Native Mass Spectrometry-Guided Screening Identifies Hit Fragments for HOP-HSP90 PPI Inhibition

Abstract

Contemporary medicinal chemistry considers fragment-based drug discovery (FBDD) and inhibition of protein-protein interactions (PPI) as important means of expanding the volume of druggable chemical space. However, the ability to robustly identify valid fragments and PPI inhibitors is an enormous challenge, requiring the application of sensitive biophysical methodology. Accordingly, in this study, we exploited the speed and sensitivity of nanoelectrospray (nano-ESI) native mass spectrometry to identify a small collection of fragments which bind to the TPR2AB domain of HOP. Follow-up biophysical assessment of a small selection of binding fragments confirmed binding to the single TPR2A domain, and that this binding translated into PPI inhibitory activity between TPR2A and the HSP90 C-terminal domain. An in-silico assessment of binding fragments at the PPI interfacial region, provided valuable structural insight for future fragment elaboration strategies, including the identification of losartan as a weak, albeit dose-dependent inhibitor of the target PPI.

Keywords: HSP70-HSP90 organizing protein; PPI inhibitors; fragment-based drug discovery; heat shock protein 90; native mass spectrometry.

Figure 1

A. X‐ray co‐crystal of Ac−MEEVD−OH (1) bound to the TPR2A domain of HOP (PDB 1ELR). Glu2 and Asp5 form a series of salt bridges with Lys229, Asn233, Gln298, Lys301, Arg305 and Asn308 residues, while Val4 occupies a hydrophobic pocket (white surface) lined with aromatic residues Tyr236 and Tyr248. B. Native mass spectrum of the 1 : 1 buffered solution of TPR2AB and 1. C. Expanded region of the 11⁺ charge state, showing the apo TPR2AB (m/z 3618.8, [M+11H]¹¹⁺) the TPR2AB‐1 complex (m/z 3679.1, [M⋅1 L+11H]¹¹⁺) and TPR2AB bound to two Ac−MEEVD−OH peptides (m/z 3739.5, [M⋅2 L+11H]¹¹⁺).

Figure 2

Expanded region of the 11⁺ charge state of a buffered mixture of TPR2AB with four selected binding fragments, F8 (A), F17 (B), F21 (C) and F 22 (D). Binding ratio was estimated based on the relative abundance of apo TPR2AB (m/z 3618.8, [M+11H]¹¹⁺) and TPR2AB‐fragment complex and shown in Tables 1 and 2. Fragment binding was confirmed based on the Δ m/z from the apo TPR2AB peak.

Figure 3

A. Thermal shift assay demonstrating ligand‐induced stabilisation of TPR2A. Data shown are averages ±SD. The average melting temperature (Tm) is shown above the bars. **** p<.0001 comparing TPR2A Tm in the presence of vehicle control (DMSO) to Tm in presence of ligands by one‐way ANOVA. Melt curves for the TPR2A protein in the absence (solid black line) and presence (dashed or coloured lines) of B fragments or C ‘sartan’ drugs. The shift in the location of the highest negative peak to a higher temperature indicates stabilisation of the protein in the presence of ligands.

Figure 4

Solid phase PPI ELISA assay, between the HOP TPR2A domain and the HSP90 C‐terminal domain, in the presence of several selected fragments. F5, F16 and F17 displayed the most promising dose‐dependent PPI inhibition. * p<.05; ** p<.01; *** p<.001; **** p<.0001.

Figure 5

A–C. Docked binding pose of F5, F10 and F16, respectively. Group 1 and 2 fragments were predicted to bind at the hydrophobic Val4 binding pocket (white surface), with their acidic moieties predicted to interact with the basic, Gln298, Lys 301 and Arg305 cluster purple). D. Overlay of the binding poses of F2 and F29, whose combined structure resembles losartan (5).

Figure 6

Solid phase PPI ELISA assay, between HOP TPR2A domain and the HSP90 C‐terminal domain. Losartan displayed a mild yet statistically significant stabilizing effect on the PPI at 25 μM. Both losartan and valsartan displayed statistically significant PPI inhibitory activity at 200 μM. * p<.05; ** p<.01; *** p<.001; **** p<.0001.