The LNCaP (ATCC, CRL-1740), 22Rv1 (ATCC, CRL-2505), DU145 (ATCC, HTB-81), C4-2 (ATCC, CRL-3314), MCF7 (ATCC, HTB-22) and SJSA1 (ATCC, CRL-2098) cell lines were purchased from American Type Culture Collection (ATCC, VA, United States). The HepG2 (SCSP-510) and NCI-H460 (TCHu205) cell lines were purchased from the National Collection of Authenticated Cell Cultures of China. LNCaP, C4-2, 22Rv1, SJSA1 and NCI-H460 cell lines were cultured with Roswell Park Memorial Institute (RPMI)-1,640 supplemented with 10% foetal bovine serum (FBS, Gibco, Thermo Fisher Scientific, MA, United States). DU145 and HepG2 cell lines were cultured with DMEM supplemented with 10% FBS. MCF7 was cultured with Eagle’s minimum essential medium supplemented with 10% FBS. LNCaP-R cells were derived from LNCaP cells, as LNCaP cells were cultured with RPMI-1640 supplemented with 10% FBS and 5 μM enzalutamide for 2 months, and then these 5 μM enzalutamide-treated LNCaP cells were cultured with RPMI-1640 supplemented with 10% FBS and 10 μM enzalutamide for another 6 months.

XR-1, XR-2 and XR-3 (all produced in house) were dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, MO, United States) for in vitro experiments, and XR-2 was dissolved in Cremophor EL (Shanghai Aladdin Bio-Chem Technology Co., LTD., China), 95% ethanol (Shanghai Aladdin Bio-Chem Technology Co., LTD.) and saline solution (1.5:1.5:7) for in vivo experiments. Enzalutamide and RG7388 (idasanutlin) were purchased from Selleck (Houston, TX, United States).

Logarithmic growth-phase cells were seeded at a density of ∼4 × 10⁵ cells per well in a 6-well plate and incubated for 48 h. Then, DMSO (Sigma-Aldrich) or test compounds were added to each well at the designated concentrations. After another 24 h of incubation, cells were collected and lysed with radio-immunoprecipitation assay lysis buffer supplemented with protease inhibitor and phosphatase inhibitor on ice for 30 min. Then, the protein lysis buffer was treated with 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes for Western blot analysis. The following antibodies were used for the detection of proteins: rabbit anti-MDM2 (1:1,000, Abways, Tracxn Technologies Limited, India), rabbit anti-p53 (1:1,000, Abways), mouse anti-PARP1 (1:1,000, Santa Cruz, United States), rabbit anti-cleaved PARP1 (1:1,000, Selleck), mouse anti-AR (441, 1:1,000, Santa Cruz, United States) and rabbit anti-PSA (1:1,000, Abways). Mouse anti-actin (1:5,000, Abcam) was used as a loading control. Proteins were visualised using anti-mouse, anti-rabbit or anti-rabbit HRP-conjugated secondary antibodies (1:5,000, Zhongshan Jinqiao Biotechnology Company, China) and ECL-Plus (Millipore, MA, United States). The resulting bands were analysed and quantified using ImageJ^® 1.49 g software (National Institutes of Health, MD, United States). Each experiment was repeated at least twice.

PCa cells were seeded at a density of 1 ×10⁴ cells per well in a 96-well plate. LNCaP and 22Rv1 cells were cultured in a complete RPMI-1640 growth medium. DU145 cells were cultured in DMEM growth medium. After 24 h incubation, 1 µL of DMSO (Sigma-Aldrich) or 1 µl of the indicated concentrations of XR-2 were added to each well. After 72 h of incubation, 20 µl of MTT (Invitrogen, MA, United States) solution (5 mg/ml in phosphate-buffered saline (PBS)) was added per well and incubated for another 4 h. The MTT formazan formed by metabolically viable cells was dissolved in 100 µl of isopropanol. The absorbance was measured at 570 nm wavelength on a plate reader (EnSpire 2,300, PerkinElmer, MA, United States). Experiments were performed in triplicates. The value of the DMSO group was 100%.

Logarithmic growth phase cells were seeded at a density of ∼1 ×10⁴ cells per well in 6-well plates (n = 3) and treated with the required drugs/compounds or vehicle for 12–14 days. The culture media were changed every 3 days. After removal of the culture media, cells were washed by PBS twice and colonies were fixed by methanol for 10 min. Thereafter, colonies were stained using 1% (w/v) crystal violet (Sigma, MO, United States) for 10 min. Each well was washed with distilled deionised water until the background was clean; after another 30 min of airing, colony pictures were generated. To quantify staining, the stained wells were washed with 1 ml of 10% acetic acid, and absorbance at 590 nm wavelength was detected on a plate reader (EnSpire 2,300, PerkinElmer).

All kinds of cells were seeded at a density of ∼4 × 10⁵ cells per well in a 6-well plate and incubated for 48 h. Then, cells were treated with vehicles (DMSO) or test compounds at designated concentrations. After another 24 h of incubation, total RNA was extracted by TRIzol (Invitrogen, MA, United States) according to the manufacturer’s instructions. Complementary cDNA synthesis was performed using a cDNA reverse transcription kit (AGbio, Inc.) and total mRNA templates. Relative mRNA levels of AR, PΜMA, p53 and p21 were quantified by qRT-PCR using a SYBR Green Premix Pro Taq HS qPCR Kit (AGbio) on the qPCR instrument (Bio-Rad, CA, United States). The mRNA expression levels were normalised to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). All reactions were performed in triplicates. The gene-specific primers are listed in Supplementary Table S1.

The Small interfering RNAs (siRNAs) of TP53 were designed and synthesized by Sangon Biotech (Shanghai, China). siRNA stocks (20 µM) and Lipofectamine™ RNAiMAX Transfection Reagent (ThermoFisher, 13778100) were diluted in opti-MEM, the mix was incubated for 20 min and then added to cells according to the instruction. After another 24 h, cells were treated with indicated compounds. The siRNA primers are listed in Supplementary Table S2.

Logarithmic growth phase cells were seeded at a density of ∼4 ×10⁵ cells per well in a 6-well plate and incubated for 48 h. Then, DMSO (Sigma-Aldrich) or test compounds were added to each well at designated concentrations. For the cell cycle analysis, after compounds were treated for 24 h, cells were harvested and fixed with ice-cold 70% (v/v) ethanol for 24 h. Then, cells were centrifuged at 1,000 rpm for 5 min. Cell pellets were then washed once with phosphate-buffered saline (PBS) and stained with propidium iodide. A total of 30,000 events were acquired by flow cytometer (BDC6, BD Biosciences, United States), and proportions of cells in each phase of the cell cycle were calculated using FlowJo software (BD Biosciences). For apoptosis analysis, after compound treatment for 24 h, cells were harvested and washed once with cold PBS. Then, cells were incubated for 15 min at room temperature with Annexin V-FITC-PI in a binding buffer. Cells were then analysed on a flow cytometer (BDC6, BD Biosciences) using FlowJo software. Results were expressed as percentages of Annexin V+ cells. Experiments were performed in duplicate.

As previously described, plasmid PSA-luc was a reporter gene plasmid in which the firefly luciferase expression is dependent on the PSA promoter; plasmid Renilla was a Ranilla luciferase reporter gene plasmid. These two plasmids were kindly provided by Dr Cen (Peking Union Medical College, China). LNCaP cells were seeded at a density of 6–7 × 10⁴ cells per well in 24-well plates. After incubation for 24 h, cells in each well were co-transfected with 100 ng of PSA-luc and 3 ng of Renilla plasmids using Lipofectamine 2000 reagent (Invitrogen) following the manufacturer’s protocol. After 24 h from the transfection, the medium was changed to phenol red-free RPMI-1640 supplemented with 10% charcoal-stripped FBS, containing 5 nM of DHT (1 µl) and 1 µl of test compounds at designated concentrations. After 24 h, the cells were lysed in 100 µl of passive lysis buffer per well, and 20 µl of the cell lysates were used for the detection of the luciferase activity using a dual-luciferase assay system (Promega, WI, United States) on a plate reader (Centro XS3 LB 960, Berthold Technologies, Germany). All experiments were run in triplicates.

LNCaP cells were treated by compounds at designated concentrations for 24 h. Total RNA was extracted from approximately 1 × 10⁶ cells using Invitrogen TRIzol^® RNA Isolation Reagent. Sequencing libraries were generated using VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina^® (NR604-01/02, CA, United States), following the manufacturer’s recommendations. Index codes were added to attribute sequences to each sample. Then, the library was examined successively. The clustering of the index-coded samples was performed on a cBot cluster generation system using HiSeq PE Cluster Kit v4-cBot-HS (Illumina, CA, United States) according to the manufacturer’s instructions. After cluster generation, the libraries were sequenced on an Illumina platform and 150 bp paired-end reads were generated. The passed raw reads were aligned and assembled by STAR (https://github.com/alexdobin/STAR). Differentially expressed genes were analysed by HTseq-count (https://pypi.org/project/HTSeq) and edgeR (https://bioconductor.org/packages/edgeR). The KEGG analysis was performed by EnrichR (https://cran.rproject.org/). Heatmaps were generated by Hiplot website (https://hiplot.com.cn/). Original data is available in The Genome Sequence Archive for Human (https://ngdc.cncb.ac.cn/gsa-human/; Accession code: HRA001434).

Cells were seeded at a density of ∼1 × 10⁵ cells per well in 6-well plates and incubated for 24 h. Then, DMSO or test compounds were added to each well at designated concentrations. After another 72 h of incubation, cell senescence was evaluated by visualising β-galactosidase activity using senescence β-galactosidase staining kit (C0602, Beyotime Technology, China) according to the manufacturer’s instructions. Three fields of each well were photographed for each of the three independent replicates for each treatment condition.

All animal experimental procedures were conducted in the animal facility of KeyGEN BioTECH (Nanjing, China). All experimental procedures involving the care and use of mice were approved by the KeyGEN BioTECH Institutional Animal Care and Use Committee. The mice were housed five per cage in an environmentally controlled SPF room (temperature 20–26°C; relative humidity 40–70%) on a 12-h light/dark cycle. The mice were fed commercial rodent chow (Beijing Keao Xieli Feed, Beijing, China) and received filter-purified water ad libitum. All studies utilised 4–6-week-old mice purchased from the Shanghai Laboratory Animal Center (Shanghai, China). 22Rv1 cells (2 × 10⁶ cells with Matrigel at a ratio of 1:1) were injected subcutaneously in the flank of the male SCID mice. SJSA-1 cells (2 × 10⁶ cells with Matrigel at a ratio of 1:1), NCI-H460 cells (2 × 10⁶ cells with Matrigel at a ratio of 1:1), or HepG2 cells (2 × 10⁶ cells with Matrigel at a ratio of 1:1) were injected subcutaneously in the flank of female BALB/c nude mice. When the average tumor volume reached 100–150 mm³, the mice were randomised and divided into the indicated groups (n = 8/group). XR-2 was administered once per day intraperitoneally. RG7388 and enzalutamide were administered once per day by oral gavage. Combination therapy studies were performed in a blinded manner. The tumor volume and mice bodyweight were monitored every other day, and the tumor volume was calculated according to the formula W² × L/2 (mm³), wherein W was the short diameter and L was the long diameter. Data are expressed as mean ± SD. Whole blood was collected 24 h later after the last administration from the orbit. Blood cell analysis was performed using a haematology analyser (XS-800I, Sysmex Corporation, Japan).

Antibodies to Ki67 (abcam, ab16667) and cleaved caspase 3 (CST, 9,661) were used for immunostaining on formalin-fixed paraffin-embedded xenograft tumor tissues. Generally, the rehydrated slides were microwave-heated for 20 min in citrate buffer (10-mM, pH 6.0) for antigen retrieval. Then incubated in 1% H₂O₂ for 10 min, after blocking with serum-free protein block, slides were incubated with the primary antibodies (Ki67, 1: 200; Cleaved caspase 3, 1: 100) for 2 h at room temperature, followed by incubation with HPR-conjugated secondary antibody for 1 h at room temperature. The immunoreaction products were visualized with 3, 3-diaminobenzidine (DAB)/H2O2 solution.

The MDM2 protein crystal structure (PDB code: 5TRF) was downloaded from the PDB database (https://www.rcsb.org). Chain A of the MDM2 crystal structure remained and was modified by the “protonate 3D” module of Discovery Studio 3.5. The 2D chemical structure of XR-2 drawn by the Chemdraw 18.1 software was generated in 3D structure using the “prepare ligands” module by Discovery Studio 3.5. As described previously, (Li et al., 2019), the binding site was centred at Ile99 in MDM2 with a radius of 10 Å to cover the binding pocket of MDM2. Then, the prepared XR-2 has docked into the MDM2 chain A binding site by the ‘CDOCKER’ module of Discovery Studio 3.5 by default. After analysing the 10 binding poses of XR-2 to MDM2, we selected the highest-ranked pose for the MDM2 structure as the binding model of XR-2.

2-(2-methoxyethoxy)ethyl-(2′S,3R,4′S,5′R)-5'-((4-carbamoyl-2-methoxyphenyl)carbamoyl)-6-chloro-4'-(3-chloro-2-fluorophenyl)-2′-neopentyl-2-oxospiro[indoline-3,3′-pyrrolidine]-1′-carboxylate (XR-1):Compound 2 was synthesized as previous explained. (Shu et al., 2013). Compound 2 (612 mg, 1.0 mmol) was dissolved in DMF (100 ml), then anhydrous KCO3 (4.0 mmol) was added and 1-chloroethyl (2- (2-methoxyethoxy) ethyl) carbonate (904 mg, 4.0 mmol) was added and stirred overnight at room temperature. After the reaction, water (20 ml) was added and extracted by ethyl acetate. The organic phase was separated, washed with brine, 1M hydrochloric acid and water successively, and then concentrated through the column to obtain the target product (644 mg, yield: 85%), which was a white solid. 1H NMR (CDCl3,400 MHz): 10.59 (s, 1H), 8.47 (d, J = 8.4 Hz, 1 H),7.80 (d, J = 1.6 Hz, 1 H), 7.55 (d, J = 1.3 Hz, 1 H), 7.48 (t, J = 6.8 Hz, 1 H), 7.28–7.36 (m, 2 H), 7.19–7.26 (m, 2 H), 7.04 (t, J = 8.0 Hz), 6.05–6.55 (bs, 1 H), 5.30–5.80 (bs, 1 H), 4.70 (t, J = 9.62 Hz, 1 H), 4.40–4.54 (m, 3 H), 3.94 (s, 3 H), 3.83 (t, J = 4.7 Hz, 2 H), 3.62–3.74 (m, 3 H), 3.60 (t, J = 4.6 Hz, 2 H), 1.24–1.35 (m, 1 H), 0.96 (s, 9 H); HRMS (ESI-TOF): m/z calculated for C37H41Cl2FN4NaO8+ [M + Na]+: 781.2183, found: 781.2190.

2-(2-(2-methoxyethoxy)ethoxy)ethyl-(2′S,3R,4′S,5′R)-5'-((4-carbamoyl-2-methoxyphenyl)carbamoyl)-6-chloro-4'-(3-chloro-2-fluorophenyl)-2′-neopentyl-2-oxospiro[indoline-3,3′-pyrrolidine]-1′-carboxylate (XR-2): Compound 2 (612 mg, 1.0 mmol) was dissolved in DMF (100 ml), then anhydrous KCO3 (4.0 mmol) was added with drops of 2- (2- (2-methoxyethoxy) ethane-1-ol (1.08 g, 4.0 mmol), and stirred overnight at room temperature. After the reaction, water (20 ml) was added and extracted by ethyl acetate. The organic phase was separated and washed with brine, 1M hydrochloric acid and water successively, and then concentrated through the column to obtain the target product (Yield: 80%), which was a white solid. 1H NMR (CDCl3,400 MHz): 10.6 (s, 1H), 8.45 (d, J = 8.0 Hz, 1H), 7.8 (s, 1 H), 7.55 (s, 1 H), 7.48 (t, J = 8.0 Hz, 1 H), 7.35 (t, J = 8.0 Hz, 2 H), 7.20–7.40 (m, 2 H), 7.04 (t, J = 8.0 Hz, 1 H), 6.1–6.6 (bs, 1 H), 5.4–5.8 (bs, 1 H), 4.69 (t, J = 10, 1 H), 4.40–4.55 (m, 3 H), 3.04 (s, 3 H), 3.83 (t, J = 4, 2 H), 3.69–3.77 (m, 4 H), 3.62–3.69 (m, 4 H), 3.50–3.58 (m, 2 H), 3.36 (s, 3 H), 3.18–3.30 (m, 1 H), 1.24–1.36 (m, 1 H), 0.95 (s, 9 H); 13C NMR (CDCl3,100 MHz): 174.41, 171.63, 168.82, 157.69, 155.21, 149.52, 148.46, 140.10, 134.91, 130.57, 130.01, 128.73, 127.35, 125.51, 124.81 (Jc-F = 4.3 Hz), 123.62, 123.36, 121.40 (d, J = 19 Hz), 119.95, 118.22, 116.15, 109.98, 72.01, 70.83, 70.81, 70.68, 68.61, 68.06, 66.60, 66.57, 65.51, 59.12, 55.76, 50.96, 42.79, 30.45, 29.89; HRMS (ESI-TOF): m/z calculated for C39H45Cl2FN4NaO9+ [M + Na]+: 825.2445, found: 825.2456.

2,5,8,11,14-pentaoxahexadecan-16-yl-(2′S,3R,4′S,5′R)-5'-((4-carbamoyl-2-methoxyphenyl)carbamoyl)-6-chloro-4'-(3-chloro-2-fluorophenyl)-2′-neopentyl-2-oxospiro[indoline-3,3′-pyrrolidine]-1′-carboxylate (XR-3): Compound 2 (306 mg, 0.5 mmol) was dissolved in DMF (50 ml), followed by the addition of anhydrous KCO3 (2.0 mmol) and the addition of 1-chloroethyl (2,5,8,11, 14-pentaxyhexadecane-16-yl) carbonate (720 mg, 2.0 mmol), stirred overnight at room temperature. After the reaction, water (20 ml) was added and extracted by ethyl acetate. The organic phase was separated, washed with brine, 1M hydrochloric acid and water successively, and then concentrated through the column to obtain the target product (Yield: 79%), which was a white solid. 1H NMR (CDCl3,400 MHz): 10.56 (s, 1 H), 8.41 (d, J = 8.4 Hz, 1 H), 7.79 (d, J = 1.7 Hz, 1 H), 7.54 (d, J = 1.3 Hz, 1 H), 7.49 (t, J = 6.8 Hz, 1 H), 7.20–7.40 (m, 4 H), 7.00–7.15 (m, 1 H), 6.30–6.60 (m, 1 H), 5.19 (bs, 1 H), 4.71 (t, J = 9.4 Hz, 1 H), 4.40–4.60 (m, 2 H), 4.25–2.34 (m, 1 H), 3.93 (s, 3 H), 3.83 (t, J = 4.6 Hz, 2 H), 3.78 (s, 1 H), 3.60–3.75 (m, 16 H), 3.50–3.58 (m, 2 H), 3.35–3.40 (m, 4 H), 1.20–1.37 (m, 1 H), 0.86–1.08 (m, 10 H); HRMS (ESI-TOF): m/z calculated for C43H53Cl2FN4NaO11 [M + Na]+: 913.2970, found: 913.2975.

Statistical analysis was performed using GraphPad Prism (version 5.01; GraphPad Software). Comparisons between groups were performed with a two-tailed Student’s t-test, except where specified, and differences were considered significant at p < 0.05.

To develop a more potently selective MDM2 inhibitor with acceptable toxicity, we synthesised some of PEGylation spirooxindole derivatives (PSD) based on previous reported spiroindolinone pyrrolidinecarboxamide compound 2 (Supplementary Figure S1). (Shu et al., 2013). Then, we evaluated the primary tumor inhibition activity of some of these compounds in SJSA1-derived xenografts. The results indicated that the 2-weeks treatment with the cis-cis isomer of PSD XR-2 (Figure 1A) led to the highest suppression rate of tumor progression, which was higher than treatment with its trans-cis isomer and both isomers mixture. XR-2 was also proven to be more potent than its non-PEGylation initial analogue compound 2 (Supplementary Figure S2). Therefore, we chose XR-2 for further investigation.

To predict the binding mode of XR-2 to the MDM2 protein, molecular docking was performed based on the protein structure of MDM2 (PDB code: 5TRF) using Discovery Studio software. We found that XR-2 could suitably locate into the p53 binding site of MDM2 and form hydrogen bond interactions with Leu54 and His96 (Figure 1B). The homogeneous time-resolved fluorescence binding assay proved that XR-2 directly inhibited MDM2–p53 interaction (Supplementary Figure S3). Mechanistically, through competitive binding to the p53 pocket in MDM2, an MDM2 inhibitor could block MDM2–p53 interaction and subsequently led to p53 accumulation and transcriptional activation in wild-type p53 cells. Herein, we analysed the activity and specificity of XR-2 to activate p53 in PCa cells. In LNCaP cells with wild-type p53, Western blot analysis results indicated that XR-2 significantly induced p53 protein accumulation in a dose-dependent manner. In addition, we detected p21 protein upregulation, which further proved the activation of the p53 pathway (Figure 1C). In 22Rv1 cells, another wild-type p53 cells, XR-2 induced similar p53 activation (Figure 1D). However, in p53-mutated DU145 cells, XR-2 did not influence the levels of both p53 and p21 (Figure 1E). Moreover, quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) analysis revealed that XR-2 caused a dose-dependent upregulation of p53 downstream target genes p21, PUMA and MDM2 but not p53 mRNA expression levels in both LNCaP and 22Rv1 cells (Figures 1F,G). Not surprisingly, XR-2 could not increase p21, PUMA and p53 mRNA levels in DU145 cells (Figure 1H). Notably, XR-2 demonstrated comparable p53 induction activity with the well-known MDM2 inhibitor RG7388 (Figure 1I). Taken together, our data provide clear evidence that XR-2 is a potent and specific MDM2 inhibitor.

The p53 key downstream target genes p21 and PUMA could regulate cell cycle arrest, apoptosis and senescence in various cancer cells. Subsequently, we investigated the CRPC inhibition activity of XR-2; as shown in Figure 2A, XR-2 selectively inhibited the proliferation of wild-type p53 LNCaP and 22Rv1 cells in a dose-dependent manner with IC₅₀ of 0.073 and 0.74 μM, respectively. Conversely, p53-mutated DU145 cells were quite less sensitive to XR-2. Furthermore, to estimate the effects of XR-2 on CRPC cell clonogenic activity, we exposed LNCaP, 22Rv1 and DU145 cells to different concentrations of XR-2 for about 2 weeks. As a result, XR-2 strongly decreased the number of LNCaP and 22Rv1 cell colonies in a dose-dependent manner compared with the untreated group, while specific treatment concentrations did not influence the number of DU145 cell colonies (Figure 2B).

Furthermore, flow cytometry analysis revealed that XR-2 therapy selectively increased the proportion of G2/M phase cells than the control group in LNCaP cells rather than in DU145 cells (Figures 2C,D). To further confirm the mechanisms of cell cycle arrest, we evaluated the influence of XR-2 on the protein levels of different cell cycle-related proteins in LNCaP, 22Rv1 and DU145 cells. XR-2 increased CDK suppressor p21 protein levels as mentioned above in LNCaP and 22Rv1 cells and thus dose-dependently reduced the cell cycle regulated protein levels of phosphorylation of CDK1 and CDK2 (Figures 2E,F). In DU145 cells, XR-2 could not upregulate p21 protein; thus, it cannot reduce the protein levels of activated phosphorylation CDK1 and CDK2 (Figure 2G). Collectively, these findings demonstrated that XR-2 effectively inhibited cell proliferation and induced cell cycle arrest of CRPC cells by activating the p53 pathway.

Moreover, we examined whether XR-2 influences CRPC cell apoptosis by flow cytometry assay. As shown in Figures 3A,B, XR-2 significantly induced apoptosis of LNCaP cells rather than of DU145 cells. Western blot analysis further proved that XR-2 promoted the accumulation of apoptotic marker cleaved PARP protein in LNCaP and 22Rv1 cells (Figures 3C,D). The mRNA levels of apoptotic markers Bax and growth arrest and DNA damage-inducible 45 (GADD45A) were also upregulated in LNCaP and 22Rv1 cells rather than in DU145 cells (Supplementary Figure S4A,B). As we confirmed that XR-2 increased the expression of the senescence marker p21, we examined the mechanism of cell senescence by detecting senescence-associated β-galactosidase. We observed a significant increase in β-galactosidase staining in LNCaP and 22Rv1 cells treated with XR-2 for 72 h compared with vehicle treatment.

To further confirm whether XR-2 inhibits CRPC cell viability through p53 activation, we performed TP53 gene knockdown assay by siRNA. The results demonstrated that p53 knockdown reduced p53 mRNA expression levels in 22Rv1 cells (Figure 4A). Not surprisingly, XR-2 therapy was unable to promote p53 accumulation in p53 knockdown cells (Figure 4B). Efficient knockdown of p53 in LNCaP and 22Rv1 cells also potently attenuated XR-2 induced upregulation of p21 and PUMA mRNA expression levels (Figures 4C,D). Furthermore, p53 knockdown forcefully reduced the cell proliferation inhibition potency of XR-2 in LNCaP and 22Rv1 cells (Figures 4E,F). We also found that XR-2-induced colony formation inhibition activity in 22Rv1 cells was blocked by p53 knockdown (Figure 4G). Finally, we found that p53 knockdown diminished XR-2-induced cell apoptosis indicated by the expression levels of the cleaved PARP protein in LNCaP and 22Rv1 cells (Figures 4H,I). These data firmly established that XR-2 inhibits CRPC cell viability through the p53 pathway.

As p53 overexpression was reported to diminish the androgen response, (Cronauer et al., 2004), we investigated whether p53 inducer XR-2 could block the AR pathway in PCa cells. The results of dual-luciferase reporter assay in LNCaP cells indicated that XR-2 inhibited dihydrotestosterone (DHT)-induced transcriptional activities of endogenous AR in a dose-dependent manner (Figure 5A). Moreover, Western blot analysis revealed that XR-2 downregulated DHT-activated PSA protein expression levels in a dose-dependent manner; surprisingly, XR-2 could also downregulate AR protein levels (Figure 5B).

RNA sequencing assay was performed to evaluate the gene expression status of LNCaP cells under XR-2 and enzalutamide treatment. Compared with single agents, XR-2 plus enzalutamide in LNCaP cells resulted in more differentially expressed genes, implicating that combined treatment had more remarkable effects (Figure 5C). The Venn analysis indicated that the largest number of differentially upregulated and downregulated genes appeared to be induced by XR-2 plus enzalutamide compared with either single-agent treatment. Interestingly, the proportion of overlapped genes between XR-2 monotherapy and combination therapy appeared higher than that between enzalutamide monotherapy and combination therapy (Supplementary Figure S5A,B). The Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis demonstrated that XR-2 plus enzalutamide upregulated the expression of genes encoding constituents of the P53 signalling pathway and downregulated those involved in DNA replication, mismatch repair and cell cycle progression (Figure 5D). As shown in Figure 5E, the main p53-target genes were upregulated in XR-2 monotherapy and combination therapy, such as growth differentiation factor 15 (GDF15), which is a surrogate for p53 activation, and apoptogenic genes BBC3, PUMA, MDM2, Bax and FAS. XR-2 monotherapy and combination therapy also upregulated the expressions of negative regulators of cell cycle progression genes, including cyclin-dependent kinase inhibitor 1A and GADD45. In addition, the combination therapy enhanced AR pathway inhibition compared with the control and enzalutamide groups, which were characterised by lower expression levels of key AR-target genes including FKBP5, PMEPA1, KLK3, NKX3-1 and KLK2 (Supplementary Figure S5C). Altogether, these findings suggest that XR-2 plus enzalutamide induced enhanced p53 pathway reactivation and AR pathway inhibition.

To further confirm the synergistic effects between XR-2 and enzalutamide, we examined the influences of XR-2 plus enzalutamide on AR and p53 pathways. We found that XR-2 plus enzalutamide demonstrated enhanced AR transcriptional inhibition activity than either monotherapy in different concentrations (Figure 5F). Similarly, as demonstrated in Figure 5G, the XR-2 plus enzalutamide reduced AR protein levels more efficiently than monotherapy in both LNCaP, 22Rv1 and C4-2 cells (Supplementary Figure S5D,F). By contrast, by detecting the p21 mRNA levels in LNCaP, 22Rv1 and C4-2 cells, we found that enzalutamide could enhance XR-2-induced p53 pathway activation (Supplementary Figure S5G,H,E). Together, these findings indicate that XR-2 could inhibit the AR pathway, and XR-2 plus enzalutamide synergistically inhibits AR and p53 pathways.

To evaluate whether XR-2-induced p53 reactivation and AR inhibition could inhibit enzalutamide-resistant CRPC cell viability, we constructed enzalutamide-resistant LNCaP (LNCaP-R) cells through long-term enzalutamide treatment (Figure 6A). Compared with LNCaP cells, LNCaP-R cells demonstrated a more robust cell proliferation activity with 5 μM enzalutamide treatment (Figure 6B). Nevertheless, XR-2 potently blocked LNCaP-R cell proliferation and colony formation activity in a dose-dependent manner (Figures 6C,D). In enzalutamide-resistant 22Rv1 cells, XR-2 similarly suppressed the proliferation of 22Rv1 cells and demonstrated synergistic effects with enzalutamide (Figures 6E,F). Western blot analysis revealed that XR-2 also induced apoptosis marker protein cleaved PARP1 expression and demonstrated synergistic effects with enzalutamide in LNCaP-R cells (Figure 6G). Subsequently, we found that the proportion of β-galactosidase-positive cells were significantly increased in both XR-2 monotherapy and combination therapy compared with that in the control or enzalutamide group. These results proved that XR-2 could overcome enzalutamide resistance in vitro.

As previous study proved that XR-2 could activate the p53 pathway and inhibit PCa cell viability in vitro. Then, we investigated the tumor inhibition efficacy of XR-2, enzalutamide and XR-2 plus enzalutamide in vivo by using 22Rv1 xenografts in male SCID mice. In this study, 5 ×10⁶ 22Rv1 cells were injected into the left flank of each male mouse, and the tumor volume was allowed to increase to approximately 100 mm³. Then, the tumor-bearing mice received intraperitoneal injection of the vehicle control, 30 mg/kg of XR-2, 30 mg/kg of enzalutamide and 30 mg/kg of XR-2 plus 30 mg/kg of enzalutamide once a day for 4 weeks. The results indicated that enzalutamide could only inhibit 22Rv1 tumor growth weakly, mainly due to ARV7 overexpression, while both XR-2 monotherapy and combination therapy suppressed 22Rv1 tumor progression and decreased tumor weight significantly (Supplementary Figure S6A, S7A,B). Interestingly, the combination therapy revealed a more effective tumor inhibition activity than XR-2 or enzalutamide monotherapy. We also evaluated whether XR-2 treatment activated p53 protein levels in vivo; as shown in Supplementary Figure S6B, the tumor p53 protein levels were remarkably accumulated in both XR-2 monotherapy and combination therapy groups. Moreover, XR-2 was tolerable to the SCID mice, as the mice bodyweight of the XR-2 monotherapy and combination therapy groups nearly had no change compared with the vehicle group (Figure 7C). Further analysis revealed that both XR-2 monotherapy and combination therapy did not influence the weight of the critical organs and blood neutrophil counts of mice, which reminded the favourable in vivo safety of XR-2 (Figures 7D,E). Moreover, immunohistochemistry assay results proved that XR-2 reduced Ki-67 protein levels and induced cleaved caspase3 protein accumulation compared with the vehicle group. Compared with monotherapy, the combination therapy enhanced these changes in protein levels (Figure 7F). Taken together, our data indicated that XR-2 could inhibit the growth of CRPC both in vitro and in vivo.