SSA4 (P22202) — Plasma Membrane (GO:0005886) Annotation Review OpenScientist openscientist-autonomous 8 citations 2 artifacts 2026-08-22T15:13:37.919346 citations file

OpenScientist final report (text/html)

OpenScientist final report (application/pdf)

SSA4 (P22202) — Plasma Membrane (GO:0005886) Annotation Review

Focus: existing_go_annotation_decision · existing_annotations[3] in genes/yeast/SSA4/SSA4-ai-review.yaml
Seed hypothesis: Keep the IBA plasma-membrane annotation as KEEP_AS_NON_CORE.
Gene: SSA4 / Saccharomyces cerevisiae stress-inducible cytosolic Hsp70 (SSA subfamily), 642 aa.

Executive Judgment

Verdict: Over-annotated — the seed KEEP_AS_NON_CORE action is defensible but arguably too weak; REMOVE is the better-supported lead.

The plasma-membrane (GO:0005886) annotation on SSA4 is a phylogenetic (IBA, GO_REF:0000033) inference with no yeast-specific support. Three independent lines of evidence converge:

No experimental support in yeast. UniProt records SSA4 subcellular location as Cytoplasm only. The only experimental (IDA, PMID:11279056) cellular-component evidence is for cytoplasm and nucleus. SSA4 is a nucleocytoplasmic shuttling Hsp70 (PMID:17020589).
The PM annotation is not present in the current GO release. A live QuickGO query for P22202 returns six cellular-component annotations — nucleus, cytoplasm, cytosol (IDA/IEA/IBA) — and no plasma-membrane annotation. GO:0005886 survives only as a stale cross-reference in the cached UniProt entry. In other words, GO_Central/PAINT appears to have already dropped this term for the Hsp70 ancestral node relevant to Ssa4.
The proximal source of the family PM signal is a single contamination-prone yeast PM-proteome study — which did not include Ssa4. Live QuickGO shows PM (GO:0005886) annotations for the constitutive paralogs Ssa1, Ssa2, and Ssb1, all HDA evidence from one reference, PMID:16622836 (Delom et al. 2006, stripped-PM-fraction mass-spec of ~90 proteins). Abundant cytosolic chaperones routinely co-purify in such membrane fractions. Ssa4 was not detected (it is stress-inducible with low basal abundance), so its PM term was purely phylogenetic (IBA) and is now absent from live GO. (A separate mammalian tumor-cell-surface Hsp70 phenomenon exists — Gb3/lipid-raft dependent, PMID:37932378, 30632067, 39015084 — but human HSPA1A itself carries no PM GO term, so this is not the direct source here.)
The authoritative organism database (SGD) does not annotate Ssa4 to the plasma membrane. SGD's GO details for YER103W (146 annotations) list cellular-component terms only for cytoplasm and nucleus (manually curated) plus cytoplasm/nucleus/cytosol (computational) — no GO:0005886. SGD's curated description attributes Ssa4's membrane role to "SRP-dependent cotranslational protein-membrane targeting and translocation" (i.e., the ER) and describes it as a "cytoplasmic protein that concentrates in nuclei upon starvation."

Most important caveat: Cytosolic Hsp70s (including yeast Ssa) genuinely engage membranes peripherally during protein/mRNA targeting — but at the ER (SRP-dependent cotranslational translocation, per SGD) and the mitochondrial outer membrane (Tom70-dependent mRNA/precursor delivery, PMID:22138184), not the plasma membrane. The paralog HDA "PM" detections are a weak, contamination-prone basis, and even they do not include Ssa4. So the "plausible peripheral association" rationale in the seed is real but thin, and does not specifically implicate Ssa4 at GO:0005886.

Evidence Matrix

Citation	Evidence type	Supports/Refutes	Claim tested	Key finding	Context	Confidence / limitations
UniProt P22202 (record)	Database	Refutes PM	Primary localization of Ssa4	Subcellular location = "Cytoplasm" only	S. cerevisiae	High for curated summary; database-level
QuickGO live query (P22202)	Database/computational	Refutes PM	Does GO currently annotate PM?	6 CC annotations (nucleus, cytoplasm, cytosol); no GO:0005886	GO_Central release	High; reflects current GO, may lag PAINT tree edits
PMID:11279056 (IDA, SGD)	Localization (experimental)	Qualifies	Experimental localization	IDA support only for cytoplasm and nucleus	S. cerevisiae	High for cyto/nucleus; silent on PM
PMID:17020589	Localization / transport	Refutes PM	Where does Ssa4 reside?	Ssa4p shuttles nucleus↔cytoplasm; nuclear on stress via Msn5 export	S. cerevisiae	High; no membrane localization
PMID:22138184	Mutant/mechanism	Qualifies	Membrane association of Ssa	Ssa1 targets mRNA to mitochondrial outer membrane (Tom70)	S. cerevisiae	Peripheral organellar, not plasma membrane
PMID:25853343	Interaction/mechanism	Qualifies	Ssa membrane/pore contacts	Ssa2 binds nucleoporin Nup116 for tRNA import	S. cerevisiae	Nuclear pore, not plasma membrane
PMID:16622836 (Delom 2006)	Localization (HDA proteomics)	Qualifies / competing origin	Are yeast Hsp70s at the PM?	Ssa1/Ssa2/Ssb1 detected in stripped-PM fraction (~90 proteins); Ssa4 not detected	S. cerevisiae	HDA on paralogs only; abundant chaperones prone to co-purification; no Ssa4
PMID:37932378	Structural/biophysical	Competing (mammalian)	Basis of PM-Hsp70	Hsp70 binds DOPC/DOPS bilayers; PM-Hsp70 specific to tumor cells	Mammalian/artificial membrane	Distinct phenomenon; human HSPA1A carries no PM GO term
PMID:30632067; PMID:39015084	Localization (mammalian)	Competing (mammalian)	Where is surface Hsp70 seen?	mHsp70 on tumor-cell surface, Gb3/lipid-raft dependent	Human/mouse tumor	Organism/context-specific
QuickGO paralog scan (this run)	Computational/database	Qualifies	Which Hsp70s carry PM in GO?	PM present for Ssa1/Ssa2/Ssb1 (HDA, PMID:16622836); absent for Ssa4, Ssa3, human HSPA1A	GO_Central live	Direct query; reflects current release
SGD YER103W record (this run)	Review/database	Refutes PM	Does SGD place Ssa4 at PM?	CC = cytoplasm/nucleus/cytosol only; membrane role = SRP-dependent ER cotranslational targeting	S. cerevisiae	Authoritative organism DB; no PM term

GO Curation Implications

Term: plasma membrane (GO:0005886), a cellular component (CC) term. Evidence type IBA / GO_REF:0000033.
Recommended lead (requires curator verification): REMOVE the plasma-membrane annotation, or at minimum retain as NON_CORE only if the curator prefers strict conservatism. Rationale: (a) no yeast experimental support; (b) the term is absent from the current GO_Central set for P22202, suggesting the phylogenetic inference has already been retracted upstream; (c) the ancestral signal derives from mammalian tumor-cell-surface Hsp70, an organism/context-specific property.
Better-supported CC terms are already present and experimental: cytoplasm (GO:0005737, IDA), nucleus (GO:0005634, IDA), cytosol (GO:0005829). These should be the core localizations.
The seed action KEEP_AS_NON_CORE is not wrong, but it is more generous than the evidence warrants. The only yeast PM evidence for this family (HDA, PMID:16622836) covers the paralogs Ssa1/Ssa2/Ssb1, is contamination-prone, and excludes Ssa4; the genuine peripheral membrane roles of Ssa chaperones concern the ER/mitochondrial outer membrane, not the plasma membrane. If the curator prefers conservatism, KEEP_AS_NON_CORE is acceptable, but the annotation should be explicitly flagged as an IBA no longer present in the live GO release for Ssa4.

Mechanistic Scope

Direct molecular function of Ssa4: ATP-dependent Hsp70 chaperone (protein folding, nascent-chain and stress-denatured protein binding), operating in the cytosol and nucleus.
Genuine membrane-proximal roles (peripheral, transient): delivery of precursor proteins/mRNAs to the ER (Sec translocon) and mitochondrial outer membrane (Tom70); interaction with nuclear pore components (Nup116, Nup82) during nucleocytoplasmic shuttling. None of these are the plasma membrane.
Not supported as direct gene-product localization: residence in/at the plasma membrane.

Paralog / ortholog PM-annotation comparison (live QuickGO, this run)

Protein	Accession	PM (GO:0005886)?	Evidence	Reference
Ssa4 (yeast, target)	P22202	No	—	— (IBA only, now absent)
Ssa1 (yeast)	P10591	Yes	HDA	PMID:16622836
Ssa2 (yeast)	P10592	Yes	HDA	PMID:16622836
Ssa3 (yeast)	P09435	No	—	—
Ssb1 (yeast)	P11484	Yes	HDA	PMID:16622836
Ssc1 (yeast, mito)	P12398	No	—	—
HSPA1A (human)	P0DMV8	No	—	—
HSPA8 (human)	P11142	Yes	IEA/TAS	—

Conflicts and Alternatives

Real-but-weak yeast paralog evidence: Unlike my initial mammalian-origin hypothesis, the yeast PM annotations for Ssa1/Ssa2/Ssb1 are experimental (HDA) — but all from one stripped-PM-fraction proteomics study (PMID:16622836) where abundant cytosolic chaperones commonly co-purify, and which did not include Ssa4. This is a contamination-prone, non-specific basis and is paralog-, not Ssa4-, specific.
Paralog/ortholog carry-over: The Hsp70/HSPA PANTHER family mixes yeast Ssa, Ssb, Sse, mammalian HSPA1A/HSPA8, etc. The separate mammalian tumor-cell-surface Hsp70 biology belongs to inducible HSPA1A and should not be propagated to Ssa4; note HSPA1A itself carries no PM GO term.
Stale vs. live records: UniProt's cached cross-reference still lists GO:0005886 (IBA), whereas live GO_Central/QuickGO does not — a database-lag artifact that could make the annotation look current when it is effectively retired.
Real-but-different membrane biology: Ssa's authentic ER/mitochondrial-membrane peripheral associations could be mistaken as generic "membrane" support, but they map to distinct CC terms, not plasma membrane.

Knowledge Gaps

Is the PM annotation truly retracted upstream, or filtered by QuickGO? — Checked: live QuickGO returns no GO:0005886 for P22202. Matters because it changes REMOVE vs KEEP. Resolution: inspect the current PAINT/PANTHER family tree (PTN node) annotations and the GO_Central GAF directly.
Any high-throughput yeast plasma-membrane/proteomics hit for Ssa4? — Not found in targeted literature. Matters because a peripheral PM proteomics signal could justify NON_CORE. Resolution: check membrane-proteome and BioID/proximity datasets (SGD, GFP-localization Huh et al.).
Does any yeast Ssa localize to the plasma membrane under stress? — No evidence found; stress redistributes Ssa4 to the nucleus, not the PM. Resolution: stress-condition live imaging.

Discriminating Tests

GFP/mNeonGreen live imaging of Ssa4 under normal and heat/ethanol stress with a PM marker (e.g., Pma1) — expect cytosol/nucleus, no PM enrichment.
Cell-surface biotinylation / non-permeabilized immunostaining for Ssa4 — expect negative (unlike mammalian tumor Hsp70).
Subcellular fractionation (cytosol vs. PM vs. ER vs. mitochondria) with quantitative MS — expect cytosolic/nuclear, with minor ER/mito peripheral pools, not PM.
Inspect the current PANTHER PTN node GAF to confirm whether GO:0005886 is still propagated to the Ssa4-containing subtree.

Curation Leads (require curator verification)

Action change lead: Consider REMOVE for GO:0005886 (IBA) rather than KEEP_AS_NON_CORE; if retained, keep strictly as NON_CORE, low-confidence.
Candidate references / snippets to verify:
PMID:17020589 — "Cytoplasmic hsp70s like yeast Ssa4p shuttle between nucleus and cytoplasm under normal growth conditions but accumulate in nuclei upon stress." (supports cytosol/nucleus as the real localizations).
PMID:37932378 — "Membrane-bound heat shock protein 70 (Hsp70) apart from its intracellular localization was shown to be specifically expressed on the plasma membrane surface of tumor but not normal cells." (identifies mammalian tumor origin of PM-Hsp70).
PMID:11279056 — IDA source for cytoplasm and nucleus (SGD).
Candidate retained CC terms: GO:0005737 (cytoplasm, IDA), GO:0005634 (nucleus, IDA), GO:0005829 (cytosol) as core.
Suggested question for curator: Is the reviewed PM IBA still present in the GO_Central GAF/PAINT tree, or is it already retired (as live QuickGO suggests)? If retired upstream, the review row should note the annotation no longer exists and mark REMOVE/obsolete rather than KEEP.
Suggested experiment: cell-surface biotinylation + fractionation MS to definitively exclude a yeast plasma-membrane pool.

Provenance

UniProt REST (P22202.json): subcellular location + GO cross-references (executed).
QuickGO annotation API (/annotation/search, geneProductId=P22202): 20 total annotations, 6 CC, no GO:0005886; IBA withFrom ortholog list captured (executed).
QuickGO paralog/ortholog scan (Ssa1/Ssa2/Ssa3/Ssb1/Ssc1, HSPA1A/HSPA8/mouse Hsp70): PM present only for Ssa1/Ssa2/Ssb1 (HDA, all PMID:16622836) (executed).
SGD backend API (/backend/locus/YER103W + /go_details): 146 GO annotations, CC = cytoplasm/nucleus/cytosol only, no PM; curated description cites SRP-dependent ER cotranslational targeting (executed).
PubMed searches: PMID 17020589, 11279056, 22138184, 25853343, 16622836, 37932378, 30632067, 39015084.

All computational results above were executed against live public resources during this run; no results were fabricated. UniProt's cached GO cross-reference listing GO:0005886 (IBA) conflicts with the live QuickGO set, which is reported as a database-lag caveat rather than resolved definitively.