Selective targeting of Plasmodium falciparum Hsp90 disrupts the 26S proteasome

Summary

The molecular chaperone heat shock protein 90 (Hsp90) has an essential but largely undefined role in maintaining proteostasis in Plasmodium falciparum, the most lethal malaria parasite. Herein, we identify BX-2819 and XL888 as potent P. falciparum (Pf)Hsp90 inhibitors. Derivatization of XL888’s scaffold led to the development of Tropane 1, as a PfHsp90-selective binder with nanomolar affinity. Hsp90 inhibitors exhibit anti-Plasmodium activity against the liver, asexual blood, and early gametocyte life stages. Thermal proteome profiling was implemented to assess PfHsp90-dependent proteome stability, and the proteasome—the main site of cellular protein recycling—was enriched among proteins with perturbed stability upon PfHsp90 inhibition. Subsequent biochemical and cellular studies suggest that PfHsp90 directly promotes proteasome hydrolysis by chaperoning the active 26S complex. These findings expand our knowledge of the PfHsp90-dependent proteome and protein quality control mechanisms in these pathogenic parasites, as well as further characterize this chaperone as a potential antimalarial drug target.

Graphical Abstract

eTOC Blurb

PfHsp90 is a molecular chaperone essential to Plasmodium parasites. Mansfield et al. develop PfHsp90 inhibitors, including tropane-based compounds with species-selectivity, and leverage them to reveal a role for the chaperone in stabilizing the 26S proteasome complex.

Introduction

Nearly half of the world population remains at risk of malaria, a devastating infectious disease caused by obligate, intracellular parasites from the genus Plasmodium. Most of the 619,000 malaria mortalities reported in 2021 are attributed to one of five human-infective species, P. falciparum.¹ Due to parasite drug resistance, current first-line artemisinin combination therapies are losing efficacy in endemic areas. This spreading resistance leads to treatment failures and threatens eradication efforts.^2–4 Thus, it is critical to expand our understanding of essential parasite processes and characterize new druggable targets.

Plasmodium are transmitted to a human host via a mosquito vector. Thereafter, parasites undergo an asymptomatic liver stage before establishing a symptomatic blood stage of infection.^5,6 A subset of asexual parasites will then differentiate into gametocytes, which can be taken up by a mosquito, undergo sexual reproduction, and enable transmission to new hosts.⁷ These distinct stages are marked by notable changes in the parasite proteome, with transcriptomic and proteomic studies^8–15 suggesting that 50% of the proteome could vary between disparate stages. This extensive remodeling requires a robust protein chaperoning system. Additionally, such chaperones maintain proteostasis despite notable environmental stressors, including temperature variations between the vector and the host, as well as during malaria fever.^16–21

Heat shock protein 90 (Hsp90) is a conserved molecular chaperone that stabilizes an array of protein substrates, termed clients, and quaternary complexes through an ATP-driven folding process.^22,23 P. falciparum (Pf)Hsp90 is expressed throughout the parasite’s life cycle, with evidence supporting its essential role in the liver and blood stages.^24–26 Studies in fungal and human systems indicate that Hsp90 chaperones are not promiscuous but rather biased towards specific interactors,^27,28 and that its stabilizing interactions position it as a regulator of its substrates and their downstream activities. Furthermore, the Hsp90 interactome–referring to the totality of Hsp90 substrates and co-chaperones that modulate its activity–differs between species and cannot be identified by protein sequence. Based on the homology and pull-down evidence to date, PfHsp90 clients are expected to participate in signaling pathways, stress responses, developmental processes, and drug resistance mechanisms.^29,30 While these studies have offered critical insights, the overall PfHsp90-dependent proteome has not been thoroughly resolved, leaving questions about the scope of its substrates and their resulting biological processes.

Small molecule inhibitors that compete with ATP at the chaperone’s nucleotide binding domain (NBD) are critical tools to manipulate PfHsp90’s activity for biological studies. They are also potential antimalarial agents that have been found to inhibit parasites synergistically in drug combinations.^24–26 However, a limitation of such compounds has been their lack of selectivity for the parasite compared to the human Hsp90 (HsHsp90). Despite the 69% sequence similarity between the closest, cytosolic homologs^26,31—UniProt IDs: Q8IC05 (PfHsp90) and P07900 (HsHsp90)—species-selective binders are attainable. For example, SNX-2112 exhibits 17-fold greater affinity to the host HsHsp90,²⁶ and harmine is the prototypical compound that exhibits significant selectivity to the parasite PfHsp90.^26,32,33 Unfortunately, harmine’s use is limited by host toxicity and its low affinity (micromolar), prompting efforts to improve its efficacy.^34–36 Thus, potent, selective, and non-toxic PfHsp90 inhibitors remain an unmet need.

Herein, we identified PfHsp90-binding molecules, probed PfHsp90-dependent processes in Plasmodium, and validated a role for PfHsp90 in maintaining the parasite proteosome complex. Our efforts identified BX-2819 and XL888 as molecules that bind to PfHsp90 and inhibit the blood and liver stages of Plasmodium. We also provided evidence that Hsp90 inhibitors exhibit activity during early stages of gametocyte development. Promisingly, derivatization of XL888 led to the development of a molecule with improved species selectivity, Tropane 1. Thermal proteome profiling (TPP) indicated proteins with altered thermal stability as a consequence of PfHsp90 inhibition and suggested a connection between PfHsp90 and the parasite proteasome, which is the main site of cellular protein degradation. Our cellular and biochemical investigations led to mechanistic insights into this interaction and supported a role for PfHsp90 in stabilizing the active 26S proteasome complex from its constitutive 20S and 19S particles.^37,38 Thus, our discovery of potent, species-selective PfHsp90 inhibitors has revealed a previously uncharacterized link between this parasite chaperone and the proteasome.

Results

PfHsp90-binding small molecules with cellular activity were identified in a target-based screen

In order to identify putative PfHsp90 inhibitors, we evaluated compound binding using a fluorescently-linked chemoproteomic strategy.^39,40 In this approach, overexpressed PfHsp90-GFP in E. coli lysate was affinity captured onto an ATP resin. Competitive PfHsp90 inhibitors were then identified based on their displacement of PfHsp90-GFP bound to the immobilized ATP, resulting in increased eluent fluorescence collected in a 96-well catch plate (Figure S1A). The screen had a Z-factor = 0.54, calculated using 200 mM ATP as a positive control, and each plate included an ATP curve and DMSO controls to ensure assay performance (Figure S1B). An internally curated 3,419-member small molecule library³⁹ consisting of kinase inhibitors, nucleotide-like compounds, and previously untested HsHsp90 inhibitors were screened for binding. Compounds were screened at 500 μM to compete with the high ATP concentration on the resin (~10 mM) (Figure 1A). Screening positives (88) were selected as compounds that increased relative fluorescence ≥2-fold over the DMSO negative controls. As a secondary assay, eluent from screening positives was evaluated by western blot to confirm protein elution (Figure S1C), consistent with the competitive displacement of PfHsp90-GFP.

Figure 1. Target-based screen to identify PfHsp90 binders.

(A) Small molecules were identified based on their competitive displacement of PfHsp90-GFP from an ATP resin, as measured by GFP fluorescence in eluent (grey circles). Compounds that increased fluorescence ≥2-fold relative to DMSO (orange dash) were selected as initial screening positives (88). A secondary western blot assay confirmed 7 compounds (orange squares) competitively displaced the protein. (B) The 7 screening actives were assessed for inhibition of P. berghei load in HuH7 cells at 30 μM. Compounds 1 and 2 markedly decreased relative parasite load (~0%) and were prioritized for additional studies. Dashed line indicates 50% inhibition. Data shown as means ± SEM, n = 3. (C) Overview of screening strategy. (D) Structures of BX-2819 and XL888.

This effort confirmed 7 compounds (0.2 % hit rate) bound to PfHsp90. These small molecules were then tested for inhibition of liver stage Plasmodium berghei in hepatocytes as well as for hepatocyte cytotoxicity to evaluate cellular activity. Rodent malaria models, like P. berghei, are the most commonly utilized liver stage Plasmodium model systems.⁴¹ Importantly, P. berghei (Pb)Hsp90 and PfHsp90 share 91% sequence similarity, with demonstrated correlation of Hsp90 inhibitor activity.²⁶ Inhibition of P. berghei load in HuH7 cells was observed for 3 of the 7 compounds tested at ~30 μM (Figure 1B), without significant HuH7 cytotoxicity (<50%). Among these, 2 decreased parasite load to near zero at the tested concentration and were selected for follow-up studies (Figure 1C). These compounds were identified as BX-2819 and XL888, which they will be referred to as hereafter (Figure 1D).^42,43

XL888 and BX-2819 are high-affinity PfHsp90 inhibitors with multi-stage anti-Plasmodium activity

To determine ligand binding affinity, full-length PfHsp90 and HsHsp90 were purified (Figure S2A) and utilized in a competitive binding assay. Our assay employs FITC-labeled geldanamycin (GA)—a well-characterized Hsp90 inhibitor that binds both PfHsp90 and HsHsp90 (Figure S2B)—as a fluorescent tracer.^26,44 Fluorescence polarization is observed when FITC-GA binds to Hsp90, and this signal decreases when it is displaced by ligands. Dose-dependent displacement of FITC-GA was observed for both XL888 and BX-2819. Competitive binding curves (Figure 2A, 2B) were fit to calculate the apparent inhibitory constants (K_i) for PfHsp90 of 27 ± 7.8 nM for XL888 and 24 ± 4.4 nM for BX-2819. Notably, XL888 and BX-2819 have previously reported activity against HsHsp90,^42,43 but they are novel inhibitors of PfHsp90. To assess species selectivity, we also quantified HsHsp90 affinity, calculated to be 130 ± 30 nM for XL888 and 49 ± 4.3 nM for BX-2819 in our assay. The difference in affinity between PfHsp90 and HsHsp90 for XL888 (p = 0.0032) and BX-2819 (p = 0.0114) corresponds to a selectivity to PfHsp90 of 4.8-fold and 2.0-fold, respectively.

Figure 2. Binding affinity and cellular potency of XL888 and BX-2819.

(A-B) Binding plots for XL888 (A) and BX-2819 (B) demonstrating their dose-dependent displacement of the FITC-GA tracer to PfHsp90 (orange) and HsHsp90 (blue). For XL888 (A), competitive binding curves calculate an apparent K_i = 27 ± 7.8 nM for PfHsp90 compared to K_i = 130 ± 30 nM for HsHsp90 (p = 0.0032). For BX-2819 (B), competitive binding curves calculate an apparent K_i = 24 ± 4.4 nM for PfHsp90 compared to K_i = 49 ± 4.3 nM for HsHsp90 (p = 0.0114). Data shown as means ± SEM, n ≥ 4; unpaired t-test. (C-D) Dose-response curves for XL888 (blue) and BX-2819 (orange) inhibition of P. falciparum blood stage parasites (C), yielding EC₅₀s of 0.33 ± 0.11 μM and 0.74 ± 0.034 μM, respectively, as well as P. berghei liver stage parasites (D), yielding EC₅₀s of 0.24 ± 0.045 μM and 1.5 ± 0.41 μM, respectively. Data shown as means ± SEM, n ≥ 2. (E) The PfHsp90 inhibitors BX-2819 (BX; 500 nM), XL888 (XL; 200 nM), and geldanamycin (GA; 300 nM) slow parasite progression through their erythrocytic cycle when treated at the ring stage (~12 hours post invasion) compared to treatment with DMSO. (F) Early and late P. falciparum gametocyte viability is decreased by geldanamycin (GA) at 1 μM and 10 μM compared to the DMSO control. Data shown as means ± SEM, n = 3. *<0.05, **<0.01, ****<0.0001; two-way ANOVA, Tukey’s multiple comparison.

The potency of XL888 and BX-2819 against blood stage P. falciparum 3D7 and liver stage P. berghei was also assessed using established high-throughput, cell-based assays (Figure 2C, 2D).⁴¹ The half maximal parasite inhibition effective concentrations (EC₅₀s) for XL888 were 0.33 ± 0.11 μM and 0.24 ± 0.045 μM for blood and liver stages, respectively. The EC₅₀s for BX-2819 were 0.74 ± 0.034 μM and 1.5 ± 0.41 μM for blood and liver stages, respectively. At 10 μM, neither XL888 nor BX-2819 decreased HuH7 viability by >50 % (Figure S2C). XL888 and BX-2819 were also found to slow parasite progression through their erythrocytic cycle compared to DMSO, consistent with the previously reported effects of geldanamycin (Figure 2E).²⁵

While the dual-stage efficacy of PfHsp90 inhibitors was previously appreciated, their activity against sexual gametocytes has not been reported. Thus, as a proof-of-principle, we investigated the effects of the better characterized geldanamycin against early- (day 1–6) and late- (day 10–12) stage gametocytes.^24,25 Geldanamycin significantly decreased gametocyte viability with a modest effect against late gametocytes (71% at 1 μM, p = 0.0487; 56% at 10 μM, p = 0.0023) and with greater efficacy against early gametocytes (37% at 1 μM, p < 0.0001; 16% at 10 μM, p < 0.0001) (Figure 2F).

Tropane-based analogs demonstrate improved selectivity to PfHsp90

The structure of HsHsp90 bound to different ligands was inspected to generate hypotheses on possible ligand selectivity mechanisms. Between binding XL888 and BX-2819, Lys112 repositions in HsHsp90, which corresponds to Arg98 in PfHsp90 (Figure 3A). This residue is located within the lid subdomain, a structure that closes over the ATP binding pocket. Upon binding BX-2819 (PDB: 3HHU), Lys112 exhibits apparent greater flexibility as it occupies different conformations between HsHsp90 protomers, whereas, upon binding XL888 (PDB: 4AWO), it is solely directed outward in a conformation that mirrors the residue positioning in the lid’s closed state (Figure S3A). In the closed and XL888-bound conformations, Lys112 forms a salt bridge with Glu25 on a N-terminal helix, a residue that is substituted by Asp11 in P. falciparum. These residue substitutions, which can also form a salt bridge, are shared by other Plasmodium species, but not Candida albicans nor Cryptococcus neoformans—fungal pathogens with species-selective Hsp90 inhibitors (Figure S3B).

Figure 3. Characterization of selective binding and identification of Tropane 1.

(A) The structural overlay of HsHsp90 bound to XL888 (PDB: 4AWO, black ribbons) and BX-2819 (PDB: 3HHU, white ribbons), displaying a section of the lid subdomain (Thr99-Gln123) and N-terminal helix (Phe22-Thr36). The residues Lys112 and Glu25 form a salt bridge in the XL888-bound HsHsp90 structure. These residues are substituted by Arg98 and Asp11 in PfHsp90, respectively. (B-C) The dose-dependent displacement of FITC-GA by XL888 (B) or BX-2819 (C) from purified wild-type (WT) or salt bridge double mutant (E25D/K112R) HsHsp90 nucleotide-binding domain (NBD). XL888 exhibits a higher affinity to the E25D/K112R mutant compared to the wild-type NBD (B), with apparent K_i = 9.6 ± 1.6 nM and K_i = 68 ± 23 nM, respectively (p = 0.0463). BX-2819 does not exhibit a change in affinity to the E25D/K112R mutant compared to the wild-type NBD (C), with apparent K_i = 13 ± 2.5 nM and K_i = 19 ± 4.0 nM, respectively (p = 0.2103). Data shown as means ± SEM, n = 4; unpaired t-test. (D) Heatmap (light blue to black) displays the amount of FITC-GA bound to PfHsp90 or HsHsp90 after displacement by 4 tropane-derived compounds at 100 nM. FITC-GA bound to PfHsp90 is competitively displaced (<50%) by compounds 1-3. A higher ratio of bound FITC-GA to HsHsp90 compared to PfHsp90 (Hs/Pf; color coded orange) for 1 suggests improved selectivity. (E) Structure of Tropane 1. (F) Table summarizing Hsp90 binding affinity and Plasmodium inhibition potency for Tropane 1. The compound exhibits 9.6-fold selectivity for PfHsp90 over HsHsp90 and retains activity against the P. falciparum blood and P. berghei liver stages without host HuH7 cell cytotoxicity. Apparent K_i values reported as means ± SEM, n = 4; *<0.05; unpaired t-test. Parasite inhibition EC₅₀ and host viability values reported as means ± SEM, n = 2.

To investigate the contribution of this salt bridge to the relative XL888 selectivity, HsHsp90 nucleotide-binding domain (NBD) constructs with E25D/K112R substitutions as well as the wild-type (WT) sequence were generated and purified. The isolated variant mimics the proposed PfHsp90 salt bridge, and would therefore be predicted to exhibit improved XL888 affinity. For comparison, two other NBD constructs with substitutions in non-conserved residues were isolated: one with mutations in the lid LXXGA/IXXSG motif (L122I/G125S/A126G) and another with a mutation in the N-terminal helix (Q23N) (Figure S3C). These substitutions were selected as they have previously been investigated in P. falciparum and C. albicans, respectively.^45,46 The purified HsHsp90 NBD variants, as well as the purified WT HsHsp90 NBD, all bound FITC-GA with similar affinity (Figure S3D). The affinities of each construct for XL888 and BX-2819 were next assessed by competition with FITC-GA. Notably, XL888 exhibited a 7-fold improvement in affinity to the E25D/K112R double mutant when compared to the WT HsHsp90 NBD (Figure 3B): K_i = 9.6 ± 1.6 nM versus K_i = 68 ± 23 nM (p=0.0463), respectively; but no significant differences were observed with the other constructs tested (Figure S3E). Further, BX-2819 affinity was unaltered for all constructs tested (Figure 3C, S3F).

Derivatives of the tropane (XL888)-based scaffold were developed to improve species selectivity. After developing 4 tropane-containing compounds, 3 demonstrated greater FITC-GA displacement (<50% bound) from PfHsp90 compared to HsHsp90 in a single-point binding assessment (Figure 3D). Among these, Tropane 1 was selected for additional characterization based on its apparent superior potency and selectivity (Figure 3E). The affinity of Tropane 1 to full-length PfHsp90 and HsHsp90 was calculated as 46 ± 6.8 nM and 440 ± 130 nM, respectively (Figure 3F, S3G). This difference in affinity (p = 0.0217) represents a 9.6-fold selectivity. Anti-parasite activity was additionally tested to ensure cellular activity was retained. Tropane 1 inhibited blood stage P. falciparum 3D7 and liver stage P. berghei parasites with EC₅₀s of 0.73 ± 0.15 μM and 1.2 ± 0.16 μM, respectively, without reducing host cell HuH7 viability (Figure 3F, S3H, S3I).

The proteasome exhibits perturbed stability upon PfHsp90 inhibition

As PfHsp90 acts by directly binding diverse substrates, we implemented TPP to assess P. falciparum protein thermal stability—which we predicted could be perturbed in PfHsp90 interactors and their downstream processes—upon PfHsp90 inhibition.⁴⁷ Thermal melt curves were generated for individual proteins using a bottom-up mass spectrometry approach to quantify the fraction of soluble, folded proteins in P. falciparum lysate treated with BX-2819 after varying degrees of thermal challenge. The point at which half the fraction of a particular protein is denatured, or its melt temperature (Tm), is representative of its thermal stability.⁴⁷ To control for false positives due to off-target effects, we utilized an inactive BX-2819 analog, HS292 (Figure 4A), as a negative control. HS292 has a methoxyl substitution of a key resorcinol hydroxyl group and was developed during derivatization efforts of the BX-2819 scaffold.^38–40 This alteration ablates a pivotal hydrogen bond with PfAsp79/HsAsp93, which is a widely conserved and dominant energetic interaction for ligand binding (Figure S4A).⁴⁸ We assessed HS292 binding to PfHsp90, and, as expected, the compound did not compete with FITC-GA (Figure 4B). Moreover, HS292 did not inhibit Plasmodium parasites at 10 μM (Figure S4B), nor did it alter the Tm of purified PfHsp90 NBD in a thermal shift assay at 500 μM (Figure S4C). Thus, we implemented both BX-2819 and HS292 in our TPP approach as an invaluable positive and negative chemical probe pair to control for off-target binding and non-specific effects (Figure S4D).

Figure 4. Thermal proteome profiling (TPP) suggests PfHsp90-dependent processes.

(A) The substitution of a hydroxyl group in BX-2819 to a methoxyl group yields the inactive derivative HS292. (B) BX-2819 (10 μM) displaces FITC-GA, whereas HS292 (100 μM) does not compete for binding to PfHsp90. Data shown as means ± SEM, n = 5; not significant (ns)>0.05, ****<0.0001; one-way ANOVA, Dunnett’s multiple comparison. (C) Plasmodium proteins that did not exhibit a stability change (−2.0 °C < ΔTm < +2.0 °C; dotted line) upon treatment with the inactive HS292 but did exhibit a stability change (ΔTm < −2.0 °C or ΔTm > +2.0 °C; dotted line) upon treatment with the active BX-2819 inhibitor compared to DMSO were selected as TPP hits (orange fill indicates selection criteria for each subset). (D) Network of interactions between TPP hits. The PfHsp90 node (blue) represents both Q8IC05 (cytosolic) and Q8III6 (mitochondrial) parasite paralogs, whereas remaining nodes represent other TPP hit proteins with UniProt Identifiers. Edges connecting nodes represent predicted interactions between the proteins, with the line width corresponding to confidence. Only proteins within two edges of the PfHsp90 node are displayed. The significant local network cluster representing the proteasome (CL: 4475; false discovery rate: 0.0067) is highlighted in orange. (E) Gene ontology (GO) enrichment detailing biological processes of TPP hits. GO terms are scaled with the color representing the significance of the enriched annotations and the size representing the number of annotations for the ascribed process in the database.

To probe PfHsp90-dependent processes, Tm values quantified from P. falciparum lysate (trophozoite stage) treated with 500 μM of the active PfHsp90 inhibitor BX-2819 were compared to Tm values quantified from lysate treated with 500 μM of the inactive HS292. The thermal melt curve data collected for all identified proteins are summarized in Table S1. A total of 537 proteins were identified with curves meeting quality standards (R>0.8) in both BX-2819 and HS292 treatments compared to DMSO treatment. There was no apparent difference in global protein stability between BX-2819 and HS292 treatments (Figure S4E), which is expected as PfHsp90 chaperones only a minor subset of the proteome. To ascertain PfHsp90-dependent perturbations, 50 hit proteins were identified that exhibited a Tm shift of greater magnitude than +2.0 °C or −2.0 °C (i.e., ΔTm<−2.0 °C or ΔTm>+2.0 °C) upon BX-2819 treatment compared to DMSO (corresponding to a Z-score of ~1) but not upon HS292 treatment (i.e., −2.0<ΔTm<2.0 °C) (Figure 4C, Table S2). Representative thermal melt curves generated for selected hits and non-hits are shown in Figures S5F–S5I. Included in the hits were both cytosolic (UniProt ID: Q8IC05) and mitochondrial-targeted (UniProt ID: Q8III6) P. falciparum Hsp90 paralogs. Despite limited knowledge of its substrates, the STRING database was used to generate hypotheses for interactions between PfHsp90 and hit proteins in our TPP dataset (Figure 4D).⁴⁹ P. falciparum proteins are represented by nodes in the network where the conjoining edges link them to predicted interactors. Further, the STRING database indicated enriched local network clusters corresponding to the proteasome and its assembly (CL: 4475, 4468; false discovery rate: 0.0067, 0.0065).⁴⁹ PfHsp90 was linked to these clusters via a putative interaction with two 19S regulatory particle subunits (UniProt IDs: Q8I323 and Q8II60). The inverse analysis—comparing proteins perturbed by HS292 but not BX-2819—did not result in any specific local network clusters (false discovery rate > 0.01), suggesting identified enrichments are a consequence of chaperone inhibition. Gene ontology (GO) annotation further indicated proteasome assembly and proteasome regulatory particle assembly (GO:0070682; p= 3.55e-3), in addition to other catabolic, homeostasis, and DNA maintenance processes, as enriched biological processes represented in the hit proteins (Figure 4E).^50,51

PfHsp90 promotes proteasome activity

Based on our TPP analysis, we sought to validate a functional interaction between PfHsp90 and the P. falciparum proteasome. Thus, proteasome activity in P. falciparum lysate (trophozoite stage) was monitored based on the fluorescence of released AMC after proteasome-mediated cleavage of the well-characterized fluorogenic peptide substrate Suc-Leu-Leu-Val-Tyr-AMC (Suc-LLVY-AMC).^52,53 We confirmed that the competitive proteasome inhibitor bortezomib significantly reduced Suc-LLVY-AMC hydrolysis in P. falciparum lysate compared to DMSO (negative control) over the reaction period.⁵⁴ In contrast, supplementing purified PfHsp90 (3 μM) increased substrate cleavage (Figure 5A). Purified PfHsp90 in assay buffer without lysate did not independently increase substrate hydrolysis over the same timeframe (Figure S5). The addition of purified PfHsp90 (3 μM) resulted in a 152% rate of Suc-LLVY-AMC hydrolysis (p < 0.0001), normalized to the rate of the DMSO control (100%), whereas addition of BX-2819 to PfHsp90-treated samples reduced hydrolysis in a dose-dependent manner (Figure 5B). The concurrent addition of 3 μM purified PfHsp90 and 100 μM BX-2819 resulted in a reaction rate that was not different from the DMSO control (p=0.87).

Figure 5. PfHsp90 increases proteasome substrate hydrolysis in vitro.

(A) Time-course measurements of the relative hydrolysis of the proteasomal substrate Suc-LLVY-AMC in P. falciparum lysate. Compared to DMSO (white fill square), addition of PfHsp90 (3 μM; blue fill circle) increased total hydrolysis whereas the proteosome inhibitor bortezomib (25 μM; orange fill square) decreased total hydrolysis over an 180 minute incubation. Data shown as means ± SEM, n = 3–4; ****<0.0001; one-way ANOVA, Dunnett’s multiple comparison. (B) The relative rate of Suc-LLVY-AMC hydrolysis upon DMSO, bortezomib (25 μM), and PfHsp90 (3 μM) treatments. Relative rate determinations were made by comparing the linear slope of Suc-LLVY-AMC hydrolysis, excluding values before the 30 min time-point (A, dashed line). Concurrent chemical inhibition with BX-2819 (BX; 10 – 100 μM) negated the increased rate (~150%) of substrate hydrolysis upon PfHsp90 addition relative to the DMSO control. Data shown as means ± SEM, n = 3–4; not significant (ns)>0.05, **<0.01, ***<0.001, ****<0.0001; one-way ANOVA, Dunnett’s multiple comparison.

Hsp90 inhibition reduces active 26S proteasomes in P. falciparum

We next implemented native polyacrylamide gel electrophoresis to interrogate the relative activity of distinct proteasome complexes based on their in-gel hydrolysis of Suc-LLVY-AMC.^55,56 To first confirm the identity of the 26S and 20S complexes, we resolved P. falciparum lysate treated with DMSO or the covalent proteasome inhibitor epoxomicin. The most prominent band with DMSO treatment corresponded to the 26S proteasomes, while the 20S complexes were not readily observed. Accordingly, while the 20S structure includes the catalytic subunits and can exist without a 19S cap, it is generally latent in a gated conformation without an associated activator.^57,58 Thus, after imaging 26S complexes, we employed a mild SDS treatment to activate and visualize 20S particles, as previously reported.^55,59 Both the 26S and activated 20S bands were reduced by epoxomicin, corroborating their identity and supporting the proteasome native gel assay (Figure S6A).⁶⁰

To assess the impact of PfHsp90 on the active 26S proteasome, we resolved lysate prepared from cycloheximide chase experiments. P. falciparum cultures were treated for 3 hours with either geldanamycin (10 μM) or DMSO, where cycloheximide (50 μM), a translation inhibitor, was included as a co-treatment. Cycloheximide addition enabled direct interrogation of the effects of Hsp90 inhibition on existing complexes and controlled for potential effects from compensatory translational responses or altered parasite development. Geldanamycin treatment resulted in a decrease in the 26S band and an increase in the 20S band after SDS activation (Figure 6A). The normalization of total protein input was confirmed by Coomassie staining (Figure S6B). Densitometry analysis indicated that the proportion of active 26S proteasomes decreased to 63% (p = 0.0005) and latent 20S particles increased to 158% (after activation; p = 0.0241) after geldanamycin treatment when compared to the DMSO control (Figure 6B, 6C).

Figure 6. PfHsp90 inhibition effects the proportion of active 26S complexes.

(A) The relative activity of P. falciparum proteasome complexes based on their in-gel hydrolysis of Suc-LLVY-AMC. Lysate was prepared from P. falciparum (trophozoite stage) treated with 10 μM geldanamycin (GA) or DMSO for 3 hours, with a cycloheximide (50 μM) co-treatment. A decreased proportion of active 26S complexes (consisting of associated 20S and 19S particles), and an increased proportion of latent 20S complexes (after SDS activation) was detected in GA compared to DMSO treated samples. (B-C) The relative quantification of Suc-LLVY-AMC hydrolysis from native gel experiments corresponding to 26S proteasome activity (B) and 20S particle activity after SDS activation (C) between GA and DMSO treated samples. Data shown as means ± SEM, n = 3; *<0.05, ***<0.001, unpaired t-test. (D) The relative hydrolysis of Suc-LLVY-AMC monitored over time after P. falciparum treatment with 350 nM Tropane 1 (orange) or DMSO (blue). The species-selective PfHsp90 inhibitor resulted in reduced total substrate hydrolysis compared to DMSO. Data shown as means ± SEM, n = 3; *<0.05, unpaired t-test.

Geldanamycin is capable of inhibiting both host and parasite chaperones; therefore, we employed Tropane 1 at a sub-lethal concentration to selectively target PfHsp90. The compound was first tested for P. falciparum blood stage inhibition in combination with bortezomib to determine a suitable concentration, with 250 – 1,000 nM Tropane 1 appearing to sensitize parasites to proteasome inhibition (Figures S6C, S6D). Parasites were then treated with Tropane 1 for 5 hours at 350 nM—a concentration within the sensitizing range, corresponding to ½-EC₅₀, and with no observed effects on P. falciparum viability. After treatment, Suc-LLVY-AMC cleavage was measured over time in P. falciparum lysate to assess total proteasome activity. Significantly less total substrate was hydrolyzed with Tropane 1 treatment compared to DMSO treatment (Figure 6D).

Discussion

P. falciparum (Pf)Hsp90 is expected to maintain proteostasis throughout parasite development based on its direct binding and chaperoning interactions with specific substrates.^{18,24–26,31} PfHsp90 processes proteins into a mature and functional state, enabling the downstream functions of its clients and positioning the chaperone as a regulator of varied cellular processes.^61–64 To support continued investigation of this essential chaperone, we first completed a target-based screen to identify high-affinity PfHsp90 inhibitors. The most promising hits were identified as BX-2819 and XL888.^42,43 Ideally, a PfHsp90 inhibitor would exhibit species selectivity such that it could be employed to specifically target the parasite protein with reduced binding of the essential HsHsp90 homolog.^26,65 Both BX-2819 and XL888 exhibited high-affinity (Kd = 24 ± 4.4 nM and Kd = 27 ± 7.8 nM, respectively) and mild selectivity to PfHsp90 over HsHsp90 (2.0-fold and 4.8-fold, respectively). While both compounds have previously described activity against HsHsp90, their activity against PfHsp90 was unknown. Moreover, XL888 is a unique tropane-based compound,⁴² which is a class of Hsp90 inhibitors that as a whole are previously untested against PfHsp90. Interestingly, we were able to improve PfHsp90 selectivity within the synthesis and testing of 4 tropane-based compounds. This effort yielded Tropane 1 with an apparent affinity of 46 ± 6.8 nM for PfHsp90 and 440 ± 130 nM for HsHsp90, corresponding to 9.6-fold species selectivity.

The species selectivity of tropane-based inhibitors may at least in part be due to the substituted pairing of Glu25 and Lys112 in HsHsp90 to Asp11 and Arg98 in PfHsp90. In the XL888-bound HsHsp90 structure (PDB: 4AWO), these residues form a salt bridge between a N-terminal helix and the protein’s lid subdomain, which is the most dynamic region of the nucleotide-binding domain. Notably, Hsp90 inhibitors are known to either bind solely within the ATP-binding pocket or to also occupy an additional space formed within the lid; XL888 occupies both.⁶⁶ Lid rearrangements have further been described to accommodate the binding of selective inhibitors between Hsp90 family proteins, including in C. albicans and C. neoformans.^45,67,68 These eukaryotic pathogens lack the same salt bridge substitution as Plasmodium, hinting that additional factors may influence lid conformations that contribute to selectivity.

In P. falciparum, unique lid dynamics led to the identification of 7-azaindole compounds as a selective scaffold for PfHsp90 inhibitor development.⁴⁶ Interestingly, an altered LXXGA/IXXSG motif at the lid periphery—not the single R98K substitution—was a determinant for 7-azaindole selectivity. This contrasts with our data where the LXXGA/IXXSG motif did not affect HsHsp90 tropane binding, but the Plasmodium Lys-Arg mutation (represented in our salt bridge double mutant) was more favorable. Thus, each substitution may favor a conformation better suited to distinct scaffolds. In particular, 7-azaindole binding and associated dynamic simulations were determined with the lid in the more open conformation—lacking the salt bridge that is observed in the closed and XL888-bound conformations. Further structural research is warranted to confirm if the Asp11 and Arg98 pair promotes a PfHsp90 lid conformation that better accommodates tropane-based inhibitors, which would be in line with previous findings in Saccharomyces cerevisiae demonstrating that N-terminal resides can modulate overall lid flexibility.⁶⁹ Given this apparent molecular basis for selectivity and productive albeit limited structure-activity relationship study, we expect the tropane-scaffold could be well-suited for further design of high-affinity, PfHsp90-specific binders.

Based on the lethality of its inhibition and previous genetic studies, PfHsp90 is essential to liver and blood stage Plasmodium parasites.^24–26,70 Notably, PfHsp90 is under consideration as an antimalarial drug target, with early indications that its inhibitors have favorable recalcitrance to resistance-generation and have the potential to act synergistically in combination therapies.⁶⁵ Tropane 1, XL888, and BX-2819 were all potent inhibitors of both the liver and blood stages of Plasmodium infection. They exhibit single-digit to sub-micromolar effective concentrations, without notable host toxicity. This suggests their potential for malaria treatment and prophylaxis, respectively.⁶⁵ Additionally, the ineffectiveness of the inactive resorcinol-based HS292 in cell-based assays supports that parasite inhibition results from on-target binding to Hsp90. As P. falciparum and P. berghei were used in our blood and liver models of infection, respectively, these compounds thus exhibit efficacy against multiple Plasmodium species at multiple stages.²⁶ Further, while the chaperone is expressed during the gametocyte stage,⁹ it was untested if PfHsp90 inhibitors affect this sexual form of the parasite that enables future transmission. Our proof-of-principle study with the well-characterized Hsp90 inhibitor geldanamycin also supports an essential role for the protein during early gametocyte development.

We implemented the chemoproteomic analysis TPP to assess the consequences of PfHsp90 inhibition on P. falciparum proteome stability, thereby providing insight into PfHsp90-mediated biological processes. Altered protein-ligand and protein-protein interactions—including alterations in a protein’s hetero-oligomeric state resulting from administration of a specific inhibitor—change a protein’s melt behavior.⁴⁷ Thus, we predicted PfHsp90 inhibition would further perturb the thermal stability of its binding partners and their downstream processes.⁷¹ For this purpose, we employed BX-2819, given its high affinity and our development of its inactive analog HS292 as an invaluable negative control. The completion of this study in P. falciparum lysate circumvented the need to employ a more species selective compound. The subset of proteins impacted by treatment with BX-2819 but not HS292 included two PfHsp90 paralog hits, which is indicative of the ligand-target binding interaction itself stabilizing the proteins. Based on this principle, TPP has primarily been applied for small molecule target deconvolution.^47,72–74 However, as alterations in protein complexation will also affect their thermal stability, our TPP analysis is sensitive to protein-protein interactions.^47,75–79 Such stability perturbations are detectable in lysate, but future studies may implement an in-cell TPP approach to increase detection of lower affinity or less abundant client interactions given that physiological concentrations and stoichiometries would not be disrupted. Regardless, the sensitivity of TPP to both ligand-protein and protein-protein associations in lysate has been demonstrated, including in P. falciparum where the small molecule clemastine was found to alter the melt behavior of the entire TCP-1 ring complex upon on its binding to a single subunit.⁸⁰

In our TPP network, there were enriched local network clusters corresponding to the proteasome and its assembly, with proteasome subunits further mapped to have putative interactions with PfHsp90. Proteomic studies in other organisms have suggested analogous interactions between homologs.^81,82 Additional reports support that Hsp90 proteins can directly bind the proteasome⁸³ and either activate or inhibit its hydrolysis depending on the substrate.^84–87 This implicates Hsp90s as regulators of proteasome activity, but their interaction and its functional consequences were unexplored in P. falciparum. To address this, we established that exogenous PfHsp90 increased the rate of proteasome hydrolysis in parasite lysate, which was negated by chemical inhibition of PfHsp90. This provides evidence that PfHsp90’s ATPase chaperoning activity supports proteasome function. Given that proteasome regulatory particle assembly was a significant biological process represented in our TPP dataset, we reasoned PfHsp90 could promote the assembly and association of the 19S regulatory particle with the 20S core particle to form the active 26S complex. Encouragingly, evidence is growing in model systems to support such a role for Hsp90 chaperones,^88,89 which further suggest contributions from Hsp70 and their mutual co-chaperone HOP that could be further explored in Plasmodium. In our cycloheximide chase experiments, we observed a reduced proportion of active 26S and increased proportion of latent 20S complexes after geldanamycin administration. These results are consistent with a destabilization of capped proteasomes. The reduced proteasome activity in response to a non-lethal dose of Tropane 1 further corroborates that the proteasome is impacted due to specific inhibition of PfHsp90. Collectively, these data support that PfHsp90 chaperones the active 26S complex with associated 20S with 19S particles.

Our TPP analysis provides evidence for overall PfHsp90-mediated processes and inspires hypotheses for future functional investigations. Of additional interest towards its relationship with the proteasome is that the 19S subunit RPN8 (UniProt ID: Q8I323) is a hit protein and PfHsp90 interactor in our analysis. Mutagenesis studies support that this subunit has a critical contribution to 19S integrity and formation.^90,91 Thus, future research is warranted into the specific interaction between PfHsp90 and RPN8 toward maintaining the 19S cap stability. In addition to the proteasome, biological processes relating to DNA repair and conformation are highly enriched in our TPP hits. This aligns with recent validation studies that support the P. falciparum recombinase PfRad51 and histone deacetylase PfSir2A as the first bona fide PfHsp90 clients.^92,93 These proteins are important in the double-strand break repair pathway and in epigenetic regulation, respectively, collectively implying that the chaperone has a role in parasite genome maintenance that could be further explored. Moreover, given that the STRING database assigns interactors based on prior evidence, which is sparse for Plasmodium, we anticipate TPP stability alterations exhibited by some proteins not depicted as a PfHsp90 interactor may result from an unrealized direct binding to the chaperone. This would enable future de novo client discovery.

Overall, we identify high-affinity, PfHsp90-binding molecules and utilize them as chemical probes to reveal chaperone-mediated biological processes. Our efforts have developed the selective Tropane 1 compound, and further demonstrate the multi-stage, anti-Plasmodium potential of PfHsp90 inhibitors. Our assessment of stability perturbations after PfHsp90 chemical inhibition led to the validation of PfHsp90’s role in supporting 26S proteasome activity. Both PfHsp90 and the P. falciparum proteasome are key components of protein quality control—ensuring proper folding or degradation of proteins—thus placing PfHsp90 as a regulator of proper protein maturation and recycling. Ultimately, continued investigation of their intersecting activities would elucidate mechanisms of proteome regulation and inform development of putative drugs that disrupt proteostasis to prevent and treat malaria.

Limitations of the study

A limitation of our study is that our TPP experiments were conducted in lysate. While detectable in lysate, stability alterations in protein interactions are generally more pronounced when drug administration and thermal challenge are conducted in intact cells.^75,78,80 Consequently, lysate-based experiments may be more biased toward high-affinity or -abundance interactors. Future studies employing an in-cell TPP approach would complement our results to support our findings and possibly expand the known PfHsp90-dependant proteome. Our characterization of PfHsp90 inhibitors also does not consider compound affinity in the cellular context. Distinct co-chaperones can alter PfHsp90 inhibitor binding and organelle-specific PfHsp90 isoforms may exhibit differential affinity.⁶⁸ These variables could be further explored in subsequent derivatization efforts to improve the selectivity and potency of PfHsp90 inhibitors in its cellular environment.

Significance

Plasmodium falciparum (Pf)Hsp90, an essential molecular chaperone, is recognized as a putative drug target amidst spreading parasite resistance. Despite the promise of PfHsp90-targeting small molecules, there is a lack of high-affinity, species-selective inhibitors to enable potential antimalarial drug development. Such molecules could also be employed as chemical probes to provide information regarding the identity and biological functions of the chaperone’s substrates. Herein, we characterize previously undescribed PfHsp90-binding molecules and provide evidence that a tropane-based scaffold is well-suited for the development of high-affinity and species-selective inhibitors. We couple our PfHsp90-binding chemical probes with thermal proteome profiling to uncover putative PfHsp90 interactions. From this dataset, we investigate a role for PfHsp90 in activating the proteasome by chaperoning the 26S complex. In total, this study provides chemical tools and potential PfHsp90-interacting proteins to expand the current understanding of PfHsp90 function. Our molecular studies position the chaperone as a global regulator of P. falciparum proteostasis—beyond its well-appreciated role in stabilizing substrates—by facilitating proteasomal degradation.

STAR Methods

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Emily R. Derbyshire (emily.derbyshire@duke.edu).

Materials availability

Reagents generated in this study will be made available upon request subject to a Materials Transfer Agreement.

Data and Code Availability

• The TPP data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are publicly available as of the date of publication. Results from the small molecule screen have been deposited on PubChem and are publicly available as of the date of publication. The associated dataset identifiers are noted in the key resource table.

• Software used for analysis is available at provided online locations noted in the key resource table. This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
GFP Antibody (B-2) Alexa Fluor® 488	Santa Cruz Biotechnology	Cat# sc-9996 AF488
HisProbe™-HRP Conjugate	ThermoFisher Scientific	Cat# 15165
Bacterial and virus strains
BL21 (DE3)	New England Biolabs	Cat# C2527
NEB5α	New England Biolabs	Cat# C2987
Biological samples
Human Blood	Gulf Coast Regional Blood Center	N/A
P. berghei ANKA infected Anopheles stephensi mosquitoes	University of Georgia SporoCore	N/A
Chemicals, peptides, and recombinant proteins
Ni-NTA agarose	Qiagen	Cat# 30210
cOmplete Ultra EDTA-free	Sigma-Aldrich	Cat# 5892953001
CNBR-Activated Sepharose 4B	Cytiva Life Sciences	Cat# 17043001
ATP-Sepharose	This paper	N/A
Q5® High-Fidelity DNA Polymerase	New England Biolabs	Cat# M0491
Geldanamycin-FITC	Sigma-Aldrich	Cat# SML1277
XL888	BPS Bioscience	Cat# 27781-1
BX-2819	This paper	N/A
Tropane 1	This paper	N/A
HS292	This paper	N/A
SYBR™ Green I	Invitrogen	Cat# S7563
Geldanamycin	TCI Chemicals	Cat# G0334
Bortezomib	Selleck Chemical	Cat# S1013
Suc-Leu-Leu-Val-Tyr-AMC	Cayman Chemical	Cat# 10008119
Cycloheximide	Cayman Chemical	Cat# 14126
Epoxomicin	Cayman Chemical	Cat# 10007806
His6-PfHsp90-GFP	This paper	N/A
PfHsp90-His6	This paper	N/A
HsHsp90-His6	This paper	N/A
T7-HsHsp90-NBD-His6	This paper	N/A
T7-HsHsp90-NBD(E25D/K112R)-His6	This paper	N/A
T7-HsHsp90-NBD(Q23N)-His6	This paper	N/A
T7-HsHsp90-NBD(L122I/G125S/A126G)-His6	This paper	N/A
T7-PfHsp90-NBD-His6	This paper	N/A
Critical commercial assays
Pierce™ Coomassie Plus Bradford Assay	Thermo Fisher Scientific	Cat# 23236
Bright-Glo™ Luciferase Assay System	Promega	Cat# G6081
CellTiter-Fluor™ Cell Viability Assay	Promega	Cat# E2620
Hemacolor® Stain Set	Sigma-Aldrich	Cat# 65044-93
TMT10plex Isobaric Label Reagent Set	Thermo Fisher Scientific	Cat# 90406
Pierce High pH Reversed-Phase Peptide Fractionation Kit	Thermo Fisher Scientific	Cat# 84868
Q5® Site-Directed Mutagenesis Kit	New England Biolabs	Cat# E0554S
Deposited data
Thermal proteome profiling raw data	This paper	PXD040417
Small molecule screen results	This paper	20240206085657
Experimental models: Cell lines
Human: Huh7 cells	Sigma-Aldrich	Cat# 01042712-1VL
Experimental models: Organisms/strains
P. falciparum 3D7	BEI Resources Repository	NIAID, NIH, MRA-102
P. falciparum NF54 peg4-tdTomato	Laboratory of Photini Sinnis	N/A
P. berghei-Luc ANKA	University of Georgia SporoCore	N/A
Oligonucleotides
Primers for cloning and mutagenesis, see Table S3	This paper	N/A
Recombinant DNA
His6-PfHsp90-GFP (pET-45b)	This paper	N/A
PfHsp90-His6 (pET-21b)	Posfai et al.26	N/A
HsHsp90-His6 (pET-21b)	Posfai et al.26	N/A
T7-HsHsp90-NBD-His6 (pET-21a)	This paper	N/A
T7-HsHsp90-NBD(E25D/K112R)-His6 (pET-21a)	This paper	N/A
T7-HsHsp90-NBD(Q23N)-His6 (pET-21a)	This paper	N/A
T7-HsHsp90-NBD(L122I/G125S/A126G)-His6 (pET-21a)	This paper	N/A
T7-PfHsp90-NBD-His6 (pET-21a)	This paper	N/A
Software and algorithms
ChemDraw Professional 18.0	PerkinElmer	https://www.perkinelmer.com/category/chemdraw
GraphPad Prism 9	GraphPad software	graphpad.com
Schrödinger Release 2023-1: Maestro	Schrödinger, LLC	https://www.schrodinger.com/products/maestro
Clustal Omega	Madeira et al.94	https://www.ebi.ac.uk/Tools/msa/clustalo/
STRING database	Szklarczyk et al.49	https://string-db.org/
Cytoscape 3.10.1	Shannon et al.103	https://cytoscape.org/
PlasmoDB	The Plasmodium Genome Database Collaborative104	https://plasmodb.org/plasmo/app/
REVIGO	Supek at al.105	http://revigo.irb.hr/
Image Lab 6.1	Bio-Rad	https://www.bio-rad.com/en-us/product/image-lab-software?ID=KRE6P5E8Z
Other
0.2 μm PVDF membrane filter plate	Corning	Cat# 3505
Non-treated black 96 well plates	Corning	Cat# 3915
Non-binding black 384 well plates	Corning	Cat# 3575
Tissue culture treated white 384 well plates	Corning	Cat# 3570
NuPAGE 3–8% tris-acetate polyacrylamide gel	Thermo Fisher Scientific	Cat# EA0375BOX

Experimental Model and Study Participant Details

Cell Lines

HuH7 cells were attained through Sigma-Aldrich from the European Collection of Authenticated Cell Cultures (ECACC). Independent cell line authentication was not conducted after purchase. Cell morphology was regularly assessed by microscopy and no commonly misidentified cell lines are cultured in lab. Cell lines were routinely tested for mycoplasma contamination. The sex of the cell line is male. Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) with L-glutamine (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (HI-FBS) (v/v) (Sigma-Aldrich) and 1% antibiotic-antimycotic (Thermo Fisher Scientific) at 37 °C with 5% CO₂.

Parasite Lines

P. falciparum 3D7 were attained from BEI Resources, NIAID, NIH, MRA-102. Blood stage P. falciparum 3D7 were grown in 10.44 g/L RPMI 1640 (Gibco), 25 mM HEPES, pH 7.2, 0.37 mM hypoxanthine, 24 mM sodium bicarbonate, 0.5% (wt/vol) AlbuMAX II (Gibco), and 25 μg/mL gentamicin. Cultures were maintained at 37 °C with 3% O₂, 5% CO₂, and 92% N₂. Whole blood samples were attained from Gulf Coast Regional Blood Center, from which fresh erythrocytes were added to cultures for blood stage P. falciparum invasion. For liver stage cultures, P. berghei ANKA infected Anopheles stephensi mosquitoes were purchased from the University of Georgia SporoCore.

Bacterial Strains

Escherichia coli NEB5α (NEB) and BL21(DE3) (NEB) were used for plasmid and protein expression, respectively. Bacteria were propagated on Difco^™ Luria-Bertani (Miller) agar plates and cultured in Difco^™ Luria-Bertani (Miller) broth.

Method Details

Molecular cloning

The PfHsp90 with a C-terminal GFP tag and the wild-type nucleotide-binding domain (NBD) constructs were cloned via restriction digestion and ligation. For mutagenesis studies, sequence alignment was conducted with EMBL-EBI Clustal Omega⁹⁴ and HsHsp90 NBD variants were produced using a Q5 Site-Directed Mutagenesis Kit (NEB) according to manufacturer instruction. Constructs were transformed into competent BL21 E. coli (NEB) for expression. Primers were obtained from Eton Bioscience and all plasmids were confirmed by sequencing (Eton Bioscience).

Protein expression and purification

For protein expression, bacteria were grown in 1 L Luria-Bertani broth with 100 μg/mL of ampicillin at 37 °C with shaking at 250 rpm. When cultures reached an optical density at 600 nm of 0.6, the temperature was reduced to 18 °C and isopropyl-β-d-thiogalactopyranoside (IPTG) was added to induce protein expression. For PfHsp90-GFP and HsHsp90 NBD mutants, 500 μM IPTG was added and expression proceeded for 12–18 hours. For PfHsp90 and HsHsp90, 100 μM IPTG was added and expression proceeded for 4 hours. Bacteria were harvested by centrifugation at 4,300 g. PfHsp90-GFP pellets were stored at −80 °C until use.

For protein purification, pellets were resuspended in lysis buffer (50 mM KH₂PO₄, pH 8.0, 200 mM NaCl, 5 % glycerol, 2 mM imidazole, 1 mM benzamidine, 5 mM β-mercaptoethanol, and 1 cOmplete protease inhibitor tablet (Roche)) and lysed via sonication using a FB120 Sonic Dismembrator (Fisher Scientific). Lysate was centrifuged at 4,300 g for 1–3 hour and then incubated with Ni-nitrilotriacetic acid agarose (Qiagen) overnight at 4 °C with rotation. The resin was washed with buffer (50 mM KH₂PO₄, pH 8.0, 200 mM NaCl, 2 mM imidazole), and protein was eluted with increasing concentrations of imidazole (2–150 mM). Protein was then buffer exchanged (25 mM TEA, pH 7.5, 5 % glycerol, 1 mM DTT). Full-length, wild-type PfHsp90 and HsHsp90 were further applied onto a HQ/10 column (Applied Biosciences) and eluted with a NaCl gradient using a NGC liquid chromatography system (Bio-Rad). Protein purity and identity was assessed by SDS-PAGE followed by Coomassie staining and HisProbe-HRP (Thermo Fisher Scientific) western blotting.

Small molecule PfHsp90 binding screen

PfHsp90-GFP pellets were thawed, resuspended in lysis buffer (25 mM HEPES, pH 7.4, 0.1 % Triton-X100, 150 mM NaCl, 60 mM MgCl₂, 0.1 mM DTT, and 1 cOmplete protease inhibitor tablet (Roche)), lysed via sonication, and centrifuged at 4,300 g for 30 min. Supernatant was loaded into a glass column with ATP-Sepharose, prepared by γ-phosphate linkage of ATP to CNBr-activated Sepharose 4B (Cytiva), as previously described.⁴⁰ Briefly, in a glass column, dry CNBr-activated Sepharose 4B (Cytiva) was reconstituted in 1 mM HCl and then washed with 1 mM HCl followed by water. A linker solution was prepared by dissolving 0.48 g NaHCO₃, 1.68 g NaCl, and 1.8 g 1,10-diaminodecane in 57.5 mL water, 14.3 mL dioxane, and 1.8 mL ethanolamine. The linker solution was added to the resin and allowed to rotate at room temperature for 2 hours. After which, the resin was drained and then washed with 1 M NaCl followed by water. An ATP solution—consisting of 3.5 g ATP disodium and 6.0 g 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide dissolved in 2.6 mL 1-methylimidazole and 71.5 mL water—was added to the resin and allowed to rotate overnight. The activated resin was then drained, washed with 1 M NaCl followed by water, and stored at 4 °C as a 1:1 slurry in buffer (40 mM Na₂HPO₄, 10 mM NaH₂PO₄, pH 7.4, 1.5 mM NaN₃).

The amount of PfHsp90-GFP loaded per volume of ATP resin was normalized to the GFP fluorescence signal for each assay plate. PfHsp90-GFP was incubated with the resin for 1 hour at 4 °C with rotation. After which, the resin was washed three times with a high salt buffer (25 mM HEPES, pH 7.4, 1.0 M NaCl, 60 mM MgCl₂, 0.1 mM DTT) and three times with a low salt buffer (25 mM HEPES, pH 7.4, 150 mM NaCl, 60 mM MgCl₂, 0.1 mM DTT). The resin was then resuspended as a 1:1 (vol:vol) Sepharose to low salt buffer slurry, and 50 μL of this suspension was added to each well of a 96-well 0.2 μm PVDF filter plate (Corning) placed on top of a black 96-well non-treated catch plate (Corning). Compounds (50 μL) were added to wells of the filter plate at a final concentration of 500 μM and 5 % DMSO. Each plate included an ATP dilution series and 5 % DMSO as positive and negative controls, respectively. After compound additions, plates incubated for 30 minutes at room temperature and then were centrifuged at 1,100 rpm for 2 min. Fluorescence intensity (Ex: 485 nm; Em: 535 nm), corresponding to eluted PfHsp90-GFP, was measured in the catch plate using an EnVision (PerkinElmer) system. As a secondary assay, all treatments that resulted in a 2-fold increase in fluorescence intensity relative to the DMSO controls were confirmed with a western blot using an anti-GFP Alexa Fluor 488 antibody (Thermo Fisher Scientific).

Synthesis of Tropanes

Experimental conditions: (a) MeO(CH₂)₃NH₂, Pd₂(dba)₃, CsCO₃, Xantphos (81%); (b) NaOH, H₂O₂, 100 °C; (c) NaOH, 80 °C (78%, two steps); (d) (S4), iPr₂NEt, HATU; HCl (48%); (e) , iPr₂NEt

Unless otherwise noted, all materials were obtained from commercial suppliers and used without further purification. All reactions with air- and/or moisture-sensitive compounds were performed under an argon atmosphere in a flame-dried or oven-dried reaction flask, and reagents were added via syringe or cannula. Preparative chromatography was carried out using Sorbtech silica gel (60 Å porosity, 40–63 μm particle size) in fritted MPLC cartridges and eluted with Thomson Instrument SINGLE StEP pumps. Thin layer chromatography analyses were conducted with 200 μm precoated Sorbtech fluorescent TLC plates. Plates were visualized by UV light and by staining with a variety of stains such as acidic anisaldehyde, acidic vanillin, ceric ammonium nitrate or iodine vapor. LC/MS and LRMS data was obtained using an Agilent 1100 HPLC/MSD system equipped with a diode array detector running an acetonitrile/water gradient and 0.1% formic acid. High resolution mass spectral data were obtained using an Agilent 6540 QTOF mass spectrometer. Nuclear magnetic resonance spectrometry was run on a Varian Inova 500 MHz or a Varian Inova 400 MHz spectrometer, and chemical shifts are listed in ppm correlated to the solvent used as an internal standard. Representative NMR spectra are included in Data S1.

Synthesis of S2: A mixture of ethyl 3-bromo-4-cyanobenzoate (500.0 mg, 1.968 mmol), Xantphos (228.0 mg, 0.394 mmol), cesium carbonate (1.28 g, 3.93 mmol),3-methoxypropan-1-amine (233 mg, 2.61 mmol), and Pd₂(dba)₃ (180.0 mg, 0.197 mmol) in dioxane (5 mL) was stirred at 100 °C for 17 h. The cooled reaction mixture was partitioned between ethyl acetate (100 mL) and H₂O (50 mL), the organic phase was washed with brine (50 mL), dried over sodium sulfate, filtered and concentrated on a rotary evaporator. Column chromatography on silica gel (hexanes:ethyl acetate 9:1) afforded S2 as a brown oil (418 mg, 81%). ¹H NMR (400 MHz, CDCl₃) δ 7.41 (d, J = 7.8 Hz, 1H), 7.31 (d, J = 1.2 Hz, 1H), 7.23 – 7.29 (m, 1H), 4.36 (q, J = 7.0 Hz, 2H), 3.50 – 3.60 (m, 2H), 3.32 – 3.44 (m, 5H), 1.88 – 2.00 (m, 2H), 1.37 (t, J = 7.0 Hz, 3H) ppm. ¹³C NMR (126 MHz, CDCl₃) δ 165.8, 150.5, 135.5, 132.7, 117.2, 116.6, 111.2, 99.1, 71.2, 61.5, 58.8, 41.9, 28.7, 14.3 ppm. LRMS (ESI) m/z calc’d for C₁₄H₁₈N₂O₃ [M+H]⁺: 263.31, found 263.21.

Synthesis of S3: A mixture of S2 (1.00 g, 3.81 mmol), 50% w/w NaOH (0.20 mL, 3.81 mmol), and H₂O₂ (2.22 mL, 72.4 mmol), in EtOH (15 mL) and DMSO (6.58 mL) was added to a 30 mL microwave vial. The reaction was heated to 100 °C in a microwave reactor for 45 minutes. Upon cooling, the mixture was extracted with H₂O and ethyl acetate, and the organic phase was concentrated on a rotary evaporator to give a solid substance which was purified with flash column chromatography to afford a mixture of the desired ester and acid as a yellow solid. This mixture was then subjected to saponification conditions with excess aqueous NaOH in EtOH to yield the corresponding carboxylate as a light yellow solid (78%, LRMS (ESI) m/z calcd for C₁₂H₁₆N₂O₄ [M+H]⁺: 252.27, found 253.20), which was used without further purification.

A 15 mL vial was charged with a magnetic stir par, the crude acid made above (295.0 mg, 1.169 mmol), 8-Boc-3-endo-amino-8-azabicyclo[3.2.1]octane (S4, 233.3 mg, 1.031 mmol), DMF (5.0 mL) and diisopropylethylamine (1.00 mL, 5.73 mmol). With stirring, HATU (730 mg, 1.92 mmol) was added and the reaction was warmed to 50 °C for 3 h. The reaction was then diluted with H₂O and extracted with ethyl acetate (3x). The combined organic extracts were dried with MgSO₄, filtered and concentrated on a rotary evaporator. This material was suspended in a 1M solution of HCl in diethyl ether, which was allowed to stir at room temperature overnight. The mixture was then concentrated on a rotary evaporator to give S3 as a brown semisolid (371.6 mg, 48%) that was used without further purification.

Synthesis of Tropane 1: To a stirred solution of N-benzyl-6-chloronicotinamide (81.0 mg, 0.202 mmol) in acetonitrile (1.0 mL) was added S3 (146.0 mg, 0.405 mmol) and diisopropylethylamine (500 mL, 2.86 mmol). The reaction mixture was purged under argon, then heated to 80 °C in a sealed vessel for 48 h. Upon cooling to room temperature, the mixture was diluted with EtOAc and then washed with H₂O. The organic layer was washed with citric acid and brine, then dried over Na₂SO₄, filtered and concentrated on a rotary evaporator. Purification by flash column chromatography (hexanes: EtOAc, 1:1) afforded Tropane 1 (79%) as an off-white solid. ¹H NMR (500 MHz, DMSO) δ 8.77 (t, J = 6.1 Hz, 1H), 8.64 (d, J = 2.0 Hz, 1H), 8.20 (t, J = 5.4 Hz, 1H), 8.14 (d, J = 4.4 Hz, 1H), 7.97 (dd, J = 8.8, 2.5 Hz, 1H), 7.86 – 7.93 (m, 1H), 7.65 (d, J = 8.3 Hz, 1H), 7.18 – 7.34 (m, 6H), 6.97 (s, 1H), 6.86 (d, J = 7.8 Hz, 1H), 6.75 (d, J = 9.3 Hz, 1H), 4.56 (br. s., 2H), 4.44 (d, J = 5.9 Hz, 2H), 3.82 (br. s., 1H), 3.40 (t, J = 6.1 Hz, 2H), 3.17 – 3.25 (m, 5H), 2.19 (d, J = 7.3 Hz, 2H), 2.03 – 2.10 (m, 2H), 1.96 – 2.02 (m, 2H), 1.90 (d, J = 14.2 Hz, 2H), 1.80 (t, J = 6.4 Hz, 2H) ppm. ¹³C NMR (126 MHz, DMSO) δ 171.5, 167.2, 165.5, 158.0, 149.9, 149.2, 140.4, 139.2, 136.8, 129.5, 128.7, 127.6, 127.1, 118.2, 116.1, 112.9, 110.3, 107.5, 70.1, 58.4, 52.4, 43.0, 42.8, 40.6, 32.4, 29.1, 27.9 ppm. HRMS m/z: [M+H]⁺ calc’d for C₃₂H₃₈N₆O₄ 571.3033; Found 571.3042.

Synthesis of Tropane 2: Using conditions similar to those described above, Tropane 2 (55%) was obtained as a light yellow solid. ¹H NMR (500 MHz, DMSO) δ 8.62 (d, J = 2.5 Hz, 1H), 8.21 (t, J = 5.4 Hz, 1H), 8.16 (d, J = 3.9 Hz, 1H), 7.95 (dd, J = 8.8, 2.0 Hz, 1H), 7.87 – 7.93 (m, 1H), 7.73 (br. s., 1H), 7.67 (d, J = 8.3 Hz, 1H), 7.25 – 7.30 (m, 1H), 7.09 (br. s., 1H), 6.99 (s, 1H), 6.86 – 6.90 (m, 1H), 6.75 (d, J = 8.8 Hz, 1H), 4.57 (br. s., 2H), 3.80 – 3.86 (m, 1H), 3.42 (t, J = 6.4 Hz, 3H), 3.25 (s, 3H), 3.19 – 3.24 (m, 2H), 2.17 – 2.23 (m, 2H), 2.08 (td, J = 14.2, 5.1 Hz, 3H), 1.98 – 2.04 (m, 2H), 1.91 (d, J = 14.2 Hz, 3H), 1.82 (quin, J = 6.5 Hz, 2H) ppm. ¹³C NMR (126 MHz, DMSO) δ 171.5, 167.3, 167.2, 158.1, 150.0, 149.5, 139.2, 137.2, 129.5, 118.1, 116.1, 112.9, 110.3, 107.5, 70.1, 58.4, 52.4, 43.1, 39.8, 32.4, 29.1, 27.9 ppm. HRMS m/z: [M+H]⁺ calc’d for C₂₅H₃₂N₆O₄ 481.2563; found 481.2571.

Synthesis of Tropane 3: Using conditions similar to those described above, Tropane 3 (55%) was obtained as a white solid. ¹H NMR (500 MHz, DMSO) δ 8.79 (d, J = 2.5 Hz, 1H), 8.35 (t, J = 5.4 Hz, 1H), 8.31 (d, J = 4.0 Hz, 1H), 8.06 (dd, J = 9.1, 2.2 Hz, 2H), 7.80 (d, J = 8.3 Hz, 1H), 7.36 – 7.47 (m, 1H), 7.12 (s, 1H), 7.02 (d, J = 7.8 Hz, 1H), 6.91 (d, J = 8.8 Hz, 1H), 4.66 – 4.82 (m, 2H), 4.39 (q, J = 7.3 Hz, 2H), 3.98 (br. s., 1H), 3.55 (t, J = 6.1 Hz, 2H), 3.32 – 3.40 (m, 5H), 2.35 (d, J = 7.3 Hz, 2H), 2.13 – 2.25 (m, 4H), 2.08 (d, J = 14.7 Hz, 2H), 1.91 – 2.00 (m, 2H), 1.43 (t, J = 7.1 Hz, 3H) ppm. ¹³C NMR (126 MHz, DMSO) δ 171.5, 167.2, 165.6, 158.5, 151.5, 149.9, 139.2, 138.3, 129.4, 116.1, 113.8, 112.9, 110.3, 107.6, 70.1, 60.3, 58.4, 52.5, 43.0, 39.5, 32.8, 29.1, 27.7, 14.8 ppm. HRMS m/z: [M+H]⁺ calc’d for C₂₇H₃₅N₅O₅ 510.2716; found 510.2727.

Synthesis of Tropane 4: Using conditions similar to those described above, Tropane 4 (26%) was obtained as an off-white solid. ¹H NMR (500 MHz, DMSO) δ 8.84 (d, J = 2.0 Hz, 1H), 8.09 (d, J = 8.3 Hz, 1H), 8.01 (d, J = 8.3 Hz, 1H), 7.18 (s, 1H), 6.76 (d, J = 8.3 Hz, 1H), 6.67 (d, J = 6.4 Hz, 1H), 6.59 (d J = 7.3 Hz, 1H), 4.74 (br. s., 2H), 4.36 (q, J = 6.9 Hz, 2H), 4.26 (q, J = 6.0 Hz, 1H), 3.48 – 3.57 (m, 3H), 3.31 – 3.43 (m, 5H), 2.26 – 2.42 (m, 4H), 2.09 – 2.20 (m, 2H), 1.88 – 2.03 (m, 4H), 1.38 (t, J = 7.1 Hz, 3H) ppm.

Synthesis of BX-2819 and HS292

Reagents were obtained from commercial sources and used without further purification. Proton NMR spectra were obtained on Varian 400 and 500 MHz spectrometers and Bruker 500 and 700 MHz spectrometers. LC/MS were obtained on an Agilent ion-trap LC/MS system. Prep purification were performed on an ISCO flash apparatus. Representative NMR spectra are included in Data S1.

Synthesis of BX-2819:⁴³

First step:⁹⁵

A solution of 4-isopropylbenzene-1.3-diol (8 g, 52.6 mmol) was dissolved in acetonitrile (400 mL) and treated with DBU (24 g, 23.54 mL, 157.7 mmol). The mixture was purged with a stream of CO₂ from a nearby flask filled with dry ice for an hour and left to stir under CO₂ overnight. The next day, the mixture was treated with 1 N HCl (200 mL) and extracted with ethyl acetate (300 mL) repeatedly until no product was obtained (3^rd was clear and discarded). The 2 combined organic layers were washed with saturated brine, dried (MgSO₄), filtered and concentrated to give a light brown solid (13 g). The solid was partitioned between ethyl acetate (200 mL) and 1N HCl (100 mL). The organic phase was washed with brine (100 mL), dried (MgSO₄) and concentrated to give 2,4-dihydroxy-5-isopropylbenzoic acid (10.02 g, 97%) as a light orange solid. This material was used without further purification.

A solution of 2,4-dihydroxy-5-isopropylbenzoic acid (10.0 g, 51 mmol) in methanol (200 mL) was saturated with HCl gas from a tank. The mixture was then heated to reflux until the reaction was complete by TLC (20% EtOAc in hexanes) after approximately 36 h. The mixture was concentrated to a solid, then adsorbed onto silica gel and chromatographed (20% ethyl acetate in hexane) to give methyl 2,4-dihydroxy-5-isopropylbenzoate (8.48 g, 79%) as a cream colored product.

Methyl 2,4-dihydroxy-5-isopropylbenzoate (3 g, 14.3 mmol) was treated with hydrazine hydrate (10 mL) and methanol (10 mL) and stirred at RT overnight. The next day, the reaction mixture was concentrated and chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give 2,4-dihydroxy-5-isopropylbenzohydrazide (1.97 g, 65.6%) as a tan powder. ¹H-NMR (dmso-d₆) δ 12.55 (s, 1H), 9.93 (s, 1H), 9.78 (s, 1H), 7.55 (s, 1H), 6.26 (s, 1H) 4.48 (br s, 2H), 3.07 (sept, J = 7 Hz, 1H), 1.13 (d, J = 7 Hz, 6H). Plus a little MeOH at 3.17 (d) and 4.08 (q) and water at 3.32 (s).

Isothiocyanate:⁹⁶

A mixture of 4-(N-BOCaminomethyl)aniline (1.5 g, 6.75 mmol) and calcium carbonate (700 mg, 6.95 mmol) were partitioned between water (6 mL) and methylene chloride (50 mL) and stirred rapidly. Thiophosgene (620 μL, 930 mg, 8.1 mmol) was added dropwise and the mixture stirred for 15 m. The mixture was partitioned between more water (25 mL) and methylene chloride (50 mL) and the organic layer (which remained cloudy) was drained off, dried (MgSO₄), filtered and concentrated to give tert-butyl (4-isothiocyanatobenzyl)carbamate (1.5 g, 84%) as a light yellow powder.

2,4-dihydroxy-5-isopropylbenzohydrazide (1 g, 4.76 mmol) and tert-butyl (4-isothiocyanatobenzyl)carbamate (1.32 g, 5 mmol) were dissolved in ethanol (15 mL) and stirred at RT overnight. The reaction mixture was concentrated then chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give compound 1 (1.98 g, 87%) as a white powder. ¹H-NMR (dmso-d₆) δ 10.47, br s, 1H), 10.11 (s, 1H), 9.83 (s, 1H), 9.64 (br s, 1H), 7.65 (s, 1H), 7.33–7.41 (m, J = 6 Hz, 3H), 7.17 (d, J = 8 Hz, 1H), 6.33 (s, 1H), 4.09 (d, J = 6 Hz, 2H), 3.10 (hept, J = 7 Hz, 1H), 1.39 (s, 9H), 1.16 (d, J = 7 Hz, 6H).

Procedure from By Ying, Weiwen et al PCT Int. Appl., 2008103353, 28 Aug 2008 see page 146.

Compound 1 (1.98 g, 4.17 mmol) was treated with 0.5 N sodium hydroxide (20 mL), flushed with nitrogen and heated to 110 °C. After about an hour, TLC (19/1 : CH₂Cl₂/MeOH with Acetic Acid 3 drops) showed formation cyclization product and product with loss of the BOC. The reaction mixture was concentrated to a solid and treated with TFA (6 mL). After one day, the reaction mixture was concentrated and chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give a solid (NMR taken here). The solid was treated with 3M HCl in methanol and concentrated thrice from methanol to give the HCl salt of compound 2 (1.31 g, 80%) as a white solid. ¹H-NMR (dmso-d₆) δ 9.64 (v br s, 3H), 8.25 (s, 1H, formate), 7.39 (d, J = 8 Hz, 2H), 7.22 (d, J = 8 Hz, 2H), 6.86 (s, 1H), 6.23 (s, 1H), 3.87 (s, 2H), 2.96 (hept, J = 7 Hz, 1H), 0.99 (d, J = 7 Hz, 6H).

Amine 2 (100 mg, 254.52 μmol) was slurried in methylene chloride (3 mL) and treated with 1N NaOH (1 mL). Everything dissolved. Ethyl chloroformate (28 mg, 25 μL, 254 μmol) was then added in a shot and the reaction stirred overnight. The next day the reaction mixture was concentrated then chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give clean product. The product fractions were concentrated to give a slurry of white powder which was filtered off and air-dried to give compound 3 (46 mg, 42%) as a white powder. ¹H-NMR (dmso-d₆) δ 13.83 (br s,1H), 9.56 (s, 1H), 9.49 (br s, 1H), 7.66 (t, J = 6 Hz, 1H), 7.24 (d, J = 8 Hz, 2H), 7.17 (d, J = 8 Hz, 2H), 6.83 (s, 1H), 6.23 (s, 1H), 4.17 (d, J = 6 Hz, 2H), 4.00 (q, J = 7 Hz, 2H), 2.96 (hep, J = 7 Hz, 1H), 1.10 (t, J = 7 Hz, 3H), 0.97 (d, J = 7 Hz, 6H). LC/MS gave a single peak with m/z = 429.3 for [M+H]⁺.

Synthesis of HS292:

Following US20110098290A1 on page 42

A mixture of 2,4-dihydroxy-5-isopropylbenzoate (1 g, 5 mmol) and potassium carbonate (830 mg, 6 mmol) in acetonitrile (25 mL) was treated with 4-fluorobenzyl bromide (685 μL, 1.04 g, 5.5 mmol) and stirred at RT for a week. The reaction mixture was concentrated and then slurried in water (100 mL) and stirred for a few hours. The solid was then filtered off and air-dried overnight to give methyl 4-((4-fluorobenzyl)oxy)-2-hydroxy-5-isopropylbenzoate (1.61 g, 101%) as a white solid. ¹H-NMR (dmso-d₆) δ 10.69 (s, 1H), 7.56 (s, 1H), 7.51 (m, 2H), 7.25 (t, J = 8 Hz, 2H), 6.64 (s, 1H), 5.17 (s, 2H), 3.87 (s, 3H), 3.14 (hept, J = 7 Hz, 1H), 1.14 (d, J = 7 Hz, 6H).

A mixture of methyl 4-((4-fluorobenzyl)oxy)-2-hydroxy-5-isopropylbenzoate (1 g, 3.14 mmol) and potassium carbonate (521 mg, 3.77 mmol) in acetonitrile (20 mL) was heated up to reflux and treated with iodomethane (320 μL, 711 mg, 5 mmol) in two batches over 8 h. TLC (20% EtOAc in hexane) showed a lower spot. The mixture was concentrated, treated with 1 N HCl (40 mL) and extracted with ethyl acetate (2 × 40 mL). The combined organic layers were washed with brine (25 mL), dried (MgSO₄), filtered and concentrated. The residue was dissolved in hot heptane (40 mL) stirred overnight. The next day, the crystalline solid was collected, washed with heptane and air-dried to give compound 4 (490 mg, 47%) as an off-white solid. ¹H-NMR (dmso-d₆) δ 7.55 (s, 1H), 7.53 (dd, J = 5, 8 Hz), 7.26 (t, J = 8 Hz, 2H), 6.79 (s, 1H), 5.23 (s, 2H), 3.83 (s, 3H), 3.73 (s, 3H), 3.16 (hept, J = 7 Hz, 1H), 1.13 (d, J = 7 Hz, 6H).

Compound 4 (500 mg, 1.5 mmol) was dissolved in ethanol and ethyl acetate (1:1, 40 mL) and treated with 10% palladium on carbon (20 mg). The mixture was stirred rapidly and vacuum purged with hydrogen gas 3 times and left to stir under a hydrogen balloon atmosphere for 1 day. No reaction was seen. The catalyst was filtered off and the reaction repeated to give product. The catalyst was filtered off and the mixture concentrated, then chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give compound 5 (153 mg, 45%) as a white powder. ¹H-NMR (dmso-d₆) δ 10.18 (br s, 1H), 7.51 (s, 1H), 6.51 (s, 1H), 3.72 (s, 3H), 3.71 (s, 3H), 3.10 (sept, J = 7 Hz, 1H), 1.13 (d, J = 7 Hz, 6H).

The synthesis of compound 5 is also described in patent US20110098290A1.

Compound 5 (150 mg, 668 μmol) was treated with hydrazine hydrate (1 mL) and methanol (3 mL) and heated to reflux then left to cool with stirring overnight. The reaction mixture was concentrated to a foamy glass which was used in the next step.

Compound 6 (150 mg, 669 μmol) and tert-butyl (4-isothiocyanatobenzyl)carbamate (185 mg, 702 μmol) were dissolved in ethanol (2 mL) and stirred at RT for 4 h. The mixture was concentrated, then chromatographed (C-18, 0 to 100% MeOH in 0.2% formic acid) to give compound 7 (203 mg, 62%) as a white powder. ¹H-NMR (dmso-d₆) δ 10.16 (br s, 1H), 9.4–9.9 (v br s, 2H), 7.73 (br s, 1H), 7.42 (br s or d, 2H), 7.37 (t, J= 6 Hz, 1H), 7.17 (d, J = 8 Hz, 2H), 6.55 (s, 1H), 4.09 (d, J = 6 Hz, 2H), 3.87 (s, 3H), 3.13 (sept, J = 7 Hz, 1H), 1.39 (s, 9H), 1.15 (d, J= 7 Hz, 6H).

Compound 7 (134 mg, 274 μmol) was added to a solution of sodium hydroxide (450 μL of 1N and 500 μL of water), flushed with nitrogen and heated to 110 °C for 1 h. The reaction mixture was allowed to cool, treated with 1N HCl (3 ml) and extracted with ethyl acetate (2 × 10 mL). The combined organic layers were washed with brine (2 mL), dried (MgSO₄), filtered and concentrated to give compound 8 (90 mg, 69%) as a white solid. This material was taken directly to the next step.

Compound 8 (90 mg, 191 μmol) was slurried in methylene chloride and treated with 4M HCl in dioxane (200 μL), then sonicated until starting material was consumed by TLC analysis. The reaction mixture was then concentrated then concentrated again from ethanol. The residue was dissolved in ethanol (3 mL) and treated with Hunig’s base (100 μL), followed by dropwise treatment with ethyl chloroformate (100 μL). The reaction was concentrated then chromatographed (C18, 0 to 100% 0.2% formic acid) to give compound 9 also known as HS292 (16 mg, 20%) as a white powder. ¹H-NMR (dmso-d₆) δ 13.9 (s, 1H), 9.81 (s, 1H), 7,64 (t, J = 6 Hz, 1H), 7.22 (d, J = 8Hz, 2H, 7.12 (d, J = 8 Hz, 2H), 7.08 (s, 1H), 6.27 (s, 1H), 4.15 (d, J = 6 Hz, 2H), 3.99 (q, J = 7 Hz, 2H), 3.31 (s, 3H), 3.05 (hept, J = 7 Hz, 1H), 1.15 (t, J = 7 Hz, 3H), 1.06 (d, J = 7 Hz, 6H). LC/MS gave a single peak with m/z = 443.2 for [M+H]⁺.

Protein binding site modeling

The Schrodinger Molecular Modeling Suite (Schrödinger Release 2023–1: Maestro, Schrödinger, LLC, New York, NY, 2021) was used to analyze the structures of HsHsp90 bound to BX-2819 (PDB: 3HHU) and XL888 (4AWO). Proteins were processed utilizing the default Schrodinger Maestro Protein Preparation Workflow. After which, the Binding Site Alignment tool was used to overlay the structures. Visualizations were generated in Schrodinger Maestro.

Competitive binding assay

Small molecule binding to PfHsp90 and HsHsp90 was assessed using a previously reported competition-based fluorescence assay.²⁶ Briefly, competition was assessed against FITC-GA (Sigma Aldrich), a well characterized PfHsp90 and HsHsp90 inhibitor (Geldanamycin; GA) that binds at the N-terminal ATP site, linked to a FITC fluorophore. Differences in fluorescence polarization are measurable between the bound or unbound tracer. Competition experiments were conducted in black 384-well, non-binding plates (Corning) with 1.0 nM FITC-GA in assay buffer (20 mM HEPES, pH 7.3, 50 mM KCl, 5 mM MgCl₂, 20 mM Na₂MoO₄, 2 mM DTT, 0.1 mg/ml bovine growth globulin, and 0.01% NP-40). PfHsp90 or HsHsp90 were added at concentrations of 60 nM or 32 nM, respectively, based on previous binding saturation determinations for the tracer to each protein.²⁶ For the HsHsp90 NBD variants, the assay was conducted with 25 nM of each protein based on their binding to FITC-GA. XL888 (repurchased from BPS Bioscience) or BX-2819 (synthesized in-house) were added between 0–10 μM to determine the dose-response for FITC-GA displacement. Experiments were completed with at least 4 independent determinations. For HS292, the compound was tested at 100 μM with 5 independent determinations. The final DMSO concentration was 1% in every well. Each assay included protein without inhibitor as a bound tracer control, and wells without protein as an unbound tracer control. Assay plates were incubated overnight in the dark at 4 °C to promote binding equilibrium between the protein, the tracer, and competitive inhibitors. Fluorescence polarization was then measured using the Envision (PerkinElmer) system (Ex: 480; Em: 535). The percent FITC-GA bound to protein in the presence of competitive inhibitors was calculated by normalizing data to the bound and unbound tracer controls. Data were fit to a nonlinear curve (GraphPad Prism) to obtain the point at which half the tracer is displaced (IC₅₀) from which the apparent inhibitory constant $\left({\mathrm{K}}_{\mathrm{i}}\right)$ for each replicate were calculated and averaged based on ${\mathrm{K}}_{\mathrm{i}}={\mathrm{I}\mathrm{C}}_{50}/(1+[\mathrm{L}]/{\mathrm{K}}_{\mathrm{d}}$, where $\mathrm{L}$ refers to the concentration of FITC-GA and ${\mathrm{K}}_{\mathrm{d}}$ refer to the determined affinity of FITC-GA to the protein–previously reported as 42 nM and 23 nM for full-length PfHsp90 and HsHsp90, respectively.²⁶

Blood stage Plasmodium culture

In vitro culture of P. falciparum 3D7 (BEI Resources) was completed by supplementing parasites with fresh human erythrocytes (Gulf Coast Regional Blood Center) into complete media (10.44 g/L RPMI 1640 (Gibco), 25 mM HEPES, pH 7.2, 0.37 mM hypoxanthine, 24 mM sodium bicarbonate, 0.5% (wt/vol) AlbuMAX II (Gibco), and 25 μg/mL gentamicin). Cultures were maintained at 37 °C with 3% O2, 5% CO2, and 92% N2. Every 48 hours, parasites were synchronized at the ring stage using 5 % (wt/vol) D-sorbitol.

For proteomic studies, synchronized P. falciparum 3D7 were cultured to 10–15 % parasitemia at 1 % hematocrit. When parasites reached the trophozoite stage (22 – 38 hours post invasion), infected red blood cells were pelleted by centrifugation and lysed by addition of 0.03% (wt/vol) saponin in PBS (Gibco). Released parasites were pelleted by centrifugation and washed 4 times with PBS (Gibco) to remove residual red blood cell proteins and lysis debris. Collected parasites were then lysed via sonication after resuspension in 4 volumes of PBS (Gibco) (yielding ≥ 2 mg/mL) with protease inhibitors (1 mM AEBSF, 10 μM pepstatin A, 20 μM leupeptin, 15 μM E-64, and 500 μM bestatin). Lysate was clarified by centrifugation and total protein was quantified using the Pierce Coomassie Plus (Bradford) Assay (Thermo Fisher Scientific). Freshly prepared lysate was immediately processed for thermal proteome profiling experiments.

Blood stage Plasmodium inhibition assays

Asexual blood stage parasite inhibition assays were completed as described previously.⁹⁷ Briefly, ring-stage P. falciparum 3D7 cultures at 2 % parasitemia and 1 % hematocrit were added to wells of UV-sterilized, black 96-well non-treated plates (Corning). Compounds were then administered from 0–50 μM using a D300e Digital Dispenser (Hewlett-Packard). Each well contained 0.5 % DMSO in a final volume of 200 μL, and every plate included DMSO (0.5 %) as a negative control and quinacrine (1 μM, Sigma-Aldrich) as a positive control. Samples were measured in 2–3 replicates per experiment, across 2–4 independent experiments. After compound addition, assay plates were incubated at 37 °C with 3% O₂, 5% CO₂, and 92% N₂ for 72 hours. After incubation, 40 μL lysis buffer (20 mM Tris, pH 7.5, 5 mM EDTA, 0.16 % saponin (wt/vol), and 1.6 % Triton X-100 (vol/vol)) supplemented with SYBR Green I (Invitrogen) at a 10X concentration was added to each well. After staining for 24 hours at room temperature in the dark, parasite DNA was assessed by measuring SYBR Green I fluorescence using an EnVision (PerkinElmer) system (Ex: 485; Em: 535). Relative parasite load was calculated by normalizing the fluorescence signal from experimental treatments to the positive and negative controls. Data were fit to a nonlinear curve (GraphPad Prism) to obtain half-maximal effective concentrations (EC₅₀).

To assess their effects on blood stage progression, P. falciparum 3D7 cultures were treated with compounds at ~12 hours post invasion at the ring stage. Inhibitors were administered below their EC₅₀s at 500 nM for BX-2819 or 200 nM for XL888. DMSO (0.5%) and 300 nM geldanamycin were included as negative and positive controls, respectively. After 24 hours incubating at 37 °C with 3% O₂, 5% CO₂, and 92% N₂, the number of ring, trophozoite, or schizont parasites in each treatment were counted from blood smears stained with a Hemacolor kit (Sigma-Aldrich). In total, >300 parasites were counted for each treatment, divided approximately equally between 2 biological replicates.

Liver stage Plasmodium inhibition assays

For liver stage assays, HuH7 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) with L-glutamine (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (HI-FBS) (v/v) (Sigma-Aldrich) and 1% antibiotic-antimycotic (Thermo Fisher Scientific) at 37 °C with 5% CO₂. P. berghei ANKA sporozoites expressing luciferase were isolated from freshly dissected salivary glands of infected mosquitoes (University of Georgia SporoCore). Dose response curves were generated for each compound by assessing P. berghei parasite load in hepatocytes as previously described.⁴¹ Briefly, HuH7 (5,000 cells/well) were seeded into 384-well white microplates (Corning). After 24 hours, compounds (0–100 μM) were added using a D300e Digital Dispenser (Hewlett-Packard) before infection with P. berghei ANKA sporozoites (4,000 spz/well). DMSO (1%) was the negative control. All samples were evaluated in triplicate with 2 or more independent determinations and had final DMSO concentration of 1%. After 44 hours post-infection, HuH7 cell viability and parasite load were assessed using CellTiter-Fluor (Promega) and Bright-Glo (Promega) reagents, respectively, according to manufacturer’s protocols. Relative fluorescence and luminescence signal was measured using the EnVision plate reader (PerkinElmer). The signal intensity of each well was normalized to the negative control to assess relative viability. EC₅₀ analysis was performed with GraphPad Prism.

Plasmodium gametocyte inhibition assays

P. falciparum gametocyte inhibition assays were performed as previously described⁹⁸ using the NF54 peg4-tdTomato reporter line.⁹⁹ Synchronously induced gametocytes were obtained following a method adapted from Fivelman et al.¹⁰⁰ In brief, parasites were allowed to commit to the sexual differentiation by growing at 6–8% parasitemia at 3% hematocrit. At the end of this commitment cycle, schizonts were purified using a Percoll-sorbitol gradient and then combined with fresh erythrocytes to allow for reinvasion. The newly invaded ring parasites (Day +1 of the gametocytogenesis) were cultivated at 3.5% parasitemia and 1% hematocrit with 50 mM N-acetylglucosamine to eliminate remaining asexual parasites. To observe the effect of inhibitors on early gametocytes, parasites were culture for 6 days with 1 μM and 10 μM of geldanamycin (GA) or DMSO (control). At day +6, gametocytemia was assessed by flow cytometry (Cytek DxP11) based on the Hoechst 33342 DNA staining (375 nm laser, 450/50 emission filter) and the tdTomato fluorescent signal (561 nm laser, 590/20 emission filter). Mean tdTomato+ signal was normalized to the DMSO control and the overall minimum to determine the early gametocyte relative viability. To observe the effect of inhibitors on late gametocytes, parasites were allowed to develop for 10 days and then cultured for 2 days with 1 μM and 10 μM of GA or DMSO. At day +12, gametocytemia was assessed by flow cytometry (Cytek DxP11) based on the Hoechst 33342 DNA staining (375 nm laser, 450/50 emission filter), DilC1(5) mitochondrial potential signal staining (640 nm laser, 670/30 emission filter) and the tdTomato fluorescent signal (561 nm laser, 590/20 emission filter). Mean DilC1(5)+ tdTomato+ signal was normalized to DMSO control and the overall minimum to determine the late gametocyte relative viability. Technical duplicates were performed within 3 biological replicates for each experiment. The effect of SGI-1027 at 1 μM and 10 μM was used as a positive control in both early and late gametocyte inhibition assays.⁹⁸

TPP experiments

The TPP experiments were conducted in a similar manner to those in our previous publication.⁸⁰ P. falciparum lysate (2 mg/ml) was divided into three ~600 μL portions. Two of the portions were treated with 500 μM of the corresponding drug (BX-2819 or HS292), and the last was treated with the same concentration of DMSO (5%) corresponding to the drug treatment groups. A 50 μl aliquot of each solution was distributed into 10 tubes, which were heated for 3 min at 37, 41, 45, 49, 53, 57, 61, 65, 69, or 73 degrees Celsius. The samples were cooled at room temperature for 3 min then placed on ice for 30 min. The protein samples from each temperature treatment were precipitated using a Beckman Coulter ultracentrifuge at 100,000 g for 20 min at 4 °C to pellet the unfolded and aggregated protein in each sample. The supernatant was subjected to a standard filter-aided sample preparation (FASP) protocol.¹⁰¹ Briefly, the samples were transferred into Amicon Ultra-0.5 ml 10k filters and washed 3 times with 8 M Urea in 100 mM Tris (pH 8.0). The samples were reduced using 5 mM TCEP and alkylated using 20 mM MMTS. The samples were washed with a 100 mM TEAB solution and then 120 μl of 0.008 mg/ml trypsin in 100 mM TEAB was added into the filters. The protein:enzyme ratio was ~1:100 and the trypsinization was allowed to proceed overnight at 37 degrees. The resulting peptides were labeled by TMT 10-plex reagent (Thermo Fisher Scientific) according to the manufacturer’s protocol before they were eluted from the filters using 500 mM NaCl. The samples from the same drug or DMSO treatment were pooled and the solvent evaporated dried using a speedvac. The pooled samples from each drug treatment were reconstituted in 2% TFA and fractionated using the Pierce High pH Reversed-Phase Peptide Fractionation Kit (Thermo Fisher Scientific) according to the manufacturer’s directions.

For MS analysis, the fractionated TMT-labelled peptide samples that included 8 fractions for each drug were reconstituted in 2% acetonitrile, 1% TFA with an estimated concentration of ~1 mg/ml. A 1 μl aliquot of the sample was injected for MS analysis. The samples were analyzed using a Thermo Orbitrap Exploris 480 mass spectrometer coupled with a Thermo Easy nLC-1200. The samples were analyzed using a PepMap PSLC C18 analytical column (2 μm, 100 Å, 75 μm × 25 cm) coupled with an Acclaim PepMap 100 loading column (NanoViper 2Pk, C18, 2 μm, 100 Å, 75 μm × 2 cm). The samples were analyzed using a gradient going from 3.2% to 28% ACN and 0.1% formic acid in 90 min, 28% to 32% ACN and 0.1% formic acid in 5 min, and 32% to 80% ACN and 0.1% formic acid in 5 min before the column was washed with 80% ACN and 0.1% formic acid for 5 min. MS1 spectra were acquired with 120,000 resolution in the 375–1500 m/z range, and 300% normalized AGC target. The acquisition window between two mass spectra was 2.5 s and the isolation window was set as 1.2 m/z for selecting precursor ions for DDA and top 20 precursors ions according to their intensities were selected for MS2 acquisition. The high-energy collision-induced dissociation (HCD) was performed with 36% HCD normalized collision energy, and the product ion MS2 spectra were collected at a resolution of 45000, 300% normalized AGC target, and 105 ms of maximum ion injection time.

The raw data files generated in the mass spectral analysis were searched against the Plasmodium falciparum isolate 3D7 proteome database from UniprotKB (UP000001450). The database searching was performed using Thermo Proteome Discoverer 2.3 and the SEQUEST node. Oxidation (+ 15.995 Da on M residues) and deamidation (+ 0.984 Da on N and Q residues) were considered as dynamic modifications. Acetylation (+ 42.011 Da), Met-loss (−131.040 Da), and Met-loss+acetyl (−89.030 Da) were all considered as dynamic modifications at the protein N-terminus of peptides. Methylthiolation (+ 45.988 Da at C residues) and TMT 10-plex (+ 229.263 Da and K residues and the N-termini) were considered static modifications. The FDR was estimated using the percolator node built in Thermo Proteome Discoverer 2.3, and only proteins with FDR < 0.05 were considered in the later data analysis. The MS data generate in this work have been uploaded to the ProteomeXchange Consortium via the PRIDE¹⁰² partner repository with the dataset identifier PXD040417.

The reporter ion intensities of each protein were normalized to the corresponding highest intensities to rescale the intensities from 0 to 1. The normalized intensities from each thermal condition were summed and further normalized to the maximum intensity detected in the experiment for plotting the “supercurve.” The supercurve was fitted with the following equation:

$$\mathrm{I}=1-\left(\right(1-\mathrm{p})/(1+\mathrm{e}-(\mathrm{a}/\mathrm{T}-\mathrm{b}))+\mathrm{p}.$$

where $I$ represents the normalized intensity of a protein, p represents post-translational baseline, $T$ represents temperature, $a$ and $b$ represent fitting parameters. Normalization factors were calculated by dividing the fitted intensities by the measured intensities on the supercurve for each temperature in the curve. The intensities for each protein were then multiplied by the appropriate normalization factor at each temperature. The thermal denaturation data of each protein was individually fit to the same equation using a previously reported R program (4) to generate an R² value and the melting temperatures for each protein. As a quality control, fitted data with R² < 0.8 were excluded.

Ultimately, prssoteins that had thermal melt coverage in both treatments comparing BX-2819 to DMSO and HS292 to DMSO were analyzed for PfHsp90-mediated stability perturbations of greater magnitude than 2.0 °C. This cutoff corresponds to a Z-score of ~1. To account for off-target effects not relating to PfHsp90 inhibition, data was filtered to only include proteins that did not have an altered thermal stability (−2.0 ≤ ΔTm ≤ 2.0 °C) between the DMSO and inactive analog (HS292) treatment. The proteins that exhibited shifts with inhibitor treatment but not with inactive analog treatment were ultimately selected as hits.

Analysis of TPP hit proteins

Hit proteins selected from TPP were analyzed using the STRING database to construct a full STRING network. Enriched local network clusters were calculated assuming the whole genome as statistical background. For assigning protein interactions, a 0.25 minimum required interaction score cut-off was assigned to generate hypotheses for potential PfHsp90 interactors represented in the dataset.⁴⁹ From the STRING analysis, a putative interaction network was constructed in Cytoscape with individual nodes representing distinct proteins.¹⁰³ Both cytosolic and mitochondrial-targeted paralogs of P. falciparum Hsp90 (Q8IC05 and Q8III6), which were both hit proteins in the TPP analysis, were considered together as a single node (PfHsp90) in the interaction network. Edges between nodes denote putative interactions between proteins, which are scaled by confidence. Only proteins that were predicted in the STRING analysis to make a direct interaction with PfHsp90, and those further connected by a single edge to a direct PfHsp90 interactor, are displayed in the network. Additionally, gene ontology analysis for biological process was conducted utilizing the PlasmoDB gene ontology tool¹⁰⁴ with a cutoff of P < 0.01, which were refined and plotted in REVIGO,¹⁰⁵ from which the figure was then visualized in Cytoscape.

Plate-based proteasome activity assays

Parasite lysate was prepared by reconstituting trophozoite stage P. falciparum 3D7 parasite pellets, harvested as described above, in proteasome activity lysis buffer (50 mM TRIS-HCl, pH 7.5, 2 mM DTT, 5 mM MgCl₂, 5 % glycerol, 2 mM ATP disodium, and 0.05% digitonin) followed by sonication. Lysate was clarified by centrifugation at 20,000 g for 10 minutes, and the supernatant was diluted to 0.1 mg/mL after quantification with the Pierce Coomassie Plus (Bradford) Assay (Thermo Fischer Scientific). For samples with addition of exogenous, purified PfHsp90, the protein was exchanged into assay buffer using a 10-kDa cut-off centrifugal filter (Vivaspin) and added to lysate at a final concentration of 3 μM. An equal volume of buffer was added to treatments without exogenous PfHsp90 addition. Experiments were conducted at a 50 μL total volume in black, non-binding surface 384-well plates (Corning). Bortezomib (25 μM) (Selleck Chemical), BX-2819 (10–100 μM), or DMSO (0.5 %) were added and allowed to pre-incubate for 30 min at 37 °C. After which, 25 μM of the proteasome reporter substrate Suc-Leu-Leu-Val-Tyr-AMC (Suc-LLVY-AMC) (Cayman Chemical) was added. The final DMSO concentration was 1.0 % in all samples. Proteasome activity was monitored by measuring the fluorescence intensity of released AMC (Ex: 355 nm; Em: 460 nm) upon hydrolysis of the reporter substrate at 30 min increments over 180 min at 37 °C on the Envision plate reader (PerkinElmer). Background fluorescence from Suc-LLVY-AMC dosed into buffer was subtracted from sample measurements at each timepoint.

To assess proteasome activity after inhibitor treatment in cells, Tropane 1 (350 nM) or DMSO (1%) were administered to early trophozoite blood stage P. falciparum 3D7 parasites at ~15% parasitemia. Cultures were then incubated at 37 °C with 3% O₂, 5% CO₂, and 92% N₂ for 5 hours. After which, parasites were processed similarly by lysing red blood cells with 0.03% (wt/vol) saponin in PBS (Gibco) and washing released parasites 4 times with cold PBS (Gibco). Parasite pellets were resuspended in proteasome activity lysis buffer (50 mM TRIS-HCl, pH 7.5, 2 mM DTT, 5 mM MgCl₂, 5 % glycerol, 2 mM ATP disodium, and 0.05% digitonin) with 25 μM Suc-LLVY-AMC (0.5% DMSO). Lysate preparations (50 μL) were loaded into black, non-binding surface 384-well plates (Corning) and fluorescence measurements (Ex: 355 nm; Em: 460 nm) were taken on the Envision system (PerkinElmer) at 15 min increments over 60 min at 37 °C. For all experiments, data was normalized between the fluorescence value of each treatment at their 0 min timepoint and the fluorescence value of the DMSO treatment at their final timepoint. For the treatments conducted in lysate, the relative reaction rates were determined by fitting a linear regression between the 30–180 min timepoints of AMC release in GraphPad Prism and comparing the calculated slopes. All conditions were tested in technical duplicate with at least 3 independent determinations.

Cycloheximide chase native gel proteasome assays

Early trophozoite blood stage P. falciparum 3D7 at ~15% parasitemia were treated with 50 μM cycloheximide and either 10 μM geldanamycin or DMSO (1% final volume). After incubating 3 hours at 37 °C with 3% O₂, 5% CO₂, and 92% N₂, parasites were harvested as described above. Parasite pellets were lysed in proteasome activity lysis buffer (50 mM TRIS-HCl, pH 7.5, 2 mM DTT, 5 mM MgCl₂, 5 % glycerol, 2 mM ATP disodium, and 0.05% digitonin) followed by 5 consecutive freeze-thaw cycles in liquid nitrogen. Lysate was clarified by centrifugation. Samples were adjusted to 75 μg total protein, as quantified using the Pierce Coomassie Plus (Bradford) Assay (Thermo Fisher Scientific), in a volume of 20 μL. A 5X loading buffer (250 mM TRIS-HCl, pH 7.5, 43.5% glycerol, and 0.05% bromophenol blue (w/v)) was added to a 1X concentration, bringing the final volume to 25 μL. Samples were then loaded (20 μL) onto a NuPAGE 3–8% tris-acetate polyacrylamide gel (Thermo Fisher Scientific) with pre-chilled running buffer (89 mM TRIS base, pH 8.3, 89 mM boric acid, 2 mM EDTA, 413 μM ATP, 2 mM MgCl₂, and 0.5 mM DTT).⁵⁶ Gel electrophoresis proceeded on ice at 110 V for 2.5 hours. The resolved gel was incubated at 37 °C for 30 minutes in proteasome in-gel activity buffer (100 mM TRIS-HCl, pH 7.5, 10 mM MgCl₂, 1 mM ATP, 1 mM DTT, and 100 μM Suc-LLVY-AMC). Fluorescence from in-gel hydrolyzed Suc-LLVY-AMC (primarily from 26S) was measured using a ChemiDoc MP gel imager (Bio-Rad) with UV excitation and a standard filter. To activate and measure latent 20S complexes, the gel was returned to the activity buffer with 0.03% SDS, incubated at 37 °C for another 30 minutes, and re-imaged on the ChemiDoc MP. Relative quantification of band intensity was conducted in the Bio-Rad Image Lab software and normalized to the DMSO control. Results are reported from the average of 3 independent experiments.

Quantification and Statistical Analysis

GraphPad Prism 9 software was used for analysis of data. The sample size, standard error, significance, and statistical test employed for analysis are noted in respective figure legends.

Highlights.

• PfHsp90 inhibitors characterized with multi-stage anti-Plasmodium activity

• Tropane-based compounds exhibited selectivity to PfHsp90 over HsHsp90

• Thermal proteome profiling indicated putative PfHsp90 interactors

• PfHsp90 supports the parasite 26S proteasome complex