Bioinformatic analysis

The mRNA expression levels of NUDT5 in patient samples were obtained from the TCGA [20] dataset via the Oncomine platform (www.oncomine.org) [21] and the METABRIC [22,23,24] dataset via the cBioPortal (https://www.cbioportal.org) [24]. NUDT5 mRNA expression levels were reported as Log2Median-centered intensity. The differences in NUDT5 expression levels among different subtypes of breast cancer were determined using one-way Analysis of Variance (ANOVA) with Bonferroni's multiple comparisons test. For breast cancer cell lines, NUDT5 mRNA expression levels were acquired from the Cancer Cell Line Encyclopedia [25] and reported as Log2 (Fragments per kilobase million). To identify overexpressed phosphatases, we compared TNBC versus ER−/HER2+ using the Bonnefoi [26] dataset, and TNBC versus non-TNBC were obtained from the Bittner data set (Bittner et al. International Genomics Consortium (IGC) 2005) through the Oncomine platform [21]. The differences in NUDT5 expression levels between each group were determined by the Student’s t test. The survival data of breast cancer patients were obtained from multiple sources: METABRIC [22,23,24], Esserman [27], Kao [28] and Pawitan [29] datasets via the Oncomine platform (www.oncomine.org) [21], with data retrieved in 2018. The Kaplan–Meier survival curves were generated by dichotomizing patients at the mean expression level of NUDT5. The prognostic impact of NUDT5 expression was determined using Mantel–Cox log-rank analysis.

Cell line culture

Breast cancer cell lines were obtained from American Type Culture Collection (Manassas, VA). MCF7, MDA-MB-361, ZR-75-1, MDA-MB-231, MDA-MB-436, BT20, T47D, MDA-MB-468, and HEK293T cell lines were passaged and cultured in Dulbecco’s modified Eagle medium (DMEM) and supplemented with 10% regular fetal bovine serum; MCF10A and MCF12A cell lines were cultured in DMEM/F12 medium with 5% horse serum, 20 ng/mL EGF, 0.5 mg/mL hydrocortisone, 100 ng/mL cholera toxin, and 10 μg/mL insulin (Cellgro, Mediatech, Inc., Manassas, VA). HCC1937, HCC70, HCC1143 and BT474 cells were cultured in Roswell Park Memorial Institute medium (RPMI) supplemented with 10% regular fetal bovine serum. Growth media for all cell lines was supplemented with 100 mg/mL penicillin/streptomycin (Gibco™ 15,140,122, Thermo Fisher Scientific, Waltham MA). Cell line identities were confirmed through short tandem repeat DNA fingerprinting, as previously described [10]. A Lonza Mycoplasma Detection Kit (LT07-418; Lonza Walkersville, Inc.,) was used according to manufacturer instructions to detect mycoplasma.

siRNA transfection

NUDT5-specific targeting siRNA oligos were purchased from Sigma-Aldrich (SASI_Hs01_00109215, SASI_Hs02_00345134, 3’UTR siRNA: 5′ UGA AAG GGC UCU CCA GAU A 3′; St. Louis, MO). Cells were seeded at 50% confluency 1 day prior to siRNA transfection. A mixture of 20 nmol/L siRNA with DharmaFECT1 (T-2001-03; Dharmacon, Lafayette, CO) transfection reagent was added to the cells, with the reagent volume adjusted according to the manufacturer’s recommendations. Non-specific siLuc (SIC001; Sigma-Aldrich) was used as the negative control. Cells were harvested or reseeded for the next analysis after at least 48 h of siRNA transfection.

Plasmid and plasmid transfection

NUDT5 ORF cDNA plasmid (OHu06714) and vector control (pcDNA3.1+/C-(K)-DYK) were purchased from GenScript (Piscataway, NJ). X-treme-Gene9 transfection reagent (XTG9-RO; Roche) was used for the transfection according to the manufacturer's instructions.

Western blot and RT-qPCR analyses

Primary antibodies included: NUDT5 antibody (ab129172, 1:1000, Abcam, Cambridge, United Kingdom), and vinculin antibody (V9131, 1:1000, Sigma-Aldrich). Secondary anti-mouse and anti-rabbit horseradish peroxidase antibodies (1:1000) were obtained from GE Healthcare Bio-Sciences Corp (Piscataway, NJ). All of the target proteins and loading controls were processed in parallel. Western blots were performed in triplicate, following previously published methods [10].

RNA was extracted from the cells using RNeasy Mini Kit (74004, Qiagen, Germany). RNA was reverse transcribed into cDNA using random primers, and SuperScript II Reverse Transcriptase (18,064,022, Invitrogen, Waltham MA). The primers and probes for TaqMan-based NUDT5 and Cyclophilin were as follows: NUDT5: Forward primer CTCCGGGAGCTTGAAGAAGA, Reverse primer TTGACAAGCCTGGGTCCATA, Probe TGCCGAATGTTCTCCAGCGGTC (with 5′Fam and 3′Tamra labels), and Cyclophilin: Forward primer ACGGCGAGCCCTTGG, Reverse primer TTTCTGCTGTCTTTGGGACCT, Probe CGCGTCTCCTTTGAGCTGTTTGCA. Real-time quantitative PCR was performed using the QuantStudio 7Pro system (Applied Biosystems, Waltham, MA). Cyclophilin served as the endogenous control, and the relative gene expression was determined through the 2 − ΔΔCt method.

Cell growth assays

Cells were treated either with siLuc control or siNUDT5 for 48 h to induce knockdown. Subsequently, cell numbers were counted using a Countess automated cell counter (Invitrogen, Waltham MA) and reseeded at a density of 1000 cells per well in 96-well plates. These cells were cultured in the indicated media for a period of 7 days. For cell growth analyses, 4′,6-diamidino-2-phenylindole (DAPI)–stained cell nuclei were imaged at days 1, 3, 5, and 7 using an ImageXpress Pico microscope (Molecular Devices, San Jose, CA) and analyzed using CellReporterXpress image acquisition and analysis software. Cells were treated with either DMSO or 10 μM TH5427 on day 1 and Hoechst-stained (20 µM, Thermo Fisher Scientific, Hoechst 33342 Solution, Waltham MA) cell nuclei were imaged at days 1, 3, 5, and 7. The half maximal inhibitory concentration (IC50) of TH5427 was calculated using the 4 parameter logistic regression models by Prism 9.1 (GraphPad). The cells were individually plated into each well as quadruplicates, and cell numbers were reported as average cell count ± SD.

Xenograft growth

This study was conducted following animal protocols approved by The University of Texas M.D. Anderson Cancer Center Institutional Animal Care and Use Committee (IACUC). Female nude mice (The Jackson Laboratory, Bar Harbor ME) aged 4–6 weeks, were used in the experiments. TH5427 (Cat. No. 6534) compound was purchased from TOCRIS (Bristol, United Kingdom). One million MDA-MB-231 cells were subcutaneously injected into the 2nd mammary fat pad of 20 nude mice. When tumor size reached 50 mm3, these 20 mice were randomly divided into two groups of 10 each. One group of mice received intraperitoneal (i.p.) injections of vehicle (water), while the other group received i.p. injections of TH5427 (50 mg/kg) 5 times per week. Xenograft tumors were measured twice per week, and tumor volumes were calculated using the formula V = 0.5(width2 × length). Individual tumor growth rates were calculated by log-transformed linear regression slopes. Mice were sacrificed when the largest tumor reached 1000 mm3. 4 mice in the TH5427 treatment group were found dead after day 7. Growth rates were compared between the slopes of the vehicle and treatment groups using a Student’s t-test.

H&E and immunohistochemistry

Tumor samples from the mice were processed by fixation in a 1:10 formalin solution and subsequently embedded in paraffin. Hematoxylin–eosin staining (H&E) and immunohistochemical staining (IHC) of tumor tissue slides was performed, as previously described [22]. For H&E and IHC staining, 3 slides from each treatment group were used. For IHC, samples were incubated with Ki67 primary antibody (Lab Vision, 1:1000) or NUDT5 primary antibody (ab129172, Abcam, 1:500). Slides were processed at the Baylor College of Medicine Breast Center Pathology Core. Cell pellet blocks of MDA-MB-361 and MDA-MB-231 cells were used as negative and positive controls for the NUDT5 primary antibody. The images were acquired using Aperio ImageScope (Leica Biosystems, Illinois US) and processed with Aperio ImageScope Pathology Slide Viewing Software (Leica Biosystems, Illinois US).

Proliferation, death, apoptosis, and ROS assays

Cells were seeded into 96-well plates at a density of 5000 cells per well. To synchronize the cell cycle, cells were incubated with Lovastatin (10 μM) for 48 h, then incubated with 1 mM mevalonate for 24 h to release them back to the normal cell cycle. Proliferation was assayed by bromodeoxyuridine (BrdU) incorporation using the Roche Cell Proliferation ELISA BrdU chemiluminescent kit, according to the manufacturer's instructions. Dying cells were detected by DRAQ7™ (3 µM, Abcam, ab109202) positivity and nuclei were counterstained with Hoechst (20 µM, Thermo Fisher Scientific™, Hoechst 33342 Solution, Waltham MA). Images were captured via ImageXpress® Pico microscope (Molecular Devices, San Jose, CA) and analyzed with CellReporterXpress image acquisition and analysis software. Apoptosis was detected by Annexin V-PI positivity (Invitrogen, Annexin V-FITC Conjugates, Waltham MA). Flow cytometry analysis was conducted with the assistance of the Flow Cytometry and Cellular Imaging Core Facility North Campus at The University of Texas MD Anderson Cancer Center. Experimental data points were collected in biological triplicates, and results were reported as average ± SD. We measured the reactive oxygen species (ROS) level of breast cancer cells using the ROS-Glo™ H2O2 Assay kit (Promega, Madison MI) according to manufacturer’s protocol. The assay is a two-reagent-addition protocol. The H2O2 substrate (provided by the kit) reacts directly with H2O2 to create the luciferin precursor, which then is converted to luciferin and reacts with luciferase (provided by the kit) to generate a luminescent signal, so that the signal is proportional to the H2O2 level. The luminescence was recorded by a BioTek Synergy Mx Microplate Reader (BioTek, Winooski, VT). The measurement of breast cancer cells was conducted both at basal level and after oxidative induction with 50 μM H2O2 for 6 h.

Immunofluorescence staining

After 72 h of siRNA treatment, cells were seeded at a density of 5000 cells per well in 96-well plates. Cell fixation was performed with cold methanol after 4 days for 8-oxoG staining and 7 days for γH2AX staining. Samples were incubated with 8-oxoG (Santa Cruz Biotechnology, sc-130914, 1:50) and γH2AX (Cell Signaling Technology, #9718, 1:1000) antibodies at 4 °C overnight. The following day, samples were incubated with the secondary antibodies Alexa Fluor 488 (Invitrogen, A28175, 1:1000) and Alexa Fluor 594 (Invitrogen, A-11032, 1:1000), accordingly, for 1 h at room temperature. The images were obtained using the ImageXpress® Pico microscope (Molecular Devices, San Jose, CA) and analyzed with CellReporterXpress image acquisition and analysis software. Experimental data points were performed in quadruplicate, and results were reported as average ± SD. 8-oxoG intensity and γH2AX-positivity were compared between siLuc and siNUDT5 using the Student's t-test.

DNA fiber assay

After 48 h of siRNA treatment, 10,000 MDA-MB-231 and MCF7 cells were seeded into 6-well plates. The following day, cells were exposed to 50 μM 5-Iodo-2'-Deoxyuridine (IdU) for 30 min, followed by a wash with phosphate-buffered saline, and treated with 100 μM 5-chloro-2′-deoxyuridine (CIdU) for another 30 min. The DNA fiber assay was performed as described [30]. First, samples were incubated with a primary antibody mix of rat monoclonal anti-BrdU antibody (Abcam, ab6326, 1:500) and mouse monoclonal anti-BrdU antibody (BD Biosciences, 347580, 1:500, San Jose, CA) in a blocking solution overnight at 4 °C. Samples were then incubated with a secondary antibody mix composed of Alexa Fluor 594 goat anti-rat IgG (1:1000, ThermoFisher Scientific, A-11007) and Alexa Fluor 488 goat anti-mouse IgG (1:1000, ThermoFisher Scientific, A-11001). DNA fibers were imaged with Andor Revolution XDi WD Spinning disk confocal microscope (Oxford Instruments plc, UK) with assistance from the University of Texas MD Anderson Cancer Center North Campus Flow Cytometry and Cellular Imaging Core Facility. The length of the individual fibers (up to 40 fibers in each group) were measured using the microscopy image analysis software Imaris (Oxford Instruments plc, UK). Statistical differences of the mean fiber length were compared using a Student’s t test and reported as average ± SD.

Statistics

All graphs were presented as mean ± standard deviation. Two-tailed Student’s t tests were used to determine the statistical significance between two different groups, and one-way ANOVA tests were used to determine the statistical significance among multiple groups. p-values less than 0.05 were considered statistically significant,, with symbols used to denote levels of significance as follows: * represents p ≤ 0.05, ** represents p ≤ 0.01, *** represents p ≤ 0.001, **** represents p ≤ 0.0001.