Hsp90: A means to an end

Abstract

The Hsp90 molecular chaperone is a key component of the protein homeostasis (proteostasis) system. Hsp90 likely serves as a gatekeeper in a cell’s protein quality control decision tree since this chaperone is linked to nascent polypeptide folding, client maturation, metastable protein maintenance, and polypeptide degradation. Interestingly, how a client protein is directed through the decision process is unclear. Minimally, modifications to the amino-terminal ATP-binding domain of Hsp90 can favor client degradation. As this includes a common class of Hsp90 inhibitors that trigger the breakdown of clinically relevant factors, a better understanding of Hsp90’s role in quality control is merited. Here, we explore how Hsp90 links to both polypeptide biogenesis and triage, the events that regulate the decision route, and how Hsp90’s connections to proteolysis pathways are being exploited for the development of new therapeutics.

Background

We dedicate this work to Dr Len Neckers: An excellent scientist, person, & friend

Heat shock protein 90 (Hsp90) is a highly conserved molecular chaperone that is abundant and essential in eukaryotes, where it functions as a central hub in the protein homeostasis (proteostasis) network.1, 2 Hsp90 is best known for protecting metastable polypeptides, including unactivated steroid aporeceptors and kinases (i.e., clients) where it keeps these signaling factors in an activatable state.³ Often, Hsp90 is thought to safeguard these polypeptides from degradation since Hsp90 inhibition can correlate with declined levels of client proteins.4, 5, 6, 7, 8, 9, 10, 11 However, not all clients, including the canonical steroid aporeceptors and kinases, are degraded following loss of Hsp90. Notably, the temperature sensitive Hsp90 alleles, which were used to establish that this chaperone maintains the activities of signaling factors, do not appreciably alter steady state levels of client proteins following chaperone inactivation.³ Yet, application of small molecule inhibitors against Hsp90 or the use of dominant negative Hsp90 alleles, which alter the ATP-binding domain, lead to a decline in the levels of aporeceptor and kinase proteins.12, 13 Hence, the relationship between Hsp90 and its clients is more complex than a rudimentary stabilization of metastable polypeptides. Here, we will explore how Hsp90 connects to the protein triage process and how this association is being exploited for therapeutic benefits.

Hsp90 was initially observed as one of several proteins whose expression is induced when cells are exposed to elevated temperatures (i.e., heat shock).14, 15 As such, Hsp90 is recognized as a major eukaryotic heat shock protein that is rapidly upregulated by physiological stress to protect the proteome from harm and/or manage damaged proteins in the aftermath by either fostering polypeptide refolding or degradation.1, 2 In eukaryotic cells, Hsp90 is a cytoplasmic and nucleoplasmic protein with subcellular homologs such as Tumor Necrosis Factor Receptor-Associated Protein-1 in the mitochondria and Glucose-Regulated Protein 94 in the endoplasmic reticulum.16, 17, 18, 19 Although present in bacteria, Hsp90 is nonessential under normal physiological conditions yet supports thermotolerance.²⁰ Even with its link to stress conditions, Hsp90 is one of the most abundant eukaryotic proteins under normal physiological conditions constituting ∼2% of a cell’s protein mass.21, 22 While the physiological relevance of Hsp90’s high abundance is still being investigated, minimally it enables Hsp90 to associate with a wide variety of proteins. High-throughput studies have shown that yeast Hsp90 interacts with ∼20% of the proteome.23, 24 Presumably, the broad binding capacity of Hsp90 allows this chaperone to monitor and safeguard a sizeable sector of a proteome.

Interest in Hsp90 grew significantly following the discovery of small molecule inhibitors against the chaperone.4, 25, 26 Intriguingly, both geldanamycin and radicicol are natural products that were previously shown to have antiproliferative activities by unknown mechanisms.¹² A critical breakthrough occurred in 1994 by Dr Len Necker’s research team when they discovered that geldanamycin disrupts the association between Hsp90 and the oncogenic tyrosine kinase Src leading to Src degradation and reversal of the Src-dependent transformation of mammalian cells.⁴ This seminal study inspired numerous clinical trials incorporating 17 different agents targeting Hsp90.¹² The most common Hsp90 inhibitors fit into the amino-terminal ATP-binding pocket and lead to a destabilization of the classic clients, with the degradation being proteasome-dependent.7, 27, 28, 29 Given the clinical relevance of these findings, the ability of Hsp90 to stabilize client polypeptides became a pervasive model in the field (Figure 1).

Fig. 1.

Inhibition of Hsp90 ATP-binding pocket destabilizes client proteins. Hsp70 prevents aberrant folding of a nascent polypeptide and is modulated by cochaperones such as Hsp40. Hop may then transfer partially folded proteins to Hsp90. Binding of ATP (shown as a yellow circle) induces the closed conformation of Hsp90, assisting in client maturation, and native proteins can be released from the open conformation generated by nucleotide disassociation. Binding of an Hsp90 inhibitor like geldanamycin or GA (shown as a purple circle) to the amino-terminal domain suppresses progression through the chaperone cycle. Clients cannot be folded into native conformation(s) and instead are subject to degradation via autophagy or polyubiquitination and proteasomal degradation.

Influence of Hsp90 inhibitors on client protein levels

To better understand the influence of amino-terminal Hsp90 inhibitors on general protein levels within a cell, a number of high-throughput studies were performed.30, 31, 32, 33, 34 Early proteomic efforts using GA identified 176 Hsp90 clientele with 49 being upregulated at least 1.5-fold following GA treatment and 70 downregulated.³⁰ Another quantitative mass spectrometry study found that Hsp90 inhibition reduced the levels of ∼300 proteins and increased 26 proteins by at least 2-fold.³¹ A general assessment of all proteins (i.e., clients and non-clients) found that ∼1600 of the ∼6200 detectable proteins in cancer cell lines were altered following application of GA.³² Multiplexed proteome dynamics profiling assessed the influence of the synthetic GA-derivative 17-AAG on client-protein stability, including nascent and mature polypeptides in different cell types.³³ Overall, 17-AAG altered the stability of ∼1000 targets with most being affected only as nascent polypeptides (i.e., Hsp90 was required only for folding of the target). For the proteins that constitutively relied on Hsp90, these factors fit the metastable mold as they generally had lower thermal stabilities and higher turnover than non-clients. Interestingly, the Hsp90 clientele appear cell specific, as only about half of the clients were shared between the two cell types.³³ Beyond protein degradation, Hsp90 inhibition also increases or decreases the solubility of ∼13% of proteins in a eukaryotic proteome and only 46 of these affected proteins were established interactors.³⁴ Hence, targeting of Hsp90’s ATP-binding domain has diverse effects on a proteome. Client fate can be altered at the nascent or mature polypeptide phase, protein solubility or stability (both decreased and increased half-lives) can shift, and non-client protein levels can fluctuate in reaction to changes to this central molecular chaperone. But how do such diverse outcomes transpire when Hsp90 is modified with just a single small molecule and why are the outcomes different if Hsp90 is genetically (temperature sensitive alleles) impaired compared to chemically (small molecule) inhibited? Perhaps it is relevant to consider that chemical inhibitors can produce off-target effects while genetic mutations may provide more direct insights into Hsp90 function.

Hsp90 conformational cycle dictates protein-protein interactions and client fate

Hsp90 is a flexible molecule consisting of an amino-terminal domain with ATPase activity, a middle domain known to bind metastable clients, and a carboxyl-terminal domain where Hsp90 dimerizes.¹ Minimally, the Hsp90 dimer transitions from an open V-shape to a closed configuration in a cycle driven by ATP binding, hydrolysis, and ADP/ATP release.35, 36 Significantly, Hsp90 preferentially interacts with numerous cochaperones at various points of this cycle, with the cochaperones contributing to ATPase regulation by either accelerating or deaccelerating it.¹ At least for metastable clients like kinases and aporeceptors, ATP binding triggers closure of the amino-terminal domains thereby clamping an extended section of a client into a groove formed by the middle domains of the two Hsp90 protomers.37, 38, 39, 40, 41 Whether non-metastable clients go through a similar clamping is not clear. For many clients, Hsp90 binds intrinsically disordered regions of the target using its surface area across all three domains.²⁴ In addition to ATPase regulation, the cochaperones can facilitate the transfer of clients to Hsp90, modify the dwell time a client is with Hsp90, and manipulate client shape in conjunction with the closing/opening of the Hsp90 dimer.39, 40, 42, 43, 44 For example, Aha1 stimulates Hsp90 ATPase activity and may promote client turnover while p23 and FKBP51 repress ATP cycling and increase dwell time.43, 45, 46, 47, 48, 49 Importantly, these interactions influence and are influenced by the conformational status of Hsp90.

The potential impact of different Hsp90 forms to selectively modulate client activity was evident from the earliest studies identifying phenotypic Hsp90 mutations.3, 50 For instance, an allele (E431K) was isolated that did not support hormone binding by the glucocorticoid receptor but was near wild type for three highly related aporeceptors (mineralocorticoid, progesterone, and estrogen receptors).⁵⁰ While the mechanism wasn’t initially apparent, future work by numerous laboratories would demonstrate that amino acid substitutions in Hsp90 can favor specific conformations (e.g., open vs closed dimer) and have select effects on client activity, client stability, and cochaperone interactions.51, 52, 53, 54, 55, 56, 57, 58 Recent work focusing on dominant negative alleles of Hsp90 show that ATP hydrolysis can be separated from closure of the amino-terminal domains and trigger ubiquitin-mediated degradation of a bound aporeceptor client.¹³ Hence, the conformation of Hsp90 is an important determinant in dictating client fate (i.e., folding or degrading) independent of ATP hydrolysis. Overall, it is evident that even subtle shifts in the conformational status of Hsp90 can have profound outcomes in cochaperone interactions and client function. But how are variations in Hsp90’s physical form driving such drastic differences in client fate including folding or degrading?

Cochaperones and E3 ubiquitin ligases: A choice is made

A typical model illustrating Hsp90’s biological significance centers on polypeptide folding where a nascent chain is initially bound by Hsp70 that has ATPase activity regulated by Hsp40, delivered to Hsp90 with the assistance of the Hsp70-Hsp90 organizing protein (Hop) cochaperone, and matured by proceeding through the ATPase cycle with the assistance of additional cochaperones.1, 59 Seminal cryo-EM based work provided detailed insights into this model, including for the client loading complex of Hsp70-Hop-Hsp90 and the client remodeling complex of Hsp90-p23.39, 40 In brief, the ligand-binding domain of GR (client) is incompletely folded when associated with Hsp70-Hop-Hsp90, yet has a folded, ligand-binding conformation when associated with Hsp90-p23. It has been suggested that Hop serves as a lynchpin between an unfolded and folded client. Indeed, p23 enters the system to relieve Hop thereby allowing the folding process to progress for GR.⁶⁰

Intriguingly, Hop interactions with Hsp90 also compete with the E3 ubiquitin ligase carboxyl terminus of Hsc70-interacting protein (CHIP).⁶¹ E3 ligases facilitate the degradation of polypeptides by mediating the transfer of ubiquitin from an E2 ubiquitin-conjugating enzyme to a substrate protein that is then directed to the proteasome for degradation.⁶² CHIP has an amino-terminal tetratricopeptide repeat domain that binds directly to the “EEVD” tails of either Hsp70 or Hsp90, a middle coiled-coil domain for self-dimerization, and a carboxyl U-box domain that interacts with E2 ubiquitin-conjugating enzymes.63, 64, 65, 66 CHIP also associates with the S5a subunit of the proteasome thereby facilitating delivery of substrates to degradation machinery.⁶⁴ Hence, CHIP serves as a bridge between the chaperones and ubiquitin-proteasome system (UPS). Of note, once a polypeptide has been selected for the UPS pathway, Hsp90 and Hsp70 facilitate the progression by enhancing CHIP’s ability to ubiquitinate a substrate.⁶⁷ But what dictates the choice to either fold or degrade an Hsp90-bound client?

One determinant is the post-translational modification state of Hsp70 and Hsp90, as phosphorylation of the carboxyl-termini of both Hsp70 and Hsp90 favors interactions with Hop and disfavors CHIP association.⁶⁸ Significantly, highly proliferative cells predominantly have phosphorylated Hsp70 and Hsp90, which are complexed with Hop, and phospho-mimetic chaperone mutants increase cell proliferation rates.⁶⁸ What is not known is what triggers the modifications to Hsp70 and Hsp90. Moreover, how might this choice be dynamically controlled, as even individual polypeptides under normal physiological conditions will require timely selection of folding or degrading pathways? Further complicating matters is the prediction that Hsp90 interacts with approximately one third of all E3 ubiquitin ligases, which would be over 200 in human cells and over 30 in yeast.⁶⁹ Unfortunately, only a few of these have been studied in the context of Hsp90. Yet, the E3 diversity might account for the differential effects of different Hsp90 alleles and small molecule inhibitors on protein stability. Hence, it is important to delineate how other E3 ubiquitin ligases modulate Hsp90 to influence client turnover.

Besides CHIP, the association of Hsp90 with the E3 ubiquitin ligases Cul5, HECTD3, and the SCF complex has been explored. The RING E3 ubiquitin ligase Cul5 serves a redundant role to CHIP for at least some clients since both CHIP and Cul5 can promote the degradation of the Hsp90/Hsp70-associated transmembrane tyrosine kinase receptor Erb-b2 receptor tyrosine kinase 2 (ErbB2).70, 71, 72 Cul5 can also influence Hsp90-cochaperone complexes as it has been shown that the E3 ubiquitin ligase contributes to Cdc37 disassociation upon Hsp90 inhibition.⁷² Yet, this is not a universal outcome for E3 ligases since HECTD3 joins the Hsp90-Cdc37 complex in the presence of an inhibitor to degrade the C-RAF kinase.⁷³ Thus, Hsp90 small molecules can both favor and disfavor E3 ubiquitin ligase interaction with cochaperones. Cochaperones can also influence Hsp90 association with E3 ubiquitin ligases. For example, the Sgt1 cochaperone is needed to promote the assembly of Hsp90 with the Skp1p-Cdc53p-F-box protein (SCF) E3 ubiquitin ligase complex.⁷⁴ The intertwined relationships between cochaperones (Table 1) and E3 ubiquitin ligases (Table 2) further the complexity of Hsp90-dependent protein degradation.

Table 1.

Cochaperones influence Hsp90 client stability.

Cochaperone (human/yeast)	Hsp90 binding site	Hsp90 ATPase activity	Effect on client association	Impact on client stability
Aha1	Middle domain and amino-terminal domain; induces closed conformation75, 76, 77	Increases46, 78	Decreases dwell time/increases turnover48, 78	Knockdown rescues mutant CFTR; inhibition decreases tau79, 80
p23/Sba1	Amino-terminal domain; stabilizes ATP-bound closed conformation40, 81	Decreases45	Increases dwell time43, 47	Knockdown decreases tau protein82
FKBP51	Carboxyl-terminal domain; may stabilize nucleotide-bound closed or open conformation49, 83, 84	Decreases49	Hsp90 serves as scaffold for FKBP51 interaction with tau49	Knockdown decreases tau and mutant huntingtin levels; knockout reduces tau levels85, 86, 87
Hop/Sti1	Middle domain and carboxyl-terminal domain39, 88, 89	Decreases90	Facilitates client transfer from Hsp70; stalls clients60, 91	Knockout only changes stability of a small number of clients92
p50/Cdc37	Amino-terminal domain; prevents conformational change93	Decreases94	Assists in client loading93	Disruption leads to degradation95, 96, 97, 98
Sgt1	Middle domain and amino-terminal domain away from ATP-binding site78, 99	No effect100	Increases dwell time78	Disruption decreases client levels; mutants show increased synphilin-1 aggregation 101, 102, 103, 104

This table summarizes how cochaperones affect client turnover.

Table 2.

E3 ubiquitin ligases modulate Hsp90-dependent client breakdown.

E3 ubiquitin ligase	Hsp90 binding site	Cochaperone association	Impact on Hsp90-dependent protein stability
CHIP	Carboxyl-terminal domain64	Hsp90 cannot bind simultaneously with Hop61	Knockdown decreases Hsp90-dependent pTau degradation82
Cul5	Unknown	Assists in disassociation of Hsp90-Cdc37 after Hsp90 inhibition72	Mutant decreases Hsp90-dependent ErbB2 degradation71
HECTD3	Unknown	Does not disassociate Hsp90-Cdc37 after Hsp90 inhibition73	Knockdown decreases Hsp90-dependent CRAF degradation73
SCF complex	Not known to directly bind Hsp90; complex stabilized by Sgt199, 100	Sgt1 needed to bind to Hsp9099	Knockdowns show that SCF destabilizes while Sgt1 stabilizes Mis12 subunits101

This table simplifies how E3 ubiquitin ligases regulate client degradation.

Hsp90 moderates proteasome assembly

In addition to aiding in the delivery of clients to the UPS, Hsp90 has been shown to influence proteasome assembly.¹⁰⁵ Proteasomes are comprised of a 20S catalytic core, which mediates polypeptide cleavage, that are capped with either one (26S) or two (30S) 19S regulatory particles that unfold and funnel polyubiquitinated substrates to the 20S core.¹⁰⁶ Disrupting Hsp90 function with a temperature sensitive allele leads to a loss of capped complexes and an accumulation of 20S particles.¹⁰⁵ The Hop cochaperone also impacts proteasome assembly. Deleting Hop decreases the ratio of capped (especially 30S) to uncapped proteasomes and reduces degradation activity in vivo.⁹² Hence, Hsp90 and its cochaperones (CHIP and Hop) can alter both client delivery to the UPS and the assembly of the UPS components.

Hsp90 intersects with autophagy

The triage of faulty proteins is managed by both UPS and the autophagy-lysosomal pathway (ALP). ALP is a critical quality control system that manages the break down and recycling of damaged organelles, protein aggregates, and pathogens.¹⁰⁷ Significantly, Hsp90 modulates several branches of ALP, including mitophagy, macroautophagy, and chaperone-mediated autophagy (CMA).108, 109, 110

Mitophagy refers to the autophagic degradation of damaged mitochondria and is initiated by activation of mammalian Uncoordinated 51-like kinase 1 (Ulk1) or yeast Autophagy-related protein 1 (Atg1) kinase. Hsp90-Cdc37 maintains Ulk1/Atg1 in activatable states.¹¹⁰ Once activated, Ulk1 phosphorylates Hsp90 to maintain its Hsp90-disassociation and initiate mitophagy.¹¹¹ Thus, Hsp90 is essential for regulating mitochondrial turnover. Hsp90 also modulates other autophagy factors, including Atg6 (Beclin 1 in mammals) and Atg7. GA-inhibition of Hsp90 destabilizes Atg6 by promoting Atg6 ubiquitination and proteasomal degradation, which dampens the autophagic response.¹¹² Comparably, Atg7 levels are reduced following Hsp90 small molecule blockade that also represses autophagy.¹¹³

Hsp90 inhibition can promote the autophagic breakdown of select proteins. For example, GA application triggers the proteolysis of the IκB kinase (IKK) and the degradation is blocked by knocking out Atg5 but is unaffected by proteasome inhibition.¹⁰⁸ In addition to IKK, the NF-κB-inducing kinase (NIK) is degraded through ALP following GA treatment.¹¹⁴ Hsp90 also regulates a broader protein triage route by affecting CMA. CMA degrades proteins with a KFERQ motif where Hsp70 binds and delivers these factors to the lysosome-associated membrane protein type 2A (LAMP2A) complex. Substrate binding to LAMP2A oligomerizes the receptor, which is then translocated into the lysosome for substrate degradation. Hsp90 stabilizes LAMP2A as it undergoes multimerization supporting this transitive state.¹⁰⁹ It was also suggested that reducing the Hsp90-LAMP2A association can dampen CMA activity.¹¹⁵ Thus, Hsp90 plays a role in polypeptide degradation mediated by either UPS or ALP.

Degradation and disease: Hsp90 as a source and a solution

The correlation between Hsp90 and disease progression involves, in part, the quality control of pathogenic proteins. Significantly, Hsp90 has both positive and negative roles in diseases with several examples corresponding to the degradation of key protein players. For two neurodegenerative diseases, Parkinson’s and amyotrophic lateral sclerosis (ALS), Hsp90 can have both protective and detrimental influences. Parkinson’s Disease is linked to toxic α-synuclein oligomers/aggregates and Hsp90 can prevent the formation of toxic α-synuclein species.¹¹⁶ Furthermore, Hsp90 facilitates the degradation of α-synuclein by CMA possibly through LAMP2A association.¹¹⁷ Alternatively, Hsp90 and Cdc37 can protect disease-associated Leucine-rich repeat kinase 2 (LKKR2) mutants from proteasomal breakdown.¹¹⁸ For example, the G2019S LKKR2 is a common mutation in Parkinson’s Disease that increases kinase activity and shortens axons.119, 120, 121 Inhibiting Hsp90 with GA decreases wildtype and mutant LKKR2 levels and restores axonal growth in G2019S LRRK2 cultured mouse neurons.¹¹⁸

For ALS, Hsp90 promotes the destruction of the trans-active response DNA-binding protein 43 (TDP-43) via the proteasome. Hsp90 binds carboxyl-terminal TDP-43 fragments, which are associated with ALS progression, and mediates the proteasomal degradation of these peptides.¹²² Notably, knocking down Hsp90 with siRNA raises TDP-43 levels, especially phosphorylated TDP-43 indicating that Hsp90 selectively targets disease-related species.¹²² In contrast, Hsp90 and Cdc37 may protect TDP-43 from autophagic breakdown. Disrupting Hsp90-Cdc37 association with a small molecule promotes degradation of TDP-43 through autophagy.⁹⁸ Hence, Hsp90 impacts the triage of TDP-43 through both UPS and ALP. Hsp90 function was shown to promote the accumulation of TDP-43 aggregates in a Caenorhabditis elegans model of ALS as heat inactivating a temperature sensitive Hsp90 allele decreased levels of unphosphorylated and phosphorylated TDP-43.¹²³

Similarly, Hsp90 appears to protect hyperphosphorylated Tau (pTau) from degradation, thereby allowing it to form aggregates that are associated with Alzheimer’s disease. Inhibiting Hsp90 with a small molecule increases pTau breakdown and cochaperones such as Hop, CHIP, FKBP51, and Aha1 can differentially affect Hsp90-dependent tau stability.80, 82, 85, 86 For example, knockdown of CHIP with siRNA mitigates Hsp90 inhibitor-induced pTau breakdown, and tau’s phosphorylation status influences CHIP binding and ubiquitination activity. In contrast, p23 knockdown decreases total tau protein.⁸² Collectively, these studies emphasize that Hsp90 function and association can generate distinct outcomes for the progression of neurodegenerative diseases.

Outside of neurodegenerative diseases, Hsp90 can alter the stability of cancer associated proteins like hypoxia-inducible factor-1α (HIF-1α) and mutant p53. Hsp90 protects HIF-1α and mutant p53, but not wild type p53, from degradation suggesting that it enhances tumor growth.5, 11, 124 Hsp90, Hsp70, and associated cochaperones have also been shown to form a network of high molecular weight complexes called the epichaperome that promotes tumor survival in over half of cancers.¹²⁵ Interestingly, extracellular Hsp90 is linked to tumor metastasis as well because in part it activates matrix metalloproteinase 2 (MMP2) to degrade the extracellular matrix. Inhibiting Hsp90 with extracellular GA-beads reduced active MMP2 and invasion, indicating its importance in cancer progression.¹²⁶ Overall, Hsp90 inhibition remains a tantalizing model for the treatment of proteopathies and cancers, and the development of therapies that effectively target Hsp90 remains a worthy pursuit having potential substantial implications for human health.

Selectivity is key: Developing therapies controlling Hsp90 interactions

One of the earliest proponents of exploiting Hsp90 inhibitors as a cancer therapeutic was Dr Len Neckers.4, 5 This innovative goal has been supported by several studies indicating the promise of Hsp90 small molecules.4, 5, 10, 11 However, given the central role of Hsp90 in proteostasis, inhibiting it might have dire consequences including: 1) degradation of crucial proteins; 2) activation of a heat stress response that upregulates chaperone levels, including Hsp90; or 3) stimulate the progression of other cancerous pathways.8, 127, 128, 129, 130, 131 For instance, inhibiting Hsp90 in mice reduces the proliferation of breast cancer cells but increases osteolytic metastases due to Src kinase activation and subsequent Akt phosphorylation.129, 130 To limit potential negative consequences, therapeutic efforts have turned towards manipulating specific Hsp90 interactions instead of base chaperone function.

Targeting select Hsp90 cochaperones has been one potential solution. For example, disrupting Hsp90 cochaperones impacts the ∆F508 variant of the Cystic Fibrosis Transmembrane Conductance Receptor (CFTR), which is a mutant that fails to fold into a stable conformation and correlates with Cystic Fibrosis (CF) disease.79, 132, 133, 134 Hsp90 binds nascent CFTR, and its inhibition with GA decreases both wildtype and mutant protein levels.¹²⁷ In contrast, expression of the Aha1 cochaperone is inversely correlated with CFTR stability and disrupting Hsp90-Aha1 interaction with a small molecule restores CFTR activity even in the ∆F508 background.79, 134 Furthermore, the Hsp90-Aha1 modulator-1 (HAM-1) compound dampens ∆F508 CFTR breakdown without causing cochaperone disassociation, instead suppressing the stimulatory effect of Aha1 on Hsp90 ATPase activity.¹³⁵

Besides CFTR, Aha1 also promotes Hsp90-dependent tau aggregation. Inhibiting Hsp90-Aha1 interaction with a small molecule decreases insoluble tau and limits tau fibrillization in vitro.⁸⁰ Overexpression of Aha1 is associated with enhanced accumulation of toxic T22-tau oligomers, reduced number of neurons in the hippocampus, and impaired memory in mouse models.⁸⁰ Hence, disrupting the Hsp90-Aha1 complex might mitigate the progression of Alzheimer’s disease.⁸⁰ Similarly, dampening Aha1 stimulation of Hsp90’s ATPase activity with the SEW84 compound reduces pTau in a human cell line.¹³⁶ In sum, selectively modulating Hsp90 and its cochaperones is a promising therapeutic route. However, to our knowledge, the treatments described above did not enter clinical trials, highlighting the drastic need for further drug development.

Hsp90 therapeutics: New clients and new hope

Recent alternative approaches aim to generate novel Hsp90 interactions to selectively degrade target proteins. By exploiting the ability of Hsp90 to both associate with numerous E3 ubiquitin ligases and facilitate the degradation of a bound protein, researchers have been generating synthetic biomolecules to steer Hsp90 to a Protein of Interest (POI) and foster its degradation. For instance, Proteolysis-Targeting Chimeras (PROTACs) consist of an Hsp90-binding ligand, a linker, and a POI-ligand to mark the POI for ubiquitination and destruction via the proteasome (Figure 2). Significantly, PROTACs have shown promise as cancer therapeutics. In one case, Hsp90-mediated targeting chimeras designed against Cyclin-Dependent Kinase 4 and 6 (CDK4/6), which are overactive in a variety of cancer cells, promote the proteasome-dependent breakdown of CDK4/6 in mouse melanoma cells and reduced tumor growth in mouse models.¹³⁷

Fig. 2.

PROTACs use Hsp90 to specifically target proteins for breakdown. Several proteolysis-targeting chimeras (PROTACs) utilize Hsp90 association with E3 ubiquitin ligases to target a Protein of Interest (POI) for degradation through polyubiquitination and delivery to the proteasome. These constructs consist of an Hsp90-Binding Ligand connected to a protein of interest (POI)-Binding Ligand via a linker. For example, Hsp90-mediated targeting chimeras (HEMTACs) and Hsp90-mediated proteolysis-targeting chimeras (HSPROTACs) are PROTACs that use modified Hsp90 inhibitors to target CDK4/6 and PARP1 for breakdown.137, 138

Similarly, an Hsp90-mediated proteolysis-targeting chimera (HSPROTAC) was generated to drive poly(ADP-ribose) polymerase-1 (PARP1) destruction. The HSPROTAC compound triggered the Hsp90-dependent degradation of PARP1 via the proteasome in MCF-7 breast cancer cells and limited tumor growth in mouse models.¹³⁸ Compounds that utilize Hsp90 have also been shown to counter tumor development by initiating cell death. Hsp90 interactome-mediated proteolysis-targeting chimeras (HIM-PROTACs) can link glutathione peroxidase 4 (GPX4) to Hsp90 for proteasomal destruction inducing ferroptosis and reducing tumorigenesis in mice.¹³⁹ In addition, Hsp90 can influence PROTAC efficacy even if it is not bound by the construct. The BRD4–MZ1–CRL2^VHL PROTAC targets BRD4 for proteasomal degradation via the Cullin-RING ubiquitin ligase CRL2^VHL, and inhibiting Hsp90 with 17-AAG increases BRD4 breakdown but only in the presence of the PROTAC in human colon cancer cells.140, 141 This work shows that Hsp90 inhibition can seemingly generate specific effects that promote cell sensitivity to other protein-targeting therapeutics.¹⁴¹ Therefore, future PROTAC development might utilize Hsp90 as a tool for protein targeting or use its inhibition as a synergistic enhancement of selective degradation. In addition, other combination therapies can utilize Hsp90 inhibitors in concert with kinase inhibitors to counter resistance in cancer cells.¹⁴² Together, these studies suggest that Hsp90 inhibition may be manipulated to enhance the design of cancer therapeutics.

Conclusion

Hsp90 has long been considered a key molecular chaperone serving as a central node in the proteostasis network. Often, our focus is on the influence of Hsp90 in protein folding events, yet the large influx of Hsp90 amino-terminal inhibitor studies highlighted the key role that this chaperone has in determining the steady-state levels of therapeutically important clients like the Src kinase. It might be useful to consider Hsp90 a cell’s “protein quality control coordinator” since it cooperates with Hsp70 and cochaperones to fold nascent polypeptides, has prolonged interactions with metastable clients (e.g., aporeceptors and kinases) to prevent aggregation, and delivers faulty polypeptides to degradation pathways including the UPS and ALP. Basically, Hsp90 operates at the decision nexus where a protein is funneled towards either maturation or degradation (Figure 3). While there is still much work to be done to fully understand all the roles of Hsp90 in the proteostasis process, what is clear is that Dr Len Neckers and his numerous discoveries on Hsp90, including amino-terminal inhibitors, should receive robust praise and our enduring gratitude.

Fig. 3.

Hsp90 functions as a decision nexus that determines protein fate. A nascent polypeptide translated by the ribosome is bound by a Hsp70-Hsp40 complex that prevents aberrant folding, and Hop can then transfer the non-native protein to Hsp90. Hsp90 serves as a decision nexus that either drives clients towards refolding into native conformations or degradation via the autophagy-lysosomal pathway or the ubiquitin-proteasome system. This decision can be impacted by Hsp90 small molecules and cochaperones. Hsp90 also directly modulates degradation machinery through its roles stabilizing LAMP2A multimerization in chaperone-mediated autophagy and assembling the proteasome generating more widespread effects on protein turnover. Thus, its functions can influence protein abundance, solubility, and stability to limit aggregation.

Declarations of interest

The authors declare the following financial interests/personal relationships, which may be considered as potential competing interests: Brian C. Freeman reports financial support was provided by the National Institutes of Health. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

No data was used for the research described in the article.