The Mef2b Regulatory Network

THE MEF2B REGULATORY NETWORK

Julia Pon

B.Sc., The University of Alberta, 2010

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

The Faculty of Graduate and Postdoctoral Studies

(Genome Science and Technology)

THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)

June 2015

Abstract Myocyte enhancer factor 2B (MEF2B) is a transcription factor with somatic mutation hotspots at K4, Y69 and D83 in diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma. The recurrence of these mutations indicates that they may drive lymphoma development. However, inferring the mechanisms by which they may drive lymphoma development was complicated by our limited understanding of MEF2B’s normal functions. To expand our understanding of the cellular activities of wildtype and mutant MEF2B, I developed and addressed two hypotheses: (1) identifying genes regulated by wildtype MEF2B will allow identification of cellular phenotypes affected by MEF2B activity and (2) contrasting the DNA binding sites, effects on gene expression and effects on cellular phenotypes of mutant and wildtype MEF2B will indicate mechanisms through which MEF2B mutations may contribute to lymphoma development. To address these hypotheses, I first identified genome-wide MEF2B binding sites and transcriptome-wide gene expression changes mediated by MEF2B. Using these data I identified and validated novel MEF2B target genes. I found that target genes of MEF2B included the cancer genes MYC, TGFB1, CARD11, NDRG1, RHOB, BCL2 and JUN. The identification of target genes led to findings that MEF2B promotes expression of mesenchymal markers, promotes HEK293A cell migration, and inhibits DLBCL cell chemotaxis. I then investigated how K4E, Y69H and D83V mutations change MEF2B’s activity. I found that K4E, Y69H and D83V mutations decreased MEF2B’s capacity to promote gene expression in both HEK293A and DLBCL cells. These mutations also reduced MEF2B’s capacity to alter HEK293A and DLBCL cell movement. Overall, these data support the concept that MEF2B mutations may promote lymphoma development by reducing expression of MEF2B target genes that would otherwise function to help confine germinal centre B-cells to germinal centres. My research demonstrates how observations from genome-scale data can aid in the functional characterization of candidate driver mutations. Moreover, my work provides a unique resource for exploring the role of MEF2B in cell biology. I map for the first time the MEF2B regulome, demonstrating connections between a relatively understudied transcription factor and genes significant to oncogenesis.

Preface Portions of Chapter 1 have been published: J. Pon, M.A. Marra. Driver and Passenger Mutations in Cancer. Annu. Rev. Pathology: Mechanisms of Disease. 10:25-50, 2014. © by Annual Reviews, http://www.annualreviews.org. I wrote most of the text for this review with guidance from M.A. Marra. Figure 1.5 was produced by Jianghong An. Portions of Chapters 2-5 have been accepted for publication pending formatting: J.R. Pon, J. Wong, S. Saberi, O. Alder, M. Moksa, S.W.G. Cheng, G.B. Morin, P.A. Hoodless, M. Hirst, M.A. Marra. MEF2B Mutations in Non-Hodgkin Lymphoma Dysregulate Cell Migration by Decreasing Transcriptional Activation of MEF2B Target Genes. Nature Communications. I designed and performed the research, analyzed and interpreted data, and drafted the manuscript. J.W. produced data in Figures 5.1b and 5.2b and optimized protein detection. S.S. ran MACS2 peak calling and IDR analysis for ChIP-seq data shown in Figure 3.9b. O.A. optimized conditions for gel shift assays. M.M. performed ChIP library construction. RNA-seq library construction, sequencing, alignment of sequence data, and the production of quality control metrics was performed by staff of the British Columbia Genome Sciences Centre. S.W.G.C. and G.B.M. provided advice on gel shift assays and purification of MEF2B-V5 expressed in E. coli. P.A.H. provided advice on gel shift assays and contributed to research design. M.H. advised and co-ordinated ChIP-seq data collection. M.A.M. conceived of the research questions and participated in research design and data interpretation. Dr. Suganthi Chittaranjan also provided guidance on experimental design and data presentation. Dr. Ryan Morin identified the MEF2C mutations reported in Table 1.2. Data in Figure 4.2 and the empty vector cell line was produced by Katie O’Brien, and data in Figure 5.1d was produced by Annie Moradian, Jessica Tamura- Wells and Marlo Firme.

iii

Table of Contents Abstract ...... ii

Preface ...... iii

Table of Contents ...... iv

List of Tables ...... ix

List of Figures ...... x

List of Abbreviations ...... xiii

Acknowledgements ...... xv

Dedication ...... xvii

Chapter 1: Transcriptional Dysregulation in Cancer and the Potential Role of MEF2B Mutations as Drivers of Non-Hodgkin Lymphoma ...... 1

1.1 Introduction ...... 1

1.2 Cancer biology ...... 1

1.3 Regulation of transcription by chromatin structure…………………………………………3

1.4 Non-Hodgkin lymphomas ...... 5

1.4.1 Clinical characteristics of DLBCL, FL and MCL ...... 6

1.4.2 The cellular origin of DLBCL, FL and MCL...... 7

1.4.3 Driver genes of FL, DLBCL and MCL ...... 9

1.5 Technologies and model systems for characterizing transcription factors ...... 10

1.5.1 Technologies for identifying transcription factor target genes ...... 10

1.5.2 Distinguishing direct and indirect target genes ...... 12

1.5.3 Target gene identification as a means of characterizing how transcription factor filller th mutations may contribute to lymphoma development ...... 13

1.5.4 Model systems for examining cancer gene function ...... 14

1.6 MEF2 family proteins ...... 16

1.6.1 Roles of MEF2 proteins in vertebrate organisms ...... 16

1.6.2 Expression patterns of MEF2B ...... 17

1.6.3 Target genes of MEF2 family proteins ...... 18

1.6.4 Functions of the MADS and MEF2 domains in MEF2 proteins ...... 19

1.6.5 Functions of the transactivation domains of MEF2 proteins ...... 22

1.7 Roles of MEF2 family proteins in human disease ...... 23

1.7.1 MEF2B may act as an oncogene in some types of carcinoma ...... 24

1.7.2 MEF2B may act as a tumor suppressor in some types of non-Hodgkin lymphoma .... 24

1.8 Thesis roadmap and chapter summaries...... 28

Chapter 2: Materials and Methods ...... 40

2.1 Production of stably transfected HEK293A cell lines ...... 40

2.2 Production of stably transduced DoHH2 cells ...... 40

2.3 Cell culture and treatments ...... 41

2.4 MEF2-dependent luciferase reporter assay ...... 41

2.5 Expression microarrays and qRT-PCR validation ...... 42

2.6 RNA-seq library construction ...... 43

2.7 Differential expression analysis using HEK293A RNA-seq data ...... 44

2.8 Chromatin immunoprecipitation (ChIP) for sequencing ...... 44

2.9 ChIP and RNA sequencing and alignment ...... 46

2.10 MEF2B-V5 ChIP-seq data analysis and validation using ChIP-qPCR ...... 47

2.11 Gel shift assays ...... 49

2.12 Quantification of protein abundance ...... 50

2.13 HEK293A cell migration assays ...... 52

2.14 Crystal violet proliferation assay...... 52

2.15 Analysis of histone modification ChIP-seq data ...... 52

2.16 Fractionation of nuclear and cytoplasmic protein lysates ...... 53

2.17 DLBCL patient sample analysis ...... 53

2.18 Mass spectrometry...... 54

2.19 Gene set enrichment analysis ...... 54

2.20 Chemotaxis assays...... 55

Chapter 3: Characterization of the MEF2B Regulatory Network in HEK293A Cells ...... 56

3.1 Introduction ...... 56

3.2 Results ...... 57

3.2.1 WT MEF2B regulates genes involved in proliferation, migration and EMT ...... 57

3.2.2 Chromatin immunoprecipitation identifies genome-wide MEF2B binding sites ...... 60

3.2.3 Integrative analysis of DNA binding and gene expression data identifies candidate fill th direct MEF2B target genes...... 62

3.2.4 Candidate direct MEF2B target genes include regulators of cell movement and cell fill th survival ...... 64

3.2.5 MEF2B transcriptional activity is not associated with increased levels of H3K27ac and th H3K4me3 ...... 66

3.2.6 Identification of genes regulated by calcium sensitive activities of MEF2B ...... 68

3.3 Discussion ...... 70

Chapter 4: Impacts of the K4E, Y69H and D83V MEF2B Mutations on the MEF2B Regulatory Network in HEK293A Cells ...... 109

4.1 Introduction ...... 109

4.2 Results ...... 110

4.2.1 Microarray data indicates that K4E, Y69H and D83V MEF2B mutations reduce filller th MEF2B transcriptional activity ...... 110

4.2.2 Validation data support that K4E, Y69H and D83V mutations decrease MEF2B filller th transcriptional activity ...... 111

4.2.3 K4E, Y69H and D83V MEF2B mutations alter the abundance of protein from MEF2B th target genes and decrease cell migration ...... 113 vi

4.2.4 Dominant negative effects of K4E and D83V mutations may be masked by the fillerrr th transcriptional activity retained by K4E and D83V MEF2B ...... 114

4.2.5 K4E and D83V MEF2B mutations decrease MEF2B DNA binding ...... 115

4.2.6 Integrative analysis of DNA binding and gene expression data for K4E, D83V and WT th MEF2B-V5 cells ...... 116

4.3 Discussion ...... 117

Chapter 5: Potential Roles of MEF2B Mutations in DLBCL Development ...... 149

5.1 Introduction ...... 149

5.2 Results ...... 150

5.2.1 MEF2 family genes are expressed in DLBCL cells ...... 150

5.2.2 MEF2B mutations decrease MEF2B’s capacity to activate BCL6 expression in filller th DLBCL cells ...... 151

5.2.3 Endogenous MEF2B in DLBCL cells binds near the TSSs of BCL6, BCL2, RHOB, fil th ABCB4, ITGA5 and JUN ...... 153

5.2.4 Isoform B MEF2B has decreased transcriptional activity compared to isoform A filller th MEF2B ...... 154

5.2.5 MEF2B activity inhibits DLBCL chemotaxis ...... 155

5.3 Discussion ...... 157

Chapter 6: Conclusions and Future Directions...... 170

6.1 WT MEF2B regulates mediators of cell proliferation, cell migration and EMT ...... 170

6.2 MEF2B mutations decrease the capacity of MEF2B to activate transcription by decreasing MEF2B DNA binding ...... 173

6.3 MEF2B mutations reduce inhibition of chemotaxis by decreasing the capacity of MEF2B to activate target gene expression in DLBCL cells ...... 174

6.4 Future directions in the study of MEF2B biology and NHL...... 176

Bibliography ...... 178

Appendices ...... 204 vii

Appendix A: Oligonucleotide sequences ...... 204

Appendix B: Quality control statistics for RNA-seq data...... 207

Appendix C: Quality control statistics for V5 ChIP-seq data...... 208

Appendix D: Quality control statistics for H3K27ac and H3K4me3 ChIP-seq data...... 209

Appendix E: Genes differentially expressed in microarray and RNA-seq data for WT MEF2B- V5 versus control cells (adjusted p-values < 0.05) ...... 210

Appendix F: Functional annotation group enrichment of DEGs in WT MEF2B-V5 versus untransfected cells ...... 215

Appendix G: Functional annotation group enrichment of DEGs in WT MEF2B-V5 versus empty vector cells...... 218

Appendix H: Motifs identified de novo in WT MEF2B-V5 ChIP-seq peak regions ...... 219

Appendix I: Known motifs enriched in WT MEF2B-V5 ChIP-seq peak regions ...... 220

Appendix J: The 261 high confidence candidate direct MEF2B target genes ...... 224

Appendix K: Functional annotation group enrichment in the candidate direct MEF2B target genes ...... 233

Appendix L: The 361 genes differentially expressed in comparisons of K4E MEF2B-V5, Y69H MEF2B-V5, D83V MEF2B-V5 and untransfected cells to WT MEF2B-V5 cells...... 236

Appendix M: Functional annotation group enrichment of DEGs in mutant versus WT MEF2B- V5 cells ...... 245

Appendix N: Functional annotation group enrichment of DEGs in DLBCL patient samples with MEF2B mutations versus DLBCL patient samples without MEF2B mutations ...... 246

Appendix O: Functional annotation group enrichment of DEGs in DLBCL patient samples versus centroblasts...... 247

viii

List of Tables Table 1.1 Potentially somatic MEF2B mutations identified in NHL…………………….. 30 Table 1.2 MEF2C mutations identified in DLBCL……………………………………… 33 Table 3.1 Microarray, RNA-seq and qRT-PCR data for the differential expression of validation set genes in WT MEF2B-V5 versus control cells……………………………... 73 Table 4.1 Microarray, RNA-seq and qRT-PCR data for the differential expression of validation set genes in Y69H versus WT MEF2B-V5 cells…………..………………….. 121 Table 4.2 Known motifs enriched in regions with peaks in WT MEF2B-V5 ChIP-seq but not mutant MEF2B-V5 ChIP-seq…………………………………………………….. 122 Table 4.3 Known motifs enriched in regions with peaks in mutant MEF2B-V5 ChIP- seq…………………………………………………………………………………………. 122 Table 5.1 Differentially expressed genes untransfected versus WT MEF2B-V5 HEK239A cells and DLBCL patient samples with versus without MEF2B mutations…... 160

List of Figures

Figure 1.1 The putative cell types of origin and transcription factor networks of B-cell lymphomas………………………………………………………………………………… 35 Figure 1.2 Structure of isoform A and B MEF2B mRNA transcripts and proteins……… 36 Figure 1.3 The frequency of alterations affecting MEF2 genes across cancer types……. 37 Figure 1.4 The localization of mutations in MEF2B ……………………………………. 38 Figure 1.5 The structure of the MEF2B MADS and MEF2 domains……………………. 39 Figure 3.1 Expression microarrays and RNA-seq detected similar alterations in gene expression in response to WT MEF2B-V5 expression…………………………………… 74 Figure 3.2 Cellular function annotation categories enriched in differentially expressed genes………………………………………………………………………………………. 75 Figure 3.3 Validation of differential gene expression in WT MEF2B-V5 versus untransfected and empty vector cells……………………………………………………… 77 Figure 3.4 MEF2B-V5 expression alters CARD11, MYC, NDRG1 and MEF2C protein abundance ………………………………………………………………………………… 81 Figure 3.5 Expression of WT MEF2B-V5 alters HEK293A cell migration…………….. 83 Figure 3.6 Expression of WT MEF2B-V5 increases expression of genes that are upregulated in EMT…..…………………………………………………………………… 84 Figure 3.7 MEF2B-V5 expression alters the abundance of mesenchymal markers ……... 85 Figure 3.8 The consistency between replicates and predicted reproducibility of peaks identified from WT MEF2B-V5 ChIP-seq data…………………………………………… 86 Figure 3.9 Binding site motifs in MEF2B ChIP-seq data………………………………… 88 Figure 3.10 MEF2A and MEF2C motifs were centrally enriched in MEF2B-V5 ChIP-seq peaks………………………………………………………………………………………... 89 Figure 3.11 WT MEF2B-V5-his binds sequences similar to MEF2 motifs in gel shift assays…...... 90 Figure 3.12 The distribution of WT MEF2B-V5 ChIP-seq peak regions relative to transcription start sites……………………………………………………………………… 91 Figure 3.13 WT MEF2B tends to acts as a transcriptional activator…………………...…. 92 Figure 3.14 ChIP-qPCR validation of V5 ChIP-seq on WT MEF2B-V5 cells………….... 93

Figure 3.15 Cellular function annotation categories enriched in the 1,141 candidate direct MEF2B target genes………………………………………………………………………. 95 Figure 3.16 MEF2B binds near the transcription start sites of BCL2 and JUN …………. 96 Figure 3.17 MEF2B knockdown tends to decrease BCL2 and JUN expression and decrease colony formation………………………………………………………………… 97 Figure 3.18 H3K27ac and H3K4me3 ChIP-seq coverage around transcription start sites... 99 Figure 3.19 The number of peaks and fold enrichment values in H3K27ac and H3K4me3 ChIP-seq data are similar between WT MEF2B-V5 and empty vector cells……………... 100 Figure 3.20 H3K27ac and H3K4me3 ChIP-seq coverage around MEF2B peak regions……………………………………………………………………………………… 101 Figure 3.21 MEF2B-V5 expression did not increase H3K27ac or H3K4me3 near MEF2B-V5 ChIP-seq peak regions or associated transcription start sites...... 103 Figure 3.22 Increased gene expression in WT MEF2B-V5 versus untransfected cells was not associated with increased H3K27ac or H3K4me3……………………………………. 104 Figure 3.23 Effects of increased intracellular calcium levels on MEF2B-dependent gene expression…………………………………………………………………………………. 106 Figure 3.24 Ionomycin increases expression of ANO1, NFATC2 and CCL8 in MEF2B- V5 cells but not empty vector cells……………………………………………………….. 108 Figure 4.1 K4E, Y69H and D83V mutations reduce MEF2B stability ………………….. 123 Figure 4.2 WT, K4E, Y69H and D83V MEF2B localize to the nucleus………………… 124 Figure 4.3 Workflow of expression microarray data analysis …………………………… 125 Figure 4.4 Gene expression in mutant versus WT MEF2B-V5 cells, detected using expression microarrays………………………….………………………………………… 126 Figure 4.5 MEF2B mutations decrease the capacity of MEF2B to promote expression of mesenchymal genes……………………………………………………………………….. 128 Figure 4.6 Validation of expression microarray data using qRT-PCR on additional cell lines………………………………………………………………………………………… 129 Figure 4.7 Correlation of fold changes in gene expression between expression microarray and RNA-seq datasets for mutant versus WT MEF2B-V5 cells…………………………… 133 Figure 4.8 Analysis of RNA-seq data supports the notion that MEF2B mutations

xi decrease the capacity of MEF2B to activate transcription……………………………...….. 134 Figure 4.9 MEF2B knockdown and MEF2B mutations have similar effects on candidate direct target gene expression………………………………………………………………. 136 Figure 4.10 R3T and R24L mutations decrease MEF2B’s capacity to activate transcription………………………………………………………………………………… 137 Figure 4.11 K4E, Y69H and D83V MEF2B mutations alter the abundance of protein from MEF2B target genes …………………………………………………………………. 138 Figure 4.12 Validation of differential protein abundance in an additional HEK293A cell line expressing Y69H MEF2B-V5.………………………………………………………… 140 Figure 4.13 Cell migration is affected more by WT than by mutant MEF2B-V5 expression…………………………………………………………………………………... 141 Figure 4.14 Gel shift assays indicate that K4E and D83V mutations decrease MEF2B DNA binding………………………………………………………………………………. 142 Figure 4.15 K4E and D83V MEF2B binds HEK293A cell DNA at fewer sites than WT MEF2B…………………………………………………………………………………….. 143 Figure 4.16 Genes with decreased expression in mutant versus WT MEF2B-V5 cells tend to be the closest to peaks that are present WT but not mutant ChIP-seq…………….. 144 Figure 4.17 ChIP-qPCR data produced for verification of ChIP-seq on mutant and WT MEF2B-V5 cells…………………………………………………………………………… 145 Figure 4.18 A model of how the K4E, Y69H and D83V MEF2B mutations may decrease target gene expression……………………………………………………………………… 147 Figure 5.1 Expression of MEF2 family members in DLBCL cells……………………….. 161 Figure 5.2 MEF2B-V5 activity in DLBCL cells promotes BCL6 expression…………….. 163 Figure 5.3 ChIP-qPCR on DLBCL cells identifies DNA regions bound by MEF2B…….. 164 Figure 5.4 Stable expression of isoform B MEF2B-V5 affects MEF2B target gene expression less than stable expression of isoform A MEF2B-V5………………………….. 166 Figure 5.5 Cellular function annotation categories enriched in genes differentially expressed in DLBCL patient samples……………………………………………………… 168 Figure 5.6 MEF2B inhibits DLBCL cell chemotaxis……………………………………... 169

xii

List of Abbreviations

ABC DLBCL Activated B-cell diffuse large B-cell lymphoma AID Activation-induced cytidine deaminase BCGSC BC Cancer Agency's Michael Smith Genome Sciences Centre BCR B-cell receptor bp Base pair ChIP Chromatin immunoprecipitation ChIP-qPCR Chromatin immunoprecipitation quantitative polymerase chain reaction ChIP-seq Chromatin immunoprecipitation sequencing DAVID Database for Annotation, Visualization and Integrated Discovery DEG Differentially expressed gene DLBCL Diffuse large B-cell lymphoma ELISA Enzyme-linked immunosorbent assay EMT Epithelial-mesenchymal transition ENCODE Encyclopedia of DNA Elements FBS Fetal bovine serum FDR False discovery rate FL Follicular lymphoma GC B-cell Germinal centre B-cell GCB DLBCL Germinal centre B-cell diffuse large B-cell lymphoma GEO Gene Expression Omnibus GSEA Gene Set Enrichment Analysis HDAC Histone deacetylase indels Small insertion/deletion mutations IPA Ingenuity Integrative Pathway Analysis of Complex omic’s Data kb Kilobase LIMMA Linear models for microarray data MALT Mucosa-associated lymphoid tissue Mb Mega base MCL Mantle cell lymphoma

xiii

NHL Non-Hodgkin Lymphoma PKC Protein kinase C qRT-PCR Quantitative reverse transcriptase polymerase chain reaction R-CHOP Rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone RNA-seq RNA sequencing s.e.m. Standard error of the mean TSS Transcription start site WT Wildtype

xiv

Acknowledgements

I am grateful to have had the opportunity to work with very talented group of people in a leading research centre. I thank my supervisor, Dr. Marco Marra, for his guidance and support. Marco’s devotion to the building of a strong scientific community is inspiring. He has been both a teacher and a role model over the course of my research. I thank my committee members Drs. Sam Aparicio, Pamela Hoodless and Torsten Nelson for their thoughtful feedback and mentoring. Their insights and questions have helped steer both my research and my career directions. I’m grateful to all the members of the Marra lab (Dr. Alessia Gagliardi, Dr. Isabel Serrano, Dr. Suganthi Chittaranjan, Susanna Chan, Marlo Firme, Diane Trinh, Dr. Maria Mendez-Lago, Ryan Huff, Veronique LeBlanc, Elizabeth Chun, Emilia Lim, and Rodrigo Goya) for their thought-provoking discussions and comradery, and for assistance with data collection and analysis. I wish to say particular thanks to all those who’ve volunteered their time to critique my writing, and to Suganthi for guiding my approach to this research. I also thank the co-op students Jackson Wong and Jessica Tamura-Wells for their perseverance in helping to collect data. I thank Dr. Martin Hirst for generously sharing his expertise and resources to assist my research, and thank Dr. Saeed Saberi and Michelle Moksa for applying their talents to my research questions. Consultation with Dr. Gregg Morin and the proteomics group of the Genome Sciences Centre was also instrumental to obtaining my data. The time spent by Drs. Grace Cheng, Olivia Alder and Annie Moradian to teach me techniques and help to run experiments is most appreciated. I thank the Library Construction, Biospecimen, Sequencing and Bioinformatics teams at British Columbia Genome Sciences Centre (BCGSC) for their expert technical assistance. I’m grateful for the expert administrative assistance provided by Lulu Crisostomo and Armelle Troussard. I thank the funding sources of the BCGSC, including the BC Cancer Foundation, Genome Canada, Genome British Columbia, the Cancer Research Society and the Leukemia and Lymphoma Society of Canada. The research reported here was funded by the BC Cancer Foundation and The Terry Fox Foundation, through a program project grant (award #019001). My research has also been generously supported by John Auston and his family. I have been

xv fortunate to have been supported by a CIHR Vanier Canada Graduate Scholarship, a Scriver MD/PhD Scholarship, a University of British Columbia Four Year Fellowship, the John Bosdet Memorial fund, and Graduate Awards from the Faculty of Medicine, Faculty of Science, and College of Interdisciplinary Studies. Finally, I’m grateful to all my family and friends for their encouragement and wisdom as I’ve pursued this research.

xvi

Dedication

To the patients.

xvii

Chapter 1: Transcriptional Dysregulation in Cancer and the Potential Role of MEF2B Mutations as Drivers of Non-Hodgkin Lymphoma1

1.1 Introduction

Cancer development is driven by genetic alterations, some of which affect transcription factor activity. Somatic mutations in the transcription factor gene MEF2B have recently been identified in non-Hodgkin lymphoma (NHL)1–7. The recurrence of these mutations indicates that they may drive lymphoma development. However, inferring the mechanisms by which they may drive lymphoma development was complicated by our limited understanding of MEF2B’s normal functions. To enhance our understanding of the cellular activities of wildtype (WT) and mutant MEF2B, I sought to characterize and contrast the regulatory networks of WT and mutant MEF2B. Chapter 1 of this thesis reviews the literature that has guided my research.

1.2 Cancer biology

Cancer cells are characterized by eight hallmark characteristics: (i) excessive and uncoordinated proliferation, (ii) insensitivity to anti-growth signals, (iii) resistance to cell death, (iv) limitless replicative potential, (v) the capacity to invade local tissue and metastasize, (vi) the capacity to sustain angiogenesis, (vii) reprogrammed energy metabolism and (viii) the capacity to evade the immune system8,9. These properties can result from somatic genetic alterations that provide a selective advantage to the cells harbouring them. Tumors are thus believed to arise from a single cell of origin with a mutation that allows it to expand into a clone of cells10. Tumor progression is then driven by sequential expansions of subclones that have acquired mutations conferring greater selective advantage10. The genetic heterogeneity within and between tumors can thus be explained by the presence of divergent lineages of cells11,12. Identifying which mutations contribute to cancer development is a key step in understanding cancer cell biology and developing targeted therapies. Mutations that provide a

1 Portions of section 1.2 and 1.6.4 have been published: J. Pon, M.A. Marra. Driver and Passenger Mutations in Cancer. Annu. Rev. Pathology: Mechanisms of Disease. 10:25-50, 2014. Copyright by Annual Reviews. Author contributions are provided in the Preface. 1 selective growth advantage and thus promote cancer development are considered driver mutations, whereas those that do not are considered passenger mutations13. Driver genes are genes that may harbour driver mutations14. Genes that have been identified as drivers in at least one cancer type are known as cancer genes14. The Catalogue of Somatic Mutations in Cancer currently lists 522 cancer genes15. Mathematical modeling estimates that 5 to 813,16 driver mutations are required for cancer development. However, the number of passenger mutations far exceeds the number of driver mutations. For instance, samples from 11 adult cancer types had on average only 2 to 6 predicted driver mutations per sample, even though each sample had on average 200 exonic somatic point mutations and small insertion/deletion mutations (indels)17. Similarly at the level of driver genes, common solid tumors had only 3 to 6 mutated genes per sample that were predicted to be drivers, even though an average of 33 to 66 genes per sample had protein altering somatic mutations14. Oncogenes are defined as genes in which driver mutations are activating or result in new functions14. Tumor suppressors are genes in which driver mutations are inactivating14. Oncogenes tend to be affected by focal amplifications or missense mutations at a limited number of codons, whereas tumor suppressors tend to be affected by focal deletions or nonsense, frameshift and splice-site mutations dispersed across the gene18. Assessing the relative frequency and distribution of missense, nonsense, frameshift, and splice-site mutations can suggest whether a candidate driver is likely to be an oncogene or tumor suppressor14,19. According to a “20/20 rule”, genes in which greater than 20% of somatic mutations are missense and at recurrent positions are considered oncogenes, whereas genes in which at least 20% of somatic mutations are inactivating are considered tumor suppressors14. Interestingly, driver mutations can be acquired decades before the cells harbouring them become cancerous13. This is possible because a cell requires multiple mutations to become cancerous, which are acquired gradually over time13. Some mutations do not provide any growth advantage and thus do not act as drivers unless certain other mutations are also present. For instance, a growth advantage is only conferred by inactivation of one allele of a haplosufficient tumor suppressor if the other allele is already inactivated20. Moreover, mutations may only act as drivers during certain stages of cancer development. For instance, mutations inactivating TGF-β signaling appear to drive early tumorigenesis by preventing TGF-β induced growth arrest,

2 whereas mutations increasing TGF-β signaling appear to drive later malignant progression by promoting angiogenesis and invasiveness21. Driver mutations may contribute to cancer development by promoting any of the hallmarks of cancer noted above, or by promoting two enabling characteristics of cancer9. These enabling characteristics, namely genomic instability and an inflammatory state, enable cancer development by accelerating the acquisition of driver mutations or enhancing concentrations of bio-active molecules that in turn promote the hallmark capabilities9. The impact of cancer genes on cellular phenotypes depends in part on their expression levels, which are regulated by an interplay between transcription factors, transcriptional coregulators, chromatin modifying enzymes and chromatin states. Alterations affecting these regulatory factors may contribute to cancer development through their effects on the expression of other cancer genes. Consequently, some mutations in transcription factors and coregulators can act as cancer drivers. Alterations affecting transcription factors in cancer cells tend to result in a loss of the transcription factors’ responsiveness to cellular cues22. This may result, for example, from translocations removing regulatory regions23 or mutations altering sites of post- translational modification24. Consistent with the above definitions, transcription factors and coregulators that contain driver mutations are considered cancer genes. Indeed, transcription factors were among the the first cancer genes to be identified (e.g. TP5325 and JUN26) and comprise a substantial fraction of the known cancer genes. Among all proteins encoded by cancer genes, domains implicated in transcriptional regulation were the second most over-represented27. However, some transcription factor genes have tissue-specific expression patterns, limiting the types of cancers that they may become drivers of17. The interplay between transcription factors and chromatin structures is the focus of the following section.

1.3 Regulation of transcription by chromatin structure

Transcription factors and their cofactors may interact with proteins that alter the packaging of DNA, to make promoter regions more or less accessible to general transcription factor (GTF) binding28,29. In eukaryotic cells, DNA is packaged into chromatin. The fundamental units of chromatin are nucleosomes, each of which contains 147 bp of DNA wrapped around an octamer of core histone proteins: H2A, H2B, H3 and H430. Histones may block access of other

3 proteins, including transcription factors, to DNA. Indeed, only certain “pioneering” transcription factors can bind DNA sequences directly associated with histones31. Pioneer transcription factors may then recruit enzymes that alter chromatin structure in a manner that allows other transcription factors to bind31. Chromatin modifying enzymes include nucleosome remodeling factors, histone modifying enzymes and enzymes that chemically modify DNA. Nucleosome remodeling factors can insert, remove or slide nucleosomes along DNA. Nucleosomes themselves may contain variant histone proteins; nucleosomes immediately up and/or downstream of TSSs tend to include the H3.3 and H2A.Z histone variants, which have less stable DNA interactions and are likely to be more easily displaced by GTFs32. Histone modifying enzymes catalyze the addition or removal of chemical groups at histone residues. Modifications have been detected at over 60 different residues30, and include methylation, acetylation, phosphorylation, deimination, ubiquitinylation, sumolation, ADP ribosylation, β-N-acetylglucosamine addition, proline isomeration, and tail clipping33. Histone modifications are indicated by the histone, the modified residue, the type of group added, and the number of groups added. For example, H3K27me3 indicates trimethylation of lysine 27 of histone 3. Histone methylation and acetylation have been the most extensively studied marks, and are thus focused on in this section. Methylated histone residues are binding sites for proteins containing chromo- or PHD domains. Some methylated residues are binding sites for proteins that promote transcription, whereas others are binding sites for proteins that inhibit transcription30. Histone acetylation occurs on positively charged arginine and lysine residues, eliminating their charge. This reduces the affinity of histones for DNA and tends to increase the accessibility of chromatin34. Additionally, acetylation marks may be bound by bromodomain containing proteins, which typically promote transcriptional activation30. Active promoters tend to be marked by H3K4me3 and histone acetylation (e.g. H3K27ac and H3K9ac), while inactive promoters are typically marked by H3K27me3 or H3K9me335. Promoters marked by both the repressive H3K27me3 and the activating H3K4me3 are considered ‘bivalent’; such promoters are inactive but are poised for activation when the H3K27me3 is removed36. Notably, H3K4me3 also marks the first exon- intron boundary and shorter first exons have been associated with higher expression levels37. Enzymes that produce H3K4me3 include SET1 proteins38,39, KMT2A40 and KMT2B41, whereas

H3K9ac and H3K27ac can be produced by GNAT family, MYST family, p300 and CREBBP proteins42. The most well-studied chemical modification of DNA is methylation of a cytosine residue that precedes a guanine residue (i.e. a CpG dinucleotide). Methylation of CpGs is associated with transcriptional repression, as it may block transcription factor binding or recruit methyl cytosine binding proteins that interact with co-repressors43. While most CpG dinucleotides in the human genome are methylated44, clusters of CpG dinucleotides, termed CpG islands, tend to escape methylation45,46. Hypomethylated CpG islands are associated with roughly half of the human protein coding genes and are characteristic of ‘broad’ promoter structures47,48. These promoters tend to have open chromatin structures that allow transcription initiation35,49. However, a subset of CpG islands are methylated, typically as part of genomic imprinting and X chromosome inactivation mechanisms50. Interestingly, deletion of an enhancer region can result in changes in chromatin structure across large domains51,52. Promoter activity may thus, in some cases, be modulated by chromatin structures that have spread from enhancer regions28,51. Spreading of chromatin states may result from complexes ‘tracking’ along intervening chromatin53 or through multiple iterations of modified histone residues serving as binding sites for proteins that then modify neighboring histones54. The specificity of interactions between regulatory sequences and particular promoters may be contributed to by the existence of DNA elements termed barrier insulators, which block the spread of chromatin states55. Chromatin structures impact the activities of promoters and other regulatory regions. The modification of chromatin structure is thus one mechanism that may mediate the effects of a transcription factor on its many target genes. Alterations of histone modifiers have been implicated in the development of numerous cancer types42, including non-Hodgkin lymphomas1.

1.4 Non-Hodgkin lymphomas

NHL is the fifth most common cancer diagnosed in Canada56, representing 3-4% of all cancer cases worldwide57. The two most common types of NHL are diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL), respectively comprising 31% and 22% of NHL cases internationally58. A more recently recognized type of NHL59, mantle cell lymphoma (MCL), represents 2-10% of NHL cases in the United States and Europe58,60,61. MEF2B

5 mutations identified in FL, DLBCL and MCL are characterized in Chapters 4 and 5. Clinical and molecular features of FL, DLBCL and MCL are reviewed in the following sections.

1.4.1 Clinical characteristics of DLBCL, FL and MCL

FL, DLBCL and MCL are all examples of NHLs arising from B-cells. Their median ages of onset are in the late 50s or 60s62, most commonly presenting with lymphadenopathy (enlarged lymph nodes). B-symptoms (fever, weight loss and drenching night sweats) may also be present. Other symptoms tend to relate to the locations of disease. Overall, patient presentation can vary considerably depending on the stage of disease and site of origin. FL, DLBCL and MCL are primarily distinguished histologically, are managed differently and have different survival rates. FL is relatively indolent: The median survival of patients presenting with advanced stage FL is ten years62. Over 90% of patients have advanced stage disease at presentation63. FL is characterized by painless and slowly progressive diffuse lymphadenopathy with disseminated disease involving the bone marrow and spleen64. The current approach is typically to wait until the patient becomes symptomatic and then treat with rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) chemotherapy64. FL may transform into a more aggressive lymphoma, usually similar to DLBCL62. A cumulative rate of transformation of 2-3% a year has been suggested65,66. DLBCL is an aggressive lymphoma, typically presenting with rapidly enlarging lymphadenopathy62. Approximately half of patients present with advanced stage disease62,67. About one third of patients present with B-symptoms and up to 40% of patients present with disease in extranodal sites, most commonly in the gastrointestinal tract62. The cornerstone of treatment for DLBCL is R-CHOP therapy67. Gene-expression profiling and clinical characteristics have divided DLBCL into multiple subtypes, two of which are the germinal centre B-cell like (GCB) and activated B-cell like (ABC) DLBCLs68–70. The five-year survival rate is ~80% for GCB DLBCL but only 50% for ABC DLBCL71. The distinction between GCB and ACB is not yet used clinically72. However, lenalidomide has been found to have greater activity against non-GBC than GCB DLBCL73 and is currently in clinical trials for DLBCL74,75. Thus, differentiating these subtypes may become useful for treatment determination. The clinical course of MCL varies depending on subtype, ranging from indolent to extremely aggressive59. Median overall survival in MCL is three to four years76. 70-95% of

6 patients present with advanced stage disease62. Presentation frequently includes generalized lymphadenopathy and involvement of the bone marrow, gastrointestinal tract, spleen or liver62. Younger patients tend to be treated with rituximab plus high dose cytarabine chemotherapy followed by autologous stem cell transplant77. Older or less fit patients may be treated with R- CHOP or rituximab combined with less intensive chemotherapy regimens77. FL and MCL are typically considered incurable78,79 and DLBCL remains incurable in approximately 40% of patients80. The search for more effective treatments has involved identifying driver genes and designing molecules to target their activities. The function of driver genes is best understood in the context of the normal biology of the cells from which these cancers originate. These cell types are reviewed in the following section.

1.4.2 The cellular origin of DLBCL, FL and MCL

B-cells are a central component of the adaptive immune response as they can be induced to differentiate into plasma cells, which produce antibodies81. Effective adaptive immunity requires the capacity to generate an enormous diversity of antibodies that each bind to unique antigens. Antibody diversity results from diversity in the transcribed sequences of the immunoglobulin genes. A central feature of B-cell development is thus the generation of variability in immunoglobulin genes and selection for cells that will produce antibodies specific to foreign antigens. In the bone marrow, hematopoetic stem cells develop into pro-B-cells, pre-B-cells and finally immature B-cells82. During these stages immunoglobulin gene rearrangements occur. Immunoglobulins are expressed on the surface of immature and mature B-cells as part of their B- cell receptors (BCRs). Immature B-cells whose BCRs bind self-antigens are selected against, whereas the remaining B-cells are allowed to migrate from the bone marrow to peripheral lymphoid organs such as the spleen, lymph nodes and mucosa-associated lymphoid tissue (MALT)81. At these sites, exposure to antigens may induce further differentiation into either marginal zone or follicular B-cells. Marginal zone B-cells exist primarily in the spleen and MALT in zones surrounding follicular B-cells83. Marginal zone B-cells can be stimulated by blood borne antigens to develop into short-lived plasma cells that secrete relatively low affinity antibodies83,84. The development of marginal zone B-cells into plasma cells does not require interactions with T-cells and may act

7 as a first line of defense against pathogens84. In contrast, follicular B-cells require interactions with both antigens and T-helper cells to become activated83. When activated, follicular B-cells divide rapidly and produce clusters of cells known as germinal centres (Figure 1.1). Within germinal centres, cells derived from the activated B-cells cycle between centroblast and centrocyte states. Centroblasts are rapidly dividing cells that undergo somatic hypermutation85, a process through which activation-induced cytidine deaminase (AID) induces point mutations in the immunoglobin variable region86,87. AID and other enzymes also promote class switching, the recombination of the constant region of the antibody heavy chain to produce IgG, IgA or IgE class antibodies from previously IgM antibodies87,88. Class switching allows antibodies to interact with different effector molecules89. Centroblasts occasionally become centrocytes, non-dividing cells that present antigen to T-cells83. However, this T-cell interaction can only occur if the centrocytes’ BCR has high antigen affinity. Centrocytes failing to interact with T-cells undergo apoptosis83, whereas centrocytes successfully presenting antigen to T-cells may differentiate into memory B-cells or long-lived plasma cells that produce high affinity antibodies. Alternatively, centrocytes may revert to centroblasts for further somatic hypermutation and class switching90. Different types of lymphoma are thought to arise from different types of B-cells (Figure 1.1). ABC DLBCL was defined as a distinct subtype based on a gene expression signature similar to that of plasma cells91,92 and is thus thought to arise from post-germinal centre B-cells undergoing differentiation into plasma cells. Full differentiation of ABC DLBCL cells into plasma cells is blocked by mutations affecting regulators of differentiation such as BLIMP1 (also called PRDM1)93. GCB DLBCL and FL are both thought to arise from germinal centre B-cells (GC B-cells), as they show signs of somatic hypermutation and have expression profiles most similar to GC B-cells70,94. MCL was originally thought to arise from antigen naïve, pre-germinal centre follicular or marginal zone B-cells, as MCL shows lower rates of immunoglobulin gene mutation than FL95,96. However, up to 40% of MCL cases show immunoglobulin gene mutations96 and many cases express cell trafficking genes upregulated by antigen exposure97. These data indicated that at least some MCLs may originate from antigen experienced B-cells or memory B-cells98.

1.4.3 Driver genes of FL, DLBCL and MCL

FL, DLBCL and MCL are each characterized by particular genetic alterations, many of which affect genes in the regulatory networks shown in Figure 1.1. A hallmark genetic alteration of FL is the t(14;18)(q32;q21) translocation, present in up to 85% of cases99. This translocation places BCL2 under control of an immunoglobulin gene enhancer, resulting in constitutively high BCL2 expression100,101. BCL2 activity allows GC B-cells to escape apoptosis regardless of BCR affinity for antigen (Figure 1.1), allowing cells to accumulate and recirculate through germinal centres102. As translocations activating BCL2 expression have also been detected at low frequencies in healthy individuals103, additional alterations must be necessary for FL development. Other alterations frequent in FL are those affecting the histone methyltransferase gene KMT2D (also known as MLL2, altered in 89% of cases)1 and the cell-cell signaling gene EPHA7 (altered in 72% of cases)104. Interestingly, FL cells can alter the activity of T-cells in their microenvironment in ways that favor FL cell survival and proliferation. These changes in microenvironment may be mediated by additional driver mutations105,106. The t(14;18)(q32;q21) translocation that is characteristic of FL is also frequent in GCB DLBCL, occurring in 34% of cases107. This translocation may be advantageous to the development of both GCB DLBCL and FL because both arise from GC B-cells70,94. Similarly, alterations affecting the histone methyltransferase gene EZH21, the Gα protein gene GNA131 and the tumor necrosis factor gene TNFRSF141,108 are recurrent in both FL and GCB DLBCL. Alterations recurrent in GCB DLBCL but not FL include translocations affecting MYC109 (Figure 1.1), deletion of the PTEN tumor suppressor110, mutation of TP53110 (Figure 1.1) and amplification of REL111. All of these alterations are much more frequent in GCB DLBCL than ABC DLBCL1,109,111. ABC DLBCL is instead characterized by mutations constitutively activating NF-κB signaling (Figure 1.1). NF-κB signaling inhibits apoptosis112 and is required for survival of ABC DLBCL cells but not GCB DLBCL cells113,114. As NF-κB activity is activated downstream of BCR signaling, alterations affecting BCR signalling are also common in ABC DLBCL112. Pro-differentiation effects of NF-κB signaling113,114 are typically blocked in ABC DLBCL by alterations that interfere with a central mediator of plasma cell differentiation, BLIMP1 (Figure 1.1)115,116.

Genetic lesions in common between GCB DLBCL, ABC DLBCL and FL include those affecting three genes involved in histone modification, KMT2D, CREBBP and EP3001. KMT2D is also mutated in 14-20% of MCL4,117. Genetic lesions affecting these genes have been predicted to alter levels of histone modifications1. However, the mechanisms by which changes in histone modification may drive lymphoma development remain unclear. Inactivating mutations affecting the acetyltransferase genes CREBBP and EP300 may also contribute to lymphoma development by reducing acetylation of p53 and BCL6, as deacetylation inactivates the p53 tumor suppressor and activates the BCL6 oncoprotein118. BCL6 is a transcriptional repressor involved in regulating cell differentiation, cell death and cell-cycle progression112 (Figure 1.1). Alterations increasing BCL6 activity are more common in GCB DLBCL than ABC DLBCL but occur in both subtypes119. Most cases of MCL have a t(11;14)(q13;q32) translocation which leads to aberrant cyclin D1 expression120. Cyclin D1 promotes cell cycle progression121 but can only drive lymphoma development in mouse models when other driver mutations are also present, such as those in Myc122,123. Whole exome sequencing of 56 MCL cases allowed identification of 37 recurrently mutated genes, including the known cancer genes ATM, TP53, and RB1117. Overall, many of the candidate drivers of DLBCL, FL and MCL are regulators of transcription. As the research described in this thesis investigates mutations in a transcription factor gene, the following section reviews techniques for the functional characterization of transcription factors.

1.5 Technologies and model systems for characterizing transcription factors

1.5.1 Technologies for identifying transcription factor target genes

Target genes are considered to be all genes whose expression can be altered by a change in the activity of the transcription factor of interest, in a given cellular context. The most commonly used technologies for assessing transcriptome-wide changes in gene expression are gene expression microarrays and whole transcriptome sequencing (RNA-seq), both of which were used in my research. The former involves production of labelled cDNA from mRNA transcripts, followed by hybridization of the cDNA library to oligonucleotide probes that are fixed to a substrate124. A set of probes complementary to each cDNA species to be detected is

10 present, and their location within the array of probes is known. If the label on the cDNA is biotin rather than a fluorophore, the cDNA bound to the array is stained with fluorescently labelled streptavidin. The intensity of fluorescence at a given location is considered proportional to the number of cDNA molecules bound, which is in turn assumed proportional to the abundance of the mRNA species from which that cDNA was produced. However, signals from cDNAs with very low abundance tend to be masked by background noise from non-specific hybridization125. Moreover, the affinity of a probe for its intended target is affected by many factors, including the probe’s sequence composition, the presence of sequence variants in the target, and competition from cross-hybridizing targets125. The resulting difference in affinity between probes introduces error into comparison of the abundance of different targets. To compensate for this source of error, arrays typically use numerous probes per gene, whose differences in affinity presumably average out. However, different regions of genes may truly have different abundance if they are part of differentially expressed isoforms. Multiple strategies have been proposed for producing optimal groupings of probes from which to combine signals126–128. However, probe set design is limited by incomplete or inaccurate annotations125. An alternative technique, RNA-seq, involves sequencing a library of cDNAs produced from RNA transcripts. The number of sequencing reads mapping to each gene is considered proportional to the abundance of that gene’s mRNA. Expression data generated by microarrays and RNA-seq is generally concordant129–131. However, unlike expression microarrays, RNA-seq allows the abundance of novel transcripts and transcript isoforms to be assessed and allows detection of sequence variants. Compared to microarrays, RNA-seq allows more sensitive detection of transcripts with low expression and more accurate detection of transcripts with high expression125,132. Using RNA-seq also has drawbacks. One of the primary reasons why expression microarrays are currently the tool of choice for most studies is that generating RNA-seq data is typically more expensive and time consuming than generating expression microarray data125. There are also some technical short-comings of RNA-seq. For instance, reads from some genomic regions tend not to map to single loci. Such ambiguous mapping can arise from the sequence similarity in paralogous gene families or from low complexity sequence regions125. Moreover, short genes are less likely to be identified as differentially expressed than longer genes. This is because fewer reads tend to be acquired from shorter transcripts, reducing

11 statistical power for identification of differential expression133. Furthermore, biases in the sampling and sequencing of transcripts result from differences in GC-content134, differences in the primers used for library construction135 and tendencies for fragmentation at certain sites136. Bioinformatic methods have been developed to reduce the impact of these biases134–136, though tools for analysis of RNA-seq data are arguably less mature than those for expression microarrays125. RNA-seq is predicted to become more widely used than expression microarrays as library construction, sequencing and analysis methods improve125.

1.5.2 Distinguishing direct and indirect target genes

Neither expression microarray nor RNA-seq experiments distinguish a transcription factor’s direct target genes from its indirect target genes. Identification of candidate direct target genes typically involves identifying interactions between transcription factors and regulatory sequences. One approach to doing so is to measure the affinity of a transcription factor to many different short sequences in vitro and use these measurements to predict binding sites throughout the genome137–139. An alternative approach is chromatin immunoprecipitation (ChIP). In preparation for ChIP, cells are formaldehyde treated to crosslink proteins to the DNA segments with which they interact140. The chromatin is then sheared into fragments. Fragments containing the transcription factor of interest are immunoprecipitated using an antibody specific to that transcription factor141. The relative abundance of different genomic regions among the immunoprecipitated DNA may then be determined using quantitative PCR (ChIP-qPCR), an array of probes (ChIP-chip) or high-throughput sequencing (ChIP-seq)142. ChIP has an advantage over in vitro affinity calculations in that it captures DNA binding events in a more biologically relevant context. Specifically, competing cellular proteins and a normal chromatin structure are present in ChIP but not in in vitro assays141,143. Moreover ChIP, but not affinity-based assays, can identify indirect interactions of transcription factors with DNA144. A comparison of ChIP-chip and ChIP-seq found that ChIP-seq required less input DNA, involved less handling of physical devices and was more cost effective145. For these reasons, the research described in this thesis used ChIP-seq to identify MEF2B binding sites. However, ChIP-seq is affected by many of the limitations of next-generation sequencing discussed in the previous section142

Once binding sites have been identified, it is commonly assumed that the genes nearest to the transcription factor binding sites are those regulated by the transcription factor146,147. This assumption has produced candidate direct target gene lists consistent with observed changes in gene expression and cellular phenotypes145,147–149 and is used in the research described in this thesis. However, estimates of what proportion of enhancers interact with the nearest gene vary from 22%150 to 84%151. Thus, long distance regulatory interactions that are inferred based on proximity have a limited degree of accuracy. Interestingly, the binding of a transcription factor to a given site can contribute to the regulation of multiple genes. Indeed, one study found that 25% of enhancers interacted with two or more promoters151. Moreover, genes can be regulated by transcription factors bound at multiple different enhancers. Specifically, the transcription start sites (TSSs) of expressed genes in lymphoblastoid cells interacted with, on average, 1.88 other chromatin regions150. Thus, in the analysis approach I used, the binding of a transcription factor to a single site may be predicted to contribute to the regulation of multiple genes, and transcription factor binding at multiple different sites may be predicted to regulate a single gene. Even if a candidate transcription factor binding site is within a well characterized regulatory region, that site may not be involved in the regulation of gene expression. For instance, only 70% of candidate transcription factor binding sites identified in promoters using ChIP-seq contributed to promoter activity when tested using luciferase assays152. A more high- throughput approach to identifying the binding sites that contribute to gene regulation involves first identifying genes whose expression is affected by the overexpression or knockdown of the transcription factor of interest. One may then assume that at least some of the candidate transcription factor binding sites that are near the differentially expressed genes are involved in regulating gene expression. This approach was used in the research described in this thesis. Examples of how target gene identification has been used in other studies of candidate cancer drivers are discussed in the following section.

1.5.3 Target gene identification as a means of characterizing how transcription factor mutations may contribute to lymphoma development

Even prior to the identification of target genes on a genome-wide scale, the identification of single target genes helped generate hypotheses about which hallmarks of cancer a candidate

13 driver mutation may promote. For instance, the findings that IRF8 promotes BCL6 expression and that BCL6 represses a regulator of plasmacytic differentiation, BLIMP1, led to hypotheses that IRF8 overexpression promotes DLBCL development by blocking differentiation of GC B- cells into plasma cells93,153,154. The finding that ETS1 also inhibits BLIMP1 expression indicated that ETS1 overexpression may also promote DLBCL development by blocking differentiation155. As a further example, the finding that CIITA regulates the gene encoding MHCII, a cell surface protein recognized by T-cells156, led to the hypothesis that CIITA inactivation allows DLBCL cells to escape immuno-surveillance157. Some studies identifying target genes on a genome-wide scale have illustrated that genetic alteration of a transcription factor gene may promote multiple hallmark properties of cancer. For instance, ChIP-chip and gene expression profiling were used to identify genes regulated by the MCL candidate driver gene SOX11. The target genes were found to include genes involved in differentiation158 and angiogenesis159. These findings led to demonstrations that overexpression of SOX11 inhibits plasmacytic differentiation158 and has pro-angiogenic effects on nearby endothelial cells159. Similarly, genes dysregulated by mutations in the candidate driver gene EZH2 were found using ChIP-chip and global gene expression profiling to include regulators of cell-cycle progression and plasmacytic differentiation. Effects of EZH2 mutation on both of these processes were then validated160,161. In light of these studies, I hypothesized that ChIP-seq and global gene expression profiling could also be used to identify MEF2B target genes and that analysis of MEF2B target genes could indicate cellular phenotypes affected by MEF2B activity. However, the identification of target genes requires the use of model systems in which the abundance or activity of a transcription factor can be altered. Factors considered in choosing a model system are discussed in the following section.

1.5.4 Model systems for examining cancer gene function

Numerous model systems have been used for the functional characterization of candidate driver mutations. A proposed order of common model systems from most to least biologically relevant is (a) animal models, (b) human tissue explants, (b) other eukaryotic model organisms, (c) cultured primary cells, (d) cell lines and (e) purified biomolecules162. Working with more biologically relevant model systems tends to involve greater experiment duration, cost and

14 labor162. Using the simplest model with sufficient biological relevance generally allows the largest number of variables to be explored with a given amount of resources. One advantage of using cultured cells over animal models is that the cells can be human. However, selecting for cells that grow in culture may select for mutations and epigenetic states that were either very rare in the original population or that occurred de novo during selection163. As uncontrolled survival and proliferation are selected for in both cell line generation and cancer development, immortalized cell lines are more likely than primary cells to already have mutations in the pathways that a candidate cancer driver mutation would have affected. Such mutations may mask a driver mutation’s effects. The most biologically relevant cell line models are those derived from the cell type that the cancer type of interest originates from. Cell type may influence not only whether a mutation acts as a driver, but also whether a driver gene functions as an oncogene or a tumor suppressor164. Dozens of cancer genes, including regulators of gene expression, have been found to act in opposite ways in cancers originating from different cell types165. For instance, the H3K27 methyltransferase gene EZH2 acts as an oncogene in non-Hodgkin lymphoma166 and as a tumor suppressor in myeloid malignancies167. However, some cell types are not amenable to growth in culture or genetic manipulation, and not all cancers have a known cell of origin. These factors have led to studies using experimentally tractable cell lines relatively unrelated to the cancer type of interest. For instance, human embryonic kidney 293 (HEK293) cells have been used to characterize mutations identified in oligodendroglioma168. As DLBCL cells are not readily amenable to manipulation, I used a cell line that was more tractable than DLBCL cell lines for the research described in Chapters 3 and 4. I chose to use HEK293 cells, for the following reasons. First, HEK293 express mRNA for coregulators known to associate with MEF2 proteins, including CABIN1, HDAC4, -5, -7 and -9, p300, CREBBP and KMT2D (data from GSM743775 in the Gene Expression Omnibus database). Thus, activities of MEF2B that are affected by these coregulators may be modeled in HEK293 cells. Second, HEK293 cells have very low endogenous MEF2B expression. Therefore, transfection with a MEF2B expression construct is likely to produce a large fold change in MEF2B abundance, increasing the likelihood that downstream changes will be of a detectable magnitude. Third, the only prior study of the MEF2B mutations that were investigated in my research primarily used HEK293 cells7. Using the same

15 model system as the previous study allows my data to be comparable with the previous study. Finally, the genome of HEK293 cell lines has been well characterized using whole-genome resequencing169. The copy number and single-nucleotide polymorphism profiles of HEK293 populations were found to remain in a steady state during standard culturing169. Such population stability increases the likelihood that findings will be reproducible. However, I recognise that a challenge to interpreting results from HEK293 cells is that the cell type of origin of HEK293 cells is unclear. Despite being derived from adenovirus transformation of human embryonic kidney tissue170, HEK293 cells express neuron-specific markers and thus may have originated from neural cells171 or neural crest derived adrenal precursor cells169. My research used HEK293 cells to characterize the MEF2B regulatory network. MEF2B belongs to the MEF2 family of transcription factors, described in the following sections.

1.6 MEF2 family proteins

The myocyte enhancer factor 2 (MEF2) family of human transcription factors consists of four proteins, MEF2A, -B, -C and –D, each of which has a homolog in other vertebrates172. MEF2A and -C have the most similar sequences, likely resulting from a duplication event that occurred near the origin of vertebrates173. In contrast, MEF2B appears to be the first of the MEF2 family to have diverged from a single ancestral MEF2 gene173. The commonly used model organisms S. cerevisiae, C. elegans, and D. melanogaster contain only one MEF2 family gene172. All MEF2 proteins contain three domains: an N-terminal DNA-binding MADS domain, a central MEF2 domain and a C-terminal transactivation domain172. The MADS and MEF2 domains are well conserved across the MEF2 family, with 91% and 68% amino acid identity, respectively, between MEF2A and -B172. The transactivation domain is less well conserved, with only 6% amino acid identity between MEF2A and -B172.

1.6.1 Roles of MEF2 proteins in vertebrate organisms

MEF2 family proteins play central roles in the differentiation, morphogenesis, and maintenance of several vertebrate tissue types. Functions of MEF2A, -C and –D are the focus of this section, as these functions may be similar to functions of MEF2B that have yet to be identified.

The MEF2 proteins were named myocyte enhancer factors because of their roles in muscle cell differentiation174. MEF2A cooperates with other factors to promote skeletal muscle differentiation175–177 and MEF2C promotes proper differentiation of mouse smooth muscle cells178. Multiple MEF2 proteins are likely involved in cardiac muscle differentiation, as competitive inhibition of binding to MEF2 sites impaired cardiac muscle differentiation179, but knockouts of individual Mef2 genes did not180–182. Other activities of MEF2A, -C and –D relate to regulation of cytoskeletal structures. For instance, deletion of Mef2c in skeletal muscle cells resulted in sarcomere disorganization183 and knockdown of MEF2A in neurons impaired dendrite morphogenesis184. MEF2 proteins also contribute to the formation of large-scale tissue structures. For instance, Mef2c null mice have cardiac looping defects180 and mice heterozygous for Mef2c deletion had decreased ossification and impaired chondrocyte hypertrophy in the sternum182. Neural crest-specific deletion of Mef2c produced defects in craniofacial morphogenesis185. The formation of craniofacial structures requires neural crest precursor cells to undergo epithelial-mesenchymal transition (EMT) and migrate to appropriate locations186. As MEF2A, –C and –D promoted EMT of hepatocellular carcinoma cells187, they may also promote EMT of the neural crest cells. Finally, MEF2C and -D have roles in regulating apoptosis. MEF2C activity in endothelial cells188 and neurons189,190 inhibits apoptosis downstream of mitogen-activated protein kinase signaling. In contrast, in T-cells MEF2C and -D promote apoptosis downstream of T-cell receptor signaling191–193. These cell-type specific differences may be related to tissue specific differences in the sites where MEF2 proteins bind DNA. In neuronal cells compared to muscle cells, MEF2A showed additional constraints for DNA-binding based on the sequences flanking MEF2 motifs194.

1.6.2 Expression patterns of MEF2B

Though the functions of MEF2B have not been well characterized, they may be inferred by identifying the cell types in which MEF2B mRNA is expressed. Similar to other Mef2 genes, Mef2b mRNA transcripts have been detected in developing cardiac muscle cells, skeletal muscle cells, neurons, neural crest cells, whisker follicle cells and chondrocytes of mouse embryos195. Mef2b mRNA expression has also been reported in the proliferating smooth muscle cells of injured rat arteries196 and human MEF2B mRNA expression has been reported in B-cell and T-

17 cell lymphomas197. Further investigation of MEF2B expression in B-cells has indicated that MEF2B is expressed in GC B-cells but not in naïve B-cells7 or in B-cells that have differentiated into plasma cells197. The cell types containing appreciable levels of MEF2B protein may be a subset of those expressing MEF2B mRNA, as the translation of MEF2 mRNAs may also be regulated198. The presence of MEF2B protein has been confirmed in GC B-cells, fibroblasts, myoblasts, myotubes and vascular smooth muscle cells7,195,196. Thus, MEF2B, like its paralogs, may contribute to development and maintenance of a variety of tissue types. Of the mouse MEF2 proteins, MEF2C had the most similar expression pattern to MEF2B and may thus have the most similar cellular functions195. Consistent with the notion that other MEF2 proteins can compensate for the loss of MEF2B activity during development, Mef2b null mice were viable and had normal heart development180. However, deletion of Mef2a181 or Mef2c199 in mice is lethal, indicating that MEF2B cannot perform some functions that are normally performed by MEF2A or MEF2C.

1.6.3 Target genes of MEF2 family proteins

Attempts to better understand the role of MEF2 proteins have included attempts to identify MEF2 target genes throughout the genome. The target genes identified for different MEF2 proteins in different cell types tended to be enriched for different functional annotation groups. For instance, a study of MEF2A DNA-binding sites in cardiomyocytes reported that candidate direct target genes were enriched for functions related to heart and muscle development and cytoskeleton organization200. In contrast, a study of MEF2A and -D in hippocampal neurons identified target genes that tended to have functions at neural synapses and expression only in central nervous system cells201. A study of the role of MEF2C in bone formation found that genes associated with MEF2C binding sites were enriched for genes that regulate bone turnover202. Only one study other than my own has identified candidate MEF2B target genes throughout the genome177 . This study, published in January 2015, contrasted the activities of the four mouse MEF2 proteins in myoblasts. Genes whose expression levels were altered by reducing levels of a MEF2 protein were considered candidate target genes of that MEF2 protein. The numbers of candidate target genes ranged from 110 for MEF2D to 4,020 for MEF2A. Of the candidate target genes for one MEF2 protein, 10% to 81% were not candidate target genes of any

18 other MEF2 protein. These differences between target gene sets are consistent with evidence that each MEF2 protein has some cellular functions distinct from those of the other MEF2 proteins (section 1.6.1). Of the 1,057 candidate MEF2B target genes, 12% were not candidate target genes of any other MEF2 protein. Gene ontology terms enriched in the MEF2B target genes related to blood vessel morphogenesis, sarcomere structure and transcription factor activity203. However, the study discussed above did not attempt to distinguish direct from indirect target genes, did not investigate the predicted effects of altered MEF2B activity on cell behavior, and was performed in a mouse rather than human cell line203. Prior to the research described in this thesis, the only validated direct MEF2B target genes were SMHC204 (a smooth muscle myosin gene), BZLF1205 (an Epstein-Barr virus gene involved in reactivation), SOST206 (a Wnt inhibitor) and BCL67 (a transcriptional regulator in B-cells). These target genes may be unique to MEF2B, as MEF2B was the only MEF2 protein to bind a region required for maintaining SMHC expression204 and MEF2B overexpression but not MEF2D overexpression increased BZLF1 transcription205. Similarly, MEF2B overexpression but not MEF2A or -C overexpression increased SOST expression downstream of the ECR5 enhancer region206. A MEF2A binding site near BCL6 has been identified207 but impacts of MEF2 proteins other than MEF2B on BCL6 expression have not been investigated. Conversely, MEF2B may not regulate some genes that are direct targets of other MEF2 proteins, as MEF2B was the only MEF2 protein that did not bind regulatory sequences near NUR77208 or the immunoglobulin J chain gene209.

1.6.4 Functions of the MADS and MEF2 domains in MEF2 proteins

The MADS box is a region of 56 amino acids highly conserved across the MADS family proteins210, whereas the MEF2 domain is a 29 amino acid region unique to MEF2 family proteins211. Both the MADS and MEF2 domains are required for DNA-binding212. Notably, MEF2 proteins bind DNA as dimers and the MADS and MEF2 domains are essential for dimerization212. Complexes thought to represent MEF2A-MEF2D heterodimers, MEF2C- MEF2D heterodimers and MEF2D homodimers have been identified in HEK293 cells213. One notable difference between the MADS domain of MEF2B and other MEF2 proteins is the presence of glutamine (Q) rather than glutamic acid at residue 14 in MEF2B. Mutation of Q14 in MEF2B to glutamic acid increased DNA binding by approximately twofold195, consistent with

19 the notion that MEF2B may have slightly reduced affinity for DNA binding sites when compared to other MEF2 proteins. Dissociation constants for DNA-binding have been determined for MEF2A214 and MEF2C215,216 but not for MEF2B. However, these studies used MEF2 proteins expressed in bacteria, where the MEF2 proteins escaped post-translational modifications that modulate their DNA binding affinity. For instance, casein kinase II phosphorylates S59 of MEF2C, a modification that enhances MEF2C DNA binding by five-fold217. Similarly, acetylation of K4 increases MEF2C DNA binding218. S59 and K4 are conserved in MEF2A, -B and -D, though their post-translational modification has only been investigated in MEF2C. The MEF2 domain is also involved in interactions with co-activators and co-repressors. Co-repressors that are thought to associate with the MEF2 domains of all MEF2 family proteins include the class IIa histone deacetylases HDAC4, -5, -7 and -9219–222. Although class IIa HDACs have minimal deacetylase activity223,224, they can mediate transcriptional repression by recruiting other co-repressors such as HP-1, CtBP, MITR and class I HDACs 224–226. Another co- repressor interacting directly with the MEF2 domain is CABIN1192,227. CABIN1 also interacts with class I HDACs192 and can interact with the H3K9 methyltransferase SUV39H1228. Co- activators binding the MEF2 domains of MEF2A229, -C230 and -D192,231 include the histone acetyltransferases CREBBP and p300, which are structural and functional homologs232. As p300 binds MEF2A at the same interface as HDAC9 and CABIN1, which have been crystallized interacting with MEF2B219,227, p300 and CREBBP may also interact with MEF2B. Given these interactions with histone modifying enzymes, MEF2 proteins may alter expression of their target genes by promoting changes in histone modification. Alternatively, HDACs and p300 may instead modulate MEF2 target gene expression by altering acetylation states of MEF2 proteins themselves. Deacetylation of MEF2D by HDAC4 can allow MEF2D to be sumoylated233. Sumoylation inhibits the capacity of MEF2 proteins to activate transcription233. Conversely, p300 can acetylate MEF2C, promoting MEF2C’s transcriptional activity218,234. p300 may also play a structural role linking MEF2 proteins to other transcription factors and transcriptional machinery, as p300 can interact with basal transcription factors and RNA polymerase II235. Because their binding sites on MEF2 proteins overlap, CABIN1, class IIa HDACs, CREBBP and p300 may compete to bind MEF2 proteins192. Indeed, decreased interaction of

CABIN1 with MEF2D correlated with an increase in the interaction of MEF2D with p300192. Association of HDACs and CABIN1 with MEF2D is inhibited by increased intracellular calcium levels192. Specifically, high calcium levels promote nuclear export of CABIN1 and class II HDACs236–238 and cause CABIN1 and class II HDACs to be sequestered into complexes with calcium-calmodulin192. Regulation of these co-repressors by calcium may explain why the expression of some MEF2 target genes is calcium sensitive. For instance, Nur77 expression in T- cells is dependent on the presence of MEF2 binding sites and is induced by calcium signaling239. Similarly, treatment with a calcium ionophore increased MEF2-dependent luciferase expression in T-cells193. MEF2-dependent gene expression may also be regulated by calcium signaling downstream of B-cell receptor (BCR) activation in B-cells. Consistent with this hypothesis, most gene expression differences between mice with B-cell specific Mef2c deletion and control mice were evident only when BCR signaling was activated240. Specifically, activation of BCR signaling tended to increase MEF2C target gene expression in control B-cells but not in MEF2C deficient B-cells. However, BCR signaling involves multiple signal transduction pathways, including activation of p38 via protein kinase C (PKC)241. p38 can phosphorylate all MEF2 proteins, including MEF2B213. In muscle cells, phosphorylation by p38 promotes association of MEF2C and MEF2D with the histone methyltransferase KMT2D (also known as MLL2)242. p38 and calcium signaling may thus contribute synergistically to regulation of MEF2 target gene expression. This notion is supported by evidence that treatment of cells with both the calcium ionophore ionomycin and the PKC activator PMA produced greater MEF2-dependent luciferase expression than expected from treatment with either agent alone193. However, interactions of MEF2B with KMT2D have not been investigated. Other transcription factors can also cooperate with MEF2 proteins through interaction with the MADS or MEF2 domains. A well-studied example is the interaction of MEF2 proteins with MyoD transcription factors. MEF2A, -C and -D only induced muscle gene expression in transfected fibroblasts when a MyoD protein was co-expressed176. Once associated with MyoD proteins, MEF2 proteins are thought to promote interactions between MEF2-MyoD complexes and transcriptional machinery198. Interestingly, even though interaction with MyoD proteins occurs through the MADS box of MEF2 proteins, MEF2 DNA-binding activity is not required for induction of a muscle gene expression program176. Similarly, the MADS and MEF2 domains

21 of MEF2 proteins can interact with GATA transcription factors to synergistically activate cardiac-specific gene expression, without requiring MEF2 DNA-binding capacity243. Thus, MEF2 proteins may affect expression of genes without MEF2 binding sites, through interactions with other transcription factors. Interactions between MEF2B and MyoD or GATA transcription factors have not yet been investigated.

1.6.5 Functions of the transactivation domains of MEF2 proteins

The MADS and MEF2 domains of MEF2 proteins are sufficient to recruit certain coregulators (discussed above), but are not sufficient to strongly activate the expression of all target genes. For instance, MEF2C proteins containing the MADS and MEF2 domains but lacking most of their transactivation domain had a reduced capacity to activate expression of a MEF2-dependent reporter gene212. Similarly, deletion of two-thirds of MEF2B’s transactivation domain eliminated MEF2B’s capacity to activate reporter gene expression195, indicating that MEF2B’s transactivation domain is essential for the activation of at least some target genes. Interestingly, the MEF2C transactivation domain was sufficient to activate reporter gene expression even when linked to the DNA-binding domain of yeast GAL4 instead of the MADS and MEF2 domains212. Thus, the coregulators that interact with the MADS and MEF2 domains (see section 1.6.4) are not essential for MEF2C to activate the expression of some target genes. Rather, coregulators recruited by the MADS and MEF2 domains may modulate the degree of target gene activation. The mechanisms by which MEF2 transactivation domains activate transcription remain unclear. One possible mechanism is through interaction with the positive transcription elongation factor b (P-TEFb), which hyperphosphorylates the C-terminal region of RNA polymerase II to promote transcription244. P-TEFb was co-immunoprecipitated with MEF2A, -C and -D245 and overexpression of P-TEFb increased MEF2-dependent transcription. Furthermore, P-TEFb was recruited to MEF2 binding sites when MEF2-dependent transcription was activated. However, it has not been determined whether P-TEFb also interacts with MEF2B and it remains unknown which domains of MEF2 proteins are required for interaction with P-TEFb. An additional function of MEF2 transactivation domains is to integrate regulatory signals. Numerous sites of post-translational modification have been identified in the transactivation domains, including phosphorylation sites for p38213,246 (discussed in section

1.6.4), BMK1247 and PKA7,248. BMK1 phosphorylates and activates MEF2A, -C and -D, but not MEF2B247. In contrast, both MEF2B and MEF2D are phosphorylated by PKA7,248. Phosphorylation at some sites in the MEF2B and MEF2D transactivation domains promotes sumoylation at nearby residues. MEF2B sumoylation is associated with decreased MEF2- dependent reporter gene expression7,249. However, sites of post-translational modification are retained in only some isoforms of MEF2 proteins. Multiple isoforms of MEF2 proteins that differ in their transactivation domains are produced through alterative splicing250–252. Two isoforms of MEF2B have been reported, isoform A (mRNA: NM_001145785.1; protein: NP_001139257.1) and isoform B (mRNA: NM_005919; protein: NP_005910.1)7. Isoform A MEF2B includes all exons, whereas isoform B excludes exon 8 (Figure 1.2). The exclusion of exon 8 results in a frameshift that alters all amino acids C-terminal to those encoded by exon 7. Consequently, 40% of the amino acids in the transactivation domain of isoform A are altered in isoform B. Given these differences in transactivation domain sequence, isoform A and B MEF2B may differ in transcriptional activity.

1.7 Roles of MEF2 family proteins in human disease

Of the MEF2 proteins, MEF2C has been associated with the widest range of disorders. For instance, increased MEF2C abundance has been associated with congenital heart defects253 and was characteristic of immature T-cell acute lymphoblastic leukemia254. In contrast, decreased MEF2C abundance produces MEF2C haploinsufficiency syndrome, a syndrome characterized by intellectual disability, epilepsy, autistic features and abnormal movements255,256. Abnormal movement is also a characteristic of Parkinson’s disease, the toxin-induced form of which has been associated with decreased MEF2C activity257. Decreased MEF2C and -D activity has also been implicated in sarcoma development258. Decreased activity of MEF2A has been associated with an entirely different phenotype, namely an autosomal dominant form of coronary artery disease259. Interestingly, the only non-cancer disease or disorder in which MEF2B alterations have been implicated is intellectual disability. This is based on evidence that MEF2B was co-deleted with 10 or 75 other genes in two patients with intellectual disability260. However, MEF2B alterations have been identified in several cancer types, discussed in the following section.

1.7.1 MEF2B may act as an oncogene in some types of carcinoma

Data from studies in which whole cancer genomes and exomes have been sequenced are accessible through the cBioPortal database261,262. cBioPortal includes reports of alterations affecting MEF2B in numerous cancer types (Figure 1.3). Cancer types in which MEF2B may act as an oncogene include those in which MEF2B was more frequently affected by copy number amplification than by deletion. MEF2B was affected by amplifications in 9% of ovarian carcinomas (28 out of 311 cases, TCGA provisional data), 5% of uterine carcinomas (11 out of 240 cases263), 5% of adrenocortical carcinomas (4 out of 88 cases, TCGA provisional data) and 3% of esophageal carcinomas (6 out of 184 cases, TCGA provisional data), yet was not deleted in any cases in these studies. Interestingly, MEF2A, -C and –D were also predominantly affected by amplifications in these studies and 10% to 25% of cases in these studies had an amplification or mutation affecting at least one of the MEF2 genes. These data indicate that MEF2B and other MEF2 genes may act as oncogenes in ovarian, uterine, adrenocortical and esophageal carcinomas.

1.7.2 MEF2B may act as a tumor suppressor in some types of non-Hodgkin lymphoma

In contrast to the cBioPortal data for carcinomas, cBioPortal data for DLBCL indicates that 4.2% of DLBCL cases contained homozygous MEF2B deletions (2 out of 48 cases, TCGA provisional data261,262) and that no cases of DLBCL had MEF2B amplifications. These data indicate that MEF2B may act as a tumor suppressor in DLBCL. Although not included in cBioPortal, numerous somatic non-synonymous and indel mutations affecting MEF2B in DLBCL and other lymphomas have been identified. Specifically, MEF2B is the target of somatic nonsynonymous and indel mutations in 8-18% of DLBCL1–3,5, 13% of FL1 and 3-7% of MCL4,6. All identified MEF2B mutations are shown in Table 1.1 and Figure 1.4. Conflicting results have been reported regarding the subtype distribution of MEF2B mutations within DLBCL. Out of 33 MEF2B mutations identified in DLBCL by Morin et al. (2011) only two were identified in ABC DLBCL. In contrast, Pasqualucci et al. (2011) found four out of eight DLBCL MEF2B mutations to occur in the ABC subtype. None of the MEF2B mutations identified in NHL have been identified in the carcinomas in which MEF2B was amplified (see section 1.7.1). This is consistent with the notion that the

MEF2B mutations in NHL decrease rather than increase MEF2B activity. The MEF2B point mutations and indels in NHL are likely all heterozygous, as samples with MEF2B mutations still contained WT MEF2B mRNA1 and no more than one mutation has been identified per sample. Assuming that MEF2B acts as a tumor suppressor in NHL, MEF2B may either be haploinsufficient or harbor mutations that have dominant negative effects. As MEF2 family proteins dimerize, mutant MEF2B is expected to interact with WT MEF2 proteins in heterozygous cells. MEF2B mutations would have dominant negative effects if dimers composed of mutant and WT MEF2B have less activity than dimers containing only WT MEF2B. Curiously, copy number alterations of other MEF2 genes in DLBCL were either not detected (MEF2A and –C) or were amplifications (MEF2D, 2 out of 48 cases, TCGA provisional data261,262). Similarly, point mutations in other MEF2 genes were either not detected (MEF2A and D) or were much less frequent (MEF2C, mutant in 3.1-3.6% of DLBCL1,2). Both MEF2B and MEF2C were affected by Y69H mutations, consistent with the notion that MEF2B and MEF2C mutations may alter the functions of MEF2B and MEF2C in a similar manner and thus may have similar effects on MEF2 target gene expression. Interestingly, one DLBCL cell line (unpublished data) and two DLBCL patient samples2 have been found to contain both MEF2B and MEF2C mutations. Thus, effects of MEF2B and MEF2C mutations are unlikely to be completely redundant. It is possible that MEF2C mutations have weaker driving effects than MEF2B mutations, such that additional mutations in MEF2 proteins can provide additional selective advantage. Consistent with this notion, one DLBCL cell line contains two different MEF2C mutations (unpublished data) and one DLBCL patient sample2 contains four MEF2C mutations plus a MEF2B mutation. In contrast, no more than one MEF2B mutation has been reported within a single DLBCL sample or cell line1–3. However, a single MEF2C mutation may still provide some selective advantage, as three DLBCL patient samples have been identified with one mutation in MEF2C and no MEF2B mutations2. All reported MEF2C mutations are shown in Table 1.2. Both MEF2B and MEF2C mutations tended to occur in the MADS and MEF2 domains, which are involved in DNA-binding, dimerization and coregulator interactions. Mutations in these domains represented 79% of MEF2B mutations in DLBCL1–3,5, 75% of MEF2B mutations in FL1 and 93% of MEF2B mutations in MCL4,6. Interestingly, MEF2B mutations in DLBCL and FL most often affected three residues within the MADS and MEF2 domains: K4, Y69 and D83

(Figure 1.4). Specifically, out of 91 mutations identified in DLBCL or FL cases, 30 (33%) affected D83, 7 (8%) affected K4 and 6 (7%) affected Y691–3,5. In contrast, MEF2B mutations in MCL were predominantly K23R mutations (10 out of 14 cases)4,6. One K23R mutation has been identified in DLBCL1, consistent with the notion that MEF2B mutations may drive DLBCL and MCL by similar mechanisms. The recurrence of MEF2B mutations at particular residues implies that only when MEF2B is altered in particular ways does it acquire the capacity to drive lymphoma development. This is consistent with the notion that MEF2B mutations have either gain-of-function or dominant negative effects on MEF2B activity. Dominant negative effects would be consistent with the hypothesis that MEF2B acts a tumor suppressor in NHL. All missense mutations in the MADS and MEF2 domains of MEF2B affected residues that are conserved in MEF2C. Effects of some of these mutations may be predicted from effects of similar mutations in MEF2C. Replacement of K4, K5, R15 or R24 in MEF2C with non- charged amino acids reduced DNA binding but did not affect dimerization, resulting in dominant negative effects212. Thus, the K4E, K5E, R15G and R24Q MEF2B mutations are also expected to reduce DNA binding and have dominant negative effects. K4, K5, R15 and R24 are located near the DNA binding interface of MEF2B (Figure 1.5) consistent with the hypothesis that their mutation alters DNA binding. Dominant negative effects resulting from reduced DNA binding were also caused by deletions of residues 59-63, 68-72, 73-76 or 77-80 in MEF2C. MEF2B mutations affecting residues within these regions (i.e. mutations affecting Y69, H76 and E77) are expected to have similar effects. The transactivation domain shows a pattern of mutations distinct from that in the dimerization domains. Out of the 20 transactivation domain mutations, 17 are expected to result in either premature termination, stop codon read-through, a frameshift, or disruption of splicing (detailed in Table 1.1). Consistent with the hypothesis that MEF2B mutations have dominant negative effects, these 17 mutations are expected to decrease the capacity of MEF2B to activate transcription without disrupting dimerization, as they are located C-terminal to the domains involved in dimerization. The remaining three transactivation domain mutations are nonsynonymous mutations whose impacts are difficult to predict. Effects of the MEF2B mutations identified in DLBCL and FL were investigated in one previous study7. This prior study found that the three MEF2B mutations considered most likely to affect dimerization, L38I, L54P and N81Y, did not disrupt dimerization, consistent with the

26 notion that the MEF2B mutations may have dominant negative effects. The same study also found that the L54P, Y69H, E77K, S78R, N81Y and D83V mutations disrupted interactions with the co-repressor CABIN1, whereas the K4E, L38I, G105E, R171X, R207Q and L269fs mutations did not. All but one of the mutations that disrupted CABIN1 interaction increased expression of a reporter gene that was driven by a region of the BCL6 promoter. This may be because CABIN1 no longer blocked interactions of MEF2B with co-activators such as p300192. However, p300 binds the same residues in MEF2 proteins as CABIN1229 and therefore p300 interactions may also be disrupted by the MEF2B mutations. The precise mechanism by which the mutations increased BCL6 reporter activation thus remains unclear. Interestingly, one of the mutations that did not affect BCL6 reporter expression, R171X, was also studied in mouse Mef2b195. In that study, R171X completely abolished expression of a MEF2B-dependent Myh3 reporter construct, a result that was not surprising considering that the R171X mutation prevented 70% of the transactivation domain from being translated212. The lack of effect of R171X on the BCL6 reporter may indicate that expression of the BCL6 reporter is not dependent on activity of the MEF2B transactivation domain. If so, the BCL6 reporter assay may not be sensitive to effects of transactivation domain mutations and may not accurately model effects of MEF2B mutations on all MEF2B target genes. Notably, R171X and Y201X MEF2B mutations did affect BCL6 reporter expression when HEK293T cells were treated with agents that promoted phosphorylation or sumoylation of MEF2B. These treatments decreased BCL6 reporter expression in cells with full-length MEF2B but not in cells with R171X or Y201X MEF2B, presumably because R171X and Y201X MEF2B did not contain the residues that were modified in full-length MEF2B7. However, if R171X and Y201X MEF2B are unable to activate the expression of genes other than the BCL6 reporter because they lack portions of the transactivation domain, they would have a loss of activity regardless of having escaped inhibitory post-translational modifications. Thus, the R171X and Y201X MEF2B mutations may reduce MEF2B’s capacity to activate some MEF2B target genes, consistent with the hypothesis that MEF2B is a lymphoma tumor suppressor. Curiously, one mutation predicted above to affect DNA binding, K4E, as well as two other mutations in the DNA binding domains, L38I and N81Y, did not affect BCL6 reporter expression. This may have been because MEF2B DNA binding was not required for activation of the construct’s expression. Rather, MEF2B may be recruited by another DNA-binding

27 transcription factor. Although the BCL6 promoter region used in the reporter contained a sequence with 84% similarity to MEF2A and MEF2C binding site motifs264, the authors did not investigate whether that sequence was essential for reporter expression or whether MEF2B directly bound that sequence in the cell type that they used for assays on mutant MEF2B. Notably, the 921 bp region of the BCL6 promoter used in the reporter construct also contained sequences with more than 84% similarity to the motifs of 67 other transcription factors, some of which may have mediated interactions with MEF2B. Thus, the BCL6 reporter system may not accurately model effects of MEF2B mutations on genes whose activation does require direct MEF2B DNA binding. MEF2 proteins have been reported to co-operate with other transcription factors to activate gene expression in a manner that does not require MEF2 DNA binding or the MEF2 transactivation domain176,243. As the other transcription factors that may have interacted with MEF2B to activate the BCL6 reporter may not be expressed in GC B-cells and may not affect the activity of MEF2B at its other target genes, the generalizability of the BCL6 reporter assay results is questionable. Altogether, the prior research characterizing effects of MEF2B mutations left open several avenues for future research. Some of those avenues, such as the effect of MEF2B mutations on MEF2B’s interaction with DNA and on the expression of endogenous target genes, are among those investigated by research described in this thesis.

1.8 Thesis roadmap and chapter summaries

As reviewed in this introduction, techniques for identifying DNA binding sites and profiling transcriptomes have been used to identify transcription factor target genes. The characterization of target gene sets has enhanced our understanding of how altered transcription factor activity may promote lymphoma development. Based on this prior research, I hypothesized that ChIP-seq and global gene expression profiling could be used to identify MEF2B target genes. I further hypothesized that analysis of target genes would indicate cellular phenotypes affected by MEF2B activity. Finally, I hypothesized that contrasting the DNA binding sites, effects on gene expression and effects on cellular phenotypes of mutant and WT MEF2B will indicate mechanisms through which MEF2B mutations may contribute to lymphoma development. Therefore, the overarching objective of my research was to characterize

28 and contrast the regulatory networks of WT and mutant MEF2B. Chapter 2 describes my research methods, whereas Chapters 3, 4 and 5 describe the research that addresses major aims related to my overall objective. The aim of the research described in Chapter 3 was to characterize the MEF2B regulatory network in HEK293A cells. WT MEF2B target genes were found to include the cancer genes MYC, TGFB1, JUN, BCL2 and CARD11, and the metastasis suppressors RHOB and NDRG1. Identification of target genes led to findings that WT MEF2B promotes expression of mesenchymal markers and promotes HEK293A cell migration. Increased intracellular calcium levels had minimal effects on most MEF2B target genes and changes in target gene expression did not correlate with changes in histone modification. Chapter 4 describes research aiming to characterize activities of K4E, Y69H and D83V MEF2B and contrast their activities with those of WT MEF2B. I found that the K4E and D83V mutations decreased MEF2B’s capacity to bind DNA, and that the K4E, Y69H and D83V mutations decreased MEF2B’s capacity to promote gene expression. These mutations also reduced MEF2B’s capacity to alter cell migration and mesenchymal marker expression. The research described in Chapters 3 and 4 was performed using HEK293A cells as those cells were an experimentally tractable model system. The objective of Chapter 5 was to characterize the roles of mutant and WT MEF2B in DLBCL cells. Results described in Chapter 5 are consistent with the notion that all MEF2B mutations in NHL decrease the capacity of MEF2B to activate transcription in DLBCL cells. I also found that WT MEF2B tended to inhibit DLBCL cell chemotaxis to a greater extent than mutant MEF2B, and propose that MEF2B mutations may promote DLBCL and FL development by de-repressing chemotaxis. My findings and potential future directions for research are summarized in Chapter 6. Overall, my research provides a unique resource for exploring the role of MEF2B in cell biology. I map for the first time the MEF2B regulome, demonstrating connections between a relatively understudied transcription factor and genes significant to oncogenesis.

Table 1.1 Potentially somatic MEF2B mutations identified in NHL. This table indicates characteristics and predicted effects of each potentially somatic MEF2B mutation identified in NHL. § indicates that at least one mutation at that position was shown to be somatic (i.e. not present in matched constitutional DNA). For mutations at the remaining positions, matched constitutional DNA was not assessed. Impacts on splicing were predicted using the Human Splicing Finder265.

Number of Nucleotide Amino acid identified Affected codon change change occurrences Domain Diagnosis Reference conserved in Notes Disrupts translation start codon. The nearest downstream start codon is in the same frame and its use would produce MEF2B lacking the 28 most N-terminal T>A M1K 1 MADS FL Morin 2011 MEF2A, MEF2C, MEF2D amino acid residues. G>A G2E 1 MADS DLBCL Lohr 2012 MEF2A, MEF2C, MEF2D GCB Morin 2011,

A>G K4E § 7 MADS DLBCL, FL Lohr 2012 MEF2A, MEF2C, MEF2D Acetylated in MEF2C A>G K5E 1 MADS FL Morin 2011 MEF2A, MEF2C, MEF2D A>C K5N 1 MADS DLBCL Lohr 2012 MEF2A, MEF2C, MEF2D A>G I8V 1 MADS DLBCL Morin 2011 MEF2A, MEF2C, MEF2D T>A I8F 1 MADS DLBCL Zhang 2013 MEF2A, MEF2C, MEF2D A>G R15G 1 MADS DLCBL Morin 2011 MEF2A, MEF2C, MEF2D In alpha helix that contacts DNA Morin 2011, Bea 2013, MEF2A, MEF2C, MEF2D DLBCL, Meissner and all MADS family A>G K23R § 11 MADS MCL 2013 members In alpha helix that contacts DNA Morin 2011, MEF2A, MEF2C, MEF2D Meissner and all MADS family G>A R24Q § 2 MADS FL, MCL 2013 members In alpha helix that contacts DNA T>G F26V 1 MADS GCB DLBCL Morin 2011 MEF2A, MEF2C, MEF2D A>C Y33S 1 MADS FL Morin 2011 MEF2A, MEF2C, MEF2D MEF2A, MEF2C, MEF2D Pasqualucci and all MADS family C>A L38I 1 MADS GCB DLBCL 2011 members T>G I43R 1 MADS DLBCL Lohr 2012 MEF2A, MEF2C, MEF2D T>C I47T 1 MADS DLBCL Morin 2011 MEF2A, MEF2C, MEF2D Bea 2013, A>G N49S 2 MADS MCL Meissner MEF2A, MEF2C, MEF2D

Number of Nucleotide Amino acid identified Affected codon change change occurrences Domain Diagnosis Reference conserved in Notes 2013 G>A R53H 1 MADS DLBCL Morin 2011 MEF2A, MEF2C, MEF2D Pasqualucci T>C L54P 1 MADS GCB DLBCL 2011 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction A>G Y57C 1 MADS DLBCL Lohr 2012 MEF2A, MEF2C, MEF2D T>G L67R § 1 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D A>G Y69H § 2 MEF2 GCB DLBCL Morin 2011 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction T>C Y69C § 4 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D

E74-P75-H76 > -GGGGCT D 1 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D A>G H76R 1 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D A>G E77G 1 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D Morin 2011, Pasqualucci 2011, Lohr DLBCL, 2012, Zhang G>A E77K 4 MEF2 ABC DLBCL 2013 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction Pasqualucci C>G S78R 1 MEF2 ABC DLBCL 2011 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction A>T N81Y § 1 MEF2 GCB DLBCL Morin 2011 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction C>A N81K § 1 MEF2 GCB DLBCL Morin 2011 MEF2A, MEF2C, MEF2D Interrupts CABIN1 interaction GCB T>G D83A § 6 MEF2 DLBCL, FL Morin 2011 MEF2A, MEF2C, MEF2D Morin 2011, Pasqualucci GCB 2011, Lohr DLBCL, FL, 2012, Zhang T>A D83V § 23 MEF2 DLBCL 2013 MEF2A, MEF2C, MEF2D T>C D83G 1 MEF2 FL Morin 2011 MEF2A, MEF2C, MEF2D +A L100 frameshift 1 transactivation FL Morin 2011 MEF2B only Pasqualucci G>A G105E 1 transactivation GCB DLBCL 2011 MEF2B only C>A E108* 1 transactivation FL Morin 2011 MEF2B only 10bp deletion G121 frameshift 1 transactivation FL Morin 2011 MEF2B only not reported P132 frameshift 1 transactivation DLBCL Lohr 2012 MEF2B only Meissner G421C D141H 1 transactivation MCL 2013 MEF2B only -AAGG P169 frameshift 1 transactivation DLBCL Morin 2011 MEF2A, MEF2C, MEF2D

Number of Nucleotide Amino acid identified Affected codon change change occurrences Domain Diagnosis Reference conserved in Notes -GGAA F170 frameshift 1 transactivation FL Morin 2011 MEF2B only

Pasqualucci C>T R171* 1 transactivation ABC DLBCL 2011 MEF2B only not reported Y201* 1 transactivation DLBCL Lohr 2012 MEF2B only Pasqualucci G>A R207Q 1 transactivation GCB DLBCL 2011 MEF2B only +GG G242 frameshift 1 transactivation ABC DLBCL Morin 2011 MEF2B only Pasqualucci Predicted to disrupt an exon silencer G>A G246E 1 transactivation ABC DLBCL 2011 MEF2B only element in exon 7. May affect splicing Predicted to result in protein similar to isoform B MEF2B being produced from -C P256 frameshift 1 transactivation FL Morin 2011 isoform A MEF2B only isoform A transcript Predicted to result in protein similar to 30bp Pasqualucci isoform B MEF2B being produced from deletion P267 frameshift 1 transactivation GCB DLBCL 2011 isoform A MEF2B only isoform A transcript Predicted to result in protein similar to 30bp isoform B MEF2B being produced from deletion L269 frameshift 1 transactivation GCB DLBCL Ying 2013 isoform A MEF2B only isoform A transcript Affects the last amino acid residue encoded by alternatively used exon 8. Predicted to disrupt exon and intron identity elements. May affect exon 8 A>C S294R 1 transactivation FL Morin 2011 isoform A MEF2B only splicing. G>T R307S 1 transactivation ABC DLBCL Morin 2011 isoform A MEF2B only A>G *369G 1 transactivation FL Morin 2011 isoform A MEF2B only Stop-codon read-through A>C *369E 1 transactivation FL Morin 2011 isoform A MEF2B only Stop-codon read-through C>G *369Y 1 transactivation FL Morin 2011 isoform A MEF2B only Stop-codon read-through

Table 1.2 MEF2C mutations identified in DLBCL. This table lists all potentially somatic MEF2C mutations identified in DLBCL and indicates whether they co-occurred with MEF2B mutations. The MEF2C mutations have not been confirmed to be somatic, as matched germline DNA was not assessed for their presence. Unpublished mutations were identified by Ryan Morin.

Patient sample or Amino acid MEF2B mutation Nucleotide change Reference cell line change present? DB cell line A>G Y69H yes (D83V) unpublished T>G K5T OCI-Ly1 cell line no unpublished T>G E14A A>T E14D G>A E14K Patient sample yes (P132fs) Lohr 2012 A>G R10G T>A I6N Patient sample T>G I11S yes (I43R) Lohr 2012 Patient sample A>G Y69H no unpublished Patient sample T>C M244V no unpublished

Figure 1.1 The putative cell types of origin and transcription factor networks of B-cell lymphomas. The lymphomas discussed in section 1.4 are circled in red. GCB DLBCL (indicated as “GCB”) and FL are thought to arise from GC B-cells as their gene expression profiles are similar to those of normal GC B-cells. Transcription factors active in these cells control the indicated processes, including somatic hypermutation (SHM) and class switch recombination (CSR). In contrast, ABC DLBCL (indicated as “ABC”) is thought to arise from plasmablasts, post-germinal centre B-cells undergoing differentiation into plasma cells. Differentiation is driven through activation of IRF4 by NF-κB signaling. However, differentiation is blocked in ABC DLBCL, typically

34 through inactivation of BLIMP1. Mantle cell lymphoma (MCL) may arise from naïve mature B- cells, antigen-experienced mature B-cells, or memory cells. The cell types from which other B- cell lymphomas may originate are also indicated (BL: Burkitt lymphoma; CLL: chronic lymphocytic leukemia; SLL: small lymphocytic lymphoma; HCL: hairy cell leukemia; MALT: mucosa-associated lymphoid tissue lymphoma). FDC indicates follicular dendritic cells in germinal centres. This figure was produced by modifying Figure 1b from © Shaffer, A. L., Young, R. M. & Staudt, L. M. Pathogenesis of human B cell lymphomas. Annu. Rev. Immunol. 30, 565–610 (2012). Page 569. Adapted with permission from publisher.

Figure 1.2 Structure of isoform A and B MEF2B mRNA transcripts and proteins. Isoform A MEF2B mRNA includes all exons, whereas isoform B MEF2B mRNA lacks exon 8 (E8). Exclusion of exon 8 shifts the reading frame of exon 9. The L269fs mutation is a deletion of 31 bp in exon 8 that shifts exon 9 into the isoform B reading frame. Red indicates the amino acid sequence produced using the isoform A reading frame. Light pink indicates an amino acid sequence read from the isoform B reading frame of exon 9. Bright pink indicates amino acids encoded by exon 8 but read from the reading frame used for isoform B exon 9. Shown is Supplementary Figure 1a from © Ying, C. Y. et al. MEF2B mutations lead to deregulated expression of the oncogene BCL6 in diffuse large B cell lymphoma. Nat. Immunol. 14, 1084– 1092 (2013). Page 2 (Supplementary Information). Adapted with permission from publisher.

Figure 1.3 The frequency of alterations affecting MEF2 genes across cancer types. Alterations affecting MEF2B are frequent in several types of cancer. The percent of cases containing MEF2B alterations is shown for studies in which that percent was greater than 3%. The plot was obtained from cBioPortal261,262. The x-axis indicates the cancer type. The sources of the data are shown in brackets. ACyC: adenoid cystic carcinoma; ACC: adrenocortical carcinoma; Uterine CS: uterine carcinosarcoma; Lung SC: small cell lung cancer; MSKCC: Memorial Sloan-Kettering Cancer Centre; CLCGP: Clinical Lung Cancer Genome Project.

Figure 1.4 The localization of mutations in MEF2B. Shown are all potentially somatic MEF2B mutations identified in NHL1–7. § indicates that at least one mutation at that position was shown to be somatic (i.e. not present in matched constitutional DNA). Missense mutations occurred predominantly in the MADS and MEF2 domains, which are involved in dimerization. In DLBCL and FL, missense mutations most frequently affected K4, Y69 and D83. Frameshift and nonsense mutation occurred only in the transactivation domain and did not cluster at particular residues. Mutations in MCL are indicated separately from those in DLBCL and FL as the MCL mutations most frequently affected a different codon (that encoding K23) than the DLBCL and FL mutations. Only missense mutations have been identified in MCL.

Figure 1.5 The structure of the MEF2B MADS and MEF2 domains. Shown are the structures of a MEF2B dimer (yellow and red), the phosphodiester backbone of a DNA helix (pink) and the co-repressor CABIN1 (green). Only the MADS and MEF2 domains of MEF2B are shown. Residues mutated in DLBCL, FL or MCL have their sidechains shown. Residue numbers and colour coded side chains (red: oxygen; blue: nitrogen; white: carbon) are indicated for residues discussed in section 1.7.2. The residues most frequently mutated in DLBCL and FL are circled in orange. The residue most frequently mutated in MCL is circled in blue. This figure was produced using previously reported crystal structure data227.

Chapter 2: Materials and Methods2

2.1 Production of stably transfected HEK293A cell lines

HEK293A cells were obtained from Dr. Gregg Morin (BCCA Genome Sciences Centre) and were authenticated in August 2014 by Genetica DNA laboratories. Isoform A (GeneCopoeia GC-Z7031-CF) and isoform B (GC-F0247-CF) MEF2B were obtained in pDONR vectors (Invitrogen). Mutations were introduced in isoform A MEF2B using the QuikChange II site directed mutagenesis kit (Agilent). All constructs were sequenced to confirm the presence of only the desired mutation, using an ABI Prism 3100 Genetic Analyser (Applied Biosystems). MEF2B was transferred into pDEST40 (Invitrogen) using Gateway LR Clonase Enzyme Mix (Invitrogen). Insertion of MEF2B into the pDEST40 vector added a C-terminal V5 tag. The C- terminus was selected as the tag location to minimize disruption of the DNA binding domain at the N-terminus of MEF2B. The pDEST40 constructs and an empty pcDNA3 vector (Invitrogen) were transfected into HEK293A cells using TurboFect (Thermo Scientific). Monoclonal cell lines expressing WT, K4E, Y69H, D83V or R24L MEF2B-V5, and an oligoclonal cell line expressing R3T MEF2B-V5, were isolated by serial dilution of cells 48 hours after transfection. The cells were then treated with G418 (200 μg/mL, Invitrogen) for three weeks to select for stably transfected cells.

2.2 Production of stably transduced DoHH2 cells

Isoform A MEF2B cDNA was transferred into the pLenti6.2 vector (Invitrogen) using the Gateway LR Clonase Enzyme Mix (Invitrogen). pLenti6.2 constructs were packaged into replication-defective lentiviruses using HEK293T cells co-transfected with CMVDeltaR8.91 and pMD2 VSV-G envelope (gifts from Dr. Gregg Morin, BC Cancer Research Centre) using TransIT-LTI transfection reagent (Mirus Bio). Virus particles collected after 48 and 72 hours were passed through a 0.45 μm filter, concentrated, and applied overnight with Polybrene (Sigma) to DoHH2 cells in RPMI (Gibco) with 10% fetal bovine serum (FBS; Gibco). A polyclonal population of stably transduced cells was selected with 7.5 µg/mL Blasticidin S

2 Portions of Chapter 2 have been submitted for publication. 40

(Invitrogen) over three weeks. As no cells stably transfected with empty pLenti6.2 vector could be isolated, untransduced cells were used for comparison with MEF2B-V5 DoHH2 cells.

2.3 Cell culture and treatments

HEK293A and HeLa cells were grown in DMEM (Gibco) with 10% FBS (Gibco). Transfected HEK293A were grown in 100 μg/mL G418 (Invitrogen). Antibiotics were removed 24 to 48 hours before harvesting cells. Cells for microarray and RNA-seq analysis were treated for 6 hours with 1.5 µM ionomycin from Streptomyces conglobatus (Sigma) or the equivalent concentration (1.07 μL/mL) of dimethyl sulfoxide (solvent-only control) before RNA was collected. The FluoForte Calcium Assay Kit (Enzo Life Sciences) was used to assess effects of ionomycin on intracellular calcium levels. Cyclohexamide (Abcam) was used at 75 μg/mL. MEF2B knockdown was produced by transiently transfecting cells with shRNA #1 (TRCN0000015738, Dharmacon) or shRNA #2 (TRCN0000232095, Sigma), using TurboFect (Thermo Scientific). Non-targeting shRNA (MISSION TRC2 pLKO.5-puro, Sigma) was transfected as a negative control. RNA and protein were collected 48 hours after transfection. For colony formation assays, 1/14th of cells transfected in 24-well plates were re-plated in 6-well plates 24 hrs after transfection. Cells were allowed to grow for 5 days before staining with crystal violet. All DLBCL cell lines were obtained from DSMZ in 2008-2009. All DLBCL cell lines were of the GCB subtype1,266 and were grown in RPMI (Gibco) with 10% FBS (Gibco). MEF2B mutation status of the DLBCL cell lines was reported previously1,7. Transduced DoHH2 were grown in 7.5 µg/mL Blasticidin S (Invitrogen). IgM treatment used 10 µg/mL of AffiniPure

F(ab’)2 fragment goat anti-human IgM fc5µ fragment (Jackson ImmunoResearch) for 6 hours. All cell lines were confirmed to be mycoplasma free using the e-Myco mycoplasma PCR detection kit (iNtRON).

2.4 MEF2-dependent luciferase reporter assay

Luciferase reporter assays used the Cignal MEF2 Reporter Kit (SABiosciences). Plasmids from the kit were transiently transfected using Turbofect (Thermo Scientific) into cells in 24 well plates. 24 hours after transfection, cells from each well were transferred into three wells of a 96 well plate and were allowed to grow for another 24 hours. Luciferase activity was

41 detected using the Dual Glo Luciferase Assay System (Promega) and a Wallac Victor2 luminometer (Perkin Elmer). Firefly luciferase activity was normalised to activity of renilla luciferase constitutively expressed from a co-transfected plasmid. The negative control reporter contained a firefly luciferase gene with only a basal promoter element (TATA box).

2.5 Expression microarrays and qRT-PCR validation

All RNA extractions were performed using the RNeasy plus mini kit (Qiagen). RNA labelling and hybridization to GeneChip Human Exon 1.0 ST arrays (Affymetrix) was performed by the McGill University and Genome Quebec Innovation Centre. Probe signal values were normalized using RMA267 in the Affymetrix Expression Console, with gene level summarization of core probeset data. RMA was chosen over PLIER as it has been found to have superior performance268. The Linear Models for Microarray Data (LIMMA) Bioconductor package269 was used to identify differentially expressed genes from microarray data. LIMMA was used as it has shown greater power and lower false positive rates when working with sample sizes less than five, compared to other algorithms270. Although LIMMA is a parametric technique and approaches for validating the assumption of normality are not appropriate for the small sample sizes used here, LIMMA is robust to considerable deviation from normal distributions270 and has been widely used on gene expression datasets. The Benjamini-Hochberg correction was used to adjust p-values. Processed and raw data files are accessible through the Gene Expression Omnibus (GEO; dataset GSE67458; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=ihsbywkazrcxbab&acc=GSE67458). Annotation group enrichment and upstream regulator analyses were performed using Ingenuity Integrative Pathway Analysis of Complex omic’s Data (IPA) version 14855783 (Qiagen). Analyses considered only molecules and relationships where the species was human and the confidence was either experimentally observed or highly predicted. For annotation group enrichment analysis, Benjamini-Hochberg corrected p-values were reported. IPA Upstream Regulator Analysis p-values were calculated using Fisher’s Exact Test and indicated the probability that the overlap between the user-provided differentially expressed gene list and the known target gene set of a potential upstream regulator was due to chance. For both annotation group enrichment and upstream regulator analysis, z-scores indicated the confidence in the

42 predicted direction of activity change. As recommended by IPA, only absolute z-scores greater than two were considered significant. For all quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) primers, neighboring exons that were highly differentially expressed were selected as primer sites, to maximize the chance of detecting differential expression. Such exons were identified using LIMMA269 on data RMA267 normalized with exon level summarization. Where exons were similarly differentially expressed, those separated by the largest intron were selected as primer sites to minimize amplification of any contaminating genomic DNA. Primer sequences are listed in Appendix A. Primers were synthesized by Integrated DNA Technologies. qRT-PCR was performed using the RNA-to-CT SYBR Master Mix (Applied Biosystems) and a 7900 HT Sequence Detection System with SDS 2.2 software (Applied Biosystems). Gene expression was normalized to PGK1 expression because PGK1 was not differentially expressed in the microarray data and was previously identified as a reference gene for gene expression studies271– 273. A gene was considered to validate for a given comparison if it met two criteria: (i) the gene was differentially expressed in both microarray (adjusted p-value < 0.05) and qRT-PCR data (p- value < 0.05), and (ii) the gene had the same direction of expression change in both microarray and qRT-PCR data. Comparison of gene expression changes in qRT-PCR and microarray data used Spearman correlation coefficients rather than Pearson correlation coefficients as Spearman coefficients are more robust to outliers and do not assume the data follows a normal distribution. Heatmaps were produced using the gplots package in R version 3.1.2.

2.6 RNA-seq library construction

Plate-based libraries were prepared following the BC Cancer Agency's Michael Smith Genome Sciences Centre (BCGSC) strand specific paired-end protocol on a Biomek FX robot (Beckman-Coulter) with Ampure XP SPRI beads (Beckman-Coulter). First-strand cDNA was synthesized using the Superscript cDNA Synthesis kit (Life Technologies) with random hexamer primers and 1 μg/μL Actinomycin D. The second strand cDNA was synthesized following the Superscript cDNA Synthesis protocol, but substituting dTTP for dUTP. The cDNA was fragmented in an E210 sonicator (Covaris) for 55 seconds, using a duty cycle of 20% and intensity of 5. Purified cDNA was subjected to end-repair and phosphorylation by T4 DNA polymerase (NEB), Klenow DNA Polymerase (NEB), and T4 polynucleotide kinase (NEB) in a

43 single reaction, followed by 3’ A-tailing by Klenow fragment (3’ to 5’ exo minus, NEB). Products were ligated to Illumina PE adapters. The second strand was then digested using Uracil- N-Glycosylase (Life Technologies), thus achieving strand specificity. After purification, products were PCR-amplified using Phusion DNA Polymerase (Thermo Fisher Scientific) and Illumina’s PE primer set, with cycle conditions of 98°C for 30sec followed by 10-15 cycles of 98°C for 10 seconds, 65°C for 30 seconds and 72°C for 30 seconds, and then 72°C for 5 minutes. Purified PCR products were analysed using a LabChip GX (PerkinElmer). PCR products within the desired size range were purified using a 96-channel size selection robot developed at the BCGSC. The DNA quality was assessed using an Agilent DNA 1000 series II assay. DNA was quantified using a Quant-iT dsDNA HS Assay Kit (Invitrogen) and diluted to 8 nM. Section 2.9 describes sequencing of the libraries and alignment of the sequence data.

2.7 Differential expression analysis using HEK293A RNA-seq data

Quality control statistics for the HEK293A RNA-seq data are shown in Appendix B and passed the BCGSC standards. Differential expression analysis of RNA-seq data used DEseq Release 2.13 (Bioconductor)274. To enable analysis of data with only one replicate per group, “SharingMode” was set to “fit-only” and “method” was set to “blind”. Default settings were otherwise used. DEseq was used because it was recommended over other frequently used tools for cases where false positives are a concern275. Comparison of gene expression changes in RNA-seq and microarray data used Spearman correlation coefficients rather than Pearson correlation coefficients as Spearman coefficients are more robust to outliers and do not assume the data follows a normal distribution.

2.8 Chromatin immunoprecipitation (ChIP) for sequencing

The basic ChIP protocol was as follows. Cells were treated with 1% formaldehyde (Sigma) for 10 minutes, prior to treatment with 1/10th volume of 1.25 μM glycine (Sigma) for 5 minutes. Cells were lysed on ice for 30 minutes in lysis buffer (50 mM Tris-HCl pH 8.0, 1% SDS, 10 mM EDTA, Roche EDTA-free complete protease inhibitor cocktail), passed through a 22G needle, and centrifuged (5000 rpm, 10 minutes). The chromatin pellet was resuspended in 400 μL of buffer and split into two equal halves for sonication (20 minutes, 30 seconds on, 30 seconds off, power level 6) in a Sonicator 3000 (Misonix). Insoluble cell debris and

44 unfragmented chromatin were removed by centrifugation (13000 rpm, 12 minutes). An aliquot of chromatin was purified to allow determination of chromatin concentration. Agarose gel electrophoresis was performed on the purified quality control sample to confirm that DNA fragments were present in the 200-500 bp size range. Protein G Dynabeads (Life Technologies) blocked with bovine serum albumin and salmon sperm DNA were used to pre-clear chromatin. Chromatin was incubated with antibody for 1 hour at 4oC before addition of blocked Dynabeads. After overnight incubation at 4oC, samples were washed twice in low salt buffer (20 mM Tris- HCl pH 8.0, 0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl) and once in high salt buffer (same as low salt buffer except 500 mM NaCl). DNA was eluted in 100 mM sodium bicarbonate with 1% SDS at 68oC for 2 hours. Eluted DNA was purified by phenol chloroform extraction in Phase Lock Gel heavy 2 mL tubes (5 Prime) and ethanol precipitated with 40 μg glycogen (Roche). Input controls were produced by purifying 100 ng of chromatin from each sample in the same manner as the immunoprecipitated chromatin was purified. For MEF2B-V5 ChIP, eight 40-80% confluent 15 cm plates were grown of each cell type. Chromatin was resuspended in lysis buffer for sonication. To maximize the amount of DNA for library construction while still keeping samples within a replicate comparable to each other, 675 μg of chromatin were used for immunoprecipitation of all replicate 1 samples and 1460 μg of chromatin were used for immunoprecipitation of all replicate 2 samples. Volumes were equalized between samples by the addition of lysis buffer. One fourth volume of IP buffer (10% Tris-HCl pH 8.0, 1% Triton X-100, 0.1% deoxycholate, 0.1% SDS, 90 mM NaCl, 2 mM EDTA, Roche EDTA-free complete protease inhibitor cocktail) was added for the immunoprecipitation step. Immunoprecipitation used 29 μL V5 mouse antibody (Invitrogen R960-25) and 272 μL of beads. The V5 antibody was considered sufficiently specific as it produced a single band on western blots of HEK293A whole cell lysates at the predicted size of MEF2B that was detectable in MEF2B-V5 but not untransfected cells. For H3K27ac and H3K4me3 ChIP, one ~50% confluent 15 cm plate was grown of each cell type. Chromatin was resuspended in IP buffer for sonication. Replicate 1 ChIP used 24 μg of chromatin, and replicate 2 ChIP used 29 μg of chromatin. Volumes were equalized between samples by the addition of IP buffer. One tenth volume of IP buffer was added for the immunoprecipitation step. Immunoprecipitation used 2.5 μL of H3K27ac rabbit antibody

(Abcam ab4729) or 4.5 µL of H3K4me3 rabbit antibody (Cell signaling 9751), along with 27 μL of beads. The antibodies used were those routinely used by the BCGSC for ChIP. ChIP DNA was size separated using 8% PAGE. The 200-500 bp DNA fractions were excised from the gel and were eluted from the gel slice overnight at 4°C in elution buffer. Elution buffer consisted of a 5 to 1 ratio of LoTE buffer (3 mM Tris-HCl, pH 7.5, 0.2 mM EDTA) and 7.5 M ammonium acetate. DNA was purified using a Spin-X Filter Tube (Fisher Scientific, UK), and ethanol precipitated. Library construction was carried out on the Bravo liquid handling platform using VWorks Automation Control Software (Agilent Automation). Samples were first subjected to end repair using T4 Polynucleotide Kinase (NEB), T4 DNA Polymerase (NEB) and Klenow DNA Polymerase (NEB) at room temperature for half an hour. DNA was purified using PEG- Sera Mag Speedbeads (Fisher), with 13.9% final PEG concentration. A-tailing was then performed using Klenow exo minus (NEB) at 37oC for 30 minutes. Products were purified as before. Illumina short sequencing adaptors were ligated to A-tailed product using T4 DNA Ligase (NEB) at room temperature overnight. To remove adaptor dimers and library fragments below 200 bp, products were purified twice using PEG-Sera Mag Speedbeads with 8.9% and 10.9% final PEG concentrations. Adaptor ligated libraries were PCR amplified and barcoded using custom indexing primers, Illumina PCR primer 1.0 and 0.5 U of Phusion Hot Start II (Fisher). The initial denaturation step at 98oC for 30 seconds was followed by 13 cycles of 15 seconds at 98oC, 30 seconds at 65oC and 30 seconds at 72oC, and a final step at 72oC for 5 minutes. Amplified libraries were purified using PEG-Sera Mag Speedbeads with 9.2% final PEG concentration. Libraries were quantified using a Qubit HS DNA assay (Invitrogen) and equal-molar amounts were pooled. Each pool was quantified for sequencing using the Kapa SYBR Fast Complete Universal qPCR kit (Kapa Biosystems).

2.9 ChIP and RNA sequencing and alignment

ChIP and RNA libraries were sequenced on the Illumina HiSeq 2000/2500 platform using v3 chemistry and HiSeq Control Software version 2.0.10. The numbers of reads produced are indicated in Appendices B to D. MEF2B-V5 ChIP libraries were sequenced 8 per lane, H3K27ac ChIP libraries were sequenced 5 per lane, H3K4me3 ChIP libraries were sequenced 20 per lane,

ChIP input control DNA was sequenced 14 per lane, and RNA libraries were sequenced 2 per lane. All lanes were 75 bp paired-end sequencing. Sequencing was performed at the BCGSC. Sequencing data were aligned to GRCh37-lite genome-plus-junctions reference (http://www.bcgsc.ca/downloads/genomes/9606/hg19/1000genomes/bwa_ind/genome) using BWA (version 0.5.7)276 with default parameters. Reads failing the Illumina chastity filter were flagged with a custom script, and duplicated reads were flagged with Picard Tools (version 1.31, http://broadinstitute.github.io/picard/). Unmapped reads, optical duplicates and PCR duplicates were removed. Wig files were produced using BAM2WIG (http://www.epigenomes.ca/tools.html) and bigWig files were produced using the UCSC tool wigToBigWig (http://genome.ucsc.edu/goldenpath/help/bigWig.html). Processed and raw data files for all sequencing data are accessible through GEO (dataset GSE67458; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=ihsbywkazrcxbab&acc=GSE67458).

2.10 MEF2B-V5 ChIP-seq data analysis and validation using ChIP-qPCR

The number of mapped reads for each sample was above the ENCODE recommended minimum of 20 million277 (Appendix C). For assessment of data quality, peaks were identified using FindPeaks version FP4.0.15278. Quality control statistics are shown in Appendix C. MACS2279 version 2.0.10.20131010 was used to identify peaks present in ChIP samples that were not present in sequenced input DNA from the same replicate at a false discovery rate (FDR) of 0.05. MACS2 was used because it is a widely accepted peak caller used in recent publications280–282. Intersects between peak lists were obtained using ChIPseek283. Motifs were identified in sequences within 100 bp of the centre of peaks, using the ChIPseek implementation of HOMER version 4.6 (http://homer.salk.edu/homer/motif/). Analysis of motif enrichment in relation to distance from the peak centres was performed using the MEME-chip implementation of CentriMo version 4.9.1284. Motif co-occurrence statistics were calculated using ChIPModule285 with the default PWMs and Lambda file. For analysis with ChIPModule, the p-value cutoff for finding PWMs was set to 0.01 and a pattern mining support value of 100 was used. Genes associated with peaks were identified using GREAT147 with the “basal plus extension” gene association rule (proximal: 5 kb upstream and 5 kb downstream; distal: 1 Mb; 286 curated regulatory domains included). BETA was used with default settings to calculate

47 regulatory potential scores and rank product scores. The y-axes of Figure 3.13b and 4.16 correspond to the ranking of regulatory potential scores. BETA’s assumption that candidate transcription factor binding sites closer to TSSs are more likely to be involved in the regulation of gene expression than more distant candidate binding sites is supported by evidence that candidate binding sites with genomic locations closer to TSSs were more likely to contribute to reporter gene activation than candidate binding sites further from TSSs152. High confidence candidate direct target genes were defined as genes that had increased expression in WT MEF2B-V5 versus untransfected cells (adjusted p-values < 0.05 in microarray data), had peaks within 1 Mb up or downstream of their TSSs in both WT MEF2B-V5 ChIP-seq replicates, had BETA rank product scores < 0.05, were associated with at least one peak identified at an irreproducible discovery rate of < 0.01 and had log2 fold changes in expression in RNA-seq data for WT MEF2B-V5 versus empty vector cells that were > 0.3. The rationale for these thresholds is as follows. A rank product score threshold of 0.05 was used because rank product scores may be interpreted as p-values286,287 and a commonly used threshold for p-values is 0.05. The 0.01 IDR threshold was selected because it was recommended by the guidelines of 277 the ENCODE consortium . To integrate RNA-seq data, I used a threshold based on log2 fold changes rather than p-values because the RNA-seq data had minimal statistical power (i.e. only one sample per cell line). Thus, using a p-value threshold may have excluded many true DEGs.

A threshold log2 fold change of 0.3 was selected as 0.3 was approximately double the minimum log2 fold change that was considered statistically significant (adjusted p-value < 0.05) in the microarray data. ChIP validation primers were designed to regions with peaks in at least one replicate of ChIP-seq that were within 5 kb up or downstream of the TSS of a validation set gene. For validation set genes with more than one peak within 5 kb up or downstream of the TSS, the peak with the greatest fold enrichment was selected for validation. Two sets of positive control primers were designed to peaks present in both replicates of all ChIP-sequencing samples, which were within 5 kb of the ABCB4 and ZNF608 TSSs. Two sets of negative control primers were designed to regions without peaks in any ChIP-seq dataset: a region proximal to the TSS of BCL6 that was previously reported to be bound by MEF2B7 and a region proximal to the TSS of CPS1. Both BCL6 and CPS1 had increased expression in WT versus untransfected cells. Primer sequences are listed in Appendix A. Primers were synthesized by Integrated DNA Technologies.

Validation HEK293A ChIP-qPCR was performed as for sequencing, but using only 170 μg of chromatin with 5 μL of V5 antibody and 56 μL of beads. ChIP-qPCR on DLBCL cell lines was performed as described above, except using a Covaris sonicator (2 minutes, duty cycle 20%, intensity 8, 200 cycles/burst) and 7 µL of isoform A MEF2B antibody (ProSci) with 258 µg of chromatin and 112 µL of beads in 62% IP buffer. qPCR was performed using SYBR Green qPCR master mix (Life Technologies) and a 7900HT Sequence Detection System (ABI) on equal volumes of ChIPed DNA. Fold enrichment of promoter DNA was calculated compared to enrichment of an intergenic DNA region not expected to interact with MEF2B, then normalized to fold enrichment in ChIP-qPCR using normal immunoglobulin (Santa Cruz). Heatmaps were produced using the gplots package in R version 3.1.2.

2.11 Gel shift assays

Gel shift assays were performed using purified MEF2B. WT isoform A MEF2B was subcloned from a pDONR vector (Invitrogen) into the pDEST42 bacterial expression construct (Invitrogen) using Gateway LR Clonase Enzyme Mix (Invitrogen). The pDEST42 plasmid contained C-terminal V5 and 6x his tags. The MEF2B-pDEST42 construct was transformed into BL21-AI competent cells (Invitrogen). Cells were grown to an optical density of 0.4 before addition of 1mM IPTG and 0.1% arabinose to induce MEF2B expression. Induced cells were grown at 30oC for 2 hours before cells were pelleted. Pellets from 500 mL of bacterial culture were lysed by vortexing with 10 mL of low imidazole buffer (50 mM pH 8 NaH2PO4, 300 mM NaCl, 5 mM imidazole, 0.5 mM AEBSF). Lysates were then sonicated using three 10 second pulses in a XL-2000 sonicator (Misonix) and were passed through a 0.2 µM filter. MEF2B was purified from lysates using an AKTA Purifier (GE Healthcare) with Unicorn 5.20 (build 500) workstation software. A 1 mL HisTrap FF column (GE Healthcare) was used to bind the his-tagged MEF2B. Low imidazole buffer was used to remove unbound protein. Bound protein was eluted using a gradually increasing ratio of high to low imidazole buffer. High imidazole buffer was identical to the low imidazole buffer (above) except contained 200 mM imidazole. Fractions eluted at 15-30% high imidazole buffer were collected and concentrated using Amicon Ultra-4 centrifugal filter units with a molecular mass limit of 30 kDa (Millipore). Protein was then dialyzed into binding buffer (50 mM Tris-HCl, 5 mM MgCl, 50 mM NaCl, 5% glycerol, 200 µg/mL BSA and 0.5 mM DTT) at 4oC using 2 mL Slide-A-Lyser mini dialysis

49 devices with a 10K molecular mass cut off. Buffer was changed after 2 hours of dialysis and dialysis was allowed to continue overnight. Dialysed protein was concentrated using Amicon Ultra-0.5 mL centrifugal filters with a molecular mass limit of 10 kDa (Millipore). An equal volume of binding buffer containing 50% glycerol and 3.5 mM DTT was then added to the protein. Probe sequences are indicated in Appendix A. Oligonucleotides were obtained from Integrated DNA Technologies and annealed by mixing 2.5 µg of each oligonucleotide in 50 µL of PCR buffer (Sigma), heating to 65oC for 5 minutes and cooling gradually to room temperature. Radiolabelling reactions used 1 µL of annealed oligonucleotides with 0.5 U of Klenow DNA Pol I (NEB), 33 µM dATP, dGTP and dTTP (NEB) and 10 µCi of dCTP containing 32P (Perkin Elmer). The labelling reaction was allowed to proceed for 15 min at room temperature before purification using Illustra MicroSpin G-50 Columns (GE Healthcare). Probe activity was determined using a Bioscan QC-2000 (InterScience). For gel shift assays, purified protein was combined with 0.1 µg/µL of poly[d(I-C)] in 23 µL of binding buffer for 20 min at room temperature. Reactions used 10 µg of WT MEF2B-V5 and a volume of mutant MEF2B-V5 containing an equivalent amount of MEF2B-V5, as determined using western blots. 20,000 cpm of probe (~5 nM) was then added, and binding reactions were allowed to proceed for 20 min at room temperature. Binding reactions were then loaded into a 6% PAGE gel made using 0.5x TBE (pH 9.25) and a 29:1 acrylamide:bisacrylamide ratio. Electrophoresis was performed at 100 V for 90 minutes in 0.25x TBE (pH 9.25). The PAGE gel was then transferred to filter paper and dried at 80oC for 60 minutes in a 583 Gel Dryer (Bio-Rad). The dried gel was exposed to a phosphor screen overnight and the screen was scanned using a Typhoon FLA 7000 (GE Healthcare).

2.12 Quantification of protein abundance

Concentrations of TGFβ1 in cell culture media were assessed using the Quantikine TGFβ1 Immunoassay (R&D Systems) as directed. To obtain media samples, cells were plated in 24 well plates at 8x104 cells per well and allowed to adhere overnight. Media was then changed to serum-free DMEM and cells were cultured for 24 hours before media samples were collected. For western blots, whole cell lysates were obtained by incubating cells at 4oC for 1 hour with buffer containing 20 mM Tris-HCl pH 7.4, 150 mM NaCl and 0.5 mM NP-40, followed by

50 syringing through a 22 G needle. Complete protease inhibitor cocktail (Roche) was added to all lysis buffers immediately before use. Protein concentrations in lysates were determined using the BCA Reagent Kit (Thermo Scientific Pierce). Protein was separated using polyacrylamide gel electrophoresis (Invitrogen 4-12% Bis-Tris PAGE gels in MOPS buffer), and transferred to PVDF membranes (Millipore). 2% skim milk in phosphate buffered saline with 0.01% Tween 20 (Sigma) was used for blocking and antibody dilutions. Primary antibody incubations were at room temperature for 1 hour (actin) or at 4oC overnight (all other antibodies). Secondary antibodies (goat anti-mouse or anti-rabbit IgG-HRP, Santa Cruz) were applied at 1:5000 for 1 hour at room temperature. Chemiluminescence was detected using Clarity ECL or SuperSignal West Femto substrates (Pierce). Blots were imaged using the ChemiDoc MP Imaging System. Densitometry was performed using Image Lab Software version 4.1 (BioRad). Band intensities were calculated relative to one of the samples, then normalized to loading control band intensity. All loading controls were probed on the same membrane as was probed for the protein of interest. Western blots were performed on whole cell lysates unless noted otherwise. MEF2B polyclonal rabbit antibody was produced by ProSci to a peptide sequence present in isoform A but not isoform B MEF2B: RPGPALRRLPLADGWPR. Other antibodies used were V5 (Invitrogen R960-25), MYC (Invitrogen, R950-25), MEF2C (Cell Signaling 5030), BCL6 (Santa Cruz, sc-7388), NDRG1 (Sigma N8539) FN1 (Genetex GTX112794), VIM (Genetex GTX100619), CARD11 (Cell Signaling 4440), SNAI2 (Cell Signaling C19G7), Histone 3 (Abcam ab1791), MEF2A (Santa Cruz sc-10794), MEF2D (Abcam ab32845), lamin A/C (Santa Cruz sc-20680), β-tubulin (Santa Cruz sc-9104), TBP (Abcam ab51841), and actin (Abcam ab8227). To determine the half-life of MEF2B-V5 protein, whole cell lysates were western blotted with V5 and actin antibodies at multiple time points during treatment with the protein synthesis inhibitor cyclohexamide (Abcam, used at 75 μg/mL). As new protein could not be produced during cyclohexamide treatment, the abundance of pre-existing proteins could be observed to decline as they degraded. The abundance of MEF2B-V5 and actin was quantified using densitometry relative to the sample collected at time zero. MEF2B-V5 abundance was then normalized to actin abundance in the same sample, as actin was assumed to have negligible degradation during the timeframe used. The mean MEF2B-V5 abundance across three biological replicates was then plotted against the length of time that the cells were in cyclohexamide

51 treatment. An exponential trendline was fit to the data and half-life was calculated from the equation of the trendline.

2.13 HEK293A cell migration assays

A scratch was drawn with a 30 G needle through a monolayer of HEK293A cells that had been 100% confluent for 24 to 36 hours. Media was replaced immediately after scratching. Cells were imaged on an Axiovert 200 fluorescence microscrope using AxioVision release 4.4 software (Zeiss). The area remaining uncovered by cells was calculated 12, 24 and 32 hours after scratches were made. Area values were divided by the length of the scratch, to give distance values. The distances at 12 hours were subtracted from distances at 24 and 32 hours, to give the distance migrated over 12 and 20 hours, respectively.

2.14 Crystal violet proliferation assay

Within each biological replicate, four technical replicates of cells were plated at 1x104 cells per well in two 24-well plates. Cells in one plate were fixed and stained after 24 hours, whereas cells in the other were fixed and stained after 72 hours. Cells were fixed by treatment with 4% paraformaldehyde (Electron Microscopy Sciences) for 15 minutes at room temperature and stained for 20 minutes in 1 mg/mL crystal violet (EMD Millipore). Cells were washed three times with water to remove excess stain, before being lysed in 10% acetic acid (Fisher). Absorbance at 490 nm was measured in a Victor X3 plate reader with 2030 Workstation software (PerkinElmer). Readings from the plate stained at 72 hours were normalized to readings from the plate stained at 24 hours, before being normalized to untransfected cells.

2.15 Analysis of histone modification ChIP-seq data

For assessment of data quality, peaks were identified at a FDR of 0.01 using FindPeaks version FP4.0.15278. Quality control statistics are shown in Appendix D. To produce coverage profiles, coverage values from WIG files were averaged within 10 bp bins. Values for each bin across all the regions of interest (i.e. TSSs or MEF2B peaks) were then averaged. The numbers and fold enrichments of significant peaks were produced using MACS2 version 2.0.10.20131010279 compared to input DNA (FDR < 0.05). Tukey boxplots of fold enrichment

52 values were produced using R version 3.1.2. Peaks were associated with genes using the “basal plus extension” gene association rule of GREAT147 (proximal: 5 kb upstream and 5 kb downstream; distal: 0 kb; curated regulatory domains included). Associations were restricted to within 5 kb up and downstream of TSSs in attempt to restrict analyses to the highest confidence associations.

2.16 Fractionation of nuclear and cytoplasmic protein lysates

Nuclear and cytoplasmic fractions were separated by first treating cells with nuclear isolation buffer (150 mM NaCl, 1.5 mM MgCl2, 0.5% NP-40, 10 mM Tris-HCl pH 7.5) on ice for 5 minutes. Nuclei were pelleted by centrifugation (800 xg, 5 minutes, 4oC) and washed four times with nuclei isolation buffer before being lysed in 250 mM NaCl, 20 mM NaPO4, 30 mM NaPyrophosphate, 10 mM NaF, 0.5% NP-40, 10% glycerol and 1 mM DTT. The resulting nuclear lysate was syringed through a 22 G needle and centrifuged at 13000 rpm, 4oC, for 30 minutes to remove insoluble debris.

2.17 DLBCL patient sample analysis

Paired-end RNA-seq datasets and mutation information for DLBCL patient samples and centroblasts were published previously1. Only samples classified as GCB in Morin et al. (2011) were included for analysis, as most MEF2B mutations in DLBCL occurred in the GCB subtype1. RPKM values for each gene were calculated using an in-house pipeline used previously1. Briefly, the total sequencing coverage across each exon was calculated from WIG files. This number was then divided by the read length to obtain the number of reads mapping to a gene. RPKM was then calculated by dividing the number of reads mapping to a gene by the length of the gene’s collapsed exons in kilobases and then dividing by the total number of millions of mapped reads. Differential expression analysis of RNA-seq data used DEseq Release 2.13 (Bioconductor) on genes with at least one read in all mutant MEF2B samples or all WT MEF2B samples. Analysis using the Database for Annotation, Visualization and Integrated Discovery (DAVID)288,289 applied Benjamini-Hochberg p-value correction. In addition to differential expression analysis, the DLBCL RNA-seq data were used for determining the ratio of isoform A to isoform B MEF2B expression. MEF2B isoform B is

53 identical to isoform A except for the skipping of exon 8. The number of isoform B transcripts was considered proportional to the sum of the number of sense and antisense reads spanning the junction of exon 7 with exon 9. The number of isoform A transcripts was considered proportional to the sum of the number of sense and antisense reads spanning the junction of exon 7 and 8, or the junction of exon 8 and 9, and dividing by two.

2.18 Mass spectrometry

Nuclear protein fractions were obtained using the ProteoJET Cytoplasmic and Nuclear Fractionation Kit (Fermentas) and rotated end-over-end overnight with 8 µL of MEF2B antibody (Abcam 33540). 40 µL of Protein G agarose was then added for 2 hours of rotation. Beads were washed six times in lysis buffer before protein was eluted at 95oC for 10 minutes in 2x SDS PAGE loading dye. Samples were run in 10% Bis-Tris PAGE gels (Invitrogen) in MOPS buffer and stained with Coomassie Brilliant Blue R-250 (Thermo Scientific). 30-45 kDa fragments were excised and multiple reaction monitoring (MRM) mass spectrometry was used to identify D83 and D83V MEF2B peptides. MRM mass spectrometry was performed using the same methods and instruments as described previously290. Three to four MRM transitions for each peptide were used in the MRM analysis. Peptides detected were TNTDILETLK (D83), TNTVILETLK (D83V) and TPPPLYLPTEGR (control peptide).

2.19 Gene set enrichment analysis

Gene Set Enrichment Analysis (GSEA) was performed as described, using default parameters291,292. For GSEA investigating whether MEF2B mutations were associated with changes in BCL6 target gene expression, three different BCL6 target gene lists were used. One BCL6 target gene list consisted of all 299 genes downregulated by BCL6 activity in normal GC B-cells that were near candidate BCL6 DNA binding sites293. A second BCL6 target gene list consisted of the 46 genes downregulated by BCL6 activity in normal GC B-cells that were near candidate BCL6 DNA binding sites293. The third BCL6 target gene list consisted of the 148 genes near candidate BCL6 DNA binding sites in DLBCL cells294. DLBCL patient samples with known genetic alterations affecting BCL6 were excluded from GSEA.

2.20 Chemotaxis assays

600 μL of RMPI media (Gibco) containing chemoattractant was added into 24-well plate wells beneath polycarbonate Transwell inserts with 5 μm pores (Corning, 3421). 1x106 cells in 100 μL of RMPI were then added to the insert. 2.5x105 cells from the same cell suspension were added in triplicate to a 96-well plate as input controls. Cells were allowed to migrate for 3.5 hours before inserts were removed and 60 μL of alamarBlue (Life Technologies) was added to the well. alamarBlue was also added to input control cells (10 µL into 90 µL of media). alamarBlue treated cells were incubated for 2 hours at 37oC before fluorescence was measured as described above. Readings were normalized to media only controls before normalization to input controls. Readings were then normalized to untransfected cells. For chemotaxis to CXCL12, the Transwell membrane separated 0 from 300 ng/mL CXCL12 (PeproTech), with 10% FBS present on both sides of the membrane. For chemotaxis to FBS, the Transwell membrane separated media containing no FBS from media containing 10% FBS.

Chapter 3: Characterization of the MEF2B Regulatory Network in HEK293A Cells3

3.1 Introduction

The overall objective of work presented in Chapter 3 was to characterize the regulatory network of WT MEF2B in HEK293A cells. Towards that end, the research presented in Chapter 3 pursued six specific aims. These aims and the research findings that address them are summarized in this introduction. First, I aimed to identify candidate MEF2B target genes by identifying transcriptome- wide gene expression changes in cells expressing V5-tagged WT MEF2B compared to control cells. I identified 3,944 differentially expressed genes (DEGs). Second, I aimed to identify differences in protein abundance and cellular phenotypes between MEF2B-V5 and control cells. I found that MEF2B-V5 expression affected the abundance of proteins involved in cell proliferation, cell migration and epithelial-mesenchymal transition (EMT) and increased cell migration. Third, I aimed to identify genome-wide candidate MEF2B binding sites. I confirmed that MEF2B binds the same sequence motifs as other MEF2 proteins and found that JUN DNA binding motifs co-occurred with sites of MEF2B DNA interaction. Fourth, I aimed to identify candidate direct MEF2B target genes and infer whether MEF2B tends to activate or repress their expression. I present evidence that MEF2B tends to act as a transcriptional activator of a set of 1,141 candidate direct target genes. Fifth, I aimed to determine whether MEF2B activity correlates with changes in histone modification. As other MEF2 proteins associate with histone modifying enzymes (see section 1.6.4), I hypothesized that MEF2B’s activation of direct target gene expression was mediated by increased levels of the activating histone modifications H3K27ac and H3K4me3. Contrary to my

3 Portions of Chapter 3 have been accepted for publication pending formatting: J.R. Pon, J. Wong, S. Saberi, O. Alder, M. Moksa, S.W.G. Cheng, G.B. Morin, P.A. Hoodless, M. Hirst, M.A. Marra. MEF2B Mutations in Non- Hodgkin Lymphoma Dysregulate Cell Migration by Decreasing Transcriptional Activation of MEF2B Target Genes. Nature Communications. Author contributions are provided in the Preface. 56 hypothesis, H3K27ac and H3K4me3 did not tend to increase near MEF2B binding sites or the TSSs of candidate direct MEF2B target genes. As the association of other MEF2 proteins with coregulators and consequently the expression of some MEF2 target genes is regulated by calcium193,239, my sixth aim was to investigate whether intracellular calcium levels affect MEF2B-dependent gene expression. The expression of three genes potentially relevant to DLBCL development, ANO1, CCL8 and NFATC2, was validated to be calcium sensitive in a MEF2B-dependent manner. Overall, the research described in Chapter 3 characterized the MEF2B regulatory network in HEK293A cells through an integrative analysis of gene expression changes, DNA binding patterns and histone modifications. This analysis identified numerous novel candidate direct and indirect target genes, including known proto-oncogenes and tumor suppressors.

3.2 Results

3.2.1 WT MEF2B regulates genes involved in proliferation, migration and EMT

To identify MEF2B target genes, I analyzed expression microarray data from three biological replicates of a monoclonal HEK293A line stably transfected with WT isoform A C- terminally V5-tagged MEF2B (Figure 3.1a), as well as from three biological replicates of untransfected cells (methods sections 2.1 and 2.5). The 3,944 differentially expressed genes (DEGs) that emerged from comparison of the WT MEF2B-V5 expressing cells to untransfected cells were considered to be sensitive to MEF2B expression levels and thus potential target genes295 (adjusted p-value < 0.05). I then verified that trends in the directions of gene expression change resulting from WT MEF2B-V5 expression were reproducibly detectable and were not predominantly technology- specific artifacts or artifacts of transfection or selection. To do so, I obtained RNA-seq data from one sample of the WT MEF2B-V5 cell line that was used for microarrays and one sample of a cell line stably transfected with an empty vector (methods sections 2.6 and 2.7). I expected that the RNA-seq and microarray data would have a degree of correlation similar to that in previous studies comparing RNA-seq and microarray data from the same samples (e.g. 0.64296 and 0.75297) if the two datasets reflected the same underlying biology. Supporting the notion that the RNA-seq and microarray datasets reflected the same underlying biology, they had a Spearman

57 correlation coefficient of 0.64 (Figure 3.1b). The RNA-seq data analysis thus succeeded in verifying the aspects of the microarray data analysis noted above. The RNA-seq data also provided additional evidence that certain genes were differentially expressed in response to MEF2B-V5 expression. Genes that were DEGs in both RNA-seq and microarray data (adjusted p-values < 0.05 for both types of data, Appendix E, 233 genes) were considered to be the highest confidence MEF2B target genes identified from global gene expression analysis and are noted throughout the following analyses. Fewer DEGs were identified using the RNA-seq data than using the microarray data, likely because fewer replicates were used for RNA-seq than for microarrays (i.e. one replicate for RNA-seq and three replicates for microarray), reducing statistical power and increasing the likelihood of true DEGs being missed in the RNA-seq analysis. I next demonstrated that the differential expression of genes in the WT MEF2B-V5 versus untransfected cells was not predominantly a result of cell line specific artifacts. To do so, I compared gene expression in empty vector cells with that in two monoclonal HEK239A cell lines stably expressing WT MEF2B-V5 (referred to as WT MEF2B-V5 H2 and WT MEF2B-V5 D3) that were different from the WT MEF2B-V5 cell line used for microarrays (Figure 3.3a). Using qRT-PCR, I assessed the expression of 30 microarray DEGs with functions of particular interest. Specifically, the 30 genes were selected to represent the two functional annotation categories that IPA indicated were most enriched in the microarray DEGs (adjusted p-values < 2x10-10, Figure 3.2a): “cellular movement” and “cellular growth and proliferation”. These categories were also enriched in the RNA-seq DEGs (adjusted p-values < 0.002, Figure 3.2b), supporting the prediction that these processes were affected by MEF2B-V5 expression. As studies comparing qRT-PCR and microarray data from the same RNA samples found correlation coefficients of 0.58296 to 0.67298, I expected that my qRT-PCR and microarray data sets would show a similar degree of correlation if they both reflected the same underlying biology. Consistent with the notion that they do reflect the same biology, correlation coefficients of 0.61 (using WT MEF2B-V5 D3 cells) and 0.75 (using WT MEF2B-V5 H2 cells) were obtained (Figure 3.3b and methods section 2.5). The qRT-PCR data analysis thus succeeded in demonstrating that differential expression in the microarray data analysis was not predominantly an artifact of cell line specific effects. This finding supports the validity of my decision to consider DEGs in the microarray data candidate MEF2B target genes.

According to the qRT-PCR data, 27 genes had significant changes in expression consistent with the microarray data in at least one of the additional WT MEF2B-V5 cell lines (adjusted p-values < 0.05, Table 3.1). Thus, the qRT-PCR data provided additional evidence that those 27 validated genes are MEF2B target genes. Amongst the validated genes significantly differentially expressed in both additional WT MEF2B-V5 lines was a previously identified MEF2B target gene, BCL67, which was assessed as a positive control. The other validated genes are discussed below. Three of the validated genes were among the top 50 genes when all genes were ranked by fold change in expression using microarray data. These three were the proto-oncogenes CARD11299 and MYC300 and the metastasis suppressor NDRG1301, all of which were also DEGs in RNA-seq data. Because of the large fold changes in these genes’ mRNA expression (Figure 3.3c,d), I suspected that changes in the abundance of proteins they encode would also be detectable. Increased abundance of CARD11 protein in WT MEF2B-V5 versus control cells was validated, as was the decreased abundance of MYC and NDRG1 protein (Figure 3.4a,b). The decreased abundance of MEF2C mRNA and MEF2C protein in WT MEF2B-V5 versus control cells was also validated (Figures 3.3c,d and 3.4c,d). MEF2C was the only MEF2 family gene other than MEF2B that was differentially expressed in the microarray data (adjusted p-value 0.03). The decreased expression of MEF2C was of interest as the differential expression of genes with MEF2 binding sites may have been through either increased interaction with MEF2B or decreased interaction with MEF2C. The mechanism by which MEF2C expression was decreased may be similar to that of a previously reported negative feedback loop302. In that feedback loop, increased levels of MEF2C or MEF2D increased expression of HDAC9302, which encodes a co-repressor that associates with MEF2 proteins219–222. Increased association with HDAC9 was thought to decrease MEF2C’s capacity to activate its own expression from a MEF2 site in the MEF2C promoter175. As the expression of HDAC5, -7 and -9 was greater in the MEF2B-V5 cells than in untransfected cells (adjusted p-values < 0.052), greater interaction with HDACs may have reduced MEF2C’s capacity to maintain its own expression. The remaining 25 validation set genes had functions related to cell migration. The differential expression of 22 of these migration related genes was validated in at least one of the two additional WT MEF2B-V5 cell lines (Figure 3.3c,d and Table 3.1). Consistent with the differential expression of migration regulators and with IPA predictions of increased cellular

59 movement in WT MEF2B-V5 versus control cells (Appendices F and G), WT MEF2B-V5 cells filled in scratched areas of a confluent monolayer faster than control cells (Figure 3.5a; methods section 2.13). Faster scratch closure was likely due to increased cell migration, not increased proliferation, as no increase in proliferation was detected between MEF2B-V5 and control cells (Figure 3.5b; method section 2.14). As increased cell migration is a hallmark of EMT, I hypothesized that MEF2B-V5 expression in HEK293A cells might promote a mesenchymal-like gene expression signature. Consistent with this hypothesis, a set of 67 genes whose expression increased during EMT303 tended to have greater expression in MEF2B-V5 cells than in untransfected cells (FDR q-value 0.086, Figure 3.6). Among these genes were the well-known EMT inducers TGFB1, FOXC2, ZEB1 and SNAI2 (also called SLUG), as well as the mesenchymal markers VIM (encoding vimentin) and FN1 (encoding fibronectin)304. As expected from the microarray data, SNAI2, vimentin, fibronectin and TGFβ1 protein tended to have greater abundance in the two additional MEF2B-V5 lines than in empty vector and untransfected cells (Figure 3.7). Interestingly, IPA identified TGFB1 as the transcriptional regulator whose known set of target genes overlapped most significantly with the microarray DEGs (Fisher’s Exact Test p-value 1x10-13, activation z- score 7.031), consistent with the notion that increased TGFβ signaling mediates a large proportion of the gene expression changes downstream of MEF2B. Further supporting this notion, the same prediction was made using DEGs identified from RNA-seq data (Fisher’s Exact Test p-value 7x10-5, activation z-score 3.2).

3.2.2 Chromatin immunoprecipitation identifies genome-wide MEF2B binding sites

The identified DEGs are likely to include both direct and indirect MEF2B target genes. To identify candidate direct MEF2B target genes, I first identified genome-wide candidate MEF2B binding sites using chromatin immunoprecipitation sequencing (ChIP-seq). Two biological replicates of ChIP-seq were performed using V5 antibody on the WT MEF2B-V5 cell line that was used for microarrays (methods sections 2.8 and 2.9). Regions of chromatin enriched during IP, referred to as peaks, were identified at a FDR of 0.05 (methods section 2.10). I assessed the quality of the ChIP-seq data by investigating whether both replicates appeared to reflect the same underlying biology. The notion that they reflect the same biology was supported by two observations. First, I noted that 99% of the peak regions identified in

60 replicate 1 also had peaks in replicate 2. Second, I found that the peaks that were most likely to reflect true signals (i.e. peaks with low p-values) were more likely to be identified in both replicates than peaks that were more likely to be noise (i.e. peaks with higher p-values; Figure 3.8a). To estimate how reproducible data may be in future experiments, I performed Irreproducible Discovery Rate (IDR) analysis305 on data from both replicates. As expected for high quality data277, a clear inflection point was present around the 1% IDR value (Figure 3.8b). The majority of the peaks in common between both replicates (i.e. 3,081 out of 5,499 peaks) had more than a 90% chance of being reproducible in future experiments. To provide a focus on high-confidence findings, I report only conclusions that were supported by data from both replicates or validated by ChIP-qPCR. I next used two independent algorithms, HOMER283 and MEME-ChIP306, to identify de novo motifs significantly enriched in the ChIP-seq data (Figure 3.9). The most enriched de novo motif matched the known motifs of MEF2A and MEF2C and was present in 44% of peaks (Figure 3.9 and Appendix H). The MEF2A and MEF2C motifs were most enriched near the centres of peaks (p-values < 1x10-10, Figure 3.10), consistent with the idea that these motifs are directly bound by MEF2B. The direct binding of MEF2B to sequences similar to MEF2 motifs was validated using gel shift assays on his-tagged WT MEF2B-V5 expressed in E. coli (Figure 3.11). The second most enriched de novo motif, present in 24% of peaks, was similar to the motifs of the AP-1 complex and a protein in the AP-1 complex, FOS (Figure 3.9 and Appendix H). Interestingly, 12.8% of peaks with MEF2 motifs also contained AP-1 motifs, indicating a tendency for MEF2 motifs and AP-1 motifs to co-occur (corrected p-value of co-occurrence: 1.33x10-8, calculated using ChIPModule285). This co-occurrence was consistent with the notion that MEF2B and the AP-1 complex may cooperatively regulate target genes. Interestingly, ENCODE data provided evidence that the AP-1 complex motif was also enriched in MEF2A and MEF2C ChIP-seq peak regions307, indicating that MEF2A and MEF2C may also interact with the AP-1 complex. Other de novo and known motifs enriched in the peak regions are listed in Appendices H and I, respectively.

3.2.3 Integrative analysis of DNA binding and gene expression data identifies candidate direct MEF2B target genes

I next identified genes that may be regulated by MEF2B bound at the peak regions. GREAT147 was used to associate each gene with all proximal peaks within 5 kb up- or downstream of its TSS, plus with all distal peaks within 1 Mb up- or downstream of its TSS that that were not within 5 kb of any other TSS. The 1 Mb threshold was chosen to maximize the inclusion of potential associations, as this has been shown to improve confidence in downstream analyses147. Using a shorter distance threshold could have excluded associations with many of the peak regions. For instance, 85% of peaks regions were further than 10 kb up- or downstream of the nearest TSS and 37% were further than 100 kb up- or downstream of the nearest TSS (Figure 3.12). In total, 4,957 genes were associated with peaks. Of the genes associated with peaks, 1,276 were DEGs in the microarray data (dark blue and red sections in Figure 3.13a). The overlap between peak-associated genes and DEGs was greater than expected by chance (p-value < 0.0001, chi-square with Yates correction), supporting the notion that expression of MEF2B-V5 can alter the expression of genes associated with MEF2B binding sites. 3,681 of the genes associated with peaks were not DEGs (grey section in Figure 3.13a), perhaps because (i) MEF2B-V5 recruited to the associated peak regions could not impact the expression of any genes, (ii) MEF2B-V5 recruited to the associated peak regions only impacted the expression of genes other than those with which they were associated (iii) the degree of change in MEF2B abundance between WT MEF2B-V5 and untransfected cells was not sufficient to detectably affect the expression of those genes or (iv) other factors prevented MEF2B from affecting the expression of those genes in the cell type and growth conditions used. The 2,668 genes that were DEGs but were not associated with peaks (light blue and pink sections in Figure 3.13a) were considered indirect MEF2B target genes, as their differential expression may have been mediated by the altered expression of other MEF2B target genes. Interestingly, 89% of the DEGs associated with peaks had increased expression (red section in Figure 3.13a) rather than decreased expression (dark blue section in Figure 3.13a) in WT MEF2B-V5 cells compared to untransfected cells. This observation was consistent with the hypothesis that MEF2B acts predominantly as a transcriptional activator. I further supported that

62 hypothesis using BETA, an approach that assumes that genes closer to ChIP-seq peak regions are more likely to truly be regulated by transcription factors bound to the peak regions286. Based on this assumption, the differential expression of genes closest to ChIP-seq peaks was considered to be the best indicator of the effect of MEF2B on direct target gene expression286. Supporting the contention that MEF2B tends to act as an activator, MEF2B peak regions tended to be closer to genes with increased expression in MEF2B-V5 versus untransfected cells than to genes with decreased or unaffected expression (Figure 3.13b). Although transcription factors have been identified that can activate or repress direct target gene expression depending on the DNA binding site used308–310, no such dual activity has been reported for MEF2 family proteins. I thus considered it most likely that the 135 peak- associated genes with decreased expression in MEF2B-V5 versus untransfected cells (the dark blue section in Figure 3.13a) were not directly repressed by MEF2B-V5 but rather were repressed indirectly, through other MEF2B target genes. In contrast, I considered the 1,141 peak- associated genes that had increased expression in MEF2B-V5 versus untransfected cells (the red section in Figure 3.13a) to be candidate direct MEF2B target genes. I next aimed to identify high confidence candidate direct target genes among the 1,141 genes (see methods section 2.10). First, I used rank product scores to integrate ChIP-seq and expression data and produce an estimate for each gene of the relative likelihood it is not a true direct target286. Higher rank product scores were assigned to genes with greater fold changes in expression and with TSSs closer to ChIP-seq peaks, as such genes were considered more likely to be direct target genes286. Rank product scores for 821 of the candidate direct target genes were < 0.05. I then identified 414 of the 821 genes that were associated with the peaks that were most likely to be reproducible (i.e. an IDR < 0.01) and thus were most likely to be associated with true MEF2B binding sites. Finally, I identified 261 of those 414 genes that I considered most likely to truly be differentially expressed in response to WT MEF2B-V5 expression, as they had increased expression in the RNA-seq data for WT MEF2B-V5 versus empty vector cells (log2 fold changes in expression > 0.3). I considered those 261 genes (listed in Appendix J) to be high confidence candidate direct target genes. I then sought to validate the ChIP-seq data using ChIP-qPCR on additional WT MEF2B- V5 cell lines. The purpose of this validation was to demonstrate that the ChIP-seq peaks were not predominantly cell line-specific or technology-specific artifacts and thus demonstrate that the

63 peak regions may include true MEF2B binding sites. I selected peaks for validation that were near the 27 genes whose differential expression in WT MEF2B-V5 versus empty vector cells had already been validated (methods section 2.10). Six of the 27 genes had one or more peaks within 5 kb up- or downstream of their TSSs in at least one replicate of ChIP-seq data. The peak region with the greatest fold enrichment for each of those six genes was assessed in validation ChIP- qPCR. Negative control ChIP-qPCR was performed on two regions without peaks in either ChIP-seq replicate that were also within 5 kb of the TSSs of expressed genes. Five out of six validation set regions (83%) but neither of the negative control regions showed at least two fold enrichment in ChIP-qPCR on at least two out of three WT MEF2B-V5 lines compared to IgG control ChIP-qPCR (Figure 3.14). This statistic was similar to published rates of transcription factor ChIP-seq validation (i.e. 75% to 86%207, using the same ChIP-qPCR fold enrichment threshold as was used in my research). The ChIP-qPCR data therefore supported the notion that the ChIP-seq peaks mark candidate MEF2B binding sites. Notably, the only peak that did not validate (that near PAK1) was a peak that was present in only one of the two ChIP- seq replicates. This was consistent with the notion that the peaks identified in both replicates were particularly likely to mark true MEF2B binding sites. The ChIP-qPCR data also provided evidence that the five genes near the validated peaks, RHOB, CDH13, ITGA5, CAV1 and RHOD, may be direct rather than indirect target genes. Notably, all five of these candidate direct target genes had increased expression in WT MEF2B- V5 versus control cells in microarray and qRT-PCR data, consistent with the notion that MEF2B activates expression of direct target genes. The remaining 21 genes that were validated as DEGs by qRT-PCR but were not associated with peaks in any ChIP-seq replicate (e.g. MYC, CARD11, NDRG1, FN1 and TGFB1) may be indirect target genes.

3.2.4 Candidate direct MEF2B target genes include regulators of cell movement and cell survival

To predict which of the direct target genes may contribute to the differences in cellular movement observed between WT MEF2B-V5 and control cells, I used IPA to analyse the functional annotations of the 1,141 candidate direct target genes. Cellular movement and cell growth and proliferation were the two most enriched categories in the analysis of candidate direct target genes (adjusted p-values ≤ 8x10-23, Figure 3.15 and Appendix K), as they were in

64 the analysis of all DEGs (i.e. both direct and indirect target genes). 210 candidate direct target genes had annotations related to cellular movement, 24 candidate direct target genes had annotations related to EMT and 341 candidate direct target genes had annotations related to cell growth and proliferation (see Appendix J for the highest confidence genes with these annotations). Four of the candidate direct target genes with functions related to cellular movement were genes whose differential expression and associated ChIP-seq peaks were validated (i.e. RHOB, CDH13, CAV1 and RHOD). ITGA5, despite its differential expression and association with a validated ChIP-seq peak, was not among the 1,141 candidate direct target genes as it had an associated peak in only one ChIP-seq replicate. Interestingly, the third most enriched category in the candidate direct targets was cell death and survival, containing predominantly pro-survival and anti-apoptotic genes (adjusted p- value 3x10-20, 324 genes). Consistent with the notion that MEF2B activity helps maintain cell viability, a prior report demonstrated that MEF2B knockdown in DLBCL cells decreased cell cycle progression7. I thus inspected the genes associated with MEF2B-V5 ChIP-seq peaks for well-known regulators of cell survival and cell cycle progression. I noted that the anti-apoptotic gene BCL2 contained a MEF2B ChIP-seq peak in its first intron (adjusted p-value 10-17; Figure 3.16a) and had increased expression in WT MEF2B-V5 versus untransfected cells in microarray data (adjusted p-value 0.005). In addition, the cell cycle regulator JUN contained a MEF2B ChIP-seq peak region that overlapped its TSS (adjusted p-values 10-67; Figure 3.16a). JUN expression is known to be regulated by MEF2 family proteins other than MEF2B311,312. However, JUN was not differentially expressed in WT MEF2B-V5 versus untransfected cells, perhaps because endogenous factors maintained expression of JUN at a level insensitive to further activation. To validate BCL2 and JUN as candidate direct MEF2B target genes, I first used ChIP- qPCR to validate ChIP-seq peak regions near these genes (Figure 3.16b). I also used gel shift assays to demonstrate that MEF2B directly binds sequences within the validated peak regions (Figure 3.11). Moreover, BCL2 and JUN mRNA expression tended to decrease with MEF2B knockdown (Figure 3.17a,b). Although the fold changes in BCL2 and JUN expression were small, these results were reproducible in HeLa cells (Figure 3.17a), consistent with the notion that reducing MEF2B expression can reduce BCL2 and JUN expression. Moreover, cells with transient MEF2B knockdown had reduced capacity for colony formation (Figure 3.17c) and I

65 was unable to isolate cells with stable MEF2B knockdown, consistent with the contention that the gene expression changes in MEF2B knockdown cells can compromise cell viability. Although no differences in proliferation or apoptosis (assessed using a TUNEL assay for DNA fragmentation, data not shown) were detected between WT MEF2B-V5 and control cells, endogenous factors may have maintained cell proliferation and survival at levels insensitive to further increases.

3.2.5 MEF2B transcriptional activity is not associated with increased levels of H3K27ac and H3K4me3

I next investigated the hypothesis that MEF2B’s activation of direct target gene expression was mediated by increased levels of activating histone modifications. This hypothesis was based on evidence that MEF2 proteins interact with histone modifying enzymes (see section 1.6.4). I selected H3K27ac and H3K4me3 for assessment because they are associated with active enhancers (H3K27ac) and/or promoters (H3K27ac and H3K4me3)35, and because MEF2 proteins interact with chromatin modifiers that were thought to deposit these marks. For instance, p300 can acetylate H3K27ac313 and interacts with MEF2A, -C and -D192,229–231. H3K27ac was also selected for assessment because H3K27ac is mutually exclusive with H3K27me3314, a marker of bivalent domains36. Bivalent domains tend to control key regulators of differentiation36 and were disrupted by the EZH2 mutations found in DLBCL161. The disruption of bivalent domains may thus be relevant to DLBCL development. H3K4me3 was investigated because MEF2D can interact with a histone methyltransferase, KMT2D, that is recurrently mutated in 1,315 DLBCL . Although KMT2D has recently been found to promote H3K4me1 and H3K4me2 rather than H3K4me3316, change in H3K4me3 patterns may still be relevant to lymphoma development because H3K4me3 is one of the modifications that defines bivalent domains36. To assess effects of MEF2B on H3K27ac and H3K4me3, I first produced ChIP-seq data for H3K27ac and H3K4me3 in two biological replicates of WT MEF2B-V5 and empty vector cells (methods section 2.8 and 2.9). As a technical control, I confirmed that H3K27ac and H3K4me3 ChIP-seq coverage of regions surrounding TSSs showed the expected M-shaped pattern316,317 (Figure 3.18). I then compared the numbers and fold enrichments of histone modification peaks between WT MEF2B-V5 and empty vector cells and found no differences

66 between the cell lines (Figure 3.19). Thus, if WT MEF2B-V5 expression does impact H3K27ac and H3K4me3, it may do so in a relatively small proportion of the genome. I then investigated H3K27ac and H3K4me3 ChIP-seq coverage in regions near MEF2B- V5 ChIP-seq peaks. Coverage tended to be greater surrounding the centres of MEF2B-V5 peaks than in more distant regions (Figure 3.20a). A similar pattern was evident surrounding the transcription start site of JUN (Figure 3.20b). These M-shaped coverage profiles may have resulted from MEF2 proteins promoting histone modification of the nucleosomes that it bound between. Alternatively, MEF2B may have bound to regions that tended to already have increased H3K27ac and H3K4me3 compared to surrounding regions. If MEF2B activity does contribute to increasing H3K27ac and H3K4me3, then ChIP-seq for these marks in WT MEF2B-V5 cells would be expected to produce greater enrichment of regions near MEF2B binding sites than ChIP-seq for these marks in empty vector cells. Contrary to this hypothesis, the numbers and fold enrichments of H3K27ac and H3K4me3 ChIP-seq peaks near MEF2B-V5 peaks tended to be similar or lower in WT MEF2B-V5 cells compared to empty vector cells (Figure 3.21a-c). Alternatively, MEF2B may tend to increase levels of H3K27ac and H3K4me3 in promoters associated with MEF2B binding sites rather than near the binding sites themselves. Moreover, as H3K4me3 tends to mark promoters but not enhancers35, changes in H3K4me3 may occur predominantly in promoters. However, the numbers and fold enrichments of H3K27ac and H3K4me3 peaks near the TSSs of genes associated with MEF2B-V5 peaks tended to be similar or decreased in the MEF2B-V5 cells compared to empty vector cells (Figure 3.21d-f, methods section 2.15), contrary to my hypothesis. I reasoned that in the above analyses noise from regions at which MEF2B did not impact histone modifications may have been masking signal from any regions in which MEF2B did impact H3K27ac and H3K4me3. In attempt to reduce noise, I repeated the above analyses using only regions associated with the 1,140 candidate direct target genes (i.e. peak-associated genes whose expression increased in WT MEF2B-V5 versus untransfected cells). Such regions were considered the most likely to show differences in histone modification, as changes in histone modification may have mediated the changes in gene expression. Contrary to my hypothesis but similar to the above results, the numbers and fold enrichments of H3K27ac and H3K4me3 peaks associated with the candidate direct target genes were similar or decreased in WT MEF2B-V5

67 compared to empty vector cells (Figure 3.22a-f). I also used the reciprocal approach, first identifying regions in which H3K27ac or H3K4me3 was greater in MEF2B-V5 than empty vector cells, and then assessing whether genes associated with those regions tended to be differentially expressed. Again contrary to my hypothesis, genes associated with increased H3K27ac or H3K4me3 did not tend to have greater expression in WT MEF2B-V5 cells than untransfected cells (Figure 3.22g). Overall, these analyses found no evidence in support of the hypothesis that MEF2B promotes increased levels of H3K27ac or H3K4me3.

3.2.6 Identification of genes regulated by calcium sensitive activities of MEF2B

My final aim was to investigate whether intracellular calcium levels affect MEF2B- dependent gene expression, as MEF2 proteins other than MEF2B have been reported to regulate target genes in a calcium dependent manner193,239 (see section 1.6.4). Moreover, several drivers of DLBCL development affect genes in the BCR signaling pathway, a pathway that includes calcium signaling112. Thus, calcium-sensitive MEF2B target genes may include genes relevant to DLBCL development. I first confirmed that treatment with the calcium ionophore ionomycin increased intracellular calcium levels of HEK293A cells, using a fluorescent calcium dye (Figure 3.23a). From this experiment, I selected a concentration of ionomycin (1.5 µM) that could be used in further experiments to produce a moderate increase in calcium levels. I then confirmed that ionomycin treatment increased MEF2-dependent transcriptional activity in HEK293A cells, using a MEF2-dependent luciferase assay (Figure 3.23b, methods section 2.4). Luciferase assays were performed in untransfected HEK293A cells and thus reflected the combined calcium sensitivity of all endogenously expressed MEF2 family members. To investigate whether elevated calcium levels could increase the transcriptional activity of MEF2B specifically, I analysed three biological replicates of expression microarray data from ionomycin treated WT MEF2B-V5 cells (methods section 2.3), compared to aforementioned data from WT MEF2B-V5 cells. As ionomycin was dissolved in DMSO, cells for all of the aforementioned microarray and RNA-seq data had been DMSO treated in the same manner as for the ionomycin treatment. This comparison of ionomycin and DMSO treated WT MEF2B-V5 cells identified 158 DEGs whose expression was considered to be calcium sensitive (adjusted p-

68 values < 0.05). To produce a list of genes whose calcium sensitivity was most likely MEF2B dependent, I removed from the list of 158 genes the 32 genes that were also DEGs in ionomycin versus DMSO treated untransfected cells (adjusted p-values < 0.05). The remaining 126 genes were genes whose expression was significantly affected by ionomycin treatment only when WT MEF2B-V5 was present (Figure 3.23c). Those 126 genes may be regulated by calcium-sensitive activities of MEF2B. However, the 126 genes represent only 3.2% of the DEGs in DMSO treated WT MEF2B-V5 versus untransfected cells. Thus, intracellular calcium levels modulated only a small proportion of gene expression changes resulting from WT MEF2B-V5 expression in HEK293A cells I next sought to validate these results by performing the same analysis on RNA-seq data from WT MEF2B-V5 and empty vector cells (one sample of each cell line with each treatment). Using the above workflow on the RNA-seq data, I identified 37 genes whose expression was affected by ionomycin treatment only when WT MEF2B-V5 was present (adjusted p-values < 0.05). Of these 37 genes, six had been identified in the same analysis of microarray data. Six out of 37 was a greater overlap than expected by chance (p-value 0.0001, Chi square with Yates correction), supporting that the apparent sensitivity of some genes to both calcium and MEF2B- V5 expression was not an artifact of other factors. Inspection of fold change values for the six genes confirmed that expression of ANO1, CCL8 and NFATC2 tended to be impacted more by ionomycin treatment of WT MEF2B-V5 cells than by similar treatment of control cells (Figure 3.23d). This was consistent with the notion that these three genes are regulated by calcium- sensitive activities of MEF2B. The other three genes, CRELD2, NR4A1, and DUSP6, showed similar fold changes in expression in response to ionomycin treatment regardless of whether MEF2B-V5 was expressed (Figure 3.23d), indicating their calcium sensitivity may be MEF2B independent. Indeed, expression of NR4A1 (also known as NUR77) is known to be increased by MEF2A and MEF2D, but not by MEF2B, in a calcium dependent manner208. The detection of NR4A1 calcium sensitivity helps validate that the method I used could detect calcium-sensitive MEF2-dependent transcriptional regulation. The three genes with the greatest evidence of being regulated by calcium-sensitive MEF2B activity encode the calcium-activated chloride channel ANO1, the cytokine CCL8, and the calcineurin-regulated transcription factor NFATC2. ANO1 is expressed in DLBCL cells, is an oncogene in carcinomas and regulates cell migration318,319, CCL8 is a chemotactic factor that

69 attracts lymphocytes320,321 and NFATC2 has been implicated in GC B-cell survival322. Thus, regulation of ANO1, CCL8 and NFATC2 by MEF2B may be relevant to DLBCL development. To validate the calcium and MEF2B sensitivity of ANO1, CCL8 and NFATC2, I used qRT-PCR to assess mRNA expression of these genes in the two additional monoclonal lines expressing WT MEF2B-V5. As expected, ionomycin treatment significantly increased the expression of all three genes in at least one WT MEF2B-V5 line but not in empty vector cells (p-value < 0.05; Figure 3.24). NFATC2 was also identified as one of the candidate direct MEF2B target genes, supporting the notion that it is regulated by MEF2B.

3.3 Discussion

The overall objective of the research described in Chapter 3 was to characterize the regulatory network of WT MEF2B in HEK293A cells. My research addresses this objective by providing the first genome-wide identification of WT MEF2B binding sites and genes differentially expressed in response to WT MEF2B-V5 expression. Through an integrative analysis of ChIP-seq and gene expression data, I identified 2,668 candidate indirect target genes and 1,141 candidate direct target genes of MEF2B. I considered 261 genes to be high confidence candidate direct target genes and present particularly strong evidence that RHOB, RHOD, CDH13, ITGA5, CAV1 and BCL2 are direct MEF2B target genes. Of these six genes, none have previously been identified as MEF2B target genes and only RHOB and BCL2 were previously identified as candidate direct target genes of other MEF2 family proteins200,207,258. I also present particularly strong evidence that FN1, MYC, MEF2C, TGFB1, CARD11 and NDRG1 are indirect MEF2B target genes. Of these six candidate indirect target genes, only FN1 was previously suggested to be a MEF2B target gene323 and only MYC200,324, MEF2C302 and TGFB1187,325 were reported to be target genes of other MEF2 family proteins. I also present evidence that the cell cycle regulator JUN is a candidate direct MEF2B target gene. This finding is of particular interest because JUN can act as an oncogene326–331 in the cancer types in which MEF2B amplifications were most frequent261,262 (i.e. ovarian, uterine, adrenocortical and esophageal carcinomas, detailed in section 1.7.1). As MEF2B tended to activate expression of its direct target genes, MEF2B amplification is expected to increase JUN expression. Thus, oncogenic effects of MEF2B amplification may be mediated by increased JUN expression. Moreover, my data indicated that MEF2B may interact with a protein complex that

70 may contain JUN, the AP-1 complex. If the interaction of MEF2B and the AP-1 complex promotes expression of AP-1 target genes, this interaction may be another means through which MEF2B amplification may have oncogenic effects. Promotion of TGFB1 expression by MEF2B may also contribute to the development of cancers in which MEF2B is amplified. TGFβ1 can initiate and maintain EMT332, which is necessary for epithelial cells to transform into carcinoma cells. Indeed, the promotion of EMT by MEF2B target genes may explain why the cancers with MEF2B amplification include several types of carcinoma. Consistent with this notion that MEF2B activity can promote EMT, I found that MEF2B-V5 expression increased expression of mesenchymal marker genes, increased abundance of the EMT inducer SNAI2, decreased abundance of the EMT suppressor NDRG1, and produced a change in cell behavior that is also seen in EMT: increased cell migration. Promotion of EMT by MEF2B would also be consistent with evidence that all human MEF2 proteins blocked mesenchymal to epithelial transition325 and with evidence that MEF2A, –C and –D promoted EMT of hepatocellular carcinoma cells187. My research also identified target genes that may mediate tumor suppressor activities of MEF2B in DLBCL cells. MYC is a well-known DLBCL oncogene300 whose expression decreased in WT MEF2B-V5 versus control cells. Thus, decreased MEF2B activity may allow MYC activity to increase and contribute to driving DLBCL development. Moreover, the MEF2B target genes TGFB121 and RHOB333 can act as tumor suppressors by limiting cell proliferation. As TGFB1 and RHOB expression was promoted by MEF2B, reduced MEF2B activity may reduce expression of TGFB1 and RHOB, allowing increased cell proliferation. Indicating that decreased TGFB1 and RHOB expression may contribute to DLBCL, two DLBCL cases with RHOB deletions have been identified (4.2% of cases examined261,262) and alterations have been identified in DLBCL that provide resistance to growth inhibitory effects of TGFβ1334,335. Interestingly, the high confidence candidate direct MEF2B target genes and cell migration regulators CDH13 and CAV1 had decreased expression in DLBCL cells versus normal centroblasts336 and DLBCL cells from advanced versus localized disease337, respectively. Homozygous deletions affecting CDH13 and CAV1 have also been identified in DLBCL (one case of each over 48 samples261,262). Thus, decreased CDH13 and CAV1 expression, as is expected to result from MEF2B mutation, may also contribute to DLBCL development.

I also aimed to determine whether MEF2B activity correlates with changes in histone modification. Only one prior study attempted to correlate MEF2 target gene expression with changes in histone modifications. That study found no association between MEF2A target gene expression and H3 acetylation200. Consistent with that prior study but contrary to my hypothesis, H3K27ac and H3K4me3 did not tend to increase near MEF2B binding sites or the TSSs of candidate direct MEF2B target genes. Because the interactions of MEF2 proteins with certain coregulators are calcium dependent (section 1.6.4), my final aim was to investigate whether intracellular calcium levels affect MEF2B-dependent gene expression. Towards this end, I present the first transcriptome- wide study of the effects of calcium levels on the transcriptional activity of any MEF2 protein. Increased calcium levels had no significant effect on expression of the vast majority of candidate MEF2B target genes, perhaps because the protein-protein interactions that mediate the majority of MEF2B’s transcriptional activity in HEK293A cells are not calcium dependent. As other studies of calcium-responsive MEF2 protein activity used T-cells193,208, myoblasts338 or neurons184,339, MEF2B-dependent target gene expression may show greater calcium sensitivity in T-cells, myoblasts and neurons than in HEK293A cells. Three genes that appeared to be regulated by calcium-sensitive MEF2B activity, ANO1, CCL8 and NFATC2, have not previously been identified as MEF2 target genes and have functions that may be relevant to DLBCL development (noted in section 3.2.6). Overall, the research described in Chapter 3 characterizes the MEF2B regulatory network in HEK293A cells. By integrating DNA-binding and gene expression data, my research demonstrates novel connections between a relatively understudied transcription factor and genes significant to oncogenesis.

Table 3.1 Microarray, RNA-seq and qRT-PCR data for the differential expression of validation set genes in WT MEF2B-V5 versus control cells. The directions of expression change are indicated for the genes that were differentially expressed in the indicated comparisons (p-values < 0.05 for qRT-PCR and RNA-seq data; adjusted p- values < 0.05 for microarray data). In yellow are cases in which the direction of change in RNA- seq or qRT-PCR data matches the direction of change in microarray data. Overall, 27 out of 30 genes (90%) validated in at least one of WT MEF2B-V5 cell lines used for qRT-PCR.

microarray data RNA-seq data qRT-PCR data Gene Symbol WT MEF2B-V5 vs WT MEF2B-V5 WT MEF2B-V5 WT MEF2B-V5 untransfected vs empty D3 vs empty H2 vs empty cells vector cells vector cells vector cells AKT1 up up AMOT up up up up BCL6 up up up CARD11 up up up up CAV1 up up up CCL2 up up up up CDH13 up up up up CTSB up up up up CXCL12 up up FN1 up up up up GNA12 up up up INPP5K down down ITGA5 up up up LGALS1 up up up up MEF2C down down down MYC down down down down NDRG1 down down down down PAK1 up up up PLCG1 up up up up RHOB up up RHOD up up up up ROCK1 up up RRAS up up up up SEMA3C up up SIX1 down up up SMAD2 up up SMAD3 up up SMAD4 up up TGFB1 up up up VEGFB up up

Figure 3.1 Expression microarrays and RNA-seq detected similar alterations in gene expression in response to WT MEF2B-V5 expression. (a) Cells stably transfected with WT MEF2B-V5 have increased abundance of isoform A MEF2B compared to untransfected cells. The bar plot shows the mean fold change in isoform A MEF2B abundance in WT MEF2B-V5 cells compared to untransfected cells. Error bars represent the s.e.m. of three biological replicates. Calculations of protein abundance were performed using densitometry on western blots. A representative western blot is shown. MEF2B was detected using a custom made isoform A MEF2B antibody (see section 2.12). (b) Correlation of fold changes in gene expression between expression microarray and RNA-seq datasets. Only genes whose differential expression was statistically significant in microarray or RNA-seq data (adjusted p-values < 0.05) were included in correlation analysis and are indicated by data points.

Figure 3.2 Cellular function annotation categories enriched in differentially expressed genes. The analysed genes were differentially expressed in (a) WT MEF2B-V5 expressing versus untransfected cells (microarray data) or (b) WT MEF2B-V5 expressing versus empty vector cells (RNA-seq data). Only genes with corrected p-values < 0.05 were considered differentially expressed. Shown are corrected p-values for enrichment, calculated using IPA. Only categories with corrected p-values < 0.05 are shown. Arrows indicate categories discussed in the main text.

Figure 3.3 Validation of differential gene expression in WT MEF2B-V5 versus untransfected and empty vector cells. (a) MEF2B-V5 abundance in the cell lines used for expression microarrays or validation qRT- PCR. The WT MEF2B-V5 cell line was the monoclonal cell line used for microarrays. WT MEF2B-V5 D3 and H2 were additional monoclonal cell lines. All WT MEF2B-V5 cell lines were HEK293A cells stably transfected with WT MEF2B-V5. Empty vector cells were stably transfected with empty pcDNA3 vector. (b) Correlation of fold changes in gene expression between qRT-PCR and microarray data sets for WT MEF2B-V5 versus control cells. (c) Fold change in mRNA expression of 30 genes investigated using qRT-PCR for validation of expression microarray data. Shown is the mean fold change across three biological replicates, compared to either untransfected cells (for qRT-PCR and microarray data) or empty vector cells (for RNA-seq data). Arrows indicate genes discussed in the main text. (d) Bar plots of the qRT- PCR data shown in (c). Note that y-axis scales differ between plots. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 in comparison to empty vector cells. Dark grey plots indicate that the gene was differentially expressed in RNA-seq data for WT MEF2B-V5 versus empty vector cells (p-values < 0.05). All 30 genes were differentially expressed in microarray data for WT MEF2B-V5 versus untransfected cells (adjusted p-values < 0.05).

Figure 3.4 MEF2B-V5 expression alters CARD11, MYC, NDRG1 and MEF2C protein abundance. (a-c) CARD11, MYC, NDRG1 and MEF2C protein abundance was affected by MEF2B-V5 expression. CARD11 and MYC are shown in the same panel as they were detected on the same blots. (d) V5 antibody was used to detect MEF2B-V5 in the lysates that were for the western blots shown in (a-c). For all panels, * P < 0.05 in comparison to empty vector cells (Student’s two-tailed t-test, unpaired). Error bars represent the s.e.m. Relative protein abundance was calculated compared to abundance in untransfected cells except for CARD11 abundance, which was calculated compared to abundance in WT MEF2B-V5 cells. Calculations of protein abundance were performed using densitometry on western blots (4 biological replicates of WT MEF2B-V5 and untransfected cells; 2 biological replicates of empty vector, WT MEF2B-V5 D3 and WT MEF2B-V5 H2 cells). Representative western blots are shown. The WT MEF2B-V5

81 cell line was the monoclonal cell line used for microarrays. WT MEF2B-V5 D3 and H2 were additional monoclonal cell lines. All WT MEF2B-V5 cell lines were HEK293A cells stably transfected with WT MEF2B-V5.

Figure 3.5 Expression of WT MEF2B-V5 alters HEK293A cell migration. (a) MEF2B-V5 expression increased movement into the scratched area of a confluent monolayer. Data from four biological replicates were pooled. The total number of scratches assessed for each sample is shown in parentheses for 12 hour and 20 hour time points, respectively. Error bars represent the s.e.m. (b) Quantification of the change in crystal violet staining after 48 hours of cell growth indicated that the WT MEF2B-V5 cells do not show increased proliferation compared to the control cell lines. Shown is the mean fold change in absorbance at 490 nm compared to untransfected cells. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 in comparison to empty vector cells (Student’s two-tailed t-test, unpaired).

Figure 3.6 Expression of WT MEF2B-V5 increases expression of genes that are upregulated in EMT. Shown is GSEA291,292 for genes upregulated in EMT303. Greater enrichment scores towards the left of the spectrum indicated that genes upregulated in EMT303 tended to have higher expression in cells with WT MEF2B-V5 than in untransfected cells. All genes with detectable expression in expression microarray data were ordered along the x-axis according to the magnitude of their expression change in WT MEF2B-V5 versus untransfected cells. Genes with higher expression in WT than untransfected cells are towards the left, and genes with lower expression in WT than untransfected cells are towards the right. Each black vertical line indicates the position of a gene upregulated in EMT in the ordered list of genes. A running-sum statistic for the enrichment of the EMT genes at each position in the ordered list is shown at the top of the figure.

Figure 3.7 MEF2B-V5 expression alters the abundance of mesenchymal markers. (a,b) The abundance of the mesenchymal proteins SNAI2, vimentin and fibronectin was increased by MEF2B-V5 expression. SNAI2 and vimentin are shown in the same panel as they were detected on the same blots. Relative protein abundance was calculated compared to untransfected cells using densitometry on three biological replicates of western blots. Representative western blots are shown. V5 antibody was used to detect MEF2B-V5 on the fibronectin western blot and in the lysates that were used for the western blots of SNAI2 and vimentin. (c) TGFβ1 concentrations in cell culture media were greater for MEF2B-V5 cells than control cells. Concentrations were assessed using an enzyme-linked immunosorbent assay (ELISA). Shown is the mean of four biological replicates. MEF2B-V5 was detected using V5 antibody in lysates from cells whose media was assayed for TGFβ1. For all panels, * P < 0.05 in comparison to empty vector cells (Student’s two-tailed t-test, unpaired). Error bars represent the s.e.m. The WT MEF2B-V5 cell line was the monoclonal cell line used for microarrays. WT MEF2B-V5 D3 and H2 were additional monoclonal cell lines. All WT MEF2B-V5 cell lines were HEK293A cells stably transfected with WT MEF2B-V5. 85

Figure 3.8 The consistency between replicates and predicted reproducibility of peaks identified from WT MEF2B-V5 ChIP-seq data. (a) Peaks with lower p-values were more likely to have been identified in both replicates of WT MEF2B-V5 ChIP-seq than peaks with higher p-values. All peaks in replicate 2 were ranked by p- value. In each bin of 1,000 rank values there were thus 1,000 peaks from replicate 2 data. However, only a fraction of those peaks were also identified in replicate 1 data. Shown are counts of how many of the replicate 2 peaks were also identified in replicate 1 data, in each bin of rank values. Peaks were identified over input control DNA at a FDR of 0.05. (b) An irreproducible discovery rate (IDR) for a given peak indicates the probability that that peak will 86 not be reproducible in future experiments. The plot shows that peaks in WT MEF2B-V5 ChIP- seq data were enriched for low (< ~0.01) IDR values, consistent with the notion that many peaks in the ChIP-seq data are highly reproducible. IDRs were calculated using both replicates of WT MEF2B-V5 ChIP-seq data and previously described methods305.

Figure 3.9 Binding site motifs in MEF2B ChIP-seq data. Shown are the two most statistically significant motifs identified de novo in sequences within 100 bp of the centres of MEF2B-V5 ChIP-seq peaks, using (a) HOMER283 and (b) MEME- ChIP306. These de novo motifs were most similar to the known motifs for either MEF2 family proteins or the AP-1 complex and a protein in the AP-1 complex, FOS. Only peaks identified in both replicates of ChIP-seq at a FDR of 0.05 were included for analysis. (a) The sizes of the letters are proportional to the frequency of the bases. The statistical significance of the de novo motifs is indicated by p-values. (b). The sizes of the letters indicate their bit scores, out of a maximum of 2. The bit scores indicate the probability of each base at each position. E-values indicate the statistical significance of the de novo motifs. The minimum FDR required to include the match of the de novo motif to the known motif is indicated by a q-value.

Figure 3.10 MEF2A and MEF2C motifs were centrally enriched in MEF2B-V5 ChIP-seq peaks. MEF2A and MEF2C binding site motifs were more likely to occur near the centre than near the edges of peaks identified in both replicates of MEF2B-V5 ChIP-seq. The plot was produced using CentriMo284 and shows the probability of a motif occurring at each position within peak regions. Central enrichment p-values are shown in the plot legend. Peaks were identified over input control DNA at a FDR of 0.05.

Figure 3.11 WT MEF2B-V5-his binds sequences similar to MEF2 motifs in gel shift assays. Probes contained 35 to 37 bp of DNA sequence selected from near the centre of a MEF2B-V5 ChIP-seq peak that was within 5 kb of the TSS of the indicated gene. The unlabelled competitor consisted of the same sequence as the labelled probe. Protein was obtained from E. coli with or without induction of WT MEF2B-V5-his expression. No protein was added to the “probe only” lane. The western blot indicated that MEF2B-V5-his was only detectable in lysates from the induced cells (detected using V5 antibody). The Coomassie stain indicated that many other proteins were present.

Figure 3.12 The distribution of WT MEF2B-V5 ChIP-seq peak regions relative to transcription start sites. Shown is the number of peak regions whose centres were within each bin of distances from the nearest transcription start site. To produce these statistics, GREAT147 was used to match each peak region to the single nearest gene. Only peak regions in common between the two ChIP-seq replicates (identified at a FDR of 0.05) were included in this analysis.

Figure 3.13 WT MEF2B tends to acts as a transcriptional activator. (a) The overlap between genes associated with peaks and DEGs. Among DEGs that were associated with peaks (the dark blue and red sections), more genes had increased than decreased expression in WT MEF2B-V5 versus untransfected cells. (b) Consistent with the notion that MEF2B is a transcriptional activator, the genes whose TSSs tended to be closest to ChIP-seq peaks were those that had increased expression in WT MEF2B-V5 versus untransfected cells. Rank numbers are shown on the x-axis. Lower ranks indicate shorter distances between the genes’ TSSs and ChIP-seq peaks. The y-axis indicates the proportion of genes with ranks at or better than the x-axis value. Rankings were calculated and plotted using BETA286. P-values were calculated compared to the background distribution (one-tailed Kolmogorov-Smirnov test). For both panels, DEGs had adjusted p-values < 0.05 and peaks were identified compared to input control DNA at a FDR of 0.05 in both ChIP-seq replicates.

Figure 3.14 ChIP-qPCR validation of WT MEF2B-V5 ChIP-seq. The mean fold enrichments of DNA regions in three biological replicates of ChIP-qPCR are shown in (a) a heatmap and (b) bar plots. For both panels, gene names indicate genes whose TSSs were within 5 kb up or downstream of the DNA region assessed. All ChIPs used V5 antibody except the ‘IgG’ sample, which used normal mouse immunoglobulin on chromatin from the WT MEF2B-V5 cell line. Fold enrichment was calculated compared to enrichment of 93 an intergenic region not expected to interact with MEF2B, then normalized to fold enrichment in ChIP-qPCR using normal immunoglobulin. The WT MEF2B-V5 cell line was the monoclonal cell line used for ChIP-seq. WT MEF2B-V5 D3 and H2 were additional monoclonal cell lines. All WT MEF2B-V5 cell lines were HEK293A cells stably transfected with WT MEF2B-V5. (a) Yellow indicates greater enrichment and red indicates less enrichment. Fold enrichments in ChIP-seq data were calculated compared to sequenced input DNA using MACS2279. The western blot shows MEF2B-V5 protein abundance in the cell lines used for ChIP-qPCR. MEF2B-V5 was detected using V5 antibody. (b) The colour of the WT MEF2B-V5 bars indicates how many replicates of V5 ChIP-seq on WT MEF2B-V5 cells had a peak in that region (light blue: no replicates; medium blue: 1 replicate; dark blue: both replicates; FDR 0.05). Error bars represent the s.e.m. * P < 0.05 compared to ChIP using normal immunoglobulin (Student’s two tailed t- test, unpaired). Note that y-axis scales differ between plots.

Figure 3.15 Cellular function annotation categories enriched in the 1,141 candidate direct MEF2B target genes. Shown are Benjamini-Hochberg corrected p-values for annotation category enrichment. Enrichment was calculated using IPA. Only categories with corrected p-values < 0.05 are shown.

Figure 3.16 MEF2B binds near the transcription start sites of BCL2 and JUN. (a) Peaks were present in WT ChIP-seq near BCL2 and JUN. * indicates the presence of a significant peak compared to input control DNA (FDR 0.05). Peaks were displayed using the UCSC genome browser340. (b) WT MEF2B-V5 interacted with the promoters of BCL2 and JUN in ChIP-qPCR. The WT MEF2B-V5 cell line was the monoclonal cell line used for ChIP-seq. WT MEF2B-V5 D3 and H2 were additional monoclonal cell lines. All ChIPs used V5 antibody except the “IgG” sample, which used normal mouse immunoglobulin on chromatin from the WT MEF2B-V5 cell line. Fold enrichment of promoter DNA was calculated compared to enrichment of an intergenic DNA region not expected to interact with MEF2B, then normalized to fold enrichment in ChIP-qPCR using normal immunoglobulin. * P < 0.05 compared to IgG (Student’s two-tailed t-test, unpaired). Error bars represent the s.e.m. of three biological replicates.

Figure 3.17 MEF2B knockdown tends to decrease BCL2 and JUN expression and decrease colony formation. (a) JUN and BCL2 mRNA expression tended to decrease in HEK293A and HeLa cells with MEF2B expression reduced below endogenous levels using siRNA. The cells used did not contain V5 tagged MEF2B. Shown is the mean fold change compared to cells transfected with non-targeting control shRNA. Error bars represent the s.e.m of three biological replicates. * P < 0.05 compared to cells transfected with non-targeting shRNA (Student’s two-tailed t-test, unpaired). (b) MEF2B shRNAs reduce MEF2B protein abundance but not the abundance of other MEF2 family proteins. To produce levels of MEF2B detectable using the MEF2B antibody, cells for the MEF2B western blot were co-transfected with shRNA and the WT MEF2B-V5 expression construct. MEF2A, MEF2C and MEF2D were detected in lysates from

97 the same cell populations as were used for (a), which were transfected only with shRNA. (c) MEF2B knockdown reduces colony formation. Shown is the mean fold change in absorbance at 490 nm of lysed stained cells, compared to cells transfected with non-targeting shRNA. * P < 0.05 compared to cells transfected with non-targeting shRNA (Student’s two-tailed t-test, unpaired). Representative stained wells are shown. Error bars represent the s.e.m. of three biological replicates.

Figure 3.18 H3K27ac and H3K4me3 ChIP-seq coverage around transcription start sites. (a) H3K27ac ChIP-seq and (b) H3K4me3 ChIP-seq produced greater coverage of regions neighboring transcription start sites (TSSs) than of more distal regions. (c) Sequencing of input DNA found no increase in coverage around TSSs, indicating that the coverage patterns in (a) and (b) were not artifacts of library construction or sequencing. Coverage values were normalized to the average of the ten data points furthest upstream of TSSs.

Figure 3.19 The number of peaks and fold enrichment values in H3K27ac and H3K4me3 ChIP-seq data are similar between WT MEF2B-V5 and empty vector cells. (a) The total numbers of H3K27ac and H3K4me3 peaks were similar between WT MEF2B-V5 and empty vector cells. (b) Fold enrichments of H3K27ac and H3K4me3 peaks were similar between WT MEF2B-V5 and empty vector cells. Peaks were identified compared to input DNA controls at a FDR of 0.05.

100

Figure 3.20 H3K27ac and H3K4me3 ChIP-seq coverage around the centres of MEF2B-V5 ChIP-seq peak regions. (a) H3K27ac and H3K4me3 ChIP-seq produced greater coverage of regions surrounding the centres of WT MEF2B-V5 peak regions than more distal regions. This pattern was not evident in the coverage of input DNA sequences, indicating that it was not an artifact of library construction or sequencing. Coverage values were normalized to the average of the ten data

101 points furthest upstream of WT MEF2B-V5 peak centres. (b) H3K27ac and H3K4me3 ChIP-seq produced greater coverage of regions surrounding the MEF2B-V5 peak at the transcription start site of JUN than more distal regions. The pattern of H3K27ac and H3K4me3 ChIP-seq coverage was similar between WT MEF2B-V5 cells and empty vector cells. Peaks were displayed using the UCSC genome browser340. A 6 kb window is shown.

102

Figure 3.21 MEF2B-V5 expression did not increase H3K27ac or H3K4me3 near MEF2B- V5 ChIP-seq peak regions or associated transcription start sites. (a) Similar numbers of H3K27ac and H3K4me3 peaks were present near MEF2B-V5 ChIP-seq peaks in WT MEF2B-V5 expressing cells versus empty vector cells. Only peaks within 150 to 700 bp of the centre of MEF2B-V5 peaks were counted, as this distance range corresponds to the region of greatest coverage in Figure 3.20. (b,c) The peaks counted in (a) did not tend to have greater fold enrichments in WT MEF2B-V5 expressing cells than empty vector cells. (d) Similar numbers of H3K27ac and H3K4me3 peaks were present within 2 kb up- or downstream of TSSs associated with MEF2B-V5 ChIP-seq peaks. (e,f) The peaks counted in (d) did not tend to have greater fold enrichments in WT MEF2B-V5 expressing cells than empty vector cells. For all panels, the numbers and fold enrichments of peaks were determined compared to input DNA controls at a FDR of 0.05.

103

104

Figure 3.22 Increased gene expression in WT MEF2B-V5 versus untransfected cells was not associated with increased H3K27ac or H3K4me3. (a) Shown are the numbers of histone modification peaks near MEF2B-V5 peaks that were associated with upregulated genes (i.e. genes that had greater expression in WT MEF2B-V5 cells than untransfected cells). These numbers are similar between WT MEF2B-V5 expressing cells and empty vector cells. Only histone modifications peaks within 150 to 700 bp upstream of the centre of MEF2B-V5 ChIP-seq peaks were counted, as this distance range corresponds to the region of greatest coverage in Figure 3.20. (b,c) Fold enrichments of the peaks counted in (a) did not tend to be greater in WT MEF2B-V5 expressing cells than empty vector cells. (d) Shown are the numbers of histone modification peaks within 2 kb up- or downstream of TSSs of upregulated genes associated with MEF2B-V5 peaks. These numbers are similar between WT MEF2B-V5 expressing cells and empty vector cells. (e,f) Fold enrichments of the peaks counted in (d) did not tend to be greater in WT MEF2B-V5 expressing cells than empty vector cells. (g) Candidate direct MEF2B target genes associated with regions of increased H3K27ac or H3K4me3 are no more highly expressed than genes near regions with decreased H3K27ac or H3K4me3. Histone modification peak regions counted as increased had either more than twofold more enrichment in ChIP-seq on WT MEF2B-V5 cells than in ChIP-seq on empty vector cells or had peaks in both replicates of ChIP-seq on WT MEF2B-V5 cells but neither replicate of ChIP-seq on empty vector cells. Regions counted as decreased were identified using the reciprocal criteria. For all panels, DEGs had adjusted p-values < 0.05 and peaks were identified compared to input control DNA at a FDR of 0.05.

105

Figure 3.23 Effects of increased intracellular calcium levels on MEF2B-dependent gene expression. (a) Ionomycin dose dependently increases intracellular calcium levels in HEK293A cells. Changes in intracellular calcium levels were detected using the FluoForte (Enzo) fluorescent calcium dye. Shown are the means of six technical replicates. Error bars indicate the s.e.m. (b) Six hour treatment with 1.5 µM ionomycin increases MEF2-dependent firefly luciferase expression. Shown are the mean fold changes in normalized luciferase activity compared to

106

DMSO treated cells transfected with a negative control reporter. Error bars represent the s.e.m. of three biological replicates. Significance was determined using an unpaired Student’s two- tailed t-test. (c) The mRNA expression of some genes tended to be affected more by ionomycin in WT MEF2B-V5 cells than in untransfected cells. Shown are expression values for the 126 genes that were DEGs in microarray data for ionomycin versus DMSO treated WT MEF2B-V5 cells but not ionomycin versus DMSO treated untransfected cells. Values are fold changes compared to DMSO treated untransfected cells, shown on a log scale. (d) The mRNA expression of six genes whose expression tended to be affected more by ionomycin in WT MEF2B-V5 cells than in control cells according to both the microarray and RNA-seq datasets. Values are fold changes compared to ionomycin treated MEF2B-V5 cells, as for some genes expression was undetectable in other samples.

107

Figure 3.24 Ionomycin increases expression of ANO1, NFATC2 and CCL8 in MEF2B-V5 cells but not empty vector cells. (a) qRT-PCR was used to quantify the expression of ANO1, NFATC2 and CCL8 in the WT MEF2B-V5 cell lines used for microarrays (WT), as well as in additional stably transfected cell lines (WT D3, WT H2 and empty vector). Shown is the mean fold change across three biological replicates, compared to ionomycin treated WT D3 cells. mRNA expression values were normalized to PGK1 expression. * P < 0.05 (Student’s two-tailed t-test, unpaired). Error bars represent the s.e.m. (b) MEF2B-V5 expression in the cell lines used for (a) was not affected by ionomycin treatment. MEF2B-V5 was detected using V5 antibody.

108

Chapter 4: Impacts of the K4E, Y69H and D83V MEF2B Mutations on the MEF2B Regulatory Network in HEK293A Cells4

4.1 Introduction

The overall objective of research presented in Chapter 4 was to characterize activities of K4E, Y69H and D83V mutant MEF2B and contrast their activities with those of WT MEF2B. To enable the most direct comparison, activities of mutant MEF2B-V5 were examined in the same type of cell line model and in the same assays as were used to characterize activities of WT MEF2B-V5. Mutant and WT MEF2B-V5 were assessed in the same batch for all assays. To address the overall objective, the research presented in Chapter 4 pursued four specific aims. These aims and the research findings that addressed them are summarized in this introduction. First, I aimed to identify differences in gene expression between HEK293A cells expressing mutant and WT MEF2B-V5. I found that candidate direct target gene expression tended to be lower in K4E, Y69H and D83V MEF2B-V5 cells than WT MEF2B-V5 cells, consistent with the notion that the mutations reduce MEF2B’s capacity to activate transcription. Second, I aimed to identify differences in protein abundance and cellular phenotypes between cells expressing mutant and WT MEF2B-V5. Consistent with the idea that mutant MEF2B has decreased transcriptional activity compared to WT MEF2B, mutant MEF2B-V5 had less impact on cell migration and the abundance of mesenchymal proteins than did WT MEF2B. Third, I aimed to determine whether the K4E, Y69H and D83V mutations alter MEF2B’s capacity to bind DNA. In gel shift assays, K4E and D83V MEF2B-V5 but not Y69H MEF2B-V5 showed less interaction with sequences containing MEF2 motifs than WT MEF2B-V5. Consistent with this result, analysis of ChIP-seq data indicated that K4E and D83V MEF2B-V5 bound fewer sites in the genome than WT MEF2B-V5. Finally, I aimed to determine whether the decreased DNA binding of K4E and D83V MEF2B may have been a cause of changes in candidate direct target gene expression in K4E and D83V versus WT MEF2B-V5 cells. Consistent with that notion that there may be a causal

4 Portions of Chapter 4 have been accepted for publication pending formatting: J.R. Pon, J. Wong, S. Saberi, O. Alder, M. Moksa, S.W.G. Cheng, G.B. Morin, P.A. Hoodless, M. Hirst, M.A. Marra. MEF2B Mutations in Non- Hodgkin Lymphoma Dysregulate Cell Migration by Decreasing Transcriptional Activation of MEF2B Target Genes. Nature Communications. Author contributions are provided in the Preface. 109 association, genes associated with regions that had peaks in WT MEF2B-V5 ChIP-seq but did not have peaks in K4E or D83V ChIP-seq tended to have decreased expression in K4E or D83V versus WT MEF2B-V5 cells. Overall, the data described in Chapter 4 support the contention that K4E, Y69H and D83V MEF2B mutations reduce the capacity of MEF2B to activate transcription.

4.2 Results

4.2.1 Microarray data indicates that K4E, Y69H and D83V MEF2B mutations reduce MEF2B transcriptional activity

To allow comparison with the WT MEF2B-V5 cell lines described in Chapter 3, I generated monoclonal HEK293A cell lines stably transfected with K4E, Y69H and D83V MEF2B-V5. Using these cell lines, I first confirmed that the mutant MEF2B-V5 proteins were expressed and compared the stability of mutant MEF2B-V5 to that of WT MEF2B-V5 (methods section 2.12). All three mutations decreased the half-life of MEF2B (Figure 4.1). However, mutant MEF2B-V5 continued to localize to the nucleus (Figure 4.2), indicating that it may still contribute to target gene regulation. To identify differences in gene expression between the mutant and WT MEF2B-V5 cells, I analysed three biological replicates of expression microarray data for one cell line expressing each mutant (methods section 2.5). I identified 975 DEGs in K4E versus WT MEF2B-V5 cells, 3,305 DEGs in Y69H versus WT MEF2B-V5 cells and 3,369 DEGs in D83V versus WT MEF2B-V5 cells (adjusted p-values < 0.05). Depending on the mutant considered, 48-71% of the DEGs in mutant versus WT MEF2B-V5 cells were also DEGs in untransfected versus WT MEF2B-V5 cells (Figure 4.3). This indicated that mutant MEF2B-V5 had altered transcriptional activity at target genes of WT MEF2B-V5. The expression of most target genes was altered further away from its levels in untransfected cells by the expression of WT MEF2B-V5 than by the expression of mutant MEF2B-V5 (Figures 4.3 and 4.4a-c), consistent with the hypothesis that MEF2B mutations reduce MEF2B’s capacity to regulate transcription. Specifically, mutant MEF2B-V5 appeared to have a reduced capacity to increase direct target gene expression compared to WT MEF2B-V5, as expression of candidate direct target genes tended to be lower in cells with mutant than WT MEF2B-V5 (Figure 4.4d).

110

The simplest explanation for how K4E, Y69H and D83V mutations may promote lymphoma development was that all three mutations do so through dysregulation of the same target genes. Thus, I identified DEGs in common between the K4E versus WT MEF2B-V5, Y69H versus WT MEF2B-V5 and D83V versus WT MEF2B-V5 comparisons. Among the 410 common DEGs, 361 (88%) were also DEGs in WT MEF2B-V5 versus untransfected cells (listed in Appendix L). The expression of all 361 of those genes was altered further away from its levels in untransfected cells by the expression of WT MEF2B-V5 than by the expression of K4E, Y69H or D83V MEF2B-V5 (Figure 4.3), consistent with the notion that all three mutations decrease the capacity of MEF2B to regulate those 361 common DEGs. Thus, if all three mutations promote lymphoma development through the same mechanism, that mechanism may involve the deregulation of genes among the 361 common DEGs. Interestingly, the tumor suppressor TGFB1 was among the 361 common DEGs and was identified using IPA as a transcriptional regulator whose known target genes overlapped significantly with the 361 common DEGs (Fisher’s Exact Test p-value 0.002, activation z-score -3.4). Thus, decreased TGFβ signaling may play a central role in mediating effects of MEF2B mutation. I also noted that the expression of genes upregulated in EMT tended to be greater in WT MEF2B-V5 cells than in the mutant MEF2B-V5 cells (FDR q-value 0.07; Figure 4.5). Although transitions away from a mesenchymal–like gene expression signature have not been reported to play a role in lymphomagenesis, this finding supports the conclusion presented in Chapter 3 that WT MEF2B-V5 promotes a mesenchymal-like gene expression signature.

4.2.2 Validation data support that K4E, Y69H and D83V mutations decrease MEF2B transcriptional activity

Given that the abundance of K4E and D83V MEF2B-V5 was greater than or not significantly different than the abundance of WT MEF2B-V5 in the cell lines that were used for microarrays (Figure 4.4e), the decrease in MEF2B transcriptional activity in K4E and D83V cells was not due to decreased MEF2B-V5 abundance. However, Y69H MEF2B-V5 was less abundant than WT MEF2B-V5 in the cell lines that were used for microarrays (Figure 4.4e). I thus performed a validation experiment to demonstrate that the decreased MEF2B transcriptional activity that was evident in the microarray data for Y69H versus WT MEF2B-V5 cells was not entirely due to low Y69H MEF2B-V5 expression or cell-line specific effects.

111

To perform this validation, I first generated an additional cell line expressing Y69H (referred to as Y69H MEF2B-V5 E3) with greater MEF2B-V5 expression than the WT MEF2B- V5 cells (Figure 4.6a). I then used qRT-PCR to compare gene expression between Y69H MEF2B-V5 E3 and WT MEF2B-V5 cells. I assessed expression of the 27 genes whose differential expression in WT MEF2B-V5 versus untransfected cells had been validated (section 3.2.1). Gene expression changes in the qRT-PCR data correlated strongly with those in the microarray data (Spearman coefficient 0.73; Figure 4.6b), indicating that the decreased MEF2B transcriptional activity in Y69H versus WT MEF2B-V5 cells was not entirely due to low Y69H MEF2B-V5 expression or cell-line specific effects. Indeed, the qRT-PCR data indicated that expression of Y69H MEF2B-V5 tended to have less impact on target gene expression levels than expression of WT MEF2B-V5, consistent with the notion that the Y69H mutation decreases MEF2B transcriptional activity (Table 4.1 and Figure 4.6c). I then performed verification experiments to demonstrate that the differential gene expression detected in each of the mutant MEF2B-V5 cells lines compared to WT MEF2B-V5 cells was not predominantly an artifact of the microarray technology or the RNA samples used. To do so, I analysed one replicate of RNA-seq data and three replicates of qRT-PCR data produced using the cell lines that were used for microarrays (methods sections 2.5 and 2.7). This verification qRT-PCR was performed on the same 27 genes as for the above validation and was only performed on K4E and D83V MEF2B-V5 cells, as gene expression changes in Y69H versus WT MEF2B-V5 cells had already been validated using qRT-PCR. Gene expression changes in the qRT-PCR and RNA-seq data correlated with those in microarray data (Table 4.1 and Figures 4.6b and 4.7, Spearman correlations > 0.68). The correlation coefficients were similar to or greater than the correlation coefficients in published studies comparing data produced using these technologies (e.g. qRT-PCR versus microarray data: 0.58296 and 0.61 to 0.67298; RNA-seq versus microarray data: 0.64296 and 0.75297). This similarity supports the notion that the microarray, qRT-PCR and RNA-seq data reflect the same underlying biology, and thus that differential expression in the microarray data was not predominantly an artifact of the technology or RNA samples used. Specifically, the qRT-PCR and RNA-seq data supported the contention that the mutations decrease MEF2B’s transcriptional activity, as mutant MEF2B-V5 tended to have less impact on gene expression levels than WT MEF2B-V5 (Figure 4.6c and 4.8).

112

To further validate that the mutations decreased MEF2B’s transcriptional activity, I compared gene expression changes in mutant versus WT cells with those in MEF2B knockdown versus non-targeting shRNA control cells (methods section 2.3). I assessed expression of JUN, BCL2 and the five validation set genes that appeared to be direct targets of MEF2B (i.e. showed enrichment in ChIP-qPCR in Chapter 3). Six out of seven genes tended to show the same direction of expression change in mutant versus WT MEF2B-V5 cells as in at least one MEF2B knockdown sample compared to the control cells (Figure 4.9). These data support the notion that MEF2B mutations decrease MEF2B transcriptional activity. I also expected that if the K4E, Y69H and D83V mutations decrease MEF2B transcriptional activity, then the K4E, Y69H and D83V mutations would have effects similar to those of mutations known to decrease the transcriptional activity of MEF2 proteins. Such mutations include R3T and R24L, which in MEF2A and MEF2C resulted in dominant negative activity212,341. I introduced R3T and R24L mutations into MEF2B-V5 and compared gene expression in cells stably transfected with R3T and R24L MEF2B-V5 to that in the cells stably transfected with K4E, Y69H and D83V MEF2B-V5. I investigated expression of the five validation set genes that appeared to be direct targets of MEF2B (i.e. showed enrichment in ChIP-qPCR in Chapter 3). All five genes showed the same direction of change in R3T or R24L versus WT MEF2B-V5 cells as in K4E, Y69H or D83V versus WT MEF2B-V5 cells (i.e. decreased expression, Figure 4.10). These data support the notion that K4E, Y69H and D83V mutations, like R3T and R24L mutations, reduce MEF2B’s capacity to activate transcription.

4.2.3 K4E, Y69H and D83V MEF2B mutations alter the abundance of protein from MEF2B target genes and decrease cell migration

I next sought to validate my findings by assessing whether differences in mRNA abundance corresponded with differences in protein abundance. I assessed abundance of the seven proteins that I had demonstrated were affected by expression of WT MEF2B-V5 (i.e. MYC, CARD11, NDRG1, MEF2C, vimentin, fibronectin and SNAI2, Figures 3.4 and 3.7). As expected from the mRNA expression data, the levels of all seven proteins were altered further away from their levels in untransfected cells by the expression of WT MEF2B-V5 than by the expression of Y69H and D83V MEF2B-V5. The differences were statistically significant for MYC, MEF2C and fibronectin abundance in Y69H and D83V versus WT MEF2B-V5 cells (p- 113 values < 0.05; Figures 4.11 and 4.12). Of the seven proteins, the abundance of only MEF2C differed between K4E and WT MEF2B-V5 cells (p-value 0.02). This finding was consistent with evidence that of the mRNAs encoding those seven proteins, only MEF2C mRNA was differentially expressed in K4E versus WT MEF2B-V5 cells (Figure 4.6). Nonetheless, the DEGs that were identified in K4E versus WT MEF2B-V5 cells still included regulators of cell migration. Indeed, IPA analysis predicted that K4E, Y69H and D83V mutations would all decrease cell migration (Figure 4.13a and Appendix M). Consistent with this prediction, K4E, Y69H and D83V MEF2B-V5 cells all filled in scratched areas of a confluent monolayer more slowly than WT MEF2B-V5 cells (Figure 4.13b, methods section 2.13). Slower scratch closure was likely due to decreased cell migration, not decreased proliferation, as no difference in proliferation was detected between mutant and WT MEF2B-V5 cells (Figure 4.13c, methods section 2.14).

4.2.4 Dominant negative effects of K4E and D83V mutations may be masked by the transcriptional activity retained by K4E and D83V MEF2B

As MEF2 family proteins dimerize, mutant MEF2B was expected to interact with WT MEF2 proteins in heterozygous cells. MEF2B mutations would have dominant negative effects if dimers composed of mutant and WT MEF2B had less activity than dimers containing only WT MEF2B. If the K4E, Y69H and D83V mutations have dominant negative effects, I would expect mutant MEF2B-V5 to inhibit the capacity of endogenous MEF2 proteins to activate direct target gene expression. Thus, I would expect candidate direct target gene expression to be lower in mutant MEF2B-V5 cells than in untransfected cells. However, depending on the mutation, only 0.5-1.4% of the candidate direct target genes had decreased expression in mutant MEF2B-V5 versus untransfected cells (adjusted p-values < 0.05). These statistics do not support the hypothesis that these mutations have dominant negative effects. Dominant negative effects could have been masked if mutant MEF2B-V5 retained some transcriptional activity that could compensate for reduction in the activity of endogenous proteins. Consistent with the notion that K4E and D83V MEF2B-V5 retained some capacity to activate transcription, 58% and 51% of candidate direct target genes had increased expression in K4E and D83V MEF2B-V5 cells, respectively, compared to untransfected cells (adjusted p- values < 0.05). In contrast, only 3.7% of candidate target genes had increased expression in

114

Y69H MEF2B-V5 cells compared to untransfected cells (adjusted p-values < 0.05), consistent with the notion that Y69H MEF2B-V5 retained very little or no transcriptional activity. Thus, K4E and D83V MEF2B may have dominant negative effects that are masked by their own retention of some transcriptional activity, whereas Y69H MEF2B may have no dominant negative effects and no retained transcriptional activity. These differences between the mutations may relate to differences in their effects on MEF2B DNA binding, discussed in the following section.

4.2.5 K4E and D83V MEF2B mutations decrease MEF2B DNA binding

I next investigated whether target gene expression differences in mutant versus WT MEF2B-V5 lines could be explained by differences in MEF2B DNA binding. To do so, I performed gel shift assays on his-tagged mutant and WT MEF2B-V5 expressed in E. coli (methods section 2.11). The probes contained sequences similar to MEF2 motifs. Y69H and WT MEF2B-V5-his shifted similar amounts of probe, whereas D83V and K4E MEF2B-V5-his shifted little and no detectable probe, respectively (Figure 4.14). These data indicated that D83V and K4E mutations reduced direct MEF2B DNA binding, whereas the Y69H mutation had no apparent effect on direct MEF2B DNA binding. I then investigated whether genome-wide patterns of DNA binding differed between K4E, D83V and WT MEF2B. To do so, I performed two biological replicates of V5 ChIP-seq on the K4E and D83V MEF2B-V5 cell lines that were used for microarrays (methods sections 2.8 to 2.10). Peaks in K4E and D83V ChIP-seq were identified using the same methods as for identification of peaks in WT ChIP-seq (i.e. peaks were regions of chromatin enriched during IP compared to input chromatin, at a FDR of 0.05). Consistent with the notion that K4E and D83V mutations reduce DNA binding, the number of peaks identified in both replicates of K4E and D83V ChIP-seq was only 0.6% and 7.7%, respectively, of the number of peaks identified in both replicates of WT ChIP-seq (Figure 4.15a). Regions where peaks were lost in K4E or D83V ChIP-seq (i.e. regions with peaks in both replicates of WT ChIP-seq but neither replicate of K4E or D83V ChIP-seq) were highly enriched for MEF2A and MEF2C motifs (Table 4.2), supporting the contention that K4E and D83V mutations disrupt interactions with MEF2 motifs. I then further investigated the 36 and 424 peaks that were identified in both replicates of K4E and D83V ChIP-seq, respectively. Of these peak regions, 97.2% and 91.0%, respectively,

115 also had peaks in both replicates of WT ChIP-seq, indicating that K4E and D83V DNA binding tended to remain restricted to sites that WT MEF2B could bind. Indicating that complexes containing K4E and D83V MEF2B-V5 retained a preference for binding MEF2 motifs, the most enriched motifs in K4E and D83V ChIP-seq peak regions were those of MEF2A and MEF2C (Table 4.3). Even though K4E MEF2B-V5 appeared not to bind MEF2 motifs in gel shift assays, heterodimers of WT and K4E MEF2 proteins may retain some capacity to bind MEF2 sequences. Immunoprecipitation of such heterodimers may have produced the enrichment of MEF2 motifs that was observed in the K4E ChIP-seq data.

4.2.6 Integrative analysis of DNA binding and gene expression data for K4E, D83V and WT MEF2B-V5 cells

I next investigated whether the decreased DNA binding of K4E and D83V MEF2B may have been a cause of the decreased transcriptional activation in K4E and D83V versus WT MEF2B-V5 cells. As fewer peaks were identified in mutant than WT MEF2B-V5 ChIP-seq, I expected that many of the genes associated with WT ChIP-seq peaks would not have associated peaks in mutant ChIP-seq. Indeed, among the genes associated with peaks in both replicates of WT ChIP-seq, 94.3% were not associated with peaks in either replicate of K4E ChIP-seq and 34.4% were not associated with peaks in either replicate of D83V ChIP-seq. Among these genes associated with WT but not mutant peaks, more genes had decreased than increased expression in mutant versus WT MEF2B-V5 cells (Figure 4.15b,c). These data are consistent with the notion that the decreased interaction of K4E and D83V MEF2B-V5 with DNA decreases the capacity of K4E and D83V MEF2B-V5 to activate target gene expression, compared to WT MEF2B-V5. To further support this notion, I also analysed the data using BETA, an approach that assumes that genes closer to peak regions are more likely to be regulated by factors that bind to the peak regions286. Thus, the differential expression of the genes closest to peak regions was assumed to be the best indicator of the effects of altered MEF2B interaction with the peak regions. Analysis using BETA found that regions with peaks in both replicates of WT ChIP-seq but neither replicate of mutant ChIP-seq tended to be closer to genes with decreased expression in mutant versus WT MEF2B-V5 cells than to other genes (Figure 4.16). The tendency of genes close to those peak regions to have decreased expression in mutant versus WT MEF2B-V5 cells

116 was consistent with the conclusion that the decreased interaction of K4E and D83V MEF2B-V5 with DNA decreased the capacity of K4E and D83V MEF2B-V5 to activate target gene expression. I then used ChIP-qPCR to verify K4E and D83V ChIP-seq data for the seven regions at which WT MEF2B-V5 binding had been validated by ChIP-qPCR (i.e. regions near BCL2, JUN, RHOB, CDH13, ITGA5, CAV1 and RHOD). None of these regions had peaks in either replicate of K4E ChIP-seq, and only the region near RHOB had a peak in D83V ChIP-seq. Thus, I expected that V5 ChIP-qPCR on cells with K4E and D83V MEF2B-V5 would produce less enrichment of these regions than V5 ChIP-qPCR on cells with WT MEF2B-V5. Five out of the seven regions showed a decrease in enrichment of at least 40% in both K4E and D83V ChIP- qPCRs compared to WT ChIP-qPCR (Figure 4.17), consistent with the notion that K4E and D83V MEF2B have decreased interaction with those five regions. The reduced interaction of K4E and D83V MEF2B-V5 with these regions may have caused the decreased expression of the associated genes (i.e. BCL2, JUN, CDH13, ITGA5 and CAV1) in K4E or D83V versus WT MEF2B-V5 (microarray data p-values < 0.006). As fold enrichments in K4E and D83V ChIP-qPCR were expected to be low, I also assessed enrichment of two positive control regions selected because they had peaks in both replicates of ChIP-seq on all cell lines. Both positive control regions showed greater than five- fold enrichment in K4E and D83V ChIP-qPCRs (Figure 4.17, p-values < 0.05 compared to IgG ChIP-qPCR), confirming that ChIP on the mutant cell lines did produce enrichment.

4.3 Discussion

The research described in Chapter 4 characterized the transcriptional activity and DNA- binding properties of mutant MEF2B and contrasted these properties with those of WT MEF2B described in Chapter 3. I present the first evidence the K4E, Y69H and D83V mutations decrease the capacity of MEF2B to activate transcription, and the first demonstration that K4E and D83V mutations decrease MEF2B’s capacity to bind DNA. Effects of K4E mutation on DNA binding were expected because K4 is located at the DNA binding interface227 and because replacement of K4 in MEF2C with a non-charged amino acid reduced MEF2C DNA binding212. It was not surprising that the D83V mutation also reduced DNA binding as deletion of residues 77-80 in MEF2C reduced MEF2C DNA binding212. The reduction in DNA binding resulting from K4E

117 and D83V mutation was consistent with the hypothesis described in section 1.7.2 that these and other mutations in the same domains reduce DNA binding. However, the Y69H mutation did not disrupt MEF2B’s capacity to bind DNA in gel shift assays. As Y69 is located at the end of MEF2 domain furthest from the DNA-binding interface (Figure 1.5), any effects of Y69H mutation on the MEF2 domain’s conformation may not affect the DNA-binding interface. Nonetheless, the Y69H mutation reduced the capacity of MEF2B to activate target gene expression. This may be because the Y69H mutation disrupts interactions with co-activators or other transcription factors with which MEF2B cooperates. Consistent with this explanation, Y69 is located near the interface where the co-activator p300 bound MEF2A229. Like Y69, D83 is located near the surface of MEF2B and was predicted to contribute to protein-protein interactions1,7 (Figure 1.5). Thus, the D83V mutation may decrease MEF2B’s capacity to activate transcription both by decreasing interaction with DNA and by decreasing interactions with co-activators or other transcription factors. Conversely, K4 is located at the DNA-binding interface of MEF2B and is thus unlikely to affect protein-protein interactions1,7. The absence of an effect of K4E mutation on interactions with co-activators or other transcription factors may explain why fewer genes were differentially expressed in K4E versus WT MEF2B-V5 cells than in Y69H and D83V versus WT MEF2B-V5 cells (detailed in Figure 4.18). The expression microarray data were also used to investigate whether mutant MEF2B-V5 had dominant negative effects on WT MEF2B-V5 and whether mutant MEF2B-V5 retained any transcriptional activity. Y69H MEF2B-V5 appeared not to have dominant negative effects and had very little or no retained transcriptional activity. In contrast, K4E and D83V MEF2B-V5 appeared to retain some transcriptional activity. This retained activity may have compensated for inhibition of WT MEF2B-V5 by mutant MEF2B-V5, thus masking any dominant negative effects that may have been present. Figure 4.18 illustrates how these properties (i.e. retained activity and possible dominant negative effects) may have resulted from observed and hypothesized deficits of mutant MEF2B. Alternative to the explanation provided in Figure 4.18, dominant negative effects of mutant MEF2B-V5 on endogenous MEF2B may not have been observed because very little endogenous MEF2B was present. Thus, further reductions in MEF2B activity may have been below the threshold of detection. Consistent with this

118 explanation, the fold changes in gene expression induced by knockdown of endogenous MEF2B were small, typically less than a 20% decrease. I also noted that the K4E, Y69H and D83V mutations all decreased the half-life of MEF2B protein. This may reduce the abundance of MEF2B protein in cells with endogenous MEF2B mutations, contributing to the decreased expression of direct MEF2B target genes. Interestingly, ubiquitination of MEF2B has been detected at K89342. As ubiquitination of MEF2A mediates the proteasomal degradation of MEF2A343, the Y69H and D83V mutations may enhance proteasomal degradation of MEF2B by promoting ubiquitination of K89. Alternatively, the mutations may decrease the favourability of proper protein folding. My finding that MEF2B mutations decreased MEF2B transcriptional activity is contrary to a report that MEF2B mutations increased expression of a reporter construct containing BCL6 promoter DNA7. In contrast to the prior study, the research described in Chapter 4 assessed the expression of multiple target genes, reported the relative abundance of mutant and WT MEF2B in the cell lines used, and identified effects of K4E mutation on target gene expression. In the prior report, K4E mutation did not affect BCL6 reporter expression. This may have been because MEF2B DNA binding was not required for the interaction of MEF2B with the BCL6 reporter. Consistent with this notion, my analysis of WT MEF2B ChIP-seq and ChIP-qPCR data (Figure 3.14 from Chapter 3) found no significant enrichment of the region of the BCL6 promoter that was used for the reporter assay. MEF2B may have been recruited to the reporter construct or an upstream regulator of BCL6 by other transcription factors, such that DNA binding activity was not required for MEF2B to promote expression of the BCL6 reporter (see section 1.7.2 for further discussion of this possibility). The prior report proposed that the mechanism by which mutations including Y69H and D83V increased BCL6 reporter expression involved disruption of MEF2B’s interaction with the co-repressor CABIN17. However, the inability of mutant MEF2B to interact with CABIN1 is not expected to allow increased expression of direct target genes unless mutant MEF2B can interact with the regulatory regions of the direct target genes. Thus, the DNA binding deficit of D83V MEF2B may have prevented altered CABIN1 interactions from impacting target gene expression in D83V versus WT MEF2B-V5 cells. Moreover, given that increased intracellular calcium levels were expected to disrupt CABIN1 interactions with MEF2B (section 1.6.4) but increased the expression of only 3% of MEF2B target genes (section 3.2.6), disruption of CABIN1

119 interactions may not increase the expression of the vast majority of MEF2B target genes in HEK293A cells. This could explain why the Y69H mutation did not tend to increase the expression of MEF2B target genes despite disrupting CABIN1 interactions and retaining interactions with DNA. The research described in Chapter 4 identifies genes other than BCL6 whose deregulation in MEF2B mutant versus WT cells may contribute to cancer development. Specifically, expression of mRNA and protein from the MYC proto-oncogene300 was greater in Y69H and D83V MEF2B-V5 cells than in WT MEF2B-V5 cells. Moreover, expression of the TGFB1 tumor suppressor21 tended to be lower in K4E, Y69H and D83V MEF2B-V5 cells than in WT MEF2B-V5 cells. Similarly, the candidate direct MEF2B target genes RHOB, CDH13 and CAV1 have been implicated as tumor suppressors of DLBCL261,262,336,337 and tended to have decreased expression in either K4E, Y69H and D83V cells (CAV1) or just Y69H and D83V cells (RHOB and CDH13) compared to WT MEF2B-V5 cells. Thus, as noted in section 3.3, MEF2B mutations may promote lymphoma development through their effects on MYC, TGFB1, RHOB, CDH13 or CAV1 expression. Overall, the research described in Chapter 4 characterizes the transcriptional activity and DNA-binding properties of mutant MEF2B and contrasts these properties with those of WT MEF2B. My finding that MEF2B mutations reduce transcriptional activity at WT MEF2B target genes is distinct from the findings of a prior investigation7 and is particularly critical to those who might develop small molecules to offset effects of MEF2B mutations. Moreover, my research demonstrates how observations from genome-scale data can aid in the functional characterization of candidate driver mutations.

120

Table 4.1 Microarray, RNA-seq and qRT-PCR data for the differential expression of validation set genes in Y69H versus WT MEF2B-V5 cells. The directions of expression change are indicated for the genes that were differentially expressed in the indicated comparisons (p-values < 0.05 for qRT-PCR and RNA-seq; adjusted p-values < 0.05 for microarray). In yellow are cases in which the direction of change in RNA-seq or qRT- PCR matches the direction of change in microarray data.

Microarray RNA-seq qRT-PCR Gene untrans- Y69H vs Y69H vs Y69H E3 Symbol fected vs WT WT vs WT WT AKT1 down down

AMOT down down down

BCL6 down down

CARD11 down down down down CAV1 down down down

CCL2 down down down down CDH13 down down down down CTSB down down down

CXCL12 down down down down FN1 down down down down GNA12 down down

ITGA5 down down

LGALS1 down down down down MEF2C up up up

MYC up up up up NDRG1 up up up up PAK1 down down

PLCG1 down down down

RHOB down down

RHOD down down down down ROCK1 down down

RRAS down down down

SMAD2 down down

SMAD3 down down

SMAD4 down down

TGFB1 down down

VEGFB down down

total 27 27 12 12

121

Table 4.2 Known motifs enriched in regions with peaks in WT MEF2B-V5 ChIP-seq but not mutant MEF2B-V5 ChIP-seq. Motifs were identified using the ChIPseek implementation of HOMER283. Peaks were identified over input control DNA at a FDR of 0.05.

Peaks in both replicates of WT MEF2B-V5 Peaks in both replicates of WT MEF2B-V5 ChIP-seq but neither replicate of K4E ChIP-seq but neither replicate of D83V MEF2B-V5 ChIP-seq MEF2B-V5 ChIP-seq p-value % of peak p-value % of peak rank (most regions rank (most regions p-value for significant containing p-value for significant containing motif enrichment = 1) the motif enrichment = 1) the motif MEF2A 1x10-715 1 34.6% 1x10-283 3 29.5%

MEF2C 1x10-680 2 36.5% 1x10-283 4 32.0%

JUN-AP1 1x10-661 3 18.5% 1x10-385 1 19.6%

AP1

1x10-589 4 30.6% 1x10-344 2 32.2%

Table 4.3 Known motifs enriched in regions with peaks in mutant MEF2B-V5 ChIP-seq. Motifs were identified using the ChIPseek implementation of HOMER283. Peaks were identified over input control DNA at a FDR of 0.05.

Peaks in both replicates of V5 ChIP-seq Peaks in both replicates of V5 ChIP-seq on K4E MEF2B-V5 cells on D83V MEF2B-V5 cells p-value % of peak p-value % of peak rank (most regions rank (most regions p-value for significant containing p-value for significant containing motif enrichment =1) the motif enrichment =1) the motif MEF2A

1x10-20 1 72.2% 1x10-104 1 47.3%

MEF2C 1x10-19 2 75.0% 1x10-96 2 48.7%

JUN-AP1 0.1 25 8.3% 1x10-40 3 15.4%

AP1

0.001 6 25.0% 1x10-37 4 26.7%

122

Figure 4.1 K4E, Y69H and D83V mutations reduce MEF2B stability. (a) The abundance of mutant MEF2B-V5 decreased more rapidly than the abundance of WT MEF2B-V5 during treatment with the protein synthesis inhibitor cyclohexamide (see methods section 2.12), indicating that mutant MEF2B-V5 degrades more rapidly than WT MEF2B-V5. (b) From the degradation curves, the half-life of mutant MEF2B-V5 was calculated to be shorter

123 than the half-life of WT MEF2B-V5. * P < 0.05 compared to WT MEF2B-V5 (Student’s two tailed t-test, unpaired). For (a) and (b) error bars represent the s.e.m. of three biological replicates. (c) Representative western blots used to produce the degradation curves.

Figure 4.2 WT, K4E, Y69H and D83V MEF2B localize to the nucleus. C: cytoplasm; N: nucleus. Lamin A and β-tubulin were detected as expected in the nuclear or cytoplasmic fractions, respectively.

124

Figure 4.3 Workflow of expression microarray data analysis. The results of this analysis indicate that expression of mutant MEF2B-V5 tended to alter gene expression away from its level in untransfected cells to a lesser extent than expression of WT MEF2B-V5 did. This trend is consistent with the notion that MEF2B mutations reduce MEF2B’s capacity to regulate transcription. DEGs were identified at adjusted p-values < 0.05. DEGs whose expression changed less in response to mutant than WT MEF2B-V5 expression were defined as genes with the same direction of expression change in mutant versus WT MEF2B-V5 cells as in untransfected versus WT MEF2B-V5 cells (adjusted p-values < 0.05 in mutant versus WT MEF2B-V5 cells and untransfected versus WT MEF2B-V5 cells).

125

126

Figure 4.4 Gene expression in mutant versus WT MEF2B-V5 cells, detected using expression microarrays. Centred and scaled expression values in mutant MEF2B-V5, WT MEF2B-V5 and untransfected cells are indicated for genes that were differentially expressed in (a) K4E versus WT MEF2B-V5 cells, (b) Y69H versus WT MEF2B-V5 cells or (c) D83V versus WT MEF2B-V5 cells (adjusted p-values < 0.05). Expression of mutant MEF2B-V5 tended to alter gene expression away from its level in untransfected cells to a lesser extent than expression of WT MEF2B-V5 did, consistent with the notion that MEF2B mutations reduce MEF2B’s capacity to regulate transcription. (d) Centred and scaled expression values in mutant MEF2B-V5, WT MEF2B-V5 and untransfected cells are indicated for the 1,141 genes that were identified as candidate direct targets of WT MEF2B. The candidate direct target genes of WT MEF2B tended to have lower expression in mutant MEF2B-V5 cells than WT MEF2B-V5 cells, consistent with the notion that the mutations decrease MEF2B’s capacity to activate direct target gene expression. For (a-d) red indicates lower expression and yellow indicates higher expression. Data were from GeneChip Human Exon 1.0 ST arrays (Affymetrix). (e) The abundance of K4E and D83V MEF2B-V5 was greater than or not significantly different than the abundance of WT MEF2B-V5. MEF2B-V5 was detected using V5 antibody. Plotted is the mean fold change compared to the WT line. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 compared to WT MEF2B-V5 (Student’s two tailed t-test, unpaired).

127

Figure 4.5 MEF2B mutations decrease the capacity of MEF2B to promote expression of mesenchymal genes. Shown is GSEA291,292 for genes upregulated in EMT303. Greater enrichment scores towards the left of the spectrum indicated that genes upregulated in EMT tended to have higher expression in cells expressing WT MEF2B-V5 than in cells expressing K4E, Y69H or D83V MEF2B-V5. Expression microarray data from K4E, Y69H and D83V cells was treated as one group and data from WT MEF2B-V5 cells was treated as another. All genes with detectable expression were ordered along the x-axis according to the magnitude of their expression change in WT MEF2B- V5 versus mutant cells. Genes with higher expression in WT than mutant cells are towards the left and genes with lower expression in WT than mutant cells are towards the right. Each black vertical line indicates the position of a gene upregulated in EMT in the ordered list of genes. A running-sum statistic for the enrichment of the EMT genes at each position in the ordered list is shown at the top of the figure.

128

129

130

131

Figure 4.6 Validation of expression microarray data using qRT-PCR on additional cell lines. (a) Shown are mean fold changes in MEF2B-V5 abundance compared to WT cells. The mutant MEF2B-V5 cells had similar or greater MEF2B-V5 abundance compared to the WT MEF2B-V5 cells. The Y69H MEF2B-V5 E3 cell line was a different monoclonal cell line than the Y69H MEF2B-V5 cell line that was used for microarrays and RNA-seq. The K4E, D83V and WT MEF2B-V5 cell lines were the cell lines that were used for microarrays and RNA-seq. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 compared to WT MEF2B-V5 (Student’s two tailed t-test, unpaired). A representative western blot is shown. MEF2B-V5 was detected using V5 antibody. (b) Correlation of fold changes in gene expression between qRT- PCR and microarray datasets for mutant versus WT MEF2B-V5 cells. (c) qRT-PCR data for fold changes in mRNA expression in the MEF2B-V5 cell lines and empty vector cells compared to untransfected cells. The mRNA expression of each gene was normalized to PGK1 expression. Note that y-axis scales differ between plots. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 in comparison to WT MEF2B-V5 cells (qRT-PCR data). Bar colour indicates the p-value for differential expression in that cell line compared to WT MEF2B-V5 cells according to RNA-seq data (dark green: < 0.05; light green: ≥ 0.05). Arrows indicate that the gene was differentially expressed in microarray data for that mutant MEF2B-V5 cell line compared to WT MEF2B-V5 cells (adjusted p-values < 0.05; up arrow indicates increased expression in mutant versus WT MEF2B-V5 cells; down arrow indicates decreased expression in mutant versus WT MEF2B-V5 cells).

132

Figure 4.7 Correlation of fold changes in gene expression between expression microarray and RNA-seq datasets for mutant versus WT MEF2B-V5 cells. Fold changes in expression were determined using microarray and RNA-seq data for (a) K4E versus WT MEF2B-V5 cells, (b) Y69H versus WT MEF2B-V5 cells and (c) D83V versus WT MEF2B-V5 cells. The plots display the correlation between the microarray and RNA-seq fold change values. Only genes whose differential expression was statistically significant in microarray or RNA-seq data (adjusted p-values < 0.05) were included in correlation analysis and are indicated by data points.

133

134

Figure 4.8 Analysis of RNA-seq data supports the notion that MEF2B mutations decrease the capacity of MEF2B to activate transcription. (a) Consistent with the microarray analysis presented in Figure 4.3, analysis of RNA-seq data indicates that expression of mutant MEF2B-V5 tended to alter gene expression away from its level in untransfected cells to a lesser extent than expression of WT MEF2B-V5 did. This trend is consistent with the notion that MEF2B mutations reduce MEF2B’s capacity to regulate transcription. DEGs were identified at adjusted p-values < 0.05. DEGs whose expression changed less in response to mutant than WT MEF2B-V5 expression were defined as genes with the same direction of expression change in mutant versus WT MEF2B-V5 cells as in empty vector versus WT MEF2B-V5 cells (adjusted p-values < 0.05 in mutant versus WT MEF2B-V5 cells and empty vector versus WT MEF2B-V5 cells). (b) Shown are centred and scaled expression values generated through analysis of RNA-seq data for the 1,141 candidate direct MEF2B targets genes. Consistent with the microarray analysis presented in Figure 4.4d, the candidate direct target genes tended to have lower expression in mutant MEF2B-V5 cells than in WT MEF2B-V5 cells. This trend is consistent with the notion that mutant MEF2B has a decreased capacity to activate direct target gene expression compared to WT MEF2B. Red indicates lower expression and yellow indicates higher expression.

135

Figure 4.9 MEF2B knockdown and MEF2B mutations have similar effects on candidate direct target gene expression. (a) Some genes upregulated by WT MEF2B-V5 expression tended to have decreased expression when endogenous MEF2B expression was knocked down using shRNAs. Mean fold changes in expression compared to either untransfected cells (for MEF2B-V5 samples) or non-targeting shRNA (for MEF2B shRNA samples) are shown on a log scale. Yellow indicates greater expression and red indicates lower expression. Means were of three biological replicates. (b) Bar plots of the data shown in (a) for CDH13, ITGA5, RHOD, CAV1, and RHOB mRNA expression in cells transfected with shRNA. Bar plots of MEF2B, BCL2 and JUN mRNA expression and western blots of protein expression in cells transfected with shRNA are shown in Figure 3.17. Error bars represent the s.e.m of three biological replicates. * P < 0.05 compared to cells transfected with non-targeting shRNA (students two-tailed t-test, unpaired).

136

Figure 4.10 R3T and R24L mutations decrease MEF2B’s capacity to activate transcription. (a) Expression of R3T or R24L MEF2B-V5 tended to increase candidate direct target gene expression to a lesser extent than expression of WT MEF2B-V5. Shown is the mean fold change in expression compared to cells transfected with empty vector. Note that y-axis scales differ between plots. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 (Student’s two tailed t-test, unpaired) compared to WT MEF2B-V5 cells. Data were produced using qRT- PCR (b) MEF2B-V5 protein abundance was greater in R3T and R24L MEF2B-V5 cells than in WT MEF2B-V5 cells. Relative abundance was calculated compared to WT MEF2B-V5 cells using densitometry on three biological replicates of western blots. A representative western blot is shown. Error bars represent the s.e.m. * P < 0.05 compared to WT MEF2B-V5 (Student’s two tailed t-test, unpaired).

137

138

Figure 4.11 K4E, Y69H and D83V MEF2B mutations alter the abundance of protein from MEF2B target genes. The abundance of (a) CARD11, MYC, (b) NDRG1, (c) MEF2C, (e) vimentin, (f) fibronectin and (g) SNAI2 protein differed between cells expressing WT and mutant MEF2B-V5. For all panels, the mean relative abundance was calculated compared to untransfected cells using densitometry on three biological replicates of western blots. Representative western blots are shown. MEF2B-V5 was detected using V5 antibody. * P < 0.05 in comparison to WT MEF2B- V5 cells (Student’s two-tailed t-test, unpaired). Error bars represent the s.e.m. All cell lines shown were used for expression microarrays except for the empty vector cells. (a) CARD11, MYC and MEF2B-V5 western blots are shown in the same panel as they were all detected on the same membrane. (d) MEF2B-V5 abundance was detected in the lysates that were used for the western blots shown in (b) and (c). (e-g) MEF2B-V5 abundance was detected in the lysates that were used for the other blots in each panel.

139

Figure 4.12 Validation of differential protein abundance in an additional HEK293A cell line expressing Y69H MEF2B-V5. Shown are western blots for (a) CARD11 and MYC, (b) NDRG1 and (c) MEF2C protein abundance. (a) CARD11, MYC and MEF2B-V5 were probed on the same membrane, and are thus shown in the same panel. MEF2B-V5 western blots shown in (b) and (c) used the same lysates as were probed for NDRG1 and MEF2C, respectively. The WT MEF2B-V5 cell line was used for microarrays whereas the Y69H MEF2B-V5 E3 cell line was not used for microarrays.

140

Figure 4.13 Cell migration is affected more by WT than by mutant MEF2B-V5 expression. (a) Cellular function annotation categories enriched in genes differentially expressed in K4E versus WT, Y69H versus WT, and D83V versus WT MEF2B-V5 expressing cells (adjusted p- values were < 0.05). Only genes with the same direction of expression change in all three comparisons were considered. Shown are Benjamini-Hochberg corrected p-values for enrichment, calculated using IPA. Only categories with corrected p-values < 0.05 are shown. Arrows indicate categories discussed in the main text. (b) Movement into the scratched area of a confluent monolayer was increased more by WT than by mutant MEF2B-V5 expression. Scratches were assessed in at least two biological replicates (K4E: 2 biological replicates; Y69H and D83V: 3 biological replicates; WT, untransfected and empty vector: 4 biological replicates). The total number of scratches assessed for each sample is shown in parentheses for 12 hour and 20 hour time points, respectively. * P < 0.05 (Student’s two tailed t-test, unpaired) compared to

141

WT cells. Error bars represent the s.e.m. (c) Cell proliferation was similar between cell lines. Shown are mean fold changes in crystal violet staining after 48 hours of cell growth, compared to untransfected cells. Error bars represent the s.e.m. of three biological replicates.

Figure 4.14 Gel shift assays indicate that K4E and D83V mutations decrease MEF2B DNA binding. Probes contained 35 to 37 bp of DNA sequence selected from near the centre of a MEF2B-V5 ChIP-seq peak that was within 5 kb of the TSS of the indicated gene. The unlabelled competitor consisted of the same sequence as the labelled probe. Protein was obtained from E. coli with or without induction of MEF2B-V5-his expression. No protein was added to the “probe only” lane. The western blot indicated that MEF2B-V5-his was only detectable in lysates from the induced cells (detected using V5 antibody). The Coomassie stain indicated that many other proteins were present.

142

Figure 4.15 K4E and D83V MEF2B binds HEK293A cell DNA at fewer sites than WT MEF2B. (a) Fewer peaks were identified in K4E and D83V MEF2B-V5 ChIP-seq than in WT MEF2B- V5 ChIP-seq. Only peaks identified in both replicates of ChIP-seq on a cell line were counted. (b,c) The numbers of DEGs (adjusted p-values < 0.05, microarray data) associated with the peaks that were present in both replicates of WT MEF2B-V5 ChIP-seq but neither replicate of (b) K4E or (c) D83V MEF2B-V5 ChIP-seq. Of the genes counted, more had decreased than increased expression in mutant versus WT MEF2B-V5 cells. All ChIP-seq peaks were identified over input control DNA at a FDR of 0.05.

143

Figure 4.16 Loss of MEF2B DNA binding is associated with decreased gene expression. Region that lost peaks in mutant compared to WT MEF2B-V5 ChIP-seq tended to be closest to genes with decreased expression in mutant versus WT MEF2B-V5 cells. Rank numbers are shown on the x-axis. Lower ranks indicate shorter distances between the genes’ TSSs and ChIP- seq peaks. The y-axis indicates the proportion of genes with ranks at or better than the x-axis value. Rankings were calculated and plotted using BETA286. P-values were calculated compared to the background distribution (one-tailed Kolmogorov-Smirnov test). Analyses considered only peaks that were present in both replicates of WT ChIP-seq but neither replicate of (a) D83V or (b) K4E mutant ChIP-seq (FDR 0.05). The direction of gene expression change was obtained from expression microarray data (adjusted p-values < 0.05).

144

145

Figure 4.17 ChIP-qPCR data produced for verification of ChIP-seq on mutant and WT MEF2B-V5 cells. The mean fold enrichments of DNA regions in three biological replicates of ChIP-qPCR are shown as (a) a heatmap and (b) bar plots. ChIP-qPCR on K4E and D83V MEF2B-V5 cells tended to produce lower fold enrichments than ChIP-qPCR on WT MEF2B-V5 cells. Gene names indicate genes whose TSSs were within 5 kb up or downstream of the DNA region assessed. All ChIPs used V5 antibody except the ‘IgG’ sample, which used normal mouse immunoglobulin on chromatin from the WT MEF2B-V5 cell line. Fold enrichment was calculated compared to enrichment of an intergenic region not expected to interact with MEF2B, then normalized to fold enrichment in ChIP-qPCR using normal immunoglobulin. The K4E, D83V and WT MEF2B-V5 cell lines were the monoclonal cell lines used for ChIP-seq. WT MEF2B-V5 D3 and H2 were monoclonal cell lines different from the cell line used for ChIP-seq. (a) Yellow indicates greater enrichment and red indicates less enrichment. (b) Bar shading indicates how many replicates of ChIP-seq on that cell line had a peak in that region (FDR 0.05; light blue indicates no replicates; medium blue indicates one replicate; dark blue indicates both replicates). Error bars represent the s.e.m. * P < 0.05 compared to ChIP using normal immunoglobulin (Student’s two tailed t-test, unpaired). Note that y-axis scales differ between plots.

146

147

Figure 4.18 A model of how the K4E, Y69H and D83V MEF2B mutations may decrease target gene expression. Shown are mechanisms through which (a) K4E, (b) Y69H and (c) D83V MEF2B mutations may decrease target gene expression. Circles represent MEF2B. In cells with heterozygous mutations, three types of MEF2B dimers may form: mutant-mutant dimers (left), mutant-WT dimers (centre) and WT-WT dimers (right). These dimers may interact with DNA directly, at sequences resembling MEF2 motifs (top), or indirectly, through interactions with transcription factors (TF) that bind motifs other than the MEF2 motif (bottom). The green arrows indicate the expected levels of target gene expression resulting from the direct and indirect interaction of each type of dimer with DNA. This figure was based on data from gel shift, ChIP-seq and expression microarray experiments. Effects on co-activator interactions are hypothesized but have not yet been investigated. Note that my findings could be equally well explained if interactions with other transcription factors were disrupted instead of interactions with co-activators. For simplicity, only the disruption of co-activator interactions is illustrated.

148

Chapter 5: Potential Roles of MEF2B Mutations in DLBCL Development5

5.1 Introduction

Having completed studies of the function of mutant and WT MEF2B in HEK293A cells (Chapters 3 and 4), my next objective was to characterize the roles of mutant and WT MEF2B in DLBCL cells (Chapter 5). Towards that end, the research presented in Chapter 5 pursued five specific aims. These aims and the research findings that address them are summarized in this introduction. First, I aimed to determine whether MEF2 genes other than MEF2B were expressed in DLBCL cells. Surprisingly, MEF2B had the lowest mRNA expression of all MEF2 family genes in DLBCL cells, providing no support for the notion that mutations were much more frequent in MEF2B than in other MEF2 genes because MEF2B was the only MEF2 protein present in GC B- cells. Second, I aimed to show that mutant MEF2B has a decreased capacity to activate transcription compared to WT MEF2B in DLBCL cells. Evidence supporting this notion was generated by investigating the expression of BCL67. Third, I aimed to determine whether promoters of candidate direct MEF2B target genes identified in HEK293A are bound by endogenous MEF2B in DLBCL cells. I identified candidate MEF2B binding sites near the TSSs of BCL6, BCL2, JUN, RHOB and ABCB4 by performing MEF2B ChIP-qPCR on DLBCL cells. Consistent with results from Chapter 4 that MEF2B mutations decrease DNA binding, lower fold enrichments were found in ChIP-qPCR on DLBCL cells with the D83V MEF2B mutation than on DLBCL cells without MEF2B mutations. I then hypothesized that if K4E, Y69H and D83V mutations promote lymphoma through their decreased capacity to activate transcription, then other MEF2B mutations identified in lymphoma would also reduce the capacity of MEF2B to activate transcription. Other MEF2B mutations identified in DLBCL and FL include nonsense mutations, frameshift mutations and mutations predicted to cause isoform A mRNA transcripts to be translated into proteins almost identical to isoform B MEF2B (Figure 1.2 and Table 1.1). My fourth aim was thus to compare

5 Portions of Chapter 5 have been accepted for publication pending formatting: J.R. Pon, J. Wong, S. Saberi, O. Alder, M. Moksa, S.W.G. Cheng, G.B. Morin, P.A. Hoodless, M. Hirst, M.A. Marra. MEF2B Mutations in Non- Hodgkin Lymphoma Dysregulate Cell Migration by Decreasing Transcriptional Activation of MEF2B Target Genes. Nature Communications. Author contributions are provided in the Preface. 149 the transcriptional activity of isoform A and isoform B MEF2B. Consistent with my hypothesis that all MEF2B mutations in DLBCL decrease MEF2B’s capacity to activate transcription, I found that isoform B MEF2B has a reduced capacity to activate target gene expression compared to isoform A MEF2B. Fifth, I aimed to identify cellular phenotypes affected by MEF2B activity in DLBCL cells. I found that WT MEF2B tended to inhibit DLBCL cell chemotaxis towards FBS and CXCL12. I also report evidence consistent with the notion that MEF2B mutations cooperate with GNA13 mutations to reduce inhibition of chemotaxis. Overall, the research described in Chapter 5 provides further evidence that the MEF2B mutations identified in NHL decrease the capacity of MEF2B to activate direct target gene expression and alter cell migration. Moreover, my research supports the concept that MEF2B mutations may promote DLBCL and FL development by reducing expression of MEF2B target genes that help confine GC B-cells to germinal centres.

5.2 Results

5.2.1 MEF2 family genes are expressed in DLBCL cells

I first sought to determine whether MEF2 genes other than MEF2B were expressed in DLBCL cells. Not only were MEF2A, -C and -D mRNAs detectable in previously reported RNA-seq data from DLBCL patient samples1, they were more abundant than MEF2B mRNA (Figure 5.1a, methods section 2.17). Supporting the notion that at least one MEF2 protein other than MEF2B was present in DLBCL cells, protein the size of MEF2C was detected using MEF2C antibody on western blots of DLBCL cell lines (Figure 5.1b). To investigate whether the process of lymphomagenesis or the presence of MEF2B mutations was associated with changes in MEF2 mRNA expression, I compared the mRNA expression of MEF2 genes in MEF2B mutant DLBCL, MEF2B WT DLBCL and reactive centroblasts (i.e. non-cancerous GC B-cells). No differences were identified between these groups (p-values > 0.12). I next determined the predominant MEF2B isoform expressed in DLBCL cells. Using RNA-seq data from DLBCL patient samples1, I found that 91.7% of MEF2B transcripts were isoform A (Figure 5.1c, methods section 2.17). Isoform A was used for all research described in this and the previous chapters. To investigate whether the process of lymphomagenesis or the presence of MEF2B mutations was associated with a change in the relative abundance of MEF2B

150 isoforms, I determined the percentage of MEF2B transcripts that were isoform A in MEF2B mutant DLBCL, MEF2B WT DLBCL and reactive centroblasts. No differences in relative isoform abundance were identified between these groups (p-values > 0.08). Next, I sought to confirm that mutant and WT MEF2B proteins were present in a DLBCL cell line with an endogenous heterozygous D83V MEF2B mutation. Such a confirmation was obtained using multiple reaction monitoring mass spectrometry (Figure 5.1d, methods section 2.18), consistent with the notion that mutant MEF2B protein is produced and thus may impact cellular function. Moreover, these data support the contention that MEF2B mutations may contribute to lymphoma development even when WT MEF2B protein is present. Finally, I compared the abundance of MEF2B protein in WT MEF2B-V5 HEK293A cells with that in DLBCL cell lines, to assess whether the HEK293A cell line models used in the research described in Chapters 3 and 4 had levels of MEF2B within the physiological range. Indicating that MEF2B abundance did not exceed physiological levels, MEF2B was approximately 200 fold less abundant in WT MEF2B-V5 HEK293A cells than DLBCL cell lines (Figure 5.1e). Thus, the candidate MEF2B binding sites and target genes identified using the WT MEF2B-V5 cells are unlikely to be artifacts of MEF2B overexpression. However, since low affinity binding sites may only become occupied at high transcription factor concentrations310, MEF2B may bind more sites in DLBCL cells than were bound in the MEF2B-V5 HEK293A cells.

5.2.2 MEF2B mutations decrease MEF2B’s capacity to activate BCL6 expression in DLBCL cells

My next aim was to determine whether the K4E, Y69H and D83V MEF2B mutations reduced MEF2B’s capacity to activate transcription in DLBCL cells, as they did in HEK293A cells (described in Chapter 4). To address this aim, I transduced WT and mutant MEF2B-V5 into the DoHH2 DLBCL cell line (methods section 2.2). DoHH2 cells were used as they had the lowest endogenous MEF2B expression of the available DLBCL cell lines. Thus, WT MEF2B-V5 expression would produce a larger fold change in MEF2B abundance in DoHH2 cells than in other DLBCL cell lines, and consequently may produce more readily detectable changes in target gene expression in DoHH2 cells than in other DLBCL cell lines.

151

Using stably transduced DoHH2 cells, I assessed expression of a lymphoma oncogene regulated by MEF2B, BCL67. Consistent with the notion that MEF2B promotes BCL6 expression, BCL6 mRNA expression tended to be greater in WT MEF2B-V5 DoHH2 cells than in untransduced cells (Figure 5.2a). Consistent with my finding that MEF2B mutations decreased the capacity of MEF2B to activate target gene expression in HEK293A cells, BCL6 mRNA expression tended to be lower in cells expressing K4E and D83V MEF2B-V5 than in cells expressing WT MEF2B-V5 (Figure 5.2a). As K4E MEF2B-V5 was more abundant than WT MEF2B-V5, the decreased BCL6 expression in K4E versus WT MEF2B-V5 cells was not due to decreased MEF2B-V5 abundance. Curiously, no decrease in BCL6 expression was evident in Y69H versus WT MEF2B-V5 cells, perhaps because Y69H, unlike K4E and D83V, does not disrupt DNA binding. To further explore possible effects of MEF2B mutation on BCL6 expression, I assessed the abundance of BCL6 protein across a panel of GCB DLBCL cell lines with no BCL6 mutations1,7. BCL6 protein levels were the lowest in the cell line with the lowest MEF2B expression (Figure 5.2b), consistent with the notion that MEF2B activates BCL6 expression. DLBCL cell lines with endogenous D83V MEF2B mutations (DB1 and SUDHL47) had lower BCL6 protein levels than those with WT MEF2B (WSU-DLCL2 and Karpas 422; Figure 5.2b), consistent with my findings in HEK293A and in transduced DoHH2 cells that MEF2B mutations decreased the capacity of MEF2B to activate transcription. The abundance of MEF2B in DB, WSU-DLCL2 and Karpas 422 cells was similar, indicating that differences in MEF2B abundance did not cause the decrease in BCL6 abundance in DB cells compared to WSU- DLCL2 and Karpas 422 cells. A final approach to addressing the question of whether MEF2B mutations increased BCL6 activity in DLBCL cells was to assess whether previously reported BCL6 target genes293,294 were enriched in the genes differentially expressed between GCB DLBCL patient samples with and without MEF2B mutations. This was done using GSEA291,292 on previously reported RNAseq data from 13 MEF2B mutant and 40 MEF2B WT patient samples1, together with three different sets of BCL6 target genes. The different BCL6 target gene sets were identified using different analysis approaches (details in methods section 2.19) and included a set of BCL6 target genes used previously for GSEA on a smaller number of DLBCL patient samples with (3 samples) versus without (8 samples) MEF2B mutations7. Contrary to the results of the

152 prior analysis7, I found that none of the BCL6 target gene sets tended to be differentially expressed between DLBCL patient samples with and without MEF2B mutations (FDR q-values > 0.29). Although these data do not provide additional evidence that MEF2B mutations decrease MEF2B activity, they also do not support a previous report that MEF2B mutations increase BCL6 activity in patient samples7.

5.2.3 Endogenous MEF2B in DLBCL cells binds near the TSSs of BCL6, BCL2, RHOB, ABCB4, ITGA5 and JUN

I next aimed to identify genomic loci bound by endogenous MEF2B in DLBCL cells. Towards this end, I used a MEF2B antibody to perform ChIP-qPCR on DLBCL cell lines with (DB) and without (Karpas 422 and WSU-DLCL2) endogenous MEF2B mutations (methods section 2.10). I assessed the enrichment of seven regions that validated in ChIP-qPCR on WT MEF2B-V5 HEK293A cells (Chapter 3). Regions near BCL6, BCL2, RHOB, ABCB4, ITGA5 and JUN (i.e. 86% of tested regions) had enrichments that were greater than four-fold and were statistically significant in at least one of the DLBCL cell lines compared to an IgG control (Figure 5.3). These data indicated that some regions bound by MEF2B-V5 in HEK293A cells were also bound by endogenous MEF2B in DLBCL cells. Moreover, these data were consistent with the notion that BCL6, BCL2, RHOB, ABCB4, ITGA5 and JUN are direct MEF2B target genes in DLBCL cells. Consistent with my finding that the D83V mutation decreased MEF2B DNA binding (Chapter 4), ChIP-qPCR fold enrichments tended to be lower for the cell line with an endogenous D83V MEF2B mutation (i.e. the DB cell line) than for the cell lines without MEF2B mutations (Figure 5.3). I considered this difference likely due to differences in MEF2B DNA binding, not an artifact of disrupted antibody binding, as the MEF2B antibody recognized an epitope present at the C-terminus of MEF2B, far from the location of D83V. As MEF2B binds near BCL6, BCL2, RHOB, ABCB4, ITGA5 and JUN, I investigated whether MEF2B-V5 expression could alter expression of these genes in DoHH2 DLBCL cells. However, neither these genes nor five indirect MEF2B target genes that were validated in HEK293A (MYC, CARD11, FN1, MEF2C and TGFB1) were differentially expressed in WT MEF2B-V5 versus untransduced DoHH2 (p-values > 0.1, data not shown). MEF2B-V5 expression may have been too low to detectably perturb the expression of these genes or cell type specific differences may have prevented MEF2B-V5 from altering their expression. It is also

153 possible that MEF2B only activates the expression of most of its target genes in DLBCL cells following BCR activation, as is the case for MEF2C240. To test this possibility, I investigated gene expression in the transduced DoHH2 cells following BCR stimulation with anti-IgM antibody. However, differences in gene expression remained undetectable (p-values > 0.1, data not shown).

5.2.4 Isoform B MEF2B has decreased transcriptional activity compared to isoform A MEF2B

If the K4E, Y69H and D83V mutations promote lymphoma development by decreasing the capacity of MEF2B to activate transcription, the most parsimonious explanation for how other MEF2B mutations contribute to lymphoma development would be that they also decrease MEF2B’s capacity to activate transcription. Some MEF2B mutations identified in lymphoma (i.e. P256, P267 and L269 frameshift mutations1,3,7) were predicted to cause proteins almost identical to isoform B MEF2B to be produced from isoform A transcripts7 (Figure 1.2). If these isoform-switching mutations, like other MEF2B mutations, result in a decrease in MEF2B transcriptional activity, then I would expect the isoform-switched mutant proteins to have a decreased capacity to activate transcription compared to WT isoform A MEF2B. Assuming that the isoform-switched mutant proteins and isoform B MEF2B behave similarly, I hypothesized that isoform B MEF2B would also have a decreased capacity to activate transcription compared to isoform A MEF2B. This hypothesis was supported by evidence that isoform A-specific regions are required for efficient transcriptional activation. Specifically, deletions from the C- terminus to D272 or Y223 in the mouse homolog of isoform A MEF2B resulted in reduced transcriptional activity212. Similar to these truncated forms of MEF2B, the amino acid sequence C-terminal to P256 in isoform A MEF2B is not present in isoform B MEF2B. To investigate my hypothesis about isoform A and B MEF2B, I generated a polyclonal HEK293A cell line stably expressing WT isoform B MEF2B-V5 (methods section 2.1). Gene expression in this line was compared to that in the WT isoform A MEF2B-V5 line used for microarrays in Chapters 3 and 4. Consistent with my hypothesis that isoform B MEF2B has decreased transcriptional activity compared to isoform A MEF2B, the expression of all ten genes investigated tended to be affected more by isoform A than by isoform B MEF2B-V5 expression (Figure 5.4). The abundance of isoform A MEF2B-V5 protein was not greater than that of

154 isoform B MEF2B-V5 protein, such that the difference in their activities could not have been due to differences in MEF2B-V5 abundance. Assuming that isoform-switched mutant MEF2B protein behaves like isoform B MEF2B, the isoform-switching MEF2B mutations would be expected to decrease MEF2B’s capacity to activate transcription, consistent with the effects of other MEF2B mutations.

5.2.5 MEF2B activity inhibits DLBCL chemotaxis

I next explored phenotypes that might be affected by MEF2B mutations in DLBCL cells. I first used previously reported RNA-seq data1 to identify 400 DEGs in 13 GCB DLBCL samples with MEF2B mutation versus 40 GCB DLBCL samples without MEF2B mutations (adjusted p-values < 0.05, methods section 2.17). Among those 400 DEGs, the highest confidence MEF2B target genes were the 21 genes that were also identified as candidate MEF2B target genes using HEK293A cells (Chapter 3). However, none of those 21 genes had functions expected to promote lymphoma development (Table 5.1). Confounding effects of other mutations in the DLBCL samples may have prevented the genes through which MEF2B does promote lymphoma development from being identified as DEGs. Alternatively MEF2B mutations may promote lymphoma development through genes that are affected by MEF2B mutations in DLBCL cells but not HEK293A cells. To explore this possibility, I further analysed the set of all DEGs in the DLBCL sample analysis. IPA analysis of the 400 DEGs in MEF2B mutant versus WT DLBCL samples indicated that the most enriched annotation category was cellular movement (adjusted p-value 0.028). When a less stringent p-value threshold of 0.1 was used, allowing 489 DEGs to be identified, cellular movement was even more significantly enriched (adjusted p-value 1x10-3, Figure 5.5a) and predictions of increased cellular movement were made with even greater confidence (z-score increased from 0.68 to 2.5, Appendix N). An alternative method of analysis using DAVID288,289 also found that the term “movement of cell or subcellular component” (GO:0006928) was enriched (adjusted p-value 0.04) in the DEGs with adjusted p-values < 0.1, supporting the prediction that MEF2B mutations affect cell movement. The prediction that MEF2B mutations increase DLBCL cell movement was surprising considering that MEF2B mutations decreased HEK293A cell movement (section 4.2.3). One reason why opposite effects on cell migration may occur in different cell types is that genes

155 mediating effects of MEF2B mutation on the movement of one cell type tended not to be affected by MEF2B mutations in the other cell type. Consistent with this explanation, none of the cellular movement DEGs in K4E, Y69H and D83V versus WT MEF2B-V5 HEK293A cells were also DEGs in MEF2B mutant versus WT DLBCL patient samples. Even if similar sets of cell migration genes were regulated by MEF2B in DLBCL and HEK293A cells, opposite effects on cell migration could occur if the genes mediating effects of MEF2B mutation on one cell type tended not to affect cell migration when differentially expressed in the other cell type. Indeed, only 4 of the 51 cellular movement DEGs in K4E, Y69H and D83V versus WT MEF2B-V5 HEK293A cells had annotated roles in lymphocyte chemotaxis in IPA (i.e. S1PR1, GNA12, GNAI2 and TGFB1). Only one of those four genes, GNA12, had a direction of expression change in mutant vs WT MEF2B-V5 HEK293A cells that would be expected to increase DLBCL chemotaxis. Moreover, the differential expression of GNA12 failed to validate in qRT-PCR (Figure 4.6) and GNA12 was not associated with a WT MEF2B-V5 ChIP-seq peak. Thus, increased DLBCL chemotaxis may be mediated by genes other than those identified as MEF2B target genes using HEK293A cells. The genes that do mediate increased DLBCL chemotaxis may include those that were in the “chemotaxis of lymphocytes” IPA annotation group and had a direction of expression change in MEF2B mutant versus WT DLBCL patient samples consistent with increased chemotaxis: CCL13, CCL14, CCL24, IL21, CXCL8 and SAA1 (adjusted p-values < 0.05). To validate the prediction that MEF2B mutations increase DLBCL chemotaxis, I assessed the chemotaxis of DoHH2 cells expressing mutant or WT MEF2B-V5 towards FBS (methods section 2.20). Chemotaxis towards FBS tended to be greater in cell lines with lower MEF2B expression (Figure 5.6a), consistent with the notion that MEF2B activity inhibits chemotaxis. I next assessed chemotaxis towards CXCL12, a chemokine that attracts GC B-cells towards the dark zone of germinal centres344. Despite having similar MEF2B-V5 expression, cells expressing K4E, Y69H or D83V MEF2B-V5 tended to show greater chemotaxis towards CXCL12 than cells expressing WT MEF2B-V5 (Figure 5.6a). These data are consistent with the predictions made using IPA and with the notion that MEF2B mutations reduce inhibition of DLBCL chemotaxis. Untransduced cells also tended to have increased chemotaxis compared to WT cells, indicating that decreased MEF2B abundance tended to reduce inhibition of chemotaxis towards CXCL12.

156

Interestingly, out of five GCB DBCL cell lines, DoHH2 showed the greatest chemotaxis towards CXCL12. DoHH2 and was also the only line in which endogenous MEF2B and GNA13 both had little or no detectable expression (Figure 5.6b). The protein encoded by GNA13, Gα13, is known to inhibit B-cell migration beyond germinal centres345. Therefore, Gα13 deficiency may cooperate with decreased MEF2B activity to reduce inhibition of cell migration. Indeed, using mutation data for GCB DLBCL patient samples from a previous study1, I found that MEF2B mutations tended to co-occur with loss-of-function mutations in GNA13 (p-value 0.06, Fisher’s exact test). Targets cooperatively regulated by MEF2B and GNA13 mutation may include Rho proteins. Specifically, RHOB and RHOD are suppressors of cell migration that are expressed in GC B-cells333,346, are predicted to act downstream of Gα13347–350 and were identified as candidate direct MEF2B target genes (Chapter 3 and section 5.2.3). However, RHOB and -D expression was not affected by the K4E mutation (section 4.2.2), indicating that some MEF2B mutations may cooperate with GNA13 mutations through other mechanisms. If MEF2B and GNA13 mutations promote DLBCL development by de-repressing chemotaxis, I would expect that genes involved in the regulation of chemotaxis would be differentially expressed in DLBCL patient samples compared to centroblasts (i.e. non-cancerous GC B-cells). Consistent with this expectation, cellular movement was the most enriched IPA annotation category (adjusted p-value 2x10-44, Figure 5.5b) in the 5,042 genes that were differentially expressed between 53 GCB DLBCL samples and 13 normal centroblast samples (adjusted p-values < 0.05; methods section 2.17) and IPA predicted that DLBCL cells would be more migratory than centroblasts (Appendix O). Consistent with the notion that some of the lymphocyte chemotaxis genes that had increased expression in MEF2B mutant versus WT DLBCL patient samples may contribute to lymphoma development, four of those six genes (CCL13, CCL14, CCL24 and SAA1) also had increased expression in DLBCL patient samples versus centroblasts (adjusted p-values <0.05).

5.3 Discussion

The research described in Chapter 5 aimed to characterize the roles of mutant and WT MEF2B in DLBCL cells. I first found that MEF2A, -C and –D mRNA was more abundant than MEF2B mRNA in DLBCL patient samples, indicating that MEF2A, -C and -D protein may also have been more abundant than MEF2B protein. Although it remains possible that MEF2B is

157 more frequently mutated than its paralogs because translational regulation allows MEF2B to be the most abundant MEF2 protein, it is also possible that MEF2B is more frequently mutated than its paralogs because it has distinct cellular functions. A further aim of the research described in Chapter 5 was to investigate the effect of MEF2B mutations on BCL6 expression in DLBCL cells. My research was the first to directly compare the transcriptional activity of WT and mutant MEF2B in the same DLBCL cell line, an approach that minimizes the chance of endogenous genetic or epigenetic differences between cell lines confounding the results. Moreover, my correlational analyses used larger sample sizes than the prior report assessing effects of MEF2B mutation on BCL6 expression (i.e. 48 patient samples and 5 cell lines rather than 11 patient samples and 3 cell lines). Overall, my findings were consistent with the notion that MEF2B mutations decrease MEF2B’s capacity to activate transcription. The decreased capacity of K4E and D83V MEF2B to activate transcription may result from their DNA binding deficits. Chapter 5 described data produced using DLBCL cells that is consistent with the DNA binding deficits described in section 4.2.5. Specifically, ChIP- qPCR produced lower fold enrichments when performed on DLBCL cells with a D83V MEF2B mutation than when performed on DLBCL cells with only WT MEF2B. The most parsimonious hypothesis for how MEF2B mutations contribute to NHL development was that the different MEF2B mutations all drive NHL through a common effect on gene expression. Consistent with this hypothesis, the K4E, Y69H and D83V mutations tended to decrease MEF2B transcriptional activity and the nonsense and frameshift mutations present in the transactivation domain were expected to do the same. My research indicated MEF2B’s capacity to activate transcription may also be reduced by mutations that are predicted to cause isoform A transcripts to be translated into proteins nearly identical to isoform B MEF2B. This conclusion was based on my novel finding that isoform B MEF2B had a lower capacity to alter gene expression than isoform A MEF2B. Thus, all MEF2B mutations identified in DLBCL may promote lymphoma development by decreasing MEF2B transcriptional activity. Interestingly, a change in the relative abundance of MEF2C isoforms was also implicated in cancer development. Specifically, levels of an inactive MEF2C isoform relative to an active MEF2C isoform were found to be greater in rhabdomyosarcoma cells than normal myoblasts, inhibiting expression of MEF2C target genes that would promote differentiation250.

158

Decreased target gene activation as a result of MEF2B mutations is consistent with the effects that mutations in EZH2 and KMT2D are predicted to have on gene expression1,166. As EZH2, KMT2D and MEF2D are thought to cooperatively regulate common target genes in skeletal muscle315, the effects of MEF2B mutations may converge with those of KMT2D and EZH2 in DLBCL and FL. MEF2B mutations tended to co-occur with EZH2 mutations1, consistent with the idea that they cooperate and thus may drive lymphoma through related pathways. However, the KMT2D mutations showed no correlation or anti-correlation with the occurrence of MEF2B mutations1, providing no support for the notion that KMT2D and MEF2B mutations drive lymphoma through similar pathways. Chapter 5 also describes the novel finding that a MEF2 family protein regulates lymphoma cell migration. Supporting the contention that MEF2 family proteins can regulate the migration of cells derived from the bone marrow, MEF2C was found to regulate myeloid leukemia cell migration351. Specifically, I found that MEF2B mutations tended to reduce inhibition of DLBCL cell chemotaxis. Indicating that the de-repression of chemotaxis may promote DLBCL development, reduced sphingosine-1-phosphate-mediated inhibition of chemotaxis towards CXCL12 was associated with DLBCL development in S1P2-deficient mice345 and Gα13 deficient mice352. Supporting the contention that decreased inhibition of chemotaxis can also contribute to human DLBCL development, the genes encoding Gα13 and 1,353 S1P2 were recurrently affected by loss of function mutations in human DLBCL . Most lymphocytes other than GC B-cells are not tightly confined to lymphoid organs352. Thus, MEF2B mutations may impact the dissemination of GC B-cells more than they would impact the dissemination of other lymphocytes. This may explain why MEF2B mutations are more frequent in lymphomas originating from GC B-cells (i.e. GCB DLBCL and FL) than other lymphomas. MEF2B mutations may promote MCL development through effects that are specific to the K23R and N49S mutations found in MCL or through effects on cellular processes other than cell migration that may also contribute to DLBCL and FL development (e.g. cell proliferation via effects on MYC, TGFB1 or RHOB). Overall, the findings described in Chapter 5 support the novel findings presented in Chapter 4 that MEF2B mutations decrease the capacity of MEF2B to activate transcription. Moreover, my findings support the notion that MEF2B mutations promote GC B-cell derived lymphoma by reducing inhibition of chemotaxis.

159

Table 5.1 Differentially expressed genes untransfected versus WT MEF2B-V5 HEK239A cells and DLBCL patient samples with versus without MEF2B mutations. Listed are all genes that had the same direction of expression change in in untransfected versus WT MEF2B-V5 HEK293A cells (adjusted p-values < 0.05; microarray data) as in DLBCL patient samples with versus without MEF2B mutations (adjusted p-values < 0.05; RNA-seq data14). Directions of expression change in comparisons between other HEK293A cell lines are indicated where the change was identified at an adjusted p-value < 0.05.

HEK293A DLBCL untransfected K4E vs Y69H vs D83V vs Gene Gene name cells vs WT WT WT WT symbol MEF2B-V5 MEF2B- MEF2B- MEF2B- mutant cells V5 cells V5 cells V5 cells vs WT

AMHR2 anti-Mullerian hormone receptor, type II up up up ATOH1 atonal homolog 1 (Drosophila) down down BMP5 bone morphogenetic protein 5 down down down down CGA glycoprotein hormones, alpha polypeptide down up down COL2A1 collagen, type II, alpha 1 up up up up CRYM crystallin, mu up up DHRS2 dehydrogenase/reductase (SDR family) member 2 up up up DPPA2 developmental pluripotency associated 2 down down down DSC3 desmocollin 3 down down down down EEF1A2 eukaryotic translation elongation factor 1 alpha 2 down down down EPHA7 EPH receptor A7 up up up up ESRP1 epithelial splicing regulatory protein 1 down down down down GLT1D1 glycosyltransferase 1 domain containing 1 down down IGSF1 immunoglobulin superfamily, member 1 down down MC4R melanocortin 4 receptor up up up NR5A1 nuclear receptor subfamily 5, group A, member 1 up up

olfactory receptor, family 51, subfamily B, member OR51B4 4 up up up RGS7 regulator of G-protein signaling 7 up up up RIMS1 regulating synaptic membrane exocytosis 1 down down SFN stratifin down down down TFPI2 tissue factor pathway inhibitor 2 down down down

160

161

Figure 5.1 Expression of MEF2 family members in DLBCL cells. (a) MEF2A, -C and -D mRNA expression was greater than MEF2B mRNA expression in RNA- seq data for DLBCL patient samples and normal centroblasts. RPKM values indicate the number of mapped reads per length of transcript in kilobases per million mapped reads. * P < 0.05 compared to MEF2B expression in the same cell type (Student’s two tailed t-test, unpaired). In brackets are the numbers of samples assessed. Error bars represent the s.e.m. (b) Protein the size of MEF2C was detected using MEF2C antibody on western blots of DLBCL cell lines, indicating that MEF2C protein may be present in DLBCL cells. (c) The majority of MEF2B transcripts in DLBCL RNA-seq data were isoform A. In brackets are the numbers of samples scored within each group. Error bars represent the s.e.m. For (a) and (c) all DLBCL samples were of the GCB subtype, and “Mutant” and “WT” indicate MEF2B mutation status. (d) Both D83 and D83V mutant peptide were detected using multiple reaction monitoring mass spectrometry in DLBCL cells with the D83V mutation (the DB cell line), indicating that both mutant and WT MEF2B proteins were present. D83V peptide was not expressed in a DLBCL cell line without MEF2B mutation (Karpas 422), confirming the specificity of D83V peptide detection. (e) MEF2B expression in DLBCL cell lines was much greater than MEF2B expression in the HEK293A cell lines that were used for the research described in Chapters 3 and 4. TBP is shown as a loading control. To enable detection of MEF2B on the same membrane, different amounts of protein were loaded for HEK293A samples than for DLBCL samples.

162

Figure 5.2 MEF2B-V5 activity in DLBCL cells promotes BCL6 expression. (a) Expression of WT MEF2B-V5 in DoHH2 DLBCL cells increased BCL6 mRNA expression. BCL6 mRNA expression was lower in cells expressing K4E and D83V MEF2B-V5 than in cells expressing WT MEF2B-V5. The mean fold change in mRNA expression compared to untransduced cells was determined using qRT-PCR data normalized to PGK1 expression. Values are the mean of six biological replicates. Error bars indicate the s.e.m. * P < 0.05 in comparison to WT MEF2B-V5 cells. The western blot indicates the abundance of MEF2B-V5 in the cell lines at the time of the experiment. MEF2B-V5 was detected using V5 antibody. (b) GCB DLBCL cell lines with MEF2B mutations (i.e. SUDHL4 and DB) or very low MEF2B expression (i.e. DoHH2) had lower BCL6 protein expression than GCB DLBCL cell lines with highly expressed MEF2B and no MEF2B mutations (i.e. Karpas 422 and WSU-DLCL2).

163

Figure 5.3 ChIP-qPCR on DLBCL cells identifies DNA regions bound by MEF2B. The mean fold enrichments of DNA regions over three biological replicates of ChIP-qPCR using MEF2B antibody (ProSci) are shown as (a) a heatmap and (b) bar plots. ChIP-qPCR on DLBCL cells with no MEF2B mutations (i.e. Karpas 422 and WSU-DLCL2) tended to produce greater fold enrichments than ChIP-qPCR on DLBCL cells with an endogenous D83V MEF2B mutation (DB). Gene names indicate genes whose TSSs were within 5 kb up or downstream of the DNA region assessed. Mean fold enrichments were calculated compared to ChIP-qPCR using normal rabbit immunoglobulin (IgG) on the same chromatin. (a) Yellow indicates greater enrichment,

164 whereas red indicates less enrichment. (b) Error bars represent the s.e.m. of three biological replicates. Note that y-axis scales differ between plots. * P < 0.05 compared to DB cells. Dark grey bars indicate P < 0.05 compared to the IgG ChIP. All statistical testing used an unpaired Student’s two tailed t-test.

165

166

Figure 5.4 Stable expression of isoform B MEF2B-V5 affects MEF2B target gene expression less than stable expression of isoform A MEF2B-V5. Data are shown as (a) a heatmap and (b) bar plots. Both panels show mean fold changes in mRNA expression compared to empty vector control cells, over three biological replicates. Data were produced using qRT-PCR and were normalized to PGK1 expression. Note that y-axis scales differ between plots. Error bars represent the s.e.m. of three biological replicates. * P < 0.05 in comparison to isoform A MEF2B-V5 expressing cells (Student’s two tailed t-test, unpaired). Yellow indicates greater expression, whereas red indicates lower expression. (c) MEF2B-V5 protein abundance was greater in the isoform B cell line than in the isoform A cell line. The mean fold change in MEF2B-V5 protein expression compared to the isoform A cell line is shown. Error bars indicate the s.e.m. of three biological replicates. A representative western blot is shown.

167

Figure 5.5 Cellular function annotation categories enriched in genes differentially expressed in DLBCL patient samples. Annotation groups relating to cellular movement (indicated by the arrow) were highly enriched for in both (a) the genes differentially expressed in GCB DLBCL samples with versus without MEF2B mutations (adjusted p-values < 0.1) and (b) the genes differentially expressed between GCB DLBCL cells and centroblasts (adjusted p-values < 0.05). Shown are Benjamini-Hochberg corrected p-values for enrichment, calculated using IPA. Only categories with corrected p-values < 0.05 are shown. 168

Figure 5.6 MEF2B-V5 inhibits DLBCL cell chemotaxis. (a) MEF2B activity decreases DoHH2 cell chemotaxis towards FBS and CXCL12. Shown is the mean fold change in the proportion of cells crossing a Transwell membrane, compared to WT MEF2B-V5 cells. Values are the mean of four (FBS) or five (CXCL12) biological replicates. (b) The DLBCL cell line with no detectable MEF2B or Gα13 shows the greatest chemotaxis towards CXCL12. For all panels, western blots show protein expression in cells at the time of the experiment. MEF2B-V5 was detected using V5 antibody. * P < 0.05 in comparison with WT MEF2B-V5 cells (Student’s two-tailed t-test, unpaired). Error bars indicate the s.e.m.

169

Chapter 6: Conclusions and Future Directions

Transcription factor dysregulation is a key driver of many human diseases, including numerous types of cancer. Mutations in the transcription factor gene MEF2B have recently been implicated in NHL development. However, little was known about the cellular roles of WT or mutant MEF2B. I developed and addressed two hypotheses: (1) Identifying target genes of WT MEF2B will allow identification of cellular phenotypes affected by MEF2B activity and (2) contrasting the DNA binding sites, effects on gene expression and effects on cellular phenotypes of mutant and WT MEF2B will indicate mechanisms through which MEF2B mutations may contribute to lymphoma development. To address these hypotheses, I characterized and contrasted the regulatory networks of WT and mutant MEF2B. Chapter 6 summarizes the conclusions of my research, details the significance of my findings and discusses directions for future research.

6.1 WT MEF2B regulates mediators of cell proliferation, cell migration and EMT

The objective of the research described in Chapter 3 was to characterise the MEF2B regulatory network in HEK293A cells. To achieve this objective, I first identified differences in gene expression, protein abundance and cellular phenotypes in WT MEF2B-V5 versus control cells. I then used ChIP-seq data to identify genome-wide candidate MEF2B binding sites. An integrative analysis of the ChIP-seq and gene expression data identified 2,668 candidate indirect target genes and 1,141 candidate direct target genes of MEF2B. I also found that expression of MEF2B-V5 did not affect levels of H3K27ac and H3K4me3, and identified effects of calcium signaling on MEF2B-dependent transcriptional activation. My research constitutes the first identification of genome-wide MEF2B binding sites and candidate direct target genes, the first investigation of how MEF2B may impact histone modifications, and the first genome-wide investigation of the effect of calcium signaling on the activity of any MEF2 protein. I generated the first de novo motifs for sequences bound by MEF2B and presented the novel finding that MEF2B binding sites tended to co-occur with JUN DNA binding motifs. The MEF2B target genes that I’ve identified include genes not previously known to be MEF2B targets (e.g. RHOB, BCL2, JUN, MEF2C, MYC and TGFB1) and genes not previously known to be targets of any MEF2 family protein (e.g. RHOD, CDH13, ITGA5, CAV1,

170

CARD11, NDRG1 and FN1). Among these novel MEF2B target genes are genes that may mediate oncogenic and tumor suppressor activities of MEF2B. I also presented the first report that MEF2B regulates cell migration and promotes a mesenchymal-like gene expression signature. These findings support the novel hypothesis that MEF2B amplification promotes carcinoma development by promoting EMT. Future research using inducible MEF2B expression constructs may provide additional evidence that some of these candidates are indeed direct targets. The expression of true direct targets is expected to change within a short time of transcription factor induction354, perhaps within less than an hour of induction. For instance, treatment with 17β-estradiol produced expression changes within 10 minutes355 and gene expression changes were evident within 20 minutes of induction of Trp63 expression356. It may be of particular interest to investigate whether any of the peak-associated genes with decreased expression in WT MEF2B-V5 versus untransfected cells are direct MEF2B target genes, as doing so may provide evidence that MEF2B can directly repress gene expression. It may also be of interest to validate predicted interactions between particular MEF2B binding sites and promoters. The co-localization of DNA regions may be investigated using chromatin conformation capture technologies, though it is unclear what proportion of interactions identified using these technologies impact gene expression357. Whether particular candidate MEF2B binding sites can contribute to transcriptional regulation could be investigated by cloning them into reporter constructs or by using genome editing technologies to remove them from the genome152. Future research may also investigate whether MEF2B and the AP-1 complex can regulate gene expression co-operatively. Evidence that MEF2B and the AP-1 complex physically interact would support the notion that they could cooperate. The distribution of candidate MEF2B binding sites among genomic elements (i.e. promoters, enhancers, regions with high or low conservation, etc.) also remains to be explored. Moreover, future research may investigate whether analysis of chromatin states and sequence conservation can predict which candidate MEF2B binding sites are involved in the regulation of gene expression. In particular, it may be of interest to investigate whether some of the peak regions that were not associated with DEGs can contribute to the regulation of gene expression. MEF2B binding sites in the peak regions that were not associated with DEGs may regulate genes

171 other than those with which they were associated, or may only participate in gene regulation in different cell types or conditions than those that were used. Future research may also delineate the pathways mediating MEF2B’s effects on indirect target genes and cell migration. Although MEF2B’s effects on cell migration and gene expression were consistent with the notion that MEF2B promotes EMT, it remains to be determined whether MEF2B overexpression in epithelial cells can promote EMT. It may also be of interest to investigate whether overexpression of MEF2B in endothelial cells can promote a related process, endothelial-mesenchymal transition (EndMT). As EMT and EndMT play key roles in development and disease358,359, modulation of these processes may be one way in which alterations affecting MEF2B could lead to disordered development and disease. Future directions may include the use of animal models to investigate whether Mef2b alterations can contribute to developmental disorders and disease. Although Mef2b null mice were reported to have no obvious defects at birth and were viable195, further examination of these mice may reveal differences from control mice. In particular, if the Mef2b null mice have an increased incidence of lymphoma, this would provide strong evidence that MEF2B is a lymphoma tumor suppressor. Similarly, an increased incidence of carcinoma in mice expressing exogenous Mef2b would provide strong evidence that MEF2B can act as a carcinoma oncogene. Future research may also further explore the impact of signaling pathway activation on activities of MEF2B. Of particular interest is the effect of retinoic acid on MEF2B-mediated changes in histone modifications and gene expression. Retinoic acid activates p38360 and phosphorylation of MEF2D by p38 allows MEF2D to interact with the histone methyltransferase KMT2D242. KMT2D recruitment to DNA correlates with increased H3K4me1, H3K4me2 and H3ac316, and KMT2D-dependent gene expression changes have been observed following retinoic acid treatment361. Retinoic acid dependent interactions may also be relevant to lymphoma development as mouse models with inhibited retinoic acid signaling tended to develop lymphoma362,363. MEF2B, like MEF2D, can be phosphorylated by p38213. Thus, retinoic acid may promote interaction of MEF2B with KMT2D. Two other signaling pathways whose effects on MEF2B’s transcriptional activity are of interest for future investigation are the PKC and BCR signaling pathways. This is because BCR signaling enhances the transcriptional activity of MEF2C in B-cells240, and PKC activity cooperates with intracellular calcium levels to enhance MEF2D transcriptional activity in T-

172 cells193,222. However, given that the effects of these pathways on MEF2 protein activity were observed in lymphocytes, their effects on MEF2B’s transcriptional activity may only be evident in lymphocytes.

6.2 MEF2B mutations decrease the capacity of MEF2B to activate transcription by decreasing MEF2B DNA binding

The objective of the research described in Chapter 4 was to characterize activities of K4E, Y69H and D83V MEF2B and contrast their activities with those of WT MEF2B. To achieve this objective, I identified differences in gene expression between HEK293A cells expressing mutant and WT MEF2B-V5. I found that mutant MEF2B had a reduced capacity to alter target gene expression compared to WT MEF2B. This correlated with reduced abundance of protein produced from MEF2B target genes, and a reduced capacity of mutant compared to WT MEF2B to promote cell migration. I then investigated the capacity of mutant MEF2B to bind DNA. K4E and D83V MEF2B-V5 but not Y69H MEF2B-V5 showed less interaction with MEF2 binding site sequences in gel shift assays than WT MEF2B-V5. Consistent with this result, analysis of ChIP-seq data indicated that K4E and D83V MEF2B-V5 bound fewer sites in the genome than WT MEF2B-V5. Finally, by integrating gene expression and DNA binding-site data, I found evidence that decreased DNA binding of K4E and D83V MEF2B may have been a cause of changes in candidate direct target gene expression in K4E and D83V versus WT MEF2B-V5 cells. Altogether, these data support my hypothesis that the MEF2B mutations identified in DLBCL decrease MEF2B’s capacity to alter target gene expression. My research constitutes the first investigation of the effect of MEF2B mutations on DNA binding, the first exploration of how target gene sets differ between WT and mutant MEF2B, and the first report that the MEF2B mutations decrease MEF2B activity. Identifying whether MEF2B mutations reduce or enhance MEF2B target gene activation is key for designing therapeutic agents that selectively target MEF2B mutant cells. Potential therapeutic approaches are discussed in section 6.3. A future direction is to identify co-activators and other transcription factors interacting with MEF2B and determine whether their interaction is disrupted by MEF2B mutations. The most comprehensive approach would be to use mass spectrometry to identify peptides co- immunoprecipitated with mutant and WT MEF2B-V5. Such an experiment could be used to

173 build a map of MEF2B’s protein-protein interactome and may identify a mechanism through which the Y69H mutation decreases MEF2B’s capacity to activate transcription. As the Y69H mutation may disrupt interactions of MEF2B with transcription factors through which it indirectly interacts with DNA, Y69H MEF2B may localize to fewer regions of the genome than WT MEF2B. Motifs present in WT MEF2B-V5 ChIP-seq peak regions but not Y69H MEF2B-V5 ChIP-seq peak regions may be those bound by transcription factors that interact with WT MEF2B-V5 but not Y69H MEF2B-V5. Thus, it may be of interest to analyse Y69H MEF2B-V5 ChIP-seq data. Future research may also contrast the effects of the most common MEF2B mutations in DLBCL and FL (i.e. K4E, Y69H and D83V) with effects of the most common MEF2B mutation in MCL, K23R. Such an investigation may provide insights into why mutations tended to occur at different MEF2B residues in MCL than in DLBCL and FL.

6.3 MEF2B mutations reduce inhibition of chemotaxis by decreasing the capacity of MEF2B to activate target gene expression in DLBCL cells

The objective of the research described in Chapter 5 was to characterize the roles of mutant and WT MEF2B in DLBCL cells. My research was the first to directly compare the transcriptional activity of WT and mutant MEF2B in the same DLBCL cell line, an approach that minimizes confounding factors. Results from this investigation were consistent with the notion that K4E and D83V mutations decrease the capacity of MEF2B to activate transcription. I presented evidence that decreased direct target gene expression may also result from the MEF2B mutations predicted to cause isoform A MEF2B transcripts to be translated into proteins almost identical to isoform B MEF2B. This conclusion was based on the novel finding that isoform B MEF2B has a decreased capacity to activate target gene expression compared to isoform A MEF2B. I also presented the first identification of MEF2B binding sites in DLBCL cells that are near genes other than BCL6 (i.e. BCL2, RHOB, ABCB4, ITGA5 and JUN). Finally, I presented the first indication that a MEF2 family protein regulates B-cell chemotaxis and the first report that MEF2B mutations may cooperate with mutations in another regulator of B-cell chemotaxis, GNA13. These data support the novel concept that MEF2B mutations may promote the development of DLBCL and FL by decreasing the capacity of MEF2B to inhibit GC B-cell migration.

174

Future directions of this research include characterising the cellular mechanisms by which isoform A versus B abundance is regulated and investigating what normal biological function the regulation of isoform A versus B MEF2B abundance may perform. Another future direction is to identify additional genes whose expression is affected by changes in MEF2B activity in DLBCL cells. This may be done using MEF2B-V5 expression constructs that are more resistant to silencing in B-cells than the constructs used for the research described in Chapter 5. Greater MEF2B-V5 expression may induce larger changes in target gene expression that are more readily detectable. Given the central role of MYC alterations in driving aggressive lymphomas300 and the promotion of MYC expression by Y69H and D83V MEF2B mutations (described in Chapter 4), it would be particularly interesting to further investigate if and how MEF2B mutations may affect MYC expression in DLBCL cells. Effects of MEF2B mutations on TGFB1 tumor suppressor in DLBCL cells are also of interest for further investigation, as K4E, Y69H and D83V MEF2B mutations decreased the expression of TGFB1 in HEK293A cells and TGFB1 has been in implicated in DLBCL development334,335. A system for more effectively expressing exogenous genes in DLBCL cells could also be useful for investigating the potential cooperation of MEF2B and GNA13 mutations and the pathways through which they impact chemotaxis. Moreover, such a system could facilitate the validation of findings by allowing MEF2B-V5 to be highly expressed in additional DLBCL cell lines. Other future directions of the research described in Chapter 5 involve the generation of mouse models expressing mutant MEF2B. For instance, effects of Mef2b mutation on the migration patterns of B-cells through normal tissue structures could be studied using mouse models. Immunohistochemical staining has been used to assess the distribution of GC B-cells in lymph nodes, and fluorescence-activated cell sorting has been used to assess the number of GC B-cells escaping into blood, lymph and other tissues352. Similarly, real time microscopy has also been used to track the movement of GC B-cells and quantify the velocity of their movement in different regions of lymph nodes345. These techniques could be used to study effects of Mef2b mutation on B-cell trafficking. Phenotypes of mouse models with Mef2b mutations could also be compared to those of a previously reported mouse model with B-cell specific Mef2c deletion240. Such a comparison could allow differences between the biological roles of MEF2B and MEF2C in B-cells to be identified. Finally, mouse models could be used to investigate whether MEF2B regulates GC B-cell differentiation into plasma cells. Notably, effects on B-cell differentiation

175 may also be studied using the murine lymphoproliferative cell line BCL1, as these cells can be induced to differentiate into a plasma-cell like state364. Given the roles of MEF2B and GNA13 in chemotaxis regulation, murine and human lymphomas with MEF2B or GNA13 mutations may show increased dissemination compared to lymphomas with mutations in neither gene. Identifying such trends could be prognostically significant. Moreover, my findings may impact the development of chemotherapeutics targeting MEF2B mutant cells. One approach to developing targeted therapeutics could be to inhibit oncogenic MEF2B target genes whose expression is increased by MEF2B mutations. MYC may be one such gene, and MYC inhibitors have been developed365,366. An alternative approach could be to identify proteins whose inhibition is lethal only in cells with MEF2B mutations. The research described in this thesis identifies four lymphoma oncogenes (CARD11299, JUN 367, BCL2107 and BCL6119) whose expression is expected to decrease as a result of MEF2B mutation. Further reducing expression of these genes may thus be more detrimental to the viability of cells with MEF2B mutations cells than to the viability of cells without MEF2B mutations. Small molecules that reduce the activity of BCL2 and JUN are already licensed as chemotherapeutics368,369 and anti-hypertensive agents370,371, respectively. Their use for treatment of DLBCL cases with MEF2B mutations may be of interest for further investigation. However, before BCL2, JUN or MYC orientated approaches to treating DLBCL cases with MEF2B mutations can be pursued, it must be demonstrated that MEF2B mutations reduce JUN and BCL2 abundance and increase MYC abundance in DLBCL cells. The low levels of MEF2B-V5 expression and confounding effects of other mutations may have prevented the detection of these effects in the DLBCL cell lines and patient samples used for my research. Nonetheless, my identification of JUN and BCL2 as candidate direct MEF2B target genes in HEK293A cells, my finding that MEF2B binds the promoters of JUN and BCL2 in DLBCL cells, and my finding that MYC is an indirect MEF2B target gene in HEK23A cells have been instrumental to identifying these directions for future research.

6.4 Future directions in the study of MEF2B biology and NHL

The research described in this thesis demonstrates how observations from genome-scale data can aid in the functional characterization of candidate driver mutations. Numerous candidate driver alterations in transcriptional regulators remain to be characterized. For instance,

176 alterations affecting the general transcription factor subunit TAF1353, the co-activators BTG1 and -22,372,373 and the histone modifying enzyme KMT2D1 are thought to contribute to DLBCL development, but the mechanisms by which they may do so remain unclear. Future research into the interplay between genetic alterations, epigenetic modification, gene expression and cellular phenotypes may help fill the gaps in current models of NHL pathogenesis and expand our understanding of transcriptional regulation.

177

Bibliography

1. Morin, R. D. et al. Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature 476, 298–303 (2011).

2. Lohr, J. G. et al. Discovery and prioritization of somatic mutations in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc. Natl. Acad. Sci. U. S. A. 109, 3879–3884 (2012).

3. Pasqualucci, L. et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat. Genet. 43, 830–837 (2011).

4. Beà, S. et al. Landscape of somatic mutations and clonal evolution in mantle cell lymphoma. Proc. Natl. Acad. Sci. U. S. A. 110, 18250–18255 (2013).

5. Zhang, J. et al. Genetic heterogeneity of diffuse large B-cell lymphoma. Proc. Natl. Acad. Sci. U. S. A. 110, 1398–1403 (2013).

6. Meissner, B. et al. The E3 ubiquitin ligase UBR5 is recurrently mutated in mantle cell lymphoma. Blood 121, 3161–3164 (2013).

7. Ying, C. Y. et al. MEF2B mutations lead to deregulated expression of the oncogene BCL6 in diffuse large B cell lymphoma. Nat. Immunol. 14, 1084–1092 (2013).

8. Hanahan, D. & Weinberg, R. A. The hallmarks of cancer. Cell 100, 57–70 (2000).

9. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144, 646–674 (2011).

10. Nowell, P. C. The clonal evolution of tumor cell populations. Science 194, 23–28 (1976).

11. Brosnan, J. A. & Iacobuzio-Donahue, C. A. A new branch on the tree: next-generation sequencing in the study of cancer evolution. Semin. Cell Dev. Biol. 23, 237–242 (2012).

12. Navin, N. et al. Tumour evolution inferred by single-cell sequencing. Nature 472, 90–94 (2011).

13. Stratton, M. R., Campbell, P. J. & Futreal, P. A. The cancer genome. Nature 458, 719–724 (2009).

14. Vogelstein, B. et al. Cancer genome landscapes. Science 339, 1546–1558 (2013).

15. Forbes, S. A. et al. COSMIC (the Catalogue of Somatic Mutations in Cancer): a resource to investigate acquired mutations in human cancer. Nucleic Acids Res. 38, D652–657 (2010).

16. Armitage, P. & Doll, R. The age distribution of cancer and a multi-stage theory of carcinogenesis. Br. J. Cancer 8, 1–12 (1954).

178

17. Kandoth, C. et al. Mutational landscape and significance across 12 major cancer types. Nature 502, 333–339 (2013).

18. Kern, S. E. & Winter, J. M. Elegance, silence and nonsense in the mutations literature for solid tumors. Cancer Biol. Ther. 5, 349–359 (2006).

19. Gnad, F., Baucom, A., Mukhyala, K., Manning, G. & Zhang, Z. Assessment of computational methods for predicting the effects of missense mutations in human cancers. BMC Genomics 14 Suppl 3, S7 (2013).

20. Knudson, A. G., Jr. Mutation and cancer: statistical study of retinoblastoma. Proc. Natl. Acad. Sci. U. S. A. 68, 820–823 (1971).

21. Akhurst, R. J. & Derynck, R. TGF-beta signaling in cancer--a double-edged sword. Trends Cell Biol. 11, S44–51 (2001).

22. Thorne, J. L., Campbell, M. J. & Turner, B. M. Transcription factors, chromatin and cancer. Int. J. Biochem. Cell Biol. 41, 164–175 (2009).

23. Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N. Engl. J. Med. 368, 2059–2074 (2013).

24. Reimand, J., Wagih, O. & Bader, G. D. The mutational landscape of phosphorylation signaling in cancer. Sci. Rep. 3, 2651 (2013).

25. Caron de Fromentel, C. & Soussi, T. TP53 tumor suppressor gene: a model for investigating human mutagenesis. Genes. Chromosomes Cancer 4, 1–15 (1992).

26. Varmus, H. E. Oncogenes and transcriptional control. Science 238, 1337–1339 (1987).

27. Futreal, P. A. et al. A census of human cancer genes. Nat. Rev. Cancer 4, 177–183 (2004).

28. Courey, A. J. & Jia, S. Transcriptional repression: the long and the short of it. Genes Dev. 15, 2786–2796 (2001).

29. Näär, A. M., Lemon, B. D. & Tjian, R. Transcriptional coactivator complexes. Annu. Rev. Biochem. 70, 475–501 (2001).

30. Kouzarides, T. Chromatin modifications and their function. Cell 128, 693–705 (2007).

31. Zaret, K. S. & Carroll, J. S. Pioneer transcription factors: establishing competence for gene expression. Genes Dev. 25, 2227–2241 (2011).

32. Jin, C. et al. H3.3/H2A.Z double variant-containing nucleosomes mark ‘nucleosome-free regions’ of active promoters and other regulatory regions. Nat. Genet. 41, 941–945 (2009).

33. Bannister, A. J. & Kouzarides, T. Regulation of chromatin by histone modifications. Cell Res. 21, 381–395 (2011).

179

34. Shogren-Knaak, M. et al. Histone H4-K16 acetylation controls chromatin structure and protein interactions. Science 311, 844–847 (2006).

35. Zhou, V. W., Goren, A. & Bernstein, B. E. Charting histone modifications and the functional organization of mammalian genomes. Nat. Rev. Genet. 12, 7–18 (2011).

36. Bernstein, B. E. et al. A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell 125, 315–326 (2006).

37. Bieberstein, N. I., Carrillo Oesterreich, F., Straube, K. & Neugebauer, K. M. First exon length controls active chromatin signatures and transcription. Cell Rep. 2, 62–68 (2012).

38. Ardehali, M. B. et al. Drosophila Set1 is the major histone H3 lysine 4 trimethyltransferase with role in transcription. EMBO J. 30, 2817–2828 (2011).

39. Wu, M. et al. Molecular regulation of H3K4 trimethylation by Wdr82, a component of human Set1/COMPASS. Mol. Cell. Biol. 28, 7337–7344 (2008).

40. Wang, P. et al. Global analysis of H3K4 methylation defines MLL family member targets and points to a role for MLL1-mediated H3K4 methylation in the regulation of transcriptional initiation by RNA polymerase II. Mol. Cell. Biol. 29, 6074–6085 (2009).

41. Hu, D. et al. The Mll2 branch of the COMPASS family regulates bivalent promoters in mouse embryonic stem cells. Nat. Struct. Mol. Biol. 20, 1093–1097 (2013).

42. Shen, H. & Laird, P. W. Interplay between the cancer genome and epigenome. Cell 153, 38– 55 (2013).

43. Müller, F. & Tora, L. Chromatin and DNA sequences in defining promoters for transcription initiation. Biochim. Biophys. Acta 1839, 118–128 (2014).

44. Ehrlich, M. et al. Amount and distribution of 5-methylcytosine in human DNA from different types of tissues of cells. Nucleic Acids Res. 10, 2709–2721 (1982).

45. Gardiner-Garden, M. & Frommer, M. CpG islands in vertebrate genomes. J. Mol. Biol. 196, 261–282 (1987).

46. Bird, A. P. CpG-rich islands and the function of DNA methylation. Nature 321, 209–213 (1986).

47. Smale, S. T. & Kadonaga, J. T. The RNA polymerase II core promoter. Annu. Rev. Biochem. 72, 449–479 (2003).

48. Suzuki, Y. et al. Identification and characterization of the potential promoter regions of 1031 kinds of human genes. Genome Res. 11, 677–684 (2001).

49. Long, H. K., Blackledge, N. P. & Klose, R. J. ZF-CxxC domain-containing proteins, CpG islands and the chromatin connection. Biochem. Soc. Trans. 41, 727–740 (2013).

180

50. Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev. 16, 6–21 (2002).

51. Kim, A. & Dean, A. A human globin enhancer causes both discrete and widespread alterations in chromatin structure. Mol. Cell. Biol. 23, 8099–8109 (2003).

52. Fromm, G. et al. An embryonic stage-specific enhancer within the murine β-globin locus mediates domain-wide histone hyperacetylation. Blood 117, 5207–5214 (2011).

53. Travers, A. Chromatin modification by DNA tracking. Proc. Natl. Acad. Sci. U. S. A. 96, 13634–13637 (1999).

54. Forsberg, E. C. & Bresnick, E. H. Histone acetylation beyond promoters: long-range acetylation patterns in the chromatin world. BioEssays News Rev. Mol. Cell. Dev. Biol. 23, 820–830 (2001).

55. Gaszner, M. & Felsenfeld, G. Insulators: exploiting transcriptional and epigenetic mechanisms. Nat. Rev. Genet. 7, 703–713 (2006).

56. Canadian Cancer Society’s Steering Committee. Canadian Cancer Statistics 2010. (2010).

57. Ferlay, J. et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int. J. Cancer J. Int. Cancer 127, 2893–2917 (2010).

58. A clinical evaluation of the International Lymphoma Study Group classification of non- Hodgkin’s lymphoma. The Non-Hodgkin’s Lymphoma Classification Project. Blood 89, 3909–3918 (1997).

59. Pejcic, I. et al. Mantle cell lymphoma-current literature overview. J. BUON Off. J. Balk. Union Oncol. 19, 342–349 (2014).

60. Zhou, Y. et al. Incidence trends of mantle cell lymphoma in the United States between 1992 and 2004. Cancer 113, 791–798 (2008).

61. Sant, M. et al. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood 116, 3724–3734 (2010).

62. Freedman, A., Friedberg, J., Mauch, P., Dalla-Favera, R. & Harris, N. in Non-Hodgkin Lymphomas (Lippincott Williams&Wilkins, 2010).

63. Friedberg, J. W. et al. Effectiveness of first-line management strategies for stage I follicular lymphoma: analysis of the National LymphoCare Study. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 30, 3368–3375 (2012).

64. Freedman, A. Follicular lymphoma: 2014 update on diagnosis and management. Am. J. Hematol. 89, 429–436 (2014).

65. Lossos, I. S. & Gascoyne, R. D. Transformation of follicular lymphoma. Best Pract. Res. Clin. Haematol. 24, 147–163 (2011).

181

66. Link, B. K. et al. Rates and outcomes of follicular lymphoma transformation in the immunochemotherapy era: a report from the University of Iowa/MayoClinic Specialized Program of Research Excellence Molecular Epidemiology Resource. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 31, 3272–3278 (2013).

67. Boyle, J., Beaven, A. W., Diehl, L. F., Prosnitz, L. R. & Kelsey, C. R. Improving Outcomes in Advanced DLBCL: Systemic Approaches and Radiotherapy. Oncol. Williston Park N 28, (2014).

68. Rosenwald, A. et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J. Exp. Med. 198, 851–862 (2003).

69. Savage, K. J. et al. The molecular signature of mediastinal large B-cell lymphoma differs from that of other diffuse large B-cell lymphomas and shares features with classical Hodgkin lymphoma. Blood 102, 3871–3879 (2003).

70. Alizadeh, A. A. et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403, 503–511 (2000).

71. Lenz, G. et al. Stromal gene signatures in large-B-cell lymphomas. N. Engl. J. Med. 359, 2313–2323 (2008).

72. Harris, N. et al. in Non-Hodgkins Lymphomas (Lippincott Williams&Wilkins, 2010).

73. Hernandez-Ilizaliturri, F. J. et al. Higher response to lenalidomide in relapsed/refractory diffuse large B-cell lymphoma in nongerminal center B-cell-like than in germinal center B- cell-like phenotype. Cancer 117, 5058–5066 (2011).

74. Feldman, T. et al. Addition of lenalidomide to rituximab, ifosfamide, carboplatin, etoposide (RICER) in first-relapse/primary refractory diffuse large B-cell lymphoma. Br. J. Haematol. 166, 77–83 (2014).

75. Vitolo, U. et al. Lenalidomide plus R-CHOP21 in elderly patients with untreated diffuse large B-cell lymphoma: results of the REAL07 open-label, multicentre, phase 2 trial. Lancet Oncol. 15, 730–737 (2014).

76. Abrahamsson, A., Dahle, N. & Jerkeman, M. Marked improvement of overall survival in mantle cell lymphoma: a population based study from the Swedish Lymphoma Registry. Leuk. Lymphoma 52, 1929–1935 (2011).

77. Campo, E. & Rule, S. Mantle cell lymphoma: evolving management strategies. Blood (2014). doi:10.1182/blood-2014-05-521898

78. Ujjani, C. & Cheson, B. D. The optimal management of follicular lymphoma: an evolving field. Drugs 73, 1395–1403 (2013).

182

79. Abrahamsson, A. et al. Real world data on primary treatment for mantle cell lymphoma: a Nordic Lymphoma Group observational study. Blood 124, 1288–1295 (2014).

80. King, R. L. & Bagg, A. Genetics of diffuse large B-cell lymphoma: paving a path to personalized medicine. Cancer J. Sudbury Mass 20, 43–47 (2014).

81. Eibel, H., Kraus, H., Sic, H., Kienzler, A.-K. & Rizzi, M. B cell biology: an overview. Curr. Allergy Asthma Rep. 14, 434 (2014).

82. Ghia, P., ten Boekel, E., Rolink, A. G. & Melchers, F. B-cell development: a comparison between mouse and man. Immunol. Today 19, 480–485 (1998).

83. Sagaert, X., Sprangers, B. & De Wolf-Peeters, C. The dynamics of the B follicle: understanding the normal counterpart of B-cell-derived malignancies. Leuk. Off. J. Leuk. Soc. Am. Leuk. Res. Fund UK 21, 1378–1386 (2007).

84. Zandvoort, A. & Timens, W. The dual function of the splenic marginal zone: essential for initiation of anti-TI-2 responses but also vital in the general first-line defense against blood- borne antigens. Clin. Exp. Immunol. 130, 4–11 (2002).

85. MacLennan, I. C. Germinal centers. Annu. Rev. Immunol. 12, 117–139 (1994).

86. Wilson, P. C. et al. Somatic hypermutation introduces insertions and deletions into immunoglobulin V genes. J. Exp. Med. 187, 59–70 (1998).

87. Arakawa, H., Hauschild, J. & Buerstedde, J.-M. Requirement of the activation-induced deaminase (AID) gene for immunoglobulin gene conversion. Science 295, 1301–1306 (2002).

88. Chaudhuri, J. & Alt, F. W. Class-switch recombination: interplay of transcription, DNA deamination and DNA repair. Nat. Rev. Immunol. 4, 541–552 (2004).

89. Stavnezer, J. & Amemiya, C. T. Evolution of isotype switching. Semin. Immunol. 16, 257– 275 (2004).

90. Vinuesa, C. G., Tangye, S. G., Moser, B. & Mackay, C. R. Follicular B helper T cells in antibody responses and autoimmunity. Nat. Rev. Immunol. 5, 853–865 (2005).

91. Shaffer, A. L. et al. XBP1, downstream of Blimp-1, expands the secretory apparatus and other organelles, and increases protein synthesis in plasma cell differentiation. Immunity 21, 81–93 (2004).

92. Wright, G. et al. A gene expression-based method to diagnose clinically distinct subgroups of diffuse large B cell lymphoma. Proc. Natl. Acad. Sci. U. S. A. 100, 9991–9996 (2003).

93. Mandelbaum, J. et al. BLIMP1 is a tumor suppressor gene frequently disrupted in activated B cell-like diffuse large B cell lymphoma. Cancer Cell 18, 568–579 (2010).

183

94. Van Besien, K. & Schouten, H. Follicular lymphoma: a historical overview. Leuk. Lymphoma 48, 232–243 (2007).

95. Bertoni, F. & Ponzoni, M. The cellular origin of mantle cell lymphoma. Int. J. Biochem. Cell Biol. 39, 1747–1753 (2007).

96. Pérez-Galán, P., Dreyling, M. & Wiestner, A. Mantle cell lymphoma: biology, pathogenesis, and the molecular basis of treatment in the genomic era. Blood 117, 26–38 (2011).

97. Ek, S., Högerkorp, C.-M., Dictor, M., Ehinger, M. & Borrebaeck, C. A. K. Mantle cell lymphomas express a distinct genetic signature affecting lymphocyte trafficking and growth regulation as compared with subpopulations of normal human B cells. Cancer Res. 62, 4398–4405 (2002).

98. Walsh, S. H. & Rosenquist, R. Immunoglobulin gene analysis of mature B-cell malignancies: reconsideration of cellular origin and potential antigen involvement in pathogenesis. Med. Oncol. Northwood Lond. Engl. 22, 327–341 (2005).

99. Horsman, D. E. et al. Follicular lymphoma lacking the t(14;18)(q32;q21): identification of two disease subtypes. Br. J. Haematol. 120, 424–433 (2003).

100. Ngan, B. Y., Chen-Levy, Z., Weiss, L. M., Warnke, R. A. & Cleary, M. L. Expression in non-Hodgkin’s lymphoma of the bcl-2 protein associated with the t(14;18) chromosomal translocation. N. Engl. J. Med. 318, 1638–1644 (1988).

101. Bakhshi, A. et al. Cloning the chromosomal breakpoint of t(14;18) human lymphomas: clustering around JH on chromosome 14 and near a transcriptional unit on 18. Cell 41, 899– 906 (1985).

102. Sungalee, S. et al. Germinal center reentries of BCL2-overexpressing B cells drive follicular lymphoma progression. J. Clin. Invest. (2014). doi:10.1172/JCI72415

103. Schüler, F. et al. Prevalence and frequency of circulating t(14;18)-MBR translocation carrying cells in healthy individuals. Int. J. Cancer J. Int. Cancer 124, 958–963 (2009).

104. Oricchio, E. et al. The Eph-receptor A7 is a soluble tumor suppressor for follicular lymphoma. Cell 147, 554–564 (2011).

105. Kiaii, S. et al. Follicular lymphoma cells induce changes in T-cell gene expression and function: potential impact on survival and risk of transformation. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 31, 2654–2661 (2013).

106. Ansell, S. M. Malignant B cells at the helm in follicular lymphoma. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 31, 2641–2642 (2013).

107. Iqbal, J. et al. BCL2 translocation defines a unique tumor subset within the germinal center B-cell-like diffuse large B-cell lymphoma. Am. J. Pathol. 165, 159–166 (2004).

184

108. Cheung, K.-J. J. et al. Acquired TNFRSF14 mutations in follicular lymphoma are associated with worse prognosis. Cancer Res. 70, 9166–9174 (2010).

109. Lin, P. & Medeiros, L. J. The impact of MYC rearrangements and ‘double hit’ abnormalities in diffuse large B-cell lymphoma. Curr. Hematol. Malig. Rep. 8, 243–252 (2013).

110. Young, K. H. et al. Structural profiles of TP53 gene mutations predict clinical outcome in diffuse large B-cell lymphoma: an international collaborative study. Blood 112, 3088–3098 (2008).

111. Lenz, G. et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc. Natl. Acad. Sci. U. S. A. 105, 13520–13525 (2008).

112. Lenz, G. & Staudt, L. M. Aggressive lymphomas. N. Engl. J. Med. 362, 1417–1429 (2010).

113. Lam, L. T. et al. Small molecule inhibitors of IkappaB kinase are selectively toxic for subgroups of diffuse large B-cell lymphoma defined by gene expression profiling. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res. 11, 28–40 (2005).

114. Davis, R. E., Brown, K. D., Siebenlist, U. & Staudt, L. M. Constitutive nuclear factor kappaB activity is required for survival of activated B cell-like diffuse large B cell lymphoma cells. J. Exp. Med. 194, 1861–1874 (2001).

115. Pasqualucci, L. et al. Inactivation of the PRDM1/BLIMP1 gene in diffuse large B cell lymphoma. J. Exp. Med. 203, 311–317 (2006).

116. Tam, W. et al. Mutational analysis of PRDM1 indicates a tumor-suppressor role in diffuse large B-cell lymphomas. Blood 107, 4090–4100 (2006).

117. Zhang, J. et al. The genomic landscape of mantle cell lymphoma is related to the epigenetically determined chromatin state of normal B cells. Blood 123, 2988–2996 (2014).

118. Pasqualucci, L. et al. Inactivating mutations of acetyltransferase genes in B-cell lymphoma. Nature 471, 189–195 (2011).

119. Wagner, S. D., Ahearne, M. & Ko Ferrigno, P. The role of BCL6 in lymphomas and routes to therapy. Br. J. Haematol. 152, 3–12 (2011).

120. Jares, P., Colomer, D. & Campo, E. Molecular pathogenesis of mantle cell lymphoma. J. Clin. Invest. 122, 3416–3423 (2012).

121. Casimiro, M. C., Velasco-Velázquez, M., Aguirre-Alvarado, C. & Pestell, R. G. Overview of cyclins D1 function in cancer and the CDK inhibitor landscape: past and present. Expert Opin. Investig. Drugs 23, 295–304 (2014).

185

122. Lovec, H., Grzeschiczek, A., Kowalski, M. B. & Möröy, T. Cyclin D1/bcl-1 cooperates with myc genes in the generation of B-cell lymphoma in transgenic mice. EMBO J. 13, 3487–3495 (1994).

123. Bodrug, S. E. et al. Cyclin D1 transgene impedes lymphocyte maturation and collaborates in lymphomagenesis with the myc gene. EMBO J. 13, 2124–2130 (1994).

124. Hardiman, G. Microarray platforms--comparisons and contrasts. Pharmacogenomics 5, 487–502 (2004).

125. Zhao, S., Fung-Leung, W.-P., Bittner, A., Ngo, K. & Liu, X. Comparison of RNA-Seq and microarray in transcriptome profiling of activated T cells. PloS One 9, e78644 (2014).

126. Cui, X. & Loraine, A. E. Consistency analysis of redundant probe sets on affymetrix three-prime expression arrays and applications to differential mRNA processing. PloS One 4, e4229 (2009).

127. Liu, H. et al. AffyProbeMiner: a web resource for computing or retrieving accurately redefined Affymetrix probe sets. Bioinforma. Oxf. Engl. 23, 2385–2390 (2007).

128. Nurtdinov, R. N., Vasiliev, M. O., Ershova, A. S., Lossev, I. S. & Karyagina, A. S. PLANdbAffy: probe-level annotation database for Affymetrix expression microarrays. Nucleic Acids Res. 38, D726–730 (2010).

129. Sîrbu, A., Kerr, G., Crane, M. & Ruskin, H. J. RNA-Seq vs dual- and single-channel microarray data: sensitivity analysis for differential expression and clustering. PloS One 7, e50986 (2012).

130. Bottomly, D. et al. Evaluating gene expression in C57BL/6J and DBA/2J mouse striatum using RNA-Seq and microarrays. PloS One 6, e17820 (2011).

131. Marioni, J. C., Mason, C. E., Mane, S. M., Stephens, M. & Gilad, Y. RNA-seq: an assessment of technical reproducibility and comparison with gene expression arrays. Genome Res. 18, 1509–1517 (2008).

132. Mortazavi, A., Williams, B. A., McCue, K., Schaeffer, L. & Wold, B. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat. Methods 5, 621–628 (2008).

133. Oshlack, A. & Wakefield, M. J. Transcript length bias in RNA-seq data confounds systems biology. Biol. Direct 4, 14 (2009).

134. Risso, D., Schwartz, K., Sherlock, G. & Dudoit, S. GC-content normalization for RNA- Seq data. BMC Bioinformatics 12, 480 (2011).

135. Hansen, K. D., Brenner, S. E. & Dudoit, S. Biases in Illumina transcriptome sequencing caused by random hexamer priming. Nucleic Acids Res. 38, e131 (2010).

186

136. Roberts, A., Trapnell, C., Donaghey, J., Rinn, J. L. & Pachter, L. Improving RNA-Seq expression estimates by correcting for fragment bias. Genome Biol. 12, R22 (2011).

137. Zhao, Y., Granas, D. & Stormo, G. D. Inferring binding energies from selected binding sites. PLoS Comput. Biol. 5, e1000590 (2009).

138. Berger, M. F. & Bulyk, M. L. Universal protein-binding microarrays for the comprehensive characterization of the DNA-binding specificities of transcription factors. Nat. Protoc. 4, 393–411 (2009).

139. Rockel, S., Geertz, M. & Maerkl, S. J. MITOMI: a microfluidic platform for in vitro characterization of transcription factor-DNA interaction. Methods Mol. Biol. Clifton NJ 786, 97–114 (2012).

140. Solomon, M. J. & Varshavsky, A. Formaldehyde-mediated DNA-protein crosslinking: a probe for in vivo chromatin structures. Proc. Natl. Acad. Sci. U. S. A. 82, 6470–6474 (1985).

141. Daftari, P., Gavva, N. R. & Shen, C. K. Distinction between AP1 and NF-E2 factor- binding at specific chromatin regions in mammalian cells. Oncogene 18, 5482–5486 (1999).

142. Park, P. J. ChIP-seq: advantages and challenges of a maturing technology. Nat. Rev. Genet. 10, 669–680 (2009).

143. Kirmizis, A. & Farnham, P. J. Genomic approaches that aid in the identification of transcription factor target genes. Exp. Biol. Med. Maywood NJ 229, 705–721 (2004).

144. Zhu, C. et al. High-resolution DNA-binding specificity analysis of yeast transcription factors. Genome Res. 19, 556–566 (2009).

145. Robertson, G. et al. Genome-wide profiles of STAT1 DNA association using chromatin immunoprecipitation and massively parallel sequencing. Nat. Methods 4, 651–657 (2007).

146. Taher, L. & Ovcharenko, I. Variable locus length in the human genome leads to ascertainment bias in functional inference for non-coding elements. Bioinforma. Oxf. Engl. 25, 578–584 (2009).

147. McLean, C. Y. et al. GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol. 28, 495–501 (2010).

148. Johnson, D. S., Mortazavi, A., Myers, R. M. & Wold, B. Genome-wide mapping of in vivo protein-DNA interactions. Science 316, 1497–1502 (2007).

149. Ji, H. et al. An integrated software system for analyzing ChIP-chip and ChIP-seq data. Nat. Biotechnol. 26, 1293–1300 (2008).

150. Sanyal, A., Lajoie, B. R., Jain, G. & Dekker, J. The long-range interaction landscape of gene promoters. Nature 489, 109–113 (2012).

187

151. Jin, F. et al. A high-resolution map of the three-dimensional chromatin interactome in human cells. Nature 503, 290–294 (2013).

152. Whitfield, T. W. et al. Functional analysis of transcription factor binding sites in human promoters. Genome Biol. 13, R50 (2012).

153. Lee, C. H. et al. Regulation of the germinal center gene program by interferon (IFN) regulatory factor 8/IFN consensus sequence-binding protein. J. Exp. Med. 203, 63–72 (2006).

154. Bouamar, H. et al. A capture-sequencing strategy identifies IRF8, EBF1, and APRIL as novel IGH fusion partners in B-cell lymphoma. Blood 122, 726–733 (2013).

155. Bonetti, P. et al. Deregulation of ETS1 and FLI1 contributes to the pathogenesis of diffuse large B-cell lymphoma. Blood 122, 2233–2241 (2013).

156. Harton, J. A. & Ting, J. P. Class II transactivator: mastering the art of major histocompatibility complex expression. Mol. Cell. Biol. 20, 6185–6194 (2000).

157. Cycon, K. A., Mulvaney, K., Rimsza, L. M., Persky, D. & Murphy, S. P. Histone deacetylase inhibitors activate CIITA and MHC class II antigen expression in diffuse large B-cell lymphoma. Immunology 140, 259–272 (2013).

158. Vegliante, M. C. et al. SOX11 regulates PAX5 expression and blocks terminal B-cell differentiation in aggressive mantle cell lymphoma. Blood 121, 2175–2185 (2013).

159. Palomero, J. et al. SOX11 promotes tumor angiogenesis through transcriptional regulation of PDGFA in mantle cell lymphoma. Blood 124, 2235–2247 (2014).

160. Velichutina, I. et al. EZH2-mediated epigenetic silencing in germinal center B cells contributes to proliferation and lymphomagenesis. Blood 116, 5247–5255 (2010).

161. Béguelin, W. et al. EZH2 is required for germinal center formation and somatic EZH2 mutations promote lymphoid transformation. Cancer Cell 23, 677–692 (2013).

162. Meyvantsson, I. & Beebe, D. J. Cell culture models in microfluidic systems. Annu. Rev. Anal. Chem. Palo Alto Calif 1, 423–449 (2008).

163. Pampaloni, F., Reynaud, E. G. & Stelzer, E. H. K. The third dimension bridges the gap between cell culture and live tissue. Nat. Rev. Mol. Cell Biol. 8, 839–845 (2007).

164. Watson, I. R., Takahashi, K., Futreal, P. A. & Chin, L. Emerging patterns of somatic mutations in cancer. Nat. Rev. Genet. 14, 703–718 (2013).

165. Stepanenko, A. A., Vassetzky, Y. S. & Kavsan, V. M. Antagonistic functional duality of cancer genes. Gene 529, 199–207 (2013).

188

166. Yap, D. B. et al. Somatic mutations at EZH2 Y641 act dominantly through a mechanism of selectively altered PRC2 catalytic activity, to increase H3K27 trimethylation. Blood 117, 2451–2459 (2011).

167. Ernst, T. et al. Inactivating mutations of the histone methyltransferase gene EZH2 in myeloid disorders. Nat. Genet. 42, 722–726 (2010).

168. Chittaranjan, S. et al. Mutations in CIC and IDH1 cooperatively regulate 2- hydroxyglutarate levels and cell clonogenicity. Oncotarget 5, 7960–7979 (2014).

169. Lin, Y.-C. et al. Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations. Nat. Commun. 5, 4767 (2014).

170. Graham, F. L., Smiley, J., Russell, W. C. & Nairn, R. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J. Gen. Virol. 36, 59–74 (1977).

171. Shaw, G., Morse, S., Ararat, M. & Graham, F. L. Preferential transformation of human neuronal cells by human adenoviruses and the origin of HEK 293 cells. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol. 16, 869–871 (2002).

172. Potthoff, M. J. & Olson, E. N. MEF2: a central regulator of diverse developmental programs. Dev. Camb. Engl. 134, 4131–4140 (2007).

173. Wu, W., de Folter, S., Shen, X., Zhang, W. & Tao, S. Vertebrate paralogous MEF2 genes: origin, conservation, and evolution. PloS One 6, e17334 (2011).

174. Gossett, L. A., Kelvin, D. J., Sternberg, E. A. & Olson, E. N. A new myocyte-specific enhancer-binding factor that recognizes a conserved element associated with multiple muscle-specific genes. Mol. Cell. Biol. 9, 5022–5033 (1989).

175. Wang, D. Z., Valdez, M. R., McAnally, J., Richardson, J. & Olson, E. N. The Mef2c gene is a direct transcriptional target of myogenic bHLH and MEF2 proteins during skeletal muscle development. Dev. Camb. Engl. 128, 4623–4633 (2001).

176. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. Cooperative activation of muscle gene expression by MEF2 and myogenic bHLH proteins. Cell 83, 1125–1136 (1995).

177. Estrella, N. L. et al. MEF2 transcription factors regulate distinct gene programs in mammalian skeletal muscle differentiation. J. Biol. Chem. 290, 1256–1268 (2015).

178. Lin, Q. et al. Requirement of the MADS-box transcription factor MEF2C for vascular development. Dev. Camb. Engl. 125, 4565–4574 (1998).

179. Karamboulas, C. et al. Disruption of MEF2 activity in cardiomyoblasts inhibits cardiomyogenesis. J. Cell Sci. 119, 4315–4321 (2006).

180. Lin, Q., Schwarz, J., Bucana, C. & Olson, E. N. Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science 276, 1404–1407 (1997).

189

181. Naya, F. J. et al. Mitochondrial deficiency and cardiac sudden death in mice lacking the MEF2A transcription factor. Nat. Med. 8, 1303–1309 (2002).

182. Arnold, M. A. et al. MEF2C transcription factor controls chondrocyte hypertrophy and bone development. Dev. Cell 12, 377–389 (2007).

183. Potthoff, M. J. et al. Regulation of skeletal muscle sarcomere integrity and postnatal muscle function by Mef2c. Mol. Cell. Biol. 27, 8143–8151 (2007).

184. Shalizi, A. et al. A calcium-regulated MEF2 sumoylation switch controls postsynaptic differentiation. Science 311, 1012–1017 (2006).

185. Verzi, M. P. et al. The transcription factor MEF2C is required for craniofacial development. Dev. Cell 12, 645–652 (2007).

186. Knecht, A. K. & Bronner-Fraser, M. Induction of the neural crest: a multigene process. Nat. Rev. Genet. 3, 453–461 (2002).

187. Yu, W. et al. MEF2 transcription factors promotes EMT and invasiveness of hepatocellular carcinoma through TGF-β1 autoregulation circuitry. Tumour Biol. J. Int. Soc. Oncodevelopmental Biol. Med. 35, 10943–10951 (2014).

188. Hayashi, M. et al. Targeted deletion of BMK1/ERK5 in adult mice perturbs vascular integrity and leads to endothelial failure. J. Clin. Invest. 113, 1138–1148 (2004).

189. Mao, Z., Bonni, A., Xia, F., Nadal-Vicens, M. & Greenberg, M. E. Neuronal activity- dependent cell survival mediated by transcription factor MEF2. Science 286, 785–790 (1999).

190. Okamoto, S., Krainc, D., Sherman, K. & Lipton, S. A. Antiapoptotic role of the p38 mitogen-activated protein kinase-myocyte enhancer factor 2 transcription factor pathway during neuronal differentiation. Proc. Natl. Acad. Sci. U. S. A. 97, 7561–7566 (2000).

191. Dequiedt, F. et al. HDAC7, a thymus-specific class II histone deacetylase, regulates Nur77 transcription and TCR-mediated apoptosis. Immunity 18, 687–698 (2003).

192. Youn, H. D. & Liu, J. O. Cabin1 represses MEF2-dependent Nur77 expression and T cell apoptosis by controlling association of histone deacetylases and acetylases with MEF2. Immunity 13, 85–94 (2000).

193. Youn, H. D., Sun, L., Prywes, R. & Liu, J. O. Apoptosis of T cells mediated by Ca2+- induced release of the transcription factor MEF2. Science 286, 790–793 (1999).

194. Andrés, V., Cervera, M. & Mahdavi, V. Determination of the consensus binding site for MEF2 expressed in muscle and brain reveals tissue-specific sequence constraints. J. Biol. Chem. 270, 23246–23249 (1995).

190

195. Molkentin, J. D. et al. MEF2B is a potent transactivator expressed in early myogenic lineages. Mol. Cell. Biol. 16, 3814–3824 (1996).

196. Firulli, A. B. et al. Myocyte enhancer binding factor-2 expression and activity in vascular smooth muscle cells. Association with the activated phenotype. Circ. Res. 78, 196–204 (1996).

197. Swanson, B. J., Jäck, H. M. & Lyons, G. E. Characterization of myocyte enhancer factor 2 (MEF2) expression in B and T cells: MEF2C is a B cell-restricted transcription factor in lymphocytes. Mol. Immunol. 35, 445–458 (1998).

198. Black, B. L. & Olson, E. N. Transcriptional control of muscle development by myocyte enhancer factor-2 (MEF2) proteins. Annu. Rev. Cell Dev. Biol. 14, 167–196 (1998).

199. Lin, Q., Schwarz, J., Bucana, C. & Olson, E. N. Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science 276, 1404–1407 (1997).

200. Schlesinger, J. et al. The cardiac transcription network modulated by Gata4, Mef2a, Nkx2.5, Srf, histone modifications, and microRNAs. PLoS Genet. 7, e1001313 (2011).

201. Flavell, S. W. et al. Genome-wide analysis of MEF2 transcriptional program reveals synaptic target genes and neuronal activity-dependent polyadenylation site selection. Neuron 60, 1022–1038 (2008).

202. Johnson, M. E. et al. A ChIP-seq-defined genome-wide map of MEF2C binding reveals inflammatory pathways associated with its role in bone density determination. Calcif. Tissue Int. 94, 396–402 (2014).

203. Estrella, N. L. et al. MEF2 Transcription Factors Regulate Distinct Gene Programs in Mammalian Skeletal Muscle Differentiation. J. Biol. Chem. (2014). doi:10.1074/jbc.M114.589838

204. Katoh, Y., Molkentin, J. D., Dave, V., Olson, E. N. & Periasamy, M. MEF2B is a component of a smooth muscle-specific complex that binds an A/T-rich element important for smooth muscle myosin heavy chain gene expression. J. Biol. Chem. 273, 1511–1518 (1998).

205. Murata, T. et al. Contribution of myocyte enhancer factor 2 family transcription factors to BZLF1 expression in Epstein-Barr virus reactivation from latency. J. Virol. 87, 10148– 10162 (2013).

206. Yu, L. et al. Sclerostin expression is induced by BMPs in human Saos-2 osteosarcoma cells but not via direct effects on the sclerostin gene promoter or ECR5 element. Bone 49, 1131–1140 (2011).

207. He, A., Kong, S. W., Ma, Q. & Pu, W. T. Co-occupancy by multiple cardiac transcription factors identifies transcriptional enhancers active in heart. Proc. Natl. Acad. Sci. U. S. A. 108, 5632–5637 (2011).

191

208. Blaeser, F., Ho, N., Prywes, R. & Chatila, T. A. Ca(2+)-dependent gene expression mediated by MEF2 transcription factors. J. Biol. Chem. 275, 197–209 (2000).

209. Rao, S., Karray, S., Gackstetter, E. R. & Koshland, M. E. Myocyte enhancer factor- related B-MEF2 is developmentally expressed in B cells and regulates the immunoglobulin J chain promoter. J. Biol. Chem. 273, 26123–26129 (1998).

210. Shore, P. & Sharrocks, A. D. The MADS-box family of transcription factors. Eur. J. Biochem. FEBS 229, 1–13 (1995).

211. Olson, E. N., Perry, M. & Schulz, R. A. Regulation of muscle differentiation by the MEF2 family of MADS box transcription factors. Dev. Biol. 172, 2–14 (1995).

212. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. Mutational analysis of the DNA binding, dimerization, and transcriptional activation domains of MEF2C. Mol. Cell. Biol. 16, 2627–2636 (1996).

213. Zhao, M. et al. Regulation of the MEF2 family of transcription factors by p38. Mol. Cell. Biol. 19, 21–30 (1999).

214. Octobre, G., Lemercier, C., Khochbin, S., Robert-Nicoud, M. & Souchier, C. Monitoring the interaction between DNA and a transcription factor (MEF2A) using fluorescence correlation spectroscopy. C. R. Biol. 328, 1033–1040 (2005).

215. Meierhans, D. & Allemann, R. K. The N-terminal methionine is a major determinant of the DNA binding specificity of MEF-2C. J. Biol. Chem. 273, 26052–26060 (1998).

216. Meierhans, D., Sieber, M. & Allemann, R. K. High affinity binding of MEF-2C correlates with DNA bending. Nucleic Acids Res. 25, 4537–4544 (1997).

217. Molkentin, J. D., Li, L. & Olson, E. N. Phosphorylation of the MADS-Box transcription factor MEF2C enhances its DNA binding activity. J. Biol. Chem. 271, 17199–17204 (1996).

218. Angelelli, C. et al. Differentiation-dependent lysine 4 acetylation enhances MEF2C binding to DNA in skeletal muscle cells. Nucleic Acids Res. 36, 915–928 (2008).

219. Han, A., He, J., Wu, Y., Liu, J. O. & Chen, L. Mechanism of recruitment of class II histone deacetylases by myocyte enhancer factor-2. J. Mol. Biol. 345, 91–102 (2005).

220. Miska, E. A. et al. HDAC4 deacetylase associates with and represses the MEF2 transcription factor. EMBO J. 18, 5099–5107 (1999).

221. Lu, J., McKinsey, T. A., Zhang, C. L. & Olson, E. N. Regulation of skeletal myogenesis by association of the MEF2 transcription factor with class II histone deacetylases. Mol. Cell 6, 233–244 (2000).

192

222. Youn, H. D., Grozinger, C. M. & Liu, J. O. Calcium regulates transcriptional repression of myocyte enhancer factor 2 by histone deacetylase 4. J. Biol. Chem. 275, 22563–22567 (2000).

223. Lahm, A. et al. Unraveling the hidden catalytic activity of vertebrate class IIa histone deacetylases. Proc. Natl. Acad. Sci. U. S. A. 104, 17335–17340 (2007).

224. Fischle, W. et al. Enzymatic activity associated with class II HDACs is dependent on a multiprotein complex containing HDAC3 and SMRT/N-CoR. Mol. Cell 9, 45–57 (2002).

225. Zhang, C. L., McKinsey, T. A. & Olson, E. N. Association of class II histone deacetylases with heterochromatin protein 1: potential role for histone methylation in control of muscle differentiation. Mol. Cell. Biol. 22, 7302–7312 (2002).

226. Zhang, C. L., McKinsey, T. A., Lu, J. R. & Olson, E. N. Association of COOH-terminal- binding protein (CtBP) and MEF2-interacting transcription repressor (MITR) contributes to transcriptional repression of the MEF2 transcription factor. J. Biol. Chem. 276, 35–39 (2001).

227. Han, A. et al. Sequence-specific recruitment of transcriptional co-repressor Cabin1 by myocyte enhancer factor-2. Nature 422, 730–734 (2003).

228. Jang, H., Choi, D.-E., Kim, H., Cho, E.-J. & Youn, H.-D. Cabin1 represses MEF2 transcriptional activity by association with a methyltransferase, SUV39H1. J. Biol. Chem. 282, 11172–11179 (2007).

229. He, J. et al. Structure of p300 bound to MEF2 on DNA reveals a mechanism of enhanceosome assembly. Nucleic Acids Res. 39, 4464–4474 (2011).

230. Sartorelli, V., Huang, J., Hamamori, Y. & Kedes, L. Molecular mechanisms of myogenic coactivation by p300: direct interaction with the activation domain of MyoD and with the MADS box of MEF2C. Mol. Cell. Biol. 17, 1010–1026 (1997).

231. Slepak, T. I. et al. Control of cardiac-specific transcription by p300 through myocyte enhancer factor-2D. J. Biol. Chem. 276, 7575–7585 (2001).

232. Lundblad, J. R., Kwok, R. P., Laurance, M. E., Harter, M. L. & Goodman, R. H. Adenoviral E1A-associated protein p300 as a functional homologue of the transcriptional co- activator CBP. Nature 374, 85–88 (1995).

233. Zhao, X., Sternsdorf, T., Bolger, T. A., Evans, R. M. & Yao, T.-P. Regulation of MEF2 by histone deacetylase 4- and SIRT1 deacetylase-mediated lysine modifications. Mol. Cell. Biol. 25, 8456–8464 (2005).

234. Ma, K., Chan, J. K. L., Zhu, G. & Wu, Z. Myocyte enhancer factor 2 acetylation by p300 enhances its DNA binding activity, transcriptional activity, and myogenic differentiation. Mol. Cell. Biol. 25, 3575–3582 (2005).

193

235. Cohen, I., Poręba, E., Kamieniarz, K. & Schneider, R. Histone modifiers in cancer: friends or foes? Genes Cancer 2, 631–647 (2011).

236. McKinsey, T. A., Zhang, C. L. & Olson, E. N. Activation of the myocyte enhancer factor-2 transcription factor by calcium/calmodulin-dependent protein kinase-stimulated binding of 14-3-3 to histone deacetylase 5. Proc. Natl. Acad. Sci. U. S. A. 97, 14400–14405 (2000).

237. Pan, F., Means, A. R. & Liu, J. O. Calmodulin-dependent protein kinase IV regulates nuclear export of Cabin1 during T-cell activation. EMBO J. 24, 2104–2113 (2005).

238. Lu, J., McKinsey, T. A., Nicol, R. L. & Olson, E. N. Signal-dependent activation of the MEF2 transcription factor by dissociation from histone deacetylases. Proc. Natl. Acad. Sci. U. S. A. 97, 4070–4075 (2000).

239. Woronicz, J. D. et al. Regulation of the Nur77 orphan steroid receptor in activation- induced apoptosis. Mol. Cell. Biol. 15, 6364–6376 (1995).

240. Wilker, P. R. et al. Transcription factor Mef2c is required for B cell proliferation and survival after antigen receptor stimulation. Nat. Immunol. 9, 603–612 (2008).

241. Avalos, A. M., Meyer-Wentrup, F. & Ploegh, H. L. B-cell receptor signaling in lymphoid malignancies and autoimmunity. Adv. Immunol. 123, 1–49 (2014).

242. Rampalli, S. et al. p38 MAPK signaling regulates recruitment of Ash2L-containing methyltransferase complexes to specific genes during differentiation. Nat. Struct. Mol. Biol. 14, 1150–1156 (2007).

243. Morin, S., Charron, F., Robitaille, L. & Nemer, M. GATA-dependent recruitment of MEF2 proteins to target promoters. EMBO J. 19, 2046–2055 (2000).

244. Taube, R., Lin, X., Irwin, D., Fujinaga, K. & Peterlin, B. M. Interaction between P-TEFb and the C-terminal domain of RNA polymerase II activates transcriptional elongation from sites upstream or downstream of target genes. Mol. Cell. Biol. 22, 321–331 (2002).

245. Nojima, M., Huang, Y., Tyagi, M., Kao, H.-Y. & Fujinaga, K. The positive transcription elongation factor b is an essential cofactor for the activation of transcription by myocyte enhancer factor 2. J. Mol. Biol. 382, 275–287 (2008).

246. Cox, D. M. et al. Phosphorylation motifs regulating the stability and function of myocyte enhancer factor 2A. J. Biol. Chem. 278, 15297–15303 (2003).

247. Kato, Y. et al. Big mitogen-activated kinase regulates multiple members of the MEF2 protein family. J. Biol. Chem. 275, 18534–18540 (2000).

248. Du, M. et al. Protein kinase A represses skeletal myogenesis by targeting myocyte enhancer factor 2D. Mol. Cell. Biol. 28, 2952–2970 (2008).

194

249. Grégoire, S. et al. Control of MEF2 transcriptional activity by coordinated phosphorylation and sumoylation. J. Biol. Chem. 281, 4423–4433 (2006).

250. Zhang, M., Zhu, B. & Davie, J. Alternative Splicing of MEF2C Controls its Activity in Normal Myogenesis and Promotes Tumorigenicity in Rhabdomyosarcoma Cells. J. Biol. Chem. (2014). doi:10.1074/jbc.M114.606277

251. Zhu, B., Ramachandran, B. & Gulick, T. Alternative pre-mRNA splicing governs expression of a conserved acidic transactivation domain in myocyte enhancer factor 2 factors of striated muscle and brain. J. Biol. Chem. 280, 28749–28760 (2005).

252. Ramachandran, B., Yu, G., Li, S., Zhu, B. & Gulick, T. Myocyte enhancer factor 2A is transcriptionally autoregulated. J. Biol. Chem. 283, 10318–10329 (2008).

253. Friedrich, F. W. et al. A novel genetic variant in the transcription factor Islet-1 exerts gain of function on myocyte enhancer factor 2C promoter activity. Eur. J. Heart Fail. 15, 267–276 (2013).

254. Homminga, I. et al. Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potential oncogenes in T cell acute lymphoblastic leukemia. Cancer Cell 19, 484–497 (2011).

255. Paciorkowski, A. R. et al. MEF2C Haploinsufficiency features consistent hyperkinesis, variable epilepsy, and has a role in dorsal and ventral neuronal developmental pathways. Neurogenetics 14, 99–111 (2013).

256. Zweier, M. et al. Mutations in MEF2C from the 5q14.3q15 microdeletion syndrome region are a frequent cause of severe mental retardation and diminish MECP2 and CDKL5 expression. Hum. Mutat. 31, 722–733 (2010).

257. Ryan, S. D. et al. Isogenic human iPSC Parkinson’s model shows nitrosative stress- induced dysfunction in MEF2-PGC1α transcription. Cell 155, 1351–1364 (2013).

258. Di Giorgio, E. et al. MEF2 is a converging hub for histone deacetylase 4 and phosphatidylinositol 3-kinase/Akt-induced transformation. Mol. Cell. Biol. 33, 4473–4491 (2013).

259. Wang, L., Fan, C., Topol, S. E., Topol, E. J. & Wang, Q. Mutation of MEF2A in an inherited disorder with features of coronary artery disease. Science 302, 1578–1581 (2003).

260. Firth, H. V. et al. DECIPHER: Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources. Am. J. Hum. Genet. 84, 524–533 (2009).

261. Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2, 401–404 (2012).

262. Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 (2013).

195

263. Cancer Genome Atlas Research Network et al. Integrated genomic characterization of endometrial carcinoma. Nature 497, 67–73 (2013).

264. Vlieghe, D. et al. A new generation of JASPAR, the open-access repository for transcription factor binding site profiles. Nucleic Acids Res. 34, D95–97 (2006).

265. Desmet, F.-O. et al. Human Splicing Finder: an online bioinformatics tool to predict splicing signals. Nucleic Acids Res. 37, e67 (2009).

266. Dasmahapatra, G. et al. Obatoclax interacts synergistically with the irreversible proteasome inhibitor carfilzomib in GC- and ABC-DLBCL cells in vitro and in vivo. Mol. Cancer Ther. 11, 1122–1132 (2012).

267. Irizarry, R. A. et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostat. Oxf. Engl. 4, 249–264 (2003).

268. Qu, Y., He, F. & Chen, Y. Different effects of the probe summarization algorithms PLIER and RMA on high-level analysis of Affymetrix exon arrays. BMC Bioinformatics 11, 211 (2010).

269. Smyth, G. K. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. 3, Article3 (2004).

270. Jeanmougin, M. et al. Should we abandon the t-test in the analysis of gene expression microarray data: a comparison of variance modeling strategies. PloS One 5, e12336 (2010).

271. Wei, R., Stewart, E. A. & Amoaku, W. M. Suitability of endogenous reference genes for gene expression studies with human intraocular endothelial cells. BMC Res. Notes 6, 46 (2013).

272. Cao, X. et al. Critical selection of internal control genes for quantitative real-time RT- PCR studies in lipopolysaccharide-stimulated human THP-1 and K562 cells. Biochem. Biophys. Res. Commun. 427, 366–372 (2012).

273. Diamanti, D. et al. Reference genes selection for transcriptional profiling in blood of HD patients and R6/2 mice. J. Huntingt. Dis. 2, 185–200 (2013).

274. Anders, S. & Huber, W. Differential expression analysis for sequence count data. Genome Biol. 11, R106 (2010).

275. Zhang, Z. H. et al. A Comparative Study of Techniques for Differential Expression Analysis on RNA-Seq Data. PloS One 9, e103207 (2014).

276. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinforma. Oxf. Engl. 25, 1754–1760 (2009).

277. Landt, S. G. et al. ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res. 22, 1813–1831 (2012).

196

278. Fejes, A. P. et al. FindPeaks 3.1: a tool for identifying areas of enrichment from massively parallel short-read sequencing technology. Bioinforma. Oxf. Engl. 24, 1729–1730 (2008).

279. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008).

280. Mundade, R., Ozer, H. G., Wei, H., Prabhu, L. & Lu, T. Role of ChIP-seq in the discovery of transcription factor binding sites, differential gene regulation mechanism, epigenetic marks and beyond. Cell Cycle Georget. Tex 13, 2847–2852 (2014).

281. Jung, Y. L. et al. Impact of sequencing depth in ChIP-seq experiments. Nucleic Acids Res. 42, e74 (2014).

282. Tsankov, A. M. et al. Transcription factor binding dynamics during human ES cell differentiation. Nature 518, 344–349 (2015).

283. Chen, T.-W. et al. ChIPseek, a web-based analysis tool for ChIP data. BMC Genomics 15, 539 (2014).

284. Bailey, T. L. & Machanick, P. Inferring direct DNA binding from ChIP-seq. Nucleic Acids Res. 40, e128 (2012).

285. Ding, J., Cai, X., Wang, Y., Hu, H. & Li, X. ChIPModule: systematic discovery of transcription factors and their cofactors from ChIP-seq data. Pac. Symp. Biocomput. Pac. Symp. Biocomput. 320–331 (2013).

286. Wang, S. et al. Target analysis by integration of transcriptome and ChIP-seq data with BETA. Nat. Protoc. 8, 2502–2515 (2013).

287. Breitling, R., Armengaud, P., Amtmann, A. & Herzyk, P. Rank products: a simple, yet powerful, new method to detect differentially regulated genes in replicated microarray experiments. FEBS Lett. 573, 83–92 (2004).

288. Huang, D. W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57 (2009).

289. Huang, D. W., Sherman, B. T. & Lempicki, R. A. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 37, 1–13 (2009).

290. Yip, S. et al. Concurrent CIC mutations, IDH mutations, and 1p/19q loss distinguish oligodendrogliomas from other cancers. J. Pathol. 226, 7–16 (2012).

291. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102, 15545– 15550 (2005).

197

292. Mootha, V. K. et al. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. 34, 267–273 (2003).

293. Basso, K. et al. Integrated biochemical and computational approach identifies BCL6 direct target genes controlling multiple pathways in normal germinal center B cells. Blood 115, 975–984 (2010).

294. Polo, J. M. et al. Transcriptional signature with differential expression of BCL6 target genes accurately identifies BCL6-dependent diffuse large B cell lymphomas. Proc. Natl. Acad. Sci. U. S. A. 104, 3207–3212 (2007).

295. Brewster, R. C. et al. The transcription factor titration effect dictates level of gene expression. Cell 156, 1312–1323 (2014).

296. Raghavachari, N. et al. A systematic comparison and evaluation of high density exon arrays and RNA-seq technology used to unravel the peripheral blood transcriptome of sickle cell disease. BMC Med. Genomics 5, 28 (2012).

297. Guo, Y. et al. Large scale comparison of gene expression levels by microarrays and RNAseq using TCGA data. PloS One 8, e71462 (2013).

298. Larkin, J. E., Frank, B. C., Gavras, H., Sultana, R. & Quackenbush, J. Independence and reproducibility across microarray platforms. Nat. Methods 2, 337–344 (2005).

299. Davis, R. E. et al. Chronic active B-cell-receptor signalling in diffuse large B-cell lymphoma. Nature 463, 88–92 (2010).

300. Ott, G., Rosenwald, A. & Campo, E. Understanding MYC-driven aggressive B-cell lymphomas: pathogenesis and classification. Hematol. Educ. Program Am. Soc. Hematol. Am. Soc. Hematol. Educ. Program 2013, 575–583 (2013).

301. Sun, J. et al. Metastasis suppressor, NDRG1, mediates its activity through signaling pathways and molecular motors. Carcinogenesis 34, 1943–1954 (2013).

302. Haberland, M. et al. Regulation of HDAC9 gene expression by MEF2 establishes a negative-feedback loop in the transcriptional circuitry of muscle differentiation. Mol. Cell. Biol. 27, 518–525 (2007).

303. Gröger, C. J., Grubinger, M., Waldhör, T., Vierlinger, K. & Mikulits, W. Meta-analysis of gene expression signatures defining the epithelial to mesenchymal transition during cancer progression. PloS One 7, e51136 (2012).

304. Zeisberg, M. & Neilson, E. G. Biomarkers for epithelial-mesenchymal transitions. J. Clin. Invest. 119, 1429–1437 (2009).

305. Li, Q., Brown, J., Huang, H. & Bickel, P. Measuring reproducibility of high-throughput experiments. Ann. Appl. Stat. 5, 1752–1779 (2011).

198

306. Machanick, P. & Bailey, T. L. MEME-ChIP: motif analysis of large DNA datasets. Bioinforma. Oxf. Engl. 27, 1696–1697 (2011).

307. Wang, J. et al. Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Res. 22, 1798–1812 (2012).

308. Sakabe, N. J. et al. Dual transcriptional activator and repressor roles of TBX20 regulate adult cardiac structure and function. Hum. Mol. Genet. 21, 2194–2204 (2012).

309. Peng, Y. & Jahroudi, N. The NFY transcription factor functions as a repressor and activator of the von Willebrand factor promoter. Blood 99, 2408–2417 (2002).

310. Huang, W. & Bateman, E. Transcription of the Acanthamoeba TATA-binding protein gene. A single transcription factor acts both as an activator and a repressor. J. Biol. Chem. 272, 3852–3859 (1997).

311. Speksnijder, N., Christensen, K. V., Didriksen, M., De Kloet, E. R. & Datson, N. A. Glucocorticoid receptor and myocyte enhancer factor 2 cooperate to regulate the expression of c-JUN in a neuronal context. J. Mol. Neurosci. MN 48, 209–218 (2012).

312. Gordon, J. W. et al. Protein kinase A-regulated assembly of a MEF2{middle dot}HDAC4 repressor complex controls c-Jun expression in vascular smooth muscle cells. J. Biol. Chem. 284, 19027–19042 (2009).

313. Henry, R. A., Kuo, Y.-M. & Andrews, A. J. Differences in specificity and selectivity between CBP and p300 acetylation of histone H3 and H3/H4. Biochemistry (Mosc.) 52, 5746–5759 (2013).

314. Pasini, D. et al. Characterization of an antagonistic switch between histone H3 lysine 27 methylation and acetylation in the transcriptional regulation of Polycomb group target genes. Nucleic Acids Res. 38, 4958–4969 (2010).

315. Aziz, A., Liu, Q.-C. & Dilworth, F. J. Regulating a master regulator: establishing tissue- specific gene expression in skeletal muscle. Epigenetics Off. J. DNA Methylation Soc. 5, 691–695 (2010).

316. Hu, D. et al. The MLL3/MLL4 branches of the COMPASS family function as major histone H3K4 monomethylases at enhancers. Mol. Cell. Biol. 33, 4745–4754 (2013).

317. Wang, Z. et al. Combinatorial patterns of histone acetylations and methylations in the human genome. Nat. Genet. 40, 897–903 (2008).

318. Britschgi, A. et al. Calcium-activated chloride channel ANO1 promotes breast cancer progression by activating EGFR and CAMK signaling. Proc. Natl. Acad. Sci. U. S. A. 110, E1026–1034 (2013).

319. Ruiz, C. et al. Enhanced expression of ANO1 in head and neck squamous cell carcinoma causes cell migration and correlates with poor prognosis. PloS One 7, e43265 (2012).

199

320. Corcione, A. et al. Chemotaxis of human tonsil B lymphocytes to CC chemokine receptor (CCR) 1, CCR2 and CCR4 ligands is restricted to non-germinal center cells. Int. Immunol. 14, 883–892 (2002).

321. Wain, J. H., Kirby, J. A. & Ali, S. Leucocyte chemotaxis: Examination of mitogen- activated protein kinase and phosphoinositide 3-kinase activation by Monocyte Chemoattractant Proteins-1, -2, -3 and -4. Clin. Exp. Immunol. 127, 436–444 (2002).

322. Kim, J. et al. Wnt5a is secreted by follicular dendritic cells to protect germinal center B cells via Wnt/Ca2+/NFAT/NF-κB-B cell lymphoma 6 signaling. J. Immunol. Baltim. Md 1950 188, 182–189 (2012).

323. Sun, Q., Sattayakhom, A., Backs, J., Stremmel, W. & Chamulitrat, W. Role of myocyte enhancing factor 2B in epithelial myofibroblast transition of human gingival keratinocytes. Exp. Biol. Med. Maywood NJ 237, 178–185 (2012).

324. Andrews, S. F. et al. Developmentally regulated expression of MEF2C limits the response to BCR engagement in transitional B cells. Eur. J. Immunol. 42, 1327–1336 (2012).

325. Zhuang, Q. et al. Class IIa histone deacetylases and myocyte enhancer factor 2 proteins regulate the mesenchymal-to-epithelial transition of somatic cell reprogramming. J. Biol. Chem. 288, 12022–12031 (2013).

326. Chen, T., Hwang, H., Rose, M. E., Nines, R. G. & Stoner, G. D. Chemopreventive properties of black raspberries in N-nitrosomethylbenzylamine-induced rat esophageal tumorigenesis: down-regulation of cyclooxygenase-2, inducible nitric oxide synthase, and c- Jun. Cancer Res. 66, 2853–2859 (2006).

327. Ogunwobi, O., Mutungi, G. & Beales, I. L. P. Leptin stimulates proliferation and inhibits apoptosis in Barrett’s esophageal adenocarcinoma cells by cyclooxygenase-2-dependent, prostaglandin-E2-mediated transactivation of the epidermal growth factor receptor and c-Jun NH2-terminal kinase activation. Endocrinology 147, 4505–4516 (2006).

328. Eckhoff, K. et al. The prognostic significance of Jun transcription factors in ovarian cancer. J. Cancer Res. Clin. Oncol. 139, 1673–1680 (2013).

329. Neyns, B., Teugels, E., Bourgain, C., Birrerand, M. & De Grève, J. Alteration of jun proto-oncogene status by plasmid transfection affects growth of human ovarian cancer cells. Int. J. Cancer J. Int. Cancer 82, 687–693 (1999).

330. Lafont, J. et al. The expression of novH in adrenocortical cells is down-regulated by TGFbeta 1 through c-Jun in a Smad-independent manner. J. Biol. Chem. 277, 41220–41229 (2002).

331. Shiozawa, T. et al. Estrogen-induced proliferation of normal endometrial glandular cells is initiated by transcriptional activation of cyclin D1 via binding of c-Jun to an AP-1 sequence. Oncogene 23, 8603–8610 (2004).

200

332. Zavadil, J. & Böttinger, E. P. TGF-beta and epithelial-to-mesenchymal transitions. Oncogene 24, 5764–5774 (2005).

333. Huang, M. & Prendergast, G. C. RhoB in cancer suppression. Histol. Histopathol. 21, 213–218 (2006).

334. Rai, D., Kim, S.-W., McKeller, M. R., Dahia, P. L. M. & Aguiar, R. C. T. Targeting of SMAD5 links microRNA-155 to the TGF-beta pathway and lymphomagenesis. Proc. Natl. Acad. Sci. U. S. A. 107, 3111–3116 (2010).

335. Jiang, D. & Aguiar, R. C. T. MicroRNA-155 controls RB phosphorylation in normal and malignant B lymphocytes via the noncanonical TGF-β1/SMAD5 signaling module. Blood 123, 86–93 (2014).

336. Ogama, Y. et al. Prevalent hyper-methylation of the CDH13 gene promoter in malignant B cell lymphomas. Int. J. Oncol. 25, 685–691 (2004).

337. Nishiu, M. et al. Microarray analysis of gene-expression profiles in diffuse large B-cell lymphoma: identification of genes related to disease progression. Jpn. J. Cancer Res. Gann 93, 894–901 (2002).

338. McKinsey, T. A., Zhang, C. L., Lu, J. & Olson, E. N. Signal-dependent nuclear export of a histone deacetylase regulates muscle differentiation. Nature 408, 106–111 (2000).

339. Mao, Z. & Wiedmann, M. Calcineurin enhances MEF2 DNA binding activity in calcium- dependent survival of cerebellar granule neurons. J. Biol. Chem. 274, 31102–31107 (1999).

340. Kent, W. J. et al. The human genome browser at UCSC. Genome Res. 12, 996–1006 (2002).

341. Fu, W., Wei, J. & Gu, J. MEF2C mediates the activation induced cell death (AICD) of macrophages. Cell Res. 16, 559–565 (2006).

342. Kim, W. et al. Systematic and quantitative assessment of the ubiquitin-modified proteome. Mol. Cell 44, 325–340 (2011).

343. She, H., Yang, Q. & Mao, Z. Neurotoxin-induced selective ubiquitination and regulation of MEF2A isoform in neuronal stress response. J. Neurochem. 122, 1203–1210 (2012).

344. Allen, C. D. C. et al. Germinal center dark and light zone organization is mediated by CXCR4 and CXCR5. Nat. Immunol. 5, 943–952 (2004).

345. Green, J. A. et al. The sphingosine 1-phosphate receptor S1P₂ maintains the homeostasis of germinal center B cells and promotes niche confinement. Nat. Immunol. 12, 672–680 (2011).

346. Tsubakimoto, K. et al. Small GTPase RhoD suppresses cell migration and cytokinesis. Oncogene 18, 2431–2440 (1999).

201

347. Kozasa, T. et al. p115 RhoGEF, a GTPase activating protein for Galpha12 and Galpha13. Science 280, 2109–2111 (1998).

348. Patel, M. et al. Gα13/PDZ-RhoGEF/RhoA signaling is essential for gastrin-releasing peptide receptor-mediated colon cancer cell migration. Mol. Pharmacol. 86, 252–262 (2014).

349. Lynn, D. J. et al. InnateDB: facilitating systems-level analyses of the mammalian innate immune response. Mol. Syst. Biol. 4, 218 (2008).

350. Liekens, A. M. L. et al. BioGraph: unsupervised biomedical knowledge discovery via automated hypothesis generation. Genome Biol. 12, R57 (2011).

351. Schwieger, M. et al. Homing and invasiveness of MLL/ENL leukemic cells is regulated by MEF2C. Blood 114, 2476–2488 (2009).

352. Muppidi, J. R. et al. Loss of signalling via Gα13 in germinal centre B-cell-derived lymphoma. Nature (2014). doi:10.1038/nature13765

353. Morin, R. D. et al. Mutational and structural analysis of diffuse large B-cell lymphoma using whole-genome sequencing. Blood 122, 1256–1265 (2013).

354. Kaiser, C. et al. The proto-oncogene c-myc is a direct target gene of Epstein-Barr virus nuclear antigen 2. J. Virol. 73, 4481–4484 (1999).

355. Hah, N. et al. A rapid, extensive, and transient transcriptional response to estrogen signaling in breast cancer cells. Cell 145, 622–634 (2011).

356. Della Gatta, G. et al. Direct targets of the TRP63 transcription factor revealed by a combination of gene expression profiling and reverse engineering. Genome Res. 18, 939–948 (2008).

357. Van Arensbergen, J., van Steensel, B. & Bussemaker, H. J. In search of the determinants of enhancer-promoter interaction specificity. Trends Cell Biol. (2014). doi:10.1016/j.tcb.2014.07.004

358. Lin, F., Wang, N. & Zhang, T.-C. The role of endothelial-mesenchymal transition in development and pathological process. IUBMB Life 64, 717–723 (2012).

359. Kerosuo, L. & Bronner-Fraser, M. What is bad in cancer is good in the embryo: importance of EMT in neural crest development. Semin. Cell Dev. Biol. 23, 320–332 (2012).

360. De Genaro, P., Simón, M. V., Rotstein, N. P. & Politi, L. E. Retinoic acid promotes apoptosis and differentiation in photoreceptors by activating the P38 MAP kinase pathway. Invest. Ophthalmol. Vis. Sci. 54, 3143–3156 (2013).

361. Guo, C. et al. Global identification of MLL2-targeted loci reveals MLL2’s role in diverse signaling pathways. Proc. Natl. Acad. Sci. U. S. A. 109, 17603–17608 (2012).

202

362. Kupumbati, T. S. et al. Dominant negative retinoic acid receptor initiates tumor formation in mice. Mol. Cancer 5, 12 (2006).

363. Manshouri, T. et al. Downregulation of RAR alpha in mice by antisense transgene leads to a compensatory increase in RAR beta and RAR gamma and development of lymphoma. Blood 89, 2507–2515 (1997).

364. Blackman, M. A., Tigges, M. A., Minie, M. E. & Koshland, M. E. A model system for peptide hormone action in differentiation: interleukin 2 induces a B lymphoma to transcribe the J chain gene. Cell 47, 609–617 (1986).

365. Soucek, L. et al. Modelling Myc inhibition as a cancer therapy. Nature 455, 679–683 (2008).

366. Annibali, D. et al. Myc inhibition is effective against glioma and reveals a role for Myc in proficient mitosis. Nat. Commun. 5, 4632 (2014).

367. Blonska, M. et al. Jun-regulated genes promote interaction of diffuse large B-cell lymphoma with the microenvironment. Blood (2014). doi:10.1182/blood-2014-04-568188

368. Haldar, S., Basu, A. & Croce, C. M. Bcl2 is the guardian of microtubule integrity. Cancer Res. 57, 229–233 (1997).

369. Kipps, T. J. et al. A phase 2 study of the BH3 mimetic BCL2 inhibitor navitoclax (ABT- 263) with or without rituximab, in previously untreated B-cell chronic lymphocytic leukemia. Leuk. Lymphoma 1–30 (2015). doi:10.3109/10428194.2015.1030638

370. Schiffrin, E. L. Vascular changes in hypertension in response to drug treatment: Effects of angiotensin receptor blockers. Can. J. Cardiol. 18 Suppl A, 15A–18A (2002).

371. Cheng, S.-M. et al. Irbesartan inhibits human T-lymphocyte activation through downregulation of activator protein-1. Br. J. Pharmacol. 142, 933–942 (2004).

372. Chambwe, N. et al. Variability in DNA methylation defines novel epigenetic subgroups of DLBCL associated with different clinical outcomes. Blood 123, 1699–1708 (2014).

373. Jiang, Y., Soong, T. D., Wang, L., Melnick, A. M. & Elemento, O. Genome-wide detection of genes targeted by non-Ig somatic hypermutation in lymphoma. PloS One 7, e40332 (2012).

203

Appendices

Appendix A: Oligonucleotide sequences

This table lists the sequences of primers used for qRT-PCR, ChIP-qPCR, sequencing and site directed mutagenesis, as well as the sequences of gel shift assay probes. Sequences similar to a MEF2 motif in the gel shift assay probes are bolded and underlined.

Assay Gene Forward primer (5' to 3') Reverse primer (5' to 3') MEF2A CAAATGGAGCTGGAAGCAGT GGAGGGGGAGACTTTGTAGG qRT-PCR MEF2B GACCGTGTGCTGCTGAAGTA AGCCTCCGAAACTTCTCTCC MEF2C ACCAGGTGAGACCAGCAGAC GTGGTCTGATGGGTGGAGAC MEF2D GTTGAAGCCCTTCTTCCTCA TCAACCACTCCAACAAGCTG

AKT1 CCCAGCAGCTTCAGGTACTC GCTCACCCAGTGACAACTCA AMOT GCCTCTCTTTTGGAGGATGA GACGAGAACCGGAACTTGAG BCL6 TGAGAAGCCCTATCCCTGTG CTGGCTTTTGTGACGGAAAT

CARD11 TCCAACATCTACCCCATCGT CAGGAACTCCTCCTCCGTCT

CAV1 GAGCTGAGCGAGAAGCAAGT TCCCTTCTGGTTCTGCAATC CCL2 CCCCAGTCACCTGCTGTTAT AGATCTCCTTGGCCACAATG CDH13 GAATGACAACCGACCGATCT TATACCGCAGGAGGGCATTA

CTSB CCAGTAGGGTGTGCCATTCT TGTGTATTCGGACTTCCTGCT CXCL12 CTTTAGCTTCGGGTCAATGC AGAGCCAACGTCAAGCATCT FN1 GCTCATCATCTGGCCATTTT TGCTTAGGCTTTGGAAGTGG

GNA12 GGTCCAAGTTGTCCAGGAAG CCACCTTCCAGCTGTACGTC

INPP5K GCTTCAGCCTCCACAGGAT CGCCCACCTACAAGTTTGAT ITGA5 CTGTTCCCCTGAGAAGTTGTAGA GTGCCCAAAGGGAACCTC JUN GTTGCAGTGGAGAGGGACAG CCACCAATTCCTGCTTTGAG

LGALS1 CAAACCTGGAGAGTGCCTTC CAGGTTGTTGCTGTCTTTGC MAP2 GAGAATGGGATCAACGGAGA CTGCTACAGCCTCAGCAGTG MYC GAGGCTATTCTGCCCATTTG CACCGAGTCGTAGTCGAGGT

NDRG1 GGGTGCCATCCAGAGAAGT CTCGCTGAGGCCTTCAAGTA

PAK1 TCCCTCATGACCAGGATCTC ACCGTGTACACAGCAATGGA PLCG1 TACCATGGGCACACCCTTAC AGTGGTCCTCAATGGACAGG RHOB CGACGTCATTCTCATGTGCT GGGACAGAAGTGCTTCACCT

RHOD GTCGCTGCTGATGGTCTTC GCACAGGTTTGCCTTTCACT ROCK1 TTTGAGATGCTTCACCTCCTC GCTGAACGAAGAGACAGAGGTC

204

Assay Gene Forward primer (5' to 3') Reverse primer (5' to 3') RRAS TCCTCAATAGTGGGGTCGTAG CAGCGAGACACACAAGCTG

SEMA3C CTTCCAGCTCCTCCAGAATG ACGCTGCTGATGGGAGATAC

SIX1 CCCCTTCCAGAGGAGAGAGT TTAAGAACCGGAGGCAAAGA SMAD2 GGCCTGTTGTATCCCACTGA TGAGCTTGAGAAAGCCATCA SMAD3 CAGGGCTTTGAGGCTGTCTA AGCAGGGGGTACTGGTCAC

SMAD4 TGGAGCTCATCCTAGTAAATGTGT TTGTGAAGATCAGGCCACCT TGFB1 CCGGGTTATGCTGGTTGTA GGCTACCATGCCAACTTCTG qRT-PCR RPSAP58 TAGCCCCTGGTGACTCTGTC GGCCACATACCTAAACGTCAA (continued) VEGFB CGGTACCCGAGCAGTCAG GGCTTCACAGCACTGTCCTT PGK1 GGGAAAAGATGCTTCTGGGAA TTGGAAAGTGAAGCTCGGAAA TBP CAGCTCTTCCACTCACAGACT GTGCAATGGTCTTTAGGTCAA RFX5 GTCACAGGTCAGGGCACAC CCTGCCTGGACTTGACCTAA HIPK1 ATTGCCCTTGTATTCCACCA TATCACACAGCCCAGTGACC EGR1 TGAACAACGAGAAGGTGCTG CACAAGGTGTTGCCACTGTT TUBA4A CCAGCACCAGTTTCACAGAA GCTGGGAGCTCTATTGCTTG BCS1L TGTTCATGACCACCAACCAC GCTGCCAGTGTGAGCAGTAG AFF2 CTGAAGCACAAAGCTGATGC TATGGGGACTTTGCTTCCAG TPPP3 TCTACAGCACCCCCTGTTTT ATTCAAGGGGAAGAGCAAGG GPX3 ATGCTGGCAAATACGTCCTC AGAATGACCAGACCGAATGG BARX2 AGAATCGCAGGATGAAATGG TGGGGATGGAGTTCTTCTTG PRKCH CCAAATAGAACCGCCTTTCA TCATTGGAAGATGTCCCTCA COPS7B TTAGAGGCTCCCGGAGTGTA TCCCATAGGCAAACAGGTTC SLC5A6 GGCCATTCCTAGACAAAGCA ACGATGGAAGACCTGATTCG HOXD13 CAGGTGTACTGCACCAAGGA CCTCTTCGGTAGACGCACAT MEF2C GGTATTTCCCATGCACGTTT CGGCCTCAGCTAAATGAAAG

RHOB GCGGCCAATCAGAGCTAAG CCTAGCGCCCGCTATTTA

ChIP-qPCR CAV1 GAGATGATGCACTGCGAAAA GGTGCTTGGGCAGATTATTT CDH13 GGCATTTTGGTAGGAGGTGA GCCACTTCTGGGACAGACTT ITGA5 TGTCTGACCCAGGAGAAACC TCCTGTGCTCTGTGCAAACT

PAK1 GCACATAGCTGCTGGAGTCA GGTGAGGAAACCTGAGAGTCA RHOD ACGCCTGGATCCAAATTCTA GGCACTGTTCCCTAGGAGGT SEMA3C TCCACAGTCACGTCTTGAGG AGGCCTCTTATGGCTCTCCT

SIX1 TCCCATCATCTGCTGTTTTG TGCTTTGGAACTAGGTGTGC

BCL2 TGTGGTGTGCTTCTTGACATC TGCAATTCAGTGCTTCCTTTT JUN GGGTGACATCATGGGCTATT CTGTCTGTCTGCCTGACTCC ZNF608 AGGGAGTGGCGACTTTTACA TGATAAATTGGACCTTATGAAACCT

ABCB4 TCAGGCTAAAGGCGAAAATG CCTGGCCCTTGTTAAACTCA BCL6 TGCATTGTAGTTGTGGCAGTC GAGCCAAATACCTGTTTGTGTTT

205

Assay Gene Forward primer (5' to 3') Reverse primer (5' to 3') CPS1 TGGAGAAAAGTTTATTCTAACGTTCTT TGTTGGGGTAAAAAGACAATGTT intergenic TATTTGATGGTCCCCAAACC AGGATGGTCCAGGGAAAGAC PCR of MEF2B cDNA GGAGGAATTCATGGGGAGGAAAAAAATCCAGATC GGAGGCTAGCCCGGGGCCAGCCGTCGGCCA A1R AGTCAGCATCAAGTCTGAGCG

Sequencing A2F CGCTCAGACTTGATGCTGACT MEF2B A3R AAGCCACCTCACCAGCAAGAC cDNA A4F GTCTTGCTGGTGAGGTGGCTT

A5.2 AGGAGCCAGGAGAGAAGTTT A7 GGGGAGCTTCCCCTTCCTC Site- K4E TGCGGAAGATCTGGATTTTTTCCCTCCCCATCG CGATGGGGAGGGAAAAAATCCAGATCTCCCGCA directed Y69H CGCTGTACTCTGTGTGCTTCAGCAGCACACG CGTGTGCTGCTGAAGCACACAGAGTACAGCG mutagenesis D83V CCGCACCAACACTGTCATCCTCGAGACGC GCGTCTCGAGGATGACAGTGTTGGTGCGG

R3T GAGATCTGGATTTTTTTCGTCCCCATGGTTCGAACTC GAGTTCGAACCATGGGGACGAAAAAAATCCAGATCTC R24L GTGACGTTCACCAAGCTGAAGTTCGGGCTGATG CATCAGCCCGAACTTCAGCTTGGTG AACGTCAC JUN GCCAGTCAACCCCTAAAAATAGCCCATGAT GGGTGACATCATGGGCTATTTTTAGGGGTT BCL2 GTCCCAAGAGGCTATAAAAGGAAGC GATTCAGTGCTTCCTTTTATAGCCTC Gel shift assay CDH13 GCTGGAATTTGCCTATGATAAGTAATGA GTCACAGTCATTACTTATCATAGGCAAA probes ZNF608 GCAGAATAAACAGACCAAAAATAGACATTTAA GGCATGCTTAAATGTCTATTTTTGGTCTGTTT

206

Appendix B: Quality control statistics for RNA-seq data.

All quality control statistics were produced using an in-house pipeline at the BCGSC and passed the BCGSC quality standards. Sequencing data were aligned to the hg19 genome using BWA276.

average M as % properly P as % average coverage average % % total duplicate mapped of total paired of total coverage in intergenic % exon intron intergenic sample reads reads reads (M) reads reads (P) reads in exons introns coverage reads reads reads WT MEF2B-V5 216913628 38926007 193379141 89.15% 1.79E+08 82.7% 88.65 0.558 0.177 90.76 6.86 2.38 K4E MEF2B-V5 237278192 41405283 209312519 88.21% 1.94E+08 81.6% 95.99 0.583 0.185 91.06 6.64 2.30 Y69H MEF2B-V5 250039810 45073674 228750317 91.49% 2.15E+08 86.1% 106.27 0.582 0.218 91.52 6.02 2.46 D83V MEF2B-V5 236565298 42125131 212306044 89.75% 1.98E+08 83.5% 98.65 0.534 0.187 91.75 5.97 2.28 empty vector 261588114 50939870 228969442 87.53% 2.11E+08 80.8% 104.41 0.564 0.209 91.65 5.95 2.40

207

Appendix C: Quality control statistics for V5 ChIP-seq data.

All quality control statistics were produced using an in-house pipeline at the BCGSC. Sequencing data were aligned to the hg19 genome using BWA276. Peaks were identified using FindPeaks278.

% of total reads that are: Number MEF2B- Total Number of N as % of Uniquely of peaks Mb V5 in Replicate number of Mapped reads in peaks total mapped mapped (FDR covered cells reads reads (N) reads reads 0.001) 1 60810194 91.3 87.3 12702 17.738 557959 1.1 WT 2 63408292 94.5 90.1 30232 46.673 1483948 2.6 1 54051882 93 89.1 1488 3.107 192191 0.4 K4E 2 58283450 91.7 87.9 1043 2.302 173636 0.3 1 74619022 92.6 88.9 13040 23.294 596319 0.9 D83V 2 53218136 93.8 89.7 5335 9.342 309336 0.6

208

Appendix D: Quality control statistics for H3K27ac and H3K4me3 ChIP-seq data.

All quality control statistics were produced using an in-house pipeline at the BCGSC. Sequencing data were aligned to the hg19 genome using BWA276. Peaks were identified using FindPeaks278.

% of total reads that are: Total Number of Mb Number of N as % of Uniquely Cells Antibody Replicate number of Mapped peaks (FDR covered reads in total mapped mapped reads reads 0.01) by peaks peaks (N) reads reads 1 47523424 92.7 89.6 46849 160.011 4453378 10.1 WT H3K27ac 2 90219214 92.3 89.2 52891 153.039 4935460 5.9 MEF2B- 1 12849500 96.7 94.4 34029 56.139 10289612 82.9 V5 H3K4me3 2 14067164 86.3 84.2 36055 55.858 10038802 82.7 1 59742664 93.9 90.9 46724 141.442 4513081 8 H3K27ac empty 2 106375106 90.6 87.8 45446 140.868 4533007 4.7 vector 1 12492796 96.0 93.7 29501 47.545 10006561 83.4 H3K4me3 2 14479372 96.0 93.7 32940 53.379 11698289 84.2

209

Appendix E: Genes differentially expressed in microarray and RNA-seq data for WT MEF2B-V5 versus control cells (adjusted p-values < 0.05)

The analysis of microarray data compared WT MEF2B-V5 cells to untransfected cells, whereas the analysis of RNA-seq data compared WT MEF2B-V5 cells to empty vector cells. Positive log2 fold change indicates expression was higher in WT MEF2B-V5 expressing cells than in control cells.

Microarray data RNA-seq data Gene Symbol Gene ID log2 fold adjusted log2 fold adjusted change p-value p-value change p-value p-value ACSM3 ENSG00000005187 -0.59672 0.000922 0.009347 -1.82627 6.71E-05 0.006034 ADAM12 ENSG00000148848 0.50799 5.21E-05 0.00173 2.570864 9.17E-08 1.90E-05 ADAMTS1 ENSG00000154734 0.904473 4.70E-07 0.000167 1.492213 0.000482 0.030147 ADAMTS16 ENSG00000145536 0.876993 7.27E-07 0.000195 2.621228 1.04E-08 2.75E-06 ADAMTS19 ENSG00000145808 -0.35562 0.000993 0.009824 -1.73944 0.000385 0.02534 ADAMTS5 ENSG00000154736 1.145397 1.81E-06 0.00035 1.942464 9.94E-06 0.001211 ADCYAP1 ENSG00000141433 0.236334 0.005579 0.030882 3.890649 3.42E-11 1.40E-08 ADPRH ENSG00000144843 -0.32054 0.000814 0.008649 -4.65761 1.09E-07 2.24E-05 AFAP1L2 ENSG00000169129 0.345173 0.000951 0.009582 1.814837 4.10E-05 0.003996 AGMAT ENSG00000116771 0.44259 0.000384 0.005578 3.861698 3.48E-06 0.000472 ALDH1A2 ENSG00000128918 -1.79026 1.19E-06 0.000268 -3.21514 3.13E-12 1.56E-09 ALPK3 ENSG00000136383 -0.36838 0.001346 0.012166 -3.41496 2.89E-13 1.69E-10 AMHR2 ENSG00000135409 -0.27139 0.002652 0.018839 -2.64272 3.44E-06 0.000472 AMOT ENSG00000126016 1.181965 8.29E-08 6.68E-05 1.639437 0.000137 0.011085 ANGPTL2 ENSG00000136859 -0.64442 3.06E-05 0.00133 -3.00653 4.87E-10 1.64E-07 ANKRD1 ENSG00000148677 2.14054 9.19E-08 7.08E-05 4.75586 2.37E-20 3.99E-17 ANO1 ENSG00000131620 0.503593 2.33E-05 0.001172 2.212018 2.70E-06 0.000384 ANXA1 ENSG00000135046 0.514993 0.000265 0.00444 2.37649 7.06E-08 1.55E-05 ARSE ENSG00000157399 0.633782 0.000137 0.00297 2.698181 7.50E-08 1.63E-05 ATOH8 ENSG00000168874 0.286927 0.002967 0.020273 1.885612 2.15E-05 0.002325 ATP2B3 ENSG00000067842 -0.44521 0.000321 0.005004 -3.29253 5.52E-11 2.15E-08 ATP9B ENSG00000166377 1.027165 4.77E-06 0.000522 1.69281 9.87E-05 0.008341 BARHL2 ENSG00000143032 -0.55368 1.58E-05 0.000972 -2.19103 1.66E-06 0.000254 BLNK ENSG00000095585 -0.35381 0.007634 0.038129 -1.97534 0.000246 0.017688 BMP2 ENSG00000125845 0.761488 8.10E-06 0.000724 3.666834 9.57E-15 6.29E-12 BMP5 ENSG00000112175 0.523535 1.70E-05 0.001006 3.466025 8.81E-08 1.84E-05 C16orf45 ENSG00000166780 0.465829 0.000592 0.007167 1.513362 0.000827 0.04631 C21orf7 ENSG00000156265 0.38846 0.000633 0.007476 4.791796 2.42E-09 7.42E-07 C21orf90 ENSG00000182912 -0.33741 0.008539 0.041432 -2.98519 0.000251 0.017854 CACNG4 ENSG00000075461 1.035104 3.84E-07 0.000151 2.934305 3.16E-10 1.11E-07 CADPS2 ENSG00000081803 -1.763 8.45E-07 0.000217 -3.39515 3.08E-13 1.76E-10 CAPN5 ENSG00000149260 0.558326 1.65E-05 0.000993 1.695906 0.000102 0.00857 CARD11 ENSG00000198286 1.582606 4.09E-08 4.53E-05 4.8113 1.88E-22 4.23E-19 CCKBR ENSG00000110148 0.317352 0.00097 0.009684 1.62423 0.000497 0.030734 CCL2 ENSG00000108691 1.156184 5.92E-06 0.000589 3.082405 2.16E-09 6.68E-07 CCNA1 ENSG00000133101 0.868061 3.06E-05 0.00133 2.060681 3.19E-05 0.003279 CD69 ENSG00000110848 0.237885 0.006595 0.034392 3.336742 0.000103 0.008642 CDC42EP3 ENSG00000163171 0.923567 6.51E-07 0.00018 1.48229 0.000547 0.033247 CDH13 ENSG00000140945 1.598786 1.28E-07 8.50E-05 2.911931 2.38E-10 8.50E-08

210

Microarray data RNA-seq data Gene Symbol Gene ID log2 fold adjusted log2 fold adjusted change p-value p-value change p-value p-value CDH4 ENSG00000179242 1.201654 1.80E-07 9.14E-05 2.377237 8.83E-08 1.84E-05 CDK18 ENSG00000117266 -0.26869 0.005442 0.030415 -1.56839 0.000305 0.020826 CDKL5 ENSG00000008086 0.787307 9.94E-05 0.00251 1.655198 0.000202 0.015276 CERKL ENSG00000188452 -0.30488 0.00144 0.012798 -2.66326 0.000131 0.010664 CHRDL1 ENSG00000101938 -1.07015 1.50E-07 8.65E-05 -2.12958 2.21E-06 0.000324 CHST1 ENSG00000175264 0.749837 7.54E-06 0.000695 2.178402 3.77E-06 0.000507 CHST8 ENSG00000124302 0.331549 0.002495 0.018053 1.805295 4.54E-05 0.00438 CHSY3 ENSG00000198108 -0.89954 0.000435 0.005965 -3.38231 3.51E-12 1.72E-09 CLIC5 ENSG00000112782 1.032509 4.29E-07 0.00016 1.735037 0.000568 0.033779 CLIP4 ENSG00000115295 -0.41035 0.000306 0.004878 -5.21812 8.86E-16 7.23E-13 CLSTN2 ENSG00000158258 0.734547 1.44E-05 0.000934 1.761897 6.44E-05 0.005874 CLU ENSG00000120885 0.573977 8.42E-06 0.000734 1.440153 0.00075 0.042805 COL12A1 ENSG00000111799 2.083948 5.58E-09 2.16E-05 4.216503 1.25E-18 1.68E-15 COL2A1 ENSG00000139219 -0.33969 0.000826 0.008718 -1.71566 6.80E-05 0.006081 CPA4 ENSG00000128510 1.349489 7.68E-07 0.0002 3.776366 3.87E-11 1.53E-08 CPS1 ENSG00000021826 1.57982 2.60E-08 4.19E-05 2.308113 2.02E-07 3.83E-05 CRABP1 ENSG00000166426 -0.43617 5.18E-05 0.001729 -4.99161 2.21E-05 0.002385 CRABP2 ENSG00000143320 -0.56263 7.48E-05 0.00211 -2.25085 3.15E-07 5.80E-05 CRLF1 ENSG00000006016 0.833845 3.30E-06 0.000423 2.585729 7.01E-09 1.97E-06 CTGF ENSG00000118523 1.392995 1.94E-06 0.000356 3.167806 3.96E-12 1.87E-09 CXorf23 ENSG00000173681 0.830693 0.000191 0.003645 1.732876 6.82E-05 0.006084 CYB561 ENSG00000008283 -0.61945 0.000108 0.00262 -7.22877 5.41E-27 2.08E-23 DDR2 ENSG00000162733 -0.53668 1.27E-05 0.000863 -1.74173 0.000208 0.015533 DERL3 ENSG00000099958 0.968845 1.06E-07 7.81E-05 1.569281 0.000261 0.01838 DLX2 ENSG00000115844 -0.46252 6.95E-05 0.002031 -1.45713 0.000679 0.039569 DPP4 ENSG00000197635 -0.34331 0.000947 0.009563 -1.71454 0.000471 0.029632 ECM1 ENSG00000143369 0.536877 4.69E-05 0.00166 1.728455 0.000155 0.012259 EDIL3 ENSG00000164176 1.821847 1.97E-08 3.49E-05 2.634595 3.60E-09 1.08E-06 EDN1 ENSG00000078401 0.432253 0.001553 0.01342 4.705976 1.44E-10 5.45E-08 EFEMP2 ENSG00000172638 1.708933 7.76E-09 2.16E-05 2.736689 1.31E-09 4.21E-07 EFNA2 ENSG00000099617 0.482964 0.000362 0.005392 2.39803 3.99E-07 7.12E-05 EGF ENSG00000138798 0.600133 4.91E-05 0.001694 1.601547 0.000829 0.04631 ELOVL3 ENSG00000119915 -0.78285 1.47E-05 0.000936 -7.25063 2.69E-19 4.26E-16 EMP1 ENSG00000134531 1.256623 5.34E-07 0.00017 1.820432 5.25E-05 0.004943 ENPEP ENSG00000138792 -1.09391 6.72E-06 0.000648 -2.69405 2.08E-08 5.24E-06 ENPP2 ENSG00000136960 -1.77669 8.83E-07 0.000218 -3.1469 7.23E-12 3.09E-09 EPAS1 ENSG00000116016 1.134248 2.05E-07 9.58E-05 1.486187 0.000523 0.032105 EPHB6 ENSG00000106123 0.323619 0.001539 0.013343 1.627915 0.000223 0.016459 EPS8L2 ENSG00000177106 0.442486 0.000259 0.004385 2.489238 2.06E-08 5.24E-06 ERBB4 ENSG00000178568 -0.93245 1.95E-06 0.000356 -1.85673 2.47E-05 0.002595 ESRP1 ENSG00000104413 0.953205 2.93E-06 0.000412 3.994904 6.31E-11 2.43E-08 ESRRB ENSG00000119715 -0.30587 0.001639 0.013866 -1.75706 0.00013 0.010664 F10 ENSG00000126218 0.529282 1.80E-05 0.001027 2.167498 4.55E-06 0.0006 FABP6 ENSG00000170231 -0.70403 3.28E-06 0.000423 Infinity 3.64E-12 1.75E-09 FAIM2 ENSG00000135472 0.493746 0.000344 0.005233 2.185118 7.45E-06 0.000942 FAM111A ENSG00000166801 0.793026 0.000163 0.003339 1.680183 0.000154 0.012215 FAM131B ENSG00000159784 0.394755 0.001538 0.013343 1.944115 2.27E-05 0.002411 FAM155B ENSG00000130054 -0.32031 0.003626 0.023154 -1.58387 0.000271 0.018795 FAM189A2 ENSG00000135063 0.840134 5.87E-06 0.000588 1.482637 0.000859 0.047625 FGF5 ENSG00000138675 -0.4598 0.000562 0.00696 -3.74007 8.65E-09 2.33E-06 FN1 ENSG00000115414 1.84056 7.53E-09 2.16E-05 3.127555 5.42E-12 2.43E-09 FNDC4 ENSG00000115226 0.702039 9.81E-06 0.000776 1.638569 0.000239 0.017376 211

Microarray data RNA-seq data Gene Symbol Gene ID log2 fold adjusted log2 fold adjusted change p-value p-value change p-value p-value FOLR1 ENSG00000110195 0.805964 2.63E-06 0.00041 2.200462 2.26E-06 0.000329 FSTL3 ENSG00000070404 0.447092 0.000855 0.008908 1.81443 3.89E-05 0.003823 GALC ENSG00000054983 -1.61633 1.32E-07 8.50E-05 Infinity 2.61E-31 2.34E-27 GALNT13 ENSG00000144278 -2.0277 9.73E-09 2.16E-05 -6.69978 9.47E-30 5.10E-26 GEMIN4 ENSG00000179409 -0.66766 2.49E-05 0.001206 -1.59383 0.000206 0.015435 GPM6A ENSG00000150625 -1.12088 2.75E-06 0.00041 -4.61996 8.09E-19 1.18E-15 GPR116 ENSG00000069122 0.964094 3.80E-06 0.000463 1.984612 2.11E-05 0.002292 GPR64 ENSG00000173698 0.89147 4.27E-05 0.001601 1.437867 0.000838 0.046613 GPRC5A ENSG00000013588 1.34932 3.39E-07 0.00014 2.488747 1.81E-08 4.64E-06 GYG2 ENSG00000056998 1.476208 6.18E-08 6.08E-05 4.171401 1.99E-17 2.15E-14 HAAO ENSG00000162882 -0.24788 0.011063 0.04972 -4.89208 4.41E-07 7.71E-05 HHIP ENSG00000164161 -0.25739 0.001608 0.013719 -1.75321 0.000252 0.017854 HOXA11 ENSG00000005073 -0.4668 0.002929 0.020131 -2.28947 9.85E-07 0.000158 HS6ST3 ENSG00000185352 -0.51031 2.52E-05 0.001209 -2.31813 4.04E-06 0.000542 ID1 ENSG00000125968 0.423581 0.00031 0.004916 2.378626 8.19E-08 1.74E-05 ID2 ENSG00000115738 0.582677 0.000304 0.004868 1.875098 1.55E-05 0.001763 ID4 ENSG00000172201 0.614377 3.39E-05 0.001402 1.471904 0.000576 0.034175 IGFBP3 ENSG00000146674 0.462207 0.000604 0.007267 1.510056 0.000852 0.047336 IGFBP4 ENSG00000141753 -0.46876 0.000242 0.004183 -2.45571 1.66E-07 3.21E-05 IKZF3 ENSG00000161405 -0.62213 4.29E-05 0.001602 -1.69424 0.000429 0.027512 INA ENSG00000148798 1.796651 1.23E-08 2.42E-05 4.42551 6.60E-15 4.55E-12 IPCEF1 ENSG00000074706 -0.23581 0.006638 0.034514 -1.93938 0.000483 0.030147 IQCA1 ENSG00000132321 0.526576 0.000138 0.002972 4.507736 9.72E-12 4.09E-09 ISL1 ENSG00000016082 -0.80987 4.45E-05 0.001637 -5.80719 1.94E-20 3.48E-17 ITGA4 ENSG00000115232 -0.83404 9.86E-06 0.000777 -2.18073 1.09E-06 0.000173 ITGA8 ENSG00000077943 -0.46939 0.000187 0.003591 -2.3134 2.72E-06 0.000386 ITPR3 ENSG00000096433 -0.23444 0.005905 0.032075 -1.57635 0.000243 0.017601 JDP2 ENSG00000140044 0.21566 0.005606 0.030982 1.639034 0.000282 0.019451 KCNS3 ENSG00000170745 1.115315 1.08E-06 0.000248 1.632984 0.000188 0.014411 KCNT2 ENSG00000162687 -1.03036 1.21E-07 8.50E-05 -5.50651 8.36E-19 1.18E-15 LGALS1 ENSG00000100097 0.61162 0.000425 0.005912 1.722938 9.50E-05 0.008125 LGALS8 ENSG00000116977 -0.30434 0.001398 0.012509 -3.41624 2.88E-06 0.000404 LGR5 ENSG00000139292 -0.95825 4.49E-05 0.001637 -1.70381 9.06E-05 0.007823 LHFPL4 ENSG00000156959 -0.83588 1.77E-05 0.001021 Infinity 3.44E-23 9.28E-20 LIPG ENSG00000101670 0.781913 5.21E-06 0.000547 2.076486 3.50E-05 0.003482 LRAT ENSG00000121207 -0.39874 0.000427 0.00592 -1.98682 2.03E-05 0.002219 LRP2 ENSG00000081479 -0.80111 4.03E-06 0.00047 -1.71638 7.92E-05 0.00699 LSR ENSG00000105699 0.662337 1.19E-05 0.000841 1.546004 0.000334 0.022332 MAP3K12 ENSG00000139625 0.746424 3.90E-06 0.000468 1.704228 9.57E-05 0.008155 MAP6 ENSG00000171533 0.913406 7.63E-07 0.0002 2.902844 1.50E-10 5.60E-08 MEIS3 ENSG00000105419 0.254221 0.004151 0.025431 1.811316 3.18E-05 0.003279 MGST1 ENSG00000008394 -0.72909 0.000319 0.004983 -1.7491 6.06E-05 0.005612 MLH3 ENSG00000119684 -1.43852 1.53E-07 8.65E-05 -4.56092 2.30E-18 2.70E-15 MPZL2 ENSG00000149573 -0.80597 9.10E-06 0.000752 -3.397 6.37E-07 0.000106 MTSS1 ENSG00000170873 -0.47042 4.58E-05 0.001644 -1.69253 0.000111 0.009197 MYC ENSG00000136997 -1.61194 1.34E-07 8.50E-05 -3.22289 1.55E-12 8.33E-10 MYCN ENSG00000134323 0.941693 5.76E-07 0.000174 2.308958 4.28E-07 7.54E-05 MYOF ENSG00000138119 1.173907 5.34E-07 0.00017 3.238306 2.42E-12 1.23E-09 NAP1L3 ENSG00000186310 0.71701 0.000516 0.006624 2.090032 4.90E-06 0.000641 NDRG1 ENSG00000104419 -1.18243 3.24E-08 4.28E-05 -2.37696 6.69E-08 1.48E-05 NEDD9 ENSG00000111859 0.869897 1.77E-05 0.001021 1.734652 8.05E-05 0.007088 NME5 ENSG00000112981 1.157281 3.27E-06 0.000423 2.268234 0.000696 0.040298 212

Microarray data RNA-seq data Gene Symbol Gene ID log2 fold adjusted log2 fold adjusted change p-value p-value change p-value p-value NOS2 ENSG00000007171 -0.40371 6.88E-05 0.002022 -2.60868 2.67E-08 6.48E-06 NOV ENSG00000136999 -0.40688 0.002263 0.016955 -1.68689 0.000248 0.017696 NPR1 ENSG00000169418 -0.23714 0.008456 0.041142 -2.55634 1.31E-07 2.59E-05 NPTX2 ENSG00000106236 -0.54542 2.24E-05 0.001142 -1.66464 0.000131 0.010664 NUPR1 ENSG00000176046 1.421164 7.15E-08 6.31E-05 2.460156 5.15E-08 1.19E-05 OAS3 ENSG00000111331 0.937445 5.37E-07 0.00017 3.012532 3.60E-11 1.45E-08 OLFM1 ENSG00000130558 1.870148 8.13E-09 2.16E-05 2.705738 2.49E-09 7.54E-07 OXGR1 ENSG00000165621 -0.54087 0.00168 0.01404 -7.0482 3.14E-11 1.30E-08 PAMR1 ENSG00000149090 1.106228 6.03E-07 0.000174 2.759522 1.74E-09 5.51E-07 PARP14 ENSG00000173193 0.956785 1.35E-06 0.000295 2.699426 3.86E-09 1.13E-06 PCDHB15 ENSG00000113248 1.104669 3.80E-05 0.001494 1.917589 0.000314 0.021334 PCDHB7 ENSG00000113212 0.39679 0.000112 0.002667 1.988528 0.0004 0.025905 PCDHB9 ENSG00000120324 0.494502 0.000563 0.006966 1.69479 0.000362 0.023982 PCOLCE2 ENSG00000163710 -0.56519 9.67E-06 0.00077 -1.74491 8.73E-05 0.007588 PDZD4 ENSG00000067840 0.339258 0.001154 0.010939 1.546042 0.00039 0.025541 PGR ENSG00000082175 0.628898 0.000258 0.004365 4.01185 5.53E-16 4.80E-13 PHLDB2 ENSG00000144824 1.704368 9.38E-09 2.16E-05 2.123601 1.28E-06 0.000201 PKP3 ENSG00000184363 0.213184 0.007779 0.038633 4.123076 6.28E-17 6.51E-14 PLA2G4A ENSG00000116711 -0.7853 0.000116 0.002712 -2.83202 1.30E-08 3.38E-06 PLBD1 ENSG00000121316 -1.98279 1.51E-07 8.65E-05 -8.33075 5.06E-27 2.08E-23 PLXDC2 ENSG00000120594 -2.29201 6.50E-09 2.16E-05 -7.23853 1.92E-24 5.76E-21 PLXNA4 ENSG00000221866 0.821182 2.39E-06 0.000396 2.261267 3.34E-07 6.08E-05 PPFIA2 ENSG00000139220 -0.23685 0.003754 0.023678 -4.42304 1.88E-09 5.88E-07 PPP1R3C ENSG00000119938 -0.69787 1.09E-05 0.00082 -3.65761 1.74E-07 3.34E-05 PPP2R2B ENSG00000156475 0.354444 0.00048 0.006335 2.515414 3.63E-07 6.56E-05 PPP2R2C ENSG00000074211 1.078996 6.80E-08 6.31E-05 1.909186 1.28E-05 0.001511 PRKG2 ENSG00000138669 -0.35993 0.000192 0.003648 -3.65032 1.00E-05 0.001211 PRR16 ENSG00000184838 -0.494 0.000315 0.004952 -5.00311 1.30E-15 1.03E-12 PRRX1 ENSG00000116132 -0.29763 0.00118 0.011145 -1.70906 0.000208 0.015533 PRSS23 ENSG00000150687 1.004055 1.91E-07 9.42E-05 1.689511 8.66E-05 0.007546 PTCHD1 ENSG00000165186 -0.59138 2.45E-05 0.001197 -1.69511 0.000114 0.009429 QRFPR ENSG00000186867 -0.71635 1.20E-05 0.000844 -1.56623 0.000556 0.033622 RAB42 ENSG00000188060 -0.48696 0.00128 0.011749 -5.49071 2.30E-07 4.30E-05 RAI2 ENSG00000131831 0.958619 2.15E-05 0.001112 4.334385 1.99E-18 2.44E-15 RBMS3 ENSG00000144642 0.957593 5.15E-06 0.000543 1.778002 6.43E-05 0.005874 RBP1 ENSG00000114115 0.405233 0.001679 0.01404 1.935989 5.85E-05 0.005453 RGS16 ENSG00000143333 -0.60175 4.27E-05 0.001601 -2.08019 2.01E-06 0.000298 RGS7 ENSG00000182901 -0.24044 0.006807 0.035042 -2.1912 9.88E-05 0.008341 RIC3 ENSG00000166405 0.372394 0.000825 0.008718 3.100705 1.53E-07 3.00E-05 RNF165 ENSG00000141622 -0.27882 0.00163 0.013818 -2.01897 1.44E-05 0.001666 RNMTL1 ENSG00000171861 -0.69733 5.61E-06 0.000571 -1.475 0.000683 0.039636 ROR2 ENSG00000169071 -0.42949 0.000523 0.006698 -1.9747 7.69E-06 0.000968 RPS6KA6 ENSG00000072133 -1.01865 0.000306 0.004878 -1.51882 0.000535 0.032604 RSAD2 ENSG00000134321 0.330665 0.004341 0.026214 2.775172 8.26E-06 0.001034 RUNX3 ENSG00000020633 -0.45093 6.60E-05 0.001982 -6.45985 7.75E-31 5.22E-27 SARDH ENSG00000123453 0.302044 0.000897 0.009167 1.936758 1.32E-05 0.001538 SCUBE2 ENSG00000175356 0.299201 0.000666 0.007728 2.181629 2.10E-05 0.002292 SERPINB8 ENSG00000166401 0.926447 9.36E-06 0.000761 3.080481 1.27E-09 4.13E-07 SERPINF1 ENSG00000132386 -1.39327 2.87E-06 0.000412 -3.84717 3.66E-15 2.82E-12 SFRP2 ENSG00000145423 -1.08819 4.40E-07 0.00016 -9.51266 4.53E-34 6.09E-30 SLC17A9 ENSG00000101194 -0.3068 0.004408 0.026431 -2.53538 2.01E-06 0.000298 SLC1A2 ENSG00000110436 1.114965 4.84E-06 0.000527 2.214035 7.95E-07 0.00013 213

Microarray data RNA-seq data Gene Symbol Gene ID log2 fold adjusted log2 fold adjusted change p-value p-value change p-value p-value SLC2A14 ENSG00000173262 0.416246 9.80E-05 0.002491 2.251854 1.73E-05 0.001942 SLC2A3 ENSG00000059804 1.209396 1.18E-05 0.000841 3.159574 4.59E-12 2.13E-09 SLC4A4 ENSG00000080493 -0.6668 1.37E-05 0.0009 -1.52284 0.000672 0.039261 SLIT3 ENSG00000184347 0.785778 2.12E-06 0.000374 1.442689 0.00082 0.046028 SMPD1 ENSG00000166311 0.700154 4.82E-05 0.001685 1.44928 0.000772 0.043757 SOX11 ENSG00000176887 0.269935 0.010645 0.048478 3.637936 1.87E-14 1.20E-11 ST6GALNAC5 ENSG00000117069 -0.41475 0.0004 0.005697 -2.73392 5.43E-08 1.22E-05 ST8SIA2 ENSG00000140557 0.393212 0.000431 0.005943 2.250443 1.06E-06 0.000169 SULF1 ENSG00000137573 0.980021 5.87E-07 0.000174 1.584402 0.000246 0.017688 SUSD2 ENSG00000099994 0.318661 0.000792 0.008511 2.551993 8.06E-08 1.72E-05 TAF7L ENSG00000102387 -0.86369 1.82E-05 0.001028 -1.65786 0.000147 0.011776 TCN2 ENSG00000185339 -0.62877 0.000127 0.002849 -1.84553 0.000272 0.018812 TGM2 ENSG00000198959 0.418333 0.000123 0.002804 1.875093 3.47E-05 0.003482 THBS1 ENSG00000137801 0.782172 4.32E-06 0.000491 1.832304 4.41E-05 0.004272 THNSL2 ENSG00000144115 0.581334 1.46E-05 0.000936 6.699368 9.01E-23 2.21E-19 TINAGL1 ENSG00000142910 0.347388 0.003805 0.023902 1.669158 0.00056 0.033658 TLE2 ENSG00000065717 0.433749 7.29E-05 0.00209 1.465468 0.0007 0.040483 TLL1 ENSG00000038295 -0.47248 2.76E-05 0.001273 -5.02958 6.58E-15 4.55E-12 TM6SF2 ENSG00000213996 0.26559 0.004009 0.02481 1.777775 0.000289 0.019824 TMBIM1 ENSG00000135926 1.063573 5.10E-07 0.00017 1.656291 0.000167 0.013145 TMEM59L ENSG00000105696 0.339911 0.000571 0.007018 1.511266 0.000501 0.030887 TNC ENSG00000041982 1.299053 7.48E-08 6.31E-05 1.836798 2.35E-05 0.00248 TNFAIP8L1 ENSG00000185361 1.181551 2.94E-08 4.28E-05 1.51829 0.000456 0.02874 TRIB1 ENSG00000173334 -0.74246 9.68E-06 0.00077 -1.99808 4.40E-06 0.000586 TSR1 ENSG00000167721 -0.86204 1.28E-05 0.000863 -1.50865 0.00043 0.027538 ULK2 ENSG00000083290 -2.24622 2.71E-09 2.16E-05 -8.59523 3.64E-39 9.82E-35 USP44 ENSG00000136014 -0.67236 6.09E-05 0.001893 -2.8258 8.27E-07 0.000134 VAT1L ENSG00000171724 1.055848 1.49E-06 0.000307 4.0365 2.84E-16 2.73E-13 VAV1 ENSG00000141968 -0.3399 0.001909 0.015164 -2.70835 2.21E-06 0.000324 VLDLR ENSG00000147852 1.298677 8.83E-07 0.000218 1.482759 0.000559 0.033658 ZIC1 ENSG00000152977 -1.10778 2.51E-07 0.000109 -2.28086 5.42E-07 9.24E-05 ZNF167 ENSG00000196345 0.860511 6.00E-06 0.000594 3.45222 7.96E-09 2.17E-06 ZNF253 ENSG00000213988 0.985577 0.001481 0.013004 2.014444 8.53E-06 0.001064 ZNF467 ENSG00000181444 0.25751 0.002617 0.018668 1.53195 0.000484 0.030147 ZNF532 ENSG00000074657 1.448763 4.05E-08 4.53E-05 3.652906 8.23E-15 5.54E-12 ZNF544 ENSG00000198131 1.225317 2.45E-06 0.000399 5.775453 4.82E-25 1.62E-21 ZNF563 ENSG00000188868 0.87168 0.000164 0.003344 2.038107 3.02E-05 0.003127 ZNF671 ENSG00000083814 -0.35459 0.001636 0.013845 Infinity 1.25E-12 6.86E-10

214

Appendix F: Functional annotation group enrichment of DEGs in WT MEF2B-V5 versus untransfected cells Shown are IPA cellular function annotation groups with absolute activation z-scores ≥ 2 and p-values ˂ 0.05 for the genes differentially expressed in WT MEF2B-V5 versus untransfected HEK293A cells. Positive z-scores indicate that MEF2B activity promotes the function and negative z-scores indicate that MEF2B activity opposes the function. In blue are annotation groups related to cell migration. In yellow are annotation groups related to EMT. In pink are annotation groups related to cell proliferation and viability.

Activation # of Category Functions Annotation p-value z-score genes Cell Death and Survival cell viability 2.68E-11 8.038 301 Cell Death and Survival cell survival 5.87E-10 7.915 308 Cell Death and Survival cell viability of tumor cell lines 9.44E-11 7.706 244 Cellular Movement invasion of cells 3.17E-17 7.048 245 Cellular Movement cell movement 8.02E-11 6.976 382 Cellular Movement migration of cells 1.22E-09 6.736 340 Cellular Movement invasion of tumor cell lines 1.04E-17 6.509 225 Cellular Movement cell movement of tumor cell lines 2.39E-14 5.488 274 Cellular Movement migration of tumor cell lines 2.20E-13 5.488 233 Cellular Assembly and Organization organization of cytoplasm 6.73E-12 5.134 216 Cellular Assembly and Organization organization of cytoskeleton 8.39E-11 5.134 191 Cellular Assembly and Organization microtubule dynamics 1.37E-09 5.131 151 Cell Death and Survival cell viability of cervical cancer cell lines 2.14E-05 4.895 72 Cell Signaling protein kinase cascade 2.63E-06 4.214 80 Cellular Growth and Proliferation proliferation of cells 1.24E-15 4.197 776 Cellular Assembly and Organization development of cytoplasm 1.08E-08 4.148 83 Cellular Movement cell movement of tumor cells 2.73E-06 4.07 29 Cell Morphology formation of cellular protrusions 1.01E-08 4.054 96 Cellular Assembly and Organization formation of filaments 5.48E-09 4.053 78 Gene Expression transactivation of RNA 3.00E-06 3.991 132 Cell Signaling I-kappaB kinase/NF-kappaB cascade 1.11E-04 3.893 39 Cellular Assembly and Organization formation of cytoskeleton 3.38E-10 3.851 75 Cellular Movement scattering of cells 1.09E-03 3.808 16 Cellular Movement cell movement of cancer cells 3.99E-05 3.802 24 Cell Morphology tubulation of cells 5.92E-03 3.76 32 Gene Expression transactivation 1.44E-06 3.729 138 Cellular Movement cell movement of melanoma cell lines 6.09E-04 3.698 31 Cellular Growth and Proliferation formation of cells 2.13E-04 3.69 44 Cellular Assembly and Organization formation of actin filaments 2.06E-09 3.637 63 Cellular Movement invasion of breast cancer cell lines 1.07E-08 3.625 80 Cellular Growth and Proliferation outgrowth of cells 2.02E-05 3.587 33 Cellular Movement scattering of tumor cell lines 2.14E-04 3.552 13 Cellular Movement scattering 4.40E-03 3.536 17

215

Activation # of Category Functions Annotation p-value z-score genes Cellular Assembly and Organization formation of actin stress fibers 1.24E-09 3.488 51 Cellular Movement cell movement of brain cancer cell lines 3.50E-05 3.461 42 Cellular Development proliferation of tumor cell lines 1.14E-08 3.396 433 Cell Death and Survival cell viability of breast cancer cell lines 1.93E-04 3.375 47 Cellular Movement migration of cervical cancer cell lines 1.23E-04 3.36 26 Cell-To-Cell Signaling and Interaction binding of tumor cell lines 2.90E-03 3.345 46 Cellular Movement cell movement of prostate cancer cell lines 4.75E-05 3.342 40 Cellular Function and Maintenance cellular homeostasis 1.82E-03 3.321 179 Cellular Movement cell movement of cervical cancer cell lines 8.39E-05 3.271 31 Cell Morphology formation of filopodia 9.19E-04 3.255 22 Cellular Movement cell movement of breast cancer cell lines 8.96E-12 3.242 103 Cellular Development epithelial-mesenchymal transition 2.04E-05 3.21 44 Cellular Movement migration of prostate cancer cell lines 1.18E-04 3.148 34 Cell Death and Survival cell viability of prostate cancer cell lines 1.82E-03 3.115 21 Cell-To-Cell Signaling and Interaction adhesion of breast cancer cell lines 7.29E-06 3.078 26 Cell Morphology shape change of tumor cell lines 3.15E-09 3.047 53 Cellular Development proliferation of neuronal cells 4.07E-06 3.041 36 Cellular Movement migration of breast cancer cell lines 3.84E-11 2.994 91 Cellular Movement cell movement of pancreatic cancer cell lines 7.26E-06 2.991 19 Cellular Movement migration of ovarian cancer cell lines 6.01E-03 2.954 13 Cellular Movement migration of brain cancer cell lines 3.54E-05 2.938 37 Protein Degradation degradation of protein 2.01E-03 2.87 92 Cell-To-Cell Signaling and Interaction adhesion of connective tissue cells 2.75E-03 2.842 20 Cellular Movement migration of melanoma cell lines 1.58E-03 2.816 25 Cellular Movement invasion of ovarian cancer cell lines 1.30E-05 2.771 20 DNA Replication synthesis of DNA 1.35E-05 2.764 76 Gene Expression expression of RNA 1.03E-13 2.76 456 Cellular Development proliferation of breast cancer cell lines 5.62E-04 2.704 120 Cell Morphology outgrowth of neurites 2.97E-04 2.676 26 Carbohydrate Metabolism metabolism of carbohydrate 4.21E-03 2.675 91 Cellular Movement invasion of brain cancer cell lines 1.47E-03 2.657 24 Cellular Movement migration of pancreatic cancer cell lines 1.62E-05 2.606 16 Cellular Movement haptotaxis of cells 2.39E-03 2.605 8 Cellular Development differentiation of hematopoietic progenitor cells 4.94E-04 2.601 35 epithelial-mesenchymal transition of tumor cell Cellular Development lines 1.52E-03 2.572 28 Cell-To-Cell Signaling and Interaction binding of breast cancer cell lines 9.62E-04 2.57 11 Cellular Growth and Proliferation proliferation of connective tissue cells 3.82E-05 2.558 64 Gene Expression transcription 1.16E-12 2.547 417 Cell Death and Survival cell death of kidney cells 8.52E-04 2.539 80 Cellular Assembly and Organization growth of neurites 2.15E-05 2.536 30 Amino Acid Metabolism phosphorylation of L-amino acid 2.12E-04 2.53 39 Amino Acid Metabolism phosphorylation of L-tyrosine 7.11E-04 2.53 34 Cellular Development proliferation of prostate cancer cell lines 7.82E-04 2.488 79 Cellular Development differentiation of connective tissue 2.27E-07 2.481 80 Cell Morphology cell spreading 1.53E-08 2.462 59 Cell Morphology outgrowth of plasma membrane projections 1.46E-04 2.454 27

216

Activation # of Category Functions Annotation p-value z-score genes Cell-To-Cell Signaling and Interaction formation of adherens junctions 2.34E-04 2.449 8 Cellular Movement homing of tumor cell lines 1.34E-03 2.444 36 Cellular Movement invasion of melanoma cell lines 6.85E-07 2.441 32 Cell Morphology formation of invadopodia 1.47E-03 2.433 9 Energy Production beta-oxidation of fatty acid 4.05E-03 2.414 10 Gene Expression transcription of RNA 3.39E-13 2.404 410 epithelial-mesenchymal transition of prostate Cellular Development cancer cell lines 2.39E-03 2.395 8 Cell-To-Cell Signaling and Interaction adhesion of epithelial cell lines 1.21E-03 2.382 15 Cell Death and Survival cell death of kidney cell lines 1.00E-03 2.365 78 Cellular Development differentiation of hematopoietic cells 2.35E-04 2.362 37 Cellular Movement transendothelial migration of tumor cell lines 6.40E-05 2.36 13 Cell Cycle mitogenesis 1.15E-03 2.347 27 Cell Death and Survival apoptosis of embryonic cell lines 3.91E-03 2.323 51 Cell Death and Survival apoptosis of kidney cell lines 2.27E-03 2.322 59 Cell Morphology cell spreading of tumor cell lines 1.79E-08 2.296 39 Cell-To-Cell Signaling and Interaction adhesion of colon cancer cell lines 2.74E-04 2.231 19 Cellular Movement invasion of colon cancer cell lines 8.33E-05 2.228 29 Cell-To-Cell Signaling and Interaction adhesion of tumor cell lines 7.52E-07 2.221 88 Cell-mediated Immune Response migration of thymocytes 4.66E-03 2.219 5 Cellular Movement haptotaxis of tumor cell lines 8.61E-04 2.215 7 Cellular Development differentiation of cells 1.01E-05 2.215 212 Cellular Movement cell movement of breast cell lines 2.18E-03 2.213 23 Cell Morphology cell spreading of fibrosarcoma cell lines 1.60E-03 2.213 5 Cell-To-Cell Signaling and Interaction adhesion of embryonic cell lines 2.75E-03 2.201 14 epithelial-mesenchymal transition of epithelial Cellular Development cell lines 1.60E-03 2.198 5 Cell Death and Survival apoptosis of epithelial cell lines 2.92E-03 2.196 53 Cellular Movement scattering of breast cancer cell lines 4.66E-03 2.194 5 Cell-To-Cell Signaling and Interaction adhesion of kidney cells 6.02E-04 2.177 23 Protein Degradation degradation of Gelatin 1.60E-03 2.169 5 Cell Death and Survival cell death of embryonic cell lines 2.47E-03 2.162 69 Cell-To-Cell Signaling and Interaction adhesion of prostate cancer cell lines 4.35E-03 2.16 13 Cell Death and Survival cell death of epithelial cell lines 2.33E-03 2.141 71 Cellular Movement invasion of fibrosarcoma cell lines 9.70E-04 2.086 12 Cell Death and Survival cell viability of melanoma cell lines 7.60E-04 2.079 16 Cell-To-Cell Signaling and Interaction adhesion of melanoma cell lines 5.52E-03 2.066 11 Cellular Assembly and Organization growth of plasma membrane projections 6.36E-06 2.059 32 Cellular Development proliferation of brain cancer cell lines 2.63E-05 2.026 56 Post-Translational Modification phosphorylation of protein 5.76E-07 2.025 139 Cell-To-Cell Signaling and Interaction phagocytosis of tumor cell lines 3.43E-03 2.008 12 Cell-To-Cell Signaling and Interaction signal transduction 1.12E-03 2 268

217

Appendix G: Functional annotation group enrichment of DEGs in WT MEF2B-V5 versus empty vector cells Shown are IPA cellular function annotation groups with absolute activation z-scores ≥ 2 and p-values ˂ 0.05 for the genes differentially expressed in WT MEF2B-V5 versus empty vector HEK293A cells. Positive z-scores indicate that MEF2B activity promotes the function and negative z-scores indicate that MEF2B activity opposes the function. In blue are annotation groups related to cell migration.

Activation # of Category Functions Annotation p-value z-score genes Cellular Movement migration of cells 2.18E-02 2.7 44 Cellular Movement cell movement 1.06E-02 2.55 50 Cellular Movement migration of tumor cell lines 1.93E-02 2.103 30

218

Appendix H: Motifs identified de novo in WT MEF2B-V5 ChIP-seq peak regions

De novo motifs were identified using the ChIP-seek283 implementation of HOMER on sequences within 100 bp of the centres of MEF2B-V5 ChIP-seq peaks. Only peaks identified in both replicates of ChIP-seq at a FDR of 0.05 were included for analysis. The top ten most enriched motifs are shown.

219

Appendix I: Known motifs enriched in WT MEF2B-V5 ChIP-seq peak regions

Known motifs were identified using the ChIP-seek283 implementation of HOMER on sequences within 100 bp of the centres of MEF2B-V5 ChIP-seq peaks. Only peaks identified in both replicates of ChIP-seq at a FDR of 0.05 were included for analysis. All motifs with p-values ≤ 0.01 are listed.

% of % of ChIP-seq background q-value peak regions Rank Name p-value sequences (Benjamini) containing the containing motif the motif 1 Mef2a(MADS)/HL1-Mef2a.biotin-ChIP-Seq/Homer/ 1E-776 0 35.19% 7.27% Mef2c(MADS)/GM12878-Mef2c-ChIP- 2 1E-740 0 37.07% 8.57% Seq(GSE32465)/Homer 3 Jun-AP1(bZIP)/K562-cJun-ChIP-Seq/Homer 1E-687 0 18.37% 1.70% AP-1(bZIP)/ThioMac-PU.1-ChIP- 4 1E-628 0 30.42% 6.57% Seq(GSE21512)/Homer Bach2(bZIP)/OCILy7-Bach2-ChIP- 5 1E-271 0 9.96% 1.42% Seq(GSE44420)/Homer

6 TEAD(TEA)/Fibroblast-PU.1-ChIP-Seq/Homer 1E-185 0 20.14% 7.78% TEAD4(TEA)/Tropoblast-Tead4-ChIP- 7 1E-185 0 21.16% 8.45% Seq(GSE37350)/Homer 8 CArG(MADS)/PUER-Srf-ChIP-Seq/Homer 1E-164 0 11.32% 3.10% 9 NF1(CTF)/LNCAP-NF1-ChIP-Seq/Homer 1E-149 0 9.98% 2.66% 10 c-Jun-CRE(bZIP)/K562-cJun-ChIP-Seq/Homer 1E-137 0 10.61% 3.17%

FOXA1(Forkhead)/LNCAP-FOXA1-ChIP- 11 1E-131 0 29.29% 16.12% Seq(GSE27824)/Homer

12 FOXA1(Forkhead)/MCF7-FOXA1-ChIP-Seq/Homer 1E-130 0 26.00% 13.63% 13 JunD(bZIP)/K562-JunD-ChIP-Seq/Homer 1E-117 0 4.10% 0.54% Fox:Ebox(Forkhead:HLH)/Panc1-Foxa2-ChIP- 14 1E-107 0 19.20% 9.47% Seq(GSE47459)/Homer 15 Foxo1(Forkhead)/RAW-Foxo1-ChIP-Seq/Homer 1E-87 0 31.66% 20.26% 16 Atf1(bZIP)/K562-ATF1-ChIP-Seq(GSE31477)/Homer 1E-87 0 15.06% 7.23%

FOXP1(Forkhead)/H9-FOXP1-ChIP- 17 1E-85 0 11.57% 4.90% Seq(GSE31006)/Homer

18 Gata2(Zf)/K562-GATA2-ChIP-Seq/Homer 1E-80 0 14.32% 6.94% 19 RUNX(Runt)/HPC7-Runx1-ChIP-Seq/Homer 1E-76 0 12.82% 6.05% 20 Gata4(Zf)/Heart-Gata4-ChIP-Seq(GSE35151)/Homer 1E-76 0 19.98% 11.35% 21 Foxa2(Forkhead)/Liver-Foxa2-ChIP-Seq/Homer 1E-76 0 16.55% 8.74% 22 Gata1(Zf)/K562-GATA1-ChIP-Seq/Homer 1E-73 0 12.80% 6.13% 23 RUNX1(Runt)/Jurkat-RUNX1-ChIP-Seq/Homer 1E-71 0 17.43% 9.61% 24 Tlx?/NPC-H3K4me1-ChIP-Seq/Homer 1E-67 0 7.76% 3.00% 25 NF-E2(bZIP)/K562-NFE2-ChIP-Seq/Homer 1E-64 0 2.71% 0.45% 26 HEB?/mES-Nanog-ChIP-Seq/Homer 1E-62 0 8.95% 3.89%

220

% of % of ChIP-seq background q-value peak regions Rank Name p-value sequences (Benjamini) containing the containing motif the motif 27 GATA3(Zf)/iTreg-Gata3-ChIP-Seq(GSE20898)/Homer 1E-62 0 27.83% 18.61% RUNX2(Runt)/PCa-RUNX2-ChIP- 28 1E-62 0 14.12% 7.54% Seq(GSE33889)/Homer 29 RUNX-AML(Runt)/CD4+-PolII-ChIP-Seq/Homer 1E-61 0 12.42% 6.34% 30 TATA-Box(TBP)/Promoter/Homer 1E-60 0 24.12% 15.59% 31 Bach1(bZIP)/K562-Bach1-ChIP-Seq(GSE31477)/Homer 1E-55 0 2.41% 0.42% 32 MafK(bZIP)/C2C12-MafK-ChIP-Seq(GSE36030)/Homer 1E-54 0 4.76% 1.53%

NF1:FOXA1/LNCAP-FOXA1-ChIP- 33 1E-53 0 2.64% 0.53% Seq(GSE27824)/Homer

Six1(Homeobox)/Myoblast-Six1-ChIP- 34 1E-52 0 5.93% 2.25% Chip(GSE20150)/Homer 35 HOXA9/HSC-Hoxa9-ChIP-Seq(GSE33509)/Homer 1E-50 0 14.68% 8.53% Nrf2(bZIP)/Lymphoblast-Nrf2-ChIP- 36 1E-48 0 2.06% 0.35% Seq(GSE37589)/Homer 37 AP-2alpha(AP2)/Hela-AP2alpha-ChIP-Seq/Homer 1E-47 0 9.55% 4.82% 38 Hoxc9/Ainv15-Hoxc9-ChIP-Seq/Homer 1E-45 0 11.90% 6.62% 39 AP2gamma(AP2)/MCF7-TFAP2c-ChIP-Seq/Homer 1E-43 0 10.05% 5.34% 40 EBF1(EBF)/Near-E2A-ChIP-Seq/Homer 1E-40 0 10.96% 6.18% 41 MafA(bZIP)/Islet-MafA-ChIP-Seq(GSE30298)/Homer 1E-39 0 11.16% 6.36% 42 Cdx2(Homeobox)/mES-Cdx2-ChIP-Seq/Homer 1E-37 0 16.15% 10.48% 43 Unknown/Homeobox/Limb-p300-ChIP-Seq/Homer 1E-36 0 14.67% 9.32% 44 EBF(EBF)/proBcell-EBF-ChIP-Seq/Homer 1E-34 0 3.13% 1.04% 45 Fli1(ETS)/CD8-FLI-ChIP-Seq(GSE20898)/Homer 1E-33 0 13.42% 8.50% 46 NF1-halfsite(CTF)/LNCaP-NF1-ChIP-Seq/Homer 1E-32 0 21.25% 15.23% 47 Sox3(HMG)/NPC-Sox3-ChIP-Seq(GSE33059)/Homer 1E-31 0 25.80% 19.35% 48 CRE(bZIP)/Promoter/Homer 1E-27 0 2.98% 1.11% 49 Sox2(HMG)/mES-Sox2-ChIP-Seq/Homer 1E-27 0 13.65% 9.13% 50 Bcl6(Zf)/Liver-Bcl6-ChIP-Seq(GSE31578)/Homer 1E-27 0 17.41% 12.34% Sox6(HMG)/Myotubes-Sox6-ChIP- 51 1E-24 0 23.71% 18.14% Seq(GSE32627)/Homer 52 ERG(ETS)/VCaP-ERG-ChIP-Seq/Homer 1E-24 0 18.70% 13.70% 53 ETV1(ETS)/GIST48-ETV1-ChIP-Seq/Homer 1E-22 0 15.39% 11.02% 54 ETS1(ETS)/Jurkat-ETS1-ChIP-Seq/Homer 1E-21 0 12.13% 8.32% 55 Elk4(ETS)/Hela-Elk4-ChIP-Seq(GSE31477)/Homer 1E-20 0 5.88% 3.34% EWS:ERG-fusion(ETS)/CADO_ES1-EWS:ERG-ChIP- 56 1E-20 0 11.08% 7.53% Seq/Homer 57 Pdx1(Homeobox)/Islet-Pdx1-ChIP-Seq/Homer 1E-20 0 16.00% 11.74% 58 GABPA(ETS)/Jurkat-GABPa-ChIP-Seq/Homer 1E-18 0 9.96% 6.77% EWS:FLI1-fusion(ETS)/SK_N_MC-EWS:FLI1-ChIP- 59 1E-17 0 7.59% 4.86% Seq/Homer 60 SCL/HPC7-Scl-ChIP-Seq/Homer 1E-17 0 45.91% 40.27%

221

% of % of ChIP-seq background q-value peak regions Rank Name p-value sequences (Benjamini) containing the containing motif the motif 61 Ets1-distal(ETS)/CD4+-PolII-ChIP-Seq/Homer 1E-16 0 4.34% 2.39% Tbx20(T-box)/Heart-Tbx20-ChIP- 62 1E-16 0 3.20% 1.60% Seq(GSE29636)/Homer 63 Elk1(ETS)/Hela-Elk1-ChIP-Seq(GSE31477)/Homer 1E-16 0 5.53% 3.35% HOXD13(Homeobox)/Chicken-Hoxd13-ChIP- 64 1E-16 0 20.60% 16.36% Seq(GSE38910)/Homer 65 SPDEF(ETS)/VCaP-SPDEF-ChIP-Seq/Homer 1E-15 0 11.50% 8.34% PAX3:FKHR-fusion(Paired/Homeobox)/Rh4- 66 1E-14 0 4.01% 2.29% PAX3:FKHR-ChIP-Seq/Homer 67 CTCF(Zf)/CD4+-CTCF-ChIP-Seq/Homer 1E-12 0 1.30% 0.48% 68 Hoxb4/ES-Hoxb4-ChIP-Seq(GSE34014)/Homer 1E-12 0 3.44% 1.95% GATA-IR4(Zf)/iTreg-Gata3-ChIP- 69 1E-11 0 1.61% 0.71% Seq(GSE20898)/Homer 70 Stat3(Stat)/mES-Stat3-ChIP-Seq/Homer 1E-11 0 5.73% 3.86%

GATA-IR3(Zf)/iTreg-Gata3-ChIP- 71 1E-11 0 3.29% 1.92% Seq(GSE20898)/Homer

72 Arnt:Ahr(bHLH)/MCF7-Arnt-ChIP-Seq(Lo et al.)/Homer 1E-10 0 6.13% 4.20% 73 Atf4(bZIP)/MEF-Atf4-ChIP-Seq(GSE35681)/Homer 1E-10 0 3.62% 2.21% 74 Maz(Zf)/HepG2-Maz-ChIP-Seq(GSE31477)/Homer 1E-09 0 8.44% 6.27% 75 Stat3+il23(Stat)/CD4-Stat3-ChIP-Seq/Homer 1E-09 0 8.03% 5.92% 76 HOXA2(Homeobox)/mES-Hoxa2-ChIP-Seq/Homer 1E-09 0 1.88% 0.98% 77 Rfx1(HTH)/NPC-Rfx1-ChIP-Seq/Homer 1E-08 0 2.35% 1.33% Foxh1(Forkhead)/hESC-FOXH1-ChIP- 78 1E-08 0 8.84% 6.75% Seq(GSE29422)/Homer 79 PPARE(NR/DR1)/3T3L1-Pparg-ChIP-Seq/Homer 1E-08 0 9.33% 7.18% 80 NFAT:AP1/Jurkat-NFATC1-ChIP-Seq/Homer 1E-08 0 2.78% 1.67% 81 NFkB-p65(RHD)/GM12787-p65-ChIP-Seq/Homer 1E-08 0 5.46% 3.86% 82 Ap4(HLH)/AML-Tfap4-ChIP-Seq(GSE45738)/Homer 1E-08 0 11.79% 9.43% 83 Nur77(NR)/K562-NR4A1-ChIP-Seq(GSE31363)/Homer 1E-08 0 2.80% 1.70% 84 BMYB(HTH)/Hela-BMYB-ChIPSeq(GSE27030)/Homer 1E-08 0 18.16% 15.30% 85 Chop(bZIP)/MEF-Chop-ChIP-Seq(GSE35681)/Homer 1E-08 0 2.73% 1.66% 86 NFkB-p65-Rel(RHD)/LPS-exp(GSE23622)/Homer 1E-08 0 1.14% 0.50% 87 ELF5(ETS)/T47D-ELF5-ChIP-Seq(GSE30407)/Homer 1E-07 0 8.68% 6.71%

Lhx3-like?(Homeobox)/Forebrain-p300-ChIP- 88 1E-07 0 16.08% 13.44% Seq/Homer

89 STAT4(Stat)/CD4-Stat4-ChIP-Seq/Homer 1E-07 0 12.04% 9.75% 90 RFX(HTH)/K562-RFX3-ChIP-Seq/Homer 1E-07 0 0.94% 0.39% Smad3(MAD)/NPC-Smad3-ChIP- 91 1E-07 0 30.22% 26.88% Seq(GSE36673)/Homer 92 MyoD(HLH)/Myotube-MyoD-ChIP-Seq/Homer 1E-07 0 7.20% 5.44% 93 Tcf12(HLH)/GM12878-Tcf12-ChIP-Seq/Homer 1E-07 0 8.84% 6.90% 94 ELF1(ETS)/Jurkat-ELF1-ChIP-Seq/Homer 1E-07 0 4.39% 3.08%

222

% of % of ChIP-seq background q-value peak regions Rank Name p-value sequences (Benjamini) containing the containing motif the motif 95 NFAT(RHD)/Jurkat-NFATC1-ChIP-Seq/Homer 1E-07 0 12.28% 10.07% 96 PU.1(ETS)/ThioMac-PU.1-ChIP-Seq(GSE21512)/Homer 1E-07 0 5.50% 4.03% 97 ETS(ETS)/Promoter/Homer 1E-06 0 3.02% 1.97% 98 X-box(HTH)/NPC-H3K4me1-ChIP-Seq/Homer 1E-06 0 1.43% 0.77% 99 TR4(NR/DR1)/Hela-TR4-ChIP-Seq/Homer 1E-06 0 1.18% 0.60% 100 Nanog(Homeobox)/mES-Nanog-ChIP-Seq/Homer 1E-06 0 51.37% 48.09% 101 MYB(HTH)/ERMYB-Myb-ChIPSeq(GSE22095)/Homer 1E-06 0 19.42% 16.94% 102 ETS:E-box/HPC7-Scl-ChIP-Seq/Homer 1E-06 0 1.19% 0.62% 103 RXR(NR/DR1)/3T3L1-RXR-ChIP-Seq/Homer 1E-06 0 9.89% 8.09% 104 BORIS(Zf)/K562-CTCFL-ChIP-Seq/Homer 1E-05 0 1.34% 0.73% 105 GATA:SCL/Ter119-SCL-ChIP-Seq/Homer 1E-05 0 1.75% 1.06% 106 Smad2(MAD)/ES-SMAD2-ChIP-Seq(GSE29422)/Homer 1E-05 0 14.50% 12.48% 107 STAT5(Stat)/mCD4+-Stat5a|b-ChIP-Seq/Homer 1E-05 0 4.25% 3.16% 108 Esrrb(NR)/mES-Esrrb-ChIP-Seq/Homer 1E-04 0 6.51% 5.23%

Smad4(MAD)/ESC-SMAD4-ChIP- 109 1E-04 0.0001 14.09% 12.28% Seq(GSE29422)/Homer

110 p53(p53)/Saos-p53-ChIP-Seq/Homer 1E-03 0.0005 1.10% 0.68% 111 Atoh1(bHLH)/Cerebellum-Atoh1-ChIP-Seq/Homer 1E-02 0.0025 9.75% 8.58% 112 Myf5(bHLH)/GM-Myf5-ChIP-Seq(GSE24852)/Homer 1E-02 0.0025 6.24% 5.30% MyoG(HLH)/C2C12-MyoG-ChIP- 113 1E-02 0.0026 8.95% 7.83% Seq(GSE36024)/Homer 114 ERE(NR/IR3)/MCF7-ERa-ChIP-Seq/Homer 1E-02 0.003 2.28% 1.73% 115 p53(p53)/p53-ChIP-Chip/Homer 1E-02 0.0031 0.18% 0.06% 116 STAT1(Stat)/HelaS3-STAT1-ChIP-Seq/Homer 1E-02 0.0035 3.71% 3.01% 117 MafF(bZIP)/HepG2-MafF-ChIP-Seq(GSE31477)/Homer 1E-02 0.0051 4.47% 3.73% 118 E2F6(E2F)/Hela-E2F6-ChIP-Seq(GSE31477)/Homer 1E-02 0.006 2.88% 2.30% 119 ATF3(bZIP)/K562-ATF3-ChIP-Seq/Homer 1E-02 0.0061 1.21% 0.85% 120 PRDM9(Zf)/Testis-DMC1-ChIP-Seq(GSE35498)/Homer 1E-02 0.0062 4.09% 3.40% 121 ZFX(Zf)/mES-Zfx-ChIP-Seq/Homer 1E-02 0.0073 10.00% 8.95% GATA-DR8(Zf)/iTreg-Gata3-ChIP- 122 1E-02 0.0087 0.99% 0.68% Seq(GSE20898)/Homer PAX5-shortForm(Paired/Homeobox)/GM12878-PAX5- 123 1E-02 0.0095 0.98% 0.67% ChIP-Seq/Homer 124 EGR(Zf)/K562-EGR1-ChIP-Seq/Homer 1E-02 0.0102 0.90% 0.61% 125 E2F4(E2F)/K562-E2F4-ChIP-Seq(GSE31477)/Homer 1E-02 0.0142 2.24% 1.79% 126 NFY(CCAAT)/Promoter/Homer 1E-02 0.0174 7.83% 7.01%

223

Appendix J: The 261 high confidence candidate direct MEF2B target genes High confidence candidate direct MEF2B target genes were defined as genes that (i) had increased expression in WT MEF2B-V5 versus untransfected cells (adjusted p-values < 0.05 in microarray data), (ii) had peaks within 1 Mb up or downstream of their TSSs in both WT MEF2B-V5 ChIP-seq replicates, (iii) had rank product scores < 0.05, (iv) were associated with at least one peak identified at an irreproducible discovery rate of < 0.01 and (v) had log2 fold changes in expression in RNA-seq data for WT MEF2B-V5 versus empty vector cells that were > 0.3. Rank product scores were calculated using BETA286 and indicate the relative likelihood of each gene being a direct MEF2B target. Genes with greater fold changes in expression and ChIP-seq peaks closer to their TSSs have lower rank product scores and are more likely to be direct MEF2B targets. Statistics shown for ChIP-seq peaks are from replicate 1 data. Gene functions were obtained using IPA. In total, 59 genes had annotated functions related to cell migration, 4 genes had annotated functions related to EMT, 86 genes had annotated functions related to cell growth and proliferation and 95 genes had annotated functions related to cell death and survival.

224

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment ANK1 5.5E-03 0.19 0.0488 98398 19.39 2.6E-41 yes 0.494 ANKRD1 1.3E-06 2.14 0.0001 -290 22.91 1.9E-51 yes yes 4.756 ANTXR2 8.4E-04 0.62 0.0019 -65 5.80 1.9E-07 yes 0.374 ANXA3 1.4E-03 0.97 0.0013 17807 6.43 1.3E-05 yes 1.416 ARFGAP2 2.0E-03 0.56 0.0012 228506 13.61 4.5E-33 0.654 ARHGAP29 1.0E-02 0.52 0.0096 -88076 12.39 7.4E-18 0.337 ASAP2 1.0E-03 0.57 0.0009 28128 17.17 1.3E-31 yes 0.667 ATF3 6.7E-05 0.45 0.0059 61 7.00 8.2E-07 yes yes yes 0.382 ATP2B4 1.4E-04 1.00 0.0001 -68307 19.41 1.9E-35 1.138 ATP8B1 1.9E-04 0.81 0.0026 13819 5.95 3.5E-06 0.838 BACH1 7.0E-03 0.41 0.0056 -77342 17.39 1.5E-30 yes 0.326 BACH2 2.8E-03 0.66 0.0017 -72373 8.35 1.8E-09 yes yes 0.362 BASP1 1.2E-03 0.56 0.0008 88460 15.91 1.8E-40 1.158 BCL2L1 2.6E-03 0.47 0.0020 44028 12.64 1.6E-24 yes yes 0.500 BHLHE40 8.9E-05 0.90 0.0001 7633 13.55 2.8E-25 yes yes 0.490 BMP5 1.1E-03 0.52 0.0010 -27573 5.46 1.9E-04 yes yes 3.466 BNC2 1.9E-03 0.71 0.0011 -83420 18.83 9.3E-34 0.526 C22orf24 2.1E-02 0.25 0.0497 -75060 19.03 1.7E-38 1.051 CA5B 1.1E-03 0.66 0.0012 13128 18.35 3.2E-32 0.970 CACNG4 1.2E-04 1.04 0.0002 -12621 5.37 4.3E-08 2.934 CALD1 2.6E-03 0.39 0.0053 8457 13.63 1.3E-21 0.508 CARS 1.7E-03 0.71 0.0017 -15776 12.76 6.4E-25 yes 0.363 CAST 5.3E-03 0.25 0.0299 39157 8.54 1.1E-10 yes yes yes 0.325 CAV2 6.8E-03 0.39 0.0166 12555 6.80 2.1E-07 yes 0.487 CCBE1 8.4E-05 1.20 0.0001 90613 6.50 2.1E-07 0.801 CCDC3 5.5E-04 0.77 0.0004 55573 21.04 7.8E-41 0.845 CCDC80 3.5E-03 0.51 0.0026 -40792 4.31 1.5E-02 yes 0.391 CD44 3.9E-03 0.44 0.0026 0.470 -55061 7.57 3.3E-16 yes yes yes yes CDC42EP3 3.0E-04 0.92 0.0002 -62960 6.04 6.8E-06 1.482 CDCA2 6.4E-03 0.52 0.0056 50083 7.58 4.5E-08 yes yes 0.528 CDKN2B 1.7E-02 0.22 0.0414 -43159 17.68 3.1E-41 yes 1.054 CITED2 3.1E-04 0.34 0.0140 -195142 7.27 9.0E-09 yes yes 0.578

225

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment CLDN1 5.1E-03 0.43 0.0035 90244 12.19 7.4E-18 0.538 CLIC5 5.5E-05 1.03 0.0002 30510 14.31 1.9E-22 1.735 COL5A1 1.4E-03 2.08 0.0000 28381 7.18 6.7E-15 1.385 CPEB4 9.6E-03 0.30 0.0207 35578 6.13 3.6E-05 0.468 CREB3 3.6E-03 0.33 0.0090 -3962 28.73 6.6E-61 yes yes 0.503 CRIM1 3.0E-03 0.45 0.0034 15059 16.02 3.2E-32 0.853 CTGF 1.3E-05 1.39 0.0004 3.168 1647 7.99 2.4E-10 yes yes yes yes CTSB 1.8E-03 0.80 0.0002 -23466 15.66 1.2E-25 yes yes yes 0.875 CX3CR1 8.9E-03 0.48 0.0181 20637 13.43 1.7E-23 yes yes 0.373 CYR61 2.3E-04 0.76 0.0010 -2131 8.13 2.8E-11 yes yes yes 1.147 DACT1 3.8E-03 0.48 0.0024 93659 8.14 2.7E-14 0.690 DAP 9.4E-03 0.28 0.0125 36156 9.50 9.5E-16 yes yes 0.308 DCLK1 9.8E-04 0.70 0.0008 -32673 46.39 6.9E-122 0.661 DDX60 3.5E-03 0.74 0.0039 -17016 13.92 1.9E-22 0.767 DNAJC15 8.2E-03 0.45 0.0360 -17517 9.89 1.8E-13 yes 1.295 DNMT3A 2.6E-03 0.51 0.0019 -35151 16.37 7.4E-29 0.710 DSC3 2.3E-03 0.53 0.0050 3113 12.96 2.3E-19 0.494 DTNB 1.0E-02 0.28 0.0155 54762 8.69 1.2E-10 yes 0.311 DUSP10 2.9E-04 1.04 0.0003 23458 13.16 6.8E-21 yes 1.367 DUSP5 2.0E-03 0.26 0.0410 -6308 16.05 1.3E-43 yes yes 0.934 DYRK1A 1.1E-02 0.34 0.0213 32158 15.01 6.4E-24 yes 0.444 EBAG9 4.1E-03 0.49 0.0079 34788 27.96 5.0E-59 yes 0.475 EDN1 1.1E-02 0.43 0.0134 81527 25.81 3.5E-86 yes yes yes 4.706 EFEMP1 1.4E-04 1.14 0.0001 38510 7.24 1.7E-08 yes yes 1.386 ENC1 3.6E-03 0.60 0.0019 9088 21.92 1.3E-44 yes yes 0.680 ERAP1 5.4E-03 0.35 0.0111 -15966 8.35 2.3E-09 0.371 ERC1 9.4E-03 0.31 0.0171 23122 7.95 2.9E-13 yes 0.324 ETAA1 3.8E-04 0.49 0.0278 3019 9.69 4.6E-12 0.805 EYA1 6.9E-04 0.86 0.0004 55280 7.96 2.3E-09 0.578 FAM110B 6.3E-03 0.37 0.0050 73448 12.96 2.3E-19 0.407 FAM126A 1.2E-02 0.39 0.0159 -62926 21.21 1.1E-42 0.542 FAM84B 6.9E-03 0.38 0.0113 -21584 22.44 4.2E-46 0.682

226

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment FLNA 2.3E-03 0.55 0.0015 49060 48.28 5.2E-149 yes yes yes 1.239 FN1 8.6E-04 1.84 0.0000 -24768 22.71 2.3E-45 yes yes yes 3.128 FNBP1 6.7E-03 0.31 0.0092 25766 7.47 3.3E-10 0.408 FNDC3B 1.1E-02 0.31 0.0134 57573 16.23 2.9E-27 0.358 GATA6 1.6E-02 0.25 0.0247 -70561 6.94 2.1E-07 yes yes yes 0.525 GJC1 4.1E-03 0.38 0.0232 7834 18.84 2.9E-41 0.440 GMPR 6.2E-04 0.67 0.0009 -32955 6.83 2.8E-07 0.326 GNA12 3.4E-03 0.61 0.0022 59162 12.82 1.3E-21 yes 0.447 GOLPH3 1.4E-02 0.19 0.0441 -19446 17.46 2.2E-62 yes 0.327 GPC1 8.8E-03 0.42 0.0081 -65430 25.27 1.3E-51 yes yes 0.586 GPRC5A 3.7E-06 1.35 0.0001 -7503 9.42 7.4E-15 2.489 GSN 3.2E-04 0.31 0.0107 -813 10.75 6.4E-17 yes yes 0.515 HIPK3 1.6E-02 0.40 0.0309 50595 11.48 3.2E-18 yes 0.591 HS1BP3 5.1E-03 0.45 0.0036 71694 10.29 6.8E-21 0.727 ID2 2.4E-03 0.58 0.0049 -4368 14.50 4.9E-24 yes yes yes 1.875 ID3 1.0E-02 0.25 0.0262 -17757 14.93 1.9E-28 yes yes yes 1.409 IGFBP7 2.8E-05 0.59 0.0011 -8717 13.83 1.2E-26 yes yes 1.007 IL6ST 6.9E-03 0.45 0.0058 -63547 10.90 6.4E-17 yes yes 0.574 IL7R 4.1E-03 0.21 0.0458 5155 11.82 1.4E-34 yes yes 0.915 INPP5A 6.4E-03 0.40 0.0246 -20367 19.42 2.0E-45 yes 0.303 ITGB8 6.0E-04 0.75 0.0010 20947 7.97 1.3E-09 yes yes 1.117 KALRN 1.4E-02 0.68 0.0004 -87353 12.15 1.8E-17 1.208 KAT5 4.1E-03 0.43 0.0053 -15976 9.88 7.8E-19 yes yes 0.395 KCNS3 3.9E-04 1.12 0.0003 98391 8.61 1.5E-11 1.633 KIAA0319 1.9E-03 0.27 0.0148 62056 11.04 3.2E-18 yes 0.635 KIF18A 1.5E-02 0.51 0.0239 58653 10.27 1.8E-13 yes 0.549 KLF12 2.7E-03 0.40 0.0064 4085 7.73 5.0E-08 0.366 KLF3 6.8E-03 0.37 0.0065 -43006 6.80 2.1E-07 0.318 KLF4 5.9E-03 0.50 0.0084 17957 9.38 1.3E-13 yes yes yes 0.372 KLF6 1.4E-04 0.60 0.0013 -204 6.68 2.9E-10 yes yes 0.478 KLHL5 4.2E-04 0.45 0.0040 -1339 10.50 3.6E-18 0.480 KLHL7 5.2E-03 0.64 0.0054 -28657 21.21 1.1E-42 0.543

227

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment LGR4 9.4E-04 0.48 0.0099 3945 8.70 5.9E-12 0.726 LIFR 1.8E-05 0.64 0.0014 -142 10.80 4.4E-19 yes yes 0.964 LMCD1 1.1E-03 0.23 0.0316 -2524 13.92 1.9E-22 yes 0.888 LMNA 3.5E-04 0.65 0.0008 11745 18.49 1.4E-35 yes yes 0.998 LOXL4 1.0E-02 0.26 0.0443 9422 17.32 9.6E-48 yes 0.541 LPIN2 2.5E-03 0.39 0.0051 -17888 6.13 7.7E-06 0.385 LRP12 1.1E-02 0.49 0.0120 -68040 6.23 1.3E-05 0.505 LYST 4.1E-03 0.47 0.0113 20897 6.23 1.3E-05 0.333 MALT1 2.6E-03 0.88 0.0017 118076 19.40 1.2E-48 yes 1.147 MAML2 8.6E-04 0.78 0.0015 4170 8.27 2.8E-11 yes 0.718 MAP2 5.1E-04 0.90 0.0004 -53684 10.66 6.6E-15 0.827 MAP3K8 4.2E-03 0.30 0.0147 -2493 11.14 3.5E-19 yes yes 0.590 MAP4K3 3.8E-03 0.43 0.0127 -2814 6.36 2.4E-06 0.525 MAPK4 1.1E-03 0.72 0.0007 69213 13.81 6.7E-22 1.023 MDH1 6.8E-03 0.28 0.0488 2004 20.08 2.4E-37 0.307 MED10 1.2E-02 0.41 0.0206 -41207 11.84 2.1E-23 0.315 MEIS1 3.0E-03 0.65 0.0019 -70739 33.35 2.4E-74 0.803 MFGE8 3.4E-04 0.34 0.0042 1496 6.75 6.7E-07 0.370 MGLL 2.8E-03 0.41 0.0098 5539 10.22 1.6E-33 1.137 MICAL2 1.5E-04 0.86 0.0002 -23790 4.20 2.4E-04 yes 0.951 MITF 7.0E-06 0.37 0.0037 22589 7.39 4.5E-08 yes yes yes 0.749 MN1 2.1E-03 0.55 0.0013 -64787 9.89 3.7E-13 0.671 MSN 6.6E-03 0.37 0.0078 -32025 21.00 4.9E-46 yes yes 0.405 MYO10 2.5E-03 0.42 0.0024 -28768 18.32 5.7E-48 yes 0.916 MYO9B 5.6E-03 0.36 0.0042 59824 13.15 2.3E-26 0.390 NARS 2.0E-02 0.19 0.0444 -77059 15.50 1.3E-29 yes 0.463 NAV2 5.0E-04 0.71 0.0007 76107 6.03 2.7E-05 0.599 NCKAP5 1.8E-02 0.34 0.0320 -65636 14.16 2.1E-23 1.314 NCOA2 3.2E-03 0.60 0.0021 -53601 11.43 2.3E-16 0.439 NDRG4 2.9E-03 0.37 0.0060 31635 16.02 3.2E-32 0.967 NECAB1 1.9E-02 0.28 0.0366 67745 13.35 6.8E-21 1.344 NEDD4 1.4E-02 0.50 0.0192 93746 25.46 1.3E-51 yes 0.522

228

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment NEDD4L 6.6E-05 0.54 0.0030 99185 8.21 3.3E-12 0.744 NEDD9 2.3E-04 0.87 0.0010 25246 12.72 1.5E-22 yes yes yes 1.735 NFATC1 2.2E-04 0.67 0.0005 -6652 5.70 1.9E-07 yes yes yes 0.903 NFIX 4.9E-03 0.38 0.0046 41014 15.87 9.7E-31 yes 0.436 NIPBL 1.6E-02 0.31 0.0304 -44225 7.90 9.8E-10 0.549 NOG 7.6E-03 1.01 0.0063 88180 7.00 8.2E-07 yes yes 1.293 NRP1 1.2E-02 0.33 0.0174 0.507 43818 6.67 1.2E-07 yes yes yes yes NSFL1C 6.1E-03 0.47 0.0048 67602 15.30 5.9E-26 0.603 NUAK1 7.3E-04 0.37 0.0186 -5548 5.93 2.0E-05 yes yes 0.496 OLFM1 6.2E-06 1.87 0.0000 62357 33.29 5.1E-97 2.706 OPHN1 4.7E-03 0.50 0.0162 4842 7.87 1.9E-09 0.689 OSBPL5 3.3E-03 0.42 0.0020 92125 12.76 6.4E-25 0.737 OSMR 3.3E-04 1.15 0.0003 -38034 14.99 4.7E-31 yes 1.281 PAG1 1.1E-03 0.80 0.0010 22336 6.78 2.9E-10 0.849 PALLD 1.4E-04 0.48 0.0023 -114 17.78 1.0E-36 yes 0.594 PAMR1 8.7E-05 1.11 0.0002 2017 17.54 2.4E-37 2.760 PAPPA 1.3E-04 0.77 0.0004 3188 46.92 7.9E-139 1.273 PAPSS2 7.1E-03 0.40 0.0065 64091 16.77 3.9E-29 yes 0.485 PBK 2.3E-03 0.83 0.0017 -34909 13.09 1.3E-23 yes yes yes 1.089 PC 6.4E-03 0.31 0.0090 -74288 15.73 8.1E-32 0.449 PCDH18 8.3E-04 0.57 0.0009 -13147 37.96 3.4E-88 yes 0.900 PCLO 6.0E-03 0.60 0.0054 43161 11.63 1.5E-19 0.421 PCYT1B 2.0E-03 0.69 0.0012 152433 23.14 2.4E-69 yes 0.978 PDE4B 1.1E-03 0.45 0.0030 -265881 43.35 1.6E-106 yes 1.330 PDE4D 3.5E-04 0.43 0.0023 244449 4.31 1.5E-02 yes yes 0.339 PDP1 1.9E-04 0.72 0.0004 18325 22.17 7.7E-51 yes 0.751 PEX13 9.5E-03 0.40 0.0335 -10745 11.04 6.6E-15 0.400 PGPEP1 1.2E-02 0.22 0.0299 24634 11.56 1.3E-25 0.581 PHKA2 8.5E-04 0.62 0.0006 45489 10.90 6.4E-17 yes 1.254 PHLDB2 1.5E-03 1.70 0.0000 5666 31.07 3.4E-71 2.124 PHLPP1 4.0E-04 0.78 0.0008 -4784 6.80 2.1E-07 yes yes 0.844 PIK3IP1 5.7E-03 0.23 0.0345 -1344 10.82 3.0E-18 0.634

229

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment PLCB4 2.6E-03 0.80 0.0086 24206 9.72 2.2E-14 0.664 PLEKHH2 6.5E-03 0.37 0.0384 -2433 11.30 1.0E-17 0.438 PLS3 3.3E-03 0.40 0.0069 21771 17.31 9.5E-37 yes 1.369 PMAIP1 7.5E-05 0.53 0.0040 -320 5.26 4.4E-04 yes 1.242 PMEPA1 8.0E-04 0.72 0.0006 -39003 7.23 4.9E-09 yes yes 0.529 PPP1R15A 1.1E-03 0.49 0.0027 767 37.55 7.7E-105 yes yes 0.526 PRICKLE1 1.3E-04 0.71 0.0003 -1533 9.63 3.6E-13 0.851 PRICKLE2 2.8E-04 0.56 0.0007 -13756 9.12 1.1E-10 0.702 PRR5L 3.5E-03 0.42 0.0054 45000 7.82 1.3E-09 0.791 PRSS23 2.1E-05 1.00 0.0001 -9976 11.27 2.0E-24 1.690 PTGER4 3.7E-03 0.45 0.0060 9724 9.12 1.6E-17 yes yes 0.564 RAB9A 5.4E-03 0.48 0.0079 -24923 13.12 4.3E-22 0.579 RANBP3L 6.7E-04 0.29 0.0301 -415 21.67 1.7E-46 1.609 RASSF8 1.3E-02 0.28 0.0314 38785 17.49 3.4E-35 yes 0.311 RASSF9 4.0E-03 0.36 0.0179 905 12.58 2.3E-19 0.324 RCAN2 2.1E-04 0.64 0.0004 53491 8.55 5.9E-12 1.095 RGS6 6.4E-03 0.33 0.0055 53675 8.17 1.8E-09 1.269 RHOB 2.0E-04 0.44 0.0018 -67 16.34 8.4E-34 yes yes yes 0.560 RHOBTB3 4.6E-03 0.26 0.0256 -18254 13.62 8.8E-23 0.412 RHOD 4.6E-04 0.81 0.0013 -11359 14.87 1.2E-27 yes yes 1.023 RHOU 2.6E-03 0.50 0.0016 95849 6.65 2.1E-07 yes 0.573 RIC3 4.3E-03 0.37 0.0087 7112 19.72 1.2E-39 3.101 ROCK2 1.3E-02 0.30 0.0434 -30199 5.67 6.2E-09 yes yes 0.594 RPS6KA2 9.6E-04 0.67 0.0008 36250 11.98 3.6E-17 yes yes 1.054 RUNX1 3.5E-03 0.39 0.0044 -78762 7.56 3.2E-12 yes yes 1.029 RUNX2 2.8E-04 0.57 0.0009 -85447 9.12 1.1E-10 yes yes 0.681 RYBP 4.4E-03 0.46 0.0036 -36385 7.09 1.7E-08 yes yes 0.523 S1PR1 7.1E-04 0.51 0.0015 8651 12.50 1.1E-20 yes yes yes 0.922 SAT1 7.8E-04 0.25 0.0404 10230 6.27 6.1E-11 yes yes 0.645 SBF2 1.1E-02 0.42 0.0332 30073 8.07 2.5E-09 0.319 SCARA5 1.3E-02 0.26 0.0333 -23636 6.23 1.3E-05 1.176 SEMA3A 2.4E-03 0.87 0.0014 70519 31.54 4.0E-83 yes yes 0.479

230

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment SEMA3C 1.2E-03 0.72 0.0013 35724 11.92 1.5E-19 0.974 SERP2 1.5E-02 0.26 0.0232 -72750 10.99 4.0E-16 1.211 SF3A2 1.0E-02 0.30 0.0137 -50656 13.78 1.6E-28 yes 0.363 SGK1 3.1E-03 0.35 0.0070 -4291 6.06 2.4E-06 yes yes yes 0.816 SH3GL2 4.5E-03 0.31 0.0062 22427 19.88 1.9E-40 0.530 SIPA1L2 2.2E-03 0.43 0.0027 -23424 6.50 3.2E-08 0.419 SLC1A2 2.1E-04 1.11 0.0005 75 6.62 8.2E-07 2.214 SLCO5A1 1.6E-02 0.31 0.0381 -36798 6.81 8.2E-07 0.807 SMAD6 5.7E-03 0.38 0.0059 46102 15.53 9.7E-29 yes yes 0.896 SMAD7 8.7E-04 0.63 0.0024 -9975 10.49 2.6E-14 yes yes yes 1.147 SMPD1 2.7E-03 0.70 0.0017 -57070 16.31 3.2E-32 yes yes 1.449 SMPX 1.5E-02 0.28 0.0488 -40881 13.12 8.4E-32 0.931 SNAP25 9.6E-03 0.29 0.0087 95591 14.41 2.0E-26 1.652 SOCS6 1.5E-03 0.94 0.0009 92935 6.23 1.3E-05 0.989 SOX6 2.1E-03 0.57 0.0014 45590 6.94 2.1E-07 0.863 SPARC 2.0E-04 0.68 0.0004 -5582 10.60 6.4E-17 yes yes yes 1.178 SPDYA 1.3E-02 0.22 0.0348 25149 18.51 9.1E-40 yes yes 0.422 SPTBN1 2.8E-03 0.41 0.0032 15365 14.41 2.0E-26 0.760 STARD4 5.6E-03 0.50 0.0067 -36677 11.17 4.0E-16 0.609 STK11 2.9E-03 0.29 0.0081 1894 18.77 1.6E-70 yes yes yes 0.363 SUN1 8.6E-04 0.36 0.0062 -617 13.74 1.7E-23 0.371 SVIL 9.6E-05 0.42 0.0033 -69 6.39 2.6E-09 yes 0.550 SWAP70 6.6E-03 0.43 0.0062 -50822 28.06 3.0E-97 0.512 SYNE1 5.9E-04 0.54 0.0025 1193 14.31 2.9E-24 0.934 SYTL5 2.3E-03 0.58 0.0029 27061 10.19 6.8E-21 0.778 TACC1 6.0E-04 0.56 0.0009 6016 14.34 2.1E-23 0.695 TBC1D1 4.4E-03 0.42 0.0044 41060 10.21 8.1E-14 0.469 TBX3 2.1E-03 0.27 0.0237 -136 7.98 3.6E-11 yes yes yes 0.386 TCEANC 1.5E-03 0.73 0.0018 11010 13.12 4.3E-22 0.533 TEAD1 9.6E-03 0.35 0.0140 32945 7.85 3.8E-10 0.539 TES 7.1E-04 0.50 0.0016 7916 14.69 4.9E-24 yes 1.025 TFAP2A 2.1E-03 0.54 0.0012 -96416 22.33 2.0E-45 yes yes yes 0.679

231

Differential Annotated Annotated expression in WT Log2 fold Peak closest to TSS that was Annotated Annotated functions functions Rank MEF2B-V5 vs change in WT identified in both WT MEF2B-V5 functions Gene functions related to related to product untransfected cells MEF2B-V5 vs ChIP-seq replicates related to symbol related to cell growth cell death score (microarray data) empty vector cell cells EMT and and Log2 fold Adjusted Distance Fold movement (RNA-seq data) q-value proliferation survival change p-value from TSS enrichment TFAP2C 9.9E-04 0.88 0.0007 -62461 8.34 2.3E-15 yes yes yes 0.920 TGIF1 2.0E-03 0.26 0.0136 -42304 13.09 1.3E-23 0.664 TGM2 3.4E-03 0.42 0.0028 1.875 38031 15.29 2.8E-29 yes yes yes yes TINAGL1 4.9E-03 0.35 0.0239 7746 19.19 3.2E-37 1.669 TM4SF1 5.3E-03 0.35 0.0346 10649 12.58 2.3E-19 2.407 TMEM55A 5.6E-03 0.62 0.0038 97973 9.69 4.6E-12 0.622 TNFRSF21 7.1E-03 0.40 0.0056 -98180 5.47 3.4E-04 yes yes 0.338 TNS3 1.5E-03 0.66 0.0009 93041 13.45 1.7E-22 yes 0.457 TOM1L2 3.4E-03 0.39 0.0037 13854 12.02 4.6E-18 0.470 TRAF6 3.8E-04 0.26 0.0143 828 14.19 1.6E-22 yes yes 0.442 TRIM44 5.4E-03 0.34 0.0068 20288 7.01 5.7E-08 0.361 TRIO 1.4E-05 0.50 0.0016 3700 9.24 1.4E-14 yes yes 0.537 UGP2 6.1E-03 0.49 0.0119 -13948 10.75 6.4E-17 0.555 USE1 5.9E-03 0.38 0.0043 -79740 13.15 2.3E-26 0.508 VAT1L 2.0E-04 1.06 0.0003 -8074 17.81 1.3E-31 4.036 VCL 3.1E-04 0.35 0.0080 10259 22.04 1.2E-52 yes 0.425 VGLL3 2.2E-03 0.45 0.0497 2642 8.26 9.4E-11 0.686 VLDLR 1.7E-04 1.30 0.0002 -18833 17.78 1.0E-36 1.483 WEE1 4.4E-04 0.33 0.0112 1943 39.33 7.0E-96 yes yes 0.612 WWP1 5.6E-03 0.34 0.0064 -37240 10.08 1.8E-13 0.492 YAP1 1.4E-03 0.36 0.0069 10203 6.67 2.1E-06 yes yes yes 0.736 ZBTB16 3.4E-03 0.40 0.0164 -16295 8.50 1.2E-10 yes yes yes 0.649 ZC3H6 7.6E-04 0.99 0.0006 26727 7.77 4.5E-08 0.723 ZCCHC2 1.6E-03 0.66 0.0029 2930 6.58 2.6E-09 0.795 ZFAND2A 6.5E-03 0.38 0.0056 -50289 24.77 6.4E-60 0.501 ZFX 4.3E-04 0.31 0.0288 -1449 5.78 5.3E-07 0.444 ZNF85 4.1E-03 0.60 0.0226 -107 17.03 9.6E-48 yes 0.911

232

Appendix K: Functional annotation group enrichment in the candidate direct MEF2B target genes Shown are IPA cellular function annotation groups with absolute activation z-scores ≥ 2 and p-values ˂ 0.05 for the 1,141 candidate direct target genes. Positive z-scores indicate that MEF2B activity promotes the function and negative z-scores indicate that MEF2B activity opposes the function. In blue are annotation groups related to cell migration, in yellow are annotation groups related to epithelial-mesenchymal transition, and in pink are annotation groups related to cell death, survival and proliferation.

Activation # of Category Functions Annotation p-value z-score genes Cell Death and Survival cell survival 1.28E-14 8.227 138 Cell Death and Survival cell viability 1.04E-15 8.201 136 Cell Death and Survival cell viability of tumor cell lines 1.08E-14 7.607 113 Cellular Movement migration of cells 2.35E-25 6.292 180 Cellular Movement cell movement 3.52E-27 6.012 199 Cellular Movement invasion of cells 6.93E-24 5.157 124 Cellular Movement invasion of tumor cell lines 1.59E-20 5.11 109 Cellular Movement migration of tumor cell lines 4.33E-18 4.954 112 Cellular Movement cell movement of tumor cell lines 8.56E-22 4.7 135 Cellular Movement cell movement of brain cancer cell lines 2.18E-06 4.29 22 Cell Death and Survival cell viability of cervical cancer cell lines 9.68E-07 4.266 35 Cellular Movement homing 3.21E-05 4.145 44 Cellular Movement homing of cells 1.03E-04 4.145 41 Cell Morphology tubulation of cells 4.97E-08 3.983 24 Cellular Movement migration of brain cancer cell lines 2.56E-05 3.868 18 Cellular Movement chemotaxis 4.65E-05 3.822 43 Cellular Movement chemotaxis of cells 1.39E-04 3.822 40 Cellular Assembly and Organization formation of filaments 6.83E-11 3.634 43 Cellular Assembly and Organization microtubule dynamics 2.22E-11 3.532 77 Cellular Assembly and Organization organization of cytoskeleton 1.08E-12 3.523 98 Cellular Assembly and Organization organization of cytoplasm 4.93E-12 3.523 106 Cell Death and Survival cell viability of breast cancer cell lines 2.98E-05 3.52 23 Cell Signaling protein kinase cascade 2.60E-06 3.511 39 Cellular Movement cell movement of tumor cells 3.62E-08 3.504 18 Cellular Growth and Proliferation proliferation of cells 2.54E-26 3.445 332 Gene Expression transactivation of RNA 1.20E-09 3.442 63 Gene Expression transactivation 3.69E-10 3.42 66 Cellular Movement cell movement of cancer cells 2.83E-06 3.364 14 Cellular Function and Maintenance cellular homeostasis 1.86E-06 3.353 84 Cell Morphology formation of cellular protrusions 2.48E-11 3.262 56 Cellular Assembly and Organization formation of actin filaments 5.21E-12 3.244 38 Gene Expression transcription 1.22E-19 3.233 196 Gene Expression expression of RNA 1.17E-20 3.223 213 Cell Death and Survival cell viability of prostate cancer cell lines 3.49E-05 3.077 13 Cellular Assembly and Organization development of cytoplasm 3.15E-11 3.068 47 Gene Expression transcription of RNA 8.43E-20 3.047 193 Cellular Development epithelial-mesenchymal transition 6.87E-07 3.002 24 233

Activation # of Category Functions Annotation p-value z-score genes Amino Acid Metabolism phosphorylation of L-amino acid 5.17E-07 3 25 Amino Acid Metabolism phosphorylation of L-tyrosine 2.39E-05 3 20 Cell Morphology cell spreading 1.76E-08 2.968 29 Cellular Movement invasion of breast cancer cell lines 1.50E-09 2.967 40 Cellular Movement cell movement of prostate cancer cell lines 2.90E-06 2.959 21 Cellular Assembly and Organization formation of actin stress fibers 8.63E-11 2.953 30 Cellular Movement migration of leukemia cell lines 4.42E-05 2.927 15 Cellular Assembly and Organization formation of cytoskeleton 2.74E-12 2.919 43 Gene Expression activation of DNA endogenous promoter 6.30E-13 2.916 112 Cellular Movement migration of breast cancer cell lines 1.70E-14 2.874 50 Cell Morphology shape change of tumor cell lines 3.20E-07 2.868 24 Cellular Movement cell movement of melanoma cell lines 3.43E-07 2.862 20 Cellular Development proliferation of tumor cell lines 1.94E-22 2.8 208 Cell Signaling synthesis of nitric oxide 1.40E-04 2.77 16 Cell-To-Cell Signaling and Interaction adhesion of breast cancer cell lines 1.68E-07 2.73 16 epithelial-mesenchymal transition of tumor cell Cellular Development lines 6.15E-05 2.716 16 Cellular Movement migration of prostate cancer cell lines 3.97E-05 2.679 17 Cellular Development proliferation of neuronal cells 3.82E-08 2.678 21 Cellular Movement cell movement of breast cancer cell lines 7.76E-15 2.65 55 Cell-To-Cell Signaling and Interaction adhesion of connective tissue cells 7.29E-08 2.64 16 Cellular Movement cell movement of cervical cancer cell lines 9.86E-06 2.631 17 Cellular Movement invasion of melanoma cell lines 5.46E-07 2.611 17 Cellular Growth and Proliferation outgrowth of cells 3.90E-04 2.608 14 Cell-To-Cell Signaling and Interaction binding of cells 2.25E-04 2.603 39 Cell-To-Cell Signaling and Interaction adhesion of colon cancer cell lines 1.27E-06 2.6 13 Cell Morphology cell spreading of tumor cell lines 4.12E-07 2.58 19 Cell Signaling I-kappaB kinase/NF-kappaB cascade 3.22E-04 2.548 19 Cell-To-Cell Signaling and Interaction adhesion of tumor cell lines 1.51E-10 2.542 47 Gene Expression expression of DNA 2.88E-16 2.509 160 Cellular Movement cell movement of leukemia cell lines 5.75E-05 2.459 21 Cellular Development proliferation of tumor cells 1.05E-09 2.451 36 Cell Morphology electrical resistance of cells 4.89E-04 2.412 6 Cellular Movement cell movement of pancreatic cancer cell lines 1.79E-05 2.402 11 Cell Morphology formation of filopodia 7.32E-05 2.401 13 Cell Signaling entrance of Ca2+ 2.47E-04 2.4 8 Lipid Metabolism concentration of phospholipid 3.56E-04 2.327 12 Cellular Development proliferation of cancer cells 9.06E-10 2.292 33 Cellular Movement migration of ovarian cancer cell lines 5.14E-04 2.28 8 Cellular Movement migration of melanoma cell lines 1.04E-06 2.276 17 Cellular Movement migration of cervical cancer cell lines 1.09E-05 2.252 15 Cell Morphology reorganization of cytoskeleton 1.27E-04 2.246 14 Cell Morphology cell spreading of embryonic cell lines 1.60E-04 2.236 5 Cell Morphology cell spreading of kidney cell lines 1.60E-04 2.236 5 Cell-To-Cell Signaling and Interaction formation of adherens junctions 3.72E-06 2.236 7 Cell Morphology outgrowth of neurites 1.69E-04 2.234 13 Cell Morphology cell spreading of epithelial cell lines 2.79E-04 2.219 5 Cellular Movement migration of pancreatic cancer cell lines 8.58E-05 2.178 9 Cellular Movement transmigration of cells 1.77E-05 2.173 16 Cell-To-Cell Signaling and Interaction signal transduction 3.45E-05 2.121 98 Cellular Assembly and Organization growth of neurites 2.26E-05 2.12 15 Cellular Development differentiation of cells 3.05E-09 2.104 100 Cell-To-Cell Signaling and Interaction communication 7.49E-06 2.087 110

234

Activation # of Category Functions Annotation p-value z-score genes Cell-To-Cell Signaling and Interaction communication of cells 1.80E-05 2.087 103 Cell-To-Cell Signaling and Interaction formation of focal adhesions 2.22E-07 2.069 19 Cellular Movement transendothelial migration of tumor cell lines 2.69E-05 2.041 8 Cellular Development differentiation of connective tissue 3.00E-10 2.012 46 Cell-To-Cell Signaling and Interaction adhesion of kidney cells 7.50E-04 2.003 11 Cell Death and Survival cell death of tumor cells 4.21E-07 -2.018 33 Cell Death and Survival apoptosis of ovarian cancer cell lines 2.33E-04 -2.228 16 Cell Death and Survival apoptosis of endothelial cells 4.19E-04 -2.274 16 Cell Death and Survival apoptosis of tumor cells 5.09E-06 -2.297 26 Cell Death and Survival necrosis of vascular endothelial cells 1.48E-04 -2.527 14 Cell Death and Survival cell death of ovarian cancer cell lines 4.98E-04 -2.572 18 Cell Death and Survival cell death of endothelial cells 1.22E-04 -2.594 18

235

Appendix L: The 361 genes differentially expressed in comparisons of K4E MEF2B-V5, Y69H MEF2B-V5, D83V MEF2B-V5 and untransfected cells to WT MEF2B-V5 cells. All genes in this list were (i) differentially expressed in K4E MEF2B-V5, Y69H MEF2B-V5, D83V MEF2B-V5 and untransfected cells compared to WT MEF2B-V5 cells (adjusted p-values < 0.05 for all comparisons) and (ii) had the same direction of expression change in all comparisons to WT MEF2B-V5 cells. Data were produced using expression microarrays. Statistics for TGFB1 are shown in bold as TGFB1 is discussed in section 4.2.1.

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value ABCA2 -0.221 3.49E-02 -0.293 1.22E-02 -0.376 1.52E-02 -0.36 7.89E-03 ABCG1 -0.421 3.92E-03 -0.409 7.00E-03 -0.442 1.67E-02 -0.496 3.52E-03 ABL1 -0.381 4.43E-03 -0.322 9.03E-03 -0.509 8.14E-03 -0.399 5.36E-03 ACTR1B -0.887 1.35E-03 -0.459 1.43E-02 -0.461 3.51E-02 -0.735 4.09E-03 ADAMTS5 -1.145 3.50E-04 -1.137 4.27E-04 -0.77 1.15E-02 -1.351 1.95E-04 ADAP1 -0.515 1.66E-03 -0.424 4.79E-03 -0.382 2.19E-02 -0.336 1.25E-02 AES -0.25 2.20E-02 -0.267 2.04E-02 -0.243 3.65E-02 -0.369 5.02E-03 AF269286 0.701 7.83E-04 0.691 1.14E-03 0.441 1.83E-02 0.692 1.44E-02 AK023447 0.294 8.54E-03 0.255 3.12E-02 0.268 4.65E-02 0.228 3.72E-02 AKAP1 0.33 6.96E-03 0.364 7.77E-03 0.349 2.03E-02 0.352 6.52E-03 ALDH1A2 1.79 2.68E-04 1.912 2.19E-04 1.409 6.78E-03 1.62 2.60E-04 AMOT -1.182 6.68E-05 -0.851 3.00E-04 -0.291 4.88E-02 -1.417 2.52E-05 AMPH -0.762 4.25E-04 -0.782 8.32E-04 -0.793 6.69E-03 -0.634 1.59E-03 ANTXR2 -0.619 1.89E-03 -0.321 2.35E-02 -0.454 3.18E-02 -0.377 3.42E-02 AP4M1 -0.273 3.23E-02 -0.25 3.09E-02 -0.337 2.78E-02 -0.518 3.45E-03 APLF -0.719 8.77E-04 -0.43 9.22E-03 -0.388 2.65E-02 -0.595 2.09E-03 ARMCX1 -0.554 9.36E-04 -0.871 2.60E-04 -0.548 6.58E-03 -0.305 9.99E-03 ARNT2 -0.483 1.62E-03 -0.361 1.33E-02 -0.297 2.82E-02 -0.547 1.17E-03 ASRGL1 0.326 1.96E-02 0.298 3.46E-02 0.38 3.86E-02 0.313 4.56E-02 ATP13A1 -0.305 7.73E-03 -0.212 3.77E-02 -0.344 1.47E-02 -0.29 1.79E-02 ATP2B4 -1.001 8.83E-05 -0.685 6.60E-04 -0.481 9.31E-03 -1.032 2.85E-04 ATP5G1 0.254 1.51E-02 0.306 9.95E-03 0.296 2.53E-02 0.413 2.99E-03 ATP9B -1.027 5.22E-04 -0.708 2.16E-03 -0.528 2.69E-02 -1.305 1.59E-04 BACE1 -0.313 1.12E-02 -0.314 9.56E-03 -0.262 3.26E-02 -0.357 4.54E-03 BCAM -0.548 1.33E-03 -0.55 1.53E-03 -0.349 2.06E-02 -0.461 2.80E-03 BHLHE40 -0.898 1.31E-04 -0.432 3.36E-03 -0.333 2.56E-02 -0.968 1.85E-04 BMP1 -0.498 1.11E-03 -0.286 2.03E-02 -0.391 1.15E-02 -0.517 2.46E-03 BNC2 -0.708 1.11E-03 -0.428 8.69E-03 -0.479 2.35E-02 -0.535 4.81E-03 BTBD2 -0.392 6.22E-03 -0.53 1.61E-03 -0.272 2.92E-02 -0.419 2.53E-02 C14orf101 -0.449 7.46E-03 -0.397 1.58E-02 -0.372 4.10E-02 -0.481 6.28E-03 C15orf38-AP3S2 -0.589 1.11E-03 -0.622 5.71E-04 -0.333 2.18E-02 -0.889 1.48E-04 C16orf58 -0.706 4.17E-04 -0.441 3.68E-03 -0.379 1.51E-02 -0.633 7.52E-04

236

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value C1QBP 0.275 1.38E-02 0.25 2.35E-02 0.292 2.81E-02 0.394 3.49E-03 C1RL -0.466 3.32E-03 -0.349 2.22E-02 -0.371 2.42E-02 -0.554 3.03E-03 C20orf96 -0.748 7.34E-04 -0.439 1.10E-02 -0.378 3.15E-02 -0.84 7.51E-04 C4orf34 -0.525 6.22E-03 -0.547 3.10E-03 -0.498 2.53E-02 -1.013 4.02E-04 C5orf4 -0.418 2.67E-03 -0.595 1.64E-03 -0.68 4.85E-03 -0.894 2.27E-04 C6orf183 0.235 2.81E-02 0.259 2.81E-02 0.338 2.34E-02 0.27 2.31E-02 C6orf57 -0.415 2.67E-02 -0.486 1.72E-02 -0.654 1.18E-02 -0.554 1.20E-02 C8orf40 -0.557 3.58E-03 -0.38 1.09E-02 -0.418 2.20E-02 -0.903 2.85E-04 CA2 0.733 8.41E-04 1.437 1.08E-04 2.118 9.58E-05 1.716 6.30E-05 CACNB3 -0.87 3.56E-04 -0.627 1.11E-03 -0.287 4.56E-02 -0.942 2.84E-04 CADPS2 1.763 2.17E-04 1.789 1.58E-04 0.631 4.41E-02 2.003 9.23E-05 CALHM2 -0.626 4.63E-04 -0.447 3.02E-03 -0.267 3.58E-02 -0.505 1.11E-03 CAV1 -0.68 3.50E-04 -0.448 3.68E-03 -0.463 1.01E-02 -0.397 1.52E-02 CC2D1A -0.332 5.52E-03 -0.32 8.04E-03 -0.257 3.26E-02 -0.366 4.39E-03 CCBE1 -1.197 6.08E-05 -0.462 7.65E-03 -0.713 4.37E-03 -1.813 1.03E-05 CCDC3 -0.77 4.10E-04 -0.488 5.39E-03 -0.261 4.76E-02 -0.517 2.16E-03 CCDC92 -0.408 1.17E-02 -0.293 1.97E-02 -0.348 3.42E-02 -0.345 1.11E-02 CD97 -0.415 2.63E-03 -0.399 6.15E-03 -0.314 2.40E-02 -0.441 3.29E-03 CD99L2 -0.312 5.52E-03 -0.229 2.99E-02 -0.244 3.30E-02 -0.42 2.32E-03 CDKN1A -0.423 3.83E-03 -0.256 3.28E-02 -0.376 2.26E-02 -0.372 1.38E-02 CEP72 -0.421 2.95E-03 -0.34 1.01E-02 -0.253 4.78E-02 -0.527 5.41E-03 CEP89 -0.361 5.81E-03 -0.262 4.74E-02 -0.291 3.95E-02 -0.359 9.24E-03 CHAC1 -0.691 1.23E-03 -0.458 3.80E-03 -0.408 2.16E-02 -0.429 3.59E-03 CHRDL1 1.07 8.65E-05 0.686 8.70E-04 1.083 1.68E-03 2.058 1.44E-05 CIRBP -0.284 8.43E-03 -0.525 1.52E-03 -0.288 2.53E-02 -0.707 2.85E-04 CLSTN2 -0.735 9.34E-04 -0.467 3.58E-03 -0.426 1.51E-02 -0.788 4.02E-04 CNDP2 -1.032 1.74E-04 -0.501 3.47E-03 -0.648 4.85E-03 -0.841 5.96E-04 CNIH3 -0.682 3.09E-03 -0.266 3.96E-02 -0.525 1.36E-02 -0.759 6.14E-04 CNN2 -0.667 8.54E-04 -0.269 3.47E-02 -0.51 1.13E-02 -0.538 2.06E-03 CNPY4 -0.42 1.18E-02 -0.32 2.62E-02 -0.383 4.18E-02 -0.389 2.01E-02 CNTD1 -0.598 3.86E-03 -0.56 5.22E-03 -0.365 4.48E-02 -0.802 8.29E-04 COL5A1 -0.614 1.06E-03 -0.655 5.76E-04 -0.409 1.34E-02 -0.559 8.92E-04 CORO2B -0.754 4.63E-04 -0.514 2.63E-03 -0.303 4.74E-02 -0.925 2.09E-04 CPS1 -1.58 4.19E-05 -1.267 1.58E-04 -0.879 7.91E-03 -1.274 1.08E-04 CPSF1 -0.217 2.84E-02 -0.344 5.66E-03 -0.455 8.14E-03 -0.276 2.78E-02 CREB3L2 -0.514 1.04E-03 -0.449 2.92E-03 -0.254 3.87E-02 -0.455 3.11E-03 CRK 0.618 7.24E-04 0.682 6.46E-04 0.406 1.83E-02 0.573 1.64E-03 CTIF -0.472 1.21E-03 -0.351 5.94E-03 -0.314 1.95E-02 -0.378 5.63E-03 CTSB -0.799 2.25E-04 -0.482 1.87E-03 -0.307 2.34E-02 -1.055 6.49E-05 CUL7 -0.673 4.23E-04 -0.644 1.01E-03 -0.459 8.14E-03 -0.753 2.84E-04 CXXC1 -0.539 7.02E-04 -0.269 2.67E-02 -0.307 1.90E-02 -0.557 9.63E-04 CYB561 0.619 2.62E-03 0.666 1.83E-03 1.002 1.39E-03 0.29 3.04E-02 CYP11A1 -0.905 1.77E-04 -0.384 2.36E-02 -0.326 2.56E-02 -1.197 3.35E-05 CYP2S1 -0.678 3.74E-04 -0.451 2.15E-03 -0.231 3.94E-02 -0.584 6.02E-04

237

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value CYP4X1 -0.794 4.10E-04 -0.888 9.35E-04 -0.428 1.69E-02 -1.019 7.91E-05 CYP51A1 -0.494 6.70E-03 -0.34 2.53E-02 -0.339 4.50E-02 -0.352 2.68E-02 CYTH3 -0.549 1.35E-03 -0.357 9.31E-03 -0.377 2.14E-02 -0.378 1.25E-02 DCAF15 -0.486 3.48E-03 -0.445 5.66E-03 -0.324 4.41E-02 -0.58 3.56E-03 DCLK1 -0.704 7.57E-04 -0.556 2.63E-03 -0.351 4.85E-02 -0.827 2.97E-04 DERL3 -0.969 7.81E-05 -0.555 1.11E-03 -0.448 8.14E-03 -1.005 6.30E-05 DFNA5 -0.797 1.02E-03 -0.762 1.64E-03 -0.432 3.51E-02 -0.896 1.32E-03 DGAT2 -0.412 9.10E-03 -0.415 3.74E-03 -0.408 1.14E-02 -0.488 2.28E-03 DHPS -0.51 1.01E-03 -0.41 5.70E-03 -0.393 1.51E-02 -0.632 9.98E-04 DHX33 0.297 1.28E-02 0.439 3.98E-03 0.306 2.69E-02 0.445 3.56E-03 DNASE2 -0.405 3.44E-03 -0.241 3.29E-02 -0.262 4.01E-02 -0.438 3.14E-03 DNMT3A -0.512 1.89E-03 -0.457 3.43E-03 -0.32 2.56E-02 -0.26 3.04E-02 DPH1 0.556 1.21E-03 0.616 1.52E-03 0.455 1.35E-02 0.554 1.39E-03 DPYSL2 -0.703 1.21E-03 -0.395 7.70E-03 -0.275 3.19E-02 -0.874 1.85E-04 EDNRA 0.55 1.19E-03 0.643 2.12E-03 0.363 3.60E-02 0.77 6.87E-04 EFEMP2 -1.709 2.16E-05 -1.257 5.53E-05 -0.273 3.30E-02 -1.802 6.57E-06 EFHA2 -0.694 1.70E-03 -0.436 2.16E-02 -0.519 2.75E-02 -0.781 2.08E-03 ELAC1 -0.645 2.47E-03 -0.516 1.98E-02 -0.657 1.55E-02 -0.873 7.92E-04 ENPP2 1.777 2.18E-04 1.57 2.38E-04 0.785 2.53E-02 2.29 2.52E-05 EPB49 -0.636 1.93E-03 -0.436 5.88E-03 -0.347 2.53E-02 -0.601 9.98E-04 EPC1 -0.413 3.71E-03 -0.309 3.64E-02 -0.345 3.34E-02 -0.384 8.46E-03 ERBB4 0.932 3.56E-04 0.874 4.75E-04 0.44 3.48E-02 1.35 1.88E-04 ERMAP -0.455 2.31E-03 -0.367 7.22E-03 -0.466 8.17E-03 -0.733 8.81E-04 ERVMER34-1 1.217 8.65E-05 1.81 4.55E-05 1.162 1.32E-03 1.456 1.65E-04 ESYT1 -0.356 6.51E-03 -0.273 2.47E-02 -0.274 4.18E-02 -0.302 1.44E-02 EYA1 -0.856 4.19E-04 -0.669 1.39E-03 -0.56 1.33E-02 -1.197 6.68E-05 FABP6 0.704 4.23E-04 0.45 5.46E-03 0.316 2.81E-02 1.47 1.22E-05 FAM50B 0.351 9.59E-03 0.28 3.48E-02 0.309 4.18E-02 0.313 3.36E-02 FAM57A 0.641 5.43E-04 0.69 4.69E-04 0.317 2.81E-02 0.623 5.85E-04 FBXO17 -0.448 2.88E-03 -0.423 3.79E-03 -0.27 3.67E-02 -0.263 2.57E-02 FBXO25 -0.7 3.96E-04 -0.443 7.18E-03 -0.36 3.44E-02 -0.879 3.72E-04 FBXO27 -0.518 1.19E-03 -0.321 1.30E-02 -0.254 3.37E-02 -0.291 1.42E-02 FBXO4 -0.528 2.63E-03 -0.293 3.62E-02 -0.459 2.17E-02 -1.103 8.92E-04 FBXW5 -0.356 3.91E-03 -0.261 1.89E-02 -0.288 2.62E-02 -0.334 1.01E-02 FKBP8 -0.595 3.12E-03 -0.437 1.80E-02 -0.468 2.96E-02 -0.595 5.06E-03 FNBP1 -0.314 9.20E-03 -0.238 3.62E-02 -0.284 3.58E-02 -0.23 3.79E-02 FOSL2 -0.739 3.48E-04 -0.435 5.58E-03 -0.258 3.12E-02 -0.582 5.57E-04 FSCN1 -0.469 1.40E-02 -0.488 7.41E-03 -0.361 2.56E-02 -0.46 4.35E-03 FSTL3 -0.447 8.91E-03 -0.457 3.98E-03 -0.388 2.98E-02 -0.321 2.16E-02 FZR1 -0.521 1.03E-03 -0.428 2.92E-03 -0.276 2.81E-02 -0.524 1.20E-03 GAS6 -0.431 2.96E-03 -0.492 2.63E-03 -0.438 1.33E-02 -0.639 7.97E-04 GDI1 -0.389 4.78E-03 -0.282 1.90E-02 -0.249 4.88E-02 -0.471 3.24E-03 GENSCAN00000002731 -0.675 4.78E-04 -0.494 1.92E-03 -0.244 3.44E-02 -0.724 2.85E-04 GFRA2 -0.924 2.68E-04 -0.491 2.97E-03 -0.279 4.02E-02 -1.045 5.14E-05

238

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value GGA2 0.21 3.07E-02 0.324 1.22E-02 0.356 1.55E-02 0.298 1.22E-02 GLOD4 0.564 2.89E-03 0.695 1.32E-03 0.366 3.37E-02 0.572 1.22E-03 GNA12 -0.607 2.15E-03 -0.601 2.39E-03 -0.38 2.52E-02 -0.546 3.63E-03 GNAI2 -0.438 4.06E-03 -0.302 1.78E-02 -0.272 3.60E-02 -0.403 4.96E-03 GPAA1 -0.474 2.71E-03 -0.348 9.14E-03 -0.491 1.14E-02 -0.433 5.74E-03 GPR108 -0.464 2.23E-03 -0.407 3.86E-03 -0.254 3.65E-02 -0.602 6.20E-04 GPR124 -0.389 5.79E-03 -0.523 3.49E-03 -0.349 2.20E-02 -0.318 1.60E-02 GPRC5B 0.389 2.62E-03 0.433 3.14E-03 0.301 2.30E-02 0.488 1.40E-03 GPRIN3 0.41 5.59E-03 0.448 1.04E-02 0.278 3.34E-02 0.381 2.68E-02 GRINA -0.388 7.20E-03 -0.284 2.55E-02 -0.3 4.01E-02 -0.404 1.04E-02 GTF2A1L -0.696 1.74E-03 -0.402 1.04E-02 -0.38 2.75E-02 -0.525 3.51E-03 H6PD -0.616 2.31E-03 -0.621 2.63E-03 -0.537 1.68E-02 -0.816 1.97E-03 HCN2 -0.779 9.33E-04 -0.519 3.14E-03 -0.417 2.53E-02 -0.668 7.42E-04 HDAC6 -0.488 1.30E-03 -0.227 3.91E-02 -0.315 1.85E-02 -0.429 3.35E-03 HGSNAT -0.745 1.79E-03 -0.586 2.77E-03 -0.424 2.96E-02 -1.154 2.84E-04 HIST1H3A -0.49 1.32E-03 -0.269 1.92E-02 -0.31 2.82E-02 -0.358 1.09E-02 HIST1H3I -0.749 8.33E-04 -0.547 1.64E-03 -0.292 4.41E-02 -0.772 2.97E-04 HLA-DRB1 0.467 4.93E-02 0.817 3.16E-03 0.772 2.78E-02 0.991 1.36E-03 HMG20B -0.545 1.17E-03 -0.466 1.92E-03 -0.419 1.15E-02 -0.371 4.21E-03 HNMT -0.752 4.95E-04 -0.356 3.05E-02 -0.381 4.32E-02 -0.272 4.76E-02 HOXB9 0.322 7.11E-03 0.209 3.88E-02 0.252 3.32E-02 0.34 6.48E-03 HTR1D -0.443 2.00E-03 -0.276 4.34E-02 -0.386 2.26E-02 -0.321 9.11E-03 IER3 -1.061 3.07E-04 -0.403 1.22E-02 -0.655 1.11E-02 -0.616 4.15E-03 IFI30 0.459 1.77E-03 0.695 5.57E-04 0.334 2.46E-02 0.443 2.37E-03 IKZF4 -0.215 4.80E-02 -0.233 4.09E-02 -0.255 4.34E-02 -0.341 9.74E-03 INA -1.797 2.42E-05 -1.684 4.67E-05 -1.677 1.15E-04 -2.21 2.16E-05 INPP5K 0.548 1.11E-03 0.758 5.78E-04 0.442 9.11E-03 0.673 3.48E-04 INTS1 -0.445 8.46E-03 -0.455 4.35E-03 -0.409 1.32E-02 -0.334 9.74E-03 IQCE -0.365 8.04E-03 -0.442 3.72E-03 -0.252 4.61E-02 -0.327 1.21E-02 IQCK 0.406 4.15E-03 0.672 9.35E-04 0.261 4.07E-02 0.503 2.29E-03 IQGAP3 -0.708 4.68E-04 -0.41 6.18E-03 -0.488 8.14E-03 -0.852 7.24E-04 IQUB -0.823 5.32E-04 -0.688 1.83E-03 -0.443 3.18E-02 -0.583 5.19E-03 ISYNA1 -0.397 2.31E-03 -0.286 1.13E-02 -0.554 4.37E-03 -0.486 2.20E-03 ITGB8 -0.753 9.72E-04 -0.589 1.97E-03 -0.389 4.66E-02 -0.6 7.47E-03 KANK2 -0.522 9.81E-04 -0.362 5.18E-03 -0.359 1.73E-02 -0.583 1.23E-03 KIAA0664 0.331 9.04E-03 0.572 1.08E-03 0.26 3.82E-02 0.482 1.61E-03 KIAA0913 -0.444 1.95E-03 -0.263 1.71E-02 -0.265 3.48E-02 -0.413 2.99E-03 KREMEN1 -0.323 1.03E-02 -0.281 1.27E-02 -0.346 1.69E-02 -0.2 4.92E-02 LAMB2 -0.705 4.12E-04 -0.554 1.58E-03 -0.533 4.85E-03 -0.767 3.64E-04 LBH -1.356 9.85E-05 -0.627 2.03E-03 -0.484 2.67E-02 -1.048 2.69E-04 LGALS3BP 0.49 1.28E-03 0.244 2.00E-02 0.39 1.11E-02 0.657 4.78E-04 LGR6 -0.564 1.43E-03 -0.43 2.96E-03 -0.524 8.34E-03 -0.573 3.01E-03 LIPG -0.782 5.47E-04 -0.247 3.04E-02 -0.231 4.01E-02 -0.828 3.72E-04 LMNA -0.653 8.20E-04 -0.374 7.27E-03 -0.32 2.26E-02 -0.438 3.91E-03

239

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value LOXL2 -0.901 9.58E-05 -0.774 3.99E-04 -0.264 2.91E-02 -0.966 5.59E-05 LRAT 0.399 5.92E-03 0.564 2.33E-03 0.295 3.81E-02 0.451 4.00E-03 LRP2 0.801 4.70E-04 0.851 4.51E-04 0.563 1.15E-02 0.859 8.81E-04 LRP5 -0.273 1.68E-02 -0.294 1.35E-02 -0.258 3.81E-02 -0.275 2.12E-02 LRRC14 -0.32 2.73E-02 -0.261 4.36E-02 -0.528 1.20E-02 -0.419 1.06E-02 LTBP3 -0.605 5.67E-04 -0.637 5.65E-04 -0.249 3.84E-02 -0.683 3.98E-04 MAGED2 -0.889 1.60E-04 -0.585 1.26E-03 -0.494 9.13E-03 -0.762 3.75E-04 MAN2B1 -0.6 5.43E-04 -0.383 3.99E-03 -0.345 1.45E-02 -0.472 1.40E-03 MAP6 -0.913 2.00E-04 -0.672 9.36E-04 -0.37 2.34E-02 -1.1 1.65E-04 MAPK4 -0.721 7.39E-04 -0.672 9.35E-04 -0.532 1.35E-02 -1.1 1.01E-04 MARK4 -0.46 1.34E-03 -0.317 7.75E-03 -0.258 3.30E-02 -0.222 4.24E-02 MCM5 -0.437 1.64E-03 -0.44 3.21E-03 -0.245 3.41E-02 -0.377 4.16E-03 MED31 0.213 2.95E-02 0.261 4.34E-02 0.341 2.04E-02 0.371 4.28E-03 MEF2C 0.26 3.38E-02 0.419 9.14E-03 0.47 2.53E-02 0.385 1.77E-02 MEX3C -0.795 3.05E-04 -0.539 3.51E-03 -0.394 1.90E-02 -0.819 2.18E-04 MIA-RAB4B -0.33 7.37E-03 -0.288 1.21E-02 -0.255 3.13E-02 -0.212 4.04E-02 MIR196B -0.463 4.22E-03 -0.333 1.85E-02 -0.414 2.91E-02 -0.463 8.76E-03 MKNK2 -0.572 2.46E-03 -0.35 8.24E-03 -0.373 1.79E-02 -0.601 9.98E-04 MN1 -0.554 1.30E-03 -0.474 2.58E-03 -0.669 4.37E-03 -0.583 9.03E-04 MOB3A -0.66 9.36E-04 -0.74 4.51E-04 -0.377 1.72E-02 -0.834 3.11E-04 MOSPD3 -0.538 1.02E-03 -0.534 3.74E-03 -0.289 3.30E-02 -0.514 2.65E-03 MPND -0.359 7.40E-03 -0.354 8.94E-03 -0.288 4.11E-02 -0.464 6.12E-03 MPZL2 0.806 7.52E-04 0.421 9.59E-03 0.928 9.09E-03 0.459 3.19E-02 MRC2 -0.256 1.40E-02 -0.36 4.77E-03 -0.352 1.53E-02 -0.243 2.03E-02 MRPS7 0.336 5.27E-03 0.306 1.12E-02 0.263 4.10E-02 0.474 1.52E-03 MSRA -0.454 2.12E-03 -0.282 2.06E-02 -0.349 2.34E-02 -0.386 5.30E-03 MTMR11 -0.564 1.10E-03 -0.312 1.23E-02 -0.278 2.81E-02 -0.595 6.29E-04 MTMR9 -0.841 4.17E-04 -0.303 2.08E-02 -0.352 4.25E-02 -0.819 2.38E-03 MTSS1 0.47 1.64E-03 0.367 7.79E-03 0.368 1.73E-02 0.529 2.22E-03 MYBBP1A 0.477 1.70E-03 0.565 1.09E-03 0.285 2.81E-02 0.509 1.34E-03 MYH9 -0.594 1.24E-03 -0.375 9.74E-03 -0.296 4.99E-02 -0.277 4.92E-02 MYO1C 0.369 7.75E-03 0.42 3.43E-03 0.313 2.19E-02 0.561 1.10E-03 N4BP2L1 -0.406 1.52E-02 -0.354 2.79E-02 -0.46 4.55E-02 -0.496 1.01E-02 NDUFAB1 0.416 1.33E-02 0.719 2.01E-03 0.409 4.57E-02 0.405 3.21E-02 NISCH -0.287 1.29E-02 -0.207 4.91E-02 -0.281 2.81E-02 -0.362 1.10E-02 NOS2 0.404 2.02E-03 0.447 3.60E-03 0.237 4.13E-02 0.496 1.23E-03 NPTX2 0.545 1.14E-03 0.86 3.83E-04 0.279 3.51E-02 0.446 4.05E-03 NRBP2 -0.482 1.19E-03 -0.553 1.49E-03 -0.431 8.14E-03 -0.552 9.67E-04 NREP -0.363 7.67E-03 -0.395 6.53E-03 -0.347 2.53E-02 -0.444 1.74E-02 NRM -0.526 1.08E-03 -0.302 1.35E-02 -0.262 3.79E-02 -0.598 1.38E-03 NTF3 -0.602 1.05E-03 -0.313 1.40E-02 -0.367 2.34E-02 -0.577 8.92E-04 NUPR1 -1.421 6.31E-05 -1.154 1.77E-04 -0.776 6.58E-03 -1.394 2.09E-04 OLFM1 -1.87 2.16E-05 -1.551 7.15E-05 -1.018 4.37E-03 -2.105 9.66E-06 ORAI3 -0.588 1.64E-03 -0.483 3.46E-03 -0.319 3.79E-02 -0.645 7.36E-04

240

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value PACSIN3 -0.337 6.01E-03 -0.247 2.55E-02 -0.228 4.42E-02 -0.265 4.63E-02 PARD6G-AS1 -0.922 2.32E-04 -0.56 1.66E-03 -0.529 1.38E-02 -1.268 5.14E-05 PARP14 -0.957 2.95E-04 -0.398 2.73E-02 -0.852 4.37E-03 -1.688 3.83E-05 PAX7 -0.529 1.11E-03 -0.41 4.56E-03 -0.279 3.40E-02 -0.421 2.53E-03 PCDH10 -0.427 2.63E-03 -0.428 3.95E-03 -0.41 1.35E-02 -0.289 3.47E-02 PCDH18 -0.573 8.98E-04 -0.571 3.04E-03 -0.338 3.07E-02 -0.361 1.03E-02 PCDHB10 -0.767 1.66E-03 -1.088 6.46E-04 -0.619 2.82E-02 -0.747 2.38E-03 PCDHB14 -0.96 1.47E-04 -0.865 1.84E-03 -1.037 3.99E-03 -0.846 5.71E-03 PCDHB15 -1.105 1.49E-03 -1.027 1.23E-03 -0.833 1.34E-02 -1.032 3.11E-03 PCDHB16 -0.646 1.89E-03 -0.975 1.45E-03 -0.61 2.83E-02 -0.69 2.48E-03 PCDHB18 -0.621 2.16E-03 -1.07 4.51E-04 -0.546 1.36E-02 -0.511 5.11E-03 PCDHB3 -0.942 1.21E-03 -0.863 1.32E-03 -0.788 3.25E-02 -0.951 1.09E-02 PCK2 -0.397 4.87E-03 -0.415 4.77E-03 -0.422 1.68E-02 -0.363 7.57E-03 PELP1 0.273 2.21E-02 0.422 4.91E-03 0.252 4.14E-02 0.347 7.10E-03 PFKL -0.445 3.86E-03 -0.342 7.80E-03 -0.268 3.49E-02 -0.593 8.81E-04 PFKP 0.33 9.13E-03 0.396 6.50E-03 0.282 3.19E-02 0.395 4.16E-03 PHLPP1 -0.778 8.24E-04 -0.42 8.42E-03 -0.42 2.96E-02 -0.903 4.42E-04 PIGL 0.321 2.78E-02 0.387 9.61E-03 0.33 3.51E-02 0.398 1.18E-02 PIK3IP1 -0.231 3.45E-02 -0.316 1.10E-02 -0.307 2.49E-02 -0.28 1.35E-02 PIK3R2 -0.242 2.38E-02 -0.252 2.56E-02 -0.312 2.98E-02 -0.39 5.92E-03 PIPOX 0.26 2.69E-02 0.357 1.05E-02 0.315 4.56E-02 0.311 2.08E-02 PITPNA 0.587 1.11E-03 0.705 7.17E-04 0.488 1.02E-02 0.645 7.02E-04 PITPNM1 -0.614 4.23E-04 -0.36 4.57E-03 -0.239 3.44E-02 -0.419 2.38E-03 PKN1 -0.379 1.36E-02 -0.29 1.24E-02 -0.247 3.66E-02 -0.428 2.69E-03 PLAT -0.575 6.13E-04 -0.385 5.87E-03 -0.288 2.78E-02 -0.974 1.24E-04 PLEKHG2 -0.459 2.57E-03 -0.31 9.28E-03 -0.249 3.50E-02 -0.369 5.84E-03 PLOD3 -0.556 1.46E-03 -0.397 7.05E-03 -0.323 3.33E-02 -0.426 7.64E-03 PMEPA1 -0.721 5.51E-04 -0.535 2.35E-03 -0.382 2.46E-02 -0.552 1.75E-03 POLDIP2 0.208 3.21E-02 0.207 4.16E-02 0.23 3.87E-02 0.284 1.15E-02 POLR3E 0.331 9.59E-03 0.578 1.53E-03 0.372 2.21E-02 0.453 3.58E-03 POPDC3 0.425 1.35E-02 0.697 1.34E-03 0.487 3.46E-02 0.602 2.45E-03 PORCN -0.349 6.54E-03 -0.213 4.84E-02 -0.29 3.32E-02 -0.423 4.51E-03 PPP1R3C 0.698 8.20E-04 0.606 3.01E-03 0.598 9.31E-03 1.258 2.25E-04 PQLC1 -0.704 6.48E-04 -0.514 3.01E-03 -0.424 1.90E-02 -0.783 6.60E-04 PRAF2 -0.649 3.84E-04 -0.296 1.19E-02 -0.337 1.54E-02 -0.467 1.74E-03 PRPF8 0.339 6.22E-03 0.395 3.75E-03 0.259 3.23E-02 0.383 3.91E-03 PRR5L -0.42 5.39E-03 -0.363 7.25E-03 -0.227 4.14E-02 -0.428 2.42E-03 PXDC1 -0.671 1.30E-03 -0.245 3.06E-02 -0.368 2.34E-02 -0.481 3.20E-03 R3HDM4 -0.401 2.12E-03 -0.364 6.55E-03 -0.286 2.56E-02 -0.43 3.55E-03 RAB11B -0.502 1.35E-03 -0.559 1.26E-03 -0.329 2.21E-02 -0.57 2.87E-03 RAPGEF5 -0.373 5.86E-03 -0.298 1.45E-02 -0.482 1.88E-02 -0.418 2.58E-02 RARB 0.709 9.36E-04 0.494 3.94E-03 0.336 4.57E-02 0.698 7.96E-04 RBFA -0.488 2.75E-03 -0.317 2.63E-02 -0.315 4.44E-02 -0.56 2.24E-03 RBMS3 -0.958 5.43E-04 -0.799 1.61E-03 -0.468 3.66E-02 -0.729 3.76E-03

241

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value RCAN2 -0.643 4.34E-04 -0.309 1.19E-02 -0.311 2.83E-02 -0.66 1.39E-03 RECQL4 -0.34 5.59E-03 -0.307 1.15E-02 -0.419 1.11E-02 -0.449 2.16E-03 RFXANK -0.303 1.11E-02 -0.394 5.89E-03 -0.238 4.18E-02 -0.403 4.45E-03 RHEB -0.583 9.66E-03 -0.387 2.74E-02 -0.607 2.14E-02 -0.619 4.18E-03 RHOU -0.503 1.60E-03 -0.333 7.76E-03 -0.267 2.81E-02 -0.637 7.97E-04 RNF216 -0.515 1.14E-03 -0.444 3.20E-03 -0.361 2.21E-02 -0.401 1.41E-02 RNMTL1 0.697 5.71E-04 0.847 8.70E-04 0.435 1.15E-02 0.797 3.98E-04 RNPEPL1 -0.524 2.65E-03 -0.419 1.01E-02 -0.332 3.30E-02 -0.333 1.74E-02 ROR2 0.429 6.70E-03 0.51 3.68E-03 0.347 3.58E-02 0.386 6.68E-03 RPA1 0.653 2.00E-03 0.703 3.00E-03 0.446 3.51E-02 0.613 2.80E-03 RPSAP58 -1.542 6.01E-04 -2.439 2.49E-04 -1.064 1.95E-02 -1.402 7.02E-04 RTN4RL1 0.64 2.80E-03 0.802 1.52E-03 0.56 2.20E-02 0.933 7.74E-04 RTTN -1.034 1.02E-03 -0.59 8.17E-03 -0.701 2.67E-02 -1.149 4.33E-04 RXRA -0.366 3.36E-03 -0.208 3.64E-02 -0.244 3.79E-02 -0.368 4.36E-03 S1PR1 -0.511 1.53E-03 -0.334 3.20E-02 -0.299 3.82E-02 -0.411 4.47E-03 SCARA3 -0.831 1.74E-04 -0.585 2.21E-03 -0.477 1.79E-02 -0.623 1.71E-03 SCO1 0.355 7.87E-03 0.485 4.86E-03 0.465 1.79E-02 0.521 2.62E-03 SDK1 -0.494 1.76E-03 -0.443 4.24E-03 -0.343 1.79E-02 -0.505 1.75E-03 SELM -0.742 3.74E-04 -0.511 1.37E-03 -0.331 2.26E-02 -0.566 8.09E-04 SEMA6B -0.314 1.31E-02 -0.263 2.83E-02 -0.287 4.41E-02 -0.389 7.22E-03 SEMA6C -0.329 1.92E-02 -0.321 1.35E-02 -0.295 3.58E-02 -0.279 3.00E-02 SEPT3 -0.744 5.78E-04 -0.485 3.43E-03 -0.392 2.20E-02 -0.696 9.92E-04 SERTM1 0.622 5.58E-04 0.885 2.19E-04 0.454 1.10E-02 0.641 4.02E-04 SETBP1 -0.804 4.63E-04 -0.53 2.52E-03 -0.357 3.17E-02 -0.862 3.73E-04 SHARPIN -0.304 1.14E-02 -0.317 9.41E-03 -0.417 1.32E-02 -0.472 1.58E-03 SHPK 0.383 4.01E-03 0.42 3.68E-03 0.364 2.26E-02 0.499 2.34E-03 SLC12A4 -0.536 1.19E-03 -0.423 3.66E-03 -0.336 1.95E-02 -0.557 1.09E-03 SLC16A2 -0.321 6.54E-03 -0.425 2.52E-03 -0.231 3.58E-02 -0.243 3.23E-02 SLC1A2 -1.115 5.27E-04 -0.91 1.32E-03 -0.589 2.66E-02 -1.249 1.99E-04 SLC1A4 -1.142 1.60E-04 -0.433 5.88E-03 -0.351 3.66E-02 -0.774 7.41E-04 SLC25A19 0.45 2.17E-03 0.636 1.49E-03 0.286 3.76E-02 0.497 2.12E-03 SLC25A23 -0.284 1.36E-02 -0.332 6.50E-03 -0.323 1.90E-02 -0.301 9.33E-03 SLC29A4 -0.747 3.19E-04 -0.842 2.19E-04 -0.333 3.02E-02 -0.516 1.85E-03 SLC2A10 -0.517 1.21E-03 -0.504 2.32E-03 -0.334 2.81E-02 -0.462 2.69E-03 SLC43A2 0.301 1.47E-02 0.612 1.77E-03 0.323 2.45E-02 0.321 9.84E-03 SLC44A2 -0.637 4.10E-04 -0.645 6.06E-04 -0.234 4.18E-02 -0.838 1.42E-04 SLC48A1 -0.759 5.06E-04 -0.501 3.43E-03 -0.316 2.70E-02 -0.477 2.19E-03 SLC9A3R1 0.213 3.38E-02 0.384 3.86E-03 0.267 2.66E-02 0.294 9.69E-03 SLCO2A1 -0.617 8.41E-04 -0.306 1.92E-02 -0.506 9.81E-03 -0.504 2.02E-03 SLIT2 0.44 2.71E-03 0.258 1.97E-02 0.462 1.53E-02 0.377 5.51E-03 SMAD2 -0.35 7.73E-03 -0.385 1.08E-02 -0.524 2.34E-02 -1.138 8.11E-04 SMARCD3 -0.391 6.14E-03 -0.308 1.26E-02 -0.444 1.20E-02 -0.302 1.26E-02 SMCR8 0.233 3.52E-02 0.421 3.87E-03 0.477 1.29E-02 0.442 6.15E-03 SMG9 -0.395 4.42E-03 -0.286 1.70E-02 -0.451 1.11E-02 -0.365 5.27E-03

242

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value SMTN -0.477 1.42E-03 -0.473 1.97E-03 -0.45 8.80E-03 -0.246 2.38E-02 SNX8 -0.714 1.29E-03 -0.566 2.51E-03 -0.395 2.53E-02 -0.496 5.31E-03 SOCS6 -0.936 8.94E-04 -0.62 2.15E-03 -0.577 1.83E-02 -1.142 2.85E-04 SPPL2B -0.308 8.18E-03 -0.427 3.24E-03 -0.377 1.91E-02 -0.422 4.99E-03 STK11 -0.294 8.12E-03 -0.26 3.56E-02 -0.238 3.66E-02 -0.367 8.30E-03 SULF1 -0.98 1.74E-04 -0.948 2.38E-04 -0.414 2.82E-02 -1.58 4.17E-05 SUN1 -0.361 6.22E-03 -0.458 3.43E-03 -0.338 2.82E-02 -0.459 1.22E-02 SYDE1 -0.486 1.80E-03 -0.425 3.30E-03 -0.324 2.28E-02 -0.312 1.10E-02 TCEA2 -0.84 1.88E-04 -0.647 6.27E-04 -0.489 1.11E-02 -1.116 4.96E-05 TCEB2 0.25 2.95E-02 0.519 3.02E-03 0.388 3.21E-02 0.27 3.87E-02 TCF25 -0.482 1.46E-03 -0.299 1.05E-02 -0.372 1.50E-02 -0.504 3.90E-03 TESK1 -0.549 9.00E-04 -0.296 1.03E-02 -0.227 3.97E-02 -0.426 2.60E-03 TGFB1 -0.664 5.88E-04 -0.258 2.22E-02 -0.811 1.68E-03 -0.37 4.97E-03 THNSL2 -0.581 9.36E-04 -0.294 1.27E-02 -0.308 2.44E-02 -0.574 5.73E-03 TIMM22 0.588 8.39E-04 0.676 5.37E-04 0.37 1.55E-02 0.68 5.87E-04 TLCD1 0.603 7.57E-04 0.537 1.52E-03 0.405 1.28E-02 0.71 3.43E-04 TLCD1 -0.603 1.05E-02 -0.836 2.05E-03 -0.379 1.15E-02 -0.773 2.23E-02 TLE2 -0.434 2.09E-03 -0.503 1.36E-03 -0.258 3.98E-02 -0.511 1.32E-03 TM7SF2 -0.679 4.10E-04 -0.582 9.35E-04 -0.279 2.81E-02 -0.944 8.98E-05 TMEFF2 0.372 9.89E-03 0.307 3.99E-02 0.512 1.15E-02 1.324 8.98E-05 TMEM187 -0.622 1.49E-03 -0.554 1.24E-02 -0.343 3.67E-02 -0.85 7.74E-04 TNFAIP8L1 -1.182 4.28E-05 -0.843 3.83E-04 -0.434 1.11E-02 -1.037 1.56E-04 TNKS -0.931 4.69E-04 -0.594 2.92E-03 -0.471 4.83E-02 -1.166 2.85E-04 TNS3 -0.664 9.36E-04 -0.539 3.50E-03 -0.324 2.82E-02 -0.369 1.58E-02 TP53INP1 -0.553 2.50E-03 -0.38 9.49E-03 -0.448 2.52E-02 -0.677 1.46E-03 TPP1 -0.575 8.02E-04 -0.351 5.80E-03 -0.299 2.56E-02 -0.575 1.14E-03 TRIOBP -0.682 4.86E-04 -0.522 1.92E-03 -0.355 2.14E-02 -0.539 2.75E-03 TRIP10 -0.471 1.79E-03 -0.397 4.38E-03 -0.325 2.82E-02 -0.487 2.01E-03 TSHZ1 -0.461 1.60E-03 -0.336 7.41E-03 -0.399 1.29E-02 -0.59 6.32E-04 TSPO -0.47 5.04E-03 -0.594 1.97E-03 -0.326 4.01E-02 -0.442 6.07E-03 TSR1 0.862 8.63E-04 1.075 5.59E-04 0.473 3.18E-02 0.972 4.02E-04 TTC28 -0.387 3.38E-03 -0.378 5.39E-03 -0.616 4.37E-03 -0.293 1.58E-02 TTYH3 -0.584 1.01E-03 -0.592 2.51E-03 -0.459 9.09E-03 -0.627 1.10E-03 TUBB2B 0.301 1.70E-02 0.382 6.15E-03 0.402 2.53E-02 0.528 1.48E-03 TXNIP -0.38 4.81E-03 -0.637 6.27E-04 -0.379 1.67E-02 -0.826 2.95E-04 TYK2 -0.353 4.10E-03 -0.344 7.50E-03 -0.246 4.05E-02 -0.457 2.62E-03 U2AF1L4 -0.659 3.96E-04 -0.467 2.52E-03 -0.291 2.46E-02 -0.521 1.41E-03 UBXN6 -0.494 1.21E-03 -0.409 3.46E-03 -0.362 1.90E-02 -0.423 5.21E-03 UGT3A2 0.228 2.96E-02 0.272 1.71E-02 0.39 1.15E-02 0.612 9.79E-04 ULK2 2.246 2.16E-05 2.284 4.54E-05 1.92 9.58E-05 2.582 4.95E-06 VCAN -0.52 2.00E-03 -0.404 5.66E-03 -0.4 2.35E-02 -0.531 5.12E-03 VEGFB -0.679 4.12E-04 -0.438 3.01E-03 -0.38 1.55E-02 -0.539 1.06E-03 VPS28 -0.534 1.34E-03 -0.544 1.55E-03 -0.396 1.80E-02 -0.6 1.09E-03 VPS53 0.551 1.44E-03 0.67 1.05E-03 0.326 4.96E-02 0.637 1.18E-03

243

untransfected vs WT D83V vs WT Y69H vs WT K4E vs WT

Gene Symbol log2 log2 log2 log2 fold adjusted fold adjusted fold adjusted fold adjusted change p-value change p-value change p-value change p-value WDR18 -0.306 7.54E-03 -0.203 4.35E-02 -0.223 4.02E-02 -0.269 1.47E-02 WDR7 -0.761 1.24E-03 -0.367 1.71E-02 -0.477 2.42E-02 -0.771 1.06E-03 XRCC1 -0.463 1.70E-03 -0.402 3.14E-03 -0.259 3.02E-02 -0.312 1.10E-02 ZADH2 -0.26 1.62E-02 -0.273 2.56E-02 -0.466 8.14E-03 -0.48 2.74E-03 ZC3H3 -0.323 8.36E-03 -0.21 4.81E-02 -0.27 3.43E-02 -0.262 1.99E-02 ZC3H6 -0.988 6.06E-04 -0.926 5.01E-04 -0.896 9.25E-03 -1.277 1.42E-04 ZKSCAN1 -0.427 8.79E-03 -0.332 3.34E-02 -0.41 4.56E-02 -0.507 2.03E-02 ZNF16 -0.565 2.09E-03 -0.326 3.42E-02 -0.632 9.31E-03 -0.639 2.62E-03 ZNF185 -0.624 5.11E-04 -0.568 1.38E-03 -0.268 3.77E-02 -0.846 2.46E-04 ZNF236 -0.747 5.04E-04 -0.338 1.16E-02 -0.444 1.83E-02 -0.901 4.76E-04 ZNF250 -0.515 2.68E-03 -0.259 4.44E-02 -0.683 1.13E-02 -0.529 4.54E-03 ZNF34 -0.33 3.28E-02 -0.279 3.67E-02 -0.331 4.71E-02 -0.423 7.43E-03 ZNF358 -0.463 5.46E-03 -0.469 3.62E-03 -0.294 3.89E-02 -0.523 2.38E-03 ZNF362 -0.54 1.98E-03 -0.404 5.66E-03 -0.408 1.72E-02 -0.486 3.42E-03 ZNF467 -0.258 1.87E-02 -0.339 8.69E-03 -0.372 1.45E-02 -0.246 2.36E-02 ZNF516 -0.381 3.48E-03 -0.246 2.22E-02 -0.215 4.41E-02 -0.42 2.80E-03 ZNF532 -1.449 4.53E-05 -1.299 8.27E-05 -0.367 2.50E-02 -1.282 5.59E-05 ZNF544 -1.225 3.99E-04 -1.188 6.20E-04 -0.504 3.87E-02 -1.46 1.95E-04 ZNF623 -0.424 5.31E-03 -0.305 1.96E-02 -0.476 1.83E-02 -0.525 1.55E-02 ZNF627 -0.823 7.34E-04 -0.575 4.96E-03 -0.487 2.77E-02 -1.039 4.33E-04 ZNF706 -0.608 6.96E-03 -0.346 3.42E-02 -0.516 3.69E-02 -0.749 3.76E-03

244

Appendix M: Functional annotation group enrichment of DEGs in mutant versus WT MEF2B-V5 cells

Shown are both IPA cellular function annotation groups absolute activation z-scores ≥ 2 and p-values ˂ 0.05, for the genes differentially expressed in K4E, Y69H and D83V MEF2B-V5 cells versus WT MEF2B-V5 cells. Positive z-scores indicate increased activity in mutant versus WT cells and negative z-scores indicate decreased activity in mutant versus WT cells.

Category Functions Annotation p-value Activation z-score # of genes Cellular Movement cell movement of cancer cells 9.90E-04 -2 6 Cell Cycle S phase of tumor cell lines 7.66E-03 2.433 7

245

Appendix N: Functional annotation group enrichment of DEGs in DLBCL patient samples with MEF2B mutations versus DLBCL patient samples without MEF2B mutations Shown are IPA cellular function annotation groups with absolute activation z-scores ≥ 2 and p-values ˂ 0.05 for genes differentially expressed in MEF2B mutant versus WT GCB DLBCL patient samples. Positive z-scores indicate increased activity in MEF2B mutant versus WT samples and negative z-scores indicate decreased activity in MEF2B mutant versus WT samples. In blue are annotation groups related to cell migration. In pink are annotation groups related to cell viability.

Activation # of Category Functions Annotation p-value z-score genes Cell Death and Survival cell viability of mononuclear leukocytes 1.34E-02 2 4 Cellular Movement chemotaxis of leukemia cell lines 7.20E-03 2 5 Cellular Movement chemotaxis of phagocytes 6.73E-05 2.125 12 Cell Signaling mobilization of Ca2+ 2.04E-03 2.158 12 Cellular Function and Maintenance ion homeostasis of cells 1.22E-02 2.17 11 Cellular Movement chemotaxis of neutrophils 1.65E-05 2.173 9 Cell-To-Cell Signaling and Interaction adhesion of phagocytes 7.20E-03 2.193 5 Cell-To-Cell Signaling and Interaction activation of tumor cell lines 6.39E-04 2.213 7 Cell Signaling, Molecular Transport quantity of Ca2+ 5.07E-03 2.286 11 Cellular Movement chemotaxis of leukocytes 3.18E-04 2.319 13 Cellular Movement chemotaxis of cells 6.01E-03 2.545 14

246

Appendix O: Functional annotation group enrichment of DEGs in DLBCL patient samples versus centroblasts Shown are the top twenty most significant IPA cellular function annotation groups with absolute activation z-scores ≥ 2 for genes differentially expressed in GCB DLBCL patient samples versus reactive centroblasts. Positive z-scores indicate increased activity in DLBCL cells versus centroblasts and negative z-scores indicate decreased activity in DLBCL cells versus centroblasts. In blue are annotation groups related to cell migration.

Activation # of Categories Diseases or Functions Annotation p-value z-score genes Cellular Movement cell movement 2.90E-48 9.916 544 Cellular Movement cell movement 3.51E-44 9.705 503 Cellular Movement migration of cells 1.88E-45 9.235 492 Cellular Movement migration of cells 8.33E-43 9.048 455 Cellular Movement homing 2.69E-30 8.444 176 Cellular Movement homing of cells 3.36E-30 8.38 170 Cellular Movement homing 1.24E-28 8.372 173 Cellular Movement homing of cells 5.77E-29 8.308 168 Cellular Movement chemotaxis 9.11E-28 8.161 170 Cellular Movement chemotaxis 2.13E-29 8.154 173 Cellular Movement chemotaxis of cells 9.98E-28 8.095 164 Cellular Movement chemotaxis of cells 6.32E-29 8.089 166 Cellular Movement leukocyte migration 8.50E-24 7.564 166 Cellular Movement leukocyte migration 6.90E-25 7.552 168 Cellular Movement cell movement of blood cells 1.48E-25 7.53 170 Cellular Movement cell movement of blood cells 4.01E-24 7.505 167 Cellular Movement cell movement of leukocytes 5.58E-24 7.163 156 Cellular Movement cell movement of tumor cell lines 9.52E-24 7.151 294 Cellular Movement cell movement of tumor cell lines 7.71E-24 7.023 326 Cellular Movement cell movement of leukocytes 1.10E-22 6.994 154

247