Bacterial trans-4-hydroxy-L-proline catabolism to 2-oxoglutarate

A reusable bacterial four-reaction route that converts trans-4-hydroxy-L-proline, the major imino acid of host and plant collagen and cell-wall protein, to 2-oxoglutarate. A PLP-independent epimerase first forms cis-4-hydroxy-D-proline. An FAD-dependent oxidoreductase generates 4-hydroxy-1-pyrroline-2-carboxylate, a deaminase converts that cyclic imine to 2,5-dioxopentanoate and ammonium, and an aldehyde dehydrogenase-family enzyme completes oxidation to 2-oxoglutarate. The route is distributed across proteobacteria, firmicutes and actinobacteria, and is unrelated to the mammalian hydroxyproline pathway, which proceeds through hydroxyproline dehydrogenase and the HOGA1 aldolase to glyoxylate and pyruvate.

The boundary starts with the stereochemical entry reaction and ends at 2-oxoglutarate. Hydroxyproline transport (the LhpPMNO-type ABC systems), LhpR-mediated induction, the separate trans-3-hydroxy-L-proline branch, and downstream use of 2-oxoglutarate are outside the module. No generic cellular location is asserted: the enzymes are soluble bacterial proteins, but only PP_1255 has a locus-level cytoplasmic call, and that is an electronic inference. Species scope. The module is written species-neutrally. KT2440 supplies the only organism with all four steps grounded, and is one exemplar rather than the subject. P. aeruginosa PAO1 grounds the epimerase step and the second oxidase architecture. The terminal reaction has two exemplars: A. brasilense LhpG, which is the only characterized terminal dehydrogenase of this pathway (purified from hydroxy-L-proline-grown cells and hydroxyproline-inducible, as against the aldarate-inducible KGSADH-II of the same organism), and A. baylyi KgsD, which performs the same reaction in the D-glucarate pathway that converges on the same 2,5-dioxopentanoate intermediate. Cofactor. The terminal annoton is typed with the NADP+ term because the characterized pathway enzyme prefers NADP+, but the reaction is not cofactor-specific: LhpG carries both reactions with a roughly 75-fold lower Km for NADP+, its aldarate paralog prefers NAD+, and GO splits the reaction into an NADP+ and an NAD+ term with no substrate-specific neutral parent. An unassayed member is therefore reviewed against the proposed neutral term (see the PP_1256 review) rather than against either child. Enzyme-family grounding. Both PANTHER families whose members carry these steps are functionally mixed, so the module grounds those nodes on exemplars and InterPro signatures instead of PANTHER terms. PTHR33442 mixes 4-hydroxyproline epimerases with trans-3-hydroxy-L-proline dehydratases in both of its subfamilies, and the epimerase activity arose more than once in that superfamily, so IPR008794 (the fold) is asserted and the activity is established per protein. PTHR13847:SF289 mixes the KT2440 oxidase with ThiO glycine oxidases and HcnC. PTHR43353:SF3 does contain both grounded terminal dehydrogenases, but it also holds aldehyde dehydrogenases with other substrates, so the narrower IPR044151 (ALDH_KGSADH) is used. Only the deaminase step carries a PANTHER subfamily term, PTHR12128:SF72, verified against panther.obo and the member index; that assertion comes from open maintainer PR #5076 and is retained here. PTHR43353:SF3 is the one case where the subfamily does track the reaction: five of its six Swiss-Prot members are KGSADHs, so its family review scopes the activity to SF3 with the one fatty aldehyde dehydrogenase recorded as a member exception. The module still grounds on IPR044151, because SF3 membership does not say which route a member serves. All four families now have FamilyReview documents recording this divergence. The oxidase step is a variant set rather than a single enzyme, because the two characterized enzymes are of independent origin: a single-chain FAD protein in P. putida and a multi-subunit FAD/FMN/[2Fe-2S] enzyme in P. aeruginosa. The subunit accessions of the P. aeruginosa enzyme are not asserted: the cached abstract names the operon as lhpABFE without mapping subunits to locus tags, and the full text was not available, so that variant is deliberately left ungrounded rather than filled with a guess. This produces a leaf-grounding warning, which is a real curation gap and not a modelling choice. The LhpBEF gap is narrow and the next pass should not re-derive it. What is settled, from the cited literature: the operon is lhpABFE and lhpA is PA1268, so the three subunits should be among the genes adjacent to PA1268. What blocks grounding: a UniProt query for lhpB, lhpE and lhpF in P. aeruginosa PAO1 returns no gene with any of those names, and the adjacent loci PA1267 and PA1266 carry only uninformative protein names with no function comment, leaving a three-subunit enzyme with at most two candidates. Accessions are deliberately not written down here, because nothing cached in this repository records the locus-to-accession mapping for them; the next pass should fetch the adjacent loci and ground them only if UniProt names the subunits explicitly with evidence, or if the full text of the characterization gives locus tags. Note also that the neighbouring lhpD/lhpH locus is the Pyr2C reductase of the 3-hydroxyproline branch, which is outside this module. Similarly, the single-chain variant's grounding is an inference about which locus the characterized P. putida enzyme is: the gene review records those rows as IGC on the strength of the gene neighbourhood, not as a direct assay of Q88NF6. The oxidase step has no exact GO molecular-function term, so GO:0016645 is used as the closest donor-class parent while the substrate-specific activity stays explicit in the descriptor and in the proposed term in the PP_1255 review. Two validator warnings are expected and accepted, and both record real states rather than defects in this module. The function-conformance check reports FAMILY_FUNCTION_UNREVIEWED for the deaminase, because the PTHR12128 review leaves the subfamily-level scope of GO:0047425 UNRESOLVED: the activity is established for PP_1257 itself, but SF72 has no other assayed member, so claiming it subfamily-wide would overstate the evidence. It also reports REVIEW_DISAGREEMENT for Q88NF5 on the terminal annoton's NADP+ term, and that warning is doing its job. PP_1256's review moves every row off GO:0047533 to a proposed substrate-specific, cofactor-neutral term, because the protein is unassayed and the characterized members of its subfamily use both cofactors. The module nevertheless types the step with the NADP+ term, because the two characterized exemplars, LhpG and KgsD, carry it on their own measurements; the disagreement is about PP_1256's cofactor, not about the reaction the step performs. When the proposed term is created, the annoton and that review should be revisited together. Page QC additionally lists the multi-subunit oxidase variant as a leaf without representative members, for the reason given above. The NCBITaxon label for the module's taxonomic context is reported as unvalidated wherever OLS cannot be reached, as it is for every module that asserts a taxon.