase. kcat/Km for glyoxylate/hydroxypyruvate ≫ phosphite would confirm GO:0030267/GO:0016618 over GO:0050609.
2. Structural superposition. AlphaFold/crystal model of Q88HI1 superposed onto GhrB (P37666) and PtxD (O69054) to confirm the absence of a phosphite-binding pocket at the modeled Val75/Ser307 positions.
3. Physiological phenotyping. Growth/complementation of a P. putida PP_3376 deletion on glyoxylate/glycolate and hydroxypyruvate/serine-cycle intermediates to assign the physiological role.

GO Curation Implications (leads for curator verification)

1. GO:0030267 (seed) — RETAIN, but recognize it as one half of a dual activity. Recommend make-more-complete, not more-specific: keep GO:0030267 and also retain/add GO:0016618 (hydroxypyruvate reductase). Both are consistent with the GhrB nearest neighbor.
2. GO:0050609 (phosphonate/phosphite dehydrogenase) — FLAG as likely over-call. It derives from the automated ptxD gene name, not experimental data, and the nearest characterized phosphite DH is only 35% identical. Candidate action: remove or demote pending evidence.
3. Do not downgrade to "protein binding" — the protein is an intact oxidoreductase.

Artifacts

• OpenScientist final report
• OpenScientist final report