HSP90 and HSF1 in Cancer: The Resistance Mechanism and How to Measure It

Choosing an HSP90 antibody is a routine decision until the experiment involves an HSP90 inhibitor — at which point what you measure depends as much on HSF1 as on HSP90 itself. Inhibiting HSP90 releases heat shock factor 1 (HSF1), which drives a compensatory chaperone response that re-stabilises client oncoproteins and blunts the drug, so interpreting the experiment means detecting HSP90, HSF1 and the induced chaperones in the same lysate.

This guide covers the biology behind that feedback loop, why it has defeated every HSP90 inhibitor taken to a phase III endpoint, and which validated reagents report each node of the pathway.

HSP90 Paralogues, Domains, and Oncogenic Clients

HSP90 is among the most abundant cytosolic proteins in eukaryotic cells, accounting for roughly 1–2% of total cellular protein under basal conditions. It functions as an ATP-dependent homodimer of ~90 kDa subunits. Each protomer contains three domains: an N-terminal domain (NTD) carrying the ATP-binding pocket — a Bergerat fold of the GHKL ATPase superfamily; a middle domain that binds clients and co-chaperones; and a C-terminal domain (CTD) that mediates constitutive dimerisation and presents the MEEVD motif for recruitment of TPR-domain co-chaperones. ATP hydrolysis at the NTD drives the conformational chaperone cycle and is indispensable for client maturation. Every clinical HSP90 inhibitor to date occupies this same NTD pocket.

Four paralogues divide the work between cellular compartments, and they are not interchangeable in an experiment — an antibody raised against cytosolic HSP90 will not report on the ER or mitochondrial pools.

The oncological interest in HSP90 rests on the breadth of its client proteome. Established clients include HER2, BRAF V600E, CDK4, AKT, MET, EGFR, ALK, BCR-ABL and VEGFR. Destabilising any single client would have limited effect on a tumour; destabilising all of them at once is what makes the chaperone attractive as a poly-oncogene target. Neckers and Workman set out this rationale, and the accompanying caution, in their review of the field (Clin Cancer Res 2012;18:64–76).

HSP90 does not act alone. Client selection and cycle progression depend on a co-chaperone complement — CDC37 for protein kinases, HOP/STIP1 for hand-off from HSP70, p23/PTGES3 for closing the ATPase cycle, and AHA1/AHSA1 as an ATPase activator. Antibodies against these partners are available in the heat shock protein and chaperone collection, including anti-CDC37 and anti-PTGES3/p23.

Why HSP90 Inhibitors Fail in the Clinic

Drug development against HSP90 has passed through three chemical generations. None has reached regulatory approval as a single agent.

GALAXY-2 is the clearest read on the ceiling of monotherapy-style HSP90 inhibition. In this randomised phase III trial, 677 patients with advanced lung adenocarcinoma received ganetespib plus docetaxel or docetaxel alone. The primary endpoint was overall survival, and the trial was stopped early for futility at a planned interim analysis: median overall survival was 10.9 months with the combination versus 10.5 months with docetaxel alone (HR 1.11; 95% CI 0.899–1.372; P = .329) (Pillai et al., J Clin Oncol 2020;38:613–622). Adding a potent HSP90 inhibitor to standard chemotherapy changed nothing.

The obstacle shared across all three generations is not pharmacokinetic. It is an intrinsic biological feedback loop: HSP90 inhibition activates HSF1, HSF1 induces compensatory chaperones, and those chaperones rescue the very clients the drug was meant to deplete. Jhaveri and colleagues catalogued the clinical landscape and reached the same conclusion (Biochim Biophys Acta 2012;1823:742–755).

HSF1 Activation: Trimerisation, Heat Shock Elements, and Ser326

HSF1 is the master transcriptional regulator of the cellular stress response. Under basal conditions it is held inactive in a cytosolic complex with HSP90, HSP70 and the HSP40/DNAJB1 co-chaperone. When HSP90 is pharmacologically inhibited, misfolded clients accumulate and titrate HSP90 and HSP70 away from HSF1 — the chaperone titration model. Released HSF1 then:

1. forms homotrimers;
2. translocates to the nucleus;
3. binds heat shock elements (HSEs) — arrays of inverted nGAAn pentameric repeats in target-gene promoters;
4. recruits the P-TEFb elongation complex to transcribe HSPA1A (HSP70), DNAJB1 (HSP40) and HSPB1 (HSP27).

The newly synthesised HSP70/HSP40 system re-chaperones the destabilised HSP90 clients, partially restoring their function and allowing the cancer cell to survive continued HSP90 inhibition.

Which phosphosite to blot

HSF1 activity is regulated by extensive phosphorylation, and the site you probe determines what you learn. A systematic alanine scan combined with mass spectrometry identified twelve serine residues phosphorylated in heat-activated human HSF1 — Ser121, Ser230, Ser292, Ser303, Ser307, Ser314, Ser319, Ser326, Ser344, Ser363, Ser419 and Ser444. Of these, only Ser326 contributed significantly to transcriptional activation: phosphorylation there rose rapidly during heat stress, and an S326A substitution mutant stimulated HSP70 expression several-fold less than wild-type factor despite retaining normal DNA binding and nuclear translocation (Guettouche et al., BMC Biochem 2005;6:4). Phosphorylation at Ser303 and Ser307 is instead associated with factor deactivation following stress.

The practical consequence: a phospho-Ser326 blot reports HSF1 activation, whereas Ser303/Ser307 blots report the opposite arm of its regulation. Total HSF1 antibodies frequently show little change in abundance after HSP90 inhibition, because activation is a post-translational event — which is why a total-HSF1 blot alone can make a strongly activated pathway look unchanged.

The HSF1 cancer programme

Beyond the classical heat shock response, HSF1 drives a separate, cancer-specific transcriptional programme. Comparing cells of high and low malignant potential against their non-transformed counterparts, Mendillo and colleagues identified an HSF1-regulated programme active in malignancy and distinct from heat shock, covering cell-cycle regulation, signalling, metabolism, adhesion and translation. This programme was detectable in breast, colon and lung tumours taken directly from patients, and was strongly associated with metastasis and death (Cell 2012;150:549–562). HSF1 is therefore not only a resistance mechanism but an oncogenic driver in its own right — an independent argument for measuring it.

Choosing an HSP90 Antibody for Western Blot, IHC, and Immunofluorescence

Three questions decide which HSP90 antibody fits an experiment: does it need to distinguish the α and β isoforms, which species does the model use, and which applications must the same clone support? Pan-HSP90 clones give the broadest species coverage and are the safest choice for comparative or non-mammalian work; isoform-selective clones are necessary whenever stress-inducible HSP90α must be separated from constitutive HSP90β.

Each StressMarq clone above is also supplied in conjugated formats — ATTO 390, ATTO 488, ATTO 594, biotin, FITC and others — so the same validated binder can move from Western blot to multiplexed immunofluorescence without re-optimising specificity. A pan-isoform HSP90 (total) antibody (SMC-149B) is available where a single total-HSP90 readout is preferred. Browse the full range in the Heat Shock Proteins & Chaperones collection.

Quantifying the Chaperone Response by ELISA

Western blot answers whether a protein changed; sandwich ELISA answers by how much, in absolute units, across many samples at once. For dose–response work, time courses, or serum and plasma measurements, quantitative immunoassay is the more appropriate format.

HSF1 kits are also available for mouse (EM1133-96T) and rat (ER1050-96T) samples, which matters when a cell-line finding moves into a xenograft or syngeneic model. Recombinant human HSF1 protein is available for standard curves and positive controls. The complete range is in the ELISA kits collection.

Experimental Design: Confirming HSF1-Driven Resistance in Your Model

If an HSP90 inhibitor underperforms in your cell line, the question is whether HSF1 is rescuing it. A four-marker panel answers that on a single blot, because the signature is directional: clients fall while chaperones rise.

Clients down and HSP70/HSP27 up is the resistance signature. Clients down with no chaperone induction points elsewhere — to drug efflux, insufficient exposure, or a client-independent survival mechanism.

The functional test is to remove HSF1 and repeat. Chen and colleagues did exactly this in a pooled RNA interference screen, identifying HSF1 as a sensitiser of HSP90 inhibitors: HSF1 knockdown combined with HSP90 inhibition produced a striking combinational effect across multiple cancer cell lines and tumour mouse models. They traced part of the mechanism to the HSF1 target gene DEDD2, and noted that hepatocellular carcinoma — where HSF1 is highly expressed in patient samples — was particularly sensitive to the combination (Oncotarget 2013;4:816–829).

Pharmacological HSF1 inhibition is the alternative to knockdown. KRIBB11 (N²-(1H-indazol-5-yl)-N⁶-methyl-3-nitropyridine-2,6-diamine) blocks HSF1-dependent recruitment of P-TEFb to the hsp70 promoter, abolishing heat-shock-induced reporter activity with an IC₅₀ of 1.2 µmol/L, and blocking induction of HSP27 and HSP70 (Yoon et al., J Biol Chem 2011;286:1737–1747). KRIBB11 is a research tool compound and is not approved for clinical use.

No HSP90 plus HSF1 combination has entered clinical approval, and the preclinical synergy above has not yet been tested in an adequately powered trial with biomarker-selected enrolment. If your work requires a target, species or format not listed here, request a quote and our scientific team will source it.

Figures 1–6 are illustrative summaries of published data and manufacturer catalog specifications. They are not experimental data generated by eBioHippo, and the directions of change shown in Figure 6 are qualitative.

Frequently Asked Questions

Which HSP90 antibody should I use for Western blot?

For most mammalian Western blot work, a pan-HSP90 monoclonal such as SMC-107B (mouse IgG2a; human, mouse, rat, dog, rabbit, hamster, chicken and fish reactivity; validated for WB, IHC, IF, IP and ELISA) is the broadest single choice. Use the isoform-selective SMC-108D when stress-inducible HSP90α must be distinguished from constitutive HSP90β, and SMC-135D when a single clone must cover both cytosolic isoforms across WB, IHC, IF and ELISA. For non-mammalian models including yeast, plant and insect systems, SMC-112B has the widest species coverage.

What is HSP90?

HSP90 (heat shock protein 90) is the most abundant molecular chaperone in eukaryotic cells, making up roughly 1–2% of total cytosolic protein. It works as an ATP-dependent homodimer that binds and stabilises client proteins — particularly signalling kinases, transcription factors and E3 ligases — in near-mature conformations. Its two cytosolic isoforms are HSP90α (HSP90AA1, stress-inducible) and HSP90β (HSP90AB1, constitutive); the ER paralogue is GRP94 (HSP90B1) and the mitochondrial paralogue is TRAP1/HSP75. In cancer cells it becomes critical for maintaining mutant, overexpressed or chimeric oncoproteins.

How do HSP90 inhibitors work?

HSP90 inhibitors competitively occupy the ATP-binding pocket in the N-terminal domain, arresting the conformational chaperone cycle. Without ATP hydrolysis, client proteins cannot be loaded, matured or released, and are instead routed to proteasomal degradation. Geldanamycin and its derivatives 17-AAG and 17-DMAG bind this pocket through a benzoquinone ansamycin scaffold; second-generation agents such as ganetespib and luminespib use resorcinol-based scaffolds but target the same pocket. The net effect is simultaneous depletion of multiple oncoproteins — HER2, BRAF V600E, CDK4, AKT, MET and ALK among them.

What is the connection between HSP90 and HSF1?

HSF1 is normally held inactive in a cytoplasmic complex with HSP90 and HSP70, so inhibiting HSP90 releases it. Accumulating misfolded proteins compete for HSP90 and HSP70, titrating those chaperones away from HSF1; free HSF1 trimerises, enters the nucleus, binds heat shock elements and transcribes compensatory chaperones — chiefly HSP70 (HSPA1A) and HSP27 (HSPB1). Those induced chaperones rescue destabilised HSP90 clients and confer resistance to the inhibitor. HSF1 is additionally overexpressed in many cancers, where it drives a broad cancer-specific transcriptional programme distinct from the heat shock response.

Why haven't HSP90 inhibitors succeeded clinically?

No HSP90 inhibitor has achieved regulatory approval as a standalone cancer therapeutic, and the principal reason is biological rather than pharmacological: HSP90 inhibition is self-limiting because it activates the HSF1 pathway that rescues clients from degradation. The phase III GALAXY-2 trial of ganetespib plus docetaxel in advanced lung adenocarcinoma was stopped early for futility on its overall survival endpoint, and first-generation geldanamycin derivatives carried dose-limiting hepatotoxicity. Patient selection based on HSF1 status or HSP90 client-gene amplification has not been systematically built into trial design, which may have obscured responder populations.

How do I detect HSF1 activation rather than HSF1 abundance?

Blot for phospho-HSF1 at Ser326, because activation is a post-translational event and total HSF1 levels often change little. Ser326 phosphorylation is the phosphosite shown to contribute significantly to HSF1 transcriptional activation, and it increases rapidly during stress. Phospho-specific antibodies against Ser303 and Ser307 report the deactivation arm of HSF1 regulation instead, so they answer a different question. Pair a Ser326 blot with HSP70 and HSP27 read-outs to confirm that the activated factor is producing transcriptional output, and use a total HSF1 antibody or a quantitative HSF1 ELISA for the loading and abundance reference.