miR-34a confers chemosensitivity through modulation of MAGE-A and p53 in medulloblastoma

Abstract

Recent studies have established miR-34a as a key effector of the p53 signaling pathway and have implicated its role in multiple cancer types. Here, we establish that miR-34a induces apoptosis, G2 arrest, and senescence in medulloblastoma and renders these cells more sensitive to chemotherapeutic agents. These effects are mediated in part by the direct post-transcriptional repression of the oncogenic MAGE-A gene family. We demonstrate that miR-34a directly targets the 3′ untranslated regions of MAGE-A genes and decreases MAGE-A protein levels. This decrease in MAGE-A results in a concomitant increase in p53 and its associated transcriptional targets, p21/WAF1/CIP1 and, importantly, miR-34a. This establishes a positive feedback mechanism where miR-34a is not only induced by p53 but increases p53 mRNA and protein levels through the modulation of MAGE-A genes. Additionally, the forced expression of miR-34a or the knockdown of MAGE-A genes by small interfering RNA similarly sensitizes medulloblastoma cells to several classes of chemotherapeutic agents, including mitomycin C and cisplatin. Finally, the analysis of mRNA and micro-RNA transcriptional profiles of a series of primary medulloblastomas identifies a subset of tumors with low miR-34a expression and correspondingly high MAGE-A expression, suggesting the coordinate regulation of these genes. Our work establishes a role for miR-34a in modulating responsiveness to chemotherapy in medulloblastoma and presents a novel positive feedback mechanism involving miR-34a and p53, via direct targeting of MAGE-A.

Medulloblastomas are the most common malignant neoplasms of the central nervous system in children, accounting for ∼15%–20% of all pediatric brain tumors.^1–3 Despite multimodal treatment with chemotherapy and radiation, only 50%–60% of medulloblastoma patients reach complete remission by 5 years.⁴ The molecular mechanisms that confer sensitivity or resistance of tumors to these treatments are still unclear.

Recent evidence has implicated micro-RNAs (miRNAs) in modulating chemosensitivity in cancer.^5–7 miRNA expression profiling in multiple cancer types including medulloblastoma has revealed the disproportionate expression of several miRNAs in tumors compared with corresponding normal tissues, suggesting their involvement in tumorigenesis and progression.^8–10 One such miRNA is miR-34a, which maps to 1p36, a chromosomal region commonly lost in cancer and in a subset of medulloblastomas. Several studies have independently established miR-34a as a direct transcriptional target of p53 via a consensus binding site located in the miR-34a promoter region.^11–13 Upon transactivation by p53, miR-34a contributes the tumor suppressor function by subverting the expression of a number of target transcripts such as CDK4, CDK6 CCND1, SIRT1, Cyclin E2, E2F3, MYCN, and BCL2.^14–18

Low expression of miR-34a has been observed in neuroblastoma, pancreatic cancer, prostate cancer, lung cancer, malignant lymphoma, retinoblastoma, and colon cancer.^{6,15,17,19,20} Additionally, miR-34a has been shown to enhance sensitivity to chemotherapeutic agents in prostate cancer, gastric cancer, and leukemia cell lines.^6,7,21,22

In this study, we show that miR-34a is variably expressed in primary medulloblastoma tumors and cell lines, and expression is positively correlated with responsiveness to chemotherapeutic agents such as mitomycin C (MMC) and cisplatin. Furthermore, we identify a subset of MAGE-A genes as direct targets of inhibition by miR-34a. MAGE-A genes belong to the larger class of cancer testes antigens, which are normally expressed in the cells of the male germ line with very little expression in somatic cells.²³ However, MAGE genes are aberrantly expressed in many malignancies, including melanoma, breast cancer, lung cancer, esophageal cancer, and medulloblastoma.^24,25 MAGE proteins elicit autologous T-cell–mediated immune responses, making these antigens an attractive focus of immunotherapy.^26,27 Kasuga et al.²⁴ previously reported that MAGE-A members are expressed in over 60% of medulloblastoma cell lines and primary tumors, and downregulation of MAGE genes mediated by small interfering RNA (siRNA) resulted in sensitization to cisplatin and etoposide. We now show that a similar sensitization to chemotherapy is mediated by the direct targeting of the MAGE-A genes by miR-34a. Furthermore, we show that the repression of MAGE-A by miR-34a results in increased expression of p53, establishing for the first time a mechanistic relationship between miR-34a and the MAGE-A genes in modulating p53. We propose a novel positive feedback loop between miR-34a and p53 via the MAGE-A family of genes, which modulates the anticancer drug response.

Materials and Methods

Primary Tumors, RNA Extraction, and Microarray Analysis

Snap-frozen primary medulloblastoma samples were obtained from Children's Hospital Boston and the Co-operative Human Tissue Network. The tumor samples were collected with prior informed consent in accordance with Institutional Review Board protocols. Diagnosis was verified by an on-site neuropathology review. Total RNAs were isolated using Trizol reagent (Invitrogen) as per the manufacturer's instructions. Gene expression data were generated by hybridizing-labeled RNAs to Affymetrix HT HG-U133A arrays. Data were preprocessed using the Robust Multichip Average algorithm, and heat maps were generated using the GenePattern software package (www.genepattern.org).

Cell Culture

Medulloblastoma cell lines R262, UW228, UW426, and R300 were supplied by Dr Мichael Bobola, University of Washington, Seattle. D556, D384Med, and D425Med were obtained from Duke University. The D283 medulloblastoma cell line and HEK 293T/17 cell line were purchased from American Type Culture Collection. Cells were maintained in Dulbecco's modified Eagle's medium with the F-12 nutrient mixture (Invitrogen) supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 1% penicillin, and streptomycin (all from Gibco) at 37°C in a humidified chamber of 5% CO₂.

Transfections

Pre-miR-34a miRNA precursor, anti-miR-34a miRNA inhibitor, and respective scrambled control miRNAs were obtained from Ambion. The miRNA precursor and the inhibitor were used at concentrations of 10 and 30–50 nM, respectively. An siRNA duplex targeting MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 (Dharmacon) was designed as described previously²⁴ and used for transient transfections at a concentration of 30 nM. A total of 2 × 10⁶ cells were transiently transfected by nucleofection (Lonza) using program T-030 or reverse transfected with siPORT NeoFX (Applied Biosystems) according to the manufacturer's protocols. Cells were harvested 48 or 72 hours post-transfection.

Luciferase Assays

293T cells were transfected with Lipofectamine 2000 (Invitrogen) in 96-well plates according to the manufacturer's protocol. Briefly, 5 × 10³ cells/well were transfected with 50 ng of firefly luciferase control plasmid and 200 ng of Renilla luciferase WT or Δ plasmids with 30 nM pre-miR-34a miRNA precursor or scrambled control miRNA duplex. Firefly and Renilla luciferase activities were measured 24 hours after transfection by a dual-luciferase assay (Promega) using an LB9507 luminometer according to the manufacturer's instructions. Renilla activity was normalized to firefly activity to control for transfection efficiency.

RNA Extraction

Total RNA from primary tumor samples was extracted using Trizol reagent (Invitrogen). Cell line RNA was isolated using the RNeasy Mini Kit (Qiagen). RNA samples enriched for small RNA molecules were further enriched in the samples by using the Nirvana RNA Isolation Kit (Ambion) according to the manufacturer's protocol. RNA concentration was measured using a NanoDrop Spectrophotometer (NanoDrop Technologies).

Real-time Quantitative Reverse Transcription–PCR

The miRNA sequence-specific reverse transcription–PCR (RT–PCR) primers for miR-34a and endogenous control RNU6B were purchased from Applied Biosystems. TaqMan Gene Expression assays for P53, CDKN1A, B2M, and MAGE were also purchased from Applied Biosystems. Analysis was carried out using the Applied Biosystems 7300 q-RT-PCR System. The gene expression delta cycle threshold (ΔCT) values of miRNAs in each sample were calculated by normalizing with the internal control RNU6B, and relative quantification values were plotted. All reactions were done in triplicate, and at least 3 independent experiments were performed to generate each data set.

Cell Viability Assays: MTS Assay

Transfected cells were seeded at 4000 cells per well in 96-well plates (Corning) in 75 μL of an antibiotic-free medium. Each sample was plated in quintuplicate. MMC (Calbiochem) and cis-diamineplatinum (II) dichloride (cisplatin; Sigma) were added 48 hours postplating at 4× concentrations made up to 25 µL in an antibiotic-free medium. Cell viability was determined using the 3-(4,5-dimethylthiazol-2-yl)- 5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt (MTS) assay with CellTiter 96 AQueous One Solution Reagent (Promega) following the manufacturer's protocol. After incubating at 37°C for 1 hour, the optical density of each well was measured with a spectrophotometric microplate reader (Bio-Rad) at 490 nm.

Colony-Forming Assays

Transfected cells were seeded as single-cell suspension at 500 cells per 10 cm cell culture dish (Corning) in 6 mL of an antibiotic-free medium. Twenty-four hours later, cells were exposed to MMC and cisplatin at the appropriate concentrations. After 14 days, the media was aspirated and dishes were washed with phosphate-buffered saline (PBS) and fixed using 4% paraformaldehyde (Electron Microscopy Sciences). Plates were subsequently stained with crystal violet solution (Sigma) and gently washed with water. Colonies consisting of more than 50 cells were counted under a light microscope. Plates were air-dried and photographed using a Gene Genius Bio Imaging System. Dose–response curves were generated by combining data from 3 independent experiments in which each condition was plated in triplicate.

Immunocytochemistry

All steps were carried out at room temperature as described previously.²⁸ Cells were plated on poly-d-lysine–coated glass coverslips 72 hours post-transfection and were fixed in 4% (w/v) paraformaldehyde (Electron Microscopy Sciences) for 1 hour, washed in PBS (Invitrogen; 6 times, 5 minutes each), incubated for 1 hour in blocking buffer (PBS, 0.8% Triton X-100, 10% normal goat serum), and then incubated overnight in blocking buffer with anti-p53 (1:100) and anti-MAGE (1:100) antibodies. The cells were washed in PBS (6 times, 5 minutes each) before the addition of fluorescent goat anti-rabbit and goat anti-mouse secondary antibodies (Alexa Ig488 [green] or Alexa Ig565 [red], Invitrogen, Molecular Probes) at 1:250 for 60 minutes. The cells were washed in PBS (6 times, 5 minutes each) and visualized with a Zeiss LSM510 META 2-Photon confocal microscope.

Cell Cycle Analysis

Cells were harvested 72 hours post-transfection, washed in PBS, and fixed with 70% ethanol overnight at 4°C. Cells were treated with 100 µg/mL of RNase A (Roche Diagnostics) and 50 µg/mL of propidium iodide (Sigma) for 30 minutes at 37°C in the dark. Propidium iodide fluorescence was monitored with FACSCalibur flow cytometry (BD Biosciences). At least 10 000 cells were collected and analyzed with Cell Quest Pro Software (BD Biosciences) and FlowJo software version 4.6.1 (TreeStar).

Western Blots

Membranes were blocked in PBS-Tween 5% milk and probed with anti-p53 (Abcam), anti-CDKN1A (Cell Signaling Technology) and anti-MAGE (Santa Cruz) antibodies at the manufacturer's recommended concentrations. Anti-β-actin (Abcam) was used as the protein-loading control. The horseradish peroxidase–conjugated secondary antibodies were used at 1:10 000 (Jackson Immunoresearch Laboratories). Blots were developed using the SuperSignal West PICO Chemiluminescent Detection System (Pierce Biotechnology). Digital images were recorded using the Fuji Image 3000 Chemiluminescence Detection System (Fuji Film).

Bioinformatics

Candidate miR-34a targets were analyzed using publicly available algorithms: TargetScan (http://www.targetscan.org/) and miRanda (http://www.microrna.org/).

Apoptosis and Senescence Assays

Apoptosis and senescence assays were performed as described previously.^15,28 For full details, please refer to Supplementary Material.

Statistical Analysis

All graphs and statistical analyses were generated using GraphPad Prism 5 for Windows. For calculating the statistical significant differences between groups, Fisher's exact test and Student's t-test were used.

Results

Differential Expression of miR-34a in Medulloblastoma

To establish the expression of miR-34a in medulloblastoma, we performed quantitative real-time PCR (q-RT-PCR) for miR-34a on a series of human medulloblastoma cell lines and primary tumors. We found 25–60-fold higher miR-34a expression in D384, D283, D341, and D556 cell lines relative to UW228, R300, and UW426 cell lines (Fig. 1A). Accordingly, we found a positive correlation (Fisher's exact test, P < .05) between p53 and miR-34a expression in these cell lines (Fig. 1A and B). This was not surprising, as previous studies had established that p53 is a direct transcriptional activator of miR-34a.^11–13 In primary medulloblastoma samples, we confirmed the variable miR-34a expression and its correlation with p53 (Fig. 1D; Supplementary Material, Fig. S1).

Fig. 1.

miR-34a expression in human medulloblastoma cell lines and primary tumors. (A) q-RT-PCR analysis of miR-34a expression in 8 medulloblastoma cell lines. Raw data were normalized to the endogenous control RNU6B. The errors bars were derived from 3 independent experiments done in triplicate. (B) q-RT-PCR analysis of p53 expression in the same cell lines showing a positive correlation with miR-34a expression. ΔCT values were normalized to the internal control B2M. (C) Dose–response curves of MMC (top panel) and cisplatin (bottom panel) in the human medulloblastoma cell lines as determined by an MTS cytotoxicity assay. Cell viability at each concentration is expressed as a percentage of the untreated controls. Data at each point are indicative of at least 2 independent experiments done in quintuplicate. Error bars represent ±SEM. Resistant and sensitive cell lines are depicted in red and blue, respectively. (D) q-RT-PCR analysis of miR-34a in a panel of 55 primary medulloblastoma tumor samples. Samples are ordered according to the expression level (lowest to highest).

We next investigated the response of medulloblastoma cell lines to pharmacologically relevant concentrations of MMC and cisplatin. Values of half maximal inhibitory concentration IC₅₀ calculated from the dose–response curves show that UW426 and R300 are more resistant to these chemotherapeutic agents than are D556, D283, and D384 cells (Fig. 1C). As expected, a strong inverse correlation (Fisher's exact test, P < .05) was observed between resistance to drugs and miR-34a and p53 expression.

MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 3′UTRs Are Direct Targets for miR-34a

To identify potential targets of miR-34a that contribute to the biological effects noted in our cell lines, we utilized several miRNA target prediction algorithms (TargetScan, PicTar, and miRANDA). Each algorithm predicts several members of the MAGE-A family (MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12) as having miR-34a–binding elements in their 3′ untranslated regions (UTRs) (Fig. 2A). Of note, decreased levels of MAGE-A have been previously associated with attenuated chemoresistance in cancer cells and medulloblastoma in particular.²⁴ Furthermore, several reports suggest that increased MAGE expression enhances cancer cell survival.^29–33

Fig. 2.

MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 are the direct targets of miR-34a. (A) Schema representing the functional interaction between miR-34a and the seed sequence (bold) in the 3′UTR of MAGE-A family members as predicted by TargetScan. (B) Schematic depiction of the Renilla luciferase reporter vectors containing wild-type MAGE-A 3′UTRs (upper panel) and mutated MAGE-A 3′UTRs (lower panel: 2 point mutations inserted into the miR-34a seed sequence). (C) Dual-luciferase assay of 293-T cells cotransfected with reporter constructs and the pre-miR-34a or scrambled control miRNA (*P < 0.05). Renilla luciferase activity was normalized to firefly luciferase activity to control for transfection efficiency. These data represent 3 independent experiments ± SEM. (D) Western blot analysis showing a decrease in the MAGE-A protein level compared with control in R262 cells overexpressing pre-miR-34a.

To confirm MAGE-A genes as direct targets of miR-34a, we cloned each MAGE-A 3′UTR into a Renilla luciferase reporter vector (Fig. 2B). Cotransfection of wild-type MAGE-A 3′UTRs with pre-miR-34a, but not with the scrambled miRNA control, decreases the luciferase activity of all 4 MAGE-A constructs by more than 65%, the exception being MAGE-A2 where there is a ∼40% reduction. In addition, we mutagenized the miR-34a “seed” region in each construct, which rescues luciferase activity in cells cotransfected with miR-34a (Fig. 2C).

We next investigated the effect of miR-34a overexpression on MAGE-A transcript and protein levels in medulloblastoma cells. We did not detect a significant difference in MAGE-A transcripts with the introduction of pre-miR-34a in R262 medulloblastoma cells, which have intermediate basal levels of miR-34a; however, we detected a decrease in the MAGE protein levels by western blot analysis (Fig. 2D). This suggests that the main mechanism of MAGE-A repression by miR-34a is via translational repression. Indeed, recent studies found that ∼50% of miRNA-target interactions are solely at the translational phase.^34,35

miR-34a Enhances Chemosensitivity in Medulloblastoma Cells by Direct Targeting of MAGE-A

To evaluate the functional consequence of miR-34a expression on sensitivity to chemotherapeutic drugs, we transfected medulloblastoma cell lines with pre-miR-34a. As mentioned, the R262 medulloblastoma cell line has intermediate levels of miR-34a and p53 expression and a midrange IC₅₀ value to both MMC and cisplatin, whereas R300 cells represent the drug-resistant phenotype with the low baseline levels of miR-34a and p53 (Fig. 1A–C). Overexpression of miR-34a in both cell lines conferred an increased sensitivity to MMC and cisplatin when compared with the scramble control (Fig. 3A and B). In addition, cell cycle analysis of the transfected cells reveals that miR-34a-overexpressing cells show a significant accumulation in the S-phase and the G2/M checkpoint (Fig. 3C). G2/M cell cycle arrest rendered cells more chemosensitive to DNA-damaging agents, consistent with the results obtained in our cell-survival analyses.^36–39 A moderate but reproducible increase in the sub-G1 fraction indicative of cell death was also observed in the pre-miR-34a–transfected cells. However, no significant G1 accumulation was observed, consistent with previous reports.⁴⁰

Fig. 3.

Forced expression of miR-34a sensitizes medulloblastoma cells to MMC and cisplatin. (A) Dose–response curves of MMC and cisplatin in R262 and R300 cell lines overexpressing pre-miR-34a. Cell viability at each concentration is expressed as a percentage of the untreated controls. Data at each point represent at least 2 independent experiments done in quintuplicate. Bars represent ±SEM. (B) Colony-forming assay of R262 overexpressing pre-miR-34a, or scrambled control, treated with 10 nM MMC. Bar graph represents clonogenicity expressed as a percentage of the respective untreated controls (*P < 0.05). (C) Cell cycle analysis of R262 cells transfected with pre-miR-34a or scrambled control. Bar graph represents the cell cycle distribution of the transfected cells and the control.

To investigate the effect of miR-34a expression on apoptosis and cellular senescence, we labeled miR-34a–transfected medulloblastoma cells with Annexin V and senescence-associated β-galactosidase, respectively. We observe a 4-fold increase in apoptotic cells in the pre-miR-34a-overexpressing cells compared with the control (Supplementary Material, Fig. S2). We also noted a 3-fold increase in senescent cells (Supplementary Material, Fig. S2).

As miR-34a has several predicted target genes, we investigated whether the phenotypic consequences of MAGE-A repression by miR-34a in medulloblastoma cells could be phenocopied by knockdown of MAGE-A transcripts by synthetic siRNAs. An siRNA designed to simultaneously target MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 efficiently downregulated their respective expression by 80%–90% (Supplementary Material, Fig. S3). In addition, the effects of MAGE-A family knockdown on the drug response and clonogenic survival in the R262 cell line were similar to R262 cells overexpressing miR-34a (Fig. 4A and B). Moreover, an analysis of the cell cycle distribution of MAGE-A knockdown cells revealed a similar pattern to miR-34a-overexpressed cells; a higher accumulation of cells in the S- and G2-phases were evident in comparison with the control (Fig. 4C). Finally, cells undergoing apoptosis were more prevalent in the MAGE-A knockdown cells and both the protein and the transcript of p53 and p21 were found to be substantially higher in the MAGE siRNA-transfected cells compared with the control (Fig. 5A and B). All these data taken together provide strong evidence that MAGE-A family knockdown is phenotypically tantamount to pre-miR-34a overexpression.

Fig. 4.

siRNA knockdown of MAGE-A family genes sensitizes medulloblastoma cells to MMC and cisplatin. (A) Dose–response curves of MMC and cisplatin in R262 and R300 cell lines overexpressing MAGE-A siRNA. Cell viability at each concentration is expressed as a percentage of the untreated controls, and data at each point represent at least 2 independent experiments done in quintuplicate. Bars represent ±SEM. (B) Colony-forming assay of R262 after siRNA-mediated knockdown of MAGE-A genes treated with 10 nM MMC compared with the transfected control (*P < 0.05). (C) Cell cycle analysis of R262 after siRNA-mediated knockdown of MAGE-A.

Fig. 5.

Increase in the p53 levels by either miR-34a–mediated or siRNA-mediated targeting of MAGE-A. (A) q-RT-PCR analysis of p53 and p21 expression in R262 cells overexpressing pre-miR-34a (top panel) and MAGE-A siRNA (bottom panel). (B) Western blot analysis for p53 in R262 cells overexpressing pre-miR-34a (left panel) or MAGE-A siRNA (right panel). (C) Cotransfection of MAGE-A family genes lacking the 3’UTR and pre-miR-34a negates the miR-34a–mediated increase in p53 and p21 transcripts. (D) Immunofluorescence of p53 and MAGE-A in R262 cells transfected with scrambled siRNA control, MAGE-A siRNA, scrambled miRNA control, and pre-miR-34a.

miR-34a Modulates p53 Levels via Regulation of MAGE-A in a Positive Feedback Mechanism

We noted that MAGE-A repression by miR-34a resulted in an increase in p53 protein levels as well as an increase in p53 and p21 transcripts (Fig. 5A and B). This establishes that miR-34a, via the regulation of MAGE-A, can modulate its own transcriptional activator, p53, through a positive feedback mechanism. To further confirm this, we ablated miR-34a expression in R262 cells using miR-34a–specific antagomirs. The downregulation of miR-34a resulted in not only an increase in the resistance of R262 cells to MMC and cisplatin (Supplementary Material, Fig. S4), but also a concomitant increase in MAGE-A proteins and transcripts and a decrease in p53 and p21 proteins and transcripts (Supplementary Material, Figs S5 and S6).

To confirm that the increase in p53 and p21 transcripts in the presence of miR-34a was indeed mediated through the direct suppression of MAGE-A genes, we cloned the MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 full-length coding sequences bereft of the 3′UTR, thus rendering these transcripts impervious to targeting by miR-34a. Cotransfection of these constructs with miR-34a completely abolished the increase in p53 and p21 (Fig. 5C).

Finally, we performed immunofluorescence studies of MAGE-A family proteins and show that they are localized predominantly in the nucleus and to a lesser extent in the cytoplasm (Fig. 5D). Throughout the cell, p53 was uniformly localized in a punctuate manner (Fig. 5D). Forced expression of pre-miR-34a by transient transfection resulted in ablation of MAGE-A protein expression and a concomitant increase in p53 signal intensity compared with the scramble control. As expected, siRNA knockdown of the MAGE family of genes almost completely abolished the expression of MAGE proteins and in conjunction increased the intensity of the p53 signal, consistent with our western blot and q-RT-PCR results (Fig. 5D). Thus, immunofluorescence patterns corroborate that miR-34a overexpression and MAGE-A knockdown phenocopy each other and provide further evidence for the positive feedback mechanism involving miR-34a, MAGE-A, and p53.

Discussion

Our study demonstrates for the first time that miR-34a directly targets the MAGE-A family of oncogenes, disengaging p53 from MAGE-A–mediated repression. An important biological consequence of this is a positive feedback loop that sensitizes medulloblastoma cells to chemotherapeutic agents via delayed G2/M progression and increased apoptosis.

Importantly, our data provide an alternate means of p53 pathway dysregulation in medulloblastoma and other cancers, short of genetic inactivation. Although ∼50% of the human solid tumors harbor mutations of the p53 gene, mutations are rare in human medulloblastoma.^41,42 Given the central importance of p53 in regulating apoptosis, senescence, and proliferation, it is logical that the p53 pathway is modulated by alternate mechanisms, such as influences from the “miRnome.” Indeed, loss of miR-34a from deletion of the 1p36 locus in neuroblastoma occurs independently of p53 mutation, suggesting that the loss of miR-34a is an alternate mechanism of p53 pathway inactivation in vivo.^17,43 In addition, similar to p53 loss, decreased expression of miR-34a is associated with chemotherapeutic resistance in several types of cancer, including chronic lymphocytic leukemia, prostate cancer, non-small-cell lung cancer (NSCLC), and gastric carcinoma.^6,7,21,22,44 Since miR-34a has also been shown to target CDK4/6, Cyclin E2, BCL2, and SIRT1, it appears that it mediates tumor suppression by p53 through the collective subversion of genes that promote cell survival.^14,17,18,45

Recent investigations have established that positive feedback and feed-forward circuits are common in miRNA regulatory networks.^46–51 Pathways known to employ such mechanisms include circuits regulating cell cycle control, aging, and apoptosis.^46,48,52,53 Our results show a positive feedback mechanism in which miR-34a positively regulates its primary transcriptional activator, p53 (Fig. 6). This makes logical sense considering that a commitment to programmed cell death should in principle be “all or none.” Thus, by repressing MAGE-A–mediated repression of p53, miR-34a serves to reinforce its own transcription, perpetuating its (and p53’s) function as a central effector of apoptosis.

Fig. 6.

miR-34a, MAGE-A genes, and p53 form a positive feedback loop modulating chemosensitivity in cancer cells. Diagram depicting the proposed novel positive feedback pathway modulating chemosensitivity in medulloblastoma cells. miR-34a–mediated direct targeting of the MAGE-A genes results in an increase in p53, which in turn acts as a transcriptional activator for miR-34a.

MAGE-A2, MAGE-A3, MAGE-A6, and MAGE-A12 have more than 90% homology in their DNA and protein sequences, suggesting their ability to act under different cellular and transcriptional contexts and also their redundancy in function. The ability of miR-34a to directly target these MAGE-A family members negates their functional redundancy and further establishes miR-34a as a central effector of the p53 pathway. The manner in which MAGE-A family proteins regulate p53 has been previously investigated. MAGE-A2 acts as a scaffolding protein for the assembly of p53–histone deacetylase (HDAC) repression complex. Thus, the ablation of MAGE-A2 presumably forestalls HDACs from deacetylating the chromatin around p53-binding sites, ensuring the transactivation of p53.³⁰ MAGE-A family proteins also complex with KAP1, which in turn bind with MDM2 to repress p53 expression, acetylation, and function.^33,54,55 Thus, the direct targeting of MAGE-A transcripts by miR-34a likely has direct consequences on downstream epigenetic mechanisms affecting p53, transcriptionally and post-translationally.

Our study establishes for the first time the functional nexus between miR-34a, MAGE, and p53 in modulating the cellular response to chemotherapeutic drugs. From a clinical standpoint, strategies aimed at taking advantage of this interplay could enhance treatment regimens. Moreover, the restricted expression of MAGE-A genes to tumor tissue makes these antigens ideal candidates for targeted therapies. Indeed, a vaccine based on a recombinant MAGE-A3 peptide is currently being evaluated in a series of clinical trials for melanoma and NSCLC.⁵⁶ A phase II clinical trial of this immunotherapy showed that disease-free survival rates increased by 33% in NSCLC patients and phase III studies are now under way. In melanoma patients, the MAGE-A3 peptide vaccine has shown promise as regression of tumor nodules has been reported in early-phase studies.⁵⁷

Our study also opens up the possibility of direct targeting of MAGE-A genes by RNA interference (RNAi) therapeutics. Indeed, the effect of miR-34a in sensitizing medulloblastomas to chemotherapy in vitro suggests that an anti-MAGE RNAi in combination with chemotherapy could be more effective than using standard agents alone. Further preclinical investigations will need to be performed to assess the feasibility of such an approach.

Authors' Contributions

S.D.W. and Y.-J.C. designed the research study. S.D.W., V.A., N.T., and A.N. performed the experiments. S.D.W., Y.-J.C., V.A., D.K.S., and S.L.P. analyzed the data. S.D.W. and Y.-J.C. wrote the manuscript.

Supplementary Material

Supplementary material is available at Neuro-Oncology Journal online.

Conflict of interest statement. None declared.

Funding

This project received support from US National Institutes of Health grants (R01-CA109467, P30-HD018655) to S.L.P.