«Molecular Characterization of Genetic and Epigenetic Alterations in Gliomas by Christopher Gentry Duncan Department of Pathology Duke University ...»
Molecular Characterization of Genetic and Epigenetic Alterations in Gliomas
Christopher Gentry Duncan
Department of Pathology
Hai Yan, Supervisor
Soman N. Abraham
Robin E. Bachelder
Roger E. McLendon
Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Pathology in the Graduate School of Duke University i v ABSTRACT Molecular Characterization of Genetic and Epigenetic Alterations in Gliomas by Christopher Gentry Duncan Department of Pathology Duke University Date:_______________________
Hai Yan, Supervisor ___________________________
Soman N. Abraham ___________________________
Robin E. Bachelder ___________________________
Roger E. McLendon ___________________________
Xiao-Fan Wang An
of a dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Pathology in the Graduate School of Duke University i v Copyright by Christopher Gentry Duncan Abstract Glioma development and progression are driven by complex genetic alterations, including point mutations and gain or loss of genomic copy number, as well as epigenetic aberrations, including DNA methylation and histone modifications.
However, the molecular mechanisms underlying the causes and effects of these alterations are poorly understood, and improved treatments are greatly needed. Here, wereport a comprehensive evaluation of the recurrent genomic alterations in gliomas and further dissect the molecular effects of the most frequently-occurring genomic events. First, we performed a multifaceted genomic analysis to identify genes targeted by copy number alteration in glioblastoma, the most aggressive malignant glioma. We identify EGFR negative regulator, ERRFI1, as a glioblastoma-targeted gene within the minimal region of deletion in 1p36.23. Furthermore, we demonstrate that Aurora-A kinase substrate, TACC3, displays gain of copy number on 4p16.3 and is overexpressed in a grade-specific pattern. Next, using a gene targeting approach, we knocked-in a single copy of the most frequently observed point mutation in gliomas, IDH1R132H/WT, into a human cancer cell line. We show that heterozygous expression of the IDH1R132H allele is sufficient to induce the genome-wide alterations in DNA methylation characteristic of these tumors. Together, these data provide insight on genetic and epigenetic alterations which drive human gliomas.
List of Figures
1. Introduction to this dissertation
1.1 Permissions and collaborative work
1.2 Genomic alterations and the pathogenesis of glioblastoma
2. Integrated genomic analyses identify ERRFI1 and TACC3 as glioblastoma-targeted genes
2.1 Background: Genomic analysis of glioblastoma
2.2.1 Tumor samples
2.2.2 Digital karyotyping
2.2.3 High density SNP-Arrays
2.2.4 TCGA data
2.2.5 Bioinformatic analysis
2.2.6 Quantitative real-time PCR
2.2.7 Migration assays
2.3.1 Detection of focal copy number alterations by DK and Illumina BeadChips... 13 2.3.2 ERRFI1 on 1p36 is a candidate tumor suppressor gene, whose products regulate glioblastoma cell migration
2.4 Discussion of copy number alterations in glioblastoma
2.4.1 Deletions of 1p36
2.4.2 Copy number gains of 4p16
3. A heterozygous IDH1R132H/WT mutation induces genome-wide alterations in DNA methylation
3.1 Background: IDH mutations and the epigenome
3.2.1 Cell culture and drug treatment
3.2.2 Gene targeting of the human IDH1 locus
3.2.3 D-2-HG analysis
3.2.4 Genome-wide CpG methylation profiling
3.2.5 Statistical analysis of DNA methylation
3.2.6 Genome-wide expression profiling
3.2.7 Statistical analysis of gene expression
3.2.8 Genomic bisulfite sequencing analysis
3.2.9 Histone extraction and western analysis
3.2.10 Chromatin immunoprecipitation
3.2.11 Quantitative real-time reverse transcriptase PCR
3.2.12 Analysis of primary glioblastomas and low grade gliomas
3.2.13 Analysis of HCT116 ChIP-seq data
3.3.2 IDH1R132H/WT induces alterations in DNA methylation
3.3.3 Methylation alterations observed in HCT116 IDH1R132H/WT knock-in cell lines are similarly affected in a brain tumor cell line overexpressing IDH1R132H................ 54 3.3.4 Methylation alterations observed in HCT116 IDH1R132H/WT knock-in cell lines are similarly associated with IDH1 mutation in IDH1-mutant and G-CIMP+ primary gliomas
3.3.5 Effects of IDH1R132H/WT on gene expression
3.3.6 Relationship between IDH1 mutation-induced alterations in DNA methylation and gene expression
3.3.7 Global and gene-specific histone lysine methylation modifications correlate with DNA methylation and gene expression alterations in IDH1R132H/WT cells.......... 73 3.3.8 Gene silencing at specific IDH1R132H/WT-targeted loci is reversed using a DNA hypomethylating agent
3.4 Discussion of IDH mutations and their impact on the epigenome
3.4.1 Heterozygous IDH1 mutations induce alterations in DNA methylation.......... 82 3.4.2 Enzymatic activity of heterozygous IDH mutations
3.4.3 TET proteins and 5-hydroxymethylcytosine
3.4.4 Histone (de)methylation
3.4.5 Relationship between DNA methylation and gene expression
3.4.6 Targets of mutant IDH1-mediated de novo epigenetic silencing
4. Summary and future directions
4.1 Summary and conclusions
4.2 Future studies
Figure 2: Focal high copy number gains and homozygous deletions in glioblastomas... 15 Figure 3: High-resolution mapping of homozygous deletions detects DFFB, C1orf174, and LOC100133612 within secondary MDR on 1p36.32.
Figure 4: High-resolution mapping of homozygous deletions reveals ERRFI1 within the most frequent MDR on 1p36.23.
Figure 5: ERRFI1 is silenced in glioblastomas and reduces cell migration in H423 glioblastoma cells.
Figure 6: High resolution mapping identifies TACC3 as the glioblastoma-targeted gene on 4p16.3
Figure 7: TACC3 is the predominant gene upregulated in 4p16.3 and correlates with Aurora kinase expression.
Figure 8: Grade-specific TACC3 upregulation and TACC2 downregulation in gliomas. 24 Figure 9: Targeted knock-in of IDH1R132H/WT hotspot mutation in a human cancer cell line.
Figure 10: Diagnostic PCR screens for generation of knock-in clones.
Figure 11: IDH1R132H/WT-induced DNA methylation alterations in HCT116 cells............... 49 Figure 12: Consistent IDH1R132H/WT-induced DNA methylation changes between independent HCT116 knock-in clones.
Figure 13: Human oligodendroglioma (HOG) cells overexpressing IDH1R132H recapitulate the changes in DNA methylation observed in HCT116 cells.
Figure 14: IDH1 mutant and G-CIMP+ gliomas recapitulate the DNA methylation alterations observed in cell line models.
Figure 16: Gene expression profiling of HCT116 IDH1R132H/WT cell lines.
Figure 17: HCT116 DNA methylation and gene expression changes negatively correlate.
Figure 18: Distribution of HCT116 IDH1R132H/WT differentially methylated loci relative to transcription start sites (TSS) and CpG islands.
Figure 19: RNA polymerase II (Pol II) and Histone 3 Lysine 4 trimethylation (H3K4me3) ChIP-seq data indicates IDH1R132H/WT differentially methylated loci have reduced Pol II binding in HCT116 wild-type cells
Figure 20: Bisulfite sequence analysis of candidate CpG loci validates IDH1R132H/WTinduced DNA methylation changes.
Figure 21: Bisulfite sequencing at candidate loci of HCT116 non-targeted clones indicates no change in methylation is resultant of the gene targeting procedure
Figure 22: Global and gene-specific histone lysine methylation coincides with IDH1R132H/WT-induced DNA methylation.
Figure 23: Inhibition of DNA methylation results in restoration of gene expression for IDH1R132H/WT-repressed loci.
Figure 24: Schematic depiction of a hypothesized link between IDH1 mutations and epigenetic alterations in gliomas.
Abraham, R.E. Bachelder, R.E. McLendon, and X.F. Wang for guidance on my thesis committee. In particular, I thank P.J. Killela for collaboration on the study of genomic alterations in glioblastoma and B.G. Barwick and P.M. Vertino for collaboration on the study of IDH1 mutations and DNA methylation. I thank D.D. Bigner, J.T. Chi, and C.
Rago for collaboration and guidance; P. Kapoor-Vazirani and D.R. Powell for histone analysis; C.A. Payne for cell migration analysis; M.J. Ehinger and D.L. Satterfield for glioblastoma sample processing; B.A. Rasheed and D. Lister for glioblastoma DNA purification; I. Spasojevic, P. Fan and the DUCC Clinical Pharmacology Lab for quantitative analysis of D-2-HG; K. Abramson and the Duke CHG Molecular Genetics Core for their assistance with the Illumina array studies; and the Duke Microarray Core Facility for their assistance with the microarray studies. I thank G. Jin, Z.J. Reitman, G.Y.
Lopez, M. Wortham, C.J. Pirozzi, Y. He, C. Guo, C. Chang, L.H. Chen, R. Yang, and other members of the Yan lab and The Preston Robert Tisch Brain Tumor Center for technical help and helpful scientific discussions. I thank C. Di and Q. Shi for fundamental laboratory training. I thank my wife T.M. Duncan, parents A.H. Duncan and K.G. Duncan, siblings J.R. Duncan, E.E.D. Baucom and D.D. Baucom, and many other family members and friends for support. Copyrighted and/or published content
Journals (Duncan and Yan 2011), Impact Journals (Duncan et al. 2010), and Cold Spring Harbor Laboratory Press (Duncan et al. 2012) as described in Appendix A. The work described here was supported by American Cancer Society Research Scholar Award RSG-10-126-01-CCE, The Pediatric Brain Tumor Foundation Institute Grant, The Southeastern Brain Tumor Foundation Grant and NCI Grants R01-CA118822, 5R01CA140316, 2RO1-CA077337 and 5RO1-CA132065, as well as by the following NIH grants: NINDS Grant 5P50 NS20023, NCI SPORE Grant 5P50 CA108786, and NCI Merit Award R37 CA 011898.
The excerpts and figures from published works were reproduced with permission as described in Appendix A. Where the publisher does not require explicit permission, excerpts and figures were reproduced in accordance with the policy of the journal or publisher as described in Appendix A. The work detailed in this dissertation was performed primarily by the candidate, although in many cases the work was aided by collaborators, for instance by performing analyses or by providing biological samples as described in the Acknowledgements.
1.2 Genomic alterations and the pathogenesis of glioblastoma While years of research have contributed to understanding the molecular mechanisms underlying initiation and progression of glioblastoma, the most common malignant brain tumor, prognosis remains dismal. Recent efforts have focused on the identification of glioblastoma-targeted genes through comprehensive genomic studies, revealing patterns of genetic alterations that include coding sequence mutations, gain or loss of genomic DNA copy number, and alterations of mRNA expression signature (Parsons et al. 2008; The Cancer Genome Atlas Research Network 2008; Ohgaki and Kleihues 2009; Purow and Schiff 2009).
The genomic landscape of glioblastoma is highly heterogeneous, as genetic alterations vary highly from tumor to tumor, but common themes have been elucidated.
Together, major genes altered in glioblastomas have been identified and shown to contribute to core disease pathways including receptor tyrosine kinase (RTK), TP53, and RB signaling (Parsons et al. 2008; The Cancer Genome Atlas Research Network 2008) (Figure 1). Among the most frequent alterations of RTK signaling are mutations or amplifications of the growth factor receptors EGFR, ERBB2, PDGFRA, and MET, mutation of phosphatidylinositol 3-kinases PIK3CA and PIK3R1, and mutation or homozygous deletion of the PTEN tumor suppressor gene. Inactivation of the RB tumor suppressor pathway was primarily realized as deletions of the CDKN2A/CDKN2B locus, amplification of CDK4, CDK6, and CCND2, and mutation or deletion of RB1. Somatic alterations of TP53 signaling comprised mutations and deletions of TP53 and amplifications of MDM2 and MDM4. Collectively, these core glioblastoma pathways promote cellular proliferation and survival, inhibit cell death, and advance cell cycle progression. An additional gene alteration that has recently been identified is point mutations in the isocitrate dehydrogenases, IDH1 and IDH2, at high frequency in secondary glioblastomas and progressive gliomas (Parsons et al. 2008; Yan et al. 2009b).
IDH1 and IDH2 mutations confer gain of a novel enzymatic activity to catalyze the production of D-2-hydroxyglutarate while impairing normal conversion of isocitrate to α-ketoglutarate. While specific mechanistic details are uncertain, recent insights Figure 1: Schematic depiction of recurrent genetic alterations and major signaling pathways involved in the pathogenesis of glioblastoma.