Regulation of Oncogenic Targets by the Tumor-Suppressive miR-139 Duplex (miR-139-5p and miR-139-3p) in Renal Cell Carcinoma

We previously found that both the guide and passenger strands of the miR-139 duplex (miR-139-5p and miR-139-3p, respectively) were downregulated in cancer tissues. Analysis of TCGA datasets revealed that low expression of miR-139-5p (p < 0.0001) and miR-139-3p (p < 0.0001) was closely associated with 5-year survival rates of patients with renal cell carcinoma (RCC). Ectopic expression assays showed that miR-139-5p and miR-139-3p acted as tumor-suppressive miRNAs in RCC cells. Here, 19 and 22 genes were identified as putative targets of miR-139-5p and miR-139-3p in RCC cells, respectively. Among these genes, high expression of PLXDC1, TET3, PXN, ARHGEF19, ELK1, DCBLD1, IKBKB, and CSF1 significantly predicted shorter survival in RCC patients according to TCGA analyses (p < 0.05). Importantly, the expression levels of four of these genes, PXN, ARHGEF19, ELK1, and IKBKB, were independent prognostic factors for patient survival (p < 0.05). We focused on PXN (paxillin) and investigated its potential oncogenic role in RCC cells. PXN knockdown significantly inhibited cancer cell migration and invasion, possibly by regulating epithelial–mesenchymal transition. Involvement of the miR-139-3p passenger strand in RCC molecular pathogenesis is a new concept. Analyses of tumor-suppressive-miRNA-mediated molecular networks provide important insights into the molecular pathogenesis of RCC.


Introduction
Renal cell carcinoma (RCC) accounts for approximately 3% of adult malignancies, with more than 330,000 cases newly diagnosed in 2018 and more than 100,000 deaths annually [1]. Clear cell RCC is the most common histological subtype of RCC, accounting for approximately 75% of all cases [2]. Approximately 20-30% of RCC patients have metastatic lesions at diagnosis, and thus the 5-year survival rate of these patients is less than 20% [2,3]. In addition, more than 20% of patients develop metastases during the postoperative follow-up period [4]. Elucidation of the molecular mechanisms involved in distant metastasis of RCC will contribute to the development of new diagnostic and therapeutic strategies.

Functional Assays (Cell Proliferation, Migration, and Invasion Assays) in RCC cells
XTT assay for cell proliferation, wound healing assay for migration, and Matrigel chamber assay for invasion were performed using RCC cells as described previously [13][14][15][16]. The reagents used are listed in Table S1.

Identification of miR-139-5p and miR-139-3p Gene Targets in RCC Cells
The search strategy used to identify miRNA targets is summarized in Figure S1. The expression profiles of miR-139-5p/miR-139-3p-transfected A498 and 786-O cells from the Gene Expression Omnibus (GEO) GSE129043 dataset were used.

In silico Analysis of RCC Public Databases
For comparison of gene expression levels between normal and cancer tissues, we utilized the GSE36895 datasets obtained from GEO. GSE36895 contains mRNA microarray data from clear cell RCC tumors, normal kidney tissues, and mouse tumor graft samples obtained using Affymetrix U133 Plus 2.0 whole-genome chips (Affymetrix, Santa Clara, CA, USA). Expression is shown as signal intensities, and for each gene with multiple probes, the mean intensity value was used. In addition, RNA sequencing data from TCGA-Kidney renal clear cell carcinoma (TCGA-KIRC) RNA sequencing datasets were utilized to re-evaluate gene expression levels in normal versus tumor samples.

Clinicopathological Analysis of RCC
For Kaplan-Meier analyses of overall survival, we downloaded TCGA clinical data (Firehose Legacy) from cBioportal (https://www.cbioportal.org). The patients were divided into two groups by median expression for each gene, according to the data collected from OncoLnc (http://www.oncolnc.org). R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria) was used for the analyses.
Multivariate Cox regression analyses were also performed using TCGA-KIRC clinical data and survival data according to the expression level of each gene from OncoLnc to identify factors associated with RCC patient survival. In addition to gene expression, the tumor stage, pathological grade, and age group were evaluated as potential independent prognostic factors. The multivariate analyses were performed using JMP Pro 15.0.0 (SAS Institute Inc., Cary, NC, USA).
We performed gene set enrichment analysis (http://software.broadinstitute.org/gsea/index.jsp) to obtain lists of differentially expressed genes between high and low PXN expression groups in the TCGA-KIRC cohort.

Plasmid Construction and Dual-Luciferase Reporter Assays
psiCHECK-2 plasmid vectors (Promega, Madison, WI, USA) harboring the wild-type or a deletion sequence of the miR-139-3p-binding site within the 3 -UTR of PXN were prepared. The predicted binding site sequence was obtained from the TargetScanHuman database (release 7.2). Cells were co-transfected with miR-139-3p and the plasmid vectors for 36 h, after which firefly and Renilla luciferase activities in cell lysates were measured consecutively using the Dual-Luciferase Reporter Assay System (Promega). Renilla luciferase activities are expressed as normalized values to firefly luciferase activities. The dual-luciferase reporter assay procedure was described in our previous studies [13][14][15][16].

Statistical Analyses
For comparisons among multiple groups, Dunnett's test was applied. The statistical analyses were performed using JMP Pro 15. Significance levels were set to p < 0.05 if not otherwise mentioned.

Analysis of miR-139-5p and miR-139-3p Expression Levels in Clinical RCC Tissues and Their Clinical Significance
The expression levels of miR-139-5p and miR-139-3p were evaluated in RNA sequencing data from RCC tissue samples obtained from TCGA. miR-139-5p and miR-139-3p were significantly downregulated in RCC tissues compared with normal tissues (p < 0.0001 and p < 0.0001, respectively; Figure 1A). Cohort analyses using the TCGA-KIRC datasets revealed that low expression of miR-139-5p and miR-139-3p was associated with poorer survival in patients with RCC (p < 0.0001 and p < 0.0001, respectively; Figure 1B). Gels (Bio-Rad), and transferred to Immun-Blot PVDF membranes (Bio-Rad). Blocking was performed with Blocking One (Nacalai Tesque) for 30 min. The antibodies used in this study are shown in Table  S1.

Statistical Analyses
For comparisons among multiple groups, Dunnett's test was applied. The statistical analyses were performed using JMP Pro 15. Significance levels were set to p < 0.05 if not otherwise mentioned.

Analysis of miR-139-5p and miR-139-3p Expression Levels in Clinical RCC Tissues and Their Clinical Significance
The expression levels of miR-139-5p and miR-139-3p were evaluated in RNA sequencing data from RCC tissue samples obtained from TCGA. miR-139-5p and miR-139-3p were significantly downregulated in RCC tissues compared with normal tissues (p < 0.0001 and p < 0.0001, respectively; Figure 1A). Cohort analyses using the TCGA-KIRC datasets revealed that low expression of miR-139-5p and miR-139-3p was associated with poorer survival in patients with RCC (p < 0.0001 and p < 0.0001, respectively; Figure 1B).

Tumor-Suppressive Functions of miR-139-5p and miR-139-3p in RCC Cells
To investigate the tumor-suppressive functions of miR-139-5p and miR-139-3p in RCC cells, we assessed cell proliferation, migration, and invasion after ectopic transfection of miR-139-5p and miR-139-3p into A498 and 786-O cells. Ectopic expression of the two miRNAs did not significantly affect the proliferation of RCC cells (Figure 2A). In contrast, the expression of these miRNAs significantly inhibited the migration and invasive abilities of RCC cells ( Figure 2B,C).

Identification of Putative Oncogenic Targets Regulated by miR-139-3p and miR-139-3p in RCC Cells
To identify the genes regulated by miR-139-5p and miR-139-3p in RCC cells, we integrated three datasets. First, we obtained RNA microarray data from miR-139-5p-or miR-139-3p-transfected A498 or 786-O cells (GSE129043). Second, we used data from the TargetScanHuman database (release 7.2) to obtain annotated putative targets regulated by each miRNA strand. Third, we extracted genes highly expressed in clinical specimens of RCC from GSE36895.

Identification of Putative Oncogenic Targets Regulated by miR-139-3p and miR-139-3p in RCC Cells
To identify the genes regulated by miR-139-5p and miR-139-3p in RCC cells, we integrated three datasets. First, we obtained RNA microarray data from miR-139-5por miR-139-3p-transfected A498 or 786-O cells (GSE129043). Second, we used data from the TargetScanHuman database (release 7.2) to obtain annotated putative targets regulated by each miRNA strand. Third, we extracted genes highly expressed in clinical specimens of RCC from GSE36895.
A schematic of the strategy used to narrow down the gene list is shown in Figure S1. A total of 19 genes regulated by miR-139-5p and 22 by miR-139-3p were finally selected (Table 1A,B).

Clinical Significance of miR-139 Target Genes in RCC Pathogenesis
Kaplan-Meier analyses of 5-year overall survival were performed according to high versus low expression of miR-139 target genes. The high expression of two miR-139-5p (PLXDC1 and TET3) and six miR-139-3p target genes (PXN, ARHGEF19, ELK1, CSF1, IKBKB, and DCBLD1) was found to be related to better 5-year overall survival rates of the patients (p < 0.05, Table 1A,B, and Figure 3). Furthermore, multivariate analyses identified four genes (PXN, ARHGEF19, ELK1, and IKBKB) as independent prognostic factors for patient survival (p < 0.05; Figure 4). We validated the expression levels of these target genes using TCGA-KIRC RNA sequencing dataset. All target genes except ARHGEF19 were upregulated in cancer tissues in this cohort ( Figure 5).
For subsequent analyses, we focused on miR-139-3p (the passenger strand) target genes, according to our previous studies emphasizing the novel roles of miRNA passenger strands. Among these target genes, we focused on PXN, which showed the most significant association with RCC patient survival in the multivariate analysis and has been reported previously as an oncogene in other types of cancers [18,19].  Furthermore, multivariate analyses identified four genes (PXN, ARHGEF19, ELK1, and IKBKB) as independent prognostic factors for patient survival (p < 0.05; Figure 4). We validated the expression levels of these target genes using TCGA-KIRC RNA sequencing dataset. All target genes except ARHGEF19 were upregulated in cancer tissues in this cohort ( Figure 5).

Direct Regulation of PXN by miR-139-3p in RCC Cells
In RCC cells transfected with miR-139-3p, both PXN mRNA and protein levels were significantly downregulated ( Figure 6A,B). Western blot images are shown in Figure S2. To validate that miR-139-3p binds directly to PXN to downregulate its expression, we performed dual-luciferase reporter assays. A498 and 786-O cells were co-transfected with plasmid vectors and miR-139-3p. We used two different plasmid vectors: one containing the partial wild-type sequence of the miR-139-3p-binding site predicted by TargetScanHuman database ("wild-type sequence" in Figure 6C) and the other containing this sequence lacking the binding site ("deletion-type sequence" in Figure 6C). Luciferase activity was significantly reduced in cells transfected with the wild-type sequence but not in cells  For subsequent analyses, we focused on miR-139-3p (the passenger strand) target genes, according to our previous studies emphasizing the novel roles of miRNA passenger strands. Among these target genes, we focused on PXN, which showed the most significant association with RCC patient survival in the multivariate analysis and has been reported previously as an oncogene in other types of cancers [18,19].

Direct Regulation of PXN by miR-139-3p in RCC Cells
In RCC cells transfected with miR-139-3p, both PXN mRNA and protein levels were significantly downregulated ( Figure 6A,B). Western blot images are shown in Figure S2. To validate that miR-139-3p binds directly to PXN to downregulate its expression, we performed dual-luciferase reporter assays. A498 and 786-O cells were co-transfected with plasmid vectors and miR-139-3p. We used two different plasmid vectors: one containing the partial wild-type sequence of the miR-139-3p-binding site predicted by TargetScanHuman database ("wild-type sequence" in Figure 6C) and the other containing this sequence lacking the binding site ("deletion-type sequence" in Figure 6C). Luciferase activity was significantly reduced in cells transfected with the wild-type sequence but not in cells transfected with the deletion sequence ( Figure 6D). These results suggest that miR-139-3p directly binds to the 3'-UTR of PXN. transfected with the deletion sequence ( Figure 6D). These results suggest that miR-139-3p directly binds to the 3'-UTR of PXN.

PXN Knockdown Assays in RCC Cells
We assessed the oncogenic functions of PXN in RCC cells by performing knockdown using two small interfering RNAs (siRNAs) targeting PXN. The mRNA and protein levels of PXN were successfully downregulated by either siRNAs in A498 and 786-O cells ( Figure 7A,B). Western blot images are shown in Figure S2.
Next, functional assays were performed in RCC cells transfected with these siRNAs. Similar to miR-139-3p transfection, transfection of the siRNAs did not suppress cell proliferation ( Figure 7C). However, cell migration and invasion were significantly suppressed by siRNA transfection in both cell lines (Figures 7D,E).

PXN Knockdown Assays in RCC Cells
We assessed the oncogenic functions of PXN in RCC cells by performing knockdown using two small interfering RNAs (siRNAs) targeting PXN. The mRNA and protein levels of PXN were successfully downregulated by either siRNAs in A498 and 786-O cells ( Figure 7A,B). Western blot images are shown in Figure S2.

PXN-Mediated Pathways in RCC Cells
We performed gene set enrichment analysis to determine differentially expressed genes between the high and low PXN expression groups of the TCGA-KIRC cohort. In support of the functional assays using siRNAs, the most enriched signaling pathway in the high PXN expression group was "epithelial-mesenchymal transition" (Figure 8 and Table S3). Other significantly enriched pathways (FDR q-value < 0.05) were "hypoxia", "KRAS signaling", "myogenesis", "angiogenesis", and "apical junction", supporting the hypothesis that PXN is related to the metastatic ability of cancer cells (Figure 7 and Table S2).  Next, functional assays were performed in RCC cells transfected with these siRNAs. Similar to miR-139-3p transfection, transfection of the siRNAs did not suppress cell proliferation ( Figure 7C). However, cell migration and invasion were significantly suppressed by siRNA transfection in both cell lines ( Figure 7D,E).

PXN-Mediated Pathways in RCC Cells
We performed gene set enrichment analysis to determine differentially expressed genes between the high and low PXN expression groups of the TCGA-KIRC cohort. In support of the functional assays using siRNAs, the most enriched signaling pathway in the high PXN expression group was "epithelial-mesenchymal transition" (Figure 8 and Table S3). Other significantly enriched pathways (FDR q-value < 0.05) were "hypoxia", "KRAS signaling", "myogenesis", "angiogenesis", and "apical junction", supporting the hypothesis that PXN is related to the metastatic ability of cancer cells (Figure 7 and Table S2).
A recent in silico study (analysis of molecular profiles in more than 5200 patient samples from 14 different cancers) revealed that both strands of miRNAs are functional across different cancer types. For example, downregulation of both strands of miR-30a and miR-145 was frequently observed in multiple cancers, and their downregulation enhanced aberrant expression of cell cycle-related genes. These malignant events were found to affect the prognosis of cancer patients [17]. Simultaneous analysis of both strands of miRNA duplexes will lead to elucidation of a new molecular mechanism in cancer cells.
In contrast to miR-139-5p, few studies have evaluated the importance of miR-139-3p in cancer because it is a passenger strand. Recently, it was reported that miR-139-3p directly regulates KDM5B (lysine demethylase 5B), a key regulator of histone 3 lysine 4 demethylation in laryngeal squamous cell carcinoma [28]. Ectopic expression of miR-139-3p suppressed cancer cell malignant behavior by inhibiting the Wnt/β-catenin pathway [28]. In the HeLa cervical cancer cell line, miR-139-3p was reduced upon increasing expression of circular RNA hsa_circ_0031288 [29]. In turn, the decreased expression of miR-139-3p resulted in increased expression of B-cell CLL/lymphoma 6, as a target molecule. These regulatory effects promoted malignant transformation of HeLa cells. In bladder cells, both strands of the miR-139 duplex were downregulated in cancer tissues and exhibited tumor-suppressive effects [30]. Interestingly, matrix metalloprotease 11 (MMP11) was directly regulated by both miR-139 strands, and knockdown of MMP11 attenuated the aggressive phenotype of bladder cancer cells [30]. Our study is the first to show that miR-139-3p (the passenger strand) is involved in RCC pathogenesis, and both strands of the miR-139 duplex are pivotal players in RCC oncogenesis.
We are also interested in the presence of oncogenes controlled by tumor-suppressive miR-139-5p and miR-139-3p in RCC cells. A total of 19 and 22 genes were identified as putative targets of miR-139-5p and miR-139-3p regulation in RCC cells, respectively. Taking advantage of survival data from TCGA datasets, we performed multivariate analysis and found that, among the miR-139 duplex target genes, PXN, ELK1, ARHGEF19, and IKBKB were independent prognostic factors for RCC patient survival. Further genomic analyses of these genes will contribute to elucidating the molecular pathogenesis of RCC.
ELK1 is a well-established transcription factor that is phosphorylated by MAPKs and induces transcription of the c-fos proto-oncogene [31,32]. ARHGEF19 is a RhoGEF that reportedly activates the MAPK pathway and interacts with BRAF in lung cancer [33]. The role of ARHGEF19 in RCC is not well characterized, but the interaction of ARHGEF19 with ELK1 via the MAPK pathway may play a role in RCC development. IKBKB encodes IKK-beta, which phosphorylates the inhibitor of NF-kB, resulting in activation of the NF-kB pathway [34,35]. NF-kB and its downstream inflammatory signaling pathway are closely related to RCC carcinogenesis and aggressiveness [36].
In this study, we focused on PXN (paxillin) because its expression was found to be directly controlled by miR-139-3p in RCC cells and strongly related to RCC molecular pathogenesis. PXN is a focal adhesion scaffold/adaptor protein that contains five LD domains (leucine-aspartate motifs) located at the N-terminus and four cysteine-histidine-rich LIM domains at the C-terminus [37,38]. The LD domains act as a docking site for focal adhesion-related proteins, e.g., Src (tyrosine-protein kinase), FAK (focal adhesion kinase), PAK (p21-activated kinase), and ILK (integrin-linked kinase) [37,38]. The LIM domains act as binding sites for protein-protein interactions [37,38]. The pathways activated by PXN enhance cancer cell malignant progression and metastasis in a wide range of cancers [39]. Moreover, two key proteins in focal adhesion complexes, PXN and integrin B4, directly bind to each other, and this complex enhanced cisplatin resistance in lung cancer cells [40]. Another study showed that PXN phosphorylation may contribute to cisplatin resistance via the ERK-mediated activation of Bcl-2 transcription in lung cancer [41]. PXN-mediated pathways may be therapeutic targets for attenuating drug-resistance in cancer cells.
In our GSEA analysis, PXN-high-expressed RCC specimen was enriched with epithelialmesenchymal transition (EMT) signaling pathways. A previous report has shown the suppressive effect of PXN on EMT pathway in cell lines [42]. These findings suggest the speculation that inhibitions of cancer cell migration and invasion by PXN knockdown in our study are due to the regulation of epithelial-mesenchymal transition.

Conclusions
Both strands of the miR-139 duplex are closely involved in RCC oncogenesis. This is the first report to reveal that miR-139-3p (the passenger strand) acts as a tumor-suppressive miRNA in RCC. Several genes are controlled by miR-139 and contribute to RCC molecular pathogenesis. Notably, the expression of PXN was directly regulated by miR-139-3p, and its overexpression enhanced RCC malignant transformation. Analyses of tumor-suppressive miRNAs (including the passenger strands) will contribute to the elucidation of new molecular networks in RCC.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2227-9059/8/12/599/s1, Figure S1: Schematic of the strategy used to narrow down the miR-139-5p/miR-139-3p target genes, Figure S2: Western blot images, Table S1: Reagents used in this study, Table S2: The number of cases per group for multivariate analysis, Table S3: Gene set enrichment analysis of genes enriched among the differentially expressed genes in the TCGA-KIRC high PXN expression group.

Conflicts of Interest:
The authors declare no conflict of interest.