liuE (PA2011) review notes

UniProt: Q9I2A0 (Swiss-Prot, 300 aa),
"3-hydroxy-3-isohexenylglutaryl-CoA/hydroxy-methylglutaryl-CoA lyase". Locus
PA2011, in the liuRABCDE cluster.

What is established

A liuE mutant cannot grow on leucine/isovalerate or on acyclic terpenes, and
accumulates intermediates of the acyclic terpene pathway
PMID:19459965.
Purified His-tagged LiuE has HMG-CoA lyase activity with measured kinetics
PMID:19459965, and the conclusion is stated directly
PMID:19459965.
The same purified protein also has HIHG-CoA lyase activity
(3-hydroxy-3-isohexenylglutaryl-CoA lyase, EC 4.1.3.26), the corresponding
step of the acyclic terpene (citronellol) pathway
PMID:19597963. These are the two EXP rows in GOA, assigned by UniProt.
The earlier cluster-level paper only inferred the activity
PMID:16517656; it also notes that the atu cluster lacks an HMG-CoA
lyase homologue, which is why liuE serves both pathways.

LiuE therefore performs the terminal cleavage step of both pathways itself:
HMG-CoA to acetyl-CoA plus acetoacetate for leucine/isovalerate, and the
analogous HIHG-CoA cleavage for citronellol-family terpenes. Both process
annotations are participation, not necessity.

The GO:0046951 (ketone body biosynthetic process) IBA

This row is not PseudoCAP-seeded: the GO:0046951 IBD at PTN000031222 is seeded
by mammalian and insect HMGCL/HMGCLL1 only (P35914, P35915, Q8TB92, RGD:620554,
FB:FBgn0031877). In vertebrate liver, HMG-CoA lyase catalyses the last step of
ketogenesis, and "ketone body biosynthetic process" names that physiological
programme. In P. aeruginosa the same chemistry is the terminal step of amino
acid and terpene catabolism; acetoacetate is produced as a catabolic
intermediate feeding central metabolism, not synthesised as a circulating fuel.
Propagating the term to the bacterial members of this node reads a
lineage-specific physiological process off shared chemistry, so the leaf row is
marked over-annotated with a propagation_review, and a GO_CENTRAL defect
asks PAINT to restrict that IBD to the animal clade. Confidence PROBABLE: the
chemistry and the seed list are checkable facts, but where exactly the node
should be cut is a judgement for the PAINT curator who placed it.

Literature search

Europe PMC, 2026-11-05: aeruginosa AND liuE. Cached and used: PMID:19459965,
PMID:19597963, PMID:16517656. Also returned PMID:16820476 (genes and proteins
needed for acyclic terpene and leucine catabolism) and PMID:26162879 (AtuE
hydratase), consistent with the above but not needed for the calls.

PseudoCAP IBA-seed assessment

PTN node	node term	n_seeds	leaf IBAs	seed annotation (term, evidence, PMID)	verdict	rationale	other seeds	IBD stands without this seed?
PTN000031222	GO:0006552 L-leucine catabolic process	3	183 (100 taxa)	GO:0006552, IMP, PMID:19459965 (PseudoCAP)	SOUND	The paper combines the growth phenotype with purified-enzyme kinetics in the same study, and HMG-CoA lyase catalyses the terminal step of leucine catabolism, so the gene product does part of the work of the process.	Mouse Hmgcl (MGI:96158) and Aspergillus Q5B2F7	Yes, but with only two other seeds it is the thinnest of the node's IBDs; mouse Hmgcl carries the leucine-catabolism evidence independently.
PTN000031222	GO:0004419 hydroxymethylglutaryl-CoA lyase activity	7	183 (100 taxa)	GO:0004419, IMP, PMID:16517656 (PseudoCAP)	INDIRECT-PHENOTYPE	A molecular function coded IMP from a cluster-mutagenesis study that itself calls the assignment "probable" and sequence-based. The claim is right, and this gene's two UniProt EXP rows (PMID:19597963, PMID:19459965) are the proper evidence, so only the coding of this row is at issue.	Human/rat/mouse HMGCL (P35914, P35915, MGI:96158, RGD:620554), human HMGCLL1 (Q8TB92), Arabidopsis AT2G26800, and Q9I2A0's own EXP-backed annotation	Yes - the mammalian HMGCL enzymes are directly characterised.

PseudoCAP defects from this gene: one WRONG_EVIDENCE_CODE (PROBABLE) on the
GO:0004419 IMP row. GO_CENTRAL defects: one (PROBABLE), on the GO:0046951 IBD
at PTN000031222, as described above.