Integrative Genomic Analysis Identifies NDRG2 As a Candidate Tumor Suppressor Gene Frequently Inactivated in Clinically Aggressive Meningioma
Total Page：16
File Type：pdf, Size：1020Kb
Research Article Integrative Genomic Analysis Identifies NDRG2 as a Candidate Tumor Suppressor Gene Frequently Inactivated in Clinically Aggressive Meningioma Eriks A. Lusis,1 Mark A. Watson,2 Michael R. Chicoine,3 Meghan Lyman,4 Peter Roerig,5 Guido Reifenberger,5 David H. Gutmann,4 and Arie Perry1 1Division of Neuropathology, Departments of 2Pathology and Immunology, 3Neurosurgery, and 4Neurology, Washington University School of Medicine, St. Louis, Missouri; and 5Department of Neuropathology, Heinrich-Heine-University, Du¨sseldorf, Germany Abstract and TSLC1 are also common (2, 4–6). However, there are few Although meningiomas are common central nervous system known genetic changes associated with the malignant progression tumors, little is known about the genetic events responsible of meningioma. Although losses of chromosomal arms 14q and 1p for malignant progression. In this study, we employed gene are common in anaplastic tumors (WHO grade III) and are expression profiling to identify transcripts whose expression generally associated with poor prognosis (7, 8), the relevant tumor was lost in anaplastic (WHO grade III) versus benign (WHO suppressor genes that map to these loci have not been identified. grade I) meningioma. Approximately 40% of genes down- Loss of heterozygosity studies have had limited success in regulated in anaplastic meningioma were localized to chro- identifying minimal regions of deletion, in part, due to the fact mosomes 1p and 14q. One specific gene located at 14q11.2, that the entire chromosome or chromosomal arm is lost in most NDRG2, was consistently down-regulated in grade III menin- examples. In the current study, we used a strategy combining expression profiling with known cytogenetic alterations, leading gioma, a finding which we validated at both the transcript and protein levels in independent sets of clinically and patholog- to the identification of NDRG2 as a potential meningioma- ically diverse meningiomas. Loss of NDRG2 expression was associated tumor suppressor gene on 14q11.2. We subsequently also seen in a subset of lower-grade meningiomas, including validated its role in meningioma tumor progression using atypical meningiomas (WHO grade II) with clinically aggres- multiple methods for measuring RNA and protein expression sive behavior. Furthermore, we found that the loss of NDRG2 within independent cohorts of clinically well-characterized expression was significantly associated with hypermethylation meningiomas. Lastly, hypermethylation of the CpG island within of the NDRG2 promoter. Collectively, these data identify the NDRG2 promoter region was shown to be the likely NDRG2 as the first specific candidate tumor suppressor gene mechanism of inactivation in the majority of meningiomas with on chromosome 14q that is inactivated during meningioma loss of NDRG2 expression. progression. In addition, these findings highlight the utility of combining genomic, epigenetic, and expression data to Materials and Methods identify clinically significant tumor biomarkers, and suggest Tissue specimens. Tissue specimens were collected by the Siteman that NDRG2 expression will be a useful and functionally Cancer Center Tissue Procurement Facility and the Department of relevant biomarker to predict aggressive behavior in patients Neuropathology, Heinrich-Heine-University, under approved protocols with meningioma. (Cancer Res 2005; 65(16): 7121-6) from the Institutions’ Review Boards. Snap-frozen tissue specimens from surgically resected meningiomas were histologically reviewed for speci- Introduction men adequacy (at least 80% tumor) and subsequent 50 Am serial sections from each banked frozen specimen were cut, placed immediately into Meningiomas account for approximately one-fourth of all Trizol reagent (Invitrogen, Carlsbad, CA), and homogenized for RNA primary central nervous system neoplasms. Although most are preparation. Total RNA was isolated from Trizol homogenates using the benign, as many as 20% display clinically aggressive features, manufacturer’s protocol. RNA was then further purified using RNeasy spin leading to increased patient morbidity or mortality (1, 2). The columns (Qiagen, Valencia, CA). For RNA isolation from paraffin- 2000 WHO grading scheme has improved the ability to predict embedded tissues, Ambion’s (Austin, TX) formalin-fixed paraffin-embedded clinical behavior (3), though significant variability remains, RNA isolation kit was used according to the manufacturer’s protocol. All particularly within the subset of atypical meningiomas (WHO RNAs were quantified by UV absorbance at 260 and 280 nm and qualitatively assessed using an RNA Nano Assay and Bioanalyzer 2100 grade II). The most clearly established early genetic alteration in (Agilent, Palo Alto, CA). sporadic meningioma is biallelic inactivation of the neurofibro- Microarray analysis. Five micrograms of tumor RNA was converted matosis 2 (NF2) gene located on chromosome 22q, with to double-stranded cDNA using a dT-T7 promoter primer. Purified, associated loss of merlin (schwannomin) expression. Recently, double-stranded cDNA was then used as a template to create we have shown that loss of expression of other protein 4.1 family biotinylated aRNA. The labeled aRNA target was quantified, fragmented, members and interacting proteins, such as 4.1B (DAL-1), 4.1R, and 15 Ag was used for microarray hybridization. The complete target preparation protocol was done according to the manufacturer’s recommendations (Affymetrix, Santa Clara, CA) and as previously published (9). Labeled targets were hybridized to Affymetrix U133A Requests for reprints: Arie Perry, Division of Neuropathology, Washington and U133B GeneChip microarrays for 16 hours and washed following University School of Medicine, Campus Box 8118, 660 South Euclid Avenue, St. Louis, standard protocols. Microarray images were processed using Affymetrix MO 63110. Phone: 314-362-9130; Fax: 314-362-4096; E-mail: [email protected]. I2005 American Association for Cancer Research. Microarray Analysis Suite version 5.0. Each array was scaled so that the doi:10.1158/0008-5472.CAN-05-0043 average probe set hybridization signal intensity value (target intensity) www.aacrjournals.org 7121 Cancer Res 2005; 65: (16). August 15, 2005 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Cancer Research Figure 1. Histogram demonstrating the location of 83 genes whose expression was significantly lower (P < 0.01) in anaplastic (WHO grade III) meningiomas relative to benign (WHO grade I) meningiomas. x-axis, the cytogenetic location; y-axis, percentage of differentially expressed probe sets located on each chromosome arm, relative to the total number of probe sets on that arm that are represented on the GeneChip U133A and U133B microarrays. Overrepresentation of probe sets at chromosome 1p and 14q (black column) was statistically significant (1p, Z score = 1.7288, P = 0.0419; 14q, Z score = 5.4000, P < 0.0001). NA, probe sets for which chromosomal localizations are not known. was 1,500. Scaled data for each array was exported to the Siteman Immunohistochemistry. Immunohistochemistry using NDRG-specific Cancer Center Bioinformatics Server,6 merged with updated gene affinity-purified rabbit polyclonal antisera (1:10,000 dilution) was done on annotation data for each probe set on the array, and downloaded for formalin-fixed paraffin-embedded sections, using a recently developed further data visualization and analysis. The completely annotated, antibody (11). Paraffin blocks from surgically resected meningiomas were MIAME-compliant data set can be found at the above noted URL. Basic retrieved from the surgical pathology files at Washington University microarray data visualization, data filtering, and hierarchical clustering were School of Medicine in St. Louis, MO. Positive controls included normal done using DecisionSite for Functional Genomics (Spotfire, Somerville, MA). human and rodent brain, as well as human leptomeningeal rolls prepared Average fold differences in gene expression between benign and anaplastic from autopsy specimens as described previously (4). Cases with strong meningiomas were calculated and Student’s t test (uncorrected for multiple diffuse staining were scored as +, those with either patchy immunoreac- comparisons) was used to generate a list of differentially expressed genes tivity or decreased staining intensity compared with adjacent vessels were between the two tumor groups. scored as F (i.e., partial loss), and those with lack of tumoral staining in Quantitative reverse transcription-PCR analysis. Oligonucleotide >90% of the specimen were scored as À (i.e., complete loss). Regions lacking sequences corresponding to the NDRG2 gene transcript were designed vascular staining were not scored due to the possibility of technical failure. using Primer Express software (Applied Biosystems, Foster City, CA) and NDRG2 promoter methylation analysis. The NDRG2 promoter was are available on request from the authors. Five hundred nanograms of analyzed for CpG site methylation using sequencing of sodium bisulfite- total cellular RNA from tissue specimens was subjected to reverse modified DNA. Sodium bisulfite treatment of genomic DNA was carried out transcription using Omniscript reverse transcriptase (Qiagen) and oligo- as described by Herman et al. (12). Two overlapping fragments from the dT, following the manufacturer’s protocol. After first strand synthesis, an NDRG2 promoter region (covering 32 CpG sites