Activation of the KDM5A/miRNA-495/YTHDF2/m6A-MOB3B axis facilitates prostate cancer progression

The current study was approved by the Ethics Committee of Harbin Medical University Cancer Hospital and performed in strict accordance with the Declaration of Helsinki. All participants signed informed consent prior to enrollment. Extensive efforts were made to ensure minimal usage of animals as well as their suffering.

A total of 78 patients who received radical prostatectomy and transurethral resection of the prostate at the Department of Urology of Harbin Medical University Cancer Hospital from June 2014 to June 2016 were recruited in our study. None of these patients received anti-tumor treatment before surgery. The cancer tissue samples without necrosis or hemorrhage were biopsied during surgery with adjacent normal tissues (no cancer cells confirmed by pathological examination) and stored in a − 80 °C freezer. The demographic information of patients was extracted and collected from the medical record system, and all patients were followed up post-surgery in order to understand the clinical outcomes after treatment and obtain comprehensive clinical data of patients. The follow-up ended in December 2019, with a total of 3–36 months. Kaplan-Meier method was used to analyze the correlation between KDM5A expression and total survival (OS) and progression-free survival (DFS).

Paraffin sections of tumor tissues from each group were taken for immunohistochemical analysis, which were dewaxed, dehydrated by alcohol gradient, and immersed in 3% methanol H₂O₂ for 20 min. Antigen was repaired through water-bath of repair solution. Sections were then blocked with normal goat serum blocking solution (C-0005, Haoran Bio, Shanghai, China) and placed at room temperature for 20 min to dry the slides. Next, sections were incubated with primary antibody to KDM5A (ab92533, 1:500) overnight at 4 °C. The secondary antibody goat anti-rabbit immunoglobulin G (IgG) was then added into the sections and placed at 37 °C for 20 min. After addition of horseradish peroxidase labeled Streptomyces ovalbumin working solution (0343-10,000 U, ImunBio, Beijing, China) the sections were placed at 37 °C for 20 min. Color was developed with diaminobenzidine (DAB; ST033, Whiga Biotechnology, Guangzhou, China) and sections were counterstained with hematoxylin (PT001, Bogoo, Shanghai) for 1 min, turned blue with 1% ammonia water and dehydrated by gradient alcohol. After regular permeabilization and mounting, sections were observed under a microscope with each section randomly selected under five high power field of view. A total of 100 cells were counted in each field. Positive cells < 10% was taken for negative, while positive cells ≥10% and < 50% for positive, and positive cells > 50% was considered as strong positive.

The total RNA was extracted from cancer tissues and cells using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA). The RNA was then reversely transcribed into complementary DNA using the TaqMan MicroRNA Assays Reverse Transcription primer (4,427,975, Applied Biosystems, USA)/PrimeScript RT reagent Kit (RR047A, Takara, Japan). The primers for KDM5A, miR-495, YTHDF2 and MOB3B were designed and synthesized by Takara Holdings Inc. (Kyoto, Japan) (Table 1). Subsequently, reverse transcription quantitative polymerase chain reaction (RT-qPCR) was conducted on an ABI 7500 instrument (Applied Biosystems, Foster City, CA, USA). The fold changes were calculated using the 2^-ΔΔCt method with U6 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) serving as internal references, respectively.

Total protein was extracted from cancer cells and tissues using radio-immunoprecipitation assay lysis buffer, with the protein concentration then determined with a bicinchoninic acid kit. The protein was separated and transferred onto a polyvinylidene fluoride membrane. The membrane was then blocked and underwent overnight incubation with primary antibodies against KDM5A (1:10000), H3K4me3 (1 μg/mL, rabbit, Abcam, USA), YTHDF2 (1:2000, rabbit) and MOB3B (1:1000, rabbit, Sigma-Aldrich Chemical Company, St Louis, MO, USA). The next day, the membrane was re-probed with horseradish peroxidase-labeled secondary goat anti-rabbit IgG (1:1000, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Afterwards, the membrane was visualized and the protein bands were quantified using the Bio-Rad ChemiDoc™ system and ImageJ2x software. The ratio of the gray value of the target band to GAPDH (1:10000, rabbit, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) was representative of the relative protein expression.

PCa cell lines (LNCaP, C42, DU145 and PC3) were purchased from American Type Culture Collection (ATCC; Manassas, VA, USA). The normal human prostate epithelial cell line RWPE-1 (Oulu Biotechnology, Shanghai, China) served as a control, and were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, USA) containing 10% fetal bovine serum (Gibco, USA), 100 μg/mL streptomycin and 100 U/mL penicillin in a 5% CO₂ incubator (Thermo Fisher Scientific Inc., Waltham, MA, USA) at 37 °C. Upon reaching approximately 75% confluence, the cells were transiently transfected using Lipofectamine 2000 (Invitrogen) with the following plasmids purchased from Sino Biological, Inc. (Beijing, China): overexpression (oe)-KDM5A, short hairpin RNA (sh)-KDM5A (sh1-KDM5A [5′-GGAACUGGGUCUCUUUUGA-3′], sh2-KDM5A [5′-GCAAAUGAGACAACGGAAA-3′] and sh3-KDM5A [5′-UGACAAUGGUGGACCGCAU-3′]), oe-KDM5A + miR-495 mimic, miR-495 mimic, miR-495 inhibitor, miR-495 mimic + oe-YTHDF2 and oe-KDM5A + oe-MOB3B as well as their corresponding controls (oe-negative control [NC], sh-NC, oe-NC + mimic NC, oe-KDM5A + mimic NC, mimic NC, inhibitor NC, mimic NC + oe-NC, miR-495 mimic + oe-NC, mimic NC + oe-YTHDF2 and oe-KDM5A + oe-NC). After 6 h of transfection, the cells were cultured for 48 h and then collected.

PCa cells were inoculated into dishes containing 10 mL preheated culture medium, and evenly dispersed by gentle rotation, followed by culture, with the culture medium changed once every 2–3 days. The culture was halted once the clone in the dish was visible to the naked eyes. After removal of supernatant, the cells were washed, and fixed. The fixed cells were stained with an appropriate amount of GIMSA (Invitrogen, USA) for 10–30 min. The number of cell clones was counted under an inverted microscope (Leica DMi8-M, Co. Ltd., Solms, Germany), and the colony formation rate was calculated using the formula: the number of cell clones/the number of inoculated cells × 100%.

Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) double staining method was used to detect cell apoptosis. The PCa cells were inoculated into 6-well plates at a density of 2 × 10⁵ cells/well. After cell transfection, the culture medium was removed, and the cells were trypsinized and collected in a centrifuge tube for centrifugation at 800 g, with the supernatant discarded. According to the instructions of the Annexin V-FITC Apoptosis Detection Kit (BD Biosciences, San Jose, CA, USA), the cells were resuspended, and added with 5 μL FITC and 5 μL PI under dark conditions. At last, the cell apoptosis was detected using BD FACSCalibur.

PCa cells were starved, then digested, and resuspended in serum-free Opti-MEMI medium (Invitrogen, USA). The cell suspension was then seeded into the 8-μm Transwell chamber (Corning, NY, USA) in a 24-well plate (100 μL per chamber, a total of 3 chambers). Next, 600 μL 10% RPMI-1640 medium was added to the lower chamber and incubated. For the cell migration experiment, the cells were fixed, after which the chamber was treated with 0.2% Triton X-100 solution (Sigma-Aldrich Chemical Company, St Louis, MO, USA), and stained with crystal violet.

During cell invasion experiment, 50 μL Matrigel (Sigma-Aldrich Chemical Company, St Louis, MO, USA) settled in the chamber before the experiment. After 48 h, fixation and staining were conducted with the aforementioned procedures. The number of stained cells was counted in five randomly selected visual fields under an inverted microscope (Leica DMi8-M, Co. Ltd., Solms, Germany) to obtain the mean value.

After 48 h of transfection, PCa cells were seeded into 6-well plates at a density of 5 × 10⁵ cells/well. After the cells completely adhered to the bottom, a scratch was made in the middle of each well using a 2-mm cell scraper and the cell culture was continued for 24 h. The cells were photographed at 0 h and 24 h after scratch, and the rate of scratch healing was calculated with the Image-Pro Plus 6.0 software (Media Cybernetics, Inc., Silver Spring, MD, USA).

The predicted binding site and mutation fragments of miR-495 promoter with KDM5A were inserted into luciferase reporter vectors (Beijing Huayueyang Biotechnology Co., Ltd., Beijing, China), known as reporter plasmids miR-495 promoter-wild-type (WT) (with sequence of 3′-CAGTGACCCA-5′) and miR-495 promoter-mutant (MUT) (with sequence of 3′-TGAGAGTATG-5′). Oe-NC and oe-KDM5A were co-transfected into 293 T cells (Oulu Biotechnology, China) with miR-495-WT and miR-495-MUT plasmids in order to detect whether KDM5A could bind with KDM5A. After 48 h of transfection, cells were collected and lysed. The luciferase activity was detected with the use of a luciferase assay kit (K801–200, Biovision, Bay Area, San Francisco, USA) on the analysis system (Promega Corporation, Madison, WI, USA). With renilla luciferase as a loading control, the luciferase activity was calculated as the relative luciferase unit (RLU) activity of firefly luciferase/RLU activity of renilla luciferase. Likewise, the interaction between miR-495 and YTHDF2 was detected in the same way.

After 48 h of transfection, PCa cells from each group were fixed in 1% formaldehyde at 37 °C for 10 min, which was then terminated by adding glycine solution for a 5 min reaction on ice. After rinsing with PBS, the cells were incubated and centrifuged to obtain cell precipitate. The cells were resuspended in 200 μL SDS lysis buffer for 10 min on ice. The chromatin DNA was broken by ultrasound on ice. The cells were centrifuged at 14000 rpm for 10 min, after which the supernatant was diluted with ChIP buffer containing protease inhibitor, and with blocking solution added for a 30-min incubation at 4 °C. After centrifugation at 1000 rpm for 1 min, the supernatant was collected and divided into three parts, with one as Input, one being incubated with rabbit IgG antibody (Abcam, China) as the NC, and the other part incubated with rabbit KDM5A antibody (2 μg used for 25 μg of chromatin, Abcam, China) at 4 °C overnight. We added Protein G Dynabeads (Thermo Fisher Scientific Inc., Waltham, MA, USA) to the supernatant and incubated it at 4 °C for 1 h to precipitate antibody-transcription factor complex. After centrifugation at 1000 rpm for 1 min, the precipitate was washed and eluted with elution buffer. Twenty μL NaCl (5 mol/L) were added into the eluted precipitate and Input DNA respectively and bathed in water at 65 °C for 4 h to ravel the DNA cross-linking. DNA fragments were then purified and recovered post-incubation with protease K. Subsequently, RT-qPCR was conducted to detect the expression of miR-495 with the recovered DNA as a template.

Total RNA was extracted from tissues and cells using the TRIzol method. The mRNA in the total RNA was isolated and purified using the PolyATtract® mRNA Isolation Systems (A-Z5300, A&D Technology Corporation, Beijing, China). IP buffer supplemented with 20 mM Tris pH 7.5, 140 mM NaCl, 1% NP-40 and 2 mM EDTA was added m6A antibody (1:500, ab151230, Abcam, USA) or IgG antibody (ab109489, 1:100, Abcam, USA) and incubated with protein A/G magnetic beads for 1 h in order for binding. IP buffer with ribonuclease inhibitor and protease inhibitor was then added with the purified mRNA-bead-antibody complex and incubated overnight at 4 °C. The RNA was eluted with elution buffer, and purified with phenol-chloroform. MOB3B was analyzed by RT-qPCR.

PCa tissues and cells were incubated with 200 mm 4-thiopyridine (Sigma, USA) for 14 h, and cross-linked with 0.4 J/cm² at 365 nm. After cleavage, YTHDF2 antibody (5 mg and 3 mg respectively) was used for immunoprecipitation at 4 °C. The precipitated RNA was then labeled with [g-32-P]-ATP and observed by autoradiography. After protease K treatment, the RNA was extracted and MOB3B expression was detected by RT-qPCR.

A total of 36 specific pathogen-free male Balb/c mice (aged 5 weeks; weighing 18–22 g) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China). PCa cells were resuspended in serum-free RPMI-1640 medium (Gibco, USA) to cell suspension at a density of 1 × 10⁶ cells/200 μL. The Balb/c nude mice were randomly divided into three groups (n = 12 in each group), and the tumorigenesis experiment lasted for 4 weeks. Under anesthesia with ether, the nude mice were disinfected and subcutaneously inoculated with cells transfected with oe-NC, oe-KDM5A + oe-NC and oe-KDM5A + oe-MOB3B at a density of 1 × 10⁶ cells/mouse (200 μL) at the back of the right hind leg. All nude mice were raised in the same environment, with themselves and their inoculation site being observed every day. The tumor size was measured every 7 days, and the length and width of the tumor were recorded, with the tumor volume estimated as length × width²/2. After 4 weeks, all mice were euthanized by cervical dislocation under anesthesia, and their tumor tissue was dissected, photographed and weighed.

All data were processed using the SPSS 21.0 (IBM Corp. Armonk, NY, USA) with a level of p < 0.05 as statistical significance. Continuous data were demonstrated as mean ± standard deviation. Data with normal distribution and homogeneity of variance between two groups were compared using t-test. Comparison of the mean among multiple groups were analyzed by one-way analysis of variance (ANOVA) with Tukey’s post hoc test or repeated measures ANOVA with Bonferroni post hoc test. Kaplan-Meier method was used to calculate the survival rate and duration of the patients, with Log-rank test used for single factor analysis. Pearson correlation coefficient was utilized for correlation analysis.

We first explored the possible role of KDM5A in the development of PCa. RT-qPCR, Western blots and immunohistochemistry revealed that the mRNA and protein expression levels of KDM5A were increased in cancer tissues compared to the adjacent normal tissues collected from the 78 PCa patients (p < 0.05; Fig. 1a, b). Next, KDM5A expression was grouped into high expression (> 2.593) and low expression (≤ 2.593), based on the mean expression value of KDM5A. Kaplan-Meier analysis showed lower OS and DFS of PCa patients with high KDM5A expression than the OS and DFS of patients with low KDM5A expression (Fig. 1c), suggesting that the high expression of KDM5A may correlate with poor prognosis of PCa patients. In addition, mRNA and protein levels of KDM5A were elevated in LNCaP, C42, DU145 and PC3 cells compared with RWPE-1 cells, with PC3 cells showing a more increased KDM5A expression (p < 0.05; Fig. 1d, e). Thus, PC3 cells were selected for the subsequent experiments. These results indicate that highly expressed KDM5A in PCa is related to a poor prognosis of patients with PCa.

We consequently became interested in the cell functions of PCa cells and the effect of KDM5A on the proliferation, migration, invasion and apoptosis of PC3 cells. The results of RT-qPCR and Western blots revealed a reduction in the mRNA and protein expression levels of KDM5A in cells transfected with sh1-KDM5A, sh2-KDM5A and sh3-KDM5A (p < 0.05), among which sh1-KDM5A showed a superior silencing efficiency (p < 0.05; Fig. 2a, b) and was therefore selected for subsequent experiments. Additionally, the mRNA and protein levels of KDM5A were significantly increased in cells following oe-KDM5A transfection (p < 0.05; Fig. 2c, d). Silencing of KDM5A was found to reduce PC3 cell proliferation, invasion and migration while inducing apoptosis (p < 0.05; Fig. 2e-h), which was undermined by overexpression of KDM5A. Collectively, upregulated KDM5A could augment the proliferation, migration and invasion of PCa cells, while diminishing the cell apoptosis.

We proceeded to examine the specific mechanism by which KDM5A regulates biological and physiological functions of PCa cells. ChIP-seq analysis showed that KDM5A could bind to the promoter of miR-495, suggesting that KDM5A might play a role in the development of PCa by targeting miR-495. miR-495 was detected to be decreased in PCa tissues using RT-qPCR (p < 0.05; Fig. 3a), and was negatively correlated with the mRNA expression level of KDM5A (Fig. 3b). In addition, miR-495 expression level was lower in LNCaP, C42, DU145 and PC3 cells than in RWPE-1 cells (p < 0.05; Fig. 3c). Dual luciferase reporter assay showed that the luciferase activity of miR-495 promoter-WT was decreased following oe-KDM5A transfection (p < 0.05) while no statistical changes were observed in luciferase activity of miR-495 promoter-MUT (p > 0.05; Fig. 3d), indicating that KDM5A could bind to the miR-495 promoter. Moreover, miR-495 expression was reduced in cells transfected with oe-KDM5A (p < 0.05; Fig. 3e). ChIP assay also confirmed the binding of KDM5A to miR-495 promoter region since the binding of miR-495 to KDM5A protein was increased following the overexpression of KDM5A (p < 0.05; Fig. 3f). Western blots further demonstrated that H3K4me3 expression in the antibody-transcription factor complex pulled down by KDM5A antibody was decreased (p < 0.05; Fig. 3g). These data indicate that KDM5A, as a H3K4me3 demethylase, can bind to the promoter region of miR-495 to suppress its transcription, and then decrease its expression in human PC3 cells.

In order to verify whether KDM5A functions in the occurrence and development of PCa by regulating miR-495, we performed a series of in vitro experiments. The results of RT-qPCR, Western blot analysis (Fig. 3h-j), colony formation assay (Fig. 3k), flow cytometry (Fig. 3l), Transwell assay (Fig. 3m) and Scratch test (Fig. 3n) and showed no significant changes in the mRNA and protein expression of KDM5A, yet increased miR-495 expression and cell apoptosis, reduced cell proliferation, invasion and migration in PC3 cells overexpressing miR-495. The transfection with oe-KDM5A in cells, however, resulted in an increase in the mRNA and protein expression of KDM5A as well as cell proliferation, migration and invasion, in addition to a reduction in miR-495 expression and cell apoptosis (p < 0.05). Co-transfection with oe-KDM5A and miR-495 mimic found no statistical alterations in KDM5A expression, elevated miR-495 expression, decreased cell proliferation, migration and invasion yet enhanced cell apoptosis (p < 0.05). Taken together, KDM5A could promote the proliferation, invasion and migration of PCa cells and reduces cell apoptosis by downregulating the expression of miR-495.

Having identified the role of KDM5A in promoting PCa cell proliferation, invasion and migration through inhibition of miR-495, we moved our attention on the underlying mechanism of miR-495 in PCa. Bioinformatics analysis predicted the binding sites between miR-495 and YTHDF2 (Fig. 4a). RT-qPCR detected that the mRNA expression of YTHDF2 was increased (p < 0.05; Fig. 4b), while miR-495 expression was reduced in patient cancer tissues (p < 0.05; Fig. 4c). Moreover, YTHDF2 protein expression was also increased in patient PCa tissues (p < 0.05; Fig. 4d). Based on the above results, we speculated that miR-495 might be involved in the pathogenesis of PCa by targeting YTHDF2. The results of dual luciferase reporter showed a decline in the luciferase activity of YTHDF2-MUT (p < 0.05) while YTHDF2-MUT luciferase activity was unaffected following miR-495 mimic transfection (p > 0.05; Fig. 4e), indicating that miR-495 could combine with YTHDF2 mRNA. RT-qPCR and Western blots revealed upregulated miR-495 expression (p < 0.05; Fig. 4f) and downregulated YTHDF2 mRNA and protein expression (p < 0.05; Fig. 4g, h) in miR-495 mimic-transfected cells, which was reversed by transfection with miR-495 inhibitor (p < 0.05; Fig. 4i-k). These results indicate that miR-495 can target YTHDF2 and inhibit its expression in PCa cells.

We then investigated whether miR-495 can inhibit the proliferation, invasion and migration of PCa cells by targeting YTHDF2. Overexpression of YTHDF2 alone led to no statistical changes in the expression of miR-495 (p > 0.05; Fig. 5a), but increased mRNA and protein expression of YTHDF2 (p < 0.05; Fig. 5b, c), enhanced cell proliferation (p < 0.05; Fig. 5d), reduced cell apoptosis (p < 0.05; Fig. 5e), and facilitated cell migration and invasion (p < 0.05; Fig. 5f, g) and. By contrast, miR-495 mimic-transfected cells exhibited miR-495 overexpression, decreased mRNA and protein expression of YTHDF2, diminished cell proliferation, migration and invasion yet augmented apoptosis. In addition, transfection with both miR-495 mimic and oe-YTHDF2 presented with similar results as those following oe-YTHDF2 transfection alone. In summary, overexpression of YTHDF2 can undermine the inhibitory effect of miR-495 on the proliferation, invasion and migration of PCa cells, and the promoting effect in cell apoptosis.

We further elaborate the mechanism of YTHDF2 in PCa. RT-qPCR showed that MOB3B mRNA expression level was decreased in PCa tissues (p < 0.05; Fig. 6a), while YTHDF2 mRNA expression was elevated (p < 0.05; Fig. 6b). Western blots also detected a decline in the protein expression of MOB3B in PCa tissues (p < 0.05; Fig. 6c). In addition, the m6A modification level of MOB3B was reduced in PCa tissues (p < 0.05; Fig. 6d). The PAR-CLIP revealed that mRNA expression of MOB3B binding to YTHDF2 was increased in PCa tissues (p < 0.05; Fig. 6e). As shown in Fig. 6f, g, cells transfected with oe-YTHDF2 exhibited downregulated mRNA and protein expression of MOB3B (p < 0.05), decreased m6A modification level of MOB3B (p < 0.05; Fig. 6h), and significantly more binding of YTHDF2 to the MOB3B mRNA (p < 0.05; Fig. 6i). These results indicate that YTHDF2 can recognize the m6A modification of MOB3B mRNA, promote the degradation of mRNA and inhibit MOB3B expression.

We extended our mechanistic findings to determine whether KDM5A regulates the expression of MOB3B via the miR-495/YTHDF2 axis to promote the proliferation, migration and invasion of the PCa cells. We analyzed the correlation between KDM5A and MOB3B expression in PCa tissues from 78 PCa patients, and the results showed that there was a significant negative correlation between KDM5A and MOB3B (p < 0.05) (Fig. 7a). In PC3 cells transfected with oe-KDM5A, the mRNA and protein levels of KDM5A and YTHDF2 were increased (p < 0.05; Fig. 7b, c), while miR-495 expression and MOB3B mRNA and protein expression levels were decreased (p < 0.05; Fig. 7d). We observed the m6A modification level of MOB3B mRNA was diminished (p < 0.05; Fig. 7e), cell proliferation was enhanced (p < 0.05; Fig. 7f), cell apoptosis was reduced (p < 0.05; Fig. 7g), cell invasion and migration was facilitated (p < 0.05; Fig. 7h, i) in response to oe-KDM5A transfection. Co-transfection with oe-KDM5A and oe-MOB3B resulted in no alterations in expression of KDM5A, miR-495 and YTHDF2 (p > 0.05), but increased MOB3B mRNA and protein expression as well as m6A modification level of MOB3B mRNA, enhanced cell apoptosis, while attenuating cell proliferation, invasion and migration (p < 0.05). The above results illustrate that KDM5A can upregulate YTHDF2 expression by inhibiting miR-495, thus downregulating MOB3B expression, whereby subsequently stimulating the proliferation, migration and invasion of the PCa cells and reducing cell apoptosis.

We carried out xenograft tumor experiments on the nude mice in order to verify the effect of KDM5A on PCa initiation and progression via the miR-495/YTHDF2/MOB3B axis in vivo. The mice treated with oe-KDM5A presented with increased tumor size and weight (p < 0.05; Fig. 8a-c), upregulated mRNA and protein expression of KDM5A and YTHDF2 (p < 0.05; Fig. 8d, e), and decreased miR-495 expression (p < 0.05; Fig. 8f) and MOB3B mRNA and protein expression as well as diminished m6A modification level of MOB3B mRNA (p < 0.05; Fig. 8g). Treatment with oe-KDM5A and oe-MOB3B, however, resulted in decreased tumor size and weight (p < 0.05), increased MOB3B mRNA and protein expression and m6A modification level of MOB3B mRNA (p < 0.05) while no statistical changes were observed in miR-495 expression, mRNA and protein expression of KDM5A and YTHDF2 (p > 0.05). These data indicate that KDM5A represses MOB3B and facilitates tumorigenicity in nude mice via the miR-495/YTHDF2 axis.

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.