Preclinical Study: Sunitinib-suppressed MiR-452-5p Facilitates Renal Cancer Cell Invasion and Metastasis Through Modulating SMAD4/SMAD7/EMT Signals

Although microRNAs (miRNAs) have been revealed as crucial modulators in tumor metastasis and target therapy, our understanding of their roles in metastatic renal cell carcinoma (mRCC) and Sunitinib treatment is limited. Here, We focused on 2 published microarray data to select out our anchored miRNA which was downregulated after Sunitinib treatment while upregulated in metastasis RCC tissues. Then we discovered that treating with Sunitinib, the targeted receptor tyrosine kinase inhibitor (TKI), inhibited renal cell migration and invasion via attenuating the expression of miR-452-5p. The novel identified miR-452-5p was upregulated and associated with poor prognosis in RCC. Preclinical studies using multiple RCC cells and xenografts model illustrated that miR-452-5p could promote RCC cell migration and invasion in vitro and in vivo. Mechanistically, P65 could directly bind to the miR-452-5p promoter and thus transcriptionally induce miR-452-5p expression, which led to post-transcriptionally abrogate SMAD4 expression, thus inhibition of its downstream signals including SMAD7 and EMT (Epithelial-Mesenchymal Transition) associated genes. Our study presented a road map for targeting this newly identified miR-452-5p and its SMAD4/SMAD7/EMT signals pathway, which imparted a new potential therapeutic strategy for mRCC treatment.


Introduction
Renal cell carcinoma (RCC) is one of the most aggressive human urological malignancies, accounting for 2%-3% of adult malignancies [15; 39]. One third of patients have metastatic lesions detected at primary diagnosis, and 30% eventually develop metastatic renal cell carcinoma (mRCC) after surgery [2]. Because of its resistance to radiotherapy and chemo-therapy, surgical resection remains the unique effective treatment [1; 25]. Hence, a better understanding of the detailed mechanisms behind the pathogenesis of RCC and more effective treatment strategies are urgently required.
MicroRNAs (miRNAs), a group of small noncoding RNAs of about 20-24 nucleotides in length, negatively regulate gene expression [11; 30]. Accumulating evidence has demonstrated the involvement of miRNAs contributing to multiple metastatic steps in various human cancers [8; 38]. Notably, plenty of miRNAs, such as miR-27a, miR-141 and miR-205 were reported to exert their roles in RCC cell invasion and metastasis [3; 18; 22]. In addition, our group already attested that miR-646 could suppress RCC cell migration and invasion [21]. Previous studies demonstrated that miR-452-5p acted as a tumor inducer or suppressor in multiple human cancers, involving osteosarcoma, breast cancer, non-small cell lung cancer, hepatocellular carcinoma and prostate cancer [9; 12; 20; 47; 49]. Nevertheless, whether miR-452-5p is implicated in the metastasis of RCC is still need to be investigated.
Sunitinib, as the first-choice tyrosine kinase inhibitor (TKI) in mRCC, provides clinical benefits for patients with mRCC [4; 6; 27]. In a phase II study of Intermittent Sunitinib in mRCC, a total of 20 patients with a median decrease in tumor burden of 45% (range 13% to 86%) entered the intermittent phase. Most patients exhibited a stable sawtooth pattern of tumor burden (TB) reduction while receiving sunitinib and TB increase while not receiving sunitinib, and metastatic progression-free survival and overall survival were 22.4 and 34.8 months, respectively [32]. Based on this finding above, Sunitinib is recognized one of the standard therapies in mRCC. However, most of patients with mRCC will develop resistance to the drug after receiving the treatment eventually [33]. In order to predict treatment response, we urgently need promising molecular biomarkers for rational indication of TKIs in patients with mRCC. Notably, our team previously reported that Sunitinib repressed RCC progression via inducing LncRNA-SARCC, which suggested that other potential non-coding RNAs were involved in RCC-Sunitinib treatment [46]. Similarly, another preclinical study reported that the efficacy of Sunitinib could be increased by miR-32-5p [42]. Then we focused on the relationship between Sunitinib and miRNAs, which the potential ability of miRNAs to predict response to TKIs in mRCC has not drawn enough attention.
P65, as a subunit of NF-КB, targets a large number of genes and plays an important role in the regulation of apoptosis, tumorigenesis, inflammation, and various autoimmune diseases [40]. Increased P65 activation was reported to be implicated in the development of renal cell carcinoma metastasis, promoting metastasis and progression of many cancers through EMT [37]. Recently, EMT was demonstrated to be a major mechanism responsible for invasion and metastasis of cancers [31].
However, the molecular mechanism downstream of P65 has not been fully elucidated in RCC.
In our work, we demonstrated that miR-452-5p acted as a potential therapeutic target for Sunitinib and novel tumor contributor, which promoted renal cancer progression both in vitro and in vivo. Furthermore, the transactivation of miR-452-5p in RCC was induced by P65. In addition, we explored that miR-452-5p directly bind to SMAD4 and suppress SMAD4 expression, thereby regulating SMAD4/SMAD7/EMT signaling pathway and finally enhancing RCC cell invasion and metastasis. Here, we identified miR-452-5p as a novel tumor inducer as well as a potential diagnostic or prognostic marker in mRCC.

Tissue samples
Tumor samples and paired normal tissues from RCC patients were obtained from the Department of Urology, Shanghai Tenth People's Hospital, Tongji University (Shanghai, China). The fresh tissues were kept in liquid nitrogen to protect the RNA from degradation. The current study was approved by the ethics committee of Shanghai Tenth People' s Hospital.

Cell culture and transfection
The human RCC cell lines, OSRC-2, SW839, A498, SN12-PM6 and were originally purchased from Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Diego, California, USA) with 1% penicillin/streptomycin (P/S, ScienCell, San Diego, California, USA). All cells described above were cultured at 37 °C in the humidified 5% CO2 environment.

Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from tissues or cells using Trizol reagent (Invitrogen). cDNAs were synthesized with PrimeScript RT reagent Kit (Takara, Kusatsu, Japan).  Supplementary   Table S1. The expression levels of miRNA were normalized to endogenous small nuclear RNA U6, and the expression levels of mRNA were normalized to endogenous control GAPDH. The 2 -∆∆Ct method was used to analyze the expression levels normalized to the endogenous control.

Wound-healing assay and invasion assay
To analyze the migration, indicated cells were plated in six-well plates. Streaks across the plate were created in the monolayer with a pipette tip. Progression of migration was observed and photographed at 0h and 24 h after wounding. The data shown were representative micrographs of wound-healing assay of the indicated cells. The invasive capability of RCC cells was determined by the transwell assay. The membrane was coated with the matrigel (200 ng/mL; BD Biosciences). Then RCC cells were harvested and seeded with serum-free DMEM into the upper chambers at 5 × 10 4 cells/well, and the bottom chambers contained DMEM with 10% FBS, and then transwells incubated for 24 h at 37 °C. Following incubation, the invasive cells invading into the lower surface of the membrane were fixed by 4% paraformaldehyde and stained with 1% toluidine blue. Experiments were repeated at least three times with similar results.
Studies on animals were conducted with approval from the ethics committee of Shanghai Tenth People' s Hospital.

Chromatin immunoprecipitation assay (ChIP)
Cells were crosslinked with 4% formaldehyde for 10 min followed by cell collection and sonication with a predetermined power to yield genomic DNA fragments of 300 bp long. Lysates were precleared sequentially with normal rabbit IgG and protein A agarose. Anti-P65 antibody (2.0 μg) was added to the cell lysates and incubated at 4 °C overnight. For the negative control, IgG was used in the reaction. Specific primer sets designed to amplify a target sequence within human miR-452-5p promoter were listed in Table S1; PCR products were analyzed by agarose gel electrophoresis.

Luciferase reporter assay
To confirm whether AR and HOXD9 could increase miR-452-5p promoter activity, Only cells with clear tumor cell morphology were scored.

Statistical analysis
Results are expressed at least 3 independent experiments. Using the GraphPad Prism statistical program, data were analyzed using ANOVA or Student' t test unless otherwise specified. P values <0.05 were considered significant.

Sunitinib abrogates RCC cell invasion and metastasis via depressing miR-452-5p
Our team has previously reported that Sunitinib remarkably blunted RCC progression via inducing LncRNA-SARCC [46]. In an attempt to further explore whether Sunitinib inhibited RCC cell invasion and metastasis in a miRNA-dependent manner, we first focused on 2 microarray data, GSE32099 (differentially expressed miRNAs in peripheral blood under sunitinib treatment, Table S2) and GSE37989 (metastasis-associated miRNAs, Table S3) through searching GEO datasets ( Figure   1A). Next we selected out top 10 common miRNAs, which were downregulated after Sunitinib treatment while upregulated in metastasis tissues ( Figure 1B). Notably, two potential candidate miRNAs (miRNA-452-5p and miRNA-605-5p) were selected on the basis of their involvement in RCC tumorigenesis by using OncomiR, an online resource for exploring miRNA dysregulation in cancer (Supplementary Figure A). As shown in Figure 1C  Consistently, wound healing assay also indicated that miR-452-5p overexpression saliently enhanced cell migration than miR-NC group (Supplementary Figure F and G).
Next, we prepared both cells lines with miR-452-5p mimic and miR-NC and then treated them with 10μM Sunitinib for 24h. As shown in Figures 1D and E, cells with miR-452-5p mimic performed more migratory capability compared with cells with control using wound-healing assay. Similarly, cells transfected with miR-452-5p mimic owned more invasive ability than those with control group (Figure 1F and G). Both pro-metastatic phenotype above elicited that miR-452-5p was potential one of the key mediators to influence Sunitinib efficacy.
To further confirm that miR-452-5p might act as a metastatic-promoting miRNA, we then established miR-452-5p with Sunitinib in a xenograft model. OSRC-2 cells with firefly luciferase expression were transfected with miR-NC or miR-452-5p under Sunitinib 10μM treatment. The stable clones were injected into left renal capsule of nude mice and metastatic sites were measured. As shown in the Figure 1H-I, Sunitinib blunted metastatic sites, which was partially reversed by miR-452-5p.
Taken together, the results above suggested that miR-452-5p might be involved in Sunitinib repressing RCC invasion and metastasis.

miR-452-5p is associated with poor prognosis of RCC
To further understand the expression and clinical relevance of miR-452-5p in RCC, the tumor development and survival outcome data were collected from The Cancer Genome Atlas (TCGA) database (http://cancergenome.nih.gov/). The results illustrated that the expression of miR-452-5p was pronouncedly increased in RCC tumor tissue compared to normal tissues (p<0.001) ( Figure 1J and Table S4). In addition, qRT-PCR was performed in renal tumors and paired non-cancerous tissues from 20 RCC patients. The results demonstrated that the expression of miR-452-5p was substantially increased in RCC compared to adjacent non-cancerous tissues (p<0.05) ( Figure 1K and L). As expected, the expression of miR-452-5p was remarkably expressed in various RCC cell lines, involving SW839, A498, SN12-PM6 and OSRC-2 compared to normal adult human kidney HK-2 cells (p<0.01) ( Figure   1M). Furthermore, receiver operating characteristics (ROC) analysis revealed that miR-452-5p might serve as a useful biomarker for discriminating RCC from normal tissues with an area under the ROC curve (AUC) of 0.919 (95% CI, 0.708-0.993; Figure 1N). Next, we analyzed the association between miR-452-5p expression and the clinicopathological characteristics from 102 RCC patients (Table S5). Correlation regression analysis showed that high expression of miR-452-5p was significantly correlated with metastasis (p=0.043). We further performed univariate and multivariable logistic regression models to analyze the correlation of miR-452-5p levels with overall survival of RCC patients. Patient characteristics were provided in Multivariate analysis indicated that a higher miR-452-5p expression (HR=1.58; 95%CI=1.07-2.31; p=0.020) were markedly associated with overall survival. In addition, Kaplan-Meier survival analysis indicated that RCC patients with the lower levels of miR-452-5p had longer overall survival than those with the higher levels of miR-452-5p (p<0.05) ( Figure 1O and Table S7).
In conclusion, the above data indicated that higher level of miR-452-5p was obviously correlated with poor prognosis of RCC.  Figure 2L).

miR-452-5p-inhibition silences the development of pro-metastatic phenotype
Taken together, the above data identified that miR-452-5p functioned as a metastasis-promoting miRNA in RCC.

miR-452-5p level is maximized by P65 through directly binding its promoter
To investigate the mechanism responsible for the up-regulation of miR-452-5p in RCC, we predicted 4 potential transcription factors, involving AR, P65 (RELA), HOXD9 and POU2F2, in the promoter region of miR-452-5p using PROMO, and using two miRNA target-predicting algorithms including Target Scan and miTar to narrow the candidates ( Figure 3A). Then, UALCAN, an interactive web resource for analyzing cancer transcriptome data, was used to assess the above 4 transcription factors. Especially, AR and HOXD9 were ruled out due to their lower expression in RCC tissues than in normal renal tissues (Supplementary Figure H). Interestingly, data from TCGA database confirmed positive correlation between P65 and miR-452-5p expression (p<0.001) ( Figure 3F and Table S8). In parallel, P65 expression was high expressed in RCC tumor tissues than in normal tissues (p<0.001) ( Figure 3G and Table S9). In addition, Immunohistochemical (IHC) staining of P65 protein level indicated a higher expression in RCC tissues than in adjacent normal tissues ( Figure 3H). Notably, TCGA database also showed higher P65 levels were obviously correlated with worse TNM stage ( Figure 3I and Table S10). Kaplan-Meier survival analysis certified that patients with higher P65 levels had poor overall survival time than those with lower P65 levels from TCGA database (p<0.05) ( Figure 3J and Table S11).
In brief, these results above corroborated that P65 were a carcinogenic gene and might induced miR-452-5p transcriptional level via directly binding its promoter in RCC.

SMAD4 is the target gene of miR-452-5p and associated with good prognosis of RCC
To further dissect the mechanism underlying miR-452-5p modulating induction of RCC metastasis, we searched for potential downstream genes of miR-452-5p through four different miRNA target-predicting algorithms including TargetMiner, miRTarBase, miRWalk and miRTar, then focused on the one possible candidate target gene SMAD4 ( Figure 4A). TCGA data sets suggested SMAD4 was low expressed in RCC tissues compared with normal tissues (p<0.05) ( Figure 4B and Table S12). In particular, IHC staining of SMAD4 expression in RCC tissues and normal tissues also elucidated the same tendency with the above results ( Figure 4C). Additionally, Kaplan-Meier survival analysis proved that patients with lower SMAD4 levels had poor overall survival time than those with higher SMAD4 levels from TCGA database (p<0.05) ( Figure 4D and Table S13). Data from TCGA database revealed that there is a negative correlation between miR-452-5p and SMAD4, in keeping with the notion that miRNAs negatively regulate gene expression ( Figure 4E and Table S14). In order to verify the speculation, by a computational prediction of miRNA databases, we identified three putative binding sites of miR-452-5p with high complementarity in SMAD4 promoter region ( Figure 4F). To identify whether Wt-A, Wt-B or Wt-C was functional, luciferase reporter assays was performed and results demonstrated that all of these predicted binding sites in the promoter region of SMAD4 were functional.
( Figure 4G). As a former study described, SMAD7 might be a feasible downstream gene of SMAD4 [28]. Consequently, WB analysis was used to detect the SMAD4 and SMAD7 protein level. When we introduced miR-452-5p mimic into OSRC-2 and SW839 cell lines, the decrease in SMAD4 and SMAD7 protein was confirmed ( Figure   4H). Conversely, both inhibited miR-452-5p and LNA-miR-452-5p markedly enhanced their protein level ( Figure 4I and J). Histological analysis of SMAD7 protein status in RCC tissues and normal tissues also elucidated that SMAD7 was potently downregulated in RCC tissues but was nearly undetectable in normal renal tissues ( Figure 4K).
Together, the data above revealed that miR-452-5p directly targeted SMAD4 and minimized SMAD4 and SMAD7 expression in RCC cells.

SMAD4 recapitulates the effects of miR-452-5p in RCC cells
To examine whether SMAD4 could suppress cell invasion and metastasis in RCC, we performed gene set enrichment analysis with published gene array of metastatic RCC signatures (GSE12606), and results revealed that SMAD4 expression was negatively correlated with RCC cell migration, invasion and metastasis ( Figure 5A). As previously described, we observed that miR-452-5p remarkably increased the cell migration and invasion of OSRC-2 and SW839 cells compared with mock. Importantly, an interruption approach with oe-SMAD4 partially reversed the effects of miR-452-5p on cell migration and invasion ( Figure 5B and C). As we expected, WB confirmed that oe-SMAD4 recapitulated the effect of blocked SMAD4 and SMAD7 protein caused by miR-452-5p in both cell lines ( Figure 5D).
Together, these results above suggested that miR-452-5p promoted RCC cell invasion and metastasis through suppressing SMAD4.

miR-452-5p promotes RCC metastasis through targeting SMAD4 in vivo
To further validate that miR-452-5p might act as a tumor inducer in vivo, we inoculated different clones of OSRC-2 cells. In this model system, sh-miR-452-5p cells as well as its control cells were inoculated into the left kidney capsule of xenograft. As shown in Figure 6A, a dramatic induction of luciferase expression in tumors of both groups was detected by in vivo imaging system (IVIS) as early as the 2th week. Figures 6B and C showed promotion of tumor metastases in the sh-miR-NC group compared with the sh-miR-452-5p group after 4 weeks (Figures 6B and C). Furthermore, sh-miR-452-5p attenuated lung, liver, spleen and right renal metastases ( Figure 6D). In parallel, IHC staining substantiated SMAD4 and SMAD7 protein levels of sh-miR-452-5p group increased compared with sh-miR-NC group in renal tumor tissues from nude mice ( Figure 6E). Conversely, transfection of miR-452-5p into OSRC-2 cells led to sufficiently enhanced metastasis of orthotopic xenograft tumors, whereas oe-SMAD4 into miR-452-5p upregulated cells mostly abolished this induction ( Figure 6F-H).
Together, results from Figure 6 illustrated that SMAD4-dependent miR-452-5p functioned as a critical tumor-metastasis promoter in RCC.

Discussion
In our study, we first reported that miR-452-5p, induced by P65, was a potential therapeutic target for Sunitinib and associated with mRCC poor prognosis.
MiR-452-5p in RCC samples and cell lines was high expressed compared with that in surrounding non-tumor tissues as well as HK2 normal cells. Moreover, P65 could directly bind to the promoter of miR-452-5p and transcriptionally induce miR-452-5p expression. Consistently, miR-452-5p elevated cell migration and invasion in RCC cell lines and promoted RCC progression through targeting SMAD4/SMAD7/EMT signals.
All these results supported the conclusion that miR-452-5p acted as a tumor inducer and a metastasis-promoting miRNA in mRCC.
Previously, some reports demonstrated that miR-452-5p might function as a tumorigenesis-promoting in hepatocellular carcinoma and contribute to the docetaxel resistance of breast cancer cells [12; 49]. On the other hand, it has also been reported that expression of miR-452-5p inhibited metastasis in osteosarcoma, non-small cell lung cancer, and down-regulation of miR-452-5p is associated with adriamycin-resistance in breast cancer cells [13; 20; 47]. The possible reason for the opposite role of miR-452-5p in various human cancers or under different treatment is distinguished. Importantly, the relationship between mRCC and miR-452-5p as well as its response to Sunitinib remained unknown. In this work, our studies reported that Sunitinib attenuated miR-452-5p to impact its therapeutic effect, and miR-452-5p enhanced the development of pro-metastatic phenotype in mRCC.  [35]. Other reports proved that P65 might serve as an activating transcription factor in several types of human cancers [19; 23]. Our conclusions supported these findings above that P65 could directly bind to the miR-452-5p promoter and thus transcriptionally induce miR-452-5p expression. Conversely, it was also reported that P65 could attenuate transcriptional activity under a certain condition.
Raman P. NAGARAJAN et al. validated that P65 was able to inhibit the SMAD7 promoter activity [28], which, in some extent, was also coincided with our result that P65 induced miR-452-5p transcription, which targeted SMAD4 and repressed SMAD7 expression. Our study further illustrated that high P65 expression was obviously correlated with higher clinical TMN stage and contributed to poor prognosis in RCC patients. All these results concluded that P65 served as a tumor-inducer in RCC.
As a member of SMAD family, SMAD4 is a critical component of TGF-β signaling and gets involved in MAPK, CDK and PI3K signaling [29; 48]. As we all know, SMAD4 has been recognized as a tumor suppressor gene in a good deal of cancers, and recent study has presented that SMAD4 suppressed the progression of RCC by targeting various downstream genes [14; 24; 26; 44]. Moreover, another report suggested that the loss of SMAD4 repressed SMAD7 transcription leading to a loss of functional protein during renal inflammation and fibrosis [28]. Here, we elucidated that miR-452-5p directly targeted SMAD4, and as a downstream gene, SMAD7 was also repressed. Furthermore, SMAD7 has been reported to play a pivotal role in EMT, which finally contributed to tumor metastasis [36; 41; 43]. Our study presented a road map that miR-452-5p facilitated RCC invasion and metastasis through SMAD4/SMAD7/EMT Signals.
Although Sunitinib is the first-line treatment for mRCC, the clinical benefit of sunitinib is limited, and the vast majority of mRCC patients under sunitinib treatment ultimately develop disease progression because of the acquisition of resistance. Several studies reported that miRNAs play a crucial role in alterting the sensitivity to Sunitinib in multiple tumors, indicating that miRNAs are potential therapeutic targets for Sunitinib， and miRNA modulation combined with Sunitinib as a novel therapeutic strategy was under exploring [5; 10; 16; 34]. In our study, miR-452-5p was confirmed to determine the sensitivity to Sunitinib, and high expression of miR-452-5p was suggested to contribute to Sunitinib resistance. Therefore, it could be used to divide mRCC patients into different responsive groups to save both money and time for the non-responsive patients and enhance Sunitinib treatment efficacy through up-regulating the expression of miR-452-5p.
In this study, we conclude that Sunitinib efficacy is potentially connected with miR-452-5p, which acts as an efficient metastasis-promoter through SMAD4/SMAD7/EMT signals in RCC patients. This finding points to a novel therapeutic target to maximize Sunitinib efficiency in RCC, and helps us to better suppress RCC progression via targeting this newly identified signal pathway.