A total of 35 paired fresh RCC specimens and adjacent normal tissues (ANTs) were collected from the Peking University Shenzhen Hospital (Shenzhen, Guangdong, China). All patients were informed in regards to the purposes of this research and provided written consent. This study was approved by the Institutional Review Board and Ethics Committee of Peking University Shenzhen Hospital, China. All the RCC samples were diagnosed as RCC pathologically without chemotherapy or radiotherapy before surgery, and the ANTs were collected from the normal region ≥5 cm outside the tumor range. All specimens were immediately frozen in liquid nitrogen following surgical resection for further study. The clinicopathological information of the patients is documented in Table I.

Human RCC cell lines, including 786-O, ACHN and Caki-2, and cervical cancer cell line HeLa were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The human embryo kidney cell line 293T (293T) was purchased from the Type Culture Collection of the Chinese Academy of Medical Sciences, Beijing, China. All cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco, Carlsbad, CA, USA), with 10% fetal bovine serum (Gibco), 1% antibiotics (100 µg/ml penicillin and 100 mg/ml streptomycin sulfates) and 1% glutamate (Gibco), and then incubated at 37°C in a humidified chamber containing 5% CO₂.

For the downregulation of CREB1, two siRNAs targeting CREB1 (si-CREB1a and si-CREB1b) were designed and purchased from GenePharma (Shanghai, China). A negative control (NC) (siRNA-NC, si-NC), positive control (siRNA-GAPDH) and Fam-labeled siRNA-NC (siRNA-NC-FAM) used in this study were also designed and purchased from GenePharma. For the overexpression of miR-10b-5p and miR-363-3p, miR-10b-5p and miR-363-3p mimic oligoribonucleotides and NC were all chemically synthesized and purchased from GenePharma. When the cells reach 60–80% confluency, they were transfected with si-CREB1a, si-CREB1b, si-NC, siRNA-GAPDH, siRNA-NC-FAM, miR-10b-5p and miR-363-3p mimics or the NC using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA), which were mixed in Opti-MEM^® I reduced serum medium (Gibco). Then, fluorescence microscopy and the quantitative real-time PCR (qPCR) were used to verify and quantify transfection efficiency. The sequences of all the siRNAs and RNAs are summarized in Table II.

The total RNA of cells and tissue samples was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. The RNA samples with 260/280 ratios of 1.8–2.0 were used for further experiments. Total RNA was converted into cDNA by using the miScript II RT kit (Qiagen, Valencia, CA, USA) or PrimeScript™ RT reagent kit (Takara Bio, Inc., Otsu, Japan). qPCR analysis was used to detect the expression level of CREB1 using SYBR^® Premix Ex Taq II (Takara) and the expression level of miR-10b-5p and miR-363-3p using miScript SYBR^®-Green PCR kit (Qiagen), respectively, according to the manufacturer's instructions on the LightCycler 480 real-time PCR system (Roche Applied Science, Mannheim, Germany). GAPDH and U6 were used as the internal control. Their expression levels were shown as fold differences relative to GAPDH and U6, which was based on the equation: Relative expression = 2^−ΔΔCt, ΔΔCt = (meanCt_cancer − meanCt_control) − (meanCt_normal − meanCt_control). The primers used for this study are shown in Table II. All reactions were run in triplicate.

The 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide assay (MTT, 5 mg/ml; Sigma-Aldrich, St. Louis, MO, USA) was used to analyze the cell proliferation according to the manufacturer's protocol. Human RCC cells (786-O and ACHN) were seeded into 96-well culture plates at a cell density of 5,000 cells/well and transfected with either 13.3 pmol siRNAs or miRNAs for each well. Cell growth was assayed by addition of 20 µl of MTT to each well, and the plate was incubated at 37°C for 4–6 h. Later, the reaction was stopped by addition of 120 µl DMSO (Sigma-Aldrich). After shaking for 30 min at room temperature, the optical density (OD) at the wavelength of 490 nm (with 630 nm as the reference wave length) of each sample was measured with an enzyme immunoassay instrument (Bio-Rad, Hercules, CA, USA). The OD values were measured for 3 days at every 24 h interval. All assays were carried out in triplicate.

For the Transwell migration assays, 1×10⁴ 786-O or ACHN cells were harvested 24 h post-transfection. They were then plated into the upper chambers (24-well insert, pore size 8 µm; Corning, Inc., Corning, NY, USA) with 100 µl serum-free DMEM. The lower chambers were filled with 500 µl DMEM containing 10% fetal bovine serum. The cells were then cultured at 37°C in a humidified chamber containing 5% CO₂. Two days later, the cells on the surface of the upper chamber were wiped off with cotton gently. Cells under the surface of the lower chamber were washed with PBS, fixed with paraformaldehyde for 25 min, stained with 0.1% crystal violet for 25 min, and then washed with PBS tenderly three times.

Wound scratch assay was also used to examine the migration of RCC cells. Approximately 3×10⁵ 786-O or ACHN cells were seeded in 12-well plates one day before the transfection. The cells were transfected when they grew to reach 80–90% confluency, and a sterile 200-µl pipette tip was used to scrape a clear line through the cell monolayer 6 h post-transfection. Then, the medium was changed with serum-free DMEM. Images of the scratches were acquired with a digital camera system at 0 and 24 h. Experiments were run in three independent repeats in triplicate and analyzed in a double-blinded manner by at least two observers.

For cell apoptosis analysis, 2×10⁵ 786-O or ACHN cells were seeded in 6-well plates and transfected with 200 pmol siRNAs/miRNAs for 6 h. Forty-eight hours post-transfection, the cells, including floating cells, were harvested, washed twice with 4°C PBS and resuspended in 100 µl 1X binding buffer and stained with 5 µl of propidium iodide (PI) and 5 µl of Annexin V-FITC (Invitrogen) for 15 min at room temperature. Flow cytometry (EPICS, Xl-4; Beckman, CA, USA) was used to quantify the percentage of apoptotic cells within 30 min after staining and 400 µl 1X binding buffer was added to each sample before measurement. Each experiment was conducted at least three times.

Prediction of miRNAs that target CREB1 was performed by computational algorithms due to their base-pairing rules between miRNA and mRNA target sites, location of binding sequences within the target's 3′-UTR, and conservation of target binding sequences within related genomes. In our study, miRNAs that target CREB1 were predicted by miRanda (http://mirdb.org/miRDB/index.html), TargetScan Release 6.2 (http://www.targetscan.org), microRNA (http://www.microrna.org), and miRWalk (http://www.umm.uni-heidelberg.de/apps/zmf/mirwalk). Putative miRNAs targeting CREB1 predicted by all four algorithms were accepted and candidates were chosen based on experimental confirmation.

Luciferase reporter plasmids within the CREB1 3′-UTR fragment (500 bp) that containing target sequences of miR-10b-5p and miR-363-3p were constructed to confirm that miR-10b-5p and miR-363-3p bind directly to the 3′-UTR of CREB1 and inhibit its translation. The miRNA target sequences were inserted between the XhoI-NotI restriction sites in the 3′-UTR of the hRluc gene in the psiCHECK™-2 luciferase vector (Promega, Madison, WI, USA), generating the wild-type (WT) of psiCHECK™-2-3′UTR. The primer sequences for the 3′-UTR of CREB1 mRNA containing the miR-10b-5p binding site (forward primer, 5′-CCGCTCGAGTGAAGAGTTGTGAGATAAATAGTTC-3′ and reverse primer, 5′-AAGGAAAAAAG CGGCCGCCCTGTTTATAATATCAACTTGCATC-3′) and containing the miR-363-3p binding site (forward primer, 5′-CCGCTCGAGCTCAAGAAATTTTCAACGCCAG-3′ and reverse primer, 5′-AAGGAAAAAAGCGGCCGCGTTGAA CACATACAGAACTGAATTA-3′) were designed. Then, we mutated the potential binding sites by exchanging the G and T, A and C, and inserted the mutated sequences in psiCHECK ™-2-3′UTR (MUT) (Fig. 6A and C). These short fragments were all cloned into psiCHECK-2 luciferase vector, respectively, and all the constructs were verified by sequencing.

For the luciferase reporter assay, 0.2 µg luciferase reporter vectors, together with 20 pmol miR-10b-5p mimics, miR-363-3p mimics and negative control, were transfected into 786-O, ACHN, HeLa and 293T cells using Lipofectamine 2000. Luciferase activity was detected using the dual luciferase assay system (Promega) on the Modulus™ Single Tube Multimode Reader (Bio-Systems International, Beloit, WI, USA), according to the manufacturer's instructions 24 h after transfection. Normalized data were calculated as the quotient of Renilla/Firefly luciferase activities. The experiments were performed in duplicate and repeated at least three times.

Immunohistochemical assay of CREB1 was performed in 20 RCC paraffin-embedded tissues according to standard procedures. The 5-µm sections were dewaxed in xylene, rehydrated in a descending series of ethanol, incubated in 3% hydrogen peroxide solution for 20 min, boiled in 0.01 M citrate buffer (pH 6.0) for antigen retrieval, and treated in 10% bovine serum albumin for 30 min at 37°C to block non-specific protein binding. Then the sections were incubated in CREB1 mouse monoclonal antibody (1:100, ab178322; Cambridge, MA, USA) overnight at 4°C. Subsequently, the sections were rinsed with PBS and treated with the anti-rabbit IHC kit (Maixin Bio, Fuzhou, China) at 37°C for 30 min, followed by staining with a DAB kit (Maixin Bio), for 4 min and counterstained with hematoxylin. Negative controls were performed with omission of the primary antibodies.

The frozen fresh samples and collected cells of human RCC cell lines (786-O and ACHN) seeded in a 6-well plate were homogenized in three volumes of RIPA buffer (Sigma-Aldrich) on ice. The protein concentration was quantified using Pierce BCA protein assay kit (Thermo Scientific, Waltham, MA, USA), and protein samples (60 µg) were separated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride membranes (Millipore, Billerica, MA, USA). After blocking with 5% fat-free milk at room temperature for 2 h, the membranes were incubated in 5% fat-free milk containing CREB1 (1:100, ab178322; Abcam) or β-actin (1:5,000, NB600-532; Novus, Littleton, CO, USA) antibody overnight at 4°C. The membranes were washed three times with TBST (Tris-buffered saline with 0.05% Tween) and incubated for 1 h with horseradish peroxidase (HRP)-linked secondary antibody (1:10,000; EarthOx, LLC, USA) at room temperature. The protein bands were detected with the chemiluminescence phototube-HRP kit (WBKLS0500; Millipore) and exposure to Kodak film.

All data are expressed as the mean ± SD from at least three separate experiments. All data were analyzed by SPSS 19.0 statistical software (SPSS Inc., Chicago, IL, USA). The clinicopathological information of the patients was analyzed by Chi-square test, while other data were determined by the Student's t-test. Results were considered statistically significant with two-tailed P<0.05.

To determine the mRNA expression level of CREB1, qPCR was performed in 35 paired RCC tissues and ANTs. As shown in Fig. 1A, the expression of CREB1 was significantly upregulated in the RCC tissues compared with ANTs. We also analyzed the expression of CREB1 in RCC cell lines (786-O, ACHN and Caki-2) and 293T. As shown in Fig. 1B, the expression level of CREB1 was significantly higher in the ACHN, 786-O and Caki-2 cells (P<0.01) compared with the level in the 293T cells.

CREB1 was discovered to be upregulated in many human cancers including leukemia (8), breast cancer (10), gliomas (11) and lung cancer (9). However, the protein expression of CREB1 in renal cancer remains unknown. Therefore, we performed IHC and western blot analysis to determine the protein expression of CREB1 in 20 pairs of RCC sections and 6 pairs of fresh tissues, respectively. The results of the IHC assay showed that CREB1 was located in the nuclei of the cells and overexpression of CREB1 was observed in 80% (16/20) of the RCC sections (Fig. 1C). The western blot assay also showed that CREB1 was upregulated in all paired RCC tissues and human RCC cell lines (786-O and ACHN) compared with the ANTs and the 293T cell line, with β-actin as the internal control (Fig. 1D and E). No relationship was found between the protein expression level of CREB1 and clinicopathologic variables due to the limited number of cases.

To explore the function of CREB1 in renal cancer, two siRNAs to downregulate CREB1 were designed and confirmed. Through transient transfection efficiency validation of the siRNAs (Fig. 2), we verified and chose si-CREB1a as the most effective siRNA targeting CREB1. Cell proliferation, migration and apoptosis of 786-O and ACHN cells were evaluated after transfection with si-CREB1a (si-CREB1) by MTT assay, Transwell migration assay, wound scratch assay and flow cytometry.

The results of the MTT assay revealed that downregulation of CREB1 suppressed cell proliferation. As shown in Fig. 3A, the cell proliferation of 786-O and ACHN cells was reduced by 8.15% (P<0.05), 16.20% (P<0.01), 32.51% (P<0.01) and 9.51% (P<0.05), 22.12% (P<0.01), 31.41% (P<0.05), respectively, after transfection of si-CREB1 at 24, 48 and 72 h, compared with the cells transfected with si-NC, respectively. Transwell migration and wound scratch assays showed the same result. Inhibition of CREB1 attenuated the migratory ability of 786-O and ACHN cells. Transwell migration assay showed that the number of migrated cells was decreased by 89.98% (P<0.01) in the 786-O cells and by 90.57% (P<0.01) in the ACHN cells (Fig. 3B). Wound scratch assay showed that the inhibition rate of migration was 40.07% for the 786-O cells (P<0.01) and 49.61% for the ACHN cells (P<0.001) (Fig. 3C). Compared with the cells transfected with si-NC, the average cell apoptosis rate was increased significantly in the cells transfected with si-CREB1, with 6.36 vs. 10.09% in the 786-O cells (P<0.05) and 6.16 vs. 12.88% in the ACHN cells (P<0.01) (Fig. 3D). All the data suggest that CREB1 is an oncogene and affects cellular proliferation, migration and apoptosis.

Bioinformatic analyses (miRanda, TargetScan Release 6.2, microRNA and miRWalk) were used to determine the potential miRNAs that target CREB1. miR-10b-5p and miR-363-3p, were predicted by all four algorithms simultaneously and were selected as candidates for further study (Fig. 4A and B). Furthermore, miR-10b-5p and miR-363-3p were previously found to be downregulated in RCC tissues compared with ANTs or downregulated in metastatic compared with primary RCCs by all the high-throughput screens (28–33).

To determine whether CREB1 is directly regulated by miR-10b-5p and miR-363-3p, a luciferase reporter assay was performed in 293T, HeLa, 786-O and ACHN cells. As shown in Fig. 4C, the relative luciferase activity of wild-type plasmids containing the putative binding site was significantly decreased when transfected with miR-10b-5p mimics in the 293T (P=0.001), HeLa (P=0.005), 786-O (P=0.009) and ACHN cells (P=0.006) while no notable reduction was observed in the mutant groups. As shown in Fig. 4D, when transfected with the miR-363-3p mimics, identical results were also obtained in the 293T (P=0.004), HeLa (P<0.05), 786-O (P<0.05) and ACHN cells (P<0.05), suggesting that miR-10b-5p and miR-363-3p may suppress the expression of CREB1 by targeting the putative binding site in the 3′-UTR.

To validate the results of the luciferase reporter assay, we performed qPCR and western blot analysis to quantify the mRNA and protein levels of CREB1 in the 786-O and ACHN cells 48 h after transfection with miR-10b-5p and miR-363-3p mimics. miR-10b-5p and miR-363-3p mimics significantly downregulated the mRNA and protein levels of CREB1 (Fig. 5A and B). To further explore the relationship of the expression pattern between CREB1 and miR-10b-5p and miR-363-3p, we quantified the expression levels of miR-10b-5p and miR-363-3p in the 35 paired RCC tissues and ANTs by qPCR. We analyzed the fold-change, Log2Ratio(T/N), of CREB1 and miR-10b-5p and miR-363-3p in each paired RCC tissue and ANT and we found that the expression level of CREB1 decreased when miR-10b-5p and miR-363-3p expression increased, suggesting that CREB1 was properly regulated by miR-10b-5p and miR-363-3p in the tissue samples (Fig. 5C). As shown in Fig. 6A and B, the results also showed that miR-10b-5p and miR-363-3p expression levels were significantly lower in the RCC tissues than the levels in the ANTs, respectively.

To further confirm that lower expression levels of miR-10b-5p and miR-363-3p mediate the upregulation of CREB1 to promote tumorigenicity, we introduced exogenous miR-10b-5p and miR-363-3p into human RCC cell lines (786-O and ACHN) and performed MTT and cell migration assays and apoptosis analysis. Forty-eight hours after transfection, the transient transfection efficiency was verified by qPCR (Fig. 7). Then MTT assay showed that overexpression of miR-10b-5p significantly decreased the cell proliferation of 786-O cells by 3.72% (P<0.05), 13.99% (P<0.01), 16.75% (P<0.01) and ACHN cells by 16.86% (P<0.05), 18.16% (P<0.01), 32.19% (P<0.01), at 24, 48 and 72 h post-transfection, respectively (Fig. 8A). Transwell migration assay found that the number of migratory cells was suppressed by 45.65% (P<0.05) and 56.28% (P<0.05) (Fig. 8B) and the migratory distances were inhibited by 28.73% (P<0.05) and 30.97% (P<0.01) (Fig. 8C) for 786-O and ACHN cells, respectively, after transfection of miR-10b-5p mimics. For the apoptosis assay, upregulated expression of miR-10b-5p promoted the mean early apoptotic rate from 4.77 to 13.20% (P<0.05) for 786-O cells and from 10.31 to 24.04% (P<0.05) for ACHN cells (Fig. 8D).

The proliferation ability of the 786-O and ACHN cells was attenuated identically by 6.70% (P<0.05), 13.22% (P<0.01), 19.79% (P<0.01) and 10.53% (P<0.05), 15.17% (P<0.01), 24.14% (P<0.01) due to introduction of miR-363-3p, respectively (Fig. 9A). After transfection of miR-363-3p mimics, for 786-O and ACHN cells, respectively, the migratory cells were suppressed by 17.95% (P<0.05) and 25.65% (P<0.05) (Fig. 9B) and the migratory distances were inhibited by 16.17% (P<0.05) and 30.02% (P<0.01) (Fig. 9C) due to ectopic expression of miR-363-3p. miR-363-3p mimics induced the mean early apoptotic rate from 4.56 to 9.58% (P<0.05) for 786-O cells and from 11.61 to 18.84% (P<0.01) for ACHN cells (Fig. 9D).

All the above data revealed that miR-10b-5p and miR-363-3p mimics exert a function similar to si-CREB1 by inhibiting cell proliferation and migration and inducing cell apoptosis, suggesting that low miR-10b-5p and miR-363-3p expression-mediated upregulation of CREB1 in RCC tissues and cell lines restores tumorigenicity.