miR-25 targets TNF-related apoptosis inducing ligand (TRAIL) death receptor-4 and promotes apoptosis resistance in cholangiocarcinoma


  • Potential conflict of interest: Nothing to report.

  • Supported by grant number K01 DK079875 (to J.L.M.) and T32 DK007198 (to N.R.) from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) at the National Institutes of Health. Additional support was from the Mayo Clinic Center for Cell Signaling in Gastroenterology (P30DK084567). The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of NIDDK.


It has been established that microRNA expression and function contribute to phenotypic features of malignant cells, including resistance to apoptosis. Although targets and functional roles for a number of microRNAs have been described in cholangiocarcinoma, many additional microRNAs dysregulated in this tumor have not been assigned functional roles. In this study, we identify elevated miR-25 expression in malignant cholangiocarcinoma cell lines as well as patient samples. In cultured cells, treatment with the Smoothened inhibitor, cyclopamine, reduced miR-25 expression, suggesting Hedgehog signaling stimulates miR-25 production. Functionally, miR-25 was shown to protect cells against TNF-related apoptosis-inducing ligand (TRAIL)-induced apoptosis. Correspondingly, antagonism of miR-25 in culture sensitized cells to apoptotic death. Computational analysis identified the TRAIL Death Receptor-4 (DR4) as a potential novel miR-25 target, and this prediction was confirmed by immunoblot, cell staining, and reporter assays. Conclusion: These data implicate elevated miR-25 levels in the control of tumor cell apoptosis in cholangiocarcinoma. The identification of the novel miR-25 target DR4 provides a mechanism by which miR-25 contributes to evasion of TRAIL-induced cholangiocarcinoma apoptosis. (HEPATOLOGY 2012)

MicroRNA expression is altered in a variety of human cancers, with some microRNAs exhibiting increased expression levels and others reduced amounts compared to nontumor samples.1 Functionally, microRNAs contribute to many processes, including oncogenesis, angiogenesis, cell death, and metastasis. Thus, the control of microRNA levels in cancer cells, as well as their particular protein targets, is an active area of investigation.

Among the microRNAs known to be dysregulated in disease states, miR-25 is expressed at altered levels in a number of cancers. miR-25 is 22 nucleotides in length, hosted by the minichromosome maintenance protein-7 (MCM7) gene, and transcribed as part of the mir-106b∼25 polycistron. Examples of miR-25 dysregulation include its upregulation in pediatric brain tumors of multiple histological classifications,2 prostate carcinoma,3 gastric adenocarcinoma,4 and epidermal growth factor receptor-positive lung adenocarcinoma.5 In the liver, miR-25 is elevated in hepatocellular carcinoma cell lines6, 7 and human hepatocellular carcinoma samples,8, 9 and was among the microRNAs shown to be overexpressed in intrahepatic cholangiocarcinoma.10 Further, exposure to tamoxifen, a known inducer of liver cancer, increased hepatic miR-25 expression in rats.11 Notably, elevated miR-25 levels are not universally associated with a worse disease course, for instance, in acute myelogenous leukemia.12 Therefore, the particular functional effect of elevated microRNA quantities cannot necessarily be generalized from one tumor type to another.

Cholangiocarcinoma represents the second most common primary hepatobiliary cancer13 and is resistant to apoptosis mediated by TNF-related apoptosis inducing ligand (TRAIL), a major player in the extrinsic death pathway.14 Similar to other tumors, cholangiocarcinoma paradoxically expresses TRAIL15-17 and, therefore, may rely on strong cellular survival signals for tumor development. We report a role for miR-25 in regulating apoptotic signaling in cholangiocarcinoma cell lines and human tumor samples. Increased miR-25 repressed Death Receptor-4 (DR4) protein expression and protected cells from TRAIL-induced death, and this effect was reversed on antagonism of miR-25.


DAPI, 4′-6-diamidino-2-phenylindole; DR4, Death Receptor-4; GFP, green fluorescent protein; LNA, locked nucleic acid; MCM7, minichromosome maintenance protein-7; qRT-PCR, quantitative reverse transcription polymerase chain reaction; SAM, significance analysis of microarray; SEM, standard error of the mean; TRAIL, TNF-related apoptosis inducing ligand; UTR, untranslated region.

Materials and Methods

Human Tissue Samples.

All protocols involving patient materials were approved by the Institutional Review Board, Mayo Clinic (Rochester MN). Cholangiocarcinoma samples from 15 patients and benign adjacent tissue from four patients were obtained intraoperatively and preserved according to a frozen tissue storage protocol. All samples were reviewed by a pathologist to verify the presence of malignant transformation.

Cell Culture.

H69 is a nonmalignant SV40-immortalized human cholangiocyte cell line, cultured as described.18 Malignant, patient-derived cholangiocarcinoma cell lines were: KMCH cells, derived from intrahepatic cholangiocellular-hepatocellular carcinoma, and Mz-ChA-1 cells, derived from metastatic gallbladder cancer. These were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) heat-inactivated fetal calf serum, penicillin, streptomycin, and geneticin.

Manipulation of MicroRNA Levels.

Consistent with current nomenclature,19 we refer to the mature sequences as “miR” (as in miR-25; note capitalization), whereas clustered microRNAs, genes, and precursors are labeled “mir” (as in mir-106b∼25). pCDNA-mir-25 was constructed by amplification of 184 basepairs of human miR-25 and nearby sequence using the following primers: forward 5′-TCTGGTCTCCCTCACA GGAC; reverse 5′-CATGGGTCGCCTACTCAC. After TOPO TA cloning (Invitrogen, Carlsbad, CA), the SpeI/EcoRV fragment was subcloned into the XbaI/EcoRV sites of pCDNA3.1(+). In vitro transcription was performed (T7, Mega Script; Ambion, Austin, TX) on linearized pCDNA-mir-25 or empty pCDNA3.1(+) as control. Antisense locked nucleic acids (LNA) were targeted to miR-25 or miR-106b and compared to Negative Control A LNA (Exiqon, Vedbaek, Denmark). Cell lines were transiently transfected with microRNA precursor (final concentration, 7.5 nM) or antagonist (final concentration, 20 nM) with FuGENE HD (Promega, Madison, WI) and Lipofectamine 2000 (Invitrogen) reagents, respectively. Total RNA was isolated 24 (precursor) or 48 hours (LNA) after transfection.

Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR).

Total RNA was isolated using the mirVana kit (Ambion; for microRNA) or by Trizol (messenger RNA [mRNA]). MicroRNA was quantified by qRT-PCR (Applied Biosystems, Foster City, CA). Profiling of 95 microRNAs plus RNU-6B was performed using Sybr Green detection and miScript primer assays (Qiagen, Valencia, CA). mRNA quantitation was conducted utilizing Sybr Green-based detection technology (Roche Applied Science, Mannheim Germany). Primers for DR420 and 18S21 have been described and for MCM7 were: forward 5′-TCTGGACATGAAGCTCATG; reverse 5′-GTAGT CAGCCAGAGACTCTG. For mRNA degradation studies, cells were treated for 0-6 hours with actinomycin D (5 μg/mL; Sigma, St. Louis, MO) followed by DR4 mRNA and 18S rRNA quantitation by qRT-PCR.

Apoptosis Studies.

Recombinant human TRAIL from R&D Systems (375-TEC; Minneapolis, MN) was used (1-5 ng/mL) to initiate apoptosis. Cells were analyzed by way of fluorescence microscopy for apoptotic nuclear morphology (condensation and fragmentation) after staining with 4′-6-diamidino-2-phenylindole (DAPI; Sigma). For apoptosis studies in cells transfected with miR-25-resistant DR4, green fluorescent protein (GFP)-tagged DR4 (DR4-GFP)22 was cotransfected with pCDNA-mir-25. Control studies included the GFP-N1 empty vector (ClonTech, Mountain View, CA). Only GFP-positive cells were scored for apoptotic nuclear morphology.


Whole cell lysates were obtained and processed for immunoblotting as described in detail.23 Primary antisera were: Death Receptor-4 (DR4; Alexis, San Diego, CA), DR5 and c-FLIP (ProSci, Poway, CA), Bim and caspase-8 (BD Biosciences, San Jose, CA), Mcl-1 and actin (Santa Cruz Biotechnology, Santa Cruz, CA: S19 for Mcl-1 and C-11 for actin).


Cells cultured on collagen-coated glass coverslips were transfected and prepared for immunofluorescence analysis according to the protocol of Xu et al.24 Cells were incubated overnight with 1:1,000 anti-DR4 (10 μg/mL) or anti-DR5 (5 μg/mL) antisera (ProSci), as indicated. Alexa Fluor 488-labeled goat antirabbit immunoglobulin G (IgG) (Invitrogen) was diluted 1:2,000 in blocking buffer. Cells were poststained with DAPI and examined with confocal microscopy (LSM 510, Carl Zeiss, Jena, Germany) and images were analyzed using ImageJ (NIH, Bethesda, MD). At least 23 cells were included in the analysis for each condition.

Luciferase Reporter Assay.

MicroCosm25 was used to predict the miR-25 binding site within DR4. The DR4 3′ untranslated region (UTR) was amplified from human genomic DNA using the following primers: forward 5′-GTCCTTGGAGTGAAAGACTC; reverse 5′-GTAAAGACGGCATTTCACGATG. The 226 basepair product was cloned (TOPO TA) and the SpeI/EcoRV fragment subcloned into the SpeI/NaeI sites of pMir-REPORT (Ambion). The resultant pMirDR4 was mutated by PCR using primers designed to incorporate a 2-base substitution into the miR-25 binding site (5′-CAGGTGTAGTGG converted to 5′-CAGGTGTccTGG). The resulting pMirDR4 constructs were cotransfected with pRL-CMV (Promega). Twenty-four hours after transfection, firefly and Renilla luciferase activities were analyzed using the Dual-Luciferase Reporter Assay System (Promega). Protein content in lysates was measured using the Bradford method.

Statistical Analysis.

At least three replicates were obtained for each condition in experiments with cell cultures. For quantitative continuous data differences between two groups, data were analyzed by Student's t test with a threshold for significance defined at P < 0.05. When multiple comparisons were possible, analysis of variance (ANOVA) coupled with Bonferroni post-hoc correction was used. For microRNA miScript evaluation, significance analysis of microarray (SAM) was applied. All data are presented as mean ± standard error of the mean (SEM).


Hedgehog Signaling Is Involved in Regulation of mir-106b∼25 Expression.

Activation of the Hedgehog signaling pathway is a recently appreciated feature of human cholangiocarcinoma,20, 26 with the potential to regulate microRNA expression.27 We sought to identify additional microRNAs that may be under control of Hedgehog signaling by profiling microRNA expression from KMCH cells that were untreated (vehicle), treated with the Smoothened inhibitor cyclopamine (5 μM), or treated with Sonic Hedgehog ligand (500 nM). Of the 95 microRNAs assayed, we determined that 14 microRNAs were downregulated upon Hedgehog inhibition (Table 1; Supporting Fig. 1). Of the cyclopamine-responsive microRNAs, miR-15b, miR-25, and miR-125a-5p were detected at the highest relative expression levels. Cyclopamine treatment decreased miR-15b expression by 52%, miR-25 by 76%, and miR-125a-5p by 51%.

Table 1. Cyclopamine Suppression of MicroRNA Expression
Gene NameReduction (%)
  1. Inhibition of Hedgehog signaling (cyclopamine 5 μM for 24 hours) caused a reduction in microRNA expression for 14 microRNAs compared to cells treated with recombinant human Sonic Hedgehog. Data are presented as the percent reduction of the microRNA signal compared to Hedgehog-treated cells, such that no reduction would be reported as 0% and complete elimination would be 100% reduction. Only microRNAs with significant alteration (significance analysis of microarrays) are listed.


We selected miR-25 for further analysis based on the magnitude of its repression upon cyclopamine treatment and its importance in tumor biology. Interestingly, miR-106b, which is part of the mir-106b∼25 polycistron and located less than 600 bases from miR-25, was also significantly repressed by cyclopamine (Table 1). Although miR-93, the third member of the mir-106b∼25 cluster at this locus, was not identified by the statistical approach, a similar expression pattern of all three microRNAs is apparent (Fig. 1).

Figure 1.

Mir-106b∼25 expression is supported by Hedgehog signaling. Quantitation of mir-106b∼25 expression in cholangiocarcinoma cells (KMCH) by qRT-PCR of total RNA isolated from cell lysates after 24-hour treatment with vehicle, Sonic Hedgehog (500 nM), or cyclopamine (5 μM). Expression of studied microRNAs was normalized to RNU-U6 expression (delta-delta Ct). Mean ± SEM, n = 3, miR-106b and miR-25 expressions were statistically different based on SAM analysis of Hedgehog versus cyclopamine treatment groups, with a false discovery rate of zero.

Mir-106b∼25 Is Overexpressed in Cholangiocarcinoma.

We next conducted studies to investigate if mir-106b∼25 indeed is involved in cholangiocarcinoma biology. We first employed benign cholangiocytes (H69) and malignant cholangiocarcinoma (Mz-ChA-1, KMCH) cell lines. Quantitation of miR-106b, miR-93, and miR-25 expression demonstrated overexpression of all three microRNAs in cholangiocarcinoma cell lines compared with benign cells (Fig. 2A-C). Among 15 human primary cholangiocarcinoma tumor samples, nine demonstrated an increase in miR-106b expression (Fig. 2D), a slight increase of miR-93 expression was demonstrated in four samples (Fig. 2E), and six samples exhibited increased miR-25 expression (Fig. 2F). Approximately 27% of human samples (4 of 15) had a concurrent increase of all three cluster members. Thus, mir-106b∼25 expression is upregulated in malignant cholangiocarcinoma cell lines and a subset of primary cholangiocarcinomas.

Figure 2.

Increased expression of mir-106b∼25 in cholangiocarcinoma cell lines and human tumor samples. (A) miR-106b expression was measured by qRT-PCT from total RNA isolated from lysates of the indicated cell lines. Results are presented as relative expression using Z30 as an internal control and employing the delta-delta Ct method (mean ± SEM; n = 3; *P < 0.05, **P < 0.01). (B) miR-93 expression from indicated cell lines, presented as above (***P < 0.001). (C) miR-25 expression from the indicated cell lines, presented as above. (D) Similarly, miR-106b was measured by qRT-PCR of total RNA isolated from frozen human tumor samples. Relative expression was determined (delta-delta Ct) compared to Z30 (internal control) and results are presented as fold change (log2) compared to the average expression measured in four nonmalignant human liver samples, and sorted based on expression. Sample numbers indicate the patients from which each was obtained. (E) miR-93 expression from human tumor cells, presented as above. (F) miR-25 expression from human tumors, presented as above.

To determine if increased expression of this cluster was due to elevated host gene RNA levels, we measured MCM7 mRNA in H69, KMCH, and Mz-ChA-1 cells. There was significantly more MCM7 in KMCH compared to H69 cells, whereas Mz-ChA-1 cells expressed similar amounts to H69 (Fig. 3A). In the primary human tumor samples, we analyzed 10 samples for MCM7 mRNA levels, divided equally as five samples that had increased miR-25 levels (#69, 75, 29, 65, and 87) and five with low miR-25 levels (#85, 63, 73, 49, and 39). The amount of MCM7 in the tumor samples that had high miR-25 was nearly 3-fold higher than those with low miR-25 (Fig. 3B). These data seem to indicate that increased host gene expression may explain, at least in part, the elevation of these microRNAs.

Figure 3.

MCM7 mRNA expression is increased in cholangiocarcinoma cell lines and human samples with high expression of miR-25. (A) Quantitation by qRT-PCR of MCM7 mRNA in whole cell lysates from nonmalignant H69 cells compared with malignant KMCH and Mz-ChA-1 cells. Relative expression was determined (delta-delta Ct) compared to 18S (internal control), mean ± SEM; n = 3; *P < 0.05. (B) MCM7 mRNA levels were similarly quantified and analyzed in human cholangiocarcinoma samples comparing groups with high versus low miR-25 expression.

TRAIL-Mediated Apoptosis Is Attenuated by miR-25.

To define the effect of miR-25 on apoptosis, we first employed the nonmalignant H69 cells that had relatively lower miR-25 expression. Enforced expression of miR-25 protected these cells from TRAIL-induced apoptosis by >40% compared to control RNA (Fig. 4A). Next, we investigated the effect of miR-25 antagonism in KMCH cells, which normally express miR-25 at elevated levels. Upon antagonism of miR-25 with LNA-25, KMCH cells demonstrated at least a 25% increase in apoptosis compared to control-transfected cells (Fig. 4B). Therefore, we demonstrated that increased miR-25 protected, and, conversely, decreased miR-25 sensitized cholangiocarcinoma cell lines to TRAIL-induced apoptosis.

Figure 4.

MiR-25 protects cells from TRAIL-induced apoptosis. (A) Nonmalignant H69 cells were transfected with control RNA or mir-25 precursor as indicated to induce miR-25 overexpression. Eighteen hours after transfection, TRAIL was added at 4 ng/mL for an additional 6 hours to induce apoptosis. Cells were stained with DAPI and evaluated based on nuclear morphology for presence of apoptosis. Data are presented as the percentage of cells with apoptotic nuclear morphology out of total cell count and as mean of at least three experiments (mean ± SEM, ***P < 0.001). At least 100 cells were counted in each experiment. (B) In parallel, KMCH cells were transfected with control LNA (CT) or miR-25 antagonistic LNA (LNA-25). Two days after transfection, TRAIL at 4 ng/mL was added where indicated for an additional 6 hours. Cells were analyzed with DAPI for percent of apoptotic nuclei, as above (mean ± SEM, **P < 0.01).

MiR-25 Suppresses TRAIL Death Receptor.

TRAIL Death Receptors DR4 and DR5, caspase-8, cellular FADD-like interleukin-1β-converting enzyme-inhibitory protein (c-FLIP), Mcl-1, and Bim all participate in the cell death pathway triggered in cholangiocarcinoma by TRAIL. Moreover, DR4, Bim, and Mcl-1 were identified by computational analysis as predicted targets for miR-25. We sought to identify the functional miR-25 targets that influence TRAIL signaling in cholangiocarcinoma. Quantitation of miR-25 expression after transfection with mir-25 precursor confirmed the overexpression paradigm (Fig. 5A). Similarly, antagonism of endogenous miR-25 was demonstrated in KMCH cells transfected with LNA targeting miR-25 (LNA-25; Fig. 5B). Cellular DR4, but not DR5, Bim, Mcl-1, Caspase-8, or c-FLIP protein expression was efficiently reduced by miR-25 overexpression in KMCH cells (Fig. 5C). Because DR4 is a novel miR-25 target, we verified this interaction in H69 and Mz-ChA-1 cells (Fig. 5D). Conversely, antagonism of miR-25 caused increased DR4 protein expression in KMCH cells. Specificity of LNA-25 was confirmed by transfection of the miR-106b antagonist (LNA-106), which had no effect on DR4 protein levels (Fig. 5E). DR4 protein in each of these cell lines was readily detected, and found to be more abundant in Mz-ChA-1 cells (Fig. 5F).

Figure 5.

Altered miR-25 reciprocally affects DR4 protein expression. (A) KMCH cells were transfected with mir-25 precursor for 24 hours. Extracted total RNA was analyzed by qRT-PCR to confirm augmented miR-25 expression. Mean ± SEM; n = 3, **P < 0.01. (B) KMCH cells were transfected (48 hours) with locked nucleic acid antagonists to miR-25 (LNA-25), miR-106b (LNA-106b), or control LNA (CT). Total RNA was then subjected to qRT-PCR for miR-25 (left) or miR-106b (right) as well as Z30 (internal control). Data are presented as relative expression (delta-delta Ct compared to Z30). Mean ± SEM; n = 3, *P < 0.05, **P < 0.01. (C) KMCH total lysates were examined by immunoblot for the indicated polypeptides after cell transfection (24 hours) with control (CT) or mir-25 precursor. Actin was used as a loading control. Apparent molecular weight indicated in kDa. (D) Similarly, Mz-ChA-1 or H69 cells were transfected with control or mir-25 precursor and total cellular protein was analyzed by immunoblot for DR4 and actin. (E) KMCH cells, where indicated, were transfected (48 hours) with control LNA (CT), LNA-25, or LNA–106b as a specificity control. Whole cell lysates were analyzed for the indicated polypeptides by immunoblot. (F) Total cellular protein was isolated from H69, KMCH, and Mz-ChA-1 cells and analyzed for DR4 protein by immunoblot with actin as a loading control.

Antagonism of miR-25 also increased DR4 protein levels when examined by immunofluorescence staining of KMCH cells. Specifically, cells transfected with LNA-25 demonstrated a significant increase in the intensity of DR4 immunostaining, an effect not seen for DR5 (Fig. 6A). Quantitation of the fluorescent signal revealed a greater than 2-fold increase in DR4 immunoreactivity in LNA-25 transfected cells versus control LNA (CT; Fig. 6B). Conversely, up-regulation of miR-25 in H69 cells resulted in significantly decreased DR4 immunoreactivity (Fig. 6C,D). This decrease is consistent with immunoblot data in H69 cells (Fig. 5D). These findings confirmed that DR4 is a target protein for regulation of the TRAIL-induced death pathway by miR-25.

Figure 6.

DR4 Repression by miR-25 and functional rescue by way of miR-25-resistant DR4. (A) After transfection of KMCH cells with control locked nucleic acid (CT) or LNA against miR-25 (LNA-25) for 48 hours, slides were prepared and analyzed by confocal microscopy for immunofluorescence intensity of DR4 or DR5. (B) Fluorescence intensity of cells transfected and stained as in (A) was quantified by ImageJ software. The data are presented as fold change in the average DR4 or DR5 intensity, mean ± SEM, ***P < 0.001. (C) H69 cells were transfected (24 hours) with control (CT) or mir-25 precursor; slides were prepared and examined for DR4 immunofluorescence by confocal microscopy. (D) DR4 fluorescence intensity in H69 cells was quantified by ImageJ software and presented as fold change in average signal intensity between control and treatment groups, mean ± SEM, ***P < 0.001. (E) H69 cells were transfected with GFP plus pCDNA, GFP plus pCDNA-mir-25, or DR4-GFP plus pCDNA-mir-25. Eighteen hours after transfection, TRAIL was added at 4 ng/mL for an additional 6 hours to induce apoptosis. Cells were stained with DAPI and GFP-positive cells were evaluated based on nuclear morphology for the presence of apoptosis. Data are presented as the percentage of cells with apoptotic nuclear morphology out of total cell count (mean ± SEM, **P < 0.01, ***P < 0.001).

Enforced DR4 Expression Restored TRAIL Sensitivity.

H69 cells transfected with an miR-25 expression vector (pCDNA-mir-25) demonstrated resistance to TRAIL-induced apoptosis, as anticipated and consistent with Fig. 4A above. Importantly, a DR4 expression plasmid that lacked the DR4 3′UTR (and thus lacked the putative miR-25 binding site) led to restoration of sensitivity to TRAIL-induced apoptosis in TRAIL-treated cells (Fig. 6E). The effect of miR-25 to repress DR4 protein levels and protect cells from TRAIL-induced apoptosis is consistent with previous findings that DR4 is important in cholangiocarcinoma cells. Short-hairpin RNA against DR4 protected KMCH cells from TRAIL-induced apoptosis, whereas transient transfection with a DR4-expression vector sensitized these cells to TRAIL.20

Direct Targeting of DR4 3′UTR by miR-25.

A proposed mechanism of microRNA regulation of protein expression is by way of binding of the microRNA to a region within a cognate mRNA, resulting in repression by either mRNA degradation or translational inhibition. The steady-state amount of DR4 mRNA in H69, KMCH, and Mz-ChA-1 cells did not indicate that cell lines with elevated miR-25 (KMCH and MzChA-1) had depressed DR4 mRNA levels (Fig. 7A). To further assess a potential inverse correlation, we measured DR4 mRNA in patient-derived cholangiocarcinoma samples stratified into high (n = 5) or low (n = 5) miR-25. Again, elevated miR-25 was not strictly correlated with low DR4 mRNA (Fig. 7B). Finally, increased miR-25 expression did not accelerate DR4 mRNA degradation compared to control. Six hours after the addition of the transcriptional inhibitor actinomycin D, miR-25 transfected H69 cells expressed 73.2% of DR4 compared to time zero, whereas control cells expressed 44.1% (data not shown). Conversely, KMCH cells transfected with the miR-25 antagonist, LNA-25, showed only a slight increase in DR4 mRNA stability, with DR4 levels of 13.7% of time zero and control LNA-transfected cells had 7.1% of DR4 after 6 hours of actinomycin treatment (not shown). Collectively, we interpret these data to support repression of DR4 protein expression through translational inhibition, rather than mRNA degradation.

Figure 7.

MiR-25 targets the DR4 3′UTR to mediate gene expression. (A) Total RNA from H69, KMCH, and Mz-ChA-1 cell lines were analyzed for DR4 mRNA expression by qRT-PCR. Relative expression was determined (delta-delta Ct) compared to 18S (internal control), mean ± SEM; n = 3; *P < 0.05. (B) Human tumor samples were grouped based on the level of miR-25 expression (high versus low) and analyzed for DR4 mRNA expression by qRT-PCR. Data are presented as relative expression with 18S used as an internal control. (C) Schematic presentation of the empty parental Luciferase reporter (pMir), the luciferase construct containing the full-length DR4 3′UTR (pMirDR4), and a two-base binding site mutant (pMirDR4-25mut). (D) KMCH cells were transiently transfected with pMir, pMirDR4, or pMirDR4-25mut, as indicated. Cells were also cotransfected with either empty control vector (pCDNA) or with a miR-25 expression vector (pCDNA-mir-25). Twenty-four hours after transfection, cells were lysed and luciferase activity measured. The activity was normalized to the total protein content, and expressed as fold change relative to the corresponding pCDNA value (set at 1.0), mean ± SEM. ** P < 0.01; n.s. = not significant.

To determine if the repression of DR4 is the result of direct binding to the DR4 3′UTR, we used a luciferase reporter assay employing the empty reporter (pMir), the reporter fused to the full-length DR4 3′UTR (pMirDR4), or the DR4 reporter with a two-base mutation at the miR-25 binding site (pMirDR4-25mut; Fig. 7C). Compared to the parental vector, pMirDR4 exhibited inhibition of luciferase expression upon enforced miR-25 expression. Mutation of the miR-25 binding site partially mitigated repression (Fig. 7D). Therefore, our data are consistent with the hypothesis that miR-25 repressed DR4 protein levels by way of direct targeting of the 3′UTR of DR4.


We and others have begun to define the altered microRNA landscape in cholangiocarcinoma.28-32 The main findings of this project identified an important molecular mechanism of apoptosis regulation by the oncogenic miR-25. First, our study demonstrated that members of the mir-106b∼25 cluster are upregulated in cholangiocarcinoma cell lines, KMCH and Mz-ChA-1, and a subset of human tumors. Second, we provide evidence that miR-25 serves as an antiapoptotic microRNA in these cells, protecting against TRAIL-induced apoptosis. Finally, we identified a novel target of miR-25 in the extrinsic pathway of cell death, DR4. Antagonism of miR-25 caused increased DR4 protein expression and sensitized cells to apoptosis. Combined, these data support a functional role for elevated miR-25 expression in apoptosis regulation of cholangiocarcinoma. Each of these findings will be discussed below.

MicroRNA gene sequences are often located in clusters where multiple precursor sequences can be transcribed on the same primary transcript. A substantial portion of microRNA coding sequences lie within a host gene, commonly in an intron. For miR-25, both of these situations apply, as mir-106b∼25 is in an intron of MCM7, located at the fragile site FRA7F on chromosome 7. Our data demonstrate increased expression of all three microRNAs from this locus in cholangiocarcinoma cells and some tumor samples. It is anticipated that transcription of this microRNA cluster is linked to that of the host gene, MCM7, as has been shown in other cell types.33, 34 Indeed, the current investigation revealed that miR-25 RNA and MCM7 levels are correlated in H69 and KMCH (although not MzChA-1). Also, in tumor samples with high miR-25, MCM7 was also elevated. These initial data may indicate that miR-25 is cotranscribed with the MCM7 mRNA; however, additional contributions from altered processing and degradation cannot be ruled out.

The fact that some, but not all, human cholangiocarcinoma samples had increased miR-25 expression may indicate that a particular genetic predisposition, tumor microenvironment, or perhaps differences in underlying patho- or oncogenesis might determine miR-25 dysregulation. Hedgehog reactivation in cholangiocarcinoma and liver injury is emerging as an important pathologic signal.35 Interestingly, in a subset of medulloblastomas constitutive activation of Sonic Hedgehog was reported to be associated with overexpression of the microRNA cluster mir-17∼92, a paralog of mir-106b∼25.36 We present initial findings consistent with stimulation of miR-106b, miR-93, and miR-25 by the oncogenic Hedgehog pathway. Interestingly, we did not observe an increase in microRNA expression upon treatment with the recombinant Sonic Hedgehog ligand, which may indicate preexisting pathway activation, for instance through autocrine expression of Hedgehog ligands.20 It is anticipated that additional signals may be discovered that also increase miR-25 levels in cholangiocarcinoma, possibly including host gene stimulation.37 Future studies may also investigate a potential link between clinical parameters (intrahepatic versus extrahepatic; localized versus advanced) and miR-25 expression.

Consistent with reports of other tumor types, we found that miR-25 had an antiapoptotic role in cholangiocarcinoma cells, specifically, by protecting from TRAIL-induced cell death. This observation was reinforced by experiments with an miR-25 antagonist, LNA-25, which lead to apoptosis stimulation in our study. Evasion of TRAIL-induced apoptosis would presumably exempt tumor cells from immunologic surveillance. As previously described, this death ligand is expressed in human cholangiocarcinoma samples, likely in response to interferon-γ within the tumor microenvironment.17 Thus, resistance to TRAIL-induced apoptosis observed in cholangiocarcinoma cell lines may be necessary for the establishment and maintenance of this cancer. Our data implicate a role for miR-25 overexpression in the resistance to apoptosis.

TRAIL apoptotic signaling in cholangiocarcinoma includes induction of the mitochondrial pathway of apoptosis. Mitochondrial dysfunction can be initiated by the BH3-containing protein Bim,38 and recently Bim was described as a target of miR-25 in esophageal neoplasia.34 Thus, we initially considered that miR-25 likely exerted its protective effect in cholangiocarcinoma cells through repression of Bim, thus decreasing the intracellular apoptotic signal. Surprisingly, we did not find that miR-25 suppressed Bim protein levels or that antagonism of miR-25 increased Bim expression in cholangiocarcinoma cells. We conclude that microRNA function is likely cell-type-specific, and may be regulated by as yet unknown conditions.

Because Bim protein levels were not reduced by miR-25 (nor induced by LNA-25), we searched for alternative targets. The DR4 mRNA has a predicted miR-25 binding site within the 3′UTR, and, indeed, experiments with miR-25 overexpression or inhibition induced reciprocal effects on DR4 protein levels. The repression of DR4 was specific, in that multiple other participants in the TRAIL death signaling pathway showed no change in response to miR-25 manipulation. Additionally, immunofluorescence studies of DR4 and DR5 demonstrated that only DR4 immunoreactivity was increased upon miR-25 antagonism. Finally, direct targeting of the DR4 3′UTR by miR-25 was identified by a luciferase reporter assay. Luciferase experiments demonstrated that miR-25 repression of the signal was dependent on the presence of the DR4 3′UTR, and mutation of the putative miR-25 binding site led to restoration of luciferase expression. We note that this restoration may be partial, as careful inspection of the data reveals that the mutant construct displayed a slight decrease in signal upon enforced miR-25 expression, although this effect was not statistically significant. This observation might indicate the presence of additional, cryptic miR-25 binding sites.

DR4 protein expression includes several layers of regulation, e.g., epigenetic silencing, loss-of-function mutations, and dysregulation of cellular localization.39 Here we identified an additional mechanism controlling DR4 protein levels in cholangiocarcinoma cells, namely, miR-25 repression of DR4. Possibly due to one or more of these alternative DR4 regulatory pathways, the DR4 protein signal in untreated cell lines in vitro is not inversely correlated with miR-25 expression. This lack of correlation cannot be taken as evidence that miR-25 does not regulate DR4, especially given the reciprocal effects of increased or decreased miR-25 on DR4 protein expression. We demonstrated that cells can be resensitized to TRAIL-induced apoptosis upon restoration of DR4 protein in miR-25 transfected cells. It is likely that resting amounts of DR4 protein are determined by multiple mechanisms.

The finding that miR-25 targets DR4, coupled with the implication of Hedgehog as a mediator of increased miR-25 levels (Fig. 8), is consistent with the observation that Hedgehog inhibition both increased DR4 protein and sensitized to apoptosis.20 The findings reported in the previous work also demonstrated direct transcriptional repression of DR4, such that Hedgehog inhibition led to increased DR4 mRNA. The current study builds on these observations, supporting a dynamic and complex inhibition of DR4 expression at the mRNA and protein levels.

Figure 8.

Regulation of apoptosis by miR-25 overexpression. Schematic diagram illustrating the role of miR-25 in cholangiocarcinoma TRAIL death-receptor signaling. (A) Increased miR-25 expression, driven in part by Hedgehog signaling, targets DR4 for repression, which results in TRAIL resistance in tumor cells, permitting cell survival. (B) Antagonism of miR-25 by way of LNA-25 permits elevated DR4 protein levels, and thus increased tumor cell death.

In conclusion, overexpression of miR-25 in cholangiocarcinoma was observed to negatively regulate apoptosis signaling and DR4 protein levels. Thus, this microRNA may serve as an attractive candidate for suppression to facilitate increased tumor cell death during cytotoxic therapy of this and other cancers.