GG-NER’s role in androgen receptor signaling inhibitor response for advanced prostate cancer

Abstract

Background

Advanced prostate cancer (PCa) often initially responds to androgen receptor signaling inhibitors (ARSI) but frequently develops resistance, driven by tumor heterogeneity and therapeutic pressure. Addressing the clinical challenge of identifying non-responsive patients and discovering new therapeutic targets is urgently needed.

Methods

We utilized single-sample gene set enrichment analysis (ssGSEA) to elucidate the influence of the GG-NER pathway on ARSI response in PCa. We then constructed and validated a prognostic model based on this pathway using LASSO regression, Kaplan-Meier analysis, Cox regression, and ROC analysis. Additionally, we mapped tumor mutations to delineate the mutational landscapes across different risk groups and explored functional pathways through GO, KEGG, and GSEA analyses. The impact of the GG-NER pathway on enzalutamide sensitivity and DNA repair in PCa was further validated through CCK-8 assays, colony formation assays, in vivo experiments, and immunofluorescence.

Results

ssGSEA indicated a trend of GG-NER pathway upregulation in patients with poor ARSI response. The GG-NER characteristic gene score (NECGS) identified a high-risk group with diminished ARSI response, serving as an independent prognostic indicator with strong predictive power. This high-risk group exhibited elevated TP53 mutation frequencies and significant enrichment in key pathways such as ribosome and mitochondrial functions, as well as MYC and E2F signaling. Experimental validation confirmed that targeting the GG-NER pathway or its key gene, ACTL6A, significantly reduces enzalutamide resistance in resistant cell lines and increases γH2AX expression.

Conclusion

NECGS effectively predicts ARSI response in PCa, and our comprehensive analysis underscores the critical role of the GG-NER pathway in enzalutamide resistance, positioning ACTL6A as a potential therapeutic target for PCa.

Introduction

Prostate cancer is the second most common cancer among men worldwide, accounting for 11% of all cancer deaths and ranking fifth globally [1]. In its early stages, prostate cancer progresses slowly, and patients often achieve favorable prognoses through surgical intervention and radiotherapy. Androgen deprivation therapy (ADT), which includes surgical castration (orchiectomy) and chemical castration (using drugs like Degarelix and GnRH analogs), remains the cornerstone of treatment for advanced prostate cancer, with most patients initially responding well. However, many tumors eventually progress to castration-resistant prostate cancer (CRPC), a more aggressive form of the disease with a median survival of only about two years [2]. Although CRPC is less responsive to androgen deprivation, most tumor cells continue to rely on androgen receptor (AR) signaling. The advent of second-generation AR signaling inhibitors (ARSIs) such as Enzalutamide has provided more targeted treatment options for CRPC [3].

As research into tumor heterogeneity advances, it has become clear that dysregulation of the DNA damage repair (DDR) pathway plays a critical role in the progression of prostate cancer, influencing the tumor’s response to radiotherapy and anti-androgen therapy [4]. Patients with germline DDR gene mutations progress to metastatic CRPC (mCRPC) more rapidly than those without such mutations and respond better to novel endocrine therapies [5]. Among DDR gene mutations, those in the Homologous Recombination Repair (HRR) pathway are most common, with BRCA2 mutations being particularly prevalent. Targeting these mutations with Poly (ADP-ribose) Polymerase inhibitors (PARPi) can block the single-strand DNA base excision repair mechanism, leading to the accumulation of double-strand breaks and ultimately inducing tumor cell death—a phenomenon known as “synthetic lethality” [6]. Interestingly, a prospective clinical study has shown that some mCRPC patients without HRR mutations can still benefit from PARPi treatment [7]. Additionally, ARSI drugs have been found to induce DNA double-strand break damage in prostate cancer cells, with PARPi inhibitors significantly enhancing this therapeutic effect [8]. However, retrospective studies comparing the treatment outcomes of patients with and without DDR mutations have found no significant differences, suggesting that tumor cells may develop resistance by bypassing synthetic lethality through alternative DNA damage repair pathways [9]. Oshima et al. used clinical samples and genome-wide RNA-seq to identify various DDR-related genes as biomarkers for CRPC. This study suggests that exploring DDR pathways could broaden the search for novel biomarkers and therapeutic targets for CRPC [10].

In our study, using single-sample Gene Set Enrichment Analysis (ssGSEA) with 1,928 representative signaling pathways from the Reactome database, we identified the “DNA Damage Recognition-Global Genome Nucleotide Excision Repair (GG-NER)” pathway as abnormally regulated in mCRPC and as a risk factor for ARSI treatment response. NER is a crucial mechanism for maintaining genome stability, primarily responsible for repairing damage that distorts the DNA double helix [11]. NER consists of two sub-pathways: Transcription Coupled Nucleotide Excision Repair (TC-NER) and GG-NER. Unlike TC-NER, which is confined to transcribed DNA, GG-NER operates across the entire genome, playing a vital role in preserving genomic integrity [12]. Certain hereditary disorders, such as Xeroderma pigmentosum (XP), are linked to defects in the GG-NER pathway and are associated with a heightened risk of cancer, particularly ultraviolet-induced skin cancer [13]. Moreover, because chemotherapy primarily exerts its effects by damaging DNA, increased GG-NER activity can enable cancer cells to repair distorted DNA structures across the genome, thereby enhancing their resistance to chemotherapy [14, 15].

In this study, we observed abnormal upregulation of the GG-NER pathway in mCRPC through ssGSEA. Consequently, we developed a GG-NER pathway characteristic gene risk model to predict ARSI response in the mCRPC cohort using the LASSO algorithm. We further combined gene mutation analysis with in vitro and in vivo experiments to validate the impact of the GG-NER pathway on ARSI response in prostate cancer. Finally, through functional enrichment analysis and experimental verification of the GG-NER pathway’s characteristic gene, ACTL6A, we explored the potential mechanisms by which the GG-NER pathway regulates ARSI response in prostate cancer. The overall research flowchart is depicted in Fig. 1.

Fig. 1

Flowchart of the study design

Materials and methods

Data collection and processing

The transcriptome datasets and related patient survival information of SU2C (cohort 1 and cohort 2) and WCDT cohort were downloaded from the cBioPortal website (https://www.cbioportal.org/) and GDC Data portal (https://portal.gdc.cancer.gov/projects), respectively. The relative data in the TCGA-PRAD, GSE32269, GSE35988-GPL6480, GSE35988-GPL6848 were obtained from the PCaDB website (http://bioinfo.jialab-ucr.org/PCaDB/). Additionally, the transcriptome and clinical information data of 32 solid tumors collected by the TCGA database from The University of California at Santa Cruz (UCSC) Xena website (https://xenabrowser.net/datapages/). Detailed information about these cohorts is summarized in Table S1.

Construction and verification of the GG-NER related prognostic model

We collected 1928 signature gene sets from the Reactome database (https://reactome.org/) updated in 2021. The single sample Gene Set Enrichment Analysis (ssGSEA) via the “GSVA” package [16] was used to score the signature gene sets mentioned above for every patient in the cohort 1 and cohort 2. The relationships between the ssGSEA scores of the signature gene sets and the response to ARSI treatment in prostate cancer were analyzed using the “LIMMA“ [17], “survminer“(version 0.4.9), and “survival“(version 3.5-7) R packages. Based on the GG-NER gene set, a three-gene prognostic model was constructed with the cohort 1 as a training set via the Least Absolute Shrinkage and Selection Operator (LASSO) method in the “glmnet” package (Version 4.1-8). The formula below, generated from the prognostic model, can be applied to quantify patients’ responses to the ARSI treatment.

Risk score = $\:{\sum\:}_{i=1}^{N}Coefficient\times\:\:gene\:expression$

As presented in the formula, the N represents the number of genes included in the model, the Coefficient represents the corresponding LASSO regression coefficient for each gene, and the gene expression denotes the expression value of each gene. Prostate cancer patients in the Training and Validating cohorts were divided into high- and low-risk score groups based on the optimal cut-off point using the “survminer” R package. Based on the median risk score, patients in the TCGA-PRAD cohort were further divided into high- and low-risk subgroups. The differences in ARSI treatment responses and progression-free interval (PFI) between patients in the two aforementioned subgroups are evaluated using the Kaplan-Meier survival analysis. The univariate and multivariate COX regression analyses were used to examine the model’s independence in predicting the prognoses of patients. The predictive performance of the risk score for patient prognosis was evaluated using the AUC in the “pROC” package [18].

Tumor mutation analysis

The cBioPortal website (https://www.cbioportal.org/) provided the mutational data, and the “ComplexHeatmap” R package [19] was utilized to visualize the mutational details of driver genes in the training set, validating set, and TCGA-PRAD set. For mutation analysis, the top 20 driver genes were chosen from the training set, validating set, and TCGA-PRAD set.

Pathway enrichment analysis

The “clusterProfiler” [20] R package was used to conduct the Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) (|logFC| > 0, p < 0.05), and Gene Set Enrichment Analysis (GSEA) analyses (|ES| > 0). The gene sets used for GSEA enrichment analysis were downloaded from the Hallmark gene set (H) in the Molecular Signatures Database (MSigDB) (https://www.gsea-msigdb.org/gsea/msigdb).

Construction and verification of enzalutamide-resistant PCa cell lines

LNCaP and C4-2B cell lines were purchased from BeiNa biotechnology company and both cultured in RPMI-1640 medium (BC-M-017-500mL, Bio-Channel) with 10% Fetal bovine serum (BC-SE-FBS07, Bio-Channel) and 1% Penicillin/ Streptomycin (15140122, Gibco) in a humid incubator with 5% CO2 and 37℃. LNCaP and C4-2B cells were cultivated in RPMI-1640 medium with 10µM and 25µM enzalutamide (S1250, Selleck) for 3 months to generated enzalutamide-resistant cell lines (LNCaP_ENZR and C4-2B_ENZR). 5000 cells per well were seeded into the 96-well plate, and CCK-8 (MA0218, MeilunBio) assay was used to examin the sensitivity to enzalutamide in LNCaP, C4-2B, LNCaP_ENZR and C4-2B_ENZR cells after treatment of gradient concentration of enzalutamide (0µM, 2.5µM, 5µM, 12.5µM, 25µM, 50µM). 5000 cells per well were seeded into the 6-well plate, and plate clone formation assay was used to examine the sensitivity to enzalutamide in LNCaP, C4-2B, LNCaP_ENZR and C4-2B_ENZR cells under 10µM and 25µM enzalutamide treatment for 12 days.

Experimental validation of GG-NER impacting role in prostate caner

CCK-8 assay was used to examin the sensitivity to HAMNO in LNCaP_ENZR and C4-2B_ENZR cells after treatment of gradient concentration of HAMNO (S0148, Selleck) (0µM, 5µM, 10µM, 20µM, 40µM, 80µM) and the examination timepoint were 24 h, 48 h and 72 h to select the best timepoint for further research. CCK-8 and plate clone formation assay were applied to detect the sensitivity to enzalutamide in LNCaP_ENZR and C4-2B_ENZR cells by treating with 10µM HAMNO or knocking down ACTL6A. siRNAs and shRNAs targeting to ACTL6A were designed and synthesized by Tsingke company and the target sequences were listed in supplementary Table 2. The transfection, quantitative PCR (qPCR) and Western blot (WB) examination procedure can be referred to our previous study [21]. Anti-ACTL6A primary antibody (A7709, Abclonal) were purchased from Abclonal company. 30,000 cells per well were seeded into the 3.5 cm plate and immunofluorescence assay were utilized to detect the expression of γH2AX (YM1429, Immunoway) in LNCaP_ENZR and C4-2B_ENZR cells after treating with 10µM HAMNO for 48 h or by knocking down ACTL6A. The detailed procedure of immunofluorescence can be referred to our previous study [21].

Animal experiments

All animal experiments were conducted in accordance with ethical standards and guidelines for the care and use of laboratory animals. The study protocol was approved by the Animal Care and Use Committee of Zhujiang Hospital. Twenty-four 6-week-old male NSG mice were purchased from GemPharmatech Co., Ltd. Each mouse received a subcutaneous injection of 5 × 10^6 C4-2B_ENZR cells in the dorsal region. Once the tumor volume reached approximately 50 mm³, the mice were randomly divided into four groups (n = 6 per group) and subjected to different treatments: vehicle, HAMNO (1 mg/kg, intraperitoneally, administered 5 days per week), enzalutamide (10 mg/kg, orally, administered 4 days per week), and a combination of HAMNO and enzalutamide. Additionally, one group of mice (n = 10) was implanted with C4-2B_ENZR sh-NC cells, while the other group (n = 10) received C4-2B_ENZR sh-ACTL6A#2 cells. Once the tumor volume reached approximately 50 mm³, both groups were randomly divided into two subgroups and treated with either vehicle or enzalutamide. During the treatment period, tumor volume was measured regularly using the formula 0.5 × length × (width)². After two weeks of drug treatment, the mice were euthanized, and the tumors were excised and weighed to assess final tumor weight.

Statistical analysis

All bioinformatic analyses were performed by R software version 4.3.0 (The R Project for Statistical Computing, Vienna, Austria). The “ComplexHeatmap” package was applied for tumor mutation burden analysis. The “survival” and “survminer” packages were applied for KM plots and Cox regression analysis. GraphPad Prism 9.0 (GraphPad, La Jolla, CA, United States) was utilized to analyze the results of cell functional assays. Two groups ofs data were shown as mean ± SD (standard deviation) using a two-sided Students’ t-test, and a one-way analysis of variance (ANOVA) was applied to three or more independent groups. The results with a p-value < 0.05 were regarded as statistically significant ones.

Results

Identification of the aberrant GG-NER pathway in prostate cancer patients

To investigate signaling pathways potentially subject to aberrant regulation in prostate cancer patients undergoing ARSI treatment, we selected 1,928 representative pathways from the Reactome database and performed ssGSEA analysis to score patients in Cohorts 1 and 2 (Tables S3-S4). The results revealed that the “DNA recognition in GG-NER pathway” was significantly enriched in mCRPC samples from patients who had paused ARSI treatment (Fig. 2A-B and E). Additionally, univariate Cox regression analysis showed a correlation between high ssGSEA scores for the GG-NER pathway and poor ARSI responses in prostate cancer patients (Figures S1A-B, Table S5). Kaplan-Meier plots indicated that patients with upregulated GG-NER pathway expression experienced worse responses to ARSI and shorter overall survival times during treatment (Fig. 2C-D).

Fig. 2

Expression of the GG-NER pathway in mCRPC cohorts. A&B. ssGSEA scoring analysis of upregulated and downregulated signaling pathways in Cohorts 1 and 2. C&D. Kaplan-Meier curves showing the association between GG-NER pathway gene set expression and overall survival and ARSI response in prostate cancer in Cohorts 1 and 2. E. the hologram of GG-NER pathway from Reactome database

Experimental validation of the impacts of GG-NER pathway on enzalutamide sensitivity in prostate cancer

As a matter of fact, patients in both cohorts 1 and 2 received enzalutamide and/or abiraterone. While enzalutamide directly targets the androgen receptor (AR), and abiraterone primarily inhibits CYP17 to reduce androgen synthesis, both function by blocking AR signaling. Given our aim to directly assess effects on AR-mediated transcriptional activity—particularly relevant to the GG-NER pathway—we prioritized enzalutamide as a representative ARSI. First of all, we administered 10µM or 25µM enzalutamide to LNCAP and C4-2B cells for 3 months respectively, and observed that LNCAP_ENZR and C4-2B_ENZR cells exhibited resistance to enzalutamide compared to their respective wild-type lines, as demonstrated by CCK-8 and colony formation assays (Figures S2A-D). Given the significant upregulation of the GG-NER pathway in prostate cancer patients who are resistant to ARSI, we selected HAMNO, a NER pathway-related inhibitor, for further investigation. We initially treated LNCAP_ENZR and C4-2B_ENZR cell lines with various concentrations of HAMNO and assessed its effects on cell growth over time. HAMNO inhibited cell growth in a concentration- and time-dependent manner. Based on these results, we selected a concentration of 10µM HAMNO and a timepoint of 48 h for further experiments (Figures S3A-B). The CCK-8 assay revealed that 10µM HAMNO significantly increased the sensitivity of LNCAP_ENZR and C4-2B_ENZR cells to enzalutamide compared to the control group (DMSO) (Fig. 3A-B). Additionally, the colony formation assay showed a reduction in the cloning capacity of LNCAP_ENZR and C4-2B_ENZR cells following 10 µM HAMNO treatment, with further inhibition observed when HAMNO was combined with enzalutamide (Fig. 3C-D). In vivo experiments further supported these findings. Mice treated with enzalutamide plus HAMNO had smaller tumors compared to those in the vehicle group (Fig. 3E-G). Moreover, the combination treatment resulted in increased DNA double-strand break damage, as indicated by elevated levels of the DNA damage marker γH2AX in LNCaP_ENZR and C4-2B_ENZR cells (Fig. 3H-I). These results suggest that the GG-NER pathway plays a significant role in modulating the response of mCRPC patients to ARSI.

Fig. 3

Impact of the GG-NER pathway on ARSI response in prostate cancer. A&B. CCK-8 assay evaluating the effect of HAMNO on the sensitivity of LNCaP_ENZR and C4-2B_ENZR cells to enzalutamide (0µM, 2.5µM, 5µM, 12.5µM, 25µM, 50µM and each treatment gradient was set with five replicates). C&D. Colony formation assay assessing HAMNO’s influence on the sensitivity of LNCaP_ENZR and C4-2B_ENZR cells to enzalutamide. E. Overall image of subcutaneous xenografts in different treatment groups. F. Tumor volume changes across different groups. G. Tumor weight across different groups. H&I. Immunofluorescence detection of DNA double-strand break damage marker (γH2AX) in LNCaP_ENZR and C4-2B_ENZR cells treated with HAMNO and (or) enzalutamide

Construction and validation of the GG-NER characteristic gene score model

To identify key regulators in the GG-NER pathway that may influence ARSI responses, we employed the LASSO-Cox method to construct a predictive model using Cohort 1 as the training set, selecting the model based on the minimum partial likelihood deviance and its corresponding regularization parameter (λ), while prioritizing the most frequently selected model. Ultimately, a three-gene (ACTL6A, PARP2, and RUVBL1) predictive model named GG-NER Characteristic Gene Score (NECGS) (Figure S4A-C, Table S6). Kaplan-Meier (KM) plots revealed that patients with higher NECGS risk scores had poorer ARSI responses, and the expressions of the three genes correlated positively with ARSI responses in prostate cancer patients (Fig. 4A). Univariate and multivariate Cox regression analyses identified NECGS and Gleason score as independent risk factors affecting prognosis in PCa patients (Fig. 4B). To further evaluate the model’s efficacy, we performed ROC curve analysis, which indicated that NECGS’s predictive accuracy improved over time, reaching its peak at 24 months (8-month AUC = 0.73; 12-month AUC = 0.78; 24-month AUC = 0.81) (Fig. 4C). To assess the model’s reproducibility, we applied it to the cohort 2 as a validation set. KM plots again demonstrated that higher risk scores were associated with shorter ARSI response times, and the NECGS distribution curve confirmed that higher patient risk scores correlated with worse ARSI responses or higher expression levels of ACTL6A, PARP2, and RUVBL1 (Fig. 4D). Univariate and multivariate Cox regression analyses in the validation set reaffirmed that NECGS is an independent risk factor influencing ARSI responses (Fig. 4E). ROC curve analysis further showed that the model had optimal predictive performance at 24 months (8-month AUC = 0.63; 12-month AUC = 0.78; 24-month AUC = 0.91) (Fig. 4F). Furthermore, we also performed KM plot, Cox regression and ROC curve analysis in WCDT cohort. KM plot showed that low risk scores of NECGS group displayed longer overall survival time in mCRPC patients (Figure S5A). Though univariate cox regression showed that high NECGS scores of patients has a high risk of progression, the result was non-significant statistically (Figure S5B). However, the multivariate cox regression analysis showed that the NECGS score might be an independent risk to progression of mCRPC patients (Figure S5C). ROC curve showed that the model had a best predictive power at 60 months (12-month AUC = 0.61; 36-month AUC = 0.66; 60-month AUC = 0.77) (Figure S5D).

Fig. 4

Construction and validation of the NECGS model. A&D. KM plots and risk curves showing the correlation between NECGS and ARSI response, including ACTL6A, PARP2, and RUVBL1 expression in training and validation sets. B&E. Univariate and multivariate Cox regression analyses of NECGS’s impact on ARSI response in training and validation sets. C&F. ROC curve analyses of NECGS’s predictive efficacy for ARSI response in training and validation sets

Pan cancer-wide analysis of NECGS

The NECGS predictive model demonstrates effective prognostic capabilities in prostate cancer. To further assess its accuracy across different tumor types, we tested the model in 33 additional solid tumors, including primary prostate cancer, within the TCGA cohort. NECGS showed higher scores in several cancers, including Lung Squamous Cell Carcinoma (LUSC), Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma (CESC), Testicular Germ Cell Tumors (TGCT), and Uterine Carcinosarcoma (UCS). Conversely, it exhibited lower scores in Kidney Chromophobe (KICH), Liver Hepatocellular Carcinoma (LIHC), Uveal Melanoma (UVM), and Pheochromocytoma and Paraganglioma (PCPG) (Fig. 5A). Given the variability in gene expression profiles across different tumors, risk score values alone may not fully capture the complexity of tumor progression. Therefore, we conducted univariate Cox regression analysis to validate the model further. This analysis revealed that the risk score is a significant prognostic factor for Overall Survival (OS) and Progression-Free Interval (PFI) in several cancers, including Prostate Cancer (PCa), Adrenocortical Carcinoma (ACC), KICH, Mesothelioma (MESO), Brain Lower Grade Glioma (LGG), Liver Hepatocellular Carcinoma (LIHC), Lung Adenocarcinoma (LUAD), and Sarcoma (SARC), as well as Pancreatic Adenocarcinoma (PAAD) (Fig. 5B-C). Additionally, the risk score was identified as a critical factor affecting PFI in Bladder Cancer (BLCA) patients, though it did not significantly impact OS. In contrast, NECGS was found to be a favorable factor for Rectum Adenocarcinoma (READ). For other cancers, the model’s impact on PFI and OS was not statistically significant (Fig. 5B-C). These results highlight that the NECGS predictive model retains robust prognostic power in certain tumor types, demonstrating its potential utility beyond prostate cancer.

Fig. 5

Prognostic performance of NECGS in TCGA pan-cancer. (A) Boxplot of NECGS levels across various cancer types. (B) Univariate Cox regression analysis of NECGS and overall survival (OS) in different tumor types. (C) Relationship between NECGS and progression-free interval (PFI) in various tumors

Tumor mutation analysis in high- and Low- NECGS groups

Abnormal regulation in DNA damage repair (DDR) pathways is commonly associated with genomic instability and gene mutations. We analyzed the effects of NECGS on gene mutations across various cohorts. In all datasets, TP53 mutation was the most frequent in PCa patients, with a significantly higher mutation rate in the high NECGS group in the TCGA-PRAD cohort (p < 0.0001) (Fig. 6A-C). Although TP53 mutation rates were also higher in the high NECGS group in both the training (p = 0.137) and validation sets (p = 0.766), these differences were not statistically significant. Additionally, FOXA1 (p = 0.02) and SPTA1 (p = 0.003) showed significantly higher mutation rates in the high NECGS group compared to the low-risk group in TCGA-PRAD (Fig. 6C); however, these differences were not observed in the training and validation sets.

Fig. 6

Mutation panorama of training set, validation set and TCGA-PRAD set. (A) Mutation overview of the top 10 driver genes in the training set. (B) Mutation overview of the top 10 driver genes in the validation set. (C) Mutation overview of the top 10 driver genes in the TCGA-PRAD set

Biological functional analyses of NECGS in PCa

To elucidate the biological functions of the GG-NER pathway, we conducted GO, KEGG, and GSEA analyses. Differentially expressed genes (DEGs) between the high- and low-risk groups were selected based on criteria (p < 0.05, |log2FC| > 1) for GO pathway enrichment analysis (Table S7&S8). In the top five pathways identified for Biological Process (BP), Cellular Component (CC), and Molecular Function (MF), the high-risk subgroup was primarily associated with cellular ribosome translation, chromosome segregation and genomic DNA damage, while the low-risk group focused on cell membrane structure and function (Fig. 7A-B, S6A-B; Table S9&S10). KEGG analysis revealed that the high-risk group was notably enriched in pathways related to ribosomes, p53 signaling pathway, oxidative phosphorylation, and the cell cycle. In contrast, the low-risk group showed enrichment in pathways related to cholesterol synthesis, glutathione metabolism and ion channels (Fig. 7C-D, S6C-D; Table S9&S10). GSEA analysis further elucidated the potential mechanisms, indicating that the high-risk group was predominantly associated with signaling pathways involved in MYC, DNA damage repair, cell cycle regulation (E2F, G2M), and metabolic processes such as MTOR, oxidative phosphorylation, and unfolded protein response (Fig. 7E-F, Table S11).

Fig. 7

Functional enrichment analysis of risk score. A&B. GO analysis of upregulated and downregulated pathways in the high-risk group, including biological process (BP), cellular component (CC), and molecular function (MF). C&D. KEGG analysis of upregulated and downregulated pathways in the high-risk group. E&F. GSEA analysis of enriched signaling pathways in the high-risk groups in the training set and TCGA-PRAD cohort

Experimental validation of ACTL6A’s role in ARSI treatment sensitivity

The risk model includes three feature genes: PARP2, RUVBL1, and ACTL6A. While previous studies have emphasized the roles of PARP2 and RUVBL1 in ARSI treatment resistance in prostate cancer [22, 23], the relationship between ACTL6A and ARSI treatment has been less explored. Notably, the coefficient for ACTL6A is the highest in this model, prompting us to select ACTL6A for further experimental validation. We first examined ACTL6A expression across various public datasets and found that ACTL6A was highly expressed in primary prostate cancer (PCa) compared to normal tissues, with no statistically significant difference between primary and metastatic CRPC (mCRPC) samples, though the average expression was higher in mCRPC (Figure S7A-C). Additionally, ACTL6A expression was significantly elevated in enzalutamide-resistant PCa cell lines compared to wild-type cells (Figure S7D-E). To investigate ACTL6A’s role in enzalutamide sensitivity, we designed two specific siRNAs targeting ACTL6A. qPCR and Western blot confirmed efficient knockdown of ACTL6A in LNCaP_ENZR and C4-2B_ENZR cell lines (Fig. 8A-B, S7F-G). CCK-8 and colony formation assays showed that ACTL6A knockdown significantly reduced enzalutamide resistance in these cells, underscoring ACTL6A’s role in modulating ARSI treatment sensitivity in prostate cancer (Fig. 8C-F). In addition, we established stable ACTL6A knockdown cell lines and employed a subcutaneous xenograft model using C4-2B_ENZR cells. The results demonstrated that ACTL6A knockdown enhanced the sensitivity to enzalutamide treatment (Fig. 8G-J). Furthermore, enzalutamide treatment in ACTL6A-knockdown cells resulted in a marked increase in γH2AX expression, indicating enhanced DNA damage (Figure S8A-B).

Fig. 8

Experimental validation of ACTL6A’s role in enzalutamide sensitivity. A&B. Western blot validation of ACTL6A knockdown efficiency by siRNA. C&D. CCK-8 assay detecting changes in sensitivity to enzalutamide (0µM, 2.5µM, 5µM, 12.5µM, 25µM, 50µM; and each treatment gradient was set with five replicates) after ACTL6A knockdown (duration of siRNA transfection for 48 h). E&F. Colony formation assay assessing changes in sensitivity to enzalutamide post-ACTL6A knockdown (siRNA transfection was performed every 96 h). G. Western blot validation of ACTL6A knockdown efficiency by shRNA. H. Overall image of subcutaneous xenografts in different treatment groups. I. Tumor volume changes across different groups. J. Tumor weight across different groups

Potential regulatory role of ACTL6A in ARSI treatment sensitivity

To explore the role of ACTL6A in regulating enzalutamide sensitivity in prostate cancer, we divided the training set and WCDT cohorts into high and low ACTL6A expression groups (Tables S12 & S13). We then performed GSEA using GO and KEGG databases based on DEGs between these groups. In the training set, high ACTL6A expression was associated with enriched pathways related to mitochondrial and ribosomal function, while in the WCDT cohort, high ACTL6A expression was primarily linked to pathways involving chromatin and nucleosome organization (Figure S9A-B; Table S14).

Discussion

Androgen Receptor Signaling Inhibitor (ARSI) therapy remains the frontline treatment for advanced prostate cancer due to the cancer’s dependency on the AR signaling pathway ( [24]. While many patients initially respond well to ARSI, tumor heterogeneity and plasticity lead to varied treatment outcomes, complicating clinical decision-making. Additionally, patients who develop resistance to prolonged ARSI therapy often face limited or no alternative treatment options ( [25]. Thus, developing predictive models for ARSI response and identifying precise therapeutic targets are critical for improving the management of advanced prostate cancer.

Our study identified the GG-NER pathway as a significant regulator of ARSI response in prostate cancer through Reactome database analysis and ssGSEA. The GG-NER pathway, a crucial component of DNA damage repair, was found to be upregulated in patients with poor ARSI responses. This suggests that the GG-NER pathway might contribute to tumor progression and drug resistance, making it a potential target for therapeutic intervention. The GG-NER pathway is essential in maintaining genomic stability by repairing DNA damage. Its components, such as the ERCC1-XPF complex and RPA, play critical roles in excising damaged DNA and facilitating repair ( [26].

Abnormal activation of GG-NER has been observed in various tumors, particularly those resistant to chemotherapy. For instance, in cisplatin-resistant non-small cell lung cancer (NSCLC), genes like XPA and ERCC1 are often upregulated, and their knockdown enhances sensitivity to cisplatin ( [14, 27]. R-Y Liu et al. found that the expression of Eukaryotic initiation factor-3a (eIF3a) is significantly higher in the cisplatin-sensitive nasopharyngeal carcinoma cell line S16 than in the resistant line CNE2. Knocking down eIF3a expression upregulates NER pathway markers like XPA, XPC, thereby enhancing S16 cell resistance to cisplatin ( [28]. Furthermore, studies have found that using a dual-targeted siRNA liposome system (si-MK2 and si-XPA) can significantly enhance NSCLC sensitivity to cisplatin, achieving augmented synthetic lethality ( [29]. Similarly, our findings suggest that inhibiting the GG-NER pathway, specifically using HAMNO, can sensitize enzalutamide-resistant prostate cancer cells to treatment.

Further analysis revealed that high-risk prostate cancer patients with elevated NECGS scores showed rapid disease progression and poor ARSI response. In the study, it is not only observed the abnormal activation of the MYC and E2F pathways which are known to contribute to treatment resistance and cancer progression ((30–31) in the groups with high NECGS score, but it is also exhibited higher TP53 mutation frequencies tendency, and enhanced ribosomal.

Regarding the role of GG-NER in inducing genomic mutations or its relationship with tumors with high mutation burdens, we believe that the latter hypothesis may be more plausible. Our findings suggest that the high-risk group, characterized by a high frequency of TP53 mutations, exhibits greater genomic instability. TP53 mutations are commonly associated with increased genomic instability, as loss of p53 function disrupts DNA damage response and apoptosis pathways, leading to the accumulation of mutations and copy number variations that are less likely to be effectively repaired ( [32]. Consequently, tumors with high mutation burdens may adapt by enhancing GG-NER activity to maintain a minimal level of genomic stability, allowing them to survive despite the elevated mutational load ( [33]. GG-NER, as a key DNA repair pathway, could provide the necessary genomic maintenance that supports tumor cell survival and proliferation under conditions of high genomic damage ( [34]. Moreover, the interplay between DDR and ribosome biogenesis is also essential for cellular homeostasis and stress responses. DDR mechanisms, particularly pathways like GG-NER, help preserve rDNA integrity, which is crucial for ribosome production and protein synthesis. Disruption in rDNA due to inadequate DDR can impair ribosome biogenesis, affecting cellular growth and survival. Therefore, we assumed that GG-NER activity might be critical in protecting these processes under genotoxic stress, thus sustaining cellular resilience and functionality ( [35].

Our risk model, incorporating three GG-NER-related genes—RUVBL1, PARP2, and ACTL6A—highlighted ACTL6A as a key player in ARSI resistance. Although the roles of PARP2 and RUVBL1 in ARSI resistance have been previously documented, ACTL6A’s involvement is less understood ((22–23). ACTL6A, a component of the SWI/SNF chromatin remodeling complex, plays a crucial role in DNA damage repair and has been implicated in various cancers ( [36]. For instance, in hepatocellular carcinoma cells, ACTL6A can manipulate the expression of SRY (Sex Determining Region Y)-box 2 (SOX2) and activate Notch1 signaling, promoting cell migration and the EMT process ( [37]. In squamous cell carcinoma, high expression of ACTL6A promotes direct interaction between TEAD-YAP and the BAF complex, leading to redistribution of H3K27me3 histone modifications and the expression of related oncogenes ( [38]. Xiao et al. have also discovered that overexpression of ACTL6A alleviates DNA damage caused by cisplatin and mediates resistance to cisplatin treatment in lung and ovarian cancers. This resistance can be reversed by using the HDACi, a deacetylase inhibitor ( [39], further suggesting ACTL6A’s crucial regulatory role in tumor treatment resistance. In prostate cancer, preliminary research has found that ACTL6A regulates gene expression mediated by AR in LNCaP cells ( [40], which suggested ACTL6A may take part in ARSI response in PCa.

Our study demonstrates that ACTL6A knockdown restores sensitivity to enzalutamide in resistant prostate cancer cells, and functional enrichment results showed that high ACTL6A expression is linked to nucleosome-, ribosome-, and mitochondrion-related pathways. In our opinion, ACTL6A may influence both ribosome biogenesis and mitochondrial function, indirectly supporting cancer cell proliferation and survival. As a component of the BAF complex, ACTL6A promotes chromatin remodeling, which can enhance the expression of genes involved in ribosomal RNA (rRNA) synthesis and ribosome assembly, meeting the high protein synthesis demands of rapidly dividing cancer cells ( [41]. Additionally, ACTL6A-driven chromatin remodeling may upregulate genes linked to mitochondrial function, boosting ATP production and supporting the energy needs of tumor cells. Furthermore, ACTL6A may help balance reactive oxygen species (ROS) levels by regulating genes involved in ROS production and antioxidant responses, allowing cancer cells to avoid oxidative damage while sustaining metabolic activity under stress ( [42]. Thus, ACTL6A-driven nucleosome remodeling may create a chromatin environment that promotes oncogenesis by facilitating genomic stability and transcriptional activation of cancer-related pathways.

In conclusion, our findings underscore the importance of the GG-NER pathway in ARSI resistance and highlight ACTL6A as a promising target for overcoming treatment resistance in prostate cancer. Future studies should focus on further elucidating the mechanisms by which ACTL6A contributes to ARSI resistance and exploring its potential as a therapeutic target in clinical settings.