STARD4 suppresses tumorigenesis and attenuates enzalutamide resistance via lipid metabolic reprogramming and AR stabilization in prostate cancer

Human PCa cell lines (DU145, PC3, C4-2, and LNCaP) and HEK293T cells were obtained from American Type Culture Collection (ATCC; Manassas, USA) and cultured in accordance with ATCC guidelines. The 22Rv1 cell line was procured from ProCell (Wuhan, China) and maintained in RPMI-1640 medium (Boster, Wuhan, China). The immortalized human prostatic epithelial cell line RWPE-1 was also obtained from ProCell and cultured in K-SMF medium (Gibco, USA). To develop a C4-2-EnzR cell line, C4-2 cells were initially treated with 4 μM enzalutamide (MCE; Shanghai, China), and the concentration was progressively increased to 40 μM over a four-month period. Thereafter, C4-2-EnzR cells were propagated in RPMI-1640 medium supplemented with 20 μM enzalutamide (MCE; Shanghai, China). All the cell lines were subjected to STR profiling verification and cultured under standardized conditions (37 °C, 5% CO₂, and 95% humidity).

Tumour tissues were collected from 77 PCa patients who underwent radical prostatectomy at the First Affiliated Hospital of Shihezi University from September 2021 to May 2023. This research was authorized by the Ethics Committee of the First Affiliated Hospital of Shihezi University (KJ2024-042–01). Written informed consent was obtained from the patients for the use of tissue samples in scientific research.

Short hairpin RNA and nontarget shRNAs were designed and synthesized by Genomeditech (Shanghai, China): sh-STARD4 #1: 5′-TCCTATACTGTGGGCTATAAATTCAAGAGAGATTTATAGCCCACAGTATAGGATTTTT-3′; sh-STARD4 #2: 5′-ACAAAGCCCAAGGTGTTATAGTTCAAGAGACTATAACACCTTGGGCTTTGTTTTTTT-3′; and sh-STARD4 #3: 5′-CCGGTCCTATACTGTGGGCTATAAACTCGAGTTTATAGCCCCACAGTATAGGATTTTTG-3′. Flag-tagged AR (Gene ID: 367; vector: pGMLV), Flag-tagged UBE4B (Gene ID: 10,277; vector: pGMLV), Flag-tagged FANCG (Gene ID: 2189; vector: pGMLV), and STARD4 overexpression plasmids were also sourced from Genomeditech. His-tagged STARD4 (Gene ID: 134,429; vector: pGMLV) and GST-tagged AR (Gene ID: 367; vector: pGEX-4 T-1) were procured from Tsingke Biotechnology (Beijing, China). Additionally, HA-tagged STARD4 (Gene ID: 134,429; vector: pCMV), and AR truncation domain (Flag-NTD, Flag-NTD-DBD, Flag-DBD, Flag-DBD-LBD, and Flag-LBD) plasmids were designed and synthesized by Miaoling Biotechnology (Wuhan, China). Transfection efficiency was rigorously assessed via RT‒qPCR and western blotting.

For proliferation assays, cells were seeded in 96-well plates at an initial density of 3 × 10³ cells/well. Viability was quantified spectrophotometrically (OD450 nm) after incubating cells with CCK-8 reagent (Yeasen Biotechnology) per the manufacturer’s guidelines. EdU incorporation assays (RiboBio) were performed by incubating cells (5 × 10⁴/well in 12-well plates) with 50 μM EdU for 2 h, followed by fixation (4% PFA, 20 min), permeabilization (0.5% Triton X-100), and dual staining with Hoechst and Apollo. EdU⁺ cells were enumerated across five microscopic fields (Olympus IX71; 200 × magnification). Clonogenic potential was assessed after fixing (methanol) and staining (crystal violet; 0.1%, 30 min) 14-day cultures in 6-well plates, with colonies counted using ImageJ software.

The migration potential of cells was evaluated using a transwell migration assay. In brief, cells (3 × 10⁴) were seeded in the upper chamber of transwell inserts in serum-free medium; the upper chambers were placed in each well of a 24-well culture plate and immersed in 500 μl of medium supplemented with 10% FBS. After incubation for 24 h, the migrated cells located on the lower side of the membrane were immobilized with a 4% paraformaldehyde solution. After staining the cells with 0.1% crystal violet solution, cells in five randomly selected fields were subsequently enumerated using a microscope.

Flow cytometry was performed to determine the percentage of apoptotic cells. Following plasmid transfection or drug treatment, the cells were harvested and incubated with PI and Annexin V (Yeasen Biotechnology, Shanghai, China) in the dark at room temperature for 15 min. Flow cytometry analysis was performed using a CytoFlex cytometer (Beckman Coulter, USA), and the data were analysed using FlowJo V10 software.

Total RNA was isolated using TRIzol™ (Servicebio Technology) and quantified spectrophotometrically. Reverse transcription was performed with the Hifair II cDNA Synthesis System (Yeasen Biotechnology). Quantitative PCR amplification was executed on a StepOnePlus™ Real-Time PCR System (Applied Biosystems) with SYBR Green chemistry. The primer sequences are detailed in Supplementary Table 1. To elucidate the transcriptomic alterations in response to STARD4 overexpression, sequencing libraries were prepared from total RNA isolated from both the control and STARD4-overexpressing groups. These libraries were then subjected to next-generation sequencing to delineate the transcriptomic landscape.

Western blotting was carried out using standard protocols as previously described [26]. The specifications of all primary and secondary antibodies are listed in Supplementary Table 2. For Co-IP, the procedure outlined by the manufacturer (Biolinkedin, Shanghai, China) was followed. Briefly, treated cells were collected and lysed on ice for 30 min using a lysis buffer containing 1% protease and phosphatase inhibitors (Boster, Wuhan, China). The resulting lysates were subjected to overnight incubation at 4 °C with rotation in the presence of target-specific antibodies to allow immune complex formation. Subsequently, prechilled magnetic beads were introduced and incubated with the complexes at room temperature for 2 h. Following three washes with ice-cold lysis buffer, the beads were pelleted and resuspended in SDS‒PAGE loading buffer. Finally, the immunoprecipitated samples were denatured by boiling at 100 °C for 5 min prior to further analysis.

Recombinant GST-AR and His-STARD4 proteins were expressed in BL21 (Rosetta) bacterial cells and purified using the standard protocol provided with a protein purification kit (Biolinkedin, Shanghai, China). A GST pull-down assay was performed using a GST protein interaction pull-down kit (Biolinkedin, Shanghai, China) in accordance with the manufacturer's instructions. Briefly, purified GST-tagged proteins were immobilized onto glutathione-Sepharose beads, which were then incubated with His-tagged proteins in PMSF-supplemented affinity isolation buffer. After incubation, the protein-bound beads were magnetically separated, washed, and the captured complexes were analyzed by western blotting.

Experimentally treated cells were seeded onto glass coverslips in 12-well culture plates at a density of 1.5 × 10⁵ cells per well. The cells were rinsed with PBS and subsequently fixed with 4% paraformaldehyde for 30 min at room temperature. After fixation, the cells were permeabilized with 0.5% Triton X-100 for 15 min. The cells were then rinsed with PBS for 15 min and blocked with goat serum (ZSGB-Bio, Beijing, China) at room temperature. Cells were then incubated with primary antibodies overnight at 4 °C, followed by PBS rinsing and incubation with fluorophore-conjugated secondary antibodies. Finally, samples were washed twice with PBS, counterstained with DAPI for nuclear visualization, and imaged using a Zeiss LSM880 confocal microscope. Antibody details are provided in Supplementary Table 2.

Stably transfected cells were seeded in Petri dishes for confocal microscopy and rinsed twice with a balanced salt solution enriched with Ca²⁺ and Mg²⁺. Next, the cells were incubated with prewarmed ER-Tracker Red Kit working solution (Beyotime) for 15 min, followed by two washes with RPMI 1640 medium to eliminate surplus dye. Then, the cells were fixed with a 4% paraformaldehyde solution for 5 min and counterstained with DAPI to visualize the nuclear architecture. Images were acquired through a confocal microscope.

ORO staining was conducted using both cells and tissue sections and an ORO stain kit (Servicebio, Wuhan, China), adhering meticulously to the manufacturer’s guidelines. The ORO working solution was prepared by combining six parts of ORO stock solution with four parts of deionized water and subsequently purifying the mixture through a 0.22 µm syringe filter. Cells in 6-well plates were fixed with 4% paraformaldehyde at ambient temperature for 10 min and subsequently rinsed with 60% isopropanol before being stained with ORO solution in a light-protected environment for 30 min at room temperature. Prior to microscopic examination, the cells were differentiated in 60% isopropanol for 5 s and subsequently washed with deionized water. Tissues were embedded in optimal cutting temperature compound, sectioned and fixed in 4% paraformaldehyde for 10 min. The sections were incubated in ORO staining solution under light-protected conditions for 10 min at room temperature, followed by counterstaining with haematoxylin to delineate the nuclear structures. Observations and photographic documentation were carried out using a microscope. ImageJ software was used to measure the diameter of lipid droplets within cells. Subsequently, ORO was redissolved in isopropanol, and the absorbance was measured at 570 nm.

Ntracellular triglyceride (TG) and total cholesterol levels were quantified using commercial assay kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), respectively. Cells were lysed in normal saline via ultrasonic disruption. For the assays, 250 μL of each lysate was mixed with 2.5 μL of the kit-specific working solution in a 96-well plate, followed by incubation at 37 °C for 10 min in the dark to avoid photodegradation. Optical density was measured at 510 nm using an ELISA reader. Additionally, testosterone concentrations were determined using an ELISA kit (Beyotime Biotechnology) following the manufacturer’s protocol.

Gene expression data and corresponding clinical information pertaining to a PCa cohort were procured from The Cancer Genome Atlas (TCGA) database and the Gene Expression Omnibus (GEO) database. The differential expression of STARD4 between malignant tissues and adjacent nontumorous tissues was assessed utilizing the Wilcoxon test. Kaplan‒Meier survival analysis was used to elucidate the correlation between STARD4 expression levels and patient survival outcomes. Furthermore, datasets from DKFZ and MSKCC were utilized to investigate the associations between STARD4 expression and both pathological T stage and Gleason score in PCa. To reveal the biological functions of STARD4, gene set enrichment analysis (GSEA) was conducted to identify the pathways that were significantly affected. Additionally, the UbiBrowser database was used to predict potential E3 ubiquitin ligases that interact with the AR protein.

For untargeted lipidomics analysis, control and STARD4-overexpressing 22Rv1 cells were snap-frozen in liquid nitrogen immediately after collection. Lipids were extracted using a dichloromethane/methanol mixture (1:1, v/v), followed by centrifugation (16,000 × g, 15 min, 4 °C) to pellet insoluble debris. The supernatants were subjected to chromatographic separation using a Vanquish HPLC system (Thermo Fisher Scientific) with a Phenomenex Kinetex C18 column (2.1 mm × 100 mm, 2.6 μm). Mass spectrometry data were acquired using an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific). Data analysis was performed using the CentWave algorithm in XCMS, and lipid identification was conducted by searching against the LipidBlast 2022 database.

All animal experiments were approved by the First Affiliated Hospital of Shihezi University Animal Care Committee (Document No. A2024-038–01). Five-week-old male athymic nude mice were procured from SIPEIFU Biotechnology (Beijing, China). To explore the influence of STARD4 on PCa growth, a suspension of 3 × 10⁶ 22Rv1 cells with STARD4 overexpression or depletion was subcutaneously injected into mice (n = 5). Tumour progression was monitored every four days until the mice were sacrificed after 32 days, after which the tumours were excised for subsequent immunohistochemical staining. Moreover, to assess the effect of STARD4 on enzalutamide resistance in PCa, 2 × 10⁶ C4-2-EnzR cells transfected with either an empty vector or a STARD4-oe vector were implanted subcutaneously into mice. Four days postinoculation, the mice were segregated into four groups (n = 5): vector, vector + enzalutamide (10 mg/kg, p.o.), STARD4-oe, and STARD4-oe + enzalutamide (10 mg/kg, p.o.). Following a treatment period of 28 days, the mice were humanely euthanized, and the tumours were collected for subsequent analysis.

All the statistical analyses were conducted using R version 4.3.0 and GraphPad Prism version 9. For comparisons between two groups, normally distributed data were analysed for statistical significance using Student’s t test, whereas nonnormally distributed data were assessed with the Mann–Whitney U test. One-way ANOVA and two-way ANOVA followed by Tukey’s multiple comparison test was used to compare multiple groups. Spearman correlation was used to conduct all correlation analyses. Univariate Cox regression analysis was employed to examine the relationships between the expression levels of candidate genes and patient survival outcomes. Survival analysis involved the comparison of Kaplan‒Meier curves utilizing the log-rank test. All assays were conducted at least three times,and data are presented as means ± standard deviations. Statistical significance was defined as a p value less than 0.05.

To identify lipid metabolism genes that play significant roles in PCa, we analysed differentially expressed genes related to lipid metabolism between PCa and normal tissues using four transcriptomic datasets (TCGA-PRAD, GSE62872↗, GSE46602↗, and GSE69223↗). This process led to the identification of 34 genes of interest (Fig. 1A, S1A). Further screening with univariate Cox regression identified 12 genes that were strongly associated with PCa prognosis (Fig. 1B). Notably, STARD4 expression was found to be significantly downregulated in PCa, and high STARD4 expression was linked to a reduced risk of PCa (Fig. 1B, S1B). Kaplan–Meier survival analysis revealed that patients with low STARD4 expression had significantly shorter progression-free survival than those with high STARD4 expression did (Fig. 1C). Additionally, low STARD4 expression was correlated with tumour progression, metastasis, and high Gleason scores (Fig. 1D).

To validate these findings, we assessed STARD4 expression levels in several PCa cell lines. Our observations indicated that STARD4 levels were significantly lower in PCa cells than in the human prostatic epithelial cell line RWPE-1 (Fig. 1E). This observation was corroborated by analysis of publicly available data from the HPA database, which revealed significantly higher STARD4 expression in benign prostate tissues relative to PCa (Figure S1C). Furthermore, IHC staining of the PCa specimens demonstrated that STARD4 protein levels were substantially lower in tumors with high Gleason scores (n = 41) than in those with low Gleason scores (n = 36) (Fig. 1F). Collectively, these results suggest that low STARD4 expression is associated with PCa progression and poor patient survival outcomes.

To elucidate the biological function of STARD4 in PCa, we overexpressed STARD4 in the 22Rv1 and C4-2 cell lines while employing shRNA to achieve stable STARD4 depletion in LNCaP, 22Rv1, and C4-2 cells (Figure S2A). The impact of STARD4 on cell proliferation was investigated through CCK-8 and EdU assays. Our results demonstrated that STARD4 overexpression markedly suppressed cell proliferation, whereas STARD4 silencing enhanced this process (Fig. 2A-B, S2B-C). Furthermore, colony formation assays confirmed that STARD4 overexpression significantly diminished the proliferative ability of PCa cells (Fig. 2C, S2D).

Moreover, transwell assays revealed that the increased expression of STARD4 inhibited the migration of 22Rv1 and C4-2 cells, whereas the knockdown of STARD4 in LNCaP cells promoted their migratory ability (Fig. 2D). Subsequent immunoblot analysis of epithelial‒mesenchymal transition (EMT)-related markers revealed that, compared with control cells, cells overexpressing STARD4 presented reduced levels of N-cadherin and vimentin and an increased level of ZO-1 (Fig. 2E). To validate these findings in vivo, xenograft models were established. STARD4 overexpression in PCa models significantly diminished the tumour growth rate, size, and weight (Fig. 2F, S2E). Consistent with these observations, IHC staining for Ki-67 confirmed the impact of STARD4 on proliferation in PCa xenografts (Fig. 2G, S2F). In summary, both the in vitro and in vivo results strongly demonstrated that STARD4 markedly suppresses PCa cell proliferation and restrains xenograft tumour progression.

To investigate the role of STARD4 in lipid metabolism in PCa, the TCGA-PRAD cohort was utilized to perform GSEA. The findings revealed a significant association between STARD4 expression and lipid metabolic processes in PCa (Fig. 3A, S3A). To validate this observation, ORO staining was employed to assess lipid accumulation within PCa cells. As depicted in Fig. 3B, compared with control cells, PCa cells overexpressing STARD4 presented a marked reduction in lipid droplet (LD) density, whereas a significant increase in LD density was noted in STARD4-knockdown cells. These results support the hypothesis that STARD4 plays a crucial role in modulating lipid metabolism in PCa.

To elucidate the impact of STARD4 on lipid accumulation in vivo, Matrigel mixed with PCa 22Rv1 or C4-2 cells was implanted subcutaneously into mice. Consistent with the in vitro findings, xenograft tumours with upregulated STARD4 expression presented reduced lipid accumulation within the cytoplasm, whereas STARD4 silencing led to a significant increase in lipid accumulation (Fig. 3C, S3B). To comprehensively characterize these lipid alterations, we performed untargeted LC–MS lipidomic profiling of STARD4-overexpressing versus control 22Rv1 cells. This analysis identified substantial decreases in multiple lipid classes, including cholesteryl esters (CE), triglycerides (TG), hexosylceramides (HexCer), and phosphatidylcholines (PC) (Figure S3C). Pathway analysis of these differentially regulated lipid species revealed significant enrichment of key metabolic processes, most notably fatty acid transport, triglyceride homeostasis, and cholesterol metabolism (Figure S3D), providing mechanistic insight into STARD4's role in lipid regulation. Based on this, TG and cholesterol contents were measured to quantify lipid accumulation in PCa cells. Consistent with the lipidomics data, stable overexpression of STARD4 resulted in lower TG and cholesterol levels (Fig. 3D-E). Collectively, these findings suggest that the upregulation of STARD4 expression mitigates lipid accumulation in PCa.

Previous studies have shown that LDs alleviate the ER stress response, thereby promoting ER homeostasis and increasing the viability of tumour cells. Furthermore, our GSEA results revealed a significant association between high STARD4 expression and the signalling pathways associated with the cargo concentration within the ER (Fig. 4A). Given that STARD4 inhibits abnormal lipid accumulation in PCa, we investigated whether it also impacts ER stress in PCa cells. ER tracking imaging analysis revealed ER dilation in cells overexpressing STARD4 (Fig. 4B), a finding corroborated by TEM, which demonstrated the presence of abnormally dilated and irregularly shaped rough ER in cells with upregulated STARD4 expression (Fig. 4C). Additionally, the overexpression of STARD4 in 22Rv1 and C4-2 cells was associated with increased levels of ER stress markers; in contrast, STARD4 knockdown in these cells led to the decreased expression of these markers (Fig. 4D, S4A-B). These observations indicate that STARD4 plays a role in modulating ER stress in PCa cells. Given that ER stress frequently triggers cell death through the activation of apoptotic signalling pathways [27, 28], flow cytometry was used to assess the degree of apoptosis. Consistent with the functional predictions, STARD4 overexpression in the 22Rv1 and C4-2 cell lines significantly increased the apoptotic indices, whereas STARD4 silencing conferred resistance to apoptosis (Fig. 4E). Immunoblotting revealed that the STARD4-mediated downregulation of Bcl-2 (antiapoptotic) expression was concurrent with upregulated Bax expression and increased caspase-3 cleavage (Fig. 4F), mechanistically validating the ability of STARD4 to induce apoptosis in PCa.

To elucidate the regulatory mechanisms of STARD4 in PCa, we performed RNA-seq analysis using 22Rv1 cells stably overexpressing STARD4. Gene ontology analysis of the differentially expressed genes revealed that STARD4 overexpression significantly affected the AR signalling pathway, regulation of the AR signalling pathway, intracellular sterol transport, and intracellular cholesterol transport (Fig. 5A). Considering that increased intratumoural androgen synthesis is a critical factor in PCa progression and that cholesterol serves as a vital precursor for androgen synthesis, we hypothesized that STARD4 influences androgen levels within tumour cells. Consequently, we assessed androgen synthesis in these cells. The results indicated that the upregulation of STARD4 expression significantly reduced testosterone levels, whereas the downregulation of STARD4 expression led to a significant increase in testosterone content. These findings suggest that STARD4 inhibits intratumoural androgen synthesis (Fig. 5B), thereby playing a crucial role in modulating hormone-dependent pathways in PCa.

Notably, androgens exert their physiological effects by binding to and activating the AR, with AR expression being a critical oncogenic driver at various stages of PCa development and progression [29]. To better understand the relationship between STARD4 and AR, we examined both the mRNA and protein levels of AR in PCa cells overexpressing STARD4 or with STARD4 depletion. RT‒qPCR analysis revealed no significant changes in AR mRNA levels across these conditions (Fig. 5C). However, western blot analysis revealed that STARD4 overexpression led to decreased AR protein levels, whereas STARD4 knockdown resulted in increased AR protein levels compared with those in the controls (Fig. 5D, S5A). This discrepancy between the AR mRNA and protein levels suggested that STARD4 might regulate AR levels through mechanisms affecting protein stability rather than transcription. To test this hypothesis, we investigated whether STARD4 interacts with AR. Co-IP experiments confirmed an interaction between STARD4 and AR (Fig. 5E). Additional evidence for a direct interaction was provided by GST pull-down assays, which demonstrated that AR directly binds to the STARD4 protein (Fig. 5F). IHC staining supported these findings by revealing the colocalization of STARD4 and AR within the cytoplasm (Fig. 5G). The AR protein comprises three structural modules: an N-terminal transactivation domain (NTD), a central DNA-binding domain (DBD), and a C-terminal ligand-binding domain (LBD) [30]. To map STARD4-AR interaction interfaces, HEK293T cells were transfected with AR domain-specific deletion constructs. Co-IP assays revealed that STARD4 binds to the AR-NTD and full-length AR but not to the DBD or LBD truncations (Fig. 5H), establishing the NTD as the critical interaction interface.

To validate the hypothesis that STARD4 regulates AR protein stability, cell lines in different groups were treated with cycloheximide (CHX), an inhibitor of protein synthesis, to assess AR protein turnover over time. The results revealed that the half-life of AR was significantly shortened after STARD4 overexpression (Fig. 6A). Conversely, STARD4 knockdown prolonged the half-life of AR (Figure S5B). To explore the mechanism by which STARD4 regulates AR protein expression, PCa cells with the stable overexpression or knockdown of STARD4 were treated with a proteasome inhibitor (MG132) and a lysosome inhibitor (chloroquine). Western blot analysis revealed that only MG132 reversed the reduction in AR protein levels caused by STARD4 overexpression in PCa cells (Fig. 6B-C, Figure S5C). These findings indicate that STARD4 affects AR protein stability through the ubiquitin‒proteasomal degradation pathway rather than via lysosomal degradation. To test this hypothesis, immunoprecipitation was subsequently performed to measure the ubiquitination level of AR. The results revealed a significant increase in AR ubiquitination in PCa cells overexpressing STARD4 (Fig. 6D). These findings suggest that STARD4 enhances the ubiquitination and subsequent proteasomal degradation of AR, thereby regulating its protein stability.

To elucidate the mechanism by which STARD4 regulates AR protein ubiquitination, computational tools such as STRING and Ubibrowser were used to predict potential ubiquitinases that could bind to AR. This screening process identified two candidate ubiquitinase genes: UBE4B and FANCG (Fig. 7A). Subsequent Co-IP experiments confirmed that only UBE4B interacts with both STARD4 and AR (Fig. 7B), suggesting a specific role for UBE4B in this context. UBE4B is an E3 ubiquitin ligase known for its involvement in the ubiquitination and degradation of protein substrates [31], playing a significant role in tumorigenesis and progression across various types of cancer [32–35]. Given this background, subsequent experiments were conducted to determine whether UBE4B acts as the key E3 ubiquitin ligase that mediates STARD4-induced AR degradation. Surprisingly, altering STARD4 expression did not affect the mRNA or protein levels of UBE4B (Fig. 7C, S6A). However, Co-IP assays revealed that the interaction between AR and UBE4B was enhanced by STARD4 overexpression in PCa cells but was reduced when STARD4 was depleted (Fig. 7D, S6B). These findings indicate that STARD4 does not affect the expression level of UBE4B but instead enhances the interaction between UBE4B and AR, facilitating the ubiquitination and subsequent degradation of AR. These findings highlight the critical role of STARD4 in modulating AR stability through its interactions with UBE4B rather than by directly regulating UBE4B expression levels. Given that UBE4B ubiquitination activities are dependent on a highly conserved proline at position 1140 and that the mutant form of UBE4B (1140A) cannot increase the ubiquitination of target proteins [36], experiments were conducted to assess the importance of UBE4B ubiquitination activity for STARD4-mediated AR protein stability. The results showed that wild-type UBE4B, not the UBE4B (1140A) mutant, restored AR protein degradation induced by STARD4 overexpression and reversed the upregulated AR expression caused by STARD4 knockdown (Fig. 7E). Consistently, the upregulation of STARD4 expression increased AR ubiquitination, and this effect was compromised when UBE4B was inhibited or mutated, indicating that the ubiquitination activity of UBE4B is essential for the STARD4-mediated regulation of AR stability (Fig. 7F). These findings underscore the indispensable role of UBE4B in mediating the effects of STARD4 on AR, particularly highlighting the necessity of the enzymatic activity of UBE4B for this regulatory process.

Given the critical role of the AR signalling pathway in the effectiveness of antiandrogen therapy, this investigation was extended to explore the role of STARD4 in modulating enzalutamide resistance in PCa. An enzalutamide-resistant C4-2 cell model was established. Compared with their enzalutamide-sensitive counterparts, C4-2-EnzR cells presented reduced sensitivity to enzalutamide (Figure S7A). Western blot analysis revealed that STARD4 expression was decreased while AR expression was increased in C4-2-EnzR cells (Fig. 8A). Additionally, ORO staining revealed an increase in LD size in C4-2-EnzR cells relative to that in C4-2 cells (Figure S7B). Subsequent experiments were conducted to assess the impact of STARD4 on cell proliferation and the response to enzalutamide treatment. CCK-8 and EdU assay results demonstrated that PCa cells with STARD4 depletion were resistant to enzalutamide treatment, whereas increased STARD4 expression enhanced the suppressive effects of enzalutamide on the proliferation of both 22Rv1 and C4-2-EnzR cells (Fig. 8B-C). Flow cytometry analysis was used to evaluate cell apoptosis following enzalutamide treatment. STARD4 overexpression restored the sensitivity of 22Rv1 and C4-2-EnzR cells to enzalutamide-induced apoptosis (Figs. 8D and S7C). Western blotting confirmed these findings, showing that STARD4 overexpression promoted apoptosis in PCa cells treated with enzalutamide (Fig. 8E).

To assess the therapeutic efficacy of STARD4 in a setting that closely mimics the biological environment of PCa more accurately, an in vivo study was conducted using a mouse model. C4-2-EnzR cells, either with a control vector or stably overexpressing STARD4, were transplanted subcutaneously into mice. The results showed that enzalutamide treatment alone had minimal effects on tumour growth. In contrast, STARD4 overexpression led to the significant inhibition of tumour growth when it was combined with enzalutamide treatment (Fig. 8F-G, S7D-E). Additionally, ORO staining revealed that, compared with the other treatments, the combination of STARD4 overexpression and enzalutamide treatment resulted in a significant reduction in LD size within tumour tissues (Fig. 8H, S7F). Collectively, these findings indicate that STARD4 has the potential to serve as a therapeutic target for treating PCa, particularly in overcoming acquired resistance to enzalutamide.

The study was approved by the First Affiliated Hospital of Shihezi University (KJ2024-042–01). All animal experiments compliance with regulations of Animal Ethics Committees in the First Affiliated Hospital of Shihezi University (Document No. A2024-038–01).

The authors declare no competing interests.

All the data can be obtained by contacting the corresponding author.