Synergistic Efficacy of Gedatolisib and Darolutamide in Prostate Cancer to Overcome Resistance to Androgen-Targeted Therapy

2.1. Effects of Gedatolisib in Combination with Darolutamide on Cell Growth

To test the effects of gedatolisib in combination with darolutamide on PC cell growth, we used established PC cell lines with different AR and PTEN status (Figure 2A). These cell lines included VCaP and LNCaP, which are widely used as models of AR+, androgen-responsive PC; 22Rv1 and the LNCaP-derived C4-2 cell line, which are used as models of AR+ CRPC; and PC3 and Du145, which do not express AR and are insensitive to androgen [33,34,35]. When grown in medium supplemented with charcoal-stripped serum (i.e., under androgen deprivation conditions), these cell lines showed variable responses to physiological concentrations of dihydrotestosterone (DHT, 1 nM), which significantly promoted cell growth only in VCaP (AR+, PTEN+) and LNCaP (AR+, PTEN−) (Figure S1). To ensure consistent androgen concentration in the medium and evaluate the effects of anti-androgens, all the subsequent cell growth assays were performed in androgen-depleted growth medium supplemented with 1 nM DHT.

The cell lines were first analyzed for their response to increasing concentrations of gedatolisib and darolutamide by calculating growth rate (GR) inhibition from cell viability assessments before and after drug treatment. As we observed in a previous study [30], gedatolisib inhibited cell growth with a GR50 < 40 nM for all cell lines and exerted cytotoxic effects (GR < 0) in all cell lines at the maximum concentration tested (Figure 2A,B). Darolutamide inhibited cell growth in VCaP (AR+, PTEN+) and LNCaP (AR+, PTEN−) cells with GR50 values of 372 and 2539 nM, respectively, and exerted a cytotoxic effect only in VCaP cells (GR = −0.37 at 10 µM). The C4-2 cells (AR+, PTEN−) were less sensitive to darolutamide with a GR50 > 3333 nM and an incomplete dose response curve up to 10 µM. All the other cell lines were completely insensitive to darolutamide up to 10 µM (Figure 2A,B). While gedatolisib growth-inhibitory effects did not correlate with the AR or PTEN status of the cell lines, darolutamide was only effective in AR+ cell lines, regardless of their PTEN status.

Next, the PC cell lines were tested for their growth response to gedatolisib combined with darolutamide. As shown in Figure 2C, the combination of darolutamide and gedatolisib induced greater growth inhibition compared to the single agents in four AR+ cell lines (VCaP, LNCaP, C4-2, and 22Rv1). Interestingly, 22Rv1, which are resistant to darolutamide as a single agent, showed significantly greater growth-inhibition in response to gedatolisib + darolutamide compared to gedatolisib alone, suggesting partial restoration of the darolutamide response. The combination of gedatolisib and darolutamide showed no or very modest benefits in MDA-PCa-2b (AR+), PC3 (AR−) and Du145 (AR−). Synergy analysis with the Chou and Talalay method indicated that gedatolisib and darolutamide inhibited cell growth synergistically (CI < 1) in VCaP, LNCaP, and 22Rv1 (Figure 2D). Synergy was further confirmed in VCaP cells by using the MuSyC method (Figure 2E). Similar results were observed when gedatolisib was combined with enzalutamide, another AR inhibitor (Figure S2).

These results demonstrated that the combination of gedatolisib and darolutamide was more effective than the single agents at inhibiting the cell growth of most AR+ PC cell lines, regardless of their PTEN status. We next focused on the AR+ PC cells to evaluate the effects of the gedatolisib/darolutamide combination on PAM and AR pathway activities as well as cellular functions controlled by these pathways.

2.2. Effects of the Gedatolisib/Darolutamide Combination on PAM Pathway and AR Pathway Activities

The AR and PAM signaling pathways are linked by reciprocal feedback loops, whereby inhibition of the AR pathway can promote PAM activity and vice versa, resulting in reduced single agent efficacy in PTEN-deficient PC cells [14,18]. We tested the effects of the gedatolisib/darolutamide combination on AR and PAM pathway activities in AR+ PC cell lines with or without PTEN loss.

AR pathway activity was assessed by analyzing the transcription of two well-known AR-target genes, TMPRSS2 and KLK3 (encoding for the prostate-specific antigen, PSA). As expected, 24 h treatment with 1 nM DHT increased the mRNA levels of both genes in all cell lines analyzed, and co-treatment with darolutamide (10 µM) prevented induction by DHT (see Table S5). In the presence of DHT, gedatolisib (111 nM) increased the mRNA levels of both TMPRSS2 and KLK3 in PTEN− cell lines (LNCaP, C4-2, red in heat map), while having no or opposite effects in PTEN+ cell lines (VCaP, MDA-PCa-2b, 22Rv1, white or blue in heat map) (Figure 3A). The increased AR signaling in LNCaP, C4-2 cells treated with gedatolisib is consistent with previous studies reporting that AR signaling is suppressed in PTEN− PC cells and can be reactivated upon inhibition of the PAM pathway [14,18,36]. The combination of gedatolisib and darolutamide reduced the mRNA levels of both genes in all cell lines, regardless of their PTEN status (Figure 3A).

PAM pathway activity was assessed by analyzing the phosphorylation of RPS6 and 4EBP1, two key PAM pathway effectors (Figure 1). Gedatolisib treatment started to decrease pRPS6 and p4EBP1 levels after 24 h, leading to reductions of up to 79% and 65%, respectively, by 72 h in all cell lines (Figure 3B). The addition of darolutamide to gedatolisib did not affect pRPS6 and p4EBP1 levels compared to gedatolisib alone (Figure 3B). The levels of pRPS6 and p4EBP1 were not affected, or very modestly affected, after 24 h treatment with darolutamide, and showed a cell-line dependent increase after 72 h. This increase was counteracted by co-treatment with gedatolisib (Figure 3B).

These data suggest that reciprocal activation of the AR or PAM pathway upon their individual inhibition may be cell line- or context-dependent. Whether or not the treatment with the single agents induced activation of the AR or PAM pathway, the gedatolisib/darolutamide combination effectively inhibited both pathways in all PC cell lines tested, regardless of their PTEN status.

2.3. Effects of the Gedatolisib/Darolutamide Combination on Cell Cycle

Our previous studies showed that gedatolisib effectively inhibited cell cycle progression and DNA replication in PC cells [30]. Since AR signaling plays a critical role in controlling cell cycle in PC cells [37], we tested whether the synergistic/additive growth inhibitory effects of the gedatolisib/darolutamide combination were linked to increased inhibition of cell cycle progression and DNA replication.

Four AR+ PC cell lines in which the combination of gedatolisib/darolutamide showed synergistic/additive effect by GR metrics analysis were treated with the same drug combination and tested for DNA replication and cell cycle progression by EdU incorporation and FxCycle DNA staining (see flow cytometry plots in Figure 4A,B). The gedatolisib/darolutamide combination inhibited EdU incorporation significantly more than the single agents in VCaP, LNCaP, and 22Rv1 cells, while it did not show any additional benefit in C4-2 cells (Figure 4A). A more in-depth analysis of the cell cycle using both EdU and FxCycle staining further indicated that the gedatolisib/darolutamide combination increased the percentage of cells in G0-G1 and decreased the percentage of cells in S phase, while the drug combination had less pronounced effects on the cells in G2/M (Figure 4B). These results demonstrated that the drug combination mostly arrested cells in G0-G1 by preventing the progression through S phase.

AR has been shown to control cell cycling through direct transcriptional regulation of cell cycle genes such as E2F1 and CDC6 [38,39]. Consistent with the GR metrics (Figure 2) and cell cycle (Figure 4A,B) results, treatment with 3.3–10 µM darolutamide significantly reduced E2F1 and CDC6 mRNA levels in VCaP, LNCaP and C4-2 cells, while it did not have significant effects in 22Rv1 (Figure 4C). In all cell lines, gedatolisib reduced the mRNA levels of E2F1 and CDC6 in a dose-dependent manner. Addition of darolutamide further decreased E2F1 and CDC6 mRNA levels (Figure 4C).

These results indicated that the greater growth-inhibitory effects of the gedatolisib/darolutamide combination observed by GR metrics (Figure 2) were associated, in most cases, with a greater inhibition of cell cycle genes and cell cycle progression through the G1-S phase.

2.4. Effects of the Gedatolisib/Darolutamide Combination on Cell Death and Apoptosis

The combination of PAM pathway inhibitors and darolutamide has been reported to induce apoptosis in androgen-sensitive cells [17]. To test the effects of the gedatolisib/darolutamide combination on apoptotic cell death, we treated VCaP, LNCaP, C4-2, and 22Rv1 cells with each drug, alone or in combination, for 72 h and analyzed the cells by flow cytometry. Overt cell death was detected by staining with Sytox, a DNA stain that can only enter dead cells with a compromised membrane. Apoptosis was assessed in live cells (i.e., Sytox-negative) by staining with Annexin V, which detects phosphatidylserine (PS) exposure on the outer surface of the cell membrane (see flow cytometry plots in Figure 5A). In all cell lines, the gedatolisib/darolutamide combination induced greater cell death and apoptosis relative to the single agents at one or more of the concentrations tested (Figure 5B). The combinatorial effect was especially pronounced in VCaP, LNCaP, and C4-2 cells. Cell death results were also confirmed by staining cells with TMRE, a dye that measures changes in mitochondrial potential and whose decrease is typically observed when cytochrome c is released during apoptosis (Figure S3).

We further tested the effects of the gedatolisib/darolutamide combination in three-dimensional (3D) cultures, which recapitulated features of the extracellular matrix in the tumor tissue. PTEN+ (22Rv1) and PTEN− (LNCaP, C4-2) cell lines, able to grow in 3D culture, were seeded at low density on reconstituted basement membrane and allowed to form spheroids for 4 days before drug treatment for 3 days. At the end of the treatment, the spheroids were imaged and stained with Sytox to assess spheroid cell death. As shown in Figure 5C, gedatolisib reduced spheroid growth and induced cell death in all cell lines as a single agent. Addition of darolutamide to gedatolisib significantly increased cell death, leading to spheroid regression. Consistent with the results in standard 2D culture, the gedatolisib/darolutamide combination increased cell death more effectively in LNCaP and C4-2 cells than in 22Rv1 cells.

These results demonstrated that darolutamide, which has modest effects on cell death as single agent, can increase the cytotoxic effects of gedatolisib in AR+ PC cells, regardless of PTEN mutations. These cytotoxic effects may be especially relevant in cell lines like C4-2 where the gedatolisib/darolutamide combination did not appear to have additive effects on cell cycle.

2.5. Effects of the Gedatolisib/Darolutamide Combination on Metabolic Functions

PC cells undergo profound metabolic reprogramming during PC progression and function with increased glycolysis and lipid biosynthesis at advanced cancer stages [40,41]. Since both the PI3K pathway and the AR pathway play critical roles in these metabolic adaptations, we tested whether the gedatolisib/darolutamide combination could impact key steps of glycolysis and lipid synthesis (schematically shown in Figure 6A) known to be controlled by these two pathways [9,42].

Glycolytic activity was assessed in LNCaP and C4-2 cell lines by measuring glucose uptake and subsequent conversion of glucose to lactate, an end product of glycolysis. These two cell lines were chosen because they showed higher baseline glycolysis compared to the other AR+ cell lines tested (VCaP and 22Rv1) and therefore represented better models to assess potential inhibition of glycolysis by drug treatments (Table S12). Treatment with gedatolisib significantly reduced in a dose-dependent manner both glucose uptake and lactate production in both cell lines, and the combination with darolutamide did not significantly increase the inhibitory effects of gedatolisib (Figure 6B,C). Consistent with these observations gedatolisib, alone or in combination with darolutamide, decreased the mRNA levels of hexokinase 2 (HK2), the enzyme that mediates the first step of glycolysis, and lactate dehydrogenase B (LDHB), which is critical for the conversion of pyruvate to lactate (Figure 6D).

Lipids are an important metabolic resource in PC. Substrates for lipid synthesis could arise from citrate and acetate or other by products of glycolysis or glutamine consumption. FASN, an enzyme that catalyzes the final steps of fatty acid de novo biosynthesis, is also a transcriptional target of increased AR activity associated with PC (Figure 6A). Given the likely involvement of PAM and AR in lipid metabolism, we assessed lipid metabolism by measuring changes in BODIPY-stained lipid stores in cells cultured in growth medium with no additional lipid sources. All four cell lines were tested for the effects of gedatolisib and darolutamide on their baseline lipid levels. VCaP, 22Rv1, LNCaP and C4-2 showed similar baseline levels of lipids (Table S12). As shown in Figure 6E, gedatolisib decreased lipid stores compared to baseline levels in all cell lines tested, while darolutamide significantly decreased lipid stores only in VCaP cells. The combination of gedatolisib and darolutamide did not significantly increase the effects of gedatolisib alone on lowering the lipid levels. Without discounting increased lipid utilization as a further cause of lipid stores reduction, we inferred that the decrease in lipid stores upon drug addition could be due to decreased de novo lipid synthesis linked to a reduction in glycolysis. In support of the hypothesis that lipid store reduction is a matter of reduced lipid synthesis from glycolytic intermediates, we observed that gedatolisib, with or without darolutamide, significantly decreased the fatty acid synthase (FASN) mRNA levels (Figure 6F).

The data demonstrate that gedatolisib significantly inhibited glycolysis and reduced lipid stores in PC cells lines. Adding darolutamide did not significantly increase gedatolisib’s metabolic effects on glycolysis and lipid stores.

2.6. Growth-Inhibitory Effects of the Gedatolisib/Darolutamide Combination in PC Cells Adapted to Darolutamide

Prostate cancer patients often develop resistance to AR inhibitors like darolutamide. To model prostate cancer cells resistant to darolutamide, VCaP (AR+, PTEN+) and LNCaP (AR+, PTEN−) cells were seeded by limiting dilutions and treated with darolutamide (500 nM for VCaP and 5 µM for LNCaP) to isolate single clones adapted to grow in the continuous presence of the drug. A total of seven VCaP-derived clones and five LNCaP-derived clones were expanded and tested for their growth response to AR and PAM inhibitors. Similar results were obtained for the VCaP clones (described hereafter) and the LNCaP clones (shown in Figure S4).

Analysis of GR metrics showed that the darolutamide GR50 values in the darolutamide-adapted VCaP clones were approximately 3–8 times higher than the GR50 in parental VCaP cells, indicating decreased sensitivity to darolutamide (Figure S5). The three least sensitive clones (CL211, CL212, and CL215, with GR50 = 1336, 1629, and 1985 nM versus GR50 = 256 nM in parental cells, see Figure 7A) were selected for further analysis. These darolutamide-adapted clones were also less sensitive to another AR inhibitor (enzalutamide), as well as to gedatolisib, samotolisib, and capivasertib (Figure 7A and Figure S5).

First, the darolutamide-adapted VCaP clones were tested for the acquisition of potential darolutamide resistance mechanisms. A common resistance mechanism to AR inhibitors is the presence or acquisition of AR alterations, such as AR overexpression or alternative splicing AR variants [43,44]. VCaP cells express both full length AR (AR FL) and the ARv7 splice variant, where exons 4–8 are replaced by the alternative exon 3b (Figure 7B) [44]. The absence of the AR ligand binding domain, encoded within exons 4–8, makes the ARv7 variant constitutively active [44]. qPCR with probes recognizing ARv7, AR FL (Ex4-5), or total AR (Ex1-2), showed that the three darolutamide-adapted VCaP clones (CL211, CL212, and CL215) had significantly higher levels of both ARv7 and AR FL mRNAs relative to parental VCaP, both in medium depleted of DHT and in medium supplemented with 1 nM DHT (Figure 7B). Consistently, the VCaP clones also showed increased transcription of two AR-target genes, TMPRSS2 and KLK3, in response to 1 nM DHT, indicating increased AR activity. Interestingly, KLK3 mRNA (and to a lesser extent TMPRSS2 mRNA) showed an increase even in the DHT-depleted medium, suggesting DHT-independent AR activity (Figure 7C). Furthermore, additional qPCR experiments showed that the darolutamide-adapted VCaP clones also expressed higher levels of glucocorticoid receptor (GR) mRNA (Figure S6), which is another known mechanism of resistance to antiandrogens [45].

Next, the darolutamide-adapted VCaP clones were tested for their response to gedatolisib in combination with darolutamide. GR metrics analysis after a 72 h treatment showed that the gedatolisib/darolutamide combination exerted greater anti-proliferative and cytotoxic effects than the single agents in the three clones tested (Figure 7D). For instance, in clone CL215, 111 nM gedatolisib and 10 µM darolutamide as single agents induced growth inhibitory effects with GR values of −0.02 and −0.03 (mostly cytostatic), and the combination of the two drugs at the same concentrations lowered the GR value to −0.47 (cytotoxic). The increased cytotoxic effects of the gedatolisib/darolutamide combination were also confirmed by flow cytometry analysis of cell death and apoptosis by Sytox/Annexin V staining (Figure S7). Chou and Talalay synergy analysis further showed that gedatolisib and darolutamide had synergistic growth inhibitory effects with CI values < 0.7 in the three clones analyzed (Figure 7E).

These results indicated that the combination of gedatolisib and darolutamide exerted synergistic growth inhibitory effects not only in darolutamide-naïve cells, but also in darolutamide-adapted cells, regardless of their PTEN status.

4.1. Cell Culture

The PC cell lines 22Rv1 (CRL-2505, RRID:CVCL_1045), Du145 (HTB-81, RRID:CVCL_0105), LNCaP (CRL-1740, RRID:CVCL_1379), LNCaP C4-2 (CRL-3314, RRID:CVCL_4782; here referred to as C4-2), MDA-PCa-2b (CRL-2422, RRID:CVCL_4748), PC3 (CRL-1435, RRID:CVCL_0035), and VCaP (CRL-2876, RRID:CVCL_2235) were obtained from ATCC (Manassas, VA, USA), authenticated by STR profiling (ATCC), and tested for mycoplasma. Cells were cultured based on the vendor’s recommendations in medium containing 10–20% FBS (R&D Systems, Minneapolis, MN, USA, Cat# S11550) in a 37 °C, 5% CO₂ humidified incubator. To ensure consistency between cultures, the same FBS lot was used in this study. The PTEN and AR status of the cell lines was derived from previous studies [35,72]. The PIK3CA status was derived from analysis of the Cancer Cell Line Encyclopedia (CCLE, Broad 2019 dataset) through cBioPortal (RRID:SCR_014555, https://www.cbioportal.org/, accessed on 7 July 2022). To develop darolutamide-adapted VCaP and LNCaP clones, VCaP and LNCaP cells were seeded in 96-well plates by limiting dilution to obtain single clones. The cells were grown in growth medium supplemented with darolutamide at concentrations above the GR50 for each cell line (500 nM for VCaP, and 5000 nM for LNCaP) over a period of 3–4 months. The clones that were able to grow in the presence of darolutamide were expanded under the drug selective pressure and archived for further experiments. The cell lines that were selected with darolutamide are referred to as “darolutamide-adapted”, while the cell lines that were not selected with darolutamide are referred to as “darolutamide-naïve” or “treatment-naïve”. The cellular origin of the clones was confirmed by STR analysis. All cell line samples sent to Charles River Laboratories (Wilmington, MA, USA) for testing were confirmed for the absence of contamination by common Mycoplasma species.

4.2. Treatments

Gedatolisib (Celcuity, Minneapolis, MN, USA), samotolisib (Cat# S8322), capivasertib (Cat# S8019), darolutamide (Cat# S7559), and enzalutamide (Cat# S1250) (Selleckchem, Houston, TX, USA) were reconstituted in DMSO (Bio-Techne Tocris, Minneapolis, MN, USA, Cat# 3176). For treatments, cells were seeded in androgen-depleted growth medium, i.e., medium supplemented with charcoal-stripped FBS (R&D Systems, Minneapolis, MN, USA, Cat# S11650) instead of standard FBS. The charcoal-stripped and standard FBS lots used in this study had a DHT concentration of 14.1 pg/mL (0.0489 pM) and 133.1 pg/mL (0.461 pM), respectively (analysis by the Clinical Endocrinology Laboratory, UC Davis, CA, USA). For treatments of darolutamide-adapted VCaP and LNCaP clones, cells were seeded in the absence of darolutamide. For flow cytometry and viability experiments, cells were seeded on white 96-well plates; for RNA experiments, cells were seeded in 12-well plates. Plates were coated with ECM proteins as described [30]. The seeding density was optimized for each cell line to prevent over-confluence by the end of the treatment. After overnight attachment, cells were treated with the indicated drugs or corresponding vehicle (DMSO) for the indicated time. Unless indicated otherwise, all treatments were performed in medium supplemented with charcoal-stripped FBS and 1 nM DHT (Sigma-Aldrich, St. Louis, MO, USA, Cat# D-073). The concentrations of gedatolisib and darolutamide tested in this study were chosen to encompass the concentrations reported in plasma of treated patients (gedatolisib > 15–20 nM up to 168 h after treatment with 154 mg; darolutamide between 2.5 and 5 µM up to 8 h after treatment with 600 mg) [70,73].

4.3. Growth Rate (GR) Metrics

GR metrics analyses were performed as described in Sen et al. [30] based on the original method by Hafner et al. [74]. Cell viability before and after drug treatments for 72 h was assessed by RT-Glo MT luciferase assay (Promega, Cat# G9712). Luminescence was measured using an Infinite M1000 microplate reader (Tecan, Männedorf, Switzerland). The viability measurements were used to calculate GR values, which were plotted in dose–response curves using PRISM (RRID:SCR_002798, GraphPad Software 10.4.1, Boston, MA, USA). Anti-proliferative effects are indicated by GR values between 0 and 1; cytostatic effects are indicated by GR = 0; cytotoxic effects are indicated by GR values between −1 and 0. Drug potency and efficacy were assessed by calculating the GR₅₀ (concentration needed to obtain GR = 0.5) and GR_Max (GR value at the maximum concentration tested) from the dose–response curves.

4.4. Drug Synergy Analyses

Synergy was assessed using the Chou-Talalay method [75] on CalcuSyn software 2.11 (ComboSyn, Inc., Paramus, NJ, USA). Drug inhibition data from RT-Glo MT luciferase assay was used for calculation of synergy. A combination index (CI) below 1 is considered synergistic, a value at 1 is considered additive, and a value above 1 is considered antagonistic. The combination index is given for ED50 (effective dose). Synergy was also quantified using the Multi-dimensional Synergy of Combinations (MuSyC) method, which distinguishes between different synergy types (e.g., potency, efficacy) [76]. Drug-induced reduction in GR values calculated from GR metric analysis was used as the input for this analysis. This was entered into the MuSyc online application (https://app.duetbiosystems.com, accessed from 11 March 2024 to 24 May 2024). Synergy in terms of efficacy is given as a percent increase in efficacy of the combination of drugs over the drugs used singly.

4.5. Analysis of PAM Pathway Activity

PAM pathway activity was assessed by flow cytometry of pRPS6 and p4EBP1, two key effectors of the PAM pathway, as previously described [30]. Cells treated with the indicated drugs for 24–72 h were harvested from 96-well plates by trypsinization. After a wash with PBS, cells were stained with Zombie NIR (Biolegend, San Diego, CA, USA, Cat# 423106), fixed with 1.6% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA, Cat# 15710), and permeabilized with cold methanol (Sigma-Aldrich, Cat# 179337). After washing, cells were stained with anti-pRPS6-BV421(S235/S236) (Biolegend, clone A17020B, Cat# 608610, 1:50 dilution) and anti-p4EBP1-AF488 (T36/T45) (BD Biosciences, New York, NY, USA, clone M31-16, Cat# 560287, 1:25 dilution) antibodies and run on a Novocyte 3005 (Agilent, Santa Clara, CA, USA) flow cytometer. Data analysis was performed with NovoExpress 1.5.6 (Agilent). pRPS6 and p4EBP1 median fluorescent intensity in live singlets (selected by forward/side scatter and Zombie staining) was used to assess pRPS6 and p4EBP1 levels. An example of the gating strategy is provided in Supplementary Figure S8.

4.6. Analysis of Cell Cycle and DNA Replication

Cell cycle and DNA replication were analyzed by flow cytometry as previously described [30]. Cells treated with the indicated drugs for 48 h were incubated with 10 µM 5-ethynyl-2′-deoxyuridine (EdU) (Thermo Fisher Scientific, Waltham, MA, USA, Cat# C10419) for the last 2 h of the treatment and harvested, stained with Zombie NIR, fixed, and permeabilized as described above. EdU incorporation was detected by using the Click-iT EdU Alexa Fluor 647 kit (Thermo Fisher Scientific, Cat# C10419) per manufacturer’s instructions. After washing, cells were stained with 1 μg/mL FxCycle Violet (Thermo Fisher Scientific, Cat# F10347) and run on a Novocyte 3005 (Agilent) flow cytometer. Data analysis was performed with NovoExpress 1.5.6 (Agilent). Live singlets, selected by forward/side scatter and Zombie staining, were gated based on EdU incorporation and FxCycle to assess DNA replication and to identify G0/G1, S, and G2/M phases. If present, cells in sub-G1 and super-G2 were not included in the analysis. An example of the gating strategy is provided in Supplementary Figure S8.

4.7. Analysis of Cell Death and Apoptosis

Cells treated with the indicated drugs for 72 h were harvested from 96-well plates by trypsinization. Both attached and floating cells (potentially containing dead/apoptotic cells) were collected for flow cytometry analysis. Dead cells were identified by staining with Sytox Blue, a DNA stain that only enters dead cells with compromised cell membranes. Apoptotic cells were identified by Annexin V staining, which detects phosphatidylserine (PS) exposure on the outer surface of the cell membrane. Additionally, cells were stained with tetramethylrhodamine ethyl ester (TMRE), which measures changes in mitochondrial potential and whose decrease is typically observed when cytochrome c is released from mitochondria during apoptosis [77]. After washing with PBS + 0.5% BSA (Sigma-Aldrich, Cat# A9647) + 0.02% Sodium Azide (G Biosciences, St. Louis, MO, USA, Cat# 786-299) (FACS staining buffer, FSB), cells were incubated with 200 nM TMRE (Abcam, Waltham, MA, USA, Cat# ab113852) for 30 min at 37 °C, washed with FSB, stained with Annexin V-A647 (Biolegend, Cat# 640943) diluted 1:20 in Annexin V binding buffer (Biolegend, Cat# 422201) for 15 min at 4 °C, and finally incubated for 5 min with 5 μM Sytox blue (Thermo Fisher Scientific, Cat# S34857) diluted in Annexin V buffer. Cells were immediately run on a Novocyte 3005 (Agilent) flow cytometer. Data were analyzed with NovoExpress 1.5.6 (Agilent). Cells were first gated by forward/side scatter to exclude cell debris and aggregates, then gated by Sytox Blue, TMRE, or Annexin V/Sytox Blue staining to identify dead cells (Sytox+/Annexin+) and live apoptotic cells (Sytox-/Annexin+ cells or Sytox-/TMRE− cells). Examples of the gating strategies are provided in Supplementary Figure S8.

4.8. Metabolic Studies

Glucose uptake was measured using Glucose Uptake-Glo kit (Promega, Madison, WI, USA, Cat# G9711). Cells were seeded in 96-well plates and allowed to incubate for 16 h under culture conditions, and then treated with the indicated drugs for 4 h. After removing the medium, cells were washed with PBS and incubated with 1 mM 2-deoxyglucose (2DG) for 10 min at room temperature. Stop buffer, neutralization buffer and detection reagent were then sequentially added as per manufacturer’s protocol. The plates were incubated for 1 h at room temperature and luminescence read using an Infinite M1000 microplate reader. Cells incubated with PBS, no 2DG, were used for background subtraction. A negative control reaction was performed with culture medium without cells. Glucose uptake measurements were normalized to cell number, which was assessed by measuring DNA content in the same plate using CyQuant assay (Thermo Fisher Scientific, Cat# C7026).

Glucose and lactate were measured in the conditioned media using the Biosen R-line instrument (EKF diagnostics, Cardiff, UK). Cells were seeded in 96-well plates and allowed to incubate for 16 h under culture conditions, followed by drug treatment for 24 h. Glucose consumed and lactate produced were calculated based on glucose and lactate concentrations measured in media with no cells, which was used as baseline control. EKF Diagnostics RUO Multi Standard (Cat#6130-3221), High Standard (Cat# 6130-6025), and Low Standard (Cat# 6130-6015) solutions were used for instrument calibration. EKF Diagnostics Measurements were normalized to cell number, which was assessed by measuring DNA content in the same plate using CyQuant assay (Thermo Fisher Scientific, Cat# C7026).

BODIPY 493/503 (Cayman Chemical Company, Ann Arbor, MI, Cat#25892) was used to measure neutral lipid storage in the PC cells. Cells were seeded in 96-well plates in charcoal-stripped media supplemented with 1 nM DHT. Cells were treated with the indicated drugs for 24 h. No additional source of lipid was provided in the cell media. After drug treatment, neutral lipids were stained by incubating cells with 5 μM BODIPY 493/503 for 30 min at 37 °C. Cells were then harvested from 96-well plates by trypsinization for flow cytometry analysis. Staining with Sytox Blue 5 μM Sytox blue (Thermo Fisher Scientific, Cat# S34857) was used to identify dead cells. Unstained cells were used as negative control. Samples were run on a Novocyte 3005 (Agilent) flow cytometer and analyzed with NovoExpress 1.5.6 (Agilent). BODIPY 493/503 median fluorescence intensity was assessed in Sytox-negative live cells.

4.9. Quantitative PCR (qPCR)

Cells for qPCR were seeded in 12-well plates and allowed to incubate for 16 h under culture conditions before 24 h drug and DHT treatment. RNA was extracted using QuickRNA Microprep kit (Zymo, Irvine, CA, USA, Cat# R1051) following manufacturer’s protocol. Nanodrop was used to measure RNA concentrations. cDNA was synthesized using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA, Cat# 1708891) and amplified using Taqman probes (Thermo Fisher Scientific) and Applied Biosystems TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific, Cat# 444557) on a QuantStudio3 thermocycler (Thermo Fisher Scientific). The Taqman probes used were AR exon 4–5, Hs00171172_m1; AR exon 1–2, Hs00907242_m1; TMPRSS2, Hs01122322_m1; KLK3, Hs02576345_m1; NDFIP1, Hs00228968_m1; FASN, Hs01005622_m1; LDHB, Hs00929956_m1; HK2, Hs00892681_m1; E2F, Hs00153451_m1; CDC6, Hs00154374_m1. The ARv7 probe was designed based on sequences provided in Marin-Aguilera et al. [78]. Relative changes in mRNA levels were calculated by the Delta/DeltaCt method using NDFIP1 as the reference gene.

4.10. Statistical Analyses

Statistical significance was calculated by Analysis of Variance (ANOVA) as indicated in the figure legends using PRISM (RRID:SCR_002798, GraphPad Software). p-values < 0.05 were considered significant at 95% confidence.