HSPA13 (STCH) Localization Hypothesis — Final Curation Report

Gene: HSPA13 / STCH (human) · UniProt: P48723 · Taxon: Homo sapiens (NCBITaxon:9606)
Focus type: function_assignment · Hypothesis slug: conditional-nuclear-cytosolic-and-surface-pools
Source file: genes/human/HSPA13/HSPA13-ai-review.yaml

Summary

Verdict: Over-annotated / refuted for the nuclear and plasma-membrane pools; unresolved (indirect inference only) for the cytosolic pool.

The seed hypothesis proposes that human HSPA13 localizes — as independently demonstrable pools — to the nucleus, plasma membrane, or cytosol, and asks whether ER-luminal proximity-labeling evidence can be reconciled with a reported NKCC2 cytoplasmic-tail interaction. After auditing the actual GO annotation provenance, the primary experimental literature, and the protein's sequence architecture, the conclusion is that the nuclear, plasma-membrane, and cytosolic Cellular Component (CC) annotations on P48723 are IBA (Inferred from Biological Ancestor) phylogenetic inferences propagated from the broader HSP70 family, not HSPA13-specific experimental findings. None of these three secondary pools is supported by an IDA/IMP/HDA/TAS experimental annotation for HSPA13 itself, and the Human Protein Atlas reports no confident immunofluorescence subcellular localization.

The directly supported localization for HSPA13/STCH is the endoplasmic reticulum / microsome / secretory pathway. This is established by (1) a cleaved N-terminal signal peptide (residues 1–22) and the absence of any transmembrane segment; (2) immunofluorescence plus subcellular fractionation showing microsomal enrichment (PMID: 8131751); (3) an independent description of STCH as "a microsomal protein" (PMID: 18793616); and (4) an HDA extracellular-exosome annotation (PMID: 19199708) consistent with secretory/endosomal egress rather than a nuclear or cytosolic functional pool.

The reported NKCC2 interaction (PMID: 33672238) is with the ER-resident (immature) form of NKCC2 in an ERAD context — it locates STCH in the ER biogenesis/quality-control machinery and does not demonstrate a cytosolic or cytoplasmic-tail-facing pool. The one genuine, unresolved topological question is raised by PMID: 10675567, which shows cytosolic ubiquitin-like proteins (UBQLN/BAG6-family homologs) binding a peptide within the STCH ATPase domain — a binding assay, not an in-situ localization. Overall: retain/strengthen ER; flag nucleus, plasma membrane, and cytosol as paralog carry-over lacking direct support.

Executive Judgment

Partially supported / over-annotated → the secondary pools were not demonstrated. Evaluated as a function-assignment (CC) hypothesis, HSPA13 does not directly and demonstrably carry the nucleus, plasma-membrane, or cytosol GO terms. Those terms exist on the record only as IBA inferences transferred from cytosolic/nuclear HSP70 paralogs. The gene product's directly supported location is the ER/microsome/secretory pathway. The cytosolic pool is the only one with a mechanistic hint (UbL/BAG6 binding to the ATPase core), but that hint is unresolved and does not rise to a demonstrated pool. The most important caveat is that IBA annotations are legitimate GO evidence; the issue is that for a divergent family member with its own experimentally supported compartment, blanket transfer of generic HSP70 localizations is a systematic over-annotation risk.

Key Findings

Finding 1 — The nucleus, plasma-membrane, and cytosol CC terms are IBA paralog carry-over, not HSPA13 evidence

An annotation-provenance audit of P48723 (via QuickGO/UniProt, executed with requests) shows the following Cellular Component landscape:

Critically, the "with/from" support panels for the nucleus, cytosol, cytoplasm, and plasma-membrane IBA annotations are dominated by other HSP70 family members — P11142 (HSPA8/HSC70), P11021 (HSPA5/BiP), P17066, P34931, and yeast/worm/plant orthologs. These are cytosolic and nuclear-cytoplasmic Hsp70 paralogs whose localizations were transferred onto HSPA13 by the Reference Genome phylogenetic annotation pipeline. No IDA, IMP, HDA, or TAS experimental annotation supports nucleus, plasma membrane, or cytosol for HSPA13 specifically. Consistent with this, the Human Protein Atlas reports no confident subcellular immunofluorescence localization for HSPA13 (Subcellular location => None), so there is no independent imaging support for the secondary pools.

Curation consequence: IBA annotations are legitimate GO evidence, but for a divergent family member with its own experimentally supported compartment, transferring generic HSP70 cytosolic/nuclear localizations risks a systematic over-annotation. These three CC terms should be flagged as non-core / paralog-derived.

Finding 2 — The primary experimental localization is ER / microsome, via a cleaved signal peptide

UniProt P48723 feature annotation shows a signal peptide at residues 1–22 (cleaved), a mature chain at 23–471, and no transmembrane domain. This is the canonical architecture of a soluble secretory-pathway (ER-luminal) protein, not a cytosolic Hsp70.

The foundational primary paper, Otterson et al. 1994 (PMID: 8131751, "Stch encodes the 'ATPase core' of a microsomal stress 70 protein"), reports that STCH migrates predominantly as a ~60 kDa species enriched in a membrane-bound microsome fraction (immunofluorescence + subcellular fractionation); has a unique hydrophobic leader sequence; is calcium (A23187)-inducible but NOT heat-inducible; and has a truncated C-terminal peptide-binding domain with peptide-independent ATPase activity.

"Immunofluorescent and subcellular analyses show that STCH migrates predominantly as a 60 kDa species and is enriched in a membrane-bound microsome fraction." — PMID:8131751 (verified snippet)

Yamagata et al. 2008 (PMID: 18793616) independently describe STCH as "a microsomal protein that contains a N-terminal ATPase domain but lacks a C-terminal protein binding domain," reinforcing the ER/secretory assignment.

"a member of the heat shock protein 70 (HSP70) superfamily, is a microsomal protein that contains a N-terminal ATPase domain but lacks a C-terminal protein binding domain" — PMID:18793616 (verified snippet)

The HDA extracellular-exosome annotation (PMID:19199708) fits the same secretory/endosomal narrative: an ER-luminal chaperone can egress via multivesicular bodies into exosomes (cf. the urinary-exosome proteomics context of PMID: 19056867). This is not evidence of a nuclear or cytosolic functional pool.

Finding 3 — The NKCC2 interaction is an ER/ERAD context, not a demonstrated cytosolic pool

Bakhos-Douaihy et al. 2021 (PMID: 33672238) identify STCH as a binding partner of the ER-resident (immature, ER-retained) form of NKCC2/SLC12A1, promoting its ER-associated degradation (proteasomal plus ER-to-lysosome routing) and impairing NKCC2 maturation and cell-surface expression.

"we report the identification of chaperone stress 70 protein (STCH) and the stress-inducible heat shock protein 70 (Hsp70), as two novel binding partners of the ER-resident form of NKCC2" — PMID:33672238 (verified snippet)

This is an ER-lumen / ER-membrane biogenesis context. The co-immunoprecipitation does not map the STCH contact to the NKCC2 cytoplasmic tail specifically, and it does not establish that STCH is present in the cytosol. The seed hypothesis's framing — that a "NKCC2 cytoplasmic-tail interaction" competes with ER-luminal evidence — is not supported by the primary data: the interaction is with the ER-resident client during quality control, fully consistent with an ER-luminal STCH pool acting on nascent/misfolded NKCC2.

Separately, Kaye et al. 2000 (PMID: 10675567) isolated two human ubiquitin-like (UbL) proteins (Chap1/Chap2 — UBQLN/Dsk2 and Scythe/BAG6-family homologs) that bind a short peptide within the ATPase domain of STCH.

"We have isolated two human ubiquitin-like (UbL) proteins that bind to a short peptide within the ATPase domain of the Hsp70-like Stch protein." — PMID:10675567 (verified snippet)

Because UBQLN/BAG6-family proteins are cytosolic, this interaction would require cytosolic exposure of the STCH ATPase core. However, it was demonstrated by in-vitro binding/isolation assays, not by in-situ localization. It is a legitimate unresolved topology question — not a demonstrated cytosolic pool.

Finding 4 — Sequence architecture is secretory (BiP-like), lacking the cytosolic-Hsp70 EEVD/TPR signature

A computed sequence audit (UniProt FASTA; Needleman–Wunsch alignment in numpy) shows HSPA13's architecture aligns with ER/secretory Hsp70s, not cytosolic ones:

Key implications:

• HSPA13 begins with a hydrophobic ER signal, like BiP — not with the cytosolic Hsp70 nucleotide-binding-domain motif GIDLGTTYSCV that opens HSC70 and HSPA1A.
• HSPA13 lacks the C-terminal EEVD motif that is the docking signature for cytosolic TPR co-chaperones (Hop, CHIP). This directly undercuts the assumption that HSPA13 behaves like a cytosolic Hsp70 in the cytoplasm/nucleus.
• HSPA13 also lacks KDEL/HDEL, so it is not a classically ER-retained luminal chaperone in the BiP sense — consistent with its being a divergent, secretory-competent, exosome-detectable protein with a truncated substrate-binding domain and a ~50-aa ATPase insertion.
• Global identity to each reference (~29–31%) shows HSPA13 is a divergent, roughly equidistant family member — precisely the situation in which uncritical IBA transfer of paralog localizations is most error-prone.

Mechanistic Model / Interpretation

The evidence converges on a single coherent model:

       HSPA13 / STCH  (P48723)
  divergent HSP70-superfamily member
signal peptide 1–22 · truncated SBD · Ca2+-inducible · not heat-inducible

   ┌───────────────────────────── ER / MICROSOME (PRIMARY) ─────────────────────────────┐
   │  • Cleaved N-terminal signal peptide targets it to the secretory pathway            │
   │  • Microsomal enrichment by IF + fractionation (PMID 8131751)                        │
   │  • "microsomal protein" (PMID 18793616)                                              │
   │  • Binds ER-resident (immature) NKCC2 → ERAD / quality control (PMID 33672238)       │
   │  • ATPase-driven, peptide-independent activity (truncated SBD)                       │
   └──────────────────────────────────────────┬──────────────────────────────────────────┘
                               │ secretory / MVB egress
                               ▼
                  EXTRACELLULAR EXOSOME (HDA, PMID 19199708)

   ┌──────────────── SECONDARY POOLS AS ANNOTATED (nucleus / PM / cytosol) ──────────────┐
   │  • All three = IBA (GO_REF:0000033), "with/from" = HSPA8, HSPA5, HSPA6, orthologs     │
   │  • No IDA/IMP/HDA/TAS for HSPA13 · HPA subcellular = None                             │
   │  → PARALOG CARRY-OVER, not demonstrated                                               │
   └──────────────────────────────────────────────────────────────────────────────────────┘

   ┌──────────────── UNRESOLVED TOPOLOGY QUESTION (cytosol only) ─────────────────────────┐
   │  • Cytosolic UbL/BAG6 (UBQLN, Scythe/BAG6) bind STCH ATPase peptide (PMID 10675567)   │
   │  • Would require cytosolic exposure of ATPase core — shown by binding assay, not      │
   │    by in-situ localization → hypothesis, not established pool                          │
   └──────────────────────────────────────────────────────────────────────────────────────┘

In plain terms: HSPA13 is an ER/microsomal, secretory-pathway member of the HSP70 superfamily that participates in ER protein quality control (illustrated by its role in NKCC2 ERAD) and can be exported into exosomes. Its nuclear, plasma-membrane, and cytosolic GO annotations are artifacts of phylogenetic (IBA) transfer from its cytosolic paralogs and are not backed by any HSPA13-specific experiment. The only mechanistic hint toward a cytosol-facing role is the UbL/BAG6 binding to the ATPase core, which raises — but does not resolve — a question about whether some of the ATPase domain is ever cytosolically exposed (e.g., during retrotranslocation or in an alternate topology). That question deserves a targeted experiment before any cytosolic annotation is asserted as HSPA13-specific.

Evidence Matrix

Evidence Base

Primary localization anchors. Stch encodes the 'ATPase core' of a microsomal stress 70 protein (PMID: 8131751) is the single most important paper: it provides direct IF and fractionation evidence for microsomal enrichment and describes the defining architecture (hydrophobic leader, truncated substrate-binding domain, Ca2+-inducibility, peptide-independent ATPase). Stomach cancer-derived del223V-226L mutation of the STCH gene (PMID: 18793616) independently labels STCH "a microsomal protein," corroborating the ER/secretory assignment while linking STCH to TRAIL-mediated apoptosis sensitization.

Interaction context. Differential Effects of STCH and Stress-Inducible Hsp70 on the Stability and Maturation of NKCC2 (PMID: 33672238) places STCH firmly in ER quality control by identifying it as a partner of the ER-resident form of NKCC2 that promotes ERAD. This is the paper the seed hypothesis invokes as a "cytoplasmic-tail interaction," but the primary text describes an ER-resident-client interaction, not a cytosolic tail contact. A family of ubiquitin-like proteins binds the ATPase domain of Hsp70-like Stch (PMID: 10675567) is the sole line of evidence hinting at cytosolic exposure of STCH's ATPase core; it is a binding study, so it raises rather than resolves the cytosol question.

Secretory egress. The extracellular-exosome HDA annotation (PMID:19199708; and the related urinary-exosome proteomics context of PMID: 19056867) is consistent with an ER/secretory protein being exported via multivesicular bodies.

Peripheral transcriptomic mentions. HSPA13 appears as a differentially expressed gene in estrogen-receptor and breast-cancer studies (PMID: 32387340; PMID: 25541910). These are expression/regulation observations and carry no localization information; they do not bear on the CC hypothesis.

GO Curation Implications

Leads requiring curator verification (not automated changes):

1. Retain / strengthen GO:0005783 endoplasmic reticulum (CC). Currently IEA (UniProtKB-SubCell). It is backed by direct experimental localization (PMID:8131751) and an independent microsomal description (PMID:18793616). Consider upgrading the evidence basis by citing PMID:8131751 as experimental (IDA-type) support and/or annotating microsome/ER lumen more specifically. The signal peptide + truncated SBD architecture supports an ER/secretory chaperone role.

2. Flag GO:0005634 nucleus, GO:0005886 plasma membrane, and GO:0005829 cytosol (all IBA) as paralog carry-over. They lack any HSPA13-specific experimental support, and their "with/from" panels are dominated by cytosolic/nuclear HSP70 paralogs. These are candidates to be removed, or retained only as non-core IBA with an explicit curator note that they are phylogenetic inferences not demonstrated for HSPA13. This directly answers the seed hypothesis: the secondary pools were not actually demonstrated.

3. GO:0070062 extracellular exosome (HDA) may be retained as-is; it reflects secretory/exosomal egress, not a nuclear/cytosolic functional pool.

4. Do not add a cytosol annotation on the strength of PMID:10675567 alone; the topology is inferred from a binding assay. If a BP/MF term is desired to capture the ER-quality-control role, an ERAD / ER protein quality-control BP term supported by PMID:33672238 is a better-informed lead than a generic "protein binding" MF term.

Mechanistic Scope

The immediate molecular activity under test is subcellular localization (a CC assignment), coupled to HSPA13's role as an ATP-dependent HSP70-superfamily chaperone with a truncated, peptide-independent substrate-binding capacity operating in the ER/secretory pathway. The directly demonstrated gene-product-level facts are: (a) signal-peptide-directed ER/microsomal residence; (b) participation in ER quality control of a membrane client (NKCC2) via ERAD. Everything else attributed by the seed hypothesis — nuclear, plasma-membrane, cytosolic pools — is either downstream inference (IBA transfer) or an unproven topology question (UbL/BAG6 binding). The breast-cancer/estrogen expression signals are transcriptional responses, not localization, and are outside the mechanistic scope of this CC hypothesis.

Conflicts and Alternatives

• Paralog confusion (primary alternative). The nucleus/PM/cytosol annotations are the expected localizations of cytosolic HSP70s (HSPA8/HSC70, HSPA1A) and were transferred to HSPA13 by IBA. This is the most parsimonious explanation for the secondary pools and is the leading alternative to a genuine multi-compartment HSPA13.
• Isoform/construct caveats. The seed hypothesis rightly asks about isoforms/constructs. Overexpressed or signal-peptide-deleted constructs could mislocalize; none of the primary localization papers here relied on tagged over-expression to claim ER (PMID:8131751 used endogenous-scale IF + fractionation). No isoform-specific nuclear/cytosolic localization was found in the reviewed literature.
• Topology question (genuine). PMID:10675567's cytosolic UbL binding to the ATPase core is a real conflict with a strictly ER-luminal model and cannot be dismissed; it is simply unproven by localization.
• Database carry-over. The UniProt "plasma membrane" line and the QuickGO IBA set are the database-level source of the hypothesis; they should be treated as orientation, not evidence.

Limitations and Knowledge Gaps

1. No high-resolution HSPA13-specific imaging. HPA reports no confident IF localization; PMID:8131751 imaging is 1994-era. What to resolve: endogenous-tag or validated-antibody confocal/immuno-EM across ER, nucleus, PM, cytosol.
2. Cytosolic exposure of the ATPase core is unresolved. PMID:10675567 shows binding but not localization. Why it matters: it is the only mechanistic basis for any cytosol annotation. Resolve with: proximity labeling (BioID/APEX at ER lumen vs cytosol), protease-protection topology assays.
3. NKCC2 contact site not mapped. PMID:33672238 does not localize the STCH–NKCC2 interface to the cytoplasmic tail. Resolve with: domain-mapping co-IP / crosslinking to determine whether the contact is luminal or cytosolic.
4. No proximity-labeling dataset for HSPA13 was directly accessible in this run. The seed hypothesis references "ER-luminal proximity-labeling evidence"; that specific dataset was not retrievable here and should be located and cited directly by the curator.
5. Exosome route vs plasma-membrane residence not disentangled. Exosomal detection (PMID:19199708) does not equal a stable plasma-membrane pool.

Discriminating Tests

1. Split proximity labeling: APEX2/TurboID fused to HSPA13 (respecting the signal peptide) plus orthogonal ER-lumen, cytosol, and nuclear baits — directly quantifies which compartments HSPA13 samples.
2. Protease-protection / topology assay on microsomes ± detergent to test whether any HSPA13 ATPase epitope is cytosolically exposed (tests the PMID:10675567 implication).
3. Endogenous CRISPR knock-in tag (e.g., HaloTag) + confocal and immuno-EM to score nucleus/PM/cytosol pools without over-expression artifacts.
4. Signal-peptide-deletion construct to determine whether removing ER targeting redirects HSPA13 to cytosol/nucleus — clarifying whether any secondary pool is physiological or construct-dependent.
5. NKCC2 interface mapping (truncation/crosslinking co-IP) to test the "cytoplasmic-tail interaction" claim explicitly.

Proposed Follow-up Experiments / Actions (Curation Leads)

Candidate references + exact snippets to verify:

• PMID:8131751 — "...STCH migrates predominantly as a 60 kDa species and is enriched in a membrane-bound microsome fraction." → supports ER/microsome CC.
• PMID:18793616 — "...is a microsomal protein that contains a N-terminal ATPase domain but lacks a C-terminal protein binding domain" → supports ER; supports truncated-SBD MF nuance.
• PMID:33672238 — "...STCH...as two novel binding partners of the ER-resident form of NKCC2" → ER/ERAD context; does not support cytosol.
• PMID:10675567 — "...ubiquitin-like (UbL) proteins that bind to a short peptide within the ATPase domain of the Hsp70-like Stch protein." → unresolved cytosolic-topology question only.

Candidate GO actions:

• Retain/strengthen GO:0005783 (ER, CC) with experimental citation PMID:8131751; consider a more specific ER-lumen/microsome term.
• Flag as non-core / candidate removal GO:0005634 (nucleus), GO:0005886 (plasma membrane), GO:0005829 (cytosol) — all IBA paralog carry-over, no HSPA13 experimental support.
• Retain GO:0070062 (exosome, HDA).
• Consider an ERAD / ER protein quality-control BP term supported by PMID:33672238 instead of leaving only generic protein binding.

Suggested curator questions: Which specific proximity-labeling dataset underlies the "ER-luminal" claim in the seed? Is any HSPA13 isoform annotated without the signal peptide? Does any experimental (non-IBA) source place HSPA13 in the nucleus/PM/cytosol?

Suggested experiments: the five discriminating tests above, prioritizing split proximity labeling and topology/protease-protection.

Bottom line

HSPA13/STCH is an ER/microsomal, secretory-pathway HSP70-superfamily chaperone (signal peptide, no TM, no EEVD, no KDEL, truncated SBD) whose nucleus, plasma-membrane, and cytosol GO CC annotations are IBA phylogenetic carry-over from cytosolic paralogs with no HSPA13-specific experimental support. The NKCC2 interaction is an ER-resident/ERAD context and does not demonstrate a cytosolic pool; only cytosolic UbL/BAG6 binding to the ATPase core (a binding assay) leaves a genuine but unproven cytosolic-topology question. Recommendation: retain/strengthen ER; flag the three secondary pools as paralog carry-over lacking direct support.

Artifacts

• OpenScientist final report
• OpenScientist final report