MicroRNA-130b Promotes Tumor Development and Is Associated with Poor Prognosis in Colorectal Cancer^,

Abstract

MicroRNA-130b (miR-130b) is involved in several biologic processes; its role in colorectal tumorigenesis has not been addressed so far. Herein, we demonstrate that miR-130b up-regulation exhibits clinical relevance as it is linked to advanced colorectal cancers (CRCs), poor patients' prognosis, and molecular features of enhanced epithelial-mesenchymal transition (EMT) and angiogenesis. miR-130b high-expressing cells develop large, dedifferentiated, and vascularized tumors in mouse xenografts, features that are reverted by intratumor injection of a specific antisense RNA. In contrast, injection of the corresponding mimic in mouse xenografts from miR-130b low-expressing cells increases tumor growth and angiogenic potential while reduces the epithelial hallmarks. These biologic effects are reproduced in human CRC cell lines. We identify peroxisome proliferator-activated receptor γ (PPARγ) as an miR-130b direct target in CRC in vitro and in vivo. Notably, the effects of PPARγ gain- and loss-of-function phenocopy those due to miR-130b down-regulation or up-regulation, respectively, underscoring their biologic relevance. Furthermore, we provide mechanistic evidences that most of the miR-130b-dependent effects are due to PPARγ suppression that in turn deregulates PTEN, E-cadherin, Snail, and vascular endothelial growth factor, key mediators of cell proliferation, EMT, and angiogenesis. Since higher levels of miR-130b are found in advanced tumor stages (III–IV), we propose a novel role of the miR-130b-PPARγ axis in fostering the progression toward more invasive CRCs. Detection of onco-miR-130b and its association with PPARγ may be useful as a prognostic biomarker. Its targeting in vivo should be evaluated as a novel effective therapeutic tool against CRC.

Introduction

Colorectal cancer (CRC) is one of the major leading causes of cancer-related death; the critical step in CRC progression is the ability of tumor cells to disseminate into and colonize adjacent tissues, predicting a poor outcome for these patients [1]. CRC development and progression involves mutational events in oncogenes and tumor suppressor genes often accompanied by deregulated gene expression due to epigenetic changes [1,2].

MicroRNAs (miRNAs) are emerging as a new class of regulatory molecules involved in numerous biologic processes [3,4]. miRNAs are single-stranded and highly conserved small non-coding RNAs that recognize complementary sequences in the 3′untranslated region (3′UTR) of target mRNAs, leading to the reduction of protein expression either by mRNA degradation and/or by translational repression [3]. More than 1000 human mature miRNA sequences are listed in the miRNA registry (miRBase release 16), and each of them is able to modulate multiple target mRNAs along diverse and sometimes divergent signaling pathways adding a further level of complexity in their mechanism of action [3].

miRNA deregulation is a common feature of human malignancies as they control the expression of oncogenes or tumor suppressors acting as onco-miRNAs or tumor suppressor miRNAs themselves [4]. To date, a still limited number of miRNAs has been identified as functional regulators of CRC development.

Peroxisome proliferator-activated receptor γ (PPARγ) is a ligand-activated transcription factor belonging to the nuclear receptor super-family. It plays a pivotal role in control of lipid metabolism and maintenance of energetic homeostasis [5]. More recently, PPARγ has been implicated in epithelial cell differentiation and in the antiproliferative response, acting as a tumor suppressor [6–8]. Consistently, we have shown that it is frequently downregulated in about 40% of aggressive CRCs, in line with the notion that it is a favorable independent prognostic factor [9–12]. We have also demonstrated that epigenetic mechanisms contribute to its down-regulation, although they do not fully account for the frequency detected [9,10]. Additional layers of regulation, such as an miRNA-mediated repression, ought to be considered. PPARG-mRNA has relatively short 5′UTR and 3′UTR; thus, it is not surprising that only few posttranscriptional regulators have been identified so far for this mRNA.

miR-130 has been linked to mesenchymal differentiation, immune cell function, and hypoxic response modulation [13–15]. It has also been validated as a PPARγ regulator because it suppresses the adipogenic process through the binding to two distinct highly conserved sites located in the coding sequence and 3′UTR of the corresponding mRNA [13]. The miR-130 family is formed by mature miR-130a and miR-130b that share the same seed sequence and are coded by two independent loci (miRBase Database). To date, only few reports have shown the involvement of miR-130b in tumorigenesis with controversial results. miR-130b overexpression has been associated with invasive gastric, hepatocellular, and renal carcinoma and identified, along with other miRNAs, as a risk factor in tumor invasion [16–19]. Contrasting results about its role in endometrial carcinoma have been reported [20,21].

In this study, we show that miR-130b acts as a bona fide onco-miRNA as it is upregulated in about 60% of sporadic CRCs and is associated with tumor progression and poorer patients' prognosis due to higher cell proliferation, epithelial-mesenchymal transition (EMT), and angiogenesis. Mouse xenograft models recapitulate the in vivo results, disclosing a strong angiogenic potential of this miRNA. The analysis of a series of CRC cell lines provides mechanistic insights into miR-130b activity leading to the identification of PPARγ as a functionally relevant downstream target. PPARγ repression mediates most of miR-130b effects in vivo and in vitro.

Materials and Methods

Cell Culture

Human CRC-derived cell lines DLD1, HCT116, HT29, LoVo, RKO, and SW480 were purchased from American Type Culture Collection (ATCC, Rockville, MD) and cultured as recommended. Human umbilical cord endothelial cells (HUVECs) were obtained from Lonza (Walkersville, MD) and maintained in EGM BulletKit medium.

mRNA/miRNA Isolation and Quantitative Reverse Transcription-Polymerase Chain Reaction Analysis

Total RNA was extracted from cells and tissues using TRIzol and treated with DNase I (Invitrogen, Carlsbad, CA). Small RNAs were extracted using the Qiagen miRNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. RNA purity and quantity were measured on a Jenway Genova Plus spectrophotometer (Bibby Scientific Ltd, Dunmow, United Kingdom). For quantitative mRNA analysis, we used SuperScript III Reverse Transcriptase (Invitrogen) and performed quantitative reverse transcription-polymerase chain reaction (qRT-PCR) on the 7300 Real-Time PCR System (Applied Biosystems, Monza, Italy) using SYBR Green PCR Master Mix (Invitrogen). Expression levels were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. The sequences of the specific primers are reported in Table W1.

For miRNA expression analysis, RNA was reverse transcribed using miRScript PCR System and analyzed by qRT-PCR with the miScript SYBR Green PCR Kit using the specific hsa-miR-130b miScript Primer Assays (Qiagen) according to the manufacturers' instructions. Expression levels were normalized to the average of U6-snuRNA. For CRC samples, miR-130b levels were calculated as fold change (2^-ΔΔCT) with respect to matched normal mucosa. The mean value of miR-130b expression in normal tissues was used as calibrator for all CRC cell lines analyzed.

Recombinant Plasmid Generation and Transfection

The 211-bp-long PPARG-3′UTR was PCR amplified and cloned into pTk-Renilla Luciferase plasmid (Promega, Madison, WI) to generate Ren-3′UTRwt reporter vector [22]. Site-directed mutagenesis was performed using the Site-Directed Mutagenesis XL Kit (Stratagene, La Jolla, CA) to generate the Ren-3′UTRmut reporter vector, carrying a mutated miR-130b seed sequence; the primers used for miR-130b mutagenesis are reported in Table W1. To obtain the PPARG-3′UTR vector, the wild-type or mutant 3′UTR was fused to the 3′-end of the PPARG-cDNA previously cloned into the pCDNA3 vector.

Luciferase assay was carried out on extracts from CRC cells co-transfected for 24/48 hours with the CMV-lacZ plasmid and Ren-Ctrl, Ren-3′UTRwt, or Ren-3′UTRmut reporter vectors. Where indicated, miR-130b mimic, miR-130b inhibitor, or scrambled controls were also transfected. β-Galactosidase activity was used to normalize for transfection efficiency.

PPARγ silencing was obtained by transfecting a retroviral vector carrying a short hairpin RNA (shRNA) targeting PPARG-mRNA (RHS1764-9494331), indicated as shPPARG [9]. The results obtained were further confirmed and reproduced by using an additional PPARG-shRNA (shPPARG_2, RHS1764-9695293); a scrambled shRNA was used as control (sh-Ctrl; Open Biosystem Inc, Huntsville, AL).

PPARγ overexpression was obtained by transfecting the vector containing the PPARG-cDNA without its 3′UTR by using Lipofectamine 2000 (Invitrogen); the empty vector was used as control. Functional assays and RNA or protein extractions were performed 24/48 hours later as indicated in the figure legends.

Transient Transfection of miRNA Mimics and Inhibitors

miR-130b mimic (Syn-hsa-miR-130b), miR-130b inhibitor (anti-hsa-miR-130b), or the appropriate scrambled controls (AllStars or miScript Inhibitor Negative Control) were transfected in the different CRC cell lines using HiPerFect Transfection Reagent (Qiagen), according to the manufacturer's recommendations. Functional assays and RNA and protein analyses were performed within 24/72 hours from transfection. In each experiment, the extent of miR-130b silencing/overexpression was assessed by qRT-PCR.

Western Blot and Immunohistochemical Analyses

Western blot analysis was performed on protein extracts from cell lines and tissues, using antibodies against PPARγ (1:100), vascular endothelial growth factor (VEGF; 1:200), cyclin A (1:1000), cyclin D1 (1:1000), PTEN (1:500), p21 (1:250), Snail (1:250; Santa Cruz Biotechnology, Santa Cruz, CA), Flag (1:4000), β-actin (1:10,000; Sigma-Aldrich, St Louis, MO), E-cadherin (1:500) and p27 (1:1000; BD Transduction Laboratories, San Jose, CA), and RUNX3 (1:500; Abcam, Cambridge, MA), as described [9,10,12]. Image acquisition and densitometric analyses were performed on ChemiDoc XRS (Bio-Rad, Richmond, CA).

Immunohistochemistry and hematoxylin and eosin staining on human CRC and mouse xenografted tissues were performed and evaluated as reported [9,10] using antibodies against Ki-67 (1:100; Dako, Milan, Italy), PPARγ (1:100), E-cadherin (1:200), Snail (1:100; Santa Cruz Biotechnology), Collagen IV (1:200; Novocastra, Milan, Italy), CD31 (1:100), and VEGF (1:500; R&D Systems, Minneapolis, MN). Image acquisition and analysis were performed on DM100 Leica Photosystem 40.106.206 (Leica, Milan, Italy).

TUNEL Assay

To evaluate the apoptotic rate on xenograft tissues, paraffin-embedded tumor masses were analyzed by the DeadEnd Fluorometric terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling test (TUNEL) System (Promega) according to the manufacturer's instructions. Image acquisition and data analyses were performed on DM100 Leica Photosystem 40.106.206 (Leica).

Proliferation, Migration, and Cell Morphology Assays

To assess cell proliferation, the microculture tetrazolium test (MTT) was performed (Sigma-Aldrich) 48 hours after transfection with miR-130b mimic, miR-130b inhibitor, or scrambled controls [10].

To perform the plate colony-forming assay, 24 hours after transfection, cells were trypsinized and seeded in six-well plates (5 x 10³ cells); after 7 days, cells were fixed, stained, and photographed under microscope. The area of single colonies was then measured by ImageJ software (National Institutes of Health Image, http://rsbweb.nih.gov/nih-image) [10].

Cell cycle analysis was performed 3 days after transfection on both attached and floating cells using the Cell Cycle Test (BD Biosciences, San Jose, CA). Propidium iodide-stained cells (>10,000 events) were analyzed on FACSVerse Flow Cytometer (BD Biosciences).

Cell motility was evaluated by the wound-healing assay and trans-well migration test; cells were incubated for 24 hours in serum-free medium before the assays. The edges of the initial scratch are indicated with a dotted line, and the wound closure values reported in Figure 3E are referred to this initial position. Hematoxylin and eosin staining was used to reveal morphologic changes, as reported [10].

Figure 3.

miR-130b promotes cell proliferation and motility. (A) The bar plot reports qRT-PCR analysis of miR-130b in the indicated cell lines with respect to the expression average detected in normal tissues (normal mucosa; n = 80). (B) Graphical representation of the MTT assay performed 48 hours after transfection of the indicated cell lines with the miR-130b mimic (miR-130b), miR-130b inhibitor (anti-miR-130b), or appropriate scrambled controls (miR-Ctrl or anti-miR-Ctrl). (C) Cell cycle analysis performed in HT29, HCT116, and RKO cells treated as in B; the percentage of cells in each phase is reported together with the results of FACS acquisitions. (D) Immunoblot of proliferation markers in the cell lines transfected as in B; relative protein fold change, obtained by densitometric analysis and normalization to β-actin, is reported below the corresponding band; *P ≤ .05. (E) Representative images of wound-healing assay after crystal violet staining (upper panel); vertical lines represent the initial wound edges. Graphical representation of relative wound closure (residual wound area) assessed 24 hours after cell transfection (lower panel). Dotted lines highlight the wound areas right after the scratch (0 hour), as controls. All data are represented as means ± SD from at least three independent experiments performed in triplicates; n.s., not significant; *P ≤ .05; **P ≤ .01.

In Vitro Angiogenesis Assay

To perform tube formation assay, HUVECs were seeded on Matrigel-coated plates (BD Biosciences) in a 1:1 mixture of endothelial basal medium (Lonza) and conditioned medium (CM) from CRC cells transfected as reported in figures. Where indicated, anti- VEGF antibody bevacizumab (Avastin; Genentech/Roche, Mountain View, CA) was added to the CM at the concentration of 250 µg/ml. Serum-free medium was used as control. After 18 hours, tubular structures were photographed with a phase-contrast microscope and tube number and/or length measured using ImageJ Application software (NIH).

Animal Experiments

For xenograft generation, 20 x 10⁶ CRC-derived cells (HCT116 or HT29) were subcutaneously transplanted into the flank of 20 female athymic nude mice (6–8 weeks old; Charles River, Lecco, Italy). Mice were maintained according to United Kingdom Co-ordinating Committee on Cancer Research (UKCCCR) guidelines, and tumor volumes, calculated as (tumor length x width²)/2, were monitored twice a week by caliper measurement [23,24]. Two weeks after transplantation, when tumors reached the volume of 200 mm³, mice were grouped (N = 5/group) and intratumorally injected every 7 days for four times with anti-miR-130b (4 ng/mm³) for HCT116 or with miR-130b mimic (2 ng/mm³) for HT29 xenografts. In both cases, the appropriate scrambled RNA was used (indicated as anti-miR-Ctrl and miR-Ctrl, respectively). At day 36, tumor masses were measured, excised, and further analyzed; qRT-PCR was performed on RNA from xenografts to establish the efficiency of miR-130b inhibition/overexpression. This experiment was carried out in duplicate. No adverse or toxic effects were observed. All animal experiments were reviewed and approved by the Ethics Commission at Menarini Ricerche, according to the guidelines of the European Directive (2010/63/UE).

Clinical Samples

Paraffin-embedded and liquid nitrogen-frozen specimens from 80 patients with primary sporadic CRCs were included in this study. Each sample was matched with the adjacent normal mucosa removed during the same surgery. Patients' familial history and tumor classification have been previously reported [9,12]. Patients were followed up for a median of 89.79 months or until death. All patients gave informed consent for sample collection, and study protocols were in accordance with the ethical guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the Fatebenefratelli Hospital.

Independent Data Set Analysis

To investigate miR-130b expression in CRC tissues, we first analyzed a publicly available data set (GEO record GSE35602). It counts on 59 RNA samples separately extracted from stroma and epithelium of 13 CRC tissues and four normal tissues [25]. Gene expression analysis was performed on Agilent-014850 Whole Human Genome Microarray 4x44K (Agilent Technologies, Cernusco S/N, Italy); for miRNA analysis, Agilent-019118 Human miRNA Microarray 2.0 G4470B was used. Robust multichip average normalization was performed using GeneSpring 11.5 (Agilent Technologies) [26]. The data from this latter data set were used to identify differentially expressed miRNAs having a fold change ≥ 2 and P < .05, as determined by Welch t test statistical analysis.

To recognize the molecular pathways potentially affected by miR-130b, we used TargetScan to generate a list of putative target genes; these latter were used as input for the pathway enrichment analyses performed by different tools: the licensed Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA) and the publicly available GeneCoDis3 [which provides data from both Kyoto Encyclopedia of Genes and Genomes (KEGG) and Panther resources] [27].

To confirm the frequency data of miR-130b deregulation obtained in our cohort of CRCs, we surveyed a second data set (GEO record GSE28364) that reports quantitative miRNA expression data from 40 CRCs (stages I-IV) and matched normal tissues, profiled by qRT-PCR using TaqMan Array Human MicroRNA A + B Cards v2.0 (Applied Biosystems). miR-130b expression in tumor tissues was calibrated to the levels detected in normal mucosa [28].

Statistics

All statistical analyses were made using Statistical Package for Social Science (SPSS; version 15.0) for Windows (SPSS Inc, Chicago, IL) and R/Bioconductor. Association between miRNA expression and tumor stage was assessed using the Fisher exact test or Pearson χ² test (where indicated). The Kaplan-Meier method was used to estimate survival; log-rank test was used to test differences between the survival curves. Data are reported as means ± SD, and mean values were compared using the Student's t test or Mann-Whitney test. Results were considered statistically significant when P ≥ .05 was obtained.

Results

miR-130b Is Significantly Upregulated in CRC and Correlates with Poor Patients' Prognosis

To identify the miRNAs deregulated in CRC, we analyzed a publicly available data set for miRNA expression by using bioinformatic tools (GEO record GSE35602). This preliminary survey revealed the deregulation of several miRNAs already implicated in CRC pathogenesis such as miR-17, miR-92a, miR-21, and miR-23a [29,30]. We focused our attention on miR-130b, as, to our knowledge, it has not been investigated in CRC (Figure 1A).

Figure 1.

miR-130b up-regulation in CRC samples associates with indexes of tumor aggressiveness and shorter patients' survival. (A) Bio-informatic analysis conducted on data from GEO record GSE35602. Volcano plot shows differentially expressed miRNAs (red squares; fold change > 2, P ≤ .05); miR-130b (green square) is arrowed. Green lines indicate the fold change and P value thresholds. (B) The box plot depicts miR-130b levels assessed by qRT-PCR in the normal mucosa (NM) and in our series of 80 CRC samples classified according tumor stage (stage I, n = 4; stage II, n = 44; stage III, n = 24; stage IV, n = 8). *P ≤ .05; **P ≤ .01. (C) Kaplan-Meier survival analysis according to miR-130b levels. The overall survival of miR-130b high-expressing group (n = 47) was significantly lower than that of the low-expressing group (n = 33; log-rank test; P < .001). (D) Cancer-related death events and median survival time of the same patients are tabulated (two-sided Fisher exact test; P = .02). (E) Representative Ki-67, E-cadherin, and VEGF immunostaining of miR-130b low- or high-expressing tumors; the relationship between miR-130b and the analyzed markers is summarized in the lower table (P values were calculated by two-sided Fisher exact test). (F) Box plots report miR-130b expression average in tumor tissues stratified according to Ki-67 (all stages; n = 80), VEGF (all stages; n = 80), and E-cadherin (stages III and IV; n = 32) staining; P values were calculated by paired t test.

To gain insights into the role of miR-130b in colorectal tumorigenesis, we assessed its expression in our series of primary sporadic CRCs (n = 80) and matched normal mucosa by qRT-PCR: miR-130b was significantly upregulated in 58.3% of the cases (Table 1), in agreement with the data obtained by the in silico analysis of an additional independent data set (n = 40; GEO record GSE28364; data not shown).

Table 1.

Association between miR-130b Expression and Patients' Clinicopathologic Features.

Variable/No. of Cases (% of Cases)	n	miR-130b Expression*		P Value
		Low	High
Overall	80	33 (41.7)	47 (58.3)
Age				.805
≤60	24	9 (37.5)	15 (62.5)
>60	56	24 (42.9)	32 (57.1)
Gender				.488
Male	48	18 (38.9)	30 (61.1)
Female	32	15 (46.9)	17 (53.1)
Tumor localization				.005†
Dx	33	20 (60.6)	13 (39.4)
Sx	47	13 (28.6)	34 (71.4)
Tumor grade				.034†
G1/G2	60	21 (33.3)	40 (66.7)
G3/G4	20	12 (65.0)	7 (35.0)
Tumor invasion				.347
T1/T2	11	6 (54.5)	5 (45.5)
T3/T4	69	27 (39.1)	42 (60.9)
Lymph node metastases				.458
Neg	56	25 (44.6)	31 (55.4)
Pos	24	8 (33.3)	16 (66.7)
Distant metastasis				.005†
Neg	63	31 (49.0)	32 (51.0)
Pos	17	2 (11.7)	15 (88.2)
TNM stage				.168
I/II	48	23 (47.2)	25 (52.8)
III/IV	32	10 (31.2)	22 (68.8)

^*miR-130b expression is evaluated in tumors and referred to the matched normal mucosa; it is considered low for samples with 2^-ΔΔCT ≤ 1 and high for samples with 2^-ΔΔCT > 1; P values were calculated by two-sided Fisher exact test.

^†Refers to statistically significant P values (i.e., tumor location .005, tumor grade .034 and distant metastasis .005).

miR-130b expression levels were higher in advanced tumor stages III and IV than tumor stages I/II and normal mucosa (P < .05), as illustrated in the box plot (Figure 1B). We then subdivided the tumors in two groups according to the miR-130b levels (high and low) and associated with patients' clinicopathologic features. miR-130b high levels were strictly correlated with left-side localization (P = .005) and distant metastasis (P = .005; Table 1). A Kaplan-Meier survival analysis revealed that miR-130b up-regulation was associated with an increased likelihood of death from disease (P < .001) and with a lower median survival (P = .020; Figure 1, C and D). A multivariate analysis supported miR-130b as an adverse prognostic factor regardless of tumor stage (data not shown). To identify the putative CRC-associated pathways modulated by miR-130b, we queried the IPA and GeneCoDis3 algorithms [27]. Genes affecting several cancer-related pathways, such as cell proliferation, EMT, and angiogenesis, were significantly enriched (Figure W1). Taking into account these predictions, our clinical samples were tested for Ki-67, E-cadherin, and VEGF, as markers of proliferation, EMT, and angiogenesis, respectively [31–37]. miR-130b was directly correlated with Ki-67 (P = .003) and VEGF (P = .0001) immunopositivity, revealing a strong link with the proliferative and blood vessel recruitment potential (Figure 1E). This correlation held true even clustering the samples for Ki-67 and VEGF staining; high proliferative and angiogenic-prone tumors expressed higher levels of miR-130b (P = .037 and P = .020, respectively; Figure 1F). In advanced tumor stages (III and IV), miR-130b up-regulation was also associated with reduced expression of E-cadherin (P = .009; Figure 1F).

Taken together, our results provide evidence that miR-130b is associated with CRC progression and poorer patients' prognosis acting as a potential oncogenic miRNA.

miR-130b Modulation Alters the Tumorigenic Potential of Mouse Xenografts

We evaluated the miR-130b oncogenic role in mouse model systems. On a preliminary survey, HCT116 and HT29 were chosen as representative of miR-130b-overexpressing or miR-130b-downexpressing cells, respectively, and subcutaneously transplanted into nude mice. After 2 weeks, miR-130b inhibitors or scrambled controls were injected in HCT116-derived tumors every 7 days for four times (Figure 2A). In the case of HT29-derived tumors, miR-130b mimics or scrambled controls were injected, following the same protocol. At day 36 from transplantation, tumor masses were measured, excised, and analyzed. The specificity and efficacy of the miR-130b inhibition/overexpression was verified by qRT-PCR (Figures 2B and W2A).

Figure 2.

miR-130b acts as oncogenic miRNA in the mouse xenograft model. (A) Timeline of miRNA delivery experiments. At day 13, miR-130b inhibitor (anti-miR-130b; n = 5) or scrambled control (anti-miR-Ctrl; n = 5) was intratumorally injected every 7 days for four cycles. (B) qRT-PCR validation of miR-130b knockdown in HCT116-derived tumors treated with anti-miR-130b. (C) Representative photographs of tumors excised from each experimental group (upper panel). The graph in the lower panel reports tumor weight (g) at the end of the treatment. (D) Representative photomicrographs of hematoxylin and eosin, Ki-67, and TUNEL assay performed on paraffin-embedded tissues from the two experimental conditions; arrows point to remarkable features of the images. Scale bars, 200 µm. (E) Proliferative and apoptotic indexes evaluated according to total number of Ki-67 or fluorescein-2′-deoxyuridine, 5′-triphosphate (dUTP)-positive cells in at least 10 low-power fields. (F) Immunohistochemical analysis of E-cadherin, Snail, CD31, type IV collagen, and VEGF performed on tumor tissues injected with anti-miR-Ctrl or anti-miR-130b. Scale bars, 200 µm. (G) Western blot analysis of E-cadherin, VEGF, and β-actin in tumors treated as in A. Protein fold changes with respect to controls are reported below the corresponding band (Student's t test; *P ≤ .05). (H) qRT-PCR assessment of the indicated markers; the mean values detected in controls were used as calibrators and highlighted with dotted red lines (P values reported were determined by Student's t test). Data are represented as means ± SD from at least two independently treated tumors. *P ≤ .05; **P ≤ .01.

miR-130b inhibition reduced the size of HCT116-derived tumors by about 30%. Hematoxylin and eosin staining, Ki-67 immunopositivity, and TUNEL assay documented the reduced proliferation rate and the increased cell death (Figure 2, C–E). Moreover, miR-130b inhibition led to increase of E-cadherin and reduction of Snail, an early inducer of the EMT process [36,37] (Figure 2, F and G); these observations were further confirmed by qRT-PCR analysis of additional epithelial and mesenchymal markers (Figure 2H).

Macroscopically, the surface of the tumors generated by miR-130b high-expressing cells appeared more vascularized than the anti-miR-130b-injected ones, indicating that miR-130b has a pro-angiogenic effect, as already suggested by VEGF assessment in our clinical samples (Figure 2C; see also Figure 1, E and F). That miR-130b was associated with angiogenesis was proved by the reduced immunopositivity for the tissue vasculature indexes, CD31 and type IV collagen, and VEGF following anti-miR-130b injection (Figure 2F) [34,35]. The lower expression of the latter marker, along with its receptor VEGFR2 and hypoxia-inducible factor 1α (HIF1α), a master regulator of angiogenesis, was confirmed by qRT-PCR (Figure 2H), supporting a pro-angiogenic function.

Interestingly, miR-130b mimic injection into mouse xenografts generated by miR-130b low-expressing HT29 cells resulted into larger tumors than controls and enhanced cell proliferation, EMT, and angiogenesis, as documented by the indicated markers (Figure W2, B–F). These data demonstrate that miR-130b acts as a bona fide onco-miRNA promoting tumor growth through induction of EMT and angiogenesis.

miR-130b Regulates Cell Proliferation and Motility In Vitro

To investigate the pathways modulated by miR-130b, we examined its expression in six different CRC cell lines by qRT-PCR. miR-130b mean levels in normal colonic mucosa were considered as calibrator; by applying this criterion, miR-130b expression in HT29 and SW480 cells was comparable to that observed in the normal mucosa; HCT116, LoVo, RKO, and DLD1 cells, instead, displayed higher levels (Figure 3A). Given that miR-130b was positively associated with tumor growth both in mouse xenografts and clinical samples, we sought to further investigate its influence on cell proliferation by transfecting the corresponding miR-130b mimic or antisense RNA and carrying out a series of functional assays. In each experiment, qRT-PCR analyses were performed to assess the specificity, the extent, and the duration of miR-130b overexpression/inhibition (data not shown). Effects due to the transfection procedures employed were ruled out by using the appropriate scrambled controls: no differences were observed between mock- and scrambled-transfected cells.

miR-130b forced expression resulted in a higher proliferation rate in HT29 and SW480 cells as assessed by the MTT test and plate colony-forming assay (Figures 3B and W3A). Interestingly, FACS analysis revealed that exogenous miR-130b promotes cell cycle progression, as documented by the significant enrichment of the cell population in the S phase (Figure 3C). miR-130b inhibition in the same cells (HT29 and SW480) produced no significant changes in key mediators of the morphologic and growth characteristics investigated (data not shown). miR-130b inhibition in HCT116, LoVo, RKO, and DLD1 cells, in contrast, reduced cell growth, produced smaller plate colonies, and increased the population of cells in G₀/G₁ and/or G₂/M phases (Figures 3, B and C, and W3A). Consistently, modulation of miR-130b levels resulted in significant changes in the expression of proliferation markers, such as p21, p27, PTEN, cyclin A, and cyclin D1, both at the mRNA and protein levels (Figures 3D and W3B). The role of miR-130b in cell motility was assessed by the wound-healing assay: miR-130b overexpression enhanced cell motility by about 30%, while its inhibition impaired wound closure by about 40%, with respect to control cells (Figure 3E). Finally, the transwell assay confirmed the effects of miR-130b on promoting cell migration and invasion (Figure W3C).

miR-130b Affects EMT and Angiogenesis In Vitro

EMT implicates loss of epithelial markers, concomitant acquisition of mesenchymal ones, and changes in cell morphology. The relationship between miR-130b and EMT observed in vivo was examined in vitro to elucidate the role that miR-130b plays in this process. We first correlated endogenous miR-130b levels with the cell phenotype: miR-130b low-expressing cells (HT29 and SW480) exhibited epithelial morphologic features and expressed the corresponding markers. These were progressively lost along with the acquisition of the mesenchymal ones in miR-130b high-expressing cells (RKO and DLD1; Figure W3D). Forced expression of miR-130b in HT29 and SW480 cells led to the acquisition of a fibroblast-like shape and to a scattered growth; accordingly, reduction of E-cadherin, zonula occludens 1, and cytokeratin 20, distinctive epithelial markers, along with up-regulation of Snail and zinc finger E-box-binding homeobox 1 (ZEB1), two pivotal players of the EMT process, was observed (Figure 4, A–C). Conversely, transfection of the anti-miR-130b in HCT116, DLD1, LoVo, and RKO cells restored an epithelial-like shape and reduced the mesenchymal markers with a concomitant increase of the epithelial ones (Figure 4, A–C). miR-130b, then, acts as an efficient inducer of EMT in vivo (see above) and in vitro, likely through up-regulation of Snail and ZEB1. These data are at variance with those reported in endometrial cancer where miR-130b overexpression and reduced levels of its target ZEB1 are associated with up-regulation of E-cadherin and attenuation of the EMT phenotype [20]. In the same context, miR-130b has been shown to target DICER1 mRNA and activate the EMT program [21].We then checked DICER1 mRNA in our CRC cell lines and found no association with miR-130b levels and cell aggressiveness. In addition, miR-130b modulation did not affect DICER1 mRNA, suggesting that miR-130b biologic effects appear to be independent from DICER1 (Figure W4A). Consistently, we checked miR-200c as a direct target of DICER1 and a marker of epithelial identity [38,39]. In poorly differentiated RKO and DLD1 cells, miR-200c was barely expressed with respect to well-differentiated HT29 and SW480 cells, in agreement with the epithelial phenotype. Notably, high levels of miR-200c were detected in HCT116 and LoVo cells that, in spite of this, displayed a hybrid phenotype, i.e., concomitant expression of mesenchymal and epithelial markers, these latter likely due to the elevated miR-130b levels (Figure W5A). Collectively, these data suggest that miR-130b directly correlates with ZEB1 changes and the DICER1/miR-130b axis plays only a marginal role in CRC progression.

Figure 4.

miR-130b expression is linked to the EMT-like phenotype and angiogenesis. (A) Morphologic features of cells transfected with scrambled controls, miR-130b mimic, and miR-130b inhibitor were evaluated after hematoxylin and eosin staining; the arrows point out some remarkable phenotypic characteristics (x40 and x100 magnifications). (B) qRT-PCR analysis of some EMT-associated genes in the CRC cell lines transfected as reported. mRNA levels detected in scrambled-transfected cells, used as calibrators, are indicated as red lines. (C) Western blot analysis for the indicated markers performed in HT29, HCT116, and RKO cells transfected as in A. Densitometric analysis is reported below the bands; *P ≤ .05. (D) Phase-contrast images of tube-like structures produced by HUVECs after incubation with collected CM from the same set of samples. Where indicated, bevacizumab (250 µg/ml) was added to the CM. (E) The histogram reports the results of the tube formation assay as mean number of tubes in at least 10 low-power fields; serum-free medium was used as negative control. Data are represented as means ± SD of at least three independent experiments performed in triplicate. *P ≤ .05; **P ≤ .01.

The pro-angiogenic potential of miR-130b, suggested by both clinical samples and in vivo experiments, was validated in vitro. We first assessed VEGF levels following miR-130b overexpression and/or silencing: miR-130b mimic enhanced VEGF expression in HT29 and SW480 cells at the mRNA and protein levels; conversely, miR-130b inhibition reduced VEGF production in RKO, DLD1, HCT116, and LoVo cells (Figures 4C and W5B). Furthermore, CM from HT29 and SW480, transfected with miR-130b mimic, increased HUVEC capillary tube formation with respect to media from scrambled-transfected cells. On the contrary, miR-130b inhibition in HCT116, LoVo, DLD1, and RKO cells reduced their ability to induce endothelial tube assembly (Figure 4D). Addition of an anti-VEGF antibody (bevacizumab) to the media impaired tube formation, confirming that miR-130b induces tumor angiogenesis through VEGF [32,34,35].

To investigate the molecular mechanisms through which miR-130b exerts its effects, we screened its putative target genes by using bioinformatic tools. Among them, we choose PPARγ because 1) PPARG-mRNA is an miR-130b target and contains two distinct seed motifs located in the coding sequence and 3′UTR region, respectively (Figure 5A) [13]; 2) its down-regulation is associated with aggressive CRCs and worse prognosis [9–12]; 3) miR-130b affects pathways partially overlapping those regulated by PPARγ, specifically those involving cell proliferation and EMT [40–43]. Additionally, mouse xenografts, obtained by transplanting HT29 cells stably expressing an shPPARG, display an enhanced tumorigenic and angiogenic potential, a more dedifferentiated histology, and reduced epithelial markers than HT29 cell-derived tumors (data not shown).

Figure 5.

miR-130b negatively regulates PPARγ in vivo and in vitro. (A) Schematic representation of human PPARG-mRNA: two specific highly conserved miR-130b binding sites are located in the coding region and in 3′UTR. (B) The graph illustrates the inverse correlation between miR-130b and PPARγ in the CRC samples; the number of patients of each group is reported (P = .005; Fisher exact test). (C) Average fold change of miR-130b in PPARγ-low (n = 49) versus PPARγ-high (n = 31) CRC samples (P = .0018; paired t test). (D) PPARγ protein levels assessed in HCT116- and HT29-derived xenografts injected as depicted. Scale bars, 200 µm. (E) Color map of miR-130b and PPARγ expression levels in the CRC cell lines. Pearson correlation coefficient (R) and P value are indicated below the graph. (F) Western blot analysis of PPARγ in cells transfected with miR-130b mimic, miR-130b inhibitor, or scrambled controls; the histogram reports the protein fold change with respect to scrambled-transfected cells. (G) Immunoblot detection of Flag-epitope in HCT116 co-transfected with the anti-miR-130b or scrambled miRNA and the expression vectors +3′UTR or -3′UTR containing, respectively, the Flag-tagged PPARG-cDNA fused or not with its 3′UTR. Transfection efficiency was evaluated by using CMV-LacZ expression vector. Data are represented as means ± SD of at least two independent experiments. *P ≤ .05; **P ≤ .01.

We first correlated PPARγ and miR-130b expression in our clinical samples (n = 80): low PPARγ levels were observed in about 75% of miR-130b-overexpressing tumors (35 of 47 cases; P = .005; Figure 5B). Such an inverse correlation was preserved even clustering tumors according to PPARγ levels: PPARγ low-expressing tumors (n = 49) had higher levels of miR-130b than PPARγ high-expressing ones (n = 31; P = .018; Figure 5C). This negative association was confirmed in the mouse xenografts described above. In this context, miR-130b knockdown increased PPARγ protein levels that were, instead, drastically reduced by miR-130b mimic injection (Figure 5D). We then investigated the miR-130b-dependent PPARγ regulation in vitro. A Pearson correlation test showed the existence of a significant inverse relation between the miR-130b expression profile and PPARγ levels (P < .05, R = -0.78; Figures 5E and W5C). Modulation of miR-130b expression through mimic or anti-sense RNAs induced remarkable and inverse variations of PPARγ (Figure 5F). Interestingly, anti-miR-130b increased PPARγ levels also in RKO and DLD1 cells that have a fully methylated PPARG promoter, revealing a posttranscriptional modulation that contributes to gene silencing, in addition to the epigenetic regulation already reported (Figure 5F) [9,10]. The specificity of miR-130b regulation was verified by analyzing RUNX3, another validated target of this miRNA [16]. Introduction of an miR-130b antagomir induced RUNX3 up-regulation at the mRNA and protein levels only in the CRC cell lines that express this gene; in other cell lines, in fact, RUNX3 is epigenetically silenced (Figure W4B) [44]. That miR-130b directly regulates PPARG was demonstrated by transfecting the plasmid carrying the Renilla luciferase reporter gene: miR-130b mimic strongly reduced, whereas miR-130b antagomir rescued luciferase activity only in the presence of the PPARG-3′UTR. Mutagenesis of the miR-130b seed recognition motif abrogated the effects, definitively showing that it is an miR-130b target (Figure W5, D and E). We also transfected an expression vector carrying the PPARG full-length cDNA, fused or not to the 3′UTR. In HCT116 cells, the exogenous PPARγ protein synthesized from a transfected wild-type 3′UTR-containing vector was detected at low levels, consistent with the high miR-130b expression. An miR-130b inhibitor rescued the exogenous protein by about 40%; of note, a 20% increase of the protein was detected even in the absence of the 3′UTR, revealing the contribution of the coding region miR-130b binding site to PPARG-mRNA down-regulation (Figure 5G). These experiments demonstrate that PPARG-mRNA is a direct miR-130b target and the two identified seed binding sites equally contribute to the overall post-transcriptional regulation of the gene.

PPARγ Is a Functionally Relevant Downstream Target of miR-130b

To investigate in depth the role that PPARγ plays in mediating the miR-130b effects, we transiently silenced and/or overexpressed PPARγ in the proper cell lines. The off-target effects were ruled out by using two different shRNAs targeting PPARG together with scrambled controls, as previously reported [9,10]. PPARG knockdown induced cell proliferation by downregulating p21 and PTEN, as demonstrated by us and others [9,10,40,41]. Furthermore, PPARγ silencing downregulated E-cadherin, induced key mesenchymal markers such as Snail, vimentin, and N-cadherin, causing loss of the epithelial phenotype, and increased motility and acquisition of mesenchymal-like morphologic characteristics (Figure W6A) [9,10]. These features are typical of the EMT process and less differentiated tumors [9,10,42,43]. Thus, in colonocytes, PPARγ influences proliferation and maintenance of epithelial cell features. We also provide the first evidence that in CRC PPARγ inhibits VEGF expression and angiogenesis (Figure W6A).

Altogether, our data support the notion that PPARγ and miR-130b affect overlapping pathways and are functionally linked with opposite results on cell functions. To provide mechanistic insights into the specific contribution of PPARγ silencing in miR-130b-induced effects, we simultaneously inhibited PPARγ and miR-130b in HCT116 cells. PPARG knockdown greatly impaired the anti-proliferative properties of the anti-miR-130b, as documented by the plate colony-forming assay, cell cycle test, and assessment of proliferation markers (Figures 6, A–C, and W6B). Similarly, HCT116 ability to migrate and induce endothelial tube formation was rescued (Figure 6, D and E). Assessment of E-cadherin, Snail, and VEGF clearly showed that PPARG silencing counteracted most of the effects due to miR-130b inhibition disclosing its crucial role in these important pathways (Figures 6F and W6B). Similar results were obtained in other CRC cell lines under the same experimental conditions. Consistent with these data, the combined exposure to an miR-130b antagomir and a specific PPARγ antagonist (GW9662) resulted in even more pronounced effects (data not shown). Altogether, these data suggest that in CRC the multiple functions exerted by miR-130b on cell proliferation, invasion, EMT, and angiogenesis are in large part dependent on its repressive effect on PPARγ.

Figure 6.

PPARγ is a functionally relevant downstream target of miR-130b. (A) Plate colony-forming assay performed in HCT116 transfected with anti-miR-130b alone or in combination with shPPARG and stained with crystal violet (left panel); the histogram reports colony areas (right panel). (B) FACS analysis of cell cycle performed on cells treated as in A. (C) Western blot analysis of PPARγ and some proliferation markers; fold changes are reported below each band. (D) Number of invading cells measured in at least five fields of the transwell migration assay. (E) Tube-like structures produced by HUVECs after incubation with collected CM from the same set of samples; the average of tube number and length is reported in the histograms. (F) Immunoblot analysis of the indicated EMT and angiogenic markers. In all experiments, cells were co-transfected with scrambled shRNA and/or scrambled miRNA as controls. Data are represented as means ± SD of at least four independent experiments conducted in triplicates. *P ≤ .05; **P ≤ .01. (G) A schematic and simplified representation of the cross talks between PPARG and miR-130b in CRC progression. miR-130b up-regulation leads to PPARγ suppression modulating proliferation, EMT, invasion, and angiogenesis. These effects could explain the worse prognosis of CRC patients exhibiting high miR-130b levels. The symbol “#” indicates that PPARγ endogenous levels influence target genes even in the absence of a specific ligand; “?” indicates processes and/or pathways that require further investigations.

Discussion

A complete understanding of the molecular mechanisms underlying tumor initiation and progression is essential for novel prognostic and therapeutic approaches aimed at improving the outcome of patients with cancer. Over the last years, miRNAs are emerging as a new class of gene regulators involved in different malignancies [4]. Here, we provide the first evidence that miR-130b plays a pivotal role in colon tumorigenesis. It is, in fact, frequently upregulated in CRC, is expressed at higher levels in advanced tumor stages (III and IV), and is strongly linked to left-side localization and distant metastasis. Consistently, miR-130b overexpression is significantly associated with poor patients' survival, suggesting that it is an adverse prognostic factor independent of the tumor stage. The positive relationship of miR-130b levels with proliferation, dedifferentiation, and angiogenesis supports the notion that it acts as an onco-miRNA.

The characteristics observed in patients with CRC are recapitulated in mouse xenograft models. miR-130b-overexpressing cells develop large, dedifferentiated, and vascularized tumors, features that are reverted by intratumor injections of a specific antisense RNA. An opposite behavior is observed with miR-130b low-expressing cells and injection of the corresponding mimic RNA.

These findings and the analysis of CRC cell lines demonstrate that miR-130b is not only linked to cell proliferation but promotes EMT by activating essential effectors of the process such as Snail and ZEB1 and repressing E-cadherin. Moreover, miR-130b is directly related to VEGF and neovessel formation, indicating that it stimulates angiogenesis. miR-130b appears, then, to influence the transition toward more invasive CRCs.

So far, miR-130b has been shown to be involved only in aggressive gastric, hepatocellular, and renal carcinomas [16–19]. Conversely, it is associated with more differentiated and less aggressive endometrial carcinomas, although opposite results have recently been reported [20,21]. These conflicting data suggest that miR-130b may have a dual function depending on the cell context and cancer type. In the colonic context, our data suggest that its up-regulation is associated with more aggressive tumors and a poorer outcome.

The involvement of miR-130b in CRC-related angiogenesis and EMT shown here is in agreement with data obtained in other cell types. In hypoxic conditions, miR-130b represses DDX6 leading to increased activity of HIF1α, a well-known VEGF inducer [15]. In endometrial cancer cell lines, instead, miR-130b leads to deregulation of miR-200 and other EMT-related genes, through DICER1 reduction [21]. Our data indicate that in CRC miR-130b induces EMT through a pathway that appears to be independent of DICER1 and its downstream target miR-200c; however, the contribution of each of these pathways to the process needs further investigations [38,39].

Herein, we validate PPARγ as a direct functional target of miR-130b in CRC, adding information to previously reported cell types [13]. We have shown that PPARγ is a CRC-independent prognostic factor implicated in cell differentiation and suppression of cell growth likely through up-regulation of its target genes E-cadherin, p21, and PTEN [9–12,40–42]. We demonstrate that, in addition to vimentin and N-cadherin, it represses also Snail, an early marker of EMT, linked to CRC metastatic progression [9,10,43,45]. Thus, PPARγ contributes to the maintenance of the epithelial characteristics preserving a less aggressive behavior and hampering tumor dissemination [42,43,45]. In addition, PPARγ inhibits VEGF expression, thus exerting an anti-angiogenic effect that could be, in turn, mediated by PTEN up-regulation, as recently reported [45–48]. More experiments are required for a deeper understanding of the complex interplay between PPARγ and angiogenesis.

Changes in PPARγ expression impact on cell proliferation, EMT, and angiogenesis, reproducing in an inverse mode the effects due to miR-130b. This establishes the existence of a functional link between them, as further demonstrated in mouse xenografts and, most importantly, in tumor samples. The reciprocal and inverse variations of PPARγ levels following miR-130b modulation clearly demonstrate that they are mechanistically linked. The tight interplay between PPARγ and miR-130b is further proven by their concomitant silencing: PPARγ knockdown completely reverts the effects on cell proliferation, EMT, and angiogenesis due to miR-130b inhibition, identifying it as a major downstream effector of this miRNA.

In light of the present data and considering that the inverse association of miR-130b with PPARγ persists in advanced tumor stages (III and IV), we hypothesize that the miR-130b-PPARγ axis can significantly influence the progression toward more invasive tumors. In the scheme of Figure 6G, some of the effects produced by PPARγ on several target genes involved in proliferation, EMT, and angiogenesis in a ligand-independent manner are illustrated. Targeting PPARγ and miR-130b signaling pathways could then represent a rational approach for the treatment of advanced CRC. In this context, modulating miR-130b could restore PPARG expression sensitizing the tumor to a combined therapy of PPARγ agonists and chemotherapeutic drugs routinely employed into the clinical practice.

In conclusion, we demonstrate that miR-130b is frequently upregulated in CRC and correlates with metastasis and poorer patients' prognosis. These characteristics are reproduced in mouse xenografts of overexpressing or down-expressing miR-130b cells and reverted by intratumor injections of miR-130b antisense or mimic. We show that PPARγ silencing phenocopies miR-130b overexpression and provide mechanistic evidence that it is a major downstream mediator. A better understanding of the miR-130b-PPARγ axis functions and interactions with other signaling pathways may clarify its clinical relevance and its prognostic role in colorectal tumorigenesis. Finally, miR-130b in vivo can be a novel predictive biomarker and represent a useful therapeutic target against CRC.