Skip to main content
NCBI home page
BMC Cancer logoLink to BMC Cancer
. 2026 Jan 21;26:97. doi: 10.1186/s12885-025-15357-5

Integrative transcriptomic profiling reveals subtype-specific therapeutic vulnerabilities and resistance mechanisms in prostate cancer

Wei Liu 1,4,#, Weiyu Kong 2,#, Silin Jiang 1,4,#, Yong Wei 1,4,#, Zijie Yu 2, Xiaoyang Cao 2, Jian Yang 1,4, Xiyi Wei 2,3, Luming Shen 1,4,✉, Chao Qin 2,✉, Qingyi Zhu 1,4,✉
PMCID: PMC12822270  PMID: 41566241

Abstract

Objective

Advanced prostate cancer (PCa) remains therapeutically challenging due to heterogeneous mechanisms of resistance to androgen receptor (AR)-targeting agents. While AR signaling persists in castration-resistant PCa (CRPC), emerging evidence suggests AR-independent survival pathways may contribute to therapeutic escape. This study integrates transcriptomic data and clinical profiling to dissect AR dependency and resistance mechanisms in PCa, aiming to identify subtype-specific vulnerabilities and therapeutic targets.

Methods

We performed CRISPR-Cas9 screens in AR-dependent (VCaP, LNCaP, 22Rv1) and AR-independent (DU145, PC-3, WPE1-NA22, P4E6, Shmac5) cell lines to identify core essential genes. RNA sequencing data from TCGA-PRAD, Changhai, and DKFZ cohorts were integrated to define molecular subtypes using consensus clustering. Spatial transcriptomics (ST) and single-cell RNA sequencing (scRNA-seq) were employed to validate gene expression patterns in primary tumors and metastatic samples. Temporal expression dynamics were analyzed using fuzzy clustering to identify resistance mediators, with a focus on MCL1. Drug sensitivity analysis revealed that AR-dependent cells were more sensitive to MCL1 inhibitor UMI-77, and MCL1 expression was higher in Enzalutamide-resistant cell lines. Functional validation via MCL1 knockdown confirmed its role in supporting the proliferation and inhibiting apoptosis of resistant cells.

Results

CRISPR screening identified 952 shared essential genes in prostate cancer, with 157 AR-high essential signature and 130 AR-low essential signature genes. AR-high essential signature genes enriched in cell cycle/polycomb pathways, while AR-low essential signature genes correlated with oxidative phosphorylation/mTOR signaling. Consensus clustering of TCGA-PRAD data revealed three molecular subtypes (Clusters 1–3); Cluster 3 showed worst prognosis (shorter PFI/OS) and advanced clinical features (higher T/N stage, Gleason grade). External validation confirmed Cluster 3’s aggressive phenotype and independent prognostic value (meta-cohort HR = 1.98, 95% CI: 1.19–3.27). Cluster 3 signature genes were upregulated in metastatic/CRPC tissues and spatially enriched in CRPC epithelium. Notably, Cluster 3 shared essential gene expression decreased after Enzalutamide treatment, whereas AR-high essential signature genes remained stable. MCL1 emerged as a key resistance driver, demonstrating persistent upregulation in Enzalutamide-resistant cells and CRPC models.

Conclusions

This study elucidates distinct AR dependency landscapes in PCa, revealing AR-independent survival pathways and a clinically actionable molecular subtype (Cluster 3) linked to therapy resistance. MCL1 emerges as a critical mediator of adaptive resistance, highlighting the need for combination therapies targeting both AR-driven and AR-independent programs to improve outcomes in advanced PCa.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-15357-5.

Keywords: Prostate cancer, Enzalutamide resistance, MCL1, CRISPR-Cas9 screen, Multi-omics

Introduction

Thetherapeutic landscape of advanced prostate cancer remains dominated by androgen receptor (AR)-targeting agents. However, despite initial responses, durable therapeutic effects remain elusive due to the complex and heterogeneous mechanisms of resistance [1]. AR signaling continues to play a central role in castration-resistant prostate cancer (CRPC), but emerging evidence points to subpopulations of tumor cells that may survive through AR-independent pathways [2–5]. This duality presents a critical challenge, highlighting the need for precise identification of AR-dependent versus AR-bypass survival mechanisms, a distinction that has profound implications for therapeutic stratification.

Recent CRISPR-based functional screens have provided valuable insights into context-specific vulnerabilities in prostate cancer [6]. However, several gaps remain in our understanding. First, systematic comparisons between shared dependencies and those exclusive to AR activity states have yet to be explored—an important distinction for identifying pan-effective therapeutic targets versus context-specific vulnerabilities. Second, although AR pathway inhibition (ARPI) resistance mechanisms frequently involve bypassing AR signaling, the persistence of AR-dependent transcriptional programs during treatment failure remains poorly understood. This lack of understanding hinders the development of strategies to overcome the so-called “AR-addicted” transcriptional inertia in progressing tumors.

In this study, we integrate transcriptomic data with clinical relevance analysis to systematically dissect AR dependency and resistance mechanisms in prostate cancer. Through AR activity-stratified CRISPR screening, we aim to: (1) characterize the distinct vulnerability landscapes of AR-high and AR-low prostate cancer cells, revealing core essential pathways that define their survival dependencies; (2) investigate the persistence of AR-dependent transcriptional programs under ARPI pressure, addressing the limitations of AR-axis mono-therapy; and (3) identify clinically relevant molecular subtypes linked to therapy resistance and disease progression. By incorporating time-resolved analyses, we further explore the adaptive resistance mechanisms that sustain tumor survival, with a particular focus on MCL1 as a potential mediator of AR-independent persistence and treatment escape. These insights provide a framework for refining therapeutic strategies in advanced prostate cancer, emphasizing the need for precision-based interventions targeting both AR-driven and AR-independent survival pathways.

Method

Data acquisition

To filter essential genes for the survival of prostate cancer, CRISPR-Cas9 essentiality screening data were obtained from the Depmap database [7], including AR-dependent (VCaP, LNCaP clone FGC and 22Rv1) and -independent (DU 145, PC-3, WPE1-NA22, P4E6, Shmac 4 and Shmac 5) cell lines. The mean gene effect score was calculated, representing the average effect of a gene in regulating prostate cancer cells respectively in AR-dependent and -independent cell lines, and a negative value for a gene effect score indicated that silencing it suppressed cell survival. A filtering threshold was set to−0.796, which equaled to the mean gene effect score of AR in AR-dependent cell lines. Consequently, a total of 1082 high AR dependency genes and 1109 low AR dependency genes were generated (Supplementary Table 1).

RNA sequencing data of three independent datasets included the TCGA-PRAD cohort (Tumor: 500; Normal: 52), the Changhai cohort [8] (Tumor: 120) and the DKFZ cohort [9] (Tumor: 227). The TPM format expression data was utilized. To compare data among cohorts, the batch effect was removed with ComBat in R package sva, and the normalization result was validated by principal component analyses (PCA). Other validation cohorts of patient-derived samples included GSE3325 [10], GSE32269 [11] and the Prostate Cancer Transcriptome Atlas (PCTA) [12], cell line originated data included GSE202885 [13], GSE229805 [14], GSE220097 [15] and GSE244024 [16], and xenograft data from GSE95413 [17].

Two single cell RNA sequencing data were included. The GSE137829 [18] dataset included 6 CRPC patient derived tumor samples. The GSE168733 [19] contained 2 Enzalutamide treated LNCaP and VCaP, and 1 VCaP DMSO control, which was used for validation of gene expression.

The spatial transcriptomics (ST) data were obtained from GSE278936 [20], including 4 benign, 17 Treatment naïve, 22 neoadjuvant-treated, 5 CRPC and 4 metastatic samples.

Enrichment analyses

Gene Ontology (GO) analyses were used to annotate the filtered CRISPR genes with the Kyoto Encyclopedia of Genes and Genomes (KEGG) gene sets.

The Gene Set Enrichment Analysis (GSEA) was conducted with R package clusterProfiler using differentially expressing data obtained with limma with the HALLMARK gene sets. The criteria for identifying significantly enriched pathways were set as q-value < 0.05.

The evaluate the expression level of a specific gene set, we utilized gsva in R package GSVA, with method set to ‘zscore’ for RNA sequencing data, and the AddModuleScore function in Seurat for single cell RNA data and ST data.

Differentially expressed genes (DEG) detection

The DEGs were obtained with R package limma. For discovery, the significant criteria were set as adjusted p-value < 0.05 and absolute fold change > 1. For validation, the significant criteria were set as adjusted p-value < 0.05.

Consensus clustering

Consensus clustering was performed using the ConsensusClusterPlus package to identify robust molecular subtypes based on gene expression profiles. Initially, gene expression data were log-transformed log2​(TPM + 1) and missing values were omitted to ensure data integrity.

The consensus clustering analysis was conducted on the processed expression matrix using k-means clustering with Euclidean distance as the similarity metric. The number of clusters (K) was set to range from 2 to 9, with 100 iterations for each K to enhance stability. The clustering was performed with 80% item resampling and 100% feature retention.

To determine the optimal number of clusters, the Proportion of Ambiguous Clustering (PAC) was calculated. The PAC score, defined as the proportion of samples with consensus values between 0.1 and 0.9, was computed for each K. The optimal cluster number was selected based on the minimum PAC score, indicating the most stable partitioning of the dataset.

To validate the consensus clustering results, we performed label transfer using two independent external prostate cancer datasets: Changhai and DKFZ cohorts. Prior to label transfer, batch effects were corrected using the ComBat function from the sva package after log-transformation of the gene expression data. The consensus clustering labels derived from the TCGA cohort were transferred to the external datasets by computing the Euclidean distance between each validation sample and the centroid of each cluster in the training cohort. Each validation sample was assigned to the closest cluster based on minimum Euclidean distance. Finally, PCA was conducted using the selected model genes, and the distribution of different clusters across datasets was visualized to assess the consistency of the clustering pattern.

Survival analysis

To assess the prognostic significance of the identified clusters, Kaplan-Meier survival analysis was conducted using progression-free interval (PFI) and overall survival (OS) as the endpoint. Survival curves were generated using the Kaplan-Meier method, and differences between clusters/groups were evaluated with the log-rank test. The survival curves were visualized using the ggsurvplot function from the survminer package. The median survival time was indicated with horizontal and vertical reference lines, and the p-value was computed based on the chi-square statistic from the log-rank test.

Single-Cell RNA sequencing data processing and integration

Normalization was performed using the NormalizeData function, and highly variable features were identified using FindVariableFeatures with the “vst” method, selecting the top 2,000 variable genes. Mitochondrial genes, ribosomal genes, and heat shock proteins (HSPs) curated from GeneCards were excluded. The filtered feature set was saved for downstream analysis.

Single cell RNA sequencing seurat objects were merged using the merge function, followed by data scaling using ScaleData. Principal component analysis (PCA) was performed using RunPCA on the selected features. To further correct batch effects, RunHarmony was applied with the first 30 principal components (PCs) utilized for integration. In addition, the signature score was calculated using the AddModuleScore function in Seurat.

Spatial transcriptomics data processing

For ST data, genes with total counts exceeding five across all spots were retained, while mitochondrial genes were excluded. Spots with fewer than 300 detected features or mitochondrial content exceeding 30% were filtered out. The data were normalized using SCTransform with the Spatial assay.

Temporal expression pattern analysis using fuzzy clustering

Gene expression data were obtained from the GSE202885 dataset. We focused on AR-high essential signature genes for temporal analysis. For each gene, the mean expression level was calculated across five treatment time points: DMSO (control), 24 h, 48 h, 96 h, and 144 h. Genes with more than 25% missing values were excluded, and any remaining missing values were imputed using the mean expression across samples. Expression values were then standardized to allow comparison across genes.

Temporal expression patterns were identified using fuzzy c-means clustering implemented in the Mfuzz package. The optimal fuzzification parameter (m) was estimated using the mestimate function. A clustering solution with three clusters was selected based on cluster stability and interpretability. Membership values were calculated for each gene, representing the degree of association with its assigned cluster, enabling classification of genes according to their dynamic responses to Enzalutamide treatment.

Drug sensitivity assessment

The drug sensitivity data of half maximal inhibitory concentration (IC50) were obtained from the Genomics of Drug Sensitivity in Cancer (GDSC) database, including four MCL1 inhibitors AZD5991, UMI-77, MIM1 and Sabutoclax.

Cell culture

Human CRPC lines C4-2 and 22RV1 were obtained from American Type Culture Collection. Low-glucose RPMI 1640 medium was purchased from Thermo Fisher Scientific (Waltham, MA, USA). We established an Enzalutamide-resistant PCa cell line (C4-2 Res and 22RV1 Res) by sequentially treating cells with 10, 20 µM Enzalutamide (Beyotime Biotechnology, Shanghai, China) for 2 months, and 30, 40 µM Enzalutamide for 1 month. Cells were incubated in RPMI 1640 (GIBCO, NY, USA) supplemented with 10% fetal bovine serum and 100 µg/mL penicillin at 37 °C and 5% CO2.

siRNA-mediated knock down of MCL1

siRNA sequence targeting MCL1 (siMCL: 5′-GCTGGAGATTATCTCTCGGTA-3′) a non-targeting control siRNA (siCtrl, scrambled sequence) were synthesized by RiboBio (Guangzhou, China). MCL1 mRNA levels were quantified by qRT-PCR using SYBR Green Master Mix (Takara) with primers:

Forward: 5′- TGCTTCGGAAACTGGACATCA − 3′

Reverse: 5′- TAGCCACAAAGGCACCAAAAG − 3′

Data normalized to GAPDH (ΔΔCt method). All experiments were performed with three independent biological replicates.

Cell apoptosis assay

Apoptotic cells were detected using an Annexin V Apoptosis Detection Kit PE (KeyGEN BioTECH). Briefly, 5 × 105 cells were collected and washed twice by cold PBS. Cells were centrifuged at 300 × g for 5 min. Then, cells were resuspended in 500 µL of 1× Binding Buffer and incubated with 1µL of Annexin V-PE at RT protected from light for 5–10 min. Fluorescence-labeled cells were detected with a flow cytometer. All experiments were performed with three independent biological replicates.

Cell viability assay using CCK-8

Cell viability was assessed using the Cell Counting Kit-8 (CCK-8, Dojindo, Japan) according to the manufacturer’s instructions. Briefly, cells were seeded into 96-well plates at a density of 1 × 10³ to 2 × 10³ cells per well and allowed to adhere overnight under standard culture conditions (37 °C, 5% CO₂). The following day, cells were treated with the indicated compounds or control medium for the specified time periods (e.g., 0, 24, 48, 72, 96 h). After treatment, 10 µL of CCK-8 solution was added to each well, and the plate was incubated at 37 °C for 2 h, depending on cell type and metabolic activity. The absorbance at 450 nm was measured using a microplate reader (BioTek, USA). All experiments were performed in triplicate and repeated at least three times. All experiments were performed with three independent biological replicates. Data were expressed as mean ± standard deviation (SD), and statistical significance was determined using Student’s t-test or one-way ANOVA where appropriate.

Statistical analysis

All analyses were performed using R version 4.3.1. All statistical tests were two-sided, and a p-value < 0.05 was considered statistically significant unless otherwise specified. Continuous variables following a normal distribution were compared using the independent samples t-test, while variables with a skewed distribution were analysed using the Wilcoxon test. *p < 0.05; **p < 0.01; ***p < 0.001.

Results

Identification of AR dependency signatures in prostate cancer cell lines

CRISPR screening stratified prostate cancer cell lines into AR-dependent (VCaP, LNCaP, 22Rv1) and AR-low-dependent subtypes (Fig. 1A). We identified 952 shared essential genes common to all lines, along with 157 genes comprising the AR-high essential signature and 130 genes comprising the AR-low essential signature (Fig. 1B and Supplementary Table 1). Functional enrichment revealed AR-high genes associated with cell cycle regulation and polycomb repressive complex activity, while AR-low genes were enriched in oxidative phosphorylation and mTOR signaling pathways (Fig. 1C).

Fig. 1.

Fig. 1

Open in a new tab

Defining prostate cancer dependent genes in AR high and low dependent cell lines from CRISPR results. A CRISPR results defined VCaP, LNCaP and 22Rv1 as AR dependent cell lines. B Filtering of prostate cancer dependent genes in AR high and low dependent cell lines. C GO analyses of KEGG terms for the shared essential genes, AR-high essential signature, and AR-low essential signature genes to explore their distinct functional roles

Consensus clustering reveals prognostic subtypes linked to clinical outcomes

Most of the shared essential genes were up-regulated in the tumor tissue (Fig. 2A). Unsupervised clustering of these shared essential genes in TCGA-PRAD cohort identified three robust subtypes (cluster 1–3) with minimal ambiguity (Fig. 2B-D and Supplementary Table 2). We found that most of the shared essential genes highly expressed in Cluster 3 were associated with poor prognosis, with hazard ratios (HR) greater than 1 (Fig. 2E). And cluster 3 exhibited the worst prognosis, with reduced overall survival and significantly shorter progression-free interval (PFI) compared to clusters 1–2 (Fig. 2F). Clinically, cluster 3 correlated with advanced T stage (p < 0.001), N stage (p < 0.001), and higher primary Gleason grade (p < 0.001) (Fig. 2G). The GSEA results of genes comparing between cluster 3 and other two clusters, as well as between cluster 2 and cluster 1 indicated that the common elevation of proliferating related hallmark pathways (Fig. 2H, J). 

Fig. 2.

Fig. 2

Open in a new tab

Consensus clustering of the common essential genes in prostate cancer. A Comparison of the expression level of common essential genes in the TCGA cohort. B The proportion of ambiguous clustering index with different cluster numbers. C The clustering results with 3 clusters. D The clinical information and gene expression patterns of the clustering results. E The prognostic results of unique clustering genes in the cluster 3. F, G KM plot showed the differences in OS and PFI among three clusters. H The proportion of different patient clinical indexes among three clusters. I, J The GSEA results of cluster 3 and cluster 2 respectively

External validation confirms subtype stability and prognostic utility

Harmonization of Changhai and DKFZ cohorts via ComBat removed batch effects (Fig. 3A and B), enabling reproducible subtyping. The PCA results indicated that clustering based on these genes could distinguish patients from different cohorts (Fig. 3C). And cluster 3 maintained its aggressive phenotype across cohorts, showing shortest PFI in both validation sets (Fig. 3D and E). And cluster 3 had the highest primary Gleason grade also in the two validation cohorts (Fig. 3F and G). Multivariable analysis confirmed Cluster 3 as an independent predictor of progression (meta-cohort HR = 1.98, 95%CI: 1.19–3.27; Fig. 3H. 3I).

Fig. 3.

Fig. 3

Open in a new tab

External validation of the clustering results in TCGA cohort. A Normalization and label determination workflow of the validation. B PCA results showed the batch effect of different cohorts were removed after integration. C The PCA results showed that each cluster were distinguished. D, E KM plots showed that cluster 3 had consistently shortest PFI among two validation cohorts. F, G The proportion of T stages and primary Gleason score in the Changhai and DKFZ cohorts. H, I Univariable and multivariable analyses of the clinical indexes and CRISPR cluster in the meta-PRAD cohort

Cluster 3 signature marks advanced disease and therapy resistance

Notably, most cluster 3-specific shared essential genes were upregulated in metastatic or CRPC samples (adjusted p-value < 0.05; Fig. 4A). And the Z-score of these genes were elevated in advanced prostate cancer, peaked in mCRPC. Single cell RNA sequencing data showed these genes were especially expressed in prostate cancer epithelial cells (Fig. 4C). And spatial transcriptomics data showed that most CRPC and metastatic samples had high expression of cluster 3-specific shared essential genes (Fig. 4D and G), and in primary tumor, the expression varied among samples (Fig. 4E), and was low in normal tissue (Fig. 4F).

Fig. 4.

Fig. 4

Open in a new tab

Custer 3-specific shared essential genes were up-regulated in advanced prostate cancer. A Most cluster 3-specific shared essential genes were up regulated in advanced prostate cancers, including metastatic and castration-resistant prostate cancer. B The Z-score of cluster 3-specific shared essential genes were elevated in prostate cancer and especially in advanced prostate cancer. C The cluster specific genes were especially expressed in epithelial cells in prostate cancer. D-G Spatial transcriptomics results of the expression level of cluster 3-specific shared essential genes

AR inhibitors, such as Enzalutamide, were essential for the treatment of CRPC. In this sense, we compared the Z-score of cluster 3-specific shared essential genes and AR-high essential signature genes after Enzalutamide treatment. We discovered that the Z-score of cluster 3-specific shared essential genes were significantly decreased after Enzalutamide treatment, while AR-high essential signature genes stayed table (Fig. 5A). The results were further validated in the single cell RNA sequencing data and VCaP xenografts results (Fig. 5B and E). Moreover, the Z-score of AR-high essential signature genes was correlated with shorter PFI in the three prostate cancer cohorts (Fig. 5F and H). Together, these findings suggest that the high dependency of AR-dependent cells on these AR-high essential signature genes may contribute to Enzalutamide resistance, highlighting their potential role in therapy evasion.

Fig. 5.

Fig. 5

Open in a new tab

Genes unique for AR dependent cell lines had consistent expression after Enzalutamide treatment. A The Z-score of cluster 3-specific shared essential genes and AR-high essential signature genes after Enzalutamide treatment of different periods. Statistical significance was assessed using one-way ANOVA. B Integration of the single cell RNA data of cell lines treated with Enzalutamide. C Volcano plot showed the differentially expressed genes after Enzalutamide treatment. D The AR-high essential signature genes stayed high after Enzalutamide treatment. E The Z-score of cluster 3-specific shared essential genes and AR-high essential signature genes after Enzalutamide treatment in VCaP xenografts. Differences between treatment and control groups were evaluated using the Wilcoxon test. F-H The Z-score of AR-high essential signature genes correlated with shorter PFI

Dynamic profiling identifies MCL1 as a key resistance driver

Considering that the overall expression levels of AR-high essential signature genes remained largely stable after Enzalutamide treatment, we further examined which individual genes exhibited the most pronounced changes. Time-course analysis of these AR-high essential signature genes revealed three temporal expression patterns post-Enzalutamide: persistently elevated (cluster1 including MCL1), stable (time-course cluster 2), and transiently suppressed (time-course cluster 3) (Fig. 6A-B and Supplementary Table 3). Interestingly, most genes in the time-course cluster 1 (46 out of 51) were not significantly elevated (logFC < 0.5) comparing tumor to normal tissue in the TCGA-PRAD cohort, and MCL1 had the most relevant trends. Single cell RNA data confirmed the elevated expression level of genes in time-course cluster 1 and MCL1 over-expression in drug-tolerant cells (Fig. 6C-D). Concordantly, MCL1 up-regulation was validated in resistant 22Rv1 and LNCaP (q-value < 0.05) cell lines (Fig. 6E-G). Drug sensitivity analysis revealed that the MCL1-specific inhibitor UMI-77 and the BCL-2 family inhibitor Sabutoclax exhibited lower IC50 values in AR-dependent cell lines (VCaP, 22RV1 and LNCaP) (Fig. 6H). In contrast, such advantages were not evident for the MCL1 inhibitors AZD5991 and MIM1. We also established Enzalutamide-resistant C4-2 and 22RV1 cell lines. Upon Enzalutamide treatment, the resistant cells exhibited significantly higher cell viability compared with the parental cells, particularly under higher drug concentrations (Fig. 6I). Moreover, we observed elevated expression of MCL1 in the resistant cell lines (Fig. 6J). Knockdown of MCL1 (Fig. 6K) markedly reduced cell proliferation, as reflected by decreased OD450 values (Fig. 6L). Flow cytometry analysis further demonstrated that MCL1 knockdown increased apoptosis rates to approximately 30% in both parental and resistant cells. Notably, this effect was more pronounced in the resistant cell lines, in which the baseline apoptosis rate of the negative control (NC) group was lower than that of the non-resistant cells (Fig. 6M).

Fig. 6.

Fig. 6

Open in a new tab

Time dependent analysis showed that MCL1 was essential in the resistance of Enzalutamide. A The time-course clustering results of AR-high essential signature genes. B The correlation of the fold change in expression in the TCGA-PRAD cohort and maxMembership as evaluated by the time-expression analyses. C-D The expression of genes in time-course cluster 1 and MCL1 in cell lines treated with or without Enzalutamide. Differences between treatment and control groups were evaluated using the Wilcoxon test. E-G Volcano plots showed that the expression of MCL1 were elevated after Enzalutamide treatment in 22RV1 and LNCaP cell lines. H The drug sensitivity to MCL1 of AR dependent, independent and normal prostate cell lines. I Establishment of Enzalutamide-resistant C4-2 and 22RV1 cell lines. Resistant cells exhibited significantly higher viability than parental cells upon Enzalutamide treatment. J The mRNA level of MCL1 in Enzalutamide resistant and sensitive cells. K The siRNA-mediated knockdown efficiency of MCL1 in C4-2 and 22RV1 Enzalutamide-resistant cells. L Cell viability analysis following siRNA-mediated knockdown of MCL1. M Flow cytometry analysis revealed increased apoptosis following MCL1 knockdown

Discussion

PCa is one of the most commonly diagnosed cancers among men worldwide, with a significant proportion of patients progressing to CRPC [21, 22]. The standard of care for advanced PCa involves androgen deprivation therapy (ADT), which initially leads to tumor regression by lowering androgen level [23]. However, the development of CRPC, characterized by disease progression despite castrate levels of serum testosterone, remains a major clinical challenge [24]. The continued activity of AR signaling pathway in CRPC has led to the development of second-generation AR-targeting agents, such as Enzalutamide and Apalutamide, which have shown efficacy in delaying disease progression [25, 26]. Despite these advancements, the emergence of resistance to these agents is inevitable, underscoring the need for a more comprehensive understanding of the molecular mechanisms underlying therapy resistance [27].

Our study provides a detailed analysis of AR dependency and resistance mechanisms in PCa, focusing on the distinct signaling dependencies of AR-dependent and AR-independent PCa subtypes. This analysis is crucial for the development of more effective treatment strategies that can overcome therapeutic resistance. By integrating transcriptomic data with clinical profiling, we systematically dissected the AR dependency and resistance mechanisms, revealing core essential pathways that define survival dependencies in PCa cells.

Our CRISPR screening analysis revealed distinct vulnerability landscapes for AR-dependent and AR-independent PCa cells, emphasizing the importance of pathway-specific dependencies in PCa progression and resistance. AR-dependent PCa cells showed enrichment in pathways associated with cell cycle regulation, such as the CDK/cyclin pathway, which is known to drive cell proliferation and survival in various cancer types [28]. This pathway’s activation in AR-dependent PCa cells could contribute to the aggressive nature of these tumors and their resistance to therapy. In contrast, AR-independent PCa cells were enriched in oxidative phosphorylation and mTOR signaling pathways. The mTOR pathway, in particular, has been extensively studied in cancers, including PCa, where it regulates cell growth, protein synthesis, and metabolism [29, 30]. Its overactivation is often associated with therapy resistance [31]. Our results suggest that targeting the mTOR pathway could be a promising strategy to overcome resistance in PCa.

The distinct pathway enrichments in AR-dependent and AR-independent PCa cells underscore the importance of personalized therapy approaches. Therapies targeting these specific pathways could be more effective in certain PCa subtypes. For example, inhibitors of the mTOR pathway have shown promise in preclinical models of cancer, including PCa, and could be particularly beneficial in AR-independent PCa cells where this pathway is highly active [32]. Similarly, therapies that target cell cycle regulation could be more effective in AR-dependet PCa cells [33]. These insights from our study align with the growing understanding of the heterogeneity of PCa and the need for tailored therapeutic strategies. The differential signaling dependencies also suggest potential synergies when combining pathway-specific therapies with existing AR-targeted therapies, which could enhance treatment efficacy [34, 35].

The identification of MCL1 as a key resistance driver following Enzalutamide treatment is a significant finding of our study. MCL1 has emerged as a critical anti-apoptotic protein in cancer, including PCa. Its overexpression has been associated with resistance to chemotherapy and targeted therapies in PCa [36, 37]. Previous study reports that LNCaP cells express higher levels of the anti-apoptotic protein MCL1 compared with PC-3 cells. This upregulation of MCL1, along with its regulatory partner USP9X, has been shown to limit the response of prostate cancer cells to radiotherapy [38]. In other cancer types, such as lung and breast cancer, MCL1 has also been implicated in therapy resistance. For example, in lung cancer, MCL1 upregulation has been linked to resistance to epidermal growth factor receptor (EGFR) inhibitors [39], and in breast cancer, it has been associated with resistance to tamoxifen [40]. Our findings align with these studies, suggesting that MCL1 may serve as a common resistance mechanism across different cancer types [41, 42]. Previous studies have reported that MCL1 is upregulated in Enzalutamide-resistant LNCaP cells, where it inhibits apoptosis and colony formation [43, 44]. Our findings extend these observations by showing that MCL1 plays a particularly important role in AR-dependent PCa cells, and that these cells are more sensitive to MCL1-specific inhibitors such as UMI-77. Targeting MCL1 or its upstream regulators could represent a promising strategy to overcome resistance in PCa. Furthermore, the role of MCL1 in mediating resistance to Enzalutamide suggests that combination therapies involving MCL1 inhibition may be particularly effective in treating CRPC.

Despite the novel insights provided by our study, there are limitations that should be acknowledged. First, our analysis was primarily based on in vitro CRISPR screening and publicly available datasets. The validation of our findings in larger, independent patient cohorts is necessary to confirm the clinical relevance of the identified genes and pathways. Second, while we have identified MCL1 as a key resistance driver, the precise mechanisms by which it mediates resistance to Enzalutamide require further investigation. Third, our study did not explore the potential of combining MCL1-targeting therapies with existing AR-targeted therapies, which could be an area of future research. Lastly, the functional validation of the identified genes and pathways in PCa models, including patient-derived xenografts (PDX), would provide additional evidence for their role in PCa progression and resistance.

In conclusion, our study provides a comprehensive analysis of AR dependency and resistance mechanisms in PCa, identifying distinct genetic and pathway dependencies that could inform the development of more effective treatment strategies. The identification of MCL1 as a key resistance driver highlights the potential of targeting anti-apoptotic pathways to overcome therapy resistance in PCa. Future studies should focus on validating these findings in larger patient cohorts and exploring the potential of combination therapies to enhance the efficacy of PCa treatment.

Supplementary Information

Supplementary Material 1. (68.7KB, xlsx)

Acknowledgements

Not applicable.

Authors’ contributions

QYZ, CQ and LMS proposed and designed this study. WL and WYK conducted most of the formal analyses. SLJ, YW and ZJY contributed to the embellishment of figures and wrote this manuscript. JY, XYW and XYC collected the data used for analyses. XWY and LMS reviewed the data. All authors confirmed the final version for submission.

Funding

Not applicable.

Data availability

The data supporting the findings of this study are available in the manuscript and its supplementary materials. RNA sequencing data, CRISPR screening results, and related analyses are publicly available through the TCGA-PRAD cohort (https://portal.gdc.cancer.gov/)，the GEO database ([https://www.ncbi.nlm.nih.gov/geo/](https:/www.ncbi.nlm.nih.gov/geo )) and the CCLE website (https://sites.broadinstitute.org/ccle).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Wei Liu, Weiyu Kong, Silin Jiang and Yong Wei contributed equally to this work.

Contributor Information

Luming Shen, Email: shenluming2008@163.com.

Chao Qin, Email: qinchao@njmu.edu.cn.

Qingyi Zhu, Email: zhuqingyi1971@njmu.edu.cn.

References

  • 1.Fleshner NE, Alibhai SMH, Connelly KA, et al. Adherence to oral hormonal therapy in advanced prostate cancer: a scoping review. Ther adv med oncol. 2023;15:17588359231152845. [DOI] [PMC free article] [PubMed]
  • 2.Zacharias N, Lee J, Ramachandran S, et al. Androgen Receptor Signaling in Castration-Resistant Prostate Cancer Alters Hyperpolarized Pyruvate to Lactate Conversion and Lactate Levels In Vivo. Mol imaging biol. 2019;21(1):86–94. [DOI] [PMC free article] [PubMed]
  • 3.Beltran H, Rickman DS, Park K, et al. Molecular characterization of neuroendocrine prostate cancer and identification of new drug targets. Cancer disc. 2011;1(6):487−95. [DOI] [PMC free article] [PubMed]
  • 4.Fritz V, Benfodda Z, Henriquet C, et al. Metabolic intervention on lipid synthesis converging pathways abrogates prostate cancer growth. Oncogene. 2013;32(42):5101−10. [DOI] [PMC free article] [PubMed]
  • 5.Jadvar H. PET of Glucose Metabolism and Cellular Proliferation in Prostate Cancer. J nuclear medicine : official publication, Society of Nucl Med. 2016;57(Suppl 3):25s–9s. [DOI] [PMC free article] [PubMed]
  • 6.Lei H, Wang Z, Jiang D, et al. CRISPR screening identifies CDK12 as a conservative vulnerability of prostate cancer. Cell death & disease, 2021;12(8):740. [DOI] [PMC free article] [PubMed]
  • 7.Tsherniak A, Vazquez F, Montgomery PG, et al. Defining a Cancer Dependency Map. Cell, 2017;170(3):564−76.e16. [DOI] [PMC free article] [PubMed]
  • 8.Li J, Xu C, Lee HJ, et al. A genomic and epigenomic atlas of prostate cancer in Asian populations. Nat. 2020;580(7801): 93−9. [DOI] [PubMed]
  • 9.Gröbner SN, Worst BC, Weischenfeldt J, et al. The landscape of genomic alterations across childhood cancers. Nat. 2018;555(7696):321−7. [DOI] [PubMed]
  • 10.Varambally S, Yu J, Laxman B, et al. Integrative genomic and proteomic analysis of prostate cancer reveals signatures of metastatic progression. Cancer cell, 2005;8(5):393–406. [DOI] [PubMed]
  • 11.Cai C, Wang H, He HH, et al. ERG induces androgen receptor-mediated regulation of SOX9 in prostate cancer. The J clinical invest. 2013;123(3):1109−22. [DOI] [PMC free article] [PubMed]
  • 12.Bolis M, Bossi D, Vallerga A, et al. Dynamic prostate cancer transcriptome analysis delineates the trajectory to disease progression. Nat commun. 2021;12(1):7033. [DOI] [PMC free article] [PubMed]
  • 13.Crowell PD, Giafaglione JM, Jones AE, et al. MYC is a regulator of androgen receptor inhibition-induced metabolic requirements in prostate cancer. Cell reports, 2023;42(10):113221. [DOI] [PMC free article] [PubMed]
  • 14.White RE, 3rd, Bannister M, Day A, et al. Saracatinib synergizes with enzalutamide to downregulate AR activity in CRPC. Front oncol. 2023;13:1210487. [DOI] [PMC free article] [PubMed]
  • 15.Suominen MI, Knuuttila M, Schatz CA, et al. Enhanced Antitumor Efficacy of Radium-223 and Enzalutamide in the Intratibial LNCaP Prostate Cancer Model. Internal j mol sci. 2023;24(3):21892023;24(3). 10.3390/ijms24032189. [DOI] [PMC free article] [PubMed]
  • 16.Qian C, Yang Q, Rotinen M, et al. ONECUT2 acts as a lineage plasticity driver in adenocarcinoma as well as neuroendocrine variants of prostate cancer. Nucleic acids res. 2024;52(13):7740−60. [DOI] [PMC free article] [PubMed]
  • 17.Knuuttila M, Mehmood A, Huhtaniemi R, et al. Antiandrogens Reduce Intratumoral Androgen Concentrations and Induce Androgen Receptor Expression in Castration-Resistant Prostate Cancer Xenografts. The Amer j pathol. 2018;188(1):216−28. [DOI] [PubMed]
  • 18.Dong B, Miao J, Wang Y, et al. Single-cell analysis supports a luminal-neuroendocrine transdifferentiation in human prostate cancer. Commun biol. 2020;3(1): 778. [DOI] [PMC free article] [PubMed]
  • 19.Taavitsainen S, Engedal N, Cao S, et al. Single-cell ATAC and RNA sequencing reveal pre-existing and persistent cells associated with prostate cancer relapse. Nat commun. 2021;12(1):5307. [DOI] [PMC free article] [PubMed]
  • 20.Kiviaho A, Eerola SK, Kallio HM L, et al. Single cell and spatial transcriptomics highlight the interaction of club-like cells with immunosuppressive myeloid cells in prostate cancer. Nat commun. 2024;15(1):9949. [DOI] [PMC free article] [PubMed]
  • 21.Siegel RL, Miller KD, Jemal A. Cancer Statistics, 2017. CA: a ca j clinic. 2017;67(1):7–30. [DOI] [PubMed]
  • 22.Nguyen-Nielsen M, Borre M. Diagnostic and Therapeutic Strategies for Prostate Cancer. Sem nucl med. 2016;46(6):484−90. [DOI] [PubMed]
  • 23.Spratt DE, Malone S, Roy S, et al. Prostate Radiotherapy With Adjuvant Androgen Deprivation Therapy (ADT) Improves Metastasis-Free Survival Compared to Neoadjuvant ADT: An Individual Patient Meta-Analysis. Journal of clinical oncology : official journal of the Amer Soc Clinic Oncol. 2021;39(2):136−44. [DOI] [PMC free article] [PubMed]
  • 24.Ryan CJ, Smith MR, De Bono JS, et al. Abiraterone in metastatic prostate cancer without previous chemotherapy. The New England j med. 2013;368(2):138−48. [DOI] [PMC free article] [PubMed]
  • 25.Watson PA, Arora VK, Sawyers CL. Emerging mechanisms of resistance to androgen receptor inhibitors in prostate cancer. Nat rev Can. 2015;15(12):701−11. [DOI] [PMC free article] [PubMed]
  • 26.Scher HI, Fizazi K, Saad F, et al. Increased survival with enzalutamide in prostate cancer after chemotherapy. The New England j med. 2012;367(13):1187–97. [DOI] [PubMed]
  • 27.Small EJ, Saad F, Chowdhury S, et al. Apalutamide and overall survival in non-metastatic castration-resistant prostate cancer. Annals of oncology : official j Eur Soc Med Oncol. 2019;30(11):1813–20. [DOI] [PMC free article] [PubMed]
  • 28.Asghar U, Witkiewicz AK, Turner NC, et al. The history and future of targeting cyclin-dependent kinases in cancer therapy. Nat rev Drug disc. 2015;14(2):130−46. [DOI] [PMC free article] [PubMed]
  • 29.Gupta S, Argilés G, Munster PN, et al. A Phase I Trial of Combined Ridaforolimus and MK-2206 in Patients with Advanced Malignancies. Clinical cancer research : an official j Amer Assoc Ca Res. 2015;21(23):5235−44. [DOI] [PubMed]
  • 30.Laplante M, Sabatini DM. mTOR signaling in growth control and disease. Cell, 2012;149(2):274 − 93. [DOI] [PMC free article] [PubMed]
  • 31.Tewari D, Patni P, Bishayee A, et al. Natural products targeting the PI3K-Akt-mTOR signaling pathway in cancer: A novel therapeutic strategy. Sem ca biol. 2022;80:1–17. [DOI] [PubMed]
  • 32.Gujrati H, Ha S, Waseem M, et al. Downregulation of miR-99b-5p and Upregulation of Nuclear mTOR Cooperatively Promotes the Tumor Aggressiveness and Drug Resistance in African American Prostate Cancer. Internal J Mol Scie. 2022;23(17):9643. 10.3390/ijms23179643. [DOI] [PMC free article] [PubMed]
  • 33.Mcnair C, Urbanucci A, Comstock CE, et al. Cell cycle-coupled expansion of AR activity promotes cancer progression. Oncogene. 2017;36(12):1655−68. [DOI] [PMC free article] [PubMed]
  • 34.Yeh Y, Guo Q, Connelly Z, et al. Wnt/Beta-Catenin Signaling and Prostate Cancer Therapy Resistance. Adv exp med biol. 2019;1210:351−78. [DOI] [PubMed]
  • 35.Einstein DJ, Arai S, Balk SP. Targeting the androgen receptor and overcoming resistance in prostate cancer. Curr opinion oncol. 2019;31(3):175 − 82. [DOI] [PMC free article] [PubMed]
  • 36.Santer FR, Erb HH, Oh SJ, et al. Mechanistic rationale for MCL1 inhibition during androgen deprivation therapy. Oncotarget. 2015;6(8):6105−22. [DOI] [PMC free article] [PubMed]
  • 37.Pasqualini L, Bu H, Puhr M, et al. miR-22 and miR-29a Are Members of the Androgen Receptor Cistrome Modulating LAMC1 and Mcl-1 in Prostate Cancer. Mol endocrinol (Baltimore, Md), 2015;29(7):1037−54. [DOI] [PMC free article] [PubMed]
  • 38.Hogh-Binder SA, Klein D, Wolfsperger F, et al. Protein Levels of Anti-Apoptotic Mcl-1 and the Deubiquitinase USP9x Are Cooperatively Upregulated during Prostate Cancer Progression and Limit Response of Prostate Cancer Cells to Radiotherapy. Can. 2023;15(9):2496. 10.3390/cancers15092496. [DOI] [PMC free article] [PubMed]
  • 39.Hirai S, Tada M, Yamaguchi M, et al. EGFR-independent EGFR-mutant lung adenocarcinoma cells depend on Bcl-xL and MCL1 for survival. Biochem biophysical res commun. 2020;526(2):417 − 23. [DOI] [PubMed]
  • 40.Xie Y, Han X, Yu J, et al. EGR3 and estrone are involved in the tamoxifen resistance and progression of breast cancer. J ca res clin oncol. 2023;149(20):18103−17. [DOI] [PMC free article] [PubMed]
  • 41.Wei G, Margolin AA, Haery L, et al. Chemical genomics identifies small-molecule MCL1 repressors and BCL-xL as a predictor of MCL1 dependency. Ca cell. 2012;21(4):547−62. [DOI] [PMC free article] [PubMed]
  • 42.Duan L, Tadi MJ, O’Hara KM, et al. Novel markers of MCL1 inhibitor sensitivity in triple-negative breast cancer cells. The J biol chem. 2024;300(6):107375. [DOI] [PMC free article] [PubMed]
  • 43.Handle F, Prekovic S, Helsen C, et al. Drivers of AR indifferent anti-androgen resistance in prostate cancer cells. Sci rep. 2019;9(1):13786. [DOI] [PMC free article] [PubMed]
  • 44.Tambe P, Salve R, Choudhary P, et al. Targeted silencing of the MCL-1 gene using multi-layered dendrimer-based nanoconstructs achieves efficient tumor regression in xenografted mice models. Internal j pharm. 2023;634:122659. [DOI] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (68.7KB, xlsx)

Data Availability Statement

The data supporting the findings of this study are available in the manuscript and its supplementary materials. RNA sequencing data, CRISPR screening results, and related analyses are publicly available through the TCGA-PRAD cohort (https://portal.gdc.cancer.gov/)，the GEO database ([https://www.ncbi.nlm.nih.gov/geo/](https:/www.ncbi.nlm.nih.gov/geo )) and the CCLE website (https://sites.broadinstitute.org/ccle).

Articles from BMC Cancer are provided here courtesy of BMC

RESOURCES