Identification and Characterization of E2A-PBX1 Genomic Binding Sites Using Chip-Seq and Bioinformatics

Identification and characterization of E2A-PBX1 genomic binding sites using ChIP-seq and bioinformatics

Kyster K. Nanan

A thesis submitted to the Department of Pathology and Molecular Medicine in conformity with the requirements for the degree of Doctor of Philosophy

Queen’s University Kingston, Ontario, Canada July, 2014

A translocation between chromosomes 1 and 19, referred to as t(1;19), results in expression of the chimeric, oncogenic transcription factor E2A-PBX1 in cases of B- progenitor acute lymphoblastic leukemia. Previous studies have shown that E2A- PBX1 functions as an activator of target gene transcription and that oncogenesis requires direct binding of E2A-PBX1 with the histone acetyltransferase CBP/p300. However, the transcriptional targets of this complex remain largely unknown. To identify and characterize genomic sites bound by E2A-PBX1 we performed RNA- seq and ChIP-seq for E2A-PBX1, p300 and the functional histone marks H3K4me1, H3K4me3, and H3K27me3 in the t(1;19) leukemic cell line RCH-ACV. Putative E2A- PBX1 binding sites were validated by site-specific ChIP-qPCR. Bioinformatics tools were used to perform integrative analysis and characterize E2A-PBX1 binding on a genome-wide scale. This analysis resulted in the identification of 555 high-confidence E2A-PBX1 sites in the RCH-ACV genome most of which occur distally (>10kb) relative to the TSSs of known genes. Ontology and pathway analysis showed that candidate E2A-PBX1 target genes are involved in hematopoiesis, lymphopoiesis, and cancer progression. Binding of E2A-PBX1 to a random sampling of these sites was

i performed using site-specific ChIP-qPCR. Consistent with earlier results, E2A-PBX1- bound sites were strongly associated with p300 recruitment and the activating chromatin marks H3K4me1 and H3K4me3; analysis of RNA-seq data confirmed a strong association between proximate binding by E2A-PBX1 and gene transcription. Sur- prisingly, rather than associating with consensus PBX1 binding sequences as expected, E2A-PBX1 associated with sites bound by the B-lymphopoietic transcription factors EBF1 and wild-type E2A; EBF1, E2A and E2A-PBX1 were confirmed to co-localize at selected E2A-PBX1 binding sites. The results of this study identify hundreds of E2A-PBX1 genomic binding sites, confirm CBP/p300 recruitment and transcriptional activation in association with these sites, and raise the novel and surprising possibility that E2A-PBX1 disrupts the transcriptional regulation of B- lymphopoiesis through physically associating with the lymphopoietic transcription factors EBF1 and wild-type E2A.

ii Acknowledgments

I would first like to thank my supervisor, Dr. David LeBrun for his mentorship and infallible support during my tenure in his lab. He is a true and outstanding scientist and, generally speaking, one of the most brilliant people who I have ever had the pleasure of meeting. It is impossible to quantify the enormous degree to which he has helped me to improve my scientific prowess, critical thinking skills, and career prospects. I am also grateful for the support, encouragement, and constructive feedback received from the exceptional members of my supervisory committee, Dr. Andrew Craig and Dr. Anne Croy. I would next like to thank the past members of the LeBrun lab, including Mark Woodcroft, Patrick Thompson, Chris Denis, Mohsen Alhejaily, and Wann-Cheng Lo, who each contributed positively to a very enjoyable and productive lab environment. I am further grateful to Huimen Geng and Bob Roeder for sharing with us, on a collaborative basis, ChIP-seq data from the 697 cell line, which turned out to be instrumental in the writing of this dissertation. I am greatly indebted to the funding agencies who supported me during my stu- dentship, including Queen’s University, the Government of Ontario, and the Terry Fox Research Foundation. I would further like to acknowledge the financial contributions by granting agencies, including CIHR, The Cancer Research Society, and The

iii Leukemia and Lymphoma Society of Canada, who made possible much of my work in the LeBrun lab. Finally, I would like to thank my dearest friend, Christina Liak, for her constant companionship, encouragement, patience, and support during my doctoral pursuits. I can’t wait to see what experiences and adventures await us in the future!

iv Contents

Abstract i

Acknowledgments iii

Contents v

List of Tables viii

List of Figures ix

List of Abbreviations xii

Chapter 1: Introduction 1 1.1 Acute Lymphoblastic Leukemia: Clinical and pathological features . . 1 1.2 Hematopoiesis and the hematopoietic stem cell ...... 3 1.2.1 The HSC and its properties ...... 5 1.2.2 Diﬀerentiation of the HSC ...... 7 1.3 A brief overview of the various stages of B lymphopoiesis ...... 9 1.4 Transcriptional gene regulation during lymphopoiesis ...... 11 1.5 Chromatin structure and its role in gene regulation ...... 13 1.5.1 Control regions in the eukaryotic genome ...... 15 1.6 Translocation 1;19 and E2A-PBX1 ...... 16 1.6.1 Wildype PBX1 ...... 19 1.6.2 Wildtype E2A/TCF3 ...... 22 1.6.3 Activation of E-protein target genes through p300 recruitment 26 1.7 Oncogenesis by E2A-PBX1: experimental models and oncogenic mechanisms ...... 28 1.7.1 The t(1;19) event converts PBX1 into a constitutive transcriptional activator ...... 29 1.8 Regulatory targets of E2A-PBX1 ...... 29 1.9 CBP/p300 in E2A-PBX1 oncogenesis ...... 31 1.10 ChIP and massively-parallel sequencing technologies ...... 32

v 1.11 Rationale and Hypothesis ...... 32 1.11.1 Hypothesis: ...... 33 1.11.2 Experimental objectives ...... 33

Chapter 2: Materials and Methods 35 2.1 Cell lines and culture conditions ...... 35 2.2 Chromatin Immunoprecipitation- quantitative PCR (ChIP-qPCR) . . 35 2.3 Next-generation sequencing and bioinformatics analysis ...... 36 2.4 siRNA-mediated gene knockdown ...... 38 2.5 Immunoblotting ...... 38 2.6 Antibodies used for ChIP, ChIP-seq, and immunoblotting ...... 39 2.7 RNA harvest, cDNA synthesis, and qRT-PCR ...... 39 2.8 Statistical analyses ...... 40

Chapter 3: Results 41 3.1 Identiﬁcation and validation of E2A-PBX1 binding sites in RCH-ACV cells ...... 41 3.2 E2A-PBX1 predominantly localizes to intronic and intergenic regions 47 3.3 Biological relevance and functional interconnectivity of E2A-PBX1 target genes ...... 48 3.4 E2A-PBX1 binding is associated with p300 recruitment and transcriptional activation ...... 53 3.5 E2A-PBX1 binding sites possess the characteristics of active transcriptional enhancers ...... 57 3.6 E2A-PBX1 associates with genomic sites bound by the B lymphopoietic transcription factors E2A and EBF1 ...... 59 3.7 Knockdown of E2A-PBX1 silences target gene expression ...... 65

Chapter 4: Discussion 67 4.1 Cooperative regulation of transcriptional enhancers by E2A-PBX1 and B lymphopoietic transcription factors ...... 67 4.2 Leukemogenesis via a hijacking of normal lymphopoietic regulatory networks ...... 70 4.3 An E2A-PBX1-containing transcriptional regulatory complex . . . . . 71 4.4 Transcriptional silencing by allosteric modiﬁcation ...... 72 4.5 The functional contribution of E2A-PBX1 to leukemogenesis . . . . . 73 4.6 Targeting E2A-PBX1-responsive genes in leukemia ...... 74 4.7 Caveats and limitations ...... 75 4.8 Conclusion ...... 77

Bibliography 78

vi AppendixA: AML1-ETO Introduction 110 A.1 Acute Myeloid Leukemia ...... 110 A.2 The t(8;21) chromosomal translocation ...... 111 A.3 Wildtype AML1 and the core binding factor ...... 112 A.4 The ETO family of transcriptional repressors ...... 114 A.5 The oncogenic fusion protein AML1-ETO ...... 116 A.6 Modelling the eﬀects of AML1-ETO expression in vivo ...... 121 A.7 The curious phenotype of t(8;21)-associated AML cells ...... 122 A.8 AML1-ETO interacts with E2A and induces E-protein silencing . . . 123 A.9 Rationale and Hypothesis ...... 126 A.9.1 Hypothesis: ...... 127 A.9.2 Experimental objectives: ...... 127

AppendixB: AML1-ETO Results 128 B.1 AML1-ETO interacts with E2A in an NHR1- and NHR2-dependent manner ...... 128 B.2 E-protein silencing by AML1-ETO is NHR1- and NHR2-dependent . 131 B.3 Examining the eﬀects of AML1-ETO on immunoglobulin chain gene rearrangement ...... 134 B.4 AML1-ETO blocks diﬀerentiation and enhances long-term proliferation of myeloid progenitors ...... 137 B.5 Examining the role of AML1-ETO on B lymphocyte development . . 140

AppendixC: AML1-ETO Discussion 146 C.1 Silencing of E2A transactivation by AML1-ETO is mediated by NHR1 and NHR2 ...... 147 C.2 AML1-ETO does not aﬀect E2A-dependent B lymphopoietic activity 147 C.3 Context-speciﬁc role of AML1-ETO in transcriptional regulation . . . 148 C.4 Conclusions and suggestions for future experiments ...... 150

AppendixD: E2A-PBX1 151 D.1 E2A-PBX1 contributes to both up- and downregulation of its operationally- deﬁned target genes ...... 151 D.2 E2A-PBX1 promotes cell cycle and opposes hematopoietic diﬀerentiation154

vii List of Tables

3.1 Disease ontology terms associated with candidate E2A-PBX1 target genes ...... 49 3.2 Top-ranked mouse phenotype terms associated with candidate E2A- PBX1 target genes ...... 49

viii List of Figures

1.1 The stages of hematopoiesis ...... 8 1.2 The stages of lymphopoiesis ...... 10 1.3 The E2A-PBX1 fusion protein ...... 18 1.4 The homeobox protein PBX1 ...... 20 1.5 The E-protein family ...... 23 1.6 Gene regulation by E-proteins and their antagonists ...... 25 1.7 Canonical model of E2A-PBX1-mediated leukemogenesis ...... 34

3.1 Example of peaks submitted for visual inspection ...... 43 3.2 Validation of high-conﬁdence E2A-PBX1 binding peaks using ChIP- qPCR ...... 45 3.3 Genomic binding of E2A-PBX1 ...... 47 3.4 Functional interaction network of E2A-PBX1 target genes ...... 51 3.5 E2A-PBX1 binding sites are associated with recruitment of p300 and enrichment for histone marks associated with transcriptional activity 54 3.6 Proximate binding of E2A-PBX1 is associated with diﬀerentially increased gene transcription selectively in t(1;19)-associated cells . . . . 56 3.7 Enrichment of H3K27Ac at E2A-PBX1 binding sites ...... 58

ix 3.8 Wildtype E2A and EBF1, but not PBX1, motifs are enriched at E2A- PBX1 binding sites ...... 60 3.9 E2A and EBF1 co-localize in GM12878 cells at genomic regions corresponding to E2A-PBX1 binding sites ...... 62 3.10 Binding of wildtype E2A and EBF1 at E2A-PBX1 peaks in RCH-ACV cells...... 64 3.11 Response of E2A-PBX1 target genes to E2A-PBX1 knockdown. . . . 66

4.1 Models of gene regulation by E2A-PBX1 in t(1;19)-associated leukemia cells ...... 69

A.1 Schematic of AML1-ETO and its wildtype precursors ...... 118 A.2 E-protein silencing by AML1-ETO ...... 125

B.1 AML1-ETO interacts with E2A in a NHR1- and NHR2-dependent manner ...... 130 B.2 AML1-ETO silences transactivation by E-proteins in an NHR1- and NHR2-dependent manner ...... 133 B.3 Rearrangement of the immunoglobulin kappa light chain is an E2A- dependent process ...... 135 B.4 AML1-ETO does not inhibit VDJ rearrangement in BOSC23 cells . . 136 B.5 AML1-ETO inhibits diﬀerentiation and increases proliferation of myeloid progenitors in vitro ...... 139 B.6 Schematic of ex vivo lymphoid diﬀerentiation assay ...... 140 B.7 AML1-ETO expression reduces proliferation and survival of B lymphocyte precursors ...... 142

x B.8 AML1-ETO does not block E-protein-dependent commitment to the B lymphoid lineage ...... 145

D.1 Response of the operationally-deﬁned E2A-PBX1 target genes to E2A- PBX1 knockdown ...... 153 D.2 Cellular pathways aﬀected by E2A-PBX1 knockdown ...... 155

xi List of Abbreviations

AD activation domain.

ALL acute lymphoblastic leukemia.

AML acute myeloid leukemia.

Ara-C arabinofuranosyl cytidine.

BCR B cell receptor. bHLH basic helix-loop-helix.

CBF core binding factor.

CBP CREB binding protein. cDNA complementary DNA.

ChIP chromatin immunoprecipitation.

CLP common lymphoid progenitor.

CMP common myeloid progenitor. eTAFH ETO TATA box-binding protein-associated factor homology domain. xii ETO eight-twenty-one.

FAB French-American-British classiﬁcation system.

FACS ﬂuorescent activated cell-sorting.

FISH ﬂuorescence in-situ hybridization.

G-CSF granulocyte colony-stimulating factor.

GAL4 galactose metabolism 4.

GFP green ﬂuorescent protein.

GM-CSF granulocyte macrophage colony-stimulating factor.

GMP granulocyte-macrophage progenitor.

HCM HOX cooperativity motif.

HDAC histone deacetylase.

HHR hydrophobic heptad repeat.

HSC hematopoietic stem cell.

IL-7 interleukin-7.

IRES internal ribosome entry site.

LT-HSC long-term hematopoietic stem cell.

M-CSF macrophage colony-stimulating factor. xiii MPP multipotent progenitor. mSin3A mammalian SIN3 homolog A, transcription regulatory (yeast).

N-CoR nuclear co-repressor.

NGS next-generation sequencing.

NHR Nervy homology region.

NK cell natural killer cell.

PCR polymerase chain reaction.

PLZF promyelocytic leukemia zinc ﬁnger. qPCR quantitative polymerase chain reaction. qRT-PCR quantitative reverse-transcription polymerase chain reaction.

RAG recombination activating gene.

RHD Runt homology domain.

SCF stem cell factor.

SMMHC smooth muscle myosin heavy chain.

SMRT silencing mediator for retinoid or thyroid-hormone receptors.

ST-HSC short-term hematopoietic stem cell.

TAD transcriptional activation domain.

UAS upstream activating sequence.

xiv 1

Chapter 1

Introduction

1.1 Acute Lymphoblastic Leukemia: Clinical and pathological features

The most common hematological malignancy observed in young children is acute lymphoblastic leukemia (ALL), which occurs at a rate of 1 in 100,000 in the US [6]. The disease is characterized by the outgrowth of immature hematopoietic progenitors, called blasts, which crowd the bone marrow and interfere with normal blood production. ALL patients show anemia, increased susceptibility to infection, bleeding, and bruising, and generally die of symptoms of bone marrow failure within days to weeks after diagnosis. Many subtypes of leukemia, including ALL, are associated with large-scale reciprocal chromosomal translocations. The recurrent, non-random chromosomal translocation involving chromosomes 1 and 19, formally t(1;19)(q23;p13), is associated with up to 5% of ALL cases. Leukemia is a biologically and clinically heterogenous group of diseases. Leukemia subtypes are defined by their clinical and patho-biological features; the latter include morphological, immunophenotypic, cytogenetic and genetic attributes. Acute leukemia is typified by a rapid accumulation of leukemic blasts, which progresses over 1.1. ACUTE LYMPHOBLASTIC LEUKEMIA: CLINICAL AND PATHOLOGICAL FEATURES 2 a period of weeks to months and ultimately leads to bone marrow failure if left untreated. Patients diagnosed with chronic lymphocytic leukemia (CLL) experience a prolonged, steady accumulation of mature B cells in circulation, whereas chronic myelogenous leukemia (CML) is associated with accumulation of granulocytic cells at all stages of maturation but not including mature cells. Lymphoid and myeloid leukemia are often differentiable by the presence of lineage-specific cell surface markers. As with many other types of cancers, it is often the case for leukemia to be associated with some type of large-scale chromosomal abnormality; reciprocal translocations are most commonly observed (Reviewed in [23] and [64]). In humans, leukemogenesis can occur at any age but certain subtypes are more common within well-defined age groups. For example, CLL patients present at over 50 years old with their disease, which is in contrast to ALL patients, who are most often afflicted during early childhood. There are undeniable parallels between hematopoiesis and leukemogenesis; the two processes are so inseparable that leukemia is often described as a case of hematopoiesis gone awry. This relationship is made clear, in part, by the observation that many of the well-characterized transcription factors that were found to be essential to hematopoietic differentiation and homeostasis were discovered through translocation of their genes in leukemia. Take for example the basic helix-loop-helix (bHLH) transcription factor SCL (also known as TAL1), which was discovered around 1989 in the DU.528 T-ALL cell line by virtue of its translocation involving the T-cell receptor gene [18]. RUNX1 was similarly first cloned from a patient with t(8;21)-associated AML [122]. Other examples of hematopoietic transcription factors involved in chromosomal translocations include RUNX1, CBP, MLL, and E2A (Reviewed in [36]). 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 3

1.2 Hematopoiesis and the hematopoietic stem cell

Hematopoiesis refers to the ordered process of blood cell formation, which involves hematopoietic stem cell (HSC) renewal, progenitor expansion, and terminal differentiation into mature, functional blood cells (reviewed in [134]). The profound phenotypic changes associated with hematopoiesis are regulated through the highly coordinated activities of myriad transcription factors, whose actions are, in turn, heavily influenced by epigenetic changes and chromatin conformation at developmentally-relevant target genes. Before addressing specifically the transcription factors and epigenetic changes that accompany hematopoiesis, it is necessary to discuss the process of hematopoiesis itself and to understand hematopoiesis necessitates a discussion about the mother of all cells of the hematopoietic system, the hematopoietic stem cell. It is well known that every multicellular organism begins its life as a single cell, the zygote. Through successive rounds of division, the zygote produces the cells that will eventually become bones, skin, neurons, and other tissue types, including blood. In the early 1900s, the unitarian hematopoietic theory was devised by the Russian experimental pathologist Alexander Maximov [113]. His theory, which was met with much skepticism at the time, posited that hematopoietic cells are derived from a single progenitor cell type that differentiated in response to changing environmental cues. It wasn’t until the mid-20th century that the case for the HSC gained major support by way of experiments that exposed mice to the DNA-damaging effects of X- ray irradiation [165, 167]. Swift, Taketa and Bond performed studies using mice that were subject to whole or sequential partial-body irradiation (also called regionally- fractionated irradiation). Though the two experimental groups of mice received the 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 4

same radiation dose, those mice that received sequential partial-body irradiation had a substantially reduced mortality due to marrow failure compared to their whole- body-irradiated counterparts. This lead to the hypothesis that under regionally- fractionated irradiation, a circulating agent of unknown identity was, by chance, spared a lethal dose of radiation that allowed the ”seeding” of the hematopoietic system during the post-irradiation recovery phase. Later, it was shown that radiation syndrome could be prevented with the injection of bone marrow cells into irradiated mice, suggesting that a cell capable of repopulating the hematopoietic system resided within [167]. It wasn’t until the experiments in the early 1960s that the field of HSC research really took off. The field of HSC stem cell research took a huge leap forward during the early- to-mid 1960s when James Till and Ernest McCulloch first demonstrated the existence of bone marrow-derived multipotent HSCs in an experimentally-tractable and quantitative manner [114, 17]. They first demonstrated increased survival of lethally- irradiated mice transplanted with increasing numbers of healthy, non-irradiated bone marrow. Next, they devised an in vivo colony-formation assay that involved transplantation of bone marrow cell suspensions into irradiated recipient mice and counting the resulting spleen nodules [17]. The spleen nodules consisted of rapidly-developing immature hematopoietic precursors and appeared at approximately 11 days post- transplantation. This in vivo system was superior to the in vitro colony formation assays of the day since it allowed for both proliferation and differentiation of colony- forming units. Importantly, it provided a method of quantifying the re-constituting component of the transplanted blood since the number of spleen nodules present were 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 5

proportional to the number of bone marrow cells transferred to the irradiated recipient. Finally, and most signiﬁcantly, they demonstrated that the colonies formed in the spleen were clonogenic, i.e. originated from a single cell [171, 115]. Till and McColloch proposed that it was the colony forming cells that were, in fact, the HSC.

1.2.1 The HSC and its properties

HSCs are a rare population of cells that reside in the bone marrow of adult mammals and sit atop the highest point of the hematopoietic cascade. The gold standard definition of an HSC is any cell that can provide long-term reconstitution of the entire hematopoietic system of lethally irradiated recipients. Further properties of the HSC include self-renewal and the ability to exit the cell cycle and enter a quiescent state, which provides protection from genotoxic stress (e.x. irradiation and chemotherapy) and allows for an extended life span (Reviewed in [181] and [153]. The operational definition of the HSC is a cell that can provide long-term reconstitution of the hematopoietic system of a lethally-irradiated animal. It wasn’t until the late 1980s that Spangrude and colleagues used fluorescence activated cell sorting (FACS) to recover a lineage negative (Lin−: Mac-1−Gr-1−B220−CD4−CD8−), multipotent subpopulation of BM-resident hematopoietic precursors characterized by expression of Thy-1 and Sca-1 cell surface markers [159, 158]. These cells achieved complete reconstitution of B, T, and myeloid lineages upon transplantation into lethally-irradiated mice at an astoundingly low dose of thirty cells per transplant recipient. It was later observed that expression of the cytokine receptor c-kit was limited to undifferentiated, immature BM-derived hematopoietic precursors. From that point forward, the Lin−Sca-1+c-kit+ (LSK) immunophenotype was entrenched, 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 6

from an experimental science standpoint, as the HSC-enriched pool. A key point of clarification was later made when it was demonstrated that LSK cells could be further divided based on high vs low/absent CD34 expression, which was associated with short-term hematopoietic stem cells (ST-HSCs) and long-term hematopoietic stem cells (LT-HSCs), respectively [135]. HSCs are caught amidst a perpetual balance between the capabilities required for hematopoietic homeostasis: self-renewal, lineage commitment/differentiation, and senescence or apoptosis. Self-renewal, during the expansion stage in the fetal liver, for example, is accomplished via symmetrical cell division, resulting in the production of two identical daughter cells, which is necessary to maintain the stem cell pool. In order to contribute to the pool of fully-differentiated hematopoietic cells, HSCs must undergo asymmetric cell division. Aside from producing an exact clone of the parental HSC, asymmetric division results in production of a daughter cell with either of two fates: (1) the daughter cell is destined to immediate commitment, maturation, and terminal differentiation along a single hematopoietic lineage or (2) the daughter cell may itself be another stem cell with subtly different properties from the parental HSC. The primary functional distinction between the parental LT-HSC and the progeny ST- HSC, referred to as the multipotent progenitor (MPP) in humans, is simply that only the LT-HSC is capable of long-term reconstitution of the adult hematopoietic system while the ST-HSC provides up to 8 weeks of reliable hematopoietic reconstitution [125]. 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 7

1.2.2 Diﬀerentiation of the HSC

As mentioned previously, the HSC may undergo symmetric or asymmetric division, the latter of which results in the production of the MPP, which is capable of terminal differentiation into all lineages of the hematopoietic system. The hematopoietic differentiation scheme is often represented as a family tree, atop of which sits the HSC and within which are the progeny of the various lineages (Figure 1.1). In humans, one of the earliest bifurcations from the MPP in the hierarchy produces either the common myeloid (CMP) or multi-lymphoid (MLP) progenitors (an alternative hierarchical structure, which results in the production of the common lymphoid progenitor (CLP), is observed in mice. Both human and murine hematopoiesis are reviewed in [43]). CMPs populate the myeloid compartment by further differ- entiating and maturing into granulocytes (which include basophils, eosinophils, and neutrophils), monocytes (which differentiate into macrophages in the appropriate environment), erythrocytes, mast cells, and megakaryocytes. MLPs populate the lymphoid compartment by maturing into T and B lymphocytes, as well as natural killer (NK) cells [43]. Immunophenotyping and FACS were used to identify and characterize the CMP and CLP. The CMP was subsequently identified as a population of LSK hematopoietic progenitors that lack IL-7Rα expression. Further fractionation, based on FcγRII/III and CD34 expression, identified the granulocyte-monocyte progenitor (GMP) as well as the megakaryocyte-erythrocyte progenitor (MEP) [4]. In 1997, Kondo and colleagues identified the CLP, through transplantation of sorted bone marrow LSK cells [94]. They demonstrated that expression of the cytokine receptor IL-7Rα was associated with exclusive reconstitution of the B-, T- and NK cell compartments when 1.2. HEMATOPOIESIS AND THE HEMATOPOIETIC STEM CELL 8

transplanted into irradiated mice [94].

Figure 1.1: The stages of hematopoiesis

Figure 1.1. Hematopoiesis is a developmental process in which progenitor cells undergo a stepwise series of lineage restrictions before ultimately becoming a mature, terminally diﬀerentiated component of the hematopoietic system. In green text, adjacent to the arrows, is listed the key cytokines predominantly associated with the transition from a progenitor to a more diﬀerentiated cell type. Adapted from [134] and [43]. 1.3. A BRIEF OVERVIEW OF THE VARIOUS STAGES OF B LYMPHOPOIESIS 9

1.3 A brief overview of the various stages of B lymphopoiesis

The stepwise process of B lymphopoiesis begins with the Il-7rα+ CLP (Reviewed in [72]. (Figure 1.2). Stimulation of the CLP by IL-7 is essential for progression along the B lymphoid lineage, as is evidenced by Il-7−/− mice that develop normal T lymphoid and NK compartments, but have signiﬁcantly reduced B lymphoid cell numbers [40]. The subsequent emergence of the cell surface marker B220 characterizes the pre-pro B cell stage, which is associated with further commitment to the B lymphoid lineage. During this stage, the transcription factors E2a (also known as Tcf3) and Ebf1 collaborate to further enforce the lymphoid lineage program, activating expression of genes like λ5, Vpreb1, and the DNA recombinase enzymes Rag1 and Rag2. Expres- sion of E2A, EBF1, and their downstream targets propels the developing lymphocyte forward to the pro-B stage, characterized by Cd19 surface expression and V-DJ rearrangement of the immunoglobulin heavy chain, required for eventual B cell receptor (BCR) assembly. In the nucleus, E2a and Ebf1 collaborate to upregulate expression of the lymphocyte-speciﬁc transcription factor Pax5. This trio of transcription factors further collaborate to upregulate a number of target genes, including Cd79a and other components of the pre-B cell receptor (pre-BCR). Progression toward the pre-B stage is associated with light chain VJ rearrangement and surface expression of the pre-BCR. The immature B cell is characterized by a downregulation of the recombinases and an increase in surface expression of IgM. At this point, the developing B cell exits the bone marrow and begins its migration to peripheral lymphoid organs where it will complete its development. Finally, IgD expression reaches the cell surface and the mature B cell phenotype is established. All of the aforementioned stages of lymphocyte development occur in the absence of antigen. Stimulation of B 1.3. A BRIEF OVERVIEW OF THE VARIOUS STAGES OF B LYMPHOPOIESIS 10 cells by antigens results in the generation of memory B cells and plasma cells, which secrete antibodies and contributes to a functional immune system.

Figure 1.2: The stages of lymphopoiesis

Figure 1.2. B lymphopoiesis occurs in a stepwise fashion from the earliest commit- ted progenitor, the pro-B cell. Key transcriptional regulators in B cell development and maturation include E2A, EBF1, and PAX5, which are highlighted in the associated table. 1.4. TRANSCRIPTIONAL GENE REGULATION DURING LYMPHOPOIESIS 11

1.4 Transcriptional gene regulation during lymphopoiesis

It should be noted that E2A is only one of the key regulators of lymphopoiesis. The transformation from an HSC to a functional B lymphocyte requires the concerted operation of a variety of additional transcriptional regulators, including EBF1 and PAX5, the subjects of this section. The transcription factor EBF1 was first cloned and characterized in 1993 in the lab of Rudolf Grosschedl [71]. EBF1 mRNA was found to be expressed primarily in pre-B and mature B lymphoid cell lines, but not in T lymphoid and non-lymphoid cell lines, such as HeLa and Cos-7. The EBF1 protein is characterized by an N-terminal zinc-finger motif and a C-terminal activation/multimerization domain. Engineered EBF1 internal deletion mutants demonstrate that the zinc finger motif is required for DNA-binding while the C-terminal region is required for homodimer formation. To characterize the transcriptional function of Ebf1 in vitro, reporter assays using the Cd79a promoter, which contains an Ebf1 consensus motif, demonstrated that Ebf1 is a transcriptional activator [73, 155]. Because EBF1 expression was so tightly associated with the B lymphoid lineage and can regulate lymphocyte-associated genes in vitro, the next logical questions re- garded the roles of EBF1 in hematopoiesis and B lymphopoiesis. To address this, Ebf1 null transgenic mice were engineered by the Grosschedl group [102]. Ebf1 null mice were viable but significantly smaller than their heterozygous or wildtype littermates. The most striking phenotype observed in Ebf1 null mice, however, was the complete lack of mature (B220+IgM+) B lymphocytes in the blood, spleen, or bone marrow. It should be noted that T and myeloid cells were present in normal numbers in Ebf1 null mice, indicating the specific requirement for Ebf1 function in the B lymphoid 1.4. TRANSCRIPTIONAL GENE REGULATION DURING LYMPHOPOIESIS 12

compartment. The blockage in lymphoid development observed in Ebf1 null mice was determined to have occurred at a very immature stage because the few B220+ lymphocyte precursors recovered from Ebf1 deficient mice lacked expression of lymphoid-associated genes (ex. Cd79a, Vpreb1, λ5, Rag1, and Rag2 ), and there was no detectable immunoglobulin gene rearrangement. In accordance with previous experiments that established a ”hierarchy” for hematopoietic transcription factors, B220+ lymphocyte precursors from Ebf1 deficient mice expressed E2a but lacked Pax5 expression. The above data indicates that E2a functions independently of Ebf1 during the earliest stages of lymphopoiesis, whereas Pax5 expression is dependant on Ebf1. Meinrad Busslinger’s group was the first to identify PAX5 (denoted ”Pax5” in mice), a homologue of the sea urchin TSAP protein, which they initially termed ”B cell-specific transcription factor” (BSAP) [12]. While PAX5 expression was detected in a variety of tissue, including the developing brain and testis, its expression within the hematopoietic system was observed at earlier stages of lymphoid development, such as the pro-B or pre-B stage, but not in mature plasma cells) [1]. Busslinger’s group further proceeded to demonstrate that CD19, a B lymphocyte cell surface marker, was a target of transcriptional activation by PAX5, intimating its role in regulating lymphoid gene expression and development [96]. In 1994, the engineering of a Pax5 knockout mouse model confirmed this supposition [174]. Pax5 heterozygous mutants appeared to develop normally when compared to their wildtype littermates, suggesting that there were no gene dose-dependent effects on overall viability [174]. Pax5 null homozygous mice, however, were born alive but had severe growth retardation and the majority died in less than a month. The 5% of 1.5. CHROMATIN STRUCTURE AND ITS ROLE IN GENE REGULATION 13

homozygous Pax5 null mice that survived to adulthood were viable but were much smaller compared to their littermates. While Pax5 null mice had defects in brain formation, the hematopoietic phenotype was confined to the B lymphoid compartment. Pax5 deficient mice lacked mature B220+IgM+ B lymphocytes in the blood, spleen, and bone marrow. Similar to Ebf1 knockout mice, the T lymphoid and myeloid compartments were unaffected by the absence of Pax5. Taken together, the data from Pax5 knockout mice advocate for its role in the hematopoietic system as an essential regulator in early B lymphocyte differentiation.

1.5 Chromatin structure and its role in gene regulation

Gene regulation is achieved largely through the coordinate binding of myriad transcription factors to regulatory regions throughout the genome. Transcription factor binding may be associated with co-activator or co-repressor recruitment, which can act to directly up- or downregulate expression of the regulatory target. An additional ”layer” of gene regulation is further accomplished through changes in chromatin conformation and accessability (recently reviewed in [177]). ”Chromatin” refers to the dense, highly organized aggregate of protein and DNA found in the nucleus of a cell (except in the case of erythrocytes, platelets, and corneocytes, which are anu- cleate). The structural unit of chromatin is the nucleosome, which is comprised of a short length of DNA ( 150 bp) wrapped around a histone core. Histones facilitate the higher-level nucleosome packing and chromatin organization required to compact the almost two meters of DNA (in humans) into each nucleus while leaving it amenable to dynamic gene regulation by transcription factors. While there are enzy- matic complexes (e.x. SWI/SNF) tasked with performing large-scale reorganizations 1.5. CHROMATIN STRUCTURE AND ITS ROLE IN GENE REGULATION 14 of nucleosomes, local changes in chromatin accessibility can occur largely through covalent post-translational modification of the protruding histone tail. For example, acetylation of lysine residues on the tails of adjacent histones, catalyzed by histone acetyltransferases (HATs), contributes to localized electrostatic repulsion that results in nucleosomes that are more spread out, resulting in a relaxed or open chromatin state [163, 185]. Conversely, cleavage of the acetyl modification by histone deacety- lases (HDACs) result in decreased nucleosome spacing and chromatin condensation. Histone acetylation is most commonly associated with transcriptional activation and is catalyzed by acetyltransferase enzymes, such as GNAT, MYST, and CBP/p300 [162]. Histone acetylation is thought to counteract the natural tendency of nucleosomes to form compact structures that would normally be inaccessible to DNA- binding proteins [60]. Gene regulatory complexes and the transcriptional machinery are more readily able to access open chromatin regions characterized by abundant histone acetylation, resulting in increased gene expression. H3K18 and H3K27 are the major in vivo epigenetic targets of CBP/p300 [84]. Recent high-throughput sequencing studies have demonstrated that the H3K27Ac modification distinguishes active from inactive and poised enhancers [37]. In contrast to histone acetylation, methylation affects chromatin structure by contributing relatively little to the electrostatic interactions between adjacent nucleosomes. Rather, methylation patterns control the recruitment of non-histone proteins, including ATPase-dependent remodelling complexes, that can further compress or re- lax the chromatin fiber [86]. Tri-methylation of histone H3 on lysine residue number 27 (H3K27me3) is one of the most well-characterized post-translational histone modifications. The Polycomb group (PcG) complex is a highly-conserved collection of 1.5. CHROMATIN STRUCTURE AND ITS ROLE IN GENE REGULATION 15

chromatin regulating proteins and enzymes (reviewed in [156]). The core PRC2 complex, comprised of SUZ12, EED, and EZH2, is specifically involved in tri-methylation of H3K27 [27]. EZH2 is the catalytic subunit of PRC2, which performs each of the three methyltransferase reactions required to yield H3K27me3. This modified histone is recognized by the CBX subunit of PRC1, which is a multi-protein complex that ultimately contributes to chromatin compaction, and thus gene silencing, through the activity of its RING1B subunit [48]. Chromatin structure is closely associated with developmentally-regulated gene expression patterns [37, 190, 169]. For example, recent genome-wide studies have shown that embryonic and induced pluripotent stem cell states are characterized by an increased prevalence of open chromatin when compared to more differentiated cell states [195]. In particular, chromatin marked by H3K27me3 is far less abundant in developmentally immature cell types compared to terminally-differentiated cells. Not surprisingly, the distribution of histone modifications is perturbed in tumor cells compared to their benign counterpart. For example, EZH2 overexpression in certain subtypes of breast cancer correlates with increased H3K27me3 deposition at the promoters of tumor suppressor genes, including RAD51, RUNX3, and CDKN1C, contributing to gene silencing, increased tumorigenicity, and poor patient outcome [66, 93].

1.5.1 Control regions in the eukaryotic genome

The massive, collaborative eﬀort to sequence the complete human genome in the early 2000s revealed that protein-coding exons comprised only 1.2% of the entire eukaryotic genome [82]. The junk-DNA hypothesis of the 1980s was threatened by 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 16

the Encyclopedia of DNA Elements (ENCODE) project, of which the main goal was to identify all functional DNA elements in the human genome [168]. ENCODE data demonstrated that over 80% of the human genome was biochemically active (i.e. at least defined by RNA transcription, transcription factor binding, or an abundance of functional histone modifications) [168]. Pre-ENCODE, the accepted model of transcriptional regulation involved a limited set of control regions, including the core promoter, enhancers, silencers, and insulators (reviewed in [145]). Core promoters are the TATA-box-containing regions immediately adjacent to the transcriptional start site (TSS) of a gene and predominantly serves to position the RNA transcription machinery [145]. ENCODE data have shown that actively transcribed promoters are associated with the functional histone modifications H3K4me2 and H3K4me3 [168]. In fact, ENCODE data have shown that the distribution of functional histone modifications near promoters can be used to predict relative RNA expression levels of the associated gene [168]. Enhancers comprise another class of gene control regions that, unlike promoters, can be found hundreds of kilobases up- or downstream of the TSS of their regulatory targets [145]. Enhancers are typified by the presence of abundant transcription factor recognition motifs and the presence of the H3K4me1 histone modification [168]. Ad- ditional studies have shown that the presence of H3K27Ac differentiates active from inactive enhancers [37, 55].

1.6 Translocation 1;19 and E2A-PBX1

As previously mentioned, the non-random 1;19 chromosomal translocation, is commonly associated with ALL cases. Around 1989, studies were performed in Michael 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 17

Cleary’s lab to precisely identify the genes located at the chromosomal breakpoints involved in the t(1;19) event. Early in 1990, researchers from the labs of Michael Clary and David Baltimore simultaneously published their work in Cell on the cloning and initial characterization of the cDNA product associated with the t(1;19) translocation from a number of B-ALL cell lines [87, 129]. This newly described transcript encoded an 85 kDa protein, which was distinguishable from the more quickly migrat- ing wildtype E2A protein at 60 kDa. Analysis of the cDNA showed that two-thirds of the fusion transcript was derived from E2A and the remainder belonged to a somewhat obscure homeobox protein on chromosome 1, initially named Prl for pre-B cell leukemia (this was later changed to PBX1 ). Northern blotting also revealed two transcript variants of the E2A-PBX1 gene, which correlated to the long and short isoforms, currently referred to E2A-PBX1a and E2A-PBX1b, respectively [87] (Fig- ure 1.3. The long isoform contained 342 AA of the PBX1 sequence and the short isoform contained 259 AA. There was no change in the number of E2A-derived amino acids in either isoform of the fusion protein [87]. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 18

Figure 1.3: The E2A-PBX1 fusion protein

Figure 1.3. E2A-PBX1 is expressed as a result of the 1;19 chromosomal translocation. The fusion oncoprotein contains the activation domains (AD1-AD3) of E2A and the homeodomain (HD) of PBX1. There exist long (E2A-PBX1a) and short (E2A-PBX1b) isoforms that result from alternate splicing of the PBX1 portion. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 19

1.6.1 Wildype PBX1

Our most basic understanding of homeotic cluster (HOX ) genes, which are those genes involved in regulating developmental patterning, is derived from the model organism Drosophila. The Drosophila gene extradenticle was identified as a HOX cofactor by way of its ability to alter embryonic segment identity when mutated [139]. After its molecular cloning in 1993, extradenticle was discovered to share over 70% sequence identity with it human homolog PBX1. It was unclear at the time how this HOX protein could be specifically involved in human ALL, but it was suspected that a general disruption of developmentally-associated gene regulation was crucial to leukemogenic transformation [142]. The PBX1 gene encodes two alternatively spliced transcripts, PBX1a and PBX1b, which differ slightly in their C-terminus (Figure 1.4). The PBX1 gene products each contain an C-terminal homeodomain and a HOX cooperativity motif, each of which are required for DNA binding. As a monomer, PBX1 weakly binds DNA in vitro but forms dimers with a number of HOX proteins. These interactions increase the affinity of the PBX1/HOX heterodimer dimer for DNA by up to 30-fold [31]. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 20

Figure 1.4: The homeobox protein PBX1

Figure 1.4. PBX1 is a homeodomain-containing protein involved in regulating developmental processes, including musculo-skeletal morphology and hematopoiesis. There are two isofoms of PBX1, a result of alternate pre-mRNA splicing, which differ only in their C-terminal regions. The HCM, which promotes DNA binding and dimer formation with HOX proteins, has been demonstrated as necessary and sufficient for oncogenesis in a fibroblast transformation assay. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 21

HOX proteins are expressed in virtually all tissue types, including the hematopoietic system (Reviewed in [8]). Abundant expression of HOX genes, including HOXB3, HOXB4, and HOXA9 is observed in the HSC compartment (reviewed in [5]). Pbx1 null mice were used to further investigate the role of Pbx1 in hematopoiesis. Deletion of Pbx1 resulted in severe anemia, markedly reduced fetal liver hematopoiesis, and embryonic lethality at day 15-16. Pbx1 -null HSCs showed greatly reduced colony- forming abilities in vitro and were incapable of reconstituting all hematopoietic lineages of lethally-irradiated mice. Cell-cycle and clonogenic assays demonstrated a principal role for Pbx1 in promoting cell proliferation [41]. The precise mechanism of transcriptional regulation by PBX1 remains elusive but recent studies have defined roles for PBX1 as a ”pioneering transcription factor” [111]. Pioneering transcription factors are a class of DNA-binding proteins that have the unique ability to penetrate condensed chromatin and bind to sequence-specific gene regulatory regions. It is thought that binding by these pioneering transcription factors induces a more open chromatin conformation and promotes access to a broad variety of transcriptional regulators. The role for PBX1 has been demonstrated in promoting chromatin accessibility at muscle-specific promoters, which allows subsequent activation by the transcription factor MyoD [20]. More recently, PBX1 has been demonstrated as a pioneering transcription factor in breast cancer cells [111]. Using chromatin immunoprecipitation followed by next generation sequencing (ChIP- seq), it was shown that estrogen receptor alpha, a hormone-responsive transcriptional activator, is localized specifically to PBX1-bound regions of the genome. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 22

1.6.2 Wildtype E2A/TCF3

In 1989, Cornelis Murre and his colleagues in David Baltimore’s lab set out to identify proteins that would bind the E-box consensus sequences (5‘-CANNTG-3‘) found in the immunoglobulin kappa chain enhancer and promoter, a well-characterized genomic region that represented an experimentally tractable mammalian cis-regulatory element [127]. They screened a lambda phage expression library, built from the BJAB human B cell lymphoma cell line, using a DNA probe containing the kappa enhancer E-box sequence. Of the 500,000 phage plaques screened, one of the positive hits corresponding to the cDNA encoding E12 was identified in the first round of screening. The BJAB phage library was re-screened using the E12 cDNA as a hybridization probe, which further identified additional clones, including E47. RNA from a number of different cell lines, including T-cell lymphoma, HeLa, and neuroblastoma, were then shown to hybridize to the E12 cDNA probe, demonstrating that E-proteins are expressed in a wide variety of tissues. Sequence analysis at both the DNA and protein level showed that E12 and E47 were identical save for a small motif near the C-terminus, a difference attributable to alternate splicing of one exon. The al- ternately spliced C-terminal region encoded a bHLH domain that was shown to be required for DNA-binding [127]. E12 and E47, referred to hereafter as E2A, belong to the E-protein family of transcription factors, which is a small family of bHLH domain-containing transcriptional activators, that also includes E2-2 (also known as TCF4 ) and HEB (also known as TCF12 ) (Figure 1.5. These E-protein family members are typified by two highly-conserved N-terminal activation domains (AD1-AD3) and a C-terminal bHLH DNA-binding domain. The bHLH domain mediates DNA binding as well as homo- 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 23

and hetero-dimerization among E-protein family members. For example, E47 can bind E-box DNA sequences as a homodimer or as a heterodimer with E12, which does not eﬃciently homodimerize in B lymphoid cells. The real ingenuity of nature emerges when E-protein family members are able to interact with other members of the (b)HLH super family (Reviewed in [88]).

Figure 1.5: The E-protein family

Figure 1.5. The E-protein family consists of the E2A gene products, E12 and E47, as well as the closely-related E2-2 and HEB proteins. Each member has three conserved N-terminal activation domains and a C-terminal bHLH domain. The activation domains are important for its interaction with the histone acetyltransferase proteins CBP/p300 while the bHLH domain mediates binding to E-box DNA consensus sequences. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 24

The bHLH motif is found in over 240 distinct proteins spanning three kingdoms (plants, fungi, and animals) [10]. Members of the bHLH superfamily family include well known transcription factors such as Myc, MyoD, Twist, and HIF. These proteins can further be subcategorized by their primary sequence, preferred E-box recognition site, and tissue-speciﬁc expression. The capacity for E-proteins to form heterodimers with other bHLH proteins allows for combinatorial control over their context-speciﬁc gene regulation (Figure 1.6). For example, E-protein heterodimers containing the bHLH protein SCL (also known as TAL1) recognizes the E-box sequence, but unlike E-protein homo- or heterodimers, this interaction enforces target gene silencing due to the recruitment of transcriptional co-repressors by SCL. ID proteins, which are helix-loop-helix (HLH) family members lacking the basic region required for DNA- binding, can interact with E2A and other E-protein family members. Like SCL, ID proteins serve as negative regulators of E2A but instead of recruiting co-repressors to E-protein-bound sites, ID proteins simply abrogate E-box binding by E-proteins. It is through their amazing capacity for interacting with various partners that the ubiqui- tously expressed E-proteins are able to exert subtle context-dependent transcriptional regulation. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 25

Figure 1.6: Gene regulation by E-proteins and their antagonists

Figure 1.6. E-proteins bind E-box consensus elements as dimers and recruit coactivators to activate target gene transcription. Interactions between E-proteins and other bHLH transcription factors may induce either activation or silencing of the target gene, depending on the precise identity of the interaction partner. E-protein and ID heterodimers do not bind E-boxes since ID proteins lack the basic region required to interact with DNA. 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 26

1.6.3 Activation of E-protein target genes through p300 recruitment

I have previously described the bHLH domain that mediates both DNA-binding and interactions with other bHLH proteins. Next, I would like to describe the partnership with p300 that effects transcriptional activation by E-proteins. The primary amino acid sequence of the prototypical E-protein, E2A, indicates the presence of two transactivation domains, AD1 and AD2 (Figure 1.5). Shortly after it was first cloned, Aronheim and colleagues demonstrated, through the use of deletion mutants and reporter gene assays, that AD1 and AD2 of E2A contributed significantly to transactivation by E-proteins [9]. They further showed that the strength of each activation domain was cell-type specific in that AD1 was more active in a variety of mammalian cell lines (ovarian, liver, cervical) whereas transactivation via AD2 was stronger in pancreatic β cells. These findings reflect the capacity for cell-type specific factors or interaction partners to modulate transactivation by E2A [9]. The functional relationship between E2A and the histone acetyltransferase protein p300 was first demonstrated by Eckner and colleagues in 1996 [45]. They used CAT reporter assays to show that transcriptional activation of the IgH enhancer by E2A is significantly increased by co-transfection of exogenous p300. Through the use of supershift-electrophoretic mobility shift assay (EMSA), they then demonstrated the existence of a DNA-binding complex containing E2A and p300, which indicated a physical interaction between these proteins [45]. To further understand the relationship between E2A and p300, Curtis Bradney and colleagues examined the cooperation of these two factors at the biochemical and genetic level [22]. The results of their 2003 study showed that E2A co-elutes in a gel filtration assay with the histone acetyltransferase proteins CBP, p300, and PCAF 1.6. TRANSLOCATION 1;19 AND E2A-PBX1 27

from human pre-B cell (Nalm-6) nuclear extracts. By co-immunoprecipitation assays, they further demonstrated that E2A physically interacted in a complex containing the aforementioned HATs [22]. In a very interesting series of experiments in mice, they demonstrated that E2A and p300 collaborate to promote B lymphopoiesis in vivo. Specifically, they crossed E2A and p300 heterozygous deletion mutant mice, enumerated B220+CD19+ cells, and observed a cumulative decrease in the B lymphoid compartment of double heterozygous (i.e. E2A+/−p300+/−) knockout mice. Interestingly, the CD4+CD8+ T-cell compartment was unperturbed in any of the mutant genotypes, suggesting that the B lymphocyte compartment is more sensitive to perturbations in E2A and p300. Research in the LeBrun lab has focused on mapping the regions of E2A that interact with the p300 paralogue, CBP. First, pull-down experiments using GST-fusion proteins that encoded regions of E2A encompassing AD1 and AD2 identified AD1 to be the higher-affinity interaction domain [14]. Deletion of a highly conserved α- helical domain in AD1 almost completely abrogated binding of the GST-E2A fusion protein to CBP/p300, suggesting that this helical element mediated direct, high- affinity contact with CBP. To map the E2A-interaction domain of CBP, GST fusion proteins containing overlapping portions of CBP were used in pull-down assays. It was determined that a region of CBP encoding the KIX domain (also known as the CREB-binding domain), a well-characterized protein-protein interaction domain, was required and sufficient for binding to E2A. This was demonstrated to be a direct interaction through the use of purified recombinant E2A and CBP GST fusion proteins. The activation domains were further implicated in mediating transcriptional activation through the use of luciferase reporter and mammalian two-hybrid assays. 1.7. ONCOGENESIS BY E2A-PBX1: EXPERIMENTAL MODELS AND ONCOGENIC MECHANISMS 28

1.7 Oncogenesis by E2A-PBX1: experimental models and oncogenic mechanisms

To study the role of E2A-PBX1 in leukemogenesis, a focus formation assay was used initially as a tractable in vitro model of oncogenesis. In a focus formation assay, non-tumorigenic contact-inhibited NIH3T3 mouse fibroblasts are forced to express a potential oncogene and then culture dishes are screened for the development of densely packed foci. The number of foci formed is considered to be roughly proportional to the transforming potential of the suspected oncogene. While both the longer E2A- PBX1a and shorter E2A-PBX1b isoforms induced NIH3T3 tumor formation in nude mice, the short isoform was found to be much more tumorigenic in the context of focus formation assay and other in vitro tests for oncogenesis [123]. To study the domain- specific contributions to transformation, E2A-PBX1 deletion mutants were used in focus-formation assays. It was shown that NIH3T3 transformation was substantially reduced by deletion of the activation domain AD1. After establishing a transformative role for E2A-PBX1 in the context of a NIH3T3 focus formation assay, it was necessary to examine its effects on cellular transformation in a hematopoietic system. Initial attempts at engineering transgenic mice that expressed E2A-PBX1 under control of the lymphoid-specific immunoglobulin gene promoter caused, not the development of ALL, but T progenitor lymphomas in transgenic mice. Further attempts by Mark Kamps to ascribe a role for E2A-PBX1 in leukemogenesis showed that transplantation of E2A-PBX1-expressing bone marrow- derived hematopoietic precursors causes the outgrowth of myeloid leukemia in mice, rather than the lymphoid leukemia that is typically associated with the human disease. Kamps proceeded to demonstrate that infection of bone marrow cells with an 1.8. REGULATORY TARGETS OF E2A-PBX1 29

E2A-PBX1-expressing retrovirus causes the in vitro outgrowth of immortalized, immature myeloid precursors. These E2A-PBX1-expressing myeloid cells proliferated indefinitely in culture and were dependent on the myeloid cytokine GM-CSF. This observation was in contrast to what happened with uninfected, control myeloid cells that underwent terminal differentiation into neutrophils and macrophages within 4 weeks, speaking to the leukemogenic properties of E2A-PBX1 as both capable of inducing proliferation and preventing differentiation.

1.7.1 The t(1;19) event converts PBX1 into a constitutive transcriptional activator

The consensus DNA-binding sequence of PBX1 (5‘-ATCAATCAA-3’), was demonstrated to interact with its gene family relatives, PBX2 and PBX3 [100, 108]. In reporter gene assays, E2A-PBX1 strongly upregulated gene transcription downstream of this PBX1 binding sequence. Interestingly, while wildtype PBX proteins were able to bind the consensus DNA element in vitro, they were unable to activate transcription in a reporter assay. This demonstrated that the fusion with E2A fundamentally altered the function of PBX1, leading to the hypothesis that E2A-PBX1 induces leukemogenesis by binding PBX consensus DNA sequences and constitutively activating transcription of PBX1 target genes.

1.8 Regulatory targets of E2A-PBX1

Since its original identification, many efforts have been made to identify transcriptional targets of E2A-PBX1. Attempts to identify E2A-PBX1 target genes using representational difference analysis (RDA) in transduced NIH3T3 cells yielded few 1.8. REGULATORY TARGETS OF E2A-PBX1 30

results [56]. RDA analysis in a more relevant hematopoietic cell line demonstrated that the gene encoding the granulocyte colony-stimulating factor receptor (G-CSFr) was upregulated in E2A-PBX1-associated leukemia cell lines and primary t(1;19) ALL cells [39]. It was also found that enforced expression of E2A-PBX1 in (non- t(1;19)) REH cells caused an upregulation of G-CSFr detectable by flow cytometry. The G-CSFr seemed like a promising target for regulation by E2A-PBX1 because it contained an almost perfect PBX1 consensus element in its promoter. In another study, Fu and colleagues identified 5 mRNA transcripts by RDA that were upregulated in the E2A-PBX1-associated B ALL cell line 697 when compared to the non-t(1;19) line Nalm-6. The candidate target genes included ALDH-1, T- plastin (PLS3 ), Caldesmon (CALD1), EB-1, and EB-2, the latter three of which were consistently upregulated in primary, t(1;19)-associated ALL samples [57]. The authors focused their research efforts on describing EB-1, a transcript encoding a novel tyrosine kinase, which they implicated as an important E2A-PBX1-specific target upregulated during leukemogenesis. Shortly after the publication by Fu et al, RDA was used once again to identify potential targets of E2A-PBX1 using another t(1;19) cell line, KOPN-63. Wnt16 was identified to be transcriptionally upregulated in 3 t(1;19)-associated B-ALL cell lines but not in non-t(1;19)-associated B-ALL cell lines. Furthermore, t(1;19)-associated Wnt16 activity was demonstrated by the observed positive correlation between E2A-PBX1 and WNT16 expression but also with upregulated Frizzled receptor (Fz), a downstream Wnt target. Furthermore treatment with anti-sense oligonucleotides against both isoforms of E2A-PBX1 (a and b) resulted in a decrease in Wnt16 mRNA levels detected by northern blot, suggesting that Wnt16 is regulated, either directly or indirectly, by E2A-PBX1. In 1.9. CBP/P300 IN E2A-PBX1 ONCOGENESIS 31

2006, it was further conﬁrmed using E2A-PBX1 siRNA that Wnt16 and EB-1, but not EB-2, transcripts were downregulated in 697 and RCH-ACV cell lines, which both express E2A-PBX1 [28].

1.9 CBP/p300 in E2A-PBX1 oncogenesis

In 1994, work from Mike Cleary’s lab demonstrated that AD1 and AD2 of E2A- PBX1 were important for the transformation of NIH3T3 mouse fibroblasts and for the induction of thymic lymphomas in transgenic mice. Deletions of these regions were also shown to inhibit transactivation of an E2A-PBX1-responsive reporter gene and decreased focus formation in NIH3T3 fibroblasts [123]. It was later shown by the LeBrun lab that mutations in AD1 of E2A-PBX1 that reduced CBP binding also reduced its ability to confer enhanced serial replating capacity and immortalization to retrovirally infected hematopoietic progenitors [14]. Faced with abundant evidence of an essential role for AD1 in mediating the interaction between E2A-PBX1 and CBP/p300, work in the LeBrun lab became focused on finely mapping the key amino acid residues that mediate this interaction. The interaction was localized within an LxxLL motif, a common protein-protein interaction motif, located within AD1 [16]. Interestingly, disruption of the LxxLL motif through a single leucine-to-alanine substitution, abbreviated L20A, completely abrogated the ability of E2A-PBX1 to confer immortalization to hematopoietic precursors in vitro. Transplantation of hematopoietic precursors transduced with the E2A-PBX1 L20A mutant also failed to induce leukemia in lethally-irradiated recipient mice. Together, these experimental observations highlight the importance of the L20 leucine residue in mediating the interaction between E2A-PBX1 and CBP/p300. 1.10. CHIP AND MASSIVELY-PARALLEL SEQUENCING TECHNOLOGIES 32

1.10 ChIP and massively-parallel sequencing technologies

Chromatin immunoprecipitation (ChIP) is an antibody-mediated technique used to detect the in vivo interaction between proteins and DNA [7]. Briefly, cells are treated with formaldehyde to stabilize protein-DNA interactions. The cross-linked chromatin is then sheared by sonication into 250 bp to 1000 bp fragments, which are immunoprecipitated using an appropriate antibody. Un-bound chromatin is removed by sequential washing steps and the immunoprecipitated DNA is eluted and purified. Site-specific protein-DNA interactions can be detected using PCR or the DNA can be hybridized to a microarray (known as ChIP-chip) in order to study genome-wide protein binding [58, 79]. With the recent advent of high-throughput next-generation sequencing (NGS) technologies, which allows for nucleotide-level sequencing of short DNA strands, microarray hybridization was abandoned as the foremost method of studying genome-wide protein-DNA interactions [58]. ChIP followed by massively- parallel sequencing, known as ChIP-sequencing (abbreviated ChIP-seq), has many advantages over the traditional microarray approach, including higher resolution, increased genomic coverage, lower cost, reduced noise, and higher dynamic range [97, 136]. ChIP-seq has been extensively used in the ENCODE project, which has contributed greatly to our understanding of the role of transcription factors and functional histone modifications in tissue-specific gene regulation [82].

1.11 Rationale and Hypothesis

E2A-PBX1 has been extensively characterized in vitro as a p300 binding partner and transcriptinal activator of PBX1 reporter constructs. The in vivo DNA-binding characteristics, transcriptional targets, and the overall mechanism of E2A-PBX1-mediated 1.11. RATIONALE AND HYPOTHESIS 33

leukemogenesis remains to be explored in t(1;19)-associated human leukemia cells. The identiﬁcation of the transcriptional targets of E2A-PBX1 will allow for a more comprehensive understanding of its role in leukemia development and may help in the discovery of druggable targets that can be antagonized during the treatment of t(1;19)-associated leukemia.

1.11.1 Hypothesis:

My hypothesis is that E2A-PBX1 binds PBX1 consensus elements in the promoter region of its targets where it recruits p300 and promotes constitutive upregulation of endogenous PBX1 target genes (Figure 1.7).

1.11.2 Experimental objectives

To address this question, I have performed RNA-seq and ChIP-seq for E2A-PBX1, p300, and the modiﬁed histones H3K4me1, H3K4me3, and H3K27me3 in the t(1;19)- associated cell line RCH-ACV. 1.11. RATIONALE AND HYPOTHESIS 34

Figure 1.7: Canonical model of E2A-PBX1-mediated leukemogenesis

Figure 1.7. E2A-PBX1, upon its expression in t(1;19)-associated leukemia cells, localizes to PBX1 consensus elements upstream of PBX1 target genes. E2A-PBX1 recruits p300 to these sites and induces constitutive activation of genes involved in leukemia progression. 35

Chapter 2

Materials and Methods

2.1 Cell lines and culture conditions

697, Raji, REH, and RCH-ACV cells were cultured in RPMI (Lonza) supplemented with 10% FBS (PAA), penicillin (100 U/ml), streptomycin (100 U/ml), and Fun- gizone (0.25 µg/ml). The triply-FLAG-tagged E2A-PBX1b-expressing RCH-ACV- derived cell line was generated by Dick Bayly. RCH-ACV cells were ﬁrst infected with an MSCV-neo-3xFLAG-E2A-PBX1b retroviral vector. Infected cells were positively selected by adding the antibiotic G418 to the culture media. Resistant clones were screened for expression of the transgene by western blot. A single clone that expressed 3xFLAG-E2A-PBX1b was used in this study.

2.2 Chromatin Immunoprecipitation- quantitative PCR (ChIP-qPCR)

Chromatin immunoprecipitation was performed as described in [128] with the excep- tion that the eluted DNA was then purified on a PCR purification column (Qiagen) according to the manufacturer’s instructions. Quantitative PCR (qPCR) analysis 2.3. NEXT-GENERATION SEQUENCING AND BIOINFORMATICS ANALYSIS 36 was performed using the Fast SYBR Green Master Mix (Life Technologies) and gene- specific primers (Operon) according to the manufacturer’s instructions. Unless otherwise noted, all ChIP-qPCR signals were normalized relative to an input sample (i.e. ”% Input”), which refers to a sample of chromatin that has not undergone immunoprecipitation. Using qPCR Ct values the calculation was performed as follows:

( 100∗2 CtInput−CtIP ). In cases where ChIP-qPCR data is reported as fold-enrichment (e.x. Figure 3.2A), this is to be interpreted as the ChIP eﬃciency in RCH-ACV cells that express FLAG-E2A-PBX1 relative to parental RCH-ACV cells, which do not express FLAG-E2A-PBX1.

2.3 Next-generation sequencing and bioinformatics analysis

DNA for ChIP-seq was harvested according to published protocols [138]. Total RNA was harvested by column purification (Qiagen) according to the manufacturer’s instructions. ChIP DNA and total RNA from RCH-ACV cells were sent to China for next-generation sequencing, which was performed at the Beijing Genomics Institute (BGI) on the Illumina Hi-seq platform using standard protocols. The sequencing data (reads) were output as FASTQ files, upon which an integrative bionformatics workup was performed. ChIP-seq reads were mapped to the human genome, UCSC build hg18, using Bowtie2 with default settings. Due to the low signal-to-noise ratio of our FLAG- E2A-PBX1 ChIP-seq data, we employed an atypical approach to peak-identification using a set of ”reference” ChIP-seq peaks identified by a collaborator in another E2A- PBX1-associated cell line, 697 [62]. The Roeder group performed N-terminal E2A and C-terminal PBX1 ChIP-seq in 697 cells, E2A and PBX1 peaks were identified 2.3. NEXT-GENERATION SEQUENCING AND BIOINFORMATICS ANALYSIS 37 separately and regions of overlapping peaks were considered E2A-PBX1 binding sites; 2823 peaks were identified in their study. They shared with us unpublished ChIP-seq signal data, which was used to assist with E2A-PBX1 peak identification in RCH- ACV cells. The peak-calling algorithm was performed as follows: 1. For each of the 2823 E2A-PBX1 binding peak identified in 697 cells by the Roeder group, the bioinformatics software package HOMER was used to calculate the FLAG ChIP-seq signal intensity at each genomic coordinate in both our FLAG-E2A-PBX1-expressing RCH-ACV cells and the parental, negative control cells [76]. 2. FLAG peaks with a signal intensity equal to or greater than 3.0 in the FLAG-E2A-PBX1 ChIP-seq track were considered for further analysis visual inspection; this cohort included 1161 of the original 2823 peaks identified by the Roeder group. 3. Visual inspection (performed independently by K. Nanan and D. LeBrun) was used to identify peaks that were a) present in E2A and PBX1 ChIP-seq tracks in 697 cells, b) present in the FLAG track from FLAG-E2A-PBX1-expressing RCH-ACV cells, and c) not present in the control RCH-ACV cells that do not express FLAG-E2A-PBX1 (an example image is included in Results section). Following this analysis, 555 high-confidence E2A-PBX1 peaks were identified as common to both RCH-ACV and 697 cells. RNA-seq data were processed using the ”Tuxedo” suite of bioinformatics software according to published protocols [172]. The data were aligned to the human genome, UCSC build hg18, using TopHat alignment software and the default settings. Transcriptome assembly and mRNA expression was estimated using Cufflinks and differential expression analysis was computed using Cuffdiff. HOMER was used to identify the overrepresentation of transcription factor binding sites, calculate fragment densities, and generate tag density histograms. The ReactomeFI Cytoscape 2.4. SIRNA-MEDIATED GENE KNOCKDOWN 38

plugin was used for network generation and visualization.

2.4 siRNA-mediated gene knockdown

The PBX1 -specific siRNA oligonucleotides used in this study were previously described in [111]. siRNA was diluted and hybridized according to the manufacturer’s instructions (Operon). siRNA were transfected into RCH-ACV cells using RNAiMAX according to the manufacturer’s instructions at a final concentration of 5 nM (Life Technologies). Cells were harvested 24 hours after siRNA treatment and knockdown efficiency was measured by FLAG immunoblotting and E2A-PBX1-specific qRT-PCR.

2.5 Immunoblotting

Cells were lysed in NP-40 buffer (0.5% NP-40, 150 mM NaCl, 50 mM Tris-Cl;pH 8.0) and briefly sonicated to disrupt the nuclear envelope. Cell debris was pelleted by centrifugation and the protein concentration in the supernatant was quantified by DC protein assay (Bio-Rad). 2-5 µg of protein was loaded on a 10% polyacrylamide- SDS gel and electrophoresed in Tris-glycine running buffer. Proteins were transferred to a nitrocellulose membrane in a Tris-glycine transfer buffer. After transferring, the membrane was blocked for 30 minutes in Tris-buffered saline containing 0.5% Tween-20 (TBS-T) and 5% BSA. Antibodies were added at the recommended dilution (between 1:1000 and 1:5000) and the membrane was incubated overnight with rocking at 4 degrees C. Subsequently, the membrane was washed with repeatedly with TBS- T a total of 5 times and then incubated at room temperature for two hours with a horse-radish-peroxidase-conjugated secondary antibody diluted in TBS-T with 5% 2.6. ANTIBODIES USED FOR CHIP, CHIP-SEQ, AND IMMUNOBLOTTING 39

BSA (1:10000 dilution). The secondary antibody was washed with 5 treatments of TBS-T, the membrane was incubated with a chemiluminescence reagent and then exposed to X-ray ﬁlm.

2.6 Antibodies used for ChIP, ChIP-seq, and immunoblotting

An anti-FLAG antibody (Sigma-Aldrich cat. F1804) was used to detect 3xFLAG- E2A-PBX1b in immunoblotting, ChIP-seq, and ChIP-qPCR assays. Antibodies against p300 (Santa Cruz cat. sc-585), H3K4me1 (Millipore cat. 07-436) , H3K4me3 (Milli- pore cat. 07-473), and H3K27me3 (Millipore cat. 07-449) were also used for ChIP- seq. E2A (Santa Cruz cat. sc-349), EBF1 (Santa Cruz cat. sc-137065), and H3K27Ac (Abcam cat. ab4729) antibodies were used for ChIP-qPCR.

2.7 RNA harvest, cDNA synthesis, and qRT-PCR

RNA was harvested using the RNeasy mini kit according to the manufacturer’s instructions (Qiagen). RNA was converted into single-stranded complementary DNA (cDNA) using the M-MLV reverse transcriptase enzyme (Sigma-Aldrich) and random hexamers (Operon). The resulting cDNA was used in as the template in quantitative PCR (qPCR) reactions to measure gene expression levels. E2A-PBX1 and GAPDH were detected using TaqMan probe and primer set while all other qRT-PCR experiments were performed using SYBR green and un-modiﬁed gene-speciﬁc primers. Relative expression levels were calculated using the ∆∆CT method, normalized to GAPDH expression. 2.8. STATISTICAL ANALYSES 40

2.8 Statistical analyses

Unless otherwise indicated, statistical analyses were performed using the R software package. The Wilcoxon signed-rank test was used to calculate p-values in ﬁgure 3.6. Box and whisker plots display, from top to bottom, the upper extreme, upper quartile (75th percentile), median, lower quartile (25th percentile), and lower extreme. All error bars represent one standard deviation above and below the mean. 41

Chapter 3

Results

3.1 Identiﬁcation and validation of E2A-PBX1 binding sites in RCH- ACV cells

The human ALL cell line RCH-ACV was used as the main experimental model in this study. RCH-ACV cells were established from a bone marrow sample of a young child diagnosed with ALL [83]. Karyotype analysis identified the 1;19 translocation and trisomy of chromosome 8 in RCH-ACV cells [83]. To identify E2A-PBX1 binding sites, stable expression of a triply FLAG-tagged copy of E2A-PBX1b was established in an RCH-ACV cell line. Next, ChIP was performed using an anti-FLAG antibody and the immunoprecipitated DNA was used as the template in a next-generation sequencing reaction. As a control for non-specific binding of the antibody, anti-FLAG ChIP-seq was also performed in the parental RCH-ACV cell line. A high level of background ”noise” was observed in the FLAG ChIP-seq track from the parental RCH-ACV cells, which express no FLAG epitope. This made it difficult for peak- calling software to effectively identify regions of FLAG-E2A-PBX1b enrichment. To 3.1. IDENTIFICATION AND VALIDATION OF E2A-PBX1 BINDING SITES IN RCH-ACV CELLS 42 bypass this impediment and bolster the reliability of peaks called, we used a combination of bioinformatics tools and visual inspection by two independent observers (K. Nanan and D. LeBrun) to identify a subset of E2A-PBX1 peaks identified by a collaborator in the t(1;19)-associated 697 cell line, which showed specific enrichment in RCH-ACV cells (example of accepted and rejected peaks in Figure 3.1) [62]. This approach, which yielded a set of 555 high-confidence E2A-PBX1 binding peaks, had the advantage of identifying the most prominent peaks in RCH-ACV cells that were also enriched in a complementary t(1;19)-associated cell line. 3.1. IDENTIFICATION AND VALIDATION OF E2A-PBX1 BINDING SITES IN RCH-ACV CELLS 43

Figure 3.1: Example of peaks submitted for visual inspection

Figure 3.1. Geng and colleagues identified E2A-PBX1 genomic binding sites by performing ChIP-seq in 697 cells using E2A and PBX1 antibodies that recognized the N- and C-termini of E2A-PBX1, respectively; the overlapping regions signified E2A- PBX1 binding [62]. E2A-PBX1 peaks identified in 697 cells were used as ”reference coordinates” for the identification of FLAG-E2A-PBX1 peaks in RCH-ACV cells. Vi- sual inspection involved comparing the FLAG ChIP-seq peak intensity and morphology at potential E2A-PBX1 binding sites in RCH-ACV cells expressing FLAG-E2A- PBX1 (blue) vs. the parental RCH-ACV cell line (black). In the above example, an E2A-PBX1 binding site is considered present at the E2F2 gene since it is enriched in RCH-ACV cells that do vs. cells that do not express FLAG-E2A-PBX1. Conversely, enrichment of E2A-PBX1 is not observed at the DNMT3L gene when comparing RCH-ACV cells that do vs. cells that do not express FLAG-E2A-PBX1. 3.1. IDENTIFICATION AND VALIDATION OF E2A-PBX1 BINDING SITES IN RCH-ACV CELLS 44

To validate binding of E2A-PBX1 to the bioinformatically-identified genomic sites identified above, 10 binding peaks were chosen at random (using a random number generator) and ChIP-qPCR was performed using primers specific for the putative E2A-PBX1 binding sites (Figure 3.2A). Anti-FLAG ChIP was performed in RCH- ACV cells stably expressing the FLAG-E2A-PBX1b transgene; ChIP in the parental RCH-ACV cell line served as a normalization control. The CALD1 -associated site was shown previously to be bound by E2A-PBX1 and was included as a positive control [62]. Conversely, the CXCR5 and BCL6 loci are not bound by E2A-PBX1 according to our ChIP-seq data and were, therefore, used as a negative control sites. ChIP-qPCR results show specific enrichment of FLAG-E2A-PBX1 at each candidate binding site in the ”validation set.” Furthermore, the average fold-enrichment in the validation set is higher than the average fold-enrichment for the negative control sites (student’s T-test p < 0.05). Anti-PBX1 ChIP-qPCR was also performed in a panel of 1;19-translocated (RCH-ACV and 697) and non-1;19-translocated (REH and Raji) B lymphocyte cell lines (Figure 3.2B). The results of this assay demonstrate that genomic binding of E2A-PBX1, at confirmed targets (Figure 3.2A), is specifically associated with the 1;19 translocation and E2A-PBX1 expression. Together, these ChIP-qPCR data demonstrate 1) enrichment of E2A-PBX1 at 10 randomly selected binding sites that are, on average, statistically significant when compared to negative control sites, and 2) specific enrichment of E2A-PBX1 binding at known E2A-PBX1 binding sites in t(1;19) vs non-t(1;19)-associated B lymphocyte cell lines. The results presented above, therefore, justify further bioinformatic characterization of all 555 E2A-PBX1 binding sites as a unified group. 3.1. IDENTIFICATION AND VALIDATION OF E2A-PBX1 BINDING SITES IN RCH-ACV CELLS 45

Figure 3.2: Validation of high-conﬁdence E2A-PBX1 binding peaks using ChIP- qPCR 3.1. IDENTIFICATION AND VALIDATION OF E2A-PBX1 BINDING SITES IN RCH-ACV CELLS 46

Figure 3.2. (A) Ten of the 555 bioinformatically-identified E2A-PBX1 binding sites, referred to as the ”validation set,” were chosen at random for validation by gene-specific anti-FLAG ChIP-qPCR. The fold-enrichment of FLAG-E2A-PBX1 at each binding site was normalized to ”background” levels of enrichment, attributable to non-specific antibody binding, and ascertained by performing anti-FLAG ChIP-qPCR in parental RCH-ACV cells that do not express FLAG-E2A-PBX1. The CXCR5 and BCL6 loci, neither of which are bound by E2A-PBX1 according to our ChIP-seq data, were used as negative control sites; the CALD1 site was previously shown to be bound by E2A-PBX1 and was consequently used as a positive control site. The label below each column refers to the gene whose TSS is nearest to the E2A-PBX1 binding site being studied. The bracketed numbers refer to the distance between the E2A-PBX1 binding site and the TSS of the associated gene (e.x. the E2A-PBX1 binding site assayed in column 1 is located 12 kilobases upstream of the TSS of the CALD1 gene). E2A-PBX1 binding was shown to be higher at all binding sites in the validation set compared to the negative control sites. A student’s T-test demonstrates a statistically significant difference in mean enrichment of E2A-PBX1 at the validation set compared to the negative controls (p value < 0.05). (B) PBX1 ChIP-qPCR was performed in a panel of t(1;19)-associated and non-t(1;19)-associated B lymphocyte cell lines. The t(1;19)-associated cells used in this assay include RCH-ACV and 697 cells; non-t(1;19)-associated cell lines included the t(12;21)-associated REH and the Burkitt’s Lymphoma-derived Raji cell line. Enrichment of E2A-PBX1 was observed only in B lymphocyte cell lines associated with the 1;19 translocation and E2A- PBX1 expression. ChIP efficiency, representing the amount of DNA recovered in each immunoprecipitation, is normalized to the input sample, which refers to cells that have not been subject to immunoprecipitation. 3.2. E2A-PBX1 PREDOMINANTLY LOCALIZES TO INTRONIC AND INTERGENIC REGIONS 47

3.2 E2A-PBX1 predominantly localizes to intronic and intergenic regions

We surveyed the genomic distribution of 555 high-conﬁdence E2A-PBX1 binding sites using the bioinformatics software package HOMER. Genomic annotation showed that 50% of E2A-PBX1 binding peaks (278 of 555) were localized to intronic regions while 38% (211 of 555) were found in intergenic regions (Figure 3.3A). The distribution of the distances between an E2A-PBX1 binding site and the transcriptional start site (TSS) of associated genes is shown in Figure 3.3B.

Figure 3.3: Genomic binding of E2A-PBX1

Figure 3.3. (A) Genomic annotation of E2A-PBX1 ChIP-seq peaks shows that most E2A-PBX1 binding peaks are found within intronic and intergenic regions. ”Promoters” are deﬁned as regions 3 kb upstream or downstream of the TSS of a gene. Gene bodies exclude the promoter region but include the 5‘ and 3‘ UTRs, coding exons, and introns. Intergenic regions are categorized as those occurring immediately upstream of the promoter region and downstream of the 3‘ UTR of a gene. (B) The peak-TSS histogram shows that E2A-PBX1 is often located distally (>5kb) from the TSS of its potential target genes. 3.3. BIOLOGICAL RELEVANCE AND FUNCTIONAL INTERCONNECTIVITY OF E2A-PBX1 TARGET GENES 48

3.3 Biological relevance and functional interconnectivity of E2A-PBX1 target genes

To attribute biological relevance to the genomic localization of E2A-PBX1, the online annotation and ontology database tool GREAT was used [117]. GREAT is uniquely equipped to process high-throughput genomic binding data (e.x. ChIP-seq) and identify region-gene associations for factors that may bind at distal enhancers up to 1 Mb from a gene TSS. E2A-PBX1 regulatory targets are deﬁned by GREAT as the nearest gene whose transcriptional start site (TSS) lies within 1 Mb of an E2A-PBX1binding site. The top ten disease ontology (DO) and mouse phenotype (MP) terms identi- ﬁed by GREAT are listed in Tables 3.1 and 3.2, respectively. Of note is the striking abundance of DO and MP terms associated with hematopoietic disorders, aberrant lymphopoiesis, and neoplasia. 3.3. BIOLOGICAL RELEVANCE AND FUNCTIONAL INTERCONNECTIVITY OF E2A-PBX1 TARGET GENES 49

Rank Description P-value FDR 1 cardiovascular system disease 2.03E-12 4.53E-09 2 hematologic cancer 2.96E-12 3.30E-09 3 hematopoietic system disease 5.43E-12 4.04E-09 4 cancer 1.85E-11 1.04E-08 5 neoplasm of lymphatic and hemopoietic tissue 1.96E-11 8.76E-09 6 cancer by anatomical entity 3.06E-10 1.14E-07 7 lymphatic system disease 4.52E-10 1.44E-07 8 lymphoproliferative disease 5.75E-10 1.61E-07 9 immunoproliferative disease 5.78E-10 1.44E-07 10 leukemia 6.04E-10 1.35E-07

Table 3.1: Disease ontology terms associated with candidate E2A-PBX1 target genes

Rank Description P-value FDR 1 abnormal bone marrow cell morphology/development 1.03E-24 7.54E-21 2 abnormal myeloblast morphology/development 1.50E-23 5.49E-20 3 abnormal mononuclear cell diﬀerentiation 3.09E-22 7.51E-19 4 abnormal lymphopoiesis 3.34E-21 6.09E-18 5 abnormal lymphocyte morphology 4.71E-21 6.88E-18 6 abnormal mononuclear cell morphology 1.51E-20 1.83E-17 7 abnormal leukopoiesis 2.28E-20 2.37E-17 8 abnormal blood cell morphology/development 1.07E-19 9.72E-17 9 abnormal hematopoiesis 3.54E-19 2.87E-16 10 abnormal leukocyte morphology 4.00E-19 2.92E-16

Table 3.2: Top-ranked mouse phenotype terms associated with candidate E2A-PBX1 target genes 3.3. BIOLOGICAL RELEVANCE AND FUNCTIONAL INTERCONNECTIVITY OF E2A-PBX1 TARGET GENES 50

To visualize and summarize the functional interconnectedness of candidate E2A- PBX1 target genes, the Cytoscape interaction analysis plugin, ReactomeFI, was used (Figure 3.4) [74, 157]. Pathway analysis within ReactomeFI identified candidate E2A-PBX1 target genes associated with B-cell receptor signaling (BLNK, CD19, CD79B; FDR 2.00e-03), primary immunodeficiency (IGLL1, RAG1 ; FDR 1.00e-03), and transcriptional misregulation in cancer (CDKN1A, ERG, WT ; FDR 1.00e-03). Cross-referencing of the NCI Cancer Gene Index, a catalogue of genes experimentally associated with human cancers, ReactomeFI identified 22% of the network components as being associated with ALL and 53% as being associated with all malignant neoplasms. A querying of the gene-drug interaction database (GDIdb) further identified a total of 21 E2A-PBX1 target genes that are targetable with anti-neoplastic drugs ([68]). Together, these data support a role for E2A-PBX1 as a modifier of normal hematopoietic gene regulation, specifically in lymphoid processes. We also identified a number of E2A-PBX1 regulable genes whose activity can be modified by anti-neoplastic drugs, which may represent potential targets for anti-leukemia therapy. 3.3. BIOLOGICAL RELEVANCE AND FUNCTIONAL INTERCONNECTIVITY OF E2A-PBX1 TARGET GENES 51

Figure 3.4: Functional interaction network of E2A-PBX1 target genes 3.3. BIOLOGICAL RELEVANCE AND FUNCTIONAL INTERCONNECTIVITY OF E2A-PBX1 TARGET GENES 52

Figure 3.4. Each E2A-PBX1 target gene with at least one experimentally de- ﬁned physical or functional interaction is represented in the network above. KEGG pathway enrichment analysis shows over-representation of genes involved in adipocyte and PPAR signaling, B-cell receptor (BCR) signaling and immunodeﬁciency, regulation of the actin cytoskeleton, and in proliferation and cell cycle regulation. Network components listed in the NCI Cancer Gene Index (diamonds) and those for which anti- neoplastic drugs exist (red text) are highlighted. A high-resolution version of this image is available for download at: http://goo.gl/K0cyWm or http://1drv.ms/Wap2ig (click ”Show full resolution” or ”View Original” to view in greatest detail). 3.4. E2A-PBX1 BINDING IS ASSOCIATED WITH P300 RECRUITMENT AND TRANSCRIPTIONAL ACTIVATION 53

3.4 E2A-PBX1 binding is associated with p300 recruitment and transcriptional activation

E2A-PBX1 has been shown to interact physically with the KIX domain of the co- activator and histone acetyltransferase CBP/p300 [14]. Perturbations in the binding interface between E2A-PBX1 and p300 show that this interaction is important for transcription of a downstream reporter gene in the context of a luciferase assay [14]. During oncogenic transformation, it is our hypothesis that E2A-PBX1 binds to open, actively transcribed chromatin regions upon its expression in hematopoietic progenitors and perpetuate the aberrant expression of genes involved in leukemogenesis. To determine whether p300 co-localizes with E2A-PBX1 at its genomic binding sites and to assess the local epigenetic correlates of E2A-PBX1 binding, the genomic distributions of p300 and the functional histone marks H3K4me1, H3K4me3, and H3K27me3 were mapped using ChIP-seq and bioinformatics (Figure 3.5). ChIP-seq tag density plots show strong localized enrichment of p300 near E2A- PBX1 binding peaks (Figure 3.5A). This result indicating co-recruitment of p300 with E2A-PBX1 across the genome complements the published experimental results in supporting a role for CBP/p300 in E2A-PBX1 oncogenesis [14, 16]. To investigate whether the enrichment of p300 at these genomic loci was speciﬁc to t(1;19)- associated cells, the p300 distribution at equivalent genomic loci in other lymphoid and non-lymphoid cell lines was determined using data from the ENCODE project [168]. Interestingly, p300 occupancy appears to be enriched at the corresponding loci in the B lineage lymphoblastoid cell line GM12878 but not in epithelium-derived HeLa cells (Figure 3.5A). These data showing that p300 is recruited to similar genomic loci in t(1;19)-associated cells and normal lymphocytes indicate that p300 localization to 3.4. E2A-PBX1 BINDING IS ASSOCIATED WITH P300 RECRUITMENT AND TRANSCRIPTIONAL ACTIVATION 54 these sites can occur independently of E2A-PBX1 and appears to be selective for B-lymphoid cells

Figure 3.5: E2A-PBX1 binding sites are associated with recruitment of p300 and enrichment for histone marks associated with transcriptional activity

Figure 3.5. p300 ChIP-seq tag density is enriched near the center of E2A-PBX1 binding peaks (i.e. point 0 on the horizontal axis) in RCH-ACV cells. p300 is also enriched at the corresponding sites in GM12878 but not HeLa cells. (B) E2A-PBX1 binding correlates with increased occupancy of the activating histone modiﬁcations H3K4me1 and H3K4me3, but not the silencing mark H3K27me3. 3.4. E2A-PBX1 BINDING IS ASSOCIATED WITH P300 RECRUITMENT AND TRANSCRIPTIONAL ACTIVATION 55

To determine whether E2A-PBX1 genomic localization affects local chromatin, we mapped the tag density of modified histones surrounding E2A-PBX1 binding sites in RCH-ACV cells (Figure 3.5B). Normalized tag density histograms showed that E2A-PBX1 binding sites in RCH-ACV cells are associated with increased occupancy of H3K4me1 and H3K4me3 but not H3K27me3 (Figure 4B). H3K4me1 is a modification associated with distal enhancer regions, H3K4me3 is predominantly associated with proximal promoter elements, and H3K27me3 is a mark of poised or silenced chromatin [13, 37, 69, 190]. In RCH-ACV cells, H3K4me1 and H3K4me3 both assume symmetrical distributions centered upon E2A-PBX1 binding sites with enrichment around 500 bp on either side of the E2A-PBX1 peak center. The depression in the binding profile may represent regions of nucleosome-depletion, which has been shown to correlate with the binding of certain transcriptional regulators [168, 179]. Using RNA-seq data, proximal binding of E2A-PBX1 to the TSS of nearby genes was shown to correlate with increased gene expression in RCH-ACV cells (Figure 3.6). ENCODE RNA-seq data showed a similar upregulation of genes near an E2A-PBX1 binding site in the GM12878 cell line, but not in HeLa cells. Together, these data suggest that E2A-PBX1 localizes preferentially to genes that are active in B-lymphoid cells. 3.4. E2A-PBX1 BINDING IS ASSOCIATED WITH P300 RECRUITMENT AND TRANSCRIPTIONAL ACTIVATION 56

Figure 3.6: Proximate binding of E2A-PBX1 is associated with diﬀerentially increased gene transcription selectively in t(1;19)-associated cells

Figure 3.6. Box and whisker diagrams are used depict levels of gene transcription based on RNA-sequencing data collected from RCH-ACV, GM12878, and HeLa cells. RNA-seq analysis shows that promoter-proximal binding of E2A-PBX1 is associated with increased gene expression in RCH-ACV cells relative to GM12878 and HeLa cells, which do not express E2A-PBX1. The values below each column (e.x. 0-5 kb) refer to the distance between an E2A-PBX1 binding peak and the TSS of the nearest gene. ”All other” refers to genes whose TSS are beyond 1 Mb from an E2A-PBX1 binding site. Box and whisker plots display, from top to bottom, the upper extreme, upper quartile (75th percentile), median, lower quartile (25th percentile), and lower extreme. Statistical signiﬁcance calculated using the Wilcoxon p-value 3.5. E2A-PBX1 BINDING SITES POSSESS THE CHARACTERISTICS OF ACTIVE TRANSCRIPTIONAL ENHANCERS 57

3.5 E2A-PBX1 binding sites possess the characteristics of active transcriptional enhancers

Transcriptional enhancers are gene regulatory regions that can be located at great distances from the promoter of their regulatory targets [145]. Active enhancers are differentiated from inactive or poised enhancers by the occupancy of p300 and the functional histone modification H3K27Ac [37, 176]. Bioinformatic characterization of E2A-PBX1 genomic binding sites demonstrated that the majority of E2A-PBX1 binding sites are found between 5 kb and 50 kb away from the TSS of the nearest gene (Figure 3.3B) and are enriched for p300 binding (Figure 3.5A), thereby fulfilling two of the characteristics of active enhancers. To further investigate the potential correlation between E2A-PBX1 genomic bind sites and enhancer activity, H3K27Ac ChIP-qPCR was performed in RCH-ACV cells. H3K27Ac was found to be markedly enriched at many E2A-PBX1 binding peaks that were assessed (Figure 3.7). These result suggests that E2A-PBX1 may frequently localize to genomic regions that contain active transcriptional enhancers in RCH-ACV cells. 3.5. E2A-PBX1 BINDING SITES POSSESS THE CHARACTERISTICS OF ACTIVE TRANSCRIPTIONAL ENHANCERS 58

Figure 3.7: Enrichment of H3K27Ac at E2A-PBX1 binding sites

Figure 3.7. ChIP-qPCR was used to measure enrichment of the active enhancer mark, H3K27Ac, at E2A-PBX1 binding sites. Most of the sites tested showed enrichment for this functional histone modiﬁcation when compared with the negative control locus, CXCR5, which was not expected to be enriched for H3K27Ac. 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 59

3.6 E2A-PBX1 associates with genomic sites bound by the B lymphopoietic transcription factors E2A and EBF1

All 555 E2A-PBX1 binding sites were submitted to the bioinformatics software suite HOMER and subjected to the program’s motif scanning algorithm used to identify over-represented motifs. We expected that a high prevalence of PBX1 consensus sequences would be found near E2A-PBX1 binding sites but found that PBX1 DNA consensus elements were present in merely 1.42% of E2A-PBX1-bound DNA fragments (Figure 3.8A). Instead, we observed enrichment of a number of hematopoietic transcription factor consensus DNA elements, including wildtype E2A and EBF1 motifs at prevalences of 51.15% and 31.42%, respectively. Furthermore, a plot of the frequency of transcription factor motifs across E2A-PBX binding peaks showed a striking enrichment of DNA recognition sites for E2A and EBF1, but not PBX1, near E2A-PBX1 binding sites (Figure 3.8). This surprising result suggests that: (1), E2A-PBX1 may interact, directly or indirectly, with E2A or EBF1 at their respective genomic targets; and, (2), the homeodomain of E2A-PBX1 may not interact directly with DNA. 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 60

Figure 3.8: Wildtype E2A and EBF1, but not PBX1, motifs are enriched at E2A- PBX1 binding sites

Figure 3.8. Motif scanning analysis of E2A-PBX1 binding sites identiﬁed signiﬁ- cant enrichment (P < 1e-50) of transcription factor motifs for wildtype E2A, EBF1, ETV1, and RUNX1. PBX1 consensus elements were largely absent at E2A-PBX1 binding sites. (B) Plotting the frequency of consensus transcription factor recognition motifs across E2A-PBX1 binding sites shows that motifs for E2A and EBF1, but not PBX1, coincide with E2A-PBX1 peaks. 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 61

Our results raised the question as to whether E2A-PBX1 is required for recruiting EBF1 or wild-type E2A to E2A-PBX1 binding sites. We then wondered whether E2A and EBF1 co-localize at the corresponding genomic loci in normal B lymphocytes in the absence of E2A-PBX1. To this end, we plotted the ChIP-seq tag density for E2A and EBF1 in the B lymphocyte cell line GM12878 to determine whether these factors occupied the corresponding E2A-PBX1 binding sites in a non-t(1;19)- associated B lymphoid cell line (Figure 3.9A). We found in GM12878 cells that both wildtype E2A and EBF1 were enriched at a subset of genomic loci that were originally characterized as E2A-PBX1 binding peaks in RCH-ACV cells. Heatmap clustering shows that nearly half of the sites analyzed are occupied by EBF1 and a subset of those are further co-occupied by E2A (Figure 3.9B). Therefore, localization in B- lymphoid cells of E2A and EBF1 to sites bound by E2A-PBX1 in t(1;19) ALL cells does not require E2A-PBX1. 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 62

Figure 3.9: E2A and EBF1 co-localize in GM12878 cells at genomic regions corresponding to E2A-PBX1 binding sites

Figure 3.9. (A) Fragment density plots using ENCODE ChIP-seq data from GM12878 cells show abundant enrichment of wildtype E2A and EBF1 protein at sites corresponding to E2A-PBX1 genomic binding sites. (B) Heatmap clustering using seqMINER shows that nearly half of the E2A-PBX1 binding peaks, identiﬁed by ChIP-seq with anti-FLAG in RCH-ACV cells (ﬁrst column), are bound by E2A and EBF1 in GM12878 cells (Cluster 2 in third and fourth columns, respectively). 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 63

The above results provide bioinformatic evidence suggesting that E2A-PBX1 might co-localize with the wildtype, B-lymphopoietic transcription factors EBF1 and/or E2A at a substantial number of sites. In order to investigate this possibility experimentally, site-speciﬁc ChIP-qPCR was carried out. The results indicate that E2A and EBF1 can localize to E2A-PBX1 binding sites in RCH-ACV cells (Figure 3.10A). This suggests that E2A, EBF1, and E2A-PBX1 may cooperate to regulate gene expression or alter the local chromatin environment near E2A-PBX1 binding sites. To determine whether E2A-PBX1 and EBF1 could simultaneously occupy the same genomic region, sequential ChIP (also known as ”re-ChIP”) was used (Figure 3.10B). In this experiment, FLAG-tagged E2A-PBX1 was ﬁrst immunoprecipitated using an anti-FLAG antibody. The bound chromatin was eluted from the beads and subject to another round of immunoprecipitation using an EBF1 antibody. After washing and eluting, the recovered DNA was used in a qPCR reaction. Sequential ChIP shows that E2A-PBX1 and EBF1 can co-occupy the same region of the E2F2 gene and, therefore, may constitute part of an in vivo transcriptional regulatory complex that localizes to E2A-PBX1 genomic binding regions. 3.6. E2A-PBX1 ASSOCIATES WITH GENOMIC SITES BOUND BY THE B LYMPHOPOIETIC TRANSCRIPTION FACTORS E2A AND EBF1 64

Figure 3.10: Binding of wildtype E2A and EBF1 at E2A-PBX1 peaks in RCH-ACV cells.

Figure 3.10. (A) ChIP-qPCR using anti-E2A and anti-EBF1 antibodies was used to demonstrate occupancy of these factors at E2A-PBX1 binding sites. The BCL6 locus was used as a negative control since it was shown to not bind E2A or EBF1 in GM12878 cells. (B) Sequential ChIP with anti-FLAG and then anti-EBF1 was used to demonstrate simultaneous co-occupancy of E2A-PBX1 and EBF1 6 kb downstream of the E2F2 gene. 3.7. KNOCKDOWN OF E2A-PBX1 SILENCES TARGET GENE EXPRESSION 65

3.7 Knockdown of E2A-PBX1 silences target gene expression

We have previously considered E2A-PBX1 primarily as a transcriptional activator by way of its interaction with p300 and its in vitro ability to increase luciferase reporter activity as a Gal4 fusion protein. To test the direct, in vivo role of E2A- PBX1 as a transcriptional regulator, we used qRT-PCR to measure expression of some candidate E2A-PBX1 target genes after depletion of E2A-PBX1 by siRNA in RCH-ACV cells. Quantitative RT-PCR confirmed a decrease in transcription of the previously-characterized E2A-PBX1-responsinve genes CALD1 and EB1 upon E2A- PBX1 knockdown (Figure 3.11) [28, 57]. We also observe a decrease in expression of E2F2 and CCND3, two candidate target genes identified in this study, in response to E2A-PBX1 silencing. These findings demonstrate a direct role for E2A-PBX1 in inducing the expression of endogenous target genes. 3.7. KNOCKDOWN OF E2A-PBX1 SILENCES TARGET GENE EXPRESSION 66

Figure 3.11: Response of E2A-PBX1 target genes to E2A-PBX1 knockdown.

Figure 3.11. RCH-ACV cells were transfected with siRNA against E2A-PBX1 and qRT-PCR was used to measure the expression of candidate E2A-PBX1 target genes. GAPDH expression was used for normalization and siRNA against LaminA/C was used as a negative control. 67

Chapter 4

Discussion

4.1 Cooperative regulation of transcriptional enhancers by E2A-PBX1 and B lymphopoietic transcription factors

Both wildtype PBX1 and the E2A-PBX1 fusion protein bind DNA containing the PBX1 consensus element in vitro but unlike PBX1, E2A-PBX1 has been shown to strongly activate transcription of genes downstream of the PBX1 consensus element in reporter gene assays [100, 108, 175]. This led to the initial hypothesis that E2A-PBX1 contributes to leukemogenesis by maintaining constitutive expression of endogenous PBX1 target genes. The current results demonstrate that, instead of localizing to PBX1 target sequences, E2A-PBX1 binding sites often contain transcription factor binding sites corresponding to endogenous wildtype hematopoietic transcription factors, including the lymphopoietic transcription factors E2A and EBF1 (Figure 4.1). This observation may help to explain why the DNA-binding homeodomain of PBX1 was dispensable for its transforming activities in some model systems and supports the suggestion that E2A-PBX1 contributes to ALL development, at least partially, by aberrantly sustaining expression of critical E2A and EBF1 transcriptional targets 4.1. COOPERATIVE REGULATION OF TRANSCRIPTIONAL ENHANCERS BY E2A-PBX1 AND B LYMPHOPOIETIC TRANSCRIPTION FACTORS 68

[32, 123, 15]. E2A-PBX1 binding sites were shown to localize predominantly within intronic and intergenic regions of the genome, rather than at the promoters of its operationally- defined target genes. While the preference for intronic binding by E2A-PBX1 may suggest a role in regulating alternative transcription, it is likely that E2A-PBX1 is binding to transcriptional enhancers, which are non-coding gene-regulatory DNA sequences that often reside within intronic and intergenic regions of the genome (Re- viewed in [133] and [161]). Active enhancers are also characterized by enrichment of CBP/p300 and the functional histone modifications H3K4me1/2, and H3K27Ac [133, 161]. These salient characteristics of enhancers are strikingly similar to the genomic distribution and epigenetic profile of E2A-PBX1 binding sites and suggests a novel role for E2A-PBX1 in enhancer regulation. Interestingly, enhancers often contain a high density of transcription factor consensus elements and have been shown to function as ”hot-spots” for factor binding [98, 124]. Transcription factors often localize to these hot-spots in a combinatorial fashion, which is thought to fine-tune gene expression patterns. The melange of transcription factor consensus sequences discovered at E2A-PBX1 binding sites through motif scanning analysis further contributes to the idea that E2A-PBX1 may localize to and regulate transcriptional enhancers. 4.1. COOPERATIVE REGULATION OF TRANSCRIPTIONAL ENHANCERS BY E2A-PBX1 AND B LYMPHOPOIETIC TRANSCRIPTION FACTORS 69

Figure 4.1: Models of gene regulation by E2A-PBX1 in t(1;19)-associated leukemia cells

Figure 4.1. (A) E2A-PBX1 was initially predicted to contribute to leukemogenesis via constitutive upregulation of PBX1 target genes. (B) Our ChIP-seq and bioinformatics data prompt a revision of the canonical model of gene regulation by E2A-PBX1. The revised model suggests that E2A-PBX1 localizes to intronic and intergenic regions containing wildtype E2A and EBF1 consensus elements. CBP/p300 may be anchored to the DNA through stable interactions with EBF1, which allows for competition between the AD1 domain, present on both E2A and E2A-PBX1, and the KIX domain of CBP/p300. The outcome of this competition for CBP/p300 binding may determine whether the operational targets of E2A-PBX1 are activated or silenced. 4.2. LEUKEMOGENESIS VIA A HIJACKING OF NORMAL LYMPHOPOIETIC REGULATORY NETWORKS 70

The potential role of E2A-PBX1 in enhancer regulation is an interesting area deserving of further study. Genome-wide profiling of active enhancers can be accomplished by H3K27Ac ChIP-seq in t(1;19)-associated cells and standard bioinformatics techniques could be used to determine whether E2A-PBX1 is, indeed, present at active enhancers. Super-enhancers are a recently identified class of gene regulatory elements capable of driving high levels of target gene expression [107, 183]. These regions are characterized by abundant binding of the Mediator co-activator and the bromodomain-containing protein BRD4 [107]. Relatively little is known about the role of super-enhancers in B lymphoid malignancies aside from one study of their role in activating E2F and MYC target genes in diffuse large B cell lymphoma [33]. Mediator and BRD4 ChIP-seq experiments can be used to identify super-enhancers in t(1;19)-associated cells and help to determine whether E2A-PBX1 participates in the regulation of these novel gene regulatory regions.

4.2 Leukemogenesis via a hijacking of normal lymphopoietic regulatory networks

The presence of E2A and EBF1 recognition sequences at many E2A-PBX1 binding sites raises the intriguing possibility that E2A-PBX1 may be able ”plug in” to endogenous gene regulatory networks that are already active in immature B lymphocytes. It is well-established that E2A and EBF1 each have critical roles during early B lymphopoiesis in promoting diﬀerentiation, survival, and expansion of lymphocyte populations [70, 126]. Lin and colleagues recently used ChIP-seq to compare the genome-wide distribution of E2A in pre-pro-B and pro-B lymphocytes [103]. They observed that each of these developmental stages was characterized by distinct E2A 4.3. AN E2A-PBX1-CONTAINING TRANSCRIPTIONAL REGULATORY COMPLEX 71 binding and gene expression patterns. In leukemia cells, the recruitment of E2A- PBX1 to genomic binding sites that are normally occupied by E2A and EBF1 may disrupt the delicate gene regulatory networks controlled by these transcription factors. E2A-PBX1 might, for example, contribute to prolonged retention of p300 at these sites, which may contribute to leukemogenesis by inappropriately perpetuating the expression of E2A and EBF1 targets that would normally be shut down during normal lymphocyte development.

4.3 An E2A-PBX1-containing transcriptional regulatory complex

While the bioinformatics data presented here have shown that E2A-PBX1 can localize to regions associated with E2A, EBF1, ETS1, and RUNX1 DNA sequence motifs, it is currently unknown whether these proteins interact directly with each other or indirectly through some intermediary component. CBP/p300 is a large protein with myriad interaction surfaces and has been shown to interact with all of the above transcriptional regulators [30, 22, 91, 194, 188]. CBP/p300, therefore, represents a potential ”scaﬀold” that can coordinate the formation of an E2A-PBX1-containing transcriptional regulatory complex at sites of cooperativity with other transcription factors. Additional research must be performed in order to identify other members of this complex that may not be detectable using ChIP-seq and bioinformatics. A recent study used immunoprecipitation and mass-spectrometry to identify novel members of a multi-protein complex nucleated by the leukemogenic fusion transcription factor AML1-ETO [164]. A similar approach could be used to identify additional interaction partners for E2A-PBX1 in t(1;19)-associated leukemia cells. By char- acterizing the E2A-PBX1-containing transcriptional regulatory complex, it may be 4.4. TRANSCRIPTIONAL SILENCING BY ALLOSTERIC MODIFICATION 72 possible to reveal new mechanisms by which E2A-PBX1 regulates the expression of target genes that contribute to leukemogenesis and provide additional therapeutic targets in t(1;19)-associated leukemia. For instance, disruption of the E2A-PBX1- containing transcription factor complex mentioned above could result in silencing of E2A-PBX1 target genes, which could have therapeutic value. Future research is necessary to identify the interaction surfaces that are critical to the stability of the complex, which may then be used to inform the development of small molecule or peptides capable of destabilizing the complex.

4.4 Transcriptional silencing by allosteric modiﬁcation

Recent RNA-seq analysis has demonstrated that E2A-PBX1 knockdown contributes to both up- and downregulation of its operational target genes (Appendix D). This novel finding suggests that E2A-PBX1, previously characterized exclusively as a transcriptional activator, also contributes to transcriptional silencing. The mechanism by which this may occur remains to be explored. If, however, CBP/p300 functions as a scaffold and is involved in the nucleation of the E2A-PBX1-containing transcriptional regulatory complex, allosteric effects induced by E2A-PBX1 and other factors may alter the binding capabilities, histone acetyltransferase, and overall transactivating function of CBP/p300. Indeed, the KIX domain of p300, which interacts with AD1 of E2A/E2A-PBX1, is capable of ”communicating” allosteric information to other regions of the protein when bound by its ligand [24]. The implications of this allosteric regulation remain to be explored at a functional level. There is an enticing (and experimentally-tractable) possibility that the combinatorial binding of E2A-PBX, wildtype E2A, and EBF1 may contribute to allosteric modifications of 4.5. THE FUNCTIONAL CONTRIBUTION OF E2A-PBX1 TO LEUKEMOGENESIS 73

CBP/p300 that modify its capacity for transcriptional activation vs. silencing.

4.5 The functional contribution of E2A-PBX1 to leukemogenesis

The functional contribution of E2A-PBX1 to gene expression and cellular pathway activation/repression was tested by knocking down E2A-PBX1 expression in RCH- ACV cells using siRNA. Silencing of E2A-PBX1 contributed to both the up- and downregulation of its operationally-defined target genes, which was unexpected since E2A-PBX1 was characterized solely as a transcriptional activator. The genome-wide analysis of E2A-PBX1 binding showed that E2A-PBX1 binding occurs predominantly within intronic and intergenic regions, which contain wildtype E2A and EBF1 binding sites. This pattern of binding suggests that E2A-PBX1 may contribute to gene regulation in a context-specific manner that is influenced by the activity of nearby transcription factors, such as E2A and EBF1. Differential gene expression analysis was performed on a transcriptome-wide level (i.e. analysis was not limited to the operational target genes of E2A-PBX1) in order to measure the contribution of E2A-PBX1 to gene expression at a cellular level. KEGG pathway analysis showed that E2A-PBX1 knockdown resulted in decreased expression of genes involved in cell cycle progression while concomitantly increasing the expression of genes involved in hematopoietic and lymphopoietic differentiation. This result invites the intriguing hypothesis that E2A-PBX1, upon its expression in immature lymphocyte progenitors, may block B cell development at the highly proliferative large pre-B cell stage and contribute to leukemia development by stabilizing the expression of genes that promote cycle progression while simultaneously silencing the expression of genes required for lymphocyte maturation. 4.6. TARGETING E2A-PBX1-RESPONSIVE GENES IN LEUKEMIA74

Interestingly, the chimeric transcription factor AML1-ETO (the subject of Ap- pendices A and B) is causally associated with myeloid leukemia and has been shown to co-localize to regions of the genome that are regulated by the E-proteins HEB and E2A [164]. Like E2A-PBX1, AML1-ETO has also been characterized as both a transcriptional activator and repressor [120, 92, 104, 144]. Since AML1-ETO and E2A-PBX1 share these commonalities, it is possible that the disruption of E2A- dependent transcriptional networks is a general requirement in the development of both myeloid and lymphoid leukemia.

4.6 Targeting E2A-PBX1-responsive genes in leukemia

In the current absence of E2A-PBX1-speciﬁc therapies, t(1;19)-associated ALL is most commonly treated with conventional chemotherapy, which indiscriminately targets rapidly dividing cells. Over 50% of our operationally-deﬁned (i.e. the nearest gene whose TSS is found within 1 Mb of an E2A-PBX1 binding site) candidate regulatory targets of E2A-PBX1 were determined to be deregulated in cancer, according to the NCI’s Cancer Gene Index. A subset of these genes (e.x. CCND3, EGFR, RET ) are targetable by anti-neoplastic drugs, according to the Drug-Gene Interaction Database. In refractory or relapsed cases of t(1;19)-associated leukemia, drugs that target E2A-PBX1 regulable genes may be a useful addition to standard chemotherapy regimens. CCND3, a putative E2A-PBX1 target gene, is expressed in RCH-ACV cells at 33 times its expression level in the comparatively normal GM12878 cell line, according to RNA-seq data. The CCND3 complex inhibitor Palbociclib was recently shown in Phase III clinical trials to improve progression-free survival in patients with metastatic breast cancer when used as an adjuvant therapeutic (Reviewed 4.7. CAVEATS AND LIMITATIONS 75

in [146]). Future research should be performed to determine whether Palbociclib ad- ministration may beneﬁt certain t(1;19)-ALL patients when used in combination with standard chemotherapy regimens.

4.7 Caveats and limitations

One notable limitation of this study regards the restrictive criteria that were used to identify E2A-PBX1 binding peaks that are common to both RCH-ACV and 697 cell lines. While this filtering technique was necessary to overcome issues regarding the signal-to-noise ratio of our RCH-ACV ChIP-seq data set, and although it seems likely to have detected the strongest binding peaks, our approach is likely to have missed a considerable number of E2A-PBX1 binding sites, including some that may be important and others that are exclusive to either RCH-ACV or 697 cells. ChIP-seq data quality can be improved with increased sequencing depth and improved antibody specificity. This would enable the identification of additional E2A-PBX1 binding peaks and reveal more subtle patterns of gene regulation that arise as a consequence of E2A-PBX1 expression [34, 85]. As an example of the anticipated benefits of increased sequencing depth, the motif scanning algorithm used to analyze the sequence composition of E2A-PBX1 binding sites identified enrichment of PU.1, MYB, FOXO1, and PAX5, but their P values failed to meet the threshold of significance required in order to be considered reliable. By submitting a greater number of bioinformatically- identified E2A-PBX1 binding sites for motif scanning analysis, statistical significance values for some of the aforementioned factors may be improved, which may provide bioinformatic evidence of additional co-regulatory roles for E2A-PBX1. 4.7. CAVEATS AND LIMITATIONS 76

To improve the success of future E2A-PBX1 ChIP-seq experiments, it may be help- ful to use a protein-protein cross-linking reagent such as N-hydroxysuccinimide (NHS, Sigma-Aldrich catalogue no. 130672) in addition to formaldehyde. This reagent has been used successfully by others in a two-step fixation procedure to improve the efficiency and reproducibility of ChIP studies of NF-κB and STAT3 [130]. Since the evidence presented here suggests that E2A-PBX1 does not interact directly with DNA, a combination of NHS and formaldehyde can promote retention of the multi-protein complex that nucleates at E2A-PBX1 binding sites and improve the resolution of future ChIP-qPCR and ChIP-seq assays. For this study, E2A-PBX1 target genes were operationally-defined as the nearest gene whose TSS occurred within 1 Mb of an E2A-PBX1 binding peak. Since the evidence presented here suggests that E2A-PBX1 may function in the regulation of enhancers, which can operate at great distances from their targets, this operational- definition of E2A-PBX1 target genes may be too restrictive [151, 169]. The potential existence of large distances that separate E2A-PBX1 binding sites from regulatory targets also suggests that chromatin looping may be involved. A recent study has shown that CBP and p300 mediate the formation of chromatin loops via a direct interaction with the stem cell transcriptional regulator Nanog and the promoters of its target genes [49]. Advanced ChIP and bioinformatics techniques, including Hi-C, can be used to study long-distance or inter-chromosomal interactions and contribute to the identification of additional transcriptional targets of E2A-PBX1 in t(1;19)- associated leukemia cells. 4.8. CONCLUSION 77

4.8 Conclusion

The experimental ﬁndings presented in this thesis provide novel and unexpected information about the genomic binding pattern of E2A-PBX1 and transcriptional consequences in the context of t(1;19)-associated leukemia cells. These results informed revision of the canonical model of E2A-PBX1-mediated leukemogenesis so as to re- ﬂect the involvement of E2A-PBX1 in co-regulating the transcriptional targets of endogenous hematopoietic transcription factors, including E2A and EBF1. This key discovery provides the basis for future research into the biochemical nature of the interactions between E2A-PBX1 and B lymphopoietic transcriptional gene regulators. BIBLIOGRAPHY 78

Bibliography

[1] B. Adams, P. Dorﬂer, A. Aguzzi, Z. Kozmik, P. Urbanek, I. Maurer-Fogy, and M. Busslinger. Pax-5 encodes the transcription factor BSAP and is expressed in B lymphocytes, the developing CNS, and adult testis. Genes Dev., 6(9):1589– 1607, Sep 1992.

[2] E. Y. Ahn, M. Yan, O. A. Malakhova, M. C. Lo, A. Boyapati, H. B. Ommen, R. Hines, P. Hokland, and D. E. Zhang. Disruption of the NHR4 domain structure in AML1-ETO abrogates SON binding and promotes leukemogenesis. Proc. Natl. Acad. Sci. U.S.A., 105(44):17103–17108, Nov 2008.

[3] M. Y. Ahn, G. Huang, S. C. Bae, H. J. Wee, W. Y. Kim, and Y. Ito. Nega- tive regulation of granulocytic diﬀerentiation in the myeloid precursor cell line 32Dcl3 by ear-2, a mammalian homolog of Drosophila seven-up, and a chimeric leukemogenic gene, AML1/ETO. Proc. Natl. Acad. Sci. U.S.A., 95(4):1812– 1817, Feb 1998.

[4] K. Akashi, D. Traver, T. Miyamoto, and I. L. Weissman. A clonogenic common myeloid progenitor that gives rise to all myeloid lineages. Nature, 404(6774):193–197, Mar 2000. BIBLIOGRAPHY 79

[5] R. A. Alharbi, R. Pettengell, H. S. Pandha, and R. @Articlepmid10082570, Author=”Liu, L. Q. and Ilaria, R. and Kingsley, P. D. and Iwama, A. and van Etten, R. A. and Palis, J. and Zhang, D. E. ”, Title=”A novel ubiquitin- speciﬁc protease, UBP43, cloned from leukemia fusion protein AML1-ETO- expressing mice, functions in hematopoietic cell diﬀerentiation”, Journal=”Mol. Cell. Biol.”, Year=”1999”, Volume=”19”, Number=”4”, Pages=”3029–3038”, Month=”Apr” Morgan. The role of HOX genes in normal hematopoiesis and acute leukemia. Leukemia, 27(5):1000–1008, Apr 2013.

[6] American-Cancer-Society. Cancer facts & ﬁgures 2014. Online, Atlanta, Jan- uary 2014.

[7] O. Aparicio, J. V. Geisberg, and K. Struhl. Chromatin immunoprecipitation for determining the association of proteins with speciﬁc genomic sequences in vivo. Curr Protoc Cell Biol, Chapter 17:Unit 17.7, Sep 2004.

[8] B. Argiropoulos and R. K. Humphries. Hox genes in hematopoiesis and leukemogenesis. Oncogene, 26(47):6766–6776, Oct 2007.

[9] A. Aronheim, R. Shiran, A. Rosen, and M. D. Walker. The E2A gene product contains two separable and functionally distinct transcription activation domains. Proc. Natl. Acad. Sci. U.S.A., 90(17):8063–8067, Sep 1993.

[10] W. R. Atchley and W. M. Fitch. A natural classiﬁcation of the basic helix-loop- helix class of transcription factors. Proc. Natl. Acad. Sci. U.S.A., 94(10):5172– 5176, May 1997. BIBLIOGRAPHY 80

[11] G. Bain, E. C. Maandag, D. J. Izon, D. Amsen, A. M. Kruisbeek, B. C. Wein- traub, I. Krop, M. S. Schlissel, A. J. Feeney, and M. van Roon. E2A proteins are required for proper B cell development and initiation of immunoglobulin gene rearrangements. Cell, 79(5):885–892, Dec 1994.

[12] A. Barberis, K. Widenhorn, L. Vitelli, and M. Busslinger. A novel B-cell lineage- speciﬁc transcription factor present at early but not late stages of diﬀerentiation. Genes Dev., 4(5):849–859, May 1990.

[13] A. Barski, S. Cuddapah, K. Cui, T. Y. Roh, D. E. Schones, Z. Wang, G. Wei, I. Chepelev, and K. Zhao. High-resolution proﬁling of histone methylations in the human genome. Cell, 129(4):823–837, May 2007.

[14] R. Bayly, L. Chuen, R. A. Currie, B. D. Hyndman, R. Casselman, G. A. Blo- bel, and D. P. LeBrun. E2A-PBX1 interacts directly with the KIX domain of CBP/p300 in the induction of proliferation in primary hematopoietic cells. J. Biol. Chem., 279(53):55362–55371, Dec 2004.

[15] R. Bayly and D. P. LeBrun. Role for homodimerization in growth deregulation by E2a fusion proteins. Mol. Cell. Biol., 20(16):5789–5796, Aug 2000.

[16] R. Bayly, T. Murase, B. D. Hyndman, R. Savage, S. Nurmohamed, K. Munro, R. Casselman, S. P. Smith, and D. P. LeBrun. Critical role for a single leucine residue in leukemia induction by E2A-PBX1. Mol. Cell. Biol., 26(17):6442– 6452, Sep 2006. BIBLIOGRAPHY 81

[17] A. J. BECKER, E. A. McCULLOCH, and J. E. TILL. Cytological demonstration of the clonal nature of spleen colonies derived from transplanted mouse marrow cells. Nature, 197:452–454, Feb 1963.

[18] C. G. Begley, P. D. Aplan, S. M. Denning, B. F. Haynes, T. A. Waldmann, and I. R. Kirsch. The gene SCL is expressed during early hematopoiesis and encodes a diﬀerentiation-related DNA-binding motif. Proc. Natl. Acad. Sci. U.S.A., 86(24):10128–10132, Dec 1989.

[19] J. M. Bennett, D. Catovsky, M. T. Daniel, G. Flandrin, D. A. Galton, H. R. Gralnick, and C. Sultan. Proposed revised criteria for the classiﬁcation of acute myeloid leukemia. A report of the French-American-British Cooperative Group. Ann. Intern. Med., 103(4):620–625, Oct 1985.

[20] C. A. Berkes, D. A. Bergstrom, B. H. Penn, K. J. Seaver, P. S. Knoepﬂer, and S. J. Tapscott. Pbx marks genes for activation by MyoD indicating a role for a homeodomain protein in establishing myogenic potential. Mol. Cell, 14(4):465– 477, May 2004.

[21] S. Bhalla, C. Spaulding, R. L. Brumbaugh, D. E. Zagort, M. E. Massari, C. Murre, and B. L. Kee. diﬀerential roles for the E2A activation domains in B lymphocytes and macrophages. J. Immunol., 180(3):1694–1703, Feb 2008.

[22] C. Bradney, M. Hjelmeland, Y. Komatsu, M. Yoshida, T. P. Yao, and Y. Zhuang. Regulation of E2A activities by histone acetyltransferases in B lymphocyte development. J. Biol. Chem., 278(4):2370–2376, Jan 2003. BIBLIOGRAPHY 82

[23] C. M. Brown, S. R. Larsen, H. J. Iland, D. E. Joshua, and J. Gibson. Leukaemias into the 21st century: part 1: the acute leukaemias. Intern Med J, 42(11):1179– 1186, Nov 2012.

[24] S. Bruschweiler, P. Schanda, K. Kloiber, B. Brutscher, G. Kontaxis, R. Kon- rat, and M. Tollinger. Direct observation of the dynamic process underlying allosteric signal transmission. J. Am. Chem. Soc., 131(8):3063–3068, Mar 2009.

[25] S. A. Burel, N. Harakawa, L. Zhou, T. Pabst, D. G. Tenen, and D. E. Zhang. Di- chotomy of AML1-ETO functions: growth arrest versus block of diﬀerentiation. Mol. Cell. Biol., 21(16):5577–5590, Aug 2001.

[26] F. Calabi, R. Pannell, and G. Pavloska. Gene targeting reveals a crucial role for MTG8 in the gut. Mol. Cell. Biol., 21(16):5658–5666, Aug 2001.

[27] R. Cao and Y. Zhang. SUZ12 is required for both the histone methyltransferase activity and the silencing function of the EED-EZH2 complex. Mol. Cell, 15(1):57–67, Jul 2004.

[28] G. Casagrande, G. te Kronnie, and G. Basso. The eﬀects of siRNA-mediated inhibition of E2A-PBX1 on EB-1 and Wnt16b expression in the 697 pre-B leukemia cell line. Haematologica, 91(6):765–771, Jun 2006.

[29] L. H. Castilla, C. Wijmenga, Q. Wang, T. Stacy, N. A. Speck, M. Eckhaus, M. Marin-Padilla, F. S. Collins, A. Wynshaw-Boris, and P. P. Liu. Failure of embryonic hematopoiesis and lethal hemorrhages in mouse embryos heterozygous for a knocked-in leukemia gene CBFB-MYH11. Cell, 87(4):687–696, Nov 1996. BIBLIOGRAPHY 83

[30] H. M. Chan and N. B. La Thangue. p300/CBP proteins: HATs for transcriptional bridges and scaﬀolds. J. Cell. Sci., 114(Pt 13):2363–2373, Jul 2001.

[31] S. K. Chan, L. Jaﬀe, M. Capovilla, J. Botas, and R. S. Mann. The DNA binding speciﬁcity of Ultrabithorax is modulated by cooperative interactions with extradenticle, another homeoprotein. Cell, 78(4):603–615, Aug 1994.

[32] C. P. Chang, I. de Vivo, and M. L. Cleary. The Hox cooperativity motif of the chimeric oncoprotein E2a-Pbx1 is necessary and suﬃcient for oncogenesis. Mol. Cell. Biol., 17(1):81–88, Jan 1997.

[33] B. Chapuy, M. R. McKeown, C. Y. Lin, S. Monti, M. G. Roemer, J. Qi, P. B. Rahl, H. H. Sun, K. T. Yeda, J. G. Doench, E. Reichert, A. L. Kung, S. J. Rodig, R. A. Young, M. A. Shipp, and J. E. Bradner. Discovery and characterization of super-enhancer-associated dependencies in diﬀuse large B cell lymphoma. Cancer Cell, 24(6):777–790, Dec 2013.

[34] Y. Chen, N. Negre, Q. Li, J. O. Mieczkowska, M. Slattery, T. Liu, Y. Zhang, T. K. Kim, H. H. He, J. Zieba, Y. Ruan, P. J. Bickel, R. M. Myers, B. J. Wold, K. P. White, J. D. Lieb, and X. S. Liu. Systematic evaluation of factors inﬂuencing ChIP-seq ﬁdelity. Nat. Methods, 9(6):609–614, Jun 2012.

[35] T. M. Conlon and K. B. Meyer. Cloning and functional characterisation of avian transcription factor E2A. BMC Immunol., 5:11, Jun 2004.

[36] H. N. Crans and K. M. Sakamoto. Transcription factors and translocations in lymphoid and myeloid leukemia. Leukemia, 15(3):313–331, Mar 2001. BIBLIOGRAPHY 84

[37] M. P. Creyghton, A. W. Cheng, G. G. Welstead, T. Kooistra, B. W. Carey, E. J. Steine, J. Hanna, M. A. Lodato, G. M. Frampton, P. A. Sharp, L. A. Boyer, R. A. Young, and R. Jaenisch. Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proc. Natl. Acad. Sci. U.S.A., 107(50):21931–21936, Dec 2010.

[38] J Nathan Davis, Laura McGhee, and Shari Meyers. The eto( mtg8) gene family. Gene, 303:1–10, 2003.

[39] W. B. de Lau, J. Hurenkamp, P. Berendes, I. P. Touw, H. C. Clevers, and M. A. van Dijk. The gene encoding the granulocyte colony-stimulating factor receptor is a target for deregulation in pre-B ALL by the t(1;19)-speciﬁc oncoprotein E2A-Pbx1. Oncogene, 17(4):503–510, Jul 1998.

[40] S. Dias, H. Silva, A. Cumano, and P. Vieira. Interleukin-7 is necessary to maintain the B cell potential in common lymphoid progenitors. J. Exp. Med., 201(6):971–979, Mar 2005.

[41] J. F. DiMartino, L. Selleri, D. Traver, M. T. Firpo, J. Rhee, R. Warnke, S. O’Gorman, I. L. Weissman, and M. L. Cleary. The Hox cofactor and proto- oncogene Pbx1 is required for maintenance of deﬁnitive hematopoiesis in the fetal liver. Blood, 98(3):618–626, Aug 2001.

[42] H. Dohner, E. H. Estey, S. Amadori, F. R. Appelbaum, T. Buchner, A. K. Burnett, H. Dombret, P. Fenaux, D. Grimwade, R. A. Larson, F. Lo-Coco, T. Naoe, D. Niederwieser, G. J. Ossenkoppele, M. A. Sanz, J. Sierra, M. S. Tallman, B. Lowenberg, and C. D. Bloomﬁeld. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international BIBLIOGRAPHY 85

expert panel, on behalf of the European LeukemiaNet. Blood, 115(3):453–474, Jan 2010.

[43] S. Doulatov, F. Notta, E. Laurenti, and J. E. Dick. Hematopoiesis: a human perspective. Cell Stem Cell, 10(2):120–136, Feb 2012.

[44] Joseph B Duﬀy and J Peter Gergen. Sex, segments, and the central nervous system: common genetic mechanisms of cell fate determination. Advances in genetics, 31:1–28, 1994.

[45] R. Eckner, T. P. Yao, E. Oldread, and D. M. Livingston. Interaction and functional collaboration of p300/CBP and bHLH proteins in muscle and B-cell diﬀerentiation. Genes Dev., 10(19):2478–2490, Oct 1996.

[46] P Erickson, J Gao, KS Chang, T Look, E Whisenant, S Raimondi, R Lasher, J Trujillo, J Rowley, and H Drabkin. Identiﬁcation of breakpoints in t (8; 21) acute myelogenous leukemia and isolation of a fusion transcript, aml1/eto, with similarity to drosophila segmentation gene, runt. Blood, 80(7):1825–1831, 1992.

[47] P. F. Erickson, M. Robinson, G. Owens, and H. A. Drabkin. The ETO portion of acute myeloid leukemia t(8;21) fusion transcript encodes a highly evolutionarily conserved, putative transcription factor. Cancer Res., 54(7):1782–1786, Apr 1994.

[48] R. Eskeland, M. Leeb, G. R. Grimes, C. Kress, S. Boyle, D. Sproul, N. Gilbert, Y. Fan, A. I. Skoultchi, A. Wutz, and W. A. Bickmore. Ring1B compacts chromatin structure and represses gene expression independent of histone ubiq- uitination. Mol. Cell, 38(3):452–464, May 2010. BIBLIOGRAPHY 86

[49] F. Fang, Y. Xu, K. K. Chew, X. Chen, H. H. Ng, and P. Matsudaira. Coacti- vators p300 and CBP maintain the identity of mouse embryonic stem cells by mediating long-range chromatin structure. Stem Cells, Mar 2014.

[50] T. S. Fenske, G. Pengue, and T. A. Graubert. Dominant negative eﬀects of the AML1/ETO fusion oncoprotein. Cell Cycle, 4(1):33–36, Jan 2005.

[51] G. A. Follows, H. Tagoh, P. Lefevre, D. Hodge, G. J. Morgan, and C. Bonifer. Epigenetic consequences of AML1-ETO action at the human c-FMS locus. EMBO J., 22(11):2798–2809, Jun 2003.

[52] N. S. Fracchiolla, G. Colombo, P. Finelli, A. T. Maiolo, and A. Neri. EHT, a new member of the MTG8/ETO gene family, maps on 20q11 region and is deleted in acute myeloid leukemias. Blood, 92(9):3481–3484, Nov 1998.

[53] R. Frank, J. Zhang, H. Uchida, S. Meyers, S. W. Hiebert, and S. D. Nimer. The AML1/ETO fusion protein blocks transactivation of the GM-CSF promoter by AML1B. Oncogene, 11(12):2667–2674, Dec 1995.

[54] R. C. Frank, X. Sun, F. J. Berguido, A. Jakubowiak, and S. D. Nimer. The t(8;21) fusion protein, AML1/ETO, transforms NIH3T3 cells and activates AP- 1. Oncogene, 18(9):1701–1710, Mar 1999.

[55] S. Frietze, R. Wang, L. Yao, Y. G. Tak, Z. Ye, M. Gaddis, H. Witt, P. J. Farnham, and V. X. Jin. Cell type-speciﬁc binding patterns reveal that TCF7L2 can be tethered to the genome by association with GATA3. Genome Biol., 13(9):R52, 2012. BIBLIOGRAPHY 87

[56] X. Fu and M. P. Kamps. E2a-Pbx1 induces aberrant expression of tissue-speciﬁc and developmentally regulated genes when expressed in NIH 3T3 ﬁbroblasts. Mol. Cell. Biol., 17(3):1503–1512, Mar 1997.

[57] X. Fu, S. McGrath, M. Pasillas, S. Nakazawa, and M. P. Kamps. EB-1, a tyrosine kinase signal transduction gene, is transcriptionally activated in the t(1;19) subset of pre-B ALL, which express oncoprotein E2a-Pbx1. Oncogene, 18(35):4920–4929, Sep 1999.

[58] T. S. Furey. ChIP-seq and beyond: new and improved methodologies to detect and characterize protein-DNA interactions. Nat. Rev. Genet., 13(12):840–852, Dec 2012.

[59] T. Gamou, E. Kitamura, F. Hosoda, K. Shimizu, K. Shinohara, Y. Hayashi, T. Nagase, Y. Yokoyama, and M. Ohki. The partner gene of AML1 in t(16;21) myeloid malignancies is a novel member of the MTG8(ETO) family. Blood, 91(11):4028–4037, Jun 1998.

[60] M. Garcia-Ramirez, C. Rocchini, and J. Ausio. Modulation of chromatin folding by histone acetylation. J. Biol. Chem., 270(30):17923–17928, Jul 1995.

[61] A. Gardini, M. Cesaroni, L. Luzi, A. J. Okumura, J. R. Biggs, S. P. Minardi, E. Venturini, D. E. Zhang, P. G. Pelicci, and M. Alcalay. AML1/ETO oncoprotein is directed to AML1 binding regions and co-localizes with AML1 and HEB on its targets. PLoS Genet., 4(11):e1000275, Nov 2008.

[62] H. Geng, S. Brennan, T. A. Milne, W. Y. Chen, Y. Li, C. Hurtz, S. M. Kweon, L. Zickl, S. Shojaee, D. Neuberg, C. Huang, D. Biswas, Y. Xin, J. Racevskis, BIBLIOGRAPHY 88

R. P. Ketterling, S. M. Luger, H. Lazarus, M. S. Tallman, J. M. Rowe, M. R. Litzow, M. L. Guzman, C. D. Allis, R. G. Roeder, M. Muschen, E. Paietta, O. Elemento, and A. M. Melnick. Integrative epigenomic analysis identiﬁes biomarkers and therapeutic targets in adult B-acute lymphoblastic leukemia. Cancer Discov, 2(11):1004–1023, Nov 2012.

[63] J. K. Ghosh, W. J. Romanow, and C. Murre. Induction of a diverse T cell receptor gamma/delta repertoire by the helix-loop-helix proteins E2A and HEB in nonlymphoid cells. J. Exp. Med., 193(6):769–776, Mar 2001.

[64] J. Gibson, H. J. Iland, S. R. Larsen, C. M. Brown, and D. E. Joshua. Leukaemias into the 21st century. Part 2: the chronic leukaemias. Intern Med J, 43(5):484– 494, May 2013.

[65] K. Giese, C. Kingsley, J. R. Kirshner, and R. Grosschedl. Assembly and function of a TCR alpha enhancer complex is dependent on LEF-1-induced DNA bending and multiple protein-protein interactions. Genes Dev., 9(8):995–1008, Apr 1995.

[66] Y. Gong, L. Huo, P. Liu, N. Sneige, X. Sun, N. T. Ueno, A. Lucci, T. A. Buchholz, V. Valero, and M. Cristofanilli. Polycomb group protein EZH2 is frequently expressed in inﬂammatory breast cancer and is predictive of worse clinical outcome. Cancer, 117(24):5476–5484, Dec 2011.

[67] Stephen Greenbaum and Yuan Zhuang. Identiﬁcation of e2a target genes in b lymphocyte development by using a gene tagging-based chromatin immunoprecipitation system. Proceedings of the National Academy of Sciences, 99(23):15030–15035, 2002. BIBLIOGRAPHY 89

[68] M. Griffith, O. L. Griffith, A. C. Coffman, J. V. Weible, J. F. McMichael, N. C. Spies, J. Koval, I. Das, M. B. Callaway, J. M. Eldred, C. A. Miller, J. Subramanian, R. Govindan, R. D. Kumar, R. Bose, L. Ding, J. R. Walker, D. E. Larson, D. J. Dooling, S. M. Smith, T. J. Ley, E. R. Mardis, and R. K. Wilson. DGIdb: mining the druggable genome. Nat. Methods, 10(12):1209– 1210, Dec 2013.

[69] M. G. Guenther, S. S. Levine, L. A. Boyer, R. Jaenisch, and R. A. Young. A chromatin landmark and transcription initiation at most promoters in human cells. Cell, 130(1):77–88, Jul 2007.

[70] I. Gyory, S. Boller, R. Nechanitzky, E. Mandel, S. Pott, E. Liu, and R. Gross- chedl. Transcription factor Ebf1 regulates diﬀerentiation stage-speciﬁc signaling, proliferation, and survival of B cells. Genes Dev., 26(7):668–682, Apr 2012.

[71] J. Hagman, M. J. Gutch, H. Lin, and R. Grosschedl. EBF contains a novel zinc coordination motif and multiple dimerization and transcriptional activation domains. EMBO J., 14(12):2907–2916, Jun 1995.

[72] J. Hagman and K. Lukin. Transcription factors drive B cell development. Curr. Opin. Immunol., 18(2):127–134, Apr 2006.

[73] J. Hagman, A. Travis, and R. Grosschedl. A novel lineage-speciﬁc nuclear factor regulates mb-1 gene transcription at the early stages of B cell diﬀerentiation. EMBO J., 10(11):3409–3417, Nov 1991. BIBLIOGRAPHY 90

[74] R. Haw, H. Hermjakob, P. D’Eustachio, and L. Stein. Reactome pathway analysis to enrich biological discovery in proteomics data sets. Proteomics, 11(18):3598–3613, Sep 2011.

[75] G. He, D. Wu, A. Sun, Y. Xue, Z. Jin, H. Qiu, X. Tang, M. Miao, Z. Fu, X. Ma, X. Wang, Z. Chen, and C. Ruan. B-Lymphoid and myeloid lineages biphenotypic acute leukemia with t(8;21)(q22;q22). Int. J. Hematol., 87(2):132– 136, Mar 2008.

[76] S. Heinz, C. Benner, N. Spann, E. Bertolino, Y. C. Lin, P. Laslo, J. X. Cheng, C. Murre, H. Singh, and C. K. Glass. Simple combinations of lineage- determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell, 38(4):576–589, May 2010.

[77] J. Hikima, M. L. Lennard, M. R. Wilson, N. W. Miller, L. W. Clem, and G. W. Warr. Evolution of vertebrate E protein transcription factors: comparative analysis of the E protein gene family in Takifugu rubripes and humans. Physiol. Genomics, 21(2):144–151, Apr 2005.

[78] D. Hildebrand, J. Tiefenbach, T. Heinzel, M. Grez, and A. B. Maurer. Mul- tiple regions of ETO cooperate in transcriptional repression. J. Biol. Chem., 276(13):9889–9895, Mar 2001.

[79] J. W. Ho, E. Bishop, P. V. Karchenko, N. Negre, K. P. White, and P. J. Park. ChIP-chip versus ChIP-seq: lessons for experimental design and data analysis. BMC Genomics, 12:134, 2011. BIBLIOGRAPHY 91

[80] R. Holmes and J. C. Zuniga-Pﬂucker. The OP9-DL1 system: generation of T- lymphocytes from embryonic or hematopoietic stem cells in vitro. Cold Spring Harb Protoc, 2009(2):pdb.prot5156, Feb 2009.

[81] B. A. Hug, S. Y. Lee, E. L. Kinsler, J. Zhang, and M. A. Lazar. Cooperative function of Aml1-ETO corepressor recruitment domains in the expansion of primary bone marrow cells. Cancer Res., 62(10):2906–2912, May 2002.

[82] IHGSC. Finishing the euchromatic sequence of the human genome. Nature, 431(7011):931–945, Oct 2004.

[83] I. Jack, R. Seshadri, M. Garson, P. Michael, D. Callen, H. Zola, and A. Morley. RCH-ACV: a lymphoblastic leukemia cell line with chromosome translocation 1;19 and trisomy 8. Cancer Genet. Cytogenet., 19(3-4):261–269, Jan 1986.

[84] Q. Jin, L. R. Yu, L. Wang, Z. Zhang, L. H. Kasper, J. E. Lee, C. Wang, P. K. Brindle, S. Y. Dent, and K. Ge. Distinct roles of GCN5/PCAF-mediated H3K9ac and CBP/p300-mediated H3K18/27ac in nuclear receptor transactivation. EMBO J., 30(2):249–262, Jan 2011.

[85] Y. L. Jung, L. J. Luquette, J. W. Ho, F. Ferrari, M. Tolstorukov, A. Minoda, R. Issner, C. B. Epstein, G. H. Karpen, M. I. Kuroda, and P. J. Park. Impact of sequencing depth in ChIP-seq experiments. Nucleic Acids Res., Mar 2014.

[86] T. G. Kahn, Y. B. Schwartz, G. I. Dellino, and V. Pirrotta. Polycomb complexes and the propagation of the methylation mark at the Drosophila ubx gene. J. Biol. Chem., 281(39):29064–29075, Sep 2006. BIBLIOGRAPHY 92

[87] M. P. Kamps, C. Murre, X. H. Sun, and D. Baltimore. A new homeobox gene contributes the DNA binding domain of the t(1;19) translocation protein in pre-B ALL. Cell, 60(4):547–555, Feb 1990.

[88] B. L. Kee. E and ID proteins branch out. Nat. Rev. Immunol., 9(3):175–184, Mar 2009.

[89] K. Kita, S. Shirakawa, and N. Kamada. Cellular characteristics of acute myeloblastic leukemia associated with t(8;21)(q22;q22). The Japanese Coop- erative Group of Leukemia/Lymphoma. Leuk. Lymphoma, 13(3-4):229–234, Apr 1994.

[90] I. Kitabayashi, K. Ida, F. Morohoshi, A. Yokoyama, N. Mitsuhashi, K. Shimizu, N. Nomura, Y. Hayashi, and M. Ohki. The AML1-MTG8 leukemic fusion protein forms a complex with a novel member of the MTG8(ETO/CDR) family, MTGR1. Mol. Cell. Biol., 18(2):846–858, Feb 1998.

[91] I. Kitabayashi, A. Yokoyama, K. Shimizu, and M. Ohki. Interaction and functional cooperation of the leukemia-associated factors AML1 and p300 in myeloid cell diﬀerentiation. EMBO J., 17(11):2994–3004, Jun 1998.

[92] L. Klampfer, J. Zhang, A. O. Zelenetz, H. Uchida, and S. D. Nimer. The AML1/ETO fusion protein activates transcription of BCL-2. Proc. Natl. Acad. Sci. U.S.A., 93(24):14059–14064, Nov 1996.

[93] C. G. Kleer, Q. Cao, S. Varambally, R. Shen, I. Ota, S. A. Tomlins, D. Ghosh, R. G. Sewalt, A. P. Otte, D. F. Hayes, M. S. Sabel, D. Livant, S. J. Weiss, M. A. Rubin, and A. M. Chinnaiyan. EZH2 is a marker of aggressive breast BIBLIOGRAPHY 93

cancer and promotes neoplastic transformation of breast epithelial cells. Proc. Natl. Acad. Sci. U.S.A., 100(20):11606–11611, Sep 2003.

[94] M. Kondo, I. L. Weissman, and K. Akashi. Identiﬁcation of clonogenic common lymphoid progenitors in mouse bone marrow. Cell, 91(5):661–672, Nov 1997.

[95] I. Kozlov, K. Beason, C. Yu, and M. Hughson. CD79a expression in acute myeloid leukemia t(8;21) and the importance of cytogenetics in the diagnosis of leukemias with immunophenotypic ambiguity. Cancer Genet. Cytogenet., 163(1):62–67, Nov 2005.

[96] Z. Kozmik, S. Wang, P. Dorﬂer, B. Adams, and M. Busslinger. The promoter of the CD19 gene is a target for the B-cell-speciﬁc transcription factor BSAP. Mol. Cell. Biol., 12(6):2662–2672, Jun 1992.

[97] C. S. Ku, N. Naidoo, M. Wu, and R. Soong. Studying the epigenome using next generation sequencing. J. Med. Genet., 48(11):721–730, Nov 2011.

[98] E. Z. Kvon, G. Stampfel, J. O. Yanez-Cuna, B. J. Dickson, and A. Stark. HOT regions function as patterned developmental enhancers and have a distinct cis- regulatory signature. Genes Dev., 26(9):908–913, May 2012.

[99] X. F. Le, D. Claxton, S. Kornblau, Y. H. Fan, Z. M. Mu, and K. S. Chang. Characterization of the ETO and AML1-ETO proteins involved in 8;21 translocation in acute myelogenous leukemia. Eur. J. Haematol., 60(4):217–225, Apr 1998. BIBLIOGRAPHY 94

[100] D. P. LeBrun and M. L. Cleary. Fusion with E2A alters the transcriptional properties of the homeodomain protein PBX1 in t(1;19) leukemias. Oncogene, 9(6):1641–1647, Jun 1994.

[101] N. Lenny, S. Meyers, and S. W. Hiebert. Functional domains of the t(8;21) fusion protein, AML-1/ETO. Oncogene, 11(9):1761–1769, Nov 1995.

[102] H. Lin and R. Grosschedl. Failure of B-cell diﬀerentiation in mice lacking the transcription factor EBF. Nature, 376(6537):263–267, Jul 1995.

[103] Y. C. Lin, S. Jhunjhunwala, C. Benner, S. Heinz, E. Welinder, R. Mansson, M. Sigvardsson, J. Hagman, C. A. Espinoza, J. Dutkowski, T. Ideker, C. K. Glass, and C. Murre. A global network of transcription factors, involving E2A, EBF1 and Foxo1, that orchestrates B cell fate. Nat. Immunol., 11(7):635–643, Jul 2010.

[104] L. Q. Liu, R. Ilaria, P. D. Kingsley, A. Iwama, R. A. van Etten, J. Palis, and D. E. Zhang. A novel ubiquitin-speciﬁc protease, UBP43, cloned from leukemia fusion protein AML1-ETO-expressing mice, functions in hematopoietic cell differentiation. Mol. Cell. Biol., 19(4):3029–3038, Apr 1999.

[105] P. Liu, S. A. Tarle, A. Hajra, D. F. Claxton, P. Marlton, M. Freedman, M. J. Siciliano, and F. S. Collins. Fusion between transcription factor CBF beta/PEBP2 beta and a myosin heavy chain in acute myeloid leukemia. Sci- ence, 261(5124):1041–1044, Aug 1993.

[106] Y. Liu, W. Chen, J. Gaudet, M. D. Cheney, L. Roudaia, T. Cierpicki, R. C. Klet, K. Hartman, T. M. Laue, N. A. Speck, and J. H. Bushweller. Structural basis BIBLIOGRAPHY 95

for recognition of SMRT/N-CoR by the MYND domain and its contribution to AML1/ETO’s activity. Cancer Cell, 11(6):483–497, Jun 2007.

[107] J. Loven, H. A. Hoke, C. Y. Lin, A. Lau, D. A. Orlando, C. R. Vakoc, J. E. Bradner, T. I. Lee, and R. A. Young. Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell, 153(2):320–334, Apr 2013.

[108] Q. Lu, P. S. Knoepﬂer, J. Scheele, D. D. Wright, and M. P. Kamps. Both Pbx1 and E2A-Pbx1 bind the DNA motif ATCAATCAA cooperatively with the products of multiple murine Hox genes, some of which are themselves oncogenes. Mol. Cell. Biol., 15(7):3786–3795, Jul 1995.

[109] B. Lutterbach, J. J. Westendorf, B. Linggi, S. Isaac, E. Seto, and S. W. Hiebert. A mechanism of repression by acute myeloid leukemia-1, the target of multiple chromosomal translocations in acute leukemia. J. Biol. Chem., 275(1):651–656, Jan 2000.

[110] B. Lutterbach, J. J. Westendorf, B. Linggi, A. Patten, M. Moniwa, J. R. Davie, K. D. Huynh, V. J. Bardwell, R. M. Lavinsky, M. G. Rosenfeld, C. Glass, E. Seto, and S. W. Hiebert. ETO, a target of t(8;21) in acute leukemia, interacts with the N-CoR and mSin3 corepressors. Mol. Cell. Biol., 18(12):7176–7184, Dec 1998.

[111] L. Magnani, E. B. Ballantyne, X. Zhang, and M. Lupien. PBX1 genomic pioneer function drives ER signaling underlying progression in breast cancer. PLoS Genet., 7(11):e1002368, Nov 2011. BIBLIOGRAPHY 96

[112] E. Matutes, R. Morilla, N. Farahat, F. Carbonell, J. Swansbury, M. Dyer, and D. Catovsky. Deﬁnition of acute biphenotypic leukemia. Haematologica, 82(1):64–66, 1997.

[113] Alexander A. Maximow. The lymphocyte as a stem cell, common to diﬀer- ent blood elements in embryonic development and during the post-fetal life of mammals. Folia Hae, pages 125–134, 1909.

[114] E. A. MCCULLOCH and J. E. TILL. The radiation sensitivity of normal mouse bone marrow cells, determined by quantitative marrow transplantation into irradiated mice. Radiat. Res., 13:115–125, Jul 1960.

[115] E. A. MCCULLOCH and J. E. TILL. PROLIFERATION OF HEMOPOI- ETIC COLONY-FORMING CELLS TRANSPLANTED INTO IRRADIATED MICE. Radiat. Res., 22:383–397, Jun 1964.

[116] L. McGhee, J. Bryan, L. Elliott, H. L. Grimes, A. Kazanjian, J. N. Davis, and S. Meyers. Gﬁ-1 attaches to the nuclear matrix, associates with ETO (MTG8) and histone deacetylase proteins, and represses transcription using a TSA-sensitive mechanism. J. Cell. Biochem., 89(5):1005–1018, Aug 2003.

[117] C. Y. McLean, D. Bristor, M. Hiller, S. L. Clarke, B. T. Schaar, C. B. Lowe, A. M. Wenger, and G. Bejerano. GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol., 28(5):495–501, May 2010.

[118] A. M. Melnick, J. J. Westendorf, A. Polinger, G. W. Carlile, S. Arai, H. J. Ball, B. Lutterbach, S. W. Hiebert, and J. D. Licht. The ETO protein disrupted in BIBLIOGRAPHY 97

t(8;21)-associated acute myeloid leukemia is a corepressor for the promyelocytic leukemia zinc ﬁnger protein. Mol. Cell. Biol., 20(6):2075–2086, Mar 2000.

[119] S. Meyers, J. R. Downing, and S. W. Hiebert. Identiﬁcation of AML-1 and the (8;21) translocation protein (AML-1/ETO) as sequence-speciﬁc DNA-binding proteins: the runt homology domain is required for DNA binding and protein- protein interactions. Mol. Cell. Biol., 13(10):6336–6345, Oct 1993.

[120] S. Meyers, N. Lenny, and S. W. Hiebert. The t(8;21) fusion protein inter- feres with AML-1B-dependent transcriptional activation. Mol. Cell. Biol., 15(4):1974–1982, Apr 1995.

[121] T. Miyamoto, K. Nagafuji, K. Akashi, M. Harada, T. Kyo, T. Akashi, K. Tak- enaka, S. Mizuno, H. Gondo, T. Okamura, H. Dohy, and Y. Niho. Persistence of multipotent progenitors expressing AML1/ETO transcripts in long-term remission patients with t(8;21) acute myelogenous leukemia. Blood, 87(11):4789– 4796, Jun 1996.

[122] H. Miyoshi, K. Shimizu, T. Kozu, N. Maseki, Y. Kaneko, and M. Ohki. t(8;21) breakpoints on chromosome 21 in acute myeloid leukemia are clustered within a limited region of a single gene, AML1. Proc. Natl. Acad. Sci. U.S.A., 88(23):10431–10434, Dec 1991.

[123] K. Monica, D. P. LeBrun, D. A. Dedera, R. Brown, and M. L. Cleary. Trans- formation properties of the E2a-Pbx1 chimeric oncoprotein: fusion with E2a is essential, but the Pbx1 homeodomain is dispensable. Mol. Cell. Biol., 14(12):8304–8314, Dec 1994. BIBLIOGRAPHY 98

[124] C. Moorman, L. V. Sun, J. Wang, E. de Wit, W. Talhout, L. D. Ward, F. Greil, X. J. Lu, K. P. White, H. J. Bussemaker, and B. van Steensel. Hotspots of transcription factor colocalization in the genome of Drosophila melanogaster. Proc. Natl. Acad. Sci. U.S.A., 103(32):12027–12032, Aug 2006.

[125] S. J. Morrison and I. L. Weissman. The long-term repopulating subset of hematopoietic stem cells is deterministic and isolatable by phenotype. Im- munity, 1(8):661–673, Nov 1994.

[126] C. Murre. Developmental trajectories in early hematopoiesis. Genes Dev., 23(20):2366–2370, Oct 2009.

[127] C. Murre, P. S. McCaw, and D. Baltimore. A new DNA binding and dimerization motif in immunoglobulin enhancer binding, daughterless, MyoD, and myc proteins. Cell, 56(5):777–783, Mar 1989.

[128] J. D. Nelson, O. Denisenko, and K. Bomsztyk. Protocol for the fast chromatin immunoprecipitation (ChIP) method. Nat Protoc, 1(1):179–185, 2006.

[129] J. Nourse, J. D. Mellentin, N. Galili, J. Wilkinson, E. Stanbridge, S. D. Smith, and M. L. Cleary. Chromosomal translocation t(1;19) results in synthesis of a homeobox fusion mRNA that codes for a potential chimeric transcription factor. Cell, 60(4):535–545, Feb 1990.

[130] D. E. Nowak, B. Tian, and A. R. Brasier. Two-step cross-linking method for identiﬁcation of NF-kappaB gene network by chromatin immunoprecipitation. BioTechniques, 39(5):715–725, Nov 2005. BIBLIOGRAPHY 99

[131] T. Okuda, Z. Cai, S. Yang, N. Lenny, C. J. Lyu, J. M. van Deursen, H. Harada, and J. R. Downing. Expression of a knocked-in AML1-ETO leukemia gene inhibits the establishment of normal deﬁnitive hematopoiesis and directly gen- erates dysplastic hematopoietic progenitors. Blood, 91(9):3134–3143, May 1998.

[132] T. Okuda, J. van Deursen, S. W. Hiebert, G. Grosveld, and J. R. Downing. AML1, the target of multiple chromosomal translocations in human leukemia, is essential for normal fetal liver hematopoiesis. Cell, 84(2):321–330, Jan 1996.

[133] C. T. Ong and V. G. Corces. Enhancer function: new insights into the regulation of tissue-speciﬁc gene expression. Nat. Rev. Genet., 12(4):283–293, Apr 2011.

[134] S. H. Orkin and L. I. Zon. Hematopoiesis: an evolving paradigm for stem cell biology. Cell, 132(4):631–644, Feb 2008.

[135] M. Osawa, K. Nakamura, N. Nishi, N. Takahasi, Y. Tokuomoto, H. Inoue, and H. Nakauchi. In vivo self-renewal of c-Kit+ Sca-1+ Lin(low/-) hemopoietic stem cells. J. Immunol., 156(9):3207–3214, May 1996.

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

[137] S. Park, W. Chen, T. Cierpicki, M. Tonelli, X. Cai, N. A. Speck, and J. H. Bushweller. Structure of the AML1-ETO eTAFH domain-HEB peptide complex and its contribution to AML1-ETO activity. Blood, 113(15):3558–3567, Apr 2009.

[138] Flo Pauli and Richard Myers. Myers lab chip-seq protocol. Online, April 2010. BIBLIOGRAPHY 100

[139] M. Peifer and E. Wieschaus. Mutations in the Drosophila gene extradenticle affect the way speciﬁc homeo domain proteins regulate segmental identity. Genes Dev., 4(7):1209–1223, Jul 1990.

[140] Y. Peng and N. Jahroudi. The NFY transcription factor functions as a repressor and activator of the von Willebrand factor promoter. Blood, 99(7):2408–2417, Apr 2002.

[141] J. M. Pongubala and M. L. Atchison. Functional characterization of the developmentally controlled immunoglobulin kappa 3’ enhancer: regulation by Id, a repressor of helix-loop-helix transcription factors. Mol. Cell. Biol., 11(2):1040– 1047, Feb 1991.

[142] C. Rauskolb, M. Peifer, and E. Wieschaus. extradenticle, a regulator of homeotic gene activity, is a homolog of the homeobox-containing human proto-oncogene pbx1. Cell, 74(6):1101–1112, Sep 1993.

[143] K. L. Rhoades, C. J. Hetherington, N. Harakawa, D. A. Yergeau, L. Zhou, L. Q. Liu, M. T. Little, D. G. Tenen, and D. E. Zhang. Analysis of the role of AML1-ETO in leukemogenesis, using an inducible transgenic mouse model. Blood, 96(6):2108–2115, Sep 2000.

[144] K. L. Rhoades, C. J. Hetherington, J. D. Rowley, S. W. Hiebert, G. Nucifora, D. G. Tenen, and D. E. Zhang. Synergistic up-regulation of the myeloid-speciﬁc promoter for the macrophage colony-stimulating factor receptor by AML1 and the t(8;21) fusion protein may contribute to leukemogenesis. Proc. Natl. Acad. Sci. U.S.A., 93(21):11895–11900, Oct 1996. BIBLIOGRAPHY 101

[145] J. J. Riethoven. Regulatory regions in DNA: promoters, enhancers, silencers, and insulators. Methods Mol. Biol., 674:33–42, 2010.

[146] A. Rocca, A. Farolﬁ, S. Bravaccini, A. Schirone, and D. Amadori. Palbociclib (PD 0332991) : targeting the cell cycle machinery in breast cancer. Expert Opin Pharmacother, 15(3):407–420, Feb 2014.

[147] J. D. Rowley. Identiﬁcaton of a translocation with quinacrine ﬂuorescence in a patient with acute leukemia. Ann. Genet., 16(2):109–112, Jun 1973.

[148] J. D. Rowley. Nonrandom chromosomal abnormalities in hematologic disorders of man. Proc. Natl. Acad. Sci. U.S.A., 72(1):152–156, Jan 1975.

[149] N. J. Sakabe, I. Aneas, T. Shen, L. Shokri, S. Y. Park, M. L. Bulyk, S. M. Evans, and M. A. Nobrega. Dual transcriptional activator and repressor roles of TBX20 regulate adult cardiac structure and function. Hum. Mol. Genet., 21(10):2194–2204, May 2012.

[150] Shuji Sakamoto, Kousho Wakae, Yuki Anzai, Kiyohito Murai, Nobuyuki Tamaki, Masaki Miyazaki, Kazuko Miyazaki, William J Romanow, Tomokatsu Ikawa, Daisuke Kitamura, et al. E2a and cbp/p300 act in synergy to promote chromatin accessibility of the immunoglobulin κ locus. The Journal of Immunology, 188(11):5547–5560, 2012.

[151] A. Sanyal, B. R. Lajoie, G. Jain, and J. Dekker. The long-range interaction landscape of gene promoters. Nature, 489(7414):109–113, Sep 2012.

[152] K. Sasaki, H. Yagi, R. T. Bronson, K. Tominaga, T. Matsunashi, K. Deguchi, Y. Tani, T. Kishimoto, and T. Komori. Absence of fetal liver hematopoiesis BIBLIOGRAPHY 102

in mice deﬁcient in transcriptional coactivator core binding factor beta. Proc. Natl. Acad. Sci. U.S.A., 93(22):12359–12363, Oct 1996.

[153] J. Seita and I. L. Weissman. Hematopoietic stem cell: self-renewal versus differentiation. Wiley Interdiscip Rev Syst Biol Med, 2(6):640–653, 2010.

[154] H. Shimada, H. Ichikawa, S. Nakamura, R. Katsu, M. Iwasa, I. Kitabayashi, and M. Ohki. Analysis of genes under the downstream control of the t(8;21) fusion protein AML1-MTG8: overexpression of the TIS11b (ERF-1, cMG1) gene induces myeloid cell proliferation in response to G-CSF. Blood, 96(2):655– 663, Jul 2000.

[155] M. Sigvardsson, D. R. Clark, D. Fitzsimmons, M. Doyle, P. Akerblad, T. Bres- lin, S. Bilke, R. Li, C. Yeamans, G. Zhang, and J. Hagman. Early B-cell factor, E2A, and Pax-5 cooperate to activate the early B cell-speciﬁc mb-1 promoter. Mol. Cell. Biol., 22(24):8539–8551, Dec 2002.

[156] J. A. Simon and R. E. Kingston. Occupying chromatin: Polycomb mechanisms for getting to genomic targets, stopping transcriptional traﬃc, and staying put. Mol. Cell, 49(5):808–824, Mar 2013.

[157] M. E. Smoot, K. Ono, J. Ruscheinski, P. L. Wang, and T. Ideker. Cytoscape 2.8: new features for data integration and network visualization. Bioinformatics, 27(3):431–432, Feb 2011.

[158] G. J. Spangrude, S. Heimfeld, and I. L. Weissman. Puriﬁcation and characterization of mouse hematopoietic stem cells. Science, 241(4861):58–62, Jul 1988. BIBLIOGRAPHY 103

[159] G. J. Spangrude and R. Scollay. A simpliﬁed method for enrichment of mouse hematopoietic stem cells. Exp. Hematol., 18(8):920–926, Sep 1990.

[160] N. A. Speck and D. G. Gilliland. Core-binding factors in haematopoiesis and leukaemia. Nat. Rev. Cancer, 2(7):502–513, Jul 2002.

[161] F. Spitz and E. E. Furlong. Transcription factors: from enhancer binding to developmental control. Nat. Rev. Genet., 13(9):613–626, Sep 2012.

[162] D. E. Sterner and S. L. Berger. Acetylation of histones and transcription-related factors. Microbiol. Mol. Biol. Rev., 64(2):435–459, Jun 2000.

[163] B. D. Strahl and C. D. Allis. The language of covalent histone modiﬁcations. Nature, 403(6765):41–45, Jan 2000.

[164] Xiao-Jian Sun, Zhanxin Wang, Lan Wang, Yanwen Jiang, Nils Kost, T David Soong, Wei-Yi Chen, Zhanyun Tang, Tomoyoshi Nakadai, Olivier Elemento, et al. A stable transcription factor complex nucleated by oligomeric aml1-eto controls leukaemogenesis. Nature, 2013.

[165] M. N. SWIFT, S. T. TAKETA, and V. P. BOND. Regionally fractionated x-irradiation equivalent in dose to tatal-body exposure. Radiat. Res., 1(3):241– 252, Jun 1954.

[166] Atsushi Takahashi, Masanobu Satake, Y Yamaguchi-Iwai, Suk-Chul Bae, Jie Lu, Mitsuo Maruyama, Yu Wen Zhang, Hiroya Oka, Naoko Arai, and Ken-ichi Arai. Positive and negative regulation of granulocyte-macrophage colony-stimulating factor promoter activity by aml1-related transcription factor, pebp2. Blood, 86(2):607–616, 1995. BIBLIOGRAPHY 104

[167] S. T. TAKETA, M. N. SWIFT, and V. P. BOND. Modiﬁcation of radiation injury in rats through gastrointestinal tract shielding and bone marrow therapy. Am. J. Physiol., 196(5):987–992, May 1959.

[168] The ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human genome. Nature, 489(7414):57–74, Sep 2012.

[169] R. E. Thurman, E. Rynes, R. Humbert, J. Vierstra, M. T. Maurano, E. Hau- gen, N. C. Sheffield, A. B. Stergachis, H. Wang, B. Vernot, K. Garg, S. John, R. Sandstrom, D. Bates, L. Boatman, T. K. Canfield, M. Diegel, D. Dunn, A. K. Ebersol, T. Frum, E. Giste, A. K. Johnson, E. M. Johnson, T. Kutyavin, B. Lajoie, B. K. Lee, K. Lee, D. London, D. Lotakis, S. Neph, F. Neri, E. D. Nguyen, H. Qu, A. P. Reynolds, V. Roach, A. Safi, M. E. Sanchez, A. Sanyal, A. Shafer, J. M. Simon, L. Song, S. Vong, M. Weaver, Y. Yan, Z. Zhang, Z. Zhang, B. Lenhard, M. Tewari, M. O. Dorschner, R. S. Hansen, P. A. Navas, G. Stamatoyannopoulos, V. R. Iyer, J. D. Lieb, S. R. Sunyaev, J. M. Akey, P. J. Sabo, R. Kaul, T. S. Furey, J. Dekker, G. E. Crawford, and J. A. Stamatoy- annopoulos. The accessible chromatin landscape of the human genome. Nature, 489(7414):75–82, Sep 2012.

[170] E. Tiacci, S. Pileri, A. Orleth, R. Pacini, A. Tabarrini, F. Frenguelli, A. Liso, D. Diverio, F. Lo-Coco, and B. Falini. PAX5 expression in acute leukemias: higher B-lineage speciﬁcity than CD79a and selective association with t(8;21)- acute myelogenous leukemia. Cancer Res., 64(20):7399–7404, Oct 2004.

[171] J. E. TILL, E. A. MCCULLOCH, and L. SIMINOVITCH. A STOCHASTIC MODEL OF STEM CELL PROLIFERATION, BASED ON THE GROWTH BIBLIOGRAPHY 105

OF SPLEEN COLONY-FORMING CELLS. Proc. Natl. Acad. Sci. U.S.A., 51:29–36, Jan 1964.

[172] C. Trapnell, A. Roberts, L. Goff, G. Pertea, D. Kim, D. R. Kelley, H. Pimentel, S. L. Salzberg, J. L. Rinn, and L. Pachter. Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc, 7(3):562–578, Mar 2012.

[173] Hideo Uchida, Jin Zhang, and Stephen D Nimer. Aml1a and aml1b can transac- tivate the human il-3 promoter. The Journal of Immunology, 158(5):2251–2258, 1997.

[174] P. Urbanek, Z. Q. Wang, I. Fetka, E. F. Wagner, and M. Busslinger. Com- plete block of early B cell diﬀerentiation and altered patterning of the posterior midbrain in mice lacking Pax5/BSAP. Cell, 79(5):901–912, Dec 1994.

[175] M. A. Van Dijk, P. M. Voorhoeve, and C. Murre. Pbx1 is converted into a transcriptional activator upon acquiring the N-terminal region of E2A in pre-B- cell acute lymphoblastoid leukemia. Proc. Natl. Acad. Sci. U.S.A., 90(13):6061– 6065, Jul 1993.

[176] A. Visel, M. J. Blow, Z. Li, T. Zhang, J. A. Akiyama, A. Holt, I. Plajzer-Frick, M. Shoukry, C. Wright, F. Chen, V. Afzal, B. Ren, E. M. Rubin, and L. A. Pennacchio. ChIP-seq accurately predicts tissue-speciﬁc activity of enhancers. Nature, 457(7231):854–858, Feb 2009.

[177] T. C. Voss and G. L. Hager. Dynamic regulation of transcriptional states by chromatin and transcription factors. Nat. Rev. Genet., 15(2):69–81, Feb 2014. BIBLIOGRAPHY 106

[178] J. Wang, T. Hoshino, R. L. Redner, S. Kajigaya, and J. M. Liu. ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human N-CoR/mSin3/HDAC1 complex. Proc. Natl. Acad. Sci. U.S.A., 95(18):10860–10865, Sep 1998.

[179] J. Wang, J. Zhuang, S. Iyer, X. Lin, T. W. Whitﬁeld, M. C. Greven, B. G. Pierce, X. Dong, A. Kundaje, Y. Cheng, O. J. Rando, E. Birney, R. M. Myers, W. S. Noble, M. Snyder, and Z. Weng. Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Res., 22(9):1798–1812, Sep 2012.

[180] L. Wang, A. Gural, X. J. Sun, X. Zhao, F. Perna, G. Huang, M. A. Hatlen, L. Vu, F. Liu, H. Xu, T. Asai, H. Xu, T. Deblasio, S. Menendez, F. Voza, Y. Jiang, P. A. Cole, J. Zhang, A. Melnick, R. G. Roeder, and S. D. Nimer. The leukemogenicity of AML1-ETO is dependent on site-speciﬁc lysine acetylation. Science, 333(6043):765–769, Aug 2011.

[181] I. L. Weissman and J. A. Shizuru. The origins of the identiﬁcation and isolation of hematopoietic stem cells, and their capability to induce donor-speciﬁc transplantation tolerance and treat autoimmune diseases. Blood, 112(9):3543–3553, Nov 2008.

[182] J. J. Westendorf, C. M. Yamamoto, N. Lenny, J. R. Downing, M. E. Selsted, and S. W. Hiebert. The t(8;21) fusion product, AML-1-ETO, associates with C/EBP-alpha, inhibits C/EBP-alpha-dependent transcription, and blocks granulocytic diﬀerentiation. Mol. Cell. Biol., 18(1):322–333, Jan 1998. BIBLIOGRAPHY 107

[183] W. A. Whyte, D. A. Orlando, D. Hnisz, B. J. Abraham, C. Y. Lin, M. H. Kagey, P. B. Rahl, T. I. Lee, and R. A. Young. Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell, 153(2):307– 319, Apr 2013.

[184] J. L. Wiemels, A. M. Ford, E. R. Van Wering, A. Postma, and M. Greaves. Protracted and variable latency of acute lymphoblastic leukemia after TEL- AML1 gene fusion in utero. Blood, 94(3):1057–1062, Aug 1999.

[185] A. P. Wolﬀe and J. J. Hayes. Chromatin disruption and modiﬁcation. Nucleic Acids Res., 27(3):711–720, Feb 1999.

[186] J. Xu, Y. Kawai, and I. J. Arinze. Dual role of C/EBP as an activator and repressor of Gi2 gene transcription. Genes Cells, 18(12):1082–1094, Dec 2013.

[187] M. Yan, E. Kanbe, L. F. Peterson, A. Boyapati, Y. Miao, Y. Wang, I. M. Chen, Z. Chen, J. D. Rowley, C. L. Willman, and D. E. Zhang. A previously unidenti- ﬁed alternatively spliced isoform of t(8;21) transcript promotes leukemogenesis. Nat. Med., 12(8):945–949, Aug 2006.

[188] C. Yang, L. H. Shapiro, M. Rivera, A. Kumar, and P. K. Brindle. A role for CREB binding protein and p300 transcriptional coactivators in Ets-1 transactivation functions. Mol. Cell. Biol., 18(4):2218–2229, Apr 1998.

[189] D. A. Yergeau, C. J. Hetherington, Q. Wang, P. Zhang, A. H. Sharpe, M. Binder, M. Marin-Padilla, D. G. Tenen, N. A. Speck, and D. E. Zhang. Embryonic lethality and impairment of haematopoiesis in mice heterozygous for an AML1-ETO fusion gene. Nat. Genet., 15(3):303–306, Mar 1997. BIBLIOGRAPHY 108

[190] M. D. Young, T. A. Willson, M. J. Wakeﬁeld, E. Trounson, D. J. Hilton, M. E. Blewitt, A. Oshlack, and I. J. Majewski. ChIP-seq analysis reveals distinct H3K27me3 proﬁles that correlate with transcriptional activity. Nucleic Acids Res., 39(17):7415–7427, Sep 2011.

[191] D. E. Zhang, C. J. Hetherington, S. Meyers, K. L. Rhoades, C. J. Larson, H. M. Chen, S. W. Hiebert, and D. G. Tenen. CCAAT enhancer-binding protein (C/EBP) and AML1 (CBF alpha2) synergistically activate the macrophage colony-stimulating factor receptor promoter. Mol. Cell. Biol., 16(3):1231–1240, Mar 1996.

[192] J. Zhang, B. A. Hug, E. Y. Huang, C. W. Chen, V. Gelmetti, M. Maccarana, S. Minucci, P. G. Pelicci, and M. A. Lazar. Oligomerization of ETO is obligatory for corepressor interaction. Mol. Cell. Biol., 21(1):156–163, Jan 2001.

[193] J. Zhang, M. Kalkum, S. Yamamura, B. T. Chait, and R. G. Roeder. E protein silencing by the leukemogenic AML1-ETO fusion protein. Science, 305(5688):1286–1289, Aug 2004.

[194] F. Zhao, R. McCarrick-Walmsley, P. Akerblad, M. Sigvardsson, and T. Kadesch. Inhibition of p300/CBP by early B-cell factor. Mol. Cell. Biol., 23(11):3837– 3846, Jun 2003.

[195] J. Zhu, M. Adli, J. Y. Zou, G. Verstappen, M. Coyne, X. Zhang, T. Durham, M. Miri, V. Deshpande, P. L. De Jager, D. A. Bennett, J. A. Houmard, D. M. Muoio, T. T. Onder, R. Camahort, C. A. Cowan, A. Meissner, C. B. Epstein, N. Shoresh, and B. E. Bernstein. Genome-wide chromatin state transitions BIBLIOGRAPHY 109

associated with developmental and environmental cues. Cell, 152(3):642–654, Jan 2013. 110

Appendix A

AML1-ETO Introduction

Work during the earlier years of my Ph.D. studies focused on the leukemogenic transcription factor AML1-ETO. Since my work on AML1-ETO is not directly relevant to my studies on E2A-PBX1, it has been included as an ”Appendix” in this thesis. Below, I have included the results of my study and the related introductory material associated with my work on this transcription factor.

A.1 Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is the most common leukemia aﬀecting adults and was diagnosed in over 6000 patients in the US in 2013 [6]. Although AML can present at any age, the disease is most often diagnosed in patients 50 years and older. The disease is characterized by the rapid proliferation of immature myeloid cells that accumulate in the bone marrow and impede normal blood production. As a result, patients can present with anemia, excessive bruising and bleeding, and an increased susceptibility to infection. If left untreated, AML patients can die within weeks to months after diagnosis due to symptoms of bone marrow failure. Aspirates of the bone marrow are used in the diagnosis of AML since bone marrow is the primary site for A.2. THE T(8;21) CHROMOSOMAL TRANSLOCATION 111

the accumulation of immature myeloid cells (blasts) in AML patients. Histological staining with Wright-Giemsa is most commonly used to diagnose the disease, but since AML is such a heterogeneous disease, molecular techniques, including qPCR, fluorescent in-situ hybridization (FISH), and next-generation sequencing are also used as a means of identifying the precise subtype of AML [42]. AML is an umbrella term that covers a spectrum of diseases involving excessive proliferation of myeloid tissue. In diagnosing AML, the French-American-British (FAB) criteria is a commonly used classification system [19]. Within FAB, AML cases are subtyped based on a number of criteria, including the type of cell affected, the potential for harbouring a chromosomal translocation, and the aberrant expression of non-myeloid lineage markers. The FAB M2 subtype is most commonly characterized by a chromosomal translocation involving chromosomes 8 and 21.

A.2 The t(8;21) chromosomal translocation

Cytogenetic abnormalities, including chromosomal translocations, are a hallmark of many types of cancer. Recurrent, non-random chromosomal translocations are a common cytogenetic feature of leukemia and can carry both diagnostic and prog- nostic significance [19]. The genomic loci that harbour members of the core-binding factor transcriptional components are frequent targets for disruption that leads to leukemia. In fact, AML1 (also known as RUNX1 or CBFα2 ) was discovered by virtue of its involvement in the t(8;21) chromosomal translocation, the most frequent genetic aberration associated with AML [122]. Work in the early 1990s from Janet Rowley and others, finely mapped the precise chromosomal break-points associated with the t(8;21) event and cloned the affected A.3. WILDTYPE AML1 AND THE CORE BINDING FACTOR 112

gene [147, 148]. Using a cDNA library generated from t(8;21)-associated AML cells, researchers cloned the novel chromosome 8 component, which they referred to as eight twenty-one (ETO) [46]. The chromosome 21 component was identiﬁed as AML1, which has previously been characterized [122]. Though there was no known homolog identiﬁed at the time of its original cloning, there was a localized region of homology between AML1 cDNA and that of the Drosophila segmentation gene runt [46]. Janet Rowley’s group proceeded to clone the entire cDNA of the fusion translocation product, AML1-ETO [46]. Later, PCR, with primers that spanned the fusion point, was used to demonstrate the presence of the t(8;21) abnormality in patient samples. Since the PCR assay was positive in samples taken at diagnosis, remission, and re- lapse, AML1-ETO was considered to be the causal agent in this subtype of leukemia [121].

A.3 Wildtype AML1 and the core binding factor

Core binding factors (CBFs) are heterodimeric transcription factors, in which only one subunit binds DNA directly, while the other has roles in mediating protein-protein interactions that support co-factor recruitment [160]. AML1 is one of three core binding factor subunits, the others being AML2 (CBFα1) and AML3 (CBFα3), which interact with the common CBFβ subunit to complete core binding factor formation. The runt homology domain (RHD) of the CBFα proteins mediates sequence-speciﬁc (5’-PyGPyGGT-3‘) DNA binding. The interaction with the CBFβ subunit primarily increases the aﬃnity with which the CBFα proteins interact with their cognate DNA element. CBFβ is also crucial in recruiting other proteins to the promoter region of its transcriptional targets. At the level of gene regulation, CBF is capable of both A.3. WILDTYPE AML1 AND THE CORE BINDING FACTOR 113

activation and repression in a cell-type-dependant manner. Specific targets include the human IL-3 promoter, granzyme B, and the GM-CSF promoter [166, 173, 174]. CBF is involved in many developmental pathways that include body segmentation in Drosophila and musculo-skeletal formation in mammals as is evidenced by CBF null mutants [44]. In terms of a hematopoietic phenotype, genetic studies using either AML1 - or CBFβ-null transgenic mice showed the importance of CBF function in definitive hematopoiesis. Null mutations in either the AML1 or CBFβ gene cause embryonic lethality at E12.5-E13.5 [132, 152]. Histological sections of the fetal liver showed that mice of either mutant genotype lack definitive hematopoiesis, resulting in death from internal hemorrhaging. Interestingly, primitive hematopoiesis, which occurs earlier in embryonic development and is responsible for the initial burst of red blood cell production, is largely unaffected by CBF disruption. Transplantation of AML1 null HSCs into lethally-irradiated recipient mice yielded no hematopoietic reconstitution, which signified a limited potential for engraftment and significantly reduced stem cell activity. Together, these studies demonstrate the requirement of CBF activity in HSCs and their effect on definitive hematopoiesis and stem cell activity. In addition to its involvement in t(8;21)-associated AML, the CBF is affected in a number of other leukemia subtypes. For instance, AML1 is joined to the ETS-related gene TEL (also known as ETV6 ) in cases of B-ALL associated with the t(12;21) event [184]. In cases of acute myeloid leukemia associated with inv(16) (i.e. an inversion on chromosome 16), CBFβ is juxtaposed against MYH11, which normally encodes the smooth muscle myosin heavy chain gene; the resultant gene product is referred to as CBFβ-SMMHC [105]. A.4. THE ETO FAMILY OF TRANSCRIPTIONAL REPRESSORS 114

Biochemical studies have shown that each of these CBF translocated gene products function as dominant negative inhibitors of wildtype CBF function. To model the in vivo effects of CBF disruption by these leukemia-associated proteins, ”knock- in” studies were performed in which mice were engineered to express AML1-ETO or CBFβ-SMMHC, the fusion gene products of the t(8;21) and inv(16) events, respectively [189, 29]. In both case, the transgenes were expressed under the control of their endogenous promoters and in both cases they resulted in embryonic lethality. Inter- estingly, the phenotype of the AML1-ETO and CBFβ-SMMHC knock-in mice were remarkably similar to that observed in CBF mutant mice in that they lacked definitive hematopoiesis. The observations made using knock-in mouse models reiterate the findings of the biochemical studies, which is that these fusion products represent a class of dominant negative inhibitors of wildtype CBF activity, which is necessary for normal hematopoiesis. Interestingly, none of the knock-in genetic models were able to recapitulate the AML phenotype observed in humans, which suggested that additional mutations may be required to induce leukemia.

A.4 The ETO family of transcriptional repressors

ETO is the founding member of the ETO family of transcriptional repressors, which was identiﬁed by virtue of its involvement in the t(8;21) event associated with acute myeloid leukemia [46, 38]. To date, ETO has been joined by two other family members, MTG16 and MTGR1, which also function as transcriptional repressors [38]. In- terestingly, MTG16 is implicated in leukemia development and is an observed translocation partner with AML1 in therapy-related AML [59]. Heterozygous deletions of the MTGR1 gene has also been associated with both myelodysplastic syndrome (MDS) A.4. THE ETO FAMILY OF TRANSCRIPTIONAL REPRESSORS 115

and AML, further demonstrating that ETO family members are well-rooted in hematological malignancies [52, 90]. In mice, ETO is normally found highly expressed in the brain, heart, and lungs but lower levels of expression are observed in early hematopoietic precursors [26]. Knockout studies have been used to identify the endogenous function of ETO [26]. Because of its expression in early hematopoietic precursors and its involvement in t(8;21)-associated leukemia, it was predicted that ETO knockout would aﬀect the hematopoietic compartment, but this was not the case. While ETO null mice were inviable, the etiology of their death was traced to a malformed gastrointestinal tract that lacked a large portion of the small intestine and colon, rather than to hematopoietic defects, which proceeded as normal [26]. Interestingly, there were no notable defects in brain, heart, or lung formation as a result of ETO knockout, suggesting that, while it has a role in gastrointestinal programming, ETO may perform redundant roles in brain, heart, and lung development. The ETO family is characterized by the presence of four conserved Nervy homology regions (NHR1-4) in their primary structure, so named for their sequence conservation when compared to the Drosophila gene Nervy [38]. NHR1 further shares sequence homology with a number of human TBP-associated factors, speciﬁcally hTAF130 and hTAF105, as well as the Drosophila TAF110 protein [47]. Consequently, NHR1 is often referred to in the literature as the ”ETO TAF homology region” or simply eTAFH. NHR2 contains a number of hydrophobic amino acid residues that give rise the the hydrophobic heptad repeat (HHR) region [192]. While NHR3 is not characterized by any particular protein folding motif, it does mediate the interaction with certain co-repressors, including N-CoR [78]. Finally, NHR4 contains two A.5. THE ONCOGENIC FUSION PROTEIN AML1-ETO 116

zinc finger, which has not been shown to bind DNA but, instead, to mediate additional protein-protein interactions with co-repressors and the protein of unknown significance SON [2]. A strong case has been built over the years for the role of ETO in transcriptional repression. The first study to provide evidence in support of ETO as a co-repressor was by Wang and colleagues in which they identified the human co-repressor N-CoR as a positive hit in a yeast-two-hybrid screen using full-length ETO as bait [178]. This study was soon followed by others that showed a direct interaction between ETO, HDAC1, HDAC2, mSin3A, and the SMRT co-repressor [192, 78, 38, 106]. Since ETO itself does not bind DNA directly, its role in transcriptional silencing can only be facilitated via an interaction with other transcription factors. To date, there are only a handful of DNA-binding transcription factors that have been demonstrated to interact with ETO, which includes PLZF, Gfi-1, and the E-protein family of lymphocyte-specific transcription factors [118, 116, 193]. Both PLZF and Gfi-1 are DNA-sequence-specific transcriptional repressors, whose transcriptional repression is enhanced by co-expression of ETO [118, 116]. The E-protein E2A, a well- characterized interaction partner of the co-activator p300, is involved in transcriptional silencing when co-expressed with ETO [193]. In the case of t(8;21)-associated leukemia, it is suspected that AML1-ETO-induced E-protein silencing contributes to the leukemogenic phenotype [193].

A.5 The oncogenic fusion protein AML1-ETO

The t(8;21) event results in the de novo synthesis of the oncogenic fusion protein AML1-ETO [46]. Structurally, the AML1-ETO fusion protein is comprised of the runt A.5. THE ONCOGENIC FUSION PROTEIN AML1-ETO 117

homology domain (RHD) of AML1 and almost all of the ETO protein, which includes NHR1-4 (Figure A.1). Recently, a shorter isoform of AML1-ETO, which results from diﬀerential splicing of the ETO portion, was identiﬁed [187]. This AML1-ETO9a isoform lacks NHR3 and NHR4 and exhibits more potent leukemogenic activity in mouse bone marrow transplant models. A.5. THE ONCOGENIC FUSION PROTEIN AML1-ETO 118

Figure A.1: Schematic of AML1-ETO and its wildtype precursors

Figure A.1. The t(8;21) event results in the de novo expression of AML1-ETO. The fusion protein retains the RHD of AML1 and almost all of ETO, including its four Nervy homology domains (NHR1-4). AML1-ETO is also expressed endogenously in its truncated form, AML1-ETO9a, due to alternate splicing of the ETO portion. A.5. THE ONCOGENIC FUSION PROTEIN AML1-ETO 119

As expected, based on the conservation of the RHD of AML1, AML1-ETO has been shown to both bind the core enhancer consensus DNA sequence and interact with CBF-beta [119]. While wildtype AML1 functions as both a transcriptional activator and repressor in a context-speciﬁc manner, AML1-ETO appears to function predominantly as a transcriptional repressor due to the presence of ETO NHR sequences [120]. This eﬀect is most noticeable in the context of promoter-reporter assays in which AML1-responsive elements are silenced upon AML1-ETO expression [120]. It was previously shown that, while AML1B can bind and activate transcription from the GM-CSF promoter region, AML1-ETO acts as a dominant negative inhibitor of wildtype AML1 function [53]. Other transcriptional targets of AML1, such as T cell receptor β and IL-3 are, indeed repressed by AML1-ETO expression, suggesting that repression of AML1 target genes is an inherent quality of AML1-ETO [65, 173]. AML1-ETO is a modular protein that contains the RHD and four Nervy homology regions, so it was reasoned that the various functions attributed to the protein may be separable into distinct domains. In the case of transcriptional silencing of AML1 target genes by AML1-ETO, the necessary domains included the DNA-binding RHD and the NHR2-4 regions, which mediate tetramer formation (NHR2) and physical interactions with co-repressor complex members (NHR2-4) [110, 101, 191]. Based on this evidence, the prevailing model of gene repression predicts AML1-ETO localization to a core binding factor consensus sequence, mediated by the RHD, and the subsequent recruitment of co-repressor complexes to these genomic loci [110, 191, 38]. Presumably, oligomerization functions to increase the amount of transcriptional co- repressors recruited to a target locus to some critical mass that shuts down gene transcription [50]. Studies have shown that transcriptional silencing by AML1-ETO A.5. THE ONCOGENIC FUSION PROTEIN AML1-ETO 120

involves a reorganization of local histones and increased histone deacetylase (HDAC) recruitment, which may be reversible through the use of HDAC inhibitors [51]. It was shown that AMl1-ETO is capable of inducing oncogenic transformation of NIH3T3 fibroblast and the oncogenic properties require contributions from both the RHD and ETO portions of the fusion protein [54]. In one of the first studies to demonstrate the anti-differentiation effects of AML1-ETO, Westendorf and colleagues used an in vitro myeloid differentiation assay that involved stimulation of mouse myeloid precursors with the myeloid cytokine G-CSF [182]. It was shown that expression of AML1-ETO in a myeloid progenitor cell line imposed a block in differentiation upon stimulation with G-CSF, a process known to be dependent upon wildtype AML1 [3]. While evidence had accumulated to support the dominant negative effects of AML1-ETO on AML1-mediated transcription, an interesting finding in the above study pertained to the ability of AML1-ETO to affect transcription of another endogenous transcription factor, C/EBPα. It was showed using co-immunoprecipitation assays from co-transfected COS-7 cells that AML1-ETO and C/EBPα interact in vivo [182]. Luciferase reporter assays were also used to demonstrate that the rat NP-3 promoter, which contains both AML1 and C/EBPα-responsive elements, up- regulates reporter gene expression in response to exogenously expressed AML1B or C/EBPα [182]. AML1-ETO was shown to reduce AML1B-mediated activation of the luciferase reporter by about 66% whereas it completely inhibited C/EBPα-mediated transcription [182]. This finding lead to the hypothesis that AML1-ETO may induce leukemia, in part, by blocking gene activation by AML1, C/EBPα and, potentially, other transcription factors. The anti-differentiation effects of AML1-ETO was further A.6. MODELLING THE EFFECTS OF AML1-ETO EXPRESSION IN VIVO 121

demonstrated in the K562 leukemia cell line, a model of human hematopoietic differentiation that can differentiate down the erythroid lineage upon stimulation with Ara-C [99]. By stably expressing AML1-ETO, K562 cells became refractory to stimulation by Ara-C and were unable to efficiently differentiate into erythroid precursors.

A.6 Modelling the eﬀects of AML1-ETO expression in vivo

A step toward the generation of a mouse model of human AML was published in a let- ter to Nature in 1997 [189]. Yergeau and colleagues created an AML1-ETO knock-in mouse model, in which a single allele of the AML1 gene was replaced by AML1-ETO cDNA, with the intent to model human AML in mice. They found, however, that instead of developing leukemia, AML1-ETO/+ transgenic mice died before birth at around E12.5-E14.5. Interestingly, AML1-ETO/+ mice suffered from central nervous system hemorrhaging and lacked a fetal liver, a phenotype that was strikingly similar to AML null embryos which had intact primitive hematopoiesis but lacked definitive hematopoiesis [132]. One distinction between AML1-ETO knock-in vs AML1 null mice was that AML1-ETO/+ yolk sac-derived hematopoietic progenitors were able to differentiate into macrophages in a colony-forming assay whereas AML1 null progenitors failed to differentiate at all [189, 132]. A related study by James Downing’s group in 1998, replicated the above findings with an AML1-ETO knock-in model of their own and also characterized a small population of surviving fetal liver-derived hematopoietic precursors [131]. They observed that these AML1-ETO-expressing hematopoietic precursors differentiated in vitro into dysplastic granulocyte-monocyte and multilineage colonies. Surprisingly, these cells had an extraordinary capacity for self renewal, well beyond that observed in wildtype fetal liver-derived hematopoietic A.7. THE CURIOUS PHENOTYPE OF T(8;21)-ASSOCIATED AML CELLS 122 precursors [131]. To further investigate the role of AML1-ETO in cell immortalization, adult bone marrow-derived hematopoietic cells were transduced with AML1- ETO and used in serial replating assays. They found that AML1-ETO expression resulted in the production of immature blast-like myeloid cells with a high capacity for serial replating compared to terminally differentiated control cells [131]. Taken together, these studies highlight the oncogenic nature of AML1-ETO, which not only functions as a dominant negative inhibitor of endogenous AML1, but appears to also skew the differentiation program toward a myeloid fate while conferring seemingly unlimited replicative capacity. Since AML1-ETO was demonstrated to have such a profound effect on embryonic survival, transgenic mice that expressed an inducible form of AML1-ETO were engineered to circumvent this block. A tetracycline-inducible mouse model of AML1- ETO expression was generated by Dong-Er Zhang’s group [143]. Inducible expression of AML1-ETO was shown to occur in a number of tissue types, including the bone marrow and in peritoneal macrophages. Incredibly, expression of AML1-ETO had no overt effects on normal hematopoiesis. The authors did, however, note that AML1- ETO expression conferred increased proliferation and self-renewal capacity, which confirmed previous findings.

A.7 The curious phenotype of t(8;21)-associated AML cells

Typically, leukemia subtypes are grouped based on their antigen expression, similar to the categorization of normal hematopoietic cells [19]. In the context of normal hematopoiesis, myeloid (e.x. CD11b, Gr-1) and lymphoid (e.x. CD19, CD45R/B220) cell surface markers segregate with the appropriate lineage. In certain types of A.8. AML1-ETO INTERACTS WITH E2A AND INDUCES E-PROTEIN SILENCING 123 leukemia, including t(8;21)-associated AML, there is a mixing of immunophenotypic markers, resulting in a bi-phenotypic leukemia [112]. In 1994, the Japanese Coop- erative Group of Leukemia/Lymphoma studied the blast cells from a large number of t(8;21) AML patients [89]. They found that hematopoietic stem-cell markers like CD34 and HLA-DR were expressed on the surface of t(8;21) blast cells in higher amounts compared to other FAB-M2 AML subtypes. Curiously, blast cells also showed frequent, aberrant surface expression of the B-lymphocyte markers CD19, which contributes to the bi-phenotypic distinction [89]. This ﬁnding was conﬁrmed by another group in 2008 that discovered co-expression of CD19 and myeloid antigens on t(8;21)-associated blasts [75]. Expression of the B-lymphocyte transcription factor PAX5 was also detected at both the mRNA and protein level in a sample of AML cells [170]. Together, these data allude to a potential connection between AML1-ETO and B lymphopoiesis.

A.8 AML1-ETO interacts with E2A and induces E-protein silencing

In 2004, a seminal article on AML1-ETO from Bob Roeder’s group, demonstrated a novel interaction between AML1-ETO and the E-protein HEB [193]. First, they used a co-immunoprecipitation-mass spectroscopy strategy in ETO-transfected HeLa cells, which identiﬁed MTGR1 and HEB as ETO interaction partners. They next demonstrated that both ETO and AML1-ETO caused silencing of HEB-mediated transcriptional activation in a reporter assay. The interaction surfaces were mapped broadly to NHR1 (referred to as eTAFH by the authors) of ETO/AML1-ETO and A.8. AML1-ETO INTERACTS WITH E2A AND INDUCES E-PROTEIN SILENCING 124

AD1 of HEB. Based on their experimental ﬁndings, they proposed a model of E- protein silencing mediated by AML1-ETO in which AML1 localizes to E-protein- bound regions of the genome and recruits co-repressors via NHR2-4, thereby silencing E-protein responsive chromatin (Figure A.2). A.8. AML1-ETO INTERACTS WITH E2A AND INDUCES E-PROTEIN SILENCING 125

Figure A.2: E-protein silencing by AML1-ETO

Figure A.2. AML1-ETO has been shown to interact with the E-protein HEB in vitro. AML1-ETO is suspected to compete for p300 binding at the AD1 domain of E-proteins. Displacement of p300 results in the stable recruitment of co-repressor complexes and silencing of the associated target genes (Adapted from Zhang et al., 2004). A.9. RATIONALE AND HYPOTHESIS 126

To study the genome-wide effects of AML1-ETO expression on endogenous AML1 and HEB gene regulation, Gardini and colleagues performed ChIP-chip for the above factors in the U937 cell line [61]. They discovered that AML1-ETO localizes to wildtype AML1 binding sites upon expression in U973 cells. They also observed a profound rearrangement in the genomic localization of HEB upon AML1-ETO expression. This result confirmed the previously-described in vitro interaction between HEB and AML1-ETO. To further investigate the interaction between AML1-ETO and the E-protein HEB, John Bushweller’s group solve the crystal structure of the NHR1 domain in complex with the HEB AD1 domain [137]. They identified key residues that mediated the interaction and demonstrated that disruption of these residues resulted in dissociation of the complex. Previous studies have shown that expression of AML1- ETO can both block granulocyte differentiation in vitro and increase the self-renewal capacity of hematopoietic progenitor cells [131, 25, 166]. Interestingly, the mutations in NHR1 that disrupted the interaction between HEB and AML1-ETO had no apparent effect on either of these functions of AML1-ETO [137].

A.9 Rationale and Hypothesis

This above demonstration that NHR1 of AML1-ETO is dispensable for leukemogenesis is somewhat surprising since a functional relationship between AML1-ETO and HEB had previously been established. One explanation for the discordance might be that the authors were measuring the eﬀects of AML1-ETO on myeloid cell diﬀer- entiation and self-renewal, which are processes that occur largely independently of E-proteins. To truly examine the relationship between AML1-ETO and E-proteins, A.9. RATIONALE AND HYPOTHESIS 127

it is necessary to use a cellular model that is dependent on E-protein function.

A.9.1 Hypothesis:

Noting the correlation between AML1-ETO and bi-phenotypic leukemic blasts, it stands to reason that disruption of wildtype E-protein function upon AML1-ETO expression in hematopoietic precursors disrupts the normal lymphoid program, leading to the malignant transformation of a lymphocyte precursor and leukemia development.

A.9.2 Experimental objectives:

To address this hypothesis, I have used (1) an ex vivo hematopoietic differentiation model, originally described by Juan Carlos Zuniga-Pflucker, to examine the role of AML1-ETO in antagonizing the E-protein-dependent process of lymphopoiesis, and (2) an orthogonal, in vitro model of E-protein function to measure the effects of AML1-ETO on immunoglobulin chain gene rearrangement in BOSC23 cells. 128

Appendix B

AML1-ETO Results

B.1 AML1-ETO interacts with E2A in an NHR1- and NHR2-dependent manner

AML1-ETO has been previously shown to interact with the E-protein HEB [193]. The interaction domains have been mapped to NHR1 of AML1-ETO and the AD1 activation domain of HEB [193, 137]. While the activation domains of E-protein family members are highly conserved, it has not been established whether AML1-ETO could interact with the activation domains of E2A. To determine whether AML1- ETO interacts with the lymphocyte-specific E-protein E2A, co-immunoprecipitation experiments were performed using cell lysate from transfected 293T cells (Figure B.1). 293T cells were co-transfected with the E2A isoform E47, AML1-ETO, and a number of AML1-ETO variants with NHR1 and NHR2 mutations that specifically disrupt its interaction with HEB. Immunoprecipitation was performed using an anti- E2A antibody and, after washing to remove non-specific or weak interactions, immunoblotting was performed using an anti-AML1 antibody, which recognized the AML1-ETO fusion protein. In cells co-transfected with wildtype AML1-ETO and B.1. AML1-ETO INTERACTS WITH E2A IN AN NHR1- AND NHR2-DEPENDENT MANNER 129

E2A, immunoblotting showed that these proteins interact in vivo, thereby conﬁrming the expectation that AML1-ETO would interact with the highly conserved E-protein activation domains. Co-transfection of E2A with AML1-ETO mutants deﬁcient in HEB binding was used to identify the region of AML1-ETO that contribute to the interaction. Results of the co-immunoprecipitation assay showed that point mutations in NHR1 or NHR2 diminish the interaction between AML1-ETO and E2A. Furthermore, by mutating both of these domains simultaneously, the interaction between AML1-ETO and E2A was almost completely disrupted. Together, these data demonstrate that the NHR1 and NHR2 domains both contribute to the interaction between AML1-ETO and E2A. B.1. AML1-ETO INTERACTS WITH E2A IN AN NHR1- AND NHR2-DEPENDENT MANNER 130

Figure B.1: AML1-ETO interacts with E2A in a NHR1- and NHR2-dependent manner

Figure B.1. Immunoprecipitation was performed in 293T cells co-transfected with E2A and the indicated AML1-ETO mutants. Single-amino acid substitution mutations in either the NHR1 or NHR2 domains of AML1-ETO disrupted its interaction with E2A. An AML1-ETO construct with combined mutations in these domains failed to interact with E2A almost completely. B.2. E-PROTEIN SILENCING BY AML1-ETO IS NHR1- AND NHR2-DEPENDENT 131

B.2 E-protein silencing by AML1-ETO is NHR1- and NHR2-dependent

Previous studies have shown that AML1-ETO can repress HEB-GAL4 mediated transcriptional activation of a upstream activating sequence (UAS) sequence in the context of a luciferase reporter assay in a manner dependent on interaction with the AD1 domain of HEB [193]. To determine whether transactivation by wildtype E2A was subject to silencing by AML1-ETO, luciferase assays using an E2A-responsive reporter were performed (Figure B.2). First, the E47 isoform of E2A and the EE6 reporter plasmid, which contains 6 tandem copies of the E-box element, was transfected into 293T cells to establish base- line transactivation levels by E2A [21] (Figure B.2A). Next, increasing amounts of a vector encoding AML1-ETO was co-transfected with E2A and the reporter plasmid. AML1-ETO repressed transactivation by E2A in a dose-dependent manner and almost completely silenced reporter activity when AML1-ETO was transfected in a 2:1 ratio with E2A. This result demonstrates that AML1-ETO is a potent inhibitor of transcriptional activation by both E2A in vitro. To identify the regions of AML1-ETO that contribute to E-protein silencing, luciferase assays using engineered AML-ETO deletion mutants and the naturally truncated AML1-ETO9a isoform was performed(Figure B.2B). AML1-ETO and the truncated AML1-ETO9a variant both eﬀectively repressed E-protein mediated transactivation. Co-transfection of E2A with the AML1-ETO ∆NHR1 mutant resulted in a near complete restoration of wildtype E-protein transactivation. Deletion of NHR2, however, only partially restored the transactivation capabilities of E2A. These ﬁnd- ings suggests that the NHR1 domain is primarily responsible the silencing of E- protein transcriptional activity. Interestingly, the data also shows that NHR1 and B.2. E-PROTEIN SILENCING BY AML1-ETO IS NHR1- AND NHR2-DEPENDENT 132

NHR2 collaborate during E-protein silencing since the ∆NHR2 mutant, which contains an intact NHR1, is unable to repress transactivation to the same extent as wildtype AML1-ETO or AML1-ETO9a. Furthermore, the data indicate that NHR3 and NHR4 may act to reduce the silencing abilities of AML1-ETO since the AML1- ETO9a mutant, which lacks these domains, is a more potent transcriptional repressor than its wildtype counterpart. B.2. E-PROTEIN SILENCING BY AML1-ETO IS NHR1- AND NHR2-DEPENDENT 133

Figure B.2: AML1-ETO silences transactivation by E-proteins in an NHR1- and NHR2-dependent manner

Figure B.2.(A) Luciferase reporter assays were performed in 293T cells to evaluate the eﬀect of AML1-ETO expression on E2A-mediated transcriptional activity. Co- transfection with AML1-ETO inhibits transactivation by E2A in a dose-dependent manner. (B) Co-transfection of AML1-ETO mutants demonstrated that NHR1 and NHR2 contribute to the repression of E2A-mediated transcription. B.3. EXAMINING THE EFFECTS OF AML1-ETO ON IMMUNOGLOBULIN CHAIN GENE REARRANGEMENT 134

B.3 Examining the eﬀects of AML1-ETO on immunoglobulin chain gene rearrangement

There is an abundance of E-box sequences in the regulatory regions of the human immunoglobulin genes that have been shown to bind E2A in vitro and in vivo [141, 67]. It has been previously demonstrated that enforced expression of E2A, RAG1 and RAG2 in the epithelial BOSC23 cell line can induce V(D)J recombination at the Igκ locus [63]. More recently, it has been shown that V(D)J rearrangement involves cooperation between E2A and p300 [150]. A PCR-based assay can be used to detect rearrangements of the immunoglobulin gene locus and measure E2A activity at that site (Figure B.3). The forward and reverse primers hybridize to the V and J regions, respectively (Figure B.3A). In the germline conﬁguration, these regions are suﬃciently distal from each other that a PCR reaction fails to produce a product. When the locus is re- arranged via RAG1 and RAG2 recombinase and E2A activity, however, the V and J regions are brought in close proximity of each other and a positive PCR result is observed. This system was used as a biological assay of E2A function to determine whether antagonism of E-protein activity by AML1-ETO would disrupt imm- munoglobulin chain gene rearrangement. B.3. EXAMINING THE EFFECTS OF AML1-ETO ON IMMUNOGLOBULIN CHAIN GENE REARRANGEMENT 135

Figure B.3: Rearrangement of the immunoglobulin kappa light chain is an E2A- dependent process

Figure B.3.(A) Schematic of the human immunoglobulin kappa light chain. PCR is used to detect rearrangements in this gene using primers that localize to the variable and joining regions of the immunoglobulin gene. (B) In BOSC23 cells, immunoglobulin kappa light chain rearrangement occurs in an E2A-dependent manner. The transfected proteins are indicated with a + sign in the table below the ﬁgure. Expected size of PCR product is indicated with an arrow. B.3. EXAMINING THE EFFECTS OF AML1-ETO ON IMMUNOGLOBULIN CHAIN GENE REARRANGEMENT 136

Full-length AML1-ETO, AML1-ETO9a, AML1-ETO ∆NHR1, and AML1-ETO ∆NHR2 were each co-transfected into BOSC23 cells along with E2A, and the recombinases RAG1 and RAG2 (Figure B.4). ID2, ID3, and TAL1 are known E-protein inhibitors and were used as positive controls for impaired E-protein function. While co-expression of ID2 did not appear to inhibit immunoglobulin gene rearrangement, co-expression of ID3 and TAL1 seem to impose a partial block in rearrangement, as indicated by diminished PCR bands. AML1-ETO, however, was shown to be incapable of impairing immunoglobulin gene rearrangement. This ﬁnding contrasts with the observations made in luciferase reporter experiments, which showed that AML1-ETO can impose a block in E-protein function.

Figure B.4: AML1-ETO does not inhibit VDJ rearrangement in BOSC23 cells

Figure B.4.PCR was performed on BOSC23 genomic DNA to detect immunoglobulin chain gene rearrangement after transfection with E2A antagonists, including AML1-ETO. Expected size of PCR product is indicated with an arrow. B.4. AML1-ETO BLOCKS DIFFERENTIATION AND ENHANCES LONG-TERM PROLIFERATION OF MYELOID PROGENITORS 137

B.4 AML1-ETO blocks diﬀerentiation and enhances long-term proliferation of myeloid progenitors

The following series of experiments that examine the effects of AML1-ETO on HSC differentiation, through the use of ex vivo assays, were performed in order to understand the biological roles of AML1-ETO in a hematopoietic context and complement the in vitro results presented previously. AML1-ETO has been shown to block differentiation in a number of hematopoietic cell lines, including the granulocytic differentiation of 32Dcl3 mouse myeloid cells and the erythroid differentiation of K562 cells [3, 99]. AML1-ETO has also been shown to block granulocytic differentiation of hematopoietic progenitors upon stimulation with G-CSF [106]. To confirm the previous findings that AML1-ETO can block differentiation of granulocyte precursors, HSC-enriched bone marrow cells from 5-fluorouracil (5-FU)-treated mice were infected with either an empty control vector or one that expressed AML1-ETO. After two weeks of culture in myeloid-promoting cytokines (i.e. SCF, IL-3, IL-6, and G-CSF), flow cytometry was performed to assess the differentiation status of the developing hematopoietic cells. Control cells, infected with the unloaded vector, developed into neutrophils and expressed the myeloid markers CD11b (also known as Mac-1) and Gr-1 (Figure B.5A). Cells infected with AML1- ETO, however, showed a marked decrease in neutrophil differentiation compared to control cells. This result confirms previous observations that AML1-ETO can block the granulocytic differentiation of hematopoietic precursors. AML1-ETO has also been shown to enhance the clonogenicity of myeloid precursors in vitro through the use of serial-replating assays [81, 154]. To confirm this result, cell numbers were counted after infection with the empty control vector or B.4. AML1-ETO BLOCKS DIFFERENTIATION AND ENHANCES LONG-TERM PROLIFERATION OF MYELOID PROGENITORS 138

AML1-ETO (Figure B.5B). Cells infected with AML1-ETO showed an increased, sustained rate of proliferation compared to control cells, which likely underwent terminal differentiation. These data confirm previous findings that AML1-ETO can block differentiation and enhance the long-term proliferation of hematopoietic precursors. B.4. AML1-ETO BLOCKS DIFFERENTIATION AND ENHANCES LONG-TERM PROLIFERATION OF MYELOID PROGENITORS 139

Figure B.5: AML1-ETO inhibits diﬀerentiation and increases proliferation of myeloid progenitors in vitro

Figure B.5.(A) Bone marrow cells terminally diﬀerentiate into CD11b+Gr-1+ neutrophils when grown in myeloid-promoting conditions (IL-3, IL-6, SCF, G-CSF). In- fection with AML1-ETO largely prevents the diﬀerentiation of immature hematopoietic precursors into neutrophils. (B) Enforced expression of AML1-ETO in mouse bone marrow hematopoietic progenitors increases their proliferation when grown under myeloid-promoting conditions (IL-3, IL-6, SCF, G-CSF). B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 140

B.5 Examining the role of AML1-ETO on B lymphocyte development

Previous studies using E2A null mice, which are almost completely devoid of B lymphocytes, have demonstrated the requirement for E2A in normal B lymphocyte development [11]. To study the effects of AML1-ETO on E-protein function in a hematopoietic context, an ex vivo lymphocyte differentiation assay was used (Fig- ure B.6) [80]. The differentiation of mouse hematopoietic stem cells is achieved by culturing the cells on a layer of irradiated OP9 stromal cells and supplementing the media with SCF and the lymphoid cytokine IL-7. After a period of approximately 7-14 days, the expansive outgrowth of CD19+B220+ B lymphocyte precursors is observed.

Figure B.6: Schematic of ex vivo lymphoid diﬀerentiation assay

Figure B.6.Lineage negative, uncommitted hematopoietic progenitors are puriﬁed from mouse fetal livers and cultured in lymphocyte-promoting conditions (OP9 stromal cells and media containing SCF and IL-7). CD19+B220+ B lymphocytes are produced after about 7 days. B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 141

To study the effects of AML1-ETO on lymphocyte differentiation, mouse HSCs were infected with retroviral vectors encoding full-length AML1-ETO, AML1-ETO9a, AML1-ETO ∆NHR1, or AML1-ETO ∆NHR2. The retroviral vectors used in this experiment also contained an IRES-GFP cassette downstream of the transgene, which facilitated the identification of transduced cells. Flow cytometry was used to measure the proportion of cells that were infected with the retroviral vector and cells were also enumerated to measure the effect of AML1-ETO on the rate of their proliferation (Figure B.7). B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 142

Figure B.7: AML1-ETO expression reduces proliferation and survival of B lymphocyte precursors B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 143

Figure B.7. (A) Flow cytometry was used to detect the persistence of cells infected with various AML1-ETO mutants. (B) Cell counts show the change in each population over the course of the assay. The initial population of cells expressing wildtype AML1-ETO and the NHR1-deletion mutant failed to proliferate extensively and died, presumably, by apoptosis, whereas cells infected with the empty GFP vector and cells expressing NHR2-deleted AML1-ETO proliferated at similar rates. Hematopoietic precursors expressing AML1-ETO9a experienced a brief period of expansion, which tapered oﬀ around day 21, and the cells were eventually eliminated from culture. B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 144

Expression of full-length AML1-ETO greatly reduced the rate of proliferation of the emerging lymphocyte precursors in this assay. Despite comprising nearly 12% of the original population, the prevalence of cells expressing AML1-ETO dwindled to less than 1% after 14 days in culture. In contrast, cells infected with the empty vector, which confers expression of GFP alone, were abundant and proliferated well in culture. Surprisingly, expression of the AML1-ETO ∆NHR1 mutant, which was unable to block E2A-mediated transactivation, was equally as effective at inhibiting the expansion of lymphoid precursors in this assay as full-length AML1-ETO. The AML1-ETO9a truncation mutant, which retains both the NHR1 and NHR2 domains, initially allowed the expansion of hematopoietic progenitors during the first two weeks of the assay but, ultimately, prevented the long-term survival of the infected cells. Finally, expression of the ∆NHR2 mutant, which only partly inhibited E2A transactivation, appeared to not affect proliferation of these hematopoietic precursors. In fact, growth curve analysis indicates that cells infected with AML1-ETO ∆NHR2 proliferate at a similar pace as those expressing GFP alone (compare green and purple curves in figure B.7). It was shown earlier that AML1-ETO blocks the granulocytic differentiation of murine bone marrow-derived hematopoietic progenitors. Immunophenotyping was performed to determine whether AML1-ETO imposed anti-differentiation effects during B lymphopoiesis (Figure B.8). Expression of CD19, an indicator of B lymphocyte differentiation, was detected on hematopoietic progenitors infected with an empty control retroviral vector. At the time of assay, there were very few cells (<1%) infected with AML1-ETO, as indicated by GFP expression, but CD19 was detected by flow cytometry, nonetheless. Positive staining for CD19 was further observe in cells B.5. EXAMINING THE ROLE OF AML1-ETO ON B LYMPHOCYTE DEVELOPMENT 145

expressing AML1-ETO9a, AML1-ETO ∆NHR1, and AML1-ETO ∆NHR2. Together, this collection of data demonstrates that (1) the anti-proliferative eﬀects of AML1- ETO on immature hematopoietic precursors is mediated primarily by the NHR2 domain and (2) expression of AML1-ETO does not inhibit B lymphocyte diﬀerentiation in the context of this assay.

Figure B.8: AML1-ETO does not block E-protein-dependent commitment to the B lymphoid lineage

Figure B.8. Immunophenotying of hematopoietic progenitors shows the outgrowth of CD19+ B lymphocyte precursors. While expression of wildtype AML1-ETO lead to a near absence of transduced cells at the time of immunophenotyping, the limited number of surviving cells appear to express the B lymphocyte marker CD19. Inter- estingly, the AML1-ETO9a and AML1-ETO ∆NHR2 variants support the outgrowth of CD19+ lymphocytes despite the presence of the inhibitory NHR1. 146

Appendix C

AML1-ETO Discussion

AML1-ETO has an established association with the bi-phenotypic subtype of leukemia, which is typiﬁed by simultaneous expression of myeloid and B lymphoid cellular traits [95, 75]. The subsequent observation that AML1-ETO can physically interact with and silence transactivation mediated by lymphopoietic E-proteins suggests that a disruption of E-protein signaling may be an important oncogenic event in t(8;21)- associated leukemia [193]. The in vitro promoter-reporter assays presented in Ap- pendix A show that AML1-ETO functions as a potent inhibitor of E-protein mediated transactivation predominantly in an NHR1-dependent manner (see Figure B.2). In contrast to the results of the promoter-reporter assay, NHR1-intact AML1-ETO does not appear to impede either the process of immunoglobulin gene rearrangement or lymphocyte diﬀerentiation, which are both critically dependent on the transactivating function of the E-protein E2A. C.1. SILENCING OF E2A TRANSACTIVATION BY AML1-ETO IS MEDIATED BY NHR1 AND NHR2 147

C.1 Silencing of E2A transactivation by AML1-ETO is mediated by NHR1 and NHR2

In their initial study, Zhang and colleagues showed that the physical and functional interactions between AML1-ETO and the E-protein HEB were mediated by NHR1 of AML1-ETO and AD1 of HEB [193]. In this study, co-immunoprecipitation was used to show that AML1-ETO also physically interacts with the E-protein family member E2A and that this interaction depends on amino acid residues in NHR1 and NHR2 of AML1-ETO. The conservation of the interaction between AML1-ETO and E2A was expected since there is exceptional sequence similarity in the activation domains of E-protein family members [35, 77]. In fact, the p300/CBP and ETO target (PCET) motif within AD1 of HEB, described as a critical mediator of its interaction with AML1-ETO, is completely conserved among all human E-protein family members [193]. NHR1 and NHR2 of AML1-ETO, which were critical for the interaction with E2A, were each shown in luciferase assays to contribute to the silencing of E2A- mediated transactivation, thereby highlighting the importance of physical contact for E-protein silencing.

C.2 AML1-ETO does not aﬀect E2A-dependent B lymphopoietic activity

An assay to measure immunoglobulin κ (Igκ) gene rearrangement in BOSC23 epithelial cells was used as an experimentally tractable model of an E2A-dependent lymphopoietic process [63]. While known inhibitors of E2A, specifically ID3 and TAL1, were capable of inhibiting Igκ chain rearrangement in BOSC23 cells, AML1-ETO did not block rearrangement at this locus. Using an ex vivo differentiation assay, AML1- ETO failed to inhibit lymphocyte differentiation from uncommitted hematopoietic C.3. CONTEXT-SPECIFIC ROLE OF AML1-ETO IN TRANSCRIPTIONAL REGULATION 148

progenitors. This is, once again, in striking contrast to the observation that AML1- ETO is a potent inhibitor of E-protein transactivation in vitro. Additionally, the observations in this appendix show that full-length AML1-ETO, AML-ETO9a, and AML1-ETO ∆NHR1 each exert strong anti-proliferative eﬀects on B lymphocyte precursors. Interestingly, the AML1-ETO ∆NHR2 mutant was the only variant that failed to produce the anti-proliferative phenotype in B lymphocyte precursors, despite the presence of NHR1, which is inhibitory to E2A.

C.3 Context-speciﬁc role of AML1-ETO in transcriptional regulation

AML1-ETO has well-established roles as a potent negative regulator of transcription by way of its ability to recruit co-repressors to sites of DNA binding [120]. Interest- ingly, AML1-ETO has also been shown to activate transcription of a number of genes, including the ubiquitin-specific protease UBP43, Bcl-2, and M-CSF-R [92, 104, 144]. While the mechanism of gene activation was largely unknown for many years, recent evidence suggests that a direct interaction between AML1-ETO and the co-activator p300 is required for gene activation [180]. Together, these observations portray AML1- ETO as a dual-specificity transcriptional regulator whose activity is determined in a context-specific manner. The context-specific duality may also help to explain the discrepancy between the in vitro antagonism of E2A-driven reporter activity and the failure of AML1-ETO to block the differentiation of murine B-lymphocyte progenitors in ex vivo cultures. Specifically, the luciferase reporter used in this study contains four tandem copies of the E-box element upstream of a luciferase reporter gene. This arrangement favours the repressive mode of gene regulation by AML1-ETO since it encourages a dense C.3. CONTEXT-SPECIFIC ROLE OF AML1-ETO IN TRANSCRIPTIONAL REGULATION 149

accumulation of co-repressor molecules at the site of DNA-binding. Under physiolog- ical circumstances, for example at the Igκ or Cd19 promoters, E2A-responsive E-box elements are present at lower a frequency compared to that of the luciferase reporter. This reduces the ability of AML1-ETO to recruit the critical mass of co-repressors necessary to shut down gene expression at critical E-protein target genes relevant to lymphopoiesis. Additionally, E-box elements are often found adjacent to DNA consensus motifs recognized by other hematopoietic transcription factors, including RUNX1, EBF1, and ETS, which suggests that these transcription factors collaborate to regulate expression of their transcriptional targets. The silencing effect of AML1- ETO observed in vitro may effectively be diluted at endogenous gene promoters due to the collaborative activity of transcriptional regulators acting at E-protein-responsive elements. The dual role of AML1-ETO as a transcriptional repressor and activator is not unique to this fusion transcription factor. Indeed, a number of transcription factors can both activate and inhibit gene transcription, including TBX20, NFY, C/EBPα, AML1, and E2A [103, 109, 140, 149, 186]. Recent evidence presented in Chapter 1 and Appendix D demonstrate that E2A-PBX1, hitherto characterized as a potent transcriptional activator, contributes to both the positive and negative regulation of gene expression since E2A-PBX1 knockdown in RCH-ACV cells resulted in both up- and downregulation of its operationally-defined target genes [123, 16]. These two leukemogenic fusion proteins are, therefore, linked in their mutual abilities to contribute to both gene activation and silencing. It is also interesting to note that AML1-ETO and E2A-PBX1 both localize to genomic regions that are regulable by wildtype E2A. C.4. CONCLUSIONS AND SUGGESTIONS FOR FUTURE EXPERIMENTS 150

This commonality suggests that a disruption of E2A-regulable transcriptional networks may play an important role in both myeloid and lymphoid leukemogenesis. E2A-regulated networks may, therefore, represent a viable therapeutic target in both t(8;21)- and t(1;19)-associated leukemia cases.

C.4 Conclusions and suggestions for future experiments

Data presented in this appendix demonstrate that AML1-ETO functions as a potent E2A antagonist in vitro but not in vivo, at least not during the E2A-dependent processes of Igκ rearrangement and B lymphopoiesis. These incongruous results lead to the conclusion that E2A antagonism by AML1-ETO occurs in a context-specific manner. The in vivo existence of the co-factor exchange mechanism of E-protein silencing originally presented by Zhang et al. remains to be demonstrated. In future attempts to interrogate this model of gene regulation, ChIP-sequencing should be used to determine whether AML1-ETO antagonizes p300 recruitment to sites of E2A binding on a genome-wide scale. To further determine whether co-factor exchange occurs and to identify the precise E2A-bound loci that are affected, ChIP-seq should also be performed to study the recruitment of HDACs and other co-repressors to sites of E2A and AML1-ETO co-occupancy. Functional changes in gene expression and the local chromatin environment should also be measured near sites of AML1-ETO and E2A co-occupancy using RNA-seq and modified histone ChIP-seq, respectively. The results of the experiments proposed above will contribute to a more thorough understanding of the role of AML1-ETO in the antagonism of E-proteins and provide insight into the context-specific duality of this leukemogenic transcription factor. 151

Appendix D

E2A-PBX1

The genomic binding characteristic of E2A-PBX1 were investigated and presented in Chapter 1 of this thesis. In Appendix B, the functional eﬀects of E2A-PBX1 binding on expression of its operationally-deﬁned target genes, as well as the overall contribution of E2A-PBX1 expression to the phenotype of RCH-ACV cells will be addressed experimentally.

D.1 E2A-PBX1 contributes to both up- and downregulation of its operationally- deﬁned target genes

Using promoter-reporter assays, E2A-PBX1 has been previously demonstrated as a potent transcriptional activator [100, 108]. Many of the reporter constructs used to assess the transactivation capabilities of E2A-PBX1 contained PBX1 consensus elements cloned directly upstream of the reporter gene (e.x. luciferase or chloram- phenicol acetyltransferase). In Chapter 1, E2A-PBX1 binding was shown to overlap with genomic regions containing wildtype E2A and EBF1 transcription factor binding sites (Figure 3.8). Furthermore, E2A-PBX1 binding was shown to occur predominantly in intronic and intergenic regions that are often many kilobases away from the D.1. E2A-PBX1 CONTRIBUTES TO BOTH UP- AND DOWNREGULATION OF ITS OPERATIONALLY-DEFINED TARGET GENES 152 transcriptional start site of the nearest gene (Figure 3.3). To determine the regulatory effects of E2A-PBX1 on its operationally-defined target genes, RNA-sequencing was performed after siRNA-mediated knockdown of E2A-PBX1 in RCH-ACV cells. E2A-PBX1 knockdown was shown to induce both up- and downregulation of its operationally-defined target genes, suggesting that E2A-PBX1 may act as either a transcriptional activator of inhibitor in a gene-specific context Figure D.1. D.1. E2A-PBX1 CONTRIBUTES TO BOTH UP- AND DOWNREGULATION OF ITS OPERATIONALLY-DEFINED TARGET GENES 153

Figure D.1: Response of the operationally-deﬁned E2A-PBX1 target genes to E2A- PBX1 knockdown

Figure D.1. E2A-PBX1 expression in RCH-ACV cells was silenced using siRNA and RNA-seq was performed following gene knockdown. RNA-seq analysis showed that E2A-PBX1 knockdown resulted in both up- and down-regulation of its operationally- deﬁned target genes. Gene expression in RCH-ACV cells depleted of E2A-PBX1 is plotted relative to control cells treated with siRNA targeting Lamin A/C. D.2. E2A-PBX1 PROMOTES CELL CYCLE AND OPPOSES HEMATOPOIETIC DIFFERENTIATION 154

D.2 E2A-PBX1 promotes cell cycle and opposes hematopoietic diﬀeren- tiation

The previous analysis was limited to the operational targets of E2A-PBX1; to study the overall effects of E2A-PBX1 on cellular pathways in RCH-ACV cells, differential expression analysis was performed on a transcriptome-wide level after E2A-PBX1 knockdown. Genes whose expression decreased to 70% of control (siLamin A/C) levels were considered downregulated and those whose expression increased to 150% of control levels were considered upregulated. Genes that were differentially-expressed in siPBX1-treated vs. control cells were subject to KEGG pathway analysis to identify the biological pathways that were affected by E2A-PBX1 knockdown. KEGG analysis showed that E2A-PBX1 knockdown resulted in decreased expression of genes involved in cell cycle progression and a concomitant increase in expression of genes involved in hematopoietic and lymphopoietic differentiation (Figure D.2). Taken together, these data suggest that E2A-PBX1 may contribute to leukemia development by stimulating cell division while blocking differentiation. D.2. E2A-PBX1 PROMOTES CELL CYCLE AND OPPOSES HEMATOPOIETIC DIFFERENTIATION 155

Figure D.2: Cellular pathways aﬀected by E2A-PBX1 knockdown D.2. E2A-PBX1 PROMOTES CELL CYCLE AND OPPOSES HEMATOPOIETIC DIFFERENTIATION 156

Figure D.2. KEGG pathway analysis was performed to identify cellular pathways that were differentially expressed in response to E2A-PBX1 knockdown in RCH-ACV cells. For KEGG pathway analysis, genes whose expression was decreased to levels below 70% of siLam control were considered downregulated; genes whose expression was increased to levels above 150% of control were considered upregulated. KEGG pathways significantly associated with genes that were upregulated (green bars) or downregulated (yellow bars) in response to E2A-PBX1 knockdown are presented in the bar graph. E2A-PBX1 knockdown tends to silence genes associated with cancer and cell cycle progression while contributing to the activation of genes involved in hematopoietic and lymphopoietic differentiation. Examples of operationally-defined E2A-PBX1 target genes involved in hematopoietic differentiation and cell cycle regulation are listed above.