Abstract
Altered expression of microRNAs (miRNA) occurs commonly in human cancer, but the mechanisms are generally poorly understood. In this study, we examined the contribution of epigenetic mechanisms to miRNA dysregulation in colorectal cancer by carrying out high-resolution ChIP-seq. Specifically, we conducted genome-wide profiling of trimethylated histone H3 lysine 4 (H3K4me3), trimethylated histone H3 lysine 27 (H3K27me3), and dimethylated histone H3 lysine 79 (H3K79me2) in colorectal cancer cell lines. Combining miRNA expression profiles with chromatin signatures enabled us to predict the active promoters of 233 miRNAs encoded in 174 putative primary transcription units. By then comparing miRNA expression and histone modification before and after DNA demethylation, we identified 47 miRNAs encoded in 37 primary transcription units as potential targets of epigenetic silencing. The promoters of 22 transcription units were associated with CpG islands (CGI), all of which were hypermethylated in colorectal cancer cells. DNA demethylation led to increased H3K4me3 marking at silenced miRNA genes, whereas no restoration of H3K79me2 was detected in CGI-methylated miRNA genes. DNA demethylation also led to upregulation of H3K4me3 and H3K27me3 in a number of CGI-methylated miRNA genes. Among the miRNAs we found to be dysregulated, many of which are implicated in human cancer, miR-1-1 was methylated frequently in early and advanced colorectal cancer in which it may act as a tumor suppressor. Our findings offer insight into the association between chromatin signatures and miRNA dysregulation in cancer, and they also suggest that miRNA reexpression may contribute to the effects of epigenetic therapy. Cancer Res; 71(17); 5646–58. ©2011 AACR.
Introduction
MicroRNAs (miRNA) are a class of small noncoding RNAs that regulate gene expression by inducing translational inhibition or direct degradation of target mRNAs through base pairing to partially complementary sites (1). miRNA genes are transcribed as large precursor RNAs, called pri-miRNAs, which may encode multiple miRNAs in a polycistronic arrangement. The pri-miRNAs are then processed by the RNase III enzyme Drosha and its cofactor Patha to produce approximately 70-nucleotide hairpin structured second precursors (pre-miRNAs). The pre-miRNAs are then transported to the cytoplasm and processed by another RNase III enzyme, Dicer, to generate mature miRNA products. miRNAs are highly conserved among species and play critical roles in a variety of biological processes, including development, differentiation, cell proliferation, and apoptosis. Subsets of miRNAs are thought to act as tumor suppressor genes (TSG) or oncogenes, and their dysregulation is a common feature of human cancers (2). More specifically, expression of miRNAs is generally downregulated in tumor tissues, as compared with the corresponding normal tissues, which suggests that some miRNAs may behave as TSGs in some tumors. Although the mechanism underlying the alteration of miRNA expression in cancer is still not fully understood, recent studies have shown that multiple mechanisms involved in regulating miRNA levels are affected in cancer. For example, genetic mutations that affect proteins involved in the processing and maturation of miRNA can lead to overall reductions in miRNA expression levels (3, 4). In addition, genetic and epigenetic alterations can disrupt expression of specific miRNAs in cancer.
Epigenetic gene silencing due to promoter CpG island (CGI) hypermethylation is one of the most common mechanisms by which TSGs are inactivated during tumorigenesis. In recent years, it has become evident that some miRNA genes are also targets of epigenetic silencing in cancer. Others and we have previously shown that pharmacologic or genetic disruption of DNA methylation in cancer cell lines induces upregulation of substantial numbers of miRNAs (5, 6). These analyses led to identification of candidate tumor-suppressive miRNAs whose silencing was associated with CGI methylation. For example, miR-127 is embedded in a typical CGI, and treatment of human bladder cancer cells with inhibitors of histone deacetylase (HDAC) and DNA methyltransferase (DNMT) induced CGI demethylation and reexpression of the miRNA (7). In addition, methylation of miR-124 family members (miR-124-1, -124-2, and -124-3) was identified in colorectal cancer and was subsequently reported in tumors of other origins (5). Similarly, we found frequent methylation and downregulation of miR-34b/c in both colorectal cancer and gastric cancer (6, 8).
Epigenetic regulation of miRNA genes is tightly linked to chromatin signatures. For instance, transcriptionally active miRNA genes are characterized by active chromatin marks, such as trimethylated histone H3 lysine 4 (H3K4me3; ref. 9). We previously showed that restoring H3K4me3 through DNA demethylation could be a useful marker for predicting the promoter region of a silenced miRNA gene (6). However, the chromatin signatures, including both active and repressive histone marks on miRNA genes, within the cancer genome are still largely unknown. In the present study, we carried out genome-wide profiling of chromatin signatures in colorectal cancer cells and identified the active promoter regions of miRNA genes. We also show that changes in chromatin signatures before and after the removal of DNA methylation lead to robust identification of miRNA genes that are epigenetically regulated in cancer.
Materials and Methods
Cell lines and tissue specimens
Colorectal cancer cell lines and HCT116 cells harboring genetic disruptions within the DNMT1 and DNMT3B loci [double knockout (DKO)] have been described previously (6). Treatment of cells with 5-aza-2′-deoxycytidine (DAC; Sigma-Aldrich) and 4-phenylbutyrate (PBA; Sigma-Aldrich) was carried out as described (8). A total of 90 primary colorectal cancer specimens were obtained as described (6, 10). Samples of adjacent normal colorectal mucosa were also collected from 20 patients. A total of 78 colorectal adenoma specimens were obtained through endoscopic biopsy. Informed consent was obtained from all patients before collection of the specimens. Total RNA from normal colonic mucosa from healthy individuals was purchased from Ambion. Total RNA was extracted using a mirVana miRNA isolation kit (Ambion) or TRIzol reagent (Invitrogen). Genomic DNA was extracted using the standard phenol–chloroform procedure.
miRNA expression profiling
Expression of 470 miRNAs was analyzed using Human miRNA Microarray V1 (G4470A; Agilent Technologies) as described previously (8). In addition, expression of 664 miRNAs was analyzed using a TaqMan microRNA Array v2.0 (Applied Biosystems). Briefly, 1 μg of total RNA was reverse transcribed using Megaplex Pools kit (Applied Biosystems), after which the miRNAs were amplified and detected using PCR with specific primers and TaqMan probes. The PCR was run in a 7900HT Fast Real-Time PCR system (Applied Biosystems), and SDS2.2.2 software (Applied Biosystems) was used for comparative ΔCt analysis. U6 snRNA (RNU6B; Applied Biosystems) served as an endogenous control. Microarray data and TaqMan Array data (ΔCt values) were further analyzed using GeneSpring GX ver. 11 (Agilent Technologies). The Gene Expression Omnibus accession number for the microarray data is GSE29900.
Real-time reverse transcriptase PCR of miRNA
Expression of selected miRNAs was analyzed using TaqMan microRNA Assays (Applied Biosystems). Briefly, 5 ng of total RNA was reverse transcribed using specific stem-loop RT primers, after which the miRNAs were amplified and detected using PCR with specific primers and TaqMan probes as described earlier. U6 snRNA (RNU6B) served as an endogenous control. Expression of the primary miR-1-1 transcript was analyzed using a TaqMan Pri-miRNA assay (assay ID Hs03303345_pri; Applied Biosystems). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH; assay ID Hs99999905_m1; Applied Biosystems) served as an endogenous control.
Chromatin immunoprecipitation-on-chip analysis
Chromatin immunoprecipitation (ChIP)-on-chip analysis was carried out according to Agilent Mammalian ChIP-on-chip Protocol version 10.0 (Agilent Technologies). Briefly, 1 × 108 cells were treated with 1% formaldehyde for 10 minutes to cross-link histones with the DNA. After washing with PBS, the cell pellets were resuspended in 3 mL of lysis buffer and sonicated. Chromatin was immunoprecipitated for 16 hours at 4°C using 10 μL of anti-trimethyl histone H3K4 (clone MC315; Upstate), anti-trimethyl histone (clone H3K27; Upstate) or anti-dimethyl histone H3K79 (clone NL59; Upstate) antibody. Before adding antibodies, 50 μL of the each cell lysate was saved as an internal control for the input DNA. After washing, elution, and reversal of the cross-links, input DNA and the immunoprecipitate were ligated to linkers and PCR amplified. Input DNA and the immunoprecipitate were then labeled with Cy3 and Cy5 using an Agilent Genomic DNA Enzymatic Labeling kit (Agilent Technologies) and hybridized to the 244K Human Promoter ChIP-on-chip microarray (G4489A; Agilent technologies). After washing, the array was scanned using an Agilent DNA Microarray scanner (Agilent Technologies), and the data were processed using Feature Extraction software (Agilent Technologies).
ChIP-seq analysis
ChIP experiments were carried out as described earlier, after which massively parallel sequencing was carried out using a SOLiD3 Plus system (Applied Biosystems) according to the manufacturer's instructions. Briefly, 100 ng of input DNA or the immunoprecipitate was ligated to adapters and PCR amplified using a SOLiD Fragment Library Construction kit (Applied Biosystems). Template bead preparation was carried out using a SOLiD ePCR kit V2 (Applied Biosystems) and a SOLiD Bead Enrichment kit (Applied Biosystems). Approximately 40 to 50 million beads per sample were sequenced using SOLiD Opti Fragment Library Sequencing Master Mix 50 (Applied Biosystems) and a SOLiD3 Plus sequencer (Applied Biosystems). Sequence reads that were of poor quality or those that were not uniquely mapped were excluded from the study. Peaks were identified using the Model-based Analysis for ChIP-seq (MACS) software (11) and visualized using the University of California Santa Cruz (UCSC) genome browser.
Reference sequence
Genomic locations are based on the UCSC hg18 (National Center for Biotechnology Information Build 36.1, March 2006), which was produced by the International Human Genome Sequencing Consortium. We also obtained locations of CGIs, ReSeq genes, and UCSC genes from the UCSC hg18 data sets.
Methylation analysis
Genomic DNA (2 μg) was modified with sodium bisulfite using an EpiTect Bisulfite kit (QIAGEN). Methylation-specific PCR (MSP), bisulfite sequencing, and bisulfite pyrosequencing were carried out as described (6). For bisulfite sequencing analysis, amplified PCR products were cloned into pCR2.1-TOPO vector (Invitrogen), and 10 to 12 clones from each sample were sequenced using an ABI3130x automated sequencer (Applied Biosystems). Primer sequences and PCR product sizes are listed in Supplementary Table S1.
Transfection of miRNA precursor molecules
Colorectal cancer cells (1 × 106 cells) were transfected with 100 pmol of Pre-miR miRNA Precursor Molecules (Ambion) or Pre-miR miRNA Molecules Negative Control #1 (Ambion) using a Cell Line Nucleofector kit V (Lonza) with a Nucleofector I electroporation device (Lonza) according to the manufacturer's instructions. Total RNA or cell lysate was extracted 48 hours after transfection. Cell viability assays, Western blotting, wound-healing assays, and Matrigel invasion assays are described in the Supplementary Methods.
Gene expression profiling
Total RNA (700 ng) was amplified and labeled using a Quick Amp Labeling kit one-color (Agilent Technologies), after which the synthesized cRNA was hybridized to the Whole Human Genome Oligo DNA microarray (G4112F; Agilent technologies). Data analysis was carried out using GeneSpring GX ver. 11 (Agilent technologies). The Gene Expression Omnibus accession number for the microarray data is GSE29760.
miRNA target predictions and luciferase reporter assays
The predicted targets of miR-1 and their downstream target sites were analyzed using TargetScan and miRanda. Construction of luciferase reporter vectors containing the predicted target sites and dual luciferase reporter assays were carried out as described in Supplementary Methods.
Results
miRNA profiling in colorectal cancer cell lines
To screen for epigenetically silenced miRNAs, we first carried out miRNA microarray analysis in a series of colorectal cancer cell lines (HCT116, DLD1, and RKO) and normal colonic tissue. Hierarchical clustering analysis revealed that expression of a majority of miRNAs was downregulated in all 3 colorectal cancer cell lines tested, as compared with normal colonic mucosa (Supplementary Fig. S1A). DAC treatment upregulated expression of a large number of miRNAs in all 3 colorectal cancer cell lines (Supplementary Fig. S1B), and combination treatment with DAC plus PBA induced even greater numbers of miRNAs in colorectal cancer cells (Supplementary Fig. S1C and D). However, the most profound effect on the miRNA expression profile was induced by genetic disruption of DNMT1 and DNMT3B in HCT116 cells (DKO cells; Supplementary Fig. S1C). We also noted a novel overlap between miRNAs upregulated by pharmacologic or genetic disruption of DNA methylation and those downregulated in colorectal cancer cells, as compared with normal colonic mucosa (Supplementary Fig. S1E–G). To test the tumor-suppressive potentials of the downregulated miRNAs, we constructed expression vectors encoding selected miRNAs and carried out colony formation assays. We found that a majority of miRNAs exerted growth-suppressive effects when they were ectopically expressed in colorectal cancer cells (Supplementary Fig. S2). These results suggest that an epigenetic mechanism plays an essential role in the downregulation of a number of miRNAs in cancer and that such downregulation of numerous miRNAs may contribute to tumorigenesis.
Chromatin signatures of active and silenced miRNA genes
We next examined the chromatin signatures of miRNA genes in HCT116 colorectal cancer cells, with and without genetic disruption of DNMT1 and DNMT3B (DKO cells). We carried out ChIP analysis using antibodies against trimethylated histone H3 lysine 4 (H3K4me3), which marks active promoters; dimethylated histone H3 lysine 79 (H3K79me2), which is associated with transcriptional elongation; and trimethylated histone H3 lysine 27 (H3K27me3), which is a repressive mark. We started our analysis using the Agilent 244K Promoter Array, which covers approximately 370 human miRNA genes, and we subsequently migrated to ChIP-seq analysis to increase our scope within the genome. We observed a good correlation between the results of the ChIP-on-chip and ChIP-seq analyses (Supplementary Fig. S3). We also validated the reliability of our ChIP-seq data by checking representative protein-coding genes that were transcriptionally active or silenced in HCT116 cells (Supplementary Fig. S4).
Representative chromatin signatures of miRNA genes are shown in Fig. 1A. We found enrichment of the H3K4me3 mark around the proximal upstream CGI regions of 2 abundantly expressed miRNA clusters, miR-200b and miR-17, in both wild-type HCT116 and DKO cells (Fig. 1A). Gene bodies were marked by H3K79me2, which indicates active transcriptional elongation, whereas they almost completely lacked the repressive H3K27me3 mark. With respect to the H3K4me3 mark in the miR-17 cluster, we observed a sharp dip at the transcription start site (TSS) of the host gene and another dip downstream, which is consistent with a previous report that miR-17 has its own TSS within the intron of the host gene (Fig. 1A; ref. 12).
Chromatin signatures of transcriptionally active and epigenetically silenced miRNA genes in colorectal cancer. A, ChIP-seq results for H3K4me3, H3K79me2, and H3K27me3 in transcriptionally active miRNA genes in HCT116 and DKO cells. Chromosomal locations are indicated on the top. Locations of host genes, pre-miRNA genes, and CGIs are shown below. B, ChIP-seq results for epigenetically silenced miRNAs with associated CGI hypermethylation. CGI methylation is lost and miRNAs are reexpressed in DKO cells. H3K4me3 marking is upregulated in the putative promoter regions in DKO cells, whereas H3K79me2 shows only a minimal increase.
In contrast, miRNAs whose silencing was associated with promoter CGI hypermethylation completely lacked both of the active histone marks. The CGIs of miR-34b/c, miR-124-1, and miR-9-3 were densely methylated in HCT116 cells (5, 6, 13) and were completely devoid of H3K4me3 and H3K79me2 marks (Fig. 1B). miR-124-1 and miR-9-3 showed moderate enrichment of H3K27me3, whereas miR-34b/c was almost H3K27me3 free, which corresponds to previous reports that DNA methylation and H3K27me3 are sometimes observed independently in cancer (14). In DKO cells, where DNA methylation was significantly diminished and gene expression was restored, increased H3K4me3 marks were found at the upstream CGI, though restoration of H3K79me2 was quite limited. Upregulation of H3K27me3 was also seen around miR-124-1 and miR-9-3, which is consistent with previous observations that genes with methylated CGIs adopt a bivalent chromatin pattern after DNA demethylation (15, 16).
Identification of putative miRNA promoter regions
Identification of epigenetically silenced miRNAs is sometimes hampered by a lack of knowledge of the transcription initiation region of the primary miRNA transcripts. Previous studies have shown that H3K4me3 is a useful marker for identifying active miRNA gene promoters (9, 12), and we employed that approach with colorectal cancer cells. Using miRNA microarrays and TaqMan low-density arrays, we detected expression of 339 and 429 distinct mature miRNAs in HCT116 and DKO cells, respectively. We then searched for the putative promoter regions of these miRNAs, using H3K4me3 as a marker.
More than half of miRNAs are located in the introns of protein-coding or long noncoding RNA genes, and it is generally believed that intragenic miRNAs share common promoters with their host genes (17). We identified the putative promoters of 166 intragenic miRNAs located in RefSeq genes and/or UCSC genes, and a majority of the H3K4me3 marks were observed at the TSS of the host genes, many of which were located more than 10 kb upstream of the pre-miRNA coding regions (Fig. 2A and C, Supplementary Fig. S5A, and Supplementary Table S2). In contrast, intragenic H3K4me3 marks were identified in the proximal upstream of 22 pre-miRNAs, indicating these miRNAs have their own promoters and are transcribed independently of their host genes (Supplementary Fig. S6, Supplementary Table S3). To identify promoters of intergenic miRNAs, we first searched 10 kb upstream for H3K4me3 marks and also explored the initiation sites of overlapping 5′ expressed sequence tags (EST). We identified the putative promoters of 66 intergenic miRNAs, the majority of which (47 of 66) were identified in the proximal upstream (<2 kb) of the pre-miRNA coding region (Fig. 2B and D, Supplementary Fig. S5B, and Supplementary Table S2). In total, we identified the putative promoters of 174 transcript units encoding 233 distinct pre-miRNAs, whereas promoters of 135 miRNAs remain unidentified, despite their positive expression in colorectal cancer cells.
Identification of miRNA gene promoter regions using chromatin signatures. A, examples of H3K4me3 marks in intragenic miRNAs. Let-7g is located within the intron of the protein-coding gene WDR82, and miR-34a is located within the exon of a noncoding host gene. H3K4me3 marks are observed in the TSS regions of the host genes, suggesting that these miRNAs share common promoters with their host genes. B, examples of H3K4me3 marking of intergenic miRNA genes. C, summarized distances between intragenic pre-miRNA coding regions and their putative promoter regions (n = 166). D, summarized distances between intergenic pre-miRNA coding regions and their putative promoter regions (n = 67).
We validated our promoter search by comparing our results with previously reported transcription initiation regions. Promoters of 177 pre-miRNAs that we identified overlapped with those identified in human embryonic stem (ES) cells by Marson and colleagues (9), whereas only the promoters of 38 pre-miRNAs did not match. Similarly, the TSS of 65 miRNAs identified in human melanoma and breast cancer cell lines by Ozsolak and colleagues overlapped with the promoters we identified (12). For example, we found H3K4me3 marks overlapping with known TSS of the miR-17 cluster, let-7a-1/let-7f-1/let-7d, and miR-200c/141 (Fig. 2B, Supplementary Fig. S5). We also identified an H3K4me3 mark at the intronic transcription initiating region of miR-21 (Supplementary Fig. S6C). The high degree of consistency between our results and those of earlier studies attests to the accuracy of our promoter prediction.
Identification of epigenetically silenced miRNAs
We next endeavored to identify epigenetically silenced miRNA genes by taking advantage of the observation that DNA demethylation can induce increases in H3K4me3 in the promoters of the epigenetically silenced genes (6). We searched for miRNA genes showing reduction or loss of both expression and H3K4me3 marks in HCT116 and DKO cells. We identified 47 pre-miRNA genes encoded in 37 primary transcription units as potential targets of epigenetic silencing in HCT116 cells. Promoters of 22 transcription units were associated with CGIs, and MSP analysis revealed that all of the CGIs were methylated (Fig. 3A and B, Table 1). In most cases, DNA demethylation led to increases in H3K4me3 and H3K27me3 marking of the methylated CGIs of miRNA genes, whereas H3K79me2 marks were not restored by demethylation (Fig. 3C). In contrast, the chromatin signatures of miRNAs without promoter CGIs were more variable among genes. We noted that a small number of non-CGI miRNAs acquired more active chromatin states upon DNA demethylation than did CGI-methylated miRNAs. For instance, miR-146a is characterized by a lack of active histone marks and enrichment of H3K27me3, but it showed restoration of both H3K4me3 and H3K79me2 in DKO cells (Fig. 3C). We observed similar upregulation of both active marks in miR-142 (Supplementary Fig. S7E). Weak basal expression of these miRNAs, detectable by TaqMan assay but not by microarray, and robust upregulation after DNA demethylation indicate that the silencing of these miRNAs is less stringent than that of miRNAs with methylated CGIs (data not shown).
Identification of epigenetically silenced miRNA genes. A, flowchart for the selection of epigenetically silenced miRNA genes in colorectal cancer. B, graph showing the number of epigenetically silenced miRNAs associated with CGI methylation and those without CGI methylation. C, chromatin signatures of 2 representative miRNA genes, with and without promoter CGI methylation. miR-1-1 (top) was silenced in association with CGI methylation in HCT116 cells. In DKO cells, H3K4me3 marking was observed around the transcription start site of the host gene C20orf166. miR-146a (bottom) is another candidate target for epigenetic silencing in HCT116, though its promoter is not associated with CGI. Both H4K4me3 and H3K79me2 were restored in DKO cells. D, expression levels of epigenetically silenced miRNAs and their host genes in HCT116 and DKO cells. TaqMan real-time PCR data for 57 mature miRNAs encoded by 47 pre-miRNA genes were imported into Gene Spring GX, after which the data were normalized and shown in box plots (left). Expression data of 13 host genes of epigenetically silenced miRNAs were obtained using an Agilent Whole Human Genome microarray (right). E, miRNAs targeted by the PcG group in ES cells are more likely to be silenced by CGI hypermethylation in colorectal cancer cells. CGI-methylated miRNAs (n = 22; left) or transcriptionally active miRNAs (n = 146; right) were selected, and their SUZ12 binding and H3K27me3 enrichment in human ES cells were assessed. Of 22 CGI-methylated miRNAs, 13 (59%) were positive for SUZ12 and 16 (73%) for H3K27me3. ND, not determined.
Epigenetically silenced miRNA genes in HCT116
DNA demethylation significantly upregulated the expression of mature miRNAs derived from 47 silenced pre-miRNAs (Fig. 3D). In addition, expression data from 13 host genes of the silenced miRNAs were obtained from Agilent gene expression microarray analysis (6), and we observed a strong tendency for the host genes to be upregulated by DNA demethylation (Fig. 3E). Recent studies have shown that genes marked by polycomb (PcG) group proteins in ES cells have a predisposition toward DNA hypermethylation in cancer (18, 19). By comparison with previously published results (9), we found that miRNAs with SUZ12 binding and H3K27me3 marks in human ES cells are significantly enriched in CGI-methylated miRNAs in colorectal cancer (Fig. 3E).
We further analyzed CGI methylation in a series of colorectal cancer cell lines using MSP and bisulfite pyrosequencing and found that they are methylated to varying degrees (Fig. 4A, Supplementary Fig. S8). We also confirmed inverse relationships between methylation and expression of selected miRNAs in colorectal cancer cell lines and normal colonic tissue (Fig. 4B). To determine the extent to which these miRNA genes are aberrantly methylated in primary tumors, we carried out bisulfite pyrosequencing of 18 miRNA promoter CGIs in primary colorectal cancer tumors (n = 90) and normal colonic tissue obtained from colorectal cancer patients (n = 20; Supplementary Fig. S9). Most of the miRNA genes were methylated in a tumor-specific or tumor-predominant manner. The two exceptions were miR-153-2 and miR-196a-1, which were methylated to similar degrees in both normal colon and tumor tissues, as well as in various normal human tissues (Supplementary Figs. S9 and S10). Elevated levels of miRNA gene methylation (>15.0%) were frequently detected in primary colorectal cancer tumors (miR-1-1, 77.8%; miR-9-1, 57.8%; miR-9-3, 89.9%; miR-34b/c, 89.7%; miR-124-1, 87.7%: miR-124-2, 96.6%; miR-124-3, 100.0%; miR-128-2, 73.6%; miR-129-2, 40.0%; miR-137, 100.0%; miR-193a, 28.7%; miR-338, 15.6%; and miR-548b, 47.8%), whereas a small number of genes were rarely methylated in primary tumors (miR-152, 4.4%; miR-155, 6.7%; and miR-596, 2.3%).
DNA methylation and expression analysis of miRNAs in colorectal cancer cells. A, representative results of MSP analysis of a series of colorectal cancer cell lines and normal colonic tissue. Bands in the “M” lanes are PCR products obtained with methylation-specific primers; those in the “U” lanes are products obtained with unmethylated-specific primers. In vitro methylated DNA (IVD) serves as a positive control. B, relationship between DNA methylation and expression of miRNAs in colorectal cancer. Bisulfite pyrosequencing results for miRNA promoter CGIs (black bars) and TaqMan real-time PCR results for mature miRNAs (gray bars) in a series of colorectal cancer cell lines and normal colonic tissue are shown. RT-PCR results were normalized to internal U6 snRNA expression.
MiR-1-1 is a candidate tumor suppressor gene in colorectal cancer
Among the epigenetically silenced miRNAs, we next focused on miR-1-1 because it has received relatively little attention in colorectal cancer despite its frequent hypermethylation in that disease. Using bisulfite pyrosequencing, we detected elevated levels (>15.0%) of miR-1-1 methylation in both primary colorectal cancer tumors and colorectal adenomas (54 of 78, 69.2%), suggesting that its methylation is an early event in colorectal tumorigenesis (Fig. 5A). In contrast, levels of miR-1-1 methylation were relatively low (<15.0%) in the normal colonic tissues tested (Fig. 5A). We carried out bisulfite sequencing analysis to confirm the methylation results in selected tissue specimens and colorectal cancer cell lines (Fig. 5B, Supplementary Fig. S11A and B). We also confirmed that DNA demethylation could restore expression of the primary transcript of miR-1-1 (pri-miR-1-1) in colorectal cancer cells (Supplementary Fig. S11C).
Methylation and functional analysis of miR-1-1 in colorectal cancer. A, summarized bisulfite pyrosequencing results for the miR-1-1 promoter CGI in normal colonic tissue (n = 20), colorectal adenomas (n = 78), and primary colorectal cancer tumors (n = 90). B, representative bisulfite sequencing results for the miR-1-1 promoter in a sample of normal colonic tissue and a primary colorectal cancer tumor. Open and filled circles represent unmethylated and methylated CpG sites, respectively. C, MTT assays with colorectal cancer cell lines transfected with a miR-1 precursor molecule or a negative control. Cell viabilities were determined 48 hours after transfection. Values were normalized to cells transfected with the negative control. Shown are the means of 8 replications; error bars represent SDs. D, colony formation assays using HCT116 cells transfected with a miR-1-1 expression vector or a control vector. Representative results are shown on the left, and relative colony formation efficiencies are on the right. Shown are means of 3 replications; error bars represent SDs. E, Western blot analysis of Annexin A2 in HCT116 cells transfected with a miR-1 precursor molecule or a negative control. Precursor of miR-17, which is abundantly expressed in HCT116 cells and is irrelevant to miR-1, served as another negative control. F, putative miR-1 binding site in the 3′ untranslated region (UTR) of ANXA2. A fragment that included the binding site was PCR amplified and cloned into pMIR-REPORT vector. G, reporter assay results using the luciferase vector with the 3′ UTR of ANXA2 or an empty vector in HCT116 cells cotransfected with a miR-1 precursor, a negative control (Cont), or a miR-17 precursor. Shown are the means of 4 replications; error bars represent SDs. H, wound-healing assay using HCT116 cells transfected with a miR-1 precursor or a negative control. The wound was made 24 hours after transfection, and photographs were taken at the indicated time points. I, Matrigel invasion assay using HCT116 cells transfected with a miR-1 precursor, a negative control, or a miR-17 precursor. Invading cells are indicated by arrowheads. Shown on the right are the means of 3 random microscopic fields per membrane; error bars represent the SDs.
To determine whether miR-1-1 serves as a tumor suppressor in colorectal cancer, we transfected colorectal cancer cell lines with a miR-1 precursor molecule or a negative control and then carried out a series of MTT assays. Forty-eight hours after transfection, we observed that ectopic expression of miR-1 moderately suppressed growth in all 3 cell lines (Fig. 5C). Colony formation assays also revealed reduced colony formation by colorectal cancer cells transfected with a miR-1-1 expression vector (Fig. 5D).
To further clarify the effect of the miRNA, we next carried out a gene expression microarray analysis of HCT116 cells transfected with a miR-1 precursor molecule or a negative control. We found that 2,769 probe sets were downregulated (>1.5-fold) by ectopic miR-1 expression, and gene ontology analysis revealed that “extracellular regions,” “membrane,” and “response to wounding” genes were significantly enriched among the downregulated genes (Supplementary Table S4). The genes downregulated by miR-1 included a number of predicted miR-1 targets (Supplementary Table S5). Among them, we noted 2 genes, Annexin A2 (ANXA2) and brain-derived neurotrophic factor (BDNF), which have been implicated in tumor growth and metastasis (20–22). Reduction of their expression by miR-1 in colorectal cancer cells was confirmed by Western blotting and real-time reverse transcriptase PCR (RT-PCR; Fig. 5E, Supplementary Fig. S12A). Reporter assays using luciferase vectors containing the putative miR-1 binding sites revealed that cotransfection of a miR-1 precursor molecule markedly reduced luciferase activities and that such reductions were not induced by a negative control or an irrelevant miRNA molecule (Fig. 5F and G, Supplementary Fig. S12B and S12C). Finally, we carried out wound-healing and Matrigel invasion assays to test the effect of miR-1 expression on colorectal cancer cell migration and invasion. We found that wound closure by HCT116 cells transfected with the negative control was complete within 28 hours whereas miR-1–expressing cells migrated toward the wound at a much slower rate (Fig. 5H). We also observed significant inhibition of cell invasion by miR-1 in HCT116 cells (Fig. 5I). These results strongly suggest that miR-1 acts as a tumor suppressor in colorectal cancer.
Discussion
In the present study, we provide a comprehensive view of the epigenetic regulation of miRNA genes in colorectal cancer cells. Because of the poor annotation of primary miRNA genes, the precise locations of the promoters and TSSs are not fully understood yet. To overcome these difficulties, earlier studies have searched for specific genomic features including RNA polymerase (pol) II binding patterns (23, 24), evolutionally conserved regions (25), EST mapping (26), and computationally predicted promoters (27, 28). Active promoters are reportedly marked by H3K4me3 (29), and recent studies that have applied such histone marks have successfully identified miRNA gene promoters or TSSs (9, 12). In the present study, we carried out high-resolution ChIP-seq analyses in an effort to detect the chromatin signatures of miRNA genes in colorectal cancer.
Although we were able to identify the putative promoters of a number of miRNAs, the present study has several limitations. First, our strategy to identify miRNA promoters can be applied only to transcriptionally active genes. Second, promoters of 135 miRNAs remain unidentified, although their expression was detected in colorectal cancer cells. The majority of such miRNAs (103 of 135) are located in the intergenic regions, and if we increase our search scope, we may identify putative promoter regions in the further upstream, although the accuracy may be decreased. For example, in DKO cells, we detected abundant expression of placenta-specific miRNAs transcribed from a miRNA cluster on chromosome 19 (C19MC), suggesting these miRNAs are epigenetically silenced in normal adult tissues. We found an H3K4me3 mark around a CGI located approximately 18 kb upstream of the cluster, suggesting that this region may be a putative promoter of C19MC (Supplementary Fig. S13), which is consistent with a recent report that hypermethylation of this CGI is associated with epigenetic silencing of C19MC in human cancer cell lines (30). However, other studies have shown that the Alu repetitive sequences within which C19MC is embedded exhibit RNA pol II or pol III promoter activities (31, 32), but we failed to detect obvious active histone marks in these Alu repeats. These results suggest that C19MC may have multiple promoter regions and point to a limitation of the strategy we employed in the current study.
Despite this limitation, chromatin signatures provided important clues to the identity of epigenetically silenced miRNAs in cancer. In HCT116 cells, for instance, the miR-9-1 promoter showed significant enrichment of active histone marks and mature miR-9 was abundantly expressed (data not shown). On the other hand, lack of H3K4me3 in the same cells and its restoration after DNA demethylation clearly suggest that miR-9-2 and miR-9-3 are epigenetically silenced in these cells, which is indicative of the utility of our strategy. We also noted that chromatin signatures of epileptically silenced miRNA genes exhibit patterns similar to those of protein-coding genes. Recent studies have shown that TSGs with CGI methylation retain repressive histone modifications (H3K9me3 and H3K27me3) even after demethylation (15). A genome-wide analysis of the chromatin signature using ChIP-on-chip in colorectal cancer cells revealed that hypermethylated genes adopt a bivalent chromatin pattern upon DNA demethylation (16). More recently, Jacinto and colleagues found that DNA demethylation never results in restoration of the H3K79me2 mark in TSGs with methylated CGIs, suggesting that such incomplete chromatin reactivation leads to relatively low levels of reexpression (33). In the present study, we found that miRNA genes with methylated CGIs never return to a full euchromatin status after DNA demethylation. In addition, we observed significant overlap between PcG marked miRNAs in ES cells and miRNAs with CGI methylation in cancer cells, suggesting a strong predisposition of these miRNAs toward aberrant DNA methylation in cancer.
Many of the epigenetically silenced miRNA genes we identified have been implicated in human malignancies. miR-124 family, miR-9 family, miR-34b/c, and miR-129-2 were identified by screening for epigenetically silenced miRNAs in colorectal cancer cell lines (5, 6, 13), and their methylation was subsequently found in various cancers (8, 34–36). Methylation-associated silencing of miR-137 was first reported in oral cancer (37), and a recent study revealed its frequent methylation in the early stages of colorectal tumorigenesis (38). The high frequency of CGI hypermethylation in these miRNAs in primary colorectal cancer is suggestive of their tumor suppressor function. It was also recently shown that the muscle-specific miRNAs miR-1 and miR-133a are downregulated in primary colorectal cancer tumors as compared with normal colonic tissues (39). Reduced expression of miR-1 is also found in lung cancer (40), and CGI methylation–mediated silencing of miR-1-1 has been reported in hepatocellular carcinoma (41). In addition, levels of miR-1 expression were diminished in the serum of non–small-cell lung cancer (NSCLC) patients who survived for only a short period, suggesting that it is predictive of prognosis in NSCLC patients (42). Ectopic expression of miR-1 in lung cancer, liver cancer, and rhabdomyosarcoma cells reportedly inhibits cellular growth through suppression of its target genes, which include MET, FOXP1, and HDAC4 (40, 41, 43). In the present study, we found frequent methylation of the miR-1-1 promoter CGI in both colorectal adenoma and primary colorectal cancer tissues, suggesting that aberrant methylation of miR-1-1 is an early event in colorectal tumorigenesis. The strong tumor specificity of the methylation indicates that it could be a novel tumor marker for early detection of colorectal neoplasia. Because the tumor suppressor potential of miR-1 has not been tested in colorectal cancer, we conducted a number of functional analyses, and our findings indicate that ectopic expression of miR-1 in colorectal cancer cells suppresses cell growth, colony formation, cell motility, and invasion. In addition, our gene expression analysis revealed that miR-1 could induce global changes in gene expression in colorectal cancer cells, especially genes related to the extracellular region, cell membrane, and wound healing. We identified 2 novel miR-1 target genes, ANXA2 and BDNF, which are frequently overexpressed in cancer and are implicated in invasion and metastasis (20–22). These results are suggestive of the tumor suppressor role of miR-1 and its potential therapeutic application in colorectal cancer.
On the other hand, we unexpectedly detected silencing of several miRNAs with known oncogenic properties. For example, miR-155 is a well-characterized oncogenic miRNA that is overexpressed in various human malignancies (44). Although we found miR-155 to be silenced with CGI methylation in HCT116 cells, its methylation was rarely observed in primary tumors, suggesting that epigenetic silencing of miR-155 may not be functionally important in colorectal cancer. Similarly, miR-196a-1 is reportedly overexpressed in several human malignancies, including esophageal adenocarcinoma and glioblastoma (45, 46). Methylation levels of miR-196a-1 in primary colorectal cancer tumors are lower than in normal colonic tissue, which is in agreement with its possible oncogenic properties in colorectal cancer.
Finally, our chromatin signature analysis revealed that a number of miRNAs without promoter CGIs are also potential targets of epigenetic silencing in colorectal cancer. These miRNAs were identified through restoration of both their expression and H3K4me3 marking after DNA demethylation, whereas the signatures of H3K79me2 and H3K27me3 varied among genes. This category may thus include miRNAs induced by secondary effects of DNA demethylation, such as upregulation of transcription factors. It is noteworthy, however, that some functionally important miRNAs showed chromatin signatures that were distinct from CGI-methylated miRNAs. Upon DNA demethylation, miR-142 and miR-146a exhibited more active chromatin states, which were characterized by enrichment of both H3K4me3 and H3K79me2 marks. Earlier studies implicated their tumor suppressor roles in cancers of various origins. For instance, miR-142 was found to be downregulated in murine and human lung cancer and its expression suppressed cancer cell growth (47). Loss of miR-146a was reported in hormone-refractory prostate cancer (48), and expression of miR-146a suppressed NF-κB activity and metastatic potential in breast and pancreatic cancer cells (49, 50). The abundant expression of miRNAs in normal colon and downregulation in multiple colorectal cancer cell lines indicates their tumor-suppressive properties in colorectal cancer (data not shown), though further study is need to define the functions of miRNAs in colorectal tumorigenesis.
With this study, we provide compelling evidence that both CGI-positive and -negative miRNAs are targets of epigenetic silencing in colorectal cancer. Our data suggest that DNA demethylation can alter the chromatin signatures of numerous miRNAs in cancer and that reexpression of these miRNAs has important relevance to the effects of epigenetic cancer therapy.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant Support
This study was supported in part by Grants-in-Aid for Scientific Research on Priority Areas (M. Toyota and K. Imai), a Grant-in-Aid for the Third-term Comprehensive 10-year Strategy for Cancer Control (M. Toyota), a Grant-in-Aid for Cancer Research from the Ministry of Health, Labor, and Welfare, Japan (M. Toyota), the A3 foresight program from the Japan Society for Promotion of Science (H. Suzuki), and Grants-in-Aid for Scientific Research (A) from the Japan Society for Promotion of Science (K. Imai).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Acknowledgments
The authors thank Dr. William F. Goldman for editing the manuscript and M. Ashida for technical assistance.
Footnotes
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received March 27, 2011.
- Revision received June 28, 2011.
- Accepted June 30, 2011.
- ©2011 American Association for Cancer Research.
References
Volume 71, Issue 17
