Targeted NUDT5 inhibitors block hormone signaling in breast cancer cells

Abstract

With a diverse network of substrates, NUDIX hydrolases have emerged as a key family of nucleotide-metabolizing enzymes. NUDT5 (also called NUDIX5) has been implicated in ADP-ribose and 8-oxo-guanine metabolism and was recently identified as a rheostat of hormone-dependent gene regulation and proliferation in breast cancer cells. Here, we further elucidate the physiological relevance of known NUDT5 substrates and underscore the biological requirement for NUDT5 in gene regulation and proliferation of breast cancer cells. We confirm the involvement of NUDT5 in ADP-ribose metabolism and dissociate a relationship to oxidized nucleotide sanitation. Furthermore, we identify potent NUDT5 inhibitors, which are optimized to promote maximal NUDT5 cellular target engagement by CETSA. Lead compound, TH5427, blocks progestin-dependent, PAR-derived nuclear ATP synthesis and subsequent chromatin remodeling, gene regulation and proliferation in breast cancer cells. We herein present TH5427 as a promising, targeted inhibitor that can be used to further study NUDT5 activity and ADP-ribose metabolism.

Fig. 1

NUDT5 is a key regulator of ADP-ribose metabolism. a Hydrolysis of potential oxidized nucleotides and nucleotide-sugar substrates by MTH1 (blue) and NUDT5 (red), as measured by the enzyme-coupled malachite green assay (MG assay), at pH 7.5. A representative experiment (of n = 2) with mean ± SD and individual values of quadruplicate replicates is shown. b Representative HPLC traces of NUDT5-mediated ADPR (red) and 8-oxo-dGDP (blue) hydrolysis to AMP or 8-oxo-dGMP, respectively (n = 2). c Top: Representative, pseudo-colored immunofluorescence stainings for γH2A.X, chromatin-bound RPA, and 53BP1 in U-2 OS cells following treatment with negative control siRNA (siNeg) or NUDT5 siRNA (siNUDT5 #1, siNUDT5 #7) for 72 h. Scale bar = 20 µm. Bottom: Quantification of γH2A.X, chromatin-bound RPA, and 53BP1 staining intensity in cells representing three independent experiments (cells scored: γH2A.X, n = 871; RPA, n = 463; 53BP1, n = 981). Lines represent median fluorescence (and interquartile range for 53BP1); a.u. arbitrary units. NS not significant, ***p < 0.001, ****p < 0.0001; Kruskal-Wallis test). d Top: Representative, pseudo-colored images of treated U-2 OS cells following the modified alkaline comet assay and confirmation of NUDT5 knockdown with a cropped western blot. Scale bar = 50 µm. Bottom: Quantification of the tail moment (relative to the DMSO+siNeg, buffer-treated control) from two independent experiments performed in duplicate (minimum of 100 cells per slide per condition). Buffer-treated samples are shown in blue, OGG1-treated samples in red. Lines represent median tail moments. ****p < 0.0001 by Kruskal–Wallis test comparing OGG1-treated samples. e A graphical procedure depicting the analysis of ADPR hydrolysis in cell lysates by HPLC. f ADP-ribose hydrolysis in U-2 OS cells alone (black, solid and dotted), following depletion with two different NUDT5 siRNAs (blue, solid and dotted) or siRNA-treated cells complemented with purified NUDT5 protein (red, solid and dotted), as quantified by HPLC. A representative chromatogram is shown (of n = 2 independent experiments) and NUDT5 knockdown confirmation with cropped western blot

Fig. 2

Development of potent NUDT5 inhibitors from a high-throughput screening campaign. a A representation of the enzyme-coupled malachite green assay (MG assay). ADPR is hydrolyzed to AMP and R5P by NUDT5, then R5P is converted to free inorganic phosphate detected by the malachite green reagent. b Preliminary structure–activity relationship study for determining where modifications are tolerated. Favorable optimization of TH1167 was performed by modification at the theophylline 8-position (purple) and tail position (green) of the molecule. MG assay IC₅₀ values are reported below the chemical structures. c Structure of NUDT5 with inhibitor TH1713. Binding of TH1713 (orange) by NUDT5 dimer (chain A, limon green and chain B, green) through stacking between Trp28 of chain A and Trp46 of chain B and hydrogen bonds to the amide nitrogen of Glu47 and side chain of Arg51. An alternate conformation of the methylbenzene modality is modeled in gray. d NUDT5 inhibitor screening funnel optimized to select for target engagement in cells

Fig. 3

CETSA-guided screening distinguishes TH5427 as a lead NUDT5 inhibitor. a Biochemical characterization of compounds shown in Table 2 by enzyme-coupled malachite green assay (MG assay). Compounds of particular interest are labeled with their code names and MG assay IC₅₀ values (nM). b Compounds with MG assay IC₅₀ < 100 nM and reasonable solubility were tested by high-throughput CETSA (HT-CETSA) in HL-60 cell lysates. Values shown are relative to NUDT5 band intensity at 37 °C, normalized to SOD1 and a representative of two experiments. Cropped, representative blots are included. c Compounds with > 50 % stabilization in cell lysate CETSA were tested for target engagement at 20 µM with intact HL-60 cells by CETSA. Values are plotted relative to NUDT5 band intensity at 37 °C and normalized to SOD1. Data are means ± SEM and individual points from n = 2 experiments. Cropped, representative blots are shown. NS not significant, *p < 0.05, **p < 0.01; one-way ANOVA analysis. d HL-60 cells were treated with the same compounds from c but titrated by serial dilution from 20 µM to 0.06 µM for isothermal dose-response fingerprint CETSA (ITDRF_CETSA). Data for TH5427 comprise data points from three independent experiments and representative experiments are shown for the remaining compounds. Representative, cropped blots are also shown. In all cases above, compounds colored in red progressed to the next stage of evaluation. e Structure of NUDT5 with TH5427 bound. Coordination of TH5427 (cyan) by NUDT5 dimer (chain A – limon green and chain B – green) via stacking interactions between Trp28 of chain A and Trp46 of chain B and hydrogen bonds to the amide nitrogen of Glu47 and side chain of Arg51

Fig. 4

TH5427 inhibits progestin-dependent nuclear ATP synthesis in breast cancer cells. a Representative thin layer chromatography (TLC) of products formed following processing of ³²P-PAR by recombinant PARG and NUDT5 in the absence or presence of TH5427 and PP_i in vitro (n = 2). b Concentrations and structures of inhibitors used in T47D^WT/M cell culture experiments. c Serum-starved T47D^WT cells expressing Nuc-luc FRTTO luciferase construct in the absence or presence of NUDT5 inhibitors prior to treatment with 10 nM R5020 and luminescence measurement. Top, a representative, pseudo-colored image of bioluminescence intensity over 60 min of R5020 treatment; bottom, box-and-whisker plots of bioluminescence quantitations from six independent experiments (center line, median; box limits, upper and lower quartiles; whiskers, minima and maxima; individual data points also shown). NS not significant, *p < 0.05, **p < 0.01, ***p < 0.001; repeated measures two-way ANOVA analysis. Control, gray, TH5423, blue, TH5424, red, TH5427, orange. d Histone displacement was determined by chromatin immunoprecipitation (ChIP) using a histone H1-specific antibody prior to (Starved, gray) or following 30 min of R5020 (+R5020, blue) in the presence or absence of NUDT5 inhibitors in T47D^M cells. Data from a representative experiment is presented as mean fold change (+R5020/Starved) ± SD of five different histone H1 contact regions. e Progesterone-dependent and –independent gene expression analysis in the presence or absence (−) of NUDT5 inhibitors following R5020 treatment (6 h) by RT-qPCR with T47D^M cells. Data represents the mean ± SEM log(mRNA abundance) normalized to serum-starved cells without R5020 treatment from three (progesterone-dependent) or two (progesterone-independent) independent experiments. *p < 0.05, **p < 0.01; one-way ANOVA analysis. f R5020-induced cell proliferation of T47D^M cells in the absence or presence of NUDT5 inhibitors was assayed by BrdU incorporation after 24 h. Mean ± SEM of individual BrdU chemiluminescence signals are displayed without R5020 treatment (Starved, gray) and following R5020 treatment (+R5020, blue) from two independent experiments. RLU; relative luminescence units. **p < 0.01, ****p < 0.0001; one-way ANOVA analysis