LNCaP, PC3, 22Rv1, C4-2, C4-2B, Du145 and 293T cells were obtained from the American Type Culture Collection. These cells were cultured in RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.) medium supplemented with 10% fetal bovine serum (FBS; Clark Bioscience) at 37°C in a humidified atmosphere with 5% CO₂. LNCaP95 cells were provided by Dr Alan Meeker at the Johns Hopkins University (Baltimore, MD, USA) and were cultured in RPMI-1640 medium supplemented with 10% charcoal-stripped FBS (Gibco; Thermo Fisher Scientific, Inc.). Erastin, docetaxel (DTX), ferrostatin-1, liproxstatin-1, ZVAD-FMK, necrosulfonamide and chloroquine were purchased from Selleck Chemicals.

After being treated with erastin at different concentrations (0, 2.5, 5.0, 10.0, 20.0 or 40.0 µM) for 24, 48 or 72 h, cell monolayers of 22Rv1 or LNCaP95 in 96-well plates were fixed using 20% (wt/vol) trichloroacetic acid (100 µl/well) at room temperature for at least 3 h, and then stained for 30 min using SRB (Shanghai YuanYe Biotechnology Co., Ltd.) at room temperature. Excess dye was removed by washing repeatedly with 1% (vol/vol) acetic acid. The protein-bound dye was dissolved in 10 mM Tris base solution and then the OD was measured at 565 nm using a Multiscan Spectrum spectrophotometer. In order to obtain more samples in the subsequent experiments, the doses of erastin were determined based on the cell inhibition rate of 30%. Thus, 22Rv1 and LNCaP95 cells were treated with 10 or 20 µM erastin, while PC3 cells were treated with 1 µM erastin.

In order to verify whether the effect caused by erastin was ferroptosis, the cells were treated with erastin (20 µM). Meanwhile, the cells were added with or without ferroptosis inhibitors (1 µM ferrostatin-1 or 1 µM liproxstatin-1), apoptosis inhibitor (10 µM ZVAD-FMK), necroptosis inhibitor (1 µM necrosulfonamide), or autophagy inhibitor (25 µM chloroquine). After treatment for 48 h, the cell viability was detected by SRB as described above.

In order to calculate the combination index (CI) between DTX and erastin, 22Rv1 and LNCaP95 cells were treated with these two drugs at different concentrations (0, 2.5, 5.0 or 10.0 µM erastin and 0, 5, 10 nM DTX). After treatment for 48 h, cell viability was detected by SRB as described above.

Western blotting was conducted as previously described (19), and proteins were visualized using an Odyssey Infrared Imaging System (LI-COR Biosciences), following the manufacturer's protocols. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used for normalization of the densitometry signals. The following antibodies were used in this study: Anti-N-terminal AR antibody (cat. no. 5153; 1:1,000; Cell Signaling Technology, Inc.), anti-phospholipid hydroperoxide glutathione peroxidase (GPX4; cat. no. ab125066; 1:1,000; Abcam) and anti-GAPDH (cat. no. A00227; 1:1,000; Wuhan Boster Biological Technology, Ltd.). IRDye^® 800CW goat anti-human IgG (H+L) (cat. no. 926-32232; 1:10,000; LI-COR Biosciences).

After treatment with erastin for 24 h, the cells were incubated with 1 µl DCFH-DA with 1 ml phosphate-buffered saline (PBS) for 1 h at 37°C in the dark to assess the cytosolic ROS levels. Samples were centrifuged at 860 × g at room temperature for 3 min, and the pellets were resuspended in 1 ml PBS. Measurements were performed on a FACSCalibur™ (BD Biosciences) flow cytometer using FlowJo software 7.6 (FlowJo LLC). All experimental results are reported as represented by three replicates.

The GSH and malondialdehyde (MDA) contents in cell lysates were assessed using GSH (cat. no. A061-1-2) and lipid peroxidation (cat. no. A003-4-1) assay kits, respectively, according to the manufacturer's instructions. Both kits were purchased from Nanjing Jiancheng Bioengineering Institute.

Three luciferase reporter plasmids were used in this research, and transfection was performed with TurboFect Transfection Reagent (cat. no. R0531; Thermo Fisher Scientific, Inc.). The androgen-responsive element-luciferase plasmid contains three repeat ARE regions ligated in tandem to a luciferase reporter (ARR3-Luc), which was provided by Dr Robert Matusik at Vanderbilt University School of Medicine (Nashville, TN, USA), and was used to reflect the AR-FL trans-activation activity. The ubiquitin-conjugating enzyme E2C-luciferase plasmid (UBE2C-luc) is driven by a minimal promoter and three repeats of an AR-V-specific promoter element. Thus, it was used to reflect AR-V7 trans-activation activity. pGL4-ARpro8.0 is driven by an 8.0 kb fragment of the 5′-flanking region of the human AR gene. The transfected cells, including 22Rv1, LNCaP95, PC3 and LNCaP cells, were divided equally into 24-well plates (1×10⁵ cells/well) and cultivated in serum-free medium for 24 h. Subsequently, the cells were exposed to charcoal-stripped FBS with or without 1 nM R1881 (Sigma-Aldrich; Merck KGaA) at 37°C, which is a type of synthetic androgen, and erastin to detect AR-FL or AR-V7 activity. After 24 h, the cells were lysed with 100 µl reporter lysis buffer (Promega Corporation), and the luciferase activity was assayed using the Luciferase Assay System (Promega Corporation) and normalized based on the protein concentrations for each sample.

The total RNA of cells was extracted and collected using an E.Z.N.A.^® Total RNA Kit I (Omega Bio-Tek, Inc.) and quantified with a NanoDrop™ spectrophotometer (Thermo Fisher Scientific, Inc.). Reverse transcription was performed with an RNA reverse transcription kit (Takara Bio, Inc.), and qPCR was performed using TransStart^® Green qPCR SuperMix (TransGen Biotech Co., Ltd.), according to the manufacturer's instructions for 45 cycles of 5 sec at 94°C and 30 sec at 60°C. 36B4 served as the normalization control for these qPCR assays. The following primers were used: AR-FL sense, 5′-GTACAGCCAGTGTGTCCGAA-3′ and anti-sense, 5′-TTGGTGAGCTGGTAGAAGCG-3′; AR-V7 sense, 5′-AAAAGAGCCGCTGAAGGGAA-3′ and anti-sense, 5′-GCCAACCCGGAATTTTTCTCC-3′; PSA sense, 5′-CTCAGGCCAGGTGATGACTC-3′ and anti-sense, 5′-GTCCAGCACACAGCATGAAC-3′; transmembrane protease serine 2 (TMPRSS2) sense, 5′-ACACACCGATTCTCGTCCT-3′ and anti-sense, 5′-TGGCCTACTCTGGAAGTTCA-3′; UBE2C sense, 5′-TTCCCCAGTGGCTACCCTTA-3′ and anti-sense, 5′-CAGGGCAGACCACTTTTCCT-3′; transcription factor E2F7 (E2F7) sense, 5′-TTCTGTTGCTCAGACGGACC-3′ and anti-sense, 5′-ATCCCTCTCTGACCCTGACC-3′; and 36B4 sense, 5′-CGACCTGGAAGTCCAACTAC-3′ and anti-sense, 5′-ATCTGCTGCATCTGCTTG-3′. mRNA expression levels were quantified using the 2^−ΔΔCq method (20).

Three plasmids were used as lentiviral vectors: VSVG, ∆8.2, pLVX-AR-FL or pLVX-AR-V7. VSVG, ∆8.2, and oLVX were provided by the National Engineering Laboratory for AIDS Vaccine (Jilin University, Changchun, China). The preparation of the lentiviral particles and lentiviral infections were performed as previously described (21). When lentiviral particles (collected at 48 h after transfection) were used to infect 22Rv1 cells in 6-well plates (1×10⁶ cells/well), polybrene was added at a final concentration of 6 µg/ml, fresh medium was replaced within 6 h, and the protein expression was determined by western blotting.

A total number of 12 male BALB/c nude mice (age, 6–8 weeks old; weight, 15–20 g) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All mice were housed at a constant temperature and constant humidity in a specific pathogen-free environment with free access to food and water. The present study was approved by the Animal Ethics Committee of Basic Medical College of Jilin University (approval no. 2016045). After 1 week of adaptation, 5×10⁶ 22Rv1 cells suspended in 50% Matrigel and PBS were injected subcutaneously into the right dorsal flank. Tumor formation was strictly monitored, and tumor volume was calculated by the modified ellipsoidal formula: Tumor volume = 0.52 × length × width². When the tumor size reached ~50 mm³, the mice were randomly allocated into two groups (n=6/per group) and treated with or without erastin. The dose of erastin administered to the mice was 20 mg/kg, which was injected intraperitoneally twice every other day. After being treated for 2 weeks, all mice were injected with pentobarbital sodium at a dose of 20 mg/kg intraperitoneally. Then, 0.5 ml mouse blood was collected from the orbital vein to detect serum prostate-specific antigen (PSA) levels using a human KLK3 ELISA Kit (cat. no. KIT10771; Sino Biological, Inc.), tumors were removed for molecular analysis, and some organs were removed for a safety evaluation at the end of the experiment. Finally, all mice were sacrificed by intraperitoneal injection with pentobarbital sodium at a dose of 150 mg/kg. Erastin was dissolved in 3.3% dimethyl sulfoxide (DMSO) and 96.7% β-cyclodextrin (20%).

Hematoxylin and eosin (H&E) staining was performed as described in a previous study (22). Briefly, the tissues of the rats were fixed in 10% buffered formalin, and then decalcified in 10% ethylenediaminetetraacetic acid. The tissues were then treated with ethanol and xylene for dehydration. After being embedded in paraffin and sliced, several 4-µm thick histological slices were stained with H&E. The images were subsequently acquired using a light microscope (BX51; Olympus Corporation) at ×200 magnification.

IHC staining was performed as described previously (23). Briefly, histological slices from tumor tissues were stained with anti-N-terminal AR antibody (1:200; cat. no. 5153; Cell Signaling Technology, Inc.). Histological images were captured by a microscope (BX51; Olympus Corporation) with an objective magnification of ×200.

The experiments were repeated three times independently. Statistical analysis was performed for multiple comparisons using one factor analysis of variance (ANOVA) followed by Dunnett's or Bonferroni's post hoc tests. All data were analyzed with the statistical software SPSS 11.0 (SPSS, Inc.), and the results are expressed as the mean ± SD. P<0.05 was considered to indicate a statistically significant difference. SPSS software was also used to calculate the CI, a CI value of >1, 1 and <1 denotes antagonism, additivity and synergism, respectively.

Western blotting was employed to detect the expression of AR, AR-Vs and GPX4 in several prostate cancer cell lines including LNCaP95, LNCaP, Du145, C4-2B, 22Rv1, C4-2 and PC3 cells. The results showed that both 22Rv1 and LNCaP95 cell lines expressed AR, AR-Vs and GPX4 (Fig. 1A). Conversely, PC3 cells did not express AR or AR-Vs and only showed a low expression of GPX4. Thus, 22Rv1, LNCaP95 and PC3 cell lines were chosen for the subsequent experiments.

The effect of erastin on the proliferation of 22Rv1 and LNCaP95 cells was detected by the SRB assay. As presented in Fig. 1B and C, erastin inhibited the proliferation of these two CRPC cell lines in a dose-dependent manner. Then, PI flow cytometry was employed to examine the effect of erastin on the cell cycle. The results showed that erastin could increase the proportion of cells in sub-G1 phase and cause cycle arrest (Fig. 1D-F).

To verify that the effect caused by erastin is ferroptosis, the cells were treated with erastin with or without ferroptosis inhibitors (ferrostatin-1 and liproxstatin-1), apoptosis inhibitor (ZVAD-FMK), necroptosis inhibitor (necrosulfonamide) and autophagy inhibitor (chloroquine). The SRB results showed that in both cell lines, ZVAD-FMK, necrosulfonamide and chloroquine had no effect on erastin-induced cell death (Fig. 2A and B). The death of the 22Rv1 cells induced by erastin was only reversed by ferrostatin-1 and liproxstatin-1.

Then, the expression of the ferroptosis marker protein GPX4 was measured. The results showed that erastin downregulated the protein expression of GPX4 in both cell lines (Fig. 2C and D). Given that ferroptosis is characterized by lipid peroxidation, the level of GSH, a key regulator that maintains cellular redox homeostasis, was investigated. The results showed that ROS levels in both cell lines were increased with erastin treatment compared with the control group (Fig. 2E and F). Moreover, GSH levels were downregulated after erastin treatment in these two cell lines (Fig. 2G and H). MDA, an end product of lipid peroxidation, was notably increased following erastin treatment, as expected (Fig. 2I and J). According to the aforementioned results, it was hypothesized that ferroptosis was initiated in these two cell lines after erastin treatment.

Given that AR-FL and AR-Vs play important roles in the development of CRPC, 22Rv1 and LNCaP95 CRPC cells that express AR-FL and AR-Vs were chosen in the present study to explore whether erastin can inhibit AR expression. In 22Rv1 cells, AR-Vs include AR-V7 (also called AR3), AR-V1 (also called AR4) and AR-V4 (also called AR5) (8,24), whereas LNCaP95 cells express only AR-V7 (25). Among these splice variants, AR-V7 has been found to be associated with the development of CRPC and is recognized as a biomarker of poor prognosis for patients with CRPC (26); thus, AR-V7 was selected as the representative of AR-Vs in the present study. The expression of AR-FL and AR-V proteins was detected by western blotting, and the results showed that erastin downregulated both AR-FL and AR-V protein expression levels (Fig. 3A and B).

To investigate how erastin decreases the protein levels of AR-FL and AR-V, AR-FL and AR-V7 mRNA expression levels were measured by RT-qPCR. As expected, erastin significantly reduced the levels of AR-FL and AR-V7 mRNA (Fig. 3C and D). Then, AR promoter activity was detected using dual-luciferase reporter plasmid pGL4-ARpro8.0. Erastin treatment resulted in the significant inhibition of AR promoter activity in both cell lines (Fig. 3E and F). Taken together, the data indicated that erastin had the ability to inhibit the transcription of the AR gene in CRPC cells.

To evaluate the effects of erastin on AR transactivation, AR-FL and AR-V transcriptional activity and their target genes were measured by a reporter gene assay and RT-qPCR, respectively. First, 22Rv1 cells were transfected with the ARR3-luc luciferase construct, which contained three tandem repeats of androgen response elements. The results showed that erastin treatment induced a reduction in luciferase activity in 22Rv1 cells (Fig. 4A). Considering that the ARR3-luc construct can be regulated not only by AR-FL, but also by AR-Vs, the LNCaP cell line was selected for transfection with a ARR3-luc luciferase construct because these cells only express AR-FL. The results in LNCaP cells clearly showed that erastin inhibited AR-FL trans-activating activity (Fig. 4B). To specifically assess the effect of erastin on AR-V transcriptional activity, 22Rv1 and LNCaP95 cells were transfected with the UBE2C-luc construct in which the luciferase gene is driven by an AR-V-specific promoter element of the UBE2C gene (27). Most of the AR-Vs identified to date display constitutive activity, when AR-V transcriptional activity was measured (28); therefore, these cells were cultured with 10% charcoal-stripped FBS. As shown in Fig. 4C and D, erastin caused the effective inhibition of AR-V trans-activating activity.

Considering that erastin had a significant inhibitory effect on endogenous AR-FL and AR-V transcriptional activity, it was speculated that erastin had an inhibitory effect on exogenous AR activity. To test this hypothesis, the effect of erastin on exogenously expressed AR-FL and AR-V7 trans-activating activity was evaluated in the PC-3 cell line (null-AR). The results showed that PC3 cells were significantly inhibited and the exogenous AR-FL and AR-V trans-activating activity was evidently decreased after the treatment of 1 µM erastin (Fig. 4E-G). The exogenous AR-V7 trans-activating activity was further tested in LNCaP cells (without AR-Vs), and the results were consistent with those of the PC3 cells (Fig. 4H).

Consistently, the mRNA levels of the AR-FL target genes PSA and TMPRSS2 and the AR-V-specific target genes UBE2C and E2F7 were significantly downregulated by erastin in both the 22Rv1 and LNCaP95 cell lines (Fig. 4I-L). Collectively, these findings indicated that erastin could downregulate AR-FL and AR-V trans-activation.

To further test the importance of the roles of AR-FL and AR-V in the action of erastin in CRPC cells, AR-FL and AR-V7 were overexpressed in the 22Rv1 cell line using lentiviral infections that delivered AR-FL and AR-V7 RNA expression constructs, and growth inhibition was evaluated using the SRB method. As shown in Fig. 5A and B, the expression of AR-FL and AR-V7 proteins in the 22Rv1 cells was significantly upregulated, especially the AR-FL protein. The overexpression of AR-FL and AR-V7 significantly promoted the resistance to high concentrations of erastin after 48 and 72 h compared with controls in the 22Rv1 cells. However, there was no significant difference between the overexpression of AR-FL and AR-V7 protein (Fig. 5C and D).

To investigate whether erastin inhibits the tumor growth in vivo, 22Rv1 cells were implanted into the subcutaneous space of immune-deficient nude mice. The tumor growth curve results are shown in Fig. 6A and revealed that erastin inhibited the growth of 22Rv1 tumors, with significance differences found on day 7 of the treatment. At the end of the experiments, the average tumor weight in the control group was 0.80±0.11 g, while that in the erastin-treated group was 0.55±0.17 g (Fig. 6B-D). The results of H&E staining of the tumor tissues showed that the nuclear/cytoplasmic ratio of tumor cells was notably decreased in the erastin group (Fig. 6E). Given that serum PSA concentration is one of most important clinical indexes for the diagnosis of prostate cancer, serum PSA levels were measured by ELISA, and a significant reduction in PSA serum levels in response to erastin treatment was observed (Fig. 6F). Then, AR protein levels were measured using IHC in tumor tissues. In erastin-treated tumor specimens, AR was obviously downregulated (Fig. 6G). In addition, the protein and mRNA levels of AR-FL and AR-V in the tumor tissues were measured by western blotting and RT-qPCR. Both the protein levels and mRNA levels of AR-FL and AR-V7 in the tumor tissues were decreased (Fig. 6H and I), which was consistent with the results obtained in vitro.

To evaluate the toxicity of erastin in vivo, the body weights of mice were measured and mouse organs were collected, including the heart, liver, spleen and kidney, to observe changes in morphology by H&E staining. The mice appeared to tolerate erastin well, and neither a significant difference in body weight (Fig. 6J) nor noticeable organ damage (Fig. 6K) was detected between the treatment group and the control group.

Docetaxel, the standard therapy for CRPC, represents the only class of chemotherapy drugs that prolongs the survival of patients with CRPC. Previous studies have reported that docetaxel induces microtubule stabilization and abrogates AR nuclear translocation and transcriptional activity (29–31). However, microtubule stabilization has been found to have no effect on AR-Vs, especially AR-V7, with neither subcellular localization nor nuclear activity affected, indicating one of the mechanisms of docetaxel resistance (29,32). Given that erastin has the ability to downregulate AR-V expression and activity and that the underlying mechanism by which docetaxel inhibits AR-FL is different from that of erastin, it was hypothesized in the present study that erastin can enhance the efficacy of docetaxel in CRPC. To test this hypothesis, the growth of 22Rv1 and LNCaP95 cells was measured after treatment with erastin and docetaxel, and the combined index values were calculated. All the combinations produced a CI value <1, suggesting synergy between erastin and docetaxel in inhibiting cell growth (Tables I and II). When 5 µM erastin and 10 nM docetaxel were added to 22Rv1 and LNCaP95 cells alone or in combination, as shown in Fig. 7A and B, combination therapy inhibited tumor cell growth to a significantly greater extent than monotherapy. These data provided preliminary support for using erastin to enhance docetaxel efficacy in CRPC.