microRNA-193a-3p is specifically down-regulated and acts as a tumor suppressor in BRAF-mutated colorectal cancer

A total of 314 patients with colorectal cancer, comprising 255 patients who underwent drug therapy including cytotoxic agents and anti-EGFR antibody, and/or surgery in the Tohoku University Hospital (TUH) between 2004 and 2013, and 59 patients who received anti-EGFR antibody in the National Cancer Center Hospital (NCCH) between 2003 and 2012, were recruited in this study. The clinical information regarding clinical characteristics of patients and tumors, OS, progression-free survival (PFS) after initiations of drug therapies, and response rate (RR), was retrospectively analyzed through reviews of clinical records. As listed in Additional file 1: Table S1, clinical characteristics of patients within the TUH cohort are associated with earlier clinical stage and proximal location compared to those within the NCCH cohort.

DNA was extracted from a 5 μm- or 10 μm-thick formalin-fixed paraffin-embedded (FFPE) tissue of each patient with colorectal cancer through the use of QIAmp DNA FFPE tissue kit (Qiagen, Valencia, CA, USA). Total RNA including miRNA fraction was extracted from the FFPE tissue of each colorectal cancer by using the Ambion RecoverAll Total Nucleic Acid Isolation Kit (Life Technologies Corporation, Carlsbad, CA, USA). Total RNA was also extracted from normal adjacent colonic mucosa of 11 patients with colorectal cancer from the cohort.

The mutational status of codon 12 and 13 of KRAS gene and the codon 600 of BRAF gene were analyzed by direct DNA sequencing through the use of CEQ2000EX automated DNA sequencer (Beckman Coulter, Brea, CA, USA). The accession number of cDNAs of KRAS, wild-type and p.V600E BRAF, were NM_033360.3, NM_004333.5 and HQ224878.1, respectively. Primers used for the amplification of fragments were 5′-accttatgtgtgacatgttc (forward) and 5′-atggtcctgcaccagtaata (reverse) for KRAS codons 12 and 13, and 5′-ataatgcttgctctgatagg (forward) and 5′-gtaactcagcagcatctcag (reverse) for BRAF codon 600.

The genome-wide miRNA expression levels of the 30 colorectal cancers from the screening set were analyzed by the SurePrint G3 Human miRNA Rel. 16.0 microarray (Agilent Technologies, Santa Clara, CA, USA), which covers 1222 human miRNAs, according to the manufacturer’s protocol. The microarray data were extracted using the GeneSpring ver. 12.5 (Agilent Technologies). The raw data was normalized by using the 90-percentile shift method, and the acquired data of each miRNA were compared between wild-type KRAS/BRAF tumors and mutant-BRAF tumors using Mann-Whitney U test. The microarray data has been deposited in the Gene Expression Omnibus database (accession No. GSE66548).

The miRNA expressions of colorectal tissues and colon cancer cell lines were quantified by Taqman real-time RT-PCR (qRT-PCR) using a CFX96 real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA). The relative expression of each miRNA was calculated by the delta CT value method, through the use of miR-16 expression for human colon samples [27, 33] and RNU48 for colorectal cancer cell lines [34] as a normalizer. At least two independent samples were loaded as an internal control in each PCR plate for miR-193a-3p analysis for colorectal tumors, to keep consistency of measurements throughout all plates. Each sample was amplified in triplicate and the results obtained from each run were normalized according to the data of internal controls.

Human colorectal cancer cell lines RKO (CRL-2577) and HCT116 (CCL-247) were purchased from the American Type Culture Collection in 2011. Human colorectal cancer cell lines DiFi, HCT8, LIM2405, and SW48 were kindly provided along with appropriate ethics rules and consents of both institutions by Dr. Mariadason in Ludwig Institute for Cancer Research, Australia. The cell lines were regularly authentificated by short tandem repeat analysis. RKO was cultured in Dulbecco’s Modified Eagle’s Medium (Sigma-Aldrich, St.Louis, MO, USA) with 10% fetal bovine serum and the other four cell lines were grown in Roswell Park Memorial Institute Medium 1640 (Sigma-Aldrich) with 10% fetal bovine serum at 37 °C.

The cells were transfected with precursor of miR-193a-3p or precursor of negative control (PM11123 or AM17110, Applied Biosystems), or anti-miR-193a-3p or anti-negative control (AM17000 or AM17010, Applied Biosystems) at a final concentration of 33–67 nM using Lipofectamine 2000 (Life Technologies Corporation), according to the manufacturer’s protocol.

The cells were seeded onto 96-well plates with the different number of cells (RKO, 7 × 10³; HCT116, 5 × 10³; SW48, 1.5 × 10⁴). When attached, cells were transfected with precursors of miR-193a-3p or negative control as mentioned above. The cell viability was measured after 48 h in RKO or after 72 h in HCT116 and SW48 using the Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan), according to the manufacturer’s protocol. Each experiment was performed in quadruplicate and data were obtained from three or more independent experiments.

Invasion activities of RKO and HCT116 cells were analyzed using Boyden chambers with 8-mm pore membranes coated with matrigel (BD Biosciences, San Jose, CA, USA) following the standard protocol. In six-well plates, 3 × 10⁵ of RKO cells and 1 × 10⁵ of HCT116 cells were transfected with precursors of miR-193a-3p or negative control. After 24 h, the transfected cells (3 × 10⁵ of the RKO cells and 1.5 × 10⁵ of the HCT116 cells) were re-suspended in 500 μl of serum-free medium in the upper wells. Medium containing 10% fetal bovine serum was then added into the bottom wells. After incubation for 48 h for RKO or 24 h for HCT116, the invaded cells were stained with 0.2% crystal violet solutions. The number of cells was counted from four representative fields of each membrane, and the results were obtained from three independent experiments.

Human wild-type BRAF cDNA was prepared by PCR using the primers (forward; 5′-gtggaattctgcagatataagatggcggcgctgagcggtgg, reverse; 5′-gccactgtgctggatcctttgttgctactctcctgaactctctcactc), which cover full lengths of the coding region of the gene, from a human cDNA library. The amplified fragments were cloned into the pcDNA 3.1(+) vector (Life Technologies) using the In-Fusion HD Cloning Kit (Takara Bio, Shiga, Japan). Human mutant BRAF cDNA containing p.V600E mutation was constructed by site-directed mutagenesis using the wild-type BRAF expression vector as a template. The insert fragments containing wild-type or mutant BRAF was sequenced by ABI Prism 3130 (Life Technologies), to confirm that the fragments has no sequence variation in the coding region of BRAF other than p.V600E.

Western blot analysis was performed following a standard protocol [35]. Anti-BRAF rabbit monoclonal antibody (#9433, Cell Signaling Technology, Danvers, MA, USA), anti-p-BRAF (Ser445) rabbit monoclonal antibody (#2696, Cell Signaling Technology), anti-p-MEK1/2 (Ser217/221) rabbit polyclonal antibody (#9121, Cell Signaling Technology), anti-p-ERK1/2 (Ser217/221) rabbit monoclonal antibody (#4094, Cell Signaling Technology), and anti-α-tubulin mouse monoclonal antibody (Sigma-Aldrich) were used as primary antibodies for detection of the specific proteins.

Statistical analyses were performed with JMP Pro ver. 11.0 (SAS Institute, Cary, NC, USA). The differences between two groups were analyzed by chi-square test, Fisher’s exact test, Student’s t test or Mann-Whitney U-test, depending on each parameter. Correlation analysis was performed by using Spearmann’s rank correlation method. Kaplan-Meier analysis was conducted to estimate distributions of OS, or PFS after the beginning of the first-line chemotherapy or after anti-EGFR therapies, and a log-rank test was utilized to analyze the statistical difference in the survival. Each difference was regarded as statistically significant when P < 0.05.

This study was performed in accordance with the Declaration of Helsinki and was approved by the Ethical Committee of TUH and NCCH. A written informed consent was obtained from all patients.

We first analyzed the mutational status of KRAS and BRAF in colorectal cancers from our cohort of patients. The sequencing analyses identified KRAS mutations in 95 tumors and BRAF mutations in 21 tumors within both the TUH and NCCH cohort (Additional file 1: Table S1). Two tumors with concurrent mutations within both KRAS and BRAF genes were excluded from any further analysis. To screen the miRNAs that are dysregulated in BRAF-mutant tumors, we then divided 64 tumor samples from the entire cohort into two sets as follows: a screening set (15 KRAS/BRAF-wild-type tumors and 15 BRAF-mutant tumors from the TUH and NCCH cohort) and a validation set (30 KRAS/BRAF-wild type tumors and four BRAF-mutant tumors from the TUH cohort) (Table 1). Using the screening set, we found nine up-regulated miRNAs (median, > 1.5-fold) and 13 down-regulated miRNAs (median, <−1.5-fold), through the global miRNA expression analysis (Table 2). We selected the top three up-regulated miRNAs (miR-31, miR-135b, and miR-7) and the bottom two down-regulated miRNAs (miR-193a-3p and miR-148b), which had a median fold change of either >4-fold or <−4-fold respectively, as the candidates of specifically dysregulated miRNAs in BRAF-mutant tumors (Table 2 and Fig. 1a). Before proceeding to the next step using the validation set, we validated the expression of the five miRNAs in the screening set using qRT-PCR. The qRT-PCR results exhibited the same trend as the microarray results, except the difference in the expression level of two miRNAs, miR-135b and miR-7, where comparison between KRAS/BRAF-wild-type tumors and BRAF-mutant tumors became not significant (Fig. 1b). We further confirmed that the results were well correlated between the microarray and the qRT-PCR analysis by comparing the signal intensity obtained by the microarray analysis and the CT values of miR-193a-3p or miR-16 obtained by qRT-PCR in each sample (Additional file 2: Figure S1), indicating the reliability of the screening results obtained by the microarray analysis.

To validate the results obtained from the screening analysis, we next performed the qRT-PCR in another set of samples, the validation set. This validation analysis successfully confirmed that miR-31 and miR-135b were significantly up-regulated, and miR-193a-3p was significantly down-regulated in BRAF-mutant colorectal cancers compared to KRAS/BRAF-wild-type cancers (Fig. 2). In contrast, the results of miR-7 and miR-148b were not validated in this analysis (Fig. 2). In addition, to further elucidate whether the altered expression of the candidate miRNAs occurred in a BRAF-dependent fashion, we added KRAS-mutant colorectal cancers (n = 20) to the qRT-PCR and compared the miRNA expressions between BRAF-mutant tumors and KRAS-mutant tumors. We found that miR-31 and miR-135b were significantly up-regulated, while, miR-193a-3p was marginally significantly down-regulated (P = 0.09) in BRAF-mutant cancers compared to KRAS-mutant cancers (Fig. 2). Furthermore, these three miRNAs were shown to be significantly dysregulated in BRAF-mutant cancers compared to normal colonic mucosa (n = 11; Fig. 2), which suggests that the alteration in these miRNA expressions may contribute, at least in part, to the carcinogenesis of BRAF-mutant tumors. Of these miRNAs, up-regulation of miR-31 and miR-135b has previously been linked to tumorigenesis of colorectal cancer [21, 25]. In contrast, little is known about the functional role of miR-193a-3p for colorectal carcinogenesis. We therefore decided to focus on miR-193a-3p for further molecular and clinical analyses, to clarify the previously unknown role of miR-193a-3p, which is involved in the tumorigenesis of human colorectal cancer.

In light of recent evidence that miR-193a-3p may have a tumor suppressive function in cancers of other organs, such as breast and lung [36, 37], and our finding that this miRNA was down-regulated in BRAF-mutant colorectal cancers (Figs. 1 and 2), we hypothesized that miR-193a-3p serves as a tumor-suppressive miRNA in colorectal cancer. To address this question, we first analyzed the effect of miR-193a-3p overexpression in colorectal cancer cell lines using the cell viability and invasion assay. The cell viability assay revealed that miR-193a-3p overexpression significantly inhibited cell viability compared to overexpression of the precursors of a negative control in all three colorectal cancer cell lines analyzed (23–38%; Fig. 3a). This inhibitory effect of miR-193a-3p overexpression on cell survival did not depend on the cellular genotype of KRAS/BRAF (RKO; wild/mutant, HCT116; mutant/wild, SW48; wild/wild, respectively) [38, 39]. In addition, the invasion assay demonstrated that miR-193a-3p exerted an inhibitory effect on cellular invasion with a 20% and 50% reduction in RKO and HCT116 cells, respectively (Figs. 3b and c). Second, we analyzed the effect of miR-193a-3p inhibition in these cell lines on cell viability and invasion. Cell viability and invasion ability were not increased in these cells transfected with miR-193a-3p inhibitors (data not shown). One possible explanation might be that endogenous miR-193a-3p expression, regardless of further forced down-regulation of the miRNA, was already down-regulated in these cells, enough for affecting their viability and invasion ability. Third, in light of the observation that miR-193a-3p inhibited the invasion ability of the cancer cells, we analyzed the effect of miR-193a-3p overexpression on expressions of epithelial-mesenchymal-transition-related (EMT-related) genes such as ZEB1, ZEB2, SNAI1, and SNAI2. We found that the expression of ZEB1, SNAI1, and SNAI2 was decreased in RKO cells transfected with the pre-miR-193a-3p compared to those transfected with a negative control (Fig. 3d), while the expression of ZEB2 was not detected in RKO cells transfected with the negative control or those transfected with pre-miR-193a-3p. This result suggests that the inhibition of the invasion ability of cancer cells by miR-193a-3p may be induced, at least in part, through the down-regulation of EMT-related genes. These results support our surmise that miR-193a-3p functions as a tumor suppressor underlying tumor initiation and development of colorectal cancer, particularly BRAF-mutant tumors.

Our screening and validation analyses identified miR-193a-3p as one possible BRAF-mutant specific miRNA; however, the underlying mechanisms controlling the specific down-regulation of this miRNA in BRAF-mutant cancers are still unclear. Earlier studies suggested that the promoter hypermethylation of miR-193a-3p may be an explanation of its reduced expression [40‐42]; however, we hypothesized that BRAF or its down-stream proteins may directly affect the down-regulation of miR-193a-3p expression. To address this issue, we next examined the expression of miR-193a-3p in colorectal cancer cell lines with mutant BRAF protein overexpression. In all three cell lines transfected with a mutant-BRAF expression vector, we confirmed that phosphorylated levels of BRAF protein, as well as those of its downstream proteins MEK and ERK, were increased, compared to the corresponding cell lines transfected with an EGFP control vector (Fig. 3e). In this system, the expression levels of miR-193a-3p were modestly but significantly decreased in SW48 and DiFi, both KRAS/BRAF-wild-type colorectal cancer cells (Fig. 3f) [39, 43]. This modest decrease in miR-193a-3p expression levels could be due to imperfect transfection efficiency (approximately 50% EGFP positive cells in all cell lines; data not shown). In contrast, the expression of miR-193a-3p was not altered by BRAF overexpression in KRAS-mutant HCT8 cells and BRAF-mutant LIM2405 cells (Fig. 3d) [44], in which the downstream constituents of the RAS-RAF-MEK-ERK pathway were already constitutively active, due to the activated mutation within the upstream KRAS. We next tried to elucidate whether miR-193a-3p expression is affected by inhibition of the downstream constituents of the RAS-RAF-MEK-ERK pathway in a BRAF-mutant cell line RKO and a KRAS-mutant cell line HCT116. miR-193a-3p expression was not significantly affected by treatments with both a BRAF inhibitor dabrafenib and a MEK inhibitor trametinib (using two doses with 1.5 and 0.25 μM, and 3.0 and 0.5 μM, respectively) either in a BRAF-mutant cell line RKO or in a KRAS-mutant cell line at 3, 6, and 9 h treatment (Additional file 3: Figure S2). In addition, to elucidate whether the miR-193a-3p down-regulation induced by BRAF overexpression affect its target genes, we co-transfected a psiCHECK-2 vector that has a luciferase sequence with target sequences of miR-193a-3p (miCheck miRNA biosensor clone, Promega), and either the BRAF V600E overexpression vector or control pEGFP vector into KRAS/BRAF-wild SW48 cells. The luciferase activity was not significantly increased in cells co-transfected with pBRAF and psiCHECK-2 vector that has miR-193a-3p target sequences (Additional file 4: Figure S3). These results suggests that mutant BRAF may affect miR-193a-3p expression to some extent, but is not a single factor that contributes to dysregulation of miR-193a-3p and its target genes. However, based upon the statistically significant down-regulation of miR-193a-3p by transient overexpression of BRAF V600E as shown in Fig. 4d, the possibility exists that the mechanism underlying the miR-193a-3p down-regulation in BRAF-mutant colorectal cancers may be partially influenced by a direct or indirect effect of activated BRAF.

Our finding that miR-193a-3p was a candidate miRNA dysregulated in a BRAF-dependent manner and also functioned as a tumor suppressor in colorectal cancer, prompted us to further elucidate whether its expression status was associated with the clinical outcome of our cohort of patients treated with cytotoxic chemotherapy and/or molecular-targeted drugs. We conducted a survival analysis using 99 patients from the TUH cohort, whose information on clinical outcomes such as OS, PFS for cytotoxic chemotherapy or anti-EGFR antibodies, RR, and RNA samples for the miRNA expression analysis were available. We first analyzed the OS and PFS for first-line drug therapy, and the PFS for anti-EGFR antibody therapy based upon the molecular subtypes characterized by KRAS/BRAF genotype. Consistent with earlier studies [9, 10], patients with a BRAF-mutant colorectal cancer had a poorer OS compared to those with a KRAS/BRAF-wild-type and KRAS-mutant cancer (Additional file 5: Figure S4a). In addition, as expected, BRAF-mutant patients showed a worse OS and PFS for anti-EGFR therapy compared to KRAS/BRAF-wild-type patients (Fig. 4a and b), whereas PFS for first-line chemotherapy of the three subtypes of patients did not differ significantly (Additional file 5: Figure S4b). Next, we categorized all colorectal tumors into a high and low miR-193a-3p expression group using the median value in all samples as a cut-off point. We found that miR-193a-3p expression status did not correlate with OS or PFS for a first-line drug therapy or OS after the initiation of anti-EGFR therapy (Additional file 5: Figure S4c, S4d and Fig. 4c). However, low expression of this miRNA was associated with a reduced PFS (HR 2.10, P = 0.02; Fig. 4d) and tended to be associated with a lower RR (24% vs. 42%; Additional file 1: Table S2) for anti-EGFR therapy. Since our results showed that miR-193a-3p was identified as a mutant-BRAF-specific miRNA, the worse outcome of colorectal cancer patients with the down-regulation of miR-193a-3p from anti-EGFR therapy may be confounded by the BRAF-mutation status. To avoid this possibility, we analyzed PFS for anti-EGFR therapy in our cohort of patients without KRAS/BRAF mutation (n = 34). In particular, we found that low miR-193a-3p expression did not associate with a poorer OS (Fig. 4e), but still tended to be associated with a shortened PFS for anti-EGFR therapy among non-KRAS/BRAF-mutant patients, although it should be noted the difference did not reach the significance level (HR 1.97, P = 0.08; Fig. 4f), which could be due to the small number of patients analyzed. The low miR-193a-3p expression tended to be associated with a lower RR (31% vs. 47%; Additional file 1: Table S3) as well.

Taken together, our results suggest that miR-193a-3p may not only serve as a tumor-suppressive miRNA involved in the oncogenesis of colorectal cancer, particularly BRAF-mutant tumors, but also may be a determinant that can affect the sensitivity to anti-EGFR therapy even in BRAF-wild-type colorectal cancer.