Role of Intracellular and Extracellular MicroRNA-92a in Colorectal Cancer

Abstract

Colorectal cancer is one of the leading causes of cancer-related death worldwide. Previous studies have shown that miR-92a has an oncogenic function in several cancers and that its up-regulation is correlated with malignant clinicopathologic behaviors of colorectal cancer. It also has been suggested that circulating miR-92a in patients' plasma can be a potential biomarker for colorectal cancer. However, the precise roles of intracellular and extracellular miR-92a are not yet understood. In this study, we examined the expression levels of miR-92a in colorectal tumors (38 cancer specimens and 56 adenoma specimens) and paired adjacent nontumorous tissues. Increased expression of miR-92a was frequently observed in the cancers compared with that in the adenomas and was correlated with advanced clinical stages, tumor depth, and size. We also demonstrated that the levels of miR-92a within microvesicles (MVs) in the plasma of mice bearing colon cancer xenografts were significantly increased compared with those in control mice. One of the roles of intracellular and extracellular miR-92a was shown to be down-regulation of Dickkopf-3 (Dkk-3), a presumed tumor suppressor gene. Within the colon cancer cells, suppression of Dkk-3 by miR-92a contributed to the cell proliferation. Extracellular miR-92a packed within MVs secreted by colon cancer cells was delivered into endothelial cells and contributed to the proliferation and motility of these cells through down-regulation of the same target gene, Dkk-3. These data suggest that intracellular and extracellular miR-92a had important roles in tumor growth and the tumor microenvironment in colorectal cancer.

Introduction

Colorectal cancer is one of the leading causes of cancer-related death, causing more than 600,000 deaths per year all over the world. In spite of early screening and the development of new chemotherapeutic strategies, the survival rate of colorectal cancer has not been essentially improved in the past 20 years. Thus, novel biomarkers more reliable for the early diagnosis and early treatment to improve the survival rate are urgently needed.

MicroRNAs (miRNAs) are single-stranded noncoding small RNAs that repress translation or induce degradation of target mRNAs through binding to specific complementary sites within the 3′ untranslated region (3′UTR) of mRNAs [1,2]. A growing body of evidence indicates that miRNAs play crucial roles in almost all biologic processes including proliferation, differentiation, cell survival, and cell death [3]. It has been also elucidated that dysregulation of miRNA expression contributes to various human diseases, including cancer [4]. Up-regulation and down-regulation of specific miRNAs have been described across different cancer types, and they are known as important regulators of oncogenes and tumor suppressor genes [5,6].

Although the majority of miRNAs exist within cells, a number of miRNAs have been found in extracellular body fluids including serum, plasma, saliva, urine, and other body fluids [7]. Circulating miRNAs packed within membranous vesicular structures named shedding microvesicles or microvesicles (MVs) are highly stable and resistant to RNase activity [8]. MVs are 100- to 1000-nm membranous vesicular carriers released by various kinds of cell surfaces; especially, cancer cells are known to secrete large amounts of MVs containing various genetic information into their surroundings, and MVs are now considered to be an important cell-to-cell communication tool [9,10]. Furthermore, it has been shown that the profiles of genetic products packed within MVs are tightly correlated with various characteristics of cancer [11–13]. Thus, there is a potential for circulating miRNAs carried by MVs (miRNAs/MVs) to be novel noninvasive cancer biomarkers for screening, diagnosis, and monitoring [10,14,15]. However, the precise role of circulating miR-92a/MVs in colorectal cancer patients is not yet fully understood.

In this study, we quantified the circulating miR-92a/MVs in the plasma of mice bearing colon cancer xenografts and confirmed a significant increase in the miR-92a level in their plasma. We also quantified the miR-92a expression in colorectal tumor tissues and adjacent normal mucosa and examined the association between miR-92a expression and clinicopathologic findings to determine the clinical significance of miR-92a expression in colorectal tumors. We further examined the role of intracellular and extracellular miR-92a, and our data implied that miR-92a secreted by colon cancer cells through MVs could contribute to the establishment of a tumor microenvironment promoting angiogenesis, as indicated by increased cell growth, motility, and tube formation of human umbilical vein endothelial cells (HUVECs), through the targeting of Dickkopf-3 (Dkk-3).

Materials and Methods

Patients and Samples

All human samples were obtained from patients who had undergone biopsy for diagnosis or surgery for resection at Fujita Health University Hospital (Toyoake, Aichi, Japan), Osaka Medical College Hospital (Takatsuki, Osaka, Japan), Kyouritsu General Hospital (Nagoya, Aichi, Japan), or Saiseikai Ibaraki Hospital (Ibaraki, Osaka, Japan). Informed consent in writing was obtained from each patient. Collection and distribution of the samples were approved by each of the appropriate institutional review boards. Thirty-eight patients with previously untreated (or recently diagnosed) colorectal cancer and 56 with adenomas were selected. The cancer study group consisted of 13 women and 25 men with a median age of 70 years (range, 38–88), and the adenoma group consisted of 25 women and 31 men with a median age of 59 years (range, 21–75). The distribution according to other clinical parameters is shown in Tables 1 (cancer and adenoma), 2 (cancer), and 3 (adenoma). Under a pathologist's supervision, all tissue sample pairs were collected from surgically or endoscopically resected tissues, with these paired samples being from the primary tumor and its adjacent nontumor mucosal tissue in the same patient. All samples were immediately stored in liquid nitrogen until RNA isolation could be performed. Samples were homogenized by use of a spatula under sterile conditions before total RNA isolation. Total RNAs of human normal tissues (Human Total RNA Master Panel II) were purchased from Clontech (Mountain View, CA).

Table 1.

Characteristics of Study Population and Expression of miR-92a in Colorectal Tumors.

Colorectal Tumors	n	Expression of miR-92a↑ [Case (%)]
Sex
Male	56	23 (41.1)
Female	38	10 (26.3)
Tumor
Cancer	38	25 (65.8)*
Adenoma	56	8 (14.3)
Location
Right colon	31	9 (29.0)
Left colon	63	24 (38.1)

^*P < .0001.

Table 2.

Characteristics of Study Population and Expression of miR-92a in Colorectal Cancer.

Colorectal Cancer	n	Expression of miR-92a↑ [Case (%)]
Sex
Male	25	16 (64.0)
Female	13	9 (69.2)
Age
<69	18	13 (72.2)
>70	20	12 (60.0)
Clinical stage
0	7	2 (28.6)*
I	10	6 (60.0)
II	8	8 (100.0)
IIIa	9	7 (77.8)
IIIb	1	1 (100.0)
IV	3	1 (33.3)
Depth
Mucosa (M)	7	2 (28.6)†
Submucosa (SM)	6	5 (83.3)
Mucosa propria (MP)	4	1 (25.0)
Subserosa (SS)	8	7 (87.5)
Serosa exposure, serosa invasion (SE, SI)	13	10 (76.9)
Tumor diameter (mm)
<40	19	16 (84.2)‡
>40	19	9 (47.4)
Dukes classification system
A	24	14 (58.3)
B	2	2 (100.0)
C	12	9 (75.0)
Location
Right colon	11	7 (63.6)
Left colon	27	18 (66.7)

^*Clinical stage 0 versus I to IV; P = .022.

^†Depth M versus SM-SI; P = .022.

^‡P = .017.

Table 3.

Characteristics of Study Population and Expression of miR-92a in Colorectal Adenoma.

Colorectal Adenoma	n	Expression of miR-92a↑ [Case (%)]
Sex
Male	31	7 (22.6)*
Female	25	1 (4.0)
Age
<59	30	5 (16.7)
>60	26	3 (11.5)
Tumor diameter (mm)
<10	28	5 (17.8)
>10	28	3 (10.7)
Grade
Low-grade dysplasia	41	5 (12.2)
High-grade dysplasia	15	3 (20.0)
Location
Right colon	20	2 (10.0)
Left colon	36	6 (16.7)

^*P < .0001.

Cell Culture and Cell Viability

All human colon cancer cell lines, DLD-1, WiDr, COLO201, and SW480, used in this study were cultured in RPMI-1640 medium supplemented with 10% (vol/vol) heat-inactivated FBS (Sigma-Aldrich Co, St Louis, MO) and 2 mM l-glutamine under an atmosphere of 95% air and 5% CO₂ at 37°C. Low-passage HUVECs were seeded in six-well plates precoated with gelatin (BIOCOAT; Becton Dickinson, Franklin Lakes, NJ) and cultured in complete medium (EGM BulletKit; Lonza, Walkersville, MD) under an atmosphere of 95% air and 5% CO₂ at 37°C. Normal diploid fibroblast ASF-4-1 cells were also seeded in six-well plates precoated with gelatin (BIOCOAT; Becton Dickinson) and cultured in RPMI-1640 medium supplemented with 15% (vol/vol) heat-inactivated FBS under an atmosphere of 95% air and 5% CO₂ at 37°C. To collect cancer cell-derived MVs, we also used MV-free RPMI medium supplemented with 10% FBS. MV-free FBS was made and added into RPMI medium as follows: FBS was centrifuged at 3000 rpm for 5 minutes and its supernatant was filtered through a Millex-HV Filter Unit (0.45-µm pores; Merck Millipore, Billerica, MA). The flow through was ultracentrifuged at 100,000 rpm for 3 hours. Without disturbing the MV pellet, the supernatant was used as MV-free FBS. The number of viable cells was determined by performing the trypan blue dye exclusion test.

Human Tumor Xenograft Model

Our institute's committee for ethics in animal experimentation approved all animal experimental protocols. Animal experiments were conducted in accordance with the guidelines for Animal Experiments of Gifu International Institute of Biotechnology. Thirteen athymic nude mice were inoculated with human colon cancer DLD-1 cells at 1 x 10⁷ cells in 100 µl of phosphate-buffered saline (PBS) into their subcutaneous flank, and the inoculation time was set as day 0. Thirteen control mice received a PBS injection. All tumor volumes were monitored by measuring the length (L), width (W), and depth (D), and volumes were estimated according to the following formula: V (mm³)= L x W x D x 0.5. Blood sampling was carried out on all animals from the abdominal aorta at week 8 after the inoculation under general anesthesia with an intraperitoneal injection of 50 mg/kg pentobarbital sodium (Somnopentyl; Kyoritsu Seiyaku Co, Tokyo, Japan) and then sacrificed. Blood was stored at 4°C in 2-ml EDTA tubesfor furtherexperiments. Blood processing wasdonewithin 2 hours after the sampling.

RNA Isolation from Circulating MVs in Mouse Blood

We isolated total RNAs contained within the MVs in the mouse blood according to the protocol recommended in the ExoMir Kit (Bioo Scientific Co, Austin, TX). Briefly, whole blood samples were centrifuged at 6000g for 5 minutes to remove blood cells. The supernatant plasma (400–500 µl) was then diluted with 5 ml of PBS and filtered through a Millex-HV Filter Unit (0.45-µm pores; Merck Millipore) to remove cell debris. The filtered samples were further filtered through two filters (pore sizes of 0.22 and 0.02 µm) provided in the ExoMir Kit. The MVs were captured by the 0.02-µm filter after passage through the 0.22-µm filter. The total RNAs were extracted from the trapped MVs and used for analysis.

Transfection with miR-92a, antagomiR-92a, or Short-Interfering RNA for Dkk-3

DLD-1 cells, WiDr cells, and/or HUVECs were seeded in six-well plates at a concentration of 0.5 x 10⁵ per well (10–30% confluence) on the day before the transfection. The mature type of miR-92a (mirVana miRNA mimic; Ambion, Foster City, CA), antagomiR-92a (mirVana miRNA inhibitor; Ambion), or short-interfering RNA (siRNA) for Dkk-3 (siR-Dkk-3; Invitrogen, Carlsbad, CA) was used for the transfection of the cells, which was achieved by using cationic liposomes, Lipofectamine RNAiMAX (Invitrogen), according to the manufacturer's Lipofection protocol. The nonspecific control miRNA (HSS, Hokkaido, Japan) sequence was 5′-GUAGGAGUAGUGAAAGGCC-3′, which was used as a control for nonspecific effects [16]. The sequence of the mature type of miR-92a used in this study was 5′-GUCCAGUUUUCCCAGGAAUCCCUU-3′ and that of siR-Dkk-3 were 5′-GAUGAGUAUGAAGUUGGCAGCUUCA-3′ and 5′-CCCTCTTTGGCAGTTGCATTAGTAA-3′. The effects manifested by the introduction of miR-92a into the cells were assessed at 72 hours after the transfection.

Quantitative Reverse Transcription-Polymerase Chain Reaction Using Real-Time Polymerase Chain Reaction

Total RNA was isolated from cultured cells or tumor tissues by using TRIzol containing phenol/guanidium isothiocyanate (Applied Biosystems, Foster City, CA) and treatment with DNase I. RNA concentration and purity were assessed by UV spectrophotometry. RNA integrity was checked by formaldehyde gel electrophoresis. For determination of the expression levels of mRNAs, total RNA was reverse-transcribed with PrimeScript RT reagent Kit (TaKaRa, Otsu, Japan). Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed with primers specific for Dkk-3 by using SYBR Premix Ex Taq (TaKaRa). The primers for Dkk-3 were given as follows: Dkk-3-sense, 5′-TTC GGG TAG TGG AAA ACC AG-3′, and Dkk-3-antisense, 5′-CAG CAG CTC GAA TTT CTT CC-3′. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. All reactions were run in triplicate. The relative expression levels of mRNAs were calculated by the ΔΔC_t method. To determine the expression levels of miRNAs, we conducted qRT-PCR by using TaqMan MicroRNA Assays (Applied Biosystems) according to the manufacturer's protocol. In brief, RT reactions contained 25 ng of total RNA samples, 50 nM stem-loop RT primer (Applied Biosystems), and reagents from a TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems). The RT reactions were run in an ABI thermocycler for 30 minutes at 16°C, 30 minutes at 42°C, 5 minutes at 85°C, and then held at 4°C for 10 minutes. Real-time PCR was performed by using PCR primers (Applied Biosystems) and Premix Ex Taq (TaKaRa). The 20-µl PCR reaction mixture included 2.0 µl of RT product, 10 µl of Premix Ex Taq, and 1 µl of PCR primer. These reaction mixtures were incubated at 95°C for 10 seconds, followed by 40 cycles of 95°C for 5 seconds, and 60°C for 30 seconds. All reactions were run in triplicate. The relative expression levels of miR-92a were calculated by the ΔΔC_t method. RNU6B was used as an internal control.

Western Blot Analysis

Whole cells were homogenized in chilled lysis buffer comprising 10 mM Tris-HCl (pH 7.4), 1% NP-40, 0.1% deoxycholic acid, 0.1% sodium dodecyl sulfate, 150 mM NaCl, 1 mM EDTA, and 1% Protease Inhibitor Cocktail (Sigma-Aldrich Co) and stood for 20 minutes on ice. After centrifugation at 13,000 rpm for 20 minutes at 4°C, the supernatants were collected as whole-cell protein samples. Protein contents were measured with a DC Protein assay kit (Bio-Rad, Hercules, CA). Ten micrograms of lysate protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis using 12.5% polyacrylamide gels and electroblotted onto a polyvinylidene difluoride (PVDF) membrane (PerkinElmer Life Sciences, Inc, Boston, MA). After blockage of nonspecific binding sites for 1 hour with 5% non fat milk in PBS containing 0.1% Tween 20 (TBS-T), the membrane was incubated overnight at 4°C with antibodies against DKK-3 (Santa Cruz Biotechnology, Santa Cruz, CA) properly diluted with TBS-T containing 2% BSA and 0.01% sodium azide. The membrane was then washed three times with TBS-T, incubated further with HRP-conjugated goat anti-rabbit IgG antibody (Cell Signaling Technology) at room temperature, and then washed three times with TBS-T. The immunoblots were visualized by use of Amersham ECL Plus Western Blotting Detection Reagents (GE Healthcare, Buckinghamshire, United Kingdom). The quantity loaded was verified by re-incubating the same membrane with anti-β-actin antibody (Sigma-Aldrich Co).

Luciferase Assay

DLD-1 or WiDr cells were seeded into 12-well plates at a concentration of 0.5 x 10⁵ per well on the day before the transfection. Searching the Target Scan 6.2 database (http://www.targetscan.org/) to find algorithm-based binding sites of miR-92a, we found the predicted binding sites to be at position 1317–1324 in the 3′UTR of Dkk-3 mRNA. The sequence region 1081–1449 containing the putative binding sequence of miR-92a was inserted into a pMIR-REPORT Luciferase miRNA Expression Reporter Vector (Applied Biosystems) according to the manufacturer's protocol. Moreover, we made another pMIR construct encompassing a mutated seed sequence for miR-92a (wild type, GTGCAAT; mutant, GTTACAT) by using a PrimeSTAR Mutagenesis Basal Kit (TaKaRa). The mutation of the vector was confirmed by sequence analysis. pRL-TK Renilla luciferase reporter vector (Promega, Madison, WI) was used as an internal control vector. DLD-1 or WiDr cells were co-transfected with each reporter vector (0.1 µg/well each) and 20 nM miR-92a or nonspecific noncoding siRNA (Dharmacon, Tokyo, Japan), which was achieved by using Lipofectamine RNAiMAX. Luciferase activities were measured at 48 hours after co-transfection by using a Pikka-Gene Dual (TOYO B-Net, Co, Ltd, Tokyo, Japan) according to the manufacturer's protocol. Luciferase activities were reported as the firefly luciferase/Renilla luciferase ratio.

Matrigel Invasion Assay

DLD-1 and WiDr cells for antagomiR-92a transfection were seeded into six-well plates at a concentration of 0.5 x 10⁵ per well and transfected with antagomiR-92a. The cells were then trypsinized at 48 hours after the transfection, suspended at a concentration of 1.25 x 10⁵ per ml in 2 ml of serum-free RPMI, and seeded into 8-µm porous BioCoat Matrigel chamber inserts (BD Biosciences, San Jose, CA). The chamber inserts were placed in wells filled with 2.5 ml of RPMI supplemented with 10% FBS as chemoattractant. After 22 hours of incubation, the upper side of the filter was scraped with a cotton tip to eliminate cells that had not migrated through it. The invasive ability of the cells was determined by counting the cells stained with Wright-Gimsa that had migrated to the lower side of the filter. The counting was done from photomicrographs.

Wound Healing Assay

HUVECs were seeded into 12-well plates at a concentration of 0.5 x 10⁵ per well 2 days before the transfection. A straight line scratch, which simulated a wound, was made by the tip of a sterile pipette at 48 hours after transfection. The cells were then washed with PBS and refreshed with serum-free medium to exclude the effect of cell growth facilitated by serum. After overnight incubation at 37°C, the cells were fixed in absolute methanol. The initial gap length (0 hour) and the residual gap length (24 hours) after the scratches had been made were calculated from photomicrographs.

Tube Formation Assay

Tube formation assay was conducted according to the manufacturer's protocol of the Angiogenesis Starter Kit (Life Technologies, Carlsbad, CA). HUVECs were seeded into six-well plates at a concentration of 0.5 x 10⁵ per well on the day before the transfection. Tube formation assay was performed at 24 hours after transfection.

Tracking Experiment of Nascent miR-92a Derived from DLD-1 Cells into HUVECs Using MVs

Nascent miR-92a was traced according to the manufacturer's protocol of the Click-iT Nascent RNA Capture Kit (Invitrogen). Briefly, DLD-1 donor cells were incubated overnight in medium containing 0.2 mM 5-ethynyl uridine (EU, an alkyne-modified nucleotide), which is efficiently and naturally incorporated into the nascent RNA. After an overnight incubation, total RNA was extracted from the donor cells. To purify MVs from the donor's medium, we subjected the medium sequentially to centrifugation at 2000 rpm for 10 minutes, filtration through a 0.45-µm filter, and ultracentrifugation at 100,000 rpm for 3 hours. The MV-concentrated solution was diluted with 0.1 ml of PBS. This PBS solution containing MVs was added to the medium of recipient HUVECs or ASF-4-1 cells. After overnight incubation, total RNA was extracted from the recipient cells. As a negative control, we used EU-free RNAs from donor cells, MVs, and recipient cells. RNAs labeled with EU were extracted by using the copper-catalyzed click reaction with azide-modified biotin, which created a biotin-based handle for capturing nascent RNA transcripts on streptavidin magnetic beads.

Statistics

Each examination was performed in triplicate. The expression levels > 1.5 were designated as up-regulation and those <0.67 as down-regulation, with fold changes obtained from the results of linear discriminant analysis of miR-92a expression patterns from 38 pairs of colon tumors and nontumorous tissues. The tumor/nontumor ratio of miR-92a expression in the samples was also expressed by use of box-and-whisker plots. Statistical differences between clinicopathologic parameters and the miR-92a level of tumor samples were evaluated by using the Pearson χ² test or Fisher exact test for comparison between two groups. All calculations were performed by using software JMP (version 5.1; SAS Inc, Cary, NC). In transfection experiments, differences were statistically evaluated by one-way analysis of variance followed by Tukey method or the unpaired t test. All calculations were conducted by using GraphPad Prism software system (GraphPad Software, Inc, La Jolla, CA). A P value < .05 was considered to be statistically significant.

Results

Plasma miR-92a/MVs Levels Were Significantly Increased in Xenograft-Bearing Mice with DLD-1 Cells

The miR-92a level was recently reported to be increased in the plasma from colon cancer patients [17]. We also confirmed in a colon cancer xenograft model whether miR-92a/MVs could serve as a circulating colon cancer marker by comparing its plasma level between xenograft mice and control mice. A significant increase in the plasma miR-92a/MVs level was observed in 10 of 13 mice (76.9%) compared with the levels for the 13 control mice (Figure 1A). The tumor-bearing mouse/control mouse ratios of miR-92a/MVs expression in the plasma samples were also expressed as box-and-whisker plots (Figure 1B).

Figure 1.

Expression levels of miR-92a/MVs in the plasma of DLD-1 cell xenograft-bearing mice. Ratios of the level of circulating miR-92a/MVs in the plasma of 13 tumor-bearing mice to that of 13 control mice are shown as (A) column bars and (B) box-and-whiskers plot. Data were expressed as the means ± SD. The P value was stated as follows: *P < .05, **P < .01, and ***P < .001.

Relationship between the Expression of miR-92a in Colorectal Tumors and Clinicopathologic Findings

The clinicopathologic findings and frequencies of up-regulation of miR-92a tested in a total of 94 patients with colorectal cancer (38 cases) or adenoma (56 cases) are shown in Tables 1 to 3. The expression level of miR-92a was significantly higher in the cancers than in the adenomas (P < .0001; Table 1). Significant up-regulation of miR-92 was observed in the smaller cancers (<40 mm; P = .017) and in cancers classified as clinical stages I to IV compared with those in stage 0 (P = .022; Table 2). Interestingly, the up-regulation of miR-92a was associated with the invasiveness of the cancer cells. We previously reported that good markers for discrimination between cancer and adenoma are miR-7, -21, and -34a, and that the down-regulation of miR-143 and -145 is an overall marker of colon tumors [18]. As a result of this present study, miR-92a was also shown to be a good marker for discrimination between cancer and adenoma.

miR-92a Was Specifically Upregulated in Colon Cancer Cell Lines

First, we examined the tissue distribution of miR-92a and found that the expression level of miR-92a in colon tissue was extremely low compared with that in the other tissues tested (Figure 2A). The relative expression level of miR-92a was significantly increased in colon cancer cell lines, DLD-1, COLO201, SW480, and WiDr cells compared with that in normal colon mucosal tissue (Figure 2B). These data suggest that miR-92a had a specific oncogenic function in colon cancer cells.

Figure 2.

Expression level of miR-92a examined by qRT-PCR. (A) miR-92a relative expression levels in various human normal tissues. miR-92a expression levels were significantly lower (***P < .001) in colon mucosa tissues compared with those in other tissues. (B) miR-92a relative expression levels in four human colon cancer cell lines and in corresponding normal colon tissue (***P < .001). The miR-92a expression levels were normalized by using RNU6B as an internal control.

miR-92a Targeted the 3′UTR Sequence of Dkk-3 in Colon Cancer Cells

To validate the role of up-regulation and secretion of miR-92a in colon cancers, we examined the effect of it on a potential target gene, Dkk-3. Dkk-3 had previously been shown to be a target gene of miR-92a in neuroblastomas [19]. The biologic role of DKK-3 in colon cancer still remains unclear; however, several reports have shown DKK-3 to have a tumor-suppressive function [20–22].

In this study, the expression level of the Dkk-3 gene in the four colon cancer cell lines examined was markedly downregulated compared with that in normal colon tissue (Figure 3A). The predicted binding site for miR-92a in the 3′UTR region of human Dkk-3 mRNA was confirmed on the basis of the database (TargetScan; http://targetscan.org/) and cloned into the downstream of the firefly luciferase gene in the pMIR-REPORT vector (Figure 3B). As expected, compared with those of the control, the luciferase activity of the wild-type pMIR-Dkk-3 region A was significantly inhibited after the introduction of miR-92a into the DLD-1 cells (data not shown) and WiDr cells (Figure 3C). Mutation of the Dkk-3 3′UTR-binding site in region A abolished the ability of miR-92a to regulate luciferase expression (Figure 3C). These results also demonstrated Dkk-3 to be a potential target of miR-92a in colon cancer.

Figure 3.

Validation of Dkk-3 as a target gene of miR-92a in colon cancer cells. (A) Expression level of Dkk-3 mRNA in four colon cancer cell lines. GAPDH was used as an internal control (***P < .001). (B) Predicted binding site for miR-92a in the 3′UTR region of human Dkk-3 mRNA (region A shown as a red box). Mutant-type pMIR vector was inserted with mutated seed sequence (from GCA to TAC) for miR-92a. (C) Luciferase activities after co-transfection with control or miR-92a and wild-type or mutant-type pMIR vectors having the indicated 3′UTR of Dkk-3 (***P < .001). A P value was determined for the difference in luciferase activity between the cells transfected with nonspecific control siRNA (Dharmacon) and those transfected with miR-92a.

antagomiR-92a Transfection Inhibited Cell Growth and Cell Invasion and Restored DKK-3 Expression in Colon Cancer Cells

Our results suggested that up-regulation of miR-92a in colon cancer cells contributed to the regulation of DKK-3 expression. Next, we antagonized miR-92a to confirm the role of miR-92a on Dkk-3 in colon cancer cells. As expected, transfection of DLD-1 or WiDr cells with the antagomiR-92a resulted in a significant growth inhibition and restored DKK-3 expression levels (Figure 4, A and B). The expression level of DKK3 and the growth inhibition at 60 nM antagomir-92a in DLD-1 cells were not dose dependent, which would indicate an off-target effect. Moreover, cancer cell invasion was also significantly reduced by antagomiR-92a in the Matrigel invasion assay (Figure 4C).

Figure 4.

Effects of transfection of colon cancer cells with antagomiR-92a. (A) Cell viability and (B) DKK-3 protein expression levels at 72 hours after antagomiR-92a transfection at a concentration of 40 or 60 nM in DLD-1 cells and 20 or 40 nM in WiDr cells (*P < .05, **P < .01, and ***P < .001). β-actin was used as an internal control, and densitometric values were calculated for DKK-3. (C) Cell invasion evaluated in the Matrigel invasion assay (*P < .05 and **P < .01). Representative microscopic images are presented in the upper panel of each assay graph.

Nascent miR-92a from Colon Cancer Cells Was Transferred to Endothelial Cells through MVs Traced by Using EU

To validate the role of miR-92a secretion by colon cancer cells through MVs, we traced EU-labeled miR-92a from donor DLD-1 cells to recipient HUVECs or ASF-4-1 cells. As shown in Figure 5A, the donor cell's intracellular EU-labeled miR-92a moved into MVs. EU-labeled miR-92a/MVs was then transferred into the recipient HUVECs and detected within the cells; however, EU-labeled miR-92a/MVs within the recipient ASF-4-1 cells was barely detected (C_t value > 39; Figure 5A). Furthermore, following incubation with MVs, significant increase in the intracellular level of miR-92a was observed in HUVECs compared with its endogenous level of miR-92a but not in ASF-4-1 cells (Figure 5B). These results suggested that miR-92a released into the bloodstream from colon cancer cells could tend to be transferred to the endothelial cells through MVs. However, we observed a significant growth promotion not only in HUVECs but also in ASF-4-1 cells by treatment of them with DLD-1-derived MVs (Figure 5C).

Figure 5.

Evidences of miR-92a transfer from donor DLD-1 cells to recipient HUVECs or ASF-4-1 cells. (A) Relative expression levels of EU-labeled miR-92a traced from donor cell (DLD-1)-derived MVs to recipient cells (HUVECs or ASF-4-1). Nascent RNU6B was used as an internal control. Expression levels of EU-free miR-92a and RNU6B were used as normalizers. (B) Relative expression levels of endogenous miR-92a in DLD-1, HUVECs, and ASF-4-1 cells and intracellular level of miR-92a in recipient cells following incubation with MVs (*P < .05). Expression level of RNU6B was used as internal control. (C) Cell viability of HUVECs and ASF-4-1 cells following incubation with MVs. HUVECs and ASF-4-1 cells were seeded at 0.2 x 10⁵/well in 24-well plate, and the percentage of viable cells was determined at 48 hours after exposure to DLD-1-derived MVs (*P < .05, and **P < .01). These MVs were collected from 2 x 10⁷ DLD-1 cells incubated for 3 days in MV-free medium.

miR-92a, Acting as a Suppressor of DKK-3, Increased Cell Growth and Cell Motility in HUVECs

As we showed that miR-92a secreted from colon cancer cells was delivered to the endothelial cells through MVs, we transfected HUVECs with mature miR-92a to validate the role of miR-92a in the endothelial cells. As expected, the cell growth was significantly enhanced, and DKK-3 expression was significantly downregulated at both mRNA and protein levels (Figure 6A). Gene silencing of Dkk-3 (siR-Dkk-3) also increased cell growth through the down-regulation of DKK-3 (Figure 6A). HUVEC motility and tube formation were also significantly increased by miR-92a or siR-Dkk-3 transfection, as confirmed by the results from the wound healing assay (Figure 6B) and tube formation assay (Figure 6C). These results suggested that the colon cancer cells secreted miR-92a to elicit a suitable tumor environment for angiogenesis through the down-regulation of Dkk-3.

Figure 6.

Effects of transfection of HUVECs with miR-92a or siR-Dkk-3 on their proliferation, motility, and tube formation. (A) Cell viability and relative expression of DKK-3 mRNA and protein at 72 hours after the transfection with nonspecific control miRNA, miR-92a, or siRNA for Dkk-3 (siR-Dkk-3) at 10 nM (**P < .01 and ***P < .001). (B) Cell motility evaluated by use of the wound healing assay. The wound (scratch) was made at 24 hours after the transfection with nonspecific control miRNA, miR-92a, or siR-Dkk-3 at 10 nM (pre; 0 hour). The percentage of remaining wound area was determined at 24 hours after the scratch (post; 24 hours). The cell migration activity is shown as a decrease in the percentage of the wound area remaining compared with that of the wound area at 0 hour (***P < .001; scale bar, 200 µm). (C) Tube formation activity was observed by use of the tube formation assay. HUVECs were moved into the 24-well plate for tube formation assay at 24 hours after the transfection with nonspecific control miRNA, miR-92a, or siR-Dkk-3 at 10 nM. Because of increased number of viable cells, firm and well-netted structures were constructed in miR-92a or siR-Dkk-3-transfected cells compared with that of nonspecific control. Bar, 200 µm.

Discussion

In this study, we showed that the level of circulating plasma miR-92a was significantly increased even in mice bearing human colon cancer xenografts. These findings support the possible use of miR-92a/MVs as a novel biomarker for the early detection of colon cancer. To clarify the contribution of miR-92a to colon cancer development and the secretion of this miRNA through MVs, we examined the expression level of miR-92a and its target gene in 94 clinical samples of colorectal cancers and adenomas as well as in various colon cancer cell lines. We further approached the significance of MVs for the transfer of miR-92a from colon cancer cells to other cells by using HUVECs and ASF-4-1 cells.

A significant increase in miR-92a expression was more frequently observed in colorectal cancers than in colorectal adenomas, which suggests that miR-92a could be a potential marker for discrimination between cancers and adenomas and a potential promoter for the phenotypic changes from adenoma into carcinoma. We also investigated the association of miR-92a expression with clinicopathologic features. In the colorectal cancer group, we found that miR-92a expression was related to advanced clinical stages and to the depth of invasion. Interestingly, the increased expression of miR-92a was frequently observed in the cancers smaller than 40 mm in diameter. In the adenoma group, male patients had frequently increased miR-92a expression in their tumors compared with that in those of female patients. As to the relationship between the clinicopathologic characteristics and miR-92a expression, Zhou et al. previously reported that overexpression of miR-92a is correlated with tumor size, node status, and metastasis classification (TNM) stage, lymph node and distant metastasis, and poor prognosis in colorectal cancer [23]. Our data support their results and add some new findings to them. The precise mechanisms behind the up-regulation of miR-92a in colorectal cancer still remain unclear. Previous studies on neuroblastoma demonstrated that c-Myc induces expression of the miR-17-92 cluster and that two members of this cluster, miR-19b and miR-92a, regulate Dkk-3 expression post-transcriptionally [19,24,25]. In this present study, we demonstrated that the up-regulation of miR-92a was rather a specific phenomenon of colon cancer cells and that the target gene was the same Dkk-3 in human colon cancer DLD-1 and WiDr cells. Moreover, cancer cell growth and invasion were significantly inhibited when miR-92a was antagonized and DKK-3 expression was restored. DKK-3 is a member of the DKK family of secreted Wnt antagonists. The functions of other members of this family have been well elucidated; however, the role of DKK-3 still remains controversial. Reduced or silenced Dkk-3 expression has been reported in several cancers including colon cancer [21,26,27]. Recently, it was reported that overexpression of Dkk-3 in cancer cells inhibits their motility and proliferation and induces apoptosis [20,28]. Thus, accumulating evidence suggests that Dkk-3 acts as an anti-oncogene and has functions beyond being merely a canonical Wnt inhibitor. Our data support these studies.

miR-92a secreted by DLD-1 cells through MVs was delivered to HUVECs and acted as a stimulant for cell growth, motility, and tube formation through targeting Dkk-3, the same target gene in colon cancer cells. The method used for the collection of MVs from the plasma and culture medium was established in our laboratory. MVs are a heterogeneous population of membrane-enclosed vesicles released by eukaryotic cells through direct cargo export mechanisms across the plasma membrane and not through signal peptide-dependent secretory transport pathways [29]. Much remains to be elucidated about the mechanisms by which MVs are formed and shed at the cell surface; however, the roles of MVs in the tumor micro-environment are being increasingly understood. One such role is the positive impact of tumor-derived MVs on endothelial cells and angiogenesis, which is needed for tumor survival and growth. Sphingomyelin, a major component of the MV membrane, contributes to the increased motility and proliferation of endothelial cells [30]. Matrix metalloproteinases and vascular endothelial growth factor, crucial molecules for vascularization, were shown to be packed within cancer-derived MVs and to contribute to the enhanced motility of endothelial cells [31,32]. In this study, miR-92a/MVs was barely detected in fibroblast ASF-4-1 cells, although increased cell proliferation was observed following incubation with MVs. It was suggested that besides miR-92a/MVs, other contents of MVs and/or components of MV membrane like sphingomyelin could also have contributed to the recipient cell proliferation [30]. It seemed that HUVECs willingly uptake miR-92a/MVs compared with ASF-4-1 cells; however, the recipient selection by MVs and/or MV selection by recipient cells still remained unclear.

Our data suggest that colon cancer cells upregulated intracellular miR-92a to downregulate the anti-oncogene Dkk-3 and also secreted miR-92a-containing MVs into their surrounding environment to facilitate angiogenesis. This process supplies more blood for the increased demand for nutrients and oxygen for further growth of the tumor. Further study will be needed to disclose the role of DKK-3 in colorectal cancer and also in HUVECs and the mechanisms involving secretion and selection of the MV's components.