All five colorectal cancer cell lines, SW620, SW480, HCT116, HT-29 and Caco2, were obtained from the American Type Culture Collection (Manassas, VA, USA) and maintained in complete RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), streptomycin (100 µg/ml), and 2 g/l sodium bicarbonate (Gibco, Invitrogen Corp., Carlsbad, CA, USA) at 37°C in humidified air with 5% CO₂.

miR-29a-3p inhibitor, miRNA inhibitor negative control (NC) oligos, miR-29a-3p mimic, miRNA mimic negative control (NC) oligos, were purchased from Exiqon (Vedbaek, Denmark). HCT116 and SW620 at a density of 5.3×10⁴ and 1.3×10⁵ cells/ml, respectively, were transfected with Lipofectamine 2000 (Invitrogen), and SW480 cells at a density of 1.1×10⁴ cells/ml were transfected with HiPerFect (Qiagen, Hilden, Germany) in 24-well plates according to the manufacturer's instructions. This was followed by MTT assay.

Archived paraffin-embedded CRC tissues and matched normal adjacent tissues were from 28 histologically confirmed CRC patients who underwent surgery in Kuala Lumpur Hospital between 2010 and 2011. Medical ethics was approved by the National Medical Research and Ethics Committee (NMRR-12-435-11565), and each patient provided consent to use their paraffin-embedded tissues. The tumors were staged according to the TNM staging system. All cases were reviewed and confirmed by two experienced pathologists. The clinical and pathological characteristics of patients are shown in Table I. Due to time and financial constraints, experiments were conducted with a small cohort containing 28 samples. Sections of 4-µm thickness were cut and mounted on glass slides followed by deparaffinization and macrodissection under a microscope. The tumor area and normal adjacent colonic mucosae were scrapped off with a pipette tip and collected in 1.5 ml microcentrifuge tubes.

For assessment of miR-29a-3p expression in CRC tissues and colorectal cancer cell lines, total RNA including miRNA was purified using miRNeasy FFPE kit (Qiagen) and miRNeasy Micro kit (Qiagen) according to the manufacturer's instructions, respectively. U6 snRNA was employed as the endogenous reference gene for relative quantification of miRNA using RT-qPCR. Reverse transcription was conducted using the Universal cDNA synthesis kit (Exiqon) according to the manufacturer's instructions.

miR-29a-3p (cat. no. 204698) and U6 (cat. no. 203907) primers were purchased from Exiqon. As Locked Nucleic Acids Technology is covered by the patents owned by Exiqon, the sequences of the primers were not provided by the manufacturer. PCR was performed with a Mastercycler EP realplex 4 (Eppendorf, Germany) using SYBR^® Green Master Mix Universal RT kit (Exiqon). PCR parameters were set as follows: polymerase activation at 95°C for 10 min, 40 amplification cycles of denaturation at 95°C for 10 sec, annealing and extension at 60°C for 1 min. Melting curve analysis was used to assess amplification specificity. Each sample was run in triplicate and each experiment was repeated for three times. Standard curve method was used to calculate the relative expression of miRNA (17).

To determine the effect of miR-29a-3p inhibitor on cell growth, MTT assay was used to assess cell viability following cell transfection with miR-29a-3p inhibitor. Cells were plated in 96-well plates in triplicates. After incubating the plate at 37°C for 24 h, miRNA inhibitors were introduced into the cells via transfection. MTT reagent (9 µl) at a concentration of 5 mg/ml was pipetted into each well after 96 h followed by incubation in cell culture incubator for 4 h. The culture medium was then removed and 100 µl DMSO was added into each well to dissolve intracellular formazan. The absorbance of each well was measured at 570 nm utilizing a spectrophotometric microtiter plate reader (Dynex Technologies, Inc., Chantilly, VA, USA). Negative control was regarded as calibrator and the absorbance was defined as 100%. Relative viability of experimental cells was calculated by dividing the absorbance of experimental cells by that of negative control. Three independent experiments were performed for each treatment.

Transwell assay system was used to assess cell migration and cell invasion in the present study. Transwell migration assay is always performed with commercially available plastic inserts compatible with multiwell plates. The bottom of plastic insert is made up of microporous membrane allowing migratory cells to reach the other side of the membrane. To determine cell invasion, a gel containing similar components with basement membrane is formed on the microporous membrane. The only difference between migration assay and invasion assay is the presence of the basement membrane-like gel. Matrigel™ was added on the microporous membrane. Unbound Matrigel was aspirated and the plate was placed in an incubator at 37°C for 30 min to allow gelling. The membrane was coated twice with Matrigel. SW480-7 cells transfected with 10 nM miRNA inhibitor or miRNA mimic for 48 h were detached with trypsin (0.25%). Cells (1.0×10⁵) resuspended in 200 µl serum-free medium were pipetted into upper chamber, and 700 µl medium with 10% FBS was added into lower chamber. After 72 h of incubation at 37°C, the insert was soaked in ice-cold 95% ethanol for 10 min to fix the cells. The invaded cells and non-invaded cells were stained by propidium iodide (PI) and 4′, 6-diamidino-2-phenylindole (DAPI), respectively. Cell counting of invaded and non-invaded cells was done using the Image-Pro Plus 6.0 software. The invasive capacity was represented by relative invasion that was determined using the formula: number of invaded cells stained by PI/number of non-invaded cells stained by DAPI. Three independent experiments were conducted to verify the reproducibility. The invasive capacity of a treated sample was normalized to that of corresponding control. One-sample t-test was used to test the differences of normalized invasive capacities of three independent experiments with the hypothetical value (set to 1).

SW480-7 cells were transfected with miR-29a-3p inhibitor or miRNA negative control oligos 48 h before the PCR array. Total RNA purification was conducted with GeneJET RNA Purification kit (cat. no. K0731; Thermo Scientific, USA). The integrity of extracted RNA was tested by agarose gel electrophoresis. Purity and quantity of RNA were determined using a Nandrop 1000 spectrophoto-meter, and only RNA samples with A260: A280 absorbance values between 1.8 and 2.1 were used in subsequent experiments. Reverse transcription (RT) was performed with RT² First Strand kit (cat. no. 330401; Qiagen) according to the manufacturer's protocol. Cytoskeleton Regulators RT² Profiler PCR array plate (cat. no. PAHS-088ZA; Qiagen) was used to perform real-time quantitative PCR. Twenty-five microliters of PCR components mix was added to each well of the 96-well RT² Profiler PCR array plate. Then, the plate was placed in the real-time cycler (Eppendorf, Wesseling-Berzdorf, Germany), and the PCR cycling program was set as follows: polymerase activation at 95°C for 10 min, 40 amplification cycles of denaturation at 95°C for 15 sec, annealing and extension at 60°C for 1 min.

Total RNA purification was conducted with GeneJET RNA Purification kit (Thermo Scientific). cDNA was synthesized by ReverAid Reverse transcriptase at 42° for 60 min. Real-time PCR was performed with SYBR Select Master Mix (cat. no. 4472908, Life Technologies, CA, USA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was set as reference gene and the primer sequences are shown in Table II. Real-time PCR was conducted with a 10-µl total volume, containing 1 µl of cDNA template, 5 µl 2X SYBR^® Select Master Mix, 0.25 µl forward/reverse primer, and 3.5µl of RNase-free water. PCR was performed on a Mastercycler EP realplex 4 (Eppendorf, Germany). The thermal cycler was programmed as follows: initial denaturation at 95°C for 2 min, 40 cycles consisting of 95°C for 15 sec, 60°C for 15 sec, 72°C for 1.5 min, and extension at 72°C for 10 min. Standard curve method was used to calculate the relative expression of mRNA.

Total proteins were harvested from SW480-7 cells 72 h post-transfection with 300 µl of 1X sodium dodecyl sulfate (SDS) lysis buffer. Approximately 20 µg of protein samples were loaded on and electrophoresed in polyacrylamide gel electrophoresis (PAGE) in running buffer at 120 V. The proteins in the gel were electrotransferred to a polyvinylidene fluoride (PVDF) membrane (Pierce, Thermo Fisher Scientific) in transfer buffer at 100 V. After 2 h of transfer, the membrane was incubated in a blocking buffer (Tris-buffered saline containing 5% bovine serum albumin) at room temperature for 1 h with agitation. After blocking, the membrane was incubated with the corresponding primary antibody diluent (1:2,000) overnight at 4°C. Then, the membrane was incubated with the species specific horseradish peroxidase conjugated secondary antibody (1:10,000 dilution; Cell Signaling Technology, USA) for 1 h at room temperature. The membrane was incubated with chemiluminescent substrate for 5 min. After that, the membrane was visualized and the images were captured using a camera in the FluorChem™ 5500 imaging system (Alpha Innotech Corp., San Leandro, CA, USA). Densitometric analysis was performed to evaluate band intensities using Image J software (National Institutes of Health, Bethesda, MD, USA). The values of targeted band density were normalized to those of α-tubulin obtained from the same membrane. Three independent experiments were conducted.

Data obtained from the Cytoskeleton Regulators RT² Profiler PCR array was analyzed by PCR Array Data Analysis Web Portal (www.SABiosciences.com/pcrarraydataanalysis.php.) using the ΔΔCt method.

Data obtained in the RT-qPCR, viability, migration, invasion, and western blot assays was analyzed using one-sample t-test by GraphPad InStat version 3.05 for windows (GraphPad Software, Inc., San Diego, CA, USA). Moreover, paired t-test was used to statistically matched tissues, and one-way analysis was used to compare means of three stages in CRC tissues. Quantitative results were expressed as the mean ± standard deviation. A two-sided P-value <0.05 was set as the criteria for statistical significance.

Multiple previous studies have shown that miR-29a-3p was overexpressed in CRC tissue (3,18,19). In the present study, miR-29a-3p was detected in five human colorectal cancer cell lines, namely HCT116, Caco2, HT29, SW480 and SW620. The result of RT-qPCR showed that miR-29a-3p was detectable in all the five colorectal cancer cell lines (Fig. 1A).

Transfection optimization indicated that three out of the five cell lines, SW480, SW620 and HCT116, were transfected successfully. Cell death siRNA acted as positive control in transfection optimization as cells which are successfully transfected will be killed. It was found that majority of SW480, SW620 and HCT116 cells were killed after transfection with cell death siRNA but no death was observed with HT29 and Caco2 (data not shown). Since miR-29a-3p was overexpressed in CRC, loss-of-function study using miR-29a-3p inhibitor was performed to determine the effect of miR-29a-3p inhibitor on viability of SW480, SW620 and HCT116. Cells were transfected with miR-29a-3p inhibitor or inhibitor control, and MTT assay was performed to detect cell viability after 96 h. The average value of cell viability was divided by that of the corresponding negative control (inhibitor control) which was set as 1. The data displayed in Fig. 2 show the means ± SD obtained from 3 independent experiments. Statistical analysis was performed using one-sample t-test. The result demonstrated that miR-29a-3p inhibition had no significant effect on cell viability in the three cell lines (P>0.05).

Since miR-29a-3p is overexpressed in CRC tissue according to the previous studies, it is implicated as an oncogene. To test the possible inhibitory effect of miR-29a-3p inhibitors on cell motility, invasive cells are required to perform invasion and migration assay. In order to evaluate the intrinsic invasive capability of the three CRC cell lines (HCT116, SW480 and SW620), Transwell invasion assay was performed without any treatment. Unfortunately, the invasive propensities of the three CRC cell lines were very poor and none of them achieved adequate invasion for the following quantification. For SW620 and HCT116, almost no cell could pass through the insert, whilst a few invaded SW480 cells were observed at the bottom of the insert. Therefore, an invasive subpopulation called SW480-7 was derived from SW480 cell line in vitro using the Transwell Matrigel invasion assay mentioned above. Cells that invaded to the bottom of the membrane were detached with 0.25% trypsin and cultured in a new cell culture flask filled with RPMI-1640 medium. An invasive subpopulation of SW480 called SW480-7 was established via 7 sequential passages through Matrigel-coated insert.

The invasive subpopulation, SW480-7, was established from the original SW480 cell line. To determine if the miR-29a-3p expression was altered in SW480-7 cells, real-time PCR was performed to assess miR-29a-3p in SW480 and SW480-7 cells (Fig. 1B). Unpaired t-test was employed to perform statistical analysis, and the results indicated that no significant difference was observed in miR-29a-3p expression between SW480 and SW480-7 cells (10.3±1.14 vs 12.9±1.57; P=0.081).

To further determine the effect of miR-29a-3p on cell motility, invasion assay and migration assay was carried out using Transwell assay. miR-29a-3p mimic and inhibitor were employed to perform gain-of-function and loss-of-function, respectively.

First, SW480-7 cells were transfected with miR-29a-3p inhibitor or negative control followed by cell invasion assay. The invasive capacity of the negative control group was used as the calibrator and set to 1. Unexpectedly, the result showed there was a significant stimulation of cell invasion by the miR-29a-3p inhibitor (2.3±0.12, Fig. 3A). The subsequent cell migration assay showed that the miR-29a-3p inhibitor promoted cell migration (2.4±0.45, Fig. 3B). Gain-of-function analyses using miR-29a-3p mimic were performed to further confirm the biological function of miR-29a-3p on cell migration and invasion. miR-29a-3p mimic or negative control was introduced into SW480-7 cells followed by Transwell cell invasion assay and migration assay. The results of gain-of-function study indicated that miR-29a-3p overexpression in SW480-7 cells suppressed cell invasion (0.58±0.12) and migration (0.64±0.13) significantly supports the observation of loss-of-function study (Fig. 3C and D).

To elucidate the underlying mechanism of alteration in migration caused by miR-29a-3p inhibition in vitro, Cytoskeleton Regulators RT² Profiler PCR array was performed 48 h after transfection. The dynamic reorganization of cell cytoskeleton is an essential requirement of cell motility process. The expression of 84 cytoskeletal regulatory genes was analysed and multiple altered genes due to miR-29a-3p inhibition was revealed (Table III). Two filter conditions were applied to select target genes for further study. One is fold change and the other is cycle threshold (Ct) value. The Ct value is inversely proportional to the amount of the gene detected in the sample, which means the higher the Ct level the lesser the amount of target gene. The genes regulated by a certain miRNA are divided into two general categories: direct targets and indirect targets. miRNA can suppress gene expression of direct targets through binding to the complementary sequence at 3′ UTR, and the other genes at downstream are indirect targets. Three widely used databases, TargetScan (http://www.targetscan.org/), miRBD (http://mirdb.org/miRDB/) and DIANA (http://diana.imis.athena-innovation.gr/DianaTools/), were employed to identify the potential direct target genes of miR-29a-3p in the 84-gene panel (Fig. 4). Five genes, namely CDC42, CDC42BPA, BAIAP2, MAPRE1 and TIAM1, were predicted as possible direct targets of miR-29a-3p by at least 2 databases, and the other seventy-nine genes were potential indirect targets. Since the interaction between a microRNA and the corresponding mRNAs has not been elucidated, all the predicted direct targets need to be verified experimentally. RT² Profiler PCR array revealed that three out of the 5 genes were upregulated in SW480-7 cells transfected with miR-29a-3p inhibitor. These three mRNAs were CDC42BPA (2.33-fold), BAIAP2 (1.79-fold) and TIAM1 (1.77-fold). Since CDC42BPA was the gene with the largest-fold change of gene expression, it was selected as the representative predicted direct target for siRNA studies.

For the indirect targets of miR-29a-3p, the cut-off value for fold changes and the Ct value were set to 3.5 and 29, respectively. As a result, four genes, IQGAP2, PHLDB2, SSH1 and PAK1, fulfilled the inclusion criteria. Taken together, RT² Profiler PCR array revealed that 5 genes, CDC42BPA, IQGAP2, PHLDB2, SSH1 and PAK1, were differentially expressed due to miR-29a-3p inhibition.

To further validate the accuracy and reproducibility of data from RT² Profiler PCR array, qRT-PCR was repeated in three independent experiments to investigate the five selected genes (CDC42BPA, IQGAP2, PHLDB2, SSH1 and PAK1) and the results are summarized in Fig. 5A and B. Statistical analysis was done using one-sample t-test. The corresponding expression levels of control groups were used as calibrators which were set as one, and relative mRNA expression of each gene was compared to control group. In SW480-7 cells transfected with miR-29a-3p inhibitor, mRNA expression levels of CDC42BPA, IQGAP2, PHLDB2 and SSH1 were increased to 3.3-, 10.1-, 4.4- and 5.0-fold, respectively, while the expression of PAK1 was reduced to 0.32-fold significantly (Fig. 5A). Fig. 5B shows the effects of the miR-29a-3p mimic on the expression of CDC42BPA, IQGAP2, PHLDB2, SSH1 and PAK1 mRNAs in SW480-7 cells. In this study, miR-29a-3p mimic could reverse the role of miR-29a-3p inhibitor, decreased the expression of CDC42BPA (0.37-fold), IQGAP2 (0.58-fold), PHLDB2 (0.62-fold) and SSH1 (0.38-fold) mRNAs, and increased PAK1 (2.63) mRNA. The results of qRT-PCR in Fig. 5A and B were consistent with result from RT² Profiler PCR array.

The protein level is influenced by several factors apart from mRNA abundance. Thus, besides validating PCR array data at the mRNA level, it is also important to determine the protein levels. Three out of the five genes, CDC42BPA, IQGAP2 and SSH1 were selected for protein expression analysis due to the following reasons: i) CDC42BPA is the only potential direct target of miR-29a-3p. ii) IQGAP2 is the gene showed the largest-fold change and lowest Ct value (highest mRNA abundance). iii) Both polymerization and depolymerization of actin need to be activated to promote cell invasion and migration. SSH1 is the only gene related to depolymerization.

Cells were collected at 72 h after miR-29a-3p inhibitor and miR-29a-3p mimic transfection and the protein levels were assessed by western blot analysis (Fig. 5C). Statistical analysis was done using one-sample t-test and the corresponding P-values were shown in the bar charts (Fig. 5D). The western blot analysis results were consistent with the mRNA results. Transfection of miR-29a-3p inhibitor increased protein expression of CDC42BPA, IQGAP2, and SSH1, while miR-29a-3p mimic reduced protein expression of these three genes.

Small interfering RNA (siRNA) is a type of synthetic non-coding RNA (21–23 nucleotides) that inhibits gene expression by binding to a specific mRNA molecule (20). In recent years, siRNAs have been used to investigate single gene function. In the present study, two CDC42BPA-specific siRNAs, siRNA-6 and siRNA-7, were used to silence CDC42BPA. There were four groups: miR-29a-3p inhibitor group (group A) was transfected with miR-29a-3p inhibitor only; miR-29a-3p inhibitor + siRNA control group (group B) was cotransfected with miR-29a-3p inhibitor and siRNA control; miR-29a-3p inhibitor + siRNA-6 group (group C) was cotransfected with miR-29a-3p inhibitor and siRNA −6; miR-29a-3p inhibitor +siRNA-7 group (group D) was cotransfected with miR-29a-3p inhibitor and siRNA-7. RT-qPCR and western blot analyses were carried out to verify CDC42BPA silencing at mRNA level and protein level, respectively. The silencing of CDC42BPA mRNA expression induced by siRNA-6 and siRNA-7 was summarized in Fig. 6A. Statistical analysis was done using one-sample t-test. The CDC42BPA expression level of group B was used as calibrator which was set as 1, and relative mRNA expression of other groups was compared to control group. As a result, mRNA expression levels of CDC42BPA were reduced significantly by siRNA-6 (P=0.0001) and siRNA-7 (P=0.0003). Both siRNA-6 and siRNA-7 were observed to induce silencing with greater than 80% efficiency. In accordance with the mRNA expression results, the western blot analysis results indicated that either siRNA-6 or siRNA-7 suppressed the protein expression of CDC42BPA (Fig. 6B).

To demonstrate the functional relevance of CDC42BPA in miR-29a-3p inhibitor-induced stimulation of cell invasion and migration, siRNA-6 and siRNA-7, were employed to silence CDC42BPA followed by cell invasion and migration assay. The results of cell invasion assay showed that cell invasion in siRNA control group was 2.11-fold higher than that of NC group, while siRNA-6 group and siRNA-7 group exhibited 1.49- and 1.61-fold higher cell invasion, respectively (Fig. 6C). The cell invasion capability of siRNA control group (p=0.0137), siRNA-6 group (p=0.0338) and siRNA-7 group (p=0.0131) were higher than that of NC group. However, the cell invasion capability of siRNA-6 group (P<0.05) and siRNA-7 group (P<0.05) was lower than that of siRNA control group. The results of migration assay indicated that cell migration of siRNA control group was 2.24-fold higher than that of NC group, while siRNA-6 group and siRNA-7 group exhibited 1.52- and 1.48-fold higher cell migration, respectively (Fig. 6D). The cell migration capability of siRNA control group, siRNA-6 group and siRNA-7 group were higher than that of NC group. However, the cell migration capability of siRNA-6 group and siRNA-7 group was lower than that of siRNA control group. Taken together, all the results suggested that CDC42BPA knockdown partially rescued the miR-29a-3p inhibitor-induced stimulation of cell invasion and migration in SW480-7 cells.

The expression of miR-29a-3p was assessed in 28 human colorectal cancer tissues and matched normal adjacent tissues (Fig. 7A). Of the 28 pairs of matched tissues, 27 pairs showed elevated expression of miR-29a-3p in tumor tissue. Statistical analysis was done using paired t-test. The paired comparison of miR-29a-3p expression between tumor tissues and normal adjacent tissues from the same patients showed a significant increase of miR-29a-3p expression in colorectal cancer tissues (2.803±1.852 vs 1.121±0.557; P=0.0115).

According to TNM staging system, these colorectal cancer samples can be divided into three subgroups: stage I consisting of 10 samples, stage II consisting of 9 samples, and stage III consisting of 9 samples. The expression pattern of miR-29a-3p in various stages is shown in Fig. 7B. One-way analysis of variance was used to compare means of three stages indicating that the highest level was found in stage I (P<0.05 vs stage II; P<0.05 vs stage III), while there was no significant difference between stage II and stage III (P>0.05). Furthermore, correlation between miRNA expression and colorectal cancer staging was analyzed using Spearman rank correlation test. The result indicated that there was negative correlation between miR-29a-3p expression and colorectal cancer staging (P=0.0032). However, the data need to be interpreted with caution due to the small sample size.