The long noncoding RNA lncZBTB10 facilitates AR function via S-palmitoylation to promote prostate cancer progression and abiraterone resistance

Abstract

Background

Activation of androgen receptor (AR) by androgen binding to its ligand-binding domain (LBD) has led to the development of clinical drugs that target androgen biosynthesis or the LBD of AR for the treatment of prostate cancer patients. While these drugs initially offer clinical benefits, the emergence of drug resistance is inevitable after a certain duration of treatment.

Objectives

Exploring alternative AR domains or identifying novel mechanisms for AR activation is crucial for advancing prostate cancer therapies.

Methods

A systematic bioinformatic analysis identified novel androgen-responsive long noncoding RNAs (lncRNAs) in prostate cancer, which were verified using loss-of-function and gain-of-function strategies in vitro and in vivo.

Results

lncZBTB10 or LINC02986 was overexpressed in prostate cancer specimens and correlated with poor clinical outcomes. Mechanistically, our findings elucidate the pivotal role of lncZBTB10 in facilitating AR function by inducing S-palmitoylation. Moreover, the interaction between lncZBTB10 and AR not only fosters but also orchestrates biomolecular condensates within the nucleus driven by a novel RNA-binding domain, particularly in prostate cancer cells. Notably, the overexpression of lncZBTB10 not only promotes tumor growth in vivo but also triggers abiraterone resistance in vitro by inducing AR expression.

Conclusions

These results collectively reveal a novel mechanism by which lncZBTB10 regulates AR function in prostate cancer cells.

Background

Androgen receptor (AR) has been widely recognized to play a pivotal role in the development of prostate cancer in recent decades. AR is composed of an N-terminal transactivation domain (NTD, also called AF1), a DNA-binding domain (DBD) and a ligand-binding domain (LBD) [1]. In general, AR is translocated from the cytosol to the nucleus to regulate its downstream targets when androgen binds to its LBD. Thus, the primary treatment approach for individuals with locally advanced prostate cancer, metastatic prostate cancer, and biochemically recurrent disease following the ineffectiveness of localized treatments is androgen-deprivation therapy (ADT) through androgen reduction or AR inhibition [2]. Although initially effective against advanced prostate cancer (PCa), the disease often progresses to a lethal, therapy-resistant state known as castration-resistant PCa (CRPC), where AR continues to exert dominance. Previously, several second-generation antiandrogen drugs, such as abiraterone, an androgen synthesis inhibitor, and enzalutamide, an AR antagonist, were approved by the US FDA as novel clinical treatment options for CRPC patients [3, 4]. However, despite their initial efficacy against CRPC, resistance commonly develops in most patients [5]. Mechanistically, the emergence of the F876L mutation in the LBD of AR and the generation of AR-V7 variants lacking the LBD are potential mechanisms contributing to drug resistance [6, 7]. Hence, developing innovative AR-targeting strategies that focus on alternative domains or underlying mechanisms preventing AR activation is imperative for advancing prostate cancer treatment.

Recent research has indicated that transcription factors, along with their disordered transactivation domains, undergo phase separation to form membrane-less molecular condensates or nuclear puncta [8, 9]. These biomolecular condensates are associated with the formation of transcriptionally active complexes at superenhancers, interactions with cofactors such as the mediator complex, and the modulation of oncogenic transcription [8, 9]. Intriguingly, the AR NTD is also an intrinsic disordered region (IDR) that is subdivided into the tau-1 (amino acids 100–359) and tau-5 regions (amino acids 360–528) [10, 11]. Indeed, recent studies have revealed phase separation behavior in both AR and the AR-V7 variant, which is positively correlated with their nuclear translocation and transactivation upon androgen stimulation [12–14]. Concurrently, an additional study revealed that AR exhibits three distinct types of movements: an immobile state, potentially associated with DNA binding; an intermediate confined state, likely facilitated through phase separation; and a freely mobile state upon androgen stimulation, as demonstrated by single-particle tracking analysis [15]. However, the identification of the potential factor that triggers the gradual entrapment of a subset of AR molecules at random upon collision with the local environment of a nuclear focus is essential.

Recently, increasing evidence has highlighted the pivotal roles of various long noncoding RNAs (lncRNAs), such as HOTAIR [16], LINC00844 [17], ARLNC1 [18], lnc-LBCS [19], NXTAR [20], PRNCR1 and PCGEM1 [21], in tumor progression [16–19, 21] and the emergence of drug resistance in prostate cancer [20]. Furthermore, most of these studies focused on regulating and fine-tuning androgen receptor (AR) expression or function through various mechanisms. These mechanisms include preventing AR degradation by obstructing its binding to the E3 ubiquitin ligase MDM2 [16], activating AR to increase its transcription through a direct interaction [17], ensuring AR transcript stability via RNA–RNA interactions [18], increasing AR expression by initiating its protein translation [19], suppressing AR transcription through epigenetic silencing [20], and fostering AR function through posttranslational modifications [21]. Nevertheless, the regulation of AR nuclear puncta formation by lncRNAs has not yet been reported in the literature. In the present study, a systematic approach revealed the overexpression of lncZBTB10 in prostate cancer specimens. Importantly, androgen stimulation was found to be pivotal for increasing lncZBTB10 expression, which in turn facilitated the formation of AR puncta in the nucleus and augmented AR function through S-palmitoylation in prostate cancer cells. Notably, the significance of lncZBTB10 was observed in both the development of prostate cancer and the emergence of resistance to abiraterone.

Methods

Cell culture and the establishment of drug-resistant clones

RWPE1, LNCaP, 22Rv1, PC3 and 293T cells were purchased from ATCC and maintained according to the ATCC guidelines. The detailed cell culture information is described in the Supplementary Materials and Methods.

RNA immunoprecipitation (RIP)

LNCaP cells treated with DHT (10 nM) for 24 h were exposed to UV light (254 nm) to crosslink the AR protein and interacting RNAs. The cell pellets were then collected and lysed in RIPA buffer supplemented with RNasin® ribonuclease inhibitor (Promega, Madison, Wisconsin, United States). The samples containing 500 μg of protein were subsequently individually incubated overnight at 4 °C with either the control or AR antibody (10 μg). Following this incubation, the RIP samples were washed with DEPC-treated PBS five times (5 min/wash) after a two-hour incubation with Dynabeads magnetic beads (Thermo Fisher Scientific, Waltham, MA, USA). Finally, TRIzol reagent was added to the RIP samples containing magnetic beads for RNA extraction. The expression of lncZBTB10 in the RIP samples was measured by qRT‒PCR.

Palmitoylated protein extraction and subcellular fractionation

The detailed extraction procedures are described in the Supplementary Materials and Methods.

Orthotopic mouse model of prostate cancer

The detailed procedures used for the animal experiments are described in the Supplementary Materials and Methods.

Next-generation sequencing analysis

A total of 2 μg of RNA extracted from lncZBTB10-knockdown LNCaP cells treated with DHT (10 nM) for 24 h was subjected to an RNA-seq analysis conducted by AllBio Science, Inc. (Taichung, Taiwan). The original raw data, along with the processed results, have been submitted to the Sequence Read Archive (SRA) database under accession number PRJNA1072257.

Bioinformatics analysis

The expression levels and survival outcomes of lncZBTB10 in clinical patients with prostate cancer from TCGA-PRAD dataset were analyzed using the UCSC Xena platform, and the Gene Expression Omnibus (GEO) dataset (GSE134051) was analyzed with the GEO2R platform. Moreover, potentially clinically relevant lncRNAs regulated by androgens were identified and cross-validated using the results from the GSE110903 (p < 0.05) and TCGA-PRAD (p < 0.001) datasets. Next, both AR and lncZBTB10 expression levels in different prostate cancer cell lines from the Cancer Cell Line Encyclopedia (CCLE) database were analyzed using the UCSC Xena platform. Finally, the gene expression profile of lncZBTB10-knockdown cells (p < 0.05) was used to perform gene set enrichment analysis (GSEA).

Statistical analysis

Statistical analyses were executed using GraphPad Prism 9 software (GraphPad Software, Inc.). Two-tailed t tests were employed to compare treatment groups and control groups, whereas ANOVA models were utilized for comparisons among multiple experimental groups. The Kaplan‒Meier method was used to analyze the survival of clinical patients, and the log-rank test was used to compare survival distributions between different groups. The results are presented as the means ± SEMs, and statistical significance was established at a p value < 0.05 for all tests.

Results

lncZBTB10, a potential androgen-responsive lncRNA, is highly overexpressed in prostate cancer tissues and associated with cancer malignancy

The gene symbols of all the lncRNAs from the NONOCDE database (v5.0) were downloaded and further individually analyzed with TCGA prostate cancer dataset containing normal and cancer tissues to systematically identify potential lncRNAs with clinical relevance that are downstream of AR (Supplementary Table S3). Next, lncRNAs with significant differences were cross-referenced with the profiles of upregulated lncRNAs in both androgen-treated LNCaP and VCaP cells from another Gene Expression Omnibus (GEO) dataset to identify androgen-regulated lncRNAs. Finally, twenty-one lncRNAs that were overexpressed in prostate cancer specimens and induced by androgen treatment in prostate cancer cells were identified via bioinformatics analyses (Fig. 1a and Supplementary Table S3). Among them, lncZBTB10 (also called LINC02986, AC009812.4 or RP11-48B3.4) was identified because it has never been previously characterized; its expression was indeed elevated in prostate cancer samples from both TCGA and GEO datasets and induced by androgen treatment (Fig. 1b and Supplementary Fig. S1A, B). Furthermore, a higher expression level of lncZBTB10 was positively correlated with not only poor survival rates, including progression-free survival (PFS) and disease-free survival (DFS) (Fig. 1c and Supplementary Fig. S1C), but also metastasis (Fig. 1d) and biomedical recurrence (Fig. 1e), suggesting its critical roles in prostate cancer. Finally, a commercial cDNA tissue array containing normal and prostate cancer samples was used to verify the bioinformatic findings. The results revealed that lncZBTB10 expression was increased in prostate cancer samples compared with their normal counterparts and was consistently elevated across various cancer staging systems (Fig. 1f and Supplementary Fig. S1D–F). However, its expression was not correlated with cancer stages (Supplementary Fig. S1G, H). Taken together, these findings suggest that lncZBTB10 expression is potentially regulated by androgen and is associated with prostate cancer progression and cancer malignancy.

Fig. 1. A newly identified androgen-regulated long noncoding RNA, termed lncZBTB10, is overexpressed in prostate cancer and has a positive correlation with unfavorable clinical outcomes.

a, b Heatmap showing the expression levels of potential androgen-regulated long noncoding RNAs (lncRNAs) (a), and lncZBTB10 expression was particularly apparent in prostate cancer (PRAD) samples in TCGA dataset (normal, n = 152; PRAD, n = 496) (b) *p < 0.05, two-tailed unpaired t-test. c–e Several clinical parameters, including the progression-free survival probability (median value of NUDT16L1 for determining high and low expression group) (c), metastasis (no, n = 374; yes, n = 81) *p < 0.05, log-rank (Mantel–Cox) test (d), and biochemical recurrence (no, n = 412; yes, n = 61) (e). *p < 0.05, two-tailed unpaired t-test., were also analyzed according to the lncZBTB10 expression levels. f lncZBTB10 expression was determined in a commercial cDNA array of PCa samples (normal, n = 15; PCa, n = 77). *p < 0.05, two-tailed unpaired t-test.

lncZBTB10 is directly regulated by AR, and their interaction maintains the proper function of AR in prostate cancer cells treated with androgen

Our bioinformatics analyses indicated that androgen treatment induces lncZBTB10 expression in prostate cancer cells (Supplementary Fig. S1B), suggesting the potential regulation of lncZBTB10 by androgen receptor (AR) in these cells following androgen treatment. Notably, an analysis of AR and lncZBTB10 expression levels in the Cancer Cell Line Encyclopedia (CCLE) database revealed elevated lncZBTB10 levels in prostate cancer cell lines, such as MDAPCAB, VCaP, LNCaP, and 22Rv1, all of which also presented increased AR expression (Fig. 2a). Consequently, LNCaP and 22Rv1 cells were selected as in vitro models for subsequent experiments. Since lncZBTB10 is a novel long noncoding RNA, its cellular localization was examined in the cytosolic and nuclear fractions of LNCaP and 22Rv1 cells. Our results showed that lncZBTB10 was expressed mainly in the nuclear region of LNCaP and 22Rv1 cells (Fig. 2b and Supplementary Fig. S2A, B). Furthermore, lncZBTB10 expression increased in a dose-dependent manner with androgen treatment (Fig. 2c) and decreased after AR knockdown in both LNCaP and 22Rv1 cells (Fig. 2d, e). AR-chromatin immunoprecipitation (ChIP) sequencing data from the ChIP-Atlas database were analyzed to explore the direct regulation of lncZBTB10 by AR In prostate cancer cells, revealing a critical AR binding signal within the lncZBTB10 locus (Supplementary Fig. S2C). AR ChIP‒PCR in androgen-treated LNCaP and 22Rv1 cells subsequently confirmed the specific binding of AR to the lncZBTB10 locus (Fig. 2f and Supplementary Fig. S2D, E), suggesting a potential role for AR in regulating lncZBTB10 expression in prostate cancer cells. RNA-seq results from both control and lncZBTB10 knockdown LNCaP cells were subjected to gene set enrichment analysis (GSEA) to comprehensively explore the function of lncZBTB10. Interestingly, androgen response signatures were enriched in the knockdown control, and the expression of numerous AR downstream target genes decreased following lncZBTB10 knockdown (Fig. 2g and Supplementary Fig. S2F), implying that lncZBTB10 loss might impair AR function. Indeed, the attenuation of androgen-induced target genes, including NKX3-1, TMPRSS2, and KLK3, was observed after lncZBTB10 knockdown in LNCaP and 22RV1 cells (Fig. 2h, i). Moreover, the loss of lncZBTB10 impeded the androgen-induced nuclear translocation of AR in both LNCaP and 22Rv1 cells (Fig. 2j, k). Conversely, lncZBTB10 overexpression using the CRISPR activation (CRISPRa) system increased NKX3-1 expression in LNCaP cells (Fig. 2l and Supplementary Fig. S2G). Given the RNAinter database prediction of an interaction between lncZBTB10 and AR (Supplementary Fig. S2H), RNA immunoprecipitation (RIP) using an AR antibody in androgen-treated LNCaP and 22Rv1 cells demonstrated physical binding between lncZBTB10 and AR (Fig. 2m). In addition, lncZBTB10 knockdown attenuated AR binding to the NKX3-1 and TMPRSS2 loci (Fig. 2n and Supplementary Fig. S2I), indicating that lncZBTB10 is essential for maintaining AR function in prostate cancer cells. In summary, these findings suggest that the induction of lncZBTB10 expression by androgens may play a crucial role in assisting AR in sustaining its proper function in response to androgen stimulation in prostate cancer cells.

Fig. 2. Androgen-induced lncZBTB10 expression facilitates the proper function of AR through a direct interaction.

a Expression levels of the AR protein (left panel) and lncZBTB10 (right panel) in different prostate cancer cell lines were downloaded from the CCLE database and reanalyzed according to the AR-positive (n = 4) and AR-negative (n = 5) status. *p < 0.05, two-tailed unpaired t-test. b The expression of lncZBTB10 was quantified in both the cytoplasmic (cyto) and nuclear (nuclear) fractions of LNCaP (left panel) and 22Rv1 (right panel) cells (n = 3). *p < 0.05, two-tailed paired t-test. c lncZBTB10 expression was determined in LNCaP (left panel) and 22Rv1 (right panel) cells treated with different doses of DHT for 24 h. *p < 0.05, one-way ANOVA. The expression levels of AR (d) and lncZBTB10 (e) were determined in both LNCaP and 22Rv1 cells treated with an siRNA targeting AR for 48 h (n = 3). *p < 0.05, two-tailed paired t-test. f LNCaP (left panel) and 22Rv1 (right panel) cells were individually treated with DHT (10 nM) for 24 h, and ChIP‒PCR was performed with an AR antibody (left panel). The GAPDH locus without the AR binding site was used as a negative control (right panel). *p < 0.05, two-tailed paired t-test. g LNCaP cells treated with a control siRNA or the lncZBTB10 siRNA for 48 h were subjected to RNA-seq analysis (n = 3). The androgen response signature was enriched in the knockdown control (siNC) phenotype after GSEA. LNCaP (h) and 22Rv1 (i) cells were treated with DHT (10 nM) for 24 h, after which the expression of AR downstream target genes, including NKX3-1, TMPRSS2 and FKBP5, was determined using qRT‒PCR analysis (n = 3). *p < 0.05, two-tailed paired t-test. LNCaP (j) and 22Rv1 (k) cells were treated with DHT (10 nM) for 24 h, after which the AR cellular distribution was determined by Western blotting. The meanings of AR (short) and AR (long) are shorter and longer exposure times, respectively, for AR. l NKX3-1 expression was assessed in both LNCaP (left panel) and 22Rv1 (right panel) cells with stable expression of lncZBTB10-CRISPRa clones using q-RT‒PCR analysis (n = 3). *p < 0.05, two-tailed paired t-test. m LNCaP (left panel) and 22Rv1 (right panel) cells were individually treated with DHT (10 nM) for 24 h, and then RNA-IP (RIP) was performed with an AR antibody. Next, the interaction between lncZBTB10 and AR was analyzed by qRT‒PCR analysis (n = 3). *p < 0.05, two-tailed paired t-test. n LNCaP cells were transfected with either the control siRNA or siRNA targeting lncZBTB10 for 48 h, followed by exposure to DHT (10 nM) for an additional 24 h. The interaction between AR and its downstream gene loci, such as NKX3-1 and TMPRSS2, was subsequently examined via ChIP‒PCR utilizing an AR antibody. *p < 0.05, two-tailed paired t-test.

lncZBTB10 promotes the S-palmitoylation of AR by increasing the expression of ZDHHC7 and ZDHHC21, thereby enhancing its function in prostate cancer cells following androgen treatment

We reanalyzed RNA-seq data from lncZBTB10 knockdown experiments to delve deeper into the mechanism through which lncZBTB10 aids in maintaining the proper function of AR in prostate cancer cells. Notably, the S-palmitoylation signature was enriched in the knockdown control cells compared with lncZBTB10 knockdown in LNCaP cells (Fig. 3a). This result implies that lncZBTB10 might positively regulate the S-palmitoylation process, and notably, AR has been reported to be palmitoylated [22, 23]. However, the full impact of this modification on AR function in prostate cancer cells remains to be thoroughly investigated. Therefore, we hypothesize that lncZBTB10 might also regulate AR function through S-palmitoylation in prostate cancer cells. In support of this idea, treatment with 2-BP, a palmitoylation inhibitor, reduced AR palmitoylation (Fig. 3b) and attenuated the expression of its downstream targets, such as NKX3-1 and TMPRSS2, in both LNCaP and 22Rv1 cells (Fig. 3c and Supplementary Fig. S3A), revealing the important role of S-palmitoylation in AR function. Crucially, lncZBTB10 knockdown indeed attenuated androgen-induced S-palmitoylation of AR (Fig. 3d) and the expression levels of ZDHHC7 and ZDHHC21 (two potential candidates of AR palmitoyl-transferase from the literature [23]) in LNCaP and 22Rv1 cells (Fig. 3e, f). AR-ChIP results demonstrated that both ZDHHC7 and ZDHHC21 were directly regulated by AR in androgen-treated LNCaP cells (Fig. 3g and Supplementary Figs. S2D, S3B). Furthermore, only double knockdown of ZDHHC7 and ZDHHC21 markedly impaired androgen-induced AR nuclear translocation (Fig. 3h, i) and S-palmitoylation and expression levels (Fig. 3j, k) in LNCaP and 22Rv1 cells, whereas single knockdown of ZDHHC7 and ZDHHC21 did not (Supplementary Fig. S3C, D). Similar findings were also observed for androgen-induced NKX3-1 and TMPRSS2 expression in both LNCaP and 22Rv1 cells (Fig. 3l, m). Finally, lncZBTB10 overexpression via the CRISPR activation system directly promoted an increase in AR S-palmitoylation in LNCaP and 22Rv1 cells (Fig. 3n, o). Collectively, our results reveal that lncZBTB10 indirectly enhances androgen receptor (AR) function in prostate cancer cells by inducing its S-palmitoylation.

Fig. 3. lncZBTB10 enhances AR S-palmitoylation and sustains its function in prostate cancer cells by inducing ZDHHC7 and ZDHHC21 expression.

a The S-palmitoylation signature was enriched in the knockdown control (siNC) phenotype after the lncZBTB10 knockdown RNA-seq results were analyzed using GSEA. b The S-palmitoylation status of AR was examined in LNCaP cells treated with 2-BP (25 μM) for 24 h through Western blot analysis. c The expression levels of NKX3-1 and TMPRSS2 were measured in LNCaP cells treated with 2-BP (25 μM) for 24 h through qRT‒PCR analysis (n = 3). *p < 0.05, two-tailed paired t-test. d, e LNCaP cells were subjected to lncZBTB10 knockdown for 48 h, followed by treatment with DHT (10 nM) for an additional 24 h. The S-palmitoylation status of AR (d) and the expression of ZDHHC7 and ZDHHC21 were subsequently assessed using Western blot and q-RT‒PCR analyses (n = 3) (e), respectively. f 22Rv1 cells were subjected to lncZBTB10 knockdown for 48 h and then treated with DHT (10 nM) for an additional 24 h. The expression levels of ZDHHC7 and ZDHHC21 were assessed via qRT‒PCR analyses (n = 3). *p < 0.05, two-tailed paired t-test. g LNCaP (left panel) and 22Rv1 (right panel) cells were treated with DHT (10 nM) for 24 h, and then ChIP‒PCR was performed with an AR antibody. The AR-binding signals for the ZDHHC7 and ZDHHC21 loci were assayed by qRT‒PCR. *p < 0.05, two-tailed paired t-test. The cellular localization (h, i), S-palmitoylation level (j, k), and expression of downstream target genes (l, m) of AR were evaluated via Western blot analysis in LNCaP and 22Rv1 cells under either control conditions or double knockdown of ZDHHC7 and ZDHHC21 (40 nM) for 48 h (n = 3). *p < 0.05, two-tailed paired t-test. S-palmitoylation of AR was assayed in both LNCaP (n) and 22Rv1 (o) cells with stable lncZBTB10 overexpression mediated by the CRISPR activation (CRISPRa) system.

The S-palmitoylated residues of androgen receptor (AR) and its interaction with lncZBTB10 are pivotal for AR nuclear localization and AR function in prostate cancer cells during androgen treatment

Given that androgen-induced AR S-palmitoylation is known to be mediated by the upregulation of lncZBTB10 expression (Fig. 3), we aimed to identify the specific residues involved in S-palmitoylation of AR. Next, the S-palmitoylation of AR was predicted by CSS-Palm 4.0, and three cysteine residues (C577, C580 and C615) (Supplementary Fig. S4A) were identified in AR. Individual and all three cysteine residues of AR were mutated to alanine residues to further verify which residue was S-palmitoylated in AR and affected its function. Interestingly, the nuclear translocation of AR constructs with triple and single cysteine mutations was significantly diminished following androgen treatment compared with that of the wild-type counterpart (Fig. 4a). Furthermore, the S-palmitoylation level of AR was noticeably decreased in the protein with triple cysteine mutations compared with the results obtained from the protein with a single cysteine mutation (Fig. 4b). Next, AR constructs with different cysteine mutations were transiently transfected into 293 T cells to investigate changes in their cellular localization after androgen stimulation. The results indicated that the wild-type AR translocated into the nucleus, where it formed puncta, whereas both the triple and single cysteine mutants of AR were exclusively expressed in the cytoplasm as puncta after androgen stimulation (Fig. 4c). In addition, the results also revealed that overexpression of wild-type AR increased NKX3-1 expression, whereas AR with triple or single cysteine mutations decreased its expression (Fig. 4d). These results suggest that individual cysteine residues with the potential to be S-palmitoylated might be crucial for AR nuclear localization and that their simultaneous loss impairs AR function.

Fig. 4. lncZBTB10 maintains AR nuclear localization through its N-terminal region.

a In 293T cells, AR-GFP wild-type (WT), triple cysteine mutation (3A), or single cysteine mutation was transiently overexpressed for 48 h, and the cells were then treated with DHT (10 nM) for an additional 2 h. The cellular distribution of AR was assessed by Western blot analysis. b S-palmitoylation of AR was analyzed in 293T cells with transient overexpression of the wild-type (WT), triple cysteine mutant (3A), or single cysteine mutant for 48 h through Western blot analysis. c In 293T cells, AR-GFP wild-type (WT), triple cysteine mutant (3A), or single cysteine mutant was transiently overexpressed for 48 h, and the cells were then treated with DHT (10 nM) for an additional 2 h. The cellular distribution of AR was assessed via confocal microscopy. d The expression levels of AR target genes, including NKX3-1 and TMPRSS2, were determined in 293T cells with transient overexpression of wild-type (WT) AR-GFP, the triple cysteine mutant (3A), or the single cysteine mutants for 48 h through qRT‒PCR analysis (n = 3). *p < 0.05, two-tailed paired t-test. e The cellular distribution of AR was assessed in 293T cells with AR-GFP overexpression and lncZBTB10 knockdown for 48 h via confocal microscopy. f 293 T cells transiently overexpressed different AR-GFP deletion constructs for 48 h, and the AR cellular distribution was assessed via confocal microscopy. The interaction between lncZBTB10 and AR (g), as well as the expression of AR downstream target genes (h), were investigated through RNA immunoprecipitation (RIP) and qRT‒PCR in 293T cells with transient overexpression of various AR-GFP deletion constructs for 48 h (n = 3). *p < 0.05, two-tailed paired t-test.

Recently, AR has been reported to exhibit phase separation behavior as puncta in the nucleus, but the underlying mechanism by which mobile AR is trapped is still unknown [14, 15]. Since lncZBTB10 is expressed mainly in the nucleus of prostate cancer cells (Fig. 2a), we hypothesized that lncZBTB10 can trap mobile AR in the nucleus of prostate cancer cells. Indeed, AR-GFP translocated into the nucleus and formed nuclear condensates, which were subsequently dispersed upon lncZBTB10 knockdown in cells following androgen stimulation (Fig. 4e). Next, we further investigated how lncZBTB10 binds to AR and regulates its nuclear condensate formation and molecular function by analyzing potential intrinsically disordered protein regions (IDRs) and RNA binding regions in AR using IUPred and HybridRNAbind, respectively. Surprisingly, the IDR (1 ~ 200 amino acids) and the RNA binding region (RBP, 2 ~ 90 amino acids) overlapped according to the prediction results (Supplementary Fig. S4B, C). The predicted RBPs (2–90) and several well-known functional domains of AR, including the ligand-binding domain (LBD), DNA-binding domain (DBD), and previously identified IDR regions (T1, 110–370 and T5, 365–485), were individually deleted to investigate their effects on AR nuclear condensate formation and changes in its molecular function and to verify the in silico findings. The findings indicated that AR deletion constructs, including del2–90, delT1, delT5, and delDBD, formed cytosolic condensates, whereas the wild-type and delLBD constructs were predominantly localized in the nuclear region following androgen stimulation (Fig. 4f). More importantly, disruption of the potential RBP (2–90), T1, T5, and DBD impaired the physical interaction between lncZBTB10 and AR, whereas deletion of the LBD did not affect this interaction (Fig. 4g and Supplementary Fig. S4D). Furthermore, only the overexpression of the wild-type and delLBD AR constructs resulted in the induction of AR downstream target genes, such as NKX3-1 and TMPRSS2 (Fig. 4h). In contrast, AR deletion constructs, including delRBP, delT1, delT5, and delDBD, did not have a similar effect on the expression of these target genes (Fig. 4h). In summary, these findings imply that lncZBTB10 may play a role in facilitating AR puncta formation and enhancing AR function through a direct interaction in prostate cancer cells.

The S-palmitoylated residues of androgen receptor (AR) and its interaction with lncZBTB10 are pivotal for AR nuclear localization and AR function in prostate cancer cells during androgen treatment

The cellular function of lncZBTB10 was studied by knocking it down in LNCaP and 22Rv1 cells, and the growth and migration of these cells were individually measured. The results revealed that the loss of lncZBTB10 impaired not only cell growth but also cell migration in both prostate cancer cell lines (Fig. 5a–d). Next, a stable clone of LNCaP cells overexpressing lncZBTB10 via the CRISPRa system was orthotopically injected into the mouse prostate to establish a mouse model of prostate cancer and directly verify the effect of lncZBTB10 on tumor growth in vivo. Compared with that in the control cells, the overexpression of lncZBTB10 in LNCaP cells accelerated the progression of prostate cancer (Fig. 5e, f), resulting in an increased proliferative ability (Fig. 5g). Moreover, the expression levels of AR, ZDHHC7 and ZDHHC21 were obviously increased in xenograft tumors overexpressing lncZBTB10, which supported our in vitro findings (Fig. 5h–j). Collectively, these findings highlight the pivotal role of lncZBTB10 in fostering prostate cancer development in vivo through the upregulation of AR expression.

Fig. 5. The upregulation of lncZBTB10 promotes tumor growth in an orthotopic mouse model of prostate cancer by inducing AR expression through ZDHHC7 and ZDHHC21.

LNCaP (a) and 22Rv1 (b) cells were subjected to siRNA-mediated knockdown of lncZBTB10 (40 nM) for 48 h, followed by subculturing to evaluate cell growth in the presence or absence of DHT (10 nM) treatment in 5% CSS media at specified time points (n = 3). *p < 0.05, two-way ANOVA following Tukey’s test for a and two-tailed paired t-test for b. c, d Both LNCaP and 22Rv1 cells were treated with the siRNA (40 nM) targeting lncZBTB10 for 48 h, and then a migration assay was performed for additional 24 h. Representative images of the migration assays are depicted (c), and migratory cells were quantified from five distinct areas of the transwells (d). *p < 0.05, one-way ANOVA. e, f LNCaP cells overexpressing lncZBTB10 via the CRISPR activation (CRISPRa) system were orthotopically injected into the mouse prostate, and mice were maintained for two months. Representative images of the xenograft tumors are presented (e), and the corresponding tumor weights were quantified (n = 5) (f) following the sacrifice of the mice. *p < 0.05, two-tailed paired t-test. The expression levels of ki67 (g), AR (H), ZDHHC7 (i), and ZDHHC21 (j) were determined in xenograft tumors via IHC staining, and representative images are shown.

The overexpression of lncZBTB10 exacerbates the development of abiraterone resistance by upregulating AR expression through S-palmitoylation in abiraterone-resistant prostate cancer cells

Previously, ligand-independent activation of AR was identified as a potential underlying mechanism contributing to resistance to abiraterone, a drug designed to target androgen biosynthesis [7, 24, 25]. Our current research further supports this hypothesis by revealing that androgen-induced lncZBTB10 expression is mediated by AR activation, suggesting its potential involvement in the development of abiraterone resistance. Specifically, our results revealed elevated AR expression in abiraterone-resistant (AaR) cells (Fig. 6a), with lncZBTB10 expression showing a corresponding increase exclusively in AaR cells, whereas its expression decreased in cells resistant to enzalutamide (EnzaR), apalutamide (ApaluR), and darolutamide (DaroR) (all targeting the androgen receptor) (Fig. 6b, c). Furthermore, the marked reduction in lncZBTB10 expression observed in AaR cells with AR knockdown (Fig. 6d) underscores its dependency on AR regulation. Consistent with these findings, AR S-palmitoylation was significantly increased in AaR cells compared to their parental counterparts (Fig. 6e). Notably, the knockdown of lncZBTB10 resulted in a reduction in AR S-palmitoylation (Fig. 6f) and the expression of its downstream target genes, including NKX-1 and TMPRSS2, during androgen treatment (Fig. 6g). Additionally, lncZBTB10 knockdown led to a decrease in AR expression in AaR cells (Fig. 6h). Finally, the loss of lncZBTB10 not only impaired cell growth (Fig. 6i) but also restored the sensitivity of AaR cells to abiraterone treatment (Fig. 6j). In summary, these collective findings suggest that the overexpression of lncZBTB10 potentially contributes to increased AR expression, indicating that lncZBTB10 may be involved in the development of resistance to abiraterone.

Fig. 6. Increased lncZBTB10 expression contributes to the development of abiraterone resistance in prostate cancer cells.

The expression levels of AR (a) and lncZBTB10 (b) were assessed in normal prostate epithelial cells and various prostate cancer cell lines through Western blot and qRT‒PCR analyses (n = 3). AaR abiraterone-resistant cells, EnzaR enzalutamide-resistant cells. *p < 0.05, two-tailed paired t-test. c The expression levels of ZBTB10 were determined in parental LNCaP cells and darolutamide-resistant (DaroR) and apalutamide-resistant (ApaluR) prostate cancer cells through qRT‒PCR analysis. *p < 0.05, one-way ANOVA. d AaR cells were treated with an siRNA targeting AR for 48 h, and lncZBTB10 expression was subsequently measured via qRT‒PCR. *p < 0.05, two-tailed paired t-test. e S-palmitoylation of AR was analyzed in parental LNCaP cells and AaR cells through Western blot analysis. f, g AaR cells were treated with an siRNA targeting lncZBTB10 (40 nM) for 48 h and then treated with DHT (10 nM) for another 24 h. AR S-palmitoylation (f) and the expression of its downstream target genes (g) were determined by Western blotting and qRT‒PCR (n = 3), respectively. *p < 0.05, two-tailed paired t-test. h AR expression was measured in AaR cells with lncZBTB10 knockdown (40 nM) for 48 h via Western blot analysis. i, j AaR cells were subjected to siRNA-mediated knockdown of lncZBTB10 (40 nM) for 48 h, followed by subculturing to evaluate cell growth in the absence (i) or presence (j) of abiraterone (10 μM) treatment for 48 h (n = 3). *p < 0.05, two-tailed paired t-test.

Discussion

Given that AR signaling remains pivotal in different stages of prostate cancer, including CRPC and resistance to second-generation anti-androgen drugs, a comprehensive examination of the underlying mechanisms governing AR function is imperative. Previously, several lncRNAs have been recognized for their pivotal roles in prostate cancer development through diverse androgen receptor regulatory mechanisms. For example, PRNCR1 (PCAT8) and PCGEM1 exhibit sequential binding to AR, contingent on its carboxy-terminal acetylation and amino-terminal methylation. This interaction robustly amplifies both ligand-dependent and ligand-independent androgen receptor-mediated gene activation programs in CRPC cells [21]. Furthermore, overexpression of LINC00675 has been reported to promote AR stability by blocking its MDM2-mediated ubiquitination, subsequently contributing to the development of CRPC [26]. In contrast, the loss of lncLBCS prevents the recruitment of hnRNPK, a critical suppressor of AR translation, to the AR mRNA and then promotes its translation in CRPC cells [19]. Furthermore, NXTAR serves as a tumor-suppressive lncRNA capable of epigenetically downregulating AR/AR-V7 expression in prostate cancer cells [20]. This process is achieved through its direct binding to and recruitment of EZH2, leading to H3K27me3 deposition at its promoter region [20]. Our study highlights the pioneering identification of a novel lncRNA, lncZBTB10, which is overexpressed in a substantial cohort of prostate cancer samples, via systematic bioinformatics analysis and wet-bench validation. Crucially, our findings indicate that lncZBTB10 positively regulates AR nuclear localization and function in prostate cancer cells through S-palmitoylation, a mechanism that has not been explored previously. Moreover, its expression is further elevated in abiraterone-resistant prostate cancer cells and leads to the development of abiraterone resistance (Fig. 7). Overall, these findings provide novel insights into AR activation in prostate cancer cells, shedding light on the crucial role of lncZBTB10 induction in this process.

Fig. 7. A cartoon that briefly summarizes the involvement of lncZBTB10 in facilitating both tumor growth and the emergence of resistance to abiraterone in prostate cancer.

The figure was created using BioRender.com.

While lncRNAs have been extensively investigated in various physiological and pathological contexts, as a novel class of RNA, the function of each individual lncRNA must be assessed in relation to its cellular localization. Nevertheless, few studies have endeavored to concurrently explore the cellular localization of novel lncRNAs with the aim of establishing connections between their cellular functions in AR regulation during the identification process. For example, PlncRNA-1 and ARLNC1, which target AR in prostate cancer cells and directly bind to the AR transcript, respectively, are cytosolic lncRNAs that increase AR levels by interacting with both miR-34c and miR-297 [18, 27]. Conversely, the newly discovered nuclear lncRNA NXTAR, which functions as a tumor suppressor, has been shown to inhibit prostate cancer growth through targeted binding to the AR locus, resulting in epigenetic silencing [20]. Our findings revealed that nuclear lncZBTB10 is positively regulated by androgen receptor (AR) and that their interaction facilitates the activation of AR downstream target genes. Furthermore, lncZBTB10 enhances AR nuclear translocation and protein expression through S-palmitoylation in prostate cancer. These observations collectively indicate the presence of a positive feedback regulatory loop in prostate cancer. However, the specific factors governing lncZBTB10-mediated gene regulation and the stability of AR induced by S-palmitoylation remain unclear.

Our findings demonstrate that lncZBTB10 promotes the S-palmitoylation of AR in prostate cancer cells, with ZDHHC7 and ZDHHC21 identified as potential palmitoyl-transferases positively regulated by AR. Notably, the double knockdown of these transferases significantly impairs AR S-palmitoylation and its nuclear localization, whereas the individual knockdown does not, implying compensatory functions of ZDHHC7 and ZDHHC21. A previous study reported that the overexpression of ZDHHC7 or ZDHHC21 leads to an increase in S-palmitoylation of AR, whereas individual knockdown of either ZDHHC7 or ZDHHC21 decreases both AR S-palmitoylation and its signaling function in C4-2 prostate cancer cells [23]. Furthermore, individual knockdown of ZDHHC7 or ZDHHC21 specifically reduces the membrane level of AR in C4-2 cells without impacting its nuclear level [23]. Recently, downregulation of ZDHHC7 has been reported in prostate cancer, and the loss of ZDHHC7 attenuates cancer growth in vitro and in vivo through the downregulation of AR mRNA expression [22]. However, these studies failed to identify the potential S-palmitoylation sites of AR and provide a detailed description of its effects on AR function in prostate cancer cells. In contrast, our study not only pinpointed potential S-palmitoylation sites on AR but also indicated that only the double knockdown of ZDHHC7 and ZDHHC21 significantly diminished AR nuclear localization, along with reductions in S-palmitoylation levels and overall expression in prostate cancer cells following androgen treatment. Moreover, our findings highlight the mediating role of lncZBTB10 in these observed phenomena. Finally, our future research direction aims to identify potential factors driving AR degradation following the loss of S-palmitoylation in prostate cancer cells.

AR activation typically occurs through androgen binding to its LBD, leading to its translocation into the cell nucleus to regulate downstream targets. Recent studies have indicated that both full-length AR and its variant, AR-V7, exhibit phase separation characteristics, including the formation of nuclear puncta upon androgen stimulation [13, 14]. Notably, deletion of the N-terminal domain (NTD) and DNA-binding domain (DBD), or the DBD–hinge–LBD region, impedes nuclear puncta formation upon androgen stimulation, emphasizing the essential role of full-length AR in this process in prostate cancer cells [14]. The NTD, also known as the activation function 1 domain (AF1), contains an intrinsically disordered region (IDR) divided into the tau-1 (amino acids 100–359) and tau-5 regions (amino acids 360–528) [10]. In addition, recent research has shown that AR phase separation is driven by tyrosine residues clustered around the ²³FQNLF [27] motif and tau-5 region, as observed using nuclear magnetic resonance (NMR) and site-directed mutagenesis assays [12]. Importantly, targeting the IDR of AR, specifically the tau-5 region, with molecules such as EPI001 and lae reduces the number of AR foci [12]. Our study revealed a novel RNA binding region (amino acids 2–90) located before the tau-1 region that is crucial for the lncZBTB10 interaction and nuclear puncta formation after androgen stimulation. While previous research has identified three movement patterns of nuclear AR—immobile, increased displacement via possible phase separation, and freely mobile—the factors or underlying mechanisms responsible for trapping freely mobile AR to initiate phase separation remain unknown [15]. Collectively, these findings suggest that nuclear lncZBTB10 may be a potential factor that traps freely mobile AR, initiating nuclear puncta formation upon androgen stimulation of prostate cancer cells.

Current novel anti-androgen therapies targeting the reduction of persistent AR signaling typically involve either inhibiting androgen production (abiraterone) or competing with endogenous ligands (enzalutamide) for binding to the AR C-terminal LBD [28]. However, despite their initial effectiveness in treating castration-resistant prostate cancer (CRPC), most patients eventually develop resistance to these therapies [5]. Further investigations revealed that the mechanisms underlying resistance to enzalutamide or abiraterone predominantly employ either drug-specific or shared mechanisms that undermine drug efficacy. For example, the emergence of the specific AR F876L missense mutation is induced by enzalutamide treatment, leading to enzalutamide resistance [6]. Additionally, an increase in intratumoral de novo androgen synthesis due to CYP17A1 overexpression has been identified as a specific mechanism causing abiraterone resistance [29]. Moreover, the overexpression of AR variants and the induction of the glucocorticoid receptor (GR) expression are recognized as common mechanisms contributing to resistance to both enzalutamide and abiraterone [7, 24, 25]. Nevertheless, our findings revealed an increase in lncBBB expression in AaR cells but a decrease in EnzaR, DaroR, and ApaluR cells, suggesting that lncZBTB10 induction may still be highly dependent on AR function. Furthermore, the overexpression of lncZBTB10 has emerged as a novel potential underlying mechanism leading to abiraterone resistance by increasing AR expression through S-palmitoylation. Hence, innovating a therapeutic strategy targeting lncZBTB10 in prostate cancer patients will be our future direction.

Author contributions

YSC and SCL (corresponding authors) conceived the ideas and supervised the research. SCL performed the experiments and analyzed the data. YSL prepared all the constructs used in this study. TMHN and WTC conducted the confocal image analysis. YCT and HYC performed the cell experiments. SCL, YSC, and SCL (corresponding author) jointly prepared the manuscript. SCL (corresponding author) conducted the bioinformatics analyses. SCL, YSL, and YCT assisted with the animal studies. YSC and TYL provided clinical insights for the study. SCL (corresponding author) secured the funding for this study. All the authors reviewed the manuscript and consented to its publication.

Funding

Research grants from the National Science and Technology Council (NSTC 109-2636-B-006-006; NSTC 110-2636-B-006-009; NSTC 111-2636-B-006-012; NSTC 112-2636-B-006-008; and NSTC 113-2636-B-006-005) and National Health Research Institutes (NHRI-EX112-11220BI and NHRI-EX113-11220BI) in Taiwan provided support for this study.

Data availability

The data supporting this article can be found both within the article itself and in its accompanying online Supplementary Material.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

We have confirmed that all methods were performed in accordance with the relevant guidelines and regulations. Approval for all the animal studies was obtained from the Institutional Animal Care and Use Committee (IACUC: 108299) at the Laboratory Animal Center, NCKU.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-025-02938-1.