MicroRNA-181a promotes tumor growth and liver metastasis in colorectal cancer by targeting the tumor suppressor WIF-1

To investigate the role of miRNA in CRC liver metastasis, miRNA expression was profiled in CRC tumor tissue from patients with (n = 5) or without (n = 5) liver metastases. Among the 743 miRNA probes on Mammalian miRNA arrayV2.0, 214 miRNAs were detected in CRC tissues. hsa-miR-181a was detected to be most elevated with > 2-fold between CRC with and without liver metastasis. (Additional file 1: Table S1).

The microarray results were validated with quantitative reverse transcription PCR (qRT-PCR) by confirming the expression of hsa-miR-181a in the same 10 human CRC tissues. All further analyses in this study focused on miR-181a.

To verify the microarray results of differentiated miRNA expression, we performed qRT-PCR analyses in a cohort of 137 paired CRC and normal tissues (Validation Cohort 1).

qRT-PCR analysis revealed a significant increase in miR-181a expression in tumor tissue (n = 137) compared with matched adjacent normal tissues (n = 137, fold change: 2.46, p = 0.0002). We then analyzed the relationship between miR-181a expression and liver metastasis. The relative expression ratio of miR-181a to U6 in patients with synchronous liver metastasis (n = 63) is significantly higher than those without metastasis (n = 60, fold change: 2.56, p = 0.0003). In addition, its expression in patients with metachronous liver metastasis (n = 14) is remarkably higher than those without metastasis (fold change: 1.93, p = 0.0114). These data demonstrate that miR-181a expression is elevated in CRC tumor tissues compared to surrounding normal tissue, and this elevated expression level of miR-181a correlates with liver metastasis in CRC patients (Figure 1).

Given that miR-181a is part of a family that includes miR-181a-d, we also checked the expression of three other miR-181 family members, miR-181b, miR-181c and miR-181d, to see whether they were also associated with metastases. Our results show that miR-181b-d expressions are not associated with CRC liver metastasis (Figure 1G-I). These results suggest that the association of miR-181a with CRC metastases is specific. The miRBase accession number and the mature sequence of the microRNAs studied are available in Additional file 1: Table S2.

The clinical significance of miR-181a up regulation in CRC was further investigated in Validation Cohort 1. The expression of miR-181a increased remarkably with the advanced CRC TNM stages. Elevated miR-181a expression was significantly correlated with local tumor invasion (T stage), lymph node metastases or liver metastases, but not with other clinical-pathological parameters (Table 1, Figure 1C-E). Kaplan-Meier analysis showed that high miR-181a expression (above the median value in this 137 patient cohort) was associated with poor overall survival in CRC (Figure 1F). These results suggested that the up regulation of miR-181a may play an important role in the development and progression of CRC.

Furthermore, multivariate analysis was performed to determine the prognostic value of miR-181a expression using the Cox proportional hazard model. The risk variables examined included miR-181a expression level, as well as factors known to significantly affect the outcome of CRC such as age and gender of patients, differentiation, lymph node metastasis, surgical-pathological staging, histological type and venous invasion. In the univariate analysis, high miR-181a expression level in tumors (HR 2.22; 95% CI 1.307 to 3.77; p = 0.003), TNM staging (HR 11.057; 95% CI 3.995 to 30.605; p < 0.0001), venous invasion (HR 2.462; 95% CI 1.476 to 4.108; p = 0.001), and lymph node metastasis (HR 4.27; 95% CI 2.156 to 8.454; p < 0.0001) were significantly associated with survival; while age, gender, differentiation and histological type were not (Table 2). In the final multivariate Cox regression model, high miR-181a expression in tumors was associated with a poor survival prognosis (HR 1.871; 95% CI 1.078 to 3.246; p = 0.026) independent of other clinical covariates (Table 2).

To further elucidate the functional role of miR-181a in CRC, we utilized in situ hybridization (ISH) to study the expression pattern of miR-181a in tumor and corresponding non-tumor tissue. A high level expression of miR-181a was detected in epithelial cells with elevated expression in tumor tissue compared with normal tissue. In cancerous tissue, miR-181a expression was detected only in the cytoplasm of tumor cells but not in tumor stroma, consistent with its role in tumor cells during colorectal carcinogenesis. Low levels of miR-181a expression were also seen in normal mucosa (Figure 2). To assess whether miR-181a expression is correlated with miR-181a ISH score, we assayed miR-181a expression by qPCR in 30 fresh CRC samples frozen in liquid nitrogen that were matched to the samples used for ISH. Our results showed that miR-181a expression was correlated with miR-181a ISH score (Figure 2H).

Moreover, elevated miR-181a expression in tumor tissue detected by ISH was significantly associated with poor survival (p < 0.001) (Figure 2G). In univariate analysis of Validation Cohort 2, high expression of miR-181a in tumor tissue (HR 1.514; 95% CI 1.223 to 1.874; p < 0.001), TNM staging (HR 6.509; 95% CI 4.02 to 10.54; p < 0.001), venous invasion (HR 2.375; 95% CI 1.758 to 3.21; p < 0.001), differentiation (HR 0.438; 95% CI 0.224 to 0.858; p = 0.016) and lymphnode metastasis (HR 2.916; 95% CI 2.062 to 4.125; p < 0.001) were significantly associated with survival; while age, gender, and histological type were not. In the final multivariate Cox regression model, high miR-181a expression in tumor tissue was associated with a poor survival prognosis (HR 1.382; 95% CI 1.11 to 1.72; p = 0.004) independent of other clinical covariates consistent with results derived from Validation Cohort 1 (Table 2).

To explore the potential mechanism of miR-181a in promoting CRC progression, HT29 CRC cells were infected with pri-miR-181a lentivirus to determine its effects on cell growth in vitro and in vivo. Stable expression of miR-181a was confirmed using quantitative PCR (Figure 3A). Elevated miR-181a expression results in stimulated cell growth in culture (Figure 3B). Consistent with in-vitro results, the tumor growth curves showed that the growth of lentiviral miR-181a HT29 tumors was markedly increased compared with tumors derived from parental cell line HT-29, and lentiviral empty vector-infected cells (HT29-NC). Elevated miR-181a expression in HT29 CRC cells stimulated tumor growth in NOD-SCID mice resulting in increased tumor weight 1.66 ± 0.08 g (HT29-181a), 0.72 ± 0.13 (HT29-NC), 0.81 ± 0.10 g (HT29) (p < 0.05) (Figure 3C).

The effect of miR-181a on tumor invasion and metastasis was assessed. Cell wound healing assay showed that elevated miR-181a expression significantly enhanced tumor cell mobility in HT29-181a cells compared with HT29-NC or the parental HT 29 cells (p = 0.004) (Figure 3D). Furthermore, the ability of cell migration and cell invasion through matrigel were significantly enhanced by overexpressing miR-181a in HT29 cells (HT29-181a) compared with controls (HT29-NC and HT29) as demonstrated in Boyden Chamber assays (migration 2.74-fold, p = 0.03, invasion 34.7-fold, p = 0.049 respectively) (Figure 3E). The migration and invasion through matrigel of lenti–miR-181a-infected SW480 cells were dramatically enhance by 3.75-fold (p = 0.0008) and 4.67-fold (p = 0.0003) respectively, compared with controls (Figure 3G). A wound-healing assay revealed that the spreading of lenti–miR-181a-infected SW480 cells into the wound was much faster than the control cells (p = 0.0083) (Figure 3F). These results indicate that miR-181a stimulates colorectal cancer cell spreading, migration and invasion, suggesting its potential role in cancer metastasis.

The promoting effect of miR-181a on liver metastasis was demonstrated applying an experimental metastasis model by injecting HT29 overexpressing miR-181a cells into mouse spleens. HT29 cells with elevated miR-181a expression resulted in increased number of liver metastatic nodules as compared to that induced by HT29 cells transfected with control vector or the parental HT29 cells with average liver surface metastatic nodules of 6 ± 1.6; 2.4 ± 1.3; 1.2 ± 1.3 for HT29-181a, HT29-NC, and HT29 respectively (p < 0.05) (Figure 3H).

These combined in vitro and in vivo results illustrate the role of miR-181a in promoting tumor metastasis consistent with its clinical association with liver metastases in CRC patients.

To elucidate the biological mechanisms underlying the role of miR-181a in promoting tumor cell growth and metastasis, we investigated the potential targets of miR-181a. Target prediction programs, miRanda and TargetScan, were applied to identify WIF-1 as putative miR-181a target. The 3′-UTR of WIF-1 mRNA contains a complementary site for the seed region of miR-181a (Figure 4A). To verify this finding, WIF-1 3′-UTRs and its mutant containing the putative miR-181a binding sites were cloned downstream of the luciferase open reading frame. These reporter constructs were co-transfected into HEK293T cells with either miR-SC or miR-181a mimics. Increased expression of miR-181a upon infection of miR-181a mimics, significantly suppressed luciferase expression derived from reporter constructs containing wild type WIF-1 3′-UTRs with inhibition rates 40% (p < 0.05) comparing to cells co-transfected with miR-SC. This suppressive effect was abolished when mutated 3′-UTR of WIF-1 mRNAs, in which the binding sites for miR-181a were inactivated by site-directed mutagenesis, were co-infected with miR-181a (Figure 4B). These results support WIF-1 as putative target of miR-181a.

Functional regulation of WIF-1 expression by miR-181a was further analyzed by modulating miR-181a levels via overexpression or knockdown in three CRC cell lines, HT29, SW480 and SW620. WIF-1 mRNA levels were substantially suppressed in HT29 overexpressing miR-181a and SW480 overexpressing miR-181a cells as compared with that in control cells (Figure 4C, E). Meanwhile, the protein levels of WIF-1 were also suppressed after ectopic overexpression of miR-181a in HT29 and SW480 cell lines (Figure 4F, G). On the other hand, knock down of miR-181a via RNA interference in HT29 andSW620 cells resulted in increased mRNA and protein levels of WIF-1 (Figure 4C-G).

Collectively, these data support the bioinformatics prediction of WIF-1 as direct targets of miR-181a and established a functional association.

To confirm further the tumour-promoting function of miR-181a, we infected the highly metastatic cell line SW620 with miR-181a-RNAi lentivirus and measured its effects on cell spreading, migration and invasion in vitro. Consistent with the results of over-expressing miR-181a in HT-29 cells, the inhibition of miR-181a in SW620 cells remarkably decreased the cell motility (p = 0.02) and cell invasion through matrigel (p = 0.03), as demonstrated by Boyden chamber assays (Figure 5A). A wound-healing assay revealed that the spreading of lenti–miR-181a-RNAi infected SW620 cells into the wound was much slower than the control cells (Figure 5B).

To explore the biological functions of WIF-1 in CRC cells, we determined whether inhibition of WIF-1 promoted CRC cell migration and invasion. In HT-29 cells, in vitro knockdown of WIF-1 promoted cell migration (p < 0.0001) and cell invasion (p = 0.0018) (Figure 5C). To further test this, we transfected SW620-181a-RNAi cells with siRNA for WIF-1 mRNA and found that the effect of miR-181a RNAi was partially attenuated by siRNA for WIF-1 mRNA (Figure 5D-E). These data confirmed that the prometastasis effect of miR-181a was mediated by inhibiting target genes WIF-1.

The regulation of WIF-1 mRNA expression by miR-181a was further characterized in Validation Cohort 1 of 137 primary CRC tissues using qRT-PCR. A significant inverse correlation between the levels of miR-181a and mRNA expression of WIF-1 was confirmed with Pearson correlation = −0.5291; and p < 0.0001 (Figure 6A).

The protein expression of WIF-1 was also examined in paired CRC and normal tissues by Western Blot analysis. A significant decrease in WIF-1 expression was seen in tumor tissues compared with the matched adjacent normal tissue (Figure 6B), while their miR-181a expressions were shown an inverse trend (Figure 6C). The same inverse correlation was further demonstrated when protein expression of WIF-1 was examined in Validation Cohort 2 of TMA. High levels of miR-181a in tumor cells correlated with low levels of WIF-1 protein expression. Conversely, low levels of miR-181a in adjacent normal tissue correlated with high levels of WIF-1 protein expression (Figure 6D). We also investigated the prognostic significance of WIF-1 in CRC Cohort 2, Kaplan-Meier analysis was performed illustrating that low WIF-1 expression was associated with poorer overall survival in the CRC cohort 2(p = 0.004, Figure 6E).

The cell morphology of the stable SW480 hsa-miR-181a–overexpressing cell line was significantly altered; the cells had long and thin processes, cell-to-cell contact was reduced. The cell morphology of the miR-181a-knockdown SW620 was also altered, and the cell-to-cell contaction was increased (Figure 7A).

When miR-181a was overexpressed in SW480 cells, two epithelial markers, E-cadherin and β-catenin, were suppressed; while the mesenchymal marker, vimentin, was induced. On the other hand, knockdown of miR-181a in SW620 cells induced E-cadherin and β-catenin expression and suppressed vimentin expression (Figure 7B). Consistent results were also obtained when these CRC cells expressing different levels of miR-181a were examined with immunofluorescence staining (Figure 7C). We also assessed the epithelial and mesenchymal markers in HT29 overexpressing miR-181a cells. The results are consistent with results derived from SW480 cells (Figure 7B). We examined the effect of WIF1 on EMT in HT29 cell lines. The expression levels of E-cadherin and β-catenin were decreased; while vimentin was increased in HT29-WIF-1 siRNA cells (Figure 7B).

Taken together, these results suggest that elevated miR-181a levels in CRC may be involved in inducing EMT and therefore further promoting invasion and metastasis.

Human CRC cell lines HT29, SW480 and SW620 were purchased from American Type Culture Collection and were maintained in Dulbecco’s Modified Eagle medium (Gibco, Carlsbad, CA) with 10% fetal bovine serum, 100 U/ml penicillin sodium and 100 mg/ml streptomycin sulfate in humidified 5% CO₂ at 37°C. Male NOD-SCID mice (4–6 weeks old) were purchased from Beijing HFK Bio-technology Co. Ltd. (Beijing, China). Mice were maintained in a pathogen-free facility and used in accordance with the institutional guidelines for animal care.

CRC samples were collected from surgical resection at Peking University Cancer Hospital & Institute (Beijing, China). All samples were immediately frozen in liquid nitrogen and stored at −80°C or fixed in 10% formalin for paraffin embedding. Samples collection and usage in the present study were approved by the Ethics Review Committees of Peking University Cancer Hospital & Institute.

The Testing Cohort consisted of CRC tumor specimens from patients with synchronous liver metastases (n = 5) or without (n = 5) liver metastases (follow up > 6 years). Samples were analyzed using microarray to generate differential miRNA profiling between CRC tumor tissues with and without liver metastases. Validation Cohort 1 consisting of fresh CRC and surrounding non-tumor tissue were obtained from 137 patients who underwent surgical resections from 2001 to 2006. These samples were analyzed by quantitative reverse transcription PCR (qRT-PCR). Validation Cohort 2, independent from Validation Cohort 1, consisting of CRC and surrounding non-tumor tissues were obtained from 294 patients who underwent surgical resections from 1999 to 2006. These samples were paraffin-embedded and micro-dissected to generate tissue microarrays (TMAs) for in situ hybridization (ISH) and immunohistochemstry. All samples were collected from patients prior to receiving any preoperative chemotherapy or radiotherapy, and were verified to contain at least 80% tumor cells. Cases with familial adenomatous polyposis CRC were excluded from the study. Final cut off date of follow-up was January 30, 2012. All samples have follow-up data. A summary of the clinical characteristics of these patients is shown in Additional file 1: Table S3 and S4.

Total RNA was extracted from 5 CRC tissues with and 5 without liver metastases using TRIZOL reagent (Invitrogen) according to the manufacturer’s instructions. miRNA microarray profiling was performed at CapitalBio Corporation (Beijing, China) following standard procedures detailed on http://​www.​capitalbio.​com. Fluorescein-labeled miRNA were used for hybridization on Mammalian miRNA arrayV2.0 (CapitalBio, Beijing, China) containing 743 probes in triplicate, corresponding to 576 human miRNAs (including 122 predicted miRNAs). Hybridization signals were detected and arrays were scanned with a LuxScanTM 10 K/A laser confocal scanner (CapitalBio, Beijing, China). The raw pixel intensities were extracted using the LuxScan3.0 software (CapitalBio, Beijing, China). The data were normalized between slides using the quantile normalization method proposed by Bolstad et al. [23]. The differentially expressed genes, classified as those with Fold changes above 2, were selected using the SAM software, version 2.1.

Total RNA was extracted using miRNeasy Mini Kit (Qiagen, Hilden, Germany). For miRNA quantification, the miRNAs levels were assayed using TaqMan MicroRNA assays, according to the manufacturer’s protocol (Applied Biosystems, Carlsbad, CA). The u6 small nuclear B noncoding RNA (RNU6B, Applied Biosystems, Carlsbad, CA) level was used as an internal normalization control.

For mRNA detection, total RNA (2 μg) was used in 20 μl of reverse transcriptase reaction to synthesize cDNA by using Moloney murine leukemia virus reverse transcriptase (M-MLV RT) for RT-PCR (Invitrogen, Carlsbad, CA), according to the manufacturer’s protocol. Real-time PCR was performed using the above RT products with the SYBR Green PCR Master Mix (Toyobo Co. Ltd., Osaka, Japan) on an ABI7500 PCR machine. PCR cycling conditions were: 95°C for 1 minute, 40 cycles of 95°C for 15 seconds, 60°C for 15 seconds and 72°C for 45 seconds. Data are presented as relative quantification (RQ) to U6 or GAPDH based on calculations of 2^-ΔCt where ΔCt = C_t(Target)-C_t(Reference), or as ΔCt. Fold changes were calculated by the 2^-ΔΔCt. All primers are listed in Additional file 1: Table S5.

Paraffin-embedded samples were microdissected and tissue microarrays (TMAs) were constructed in quadruplicates using a previously described method [24]. Seven TMAs were constructed to include 160 Primary CRC tumors with Synchronous Liver Metastasis (PSLM), 97 Primary CRC tumors with Non-Liver Metastasis (PNLM), 37 Primary CRC tumors with metachronous liver metastasis (PMLM), and 72 paired adjacent normal colorectal mucosa and 32 matched liver metastasis tissues.

Digoxigenin (DIG)-labeled mercury locked nucleic acid probes for human miR-181a, U6 (positive control) and scramble DNA (negative control) were purchased from TaKaRa (Dalian, China). The in situ hybridization (ISH) was performed with a modified version of the protocol for formalin-fixed paraffin-embedded tissue by Wigard Kloosterman (http://​www.​exiqon.​com/​ls/​Documents/​Scientific/​FFPE%20​in%20​situ%20​hybridization.​pdf) on human colorectal cancer tissues. The sequence and hybridization temperature of probes were showed in Additional file 1: Table S5.

The slides were then scored by two pathologists independently as negative (−), weak or focally positive (1+), or strongly positive (2+) [25]. Both pathologists were blinded to the patient’s clinical outcome.

miRNA mimics, the miRNA inhibitor and negative control miRNA oligonucleotides of has-miR-181a were from RiboBio Co. Ltd (Guangzhou, China). The small interfering RNAs (siRNAs) targeting WIF-1 were synthesized by RiboBio Co. Ltd. An unrelated sequence was used as a negative control (provided by RiboBio). The sequence was WIF1 siRNA, 5′ GAGUACUCAUAGGAUUUGA dTdT -3′ (sense). Stable transfectants overexpressing or knockdown miR-181a were generated by lentiviral transduction using a pGCsil-GFP vector (GeneChem Co., Ltd, Shanghai,China). A lentiviral vector expressing green fluorescent protein alone (LV-GFP) was used as a control. Transfection of oligonucleotides or lentivirus construction was performed using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s instructions. HT-29, SW480 and SW620 cells were infected with recombinant lentivirus-transducing units plus 8 μg/ml Polybrene (Sigma, St Louis, Missouri, USA).

Cell growth was assessed using CCK-8 (Dojindo, Tokyo, Japan) according to the manufacturer’s instructions.

Cell spreading was evaluated by a wound healing assay. The wound healing assay was performed by the CellPlayer™ 96-Well Kinetic Cell Migration assay (The Essen BioScience IncuCyte^TM Live-Cell Imaging System). Cells were seeded into 96-well plate and cultured until attached. The woundMaker™ and wounding procedure were used to create precise and reproducible wounds in all wells of the 96-well plate. The plate was placed into the IncuCyte™ ZOOM to scan every hour for migration assays. The data was analyzed by wound confluence, Wound confluence (%) represents the fractional area of the wound that is occupied by cells.

Cell motility or invasion was measured using a Boyden chamber assay with or without Matrigel. Photographs of three randomly selected fields of the fixed cells were taken and cells were counted. Experiments were repeated three times independently. These assays were carried out as previously described [26].

For the tumorigenecity study, HT29 cells, has-miR-181a-expressing cells (HT29-181a+) and empty vector cells (HT29-NC) were injected subcutaneously into the back of NOD-SCID mice (8 mice/group). Over a 4-week period, tumor formation in mice was observed by measuring the tumor volume calculated by the formula V = 0.5xLxW². Tumors were then excised and weighed.

HT29 cells, has-miR-181a-expressing cells (HT29-181a+) and empty vector cells (HT29-NC) were assayed for metastasis in 5- to 6-week-old NOD-SCID mice using the splenic injection model as described previously [27]. Cells (1 × 10⁶) were injected into the spleen subcapsular through an abdominal incision under sterile conditions. After injection, the spleen was returned gently back to abdominal cavity and homoeostasis was assured. The area was thoroughly irrigated with warm sterile water and the abdominal cavity was closed in appropriate layers. Ten weeks after surgery, mice were euthanized and the livers were obtained to determine the liver weight and hepatic metastatic foci.

All animal experiments were reviewed and approved by the Ethics Review Committee at the Peking University School of Oncology.

The 3′-UTR of WIF-1 was amplified via polymerase chain reaction (PCR) from human genomic DNA constructed and cloned into pmiR-RB-REPORT™Luciferase, downstream of the firefly luciferase gene to form pMIR-3′-UTR WIF-1 and pMIR-3′-UTR WIF-1-Mut constructs (primers are listed in Additional file 1: Table S3).

For the luciferase assay, HEK 293 T cells were seeded in 96-well plates, pMIR-3′-UTR constructs were co-transfected with miR-181a mimics or miR scramble (miR-SC) (50 nM, synthesized by Ribobio Co. Guangzhou, China) using the Lipofectamine2000 method. The activities of firefly luciferase and Renilla luciferase were measured at 24 h post-transfection using the dual-luciferase reporter assay kit (Promega) following the manufacturer’s protocol. Firefly luciferase activity was normalized to that of the Renilla luciferase for each sample. Each experimental group consisted of three wells and the experiment was repeated three times.

Total protein lysates extracted from samples were quantitated using BCA protein assay kit (Pierce, Rockford, IL). 20 μg total protein lysates extracted from samples were separated with 10% sodium dodecyl sulfate-polyacryl-amide gels and transferred to a polyvinylidene fluoridemembrane. The membrane was blocked with 3% BSA, followed by incubation with mouse monoclonal anti-human WIF-1 (Abcam, Cambridge, MA), rabbit polyclonal anti-human E-cadherin, β-catenin and vimentin (Cell Signaling Technology Inc., Danvers, MA), anti-human β-Actin antibody (Epitomics, Burlingame, CA). The membrane was then incubated with secondary horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse antibody (Jackson Immune Research Laboratories Inc., West Grove, PA), and visualized using Immobilon™ Western Chemiluminescent HRP substrate (Millipore). β-Actin was used as the loading control.

Immunohistochemical studies of WIF-1 were done on tissue microarray using the Polymer Detection System for immune-histological staining (PV9000 kit, ZSJQ-Bio, China). All images were examined by two experienced pathologists independently. Immunostaining was evaluated as described [28] previous.

SW480, SW480-NC, SW480-181a, SW620, SW620-NC, SW620-181a RNAi cells were plated in a 96-well microplate and were cultured overnight until cells were adherent. Cells were fixed and stained with anti-E-cadherin, −β-catenin and-vimentin primary antibody and DyLight549 Goat Anti-rabbit Secondary Antibody. Cells were also labeled with Hoechst 33342 Dye for nuclear staining. Expression of E-cadherin, β-catenin and vimentin was imaged using a laser scanning confocal microscopy (TCS-SP2, Leica Microsystems).