AIGR Deep Research Report — S. pombe cps1 / SPAC24C9.08 (UniProt O13968)

Focus type: function_assignment
Hypothesis slug: additional-lipid-amide-and-extracellular-capacities
Gene: cps1 (Carboxypeptidase S, CBPS_SCHPO), Schizosaccharomyces pombe 972h- (NCBITaxon:284812)

Executive Judgment

Verdict: Over-annotated / weakly supported for the "additional" functions; the core vacuolar carboxypeptidase function is well supported.

Evaluated claim-by-claim, the seed's proposed additional capacities for O13968 are not backed by direct evidence:

The seed is correct on several methodological points: (a) the target is not bgs1 glucan synthase; (b) the PANTHER family label "PM20D1" is not proof of transferred specificity; (c) GFP-/Ub-GFP- constructs report engineered MVB cargo behaviour, not native secretion; (d) absence of a fungal biochemical assay is not positive proof of loss of any activity. However, the practical curation consequence is that the additional PM20D1-derived MF/BP/CC terms rest solely on rule-based electronic inference and should not be treated as gene-product functions without experimental support. The primary, experimentally-anchored function is a luminal vacuolar-membrane metallocarboxypeptidase (peptidase M20A).

Evidence Matrix

Residue-level & domain-database analysis (Iteration 2, computed)

• Independent domain databases assign carboxypeptidase S, not PM20D1. For O13968, InterPro member DBs give: CDD cd05674 "M20 Peptidase, carboxypeptidase yscS"; PIRSF037217 "Carboxypeptidase S"; InterPro IPR017141 "Peptidase M20, carboxypeptidase S" → GO:0004181 metallocarboxypeptidase activity. The PANTHER PTHR45962 entry (name "N-fatty-acyl-amino acid synthase/hydrolase PM20D1") carries no GO terms — the PM20D1 label is a family name without functional annotation.
• M20 catalytic core is 100% conserved across O13968 / ScCPS1 / human PM20D1: His197, Asp199, Asp232, Glu266 (proton acceptor), Glu267, Glu296 all align 1:1 (PM20D1 His125/Asp127/Asp157/Glu191/Glu192/Glu218). ⇒ All three are genuine binuclear-zinc amide-bond metallohydrolases; O13968 mechanistically retains amide/peptide hydrolase capability (argues against "loss of peptide hydrolysis").
• A substrate-binding residue tracks the fungal CPS clade, not PM20D1: O13968 His565 = ScCPS1 His547 (match) but = PM20D1 Asn467 (divergent). Consistent with conserved carboxypeptidase-S substrate recognition in the fungal enzymes and altered specificity in PM20D1.

Net: shared catalytic chemistry does not equal shared substrate specificity. The specificity-informative, GO-bearing signatures unanimously call O13968 a carboxypeptidase S.

GO Curation Implications (leads — require curator verification)

Retain (core, experimentally/phylogenetically supported):
- MF GO:0004181 metallocarboxypeptidase activity (ISO) and GO:0004180 carboxypeptidase activity (IBA) — retain; these are the direct molecular function.
- CC GO:0000324 fungal-type vacuole (HDA) / GO:0000328 vacuole lumen (IBA) / GO:0005774 vacuolar membrane — retain (type II membrane, luminal catalytic domain).
- BP GO:0007039 protein catabolic process in the vacuole (TAS) / GO:0006520 amino acid metabolic process — retain.

Treat as non-core / candidate for removal or "do not accept" (IEA:UniProtKB-ARBA only, family-name driven):
- GO:0005576 extracellular region — flag as over-annotation; conflicts with HDA vacuole localization and lumenal type II topology. Lead: do not promote to a curated CC; consider NOT-qualifier or removal.
- GO:0006629 lipid metabolic process — flag as over-annotation; no fungal lipid-catalysis evidence.
- GO:0016810 hydrolase activity acting on C–N (but not peptide) bonds — flag; this is the N-acyl-amino-acid-hydrolase-flavored MF and is electronic-only.

Do not add:
- GO:1990845 adaptive thermogenesis — no evidence, mechanistically inapplicable to a unicellular fungus; PM20D1 thermogenesis is adipocyte/mitochondrial-context specific. Recommend against.

Avoid "protein binding" as an outcome — the informative MF here is metallocarboxypeptidase activity.

GO Decision Table (computed provenance — see go_decision_table.csv)

Native localization support: PMID 16823372 (Matsuyama et al., 2006) — genome-wide YFP ORFeome localization of ~90% of the S. pombe proteome — provides the native (non-engineered) localization dataset the seed references; it is a high-throughput localization resource consistent with the HDA vacuolar assignment (as opposed to the engineered GFP-/Ub-GFP-Cps1 MVB-cargo constructs).

Mechanistic Scope

Direct molecular function tested: zinc-dependent M20A exopeptidase releasing a C-terminal amino acid from peptides with a penultimate glycine (carboxypeptidase S activity), acting in the vacuole lumen for peptide/amino-acid recycling and use of peptides as nitrogen source.

Downstream / non-core: MVB sorting behaviour of GFP-/Ub-GFP-Cps1 fusions is a property of ubiquitin-dependent cargo trafficking machinery, not of the enzyme's catalytic activity, and does not indicate secretion. The PM20D1 "uncoupling/thermogenesis" phenotype is a mammalian, whole-organism metabolic outcome of a secreted paralog and is not a molecular property attributable to O13968.

Conflicts and Alternatives

• Paralog / family-name carry-over: PANTHER PTHR45962 is named for mammalian PM20D1; UniProt's ARBA rules propagate lipid-amide/extracellular terms to all members, including this fungal M20A vacuolar carboxypeptidase. This is the classic paralog-overannotation and frequency-bias pattern.
• Name collision: "cps1" also denotes S. pombe bgs1/SPBC19G7.05c (1,3-β-glucan synthase). The seed correctly excludes it; this report concerns only SPAC24C9.08/O13968.
• Topology conflict: A secreted, circulating N-acyl-amino-acid enzyme (PM20D1) vs a type II vacuolar-membrane protein with luminal catalytic domain — the localizations are mutually exclusive.
• Divergence: ~28% identity to PM20D1 (twilight zone) vs ~37% to a bona fide fungal carboxypeptidase S argues the substrate specificity should be read from the fungal CPS clade.

Knowledge Gaps

1. No fungal biochemical assay of O13968 (or ScCPS1) for N-acyl-amino-acid synthase/hydrolase activity exists. Checked: UniProt catalytic annotation + literature; only peptidase activity is documented. Matters: one cannot positively exclude a promiscuous side-activity; but absence of evidence ≠ presence of function for curation. Resolve: in vitro assay with purified enzyme against N-acyl-amino-acid panel.
2. Active-site substrate-determinant comparison with PM20D1 not performed residue-by-residue here. Matters: would test whether the fatty-acyl binding pocket exists. Resolve: structural/AlphaFold pocket comparison and specificity-residue alignment.
3. Native secretion: no evidence of extracellular pool; HDA is vacuolar. Resolve: secretome/immunolocalization in S. pombe.

Discriminating Tests

• In vitro activity panel: purified O13968 assayed for (i) Gly-penultimate carboxypeptidase activity (expected positive) and (ii) N-oleoyl-/N-arachidonoyl-amino-acid hydrolysis/condensation (tests PM20D1-type activity). Bidirectional LC-MS assay as in PMID 27374330.
• Structure-guided pocket analysis: AlphaFold model + superposition on PM20D1/carboxypeptidase-S structures; check for a hydrophobic acyl-binding channel and conserved Zn-coordinating residues (His199/His-cluster, Glu266 proton acceptor already annotated).
• Localization: endogenous-tag microscopy and secretome MS to confirm vacuolar (not extracellular) steady-state pool.
• Specificity-residue alignment across M20A CPS clade vs M20 PM20D1 clade to identify divergent substrate determinants.

Curation Leads (require curator verification)

1. Downgrade/flag electronic PM20D1-derived terms (GO:0005576, GO:0006629, GO:0016810) as IEA-only family carry-over; do not elevate to experimental status. Suggested action: mark as over-annotation / non-core; consider removal or NOT for extracellular.
2. Do not add GO:1990845 (adaptive thermogenesis). No support; mechanistically inapplicable.
3. Retain and foreground MF GO:0004181/GO:0004180 (metallocarboxypeptidase), CC vacuole/vacuolar-lumen terms, BP vacuolar protein catabolism.
4. Candidate references to verify (with snippets):
5. PMID 27374330 — "we identify a secreted enzyme, peptidase M20 domain containing 1 (PM20D1) … catalyzing both the condensation of fatty acids and amino acids to generate N-acyl amino acids and also the reverse hydrolytic reaction." → establishes activity in a secreted mammalian adipocyte context, not transferable by homology.
6. PMID 31341193 — "Carboxypeptidases Y (Cpy1) and S (Cps1) … follow the carboxypeptidase Y (CPY) pathway from the trans-Golgi network (TGN) to the prevacuolar endosome (PVE)." → native S. pombe Cps1 traffics to the vacuole.
7. PMID 18508771 — "missorting of the MVB cargo GFP-Cps1." → GFP-Cps1 is engineered vacuolar MVB cargo, not evidence of secretion.
8. Suggested question for curators: Should the S. pombe CBPS record's ARBA-derived extracellular/lipid/C–N-hydrolase terms be suppressed pending a fungal biochemical assay?
9. Suggested experiment: bidirectional N-acyl-amino-acid LC-MS assay on purified O13968 to positively test (not assume) PM20D1-type promiscuity.

Provenance

Computed checks (UniProt REST retrieval of O13968 annotations/topology/GO evidence codes; Needleman-Wunsch pairwise identities O13968/P27614/Q6GTS8) were executed in this session; identities 36.7% (vs ScCPS1) and 27.9% (vs human PM20D1) are direct outputs. Simple match/mismatch scoring was used, so absolute identities are approximate; the relative ordering (closer to fungal CPS than to PM20D1) is the robust conclusion. Literature evidence retrieved via PubMed (PMIDs 27374330, 31341193, 18508771).

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist go decision table