Loss of miR-638 in vitro promotes cell invasion and a mesenchymal-like transition by influencing SOX2 expression in colorectal carcinoma cells

Participants who provided samples also provided written, informed consent to participate in this study. The Ethics Committee of the Shanghai Cancer Institute approved the study, the consent procedure, and the tissue array study. All of the research was performed in China. Paired colorectal tumor tissues and their corresponding adjacent non-tumor colorectal tissues (5 cm away from the lesions) were collected from patients who underwent curative surgery for CRC at Anhui Medical University, Anhui Province, China. Normal colon tissue was collected from patients with non-cancerous colon disease. A CRC diagnosis was confirmed by histological examination, and the relevant clinical and pathological information was retrieved from the hospital database (Additional file1: Table S1a). Glass-slide tissue arrays for CRC were purchased from the Shanghai Outdo Biotech Co., Ltd. (Shanghai, China) (Additional file1: Table S2a), and immunostaining (SOX2, ab75485, 1:100, Abcam, Cambridge, MA; vimentin, #5741, 1:50, Cell Signaling Technology, Beverly, MA) was performed on the tissue microarray slides. Staining was analyzed based on the percentage of positively stained cells and staining intensity by a pathologist or using Image-Pro Plus 6.0 software (Media Cybernetics, Inc., Bethesda, MD) (Additional file2).

Four human CRC cell lines were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). HCT-116 cells (ATCC No. CCL-247) were maintained in McCoy’s 5A medium, LoVo cells (ATCC No. CCL-229) were maintained in F-12 K medium (Kaighn’s Modification of Ham’s F-12 Medium), and SW480 cells (ATCC No. CCL-228) and SW1116 cells (ATCC No. CCL-233) were maintained in Leibovitz’s l-15 medium. The media were supplemented with 10% fetal bovine serum, and the cells were incubated in a humidified atmosphere of 95% air and 5% CO₂ at 37°C.

miRNA mimics and miRNA antagomiRs were designed and synthesized by RiboBio (Guangzhou, China). The miRNA antagomiRs were composed of nucleotides with a 2′-O-methylmodification. The SOX2 siRNAs (sense, 5′-GGAAUGGACCUUGUAUAGAUC-3′; anti-sense, 5′-UCUAUACAAGGUCCAUUCCCC-3′) were synthesized by RiboBio (Guangzhou, China), and the SOX2 overexpression construct was obtained from Origene Inc. (Beijing, China). The miRNA mimics, miRNA antagomiRs, SOX2-targeted siRNA, and SOX2 overexpression construct were transiently co-transfected with GFP (transfection efficiency control) using an Amaxa Nucleofector (Amaxa, Koeln, Germany) according to the manufacturer’s protocol.

Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. qRT-PCR with miRNA was performed using the TaqMan Reverse Transcription Kit (Applied Biosystems), TaqMan MicroRNA Assays (Applied Biosystems), and a LightCycler TaqMan Master (Roche Diagnostics, Mannheim, Germany) according to the manufacturers’ instructions. The miRNA expression levels were calculated using the delta-delta Ct method with RNU6B as an internal control. A Ct value of 35 was set as the cut-off value for defining as non-detected.

cDNA was reverse-transcribed from 1 μg of RNA using the SYBR®Prime Script™ RT-PCR kit (Takara Biochemicals, Tokyo, Japan), and the reactions were performed on an ABI PRISM®7900HT Real-Time PCR System. The thermal cycling conditions were as follows: an initial step at 95°C for 15 s followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Each experiment was performed in a 20-μl reaction volume containing 10 μl of SYBR® Prime Ex Taq™ II (2×), 0.8 μl of forward primer and reverse primer (10 μM each), 0.4 μl of ROX Reference Dye or Dye II (50×), 2 μl of cDNA, and 6 μl of H₂O. β-Actin was used as an internal control. The quantification of the mRNA was calculated using the comparative Ct (the threshold cycle) method according to the following formula: Ratio = 2^{-ΔΔ ct} = 2^{- [Δ Ct(sample) - Δ Ct(calibrator)]}, where ΔCt is equal to the Ct of the target gene minus the Ct of the endogenous control gene (β-actin). The primers for SOX2 (SOX2RTF: 5′-CGAGATAAACATGGCAATCAAAAT-3′; SOX2RTR: 5′-AATTCAGCAAGAAGCCTCTCCTT-3′) have been described previously[18]. The internal control actin primers were designed as previously described[19, 20].

Proteins were separated by SDS-PAGE and transferred to a polyvinylidene fluoride membrane (Bio-Rad, Hercules, CA). The membrane was blocked with 5% non-fat milk and incubated with the following antibodies: the epithelial cell marker ZO-1 (1:200) and E-cadherin (1:1000), the mesenchymal cell marker vimentin (1:50,000), SOX2 (1:2,000), Myc (1:2,000), β-actin (1:2,000, Santa Cruz Biotechnology, Santa Cruz, CA), and GAPDH (1:10,000; Kang-Chen Bio-tech Shanghai, China). And the quantification of Western Blot exerted by ImagineJ software (NIH, USA).

We transfected the miR-638 mimics and SOX2 siRNA into HCT-116 and SW1116 cells using an Amaxa Nucleofector (Amaxa, Koeln, Germany). Cell migration and invasion were evaluated using Matrigel-uncoated and -coated transwell chambers (cat. 3422, Corning, NY). Briefly, 5 × 10⁴ cells were suspended in 200 μl of DMEM without serum and placed into cell culture inserts (8-μm pore size; BD Falcon, San Jose, CA) of a companion plate (BD Falcon San Jose, CA) in pre-warmed culture medium containing 10% fetal bovine serum in the well. The cells were incubated overnight at 37°C in 5% CO₂ and then fixed in 4% paraformaldehyde in PBS. Cell migration and invasion were determined by staining the cells with 0.1% crystal violet (Sigma, St Louis, MO) and counting the cells under a light microscope (100× magnification) in eight randomly selected areas.

Transfected SW1116 cells were seeded at a density of 2 × 10⁴ onto poly-L-lysine-coated glass coverslips in a 6-well plate. After further culture overnight, the cells were permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, St. Louis, MO). For filamentous actin (F-actin) staining, the coverslips were incubated with TRITC-labeled phalloidin (Sigma-Aldrich, St. Louis, MO) at room temperature, and the cell nuclei were counterstained with DAPI. The cells were co-transfected with 40 ng of pEGFP plasmid as a control.

All experiments were performed in triplicate. The data are presented as the mean values ± standard error of the mean (SEM) and were analyzed using Student’s t-test. p values less than 0.05 were considered significant. Statistical analyses were performed using GraphPad Prism 5.01 software (GraphPad Software Inc., San Diego, CA).

The accession numbers for miR-638 is MIMAT0003308, and that for SOX2 is NM_003106.2.

Previous microarray analyses revealed that 23 miRNAs are downregulated in CRC tissues (Additional file1: Table S3), including miR-497[21], miR-9[22], miR-30a[23], and miR-139[24]. To further screen miRNAs that are deregulated in CRC, qRT-PCR assays were conducted to evaluate the expression levels of these miRNAs in 36 pairs of CRC clinical samples. In addition to the four miRNAs described above, miR-638 was markedly downregulated in CRC tissues. The expression levels of miR-638 were decreased in 83.33% the samples (30/36; Figure 1B, Additional file3: Table S1b) and a 22.98% decrease in expression in the CRC tissue samples compared with adjacent noncancerous tissue samples (2.323 to 1.789, p < 0.0001; Figure 1A). And a 27.28% decrease in moderately differentiated samples and 61.29% decrease in poorly differentiated samples compared to well-differentiated samples (Figure 1C). The miR-638 levels in all four CRC cell lines (HCT116, LoVo, SW1116, and SW480) were downregulated compared with that of normal colorectal tissues (Figure 1D). These results demonstrate that miR-638 showed reduced expression levels in CRC and was inversely correlated with tumor differentiation.

To understand the biological effect of miR-638 deregulation on the development of colorectal carcinoma, gain- or loss-of-function analyses were performed using an overexpression or silencing strategy through the transfection of miR-638 mimics or antagomiRs (using an Amaxa Nucleofector device) into the CRC cell lines HCT-116 and SW1116 (the miR-638 levels in the CRC cells were confirmed through qRT-PCR; Figure 2A and2B). miR-638 elicited an apparent effect on the cell motility of CRC cells. Matrigel-coated (for invasion) and -uncoated (for migration) transwell assays were used to determine the invasiveness and migration of HCT-116 and SW1116 cells after transfection with miR-638 mimics or antagomiR-638. miR-638 overexpression reduced the number of invasive HCT-116 and SW1116 cells by 37.6% and 43.0%, respectively; however, antagomiR-638 enhanced cell invasion up to 20.8% and 27.6%, respectively (Figure 2C, Additional file4: Figure S1A). Moreover, miR-638 overexpression inhibited HCT-116 and SW1116 cell migration by 23.4% and 33.1%, respectively, whereas the inhibition of miR-638 expression enhanced cell migration up to 22.6% and 23.5%, respectively (Figure 2D, Additional file4: Figure S1B). These data show that miR-638 inhibited CRC cell invasion and migration.

The loss of miR-638 expression in poorly differentiated CRC tissues and its role in promoting cell migration and invasion suggest that miR-638 may be involved in the EMT process. The mesenchymal-like transition includes changes in cell morphology, markers, and motility. To confirm this hypothesis, we first examined the morphology of SW1116 cells with altered miR-638 expression levels. The data showed that the cell morphology was significantly altered. The cell-to-cell contacts were increased in the cells with miR-638 overexpression; in contrast, when antagomiR-638 was transfected, lamellipodium formation was promoted and cell-to-cell contracts were decreased (Figure 3A). Furthermore, we examined epithelial and mesenchymal cell marker expression levels through Western blotting. In the miR-638 mimic-transfected SW1116 cells, the epithelial cell marker zonula occludens-1 (ZO-1) and E-cadherin were upregulated compared with SW1116 cells transfected with the NC mimics. In contrast, the level of the mesenchymal cell marker vimentin was decreased (Figure 3B). Conversely, in the antagomiR-638-transfected SW1116 cells, ZO-1 and E-cadherin were downregulated compared with the NC antagomiR-transfected SW1116 cells, whereas vimentin levels increased (Figure 3B). Taken together, these results indicate that the inhibition of miR-638 in SW1116 cells resulted in mesenchymal-like features (such as stretched lamellipodia, reduced cell-to-cell contact, decreased epithelial cell marker ZO-1 and E-cadherin expression, and increased mesenchymal cell marker vimentin expression). In contrast, miR-638 overexpression resulted in epithelial-like features (such as increased cell-to-cell contact, increased ZO-1 and E-cadherin expression, and decreased vimentin expression).

Identification of the target genes of miR-638 is essential for the elucidation of its biological functions. First, we predicted the target genes of miR-638 from miRNA.org, and the top genes with the lowest mirSVR score (cutoff = -1.2, Additional file5: Table S4) included PLXDC2, SOX2, TCERG1L, and WDR4, which were considered putative candidate targets. Second, the 3′-UTRs of these four genes were subcloned into a luciferase reporter vector to evaluate the influence of miR-638 on the expression of these genes through a luciferase assay (the PCR primers are listed in Additional file1: Table S5). The results showed that miR-638 more drastically suppressed the luciferase activity of the SOX2 3’-UTR-containing reporter gene construct than the other three constructs (Figure 4A). To further determine whether SOX2 was a bona fide target of miR-638, we mutated the predicted binding site of miR-638 on the SOX2 3′-UTR (Figure 4B) and found that the mutant SOX2 3′-UTR reporter gene completely abolished miR-638-mediated repression (Figure 4C). Next, we determined the SOX2 mRNA and protein expression levels in HCT-116 and SW1116 cells with altered miR-638 expression levels. Our data revealed that miR-638 overexpression inhibited SOX2 mRNA expression by 69.97% and 61.02% compared with the negative control group in the HCT-116 and SW1116 cells, respectively; furthermore, SOX2 mRNA levels were upregulated in antagomiR-638-transfected HCT116 and SW1116 cells (Figure 4D). Moreover, the SOX2 protein expression levels were downregulated in the miR-638 mimic-transfected CRC cell lines compared with the negative control (mimics negative control), and SOX2 expression was upregulated in the antagomiR-transfected CRC cell lines compared with the negative control (antagomiR negative control; Figure 4E). Furthermore, we detected SOX2 mRNA expression levels in CRC tissues in which miR-638 expression was detected. The data show that the SOX2 mRNA expression level in the miR-638 high-expression group was only 73.04% of that observed in the miR-638 low-expression group (Figure 4F). Taken together, these results indicate that miR-638 inhibited SOX2 expression through direct binding to the 3′-UTR of SOX2.

To further evaluate the correlation between miR-638 and SOX2 expression in human CRC, we assessed SOX2 expression using immunohistochemistry (IHC) in a CRC tissue array (90 pairs of CRC tissues; Additional file3: Table S2b). Of the 90 cases, 73 tumors exhibited increased SOX2 expression levels compared with patient-matched, adjacent non-cancerous tissues (Figure 5A). The SOX2 staining scores for the CRC tissue samples were higher than those for their adjacent normal tissue samples (Figure 5B). Furthermore, the SOX2 expression levels in the tumor tissues were inversely correlated with TNM stage (Figure 5C). We then quantified the intensity of SOX2 expression using Image-Pro Plus 6.0 software. The results indicated that the SOX2 staining intensity values were upregulated in CRC tissue samples by 40.45% compared with those of the adjacent non-cancerous tissues (1.4045/1; the SOX2 IHC average intensity value in CRC samples was designated as 1) and in 86.67% of the samples (78/90; Figure 5D). Representative SOX2 IHC staining in CRC tissue samples No. 448 and No. 586 is shown (Figure 5E and5F). These data demonstrate that SOX2 was frequently highly expressed in CRC tissues compared with adjacent non-cancerous tissues and was associated with tumor grade, in contrast to the miR-638 expression levels in CRC.

To further determine whether SOX2 is involved in miR-638-induced CRC cellular invasiveness, we inhibited SOX2 using siRNA, which was confirmed through Western blotting (Figure 6A). The results revealed that SOX2-depleted cells showed significantly reduced cell invasion (Figure 6B, Additional file6: Figure S2A) and migration (Figure 6C, Additional file6: Figure S2B), which mimicked the effect of miR-638. These data indicate that the knockdown of SOX2 can suppress CRC cell migration and invasion, similar to miR-638 overexpression.

To determine whether SOX2 is involved in the miR-638-induced inhibition of invasion and mesenchymal-like transition in CRC cells, we increased SOX2 expressions using an overexpression construct in miR-638 mimic-transfected CRC cells. The transfection efficiency was 100% using an Amaxa Nucleofector (with the co-transfection of GFP as an indicator). After the cells were co-transfected with the SOX2 overexpression construct and miR-638, we confirmed SOX2 overexpression by Western blotting and then detected the expression of the epithelial cell marker ZO-1/E-cadherin and the mesenchymal cell marker vimentin. The data showed that the ZO-1/E-cadherin expression levels decreased, whereas the vimentin levels increased (Figure 7A, Additional file7: Figure S4). Furthermore, the cell morphology changed (Figure 7B), the cell-to-cell contacts decreased, and the lamellipodia were stretched. SOX2 overexpression reversed the miR-638 inhibition of cell invasion and cell migration (Figure 7C and D, Additional file8: Figure S3).

Taken together, these data indicate that SOX2 can rescue the miR-638-induced inhibition of cell invasion and mesenchymal-like transition.