Targeting DHODH reveals a metabolic vulnerability in AR-positive and AR-negative prostate cancer cells via pyrimidine synthesis and metabolic crosstalk with the TCA and urea cycles

Abstract

Following recurrence, the cornerstone clinical therapy to treat prostate cancer (PCa) is to inhibit the androgen receptor (AR) signaling. While AR inhibition is initially successful, tumors will eventually develop treatment resistance and evolve into lethal castration-resistant PCa. To discover new anti-metabolic treatments for PCa, a high-throughput anti-metabolic drug screening was performed in PC3 cells, an AR-negative PCa cell line. This screening identified the dihydroorotate dehydrogenase (DHODH) enzyme as a metabolic vulnerability, using both AR-positive and AR-negative models, including the neuroendocrine cell line LASCPC-01 and patient-derived organoids. DHODH is required for de novo pyrimidine synthesis and is the sole mitochondrial enzyme of this pathway. Using extracellular flux assays and targeted metabolomics, DHODH inhibition was shown to impair the pyrimidine synthesis pathway, as expected, along with a significant reprogramming of mitochondrial metabolism, with a massive increase in fumarate (>10-fold). Using ¹³C₆-glucose, it was shown that following DHODH inhibition, PCa cells redirect carbons from glucose toward biosynthetic pathways rather than the TCA cycle. In parallel, using ¹³C₅-glutamine, it was shown that PCa cells use this amino acid to fuel a reverse TCA cycle. Finally, ¹³C₁-aspartate and ¹⁵N₁-glutamine highlighted the connection between pyrimidine synthesis and the urea cycle, redirecting pyrimidine synthesis intermediates toward the urea cycle as a stress response mechanism upon DHODH inhibition. Consequently, combination therapies targeting DHODH and glutamine metabolism were synergistic in impairing PCa cell proliferation. Altogether, these results highlight DHODH as a metabolic vulnerability of AR-positive and AR-negative PCa cells by regulating central carbon and nitrogen metabolism.

Highlights

• •DHODH is a metabolic vulnerability of AR-positive and AR-negative prostate cancer cells.
• •Upon DHODH inhibition, glucose, glutamine and aspartate usage are rewired toward several metabolic pathways.
• •DHODH inhibition connects pyrimidine synthesis, the TCA cycle and the urea cycle in PCa cells.
• •DHODH insights revealed GSL as a synergistic strategy for PCa cells.

1. Introduction

Prostate cancer (PCa) is a major public health issue worldwide, being the most frequent cancer afflicting men in 117 countries [1]. Therapies targeting the androgen receptor (AR), such as androgen deprivation therapies (ADT) and anti-androgens, are the cornerstone for treating PCa [2]. Indeed, the AR is a primary oncogenic driver of PCa cell survival and proliferation [3,4]. Even though those treatments are efficient in the first stages of the disease, tumors will eventually develop resistance and evolve to castration-resistant prostate cancer (CRPC) [4]. While most resistance mechanisms involve a reactivation of the AR signaling pathway, including AR gene amplification and de novo intra-tumor androgen synthesis, a neuroendocrine differentiation can also occur in ∼20% of CRPC cases [[4], [5], [6]]. Treatment-induced neuroendocrine PCa (tNEPC) shares the same characteristics as de novo NEPC, like the expression of neuroendocrine markers enolase 2 (ENO2), synaptophysin (SYP), and chromogranin A (CHGA), along with the absence or loss of AR [7]. With limited treatment options following the loss of AR, tNEPC is a lethal form of PCa for which new therapeutic opportunities are urgently required.

The dihydroorotate dehydrogenase, DHODH, catalyzes the 4th reaction in de novo pyrimidine biosynthesis, a pathway often exacerbated in cancer cells [8]. It is the sole enzyme involved in pyrimidine synthesis that is localized within the mitochondria. The other two, carbamoyl-phosphate synthetase 2, aspartate transcarbamoylase, and dihydroorotase, CAD, and uridine monophosphate synthase, UMPS, are found in the cytosol [9]. Because of its mitochondrial localization, DHODH functionally links pyrimidine biosynthesis to the electron transport chain via its cofactor ubiquinone [10]. The DHODH enzyme, thus, has a dual function in cell metabolism, contributing to pyrimidine synthesis and regulating mitochondrial activity [[11], [12], [13]]. Given its importance in nucleotide synthesis, DHODH has attracted attention in several cancer settings, for both solid and liquid tumours [14]. Indeed, its inhibition impairs nucleotide synthesis, forcing cancer cells to rely on salvage nucleotide synthesis and promoting apoptosis when this alternative pathway cannot compensate [12]. In preclinical models, DHODH inhibition has shown promising results, such as for acute myeloid leukemia [14], glioblastoma [15], medulloblastoma [16], liver cancer [17], and pancreatic cancer [18].

In a recent study, high DHODH expression was associated with poorer survival of PCa patients, suggesting that it plays an essential role in supporting PCa cell proliferation [19]. In line with this hypothesis, knockdown of the DHODH gene was shown to induce AR-positive PCa cell death, and DHODH inhibition reduced AR-positive PCa xenograft growth in vivo [19]. Interestingly, the inhibition of DHODH was shown to induce androgen biosynthesis, and, thus, the combination with the androgen synthesis inhibitor abiraterone was shown to increase efficacy in AR-positive models [19]. In clinical settings, AR-positive tumours are often AR-dependent and can be treated with hormonal therapies [4]. One of the biggest clinical challenges in the PCa field is to treat AR-negative tumors, such as tNEPC, as they are not responsive to anti-androgen therapies [20]. If DHODH is a promising therapeutic target in AR-negative PCa cells, and what the molecular consequences of DHODH inhibition are in this context, remains to be determined.

Here, using a high-throughput anti-metabolic screening in an AR-negative cell line, DHODH was identified as a candidate metabolic vulnerability. The efficacy of DHODH inhibition in the context of AR-negative PCa, and notably in a neuroendocrine model, was further validated in additional cell lines and patient-derived organoid (PDO) models, along with rescue experiments. Metabolic assays, including stable isotope tracer analyses, were then performed to understand the metabolic consequences of DHODH inhibition. While it inhibited pyrimidine synthesis, as expected, our results also revealed that it further reprogrammed the TCA and the urea cycles in AR-negative PCa cells, thus having broader metabolic consequences than initially anticipated. Altogether, our results demonstrate that, by significantly altering the cancer cell metabolome, DHODH inhibition represents a novel metabolic vulnerability in AR-independent PCa cells. Furthermore, by understanding the metabolic consequences of DHODH inhibition, this study also revealed possible combination therapies to maximize the blockade of PCa cell proliferation.

2. Results

2.1. High-throughput anti-metabolic drug screening identifies DHODH as a PCa metabolic vulnerability in AR-negative PCa cells

To identify new metabolic vulnerabilities of AR-negative PCa cells, a high-throughput anti-metabolic drug screening was performed. To this end, an anti-metabolic library, comprised of 1,280 molecules, was used. This library includes FDA-approved drugs, molecules in ongoing clinical trials, and molecules used in preclinical models. PC3 cells were first plated in 96-well plates. Twenty-four hours later, treatment was initiated at a concentration of 100 nM for each molecule. Finally, four days later, cells were stained with Hoechst and nuclei were counted using automated cell imaging (see experimental design schematic, Figure 1A). Along with the molecules tested in the library (Figure 1B), the used experimental design included two additional positive controls: metformin, to induce a low but significant decrease in proliferation (to control for our ability to detect low impacts), and puromycin, to induce strong cell death (Figure 1C). Out of the 1,280 studied molecules from the library, 90 led to a significant decrease of 40% or more in PC3 cell number (Figure 1B). These include molecules targeting known targets in PCa, such as the doxorubicin chemotherapy [21] or inhibitors of the PI3K/Akt/mTOR oncogenic pathway [22], both leading to >80% decrease in cell number, validating the studied experimental design (Figure 1B and C). In addition, the screening also revealed molecules targeting pathways or enzymes not known or less examined in the context of PCa (Supplemental Table S1). Of particular interest, five molecules targeting the same metabolic enzyme, DHODH, were identified (brequinar, orludodstat, DHODH–IN–1, hDHODH–IN–4, and hDHODH–IN–7; Figure 1D). These molecules inhibited PC3 proliferation by decreasing cell number between 45.5 and 63.7% following 96 h of treatment. These results, performed in an AR-negative human PCa cell line, suggest that DHODH inhibition is a novel metabolic vulnerability for this aggressive subset of PCa tumors.

Figure 1.

High-throughput anti-metabolic drug screening identifies in PC3 cells the DHODH enzyme as a metabolic vulnerability of PCa cells. A) Schematic overview of the experimental design used for the high-throughput anti-metabolic drug screening. PC3 cells were seeded at the beginning of the experiment in 96-well plates. Twenty-four hours later, cells were treated with the compounds from the library, which included 1,280 molecules. All treatments, at 100 nM, were performed for a total of four days before automated cell counting of three experimental replicates. B) Results from the high-throughput anti-metabolic drug screening, shown as alterations compared to controls, which are set at 0. Results are shown as the average cell count of three experimental replicates. Ninety molecules were identified to significantly decrease PCa cell proliferation by at least 40%. Specific results for positive controls (C) and DHODH inhibitors (D) are shown, with average and standard error of the means. ∗∗∗p < 0.001.

Several generations of DHODH inhibitors have been developed over the years. Teriflumonide and brequinar were created in the 1980–1990s as part of the first generation of DHODH inhibitors [23]. While brequinar showed reduced cancer cell proliferation in preclinical models, it unfortunately failed to provide significant results in clinical trials for various solid tumours [24,25]. Similarly, teriflunomide showed anti-cancer effects in the AR-negative DU145 cell lines, but at high concentrations (50 μM) [26], and also failed to be in a clinical context for cancer treatment. However, with the generation of new DHODH inhibitors, including DHODH-IN and orludodstat, there is a renewed interest in combinational therapies for the treatment of blood cancers [27,28]. Interestingly, Guo et al. recently revealed the potential of orludodstat, also named BAY 2402234 [14], as a new potential therapeutic target for AR-positive PCa preclinical models [19]. Notably, we validated their results in the AR-positive PCa cell line LNCaP, showing that inhibition of DHODH with orludodstat decreases LNCaP cell proliferation in cell counting (Supplemental Fig. S1A). Here, our results with the high-throughput screening also support the hypothesis that DHODH inhibition is a therapeutic target in aggressive AR-negative PCa.

To validate this hypothesis, the impact of DHODH inhibition using orludodstat, an orally available DHODH inhibitor, was validated on cell number in PC3 cells (Figure 2A and B) and further studied in two additional human AR-negative models, DU145 (Figure 2C and D) and the neuroendocrine PCa model LASCPC-01 (Figure 2E and F) cells. Cell number quantification revealed highly similar results for all three cell lines, with >70 % decreased cell number for all three cell lines. In viability assays, PC3 and DU145 cells showed similar sensitivity to DHODH inhibition, in the low nM range, with a maximal inhibition of 70%, suggesting that orludodstat does not kill these cells but rather impairs their proliferation (Figure 2G and H). The neuroendocrine PCa model LASCPC-01was more sensitive to orludodstat, exhibiting less than 5% viability after a 4-day treatment with orludodstat (Figure 2I), suggesting cell death occurs in this context. Accordingly, LASCPC-01 cells exhibited the induction of the apoptosis marker, poly(ADP-ribose) polymerase (PARP) cleavage, following orludodstat treatment, which was not observed in PC3 and DU145 cells (Figure 2J). To validate that the effect is cytostatic in PC3 and DU145 cels, we quantified the proportion of cells in the different phases of the cell cycle using flow cytometry. In these experiments, PC3 and DU145 cells were shown to exhibit a blockade of the cell cycle upon DHODH inhibition, being mostly enriched in the S phase (∼75%; Figure 2K–N). This cytostatic effect is similar to the effect observed in other cancer settings [14]. While the effects of orludodstat, at the concentration used, were shown to be specific for DHODH inhibition in other cancer cells [29], we next validated that this impairment of cell viability/proliferation in the three studied PCa cell lines was specific to DHODH inhibition. To this end, we first performed a rescue experiment using uridine, a metabolite downstream of DHODH that was shown to rescue the inhibition of this enzyme [29]. As expected, orludodstat alone significantly decreased PC3, DU145, and LASCPC-01 cell number (Figure 2O–Q). Importantly, uridine completely rescued this phenotype (Figure 2O–Q), a result that was also observed in LNCaP cells (Supplemental Fig. S1B). In addition, we performed cell counting and viability assays in PC3 and DU145 using additional DHODH inhibitors: teriflunomide, brequinar, and DHODH–IN–1. All three inhibitors significantly decreased cell number in both PC3 and DU145 cells (Supplemental Fig. S1C–D). However, teriflumonide and brequinar, first generation DHODH inhibitors, had a lower impact on cell number compared to DHODH–IN–1 in DU145 cells (∼30% vs 90%). These three inhibitors induced a similar decrease in PC3 cell viability (Supplemental Fig. S1E). However, only the most recent inhibitor, DHODH–IN–1 induced a significant decrease of DU145 cell viability (Supplemental Fig. S1F), consistent with a stronger impact on cell number (Supplemental Fig. S1D) and reinforcing the need to study newer inhibitors for cancer treatment, such as orludodstat and DHODH–IN–1. Thus, as observed for AR-positive cell lines ([19] and Supplemental Fig. S1A–B), DHODH inhibition, and notably the newer inhibitors orludodstat and DHODH–IN–1, also impairs AR-negative PCa cell proliferation and viability, being cytostatic in PC3 and DU145 cells and inducing cell death in neuroendocrine LASCPC-01 cells.

Figure 2.

DHODH inhibition using orludodstat impairs cell proliferation in various AR-negative PCa cells. Brightfield visualisation of PC3 cells (A), DU145 cells (C) and LASCPC-01 (E), with and without 96 h orludodstat treatment at 100 nM. Scale bar 100 μm. PC3 (B), DU145 (D) and LASCPC-01 (F) cell number, following 96 h orludodstat treatment at 100 nM. PC3 (G), DU145 (H) and LASCPC-01 (I) cell viability after 96 h of different concentrations of orludodstat treatment. Results from one representative out of three independent experiments are shown (average and S.E.M.). J) Western Blot assessment of PARP cleavage, a marker of apoptosis, following 24 h orludodstat treatment. PC3 cell analysis following 96 h orludodstat treatment using flow cytometry (K), showing a blockade in S phase (L). DU145 cell analysis following 96 h orludodstat treatment using flow cytometry (M), showing a blockade in S phase (N). Rescue experiments of 96 h orludodstat treatment with 750 μM uridine in PC3 (O), DU145 (P), and LASCPC-01 (Q) cells. Results from three independent experiments are shown (average and S.E.M.). For panels A, C, E, and J, representative images of one independent experiment, out of three independent experiments, are shown. ∗p < 0.05, ∗∗p < 0.01 ∗∗∗p < 0.001.

2.2. DHODH expression is associated with PCa aggressivity, and its inhibition impairs PDO growth

The results described above, highlighting the requirement of DHODH for maximal PCa cell survival and proliferation, prompted us to study its expression in publicly available RNA-seq datasets. First, reanalysis of The Cancer Genome Atlas (TCGA) dataset indicated that DHODH expression is increased in PCa tumors compared to normal prostate glands, supporting the hypothesis that this gene plays an oncogenic role (Figure 3A), consistent with previous reports [19]. In addition, DHODH relative gene expression was also found to be higher in tumour samples compared to benign peri-tumor glands in the dataset from Varambally et al. [32] (Figure 3B) and to be higher in recurrent PCa tumors compared to non-recurrent tumors in the microarray dataset from Sun and Goodison dataset [33] (Figure 3C-D). The only exception was in the dataset from Taylor et al. [34], in which DHODH expression was lower in tumours compared to benign peri-tumour glands (Supplemental Fig. S2A). Yet, altogether, this reanalysis of publicly available datasets instead supports that DHODH levels are increased during prostate carcinogenesis. As DHODH inhibition was identified to block AR-positive [19] and AR-negative (this study) PCa cell proliferation, we next tested if DHODH expression in tumours was correlated or not with either AR or tNEPC biomarkers. To this end, we leveraged the publicly available RNA-seq dataset from the SU2C/PCF Dream Team, comprising tumour samples from 444 metastatic CRPC [35]. No significant correlations were observed between DHODH and either AR, two androgen-regulated genes, KLK3 (encoding PSA) and FKBP5, nor the tNEPC biomarkers SYP, CHGA, and ENO2 (Supplemental Fig. S2B–G). In PCa cell lines, we did not observe major differences between DHODH levels and AR status (Supplemental Fig. S2H). Altogether, these findings indicate that DHODH status is independent of AR or tNEPC status and constitutes a broad metabolic vulnerability in prostate cancer.

Figure 3.

DHODH expression is associated with PCa aggressivity, and its inhibition impairs PDO growth. A) DHODH relative expression in publicly available data from the TCGA consortium [30,31], which includes 100 normal peri-tumor samples and 495 tumor samples. B) DHODH relative expression in publicly available data from Varambally et al. [32] (probe, 213631_x_at), which includes 6 normal peri-tumor benign samples and 13 tumor samples. C) DHODH relative expression in publicly available data from Sun et al. (probe 213632_x_at) [33], which includes 40 and 39 tumor samples from non-recurrent and recurrent PCa following surgery, respectively. D) DHODH relative expression in publicly available data from Sun et al. (probe 217647_x_at) [33], which includes 40 and 39 tumor samples from non-recurrent and recurrent PCa following surgery, respectively. E) Left: representative brightfield visualization of PDO line #1 samples, with and without orludodstat treatment (100 nM). Scale bar: 200 μm. Treatment began at seeding and was maintained for the entire experiment (14 days). At 14 days, samples were harvested for a viability assay by quantification of intracellular ATP, as shown on the right. F) Viability assay of PDO line #2, with and without orludodstat treatment from seeding (day 0) until the end of the experiment (day 14). G) Viability assay of PDO line #3, with and without orludodstat treatment from day 3 of culture until the end of the experiment (day 14). H) Viability assay of PDO line #4, with vehicle or 100 nM orludodstat treatment or 750 μM uridine or both treatments, at seeding until the end of the experiment (day 14). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Using an established pipeline to generate patient-derived organoids (PDOs) [36,37], the therapeutic potential of DHODH inhibition was further confirmed in three PDO lines. These PDO lines were derived from primary hormone-naïve PCa adenocarcinoma (clinical details can be found in the Supplemental Table S2) and some of them were previously shown to be sensitive to the anti-androgen enzalutamide [37]. When orludodstat was added at the seeding of PCa cells for organoid genesis, DHODH reduced cell viability by more than 75% (Figure 3E and F). In a third PDO line, orludodstat was added after the initial 3 days in culture. Treatment was renewed 3 times a week until organoids reached their maximal growth (14 days); in these settings, DHODH inhibition decreased by over 50% cell viability (Figure 3G). As observed in PCa cell lines, addition of uridine rescued this phenotype (Figure 3H). These results are in line with those obtained using cell lines (Figures 1 and 2) and indicate that DHODH inhibition impairs PCa cell proliferation and organoid growth.

2.3. DHODH is essential to central carbon metabolism in AR-negative PCa cells

Next, the impact of DHODH inhibition on pyrimidine synthesis was studied in PC3, DU145, and LASCPC-01 cells using targeted metabolomics. Both glutamine and aspartate are required for full de novo pyrimidine synthesis (Figure 4A), among other pathways in which they can participate. While glutamine levels were inconsistently modulated by orludodstat, being unchanged in PC3 cells, undetected in DU145 cells, and decreased in LASCPC-01 cells, aspartate levels were significantly increased in all three cell lines following DHODH inhibition, from an increase of 150% to over 1,500% (Figure 4B–D). While we were not able to quantify the DHODH product, orotate, its substrate, dihydroorotate (DHO), was detectable and quantifiable. In all three cell lines, following treatment with orludodstat, we observed a massive increase in DHO levels, consistent with a blockade of its usage in PCa cells following DHODH inhibition (>100-fold; Figure 4E–G; and using the DHODH–IN–1, Supplemental Fig. S3A). Similar results were obtained in LNCaP cells, with strong increases in aspartate and DHO levels following orludodstat treatment (Supplemental Fig. S3B).

Figure 4.

DHODH is required for proper mitochondrial metabolism in PCa cells. A) Schematic visualisation of the pyrimidine de novo biosynthetic pathway. Glutamine and aspartate are used as CAD substrates to initiate pyrimidine synthesis, with three sequential enzymatic reactions, before the enzymatic activity of the mitochondrial DHODH. Glutamine (glut) and aspartate (asp) quantification in PC3 (B), DU145 (C), and LASCPC-01 (D) cells, with and without 24 h treatment with the DHODH inhibitor orludodstat at 100 nM. The DHODH substrate, dihydroorotate (DHO), was measured in PC3 (E), DU145 (F), and LASCPC-01 (G) cells following 24 h treatment with orludodstat. H–I) RNA-seq data were retrieved from the TCGA cbioportal for the neuroendocrine PCa cohort (NEPC). Genes positively correlating with DHODH (>0.75) were studied for their enrichment in biological pathways (H) or human diseases (I) using the Metascape portal. For -log10(p value), a value > 1.3 is significant with p > 0.05 (dashed line). Lactate quantification in PC3 (J), DU145 (K), and LASCPC-01 (L) cells, with and without 24 h treatment with the DHODH inhibitor orludodstat. TCA cycle metabolite measurement in PC3 (M), DU145 (N), and LASCPC-01 (O) cells, with and without 24 h treatment with the DHODH inhibitor orludodstat. Metabolomics results are shown as average and S.E.M. of one representative experiment, out of two to three independent experiments. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; n.s.: not significant. Basal respiration of PC3 cells (P) and DU145 cells (Q) after 10 h following DHODH injection. Results are shown as the mean and S.E.M (n = 20–24) in one representative experiment. All the experiments were performed at least three times independently.

To further gain insights into the possible molecular functions of DHODH in the PCa context beyond DHO usage, and notably in AR-negative tumors, we also re-analyzed RNA-seq data from the neuroendocrine PCa cohort using the TCGA ciobportal [38,39]. In these data, genes positively correlated with the DHODH mRNA with a Spearman's correlation coefficient of 0.75 or above were selected, identifying a total of 154 genes. Highly correlated genes often represent genes regulated by similar regulatory mechanisms and/or genes connected to similar biological pathways. Interestingly, these top-correlated genes with DHODH showed enrichment in mitochondrial biological pathways, including the TCA cycle (Figure 4H). Furthermore, using the Metascape tool to study if these genes are linked to human diseases [40], they showed enrichment connected with lactic acidosis (Figure 4I), suggesting an imbalance between mitochondrial respiration and aerobic glycolysis (lactate production). Glucose is typically consumed via glycolysis to generate energy. The end product of glycolysis, pyruvate, can then be used to generate lactate (lactic acid) or fuel the TCA cycle for mitochondrial respiration. When oxygen is present, respiration is generally favored, but cancer cells often exhibit high lactate production in normoxia, typically referred to as the Warburg effect or aerobic glycolysis [41]. In tissues, when lactate levels build up, it decreases the pH and can cause damage due to this lactic acidosis, a phenomenon that is often linked to impaired mitochondrial respiration. Results from genes positively correlated with DHODH, strengthening the link between this enzyme, glycolysis, and respiration, prompted us to study these metabolic pathways. Lactate levels were inconsistently modulated upon DHODH inhibition, being either unregulated in PC3 cells (Figure 4J) or decreased in DU145 and LASCPC-01 cells (Figure 4K and L). Interestingly, for the TCA cycle, treatment with orludodstat led to several alterations in metabolite abundance, with a notable and consistent massive increase of fumarate in all studied cell lines, reaching a 6- to 50-fold increase in fumarate levels (Figure 4M−O), results that were also observed with the other DHODH inhibitor DHODH–IN–1 (Supplemental Fig. S3C). The modulation of the TCA cycle, and notably of fumarate, was also observed in the AR-positive LNCaP cells (Supplemental Fig. S3D). Consistent with an alteration of the TCA cycle activity, treatment of PC3, DU145, and LNCaP cells with orludodstat was sufficient to decrease basal and maximal mitochondrial respiration in extracellular flux (seahorse) assays (Figure 4P, Q and Supplemental Fig. S3E–G). Altogether, these results clearly show that DHODH is essential for proper mitochondrial activity in AR-negative PCa cells.

2.4. DHODH inhibition alters central carbon metabolism in PCa cells, rewiring glucose, glutamine, and aspartate usage

To further gain insights into how DHODH inhibition altered TCA cycle metabolite levels, stable isotope tracer analyses were performed. To this end, PC3 cells were selected as a working model, as it was the cell line used to identify DHODH as a metabolic vulnerability (Figure 1). PC3 cells were exposed to ¹³C₆-glucose, with and without orludodstat, before mass spectrometry analysis to track how carbons from glucose were consumed through aerobic glycolysis and the TCA cycle (Figure 5A). Lactate levels were not significantly altered orludodstat exposure (Figure 4J), and no significant alterations in labeling of lactate from ¹³C₆-glucose was observed (Figure 5B). Alanine and serine, two amino acids that can be produced from glucose through glycolysis intermediates, showed higher levels and higher ¹³C labeling upon DHODH inhibition (Figure 5C and D). The increase in both alanine and serine synthesis indicates higher rates of glucose usage in the cytosol. For mitochondrial TCA cycle intermediates, such as citrate, α-ketoglutarate, and succinate, orludodstat did not alter the levels of ¹³C₂ intermediates, indicating that glucose flux into the TCA cycle, if anything, was only modestly altered upon DHODH inhibition (Figure 5E–I). Interestingly, a massive increase in fumarate levels was observed, as reported for PC3, DU145, LASCPC-01, and LNCaP cells (Figure 4M−O and Supplemental Fig. S3D). While the absolute levels of fumarate labeled from glucose were significantly higher (Figure 5H), the proportional contribution of carbons from glucose was lower (6% of fumarate labeled with ¹³C from glucose in orludodstat-treated cells compared to 10% in control cells). Taken together, these results show that DHODH inhibition with orludodstat alters glucose metabolism. However, differential glucose usage could not explain the massive increase in fumarate, suggesting additional pathways feeding into the TCA cycle following DHODH blockade.

Figure 5.

DHODH inhibition reprograms glucose metabolism in PC3 cells. A) Schematic overview of ¹³C₆-glucose metabolism through glycolysis, amino acid synthesis, and the TCA cycle. B–I) PC3 cells were incubated with ¹³C₆-glucose, with and without orludodstat for 24 h, before being harvested for metabolomics. Relative quantification of unlabelled (m+0) and labelled (m+2, +3) metabolites is shown for lactate (B), alanine (C), serine (D), citrate (E), α-ketoglutarate (αKG; F), succinate (G), fumarate (H), and malate (I). Results are shown as average and S.E.M. of one representative experiment (n = 5/group), out of two independent experiments. ∗∗∗p < 0.001; n.s.: not significant.

Thus, we hypothesized that glutamine could be a major carbon donor to the TCA cycle in that context. Indeed, glutamine is the most abundant amino acid in circulation and can enter the cycle by being converted into glutamate and then α-ketoglutarate (Figure 6A). When PC3 cells were treated with ¹³C₅-glutamine, the flux into glutamate was not significantly altered by orludodstat (Figure 6B). However, there was a 2-fold increase in the flux toward α-ketoglutarate, indicating that, contrary to glucose, carbons from glutamine were more used to feed the TCA cycle following DHODH inhibition (Figure 6C). Then, α-ketoglutarate (m+5) can be converted into succinate (m+4) in the canonical TCA cycle or the cycle can go anti-clockwise to generate citrate (m+5). This phenomenon, known as reductive carboxylation, is notably seen under hypoxic stress [42]. Interestingly, succinate m+4 was not significantly modulated by orludodstat, but citrate m+5 levels increased by 2-fold, indicating that reductive carboxylation was favored in PCa cells following DHODH inhibition (Figure 6D and E). Moreover, the flux of carbons from glutamine into malate, and importantly into fumarate, was highly induced (6-fold increased in fumarate m+4; Figure 6F and G). Thus, following DHODH inhibition, not only is glucose metabolism rewired in PCa cells (Figure 5), but glutamine usage is also reprogrammed, notably sustaining a reverse TCA cycle (Figure 6A–G). In electron transport chain (ETC)-deficient cells, the activity of the MDH2 and SDH complexes was shown to be reversed, to promote reductive carboxylation [43]. Here, in the absence of DHODH, while a reductive carboxylation is favored, the lack of increase in flux toward succinate suggests, in AR-negative PCa cells at least, the SDH complex cannot work in reverse.

Figure 6.

Orludodstat induces an aspartate and glutamine metabolic reprogramming by fuelling the TCA cycle. A) Schematic overview of ¹³C₅-glutamine metabolism through the TCA cycle. It can be used through the canonical TCA cycle (clockwise) or a reverse TCA cycle (anti-clockwise; reductive carboxylation). B–G) PC3 cells were incubated with ¹³C₅-glucose, with and without orludodstat for 24 h, before being harvested for metabolomics. Relative quantification of unlabelled (m+0) and labelled (m+1–5) metabolites is shown for glutamate (B), α-ketoglutarate (αKG; C), citrate (D), succinate (E), fumarate (F), and malate (G). H) Schematic overview of ¹³C₁-aspartate metabolism through the TCA cycle. It can be used through the canonical TCA cycle (clockwise) or a reverse TCA cycle (anti-clockwise). I–M) PC3 cells were incubated with ¹³C₁-aspartate, with and without orludodstat for 24 h, before being harvested for metabolomics. Relative quantification of unlabelled (m+0) and labelled (m+1) metabolites is shown for citrate (I), α-KG (J), succinate (K), fumarate (L), and malate (M). Results are shown as average and S.E.M. of one representative experiment (n = 5/group), out of two independent experiments. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; n.s.: not significant.

Given the high increase in fumarate following DHODH inhibition, but not in succinate, we used labelled aspartate (¹³C₁; Figure 6H-M) to test if the SDH complex could sustain fumarate reduction into succinate [44]. Aspartate was previously shown to be used by normal prostate epithelial cells to feed into the TCA cycle, generating malate and then citrate [36]. In PC3 cells, labelling of malate was also observed, being slightly but significantly increased upon treatment with orludodstat (Figure 6M). While citrate m+1 was detected, it was not altered by orludodstat (Figure 6I). Importantly, we observed a 9-fold increase in fumarate m+1 from ¹³C₁-aspartate (Figure 6L), supporting the hypothesis that DHODH inhibition promotes a reverse TCA cycle in PCa cells. However, we observed no labeling of succinate, suggesting a blockade of the SDH complex in that context (Figure 6K). Altogether, these results highlight the capacity of PC3 cells to reprogram their mitochondrial metabolism upon DHODH inhibition, notably fueling glutamine into the TCA cycle.

We then hypothesized that DHODH inhibition could connect pyrimidine synthesis to TCA cycle intermediates and the urea cycle (Figure 7A). Indeed, the first molecule generated from glutamine by CAD is carbamoyl phosphate [45], which is also required to provide nitrogen for the urea cycle. Secondly, aspartate can feed into the urea cycle to provide the second nitrogen of urea, while its carbons are used to generate fumarate. While we observed, upon DHODH inhibition, an increased flux of carbons from aspartate into malate (Figure 6M), the increased flux in fumarate was much higher (Figure 6L), supporting also a more direct connection between aspartate and fumarate. Blockade of DHODH could redirect carbamoyl phosphate and aspartate toward the urea cycle, generating fumarate from aspartate and urea using nitrogen from glutamine. Using our metabolomics method, two intermediates of the urea cycle, ornithine and urea, were detectable. Supporting an increased activity of the urea cycle following DHODH inhibition, both metabolites were at higher levels upon treatment with orludodstat (Figure 7B and C). Finally, to confirm that, in PC3 cells, DHODH inhibition rewires carbamoyl phosphate from pyrimidine synthesis toward the urea cycle, cells were treated with ¹⁵N₁-glutamine (Figure 7A). While no ¹⁵N flux was observed in cells under normal culture conditions, with no labeling detectable in urea, orludodstat treatment led to a massive increase in total urea levels and a labeling of >30% (Figure 7D), in parallel to a major enrichment in DHO as well (Figure 7E). No ¹⁵N labelling was observed in aspartate, further strengthening the connection between glutamine, pyrimidine synthesis, and the urea cycle (Figure 7F). Together, these results confirm that DHODH inhibition promoted a rewiring of de novo pyrimidine synthesis intermediates, downstream glutamine, toward the urea cycle.

Figure 7.

Blockade of DHODH redirects carbamoyl phosphate from de novo pyrimidine synthesis toward the urea cycle. A) Schematic visualization of the pyrimidine de novo biosynthetic pathway and how it could connect to the urea cycle. Glutamine and aspartate are used as CAD substrates to initiate pyrimidine synthesis, with three sequential enzymatic reactions, before the enzymatic activity of the mitochondrial DHODH. The first enzymatic reaction of CAD generates carbamoyl phosphate, an intermediate required to feed nitrogen from glutamine toward urea production. Quantification of two intermediates of the urea cycle, urea (B) and ornithine (C) in PC3. D) PC3 cells were incubated with ¹⁵N₁-glutamine, with and without orludodstat for 24 h, before being harvested for metabolomics. Relative quantification of unlabelled (m+0) and labelled (m+1) isotopomers is shown for urea, dihydroorotate (E), and aspartate (F). G) PC3 cell viability following 96 h treatment with a low dose of orludodstat (10 nM), with and without combination with the GLS inhibitor BTPES, to block glutamine metabolism. H) DU145 cell viability following 96 h treatment with a low dose of orludodstat (10 nM), with and without combination with the GLS inhibitor BTPES. Metabolomics results are shown as average and S.E.M. of one representative experiment, out of two to three independent experiments. Cell viability experiments are shown as the average and S.E.M. of three independent experiments. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Finally, we tested if this metabolic rewiring occurring in PCa cells following the inhibition of DHODH could also be targeted for therapeutic purposes. Given the increased flux of nitrogens from glutamine toward the urea cycle, we hypothesized that targeting metabolism, in combination with orludodstat, could reveal an additional metabolic vulnerability. To this end, we used BPTES, an inhibitor of the GLS enzyme, which catalyzes the conversion of glutamine into glutamate, generating free ammonia (that can then be used through the urea cycle to generate urea). PC3 cells were thus treated with orludodstat at 10 nM, a concentration that decreases by ∼40% cell number, to test if combination with BPTES could increase the anti-proliferative potential (Figure 7G). Alone, BPTES, up to 2 μM, did not affect PC3 cell viability. However, the combination of BTPES and DHODH showed a significantly higher impact on PC3 cell viability (Figure 7G). We confirmed these results in DU145 cells, showing again that the combination of orludodstat and BPTES had a greater impact on PCa cell viability compared to either molecule alone (Figure 7H). At the fixed concentration of 10 nM for orludodstat, the Bliss independence test showed a synergistic effect for the combination of this DHODH inhibitor and BPTES. Together, these results reveal that targeting de novo pyrimidine synthesis, along with impairing glutamine metabolism, has a stronger impact on PCa cell.

3. Discussion

PCa remains a major public health issue worldwide, being the most frequent cancer affecting males in over 100 countries [1]. Here, using a high-throughput anti-metabolic drug screening, the DHODH enzyme was identified as a novel metabolic vulnerability in AR-negative PCa cells, the most complex form of PCa to treat due to the lack of targeted therapies. DHODH inhibition, notably using the orally bioavailable molecule orludodstat, was shown to be effective in slowing down cancer cell proliferation, even inducing high levels of cell death in the neuroendocrine PCa model LASCPC-01 cells and inhibiting the proliferation of AR-negative PCa cell lines. We also further validated previous results showing its potential to inhibit AR-positive PCa cells, such as using LNCaP cells. At the cellular level, DHODH inhibition was shown to reprogram cancer cell metabolism; not only was pyrimidine synthesis impaired, but also glucose, glutamine, and aspartate usage, highlighting a global metabolic reprogramming occurring following the blockade of this mitochondrial enzyme (Figure 8). Understanding these molecular consequences of DHODH inhibition also revealed an additional metabolic vulnerability, as GLS inhibition was shown to strengthen the proliferation blockade of PCa cell following treatment with orludodstat. Together, our results highlight the potential of drug repurposing targeting DHODH for the treatment of both AR-positive and AR-negative PCa, along with highlighting the molecular consequences of DHODH inhibition on cell metabolism, linking de novo pyrimidine synthesis to the TCA and urea cycles.

Figure 8.

Working model of metabolic reprogramming following DHODH inhibition in prostate cancer cells. In normal conditions (control), DHODH links the de novo pyrimidine synthesis pathway and the electron transport chain. Without metabolic stress, glucose feeds glycolysis and the TCA cycle, while glutamine and aspartate can be used for de novo pyrimidine synthesis. However, when DHODH is inhibited, the fluxes of glutamine and aspartate are rewired. First, because of the massive increase in dihydroorotate, carbamoyl phosphate synthesized from glutamine is redirected toward the urea cycle, which shows a significant increase in activity. Second, aspartate, which cannot be used further due to the blockade of pyrimidine synthesis, is also used through this urea cycle, generating a massive increase in cellular fumarate levels. Third, glutamine metabolism is also redirected toward the TCA cycle, where it can feed into the canonical cycle along with reductive carboxylation, a stress response that can be block with the GLS inhibitor BPTES. Fourth, glucose, instead of feeding the TCA cycle, is used to synthesize amino acids. Together, the results of the current study demonstrate that, in AR-negative PCa cells, DHODH inhibition forces cells to reprogram their central carbon and nitrogen metabolism.

Targeting nucleotide synthesis is one of the oldest chemotherapy approaches, with various classes of inhibitors developed over time, including methotrexate, which inhibits one-carbon metabolism (with the folate cycle), and 5-fluorouracil, which blocks pyrimidine synthesis. In line, DHODH inhibition has been studied in various cancer settings, alone or in combination, such as for leukemias and brain tumours [[14], [15], [16]]. In the context of PCa, high DHODH expression was recently reported to be associated with poorer survival [19], and the results shown here are in accordance with this study. In this study, DHODH inhibition was shown to be efficient in AR-positive PCa models [19]. However, a consequence of this treatment was to induce steroidogenesis; hence, treatment with orludodstat, also known as BAY 2402234, had to be combined with a steroidogenesis inhibitor, abiraterone, to induce maximal effects in xenograft assays in vivo [19]. Our results with LNCaP cells suggest that DHODH plays a similar metabolic role in AR-positive and AR-negative cells, linking de novo pyrimidine synthesis, the TCA cycle and the urea cycle. The results shown here in AR-negative models bypass the need to combine with abiraterone, as AR is not expressed, and thus, the possible induction of androgen biosynthesis will not activate AR. This is particularly important for tNEPC, for which only limited therapeutic options are available. However, as tumours are often heterogeneous, exhibiting both AR-positive and AR-negative cancer cells in castration-resistant tumors, combining abiraterone acetate is likely important to ensure complete blockade of oncogenic pathways. In such clinical settings, demonstrating that both AR-positive and AR-negative cells are sensitive to DHODH inhibition is even more important, as this should be more effective in controlling the progression of castration-resistant PCa. Future clinical studies are now required to test this hypothesis and demonstrate if DHODH inhibition can be positively used for patients with advanced PCa.

At the metabolic level, DHODH, by being the only de novo pyrimidine synthesis enzyme localized at the mitochondria, can connect this pathway to mitochondrial respiration and the TCA cycle [46,47]. We initially hypothesized that glucose metabolism would be exacerbated upon DHODH inhibition. Yet, the highest increase in carbon flux from glucose was observed in biosynthetic cytosolic pathways, notably for alanine and serine synthesis. Interestingly, glutamine flux toward the TCA was increased. Glutamine is the precursor of the de novo pyrimidine synthesis pathway, and the most abundant amino acid in the organism [48]. Following the DHODH inhibition, which impairs the usage of this amino acid for pyrimidine synthesis, PCa cells increased its flux toward the TCA cycle. Glutamine was used both in the canonical TCA cycle and in reductive carboxylation, highlighting this amino acid as an important stress response mechanism in PCa cells.

The second metabolite of the de novo pyrimidine synthesis pathway is carbamoyl phosphate, produced using nitrogen from glutamine. Carbamoyl phosphate is also required to fuel the urea cycle. Indeed, in the liver, carbamoyl phosphate is required with ornithine to produce citrulline through this cycle [49]. However, to our knowledge, it is unclear how these two metabolic pathways can crosstalk under cellular stress conditions. Here, using ¹³C₅-glutamine and ¹⁵N₁-glutamine, we showed that PCa cells, following DHODH inhibition, increased the flux into carbamoyl phosphate, which probably accumulates, and that is then rewired from de novo pyrimidine synthesis to the urea cycle. As such, stable isotope tracer analysis revealed an increase of labelled dihydroorotate in m+1, a read-out of CAD activity. Furthermore, we observed an increase in urea labelling, attesting to the link between pyrimidine synthesis and the urea cycle in AR-negative PCa cells. Finally, in the urea cycle, argininosuccinate lyase (ASL) catalyzes the reaction of arginosuccinate into arginine by releasing fumarate [50]. Thus, the massive accumulation of the fumarate level when DHODH is inhibited is due partially to an increase in the urea cycle activity, as also supported by studies using labelled aspartate.

Aspartate is an essential amino acid required for de novo nucleotide synthesis, downstream of glutamine. Most previous studies have shown that aspartate can be produced by malate from the TCA cycle, thus being often studied as an output metabolite from this pathway [[51], [52], [53]]. However, in specific contexts, aspartate was reported to feed into the TCA cycle; for example, aspartate was shown to be metabolized not into citrate but into fumarate and succinate by reverse SDH activity [54]. Interestingly, the normal prostate exhibits a unique mitochondrial TCA cycle that is truncated after citrate synthesis, through a blockade of the aconitase 2 (ACO2) enzyme, to produce and secrete massive amounts of citrate, a key function of the prostate in male fertility [55]. We have previously shown that in normal prostate models, aspartate can be used to fuel the TCA cycle and citrate synthesis for secretion [36]. During carcinogenesis, this citrate-secretory metabolic profile is reprogrammed so that PCa cells utilize citrate instead of secreting it, although the molecular details are still lacking [55]. In control PC3 cells, we observed 38%, 30%, and 6% of total labelling of citrate from glucose, glutamine, and aspartate, respectively. Considering the media concentrations of glucose (10 mM), glutamine (2 mM), and aspartate (1 mM), these results show that these nutrients, including aspartate, contribute to the citrate production even in PCa cells. Moreover, for other intermediates such as malate, the contribution of glucose, glutamine, and aspartate was 17%, 28%, and 10%, respectively (in controls), again highlighting that aspartate, even though present at only 1/10 of glucose concentration, can be readily used to feed into the TCA cycle. These results also highlight the important contribution of glutamine, with often higher labelling than glucose, to the mitochondrial TCA cycle of PC3 cells. Importantly, following DHODH inhibition, PCa cells increased the flux of aspartate, not in the TCA cycle, but in the urea cycle, probably to “absorb” the accumulating carbamoyl phosphate, thus generate high levels of fumarate. Altogether, these results highlight the importance of both glutamine and aspartate in PCa metabolic stress response, feeding the TCA and urea cycles under normal or stressful conditions. The current study thus demonstrates the connection between de novo pyrimidine synthesis and the urea cycle in PCa cells, and it remains to be determined if this relationship also occurs in other cancer types.

Furthermore, this understanding of the molecular consequences of DHODH inhibition allowed us to identify glutamine metabolism, and more specifically the GLS enzyme, as an efficient anti-metabolic combination approach for PCa cells. Indeed, metabolism generally reacts rapidly to external stresses, such as drugs, and combination therapies are now often studied for efficient and successful clinical translation of anti-metabolic approaches. Our results are similar to the ones obtained by Udutha et al. In their study, the authors showed that, in glioblastoma cancer cells, the combination of GLS inhibition with glutathione synthesis created a synthetically lethal metabolic vulnerabilities [56]. In PCa, to our knowledge, the combination of glutamine and pyrimidine metabolism inhibition has never been investigated. Dasgupta and colleagues showed that a PC3 cell derivative, stably expressing an shNTC and thus being distinct from the parental PC3 cells, were sensitive to BPTES [57]. In other cancer settings, glutamine was shown to be required for de novo pyrimidine synthesis, such as in renal and breast cancers [58,59], but has not yet been tested in combination with DHODH inhibitors. As shown in the present study, the combination of DHODH and GLS inhibition was synergistic, at least in PCa cells and it will be interesting in the future to determine if such a combination is also efficient in other cancer cell types.

In conclusion, DHODH inhibition appears as a novel therapeutic avenue for AR-negative, AR-independent PCa cells, at least in preclinical models. Given that DHODH inhibitors are already available in the clinic, for other indications, this could be rapidly repurposed for patients with these lethal AR-negative PCa, including those exhibiting neuroendocrine differentiation, or with a heterogeneous disease comprised of AR-positive and AR-negative cancer cells. Interestingly, DHODH inhibition also revealed metabolic specificities of PCa cells, such as the reliance on glutamine and aspartate as a metabolic stress response, further demonstrating the connection between de novo pyrimidine synthesis, the TCA cycle, and the urea cycle. This mechanistic understanding of the molecular consequences of DHODH inhibition further revealed increased sensitivity of PCa cells, upon pyrimidine synthesis, to GLS inhibition, supporting the idea of future anti-metabolic combination therapies.

4. Materials and methods

4.1. Cell culture

PC3, DU145 and LNCaP cells were purchased from the ATCC and cultivated in RMPI supplemented with 10% fetal bovine serum (FBS, Wisent #090150), penicillin, streptomycin (P/S Wisent #450-201-EL), and sodium pyruvate (ThermoFischer Scientific #11360070). LASCPC-01 cells were cultivated following the protocol established by Lee et al. [60], being grown in modified medium HITES made with RMPI supplemented with 5% FBS, penicillin, streptomycin, transferrin at 0.01 mg/mL (Sigma–Aldrich #T5391), insulin at 4 mg/mL (Gibco #12585-014), hydrocortisone at 10 mM (Sigma–Aldrich #H0135), sodium selenite at 100 μg/mL (Sigma–Aldrich #S9133), and β-estradiol at 1 μg/mL (Sigma–Aldrich #E8875). The confluence of all cell lines was maintained below 75% and the media was changed every two or three days. Mycoplasma presence was verified every 4 months.

4.2. High-throughput anti-metabolic drug screening

The anti-cancer metabolism compounds library was purchased from TargetMol. PC3 cells were seeded in RPMI medium at a density of 2,000 cells per well in 96-well plates. After 24 h, cells were treated with compounds from the library, with each compound being at a final concentration of 100 nM. After 96 h treatment, nuclei were stained by adding 10 μg/mL of Hoechst solution for 30 min, before being automatically counted using a Cytation 5, as previously described [61]. For each molecule tested, three independent replicates were performed (on independent cell culture plates).

4.3. Proliferation assays

PC3, DU145 and LNCaP cells were seeded in 6-well plates at 60,000 cells per well in RPMI medium. LASCPC-01 cells were seeded in 96-well plates at 200,000 cells per well in HITES medium. 24 h after seeding, cells were treated with 100 nM Orludodstat, 750 μM uridine, or both. For the other DHODH inhibitor, teriflumonide (Cedarlane #HY-110159S), brequinar (Cedarlane #24445-1, and DHODH–IN–1 (Cedarlane #HY-135282) were used at 10 μM, 100 nM, and 100 nM, respectively. 120 h after seeding, cells were trypsinized and counted with a TC10 Automated cell counter (Bio-Rad).

4.4. Viability assays

PC3, DU145, and LNCaP cells were seeded in 96-well plates at 2,000 cells per well in RPMI medium. LASCPC-01 cells were seeded at 20,000 cells per well in 96-well plates in HITES medium. PC3 and DU145 were treated with a concentration range of 0.5 nM–100 nM, and LASCPC-01 with a concentration range of 0.01 nM–100 nM, 24 h after seeding in a total volume of 200 μL. 120 h post seeding, 20 μL of MTS solution (Promega #G3581) was added per well, and plates were incubated at 37 °C. Then, absorbance was read using the spectrophotometer TECAN M1000 at 490 nm. Cell viability was obtained by converting absorbance and expressing it as a percentage of the control. For the other DHODH inhibitor, teriflumonide, brequinar, and DHODH–IN–1 were used at 10 μM, 100 nM, and 100 nM, respectively. For the combination between orludodstat and BPTES treatment, a concentration range from 0.1 μM to 2.0 μM of BPTES (Cedarlane #HY-12683) was used with or without 10 nM orludodstat. We used the Bliss independence model, using a fixed concentration for orludodstat, to provide a point–wise estimation of synergy.

4.5. PDO primary culture and viability assay

PDOs were obtained from consenting subjects’ prostate biopsies by the project approved and framed by the research ethics committee of the CRCHUQ-UL (2021–5661). Prostate samples were obtained from freshly removed radical prostatectomy patients in neoplastic regions, as previously described [37,62], and the presence of cancer was confirmed by a pathologist. To grow PDOs, 15,000 cells were plated in 40 μL droplets with 75% Matrigel (Corning#356234) in 24-well plates with a 3D culture medium, as previously described [37,62]. This medium is composed of KSFM supplemented with 4% B27 Supplement (Stemcell Technologies #5711), 10 μM Y27632 dihydrochloride (TOCRIS #1254), 10 μM calcitriol (Cayman Chemical #71820), 500 nM A83-01 (Sigma–Aldrich #SML0788), 10 nM testosterone, 10 nM estradiol (Sigma–Aldrich #E8875), 1% R-Spondin, and 1% Noggin. Orludodstat was added at seeding (day 0) or 3 days after seeding, at a final concentration of 100 nM. Culture media was renewed 3 times a week. After 14 days of culture, 650 μL was removed to leave 100 μL per well. Then, 100 μL of CellTiter-Glo 3D solution (Promega #G9681) was added. The plate was incubated at room temperature for 25 min in obscurity. Then, 200 μL of the solution was transferred to a 96-well black plate with a transparent flat, and bioluminescence was read by a BioTek Synergy H1 (Agilent). For the rescue experiment, 100 nM orludodstat, 750 μM uridine, or both were added at the seeding of the PDO4 line (day 0) and renewed 3 times a week for 14 days. Readout was the same as previously described. Four PDO lines were used in the current study, labelled CW513T, CW561T, CW680T and CJ241T in our local PDO biobank, and referred herein as PDO1, PDO2, PDO3 and PDO4, respectively.

4.6. Extracellular flow (seahorse) assays

PC3, DU145 and LNCaP were seeded at 10,000 cells per well in 96-well Seahorse XFe96 microplates. After 24 h, the medium was removed and replaced with minimal DMEM without HEPES and supplemented with 1% penicillin and streptomycin, 1 mM sodium pyruvate, 10 mM glucose, and 2 mM glutamine. pH was adjusted to 7.4. After 1 h of equilibration at 37 °C, the plate was transferred to the XFe96 instrument (Agilent). The program used was: 1) 6 measurement cycles of basal oxygen consumption rates (OCR) and extracellular acidification rates (ECAR); 2) a first injection of orludodstat at a final concentration of 100 nM, or DMSO, followed by 95 measurement cycles of OCR and ECAR; and 3) then, a regular mitochondrial stress test was performed, with sequential injections of oligomycin, FCCP, and a combination or rotenone and antimycin A. At the end of the assay, cells were counted using CyQUANT (ThermoFisher Scientific #C7026) to normalize OCR and ECAR data for cell number.

4.7. Targeted metabolomics and stable isotope tracer analyses

PC3, DU145, LASCPC-01 and LNCaP were seeded at 2 × 10⁶ cells per cell culture dish. Forty hours later, cells were treated with 100 nM orludodstat or vehicle (DMSO), or DHODH–IN–1 at 100 nM for PC3 cells. After a treatment of 24 h, cells were harvested for GC–MS analysis as previously described [62]. In brief, after washing the cells with cold saline, cell extracts were harvested in dry ice-cold 80% methanol. After derivatization, metabolites were analyzed using an Agilent 8890 GC equipped with DB5-MS + DG capillary coupled to an Agilent 5977 B MS instrument. Analyses were done using the MassHunter Workstation Software (Agilent) and the 257 NIST/EPA/JIH Mass Spectral Library (NIST 2.3, 2017). Results were normalized by the cell counting of cell culture dishes prepared in parallel to those used for metabolomcis. Student's T-tests were performed to determine statistical ignificance.

4.8. Stable isotope tracer analysis

For stable isotope tracer analyses, PC3 cells were cultured as described above. Twenty-four hours after seeding, tracers were added individually to the cell culture media. Various tracers were used: ¹⁵N₁-glutamine (#NLM-557-1 Cambridge Isotope Laboratories, Inc), ¹³C₅-glutamine (#CLM-1822-H-0.1 Cambridge Isotope Laboratories, Inc), ¹³C₁-aspartate (#489972 Sigma–Aldrich), and ¹³C₆-glucose (#CLM-1396-1 Cambridge Isotope Laboratories, Inc). Glutamine, aspartate, and glucose were used at a final media concentration of 2 mM, 1 mM, and 10 mM, respectively. After another 24 h, cells were harvested for targeted metabolomics as described above.

4.9. Public data reanalysis

Public TCGA prostate adenocarcinoma sequencing data were analyzed from the UCSC Xena platform during Spring 2025 (https://xena.ucsc.edu/) to get clinical and expression data. Only primary tumors and normal tissues were kept for analysis. For DHODH expression data in human normal or tumorous prostate, we used data from GSE25136, which included 40 non-recurrent PCa and 39 recurrent PCa [33]. GSE3325 was also used and included 6 benign and 13 tumoral prostate tissues [32]. Analyses were also confirmed with GSE21032GPL10264 [34], which is composed of 19 normal adjacent benign prostate tissues, 150 tumoral prostate tissues. For DHODH co-dependencies, the Depmap.org portal was used and the top 100 co-dependency genes were studied using the Metascape portal [40]. For DHODH correlation analysis with AR or NEPC biomarker genes, the cbioportal for cancer research was used for analysis [30,31].

4.10. Western blot

For the early apoptosis detection, PC3, DU145, and LASCPC-01 cells were harvested after been treated for 24 h at 100 nM orludodstat. Medium was removed, and lysis buffer K supplemented with protease and phosphate inhibitors was added to the cellular pellet. Lysis buffer K was made with 20 mM Na₂HPO₄, 0.15 M NaCl, 0.1% NP40, 5 mM EDTA, and CHAPS 0.01 % proteinase inhibitor (Sigma–Aldrich #P8340-5 ML) and a phosphatase inhibitor cocktail tablet (Roche#04-906-837-001) was added for 10 mL of Lysis buffer K. For the apoptosis signaling, western blot analysis was performed with primary antibodies DHODH (Cell Signaling Technology #26381), cleaved PARP (Santa Cruz Biotechnology #sc-56196), α-tubulin (Cell Signaling Technology #2125); ribosomal protein S6 (Santa Cruz Biotechnology #sc-74459); and AR statut (Santa Cruz Biotechnology #sc-7305).

4.11. Flow cytometry

PC3 and DU145 were seeded in a 15 cm dish and were treated with 100 nM orludodstat on day one. After 120 h, cells were trypsinized to obtain a single cell suspension and 500,000 cells were next used in FACS tubes in a mix of 70% Ethanol with PBS for a total volume of 500 μL. After 2 h on ice, cells were centrifuged and the ethanol/PBS mix was removed. The staining mix was added, made with 100 μg/mL of RNase A (ThermoFisher Scientific #EN0531), 0.1% Triton X-100, and 25 μg/mL propidium iodide solution (ThermoFisher Scientific #P3566) in PBS for a final volume of 500 μL per sample. For each condition, stained and unstained tubes were prepared. Tubes were kept at 4 °C on ice overnight before analysis using a FACSymphony A1 (BD Biosciences) at the flow cytometry facility of the Centre de recherche du CHU de Québec – Université Laval.

CRediT authorship contribution statement

Maxime Labroy: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Marc-Oliver Paré: Writing – review & editing, Validation, Investigation. Line Berthiaume: Writing – review & editing, Methodology, Investigation. Mélissa Thomas: Writing – review & editing, Methodology. Cynthia Jobin: Writing – review & editing, Investigation. Alain Veilleux: Writing – review & editing, Resources. Martin Pelletier: Writing – review & editing, Resources. Frédéric Pouliot: Writing – review & editing, Resources, Methodology, Funding acquisition. Jean-Yves Masson: Writing – review & editing, Methodology, Conceptualization. Étienne Audet-Walsh: Writing – original draft, Supervision, Investigation, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.