Transfusion, Iron, and Leukemogenesis: Implications for Myelodysplastic Syndrome

Lap Shu Alan Chan

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Medical biophysics University of Toronto

Lap Shu Alan Chan

Doctor of Philosophy

Department of Medical Biophysics University of Toronto

2015

Abstract

Patients with myelodysplastic syndrome (MDS) require chronic red blood cell (RBC) transfusion due to anemia. Multiple RBC transfusions cause secondary iron overload and subsequent excessive generation of reactive oxygen species (ROS), which leads to mutations, cell death, organ failure, and inferior disease outcomes. The goal of this project is to assess the roles of anemia, secondary iron overload, and iron chelation in several aspects related to MDS including transfusion requirements, oxidative stress, and leukemogenesis.

Using the RBC transfusion history from MDS patients, we discovered that patients who have high initial transfusion intensity (ITI) had higher long-term transfusion requirement and reduced overall survival. Our results suggest that the underlying pathology of patients with high ITI may be more severe. We next analyzed early bone marrow hematopoietic cells from MDS patients and revealed that the ROS level of these cells is correlated with serum ferritin, a clinical marker of iron burden, in patients with high blast count. Conversely, iron chelation therapy (ICT) decreased serum ferritin and ROS level in these cells. Our findings established the relationship between iron and ROS in early hematopoietic cells. To elucidate the effect of iron overload and

ii iron chelation on leukemogenesis, we utilized a radiation-induced acute myeloid leukemia

(AML) animal model. Iron loading in irradiated B6D2F1 mice accelerated AML development.

However, there was a progressive decrease in AML risk for irradiated mice with increase in iron burden from 7.5 to 15 to 30 mg. In addition, ICT decreased AML incidence in the 7.5 mg iron- loaded irradiated mice, while AML onset was earlier for the 30 mg iron-loaded irradiated mice that received ICT. We concluded that iron is a promoter of mutation and leukemogenesis in vivo up to a peak iron dose, but further iron loading decreases AML risk by increasing cell death.

Taken together, these studies support the possibility that iron toxicity as a result of chronic RBC transfusion leads to poor outcome in MDS by impairing primitive hematopoietic cells via oxidative stress. ICT can partially mitigate the adverse effects of iron overload, and to maximize its benefit this intervention should be undertaken prior to the development of extreme iron overload.

iii

Acknowledgements

First and foremost, I would like to thank my mentor, Dr. Richard Wells, for granting me the opportunity to enter to realm of scientific research. His unwavering support, guidance and encouragement are crucial to the completion of this work. He is open-minded to ideas (whether they are bright or not) and always generous in providing valuable feedbacks and insights – some of these ideas eventually bore fruit and are included in this thesis. I owe Dr. Wells my deepest gratitude and I shall do my best to utilize his teachings to the fullest, in life and in science.

I would also like to thank my supervisory committee members, Drs. Suzanne Kamel-Reid and

Douglas Templeton for their advices and suggestions on my project. I am especially grateful to them and the rest of PhD examination committee members, Drs. Heather Leitch, Chen Wang,

Jonathan Rast, and Barbara Gibson for taking the time to edit the lengthy thesis and participate in the final exam.

This work would not have been possible without the support from the past and present members of the Wells’ laboratory: Patty, Christine, Michael, Joelle, Joy, Kevin, James, Roman etc, etc, etc. I would especially like to thank Dr. Lilly Gu for her assistant and technical know-how. Lilly is also a good friend and I truly appreciate what she has done for me.

Last but not least, a big thank-you from the bottom of my heart to my parents, Chiu Yat Chan and Wai Lin Pang, and my sisters, Sandy and Shirley Chan for their belief in me. I would not be able to accomplish anything without their love and support – this work is dedicated to them.

This work was funded by the Canadian Blood Service Graduate Fellowship Program, as well as research grants from the Leukemia & Lymphoma Society of Canada, Novartis Canada, and

Celgene Canada. iv

Table of Contents

Abstract ...... ii

Acknowledgements ...... iv

Table of Contents ...... v

List of Abbreviations ...... x

List of Tables ...... xiv

List of Figures ...... xv

Chapter 1 ...... 1

Introduction ...... 1

1.1 Normal hematopoiesis ...... 2

1.2 Myelodysplastic syndromes ...... 4

1.2.1 Historical perspectives and definition ...... 4

1.2.2 Etiology and epidemiology ...... 6

1.2.3 Initial evaluation and diagnosis ...... 7

1.2.4 Classification of MDS ...... 9

1.2.5 Clonal evolution and genetic changes in MDS ...... 13

1.2.5.1 Clonal evolution ...... 13

1.2.5.2 Gene mutations ...... 14

1.2.5.3 Cytogenetic changes ...... 19

1.2.6 Risk stratification ...... 23

1.2.7 Therapeutic options ...... 29

1.2.7.1 Therapy for lower-risk patients ...... 32

1.2.7.2 Therapy for higher-risk patients ...... 35

1.2.8 Supportive care for MDS ...... 38

1.2.8.1 Infection and antimicrobials ...... 38 v

1.2.8.2 Bleeding and platelet transfusion ...... 38

1.2.8.3 Anemia and RBC transfusion ...... 39

1.3 Iron ...... 41

1.3.1 The role of iron in biology ...... 41

1.3.2 Iron uptake from the environment ...... 44

1.3.3 Iron distribution in humans ...... 48

1.3.4 Iron processing in mammalian cells ...... 50

1.3.5 Regulation of iron balance ...... 53

1.3.6 Disruption of iron balance in humans ...... 58

1.3.6.1 Iron deficiency ...... 58

1.3.6.2 Primary iron overload ...... 59

1.3.6.3 Secondary iron overload ...... 61

1.4 Iron overload and myelodysplastic syndromes ...... 65

1.4.1 Non-transferrin bound iron and labile cell iron ...... 65

1.4.2 Pathophysiology of iron overload ...... 66

1.4.2.1 Generation of reactive oxygen species by iron ...... 66

1.4.2.2 Oxidative damage of macromolecules ...... 67

1.4.2.3 Signaling pathways and gene regulation ...... 68

1.4.2.4 Cell death and survival ...... 71

1.4.3 Iron overload in MDS ...... 73

1.4.4 Iron chelation in MDS ...... 74

1.5 Current questions, rationale and project objectives ...... 78

1.5.1 Current questions and rationale ...... 78

1.5.2 Project goal ...... 80

1.5.3 Project-specific questions ...... 80

1.5.4 Objective and hypotheses ...... 81

Chapter 2 ...... 82

Initial Transfusion Intensity Predicts Survival in MDS ...... 82

2.1 Introduction ...... 83

2.2 Methods ...... 84

2.2.1 Patients ...... 84

2.2.2 Data analysis ...... 85

2.3 Results ...... 85

2.3.1 Clinical characteristics ...... 85

2.3.2 Establishment of initial transfusion intensity ...... 87

2.3.3 ITI predicts long-term transfusion requirements ...... 89

2.3.4 ITI is a prognostic indicator of OS ...... 91

2.4 Discussion ...... 94

Chapter 3 ...... 96

Serum Ferritin Level is Associated with Intracellular ROS in the Hematopoietic Progenitors of MDS Patients ...... 96

3.1 Introduction ...... 97

3.2 Materials and methods ...... 98

3.2.1 Study population ...... 98

3.2.2 Sample collection and processing ...... 98

3.2.3 Intracellular ROS measurement ...... 98

3.2.4 Gating criteria for DCF measurement ...... 99

3.3 Results ...... 101

3.3.1 Experimental variation in iROS measurement ...... 101

3.3.2 Measurement of iROS of BM lymphocytes and CD34+ cells ...... 104

vii

3.3.3 Iron overload and iROS in MDS ...... 108

3.4 Discussion ...... 111

Chapter 4 ...... 114

Secondary iron overload promotes radiation-induced acute myeloid leukemia in B6D2F1 mice ...... 114

4.1 Introduction ...... 115

4.1.1 Animal models of iron overload and ICT ...... 115

4.1.2 Animal models of radiation-induced AML ...... 118

4.1.3 Hypothesis ...... 121

4.2 Methods ...... 121

4.2.1 Animals ...... 121

4.2.2 Treatments ...... 122

4.2.3 Monitoring and analysis ...... 123

4.2.4 Flow cytometry ...... 123

4.2.5 DNA, RNA and protein isolation ...... 125

4.2.6 Oxidative DNA damage (AP sites) ...... 125

4.2.7 Quantitative RT-PCR ...... 126

4.2.8 Figures and data analysis ...... 126

4.3 Results ...... 126

4.3.1 High dose/short term iron loading ...... 126

4.3.2 Low dose/long term iron loading ...... 129

4.3.3 Effects of iron and ICT on radiation-induced AML ...... 134

4.3.4 The role of iron loading in leukemogenesis of RI-AML ...... 148

4.3.5 Gene expression in the BMCs of mice that develop AML ...... 148

4.3.6 The early effects of iron on mice at 3 months after irradiation ...... 160

viii

4.3.7 The effects of iron on mice at 5 months after irradiation ...... 162

4.3.8 The effects of iron on mice at 7 months after irradiation ...... 169

4.4 Discussion ...... 175

Chapter 5 ...... 183

Conclusions and future directions ...... 183

5.1 Summary and conclusions ...... 184

5.2 Future directions ...... 187

5.2.1 Anemia, transfusion, and the outcome of MDS ...... 187

5.2.2 Iron-induced dysregulation in BMCs from MDS patients ...... 188

5.2.3 The effects of iron loading on leukemogenesis ...... 189

5.2.4 The effects of iron loading on RI-AML in different mouse strains ...... 190

5.2.5 Secondary iron overload in MDS animal models ...... 190

References ...... 192

List of Abbreviations

5q‐ Deletion within the long arm of chromosome 5 5q‐ syndrome MDS associated with isolated 5q‐ 7‐ Monosomy 7 7q‐ Deletion within the long arm of chromosome 7 8+ Trisomy 8 AAMAC Aplastic Anemia & Myelodysplasia Association of Canada AML Acute myeloid leukemia AMPK AMP‐activated protein kinase ANC Absolute neutrophil count AP Apurinic/apyrimidinic ARNT Aryl hydrocarbon nuclear translocator ASR Age‐standardized rate ATG Anti‐thymocyte globin BM Bone marrow BMCs Bone marrow cells BSC Best supportive care CBC Complete blood count CFU‐E Erythroid colony‐forming units CI Confidence interval ciiBMCs Iron chelated iron‐loaded irradiated BMCs COD Cause of death CSF‐1 Colony‐stimulating factor‐1 CV Coefficient of variation DCF 2′,7′‐dichlorofluorescein DCFH 2',7'‐dichlorodihydrofluorescein DCFH‐DA 2',7'‐dichlorodihydrofluorescein diacetate DcytB Duodenal cytochrome B DMSO Dimethyl sulfoxide DMT1 Divalent metal ion transporter 1 EPIC Evaluation of Patients Iron Chelation with Exjade EPO Erythropoietin ESA Erythropoiesis‐stimulating agent FAB French‐American‐British Fe/S Iron‐sulfur FISH Fluorescence in situ hybridization FLVCR Feline leukemia virus subgroup C receptor FOXO Forkhead box O transcription factor FPN Ferroportin x

FSC Forward scatter G‐CSF Granulocyte colony‐stimulating factor GPX Glutathione peroxidase GSH Glutathione GVHD Graft‐versus‐host‐disease H2K27 Histone H3 lysine 27 HCP1 Heme carrier protein 1 HCV Hepatitis C virus Heph Hephaestin Hfe Human hemochromatosis HH Hfe‐related hemochromatosis HI Hematologic improvement HI‐E Erythroid improvement HIF Hypoxia inducible factor HI‐N Neutrophil improvement HI‐P Platelet improvement HLA Human leukocyte antigen HMA Hypomethylating agent HO‐1 Heme oxygenase 1 HR Hazard ratio HSC Hematopoietic stem cell HSCT Hematopoietic stem cell transplantation HSF Heat‐shock factor HSP Heat‐shock protein iBMCs Irradiated BMCs ICT Iron chelation therapy ICUS Idiopathic cytopenia of unknown significance IDUS Idiopathic dysplasia of unknown significance iiBMCs Iron‐loaded irradiated BMCs IPSS International Prognostic Scoring System IPSS‐R International Prognostic Scoring System ‐ Revised IRE Iron response element iROS Intracellular reactive oxygen species IRP Iron response protein IST Immunosuppressive therapy ITI Initial transfusion intensity IWG International Working Group IWG‐PM International Working Group for Prognosis in MDS JH Juvenile hemochromatosis JNK2 Jun N‐terminal kinase 2

Keap1 Kelch like‐ECH‐associated protein 1 KM Kaplan‐Meier LDH Lactate dehydrogenase LFS Leukemia free survival LFS AML‐free survival LIC Liver iron concentration Lin Lineage LIP Labile iron pool LLS Leukemia & Lymphoma Society LPI Labile plasma iron MCH Mean corpuscular hemoglobin MCV Mean corpuscular volume MDS Myelodysplastic syndrome MDS‐U Myelodysplastic syndrome, unclassified MFI Mean fluorescence intensity MMP Matrix‐degrading metalloprotease MNCs Mononuclear cells MPD Myeloproliferative disorders MS Median survival MSC Mesenchymal stem cell NCCN National Comprehensive Cancer Network NF‐κB Nuclear factor‐κB Nrf2 Nuclear factor erythroid 2‐related factor nROS Normalized intracellular reactive oxygen species NTBI Nontransferrin bound iron OS Overall Survival OT Occasionally transfused PB Peripheral blood PBS Phosphate‐buffered‐saline PR Partial remission PRBC Packed red blood cells PRC Polycomb repressive complex QoL Quality of life RA Refractory anemia RAEB‐1 Refractory anemia with excess blasts‐1 RAEB‐2 Refractory anemia with excess blasts‐2 RAEB‐T Refractory anemia with excess blasts in transformation RARS Refractory anemia with ringed sideroblasts RBC Red blood cell RCC Refractory cytopenia of childhood

xii

RCMD Refractory cytopenia with multilineage dysplasia RCUD Refractory cytopenia with unilineage dysplasia RI‐AML Radiation‐induced AML RIC Reduced intensity conditioning RIP Receptor interacting protein kinase RN Refractory neutropenia ROS Reactive oxygen species RT Refractory thrombocytopenia SOD Superoxide dismutase SQUID Superconducting quantum interference device SSC Side scatter TD Transfusion dependency TFR Transferrin receptor TGF‐β Transforming growth factor β TI Transfusion intensity t‐MDS Therapy‐related MDS TNF‐α Tumor necrosis factor‐α TRIM Transfusion‐related immunomodulation US United States of America UTR Untranslated region WHO World Health Organization WPSS WHO‐based prognostic scoring system

xiii

List of Tables

Table 1.2.4a. Morphologic characteristics of dysplasia in MDS Table 1.2.4b. Criteria for 2008 WHO classification of MDS Table 1.2.5.2. Common germline and somatic genetic mutations in MDS Table 1.2.5.3. Common recurring chromosomal abnormalities in MDS Table 1.2.6a. International prognostic scoring system (IPSS) for MDS Table 1.2.6b. Revised international prognostic scoring system (IPSS‐R) Table 1.2.6c. WHO classification‐based prognostic scoring system (WPSS) Table 1.2.7a. Definition of lower and higher risk patients Table 1.2.7b. Recommended treatment options for MDS Table 1.2.7c. Measurement of response in MDS Table 1.3.1. Iron containing proteins Table 2.3.1. Clinical characteristics of patients with transfusion history related to MDS Table 2.3.4. Multivariate analysis of factors affecting survival Table 3.3.1a. iROS level of PB lymphocytes from four healthy volunteers Table 3.3.1b. iROS level of HL60 and NB4 cell lines Table 3.3.2. Patients characteristics Table 4.3.1. Effects of high dose, short term iron‐loading in mice Table 4.3.2. Effects of low dose, long term iron‐loading in mice Table 4.3.3a. Effects of iron dextran and deferasirox on irradiated B6D2F1 mice Table 4.3.3b. Cause of death for control, iron, iron/ICT irradiated mice Table 4.3.5a. Genes with altered expression in all analyses Table 4.3.5b. Gene expression levels of total BMCs from 5 months irradiated mice and AML mice that are significantly different Table 4.3.5c. Progressive up or downregulation of total BMCs gene expression from 5 months irradiated mice, to 7 months irradiated mice, to AML mice Table 4.3.6. Early analysis of control and 5 mg iron‐loaded irradiated mice at 3 months post‐irradiation Table 4.3.7a. Analysis of BMCs from control, iron‐loaded, iron/ICT mice at 5 months post‐irradiation Table 4.3.7b. Altered gene expression in total BMCs from control, iron, iron/ICT mice at 5 months post‐irradiation Table 4.3.8a. Analysis of BMCs from control mice at 5 and 7 months post‐irradiation Table 4.3.8b. Altered gene expression in total BMCs from control mice at 5 and 7 months post‐irradiation Table 4.3.8c. Altered gene expression in total BMCs from control, iron, iron/ICT mice at 7 months post‐irradiation

xiv

List of Figures

Figure 1.1. Hierarchical model of hematopoiesis Figure 1.2.7.1. Recommended therapeutic approach for lower‐risk MDS patients Figure 1.2.7.2. Recommended therapeutic approach for higher‐risk MDS patients Figure 1.3.2. Established and putative iron absorption pathways in the intestinal enterocyte Figure 1.3.3. Normal iron intake, loss, and distribution in adult humans Figure 1.3.4. Iron processing in mammalian cells Figure 1.3.5a. Translation regulation by Iron response/regulatory proteins Figure 1.3.5b. Regulation of iron balance by hepcidin Figure 2.3.2a. Initial transfusion intensity (ITI) distribution of patients with transfusion history related to MDS Figure 2.3.2b. Pre‐transfusion hemoglobin (Hb) levels in ITI groups 1 and 2 prior to the first and second transfusion. Figure 2.3.3. Transfusion characteristics of the two ITI groups Figure 2.3.4. Impact of ITI on overall survival Figure 3.2.4. Cell gating criteria from the mononuclear cells (MNCs) Figure 3.3.1. The PB lymphocyte iROS level of a single healthy volunteer Figure 3.3.2a. Normalized iROS level (nROS) Figure 3.3.2b. Illustration of wide and narrow iROS distribution of CD34+ cells Figure 3.3.3a. nROS of CD34+ cells vs. serum ferritin level of low blast count patients Figure 3.3.3b. nROS of CD34+ cells vs. serum ferritin level of high blast count patients Figure 3.3.3c. Effects of iron chelation therapy on serum ferritin level and nROS level of CD34+ cells Figure 4.3.1. Representative iron staining of tissue sections from control and high dose, short term iron‐loaded mice Figure 4.3.2a. Representative iron staining of liver and bone marrow tissue sections from control and low dose, long term iron‐loaded mice Figure 4.3.2b. iROS level of Lin‐CD45+ BMCs Figure 4.3.3a. Workflow for iron loading and ICT on irradiated mice Figure 4.3.3b. Representative weight change over time for irradiated mice Figure 4.3.3c. Representative May‐Grünwald‐Giemsa staining of peripheral blood smear Figure 4.3.3d. Representative flow cytometry analysis of myeloid vs. lymphoid population in PB Figure 4.3.3e. Representative hematoxylin and eosin staining of tissue sections from liver, spleen and bone marrow Figure 4.3.3f. Representative May‐Grünwald‐Giemsa staining of BM cytospin Figure 4.3.3g. Representative flow cytometry analysis of lineage vs. CD45 BMCs xv

Figure 4.3.3h. Representative flow cytometry analysis of Gr‐1 vs. CD11b BMCs Figure 4.3.3i. Kaplan‐Meier leukemia‐free survival curves of irradiated mice treated with sham injection, iron dextran injection, and iron/deferasirox treatments Figure 4.3.3j. KM LFS curves of irradiated mice treated with sham injection or iron dextran injections Figure 4.3.3k. KM LFS curves of irradiated mice treated with sham injection, 7.5 mg iron dextran injection, and 7.5 mg iron/deferasirox Figure 4.3.3l. KM LFS curves of irradiated mice treated with sham injection, 30 mg iron dextran injection, and 30 mg iron/deferasirox Figure 4.3.5a. Clustergram of 86 genes that were up or downregulated in the total BMCs of AML mice and all groups used for early analyses Figure 4.3.5b. Progressive downregulation of Pten in total BMCs from 7 months irradiated mice, to 7 months iron‐loaded mice, to AML mice Figure 4.3.6. Representative iron staining of tissue sections from control and iron‐ loaded irradiated mice for early analysis at 3 months post‐irradiation Figure 4.4a. Proposed pro‐death and pro‐mutagenic influences of iron loading and ICT on the rate of AML Figure 4.4b. RI‐AML progression model and the influence of iron or iron/ICT to leukemogenesis

xvi

Chapter 1

Introduction

1.1 Normal hematopoiesis

Hematopoiesis follows a hierarchical model in which hematopoietic stem cells (HSCs) differentiate into various lineage specific progenitors and subsequently mature blood cells (Figure 1.1) (Munker et al., 2000b; Wang and Wagers, 2011). HSCs are also capable of self- renewal, thereby maintaining the number of HSCs for sustained hematopoiesis through the lifespan of an individual organism (Munker et al., 2000b). In addition, a portion of HSCs remain dormant at the G0 phase of the cell cycle, presumably to extend their longevity and prevent stem cell exhaustion. Recent data suggest that the cell-cycle status of HSCs is associated with their location in different niches at the bone marrow (BM) endosteum: quiescent HSCs reside near thin-walled arterioles, while cycling HSCs are distributed randomly near the sinusoids (Kunisaki et al., 2013). The activities of HSCs are directly or indirectly managed by various other cell types, which include osteoblasts, mesenchymal cells, endothelial cells, and Schwann cells (Morrison and Scadden, 2014).

In postnatal mammals, hematopoiesis occurs mainly in the BM, with the exception of T-cells maturation in the thymus. During early life, hematopoiesis is active in the BM of virtually all bones. During adult life, active hematopoiesis is restricted to the BM (active BM) of certain bones such as the sternum, ribs, pelvic bones and the proximal ends of the femur. A number of non-hematopoietic cells are also present in the BM, such as osteoblasts and adipocytes (Pontikoglou et al., 2011). These cells are derived from the mesenchymal stem cells (MSCs) and contribute to the maintenance of the hematopoietic microenvironment in the bone marrow. The BM contains a substantial amount of adipose tissue, ranging from 40% fat content in active BM to 80% in inactive BM (Loevner et al., 2002). BM fat content increases with age and may be associated with declining blood flow to the BM (Tuljapurkar et al., 2011). The differentiation of blood cells depends on the complex interaction between multiple genes, cytokines, and growth factors. Upon maturation, the developed blood cells enter the circulation and contribute to various vital physiological functions: oxygen transportation by erythrocytes (which are produced by the process of erythropoiesis), innate immune response by granulocytes (which are produced by the process of myelopoiesis), adaptive immune response by lymphocytes (which are produced by the process of lymphopoiesis), and hemostasis by platelets (which are produced by megakaryocytes via the process of thrombopoiesis). 2

1.2 Myelodysplastic syndromes

1.2.1 Historical perspectives and definition The contemporary understanding of myelodysplastic syndrome (MDS) is the fruit of decades of technological advancements and medical discoveries (Steensma, 2012). Assessment of blood cells under the microscope is made possible by the work of the German scientist Paul Ehrlich, who developed aniline dyes to stain and define different subtypes of hematopoietic cells (Ehrlich, 1879). Abnormal BM morphology is one of the defining features of MDS as bone marrow conditions. This feature became relevant in the clinical setting after the development of the Arinkin’s technique by the Russian scientist Mikhael Arinkin (Arinkin, 1929), which enabled safe and in vivo bone marrow sampling from the sternum, thereby allowing physicians to assess the bone marrow condition of live patients.

MDS as conceptualized today has historically been intertwined with other hematologic conditions. For decades since the 1920s, “Di Guglielmo’s syndrome”, named after the Italian physician Giovanni Di Guglielmo, was often used to describe an assortment of BM disorders associated with anemia and abnormal blood cells, a number of these cases can be classified as MDS by contemporary diagnostic standards (Dameshek and Baldini, 1958; Guglielmo, 1926; Steensma, 2012). Further discoveries during the 1920s and 1930s have identified the causes for many types of anemia and allowed them to become “standalone” diseases, such as pernicious anemia due to vitamin B12 deficiency and beta thalassemia major (Cooley’s anemia) due to mutation of the β-globin gene. The term “refractory anemia” was coined to describe other acquired anemia of unidentifiable origin or unresponsive to hematinics at the time (Rhoads and Barker, 1938). The causes of various types of anemia were eventually identified, and the definition of refractory anemia was narrowed and altered over time to describe a particular form of MDS.

In the 1940s and 1950s, a number of reports emerged to connect BM disorders/anemia with subsequent risk of leukemia (Block et al., 1953; Chevallier, 1942-1943; Hamilton-Paterson, 1949). The predisposition to leukemia greatly increased the awareness of the disorders among physicians and patients, and the term “pre-leukemia” was commonly used until the 1970s to describe the disorders. Nonetheless, it was later established that around two-third of “pre-

4 leukemic” (MDS in retrospect) patients do not develop leukemia, and the sole emphasis on the leukemia aspect of the disorders may be misleading.

In 1969, the American hematologist William Dameshek, who defined myeloproliferative disorders (MPD) in 1951, suggested that the definition of MPD could be expanded to include pre-leukemia with erythroid and myeloid hyperplasia (Dameshek, 1951, 1969). However, Dr. Harriet Gilbert in New York noted the following year that the subset of pre-leukemia cases is marked by the lack of maturation and therefore distinct from the classical definition of MPD (Gilbert, 1970). Gilbert subsequently proposed the term “myelodysplastic disease” for these non- MPD pre-leukemia cases, which was eventually adopted by the French-American-British (FAB) Co-operative Group as MDS. The FAB group was founded by a panel of seven leukemia experts in an effort to refine nomenclature and classifications of hematologic diseases. In 1982, the group proposed the classification for MDS, which was widely adopted for the next two decades (Bennett et al., 1982). Moreover, efforts to establish prognosis in MDS ballooned in the 1980s, resulting in the creation of the International Prognostic Scoring System (IPSS) in 1997 (Greenberg et al., 1997). The continual effort to refine the standard of diagnosis, classification, prognosis, and treatment of MDS is now the responsibility of the World Health Organization (WHO) and other national organizations such as the National Comprehensive Cancer Network (NCCN) and the MDS Foundation in the US, the European LeukemiaNet in the European Union, and the Canadian Consortium on MDS (CCMDS) in Canada.

MDS is defined today as a group of hematological disorders derived from abnormal HSCs or hematopoietic progenitors via clonal evolution (Brunning et al., 2008). MDS is a type of progressive bone marrow failure characterized by cytopenias, dysplasia, increased apoptosis, and ineffective hematopoiesis in one or more of the myeloid lineages. Clinical manifestations of MDS are often non-specific and correspond to the type and severity of cytopenias (Foran and Shammo, 2012; Greenberg et al., 2013): anemia is associated with fatigue and shortness of breath; neutropenia increases the susceptibility to infection; thromobocytopenia increases the susceptibility to unexplained bleeding and bruising. Increased risk of transformation to acute myeloid leukemia (AML) is also noted, especially in patients with increased myeloblasts. Nevertheless, the pathological progression of MDS can vary substantially based on the subtypes:

5 from patients who are almost asymptomatic with near-normal lifespan, to others who acquire AML shortly after their condition is diagnosed.

1.2.2 Etiology and epidemiology Therapy-related (secondary) MDS can occur as a late complication of cytotoxic chemotherapy and/or radiation therapy, which account for 10-20% of all cases of MDS (Brunning et al., 2008). A small number of non-therapy-related (primary or de novo) MDS may be associated with inherited factors such as family history of hematopoietic neoplasms, Fanconi anemia, dyskeratosis congenita, Shwachmann-Diamond syndrome, Diamond-Blackfan syndrome, and heritable RUNX1 or GATA2 mutations. Patients with aplastic anemia are also predisposed to acquire MDS as a late complication (Afable et al., 2011). The majority of cases of primary MDS have no confirmed genetic or environmental origins. Certain occupational groups or habits may increase the risk of MDS possibly due to the exposure to substances such as benzene, cigarettes, and agricultural chemicals or solvents (Strom et al., 2005).

Based on statistics reported by the Leukemia & Lymphoma Society (LLS), there were approximately 70,615 new cases of MDS in the United States of America (US) from 2006 to 2010, averaging an estimated 14,123 cases per year (LLS, 2014). The overall incidence rate for MDS is 4.8 cases per 100,000 population per year. The incidence rates vary by gender and race, with 6.5 per 100,000 population for males versus 3.7 per 100,000 population for females, and 4.9 per 100,000 population for White versus 3.7 per 100,000 population for Asian and Pacific Islander. White males have the highest incidence rates at 6.7 per 100,000 population, while Asian and Pacific Islander females have the lowest incidence rate at 2.7 per 100,000 population. MDS is infrequent among children and young adults before the age of 40, with an incidence rate of 0.2 cases per 100,000 population. Conversely, MDS typically occurs in the elderly population with a median age between 71 and 76 years at diagnosis (Barzi and Sekeres, 2010; Ma et al., 2007). The incidence of MDS increases dramatically with age, from 2.6 per 100,000 population for ages 50-59, to 29.6 per 100,000 population for ages 70-79, and 55.8 per 100,000 population for ages 80+. The incidence of MDS has increased over time, and the upward trend is expected to continue in part due to an aging population, improving reporting of MDS to cancer registries, as well as diagnosis of patients with unexplained anemia who may previously have been missed (Goldberg et al., 2010; Guralnik et al., 2004; Ma, 2012). The MDS incidence rate in Canada on a 6 per population basis is comparable to that of the US, with approximately 1,400 new cases in 2013 (LLSC, 2013).

The age-standardized rate (ASR) for mortality of MDS as the underlying cause is approximately 1.7 per 100,000 population in the US between the years 1999 to 2009, with 6007 deaths in 2009 (Polednak, 2013). The ASR is higher at approximately 2.9 per 100,000 population for MDS patients who pass away due to multiple causes, with 10729 deaths in 2009. Therefore, about 40% of the mortalities are not directly related to MDS. Common non-MDS underlying causes of death (CODs) are leukemia and cardiovascular diseases, both with similar ASRs at approximately 0.4 per 100,000 population. The mortality rate is also higher among older patients. The ASR in 2009 for patients over 65 years death from multiple causes is 23.9, compared to 0.47 for patients between age 25 and 64 years. Older MDS patients are also more likely to die from cardiovascular diseases but less likely to die from leukemia. Dayyani et al. has conducted a smaller survey that focuses on lower-risk MDS patients who have longer life expectancy and lower risk of leukemia progression (Dayyani et al., 2010). MDS-related infection, especially pneumonia and sepsis, is the most frequent COD, which accounts for 38% of all mortalities. Other common CODs among lower-risk MDS patients include AML (15%), MDS-related hemorrhage (13%), and cardiovascular events (7%). An increase in median survival (MS) is also observed: from 14.5 months for the 1980s to 24.7 months for the 2000s, plausibly due to the introduction of disease modifying treatments and improvements in supportive care.

1.2.3 Initial evaluation and diagnosis The clinical status of patients with MDS is established by several types of evaluations, including the patient’s medical history, clinical manifestations, and laboratory assessments. The goal of this integrated approach is to exclude other possible non-MDS explanations to the patient’s symptoms, as well as to establish the diagnosis, prognosis, and treatment options of confirmed MDS.

The medical history is essential to determine if the cytopenias may be related to MDS. Predisposition to MDS or AML is seen with conditions such as Fanconi anemia and dyskeratosis congenita or inherited genetic abnormalities such as RUNX1 and GATA2 mutations (Brunning et al., 2008; Liew and Owen, 2011). Therapy-related MDS may occur as a late complication of

7 chemotherapy or radiation therapy (Zhang and Wang, 2014). Conversely, other factors, such as nutritional deficiencies, infection, alcohol abuse, and hypothyroidism, may cause prolonged anemia or cytopenias but they are not MDS. Therefore, a differential diagnosis is crucial to exclude other possible non-MDS causes.

Common laboratory tests for the screening of MDS include complete blood count (CBC), reticulocyte counts and blood chemistry tests (Brunning et al., 2008; Greenberg et al., 2013). Examination of the peripheral blood (PB) smear is also essential to determine dysplasia of circulating hematopoietic cells or the existence of blasts. Suspicion of MDS often arises as a result of unexplained cytopenias over a six-month period, but if an MDS-specific karyotype or bilineage dysplasia is present, the pre-requisite for the duration of persistent cytopenias can be shortened to two months. Other abnormal blood count results such as thrombocytosis may also indicate possible MDS in the absence of cytopenia. Furthermore, blood count stability is useful to determine disease progression.

Patients with suspected MDS require BM examination, in which both BM trephine biopsy and aspirate specimens are collected (Greenberg et al., 2013; ICSH et al., 2008). The core biopsy is necessary to assess the histology, immunohistochemistry and architecture of the hematopoietic compartment, as well as the localization of blasts and immature cells. Although the marrow cellularity is often normal or increased in MDS, a hypocellular BM can be seen in approximately 10% of the cases (Brunning et al., 2008; Maschek et al., 1993; Orazi et al., 1997). Hypoplastic MDS may be confused with aplastic anemia during diagnosis. Hypocellelar BM occurs more frequently in women and may be associated with a greater probability of response to Immunosuppressive therapy (IST), but it has no known prognostic value in MDS.

The aspirate smear is useful for the characterization of dysplasia and blast count. In addition to the standard May-Grünwald-Giemsa staining, Prussian blue and reticulin staining of the specimens are needed to evaluate the presence of ringed sideroblasts and fibrosis, respectively. Myelofibrosis is observed in 10-50% of MDS cases, which is an adverse prognostic factor generally associated with high-grade MDS, shorter life expectancy and faster leukemic transformation (Brunning et al., 2008; Marisavljevic et al., 2004).

Since cytogenetic changes can occur in almost half of all primary MDS cases (Haase, 2008), suspected cases are now routinely submitted for conventional karyotyping and sometimes also for fluorescence in situ hybridization (FISH) analysis. Many chromosome aberrations are not MDS-specific, and other clinical evidence is generally required for the confirmation of MDS. Finally, immunophenotyping by flow cytometry is a useful adjunct for MDS diagnosis, especially in difficult cases (Brunning et al., 2008; Greenberg et al., 2013).

1.2.4 Classification of MDS MDS is a heterogeneous group of disorders consisting of many subtypes with unique features. The classification of MDS into different subtypes is mostly based on morphological features, including the characteristics of the dysplasia, the degree of cytopenia, and the number of blasts in PB and BM. Other criteria are also required for specific MDS subtypes such as chromosome 5q deletion (5q-).

Cellular dysplasia in MDS is indicative of abnormal hematopoiesis. Dysplasia is defined as the presence of ≥10% dysplastic erythroid precursors, granulocytes, or megakaryocytes in PB or BM smear (Brunning et al., 2008). The number of dysplastic lineages is also a significant consideration for diagnosis and prognosis. The characteristics of dysplasia in different lineages are summarized in Table 1.2.4a (Brunning et al., 2008). Dyserythropoiesis is characterized by nuclear and cytoplasmic alterations of immature erythroid cells. Internuclear bridging, karyorrhexis, and megaloblastoid changes are often observed in the nucleus of erythrocytes. The presence of ≥15% ringed sideroblasts in erythroblasts, as a result of iron accumulation in mitochondria around the nucleus, is an important cytoplasmic feature with different pathophysiology and clinical implication (Malcovati and Cazzola, 2013). Dysgranulopoiesis is characterized by small size, cytoplasmic hypogranularity, and nuclear hypolobation or hypersegmentation. Dysmegakaryocytopoiesis is characterized by nuclear hypolobation, micromegakaryocytes, and multinucleation. The presence of Auer rods is considered as evidence of high-grade MDS and is associated with adverse prognosis even in patients with low blast counts. Some dysplastic characteristics are associated with cytogenetic abnormalities, such as increased number of hypolobated micromegakaryocytes for isolated 5q-, and hypolobated neutrophils for 17p-.

The diagnosis of MDS requires constant and substantial cytopenias of at least one lineage for six months, unless in the presence of other manifestations such as specific karyotypes. Cytopenias of multiple lineages is not uncommon and the number of lineages involved is correlated with the severity of MDS. The recommended cutoff levels for hemoglobin, absolute neutrophil count (ANC), and platelets by peripheral blood count are <100 g/L, <1.8x109 /L, and <100x109 /L, respectively (Greenberg et al., 1997). Excess blasts at 5-19% from BM smear or histology is an important consideration but not necessarily required for all MDS subtypes. The presence of excess blasts is associated with high-grade disease and higher-risk of AML progression. Cases with blast count between 5-19% in PB and 20-30% in BM are classified as refractory anemia with excess blasts in transformation (RAEB-T) in the FAB classification and considered as AML with MDS-related changes in the WHO classification (Greenberg et al., 2013). Patients who do not fulfill the minimal MDS criteria for dysplasia and cytopenias are classified as either idiopathic cytopenia of unknown significance (ICUS) for those with constant cytopenias but no dysplasia and no abnormal karyotype, or idiopathic dysplasia of unknown significance (IDUS) for those with marked dysplasia but no substantial cytopenia (Valent and Horny, 2009). Patients with ICUS and IDUS are at risk to develop actual MDS and should be carefully monitored for disease progression.

Although the FAB classification was widely adopted for the diagnosis of MDS since the 1980s (Bennett et al., 1982), it is now replaced by the WHO classification (Brunning et al., 2008). The classification is essential for the diagnosis, treatment, and prognostication of MDS. The WHO classification is periodically modified to incorporate the latest understanding of MDS. The current version was published in 2008, and continual updates are expected in the future. The 2008 WHO classification consists of seven MDS subtypes (Brunning et al., 2008): refractory cytopenia with unilineage dysplasia (RCUD), refractory anemia with ringed sideroblasts (RARS), refractory cytopenia with multilineage dysplasia (RCMD), refractory anemia with excess blasts-1 (RAEB-1), refractory anemia with excess blasts-2 (RAEB-2), myelodysplastic syndrome, unclassified (MDS-U), and MDS associated with isolated 5q- (5q- syndrome). The criteria for these subtypes are listed in Table 1.2.4b.

RAEB-1 and RAEB-2 are generally considered as high-grade disease with higher risk of leukemia progression and shorter life expectancy, while the rest of the subtypes belong to the low-grade category. RCUD is an expansion of the refractory anemia (RA) subtype from the FAB classification, which now encompasses RA, refractory neutropenia (RN) and refractory thrombocytopenia (RT). The diagnosis of 5q- syndrome requires an isolated 5q- chromosome aberration. MDS-U also requires cytogenetic aberrations but dysplasia is not necessary. Cases with myeloproliferative features, such as chronic myelomonocytic leukemia (CMML) with >1x109 /L monocytes, belong to a separate disease entity called myelodysplastic/myeloproliferative neoplasms (MDS/MPN) instead of pure MDS as it was defined by the FAB system. The classification for MDS subtypes is generally applicable to both primary and secondary MDS, although secondary MDS is always considered to carry a poor prognosis. For childhood MDS, diagnostic criteria for RAEB-1 and RAEB-2 are identical to adults. However, many cases of low-grade MDS in children are provisionally classified into a separate subtype called refractory cytopenia of childhood (RCC), which address unique features such as the higher frequency of neutropenia/thrombocytopenia and hypocellular BM. Furthermore, MDS associated with Down syndrome is considered as a unique biologic entity that is distinct from childhood MDS.

Table 1.2.4a. Morphologic characteristics of dysplasia in MDS (WHO Classication of Tumours of Hematopoietic and Lymphoid Tissues, Swerdlow et al., 2008).

Dysplasia Morphologiccharacteristics Dyserythropoiesis (Nuclear) Nuclear budding, internuclear bridging, karyorrhexis, multinuclearity, nuclear hyperlobation, megaloblasticchanges; (cytoplasmic) ring sideroblasts, vacuolization, periodic acid‐ Schiff positivity.

Dysgranulopoiesis Small or unusually large size, nuclearhypolobation, irregular hypersegmentation, decreased granules, agranularity, pseudo Chediak‐Higashi granules, Auer rods

Dysmegakaryocytopoiesis Micromegakaryocytes, nuclear hypolobation, multinucleation

Table 1.2.4b. Criteria for 2008 WHO classification of MDS (WHO Classification of Tumours of Hematopoietic and Lymphoid Tissues, Swerdlow et al., 2008).

Subtype Blood Bone marrow

Refractory cytopenia Unicytopenia or Unilineage dysplasia (≥10% with unilineage dysplasia bicytopenia; <1% of cells in one myeloid (RCUD)* blasts lineage); <5% blasts; <15% ring sideroblasts

Refractory anemia with Anemia; no blasts ≥15% ring sideroblasts; ringed sideroblasts erythroid dysplasia only; (RARS) <5% blasts

Refractory cytopenia Cytopenia(s); <1% Dysplasia in ≥10% of cells with multilineage blasts; no Auer rods; in ≥ two myeloid lineages; dysplasia (RCMD) <1x109 /L monocytes <5% blasts; No Auer rods

Refractory anemia with Cytopenia(s); <5% Unilineage or multilineage excess blasts‐1 (RAEB‐1) blasts, no Auer rods; dysplasia; 5‐9% blasts; no <1x109 /L monocytes Auer rods

Refractory anemia with Cytopenia(s); 5‐19% Unilineage or multilineage excess blasts‐2 (RAEB‐2) blasts, ±Auer rods; dysplasia; 10‐19% blasts; <1x109 /L monocytes ±Auer rods

Myelodysplastic Cytopenias; ≤1% blasts Dysplasia in ≤1 myeloid syndrome, unclassified lineage; MDS cytogenetic (MDS‐U) abnormality; <5% blasts

MDS associated with Anemia; platelets Isolated 5q‐;normal to isolated 5q‐ normal or increased; increased megakaryocytes (5q‐ syndrome) ≤1% blasts with hypolobated nuclei, no Auer rods * RCUD consists of refractory anemia (RA), refractory neutropenia (RN), and refractory thrombocytopenia (RT)

1.2.5 Clonal evolution and genetic changes in MDS MDS arises from clonal evolution and expansion of abnormal HSCs driven by the accumulation of genetic abnormalities (Brunning et al., 2008; Cazzola et al., 2013a; Haase, 2008; Munker et al., 2000b). Genetic abnormalities in MDS are diverse and can involve gene mutations and cytogenetic aberrations. Although these changes in general are not sufficient for the diagnosis of MDS without additional clinical evidence, they are important for the understanding of the pathophysiology of MDS, and have important implications in the prognostication and therapeutic options for the disorder.

1.2.5.1 Clonal evolution The pathogenesis of MDS begins with the acquisition of an initial driver mutation in a HSC (Cazzola et al., 2013a). The driver mutation gives a selective advantage to the mutated HSC relative to normal HSC. In primary MDS, somatic mutations of genes such as TET2 or DNMT3a can induce epigenetic dysregulation, resulting in the altered expression of multiple genes (Cazzola et al., 2013a; Issa, 2013; Khan et al., 2013). Alternatively, age-related epigenetic drift may induce comparable epigenetic changes without an actual somatic mutation (Chambers et al., 2007; Issa, 2013; Maegawa et al., 2014). For familial MDS, heritable mutations of genes such as RUNX1 or GATA2 are well documented, but second hit mutations are likely required for the development of overt MDS (Gao et al., 2014; Hahn et al., 2011; Liew and Owen, 2011). The genetic or epigenetic aberrations provide the mutated HSC with a competitive advantage over normal HSCs, thus allowing the mutated HSC to expand into a MDS clone and displace local normal HSCs. The MDS clone subsequently migrates to other bone marrow sites and eventually reaches clonal dominance in the body. Overt symptoms may occur as normal hematopoiesis is impaired due to the combination of displacement of normal HSCs and impairment of normal stem cell function by changes imposed by the MDS clone on the hematopoietic milieu. Clinical presentation and disease severity depend on the pathological phenotype of the driver mutation and may be affected by the presence of cooperating mutations (Cazzola et al., 2013a). Some cells from the original MDS clone may acquire additional driver mutations over time and transform into subclones (Bejar, 2014; Cazzola et al., 2013a). The additional mutations may promote disease progression via further impairment of differentiation capacity, promotion of proliferation, improvement of DNA repair mechanism, and resistance to apoptosis. In general, the presence of

13 additional driver mutations is associated with increased disease severity, accelerated leukemia progression and adverse prognosis (Bejar, 2014). Nonetheless, there are instances where the co- occurrence of mutants may not result in a worse clinical outcome, as it is possible for a mutant to mitigate the negative effect of another more detrimental mutant (Bejar et al., 2012; Papaemmanuil et al., 2013). The initial driver mutation may also have a “predestination” role in the progression and outcome of the disease by shaping the future trajectories of disease evolution via restriction of the repertoire of the additional mutations (Papaemmanuil et al., 2013).

1.2.5.2 Gene mutations Familial cases of MDS are rare in comparison to other non-familial de novo cases (Liew and Owen, 2011). Several heritable disorders with developmental deficiency in both hematopoietic and non-hematopoietic compartments are known to have higher risk of MDS and AML progression, such as Fanconi anemia and Bloom syndrome (Alter, 2007). Conversely, pure familial MDS, such as familial platelet disorder, are not associated with other non-hematological syndromes (Liew and Owen, 2011; West et al., 2014). Germline mutations associated with these conditions have been documented (Table 1.2.5.2) (Gao et al., 2014; Liew and Owen, 2011; West et al., 2014). These mutations disrupt various cellular functions ranging from DNA repair to transcription regulation to telomere maintenance. Likewise, mutations or haplodeficiency of genes such as RUNX1 and GATA2 are implicated in both familial and sporadic MDS. Therefore, despite the rarity of familial MDS, elucidating the molecular mechanism of relevant mutations can be important to the understanding of the pathogenesis of MDS.

Sporadic somatic mutations of epigenetic regulators are the most common recurring mutations in MDS (Cazzola et al., 2013a; Montalban Bravo et al., 2014). Regulators of pre-mRNA splicing and DNA methylation are often mutated early in MDS clones, whereas genes involved in histone modification and signaling are mutated in MDS subclones at a later stage of disease evolution (Papaemmanuil et al., 2013). Mutations of genes that comprise the spliceosome for pre-mRNA splicing can be identified in more than 50% of MDS cases (Table 1.2.5.2) (Cazzola et al., 2013a). Since RNA splicing is required to process more than 90% of all human genes, spliceosomes with mutated components can affect multiple genes by creating novel protein isoforms (Chen and Manley, 2009; Montalban Bravo et al., 2014). Most of these mutations involve the SF3B1, SRSF2 and U2AF1 genes, which affect U2 snRNP and 3’ splice site 14 recognition functions (Cazzola et al., 2013b; Yoshida et al., 2011). These three genes do not co- mutate with each other within the same MDS clone or subclone, which may imply functional redundancy and that a single mutation is sufficient by itself to lead to the disease phenotype, or that multiple mutations are lethal (Papaemmanuil et al., 2013). Mutants of SF3B1 and SRSF2 also co-mutate with different driver mutations. Mutations of SF3B1 do not co-mutate with the cellular metabolic gene IDH2, while mutations of SRSF2 are strongly associated with IDH2 mutations (Papaemmanuil et al., 2013). Since the SF3B1 and SRSF2 mutations are often the early events in MDS pathogenesis, their associations with specific co-mutations supports the notion of “predestination” in which founder mutations dictate the trajectory of clonal evolution. Different mutation combinations also account for distinct pathophysiology and prognostication. SF3B1 mutations are commonly associated with the presence of ringed sideroblasts and favorable clinical outcome. Conversely, patients with SRSF2 or U2AF1mutations have poor OS and high risk of AML progression (Cazzola et al., 2013a; Montalban Bravo et al., 2014).

Promoter-associated CpG islands are largely unmethylated in normal tissues at any differentiation state (Bird, 2002). In MDS, dysregulation of DNA methylation is common and the degree of aberration is associated with disease progression (Jiang et al., 2009; Montalban Bravo et al., 2014). Although aberrant DNA methylation in MDS may occur as a result of age- related epigenetic drift (Chambers et al., 2007; Maegawa et al., 2014), recurring somatic mutations have been found in several genes that regulate DNA methylation, including TET2, DNMT3A, and IDH1/IDH2 (Table 1.2.5.2) (Cazzola et al., 2013a; Khan et al., 2013). Mutations in these genes are often the founding mutations for MDS (Papaemmanuil et al., 2013). TET2 is an enzyme responsible for DNA demethylation (Issa, 2013). Loss-of-function mutations of TET2 can occur in up to 30% of all MDS cases and 60% of CMML cases with no impact on OS (Cazzola et al., 2013a). TET2 and SRSF2 are also co-mutated in CMML (Papaemmanuil et al., 2013). Loss of functional TET2 causes DNA hypermethylation, which alters gene transcription and eventually promotes the expansion of the HSC and myeloid compartments (Moran-Crusio et al., 2011; Williams et al., 2012). DNMT3a mediates de novo DNA methylation and it is highly expressed in HSC (Issa, 2013). Mutations of DNMT3A account for approximately 10% of all MDS cases (Cazzola et al., 2013a). Recurring DNMT3A mutations at R882 have reduced methyltransferase activity and can inhibit wild-type DNMT3A to form active tetramers (Russler-

Germain et al., 2014; Walter et al., 2011). Loss of DNMT3A in mouse HSC causes HSC expansion and progressive loss of differentiation capacity (Challen et al., 2011). Although DNMT3A ablation results in hypomethylation in some region of the genome, the majority of CpG islands are hypermethylated, possibly due to widespread indirect changes by the dysregulated DNMT3A targets. DNMT3A mutations are associated with unfavorable clinical outcome, except in the presence of SF3B1 co-mutation, which possibly diminishes the negative effect of DNMT3A mutants (Bejar et al., 2012; Papaemmanuil et al., 2013). Recurring gain-of- function mutations of IDH1 and IDH2 indirectly interfere with DNA methylation by the increased production of 2-hydroxyglutarate, which inhibits the α-ketoglutarate dependent DNA demethylation activity of the TET family proteins and results in DNA hypermethylation (Xu et al., 2011). Mutations of IDH1 and IDH2 occur in approximately 5% of all MDS cases and are associated with poor clinical outcome (Cazzola et al., 2013a).

Gene expression can also be modified at the chromatin level by histone modification via acetylation, methylation, and ubiquitination (Bannister and Kouzarides, 2011). Recurring mutations of the histone modifiers ASXL1 and EZH2 have been implicated in MDS typically as subclonal driver mutations (Table 1.2.5.2) (Cazzola et al., 2013a; Montalban Bravo et al., 2014; Papaemmanuil et al., 2013). Both ASXL1 and EZH2 are components of the polycomb repressive complexes (PRC1 and PRC2), which are involved in the tri-methylation of histone H3 at lysine 27 (H3K27) (Abdel-Wahab et al., 2012; Montalban Bravo et al., 2014; Yamaguchi and Hung, 2014). In MDS, loss-of-function mutations of these genes inactivate PRCs, decrease H3K27 trimethylation, and dysregulate gene expression (Issa, 2013). Mutations of both ASXL1 and EZH2 are associated with unfavorable clinical outcome (Cazzola et al., 2013a). However, only ASXL1 is linked to accelerated AML progression (Gelsi-Boyer et al., 2010). In fact, loss of EZH2 promotes MDS but attenuates predisposition of leukemic transformation, either by itself or in the presence of TET2 or RUNX1 co-mutations (Muto et al., 2013; Sashida et al., 2014). In addition to defective PRC components, other mutations may indirectly interfere with histone modification. Gain-of-function mutations of the IDH1 and IDH2 product 2-hydroxyglutarate inhibits the activities of α-ketoglutarate dependent histone demethylases, and increases trimethylation of H3K27 (Venneti et al., 2013). DNMT3A-deficiency also increases H3K27 tri-

16 methylation and therefore down-regulates gene expression (Challen et al., 2011; Wu et al., 2010).

Compared to mutations of epigenetic regulators, other driver mutations of MDS are more often acquired by MDS subclones at a later stage of disease evolution (Table 1.2.5.2) (Cazzola et al., 2013a; Papaemmanuil et al., 2013). RUNX1 is an essential transcription factor responsible for HSC differentiation (Okuda et al., 2001). RUNX1 mutations occur in familial MDS, tMDS and approximately 10% of de novo MDS (Cazzola et al., 2013a; Liew and Owen, 2011; Okuda et al., 2001). Aberrations of the RUNX1 gene range from point mutations to complete deletion, resulting in loss-of-function or dominant negative mutants. RUNX1 mutants block differentiation; some mutants may also increase HSC proliferation. RUNX1 mutation is associated with unfavorable clinical outcome. Haplodeficiency of RUNX1 is leukemogenic but additional driver mutations are typically required for progression to overt AML (Osato, 2004). The gene product of TP53, p53, is a major tumor suppressor which is mutated or deleted in a wide range of malignancies (Hollstein et al., 1991). TP53 mutations are typically acquired in MDS subclones with other driver mutations. These mutants are found in approximately 5% of MDS cases, usually in patients with more advanced disease (Cazzola et al., 2013a). Mutation of TP53 is an independent predictor of poor clinical outcome and short leukemia-free survival in MDS (Horiike et al., 2003; Saft et al., 2014). Other genes, including the RAS superfamily, SETBP1, and CSF3R, are found in a smaller subset of MDS cases as subclonal mutations. In general, these mutations cluster with other genetic aberrations as a consequence of disease evolution, and are associated with unfavorable clinical outcome (Cazzola et al., 2013a).

Table 1.2.5.2. Common germline and somatic genetic mutations in MDS (Gao et al., 2014; Liew and Owen, 2011; West et al., 2014).

Gene(s) Biological pathways Disease associations

FANC/BRCA DNA repair Fanconianemia, 50% MDS/AML risk (familial) pathway

BLM DNA repair Bloom syndrome, 25% MDS/AML risk (familial)

Familial platelet disorder with predisposition to RUNX1 Transcription factor AML, 20‐60% MDS/AML risk; ~10% cases of de novo MDS (familial/sporadic)

TERC/TERT Telomere Severe/occult dyskeratosis congenita

Embergersyndrome, MonoMACsyndrome GATA2 Transcription (familial/sporadic) Founding mutation (15‐30% cases), ringed SF3B1 RNA splicing sideroblasts phenotype (sporadic) Founding mutation (10‐20% cases), CMML, RCMD SRSF2 RNA splicing or RAEB, high risk of AML progression (sporadic) Founding mutation (<10%cases), RCMD or RAEB, U2AF1 RNA splicing high risk of AML progression (sporadic) Founding mutation (20‐30% cases of MDS, 50‐60% TET2 DNA methylation in CMML) (sporadic), no impact on overall survival Founding mutation (~10% cases) (sporadic), DNMT3A DNA methylation unfavorable outcome unless SF3B1 is co‐mutated DNA methylation / Founding mutation (~5% cases) (sporadic), IDH1/IDH2 energy metabolism unfavorable outcome Histone Subclonalmutation (15‐20% cases of MDS, 40% in ASXL1 modification CMML) (sporadic), unfavorable outcome Histone Subclonalmutation (~5% cases) (sporadic), EZH2 modification unfavorable outcome Subclonalmutation (~5% cases) (sporadic), TP53 Tumor suppressor unfavorable outcome and highrisk of leukemia progression

1.2.5.3 Cytogenetic changes Cytogenetic changes are present in approximately 50% of primary MDS (Bejar, 2014; Cazzola et al., 2013a; Haase, 2008). Of these changes, unbalanced chromosome aberrations occur much more frequently than balanced chromosome translocations. The unbalanced aberrations suggest that the cytogenetic changes in most primary MDS may be a consequence of non-disjunction at mitosis, loss of the telomere, formation of micronuclei or chromosome lagging during anaphase (Pedersen-Bjergaard and Rowley, 1994). The use of alkylating agents for chemotherapy is also known to be associated with unbalanced aberrations in t-MDS, in particular the deletion of chromosomes 5 and 7 (Pedersen-Bjergaard and Rowley, 1994). Moreover, partial or total deletion of chromosomes occurs more often than the gain of genetic material, suggesting the loss of tumor suppressor genes as a possible genetic mechanism in the pathogenesis of MDS (Haase, 2008). Conversely, the presence of cytogenetic changes associated with specific subtypes of AML, such as the t(15;17)(q24;q21) translocation leading to the PML-RARα fusion, and the t(8;21)(q22;q22) translocation leading to the RUNX1-RUX1T1 fusion, are not classified as MDS regardless of other supporting clinical evidence (Brunning et al., 2008; Greenberg et al., 2013).

Despite the heterogeneity of cytogenetic changes in MDS, a number of recurring chromosome abnormalities have been documented (Table 1.2.5.3) (Bejar, 2014; Brunning et al., 2008; Haase, 2008). The most common sole cytogenetic change in MDS is deletion of varying amounts of genetic material from within the long arm of chromosome 5 (5q-), including a minimal core of approximately 1.5 megabases at 5q31-q32 (Boultwood et al., 2002; Cazzola, 2008; Haase, 2008). Patients with the 5q- aberration as the sole karyotypic abnormality and without excess of blasts belong to a separate MDS subtype – 5q- syndrome. The 5q- syndrome is more common in older females and is associated with refractory macrocytic anemia, normal or elevated platelets, and mild leukocytopenia. These patients often have more favorable prognosis with milder symptoms, better long term survival and lower risk of leukemia transformation. The pathophysiological basis of the 5q- syndrome is suspected to be the consequence of haploinsufficiency of genes from the deleted region. For instance, the ribosomal protein S14 (RPS14) gene is responsible for ribosome biogenesis; partial loss of RPS14 expression as a result of 5q deletion is associated with defective erythropoesis and therefore anemia (Ebert et al., 2008; Pellagatti et al., 2008). Also, knockdown of the micro RNAs miR-145 and miR-146a increases the expression of Toll-

19 interleukin-1 receptor domain-containing adaptor protein (TIRAP) and tumor necrosis factor receptor-associated factor-6 (TRAF6) in HSC and other hematopoietic progenitors, which are involved in innate immune signaling (Starczynowski et al., 2010). Up-regulation of TRAF6 in mice results in a condition that resembles 5q- syndrome with dysmegakaryocytopoiesis, neutropenia, and AML progression. Recently, Schneider et al. reported that haploinsufficiency of casein kinase 1A1 confers a competitive repopulation advantage to HSC (Schneider et al., 2014). Other genes and micro RNAs from the deleted region, such as SPARC and NPM1, are known to be tumor suppressors, but their exact involvement in the pathophysiology of the 5q- syndrome is yet to be fully elucidated (Fuchs, 2012).

Isolated trisomy 8 is seen in approximately 5% of MDS patients but by itself is not specific to MDS (Brunning et al., 2008; Haase, 2008; Paulsson and Johansson, 2007). Another frequent cytogenetic change in MDS is total or partial deletion of chromosome 7 (7- or 7q-), which occurs more often in younger MDS patients (Haase, 2008). Both of these karyotypes are also seen in other neoplastic disorders such as AML and MPD, and are associated with worse symptoms and prognosis in terms of disease severity and risk of leukemia progression.

Complex karyotype is defined as the presence of at least three independent chromosome aberrations within one cell clone (International Standing Committee on Human Cytogenetic Nomenclature. et al., 2009). The frequency of complex karyotype in MDS is comparable to the 5q- syndrome (Bejar, 2014; Haase, 2008). However, unlike the 5q- syndrome, complex karyotype is among the poorest risk factor in MDS with short median overall survival and high risk of leukemia progression. The complexity is thought to be a stepwise accumulation of chromosome aberrations from a primary chromosome abnormality, driven by genetic instability. Unbalanced chromosome aberrations involving chromosomes 5, 7 and 8 are frequently involved in complex karyotype, which is consistent with the notion of karyotype evolution from a primary abnormality. In general, the extent of complexity is associated with disease outcome (Haase et al., 2007). MS for patients with normal karyotype is 53 months, whereas patients with 3 abnormalities have shorter MS at 17 months, and even worse at 9 months for patients with 4-6 abnormalities.

The concept of monosomal karyotype was proposed in recent years to describe a particular type of aberration with two or more distinct autosomal chromosome monosomies, or one single 20 autosomal monosomy in the presence of structural abnormalities (Breems et al., 2008). In AML, monosomal karyotype is a stronger prognostic indicator than complex karyotype and is linked with unfavorable outcome, as well as poor response rate to treatment (Breems and Lowenberg, 2011). Monosomal karyotype is often admixed with complex karyotype and the incidence is associated with the complexity of the aberration. In MDS, monosomal karyotype most frequently involves monosomy 7; other recurring abnormalities include chromosome 13, 20, and 18 (Raza et al., 2011). Several studies have suggested that monosomal karyotype is an independent indicator of prognosis and its prognostic value surpasses that of complex karyotype (Gangat et al., 2013; Xing et al., 2014). Nonetheless, this notion remains controversial since other analyses have found that the prognostic value of monosomal karyotype merely reflects the complexity of chromosomal abnormalities (Schanz et al., 2013; Valcarcel et al., 2013).

Table 1.2.5.3. Common recurring chromosomal abnormalities in MDS (Bejar, 2014; Brunning et al., 2008; Haase, 2008).

Cytogenetic Approximate Associated Disease associations abnormalities frequency risk

Normal 50‐60% Good

5q‐ 7%a Anemia, elevated platelet counta Good

Common in other myeloid 8+ 5%a Intermediate malignancies

7‐/7q‐ 4%a Lower median age Intermediate

Y‐ 2%a Very Good

Common in other myeloid 20q‐ 2%a Good malignancies

3 abnormalities 2% Complex karyotype Poor

>3 abnormalities 7% Complex Karyotype Very Poor

Often overlap with complex Monosomal >7% Very Poor karyotype

a Isolated, without othercytogenetic abnormalities

1.2.6 Risk stratification The clinical outcome of MDS is measured by overall survival (OS) and leukemia free survival (LFS). Various clinical findings from initial diagnosis and subsequent assessments are useful in predicting the risk of shortened life span or leukemia progression. Establishing the prognosis for MDS is crucial for patient expectations, care and treatment options. Although not intended to predict outcomes, the FAB and WHO diagnostic classifications have some prognostic value: RCUD and RARS belong to the low risk group with longer OS and LFS; RCMD and RAEB-1 comprise the intermediate risk group; and RAEB-2 is high risk with shorter OS and LFS (Brunning et al., 2008). Nevertheless, the diagnosis of MDS subtype is mainly based on cytopenias and dysplasia, while cytogenetic findings are ignored with the exception of 5q- syndrome. In addition, classification of certain subtypes, such as RAEB, only requires the presence of cytopenias and dysplasia without distinguishing the number of lineages involved. Since the degree of cytopenias and dysplasia are associated with clinical outcome (Greenberg et al., 1997), it is difficult to stratify risk within the same subtype by diagnostic classification alone.

The International Prognostic Scoring System (IPSS) is established based on a large cohort of MDS cases as a standardized risk stratification tool to estimate prognosis for MDS (Greenberg et al., 1997). The IPSS assigns relative risk scores to three independent prognostic variables at initial MDS diagnosis that include the percentage of bone marrow myeloblasts, the number of cytopenia, and the number and types of cytogenetic abnormalities (Table 1.2.6a). Patients are stratified into four risk categories based on the sum of all risk scores from the three variables: low, intermediate-1 (INT-1), intermediate-2 (INT-2), and high. The expected clinical outcome is most favorable for IPSS low and most unfavorable for IPSS high, with MS at 5.7 and 0.4 years, and 25% AML progression at 9.4 and 0.2 years, respectively. The IPSS demonstrated greater prognostic discriminating power than the FAB classification system. Recently, the International Working Group for Prognosis in MDS (IWG-PM) has proposed a revised version of the IPSS (IPSS-R) (Table 1.2.6b) (Greenberg et al., 2012). Compared to the IPSS, the IPSS-R was derived from a larger dataset and incorporated additional cutoffs to allow more refined risk scores for cytogenetics, BM blast percentage, hemoglobin, platelets, and ANC. The cutoff for ANC decreases from 1.8x109 /L to 0.8x109 /L, which accounts for infection risk instead of neutropenia. The list of cytogenetic abnormalities has been expanded to reflect the growing

23 understanding of the impact of cytogenetics to MDS. Statistical adjustment based on patient’s age is also incorporated into the calculation of IPSS-R to account for the impact of age on survival. The IPSS-R defined five risk categories: very low, low, intermediate (INT), high, and very high. Based on the training cohort, the MS and 25% AML progression of IPSS-R high and very high resemble those of IPSS INT-2 and high, respectively. However, the IPSS-R is more refined than the IPSS for lower risk stratification in which three risk categories (very low, low, intermediate) are defined for IPSS-R instead of two for IPSS (low, INT-1). Furthermore, validation of these scoring systems using other cohorts suggest that the IPSS-R indeed has higher predictive power for OS and LFS than the IPSS (Neukirchen et al., 2014; Voso et al., 2013). Although it is still common to use the IPSS in current clinical settings, wider adoption of the IPSS-R is expected in the future.

Both the IPSS and IPSS-R are designed for initial prognostic and planning purposes. Although the IPSS has been shown to have prognostic value for patients during the course of their disease, there may be bias when the same risk category contains cases from initial and continual prognosis (Cutler et al., 2004; Sierra et al., 2002). The WHO classification-based prognostic scoring system (WPSS) was created as a dynamic prognostic scoring system based on time- dependent analysis of multiple prognostic factors to provide prognosis any time after initial diagnosis (Table 1.2.6c) (Malcovati et al., 2005). Prognostic variables for the WPSS include WHO classification, karyotype, and transfusion dependency. Transfusion dependency was later replaced by hemoglobin level as an “objective” indicator of the severity of anemia (Asemissen and Giagounidis, 2011; Malcovati et al., 2011). Five risk categories were defined by the WPSS: very low, low, intermediate (INT), high, and very high. Although the predictive power of the WPSS is inferior to the IPSS-R (Neukirchen et al., 2014; Voso et al., 2013), it has been shown to be effective in predicting post-transplantation outcome in MDS patients (Alessandrino et al., 2008). Risk-based therapy for MDS may also require the WPSS to provide updated prognostic information during the course of treatment (Cazzola and Malcovati, 2010). Hence, the IPSS or IPSS-R can be deployed in tandem with WPSS for current clinical use. Other prognostic scoring systems have also been developed but are yet to be widely adapted, which include the lower-risk MDS prognostic scoring system and the MD Anderson comprehensive scoring system (Garcia- Manero et al., 2008; Kantarjian et al., 2008).

Aside from the established scoring systems, a number of additional markers are known to be independent predictors of disease outcome. Clinical indicators, such as serum lactate dehydrogenase (LDH) (Wimazal et al., 2008), performance status and ferritin (Malcovati et al., 2005), were recommended to be included in the IPSS-R as supplements to the model (Greenberg et al., 2012). Furthermore, current prognostic scoring systems incorporate cytogenetic abnormalities but not somatic mutation data, despite their crucial role in the pathophysiology and outcome of MDS. Nonetheless, the complexity of mutation data has to be addressed before clinical adaptation is feasible (Bejar, 2013). There is a continual effort to refine the criteria for current prognostic variables and discover novel prognostic markers. Hence, future scoring systems are expected to improve upon current scoring systems and incorporate additional markers.

Table 1.2.6a. International prognostic scoring system (IPSS) for MDS (Greenberg et al., 1997).

Score Value

Prognostic 00.511.52.0 variable

Marrow blasts (%)a <5 5‐10 – 11‐20 21‐30

Inter‐ Cytogeneticsb Good Poor – – mediate

Cytopeniac 0/12/3–––

IPSS risk Median survival 25% AML category Overall Score (year) without progression (year) (% IPSS pop.) therapy without therapy

LOW (33) 0 5.7 9.4

INT‐1 (38) 0.5‐1.0 3.5 3.3

INT‐2 (22) 1.5‐2.0 1.1 1.1

HIGH (7) ≥2.5 0.4 0.2 a Patients with >20% blasts are now considered to have AML by WHO classification. b Cytogenetics: Good = normal, isolated ‐Y, i sol a ted 5q‐,isolated20q‐; Intermediate = other abnormalities; Poor = complex (≥3 abnormalities), chromosome 7 anomalies. Abnormalities such as t(8;21), inv16, and t(15;17) are considered to be AML. c Cytopenias: hemoglobin <10 g/dL, neutrophil count <1.8x109 /L, and platelets <100x109 /L.

Table 1.2.6b. Revised international prognostic scoring system (IPSS‐R) (Greenberg et al., 2012).

Score Value

Prognostic 0 0.5 1 1.5 2 3 4 variable

Very Inter‐ Very Cytogeneticsa –Good– Poor good mediate poor

Marrow blasts ≤2–>2‐<5 – 5‐10 >10 – (%)

Hemoglobin ≥10 – 8‐<10 <8 – – – (g/dL)

Platelets 50‐ ≥100 <50 – – – – (x109 /L) <100

ANC ≥0.8 <0.8 – – – – – (x109 /L)

IPSS risk Median survival 25% AML category Overall Score (year) without progression (year) (% IPSS‐R pop.) therapy without therapy

VERY LOW (19) ≥1.5 8.8 Not reached

LOW (38) >1.5‐3 5.3 10.8

INT (20) >3‐4.5 3 3.2

HIGH (13) >4.5‐61.61.4

VERY HIGH (10) >6 0.8 0.7 a Cytogenetics: Very good = ‐Y, 11q‐;Good=normal,5q‐,12p‐,20q‐,double including 5q‐;Intermediate=7q‐, +8, +19, i(17q), any other single or double independent clones; Poor = ‐7, inv(3)/t(3q)/3q‐, double including ‐7/‐7q, complex: 3 abnormalities; Very poor = complex: >3 abnormalities.

Table 1.2.6c. WHO classification‐based prognostic scoring system (WPSS) (Malcovati et al., 2011).

Score Value

Prognostic 0123 variable

RCUD, RARS, WHO category RCMD RAEB‐1RAEB‐2 5q‐ syndrome

Cytogeneticsa Good Intermediate Poor –

Severe anemiab Absent Present – –

AML progression WPSS risk Median survival (2/5 years category Overall Score (year) without cumulative (% WPSS pop.) therapy probability)

VERY LOW (23) 0 8.6 0.0 / 0.06

LOW (28) 1 6 0.11 / 0.24

INT (19) 2 3.3 0.28 / 0.48

HIGH (23) 3‐41.80.52 / 0.63

VERY HIGH (7) 5‐6 1 0.79 / 1.0 a Cytogenetics: Good = normal, isolated ‐Y, i sol a ted 5q‐,isolated20q‐; Intermediate = other abnormalities; Poor = complex (≥3abnormalities), chromosome 7 anomalies. Abnormalities such as t(8;21), inv16, and t(15;17) are considered to be AML. b Severe anemia : hemoglobin <9 g/dL in males or <8 g/dL in females.

1.2.7 Therapeutic options Therapeutic decisions for MDS are made by means of a risk-based approach that triages patients into lower and higher-risk treatment groups according to their diagnosis and prognosis (Cheson et al., 2000; Cheson et al., 2006; Greenberg et al., 2013). The lower-risk group consists of IPSS low/INT-1, IPSS-R very low/low/INT, and WPSS very low/low/INT; the higher-risk group consists of IPSS INT-2/high, IPSS-R INT/high/very high, and WPSS high/very high (Greenberg et al., 2013). Moreover, IPSS-R intermediate patients may belong to either risk groups depending on additional prognostic factors and responsiveness to treatment. The IPSS and IPSS-R are suitable for treatment planning after initial diagnosis, while the WPSS is recommended for continual assessment and adjustment (Table 1.2.7a). Other variables, such as the patient’s age, performance status, and presence of co-morbidities, are crucial to determine patient eligibility to receive a certain therapy. The treatment options for MDS consist of drug therapy, allogeneic hematopoietic stem cell transplantation (HSCT), and supportive care (Greenberg et al., 2013; Munker et al., 2000b) (Table 1.2.7b). These options are used in combination with each other to improve quality of life and disease outcomes (Cheson et al., 2000; Cheson et al., 2006) (Table 1.2.7c). Algorithms such as the NCCN treatment guidelines for MDS (Greenberg et al., 2013) and the “MDS Clear Path” (http://www.mdsclearpath.org) have been developed based on consensus recommendations to aid physicians in deciding the appropriate therapeutic approach. Patients with no response to treatment should be considered for participation in clinical trials. Also, all patients should receive adequate supportive care to alleviate their symptoms.

Table 1.2.7a. Definition of lower and higher risk patients,basedonNCCN recommendation (NCCN, 2014).

Treatment IPSS IPSS‐RWPSS group VERY LOW, LOW, Lower‐risk LOW, INT‐1VERY LOW, LOW, INTa INTERMEDIATE

Higher‐risk INT‐2, HIGH INTa, HIGH, VERY HIGH HIGH, VERY HIGH

a Patients in the IPSS‐RINT risk category may be managed as lower or high‐risk treatment group.

Table 1.2.7b. Recommended treatment options for MDS (NCCN, 2014).

Treatment Treatment type Treatment name Intensity group Anti‐thymocyte globlin (ATG), Lower‐risk Immunosuppression Low cyclosporine A

Lower‐risk Immunomodulation Lenalidomide (5q‐ patients) Low

Higher‐risk Hypomethylation 5‐Azacytidine, decitabine Low

Intensive induction Higher‐risk Idarubicin, cytarabine High chemotherapy Allogeneic hematopoietic stem Higher‐risk Transplantation High cell transplantation (HSCT) Erythropoiesis‐stimulating agents Hematopoietic Lower‐risk (ESAs)with granulocyte colony‐ Supportive cytokines stimulating factor (G‐CSF) Packed red blood cells (PRBC), Lower/Higher Transfusion Supportive platelet

Lower/Higher Iron chelation Deferoxamine, deferasirox Supportive

Table 1.2.7c. Measurement of response in MDS (Cheson et al., 2000 and 2006).

Response type Measurement

Altering disease course Complete remission (CR) –nocytopenia,≤5% BM blasts, normalmaturation Partial remission(PR) – nocytopenia,decreasedblasts Marrow CR – ≤5% BM blasts and ↓ by 50% Stable disease – failure to achieve PR/CR, but no progression for>8 weeks Failure – deathordisease progression Relapse – blast return to pretreatment level, worsening cytopenia Disease progression – increase blast level, worsening cytopenia Disease transformation –transformationtoAMLwith30%or more blasts Survival –overall,eventfree,progressionfree,diseasefree, cause‐specific Cytogenetic response Major– complete disappearance of cytogeneticabnormality Minor– ≥50% reduction of the chromosomalabnormality Quality of Life (QoL) Physical,functional,emotional,social,spiritual

Hematologicimprovement (HI) Erythroid (HI‐E): Major– ↑≥2g/dL in hemoglobin, transfusionindependence Minor – ↑1‐2g/dL in hemoglobin, ↓50% in transfusion requirement Platelet (HI‐P): Major– ↑≥30x109 /L, transfusion dependence Minor– ↑50% with net increase between 10‐30x109 /L Neutrophil(HI‐N): Major– ↑100% or ↑0.5x109 /L in netANC Minor– ↑100% but ↑<0.5x109 /L in netANC Progression/relapse: Erythroid – ↓≥2g/dL in hemoglobin, transfusion dependence Neutrophil/platelet – ↓50% maximum response, transfusion dependence

1.2.7.1 Therapy for lower-risk patients Treatment goals for lower-risk MDS patients are to manage cytopenias and improve quality of life (Cheson et al., 2006; Greenberg et al., 2013) (Figure 1.2.7.1). Lower-risk patients with confirmed MDS who do not show overt symptom should be monitored every three to six months for disease progression. IPSS low or Int-1 MDS patients with 5q- syndrome should receive lenalidomide if they have transfusion-dependent anemia (Greenberg et al., 2013). Lenalidomide is an immunomodulatory and antiproliferative agent that requires the direct binding and inhibiton of the E3 ligase protein cereblon (Lopez-Girona et al., 2012; Zhu et al., 2011). Lenalidomide can effectively induce cytogenetic response and erythroid response (HI-E) in patients with the 5q- karyotype (Fenaux et al., 2011; List et al., 2006). Lenalidomide may also be beneficial to patients without the 5q- karyotype (Raza et al., 2008) or higher-risk patients in combination with azacitidine (Sekeres et al., 2012). Conversely, lenalidomide may induce or worsen existing neutropenia and thrombocytopenia (Gaballa and Besa, 2014).

Pharmacological doses of erythropoietic stimulating agents (ESAs) such as erythropoietin (EPO) can compensate for the defective EPO response and alleviate the symptoms of anemia for patients with serum EPO level at 500 mU/mL or lower (Greenberg et al., 2013; Santini, 2011). Concurrent use of G-CSF can enhance the effect of EPO in a synergistic manner especially for patients with more than 15% ringed sideroblasts in the BM (Hellstrom-Lindberg et al., 1998). Furthermore, combinations of active treatments and ESAs have demonstrated promising initial results, but these treatment regimens require additional prospective studies before they can be included in the standard of care (Itzykson et al., 2012; Sibon et al., 2012). Nonetheless, the general response rate to ESAs is lower than 50% and varies greatly among different groups of patients depending on parameters such as karyotype, blast percentage, IPSS, and the time elapsed since first diagnosis (Santini, 2011). Moreover, many MDS patients become non-responsive to ESAs after prolonged treatment, which may be the results of iron depletion, disease progression, loss of sensitivity to ESAs, or autoimmune response to erythrocytes (Santini, 2011).

Other patients with symptomatic cytopenias may be considered for IST by antithymocyte globulin (ATG) or cyclosporine A (Greenberg et al., 2013). It has been shown that increased apoptosis in early MDS is associated with T lymphocyte-mediated BM failure and can be reversed by IST (Kochenderfer et al., 2002). Although IST is not a curative therapy, it can induce 32 long-term hematologic improvement (HI) and about one third of patients become transfusion independent with normal counts. In addition, patients with certain biomarkers are more likely to respond to IST, including age below 60, human leukocyte antigen (HLA)-DR15 positivity, presence of a Paroxysmal Nocturnal Hemoglobinuria clone, isolated 8+ karyotype, and hypoplastic BM (Lim et al., 2007; Nakao et al., 2006; Sloand et al., 2005; Sloand et al., 2008). Treatment with azacitidine may be considered for patients who are not suitable for IST (Greenberg et al., 2013). Patients with anemia are recommended to receive erythropoiesis- stimulating agents (ESAs) (Greenberg et al., 2013), as well as other supportive care when necessary. Allogeneic HSCT is infrequent for lower-risk patients since the expected slow progression of their disease does not justify the potential risk of the transplantation (Bartenstein and Deeg, 2010). Nonetheless, HSCT may be considered if MDS associated quality-of-life (QoL) impairment is substantial, or upon disease progression (Cutler et al., 2004; Koreth et al., 2013).

Lower‐risk MDS

Allogeneic HSCT assessment NO Monitor every Symptomatic? 3‐6 months YES

Anemia Neutropenia Thrombocytopenia

NO NO 5q‐ Karyotype Immunosuppressive therapy (IST) Candidate YES YES

ESA Lenalidomide IST HMA/ESA

Loss of response without Disease Higher risk MDS disease progression progression

Figure 1.2.7.1. Recommended therapeutic approach for lower‐risk MDS patients (based on NCCN guidelines for MDS and MDS Clear Path, http://www.mdsclearpath.org). Supportive care should be provided whenever appropriate. Patients undergoing treatment should be monitored every 1 to 3 months. ESA – erythropoiesis‐stimulating agent, HMA – hypomethylating agent. ESA is often administered with granulocyte colony‐stimulating factor (G‐CSF).

1.2.7.2 Therapy for higher-risk patients Higher-risk MDS patients have substantially shorter survival and higher-risk of leukemia transformation. As a result, the goals of therapy in higher-risk MDS are to prevent or delay progression to AML and to extend survival. High-intensity therapies, namely allogeneic HSCT and induction chemotherapy, are required if the patient is eligible (Greenberg et al., 2013). Eligibility for these therapies is based on age, performance status, major co-morbid conditions, psychosocial status, patient preference, and availability of a caregiver. Patients who are not “fit” enough for high-intensive therapies will receive low-intensity therapies or supportive care.

Allogeneic HSCT is the only potentially curative modality for MDS (Munker et al., 2000b). HSCT can restore effective hematopoiesis after the complete or partial removal of existing MDS cells. Immune cells from the transplanted HSC can also eliminate residual MDS cells via the graft-versus-tumor effect (Martino et al., 2002). However, HSCT is a highly risky procedure and generally not recommended for patients with less advanced disease (Cutler, 2010; Sekeres and Cutler, 2014). Different types of HLA-matched sources can contribute HSC for the transplantation, but donor availability varies greatly (Bartenstein and Deeg, 2010). About 25% of patients will have a matched sibling or related donor. An unrelated donor can be identified for about 50-60% of Caucasians but the proportion can be as low as 10% in certain ethnic minority groups. In some situations, an incomplete match may be acceptable, but there is a corresponding increase in the risk of non-engraftment and graft-versus-host-disease (GVHD). The use of partially mismatched umbilical cord blood allows more rapid engraftment with a lower risk of GVHD than HSCT from an unrelated donor, but limited availability remains an issue for this method (Barker et al., 2002). Candidates for HSCT often require conditioning, which increases the chance of successful engraftment by overcoming the recipient’s immunological barrier and by decreasing the number of abnormal MDS cells (Bartenstein and Deeg, 2010). The NCCN guidelines recommend that younger patients should receive conventional conditioning (Greenberg et al., 2013), which consists of induction chemotherapy and is associated with a lower-risk of relapse but higher toxicity (Warlick et al., 2009). Reduced intensity conditioning (RIC) is a viable alternative for patients with co-morbidities or who are older than 55 or 60 years of age (Greenberg et al., 2013). The probability of relapse is higher with RIC, but the risk may be

35 offset by lower toxicity, higher chance of successful engraftment, and longer survival for patients who otherwise cannot undergo HSCT with conventional conditioning.

High-intensity induction chemotherapy should be considered for HSCT candidates who lack suitable donors or patients whose marrow blast count requires reduction (Greenberg et al., 2013). Conversely, non-intensive therapy using azacitidine and decitabine should be considered for patients who are not suitable for intensive therapy. Both azacitidine and decitabine are hypomethylating agents (HMAs) intended to counteract aberrant DNA methylation in MDS by depleting cellular DNMT (Santini et al., 2013). Both of these molecules are also cytosine analogs that can be integrated into DNA, but only azacitidine can be integrated into RNA. Randomized clinical trials have shown that both of these HMAs can reduce transfusion dependency and improve QoL for lower-risk patients when compared to best supportive care (BSC) (Kantarjian et al., 2006; Kornblith et al., 2002; Silverman et al., 2006). For higher-risk patients, azacitidine improved OS and CR rate when compared to BSC and low-dose cytarabine, while decitabine improved CR rate but not OS (Fenaux et al., 2009; Lubbert et al., 2011). Therefore, azacitidine is currently the preferred treatment for higher-risk MDS over decitabine (Greenberg et al., 2013).

Higher‐risk MDS

NO YES Allo‐HSCT assessment

Pretransplant HMA candidate? regimen

YES NO

HMA RIC/HMA Allo‐HSCT IC

Best supportive care

Figure 1.2.7.2. Recommended therapeutic approach for higher‐risk MDS patients (based on NCCN guidelines for MDS and MDS Clear Path, http://www.mdsclearpath.org). Supportive care should be provided whenever appropriate. Patients with higher‐risk MDS should be monitored every 1 to 3 months. Allo‐SCT – allogeneic hematopoietic stem cell transplantation, RIC – reduced intensity conditiining, IC – induction chemotherapy.

1.2.8 Supportive care for MDS Participation in clinical trials for experimental drugs should be considered for patients who are not eligible for or no longer responsive to recommended treatments. Adequate supportive care should be maintained for all patients regardless of disease state and prognosis. The goal of supportive care for MDS is to mitigate the primary and secondary symptoms caused by cytopenia.

1.2.8.1 Infection and antimicrobials Infection is a common cause of morbidity and mortality in MDS, which accounts for almost 40% of mortality among lower-risk MDS patients (Cunningham et al., 1995; Dayyani et al., 2010). Predisposition to infection is likely the consequence of a combination of factors including neutropenia, immune dysfunction, iron overload and the presence of other co-morbidities (Toma et al., 2012). Active pharmacologic treatments such as HMAs, lenalidomide and chemotherapy may exacerbate neutropenia and further increase the risk of infection. Moreover, infection accounts for half of all post-HSCT mortality and the risk is even higher for patients with pretransplant neutropenia (Deeg et al., 2002; Scott et al., 2008). Antimicrobials are used to treat infectious episodes or as a prophylactic measure during active treatment (Toma et al., 2012). However, continual use of antimicrobial prophylaxis is not recommended due to the risk of induced resistance and adverse drug effects. Coversely, growth factors such as granulocyte colony-stimulating factor (G-CSF) can mitigate neutropenia especially for lower-risk MDS by increasing the ANC and restoring neutrophil function (Hellstrom-Lindberg et al., 1998; Nagler et al., 1990; Verhoef and Boogaerts, 1991).

1.2.8.2 Bleeding and platelet transfusion Based on a survey on 2410 patients, more than half of all MDS patients have thrombocytopenia at initial diagnosis (Kantarjian et al., 2007). Thrombocytopenia is more frequent and severe for higher-risk patients, and may be induced or worsened by active pharmacologic treatments even though achieving remission may restore platelet count. Increased risk of bleeding is the clinical consequence of defective hemostasis, and the severity of bleeding is associated with the degree of thrombocytopenia (Kantarjian et al., 2007). Bleeding complications range from gingival bleeding to gastrointestinal hemorrhage, and may account for up to 20% of mortality as a contributing factor to death in MDS (Kantarjian et al., 2007; Konstantopoulos et al., 1989). 38

Platelet transfusions improve hemostasis and are widely used for patients with thrombocytopenia (Ramsey et al., 2012; Sullivan et al., 2007). Patients with ongoing thrombocytopenia often require chronic platelet transfusion, which increases the risk of transfusion-transmitted infection and refractoriness to foreign platelets due to alloimmunization (Agarwal et al., 2014; Kuehnert et al., 2001).

1.2.8.3 Anemia and RBC transfusion Up to 90% of MDS patients become anemic during the course of their disease (Malcovati, 2009). More than 50% of these patients have a hemoglobin level below 100 g/L at diagnosis, and almost 30% have severe anemia with less than 80 g/L hemoglobin (Mitchell et al., 2013). Low hemoglobin level causes physical impairment and fatigue, thereby reducing the QoL of MDS patients (Jansen et al., 2003; Oliva et al., 2005). According to the Fick principle, cardiac output increases with lower hemoglobin level as a physiological response to maintain adequate oxygen supply (McLellan and Walsh, 2004). Prolonged increase in cardiac output induces cardiac remodeling, cardiac enlargement and left ventricular hypertrophy (Oliva et al., 2005). Since many MDS patients also have pre-existing cardiovascular disease, chronic anemia can further impair their cardiac condition. Indeed, severe anemia with hemoglobin level lower than 90 g/L is an independent predictor of reduced overall survival and higher risk of cardiac death (Malcovati et al., 2011). Anemia in MDS is usually macrocytic or normocytic in the absence of nutritional deficiency, and is the consequence of impaired erythropoiesis due to excessive premature apoptosis or defective response to erythropoietin (EPO) in hematopoietic precursors (Catenacci and Schiller, 2005; Cazzola et al., 2008; Cazzola and Malcovati, 2005; Santini, 2011).

RBC transfusion is required when symptomatic anemia cannot be adequately treated by other means. In general, RBC transfusion is considered when the patient’s hemoglobin level is below 80 g/L; or 100 g/L if the patient has pre-existing cardiovascular disease (Balducci, 2006), although other clinical features such as fatigue and shortness of breath are also useful to determine if RBC transfusion is necessary (Murphy et al., 2001). Packed red blood cells (PRBC) are commonly used for transfusion in MDS, which are prepared from whole blood by removing plasma and then mixing with anticoagulant and storage solution to preserve RBC viability and functionality (Clarke and Chargé, 2007). Leukocytes are depleted from the blood product by filtration to decrease febrile non-hemolytic transfusion reactions and HLA alloimmunization 39

(Chapman et al., 1998). The volume of a typical unit of PRBC ranges from 250-350 mL and each unit is expected to increase the hemoglobin level by approximately 10 g/L (Elzik et al., 2006). MDS patients often become dependent on RBC transfusions due to chronic impairment of erythropoiesis. Transfusion dependency in MDS is defined as the requirement for at least one RBC transfusion every eight weeks over a period of four months (Cheson et al., 2006; Malcovati and Cazzola, 2008). Up to 80% of IPSS high patients are RBC transfusion-dependent, compared to 40% for IPSS low patients (Balducci, 2006). Even within different prognostic groups, transfusion dependency is a predictor of poor disease outcome in MDS (Cazzola and Malcovati, 2005). In addition, the intensity of transfusion is associated with disease progression and outcome (Balducci, 2006; Pereira et al., 2011). Therefore, transfusion dependency may be an indicator of more serious bone marrow failure or advanced disease. Furthermore, Oliva et al. reported that cardiac remodeling due to increased cardiac output is more common in transfusion- dependent patients at 92% compared to non-transfused patients at 48% (Oliva et al., 2005). Their observations support the notion that cardiac output is not normalized in transfusion dependent patients due to fluctuation of hemoglobin levels between RBC transfusions. Thus, transfusion is not sufficient to prevent the risk of anemia-related cardiac complications.

Despite its benefits in alleviating the symptoms of anemia, RBC transfusion is associated with other complications. Since the 1980s, it is extremely rare for transfused patients to become infected due to pathogen contaminated blood products (Vamvakas and Blajchman, 2009). Nonetheless, transfusion-dependent patients have a higher risk of infection than non-transfused patients at 81% versus 56%, respectively, which could be the consequence of immunosuppression or concurrent neutropenia (Goldberg et al., 2010). Acute reactions such as acute hemolytic reactions and transfusion-related acute lung injury remain possible but their occurrence is relatively infrequent (Sharma et al., 2011; Vamvakas and Blajchman, 2009). Long- term allergic reaction and alloimmunization to transfused RBC occurs in approximately 20% of transfusion-dependent MDS patients (Platzbecker et al., 2012; Stiegler et al., 2001). The risks or consequences of these complications can be mitigated by the use of RBC antigen-negative units or leukocyte reduction of blood products (Balducci, 2006). Iron accumulation is an inevitable consequence of chronic RBC transfusion due to the presence of iron in hemoglobin and the excess iron may need to be removed by iron chelation therapy (ICT) (Mitchell et al., 2013).

Transfusion of RBC after prolonged storage may also be harmful because acute clearance of damaged/aged RBC from stored RBC leads to acute tissue iron deposition and initiates inflammation (Hod et al., 2010). Iron overload and ICT will be discussed in later sections.

1.3 Iron

1.3.1 The role of iron in biology Iron has various oxidation states ranging from Fe2- to Fe6+, with the Fe2+ (ferrous) and Fe3+ (ferric) states being the most biologically relevant (Crichton, 2009a). In a simple aqueous 2+ 3+ solution, reduction and oxidation between dissolved Fe and Fe can be readily and reversibly accomplished by common reducing agents and molecular dioxygen (O2), respectively (Aisen et al., 2001). This flexibility makes iron suitable for a variety of biochemical reactions ranging from dioxygen binding to electron transport. Therefore, iron is crucial to almost all life forms on Earth and it is incorporated into hemoproteins, iron-sulfur proteins, and other iron-containing proteins (Crichton, 2009b) (Table 1.3.1).

Hemoproteins are a major class of iron-containing proteins that contain heme (Crichton, 2009b). Heme is a porphyrin molecule with an Fe2+ at the center (Crichton, 2009f). In mammalian cells, heme biosynthesis begins in the mitochondria in which succinyl CoA from the citric acid cycle condenses with glycine to form δ-aminolaevulinate. The process continues in the cytosol to produce coproporphyrinogen III, which then re-enters the mitochondria and becomes protoporphyrin IX. The final step of heme biosynthesis involves the incorporation of Fe2+ into the protoporphyrin IX molecule by ferrocheletase. Exportation of heme to the cytosol is mediated by the mitochrondrial heme exporter FLVCR1 (Korolnek and Hamza, 2014). Alternatively, extracellular heme can be imported into the cell by importers such as Hrg1 on the plasma membrane (Rajagopal et al., 2008). Heme is incorporated into three types of hemoproteins: oxygen carriers, oxygen activators, and electron transport proteins (Crichton, 2009b). In higher eukaryotes, transportation and storage of molecular dioxygen are mediated by oxygen carriers such as hemoglobins and myoglobins. Molecular dioxygen is reversibly bound to the Fe2+ ion of heme in a hydrophobic pocket to form a ferric-superoxo complex. The complex is stabilized by a distal histidine proton via hydrogen bonding. The strength of dioxygen binding and release depend on factors such as globin type, acidity, and partial pressure of dioxygen.

Oxygen activators are characterized by the heme’s capability to reach the Fe4+ (ferryl) cation state (Crichton, 2009b). This type of hemoprotein is able to participate in various chemical reactions, including reduction of dioxygen to water by cytochome c oxidase; catalatic or peroxidatic activities by catalases and peroxidases, respectively; and incorporation of a single oxygen atom from dioxygen into endogenous or foreign substrates by cytochrome P450s. Electron transport proteins transfer electrons from reduced donors to target acceptors (Crichton, 2009b). Cytochrome a, b, and c are classical hemoproteins responsible for electron transportation as components of the respiratory chain in the mitochondria.

Iron-sulfur (Fe/S) clusters are believed to be one of the earliest biological catalysts, and consist of iron ions typically bound to cysteine-sulfur groups (Crichton, 2009b; Lill and Muhlenhoff, 2008). Iron in Fe/S clusters usually alter between the Fe2+ and Fe3+ states with versatile electrochemical properties suitable for various redox reactions. Fe/S clusters have four basic structures ranging from simple rubredoxins which consist of an iron centre bound to four cysteine residues, to more complex ferredoxins, cuboidal three iron-four sulfide clusters, and cubane four iron-four sulfide clusters. Biosynthesis of Fe/S clusters in eukaryotic cells takes place in the mitochondria. Fe/S clusters are first assembled by the iron-sulfur cluster assembly machinery on the Isu1 or Isu2 scaffold proteins. These clusters are then released from Isu1 or Isu2, and transferred to apoproteins to become mature Fe/S proteins in the mitochondria, cytosol, or nucleus. Fe/S proteins in the mitochondria participate in various biological machinery including the respiratory chain (complexes I and II) and Fe/S cluster biogenesis (Yah1). Cytosolic Fe/S proteins are involved in processes such as amino acid synthesis (Ecm17), nucleotide metabolism (GPAT, XOR), regulation of iron uptake (IRP1), translation initiation (Rli1), and Fe/S protein biogenesis (Nbp35, Nar1). Nuclear Fe/S proteins have roles in DNA synthesis (Pri2) and DNA repair (Rad3).

There is also a heterogeneous group of iron-containing proteins that do not contain heme or Fe/S cluster (Crichton, 2009b). These proteins carry out various roles in most life-forms. Mononuclear non-heme iron enzymes can be classified into two different types based on their iron oxidation state (Crichton, 2009b; Solomon et al., 2003). Non-heme iron enzymes such as Rieske dioxygenases contain a Fe2+ center to activate molecular dioxygen, which are similar to heme enzymes. Conversely, other non-heme iron enzymes, such as lipoxygenases, contain an Fe3+ 42 active site to activate substrates that subsequently react with molecular dioxygen. Dinuclear non- heme iron enzymes are characterized by the presence of carboxylate-bridge diiron centres or polyiron aggregates (Crichton, 2009b). They have diverse functions, including iron storage by ferritins, formation of deoxyribonucleotides by ribonucleotide reductase, and transformation of methane by methane monooxygenase.

Table 1.3.1. Iron containing proteins (Crichton, 2009b).

Class Type Example Hemoproteins Oxygen carriers Hemoglobin, myoglobin Activators of molecular oxygen Cytochrome P450s, peroxidase Electron transport proteins Cytochromes a, b and c Iron‐sulfer Fe‐S cluster (bacteria) Rubredoxin proteins

Rhombic Fe2‐S2 cluster Rieske protein, ferredoxin

Cuboidal Fe3‐S4 cluster Ferredoxin, inactive aconitase

Cubane Fe4‐S4 cluster Active aconitase Other Mononuclear non‐heme iron enzymes Rieske, catechol dioxygenase Dinuclear non‐heme iron enzymes Ferritin, ribonucleotide reductase Iron transport proteins Transferrin, transferrin receptor

1.3.2 Iron uptake from the environment The Earth’s crust consists of approximately 5% of iron by weight, making iron the fourth most abundant element and the second most abundant metal (Lutgens and Tarbuck, 2000). Early life on Earth utilized iron extensively, which was rich in the anaerobic environment of the primodial ocean as soluble Fe2+ (Ilbert and Bonnefoy, 2013). The evolution of photosynthesis to produce oxygen by cyanobacteria and plants resulted in the oxygenation of the atmosphere and ocean between 2.4 billion to 542 million years ago. The oxygenation events permanently altered the redox states and solubility of iron in the biosphere. Although Fe2+ is soluble at a wide range of pH, Fe3+ undergoes hydrolytic deprotonations with water molecules at a pH higher than 2.0 to form insoluble ferric hydroxide and subsequently iron polynuclear complex (Aisen et al., 2001). Precipitation of Fe3+ further enhances auto-oxidation of Fe2+ by mass action under aerobic aqueous conditions, resulting in the removal of soluble iron from the solution.

After the biosphere became oxygenated, soluble iron in the environment was no longer readily available, and organisms needed to develop various mechanisms to extract and convert iron into a bioavailable form. Microbes synthesize a class of low weight iron complexing molecules known as siderophores (Crichton, 2009c). Natural siderophores such as ferrichrome and albomycin contain hard oxygen donor atoms that can form stable complexes with Fe3+. The Fe3+- siderophore complex is then transported into the cytoplasm by active transport that requires the hydrolysis of ATP. The Fe3+ is then reduced to Fe2+ by ferrireductase and released from the complex in the cytoplasm. Many bacterial pathogens can also transport or bind mammalian iron complexes such as heme, transferrin, and lactoferrin. Furthermore, bacteria such as E. Coli has a transport system for the uptake of Fe2+ in anaerobic conditions.

Iron is a common limiting nutrient for plants that cannot be easily supplemented by fertilizer because of the insolubility of iron hydroides (Crichton, 2009c). In iron sufficient conditions, Fe3+ is reduced to Fe2+ by ferric reductase at the root and transported through the plasma membrane. However, the concentration of bioavailable iron is at least six orders of magnitude below the level required for optimal growth, with the exception of flooded or waterlogged soil in which the acidic and anaerobic conditions allow the existence of Fe2+. Plants under the stress of iron deficiency employ two main strategies to increase iron uptake. Strategy I plants, including dicotyledons and non-graminaceous monocotyledons, release protons, iron reductants and Fe3+ 44 chelators into the rhizosphere. Iron chelation and low pH at the rhizosphere increase iron solubility and availability. The available iron is then reduced to Fe2+ and transported into the symplasm of epidermal cells at the roots by the ferrous transporter IRT1. Conversely, gramminaceous plants increase iron uptake at the root using a chelation-based approach (strategy II) by the release of phytosiderophores. Phytosiderophores are able to chelate poorly soluble iron by the formation of Fe3+-Phytosiderophore complexes, which are then transported into the symplasm by the YS1 transporter.

Mammals, including humans, obtain iron solely from the diet (Crichton, 2009g). Dietary iron can be classified based on the presence of heme. Absorption of non-heme and heme iron primarily takes place in the intestinal lumen by the duodenal enterocytes (Figure 1.3.2). In a standard western diet, approximately 80-90% of the dietary iron is non-heme iron, most of which is inorganic iron from plants (Crichton, 2009g). The absorption of non-heme iron can be enhanced by meat and organic acids, especially ascorbic acid. Ascorbic acid can increase the solubility of iron by reducing Fe3+ to Fe2+. It can also chelate Fe3+ from ferric chloride to form a soluble complex in the stomach. The complex remains soluble in the gastrointestinal tract for further processing and absorption. Conversely, non-heme iron can form complexes with a number of plant metabolites or tissues ranging from polyphenols to dietary fiber, thereby interfering with the absorption of iron. In adult, transportation of inorganic iron into enterocytes is mediated by the divalent metal ion transporter DMT1, which only transports Fe2+ but not Fe3+. Moreover, the high pH condition in the lumen does not favor the presence of soluble Fe3+. Therefore, inorganic iron must be reduced to Fe2+ by reducing agents such as ascorbic acid or apical membrane ferrireductases such as duodenal cytochrome B (DcytB) before absorption can take place. Absorbed Fe2+ enters the labile iron pool (LIP) of enterocytes and is subsequently transferred across the basolateral membrane by ferroportin (FPN). The basolateral membrane also expresses the ferroxidase hephaestin (Heph), which oxidizes Fe2+ into Fe3+ and facilitates the uptake of iron by apotransferrin. Each transferrin can bind to two Fe3+ and serve as the main transporter of iron in the circulation. Iron exportation from FPN can be directly blocked by hepcidin (Nemeth et al., 2004). Blockage by hepcidin retains the iron within the enterocyte for storage in ferritin. The stored iron may be lost when the enterocyte is shed into the lumen and excreted. Expression of FPN at a transcriptional and translational level is also subjected to iron dependent regulation.

Dietary ferritin also contains large amounts of non-heme iron. Although the mechanism of ferritin absorption remains unclear, it has been suggested that soybean ferritin is internalized into intestinal epithelial cells by endocytosis (Fuqua et al., 2012; San Martin et al., 2008). The iron from the absorbed ferritin is then reduced and enters the cytosolic LIP.

Heme iron is derived from hemoglobin and myoglobin associated with meat consumption (Crichton, 2009g; West and Oates, 2008). The absorption of heme iron is 6-7 times more efficient than that of non-heme iron. In addition, the inhibitors of non-heme iron absorption do not interfere with the absorption of heme iron. Furthermore, humans are more adapted to absorb heme iron than other species such as rodents (Cao et al., 2014). Therefore, even though heme iron only accounts for 10-20% of total dietary iron, it may contribute up to 50% of total iron absorption in humans. Monomeric heme is released from hemoproteins by proteolysis and absorbed as metalloporphyrin into enterocytes. Heme may be imported by transporter or receptor-mediated endocytosis, but the exact mechanism remains elusive (Fuqua et al., 2012; Korolnek and Hamza, 2014). Heme carrier protein 1 (HCP1) is a putative heme transporter, but it has low affinity for heme and may only have a minor role in enterocyte heme importation (Le Blanc et al., 2012). Heme oxygenase degrades the absorbed heme and releases Fe2+ to the cytosolic LIP. Moreover, intact heme may be exported by various mechanisms, including export into the circulation by the feline leukemia virus subgroup C receptor (FLVCR), export into the circulation by transcytosis (similar to vitamin B12), or export into the lymphatic system instead of the blood circulation (similar to cholesterol) (Khan and Quigley, 2013; Korolnek and Hamza, 2014). Unbound heme in the circulation is harmful and needs to be scavenged by high-density lipoprotein, low-density lipoprotein, hemopexin and albumin (Ascenzi et al., 2005; Crichton, 2009d).

Fe3+ 2+ Fe3+ Fe2+ Fe Ferritin Fe3+

Fe2+ DMT1 HCP1? DcytB

Fe2+ Endocytosis? Fe2+ Ferritin HO Fe2+

2+ Fe Fe2+ Nucleus Fe2+

Heph FLVCR(?) FPN1 Fe3+ Fe3+ Fe3+ Fe2+ Fe2+ Fe2+

Bloodstream

Apotransferrin Fe2+ Heme Biliverdin

Fe3+ Fe3+ Diferric Transferrin Hemopexin / Hepcidin albumin

Figure 1.3.2. Established and putative iron absorption pathways in the intestinal enterocyte (West and Oates, 2008; Crichton, 2009g). HCP1 – heme carrier protein 1, HO – heme oxygenase, DMT1 – divalent metal transporter 1, FLVCR – Feline leukemia virus subgroup C receptor‐related protein 1, Heph – hephaestin, FPN1 – ferroportin 1.

1.3.3 Iron distribution in humans Total body iron content in adult humans is about 3-5 g, which corresponds to 40-50 mg Fe/kg body weight (Abbaspour et al., 2014; Crichton, 2009g) (Figure 1.3.3). Women typically have lower iron content than men, and iron level in women fluctuates more than in men due to the menstrual cycle. The majority of the body iron (75%) is utilized as hemoglobin by circulating and developing erythrocytes for oxygen transportation. Around 5-15% of the body iron is used by other cells for iron-containing proteins such as myoglobin and Fe/S proteins. Unutilized iron (10-20%) is stored in the liver, spleen, bone marrow and muscles as ferritin and hemosiderin. At any given time, only a tiny fraction of total body iron (around 3 mg) is involved in the transportation of iron by transferrin in the circulation and extracellular fluids. Most of the iron in transit (about 22 mg/day) originates from erythrocyte recycling or is destined for erythropoiesis. Moreover, iron is highly conserved and there is no physiological mechanism to actively excrete iron from the body, with the exception of menstruation, pregnancy, child birth, and bleeding. Cells shedding from epithelial surfaces such as skin and gastrointestinal tract contribute to small amount of iron loss at about 1-2 mg/day. The Western diet typically contains 10-20 mg of iron per day. Only 1-2 mg/day of dietary iron is absorbed to compensate for normal iron loss, even though the efficiency of iron absorption by enterocytes can be boosted by 35% when necessary.

1.3.4 Iron processing in mammalian cells Most cells acquire iron from the plasma iron pool which essentially consists of iron loaded transferrin (Crichton, 2009d) (Figure 1.3.4a). Holotransferrin binds to the transferrin receptor (TFR) on the cell surface and is internalized into the cell as a vesicle via clathrin-mediated endocytosis. The vesicle contents are delivered to the endosome and the Fe3+ is released under mildly acidic conditions. The iron is reduced to Fe2+ by ferrireductases such as Steap3 in erythroid precursors and transported into the cytosol by DMT1. The Fe2+ contributes to the cytosolic LIP for storage in ferritin or further processing to synthesize cofactors such as heme and Fe/S clusters, while apotransferrin dissociates from the TFR and is released back into the circulation.

All cells eventually die and disintegrate. Iron that is stored or utilized in these cells must be rendered safe and recycled. Erythrocytes have a short lifespan at about 120 days and contain large amount of iron in the form of hemoglobin (Beaumont, 2010; Crichton, 2009d). Therefore, proper processing of senescent erythrocytes and their iron content is a crucial part of iron recycling. The majority of senescent erythrocytes are cleared in the spleen by tissue macrophages, also referred to as the reticuloendothelial system (Figure 1.3.4b). Tissue macrophages are able to recognize age-related modifications of erythrocytes such as shape transformation and conformational change of surface antigens. These aged erythrocytes are then engulfed by phagocytosis into phagosomes. Once internalized, the phagosomes fuse with other vesicles such as endosomes and lysosomes to from phagolysosomes. Phagolysosomes contain a number of proteins that can degrade the erythrocytes and their cellular components. Heme is released from hemoglobin after proteolysis. Catalysis of heme into carbon monoxide, biliverdin and Fe2+ is mediated by heme oxygenase 1 (HO-1) or NADPH-cytochrome c reductase. Biliverdin is further reduced to bilirubin by biliverdin reductase. Disintegration of erythrocytes may also occur outside of the reticuloendothelial system, which results in the release of hemoglobin into the circulation. Free hemoglobin in the plasma is toxic and must be rendered safe by haptoglobin through the formation of a hemoglobin-haptoglobin complex. The complex is then recognized by CD163 on tissue macrophages and taken up for clearance by receptor- mediated endocytosis (Ascenzi et al., 2005). Imported iron is transported into the cytosol by

DMT1 as part of the cytosolic LIP, which can be utilized by the host cell, stored in ferritin or hemosiderin, or exported out of the cell to transferrin by FPN.

Unutilized iron is primarily stored in hepatocytes within the liver (Crichton, 2009d; Takami and Sakaida, 2011) (Figure 1.3.4c). Hepatocytes import a large portion of iron in the form of holotransferrin via both TFR1 and TFR2. Although the affinity of TFR2 to bind holotransferrin is much lower than TFR1, TFR1 has a lower capacity to transport holotransferrin and its expression is subject to iron dependent regulation. Therefore, TFR1 is an inducible holotransferrin importer when the iron level is low. Hepatocytes can also import nontransferrin bound iron (NTBI) in the plasma that appears when the iron binding capacity of transferrin is exceeded due to iron overload. A number of proteins have been proposed to be NTBI importers, which include ZIP14, ZIP8, and L-type voltage-gated calcium channels. DMT1 is also implicated, but it has recently been shown to be dispensable based on a mouse model with inactivated DMT1 (Wang and Knutson, 2013). Furthermore, hepatocytes can import and degrade hemoglobin-haptoglobin and heme-hemopexin complexes via receptor-mediated endocytosis (Ascenzi et al., 2005). Hepatic iron is utilized by iron-containing proteins such as cytochrome P450s, stored in ferritin, or exported to transferrin via FPN (Crichton, 2009d).

1.3.5 Regulation of iron balance The supply of iron must be tightly regulated to maintain normal physiological functions without causing cellular and organ damage. There exist a number of factors that are responsible for the regulation of iron balance at both a cellular and systemic level in mammals.

Iron storage is a crucial aspect of iron homeostasis to minimize the impact of fluctuations in iron supply from the environment. To maintain a steady level of iron, living organisms must be able to release stored iron during iron deficiency and stockpile excess iron during iron sufficiency in a timely manner. The unutilized iron in normal human subjects is mostly stored in hepatocytes and the reticuloendothelial system (Crichton, 2009e; Theil, 2013). Cellular iron storage consists of ferritin or hemosiderin. Most mammalian cell types can store iron in ferritin, whereas only the reticuloendothelial system can store iron in hemosiderin under normal circumstances. There are two types of ferritin subunits: heavy (H) chain ferritin has a ferroxidase centre that can oxidize Fe2+ to Fe3+; light (L) chain ferritin is supposed to be responsible for iron nucleation in the ferritin core. Typical ferritin is a hollow protein shell comprised of 24 H and L subunits and is capable of storing up to 4500 atoms of iron in a water soluble, non-toxic, and bioavilable form that is similar to ferrihydrite (5Fe2O3·9H2O). The ratio between H and L chains varies in different organs and cell types. The H chain is the predominant subunit in the heart, while the L chain is the predominant subunit in the liver. Ferritin can also be actively secreted or leaked into the serum. Serum ferritin largely comprises L chains ferritin and accounts for only a small proportion of body iron (Rosario et al., 2013). Conversely, it is a useful marker to assess total body iron, even though its level can be influenced by non-iron factors such as inflammation and infections (Jensen, 2004). Deposition of iron into ferritin is a multi-step process that involves the uptake of Fe2+ into the protein shell through hydrophilic channels, oxidation of Fe2+ to Fe3+ by the H chain ferroxidase, and Fe3+ nucleation on the interior surface of the L chain to form the ferrihydrite mineral core. The molecular mechanisms of in vivo iron release from ferritin are not well established. It has been proposed that ferritin iron mobilization involves iron release upon ferritin degradation in the lysosome, or the iron may exit the ferritin after reduction and chelation. Lysosomal degradation of ferritin also leads to the formation of hemosiderin. Hemosiderin is a heterogeneous iron complex which consists of varying amounts of ferritin, denatured ferritin, and other materials. It stores a large amount of iron in a ferrihydrite-like core,

53 but the stored iron is neither soluble nor readily bioavailable. The exact nature of hemosiderin is not well characterized despite being identified as the first form of iron storage in living organisms over a century ago.

The expression of genes related to cellular iron balance is more often regulated at the translational level instead of the transcription level (Andrews and Schmidt, 2007; Crichton, 2009f) (Figure 1.3.5a). Translational regulation of many of the “iron genes” depends on the presence of the iron response/regulatory element (IRE) in the mRNA, which is a conserved short nucleotide RNA sequence with a stem-loop structure. Iron response/regulatory proteins (IRPs) 1 and 2 can recognize and bind to IREs in response to iron deficiency, thereby modulating the translation of the corresponding mRNA. IRP-1 and IRP-2 are closely related to each other with sequence identity close to 60%, but IRP-2 is proposed to have a more active role in iron- dependent translational regulation. At low oxygen concentrations in typical mammalian tissues, IRP-2 has a greater affinity for IREs than IRP-1. Also, IRP-1 binding of IREs requires lower iron concentration than IRP-2, suggesting that IRP-1 has an active role in iron homeostasis only when cellular iron is more severely depleted. Furthermore, IRP-2 can be up-regulated to compensate for IRP-1 when IRP-1 is genetically ablated, but IRP-1 cannot compensate for the loss of IRP-2. In iron depleted cells, IRPs prevent translation initiation when they bind to IREs that are located at the 5’ untranslated region (UTR) of mRNAs for genes such as ferritin and FPN. IRPs also bind to the 3’UTR of other mRNAs for genes such as DMT1 and TFR1, which enable continual translation by protecting the mRNA from endonucleolytic cleavage and degradation. The changes in gene expression increase the cellular iron concentration by increasing iron import while suppressing iron export and storage. Conversely, excess intracellular iron prevents IRPs from binding to IREs, which enhances translation initiation and increase degradation of mRNAs with IREs at their 5’ and 3’ UTR, respectively. The amount of cytosolic LIP is decreased as a result of decreased iron import and increased iron storage and export. When IRP2 is not bound to IREs during iron sufficiency, it is degraded by the ubiquitin/proteasome system. IRP1 is not degraded during iron sufficiency. Instead, excess iron completes the formation of a 4Fe-4S cluster within IRP1, which transforms IRP1 into a cytosolic aconitase. This transformation enables IRP1 to covert citrate into isocitrate, possibly for the cytosolic production of NADPH.

Under normal circumstances, systemic iron balance is primarily maintained by the iron- regulatory hormone hepcidin (Andrews and Schmidt, 2007; Crichton, 2009h; Rochette et al., 2014; Weiss, 2010) (Figure 1.3.5b). Active hepcidin is a 25 amino acid peptide that is mainly synthesized by the liver and secreted into the circulation. Non-hepatic secretion of hepcidin occurs in macrophages, adipocytes, heart, and stomach, which may be important in the regulation of local iron flux (Rochette et al., 2014). Truncated hepcidin peptides with less than 25 amino acids also exist, but these peptides do not participate in iron regulation. In addition, hepcidin possesses antimicrobial activity possibly by depriving pathogens of iron required for survival (Dao and Meydani, 2013). Serum hepcidin regulates iron balance by binding to FPN on the cell surface (Nemeth et al., 2004; Rochette et al., 2014). The hepcidin/FPN complex is internalized and degraded, thereby decreasing the concentration of membrane FPN and cellular iron efflux. The expression of hepcidin is regulated by the level of transferrin in the serum, which is the main carrier of bioavailable iron in the body (Rochette et al., 2014). During iron sufficiency, holotransferrin binds to TFR1 and displaces the human hemochromatosis (Hfe) protein. The Hfe protein forms a membrane-associated protein complex with TFR2 and hemojuvelin, and then activates the transcription of hepcidin via the ERK1-2 or P38 MAP-kinase signaling pathways (D'Alessio et al., 2012; Rochette et al., 2014). Hemojuvelin is also the co-receptor for the bone morphogenetic protein (BMP) receptor. The receptor complex is activated by BMP6 in response to high liver iron content, which triggers hepcidin transcription via the SMAD pathway (Core et al., 2014). Cytokines such as interleukin-6 can induce hepcidin transcription through STAT3 in response to inflammation (Andrews and Schmidt, 2007). Moreover, hepcidin expression can be suppressed by the recently identified hormone erythroferrone, which is produced by erythroblasts in response to erythropoietin during stress erythropoiesis (Kautz et al., 2014). Suppression of hepcidin by erythroferrone may explain the development of iron overload in some MDS patients even before they become transfusion dependent (Shenoy et al., 2014). Other physiological or pathological conditions can also modulate hepcidin expression, including hypoxia, renal failure, and vitamin D deficiency (Rochette et al., 2014).

IRP

Fe Fe Fe Fe

(i) AUG Protein coding AAAAAAA Translation IRE

Translation (ii) AUG Protein coding AAAAAAA inhibition

Fe Fe Fe Fe Susceptible to (iii) AUG Protein coding AAAAAAA endonucleases

(iv) AUG Protein coding AAAAAAA Translation

Figure 1.3.5a. Translation regulation by Iron response/regulatory proteins (IRPs). IRPs do not bind to mRNA when cellular iron is sufficient (I, iii). During iron depletion, IRPs bind to mRNA with iron response/regulatory element (IRE) at the 5’ (ii) (e.g. ferritin) or 3’ (iv) (e.g. TFR) untranslated region (UTR), thereby inducing translation inhibition or promotion, respectively (Crichton, 2009f).

Degradaon, ↓FPN1 Excretion Hepatocytes / Macrophages / Enterocytes

2+ Fe Kidney

α2M FPN1

2+ Fe ‐ Fe3+ Fe3+ 3+

Fe Excess Holo‐TF 3+ Inflammation Fe iron IL‐6 BMPs Basal Infection

BMP‐RHJVTFR HFE IL‐6R Erythroferrone

‐ Suppress SMAD‐P ERK1/2 STAT3

Enhance + + +

HAMP Hepcidin Hepatocytes

Figure 1.3.5b. Regulation of iron balance by hepcidin (Andrews and Schmidt, 2007; Crichton, 2009h; Kautz et al., 2014; Rochette et al., 2014; Weiss, 2010). α2M –activatedα2‐macroglobulin, BMPs – bone morphogenetic proteins, BMP‐R– BMP receptor, HFE – human hemochromatosis protein, IL‐6–interleukin6,IL‐6R – IL‐6 receptor, SMAD‐P – phosphorylated SMAD complex, ERK1/2 – extracellular signal‐regulated kinases 1/2, STAT3 – signal transducer and activator of transcription 3.

1.3.6 Disruption of iron balance in humans Organisms can cope with temporary iron loss or gain by utilizing from or contributing to the body iron storage, respectively. However, various physiological functions can be disrupted when the body fails to maintain the total body iron at a specific range. Insufficient iron intake or excess iron loss induces iron deficiency, while uncontrolled iron intake induces iron overload. Dysregulation of iron status can be caused by acquired or genetic factors. In general, iron dysregulation is not immediately fatal. However, if unattended, prolonged iron dysregulation has a negative impact on health and causes life-threatening conditions.

1.3.6.1 Iron deficiency Anemia is the most common physiological consequence of iron deficiency (Crichton, 2009h). Anemia occurs when the iron requirement for erythropoiesis exceeds the iron supply and dietary iron intake, causing insufficient production of erythrocytes and hemoglobin to transport oxygen in the circulation. Iron deficiency is often the consequence of a poor diet, which includes the lack of vitamin C and heme protein. Conversely, iron absorption can be impaired by a high content of fiber, phytates, phosphates, and calcium. Women are at higher risk for acquiring iron deficiency anemia due to menstruation, pregnancy, and breastfeeding. In addition, the prevalence of iron deficiency is increased for women who are below the poverty level, minorities, and have multiple children. Chronic blood loss is another major cause of iron deficiency, which can be caused by peptic ulcer, drugs, and parasite infestation in the intestine. Iron deficiency causes anemia by a multi-step process in which all storage iron is first mobilized for erythropoiesis. Patients become anemic when the storage iron is exhausted and an appropriate hemoglobin level cannot be maintained. The diagnostic criteria for iron deficiency anemia include decreased serum ferritin, low serum iron, high serum transferrin, increased soluble transferrin in plasma, low mean corpuscular volume (MCV), low mean corpuscular hemoglobin (MCH), and appearance of pathological erythroblasts and erythrocytes. Anemia as the consequence of dietary iron deficiency can be remedied by oral iron therapy using ferrous sulfate, or parenteral iron therapy such as intravenous or intramuscular injection of iron dextran. Management of iron deficiency anemia due to bleeding is more complicated and likely requires treatment of the underlying cause. Moreover, chronic infection or illness can induce anemia via functional iron deficiency, in

58 which inflammation-related cytokines up-regulate hepcidin expression and decrease cellular iron export through FPN.

1.3.6.2 Primary iron overload Iron overload describes a condition in which the accumulation of iron exceeds the body’s capacity to store iron at a safe level. Excess iron is deposited in various tissues to induce progressive cell and tissue damage, which eventually causes organ failure and other pathological states. Iron accumulation due to heritable mutations of genes related to the iron regulation system is commonly referred to as primary iron overload, or hereditary hemochromatosis. Most known mutations for primary iron overload involve genes that participate in cellular iron export by FPN. Hfe-related or type 1 hemochromatosis (HH) is the most frequent type of primary iron overload (Crichton, 2009h; Siddique and Kowdley, 2012). The prevalence of Hfe gene mutations C282Y and H63D is high among Northern Europeans and Europeans in general, respectively. HH is an autosomal recessive disorder of iron metabolism in which the defective Hfe protein has diminished ability to up-regulate the expression of hepcidin. The clinical manifestation of HH is a prolonged process that can take four to five decades. Suppression of hepcidin expression allows excessive dietary iron absorption, which increases transferrin saturation. Accumulation of total body iron leads to iron overload and is marked by increased transferrin saturation and serum ferritin. Continual iron overload causes symptoms such as arthralgias, and eventually organ damage with life-threatening disorders such as cardiomyopathy, liver cirrhosis, and hepatocellular carcinoma. The severity and latency of iron-related symptoms can be modified by multiple factors, including co-mutations of other genes, male gender, increased alcohol consumption, and excessive dietary heme consumption.

Three major types of non-Hfe related hemochromatosis have also been identified with relatively low frequency compare to that of HH (Crichton, 2009h; Siddique and Kowdley, 2012). Juvenile hemochromatosis (JH) involves autosomal recessive mutations of the hemojuvelin (type 2A) or hepcidin (type 2B) genes. Similar to HH, mutations for JH also cause iron overload by inhibiting hepcidin expression. However, the symptoms of JH are more severe and can occur by the third decade of life, which is earlier than HH. Type 3 hemochromatosis involves autosomal recessive mutations of TFR2 marked by increased transferrin saturation and serum ferritin. Type 4 hemochromatosis involves the FPN gene, which can be caused by gain-of-function or loss-of- 59 function mutations. Gain-of-function FPN mutants are resistant to internalization and degradation by hepcidin, resulting in enhanced cellular iron exportation, low transferrin saturation, and iron accumulation in hepatocytes. Conversely, loss-of-function mutations abolish FPN activity, causing impairment of iron exportation and retention of iron in the reticuloendothelial system. In addition, mild anemia can occur because of insufficient iron supply for erythropoiesis from macrophages.

Initial diagnosis of primary iron overload requires the evaluation of transferrin saturation and serum ferritin by blood test (Crichton, 2009h; Kanwar and Kowdley, 2013). Serum ferritin can be directly measured, while transferrin saturation is calculated by dividing serum iron by total iron-binding capacity. Patients are suspected to have hemochromatosis and require genetic testing when their transferrin saturation is >45% and serum ferritin is >300 µg/L for men or >200 µg/L for women. Precaution must be taken when interpreting the level serum ferritin, since it can be influenced by non-iron factors such as infection and chronic inflammation. Genetic testing is useful to identify common alleles such as Hfe C282Y and H63D for Northern Europeans. In addition, other factors such as male gender and alcohol may modify the pathophysiology of hemochromatosis and therefore must be considered. Liver biopsy is the most accurate method to evaluate hepatic iron concentration and liver cirrhosis. However, it is no longer the preferred method in the context of hemochromatosis. Liver iron concentration (LIC) is now estimated by T2* MRI, which is a non-invasive method based on the altered relaxation time of protons in iron- loaded tissues.

Iron removal is the logical approach for the treatment of iron overload, and can be achieved by therapeutic phlebotomy (Crichton, 2009h; Kanwar and Kowdley, 2013). Phlebotomy is a direct method to remove iron since each unit of blood contains about 250 mg of iron in the form of heme. Blood removal also stimulates erythropoiesis, which increases iron utilization and decreases storage. The interval between phlebotomies varies greatly among individual patients, and the therapeutic target is to maintain serum ferritin below 50 µg/L range (Assi and Baz, 2014; Leitman, 2013). Phlebotomy can lower the morbidity and mortality rates for hereditary hemochromatosis especially before the development of iron-induced cirrhosis or diabetes. It is also well established, safe, inexpensive, and may even be beneficial to the society as a source of blood donation (Leitman, 2013). Phlebotomy is recommended as the first line therapy for most 60 types of hereditary hemochromatosis. However, phlebotomy may not be suitable for type 4 hemochromatosis patients who carry the loss-of-function FPN mutations and may therefore be at risk of anemia. Other methods such as ICT may be used only when phlebotomy is non-tolerated or contraindicated (Crichton, 2009h).

1.3.6.3 Secondary iron overload The term secondary, or acquired, iron overload is usually used to describe a condition in which excessive iron accumulation in the body is not the direct consequence of heritable defects in genes related to the iron regulation system (Crichton, 2009h; Siddique and Kowdley, 2012). Secondary iron overload can be a multi-step process triggered by chronic liver diseases including hepatitis C virus (HCV) infection, alcoholic liver disease, and non-alcoholic fatty liver disease. The stresses related to these liver conditions can initially alter the expression of iron regulatory genes, such as down-regulation of hepcidin through the inhibition of C/EBPα by the increased production of ROS. In addition, hepatocytes and the liver can be damaged under stress, resulting in the depression of hepcidin level in the serum and the release of iron from necrotic hepatocytes. Iron homeostasis is disrupted as a result, which causes iron accumulation in the hepatocytes and eventually in other tissues. Excess iron and the original stress can act in a synergistic manner to cause further liver injury and iron overload. Furthermore, genetic predispositions such as Hfe mutations can modify the risk of iron overload. Genetic predisposition is also a modifier for African iron overload, in which Africans with mutations of the FPN gene are more vulnerable to acquire iron overload as a result of heavy and continuous consumption of iron rich traditional home brewed beer.

Secondary iron overload can also be an unintended consequence of chronic RBC transfusion for the treatment of ineffective erythropoiesis (Crichton, 2009h; Siddique and Kowdley, 2012). Chronic anemia due to ineffective erythropoiesis is a pathophysiological feature of numerous genetic or acquired hematological disorders. The thalassemia syndromes are the most common and documented among these disorders (Munker et al., 2000a; Siddique and Kowdley, 2012). Normal hemoglobin in adults consists of two α- and β-chains each. The ratio between the α- and β-chains is crucial for proper oxygen transportation by erythrocytes. In thalassemias, heritable mutations of genes that encode the α- or β-chains reduce or abolish the production of the corresponding globin chains, thereby preventing the formation of normal hemoglobin. The 61 remaining chains aggregate and become cytotoxic to developing erythrocytes, which causes ineffective erythropoiesis and other complications. In general, the severity of thalassemia correlates with the nature and the number of mutated alleles (Muncie and Campbell, 2009). These mutations are carried by tens of millions of people especially among African, Mediterranean, and Southeast Asian populations. The high prevalence may be the consequence of natural selection for the mutants’ ability to provide resistance against malaria (Flint et al., 1998; Muncie and Campbell, 2009). The α-chain is encoded by two separate genes in humans. Gene deletions result in α-thalassemias with progressive severity. Deletion of one α-chain gene has minimal effect on hemoglobin synthesis, whereas complete deletion results in stillborn with hydrops fetalis. The β-chain is encoded by a single gene with over 200 mutations that can suppress its production to varying degrees. One fully functional β-chain allele is sufficient to sustain hemoglobin production with minimal impact to erythropoiesis. Combinations of defective or non-functional β-chain alleles result in β-thalassemia intermediate, with more severe but not life-threatening phenotype. Patients with β-thalassemia major have two non-functional β-chain alleles. Clinical symptoms of β-thalassemia major can be detected by six months of age. If these patients are not properly treated, they will develop severe microcytic anemia, growth retardation, splenomegaly, high cardiac output congestive heart failure and mortality by the second decade of life (Hoffman, 2000). Furthermore, anemia can be aggravated by splenomegaly for patients with symptomatic thalassemias. Splenectomy can be performed for these patients, but a chronic hypercoagulable state may develop after the procedure (Eldor and Rachmilewitz, 2002).

Patients with β-thalassemia major require regular RBC transfusions to maintain a minimum hemoglobin level of at least 95 g/L and a post-transfusion level of about 140 g/L (Galanello and Origa, 2010; Muncie and Campbell, 2009; Musallam et al., 2013). Transfusions start early in life and these patients are transfusion-dependent for the rest of their lives, unless the underlying genetic defect is corrected by methods such as HSCT. Patients with other types of thalassemias also develop mild to moderate microcytic anemia and may require occasional RBC transfusions to maintain normal life and growth development. Transfusion frequency for β-thalassemia major is usually two to four weeks, while the amount of RBC to be transfused is calculated based on the patient’s weight, targeted hemoglobin level, and other considerations. RBC transfusion is intended to prevent growth impairment, organ damage, maintain adequate quality of life and

62 extend lifespan (Hoffman, 2000). Furthermore, hypertransfusion is often beneficial in that it can suppress extramedullary erythropoiesis, decrease intestinal iron absorption, and alleviate splenomegaly and hepatomegaly (Hoffman, 2000; Pippard and Weatherall, 1984).

Despite the benefits, RBC transfusion is associated with a number of possible complications including an increased risk of infection and alloimmunization to transfused erythrocytes (Galanello and Origa, 2010; Schreiber et al., 1996; Singer et al., 2000). Secondary iron overload is an unavoidable consequence of chronic RBC transfusion for β-thalassemia major due to the massive infusion of iron via hemoglobin. Moreover, other thalassemia patients who are not transfusion-dependent may also develop iron overload (Taher et al., 2013). Anemia and hypoxia as a consequence of ineffective erythropoiesis likely suppress hepcidin expression, which increases intestinal iron absorption and allows iron accumulation. The progression of iron overload is slower for these patients when compared to transfusion-dependent patients, but the health risk remains significant. Regular transfusion of as few as 10 units of PRBC can cause complete saturation of transferrin (Prati, 2000). Further iron loading by transfusion overwhelms the body’s capacity to store iron safely. Iron overload leads to the formation of NTBI, which is the main culprit responsible for cellular and organ damage (McLaren et al., 1983; Porter, 2009; Prati, 2000). For patients with β-thalassemia major who require chronic transfusion early in life, liver fibrosis can occur as early as 3 years due to hepatic iron accumulation. Without intervention, liver fibrosis progresses into cirrhosis and possibly hepatocellular carcinoma later in life. Cardiac iron overloading occurs after 70-100 units of PRBC for thalassemia patients around the age of 10 years. Impairment of cardiac function is the consequence of iron deposition in the cardiac tissue, as well as other factors such as antioxidant levels and immunological modifications. Cardiac dysfunction progresses slowly and may become symptomatic around the third decade of the patient’s life. Fatality due to cardiac failure was the leading cause of death before ICT was available. Furthermore, iron overload in other tissues can cause complications including skin hyperpigmentation, hypogonadism, diabetes, hypothyroidism, and hypoparathyroidism.

ICT is recommended for the treatment of secondary iron overload, especially for patients who are anemic and require chronic transfusion (Musallam et al., 2013; Porter, 2001; Thuret, 2013). Correct ICT dosing requires accurate assessment of iron levels at appropriate intervals. The level 63 of transfusion-related iron overload in thalassemia is usually assessed by serum ferritin, hepatic and cardiac iron concentration by MRI (Thuret, 2013). Serum ferritin is an indirect estimation of iron stores and it can be influenced by non-iron factors such as infection and inflammation. Nonetheless, it is the standard clinical procedure to assess iron overload since the method is cheap, convenient, widely available, and may be the only available method for developing countries. In general, patients are considered to have iron overload and severe iron overload when their serum ferritin levels reach 1000 and 2500 µg/L, respectively. Serial measurements of serum ferritin every three months are recommended to identify the trend of iron status and adjust ICT dosing. In addition, the accuracy of serum ferritin measurement can be enhanced when calibrated against LIC values. LIC and cardiac iron are measured based on T2 or T2* relaxation time using MRI, which is non-invasive with high international reproducibility. Annual assessments of hepatic and cardiac iron are generally recommended, but more frequent measurement may be required for patients with considerable iron loading. Initiation of ICT is recommended for patients who have received 10 or more transfusions or with serum ferritin >1000 µg/L. ICT may also be necessary when LIC is higher than the normal liver iron storing capacity at 3 mg/g dw to avoid liver iron overload and spread of iron to other organs. Alternatively, a more conservative threshold of 7 mg/g dw (roughly equivalent to 1000 µg/L serum ferritin) may be used to avoid the risk of adverse drug reaction to the iron chelator deferoxamine.

Deferoxamine, along with deferiprone and deferasirox, are major iron chelators for the treatment of secondary iron overload (Musallam et al., 2013; Porter, 2001; Thuret, 2013). Deferoxamine has been deployed for clinical use since the 1970s. Its use has dramatically reduced the rate of iron-related morbidity and mortality in patients with β-thalassemia major. However, the need for prolonged parenteral infusions leads to significant compliance issues. Also, deferoxamine is not able to completely prevent cardiac iron loading. Deferiprone is the first oral iron chelator for humans. Compared to deferoxamine, deferiprone is more effective at cardiac iron removal and more convenient to administer. Deferasirox is a new class of rationally designed oral iron chelator. The effectiveness of deferasirox to chelate iron diminishes with transfusion burden, and dosing regimens should be established based on transfusion rate, trends in serum ferritin, and other safety markers. Deferasirox can effectively remove cardiac iron, with mild to moderate side

64 effects including gastrointestinal, renal and hepatic impairments. Recent studies show that deferasirox has an acceptable safety profile and non-inferior efficacy when compared to deferoxamine. Therefore, adoption of deferasirox for ICT is expected to increase. Deferoxamine is the first-line therapy for young β-thalassemia major patients below the age of 2 for the US and 6 for the European Union, as the safety of the oral chelators has not been established in this age group. Deferiprone and deferasirox are the second-line therapy for young patients resistant to deferoxamine. Deferoxamine and deferasirox are generally recommended for children above 6- 10 years of age. Intensive deferoxamine therapy is required for severe iron overload without cardiac iron, while combined therapy using deferoxamine or deferasirox + deferiprone is recommended for patients with cardiac iron overload or dysfunction. To avoid drug toxicity, dose reduction of deferoxamine is recommended when serum ferritin drops below 1000 µg/L. The same recommendation also applies to deferasirox and deferiprone, which may be changed in the future, when their safety profile at such low a level of serum ferritin is better understood.

1.4 Iron overload and myelodysplastic syndromes

1.4.1 Non-transferrin bound iron and labile cell iron Iron overload is a serious concern for the majority of MDS patients who are anemic and require chronic RBC transfusion (Temraz et al., 2014). Transfusion-dependent MDS patients typically develop secondary iron overload after receiving 20 units of PRBC, which can occur within one year of chronic transfusion (Malcovati, 2009). In addition, the mean lifetime RBC transfusion reaches 160 units based on a survey of 292 transfusion-dependent MDS patients in Japan (in japan, one RBC unit is derived from 200 mL of whole blood) (Takatoku et al., 2007). Transfused erythrocytes are recycled by the reticuloendothelial system and the associated iron is released to erythroid progenitors for erythropoiesis or hepatocytes for storage (Mahesh et al., 2008). Continuous transfusions overwhelm the body’s capacity to store and transport iron, which leads to the existence of NTBI in the circulation (Brissot et al., 2012). Moreover, the level of NTBI in the circulation can be elevated in MDS even before the initiation of transfusion, which may be related to ineffective erythropoiesis and increased iron absorption (Cortelezzi et al., 2000). Similar to β-thalassemia major, NTBI is suspected to be the main culprit of cellular and organ damage behind transfusion-related iron overload in MDS (Brissot et al., 2012; Hider, 2002). The exact chemical nature of NTBI is not well established, but there are reports suggesting that it 65 may comprise a combination of oligomeric Fe3+ and Fe3+ complexes with citrate or albumin. Circulating NTBI does not bind with appropriate ligands, such as transferrin, heme, and ferritin. Also, a portion of NTBI in the plasma can participate in redox reactions and is called labile plasma iron (LPI). NTBI enters cells of various organs including the liver, heart, pancreas, and reticulocytes (Brissot et al., 2012). The influx of NTBI into the cell contributes to the cellular labile iron pool, which only constitutes a minor fraction of the total cell iron under normal circumstances (Brissot et al., 2012; Kakhlon and Cabantchik, 2002). Labile cell iron (LCI) is considered as the cellular form of NTBI that is not utilized by iron-containing proteins or sequestered by ferritin/hemosiderin, but is instead loosely associated with citrate or cellular proteins. Furthermore, excess LCI can enter the mitochondria and become non-chelatable by high-affinity iron chelators in the cytosol.

1.4.2 Pathophysiology of iron overload

1.4.2.1 Generation of reactive oxygen species by iron Toxicity of excess iron is thought to be mediated by the generation of reactive oxygen species (ROS) (Mahesh et al., 2008). ROS are chemically active oxygen-containing molecules that are primarily the by-products of aerobic respiration in the mitochondria (Kowaltowski et al., 2009). - Superoxide anion (•O2 ) constitutes the bulk of ROS generated in the mitochondria by complexes

I and III. Once generated, superoxide is transformed into hydrogen peroxide (H2O2) by superoxide dismutases (SODs). Hydrogen peroxide is a more stable form of ROS that is subsequently converted into water and rendered harmless by catalases and peroxidases. Non- mitochondrial superoxide can also be generated by NADPH oxidases and cytochrome P450s for various physiological processes including respiratory burst by neutrophils against pathogens, vascular formation and development, and regulation of kidney functions (Bae et al., 2011; Nunes et al., 2013; Sedeek et al., 2013; Zhou et al., 2013).

- 3+ 2+ A redox reaction occurs between •O2 and Fe to produce O2 and Fe (1.1) (Crichton, 2009a).

3+ - 2+ Fe + •O2 → Fe + O2 (1.1)

2+ 3+ Conversely, Fe can be oxidized into Fe by H2O2 via the Fenton reaction (1.2) (Fenton, 1894).

2+ 3+ - Fe + H2O2 → Fe + •OH + OH (1.2) 66

The sum of reactions (1.1) and (1.2) is known as the Haber-Weiss reaction (1.3) (Haber, 1934).

- - •O2 + H2O2 → O2 + •OH + OH (1.3)

The Haber-Weiss reaction describes the interaction between superoxide anion and hydrogen peroxide in the presence of catalytic quantities of soluble iron. One of the by-products of the Haber-Weiss reaction is the hydroxyl radical (•OH), which is also a type of ROS (Avery, 2011; - Cadet et al., 2012; Dixon and Stockwell, 2014). Unlike •O2 and H2O2, the hydroxyl radical cannot be enzymatically neutralized. The hydroxyl radical is highly reactive with a very short half-life and leads to higher level of oxidative stress. Excess circulating and cellular NTBI as a consequence of secondary iron overload increases the availability of iron for the generation of hydroxyl radical. Moreover, Fe2+ can catalyze the generation of lipid alkoxyl radicals via the Fenton reaction (Dixon and Stockwell, 2014).

1.4.2.2 Oxidative damage of macromolecules Hydroxyl and lipid radicals can damage and disable almost all types of macromolecules, including nucleic acids, lipids, and amino acids. Oxidative stress occurs when the cell is unable to readily detoxify the reactive radicals or repair the resulting damage. Iron-catalyzed oxidation of DNA is especially relevant because of the capability of iron to bind to the phosphodiester backbone and participate in the generation of hydroxyl radicals (Avery, 2011; Cadet et al., 2012). Hydroxyl radicals induce multiple forms of DNA damage, including DNA double strand breaks, single and tandem base lesions, and DNA-protein and interstrand cross-links. Direct induction of massive DNA damage by ROS results in the decrease of cell viability. Alternatively, ROS can trigger genomic instability and cell death by disrupting the functions of DNA repair proteins. Non-lethal amounts of oxidative stress may promote mutagenesis and therefore carcinogenesis instead of cell death. Furthermore, oxidative stress can induce epigenetic instability by altering DNA methylation and histone modification. ROS can alter the epigenome by dysregulating the activities of epigenetic modifiers such as polycomb repressive complexes and histone deactylase SIRT1 (Cencioni et al., 2013; Nishida and Kudo, 2013). ROS may also directly participate in the modification of DNA and histones by proposed nucleophilic mechanisms (Afanas'ev, 2014). Dysregulation of the epigenome is associated with altered expression of genes such as tumor suppressors, which may cause cell growth arrest, senescence, and carcinogenesis. Mitochondrial

DNA is more vulnerable to oxidative damage than nuclear DNA due to the lack of protective histones and higher level of respiratory ROS production (Shokolenko et al., 2014). Iron-induced oxidative damage of mitochondrial DNA can introduce mutations, alter gene expression and disrupt crucial mitochondrial functions such as the electron transport chain. These disruptions are associated with events outside of the mitochondria, including dysregulation of p53 and epigenetic control of nuclear DNA (Szumiel, 2014). In addition, excess iron can directly disrupt the production of the Fe/S cluster, thereby interfering with the synthesis of crucial iron- containing proteins (Levi and Rovida, 2009). Moreover, oxidative stress can impair cell integrity by the peroxidation of lipid membranes at various locations including the cell surface, lysosome, and nucleus (Brissot et al., 2012). Additionally, direct oxidation of low-density lipoprotein by lysosomal iron may play a role in atherosclerosis (Satchell and Leake, 2012).

1.4.2.3 Signaling pathways and gene regulation ROS are involved in a complicated web of signaling networks where their generation is regulated by multiple pathways, including increased ROS generation in response to tumor necrosis factor-α (TNF-α) and hypoxia inducible factors (HIFs), differential regulation of pro-oxidant and antioxidant genes by p53, inhibition of TNF-α and subsequent ROS generation by Bcl-2, modulation of mitochondrial ROS release to the cytosol by Romo1, and regulation of NADPH oxidases and cytochrome P450s (Bae et al., 2011). Consequently, increased generation of ROS by these pathways can enhance iron-induced oxidative damage. Conversely, ROS act as signaling molecules for other signaling pathways such as PTEN, PTP1B, MAPK and nuclear factor-κB (NF-κB) (Reczek and Chandel, 2014). Multiple models have been proposed regarding the mechanisms of ROS signal transduction, which range from oxidation of functional cysteine thiol groups in phosphatases by H2O2, to the transfer of oxidation from oxidized scavenger proteins to signaling proteins (Reczek and Chandel, 2014).

Excess iron is involved in the regulation of several signaling pathways through the generation of ROS (Templeton and Liu, 2003). An elevated level of non-heme iron in hepatic macrophages is sufficient to trigger oxidative activation of NF-κB, which causes TNF-α induction and leads to increased transcription of H chain ferritin. Iron can induce the expression of antioxidant enzymes such as glutathione peroxidase (GPX) to reduce the level of H2O2 and subsequent production of hydroxyl radicals. In dermal fibroblasts, iron-dependent ROS generation can up-regulate the 68 transcription of matrix-degrading metalloproteases MMP-1 and 3 via the activation of Jun N- terminal kinase 2 (JNK2) and c-Jun, which may help to explain the underlying mechanism of skin damage by ultraviolet B. Expression of pro-α2(I)-collagen in the liver is associated with tissue fibrosis and its expression can be up-regulated by iron without concurrent induction of oxidation-responsive genes c-fos and hsp70. Transcription of collagen α1(I) in the liver can also be up-regulated by iron as a result of lipid peroxidation and protein adduct formation, suggesting that iron may directly induce hepatic fibrosis as well as via oxidation.

The cytokine transforming growth factor β (TGF-β) is a key regulator of fibrotic gene expression that is up-regulated in response to iron loading in the liver (Templeton and Liu, 2003). On the other hand, transcription of TGF-β and collagen α1(I) is suppressed by iron loading in cultured cardiac cells. This discrepancy may be explained by iron-induced up-regulation of decorin in the cardiac cells. Decorin is a proteoglycan that influences fibrillogenesis and inhibits TGF-β. Inhibition of TGF-β reduces autoinduction and transcription of collagen α1(I). Iron depletion in these cells also reduces TGF-β activity and collagen expression, suggesting a specific range of iron concentration is required for the expression of these genes. Furthermore, decorin and TGF-β participate in other cellular processes including inflammation and the cell cycle. Therefore, iron loading may also influence these processes by regulating decorin and TGF-β.

In addition to iron-induced oxidative damage, iron overload is involved in other signaling pathways that are implicated in cancer and carcinogenesis. In vitro observations suggested that iron loading modulates Wnt signaling (Brookes et al., 2008). Activation of Wnt signaling by Wnt protein or chemicals such as lithium chloride is manifested via the inhibition of β-catenin degradation by the destruction complex composed of the proteins APC, GSK-3β, and Axin. Increased level of β-catenin up-regulates the transcription of genes such as c-myc and Nkd1, and subsequently increases cellular proliferation. In cell lines that harbor mutated APC, iron loading is sufficient to induce Wnt signal. In other cell lines with wild-type APC, iron loading can enhance Wnt signal in conjunction with lithium chloride. Therefore, iron alone is not a potent carcinogen, but it may exaggerate carcinogenesis in the presence of carcinogenic factors such as existing APC mutations or elevated Wnt signal. Iron overload in the colonic lumen due to meat consumption may be especially important for the development of colorectal cancer. Conversely, iron depletion by iron chelators inhibits proliferation of leukemic and cancer cells by both β- 69 catenin-dependent and independent mechanisms (Coombs et al., 2012; Song et al., 2011). APC is located on the human chromosome 5q that is commonly deleted to MDS, and mice that express a non-functional version of APC develop a condition that resembles MDS/MPD (Lane et al., 2010). In addition, enhanced Wnt signal has been reported in MDS due to overexpression of Wnt activating oncogene or downregulation of Wnt antagonists (Masala et al., 2012; Shuai et al., 2009). Since secondary iron overload as a result of chronic transfusion is a common feature in MDS, it may be reasonable to speculate that excess iron is involved in the pathogenesis of MDS via the Wnt signaling pathway.

Iron has been shown to activate the P13K/Akt pathway in the development of head and neck squamous carcinoma, as well as neuronal response to oxidative stress (Kaomongkolgit et al., 2008; Uranga et al., 2013). Akt is a serine/threonine protein kinase that is involved in the regulation of several cellular functions including nutrient metabolism, cell growth, and survival (Song et al., 2005). Activation of Akt promotes survival and prevents apoptosis in normal and cancer cells. Crosstalk between Akt and numerous signaling pathways in the context of cancer has been documented, which includes forkhead box O transcription factors (FOXOs) (Fu and Tindall, 2008), NF-κB (Hussain et al., 2012), MAPK/ERK (De Luca et al., 2012), Wnt (Vadlakonda et al., 2013), the estrogen receptor (Bratton et al., 2010), and the androgen receptor (Wang et al., 2007). The antagonistic interaction between Akt and FOXOs is of particular interest in cancer. Activation of Akt leads to inhibition of FOXOs and diminishes its function as a tumor suppressor, a process that is implicated in 50% of AML (Martelli et al., 2010). In the context of iron overload, excess iron, through Akt activation, may favor the development of cancer or leukemia with the Akt on / FOXO off signaling characteristic. Paradoxically, decreased Akt activity and increased FOXO activation was observed in about 40% of AML patients, and FOXO activation is inversely correlated with JNK/c-JUN signaling (Sykes et al., 2011). Sykes et al. proposed that FOXOs may possess an additional role to maintain the immature state of leukemia initiating cells and prevent them from differentiation and apoptosis.

Thus, iron overload can promote cell death or carcinogenesis at several levels, including dysregulation of various cancer-related signaling pathways that may or may not involve iron- induced oxidative stress, as well as iron-catalyzed oxidative damage of macromolecules. The actual physiological or pathological effects of iron are dictated by cell context, which varies 70 depending on cell-types, active signaling pathways, gene expression pattern, ROS status, and degree of iron loading.

1.4.2.4 Cell death and survival Accumulation of ROS and iron-induced oxidative stress can be a modulating or determining factor in multiple forms of cell death. Oxidative stress in the mitochondria can induce apoptosis by two different pathways (Dixon and Stockwell, 2014). ROS can trigger a mitochondria-to- nucleus signaling pathway that involves the activation of AMP-activated protein kinase (AMPK) and up-regulation of E2F1 transcription factor, which subsequently leads to the expression of pro-apoptotic genes and apoptosis. ROS are also responsible for the intrinsic apoptotic pathway via the release of cytochrome c from the mitochondria, which induces the formation of the apoptosome and leads to caspase-3 and caspase-7 activation. The levels of ROS and LIP may have a cell-type specific role in modulating a form of TNF-dependent necrotic cell death called necroptosis, which is mediated by receptor interacting protein kinase 1 (RIP1), RIP3, STAT3, and GRIM-19 (Dai et al., 2013). Ferroptosis is a novel form of iron-dependent non-apoptotic cell death triggered by certain small molecules such as erastin and sulfasalazine (Dixon et al., 2012). Ferroptosis is initiated by the inhibition of cysteine uptake, thereby lowering the production and causing the depletion of glutathione (GSH). GSH depletion triggers cell death by allowing cellular ROS accumulation. The exact role of iron in ferroptosis is not fully understood, but it may be involved as a part of ROS producing enzymes such as heme-containing NADPH oxidases.

Protection against ROS is mediated by antioxidant defenses that comprise various enzymes and compounds (Landriscina et al., 2009). Cells have limited capability to handle iron-dependent generation of hydroxyl radicals due to their high reactivity and short half-life. As a result, protection against iron-induced oxidative stress is mediated by preventing the formation of hydroxyl radicals at the source, such as the removal of hydrogen peroxide (Crichton, 2009i). Both normal and malignant cells can up-regulate a network of antioxidant enzymes to handle elevated oxidative stress. Adaptation to increased ROS level is in part mediated by the activation of nuclear factor erythroid 2-related factor (Nrf2) (Landriscina et al., 2009; Ma and He, 2012). In normal circumstances, Nrf2 is bound to Kelch like-ECH-associated protein 1 (Keap1) in the cytosol and degraded quickly by ubiquitination. Oxidative stress halts Nrf2 degradation and 71 allows it enter the nucleus and up-regulate the expression of antioxidant enzymes, including PRX-1, GPX, TrxR, and GST. Although Nrf2 can protect normal cells from oxidative stress, it has also been implicated in cancer and chemoresistance as a way to counteract oxidative stress from the tumor microenvironment and chemotherapy, respectively (Jaramillo and Zhang, 2013).

Alternatively, adaption to excess iron and oxidative stress can be achieved without lowering iron or ROS levels. Oxidative stress and iron can modulate several signaling pathways such as Akt, p53 and Wnt, which promote survival, avoid apoptosis, allow escape from growth arrest, and facilitate cancer transformation (Hollstein et al., 1991; Song et al., 2005; Vadlakonda et al., 2013). In addition, maintaining proteome stability is crucial for cells to survival under oxidative stress, in which protein oxidation accounts for almost 70% of all oxidized molecules (Corcoran and Cotter, 2013). Oxidized proteins can become unfolded or misfolded, which diminish functionality and cause aggregation (Niforou et al., 2014). Molecular chaperones such as heat- shock proteins (HSPs) are proteins that can rescue the damaged proteins or direct them to degradation if needed. Oxidative stress activates a group of transcription factors called heat- shock factors (HSFs), thereby up-regulating the expression of multiple HSPs and protecting the cell from protein oxidation.

Many long-living species have a high level of oxidative stress, including birds, bats, and the naked mole-rat. Compared to other rodent species, the naked mole-rat has a much lower level of antioxidant capacity and corresponding high levels of liver iron, oxidative stress and related damage (Andziak and Buffenstein, 2006; Chiu et al., 1976; Perez et al., 2009). Nevertheless, the naked mole-rat is highly resistant to cancer and can survive for almost three decades, versus >70% cancer rate and four years lifespan for common laboratory mouse strains such as C57BL/6 (Lewis et al., 2013). A number of non-antioxidant cytoprotective mechanisms may allow the subterranean rodent to withstand high oxidative stress and contribute to its longevity, including low metabolic and respiratory rates (Nathaniel et al., 2012), hypersensitivity to cell contact inhibition via both p16 and p27 (Seluanov et al., 2009), secretion of extremely high-molecular- mass hyaluronan (Tian et al., 2013), unique 28S ribosomal RNA cleavage with increased translational fidelity (Azpurua et al., 2013), and maintenance of robust proteostasis (Rodriguez et al., 2012). Many of the physiological aspects of longevity of the naked mole-rat remain to be

72 fully elucidated, but the findings may have important implications for cancer and carcinogenesis that is related to excess iron loading and oxidative stress.

1.4.3 Iron overload in MDS Clinical estimation of iron overload in MDS resembles that of β-thalassemia major, and includes assessment based on transfusion history, measurement of systemic iron status by transferrin saturation and serum ferritin, as well as LIC and cardiac iron load by MRI (Leitch, 2011; Mahesh et al., 2008).

Unlike thalassemia, the knowledge regarding the effects of iron overload in the context of MDS is limited and based mostly on retrospective and non-randomized analyses (Leitch, 2011). In an Italian study of 467 patients with de novo MDS, transfusion dependency was associated with shorter OS and LFS in both low and high risk MDS patients (Malcovati et al., 2005). However, transfusion dependency is not only an indicator of iron overload but may also be associated with more severe bone marrow disease, confounding the association with inferior outcomes. Additional markers are required to assess the iron status, including transfusion burden per month and serum ferritin. High transfusion burden is an independent predictor of poor OS for both low and high risk patients, and LFS for low risk patients. Serum ferritin above 1000 µg/L is also a predictor of poor OS, and every 500 µg/L increase in serum ferritin corresponds to a 30% greater risk of death after transfusion burden is adjusted. The deleterious effect of secondary iron overload is significant among patients with RA/RARS, but not among those with RCMD/RCMD-RS. The association of secondary iron overload with inferior OS in low risk MDS patients has also been shown in other cohorts (Cermak et al., 2009; de Swart, 2011; Sanz et al., 2008). Excess iron may contribute to shorter LFS by promoting leukemogenesis and enhancing the growth of leukemic blasts (Leitch, 2011).

Despite the role of chronic anemia in cardiac remodeling and related complications (Goldberg et al., 2010; Temraz et al., 2014), chronic transfusion remains a risk factor for cardiac events when compared to the general Medicare population and transfusion independent MDS patients in the US (Goldberg et al., 2010). The higher rate of cardiac events may be attributed to cardiac remodeling due to incomplete alleviation of anemia, myocardial hypoxia due to fluctuation of hemoglobin levels, and iron overload due to transfusion. In a Japanese cohort, cardiac failure

73 also accounted for the most non-leukemic deaths (Takatoku et al., 2007). Most of these patients had serum ferritin levels above 1000 µg/L, which is indicative of an association between secondary iron overload and mortality due to cardiac failure in MDS. However, cardiac iron accumulation is not frequent in MDS patients, and cardiac complications in transfusion dependent MDS patients may be associated with anemia or oxidative stress rather than direct iron deposition in the heart (Di Tucci et al., 2008; Konen et al., 2007; Leitch, 2011). Transfused MDS patients are more likely than non-transfused patients to develop comorbidities such as infection, dyspnea, and tendency towards diabetes mellitus (Goldberg et al., 2010). Hepatic cirrhosis develops later than cardiac complications in transfusion dependent MDS patients and liver complications do not appear to contribute to higher mortality than non-transfused patients (Di Tucci et al., 2008; Konen et al., 2007; Steensma and Gattermann, 2013). Nevertheless, signs of hepatic complications, such elevated levels of aspartate aminotransferase and alanine aminotransferase, are found to be correlated with transfusion and hepatic iron overload (Delea et al., 2009; Goldberg et al., 2010; Takatoku et al., 2007). Impairments of heart, liver, and glucose intolerance can develop within four years in transfusion dependent patients (Schafer et al., 1981).

Iron overload before transplantation is associated with inferior OS and disease-free survival with a higher rate of treatment-related mortality in MDS patients who underwent HSCT (Armand et al., 2007). Although elevated pre-transplant serum ferritin is generally associated with inferior survival after HSCT, subsequent studies reported variable findings regarding the effect of iron overload on parameters such as GVHD and pre-transplant conditioning (Alessandrino et al., 2008; Koreth and Antin, 2010; Mahindra et al., 2009a; Mahindra et al., 2009b; Platzbecker et al., 2008; Pullarkat et al., 2008). Iron overload can also be exacerbated in HSCT candidates by factors other than transfusion, which include increased iron absorption due to ineffective erythropoiesis, release of iron from destroyed hepatic and hematopoietic cells due to the conditioning regimen, and decreased iron utilization due to chemotherapy-induced BM aplasia (Steensma and Gattermann, 2013).

1.4.4 Iron chelation in MDS When iron overload is suspected, ICT is required to remove the excess iron and restore iron balance. Since phlebotomy is not suitable due to chronic anemia, ICT is the only option for MDS patients, and oral chelators are generally preferred. The choices of oral chelators in MDS 74 resemble those for the treatment of secondary iron overload in thalassemia (Cazzola et al., 2008; Greenberg et al., 2013; Steensma and Gattermann, 2013; Temraz et al., 2014). The use of deferoxamine is well established for the treatment of iron overload with good safety and efficacy profiles. However, patient compliance with the treatment is a major concern due to the inconvenience of frequent subcutaneous infusion and other adverse effects. Deferiprone is administered via the oral route and is more effective than deferoxamine at reducing cardiac iron overload. Combined therapy with the two chelators is synergistic for iron removal and can be used for patients who cannot achieve negative iron balance with monotherapy. Nonetheless, deferiprone has to be taken three times daily and it has not received approval for MDS in North America. Also, deferiprone has been associated with sporadic agranulocytosis (Cohen et al., 2000), which may lead to serious infections and death (Greenberg et al., 2013). Deferasirox is a once daily oral chelator that is approved for clinical use in many countries. The convenience of its route of administration compare to deferoxamine allows higher patient compliance, and it is now frequently used as the first line therapy for secondary iron overload in MDS. The long-term effectiveness and safety of deferasirox have not been defined in higher-risk MDS patients, and it is currently not recommended for these patients.

ICT has been shown to decrease iron burden in the context of MDS (Temraz et al., 2014). Deferasirox significantly decreased serum ferritin and LIC in iron overloaded MDS patients with no prior history of ICT, while the decrease was less pronounced for pretreated patients (Breccia et al., 2012; Gattermann et al., 2010; Gattermann et al., 2012b; Porter et al., 2008). In addition, a corresponding decrease of alanine aminotransferase levels with reduced serum ferritin was observed, suggesting possible improvement of hepatic function upon ICT (Gattermann et al., 2010). A similar observation was also reported by Takatoku et al., who found that serum ferritin, aspartate aminotransferase, alanine aminotransferase, and fasting blood glucose were lowered from baseline for MDS patients who received continuous deferoxamine treatment (Takatoku et al., 2007). Some iron overloaded MDS patients experience hematological improvements upon ICT (Leitch, 2011; Steensma and Gattermann, 2013; Temraz et al., 2014). Jensen et al. reported reduction of transfusion requirement in seven out of eleven patients who were chelated by deferoxamine; five of them even became transfusion independent (Jensen et al., 1996). Other improvements were noted, including elevated platelet and neutrophil counts. Recent studies on

75 deferasirox and deferoxamine also reported erythroid, platelet, and neutrophil responses after ICT (Cilloni et al., 2011; Gattermann et al., 2012a). Reduction of serum ferritin tends to be greater for hematologic responders than non-responders, and response time was shorter for deferasirox than deferoxamine. The observed hematologic improvement may be the consequence of reduced iron-related oxidative stress by ICT (Leitch, 2011). ICT may also improve erythropoiesis by reversing suppression of proliferation of erythroid progenitors by iron overload (Hartmann et al., 2013). Furthermore, deferasirox inhibited NF-κB activity in PB mononuclear cells collected from MDS patients, as well as in leukemia cell lines (Messa et al., 2010). The inhibition of NF-κB is unique to deferasirox and exerted by a mechanism independent of iron and ROS. Elevated neutrophil count may lead to improvement of immunity and thereby lower the risk of infection (Steensma and Gattermann, 2013). Alternatively, the risk of bacterial and fungal infections may also be reduced as a result of iron deprivation by ICT.

Pre- or post-transplantation iron reduction in iron overloaded MDS patients is hypothesized to be beneficial to the outcome of HSCT (Leitch, 2011; Steensma and Gattermann, 2013). A number of studies have been conducted to examine this notion (Armand et al., 2013; Busca et al., 2010; Kamble et al., 2006; Lee et al., 2009; Majhail et al., 2010; McKay et al., 1996; Rose et al., 2007; Tomas et al., 2000). Post-HSCT iron reduction by ICT or phlebotomy is associated with decreased serum ferritin and LIC, as well as improvement in liver function. Patients whose serum ferritin was maintained below 1000 µg/L prior to HSCT had better OS than those with serum ferritin above 1000 µg/L (Lee et al., 2009). The OS of the chelated patients also resembled those who did not receive ICT and had serum ferritin below 1000 µg/L. Moreover, designing an appropriate chelation regimen at the time of transplantation is a challenging endeavor, and it may be necessary to prevent iron overload at an earlier time point rather than to reduce excess iron prior to HSCT (Armand et al., 2013; Leitch, 2011).

Several retrospective studies have suggested the beneficial effect of ICT to prolong survival in iron overloaded MDS patients. A Canadian study reported that ICT by deferoxamine is associated with improved survival in iron overloaded MDS patients with IPSS low or Int-1, compared to those who did not receive the treatment (Leitch et al., 2008). The four year survival rates of chelated versus non-chelated patients were 64% and 43%, respectively, with a hazard ratio of 0.29 after adjustment for clinical and prognostic features. Similar observations were also 76 reported by US, French and German researchers for low-risk MDS patients (Neukirchen et al., 2012; Raptis et al., 2010; Rose et al., 2010). Even patients with low-dose ICT have better median OS than the non-chelated group at 85 versus 124 months, respectively, suggesting the benefit of chelation is dose-dependent (Rose et al., 2010). In addition, the benefit of ICT on OS is more pronounced for heavily transfused patients (Rose et al., 2010). Conversely, the effect of ICT on OS for high risk MDS patients was either not examined or not significant. Rose et al. reported that 34% of non-chelated patients progressed to AML versus 17% for chelated patients, but the difference was not statistically significant (p=0.087) (Rose et al., 2010). Sanz et al. also reported that iron overload and transfusion dependency were associated with the risk AML transformation (Sanz et al., 2008). On the other hand, Neukirchen et al. reported no difference in the cumulative risk of AML transformation between chelated and supportive care groups (p=0.73), which was based on 16 pairs of patients who were matched for age, gender, WHO classification, and IPSS (Neukirchen et al., 2012). Nevertheless, these studies have several limitations that are common to retrospective studies (Hess, 2004): inherently susceptible to selection/referral bias and confounding; loss of accurate and important information from some subjects; difficulties in establishing cause and effect. Prospective clinical trials will be required to address these shortfalls, although available resources are more likely to be directed towards newly developed iron chelators. Deferasirox was approved relatively recently for clinical use when compared to deferoxamine and deferiprone. The efficacy of deferasirox (marketed as Exjade®) in reducing iron burden was demonstrated in the Evaluation of Patients Iron Chelation with Exjade® (EPIC) study, which was a prospective, 1-year, multicenter, open-label, phase IIIb clinical trial involving 1744 iron overloaded patients with various transfusion-dependent anemias (Gattermann et al., 2010). Serum ferritin, labile plasma iron, and alanine aminotransferase levels were reduced significantly in the 341 MDS patients who received deferasirox in the EPIC study – suggesting alleviation of iron overload and hepatocellular injury. A post-hoc analysis has further shown erythroid (21.5%, 52/247 patients), platelet (13.0%, 13/100 patients), and neutrophil (22.0%, 11/50 patients) responses for a subset of MDS patients who did not receive concomitant medication for MDS (Gattermann et al., 2012a). Other large prospective clinical trials are currently underway to assess the effect of deferasirox on OS and LFS in iron overload low risk MDS patients (Gattermann et al., 2010; Porter et al., 2012; Temraz et al., 2014).

A number of guidelines have been established by different organizations on the use of ICT for the treatment of secondary iron overload in MDS (Alessandrino et al., 2002; Arrizabalaga et al., 2008; Bennett et al., 2008; Bowen et al., 2003a; Greenberg et al., 2013; Mittelman et al., 2008; Santini et al., 2010; Steensma and Gattermann, 2013; Suzuki et al., 2008; Valent et al., 2008; Wells et al., 2008). Nonetheless, retrospective studies account for much of the evidence regarding the deleterious effects of secondary iron overload and the clinical benefits of ICT in the context of MDS (Temraz et al., 2014). Current recommendations for ICT in MDS are largely based on retrospective studies, expert opinions and the experience from transfusion-related iron overload in thalassemia patients (Leitch, 2011; Mahesh et al., 2008; Temraz et al., 2014; Yeh et al., 2009). Although there are some variations between different guidelines, they adhere to several general principles. ICT is primarily recommended for low risk MDS patients with life expectancy longer than 6 to 24 months, who are more likely to be adversely affected by the deleterious effects of secondary iron overload. ICT is generally not offered to high risk MDS patients, except for those who are scheduled to undergo HSCT or have compromised organ function. ICT is indicated when the serum ferritin is above 1000-2500 µg/L or transfusion burden is more than 20-50 units PRBC. The target of ICT is to lower the serum ferritin to below 500-1000 µg/L, while dose reduction may be recommended when serum ferritin is below 2000 µg/L.

1.5 Current questions, rationale and project objectives

1.5.1 Current questions and rationale Chronic RBC transfusion is an inevitable consequence in many MDS patients with anemia. Transfusion dependency correlates with poor outcome in MDS, and the intensity of transfusion is associated with underlying BM dysfunction. In addition, chronic RBC transfusion leads to secondary iron overload. As will be discussed below, excess iron may further impair the function of the marrow and worsen the outcome of these patients. Conversely, it is currently unclear if survival is associated with the initial functional state of the BM at the onset of transfusion. If so, it may be useful to develop a clinically feasible parameter to assess this initial state.

Secondary iron overload as a consequence of chronic RBC transfusion is associated with poor outcome in MDS. There is general agreement on the benefits of ICT for the treatment of

78 secondary iron overload. The recommendations for ICT by various guidelines reflect the current understanding of the effects of iron overload and chelation in the context of MDS, but are not without controversies (Gattermann, 2008; Leitch, 2011; Steensma and Gattermann, 2013). Compared to secondary iron overload in β-thalassemia major, clinical data regarding the pathophysiological effects of excess iron and its removal in MDS are relatively limited and the majority of these are based on retrospective studies. Although these studies have provided valuable clinical information, many are small-scale studies and vulnerable to confounding and selection bias. Meta-analysis and prospective randomized controlled clinical trials will provide more reliable data to improve existing ICT guidelines. Ongoing prospective trials, such as the TELESTO trial, are primarily conducted in patients with low risk MDS receiving deferasirox, while most high risk MDS patients will not participate (Steensma and Gattermann, 2013).

ICT is not currently recommended for high risk MDS patients mainly on the basis that their life expectancy is not long enough to be affected by the symptoms related to secondary iron overload. The notion that the deleterious effects of iron overload take years to develop, as well as the current guidelines for ICT in MDS, is largely extrapolated from the experience with secondary iron overload in β-thalassemia major. Although anemic MDS and thalassemia patients are destined to acquire secondary iron overload due to chronic transfusion, the two diseases are fundamentally distinct from one another, and by extension, their responses to excess iron and ICT may not be identical (Leitch, 2011; Pullarkat, 2009). Compared to thalassemia, MDS is associated with advanced age and co-morbidities. Even though the effect of ICT on development is not a significant issue for older MDS patients, their tolerance for iron loading may be lower than thalassemia patients due to aged-related organ deterioration. Iron overload may also aggravate existing co-morbidities in MDS, especially for patients with cardiac dysfunction. In contrast, thalassemia patients have intact cardiac health at birth and their hearts can withstand an extended period of iron-induced damage.

Another major difference lies in the nature of the disease. MDS is a heterogeneous pre-leukemic condition associated with impaired hematopoiesis of multiple lineages, while thalassemia is only associated with ineffective erythropoiesis. The heterogeneity among different MDS subtypes may imply heterogeneous response to iron loading and ICT. In fact, Leitch et al. reported a stronger association between ICT and superior OS in low risk non-RARS patients than those 79 with RARS (Leitch et al., 2012). Excess iron may worsen cytopenias by further impairing hematopoiesis in conjunction with the underlying pathology. Conversely, a number of observations suggest that iron removal can improve residual erythropoiesis and alleviate anemia, which is irrelevant to thalassemia given the non-existence of normal erythropoietic apparatus (Cilloni et al., 2011; Gattermann et al., 2012a). Alleviation of cytopenias as a result of ICT can also increase neutrophil and platelet counts, thereby lowering the risk of infection and bleeding in MDS. More importantly, MDS patients are predisposed to develop leukemia, while thalassemia patients are not. Iron overload and iron-induced oxidative stress may accelerate leukemogenesis by aggravating existing genomic instability and dysfunctional signaling pathways in the pre-leukemic clones of MDS (Steensma and Gattermann, 2013). This is particularly crucial for high risk patients, since they are at the highest risk for AML transformation. On the other hand, ICT may counteract the leukemogenic effects of excess iron and prolong LFS. The roles of iron and ICT in leukemogenesis are yet to be confirmed and defined. Nevertheless, in many cancers, iron is implicated as the sole or contributing factor in cancer and carcinogenesis, while iron reduction is associated with decreased cancer risk (Toyokuni, 2009; Zacharski et al., 2008). Thus, if secondary iron overload can modulate the underlying pathology of MDS, and excess iron can adversely affect high risk MDS patients within their life expectancy, then ICT may provide more benefits than is reflected in its current indications, and it may be required by more patients including those with high risk MDS.

1.5.2 Project goal The goal of this project is to assess the roles of anemia, secondary iron overload, and iron chelation in several aspects related to MDS including transfusion requirements, oxidative stress, and leukemogenesis.

1.5.3 Project-specific questions Through the work detailed herein, the following questions will be addressed:

1. Is there a clinically measureable parameter that can describe the severity of erythropoietic failure close to the time of transfusion initiation? (Chapter 2) 2. Does the above parameter predict disease outcome? (Chapter 2)

3. How to properly measure the levels of intracellular ROS (iROS) from the early BM hematopoietic cells of MDS patients? (Chapter 3) 4. What are the ROS characteristics of these early hematopoietic cells? (Chapter 3) 5. What is the relationship between iron overload and the iROS levels in these cells? (Chapter 3) 6. Does iron loading lead to iron accumulation in hematopoietic tissues? (Chapter 4) 7. By what extent does iron loading modulate the incidence rate and onset of AML in a pre- leukemic mouse model? (Chapter 4) 8. By what extent does iron chelation modulate the effects of iron loading on AML transformation in vivo? (Chapter 4) 9. What are the possible mechanisms for iron-induced leukemogenesis? (Chapter 4)

1.5.4 Objective and hypotheses Objective 1: To examine the transfusion history of MDS patients to explore possible pathophysiological and prognostic values of these parameters.

Hypothesis: The initial requirement for RBC transfusion reflects the severity of erythropoietic failure and is a predictor of disease outcome in MDS.

Objective 2: To assess the characteristics of iROS in the early BM hematopoietic cells among MDS patients, and to determine the relationship between the iROS levels in these cells and the degree of iron overload.

Hypothesis: The ROS characteristics of early hematopoietic cells are different between low and high risk MDS patients. In addition, iron overload secondary to transfusion leads to increased level of intracellular ROS in early hematopoietic cells in MDS.

Objective 3: To elucidate the effect of iron overload and iron chelation on hematopoiesis and leukemogenesis in animal models.

Hypothesis: Iron overload, through ROS dependent and independent mechanisms, damages macromolecules and dysregulates signaling pathways, which leads to impairment of hematopoiesis and acceleration of leukemogenesis. The deleterious effects of iron overload can be counteracted by iron chelation. 81

Chapter 2

Initial Transfusion Intensity Predicts Survival in MDS

Chapter 2 was published as follows:

Chan LSA, Shapiro R, Buckstein R, Lin Y, Callum J, Chodirker L, Lee CD, Prica A, Lam A, Mamedov A, Wells RA, Initial Transfusion Intensity Predicts Survival in Myelodysplastic Syndrome. Leukemia & Lymphoma 2014 Oct; 55(10): 2296-2300.

2.1 Introduction For most patients with myelodysplastic syndrome (MDS), anemia and associated transfusion dependency are the most important clinical problems and the main determinants of quality of life (Balducci, 2010; Kao et al., 2008; Malcovati et al., 2005; Platzbecker et al., 2012). Most MDS patients are anemic at the time of diagnosis, and severe anemia develops in nearly two-thirds of cases (Steensma and Bennett, 2006). Despite the recent introduction of active disease-modifying therapies for MDS (Fenaux and Ades, 2013; Giagounidis et al., 2008; Hellstrom-Lindberg et al., 2003; List et al., 2005), red blood cell transfusion remains a mainstay in the management of patients with this condition.

Several groups have reported an association between transfusion and poor outcome in MDS (Cazzola and Malcovati, 2005; Mahesh et al., 2008; Malcovati et al., 2006; Malcovati et al., 2005). Transfusion dependency, as a categorical variable, is an independent predictor of inferior overall survival (OS), most notably in the WHO-based prognostic scoring system (WPSS), which incorporates “regular transfusion requirement” along with cytogenetics and WHO diagnostic category as prognostic variables for OS and leukemia-free survival (Malcovati et al., 2007). Moreover, both transfusion burden (as indicated by serum ferritin) and transfusion rate have dose-dependent effects on OS (Cermak et al., 2009; Malcovati et al., 2005; Pereira et al., 2011).

In a recent report, Pereira and coworkers analyzed records of 191 transfusion-dependent MDS patients to dissect the effects of transfusion burden from the related variables of time since diagnosis and transfusion frequency (Pereira et al., 2011). Transfusion frequency was captured in the parameter “transfusion intensity” (TI), which they defined as (number of RBC units transfused at a given transfusion episode x 365 days)/number of days from the previous transfusion episode; in other words, the number of RBC units the patient would have received over a full year had he/she kept being transfused at that rate over a complete year. Although the cumulative number of RBC units received was associated with significantly reduced OS, this variable was confounded by both time from first transfusion and TI, with TI accounting for most of the prognostic importance of transfusion burden.

The practical utility of the TI as a prognostic variable in clinical practice may be limited if it tends to change over time due to therapy, allo-immunization, splenomegaly, and other influences that may change the half-life of transfused RBC. However, transfusion intensity at the beginning of transfusion dependency should be relatively unaffected by such influences and therefore may accurately reflect the extent of underlying bone marrow dysfunction. Supporting this supposition is the finding that transfusion intensity at the beginning of transfusion dependency is a good predictor for reaching a cumulative transfusion burden of 25 units (Pereira et al., 2011).

We hypothesized that the TI immediately after a patient begins to require RBC transfusions would reflect the severity of erythropoietic failure and would be relatively unaffected by these other factors. We report here that this initial transfusion intensity (ITI) – defined simply as the number of units of blood transfused from weeks one to five excluding the initial transfusion is a simple, robust, and highly prognostic marker for overall survival in MDS.

2.2 Methods

2.2.1 Patients The MDS Research Program at the Odette Cancer Centre initiated an IRB-approved clinical database in 2006. All MDS patients referred to the MDS Centre of Excellence are offered an opportunity to participate in this database, which currently contains extensive longitudinal clinical and laboratory data on 411 patients. In order to examine the potential prognostic value of initial transfusion intensity in MDS, we defined the transfusion cohort as patients diagnosed with MDS between 1998 and 2010, whose diagnosis of MDS had been confirmed at our center, who were previously transfusion-naïve, had received at least one unit of RBC transfusion and who had received all of their transfusions at our centre, for whom we therefore had accurate and complete transfusion records. During the period of this study patients in our program were transfused according to a restrictive program, in which transfusion was given when the hemoglobin level fell below 80 g/L, or when severe symptoms of anemia including dyspnea, fatigue or increasing angina were present.

2.2.2 Data analysis Data collected included age at diagnosis, gender, diagnosis according to the WHO classification, international prognostic scoring system (IPSS) risk category, therapies including iron chelation therapy, and cause of death.

OS was calculated based on the time between first transfusion and death (as a result of all causes) or end of follow-up (censored observations). Survival analyses were performed using the Kaplan–Meier and Mantel–Cox methods. The mean cumulative transfusion burden was defined as the aggregate number of transfused RBC units received by all patients in the cohort at a given time point, divided by the number of patients in the cohort at that time point. Multivariate analysis was performed using Cox proportional hazards survival regression. Statistical significance was defined as a two-sided P-value less than 0.05.

2.3 Results

2.3.1 Clinical characteristics Data on patient characteristics and outcomes are summarized in Table 2.3.1. Survival data were available for all patients in the cohort, and the median overall survival (OS) was 72.7 months (median follow-up 33.5 months). Although all patients had received at least one RBC transfusion, only 36 (67.9%) met IWG criteria for transfusion dependency (TD, defined as having at least one RBC transfusion every 8 weeks over a period of 4 months) (Malcovati et al., 2007), while the remaining 17 patients were occasionally transfused (OT). Median survival in TD patients was 51 months, compared to 113 months in OT patients (P=0.26). At the close of this study on June 2011, 11 patients had progressed to AML (20.75%) and 28 patients (52.8%) had died. AML was the most frequent cause of death (10 patients), followed by sepsis (six patients) and cardiac causes (four patients).

Table 2.3.1. Clinical characteristics of patients with transfusion history related to MDS.

Variable Value

Total (N) 53 Age (Median, Range) 72 (26‐88) Sex (M/F) 31 (58.5%) / 22 (41.5%)

Diagnosis (WHO) 5q(2), CMML‐1(5), MDS/MPD(2), RA(5), RARS(2), RCMD(14), RCMD‐RS(4), Unclassified(6), CMML‐2(1), RAEB1(5), RAEB2(7) IPSS Low(11), Int‐1(25), Int‐2(9), High(6), No Score(2) Iron Chelation (Y/N) 17 (32.08%) / 36 (67.92%)

Active Treatment (Y/N) 34 (64.15%) / 19 (35.85%)

Transfusion Dependence (Y/N) 36 (67.92%) / 17 (32.08%)

AML Progression (Y/N) 11 (20.75%) / 42 (79.25%) Median Follow Up 33 months Median OS 72.7 months Cause of Death AML(10), sepsis(6),cardiac(4), SCT complications(2), other cancers(3), pulmonary fibrosis(1), unknown(2)

2.3.2 Establishment of initial transfusion intensity TI has been reported to be a predictor of outcome in MDS (Pereira et al., 2011), but the stability of this parameter over time in patients with MDS – whether TI changes over time – has not been studied. In particular, it is not known whether TI early in transfusion therapy is a robust predictor of TI throughout the course of a patient’s disease. We therefore evaluated the ITI, which we defined as the total number of RBC units transfused in the 4 weeks after the first transfusion, in our MDS cohort (Figure 2.3.2a). In this patient population the median interval between the first and second transfusions was 4 weeks. As an initial analytic strategy the ITI was therefore categorized into two groups: group 1 (ITI < median) and group 2 (ITI > median). One patient died before the end of the 4-week period and the corresponding ITI could not be obtained, and hence 52 patients were available for further analysis involving ITI.

Hemoglobin levels in the two groups immediately prior to the first and second transfusions were not significantly different (Figure 2.3.2b), verifying that transfusion practices were comparable in both groups in the initial transfusion period. Other hematological parameters, including platelet count, neutrophil count, mean cell volume, mean platelet volume, monocyte count and lymphocyte count, were not significantly different between the two groups (data not shown).

We considered the possibility that interpretation of the ITI might be confounded if significant numbers of patients with more aggressive MDS initiated intensive therapy during the first 4 weeks after the first transfusion, which would tend to result in increased transfusion needs in this period. However, only four patients in our cohort received active therapy of any kind within 4 weeks of their first transfusion. Similarly, the strength of any association between initial transfusion requirements and severity of bone marrow dysfunction would tend to be reduced if clinically significant bleeding were prevalent. In our cohort there were four episodes of bleeding necessitating blood transfusion, but none of these occurred within 6 months of the first transfusion.

0 unit (N=24) >0 unit (N=28) Group 1 Group 2 30

25 24

15 13 Frequency

10 5 4 5 3 2 1 0 0123567891011 Amount of Transfusion (1‐5 Weeks, Units PRBC)

Figure 2.3.2a. Initial transfusion intensity (ITI) distribution of patients with transfusion history related to MDS. ITI is the units of packed red blood cells (PRBC) transfused during the first 4 weeks of transfusion. Two ITI groups are identified: group 1 (0 unit) and group 2(>0 unit).

Hb before 1st transfusion Hb before 2nd transfusion 150 150 (g/L) (g/L)

100 100

50 50 Hemoglobin Hemoglobin P=0.3823 P=0.7101 0 0 Group 1 Group 2 Group 1 Group 2 Figure 2.3.2b. Pre‐transfusion hemoglobin (Hb) levels in ITI groups 1 and 2 prior to the first and second transfusion.

2.3.3 ITI predicts long-term transfusion requirements The initial difference in transfusion intensity between the two groups was maintained over the entire duration of follow-up, and the cumulative transfusion curves did not converge (Figure 2.3.3a), indicating that ITI is a predictor of long-term transfusion requirements in this patient population. The average TI of both groups showed an upward trend over time, although this trend was more marked in group 2 (Figure 2.3.3b). The heterogeneity of the average TI increased over time, possibly due to decreased sample size as a result of patient attrition. The observed increase in average TI was not apparent in the first year of transfusions, during which TI was stable in both groups (Figure 2.3.3c).

(a) 150

Group 2 (>0 unit, N=28) 100 PRBC/patient 50

Units Group 1 (0 unit, N=24) 0 0 26 52 78 104 130 156 182 Time from First Transfusion (Weeks) (b) 3

2 Intensity Group 2

1 Group 1 Transfusion 0 0 26 52 78 104 130 156 182 Time from First Transfusion (Weeks) (c) 2.0 weeks)

1.5 Intensity 1.0 Group 2

0.5 Group 1 PRBC/patient/10

Transfusion 0.0 0 13 26 39 52 65

(units Time from First Transfusion (Weeks)

Figure 2.3.3. Transfusion characteristics of the two ITI groups.(a)Mean cumulative transfusion burden of the two ITI groups with respect to the time from first MDS related transfusion. Vertical bars indicate standard deviation. (b) Ten weeks transfusion moving average of the two ITI groups. (c) Ten weeks transfusion moving average, only data prior to median survival of the respective groups are shown.

2.3.4 ITI is a prognostic indicator of OS ITI was strongly associated with OS. In the entire patient group, OS of group 2 was lower than that of group 1 (Figure 2.3.4a), with median actuarial survival of 440 weeks for group 1 and 167 weeks for group 2 (P<0.01; hazard ratio [HR] = 3.137). The lower OS of group 2 persisted when only patients who were TD according to IWG criteria were included in the analysis (Figure 2.3.4b), with median actuarial survivals of 315 weeks for group 1 and 128 weeks for group 2 (P<0.05; HR=2.770), and for IPSS low/intermediate-1 (Low/Int-1) patients (Figure 2.3.4c), with median actuarial survivals of 491 weeks for group 1 and 167 weeks for group 2 (P<0.01; HR=4.715).

To explore further the prognostic relevance of ITI, we performed multivariate analyses using a Cox proportional hazards regression model (Table 2.3.4). The following parameters were examined: IPSS (Int-2/High vs. Low/Int-1), age (per 10-year increase), AML progression (yes vs. no) and ITI (group 2 vs. 1). When the entire cohort was considered, all examined parameters were statistically significant negative predictors of OS, with the highest risk associated with ITI group 2 with respect to group 1 (P<0.01; HR=3.692; 95% confidence interval [CI] 1.414–9.640). ITI remained a significant negative predictor of OS after exclusion of patients with IPSS Int-2 and High risk (P<0.05; HR=5.072; 95% CI 1.299–19.803) and when analysis was limited to patients who did not progress to AML (P<0.01; HR=7.825; 95% CI 1.678–36.486).

All Patients (a) 100 Group 1 (0 unit) 80 Group 2 (>0 unit) 60 survival

20 Percent 0 0 104 208 312 416 520 624 728 Weeks Transfusion Dependent Patients (b) 100 Group 1 80 Group 2 60 survival

20 Percent 0 0 104 208 312 416 520 624 728 Weeks IPSS Low/Int‐1 (c) 100 Group 1 80 Group 2 60 survival

20 Percent 0 0 104 208 312 416 520 624 728 Weeks Figure 2.3.4. Impact of ITI on overall survival. (a) All patients (median survival: Group 1 – 440 wks, N=24 vs Group 2 – 167 wks, N=28; P<0.01, HR=3.137). (b) patients who are transfusion dependent according to International Working Group(IWG)2003criteria(mediansurvival:Group1–315wks,N=13vsGroup2 – 128 wks, N=22; P<0.05, HR=2.770). (c) the two ITI groups on overall survival for patients who are IPSS Low or Int‐1 (median survival: Group 1 – 491 wks, N=15 vs Group 2 – 167 wks, N=20; P<0.01, HR=4.715). 92

Table 2.3.4. Multivariate analysis of factors affecting survival.

Variable P‐value Risk Ratio 95% CI All Patients with ITI (N=50) IPSS (Int‐2/Hi vs Low/Int‐1) 0.0452 2.5539 1.0202 to 6.3930

Age (per 10‐year increase) 0.0355 1.5677 1.0310 to 2.3838

AML Progression (Yes vs No) 0.0064 3.4506 1.4161 to 8.4083

ITI (Group 2 vs 1) 0.0076 3.6919 1.4140 to 9.6397 IPSS Low/Int‐1 (N=35) Age (per 10‐year increase) 0.0067 2.3918 1.2730 to 4.4941

ITI (Group 2 vs 1) 0.0195 5.0716 1.2988 to 19.8031 Patients Who Did Not Progress to AML (N=40) Age (per 10‐year increase) 0.0051 2.2874 1.2826 to 4.0791 ITI (Group 2 vs 1) 0.0088 7.8248 1.6781 to 36.4863

2.4 Discussion Chronic red blood cell transfusion as a result of anemia is common among MDS patients (Steensma and Bennett, 2006). Previous reports have identified transfusion dependency and the intensity of transfusion as negative prognostic factors in MDS (Cermak et al., 2009; Malcovati et al., 2005; Pereira et al., 2011). In the present study we have examined the initial intensity of transfusion and investigated its association with outcome in MDS. In a group of 53 MDS patients who received all of their transfusions at our centre, higher ITI was strongly associated with reduced OS. In multivariate analysis, this association was robust when age, IPSS score and AML progression are also considered. In patients with lower risk IPSS, higher ITI identified patients who were at risk of dying early, and might thus be considered candidates for disease-modifying therapy including allogeneic stem cell transplantation.

Various mechanisms may account for the association between transfusion dependency and poor prognosis in MDS. Transfusion requirement is a marker for the degree of symptomatic anemia and therefore for the severity of bone marrow failure. However, independent of the clinical and biological characteristics of the underlying disease, transfusion-related complications such as immunomodulation (TRIM) (Blumberg et al., 1990) and the adverse effects of banked blood on the microcirculation (Frenzel et al., 2009) may be associated with poor MDS outcome. It should be noted that TRIM is unlikely to play a major role in Canadian patient cohorts, since all RBC units transfused in Canada are leukoreduced. Iron overload secondary to transfusion, a phenomenon dependent upon cumulative transfusion burden rather than transfusion intensity, is suspected of contributing to morbidity and premature mortality in MDS (Goldberg et al., 2010; Pullarkat, 2009; Schafer et al., 1981). Nonetheless, early transfusion requirements, and by extension erythropoietic failure between the two ITI groups, are different even before transfusion-related iron overload is possible (Figure 2.3.3a/b/c), indicating that the initial state of erythropoiesis at the onset of transfusion impacts overall survival as well. In summary, it is likely that the adverse effects of transfusion on MDS prognosis are multifarious and complex. Since it is purely an index of the depth of erythropoietic failure, the ITI potentially offers a fast and simple means of dissecting the prognostic impact of these complex effects.

There are several limitations to this study that need to be addressed in the future. Our observations may be susceptible to bias and confounding that are inherent in a retrospective 94 study. For example, our cohort was recruited from a single center, and therefore referral and selection biases cannot be excluded. In addition, our small sample size limits the predictive power of other covariates on OS in regression model analysis. Finally, our arbitrarily defined cutoffs for the two ITI categories require validation and refinement in an independent cohort. To address these limitations, we are currently in the process of prospectively collecting transfusion data for MDS patients from multiple Canadian sites in a larger nationwide database.

It is tempting to conjecture that the poor prognosis of lower risk MDS patients with high ITI might be improved by early intervention with disease-modifying therapy such as a hypomethylating agent (HMA) (Garcia-Manero and Fenaux, 2011). However, further data are required before such a recommendation can be endorsed. The modifying effects of HMA therapy on prognosis could not be examined in this study since these agents are not reimbursed for lower risk MDS in Ontario. Evaluation of ITI in a prospective clinical trial of an HMA in lower risk MDS will be essential to ascertain the value of this parameter as a predictive biomarker for this type of therapy.

We conclude that ITI warrants evaluation as a fast and simple means of identifying MDS patients who, despite a lower risk IPSS category, are at high risk of premature mortality and who should therefore be candidates for disease-modifying therapy.

Chapter 3

Serum Ferritin Level is Associated with Intracellular ROS in the Hematopoietic Progenitors of MDS Patients

3.1 Introduction Reactive oxygen species (ROS), redox active molecules containing oxygen, are natural by- products of mitochondrial respiration and have crucial roles in various signaling pathways, homeostasis, and cellular functions (Bae et al., 2011). Under normal circumstances, the production of intracellular ROS (iROS) depends on cell type and the rate of metabolic activity (Simko, 2007). The level of iROS can also be altered by exogenous factors such as oxygen tension and the presence of redox-active metal ions (Halliwell and Gutteridge, 1990). In the bone marrow, long-term primitive hematopoietic stem cells (HSCs) have a lower iROS level than more mature HSCs or hematopoietic progenitors, possibly due to both cell intrinsic and extrinsic factors (Jang and Sharkis, 2007). Excess iROS induces oxidative stress by overwhelming the cell’s anti-oxidant defense, resulting in damage to biomolecules and subsequently triggering mutagenesis or cell death (Zhou et al., 2014).

In myelodysplastic syndrome (MDS), iROS homeostasis may be affected by several mechanisms. Disruption of the bone marrow microenvironment may change local oxygen tension, thereby altering the iROS level of bone marrow cells including HSCs and progenitors (Das et al., 2013; Spencer et al., 2014). The metabolic profile of MDS cells varies according to disease stage, with activation of Akt signaling in higher risk MDS leading to a shift to glycolysis and therefore reduced generation of ROS in comparison to lower risk MDS (Brand and Hermfisse, 1997; Nyakern et al., 2006). Finally, a large proportion of MDS patients require regular red blood cell (RBC) transfusion, resulting in secondary hemochromatosis (Fenaux and Rose, 2009). Excess iron contributes to the induction of oxidative stress by catalyzing the generation of hydroxyl radicals via the Haber-Weiss reaction (Kehrer, 2000). It has been hypothesized that iron-induced oxidative stress may damage CD34+ HSCs and progenitors, further impairing hematopoiesis and accelerating the progression of MDS to AML (Leitch, 2011). Hence, oxidative stress may contribute to impairment of hematopoiesis in MDS and may exert a selective pressure promoting disease progression, and thus investigation of the iROS level of CD34+ bone marrow cells from MDS patients is of significant interest.

The oxidation of 2',7'-dichlorodihydrofluorescein diacetate (DCFH-DA) to fluorescent 2′,7′- dichlorofluorescein (DCF) by ROS is an effective method to measure iROS level in live cells by flow cytometry (Bass et al., 1983). In this paper we discuss the adaptation of DCF-based iROS 97 measurement by flow cytometry in CD34+ bone marrow cells of MDS patients, as well as changes of iROS level in these cells in relation to blast count and iron overload.

3.2 Materials and methods

3.2.1 Study population Peripheral blood was drawn from the median cubital vein of healthy volunteers. For assessment of bone marrow (BM) cells, thirty-eight consecutive BM specimens were obtained by iliac crest aspiration between November 2007 and August 2008 from twenty-four MDS patients. All patients were recruited from the MDS Program at the Odette Cancer Centre; clinical characteristics of these patients are recorded in Table 3. The study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre, and informed consent was obtained from all participants.

3.2.2 Sample collection and processing Both peripheral blood (PB) and bone marrow (BM) specimens were collected in BD Vacutainer blood collection tubes containing lithium heparin (BD-Canada, Mississauga, ON, Canada). The plasma from each specimen was collected after centrifugation at 300 g for 10 minutes. The rest of the specimen was then diluted 1:1 with Ca2+ and Mg2+ free Dulbecco’s phosphate-buffered- saline (PBS) (Life Technologies, Burlington, ON, Canada). Mononuclear cells (MNCs) were purified from the specimen by density gradient centrifugation using Ficoll-Paque PLUS (GE Healthcare Life Sciences, Baie d’Urfe, QC, Canada) according to the manufacturer’s recommendation. Remaining red blood cell (RBC) were lysed by incubating with RBC lysis buffer (Life Technologies) on ice for 10 minutes. The MNCs were washed and suspended in PBS for subsequent assays. All specimens were collected, processed, and assayed within the same day unless indicated otherwise.

3.2.3 Intracellular ROS measurement For iROS measurement, 1x106 MNCs were incubated with 10 µM final concentration DCFH-DA (Sigma-Aldrich, Oakville, ON, Canada) in serum-free RPMI-1640 media (Life Technologies) at o 37 C for 15 minutes in a humidified atmosphere of 5% CO2 in air. DCFH-DA is converted into cell-bound 2',7'-dichlorodihydrofluorescein (DCFH) by cellular esterase and subsequently oxidized by iROS to form the fluorescent compound DCF. The fluorescence intensity of DCF is 98 proportional to the amount of iROS contained in the cells. The cells were washed twice with PBS, and then incubated with 10 µl of anti-human CD34-PE antibody (BD Biosciences, Mississauga, ON, Canada) in 100 µl of PBS at room temperature for 15 minutes in the dark. The cells were washed two additional times with PBS before they were analyzed by flow cytometry using a BD FACSCalibur cytometer (BD Biosciences). All data were acquired with Cell Quest software (BD Biosciences) and analyzed by FlowJo for Mac (Tree Star, Ashland, OR, USA).

3.2.4 Gating criteria for DCF measurement The lymphocyte population from the MNCs was selected based on their forward (FSC) and side (SSC) scatter characteristics using flow cytometry (Figure 3.2.4a) (Vitale et al., 1987). CD34+ cells were selected based on CD34 positivity (Figure 3.2.4b). The identity of these populations was further verified based on CD45 and SSC characteristics (not shown). The iROS of lymphocytes and CD34+ cells was assessed using the mean fluorescence intensity (MFI) of DCF at the FL1 channel. The histograms of iROS of the lymphocyte and CD34+ populations are shown in Figure 3.2.4c/d. All figures and statistical analysis were prepared by GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) or Microsoft Excel 2007.

100

3.3 Results

3.3.1 Experimental variation in iROS measurement The iROS of PB lymphocytes were measured in four healthy volunteers. All specimens were concurrently obtained and processed, and analyzed using the same aliquot of DCFH-DA. The variation in PB lymphocyte iROS between volunteers was small, ranging from 208 to 221 with a mean MFI of 214±5.35 (2.5%, Table 3.3.1a). To assess inter-experimental variation in iROS measurements, PB lymphocyte iROS was measured from a single healthy volunteer subject in 25 separate experiments performed over the course of 35 weeks. Cell processing procedures and flow cytometry settings were kept consistent throughout the study, and DCFH-DA was drawn from separate aliquots of the same original batch. Nevertheless, PB lymphocyte iROS varied considerably, with a range of 141-437 and a mean MFI of 257.52±70.06 (27.2%, Figure 3.3.1). These data demonstrate that inter-experimental variation in iROS is too high to permit reliable comparisons between specimens assayed on different days. The source of this inconsistency is unclear, but may be a consequence of the instability of the DCFH-DA reagent, or small variations in the time between the addition of DCFH-DA to the specimen and the measurement of iROS by flow cytometry. However, the tight clustering in PB lymphocyte iROS among different subjects assayed in a single experiment suggests that concurrent measurement of iROS of freshly obtained PB lymphocytes could be used to normalize for such variation.

In the setting of our research program, where fresh patient specimens become available on an unpredictable schedule, it may sometimes be inconvenient to obtain fresh PB lymphocytes from a healthy volunteer. We therefore investigated other possible normalization standards for the iROS assay. HL-60 cells assayed in four separate experiments showed significant variation of iROS ranging from 442-1513; this heterogeneity persisted even when the iROS of fresh PB lymphocyte was used to normalize the data (Table 3.3.1b). Similar heterogeneity in iROS was observed in NB4 cells (Table 2). Furthermore, ex vivo manipulations of PB lymphocytes from a healthy volunteer by overnight culture and cryopreservation result in substantial increases in iROS relative to fresh PB lymphocytes (data not shown). Hence, cultured cells and manipulated PB lymphocytes are not suitable as normalization controls.

101

Table 3.3.1a. iROS level of PB lymphocytes from four healthy volunteers (all samples were collected, processed and assayed on the same day. One of two representative experiments is shown).

Subject Gender Age PB lym iROS level A M 44 214 B F 39 213 C M 30 221 D M 30 208 Average: 214±5.35 (2.5%)

500 450 400 350 300 Lymphocytes

250 PB

of 200

150 100

iROS level Range = 141 to 437 50 Average iROS level = 257.52±70.06 (27.2%) 0 0 5 10 15 20 25 30 35

Acquisition Date (Week)

Figure 3.3.1. The PB lymphocyte iROS level of a single healthy volunteer. Specimens were obtained over the course of 35 weeks in 25 experiments based on the criteria described in Figure 3.2.4.

102

Table 3.3.1b. iROS level of HL60 and NB4 cell lines.

iROS level Normalized iROS Cell lines iROS level (fresh PB lymphocyte level (to fresh PB (experiment) (cell line) from healthy volunteer) lymphocyte) HL60 (1) 442 136 3.25 HL60 (2) 892 209 4.27 HL60 (3) 1513 226 6.69 HL60 (4) 1412 214 6.60 Average: 5.20±1.72 (33%) NB4 (1) 569 242 2.35 NB4 (2) 671 309 2.17 NB4 (3) 2684 309 8.69 Average: 4.40±3.71 (84%)

103

3.3.2 Measurement of iROS of BM lymphocytes and CD34+ cells We measured iROS in the BM lymphocytes and CD34+ cells of 38 consecutive BM specimens (Table 3.3.2). Each bone marrow aspirate was obtained, processed and analyzed on the same day and the iROS of fresh normal PB lymphocytes from a healthy volunteer subject was measured during each experiment.

MDS patients were separated into two groups according to BM blast count (<5%, 16 aspirates from 16 patients; and 5-19%, 22 aspirates from 12 patients). We considered the possibility of using iROS measurements from each patient’s own bone marrow lymphocyte population to normalize for the iROS of CD34+ cells. Although the iROS of BM lymphocytes correlated with that of normal PB lymphocytes (R=0.553, P<0.0005), this correlation was stronger for patients with low blast count (R=0.652, P<0.01) than those with high blast count (R=0.469, P<0.05). Since the variation is associated with the stage of the disease, we concluded that the BM lymphocyte iROS is insufficiently consistent to be utilized as the normalization control for the measurement of iROS between experiments. We therefore elected to calculate normalized intracellular ROS (nROS) level of BM CD34+ cells with a concurrently determined iROS of fresh normal PB lymphocytes.

Although we observed considerable variation of BM lymphocyte nROS within the low and high blast count groups (Figure 3.3.2a.i), there is no significant difference between the two groups. In contrast, the nROS of BM CD34+ cells in the high blast count group is significantly lower than that of the low blast count group (nROS of 1.325±0.464 vs. 2.156±0.780, unpaired t-test P<0.0005, Figure 3.3.2a.ii). Scrutiny of the shape of the CD34+ cell iROS histogram of the BM cells for each patient revealed that fluorescence intensity is more widely distributed in the BM CD34+ cells of the low blast count group (Figure 3.3.2b.i) than that of the high blast count groups (Figure 3.3.2b.ii). To confirm this notion, we quantified the distribution of fluorescence signal using the robust coefficient of variation (CV, the spread that encompasses the central 68% of the event). BM CD34+ cells of the low blast count group have wider signal distribution than those of the high blast count group (robust CV of 98.78±23.33 vs. 77.61±17.84, unpaired t-test P<0.005, Figure 3.3.2b.iv). In contrast, the iROS measurements in BM lymphocytes of the two blast count groups have similar robust CV (Figure 3.3.2b.iii); suggesting that the difference in iROS distribution is specific to BM CD34+ cells. 104

Table 3.3.2. Patients characteristics.

At sample acquisition Day Sample Patient Blast WHO IPSS from Sex Age WBC HB Platelet Ferritin Cytogenetic Treatment No. No. % Subtype grade previous 1 11 0 M 67 6 102 67 10132 +1,der(1;7)(q10:p10)[4] 15% RAEB‐2Int‐2DFO, neupogen 2 13 0 F 78 2.5 97 233 3173 Complex 6% RAEB‐1Int‐2 3 10 0 M 67 21 119 28 608 +8[19]/46,XY[1] 3% RCMD 4 12 0 M 86 2.1 95 69 169 46XY 15% RAEB‐2Int‐2 5 25 0 M 58 4.5 89 38 147 Complex 12% RAEB‐2 6 14 0 F 70 1.2 82 46 837 Complex 15% RAEB‐2High 7 15 0 F 78 11.5 98 253 1023 del5q13‐q33+8 3% 5q Low 8 6 0 M 69 23.5 92 30 208 46XY 12% RAEB‐2 9 16 0 M 72 8.2 80 753 1974 46XY 2% RARS Low 10 1 0 F 84 9.5 92 335 211 46XX 2% RARS Low 11 11 77 M 68 14.7 102 111 5567 +1,der(1;7)(q10;p10) 14% RAEB‐2Int‐2DFO, neupogen 12 12 63 M 87 1.5 84 62 184 46XY 18% RAEB‐2Int‐2 13 4 0 F 90 8.1 109 397 44 46XX 4% RA Low 14 3 0 M 80 9.8 96 37 1865 46XY 14% RAEB‐2Int‐2 15 5 0 F 62 1.8 103 102 1454 del11(q13;q23) 13% RAEB‐2Int‐2 Decitabine 16 19 0 M 59 2.4 106 24 1411 46XY 3% RCMD Int‐1 17 3 35 M 80 1.6 95 14 2760 46XY 5% RAEB‐1Int‐2 18 18 0 M 52 3 101 73 260 45XX, ‐75%RAEB‐1Int‐2 19 6 84 M 69 28.5 87 24 218 46XY 13% RAEB‐2 20 14 119 F 71 0.9 76 3 3213 Complex 5% RAEB‐1Int‐2Revlimid 21 17 0 F 70 5.4 89 331 723 46XX 2% RCMD‐RS Int‐1eprex, neupogen 22 11 154 M 68 4.4 98 36 5641 Complex 18% RAEB‐2HighDFO, revlimid 23 20 0 M 74 38.1 74 11 4033 Complex 12% RAEB‐2High 24 18 35 M 52 2.5 101 105 260 45,XY,‐76%RAEB‐1Int‐2 25 2 0 M 81 3.5 97 62 179 46XY 2% MDS‐UInt‐1 26 8 0 M 69 2.7 99 36 4145 45,XY,‐78%RAEB‐1Int‐2 27 21 0 F 46 5.2 104 415 471 46XX 3% RCMD Low 28 22 0 M 72 2.7 99 15 526 46,XY,+Y,‐74%MDS‐UInt‐1 29 27 0 M 74 5 87 183 465 45,XY,‐79%RAEB‐1Int‐2 30 11 210 M 68 7 92 48 5264 Complex 15% RAEB‐2HighDFO, Revlimid 31 9 0 M 77 4.8 135 161 584 Complex 1% RCMD Int‐1 Decitabine 32 22 35 M 72 1.9 91 3 1080 46XY 2% MDS‐UInt‐1 33 23 0 M 81 2.4 89 12 1463 46,XY[20] 3% RCMD Int‐1 34 8 63 M 69 6 106 22 5717 45,XY,‐7[20] 7% RAEB‐1Int‐2Revlimid 35 5 168 F 62 2.3 87 125 582 Complex 6% RAEB‐1Int‐1 36 7 0 M 84 2.9 103 96 19 46,XY[20] 3% RCMD Int‐1 37 26 0 M 76 3.7 123 86 245 46,XY[21] 3% RA Low 38 24 0 F 60 3.8 110 173 885 Complex 3% RCMD‐RS Int‐1

105

(i) (ii) P<0.0005 (T‐Test) 2.0 4 level) level)

1.5 3 iROS iROS

1.0 2 (Normalized 0.5 (Normalized

nROS + BM LymphocytesnROS BM CD34 Cells 0.0 0 <5% 5‐19% <5% 5‐19%

Blast Count Blast Count

Figure 3.3.2a. Normalized iROS level (nROS). (i) Bone marrow (BM) lymphocytes and (ii) CD34+ cellsinlowblast(<5%)andhighblast(5‐19%) MDS patients. nROS is calculated by dividing patient’s cell iROS with volunteer’s lymphocyte iROS. Data are expressed as mean with standard deviation (SD).

106

107

3.3.3 Iron overload and iROS in MDS Free iron catalyzes the generation of iROS such as hydroxyl radicals via the Haber-Weiss reaction. Ghoti et al. have reported a correlation between serum ferritin levels of low-risk MDS patients and ROS in peripheral blood erythrocytes and platelets (Ghoti et al., 2007). However, the correlation is not observed in lymphocytes or polymorphonuclear cells and it is possible that iron induced ROS generation may depend on microenvironment and cell type. We therefore examined the nROS of BM lymphocytes and CD34+ cells in the context of iron overload. In high blast MDS patients, serum ferritin correlated with nROS of BM CD34+ cells (R=0.613, P<0.005, Figure 3.3.3b) but not lymphocytes (data not shown). In low blast MDS patients, we did not observe any significant correlation between serum ferritin and nROS of BM CD34+ cells (R=0.345, P=0.191, Figure 3.3.3a). Nonetheless, the serum ferritin levels are higher and more widely distributed for high blast (range 147-10132 µg/L, mean 2558.8±2676.4 µg/L) than low blast MDS patients (range 19-1974 µg/L, mean 715.4±560.2 µg/L) (Student’s t-test with unequal variance P<0.005). It is possible that there are insufficient high serum ferritin data from low blast MDS patients to establish a correlation with nROS of BM CD34+ cells.

If iron overload contributes to nROS elevation in MDS, then iron chelation therapy ought to result in reduction of nROS in BM CD34+ cells. To examine this notion, we measured BM CD34+ nROS level and serum ferritin level in one RAEB-2 patient over the course of 210 days, during therapy with deferoxamine for transfusion-related iron overload. As expected, both serum ferritin level and nROS in BM CD34+ cells decreased after institution of iron chelation therapy, and the time course and magnitude of the decrease of serum ferritin and nROS level are comparable (Figure 3.3.3c).

108

4 level)

3 iROS

1 (Normalized BM CD34+ cells (<5% blast) N=16, R=0.345, P=0.191

nROS 0 0 500 1000 1500 2000 2500 Serum Ferritin Level (g/L)

Figure 3.3.3a. nROS of CD34+ cells vs. serum ferritin level of low blast count patients.

4 level)

3 iROS

1 (Normalized

BM CD34+ cells (5‐19% blast) N=22, R=0.613, P<0.005

nROS 0 0 2000 4000 6000 8000 10000 12000 Serum Ferritin Level (g/L)

Figure 3.3.3b. nROS of CD34+ cells vs. serum ferritin level of high blast count patients.

109

12000 3 DFO nROS 10000 2.5 g/L)

(Normalized μ ( 8000 2 ‐35% ‐33% Level 6000 ‐47% 1.5

‐45% ‐44% ‐48% iROS Ferritin 4000 1

MFI)

Serum 2000 Serum ferritin level 0.5 CD34+CD34+ cellscells nROSnROS 0 0 0 50 100 150 200 Day

Figure 3.3.3c. Effects of iron chelation therapy (deferoxamine, DFO, 500 mg/d, treatment initiated at day 0) on serum ferritin level and nROS level of CD34+ cells.

110

3.4 Discussion Measurement of iROS by fluorescence activation of DCFH-DA is a simple and widely used technique to measure the redox state in live cells, but adaptation of the DCFH-DA assay for application in a clinical setting must be done with care. We show that the direct fluorescent intensity readout of the assay can vary by 50% between experiments. However, there is strong intra-experimental agreement among the iROS of peripheral blood lymphocytes of normal subjects. Therefore, the inter-experimental variations in DCF levels are not likely due to variations between the nROS levels of lymphocytes from normal individuals. As a result, in order for iROS levels of specimens assayed on different dates to be comparable, they must be normalized against an internal or external standard such as normal lymphocytes from healthy volunteers.

Although it would be convenient to utilize each patient’s own bone marrow lymphocytes as a normalization standard, this is not a suitable approach for normalization, due to the lower correlation between the iROS measurements of these lymphocytes with normal lymphocytes. This lack of correlation is especially true for MDS patients with high blast count, and may be due to a disrupted bone marrow microenvironment or dysplasia of the lymphocytes, some of which may derive from the MDS clone (Disperati et al., 2006). The use of cell lines for normalization cannot be recommended due to the variable nature of iROS in these cells, possibly due to a number of varying factors such as cell growth phase and culture condition. Furthermore, ex vivo manipulations such as culture and cryopreservation generally increase the iROS level of lymphocytes and may unpredictably affect the iROS readout.

There are differences in magnitude and distribution of nROS levels that distinguish BM CD34+ cells of MDS patients with low blast count (<5%) and high blast count (5-19%). In patients with low blast count nROS levels were significantly higher than in patients with high blast count. Since oxidative stress is strongly associated with progression of malignancy this may seem counterintuitive (Sesti et al., 2012). It is, however, consistent with previous observations on energy metabolism in MDS. In high blast MDS and in AML, Akt signalling is activated (Kubota et al., 2004; Nyakern et al., 2006), indicating a reliance on glycolysis (Herst et al., 2011), whereas in low blast MDS Akt is inactive and mitochondrial oxidative phosphorylation predominates. Since mitochondrial respiration is the principal source of reactive oxygen 111 intermediates, from which the production of ROS proceeds, our observation of higher nROS in low blast MDS is unsurprising. Additionally, it has been demonstrated that functionally defined leukemia stem cells (LSCs) have relatively low ROS level due to the over-expression of BCL-2 and glutathione peroxidise 3 (Herault et al., 2012; Lagadinou et al., 2013). One might speculate that, in high blast MDS, the biology of the CD34+ cell population comes to resemble that of the LSC in regard to ROS metabolism. Indeed, the capacity for enhanced ROS defense may be a characteristic selected for in the clonal evolution of MDS.

We also observed greater heterogeneity in CD34+ cell nROS in low blast MDS patients, as reflected in the high robust CV in these patients. The CD34 antigen is expressed on the surface of various hematopoietic stem cells (HSCs) and progenitors, which exhibit a range of metabolic activities and oxygen availability in their corresponding bone marrow niches. For example, the ROS level of primitive HSCs is lower than other HSCs (Jang and Sharkis, 2007). Therefore, when viewed as a single population in the bone marrow of low blast MDS patients, CD34+ cells display a wider nROS distribution than the corresponding population in high blast MDS patients, reflecting the presence of a mixture of cell types. We conjecture that during the course of clonal evolution to a more aggressive (high blast count) form of MDS, this natural heterogeneity is lost as the CD34+ population becomes homogeneous. Furthermore, it may be that enhanced antioxidant defense protein expression and reliance on glycolysis rather than mitochondrial oxidative phosphorylation are traits selected for in clonal evolution.

Xiao et al have recently reported higher ROS levels in high-risk than in low-risk MDS patients, a finding that is inconsistent with our observations (Xiao et al., 2013). The differences in these findings may be due to methodological differences, including failure to normalize iROS with an external standard and flow cytometry techniques.

The correlation between serum ferritin and nROS in terminally differentiated peripheral blood cells in the context of MDS has been documented (Ghoti et al., 2007; Saigo et al., 2011). However, to probe the mechanisms of the effects of ROS and iron overload on hematopoiesis and disease progression in MDS it is necessary to interrogate the bone marrow HSC/progenitor compartment. We observed a significant correlation between serum ferritin and nROS in the CD34+ cells of the high blast count patients but not in their lymphocytes, consistent with the prevailing model in which free iron enters the HSCs and progenitors and subsequently increases 112 oxidative stress in these cells via Fenton chemistry. Since both free iron and oxidative stress are known to cause mutagenesis and apoptosis (Lu et al., 2013; Valko et al., 2004), our finding supports the notion that iron overload may accelerate AML progression and worsen the symptoms of cytopenias in MDS by damaging the HSCs and their progenitors via ROS. Moreover, the time course of the decrease in CD34+ cell nROS observed after institution of iron chelation therapy is consistent with the reported time course of ICT-mediated hematological improvement in MDS patients (Jensen et al., 1996; Nolte et al., 2013), suggesting ROS reduction as a possible mechanism of this clinical benefit.

There are limitations that need to be addressed in future studies. The retrospective nature of this study makes our observations susceptible to certain biases and confounding variables. Our cohort was recruited from a single centre therefore referral and selection biases are possible. In addition, our limited sample size may be vulnerable to the effects of outliers. Finally, DCF is a general detector of several types of ROS (Chen et al., 2010; Karlsson et al., 2010), and therefore may not be sufficient to elucidate the contribution of specific ROS, such as hydroxyl radicals, which are the products of the Haber-Weiss reaction. We are currently in the process of optimizing our ROS detection technique along with other methods of assessing cellular damage, and integrate them into current clinical diagnostic procedures for MDS.

In conclusion, ex vivo assessment of intracellular ROS using DCF requires normalization in order to allow inter-experimental comparison. By normalizing iROS measurements with those of fresh normal lymphocytes, we determined that the normalized ROS level of bone marrow CD34+ cells is lower in MDS patients with high blast count than those with low blast count. In MDS patients with high blast count, the oxidative stress level of the CD34+ cells is also associated with iron overload. Finally, both iron burden and nROS level can be reduced by iron chelation therapy.

113

Chapter 4

Secondary iron overload promotes radiation-induced acute myeloid leukemia in B6D2F1 mice

114

4.1 Introduction

Anemia – leading to the need for regular RBC transfusion – is common in myelodysplastic syndrome (MDS). Chronic red blood cell (RBC) transfusion leads inevitably to secondary hemochromatosis, resulting in a predictable pattern of end-organ damage that is mediated at the cellular level by iron-catalyzed generation of reactive oxygen species. To mitigate the risk of iron-related morbidity and premature mortality, iron chelation therapy (ICT) is recommended in transfusion dependent iron-overloaded patients with lower risk MDS, even though the evidence that ICT is effective in influencing these outcomes is weaker than it is for patients with thalassemia major (Gattermann, 2008).

In addition to iron toxicities to the liver, heart, endocrine organs, and skin seen in thalassemia major, it has also been suggested, on the basis of retrospective reviews of registry data, that in MDS iron overload may promote the development of acute myeloid leukemia (AML) (Leitch, 2011). Although an excess incidence of AML is not seen in thalassemia major, such an association in MDS may nonetheless be plausible. Hereditary hemochromatosis is associated with several forms of human carcinomas in liver, lung and colon (Fargion et al., 2010; Toyokuni, 2009), establishing the principle that iron can promote cancer development. In addition, it must be borne in mind that MDS – a clonal myeloid disorder that is intrinsically linked to AML development and that is characterized by genomic instability – can be expected to be more susceptible than thalassemia major to an AML-promoting effect.

Proof that iron overload promotes the progression of MDS to AML would have a profound effect on the aims and breadth of ICT in this disease, and could have a positive impact in both low and high risk MDS. In this chapter a radiation-induced AML (RI-AML) mouse model is utilized to prove the principle that extrinsic iron overload can promote AML development.

4.1.1 Animal models of iron overload and ICT

Animal models of iron overload can be created by various means depending on the type of hemochromatosis they are intended to investigate. Primary hemochromatosis can be genetically induced by the knockout of the human hemochromatosis (Hfe) gene, which suppresses hepcidin expression and allows excess intestinal iron absorption (Gao et al., 2009a; Zhou et al., 1998). 115

Hfe-/- mice exhibit elevated transferrin saturation and liver iron content. However, unlike primary hemochromatosis in humans, these mice do not develop liver fibrosis or cirrhosis. This may be due to the ability of rodents to produce ascorbic acid that provides some protection against excess iron, whereas humans cannot synthesize ascorbic acid (Chatterjee et al., 1975). Dietary iron supplementations by carbonyl iron, ferrocene, and ferric ammonium citrate induce a state that resembles primary hemochromatosis (Barton and Edwards, 2000). A diet enriched for carbonyl iron increases intestinal iron absorption and iron accumulation in hepatocytes, resulting in lipid peroxidation, impairment of organelle functions, and hepatic fibrosis.

Parenteral injection of iron dextran provides an in vivo model that resembles secondary iron overload as a result of repeated transfusion (Barton and Edwards, 2000). In addition, iron loading by injection permits a more precise and controlled iron dosage when compared to dietary iron. Iron dextran is an iron complex with stable pH and low toxicity that is used for the management of iron-deficiency anemia (Auerbach and Ballard, 2010). Other iron complexes are sometimes used in animal models, including iron sorbitol and ferric nitrilotriacetate. Iron dextran is initially absorbed by the reticuloendothelial system and then redistributed into the parenchymal cells of the liver and other organs. Mice are able to remove excess body iron actively after iron dextran injection, although the same mechanism does not exist in humans (Musumeci et al., 2014). Iron accumulation and damage have been reported in the liver and heart (Bartfay et al., 1999; Moon et al., 2011). Generation of reactive oxygen species (ROS) and oxidative damage of cellular components have also been documented upon iron loading (Bartfay and Bartfay, 2000; Bartfay et al., 2000; Gao et al., 2009b; McCullough and Bartfay, 2007). In the hematopoietic tissues, iron dextran injection results in iron accumulation and impairment of the bone marrow (BM) hematopoietic microenvironment (Okabe et al., 2014). Iron overload also disrupts hematopoiesis by decreasing the formation of various hematopoietic colony forming units, but the impact on peripheral blood counts appears to be minimal (Chai et al., 2014; Chai et al., 2013).

Iron homeostasis is influenced by multiple genes and the expression of these genes is dependent on numerous factors. In the mouse, genetic variations between different strains modulate the expression of important genes for iron metabolism including hepcidin, ferritin, and iron regulatory protein (Clothier et al., 2000; Courselaud et al., 2004). Genetic background determines the severity of iron overload in the Hfe knockout mouse model, in which Hfe-/- 116

C57BL/6 mice have lower non-heme liver iron accumulation when compared to AKR mice (Fleming et al., 2001). Wild-type C57BL mice (/6 and /10) also have the lowest basal non-heme iron content among several inbred mouse strains, and are more highly resistant than SWR mice to iron-dependent toxicity by diquat (Clothier et al., 2000). In addition, the concentration of dietary carbonyl iron required to saturate transferrin completely is ten times greater in C57BL/6 mice than DBA/2 and AKR mice (Leboeuf et al., 1995). Therefore, the C57BL strain is more resistant to iron overload than other common laboratory inbred mouse strains. For this reason, B6D2F1 mice, a hybrid between C57BL/6 female and DBA/2 male mice, are routinely used to evaluate the effects of iron overload on cardiac and hepatic functions (Bartfay and Bartfay, 2000; Bartfay et al., 1999; McCullough and Bartfay, 2007). B6D2F1 mice have also been shown to be a suitable model for the study of oxidative stress-mediated pathogenesis such as impaired endothelium-dependent dilation (Lesniewski et al., 2009). Other factors that modulate iron status include age, gender and tissue type (Hahn et al., 2009). Brain, liver and heart iron contents increase by 30-70% for aged mice when compared to young adults. Male BALB/c and DBA/2J mice have higher iron content in the brain and retina, while female mice from these strains have higher liver iron. The liver also has higher iron level than other organs such as the heart and brain on a per weight basis.

Iron is a known carcinogen and has been shown to induce or promote tumors (Beguin et al., 2014). Parenteral iron administration by iron dextran or ferric nitrilotriacetate induces sarcoma at the site of injection in animal models and humans. Other types of tumors, including renal cell carcinoma, mesothelioma, and papillomas have also been reported. In addition, excess dietary or parenteral iron promotes carcinogenesis in conjunction with cancer inducing agents that are chemical, genetic or surgical in nature. The potential of iron as a tumor promoter has been shown in various types of cancer in the liver, skin, GI tract, breast, oral cavity, and lung. Furthermore, iron loading facilitates tumor proliferation both in vivo and in vitro. In the context of leukemia, parenteral iron loading in DBA/2 mice at levels comparable to clinical doses for humans promotes in vivo proliferation of implanted L1210 lymphocytic leukemia cells and shortens the lifespan of the inoculated animals (Bergeron et al., 1985). Iron may contribute to carcinogenesis via several mechanisms, including induction of ROS and oxidative damage, altered gene

117 expression and signaling pathways, and suppression of immune surveillance against cancer (Beguin et al., 2014; Bystrom et al., 2014).

ICT by deferoxamine, deferiprone, or deferasirox has been shown to reduce iron burden and provide protection against iron-induced damage in iron-overloaded animals (Ikeda et al., 2014; Nick et al., 2009; Yatmark et al., 2014). ICT effectively reduces redox-active non-transferrin bound iron (NTBI), which may be responsible for the reversal of iron-induced cell damage and inflammation (Yatmark et al., 2014). In addition, the proliferation of malignant cells is inhibited in iron deficient or iron chelated mice (Hann et al., 1988; Lui et al., 2013), suggesting that ICT may be a viable strategy for the treatment of cancer and leukemia. Malignant cells require a greater amount of iron than normal cells to sustain rapid proliferation (Elford et al., 1970; Heath et al., 2013). Iron depletion disproportionally disrupts iron homeostasis in malignant cells and interrupts multiple cellular processes including cell cycle, cellular metabolism, and cell growth (Heath et al., 2013). Moreover, iron chelation inhibits cancer-associated Wnt signaling (Coombs et al., 2012; Song et al., 2011), and deferasirox has the additional capability to suppress NF-κB activity (Frelin et al., 2005).

In summary, iron overload adversely affects normal physiology and causes or promotes carcinogenesis. ICT has been shown to reduce iron burden and relieve the adverse effects of iron overload, including neoplasia. Several animal models have been established to investigate the in vivo effects of iron loading and ICT on crucial organs such as the heart and liver. However, there are limited animal studies that assess the effects of iron on the hematopoietic tissues, especially in the context of leukemogenesis.

4.1.2 Animal models of radiation-induced AML

Exposure to gamma and X-ray irradiation are known risk factors for MDS and leukemia in humans (Pedersen-Bjergaard et al., 2002). Solid epidemiological data of radiation-induced leukemia have been obtained from a number of cohorts, including the Japanese atomic bomb survivors, Chernobyl disaster survivors, and patients who received therapeutic radiation for cancers (Boice et al., 1988; Ivanov et al., 1997; Little et al., 1999; Noshchenko et al., 2010; Preston et al., 1994). The risk of leukemia development is highest within ten years after exposure. Younger children between 5 and 9 years of age are vulnerable to develop acute 118 lymphocytic leukemia as a result of radiation exposure, while acute or chronic myeloid leukemias are the most common radiation-induced malignancy in older children and adults. Radiation exposure is also associated with MDS development among Nagasaki atomic bomb survivors (Iwanaga et al., 2011). Radiation-induced MDS has long latency at 40 to 60 years after exposure, and the MDS rate increases inversely with exposure distance.

As in humans, radiation exposure also increases AML risk in several mouse strains including RF, SJL/J, CBA, and C3H/He (Rivina et al., 2014). Conversely, C57BL/6 mice are resistant to RI- AML (Boulton et al., 2001). RI-AML is triggered by single dose total body irradiation at approximately 3 Gy on mice at around 10 weeks of age. Morphological features of RI-AML include hepatomegaly, splenomegaly and leukemic infiltrations in the bone marrow, spleen and liver (Rithidech et al., 1999). The latency and frequency of RI-AML varies among different strains. RF mice are more sensitive and 50-90% of them begin to develop RI-AML at 4-12 months post-irradiation. Approximately 25% of SJL/J, CBA and C3H mice develop RI-AML at 1.5-24 months post-irradiation. The incidence of RI-AML in SJL/J mice can be increased to 70% by the administration of corticosteroids or colony-stimulating factor-1 (CSF-1) after irradiation (Haran-Ghera et al., 1997; Resnitzky et al., 1985), while calorie restriction reduces RI-AML incidence by more than half in C3H/He mice (Yoshida et al., 1997). Moreover, 2-4% of RF mice develop spontaneous myeloid leukemia, while SJL/J and C3H/He mice are at high risk to develop spontaneous reticulum cell neoplasms and hepatomas, respectively (Rivina et al., 2014). In contrast, the CBA strain has a low spontaneous frequency of malignant transformation, making it a favorable RI-AML model for human AML (Rithidech et al., 1999).

Radiation-dependent leukemogenesis follows a defined pattern in reported mouse strains that are sensitive to RI-AML (RF, C3H/He, CBA, and SJL/J) (Rivina et al., 2014). Leukemia-initiating cells containing partial deletion of chromosome 2 emerge as early as 24 hours after irradiation in mice that are sensitive to RI-AML (Hayata et al., 1983; Olme et al., 2013). Frequent aberrations are also noted on chromosome 2 in sensitive mouse strains when compared to the resistant strains (Darakhshan et al., 2006). The common deleted region on chromosome 2 in RI-AML encompasses the Sfpi1 gene, whose gene product, PU.1, is a key protein involved in the regulation of hematopoiesis (Burda et al., 2010). Conversely, PU.1 deleted pre-leukemic cells do not persist in RI-AML resistant C57BL/6 mice (Peng et al., 2009). In addition, long-lived 119 inflammatory signals involving the upregulation of TNFα and Fas ligand are noted in the BM microenvironment of CBA/Ca mice, causing genomic instability of the pre-leukemic cells as well as normal BM cells (BMCs) (Lorimore et al., 2011). Interestingly, long-lived inflammatory signals are also observed in the BM of irradiated C57BL/6 mice, but the BMCs in these mice undergo apoptosis instead of undergoing mutagenesis (Lorimore et al., 2011). Moreover, irradiated BM stromal cell lines release the cytokine CSF-1, which may promote RI-AML by inducing the expansion of pre-leukemic cells (Greenberger et al., 1992a; Greenberger et al., 1992b). Dexamethasone treatment increases CSF-1 production in the BM (Tartakovsky et al., 1993), suggesting corticosteroids may promote RI-AML via CSF-1. Corticosteroids may also protect the pre-leukemic cells from apoptosis by sequestering p53 in the cytoplasm and thereby suppressing its activation (Sengupta et al., 2000), although corticosteroids can induce apoptosis in certain cell types and contexts (Greenstein et al., 2002; White and Dorscheid, 2002).

Leukemia-initiating cells with chromosome 2/PU.1 deletion continue to survive after irradiation, acquire further mutations, and become leukemic over time (Dekkers et al., 2011). The second copy of Sfpi1 is often mutated during radiation-induced leukemogenesis, resulting in further inactivation of the PU.1 protein. Mutation or suppression of PU.1 blocks myeloid differentiation and is associated with AML and MDS (Laricchia-Robbio et al., 2009; Mueller et al., 2002). Knockdown of PU.1 by 80% in mice promotes AML within 6 months of age, and induces aggressive AML within 1-2 months when the p53 alleles are also removed (Basova et al., 2014). In addition, numerical changes involving the Y chromosome are frequently observed in RI-AML (Hayata et al., 1983), and several loci on chromosome 8, 13, and 18 may be involved in radiation-induced leukemogenesis (Darakhshan et al., 2006).

Although RI-AML has been shown in multiple mouse strains, we are not aware of any previous study that evaluates the effect of irradiation on AML development in B6D2F1 mice. Darakhshan et al. reported the susceptibility of the parental strains of B6D2F1 mice, C57BL/6 and DBA/2, to RI-AML based on the sensitivity of chromosome 2 to X-ray induced aberration (Darakhshan et al., 2006). Both of these strains are classified as non-sensitive resistant group. The C57BL/6 strain is known to be resistant to RI-AML because the leukemia-initiating cells that harbor the Sfpi1 gene deletion on chromosome 2 do not persist after irradiation (Peng et al., 2009). Conversely, the DBA/2 strain is only marginally resistant to RI-AML based on chromosome 2 120 sensitivity, and the strain is closely related to the RI-AML sensitive CBA/H strain. The resistance of C57BL/6 strain to RI-AML is not a dominant trait since its F1 hybrid with CBA/H strain has radiation sensitive chromosome 2 (Darakhshan et al., 2006). In addition, administration of other factors, such as dexamethasone and CSF-1, increases RI-AML incidence and reduces its latency (Haran-Ghera et al., 1992). It is reasonable to expect that both B6D2F1 and DBA/2 strains are vulnerable to develop AML upon irradiation with dexamethasone treatment.

In summary, radiation is a potent inducer of AML in a number of mouse strains. The development of RI-AML can be modulated by extrinsic factors such as dexamethasone injection and calorie restriction. Excess iron is a known carcinogen, while ICT has been shown to reduce iron accumulation and possess anti-cancer properties. Hence, iron loading and ICT may participate in leukemogenesis and influence the risk of RI-AML in animal models. In this study, we selected B6D2F1 mice to investigate the effects of iron on RI-AML. The B6D2F1 mouse is a well-established animal model for secondary iron overload, with reasonable expectation to be susceptible to RI-AML.

4.1.3 Hypothesis

In this study, we investigate the effects of secondary iron overload and ICT in modulating leukemogenesis in a pre-leukemic animal model. We hypothesize that parenteral iron injection in mice results in iron accumulation in the BM. In addition, we hypothesize that excess iron in the BM of irradiated B6D2F1 mice promotes AML development by increasing oxidative stress and disrupting important signaling pathways in the BMCs. Conversely, we hypothesize that ICT using deferasirox reduces AML risk by lowering iron burden in the iron-loaded irradiated animals.

4.2 Methods

4.2.1 Animals

Male C57BL/6 and B6D2F1 mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA) and Charles River Canada (St. Constant, QC, Canada), respectively. The B6D2F1 mice are genetically and phenotypically uniformed F1 hybrid between C57BL/6 female and DBA/2 male 121 inbred mice (Lewis et al., 1985). All mice were housed in unisex groups of 5 or less per cage in a temperature and humidity controlled room maintained on a 12:12h light/dark cycle in a pathogen-free facility at Sunnybrook Research Institute. The mice had access to food and water ad libitum. Standard rodent diet (Teklad 2018, Madison, WI, USA) containing 200 mg/kg ferrous sulfate was given as the sole food source. All experimental protocols were conducted with the approval of the Animal Care Committee of Sunnybrook Research Institute and in compliance with the guidelines established by the Canadian Council on Animal Care and the Animals for Research Act of Ontario.

4.2.2 Treatments

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless stated otherwise. Upon arrival, all mice were allowed to adapt to their surroundings for at least 3 weeks before receiving any treatment. For radiation treatment, nine week old B6D2F1 mice were subjected to non-lethal total-body irradiation at 300 cGy delivered in a Cs-137 small animal irradiator. Several hours after irradiation, the mice were inoculated with 0.5 mg dexamethasone sodium phosphate (Omega Laboratories, Montreal, QC, Canada) by subcutaneous injection. Nonlethal radiation has been shown to induce AML in multiple mouse strains with a latency period of approximately one year post-irradiation (Rivina et al., 2014). Dexamethasone injection after irradiation increases AML rate and decreases the latency in SJL/J mice (Haran-Ghera et al., 1992). Mice began to receive iron or sham treatment at 2 weeks after irradiation. For mice that were not assigned to receive radiation, iron loading began after they had adapted to their surroundings. Iron dextran or the equivalent dose of dextran (from Leuconostoc spp., Mr ~6,000) diluted by phosphate-buffered saline (PBS) up to a total volume of 200 µl, was delivered by intraperitoneal injection for 5 days per week. For low dose iron loading between 5 to 20 mg, 1 mg of iron dextran was injected per day for 5 to 20 days over 1 to 4 weeks. High dose iron loading at 150 mg required 10 mg iron dextran injection per day for 15 days over 3 weeks. Iron chelation therapy was initiated after the end of iron loading. Deferasirox (Novartis, Dorval, QC, Canada) was suspended in 0.5% hydroxypropylcellulose (a gift from Nippon Soda Co. Ltd, Tokyo, Japan) and administered by oral gavage at 10 or 40 mg/kg/day for 7 days a week over 4 or 8 weeks.

122

4.2.3 Monitoring and analysis

The body weight of all irradiated B6D2F1 mice was measured at least once per week for the duration of the experiment. Overt leukemia was suspected when the subject lost 20% of its body weight, showed signs of illness, and presented leukemic blasts in its tail vein peripheral blood smear. Diagnosis of AML was made based on the Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice (Kogan et al., 2002). Mice were sacrificed when they became ill or earlier when they were assigned for early analysis. Tissue samples, including PB, hind limb bones, spleen, liver, and heart, were collected and analyzed. PB count was measured using an analyzer by the Sunnybrook Health Science Centre Hematology Laboratory Service. PB morphology was evaluated by May-Grünwald-Giemsa staining of air-dried smear. At least one of the hind limb bones and other harvested organs were fixed for 24 hours in neutral buffered 10% formalin solution followed by decalcification for the bones, paraffin embedding, slicing, and staining with hematoxylin and eosin by the Sunnybrook Research Institute Histology Core Facility. Unstained slices from the paraffin embedded tissue blocks were used to evaluate iron loading by the iron stain kit (Prussian blue and pararosaniline staining) according to manufacturer’s instructions.

4.2.4 Flow cytometry

All antibodies were purchased from BD Biosciences or eBioscience (San Diego, CA, USA) unless stated otherwise. Flow cytometry was performed with a FACSCalibur cytometer (BD Biosciences, Mississauga, ON, Canada). All data were acquired with Cell Quest software (BD Biosciences) and analyzed with FlowJo for Mac (Tree Star, Ashland, OR, USA). All specimens were collected, processed, and assayed within the same day unless indicated otherwise. BMCs for each mouse were obtained from the remaining hind limb bones by flushing the medullary cavity with PBS containing 2% fetal bovine serum (FBS, Life Technologies, Burlington, ON, Canada) using a syringe attached to a 25 gauge needle (BD Diagnostics, Mississauga, ON, Canada). Mature RBC from the BMCs were lysed using ACK lysing buffer (Life Technologies), washed twice and resuspended in PBS for subsequent procedures. An aliquot of BMCs was prepared for cytospin and stained with May-Grünwald-Giemsa stains for the evaluation of cell morphology. The BMCs were analyzed by flow cytometry to distinguish different cell

123 populations. The BMCs were stained with a panel of antibodies using standard staining techniques: anti-CD45-APC versus lineage markers (biotin conjugated anti-CD3e, anti-B220, TER119, anti-CD11b, and anti-Gr-1 attached to FITC conjugated streptavidin) (Stelzer et al., 1993; Weissman and Shizuru, 2008); myeloid lineage (anti-CD11b-APC) versus lymphoid lineage (biotin conjugated anti-CD3e and anti-B220 attached to FITC conjugated streptavidin); myeloid cell profile (anti-CD11b-APC and anti-Gr-1-FITC) (Lagasse and Weissman, 1996); erythroid progenitor profile (anti-CD71-FITC and biotin conjugated TER119 attached to APC conjugated streptavidin) (Koulnis et al., 2011). Apoptosis was determined based on the annexin V+/7AAD- population as a percentage of 7AAD- population measured by flow cytometry using the PE annexin V Apoptosis detection kit (BD Biosciences) according to manufacturer’s recommendations. Intracellular ROS (iROS) level was determined using dichloro-dihydro- fluorescein diacetate (DCFH-DA) (Bass et al., 1983). BMCs (1 x 106 cells) were washed with PBS and resuspended in serum-free RPMI 1640 media (Life Technologies). DCFH-DA was added to the BMCs for a final concentration of 10 µM and incubated at 37oC for 15 minutes in a humidified atmosphere of 5% CO2 in air. The cells were then washed with PBS, stained by anti- CD45-APC antibody at room temperature for 15 minutes in the dark and analyzed by flow cytometry.

To detect intracellular antigens, BMCs were stained with FITC or APC conjugated anti-CD45 antibodies. The cells were then fixed and permeabilized using the IntraPrep permeabilization reagent (Beckman Coulter, Mississauga, ON, Canada) according to manufacturer’s recommendations. For intracellular staining, the following antibodies were used: Alexa Fluor 647 conjugated anti-phospho-Akt (pAkt) (Ser473) (Cell Signaling Technology, Deverly, MA, USA); anti-phospho-Foxo3a (pFoxo3a) (Ser318/321) (Cell Signaling Technology) with PerCP- Cy5.5 conjugated secondary antibody (Santa Cruz Biotechnology, Dallas, TX, USA); anti- active-β-Catenin (dephosphorylated Ser37/Thr41) (Millipore Canada, Etobicoke, ON, Canada) with PE conjugated secondary antibody (Santa Cruz Biotechnology); and FITC conjugated anti- phospho-Histone H2A.X (γH2AX) (Ser139) (Millipore Canada). Anti-CD45-FITC stained BMCs were used for pAkt and pFoxo3a staining. Anti-CD45-APC stained BMCs were used for active-β-Catenin and γH2AX staining. The cells were incubated overnight at 4oC and analyzed by flow cytometry. Expression of intracellular antigens was measured in CD45+ BMCs. For early

124 analysis, the cohort was separated into smaller groups and analyzed over several days. An untreated C57BL/6 mouse was attached to each group and analyzed on the same day. To correct for inter-experimental variation, the expression level of the subjects from each group was normalized to the corresponding expression level of the attached untreated C57BL/6 mouse. The expression level of each subject was further normalized to the mean expression level of the control group.

4.2.5 DNA, RNA and protein isolation

The remaining BMCs were cryopreserved in freezing medium consisted of 10% dimethyl sulfoxide (DMSO) and 90% FBS. DNA, RNA and protein were isolated from the cryopreserved BMCs at a later time using the AllPrep DNA/RNA/Protein mini kit (QIAGEN, Toronto, ON, Canada) according to manufacturer’s instructions. Upon isolation, genomic DNA and total RNA were eluted into Buffer EB and RNase-free water, respectively. Their quantity and purity were measured by a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) using standard methods. Total protein was dissolved in 5% SDS solution, and quantified by a NanoDrop 2000c using the BCA protein assay kit (Thermo Fisher Scientific). All specimens were stored at -70oC until further use.

4.2.6 Oxidative DNA damage (AP sites)

Apurinic/apyrimidinic (AP) site was measured from the genomic DNA of the BMCs using the OxiSelect oxidative DNA damage quantitation kit (Cell Biolabs, San Diego, CA, USA). The assay utilized an aldehyde reactive probe to react specifically with an aldehyde group on the open ring form of AP sites on isolated DNA. The AP sites were then tagged with biotin and attached to enzyme conjugated streptavidin. The processed sample was then subjected to enzymatic reaction and the absorbance at 450 nm was measured using an EL800 absorbance reader (BioTek, Winooski, VT, USA). The quantity of AP sites in DNA sample was calculated by comparing its absorbance with a standard curve generated from the provided DNA standard containing predetermined AP sites.

125

4.2.7 Quantitative RT-PCR

The RT2 first strand kit (QIAGEN) was used to eliminate genomic DNA contamination and reverse transcribe 400 ng of total RNA into cDNA. The cDNA was then mixed with RT2 SYBR Green qPCR mastermix (QIAGEN) and aliquoted into a single RT2 Profiler PCR Array (QIAGEN). The following PCR arrays were used: mouse PI3K-Akt signaling pathway (PAMM- 058Z), mouse Wnt signaling pathway (PAMM-243Z), and mouse oxidative stress (PAMM- 065Z). Each array assesses the expression of 84 genes related to the pathway in question, along with corresponding housekeeping genes and controls. Real-time quantitative polymerase chain reaction (PCR) was performed by a ViiA 7 real-time PCR system (Life Technologies) using recommended settings supplied by QIAGEN. The ΔCT value was calculated by subtracting the

CT value of the gene of interest with that of the housekeeping gene (GAPDH). The relative expression value was presented as 2^-ΔCT.

4.2.8 Figures and data analysis

All figures and statistical analysis were prepared by GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) or Microsoft Excel 2007. Data are presented as mean ± SD. Statistical significance was determined by Student’s t-test or analysis of variance (ANOVA) and set at P<0.05. Post-hoc analysis of significant ANOVA results was performed using the Tukey’s method. Homogeneity of variances was assessed by F-test or Bartlett’s test. AML free survival between different treatment groups was plotted on Kaplan-Meier (KM) survival curves and analyzed by the Mantel-Cox test or log-rank test for trend. All statistical tests were two sided. Clustergram for RT2 Profiler PCR Arrays are created by the web-based algorithm provided by QIAGEN (http://www.sabiosciences.com/).

4.3 Results

4.3.1 High dose/short term iron loading

High dose intraperitoneal injection of iron dextran that exceeds 100 mg has been shown to result in massive iron deposition in various organs including heart and liver (Bartfay et al., 2000; Moon et al., 2011). To confirm iron accumulation in hematopoietic tissues including BM and spleen,

126 we injected B6D2F1 mice with 150 mg iron dextran and performed analysis at 3 days after the end of iron loading (N=5 each for control and iron treated groups). Iron-loaded mice had coarse fur and lower final body weight when compared to the control mice (Table 4.3.1). Iron loading did not cause early mortality and we did not observe overt adverse responses such as trembling, decreased activity and respiratory difficulties. Examined after sacrifice, the major organs and body cavity of the iron-loaded mice had bronze discoloration with a significant increase in spleen volume by almost 30% when compared to control mice (P<0.05, Table 4.3.1). Iron staining of tissue sections confirmed iron deposition in the liver, myocardium, spleen and BM (Figure 4.3.1). Iron deposition in the spleen was limited to the red pulp, which is consistent with the location of red pulp macrophages that are responsible for RBC removal and iron recycling (Ganz, 2012). In addition, flow cytometric analysis revealed increased apoptosis of total BMCs in iron-loaded mice based on annexin V+ and 7 AAD- staining (6.26±0.96% versus 3.54±0.99% for control mice, P<0.005, Table 4.3.1). Furthermore, iROS level in total BMCs measured using DCFH-DA was significantly higher for the iron-loaded mice than the control mice (Table 4.3.1).

Table 4.3.1. Effects of high dose, short term iron‐loading in mice.

Treatment Control Iron‐loaded t‐test N55 Total iron burden (mg) 0 150 Mean initial body weight (g) 21.50±0.94 21.92±0.89 P=0.49 Mean final body weight (g) 25.28±0.62 20.60±2.90 P<0.01 Mean final spleen weight (mg) 75.64±3.09 97.08±20.50 P<0.05 Mean apoptosis in total BMCsa 3.54±0.99% 6.26±0.96% P<0.005 Mean iROS in total BMCsb 1.00±0.29 1.81±0.50 P<0.05 a Annexin V+/7 AAD‐ staining population as a percentage of 7 AAD‐ population b DCF mean fluorescence intensity (MFI), normalized to control

127

128

4.3.2 Low dose/long term iron loading

The mortality rate of iron loading at 150 mg in B6D2F1 mice has been reported to be between 10 to 20% within weeks after the initiation of iron injection (Bartfay and Bartfay, 2000; Bartfay et al., 1999). This high mortality rate means this high dose of iron may not be suitable for the study of leukemogenesis, which we anticipated would take months after iron loading since iron overload is not known to induce carcinogenesis within a short period of time. Also, on a per weight basis, this amount of iron is comparable to the transfusion of 1200 units of packed RBC for humans, far exceeding plausible exposure in MDS patients We therefore decided to assess low dose iron loading at 5, 10 and 20 mg, doses that are comparable to 40, 80 and 160 units of packed PRBC for humans, respectively. Therefore, for 5 mg iron loading:

5 mg iron ÷ 35 g mouse body mass = 0.143 g iron/kg mouse body mass

0.143 g iron/kg mouse body mass × 70 kg human body mass = 10 g iron

10 g iron ÷ 0.25 g iron/unit PRBC = 40 units PRBC).

This is admittedly a crude estimation of equivalency, since it ignores many important factors including the difference in bioavailability of iron dextran and heme iron; however, it does provide some guidance and reassures that the doses chosen are in the right “ballpark” to simulate the degree of iron loading caused by transfusion in MDS patients. Iron dextran or sham injections took place over 1 to 4 weeks, and the mice were sacrificed for analysis at 3 months after iron loading (N=5 each for control, 5, 10 and 20 mg iron-loaded groups).

Low dose/long term iron loading was well tolerated by the B6D2F1 mice with no early mortality. Bronze discoloration of organs and body cavity was observed in the 20 mg iron-loaded mice, but the degree of discoloration was substantially milder than in the 150 mg iron-loaded mice. Slightly lower final body weight and spleen enlargement was noted in the 20 mg iron-loaded mice but this change was not statistically significant (Table 4.3.2). Iron deposition in the liver and BM of the iron-loaded mice was confirmed by iron staining (Figure 4.3.2a). We observed increased presence of hemosiderin and iron staining in general throughout the iron-loaded BM. For peripheral blood counts, mean corpuscular volume (MCV) was increased in iron-loaded mice

129 at all iron doses when compared to the control mice; the difference was statistically significant for the 10 mg iron-loaded mice (ANOVA P<0.01, Table 4.3.2). We also observed increased apoptosis of CD45- BMCs, a population that includes nucleated RBC precursors, with iron loading; the increase was significant in the 20 mg iron-loaded mice (ANOVA P<0.01, Table 4.3.2). Although the iron-loaded mice were not significantly anemic, elevated apoptosis of BM nucleated RBC and increased PB MCV might be indicative of altered erythropoiesis (Aslinia et al., 2006). Indeed, iron-induced increase in red cell size has been reported in hemodialysis patients (Gokal et al., 1979). For lineage negative (Lin-) and CD45+ hematopoietic stem cells (HSCs) or progenitors, iron loading appeared to increase the width of the distribution of iROS level in the 5 mg loaded mice, but the distribution decreased progressively at high iron dosages (Table 4.3.2, Figure 4.3.2b). Bartlett’s test for equal variances indicated that the variances differ significantly among different treatment groups (P<0.05). Further analysis by F-test suggested that the variances of 5 and 10 mg iron-loaded groups differ significantly with the control group (P<0.01 and P<0.05, respectively). The changes in variances suggested increased iROS level in some, but not all, of the 5 and 10 mg iron-loaded mice. Iron loading at these dose levels appears to challenge ROS homeostasis in hematopoietic progenitors.

130

Table 4.3.2. Effects of low dose, long term iron‐loading in mice.

Treatment Control Iron (5mg) Iron (10mg) Iron (20mg)

N5555

Total iron burden (mg) 0 5 10 20

Mean initial body weight (g) 24.80±2.73 24.72±1.22 24.08±1.76 25.70±1.30

Mean final body weight (g) 38.50±6.50 37.92±2.91 35.36±2.56 34.78±3.56

Mean final spleen weight (mg) 97.19±42.00 98.24±9.04 100.62±25.19 112.5±7.86

Mean MCV (fL) 44.76±0.76 45.68±0.79 46.3±0.78** 45.74±0.51

Mean apoptosis in CD45‐ BMCsa 26.27±7.36% 33.60±8.62% 34.8±4.8% 42.1±9.05%**

Mean iROS in Lin‐CD45+ BMCsb 1.00±0.31 1.74±0.95 1.34±0.72 0.97±0.33

F‐testc ‐ P<0.01 P<0.05 P=0.80 a AnnexinV+/7 AAD‐ staining population as a percentage of 7 AAD‐ population b DCF MFI, normalized to control c F‐test of iROS in Lin‐CD45+ BMCs compared to control ** ANOVA P<0.01 (Tukey’spost‐hoc analysis)

131

132

4 **

3 *

2 (normalized)

level 1 iROS 0 0 5 10 20 Iron loading (mg)

Figure 4.3.2b. iROS level (DCF MFI normalized to control) of Lin‐CD45+ BMCs from control (0 mg iron loading) and low dose, long term iron‐ loaded (5‐20 mg iron loading) mice. F‐test P<0.05 (*), P<0.01(**) when compared to control.

133

4.3.3 Effects of iron and ICT on radiation-induced AML

We employed a RI-AML mouse model to determine the effects of iron loading on leukemogenesis (Figure 4.3.3a). B6D2F1 mice were treated with non-lethal total-body irradiation, injected with dexamethasone, and then injected with 5, 10 or 20 mg of iron dextran. Rodents reduce excess iron over time by decreasing iron uptake and increasing iron excretion (Crichton, 2009g; Musumeci et al., 2014). To mimic persistent secondary iron overload as a result of chronic transfusion, we administered a second dose of iron at 5 months after irradiation using half of the initial treatment amount. Therefore, the total amount of iron received was 7.5, 15 or 30 mg. To assess if ICT can modulate the effects of iron loading on leukemogenesis, some of the 7.5 or 30 mg iron-loaded mice were fed with deferasirox at 10 or 40 mg/kg/day by oral gavage for 8 weeks immediately after the end of the first series of iron injections, and for 4 weeks after the end of second iron injection series. The mice were monitored for signs of illness for 70 weeks after irradiation, and consisted of 8 treatment groups with varying iron (mg)/ICT (mg/kg/day) dosage: 0/0 (N=13), 7.5/0 (N=13), 15/0 (N=10), 30/0 (N=10), 7.5/10 (N=10), 7.5/40 (N=10), 30/10 (N=10), and 30/40 (N=10).

There was no significant growth difference among the treatment groups as indicated by their maximum weight during the 70 weeks observation period (Table 4.3.3a). However, the variance of maximum weight was larger for the mice that received 30 mg iron than the control mice (F- test P<0.0005), which may be indicative of higher variability of physical condition among individuals in that group. During the observation period, 29 (33.7%) of the 86 mice were found dead or needed to be sacrificed due to illness (Table 4.3.3a). Among these mice, 15 were diagnosed with AML, while the other 13 died from other causes ranging from preputial abscess to tumors at various locations (Table 4.3.3b).

Since irradiation in mice is specifically associated with AML (Rivina et al., 2014), we decided to censor other causes of death and focus our analysis on AML. Manifestation of AML in our animal model was marked by severe weight drop of at least 20% (Figure 4.3.3b) and the presence of blasts in PB (Figure 4.3.3c). In mice with leukemia, the size of the myeloid compartment in the PB, based on CD11b+ population, was also expanded in the PB in comparison to B220+CD3e+ lymphoid population (Figure 4.3.3d). The condition was fatal

134 within 4 weeks after the weight drop. Hepatomegaly or splenomegaly was noted in most of the AML mice. The mean liver weight was 3.70±1.52 g (range 2.01-7.29 g, while the mean spleen weight was 771±485mg (range 86-1609 mg). Based on our experience, the normal liver and spleen weight of adult B6D2F1 mice are about 1.5 g and 100 mg, respectively. We also observed blast infiltration in the bone marrow, liver and spleen (Figure 4.3.3e). BMCs from the AML mice were less heterogeneous with a high proportion of immature cells when compared to non- AML mice (Figure 4.3.3f). Some neutrophils and monocytes were also seen in the BMCs from the AML mice. Flow cytometry analysis of BMCs from AML mice revealed expansion of immature (Lin-CD45+) hematopoietic populations (Figure 4.3.3g). A Lin-CD45low/- population was also noted in some of the AML mice (Figure 4.3.3g), which may represent leukemic blasts with the characteristics of CD45-TER119- erythroid colony-forming units (CFU-E). In addition to CD11b+Gr-1low monocyte and CD11b+Gr-1hi granulocyte, we detected CD11b-Gr-1+ population in the BMCs from some of the AML mice (Figure 4.3.3h). Gene expression levels of some of the AML mice were also assessed by quantitative RT-PCR and discussed in later sections.

Based on the Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms (Kogan et al., 2002), 7 mice were diagnosed with monocytic leukemia characterized by the presence of a monocytic component in the BMCs, 2 mice had myelomoncytic leukemia with both neutrophilic and monocytic components, and one mouse had myeloid leukemia with maturation characterized by the presence of a neutrophilic component. The other 5 mice were confirmed to have myeloid leukemia but the exact subtype could not be determined. All of the AML cases during the observation period developed in mice that belonged to the iron-loaded or iron-loaded/ICT groups. Of the control mice that were irradiated but not iron-loaded, two eventually developed AML after the end of the 70 weeks observation period. Within the observation period, the earliest AML onset was at 25.4 weeks after irradiation and the latest was at 69 weeks. Leukemia (AML)-free survival (LFS) in iron-loaded and iron-loaded/ICT mice were 74% and 82%, respectively, and both were not significantly different when compared with the control mice (P=0.06 and P=0.12, respectively) or with each other (P=0.49 Figure 4.3.3i). Among the iron-loaded mice, the highest rate of AML was observed in the group receiving 7.5mg iron with LFS at 58% (P<0.05, HR 9.29 vs. controls, Figure 4.3.3j). Surprisingly, there

135 appeared to be an inverse relationship between iron dose and AML. The LFS and earliest AML onset in the 7.5 mg, 15 mg, and 30 mg iron-loaded groups were 58%/25.4 weeks, 80%/52.7 weeks, and 88%/67.1 weeks, respectively (logrank test for trend P=0.10). Deferasirox treatment improved LFS in the 7.5 mg iron-loaded group (80% vs. 58% without ICT, Figure 4.3.3k). In addition, a significant trend in LFS was observed when the groups were arranged according to the amount of iron loading, ranking highest-to-lowest from control, 7.5 mg iron/ICT, to 7.5 mg iron (logrank test for trend P<0.05). Conversely, ICT had the opposite effect on the 30 mg iron- loaded group, in which a more rapid AML onset was observed after deferasirox treatment (33.3 vs. 67.1 weeks without ICT, Figure 4.3.3l). Moreover, the deferasirox-treated 30 mg iron-loaded mice had bronze discoloration in their organs and body cavity, suggesting the presence of excess iron even after ICT.

136

137

Table 4.3.3a. Effects of iron dextran and deferasirox on irradiated B6D2F1 mice.

Treatment (iron/ICT) 0/0 7.5/0 7.5/10 7.5/40 15/0 30/0 30/10 30/40 C1.X C3.X C9.X C11.X C5.X C7.X C13.X C15.X C2.X C4.X C10.X C12.X C6.X C8.X C14.X C16.X N 1313101010101010 Total iron burden (mg) 0 7.5 7.5 7.5 15 30 30 30 Deferasirox (mg/kg/d)0010400 01040 Mean initial body 27.12 28.21 32.93 30.51 27.68 27.32 31.19 31.69 weight (g) ±1.57 ±2.1 ±2.85 ±1.79 ±3.32 ±1.78 ±1.48 ±3.11

Mean maximum body 52.35 50.48 56.37 50.38 52.56 55.91 51.95 56.73 weight (g) ±3.37 ±7.05 ±4.64 ±4.13 ±6.05 ±10.89* ±5.27 ±7.52

Mortality due to AML 0 (0%) 5 (38%) 1 (10%) 3 (30%) 2 (20%) 1 (10%) 3 (30%) 0 (0%) Total mortality 2 (15%) 7 (54%) 3 (30%) 4 (40%) 3 (30%) 3 (30%) 3 (30%) 4 (40%) * F‐test compared to control P<0.0005

138

Table 4.3.3b. Cause of death (AML or non‐AML) for control, iron, iron/ICT irradiated mice.

Label Iron/ICTa LFS/OS Final body Final liver Final spleen Diagnosisc (weeks)b weight (g) weight (g) weight (mg) AML C3.3 7.5/0 35.8/37.1 33.5 2.53 260 Monocytic leukemia C3.6 7.5/0 54.0/57.0 45.0 ‐‐Myeloid leukemiad C3.7 7.5/0 41.0/43.0 36.3 ‐‐Myeloid leukemiad C4.1 7.5/0 58.7/60.0 47.1 3.83 1085 Monocytic leukemia C4.5 7.5/0 25.4/27.7 29.0 2.22 367 Monocytic leukemia C10.1 7.5/10 31.3/35.0 31.6 7.29 1560 Monocytic leukemia C11.4 7.5/40 37.3/38.3 34.1 3.64 1609 Monocytic leukemia C11.5 7.5/40 63.1/69.0 40.7 2.06 86 Myeloid leukemiae C12.5 7.5/40 36.3/41.6 36.4 2.01 930 Myelomoncytic leukemia C6.1 15/0 55.7/57.0 42.9 3.94 789 Monocytic leukemia C6.5 15/0 52.7/54.0 52.2 4.01 579 Myeloid leukemiad C7.4 30/0 67.1/67.1 58.9 3.42 548 Myelomoncytic leukemia C13.3 30/10 42.1/45.0 36.9 3.80 742 Myeloid leukemiad C13.4 30/10 33.3/36.4 33.1 6.01 1170 Monocytic leukemia C14.2 30/10 55.6/55.6 41.3 3.32 293 Myeloid leukemiad Non‐AML C1.4 0/0 56.6 43.9 2.04 97 Other C2.4 0/0 50.0 49.9 2.12 1380 Metastasized tumor C4.4 7.5/0 16.7 37.2 1.46 63 Other C10.2 7.5/10 69.3 51.8 1.78 80 Other C10.5 7.5/10 68.6 45.5 6.30 175 Liver tumor C12.1 7.5/40 69.1 35.6 1.13 140 Lung tumor C5.1 15/0 67.1 52.0 2.88 187 Abdominal tumor C7.3 30/0 52.6 37.3 ‐‐Other C8.5 30/0 39.6 28.4 2.08 87 Other C15.1 30/40 54.6 38.9 2.71 421 Lymphoid neoplasm C15.2 30/40 31.6 29.2 1.88 89 Liver tumor C15.4 30/40 44.0 60.8 3.87 205 Bone tumor C16.2 30/40 67.4 58.0 3.40 389 Abdominal tumor a Iron burden (mg) / ICT by deferasirox (mg/kg/d) b Leukemia (AML) free survival / overall survival (post‐irradiation, weeks) c Diagnosis of AML was made based on the Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice (Koganet al., 2002) d AML subtype not determined e Myeloid leukemia with maturation

139

45 (g)

weight 40

Body 35

30 C1.6 (0/0)

C3.7 (7.5/0, AML) 25 0 102030405060 Post‐irradiation survival (weeks)

Figure 4.3.3b. Representative weight change over time for irradiated mice.Non‐iron‐loaded C1.6 (0/0) did not display any signs of illness. C3.7 (7.5/0) was loaded with 7.5 mg iron dextran and developed AML. Severe weight drop was observed for C3.7 upon AML onset.

140

C1.6 (0/0) C6.1 (15/0, AML)

PB smear (400x)

Figure 4.3.3c. Representative May‐Grünwald‐Giemsa staining of peripheral blood (PB) smear.Non‐iron‐loaded C1.6 did not display any signs of illness. C6.1 was loaded with 7.5mgirondextrananddevelopedAML.ThePBspecimenfromC6.1wasobtainedafter AML onset, the presence of leukemic blasts was observed.

C1.6 (0/0) C3.6 (7.5/0, AML) 4 104 10

3 103 12.6% 10 41.2%

2 102 10 B220/CD3e 1 101 10

0 100 10 0 1 2 3 4 100 101 102 103 104 10 10 10 10 10 CD11b CD11b Figure 4.3.3d. Representative flow cytometry analysis of myeloid vs. lymphoid population in PB using anti‐B220/CD3e‐FITC and anti‐CD11b‐APC antibodies. Non‐ iron‐loaded C1.6 (0/0) did not display any signs of illness. C3.6 (7.5/0) was loaded with 7.5 mg iron dextran and developed myeloid leukemia. The PB specimen from C3.6 was obtained after AML onset, enrichment of myeloid (CD11b+) population was observed.

141

142

143

144

145

100

80 survival

70 Control

Percent 60 Iron Iron/ICT 50 0 20 40 60 80 Post‐irradiation LFS (weeks)

Figure 4.3.3i. Kaplan‐Meier (KM) leukemia‐free survival (LFS) curves of irradiated mice treated with sham injection (Control: 0/0 group, N=13), iron dextran injection (Iron: 7.5/0, 15/0 and 30/0 groups, N=33), and iron/deferasirox treatments (Iron/ICT: 7.5/10, 7.5/40, 30/10 and 30/40 groups, N=40).Leukemia‐ free survival of the control group was not significantly different than the Iron (P=0.06) or Iron/ICT (P=0.12) groups. Survival of the Iron and IronICT groups were not significantly different from each other (P=0.49).

100

80 survival

70 Control 7.5 mg Iron

Percent 60 15 mg Iron 30 mg Iron 50 0 20 40 60 80 Post‐irradiation LFS (weeks)

Figure 4.3.3j. KM LFS curves of irradiated mice treated with sham injection (Control: 0/0 group, N=13 ) or iron dextran injections (7.5 mg Iron [7.5/0], N=13; 15 mg Iron [15/0], N=10; 30 mg Iron [30/0], N=10).Leukemia‐free survival of the 7.5 mg iron‐loaded group was significantly different compared to the control group (P<0.05, HR=9.09).

146

100

80 survival

70 Control

Percent 60 7.5 mg Iron 7.5 mg Iron/ICT 50 0 20 40 60 80 Post‐irradiation LFS (weeks)

Figure 4.3.3k. KM LFS curves of irradiated mice treated with sham injection (control: 0/0 group, N=13), 7.5 mg iron dextran injection (7.5/0 group, N=13), and 7.5 mg iron/deferasirox (7.5 mg Iron/ICT: 7.5/10 and 7.5/40, groups, N=20).A significant trend was observed for the following group arrangement: Control, 7.5 mg Iron/ICT, and 7.5 mg Iron (logrank test for trend P<0.05).

100

80 survival

70 Control

Percent 60 30 mg Iron 30 mg Iron/ICT 50 0 20 40 60 80 Post‐irradiation LFS (weeks)

Figure 4.3.3l. KM LFS curves of irradiated mice treated with sham injection (control: 0/0 group, N=13), 30 mg iron dextran injection (30/0 group, N=10), and 30 mg iron/deferasirox (30 mg Iron/ICT: 30/10 and 30/40, groups, N=20).

147

4.3.4 The role of iron loading in leukemogenesis of RI-AML

To elucidate the mechanism by which iron contributes to radiation-induced leukemogenesis, we examined the BMCs from the mice that developed AML, as well as non-leukemic mice at 3, 5, and 7 months after irradiation (Figure 4.3.3a). For the analyses at earlier intervals, we injected the mice with 7.5 mg iron dextran (or 5 mg before the second iron loading), which had the highest incidence rate of AML among the tested iron doses. To assess the effects of ICT, some of the iron-loaded mice for the 5 and 7 months analyses had also received oral deferasirox in addition to iron injection.

4.3.5 Gene expression in the BMCs of mice that develop AML

We assessed the gene expression profile of BMCs from 5 irradiated mice that developed AML by quantitative RT-PCR using RT2 profiler PCR arrays, with which we examined the transcription of 252 genes related to PI3K-Akt, Wnt and oxidative stress. Genes with altered expression are listed in Table 4.3.5a. BMCs from the following mice were analyzed: C2.5 (control mice that developed AML after the 70 weeks observation period), C6.1 (15 mg iron- loaded), C7.4 (30 mg iron-loaded), C11.4 (7.5 mg iron-loaded with 40 mg/g/day deferasirox), and C13.4 (30 mg iron-loaded with 10 mg/g/day deferasirox). To screen for transcription difference, the expression level of a gene of interest from the AML group was compared to that of the control group (irradiated mice without iron-loading) at 5 months post-irradiation by Student’s t-test.

The expression levels of 73 genes were changed in the AML group when compared to the 5 months control group (Table 4.3.5b). Among genes related to the PI3K-Akt pathway, 21 were upregulated and 6 were downregulated. Among genes related to the Wnt pathway, 5 were upregulated and 16 were downregulated. For oxidative stress related genes, 19 were upregulated and 6 were downregulated. A clustergram based on 86 genes was constructed using unsupervised hierarchical clustering analysis and consisted of mice from the AML, 5 months and 7 months groups (Figure 4.3.5a). The list of 86 genes comprised the 73 genes whose expression levels were altered in the AML group when compared to the 5 months control group, and 13 other genes whose expression levels were altered from other comparisons at earlier time interval (5 and 7 months post-irradiation). As expected, all AML specimens were clustered into the same 148 group, suggesting comparable gene expression profile among the AML mice. We also compared the AML group with the control groups at different time points to determine if the expression of certain genes are progressively changed and may be involved in leukemogenesis (Table 4.3.5c). Cebpd, Jun and Nfkbia were progressively downregulated in the irradiated control mice from 5 months to 7 months, and reached the lowest level for the AML group. Conversely, Fzd7, Itgb1, Ccs and Gss were progressively upregulated from 5 months to 7 months control groups, and reached the highest level for AML group. We have also examined the expression level of the tumor suppressor Pten. The mean expression level of Pten in the AML group was not significantly different from the 7 months iron-loaded group. However, we noticed an outlier in the AML group with unusually high level of Pten. Nonetheless, Phosphatase and tensin homolog (PTEN, gene product of Pten) can be regulated by post-translational means and suppressed without altering its transcription level (Wang and Jiang, 2008). Upon removal of the outlier, progressive downregulation of Pten was observed from the 7 months control group to the corresponding iron-loaded group and then the AML group (Figure 4.3.5b).

Many antioxidant genes were upregulated in the BMCs of the AML mice (aBMCs) when compared to those in the 5 months control mice. These genes are responsible for several major groups of antioxidant defenses including glutathione (Gss, Gclm, Gstp1, Gpx4, and Gsr) (Traverso et al., 2013), superoxide dismutase (Ccs, Sod1, and Sod2) (Fukai and Ushio-Fukai, 2011; Wong et al., 2000), catalase (Cat) (Bauer, 2012), and peroxidase (Prdx2, Prdx3, and Prdx4) (Lu et al., 2014; Neumann and Fang, 2007). Although antioxidant defenses are crucial against tumor transformation in normal cells, they are also upregulated or selected for in malignant cells in order to enable these cells to cope with high oxidative stress (Hole et al., 2011; Traverso et al., 2013). In fact, augmented antioxidant gene expression has been reported in MDS and AML (Bowen et al., 2003b).

The expression pattern of genes related to the PI3K-Akt pathway suggested elevated Akt and mTOR activities in the aBMCs. This pathway is crucial in tumor development and has been shown to be active in many cancers and AML (Altman et al., 2011; Polivka and Janku, 2014). Several downstream components of mTOR were overexpressed, including Rps6kb1, Pabpc1, Eif4g1 Eif4b, and Eif4e. Their gene products are responsible for the transduction of activated mTOR signal to initiate protein synthesis and cell proliferation (Datan et al., 2014; Proud, 2014; 149

Shahbazian et al., 2006; Tcherkezian et al., 2014; Thoreen et al., 2012). We also observed upregulation of Akt1 and Akt2, which activates mTOR when they are activated by phosphorylation (Hay and Sonenberg, 2004). Akt activation might be partly mediated by the upregulation of Pdk1, Itgb1 and Pik3r2 as well as downregulation of Cd14, whose gene products are known to phosphorylate Akt (Cortes et al., 2012; Mora et al., 2004; Sahay et al., 2009; Speicher et al., 2014).

NF-κB activity was likely enhanced in the BMCs of the AML mice as a result of Nfkbia downregulation and Gsk3b upregulation. The gene product of Nfkbia, IκBα, binds to NF-κB and strongly inhibits its activity (Oeckinghaus and Ghosh, 2009). Removal of IκBα from NF-κB is the first step of NF-κB activation. The gene product of Gsk3b, GSK3β, activates NF-κB at the level of the transcriptional complex (Hoeflich et al., 2000). In addition, crosstalk between NF-κB and Akt has been shown in various cancers and both pathways are often constitutively active in AML (Davoudi et al., 2014). NF-κB activity is also required for H-ras induced cell proliferation and anti-apoptotic activity in neoplastic cells (Jo et al., 2000). Transcription of H-ras (Hras1) was upregulated in the aBMCs.

Components of the transcription factor complex AP-1, including c-Jun (Jun), c-Fos (Fos) and FOSL1 (Fosl1), were downregulated in the aBMCs. AP-1 is activated by JNK to promote transcription of genes that are essential to differentiation, proliferation and apoptosis (Ameyar et al., 2003). Indeed, several JNK target genes, including Ptgs2, Plaur and Mmp9, were also downregulated in the aBMCs. JNK suppression may also be partly mediated by activated Akt and NF-κB, both of which have been shown to inhibit JNK-dependent apoptosis and promote survival of cancer cells (Sykes et al., 2011; Zhang et al., 2004). Although pro-apoptotic genes such as caspase 9 (Casp9) and Bcl-2-associated death promoter (Bad) were upregulated in the aBMCs, their ability to promote apoptosis might be offset by JNK inactivation (Donovan et al., 2002; Sugahara et al., 2006).

We observed transcriptional dysregulation of genes related to the Wnt signaling pathway in the aBMCs when compared to the BMCs at 5 months. Components of the Wnt pathway, such as β- catenin (Ctnnb1) and frizzled-7 (Fzd7), were upregulated in the aBMCs. Apc was downregulated in the aBMCs, whose gene product, adenomatous polyposis coli (APC), is part of a destruction

150 complex that disables the transcription activity of β-catenin and thereby suppresses the Wnt signal. Suppression of APC enhances Wnt signaling and has been reported to promote AML (Stoddart et al., 2014). Conversely, we observed increased expression of Gsk3b, whose gene product is also a part of the β-catenin destruction complex. The expression of Wnt target genes, including Fzd7, Six1 and Fst, was also upregulated, while other Wnt target genes, including Runx2, Mmp9 and Id2, were downregulated. The simultaneous up and downregulation of Wnt target genes might be a consequence of incomplete Wnt activation, resulting in the production of insufficient quantity of active β-catenin for its multiple partners and therefore partial transcription activation (Shapiro, 2001). Alternatively, some of the Wnt target genes are regulated by other pathways, such as Runx2 by TNFα (Ding et al., 2009) and Mmp9 by JNK (Cheng et al., 2012), adding to the complexity of the expression profile.

Leukemogenesis is a multistep process where normal cells acquire multiple mutations over time and eventually become malignant (Moriya et al., 2012; Vogelstein and Kinzler, 1993). Progressive transcriptional dysregulation can also be expected as a result of mutations that alter signaling pathways. In our leukemia mouse model, we expect the pathogenesis of AML to correspond with two variables: the time elapsed after irradiation and iron loading. We observed progressive dysregulation of several genes related to different signaling pathways, including JNK deactivation (Jun downregulation), NF-κB activation (Nfkbia downregulation) and Wnt activation (Fzd7 upregulation).These pathways have been implicated to be dysregulated during the process of leukemogenesis by allowing abnormal cells to self-renew, transform and evade apoptosis (Lento et al., 2013; Malinge et al., 2006; Mitani, 2004; Muller-Tidow et al., 2004). Moreover, progressive upregulation of antioxidant genes such as Ccs and Gss suggest the increased need for the cells to cope with oxidative stress. The adhesion molecule β1-integrin (Itgb1) allows abnormal cells to bind the extracellular matrix and avoid anoikis (Bendall et al., 1993; Iyoda and Fukai, 2012). Anoikis is a form of programmed cell death induced by inadequate or inappropriate cell-matrix interactions (Frisch and Screaton, 2001). Progressive overexpression of β1-integrin might provide survival and proliferation advantages to leukemic blasts. Furthermore, progressive downregulation of tumor suppressors Cebpd and Pten are consistent with the multistep aspect of leukemogenesis. CCAAT-enhancer-binding protein δ (C/EBPδ, gene product of Cebpd), a crucial mediator of apoptosis, is often downregulated in

151 cancer (Palmieri et al., 2012; Thangaraju et al., 2005). Suppression of Cebpd might be an early event that contributes to AML development in our mouse model in that its transcription was gradually downregulated from 5 months to 7 months to AML. PTEN inactivation is a common feature of carcinogenesis and its suppression is defined as a late event in some forms of cancers including prostate cancer and melanoma (Mutter, 2001). Progressive suppression of Pten, in comparison to non-irradiated mice, occurred: the lowest degree of suppression was seen at 7 months in the control (irradiated) group, with a greater degree of suppression at 7 months in the iron-loaded group, and the greatest suppression in the AML group. We conjecture that the 7 months iron-loaded group represents a later stage of leukemogenesis (to be discussed in more detail below), and suppression of PTEN activity might be a late event that occurs before the development of overt AML.

152

Table 4.3.5a. Genes with altered expression in all analyses (total BMCs of AML mice, and mice from 5 and 7 months early analyses).

Symbol Unigene Array Description Akt1 Mm.6645 Akt Thymoma viral proto‐oncogene 1 Akt2 Mm.177194 Akt Thymoma viral proto‐oncogene 2 Apc Mm.384171 Akt Adenomatosis polyposis coli Bad Mm.4387 Akt BCL2‐associated agonist of cell death Casp9 Mm.88829 Akt Caspase 9 Cd14 Mm.3460 Akt CD14 antigen Chuk Mm.3996 Akt Conserved helix‐loop‐helix ubiquitous kinase Ctnnb1 Mm.291928 Akt Catenin (cadherin associated protein), beta 1 Eif4b Mm.391795 Akt Eukaryotic translation initiation factor 4B Eif4e Mm.488704 Akt Eukaryotic translation initiation factor 4E Eif4g1 Mm.260256 Akt Eukaryotic translation initiation factor 4, gamma 1 Elk1 Mm.490895 Akt ELK1, member of ETS oncogene family Fasl Mm.3355 Akt Fas ligand (TNF superfamily, member 6) Fkbp1a Mm.487732 Akt FK506 binding protein 1a Fos Mm.246513 Akt FBJ osteosarcoma oncogene Gsk3b Mm.394930 Akt Glycogen synthase kinase 3 beta Hras1 Mm.334313 Akt Harvey rat sarcoma virus oncogene 1 Ilk Mm.274846 Akt Integrin linked kinase Itgb1 Mm.263396 Akt Integrin beta 1 (fibronectin receptor beta) Jun Mm.275071 Akt Jun oncogene Mapk1 Mm.196581 Akt Mitogen‐activated protein kinase 1 Mtor Mm.21158 Akt Mechanistic target of rapamycin (serine/threonine kinase) Nfkbia Mm.170515 Akt NF‐Kappa‐B Inhibitor Alpha Pabpc1 Mm.371570 Akt Poly(A) binding protein, cytoplasmic 1 Pdk1 Mm.34411 Akt Pyruvate dehydrogenase kinase, isoenzyme 1 Pdk2 Mm.29768 Akt Pyruvate dehydrogenase kinase, isoenzyme 2 Pik3r2 Mm.12945 Akt Phosphoinositide‐3‐Kinase, Regulatory Subunit 2 (Beta) Prkcb Mm.446371 Akt Protein kinase C, beta Rheb Mm.319175 Akt Ras homolog enriched in brain Rps6kb1 Mm.446624 Akt Ribosomal protein S6 kinase, polypeptide 1 Srf Mm.45044 Akt Serum response factor Tollip Mm.103551 Akt Toll interacting protein Wasl Mm.1574 Akt Wiskott‐Aldrich syndrome‐like (human)

153

Table 4.3.5a. Continued (1).

Symbol Unigene Array Description Abcb1a Mm.207354 Wnt ATP‐binding cassette, sub‐family B (MDR/TAP), member 1A Angptl4 Mm.196189 Wnt Angiopoietin‐like 4 Cdh1 Mm.35605 Wnt Cadherin 1 Cebpd Mm.347407 Wnt CCAAT/enhancer binding protein (C/EBP), delta Efnb1 Mm.3374 Wnt Ephrin B1 Egr1 Mm.181959 Wnt Early growth response 1 Ets2 Mm.290207 Wnt E26 avian leukemia oncogene 2, 3' domain Fn1 Mm.193099 Wnt Fibronectin 1 Fosl1 Mm.6215 Wnt Fos‐like antigen 1 Fst Mm.4913 Wnt Follistatin Fzd7 Mm.297906 Wnt Frizzled homolog 7 (Drosophila) Id2 Mm.34871 Wnt Inhibitor of DNA binding 2 Il6 Mm.1019 Wnt Interleukin 6 Klf5 Mm.30262 Wnt Kruppel‐like factor 5 Lrp1 Mm.271854 Wnt Low density lipoprotein receptor‐related protein 1 Met Mm.86844 Wnt Met proto‐oncogene Mmp9 Mm.4406 Wnt Matrix metallopeptidase 9 Nrcam Mm.208439 Wnt Neuron‐glia‐CAM‐related cell adhesion molecule Nrp1 Mm.271745 Wnt Neuropilin 1 Pdgfra Mm.221403 Wnt Platelet derived growth factor receptor, alpha polypeptide Pitx2 Mm.246804 Wnt Paired‐like homeodomain transcription factor 2 Plaur Mm.1359 Wnt Plasminogen activator, urokinase receptor Ptch1 Mm.488634 Wnt Patched homolog 1 Runx2 Mm.391017 Wnt Runt related transcription factor 2 Six1 Mm.4645 Wnt Sine oculis‐related homeobox 1 homolog (Drosophila) Tcf7l1 Mm.440067 Wnt Transcription factor 7‐like 1 (T‐cell specific, HMG box)

154

Table 4.3.5a. Continued (2).

Symbol Unigene Array Description Atr Mm.212462 ROS Ataxia telangiectasia and rad3 related Cat Mm.4215 ROS Catalase Ccs Mm.434411 ROS Copper chaperone for superoxide dismutase Cygb Mm.34598 ROS Cytoglobin Ehd2 Mm.138215 ROS EH‐domain containing 2 Ercc6 Mm.318310 ROS Excision Repair Cross‐Complementation Group 6 Fth1 Mm.1776 ROS Ferritin heavy chain 1 Gclm Mm.292676 ROS Glutamate‐cysteine ligase, modifier subunit Gpx4 Mm.359573 ROS Glutathione peroxidase 4 Gpx7 Mm.20164 ROS Glutathione peroxidase 7 Gsr Mm.283573 ROS Glutathione reductase Gss Mm.252316 ROS Glutathione synthetase Gstp1 Mm.299292 ROS Glutathione S‐transferase, pi 1 Idh1 Mm.9925 ROS Isocitrate dehydrogenase 1 (NADP+), soluble Ift172 Mm.293023 ROS Intraflagellar transport 172 homolog (Chlamydomonas) Mpo Mm.4668 ROS Myeloperoxidase Ngb Mm.41395 ROS Neuroglobin Park7 Mm.277349 ROS Parkinson disease (autosomal recessive, early onset) 7 Prdx2 Mm.393373 ROS Peroxiredoxin 2 Prdx3 Mm.29821 ROS Peroxiredoxin 3 Prdx4 Mm.247542 ROS Peroxiredoxin 4 Psmb5 Mm.8911 ROS Proteasome (prosome, macropain) subunit, beta type 5 Ptgs2 Mm.292547 ROS Prostaglandin‐endoperoxide synthase 2 Rag2 Mm.4988 ROS Recombination activating gene 2 Sod1 Mm.466779 ROS Superoxide dismutase 1, soluble Sod2 Mm.290876 ROS Superoxide dismutase 2, mitochondrial Txn1 Mm.260618 ROS Thioredoxin 1

155

Table 4.3.5b. Changes in gene expression level of total BMCs (2^‐ΔCT value) from 5 months irradiated mice (Control, 0/0) to AML mice (AML) (t‐test P<0.05). ArraySymbol Control AML Fold Δa P‐Valueb Networksc S/Pd A/Te Mf Akt Fkbp1a 0.01177 0.22044 18.73 * Akt Eif4b 0.04309 0.56608 13.14 * Akt/mTOR Xg Akt Rps6kb1 0.03916 0.25793 6.59 * Akt/mTOR X Akt Hras1 0.01548 0.09236 5.97 ** Akt/NF‐κB X Yg Akt Casp9 0.01090 0.05020 4.60 ** X Akt Pabpc1 0.40746 1.66997 4.10 ** Akt/mTOR X Akt Bad 0.01422 0.05243 3.69 * X Akt Eif4g1 0.05268 0.18177 3.45 * Akt/mTOR X Akt Pdk1 0.00184 0.00595 3.23 * Akt X Akt Itgb1 0.03274 0.10534 3.22 * Akt X Y Akt Gsk3b 0.03786 0.10945 2.89 * NF‐κB/Wnt YY Akt Eif4e 0.03588 0.10333 2.88 *** Akt/mTOR X Akt Pik3r2 0.00129 0.00369 2.86 * Akt X Akt Akt1 0.06709 0.18991 2.83 * Akt/mTOR X Akt Mapk1 0.07491 0.20929 2.79 * Akt Srf 0.03096 0.08355 2.70 * Akt Wasl 0.01290 0.03442 2.67 * Akt Tollip 0.04894 0.11807 2.41 * Akt Ctnnb1 0.05132 0.10756 2.10 * Wnt X Akt Akt2 0.04501 0.08354 1.86 * Akt/mTOR X Akt Rheb 0.03066 0.04873 1.59 * Wnt Six1 0.00002 0.05190 3125 * Wnt X Wnt Angptl4 0.00061 0.02120 34.52 * Wnt Fzd7 0.00103 0.03302 31.96 * Wnt X Y Wnt Fst 0.00033 0.00298 9.16 * Wnt X Wnt Ptch1 0.00105 0.00792 7.57 ** ROS Cygb 0.00015 0.00692 45.54 ** ROS Gss 0.00283 0.08307 29.37 ** AODh X ROS Ngb 0.00011 0.00306 29.14 * ROS Ercc6 0.00465 0.06246 13.44 * ROS Prdx2 0.01757 0.18931 10.78 * AOD X ROS Gclm 0.02659 0.27616 10.38 * AOD X ROS Prdx3 0.01422 0.10794 7.59 ** AOD X ROS Ccs 0.00442 0.03178 7.19 * AOD X ROS Cat 0.01522 0.08916 5.86 * AOD X ROS Psmb5 0.04963 0.27888 5.62 ** ROS Prdx4 0.00767 0.04139 5.40 ** AOD X ROS Ift172 0.00321 0.01683 5.24 ** ROS Park7 0.03232 0.15952 4.94 * ROS Gstp1 0.03075 0.14946 4.86 ** AOD X ROS Sod2 0.05930 0.22014 3.71 * AOD X ROS Gpx4 0.14152 0.48981 3.46 ** AOD X ROS Sod1 0.03247 0.09978 3.07 * AOD X ROS Idh1 0.01816 0.03401 1.87 * ROS Gsr 0.36637 0.57488 1.57 * AOD X a Fold change, upregulation: AML/Control, e A/T – effects on apoptosis/tumor suppression downregulation: ‐Control/AML f M – effects on mutagenesis b * P<0.05, ** P<0.01, *** P<0.0001 g X – enhance, Y – suppress c Involvement in cell signaling networks h AOD –antioxidant defenses d S/P – effects on survival/proliferation 156

Table 4.3.5b. Continued. ArraySymbol Control AML Fold Δa P‐Valueb Networksc S/Pd A/Te Mf Akt Fasl 0.00241 0.00003 ‐92.81 ** Akt Cd14 0.24415 0.00778 ‐31.37 ** Akt Xg Akt Nfkbia 0.90250 0.08500 ‐10.62 *** NF‐κB Yg Akt Pdk2 0.00043 0.00004 ‐9.75 *** Akt Jun 2.75436 0.47910 ‐5.75 *** JNK Y Akt Fos 2.43546 0.75735 ‐3.22 ** JNK Y Wnt Pdgfra 0.00004 0.00000 ‐h * Wnt Nrcam 0.00004 0.00000 ‐2156.43 ** Wnt Ptgs2 0.06663 0.00003 ‐2067.02 ** JNK Y Wnt Cdh1 0.00059 0.00000 ‐652.02 * Wnt Il6 0.00033 0.00001 ‐31.06 * Wnt Ets2 0.15062 0.00503 ‐29.96 ** Wnt Abcb1a 0.00066 0.00003 ‐20.04 ** Wnt Fosl1 0.08123 0.00407 ‐19.95 ** JNK Y Wnt Egr1 0.64400 0.06151 ‐10.47 ** Wnt Pitx2 0.00003 0.00000 ‐7.56 * Wnt Mmp9 1.60566 0.27558 ‐5.83 ** JNK/Wnt YY Wnt Plaur 1.11144 0.19826 ‐5.61 ** JNK Y Wnt Klf5 0.03508 0.00889 ‐3.95 ** Wnt Id2 0.22211 0.08660 ‐2.56 ** Wnt Y Wnt Cebpd 0.25545 0.11006 ‐2.32 ** Y Wnt Runx2 0.02311 0.01171 ‐1.97 * Wnt/TNFα Y ROS Ehd2 0.00045 0.00000 ‐h ** ROS Gpx7 0.00019 0.00000 ‐h * ROS Rag2 0.00103 0.00003 ‐34.61 * ROS Txn1 0.00164 0.00016 ‐10.11 *** ROS Apc 0.00391 0.00152 ‐2.57 * Wnt X ROS Fth1 3.01120 1.40767 ‐2.14 * a Fold change, upregulation: AML/Control, downregulation: ‐Control/AML b * P<0.05, ** P<0.01, *** P<0.0001 c Involvement in cell signaling networks d S/P – effects on survival/proliferation e A/T – effects on apoptosis/tumor suppression f M – effects on mutagenesis g X – enhance, Y – suppress h Zero value denominator

157

158

Table 4.3.5c. Progressive up or downregulation of total BMCs gene expression

(2^‐ΔCT value) from 5 months irradiated mice (0/0), to 7 months irradiated mice (0/0), to AML mice.

Fold P‐ Fold P‐ Symbol 5 months 7 months AML Networkc S/Pd A/Te Mf Δa valueb Δa valueb Fzd7 0.00103 0.00169 1.63 t* 0.03302 31.97 a* Wnt Xg

Gss 0.00283 0.00426 1.51 t*** 0.08307 29.36 a** AODh X

Ccs 0.00442 0.00631 1.43 t* 0.03178 7.19 a* AOD X

Itgb1 0.03274 0.04886 1.49 t* 0.10530 3.22 a* Akt X Yg

Cebpd 0.2555 0.18780 ‐1.36 t* 0.11010 ‐2.32 a** Y

Jun 2.7540 1.9620 ‐1.40 a* 0.4791 ‐5.75 a*** JNK Y

Nfkbia 0.9025 0.5521 ‐1.63 a* 0.0850 ‐10.62 a*** NF‐κB Y a Fold change, later group/5 months for upregulation, ‐5 months/later group for downregulation b Significant difference compared to 5 months group by ANOVA(a) or t‐test(t). * P<0.05, ** P<0.01, *** P<0.0001 c Involvement in cell signaling networks d S/P – effects on survival/proliferation e A/T – effects on apoptosis/tumor suppression f M – effects on mutagenesis g X – enhance, Y – suppress h AOD – antioxidant defenses )

T 0.4 C ‐Δ (2^ 0.3 a* level

0.2 a*** 0.1 expression

0.0 Gene 7 months 7 mo. Iron AML AML#

Figure 4.3.5b. Progressive downregulation of Pten in total BMCs from 7 months irradiated mice (0/0), to 7 months iron‐loaded mice (7.5/0), to AML mice. An outlier from the AML group was removed to create the AML# group. Significant differences compared to the 7 months group were marked by “a” for ANOVA. * P<0.05, *** P<0.0001.

159

4.3.6 The early effects of iron on mice at 3 months after irradiation

In an effort to discern changes occurring prior to the development of frank leukemia, we assessed the in vivo effects of iron on the BMCs of the irradiated mice at earlier time points. Control and 5 mg iron-loaded mice (N=3 per treatment group) were sacrificed at 3 months after irradiation and first iron injection (Figure 4.3.3a) – none of these mice had developed leukemia. Iron loading at 5 mg was well tolerated by the irradiated mice with no visible impairment or weight difference when compared to the control mice (Table 4.3.6). Although there was no obvious bronze discoloration of organs and body cavity, Prussian blue staining confirmed the presence of iron in the liver, spleen and BM of the iron-loaded mice (Figure 4.3.6). In addition, assessment of the BMCs by flow cytometry revealed significant expansion of Lin-CD45+ and Lin-CD45low/- populations (Table 4.3.6). Expansion of these populations may be indicative of early impairment in differentiation and/or expansion of pre-leukemic hematopoietic cells. Interestingly, the iROS level of the Lin-CD45+ population was significantly lower in the iron-loaded group than in the control group (Table 4.3.6).

Table 4.3.6. Early analysis of control and 5 mg iron‐loaded irradiated mice at 3monthspost‐irradiation.

Treatment Control Iron % Change t‐test N33 Irradiation (Gy) 3 3 Dexamethasone (mg) 0.5 0.5 Total iron burden (mg) 0 5 Mean initial body weight (g) 29.23±1.31 30.07±3.71 2.87% P=0.73 Mean final body weight (g) 45.33±1.90 46.20±6.45 1.99% P=0.83 Mean final liver weight (g) 2.49±0.60 3.38±0.84 35.74% P=0.21 Mean final spleen weight (mg) 100±40 96.67±30.55 ‐3.33% P=0.91 Lin‐CD45low/‐ BMCs, frequency of parent (%) 0.71±0.14 2.18±0.89 67.43% P<0.05 Lin‐CD45+ BMCs, frequency of parent (%) 6.08±0.17 6.85±0.32 11.24% P<0.05 Mean iROS in Lin‐CD45+ BMCsa 1.00±0.03 0.68±0.07 ‐31.66% P<0.005 a DCF MFI normalized to control

160

161

4.3.7 The effects of iron on mice at 5 months after irradiation

Next we assessed the in vivo effects of iron at 5 months after irradiation on BMCs from control mice, 5 mg iron-loaded mice, and mice receiving a 5 mg iron load with 40 mg/kg/day ICT (N=5 per treatment group, Figure 4.3.3a). There was no early mortality and we did not observe visible appearance or weight difference among the 3 treatment groups; none of the mice developed leukemia. BMCs were collected and analyzed at the molecular level. When compared to the control group, we observed lower iROS levels in CD45+ BMCs of the iron-loaded group (ANOVA P<0.0001, Table 4.3.7a). The iROS level in the iron/ICT group was between that of the control (P<0.01) and iron-loaded group (P<0.05). We determined the number of AP sites in the irradiated bone marrow cells and found that some iron-loaded mice have more AP sites (Bartlett’s test for equal variances P<0.001, Table 4.3.7a). We also used intracellular flow cytometry to examine the phosphorylation of H2AX to γH2AX in response to DNA damage (Table 4.3.7a). Quantification of γH2AX is a useful tool to detect low-levels of DNA damage. We observed significant increase of γH2AX in CD45+ BMCs of the iron-loaded group when compared to the control group (ANOVA P<0.05). In the iron/ICT group, γH2AX level was elevated but the difference compared to controls was no longer statistically significant. We further examined the status of the following signaling pathways in CD45+ BMCs by intracellular flow cytometry: activation of Akt by Ser473 phosphorylation and activation of Foxo3a by Ser318/321 dephosphorylation. Compared to the control group, both iron and iron/ICT groups had higher level of activated Akt and Foxo3a (Table 4.3.7a).

We used quantitative RT-PCR arrays to assess the transcription of genes related to PI3K-Akt, Wnt and oxidative stress in the total BMCs. To screen for transcription regulation, the expression level of a gene of interest from the iron group was compared to that of the control group by Student’s t-test (Table 4.3.7b, Figure 4.3.5a). The iron/ICT group was subsequently included and the 3 groups were evaluated by ANOVA. Although we observed activation of Akt by Ser473 phosphorylation, some PI3K-Akt related genes were downregulated in the iron-loaded mice when compared to the control mice, including Jun (ANOVA P<0.05), Bad (t-test P<0.05), Elk1 (t-test P<0.05), Hras1 (t-test P<0.05), Mtor (t-test P<0.05), and Prkcb (t-test P<0.05). The expression level of Jun was also decreased in the iron/ICT group when compared to the control group (ANOVA P<0.05), while the expression level of other genes were between the control and 162 iron groups. For Wnt related genes, iron-loading was associated with the upregulation of Efnb1 (t-test P<0.01), Fzd7 (ANOVA P<0.01), and Met (ANOVA P<0.01), as well as downregulation of Tcf7l1 (ANOVA P<0.05). In the iron/ICT group, Met remained upregulated (ANOVA P<0.01) and Tcf7l1 remained downregulated (ANOVA P<0.05). For genes related to oxidative stress, Atr was upregulated (t-test P<0.05) in the iron group, while Mpo (ANOVA P<0.05) and Ptgs2 (ANOVA P<0.01) were downregulated. Both Mpo (ANOVA P<0.01) and Ptgs2 (ANOVA P<0.05) remained downregulated in the iron ICT group. There was no significant difference in the expression of Akt1/2 (gene product of Akt) and Foxo3 (gene product of Foxo3a) among the treatment groups. Nevertheless, the expression of these genes were altered by less than 2.5 fold when compared to control, suggesting that 5 mg iron loading did not result in substantial transcription dysregulation.

At 5 months after irradiation, iron loading induces DNA damage response in the BM. When compared to BMCs from irradiated mice without iron loading (iBMCs), we observed upregulated transcription of Atr and increased phosphorylation of histone H2AX to γH2AX in the iron- loaded irradiated BM cells (iiBMCs). The gene product of Atr, ataxia telangiectasia and Rad3- related (ATR) protein kinase, is a key enzyme in sensing single-stranded DNA damage during DNA replication, which then activates the ATR-Chk1 pathway and leads to cell cycle arrest (Smith et al., 2010). The ATR-Chk1 pathway can also be induced by ataxia telangiectasia mutated (ATM) in response to DNA double-strand breaks (Rhind, 2009). Phosphorylation of H2AX can be mediated by ATR and ATM in response to DNA single and double strand breaks (Smith et al., 2010). We also examined the number of AP sites in the irradiated bone marrow cells and found that some iron-loaded samples have more AP sites. AP sites are locations in DNA that have neither a purine nor a pyrimidine base as a result of DNA glycosylase activity or ROS-induced DNA damage.

Both iron loading and irradiation are known to induce TNFα expression, which is a crucial molecule for systemic inflammation and apoptotic cell death (Brown et al., 2006; Veeraraghavan et al., 2011; Zelova and Hosek, 2013). Although the level of TNFα was not directly assessed, we observed its downstream effects including activation of Akt by phosphorylation at Ser473 and downregulation of Mpo (myeloperoxidase, MPO) transcription in the iiBMCs. Activation of Akt by TNFα has been implicated in promotion of squamous cell carcinoma (Faurschou and 163

Gniadecki, 2008), while Mpo transcription is suppressed by radiation-induced TNFα production in human HL60 promyelocytes (Hachiya et al., 2000). In addition, TNFα is known to activate NF-κB and JNK (Nakano, 2004), and activated NF-κB subsequently suppresses prolonged JNK activation in order to avoid apoptosis (Wicovsky et al., 2007). Indeed, JNK regulated genes Jun and Elk1 were downregulated in the iiBMCs. We further observed the downregulation of Hras1 in the iiBMCs. Aberrant activation of H-Ras can enhance AP-1 (c-Jun/c-Fos complex) activity via the MAPK/ERK pathway to increase the transcription of JNK regulated genes (Deng and Karin, 1994), as well as inactivate Akt through a negative feedback loop (Courtois-Cox et al., 2006). Therefore, lower expression and possibly lower activation of H-Ras may partly account for the downregulation of Jun and Elk1, and activation of Akt. JNK inactivation may also explain Ptgs2 downregulation in the iiBMCs, whose gene product is the inflammation enzyme COX-2. Induction of Ptgs2 mRNA levels by LPS in activated macrophages is suppressed by a JNK inhibitor (Gayatri and Chatterjee, 1991). JNK and COX-2 are also involved in the production of iROS (Ventura et al., 2004; Xu et al., 2006), and their downregulation may contribute to the lower iROS levels in iiBMCs. Downregulation of Bad may suppress apoptosis in conjunction with JNK inactivation. Phosphorylation of Bad by JNK has been shown to promote cell death (Dhanasekaran and Reddy, 2008). Additionally, Prkcb was downregulated while Efnb1 was upregulated in the iiBMCs, the effect of which may be to dampen the pro-apoptotic effects of TNFα since in human vascular endothelial cells downregulation of protein kinase C beta type (gene product of Prkcb) attenuates TNFα-induced ROS production and apoptosis (Wang et al., 2009), while upregulation of ephrin-B1 (gene product of Efnb1) reduces TNFα-induced ERK1/2 activation in adipocytes (Mori et al., 2013). Therefore, we conjecture that treatment with irradiation followed by iron loading induces a TNFα response in iiBMCs at 5 months after irradiation, but part of the response – apoptosis – is suppressed by compensatory mechanisms, allowing these cells to survive.

In addition to increased TNFα signaling and downregulation of Hras1, Akt activation in the iron- loaded BM cells may also be augmented by Met upregulation and mTOR downregulation. Akt is involved in multiple cellular process including promotion of proliferation and inhibition of apoptosis (Chan et al., 2014). Autocrine activation of c-met (gene product of Met) by hepatocyte growth factor (HGF) is frequently observed in AML and correlated with Akt activation (Kentsis

164 et al., 2012; Trovato et al., 2013). Activation of Akt by mammalian target of rapamycin (mTOR, gene product of Mtor) varies depending on the ratio between mTORC1 and mTORC2, with higher mTORC2 level favoring pAkt expression (Mori et al., 2014). Inactivation of mTORC1 is also associated with the transcription upregulation and activation of Foxo3a (Mori et al., 2014). Since we observed Foxo3a and Akt activation in the iiBMCs, the level of mTORC2 is likely higher than mTORC1 in these cells. In addition, Foxo3a activation by dephosphorylation can protect quiescent cells from ROS-induced apoptosis by increasing MnSOD expression, which leads to JNK inactivation after cells become tolerant to oxidative stress (Kops et al., 2002). Although MnSOD transcription was not increased in the iiBMCs (not shown), we observed lower iROS levels and transcription repression of JNK regulated genes, suggesting that these cells might have already adapted to handle the oxidative stress. Furthermore, Foxo3a can be activated by COX-2 silencing (Li et al., 2011), and Foxo3a activation can reduce iROS level. Conversely, COX-2 downregulation should also suppress Akt activation (Li et al., 2011), but instead we observed upregulation of pAkt level. Thus, we conjecture that the elevated level of pAkt is the net result of the interactions between multiple regulators including TNFα, H-Ras, c- Met, mTOR, and COX-2.

The canonical Wnt pathway is initiated via the engagement of the cell surface receptor frizzled (Fzd) by Wnt, leading activation of Dvl, which in turn prevents the destruction complex, comprised of Axin, APC, PP2A, GSK3, and Ck1α, from phosphorylating β-catenin at Ser33/Ser37/Thr41, thereby allowing this protein to avoid being targeted for subsequent degradation (Rao and Kuhl, 2010). Active dephosphorylated β-catenin accumulates in the nucleus, forms a transcription activator complex with Tcf, and promotes transcription of target genes. We observed upregulated Fzd7 transcription, which is an important mediator of Wnt signaling (Katoh, 2008). Fzd7 is required to maintain the pluripotent properties of human embryonic stem cells (Fernandez et al., 2014), and upregulation of Fzd7 has been reported in multiple cancers (Katoh and Katoh, 2007). In addition, downregulation of Tcf7l1 in the iiBMCs may functionally enhance β-catenin activity. Tcf7l1 binds to chromatin and represses the transcription of Wnt target genes (Kuwahara et al., 2014). Inactivation of Tcf7l1 by β-catenin or downregulation of Tcf7l1 transcription decreases chromatin occupancy and increases target gene transcription (Shy et al., 2013). We further observed the upregulation of Efnb1 in the iBMCs.

165

Ephrin-B1 is overexpressed in a number of carcinomas and may promote tumor growth, angiogenesis and cell migration (Sawai et al., 2003). However, Efnb1 has been reported to be upregulated upon β-catenin/TCF inhibition in differentiated cells of the intestinal epithelium (Batlle et al., 2002). Also, downregulation of Ptgs2 may not be consistent with elevated Wnt signaling. Ptgs2 expression is upregulated by the β-catenin/Tcf transcription activator complex, which cannot take place in the absence of activated β-catenin in the nucleus (Nunez et al., 2011). Moreover, prostaglandin E2, the bioactive product of COX-2, is able to activate β-catenin by suppressing the β-catenin destruction complex (Buchanan and DuBois, 2006). The inconsistency may be the consequence of interference by other signaling pathways, or compensatory mechanisms after prolonged exposure to excess iron.

Our data suggest that iron loading and irradiation in mouse BM cells enhances DNA damage response and dysregulates signaling activities. DNA damage response is upregulated in the iiBMCs, suggesting that these cells are under continuous iron-induced oxidative stress resulting in mutagenesis. Activation of TNFα and Akt with low JNK activation is indicative of cell survival with low apoptosis. Altered expression and activity of Foxo3a, and Ptgs2 may suggest suppressed inflammatory response and iROS production. Some aspects of the Wnt signaling pathway are also dysregulated. Conversely, we did not observe significant transcriptional variation in association with iron loading for PTEN, cyclin D1 or c-Myc, genes related to carcinogenesis.

In iron-loaded BMCs obtained at 5 months after irradiation, ICT appeared to have the effect of slightly decreasing DNA damage as a result of iron loading. The expression of γH2AX in iron chelated iiBMCs (ciiBMCs) was higher but not statistically significantly when compared to iBMCs. In contrast, iiBMCs had significantly higher γH2AX level than iBMCs. The variability of the number of AP sites in ciiBMCs was also lower than iiBMCs and became more comparable to iBMCs. In addition, ICT appeared to revert some of the iron-induced gene expression changes. The expression of these genes in ciiBMCs deviated towards iBMCs from iiBMCs, or their distributions spanned between iBMCs and iiBMCs. Downregulation of Efnb1 and upregulation of Prkcb in ciiBMCs suggested that TNFα simulation may be decreased in ciiBMCs compare to iiBMCs. It should be noted that in addition to chelating iron, deferasirox is known to suppress NF-κB transcription and TNFα production (Frelin et al., 2005). Upregulation of Elk1, Hras1, and 166

Bad may indicated higher JNK activity upon iron chelation. Although deferasirox did not reduce phosphorylation of Akt at Ser473, the expression levels of both Hras1 and Mtor in the ciiBMCs were higher than in iiBMCs, suggesting possible suppression of Akt activity. Furthermore, iron chelation by deferasirox can efficiently remove the labile iron pool (LIP) and labile plasma iron (LPI) (Ghoti et al., 2010), but non-labile iron stores can only be indirectly reduced as a result of labile iron depletion. If the total iron load cannot be reduced to normal level by the end of ICT, a certain amount of non-labile iron will remain and reconstitute the LIP and LPI. Therefore, the alterations of DNA damage response and gene expression pattern of ciiBMCs may partly be the consequence of reduced total iron load at the time of the analysis.

Table 4.3.7a. Analysis of BMCs from control (0/0, N=5), iron‐loaded (5/0, N=5), iron/ICT (5/40, N=5) mice at 5 months post‐irradiation. AP sites were measured from the DNA extract of total BMCs. Readings from γH2AX, pAkt, pFoxo3a, and iROS were measured from CD45+ BMCs using flow cytometry.

Control Iron P‐valueb Iron/ICT P‐valueb

AP sites (/100,000 bp) 3.54±1.17 4.17±2.18 ‐c 3.71±0.43 ‐

Mean γH2AXa 1.00±0.12 1.25±0.15 a* 1.19±0.13 t*

Mean pAktSer473a 1.00±0.01 1.13±0.03 a*** 1.17±0.02 a***

Mean pFoxo3a Ser318/321a 1.00±0.05 0.64±0.1 a*** 0.66±0.05 a***

Mean iROSa 1.00±0.28 0.63±0.08 a* 0.55±0.20 a*

a MFI normalized to control b Significant difference compared to control group by ANOVA(a) or t‐test(t). * P<0.05, ** P<0.01, *** P<0.0001 c Bartlett’s test for equal variances P<0.001

167

Table 4.3.7b. Altered gene expression in total BMCs (2^‐ΔCT value) from control (0/0), iron (5/0), iron/ICT (5/40) mice at 5 months post‐irradiation.

Fold P‐ Fold P‐ Symbol Control Iron Iron/ICT Networkc S/Pd A/Te Mf Δa valueb Δa valueb Met 0.00085 0.00171 2.01 a** 0.00183 2.16 a*** Akt Xg

Atr 0.00185 0.00338 1.83 t* 0.00293 1.58 ‐ DDRh X

Fzd7 0.00103 0.00156 1.51 a** 0.00133 1.28 ‐ Wnt X

Efnb1 0.00058 0.0008 1.38 t** 0.00078 1.34 ‐ TNFα/Wnt X Yg

Hras1 0.01548 0.01248 ‐1.24 t* 0.01483 ‐1.04 ‐ Akt/JNK X Y

Bad 0.01422 0.01119 ‐1.27 t* 0.01315 ‐1.08 ‐ JNK Y

Mtor 0.01584 0.01235 ‐1.28 t* 0.01329 ‐1.19 ‐ Akt/Foxo3a X

Elk1 0.00644 0.00489 ‐1.32 t* 0.00552 ‐1.17 ‐ JNK Y

Prkcb 0.1431 0.1049 ‐1.36 t* 0.1288 ‐1.11 ‐ TNFα Y

Jun 2.7540 1.8550 ‐1.48 a* 1.909 ‐1.44 a* JNK Y

Tcf7l1 0.00028 0.00014 ‐2.02 a* 0.00013 ‐2.12 a* Wnt X Mpo 0.9557 0.4524 ‐2.11 a* 0.3949 ‐2.42 a** TNFα X Ptgs2 0.06061 0.02787 ‐2.17 a** 0.03768 ‐1.61 a* JNK/Foxo3a X a Fold change, upregulation: treatment group/control, downregulation: ‐control/treatment group b Significant difference compared to control group by ANOVA(a) or t‐test(t). * P<0.05, ** P<0.01, *** P<0.0001 c Involvement in cell signaling networks (control vs iron) d S/P – effects on survival/proliferation e A/T – effects on apoptosis/tumor suppression f M – effects on mutagenesis g X – enhance, Y – suppress h DDR –DNA damage response

168

4.3.8 The effects of iron on mice at 7 months after irradiation

A second cohort was sacrificed and analyzed after the second iron-loading at 7 months post- irradiation. The cohort comprised mice that were treated with dextran injection, 7.5 mg iron injection, and 7.5 mg iron injection with 40 mg/kg/day ICT (N=5 per treatment group, Figure 4.3.3a). All of the mice were alive with no observable impairment by the time of the analysis. None had developed leukemia. BMCs were harvested and analyzed using the same procedure as the 5 months post-irradiation analysis. We compared the control groups at 5 months and 7 months after irradiation. The level of γH2AX was increased in the 7 months control group (t-test P<0.01) along with activation of Akt (t-test P<0.05) (Table 4.3.8a). In the PI3K-Akt pathway array, we observed upregulation of Chuk (t-test P<0.05), Ilk (t-test P<0.05), Itgb1 (t-test P<0.05) and Pik3r2 (t-test P<0.01), as well as downregulation of Jun (t-test P<0.05), Nfkbia (t-test P<0.05) (Table 4.3.8b). In the Wnt pathway array, Cebpd (t-test P<0.05) was downregulated, while Fn1 (t-test P<0.05), Fzd7 (t-test P<0.05), Lrp1 (t-test P<0.05) and Nrp1 (t-test P<0.05) were upregulated (Table 4.3.8b). Ccs and Gss were upregulated in the oxidative stress array (Table 4.3.8b). We did not observe significant difference in the expression of Akt1/2 or Foxo3 among the treatment groups.

We then assessed the effects of iron-loading and iron/ICT on mice at 7 months post-irradiation (Table 4.3.8c, Figure 4.3.5a). There were no significant differences for the levels of iROS, γH2AX, pAkt and pFoxo3a among the 3 treatment groups. We also observed comparable expression of genes related to the Wnt pathway and oxidative stress among the 3 groups. On the other hand, iron loading was associated with the downregulation of genes in the PI3K-Akt pathway array, including Grb2 (t-test P<0.05), Itgb1 (t-test P<0.05), Pik3r2 (t-test P<0.05), Pten (ANOVA P<0.05), and Srf (t-test P<0.05). ICT neutralized the effects of iron loading in the irradiated mice in that the expressions of these genes in the ICT group were upregulated when compared to the iron group. Pten expression was higher in the ICT group than the iron group (t- test P<0.05). Moreover, the expression levels of Grb2, Itgb1, Pik3r2 and Srf in the iron/ICT group were not significantly different when compared to the control and iron-loaded groups.

A number of changes were observed between the 5 month to 7 month analysis time points in bone marrow cells of mice that received irradiation but no iron. DNA damage response,

169 indicated by γH2AX upregulation, was increased in the iBMCs at 7 months post-irradiation when compared to that of the 5 months iBMCs. We also observed higher Akt activation by phosphorylation at 7 months than at 5 months in the iBMCs, along with upregulated Ilk, Pik3r2 and Itgb1 transcription. Ilk is involved in cancer progression and has been shown to phosphorylate Akt at Ser 473 (Persad and Dedhar, 2003). Pik3r2 is overexpressed in breast and colon carcinomas (Cortes et al., 2012). Its upregulation induces PI3K and subsequent Akt activation, resulting in enhanced cell invasion and cancer progression. Knockdown of β1-integrin abolishes EGF-induced Akt activation and impairs regeneration of the liver and proliferation of lung cancer cells (Morello et al., 2011; Speicher et al., 2014). Therefore, β1-integrin expression is important for tumor growth by mediating Akt activation. In addition, β1-integrin is involved in c-Met phosphorylation (Morello et al., 2011), which is involved in Akt activation (Trovato et al., 2013). We further observed upregulation of Fzd7 transcription in the 7 months iBMCs, suggesting augmented Wnt signaling. Furthermore, Jun was downregulated in the 7 months iBMCs, which may indicate decreased AP-1 formation and thereby prevent JNK from promoting apoptosis. JNK inactivation may also be induced by NF-κB activation based on the downregulation of Nfkbia and upregulation of Chuk. IκBα inhibits NF-κB by preventing the entry of NF-κB to the nucleus, while IKKα (gene product of Chuk) activates NF-κB by disabling the interaction between IκBα and NF-κB (Oeckinghaus and Ghosh, 2009). Moreover, we observed altered expression of cancer-related genes in the 7 months iBMCs, including downregulation of tumor suppressor Cebpd, as well as upregulation of Fn1, Lrp1 and Nrp1. Dysregulation of these genes has been documented in many types of cancer and may be involved in carcinogenesis (Jubb et al., 2012; Langlois et al., 2010; Palmieri et al., 2012; Ruoslahti, 1999). Additionally, cellular defense against oxidative stress was enhanced in the 7 months iBMCs as indicated by the upregulation of Ccs and Gss. Ccs encodes a copper chaperone that is responsible for the activation of SOD1 (Wong et al., 2000), and Gss encodes a crucial enzyme for glutathione biosynthesis (Traverso et al., 2013). Both of these genes are known to enhance the antioxidant defense of malignant cells and allow these cells to adapt to high oxidative stress. To summarize, by 7 months irradiation alone caused dysregulations in iBMCs that include JNK inactivation, Akt and Wnt activation, and increased DNA damage and antioxidant defenses.

170

It is interesting to note that the gene and protein expression patterns of iiBMCs at 5 months share some similarities with iBMCs at 7 months, including upregulation of γH2AX, Akt phosphorylation, and Fzd7 transcription, as well as the down regulation of Jun. This observation should not be surprising, since irradiation itself is the primary instigator of AML in this model. It is expected that similar changes will occur in the iBMCs and the iiBMCs, but that the changes will occur on an accelerated schedule in the presence of iron loading.

Iron loading led to further dysregulation in the iBMCs at 7 months post-irradiation. We observed significant transcriptional downregulation of Pten in the iiBMCs when compared to iBMCs. PTEN suppresses the PI3K/Akt/mTOR pathway by preventing Akt activation via PIP3 dephosphorylation (Mester and Eng, 2013). Loss of PTEN function is associated with cellular proliferation, migration, survival, and tumorigenesis. PTEN is therefore a crucial tumor suppressor that is downregulated or mutated in numerous forms of cancer. In HSCs, inactivation of PTEN causes stem cell depletion, which leads to myeloproliferative disease and eventually acute leukemia (Yilmaz et al., 2006; Zhang et al., 2006). The protein level of PTEN may be further decreased by the downregulation of Srf. Serum response factor (SRF, gene product of Srf) depletion has been shown to suppress PTEN expression at a post-transcriptional level in smooth muscle cells (Yu et al., 2002). However, we did not observe further Akt activation in 7 months iiBMCs when compared to the corresponding iBMCs. We suspect Akt activation due to the loss of PTEN may be compensated by the downregulation of Itgb1, Pik3R2 and Grb2. The protein products of these genes are components of signaling pathways that lead to Akt phosphorylation (Cortes et al., 2012; Cully et al., 2006; Morello et al., 2011), and their downregulation may decrease Akt activation. Suppression of Itgb1 transcription may be due to SRF downregulation, which has been shown to cause accumulation of defective HSCs with decreased adhesion and compromised engraftment properties (Ragu et al., 2010). Alternatively, oxidative stress may negatively regulate Itgb1 transcription, since hydrogen peroxide has been shown to suppress Itgb1 transcription via PML (Reineke et al., 2010). The level of γH2AX in the iiBMCs at 7 months was higher than in iBMCs at 5 months, but was comparable to the iiBMCs at 5 months and the iBMCs at 7 months (not shown). This is consistent with a model in which iron loading initially increases DNA damage response in the iiBMCs (as seen at 5 months after irradiation), persisting until 7 months, by which time the non-iron loaded cells have “caught up”.

171

Iron loading increases the duration of DNA damage response in iiBMCs, which may lead to further accumulation of mutations and promote leukemogenesis. Hence, our observations suggest that even though the iiBMCs at 7 months are not overtly leukemic, they have accumulated more mutations than the iBMCs at 7 months and are likely at a more advanced stage of leukemogenesis.

At 7 months, iron chelation mitigated some of the transcriptional effects of iron loading on ciiBMCs. Pten expression in the 7 months ciiBMCs resembled that of the corresponding iBMCs and was higher than that of the iiBMCs (t-test P<0.05). Iron-induced downregulations of Itgb1, Pik3r2, Grb2, and Srf were also partially reversed by ICT, in which the transcription of these genes in ciiBMCs was not significantly different than iBMCs and iiBMCs. The reversal of iron- induced effects on ciiBMCs by ICT may help to explain the lower AML rate in iron- loaded/chelated mice when compared to iron-loaded/non-chelated mice.

Table 4.3.8a. Analysis of BMCs from control (0/0) mice at 5 and 7 months post‐irradiation (N=5 each). Readings from γH2AX, pAkt, pFoxo3a, and iROS were measured from CD45+ BMCs using flow cytometry.

5 months 7 months P‐valueb Mean γH2AXa 1.00±0.12 1.52±0.21 ** Mean pAkt Ser473a 1.00±0.01 1.74±0.64 * Mean pFoxo3a Ser318/321a 1.00±0.05 1.09±0.07 ‐ Mean iROSa 1.00±0.28 0.63±0.24 ‐ a MFI normalized to control b Significant difference compared to 5 months group by t‐test(t). * P<0.05, ** P<0.01.

172

Table 4.3.8b. Altered gene expression in total BMCs (2^‐ΔCT value) from control (0/0) mice at 5 and 7 months post‐irradiation (N=5 each).

Fold P‐ Symbol 5 months 7 months Networkc S/Pd A/Te Mf Δa valueb Fn1 0.05503 0.1103 2.00 * Yg Lrp1 0.00756 0.01512 2.00 * Y Pik3r2 0.00129 0.00212 1.64 ** Akt Xg Fzd7 0.00103 0.00169 1.63 * Wnt X Chuk 0.01816 0.02846 1.57 * NF‐κB Y Nrp1 0.01058 0.01638 1.55 * Y Gss 0.00283 0.00426 1.51 ** AODh X Itgb1 0.03274 0.04886 1.49 * Akt X Y Ccs 0.00442 0.00631 1.43 * AOD X Ilk 0.07148 0.08823 1.23 * Akt X Cebpd 0.2555 0.1878 ‐1.36 * Y Jun 2.754 1.962 ‐1.40 * JNK/NF‐κB Y Nfkbia 0.9025 0.5521 ‐1.63 * NF‐κB Y a Fold change, 7/5 months for upregulation, ‐5/7 months for downregulation b Significant difference compared to 5 months group by t‐test(t). * P<0.05, ** P<0.01 c Involvement in cell signaling networks d S/P –effects on survival/proliferation e A/T –effects on apoptosis/tumor suppression f M –effects on mutagenesis g X – enhance, Y – suppress h AOD – antioxidant defenses

173

Table 4.3.8c. Altered gene expression in total BMCs (2^‐ΔCT value) from control (0/0), iron (7.5/0), iron/ICT (7.5/40) mice at 7 months post‐irradiation.

Fold P‐ Fold P‐ Symbol Control Iron Iron/ICT Networkc S/Pd A/Te Mf Δa valueb Δa valueb Srf 0.02936 0.02279 ‐1.29 t* 0.02812 ‐1.04 ‐ Akt Yg

Grb2 0.3183 0.2446 ‐1.30 t* 0.2875 ‐1.11 ‐ Akt Y

Itgb1 0.04886 0.03725 ‐1.31 t* 0.04385 ‐1.11 ‐ Akt Y Xg

Pten 0.2214 0.1555 ‐1.42 a* 0.2125 ‐1.04 ‐ Akt Y

Pik3r2 0.00212 0.00148 ‐1.43 t* 0.00222 1.05 ‐ Akt Y a Fold change, upregulation: treatment group/control, downregulation:‐control/treatment group b Significant difference compared to control group by ANOVA(a) or t‐test(t).* P<0.05. c Involvement in cell signaling networks (control vs iron) d S/P – effects on survival/proliferation e A/T – effects on apoptosis/tumor suppression f M – effects on mutagenesis g X – enhance, Y – suppress

174

4.4 Discussion

In inherited anemia and in genetic hemochromatosis, severe chronic iron overload leads to early mortality mainly by inducing hepatic and cardiac injury (Bartfay and Bartfay, 2000; Bartfay et al., 1999; Prati, 2000). ICT removes excess iron and improves survival, but incomplete removal of iron may increase the risk of carcinogenesis, including hepatocellular carcinoma and hematopoietic neoplasia for thalassemia patients with secondary iron overload (Borgna-Pignatti et al., 2014; Karimi et al., 2009). Furthermore, the risk of hepatocellular carcinoma is elevated for chronically transfused thalassemia patients with cirrhotic liver or viral hepatitis (Maakaron et al., 2013; Mancuso, 2010), suggesting that secondary iron overload may further promote carcinogenesis in organs that are already in a premalignant state. Anemic MDS patients acquire iron overload as a consequence of chronic transfusion (Malcovati, 2009). However, unlike thalassemia, the hematopoietic tissues of MDS patients are vulnerable to leukemic transformation. We therefore hypothesized that iron loading may accelerate AML progression in MDS. Our study was designed to investigate the potential of iron loading to induce impairment and eventually AML, specifically when the hematopoietic tissue is in a premalignant state.

B6D2F1 mice are sensitive to RI-AML

To answer this question we used a standard radiation/dexamethasone protocol to induce AML in B6D2F1 mice, which have been utilized commonly in studies of iron loading. Although there have been no reports of the rate of RI-AML in B6D2F1 mice, we reasoned that this model should show an intermediate sensitivity to radiation, since it is the F1 progeny of a radiation sensitive (DBA2) and a radiation insensitive (C57Bl/6) strain. Indeed, our B6D2F1 cohort developed RI- AML with prolonged latency at beyond 18 months. Iron is a potent promoter of RI-AML with reduced latency at 7 months and 40% incidence rate. In comparison, the frequency of RI-AML in the SJL/J strain is 10-30% at approximately 1 year post-irradiation (Rivina et al., 2014), and in addition, corticosteroid treatment increases the AML rate to 50-70% at a mean latent period of 300 days post-irradiation (Resnitzky et al., 1985). Also, the frequency of RI-AML in the CBA strain is 25% at 18-24 months (Major, 1979). We also speculate that iron loading improves survival and proliferation of radiation-induced leukemia-initiating cells, based on the increased Lin-CD45+ and Lin-CD45low/- populations in the iiBMCs at 3 months post-irradiation.

175

Leukemia-promoting iron dosages are consistent with previous studies of iron and cancer

We confirmed the presence of excess iron in the bone marrow of wild-type B6D2F1 mice after intraperitoneal injection of iron dextran at total doses from 5 to 150 mg. High dose iron loading at 150 mg was severely detrimental to the subjects and increased apoptosis in the BM, while low dose iron loading from 5 to 20 mg was well tolerated at 3 months after iron injection and its effects on the BM and PB were mild. Recent reports have suggested that high dose/short term iron loading in C57BL/6 mice impairs the BM microenvironment, increases oxidative stress and reduces the number of various colony-forming units (Chai et al., 2013; Okabe et al., 2014). Chai et al. have also reported that lower dose/short term loading at 35 mg total iron burden induces extramedullary erythropoiesis and myelopoiesis in the spleen and increases oxidative stress (Chai et al., 2014; Chai et al., 2013). In the PB, the numbers of red and white blood cells were not significantly affected by either high or lower dose iron loading, suggesting that the effects of iron overload on hematopoiesis was mild and could be compensated.

In animal models, iron overload induces or promotes various types of tumors (Beguin et al., 2014). However, iron loading is not known to induce leukemia as a sole agent and there are few studies that have evaluated the role of iron overload to promote leukemogenesis. We employed the RI-AML mouse model to assess the effect of iron loading on leukemogenesis. Our data show that irradiation induces RI-AML in B6D2F1 mice. Iron loading shortened the latency of RI-AML and increased the incidence rate. RI-AML was most prominent in the 7.5 mg iron-loaded group in which 40% of the mice developed AML starting at 7 months after irradiation. In other studies, cumulative iron dose by parenteral injection for the induction or promotion of cancer varies from 2.3 mg to 640 mg based on a 35 g mouse (Beguin et al., 2014). A large portion of these studies were conducted using iron dextran at approximately 21 g of total iron burden. The total iron dose to promote RI-AML in this study is therefore generally consistent with the literature. Nevertheless, a direct comparison cannot be made between BM and other tissues regarding their carcinogenic sensitivity to iron due to the use of different methods to induce carcinogenesis, and because the basal iron level and the severity of iron accumulation are known to be mouse strain dependent (Fleming et al., 2001; Hahn et al., 2009). The iron status of an individual is also influenced by other factors including age, gender and cell/tissue type (Hahn et al., 2009). These factors may influence the tolerance and physiological response of BMCs to excess iron, thereby 176 modulating the effects of iron loading to induce cell damage and promote leukemogenesis. Therefore, comparisons of iron sensitivity across studies should be conducted with caution.

Dose-response relationship for iron loading and leukemogenesis

Interestingly, the occurrence of RI-AML and total iron burden was inversely correlated. There was a progressive decrease in AML frequency and corresponding increase in latency for irradiated mice with iron burden from 7.5 to 15 to 30 mg. In addition, ICT by deferasirox on the 7.5 and 30 mg iron-loaded irradiated mice produced opposite effects – we observed lower AML incidence in the 7.5 mg iron-loaded mice that received ICT (compared to 7.5 mg iron-loaded mice that received no ICT), while AML onset was earlier for 30 mg iron-loaded mice that received ICT (compared to 30 mg iron-loaded mice that received no ICT). Bronze discoloration was observed in the ICT-treated 30 mg iron-loaded mice, suggesting that our ICT protocol did not result in the complete removal of excess iron. We conjecture that the risk of AML is influenced by the pro-apoptotic and pro-mutagenic properties of iron at different doses (Figure 4.4a). In this model, iron-induced oxidative stress damages cellular components, including DNA, lipid, and protein. Incomplete repair of this damage contributes to leukemogenesis. The magnitude of AML risk correlates with iron dose but plateaus after a certain iron burden is reached. On the other hand, iron also has the potential to induce cell death by apoptosis or ferroptosis; induction of cell death by iron has a steeper dose-response curve and surpasses the rate of leukemia induction at high doses. Thus, cells with very high iron burden undergo apoptosis or ferroptosis instead of transforming into leukemia-initiating cells. Consequently, at higher iron doses the reduction of cell viability lowers the risk of AML. The resulting two-stage effects of iron loading on AML risk resembles a biphasic dose-response curve, which accounts for the initial rise of AML risk due to mutagenesis and subsequent drop due to reduced viability. A two-stage mutation model has also been proposed for radiation carcinogenesis, in which the risk of cancer increases up to a certain radiation dose and then declines at higher dose (Hall, 2000; Lindsay et al., 2001). This notion was further demonstrated by Di Majo et al. using male CBA/Cne mice, in which RI-AML peaked at 3 Gy and declined at 5 and 7 Gy (Di Majo et al., 1996). Our observation on the 30 mg iron-loaded mice also leads to the speculation that ICT in moderately iron overloaded MDS patients would be anticipated to reduce AML risk, incomplete ICT in heavily iron overloaded MDS patients could conceivably increase their risk of AML. 177

ICT ICT

7.5 15 30 mg mg mg Pro‐death

AML rate effect

Pro‐mutagenesis Magnitude

Iron dose

Figure 4.4a. Proposed pro‐death and pro‐mutagenic influences of iron loading and ICT on the rate of AML.

178

Iron loading accelerates progressive changes initiated by irradiation that culminate in leukemia, while ICT reduces this effect

Assessments of the irradiated BMCs from the control, 5/7.5mg iron-loaded, and iron-loaded/ICT group prior to development of AML suggested a pattern of progressive molecular alterations including alteration of signaling pathways, DNA damage response, and gene expression pattern. We characterized 3 stages of leukemogenesis of RI-AML without iron loading from pre-AML stage 1 at 5 months post-irradiation, to pre-AML stage 2 at 7 months post-irradiation, to eventual AML beyond 18 months (Figure 4.4b). There were progressive alterations including activation of Akt, NF-κB, Wnt, antioxidant defenses, and DNA damage response, as well as inactivation of JNK, C/EBPδ, and PTEN. Iron loading appeared to induce intermediate stages between stages 1 and 2 as well as stage 2 and AML, which we defined as stage 1a and 2a, respectively. At 5 months post-irradiation for stage 1a, 5 mg iron loading induced alterations that partly resemble stage 2, in which changes to Akt, NF-κB, JNK, and DNA damage response were observed when compared to stage 1. Conversely, the expression C/EBPδ and the expression of antioxidant genes were altered in stage 2, but this was not the case for the 5 months iiBMCs. Therefore, we conclude that stage 1a is at a more advanced leukemogenic state than stage 1 but has not yet reach stage 2. Moreover, iron loading had additional effects including activation of Foxo3a. The additional effects may be cellular responses to handle the excess iron. Alternatively, non-iron- loaded iBMCs may display a similar expression pattern that resembles stage 1a when they transform from stage 1 to 2. At 7 months post-irradiation for stage 2a, the tumor suppressor PTEN was downregulated upon in the iiBMCs, which is a critical step towards AML development. We therefore conclude that stage 2a is a possible intermediate state between stage 2 and AML. Furthermore, the more advanced leukemogenic state and sustained DNA damage may be responsible for the earlier onset and higher incidence of AML for the iron-loaded mice. Our ICT regimen did not completely reverse the effects of iron loading on the ciiBMCs at 5 and 7 months post-irradiation. ICT partly reduced iron-dependent DNA damage response at 5 months, and dampened iron-induced transcription dysregulations of many genes at 5 and 7 months, and eventually lowered AML incidence. Therefore, we concluded that the leukemogenic states of ciiBMCs were between stage 1 and 1a at 5 months post-irradiation, and between stage 2 and 2a at 7 months post-irradiation.

179

Irradiated Irradiated Irradiated Normal / Control +Iron +Iron +ICT Signaling networks (iBMCs) (iiBMCs) (ciiBMCs) S/P A/T Akt ↔ JNK ↔ Pre‐AML NF‐κB ↔ C/EBPδ ↔ 5 months Stage 1 Wnt ↔ PTEN ↔ AOD ↔ DDR ↔ M 5 months Akt↑ JNK↓ Pre‐AML NF‐κB↑ C/EBPδ ↔ 5 months Stage 1a Wnt ↔ PTEN ↔ AOD ↔ DDR↑

Akt↑ JNK↓ Pre‐AML NF‐κB↑ C/EBPδ↓ Stage 2 7 months Wnt↑ PTEN ↔ AOD↑ DDR↑

7 months Akt↑ JNK↓ Pre‐AML NF‐κB↑ C/EBPδ↓ Stage 2a 7 months Wnt↑ PTEN↓ AOD↑ DDR↑

Akt↑↑ JNK↓↓ 7 months 9 months NF‐κB↑ C/EBPδ↓↓ >18 months AML 40% AML 20% AML Wnt↑↑ PTEN↓↓ AOD↑↑ DDR ↔

Figure 4.4b. RI‐AML progression model and the influence of iron or iron/ICT to leukemogenesis. AOD – antioxidant defenses, DDR – DNA damage response, S/P – effects on survival/proliferation, A/T – apoptosis/tumor suppression, M –mutagenesis

180

Effect of iron loading on intracellular ROS levels in bone marrow cells

We observed elevated iROS level in non-irradiated BMCs after high dose/short term iron loading. Other studies have also reported similar findings (Chai et al., 2014; Okabe et al., 2014). Conversely, the effect of long term iron loading on iROS level in the BMCs was not as prominent when compared to short term loading. The level of iROS even decreased in the iiBMCs. This discrepancy may be due to the time difference between iron loading and actual analysis. Our short term analysis and other studies examined the BMCs shortly after the end of iron loading, whereas the subjects in our long term analysis survived for at least 3 months before their BMCs were assessed. The overall iROS level is influenced by physiological iROS producing apparatus such as the respiratory chain (Devasagayam et al., 2004), as well as non- physiological factors such as iron loading and irradiation. Also, iron-induced generation of hydroxyl radicals via Haber-Weiss reaction relies on physiological production of hydrogen peroxide (Haber, 1934). In short term iron loading, non-physiological iROS level is increased while physiological iROS production remains unchanged, resulting in higher overall iROS level. To avoid cell death, BMCs need to adapt to oxidative stress over time by suppressing physiological iROS production. Analysis of gene expression and signaling pathways revealed the activation of Foxo3a, downregulation of Ptgs2, and inactivation of the JNK pathway upon iron loading and irradiation. These changes are known to reduce iROS levels and protect cells from elevated oxidative stress (Ferber et al., 2012; Sakon et al., 2003; Xu et al., 2006). The non- physiological iROS level remains high during the long incubation period, but it is offset by the suppression of physiological iROS production. Nevertheless, iiBMCs have elevated DNA damage response, which was indicative of iron-induced oxidative DNA damage despite suppressed iROS production. This phenomenon may be explained by the ability of Fe3+ to bind to DNA with high affinity, which then contributes to oxidative damage of DNA due to its proximity to nucleotides (Netto et al., 1991; Sonntag et al., 2004). We also observed a progressive decrease of iROS level in the BMCs from mice loaded with 5 to 20 mg iron at 3 months after iron loading. Elevated iROS level at 5 mg iron suggested a lag in cellular adaptation response to decrease physiological iROS production at a low iron level. The adaptation response was initiated at a certain iron level and increased with higher iron dose, resulting in the reduction of the overall iROS level.

181

Implications for MDS

The pathogenesis of RI-AML shares several similarities with that of MDS, including the early emergence of leukemia-initiating cells, the long latency before AML transformation, and the disrupted bone marrow microenvironment. Radiation-induced BM injury is a valid approach to mimic the pre-leukemic state in the BM of MDS patients. Our study confirmed the ability of iron to promote RI-AML. This supports the notion that secondary iron overload as a consequence of chronic transfusion may accelerate AML transformation in MDS. Molecular changes similar to those seen in the iron-loaded irradiated BMCs may also occur in the BMCs from MDS patients. ICT dampened some of the iron-induced alteration in the irradiated BMCs and improved RI- AML survival. Conversely, ICT increased AML risk for irradiated mice with high iron burden. Therefore, the total iron burden may have a crucial role in determining the AML risk in RI-AML as well as in MDS. However, the relationship may be complicated due to the biphasic dose- response nature between iron and AML. Risk estimation will require the assessment of the pro- apoptotic and pro-mutagenic effects of iron on the HSCs from MDS patients with varying iron burden. If the AML risk peaks at an iron dose that is higher than relevant iron burden in the clinical setting, then ICT may be beneficial to all iron overloaded MDS patients, especially those who are already at higher risk of AML transformation. On the other hand, if the AML risk peaks at an iron dose that is within the range of clinically relevant iron burden, then adequate ICT will be crucial for patients with severe iron overload. Moreover, early intervention to control iron burden by ICT may be an effective measure to avoid the potential AML risk due to iron overload. Current guidelines regarding the use of ICT in MDS patients with iron overload do not take into account the potential AML risk due to iron overload (Gattermann, 2008). It is also uncertain if current recommended serum ferritin target (usually at 1,000 µg/L) for ICT is sufficiently low to avoid the potential iron-related AML risk. Thus, our findings point to the necessity of further research to verify the role of secondary iron overload to accelerate AML transformation in MDS.

182

Chapter 5

Conclusions and future directions

183

5.1 Summary and conclusions

Our analysis of initial transfusion intensity (ITI) on MDS patients (Chapter 2) demonstrated that:

 Patients who began RBC transfusion to alleviate the symptoms of anemia could be separated in to two ITI groups based on whether they required additional transfusion (group 2, “high ITI”) or not (group 1, “low ITI”) within the 4 weeks after the first transfusion.  ITI was a predictor of long-term transfusion requirements in our cohort, with patients from ITI group 1 having lower transfusion requirements over time than that of group 2. The cumulative transfusion curves of the two groups did not converge.  High ITI was associated with poor overall survival (OS), with the OS of group 2 being lower than that of group 1. This association was robust, and remained significant when analysis was limited to patients with lower-risk MDS (IPSS Low/Int-1) or to those who were transfusion dependent according to IWG criteria (at least one RBC transfusion every 8 weeks over a period of 4 months).  ITI was a negative predictor of OS independent of age, IPSS and progression to acute myeloid leukemia (AML).  Early transfusion requirements are indicative of the extent of bone marrow (BM) failure before transfusion-related iron overload is possible.  The underlying pathology of patients with high ITI may be more severe and these patients may benefit from early intervention with disease-modifying therapy.

Our analysis of CD34+ bone marrow cells of MDS patients (Chapter 3) demonstrated that:

 The normalized intracellular reactive oxygen species (nROS) in the CD34+ primitive hematopoietic cells from MDS patients with low blast count were significantly higher than patients with high blast count, which may reflect a reliance on glycolysis and enhanced ROS defense in high blast MDS.  CD34+ cells from patients with high bone marrow blast counts displayed narrower nROS distribution than the corresponding cell population in low blast patients, suggesting that loss of heterogeneity in ROS content accompanies the clonal evolution of MDS.

184

 Serum ferritin in high blast count patients correlated positively with nROS in their CD34+ cells, consistent with the idea that iron overload increases oxidative stress in primitive hematopoietic cells. This in turn supports the notion that iron overload may further impair hematopoiesis and accelerate AML progression by increasing oxidative stress in CD34+ cells.  Iron chelation therapy (ICT) decreased both serum ferritin and nROS level of CD34+ cells from one RAEB-2 patient, suggesting removal of excess iron reduces oxidative stress in primitive hematopoietic cells and may mitigate the negative consequences of iron overload in these cells.

Our analysis of a radiation-induced AML (RI-AML) animal model (chapter 4) demonstrated that:

 Parenteral iron dextran injection resulted in iron accumulation in mouse BM.  Low dose iron loading of up to 20 mg was well tolerated in mice and did not result in severe deterioration of hematopoiesis.  RI-AML developed in B6D2F1 mice with a latency of more than 70 weeks.  Iron loading accelerated leukemogenesis in irradiated B6D2F1 mice. Total iron burden of 7.5 mg was the most effective in promoting RI-AML among the tested iron doses, with 40% of the mice receiving this dose developing AML starting at 28 weeks post- irradiation.  The occurrence of RI-AML was inversely correlated with total iron burden among the tested iron doses. There was a progressive decrease in AML frequency and corresponding increase in latency for irradiated mice with iron burden from 7.5 (58%/25.4 weeks – LFS/earliest AML onset) to 15 (80%/52.7 weeks) to 30 mg (88%/67.1 weeks).  ICT decreased AML incidence in the 7.5 mg iron-loaded irradiated mice. Conversely, AML onset was earlier for the 30 mg iron-loaded irradiated mice that received ICT.  The risk of AML may be influenced by the pro-apoptotic and pro-mutagenic properties of iron at different doses. The resulting relationship between iron burden and RI-AML risk may resemble a biphasic dose-response curve, in which the risk of AML increases up to a peak iron dose and then declines at higher dose.

185

 Leukemogenesis in irradiated B6D2F1 mice involved progressive activation of Akt, NF- κB, Wnt, antioxidant defenses, and the DNA damage response in the irradiated BM cells (iBMCs), as well as inactivation of JNK, C/EBPδ, and PTEN.  Progression through this series of changes is accelerated by iron. Iron-loaded iBMCs at 5 months post-irradiation are similar to non-iron-loaded iBMCs at 7 months with changes to Akt, NF-κB, JNK, and DNA damage response. Iron loading also activated Foxo3a in the iBMCs.  The tumor suppressor PTEN was downregulated at 7 months in the iron-loaded iBMCs at 7 months post-irradiation.  ICT partly reduced DNA damage and transcription dysregulation in the chelated iron- loaded iBMCs, which may help to explain the lower AML incidence in the 7.5mg iron- loaded irradiated mice treated with ICT.  The effects of iron loading and ICT on RI-AML may be applicable to AML transformation in MDS.

Our overall goal was to investigate if the outcome of MDS, especially AML transformation, is affected by anemia and related issues, namely transfusion, secondary iron overload, and iron chelation. High ITI predicted poor OS in MDS, suggesting that the extent of BM failure influences the outcome of MDS even before the occurrence of secondary iron overload due to transfusion. BM failure may contribute directly to the outcome of MDS. On the other hand, BM failure may also lead to inferior MDS survival by increasing RBC transfusion requirement, leading to secondary iron overload and iron toxicity. Indeed, elevated serum ferritin levels is an indicator of inferior MDS survival (Malcovati, 2007), while adequate ICT was associated with markedly better OS in transfusion-dependent MDS patients (Delforge et al., 2014). We further observed a correlation between serum ferritin level and ROS production in primitive hematopoietic cells of high blast count MDS patients, suggesting that secondary iron overload increases oxidative stress in these cells. Our data support the possibility that iron toxicity leads to poor outcome in MDS by impairing primitive hematopoietic cells via oxidative stress.

We then examined the role of excess iron in promoting AML development. AML transformation is a major concern in MDS and occurs in one-third of all MDS patients (Shukron et al., 2012). Using a RI-AML animal model, we assessed the role of iron loading on leukemogenesis without 186 the interference of other MDS related symptoms, such as anemia and other cytopenias. As hypothesized, iron loading, especially at 7.5 mg, promoted RI-AML in irradiated B6D2F1 mice. We have also identified signaling pathways that are dysregulated in response to iron loading and may contribute to leukemogenesis. However, the relationship between the tested iron doses and AML risk resembles a biphasic curve instead of linear or sigmoidal, possibly due to the varying pro-apoptotic and pro-mutagenic properties of iron at different doses. This implies that ICT could be beneficial or detrimental to AML free survival (LFS) depending on the iron burden before and after ICT. We concluded that iron is a promoter of leukemogenesis and our findings are consistent with the notion that secondary iron overload may accelerate AML transformation in MDS. Although the exact nature of the interaction between iron and AML development in MDS requires further characterization, this study has provided useful insight into the mechanism by which iron might promote leukemogenesis in MDS.

5.2 Future directions

5.2.1 Anemia, transfusion, and the outcome of MDS

We have shown that ITI, and by extension the initial state of erythropoietic failure prior to transfusion and related iron overload, is a prognostic indicator of OS in MDS. Our findings must be verified with datasets from other centers. Analysis with a larger cohort will also allow the inclusion of additional parameters such as WHO classification and iron status. In addition, we will be able to assess LFS as the outcome of MDS, which was not examined in this study due to limited AML transformation in the cohort. We expect high ITI to be associated with inferior LFS, given that transfusion dependency is associated with AML transformation (Leitch and Vickars, 2009; Sanz et al., 2008). ITI may partially influence the iron status of MDS patients, and both may contribute to the outcome of MDS. The possibility that iron burden and AML risk may have a biphasic dose-dependent relationship will also need to be assessed. Although ICT has been shown to improve OS in low-risk MDS patients (Temraz et al., 2014), most of the studies were retrospective in nature and further confirmation will be necessary. Furthermore, other reported or proposed benefits of ICT require further characterization, such as hematological improvement (Gattermann et al., 2012a; Jensen et al., 1996), or remain controversial, such as delayed AML transformation (Steensma and Gattermann, 2013). Several studies have been

187 conducted to examine the role of transfusion dependency and related issues in MDS, but not all data are in agreement regarding the impact of secondary iron overload on MDS outcome (Leitch and Vickars, 2009). Our analysis will contribute to the international efforts toward understanding the roles of BM failure, transfusion, and iron status in the outcome of MDS.

5.2.2 Iron-induced dysregulation in BMCs from MDS patients

Iron status of BMCs, especially primitive hematopoietic cells, requires further characterization. The level of labile iron pool (LIP) in cells can be measured by flow cytometry using calcein-AM and high-affinity iron-chelator (Prus and Fibach, 2008). In addition, BM iron accumulation has been recently demonstrated in thalassemia major patients using quantitative T2*MRI (Meloni et al., 2014). Oxidative stress in the hematopoietic tissues and cells should also be assessed in conjunction with iron status. Although the production of overall intracellular ROS (iROS) can be measured using 2',7'-dichlorodihydrofluorescein by flow cytometry (Chen et al., 2010), the use of other probes may be necessary to determine the level of specific type of ROS. MitoSOX Red detects the production of superoxide mitochondrial respiration (Mukhopadhyay et al., 2007), while aminophenyl fluorescein detects highly reactive oxygen species including hydroxyl radicals but not superoxide (Gomes et al., 2005). The use of different probes allows differential detection of iROS production from mitochondria-dependent generation of superoxide and iron- induced generation of hydroxyl radicals. Other indicators of oxidative stress should also be examined, including glutathione by ThiolTracker Violet stain (Mandavilli and Janes, 2010), and oxidative damage of DNA, lipid, and protein.

Our findings from the iron-loaded irradiated mice revealed changes in the BMCs that may contribute to radiation-induced leukemogenesis, including elevated DNA damage response, and earlier dysregulation of gene transcription and signaling pathways. We conjecture that some of these alterations may occur in both RI-AML and AML transformation in MDS, even though their pathogenesis is not identical. BMCs can be examined from MDS patients who are not transfused, heavily transfused, or receiving ICT. Freshly obtained bone marrow aspirates are suitable for the determination of surface and intracellular antigens by flow cytometry, which allows the assessment of the activation status of crucial signaling proteins, such as Akt, Foxo3a, and β- catenin, in individual cells from different cell types. Conversely, analysis of gene expression can

188 be conducted using both fresh and cryopreserved specimens. The Wells’ laboratory hosts a tissue bank consisted of cryopreserved BM specimens from MDS patients, which allow instant access to a substantial quantity of clinical material. The expression of genes related to several signaling pathways that were altered in the iron-loaded irradiated BMCs should be assessed, including Akt, Wnt, JNK, NF-κB, antioxidant defenses and DNA damage response. The expression of tumor suppressors, such as C/EBPδ, and PTEN, should also be determined. Alternatively, transcriptome analysis by RNA-Seq using next-generation sequencing will permit unbiased and comprehensive examination of multiple signaling pathways.

5.2.3 The effects of iron loading on leukemogenesis

Although we have shown the ability of excess iron to promote leukemogenesis, further elucidation is required for several aspects of iron loading in the development of RI-AML. In our early analysis, we observed altered gene expression in the BMCs from the irradiated mice upon iron loading or ICT. However, the gene expression profile was created based on total BMCs, and it is necessary to examine the gene expression profile of different hematopoietic cell types. Assessment of primitive hematopoietic cells is of particular importance, since these cells are proposed to give rise to leukemic stem cells (Hope et al., 2004), and hematopoiesis may be impaired when they are damaged. Primitive hematopoietic cells can be identified and isolated based on the expression of various surface antigens. In addition, the gene expression profile of these cells should be monitored at additional timepoints. Assessments shortly after iron injection may reveal acute cellular response that could become dampened by several months after iron loading, such as the compensation of TNFα-induced apoptosis by JNK inactivation (Varfolomeev and Ashkenazi, 2004). Also, assessments before AML onset may reveal gene expression changes that are involved in the transition from pre-leukemia to overt AML. In addition, the BM microenvironment has been shown to contribute to leukemogenesis and leukemia progression via growth factors (e.g. vascular endothelial growth factor), cytokines (e.g. C-X-C motif chemokine 12, colony stimulating factor 1) and interaction with BM stromal cells (Ayala et al., 2009). Recent studies have also demonstrated that iron loading can impair and modify the BM microenvironment (Okabe et al., 2014; Zhang et al., 2015). Therefore, further assessment and manipulation of these factors will be necessary to determine the interaction between iron and the BM microenvironment in the context of RI-AML. 189

5.2.4 The effects of iron loading on RI-AML in different mouse strains

Our observations suggest that B6D2F1 mice are sensitive to RI-AML and that iron loading accelerates leukemogenesis. Interestingly, C57BL/6 mice, the maternal strain of B6D2F1 mice, are resistant to both RI-AML (Boulton et al., 2001; Darakhshan et al., 2006) and iron loading (Boulton et al., 2001; Clothier et al., 2000; Fleming et al., 2001) when given as single agents. It has recently been shown that excess iron induces injury in irradiated BM (Chai et al., 2013; Okabe et al., 2014), but the combined effects of iron and irradiation on leukemogenesis in the C57BL/6 strain are yet to be determined. Also, long-lived inflammatory signaling triggered by TNFα and FasL promotes apoptosis in the BM of irradiated C57BL/6 mice (Lorimore et al., 2011). Conversely, the same inflammatory signaling promotes chromosomal instability in the BM of irradiated CBA/Ca mice (Lorimore et al., 2011). We speculate that iron loading in an already pro-apoptotic BM microenvironment of irradiated C57BL/6 mice may trigger further apoptosis of hematopoietic stem cells (HSCs), while the same condition may promote these cells to undergo leukemogenesis instead of apoptosis in CBA/Ca mice and other RI-AML sensitive strains such as DBA2. This notion can be verified by comparing the rate and latency of RI-AML between B6D2F1, C57BL/6 and DBA mice in the context of iron loading and ICT. Further analysis will also be necessary to elucidate the genetic and cell signaling differences among these strains that contribute to different RI-AML sensitivity in response to iron loading and ICT.

5.2.5 Secondary iron overload in MDS animal models

Iron overload is associated with inferior AML-free survival in MDS (Sanz et al., 2008) and our data show that excess iron accelerates RI-AML in mice. These observations are consistent with the notion that iron is a promoter of leukemogenesis. Nevertheless, the pathogenesis of AML progression in MDS is not identical to that of RI-AML, and further evaluation of iron as a promoter of AML transformation is necessary in the context of MDS. A number of mouse models have been developed to bear genetic and phenotypic resemblance with some aspects of MDS in humans (Beurlet et al., 2013). Forced expression of EVI1 in the BM of mice leads to epigenetic dysregulation and gives rise to an MDS phenotype that resemble RCMD with pancytopenia, a pro-apoptotic profile, and dyserythropoiesis (Laricchia-Robbio et al., 2006). Features of 5q- syndrome have been demonstrated in knockout mouse models of genes that are

190 commonly deleted in 5q- MDS patients, including miR145/miR146a (Starczynowski et al., 2010), SPARC (Lehmann et al., 2007), NPM1+/- (Grisendi et al., 2005; Sportoletti et al., 2008), and APC +/- (Wang et al., 2010). Iron loading and ICT in these mouse models may allow the assessment of iron toxicity in low-risk MDS. We expect excess iron to further impair hematopoiesis in these models; induction of leukemic transformation may also be possible. To mimic high-risk MDS, transplantation of BMCs with a truncated AML1 mutant has been shown to induce pancytopenia with erythroid dysplasia in conjunction with EVI1, which then progresses to a RAEB-2-like state and eventual AML transformation (Watanabe-Okochi et al., 2008). In addition, transgenic mice expressing mutated NRAS and BCL-2 acquire RAEB-1-like disease or AML with MDS dysplastic features, depending on the promoters that drive the transcription of these genes (Omidvar et al., 2007). Furthermore, a fusion gene that involves NUP98 and HOXD13 causes progressive MDS and AML in mice (Lin et al., 2005). Hematopoietic precursors that harbor the NUP98-HOXD13 fusion gene fail to differentiate into mature cells or undergo apoptosis when induced to differentiate (Choi et al., 2008). Increased oxidative stress and mutation frequency are also observed in the BM nucleated cells in the transgenic mice (Chung et al., 2014). Iron loading and ICT in these mice will allow us to evaluate the relevance of iron toxicity in more severe MDS with shortened life expectancy.

To summarize, our data on ITI demonstrate that long-term transfusion requirement and subsequent iron overload can be predicted by the extent of bone marrow failure before transfusion-related iron overload is possible. On the other hand, we showed that leukemogenesis can be promoted by iron overload and subsequent accumulation of ROS in the bone marrow. Therefore, it is reasonable to speculate that the risk of AML transformation in MDS may be influenced by the initiate degree of bone marrow failure, and then aggravated by iron overload as a result of chronic RBC transfusion. The observation that ICT can decrease AML risk in our radiation-induced AML animal model support the notion that ICT can delay or mitigate AML progression for iron-overloaded MDS patients. However, early intervention to avoid the development of extreme iron overload by ICT may be necessary to minimize the potential risk of iron-related leukemogenesis. Moreover, the iron/ICT-induced changes in gene expression and cell signaling network reported herein present testable markers for future studies and may possibly be adopted for clinical use in MDS.

191

References

Abbaspour, N., Hurrell, R., and Kelishadi, R. (2014). Review on iron and its importance for human health. Journal of research in medical sciences : the official journal of Isfahan University of Medical Sciences 19, 164‐174. Abdel‐Wahab, O., Adli, M., LaFave, L.M., Gao, J., Hricik, T., Shih, A.H., Pandey, S., Patel, J.P., Chung, Y.R., Koche, R., et al. (2012). ASXL1 mutations promote myeloid transformation through loss of PRC2‐mediated gene repression. Cancer cell 22, 180‐193. Afable, M.G., 2nd, Tiu, R.V., and Maciejewski, J.P. (2011). Clonal evolution in aplastic anemia. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program 2011, 90‐95. Afanas'ev, I. (2014). New nucleophilic mechanisms of ros‐dependent epigenetic modifications: comparison of aging and cancer. Aging and disease 5, 52‐62. Agarwal, N., Chatterjee, K., Sen, A., and Kumar, P. (2014). Prevalence of platelet reactive antibodies in patient's refractory to platelet transfusions. Asian journal of transfusion science 8, 126‐127. Aisen, P., Enns, C., and Wessling‐Resnick, M. (2001). Chemistry and biology of eukaryotic iron metabolism. The international journal of biochemistry & cell biology 33, 940‐959. Alessandrino, E.P., Amadori, S., Barosi, G., Cazzola, M., Grossi, A., Liberato, L.N., Locatelli, F., Marchetti, M., Morra, E., Rebulla, P., et al. (2002). Evidence‐ and consensus‐based practice guidelines for the therapy of primary myelodysplastic syndromes. A statement from the Italian Society of Hematology. Haematologica 87, 1286‐1306. Alessandrino, E.P., Della Porta, M.G., Bacigalupo, A., Van Lint, M.T., Falda, M., Onida, F., Bernardi, M., Iori, A.P., Rambaldi, A., Cerretti, R., et al. (2008). WHO classification and WPSS predict posttransplantation outcome in patients with myelodysplastic syndrome: a study from the Gruppo Italiano Trapianto di Midollo Osseo (GITMO). Blood 112, 895‐902. Alter, B.P. (2007). Diagnosis, genetics, and management of inherited bone marrow failure syndromes. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program, 29‐39. Altman, J.K., Sassano, A., and Platanias, L.C. (2011). Targeting mTOR for the treatment of AML. New agents and new directions. Oncotarget 2, 510‐517. Ameyar, M., Wisniewska, M., and Weitzman, J.B. (2003). A role for AP‐1 in apoptosis: the case for and against. Biochimie 85, 747‐752. Andrews, N.C., and Schmidt, P.J. (2007). Iron homeostasis. Annual review of physiology 69, 69‐85. Andziak, B., and Buffenstein, R. (2006). Disparate patterns of age‐related changes in lipid peroxidation in long‐lived naked mole‐rats and shorter‐lived mice. Aging cell 5, 525‐532. Arinkin, M.I. (1929). Die intravitale Untersuchungsmethdik des Knochenmarks. Folia Haematol (Leipzig) 38, 233‐ 240. Armand, P., Kim, H.T., Cutler, C.S., Ho, V.T., Koreth, J., Alyea, E.P., Soiffer, R.J., and Antin, J.H. (2007). Prognostic impact of elevated pretransplantation serum ferritin in patients undergoing myeloablative stem cell transplantation. Blood 109, 4586‐4588. Armand, P., Sainvil, M.M., Kim, H.T., Rhodes, J., Cutler, C., Ho, V.T., Koreth, J., Alyea, E.P., Neufeld, E.J., Kwong, R.Y., et al. (2013). Pre‐transplantation iron chelation in patients with MDS or acute leukemia and iron overload undergoing myeloablative allo‐SCT. Bone marrow transplantation 48, 146‐147. Arrizabalaga, B., del Canizo, C., and Remacha, A. (2008). Guía clínica de quelación del paciente con síndrome mielodisplásico. Haematologica 93, 3‐10. Ascenzi, P., Bocedi, A., Visca, P., Altruda, F., Tolosano, E., Beringhelli, T., and Fasano, M. (2005). Hemoglobin and heme scavenging. IUBMB life 57, 749‐759. Asemissen, A.M., and Giagounidis, A. (2011). If it ain't broke, don't fix it! Haematologica 96, e44.

192

Aslinia, F., Mazza, J.J., and Yale, S.H. (2006). Megaloblastic anemia and other causes of macrocytosis. Clin Med Res 4, 236‐241. Assi, T.B., and Baz, E. (2014). Current applications of therapeutic phlebotomy. Blood Transfus 12 Suppl 1, s75‐83. Auerbach, M., and Ballard, H. (2010). Clinical use of intravenous iron: administration, efficacy, and safety. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program 2010, 338‐347. Avery, S.V. (2011). Molecular targets of oxidative stress. The Biochemical journal 434, 201‐210. Ayala, F., Dewar, R., Kieran, M., and Kalluri, R. (2009). Contribution of bone microenvironment to leukemogenesis and leukemia progression. Leukemia 23, 2233‐2241. Azpurua, J., Ke, Z., Chen, I.X., Zhang, Q., Ermolenko, D.N., Zhang, Z.D., Gorbunova, V., and Seluanov, A. (2013). Naked mole‐rat has increased translational fidelity compared with the mouse, as well as a unique 28S ribosomal RNA cleavage. Proceedings of the National Academy of Sciences of the United States of America 110, 17350‐17355. Bae, Y.S., Oh, H., Rhee, S.G., and Yoo, Y.D. (2011). Regulation of reactive oxygen species generation in cell signaling. Molecules and cells 32, 491‐509. Balducci, L. (2006). Transfusion independence in patients with myelodysplastic syndromes: impact on outcomes and quality of life. Cancer 106, 2087‐2094. Balducci, L. (2010). Anemia, fatigue and aging. Transfus Clin Biol 17, 375‐381. Bannister, A.J., and Kouzarides, T. (2011). Regulation of chromatin by histone modifications. Cell research 21, 381‐ 395. Barker, J.N., Krepski, T.P., DeFor, T.E., Davies, S.M., Wagner, J.E., and Weisdorf, D.J. (2002). Searching for unrelated donor hematopoietic stem cells: availability and speed of umbilical cord blood versus bone marrow. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 8, 257‐ 260. Bartenstein, M., and Deeg, H.J. (2010). Hematopoietic stem cell transplantation for MDS. Hematology/oncology clinics of North America 24, 407‐422. Bartfay, W.J., and Bartfay, E. (2000). Iron‐overload cardiomyopathy: evidence for a free radical‐‐mediated mechanism of injury and dysfunction in a murine model. Biol Res Nurs 2, 49‐59. Bartfay, W.J., Dawood, F., Wen, W.H., Lehotay, D.C., Hou, D., Bartfay, E., Luo, X., Backx, P.H., and Liu, P.P. (1999). Cardiac function and cytotoxic aldehyde production in a murine model of chronic iron‐overload. Cardiovasc Res 43, 892‐900. Bartfay, W.J., Hou, D., Lehotay, D.C., Luo, X., Bartfay, E., Backx, P.H., and Liu, P.P. (2000). Cytotoxic aldehyde generation in heart following acute iron‐loading. Journal of trace elements in medicine and biology : organ of the Society for Minerals and Trace Elements 14, 14‐20. Barton, J.C., and Edwards, C.Q. (2000). Hemochromatosis genetics, pathophysiology, diagnosis and treatment (Chapter 48). (Cambridge, UK ; New York, Cambridge University Press), pp. xvi, 600 p. Barzi, A., and Sekeres, M.A. (2010). Myelodysplastic syndromes: a practical approach to diagnosis and treatment. Cleveland Clinic journal of medicine 77, 37‐44. Basova, P., Pospisil, V., Savvulidi, F., Burda, P., Vargova, K., Stanek, L., Dluhosova, M., Kuzmova, E., Jonasova, A., Steidl, U., et al. (2014). Aggressive acute myeloid leukemia in PU.1/p53 double‐mutant mice. Oncogene 33, 4735‐ 4745. Bass, D.A., Parce, J.W., Dechatelet, L.R., Szejda, P., Seeds, M.C., and Thomas, M. (1983). Flow cytometric studies of oxidative product formation by neutrophils: a graded response to membrane stimulation. J Immunol 130, 1910‐ 1917. Batlle, E., Henderson, J.T., Beghtel, H., van den Born, M.M., Sancho, E., Huls, G., Meeldijk, J., Robertson, J., van de Wetering, M., Pawson, T., et al. (2002). Beta‐catenin and TCF mediate cell positioning in the intestinal epithelium by controlling the expression of EphB/ephrinB. Cell 111, 251‐263.

193

Bauer, G. (2012). Tumor cell‐protective catalase as a novel target for rational therapeutic approaches based on specific intercellular ROS signaling. Anticancer research 32, 2599‐2624. Beaumont, C. (2010). Multiple regulatory mechanisms act in concert to control ferroportin expression and heme iron recycling by macrophages. Haematologica 95, 1233‐1236. Beguin, Y., Aapro, M., Ludwig, H., Mizzen, L., and Osterborg, A. (2014). Epidemiological and nonclinical studies investigating effects of iron in carcinogenesis‐‐a critical review. Critical reviews in oncology/hematology 89, 1‐15. Bejar, R. (2013). Prognostic models in myelodysplastic syndromes. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program 2013, 504‐510. Bejar, R. (2014). Clinical and genetic predictors of prognosis in myelodysplastic syndromes. Haematologica 99, 956‐ 964. Bejar, R., Stevenson, K.E., Caughey, B.A., Abdel‐Wahab, O., Steensma, D.P., Galili, N., Raza, A., Kantarjian, H., Levine, R.L., Neuberg, D., et al. (2012). Validation of a prognostic model and the impact of mutations in patients with lower‐risk myelodysplastic syndromes. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 30, 3376‐3382. Bendall, L.J., Kortlepel, K., and Gottlieb, D.J. (1993). Human acute myeloid leukemia cells bind to bone marrow stroma via a combination of beta‐1 and beta‐2 integrin mechanisms. Blood 82, 3125‐3132. Bennett, J., Cazzola, M., deCastro, C., Deeg, J., Delforge, M., Estey, E., Fenaux, P., Giagounidis, A., Germing, U., Goldberg, S., et al. (2008). Consensus statement on iron overload in myelodysplastic syndromes. American journal of hematology 83, 858‐861. Bennett, J.M., Catovsky, D., Daniel, M.T., Flandrin, G., Galton, D.A., Gralnick, H.R., and Sultan, C. (1982). Proposals for the classification of the myelodysplastic syndromes. British journal of haematology 51, 189‐199. Bergeron, R.J., Streiff, R.R., and Elliott, G.T. (1985). Influence of iron on in vivo proliferation and lethality of L1210 cells. The Journal of nutrition 115, 369‐374. Beurlet, S., Chomienne, C., and Padua, R.A. (2013). Engineering mouse models with myelodysplastic syndrome human candidate genes; how relevant are they? Haematologica 98, 10‐22. Bird, A. (2002). DNA methylation patterns and epigenetic memory. Genes & development 16, 6‐21. Block, M., Jacobson, L.O., and Bethard, W.F. (1953). Preleukemic acute human leukemia. J Am Med Assoc 152, 1018‐1028. Blumberg, N., Triulzi, D.J., and Heal, J.M. (1990). Transfusion‐induced immunomodulation and its clinical consequences. Transfus Med Rev 4, 24‐35. Boice, J.D., Jr., Engholm, G., Kleinerman, R.A., Blettner, M., Stovall, M., Lisco, H., Moloney, W.C., Austin, D.F., Bosch, A., Cookfair, D.L., et al. (1988). Radiation dose and second cancer risk in patients treated for cancer of the cervix. Radiat Res 116, 3‐55. Borgna‐Pignatti, C., Garani, M.C., Forni, G.L., Cappellini, M.D., Cassinerio, E., Fidone, C., Spadola, V., Maggio, A., Restivo Pantalone, G., Piga, A., et al. (2014). Hepatocellular carcinoma in thalassaemia: an update of the Italian Registry. British journal of haematology 167, 121‐126. Boulton, E., Cleary, H., Papworth, D., and Plumb, M. (2001). Susceptibility to radiation‐induced leukaemia/lymphoma is genetically separable from sensitivity to radiation‐induced genomic instability. International journal of radiation biology 77, 21‐29. Boultwood, J., Fidler, C., Strickson, A.J., Watkins, F., Gama, S., Kearney, L., Tosi, S., Kasprzyk, A., Cheng, J.F., Jaju, R.J., et al. (2002). Narrowing and genomic annotation of the commonly deleted region of the 5q‐ syndrome. Blood 99, 4638‐4641. Bowen, D., Culligan, D., Jowitt, S., Kelsey, S., Mufti, G., Oscier, D., Parker, J., and Group, U.M.G. (2003a). Guidelines for the diagnosis and therapy of adult myelodysplastic syndromes. British journal of haematology 120, 187‐200. Bowen, D., Wang, L., Frew, M., Kerr, R., and Groves, M. (2003b). Antioxidant enzyme expression in myelodysplastic and acute myeloid leukemia bone marrow: further evidence of a pathogenetic role for oxidative stress? Haematologica 88, 1070‐1072.

194

Brand, K.A., and Hermfisse, U. (1997). Aerobic glycolysis by proliferating cells: a protective strategy against reactive oxygen species. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 11, 388‐395. Bratton, M.R., Duong, B.N., Elliott, S., Weldon, C.B., Beckman, B.S., McLachlan, J.A., and Burow, M.E. (2010). Regulation of ERalpha‐mediated transcription of Bcl‐2 by PI3K‐AKT crosstalk: implications for breast cancer cell survival. International journal of oncology 37, 541‐550. Breccia, M., Finsinger, P., Loglisci, G., Federico, V., Santopietro, M., Colafigli, G., Petrucci, L., Salaroli, A., Serrao, A., Latagliata, R., et al. (2012). Deferasirox treatment for myelodysplastic syndromes: "real‐life" efficacy and safety in a single‐institution patient population. Annals of hematology 91, 1345‐1349. Breems, D.A., and Lowenberg, B. (2011). Acute myeloid leukemia with monosomal karyotype at the far end of the unfavorable prognostic spectrum. Haematologica 96, 491‐493. Breems, D.A., Van Putten, W.L., De Greef, G.E., Van Zelderen‐Bhola, S.L., Gerssen‐Schoorl, K.B., Mellink, C.H., Nieuwint, A., Jotterand, M., Hagemeijer, A., Beverloo, H.B., et al. (2008). Monosomal karyotype in acute myeloid leukemia: a better indicator of poor prognosis than a complex karyotype. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 26, 4791‐4797. Brissot, P., Ropert, M., Le Lan, C., and Loreal, O. (2012). Non‐transferrin bound iron: a key role in iron overload and iron toxicity. Biochimica et biophysica acta 1820, 403‐410. Brookes, M.J., Boult, J., Roberts, K., Cooper, B.T., Hotchin, N.A., Matthews, G., Iqbal, T., and Tselepis, C. (2008). A role for iron in Wnt signalling. Oncogene 27, 966‐975. Brown, K.E., Mathahs, M.M., Broadhurst, K.A., and Weydert, J. (2006). Chronic iron overload stimulates hepatocyte proliferation and cyclin D1 expression in rodent liver. Transl Res 148, 55‐62. Brunning, R.O., A. Germing, U. Le Beau, MM. Porwit, A. Baumann, I. Vardiman, JW. Hellstrom‐Linderg, E., Swerdlow, S.H., International Agency for Research on Cancer., and World Health Organization. (2008). WHO classification of tumours of haematopoietic and lymphoid tissues (Myelodysplastic syndromes/neoplasms, overview), 4th edn (Lyon, France: International Agency for Research on Cancer). Buchanan, F.G., and DuBois, R.N. (2006). Connecting COX‐2 and Wnt in cancer. Cancer cell 9, 6‐8. Burda, P., Laslo, P., and Stopka, T. (2010). The role of PU.1 and GATA‐1 transcription factors during normal and leukemogenic hematopoiesis. Leukemia 24, 1249‐1257. Busca, A., Falda, M., Manzini, P., D'Antico, S., Valfre, A., Locatelli, F., Calabrese, R., Chiappella, A., D'Ardia, S., Longo, F., et al. (2010). Iron overload in patients receiving allogeneic hematopoietic stem cell transplantation: quantification of iron burden by a superconducting quantum interference device (SQUID) and therapeutic effectiveness of phlebotomy. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 16, 115‐122. Bystrom, L.M., Guzman, M.L., and Rivella, S. (2014). Iron and reactive oxygen species: friends or foes of cancer cells? Antioxidants & redox signaling 20, 1917‐1924. Cadet, J., Ravanat, J.L., TavernaPorro, M., Menoni, H., and Angelov, D. (2012). Oxidatively generated complex DNA damage: tandem and clustered lesions. Cancer letters 327, 5‐15. Cao, C., Thomas, C.E., Insogna, K.L., and O'Brien, K.O. (2014). Duodenal absorption and tissue utilization of dietary heme and nonheme iron differ in rats. The Journal of nutrition 144, 1710‐1717. Catenacci, D.V., and Schiller, G.J. (2005). Myelodysplasic syndromes: a comprehensive review. Blood reviews 19, 301‐319. Cazzola, M. (2008). Myelodysplastic syndrome with isolated 5q deletion (5q‐ syndrome). A clonal stem cell disorder characterized by defective ribosome biogenesis. Haematologica 93, 967‐972. Cazzola, M., Della Porta, M.G., and Malcovati, L. (2008). Clinical relevance of anemia and transfusion iron overload in myelodysplastic syndromes. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program, 166‐175. Cazzola, M., Della Porta, M.G., and Malcovati, L. (2013a). The genetic basis of myelodysplasia and its clinical relevance. Blood 122, 4021‐4034. 195

Cazzola, M., and Malcovati, L. (2005). Myelodysplastic syndromes‐‐coping with ineffective hematopoiesis. The New England journal of medicine 352, 536‐538. Cazzola, M., and Malcovati, L. (2010). Prognostic classification and risk assessment in myelodysplastic syndromes. Hematology/oncology clinics of North America 24, 459‐468. Cazzola, M., Rossi, M., Malcovati, L., and Associazione Italiana per la Ricerca sul Cancro Gruppo Italiano Malattie, M. (2013b). Biologic and clinical significance of somatic mutations of SF3B1 in myeloid and lymphoid neoplasms. Blood 121, 260‐269. Cencioni, C., Spallotta, F., Martelli, F., Valente, S., Mai, A., Zeiher, A.M., and Gaetano, C. (2013). Oxidative stress and epigenetic regulation in ageing and age‐related diseases. International journal of molecular sciences 14, 17643‐17663. Cermak, J., Kacirkova, P., Mikulenkova, D., and Michalova, K. (2009). Impact of transfusion dependency on survival in patients with early myelodysplastic syndrome without excess of blasts. Leukemia research 33, 1469‐1474. Chai, X., Zhao, M., Li, D., Zhang, Y., Lu, W., Cao, X., Meng, J., You, Q., and Meng, A. (2014). [Effects and mechanism of iron overload on hematopoiesis in mice with bone marrow injury]. Zhonghua Xue Ye Xue Za Zhi 35, 1000‐1004. Chai, X., Zhao, M.F., Li, D.G., Meng, J.X., Lu, W.Y., Mu, J., and Meng, A.M. (2013). [Establishment of an mouse model of iron‐overload and its impact on bone marrow hematopoiesis]. Zhongguo Yi Xue Ke Xue Yuan Xue Bao 35, 547‐552. Challen, G.A., Sun, D., Jeong, M., Luo, M., Jelinek, J., Berg, J.S., Bock, C., Vasanthakumar, A., Gu, H., Xi, Y., et al. (2011). Dnmt3a is essential for hematopoietic stem cell differentiation. Nature genetics 44, 23‐31. Chambers, S.M., Shaw, C.A., Gatza, C., Fisk, C.J., Donehower, L.A., and Goodell, M.A. (2007). Aging hematopoietic stem cells decline in function and exhibit epigenetic dysregulation. PLoS biology 5, e201. Chan, C.H., Jo, U., Kohrman, A., Rezaeian, A.H., Chou, P.C., Logothetis, C., and Lin, H.K. (2014). Posttranslational regulation of Akt in human cancer. Cell Biosci 4, 59. Chapman, J.F., KForman, K., Kelsey, P., Knowles, S.M., Murphy, M.F., Williamson, L.M., and Wood, J.K. (1998). Guidelines on the clinical use of leucocyte‐depleted blood components. British Committee for Standards in Haematology, Blood Transfusion Task Force. Transfusion medicine 8, 59‐71. Chatterjee, I.B., Majumder, A.K., Nandi, B.K., and Subramanian, N. (1975). Synthesis and some major functions of vitamin C in animals. Annals of the New York Academy of Sciences 258, 24‐47. Chen, M., and Manley, J.L. (2009). Mechanisms of alternative splicing regulation: insights from molecular and genomics approaches. Nature reviews. Molecular cell biology 10, 741‐754. Chen, X., Zhong, Z., Xu, Z., Chen, L., and Wang, Y. (2010). 2',7'‐Dichlorodihydrofluorescein as a fluorescent probe for reactive oxygen species measurement: Forty years of application and controversy. Free Radic Res 44, 587‐604. Cheng, C.Y., Hsieh, H.L., Hsiao, L.D., and Yang, C.M. (2012). PI3‐K/Akt/JNK/NF‐kappaB is essential for MMP‐9 expression and outgrowth in human limbal epithelial cells on intact amniotic membrane. Stem Cell Res 9, 9‐23. Cheson, B.D., Bennett, J.M., Kantarjian, H., Pinto, A., Schiffer, C.A., Nimer, S.D., Lowenberg, B., Beran, M., de Witte, T.M., Stone, R.M., et al. (2000). Report of an international working group to standardize response criteria for myelodysplastic syndromes. Blood 96, 3671‐3674. Cheson, B.D., Greenberg, P.L., Bennett, J.M., Lowenberg, B., Wijermans, P.W., Nimer, S.D., Pinto, A., Beran, M., de Witte, T.M., Stone, R.M., et al. (2006). Clinical application and proposal for modification of the International Working Group (IWG) response criteria in myelodysplasia. Blood 108, 419‐425. Chevallier, P. (1942‐1943). Sur la terminologie des leucoses et les affections‐frontieres: les odoleucoses. Sangre (Brac) 15, 587‐593. Chiu, D.T., Stults, F.H., and Tappel, A.L. (1976). Purification and properties of rat lung soluble glutathione peroxidase. Biochimica et biophysica acta 445, 558‐566. Choi, C.W., Chung, Y.J., Slape, C., and Aplan, P.D. (2008). Impaired differentiation and apoptosis of hematopoietic precursors in a mouse model of myelodysplastic syndrome. Haematologica 93, 1394‐1397.

196

Chung, Y.J., Robert, C., Gough, S.M., Rassool, F.V., and Aplan, P.D. (2014). Oxidative stress leads to increased mutation frequency in a murine model of myelodysplastic syndrome. Leukemia research 38, 95‐102. Cilloni, D., Messa, E., and Biale, L. (2011). High rate of erythroid responseduring iron chelation therapy in a cohort of 105 patients affectedby hematologic malignancies with transfusional iron overload: anItalian multicenter retrospective study. ASH Annu Meet Abstr 118, 611. Clarke, G., and Chargé, S. (2007). Canadian Blood Services CLinical Guide to Transfusion. Clothier, B., Robinson, S., Akhtar, R.A., Francis, J.E., Peters, T.J., Raja, K., and Smith, A.G. (2000). Genetic variation of basal iron status, ferritin and iron regulatory protein in mice: potential for modulation of oxidative stress. Biochem Pharmacol 59, 115‐122. Cohen, A.R., Galanello, R., Piga, A., Dipalma, A., Vullo, C., and Tricta, F. (2000). Safety profile of the oral iron chelator deferiprone: a multicentre study. British journal of haematology 108, 305‐312. Coombs, G.S., Schmitt, A.A., Canning, C.A., Alok, A., Low, I.C., Banerjee, N., Kaur, S., Utomo, V., Jones, C.M., Pervaiz, S., et al. (2012). Modulation of Wnt/beta‐catenin signaling and proliferation by a ferrous iron chelator with therapeutic efficacy in genetically engineered mouse models of cancer. Oncogene 31, 213‐225. Corcoran, A., and Cotter, T.G. (2013). Redox regulation of protein kinases. The FEBS journal 280, 1944‐1965. Core, A.B., Canali, S., and Babitt, J.L. (2014). Hemojuvelin and bone morphogenetic protein (BMP) signaling in iron homeostasis. Frontiers in pharmacology 5, 104. Cortelezzi, A., Cattaneo, C., Cristiani, S., Duca, L., Sarina, B., Deliliers, G.L., Fiorelli, G., and Cappellini, M.D. (2000). Non‐transferrin‐bound iron in myelodysplastic syndromes: a marker of ineffective erythropoiesis? The hematology journal : the official journal of the European Haematology Association / EHA 1, 153‐158. Cortes, I., Sanchez‐Ruiz, J., Zuluaga, S., Calvanese, V., Marques, M., Hernandez, C., Rivera, T., Kremer, L., Gonzalez‐ Garcia, A., and Carrera, A.C. (2012). p85beta phosphoinositide 3‐kinase subunit regulates tumor progression. Proceedings of the National Academy of Sciences of the United States of America 109, 11318‐11323. Courselaud, B., Troadec, M.B., Fruchon, S., Ilyin, G., Borot, N., Leroyer, P., Coppin, H., Brissot, P., Roth, M.P., and Loreal, O. (2004). Strain and gender modulate hepatic hepcidin 1 and 2 mRNA expression in mice. Blood cells, molecules & diseases 32, 283‐289. Courtois‐Cox, S., Genther Williams, S.M., Reczek, E.E., Johnson, B.W., McGillicuddy, L.T., Johannessen, C.M., Hollstein, P.E., MacCollin, M., and Cichowski, K. (2006). A negative feedback signaling network underlies oncogene‐ induced senescence. Cancer cell 10, 459‐472. Crichton, R.R. (2009a). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 1). (Chichester, UK, John Wiley & Sons), pp. 1‐15. Crichton, R.R. (2009b). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 2). (Chichester, UK, John Wiley & Sons), pp. 17‐56. Crichton, R.R. (2009c). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 3). (Chichester, UK, John Wiley & Sons), pp. 59‐94. Crichton, R.R. (2009d). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 5). (Chichester, UK, John Wiley & Sons), pp. 141‐173. Crichton, R.R. (2009e). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 6). (Chichester, UK, John Wiley & Sons), pp. 183‐215. Crichton, R.R. (2009f). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 7). (Chichester, UK, John Wiley & Sons), pp. 223‐261. Crichton, R.R. (2009g). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 8). (Chichester, UK, John Wiley & Sons), pp. 271‐294. Crichton, R.R. (2009h). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 9). (Chichester, UK, John Wiley & Sons), pp. 299‐325. Crichton, R.R. (2009i). Iron metabolism from molecular mechanisms to clinical consequences (Chapter 10). (Chichester, UK, John Wiley & Sons), pp. 335‐365.

197

Cully, M., You, H., Levine, A.J., and Mak, T.W. (2006). Beyond PTEN mutations: the PI3K pathway as an integrator of multiple inputs during tumorigenesis. Nat Rev Cancer 6, 184‐192. Cunningham, I., MacCallum, S.J., Nicholls, M.D., Byth, K., Hewson, J.W., Arnold, B., Motum, P.I., Mulligan, S.P., and Crane, G.G. (1995). The myelodysplastic syndromes: an analysis of prognostic factors in 226 cases from a single institution. British journal of haematology 90, 602‐606. Cutler, C. (2010). Patient selection for transplantation in the myelodysplastic syndromes. Hematology/oncology clinics of North America 24, 469‐476. Cutler, C.S., Lee, S.J., Greenberg, P., Deeg, H.J., Perez, W.S., Anasetti, C., Bolwell, B.J., Cairo, M.S., Gale, R.P., Klein, J.P., et al. (2004). A decision analysis of allogeneic bone marrow transplantation for the myelodysplastic syndromes: delayed transplantation for low‐risk myelodysplasia is associated with improved outcome. Blood 104, 579‐585. D'Alessio, F., Hentze, M.W., and Muckenthaler, M.U. (2012). The hemochromatosis proteins HFE, TfR2, and HJV form a membrane‐associated protein complex for hepcidin regulation. Journal of hepatology 57, 1052‐1060. Dai, M.C., Zhong, Z.H., Sun, Y.H., Sun, Q.F., Wang, Y.T., Yang, G.Y., and Bian, L.G. (2013). Curcumin protects against iron induced neurotoxicity in primary cortical neurons by attenuating necroptosis. Neuroscience letters 536, 41‐46. Dameshek, W. (1951). Some speculations on the myeloproliferative syndromes. Blood 6, 372‐375. Dameshek, W. (1969). Foreword and a proposal for considering paroxysmal nocturnal hemoglobinuria (PNH) as a "candidate" myeloproliferative disorder. Blood 33, 263‐264. Dameshek, W., and Baldini, M. (1958). The Di Guglielmo syndrome. Blood 13, 192‐194. Dao, M.C., and Meydani, S.N. (2013). Iron biology, immunology, aging, and obesity: four fields connected by the small peptide hormone hepcidin. Advances in nutrition 4, 602‐617. Darakhshan, F., Badie, C., Moody, J., Coster, M., Finnon, R., Finnon, P., Edwards, A.A., Szluinska, M., Skidmore, C.J., Yoshida, K., et al. (2006). Evidence for complex multigenic inheritance of radiation AML susceptibility in mice revealed using a surrogate phenotypic assay. Carcinogenesis 27, 311‐318. Das, M., Chaudhuri, S., and Law, S. (2013). Unveiling the paradoxical nature of myelodysplastic syndromes (MDS): why hypercellular marrow strongly favors accelerated apoptosis. Biochemistry and cell biology = Biochimie et biologie cellulaire 91, 303‐308. Datan, E., Shirazian, A., Benjamin, S., Matassov, D., Tinari, A., Malorni, W., Lockshin, R.A., Garcia‐Sastre, A., and Zakeri, Z. (2014). mTOR/p70S6K signaling distinguishes routine, maintenance‐level autophagy from autophagic cell death during influenza A infection. Virology 452‐453, 175‐190. Davoudi, Z., Akbarzadeh, A., Rahmatiyamchi, M., Movassaghpour, A.A., Alipour, M., Nejati‐Koshki, K., Sadeghi, Z., Dariushnejad, H., and Zarghami, N. (2014). Molecular target therapy of AKT and NF‐kB signaling pathways and multidrug resistance by specific cell penetrating inhibitor peptides in HL‐60 cells. Asian Pac J Cancer Prev 15, 4353‐ 4358. Dayyani, F., Conley, A.P., Strom, S.S., Stevenson, W., Cortes, J.E., Borthakur, G., Faderl, S., O'Brien, S., Pierce, S., Kantarjian, H., et al. (2010). Cause of death in patients with lower‐risk myelodysplastic syndrome. Cancer 116, 2174‐2179. De Luca, A., Maiello, M.R., D'Alessio, A., Pergameno, M., and Normanno, N. (2012). The RAS/RAF/MEK/ERK and the PI3K/AKT signalling pathways: role in cancer pathogenesis and implications for therapeutic approaches. Expert opinion on therapeutic targets 16 Suppl 2, S17‐27. de Swart, L., Smith, A., Fenaux, P. (2011). Transfusion‐dependency isthe most important prognostic factor for survival in 1000 newlydiagnosed mds patients with low‐ and intermediate‐1 risk MDS inthe European LeukemiaNet MDS registry. ASH Annu Meet Abstr 118, 2775. Deeg, H.J., Storer, B., Slattery, J.T., Anasetti, C., Doney, K.C., Hansen, J.A., Kiem, H.P., Martin, P.J., Petersdorf, E., Radich, J.P., et al. (2002). Conditioning with targeted busulfan and cyclophosphamide for hemopoietic stem cell transplantation from related and unrelated donors in patients with myelodysplastic syndrome. Blood 100, 1201‐ 1207.

198

Dekkers, F., Bijwaard, H., Bouffler, S., Ellender, M., Huiskamp, R., Kowalczuk, C., Meijne, E., and Sutmuller, M. (2011). A two‐mutation model of radiation‐induced acute myeloid leukemia using historical mouse data. Radiat Environ Biophys 50, 37‐45. Delea, T.E., Hagiwara, M., and Phatak, P.D. (2009). Retrospective study of the association between transfusion frequency and potential complications of iron overload in patients with myelodysplastic syndrome and other acquired hematopoietic disorders. Current medical research and opinion 25, 139‐147. Delforge, M., Selleslag, D., Beguin, Y., Triffet, A., Mineur, P., Theunissen, K., Graux, C., Trullemans, F., Boulet, D., Van Eygen, K., et al. (2014). Adequate iron chelation therapy for at least six months improves survival in transfusion‐dependent patients with lower risk myelodysplastic syndromes. Leukemia research 38, 557‐563. Deng, T., and Karin, M. (1994). c‐Fos transcriptional activity stimulated by H‐Ras‐activated protein kinase distinct from JNK and ERK. Nature 371, 171‐175. Devasagayam, T.P., Tilak, J.C., Boloor, K.K., Sane, K.S., Ghaskadbi, S.S., and Lele, R.D. (2004). Free radicals and antioxidants in human health: current status and future prospects. J Assoc Physicians India 52, 794‐804. Dhanasekaran, D.N., and Reddy, E.P. (2008). JNK signaling in apoptosis. Oncogene 27, 6245‐6251. Di Majo, V., Coppola, M., Rebessi, S., Saran, A., Pazzaglia, S., Pariset, L., and Covelli, V. (1996). The influence of sex on life shortening and tumor induction in CBA/Cne mice exposed to X rays or fission neutrons. Radiat Res 146, 81‐ 87. Di Tucci, A.A., Matta, G., Deplano, S., Gabbas, A., Depau, C., Derudas, D., Caocci, G., Agus, A., and Angelucci, E. (2008). Myocardial iron overload assessment by T2* magnetic resonance imaging in adult transfusion dependent patients with acquired anemias. Haematologica 93, 1385‐1388. Ding, J., Ghali, O., Lencel, P., Broux, O., Chauveau, C., Devedjian, J.C., Hardouin, P., and Magne, D. (2009). TNF‐ alpha and IL‐1beta inhibit RUNX2 and collagen expression but increase alkaline phosphatase activity and mineralization in human mesenchymal stem cells. Life Sci 84, 499‐504. Disperati, P., Ichim, C.V., Tkachuk, D., Chun, K., Schuh, A.C., and Wells, R.A. (2006). Progression of myelodysplasia to acute lymphoblastic leukaemia: implications for disease biology. Leukemia research 30, 233‐239. Dixon, S.J., Lemberg, K.M., Lamprecht, M.R., Skouta, R., Zaitsev, E.M., Gleason, C.E., Patel, D.N., Bauer, A.J., Cantley, A.M., Yang, W.S., et al. (2012). Ferroptosis: an iron‐dependent form of nonapoptotic cell death. Cell 149, 1060‐1072. Dixon, S.J., and Stockwell, B.R. (2014). The role of iron and reactive oxygen species in cell death. Nature chemical biology 10, 9‐17. Donovan, N., Becker, E.B., Konishi, Y., and Bonni, A. (2002). JNK phosphorylation and activation of BAD couples the stress‐activated signaling pathway to the cell death machinery. The Journal of biological chemistry 277, 40944‐ 40949. Ebert, B.L., Pretz, J., Bosco, J., Chang, C.Y., Tamayo, P., Galili, N., Raza, A., Root, D.E., Attar, E., Ellis, S.R., et al. (2008). Identification of RPS14 as a 5q‐ syndrome gene by RNA interference screen. Nature 451, 335‐339. Ehrlich, P. (1879). Methodologische Beitrage Zur Physiologie and Pathologie der verschiedenen Formen Der Leukocyten. Z Klin Med 1, 553‐560. Eldor, A., and Rachmilewitz, E.A. (2002). The hypercoagulable state in thalassemia. Blood 99, 36‐43. Elford, H.L., Freese, M., Passamani, E., and Morris, H.P. (1970). Ribonucleotide reductase and cell proliferation. I. Variations of ribonucleotide reductase activity with tumor growth rate in a series of rat hepatomas. The Journal of biological chemistry 245, 5228‐5233. Elzik, M.E., Dirschl, D.R., and Dahners, L.E. (2006). Correlation of transfusion volume to change in hematocrit. American journal of hematology 81, 145‐146. Fargion, S., Valenti, L., and Fracanzani, A.L. (2010). Hemochromatosis gene (HFE) mutations and cancer risk: expanding the clinical manifestations of hereditary iron overload. Hepatology 51, 1119‐1121. Faurschou, A., and Gniadecki, R. (2008). TNF‐alpha stimulates Akt by a distinct aPKC‐dependent pathway in premalignant keratinocytes. Exp Dermatol 17, 992‐997.

199

Fenaux, P., and Ades, L. (2013). How we treat lower‐risk myelodysplastic syndromes. Blood 121, 4280‐4286. Fenaux, P., Giagounidis, A., Selleslag, D., Beyne‐Rauzy, O., Mufti, G., Mittelman, M., Muus, P., Te Boekhorst, P., Sanz, G., Del Canizo, C., et al. (2011). A randomized phase 3 study of lenalidomide versus placebo in RBC transfusion‐dependent patients with Low‐/Intermediate‐1‐risk myelodysplastic syndromes with del5q. Blood 118, 3765‐3776. Fenaux, P., Mufti, G.J., Hellstrom‐Lindberg, E., Santini, V., Finelli, C., Giagounidis, A., Schoch, R., Gattermann, N., Sanz, G., List, A., et al. (2009). Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher‐risk myelodysplastic syndromes: a randomised, open‐label, phase III study. The Lancet. Oncology 10, 223‐232. Fenaux, P., and Rose, C. (2009). Impact of iron overload in myelodysplastic syndromes. Blood reviews 23 Suppl 1, S15‐19. Fenton, H. (1894). The oxidation of tartaric acid in presence of iron. J. Chem. Soc. Trans. 10, 157‐158. Ferber, E.C., Peck, B., Delpuech, O., Bell, G.P., East, P., and Schulze, A. (2012). FOXO3a regulates reactive oxygen metabolism by inhibiting mitochondrial gene expression. Cell Death Differ 19, 968‐979. Fernandez, A., Huggins, I.J., Perna, L., Brafman, D., Lu, D., Yao, S., Gaasterland, T., Carson, D.A., and Willert, K. (2014). The WNT receptor FZD7 is required for maintenance of the pluripotent state in human embryonic stem cells. Proceedings of the National Academy of Sciences of the United States of America 111, 1409‐1414. Fleming, R.E., Holden, C.C., Tomatsu, S., Waheed, A., Brunt, E.M., Britton, R.S., Bacon, B.R., Roopenian, D.C., and Sly, W.S. (2001). Mouse strain differences determine severity of iron accumulation in Hfe knockout model of hereditary hemochromatosis. Proceedings of the National Academy of Sciences of the United States of America 98, 2707‐2711. Flint, J., Harding, R.M., Boyce, A.J., and Clegg, J.B. (1998). The population genetics of the haemoglobinopathies. Bailliere's clinical haematology 11, 1‐51. Foran, J.M., and Shammo, J.M. (2012). Clinical presentation, diagnosis, and prognosis of myelodysplastic syndromes. The American journal of medicine 125, S6‐13. Frelin, C., Imbert, V., Griessinger, E., Peyron, A.C., Rochet, N., Philip, P., Dageville, C., Sirvent, A., Hummelsberger, M., Berard, E., et al. (2005). Targeting NF‐kappaB activation via pharmacologic inhibition of IKK2‐induced apoptosis of human acute myeloid leukemia cells. Blood 105, 804‐811. Frenzel, T., Westphal‐Varghese, B., and Westphal, M. (2009). Role of storage time of red blood cells on microcirculation and tissue oxygenation in critically ill patients. Curr Opin Anaesthesiol 22, 275‐280. Frisch, S.M., and Screaton, R.A. (2001). Anoikis mechanisms. Current opinion in cell biology 13, 555‐562. Fu, Z., and Tindall, D.J. (2008). FOXOs, cancer and regulation of apoptosis. Oncogene 27, 2312‐2319. Fuchs, O. (2012). Important genes in the pathogenesis of 5q‐ syndrome and their connection with ribosomal stress and the innate immune system pathway. Leukemia research and treatment 2012, 179402. Fukai, T., and Ushio‐Fukai, M. (2011). Superoxide dismutases: role in redox signaling, vascular function, and diseases. Antioxidants & redox signaling 15, 1583‐1606. Fuqua, B.K., Vulpe, C.D., and Anderson, G.J. (2012). Intestinal iron absorption. Journal of trace elements in medicine and biology : organ of the Society for Minerals and Trace Elements 26, 115‐119. Gaballa, M.R., and Besa, E.C. (2014). Myelodysplastic syndromes with 5q deletion: pathophysiology and role of lenalidomide. Annals of hematology 93, 723‐733. Galanello, R., and Origa, R. (2010). Beta‐thalassemia. Orphanet journal of rare diseases 5, 11. Gangat, N., Patnaik, M.M., Begna, K., Kourelis, T., Knudson, R.A., Ketterling, R.P., Hodnefield, J.M., Hanson, C.A., Pardanani, A., and Tefferi, A. (2013). Evaluation of revised IPSS cytogenetic risk stratification and prognostic impact of monosomal karyotype in 783 patients with primary myelodysplastic syndromes. American journal of hematology 88, 690‐693. Ganz, T. (2012). Macrophages and systemic iron homeostasis. J Innate Immun 4, 446‐453.

200

Gao, J., Chen, J., Kramer, M., Tsukamoto, H., Zhang, A.S., and Enns, C.A. (2009a). Interaction of the hereditary hemochromatosis protein HFE with transferrin receptor 2 is required for transferrin‐induced hepcidin expression. Cell Metab 9, 217‐227. Gao, J., Gentzler, R.D., Timms, A.E., Horwitz, M.S., Frankfurt, O., Altman, J.K., and Peterson, L.C. (2014). Heritable GATA2 mutations associated with familial AML‐MDS: a case report and review of literature. Journal of hematology & oncology 7, 36. Gao, X., Campian, J.L., Qian, M., Sun, X.F., and Eaton, J.W. (2009b). Mitochondrial DNA damage in iron overload. The Journal of biological chemistry 284, 4767‐4775. Garcia‐Manero, G., and Fenaux, P. (2011). Hypomethylating agents and other novel strategies in myelodysplastic syndromes. J Clin Oncol 29, 516‐523. Garcia‐Manero, G., Shan, J., Faderl, S., Cortes, J., Ravandi, F., Borthakur, G., Wierda, W.G., Pierce, S., Estey, E., Liu, J., et al. (2008). A prognostic score for patients with lower risk myelodysplastic syndrome. Leukemia 22, 538‐543. Gattermann, N. (2008). Overview of guidelines on iron chelation therapy in patients with myelodysplastic syndromes and transfusional iron overload. International journal of hematology 88, 24‐29. Gattermann, N., Finelli, C., Della Porta, M., Fenaux, P., Stadler, M., Guerci‐Bresler, A., Schmid, M., Taylor, K., Vassilieff, D., Habr, D., et al. (2012a). Hematologic responses to deferasirox therapy in transfusion‐dependent patients with myelodysplastic syndromes. Haematologica 97, 1364‐1371. Gattermann, N., Finelli, C., Porta, M.D., Fenaux, P., Ganser, A., Guerci‐Bresler, A., Schmid, M., Taylor, K., Vassilieff, D., Habr, D., et al. (2010). Deferasirox in iron‐overloaded patients with transfusion‐dependent myelodysplastic syndromes: Results from the large 1‐year EPIC study. Leukemia research 34, 1143‐1150. Gattermann, N., Jarisch, A., Schlag, R., Blumenstengel, K., Goebeler, M., Groschek, M., Losem, C., Procaccianti, M., Junkes, A., Leismann, O., et al. (2012b). Deferasirox treatment of iron‐overloaded chelation‐naive and prechelated patients with myelodysplastic syndromes in medical practice: results from the observational studies eXtend and eXjange. European journal of haematology 88, 260‐268. Gayatri, R., and Chatterjee, S. (1991). Effects of chlorpromazine on growth and development of Dictyostelium discoideum. Microbios 68, 97‐107. Gelsi‐Boyer, V., Trouplin, V., Roquain, J., Adelaide, J., Carbuccia, N., Esterni, B., Finetti, P., Murati, A., Arnoulet, C., Zerazhi, H., et al. (2010). ASXL1 mutation is associated with poor prognosis and acute transformation in chronic myelomonocytic leukaemia. British journal of haematology 151, 365‐375. Ghoti, H., Amer, J., Winder, A., Rachmilewitz, E., and Fibach, E. (2007). Oxidative stress in red blood cells, platelets and polymorphonuclear leukocytes from patients with myelodysplastic syndrome. European journal of haematology 79, 463‐467. Ghoti, H., Fibach, E., Merkel, D., Perez‐Avraham, G., Grisariu, S., and Rachmilewitz, E.A. (2010). Changes in parameters of oxidative stress and free iron biomarkers during treatment with deferasirox in iron‐overloaded patients with myelodysplastic syndromes. Haematologica 95, 1433‐1434. Giagounidis, A., Fenaux, P., Mufti, G.J., Muus, P., Platzbecker, U., Sanz, G., Cripe, L., Von Lilienfeld‐Toal, M., and Wells, R.A. (2008). Practical recommendations on the use of lenalidomide in the management of myelodysplastic syndromes. Ann Hematol 87, 345‐352. Gilbert, H.S. (1970). A reappraisal of the "myeloproliferative disease" concept. The Mount Sinai journal of medicine, New York 37, 426‐435. Gokal, R., Weatherall, D.J., and Bunch, C. (1979). Iron induced increase in red cell size in haemodialysis patients. Q J Med 48, 393‐401. Goldberg, S.L., Chen, E., Corral, M., Guo, A., Mody‐Patel, N., Pecora, A.L., and Laouri, M. (2010). Incidence and clinical complications of myelodysplastic syndromes among United States Medicare beneficiaries. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 28, 2847‐2852. Gomes, A., Fernandes, E., and Lima, J.L. (2005). Fluorescence probes used for detection of reactive oxygen species. J Biochem Biophys Methods 65, 45‐80.

201

Greenberg, P., Cox, C., LeBeau, M.M., Fenaux, P., Morel, P., Sanz, G., Sanz, M., Vallespi, T., Hamblin, T., Oscier, D., et al. (1997). International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood 89, 2079‐ 2088. Greenberg, P.L., Attar, E., Bennett, J.M., Bloomfield, C.D., Borate, U., De Castro, C.M., Deeg, H.J., Frankfurt, O., Gaensler, K., Garcia‐Manero, G., et al. (2013). Myelodysplastic syndromes: clinical practice guidelines in oncology. Journal of the National Comprehensive Cancer Network : JNCCN 11, 838‐874. Greenberg, P.L., Tuechler, H., Schanz, J., Sanz, G., Garcia‐Manero, G., Sole, F., Bennett, J.M., Bowen, D., Fenaux, P., Dreyfus, F., et al. (2012). Revised international prognostic scoring system for myelodysplastic syndromes. Blood 120, 2454‐2465. Greenberger, J., Leif, J., Crawford, D., Anklesaria, P., English, D., Sakakeeny, M., Rubin, J., Pierce, J., Shadduck, R., and FitzGerald, T.J. (1992a). Humoral and cell surface interactions during gamma‐irradiation leukemogenesis in vitro. Exp Hematol 20, 92‐102. Greenberger, J.S., Sakakeeny, M.A., Leif, J., Anklesaria, P., Pierce, J.H., and Shadduck, R.K. (1992b). Expression of M‐ CSF and its receptor (C‐FMS) during factor‐independent cell line evolution from hematopoietic progenitor cells cocultivated with gamma irradiated marrow stromal cell lines. Leukemia 6, 626‐633. Greenstein, S., Ghias, K., Krett, N.L., and Rosen, S.T. (2002). Mechanisms of glucocorticoid‐mediated apoptosis in hematological malignancies. Clin Cancer Res 8, 1681‐1694. Grisendi, S., Bernardi, R., Rossi, M., Cheng, K., Khandker, L., Manova, K., and Pandolfi, P.P. (2005). Role of nucleophosmin in embryonic development and tumorigenesis. Nature 437, 147‐153. Guglielmo, G.D. (1926). Eritremie acute. Boll Soc Med Chir 1, 655‐673. Guralnik, J.M., Eisenstaedt, R.S., Ferrucci, L., Klein, H.G., and Woodman, R.C. (2004). Prevalence of anemia in persons 65 years and older in the United States: evidence for a high rate of unexplained anemia. Blood 104, 2263‐ 2268. Haase, D. (2008). Cytogenetic features in myelodysplastic syndromes. Annals of hematology 87, 515‐526. Haase, D., Germing, U., Schanz, J., Pfeilstocker, M., Nosslinger, T., Hildebrandt, B., Kundgen, A., Lubbert, M., Kunzmann, R., Giagounidis, A.A., et al. (2007). New insights into the prognostic impact of the karyotype in MDS and correlation with subtypes: evidence from a core dataset of 2124 patients. Blood 110, 4385‐4395. Haber, F., Weiss, J. (1934). The catalytic decomposition of hydrogen peroxide by iron salts. Proc. Roy. Soc. Ser. A 147, 332‐351. Hachiya, M., Osawa, Y., and Akashi, M. (2000). Role of TNFalpha in regulation of myeloperoxidase expression in irradiated HL60 promyelocytic cells. Biochimica et biophysica acta 1495, 237‐249. Hahn, C.N., Chong, C.E., Carmichael, C.L., Wilkins, E.J., Brautigan, P.J., Li, X.C., Babic, M., Lin, M., Carmagnac, A., Lee, Y.K., et al. (2011). Heritable GATA2 mutations associated with familial myelodysplastic syndrome and acute myeloid leukemia. Nature genetics 43, 1012‐1017. Hahn, P., Song, Y., Ying, G.S., He, X., Beard, J., and Dunaief, J.L. (2009). Age‐dependent and gender‐specific changes in mouse tissue iron by strain. Exp Gerontol 44, 594‐600. Hall, E.J. (2000). Radiation, the two‐edged sword: cancer risks at high and low doses. Cancer J 6, 343‐350. Halliwell, B., and Gutteridge, J.M. (1990). Role of free radicals and catalytic metal ions in human disease: an overview. Methods in enzymology 186, 1‐85. Hamilton‐Paterson, J.L. (1949). Pre‐leukaemic anaemia. Acta haematologica 2, 309‐316. Hann, H.W., Stahlhut, M.W., and Blumberg, B.S. (1988). Iron nutrition and tumor growth: decreased tumor growth in iron‐deficient mice. Cancer research 48, 4168‐4170. Haran‐Ghera, N., Krautghamer, R., Lapidot, T., Peled, A., Dominguez, M.G., and Stanley, E.R. (1997). Increased circulating colony‐stimulating factor‐1 (CSF‐1) in SJL/J mice with radiation‐induced acute myeloid leukemia (AML) is associated with autocrine regulation of AML cells by CSF‐1. Blood 89, 2537‐2545. Haran‐Ghera, N., Peled, A., Krautghamer, R., and Resnitzky, P. (1992). Initiation and promotion in radiation‐induced myeloid leukemia. Leukemia 6, 689‐695.

202

Hartmann, J., Braulke, F., Sinzig, U., Wulf, G., Maas, J.H., Konietschke, F., and Haase, D. (2013). Iron overload impairs proliferation of erythroid progenitors cells (BFU‐E) from patients with myelodysplastic syndromes. Leukemia research 37, 327‐332. Hay, N., and Sonenberg, N. (2004). Upstream and downstream of mTOR. Genes & development 18, 1926‐1945. Hayata, I., Seki, M., Yoshida, K., Hirashima, K., Sado, T., Yamagiwa, J., and Ishihara, T. (1983). Chromosomal aberrations observed in 52 mouse myeloid leukemias. Cancer research 43, 367‐373. Heath, J.L., Weiss, J.M., Lavau, C.P., and Wechsler, D.S. (2013). Iron deprivation in cancer‐‐potential therapeutic implications. Nutrients 5, 2836‐2859. Hellstrom‐Lindberg, E., Ahlgren, T., Beguin, Y., Carlsson, M., Carneskog, J., Dahl, I.M., Dybedal, I., Grimfors, G., Kanter‐Lewensohn, L., Linder, O., et al. (1998). Treatment of anemia in myelodysplastic syndromes with granulocyte colony‐stimulating factor plus erythropoietin: results from a randomized phase II study and long‐term follow‐up of 71 patients. Blood 92, 68‐75. Hellstrom‐Lindberg, E., Gulbrandsen, N., Lindberg, G., Ahlgren, T., Dahl, I.M., Dybedal, I., Grimfors, G., Hesse‐ Sundin, E., Hjorth, M., Kanter‐Lewensohn, L., et al. (2003). A validated decision model for treating the anaemia of myelodysplastic syndromes with erythropoietin + granulocyte colony‐stimulating factor: significant effects on quality of life. Br J Haematol 120, 1037‐1046. Herault, O., Hope, K.J., Deneault, E., Mayotte, N., Chagraoui, J., Wilhelm, B.T., Cellot, S., Sauvageau, M., Andrade‐ Navarro, M.A., Hebert, J., et al. (2012). A role for GPx3 in activity of normal and leukemia stem cells. The Journal of experimental medicine 209, 895‐901. Herst, P.M., Howman, R.A., Neeson, P.J., Berridge, M.V., and Ritchie, D.S. (2011). The level of glycolytic metabolism in acute myeloid leukemia blasts at diagnosis is prognostic for clinical outcome. Journal of leukocyte biology 89, 51‐ 55. Hess, D.R. (2004). Retrospective studies and chart reviews. Respiratory care 49, 1171‐1174. Hider, R.C. (2002). Nature of nontransferrin‐bound iron. European journal of clinical investigation 32 Suppl 1, 50‐ 54. Hod, E.A., Zhang, N., Sokol, S.A., Wojczyk, B.S., Francis, R.O., Ansaldi, D., Francis, K.P., Della‐Latta, P., Whittier, S., Sheth, S., et al. (2010). Transfusion of red blood cells after prolonged storage produces harmful effects that are mediated by iron and inflammation. Blood 115, 4284‐4292. Hoeflich, K.P., Luo, J., Rubie, E.A., Tsao, M.S., Jin, O., and Woodgett, J.R. (2000). Requirement for glycogen synthase kinase‐3beta in cell survival and NF‐kappaB activation. Nature 406, 86‐90. Hoffman, R. (2000). Hematology : basic principles and practice (Chapter 29), 3rd edn (New York ; Edinburgh: Churchill‐Livingstone). Hole, P.S., Darley, R.L., and Tonks, A. (2011). Do reactive oxygen species play a role in myeloid leukemias? Blood 117, 5816‐5826. Hollstein, M., Sidransky, D., Vogelstein, B., and Harris, C.C. (1991). p53 mutations in human cancers. Science 253, 49‐53. Hope, K.J., Jin, L., and Dick, J.E. (2004). Acute myeloid leukemia originates from a hierarchy of leukemic stem cell classes that differ in self‐renewal capacity. Nat Immunol 5, 738‐743. Horiike, S., Kita‐Sasai, Y., Nakao, M., and Taniwaki, M. (2003). Configuration of the TP53 gene as an independent prognostic parameter of myelodysplastic syndrome. Leukemia & lymphoma 44, 915‐922. Hussain, A.R., Ahmed, S.O., Ahmed, M., Khan, O.S., Al Abdulmohsen, S., Platanias, L.C., Al‐Kuraya, K.S., and Uddin, S. (2012). Cross‐talk between NFkB and the PI3‐kinase/AKT pathway can be targeted in primary effusion lymphoma (PEL) cell lines for efficient apoptosis. PloS one 7, e39945. ICSH, Lee, S.H., Erber, W.N., Porwit, A., Tomonaga, M., Peterson, L.C., and International Council for Standardization In, H. (2008). ICSH guidelines for the standardization of bone marrow specimens and reports. International journal of laboratory hematology 30, 349‐364.

203

Ikeda, Y., Ozono, I., Tajima, S., Imao, M., Horinouchi, Y., Izawa‐Ishizawa, Y., Kihira, Y., Miyamoto, L., Ishizawa, K., Tsuchiya, K., et al. (2014). Iron chelation by deferoxamine prevents renal interstitial fibrosis in mice with unilateral ureteral obstruction. PloS one 9, e89355. Ilbert, M., and Bonnefoy, V. (2013). Insight into the evolution of the iron oxidation pathways. Biochimica et biophysica acta 1827, 161‐175. International Standing Committee on Human Cytogenetic Nomenclature., Shaffer, L.G., Slovak, M.L., and Campbell, L.J. (2009). ISCN 2009 : an international system for human cytogenetic nomenclature (2009) (Basel ; Farmington, CT: Karger). Issa, J.P. (2013). The myelodysplastic syndrome as a prototypical epigenetic disease. Blood 121, 3811‐3817. Itzykson, R., Thepot, S., Beyne‐Rauzy, O., Ame, S., Isnard, F., Dreyfus, F., Salanoubat, C., Taksin, A.L., Chelgoum, Y., Berthon, C., et al. (2012). Does addition of erythropoiesis stimulating agents improve the outcome of higher‐risk myelodysplastic syndromes treated with azacitidine? Leukemia research 36, 397‐400. Ivanov, V.K., Tsyb, A.F., Gorsky, A.I., Maksyutov, M.A., Rastopchin, E.M., Konogorov, A.P., Korelo, A.M., Biryukov, A.P., and Matyash, V.A. (1997). Leukaemia and thyroid cancer in emergency workers of the Chernobyl accident: estimation of radiation risks (1986‐1995). Radiat Environ Biophys 36, 9‐16. Iwanaga, M., Hsu, W.L., Soda, M., Takasaki, Y., Tawara, M., Joh, T., Amenomori, T., Yamamura, M., Yoshida, Y., Koba, T., et al. (2011). Risk of myelodysplastic syndromes in people exposed to ionizing radiation: a retrospective cohort study of Nagasaki atomic bomb survivors. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 29, 428‐434. Iyoda, T., and Fukai, F. (2012). Modulation of Tumor Cell Survival, Proliferation, and Differentiation by the Peptide Derived from Tenascin‐C: Implication of beta1‐Integrin Activation. Int J Cell Biol 2012, 647594. Jang, Y.Y., and Sharkis, S.J. (2007). A low level of reactive oxygen species selects for primitive hematopoietic stem cells that may reside in the low‐oxygenic niche. Blood 110, 3056‐3063. Jansen, A.J., Essink‐Bot, M.L., Beckers, E.A., Hop, W.C., Schipperus, M.R., and Van Rhenen, D.J. (2003). Quality of life measurement in patients with transfusion‐dependent myelodysplastic syndromes. British journal of haematology 121, 270‐274. Jaramillo, M.C., and Zhang, D.D. (2013). The emerging role of the Nrf2‐Keap1 signaling pathway in cancer. Genes & development 27, 2179‐2191. Jensen, P.D. (2004). Evaluation of iron overload. British journal of haematology 124, 697‐711. Jensen, P.D., Heickendorff, L., Pedersen, B., Bendix‐Hansen, K., Jensen, F.T., Christensen, T., Boesen, A.M., and Ellegaard, J. (1996). The effect of iron chelation on haemopoiesis in MDS patients with transfusional iron overload. British journal of haematology 94, 288‐299. Jiang, Y., Dunbar, A., Gondek, L.P., Mohan, S., Rataul, M., O'Keefe, C., Sekeres, M., Saunthararajah, Y., and Maciejewski, J.P. (2009). Aberrant DNA methylation is a dominant mechanism in MDS progression to AML. Blood 113, 1315‐1325. Jo, H., Zhang, R., Zhang, H., McKinsey, T.A., Shao, J., Beauchamp, R.D., Ballard, D.W., and Liang, P. (2000). NF‐kappa B is required for H‐ras oncogene induced abnormal cell proliferation and tumorigenesis. Oncogene 19, 841‐849. Jubb, A.M., Strickland, L.A., Liu, S.D., Mak, J., Schmidt, M., and Koeppen, H. (2012). Neuropilin‐1 expression in cancer and development. J Pathol 226, 50‐60. Kakhlon, O., and Cabantchik, Z.I. (2002). The labile iron pool: characterization, measurement, and participation in cellular processes(1). Free radical biology & medicine 33, 1037‐1046. Kamble, R.T., Selby, G.B., Mims, M., Kharfan‐Dabaja, M.A., Ozer, H., and George, J.N. (2006). Iron overload manifesting as apparent exacerbation of hepatic graft‐versus‐host disease after allogeneic hematopoietic stem cell transplantation. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 12, 506‐510. Kantarjian, H., Giles, F., List, A., Lyons, R., Sekeres, M.A., Pierce, S., Deuson, R., and Leveque, J. (2007). The incidence and impact of thrombocytopenia in myelodysplastic syndromes. Cancer 109, 1705‐1714.

204

Kantarjian, H., Issa, J.P., Rosenfeld, C.S., Bennett, J.M., Albitar, M., DiPersio, J., Klimek, V., Slack, J., de Castro, C., Ravandi, F., et al. (2006). Decitabine improves patient outcomes in myelodysplastic syndromes: results of a phase III randomized study. Cancer 106, 1794‐1803. Kantarjian, H., O'Brien, S., Ravandi, F., Cortes, J., Shan, J., Bennett, J.M., List, A., Fenaux, P., Sanz, G., Issa, J.P., et al. (2008). Proposal for a new risk model in myelodysplastic syndrome that accounts for events not considered in the original International Prognostic Scoring System. Cancer 113, 1351‐1361. Kanwar, P., and Kowdley, K.V. (2013). Diagnosis and treatment of hereditary hemochromatosis: an update. Expert review of gastroenterology & hepatology 7, 517‐530. Kao, J.M., McMillan, A., and Greenberg, P.L. (2008). International MDS risk analysis workshop (IMRAW)/IPSS reanalyzed: impact of cytopenias on clinical outcomes in myelodysplastic syndromes. Am J Hematol 83, 765‐770. Kaomongkolgit, R., Cheepsunthorn, P., Pavasant, P., and Sanchavanakit, N. (2008). Iron increases MMP‐9 expression through activation of AP‐1 via ERK/Akt pathway in human head and neck squamous carcinoma cells. Oral oncology 44, 587‐594. Karimi, M., Giti, R., Haghpanah, S., Azarkeivan, A., Hoofar, H., and Eslami, M. (2009). Malignancies in patients with beta‐thalassemia major and beta‐thalassemia intermedia: a multicenter study in Iran. Pediatr Blood Cancer 53, 1064‐1067. Karlsson, M., Kurz, T., Brunk, U.T., Nilsson, S.E., and Frennesson, C.I. (2010). What does the commonly used DCF test for oxidative stress really show? The Biochemical journal 428, 183‐190. Katoh, M. (2008). WNT signaling in stem cell biology and regenerative medicine. Curr Drug Targets 9, 565‐570. Katoh, M., and Katoh, M. (2007). Comparative integromics on FZD7 orthologs: conserved binding sites for PU.1, SP1, CCAAT‐box and TCF/LEF/SOX transcription factors within 5'‐promoter region of mammalian FZD7 orthologs. Int J Mol Med 19, 529‐533. Kautz, L., Jung, G., Valore, E.V., Rivella, S., Nemeth, E., and Ganz, T. (2014). Identification of erythroferrone as an erythroid regulator of iron metabolism. Nature genetics 46, 678‐684. Kehrer, J.P. (2000). The Haber‐Weiss reaction and mechanisms of toxicity. Toxicology 149, 43‐50. Kentsis, A., Reed, C., Rice, K.L., Sanda, T., Rodig, S.J., Tholouli, E., Christie, A., Valk, P.J., Delwel, R., Ngo, V., et al. (2012). Autocrine activation of the MET receptor tyrosine kinase in acute myeloid leukemia. Nature medicine 18, 1118‐1122. Khan, A.A., and Quigley, J.G. (2013). Heme and FLVCR‐related transporter families SLC48 and SLC49. Molecular aspects of medicine 34, 669‐682. Khan, H., Vale, C., Bhagat, T., and Verma, A. (2013). Role of DNA methylation in the pathogenesis and treatment of myelodysplastic syndromes. Seminars in hematology 50, 16‐37. Kochenderfer, J.N., Kobayashi, S., Wieder, E.D., Su, C., and Molldrem, J.J. (2002). Loss of T‐lymphocyte clonal dominance in patients with myelodysplastic syndrome responsive to immunosuppression. Blood 100, 3639‐3645. Kogan, S.C., Ward, J.M., Anver, M.R., Berman, J.J., Brayton, C., Cardiff, R.D., Carter, J.S., de Coronado, S., Downing, J.R., Fredrickson, T.N., et al. (2002). Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice. Blood 100, 238‐245. Konen, E., Ghoti, H., Goitein, O., Winder, A., Kushnir, T., Eshet, Y., and Rachmilewitz, E. (2007). No evidence for myocardial iron overload in multitransfused patients with myelodysplastic syndrome using cardiac magnetic resonance T2 technique. American journal of hematology 82, 1013‐1016. Konstantopoulos, K., Lauren, L., Hast, R., and Reizenstein, P. (1989). Survival, hospitalization and cause of death in 99 patients with the myelodysplastic syndrome. Anticancer research 9, 893‐896. Kops, G.J., Dansen, T.B., Polderman, P.E., Saarloos, I., Wirtz, K.W., Coffer, P.J., Huang, T.T., Bos, J.L., Medema, R.H., and Burgering, B.M. (2002). Forkhead transcription factor FOXO3a protects quiescent cells from oxidative stress. Nature 419, 316‐321. Koreth, J., and Antin, J.H. (2010). Iron overload in hematologic malignancies and outcome of allogeneic hematopoietic stem cell transplantation. Haematologica 95, 364‐366.

205

Koreth, J., Pidala, J., Perez, W.S., Deeg, H.J., Garcia‐Manero, G., Malcovati, L., Cazzola, M., Park, S., Itzykson, R., Ades, L., et al. (2013). Role of reduced‐intensity conditioning allogeneic hematopoietic stem‐cell transplantation in older patients with de novo myelodysplastic syndromes: an international collaborative decision analysis. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 31, 2662‐2670. Kornblith, A.B., Herndon, J.E., 2nd, Silverman, L.R., Demakos, E.P., Odchimar‐Reissig, R., Holland, J.F., Powell, B.L., DeCastro, C., Ellerton, J., Larson, R.A., et al. (2002). Impact of azacytidine on the quality of life of patients with myelodysplastic syndrome treated in a randomized phase III trial: a Cancer and Leukemia Group B study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 20, 2441‐2452. Korolnek, T., and Hamza, I. (2014). Like iron in the blood of the people: the requirement for heme trafficking in iron metabolism. Frontiers in pharmacology 5, 126. Koulnis, M., Pop, R., Porpiglia, E., Shearstone, J.R., Hidalgo, D., and Socolovsky, M. (2011). Identification and analysis of mouse erythroid progenitors using the CD71/TER119 flow‐cytometric assay. J Vis Exp. Kowaltowski, A.J., de Souza‐Pinto, N.C., Castilho, R.F., and Vercesi, A.E. (2009). Mitochondria and reactive oxygen species. Free radical biology & medicine 47, 333‐343. Kubota, Y., Ohnishi, H., Kitanaka, A., Ishida, T., and Tanaka, T. (2004). Constitutive activation of PI3K is involved in the spontaneous proliferation of primary acute myeloid leukemia cells: direct evidence of PI3K activation. Leukemia 18, 1438‐1440. Kuehnert, M.J., Roth, V.R., Haley, N.R., Gregory, K.R., Elder, K.V., Schreiber, G.B., Arduino, M.J., Holt, S.C., Carson, L.A., Banerjee, S.N., et al. (2001). Transfusion‐transmitted bacterial infection in the United States, 1998 through 2000. Transfusion 41, 1493‐1499. Kunisaki, Y., Bruns, I., Scheiermann, C., Ahmed, J., Pinho, S., Zhang, D., Mizoguchi, T., Wei, Q., Lucas, D., Ito, K., et al. (2013). Arteriolar niches maintain haematopoietic stem cell quiescence. Nature 502, 637‐643. Kuwahara, A., Sakai, H., Xu, Y., Itoh, Y., Hirabayashi, Y., and Gotoh, Y. (2014). Tcf3 represses Wnt‐beta‐catenin signaling and maintains neural stem cell population during neocortical development. PloS one 9, e94408. Lagadinou, E.D., Sach, A., Callahan, K., Rossi, R.M., Neering, S.J., Minhajuddin, M., Ashton, J.M., Pei, S., Grose, V., O'Dwyer, K.M., et al. (2013). BCL‐2 inhibition targets oxidative phosphorylation and selectively eradicates quiescent human leukemia stem cells. Cell stem cell 12, 329‐341. Lagasse, E., and Weissman, I.L. (1996). Flow cytometric identification of murine neutrophils and monocytes. J Immunol Methods 197, 139‐150. Landriscina, M., Maddalena, F., Laudiero, G., and Esposito, F. (2009). Adaptation to oxidative stress, chemoresistance, and cell survival. Antioxidants & redox signaling 11, 2701‐2716. Lane, S.W., Sykes, S.M., Al‐Shahrour, F., Shterental, S., Paktinat, M., Lo Celso, C., Jesneck, J.L., Ebert, B.L., Williams, D.A., and Gilliland, D.G. (2010). The Apc(min) mouse has altered hematopoietic stem cell function and provides a model for MPD/MDS. Blood 115, 3489‐3497. Langlois, B., Perrot, G., Schneider, C., Henriet, P., Emonard, H., Martiny, L., and Dedieu, S. (2010). LRP‐1 promotes cancer cell invasion by supporting ERK and inhibiting JNK signaling pathways. PloS one 5, e11584. Laricchia‐Robbio, L., Fazzina, R., Li, D., Rinaldi, C.R., Sinha, K.K., Chakraborty, S., and Nucifora, G. (2006). Point mutations in two EVI1 Zn fingers abolish EVI1‐GATA1 interaction and allow erythroid differentiation of murine bone marrow cells. Molecular and cellular biology 26, 7658‐7666. Laricchia‐Robbio, L., Premanand, K., Rinaldi, C.R., and Nucifora, G. (2009). EVI1 Impairs myelopoiesis by deregulation of PU.1 function. Cancer research 69, 1633‐1642. Le Blanc, S., Garrick, M.D., and Arredondo, M. (2012). Heme carrier protein 1 transports heme and is involved in heme‐Fe metabolism. American journal of physiology. Cell physiology 302, C1780‐1785. Leboeuf, R.C., Tolson, D., and Heinecke, J.W. (1995). Dissociation between tissue iron concentrations and transferrin saturation among inbred mouse strains. J Lab Clin Med 126, 128‐136. Lee, J.W., Kang, H.J., Kim, E.K., Kim, H., Shin, H.Y., and Ahn, H.S. (2009). Effect of iron overload and iron‐chelating therapy on allogeneic hematopoietic SCT in children. Bone marrow transplantation 44, 793‐797.

206

Lehmann, S., O'Kelly, J., Raynaud, S., Funk, S.E., Sage, E.H., and Koeffler, H.P. (2007). Common deleted genes in the 5q‐ syndrome: thrombocytopenia and reduced erythroid colony formation in SPARC null mice. Leukemia 21, 1931‐ 1936. Leitch, H.A. (2011). Controversies surrounding iron chelation therapy for MDS. Blood reviews 25, 17‐31. Leitch, H.A., Chan, C., Leger, C.S., Foltz, L.M., Ramadan, K.M., and Vickars, L.M. (2012). Improved survival with iron chelation therapy for red blood cell transfusion dependent lower IPSS risk MDS may be more significant in patients with a non‐RARS diagnosis. Leukemia research 36, 1380‐1386. Leitch, H.A., Leger, C.S., Goodman, T.A., Wong, K.K., Wong, D.H.C., Ramadan, K.M., Rollins, M.D., Barnett, M.J., Galbraith, P.F., and Vickars, L.M. (2008). Improved Survival in Patients with Myelodysplastic Syndrome Receiving Iron Chelation Therapy. Clinical Leukemia 2, 205‐211. Leitch, H.A., and Vickars, L.M. (2009). Supportive care and chelation therapy in MDS: are we saving lives or just lowering iron? Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program, 664‐672. Leitman, S.F. (2013). Hemochromatosis: the new blood donor. Hematology / the Education Program of the American Society of Hematology. American Society of Hematology. Education Program 2013, 645‐650. Lento, W., Congdon, K., Voermans, C., Kritzik, M., and Reya, T. (2013). Wnt signaling in normal and malignant hematopoiesis. Cold Spring Harb Perspect Biol 5. Lesniewski, L.A., Connell, M.L., Durrant, J.R., Folian, B.J., Anderson, M.C., Donato, A.J., and Seals, D.R. (2009). B6D2F1 Mice are a suitable model of oxidative stress‐mediated impaired endothelium‐dependent dilation with aging. J Gerontol A Biol Sci Med Sci 64, 9‐20. Levi, S., and Rovida, E. (2009). The role of iron in mitochondrial function. Biochimica et biophysica acta 1790, 629‐ 636. Lewis, K.N., Andziak, B., Yang, T., and Buffenstein, R. (2013). The naked mole‐rat response to oxidative stress: just deal with it. Antioxidants & redox signaling 19, 1388‐1399. Lewis, S.E., Johnson, F.M., Skow, L.C., Popp, D., Barnett, L.B., and Popp, R.A. (1985). A mutation in the beta‐globin gene detected in the progeny of a female mouse treated with ethylnitrosourea. Proceedings of the National Academy of Sciences of the United States of America 82, 5829‐5831. Li, C.J., Chang, J.K., Wang, G.J., and Ho, M.L. (2011). Constitutively expressed COX‐2 in osteoblasts positively regulates Akt signal transduction via suppression of PTEN activity. Bone 48, 286‐297. Liew, E., and Owen, C. (2011). Familial myelodysplastic syndromes: a review of the literature. Haematologica 96, 1536‐1542. Lill, R., and Muhlenhoff, U. (2008). Maturation of iron‐sulfur proteins in eukaryotes: mechanisms, connected processes, and diseases. Annual review of biochemistry 77, 669‐700. Lim, Z.Y., Killick, S., Germing, U., Cavenagh, J., Culligan, D., Bacigalupo, A., Marsh, J., and Mufti, G.J. (2007). Low IPSS score and bone marrow hypocellularity in MDS patients predict hematological responses to antithymocyte globulin. Leukemia 21, 1436‐1441. Lin, Y.W., Slape, C., Zhang, Z., and Aplan, P.D. (2005). NUP98‐HOXD13 transgenic mice develop a highly penetrant, severe myelodysplastic syndrome that progresses to acute leukemia. Blood 106, 287‐295. Lindsay, K.A., Wheldon, E.G., Deehan, C., and Wheldon, T.E. (2001). Radiation carcinogenesis modelling for risk of treatment‐related second tumours following radiotherapy. Br J Radiol 74, 529‐536. List, A., Dewald, G., Bennett, J., Giagounidis, A., Raza, A., Feldman, E., Powell, B., Greenberg, P., Thomas, D., Stone, R., et al. (2006). Lenalidomide in the myelodysplastic syndrome with chromosome 5q deletion. The New England journal of medicine 355, 1456‐1465. List, A., Kurtin, S., Roe, D.J., Buresh, A., Mahadevan, D., Fuchs, D., Rimsza, L., Heaton, R., Knight, R., and Zeldis, J.B. (2005). Efficacy of lenalidomide in myelodysplastic syndromes. N Engl J Med 352, 549‐557. Little, M.P., Weiss, H.A., Boice, J.D., Jr., Darby, S.C., Day, N.E., and Muirhead, C.R. (1999). Risks of leukemia in Japanese atomic bomb survivors, in women treated for cervical cancer, and in patients treated for ankylosing spondylitis. Radiat Res 152, 280‐292. 207

LLS (2014). Leukemia & Lymphoma Society Facts Spring 2014. LLSC (2013). Leukemia & Lymphoma Society of Canada 2013 Blood Cancer Statistics. Loevner, L.A., Tobey, J.D., Yousem, D.M., Sonners, A.I., and Hsu, W.C. (2002). MR imaging characteristics of cranial bone marrow in adult patients with underlying systemic disorders compared with healthy control subjects. AJNR. American journal of neuroradiology 23, 248‐254. Lopez‐Girona, A., Mendy, D., Ito, T., Miller, K., Gandhi, A.K., Kang, J., Karasawa, S., Carmel, G., Jackson, P., Abbasian, M., et al. (2012). Cereblon is a direct protein target for immunomodulatory and antiproliferative activities of lenalidomide and pomalidomide. Leukemia 26, 2326‐2335. Lorimore, S.A., Mukherjee, D., Robinson, J.I., Chrystal, J.A., and Wright, E.G. (2011). Long‐lived inflammatory signaling in irradiated bone marrow is genome dependent. Cancer research 71, 6485‐6491. Lu, W., Fu, Z., Wang, H., Feng, J., Wei, J., and Guo, J. (2014). Peroxiredoxin 2 is upregulated in colorectal cancer and contributes to colorectal cancer cells' survival by protecting cells from oxidative stress. Molecular and cellular biochemistry 387, 261‐270. Lu, W., Zhao, M., Rajbhandary, S., Xie, F., Chai, X., Mu, J., Meng, J., Liu, Y., Jiang, Y., Xu, X., et al. (2013). Free iron catalyzes oxidative damage to hematopoietic cells/mesenchymal stem cells in vitro and suppresses hematopoiesis in iron overload patients. European journal of haematology 91, 249‐261. Lubbert, M., Suciu, S., Baila, L., Ruter, B.H., Platzbecker, U., Giagounidis, A., Selleslag, D., Labar, B., Germing, U., Salih, H.R., et al. (2011). Low‐dose decitabine versus best supportive care in elderly patients with intermediate‐ or high‐risk myelodysplastic syndrome (MDS) ineligible for intensive chemotherapy: final results of the randomized phase III study of the European Organisation for Research and Treatment of Cancer Leukemia Group and the German MDS Study Group. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 29, 1987‐1996. Lui, G.Y., Obeidy, P., Ford, S.J., Tselepis, C., Sharp, D.M., Jansson, P.J., Kalinowski, D.S., Kovacevic, Z., Lovejoy, D.B., and Richardson, D.R. (2013). The iron chelator, deferasirox, as a novel strategy for cancer treatment: oral activity against human lung tumor xenografts and molecular mechanism of action. Mol Pharmacol 83, 179‐190. Lutgens, F.K., and Tarbuck, E.J. (2000). Essentials of geology (Chapter 2), 7th edn (Upper Saddle River, NJ: Prentice Hall). Ma, Q., and He, X. (2012). Molecular basis of electrophilic and oxidative defense: promises and perils of Nrf2. Pharmacological reviews 64, 1055‐1081. Ma, X. (2012). Epidemiology of myelodysplastic syndromes. The American journal of medicine 125, S2‐5. Ma, X., Does, M., Raza, A., and Mayne, S.T. (2007). Myelodysplastic syndromes: incidence and survival in the United States. Cancer 109, 1536‐1542. Maakaron, J.E., Cappellini, M.D., Graziadei, G., Ayache, J.B., and Taher, A.T. (2013). Hepatocellular carcinoma in hepatitis‐negative patients with thalassemia intermedia: a closer look at the role of siderosis. Ann Hepatol 12, 142‐ 146. Maegawa, S., Gough, S.M., Watanabe‐Okochi, N., Lu, Y., Zhang, N., Castoro, R.J., Estecio, M.R., Jelinek, J., Liang, S., Kitamura, T., et al. (2014). Age‐related epigenetic drift in the pathogenesis of MDS and AML. Genome research 24, 580‐591. Mahesh, S., Ginzburg, Y., and Verma, A. (2008). Iron overload in myelodysplastic syndromes. Leukemia & lymphoma 49, 427‐438. Mahindra, A., Bolwell, B., Sobecks, R., Rybicki, L., Pohlman, B., Dean, R., Andresen, S., Sweetenham, J., Kalaycio, M., and Copelan, E. (2009a). Elevated pretransplant ferritin is associated with a lower incidence of chronic graft‐versus‐ host disease and inferior survival after myeloablative allogeneic haematopoietic stem cell transplantation. British journal of haematology 146, 310‐316. Mahindra, A., Sobecks, R., Rybicki, L., Pohlman, B., Dean, R., Andresen, S., Kalaycio, M., Sweetenham, J., Bolwell, B., and Copelan, E. (2009b). Elevated pretransplant serum ferritin is associated with inferior survival following nonmyeloablative allogeneic transplantation. Bone marrow transplantation 44, 767‐768.

208

Majhail, N.S., Lazarus, H.M., and Burns, L.J. (2010). A prospective study of iron overload management in allogeneic hematopoietic cell transplantation survivors. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 16, 832‐837. Major, I.R. (1979). Induction of myeloid leukaemia by whole‐body single exposure of CBA male mice to x‐rays. Br J Cancer 40, 903‐913. Malcovati, L. (2007). Impact of transfusion dependency and secondary iron overload on the survival of patients with myelodysplastic syndromes. Leukemia research 31 Suppl 3, S2‐6. Malcovati, L. (2009). Red blood cell transfusion therapy and iron chelation in patients with myelodysplastic syndromes. Clinical lymphoma & myeloma 9 Suppl 3, S305‐311. Malcovati, L., and Cazzola, M. (2008). Correspondence: In Reply. J Clin Oncol. 26, 1180. Malcovati, L., and Cazzola, M. (2013). Refractory anemia with ring sideroblasts. Best practice & research. Clinical haematology 26, 377‐385. Malcovati, L., Della Porta, M.G., and Cazzola, M. (2006). Predicting survival and leukemic evolution in patients with myelodysplastic syndrome. Haematologica 91, 1588‐1590. Malcovati, L., Della Porta, M.G., Strupp, C., Ambaglio, I., Kuendgen, A., Nachtkamp, K., Travaglino, E., Invernizzi, R., Pascutto, C., Lazzarino, M., et al. (2011). Impact of the degree of anemia on the outcome of patients with myelodysplastic syndrome and its integration into the WHO classification‐based Prognostic Scoring System (WPSS). Haematologica 96, 1433‐1440. Malcovati, L., Germing, U., Kuendgen, A., Della Porta, M.G., Pascutto, C., Invernizzi, R., Giagounidis, A., Hildebrandt, B., Bernasconi, P., Knipp, S., et al. (2007). Time‐dependent prognostic scoring system for predicting survival and leukemic evolution in myelodysplastic syndromes. J Clin Oncol 25, 3503‐3510. Malcovati, L., Porta, M.G., Pascutto, C., Invernizzi, R., Boni, M., Travaglino, E., Passamonti, F., Arcaini, L., Maffioli, M., Bernasconi, P., et al. (2005). Prognostic factors and life expectancy in myelodysplastic syndromes classified according to WHO criteria: a basis for clinical decision making. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 23, 7594‐7603. Malinge, S., Monni, R., Bernard, O., and Penard‐Lacronique, V. (2006). Activation of the NF‐kappaB pathway by the leukemogenic TEL‐Jak2 and TEL‐Abl fusion proteins leads to the accumulation of antiapoptotic IAP proteins and involves IKKalpha. Oncogene 25, 3589‐3597. Mancuso, A. (2010). Hepatocellular carcinoma in thalassemia: A critical review. World J Hepatol 2, 171‐174. Mandavilli, B.S., and Janes, M.S. (2010). Detection of intracellular glutathione using ThiolTracker violet stain and fluorescence microscopy. Curr Protoc Cytom Chapter 9, Unit 9 35. Marisavljevic, D., Rolovic, Z., Cemerikic, V., Boskovic, D., and Colovic, M. (2004). Myelofibrosis in primary myelodysplastic syndromes: clinical and biological significance. Medical oncology 21, 325‐331. Martelli, A.M., Evangelisti, C., Chiarini, F., and McCubrey, J.A. (2010). The phosphatidylinositol 3‐kinase/Akt/mTOR signaling network as a therapeutic target in acute myelogenous leukemia patients. Oncotarget 1, 89‐103. Martino, R., Caballero, M.D., Perez‐Simon, J.A., Canals, C., Solano, C., Urbano‐Ispizua, A., Bargay, J., Leon, A., Sarra, J., Sanz, G.F., et al. (2002). Evidence for a graft‐versus‐leukemia effect after allogeneic peripheral blood stem cell transplantation with reduced‐intensity conditioning in acute myelogenous leukemia and myelodysplastic syndromes. Blood 100, 2243‐2245. Masala, E., Valencia, A., Buchi, F., Nosi, D., Spinelli, E., Gozzini, A., Sassolini, F., Sanna, A., Zecchi, S., Bosi, A., et al. (2012). Hypermethylation of Wnt antagonist gene promoters and activation of Wnt pathway in myelodysplastic marrow cells. Leukemia research 36, 1290‐1295. Maschek, H., Kaloutsi, V., Rodriguez‐Kaiser, M., Werner, M., Choritz, H., Mainzer, K., Dietzfelbinger, M., and Georgii, A. (1993). Hypoplastic myelodysplastic syndrome: incidence, morphology, cytogenetics, and prognosis. Annals of hematology 66, 117‐122. McCullough, K.D., and Bartfay, W.J. (2007). The dose‐dependent effects of chronic iron overload on the production of oxygen free radicals and vitamin E concentrations in the liver of a murine model. Biol Res Nurs 8, 300‐304.

209

McKay, P.J., Murphy, J.A., Cameron, S., Burnett, A.K., Campbell, M., Tansey, P., and Franklin, I.M. (1996). Iron overload and liver dysfunction after allogeneic or autologous bone marrow transplantation. Bone marrow transplantation 17, 63‐66. McLaren, G.D., Muir, W.A., and Kellermeyer, R.W. (1983). Iron overload disorders: natural history, pathogenesis, diagnosis, and therapy. Critical reviews in clinical laboratory sciences 19, 205‐266. McLellan, S., and Walsh, T. (2004). Oxygen delivery and haemoglobin. Contin Educ Anaesth Crit Care Pain 4, 123‐ 126. Meloni, A., Renni, R., Romano, N., Cirotto, C., Gagliardotto, F., Caniglia, M., De Franceschi, L., Neri, M.G., Positano, V., Valeri, G., et al. (2014). Quantitative T2*MRI for Bone Marrow Iron Overload Assessment in Thalassemia Major and Intemedia Patients, Vol 124. Messa, E., Carturan, S., Maffe, C., Pautasso, M., Bracco, E., Roetto, A., Messa, F., Arruga, F., Defilippi, I., Rosso, V., et al. (2010). Deferasirox is a powerful NF‐kappaB inhibitor in myelodysplastic cells and in leukemia cell lines acting independently from cell iron deprivation by chelation and reactive oxygen species scavenging. Haematologica 95, 1308‐1316. Mester, J., and Eng, C. (2013). When overgrowth bumps into cancer: the PTEN‐opathies. Am J Med Genet C Semin Med Genet 163C, 114‐121. Mitani, K. (2004). Molecular mechanisms of leukemogenesis by AML1/EVI‐1. Oncogene 23, 4263‐4269. Mitchell, M., Gore, S.D., and Zeidan, A.M. (2013). Iron chelation therapy in myelodysplastic syndromes: where do we stand? Expert review of hematology 6, 397‐410. Mittelman, M., Lugassy, G., Merkel, D., Tamary, H., Sarid, N., Rachmilewitz, E., Hershko, C., Group, M.D.S.I., and Israel Society of, H. (2008). Iron chelation therapy in patients with myelodysplastic syndromes: consensus conference guidelines. The Israel Medical Association journal : IMAJ 10, 374‐376. Montalban Bravo, G., Lee, E., Merchan, B., Kantarjian, H.M., and Garcia‐Manero, G. (2014). Integrating genetics and epigenetics in myelodysplastic syndromes: advances in pathogenesis and disease evolution. British journal of haematology. Moon, S.N., Han, J.W., Hwang, H.S., Kim, M.J., Lee, S.J., Lee, J.Y., Oh, C.K., and Jeong, D.C. (2011). Establishment of secondary iron overloaded mouse model: evaluation of cardiac function and analysis according to iron concentration. Pediatr Cardiol 32, 947‐952. Mora, A., Komander, D., van Aalten, D.M., and Alessi, D.R. (2004). PDK1, the master regulator of AGC kinase signal transduction. Semin Cell Dev Biol 15, 161‐170. Moran‐Crusio, K., Reavie, L., Shih, A., Abdel‐Wahab, O., Ndiaye‐Lobry, D., Lobry, C., Figueroa, M.E., Vasanthakumar, A., Patel, J., Zhao, X., et al. (2011). Tet2 loss leads to increased hematopoietic stem cell self‐renewal and myeloid transformation. Cancer cell 20, 11‐24. Morello, V., Cabodi, S., Sigismund, S., Camacho‐Leal, M.P., Repetto, D., Volante, M., Papotti, M., Turco, E., and Defilippi, P. (2011). beta1 integrin controls EGFR signaling and tumorigenic properties of lung cancer cells. Oncogene 30, 4087‐4096. Mori, S., Nada, S., Kimura, H., Tajima, S., Takahashi, Y., Kitamura, A., Oneyama, C., and Okada, M. (2014). The mTOR pathway controls cell proliferation by regulating the FoxO3a transcription factor via SGK1 kinase. PloS one 9, e88891. Mori, T., Maeda, N., Inoue, K., Sekimoto, R., Tsushima, Y., Matsuda, K., Yamaoka, M., Suganami, T., Nishizawa, H., Ogawa, Y., et al. (2013). A novel role for adipose ephrin‐B1 in inflammatory response. PloS one 8, e76199. Moriya, K., Suzuki, M., Watanabe, Y., Takahashi, T., Aoki, Y., Uchiyama, T., Kumaki, S., Sasahara, Y., Minegishi, M., Kure, S., et al. (2012). Development of a multi‐step leukemogenesis model of MLL‐rearranged leukemia using humanized mice. PloS one 7, e37892. Morrison, S.J., and Scadden, D.T. (2014). The bone marrow niche for haematopoietic stem cells. Nature 505, 327‐ 334.

210

Mueller, B.U., Pabst, T., Osato, M., Asou, N., Johansen, L.M., Minden, M.D., Behre, G., Hiddemann, W., Ito, Y., and Tenen, D.G. (2002). Heterozygous PU.1 mutations are associated with acute myeloid leukemia. Blood 100, 998‐ 1007. Mukhopadhyay, P., Rajesh, M., Hasko, G., Hawkins, B.J., Madesh, M., and Pacher, P. (2007). Simultaneous detection of apoptosis and mitochondrial superoxide production in live cells by flow cytometry and confocal microscopy. Nat Protoc 2, 2295‐2301. Muller‐Tidow, C., Steffen, B., Cauvet, T., Tickenbrock, L., Ji, P., Diederichs, S., Sargin, B., Kohler, G., Stelljes, M., Puccetti, E., et al. (2004). Translocation products in acute myeloid leukemia activate the Wnt signaling pathway in hematopoietic cells. Molecular and cellular biology 24, 2890‐2904. Muncie, H.L., Jr., and Campbell, J. (2009). Alpha and beta thalassemia. American family physician 80, 339‐344. Munker, R., Hiller, E., and Paquette, R. (2000a). Modern hematology : biology and clinical management (6. Hemolytic Anemias) (Totowa, N.J.: Humana Press). Munker, R., Hiller, E., and Paquette, R. (2000b). Modern hematology : biology and clinical management (11. Myelodysplastic Syndromes) (Totowa, N.J.: Humana Press). Murphy, M.F., Wallington, T.B., Kelsey, P., Boulton, F., Bruce, M., Cohen, H., Duguid, J., Knowles, S.M., Poole, G., Williamson, L.M., et al. (2001). Guidelines for the clinical use of red cell transfusions. British journal of haematology 113, 24‐31. Musallam, K.M., Angastiniotis, M., Eleftheriou, A., and Porter, J.B. (2013). Cross‐talk between available guidelines for the management of patients with beta‐thalassemia major. Acta haematologica 130, 64‐73. Musumeci, M., Maccari, S., Massimi, A., Stati, T., Sestili, P., Corritore, E., Pastorelli, A., Stacchini, P., Marano, G., and Catalano, L. (2014). Iron excretion in iron dextran‐overloaded mice. Blood Transfus 12, 485‐490. Muto, T., Sashida, G., Oshima, M., Wendt, G.R., Mochizuki‐Kashio, M., Nagata, Y., Sanada, M., Miyagi, S., Saraya, A., Kamio, A., et al. (2013). Concurrent loss of Ezh2 and Tet2 cooperates in the pathogenesis of myelodysplastic disorders. The Journal of experimental medicine 210, 2627‐2639. Mutter, G.L. (2001). Pten, a protean tumor suppressor. Am J Pathol 158, 1895‐1898. Nagler, A., Binet, C., Mackichan, M.L., Negrin, R., Bangs, C., Donlon, T., and Greenberg, P. (1990). Impact of marrow cytogenetics and morphology on in vitro hematopoiesis in the myelodysplastic syndromes: comparison between recombinant human granulocyte colony‐stimulating factor (CSF) and granulocyte‐monocyte CSF. Blood 76, 1299‐ 1307. Nakano, H. (2004). Signaling crosstalk between NF‐kappaB and JNK. Trends Immunol 25, 402‐405. Nakao, S., Sugimori, C., and Yamazaki, H. (2006). Clinical significance of a small population of paroxysmal nocturnal hemoglobinuria‐type cells in the management of bone marrow failure. International journal of hematology 84, 118‐122. Nathaniel, T.I., Otukonyong, E., Abdellatif, A., and Soyinka, J.O. (2012). Effect of hypoxia on metabolic rate, core body temperature, and c‐fos expression in the naked mole rat. International journal of developmental neuroscience : the official journal of the International Society for Developmental Neuroscience 30, 539‐544. Nemeth, E., Tuttle, M.S., Powelson, J., Vaughn, M.B., Donovan, A., Ward, D.M., Ganz, T., and Kaplan, J. (2004). Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science 306, 2090‐ 2093. Netto, L.E., Ferreira, A.M., and Augusto, O. (1991). Iron(III) binding in DNA solutions: complex formation and catalytic activity in the oxidation of hydrazine derivatives. Chem Biol Interact 79, 1‐14. Neukirchen, J., Fox, F., Kundgen, A., Nachtkamp, K., Strupp, C., Haas, R., Germing, U., and Gattermann, N. (2012). Improved survival in MDS patients receiving iron chelation therapy ‐ a matched pair analysis of 188 patients from the Dusseldorf MDS registry. Leukemia research 36, 1067‐1070. Neukirchen, J., Lauseker, M., Blum, S., Giagounidis, A., Lubbert, M., Martino, S., Siragusa, S., Schlenk, R.F., Platzbecker, U., Hofmann, W.K., et al. (2014). Validation of the revised international prognostic scoring system (IPSS‐R) in patients with myelodysplastic syndrome: a multicenter study. Leukemia research 38, 57‐64. Neumann, C.A., and Fang, Q. (2007). Are peroxiredoxins tumor suppressors? Curr Opin Pharmacol 7, 375‐380. 211

Nick, H., Allegrini, P.R., Fozard, L., Junker, U., Rojkjaer, L., Salie, R., Niederkofler, V., and O'Reilly, T. (2009). Deferasirox reduces iron overload in a murine model of juvenile hemochromatosis. Exp Biol Med (Maywood) 234, 492‐503. Niforou, K., Cheimonidou, C., and Trougakos, I.P. (2014). Molecular chaperones and proteostasis regulation during redox imbalance. Redox biology 2, 323‐332. Nishida, N., and Kudo, M. (2013). Oxidative stress and epigenetic instability in human hepatocarcinogenesis. Digestive diseases 31, 447‐453. Nolte, F., Hochsmann, B., Giagounidis, A., Lubbert, M., Platzbecker, U., Haase, D., Luck, A., Gattermann, N., Taupitz, M., Baier, M., et al. (2013). Results from a 1‐year, open‐label, single arm, multi‐center trial evaluating the efficacy and safety of oral Deferasirox in patients diagnosed with low and int‐1 risk myelodysplastic syndrome (MDS) and transfusion‐dependent iron overload. Annals of hematology 92, 191‐198. Noshchenko, A.G., Bondar, O.Y., and Drozdova, V.D. (2010). Radiation‐induced leukemia among children aged 0‐5 years at the time of the Chernobyl accident. Int J Cancer 127, 412‐426. Nunes, P., Demaurex, N., and Dinauer, M.C. (2013). Regulation of the NADPH oxidase and associated ion fluxes during phagocytosis. Traffic 14, 1118‐1131. Nunez, F., Bravo, S., Cruzat, F., Montecino, M., and De Ferrari, G.V. (2011). Wnt/beta‐catenin signaling enhances cyclooxygenase‐2 (COX2) transcriptional activity in gastric cancer cells. PloS one 6, e18562. Nyakern, M., Tazzari, P.L., Finelli, C., Bosi, C., Follo, M.Y., Grafone, T., Piccaluga, P.P., Martinelli, G., Cocco, L., and Martelli, A.M. (2006). Frequent elevation of Akt kinase phosphorylation in blood marrow and peripheral blood mononuclear cells from high‐risk myelodysplastic syndrome patients. Leukemia 20, 230‐238. Oeckinghaus, A., and Ghosh, S. (2009). The NF‐kappaB family of transcription factors and its regulation. Cold Spring Harb Perspect Biol 1, a000034. Okabe, H., Suzuki, T., Uehara, E., Ueda, M., Nagai, T., and Ozawa, K. (2014). The bone marrow hematopoietic microenvironment is impaired in iron‐overloaded mice. European journal of haematology 93, 118‐128. Okuda, T., Nishimura, M., Nakao, M., and Fujita, Y. (2001). RUNX1/AML1: a central player in hematopoiesis. International journal of hematology 74, 252‐257. Oliva, E.N., Dimitrov, B.D., Benedetto, F., D'Angelo, A., and Nobile, F. (2005). Hemoglobin level threshold for cardiac remodeling and quality of life in myelodysplastic syndrome. Leukemia research 29, 1217‐1219. Olme, C.H., Brown, N., Finnon, R., Bouffler, S.D., and Badie, C. (2013). Frequency of acute myeloid leukaemia‐ associated mouse chromosome 2 deletions in X‐ray exposed immature haematopoietic progenitors and stem cells. Mutat Res 756, 119‐126. Omidvar, N., Kogan, S., Beurlet, S., le Pogam, C., Janin, A., West, R., Noguera, M.E., Reboul, M., Soulie, A., Leboeuf, C., et al. (2007). BCL‐2 and mutant NRAS interact physically and functionally in a mouse model of progressive myelodysplasia. Cancer research 67, 11657‐11667. Orazi, A., Albitar, M., Heerema, N.A., Haskins, S., and Neiman, R.S. (1997). Hypoplastic myelodysplastic syndromes can be distinguished from acquired aplastic anemia by CD34 and PCNA immunostaining of bone marrow biopsy specimens. American journal of clinical pathology 107, 268‐274. Osato, M. (2004). Point mutations in the RUNX1/AML1 gene: another actor in RUNX leukemia. Oncogene 23, 4284‐ 4296. Palmieri, C., Monteverde, M., Lattanzio, L., Gojis, O., Rudraraju, B., Fortunato, M., Syed, N., Thompson, A., Garrone, O., Merlano, M., et al. (2012). Site‐specific CpG methylation in the CCAAT/enhancer binding protein delta (CEBPdelta) CpG island in breast cancer is associated with metastatic relapse. Br J Cancer 107, 732‐738. Papaemmanuil, E., Gerstung, M., Malcovati, L., Tauro, S., Gundem, G., Van Loo, P., Yoon, C.J., Ellis, P., Wedge, D.C., Pellagatti, A., et al. (2013). Clinical and biological implications of driver mutations in myelodysplastic syndromes. Blood 122, 3616‐3627; quiz 3699. Paulsson, K., and Johansson, B. (2007). Trisomy 8 as the sole chromosomal aberration in acute myeloid leukemia and myelodysplastic syndromes. Pathologie‐biologie 55, 37‐48.

212

Pedersen‐Bjergaard, J., Christiansen, D.H., Andersen, M.K., and Skovby, F. (2002). Causality of myelodysplasia and acute myeloid leukemia and their genetic abnormalities. Leukemia 16, 2177‐2184. Pedersen‐Bjergaard, J., and Rowley, J.D. (1994). The balanced and the unbalanced chromosome aberrations of acute myeloid leukemia may develop in different ways and may contribute differently to malignant transformation. Blood 83, 2780‐2786. Pellagatti, A., Hellstrom‐Lindberg, E., Giagounidis, A., Perry, J., Malcovati, L., Della Porta, M.G., Jadersten, M., Killick, S., Fidler, C., Cazzola, M., et al. (2008). Haploinsufficiency of RPS14 in 5q‐ syndrome is associated with deregulation of ribosomal‐ and translation‐related genes. British journal of haematology 142, 57‐64. Peng, Y., Brown, N., Finnon, R., Warner, C.L., Liu, X., Genik, P.C., Callan, M.A., Ray, F.A., Borak, T.B., Badie, C., et al. (2009). Radiation leukemogenesis in mice: loss of PU.1 on chromosome 2 in CBA and C57BL/6 mice after irradiation with 1 GeV/nucleon 56Fe ions, X rays or gamma rays. Part I. Experimental observations. Radiat Res 171, 474‐483. Pereira, A., Nomdedeu, M., Aguilar, J.L., Belkaid, M., Carrio, A., Cobo, F., Costa, D., Rozman, M., Sanz, C., and Nomdedeu, B. (2011). Transfusion intensity, not the cumulative red blood cell transfusion burden, determines the prognosis of patients with myelodysplastic syndrome on chronic transfusion support. American journal of hematology 86, 245‐250. Perez, V.I., Buffenstein, R., Masamsetti, V., Leonard, S., Salmon, A.B., Mele, J., Andziak, B., Yang, T., Edrey, Y., Friguet, B., et al. (2009). Protein stability and resistance to oxidative stress are determinants of longevity in the longest‐living rodent, the naked mole‐rat. Proceedings of the National Academy of Sciences of the United States of America 106, 3059‐3064. Persad, S., and Dedhar, S. (2003). The role of integrin‐linked kinase (ILK) in cancer progression. Cancer Metastasis Rev 22, 375‐384. Pippard, M.J., and Weatherall, D.J. (1984). Iron absorption in non‐transfused iron loading anaemias: prediction of risk for iron loading, and response to iron chelation treatment, in beta thalassaemia intermedia and congenital sideroblastic anaemias. Haematologia 17, 17‐24. Platzbecker, U., Bornhauser, M., Germing, U., Stumpf, J., Scott, B.L., Kroger, N., Schwerdtfeger, R., Bohm, A., Kobbe, G., Theuser, C., et al. (2008). Red blood cell transfusion dependence and outcome after allogeneic peripheral blood stem cell transplantation in patients with de novo myelodysplastic syndrome (MDS). Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 14, 1217‐1225. Platzbecker, U., Hofbauer, L.C., Ehninger, G., and Holig, K. (2012). The clinical, quality of life, and economic consequences of chronic anemia and transfusion support in patients with myelodysplastic syndromes. Leukemia research 36, 525‐536. Polednak, A.P. (2013). Trend (1999‐2009) in U.S. death rates from myelodysplastic syndromes: utility of multiple causes of death in surveillance. Cancer epidemiology 37, 569‐574. Polivka, J., Jr., and Janku, F. (2014). Molecular targets for cancer therapy in the PI3K/AKT/mTOR pathway. Pharmacology & therapeutics 142, 164‐175. Pontikoglou, C., Deschaseaux, F., Sensebe, L., and Papadaki, H.A. (2011). Bone marrow mesenchymal stem cells: biological properties and their role in hematopoiesis and hematopoietic stem cell transplantation. Stem cell reviews 7, 569‐589. Porter, J., Bowden, D.K., Economou, M., Troncy, J., Ganser, A., Habr, D., Martin, N., Gater, A., Rofail, D., Abetz‐ Webb, L., et al. (2012). Health‐Related Quality of Life, Treatment Satisfaction, Adherence and Persistence in beta‐ Thalassemia and Myelodysplastic Syndrome Patients with Iron Overload Receiving Deferasirox: Results from the EPIC Clinical Trial. Anemia 2012, 297641. Porter, J., Galanello, R., Saglio, G., Neufeld, E.J., Vichinsky, E., Cappellini, M.D., Olivieri, N., Piga, A., Cunningham, M.J., Soulieres, D., et al. (2008). Relative response of patients with myelodysplastic syndromes and other transfusion‐dependent anaemias to deferasirox (ICL670): a 1‐yr prospective study. European journal of haematology 80, 168‐176. Porter, J.B. (2001). Practical management of iron overload. British journal of haematology 115, 239‐252. 213

Porter, J.B. (2009). Pathophysiology of transfusional iron overload: contrasting patterns in thalassemia major and sickle cell disease. Hemoglobin 33 Suppl 1, S37‐45. Prati, D. (2000). Benefits and complications of regular blood transfusion in patients with beta‐thalassaemia major. Vox sanguinis 79, 129‐137. Preston, D.L., Kusumi, S., Tomonaga, M., Izumi, S., Ron, E., Kuramoto, A., Kamada, N., Dohy, H., Matsuo, T., Matsui, T., et al. (1994). Cancer incidence in atomic bomb survivors. Part III. Leukemia, lymphoma and multiple myeloma, 1950‐1987. Radiat Res 137, S68‐97. Proud, C.G. (2014). Mnks, eIF4E phosphorylation and cancer. Biochimica et biophysica acta. Prus, E., and Fibach, E. (2008). Flow cytometry measurement of the labile iron pool in human hematopoietic cells. Cytometry A 73, 22‐27. Pullarkat, V. (2009). Objectives of iron chelation therapy in myelodysplastic syndromes: more than meets the eye? Blood 114, 5251‐5255. Pullarkat, V., Blanchard, S., Tegtmeier, B., Dagis, A., Patane, K., Ito, J., and Forman, S.J. (2008). Iron overload adversely affects outcome of allogeneic hematopoietic cell transplantation. Bone marrow transplantation 42, 799‐ 805. Ragu, C., Elain, G., Mylonas, E., Ottolenghi, C., Cagnard, N., Daegelen, D., Passegue, E., Vainchenker, W., Bernard, O.A., and Penard‐Lacronique, V. (2010). The transcription factor Srf regulates hematopoietic stem cell adhesion. Blood 116, 4464‐4473. Rajagopal, A., Rao, A.U., Amigo, J., Tian, M., Upadhyay, S.K., Hall, C., Uhm, S., Mathew, M.K., Fleming, M.D., Paw, B.H., et al. (2008). Haem homeostasis is regulated by the conserved and concerted functions of HRG‐1 proteins. Nature 453, 1127‐1131. Ramsey, S.D., McCune, J.S., Blough, D.K., McDermott, C.L., Beck, S.J., Lopez, J.A., and Deeg, H.J. (2012). Patterns of blood product use among patients with myelodysplastic syndrome. Vox sanguinis 102, 331‐337. Rao, T.P., and Kuhl, M. (2010). An updated overview on Wnt signaling pathways: a prelude for more. Circ Res 106, 1798‐1806. Raptis, A., Duh, M.S., Wang, S.T., Dial, E., Fanourgiakis, I., Fortner, B., Paley, C., Mody‐Patel, N., Corral, M., and Scott, J. (2010). Treatment of transfusional iron overload in patients with myelodysplastic syndrome or severe anemia: data from multicenter clinical practices. Transfusion 50, 190‐199. Raza, A., Reeves, J.A., Feldman, E.J., Dewald, G.W., Bennett, J.M., Deeg, H.J., Dreisbach, L., Schiffer, C.A., Stone, R.M., Greenberg, P.L., et al. (2008). Phase 2 study of lenalidomide in transfusion‐dependent, low‐risk, and intermediate‐1 risk myelodysplastic syndromes with karyotypes other than deletion 5q. Blood 111, 86‐93. Raza, S., TaherNazerHussain, F., Patnaik, M., Knudson, R., Van Dyke, D., and Tefferi, A. (2011). Autosomal monosomies among 24,262 consecutive cytogenetic studies: prevalence, chromosomal distribution and clinicopathologic correlates of sole abnormalities. American journal of hematology 86, 353‐356. Reczek, C.R., and Chandel, N.S. (2014). ROS‐dependent signal transduction. Current opinion in cell biology 33C, 8‐ 13. Reineke, E.L., Liu, Y., and Kao, H.Y. (2010). Promyelocytic leukemia protein controls cell migration in response to hydrogen peroxide and insulin‐like growth factor‐1. The Journal of biological chemistry 285, 9485‐9492. Resnitzky, P., Estrov, Z., and Haran‐Ghera, N. (1985). High incidence of acute myeloid leukemia in SJL/J mice after X‐irradiation and corticosteroids. Leukemia research 9, 1519‐1528. Rhind, N. (2009). Changing of the guard: how ATM hands off DNA double‐strand break signaling to ATR. Molecular cell 33, 672‐674. Rhoads, C.P., and Barker, W.H. (1938). Refractory anemia: analysis of 100 cases. 1938 110, 794‐796. Rithidech, K.N., Cronkite, E.P., and Bond, V.P. (1999). Advantages of the CBA mouse in leukemogenesis research. Blood cells, molecules & diseases 25, 38‐45. Rivina, L., Davoren, M., and Schiestl, R.H. (2014). Radiation‐induced myeloid leukemia in murine models. Hum Genomics 8, 13.

214

Rochette, L., Gudjoncik, A., Guenancia, C., Zeller, M., Cottin, Y., and Vergely, C. (2014). The iron‐regulatory hormone hepcidin: A possible therapeutic target? Pharmacology & therapeutics. Rodriguez, K.A., Edrey, Y.H., Osmulski, P., Gaczynska, M., and Buffenstein, R. (2012). Altered composition of liver proteasome assemblies contributes to enhanced proteasome activity in the exceptionally long‐lived naked mole‐ rat. PloS one 7, e35890. Rosario, C., Zandman‐Goddard, G., Meyron‐Holtz, E.G., D'Cruz, D.P., and Shoenfeld, Y. (2013). The hyperferritinemic syndrome: macrophage activation syndrome, Still's disease, septic shock and catastrophic antiphospholipid syndrome. BMC Med 11, 185. Rose, C., Brechignac, S., Vassilief, D., Pascal, L., Stamatoullas, A., Guerci, A., Larbaa, D., Dreyfus, F., Beyne‐Rauzy, O., Chaury, M.P., et al. (2010). Does iron chelation therapy improve survival in regularly transfused lower risk MDS patients? A multicenter study by the GFM (Groupe Francophone des Myelodysplasies). Leukemia research 34, 864‐ 870. Rose, C., Ernst, O., Hecquet, B., Maboudou, P., Renom, P., Noel, M.P., Yakoub‐Agha, I., Bauters, F., and Jouet, J.P. (2007). Quantification by magnetic resonance imaging and liver consequences of post‐transfusional iron overload alone in long term survivors after allogeneic hematopoietic stem cell transplantation (HSCT). Haematologica 92, 850‐853. Ruoslahti, E. (1999). Fibronectin and its integrin receptors in cancer. Adv Cancer Res 76, 1‐20. Russler‐Germain, D.A., Spencer, D.H., Young, M.A., Lamprecht, T.L., Miller, C.A., Fulton, R., Meyer, M.R., Erdmann‐ Gilmore, P., Townsend, R.R., Wilson, R.K., et al. (2014). The R882H DNMT3A mutation associated with AML dominantly inhibits wild‐type DNMT3A by blocking its ability to form active tetramers. Cancer cell 25, 442‐454. Saft, L., Karimi, M., Ghaderi, M., Matolcsy, A., Mufti, G.J., Kulasekararaj, A., Gohring, G., Giagounidis, A., Selleslag, D., Muus, P., et al. (2014). p53 protein expression independently predicts outcome in patients with lower‐risk myelodysplastic syndromes with del(5q). Haematologica 99, 1041‐1049. Sahay, B., Patsey, R.L., Eggers, C.H., Salazar, J.C., Radolf, J.D., and Sellati, T.J. (2009). CD14 signaling restrains chronic inflammation through induction of p38‐MAPK/SOCS‐dependent tolerance. PLoS Pathog 5, e1000687. Saigo, K., Takenokuchi, M., Hiramatsu, Y., Tada, H., Hishita, T., Takata, M., Misawa, M., Imoto, S., and Imashuku, S. (2011). Oxidative stress levels in myelodysplastic syndrome patients: their relationship to serum ferritin and haemoglobin values. The Journal of international medical research 39, 1941‐1945. Sakon, S., Xue, X., Takekawa, M., Sasazuki, T., Okazaki, T., Kojima, Y., Piao, J.H., Yagita, H., Okumura, K., Doi, T., et al. (2003). NF‐kappaB inhibits TNF‐induced accumulation of ROS that mediate prolonged MAPK activation and necrotic cell death. The EMBO journal 22, 3898‐3909. San Martin, C.D., Garri, C., Pizarro, F., Walter, T., Theil, E.C., and Nunez, M.T. (2008). Caco‐2 intestinal epithelial cells absorb soybean ferritin by mu2 (AP2)‐dependent endocytosis. The Journal of nutrition 138, 659‐666. Santini, V. (2011). Clinical use of erythropoietic stimulating agents in myelodysplastic syndromes. The oncologist 16 Suppl 3, 35‐42. Santini, V., Alessandrino, P.E., Angelucci, E., Barosi, G., Billio, A., Di Maio, M., Finelli, C., Locatelli, F., Marchetti, M., Morra, E., et al. (2010). Clinical management of myelodysplastic syndromes: update of SIE, SIES, GITMO practice guidelines. Leukemia research 34, 1576‐1588. Santini, V., Melnick, A., Maciejewski, J.P., Duprez, E., Nervi, C., Cocco, L., Ford, K.G., and Mufti, G. (2013). Epigenetics in focus: pathogenesis of myelodysplastic syndromes and the role of hypomethylating agents. Critical reviews in oncology/hematology 88, 231‐245. Sanz, G., Nomdedeu, B., and Such, E. (2008). Independent impact of iron overload and transfusion dependency on survival and leukemic evolution in patients with myelodysplastic syndrome. ASH annual meeting Abstracts 112, 640. Sashida, G., Harada, H., Matsui, H., Oshima, M., Yui, M., Harada, Y., Tanaka, S., Mochizuki‐Kashio, M., Wang, C., Saraya, A., et al. (2014). Ezh2 loss promotes development of myelodysplastic syndrome but attenuates its predisposition to leukaemic transformation. Nature communications 5, 4177.

215

Satchell, L., and Leake, D.S. (2012). Oxidation of low‐density lipoprotein by iron at lysosomal pH: implications for atherosclerosis. Biochemistry 51, 3767‐3775. Sawai, Y., Tamura, S., Fukui, K., Ito, N., Imanaka, K., Saeki, A., Sakuda, S., Kiso, S., and Matsuzawa, Y. (2003). Expression of ephrin‐B1 in hepatocellular carcinoma: possible involvement in neovascularization. Journal of hepatology 39, 991‐996. Schafer, A.I., Cheron, R.G., Dluhy, R., Cooper, B., Gleason, R.E., Soeldner, J.S., and Bunn, H.F. (1981). Clinical consequences of acquired transfusional iron overload in adults. The New England journal of medicine 304, 319‐ 324. Schanz, J., Tuchler, H., Sole, F., Mallo, M., Luno, E., Cervera, J., Grau, J., Hildebrandt, B., Slovak, M.L., Ohyashiki, K., et al. (2013). Monosomal karyotype in MDS: explaining the poor prognosis? Leukemia 27, 1988‐1995. Schneider, R.K., Adema, V., Heckl, D., Jaras, M., Mallo, M., Lord, A.M., Chu, L.P., McConkey, M.E., Kramann, R., Mullally, A., et al. (2014). Role of casein kinase 1A1 in the biology and targeted therapy of del(5q) MDS. Cancer cell 26, 509‐520. Schreiber, G.B., Busch, M.P., Kleinman, S.H., and Korelitz, J.J. (1996). The risk of transfusion‐transmitted viral infections. The Retrovirus Epidemiology Donor Study. The New England journal of medicine 334, 1685‐1690. Scott, B.L., Park, J.Y., Deeg, H.J., Marr, K.A., Boeckh, M., Chauncey, T.R., Appelbaum, F.R., Storb, R., and Storer, B.E. (2008). Pretransplant neutropenia is associated with poor‐risk cytogenetic features and increased infection‐related mortality in patients with myelodysplastic syndromes. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 14, 799‐806. Sedeek, M., Nasrallah, R., Touyz, R.M., and Hebert, R.L. (2013). NADPH oxidases, reactive oxygen species, and the kidney: friend and foe. Journal of the American Society of Nephrology : JASN 24, 1512‐1518. Sekeres, M.A., and Cutler, C. (2014). How we treat higher‐risk myelodysplastic syndromes. Blood 123, 829‐836. Sekeres, M.A., Tiu, R.V., Komrokji, R., Lancet, J., Advani, A.S., Afable, M., Englehaupt, R., Juersivich, J., Cuthbertson, D., Paleveda, J., et al. (2012). Phase 2 study of the lenalidomide and azacitidine combination in patients with higher‐risk myelodysplastic syndromes. Blood 120, 4945‐4951. Seluanov, A., Hine, C., Azpurua, J., Feigenson, M., Bozzella, M., Mao, Z., Catania, K.C., and Gorbunova, V. (2009). Hypersensitivity to contact inhibition provides a clue to cancer resistance of naked mole‐rat. Proceedings of the National Academy of Sciences of the United States of America 106, 19352‐19357. Sengupta, S., Vonesch, J.L., Waltzinger, C., Zheng, H., and Wasylyk, B. (2000). Negative cross‐talk between p53 and the glucocorticoid receptor and its role in neuroblastoma cells. The EMBO journal 19, 6051‐6064. Sesti, F., Tsitsilonis, O.E., Kotsinas, A., and Trougakos, I.P. (2012). Oxidative stress‐mediated biomolecular damage and inflammation in tumorigenesis. In vivo 26, 395‐402. Shahbazian, D., Roux, P.P., Mieulet, V., Cohen, M.S., Raught, B., Taunton, J., Hershey, J.W., Blenis, J., Pende, M., and Sonenberg, N. (2006). The mTOR/PI3K and MAPK pathways converge on eIF4B to control its phosphorylation and activity. The EMBO journal 25, 2781‐2791. Shapiro, L. (2001). beta‐catenin and its multiple partners: promiscuity explained. Nat Struct Biol 8, 484‐487. Sharma, S., Sharma, P., and Tyler, L.N. (2011). Transfusion of blood and blood products: indications and complications. American family physician 83, 719‐724. Shenoy, N., Vallumsetla, N., Rachmilewitz, E., Verma, A., and Ginzburg, Y. (2014). Impact of iron overload and potential benefit from iron chelation in low‐risk myelodysplastic syndrome. Blood 124, 873‐881. Shokolenko, I.N., Wilson, G.L., and Alexeyev, M.F. (2014). Aging: A mitochondrial DNA perspective, critical analysis and an update. World journal of experimental medicine 4, 46‐57. Shuai, X., Zhou, D., Shen, T., Wu, Y., Zhang, J., Wang, X., and Li, Q. (2009). Overexpression of the novel oncogene SALL4 and activation of the Wnt/beta‐catenin pathway in myelodysplastic syndromes. Cancer genetics and cytogenetics 194, 119‐124. Shukron, O., Vainstein, V., Kundgen, A., Germing, U., and Agur, Z. (2012). Analyzing transformation of myelodysplastic syndrome to secondary acute myeloid leukemia using a large patient database. American journal of hematology 87, 853‐860. 216

Shy, B.R., Wu, C.I., Khramtsova, G.F., Zhang, J.Y., Olopade, O.I., Goss, K.H., and Merrill, B.J. (2013). Regulation of Tcf7l1 DNA binding and protein stability as principal mechanisms of Wnt/beta‐catenin signaling. Cell Rep 4, 1‐9. Sibon, D., Cannas, G., Baracco, F., Prebet, T., Vey, N., Banos, A., Besson, C., Corm, S., Blanc, M., Slama, B., et al. (2012). Lenalidomide in lower‐risk myelodysplastic syndromes with karyotypes other than deletion 5q and refractory to erythropoiesis‐stimulating agents. British journal of haematology 156, 619‐625. Siddique, A., and Kowdley, K.V. (2012). Review article: the iron overload syndromes. Alimentary pharmacology & therapeutics 35, 876‐893. Sierra, J., Perez, W.S., Rozman, C., Carreras, E., Klein, J.P., Rizzo, J.D., Davies, S.M., Lazarus, H.M., Bredeson, C.N., Marks, D.I., et al. (2002). Bone marrow transplantation from HLA‐identical siblings as treatment for myelodysplasia. Blood 100, 1997‐2004. Silverman, L.R., McKenzie, D.R., Peterson, B.L., Holland, J.F., Backstrom, J.T., Beach, C.L., Larson, R.A., Cancer, and Leukemia Group, B. (2006). Further analysis of trials with azacitidine in patients with myelodysplastic syndrome: studies 8421, 8921, and 9221 by the Cancer and Leukemia Group B. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 24, 3895‐3903. Simko, M. (2007). Cell type specific redox status is responsible for diverse electromagnetic field effects. Current medicinal chemistry 14, 1141‐1152. Singer, S.T., Wu, V., Mignacca, R., Kuypers, F.A., Morel, P., and Vichinsky, E.P. (2000). Alloimmunization and erythrocyte autoimmunization in transfusion‐dependent thalassemia patients of predominantly asian descent. Blood 96, 3369‐3373. Sloand, E.M., Mainwaring, L., Fuhrer, M., Ramkissoon, S., Risitano, A.M., Keyvanafar, K., Lu, J., Basu, A., Barrett, A.J., and Young, N.S. (2005). Preferential suppression of trisomy 8 compared with normal hematopoietic cell growth by autologous lymphocytes in patients with trisomy 8 myelodysplastic syndrome. Blood 106, 841‐851. Sloand, E.M., Wu, C.O., Greenberg, P., Young, N., and Barrett, J. (2008). Factors affecting response and survival in patients with myelodysplasia treated with immunosuppressive therapy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 26, 2505‐2511. Smith, J., Tho, L.M., Xu, N., and Gillespie, D.A. (2010). The ATM‐Chk2 and ATR‐Chk1 pathways in DNA damage signaling and cancer. Adv Cancer Res 108, 73‐112. Solomon, E.I., Decker, A., and Lehnert, N. (2003). Non‐heme iron enzymes: contrasts to heme catalysis. Proceedings of the National Academy of Sciences of the United States of America 100, 3589‐3594. Song, G., Ouyang, G., and Bao, S. (2005). The activation of Akt/PKB signaling pathway and cell survival. Journal of cellular and molecular medicine 9, 59‐71. Song, S., Christova, T., Perusini, S., Alizadeh, S., Bao, R.Y., Miller, B.W., Hurren, R., Jitkova, Y., Gronda, M., Isaac, M., et al. (2011). Wnt inhibitor screen reveals iron dependence of beta‐catenin signaling in cancers. Cancer research 71, 7628‐7639. Sonntag, D.M., de Boer, J., Medvedovic, M., Baxter, C.S., LeMasters, G., and Talaska, G. (2004). Mutational biases associated with potential iron‐binding DNA motifs in rodent lacI and human p53 mutational databases. Mutat Res 550, 73‐88. Speicher, T., Siegenthaler, B., Bogorad, R.L., Ruppert, R., Petzold, T., Padrissa‐Altes, S., Bachofner, M., Anderson, D.G., Koteliansky, V., Fassler, R., et al. (2014). Knockdown and knockout of beta1‐integrin in hepatocytes impairs liver regeneration through inhibition of growth factor signalling. Nature communications 5, 3862. Spencer, J.A., Ferraro, F., Roussakis, E., Klein, A., Wu, J., Runnels, J.M., Zaher, W., Mortensen, L.J., Alt, C., Turcotte, R., et al. (2014). Direct measurement of local oxygen concentration in the bone marrow of live animals. Nature 508, 269‐273. Sportoletti, P., Grisendi, S., Majid, S.M., Cheng, K., Clohessy, J.G., Viale, A., Teruya‐Feldstein, J., and Pandolfi, P.P. (2008). Npm1 is a haploinsufficient suppressor of myeloid and lymphoid malignancies in the mouse. Blood 111, 3859‐3862.

217

Starczynowski, D.T., Kuchenbauer, F., Argiropoulos, B., Sung, S., Morin, R., Muranyi, A., Hirst, M., Hogge, D., Marra, M., Wells, R.A., et al. (2010). Identification of miR‐145 and miR‐146a as mediators of the 5q‐ syndrome phenotype. Nature medicine 16, 49‐58. Steensma, D.P. (2012). Historical perspectives on myelodysplastic syndromes. Leukemia research 36, 1441‐1452. Steensma, D.P., and Bennett, J.M. (2006). The myelodysplastic syndromes: diagnosis and treatment. Mayo Clin Proc 81, 104‐130. Steensma, D.P., and Gattermann, N. (2013). When is iron overload deleterious, and when and how should iron chelation therapy be administered in myelodysplastic syndromes? Best practice & research. Clinical haematology 26, 431‐444. Stelzer, G.T., Shults, K.E., and Loken, M.R. (1993). CD45 gating for routine flow cytometric analysis of human bone marrow specimens. Annals of the New York Academy of Sciences 677, 265‐280. Stiegler, G., Sperr, W., Lorber, C., Fabrizii, V., Hocker, P., and Panzer, S. (2001). Red cell antibodies in frequently transfused patients with myelodysplastic syndrome. Annals of hematology 80, 330‐333. Stoddart, A., Fernald, A.A., Wang, J., Davis, E.M., Karrison, T., Anastasi, J., and Le Beau, M.M. (2014). Haploinsufficiency of del(5q) genes, Egr1 and Apc, cooperate with Tp53 loss to induce acute myeloid leukemia in mice. Blood 123, 1069‐1078. Strom, S.S., Gu, Y., Gruschkus, S.K., Pierce, S.A., and Estey, E.H. (2005). Risk factors of myelodysplastic syndromes: a case‐control study. Leukemia 19, 1912‐1918. Sugahara, K., Rubel, E.W., and Cunningham, L.L. (2006). JNK signaling in neomycin‐induced vestibular hair cell death. Hear Res 221, 128‐135. Sullivan, M.T., Cotten, R., Read, E.J., and Wallace, E.L. (2007). Blood collection and transfusion in the United States in 2001. Transfusion 47, 385‐394. Suzuki, T., Tomonaga, M., Miyazaki, Y., Nakao, S., Ohyashiki, K., Matsumura, I., Kohgo, Y., Niitsu, Y., Kojima, S., and Ozawa, K. (2008). Japanese epidemiological survey with consensus statement on Japanese guidelines for treatment of iron overload in bone marrow failure syndromes. International journal of hematology 88, 30‐35. Sykes, S.M., Lane, S.W., Bullinger, L., Kalaitzidis, D., Yusuf, R., Saez, B., Ferraro, F., Mercier, F., Singh, H., Brumme, K.M., et al. (2011). AKT/FOXO signaling enforces reversible differentiation blockade in myeloid leukemias. Cell 146, 697‐708. Szumiel, I. (2014). Ionising radiation‐induced oxidative stress, epigenetic changes and genomic instability: the pivotal role of mitochondria. International journal of radiation biology, 1‐55. Taher, A.T., Viprakasit, V., Musallam, K.M., and Cappellini, M.D. (2013). Treating iron overload in patients with non‐ transfusion‐dependent thalassemia. American journal of hematology 88, 409‐415. Takami, T., and Sakaida, I. (2011). Iron regulation by hepatocytes and free radicals. Journal of clinical biochemistry and nutrition 48, 103‐106. Takatoku, M., Uchiyama, T., Okamoto, S., Kanakura, Y., Sawada, K., Tomonaga, M., Nakao, S., Nakahata, T., Harada, M., Murate, T., et al. (2007). Retrospective nationwide survey of Japanese patients with transfusion‐dependent MDS and aplastic anemia highlights the negative impact of iron overload on morbidity/mortality. European journal of haematology 78, 487‐494. Tartakovsky, B., Goldstein, O., Krautghamer, R., and Haran‐Ghera, N. (1993). Low doses of radiation induce systemic production of cytokines: possible contribution to leukemogenesis. Int J Cancer 55, 269‐274. Tcherkezian, J., Cargnello, M., Romeo, Y., Huttlin, E.L., Lavoie, G., Gygi, S.P., and Roux, P.P. (2014). Proteomic analysis of cap‐dependent translation identifies LARP1 as a key regulator of 5'TOP mRNA translation. Genes & development 28, 357‐371. Templeton, D.M., and Liu, Y. (2003). Genetic regulation of cell function in response to iron overload or chelation. Biochimica et biophysica acta 1619, 113‐124. Temraz, S., Santini, V., Musallam, K., and Taher, A. (2014). Iron overload and chelation therapy in myelodysplastic syndromes. Critical reviews in oncology/hematology 91, 64‐73.

218

Thangaraju, M., Rudelius, M., Bierie, B., Raffeld, M., Sharan, S., Hennighausen, L., Huang, A.M., and Sterneck, E. (2005). C/EBPdelta is a crucial regulator of pro‐apoptotic gene expression during mammary gland involution. Development 132, 4675‐4685. Theil, E.C. (2013). Ferritin: the protein nanocage and iron biomineral in health and in disease. Inorganic chemistry 52, 12223‐12233. Thoreen, C.C., Chantranupong, L., Keys, H.R., Wang, T., Gray, N.S., and Sabatini, D.M. (2012). A unifying model for mTORC1‐mediated regulation of mRNA translation. Nature 485, 109‐113. Thuret, I. (2013). Post‐transfusional iron overload in the haemoglobinopathies. Comptes rendus biologies 336, 164‐ 172. Tian, X., Azpurua, J., Hine, C., Vaidya, A., Myakishev‐Rempel, M., Ablaeva, J., Mao, Z., Nevo, E., Gorbunova, V., and Seluanov, A. (2013). High‐molecular‐mass hyaluronan mediates the cancer resistance of the naked mole rat. Nature 499, 346‐349. Toma, A., Fenaux, P., Dreyfus, F., and Cordonnier, C. (2012). Infections in myelodysplastic syndromes. Haematologica 97, 1459‐1470. Tomas, J.F., Pinilla, I., Garcia‐Buey, M.L., Garcia, A., Figuera, A., Gomez‐Garcia de Soria, V.G.G., Moreno, R., and Fernandez‐Ranada, J.M. (2000). Long‐term liver dysfunction after allogeneic bone marrow transplantation: clinical features and course in 61 patients. Bone marrow transplantation 26, 649‐655. Toyokuni, S. (2009). Role of iron in carcinogenesis: cancer as a ferrotoxic disease. Cancer science 100, 9‐16. Traverso, N., Ricciarelli, R., Nitti, M., Marengo, B., Furfaro, A.L., Pronzato, M.A., Marinari, U.M., and Domenicotti, C. (2013). Role of glutathione in cancer progression and chemoresistance. Oxidative medicine and cellular longevity 2013, 972913. Trovato, M., Torre, M.L., Ragonese, M., Simone, A., Scarfi, R., Barresi, V., Giuffre, G., Benvenga, S., Angileri, F.F., Tuccari, G., et al. (2013). HGF/c‐met system targeting PI3K/AKT and STAT3/phosphorylated‐STAT3 pathways in pituitary adenomas: an immunohistochemical characterization in view of targeted therapies. Endocrine 44, 735‐ 743. Tuljapurkar, S.R., McGuire, T.R., Brusnahan, S.K., Jackson, J.D., Garvin, K.L., Kessinger, M.A., Lane, J.T., BJ, O.K., and Sharp, J.G. (2011). Changes in human bone marrow fat content associated with changes in hematopoietic stem cell numbers and cytokine levels with aging. Journal of anatomy 219, 574‐581. Uranga, R.M., Katz, S., and Salvador, G.A. (2013). Enhanced phosphatidylinositol 3‐kinase (PI3K)/Akt signaling has pleiotropic targets in hippocampal neurons exposed to iron‐induced oxidative stress. The Journal of biological chemistry 288, 19773‐19784. Vadlakonda, L., Pasupuleti, M., and Pallu, R. (2013). Role of PI3K‐AKT‐mTOR and Wnt Signaling Pathways in Transition of G1‐S Phase of Cell Cycle in Cancer Cells. Frontiers in oncology 3, 85. Valcarcel, D., Adema, V., Sole, F., Ortega, M., Nomdedeu, B., Sanz, G., Luno, E., Canizo, C., de la Serna, J., Ardanaz, M., et al. (2013). Complex, not monosomal, karyotype is the cytogenetic marker of poorest prognosis in patients with primary myelodysplastic syndrome. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 31, 916‐922. Valent, P., and Horny, H.P. (2009). Minimal diagnostic criteria for myelodysplastic syndromes and separation from ICUS and IDUS: update and open questions. European journal of clinical investigation 39, 548‐553. Valent, P., Krieger, O., Stauder, R., Wimazal, F., Nosslinger, T., Sperr, W.R., Sill, H., Bettelheim, P., and Pfeilstocker, M. (2008). Iron overload in myelodysplastic syndromes (MDS) ‐ diagnosis, management, and response criteria: a proposal of the Austrian MDS platform. European journal of clinical investigation 38, 143‐149. Valko, M., Izakovic, M., Mazur, M., Rhodes, C.J., and Telser, J. (2004). Role of oxygen radicals in DNA damage and cancer incidence. Molecular and cellular biochemistry 266, 37‐56. Vamvakas, E.C., and Blajchman, M.A. (2009). Transfusion‐related mortality: the ongoing risks of allogeneic blood transfusion and the available strategies for their prevention. Blood 113, 3406‐3417. Varfolomeev, E.E., and Ashkenazi, A. (2004). Tumor necrosis factor: an apoptosis JuNKie? Cell 116, 491‐497.

219

Veeraraghavan, J., Natarajan, M., Aravindan, S., Herman, T.S., and Aravindan, N. (2011). Radiation‐triggered tumor necrosis factor (TNF) alpha‐NFkappaB cross‐signaling favors survival advantage in human neuroblastoma cells. The Journal of biological chemistry 286, 21588‐21600. Venneti, S., Felicella, M.M., Coyne, T., Phillips, J.J., Gorovets, D., Huse, J.T., Kofler, J., Lu, C., Tihan, T., Sullivan, L.M., et al. (2013). Histone 3 lysine 9 trimethylation is differentially associated with isocitrate dehydrogenase mutations in oligodendrogliomas and high‐grade astrocytomas. Journal of neuropathology and experimental neurology 72, 298‐306. Ventura, J.J., Cogswell, P., Flavell, R.A., Baldwin, A.S., Jr., and Davis, R.J. (2004). JNK potentiates TNF‐stimulated necrosis by increasing the production of cytotoxic reactive oxygen species. Genes & development 18, 2905‐2915. Verhoef, G., and Boogaerts, M. (1991). In vivo administration of granulocyte‐macrophage colony stimulating factor enhances neutrophil function in patients with myelodysplastic syndromes. British journal of haematology 79, 177‐ 184. Vitale, M., Papa, S., Mariani, A.R., Facchini, A., Rizzoli, R., and Manzoli, F.A. (1987). Use of poligonal windows for physical discrimination among mononuclear subpopulations in flow cytometry. Journal of immunological methods 96, 63‐68. Vogelstein, B., and Kinzler, K.W. (1993). The multistep nature of cancer. Trends Genet 9, 138‐141. Voso, M.T., Fenu, S., Latagliata, R., Buccisano, F., Piciocchi, A., Aloe‐Spiriti, M.A., Breccia, M., Criscuolo, M., Andriani, A., Mancini, S., et al. (2013). Revised International Prognostic Scoring System (IPSS) predicts survival and leukemic evolution of myelodysplastic syndromes significantly better than IPSS and WHO Prognostic Scoring System: validation by the Gruppo Romano Mielodisplasie Italian Regional Database. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 31, 2671‐2677. Walter, M.J., Ding, L., Shen, D., Shao, J., Grillot, M., McLellan, M., Fulton, R., Schmidt, H., Kalicki‐Veizer, J., O'Laughlin, M., et al. (2011). Recurrent DNMT3A mutations in patients with myelodysplastic syndromes. Leukemia 25, 1153‐1158. Wang, C.Y., and Knutson, M.D. (2013). Hepatocyte divalent metal‐ion transporter‐1 is dispensable for hepatic iron accumulation and non‐transferrin‐bound iron uptake in mice. Hepatology 58, 788‐798. Wang, F., Liu, H.M., Irwin, M.G., Xia, Z.Y., Huang, Z., Ouyang, J., and Xia, Z. (2009). Role of protein kinase C beta2 activation in TNF‐alpha‐induced human vascular endothelial cell apoptosis. Can J Physiol Pharmacol 87, 221‐229. Wang, J., Fernald, A.A., Anastasi, J., Le Beau, M.M., and Qian, Z. (2010). Haploinsufficiency of Apc leads to ineffective hematopoiesis. Blood 115, 3481‐3488. Wang, L.D., and Wagers, A.J. (2011). Dynamic niches in the origination and differentiation of haematopoietic stem cells. Nature reviews. Molecular cell biology 12, 643‐655. Wang, X., and Jiang, X. (2008). Post‐translational regulation of PTEN. Oncogene 27, 5454‐5463. Wang, Y., Kreisberg, J.I., and Ghosh, P.M. (2007). Cross‐talk between the androgen receptor and the phosphatidylinositol 3‐kinase/Akt pathway in prostate cancer. Current cancer drug targets 7, 591‐604. Warlick, E.D., Cioc, A., Defor, T., Dolan, M., and Weisdorf, D. (2009). Allogeneic stem cell transplantation for adults with myelodysplastic syndromes: importance of pretransplant disease burden. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation 15, 30‐38. Watanabe‐Okochi, N., Kitaura, J., Ono, R., Harada, H., Harada, Y., Komeno, Y., Nakajima, H., Nosaka, T., Inaba, T., and Kitamura, T. (2008). AML1 mutations induced MDS and MDS/AML in a mouse BMT model. Blood 111, 4297‐ 4308. Weiss, G. (2010). Genetic mechanisms and modifying factors in hereditary hemochromatosis. Nature reviews. Gastroenterology & hepatology 7, 50‐58. Weissman, I.L., and Shizuru, J.A. (2008). 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, 3543‐3553.

220

Wells, R.A., Leber, B., Buckstein, R., Lipton, J.H., Hasegawa, W., Grewal, K., Yee, K., Olney, H.J., Larratt, L., Vickars, L., et al. (2008). Iron overload in myelodysplastic syndromes: a Canadian consensus guideline. Leukemia research 32, 1338‐1353. West, A.H., Godley, L.A., and Churpek, J.E. (2014). Familial myelodysplastic syndrome/acute leukemia syndromes: a review and utility for translational investigations. Annals of the New York Academy of Sciences 1310, 111‐118. West, A.R., and Oates, P.S. (2008). Mechanisms of heme iron absorption: current questions and controversies. World journal of gastroenterology : WJG 14, 4101‐4110. White, S.R., and Dorscheid, D.R. (2002). Corticosteroid‐induced apoptosis of airway epithelium: a potential mechanism for chronic airway epithelial damage in asthma. Chest 122, 278S‐284S. Wicovsky, A., Muller, N., Daryab, N., Marienfeld, R., Kneitz, C., Kavuri, S., Leverkus, M., Baumann, B., and Wajant, H. (2007). Sustained JNK activation in response to tumor necrosis factor is mediated by caspases in a cell type‐ specific manner. The Journal of biological chemistry 282, 2174‐2183. Williams, K., Christensen, J., and Helin, K. (2012). DNA methylation: TET proteins‐guardians of CpG islands? EMBO reports 13, 28‐35. Wimazal, F., Sperr, W.R., Kundi, M., Vales, A., Fonatsch, C., Thalhammer‐Scherrer, R., Schwarzinger, I., and Valent, P. (2008). Prognostic significance of serial determinations of lactate dehydrogenase (LDH) in the follow‐up of patients with myelodysplastic syndromes. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO 19, 970‐976. Wong, P.C., Waggoner, D., Subramaniam, J.R., Tessarollo, L., Bartnikas, T.B., Culotta, V.C., Price, D.L., Rothstein, J., and Gitlin, J.D. (2000). Copper chaperone for superoxide dismutase is essential to activate mammalian Cu/Zn superoxide dismutase. Proceedings of the National Academy of Sciences of the United States of America 97, 2886‐ 2891. Wu, H., Coskun, V., Tao, J., Xie, W., Ge, W., Yoshikawa, K., Li, E., Zhang, Y., and Sun, Y.E. (2010). Dnmt3a‐dependent nonpromoter DNA methylation facilitates transcription of neurogenic genes. Science 329, 444‐448. Xiao, Y., Wang, J., Song, H., Zou, P., Zhou, D., and Liu, L. (2013). CD34+ cells from patients with myelodysplastic syndrome present different p21 dependent premature senescence. Leukemia research 37, 333‐340. Xing, R., Li, C., Gale, R.P., Zhang, Y., Xu, Z., Qin, T., Li, B., Fang, L., Zhang, H., Pan, L., et al. (2014). Monosomal karyotype is an independent predictor of survival in patients with higher‐risk myelodysplastic syndrome. American journal of hematology. Xu, W., Yang, H., Liu, Y., Yang, Y., Wang, P., Kim, S.H., Ito, S., Yang, C., Wang, P., Xiao, M.T., et al. (2011). Oncometabolite 2‐hydroxyglutarate is a competitive inhibitor of alpha‐ketoglutarate‐dependent dioxygenases. Cancer cell 19, 17‐30. Xu, Z., Choudhary, S., Voznesensky, O., Mehrotra, M., Woodard, M., Hansen, M., Herschman, H., and Pilbeam, C. (2006). Overexpression of COX‐2 in human osteosarcoma cells decreases proliferation and increases apoptosis. Cancer research 66, 6657‐6664. Yamaguchi, H., and Hung, M.C. (2014). Regulation and Role of EZH2 in Cancer. Cancer research and treatment : official journal of Korean Cancer Association 46, 209‐222. Yatmark, P., Morales, N.P., Chaisri, U., Wichaiyo, S., Hemstapat, W., Srichairatanakool, S., Svasti, S., and Fucharoen, S. (2014). Iron distribution and histopathological characterization of the liver and heart of beta‐thalassemic mice with parenteral iron overload: Effects of deferoxamine and deferiprone. Exp Toxicol Pathol 66, 333‐343. Yeh, S.P., Yang, Y.S., Yao, C.Y., and Peng, C.T. (2009). Iron chelation therapy for patients with myelodysplastic syndrome. Hemoglobin 33, 339‐345. Yilmaz, O.H., Valdez, R., Theisen, B.K., Guo, W., Ferguson, D.O., Wu, H., and Morrison, S.J. (2006). Pten dependence distinguishes haematopoietic stem cells from leukaemia‐initiating cells. Nature 441, 475‐482. Yoshida, K., Inoue, T., Nojima, K., Hirabayashi, Y., and Sado, T. (1997). Calorie restriction reduces the incidence of myeloid leukemia induced by a single whole‐body radiation in C3H/He mice. Proceedings of the National Academy of Sciences of the United States of America 94, 2615‐2619.

221

Yoshida, K., Sanada, M., Shiraishi, Y., Nowak, D., Nagata, Y., Yamamoto, R., Sato, Y., Sato‐Otsubo, A., Kon, A., Nagasaki, M., et al. (2011). Frequent pathway mutations of splicing machinery in myelodysplasia. Nature 478, 64‐ 69. Yu, W.M., Hawley, T.S., Hawley, R.G., and Qu, C.K. (2002). Role of the docking protein Gab2 in beta(1)‐integrin signaling pathway‐mediated hematopoietic cell adhesion and migration. Blood 99, 2351‐2359. Zacharski, L.R., Chow, B.K., Howes, P.S., Shamayeva, G., Baron, J.A., Dalman, R.L., Malenka, D.J., Ozaki, C.K., and Lavori, P.W. (2008). Decreased cancer risk after iron reduction in patients with peripheral arterial disease: results from a randomized trial. Journal of the National Cancer Institute 100, 996‐1002. Zelova, H., and Hosek, J. (2013). TNF‐alpha signalling and inflammation: interactions between old acquaintances. Inflamm Res 62, 641‐651. Zhang, J., Grindley, J.C., Yin, T., Jayasinghe, S., He, X.C., Ross, J.T., Haug, J.S., Rupp, D., Porter‐Westpfahl, K.S., Wiedemann, L.M., et al. (2006). PTEN maintains haematopoietic stem cells and acts in lineage choice and leukaemia prevention. Nature 441, 518‐522. Zhang, L., and Wang, S.A. (2014). A focused review of hematopoietic neoplasms occurring in the therapy‐related setting. International journal of clinical and experimental pathology 7, 3512‐3523. Zhang, S., Lin, Z.N., Yang, C.F., Shi, X., Ong, C.N., and Shen, H.M. (2004). Suppressed NF‐kappaB and sustained JNK activation contribute to the sensitization effect of parthenolide to TNF‐alpha‐induced apoptosis in human cancer cells. Carcinogenesis 25, 2191‐2199. Zhang, Y., Zhai, W., Zhao, M., Li, D., Chai, X., Cao, X., Meng, J., Chen, J., Xiao, X., Li, Q., et al. (2015). Effects of iron overload on the bone marrow microenvironment in mice. PloS one 10, e0120219. Zhou, D., Shao, L., and Spitz, D.R. (2014). Reactive oxygen species in normal and tumor stem cells. Advances in cancer research 122, 1‐67. Zhou, X.Y., Tomatsu, S., Fleming, R.E., Parkkila, S., Waheed, A., Jiang, J., Fei, Y., Brunt, E.M., Ruddy, D.A., Prass, C.E., et al. (1998). HFE gene knockout produces mouse model of hereditary hemochromatosis. Proceedings of the National Academy of Sciences of the United States of America 95, 2492‐2497. Zhou, Y., Yan, H., Guo, M., Zhu, J., Xiao, Q., and Zhang, L. (2013). Reactive oxygen species in vascular formation and development. Oxidative medicine and cellular longevity 2013, 374963. Zhu, Y.X., Braggio, E., Shi, C.X., Bruins, L.A., Schmidt, J.E., Van Wier, S., Chang, X.B., Bjorklund, C.C., Fonseca, R., Bergsagel, P.L., et al. (2011). Cereblon expression is required for the antimyeloma activity of lenalidomide and pomalidomide. Blood 118, 4771‐4779.

222

Figure 1.1 was adapted from Wang and Wagers, 2011 with permission (licensed content publisher: Nature Publishing Group, license number: 3383760590038).

Chapter 2 was published as follows: Chan LSA, Shapiro R, Buckstein R, Lin Y, Callum J, Chodirker L, Lee CD, Prica A, Lam A, Mamedov A, Wells RA, Initial Transfusion Intensity Predicts Survival in Myelodysplastic Syndrome. Leukemia & Lymphoma 2014 Oct; 55(10): 2296-2300. Permission was obtained from the publisher to include the content of the publication in this thesis (licensed content publisher: Informa Healthcare, license number: 3390251165486).

223