SSA2 (P10592) — Review of GO:0005886 "plasma membrane" annotation

Focus: existing_go_annotation_decision · existing-go-0005886-keep-as-non-core
Term: plasma membrane (GO:0005886) · Evidence: IBA · Ref: GO_REF:0000033
Current action under test: KEEP_AS_NON_CORE
Iteration: finalized in Iteration 3 of 3

Data-integrity note (Iteration 2, QuickGO/EBI live check): In the current GO database the
only plasma-membrane (GO:0005886) annotation for SSA2 is HDA from PMID:16622836 (SGD).
There is no IBA annotation to plasma membrane — the IBA (GO_REF:0000033, GO_Central) pipeline
propagated only cytosol/cytoplasm/nucleus to SSA2. The seed row's evidence_type: IBA for GO:0005886
therefore appears to be a mismatch; the reviewed term is HDA-only. The paralog SSA1 (P10591)
carries the identical single HDA PM annotation from the same paper, consistent with shared-peptide
paralog co-detection (SSA1/SSA2 ≈97% identical). This does not change the verdict but should be
corrected in the review.

Executive Judgment

Verdict: Partially supported — the KEEP_AS_NON_CORE action is justified (conservative and appropriate).

The plasma-membrane annotation is real in the databases but weakly grounded. It rests on
(1) a single high-throughput proteomics detection in a stripped plasma-membrane fraction
(PMID:16622836, Delom et al. 2006) recorded as HDA, and (2) an IBA phylogenetic propagation
from the HSP70 family (GO_REF:0000033). Neither establishes plasma membrane as a primary
site of Ssa2 function. SSA2 is one of the most abundant cytosolic Hsp70 chaperones and has
no transmembrane domain, no signal peptide, and no lipid anchor (UniProt P10592), so its
appearance in a membrane pellet is best explained by abundant-cytosolic-protein carryover —
the same fractions routinely capture elongation factors and enolase (PMID:16544286). There is
genuine cell-periphery biology (histatin-5 receptor at the cell envelope, PMID:12761219;
secreted/cell-wall and extracellular-vesicle IDA annotations), but that is cell wall/secreted,
not integral plasma membrane.

Most important caveats: (a) The evidence would also support a defensible remove/do-not-propagate
argument for the IBA specifically; KEEP_AS_NON_CORE is the more conservative of two reasonable
choices. (b) Curators should ensure the term is never treated as core and that the richer, better-
evidenced peripheral terms (cell wall GO:0009277, extracellular region/vesicle) carry the
cell-surface story instead.

Evidence Matrix

Provenance: /tmp/SSA2_GO0005886_evidence_matrix.csv, /tmp/SSA2_CC_landscape.csv (computed this run).

GO Curation Implications

• Lead (requires curator verification): KEEP the GO:0005886 annotation but treat it as NON-CORE — consistent with the seed action.
• It is a CC term. Evidence supports presence at but not function at the plasma membrane.
• Do not generalize it to a core location term (cytoplasm/cytosol IDA already carry the core CC).
• Alternative lead worth flagging: because the sole experimental support is one stripped-membrane HT proteomics hit for a hyper-abundant soluble chaperone, a curator could reasonably down-weight or not-propagate the IBA (mark with NOT/annotation-extension review) while retaining the HDA as non-core. This is a stronger-scrutiny option, not a recommendation to delete outright.
• Ensure the better-evidenced peripheral terms are the ones telling the cell-surface story: fungal-type cell wall (GO:0009277, IDA), extracellular vesicle (GO:1903561, IDA), secreted/extracellular region.

SSA2 cellular-component landscape (from UniProt/SGD): cytoplasm (IDA, core), cytosol (IDA, core),
fungal-type cell wall (IDA), extracellular vesicle (IDA), fungal-type vacuole membrane (IDA),
mitochondrion (IDA), nucleus (IBA), plasma membrane (HDA + IBA → non-core, weak).

Mechanistic Scope

The immediate molecular function of Ssa2 is ATP-dependent Hsp70 chaperone activity (nucleotide-
binding domain + substrate-binding domain) acting in the cytosol/nucleus: de novo folding,
post-translational translocation across organelle membranes, tRNA nuclear import, and assembly of
client complexes. Any plasma-membrane signal is downstream/associative, not a direct PM
activity: (i) fraction co-purification of an abundant soluble protein, and (ii) transient,
client-directed membrane engagement (e.g., inserting viral replication proteins into membranes,
PMID:19153242). None of these convert PM into a site of intrinsic Ssa2 molecular function.

Conflicts and Alternatives

• Contamination / frequency bias: Ssa2's extreme cytosolic abundance makes membrane-pellet
  detection the expected null result; the control literature (PMID:16544286) confirms cytoplasmic
  carryover in these fractions.
• Paralog confusion: Ssa1/Ssa2 are ~97% identical and functionally redundant; several
  "cell-surface Ssa" studies (e.g., histatin) cannot resolve Ssa1 vs Ssa2 vs Ssa3/4. Cell-surface
  claims may be family-level, not SSA2-specific. Confirmed: SSA1 (P10591) carries the identical
  single HDA PM annotation from the same paper (PMID:16622836), i.e. the proteomics signal was
  propagated to both paralogs and cannot distinguish them.
• Evidence-code carry-over (data integrity): the seed labels the reviewed annotation IBA
  (GO_REF:0000033), but the live GO record has no IBA to plasma membrane; the term is HDA-only.
  The "IBA is consistent" rationale in the review summary does not apply and should be revised.
• Location mis-typing: the strongest peripheral evidence points to cell wall/cell envelope
  and secreted, which are distinct from the plasma membrane; GO:0005886 may be a coarse stand-in
  for these better-supported terms.
• IBA source strength: the phylogenetic call inherits mammalian Hsc70 PM annotations, whose own
  "PM" status is itself partly surface/associative.

Knowledge Gaps

1. Is Ssa2 genuinely at the PM vs cell wall? Checked: UniProt features (no TM/anchor), Delom
   proteomics (stripped PM), histatin papers (cell wall). Matters because it decides cell-wall vs PM
   term. Resolve with GFP/split-GFP PM-marker co-localization or protease-shaving surface proteomics
   with strict PM-marker controls.
2. Paralog specificity. Checked: sequence identity/redundancy literature. Matters for SSA2-
   specific annotation. Resolve with Ssa2-only tagged strains in an ssa1Δ ssa3Δ ssa4Δ background.
3. Functional consequence at the PM. Checked: no functional PM assay exists. Matters because
   "non-core" hinges on absence of a PM function. Resolve with PM-targeted degron/anchor-away tests
   for a PM-localized phenotype.
4. Quantitative abundance rank. The contamination argument leans on Ssa2 being among the most
   abundant cytosolic proteins. Checked: attempted a programmatic PaxDb lookup (YLL024C) — the PaxDb
   API was not accessible (HTTP 404), so no numeric rank is reported here to avoid fabrication.
   The abundance is nonetheless well documented in the literature (Hsp70 Ssa1/Ssa2 are among the
   highest-copy chaperones; UniProt lists "Direct protein sequencing"). A curator can confirm the
   exact rank via PaxDb/SGD abundance datasets; this would further quantify carryover risk.

Discriminating Tests

• Quantitative co-localization of Ssa2-GFP with a bona-fide PM marker (e.g., Pma1-mCherry) vs a
  cell-wall marker, in live cells — distinguishes PM from wall/cytosol.
• Fractionation with sucrose-gradient PM enrichment + abundance normalization: does Ssa2 enrich
  with PM markers beyond its bulk cytosolic level, or track with soluble contaminants?
• Surface biotinylation / trypsin-shaving of intact spheroplasts to test true surface exposure.
• Cross-reference SGD/GO annotation provenance to confirm the HDA derives solely from Delom 2006
  and whether the IBA has independent experimental support in the HSP70 tree.

Curation Leads (require curator verification)

• Action: Retain seed action KEEP_AS_NON_CORE for GO:0005886. Rationale is sound.
• Correct the review metadata: change evidence_type for GO:0005886 from IBA → HDA (live GO
  shows HDA/PMID:16622836 only; no IBA to PM). Revise the review summary sentence "IBA is consistent"
  because no PM IBA exists. Verify against QuickGO (geneProductId=P10592&goId=GO:0005886).
• Candidate reference to cite in the review: PMID:16622836 — verify snippet: "Proteins from a
  stripped plasma membrane fraction were solubilized with the neutral and non-denaturing detergent,
  the n-dodecyl beta-D-maltoside." (documents the HT/fractionation nature of the evidence).
• Contamination-control reference: PMID:16544286 — verify snippet: "Elongation factors,
  mitochondrial membrane proteins and a mannosyltransferase... were also identified in this
  fraction." (supports carryover interpretation).
• Competing/peripheral biology reference: PMID:12761219 — verify snippet: "co-localized on
  immunoblots of yeast beta-mercaptoethanol cell wall extracts and cytosolic fractions."
• Candidate refinement: ensure cell-surface signal is represented by GO:0009277 (fungal-type
  cell wall, IDA) and GO:1903561 (extracellular vesicle, IDA), so PM is not over-interpreted.
• Optional stronger-scrutiny lead: flag the IBA GO:0005886 for reduced propagation confidence
  given the single abundant-protein HT source; do not delete without curator review.
• Suggested question for curator: Should GO:0005886 be down-graded/annotated with an evidence
  note distinguishing "detected in PM fraction" from "functions at PM"?

Analyses run: NCBI eutils fetch of PMID:16622836; UniProt P10592 feature/localization/GO-CC parse;
QuickGO/EBI live annotation query for P10592 and P10591 (Iteration 2); evidence-matrix and
CC-landscape tables (saved to /tmp).

Iteration-2 QuickGO result (verbatim): SSA2 GO:0005886 = 1 annotation, evidence=HDA,
ref=PMID:16622836, by=SGD; SSA1 GO:0005886 = 1 annotation, evidence=HDA, ref=PMID:16622836, by=SGD.
Full SSA2 CC set (16 annotations) shows IBA/GO_REF:0000033 only on GO:0005634, GO:0005737, GO:0005829.

Artifacts

• OpenScientist final report
• OpenScientist final report