Cervical Cancer Cells-Derived Extracellular Vesicles Containing microRNA-146a-5p Affect Actin Dynamics to Promote Cervical Cancer Metastasis by Activating the Hippo-YAP Signaling Pathway via WWC2

Application of extracellular vesicles (EVs) for cancer treatment has been well-documented. We probed into the potential role of cervical cancer cells-secreted EVs by transferring miR-146a-5p in cervical cancer. After characterization of miR-146a-5p expression in clinical cervical cancer tissue samples, gain- and loss-of-function experiments were implemented to test the effect of miR-146a-5p on the invasion, epithelial-mesenchymal transition (EMT), and anoikis in cervical cancer cells. EVs were isolated from high-metastatic cervical cancer cells, after which their effects on the malignant behaviors of low-metastatic cervical cancer cells were assessed in a co-culture system. Luciferase assay was implemented to validate the putative binding relationship between miR-146a-5p and WWC2, followed by further investigation of downstream pathway (Hippo-YAP). Finally, nude mouse lung metastasis model was developed for in vivo validation. miR-146a-5p was elevated in cervical cancer tissues and high miR-146a-5p expression promoted the metastatic potential of cervical cancer cells through enhancing their invasiveness and anoikis resistance, and inducing EMT. Furthermore, miR-146a-5p carried by EVs secreted by highly metastatic cervical cancer cells could promote the metastasis of low-metastatic cervical cancer cells. Mechanistically, miR-146a-5p targeted WWC2 to activate YAP, by which it inhibited the phosphorylation of cofilin, and promoted the process of cofilin-mediated depolymerization of F-actin to G-actin. In vivo data demonstrated that EVs-carried miR-146a-5p promoted tumor metastasis through the WWC2/YAP axis. Cancer-derived EVs delivered pro-metastatic miR-146a-5p to regulate the actin dynamics in cervical cancer, thereby leading to cancer metastasis. This experiment highlighted an appealing therapeutic modality for cervical cancer.


Introduction
Cervical cancer poses a main gynecological problem in both developing and underdeveloped countries [1]. Recent advancements in imaging diagnosis, management, and treatment contribute to the global improvement of 5-year net survival of cervical cancer patients [2]. Metastasis remains as the major reason for the treatment failures of most patients with cervical cancer [3]. ough various pathologies associated with cervical cancer progression have been demonstrated, further investigation involving tumor metastasis is still necessary [4,5].
Extracellular vesicles (EVs) have attracted much attention from oncologists owing to their promising potential as prognostic indicators [6]. It is known that EVs containing messengers such as microRNAs (miRs) bear great responsibility in modulating the tumor development and metastasis [7]. Importance of miRs in tumor oncogenesis and metastasis has been well-characterized [8]. Specifically, elevated miR-146a-5p is witnessed in the EVs derived from inflammatory microglia [9]. Furthermore, miR-146a can enhance the viability of cervical cancer cells through regulating IRAK1 and TRAF6 [10]. More recently, upregulation of miR-146a-5p is identified in the plasma EVs of patients with cervical cancer, and it can be a promising novel biomarker for the diagnosis of cervical cancer [11]. Based on the aforementioned findings, whether EVs containing miR-146a-5p involves in cervical cancer is yet to be investigated.
All WW-and-C2-domain-containing (WWC) proteins in the WWC protein family can suppress the transcriptional activity of YAP to inhibit cell proliferation and organ growth [12]. It is reported that WWC2, belonging to the WWC protein family, exerts tumor-suppressing function in lung adenocarcinoma through negatively regulating the Hippo signaling pathway [13]. Emerging evidence shows the involvement of WWC2 in cancer metastasis. For instance, WWC2 can reverse epithelial-mesenchymal transition (EMT) and block stemness maintenance of pancreatic cancer stem cells through inhibiting the Hippo signaling pathway [14]. e Hippo-YAP signaling pathway exerts a great function in regulating organ volume, stem cells, tissue regeneration, and tumorigenesis [15]. It is reported that the Hippo/YAP signaling pathway can regulate cervical cancer progression through interacting with EGFR signaling and HPV oncoproteins [16]. YAP functions as a promoter of tumor metastasis through regulating the actin dynamics [17].
In view of these findings, this study is intended to explore the regulatory mechanism how the cervical cancer cell-secreted EVs carrying miR-146a-5p contributes to cervical cancer metastasis.

Bioinformatics Analysis.
e cervical cancer-related miRNA expression microarray GSE86100 was obtained through retrieval from the Gene Expression Omnibus database, which contained a total of 12 samples (6 normal samples and 6 cervical cancer samples). e R "limma" software package was run for differential analysis with |logFC| > 1 and p < 0.05 used as thresholds to screen the significantly differentially expressed miRNAs in the microarray. In addition, miRNA expression microarray GSE30656 including 10 normal samples and 19 cervical cancer samples were obtained for miRNA expression verification. e ENCORI, miRDB, miRWalk, and Tar-getScan databases were searched to predict target genes of miR-146a-5p utilizing |log2FC| > 1.5 and p value <0.01 as the screening criteria. e cervical cancer-related genes were obtained through the GEPIA2 website, which were intersected with the target genes to predict downstream regulatory genes of miR-146a-5p in cervical cancer.

Clinical Sample Collection.
Cervical cancer tissues were collected from 30 patients diagnosed with cervical cancer with an average age of 52.7 ± 6.25 from the Shanghai Tongji Hospital, School of Medicine, Tongji University. In accordance with the diagnostic criteria of the International Federation of Gynecology and Obstetrics systems (2014), all patients were examined by two gynecological oncologists with senior seniority and above through combined gynecological ultrasound, pelvic magnetic resonance imaging, computed tomography, and other examinations. Inclusion criteria were: (1) Histologically or cytologically confirmed cervical cancer; (2) Patients with primary cervical cancer were treated at the first time; (3). Pathological types: squamous cell carcinoma, adenocarcinoma, and adenosquamous cell carcinoma; (4) Patients showed no other tumors or treatment history of other tumors. Exclusion criteria were: (1) Neuroendocrine cervical tumor or other rare histology (such as clear cell carcinoma) through pathological diagnosis; (2) Patients were not newly diagnosed cancer patients; (3) Patients had history of tumor treatment; (4) Patients undergoing drug intervention. Normal cervical tissues were collected from 30 patients (mean age of 49.67 ± 5.65) who were diagnosed with uterine fibroids without HPV infection or cervical lesions. e tissue samples were stored at −80°C.

Cell Culture and Lentivirus Transduction.
Human embryonic kidney cells (HEK293T cells), cervical cancer cell lines (HeLa, CaSki, SiHa, and C33A), and normal cervical epithelial cell lines (End1/E6E7 and HcerEpic) were all procured from Mingzhou Biotechnology (Ningbo, China). e cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco, Grand Island, NY) replenishing 10% fetal bovine serum (Gibco), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco) at 37°C with 5% CO 2 . miR-146a-5p mimic, miR-146a-5p inhibitor (this type of inhibitor is chemically modified specifically for the specific target miR-146a-5p in cells, which can specifically target and knock down the expression of a single miRNA molecule), and negative controls (mimic-NC and inhibitor-NC) were all procured from GenePharma Co., Ltd. (Shanghai, China). e cells were transduced with Vector, WWC2, short hairpin RNA (sh) NC, shWWC2-1, shWWC2-2, shYAP-1, shYAP-2, and shWWC2 + shYAP, respectively. Logarithmically growing cells were detached with trypsin, prepared into cell suspension (5 × 10 4 ) cells/mL and seeded in a 6-well plate with 2 mL per well, followed by culture overnight at 37°C. After 48 h of transduction, the GFP expression efficiency was checked by a fluorescence microscope. e stable-transduced cell line was constructed. At 72 h after the virus transduction, the medium was renewed with a complete medium appended to 2 μg/mL puromycin. e shRNA sequence (Supplementary Table 1) was designed by Life Technologies and synthesized by GenePharma.

Extraction and Identification of EVs.
e equal number of well-grown cervical cancer cells CaSki were collected, plated in a 10 cm plate, and cultured overnight in DMEM containing 10% EV-free serum (overnight centrifugation was performed at 12, 000 × g to remove EVs). Under 60-70% confluence, the supernatant was harvested. e clinical serum samples and cell culture supernatant were centrifuged at 2000g and 4°C for 20 minutes to remove cell debris, and the resulting supernatant was filtered utilizing a 0.22 μm filter (Merck Millipore, Billerica, MA). EVs were gained by ultracentrifugation as previously described [18]. Ultracentrifugation was processed utilizing Optima Max-XP (Beckman Coulter, Miami, FL). e EVs were observed under a Philips CM120 BioTwin transmission electron microscope (FEI company). e EV particles were used for quantitative identification. e antibodies (Abcam, Cambridge, UK) for Western blot assay included TSG101 (ab125011, 1 : 1000), CD63 (ab134045, 1 : 1000), CD81 (ab109201, 1 : 5000), and Calnexin (ab22595). Diameter of EVs was determined employing dynamic light scattering utilizing Zetasizer Nano-ZS90 instrument of Malvern Company (Malvern, UK).

Uptake of EVs.
e extracted CaSki EVs were labeled by referring to the PHK67 labeling kit (KH67GL, Sigma-Aldrich, St Louis, MO). C33A cells were cultured overnight in the Petri dish with a special cell climbing piece and then incubated with 10 μg PHK67-labeled CaSki-derived EVs for 24 h. Following soaked in 4% paraformaldehyde for 30 minutes, the slices were subjected to permeabilization with 2% Triton X-100 for 15 minutes and blockage with 2% bovine serum albumin (BSA) for 45 minutes. After that, the slides were stained with 4′,6-diamidino-2-phenylindole (DAPI) (2 μg/mL) and observed under an inverted fluorescence microscope.

RT-qPCR.
TRIzol (16096020, Invitrogen) was employed to extract total RNA from tissues and cells. For miRNA, PolyA tailing detection kits (B532451, Sangon Biotech Co., Ltd. Shanghai, China) containing universal PCR primer R and U6 universal PCR primer R were adopted to obtain cDNA of miRNA with PolyA tail. Using SYBR Premix Ex Taq ™ II kits (DRR081, Takara, Kyoto, Japan), RT-qPCR was run in a real-time fluorescent quantitative PCR instrument (ABI 7500, ABI, Foster City, CA). U6 was served as a normalizer for miRNA in tissues while syn-cel-miR-39 (miRB0000010-3-1, RiboBio, Guangzhou, China) as an internal control in EVs.
For mRNA detection, cDNA was obtained employing reverse transcription kits (RR047A, TaKaRa, Tokyo, Japan). On the basis of TaqMan Gene Expression Assays protocol (Applied Biosystems, Foster City, CA), PCR was conducted. GAPDH was used as a normalizer. Primer sequences are described in Supplementary Table 2. e relative expression was quantified utilizing the 2 −ΔΔCt method. as well as horseradish peroxidase labeled goat anti-rabbit IgG (ab6721, Abcam) or goat anti-mouse IgG (ab6789, Abcam). ImageJ software was utilized for protein quantitative analysis.

Dual-Luciferase Reporter Gene Assay.
e wild-type (wt) plasmid containing WWC2 3′untranslated region (UTR) full-length sequence and mutant (mut) plasmid were constructed and designated as PmirGLO-WWC2-wt and PmirGLO-WWC2-mut, respectively. e dual luciferase reporter plasmid and the corresponding plasmid were cotransfected into 293T cells. After 48 h of transfection, the luciferase activity was assayed employing Dual-Luciferase 2.12. Transwell Assay. Transwell chambers (pore size of 8 mm; Corning Incorporated, Corning, NY, USA) in 24-well plates were adopted for invasion and migration experimentations. A total of 100 μL of Matrigel was spread in each chamber and incubated at 37°C for 2 h for invasion detection. Matrigel was not included in cell migration experiment. e cells were treated with the equal number of EVs in different groups, detached, and resuspended in serum-free DMEM (Gibco) to a density of 3 × 10 5 cells/mL [19]. e invaded or migrative cells were counted and photographed under an Olympus IX 71 microscope (Olympus Optical Co., Ltd, Tokyo, Japan).
2.13. Immunohistochemistry. After preparation of tissue sections, deparaffinization with xylene and blockage of endogenous peroxidase activity utilizing 3% hydrogen peroxide were implemented. en, the sections were blocked in 10% normal goat serum for 15 minutes. e antibody YAP (1 : 200; # 14074, CST) was added for incubation overnight at 4°C. e next day, the secondary antibody was added to the sections for 1-h of incubation. e immunoreactivity was quantified utilizing the diaminobenzidine tetrahydrochloride kit (Invitrogen), followed by a microscopic observation.

Nude Mouse Lung Metastasis Model and In Vivo
Imaging.
e healthy female BALB/c nude mice aged 4-6 weeks (National Rodent Laboratory Animal Resource, Shanghai Branch, PRC) were separately caged in a specificpathogen-free animal laboratory for 1 week. e C33A cells stably transduced with Vector, WWC2, shNC, or shYAP were trypsinized and dispersed into single cell suspension. A total of 5 × 10 5 luciferase-labeled stable-transduced cells were injected into nude mice through the tail vein. At the same time, 20 μg EVs or PBS was injected through the tail vein every three days. Mice were intraperitoneally injected with 200 μL of luciferase substrate D-luciferin (15 mg/mL; Gold Biotechnology, St. Louis, MO), and 15 minutes later, imaged on an in vivo animal imager (UVP iBox ® Scientia ™ , Germany). e survival curve of nude mice was plotted using GraphPad software.

Statistical
Analysis. GraphPad Prism software was processed for data analysis, and the p value was determined by a two-tailed t-test on independent samples. Measurement data were expressed as mean ± standard deviation. Comparison between the two groups was performed by independent sample t-test, and comparison among multiple groups was assayed by one-way analysis of variance. e Kaplan-Meier method was utilized to calculate the survival rate of nude mice, and the Log-rank test was applied for univariate analysis. p < 0.05 indicated statistical significance.
Subsequent RT-qPCR identified similar elevation of miR-146a-5p expression in cervical cancer tissues ( Figure 1(e)). Meanwhile, its elevation was also witnessed in HeLa, CaSki, SiHa, and C33A cells than in End1/E6E7 and HcerEpic cells. Among all cancer cells, miR-146a-5p expressed at the highest level in high-metastatic CaSki cells and at the lowest level in low-metastatic C33A cells (Figure 1(f )).
We then probed into the effect of miR-146a-5p on cervical cancer metastasis. RT-qPCR results showed increased expression of miR-146a-5p in the cells transfected with miR-146a-5p mimic but opposite finding was witnessed upon miR-146a-5p inhibitor transfection (Figure 1(g)).
Additionally, mimic of miR-146a-5p promoted cell migration and invasion, while loss-of-function of miR-146a-5p caused contrary trends ( Figure 1 hsa-miR-590-5p type type cancer   Journal of Oncology neighboring cells during metastasis, and circulated to a distant place to escape apoptosis during transport (i.e., antianoikis) [21]. erefore, the cell-matrix adhesion experiment was conducted to assess the adhesion of cells to the extracellular matrix. We found that mimic of miR-146a-5p could inhibit cell adhesion to the extracellular matrix, while loss-of-function of miR-146a-5p promoted their adhesion to the extracellular matrix (Figure 1(i), Supplementary Figure 1B). At the same time, we assessed cell apoptosis after anoikis treatment (to mimic the microenvironment of extracellular matrix loss). TUNEL assay results depicted that the number of anoikis of cells overexpressing miR-146a-5p was decreased, but it was increased following miR-146a-5p inhibition (Figure 1(j), Supplementary Figure 1E). Next, the EMT marker protein expression was characterized by Western blot assay, which clarified that overexpression of miR-146a-5p diminished E-cadherin expression and promoted the expression of N-cadherin and Vimentin; while loss-of-function of miR-146a-5p brought about opposing tendency ( Figure 1(k), Supplementary Figure 1F). Additionally, we overexpressed miR-146a-5p in HcerEpic cells, and determined the effect of miR-146a-5p on the migration and invasion of HcerEpic. e obtained results were similar to those of C33A cells, but gain-of-function of miR-146a-5p had a small effect on cell migration and invasion (Supplementary Figures 2A-2D). Western blot assay results indicated that miR-146a-5p mimic in HcerEpic cells diminished E-cadherin expression while promoting that of N-cadherin and Vimentin (Supplementary Figure 2E).
In order to elucidate whether other miRNAs have an impact on cervical cancer metastasis, we initially performed RT-qPCR to analyze the expression of the top two upregulated miRNAs hsa-miR-7-5p and hsa-miR-20b-5p in clinical cervical cancer tissues. e results displayed an elevation in the expression of the two miRNAs in the cervical cancer tissues (Supplementary Figure 3A). Furthermore, RT-qPCR results confirmed higher expression of miR-7-5p and miR-20b-5p in cervical cancer cell lines. Among all cancer cells, miR-7-5p and miR-20b-5p expressed at the highest level in high-metastatic CaSki cells and at the lowest level in low-metastatic C33A cells (Supplementary Figure 3B). Besides, we silenced or overexpressed miR-7-5p and miR-20b-5p in C33A cells, followed by determination of silencing and overexpression efficiency utilizing RT-qPCR (Supplementary Figure 3C). Moreover, the results of Transwell assay unveiled that overexpression or silencing of miR-7-5p/ miR-20b-5p had no effect on cell migration and invasion (Supplementary Figures 3D and 3E). Cell-matrix adhesion test data indicated that the overexpression or silencing of miR-7-5p/miR-20b-5p could not promote cell adhesion to the extracellular matrix (Supplementary Figure 3F). At the same time, we performed anoikis treatment on the cells (to mimic the microenvironment of the absence of extracellular matrix), and then used TUNEL assay to detect cell apoptosis. e results suggested that neither hsa-miR-7-5p nor hsa-miR-20b-5p overexpression or inhibition showed effect on the number of anoikis (Supplementary Figure 3G).
Conclusively, miR-146a-5p was elevated in cervical cancer, and its overexpression could promote malignant and anti-anoikis abilities of cervical cancer cells while inhibiting cell adhesion to the extracellular matrix.

Cervical Cancer Cells-Derived EVs Carried miR-146a-5p
to Promote the Metastasis of Cervical Cancer. To discern the impact of EVs on cervical cancer metastasis, we separated EVs from the supernatant medium of HeLa, CaSki, SiHa, and C33A cells as well as End1/E6E7 and HcerEpic cells, and observed the morphology of the separated particles under an electron microscope. As shown in Figures 2(a) and 2(b), the isolated EVs from CaSki and HcerEpic cells were basically uniform round-shaped or elliptical-shaped, with the diameter in the range of 30-120 nm determined by dynamic light scattering. Western blot assay results showed that the isolated vesicles were positive for EVs markers CD63, CD81, and TSG101 but negative for endoplasmic reticulum marker protein Calnexin, validating the successful extraction of EVs ( Figure 2(c)). Next, RT-qPCR results unfolded that compared with EVs secreted by normal cervical epithelial cells, miR-146a-5p was expressed higher in EVs secreted by cervical cancer cells (Figure 2(d)). To further identify whether the high expression of miR-146a-5p in EVs secreted by cervical cancer cells is due to the increase of EVs secreted by cervical cancer cells or more miR-146a-5p encapsulated by EVs, we compared the expression of miR-146a-5p in the equal number of EVs isolated from cervical cancer cell line (CaSki) and normal cervical epithelial cell line (HcerEpic) and the expression of corresponding target genes in the isolated EVs. Western blot assay results described that the expression of CD63, CD81, and CD9 in EVs isolated from CaSki cells was the same as that in the EVs isolated from HcerEpic cells (Supplementary Figure 2F), indicating that the number of counted EVs was the same. Further RT-qPCR detection results revealed increase of miR-146a-5p in the EVs isolated from CaSki cells than that in the EVs isolated from HcerEpic cells (Supplementary Figure 2G), indicating that miR-146a-5p was more packaged into EVs isolated from CaSki cells than the EVs from HcerEpic cells after equalizing the EV count.
Next, the EVs secreted by the highly metastatic CaSki cells were extracted, and labeled with PKH67, and then cocultured with the low-metastatic C33A cells for 30 and 120 minutes. e uptake of EVs by C33A cells was observed under a confocal fluorescence microscope. e results showed that EVs secreted by CaSki cells were internalized by C33A cells, which occurred at 30 minutes of co-culture (Figure 2(e)). en, co-culture system depicted that C33A cells co-cultured with CaSki cells transfected with miR-146a-5p-Cy3 emitted red fluorescence; however, no obvious red fluorescence was noted in the C33A cells co-cultured with CaSki cells co-treated with miR-146a-5p-Cy3 and GW4869 (Figure 2(f )), indicating that EVs from CaSki cells could deliver miR-146a-5p to C33A cells. Next, we extracted the EVs from CaSki cells overexpressing miR-146a-5p. As described by RT-qPCR, an enhancement in the miR-146a-5p expression was witnessed in EVs from CaSki cells   Journal of Oncology overexpressing miR-146a-5p while no significant difference occurred between the EVs from control cells and mimic-NC-treated cells (Figure 2(g)). ese data implied that the overexpression of miR-146a-5p may lead to more miR-146a-5p packaged into EVs. C33A cells were coincubated with EVs secreted by CaSki cells overexpressing miR-146a-5p (EV-miR-146a-5p mimic). Compared with that in the cells cultured with PBS, the expression of miR-146a-5p was increased in the C33A cells cultured with CaSki-secreted EVs (EV-control); while it was further increased following co-culture of EV-miR-146a-5p mimic (Figure 2(h)). Transwell assay showed that treatment with EV-control resulted in increased migrated and invaded C33A cells, while treatment with EV-miR-146a-5p-mimic caused an enhanced promotion (Figure 2(i), Supplementary Figure 1C). At the same time, we also found that EV-control weakened extracellular matrix adhesion and lowered the expression of E-cadherin but enhanced the apoptosis resistance ability and expression of N-cadherin and Vimentin in C33A cells; whereas, EV-miR-146a-5p-mimic caused consistent changes in larger folds (Figures 2(j)-2(l), Supplementary Figures 1D, 1G, and 1H). erefore, miR-146a-5p carried by EVs secreted from high-metastatic cervical cancer cells could promote the malignancy of low-metastatic cervical cancer cells.

miR-146a-5p
Could Negatively Target WWC2 to Activate the Hippo-YAP Signaling Pathway. We then predicted the downstream targets of miR-146a-5p utilizing TargetScan, miRWalk, miRDB, and ENCORI databases, and used GEPIA database to find the differentially expressed genes (log2FC < −1.5, p < 0.01). Finally, two candidate genes were harvested after intersection: WWC2 and NOVA1 (Figure 3(a)). It has been reported that WWC2 gene can inhibit tumor metastasis [22], while NOVA1 usually plays an oncogenic role [23]. erefore, we chose WWC2 as the target gene. Figure 3(b) shows the WWC2 expression in cervical cancer samples in the TCGA database utilizing the GEPIA2 tool. RT-qPCR results consistently showed downregulation of WWC2 in cervical cancer clinical tissues and cell lines (Figures 3(c) and 3(d)).
Based on the binding site of miR-146a-5p and WWC2 shown by ENCORI database (Figure 3(e)), the WWC2-3′UTR-mut plasmid was constructed. e results of the dual luciferase reporter experiment described that the cells cotransfected with miR-146a-5p mimic and WWC2-WT had a decreased luciferase activity (Figure 3(f )). RT-qPCR results summarized that the WWC2 mRNA expression was decreased in the cells transfected with miR-146a-5p mimic but elevated in those transfected with miR-146a-5p inhibitor (Figure 3(g)). Hence, miR-146a-5p could limit WWC2 expression in cervical cancer cells.  e activity of the Hippo-YAP signaling pathway assayed by Western blot assay unfolded that the levels of LATS1 and YAP phosphorylation were increased after WWC2 was overexpressed, while those were diminished after overexpression of miR-146a-5p. Simultaneous overexpression of WWC2 and miR-146a-5p could reverse the inhibiting effect of single miR-146a-5p overexpression on the levels of LATS1 and YAP phosphorylation (Figures 3(h) and 3(i)). Next, immunofluorescence was performed to determine YAP nuclear localization. e results showed suppressed nuclear YAP expression after overexpression of WWC2. e nuclear YAP expression was increased after mimic of miR-146a-5p, which was diminished after restoration of WWC2 (Figure 3(j)). Furthermore, mRNA expression of Hippo-YAP signaling pathway downstream molecules (CTGF and Cyr61) in C33A cells was decreased after overexpression of WWC2, but elevated after overexpression of miR-146a-5p. However, overexpression of WWC2 reversed the promotion of CTGF and Cyr61 mRNA expression induced by mimic of miR-146a-5p (Figure 3(k)). In conclusion, miR-146a-5p activated the Hippo-YAP signaling pathway through targeting WWC2 in cervical cancer cells.

miR-146a-5p Carried by Cervical Cancer Cells-Secreted EVs Could Facilitate Cervical Cancer Metastasis through
Downregulating WWC2. To explore whether miR-146a-5p mediated WWC2 to affect cervical cancer metastasis, C33A cells overexpressing WWC2 were co-cultured with EVs secreted by the CaSki cells treated with mimic-NC or miR-146a-5p mimic (EV-mimic-NC or EV-miR-146a-5p mimic). RT-qPCR results showed that the WWC2 expression was elevated by treatment with WWC2 overexpression plasmid in the C33A cells cultured with EV-mimic-NC. e expression of miR-146a-5p was increased while that of WWC2 was decreased by co-culture with EV-miR-146a-5p mimic in either the C33A cells treated with Vector or C33A cells overexpressing WWC2 (Figure 4(a)).
ese results summarized that miR-146a-5p carried by EVs could diminish WWC2 expression. Furthermore, our results demonstrated that WWC2 overexpression reduced the numbers of migrated and invaded C33A cells cultured with EV-mimic-NC, which was reversed by the delivery of more miR-146a-5p via EVs. Additionally, an increase was noted in the numbers of migrated and invaded C33A cells treated with EV-miR-146a-5p mimic + Vector relative to those in cells treated with EV-mimic-NC + Vector (Figure 4(b)). At the same time, WWC2 overexpression resulted in promoted extracellular matrix adhesion ability and elevated expression of E-cadherin, and weakened anti-anoikis ability and decreased expression of N-cadherin and Vimentin in the C33A cells cultured with EV-mimic-NC. However, EV-miR-146a-5p mimic could reverse the effects induced by WWC2 overexpression (Figures 4(c)-4(e)). Furthermore, nude mice lung metastasis models were established for in vivo validation. e results of in vivo imaging, H&E staining, and the curve for mouse survival showed that lung metastasis of the mice injected with EV-mimic-NC was suppressed and their survival time was increased upon WWC2 overexpression, whereas the lung metastasis in the mice overexpressing WWC2 was enhanced by treatment with EV-miR-146a-5p mimic, corresponding to shortened survival time. Treatment with EV-miR-146a-5p mimic + Vector enhanced the lung metastasis of the mice while arresting mouse survival time (Figures 4(f )-4(h)). Taken together, miR-146a-5p carried by EVs accelerated the metastasis of cervical cancer by inhibiting WWC2.

WWC2 Regulated Actin Dynamics to Limit the Metastasis of Cervical Cancer Cells through the Hippo-YAP Pathway.
YAP can promote activity of the actin depolymerization factor cofilin by inhibiting the phosphorylation of cofilin, thereby changing the dynamics of F-actin/G-actin and accelerating cancer metastasis [17,24]. Expression and location of YAP in clinical cervical cancer samples assayed utilizing immunohistochemical staining described increased protein expression of YAP and elevated nuclear expression in cervical cancer tissues ( Figure 5(a)). Next, shRNAs against WWC2 and YAP were constructed to silence WWC2 and YAP, of which shWWC2-1 and shYAP-1 had better silencing efficacy and were thus selected for subsequent experiments (Figure 5(b)). Consistently, Western blot assay confirmed the silencing efficacy ( Figure 5(c)). In addition, RT-qPCR results revealed that CTGF and Cyr61 were downregulated after YAP knockdown, while concomitant silencing of WWC2 and YAP could also reduce the mRNA expression of CTGF and Cyr61 ( Figure 5(d)). Western blot assay results exhibited that the phosphorylation of cofilin and the ratio of F-actin/G-actin were both decreased after WWC2 knockdown. YAP knockdown led to enhanced cofilin phosphorylation and F-actin/G-actin ratio, which were both diminished after knocking WWC2 down ( Figure 5(e)).
Immunofluorescence displayed that the stress fibers were shortened and weakened upon knockdown of WWC2, and the stress fibers were elongated after knocking YAP down, with the presence of actin filaments. Concomitant silencing of WWC2 and YAP negated the effects of YAP silencing alone on the cytoskeleton ( Figure 5(f )). e above-mentioned results suggested that WWC2 promoted phosphorylation of actin depolymerization factor cofilin through YAP and inhibit its activity, thereby inhibiting the depolymerization of F-actin to G-actin in cervical cancer cells.
Moreover, loss of YAP suppressed the migrated and invaded abilities of C33A cells while loss of WWC2 enhanced these abilities. WWC2 silencing counteracted the anti-migratory and anti-invasive effects of YAP silencing ( Figure 5(g)). At the same time, we also found that the ability to adhere to the extracellular matrix and the expression of E-cadherin were enhanced, while the anti-anoikis ability and N-cadherin and Vimentin expression were suppressed after YAP was knocked down alone in C33A cells; silencing WWC2 caused weakened extracellular matrix adhesion ability, decreased E-cadherin expression, enhanced antianoikis ability, and increased expression of N-cadherin and Vimentin. ese changes induced by YAP silencing were reversed by WWC2 silencing (Figures 5(h)-5(j)). Collectively, WWC2 suppressed cervical cancer metastasis through blocking the Hippo-YAP pathway.

miR-146a-5p Carried by Cervical Cancer Cells-Secreted EVs Accelerated the Metastasis of Cervical Cancer through Regulating the WWC2/YAP-Mediated Actin Dynamics.
To identify whether miR-146a-5p carried by EVs affected actin dynamics through WWC2/YAP, C33A cells were treated with Vector and WWC2, and then cultured with CaSki cells-secreted EVs. Western blot results showed that EV-mimic-NC or EV-miR-146a-5p mimic downregulated WWC2, and the levels of LATS1 and YAP phosphorylation in the C33A cells. On the contrary, WWC2 overexpression rescued the levels of LATS1 and YAP phosphorylation inhibited by EV-miR-146a-5p mimic (Figure 6(a)). RT-qPCR results showed that EV-mimic-NC or EV-miR-146a-5p mimic elevated the expression of miR-146a-5p, CTGF, and Cyr61 in the C33A cells; whereas, WWC2 overexpression reversed the effect of EV-miR-146a-5p mimic on CTGF and Cyr61 expression (Figures 6(b) and 6(c)). At the  same time, EV-mimic-NC or EV-miR-146a-5p mimic elevated nuclear YAP expression in C33A cells, but the nuclear YAP expression enhanced by EV-miR-146a-5p mimic was diminished after WWC2 overexpression (Figure 6(d)). erefore, miR-146a-5p carried by cervical cancer cellssecreted EVs activated the Hippo-YAP pathway through inhibition of WWC2. Next, the stably transduced C33A cells (shNC and shYAP) were cultured with EV-mimic-NC or EV-miR-146a-5p mimic. Western blot assay and immunofluorescence detection showed that the phosphorylation of cofilin and the ratio of F-actin/G-actin were decreased, and the stress fibers were shortened and weakened in the C33A cells treated with shNC after culture with EV-mimic-NC or EV-miR-146a-5p mimic. e changes caused by EV-miR-146a-5p mimic were reversed after YAP silencing (Figures 6(e)-6(g)).
Additionally, YAP silencing suppressed the migration and invasion (Figure 6(h)), enhanced adhesion potential to extracellular matrix and E-cadherin expression, and weakened anti-anoikis and N-cadherin and Vimentin expression (Figures 6(i)-6(k)) in C33A cells mediated by EV-miR-146a-5p mimic. Meanwhile, in vivo data suggested that the YAP knockdown suppressed lung metastasis, and increased survival time of mice treated with EV-miR-146a-5p mimic (Figures 6(l)-6(n)). All in all, miR-146a-5p delivered by EVs contributed to cervical cancer metastasis through WWC2mediated Hippo-YAP pathway.

Discussion
Chemotherapy and radiotherapy are the common therapeutic treatments of cervical cancer but alternative therapies   Figure 7: Molecular mechanism diagram illustrating the transfer of miR-146a-5p via cervical cancer cells-secreted EVs to promote the metastasis of cervical cancer. EVs secreted by high-metastatic cervical cancer cells carrying miR-146a-5p can be internalized by lowmetastatic cervical cancer cells, and then miR-146a-5p inhibited WWC2 expression to promote dephosphorylation and nuclear translocation of YAP. Consequently, YAP promoted the activity of actin depolymerization factor cofilin by inhibiting its phosphorylation, thereby changing the dynamics of F-actin/G-actin and leading to metastasis of low-metastatic cervical cancer cells.
are urgently required due to their side effects and toxicity [25]. erefore, extensive studies of the molecular mechanisms that regulate the metastatic potential of cervical cancer cells are required to seek new therapeutic targets and strategies for this malignancy. In this study, we provided evidence that the miR-146a-5p carried by cervical cancer cells-derived EVs activated the Hippo-YAP signaling pathway through WWC2 to affect actin dynamics and promote cervical cancer metastasis.
Our study discerned that miR-146a-5p was elevated in the cervical cancer tissues. It is reported that cancerderived EVs carry various tumor-derived molecules such as mutated DNA and RNA fragments, miRNA, and protein [7]. Consistent with our finding, a prior study has reported an elevated miR-146a expression in cervical intraepithelial neoplasia and cervical cancer [25], which is also demonstrated in a recent study, suggesting the diagnostic value of miR-146a-5p for cervical cancer [11]. e tumor-promotive role of miR-146a in cervical cancer has been unraveled in increasing studies; for instance, miR-146a can enhance cancer cell proliferation [26]. miR-146a possesses potential to promote cervical cancer cell viability through modulation of IRAK1 and TRAF6 [10]. In addition to the pro-proliferative role of miR-146a-5p, this study further demonstrated that mimic of miR-146a-5p could strengthen the anti-anoikis ability and drive the EMT process. Furthermore, evidence has been presented demonstrating an enrichment of miR-146a-5p in the EVs secreted by cervical cancer cells. Similarly, miR-146a-5p in EVs from cancer-associated fibroblasts can enhance EMT to accelerate prostate cancer cell metastasis [27]. In addition, it is reported that breast cancer-derived EVs carrying miR-146a contributed to the invasion and metastasis of breast cancer cells through activating cancerassociated fibroblasts [28].
In this experiment, the results presented that miR-146a-5p could target WWC2 and inhibited its expression to activate the Hippo-YAP signaling pathway. WWC2 is shown as an anti-oncogene in human cancers that can be mediated by tumor-promotive miRNAs. Similar with our findings, downregulation of WWC2 by miR-21-5p contributes to progression of lung adenocarcinoma [13], while upregulation of WWC2 caused by loss of miR-10a reverses the induction of EMT in pancreatic cancer stem cells [14]. Furthermore, WWC2 inhibits the metastasis of hepatocellular carcinoma by negatively modulating the Hippo pathway [22]. Our data also showed that WWC2 phosphorylated the transcriptional co-activator YAP to promote the phosphorylation of the actin depolymerization factor cofilin, whereby promoting the depolymerization of F-actin to G-actin. A prior study has reported the importance of dynamic rearrangement of F-actin in presenting the therapeutic effects of antitumor agents in tumor cells [29]. Consistent with our findings, overexpressing YAP leads to cytoskeletal rearrangement through regulating the dynamics of F-actin/G-actin turnover, thus accelerating the migration of gastric cancer cells [17]. In the in vivo tumor metastatic model, we also demonstrated that cervical cancer cells-secreted EVs carrying miR-146a-5p promoted the metastasis of cervical cancer through activating YAP.

Conclusions
Conclusively, our study showed evidence that EVs secreted by high-metastatic cervical cancer cells transmitted miR-146a-5p into low-metastatic cervical cancer cells, which activated the WWC2-mediated Hippo-YAP signaling pathway and altered the dynamics of F-actin/G-actin, finally leading to cancer metastasis (Figure 7). is study mainly highlights a promising preventive strategy based on cancerderived EVs against cancer metastasis. However, due to the small clinical sample size in the study, further larger-scale verification is required in the future.
Data Availability e datasets generated/analyzed during the current study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
Weiwei Wang and Lipei Wu contributed to the conception and design of the study; Jiale Tian contributed to the acquisition of data; Wenhui Yan and Chunrun Qi contributed to the analysis and interpretation of data; Weiwei Wang and Shihai Xuan contributed to drafting the article; Anquan Shang and Wuchao Liu contributed to revising the article critically for important intellectual content; all of the authors approved the final version to be submitted. Weiwei Wang, Lipei Wu, and Jiale Tian authors contributed equally to this work.