Human PCa samples and adjacent tissues used in this study were obtained from 42 localized PCa patients who underwent radical prostatectomy at the Second Hospital of Tianjin Medical University, as our previously described (14). In addition, we also obtained the serum of 33 metastatic PCa patients who underwent at least 4 cycles of DTX-based chemotherapy from January 2020 to January 2021. This study was approved by the Ethics Committee of the Second Hospital of Tianjin Medical University, and written informed consent was obtained from each patient. Clinical tissues and serum samples were snap-frozen immediately in liquid nitrogen and stored until use. DTX resistance is defined as progression (i.e. PSA level or imaging examination) after at least 4 cycles of standard DTX-based chemotherapy.

The human prostate carcinoma cell lines PC3 and 22RV1 were purchased from the National Collection of Authenticated Cell Culture (Shanghai, China) and maintained in DMEM/F12 medium supplemented with 10% fetal bovine serum (Biological Industries, Israel) and 1% antibiotics (penicillin and streptomycin, HyClone, USA). As previously described, DTX-resistant cell lines (PC3/DR and 22RV1/DR) were developed by stepwise increased DTX concentrations over a period of 8-9 months (19, 20). DTX, Erastin, Ferrostatin-1 (Fer-1),CHX and Borz were purchased from MedChemExpress (China).

Total RNA used for RNA sequencing was isolated from three independent groups of PC3 and PC3/DR cells using TRIzol reagents (Invitrogen, USA). High-quality RNA samples were subjected to library construction, and then the library was sequenced on an Illumina NovaSeq™ 6000 following the standard procedures of Geneseed (China). In addition, the mRNA library of PCAT1-knockdown PC3/DR cells and control PC3/DR cells was sequenced on an Illumina NovaSeq™ 6000 following the vendor’s recommended protocol (LC-Bio, China). The differentially expressed lncRNAs and mRNAs were selected with fold change > 2 or fold change < 0.5 and p value < 0.05 by the R package edgeR.

Cell viability was analyzed by a CCK-8 assay (Dojindo, Japan) according to the manufacturer’s protocol. For cell colony formation, 2000 cells were seeded into the wells of 6-well plates and cultured in specific culture media. After 2 weeks, the cells were fixed with ethanol and stained with 0.1% crystal violet (Solarbio, China). The number of visible colonies was counted using ImageJ software.

Lipid ROS production was detected using an ROS Assay Kit (Beyotime, China). Lipid peroxidase was assessed by a Liperfluo Assay Kit (Dojindo, Japan), and the GSH level was evaluated by a GSH assay kit (Nanjing Jiancheng, China) following the manufacturer’s protocol.

Intracellular total iron and ferrous iron concentrations were detected by an iron assay kit following the manufacturer’s protocol (Sigma-Aldrich, USA). A FerroOrange assay (Dojindo, Japan) was used to detect the ferrous iron level, and the mitochondrial membrane potential was detected by a Mitochondrial Membrane Potential Assay Kit with JC-1 (Beyotime, China). The TEM samples were prepared using our previously described method, and ultra-structural images were captured with the transmission electron micorscope (JEM-1011, Japan) (18).

Total RNA was isolated from cells and clinical samples using TRIzol reagent (Invitrogen, USA). Reverse transcription was performed to synthesize cDNA using the lncRcute and miRcute Kit (Tiangen, China). qRT-PCR was conducted using Hieff SYBR Green Master Mix (Yeasen, China). The expression of lncRNAs, mRNAs and miRNAs was calculated and normalized to the internal control β-actin by using the standard 2^-ΔΔCt method. The qRT-PCR primer sequences are listed in Table S1 .

Cells were lysed on ice with RIPA buffer (Beyotime, China) containing 1 × protease inhibitor cocktail (Sigma-Aldrich, USA). Equal amounts of protein samples (20 μg) were separated by 10% SDS-PAGE gels and then transferred to PVDF membranes. The membranes were incubated with primary antibodies and the associated secondary rabbit or mouse antibodies. The primary antibodies were as follows: SLC7A11 (1:1000, Abcam, USA, Cat. ab175186), GPX4 (1:1000, Abcam, USA, Cat. ab125066), TFAP2C (1:1000, Proteintech, China, Cat. 14572-1-AP), c-Myc (1:1000, Proteintech, Cat. 67447-1-Ig) and β-actin (1:1000, Proteintech, China, Cat. 66009-1-Ig).

Full-length PCAT1, SLC7A11, c-Myc, TFAP2C and their negative control cDNA were cloned into the pcDNA3.1 vector (Invitrogen, USA). siRNAs molecules specifically targeting PCAT1, SLC7A11 c-Myc and TFAP2C were designed and synthesized by RiboBio (Guangzhou, China). miRNA mimics and inhibitors of miR-25-3p were also purchased from RiboBio (Guangzhou, China). Lipofectamine 2000 Reagent (Invitrogen, USA) was used to transiently transfect these reagents. In addition, a GV112 lentivirus vector (GeneChem, China) was used to generate short hairpin RNAs (shRNAs) against PCAT1 (labeled as shPCAT1 #1) and a negative control (labeled as shNC). The lentiviral packaging for the PCAT1 expression plasmids and empty control were generated with a lentiviral packaging kit (GeneChem, China). The infected PCa cells were selected with 2 μg/mL puromycin (Sigma-Aldrich, USA) for up to 2 weeks. The siRNA sequences are presented in Table S1 .

The RNA pull-down assay was performed by a Pierce™ Magnetic RNA-Protein Pull-Down Kit (Thermo Scientific, USA). Biotin-labeled full-length and short-length PCAT1 probes were synthesized by GenePharma (Shanghai, China). In brief, the labeled RNA probe was incubated with streptavidin magnetic beads and washed with 20 mM Tris. Then, the master mix of the RNA-protein binding reaction was incubated with RNA-bound beads. The lncRNA-interacting proteins were eluted with 1× SDS loading buffer and further separated by 10% SDS-PAGE gels. The Fast Silver Stain Kit (Beyotime, China) was used to visualize the associated proteins.

The RIP assay was performed using a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) according to the manufacturer’s instructions. In brief, PC3/DR cells were harvested and lysed in RIP lysis buffer and then incubated with magnetic beads coated with anti-c-Myc and anti-IgG antibodies. Then, the coprecipitated RNAs were extracted and evaluated by qRT-PCR.

The wild-type and mutant PCAT1 and SLC7A11 3’UTRs were cloned into a pmiR-RB-Report™ vector (RiboBio, China). Then, PC3 cells were plated on a 96-well plate and cotransfected with wild-type and mutant luciferase reporter plasmids and miR-25-3p mimic. In addition, the promoter segments of PCAT1 and SLC7A11 were cloned into the pGL3-basic vector (Promega, USA). The pGL3-PCAT1 and pGL3-SLC7A11 vectors were cotransfected with TFAP2C or c-Myc plasmids in PC3 cells. The relative luciferase activity was measured by the Dual Luciferase Assay System according to the manufacturer’s protocol (Promega, USA).

ChIP assays were performed using a ChIP kit following the manufacturer’s instructions. Briefly, cells were crosslinked with formaldehyde and sonicated to 200-1000 bp. Immunoprecipitation was conducted with an anti-c-Myc, antiTFAP2C or anti-IgG control. Precipitated DNA was amplified by qPCR. The ChIP-qPCR sequences are listed in Table S1 .

All mice were housed in a specific pathogen-free environment at the Second Hospital of Tianjin Medical University, and experimental procedures were approved by the Tianjin Medical University Institutional Animal Care and Use Committee. PC3 and PC3/DR cells (5×10⁶ cells) were subcutaneously injected into each flank of six-week-old male BALB/c nude mice (HFK Biotech, China). When the tumor volume reached 100 mm³, the mice were treated with Dimethyl Sulfoxide (DMSO) alone, DTX (5 mg/kg body weight, every two days) with DMSO or erastin (20 mg/kg body weight in 20 μl DMSO plus 130 μl corn oil, daily) by intraperitoneal injection, as previously described (9, 20). The tumor volumes of the mice were assessed by measuring the length and width every three days. After 4 weeks, all mice were sacrificed, and the tumors and metastatic tissues were harvested for further analysis.

IHC was conducted on tumor samples from xenograft mice following standard methods. The tumor specimens were fixed in formalin, embedded in paraffin and then cut into 4 μm slices. After dewaxing, rehydration and antigen retrieval, the slides were incubated with specific primary antibodies against Ki-67 (1:5000, Proteintech, China) and SLC7A11 (1:1000, Proteintech, China). After incubation with the corresponding secondary antibodies, the sections were stained with diaminobenzidine and counterstained with hematoxylin. Representative images were photographed using an Olympus light microscope.

All statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software, USA) and SPSS software 17.0 (IBM, USA). The data are presented as the mean ± standard deviation (SD) from three independent experiments. Statistical significance was analyzed using Student’s t-test or two-way ANOVA tests. The correlation between PCAT1 and TFAP2C expression levels was determined by using Pearson’s correlation coefficient. All data were considered statistically significant at a p value < 0.05.

To clarify whether ferroptosis inducer can also inhibit DTX-resistant PCa, we generated two acquired DTX-resistant cell lines (PC3/DR and 22RV1/DR) derived from castration-resistant PCa cells (PC3 and 22RV1) as our previously described (19, 20). DTX resistance was further validated by cell viability and colony formation assays ( Figure S1A–D ). Subsequently, we assessed the susceptibility to ferroptosis in DTX-resistant PCa cells and their drug-naïve parental cells following treatment with the ferroptosis inducer erastin. As expected, the growth of PC3 and 22RV1 cells was significantly inhibited by DTX and erastin, and cell growth inhibition was reversed by the ferroptosis inhibitor Fer-1 ( Figures 1A, B ). Exposure to DTX or erastin in PC3 and 22RV1 cells greatly increased the lipid ROS levels and intracellular total iron and ferrous iron content ( Figures 1C–H ), whereas the mitochondrial membrane potential was decreased upon DTX and erastin treatment ( Figures 1I–L ). Intriguingly, we found that the cytotoxicity of DTX and erastin was restricted in PC3/DR and 22RV1/DR cells ( Figures 1A, B ). In addition, PC3/DR and 22RV1/DR cells had a ferroptosis resistance phenotype against erastin and DTX compared with PC3 and 22RV1 cells ( Figures 1C–L ). FerrOrange staining assays also revealed that PC3 and 22RV1 cells had higher ferrous iron levels after DTX and erastin treatment than PC3/DR and 22RV1/DR cells ( Figure 1M ). When PC3 and PC3/DR cells were treated with erastin, the morphological features of mitochondria involved a larger size, decreased membrane density and cristae thinning in PC3 cells, while the morphology of mitochondria was not changed in PC3/DR cells ( Figure 1N ). Moreover, the GSH content was higher in DTX-resistant PCa cells than in DTX-sensitive cells ( Figures 1O, P ). To further assess the therapeutic potential of erastin in DTX-resistant PCa cells, we treated BALB/c nude mice bearing PC3 and PC3/DR cells with erastin, DTX and DMSO when the tumor volume reached 100 mm³. As shown in Figures 1Q, R , exposure to erastin or DTX significantly delayed the tumor growth of PC3 xenografts by tumor weight compared to treatment with DMSO. Consistent with the in vitro findings, we did not observe any significant differences in tumor growth in PC3/DR xenografts treated with erastin or DTX ( Figures 1Q, R ). In addition, Immunohistochemistry (IHC) staining revealed that the tissue morphology and Ki-67 levels in the PC3/DR group were similar upon DTX and erastin treatment ( Figure 1S ). These results indicated that DTX-resistant PCa cells have the ability to inhibit ferroptosis in vitro and in vivo.

To elucidate the potential role of lncRNAs mediating PCa chemoresistance, we performed a comprehensive analysis of lncRNAs using whole transcriptome sequencing to compare DTX-resistant PC3/DR cells with sensitive parental PC3 cells. Subsequent differential expression analysis revealed 1 088 upregulated lncRNAs and 899 downregulated lncRNAs in PC3/DR cells compared to PC3 cells ( Figure 2A ). Subsequently, we selected the top 20 upregulated lncRNAs according to the inclusion criteria for further validation by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) ( Figure S2A ). Among these lncRNAs, we observed that PCAT1 expression was greatly upregulated in PC3/DR and 22RV1/DR cells compared to PC3 and 22RV1 cells by qRT-PCR ( Figure 2B ). To validate the clinical significance of PCAT1 expression in PCa, we evaluated the expression level of PCAT1 in PCa samples. Consistent with our previous study, the expression of PCAT1 was found to be significantly upregulated in tumors compared with adjacent normal tissues ( Figure 2C ) (21). Moreover, we found that the expression level of PCAT1 in pretherapy plasma was higher in metastatic PCa patients who suffered from progressive disease (PD) after at least 4 cycles of DTX chemotherapy than in those without PD ( Figure 2D ). Together, the above data indicated that elevated PCAT1 expression might be associated with DTX resistance in PCa.

To further explore whether PCAT1 mediates DTX-induced ferroptosis in PCa cells, we ectopically expressed PCAT1 in PC3 and 22RV1 cells ( Figure 2E ). Cell viability assays showed that OE of PCAT1 enhanced DTX resistance in PC3 and 22RV1 cells ( Figures 2F, G ). In addition, we also found that PCAT1 OE reversed the erastin-induced and DTX-induced growth inhibition of PC3 and 22RV1 cells by CCK-8 assay and colony formation assay ( Figures 2H–K ). Additionally, we further assessed the phenotype of ferroptosis in PCAT1-transfected PC3 and 22RV1 cells. As shown in Figures 2L–Q , lipid ROS production and intracellular iron content levels were greatly decreased upon DTX and erastin treatment, whereas mitochondrial membrane potential was increased. In addition, we also found that GSH levels were increased in PC3 and 22RV1 cells after PCAT1 transfection ( Figures 2R, S ). Taken together, our results demonstrated that upregulation of PCAT1 could increase DTX resistance and inhibit ferroptosis.

To further evaluate the ferroptotic process of PCAT1 in DTX-resistant PCa cells, we performed transient KD of PCAT1 in PC3/DR and 22RV1/DR cells by using small interfering RNA (siRNA) molecules. The efficiency of KD was verified by qRT-PCR assay ( Figure 3A ). Inhibition of PCAT1 in PC3/DR and 22RV1/DR cells significantly suppressed DTX resistance ( Figures 3B, C ). Moreover, PCAT1 KD increased the erastin-induced and DTX-induced cytotoxicity of PC3/DR and 22RV1/DR cells ( Figures 3D, E ). At the same time, the cell growth inhibition of DTX and erastin could be reversed by Fer-1 in PCAT1 KD cells ( Figures 3D, E ). Colony formation assays also revealed that PCAT1 KD increased the cytotoxicity of erastin or DTX in PC3/DR and 22RV1/DR cells ( Figures 3F, G ). In addition, we found that inhibition of PCAT1 increased lipid ROS production and intracellular iron content, whereas mitochondrial membrane potential was decreased ( Figures 3H–M ). GSH levels were increased in PC3/DR and 22RV1/DR cells after PCAT1 KD ( Figures 3N, O ). In addition, we also generated xenografts by subcutaneous injection of PC3/DR cells with stable KD of PCAT1 and control cells followed by intraperitoneal treatment with DTX and erastin. Thus, we constructed PCAT1 shRNA (i.e. shPCAT1 #1) using the sequence of siRNA #1 to stable KD the expression of PCAT1 in PC3/DR cells ( Figure S3 ). Compared with the control group, xenograft tumor weights were significantly lower in the PCAT1 KD group ( Figures 3P, Q ). Of note, the PCAT1 KD group treated with erastin or DTX treatment showed the slowest tumor growth among all groups ( Figures 3P, Q ). In addition, IHC staining also revealed that the malignancy of tumor cells was significantly suppressed in the PCAT1 KD group ( Figure 3R ). Taken together, these results suggested that inhibition of PCAT1 impairs DTX resistance and promotes ferroptosis in vitro and in vivo.

Given the putative role of PCAT1 in DTX-resistant PCa, we evaluated the mRNA expression profiles in PC3/DR cells after KD of PCAT1 with siRNA. As shown in Figure 4A , RNA-sequencing results revealed a total of 332 genes that were differentially expressed (168 upregulated genes and 164 downregulated genes) (fold change >2, p <0.05). Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed significant functional enrichment of DNA replication, double-strand break repair, oxidation-reduction process, drug metabolism and microRNA (miRNA) in cancer ( Figures 4B, C ). To identify the key regulators of ferroptosis in DTX-resistant PCa, we detected the expression of SLC7A11 and GPX4 by qRT-PCR and Western blot assay. As shown in Figures 4D, E , the mRNA and protein levels of SLC7A11 were greatly upregulated in PC3/DR and 22RV1/DR cells, whereas no significant difference was found in GPX4 expression. To further elucidate whether PCAT1 regulates SLC7A11 expression, we detected the expression of SLC7A11 after PCAT1 KD or OE. As expected, the mRNA and protein levels of SLC7A11 were greatly elevated in PCAT1-transfected PC3 and 22RV1 cells ( Figures 4F, G ). In addition, inhibition of PCAT1 decreased the mRNA and protein levels of SLC7A11 in PC3/DR and 22RV1/DR cells ( Figures 4H–I ). IHC staining also showed that PCAT1 knockdown decreased the SLC7A11 expression in xenografts mice model ( Figure 4J ). Moreover, we also performed rescue experiments to explain why PCAT1 inhibits ferroptosis by regulating SLC7A11 expression. As shown in Figure 4K , inhibition of SLC7A11 increased the cytotoxicity of DTX and erastin in PCAT1-transfected PC3 cells. In addition, intracellular ROS levels and total iron content were also increased ( Figures 4L, M ). Collectively, the above data together indicated that PCAT1 inhibits ferroptosis by activating SLC7A11 expression in DTX-resistant PCa cells.

Previous studies have pointed out that lncRNAs regulate gene expression functions by interacting with certain cellular proteins (22). Subsequently, we investigated whether PCAT1 may also interact with specific proteins to regulate ferroptosis in DTX-resistant PCa cells. RNA pulldown and silver staining assays revealed that PCAT1 binds several specific proteins with molecular weights of about 55 kDa in PC3/DR cells ( Figure 5A ). Among these proteins, we found that c-Myc was pulled down by the biotin-labeled probe PCAT1 ( Figure 5B ). Moreover, an RNA Immunoprecipitation (RIP) assay confirmed the interaction between PCAT1 and c-Myc in PC3/DR cells ( Figure 5C ). Next, we used two bioinformatic websites (i.e., CISBP-RNA and catRAPID) to investigate which domain of c-Myc contributes to the interaction with PCAT1. The results of catRAPID fragments revealed that the 1093-1367 nucleotide (nt) positions of the PCAT1 sequence may bind to the 151-202 amino acid (aa) residues of the c-Myc protein with the top three propensities ( Figures 5D, E ). To identify which c-Myc regions interact with PCAT1 in vitro, we constructed four different deletion fragments of the c-Myc protein ( Figure 5F ). RIP assays confirmed that PCAT1 could bind to fragment 2 of c-Myc, which matched the predicted binding sites ( Figure 5G ). In addition, we also constructed a short biotin-labeled probe containing 1093–1367 nt of PCAT1, and the results confirmed that the above short probe bound to the c-Myc protein by RNA pulldown assay, indicating that 1093–1367 nt of PCAT1 interacts with 151-202 aa of the c-Myc protein ( Figure 5H ).

Previous studies reported that PCAT1 regulates c-Myc expression at the posttranscriptional level (23, 24). Consistent with the above results, we found that OE of PCAT1 promoted the protein level of c-Myc in PC3 and 22RV1 cells, while PCAT1 KD inhibited the c-Myc levels in PC3/DR and 22RV1/DR cells ( Figures 5I, J ). As expected, no significant difference was found in the mRNA level of c-Myc upon PCAT1 KD or OE (data not shown). In addition, DTX-resistant PCa cells had a higher expression of c-Myc protein than their sensitive cells ( Figure 5K ). To further identify the possible mechanism, we constructed PCAT1 plasmids with deletion of predicted binding regions with c-Myc. As shown in Figure 5L , the protein level of c-Myc was increased in PC3 cells transfected with wild-type PCAT1 but not the deletion form. To evaluate whether PCAT1 was associated with the stabilization of the c-Myc protein, we treated PC3 cells with the protein synthesis inhibitor Cycloheximide (CHX) after PCAT1 transfection. The results showed that the half-life of the c-Myc protein was significantly longer in PCAT1 wild-type cells than in control cells ( Figure 5M ). Additionally, we treated PCAT1-KD cells with the proteasome inhibitor Bortezomib (Borz), and Western blot assays showed that Borz treatment reversed the inhibition of c-Myc protein upon PCAT1 inhibition ( Figure 5N ). Collectively, these data demonstrated that PCAT1 interacts and stabilizes the c-Myc protein.

To elucidate the specific mechanism of c-Myc-mediated chemo- and ferroptotic- resistance, we performed the following experiments upon c-Myc OE or KD. As shown in Figures 6A, B , ectopic expression of c-Myc in PC3 cells reversed erastin-induced and DTX-induced cell growth inhibition. Additionally, lipid ROS production and intracellular iron levels were greatly decreased upon DTX and erastin treatment in PC3 cells after c-Myc OE ( Figures 6C, D ). In contrast, inhibition of c-Myc resensitized PC3/DR cells to erastin or DTX ( Figures 6E, F ). The lipid ROS levels and iron content were increased in c-Myc-KD PC3/DR cells under DTX and erastin exposure ( Figures 6G, H ). To explain whether c-Myc transcriptionally regulates SLC7A11 expression in PCa, we assessed the mRNA expression of SLC7A11 upon c-Myc OE or KD. qRT-PCR assays revealed that OE of c-Myc increased the SLC7A11 mRNA level, while KD of c-Myc increased the SLC7A11 mRNA level ( Figures 6I, J ). In addition, we further searched the JASPAR database and found two c-Myc binding sites (namely, Site A and Site B) within the promoter region of SLC7A11 ( Figure 6K ). Chromatin immunoprecipitation (ChIP) assays showed that c-Myc was enriched mainly at Site A of the SLC7A11 promoter ( Figure 6L ). Moreover, we constructed wild-type and mutant SLC7A11 promoter reporter plasmids. The results showed that OE of c-Myc promoted the luciferase activities of wild-type SLC7A11 but not the mutant form, indicating that SLC7A11 was transcriptionally regulated by c-Myc ( Figure 6M ). We next investigated the mechanism by which PCAT1 regulates the expression of SLC7A11. As illustrated in Figure 6N , the binding ability of c-Myc in the promoter of SLC7A11 was significantly inhibited in PC3/DR cells upon PCAT1 KD. Next, the luciferase reporter assay revealed that KD of PCAT1 impaired the transcriptional level of the SLC7A11 promoter in PC3/DR cells ( Figure 6O ). Rescue experiments revealed that inhibition of c-Myc greatly decreased the mRNA levels of SLC7A11 in PCAT1-transfected PC3 cells ( Figure 6P ). Taken together, these results confirmed that c-Myc transcriptionally activates SLC7A11 expression by interacting with PCAT1.

Previous studies have indicated that PCAT1 can function as a competing endogenous RNA (ceRNA) to regulate mRNA expression by sponging miRNAs (25). To identify the possible target miRNAs of both PCAT1 and SLC7A11, we searched three online databases (i.e., lncBase, miRDB and StarBase). The results showed that miR-25-3p and miR-302e might putatively bind to PCAT1 and SLC7A11 ( Figure 7A ). Subsequently, qRT-PCR assays showed that the expression of miR-25-3p was greatly downregulated in PC3/DR cells compared to PC3 cells, whereas no significant difference was found in the expression of miR-302e ( Figure 7B ). Western blot assays revealed that OE of miR-25-3p decreased the protein level of SLC7A11, whereas KD of miR-25-3p increased the SLC7A11 protein levels ( Figure 7C ). To further elucidate the interaction between PCAT1, SLC7A11 and miR-25-3p, we constructed PCAT1 and SLC7A11 luciferase reporter plasmids containing the wild-type and mutated miR-25-3p-binding site sequences, respectively ( Figure 7D ). As shown in Figure 7E , a significant decrease in luciferase activities was observed following the transfection of miR-25-3p and wild-type PCAT1 but not mutant PCAT1. Moreover, OE of miR-25-3p inhibited the luciferase activities after cotransfection of wild-type SLC7A11 plasmids but not the mutant form ( Figure 7F ). To further clarify whether PCAT1 regulates SLC7A11 by sponging miR-25-3p, we performed rescue experiments following transfection with PCAT1 and miR-25-3p in PC3 cells. The results showed that ectopic expression of miR-25-3p in PCAT1-overexpressing PC3 cells increased erastin-induced and DTX-induced cytotoxicity ( Figure 7G ). Additionally, the elevated protein level of SLC7A11 induced by PCAT1 in PC3 cells was inhibited after transfection with the miR-25-3p mimic ( Figure 7H ). These findings together indicated that PCAT1 regulates SLC7A11 expression by sponging miR-25-3p.

To investigate the transcription factors that were responsible for the upregulation of PCAT1, we searched the possible transcription factors within the promoter regions of PCAT1 by online bioinformatics analysis (HUMAN TFDB and ALGGEN) ( Figure 8A ). Among these databases, we found that TFAP2C was reported previously to be a ferroptotic factor (26). Then, we measured the expression level of TFAP2C in two DTX-resistant PCa cell lines and their parental cells. As shown in Figures 8B–D , the mRNA and protein levels of TFAP2C were greatly increased in PC3/DR and 22RV1/DR cells compared to PC3 and 22RV1 cells. KD of TFAP2C efficiently inhibited the levels of PCAT1 and SLC7A11 in PC3/DR and 22RV1/DR cells, while TFAP2C OE activated PCAT1 and SLC7A11 mRNA levels in PC3 and 22RV1 cells ( Figures 8E, F ). Moreover, JASPAR predicted that there are three potential binding sites in the promoter regions of PCAT1 ( Figure 8G ). Using a ChIP assay, we found the affinity of the TFAP2C protein to binding site 2 of the PCAT1 promoter compared to other binding sites (site 1 and site 3) ( Figure 8H ). To further validate the direct transcription of TFAP2C to PCAT1, we constructed wild-type and mutant luciferase reporter plasmids targeting putative binding site 2. The results showed that transfection of the TFAP2C wild-type plasmid increased the relative luciferase activity of the PCAT1 promoter, but transfection of the mutant TFAP2C binding sites did not ( Figure 8I ). In addition, KD of TFAP2C enhanced the cytotoxicity of erastin or DTX in PC3/DR and 22RV1/DR cells, whereas ectopic expression of PCAT1 in TFAP2C-deleted cells rescued the above phenotype ( Figure 8J ). Our clinical data also indicated that TFAP2C was dramatically elevated in PCa samples compared with adjacent normal tissues ( Figure 8K ). The expression of TFAP2C was positively correlated with PCAT1 levels in PCa samples ( Figure 8L ). These findings revealed that TFAP2C transcriptionally activates PCAT1 expression, thereby contributing to DTX resistance in PCa cells ( Figure 8M ).