Synthesis and Evaluation of Small Molecule Disruptors of the Aha1/Hsp90 Complex for the Reduction of Tau Aggregation

Abstract

KU-177 was recently shown to disrupt interactions between Hsp90 and Aha1 in vitro. Subsequent studies in recombinant thioflavin T (ThT) assays demonstrated that KU-177 ablates Aha1-driven enhancement of Hsp90-dependent tau aggregation, which was confirmed by TEM. Using KU-177 as a lead compound, derivatives of KU-177 were synthesized and evaluated for their ability to disrupt Aha1/Hsp90 interactions and inhibit P301L tau aggregation. Preliminary structure-activity relationships were revealed, which led to the identification of a new lead compound that contains a cis-like amide bond. The new lead compounds retain the ability to disrupt Aha1/Hsp90 interactions in SH-SY5Y and SK-BR-3 cells without direct inhibition of Hsp90, providing a new scaffold for subsequent drug discovery efforts.

The aggregation of misfolded proteins is a hallmark of several neurodegenerative diseases, an example of which is the accumulation of microtubule-associated protein tau (MAPT, tau) aggregates, which appear to drive the pathology of Alzheimer’s disease (AD), Parkinson’s disease (PD), and other diseases collectively called “tauopathies.” Despite the neurotoxic effects of tau accumulation in these pathologies, no therapy has emerged to reduce or prevent tau aggregation. To address this unmet need, attention has turned to molecular chaperones, which regulate the folding, stability, and maturation of tau and other proteins within the cell. Several of these molecular chaperones have emerged as potential targets for the development of chemotherapeutics to combat aberrant accumulation of such neurotoxic aggregates. In particular, the ATP-dependent 90 kDa heat shock protein, Hsp90, has emerged as a target for therapeutics to treat neurodegenerative tauopathies, as it has been shown to modulate the accumulation of misfolded tau species.¹⁻⁴ Notably, Hsp90 has also been shown to colocalize with amyloid plaques, and its expression is increased in both cytosolic and membranous fractions of AD brains.⁵

Dickey et al. were the first to demonstrate that small molecule inhibitors of Hsp90 decrease tau levels in vitro.⁶ Geldanamycin, an N-terminal Hsp90 inhibitor, was shown to manifest neuroprotection in cells that express mutant forms of tau.⁷ While molecules that inhibit Hsp90’s ATPase activity have shown preclinical efficacy, many of these Hsp90 inhibitors failed in clinical trials. These failures were the result of poor blood–brain barrier permeability and on-target effects associated with the inhibition of Hsp90, specifically, disruption of Hsp90’s ability to fold other Hsp90-dependent substrates.⁸⁻¹⁰

The activator of Hsp90 ATPase homologue 1 (Aha1) is the only Hsp90 cochaperone known to stimulate Hsp90’s ATPase activity, which it achieves through direct interactions that are regulated by c-Abl-mediated tyrosine phosphorylation.¹¹ Previously, it was demonstrated that disruption of Aha1/Hsp90 interactions with small molecules leads to the inhibition of tau fibril seeding and the formation of insoluble tau species both in vitro and in cultured cells.¹² Importantly, these molecules manifest their neuroprotective activity without affecting the Hsp90-mediated refolding process. Another small molecule disruptor of Aha1/Hsp90 interactions, HAM-1, has since been shown to selectively inhibit Aha1-stimulated Hsp90 ATPase activity without affecting Hsp90’s ATPase activity in the absence of Aha1.¹³ Such data provide evidence that disruption of Aha1/Hsp90 interactions can prevent tau aggregation without affecting Hsp90’s ability to fold other client protein substrates. Therefore, disruption of the Aha1/Hsp90 complex represents a promising therapeutic strategy for the development of small molecules to combat AD, PD, and other tauopathies.

Previously, several novobiocin-based C-terminal Hsp90 inhibitors that disrupt the Hsp90/Aha1 complex were identified¹⁴ (Figure 1). Preliminary studies suggested that the noviose sugar present on these inhibitors was responsible for binding interactions with Hsp90, while the amide side chain was responsible for interactions with Aha1.¹² In agreement with this hypothesis, replacement of the benzamide side chain in KU-174 with the acetamide side chain in KU-32 afforded Hsp90-mediated neuroprotective activity; however, such compounds were less effective at disruption of Hsp90/Aha1 interactions.¹⁴ Through subsequent structural investigations, it was shown that novobiocin analogs containing an amide side chain with five or more carbons exhibit greater efficacy at disruption of the Aha1/Hsp90 complex.¹⁵

Figure 1.

Structures of select C-terminal Hsp90 inhibitors and Aha1/Hsp90 disruptors.

Upon removal of the noviose sugar present in KU-174, a small molecule (KU-177) was identified that disrupts Aha1/Hsp90 interactions and ablates Aha1-driven enhancement of Hsp90-dependent tau aggregation. Notably, KU-177 manifests this activity without inhibiting Hsp90’s ability to refold luciferase.¹² These studies confirmed that disruption of the Aha1/Hsp90 complex is achievable without affecting Hsp90’s ability to properly fold other client proteins. The combination of such attributes is desirable for the treatment of tauopathic disease states, while avoiding the deleterious side effects associated with direct inhibition of Hsp90 ATPase activity. Although KU-177 significantly disrupts Aha1/Hsp90 interactions, its psychochemical and pharmacokinetic properties are not ideal for use in vivo. Therefore, KU-177 served as a lead compound, and additional analogs have now been designed, synthesized, and evaluated as disruptors of Aha1/Hsp90 interactions and inhibitors of Hsp90-mediated tau aggregation. Specifically, preliminary structure-activity relationships (SAR) for the coumarin core, biaryl side chain, and the amide linker are reported herein. Results from these studies are summarized below.

On the basis of computational models and preliminary SAR, the biaryl side chain of KU-177 was identified as a pharmacophore for disruption of the Aha1/Hsp90 complex.¹² Specifically, initial efforts demonstrated that disruption of the Aha1/Hsp90 complex can be achieved upon replacement of the coumarin ring with aliphatic and aromatic substituents, which was demonstrated by coimmunoprecipitation studies with Aha1 from SK-BR-3 cells treated with KU-177 derivatives for 24 h (data not shown). As a result, derivatives that contain a phenyl surrogate for the coumarin core of KU-177 were first investigated. As outlined in Scheme 1, the biaryl acid (4) was prepared in two steps via a Suzuki coupling reaction between commercially available methyl-3-iodo-4-methoxybenzoate (1) and (3-methoxyphenyl)boronic acid (2). The biaryl methyl ester (3) was then converted to the carboxylic acid (4) upon hydrolysis with lithium hydroxide in a mixture of tetrahydrofuran, methanol, and water. The biaryl acid chloride (5) was coupled with various anilines (6a–6q) that served as surrogates for the coumarin ring, which led to the corresponding amides (7a–7q).

Scheme 1. Synthesis of KU-177 Biaryl Side Chain Derivatives; Reagents and Conditions: (a) K₂CO₃, Pd(dppf)Cl₂, 1,4-dioxane, 130 °C, 16 h, 88–96%; (b) LiOH, THF/MeOH/H₂O (3:1:1), rt, 12 h, 96–100%; (c) (COCl)₂, DMF (cat.), DCM, rt, 5 h; (d) RNH₂, DIPEA, DCM, rt, 16 h.

A previously employed AlphaLISA (PerkinElmer Inc.) assay, whereby recombinant GST-tagged Aha1 and His-tagged Hsp90α were incubated in the presence of each compound,¹⁶ was used to determine whether these compounds disrupt Aha1/Hsp90 interactions. KU-177 was found to manifest an IC₅₀ value of 4.08 ± 0.6 μM in this assay. Compounds were screened at 10 μM, and IC₅₀ values were determined for each hit. Derivatives that retain the biaryl side chain of KU-177 manifested the ability to disrupt Aha1/Hsp90 interactions (Table 1). Results from these experiments revealed an SAR trend whereby electron withdrawing groups at the para position relative to the amide were most effective. Notably, these substituents were accurately predicted by the Topliss tree model for phenyl substitution.¹⁷ In fact, substitution of the KU-177 coumarin core with p-NO₂ (7i, IC₅₀ = 4.04 μM) or p-CF₃ (7h, IC₅₀ = 4.83 μM) phenyl substituents retained the activity manifested by KU-177, while simultaneously reducing atomic weight. A set of derivatives with nitrogen-containing, electron-withdrawing heterocycles at the para position were synthesized to further investigate this region of the molecule. Compounds containing pyridines, imidazole, or an N-linked 1,2,4-triazole ring attached to the para position manifested an impaired ability to disrupt Aha1/Hsp90 interactions as compared to the p-NO₂ or p-CF₃ phenyl derivatives (data not shown), indicating that the stronger electron withdrawing effects of p-NO₂ and p-CF₃ substituents at the para position are preferred.

Table 1. SAR of Phenyl Derivatives Containing the KU-177 Biaryl Side Chain.

entry	R1	R2	R3	R4	R5	IC50
7a	H	H	H	H	H	>10 μM
7b	H	H	OH	H	H	>10 μM
7c	H	H	OMe	H	H	>10 μM
7d	H	H	Cl	H	H	6.03 ± 1.9 μM
7e	H	H	F	H	H	6.32 ± 3.1 μM
7f	H	H	COOMe	H	H	>10 μM
7g	H	H	CN	H	H	8.57 ± 0.3 μM
7h	H	H	CF3	H	H	4.83 ± 0.9 μM
7i	H	H	NO2	H	H	4.04 ± 0.5 μM
7j	OMe	H	OMe	H	H	9.53 ± 0.3 μM
7k	Cl	H	Cl	H	H	>10 μM
7l	H	Cl	Cl	H	H	>10 μM
7m	Cl	H	H	H	H	>10 μM
7n	H	OMe	H	OMe	H	>10 μM
7o	H	Cl	H	H	H	>10 μM
7p	H	COOMe	H	H	H	>10 μM
7q	H	NO2	H	H	H	>10 μM

SAR investigation of the KU-177 biaryl side chain were then pursued with the p-CF₃ aniline (6f) in lieu of the coumarin ring to generate a series of compounds that contain substitutions on each aromatic ring of the biaryl side chain. A combination of preliminary data and information from prior studies on KU-174 derivatives¹⁸ suggested that the hydrogen bond accepting methoxy group at the para position of the first biaryl ring was optimal for disruption of the Aha1/Hsp90 complex. Therefore, efforts turned toward optimization of the substitution pattern on the second biaryl ring. Consequently, two series of compounds were synthesized from commercially available boronic acids (8 and 13) via Scheme 2, which was used to establish preliminary SAR for the biaryl appendage.

Scheme 2. Synthesis of biaryl side chain derivatives with second phenyl ring ortho or meta to the amide. Reagents and conditions: (a) (COCl)₂, DMF (cat.), DCM, rt, 5 h, 90–99%; (b) DIPEA, DCM, rt, 12–48 h, 62–92%; (c) K₂CO₃, Pd (dppf)Cl₂, DMF, 130 °C, 12–24 h, 64–99%.

None of the derivatives that contained the second aryl ring attached at the ortho position disrupted Aha1/Hsp90 interactions with an IC₅₀ value below 10 μM (Table 2), suggesting that attachment of the second aryl ring at the meta position is preferred. Indeed, several compounds that contain the meta-substituted aryl ring exhibited micromolar IC₅₀ values (Table 3). Although the 4-phenol derivative (16r) did show micromolar potency, substituents at the 3 position manifested better activities among this class of compounds. Specifically, the m-methoxyphenyl (7h) and m-dimethylaniline (16g) derivatives exhibited the lowest IC₅₀ values, suggesting that a hydrogen bond acceptor at this position is important.

Table 2. SAR of Biaryl Side Chain Derivatives with Second Phenyl Ring Ortho to the Amide.

entry	R1	R2	R3	R4	R5	IC50
12a	H	H	H	H	H	>10 μM
12b	OMe	H	H	H	H	>10 μM
12c	H	OMe	H	H	H	>10 μM
12d	H	OH	H	H	H	>10 μM
12e	H	H	OMe	H	H	>10 μM
12f	H	H	OH	H	H	>10 μM
12g	H	H	Cl	H	H	>10 μM
12h	H	Cl	Cl	H	H	>10 μM
12i	H	CF3	Cl	H	H	>10 μM

Table 3. SAR of Biaryl Side Chain Derivatives with Second Phenyl Ring Meta to the Amide.

entry	R1	R2	R3	R4	R5	IC50
7h	H	OMe	H	H	H	4.83 ± 0.9 μM
16a	H	H	H	H	H	>10 μM
16b	OMe	H	H	H	H	9.65 ± 0.1 μM
16c	OH	H	H	H	H	>10 μM
16d	OMe	OMe	H	H	H	>10 μM
16e	OMe	H	H	H	OMe	9.05 ± 0.4 μM
16f	H	OH	H	H	H	>10 μM
16g	H	N(Me)2	H	H	H	3.36 ± 0.3 μM
16h	H	NH2	H	H	H	7.23 ± 0.6 μM
16i	H	CH2OH	H	H	H	>10 μM
16j	H	Cl	H	H	H	>10 μM
16k	H	F	H	H	H	>10 μM
16l	H	CF3	H	H	H	>10 μM
16m	H	CF3	Cl	H	H	>10 μM
16n	H	Cl	Cl	H	H	>10 μM
16o	H	H	Cl	H	H	>10 μM
16p	H	H	CF3	H	H	>10 μM
16q	H	H	OMe	H	H	>10 μM
16r	H	H	OH	H	H	6.39 ± 1.4 μM

When combined with the data from Table 1, SAR reveals that electron donors that contribute to the carbonyl of the amide are important for activity, as well as the ability to withdraw electrons from the nitrogen of the amide bond. This combination of attributes suggests that a strong electronic contribution to the amide bond is important for the disruption of Aha1/Hsp90 interactions. Therefore, the amide bond was investigated.

Similar patterns of electron donation and withdrawal with respect to the amide bond have been shown to increase rotation about the central C–N bond, allowing such molecules to adopt cis-like amide bond character.¹⁹⁻²¹ As the activity of these derivatives correlates with the electron withdrawing/donating effects of the aryl substituents, the potential arose that a nontraditional amide bond could be important for Aha1/Hsp90 disruption. Compound 17, containing a 1,5-disubstituted 1,2,3,4-tetrazole, an established cis-amide isostere,^22,23 was synthesized to test this hypothesis. This compound, which was prepared in one step from 7h (Scheme 3), retained activity in the AlphaLISA assay, supporting the need for a bent or “C-shaped” conformation.

Scheme 3. Synthesis of 1,5-Disubstituted 1,2,3,4-Tetrazole (17), a cis-Amide Analog; Reagents and Conditions: (a) NaN₃, POCl₃, MeCN, 90 °C, 12 h, 71%.

The Aha1/Hsp90 disruptor KU-177 had previously been shown to ablate Aha1-stimulated P301L tau aggregation in a Thioflavin T (ThT) fluorescence assay.¹² Therefore, tau aggregation ThT fluorescence assays with Aha1/Hsp90 disruptors 16g and 17 were performed to determine whether these compounds retain the ability to prevent Aha1-stimulated P301L tau aggregation in vitro. As seen in Figure 3, both compounds significantly inhibited Hsp90-mediated and Aha1-stimulated P301L tau aggregation (p values <0.01) to a level that was statistically nonsignificant from tau alone. Furthermore, both compounds were found to be more effective at preventing tau aggregation in this assay than KU-177 (p values <0.05).

Figure 3.

Aha1/Hsp90 disruptors inhibit tau aggregation. Recombinant P301L tau aggregation measured by thioflavin T (ThT) fluorescence comparing the effects of 25 μM KU-177 to compounds 16g and 17. Equal amounts of DMSO were present in all experimental groups. Results represent the mean ± STDEV; *P < 0.05, **P < 0.01, ***P < 0.001.

Compounds 7h, 16g, and 17 were administered to SH-SY5Y neuroblastoma cells and Her2 overexpressing SK-BR-3 breast cancer cells to evaluate their activity within the cellular environment. Compounds 7h, 16g, and 17 exhibited the ability to disrupt interactions between Aha1 and Hsp90 in coimmunoprecipitation experiments (Figure 4a). Studies have shown that disruption of interactions between Hsp90 and cochaperones by small molecules is achievable in cells without significant inhibition or degradation of Hsp90 client proteins.²⁴ Importantly, these compounds disrupt the Aha1/Hsp90 complex without direct inhibition of Hsp90 protein folding activity, as Western blot analyses revealed these compounds do not induce the degradation of Hsp90 client proteins Her2 (in SK-BR-3 cells), Cdk6, or pAkt^S473 (in SH-SY5Y cells), nor do they induce the expression of Hsp70, a marker of the heat shock response. In contrast, the Hsp90 ATPase inhibitor, 17-AAG, induces the degradation of such clients as well as induction of the HSR (Figure 4b), clearly highlighting the potential advantage of developing Aha1/Hsp90 disruptors to treat neurodegeneration.

Figure 4.

Co-immunoprecipitation and Western blot analyses of SK-BR-3 cells and SH-SY5Y cells treated with Aha1/Hsp90 disruptors. (A) Co-immunoprecipitations of Hsp90 using Aha1 (C12) antibodies were performed on lysates from cells following 24 h of treatment with DMSO (vehicle/control) or a drug. (B) Western blot analysis of lysates from cells following 24 h of treatment with DMSO (vehicle/control) or a drug compared to the Hsp90 inhibitor, 17-AAG.

Data from these experiments support a pharmacophore model as shown in Figure 5. This model encompasses SAR implications from the data presented, notably, that electron withdrawing groups are preferred on the left side of the molecule, the position of the second aryl ring at the meta position is preferred, and that the cis-like character of the central amide linker is important for activity. Compounds investigated during the course of this study manifest the ability to inhibit Aha1-stimulated P301L tau aggregation and exhibit the ability to disrupt Aha1/Hsp90 interactions in the cellular environment without inhibition of Hsp90’s ATPase activity. Therefore, disruption of the Aha1/Hsp90 complex with small molecules represents a promising therapeutic strategy to inhibit pathological tau accumulation while avoiding the detrimental side effects associated with Hsp90 inhibition. Studies are currently underway to optimize analogs of 17, which will be reported in due course.

Figure 5.

Summary of SAR of Aha1/Hsp90 disruptors. Pharmacophore model for future development of small molecule Aha1/Hsp90 disruptors.

Glossary

Abbreviations

AD

Alzheimer’s disease

PD

Parkinson’s disease

Hsp90

heat shock protein 90

Aha1

activator of Hsp90 ATPase homologue 1

ThT

thioflavin T

Her2

human epidermal growth factor receptor 2

Cdk6

cyclin dependent kinase 6

Hsp70

heat shock protein 70

HSR

heat shock response

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00064.

• General experimental procedures, compound synthesis, characterization, and additional experimental details (PDF)

Author Contributions

All authors have approved the final version of this manuscript.

Research for this project was supported by supported by The National Institutes of Health [CA213586] and [N5075311] as well as the University of Notre Dame.

The authors declare no competing financial interest.