A circular RNA, circSMARCA5, inhibits prostate cancer proliferative, migrative, and invasive capabilities via the miR-181b-5p/miR-17-3p-TIMP3 axis

SMARCA5 (circSMARCA5) is involved in the occurrence of different cancers, but its role in prostate cancer carcinogenesis and metastatic transformation remains elusive. Thus, we evaluated the circSMARCA5 functional relevance in prostate cancer and its associated molecular mechanism. First, circSMARCA5 expression and function in this cancer were evaluated. To determine the miR-181b-5p/miR-17-3p target and clarify how circSMARCA5 regulates the miR-181b-5p-TIMP3 and miR-17-3p-TIMP3 axis, RNA immunoprecipitation, biotin-coupled microRNA capture, luciferase reporter, Western blot, and quantitative real-time PCR assays were employed. In primary and metastatic prostate cancer tissues, circSMARCA5 was significantly downregulated compared with normal controls. Functionally, circSMARCA5 exhibited a suppressive effect on prostate cancer cells' metastasis and growth. At the molecular level, circSMARCA5 could affect the tissue inhibitor of metalloproteinases 3 (TIMP3) expression through miR-181b-5p or miR-17-3p interactions. Moreover, lysine acetyltransferase 5 (KAT5) induced circSMARCA5 biogenesis and regulated the miR-181b-5p-TIMP3 and miR-17-3p-TIMP3 axis. These results suggested that targeting circSMARCA5-miR-181b-5p-TIMP3 and circSMARCA5-miR-17-3p-TIMP3 axis might be a novel therapeutic strategy for prostate cancer.


INTRODUCTION
Prostate cancer is a common male cancer with increasing morbidity and mortality rates in China [1,2]. Despite recent progress in prostate cancer patient's fiveyear survival rate, it remains a significant medical challenge. Particularly, resistance to radiotherapy, chemotherapy, or androgen deprivation therapy can lead to tumor metastasis and recurrence [3]. Prostate cancer pathogenesis and development are complicated and have different deregulated oncogenes and tumorsuppressors involved [4]. Therefore, elucidate prostate cancer genetic mechanisms to understand its metastatic capability and improve clinical therapy efficiency is critical.
MicroRNAs (miRNAs) are involved in many human malignancies and have been considered a therapeutic target due to their roles in the regulation of tumor cell behaviors [11]. circRNAs could sponge miRNA and control their expression, subsequently affecting their binding to mRNAs targets [12]. For instance, miR-181b-5p and miR-17-3p oncogenic roles in tumors proliferative, invasive, and metastatic capabilities were previously reported [13,14]. Additionally, circSMARCA5 regulatory effect on miR-181b-5p and miR-17-3p in HCC was previously reported [10]. However, if circSMARCA5 regulates prostate cancer cells' behaviors via miR-181b-5p and/or miR-17-3p sponging remains unknown. Therefore, we evaluated circSMARCA5 expression, role, and mechanism in the regulation of prostate cancer cells' proliferative and invasive capabilities.

Low circSMARCA5 expression in metastatic prostate tumors
First, we detected the circSMARCA5 (has_circ_ 0001445, located in chr4:144464661-144465125) expression pattern in prostate cancer. Primers that target and amplify the circSMARCA5 back-splice site were designed to detect its expression in prostate cancer. Malignant tumor tissues had significantly low circSMARCA5 expression, compared with primary samples, which was also observed in metastatic samples ( Figure 1A). Additionally, we observed lower circSMARCA5 expression in DU145, LNCaP, and PC3 cells, compared to RWPE-1 cells ( Figure 1B). Collectively, these data indicated that circSMARCA5 expression is involved in prostate malignant behaviors.

circSMARCA5 inhibits prostate tumors proliferative, migrative, and invasive capabilities in vitro
To further reveal circSMARCA5 functional relevance in prostate carcinogenesis and metastatic transformation, tumor cellular behaviors in which circSMARCA5 expression was impaired or enhanced were examined. circSMARCA5 was successfully overexpressed in DU145 cells ( Figure 2A). Accordingly, circSMARCA5 upregulation suppressed cell growth, migration, and invasion ( Figure 2B-2E). DU145 cells with circSMARCA5 knockdown exhibited the opposite effect on these processes ( Figure 2B-2E). Similar effects were detected in PC3 cells ( Figure 3A-3E). Altogether, these data validated that circSMARCA5 has a tumor-suppressive effect in prostate cancer.

AGING circSMARCA5 inhibits tumor progression and metastasis in vivo
Next, we evaluated the circSMARCA5 role using a prostate cancer xenograft mouse model. circSMARCA5 overexpression in DU145 cells strongly suppressed tumor growth ( Figure 4A-4D). Immunohistochemistry staining showed that overexpressed circSMARCA5 decreased vimentin and N-cadherin levels, and enhanced E-cadherin levels ( Figure 4E). Moreover, a lung metastasis model confirmed that SMARCA5 upregulation dramatically inhibited lung metastasis ( Figure 4F). Overall, these results suggested that circSMARCA5 overexpression can inhibit prostate tumor growth and metastasis.

Lysine acetyltransferase 5 (KAT5) mediates circSMARCA5 biogenesis
Furthermore, the mechanism underlying the circSMARCA5 decrease in prostate cancer cells was explored. Our previous study focused on KAT5's effects on prostate cancer pathogenesis [15]. Currently, compared to the control prostate epithelial cells RWPE-1, the KAT5 level was significantly decreased in AGING DU145, LNCaP, and PC3 cells ( Figure 7A). To explore the interaction between KAT5 and circSMARCA5 formation, RIP assays with a KAT5 antibody revealed significant circSMARCA5 enrichment in the anti-KAT5 immunoprecipitants, compared to those in IgG controls ( Figure 7B). Next, we determined if KAT5 could regulate the circSMARCA5 biogenesis. Results from qRT-PCR showed that KAT5 increased the circSMARCA5 production ( Figure 7C), suggesting that KAT5 binds to circSMARCA5 and induces its biogenesis.
AGING DISCUSSION circSMARCA5 is implicated in the tumorigenesis of different cancers [9,10]. However, the circSMARCA5 biological relevance in prostate carcinogenesis and metastatic transformation was unknown. In this study, we demonstrated that the circSMARCA5-miR-181b-5p/ miR-17-3p-TIMP3 axis is critical in prostate cancer development.
The circSMARCA5 tumor-suppressive role, discovered in the present study, was consistent with previous studies. For example, Davide et al. found that circSMARCA5 was negatively correlated with angiogenesis in glioblastoma multiforme [16]. Similarly, a negative correlation between circSMARCA5 and clinicopathological features, including cancer diameter, invasion, and the tumor-nodemetastasis stage were observed in HCC [17]. These studies suggested that circSMARCA5 may act as a prognostic marker for cancers. Furthermore, functional assays revealed that circSMARCA5 inhibits the migrative and invasive capabilities in some tumors, such as cervical cancer, HCC, and lung cancer [18,19]. Both tumor suppressive and promotive functions of circSMARCA5 were observed in prostate cancer [10,20]. Our present results revealed that circSMARCA5 was negatively associated with prostate cancer proliferative, migrative, and invasive capabilities. Consistently, our in vivo assay also confirmed that circSMARCA5 exhibited a suppressive role in prostate tumors growth and metastasis. AGING circRNAs can play diverse roles such as RNA-binding proteins, miRNA sponges, and transcriptional regulators [21]. For instance, circ-Sirt1 can sponge the miR-132/212 cluster to alleviate their suppressive effects on SIRT1 expression [22]. circSMARCA5 sponges the miR-19b-3p, leading to invasion, migration, and growth inhibition in NSCLC cells [14]. Through bioinformatics analysis, we predicted the interacting miRNA and its regulated target gene. Consequently, miR-181b-5p and miR-17-3p had conserved circSMARCA5 binding sites and were shown to be onco-miRNAs with proliferative, invasive, and metastatic potential in tumor cells [13,14]. Additional results confirmed that circSMARCA5 could sponge miR-181b-5p and miR-17-3p. TIMP3 is a tumor-suppressive factor that if lost will accelerates tumor invasion and metastasis in prostate cancer [23]. Thus, we further identified that TIMP3 can be targeted by miR-181b-5p and miR-17-3p, implying a regulatory role of the circSMARCA5-miR-181b-5p-TIMP3 and/or circSMARCA5-miR-17-3p-TIMP3 axis involved in prostate cancer progression. Indeed, we found that miR-181b-5p or miR-17-3p attenuated the circSMARCA5 inhibitory role on tumor cell migration and invasion, validating that circSMARCA5 could sponge miR-181b-5p, as well as miR-17-3p, to promote TIMP3 and inhibit malignant behaviors in prostate cancer.
Furthermore, the mechanism regarding the regulation of circSMARCA5 deregulation in prostate cancer remains unknown. As a main NuA4 subunit, KAT5 can initiate target genes expression and regulate histone acylation [24]. KAT5 can induce prostate cancer cells' apoptosis and suppress their growth, representing a gene therapy target [25]. Furthermore, KAT5 can directly interact with the circRNA, circRHOT1, which is involved in hepatocellular carcinoma progression [26]. The pathogenic effect of KAT5 in prostate cancer was observed in our previous study [16]. Herein, decreased KAT5 expression in prostate cancer tissues was observed and KAT5 upregulation promoted circSMARCA5 production. Functionally, KAT5 suppression enhanced the migrative and invasive capabilities and reversed the circSMARCA5 suppressive effects, suggesting a novel link of the KAT5-mediated circSMARCA5 production in prostate cancer.

Patients, specimens, and cell culture
Patients (n = 20) who underwent radical prostatectomy for primary, metastatic, and benign prostate cancer, and prostate transurethral resected patients were enrolled in this study. The Ethics Committee of Shanghai Jiao Tong University approved all procedures. RPMI1640 medium was used to culture the DU145, PC3, and LNCaP cells. The K-SMF medium was used for RWPE-1 cells.

Cell proliferation
First, prostate cancer cells were seeded (4 × 10 3 ). Then, the tetrazolium compound (MTS; Promega, USA) was employed to determine cell growth. The values were determined at 490 nm after a 2-hour incubation at 37° C.

Immunohistochemical analysis
After fixation with 10% PFA, tumor tissues were cut into 5 mm-thick sections. Slides were heated in a dry oven, deparaffinized, rehydrated through graded ethanol, and microwaved in 10 mM citric acid buffer (pH 6.0). Nonspecific binding was blocked with 10% goat serum and the primary antibodies against Ncadherin (#4061, CST, USA), vimentin (#5741, CST, USA), or E-cadherin (#3195, CST, USA) were incubated at 4° C overnight. Finally, the slides were incubated with horseradish peroxidase-labeled streptavidin for targeted proteins detection.

Measurement of cell migrative and invasive capabilities
To evaluate migrative and invasive capabilities, a Transwell plate was employed. Cells were plated into the top part of the chambers. The Matrigel was used to coat the chambers to determine the cell's invasive capability with indicated treatment. The normal medium was added to the bottom part of the Transwell plate. After 48-hour incubation, methanol was used to fix cells that migrated, or infiltrated, to the bottom of the champers. Finally, fixed cells were observed after crystal violet staining using a microscope.

Western blot
The proteins were isolated from indicated lysed cells and 20 μg proteins were loaded into SDS-PAGE per lane. Protein bands were transferred onto polyvinylidene difluoride membranes. After washing and blocking three times, membranes were incubated with indicated primary and secondary antibodies against TIMP3 (ab39184, Abcam, USA). GAPDH (ab8245, Abcam, USA.) was detected using HRPconjugated secondary antibodies (Promega, USA). A chemiluminescence detection system (Millipore, WBKLS0500) was used to visualize the protein bands. The GAPDH (1:5000, Abcam, USA) was the normalization control.

Animals
The Animal Ethics Committee of Shanghai Jiao Tong University reviewed and approved all procedures in this study. To induce the tumor xenograft murine model, 5 × 10 7 DU145 cells were subcutaneously injected into nude mice. Additionally, mice were intraperitoneally injected with tumor cells for metastasis assessment. Tumors size was monitored once a week, and the formula length X width 2 was employed to calculate its volume. For Hematoxylin and Eosin (H&E) staining, lung tissues were sectioned and fixed with 10% PFA and stained with H&E, based on standard protocols.

RNA immunoprecipitation assay
The Biotin-labeled circSMARCA5 used in this study was synthesized by Sangon Biotech (Shanghai, China). After circSMARCA5 overexpression, cells were washed with PBS and fixed with 1% formaldehyde. Then, cells were incubated with streptavidin Dynabeads (Invitrogen, USA). The precipitated RNA was used to detect the targeted RNA by qRT-PCR. This experiment was repeated at least three times.

Statistical analysis
All values are shown as means ± standard deviations (S.D.). The Student's t-test was performed to determine the differences between indicated groups. A p < 0.05 was considered statistically significant.

AUTHOR CONTRIBUTIONS
Conceptualization, investigation, data collection and analysis, manuscript writing, review, and editing: Xin Xie, Fu-Kang Sun and Wei He. Data curation, methodology, and investigation: Xin Huang and Cheng-He Wang. Data curation, formal analysis, and investigation: Jun Dai, Ju-Ping Zhao, and Chen Fang.