Asparagine Dependency is a Targetable Metabolic Vulnerability in TP53-Altered Castration-Resistant Prostate Cancer

Abstract

The TP53 tumor suppressor is frequently altered in lethal, castration-resistant prostate cancer (CRPC). However, to date there are no effective treatments that specifically target TP53 alterations. Using transcriptomic and metabolomic analyses, we showed here that TP53-altered prostate cancer (PCa) exhibits an increased dependency on asparagine and overexpresses asparagine synthetase (ASNS), the enzyme catalyzing the synthesis of asparagine. Mechanistically, loss or mutation of TP53 transcriptionally activated ASNS expression, directly as well as via mTORC1-mediated ATF4 induction, driving de novo asparagine biosynthesis to support CRPC growth. TP53-altered CRPC cells were sensitive to asparagine restriction by knockdown of ASNS or L-asparaginase treatment to deplete the intracellular and extracellular sources of asparagine, respectively, and cell viability was rescued by asparagine addition. Notably, pharmacological inhibition of intracellular asparagine biosynthesis using a glutaminase inhibitor and depletion of extracellular asparagine with L-asparaginase significantly reduced asparagine production and effectively impaired CRPC growth. This study highlights the significance of ASNS-mediated metabolic adaptation as a synthetic vulnerability in CRPC with TP53 alterations, providing a rationale for targeting asparagine production to treat these lethal prostate cancers.

Introduction

Advanced prostate cancer (PCa) is a major cause of cancer mortality worldwide (1). Despite improvements in new and more potent targeted therapeutics, e.g. novel androgen receptor pathway inhibitors, therapeutic resistance is common leading to lethal castration resistant prostate cancer (CRPC) (2–4). Analysis of CRPC tumors, coupled with animal modeling studies, have established a prominent role for TP53 alterations in therapeutic resistance (5–8). For example, whole-exome sequencing (WES) from the SU2C/PCF metastatic prostate cancer cohort showed a significant occurrence of deletions/mutations in the TP53 gene (5). However, despite the prominence of TP53 as a cancer driver, targeting it therapeutically has been challenging. Thus, a better understanding of the molecular features of treatment-refractory CRPC with TP53 alterations is necessary to help identify more effective therapeutic modalities. To this end, we have developed and analyzed a pair of transgenic PCa models based on the deletion of the Pten tumor suppressor gene with or without additional deletion of Tp53. We reasoned that a comparison of Pten knockout with Pten/Tp53 double knockout tumors following androgen-deprivation may identify targetable Tp53-driven pathways important for castration resistance. Our unbiased transcriptomic and metabolomic analyses point to an important role for metabolic adaptation in therapeutic resistance.

Under stressful conditions such as oncogenic stress, cancer cells can reprogram cellular metabolism that provides diverse energy sources and metabolites required for macromolecular biosynthesis to support cell proliferation and survival. Amino acids are essential precursors for nucleotide, protein and lipid biosynthesis and serve as metabolic fuels providing energy via the tricarboxylic acid (TCA) cycle and oxidative phosphorylation (9). Thus, alterations in amino acid synthesis and catabolism are commonly found in multiple types of cancer (9–11). Targeting amino acid metabolism by limiting nutrient availability has been reported to be effective in suppressing cancer growth and metastasis (12,13). Additionally, several studies suggest that inactivation of TP53 plays an important role in rewiring amino acid metabolism to optimize nutrient utilization by regulating enzymes involved in amino acids biosynthesis (14,15). For instance, TP53 has been implicated in regulating glutamine, serine and glycine metabolism by repressing the expression of corresponding biosynthetic enzymes, and TP53-deficient tumor cells are more sensitive to glutamine or serine/glycine depletion (16–20).

Asparagine (Asn), a non-essential amino acid that is synthesized from aspartate (Asp) through asparagine synthetase (ASNS) activity, contributes to support protein and nucleotide synthesis during tumor progression (21,22). Early investigations showed that tumors expressing low levels of ASNS, such as acute lymphoblastic leukaemia (ALL) and some forms of acute myeloblastic leukaemia (AML), are sensitive to bacterial L-asparaginase (ASNase) treatment due to their reduced intracellular asparagine levels resulting from low ASNS activity (23–25). Furthermore, increased expression of ASNS contributes to the resistance of these cells to ASNase (26–28). Thus, targeting both the intracellular and the extracellular sources of asparagine might be crucial for maximal therapeutic efficacy. Recent studies indicate that elevated ASNS expression and asparagine availability are strongly correlated with malignant phenotype and poor clinical outcomes in multiple solid cancers including prostate cancer (29–33). By integrating metabolomic with transcriptomic and epigenomic data of CRPC from patient samples, genetically engineered mice (GEM), PDX models and cell lines, we found that loss/mutation of TP53 promote asparagine synthesis to facilitate castration-resistant growth of PCa cells.

Methods

Mouse strains, procedures, and xenograft models

The following mouse strains were from the Jackson Laboratory (Bar Harber, ME): Bmi1CreER, Pten^f/f, and Tp53^f/f. General procedures for tamoxifen administration, surgical castration, and dissociation of prostate glands and urogenital sinus mesenchyme (UGM) cells have been described previously (34,35). For renal capsule grafting, dissociated prostate epithelial cells were mixed with dissociated UGM cells in 50 ul of 3:1 collagen:setting solution (designated as tissue recombinants (TRs)). TRs were implanted under the renal capsules of male NOD/SCID mice (4–6 weeks). For CRPC patient-derived xenografts (PDX) models, TM00298 PDX model of PCa was obtained from the Jackson Laboratory and adopted to CRPC growth in NSG mice after castration in our lab. LAPC9 xenograft cells were a generous gift from Dr. Dean Tang laboratory at Roswell Park and propagated into NSG mice as previously described. For LNCaP/AR CRPC xenograft, isogenic LNCaP/AR cells were kindly provided by Dr. Charles Sawyers (Memorial Sloan Kettering Cancer). Minced tumor fragments or 2 × 10⁶ LNCaP/AR cells (100 μl, 1:1, PBS: Matrigel) were injected subcutaneously into the flank of intact or castrated NSG mice. When the mice reached the experimental endpoint, tumor tissues were harvested and processed for further analyses. All in vivo xenograft studies were performed in male NSG mice (Jackson Laboratory) and all mice were housed in a pathogen-free animal barrier facility. Tumor development was monitored over time by palpation and caliper measurements once per week. All mouse studies were approved by the Northwestern University Institutional Animal Care and Use Committee.

Histology and immunostaining

Tissues were fixed with 10% neutral buffered formalin for 24h at 4 °C, then transferred to 70% ethanol until paraffin processing, embedding and sectioning. Hematoxylin/eosin staining and immunostaining were performed as described (35). The following primary antibodies were used: CK8 (MMS-162P, Covance; 1:500), AR (RB-9030-P, Thermo Scientific; 1:200), p63 (sc-8343, Santa Cruz; 1:50), BrdU (BU1/75 (ICR1), Abcam; 1:100), Chromogranin A (ab15160, Abcam; 1:500), ASNS (14681–1-AP, Proteintech; 1:500), and p53 (sc-126, Santa Cruz; 1:500). For immunofluorescence staining, slides were incubated with secondary antibodies labeled with Alexa Fluor 488, 594, or 647, then counterstained with DAPI (Sigma), followed by mounting with ProLong Gold Antifade reagent (Invitrogen/Molecular Probes). For IHC, slides were incubated with ImmPRESS HRP anti-rabbit (Vector#MP-7401) or anti-mouse (Vector#MP-7402) following AEC peroxidase substrate (Vector #SK-4200) according to the manufacturer’s instructions. Images were visualized in Axioskop 40 FL microscope.

RNA-seq

Total RNA was extracted from mouse prostate grafts 5 months post-castration using the RNAeasy Plus mini kit (Qiagen) according to the manufacturer’s instructions. RNA was submitted to the NUSeq Core at Northwestern University for quality testing, library construction, and RNA-seq analysis. High-throughput sequencing with 50 bp single-end reads was performed using the Illumina HiSeq 4000, and reads were aligned to hg38 using Bowtie2. Gene expression was quantified using htseq, and differential expression determined using DESeq2. Gene Set Enrichment Analysis (GSEA) of differentially expressed genes (FDR-corrected p-value ≤0.05) was used for pathway analysis using GSEA software from the Broad Institute.

Cell lines and culture

Human prostate cancer cell lines, LNCaP, C4–2B, DU-145, and PC3 cells were maintained in RPMI 1640 media. LAPC4 cells were cultured in Iscove’s medium. HEK‐293T cells for lentivirus production were cultured in high glucose DMEM. All the mediums were supplemented with 10% fetal bovine serum (FBS, Life Technologies no. 10437–028) and 1% penicillin/streptomycin antibiotic solution (Life Technologies no. 15140–122), unless indicated otherwise. All in vitro experiments were conducted using cell lines cultured for 2–3 weeks in medium containing charcoal-stripped FBS. All cells were verified to be mycoplasma-free (Lonza) and genetically authenticated by ATCC.

Plasmids and the gene knockdown experiment

shRNA vectors against ASNS (TRCN0000290113, TRCN0000290105; showing similar results) were purchased from Sigma. Lentiviral constructs for TP53 were generated by cloning the cDNA into pFUGW-H1 vector. Lentiviruses were produced by transfecting 293T cells using Δ8.9 packaging vector and VSVG envelope vector (2:1:1) in Opti-MEM media (Gibco). Cells were transduced with the virus in the presence of 8 μg/ml polybrene (Invitrogene) and 1 μg/ml of puromycin was added to select stably expressing cells. To knockdown ATF4, the human siGENOME ATF4 Smart Pool (L-005125–02) and siControl Non-Targeting siRNA (D-001210–02) were purchased from Dharmacon (Lafayette, CO). siRNAs were transfected into cells by using Lipofectamine transfection reagent.

Cell proliferation assay

Cells were seeded in triplicate at a density of 4 × 10³ cells per well in 96-well culture plates or at 5 × 10⁴ cells per well in six-well plates 24 hours prior to starting the experiment. The medium was then replaced with DMEM supplemented with 10% charcoal-stripped or dialyzed FBS (Cytiva HyClone^™) medium, with or without 0.1 mM asparagine, in the presence or absence of DMSO (vehicle), CB-839 (HY-12248, MedChemExpress), or E.coli-ASNase (ENZ-287, Prospec Bio). Relative proliferation was determined in triplicate wells by CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, G7571), Cell Counting Kit-8 (CCK-8) assay (Dojindo Molecular Technologies), or manual counting 3–5 days post-treatments, according to the manufacturer’s instructions. For colony formation assays, 3000 cells were plated in six-well plates in triplicate with medium containing corresponding drugs. The medium were replaced with fresh medium every 3 days and when the colonies were clearly visible, they were fixed with 100% methanol and stained them with crystal violet (0.1%) for 1h. Subsequently, plates were rinsed with PBS and colonies were counted using the Image J software to calculate the percentage of the area covered by colonies.

Annexin V-PE staining and FACS analysis

The apoptosis rate was determined by a PE Annexin V Apoptosis Detection Kit I (BD Biosciences, no. 559763). Briefly, cells were collected and washed with ice cold 1 X Cell Cycle Assay Buffer, then resuspended in 1 × Binding Buffer at a concentration of 1 × 10⁶ cells/ml. Next, 100 μl of Annexin V-PE and 5 μl 7-aminoactinomycin-D (7-AAD) were added to the cell suspension. Following a 15 min incubation at RT in the dark, flow cytometry was used to analyze apoptosis. Annexin-V+/7-AAD- and Annexin-V+/7-AAD+ cells were identified as apoptotic cells.

Cell cycle assay

The Cell Cycle Analysis Kit (Abcam, no ab287852) was used to analyze the cell cycle, following the manufacturer’s protocol. Cells were collected and washed with ice-cold PBS, then fixed by adding ice-cold 70% ethanol and placed on ice for a minimum of 30 min. After washing the cells with 1X Cell Cycle Assay Buffer, they were completely resuspended in Staining Solution and incubated at room temperature for 30 min. Flow cytometry measurements were then conducted.

Western blot analysis

Whole-cell extraction was conducted using RIPA buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, and 1% NaDeoxycholate [pH 7.4]) supplemented with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 μg/ml peptasin A, 10 μg/ml aprotinin, and 5 μg/ml leupeptin). Protein concentrations were then measured using Bio-rad protein assay kits (BioRad, Hercules, CA). Next, the protein lysates were resolved by SDS-PAGE, and then transferred onto nitrocellulose membranes, blocked with PBS containing 0.2% Tween 20 and 5% non-fat dry milk, and incubated with primary antibody. Primary antibodies were as follows: anti-ASNS (14681–1-AP, Proteintech; 1:1000), p53 (DO-1, sc-126, Santa Cruz; 1:1000), p53 (FL-393, sc-6243, Santa Cruz; 1:1000), ATF4 (D4B8, 11815, Cell Signaling; 1:1000), p21 (10355, Proteintech; 1:1000), phospho-S6K (108D2, 9234, Cell Signaling; 1:1000), S6K (49D7, 2708, Cell Signaling; 1:1000), phospho- 4E-BP1 (236B4, 2855, Cell Signaling; 1:1000), 4E-BP1 (53H11, 9644, Cell Signaling; 1:1000), phospho-GCN2 (E1V9M, 94668, Cell Signaling; 1:1000), GCN2 (ab75836, Abcam, 1:1000), phospho-PERK (16F8, #3179, Cell Signaling; 1:1000 or MA5–15033, Thermo Fisher, 1:1000), PERK (C33E10, 3192, Cell Signaling; 1:1000), eIF2α (9722, Cell Signaling; 1:1000), phospho-eIF2α (Ser51) (119A11, 3597, Cell Signaling; 1:1000), PHGDH (13428, Cell Signaling; 1:1000), PSAT1 (20180–1-AP, Proteintech; 1:1000), MTHFD2 (12270–1-AP, Proteintech; 1:1000), AARS (A303–475A-M, Bethyl Antibodies: 1:1000), and anti-β-actin (Sigma, 1:5000).

PDX organoid assays

PDX organoids were generated as previously described (36,37). For lentiviral transduction, dissociated PDX cells were transduced with shNT, shTP53, shControl, or shASNS and spinfected for 3h at 300g with virus containing media. After spinfection, virus-containing media was replaced with organoid media and 5000 cells were plated in Ultra-Low Attachment Surface plates (Corning). At 2 days post-transduction, organoids were treated with either DMSO, 0.5 U/ml ASNase, or 0.5 μM CB-839. At day 5 after treatment, 3D viability was assessed using CellTiter-Glo^® 3D Cell Viability Assay (Promega) and organoids were collected for protein analysis.

ATAC-seq

ATAC-seq analysis was performed as previously described (38). Nuclei were isolated with nuclear lysis buffer (10 mM tris, 10 mM NaCl, 3 mM MgCl2, and 0.5% IGEPAL-630) and centrifuged at low speeds. The nuclei were resuspended in 50uL of transposase reaction mixture (22.5uL nuclease-free water, 25uL of TD buffer and 2.5uL of TDE1 enzyme, Illumina #20034197). Transposition was carried out at 37°C for 30 min, followed by DNA purification with DNA Clean and Concentrator-5 kit (Zymo Research) according to the manufacturer’s recommendation. After purification, library fragments were PCR-amplified with Nextera XT v2 adapter primers. Multiplexed libraries were sequenced on the NextSeq 500 (Illumina) for paired end 37nt. Reads were aligned to hg38 using Bowtie2. Coverage tracks were generated using deepTools and peaks visualized in Integrative Genomics Viewer (IGV) browser.

ChIP-qPCR

ChIP was performed using the Ideal ChIP-qPCR Kit from Diagenode (C01010180) following the manufacturer’s instructions. In brief, cells were crosslinked with fresh 1% formaldehyde for 10 min at room temperature. Chromatin was sheared to ~200 base pairs and followed by immunoprecipitation with indicated antibodies. Antibody used for ChIP was p53 FL-393 (sc-6343, Santa Cruz), p53 DO-1 (sc-126, Santa Cruz), and ATF4 (11815, Cell Signaling). Primers sequences as follows: ASNS_−657: 5’-CCGCTCGAGATTTCTCAATTTATTTCGG-3’,5’-CCCAAGCTTCAAAATACATCAGTGGTC-3’; ASNS_−3: 5’-TGGTTGGTCCTCGCAGGCAT-3’, 5’-CGCTTATACCGACCTGGCTCCT-3’; ASNS_+1195: 5’-CCGCTCGAGACCTGTCTGTAGTTGGTTA-3’, 5’-CCCAAGCTTACTGTTCTTCCTACTCCAAC- 3’; ATF4_−13: 5’-TCCTCGGCCTTCACAATAA-3’, 5’-TACTTTACTAGGCTTCTATG-3’; CDKN1A_-RE: 5’-CCGCTCGAGGTGGCTCTGATTGGCTTTCTG −3’, 5’-CCCAAGCTTCTGAAAACAGGCAGCCCAAG −3’

Metabolite profiling and isotope tracing

Cells were washed twice with cold PBS, and pellets were resuspended in 80% methanol and then lysed by 5 freeze–thaw cycles (LN2 freezing followed by thawing at room temperature). Samples were then collected and centrifuged at 20,000 × g for 30 min at 4 °C, and the supernatant was dried using SpeedVac. 50% acetonitrile was added to the tube for reconstitution following by overtaxing for 30 sec. Samples solution was then centrifuged for 15 min 20,000 × g, 4 °C. Supernatant was collected for liquid‐chromatography mass spectrometry (LC‐MS) analysis. For isotope tracing experiments, cells were seeded in in 6-cm dishes and cultured in medium containing 10 mM [U-¹³C₆] glucose for 24 h. Metabolites were then extracted as described above. For both, samples were analyzed by High-Performance Liquid Chromatography and High-Resolution Mass Spectrometry and Tandem Mass Spectrometry (HPLC-MS/MS) as previously described (39). Specifically, the system consisted of a Thermo Q-Exactive in line with an electrospray source and an Ultimate3000 (Thermo) series HPLC consisting of a binary pump, degasser, and auto-sampler outfitted with a Xbridge Amide column (Waters; dimensions of 4.6 mm × 100 mm and a 3.5 μm particle size). The mobile phase A contained 95% (vol/vol) water, 5% (vol/vol) acetonitrile, 20 mM ammonium hydroxide, 20 mM ammonium acetate, pH = 9.0; B was 100% Acetonitrile. The gradient was as following: 0 min, 15% A; 2.5 min, 30% A; 7 min, 43% A; 16 min, 62% A; 16.1–18 min, 75% A; 18–25 min, 15% A with a flow rate of 400 μL/min. The capillary of the ESI source was set to 275 °C, with sheath gas at 45 arbitrary units, auxiliary gas at 5 arbitrary units and the spray voltage at 4.0 kV. In positive/negative polarity switching mode, an m/z scan range from 70 to 850 was chosen and MS1 data was collected at a resolution of 70,000. The automatic gain control (AGC) target was set at 1 × 106 and the maximum injection time was 200 ms. The top 5 precursor ions were subsequently fragmented, in a data-dependent manner, using the higher energy collisional dissociation (HCD) cell set to 30% normalized collision energy in MS2 at a resolution power of 17,500. Besides matching m/z, metabolites are identified by matching either retention time with analytical standards and/or MS2 fragmentation pattern. Data acquisition and analysis were carried out by Xcalibur 4.1 software and Tracefinder 4.1 software, respectively (both from Thermo Fisher Scientific).

In vivo therapeutic experiments

TM00298 PDX tumors were minced to 1.5 mm X 1.5 mm of small pieces and 2–3 tumor fragments were injected to flanks of NSG mice. When tumors volume reached about 100 mm³, mice were castrated and at 5 days post-castration, the castrated mice were administered with CB-839 (200 mg/kg in 25% (w/v) hydroxypropyl-β-cyclodextrin (HPBCD; MedChemExpress) in 10 mmol/L citrate, pH 2.) twice daily by oral gavage, ASNase (2000IU/kg) every 3 days by intraperitoneal (i.p.) injection, both CB-839 and ASNase, or vehicle controls for up to 4 weeks (n=7–11 animal/group). Tumor volume and body weight were measured using calipers twice a week following different treatment. Tumor growth was calculated with the formula (length × width × width)/2. When the mice reached the experimental endpoint, tumor tissues were harvested, weighed, and processed for further analyses. For tumor metabolite extraction, when the mice reached the endpoint, metabolites were extracted from about 100mg of frozen tissues by adding 80% methanol at −80°C followed by vortexing and centrifugation at 20,000 × g for 30 min. Dried metabolites were reconstituted in 50% acetonitrile, then centrifuged for 15 min 20,000 × g, 4 °C. Supernatant was run on a LC-MS system.

Statistical analysis

Statistical analyses were carried out using GraphPad Prism 9.1 software. Data are shown as mean ± SD. Statistical significance was evaluated using a two-tailed Student’s t-test or Fisher’s exact test. Survival plots were depicted as Kaplan-Meier curves. A P value of less than 0.05 was considered statistically significant.

Data availability

Oncomine datasets of PCa were downloaded for the comparison of ASNS expression changes between normal, cancer and metastatic tissues and in cancers of different grade. The gene expression data of hormone- naïve PCa or CRPC patients in this study was obtained from Gene Expression Omnibus (GEO) at GSE6811 (Tamura dataset, n=35) (40), Tomlins (n=18) (41), and GEO at GSE3325 (Varambally dataset, n=13) (42). The Beltran (n=49) (43) and SU2C/PCF 2019 (n=208) (44) datasets with gene expression data from adenocarcinoma or NEPC samples were also downloaded for validation from the cBioPortal platform (https://www.cbioportal.org/). Patient survival data were obtained from GEO at GSE21032 (Taylor dataset, n=92) (45) and the cBioPortal platform (SU2C/PCF 2019 dataset, n=65) (44). In addition, gene expression data from the PCa patients were also derived from GEO at GSE35988 (Grasso dataset, n=25) (46) and cBioPortal (Fred Hutchinson, 2016 (47) and DKFZ (48)). Pan-cancer ASNS RNA and protein expressions were obtained from The Cancer Genome Atlas (TCGA). The RNA-seq and ATAC-seq data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE215153 and GSE255000, respectively. Metabolomics datasets are available in the Supplementary Tables. All other data generated in this study are available from the corresponding author on request.

Results

Model of CRPC by deletion of Pten and Tp53 in Bmi1+ cells

To gain mechanistic insights into how TP53 inactivation contributes to CRPC development, we generated Pten;Tp53 double knockout (Bmi1-Cre^ER; Pten^f/f; Tp53^f/f, hereafter referred to as DKO) mice that allow conditional, tamoxifen-inducible deletion of Pten and Tp53 in Bmi1+ progenitor cells. To limit Cre-ER-mediated gene deletion to the prostatic epithelium, we used a tissue recombination strategy to ‘rescue’ transgenic Pten KO and DKO mouse prostates by regeneration as grafts in host mice (Fig. 1A). With this strategy, tamoxifen treatment of host mice bearing established prostate grafts from transgenic mice leads to deletion of Pten or Pten and Tp53 exclusively in the prostatic Bmi1+ cells (34,35). As expected, accelerated prostate tumor progression was observed in DKO grafts compared to Pten KO grafts, consistent with the aggressive behavior of other preclinical models of Pten/Tp53 deletion in the prostate. Unlike Pten KO grafts that consist exclusively of AR+ luminal adenocarcinoma at 5 months post-tamoxifen induction, DKO grafts displayed more heterogeneous and highly proliferating tumors with diverse phenotypes including sarcomatiod, squamous, and neuroendocrine (NE)-like phenotypes (Fig. 1B). After castration, DKO grafts developed CRPC with a higher incidence of non-adenocarcinoma histology (Fig. 1B–C) and tumor cell proliferation (Fig. 1D) when compared with either intact DKO tumors or Pten KO CRPC. DKO CRPC showed an increase in AR-null, p63+ and NE+ tumor cells (Fig. 1E–F), which resembled the histopathological features of CRPC patients with co-alteration of PTEN and TP53 that exhibit an increased frequency of tumors with NE features (Fig. 1G).

Fig. 1. Bmi1-CreER;Pten^f/f (Pten KO) and Bmi1-CreER;Pten^f/f;Tp53^f/f (Pten/Tp53 double KO, DKO) mice models of castration-resistant prostate cancer.

(A) Scheme for prostate cancer initiation, castration-induced regression, and recurrence in rescued Pten KO and DKO mice prostate grafts by tissue recombination. We recombined prostate epithelium from adult Pten KO or DKO mice with rat urogenital mesenchyme in collagen and grafted under the renal capsule of severe combined immunodeficient (SCID) mice. Grafts were then analyzed at various time points after tamoxifen induction and castration. (B) Disease distribution of Pten KO and DKO mice prostate grafts before castration (intact) and 5 months postcastration (recurrence or castration-resistant prostate cancer (CRPC)). Adeno, Adenocarcinoma; NEPC, Neuroendocrine prostate cancer. (C) Histological characterization in CRPC generated from Pten KO or DKO mice prostate. Representative hematoxylin and eosin (H&E) staining are shown. Scale bar, 50 μm. (D) Graph showing the percentage of proliferating tumor cells (BrdU+) in DKO prostate grafts before/during castration and Pten KO CRPC. (E) Immunostaining for the indicated antibodies in DKO and Pten KO CRPC. Scale bar, 50 μm. (F) Percentage of tumor cells that are positive for the indicated lineage markers in DKO prostate grafts before and during castration. # No AR+ cells were observed. (G) Frequency of cases with small cell features and NE expression in CRPC patients with any alterations in PTEN or/and TP53 from the SU2C/PCF 2019 prostate cancer dataset (cBioPortal). All data represent the mean ± SD, *p<0.05, **p<0.01, ***p<0.001, two-tailed Student’s t-test.

Integrated transcriptome and metabolome analysis of post-castration tumors identifies asparagine biosynthesis as a pathway upregulated upon Tp53 loss

To identify potential vulnerabilities imposed by Tp53 loss, we performed RNA sequencing (RNA-seq) from DKO and Pten KO tumors 5 months post-castration. Unsupervised principal component analysis (PCA) showed that the Pten KO and DKO CRPC clustered separately (Fig. 2A), consistent with their distinct phenotypes. We identified 4960 and 4715 significantly upregulated or downregulated genes in DKO CRPC relative to Pten KO CRPC (Padj ≤ 0.05; Fig. 2B). Gene set enrichment analysis (GSEA) of the significantly upregulated genes showed enrichment in pathways associated with the metabolic response to stress, such as mTORC1 signaling, unfolded protein response, amino acid metabolism and biosynthesis (Fig. 2C). These findings prompted us to examine metabolic alterations in this model. We used liquid chromatography–mass spectrometry (LC/MS) analysis to characterize intracellular metabolites in Pten KO and DKO prostate tissues 5 months after castration (Supplementary Table S1). Metabolomics identified 133 significant differentially altered metabolites in DKO compared to Pten KO CRPC. KEGG pathway analysis of significantly altered metabolites in DKO CRPC identified enrichment of tricarboxylic acid (TCA), nucleotide, and amino acid metabolic pathways (Supplementary Fig. S1A), consistent with the results obtained from the GSEA analysis of RNA-seq. Notably, analysis of metabolite abundances revealed that although DKO CRPC did not exhibit a significant increase in the levels of TCA intermediates, they displayed dramatic changes in various amino acids, including alanine, arginine, asparagine, glycine, and methionine (Supplementary Fig. S1B). Asparagine and arginine were among the top 25 most abundant metabolites in DKO CRPC (Fig. 2D). To further explore the role of p53 inhibition in metabolic alterations, we treated PCa cells with pifithrin-α (PFT-α), which suppresses p53-mediated transactivation. Treatment of LNCaP cells with PFT-α resulted in increased levels of several TCA-associated metabolites (citrate, succinate, and malate), asparagine, and glutamine, while alanine and arginine showed a significant decrease (Supplementary Fig. S1C and Supplementary Table S2). To directly investigate these findings, we knocked down TP53 in C4–2B and LNCaP cells using shRNA and cultured them in conditions of charcoal-stripped serum to mimic androgen deprivation. We observed higher levels of asparagine and serine in C4–2B cells following TP53 knockdown (KD) (Supplementary Fig. S1D–E and Supplementary Table S3). By conducting metabolic flux studies with a tracer of uniformly labeled [U-¹³C₆] glucose, we observed a significant increase only in the levels of aspartate/asparagine among the amino acids in shTP53 LNCaP cells, when compared to the control cells (Fig. 2E and Supplementary Table S4). However, we did not observe any significant increase for the other amino acids. Hence, we focused on asparagine and investigated how glucose contributes to the TCA cycle and asparagine production upon TP53 depletion. In the process of glycolysis, [U-¹³C₆] pyruvate is oxidized by pyruvate dehydrogenase (PDH) to acetyl-CoA, which enters the TCA cycle, resulting in the incorporation of two ¹³C₆ carbons (M2, black filled circle in Fig. 2E) into all TCA cycle intermediates and downstream amino acids. After the first turn, the second turn of the TCA cycle, with either unlabeled or M2 acetyl-CoA and M2 oxaloacetate (gray filled circle in Fig. 2E) from the first TCA cycle, leads to the formation of M1 or M3 isotopologues. Alternatively, pyruvate can be carboxylated by pyruvate carboxylase (PC), converting it to M3 oxaloacetate, which in turn produces M5 citrate by reacting with M2 acetyl-CoA. As shown in Fig. 2E, we observed a significant increase in M2/M3 isotopologues of TCA cycle intermediates and aspartate/asparagine, but not M5 citrate, upon TP53 knockdown in LNCaP cells. This suggests that in the condition of TP53 loss, ¹³C₆ glucose labels TCA metabolites and asparagine through increased flux, partially mediated by PDH activity.

Fig. 2. Integrated transcriptome and metabolome analysis of post-castration tumors reveals that asparagine biosynthesis is upregulated upon Tp53 loss.

(A) RNA sequencing (RNA-seq) was conducted on DKO and Pten KO tumors, which were collected 5 months after castration. Principal component analysis (PCA) indicated that DKO CRPC showed a different set of differentially expressed genes (DEG) compared to Pten KO CRPC. (B) Heatmap shows Z-scores of DEseq2 normalized read counts for differentially expressed genes identified by RNA-seq in DKO and Pten KO CRPC. (C) Top enriched pathways from the C2 Canonical Pathways (CP) gene set in DKO CRPC. (D) Liquid chromatography-mass spectrometry (LC/MS) analysis was performed on DKO and Pten KO tumors collected 5 months after castration. The heatmap displays the 25 most differentially abundant metabolites between DKO and Pten KO CRPC. Colored cells on the map correspond to concentration values which are measured by Euclidean distance with an average clustering algorithm (n = 3 per group). p<0.05 is used for each comparison. (E) LNCaP (WT TP53) cells transduced with shNT or shTP53-expressing lentiviral vectors were maintained for 2 weeks in medium with charcoal stripped FBS to mimic androgen deprivation. For metabolic flux analysis, the cells were incubated in DMEM containing 10 mM ¹³C₆-glucose for 24 hours (n = 3 per group) and the incorporation of ¹³C₆-glucose carbon into downstream metabolites were measured. The results demonstrate the levels of metabolites labeled with singly (M1) to quintuply (M5) ¹³C-labeled isotopologues in the cells. Schematic of ¹³C-labeling patterns after the metabolism of [U-¹³C₆] glucose via glycolysis and the TCA cycle is shown in center. Empty circles, ¹²C; black filled circle, ¹³C; gray filled circles, ¹³C derived from ¹³C-labeled oxaloacetate that enters the second turn of the TCA cycle. Ala, alanine; Asn, asparagine; Asp, aspartate; Gly, glycine; Ser, serine; Gln, glutamine; Cit, citrate; Mal, malate. All data represent the mean ± SD, **p<0.01, ***p<0.001, two-tailed Student’s t-test.

TP53 loss promotes de novo biosynthesis of asparagine by upregulating ASNS expression

Since the metabolic flux from TCA cycle-derived oxaloacetate to aspartate could support asparagine biosynthesis mediated by the enzyme asparagine synthetase (ASNS) (21), our results suggest that elevated ASNS expression may be responsible for the increased asparagine levels observed in TP53-deficient CRPC. Consistent with this, analysis of gene expression data from DKO and Pten KO CRPC revealed that ASNS is one of the top upregulated genes in DKO CRPC (Fig. 3A–B). In line with the RNA-seq results, IHC analysis showed that ASNS expression was strongly up-regulated in DKO CRPC compared to Pten KO CRPC (Fig. 3C). Similar results were obtained in CRPC tissues established from LNCaP-AR/TP53-null xenografts compared to LNCaP-AR/TP53-WT (Fig. 3D). Further analysis of prostate cancer gene expression datasets revealed that ASNS was upregulated in cancer relative to normal prostate (Supplementary Fig. S2A) and significantly associated with metastasis and a higher Gleason score (Supplementary Fig. S2B–C). Furthermore, ASNS expression was significantly higher in CRPC compared with hormone-naive PCa (Supplementary Fig. S2D). Additionally, in the Beltran et al. CRPC dataset (43), neuroendocrine-CRPC patients exhibited higher ASNS expression compared to adenocarcinoma-CRPC tumors (Supplementary Fig. S2E). In the SU2C/PCF metastatic prostate cancer cohort, ASNS expression was found to be significantly correlated with the NEPC score, although the correlation was weak (Supplementary Fig. S2F). However, no correlation was observed between ASNS expression and the AR score (Supplementary Fig. S2G). Kaplan–Meier survival analysis using multiple patient cohorts showed that patients with higher ASNS expression had shorter relapse-free or overall survival rates than patients with lower ASNS expression (Supplementary Fig. S2H).

Fig. 3. TP53 loss enhances ASNS expression, which promotes de novo biosynthesis of asparagine.

(A) Volcano plot showing differentially expressed genes in DKO and Pten KO CRPC. (B) Schematic diagram of de novo asparagine synthesis catalyzed by ASNS. (C) IHC analysis for ASNS in DKO and Pten KO tumors 5 months post-castration. Scale bars, 50 μm. (D) IHC images of ASNS in the castration-resistant LNCaP/AR-sgTP53 (CRISPR/Cas9-mediated knock down (KD) of TP53) and LNCaP/AR-sgNT (TP53 WT) xenografts. Scale bars, 50 μm. (E) ASNS mRNA levels in published datasets obtained from prostate cancers and all cancer types with wild type (WT) TP53 allele or biallelic copy loss of TP53 (DEL). Data represent the median with minimum and maximum values at the extremes of the whiskers. (F, G) Western blot of showing the protein levels of p53, ASNS, or p21 in shNT and shTP53 LNCaP or C4–2B cells transduced with the indicated lentiviral vectors (F) or treated with an inhibitor of p53, Pifithrin-α (PFT-α, 2.5 μM), for 24 hours (G), respectively. LNCaP and C4–2B cells were transduced with lentiviral vectors expressing either shNT or shTP53, and then maintained in medium supplemented with charcoal-stripped FBS for 2 weeks. Subsequently, the cells were transduced with lentiviral vectors expressing either shControl (shCtrl) or shASNS for 72 hours. (H) The abundance of intracellular asparagine (Asn) in shTP53 LNCaP and C4–2B cells expressing shCtrl or shASNS (n=3 per group). All data represent the mean ± SD, Two-tailed Student’s t-test. **p<0.01.

Of note, we observed higher ASNS expression across multiple cancer types including PCa with biallelic TP53 deletions compared with those with wild type, demonstrating that ASNS expression correlates with the loss of TP53. (Fig. 3E). In agreement with the marked increase in asparagine levels in CRPC harboring TP53 loss, TP53 knockdown by shRNA or TP53 inhibition with PFT-α in LNCaP and C4–2B cells increased ASNS expression (Fig. 3F–G). By contrast ASNS KD did not affect TP53 expression, indicating that ASNS could be a downstream effector of mutant TP53 activity. In addition, ASNS depletion in LNCaP- and C4–2B-shTP53 cells resulted in a reduction in intracellular asparagine levels (Fig. 3H), suggesting that TP53 deficiency promotes de novo biosynthesis of asparagine by upregulating ASNS expression.

TP53 loss sensitizes PCa cells to intracellular and extracellular asparagine depletion

We next determine whether the enhanced asparagine levels are functionally relevant for the growth of CRPC cells with TP53 loss. As tumors with high ASNS expression could either synthesize asparagine intracellularly or take up asparagine from the extracellular environment, depletion of extracellular asparagine only may not be effective for treating these tumors (49). Accordingly, when TP53-deficient or -inactivated LNCaP and C4–2B cells were not silenced for ASNS, extracellular asparagine depletion from the media did not show a significant growth inhibitory effect. However, intracellular depletion of asparagine via ASNS KD significantly impaired the growth of these cells cultured in asparagine-free medium (Fig. 4A). This was further validated in TP53-null PC3 cells (Fig. 4B). Moreover, asparagine addition significantly rescued cell growth in ASNS KD cells under the asparagine-depleted condition (Fig. 4B). Of note, the growth of TP53 WT PCa cells was not significantly affected by ASNS KD, but TP53 KD sensitized these cells to ASNS depletion (Fig. 4A). These results suggest that ASNS-mediated asparagine bioavailability contributes to the growth of TP53-deficient CRPC.

Fig 4. TP53 loss sensitizes PCa cells to intracellular and extracellular asparagine depletion.

(A) shNT or shTP53 LNCaP cells transduced with shCtrl or shASNS were cultured under Asn-depleted conditions. LNCaP and C4–2B cells-expressing shCtrl or shASNS were treated with 2.5 μM PFT-α under Asn-depleted conditions. (B) Western blot for the indicated proteins and cell growth in PC3 cells-expressing shCtrl or shASNS in the presence or absence of 0.1 mM Asn. (C) Schematic diagram shows that tumor growth can be impaired by dual targeting of de novo asparagine synthesis and exogenous asparagine availability. (D) shNT or shTP53 LNCaP cells transduced with shCtrl or shASNS were treated with vehicle or the indicated concentrations of L-asparaginase (ASNase). C4–2B cells-expressing shCtrl or shASNS were treated with 2.5 μM PFT-α alone or in combination with the indicated concentrations of ASNase. (E) PC3 cells-expressing shCtrl or shASNS were treated with vehicle or 0.5 U/ml ASNase. Cell growth (A-E) was evaluated using a colony formation assay (n=3 per group). After 2–3 weeks of treatment with the indicated shRNA or PFT-α, the colonies were stained with crystal violet and measured as a percentage of the colony number. All data represent the mean ± SD, Two-tailed Student’s t-test. *p<0.05, **p<0.01, ***p<0.001.

Based on the findings above, we hypothesized that dual targeting of de novo asparagine synthesis and exogenous asparagine availability may suppress the growth of these cells (Fig. 4C). To test this hypothesis, we explored the combined effect of ASNS silencing with L-asparaginase (ASNase)-mediated depletion of extracellular asparagine on TP53-altered CRPC growth. Although depletion of extracellular asparagine only by ASNase had negligible effects on the growth of CRPC with TP53 loss, intracellular asparagine depletion by ASNS KD sensitized these cells to ASNase treatment (Fig. 4D–E), suggesting that CRPC induced by TP53 loss could become susceptible to asparagine deprivation once its biosynthesis was inhibited.

TP53 loss activates ASNS via mTORC1-mediated induction of ATF4, contributing to asparagine production and promoting prostate cancer growth

The results described above suggest that TP53 loss induces ASNS expression, which promotes asparagine production. Thus, we next investigated the mechanisms by which TP53 loss regulates ASNS expression. First, we measured chromatin accessibility at the ASNS locus by ATAC-seq in isogenic LNCaP cells with or without TP53 KD by shRNA. We observed increased ATAC-seq signal at the promoter region of ASNS in LNCaP-shTP53 relative to its isogenic control, consistent with increased chromatin accessibility and ASNS expression in TP53-deficient LNCaP cells (Fig. 5A). We also observed the presence of canonical p53 and ATF4 binding motifs within this region (Fig. 5B). We then sought to determine if p53 binds to the ASNS promoter to regulate transcriptional activation of the ASNS gene. A recent chromatin immunoprecipitation sequencing (ChIP-seq) study of p53 binding sites (50) showed significant levels of p53 binding to the promoter region upstream and downstream of the ASNS transcription start site (TSS) in LNCaP cells, further increased in the presence of Nutlin-3, an MDM2 antagonist that stabilizes and activates p53 (Fig. 5C). Our ChIP-qPCR results demonstrated p53 binding primarily at genomic regions closer to the TSS (Fig. 5D). Notably, the site contained in the ATAC-seq peak (−3bp upstream of the TSS) showed stronger p53 binding than two previously described p53 binding sites (+1195 downstream and −657 upstream of ASNS TSS) (51) (Fig. 5D). As a control, we observed significant binding of p53 to the promoter of the canonical p53 target gene CDKN1A (p21) (Fig. 5D). Interestingly, ATF4 also showed significant binding affinities to regions at −3bp upstream and +1195 downstream of ASNS TSS, indicating that p53 and ATF4 bind to the ASNS promoter and have overlapping binding sites (Fig. 5E). These results, combined with the observation that TP53 KD upregulates ASNS expression and that ASNS is a key downstream target of ATF4, suggest that ATF4-mediated activation of the ASNS promoter may be repressed by p53, and that ATF4 binding may facilitate ASNS transcription upon p53 loss (Fig. 5F).

Fig 5. TP53 loss activates ASNS transcription via mTORC1-mediated induction of ATF4, resulting in increased asparagine production and promoting PCa growth.

(A) ATAC-seq results in LNCaP-shNT and -shTP53 cells (n=3 per group). TP53 KD increased chromatin accessibility at ASNS promoter regions. (B) ATF4- and TP53-binding motif were identified within increased ATAC-seq signals of ASNS promoter region. (C) ChIP-seq signals for p53 levels at the ASNS gene loci in LNCaP cells treated with 10 μM Nutlin or vehicle for 24 hours. The ChIP-seq dataset GEO: GSE157337 was used for the analysis. RE, the genomic regions containing the p53 response element; peak, the site contained in the ATAC-seq peak. (D) ChIP-qPCR analysis of p53 binding at the indicated regions of the ASNS gene loci and CDKN1A promoter in LNCaP cells (n=3 per group). (E) ChIP-qPCR analysis of ATF4 binding at the indicated regions of the ASNS gene loci in LNCaP cells following treatment with 100 nM thapsigargin for 4 hours (n=3 per group). (F) This schematic diagram illustrates that the activation of the ASNS promoter by ATF4 may be repressed by p53, and ATF4 binding may promote ASNS transcription when p53 is lost. (G) ChIP-qPCR analysis of p53 binding at the ATF4 promoter in LNCaP cells (n=2 per group). (H) Western blot for the indicated proteins in LNCaP cells transduced with shNT or shTP53-expressing lentiviral vectors or treated with 5 μM PFT-α or vehicle (DMSO) for 24 hours. (I) Western blot for the indicated proteins in shTP53 LNCaP cells treated with vehicle or 250 nM Torin1 (Tor) for 24 hours. (J) Western blot of LNCaP-shTP53 cells transfected with control or ATF4 siRNA for 48 hours. Relative cell proliferation rates were determined by culturing in medium with or without 0.1 mM Asn for 4 days (n=3 per group). (K) Schematic diagram to summarize the mechanism by which TP53 loss triggers ATF4-driven activation of ASNS. p53 directly binds to ATF4 and ASNS, leading to transcriptional repression (i). The loss of p53 results in the upregulation of the mTORC1 pathway, which further induces ATF4 expression and subsequently activates ASNS transcription (ii). All data represent the mean ± SD, Two-tailed Student’s t-test. *p<0.05, **p<0.01, ***p<0.001.

Given the observed increased expression of ATF4 and its canonical target genes in CRPC of DKO mice (Supplementary Fig. S3A–B), we hypothesize that TP53 loss may upregulate ATF4 to enhance ASNS expression. Consistent with this hypothesis, TP53 KD in LNCaP cells led to a notable increase in ATF4 mRNA (Supplementary Fig. S3C). In human tumor samples from the TCGA cohort, we also observed higher ATF4 mRNA expression in tumors with TP53 deletion compared to those with wild-type TP53 (Supplementary Fig. S3D), indicating a potential role for p53 in mediating the transcriptional activation of the ATF4 gene. Supporting this, p53 ChIP-seq and ChIP-PCR analyses in LNCaP cells demonstrated p53 binding to the region 13 base pairs upstream of the ATF4 transcription start site (Supplementary Fig. S3E and Fig. 5G). Additionally, TP53 KD in LNCaP cells resulted in a significant increase in ATF4 protein levels, along with elevated ASNS expression (Fig. 5H), further suggesting that ATF4 may be a direct target gene repressed by p53. To further investigate the regulation of ATF4 levels through the phosphorylation of the initiation factor eIF2α, which is triggered by stress-activated protein kinases in the integrated stress response (ISR) (52–54), we examined the phosphorylation of GCN2, PERK, and eIF2α in DKO CRPC, LNCaP, and C4–2B cells following TP53 KD or PFT-α treatment. However, no significant changes were observed (Supplementary Fig. S3F, Fig. 5H, and Supplementary Fig. S3F), suggesting that this canonical pathway does not play a significant role in ATF4 upregulation in this context. Recent studies have shown that mTORC1 can independently regulate ATF4 translation without relying on changes in eIF2α phosphorylation (55–57). We observed a significant increase in mTORC1 activity in DKO CRPC and LNCaP cells upon TP53 KD or PFT-α treatment, as assessed by the phosphorylation of mTORC1 target genes (S6K and 4E-BP1) (Supplementary Fig. S3A and Fig. 5H). Significantly, the use of Torin1 to inhibit mTORC1 signaling in shTP53 LNCaP cells led to a notable reduction in ATF4 and ASNS levels (Fig. 5I). However, it did not affect other ATF4 targets. This finding suggests that TP53 KD may enhance ATF4 protein expression through mTORC1 signaling to regulate ASNS transcription. We next investigated the contribution of ATF4 protein in the transcriptional activation of ASNS caused by TP53 loss. ATF4 KD significantly inhibited the induction of ASNS by TP53 silencing (Fig. 5J). Moreover, ATF4 depletion markedly impaired the growth of TP53-deficient cells in asparagine-free medium, and the exogenous addition of asparagine rescued the growth inhibitory effect of TP53/ATF4 dual knockdown (Fig. 5J). Taken together, these findings suggest that p53 directly targets the ATF4 and ASNS genes for transcriptional repression (Fig. 5Ki). The loss of p53 results in the activation of ASNS transcription through the induction of ATF4 expression by mTORC1 (Fig. 5Kii). The increased ASNS expression supports intracellular asparagine synthesis, fueling PCa growth.

Mutant TP53 is associated with high levels of ASNS and asparagine whose depletion impairs TP53-mutated CRPC growth

Alteration in the TP53 gene as missense mutations is one of the most common events responsible for treatment resistance in prostate cancer (5–8). Furthermore, although inactivation of TP53 frequently occurs by a missense mutation and loss of the second allele in various cancers, recent studies indicate that TP53 mutations can coexist with WT TP53 in prostate cancer (50). We therefore further surveyed ASNS expression by classifying CRPC according to TP53 mutation status in the background of WT TP53 or TP53 deletion. As compared with prostate cancers harboring both WT TP53 alleles, high level of ASNS was significantly associated with tumors with TP53 missense mutations (Supplementary Fig. S4A). Within the group of tumors with TP53 missense mutations, we observed no significant differences in ASNS levels based on the presence or absence of a WT TP53 allele (Supplementary Fig. S4B–C), suggesting that the ability of mutant TP53 to upregulate ASNS expression is independent of the remaining WT TP53 status. In addition to prostate cancers, ASNS expression was significantly upregulated at both mRNA (Supplementary Fig. S4A) and protein (Fig. 6A) levels across multiple tumor types carrying TP53 mutations (alone and with WT TP53). Most importantly, ASNS protein levels were markedly higher in human cancers with TP53 hotspot missense mutations in the DNA binding domain (Fig. 6B), which are the most frequently mutated in all types of cancer (Supplementary Fig. S4D).

Fig. 6. Mutant TP53 drives ASNS-mediated asparagine biosynthesis and asparagine depletion impairs TP53-mutated CRPC growth.

(A) ASNS protein levels across different cancer types including PCa based on the TP53 mutation status (TCGA). TP53 mutations involved both monoallelic (a single missense mutation and retaining a WT TP53 allele) and biallelic (a single missense mutation and loss of the second allele of TP53) events. PRAD, Prostate Adenocarcinoma; BRCA, Breast Invasive Carcinoma; LUAD, Lung Adenocarcinoma; UCEC, Uterine Corpus Endometrial Carcinoma; BLCA, Bladder Urothelial Carcinoma; OV, Ovarian Cancer. Data represent the median. (B) ASNS protein levels in human cancers that harbor five hotspot missense mutations in TP53. (C) Representative images of ASNS and p53 in the castration-resistant TM00298 (TP53 R273C) and LAPC9 (TP53 WT) PDX tumors. Scale bars, 50 μm. (D) Western blot of p53 and ASNS in the PDX-derived organoid, LAPC-4, and DU-145 cells transduced with control (Ctrl), shTP53, or shASNS-expressing lentiviral vectors. (E) ¹³C₆-enrichment of Asp/Asn in shNT and shTP53 LAPC-4 cells cultured with 10 mM ¹³C₆-glucose for 24 hours (n = 3 per group). (F) Cell growth of shCtrl or shASNS PDX-derived organoid, LAPC-4, or DU-145 cells in the presence or absence of 0.1 mM Asn for 5 days (n=e per group). (G) Cell growth of shCtrl or shASNS PDX-derived organoid treated with vehicle or 0.5 U/ml ASNase for 5 days (n=3 per group). All data represent the mean ± SD, Two-tailed Student’s t-test. **p<0.01, ***p<0.001.

We did not observe a significant correlation between PTEN deletions and ASNS mRNA or protein expression (Supplementary Fig. S4E). Instead, ASNS was markedly upregulated in PTEN-deficient samples with concurrent TP53 alterations (Supplementary Fig. S4F). Additionally, a TP53 mutant (R273C), castration-resistant PDX prostate tumor expressed higher levels of ASNS compared to LAPC9 (WT TP53) PDX prostate tumor (Fig. 6C). Notably, KD of the endogenous mutant TP53 in multiple PCa models carrying different TP53 mutations and AR levels, including AR-negative PDX organoid (R273C), AR-positive LAPC4 (R175H/P72R), and AR-negative DU145 (P223L/V274F), strongly reduced ASNS expression in all cases (Fig. 6D), indicating that mutant TP53 activates ASNS expression independent of AR status. Of note, and consistent with decreased ASNS expression, tracing of [U-¹³C₆] glucose showed that relative labeling of aspartate/asparagine was significantly decreased in LAPC-4 cells upon KD of mutant TP53 compared with control cells (Fig. 6E and Supplementary Table S5), demonstrating that TP53 mutation promotes ASNS-driven asparagine biosynthesis. Next, we determined whether the enhanced ASNS and asparagine levels support the growth of TP53-mutated CRPC cells. Similar to the results obtained in TP53-deficient PCa cells, inhibiting de novo asparagine synthesis via ASNS KD in asparagine-free medium significantly impaired the growth of PDX-derived organoids, LAPC-4, and DU-145 cells (Fig. 6F). ASNS depletion also rendered LAPC-4 cells highly sensitive to ASNase (Fig. 6G), indicating that asparagine restriction impairs TP53-mutated CRPC growth.

Mutant p53-induced activation of ATF4 contributes to the transcriptional upregulation of ASNS and CRPC growth

We next sought to determine how mutant p53 regulates ASNS expression. Analysis genome-wide mutant p53 ChIP-seq data (50) revealed p53-R273 binding to the ASNS promoter in LNCaP cells expressing p53-R273 (Fig. 7A). Interestingly, p53 ChIP-qPCR data from mutant p53-expressing LAPC-4 cells showed that while p53 binding sites in ASNS promoter were shared between WT and mutant p53, mutant p53 showed strong binding to all sites in LAPC-4 cells (Fig. 7B). Importantly, while KD of WT TP53 elevated ASNS expression (Fig. 3F), silencing of mutant TP53 led to the downregulation of ASNS expression (Fig. 7C), demonstrating that mutant p53 activates ASNS expression while WT p53 represses it. In addition, we observed strong binding of ATF4 to the ASNS promoter in LAPC-4 cells (Fig. 7D), indicating that mutant p53 may enhance the transcriptional activation of ASNS gene by cooperating with ATF4.

Fig. 7. Mutant p53-induced activation of ATF4 contributes to the transcriptional upregulation of ASNS and CRPC growth.

(A) ChIP-seq signals for p53 levels at the ASNS or ATF4 gene loci in LNCaP cells expressing p53 R273. The ChIP-seq dataset GEO: GSE157337 was used for the analysis. (B, D) ChIP-qPCR analysis of p53 (B) and ATF4 (D) binding at the indicated regions of the ASNS gene loci and CDKN1A promoter in LAPC-4 cells (n=2 per group). (C, F) Western blot for the indicated proteins in LAPC-4 cells transduced with shNT or shTP53-expressing lentiviral vectors. (E) ChIP-qPCR analysis of p53 binding at the ATF4 promoter in LAPC-4 cells (n=2 per group). (G) Western blot and cell growth of LAPC-4 cells transfected with control or ATF4 siRNA for 4 days (n=3 per group). (H) Schematic diagram to summarize the mechanism by which mutant p53 controls ATF4-mediated transcriptional activation of ASNS. All data represent the mean ± SD, Two-tailed Student’s t-test. **p<0.01, ***p<0.001.

TP53 mutation positively correlated with high level expression of ATF4 mRNA across all cancer types (Supplementary Fig. S5A). To dissect the molecular link between mutant p53 and ATF4, we next determined whether mutant p53 may transcriptionally activate the ATF4 gene by directly binding to the ATF4 promoter or through interaction with other TFs. ChIP-seq data from p53 R273-expressing LNCaP cells (50) showed the association of mutant p53 with the promoter of the ATF4 gene (Fig. 7A). In agreement with the ChIP-seq results, ChIP-qPCR revealed significant levels of mutant p53 at the ATF4 gene promoter in LAPC-4 cells (Fig. 7E). Notably, TP53 KD in LAPC-4 cells led to the down-regulation of ATF4 as well as ASNS at both mRNA (Supplementary Fig. S5B) and protein (Fig. 7F) levels. Moreover, there was a significant decrease in ASNS expression upon ATF4 KD (Fig. 7G), suggesting ATF4-dependent induction of ASNS in p53-mutated PCa cells. Importantly, asparagine supplementation largely rescued proliferation of ATF4-depleted LAPC-4 cells (Fig. 7G). Together, these results indicate that mutant p53 directly targets ATF4 gene for transcriptional activation, which subsequently contributes to transcriptional activation of ASNS gene by working cooperatively with mutated p53 (Fig. 7H).

The combination of CB-839 and ASNase depletes asparagine and impairs TP53-mutant CRPC growth in vitro and in vivo.

We next investigated the therapeutic potential of inhibiting intracellular and extracellular asparagine in TP53-mutant CRPC. As no direct ASNS inhibitors are currently available, we utilized a clinical-stage drug CB-839, glutaminase (GLS) inhibitor (NCT02071862) (58), that could phenocopy the antiproliferative effects of ASNS inhibition by reducing the production of glutamate and aspartate, consequently, blocking asparagine biosynthesis (Fig. 8A). Pharmacological inhibition of intracellular asparagine biosynthesis using CB-839 and depletion of extracellular asparagine with ASNase significantly reduced asparagine production in TP53-mutated PDX organoids (Fig. 8A). Moreover, the combined CB-839 and ASNase treatment of TP53-mutant or TP53-null models more effectively reduced cell proliferation than either inhibitor alone (Fig. 8B–C). Next, we examined the effects of combined CB-839 and ASNase treatment on apoptosis induction and the cell cycle state in PC3 cells. The combinational treatment led to a significant increase in the percentage of cells in the quiescent G0/G1 phase and a decrease in the percentage of cells in the proliferative S and G2/M phases (Fig. 8D). However, it did not induce apoptosis (Supplementary Fig. S6A), suggesting that the combination treatment may hinder TP53-altered PCa cell proliferation by inducing cell cycle arrest. Considering the crucial role of aspartate and glutamate as essential substrates for nucleotide synthesis (59–61), we proposed that a decrease in their availability could result in nucleotide deficiency and that introduction of exogenous asparagine could potentially restore impaired nucleotide synthesis caused by the treatment. Our findings support this hypothesis, as the combined treatment of CB-839 and ASNase resulted in reduced nucleotide levels (Fig. 8E and Supplementary Table S6–7). However, when exogenous asparagine was added, there was a partial restoration of nucleoside diphosphate and nucleoside triphosphate levels in both PDX organoids and PC3 cells. This suggests that asparagine plays a role, at least in part, in coordinating nucleotide synthesis. Notably, this restoration of nucleotide synthesis was accompanied by a rescue of impaired proliferation with exogenous asparagine (Fig. 8F).

Fig. 8. The combination of CB-839 and ASNase depletes asparagine, which impairs CRPC growth.

(A) Diagram shows the potential pathways which inhibit asparagine synthesis (center) and intracellular amount of Asn, Asp, and glutamate (Glu) in TM00298 PDX CRPC organoids treated with 0.5μM CB-839 (a selective glutaminase inhibitor), 0.5 U/ml ASNase, the combination of both (combo), or vehicle, for 5 days. (B) Cell growth in PDX CRPC organoids following treatment with increasing concentrations of ASNase, 0.5μM CB-839, the combination of both, or vehicle for 5 days. (C) Cell growth of PC3 cells following treatment with 0.5 U/ml ASNase, 0.5μM CB-839, the combination of both, or vehicle for 3 days. (D) Percentages of the cells in G0/G1, S, and G2/M phases following treatment with the combination of 0.5 U/ml ASNase and 0.5μM CB-839, or vehicle, for 24 h. (E) Cell growth in PC3 cells following treatment with the combination of 0.5 U/ml ASNase and 0.5μM CB-839 or vehicle in the presence or absence of 0.1 mM Asn for 3 days. (F) Relative nucleotide levels in PC3 and PDX organoids following treatment with the combination of 0.5 U/ml ASNase and 0.5μM CB-839 or vehicle in the presence or absence of 0.1 mM Asn for 3 and 5 days, respectively. (G) When tumors volume reached about 100 mm³, castrated NSG mice harboring PDX tumors were treated with CB-839 (200 mg/kg, oral gavage, twice per day; n = 8), ASNase (2000IU/kg, intraperitoneal injection, 3 days per week; n = 7), the combination (n = 7), or vehicle controls (n=7) for 4 weeks. Tumor volume was measured once a week following different treatment. (H) Relative intracellular levels of glutamate, aspartate, and asparagine at the endpoint in mice treated with CB-839, ASNase, the combination of both, or vehicle. n=7/group. All data represent the mean ± SD, Two-tailed Student’s t-test. *p<0.05, **p<0.01, ***p<0.001. (I) Proposed working model of TP53 mutation/loss and ATF4/ASNS cooperation in CRPC growth. Mutant TP53 or loss of wild-type TP53 can cooperate with ATF4, either directly or through the mTORC1 pathway, to enhance ASNS-mediated asparagine biosynthesis. This process creates metabolic vulnerability in these cells, which can contribute to tumor growth. Therapeutic intervention of asparagine levels by combinatorial treatment of GLS inhibitor and ASNase significantly decreases tumor growth. Mut, mutation; WT, wild type.

Based on the in vitro results, we hypothesized that targeting asparagine production with CB-839 and ASNase could enhance therapeutic efficacy. To test this, NSG mice harboring mutant TP53 PDX tumors were castrated once tumors reached ~100 mm³ in size, followed by treatment with vehicle, CB-839, ASNase, or a combination of CB-839 and ASNase. Notably, the combination treatment significantly impaired tumor growth and reduced tumor weight more efficiently than the single drug treatment alone (Fig. 8G). It is worth noting that ASNase, in addition to its primary asparaginase activity, also has glutaminase activity (62,63). However, this activity can cause toxic side effects that can limit the effectiveness of the therapy (62,63). To increase the therapeutic index of ASNase, inhibiting its glutaminase activity with CB-839 may be desirable. Indeed, in our preclinical studies, combining CB-839 with ASNase was found to be well-tolerated and had no effect on the weight of mice (Supplementary Fig. S6B). Furthermore, at the end point of the experiments, PDX tumors from mice treated with the combination of CB-839 and ASNase exhibited a marked decrease of intracellular asparagine (Fig. 8H). These results suggest that limiting asparagine availability by the combination of CB-839 and ASNase may offer therapeutic benefit to TP53-altered CRPC patients (Fig. 8I)

Discussion

Despite the fact that TP53 alterations play a major role in advanced prostate cancer and therapeutic resistance, they have been difficult to target effectively. Our study supports the notion that targeting a metabolic vulnerability present in TP53 altered tumors could be a promising therapeutic strategy. It is well known that oncogenic signaling controlled by TP53 alterations can rewire cancer cell metabolism in part through the regulation of several biosynthetic pathways, thereby promoting the synthesis of macromolecular precursors that support cell proliferation and cancer progression (14–16). Here we present insights into the mechanisms underlying how TP53 inactivation promotes adaptation of PCa cells to the androgen-deprived environment and identify metabolic vulnerabilities that result from TP53 alteration to target these cancers therapeutically. We demonstrate that p53 loss/mutation enhances the reliance of PCa cells on ASNS-mediated asparagine biosynthesis in mTORC1/ATF4-dependent mechanism, which confers metabolic plasticity to tumor cells, thereby supporting their growth in the androgen-depleted environment (Fig. 8G). Consistent with our findings, a recent study has also reported that TP53 represses ASNS expression to block de novo biosynthesis of asparagine (51).

Although the initial attention to the role of ASNS was drawn by the investigation of low enzyme activity for asparagine in ALL cells, which are associated with the sensitivity to ASNase, growing evidence suggests that ASNS can play a critical role as a key driver for solid tumor progression, including prostate cancer (29–33). Our study provides mechanistic insights connecting ASNS expression to TP53 alterations in solid tumors. An interesting aspect of our findings is the opposing roles of WT vs mutant p53 on ASNS expression. Both WT and mutant p53 bind directly to the ASNS promoter but WT p53 represses, while mutant p53 activates the ASNS gene. This is likely to be due to the recruitment of transcriptional corepressors vs coactivators by WT and mutant p53 respectively. Importantly, the functional outcomes of TP53 loss and TP53 mutation on ASNS expression and asparagine biosynthesis are similar: both alterations lead to upregulation of ASNS and increased intracellular asparagine production.

An additional layer of how TP53 loss/mutation regulates ASNS expression is via ATF4. ATF4 is a known transcriptional activator of ASNS expression. Our results indicate that both WT p53 and mutant p53 bind to the ATF4 promoter to repress or activate ATF4 expression respectively. In conjunction with previous works that ASNS is a key downstream target of ATF4 function in transcriptional activation of asparagine and nucleotide biosynthesis (52–54), these findings provide evidence in support of the model that loss/mutation of p53 results in an increase in ATF4 mRNA and protein expression levels, which then binds and transcriptionally activates ASNS gene. The fact that WT and mutant p53 utilize very different interactions with the transcriptional machinery when regulating the same gene (64–66) raise interesting questions of how mutant p53 interacts with ATF4 to activate the transcription of ASNS gene.

Given our findings that p53 and ATF4 binding overlap at the promoter of ASNS, it is possible that while WT p53 may compete with ATF4 to control the transcription of ASNS, mutant p53 may cooperate with ATF4 in regulating ASNS expression. Since all p53 mutations are not functionally equivalent (67,68), how distinct TP53 mutants contribute to ASNS/ATF4 activation remains an open question. DNA contact (R248Q and R273H) and conformational (R175H and R282W) mutants of p53 confer different functional properties and transcriptional programs that promote cancer progression (69). Importantly, numerous studies have demonstrated that transcriptional activity of contact mutant p53 proteins is modulated through interactions with other transcription factors or chromatin-modifying proteins that alter gene expression (69,70). Recent studies have demonstrated that ATF4 is a key downstream target of NRF2 function in transcriptional activation of asparagine and nucleotide biosynthesis (71,72) and mutant p53 interacts with NRF2 to modulate NRF2 transcriptional activity (73), indicating the possibility that mutant TP53 may cooperate with other transcription factors in regulating ATF4-mediated ASNS expression.

ASNS expression is regulated by an adaptive cellular stress pathway triggered in response to amino acid starvation and by increased endoplasmic reticulum (ER) stress and Unfolded Protein Response (UPR) (21,52–54,74). These pathways lead to activation of general control nonderepressible 2 (GCN2) kinase or PKR-like ER kinase (PERK), which promote the phosphorylation of the initiation factor eIF2α and an increase in transcription factor ATF4 protein levels (52–54). Alternatively, mTORC1 pathways have been reported to be involved in ATF4 upregulation and amino acid homeostasis (55–57). Activated mTORC1 phosphorylates 4EBP1 and S6 kinase (S6K) to promote protein translation. Notably, we also observed a significant upregulation of ASNS expression in TP53-deficient CRPC through the mTORC1-mediated activation of ATF4. We do not rule out an effect of ISR on ATF4 translation upon p53 loss/mutation, although we failed to observe activation of the canonical p-eIF2α pathway. More work is needed to define the role of translational regulation in ATF4 activation by TP53 loss/mutation.

Targeting asparagine metabolism by ASNase has been used clinically to treat leukemic cells, which are unable to synthesize sufficient asparagine due to low ASNS level (23–25). However, extracellular depletion of asparagine only using ASNase has not been proved to be effective in tumors with high ASNS expression (26–28) and development of clinically useful ASNS inhibitors has been limited, thus encouraging alternative strategies. Since glutamate generated from glutamine by glutaminase (GLS) contributes to the de novo asparagine synthesis (75), therapeutic intervention of asparagine levels by combinatorial treatment of GLS inhibitor and ASNase could be a possible approach to treat tumors that exhibit asparagine dependency. Accordingly, demonstrate that a selective GLS inhibitor CB-839 (58) in combination with ASNase treatment effectively restricts asparagine synthesis and hinder the growth of TP53-mutated CRPC PDX tumors. Therefore, GLS inhibitor treatment may serve as an effective strategy to block ASNS activity and restore cancer cell sensitivity to ASNase. CB-839 is a well-tolerated drug that has passed phase 1 clinical trials for multiple tumors such as hematological malignancies, lung cancer, breast cancer, and renal cancer (www.clinicaltrials.gov). Moreover, inhibiting ASNase’s glutaminase activity with CB-839 could improve its therapeutic index and reduce its potential toxic side effects. Therefore, combining CB-839 and ASNase could be a rapid and effective strategy for the clinical benefit of CRPC patients. However, ASNase administration is associated with a unique and significant toxicity profile compared to other chemotherapy types (76). Hypersensitivity, pancreatitis, thrombosis, encephalopathy, and liver dysfunction are some of the toxicities associated with its use. Currently, three preparations are used for treatment, two of which are native and purified from bacterial sources, while the third is modified from a native preparation (77). Polyethylene glycolated (PEG)-ASNase, which is better tolerated, has largely replaced native E. coli ASNase (78). Advances in ASNase therapy have been made through recombinant technology, extending its half-life and reducing immunogenicity. The future of ASNase therapy lies in personalized dosing, incorporating pharmacogenomics to predict toxicity risk, and developing additional strategies to mitigate target toxicities.

Significance:

TP53 mutated castration-resistant prostate cancer is dependent on asparagine biosynthesis due to upregulation of ASNS and can be therapeutically targeted by approaches that deplete intracellular and extracellular asparagine.