Unfolded Protein Binding

Annotation Review

Reclassifying GO:0051082 & GO:0031249

Chris Mungall | AI-Assisted Gene Review
2026-10-04 refresh

The Problem

GO:0051082 "unfolded protein binding" and GO:0031249 "denatured protein binding" were obsoleted (go-ontology#30962)

These terms conflate mechanistically distinct activities under a single "binding" label:

  • ATP-dependent foldases that actively refold clients
  • ATP-independent holdases that prevent aggregation in situ
  • Disaggregases that solubilize existing aggregates
  • Sensors that recognize misfolded proteins for degradation
  • Co-chaperones that modulate chaperone ATPase cycles

The Holdase-Foldase Network

Conserved across all domains of life: stress unfolds proteins; sHSPs (holdases) capture them to prevent aggregation; clients are handed off to Hsp70 (foldase) for ATP-dependent refolding, or to Hsp100 (disaggregase). Eukaryotes also use Hsp90 and Hsp60/CCT.

Figure 3 from Luo et al. (2022) Front Mol Biosci 9:832160, CC-BY 4.0

The Proteostasis Triage Decision

(A) HSP70-HOP-HSP90 axis: fold-forward pathway to native protein.
(B) HSP70-CHIP axis: ubiquitination and proteasomal degradation.

Co-chaperones (DNAJ, NEFs, BAG1/3) tune the fold vs degrade decision.

Figure 2 from Radons (2020) Cells 9:587, CC-BY 4.0

Recent Literature Context

Key reviews informing this reclassification:

  • Mitra et al. (2022) Annu Rev Biophys 51:409 — ATP-independent chaperones; force-clamp experiments show context-dependent holdase/foldase switching
  • Bhattacharjee et al. (2025) J Biosci — Co-chaperones fine-tune HSP70: fold, hold, or degrade; O-GlcNAc modification of HSP27 promotes BAG3-mediated refolding
  • Boopathy (2024) — HUWE1 + HSP70 cooperate to clear nuclear inclusions, linking holdase activity to E3-mediated clearance
  • Le et al. (2024) Mol Cells — UBR1/UBR2 as ER-stress sensors: substrate load stabilizes these E3s to enhance PQC

Phase-1 Scope

Metric Count
Unique genes in frozen audit 148
Human genes (primary) 33
Non-human genes 115
Species covered 17
Total annotations reviewed 5,529
Remaining PENDING 0

Feb 2026 action snapshot; the live registry has since grown through follow-up
case studies and identity corrections.

Mechanism Classes

Class GO term ATP? Example
Foldase GO:0044183 Yes GroEL/ES, TRiC/CCT
Holdase NTR needed No CRYAB, CLU
Carrier-holdase GO:0140309 No Tim9-Tim10
Foldase/holdase GO:0044183 + holdase NTR Yes HSPA1A (HSP70)
Co-chaperone (see gap) N/A DNAJB1, AHSA1
Disaggregase GO:0140545 Yes HSP104, HSPA1A
Sensor NTR proposed N/A SYVN1, UGGT1

Decision Rules

Mechanism Action Replacement
Foldase (GroEL, TRiC) MODIFY GO:0044183
Foldase/holdase (HSP70) MODIFY GO:0044183 (holdase NTR pending)
Co-chaperone, J-domain MODIFY GO:0044183 (interim)
Holdase (sHSP, crystallin) MODIFY proposed holdase chaperone activity
Disaggregase MODIFY GO:0140545
ER/QC sensor REMOVE (not chaperones)
HSP90 co-chaperone OVER_ANNOTATED (not direct UPB)

Impact: Human Genes (n=33)

Primary action Count Notes
MODIFY to GO:0044183 (foldase) 16 HSP70 (6), J-domain interim (4), prefoldin (6)
MODIFY to proposed holdase chaperone activity 6 sHSPs, CLU, DNAJB6, DNAJB8
MODIFY to other specific MF 3 NPM1, AIP, SCG5
MARK_AS_OVER_ANNOTATED 5 Sensor/co-chaperone cases
REMOVE 3 SYVN1, ERLEC1, GRPEL1

Plus 3 co-annotations to GO:0140545 (disaggregase): HSPA1A, HSPA1B, HSPA8

Before/After Examples

Gene Before After Why
HSPA1A GO:0051082 GO:0044183 + GO:0140545 ATP-dependent foldase + disaggregase
CRYAA GO:0051082 proposed holdase chaperone activity sHSP holdase; prevents aggregation in situ
DNAJB1 GO:0051082 GO:0044183 (interim) J-domain co-chaperone, not independent foldase
SYVN1 GO:0051082 REMOVE E3 ligase; recognizes misfolded substrates
NPM1 GO:0051082 GO:0140713 Histone chaperone; UPB was secondary

Critical Finding: The Holdase Gap

GO:0140309 "unfolded protein holdase activity" does not fit most holdases.

  • Created Nov 2025 for TIM carrier-holdases (Tim9-Tim10) in go-ontology#30552
  • Definition still requires escort "to an acceptor molecule or to a specific location"
  • It remains a child of GO:0140597 "protein carrier chaperone"
  • The primary holdase label outgrew the carrier-specific definition

6 human genes + HSPH1 are in-situ holdases that prevent aggregation without delivery to a defined acceptor or destination. They cannot use GO:0140309.

Proposed: "Holdase Chaperone Activity"

Definition: Binding to an unfolded or misfolded protein to prevent its aggregation without actively catalyzing refolding. The holdase maintains the client protein in a soluble, folding-competent state.

Parentage: Direct child of GO:0003674 (molecular_function).
GO:0140309 (carrier-holdase) becomes a child of this new term.

Affected genes: CRYAA, CRYAB, HSPB6, CLU, DNAJB6, DNAJB8, HSPH1

Remaining in-situ holdase migrations are tracked in ai-gene-review#2222.

Other Ontology Gaps

Misfolded protein sensor activity

  • Recognition of misfolded conformation for QC degradation
  • CHIP/STUB1 ubiquitinates chaperone-bound clients; UBR1/UBR2 recognize N-degrons (Le et al. 2024)
  • Affects: SYVN1, SAN1 (yeast), Fbxo2 (mouse)

Co-chaperone MF representation

  • GO:0003767 "co-chaperone activity" deliberately obsoleted
  • No MF term for J-domain co-chaperone function (ATPase activation + substrate delivery)
  • GO:0044183 used as pragmatic interim
  • Affects all J-domain proteins across species

Phase-1 Cross-Species Stress Test

The phase-1 audit applied the human-derived categories across 17 species and
exposed bacterial, fungal and metazoan edge cases.

Species Genes Highlights
S. cerevisiae 67 Most mechanism classes represented
E. coli 13 Periplasmic holdases (SurA, Skp, Spy, HdeA/B)
D. melanogaster 7 sHSPs Hsp22-27
M. musculus 6 Hspa8 largest review (240 annotations)
R. norvegicus 4 Hspa5/BiP (101 annotations)
Other (12 spp.) 18 Zebrafish crystallins, CnoX redox holdase

Notable Cross-Species Findings

  • SlrP (S. typhimurium): misannotation removed — T3SS effector E3 ligase, not a chaperone
  • CnoX (E. coli): redox-activated holdase — becomes active under oxidative stress when Cys residues are oxidized
  • Peroxiredoxins (TSA1, pmp20, tpx1): dual-function enzymes with stress-dependent holdase activity
  • Assembly factors (ATP10, PET100, SHY1): single-client chaperones, not general UPB
  • IRE1 (yeast + T. reesei): UPR sensor, not chaperone — cross-kingdom confirmation

What We Need from GO Editors

  1. Holdase NTR (UNRESOLVED): Create "holdase chaperone activity" for in-situ holdases
  2. Repair obsoletion guidance for obsolete GO:0051082 / GO:0031249
  3. Preferred labels: "foldase" exact synonym on GO:0044183; reserve plain "holdase" for the general term
  4. Co-chaperone MF gap: How should J-domain function be annotated?
  5. HSP70 dual annotation: Confirm foldase + holdase per experimental context
  6. Misfolded protein sensor: New term for SYVN1, SAN1, Fbxo2?

Summary

148 phase-1 genes across 17 species, 5,529 annotations reviewed

The key bottleneck is the holdase NTR — without it, six human
in-situ holdases plus HSPH1 lack a valid GO replacement after GO:0051082 and GO:0031249 obsoletion.

All gene review YAMLs: genes/<SPECIES>/<GENE>/
Validate: just validate-all