Verified 2026-05-07 against primary source PDFs (Whitesell 1994, Taipale 2012, Walther 2015, Schopf/Biebl & Buchner 2019) and canonical DBs (UniProt REST, NCBI Gene, HGNC REST, Open Targets Platform). Rutherford & Lindquist 1998 and Hipp 2019 not_oa — qualitative claims from these sources retained with gap tags. See log.md for corrections made.
HSP90 (HSP90AA1 — inducible cytosolic isoform)
Paralog disambiguation: This page covers HSP90AA1 (P07900), the stress-inducible cytosolic alpha isoform. For the constitutive cytosolic beta isoform (HSP90AB1, P08238), the ER-resident paralog (GRP94/HSP90B1, P14625), and the mitochondrial paralog (TRAP1, Q12931), see the paralog table below. Those three paralogs do not yet have dedicated wiki pages — implicit stubs flagged in summary.
HSP90 (heat shock protein 90) is one of the most abundant cytosolic proteins in eukaryotic cells, constituting 1–2% of total cellular protein under basal conditions and rising further under stress 1. It operates as an obligate homodimer and is the central hub of a large chaperone network: whereas HSP70 handles the bulk of newly synthesized polypeptides and thermally denatured proteins, HSP90 specializes in the late-stage maturation and conformational activation of a specific, highly selective client proteome — principally protein kinases, transcription factors, and steroid hormone receptors in their near-native but not fully active states 2. In aging, HSP90 becomes a critical node in the proteostasis collapse hypothesis: as HSF1-driven transcription weakens with age and the burden of misfolded or aggregation-prone proteins grows, the chaperone buffering capacity of the HSP90/HSP70 network is progressively overwhelmed 3.
HSP90 paralogs in humans
| Paralog | UniProt | Gene | Compartment | Expression pattern | Wiki page |
|---|---|---|---|---|---|
| HSP90AA1 (HSP90α, HSP86) | P07900 | HSP90AA1 | Cytosol/nucleus | Stress-inducible + constitutive; HSF1-driven surge under stress | this page |
| HSP90AB1 (HSP90β, HSC84) | P08238 | HSP90AB1 | Cytosol/nucleus | Constitutive; minimally stress-inducible; more abundant than HSP90AA1 at baseline | no page yet — stub |
| GRP94 (HSP90B1, endoplasmin) | P14625 | HSP90B1 | ER lumen | ER-resident; UPR-induced; clients include integrins and TLRs | no page yet — stub |
| TRAP1 (HSP90L) | Q12931 | TRAP1 | Mitochondrial matrix | Mitochondrial; involved in metabolic regulation and UPRmt | no page yet — stub |
The two cytosolic isoforms (alpha and beta) share ~85% sequence identity and overlapping client portfolios but differ in regulation: HSP90AA1 is strongly stress-inducible via HSF1, while HSP90AB1 maintains a large constitutive pool. Most pharmacological HSP90 inhibitors cannot discriminate between the two cytosolic isoforms. This page treats HSP90AA1 as the primary canonical entity given its direct regulation by hsf1 and greater stress-responsiveness, which are most relevant to the aging biology context.
Domain architecture and chaperone cycle
HSP90 is a 732-residue protein (~90 kDa) organized into three domains that drive an ATP-dependent conformational cycle 1:
| Domain | Approximate residues | Function |
|---|---|---|
| N-terminal domain (NTD) | 1–236 | ATP/ADP binding (Bergerat fold); site of competitive inhibition by geldanamycin/17-AAG/ganetespib |
| Middle domain (MD) | 271–616 | Client protein binding surface; co-chaperone interaction; catalytic contribution to ATP hydrolysis |
| C-terminal domain (CTD) | 628–732 | Homodimerization; MEEVD motif (residues 727–732) docking site for TPR-domain co-chaperones (HOP/STIP1, CHIP/STUB1, PPP5C) |
The HSP90 chaperone cycle proceeds through well-defined conformational states 1:
- Open/apo state: The HSP90 homodimer adopts an open V-shaped conformation with low ATPase activity; clients can bind in this state.
- ATP binding: ATP binding to the NTD initiates transient NTD dimerization (“closed” state), trapping the client in a partially structured conformation.
- Co-chaperone modulation: AHA1 (AHSA1) binds the MD and stimulates ATPase activity; p23 (PTGES3) binds the closed NTD dimer and stabilizes it, slowing ATP hydrolysis and prolonging client dwell time; CDC37 (kinase-specific co-chaperone) stabilizes kinase clients during loading.
- ATP hydrolysis: Hydrolysis of ATP to ADP + Pi drives the power stroke that releases the client in its mature conformation.
- Client release: ADP exchange (facilitated by NEFs or spontaneous dissociation) resets the cycle.
HOP (STIP1) serves as the physical adaptor between the HSP70 and HSP90 chaperone systems: HOP binds simultaneously to the EEVD motifs of HSP70 and HSP90 via its TPR domains, enabling hand-off of partially folded clients from HSP70 to HSP90 for late-stage maturation 1.
Key PTMs (HSP90AA1, UniProt P07900):
- Phosphorylation: Ser-231, Ser-252, Ser-263, Thr-5, Thr-7 (contextual; stress-regulated)
- Acetylation: Lys-58, Lys-84, Lys-443, Lys-458, Lys-489, Lys-585 — multiple sites; acetylation state modulates co-chaperone binding and client affinity
- S-nitrosylation: Cys-598 — nitric oxide modification; inhibits ATPase activity in some contexts
Client proteome
HSP90 is distinctive among chaperones for the selectivity and identity of its clients. Unlike HSP70 — which recognizes exposed hydrophobic stretches in any unfolded protein — HSP90 binds “metastable” clients: proteins that are nearly folded but require HSP90-assisted conformational activation to reach their active states 2.
Taipale et al. 2012 quantified the HSP90 client interaction network using the LUMIER/BACON in-cell luminescence assay (not mass spectrometry/crosslinking), systematically testing 420 kinase clones (355 unique proteins, 69% of human kinome), 1,303 TF clones (843 unique, 79% of human TFs), and 498 E3 ligase clones (426 unique). The study identified “almost 400 client proteins” total, including 193 kinases (61% of the kinase clones tested, representing >half the human kinome) 2:
| Client class | Interaction rate (Taipale 2012) | Examples | Functional consequence of HSP90 inhibition |
|---|---|---|---|
| Protein kinases | 61% of tested kinase clones (193/420 clones; 193 of 355 unique kinases) | Akt (akt), Raf, CDK4, HER2/ERBB2, MET, EGFR | Kinase destabilization and proteasomal degradation |
| Transcription factors | ~7% of tested TFs (58/843 unique TFs) — much lower than previously assumed | HSF1 (hsf1), HIF-1α, p53 (p53), AR, GR | Loss of TF activation capacity |
| Steroid hormone receptors | Subset of TF group; historically best-studied HSP90 clients | GR (NR3C1), ER-α, PR, AR | Receptor destabilization; hormone unresponsiveness |
| E3 ubiquitin ligases | 31% of tested E3 ligases (117/372 unique) | MDM2, CHIP/STUB1 | Impaired ubiquitin pathway function |
| Kinase-regulatory proteins | (not quantified in this study) | CDC37, TSC1/TSC2 | Signaling cascade disruption |
This broad kinome coverage makes HSP90 a central “hub” in signal transduction: inhibiting HSP90 simultaneously degrades dozens of oncoproteins, which is both its therapeutic appeal in cancer and the mechanistic basis for its many adverse effects 4.
The metastable client concept and aging: Aging-associated changes in protein stability (accumulated somatic mutations, oxidative damage, post-translational noise) push many client proteins toward the unstable side of the metastability spectrum, increasing dependence on HSP90 buffering 3. This creates a “chaperone addiction” phenotype in aged cells analogous to non-oncogene addiction in cancer cells.
Aging biology
Proteostasis buffering capacity and collapse
HSP90 is a core component of what Hipp, Kasturi & Hartl (2019) call the “proteostasis network” — the integrated system of chaperones, disaggregases, the ubiquitin-proteasome system (UPS), and autophagy that collectively maintain proteome integrity 3. In this framework, HSP90 and HSP70 together constitute the primary chaperone buffering capacity for the cytosolic proteome.
Two lines of evidence link HSP90/chaperone network capacity to aging:
-
Age-associated insolubilization: Walther et al. 2015 performed SILAC quantitative proteomics across the lifespan of C. elegans N2 (WT), profiling >5,000 proteins across days 1–22. Of 1,083 proteins quantified in ≥3 of 4 experiments, 975 accumulated significantly in the insoluble fraction by day 12. Aggregation initiates mainly after day 6 of adulthood (post-reproductive), not at reproductive maturity per se 5. The major driver of aggregate load is protein abundance exceeding critical solubility thresholds (“supersaturation”), not intrinsic aggregation propensity or a specific chaperone substrate signature — abundant proteins dominate aggregate mass. Small HSPs are markedly enriched in aggregates and the paper argues this represents a protective sequestration strategy to slow proteostasis decline, not simple chaperone network collapse. Not yet replicated at equivalent proteome-wide resolution in humans needs-human-replication.
-
HSF1 decline: HSF1 transcriptional output — the primary driver of stress-inducible HSP90AA1 and HSP70 expression — declines with age (see hsp70 for the quantitative data from Heydari 1993 and Locke 1996). The consequence for HSP90 specifically is reduced induction capacity for the inducible alpha isoform; the constitutive HSP90AB1 isoform is less affected, but constitutive levels alone are insufficient to handle the increased misfolded-protein burden of aged cells.
| Dimension | Status | Notes |
|---|---|---|
| Pathway conserved in humans? | yes | HSP90 chaperone cycle and HSF1 regulation highly conserved |
| Phenotype conserved in humans? | partial | Protein aggregation in aging demonstrated in C. elegans; human proteome-wide aging insolubility not yet shown at equivalent resolution |
| Replicated in humans? | no | needs-human-replication |
HSP90 as an evolutionary capacitor
Rutherford & Lindquist (1998) demonstrated a foundational principle: HSP90, by buffering the effects of genetic variants that would otherwise alter protein folding, acts as an evolutionary “capacitor” — masking genetic variation under normal conditions and releasing it when chaperone capacity is compromised (by stress, pharmacological inhibition, or aging) 6. When HSP90 function is reduced, cryptic genetic variation is exposed, producing novel phenotypes.
This capacitor function has direct relevance to aging: as HSP90 buffering capacity declines with age (through HSF1 attenuation and increased client load), previously tolerated “cryptic” protein variants may become destabilizing, contributing to age-associated loss of cellular robustness. The hypothesis predicts that aged organisms exhibit broader phenotypic variance than young organisms with identical genotypes — a prediction with some empirical support in model organisms but not yet tested rigorously in humans. needs-human-replication no-mechanism (quantitative contribution of capacitor release vs. other aging mechanisms is not established)
Pharmacology — HSP90 inhibitors
Mechanism of pharmacological inhibition
All first- and second-generation HSP90 inhibitors target the ATP-binding pocket of the NTD, competing with ATP binding and locking HSP90 in its open, client-releasing conformation. This prevents chaperone cycle completion, causing client proteins to be routed to proteasomal degradation via CHIP/STUB1-mediated ubiquitination. Note on attribution: Whitesell et al. 1994 4 established that geldanamycin binds HSP90 and disrupts HSP90–client complexes via affinity precipitation; the paper did not characterize the binding site at the NTD ATP pocket. The structural basis of NTD binding was established by Stebbins et al. 1997 (Nat. Struct. Biol. 4:803–808), which is not yet in this wiki. needs-citation
Inhibitor classes
| Compound | Class | Development stage | Notes |
|---|---|---|---|
| Geldanamycin | Benzoquinone ansamycin | Preclinical (toxic) | First HSP90 inhibitor identified; Whitesell et al. 1994 4; hepatotoxicity limits clinical use |
| 17-AAG (tanespimycin) | Geldanamycin analog | Phase 2 (multiple oncology indications; not approved) | ~10-fold less hepatotoxic than geldanamycin; first HSP90 inhibitor in clinical trials; tumor responses observed in hematologic malignancies |
| 17-DMAG (alvespimycin) | Geldanamycin analog | Phase 1/2 | Water-soluble; similar mechanism; discontinued due to modest efficacy |
| Ganetespib (STA-9090) | Resorcinol | Phase 3 (lung cancer; failed primary endpoint) | Non-ansamycin scaffold; favorable PK; Phase 3 GALAXY-2 trial in NSCLC negative |
| AUY922 (luminespib) | Isoxazole-resorcinol | Phase 2 | Potent, well-tolerated in early trials; multiple Phase 2 programs; not approved |
| Onalespib (AT13387) | Resorcinol | Phase 2 | Long-duration HSP90 inhibition; prostate cancer combination trials |
Druggability note: HSP90AA1 druggability-tier has been corrected to tier-2 (Open Targets Platform 2026-05-07, ENSG00000080824: SM tractability = Advanced Clinical TRUE, Approved Drug FALSE). No FDA-approved HSP90 inhibitor exists — ganetespib reached Phase 3 (GALAXY-2, NSCLC) but failed the primary endpoint; tanespimycin reached Phase 2 with responses in hematologic malignancies but was not approved. The seeder’s tier-1 assignment was incorrect. For aging-context druggability, tier-2 also applies: high-quality preclinical probes exist but no aging-indication clinical compound.
Aging-relevant pharmacological hypotheses
The aging-relevant pharmacology hypothesis is the inverse of the oncology strategy: rather than inhibiting HSP90 to kill cancer cells by destabilizing oncoproteins, aging biology suggests that boosting or preserving HSP90/HSP70 chaperone capacity in aged cells could ameliorate proteostasis collapse. Strategies include:
- HSF1 activators — pharmacologically elevating HSF1 activity to drive HSP90AA1 and HSP70 induction (see hsp70 for arimoclomol/HSF1A context)
- Reducing the misfolded protein burden — autophagy inducers (rapamycin via mtor inhibition; see autophagy) to clear aggregates and relieve the HSP90 client queue
- Selective co-chaperone modulation — AHA1/AHSA1 inhibitors that slow the HSP90 cycle, prolonging dwell time on clients and improving yield; explored in polyglutamine diseases preclinically
None of these strategies has aging-specific human evidence. needs-human-replication long-term-unknown
Key interactors and co-chaperones
| Interactor | Type | Functional context |
|---|---|---|
| hsf1 | TF client + upstream regulator | HSP90 sequesters HSF1 monomers at baseline, preventing HSR activation; also a client for HSF1 conformational activation; forms a negative-feedback loop with hsp70 |
| hsp70 | Chaperone partner | HSP70 hands off partially-folded clients to HSP90 via HOP/STIP1; the two chaperones operate in series for kinase and TF maturation |
| HOP (STIP1) | TPR-domain adaptor co-chaperone | Physical bridge between HSP70 EEVD and HSP90 MEEVD motifs; mediates client hand-off |
| p23 (PTGES3) | Co-chaperone | Stabilizes the NTD-dimerized “closed” HSP90 complex; slows ATPase, prolonging client association |
| AHA1 (AHSA1) | Co-chaperone | Stimulates HSP90 ATPase activity; accelerates the chaperone cycle |
| CDC37 | Kinase-specific co-chaperone | Recruits and stabilizes kinase clients on HSP90; required for kinome client loading |
| chip (STUB1) | E3 ubiquitin ligase | Binds HSP90 MEEVD via TPR domain; routes unfolded/non-foldable clients to ubiquitin-proteasome degradation |
| akt | Client kinase | Requires HSP90 for stability and membrane localization; inhibited by HSP90 inhibitors |
| p53 | Client TF | Mutant p53 is an HSP90 client; wt p53 maturation also involves HSP90 |
| hif1a | Client TF | HIF-1α stability and activity depend on HSP90; links HSP90 to hypoxia and metabolic stress responses |
| TSC1-TSC2 | Client complex | The TSC1/TSC2 heterocomplex (see tsc1-tsc2) is an HSP90 client; connects HSP90 to mTOR pathway regulation |
Limitations and open questions
- Paralog-specific aging functions. The relative contributions of HSP90AA1 (inducible) versus HSP90AB1 (constitutive) to aging-related proteostasis changes are not resolved. Most studies use pan-HSP90 antibodies or inhibitors that cannot distinguish the isoforms. needs-replication
- Aging-specific human evidence is absent. All chaperone capacity decline data in the context of HSP90 specifically comes from model organisms (C. elegans, yeast, rodents) or in-vitro cell culture. No proteome-wide profiling of HSP90 client stability in aged human tissue has been published. needs-human-replication
- Capacitor release in human aging. The evolutionary capacitor concept (Rutherford & Lindquist) predicts increased phenotypic variance in aged individuals. This has not been tested at the molecular level in human aging cohorts. no-mechanism
- Druggability for aging is unexplored. All clinical-stage HSP90 inhibitors were developed for oncology. An aging indication — which would require preserving or enhancing HSP90 function rather than inhibiting it — has no clinical-stage compound. The pharmacological strategy (activation vs. inhibition) for an aging intervention is not established. long-term-unknown
- Compartment-specific roles. GRP94 (ER) and TRAP1 (mitochondria) mediate distinct aging-relevant functions (ER stress, UPRmt) that are not covered here and lack wiki pages. stub — seed GRP94 and TRAP1 pages as R27 proteostasis arm candidates.
- Interaction with mTOR/proteostasis axis. mTOR inhibition extends lifespan partly via autophagy induction; the degree to which this relieves HSP90 client queue pressure is mechanistically important but not quantified. no-mechanism
Footnotes
Footnotes
-
doi:10.1101/cshperspect.a034017 · Biebl MM & Buchner J (2019) “Structure, Function, and Regulation of the Hsp90 Machinery” · Cold Spring Harb Perspect Biol 11(8):a034017 · review · comprehensive description of HSP90 domain architecture (NTD/MD/CTD), ATPase-driven chaperone cycle, co-chaperone functions (Sti1/HOP, p23, Aha1, Cdc37), organelle-specific paralogs (GRP94, TRAP1), and pharmacological inhibition · 289 citations · OA: bronze · archive: downloaded · Note: seeder incorrectly listed authors as “Schopf FH, Biebl MM & Buchner J”; the actual authors are Biebl MM & Buchner J only (two authors) ↩ ↩2 ↩3 ↩4
-
doi:10.1016/j.cell.2012.06.047 · Taipale M, Krykbaeva I, Koeva M, Kayatekin C, Westover KD, Karras GI & Lindquist S (2012) “Quantitative Analysis of Hsp90-Client Interactions Reveals Principles of Substrate Recognition” · Cell 150(5):987–1001 · in-vitro (original research — LUMIER/BACON in-cell luminescence interaction assay) · model: 293T stable cell line expressing Renilla luciferase-tagged HSP90β; 420 kinase clones (355 unique, 69% of human kinome), 1,303 TF clones (843 unique, 79% of human TFs), 498 E3 ligase clones (426 unique) tested · 193 kinases (61% of tested, representing >half the human kinome) interacted with HSP90; only ~7% of TFs (58/843) interacted — much lower than previously assumed; 31% of E3 ligases (117/372) interacted; interaction spans “almost 400 client proteins” total (stated in discussion); defined metastability/thermodynamic instability as key substrate recognition principle · 873 citations (100th percentile) · OA: bronze · archive: downloaded ↩ ↩2 ↩3
-
doi:10.1038/s41580-019-0101-y · Hipp MS, Kasturi P & Hartl FU (2019) “The proteostasis network and its decline in ageing” · Nat Rev Mol Cell Biol 20(7):421–435 · review · comprehensive synthesis of the chaperone network (HSP70, HSP90, TRiC/CCT, small HSPs), UPS, and autophagy as integrated proteostasis system; aging-related decline of each component; therapeutic strategies · 1,522 citations (100th percentile) · not OA · archive: not_oa · Unverified against full text — no-fulltext-access; proteostasis network framework claims on this page are drawn from abstract and secondary characterizations of this review ↩ ↩2 ↩3
-
doi:10.1073/pnas.91.18.8324 · Whitesell L, Mimnaugh EG, De Costa B, Myers CE, Neckers LM (1994) “Inhibition of heat shock protein HSP90-pp60v-src heteroprotein complex formation by benzoquinone ansamycins: essential role for stress proteins in oncogenic transformation” · PNAS 91(18):8324–8328 · in-vitro · model: NIH 3T3/v-src and 3T3/tsvsrc transformed fibroblasts; PC-3M prostate carcinoma; Jurkat T cells; CHP-100 Ewing sarcoma; rabbit reticulocyte lysate; affinity precipitation with immobilized geldanamycin derivative · first demonstration that geldanamycin binds HSP90 and disrupts src-HSP90 heteroprotein complexes; does NOT characterize the NTD ATP-binding site (that attribution requires Stebbins 1997) · 1,439 citations (100th percentile) · OA: green (PMC) · archive: downloaded ↩ ↩2 ↩3
-
doi:10.1016/j.cell.2015.03.032 · Walther DM, Kasturi P, Zheng M, Pinkert S, Vecchi G, Ciryam P, Morimoto RI, Dobson CM, Vendruscolo M, Mann M & Hartl FU (2015) “Widespread Proteome Remodeling and Aggregation in Aging C. elegans” · Cell 161(4):919–932 · in-vivo · model: C. elegans N2 (Bristol strain, WT); age-fractionated cohorts days 1–22; SILAC quantitative proteomics · >5,000 proteins profiled; 975 of 1,083 proteins (quantified in ≥3 of 4 experiments) accumulated significantly in the insoluble fraction at day 12 relative to day 1; aggregation initiates mainly after day 6 of adulthood (post-reproductive); abundant proteins make the largest contribution to aggregate load by mass (not chaperone substrates per se); small HSPs strongly enriched in aggregates — interpreted as a protective sequestration response, not solely a consequence of chaperone network saturation · 582 citations (100th percentile) · OA: bronze · archive: downloaded ↩
-
doi:10.1038/24550 · Rutherford SL & Lindquist S (1998) “Hsp90 as a capacitor for morphological evolution” · Nature 396(6709):336–342 · in-vivo + genetic · model: Drosophila melanogaster (pharmacological HSP90 inhibition + Hsp83 partial loss-of-function) · HSP90 reduction reveals cryptic genetic variation producing diverse morphological phenotypes; defines HSP90 as an evolutionary capacitor buffering genetic variation under normal conditions · 2,297 citations (100th percentile) · not OA · archive: not_oa · Unverified against full text — no-fulltext-access; qualitative capacitor claims on this page are drawn from secondary literature characterizations of this paper ↩