Identification of OTX2 as a Medulloblastoma Oncogene Whose Product can be Targeted by All-Trans Retinoic Acid

Introduction

Primary brain tumors are a leading cause of cancer death in children, and medulloblastoma is the most frequent malignant brain tumor in this age group. Despite therapeutic advances, more than one third of children with medulloblastoma die from the disease within 5 years of diagnosis, and the remaining survivors experience significant toxicities from extant therapies (1, 2) . Elucidation of the molecular pathogenesis of medulloblastoma may suggest novel therapeutic targets. Whereas comprehensive screens for activating or inactivating mutations would require sequencing and functional studies of tens of thousands of genes, measurements of the genomic DNA copy number, or gene dosage, of chromosomal segments is far more amenable to analysis. Complete sequencing of the human genome has made possible the development of novel techniques that narrow the resolving power of genome-wide screens to regions covering one or a small handful of genes (≤1 Mb; refs. 3, 4 ). With digital karyotyping and gene expression analysis, we identified genomic amplification and overexpression of OTX2 gene in medulloblastomas. Medulloblastoma cell lines overexpressing OTX2 were growth inhibited by pharmacologic doses of all-trans-retinoic acid (ATRA) and knockdown of OTX2 expression by siRNA.

Materials and Methods

Tissue Samples. Brain tumor cell lines and frozen primary tumor samples were obtained from the Duke University Brain Tumor Center Tissue Bank. Acquisition of tissue specimens was approved by the Duke University Health System Institutional Review Board and was done in accordance with the Health Insurance Portability and Accountability Act of 1996 regulations.

Digital Karyotyping. Digital karyotyping libraries were constructed as previously described (3, 5) . Briefly genetic tags were matched to the human genome, and tag densities were evaluated by using a digital karyotyping software package. Genomic densities were calculated as the ratio of experimental tags to the number of virtual tags present in a fixed window. Sliding windows of 200 virtual tags in size were used to identify regions of increased and decreased genomic density. Chromosomal regions were considered amplified if maximal genomic densities were >6 genome copies per haploid genome. Digital karyotyping protocols and software for extraction and analysis of genomic tags are available at http://www.digitalkaryotyping.org.

Serial Analysis of Gene Expression Data Analysis. The serial analysis of gene expression (SAGE) data were obtained from the National Center for Biotechnology Information Cancer Genome Anatomy Project repository (http://cgap.nci.nih.gov/SAGE). The presence of OTX2 SAGE tags in a total of 157 tumors and 54 normal human tissues was identified by using the SAGE Anatomic Viewer.

Quantitative Real-time PCR. Genomic DNA content differences of OTX2 or C14orf101 between medulloblastoma cells and normal cells were determined by using the quantitative real-time PCR (Q-PCR) method as previously described (5). cDNA from medulloblastoma samples was used to measure the level of OTX2 mRNA expression. cDNA from normal human cerebellum was used as the normal control, and cDNA content was normalized to that of GAPDH.

Cell Proliferation Assay. Cell viability was determined by the 3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide MTT assay.

siRNA-Mediated “Knockdown” of OTX2 Expression. To target OTX2, siRNA#1 GGAGGUGGCACUGAAAAUCtt and siRNA#2 GGACACUAAUUCAUCUGUAtt (Ambion, Austin, TX) were generated using OTX2 sequences. Cells were collected at 48 hours after siRNA transfection to generate cDNA for Q-PCR. After three doubling times, the cells were assayed for viability by the 3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide MTT assay.

Colony Formation in Soft Agar Assay. The cells were plated at 5 × 10³ in triplicates in 0.5% agarose-coated 24-well plates with or without ATRA (Sigma, St. Louis, MO). After 2 weeks, the number of colonies was counted.

Apoptosis. Cell apoptosis induced by ATRA was determined by the Cell Death Detection ELISA^plus method (Roche Diagnostic Co., Indianapolis, IN).

Results and Discussion

We initially used digital karyotyping to analyze quantitatively the genomic elements at high resolution in the medulloblastoma cell line D458MED (3). A digital karyotyping library of 179,299 genomic tags from the D458MED cell line was generated, permitting analysis of loci distributed at an average distance of 33 kb throughout the genome. Analysis of the tag densities revealed subchromosomal regions of amplification and deletion commonly associated with medulloblastomas (ref. 6; Fig. 1A ), including amplification of the C-MYC oncogene on chromosome 8q24.21, loss of chromosome 17p, and gain of chromosome 17q. Importantly, subtle changes below the level of resolution achievable with traditional measurements were also identified. Most striking was a 28-fold amplification located at base pairs 55,051,299-55,437,589 (UCSC human genome assembly, July 2003 freeze) on chromosome 14q22.3. Examination of a public human genome database (http://genome.ucsc.edu/cgi-bin/hgGateway) identified one known and one predicted gene in the amplified segment, full-length OTX2 (55,257,468-55,267,225) and a COOH-terminal portion of C14orf101 (55,036,584-55,105,043), respectively ( Fig. 1B. Amplification of 20 copies per haploid genome was confirmed with Q-PCR analysis of genomic DNA using multiple sets of primers specific to the genomic sequences of OTX2 and C14orf101 (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

OTX2 is amplified in medulloblastoma cell line D458MED. A, low-resolution tag density maps of digital karyotyping reveal subchromosomal changes in D458MED. For all chromosomes (labels 1-22, x, and y), values on the y-axis indicate genome copies per haploid genome, and values on the x-axis represent positions along the chromosome in Mb. Two prominent peaks, which contain large-scale genomic amplifications, are clearly revealed on chromosomes 8 and 14. Loss of heterozygosity of 17p and gain of 17q is also observed. B, high-resolution tag density maps identify OTX2 amplification on chromosome 14q22.3. An enlarged view of the region of amplification on chromosome 14 was determined from sliding windows of 200 virtual tags. Genes present within the region are indicated below on a high-resolution map, displaying that the OTX2 gene was entirely contained in the amplified region.

After Q-PCR evaluation of genomic DNA from 12 additional medulloblastoma cell lines and 44 clinical medulloblastoma samples, C14orf101 amplification was observed in cell line D487MED, and marked OTX2 amplification was identified in both D487MED and D425MED (10- and 25-fold, respectively). Notably, cell line D425MED was derived from a patient with a primary cerebellar medulloblastoma, whereas D458MED, the cell line used for the original digital karyotyping, was established from cerebrospinal fluid samples of the same patient at a later date, indicating that the OTX2 amplification occurred before the tumor metastasized and was maintained by the metastasized cells. We further assessed OTX2 amplification by fluorescence in situ hybridization. Increased OTX2 copy number was observed in D425MED, D458MED, and D487MED (data not shown).

OTX2 amplification is particularly interesting because of OTX2's role in normal cerebellar development (7). OTX2 is a member of a highly conserved family containing the bicoid-like homeodomain transcription factors that control the developmental programs underlying brain morphogenesis (8). In the embryo, OTX2 is normally expressed throughout the posterior cerebellum, within the external granular layer, and in the emerging internal granular layer and then disappears in later life (9). Elimination of OTX2 function in the mouse results in defective development of the rostral neuroectoderm, leading to a headless phenotype (10), whereas mice with induced ectopic OTX2 expression in the anterior hindbrain display cerebellar ataxia (11). Several developmentally related and hindbrain-specific mRNA transcripts have been found at abnormally high levels in medulloblastomas by SAGE and suppression subtraction hybridization (12, 13) , supporting the hypothesis that these tumors originate from dysregulation of the developmental programs in cerebellar granular cells or their precursor. OTX2 has been identified as one of these genes. However, as with the other genes identified in prior studies, this relationship was associative and not proof of causation. The finding of highly specific and marked OTX2 genetic amplifications in a subset of the medulloblastomas implies that OTX2 is a targeted oncogene in medulloblastoma pathogenesis and not merely transcriptionally upregulated by other oncogenic events.

In order to examine OTX2 expression, we evaluated OTX2 transcript levels by SAGE data mining 211 libraries obtained from normal human tissues and tumors (http://cgap.nci.nih.gov/SAGE; ref. 14). The data revealed a dramatic overexpression of OTX2 in medulloblastomas but not in any other tumor type or normal tissues ( Table 1 ). Consistent with SAGE analysis, quantification of OTX2 mRNA by Q-PCR showed minimal expression in cerebellum, no detectable expression in other normal tissues tested, and no expression in 18 glioblastoma multiforme tumors, the other major brain tumor type (data not shown). By contrast, high-level OTX2 expression was identified in 21 of 33 (63%) medulloblastoma samples tested ( Table 2 ).

Medulloblastoma patients have widely disparate and often unpredictable clinical outcomes. Two subtypes of medulloblastoma, large cell and nodular/differentiated, have documented worse and better prognoses, respectively (1). A recent Pediatric Oncology Group study of 330 childhood medulloblastomas identified tumor anaplasia, found in 24% of tumors in their group, to be a marker of aggressive clinical behavior (15). Among the 25 medulloblastoma cases evaluated for OTX2 expression by Q-PCR or SAGE, 14 of 15 (93%) cases with high OTX2 expression were anaplastic, whereas only 2 of 10 (20%) cases with no OTX2 expression were classified as anaplastic ( Table 2).

To further examine the role of OTX2 expression, we disrupted OTX2 expression in medulloblastoma cells using two different OTX2 siRNAs against OTX2. Repression of OTX2 expression by siRNAs inhibited cell growth in the OTX2-expressing cell lines D283MED and D425MED, whereas there was no effect using the same treatment on the OTX2-nonexpressing cell line D581MED Fig. 2 ). We also identified a pharmacologic agent for manipulation of OTX2 expression in medulloblastoma cells. Exogenous retinoids comprise a chemically related group of nuclear-acting lipid-soluble hormones, which when exogenously applied have been shown to displace or repress OTX2 expression in the embryonic nervous system and also in embryonal carcinoma cells through cis-acting elements of the OTX2 promoter (16, 17) . We therefore examined the capacity of a strong retinoid, ATRA, for its effects on cell proliferation and apoptosis on OTX2-expressing and nonexpressing cell lines. Remarkably, ATRA abrogated cell proliferation in each of the seven OTX2-expressing cell lines in a dose-dependent manner but had no growth-inhibitory effects on the four medulloblastoma cell lines with absent or minimal OTX2 expression ( Fig. 3A ). Furthermore, 2.0 μM ATRA markedly suppressed anchorage-independent tumor cell growth in each of the seven medulloblastoma lines with OTX2 expression ( Fig. 3B), whereas there was no inhibitory effect on D581MED, which expressed OTX2 at a minimal level ( Fig. 3B). To illustrate that the growth inhibition was mediated through the repression of OTX2 expression, we examined the expression of OTX2 after ATRA treatment and found that ATRA blocked expression of OTX2 in a dose-dependent manner ( Fig. 3C). We also showed that the ATRA-induced decrease in cell viability was due to increased cell apoptosis ( Fig. 3D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

OTX2 siRNAs repressed OTX2 expression and inhibited cell growth in medulloblastoma cell lines with OTX2 expression. A, inhibition of OTX2 mRNA expression by OTX2 siRNAs. D283MED and D425MED, two OTX2-expressing medulloblastoma. Cell lines, were treated with OTX2 siRNA#1 or siRNA#2. After 48 hours, cells were collected and mRNA was isolated for Q-PCR quantification of the level of OTX2 expression. B, OTX2 siRNAs inhibited medulloblastoma cell proliferation. Cell lines D283MED, D425MED, and D581MED# were treated with the OTX2 siRNA#1 or siRNA#2. After three doubling times, the number of viable cells was determined by the MTT assay. D581MED# was an OTX2-nonexpressing medulloblastoma cell line.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

ATRA repressed OTX2 expression and inhibited cell growth in medulloblastoma cell lines with OTX2 expression. A, ATRA inhibited medulloblastoma cell proliferation. Medulloblastoma cell lines with OTX2 expression at high levels (solid symbol) and lines without OTX2 expression (open symbol) were treated with the indicated doses (0, 0.05, 0.5, and 2 μmol/L) of ATRA. After three doubling times, the number of viable cells was determined by the MTT assay. B, ATRA abrogated anchorage-independent growth of medulloblastoma cells in soft agar. The indicated medulloblastoma cell lines were incubated with or without 2 μmol/L of ATRA. D283MED, D384MED, D425MED, D458MED, D487MED, D556MED, and D721MED expressed OTX2 at high levels, whereas D581MED expressed minimal OTX2. C, inhibition of OTX2 mRNA expression by ATRA. D283MED, an OTX2-expressing medulloblastoma cell line, was treated with the indicated ATRA doses (0, 0.5, 2, 5, and 10 μmol/L). After 6, 12, 24, and 48 hours, cells were collected and mRNA was isolated for Q-PCR quantification of the level of OTX2 expression. D, ATRA induces medulloblastoma cell apoptosis. D283MED, D425MED, D581MED, and MCD1 were treated with the indicated dose of ATRA (2 and 10 μmol/L) or vehicle control (DMSO). After 2 days, the cytoplasm of the cells was extracted and the histone-associated DNA fragments (nucleosome) enriched in the cytoplasm were quantified by the Cell Death Detection ELISA^plus method (Roche Diagnostic). D283MED and D425MED were OTX2-expressing medulloblastoma cell lines, whereas D581 and MCD1 were OTX2-nonexpressing medulloblastoma cell lines. *, P < 0.05. #, OTX2-nonexpressing cells.

In summary, this study shows the power of digital karyotying as a high-resolution whole-genome approach for the isolation of individual oncogenes. Through analyses of genomes and transcriptomes, we identified OTX2 as being overexpressed in the majority of medulloblastomas with anaplastic histopathologic features and as being genetically amplified in a subset of them. The genetic alterations of OTX2 provide cogent evidence for a pathogenic role, as tumor expression data alone cannot be interpreted with respect to cause and effect (18). Further experiments are necessary to investigate the possibility that abnormal OTX2 expression may result in the dysregulation of the developmental programs in cerebellar progenitor cells and cause neoplastic transformation. It is of interest that other developmental regulatory pathways have been shown to be altered by expression or mutation in medulloblastomas. In particular, the hedgehog/patched signaling pathway is required for the proliferation and maintenance of cerebellar progenitor cells, and mutations of the components of this pathway have been identified in medulloblastomas. It will be interesting to determine if OTX2 is part of this signaling network or if it belongs to a parallel/alternate pathway.

Finally, the prospect for the rational use of a new therapeutic agent against medulloblastoma should be discussed. This study showed that ATRA repressed OTX2-expression and inhibited OTX2-expressing medulloblastoma cell growth. Although ATRA and other retinoids may affect multiple molecular pathways, the connection between OTX2 repression and growth inhibition effect of ATRA suggests that OTX2 expressing medulloblastomas may be amenable to therapy with retinoids. Previous studies have shown that pharmalogically relevant doses of ATRA induce apoptosis in medulloblastoma cells, although a connection with anaplastic histology, or OTX2 expression was not established (19, 20). ATRA is approved clinically for the treatment of propmyelocytic leukemia, a disease in which another nuclear receptor is a predominant determinant of pathogenesis (21). Our studies of ATRA in medulloblastoma, in conjunction with the studies of others, lay the conceptual framework for clinical trials of retinoids in the treatment of a commonly lethal pediatric brain tumor.