AIGR Gene Hypothesis Deep Research — LSC2 (P53312): β-Subunit Nucleotide Specificity and Phylogeny

Target gene: LSC2 (YGR244C), succinyl-CoA synthetase/ligase β-subunit, Saccharomyces cerevisiae S288C (NCBITaxon:559292)
UniProt: P53312 · EC: 6.2.1.5 · PANTHER family: PTHR11815 · Hypothesis slug: beta-nucleotide-specificity-phylogeny
Seed reference: PMID:9874242

Summary

Verdict: Supported (with one wording caveat on the phylogenetic sister-relationship). The seed hypothesis — that S. cerevisiae Lsc2 is an ADP-specific (ATP-forming) succinyl-CoA synthetase β-subunit by descent, branching with the ADP-forming lineage rather than the animal GDP-forming SUCLG2 clade, and carrying the ADP-state nucleotide-specificity residues — is supported by three independent, converging lines of evidence: direct biochemistry, residue-level structural comparison, and molecular phylogeny.

The strongest single piece of evidence is direct and experimental. Przybyla-Zawislak and colleagues (PMID: 9874242) cloned both yeast subunit genes (LSC1/YOR142W and LSC2/YGR244C), showed that succinyl-CoA synthetase (SCS) activity is lost in lsc1/lsc2 deletion mitochondria and restored by plasmid-borne genes, and assayed nucleotide usage directly — the enzyme "utilize[s] ATP but not GTP for succinyl-CoA synthesis." This establishes the Lsc1/Lsc2 enzyme as ADP-forming without any need for phylogenetic inference. Structural and phylogenetic analyses then explain why: at the single residue that confers guanine-nucleotide specificity in the pig GTP-specific enzyme (Gln-20β; PMID: 16481318), yeast Lsc2 — like every other fungus, plant, and animal SUCLA2 examined — carries the ADP-state Proline, whereas only the animal GDP-forming SUCLG2 clade carries the specificity Glutamine. In an 18-taxon neighbor-joining tree with position-resampling bootstraps, yeast Lsc2 falls outside the animal SUCLG2/GDP clade in 99% of replicates, and all four fungi sit outside that clade in 86%.

The curation implication is clean: GO:0004775 (succinate–CoA ligase, ADP-forming activity) applies to LSC2 and GO:0004776 (GDP-forming) should not be propagated to it. The current UniProt record already carries GO:0004775, but only as IEA:UniProtKB-UniRule; the direct experimental evidence in PMID:9874242 justifies upgrading that annotation to an experimental evidence code. GDP specificity is best understood as an animal-SUCLG2-specific derived trait — its diagnostic Gln-20β determinant is absent even from a bacterial GDP-preferring enzyme (Thermus aquaticus), which uses a Pro20β + water arrangement instead. The one caveat is that the precise "sister-to-SUCLA2" wording in the seed is only weakly resolved by distance methods; the robust, defensible statement is that Lsc2 retains the ancestral ADP state outside the SUCLG2 clade, not that it is specifically a SUCLA2 ortholog.

Key Findings

Finding 1 — Direct biochemistry: the yeast Lsc1/Lsc2 enzyme uses ATP, not GTP (ADP-forming)

The most decisive evidence is experimental and organism-specific. Przybyla-Zawislak et al. (PMID: 9874242) isolated LSC1 (YOR142W) and LSC2 (YGR244C) from S. cerevisiae as genes "with high similarity to succinyl-CoA ligase subunits from other species," demonstrated that SCS activity is absent in mitochondria from lsc1/lsc2 deletion strains and restored by plasmid-borne LSC genes, and directly assayed nucleotide usage. Their conclusion is unambiguous: "The yeast succinyl-CoA ligase was shown to utilize ATP but not GTP for succinyl-CoA synthesis."

This is a direct biochemical demonstration of ADP-forming (ATP-specific) activity for the holoenzyme in which Lsc2 is the β-subunit. Because the β-subunit houses the ATP-grasp nucleotide-binding domain that determines nucleotide specificity, this result assigns the ADP-forming phenotype to the Lsc1/Lsc2 complex and, by extension, identifies Lsc2 as the β-subunit of an ADP-specific enzyme. This finding does not depend on any phylogenetic or residue-level inference; the sequence analyses below serve to explain and corroborate it rather than to establish it. The assay is qualitative (ATP yes / GTP no) rather than a quantitative kinetic measurement, but the ATP-vs-GTP distinction it reports is exactly the one at issue for the ADP-forming vs GDP-forming assignment.

Finding 2 — Lsc2 carries the ADP-state residue (Pro) at the guanine-specificity position (pig Gln-20β)

Structural work on the GDP-forming enzyme pinpointed the molecular determinant of guanine-nucleotide specificity. In pig GTP-specific SCS, "interactions of the guanine base with a glutamine residue (Gln-20β) and with backbone atoms provide the specificity" (PMID: 16481318). This Gln-20β is the single side chain that reads the guanine base and discriminates GDP/GTP from ADP/ATP.

A star alignment (Needleman–Wunsch) of 12 SCS β-subunits anchored on pig GDP-specific SUCLG2 (P53590; accounting for the 1–38 mitochondrial transit peptide so that mature Gln-20β maps to precursor index 57, confirmed as 'Q', with the companion Arg-54β at precursor index 91, confirmed 'R') showed a clean dichotomy at this position:

Yeast Lsc2 therefore matches the ADP/non-GDP state rather than the GDP-specific Gln. The companion determinant Arg-54β (which contacts the γ-phosphate) is Arg in all 12 β-subunits — it is family-conserved and therefore not a specificity discriminator, but it served as an internal cross-check that the alignment columns were correctly mapped. Independent structural work confirms that Pro at this position is the non-Gln state: in the GTP-preferring Thermus aquaticus enzyme, "A water molecule and Pro20β in T. aquaticus take the place of Gln20β in pig GTP-specific SCS" (PMID: 22751660).

{{figure:lsc2_specificity_summary.png|caption=Residue state at the two nucleotide-specificity positions (pig Gln-20β and Arg-54β) across succinyl-CoA synthetase β-subunits, with bootstrap/identity summary. Only the animal SUCLG2 (GDP-forming) clade carries Gln at the guanine-base position; yeast Lsc2, animal SUCLA2, and plant ADP-forming β all carry the ADP-state Pro. Arg-54β is conserved family-wide and is not a discriminator.}}

Finding 3 — Phylogeny places Lsc2 outside the animal GDP-forming SUCLG2 clade

A neighbor-joining tree of 12 β-subunits (k-mer distance, k=4) with 200 position-resampling bootstraps placed yeast Lsc2 outside the SUCLG2/GDP clade in 86% of replicates, with the animal SUCLA2 (ADP) clade recovered as monophyletic in 98.5%. Needleman–Wunsch pairwise identities for yeast Lsc2 were higher to the ADP-forming animal SUCLA2 subunits (mean 49.0%) than to the GDP-forming SUCLG2 subunits (mean 45.7%), and the top four identity hits were all SUCLA2/ADP subunits.

The evolutionary logic is captured by the primary phylogenetic paper: Johnson et al. (PMID: 9765291) concluded that the data "suggests that the A-beta and G-beta genes arose by duplication early in the evolution of multicellular eucaryotes." Because the ADP (A-β)/GDP (G-β) split is an animal-lineage duplication, the fungal β-subunit necessarily lies outside the GDP-forming SUCLG2 clade — it predates the duplication that created SUCLG2. This directly answers the seed's evolutionary question: GDP specificity arose within the animal SUCLG2 lineage, and the fungal β-subunit is outside that clade.

Finding 4 — Broadened 18-taxon sampling: all fungi retain the ADP-state Pro; Gln-20β is animal-SUCLG2-specific

To test robustness, the phylogeny was expanded to 18 β-subunits spanning four fungi (yeast Lsc2; S. pombe O94415; Neurospora Q9P567; Aspergillus A0A7R8AQL1), animal SUCLA2 and SUCLG2, plant, four bacterial SucC (E. coli, B. subtilis, Pseudomonas, M. tuberculosis), and the bacterial GDP-forming β of Thermus thermophilus (P25126). With 200 position-bootstraps:

• Yeast not inside the animal-GDP clade: 99.0%
• All four fungi outside the animal-GDP clade: 86.0%
• Animal SUCLA2 monophyly: 99.5%

At the pig Gln-20β position, all four animal SUCLG2 = Gln(Q); every eukaryotic ADP-forming β — all four fungi, plant, and animal SUCLA2 — = Pro(P); bacterial SucC gapped; and crucially the bacterial GDP-forming Thermus β = Pro, not Gln. This last point is decisive for interpretation: a GDP-preferring bacterial enzyme achieves its specificity by a different structural route (Pro20β + ordered water; PMID: 22751660), which confirms that the Gln-20β side chain is an animal-SUCLG2-specific guanine determinant, not a universal marker of GDP-forming enzymes.

{{figure:lsc2_20beta_residue_map.png|caption=Residue at the pig Gln-20β guanine-specificity position mapped across 13+ taxa including four fungi and the bacterial GDP-forming Thermus β-subunit. Gln (the GDP determinant) is confined to the animal SUCLG2 clade; all fungi carry the ADP-state Pro, and even the bacterial GDP-preferring enzyme uses Pro + water rather than Gln.}}

Finding 5 — Current GO annotation status and the curation upgrade

UniProt P53312 cross-references include GO:0004775 "succinate-CoA ligase (ADP-forming) activity" (evidence IEA:UniProtKB-UniRule), GO:0005524 ATP binding (IEA), GO:0000287 Mg binding (IEA), GO:0009361 succinate-CoA ligase complex (ADP-forming) (NAS), GO:0006099 TCA cycle (IBA), GO:0005739 mitochondrion (IDA), and GO:0006104 succinyl-CoA metabolic process (IDA:SGD). The catalytic reaction is Rhea:17661, "succinate + ATP + CoA = succinyl-CoA + ADP + phosphate" — explicitly ATP-based. GO:0004776 (GDP-forming) is not annotated.

The key curation observation is that the ADP-forming molecular-function term is currently present only as a computational (IEA) annotation, while direct experimental evidence exists in PMID:9874242 (the enzyme uses ATP not GTP; activity lost in lsc1/lsc2 deletions, restored by plasmids). This supports upgrading GO:0004775 to an experimental evidence code (IDA/IMP), and it reinforces that GO:0004776 should remain absent.

Mechanistic Model / Interpretation

Succinyl-CoA synthetase catalyzes the only substrate-level phosphorylation step of the citric acid cycle. The enzyme is an α/β heterodimer (α₂β₂ in bacteria; PMID:8144675); the α-subunit (Lsc1 in yeast) carries the catalytic phosphohistidine, and the β-subunit (Lsc2 in yeast) carries the ATP-grasp nucleotide-binding domain that determines whether ADP/ATP or GDP/GTP is used. Nucleotide specificity is therefore a β-subunit property, which is exactly why this hypothesis focuses on Lsc2.

The specificity switch is remarkably local. In the animal GDP-forming β-subunit (SUCLG2), a single glutamine side chain (Gln-20β) hydrogen-bonds the guanine base and confers GDP/GTP preference. Swap that glutamine for the smaller proline found in ADP-forming subunits, and the pocket can no longer read guanine, defaulting to adenine nucleotides.

    Guanine-specificity residue (pig Gln-20β position)
    ----------------------------------------------------
Animal SUCLG2  --- Gln (Q) ---> reads guanine base ---> GDP/GTP-forming  (GO:0004776)
     |
     | animal-specific A-β/G-β gene duplication (PMID:9765291)
     |
Ancestral β   --- Pro (P) ---------------------------> ADP/ATP-forming   (GO:0004775)
   |
   +-- Yeast Lsc2 (P53312)      Pro  --> ATP, not GTP (PMID:9874242)  ✔ direct assay
   +-- Other fungi (S.pombe, Neurospora, Aspergillus)  Pro
   +-- Plant ADP-forming β      Pro
   +-- Animal SUCLA2            Pro
   +-- Bacterial GDP-former (Thermus)  Pro + water (different route to GDP)

Three independent data layers converge on the same conclusion:

The convergence is important because each layer has different failure modes (an assay can be confounded by contaminating activities; a single residue can be misaligned; a distance tree can be misled by rate variation), yet all three agree. The evolutionary narrative is that GDP specificity is a derived, animal-SUCLG2-specific innovation layered onto an ancestral ADP-forming β-subunit scaffold, and fungi — including S. cerevisiae — retain that ancestral ADP state.

Evidence Base

Comparison Table — guanine-specificity residue (pig Gln-20β) across 18 β-subunits

Pattern: Gln-20β is present only in the animal GDP-forming SUCLG2 clade. Yeast Lsc2 and every other fungus carry the ADP-state Pro.

GO Curation Implications

Leads requiring curator verification.

1. GO:0004775 — succinate-CoA ligase (ADP-forming) activity (MF). Retain and strengthen. The term is correct and already present on P53312, but only as IEA:UniProtKB-UniRule. Direct experimental evidence (PMID:9874242) supports upgrading to an experimental evidence code (IDA for the ATP-vs-GTP assay; IMP is also defensible given the deletion-strain activity loss and plasmid rescue). This is the primary, core molecular function of Lsc2 as a β-subunit.

2. GO:0004776 — succinate-CoA ligase (GDP-forming) activity (MF). Keep absent; do not propagate. GDP specificity is an animal-SUCLG2-specific derived trait defined by Gln-20β, which yeast lacks (it has Pro). The seed context notes PAINT places an IKR for GO:0004775 and an IBD for GO:0004776 at the SUCLG2 ancestral node PTN000929335; the present analysis independently supports that the GDP-forming character is restricted to the animal SUCLG2 subtree, so any pipeline that would propagate GO:0004776 to the fungal β-subunit should be blocked for LSC2.

3. GO:0009361 — succinate-CoA ligase (ADP-forming) complex (CC). Retain. Consistent with the ADP-forming assignment and the α/β heterodimer biology; currently NAS.

4. GO:0006099 — tricarboxylic acid cycle (BP) and GO:0005739 / 0005759 — mitochondrion / mitochondrial matrix (CC). Retain (IBA and IDA respectively); consistent with pathway context and localization.

Avoid any recommendation that defaults to "protein binding"; the evidence supports the specific, informative MF term GO:0004775.

Mechanistic Scope

The immediate molecular function under test is the catalytic/nucleotide-specificity identity of the β-subunit within succinyl-CoA synthetase (α = Lsc1, β = Lsc2) — specifically whether Lsc2 contributes an ADP-forming (ATP-using) or GDP-forming (GTP-using) nucleotide-binding β-subunit. This is a direct gene-product property: the β-subunit's ATP-grasp domain physically binds the nucleotide, and the residue at position 20β (pig numbering) directly reads (or fails to read) the guanine base.

What is not being claimed here, and should be kept distinct from the molecular-function assignment:
- Downstream TCA-cycle flux phenotypes, respiratory-growth defects (impaired growth on glycerol/pyruvate, dispensable on acetate; PMID:9874242), or petite phenotypes of lsc mutants are pathway-level consequences, not the molecular specificity itself.
- The phylogenetic placement is inference about descent; it corroborates but does not by itself prove the biochemical specificity — that proof comes from the direct assay in PMID:9874242.
- The residue-level call is a structural inference about the pocket; it is consistent with, but subordinate to, the direct enzymology.

Conflicts and Alternatives

• No conflicting evidence was found; all lines agree on ADP-forming.
• Paralog/node overannotation (the main risk). PANTHER PTHR11815 contains both ADP-forming and GDP-forming β-subunits. The chief curation hazard is carry-over of the animal GDP-forming (SUCLG2 / GO:0004776) annotation onto fungal single-copy β-subunits via IBA/IBD. The data explicitly argue against this: fungi, including Lsc2, are outside the SUCLG2 subtree.
• "Sister-to-SUCLA2" overstatement. The seed wording implies Lsc2 groups specifically with animal SUCLA2. Distance methods only weakly support this (yeast+SUCLA2 grouping ~54%; yeast sits on a long branch). The defensible claim is the weaker, more robust one: Lsc2 retains the ancestral ADP state outside the SUCLG2 clade. An ML tree could place yeast basal to the animal duplication rather than nested within ADP-forming animals; either topology keeps it outside the GDP clade, so the conclusion is unchanged.
• GDP specificity is not monophyletic across all life. The bacterial Thermus enzyme is GDP-preferring yet lacks Gln-20β (it uses Pro + water; PMID:22751660). "GDP-forming" is therefore not a single evolutionary lineage; the Gln-20β marker is specific to the animal SUCLG2 route. This does not threaten the ADP assignment of Lsc2 but cautions against using Gln-20β as a universal GDP flag.
• Bacterial mapping limitation (stated plainly). In a pairwise alignment to pig SUCLG2, bacterial SucC N-termini are divergent and gap at the 20β column, so the E. coli residue at pig-20β could not be read directly from sequence — this is noted rather than inferred. E. coli's ADP-binding mode is nonetheless established structurally (PMID:8144675). This does not affect the eukaryotic conclusion.

Limitations and Knowledge Gaps

1. IEA vs experimental evidence code. Checked: UniProt P53312 shows GO:0004775 as IEA only. Why it matters: the direct assay in PMID:9874242 is not yet reflected in the evidence code, understating annotation strength. Resolution: a curator should add an IDA/IMP annotation citing PMID:9874242.
2. Precise phylogenetic position (basal vs SUCLA2-sister). Checked: k-mer NJ on 12 and 18 taxa; yeast robustly outside the GDP clade (99%) but SUCLA2-sister placement only ~54%. Why it matters: affects exact wording only. Resolution: curated MSA (MAFFT) + ML tree (IQ-TREE) with ≥1000 ultrafast bootstraps and broad taxon sampling.
3. Quantitative nucleotide kinetics. Checked: PMID:9874242 is qualitative ("ATP but not GTP"). Why it matters: no k_cat/K_m values for ADP vs GDP. Resolution: steady-state kinetics on purified recombinant Lsc1/Lsc2.
4. No experimental yeast Lsc2 structure. Checked: specificity inferred from pig (2FP4/2FPG), human SUCLA2 (6G4Q), E. coli SucC, and Thermus structures by alignment. Why it matters: the Pro-20β call rests on sequence alignment. Resolution: AlphaFold model of P53312 superposed on 2FP4/6G4Q to confirm pocket geometry.

Discriminating Tests / Proposed Follow-up Actions

• Add the experimental GO annotation now (highest value, lowest cost): cite PMID:9874242 for GO:0004775 (IDA/IMP) on P53312, and confirm GO:0004776 stays absent.
• Model-based phylogeny: MAFFT MSA of PTHR11815 β-subunits + IQ-TREE with ultrafast bootstrap to formally test whether LSC2 ⊄ SUCLG2 clade (expected: yes, high support) and whether SUCLA2-sister is supported.
• AlphaFold superposition: generate an Lsc2 model and superpose on human SUCLA2 (6G4Q, ADP-forming) and pig SUCLG2 (2FP4, GDP-forming) to verify Pro occupies the Gln-20β pocket and no Gln is present.
• In vitro nucleotide-preference assay: ADP vs GDP kinetics on recombinant yeast Lsc1/Lsc2 for quantitative confirmation.
• Targeted mutagenesis (confirmatory): Pro→Gln at the Lsc2 20β-equivalent to test whether GTP usage appears; conversely Gln→Pro in SUCLG2.
• Block PANTHER propagation: verify PAINT GDP-forming inferences at PTN000929335 are confined to the animal SUCLG2 subtree and do not reach fungal leaves.

Curation Leads (require curator verification)

• Retain + consider IEA→experimental upgrade: GO:0004775 (succinate–CoA ligase (ADP-forming) activity) — candidate reference PMID:9874242, exact snippet to verify: "The yeast succinyl-CoA ligase was shown to utilize ATP but not GTP for succinyl-CoA synthesis." Reaction concordant with EC 6.2.1.5 / Rhea:17661.
• Subunit identity snippet: "Two genes, LSC1 (YOR142W) and LSC2 (YGR244C), with high similarity to succinyl-CoA ligase subunits from other species were isolated from Saccharomyces cerevisiae." (PMID:9874242)
• Exclude MF: GO:0004776 (GDP-forming) — do not annotate / do not propagate to LSC2; candidate reference PMID:9765291, snippet: "arose by duplication early in the evolution of multicellular eucaryotes."
• Support BP/CC: GO:0006099 (TCA cycle); GO:0005759 (mitochondrial matrix).
• Candidate supporting refs: PMID:16481318 (Gln-20β determinant), PMID:22751660 (Pro-20β non-Gln state), PMID:8144675 (family architecture).
• Suggested question for curator: should the review explicitly document that the PAINT IBD for GO:0004776 at PTN000929335 is a derived SUCLG2 character and therefore not inherited by LSC2?
• Artifacts: lsc2_specificity_summary.png and lsc2_20beta_residue_map.png, generated from the executed alignments.

Bottom Line

The seed hypothesis is supported. Yeast Lsc2 (P53312) is an ADP-specific (ATP-forming) succinyl-CoA synthetase β-subunit by direct biochemistry (PMID:9874242), by its ADP-state Pro at the guanine-specificity residue (vs the animal SUCLG2 Gln-20β; PMID:16481318), and by its phylogenetic placement outside the animal GDP-forming SUCLG2 clade (99% bootstrap, 18 taxa; consistent with the animal-specific A-β/G-β duplication of PMID:9765291). GO:0004775 applies and merits an IEA→experimental upgrade; GO:0004776 should not be propagated to LSC2. The only material refinement is to describe Lsc2 as retaining the ancestral ADP state outside the SUCLG2 clade rather than asserting a specific SUCLA2 sister relationship.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist lsc2 20beta residue map
  
• OpenScientist lsc2 specificity summary