AIGR Gene Hypothesis Deep Research — PP_3394 (Q88HG4), *Pseudomonas putida* KT2440 OpenScientist openscientist-autonomous 1 citations 3 artifacts 2026-09-20T23:24:52.883310 citations file

AIGR Gene Hypothesis Deep Research — PP_3394 (Q88HG4), Pseudomonas putida KT2440

Focus type: function_assignment
Hypothesis slug: dre-tim-reaction-and-pathway-specificity
Target: PP_3394 / Q88HG4 (309 aa, HMGL-like TIM-barrel; Pfam PF00682; CDD cd07938 DRE_TIM_HMGL), Pseudomonas putida KT2440 (NCBITaxon:160488)


Executive Judgment

Verdict: Partially supported at the family level, but the specific reaction is UNRESOLVED, and the confident "HMG-CoA lyase / leucine catabolism" assignment for PP_3394 is over-annotated (weakly supported).

PP_3394 is a genuine member of the HMG-CoA-lyase / DRE-TIM metallolyase superfamily (the "another oxo-acid lyase reaction, or acyl transfer forming a carbon-carbon bond" framing in the seed hypothesis is broadly correct). It retains the catalytic core residues of the fold, so it is very likely to be a catalytically competent enzyme. However, three lines of computed evidence converge to show that PP_3394 is NOT the physiological leucine/terpene HMG-CoA lyase of P. putida KT2440, and that its precise substrate specificity cannot be inferred from sequence:

  1. PP_3394 is a highly divergent paralog (~38% identity to every reference), equidistant from bona fide HMG-CoA lyases (LiuE/mvaB/human HMGCL) and from isopropylmalate/homocitrate synthases (IPMS LeuA). Pairwise similarity alone does not — and cannot — settle its specificity.
  2. PP_3394 FAILS the HMG-CoA-lyase-specific PROSITE active-site signature PS01062 (SVAGLGGCPY) that all three verified HMG-CoA lyases pass, and instead MATCHES the IPMS/homocitrate-synthase site-1 signature PS00815 and carries the condensation-enzyme-type "LRDG" N-terminal motif rather than the HMG-CoA-lyase "PRDG" motif.
  3. The genuine leucine/terpene HMG-CoA lyase of P. putida is the paralog mvaB / PP_3540, which is 78.6% identical to the experimentally characterized P. aeruginosa LiuE (PA2011) — an ortholog-level match — and which carries PS01062. KEGG nonetheless dual-annotates BOTH PP_3394 and PP_3540 to K01640 (EC 4.1.3.4), the textbook signature of paralog over-annotation.

Most important caveats: (a) All of the discriminating evidence is computational (pairwise alignment, PROSITE/motif scanning, and a whole-sequence NJ phylogeny). There is no direct biochemical assay or mutant phenotype for PP_3394 itself; the only experimental anchor is for the P. aeruginosa ortholog of mvaB (LiuE, PMID 19459965). (b) "Ketone-body biosynthesis" is a mammalian physiological concept and should not be transferred to this bacterium regardless of enzyme family. (c) Because the catalytic core is retained but the specificity determinants diverge, PP_3394 could plausibly act on an HMG-CoA-like or a different acyl-CoA / oxo-acid substrate — the honest curation position is "unresolved DRE-TIM metallolyase," not "HMG-CoA lyase."


Summary

This report adjudicates whether P. putida KT2440 PP_3394 directly catalyzes HMG-CoA cleavage in support of leucine catabolism and/or ketone-body biosynthesis, or a different DRE-TIM metallolyase reaction. Across three iterations, the investigation confirmed four findings that together resolve the paralog question that prior mvaB and branched-chain-amino-acid-catabolism reports had explicitly left open.

The core conclusion is that the leucine/terpene HMG-CoA-lyase function of P. putida belongs to mvaB / PP_3540, the reciprocal-best ortholog (78.6% identity) of the experimentally verified P. aeruginosa enzyme LiuE (PMID: 19459965). PP_3394, by contrast, is a divergent paralog (~38% identity to every reference protein) whose active-site signatures do not match the HMG-CoA-lyase group and instead resemble the isopropylmalate/homocitrate-synthase group, even though whole-sequence phylogeny places it basally within the HMG-CoA-lyase clade. This tension between backbone relatedness and specificity motifs is the hallmark of a functionally diverged paralog of unresolved substrate specificity.

For curation, the practical upshot is that PP_3394's confident HMG-CoA-lyase / leucine-catabolism annotation is over-annotated and should be generalized to a low-confidence oxo-acid/carbon-carbon lyase molecular function, with leucine catabolism treated as non-core and ketone-body biosynthesis rejected as a mammalian concept inapplicable to this bacterium. The strong, experimentally anchored annotation belongs on mvaB/PP_3540. Resolving PP_3394's true reaction requires direct biochemistry or a mutant phenotype — neither of which currently exists for this gene.


Key Findings

Finding 1 — mvaB (PP_3540), not PP_3394, is the ortholog of the experimentally verified leucine/terpene HMG-CoA lyase LiuE

The single experimentally anchored data point in this entire family for the Pseudomonas genus is P. aeruginosa LiuE (PA2011, Q9I2A0). Chávez-Avilés and colleagues showed by direct enzymology and mutant phenotype that LiuE is a 3-hydroxy-3-methylglutaryl-CoA lyase (HMG-CoA lyase) that cleaves HMG-CoA to acetyl-CoA + acetoacetate (Km ≈ 100 µM HMG-CoA; Vmax ≈ 21 µmol/min/mg), and that a liuE deletion abolishes growth on both leucine/isovalerate and acyclic terpenes (PMID: 19459965).

When the P. putida KT2440 HMGL-family paralogs are aligned (global Needleman–Wunsch, BLOSUM62) against LiuE:

Query UniProt % identity to LiuE (PA2011) PS01062 (HMG-CoA lyase site)
mvaB / PP_3540 Q88H25 78.6% (ortholog-level) Match (SVAGLGGCPY)
PP_3394 Q88HG4 38.2% (superfamily-level) FAIL
(comparison) human HMGCL P35914 — Match
(comparison) E. coli IPMS LeuA P09151 — Fail

mvaB's 78.6% identity is the reciprocal-best, ortholog-level relationship expected for a true functional counterpart. PP_3394's 38.2% is no better than its 38.0% identity to human HMGCL or its 38.8% identity to E. coli IPMS — i.e., PP_3394 is generically "an HMGL-fold enzyme," not specifically "a LiuE ortholog." This finding directly answers the seed hypothesis's core question: the leucine-catabolic HMG-CoA lyase role of P. putida maps to mvaB / PP_3540, not to PP_3394.

Finding 2 — PP_3394 is a divergent DRE-TIM metallolyase of unresolved substrate specificity, not a canonical HMG-CoA lyase

PP_3394 (309 aa) is a single HMGL-like TIM-barrel (Pfam PF00682; CDD cd07938 DRE_TIM_HMGL). Two orthogonal diagnostic-motif tests place it outside the HMG-CoA-lyase specificity group:

Despite these specificity-determining differences, PP_3394 retains the catalytic core of the fold: it aligns to the human HMGCL catalytic/metal-binding residues Arg41, Asp42, His233, His235 and the catalytic Cys266. It is therefore very likely catalytically competent, but the sequence does not determine which C–C bond it makes or breaks. The honest interpretation is: divergent DRE-TIM metallolyase, specificity undetermined by sequence — precisely the "unresolved specificity" the seed hypothesis asked us to report honestly rather than paper over.

Finding 3 — The P. putida leucine operon lacks its own HMG-CoA lyase, and KEGG dual-annotates PP_3394 and PP_3540 to K01640 (paralog over-annotation)

Genomic-context analysis (KEGG) shows the P. putida KT2440 leucine/isovalerate catabolic operon is PP_4063–PP_4068:

PP_4063  long-chain acyl-CoA ligase
PP_4064  ivd   isovaleryl-CoA dehydrogenase   (K00253)
PP_4065  mccB  methylcrotonyl-CoA carboxylase (K01969)
PP_4066  liuC  methylglutaconyl-CoA hydratase (K13766)
PP_4067  mccA  methylcrotonyl-CoA carboxylase (K01968)
   ── NO adjacent HMG-CoA-lyase gene in the operon ──

Neither HMGL-family gene sits in this operon. Instead they are dispersed:

KEGG maps BOTH PP_3394 and PP_3540 to K01640 (EC 4.1.3.4, HMG-CoA lyase) and to leucine-degradation module M00036 — an unresolved best-KO duplication. This is the classic mechanism by which an over-annotation propagates: a single KO is assigned to two paralogs, and the divergent one (PP_3394) inherits the confident functional label from the true ortholog (PP_3540) without independent support. PP_3394's neighborhood provides no pathway-context corroboration for a leucine-catabolic role.

Finding 4 — Whole-sequence phylogeny places PP_3394 basal to the HMG-CoA-lyase clade, not among condensation synthases

A Neighbor-Joining tree over a 9-member DRE-TIM panel (p-distance = 1 − global NW/BLOSUM62 identity) recovers two well-separated clades:

Newick (branch lengths):
((PP3394:0.307,(BsHMGL:0.267,(hHMGCL:0.220,(mvaB:0.113,PaLiuE:0.101):0.092):0.066):0.041):0.015,
 (condensation clade B):0.015);

PP_3394 branches at the base of clade A. Its terminal branch (0.307) is ~3× longer than mvaB's (0.113), quantifying how divergent it is even within the HMG-CoA-lyase clade. Two nuances matter for curation: (i) whole-sequence phylogeny groups PP_3394 nearer the lyase clade, whereas the motif/active-site signatures (Finding 2) pull it toward the IPMS/condensation specificity determinants — this tension is itself the reason specificity is unresolved; and (ii) basal, long-branch placement is exactly the topology of a paralog that has functionally diverged and should not simply inherit the clade's canonical annotation.


Mechanistic Model / Interpretation

The DRE-TIM metallolyase (HMGL) superfamily uses a divalent-metal-dependent TIM-barrel active site to catalyze either retro-aldol / oxo-acid cleavage (HMG-CoA lyase; 4-hydroxy-2-oxovalerate aldolase) or Claisen-type condensation (isopropylmalate synthase, homocitrate synthase, citramalate synthase). A shared catalytic core (metal-binding Asp/His pair, catalytic Arg and Cys) supports both chemistries; substrate specificity and reaction direction are set by peripheral residues and short signature motifs, not by the core. This is why sequence identity alone is a poor discriminator here — and why the seed hypothesis was right to insist that "pairwise similarity alone does not settle the paralog."

Placing the P. putida paralogs onto this framework:

          DRE-TIM / HMGL-fold superfamily
                     │
┌────────────────────────────┴─────────────────────────────┐
   HMG-CoA lyase specificity                          condensation-synthase specificity
   (PS01062 SVAGLGGCPY; PRDG motif)                   (PS00815; LRDG motif)
│                                                        │
   ┌────┴──────────────┐                                         │
 mvaB/PP_3540        PaLiuE/PA2011  ← EXPERIMENTALLY VERIFIED     │
 (78.6% id to LiuE,  (Km 100 µM,    HMG-CoA lyase                 │
  PS01062 +)          Vmax 21)                                    │
│                                                        │
   leucine + terpene catabolism (real)                           │
                                                 │
   PP_3394 ── retains catalytic core, BUT fails PS01062,         │
     matches PS00815, carries LRDG motif,                │
     ~38% id to everything, long basal branch ───────────┘
     ⇒ specificity UNRESOLVED (could be an HMG-CoA-like,
       other acyl-CoA, or oxo-acid substrate)

Adjudicating the two pathway claims separately, as the seed hypothesis requested:

The coherent model: PP_3394 is a functionally diverged HMGL-fold paralog whose true substrate is undetermined; the confident HMG-CoA-lyase label it currently carries is a carry-over from its better-characterized paralog mvaB via a shared KEGG KO.


Evidence Base / Evidence Matrix

Citation Evidence type Supports / refutes / qualifies Claim tested Key finding Context Confidence & limitations
PMID: 19459965 Direct assay + mutant phenotype Supports (anchors the ortholog to mvaB, not PP_3394) Is LiuE the leucine/terpene HMG-CoA lyase? LiuE (PA2011) cleaves HMG-CoA→acetyl-CoA+acetoacetate; liuE mutant can't use leucine/isovalerate or acyclic terpenes P. aeruginosa, in vitro enzymology + growth phenotype High for LiuE; applies to PP_3394 only via orthology to mvaB (78.6%), not to PP_3394 (38.2%)
Computed (global NW/BLOSUM62 pairwise identity) Structural/evolutionary (computational) Qualifies/refutes confident PP_3394 assignment Is PP_3394 a LiuE ortholog? mvaB=78.6% id (ortholog); PP_3394=38.2% id, no better than to human HMGCL (38.0%) or E. coli IPMS (38.8%) 9-protein reference panel High as computed; sequence identity cannot prove specificity, only rank relatedness
Computed (PROSITE PS01062 / PS00815 + N-terminal motif) Structural/evolutionary (computational) Refutes "canonical HMG-CoA lyase" for PP_3394 Does PP_3394 carry HMG-CoA-lyase-specific signatures? Fails PS01062 (all 3 real lyases pass); matches IPMS site-1 PS00815; carries LRDG (not PRDG) motif; but retains catalytic core (R41,D42,H233,H235,C266 equivalents) Same panel High that PP_3394 is atypical; motif tests indicate but do not prove alternative specificity
Computed (KEGG genomic context) Review/database (computational) Qualifies (over-annotation signal) Does pathway context support PP_3394's role? liu operon PP_4063–4068 has no HMGL gene; KEGG maps BOTH PP_3394 & PP_3540 to K01640 & module M00036 KT2440 genome High for the dual-annotation fact; database-level, not experimental
Computed (Neighbor-Joining phylogeny) Structural/evolutionary (computational) Qualifies Where does PP_3394 sit phylogenetically? Basal in HMG-CoA-lyase clade A; branch 0.307 ≈ 3× mvaB's 0.113 9-protein panel Moderate–high; NJ on p-distance from a small panel; no bootstrap reported

How the key paper supports/challenges the findings. The Pseudomonas aeruginosa liuE gene encodes the 3-hydroxy-3-methylglutaryl coenzyme A lyase, involved in leucine and acyclic terpene catabolism (PMID: 19459965) is the only primary experimental anchor available and it supports the findings by orthology transfer to mvaB, not to PP_3394. It establishes LiuE (PA2011) as a verified HMG-CoA lyase whose loss blocks leucine/isovalerate and terpene growth; the P. putida protein that inherits this by 78.6% identity and by carrying PS01062 is mvaB/PP_3540. The paper therefore challenges the idea that PP_3394 (a 38.2%-identity paralog lacking PS01062) is the physiological leucine-catabolic lyase.


GO Curation Implications

Lead (requires curator verification): The current confident annotation of PP_3394 as an HMG-CoA lyase acting in leucine catabolism is over-annotated and should be generalized/qualified, while the leucine-catabolic HMG-CoA-lyase role is retained/strengthened on mvaB / PP_3540.


Mechanistic Scope

The molecular function under test is a single, direct enzymatic activity: a DRE-TIM metallolyase (HMGL-fold) reaction — either retro-aldol cleavage of an oxo-acid/acyl-CoA (as in HMG-CoA lyase) or a Claisen condensation (as in IPMS/homocitrate synthase). The catalytic core is present in PP_3394, so a direct metal-dependent lyase/synthase activity is plausible. Everything downstream — participation in leucine catabolism, terpene degradation, "ketone-body" chemistry — is pathway consequence, and for PP_3394 none of it is established by direct evidence. The experimental phenotype (loss of leucine/terpene growth) belongs to the P. aeruginosa LiuE mutant, i.e., the mvaB ortholog, and must not be transferred to PP_3394 by inference from loss of function in a different gene.


Conflicts and Alternatives


Limitations and Knowledge Gaps

  1. No direct biochemistry for PP_3394. All discriminating evidence is computational (pairwise identity, PROSITE/motif scan, NJ tree). Checked: sequence, signatures, genomic context, phylogeny. Why it matters: specificity in this fold is set by peripheral residues; only an assay settles it. Resolver: purify recombinant PP_3394 and assay HMG-CoA lyase activity vs. candidate condensation/other-acyl-CoA reactions.
  2. No mutant phenotype for PP_3394. Why it matters: a leucine/terpene growth defect (or its absence) in a ΔPP_3394 strain would directly test pathway membership. Resolver: clean deletion + growth on leucine/isovalerate/acyclic terpenes, with ΔPP_3540 as the positive control.
  3. Phylogeny lacks bootstrap support and uses a small 9-taxon panel. Why it matters: basal placement confidence is unquantified. Resolver: broader taxon sampling with ML + bootstrap.
  4. No experimental structure. Why it matters: an AlphaFold model + active-site pocket analysis (substrate-channel residues, metal site geometry) could sharpen the lyase-vs-synthase call. Resolver: model/compare active-site pocket against solved HMGCL and IPMS structures.
  5. Substrate identity of PP_3394 entirely open. No candidate has been experimentally excluded. Resolver: metabolomic screen / activity-based substrate profiling.

Discriminating Tests

Ranked by efficiency in separating "HMG-CoA lyase" from "divergent-specificity metallolyase":

  1. In vitro enzyme assay (definitive). Recombinant His-tagged PP_3394 + Mg²⁺/Mn²⁺; test HMG-CoA cleavage (spectrophotometric acetoacetate/DTNB acetyl-CoA assays, matched to the LiuE assay of PMID 19459965) alongside an IPMS/homocitrate-synthase condensation assay panel. Compare kinetics against mvaB/PP_3540 run in parallel.
  2. Reciprocal deletion phenotyping. ΔPP_3394, ΔPP_3540, and the double mutant; growth on leucine, isovalerate, and acyclic terpenes (citronellol/geraniol). Prediction: ΔPP_3540 loses growth; ΔPP_3394 does not (if PP_3394 is off-pathway).
  3. Structure-guided active-site comparison. AlphaFold model of PP_3394 → superpose on HMGCL and IPMS; score the PS01062-equivalent loop and metal site. Cheap, immediate, and directly interrogates the Finding-2 tension.
  4. Broadened, bootstrapped phylogeny + HMM specificity profiling to test whether PP_3394 falls in a distinct, uncharacterized subfamily.
  5. Coexpression / operon transcriptomics under leucine vs. alternative carbon sources to test pathway coregulation of PP_3394 vs. the liu operon and mvaB.

Proposed Follow-up Actions / Curation Leads (require curator verification)


Bottom Line

PP_3394 (Q88HG4) is a real HMGL-fold / DRE-TIM metallolyase with an intact catalytic core, but it is a highly divergent paralog (~38% identity to all references) that fails the HMG-CoA-lyase active-site signature PS01062 and instead matches an IPMS/homocitrate-synthase motif — so its exact reaction is unresolved by sequence and its confident annotation as an HMG-CoA lyase in leucine catabolism is over-annotated. The physiological leucine/terpene HMG-CoA lyase of P. putida KT2440 is the paralog mvaB / PP_3540 (78.6% identical to the experimentally verified LiuE, PMID 19459965). "Ketone-body biosynthesis" does not apply to this bacterium. Recommended curation: generalize/qualify PP_3394's MF as a low-confidence oxo-acid/carbon-carbon lyase, treat leucine catabolism as non-core for PP_3394, and require a biochemical assay or mutant to resolve specificity.

Artifacts