MiR-223-3p promotes cell proliferation and metastasis by downregulating SLC4A4 in clear cell renal cell carcinoma

MiR-223-3p is significantly upregulated and associated with a poor prognosis in ccRCC patients in TCGA-KIRC

Gene expression data from ccRCC patients were obtained from The Cancer Genome Atlas (TCGA) ccRCC dataset (TCGA-KIRC). Detailed clinicopathological information on these patients is presented in Table 1. The relative expression of miR-223-3p, shown in log2 (FPKM+1) form, ranged from 4.85 to 10.62 units in normal tissues and from 3.59 to 10.92 units in tumor tissues. MiR-223-3p expression was significantly higher in ccRCC tissues than in the corresponding non-cancerous tissues. Similar results were obtained from 71 paired ccRCC tissues and corresponding non-cancerous tissues (Figure 1A). Receiver operator characteristic (ROC) curve analysis demonstrated that miR-223-3p could sufficiently discriminate ccRCC tissues from normal tissues (area under the curve [AUC]: 0.7609, 95% confidence interval: 0.6964 - 0.8254) and from paired normal tissues (AUC: 0.7337, 95% confidence interval: 0.6494 - 0.8180) (Figure 1B). We also screened the database for a classic oncomiRNA (miR-21) and several miRNAs from recently published articles (miR-452, miR-543 and miR-708). We found that miR-223-3p was less sensitive than miR-21, but more sensitive than miR-708 (Supplementary Figure 1).

T-test analysis demonstrated that miR-223-3p expression was higher in patients with lymph node metastases than in those without (Figure 1C). High miR-223-3p expression was also observed in patients with distant metastases (Figure 1D). MiR-223-3p expression was significantly higher in T stages III and IV than in T stages I and II, and was greater in T stage IV than in T stage III (Figure 1E). Higher miR-223-3p levels were associated with more advanced pathological TNM stages (T stage, lymphatic metastasis and distant metastasis grouping) and grades in ccRCC patients (Figure 1F and 1G).

Kaplan-Meier survival analysis was used to determine the correlation of miR-223-3p expression with the overall survival (OS) and disease-free survival (DFS) of ccRCC patients. Patients from TCGA-KIRC were divided into two groups based on the median miR-223-3p level. Patients with higher miR-223-3p expression (> 6.7 units) exhibited shorter OS (Figure 1H, p < 0.0001) and DFS (Figure 1I, p = 0.0346). The expression of miR-223-3p was then compared between patients with longer and shorter OS and DFS (OS-Good and OS-Poor, DFS-Good and DFS-Poor). High miR-223-3p expression indicated a poor prognosis (Supplementary Figure 2). Univariate and multivariate survival analyses demonstrated that miR-223-3p expression was an independent prognostic factor for OS in ccRCC patients (p = 0.000) (Table 2).

These data indicate that miR-223-3p expression is upregulated and is significantly associated with various clinicopathological parameters (such as T stage, lymphatic metastasis and distant metastasis) in ccRCC. High miR-223-3p expression predicted a poor prognosis and thus could be a potential prognostic indicator for ccRCC.

MiR-223-3p promotes renal cancer cell proliferation, migration and invasion in vitro

We next evaluated miR-223-3p levels in ccRCC patients from The First Affiliated Hospital of Xiamen University. The results confirmed our observations in TCGA-KIRC (Figure 2A). We also determined the miR-223-3p levels of three renal cancer cell lines (A498, Caki-1 and ACHN) [27], and found that ACHN cells had the lowest expression, while Caki-1 cells had the highest (Figure 2B).

Then, we used gain- or loss-of-function assays to evaluate the effects of miR-223-3p on the proliferation, migration and invasion of 786-O (RCC) cells and Caki-1 cells. The effects of miR-223-3p mimics and inhibitors are shown in Figure 2C and 2D. MiR-223-3p overexpression significantly enhanced the proliferation of 786-O and Caki-1 cells (Figure 2E), while miR-223-3p knockdown significantly reduced their proliferation (Figure 2F). Knockdown of miR-223-3p significantly repressed cell migration and invasion compared to the negative control (NC) in Transwell assays (Figure 2G and 2H). These data indicate that miR-223-3p promotes the proliferation, migration and invasion of renal cancer cells.

Bioinformatic analysis of miR-223-3p target genes in ccRCC

To identify the potential targets of miR-223-3p, we used two prediction software programs (TargetScan and miRDB) to score predicted molecules. We then selected the top 20 molecules scored by each program, and identified five molecules of interest (Figure 3A): solute carrier family 4, member 4 (SLC4A4), Ras homolog family member B (RHOB), interleukin 6 signal transducer (IL6ST), LIM domain only 2 (LMO2) and F-box and WD repeat domain containing 7 (FBXW7). To narrow the field of prospective candidates, we performed a correlation analysis in TCGA-KIRC. A heat map revealed the correlations between miR-223-3p and the five target molecules (Figure 3B). As shown in Figure 3C, only SLC4A4 and RHOB levels exhibited significant negative correlations with miR-223-3p levels. Then, we compared the mRNA levels of the five target molecules between cancerous and non-cancerous tissues in TCGA-KIRC. SLC4A4 expression was significantly lower in ccRCC tissues than in the corresponding non-cancerous tissues, while RHOB expression was higher in ccRCC tissues (Figure 3D). In addition, miR-223-3p expression exhibited a significant negative correlation with SLC4A4 expression in ccRCC patient samples from our university (Figure 3E).

MiR-223-3p directly binds to SLC4A4

To determine whether miR-223-3p directly binds to SLC4A4, we evaluated the changes in protein and mRNA levels by Western blotting and qRT-PCR, respectively, following cellular treatment with miR-223-3p mimics and inhibitors. In 786-O and Caki-1 cells, miR-223-3p mimics reduced both the protein and mRNA levels of SLC4A4 (Figure 4A and 4B), while miR-223-3p inhibitors had the opposite effects (Figure 4C and 4D).

Then, we determined the effect of miR-223-3p on the 3′-untranslated region (UTR) of SLC4A4 using a luciferase reporter assay. Luciferase reporter constructs containing either wild-type or mutated SLC4A4 binding sequences upstream of the firefly luciferase gene were generated (Figure 4E). Caki-1 and 786-O cells were co-transfected with the reporter vectors and mimics or mimic controls. Luciferase activity was significantly reduced after miR-223-3p mimic co-transfection with WT vectors (Figure 4F). These results suggest that SLC4A4 is a direct target of miR-223-3p.

SLC4A4 is significantly downregulated and associated with a poor prognosis in ccRCC patients in TCGA-KIRC

As SLC4A4 was found to be a direct target of miR-223-3p, we investigated SLC4A4 mRNA levels in TCGA-KIRC. The relative expression of SLC4A4 in log2 (FPKM+1) form ranged from 4.85 to 10.62 units in normal tissues and from 3.59 to 10.92 units in tumor tissues. The expression of SLC4A4 was significantly lower in ccRCC tissues than in non-cancerous tissues (Figure 5A). To confirm the results from TCGA-KIRC, we examined three additional datasets in the Oncomine database (Figure 5B). Low SLC4A4 expression was detected in patients with distant metastases (Figure 5C). SLC4A4 expression was significantly lower in T stage IV than in T stages I, II and III (Figure 5D). Lower SLC4A4 levels were associated with more advanced pathological TNM stages and grades in ccRCC patients (Figure 5E and 5F). Patients with lower SLC4A4 expression exhibited shorter OS (Figure 5G, t-test, p < 0.0001) and DFS (Figure 5H, t-test, p = 0.005). Univariate and multivariate survival analyses indicated that SLC4A4 expression was an independent prognostic factor for OS and DFS in ccRCC patients (Tables 2 and 3).

SLC4A4 is downregulated and involved in the biological pathogenesis of ccRCC

As SLC4A4 was downregulated in ccRCC patients in TCGA-KIRC and was an independent prognostic factor for ccRCC, we tested SLC4A4 expression in ccRCC patient samples from our hospital by quantitative real-time polymerase chain reaction (qRT-PCR) and immunohistochemistry analyses. SLC4A4 was downregulated in ccRCC tissues compared to normal tissues (Figure 6A and 6B). Immunohistochemistry data from the Human Protein Atlas also demonstrated that SLC4A4 was downregulated in ccRCC tissues compared to normal tissues (Figure 6C). Gene set enrichment analysis demonstrated that SLC4A4 expression was associated with the epithelial-mesenchymal transition and Kirsten rat sarcoma viral oncogene (KRAS) signaling (Figure 6D).

Then, we evaluated the function of SLC4A4 in ccRCC by overexpressing SLC4A4 in RCC cell lines. The mRNA and protein levels of SLC4A4 were confirmed to be upregulated in 786-O and Caki-1 cells transfected with plasmid vectors expressing SLC4A4 (Figure 6E and 6F). Overexpression of SLC4A4 significantly repressed the proliferation of 786-O cells compared to the NC group (Figure 6G). Furthermore, SLC4A4 vector transfection markedly inhibited the migration and invasion of Caki-1 cells in Transwell assays (Figure 6H and 6I). These results demonstrate that SLC4A4 may function as a tumor suppressor in ccRCC cells.

SLC4A4 attenuates the function of miR-223-3p in renal cancer cells

To determine whether SLC4A4 could reverse the oncogenic effects of miR-223-3p, we co-transfected renal cancer cells with miR-223-3p mimics and SLC4A4. The mRNA levels of miR-223-3p and SLC4A4 are shown in Figure 7A and 7B, while the protein levels of SLC4A4 and KRAS are shown in Figure 7C. Overexpression of SLC4A4 impaired the ability of miR-223-3p mimics to induce KRAS expression (Figure 7C) and to enhance 786-O and Caki-1 cell proliferation (Figure 7D). The enhancement of Caki-1 cell migration and invasion by miR-223-3p was also reversed by co-transfection with SLC4A4 (Figure 7E and 7F). These data demonstrate that SLC4A4 partially alleviates the oncogenic effects of miR-223-3p in RCC.

Patient samples

Surgical specimens were obtained from 24 patients between 2017 and 2018 at the Department of Urology of The First Affiliated Hospital of Xiamen University. Resected tissues were frozen in liquid nitrogen and stored at -80°C for further experiments. Informed consent was obtained from all patients, and the experimental procedures were approved by the Institutional Review Board of Xiamen University.

RNA extraction and qRT-PCR

Tissue RNA was extracted with the TRIzol reagent (Thermo, Massachusetts, USA) according to the manufacturer’s instructions. The purity and concentration of each RNA sample was tested with a NanoDrop 2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). A qPCR analysis was performed according to the manufacturer’s instructions (LightCycler 480II; Roche, Basel, Switzerland) with SYBR Green mix (Thermo, Massachusetts, USA). The relative levels of miR-223-3p and SLC4A4 were calculated by: 2^-ΔCt (ΔCt = Ct_gene–Ct_normalizer). MiRNA primers were purchased from RiboBio (Guangzhou, China). The primers for SLC4A4 and GAPDH were purchased from GENEWIZ (GENEWIZ, Suzhou, China): SLC4A4, Forward 5’-TTCACGGAACTGGATGAGCT-3’, Reverse 5’-ACTGTGGGAGAGAAGAAGCC-3’; GAPDH, Forward 5’-GAGTCAACGGATTTGGTCGT-3’, Reverse 5’-GACAAGCTTCCCGTTCTCAG-3’.

Cell culture and transient transfection

HK2, 786-O, ACHN and Caki-1 cells were purchased from The American Type Culture Collection (USA). Cancer cells were cultivated in high-glucose Dulbecco’s modified Eagle’s medium (DMEM; Wuhan Boster Biological Technology, Ltd., Wuhan, China) containing 10% fetal bovine serum (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) at 37°C in a 5% CO₂ incubator, as previously reported [36]. Human miR-223-3p mimics, miR-223-3p inhibitors and the corresponding controls were synthesized by RiboBio. Plasmid vectors expressing SLC4A4 or the NC were constructed by Genechem (Shanghai, China). Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA) was used to transfect cells with miR-223-3p mimics, miR-223-3p inhibitors, the SLC4A4 plasmid and the corresponding controls in six-well plates according to the recommendations of the manufacturer.

Cell proliferation analysis

Caki-1 and 786-O cells were first transfected with miR-223-3p mimics, miR-223-3p inhibitors, SLC4A4 or the NC. Then, the cells were added to a 96-well plate at a density of 3x10³ cells/well. The cell proliferation rate (optical density value) was determined with a CCK-8 assay (Dojindo Molecular Technologies, Inc, Rockville, MD, USA) according to the manufacturer’s protocol. Two hours after the addition of 10 µL of CCK-8 solution to each well, the optical density value was measured at 450 nm. Three independent experiments were conducted for each assay.

Migration and invasion assays

Migration assays were performed as previously described [37]. Cells were homogenized in the absence of plasma for 24 h, and then cultured at 1x10⁵ cells/well in 24-well Transwell plates with polycarbonate membrane inserts (Corning Inc., Corning, NY, USA). For the invasion assays, the membrane was coated with Matrigel (Thermo Fisher Scientific), and cells were cultured at 2x10⁵ cells/well. After incubation for 24 hours, the cells were fixed with 100% methanol and stained with 0.05% crystal violet. Five random fields were counted, and three independent experiments were conducted for each assay.

Luciferase assays

Wild-type and mutant SLC4A4 3′-UTR reporters were purchased from RiboBio. Tumor cells transfected with miR-223-3p mimics or miR-NC in 24-well plates were co-transfected with 500 ng of a luciferase reporter with Lipofectamine 2000 reagent (Thermo Fisher Scientific). All dual luciferase assays were performed with a Dual Luciferase Assay kit (Promega, Madison, WI, USA). Normalized Renilla-luciferase values were calculated relative to the control, according to the manufacturer’s protocol.

Western blotting

Tissues and cells were pyrolysed in a protein lysis system containing radioimmunoprecipitation assay buffer (Wuhan Boster Biological Technology, Ltd.), a protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN, USA) and PMSF (Wuhan Boster Biological Technology, Ltd.). Protein concentrations were measured with a bicinchoninic acid kit (Beyotime Institute of Biotechnology) at 562 nm. Thirty micrograms of each protein were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (EMD Millipore, Bedford, MA, USA) for 90 min. After the transfer, the membranes were blocked in phosphate-buffered saline with 5% non-fat milk for one hour and then incubated with antibodies against SLC4A4 (1:1000; A5332; ABclonal Biotech Co., Ltd.), KRAS (1:1000; A11059; ABclonal Biotech Co., Ltd.) and GAPDH (1:2000; BM3876; Wuhan Boster Biological Technology, Ltd.) at 4°C overnight. The next day, the membranes were washed and then incubated with secondary antibodies (1:5000; BA1020; Wuhan Boster Biological Technology, Ltd.) at room temperature (25°C) for two hours. Finally, the membranes were washed, and protein levels were detected with a ChemiDoc-XRS+ (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

Immunohistochemistry

Adjacent normal tissues and ccRCC tissues were fixed in formalin, dehydrated and embedded. Tissue sections were then incubated with a primary rabbit SLC4A4 polyclonal antibody (1:100) at 4°C overnight. The sections were washed three times with phosphate-buffered saline and then incubated with a goat anti-rabbit secondary antibody (1:200; GB23303; Servicebio, Inc., Woburn, MA, USA) for two hours at room temperature.

Bioinformatic analysis

RNA-seq data from ccRCC patients (TCGA-KIRC) and the patients’ clinical information were obtained from the Xena Functional Genomics Explorer (https://xenabrowser.net/heatmap/) of the University of California Santa Cruz [38] and the Oncomine database (https://www.oncomine.org). Detailed clinical information on staging can be found at (https://www.cancer.org/cancer/kidney-cancer/detection-diagnosis-staging/staging.html). Protein immunohistochemistry data were obtained from the Human Protein Atlas (https://www.proteinatlas.org/). Candidate miR-223-3p target genes were identified through two publicly available algorithms, miRDB (http://mirdb.org/miRDB) and TargetScan (http://www.targetscan.org). Gene set enrichment analysis was used to determine the SLC4A4 pathway involved in the pathogenesis of ccRCC in patients from TCGA-KIRC (http://www.broadinstitute.org/gsea) [39]. Following the performance of 1,000 permutations, p values < 0.05 and a false discovery rate < 25% were required for the enriched gene set analysis to be considered significantly enriched.

Statistical analysis

The RNA levels of paired samples were analyzed with a paired-sample t-test, while those of unpaired samples were analyzed by one-way ANOVA for four groups (#) or by the t-test for two groups (*). The AUC and ROC curve were used to distinguish the clinical classifications. A Kaplan-Meier curve was used to evaluate the survival rate according to miR-223-3p and SLC4A4 expression with a log-rank test. Long and short OS and DFS were discriminated based on good (≥ 5 years alive or disease-free) and poor (death or recurrence or progression in ≤ 2 years) prognoses [40]. The prognostic significance of miR-223-3p and SLC4A4 in ccRCC were analyzed by univariate and multivariate Cox proportional hazard regression analyses. P values < 0.05 were considered to be statistically significant. All statistical analyses were conducted with SPSS Statistics 22.0 (IBM SPSS, Chicago, IL, USA).