PCa tissues and matched normal tissues were collected from Tongji Hospital, Tongji University School of Medicine. The protocols used in this research were approved by the Hospital’s Protection of Human Subjects Committee and written informed consent was provided by all patients or their legal guardians.

NE-like cells (PC3, DU145, and NCI-H660), LNCaP, C4-2 cells were purchased from ScienCell (Santiago, California, USA). All cells were maintained in RPMI-1640 media containing 10% FBS at 37°C in a 5% CO₂ incubator.

MiR-326 mimic (miR-326), mimic control (miRcont), siRNA for PCAT6 (si- PCAT6), hnRNPA2B1 (si-hnRNPA2B1), and scrambled siRNA (si-NC) were obtained from Shanghai Genechem Co., LTD (Shanghai, China). To overexpress PCAT6, human PCAT6 (NR_046326.1) cDNA was cloned into the pcDNA3.1 vector (Invitrogen). si-PCAT6, UGGCCUAGGAACCCGA ACCUGACCC; si-hnRNPA2B1, 5’-GGAACAGUUCCGUAAGCUC-3’; miR-326 mimic, 5’-GCAGGGCACGACUGAUCUUGG-3’.

Total RNA was extracted from PCa tissues cells with Trizol reagent (Takara, Japan). Special miRNA primers were applied for the reverse transcription (RT) of miRNAs. Oligo (dT) and random primers were applied for the RT of mRNA and lncRNA. qPCR was carried out by SYBR Green qPCR Master Mix (APExBIO, Houston, USA) on a Bio-Rad CFX96 system (BioRad, Hercules, CA, USA) following the manufacturer’s protocol. Related genes expression levels were normalized to GAPDH expression. The fold change was calculated through the 2^-ΔΔCt method. miR-326, Forward 5’-CCTCTGGGCCCTTCCTCCAG-3’, Reverse 5’-CTGGAGGAAGGGCCCAGAGG-3’; PCAT6, Forward 5’- CCCCTCCTTACTCTTGGACAAC-3’, Reverse 5’-GACCGAATGAGGATGG AGACAC-3’; NES, Forward 5’-GCAGAGGTCTACCACACACTCAA-3’, Reverse 5’-GGCTTCCTTCACCAGTTCCA-3’; SYP, Forward 5’-CTTTGCCATCTTCGCCTTT-3’, Reverse 5’-AGGTCACTCTTGGTCTTGTTGG-3’; CHgA, Forward 5’-TACAAGGAGATCCGGAAAGG-3’, Reverse 5’-CCATCTCCTCCTCCTCCTCT-3’; GAPDH, Forward 5’-TGAAGGGTGGAGCCAAAAG-3’, Reverse 5’-AGTCTTCTGGGTGGCAGTGAT-3’.

The expression and localization of PCAT6 were detected with FISH by RiboTM lncRNA FISH Probe Mix (Green) (RIBOBIO, Guangzhou, China). LNCaP and NCI-H660 cells (1-2×10⁴/well) were loaded on slides, fixed with 4% formaldehyde. After pretreating with protease K (2 μg/mL), glycine and acetic anhydride, the slides were pre-hybridized for 60 min and hybridized using 300 ng/mL probes (250 μL) at 42°C against PCAT6. Lastly, slides were stained with PBS with DAPI (Solarbio, Beijing, China) and photographed with a fluorescence microscope.

LNCaP and NCI-H660 cells were harvested using radioimmunoprecipitation assay (RIPA) Lysis (Elabscience, Wuhan, China), and protein concentration were detected by the BCA protein assay kit (Pierce, Rockford, IL, USA). 10 µg of protein were separated by 10% SDS-PAGE gels, followed by electrotransfer onto PVDF membranes (Arkema, Paris, France). The membrane was incubated with primary antibody against ChgA (1:1000, ab254557, Abcam), NSE (1:100, ab105389, Abcam), SYP (1:1000, ab32127, Abcam), Hnrnpa2b1 (5 µg/ml, ab31645, Abcam), and β-actin (1:1000, 3700, CST, Danvers, MA, USA) overnight at 4 °C, HRP-conjugated secondary antibodies (1:1000; ab205718; Abcam), and visualized through the ECL imaging system (Bio-Rad, Hercules, CA, USA).

A Cell Counting Kit-8 (CCK-8, Dojindo, Japan) was applied to detect cell proliferation ability as instructed by the manufacturer. The transfected LNCaP or NCI-H660 cells were plated into 24-well plates (6×10⁴ cells/well) for 24, 48, and 72 h. 10 μL CCK-8 solution was added into per well followed by incubation for 2 hours. Lastly, the absorbance of the resulting solution was tested at 450 nm with a microplate reader (REAGEN, USA).

Transwell chamber (24-well) with Matrigel was applied to determine cell invasion following the manufacturer’s instructions. Briefly, 200 μl transfected LNCaP or NCI-H660 cells (3.0×10⁵) were added in the upper chamber with RPMI-1640 medium without serum. 500 μl RPMI-1640 medium contained 10% FBS was added to the lower chamber. After incubating for one day at 37 °C with 5% CO₂, the cells on the lower surface of membranes were fixed with absolute ethanol, stained by 0.1% crystal violet for 15 min. The invasive cell number was counted using a Zeiss Microscope (Oberkohen, Baden-Wurttemberg, Germany).

After centrifuging at 12,000 g at 4 °C for 12 minutes, the LNCaP cell supernatant was harvested. Then, RIP buffer containing biotin-labeled PCAT6 was mixed with cell lysates for 60 minutes. Next, streptavidin beads were added to each binding reaction followed by incubation for 60 minutes. Lastly, the protein in the elution complex was determined with western blot assay.

The wildtype (WT) or mutant (Mut) PCAT6 promoters were constructed into PGL3-basic plasmid (YouBia, Hunan, China). LNCaP and NCI-H660 cells were seeded into a 24-well plate followed by transfection with PGL3-Luc supplemented with WT or Mut PCAT6 promoters with Lipofectamine (Invitrogen) and pRL-TK was applied as control. Luciferase activity was examined using the Dual-Luciferase^® Reporter Assay system (Promega Corporation, Madison, WI, USA) as the manufacturer’s protocol.

BALB/c nude mice (male; 4-6 w; 18-22 g) were selected from Shanghai Model Organisms Center, Inc (China). The animal experiment was approved by the Ethics Committee of the Tongji Hospital, Tongji University of Medicine. The mice were injected with the PCAT6 lentivirus vector. Briefly, LNCaP cells were infected with lentiviral vectors with a titer of 2*10⁸ plaque-forming units (PFUs)/mL. After obtaining stably transfected cell lines, cell suspensions (5*10⁷ cells/mL) were injected intravenously into the nude mice’s tail vein. The tumor growth was measured at specific times (7th, 14th, 28th, and 35th) after inoculation.

Statistical analysis was carried out with the SPSS 17.0 software (IBM-SPSS, Inc., Chicago, IL, USA). Each test process was independently repeated 3 times. Results are presented as mean ± SD. Comparison of two groups was identified by t-test analysis, or one-way analysis of variance (ANOVA) followed by the Scheffé test. A P-value of < 0.05 was considered statistically significant.

Luo et al., has found that the expression of lncRNAs is dys-regulated in NEPC samples and Enzalutamide enhances lncRNA-p21 expression to facilitate NED of PCa ( Supporting Figure S1A and Supporting Table S1 ) (20). Here the top 10 lncRNAs significantly dys-regulated in NEPC samples were reassessed in androgen-sensitive LNCaP cells and NE-like NCI-H660 cells. As shown in Figure 1A , PCAT6 was identified as the most highly expressed lncRNA except lncRNA-p21 in NCI-H660 cells compared with LNCaP cells. The PCAT6 level in NE-like cells (PC3, DU145, and NCI-H660) was commonly increased compared with LNCaP as well as C4-2 cells ( Figure 1B ), indicating the potential effect of PCAT6 on NED. We then investigated the effect of treatment time and concentration of Enzalutamide on PCAT6 expression in LNCaP cells. Figures 1C, D revealed that Enzalutamide treatment significantly enhanced PCAT6 level in a dose-dependent and time-dependent fashion. The results from RNA-FISH also indicated that PCAT6 expression was higher in NCI-H660 cells when compared with LNCaP cells ( Figure 1E ), and Enzalutamide treatment upregulated PCAT6 level in LNCaP cells ( Figure 1F ). We then assessed PCAT6 level in human PCa tissues and paired normal tissues. Figure 1G revealed that PCAT6 expression was markedly increased in most tumor tissues compared to controls. Furthermore, the PCAT6 level was markedly upregulated in NEPC tissues compared with CRPC tissues ( Figure 1H ), indicating that upregulated PCAT6 was correlated with PCa NED.

To explore the biological function of PCAT6 on PCa cells, PCAT6 was overexpressed in LNCaP cells or inhibited in NCI-H660 cells and then NED, proliferation, invasion of PCa cells were assessed. Forced expression of PCAT6 remarkably enhanced mRNA and protein levels of NE markers (NSE, SYP, and ChgA) ( Figures 2A–C ). On the contrary, PCAT6 knockdown significantly repressed the expression of NE markers ( Figure 2D ), indicating that PCAT6 positively regulated NED of PCa cells. Additionally, the effect of PCAT6 in regulating PCa cell proliferation and invasion was assessed. Figures 2E, F showed that PCAT6 overexpression promoted LNCaP cell proliferation, whereas PCAT6 knockdown repressed NCI-H660 cell proliferation. As expected, PCAT6 overexpression facilitated LNCaP cell invasion, whereas PCAT6 knockdown significantly suppressed NCI-H660 cell invasion ( Figures 2G–J ).

Then, we investigated the role of PCAT6 in tumor growth and metastasis in vivo. To construct the murine metastatic PCa model in vivo, PCAT6-overexpressed LNCaP cells stably expressed luciferase were injected intravenously into tail vein of male nude mice and the bioluminescent signal was assessed at 10 weeks following injection. PCAT6 overexpression enhanced the bioluminescent signals, indicating the role of PCAT6 in promoting PCa metastasis ( Figures 3A, B ). To assess the effect of PCAT6 on tumor growth, PCAT6-overexpressed LNCaP cells were injected subcutaneously into nude mice and tumor growth was surveyed. Figures 3C, D showed that PCAT6 overexpression significantly promoted tumor growth at different time points. More importantly, the mRNA and protein expressions of NE markers in PCAT6-overexpressed groups were markedly upregulated compared to the control group ( Figures 3E, F ).

Mounting evidence demonstrates that lncRNAs commonly function as ceRNA to absorb miRNAs, thus upregulate mRNA expression when lncRNAs are expressed in the cytoplasm (21, 22). The results from RNA-FISH indicated that PCAT6 was mainly located in the cytoplasm of PCa cells ( Figures 1E, F ), indicating that PCAT6 might act as a ceRNA. To identify the potential miRNAs sponged by PCAT6, the Starbase v2.0 database (http://starbase.sysu.edu.cn/) was applied to forecast potential miRNA binding sites in PCAT6 (23). The results from the bioinformatic analysis revealed that 15 miRNAs could be sponged by PCAT6 ( Supporting Table S2 ). Among 15 miRNAs, miR-326 was selected for further analysis because miR-326 was verified as a biomarker for prostate cancer diagnosis ( Figure 4A ) (24, 25). Interestingly, the miR-326 level in NE-like cells was reduced compared with LNCaP and C4-2 cells ( Figure 4B ). To assess the direct combination of PCAT6 and miR-326, the recombinant plasmids of pGL3-PCAT6-WT and pGL3-PCAT6-Mut were constructed, co-transfected with miR-326 into LNCaP cells, respectively ( Figure 4C ). The results from the luciferase test revealed that pGL3-PCAT6-WT luciferase activity was markedly repressed by co-transfection with miR-326 compared with miRcont, whereas miR-326 did not affect the luciferase activity of pGL3-PCAT6-Mut ( Figure 4D ). Furthermore, miR-326 mutant lost its suppressor role in regulating luciferase activity of pGL3-PCAT6-WT ( Figure 4E ). RNA pull-down was performed to further validate the direct combination of PCAT6 and miR-326. As shown in Figure 4F , PCAT6 was more enriched in the miR-326 compared with miRcont. miR-326 was lowly expressed in small cell carcinoma tissues compared to CRPC tissues ( Figure 4G ). These data suggest that PCAT6 functioned as a tumor activator in PCa progression, at least in part by sponging miR-326.

We next identified the target gene of miR-326 in PCa cells. To this end, four bioinformatics algorithms (miRanda, miRmap, picTar, and TargetScan) were used to predict the commonly potential targets. The results from bioinformatics analysis revealed that 37 genes were the common target genes of miR-326 ( Figure 5A and Supporting Table S3 ). Among them, hnRNPA2B1 was identified as the most significant target through luciferase activity assay ( Figure 5B ). To verify the statement, the recombinant plasmids of pGL3-hnRNPA2B1-3’UTR-WT and its mutant (pGL3-hnRNPA2B1-3’UTR-Mut) was constructed through inserting hnRNPA2B1-3’UTR cDNA or its mutant into pGL3 vector ( Figure 5C ). The results from luciferase test revealed that pGL3-hnRNPA2B1-3’UTR-WT luciferase activity was markedly suppressed by co-transfection with miR-326 compared with miRcont, whereas miR-326 did not affect pGL3-hnRNPA2B1-3’UTR-Mut luciferase activity ( Figure 5D ). These data suggested that miR-326 could straightway bind hnRNPA2B1-3’UTR by their own miRNA binding sites. Furthermore, miR-326 overexpression markedly repressed the protein level of hnRNPA2B1 in NCI-H660 cells ( Figure 5E ). The regulatory correlations of PCAT6, miR-326, and Hnrnpa2b1 were next assessed. Figure 5F revealed that pGL3-hnRNPA2B1-3’UTR luciferase activity was suppressed due to miR-326 overexpression, but the effect was remitted by transfecting pGL3-PCAT6-WT. pGL3-PCAT6-Mut did not affect the potential of miR-326 to repress the luciferase activity of pGL3-hnRNPA2B1-3’UTR, showing that PCAT6 competed with hnRNPA2B1 to binding to miR-326 and that PCAT6 functioned as an endogenous “sponge” by absorbing miR-326. Given that PCAT6 sponged miR-326 and miR-326 directly repressed hnRNPA2B1, we speculated that PCAT6 might enhance hnRNPA2B1 expression in miR-326-dependent manner. As expected, PCAT6 enhanced the mRNA and protein level of hnRNPA2B1 in NCI-H660 cells, whereas miR-326 overexpression markedly repressed the effect ( Figures 5G–I ).

Finally, we examined the role of PCAT6/miR-326/hnRNPA2B1 axis in regulating PCa cells proliferation, invasion, and NED. As shown in Figures 6A–C , PCAT6 enhanced the mRNA and protein level of NE markers in LNCaP cells compared to control group, whereas both miR-326 overexpression and hnRNPA2B1 knockdown significantly reversed the effect. PCAT6 promoted LNCaP cells proliferation and invasion, whereas both miR-326 overexpression and hnRNPA2B1 knockdown significantly repressed the effect of PCAT6 on pro-proliferation and pro-invasion ( Figures 6D–F ). All results indicate that PCAT6 promoted Enzalutamide-induced PCa NED by regulating miR-326/hnRNPA2B1 axis.