LSC2 (YGR244C, P53312) notes

• Beta subunit of succinyl-CoA ligase; ATP-specific enzyme PMID:9874242.
• Beta provides nucleotide specificity and succinate binding, Mg2+ cofactor (By similarity) [UniProt:P53312].
• Mitochondrial [PMID:9874242; UniProt:P53312 HTP mito proteome PMID:14576278].
• ARBA "cytoplasm" IEA marked over-annotated (protein is mitochondrial). Corrected 2026-11-12: the action is now TOO_GENERAL, and "bacterial default" was wrong; the condition set of ARBA00026971 that fires is the eukaryotic A-type subfamily PTHR11815:SF1 (see below).

2026-11-11 re-review from an evolutionary perspective

Scope and method as in the LSC1 notes; the shared analysis is in LSC2-bioinformatics/ (script, results.yaml, RESULTS.md).

• Lsc2 groups with animal SUCLA2 in the PANTHER tree, a bootstrapped neighbor-joining tree and a maximum-likelihood tree [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "LSC2 belongs to the A-type lineage by descent: it groups with animal SUCLA2 in the PANTHER tree, the neighbor-joining tree and the maximum-likelihood tree."]. Lsc2 is best read as the SUCLA2 ortholog and a paralog of SUCLG2 (quote updated 2026-11-12 to the rewritten RESULTS.md).
• The duplication is older than proposed in 1998 PMID:9765291; Dictyostelium and the choanoflagellate Monosiga each have one gene in each clade.
• Gln at position 20 of the mature beta subunit contacts guanine in the pig GTP-specific enzyme (corrected 2026-11-12: this line used to call it a marker of GDP specificity; it is not diagnostic, see below) PMID:16481318. Lsc2 has Pro at the aligned position (Pro50 of the precursor). Pro at this position does not exclude GTP use PMID:22751660. Recorded as a residue site (strength CONTRIBUTES) in the PTHR11815 family review.
• Action changes: GO:0042709 succinate-CoA ligase complex IBA, ACCEPT to TOO_GENERAL (the ADP-forming child GO:0009361 is asserted; the root node correctly carries the parent because it has ADP- and GDP-forming descendants). GO:0005737 cytoplasm ARBA row, MARK_AS_OVER_ANNOTATED to TOO_GENERAL (true, entailed by mitochondrion). Both now match the S. pombe ortholog SPCC1620.08.
• PAINT handles the beta tree correctly: ADP-forming at the root, with a loss (IKR) and a GDP-forming gain at PTN000929335, the node that holds the characterised GTP-specific enzymes. Corrected 2026-11-12: this line used to say that the node is exactly the clade that carries Gln20. It contains all Gln20 subunits of the first panel plus two misplaced leaves (Xanthomonas SucC Q8P676, with Pro20, and the Ustilago protein A0A0D1E3R8, which has no ATP-grasp domain), and Gln20 also occurs outside it. Lsc2 is outside that node. The alpha tree has no equivalent, which is the cause of the wrong GDP-forming IBA on Lsc1.
• Not settled by the trees: which copy plants, stramenopiles and excavates descend from. Corrected 2026-11-12: PANTHER puts them in the G daughter and both sequence trees put plants, stramenopiles and trypanosomatids with SUCLA2; the Ustilago maydis protein A0A0D1E3R8 is placed with prokaryotic sequences by both sequence trees, so its position is not unstable between methods as first written. Neither affects an annotation on Lsc2.
• OpenScientist run (LSC2-hypotheses/beta-nucleotide-specificity-phylogeny): supports ADP-forming by assay, residue and phylogeny; could not resolve whether Lsc2 is sister to SUCLA2 with a k-mer distance tree of 18 sequences and recommended a maximum-likelihood tree with broader sampling, which was then run locally (clade of fungal subunits and animal SUCLA2, ultrafast bootstrap 90 in the first run; after the 2026-11-12 rerun with 64 sequences the clade of 11 fungal and 5 animal subunits has ultrafast bootstrap 100 in three runs and is in 120 of 200 neighbor-joining bootstrap trees).
• Family-level observations recorded in interpro/panther/PTHR11815/PTHR11815-review.yaml: coherence changed to MOSTLY_COHERENT (ATP-citrate synthase, malate-CoA ligase and sulfoacetate-CoA ligase subunits are InterPro members of SF10); the mitochondrion IBD at the eukaryotic root needs pruning for the hydrogenosomal Trichomonas vaginalis subunits.
• Mitochondrial matrix was not added as a NEW row on 2026-11-11, on the ground that no sub-mitochondrial assay of Lsc2 had been found and the transit peptide is rule-based. Corrected 2026-11-12: both parts of that ground were wrong. A sub-mitochondrial proteomic study lists Lsc2 as a soluble matrix protein (PMID:28819139), and projects/ORGANELLES.md rule 4 counts a cleavable presequence and a matrix TCA function as lines of evidence without requiring an assay. The row is now added; see the section below.

Retrospective audit corrections (2026-11-12)

A retrospective audit of the succinyl-CoA synthetase reviews found errors in this review. The analysis in LSC2-bioinformatics/ was rerun with a 64-sequence panel, a scan of all PANTHER leaves and a UniProt-wide census, and RESULTS.md was rewritten. Changes here:

• Fungi and the G-type beta gene. "Fungi kept only the A-type copy" was a generalisation from the 15 fungal genomes in PANTHER [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "The 15 fungal genomes in the PANTHER tree (14 Dikarya and the chytrid Batrachochytrium dendrobatidis) have an A-type beta gene and no G-type beta gene."]. Early-diverging fungi keep both [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "Early-diverging fungi (Mortierellomycota, Basidiobolus, Rozella) keep a SUCLG2-type beta gene with Gln at position 20, so the G-type beta gene was lost within fungi and not in the common ancestor of all fungi."]. The description, the findings and the suggested questions were reworded. S. cerevisiae itself has one beta gene, so no action changes.
• Gln20 is not a marker of GDP specificity. It is found in the characterised GTP-specific enzymes but also in the ATP-specific Blastocystis enzyme PMID:18452512, in ciliates and in some bacterial SucC [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "Gln at position 20 is not sufficient for GTP specificity, and is not required for GTP use or preference."]. The finding "GDP specificity is a derived state confined to the animal, choanoflagellate and Dictyostelium members of the G clade" was removed. Lsc2 has Pro20, the commonest residue of the family, which by itself predicts nothing; the ATP specificity of the yeast enzyme rests on the assay.
• "ADP specificity is the ancestral state" was removed from the core function. Bacterial Pro20 enzymes use both nucleotides, so what Lsc2 shares with them is the residue and ATP use [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "This analysis does not show which nucleotide specificity is ancestral."].
• Dating of the duplication. The description now says that gene trees place it before the divergence of animals, fungi and amoebozoa. The PMID:9765291 finding is still DISPUTED, now with published support beside the local tree PMID:18452512. "Genome-scale trees" was wrong wording for a single-gene tree and was replaced.
• Tree statements. PANTHER and the two sequence trees differ on plants, stramenopiles and trypanosomatids; the two sequence trees agree with each other. The Ustilago protein A0A0D1E3R8 (UMAG_01904) lacks the ATP-grasp domain, so no residue can be read at position 20 (the earlier "Ala" was an alignment artefact), and both sequence trees place it with prokaryotic sequences. Its neighbour UMAG_01903 (A0A0D1CWH6) is a divergent alpha-like protein. The suggested questions were rewritten to match.
• New row: GO:0005759 mitochondrial matrix (IC, NEW), and matrix added to core_functions.locations beside mitochondrion. Lines of evidence: (1) Voegtle et al. 2017 PMID:28819139; Supplementary Data 5 (publisher's spreadsheet, not in the cached text) has LSC2/YGR244C with X in the Soluble and Matrix columns (L sol 0.98). (2) HAMAP transit peptide 1..30 in the UniProt record. (3) Soluble TCA-cycle enzyme, activity measured in isolated mitochondria. (4) The partner Lsc1 has the same assignment and a nucleoid row. All root rows are kept (projects/ORGANELLES.md rules 4 and 5). The location list keeps the root because the activity itself was measured in isolated mitochondria.
• GO:0005737 cytoplasm (ARBA00026971) stays TOO_GENERAL, with a corrected account of the rule. The rule has 7,695 condition sets; the three that name this family are IPR005809 + IPR005811 + PTHR11815:SF1 with taxon Archaea, Eukaryota or Bacteria (UniProt ARBA record, read from the audit download). SF1 is the eukaryotic A-type subfamily, so the rule is not a bacterial default.
• GO:0006104 succinyl-CoA metabolic process stays ACCEPT on both rows. "More general than the TCA cycle" was wrong (GO:0006104 is not an ancestor of GO:0006099; it is the parent of GO:1901289, also accepted). The IBA summary named Arabidopsis as a donor; the plant seed Q84LB6 is tomato (SUCB_SOLLC, checked in UniProt), and the rat seeds are Sucla2 (RGD:1309397) and Suclg2 (RGD:1305008). The family review quotes that summary and must be re-quoted.
• Reasons were added to the rows that had only a summary and a non-ACCEPT action (GO:0000287 and GO:0005524 KEEP_AS_NON_CORE, GO:0003824 MODIFY, both protein binding REMOVE rows).
• PMID:16270230: the note no longer calls the tomato beta subunit ADP-type; the abstract does not state its nucleotide specificity.
• OpenScientist report: its tree bootstrap figures (99%, 86%, 54%) have no saved code and were not reproduced; PMID:8144675 is cached and used in the E. coli sucD review.
• Not changed: every existing action.
• Cross-file consistency pass (2026-11-12). GO:0003824 catalytic activity (IEA) changed from MODIFY to TOO_GENERAL, the action used on the other ten subunit reviews of the family: the term is true and is entailed by GO:0004775, accepted in this review. The fungal-loss finding now names the ascomycete Saitoella complicata, whose Gln20 SUCLG2-type gene is of unresolved origin [file:yeast/LSC2/LSC2-bioinformatics/RESULTS.md "The Saitoella gene groups with the Mortierellomycota genes only weakly (ultrafast bootstrap 57), so vertical retention in Taphrinomycotina, transfer and contamination are not distinguished here."]. RESULTS.md no longer lists Dictyostelium among the unassayed Gln20 subunits: the enzyme purified from D. discoideum uses GTP PMID:7771809; the assay is not assigned to a beta gene.