Investigation of a cryptic ligand binding site on Plasmodium falciparum Hsp90

Abstract

The molecular chaperone heat shock protein 90 (Hsp90) has an important role in maintaining proteostasis in Plasmodium parasites, the causative agents of malaria, and is of interest as a potential antimalarial drug target. Inhibitors targeting its well-characterized N-terminal ATP-binding site are lethal, but the development of high-affinity binders with selectivity for the Plasmodium over the human homolog has been challenging given the high conservation of this domain. A binding site in the less conserved Hsp90 C-terminus has been reported to interact with nucleotides and inhibitors in other eukaryotic systems, which could offer an alternative route for antimalarial design. Herein, we characterize the potential ligandability of the C-terminus in the Plasmodium falciparum chaperone PfHsp90 with in silico and in vitro methods. We conducted affinity experiments with both a lysine-reactive nucleotide analog and an ATP resin that support a specific interaction between ATP and a C-terminal truncation of PfHsp90. We further explored the nucleotide structural requirements for this interaction with limited proteolysis experiments, which suggest association with ATP, dATP, and ADP, but not AMP or GTP. Lastly, we employed computational analyses and mutagenesis studies to interrogate the molecular basis for the interaction. Our findings provide the foundation for future studies to assess and develop C-terminal Hsp90 inhibitors against Plasmodium parasites.

Keywords: Cryptic site; Heat shock protein 90; Hsp90; Nucleotide interactions; Plasmodium.

Figure 1.. Prediction of a putative nucleotide binding site in PfHsp90 MC.

(A) Domain schematic of PfHsp90. The MC construct consists of the contiguous MD and CTD, which, in the full-length protein, are connected to the NBD by a flexible linker. (B) PfHsp90 MC ribbon model with the location of residues predicted by NsitePred to be nucleotide-interacting with a probability of <20%, >20%, or >50% highlighted in yellow, orange, or red, respectively. Red arrow identifies >50%. (C) PfHsp90 MC model with molecular surface representing the CryptoSite score scaled on a gradient of white to blue representing the regions with lowest to highest possibility, respectively, of maintaining a cryptic ligand binding site. (D) PfHsp90 MC with ATP docked using Schrödinger Maestro Glide. Locations of residues with a CryptoSite score >0.15 or an NsitePred probability >20% are designated by the blue or red surface area, respectively. All structures are based on the PfHsp90 AlphaFold model (AF-Q8IC05-F1-v4).

Figure 2.. Evaluation of a PfHsp90 MC and ATP interaction.

(A) Representative streptavidin-HRP Western blot (top) and Ponceau stain (bottom) of PfHsp90 MC (left) and PfHsp90 NBD (right) after labeling by desthiobiotin-ATP in the presence of 0 mM or 10 mM ATP. (B-C) Quantification of desthiobiotin-ATP labeling of PfHsp90 MC (B) and PfHsp90 NBD (C) with 10 mM ATP relative to 0 mM ATP [shown in (A)]. Data displayed as means ± SEM, n = 3; ***<0.001, ****<0.0001; unpaired t-test. (D) Representative Coomassie staining for pull-down experiments showing PfHsp90 MC total protein input (Input), native PfHsp90 MC pulled-down with ATP-Sepharose (Pull-down), native PfHsp90 MC pulled-down with a negative control resin (Non-binding), and heat denatured PfHsp90 MC pulled-down with ATP-Sepharose (Boiled). (E) Quantification of Coomassie staining of protein recovered in pull-down experiments relative to Input [shown in (D)]. Data displayed as means ± SEM, n = 3; not significant (ns)>0.5, *<0.05, ****<0.0001; one-way ANOVA, Dunnett’s multiple comparison test. (F) Representative confocal microscopy images of ATP-Sepharose (top left), Ni-NTA agarose (bottom left), and negative control agarose (bottom right) after incubation with PfHsp90 MC-488 (+protein). Confocal microscopy image of ATP-Sepharose (top right) without PfHsp90 MC-488 (-protein) is also shown. Scale bar is 100 μm.

Figure 3.. ATP-induced protection of PfHsp90 MC from proteolysis.

(A) Representative SDS-PAGE with Coomassie staining of PfHsp90 MC after incubation with thermolysin in the presence of NaCl (50 mM or 500 mM), MgCl₂ (0 mM or 10 mM), and ATP (0 mM or 2.5 mM). Undigested protein included as total input (In). (B) Quantification of the amount of undigested PfHsp90 MC, calculated as a percent of total protein input, in limited proteolysis experiments [shown in (A)]. Data displayed as means ± SEM, n = 3; not significant (ns)>0.5, ***<0.001, ****<0.0001; two-way ANOVA, Tukey’s multiple comparison test.

Figure 4.. Biophysical characterization of PfHsp90 MC variants.

(A) Sequence alignment of the PfHsp90 (Pf) and HsHsp90 (Hs) region including residues altered by site-directed mutagenesis (blue arrow). NsitePred predictions are highlighted yellow, orange, or red, corresponding to a nucleotide interaction probability of <20%, >20%, or >50%, respectively. Residues within 6 Å of ATP docked into PfHsp90 MC [Shown in (Fig. 1D)] are designated by a black bar above the sequence. (B) Quantification of the melt temperature (Tm) from DSF experiments of each PfHsp90 variant: WT, G515A, D545A, E566A, R610A, and C616A. Data displayed as means ± SEM, n = 3; not significant (ns)>0.5, **<0.01; one-way ANOVA, Dunnett’s multiple comparison test. (C) Individual histograms displaying the average relative frequency (%) of different mass species from mass photometry analysis of each PfHsp90 MC variant: WT (top left), G515A (top middle), D545A (top right), E566A (bottom left), R610A (bottom middle), and C616A (bottom left). The monomeric and dimeric populations are indicated by orange (centered at 50 kDa) and blue (centered at 100 kDa) fill area, respectively. The remaining grey fill area encompasses additional higher-order oligomeric species. Data displayed as the average of 3 independent determinations, and the statistical comparison between the quantity of monomers and dimers is based on their total proportion encompassed within their respective orange and blue fill areas [shown in (Fig. S5D)]: not significant (ns)>0.5, ****<0.0001; two-way ANOVA, Tukey’s multiple comparison test.

Figure 5.. Investigation of nucleotide interactions in PfHsp90 variants.

(A) The degree of desthiobiotin-ATP labeling of each PfHsp90 MC mutant: G515A, D545A, E566A, R610A, and C616A; relative to the WT protein, as quantified from streptavidin-HRP Western blot [shown in (Fig. S6A)]. Data displayed as means ± SEM, n = 3; not significant (ns)>0.5, **<0.01, ****<0.0001; one-way ANOVA, Dunnett’s multiple comparison test. (B) Representative SDS-PAGE with Coomassie staining of both PfHsp90 MC WT and D545A after limited proteolysis in the absence of nucleotides (−) or in the presence of 2.5 mM ATP, dATP, ADP, or GTP. Undigested protein is included as total input (In). (C) Quantification of the amount of undigested PfHsp90 MC WT and D545A, calculated as a percent of total protein input for each variant, in limited proteolysis experiments [shown in (B)]. Data displayed as means ± SEM, n = 3; not significant (ns)>0.5, **<0.01, ***< 0.001, ****<0.0001; two-way ANOVA, Tukey’s multiple comparison test.