Integrated analysis of miRNA and mRNA expression in childhood medulloblastoma compared with neural stem cells

Abstract

Medulloblastoma (MB) is the most common malignant brain tumor in children and a leading cause of cancer-related mortality and morbidity. Several molecular sub-types of MB have been identified, suggesting they may arise from distinct cells of origin. Data from animal models indicate that some MB sub-types arise from multipotent cerebellar neural stem cells (NSCs). Hence, microRNA (miRNA) expression profiles of primary MB samples were compared to CD133+ NSCs, aiming to identify deregulated miRNAs involved in MB pathogenesis. Expression profiling of 662 miRNAs in primary MB specimens, MB cell lines, and human CD133+ NSCs and CD133- neural progenitor cells was performed by qRT-PCR. Clustering analysis identified two distinct sub-types of MB primary specimens, reminiscent of sub-types obtained from their mRNA profiles. 21 significantly up-regulated and 12 significantly down-regulated miRNAs were identified in MB primary specimens relative to CD133+ NSCs (p<0.01). The majority of up-regulated miRNAs mapped to chromosomal regions 14q32 and 17q. Integration of the predicted targets of deregulated miRNAs with mRNA expression data from the same specimens revealed enrichment of pathways regulating neuronal migration, nervous system development and cell proliferation. Transient over-expression of a down-regulated miRNA, miR-935, resulted in significant down-regulation of three of the seven predicted miR-935 target genes at the mRNA level in a MB cell line, confirming the validity of this approach. This study represents the first integrated analysis of MB miRNA and mRNA expression profiles and is the first to compare MB miRNA expression profiles to those of CD133+ NSCs. We identified several differentially expressed miRNAs that potentially target networks of genes and signaling pathways that may be involved in the transformation of normal NSCs to brain tumor stem cells. Based on this integrative approach, our data provide an important platform for future investigations aimed at characterizing the role of specific miRNAs in MB pathogenesis.

Figure 1. Clustering analysis of primary MB, MB cell lines and CD133+ NSCs and CD133− NPCs.

(A) Unsupervised hierarchical clustering and (B) PCA. The analyses were based on normalized expression data for 662 miRNAs in ten primary MB specimens, three MB cell lines (PER-547, PER-568 and PER-621), two CD133+ NSC and two CD133− NPC samples. Sample numbers refer to those in Table 1. The similarity metric utilized for unsupervised hierarchical clustering was Pearson's correlation (r). MB sub-typing determined from mRNA gene expression analysis is indicated. Sub-type B is characterized by over-active SHH signaling, whilst sub-type C is characterized by the enriched expression of genes associated with neuronal differentiation. Sub-type E is characterized by enriched expression of photoreceptor genes, whilst sub-type D is characterized by mixed neuronal and photoreceptor genes. Primary MB specimens defined as “NA” were not available for sub-typing analysis. The M1 primary sample failed on the Pool B TLDA card and was therefore excluded from PCA, as this type of analysis cannot be performed on incomplete datasets. Primary MB sample, M15, grouped with cluster three, as compared to cluster four in Figure 1A.

Figure 2. Heatmap analysis based on deregulated miRNAs in MB primary specimens and cell lines compared to CD133+ NSCs.

Sample numbers refer to those in Table 1. The similarity metric utilized for this analysis was log₂(2^−ΔCt) transformed miRNA values obtained from qRT-PCR profiling analysis. miRNA expression in normal CD133+ NSCs was determined from averaging log₂(2^−ΔCt) transformed miRNA values of CD133+ NSCs from both hES3 and MEL1 ESC lines. A red to green color scale (−20 to +2) depicts normalized miRNA expression on a log scale, with median expression across all samples represented as black. White spaces in the heatmap are due to M1 primary sample failing on Pool B TLDA card.

Figure 3. Validation of putative miR-935 targets in PER-547 cells.

Quantitative RT-PCR analysis of (A) KIAA0232 (B) SLC5A3 (C) TBC1D9 and (D) ZFAND6 expression levels in PER-547 cells 24 h after transfection with scrambled negative control or pre-miR-935. Target gene expression levels in pre-miR-935-transfected samples were normalized to scrambled control levels. p values were obtained from a two way ANOVA on log transformed expression data. Error bars represent mean +/− SEM from six independent experiments.