Background
Prostate cancer (PCa) is one of the most common malignancies and the second leading cause of cancer-related death in men [
1,
2]. Although organ-confined PCa can be effectively treated by radical prostatectomy or radiation therapies, androgen deprivation therapy (ADT) is first-line
treatment for metastatic PCa. Once hormonal resistance occurs, PCa progresses rapidly, and advanced PCa is usually fatal within 18 months [
3,
4]. Currently, several compounds, including abiraterone acetate [
5], enzalutamide [
6], sipuleucel-T [
7], alpharadin [
8], and docetaxel [
9] have been used to help treat PCa. Unfortunately, adverse side effects of the treatment and drug resistance often lead to treatment failure [
10]. Therefore, there is an urgent need to further understand the molecular mechanism involved in prostate carcinogenesis and drug resistance.
Although the molecular mechanisms driving prostate carcinogenesis are complex, the dysregulation of cell proliferation is a fundamental feature of all types of cancer. Cell proliferation is coupled with cell cycle progression, and mammalian CDKs are essential for driving each cell cycle phase. Accumulating evidence has suggested that tumor-associated cell cycle disorders are often mediated by alterations in cyclin-dependent kinase (CDK) activity. Mis-regulated CDKs induce unscheduled proliferation [
11]. It has been well known that mammalian cells contain at least 13 CDKs [
11]. Of these, CDK1-CDK6, CDK10 and CDK11 are all involved in cell cycle control [
11‐
13]. CDK7,CDK8 and CDK9 have activities that are different from cell cycle control, these 3 CDKs can phosphorylate the carboxyl-terminal domain (CTD) of RNA polymerase II and exert actions in transcriptional regulation [
12‐
14]. CDK12 and CDK13 bind to L-type cyclins (CycL) and regulate alternative RNA splicing [
15,
16]. A recent study shows that knocking out CDK13 leads to abnormal expression of several genes involved in a variety of biological processes including cell growth regulation [
17]. Notably, both CDK12 and CDK13 knockdown affect the expression of genes involved in RNA processing, but CDK13-regulated gene sets are not affected by CDK12 knockdown. These evidences clearly suggest that human CDK functions do not overlap with each other, probably reflecting tissue-specific and fine-tuned regulation of cell cycle regulation. Importantly, several recent studies reported that CDK12 expression is dysregulated in metastatic castration-resistant prostate cancer (mCRPC) samples, and CDK12 loss results in highly recurrent gains at loci of genes involved in the cell cycle and DNA replication [
18‐
20]. However, much less is known regarding CDK13 expression and function in PCa.
Circular RNAs (circRNAs) are a novel class of non-coding RNA characterized by the presence of a covalent bond linking the 3′ and 5′ ends generated by back-splicing [
21]. Emerging evidences have shown that circRNAs are frequently deregulated in various diseases and have distinct and specific functions in a number of biological processes, such as proliferation, apoptosis or drug resistance [
22,
23]. Our previous study revealed that the RNA-binding protein RBM25 induces circAMOTL1L biogenesis by directly interacting with circAMOTL1L, p53 upregulates circAMOTL1L expression through activating the RBM25 gene, whereas p53 downregulation in PCa cells facilitates epithelial-mesenchymal transition (EMT) [
24]. Recently, we found that circACTA2 is able to mediate NRG-1-ICD regulation of its parental gene ACTA2 (alpha-actin gene) in vascular smooth muscle cells via NRG-1-ICD/circACTA2/miR-548f-5p axis [
25]. Remarkably, several lines of evidence suggest that some circRNAs play important roles in the resistance of cancer cells to anticancer drugs. For example, circAKT3 upregulates PIK3R1 to enhance cisplatin resistance in gastric cancer via inhibition of miR-198 [
22]. circ_0025202 suppresses tumor growth and enhances tamoxifen sensitization via regulating the miR-182-5p/FOXO3a axis in breast cancer [
26]. Circular RNA cESRP1 increases small cell lung cancer responsiveness to chemotherapy by sequestering miR-93-5p to inhibit the TGF-β pathway [
27]. So far, the multiple mechanisms that have been reported to be associated with the development of drug resistance involve alterations in drug targets, drug metabolism, cancer stem cell population, DNA damage repair, as well as cell survival and death signals [
28]. Despite the important roles of circRNAs in prostate carcinogenesis and drug resistance, little is known about the role of circRNAs derived from the same parental gene, which can regulate the transcription of the parental gene by binding to RNA polymerase II [
29], and thus form the positive feedback loop between circRNAs and their parental genes to induce prostate carcinogenesis and drug resistance.
In this study, we report that CDK13 is significantly upregulated in PCa, and transcriptional activation of endogenous CDK13 promotes E2F5 expression by facilitating the formation of circCDK13, which in turn sponges miR-212-5p/449a and thus relieves their repression of the E2F5 expression, subsequently leading to the upregulation of E2F5 expression. Further, increased E2F5 enhances CDK13 transcription and promotes circCDK13 biogenesis. Our findings provide the first evidence that CDK13 upregulation-induced formation of the positive feedback loop among circCDK13, miR-212-5p/miR-449a and E2F5 contributes to prostate carcinogenesis and drug resistance.
Methods
The detailed procedures of plasmid and lentivirus expression vector constructs, antibody and immunoblot, xenograft animal model, RNA isolation and RT-qPCR, cell proliferation assays, chromatin immunoprecipitation-qPCR, co-immunoprecipitation assay, immunofluorescence staining, in situ hybridization, morphometry and histology, luciferase reporter assay, analyses of apoptosis, RNA synthesis and biotin pull-down, TUNEL staining, RNA immunoprecipitation (RIP) assays, proximity ligation assay as well as key reagents are described in Supplementary Experimental Procedures.
Clinical samples and microarray
Clinical samples collection and clinicopathological characteristics as described previously [
24,
30]. In brief, Patients underwent radical prostatectomy for localized PCa and benign prostatic hyperplasia underwent transurethral resection of the prostate (TURP) at the Department of Urology, the Second Hospital of Hebei Medical University, China from July 2014 to October 2017. No treatment was administered prior to surgery. All the tissue specimens were confirmed by two experienced pathologists. The study protocol was approved by the Ethics Committee of Second Hospital of Hebei Medical University and Verbal consent was obtained from each patient. Microarray hybridization analysis of mRNA expression in 2 PCa samples and 2 BPH were performed according to the manufacturer’s protocol (Arraystar, Inc., Rockville, MD, USA).
Cell culture and transfections
PC3 (CRL-1435; ATCC), LNCaP (CRL-1740; ATCC), 22Rv1 (CRL-2505; ATCC) and DU145 cells (HTB-81; ATCC) were grown and maintained in RPMI 1640 medium (Gibco, USA) containing penicillin (100 units/ml) and streptomycin (100 μg/ml). RWPE-1 cells (CRL-11609; ATCC) were maintained in K-SMF medium (Life Technologies, USA) supplemented with 5 ng/mL epidermal growth factor (EGF) and 50 μg/mL bovine pituitary extract. Cell Cultures and transfections were performed as described previously [
24,
30]. In brief, the transfection was using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocols. The miR-449a and miR-212-5p mimics, mimic NC, inhibitors, inhibitor NC, circRNA probe and its controls were purchased from GenePharma Co., Ltd. (Shanghai, China).
Activation of endogenous genes by a CRISPR-Cas9 complex
In order to activate endogenous CDK13 we used a CRISPR-Cas9 complex with three SAM components: dCas9–VP64 (Addgene #61425),MS2–p65–HSF1(Addgene, #61426), and sgRNA (Addgene, #89493) as described previously [
31]. gRNAs were designed using the online optimized CRISPR design tool (
http://crispr.mit.edu) and targeted the proximal promoter regions of CDK13. Oligos, synthesized by Sangon Biotech., (Shanghai, China), were annealed and sub-cloned into the lentiGuide-puro vector.
Statistical analysis
All of the data are presented as the means±SEM. Student’s t-test was used to analysis the differences between two groups for multiple comparisons or repeated measurements, ANOVA or repeated ANOVA followed by Tukey’s post hoc test was used. P < 0.05 was considered statistically significant. Statistical analysis was performed using Graphpad Prism 7 software (GraphPad Software, San Diego, CA, USA).
Discussion
In the present study, we showed that CDK13 was significantly upregulated in PCa tissues, consistently with our results in the TCGA database. The upregulation of CDK13 depressed apoptosis and promoted proliferation of PCa cell lines. The CoIP-MS revealed that there exist a strong interaction of E2F5 with CDK13, which is involved in PCa cell proliferation. Interestingly, transcriptional activation of endogenous CDK13, but not the forced expression by transfecting a CDK13 expression plasmid into cells, remarkably promoted E2F5 protein expression by facilitating circCDK13 formation. The increased circCDK13 functions as a ceRNA of miR-221-5p and miR-449a, both of which target E2F5 3′-UTR, and thus relieves miR-221-5p and miR-449a repression of the expression of E2F5, leading to E2F5 upregulation. Subsequently, E2F5 functions as the transcriptional activator of CDK13 gene and positively regulates circCDK13 expression. To provide supporting evidence that circCDk13 can act as the ceRNA of miR-212-5p and miR-449a, we quantified the endogenous levels of circCDK13, miR-212-5p and miR-449a, and found that the expression level of circCDK13 is 26.59 times in PC3 cells and 13.49 times in 22RV1 cells over that of miR-212-5p, as well as 10.21 times in PC3 cells and 8.83 times in 22RV1 cells over that of miR-449a (Appendix Fig.
S9). Moreover, there are two miR-212-5p binding sites and four miR-449a binding sites in circCDK13 sequences. Collectively, these results suggest that there are sufficient circCDK13 copies that are present in PC3 and 22RV1 cells, and thus circCDk13 can function as the sponge RNA of miR-212-5p and miR-449a in these cells. Our findings provide the first evidence that CDK13 upregulation-induced formation of the feedback regulatory loop among circCDK13, miR-212-5p/miR-449a and E2F5 is responsible for the progression of PCa. Importantly, interference of E2F5/CDK13/circCDK13/miR-212-5p/miR-449a pathway by a pharmacological inhibitor 1-Azak may be a novel therapeutic strategy for PCa.
Accumulating evidence reveals that circRNAs are not the by-products of mis-splicing or splicing errors, and a lot of circRNAs have been indicated to play an important role in cancer development, such as in prostate cancer, bladder cancer, esophageal squamous cell carcinoma and basal cell carcinoma [
39]. Despite the recent advances regarding disease-related circRNAs, little is known about the biogenesis of circRNAs and the underlying molecular mechanism of circRNA-mediated gene regulation in PCa development. Our previous study found that the RNA binding protein RBM25 interacted directly with circAMOTL1L and induced its biogenesis, whereas p53 regulated epithelial–mesenchymal transition (EMT) via direct activation of RBM25 gene [
24]. The neuregulin-1 intracellular domain (Nrg-1-ICD) induced circACTA2 formation in vascular smooth muscle cells through recruiting the zinc-finger transcription factor IKZF1 to the first intron of smooth muscle α-actin gene [
25]. In addition, circular RNAs may be generated along with the transcription of its parental gene, and in turn regulate the expression of its parental gene [
35]. Therefore, the pivotal role of circRNAs in the regulation of gene expression cannot be ignored, especially those of the circRNAs whose upregulation is accompanied by a corresponding increase in linear mRNA expression. In this study, we found that E2F5 directly bound to the CDK13 promoter and upregulated CDK13 expression and circCDK13 biogenesis. In turn, upregulation of circCDK13, as a feedback mechanism, enhanced the expression of its parental gene CDK13 via the miR-212-5p/miR-449a-E2F5 regulatory axis. It is worth noting that the positive feedback loop formed by circRNAs is often overlooked in terms of the drug resistances. For example, THZ531, an inhibitor of CDK13, potently inhibits CDK13 by irreversibly targeting a cysteine located outside the kinase domain and thus suppresses cell proliferation [
40]. However, the circCDK13 produced together with CDK13 mRNA can still promote the proliferation by regulating the expression of the proliferation-related transcription factor E2F5. Therefore, blocking the positive feedback loop between circCDK13 and E2F5 in the regulation of gene expression may be one of the effective ways to prevent drug resistance.
CDK13 (also known as CDC2L5, CHED) belongs to the member of cyclin-dependent serine/threonine protein kinase family [
41]. Previous studies have shown that knocking out CDK13 leads to abnormal expression of several genes involved in a variety of biological processes. Certain downregulated genes are robustly associated with transmembrane receptor protein kinase signaling, enzyme-linked receptor protein signaling pathways, cell growth regulation, helix localization, regulation of response to external stimuli, cell size regulation, and cell projection [
17]. Notably, CDK13 is a crucial regulator of cell cycle progression in eukaryotes [
41]. Five mammalian CDKs have been reported to be transcription-associated kinases that, together with their corresponding cyclin subunits such as CDK7/cyclin-H, CDK8/cyclin-C, CDK9/cyclin-T1 or -T2, CDK12/cyclin-K and CDK13/cyclin-K, regulate cell cycle progression and transcription [
42]. A recent study of the structural and functional analysis of the CDK13/cyclin-K complex revealed that CDK13 contains a C-terminal extensional helix, a specific feature of transcriptional elongation kinases [
33]. Although these complexes are related to transcription, especially, both CDK12 and CDK13 knockdown affects the expression of genes involved in RNA processing, CDK13-regulated gene sets are not affected by CDK9 or CDK12 knockdown, further suggesting that these CDK functions do not overlap with each other [
41]. Moreover, several recent studies reported that CDK12 expression is dysregulated in metastatic castration-resistant prostate cancer (mCRPC) samples, and CDK12 loss results in highly recurrent gains at loci of genes involved in the cell cycle and DNA replication [
18‐
20]. However, much less is known regarding CDK13 expression and function in PCa. In this study, we found that the expression of CDK13 was significantly increased in PCa tissues and TCGA database. Overexpression of CDK13 promoted, whereas depletion of CDK13 inhibited the proliferation of PCa cells in vitro. Importantly, we found that transcriptional activation of endogenous CDK13 by sgRNA, but not overexpression of CDK13 by its expression vector, substantially upregulated E2F5 expression by a way of epigenetics, which in turn enhanced PCa cell proliferation. Previous research has shown epigenetic regulation enables tumors to respond to changing environments during tumor progression and metastases and facilitates treatment resistance. However, a highly selective inhibitor of CDK13 that can disables triple-negative breast cancer cells progression and metastases [
43]. In our immunochemistry staining result, we found that there is a strong CDK13 staining in the stromal tumor micro-environment. We hypothesize that the high expression of CDK13 in stromal tumor micro-environment may be beneficial to the migration of PCa cells.
E2F5 is a member of the E2F family of transcription factors which contain one or more evolutionarily conserved domains that bind target promoters and regulate their transcription [
37]. These domains include a DNA-binding domain that determines the dimeric domain that interacts with the transcription factor protein, a trans-activated domain rich in acidic amino acids, and a tumor suppressor protein embedded in the trans-activation domain. This means that E2F5 functions as a activator or a depressor depending on the proteins which interact with it. Many studies have shown that E2F5 is a oncogene in cancer development, such as breast cancer [
44], ovarian cancer [
45], hepatocellular carcinoma [
46], esophageal squamous cell carcinoma [
47] and prostate cancer [
48]. An increased gene copy number of E2F5 is detected in two independent cohorts of patients with breast cancer [
44,
49], and there is a positive association of E2F5 amplification with a pathological basal phenotype and a worse clinical outcome [
44]. Gandellini et al. reported that miR-205 exerts a tumor-suppressive effect in human prostate by counteracting epithelial-to-mesenchymal transition and reducing cell migration/invasion, in part through down-regulating ErbB3, E2F1, E2F5, ZEB2 and protein kinase Cε, and ectopical expression of miR-205 can halt PCa progression by the downregulation of E2F5 and E2F1 [
48]. Inositol hexaphosphate (IP6) inhibits growth, and induces G1 arrest and apoptotic death of prostate carcinoma DU145 cells via decreasing the level of E2F4 as well as via increasing binding of E2F4 with pRb/p107 and pRb2/p130, thus modulating CDKI-CDK-cyclin and pRb-related protein-E2F complexes [
50]. Using whole transcriptome sequencing analysis and validation of PCR assay, we confirmed a significantly increased expression of E2F5 in PCa tissues compared with BPH tissues. The upregulation of E2F5 resulted in the activation of CDK13 transcription and increase in circCDK13 biogenesis, which in turn sponges miR-212-5p and miR-449a and thus relieves their repression of the E2F5 expression, subsequently leading to the upregulation of E2F5 expression and PCa cell proliferation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.