High-throughput screening assay for PARP-HPF1 interaction inhibitors to affect DNA damage repair

Abstract

ADP-ribosyltransferases PARP1 and PARP2 play a major role in DNA repair mechanism by detecting the DNA damage and inducing poly-ADP-ribosylation dependent chromatin relaxation and recruitment of repair proteins. Catalytic PARP inhibitors are used as anticancer drugs especially in the case of tumors arising from sensitizing mutations. Recently, a study showed that Histone PARylation Factor (HPF1) forms a joint active site with PARP1/2. The interaction of HPF1 with PARP1/2 alters the modification site from Aspartate/Glutamate to Serine, which has been shown to be a key ADP-ribosylation event in the context of DNA damage. Therefore, disruption of PARP1/2-HPF1 interaction could be an alternative strategy for drug development to block the PARP1/2 activity. In this study, we describe a FRET based high-throughput screening assay to screen inhibitor libraries against PARP-HPF1 interaction. We optimized the conditions for FRET signal and verified the interaction by competing the FRET pair in multiple ways. The assay is robust and easy to automate. Validatory screening showed the robust performance of the assay, and we discovered two compounds Dimethylacrylshikonin and Alkannin, with µM inhibition potency against PARP1/2-HPF1 interaction. The assay will facilitate the discovery of inhibitors against HPF1-PARP1/2 complex and to develop potentially new effective anticancer agents.

Figure 1

HPF1-PARP interaction. (a) HPF1 (magenta) and PARP2 (blue) interaction interface. NAD⁺ analog EB-47 (green) is bound to the PARP2 catalytic site (PDB id: 6TX3), (b) HPF1-PARP FRET pair interaction (left panel) and inhibition of HPF1-PARP interaction by inhibitor resulting in loss of FRET (right panel).

Figure 2

Buffer and protein concentration optimization. (a) Buffer pH optimization in 10 mM Bis–Tris-Propane. (b) Buffering reagent optimization with 10 mM concentration at pH 7.0. (c) Bis–Tris-Propane (BTP) concentration optimization at pH 7.0. (d) NaCl concentration optimization in 10 mM BTP at pH 7.0. (e) %(w/v) PEG 20k optimization in 10 mM BTP at pH 7.0. f) Detergent Triton X-100%(v/v) optimization in 10 mM BTP at pH 7.0 In panels a-f, 5 µM YFP-HPF and 1 µM CFP-PARP2 were used in reactions. (g) YFP-HPF1 and CFP-PARP2 protein concentration optimization in FRET buffer (10 mM BTP, 0.01% Triton X-100, 3% PEG 20k, pH 7.0) where x-axis is representing CFP-PARP2 concentrations (µM) and YFP-HPF1 was mixed in 2:1 (white bars) and 3:1 (grey bars) ratio to CFP-PARP2 concentration. All the buffer and protein concentration optimization were performed in 20 µl volume in 384 well plates and the data shown are mean with standard deviations of 4 replicates.

Figure 3

rFRET signal for CFP-PARP2 and YFP-HPF1. (a) Time dependent monitoring of rFRET signal, where FRET pair concentration is 400 nM CFP-PARP2 and 800 nM YFP-HPF1. (b) PARP2 ART domain (1 mg/ml) thermal stability without (grey) and with (black) Olaparib (100 µM). (c) rFRET signal (400 nM CFP-PARP2 and 800 nM YFP-HPF1) monitoring with time in absence and presence of Olaparib [1—negative rFRET signal without Olaparib, 2—positive rFRET signal without Olaparib, 3—negative rFRET pair signal with Olaparib, 4—positive rFRET pair signal with Olaparib]; (d) Effect of protein mixing on rFRET signal, [Dark grey bars: rFRET signal with intermediate dilution of protein (25 µM CFP-PARP2 and 50 µM YFP-HPF1) for no FRET ( −) and FRET ( +) reaction; Light grey bars: rFRET signal with mixing of protein (400 nM CFP-PARP2 and 800 nM YFP-HPF1) for no FRET ( −) and FRET ( +) reaction in the FRET buffer]. All the reactions were performed in 20 µl volume in 384 wells plate and the data shown are mean with standard deviations of 4 replicates.

Figure 4

CFP-PARP2 and YFP-HPF1 FRET pair signal competition with unlabeled HPF1 and determination of a dissociation constant. (a) FRET pair competition with unlabeled HPF1 containing salt (black) and equivalent salt (25 nM–350 mM; grey). (b) FRET pair competition with desalted unlabeled HPF1. (c) FRET-based determination (Kd) for CFP-PARP2 with YFP-HPF1 interaction. The FRET fluorescence emissions (Em_FRET) were determined as described by Song et al.. All the reactions were performed in 20 µl volume in 384 well plates and the data shown are mean with standard deviations of 4 replicates.

Figure 5

Signal validation and validatory screening in 384-well plates. (a) Signal validation in 10 µl reaction volume. (b) Signal validation in 20 µl reaction volume. (c) Validatory screening using Targetmol compound library with 10 µM compound in 10 µl final reaction volume.

Figure 6

rFRET signal for CFP-PARP1 and YFP-HPF1 FRET pair. (a) rFRET signal from control ( −) and FRET pair ( +) at different protein concentrations of CFP-PARP1 and YFP-HPF1. (b) Time dependent monitoring of rFRET signal, with 200 nM CFP-PARP1 and 400 nM YFP-HPF1 protein concentration. All the reactions were performed in 10 µl volume in 384 well plates and the data shown is mean ± standard deviation with 4 replicates.

Figure 7

Example IC₅₀ curves for Inhibition of PARP-HPF1 interaction by increasing concentration of inhibitor compounds. PARP1-HPF1 interaction inhibition by (a) Dimethylacrylshikonin (b) Alkannin. PARP2 -HPF1 inhibition by (c) Dimethylacrylshikonin (d) Alkannin.

Figure 8

Inhibitory effect of Dimethylacrylshikonin and Alkannin on PARP-HPF1 interaction. (a) PARP1 and (b) PARP2 activity assay to monitor the inhibition by inhibitor compounds. The assay was repeated twice for both PARP1 and PARP2 with similar results.