This augmentation preserves the 84 historical annotation source objects, including the prior glycine-cleavage proposal, and incorporates five genuine additional GOA rows. One separately proposed endopeptidase annotation brings the total to 90. The three official natural products remain distinct from processed chains, engineered fragments and disease variants. Primary source access, rather than a cache’s nominal full-text flag alone, determines the claims below.
The main activity regenerates oxidized lipoyl cofactors on E2 or GCSH proteins using FAD and NAD+. The pyruvate, 2-oxoglutarate and branched-chain ketoacid process definitions include all three coupled reactions, including DLD’s regeneration step. DLD does not itself perform the E1 decarboxylation. Human biochemical evidence includes reconstitution and activity measurements, mutant E3 assays and calorimetry and E3/E3BP structure. These support one shared redox unit across enzyme systems.
The OADH paper remains abstract-only: it explicitly reports recombinant human DHTKD1 recruitment of DLST and DLD. Its E1 substrate preference and enamine-associated oxidant chemistry are not DLD-specific properties. The human-liver BCKDH paper also remains abstract-only; it reports dissociable E3 and recovery of overall complex activity upon adding it. Existing well-established target assertions are retained without inventing inaccessible Methods.
The prior glycine-cleavage proposal remains TAS-level evidence. In reaction R-HSA-5694018, catalyst 5694028 identifies active unit 5694020, the DLD dimer with FAD. Its DLD component 69978 resolves to P09622 residues 36–509 in matrix. DLD performs lipoyl regeneration rather than merely being consumed. Same-role LPD1 and LpdA annotations support this convention; checked ancestry does not duplicate another retained process. No target occurrence was found in the checked GO-CAM index. The local glycine module cites this review and is not independent evidence.
The new endopeptidase annotation describes conditional cleavage, supported by human recombinant WT/mutant frataxin-substrate experiments and independently by human DLD preparations cleaving human IκBε. The former was read through indexed original Results, figure legends and Methods despite failed article downloads; its protected cache is still abstract-only. The latter cache omits critical Results/Methods, which were recovered separately. Urea/refolding, reaction conditions and reported WT/mutant preparation limits are retained. Mouse N9/brain stress experiments are not direct normal human microglial physiology. No additional NF-κB process or protease location is proposed. The retained NEW molecular-function assertion records the conditional biochemical activity; it is not a second core function. Native human deployment and substrates are recorded as a BIOLOGY knowledge gap, recognizing the existing IκBε experiments rather than claiming that no candidate cellular substrate is known.
The current serine-type peptidase definitions require a catalytic triad. The first paper instead supports a serine/glutamate dyad, with H450A increasing activity. Therefore GO:0004175 is used without deleting the biochemical serine dependence. Mature-protein E431/D444/H450/S456 correspond to precursor E466/D479/H485/S491; the later paper’s reported S456A name is not independently remapped to its clone. The engineered 126-residue fragment is not natural isoform 2.
Official UniProt P09622 and independent official sequence reconstruction give products of 509, 410 and 461 residues. Isoform 2 deletes residues 1–99; isoform 3 deletes 147–194. Canonical precursor processing removes 1–35. Sequence changes alone do not establish the alternative products’ activity or localization.
Mitochondrion annotations are refined at the gene level to the established matrix pool, while each source’s assay resolution remains explicit. An organelle proteomics experiment is not described as a matrix-resolved assay. In particular, the human-muscle phosphoproteomics cache is marked full text but lacks detailed Methods and target supplemental assignments. The matrix refinement uses independent evidence. The human mitochondrial proteome study provides methodological context, without a new target-peptide or phosphosite reanalysis.
The nuclear OGDH study directly detects a small DLD pool in human HEK293/U251 nuclei. Its OGDH-dependent association with KAT2A supplies succinyl-CoA; KAT2A performs histone succinyl transfer. The current HPA page reports supported nucleoplasm staining and a principal mitochondrial pool. These secondary compartments remain compatible with the matrix refinement.
The sperm study provides positive homolog evidence for the four contextual localizations: hamster acrosomal and flagellar principal-piece staining with antigen-preabsorption controls (Fig. 2), mouse acrosomal staining (Fig. 3A/B), and hamster detergent/shearing/exocytosis experiments resolving acrosomal matrix (Fig. 4). This original was read through an author-deposited indexed copy in addition to the newly fetched abstract cache. The mouse donor GOA labels extend beyond the explicitly mouse assay resolution; the hamster experiments supply the corresponding compartment evidence. Human anti-E3 staining failed, and anti-whole-PDH positivity is not DLD-specific. HPA's human principal-piece and primary-cilium staining both have uncertain reliability. These are separate observations: rodent motile sperm flagella do not establish human primary-cilium localization. No sperm process or universal natural-product localization is inferred.
The project’s explicit binding instruction supports retaining positive generic interactions when no narrower molecular-function claim is established. This supersedes the earlier notes’ genericity-only removal rationale. Different partner identities and assay scopes are assessed separately, not merely different evidence codes.
| Source | Exact target evidence | Decision scope |
|---|---|---|
| 16263718 and 16442803 | Human purified E3/E3BP structures and biochemical binding; original Methods/Results recovered separately from abstract caches | Retain PDHX binding. PDHX is a noncatalytic tether; it is not an enzyme-binding partner merely because it resides in PDC. |
| BioPlex 2 and BioPlex 3 | Exact DLD–PDHX AP-MS records EBI-21660882 and EBI-54592518; genuine original Methods read | Retain co-complex association, with independent structural corroboration. These AP-MS records alone do not prove direct recognition. |
| 29128334 | PRDX6 and YWHAE crosslink/association records plus nonexpanded anti-bait co-IP EBI-20304023 and EBI-20304029 | Retain both associations; partner enzymatic/adaptor roles do not make lysate co-IP a direct molecular-class recognition assay. |
| HuRI | PDHX validated binary record EBI-24759808 and ITGB1BP1 record EBI-24650155, with corresponding array/prey-pool records | Both pairs remain positive generic non-core interactions. ITGB1BP1’s established partner class does not alone establish a physiological cytoskeletal-regulatory binding function for DLD. Matrix versus cytosolic focal-adhesion contexts leave native encounter unresolved and do not invalidate the yeast assay. |
| 32814053 | Validated binary record EBI-26272042 explicitly uses HTT residues 1–88; institutional original Methods recovered | Retain fragment-specific binding, without claiming native full-length HTT transport or aggregation biology. |
HuRI’s main original describes quadruplicate retesting and sequence confirmation. Dataset-level orthogonal validation is not represented as an additional assay on each DLD pair. BioPlex Methods describe tagged-bait affinity purification and technical replication; group recovery is not a measured pairwise affinity or stoichiometry.
The protected PAINT assertions at PTN000530299 and PTN000530301 precede the human target PTN002505536 on the actual PTHR22912 tree path. Donor count is not treated as evidence quality, and human DLD appearing among descendant sources is not circularity. The broad metabolic assertion is refined by term scope, not rejected as a misplaced ancestral node. InterPro target matches and actual interpro2go mappings support the redox/FAD assertions; broad redox categories are refined to GO:0004148.
All 22 cited Reactome events resolve through typed entity paths to human DLD in matrix. Representative distinctions are PDH regulation, OGDH E1 chemistry, DLD chemistry, and CLPXP substrate degradation. An enclosing catalytic complex or substrate set does not assign its reaction to every component. Mutant DBT/BCKDK/PPM1K events support matrix context, not a DLD mutation or loss of function. Historical machine snapshots are preserved. The current GOA file is the unedited normal-fetch output, with its exact predecessor archived as described below; newly missing caches were obtained through the normal cache path.
Abstract-only access remains for 14638692, 15712224, 16770810, 24534072, 29191460, 3593587, 8506365 and 9242632. Their explicit positive human biochemical context supports retained established claims with curator deference. The 3278312 full-text section repeats its abstract. Genuine separately recovered originals supply 15888450, 16263718, 16442803, 17404228, 32814053 and the missing 29128334/33961781 Methods/Results. No failed download, challenge page or guessed publisher route is treated as an original.
The old notes remain an exact prefix. New text supersedes the old generic-removal, broad-term and protease assessments while preserving their historical record. Short source anchors occur once in the YAML, with reference-only reuse elsewhere; no additional quotations are added to this guide or the new notes. The original long Reactome quotation in the prior notes is inherited and is not repeated or expanded. Provider text and all prior caches/history are preserved; the provider attempt returned the pre-existing research body, which is not presented as newly generated evidence.
The genuine refreshed GOA has 85 records. Historical rows 22/25/26 retain GO_REF:0000120 and all original source fields but carry retired: true, meaning replaced by the current GO_REF:0000107 rows 86/87/88. Their ACCEPT verdicts remain scientifically valid. The exact prior snapshot and retrieval provenance make this transition inspectable. No current and historical TSV rows were concatenated. The separate 25 supporting-entity enrichments remain provenance; they do not overwrite old source dictionaries. All 85 current records are covered, alongside the three retained historical source records and the two authored proposals.