National Library of Canada

National Library Bibliothèque nationale of Canada du Canada Acquisitions and Acquisitions et Bibliographic Services services bibliographiques 395 Wellington Street 395. nie Wellingtoci Ottawa ON K1A ON4 Ottawa ON KIA ON4 Canada CaMda

The author has granted a non- L'auteur a accordé une kence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous papa or electronic formats. la forme de microfiche/film, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. ABSTRACT Hematopoiesis is an ordered process of differentiation and proliferation, leading to the generation of mature blood cells of multiple distinctive lineages from totipotent hematopoietic stem cells. Regulation of this impressive machinery occurs at multiple levels, involving both extrinsic and intrinsic regulators. The members of the Hox homeobox gene farnily of transcription factors, known for their roles in determination of cell identity and pattern formation in a variety of structures during embryogenesis, have recently come under scrutiny as important regulators of hematopoiesis. In normal hematopoietic cells Hox genes have been shown to be expressed at levels that Vary both as a function of the stage of differentiation of the cells, and of the specific Hox gene examined. Thus, some genes like HOXB3 and HOXB4 are preferentially expressed in the most primitive fraction of hematopoietic cells, and others like HOXAlO and HOXB9 show a broader range of expression with downregulation at later stages of hematopoietic differentiation. Furtherrnore, Our group has recently shown that retroviral overexpression of HOXB4 in murine hematopoietic cells causes selective expansion of primitive hematopoietic cells, most profoundly the totipotent hematopoietic stem cell (HSC), without altering either myeloid or lymphoid differentiation or predisposing to leukemia. Taken together, these results strongly implicate Hox genes in the regulation of early hematopoietic cells. The major objective of this thesis was to investigate further the possible roles of Hox genes in the regulation of hematopoietic cell growth and differentiation, and to determine whether these rotes are Hox gene-specific. The strategy taken was to independently engineer the overexpression of selected Hox genes in murine bone marrow cells. Two Hox genes were chosen, HOXB3 and HOXAIO, based on their divergent expression patterns in hernatopoietic cells. The subsequent effects of these manipulations on the proliferation and differentiation of various populations of myeloid and lymphoid cells were then analyzed in a transplantation model and various in vitro cultures. Overexpression of either of these two Hox genes was found to impact on u.. both the processes of proliferation and differentiation of hematopoietic cells, involving multiple lineages. Specifically , overexpression of HOXA 10 led to enhanced formation of megakaryocytic progenitors in vivo and in vitro, diminished numbers of macrophage and B lymphoid progenitors, and the generation of myeloid leukemias in a significant proportion of mice 5 to 8 months after transplantation. In contrast, overexpression of HOXB3 led to an almost complete block in the thymic production of CD4+CD8+ T lymphocytes accompanied by an expansion of yG-TCR+ thymocytes, impaired B lymphoid development, and enhanced myelopoiesis leading to a myeloproliferative disorder. The hematopoietic perturbations generated by overexpression of HOXB3 and HOXA 10 were thus strikingly different from each other and also from those previously reported for the overexpression of HOXB4, which did not detectably perturb myeloid, B or T cell differentiation but did induce expansion of myeloid and lymphoid progenitor cells, and enhanced up to 47-fold the regeneration of the HSC. This thesis work was also aimed at delineating further the effects of HOXB4 overexpression on the regenerative potential of HSC. For that purpose, the size and the clonal composition of the regenerated pool of HSCs in mice transplanted with bone marrow cells overexpressing HOXB4 was analyzed at various time points (16 to 52 weeks) after transplantation. These studies, in addition to confirming our initial observation that HOXB4 overexpression enhances the regeneration of HSC following transplantation, show that this effect is long-lasting, and that the expansion of HSCs appears both to be controlled and to involve multiple HSCs. Taken together, these results presented in this thesis add to the recognition of Hox genes as important regulators of hematopoiesiç. These results suggest a role for Hox genes in the regulation of proliferation of the most primitive hematopoietic cells, and in lineage-commitments and proliferation of myeloid and lymphoid progenitor cells. Furthermore, these results also point to Hox gene-specific roles in the regulation of hematopoiesis. iii TABLE OF CONTENTS

TlTLE ABSTRACT ...... ii TABLE OF CONTENTS...... iv LIST OF TABLES AND FIGURES...... viii ABBREVIATIONS...... xi ACKNOWLEDGMENTS...... xiv CHAPTER 1. Introduction 1.1 Overview ...... 1 1.2 Organization and regulation of the hematopoietic system...... 2 1.2.1 Developmental and hierarchal organization of hematopoiesis...... 2 1.2.2 Definition and properties of early hematopoietic cell types ...... 5 1.2.2.1 The totipotent hematopoietic stem ceIl (HSC) ...... 5 1.2.2.2 Properties of the HSC...... 10 1.2.2.3 Myeloid and erythroid progenitor cells...... 13 1.2.2.4 Lymphoid progenitor cells ...... 15 1.2.3 Regulation of hematopoiesis...... 17 1.2.3.1 Regulation of hematopoiesis by extemal factors ...... 18 1.2.3.2 Regulation of hematopoiesis by intracellular factors ...... 20 1.3 Homeodomain-containingproteins ...... 26 1.3.1 Classification. chromosomal organization and evolution of the homeobox genes ...... 26 1.3.2 Important conserveci properties of Hox genes ...... 30 1.3.3 Hox genes and embryonic development ...... 32 1.3.4 The structure and functional specificity of Hox proteins...... 34 1.3.5 Regulation of Hox gene expression...... 38 1.3.6 Target genes regulated by Hox proteins...... 41 . . 1.4 Hox genes and hematopoiesis...... 43 1.4.1 Hox gene expression in human and murine hematopoietic cell lines and human leukemias...... 43 1.4.2 Hox gene expression in normal hematopoietic cells...... 44 1.4.3 Hox gene functions in normal hematopoietic cells ...... 48 1.5 Thesis Objectives...... 52 CHAPTER 2 . Materials and methods 2.1 Generation of retroviruses and viral assays ...... 54 2.1.1 Recombinant retrovira1vectors ...... 54 2.1.2 Generation of viral producer cells ...... 54 2.1.3 Viral titering and helper virus assays ...... 55 2.2 Hematopoietic cell cultures and assays ...... 56 2.2.1 Mice...... 56 2.2.2 Viral infection of murine bone marrow cells...... 56 2.2.3 Transplantation of retrovirally transduced bone marrow ...... 57 2.2.4 In vitro clonogenic progenitor assays...... 57 2.2.5 Day 12 CFU-S assay...... 58 2.2.6 Competitive repopulating unit (CRU) assay ...... 58 2.2.7 Culturing of Sca+Lin-WGA+ cells...... 59 2.2.8 Morphological evaluations of bone marrow. spleen and peripheral blood cells from transplanted mice...... 59 2.3 Antibodies. flow cytometry and cell sorting ...... 60 2.4 Molecular analysis...... 62 2.4.1 Southem blot analysis ...... 62 2.4.2 Northem biot analysis ...... 63 2.4.3 Western biot analysis...... 64 2.4.4 cDNA generation. amplification and analysis...... 64 2.5 Statistical Methods...... 65 CHAPTER 3. Overexpression of HOXA10 in murine hematopoietic cells perturbs both myeloid and lymphoid differentiation and leads to acute myeloid leukemia 3.1 Introduction...... 67 3.2 Results...... 68 3.2.1 Retroviral-mediated transduction of HOXA10 to murine bone marrow cells...... 68 3.2.2 Altered colony formation in vitro of myeloid progenitor cells overexpressing HOXA1O ...... 69 3.2.3 HOXA10 overexpression increases the maintenance of day 12 CFU-S in vitro ...... 71 3.2.4 Expansion in vivo of myeloid progenitor cells overexpressing HOXA10 and their altered colony formation in vitro ...... 72 3.2.5 Overexpression of HOXA10 in vivo is not permissive for pre-B lymphoid colony formation...... 79 3.2.6 Acute myeloid leukemia arises in recipients of HOXAIO- transduced bone marrow cells...... 81 3.3 Discussion...... 86

CHAPTER 4 . Overexpression of HOXB3 in hemopoietic cells causes defective lymphoid development and progressive myeloproliferation 4.1 Introduction...... 92 4.2 Results...... 93 4.2.1 cDNA cloning of a HOXB3 transcript from hemopoietic cells and generation of a HOX03 retroviral vector ...... 93 4.2.2 Hemopoietic reconstitution of recipients of HOXB3-transduced bone marrow cells...... 93 4.2.3 Recipients of HOXB3-transduced celIs show increased granulopoiesis intheir bone marrow and spleen...... 97 4.2.4 Recipients of HOXB3-transcuded cells have defective T cell development...... 101 4.2.5 Altered 6 cell development by overexpression of HOXB3 ...... 105 4.3 Discussion...... 107

CHAPTER 5 . Enhanced polyclonal regeneration of hematopoietic stem cells overexpresing HOXB4 following bone marrow transplantation 5.1 Introduction...... 112 5.2 Resuks...... 114 5.2.1 Experimantal model...... 114 5.2.2 Enhanced regeneration of CRU cells in mice transplanted with HOXB4-transduced bone marrow cells...... 117 5.2.3 Polyclonal hematopoietic regeneration by HOXB4-transduced cells both at the levels of HSC and mature end cells...... 119 5.2.4 HOXB4-inducod expansion in vivo of clonogenic progenitor cells ... 123 5.2.5 Enhanced regeneration of HOXB4-transduced CRU cells is not reflected in the number of regenerated Scaf +Lin-WGA+ cells...... 126 5.2.6 Lack of HOXB4 effects on mature end cell output...... 127 5.3 Discussion...... 129 CHAPTER 6 General discussion and conclusion 6.1 Overexpression of Hox genes perturbs hematopoietic devetopment at multiple stages involving multiple lineages...... 135 6.2 Overexpression of Hox genes: phenotype vs function ...... 138 6.3 Hox genes and hematopoiesis...... 141 6.4 Future directions...... 143

CHAPTER 7 References...... 146

vii LIST OF TABLES AND FIGURES

Chapter 1 Figure 1.1 Schematic representation of the organization of the hematopoietic system ...... 3 Figure 1.2 Simplified representation of B and T cell development...... 16 Figure 1.3 Schematic representation indicating the position of essential function of some of the transcription factors known to be active . . in hematopoiesis...... 23 Figure 1.4 Alignrnent of vertebrate Hox complexes into paralogous groups and corn parison with Drosophila HOM-C...... 29 Figure 1.5 Schematic representaion of Hox proteins domain organization and their DNAbinding sequences...... 35 Figure 1.6 Expression of Hox genes in human bone marrow cells ...... 46

C hapter 3 Figure 3.1 Structure and expression of the HOXA10 retrovirus used in this study ...... 69 Figure 3.2 Effects of HOXA10 and HOXB4 overexpression on the relative frequency of various colony types generated in vitro, immediately following retroviral infection of bone marrow cells...... 70 Figure 3.3 Effect of overexpression of HOXA10 on the recovery of day 12 CFU-S in vitro ...... 72 Figure 3.4 Southem blot analysis of DNA isolated from hematopoietic tissue of HOXA10 and neo mice...... 74 Figure 3.5 Nortitern and Westem blot analyses to demonstrate high levels of viral derived HOXA10 messages and protein in reconstituted HOXA10 mice...... 75 Table 3.1 Body weight and hematological parameters in neo and HOXAIOmice...... 76 Table 3.2 HOXAlO mice have increased numbers of myeloid clonogenic progenitor celis when compared to neo control mice...... 77 Table 3.3 HOXAIO mice have greatly increased numbers of bone marrow megakaryocyte/blast and blast colony forming cells (CFC) and decreased numbers of unilineage macrophage CFC ...... 78 Table 3.4 Sca+Lin-WGA+ cells overexpressing HOXA10 have increased potential to generate megakaryocyte-containing colonies in vitro ..... 79

.** VIl1 Table 3.5 Overexpression of HOXA10 is into permissive for lymphoid pre-B colony formation ...... 80 Figure 3.6 Survival of HOM10 mice compared to that of neo control mice and mice transplanted with HOXB4-transduced bone marrow cells using Kaplan-Meyer curves ...... 82 Figure 3.7 Wright-Geimsa staining of cytospin preparations of HOXA10- transduced myeloid colonies (AC) and of hematopoietic tissues from a representative Ieukemic HOXA10 mouse (D-F) ...... 83 Figure 3.8 Southern blot analysis to detemine: (A) the intensity of the proviral signal in bone marrow, spleen and thymus of HOM10 and neo mice; and (6) the clonality of the leukemias in HOXA10 mice...... 85

Chapter 4 Figure 4.1 Structure and expression of the HOXB3 and neo control retroviruses used in this study ...... 94 Figure 4.2 Overview of experimental design ...... 95 Figure 4.3 Donor-derived hemopoietic reconstitution of HOXB3 and neo mice.. 96 Figure 4.4 Hemopoietic parameters of neo and HOXB3 mice...... 98 Figure 4.5 Flow cytometric analysis of various hematopoietic populations...... 99 Table 4.1 Evaluation by limiting dilution analysis of competitive long-term repopulating cells (CRU) in primary mice transplanted with HOXB3 or neo control transduced bone marrow cells, 18 weeks after transplantation...... , ...... 100 Figure 4.6 Cellularity of thymuses and their subpopulations in individual neo and HOXB3 mice...... 102 Figure 4.7 Cytoflourometric analysis of thymocyte subpopulations...... 103 Figure 4.8 RT-PCR analysis purified thymic su bpopulations...... 104

Chapter 5 Figure 5.1 Structure of the HOXB4 and control neo retroviruses and the experimantal outline...... 1 16 Table 5.1 CRU numbers regenerated in primary recipients of neo- or HOXB4- transduced bone rnarrow cells...... 1 18 Figure 5.2 Southem blot analysis of proviral integration patterns in primary neo and HOXB4 mice sacrificed 32 weeks (A) or 52 weeks (B) post transplantation and their secondary recipients from the CRU assay ...... 121 Figure 5.3 Northem blot analyses to detect expression of the transduced HOXB4 gene in bone marrow of HOXB4 mice ...... 123 Figure 5.4 Effects of HOXB4 overexpression on the expansion of myeloid and pre-6 colony forming cells following bone marrow transplantation.... 125 Table 5.2 Regeneration of Scal+Lin-WGA+cells in neo and HOXB4 mice...... 126 Table 5.3 Hematological parameters in HOXB4 and neo mica, 16-52 weeks post transplantation...... 1 27 Table 5.4 Absolute numbers of various phenotypically defined hematopoietic populations in neo and HOXB4 mice 32 weeks after transplantation...... 128

C hapter 6 Figure 6.1 Schematic representation of the effects observed on various hematopoietic lineages with overexpression of HOXB3, HOXB4 or HOXA10...... 137 ABB REVlATlONS aPTCR alfa beta T cell receptor bp base pair BFU-E burst forming unit-erythroid BFU-Mk burst forming unit-megakaryocyte bHLH basic helix loop helix BMT bone marrow transplantation BSA bovine serum albumin cDNA complementary deopyribonucleic acid y6 TCR gamma delta T cell receptor CFU colony forming unit CFU-E colony forming unit-erytthroid CFU-GEMM CFU- granulocyte, erythrocyte, macrophage, megakaryocyte CFU-GM CFU- granulocyte, macrophage CFU-Mk CFU-megakaryocyte

CFU-S colony forming unit-spleen , GY centiGray CLL chronic lymphocytic leukemia CRU com petitive repopulation unit

~PP decapentaplegic Dhh desert hedgehod €PO erythropoietin ES em bryonic stem (cell) FCS fetal calf serurn 5-FU 5-flourouracil G-SCF granulocyte-colony stimulating factor GM-CSF gransulcyte, macrophage-coIony stimulating factor HGF hematopoietic growth factor hh hedgehog HSC hematopoietic stem cell HTH helix tum helix HXM hypoxanthine-xanthine-mycophenoloc acid IAP intracistemal A particle IL interleukin Ihh indian hedgehog KCI potassium chloride LTC long-tem culture LTC-lC Iong-terni culture-initiating cell LTR long terminal repeat P-ME b-mercaptoethanol MIP-1a macrophage inhibitory protein-1 alpha mRNA messenger ribonucleic acid MSCV murine stem cell virus NCS newborn calf serum OSM oncostatin-M PC-G polycomb group genes PHA phytohemagglutinin pol II RNA polymerase II R-PE R-phycoerythrin RNA ribonucleic acid RU repopulation unit S.D. standard deviation SCCM spleen cell conditioned medium SDS sodium dodecyl sulfate Shh sonic hedgehog TGF- B1 transforming growth factor -beta 1 TPO thrombopoietin trx-G trithorax group genes U units I am very grateful to my supervisor Dr. Keith Humphries for the opportunity to do graduate training in his laboratory and for his enthusiastic support and tireless guidance throughout this project.

I would also like to thank Dr. Connie Eaves for her invaluable help and support throughout this project.

Thanks to al1 members of Keith's laboratory (Margaret, Patty, Jana, Rob, Cheryl, Jennifer, Sue, Nick, Thilo and Sharlene) for their help and many pleasant unforgettable moments. A very special thanks to Dr. Guy Sauvageau, with whom I have worked in close collaboration, for his genuine passion for science and al1 the stimulating discussions.

I would also like to thank Wieslava Dragowska, Gayle Thombury, Maya St-Clair, and Fred Jenson for their valuable assistance and Drs. Connie Eaves, Frank Jirik and Furnio Takei for serving on my graduate committee.

Finally, above al1 I would like to thank my parents for teaching me the value of education and knowledge and for their years of constant support and caring, and my husband and son for their love and understanding.

Contribution of collaborators to the work presented in this thesis.

In the studies presented in Chapter 3, Dr. Guy Sauvageau's contribution to the work involving construction and production of the retroviruses used, and cytological evaluation of hematopoietic cells derived from HOXA10 and neo mice, is gratefully acknowledged. Dr. Margaret R. Hough and Ms. Wieslawa Dragowska provided assistance with flow cytometric analysis and purification of cells, respectively.

xiv In the studies presented in Chapter 4, Dr. Guy Sauvageau's contribution to the work involving the construction and production of retroviruses used, cytological evaluation of hematopoietic cells derived from HOXB3 and neo mice, and flow cytometric analysis, and Dr. Margaret R. Hough's additional assistance with flow cytometric analysis, are gratefully acknowledged.

In the studies presented in Chapter 5, Dr. Margaret R. Hough and Dr. Guy Sauvageau provided valuable assistance with fIow cytometric analysis and cytological evaluation of hematopoietic cells derived from HOXB4 and neo mice, respectively. Chapter 1

Introduction

1.1 Overview

Hematopoiesis is the process by which blood cells of multiple distinct lineages are produced throughout life. The mature products of this process have highly specialized and essential functions including transport of carbon dioxide and oxygen by erythrocytes, blood clotting by platelets, mediation of innate immunity by macrophages, granulocytes and natural killer cells, and mediation of antigen-specific immunity by B lymphocytes (humoral) and T lymphocytes (cellular). Large numbers of these cells are found in the circulation and because most of these cells have relatively short life spans (a few days to weeks) they need to be constantly replenished. Thus for example it has been estimated that about 1 trillion mature blood cells are produced in adult man every day, including 200 billion erythrocytes and 60 billion neutrophilic leukocytes. The ultimate burden of the life-long production of these mature elements is carried out by a small population of primitive cells, predominantly residing in the bone marrow, which has the capacity for multilineage differentiation and self-renewal divisions i.e. is self-maintaining. The control of this highly complex and dynamic process occurs at multiple levels by both positive and negative regulators, which under normal conditions assure the regulated production of al1 types of blood cells to meet changing physiological needs.

As yet, very little is known about the genetic mechanisms that bring about the self- renewal or differentiation outcomes of early hematopoietic cell divisions, although much progress has been made in identifying a variety of cytokines that can regulate the cycling status of these cells. Accumulating evidence, from a number of recent studies, is now pointing to transcription factors such as SCUtal-1, Ikaros, AML-1 and

1 GATA-2 as key components of the intrinsic processes involved in Iineagc determination and the subsequent lineage-specific differentiation of earl: hematopoietic cells. Among transcriptional regulators is the Hox homeobox family a genes, initially described in the fruit fIy as genetic elements essential for specificatioi of cell identity and pattern formation of a number of embryonic structures. Strikingly Hox gene functions have been found to apply to most if not al1 multicellular organisms albeit with very different end results, reflecting the "flexibility" of Hox genes ii orchestrating tissue specification. Hox genes have recently been shown to bc expressed in normal hematopoietic cells, at levels that Vary as a function of the stagi of differentiation of the cells and of the specific HOXgene examined. By analogy O hematopoiesis to other developmental programs it was hypothesized that thi! differential expression pattern of Hox genes might reflect roles for Hox genes in tht regulation of hematopoiesis.

As a test of this hypothesis, the research described in this thesis was aimed a analyzing possible functional roles of Hox genes in hematopoiesis. Toward that en( the normal expression patterns of three Hox genes were disturbed by engineeriq their overexpression in murine bone marrow cells, using retroviral gene transfer. Tht subsequent effects of these manipulations on the behavior of hematopoietic cells bott in vitro and in vivo were then analyzed. This introduction is divided into three mail parts that review current understanding of three separated but related topics central tc this thesis. These are: (1) the organization and regulation of the hematopoietic system (2) the Hox homeobox genes; and (3) the involvement of Hox genes in hematopoiesis,

1.2 Organization and Regulation of the Hematopoietic System

1.2.1 Developmental and hierarchical organization of hematopoiesis

In mammals, hematopoietic cells originate from the mesodermal layer of the embryo The first known site of hematopoiesis is extra-embryonic, in the yolk sac blood islandr

(known as primitive hematopoiesis). A few days later, the appearance O' 2 hematopoietic cells in an intraembryonic site, the so-called aortic-gonadal mesonephros (AGM) region, has recently been described (Godin et al., 1993; Huyhl et al., 1995). Hematopoiesis then shifts to the fetal liver (definitive hernatopoiesis) anc finally to the bone marrow where hematopoiesis continues throughout adult iife (Zon 1995). Evidence favoring the AGM region as the source of definitive hernatopoiesi: rather than the extra-embryonic yolk sac, as previously thought, has recently beer reported (Cumano et al., 1996; Medvinsky and Dzierzak, 1996).

Potentiai for Self- Prolif- e ration O CRU Totipotent HSC

Lymphoid Mveloid

potential Day 12 progenitors CFU-S

colonvin vitro uni- J3-A \ I O O p.;:.uim I 1 ' I

Figure 1.1 Schematic representation of the organiration of the hematopoietic system

Some of the assays used to detect various populations of hematopoietic cells are listed in bold. HSC hematopoietic stem cells,: CRU, cornpetitive repopulation unit assay; LTC-ICI in vifruassay that detects cells that initiate long-term cultures; day 12 CFU-S, colony-forming-unit spleen cells that generate nodules on the spleen 12 days after inoculum. The relative potential for self-renewat and proliferation of various populations of hematopoietic cells is shown to the right

The cells of the hematopoietic system can been subdivided into myeloid anc lymphoid cells as their pathways are thought to diverge early on in hernatopoietic differentiation (Figure 1.1), although the existence of "mixed" lymphoid and myeloid pathway (macrophage and B lymphoid) has been described both in bone marrow and fetal liver cells (Cumano et al., 1992; Ohara et al., 1991). All of the myeloid cells are 3 produced in the bone marrow, which include the erythroid, granulocytic ant megakaryocytic lineages (Lichtman, 1981). The cells of the lymphoid lineages whict give nse to T and 6 lymphocytes and natural killer (NK) cells, however, are producec and/or found to varying degrees in bone marrow, spleen, thymus and lymph nodes.

The differentiation of al1 of these different cell types is a multistep process usualll spanning many cell divisions, which has allowed each lineage to be visualized as 2 series of distinct stages. Cells at these different stages have been distinguished b) differences in proliferation and differentiation potential, response to different regulator: and ce11 surface antigenic profiles (Figure 1.1). The most primitive stage is tha represented &y hematopoietic cells with unrestricted proliferation and differentiatior potentials. These primitive cells comprise a very small proportion of the iota population of hematopoietic cells and their ability to sustain life-long hematopoiesis it associated with a capacity to undergo self-renewal divisions and therefore avoic exhaustion of the stem cell pool. Cells arising from these stem celIs by the process 01 differentiation to specific Iineages, are commonly referred to as hematopoietic progenitors (Figure 1A). When these progenitor cells divide, they differentiate intc morphologically recognizable cells, with very limited proliferative potential (3-5 ce1 divisions), that are the immediate precursors of specific types of mature blood cells.

This hierarchical organization of hematopoiesis is now widely accepted, with the emergence of key questions of how this process is regulated and what are the actual changes in gene expression associated with the commitment of totipotent stem cells ta particular blood cells lineages. Pivotal to these studies is the availability of assays and purification schemes to quantitate and functionally characterize early hematopoietic cells. In the following section some of these methods will be discussed, as well as what they have taught us about the properties of these early cells. 1.2.2 Definition and properties of early hematopoietic eell types

1.2.2.1 The totipotent hematopoietic stem cell (HSC)

The existence of very primitive cells with self-renewal and lympho-myeloid differentiation potential was first clearly shown almost three decades ago. By using radiation induced chromosomal abnormalities and bone marrow transplantation, Wu et al. showed that both myeloid and lymphoid cells in transplanted mice could be derived from a common cell (Wu et al., 1968). This observation has since been confirmed by other approaches, including the use of retroviruses to introduce unique clonal markers by their random integration into the genorne, and hence demonstrated common proviral integrants in both lymphoid and myeloid cells in mice transplanted with retrovirally infected cells (Capel et al., 1989; Dick et al., 1985; Keller et al., 1985). The existence of the HSC is thus a well established concept however, the quantitation of these cells has been more problematic.

Quantitation of HSC - One of the first assays developed for primitive hematopoietic cells was the in vivo spleen-colony-foming unit (CFU-S) assay (Till and McCulloch, 1961). This assay detects cells that have the ability to form macroscopic nodules of hematopoietic cells on the spleen 8-12 days after their intravenous injection into myeloablated recipients. These CFU-S ceils were initially considered to be HSCs because they share many properties that are attributed to HSCs. These include their ability to generate large clonal populations (105-1 07 cells), containing progeny of multiple hematopoietic lineages, thus demonstrating the high differentation and proliferation potential of individual CFU-S (Becker et al., 1963; Wu et al., 1968). Importantly, many CFU-S were found to possess the ability for self-renewal as progeny CFU-S were frequently detected in spleen colonies by injection into secondary recipients (Siminovitch et al., 1963). Although many of these cells possess the ability to differentiate into erythrocytes as weli as al1 cells of the myeloid lineage, their lymphoid potential remained controversial (Wu et al., 1968; Lala and Johnson, 1978; Paige et al., 1979), until recently (Lepault et al., 1993). The validity of the CFU-S assa to detect HSC with long-term reconstituting potential was also complicated by th discovery of functional heterogeneity among cells detectable as CFU-SI with some c these cells capable of only unilineage differentiation and lacking demonstrable seIl renewal potential (Magli et al., 1982).

Further assays have been developed that are based on the operational definition c an HSC as a cell type that can sustain both myeloid and lymphoid differentiation i vivo for a prolonged period of time. With the use of such assays and techniques thi can physically separate functionally distinct primitive hematopoietic cells (see below) was demonstrated that most CFU-S cells in normal adult bone marrow are cornmittel myeloid progenitors, as they can be physically separated from more primitive cells witl long-term lympho-myeloid repopulating potential (Jones et al., 1990). ne ver the les^ the CFU-S assay can detect primitive progenitor cells and has thus played a key roll in the development of concepts of primitive cell organization and regulation.

Two of the most rigorous and now widely used assays to quantitate the murine HS( are those developed by Harrison et. al and Szilvassy et. al (Harrison et al., 1993 Szilvassy et al., 1990). Both of these are functional assays based on the ability of th( HSC to regenerate and sustain hematopoiesis in myeloablated or genetically anemil mice (i.e. WNVV) for long periods. Their designs, however, differ and so do, thei potential applications.

The method developed by Harrison (for review see Harrison et al., 1993) is basec on the principles of "competitive repopulation" and binomial distribution. In this assa: two populations of hernatopoietic cells, distinguished from each other by phenotypic genetic or biochemical markers, are compared in their ability to repopulate bot1 lymphoid and myeloid compartments for long periods after their transplantation inti myeloablated mice. The first of these populations is called the "cornpetitor" an( normally consists of a fixed number (usually 1-2x106) of fresh bone marrow cells tha serves as a standard for repopulating potential. The second population, the "donor", is of unknown HSC content. Various numbers of "donor" cells are then injected along with "competitor" cells into myeloablated mice and the relative contribution of the two populations to hematopoiesis then measured. The frequency of HSC, here called "repopulating units" (RU), in the "donoru population is then calculated by the formula RU=%(C)/(100-%), where the % is the measured percentage of "donor" hematopoiesis and C is the number of fresh "competitor" cells used. In contrast to the CRU assay which uses limit dilution principles (see below), this assay uses high cell numbers and thus reduces the deniand for short-term repopulation imposed on the "tested' population. This assay is therefore particularly suited for the cornparison of the relative numbers of RU in two different populations of hematopoietic cells (Harrison et al., 1993). Although highly rigorous, the potential drawbacks of this assay are its requirement for a relatively long incubation period (>3months).

The Cornpetitive Repopulating Unit (CRU) assay developed by Szilvassy et. al combines limiting dilution and "competitive repopulation" principles to quantitate absolute frequencies of HSC, which go by the operational name CRU cells. As originaliy described, a population of cells being analyzed, the "test" cells, which are uniquely distinguished by genetically determined markers (Szilvassy et al., 1990; Rebel et al., 1994), are injected at various dilutions into multiple myeloablated recipients along with a fixed number of cells that sewe the purpose of ensuring short- term survival of the mice and provide a standardized level of "competition", independent of the "test" cell source. Recipients that are defined to be repopulated by at least one "test" cell derived CRU are those where a significant proportion (>5%) of their mature tymphoid and rnyeloid cells are of "test" cell origin. According to Poisson statistics, the number of "test" cells which has a 63% chance of repopulating a recipient would contain one CRU. The frequency of CRU cells in the "test" cell population can then be determined by estimating that value from the proportion of negative mice obtained for each "test" cell dilution (i.e., where a negative mouse is one that does not meet the repopulating criteria). Using this method the frequency of CRU cells in normal bone marrow has been estimated to be 1 per 104 cells (Szilvassy et al., 1990). CRU frequencies determined five weeks after transplantation have been shown to be virtually identical to those obtained at later time points (at least up to 6 months), indicating that this assay can be used to evaluate true long-term repopulating cells after little more than a month. Originally, a source of cells with relatively reduced long- term repopulating ability was used to ensure survival of the mice. However, more recently, it was found that 1 O5 normal adult mouse bone marrow cells could be substituted, with alterations of the sensitivity of the assay. (Rebel et al., 1994, see also chapter 4 and 5 of this thesis). The main advantages of this assay are that it allows rigorous but relatively quick (5-6 weeks) quantitation of HSC, and that since it is built around limiting dilution principles it allows quantitation of individual HSCs independent of their proliferative behavior (beyond the minimum threshold required for their detection).

In vitro assays to detect primitive hematopoietic cells have also been developed. The best characterized of these is the long-terni culture initiating cell (LTC-IC) assay, which detects cells that are capable of initiating and sustaining myeloid clonogenic progenitor production for at least 6 weeks under long-term culture conditions on a pre- established stroma1 layer (Sutherland et al., 1990). In hurnans this assay has thus far been the best available assay to quantitate the earliest hematopoietic cells. Recently, however, with the use of SClD mice the deveIopmant of an in vivo assay to quantitate human lympho-myeloid repopulating cells, has been proposed (Conneally et al., 1996).

Phenotyping and purification of HSC - Both the low frequency of HSC in hematopoietic tissue and the use of indirect functional assays as the only means to detect and quantitate these cells has driven interest in methods to physically isolate a pure population of HSCs as a way to quantitate HSC by phenotype and to faciliate their molecular and functional characterization. 8 Numerous strategies to enrich for these cetls have been developed. These include those that are based on physical differences between HSC and more mature cells (e.g. buoyant density andor size), which have led to the characterization of HSC as relatively small, low density cells with an undifferentiated blast cell-like morphology (Jones et al., 1990; Orlic et al., 1993). The most commonly used methods are, however, those that have combined the use of flow cytometry and the capacity of HSC to exclude the vital mitochondrial dye rhodamin-123 (Rh-123) (Ploemacher and Brons, 1988; Spangrude and Johnson, 1990), to bind lectins such as wheat germ agglutinin (WGA) (Ploemacher et al., 1993; Rebel et al., 1994; Visser et al., 1984) and by their expression of cell-surface antigens that can be recognized by monoclona[ antibodies (Morrison and Weissman, 1994; Spangrude et al., 1988; Szilvassy and Cory, 1993) Although some differences occur between different mouse strains (Spangrude and Brooks, 1993), the following surface phenotype is usually associated with the adult murine HSCs: they express high levels of the Ly-6NE (Sca-1) (Spangrude et al., 1988) and H-2K antigens (Szilvassy and Cory, 1993), and WGA (Rebel et al., 1994), low levels of the receptors c-kit (Katayama et al., 1993; Li and Johnson, 1995) and Thy-1 (Spangrude et al., 1988), and are largely negative for markers expressed on terminally differentiated cells (Lin-) such as 8220, CD3, CD4, CD8 (lymphocytes), Mac-1 , Gr-1 (myeloid) and f ER-1 19 (erythrocytes) (Spangrude et al., 1988). It is important to note that none of the known surface markers that are associated with HSC are restricted in their expression to HSCs, and as a result the stem ceIl surface phenotype is defined by the combination of those markers that are expressed and those that are not expressed on these cells.

Ce11 sorting based on these criteria, now allows the enrichment of murine HSC by 500 to 1000-fold (Morrison and Weissman, 1994). Thus for example, with the use of the CRU assay to quantitate the enrichment factor, the frequencies of HSCs in a population of cells with the Sca-l+Lin'WGA+ phenotype is 1 per 30 cells (Rebel et al., 1994) and frequencies up to 1 per 5 cells have been achieved with Sca-1+Thy- 1.l IoLin-Rhdutl cells (Spangrude et al., 1995). The enrichment factors stated above aF IikeIy an underestimation of the real frequencies of HSC in these purified population^ since they donnttake into account seeding efficiency (i.e. homing to appropriate site fc further growth) and their stimulation to generate mature blood cells as opposed t( them either remaining or becoming quiescent or executing predominantly self-renewz divisions.

Despite the significant advances in the purification of murine HSC, the applicatioi of the HSC phenotype has its restrictions. This is largely due to the fact that no surfaci markers that are exclusive to HSCs have been identified to date. This is bes exemplified by studies that compared the frequencies of HSCs in the bone marrot population with the candidate HSC phenotype, before and after either culture (Rebc et al., 1994) or transplantation (Spangrude et al., 1995) of these cells. 60th of thesi manipulations can result in great expansion of cells with the original HSC candidat( phenotype, but without concurrent expansion in HSC numbers as measured b: functional assays.

The avaitability both of assays to identify and quantitate HSC and of schemes tc purify these cells, has opened a way for answering some key questions regarding thc properties and regulation of HSC. These include their cycling status and utilization ii steady state hematopoiesis, their potential for self-renewal divisions and their possiblf heterogeneity as a function of source, some of which will be discussed in the followin! section.

1A2.2 Properties of the HSC

One characteristic that can distinguish HSC from more mature progenitor cells is thei cycling status. Under homeostatic conditions in vivo many HSC are quiescent O cycling very slowly, as demonstrated by their relative resistance to killing by tht cytotoxic drug 5-fluorouracil (5-FU),in contrast to the much higher proportion O progenitor cells killed by the samo treatment (Hodgson and Bradley, 1979; Lerner anc 10 Harrison, 1990). Interestingly, the 5-FU treatrnent also appears to recruit HSC intc active cell cycle, since they become highly sensitive to a second dose of 5-FU, giver 3-5 days later (Harrison and Lerner, 1 991 ) .

Several studies using retroviral marking of HSCs have demonstrated convincingl) the ability of HSCs to undergo self-renewal divisions both in vitro (Fraser et al., 1992: and in vivo (Jordan and Lemischka, 1990; Keller and Snodgrass, 1990; Lemischka e al., 1986). The potential of HSC to execute self-renewal divisions is, however, not fullg resolved. For example, given the right conditions do HSC have unlimited ability tc undergo self-renewal divisions or do these cells possess an inherer-it limit for suct divisions? In support of the latter possibility are observations that have shown thai following bone marrow transplantation the HSC pool is not found to regenerate tc levels >IO% of normal pre-transplantation values, despite regeneration of bon€ marrow cellularity and progenitor cell numbers to normal pre-transplantation levels (Harrison and Astle, 1982; Harrison et al., 1978; Harrison et al., 1990; Mauch and Hellman, 1989; Pawliuk et al., 1996). Even when mice are transplanted with a cell dose representing one half of the total bone marrow of a mouse, the HSC pool is no1 fully regenerated. Interestingly, no difference in HSCs content has been found between bone marrow cells from old (2 to 2.5 years) and young mice, suggesting thai if there are limits to HSC self-renewal, those limits are at least not reached in the normal life span of a mouse. Other studies using retroviral marking, have indicated thal at least some HSC can undergo a very high number of self-renewal divisions (Keller and Snodgrass, 1990; Pawliuk et al., 1996). Keller and Snodgrass showed that a single HSC could completely reconstitute al1 lineages of a primary recipient for more than 10 months and also almost completely that of a secondary recipient for another 17 months. Pawliuk et. al were able to put a numerical value on this process and showed that at least some HSC can be amplified -370-fold following transplantation. These studies therefore suggest that the substantial loss of marrow repopulating potential associated with transplantation is not simply a reflection of HSC fostering low inherent potential for self-renewal divisions, and that other constrains might be involved. One possibility is that as a consequence of HSCs regeneration of more differentiated cells under the conditions operative in the post-transplant recipients, it may result in loss of their subsequent ability to execute self-renewal divisions. Others have suggested (Pawliuk et al., 1996) involvement of negative regulatory feedback mechanisms imposed in vivo by more mature cells as a possible mechanism that could prematurely inhibit HSC expansion. These might involve the production of factors such as MIP-la and TGF-Pl which have been shown to decrease the proportion of primitive hematopoietic cells in cycle (Cashman et al., 1992; Eaves et al., 1991; Hatzfeld et al., 1991; Ruscetti et al., 1991). In the above discussion HSCs have been considered more or less as a homogeneous population. However, functionally defined HSCs appear heterogeneous by several criteria. HSCs are heterogeneous in their expression of surface markers. For example, the Sca-1+Thyl .lIoLin- cells which are highiy enriched for HSCs, can be further divided into subgroups based on differential expression of c-kit or CD4 antigens. Although the levels of HSCs in each of these subgroups are different, they al1 contain HSCs. Furthermore, HSCs are also found in other populations (e-g. Sca-1-Lin+), albeit at low frequencies (Rebel et al., 1994). HSCs in a given hematopoietic population are also likely heterogeneous in their proliferative potential. Experiments using retroviral tagging have revealed that some HSCs only contribute to hematopoiesis for the first -6 months post transplantation whereas others contribute for the duration of the recipient's life (Jordan and Lemischka, 1990; Keller and Snodgrass, 1990). Whether this heterogeneity represents true intrinsic differences in their proliferative potential or differences in the expression of this potential due to stochastic commitment versus self-renewal events, is, however, not clear. HSC from fetal liver have been shown to manifest both higher competitive repopulating and expansion potential than their adult bone marrow counterparts (Pawliuk et al., 1996; Rebel et al., 1996), indicating ontogeny related differences in the proliferative potential of HSC. Heterogeneity in the behavior of HSC from different mouse strains, has also beer documented. Transplantation experiments carried out with bone marrow cells fron allophenic mice created by mixing C57BU6 and DBA12 early embryonic cells demonstrated that cells of DBN2 origin dorninated in the early phase of regeneration whereas cells of C57BU6 origin dominated the late phase (Van Zant et al., 1992; Var

Zant et al., 1991). Subsequent experiments demonstrated that a larger proportion O progenitor cells from DBA12 mice were actively cycling and responded more rapidly tc Steel factor, whereas C57BU6 mice had a larger pool of candidate HSCs with z primitive phenotype (Rhdull). These differences might at least in part explain th< different engrafment pattern of these two populations of cells (Phillips et al., 1992) Genotypic differences in HSC frequencies have also been documented (Müller. Sieburg and Riblet, 1996). Taking advantage of this difference, it was shown b) genetic analysis that HSC frequencies are determined by multiple genes of which twc loci appear to have major quantitative effects (Müller-Sieburg and Riblet, 1996).

1.2.2.3 Myeloid and erythroid progenitor cells

A variety of more differentiated progeny of HSC can be detected by their ability to forrr colonies in semi-solid medium (usually containing either methylcellulose or agar: supplemented with appropriate nutrients and growth factors, either provided by poorl) defined "conditioned" cell medium or as pure recombinant growth factors (for review see Eaves, 1995). Different classes of these progenitor cells can be distinguished bj their growth factor requirements, the time required for their formation of a maturf colony and the types of mature cells that it generates. The period that elapses from th€ time of plating until morphologically mature cells are detected, and the number anc types of cells in each mature colony can be used to infer the developmental stage 01 the progenitor cell. Thus, large colonies containing cells from multiple lineages, originate from more primitive cells and small colonies of a single cell type from latei progenitors. These assays have played an instrumental role in the characterization ol cytokines that act at various stages of hematopoietic development. They have alsa 13 been central to studies aimed at dissecting the nature of the regulatory processe$ extrinsic versus intrinsic, that govern the survival, proliferation and differentiation c hematopoietic progenitor cells (see below 1.2.3.1, for more discussion)

The vast majority of progenitors detected in these cultures are uni- or bi-potentia These include progenitors that give rise to colonies which contain granulocytes an1 macrophages, and are called granulocyte-macrophage colony-forming units or CFU GM. These CFU-GM celIs have a broad range of proliferative potentials, as reflected il different sizes of the GM colonies that can be produced. More differentiated progeny c CFU-GM cells are detected in these cultures as CFU-G and CFU-M. Two erythroil restricted progenitors have been characterized: a more primitive progenitor, the bursl forming units-erythroid or BFU-E, and a more mature progenitor, the colony-formin! unit-erythroid or CFU-E. In the megakaryocyte lineage two progenitor types have alsi been identified, the more primitive BFU-Mk and more mature CFU-Mk. In steady statc hematopoiesis, the majority of these uni- and bi-potential progenitors appear to bi actively cycling as they are highly susceptible to killing by cytotoxic drugs such as 5 FU (Van Zant, 1984).

Progenitors that are more resistant to the effects of 5-FU and have rnultilineage an( high proliferative potential can also be detected in vitro by their colony formation Depending on the semi-solid medium and the growth factors provided, thesi progenitors are detected by their formation of colonies containing a mixture o granulocytes, erythrocytes, macrophages and megakaryocytes (CFU-GEMM (Humphries et al., 1981; Johnson and Metcalf, 1977); very large macrophage colonie: (high proliferative potential colony-forming cells, HPP-CFC) (Bradley and Hodgson 1979); or colonies consisting of undifferentiated blast cells (blast colony-forrning cells (Nakahata and Ogawa, 1982). Such progenitors can also exhibit lirnited self-renewa ability (Humphries et al., 1981; Nakahata and Ogawa, 1982). 1.2.2.4 Lymphoid progenitor cells

It has been proposed that both T and B cells originate from a common progenitor ce1 (Figure 1.1). However, the existence of such a ce11 type in the bone marrow has na been convincingly demonstrated. The strongest evidence for the existence of such i progenitor is an intrathymic cell population with the Thy-lloSca-2+c-kit+Lin-CD4lc surface phenotype, which has both T and B ce11 developmental potential, but vefl reduced myeloid potential (Matsuzaki et al., 1993; Wu et al., 1991; Wu et al., 1991).

Two in vitro assays that detect primitive B cell progenitors have recently beer developed (Saffran et al., 1992; Suda et al., 1989), but no such assays has yet beer described for progenitors with T lymphoid potential. Nevertheless, early stages of E and T cell development have been well characterized, mainly on the basis of th6 changes in expression of cell surface markers, and immunoglobulin (Ig) and T ce1 receptor (TCR) gene rearangements, that occure during these processes (Figure 1.2).

The various stages of B ceIl development have been characterized by the surface expression of developmetally regulated antigens, the rearrangement status anc surface expression of the lg genes, and by responsiveness to interleukin-7 (outlinec in Figure 1.2). These early steps in B ce11 development largely take place in the bon€ marrow, with further development of immature and mature B cells taking place in the psriphery (spleen and lymph nodes).

The main site of T cell developrnent is the thymus. The thymus is repopulated, as described above, with progenitor cells that have both T and B cell developmental potential (Figure 1.2). Following cornmitment of thymus repopulating cells to the T cell lineage, T cells develop through various stages characterized by expression of the CD4 and CD8 antigens, the TCR, the CD3 complex (the TCR associated complex), and a variety of other surface molecules such as CD44, heat stable antigen (HSA) and CD25 (IL-2 receptor a-chain (IL-2Ra)) (Figure 1.2) (reviewed in Godfrey and Zlotnik, 1993; Zlotnik and Moore, 1995)). The majority of thymocytes develop along the 15 pathway leading to the generation of cells expressing aB TCR afong with CD4 and CD8. A smaller proportion develops into T cells expressing y6 TCR and lacking expression of CD4 and CD8. These pathways are thought to diverge early in T cell development, at the so-called pre-T cell stage (Figure 1.2). The actual events that determine which pathway is selected are, however, not clear, but rnight involve transcriptional silencing of the TCR y genes in the ap T ceil precursors cells (Haas and Tonegawa,

A Immature Immature Pro-6 cells preB cells Pie-6 celk 6 cells Mature B cells

Surface B220+ B22h 8220+ B220+ B220+ antigens Thy-1 'O CD#+ CD#" I~M+ IgM+lgD+ or IgG+ Growth Stromal dependent Stromal independent properties IL-7 responsive

c-kit- + TCR aP/CD3+

Figure 1.2 Simplified representation of B and T cell development

(A) Various stages in 6 cells development that have been characterized based on expression of developmentally regulated antigens and the growth requirements of the cells Le. whether their growth is dependent on stroma1 cells andfor IL-7 or not. (8) Stages in T cell development as characterized by differential surface expression of number of antigens that lead either to the generation of ap T ceii receptor (TCR) T cells or y6 TCR T cells. DN. double negative; DP, double positive; and SP, single positive. 1.2.3 Regulation of hematopoiesis

As discussed above, the process of mature blood cell production must be tighttm controlled. Similarly, the entry of mature cells into the circulation, their localization tc the appropriate tissue, as well as their functional activation must also be regulatec Regulation of this complex process can be simplistically viewed as the combine4 effects of extemal influences (composed of both humoral factors and cell-cell or cell matrix interactions) and intracellular signaling events with consequent transcriptiona factor regulation, that ultirnately lead to changes in gene expression of multiple effecto molecules (Orkin, 1995). Although there is abundant evidence supporting the role a molecules from each of these "control levels" in the regulation of hematopoiesis, thei relative importance and interactions are not fully resolved.

The production of mature blood cells takes several cell cycles to complete. Thus there must be mechanisms in place at the single cell level that assure an appropriatr balance between the processes of differentiation, proliferation and maintenance of ce1 viability. Both viability and proliferation of hematopoietic cells are regulated by numbe of cytokines (for review see Ogawa, 1993). In vivo the size of a progenitor populatioi can thus be regulated by the availability of hematopoietic growth factors whict determine both the fate (survival or death) and amplification (by proliferation) of tht progenitor population. The more primitive that a given progenitor population is, tht greater is the impact of its altered behavior. Therefore fine tuning of mature cell outpu can thus be achieved by modulation in viability and proliferation of later cells wherea! more dramatic alterations in mature cell production (e.g. blood loss) would requiri altered behavior of more primitive cells. Although much progress has been made ir identifying a variety of cytokines that can regulate the cycling status of hematopoietic progenitor cells the genetic mechanisms that are responsible for their lineage commitment and differentiation are largely unknown. It has been proposed tha differentiation of hematopoietic cells, in contrast to regulation of their viability an( proliferation, is largely determined by intrinsic factors due to the observed apparen stochastic nature of this process (Fairbairn et al., 1993; Ogawa, 1993).

1.2.3.1 Regulation of hematopoiesis by external factors

Hernatopoietic growth factors (HGF) currently represent the most extensivel] described external regulators of hematopoiesis (see Metcalf, 1993 and Ogawa, 1993 for detailed review). To date more than 25 HGF have been identified. These includc the hematopoietic colony-stimulating factors (G-CSF, M-CSF and GM-CSF), thc interleukins (IL4 to IL-17), the hematopoietic inhibitors (TGF-P and MIP-1a), the "stem cell" factors (LIF, Steel factor and flk2fflt3 ligand), and erythropoietin (Epo) anc thrombopoietin (TPO). The regulation of hematopoiesis by HGFs has largely beer studied using either hematopoietic cell fines or the combined use of prima4 hematopoietic cells and the in vitro colony forming assays described above. In severa of these studies HGFs have been shown to be necessary for proliferation and surviva of hernatopoietic cells (Metcalf, 1993; Ogawa, 1993), whereas their effects on lineag~ commitment have only been demonstrated in limited studies (Metcalf, 1991). Th€ rnajority of the HGFs, with the exception of Epo and G-CSF, appear to hav~ considerable overlap in their function. This is evidenced both by the ability of differeni HGFs to support growth of the same type of progenitor cells and by the ability of a given HGF to act on different types of progenitors. For some of these HGFs the nature of this redundancy may be explained by their receptors sharing a common subunit, that is involved in the initiation of the intracellular signal. Thus, for example the IL-3, IL. 5 and GM-CSF share a common P subunit and the ability to stimulate eosinophil proliferation. Similarly, IL-6, IL-1 1, LIF and OSM which share pleiotropic activity also share a common P subunit in their respective receptors.

An interesting concept that has developed from these HGF studies, is that primitive hematopoietic progenitor cells can only be stimulated to proliferate in the presence 01 two (or more) HGFs, whereas more mature progenitors can proliferate in response to a single HGF although the presence of additional factors may have synergistic effects (Metcalf, 1993). Even though primitive hematopoietic cells cm be recruited into proliferation by the combination of various cytokines, net expansion of HSCs has been difficult to demonstrate (Bodine et al., 1992; Li and Johnson, 1994). These difficulties may imply the existence of as yet unidentified HGFs acting on HSCs. They have also raised the possibility that HSC self-renewal may predorninantly be regulated by other mechanisms, such as signaling through adhesion molecules and/or by some poorly understood "interna1 regulators".

The roles of some HGFs in hematopoiesis have also been studied in mice lacking the expression of a particular HGF or its receptor as a result of inactivation by gene targeting in embryonic stem cells. In mice where either Epo or the Epo receptor (EpoR) were inactivated, the generation of erythroid BFU-E and CFU-E progenitors was not affected, indicating that Epo-mediated signaling is not necessary for commitment to the erythroid lineage (Wu et al., 1995). However, terminal differentiation of these progenitor cells was blocked in these mice. Interestingly, this block could be overcorne in vitro by TPO, underscoring the redundancy between HGFs (Kieran et al., 1996). Of the HGF receptors known to date, one, the flWflt3 receptor, shows expression that is largely confined to primitive hematopoietic cells (Matthews et al., 1991; Palacios and Nishikawa, 1992). Homozygous mutant mice for the flk2/flt3 receptor are viable, and the only detectable hematological abnormalities in these mice are diminished numbers of pro- and pre-B cells (Mackarehtschian et al., 1995), indicating that the flWflt3 iigand is not essential or solely responsible for hematopoietic development in mice.

HGFs with the potential to inhibit primitive hematopoietic proliferation have also emerged as important regulators of hematopoiesis (Graham et al., 1990). Of these, TGFPl and MIP-la have been the best characterized, and were shown to decrease the proportion of primitive hematopoietic progenitors that are cycling, both in vitro and in vivo (Cashman et al., 1992; Eaves et al., 1991; Hatzfeld et al., 1991; Jacobsen et al., 1994; Lord et al., 1992).

The responses of hematopoietic cells to the external cues provided by their microenvironment thus depends on their repertoire of expressed HGF receptor types and other membrane-bound regulators, such as adhesion molecules. Upon activation, these receptors transmit signals (e.g. proliferativelinhibitory or survival), through signal transduction processes, which initiate a myriad of cellular responses including the activation of nuclear transcriptional regulators. Nuclear factors in tum activate multiple effector genes that elicit the appropriate cellular responses directed by the external stimulus. Clear connections have now been established between external factors and regulation of many nuclear factors involved in proliferation (e-g. c-fos, c-jun, c-myc), survival (c-myc) and mature end cell functions (e.g. STATs) of hernatopoietic cells. The roles of external factors in the regulation of nuclear factors involved in lineage- commitment of hematopoietic cells are, however, not clear. Nevertheless, whether regulated by extemal factors or by some poorly understood intrinsic mechanisms, transcriptional regulators are central to the processes of proliferation, differentiation and survival of hematopoietic cells. The next section will thus focus on the current understanding of the molecules involved in transcriptional regulation of hematopoiesis.

1.2.3.2 Regulation of hematopoiesis by intracellular factors

Appropriate transcriptional regulation of a given eukaryotic gene depends on contributions from a variety of factors. RNA polymerase II (pol II) is the enzyme responsible for transcription of the messenger RNA. For pol II to bind DNA and initiate transcription it must be part of a multimeric protein complex called the basal transcription machinery, which in addition to pol II contains general transcription factors called, TFIID, TFIIA, TFIIB, TFIIE, TFIIF, TFllH and TFIIJ. This complex binds to DNA at the core promoter (usually the TATA (TATAAA) box) which spans the DNA sequence from -35 to +35 relative to the transcription start site (for review see Emst and Smale, 1995). The ability of pol II to initiate transcription at a defined frequency is influenced by DNA-binding proteins called transcription factors. These factors bind to DNA at multiple regulatory regions, some of which are close to the basal transcription machinery (promoters) while others cm be very distant (enhancers), and can mediate either transcriptional activation or repression. These proteins modulate pol II activity by directly interacting with the basal transcription machinery, and their interactions can, in turn, be altered either by the presence of other transcription factors bound to nearby DNA at the same time or by interactions behveen these transcription factors (for review see Ernst and Smale, 1995).

Although the main focus in studies involving transcription regulation has been the transcription factors themselves, growing evidence indicates that other regulatory mechanisms invoiving chromatin structure and speciaiized chromatin eiements, ais0 play a crucial role (Emst and Smale, 1995; Grunstein, 1990; Peterson and Tamkun, 1995; Tamkun, 1995). In eukaryotes DNA is packed with proteins into a chromatin structure which can present a barrier against binding to DNA of both the basal transcription machinery and transcription factors. In S. cerevisiae, modulations of chromatin proteins (histones) have been shown to result in alterations in transcription (Felsenfeld, 1992). The most convincing evidence for a dynamic role of chromatin in the regulation of gene expression has corne from genetic and biochemical studies in Drosophila and S. cerevisiae. These studies have identified multiprotein complexes that can alter chromatin structure and either enhance (inS. cerevisiae the SWI-SNF complex, in Drosophila the trithorax group) or decrease transcriptional activity (in Drosophila the Polycomb group) (Peterson and Tamkun, 1995; Tamkun, 1995). In Drosophila and mice, both the trithorax and Polycomb group proteins have been implicated as important transcriptional regulators of the Hox homeobox genes (see for more detail in section 1.3.5). DNA cis-acting regions such as silencer and locus control regions have also been shown to be involved in transcriptional regulation which, at least in some cases, can be mediated through alteration in chromatin structur, (Felsenfeld, 1992; Sawada et al., 1994; Siu et al., 1994)

Transcription factors have been grouped into families according to the structure c their DNA-binding domain. Several classes of DNA binding domains have beei described including the zinc finger, basic leucine zipper, basic helix-loop-heli (bHLH), helix-tum-helix (HTH), paired, runt homology and many other domains (Ems and Smale, 1995). Some of these broad classes have been further subdivided. Fo example the majority of proteins in the HTH family constitute a subfamily o homeodomain proteins, which can be further divided into multiple subgroups basec on differences in their homeodomains and on additional consewed motifs within thesc proteins (see below in 1.3.1) (DeRobertis, 1994). In addition to the DNA bindin! domain, transcriptional activators contain a "transcriptional activation domain" whicl can interact with the basal transcriptional machinery. Transcriptional repressors conversely, can act either by steric hindrance mechanisms or by active inhibitor! mechanisms through protein-protein interaction with the basal transcription machine? or with transcriptional activators (Hanna-Rose and Hansen, 1996).

To date a number of transcription factors have been identified that play critical (non redundant) roles in hematopoietic cell lineage commitment, proliferation and survival The majority of these factors were originally identified either by their aberran expression in leukemias (Nichols and Nimer, 1992), or by their binding to cis* regulatory DNA sequences of lineage-specific genes (Georgopoulos et al., 1992; Tsa et al., 1989). Of these factors, none shows absolute specificity for hematopoietic tissu (Shivdasani and Orkin, 1996). Thus, like in other differentiation systems, transcriptior factors appear to regulate specific gene expression during hematopoietic development by their combinatorial actions.

The functional roles of these transcription factors in hematopoiesis have mainlp been investigated through their inactivation by gene targeting methods in embryonic stem cells, and in some cases by their forced overexpression. Several of these transcription factors are essential for normal hernatopoietic development, as their absence can lead to a block in normal hematopoietic development. Depicted in figure 1.3 are the positions of essential function for selected transcription factors involved in hematopoietic development, as suggested from the above mentioned studies. Sorne of these factors like the bHLH domain-containing tal-I/SCL and the LIM domain- containing rbtn2ILM02 proteins must act very early in hematopoietic development or be essential at later time points in multible lineages, as mice lacking these factors lack al1 lineages of both primitive and definitive hematopoiesis (Porcher et al., 1996; Shivdasani and Orkin, 1996; Warren et al., 1994). The tal-1/SCL and rbtnULM02 proteins have been shown to interact in vivo and form a heterocomplex (Valge-Archer et al., 1994; Wadman et al., 1994), which together with the similarities in the phenotypes of the ta/-VSCL and the rbtn21LM02 nuIl mice suggest, that they might act together in transcriptionai regulation. Both of these early acting transcription factors are also expressed at later stages of hematopoietic development, however, their functional role at these later stages could not be evaluated using this experimental system. None of the target genes regulated by either tal-1/SCL or rbtnZ/LM02 have been identified to date. GATA-I tal- I/SCL rbtn2R MO2 Erythrocytes (primitive) AML-1 Nf-€2 (definitive) Mega +Platelets karyocytes

Neutrophils GATA-2 Monocytes

T cell 8 cells

Figure 1.3 Schematic representation indicating the position of essential function of some transcription factors known to be active in hematopoiesis Positioning of each gene product is based on the earliest block observed in hematopoieis resulting from its absence. The various transcription factors are shown in italics. Adapted from (Shivdasani and Orkin, 1996).

Other transcription factors whose inactivation also affects multiple hematopoietic lineages include the AML-1, GATA-2, PU. 1, c-myb and lkaros proteins. A possible role for the runt homology domain AML-1 in the initiation of definitive hematopoiesis has been suggested, as mice lacking AM-1 have normal primitive hematopoiesis but completly lack definitive hernatopoiesis (Okuda et al., 1996). AML-1 has been implicated in the regulation of a number of both myeloid and lymphoid-specific genes (e.g. TCR, GM-CSF, M-CSF and IL-3), as the AML-1 core DNA binding motive has been shown to be essential for tissue specific expression of those genes (Shivdasani and Orkin, 1996). Absence of the zinc finger protein GATA-2 appears to impair the proliferation of the HSC rather than its differentiation (Tsai et al., 1994). This is evidenced by greatly reduced ability (as opposed to complete absence) of GATA-2 -1- cells to contribute to al1 hernatopoietic lineages. Mice that lack functional c-myb protein have a phenotype similar to that of the GATA-2-1- mice, with the exception of the megakaryocyte lineage which appears to develop normally in the absence of c-myb (Mucenski et al., 1991). 60th GATA-2 and c-myb are normally expressed in primitive hematopoietic cells and then downregulated as these cells differentiate (Thompson and Ramsay, 1995; Yamomoto et al., 1990). In contrast to the inactivation of GATA-; and c-myb, their forced overexpression in progenitor cells (c-myb in myeloid ant GATA-2 in erythroid progenitor cells), promotes their proiiferation and block! differentiation (Briegel et al., 1993; Yamomoto et al., 1990). Mice nul1 mutant for the et! family member PU. 1 lack cells of the granulocytic, monocytic and B cell lineages - th( cells in which PU.l is normally expressed (McKercher et al., 1996). Interestingly, 2 separate line of PU. 1 knockout mice, derived using a different targeting vector, has 2 more severe phenotype with additional defects in cells of the T and erythroid lineage:

(Scott et al., 1994). In monocytic and granulocytic differentiation a number of PU. ' target genes have been identified, and based on the known functional roles of thesc targets it appears that PU. 7 is crucial for terminal rather than early monocytic anc granulocytic differentiation (Olson et al., 1995). The lkaros gene gives rise to si) different proteins, by means of differentid splicing, that are differentially expressed ir lymphoid cells wherein they are thought to regulate the expression of a number O lineage specific genes (Georgopoulos et al., 1992). Mice lacking lkaros functior display a complete absence of al1 cells of the lymphoid lineages (T, B and NK) whereas both progenitor and mature cells of the myeloid and erythroid lineages were increased (Georgopoulos et al., 1994). Based on these results it has been proposec that lkaros acts as a developmental switch in HSCs, with Ikaros expression drivins lymphoid rather than myeloid differetiation.

ûther factors appear to be more lineage-specific in action such as GA TA-1, EKLF NF-E2, Pax5 and EZA, as their absence affects only one hematopoietic lineage (Nuer et al., 1995; Perkins et al., 1995; Pevny et al., 1991; Shivdasani et al., 1995b; Urbanek et al., 1994; Weiss et al., 1994; Zhuang et al., 1994) (Figure 1.3). GATA-1 is perhap: the best studied of the transcription factors known to be active in hematopoietic cells Absence of GATA-1 blocks hematopoietic differentiation at the proerythroblast stagc followed by apoptosis of these cells, suggesting that the functional role of GATA-1 is tc permit suwival and maturation of erythroid progenitor cells by preventing apoptosi3 (Weiss and Orkin, 1995). A role for GATA- 1 in lineage selection has also been suggested because its forced overexpression can reprograrn two different myelomonocytic cell lines, one into megakaryocytic differentiation (Visvader et al., 1992), and the other into megakaryocytic, erythrocytic or eosinophilic differentiation (Kulessa et al., 1995). Remarkably, in the latter case the intracellular level of GATA-1 affected the differentiation outcome (high megakaryocytic vs. low eosinophilic). The ability of one transcription factor to reprograrn the lineage cornmitments of hematopoietic cells lines, has suggested the existence of crosstalk between the regulatory networks involved in lineage choice. This idea is further supported by recent evidence that the transcription factor mafB (a member of the AP-1 superfamily), whose expression is restricted to myelomonocytic cells, can inhibit erythroid differentiation by interacting with Ets-1 and thereby repressing Ets-1 mediated gene activation of erythroid specific genes (Sieweke et al., 1996). In principle, mafB and GATA-7 may thus serve two complementary functions: GATA-1 would maintain a megakaryocytic/eosinophilic/erythroid phenotype and suppress a rnyeloid phenotype, whereas mafB would maintain a rnyeloid phenotype and suppress the erythroid phenotype. The phenotype of the lkaros nul1 mice would also suggest the existence of such lineage crosstalk, however, reprogramming of primitive cells with myeloid/erythroid potential to the lymphoid lineage by forced expression of Ikaros has not been demonstrated.

Our current knowledge of the transcription factors that are involved in regulation of hematopoiesis is far from complete. Several lines of evidence are now suggesting that factors that are highly consewed through evolution and play an important role in embryonic developrnent could also be active in adult tissues with continuing developmental potential (Krumlauf, 1994; Orkin, 1996). Among such genes are the homeodomain proteins, including the 39 members of the Hox homeobox farnily. The possible functional role for these genes in hematopoiesis had only recently corne under scrutiny at the time this thesis work was initiated. These initial studies had

26 revealed that many of these genes are expressed in hematopoietic cells lines and primitive subpopulations of normal hematopoietic cells (review in Lawrence et al., 1996). Such findings, and added evidence of function, prompted the work in this thesis to further resolve Hox gene roles in hematopoiesis.

1.3 Homeodomain-Containing Proteins

1.3.1 Classification, chromosomal organization and evolution of the homeobox genes The term homeobox gene arose from an earlier genetic term, the homeotic mutation, described some 102 years ago by Drosophila geneticists to qualify morphological variations transforrning something "... into the likeness of something else" (Bateson, 1894), as exemplified by generation of a leg in place of the antennae and halteres instead of wings (Lawrence and Morata, 1994). Many years later (1978), Lewis identified a homeotic gene complex in Drosophila, called Bithorax, that determined the development of middle and posterior body parts of the fruit fly (Lewis, 1978). Shortly thereafter another homeotic complex, Antennapedia, was similarly found to control segmentai development of the more anterior structure of the fruit fly (Kaufman, 1983). These complexes were both located on Drosophila chromosome 3, and by "chromosomal walking" on this chromosome the first homeotic gene called Antennapedia (Antp) was cloned (Garber et al., 1983). Using the cDNA of the Antp gene, other homeotic genes of the Antennapedia and Bithorax complexes were found to cross-hybridize to a highly-conserved repetitive sequence located at the 3' end of the Antp gene (McGinnis et al., 1984). Subsequently, this sequence was found to encode for a 60 arnino acid domain that was given the name homeodomain. In addition to being conserved among Drosophila homeotic genes, this domain was also found to be highly conserved among metazoans, including vertebrates (McGinnis et al., 1984; McGinnis et al., 1984; Scott and Weiner, 1984). The homeodomain forrns a helix-turn-helix DNA binding dornain, and the homeoproteins are found to act as transcription factors (Hoey and tevine, 1988; Levine and Hoey, 1988; Thali et al., 27 1988). With this new knowledge about homeotic genes, the concept of detemination during embryonic development became less abstract and more molecular, and suggested that the principles of genetic control of development, with some variations, could be applied to al1 multicellular organisrns (Lawrence and Morata, 1994).

Homeobox genes in insects and vertebrates can be divided into two broad categories: those that belong to the homeotic complex and are related to the Antp gene of Drosophila, and those which do not cause homeotic transformation and contain a divergent homeodomain. The latter group contains numerous genes that have been classified into more than 12 different classes and families including: Pax homeodomain genes, which contain a so-called paired box domain and homeodomain (Stuart and Gruss, 1996); POU domain genes that are defined by the presence of a bipartite DNA binding domain consisting of a 70 amino acid POU domain and a 60 amino acid POU-specific homeodomain and include members like the ubiquitously expressed Oct-1 protein and the pituitary-specific factor Pit-1, responsible for murine dwarfism (Rosenfeld, 1991); LIM-horneodomain genes, which contain two tandemly arranged cystein-rich LIM motifs and include the Lmxl gene that patterns the dorsal-ventral axis of developing vertebrate limbs (Sanchez-Garcia and Rabbits, 1994) ; Msx family of genes that are expressed at numerous sites in the developing mouse embryo and which are essential for teeth formation and for normal craniofacial bone development (Davidson, 1995); the exd/Pbx family originally identified for the involvement of one of its members (PBXI) in a translocation associated with human leukemia (Kamps et al., 1990) and later for their role as Hox gene CO-factors(see below) (Chan and Mann, 1996; Chang et al., 1996); and the Hlx class identified by their expression in hematopoietic cells (Deguchi et al., 1991).

In the fruit fly there are 8 homeotic homeobox genes, found in two separate clusters -the Antennapedia (ANT-C) and Bithorax (BX-C) complexes- that together form a larger complex termed Homeotic complex (HOM-C) (Figure 1.4) (Lawrence and Morata, 1994). To date 39 Antp like genes have been identified in vertebrates and they 28 are collectively called Hox genes (Krumlauf, 1994). These genes are organized in clusters, called A, B, C and Dl found on separate chromosomes, each containing 9-1 different Hox genes (Figure 1.4). The genomic structure and organization of Hox an( HOM4 share many intriguing similarities. In addition to organization into clusters, th( relative chromosomal position of HOM-CIHox genes is highly conserved, suggestin! that the Hox clusters arose by duplication of a common ancestral cluster, which is a yet, unidentified (Krumlauf, 1994). Drosophila, HOM-C Bithorax Antennapedia Abd-B Abd-A Ubx Antp Scr Dfd pb lab Verlebrates, Hox -1

HoxB HoxC HoxD -a-- Paralogous 13 12 11 10 9 8 7 6 54321 groups 5' 3' Late Low RA response ..mm.mm ...... High RA response Anterior

Figure 1.4. Alignment of vertebrate Hox complexes into paralogous groups anc cornparison with Drosophila HOM-C.

The Drosophila HOM-C is at the top, and the haçhed marks between Antp and Ubx indicate thc junction where the ANT-C and EX-C split. The 13 paralogous groups are labeled at the bottom. Example! of current nomenclature for mouse and human Hox genes is given for the parologous group gene 1 an( 13 in the Hox A cluster. In places where no gene is found, a gap is left. Shown is the vertical homolog! relationship that exists between the Drosophiia HOM-C and paralogous group Hox genes: group 1, labia (lab); group 2, proboscipedia (pb); group 4, Deforrned (Dfd); group 5, Sex combs reduced (Scr), an( groups 9-13, Abdominal 6 (Abd-6). At the sequence level, the homology between group 5 genes an( Scr is not strong, but there are strong functional similarities that are the bases for this assignment Antennapedia (Antp), Ultrabithorax, (Ubx) and Abdominal-B (Abd-6) genes have no vertebratt homologues. At the bottom is indicated the anterior-posterior, temporal and RA response colinearity ir Hox gene expression and the expression boundaries of selected Hoxü cluster genes along the A-P axi: of an early mouse embryo. Adapted from, Kmmlauf, 1994. Figure 1.4 summarizes the chromosomal organization and homology relationshipi between the four vertebrate Hox clusters and HOM-C genes. Alignments of genes are made on the basis of multiple domains of sequence identity, in addition to th€ homeodomain itself, and on the relative position of the genes within each complex There are 13 different sets of genes with shared properties, and they are termec paralogous groups 1 to 13 (Boncinelli et al., 1989). The nomenclature of Hox gene: was initially confusing since each gene was labeled according to its discovery. A new nomenclature based on the name of the cluster and the paralogous group that E particular gene is in, was coined at the third homeobox workshop (Scott, 1992). Th€ old and new nomenclature of each Hox gene has been reported elsewhere (Scott 1992) and only the new nomenclature is shown in Figure 1.4 and used in this thesis The names of human Hox genes are written with upper case letters, to b~ distinguishable from their murine homologs, which are represented by lower case letters (Figure 1.4).

Homeotic genes have also been extensively studied in other animals such as nematodes (C. eleganse) (Salser and Kenyon, 1994) and annelides (Leech) (Shankland, 1994) where they are found to share both functional and structura similarities with those of vertebrates and insects. Throughout this thesis homeotic genes will thus collectively be termed Hox genes.

1.3.2 Important conserved properties of Hox genes.

A distinguishing hallmark of the Hox complexes both in Drosophila and vertebrates is the correlation between the physical order of genes along the chromosomes and theii expression/function along the anterior-posterior (A-P) axis of the embryo. This property has been referred to as colinearity (Krumlauf, 1994). There is a spatial colinearity, which refers to the ordered array of spatially restricted expression domains along the A-P axis in embryonic tissue such as paraaxial mesodemi, surface ectoderm, neural tube and hindbrain segments (Burke et al., 1995; Gaunt, 1991; Graham et ai., 1989; Kessel and Gruss, 1991; Wilkinson et al., 1989). In vertebrates there is also a tempora colinearity based upon the time of appearance of expression during embryogenesi~ (Dekker et al., 1993; Izpisua-Belmonte et al., 1991). Retinoic acid is known to inducc vertebrate Hox gene expression, and there is a colinear sensitivity in the level anc time of responses of Hox genes to retinoic acid (Dekker et al., 1993). When these properties are al1 related to the clustered organization of the Hox complexes, the genes located at the extreme 3' end of each cluster ( i.e. paralog group 1 genes) are activated earliest, have the most anterior boundary of expression and display the highest sensitivity to retinoic acid (Figure 1.4). Paralog group genes with more 5' locations show progressively later expression, a more posterior location of theii anterior expression boundaries, and reduced response to retinoic acid. The conservation of this colinearity of Hox genes in evolution and in diverse vertebrate embryonic structures suggests that it is an important component for their function. The mechanisms underlying these expression patterns, however, remain poorly understood.

Another consewed property of Hox genes is an apparent functional hierarchy. This phenomenon was first discovered in Drosophila, where more posteriorly expressed genes appeared to suppress the effects of more anteriorly expressed genes. lnitially this effect, termed "phenotypic suppression", was thought to be caused by transcriptional regulation, as more posterior gene products can repress transcription of more anteriorly expressed genes (Struhl and White, 1985). However, this effect was later found to be postranscriptional, as Hox proteins expressed at high levels and ubiquitously (under the control of a heat shock promoter) only altered Drosophila development of body parts anterior to their normal expression domains (Gonzales- Reyes and Morata, 1990; Mann and Hogness, 1990). In vertebrates, a similar functional hierarchy is observed, where posteriorly expressed genes suppress the effects of Hox genes expressed more anteriorly, and is called "posterior prevalence". Evidence for existence of posterior prevalence in mice cornes from two types of studies: gain-of-function studies where ectopically expressed Hox genes predominantly gave phenotypes anterior to their normal endogenous expression domain, and loss-of-function studies which gave phenotypes only in a rostral part 01 the endogeneous expression domain of the inactivated gene (Duboule and Morata, 1994; Krumlauf, 1994).

1.3.3 Hox genes and embryonic development

The Hox genes have been most extensively studied during embryogenesis, where they determine cell identity and pattern formation of a variety of structures such as the anterior-posterior segmentation of early embryos; the limbs (wings, legs), the skeleton and the nervous system (Krumlauf, 1993; Krumlauf, 1994; Lawrence and Morata, 1994; Tabin, 1995).

Different animais initiate their embryonic development leading to anterior-posterior segmentation, in a diverse manner (Kenyon, 1994). No matter how they initiate development, as these embryos establish their body plans and begin to undergo morphogenesis a conserved Hox expression pattern appears along their anterior- posterior (A-P) axis, that parallels their relative 3' to 5' location in the Hox clusters (Figure 1.4). Because this Hox expression is position specific, one might imagine that this pattern would be established by localized cell-extrinsic signals. In Drosophila that is essentially what happens (Lawrence and Morata, 1994; St Johnston and Nusslein- Volhard, 1992). A molecular cascade initiated by a homeodomain transcription factor, bicoid, leads to localized expression pattern of the gap (e.g. hunchback, orthodental, giant, Kruppel and knirps) and pair-rule (e.g. hairy, paired and even-skipped) segmentation genes along the A-P mis. It is this expression pattern that allocates cells to the 14 A-P parasegments and activates different Hox genes in a 3' to 5' manner along the A-P axis. Like in flies the Hox gene expression in vertebrates is strongly influenced by the position of a cell within the embryo (Lawrence and Morata, 1994; Riddle et al., 1993). However, vertebrates do not appear to have homologs of many Drosophila gap and pair rule genes, suggesting that different molecules are involvec in setting up the initial Hox expression pattern in vertebrates (Kenyon, 1994) Interestingly, in Ieeches and C. elegans the initiation of Hox expression patterns alon! the A-P axis appears not to be position dependent, but rather determined by temporal and lineage-specific control systems, respectively (Cowing and Kenyon, 1996 Nardelli-HaefIiger et al., 1994).

Aiter parasegment formation in the Drosophila embryo, the control of Hox gene: within individual segments is tumed over to the segment polarity genes, which consis of the hedgehog (hh) and wingless (wg) signalling system, and decapentaplegic (dpp which is a member of the transforming growth factor P (TGFP) superfamily of gene:

(Lawrence et al., 1996; Lawrence and Morata, 1994; Petrimon, 1994). The products O the hh, wg and the dpp genes are secreted proteins that play key roles in instructin' cells about their fate within the parasegments and their funetional role appears to bt more widely conserved than that of the gap and pair rule genes (Patel et al., 1989 Shankland, 1994; Tear et al., 1990). A number of vertebrate homologs of these secreted proteins have been identified including, Sonic hedgehog (Shh), Deser hedgehog (Dhh) and lndian hedgehog (Ihh), which are the homologs of hh (Ingham 1995);the members of the Wnt gene family (to date 14 members have been identified which are the homologs of wg (Nusse and Varmus, 1992); and the bont morphogenetic proteins (BMPs) which comprise a large family (>20members) and arg the homologs of dpp (Hogan, 1996). These secreted proteins are expressed in man] vertebrate embryonic tissues, and a number of recent studies have indicated tha these proteins are the extracellular signals central to the formation and organization O a number of embryonic tissues (Chiang et al., 1996; Concordet and Ingham, 1995 Fietz et al., 1994; Hogan, 1996; McMahon and Bradley, 1990; Parr and McMahon 1995; Stark et al., 1994). In vertebrates, direct interactions betwen Hox genes anc these secreted proteins have not, yet, been demonstrated. However, in vertebrate limt development Hox genes have been shown to act in pathways that are both upstrearr and downstream of SHH signalling, which is essential for limb development (Chant€ et al., 1994; Cohn and Tickle, 1996; Niswander et al., 1994; Riddle et al., 1993).

In the Iast decade, the field of embryology has taken a major leap with the identification of many of the molecules and the principles applied in the formation of a variety of embryonic structures both in insects and vertebrates. These studies are indicating that many of these molecules as well as the principles applied, are shared, with some variations, both between different anirnals and different embryonic structures in the same species. Thus, for example, the known molecular steps in wing development in the fruit fly show a striking resernblance to those involved in vertebrate limb development (Lawrence and Struhl, 1996; Riddle, 1995; Vogel et al., 1995). Central to this genetic machinery are secreted molecules of the hedgehog, Wnt and BMPfamilies which act as morphogens, as well as transcription factors such as Hori and others, many of which contain homeodornains, which give sets of cells theii genetic addresses to determine their fates and responses to secreted morphogens (Lawrence and Struhl, 1996).

1.3.4 The structure and functional specificity of Hox proteins

In vitro DNA binding studies have shown that the homeodomain and its flanking sequences are essential for DNA binding and likely therefore the transcriptional activity of Hox proteins (Desplan et al., 1988; Hoey and Levine, 1988). The crystal structure of three homeodomains (from engrailed, MATa2 and Antp ) al1 complexed with DNA, have been characterized (Billeter et al., 1993; Kissinger et al., 1990; Wolberger et al., 1991). AI1 three structures were very similar to each other and showed that the homeodomain consists of three a-helices and an N-terminal arm thal is without secondary structure (Figure 1.5).

The first and the second helices are separated by a loop, and the second and third helices, together with a 4 amino-acid turn which separates them, form a motif similar to the helix-turn-helix structure found in prokaryotic transcriptional regulators. From in 34 vitro DNA binding studies Hox proteins have been shown to bind to a very similar 'core' DNA sequence (Figure 1.5) (Ekker et al., 1994; Ekker et al., 1991; Kalionis and OIFarrell, 1993). The homeodomain crystal structure studies showed that the homeodomain contact with this "core" DNA sequence is mediated both by the third a- helix, which sits in the major groove of DNA, and the N-terminal am that makes base specific contact in the minor groove (Figure 1.5).

NH2 2N-ami olhelix 1 loop turn a helix 2 a helix 3 I 1 T- N A-T- GiT- GIA Hox DNA "core" 5' binding sequence 3' T-G-A-T-N-N A-T- GK- GIA (1 (2) (3) (4) (5) (6) HoX/pbx(exd) DNA "tore" (7)(8) (9) (10) I binding sequence I minor groove major groove

Figure 1.5. Schematic representation of Hox protein domain organization and their DNA binding sequences.

Positions of conserved domains shared by al1 Hoxproteins in the same paralogous group (shaded) or by subsets of Hox genes (striped) are indicated. The structure of the homeodomain is shown and its contacts to DNA. "Core" DNA sequences are indicated which are preferred by Hox proteins alone or in complexes with pbdexd. Abbrevations: HP (T), conserved hexapeptide (tryptophan) motif.

During embryogenesis Hox proteins act with great biological specificity, presumably by their ability to regulate different sets of target genes. It has thus been one of the central issues in developmental biology to understand how Hox proteins achieve their functional specificity, and yet bind in vitro to a very similar 'core' DNA sequence with similar affinity (Ekker et al., 1994; Ekker et al., 1991; Kalionis and OIFarrell, 1993). Although this mystery has not been fully solved, recent molecular and genetic studies have started to give some answers.

In a number of studies, mouse or chicken Hox genes (i.e. Hoxb-1, Hoxb-6, Hoxb-9, Hoxb-4 and Hoxa-5) have been shown to be able to replace the function of their corresponding Drosophila homologs (i.e. lab, Antp, Abd-B, Dfd and Scr, respectively) , indicating a functional phylogenic conservation between Hox genes in the same paralogous group (Bachiller et al., 1994; Lutz et al., 1996; Malicki et al., 1990; McGinnis et al., 1990; Zhao et al., 1993). These studies and others involving chirneric or truncated Hox proteins (Chan and Mann, 1993; Furukubo-Tokunaga et al., 1993; Gibson et al., 1990; Lin and McGinnis, 1992; Mann and Hogness, 1990; Phelan et al., 1994; Zeng et al., 1993; Zhao et al., 1996), have shown that Hox genes functional specificity is in large part mediated both by homeodomain and by conserved motif N- terminal to the homeodomain, which in paralogous groups 1-8 is termed the hexapeptide (also pentapeptide or "YPWM") motif (Lutz et al., 1996; Zhao et al., 1993). This hexapeptide motif is not found in proteins from paralogous groups 9-13; however, a distinctive conserved tryptophan-containing motif is found at a similar location in proteins from paralogous groups 9 and 10 (Figure 1.5) (Chang et al., 1996). These studies also indicated that of the sequences in the homeodornain, the N-terminal arm is the most critical in mediating functional specificity. Taken together, these studies showed that Hox in vivo specificity is mediated both by conserved residues that directly bind DNA and also non-DNA binding residues, suggesting that protein-protein interaction could be important in mediating Hox functional specificity.

In 1990, a Drosophila gene called extradenticle (ex4 was identified by its ability, when mutated, to modify Drosophila Hox in vivo functions without altering the expression patters of Hox proteins (Peifer and Wieschaus, 1990). This suggested that exd could act as an Hox protein CO-factorthat might play a role mediating Hox specificity. Exd encodes a protein with a divergent homeodomain (see above), that has been highly conserved throughout evolution (Flegel et al., 1993; Rauskolb et al., 1993). In vertebrates the exd homologs are the pbx genes (pbxl, pbx2 and pbx3) which were originally identified from the involvement of the PBXl gene in the translocation t(1;19), frequently observed in chiid pre-8 acute lymphoblastic leukemia (Kamps et al., 1990; Nourse et al., 1990). Consistent with the CO-factor model, Drosophila and vertebrate Hox proteins were found to cooperatively bind DNA in vitrt with pbx or exd, and with higher affinity than when in monomeric form (Chan et al. 1994; Chang et al., 1995; Lu et al., 1995; Phelan et al., 1995; Popperl et al., 1995; var Dijk and Murre, 1994). These studies also identified that, in a majority of cases, th6 hexapeptide or the conserved tryptophan-containing motive of Hox proteins i: essential for successful cooperative binding (Chang et al., 1995; Lu et al., 1995 Phelan et al., 1995).

From a number of in vitro DNA binding studies (Chan et al., 1996; Chang et al. 1996; Lu and Kamps, 1996) and in vivo functional analysis (Chan et al., 1996; Popper et al., 1995) the following model has been proposed to describe how interaction: between exdlpbx and Hox proteins can contribute to Hox protein functioal specificity In this model the pbxiexd and Hox homeodomains are oriented as head-to-tai heterodimers with the centers of their binding sites only 4 bp apart. The N-terminai arrr of the Hox homeodomains is in the middle of the complex and interacts with a base pair in the minor groove. It is these two bases (5 and 6) (Figure 1S), that determine which pbdexd-Hox complexes are capable of binding with high affinity. Interestingly, the binding specificity of the Hox gene complexed with pbxiexd, correlates with its relative position in the Hox cluster; base nurnber 5 is increasingly preferred as G bp Hox proteins towards the 3' end of the clusters, as T at the 5' end and as A throughoui the middle of the cluster (Chang et al., 1996). The formation of a heterodimer compler between pbdexd and Hox proteins thus induces stable conformational changes in the Hox protein, which allows different Hox proteins to bind to with high affinity anc discriminate between DNA motifs with only 1-2 bp differences. Furthermore, th€ sequence specificity of a Hox protein monomer can be different from its specificity as heteromer with pbxlexd.

Interactions of Hox proteins with pbxiexd CO-factorsare unlikely to provide al1 Hon proteins with their functional specificity. Other CO-factorsare likely to be involved, such as the pbx-like protein Meis-1 which was recently found to be CO-activatedwith Hoxa-i 37 and Hoxa-9 in retroviral insertion-induced murine myeloid leukemias (Nakamura et al., t996b). Similarily, sorne studies have indicated that protein-protein interactior between Hox proteins, could also modulate their target selections (Zappavigna et al.! 1994). Another level of regulatory control may involve post-translational modifications in response to extracellular signals. Indeed, recently it has been shown that during Drosophiia embryonic rnidgut developrnent, the subcellular localization (nuclear vs. cytoplasmic) of exd is controlled by two secreted proteins, dpp and wg which themselves are thought to act as morphogens that direct the midgut developmeni (Mann and Abu-Shaar, 1996).

1.3.5 Regulation of Hox gene expression

The molecular mechanisms that regulate Hox gene expression are poorly defined. The striking feature of Hox genes is the conservation of their cluster organization and the relative position of each gene within the clusters. This, combined with the colinearity in their expression, suggests that their proper regulation is dependent on their position within the cluster (Krumlauf, 1994). In support of this idea it has been shown that different Hox genes can share promoters and that some of their regulatory elements are interspersed in the complex (Sirneone et al., 1988; van der Hoeven el al., 1996; Whiting et al., 1991). It was thus surprising to discover that for many HOA genes (Hoxa-4, Hoxa-7, and Hoxb-3) their proper expression pattern could be recapitulated using a small region of their respective complex in transgenic mice (Behringer et al., 1993; Puschel et al., 1990; Sham et al., 1992). However, in many cases (Hoxb-6, Hoxb-7, Hoxd-9 and Hoxd-11), the transgenic expression pattern did not fully reproduce the corresponding endogenous Hox expression (Eid et al., 1993; Gerard et al., 1996; van der Hoeven et al., 1996; Vogels et al., 1993). By using gene transpositions, where some 5' Hoxd genes along with their promoters were relocated within the Hoxd cluster, it was recently shown that Hox genes can be regulated both in a complex-dependent and in a gene-dependent manner (van der Hoeven et al., 1996). The cornplex-dependent mechanism appears to predominate early in 38 development to coordinate colinear Hox gene expression, but appears to give way to a more gene dependent mechanism later in development (van der Hoeven et al., 1996).

The factors that regulate Hox gene expression are still poorly defined. In Drosophila, as described earlier, transcription factors of the gap and pair rule families set up the initial Hox gene expression pattern, but a similar role for vertebrate homologs of these factors has not been established, despite repeated attempts (Kenyon, 1994; Krumlauf, 1994; Lawrence and Morata, 1994). In vertebrates, retinoic acid (RA) is a candidate for regulation of Hox genes (through its nuclear receptors), as Hox genes can respond to RA in a colinear fashion and RA can induce alterations in Hox expression in a wide variety of vertebrates in embryogenesis (Krumlauf, 1994). Furtherrnore, RA response elements have been found in Hoxa-1, Hoxb- 1 and Hoxd-4 genes, and most significantiy, this element is essentiai for normal Hoxb- 1 expression (Langston and Gudas, 1992; Marshall et al., 1994; Ogura and Evans, 1995; Ogura and Evans, 1995; Popperl and Featherstone, 1993).

Hox genes have also been shown to regulate their own expression in both auto- and a cross-regulatory manner (Bienz, 1994; Chouinard and Kaufman, 1991 ; Popperl et al., 1995; Zappavigna et al., 1994). During development of the hindbrain into segmented structures called rhombomeres, Hox genes are thought to play roles both in rhombomere segmentation and in segment identification (Lumsden and Krumlauf, 1996). There, in addition to RA, two candidate genes have been identified as regulators of Hox gene expression: kreisler, a b-Zip member of the c-maf proto- oncogene family (Frohman et al., 1993; McKay et al., 1994) and Krox-20, a zinc finger gene which directly regulates the activity of both Hoxa-2 and Hoxb-2 (Nonchev et al., 1996; Sham et al., 1993). However, in human hematopoietic cell lines with erythroid- megakaryocytic potential, HOXB2 is apparently not regulated by Krox-20, but here another zinc finger transcription factor, GATA-1, has been implicated in its regulation (Vieille-Grosjean and Huber, 1995). 39 The gap and pair rule genes that set up the Hox expression pattern along the A-F axis in Drosophila are only expressed transiently and thus distinct regulators arg

needed to rnaintain the Hox gene expression pattern. In Drosophila, two groups O genes of high genetic complexity have been described as such regulators: th€ Polycomb-group (Pc-G) and the trithorax-group (trx-G) (Simon, 1995). The Pc-C proteins form multimeric complexes that maintain transcriptional repression of HOJ genes initially turned off, presumably by induction of heterochromatin formatior (Jurgens, 1985; Landecker et al., 1994; Orlando and Paro, 1995; Simon, 1995). Ir contrast, the trx-G proteins, which also form multimeric complexes, maintain th€ expression of Hox genes initially turned on, by promoting an open chromatin structure by counteracting the repressive effects of chromatin components (Simon, 1995 Tamkun, 1995). Genetic studies in Drosophila have indicated that Pc-G and tm-G proteins are not only dedicated to Hox gene regulation, but rather are genera transcriptional regulators (Orlando and Paro, 1995).

In vertebrates, genes that are homologues of some of the Pc-G and trx-G genes have been identified. The Pc-G homologue bmi-1, first identified as an oncogene inducing murine T-cell lymphomas (Haupt et al., 1991; van Lohuizen et al., 1991), ha5 both been overexpressed and inactivated in transgenic mice (Alkema et al., 1995; var- der Lugt et al., 1994). The effects of these manipulations indicated that bmi-7 does negatively regulate Hox gene expression in vertebrates and interestingly severe hematopoietic defects were observed in the bmi-1 nuIl mutant mice. In vertebrates bmi-1 might have an additional role, as its oncogenic potential does not appear ta correlate with its transcriptional suppression activity (Cohen et al., 1996). Mel-18, another vertebrate Pc-G homologue has also been shown to negatively regulate transcription, but in contrast to bmi-1 has a tumor suppressor activity (Kanno et al., 1995). Several vertebrate homologs of the tnr-G genes have also been identified, one of which is the MLL (HRX or ALL-1) gene originally identified for its rearrangement in human acute leukemias (Gu et al., 1992; Tkachuk et al., 1992). Inactivation of the Mh gene in mice causes embryonic lethality and absence of Hox gene expression, and in MI1 heterozygous mice the expression boundaries of Hox genes were shifted posteriorly, consistent with a role for Mlj as a positive regulator of Hox gene expression (Yu et al., 1995). The heterozygous MI/ mice were documented to have some defects in erythroid and B lymphoid cells, which were however not fully characterized (Yu et al., 1995).

1.3.6 Target genes regulated by Hox proteins

The identification of Hox target genes is fundamental for Our understanding of how Hox genes function in regulating cell identity. However, this field of Hox research has not yet been very fruitful. In Drosophila, the best characterized targets are the Hox genes themselves (Bienz, 1994; Chouinard and Kaufman, 1991; Zheng et al., 1994) and this can be mediated by both direct and indirect pathways. The indirect pathways are frequently associated with the secreted proteins dpp and wg (see above), that can both mediate auto- (Ubx and /ab) and cross-(abdA -> Ubx and Ubx -> /ab) regulatory signals (Bienz, 1994). The dpp- and wg- mediated auto-regulation of /ab in the developing rnidgut was recently shown to involve the translocation of the Hox cofactor exd from the cytoplasm into the nucleus (Mann and Abu-Shaar, 1996). Interestingly, Hox proteins can also regulate dpp and wg expression. The Ubx protein has been shown to bind to dpp cis-regulatory elements and directly activate dpp expression ir! the developing midgut (Capovilla et al., 1994); for full dpp activation the exd protein was also necessary (Sun et al., 1995). The abd-A protein has also been implicated as a direct regulator of dpp, by suppressing its expression (Capovilla et al., 1994). Wg expression has alsr, been shown to be regulated by Hox proteins, but these interactions are thought to be indirect (Bienz, 1994).

In vertebrates, interactions of Hox genes with members of the BMP and the Wnt families (dpp and wg homologs) have not been well documented although the ability of a single homeodomain binding site in the regulatory region of the Wnt-7 gene to spatially restrict Wnt-1 expression in the developing brain has been reported (Iler el al., 1995). Vertebrate Hox genes have also been shown to auto- and cross-regulate each other (Popper1 et al., 1995; Zappavigna et al., 1994) and because of the conservation of BMP and Wnt members in evolution (see above), it is Iikely thai vertebrate Hox genes could also interact both directly and indirectly with the members of these two families.

Other known Hox targets in Drosophila, also have a function in cell-cell interaction. These including the connectine gene (Gould and White, 1992), which encodes a cell adhesion molecule involved in the innervation of muscles (Nose et al., 1992) and the scabrous gene, which appears to produce a secreted protein involved in cellulai communication during neurogenesis (Graba et al., 1992). Likely, vertebrate Hom targets also include those involved in cell-cell interaction. The mouse homologue 01 the Drosophila tumor suppressor gene 1(2)g1, which shows homology to the cadherin family of cell-adhesion molecules, is controlled in vivo by Hoxc-8 (Tomotsune et al., 1993). Others, such as Hoxb-8, Hoxb-9 and Hoxc-6 have been shown to bind to the promoter of the gene encoding the neural ce11 adhesion rnolecule (hl-CAM) and ta modulate its expression (Hirsch et al., 1990; Hirsch et al., 1991; Jones et al., 1993; Jones et al., 1992) and the Hoxd-9 was found to bind to the L-CAM enhancer (Goomer et al., 1994). Most recently the first vertebrate cytokine regulated by Hox proteins was identified where constitutive expression of HOXB7 in human melanorna cells was shown to directly activate basic fibroblast growth factor (FGF) (Care et al., 1996).

Null mutant mice have been generated for many Hox genes. An interesting feature of the phenotypes of these mutants is that they are less severe than that of the Drosophila nul1 mutants, and that what would be predicted based on their expression patterns (Krumlauf, 1994). This has tentatively been explained with the argument thaî paralogous genes (which can be up to four) (Figure 1.1 .) may compensate for the loss of one gene, due to both the highly similar structures and expression patterns 01 paralogous genes. To analyze possible interactions between paralogous genes, 42 embryonic development were not obvious in hematopoietic cell Iines, but rather that whole clusters (or large regions of a cluster) were turned on or off in a lineage-specific rnanner. Expression of several of the HOA,HoxB and HoxC cluster genes could also be detected in cell lines with B- or T-lymphoid potential (Lawrence et al., 1993; Petrini et al., 1992; Vieille-Grosjean et al., 1992), and the expression of the HOXC4 gene appeared to be restncted to lymphoid ce11 lines (Lawrence et al., 1993).

A number of studies have reported the expression of specific Hox genes in certain types of human leukemias. For example, HOXC4 expression is limited to lymphoid leukemias (Celetti et al., 1993; Lawrence et al., 1993), and conversely the HOXA10 gene is strongly expressed in myeloid leukemias but silent in lymphoid leukemias (Lawrence et al., 1995). A block of HOXB genes is also expressed in acute myeloid leukemias, but switched off in chronic myelogenous leukemias (Celetti et al., 1993).

1A.2 Hox gene expression in normal hematopoietic cells

The expression studies of Hox genes in hematopoietic cell lines were extended to normal human bone marrow cells, where expression of a number of Hox genes was detected using RNase protection, thus indicating that Hox gene expression was not simply a haltmark of transformed cells (Lawrence et al., 1993; Lowney et al., 1991; Mathews et al., 1991). Systematic analyses of Hox gene expression in different hematopoietic lineages and at different stages of hematopoietic differentiation was, however, hampered by the difficulties in obtaining high numbers of "pure", functionally different subpopulations of hematopoietic cells, and by the low levels of expression of Hox genes in normal hematopoietic cells. However, with the recent advances in fractionation of bone marrow cells into functionally distinct population using cell surface specific antibodies and flow cytometry (Lansdorp and Dragowska, 1992), combined with reverse transcription polymerase chain reaction (RT-PCR) procedure, more systematic analyses of Hox gene expression in normal hematopoietic cells became possible. Three laboratories have analyzed the expression of Hox genes in the fraction c human bone marrow cells that expresses the CD34 antigen (CD34+), which contain the majority if not al1 heniatopoietic progenitor cells, including the most primitiv hematopoietic cells (Sauvageau et al., 1994). These studies detected expression of of the 11 HOXA genes (Moretti et al., 1994; Sauvageau et al., 1994), 8 of the 9 HOXI genes (Giampaolo et al., 1994; Sauvageau et al., 1994), and 4 of 9 HOXC genes i CD34+cells, but expression of HOXD genes was not detected (Figure 1-6). In the mo: detailed of these studies, Sauvageau et al. further analyzed the expression of sever; of the HOXA and HOXB cluster genes in functionally distinct subpopulations of CD34 cells Le. those enriched for CFU-GM and BFU-E progenitors, and more primitive ster cells, the LTC-IC, where two patterns of expression were observed. Specific genez primarily located at the 5' end of each cluster (e.g. HOXA 10, HOXQ), showec essentially invariant expression in al1 subpopulations, whereas a second group c genes, located towards the 3' side of the ciusters (e.g. HOXB3, HOXB4, HOXAq, werc expressed at their highest levels in the LTC-IC enriched subpopulation and thel sharply downregulated in later cell populations (Sauvageau et al., 1994) (Figure 1.6) This study also suggested that genes located throughout the clusters, werc preferentially expressed in primitive bone marrow cells as expression of both HOXB, and HOXA 10 genes were virtually extinguished in the more differentiated CD34' ceIli (Figure 1.6). Furthermore, the apparent lineage-restricted expression of particular Ho. clusters observed in hematopoietic cell lines (Le. HOXB cluster genes in erythroid ce1 lines and HOXA cluster genes in myeloid cell lines), could not be detected in thi: study. Rather there appeared to be "reverse colinearity" in expression, where tht majority of Hox genes are expressed in the most primitive hematopoietic cells an( then downregulated as cells become progressively more differentiated, with the 3 genes being downregulated earlier than 5' located genes. HoxB

HoxD Paralogous 13 12 11 10 9 8 7 6 54321 groups

CD34- I CD34+ 1 CD% CFU-GM and BFU-E enriched LTC- IC enriclhed Mature cells Progenitor cells Stem cells

Figure 1.6. Expression of Hox genes in human bone marrow cells.

(A) Hoxgenes that are expressed in human hematopoietic cells are shown as black or striped boxes. (B) Graphic representation of the expression of two 3' and one 5' located Hox genes (shown as striped boxes in(A)) in subpopulations of human bone marrow cells that are enriched for functionally distinct cells. The expression levels are shown as relative to actin expression in each subpopulation and the expression of each of the Hoxgenes in the most primitive subpopulation, the LTC-IC enriched.

Much less data exists on the expression pattern of Hox genes in normal mutine hematopoietic cells. Expression of a number of Hoxa, Hoxb and Hoxc cluster genes has been demonstrated in the murine yolk sac (Palis et al., 1994). Expression of one of these genes, Hoxb-6, has been analyzed in more detail by RT-PCR, and which showed that it is also expressed in fetal liver, and in embryonic and adult CFU-E and BFU-E progenitor cells, but not in murine stem cell enriched cell fractions (Rich and Zimmermann, 1995). In the stem cell enriched fraction of human bone marrow (LTC- IC), HOXBG was one of the few Hox genes for which expression could not be detected, suggesting a conservation in Hox gene expression between human and murine hematopoietic cells (Sauvageau et al., 1994). This consenration is further supported by recent unpublished RT-PCR data where the expression of Hoxb-3, Hoxb-4 and Hoxa-10 also appears to be restricted to early cells, as their expression could only be 46 detected in the primitive Scaf +Lin- cells but not in more mature Scal-Lin+ cells, derived either from murine fetal liver or adult bone marrow cells (Pinneault and Humphries, unpublished).

Hox gene expression during the various stages of normal 6 and T lymphoid development has not been systematicalIy analyzed either in mice or humans. Human HOXC4 expression appears to be restricted to B and T lymphocytes, where it is activated during early to intermediate stages of both T and 8 cell developmenl (Lawrence et al., 1993). In mature resting 0, T and NK lymphocytes Hox genes are apparently not expressed with the exception of HOXB7 in CD8+ T cells (Carè et al., 1994; lnamori et al., 1993; Petrini et al., 1992). However, upon mitogenic activation (phytohemaglutinin (PHA) or IL-2AL1 stimulation) the expression of the HoxB cluster genes is activated both in T cells and NK cells (Carè et al., 1994; Petrini et al., 1992; Quaranta et al., 1996). Interestingly, this activation both in T and NK cells appears colinear as in embryonic development, with early activation of 3' located genes and then sequential later activation of 5' genes (Carè et al., 1994; Quaranta et al., 1996).

Lirnited information is now also available on the expression patterns of the known Hox CO-factors,pbx-1, -2 and -3, in hematopoietic cells. Expression of al1 three PBX genes has been documented in human fetal thymuses and spleens, and in human adult peripheral blood mononuclear cells, whereas in adult human thymuses PBX-2 and -3 are expressed but not PBX- 1 (Monica et al., 1991). Similarly, PBX- 1 expression has not been detected in human B and T lymphoid cell lines representing various developmental stages or in cell lines with monocytic potential, in contrast to expression of PBX-2 and -3 in al1 of these cell lines (Monica et al., 1991).

Other divergent homeobox genes are also expressed in hematopoietic cells. These include genes such as HLX (previously called HB24), first identified due to its expression in mitogen-stimulated human B lymphocytes and later in CD34+ human bone marrow cells, but not in more differentiated cells CD34- (Deguchi and Kehrl, 1991; Deguchi et al., 1991) and the HEX gene whose expression appears to be restricted to the hematopoietic systern (Bedford et al., 1993).

1.4.3. Hox gene functions in normal hematopoietic cells

Direct evidence for Hox gene function in hematopoiesis as a result of modulation of their expression, has only recently becorne available. Thus for example only four reports (see below) were published when the research presented in this thesis was initiated (Perkins et al., IWO; Shen et al., 1992; Takeshita et al., 1993; Wu et al., 1992).

Three studies have described the use of antisense oligonucleotides to down- regulate the mRNA levels of specific Hox genes in normal hematopoietic cells (Carè et al., 1994; Takeshita et al., 1993; Wu et al., 1992). Treatment of murine bone rnarrow with antisense nucleotides against Hoxb-7 lead to four-fold reduction in the formation of CFU-GM, whereas BFU-E and CFU-Mk were unaffected (Wu et al., 1992). Effects of antisense oligonucleotides directed against HOXC6 in human bone marrow were examined, and shown to suppress th8 formation of CFU-E without affecting either earlier BFU-E or myeloid progenitors (Takeshita et al., 1993). In mature T or NU cells, the HOXB cluster genes are activated in a 3' to 5' manner upon mitogen stimulation (see above), and treatment of these cells with antisense oligonucleotides against HOXB2 or HOX54 severely inhibits the proliferation of both the T and NK cells, suggesting an important role for Hox genes in proliferation of these cells (Care et al., 1994; Quaranta et al., 1996). In the NK ceIl study, HOXB gene induction was not observed when cells were treated with cytokines that either stimulated NK cell activation (IL-12) or survival (stem cell factor), further supporting the proliferative role of Hox genes in these cells. Potential drawbacks of the antisense oligonucleotide approach, however, are the uncertainty in the level of mRNA suppression (in some cases only -50%) and the specificity of the oligonucleotides used. Most recently, analyses of possible hematological defects in mice with targetec distributions of specific Hox genes have been initiated (Lawrence et al., 1996). Micc lacking a functional Hoxa-9 gene have defects in both granulocytic and lymphocytic pathways (Lawrence et al., 1996). These mice, which are othennrise healthy and fertile, have reduced numbers of peripheral blood granulocytes and lymphocytes, smallei spleens and thymuses, and reduced numbers of bone marrow myeloid and pre-El progenitor cells. However, their numbers of more primitive hematopoietic celk (Lea CFU-S and LTC-IC ), were not altered, indicating that absence of the Hoxa-9 affected hematopoiesis primarily at the level of the committed progenitors and not at earliei stages.

Expression of specific Hox genes has also been modulated by overexpression, either in hematopoietic cell lines or in normal hematopoietic cells. In the human erythroid cell line K562, overexpression of HOXB6 caused reduction in erythroid features as evidenced by decreased globin synthesis and surface glycophorin expression (Shen et al., 1992). The HOXB7 gene has been overexpressed in the human myelomonocytic HL-60 cell Iine (Lill et al., 1995). Normally HOXB7 expression is not detected in undifferentiated HL-60 cells; however, upon stimulation that induces monocytic differentiation its expression is readily detected, but not when induced to differentiate into the granulocytic lineage (Lill et al., 1995). Interestingly, overexpression of HOXB7 blocks granulocytic differentiation, whereas rnonocytic differentiation is unaffected (Lill et al., 1995). In the murine myelomonocytic leukemic cell line WEHI-36 transposition of endogenous retroviral like elements into both the Hoxb-8 and the IL-3 loci results in constitutive expression of both genes (Perkins et al., 1990). Since enforced expression of IL-3 alone does not render normal hematopoietic cells malignant but rather induces myeloproliferation (Wong et al., 1989), it was suggested that concomitant expression of Hoxb-8 with IL-3 might provoke the malignant phenotype of the WEHI-36 cell line. To analyze the possible role of Hoxb-8 in leukemogenesis, Hoxb-8 was overexpressed, either with IL-3 or alone, in murine bone marrow cells using retrovirus-mediated gene transfer (Perkins et al., 1990; Perkins and Cory, 1993). Concomitant overexpression of Hoxb-8 and IL-3 was fully transfoming, as mice transplanted with bone marrow celis that had been infected with retrovirus containing both genes developed fulminant polyclonal leukemias -3 weeks after transplantation (Perkins et al., 1990). Overexpression of Hoxb-8 alone, however, enhanced the self-renewal ability of myeloid progenitor cells as evidenced by theii enhanced replating ability and by the generation of non-tumorigenic myeloid cell lines in the presence of high concentrations of IL-3 (Perkins and Cory, 1993). Mice reconstituted with bone marrow cells overexpressing Hoxb-8 were free of leukemia foi at least 7 months, but -20% eventually developed myeloid leukemias, indicating thai Hoxb-8 alone was not fully leukemogenic. Interestingly, in some cases the leukemias were associated with activation or rearrangement of the IL-3 gene (Perkins and Cory, 1993).

As described in the last section, HOXB4 is one of the Hox genes found to be preferentially expressed in the most primitive subpopulation (LTC-IC enriched) of human bone marrow cells. Our group has recently retrovirally overexpressed this gene in murine bone marrow cells (Sauvageau et al., 1995). Serial transplantation studies revealed a greatly enhanced ability of HOXB4-transduced bone marrow cells to regenerate the HSC (here CRU) compartment, resulting in 47-fold higher numbers of CRU both in primary and secondary recipients, compared to serially passaged neo- infected control cells. This enhanced regeneration of HOXB4-transduced CRU, brought the CRU pool in primary recipients slightly above normal pretransplantation levels and that of secondary recipients near to normal levels. Myeloid and lymphoid pre-B clonogenic progenitor cells were also increased in recipients of HOXB4- transduced bone marrow cells (5- and 2-fold, respectively) compared to that of control neo mice. However, despite enhanced expansion of both CRU and clonogenic progenitor cells the relative numbers of the various types of in vitro myeloid clonogenic progenitors (CFU-GM, BFU-E and CFU-GEMM) in primary and secondary recipients of HUXB4-transduced bone marrow were the same as in recipients ot neo-control CeilS. In addition, total bone marrow and spleen cellularity of recipients of HOXB4- transduced bon8 marrow cells, as well as their peripheral blood differential counts, were al1 within normal range. Thus, despite a marked effect of overexpression of HOXB4 on the number of CRU and myeloid and lymphoid clonogenic progenitors, their was no gross effect on lineage determination not evidences of consequent expansion of later cell types.

Most recently reported during the course of this thesis, Hox genes have been directly implicated in the pathogenesis of human leukemias (Borrow et al., 1996; Nakamura et al., 1996a). In the t(7;11)(p1 $pl 5) translocation, which is recurrently observed in a subset of acute myeloid leukemias and in rare cases of chronic myeloid leukemias, the N-terminal half of the nucleoporin gene NUP98 is found to be fused in frame with most of the coding region of the HOXA9 gene. Intriguingly, murine Hoxa9 and its near neighbor Hoxa7 have also been implicated in myeloid leukemias associated with retroviral insertional activation in the BXH-2 mouse line (Nakamura et al., 1996b).

The functions of some of the divergent homeobox genes in hematopoiesis have also been analyzed. Enforced expression of the Hlx/HLX genes could modify the phenotype of several murine and human hematopoietic cell lines (Allen and Adams, 1993; Deguchi et al., 1992), as well as disturb T cell development in transgenic mice (Allen et al., 1995; Deguchi et al., 1993). Although Hlx nuIl mice suffer frorn severe embryonic anemia, they have no intrinsic defect in their hematopoietic cells, and, rather, the anemia is caused by an inadequate embryonic liver microenvironment (Hentsch et al., 1996). Like PBXI, another divergent homeobox gene, HOXI 1, was also identified for its involvement in a leukemia associated with a translocation (Hatano et al., 1991). In the t(10;14) translocation associated with human T cell acute lymphoblastic leukemia, the promoter of the T cell receptor 6 gene is juxtaposed to a region upstream of the HOXI 1 gene resulting in its ectopic expression. The murine 5 1 HOXl1 homolog, Tlx-1, has been overexpressed in murine bone marrow cells b! retroviral mediated gene transfer and generated IL-3 dependent irnmortal myeloid ce1 lines (Hawley et al., 1994). Interestingly, mice with targeted distribution of the Tlx- gene were asplenic (Roberts et al., 1994) which was later found to be caused b! apoptotic death of mesoderm-derived spleen progenitors, suggesting that Txl- 1 l; oncogenic potential could be due to enhanced cell survival (Dear et al., 1995).

1.S Thesis Objectives

As reviewed in previous sections, several lines of evidence are now pointing to Ho. genes as important regulators of growth and differentiation of hernatopoietic cells. Tht work presented in this thesis was initiated to test the hypothesis that individual Ho. genes may play unique roles in regulating of hematopoiesis and to gain furthe insights into the nature of these roles. The approach taken was influenced by two mail observations. The first one was derived from ouï initial demonstration, that retrovira overexpression of HOXB4 in murine hematopoietic cells can selectively enhance th( expansion of primitive cell populations, most profoundly the HSC, which suggestec that HOXB4 might be an important natural regulator of HSC regenerative potential The second was the apparent stage-dependent expression of Hox genes ir hematopoietic cells, with some genes like HOXB3 and HOXB4 being preferentiall! expressed in the most primitive subpopulation of human CD34+ hematopoietic cells and others IikeHOXAlO and HOXBS showing essentially invariant expression in al CD34+ subpopulations with downregulation at later stages of hematopoietic differentiation when cells becorne CD34'. Together these results raised the interestins hypothesis that specific Hox genes play distinctive roles in the regulation of differen aspects of hematopoiesis.

The first objective of this thesis was to test the hypothesis that different Hox gene: might have unique effets on the regulation of hematopoietic cell proliferation anc differentiation. The strategy taken was to engineer the overexpression in murine bon6 marrow cells of previousty untested Hox genes. Two such Hox genes were chosen HOXB3 and HOXA 10, based on their divergent expression patterns in hematopoietic cells (see above). Based on the known expression patters of these two genes it wa: hypothesized that overexpression of HOXB3 would primarily effect the properities 0 the most primitive hematopoietic cells, similar to HOXB4, whereas overexpression 0 HOXA 10 might effect a broader range of cell types. The subsequent effects of thes€ genetic manipulations on the proliferation and differentiation of various populations o. myeloid and lyrnphoid cells were then analyzed in a transplantation model anc various in vitro cultures. The results of these studies are presented in Chapter 3 and 4.

The second objective of this thesis work was aimed at delineating further the effects of HOXB4 overexpression on the expansion of HSC. In our initial studies WE demonstrated by serial transplantation that HOXB4-transduced bone marrow cells hac greatly enhanced potential to regenerate the HSC cornpartment resuiting in -50-folc higher numbers of HSC in both prirnary and secondary recipients compared to seriallp passaged neo-infected cells. The aim of the studies described in this thesis was tc determine the long-term effects of overexpression of HOXB4 on the HSC pool ir steady state hematopoiesis Le. whether these cells would continue to expand oi become exhausted, or if their expansion would be subjected to environmental contro mechanisms. For that purpose, the size and the clonal composition of the regeneratec pool of HSCs in mice transplanted with bone marrow cells overexpressing HOXB4 was analyzed at various time points (16 to 52 weeks) after transplantation. The results from these studies are presented in Chapter 5. Chapter 2

Materials and methods

2.1 Generation of Retroviruses and Viral Assays

2.1 .f Recombinant retroviral vectors

The human HOX54 cDNA used in the experiments presented in Chapters 3 and 5 was isolated from human fetal liver cells (Piverali et al., 1990); the human HOXA 10 cDNA used in experiments presented in Chapter 3 was isolated from the human myeloid ce11 line ML3 (Lowney et al., 1991); and the HOXB3 cDNA used in experiments presented in Chapter 4 was isolated from a cDNA Iibrary generated from CD34+ human bone marrow celIs (Sauvageau et al., 1997). These cDNAs were individually cloned, by blunt end ligation, into the murine stem cell virus (MSCV) 2.1 retroviral vector (Hawley et al., 1992) (kindly provided by Dr. R. Hawley; Sunnybrook Research Institute, Toronto, Ontario) at a polylinker site immediately 5' to a murine pgk promoter-neo cassette, using standard procedures (Davis et al., 1994). The HOXB4 cDNA, encompassing the complete coding sequence, was isolated as a 5amHl fragment from a plasmid (kindly provided by Dr. E. Boncinelli, Ospedale S. Faffaele, Milan, Italy) and subcloned at the Xbal site in the polylinker. The HOXAlO and HOXB3 cDNAs, encompassing the cornplete coding sequences, were isolated as EcoRl and Kpnl fragments, respectively, and subcloned at the Hpal site in the polylinker.

2.1.2 Generation of viral producer cells

The ecotropic cell line, GP+E-86 (Markowitz et al., 1988) and the amphotropic ceIl line, GP+ envAM12 (Markowitz et al., 19881, were used to generate helper-free recombinant retroviruses. These cell Iines were maintained in HXM medium which consists of Dulbecco's rnodified Eagle medium (DMEM; StemCeIl Technologies,

54 Vancouver, British Columbia), 10% heat-inactivated (55°C for 30 minutes) newbom calf serum (GibcolBRL Canada, Burlington, Ontario), hypoxanthine (15 mglml; Sigma Chemical Co., St. Louis, MO), xanthine (250 mg/ml; Sigma), and mycophenolic acid (25 mg/ml; Sigma). Purified plasmid vector DNA (10-14 pg), i.e. the MSCV 2.1 (control), MSCV 2.1 -HOXB4, MSCV 2.1-HOXA 10 or MSCV 2.1-HOXB3 were introduced into the GP+E-86 and the GP+envAM-12 packaging ce11 lines, using the calcium phosphate (Capo4) transfection technique. Virus-containing supernatants were hanrested 24-48 hours after the transfection, filtered, and then used to cross- infect the ecotropic and amphotropic packaging cells, transfected with the sarne plasmid vector, in the presence of 6 pgiml polybrene (Sigma). lnfected cells were then selected in 1 mg/mI of the neomycin analog G418 (GibcoIBRL), to obtain a polyclonal population of amphotropic or ecotropic viral producer cells. To increase the viral titer of these cells, filtered supernatant from ecotropic and amphotropic virus producing cells harboring the same retroviral construct were used to cross-infect these same cells 4-6 times. The viral producer cells were then maintained in HXM medium supplemented with 1 mghl G418.

2.1.3 Viral titeritig and helper virus assays

Viral titers of the GP+E-86-MSCV-pgk-neo, G P+E-86-MSCV-HOXB4-pg k-neo, GP+E- 86-MSCV- HOXA I O-pgk-neo and GP+E-86-MSCV- HOXB3-pgk-neo viral producer cells (geneïating viruses hereafter called neo, HOXB4, HOXA 10 and HOXB3, respectively) were determined by assaying various dilutions of filtered viral supernatant for the transfer of neomycin resistance to NIH-3T3 cells (American Type Culture Collection (ATCC), Rockville, MD) (Cone and Mulligan, 1984). The viral titers of the neo viral producer cells were 3-5x106 colony forming unit (CFU)/ml, and 3- 5x105 for HOXB4, HOXA 10 and MX83 viral producer cells. Absence of helper virus generation in the HOXB4, HOXA 10, and HOXB3 viral producer cells was verified by failure to serially transfer virus conferring G418 resistance to NIH-3T3 cell (Cone and Mulligan, 1984). 55 2.2 Hematopoietic Cell Cultures and Assays

Mice used as recipients were 7 to 12 week old male or female (C57B116J I C3HlHeJ)Fl ((B6C3)Fl) and used as donors were (C576116Ly-Pep3b x C3H/HeJ)FS ((PepC3)Fl) mice. The (B6C3)Fl and (PepC3)FI mice are phenotypicall] distinguishable by their cell surface expression of different allelic forrns of the Lyf locus; (B6C3)Fl are homozygous for the Ly5.2 allotype and (PepC3)Ft art heterozygous for the Ly5.1lLy5.2 allotypes. These mice were bred from parental strair breeders originally obtained from the Jackson Laboratories (Bar Harbor, MA) anc maintained in microisolztor cages and provided with sterilized food and acidified wate in the animal facility of the British Columbia Cancer Research Center.

2.2.2 Viral infection of murine bone marrow cells

Bone marrow cells used for retroviral infection were isolated by flushing femurs anc tibias of (PepC3)Fl (Ly5.1lLy5.2) mice, injected intravenously 4 days previously witl 150 mglkg body weight of 5-fluorouracil (5-FU), with DMEM 2% fetal calf serum (FCS: (StemCells Technologies), using a 21 gauge needle. Single cell suspensions of 1-5 105 bone marrow cellslml were then incubated in DMEM containing 15% FCS, 6 ngh murine interleukin-3 (mlL-3), 100 ng/ml murine Steel factor (mSF) and 10 nglm human IL-6 (hlL-6) for 48 hours at 37% in 5% COe . All cells were then harvested anc plated on monolayers of irradiated viral producer cells (1500 cGy X-ray), usin' identical medium with the addition of 6 pg/ml polybrene (Sigma), and cultured at 37°C for additional 48 hours. Loosely adherent and non-adherent bone marrow cells were recovered from the CO-culturesby repeated washing of dishes, using Hank's balancec salt solution (StemCells Technologies) containing 2% FCS, and then counted using ; hemocytometer. All growth factors, unfess othewise specified, were used as dilutec supernatants from transfected COS cells as prepared in the Terry Fox Laboratory. 2.2.3 Transplantation of retravirally transduced bone rnarrow

For bone marrow transplantation procedures, lethally irradiated (950cGy, 1 lOcGyImin., 137~sgamma-rays) (B6C3)Fl (Ly5.2) recipients were injected intravenously with 2 x los bone marrow cells derived from (PepC3)Fl (Ly5.1lLy5.2) immediately after their CO-cultivationwith neo, HOXB4, HOXA 10 or HOXB3- viral producer cells. These mice are hereafter called neo, HOXB4, HOXA 10 and HOXB4 mice, respectively. Donor-derived repopulation in recipients was assessed using flow cytometry, from the proportion of leukocytes in bone marrow, thymus, spleen and peripheral blood, which expressed the Ly5.1 allelic form of the Lys-locus.

2.2.4 In vitro clonogenic progenitor assays

For myeloid clonogenic progenitor assays, cells were cultured at 37OC and in 5% CO2 on 35mm petri dishes (Greiner, Germany ) in a 1.1 ml mixture of 0.8% methylcellufose in alpha medium supplemented with 30% FCS, IoAbovine serum albumin (BSA),10' 4~ B-mercaptoethanol (O-ME), 3 Ufml hurnan urinaiy erythropoietin (hEpo) (SternCells Technologies) and 2% spleen cell conditioned medium (SCCM) (Stemcells Technologies), in the presence or absence of 1.4 mglml of G418. Bone marrow cells harvested after CO-cultivationwith viral producer cells were plated at a concentration of 1-2 x 1o3 cellsldish, whereas bone marrow cells from neo, HOXB4, HOXA 1O. HOXB3 mice were plated at a concentration of 4 x lo4 cellsldish. Spleen cells from neo mice were plated at 3-10 x 1o6 cellddish and from HOXB4, HOXA 10 and HOXB3 mice at a concentration 3-1 0 x 10' cells/dish. Colonies were scored on day 10-12 of incubation as derived from CFU-M, CFU-GM, BFU-E or CFU-GEMM according to standard criteria (Humphries et al., 1981). In sorne experiments, identification of colony types was confirrned by Wright-Geimsa staining of cytospin preparations of colonies and in some instances the classifacations of colony type as megakaryocytelblast cell colony was confirmed using the megakaryocytic specific, GPllbfllla (CD41) surface antigen and flow cytometry. For pre-B clonogenic progenitor assays, bone marrow cells from neo, HOX64, HOXA 10 and HOXB3 mice were plated at a cell concentration of 5-10 x 10' cellldish, in 0.8% methylcellulose in alpha medium supplemented with 30% FCS. IO-' M 8-ME and 0.2 nglml of IL-7 with or without 1.4 mgfml G418. Pre-B colonies werc scored on day 7 of incubation.

2.2.5 Day 12 CFU-S assay

In chapter 3 the day 12 CFU-S assay was utilized. HOXA 10- or neo-transduced bon€ marrow cells were injected into lethally irradiated recipients either immediately aftei retroviral infection, or after 1 week of culture, at an initial density of 1-5 x 105 cellshl ir medium containing 30% FCS, 1% BSA, 10"~M PME, 3 Ulml of hEpo, 2% SCCM with or without 1.4 mglml of (3418. The number of cells that each mouse received waç adjusted to give 10-15 macroscopic spleen colonies. Untransplanted lethally irradiated mice were tested in each experiment for endogenous CFU-S surviving irradiation and consistently gave no spleen colonies. Twelve days after injection, animals were sacrificed by cervical dislocation and the number of macroscopic colonies on the spleen were evaluated after fixation in Telleyesniczky's solution. In certain cases, prior to fixation, well isolated spleen colonies were excised with scalpel blade, cut open and cells were gently spread on a microscopic slide for cytoIogical evaluation after Wright-Geimsa stain.

2.2.6 Cornpetitive Repopulating Unit (CRU) assay

Bone marrow cells from neo, HOXB4, HOXA 10 or HOXB3 mice that had been transplanted earlier with transduced cells derived from (PepC3)Fl (Ly5.1lLy5.2) mice, were injected at different dilutions into lethally irradiated (B6C3)Fl (Ly5.2) mice (5-7 recipients per groupldilution), together with a life sparing dose of 1 x lo5 cornpetitor bone marrow cells from (66C3)Fl (Ly5.2) mice. The level of lymphoid and myeloid repopulation with Ly5.1+ donor-derived cells in these secondary recipients was evaluated >13 weeks later by ffow cytometric analysis of peripheral blood as described (Rebel et al., 1994) Recipients with 1 1% donor (Ly5.1+) derived periphera blood lymphoid and myeloid leukocytes as determined by the side scatter distributior of ~~5.1~cells (i.0. lymphoid low side scatter; myeloid high side scatter), werc considered to be repopulated by at least one lympho-rnyeloid repopulating (CRU) cell CRU frequency in the test ce11 population was then calculated by applying Poisor statistics to the proportion of negative recipients at different dilutions as describec previously (Szilvassy et al., 1990).

2.2.7 Culturing of Scal+Lin-WGA+ cells

In Chapter 3, ScaVLin-WGA+ cells were purified from bone marrow of neo anc HOXA 10 mice (see below, 2.3 ) and cultured as single cells in serum free medium After purification of Scal+Lin-WGA+ cells, they were re-sorted and deposited direct11 into wells of 96-well plates using an automatic ceIl deposition attachment to fACStari (Becton Dickinson). Single cells were cultured in Iscove's modified Dulbecco': medium containing 10 mglml BSA, 10 mglml bovine insulin, 0.2 mgtml transferrin, 1O-' M P-ME and 40 pglml low density liporotein (LDL) (serum free medium) supplemented with the following growth factors: 20nglml mlL-3, 10ngIml hlL-6, 5nglm hiL-7, 25ngfml hlL-11, 3uIml hEpo, 50ngIml mSF, 10ngIml hG-CSF and 1.4 mgfm G418 for selection of transduced cells. Following G418 selection, 9-15 days Iater those wells containing 210 cells were scored by visual inspection for the presence O megakaryocytes. Visual scoring criteria were validated by Wright-Geimsa staining O cytospin preparations and by flow cytometry for expression of the megakaryocytic specific, GPllbIllla (CD41) surface antigen, using D9 mAb (kindly provided by Dr. K.A Ault, Maine Medical Center Research Institute, South Portland, ME, USA).

2.2.8 Morphological evaluations of bone marrow, spleen and periphera blood cells from transplanted mice

At various times after transplantation, peripheral blood ce11 counts and hematocrits a neo , HOXB4, HOXAlO and HOXB3 mice were determined using a Coulter CBCS 59 Differential counts of bone marrow, spleen and peripheral blood cells frorn neo HOXB4, HOXA 10 and HOXB3 mice, that were sacrificed or that became terrninally il were performed on Wright-Geimsa stained cytospin preparations.

2.3 Antibodies, Flow Cytometry and Cell Sorting

Various hematopoietic populations in bone marrow, spleens and thymuses of neo HOXB4, HOXA10 and HOXB3 mice were analyzed by flow cytometry at different timer post transplantation. In some of the HOXB3 mice (Chapter 4) the lymph nodes were also analysed. A single cell suspension of bone marrow was prepared by injectinc Hanks-HEPES buffered salt solution containing 2% fetal calf serum (FCS) and 0.1% sodium azide (Sigma) (HFN) into femurs to flush out cells, followed by gentle disaggregation through a 21 gauge needle. Cells were released from the thymus spleen and lymph nodes by disruption through a fine steel mesh. To lyse erythrocytes cell suspensions were treated with 0.1 65 M NH&I (StemCells Technologies) anc washed once with HFN. Cells were stained with primary antibodies in HFN on ice foi 40 minutes, washed twice with HFN and resuspended in HFN containing 1pg/m propidium iodide (Sigma). Flow cytometric analysis was performed using a FACSort oi FACStar flow cytometer equipped with PC LYSISII software. Monoclonal antibodiea (mAbs) were titered and used as described (Hough et al., 1996; Hough et al., 1994; Rebel et al., 1994). The following mAb were used in this thesis.work: fluorescein isothiocyanate (FITC)-conjugated anti-CD4, anti-CD8, anti-y&T cell receptor (TCR) anti-CD43 (S7), anti-Gr-l , anti-Ly5.1 (A20-1.7) (kindly provided by Dr. G. Spangrude, Salt Lake City, Utha) and anti-wheat germ agglutinin (WGA); phycoerythrine- (PE) conjugated anti-IL-2 receptor a chain (IL-2Ra), anti-CD4, anti-CD8, and anti-B220; and biotinylated anti-9220, anti-Ly-1, anti Gr-1 and anti-ab-TCR mAb al1 of which werc purchased from Pharmingen. HSA was detected by a cyanine 5-succinimidylestei (Cy5)-labeled Ml169 mAb purified from the TI6125 hybridoma (American Type Culture Collection, Rockville, MD). The Sca-1 antigen was detected with Cy5-conjugated El31 61 -7 mAb. The anti-Mac-1 mAb was purified from the MlffO.lS.11 hybridoma 60 (American Type Culture Collection, Rockville, MD) and either labeled with FlTC O biotin. FITC-labeled mouse anti-lgM, and PE-labeled mouse anti-lgD were purchasec from Southern Biotechnology (Birmingham, AL). PE and FlTC conjugated Streptavidir were purchaçed from Jackson lmmuno Research Laboratories (Westgrove, PA).

Bone marrow and spleen cells from neo, HOXB4, HOXA 10 and HOXB3 mice wert stained with anti-Ly5.1, anti-Mac-1, anti-Gr-l , anti-B220, anti-CD43, anti-lgM and anti IgD mAbs, and their spleen cells in addition with anti-CD4 and anti-CD8 mAbs. Thymic cells were stained with anti-Ly5.1, anti-CD4 and anti-CD8. Thymic cells from HOXB; mice were further analyzed using anti-ap-TCR, anti-y&-TCR, anti-IL-2Ra , M/169, anti 8220 and anti-Mac-l . For some of the HOXB3 mice (see Chapter 4) analyzed at 1L weeks after transplantation (n=6), staining of their thymic cells was performed a5 described elsewhere (Hugo et al., 1993). A single cell suspension of 106 celk prepared in PBS 2 % FCS, 0.1 % NaN3 (PBSWB) was deposited in microculturc wells. After centrifugation, the pellet was resuspended in 10 pl of a blocking cocktai containing 5 pg/ml of human gamma globulin (Sigma) diluted in supernatant from the hybridoma 2.4G2, which produces a mAb against the Fc Rllb/lll (Unkeless, 1979), anc incubated for 5 min at 230C. A cocktail of 40 pl PBSWB containing FITC-conjugatec anti-TCR (GL3-1.4; (Gorski et al., 1993)), PE-conjugated anti-CD4 (GK.15; Gibco BRL) and RED613-conjugated anti-CD8 (Gibco BRL), biotinylated-anti-TcRa (H57-597 (Kubo et al., 1989)) mAbs at the appropriate concentrations was added. After a 2E minute incubation at 40C, the cells were washed three times and resuspended in 50 p of a PBSWB with streptavidin-conjugated RED670 (Gibco BRL). The cells were subsequently incubated for an additional 30 minute period, washed and resuspendec in PBSWB to allow analysis on a Coulter XLTM flow cytometer equipped with a 48E nm laser and fluorescence detectors at 525, 575, 620 and 670 nm. For al cytofluorometric analyses done at 14 weeks post-transplantation, a minimum O' 150,000 events were acquired. As presented in Chapter 4, CD4 and CD8 thymic subpopulations Le. double negative CD4-CD8- cells, double positive CD4+CD8+ cells and single positive CD4+CD8- and CD4-CD8+ cells were purified from thymuses of neo and HOXB3 mice using PE-anti-CD4 and FITC-labeled anti-CD-8. For each ce11 population 10,000 cells were haniested and a proportion re-analyzed for purity (>95%)

As presented in Chapters 3 and 5, the proportion of Scal+Lin-WGA+ cells in bone marrow of neo, HOXB4 and HOXA 10 was evaluated. Borie marrow cells were first stained with biotinylated mAb directed against the following lineage (Lin) markers: 8220, Ly-1, Gr-1 and Mac-1. After two washes cells were stained simultaneously with Cy5-labeled anti-Sca-1, FITC-labeled anti-WGA and PE-labeled streptavidin. The Scal +LinmWGA+cells from neo and HOXA10 mice (Chapter 3) were purified from their bone marrow and, after re-sorting, cultured as single cells (see above, 2.2.7).

2.4 Molecuiar Anaiysis

2.4.1 Southern blot analysis

High-molecular weight DNA was isolated from bone marrow, spleen and thymic cells of neo, HOXB4, HOXA10 and HOXB3 mice using the DNAzol reagent (Canadian Life Technologies, Burlington, Ontario), then precipitated with 95% ethanol and washed Nice in 80% ethanol. DNA was then dissolved in 1xTE (10 mM Tris ph7.5, 1 mM EDTA ph8.0), and 10-20 pg of DNA digested with various restriction enzymes at 37OC for 12- 16 hours. The digested DNA was separated on a 0.9% agarose gel and the gel then treated for 15 minutes with 0.1 M HCI solution, followed by a 30-40 minutes treatment with denaturing solution containing 0.5 M NaOH, 1.5 M NaCI. The DNA was then transferred to nylon membrane (Zeta-Probe; Bio-Rad Laboratories, Richmond CA) overnight in 10 x SSC by standard blotting method. The membranes were then baked at 80°C for 1 hour and then pre-hybridized at 65" for 2 hours in 4.4 x SSC, 7.5% formamide (GibcoBRL), 7.5% dextral sulfate (Sigma), 0.75% sodium dodecylsulphate (SDS) (Gibco/BRL), 1.5 mM EDTA (Sigma), 0.75% skim milk and 370mglml of salmon 62 spem DNA (Sigma). A radioactive probe was then added to the pre-hybridization solution and the membranes hybridized for 20 hours at 65OC. Probes were labeled with 3*P-dCTP (3000 Cilrnrnol; ICN Biomedical INC.Costa Mesa, CA) by random priming and purified on a Sephadex-GSO (Phamacia) colurnn before hybridization. Following hybridization membranes were washed 4 times at 65°C for 30 minutes each time, in 0.3 x SSC, 0.1% SDS and 1 mg/ml sodium pyrophosphate. Autoradiography was performed with Kodak XAR-5 fiIm and an intensifying screen, at -70°C for 1-5 days. For re-probing, membranes were stripped by washing for 30-40 minutes in a 1% SDS solution at 100°C.

Kpnl, which cuts once in the retroviral long terminal repeats (LTR's ), was used for releasing the integrated proviruses in DNA isolated from hematopoietic tissue of neo, HOXB4, HOXA10 and HOXB3 mice (Ssfl was used for some of the DNA isolated from neo rnice in Chapter 3). For release of DNA fragrnent(s) specific for the proviral integration site(s), DNA was cut with restriction enzymes that only cut once in the provirus. For that purpose, DNA isolated from neo mice was cut with EcoRl or BamHl (Chapter 5), and Hindlll (Chapter 3), from HOXB4 mice with EcoRl or BamHi (Chaptar 5), and from HOXA 10 mice with Hindlll (Chapters 3). Probes used were a XhollSaA fragment of pMClneo (Thomas and Capecchi, 1987) (Chapters 3,4 and 5), a 1.8 kb genomic KpnllHindlll fragment of the murine SH2-containing inositol phoshatase (SHIP) gene (Darnen et al., 1996)(Chapters 3 and 5), a KpnllMsel fragment of pXM(ER)-190 which releases the full-Iength erythropoietin receptor (EpoR) cDNA (Chapter 3), (kindly provided by Dr. A. D'Andrea) and a 307 bp cDNA fragment 3' to HOXB3 horneodomain obtained by Apal / BamHl restriction digest (Chapter 4)

2.4.2 Northern blot analysis

Total cellular RNA was isolated from various hematopoietic tissues of neo, HOXB4, HOXA 10 and HOXB3 mice using TRlzol reagent (GibcoIBRL), precipitated out using isopropanol and then washed twice with 75% ethanol. RNA was then dissolved in sterile water and 5-10 mg RNA separated on a 1% formaldehyde/agarose gel Following treatment of the gel with 10 x SSC for 30-40 minutes, the RNA waa transferred to nylon membrane (Zeta Probe) ovemight in 10 x SSC . The membran~ was baked for 1 hour at 80°C and then pre-hybridized for 1-2 hours in 50% formamide 0.5 M NaH2P04/Na2HP04 ph 7.2,5% SDS, 1 mgfml BSA, followed by 18-20 hou hybridization after addition of 32P-labeled probe. Probes used were, XhollSal fragment of pMCl ne0 (Chapters 3 and 4), full length human HOXB4 cDNA (Chapte 5), a 300 bp cDNA fragment from a region 3' to the homeodomain of murine HOXAI( obtained by PCR (Chapter 3), and the 2.0 kb Pstl chicken 0-actin fragment (Chapteri 3, 4 and 5)

2.4.3 Western blot analysis

To detect the HOXA 10 protein, neo and HOXA 10 viral producer cells, and bon€ marrow celis of untransplanted, neo and HOXA 10 mice were lysed in phosphate solubilization buffer (PSB; 50 mM HEPES, 100 mM NaF, 10 mM Sodium pyrophosfate 2 mM Sodium vanadate, 2 mM EDTA, 2 mM Na2Mo042H20 ph7.35, 1% NP40 anc protease inhibitors PMSF, Leupeptin and Aprotinin). After electrophoresis of the wholc cells lysate in a 10% polyacrylamide gel, proteins were transferred to nitrocellulose and probed with a whole rabbit antisera to a synthetic oligopeptide derived from the Ca terminal, non-homeodomain region of HOXA10. The blot was then treated with a goa' anti-rabbit IgG second antibody followed by incubation with ECL detection reagens (Amersham Corp., Arlington Heights, IL)

2.4.4 cDNA generation, amplification and analysis

Reverse transcription and amplification of total messenger RNA isolated from the puri subpopulation of thymic cells from neo and HOXB3 mice (Chapter 4) was done exactl! previously reported (Sauvageau et al., 1994). Thymocytes purified by cell sorting (IO,( were pelleted and lysed in 100 pl of 5M guanidinium isothiocyanate solution. Nuc acids were then precipitated, and subsequent synthesis of the cDNA done with a 60 64 primer containing a 3' poly thymidine stretch as described (Brady et al., 1990). A SI poly-adenosine tail was added to the 3' end of the first strand cDNA using term deoxynucleotidyl transferase and the second strand synthesis and subsequent P amplification done by the polymerase chah reaction with the same primer, but at a hiç concentration, as used during the reverse transcription (Sauvageau et al., 19! Amplified total cDNA was size fractionated on a 1% agarose gel, transferred to n) membranes and hybridized as described above (section 2.4.1). Probes used were XhollSaA fragment of pMC1 neo and a 307 bp cDNA fragment 3' to HOXB3 homeodorr obtained by Apal l BamHl restriction digest.

2.5 Statistical Methods

Through out this thesis, results are expressed as mean f standard deviation from mean, as calculated using Student's t-Test: Two-Sample Assuming Equal Varianc provided in Statistical Analysis Tools in the Microsoft Excel program. The Student's t-Tl Two-Sample Assuming Equal Variances was also used to determine whether two sarr means where equaI (the nuIl hypothesis was correct or should be rejected). P valu6 10.05 was pre-chosen as a significant difference between two means. Estimated suw of transplanted mice were based on Kaplan-Meier curves, a product-limit met1 estimating the survivorship function (Kaplan and Meier, !958), where the survival various time points post transplantation is calculated based on the size of each grouf mice when an animal died or became terminally ill. Chapter 3

Overexpression of HOXA IO in murine hematopoietic cells perturbs both myeloid and lymphoid differentiation and leads to acute myeloid leukemial

l~hematerial presented in this Chapter is essentially as described in: U. Thorsteinsdottir, G. Sauvageau, M.R. Hough, W. Dragowska, P.M. Lansdorp, H.J. Lawrence, C. Largman, and R.K. Humphries (1997). Overexpression of HOXA 10 In Mu rine Hematopoietic Cells Perlurbs 60th Myeloid and Lyrnphoid Differentiation and Leads to Acute Myeloid Leukemia. Mol. Biol. Cell 17: 495-505. 3.1 Introduction

As discussed in Chapter 1, members of the HoxA, HoxB and HoxC cluster genes have been shown to be preferentially expressed in the CD34+fraction of human bone marrow cells, which contains most if not al1 of the hematopoietic progenitor cells (Sauvageau et al., 1994). Further detailed analyses of the expression of several oi these Hox genes in functionally distinct subpopulations of CD34+ cells revealed twa patterns of expression: specific genes, primarily located at the 5' end of the clusters (e.g. HOXA 10, HOXB9 ) showed essentially invariant expression in these subpopulations, whereas a second group of genes located towards the 3' side of the clusters (e.g. HOXB3, HOXB4) were expressed at their highest levels in the subpopulation containing the most primitive hematopoietic cells and then sharply downregulated in later cell populations (Sauvageau et al., 1995). By retrovirally overexpressing in murine bone marrow one of the 3' located genes, HOXB4, our group has recently shown that its overexpression selectively enhances the proliferative potential of primitive hematopoietic cells, most profoundly HSC, withoul detectable effects on hematopoietic differentiation (Sauvageau et al., 1995).

On the basis of these results it was hypothesized that these expression patterns might reflect different functional activities of Hox genes in hematopoiesis. As an initial test of this hypothesis, I have examined both the in vitro and in vivo behavior of primary murine hematopoietic cells which were engineered by retroviral gene transfer to overexpress HOXA 10. Here, I demonstrate that the effects of HOM 10 overexpression stand in sharp contrast to those of HOXB4. Overexpression of HOXA 10 is found to profoundly perturb differentiation of bone marrow progenitors with megakaryocytic and monocytic potentials, is not permissive for 8-cell development and enhances proliferation of hematopoietic progenitor cells. Furthermore, a significant proportion of recipients of HOXAlO-transduced bone marrow cells eventually develop acute myeloid leukemia. 3.2 Results

3.2.1 Retroviral-mediated transduction of HOXA 7 O to murine bon4 marrow cells

For this study I used a full length HOXA10 cDNA isolated from the human myeloid ce line ML3 which encodes a 55 kDa protein (Lowney et al., 1991). This HOXA10 cDNl was chosen because it represents the most abundant HOXA IOtranscript found il human bone marrow cells (Lowney et al., 1991). The HOXA 10 cDNA was inserted inti the murine stem cell virus (MSCV) 2.1 retroviral vector (Hawley et al., 1992) 5' to i phosphoglycerate kinase promoter (PGK)-driven ne0 gene such that HOXAlO wa: expressed from the viral enhancer and promoter sequences within the long termina repeat (LTR) (Figure 3.1A). The LTR sequences in this MSCV vector have been show1 previously to give high and long-term expression in primitive murine hematopoietic cells and their mature progeny of both myeloid and lymphoid iineage (Hawley et ai. 1994; Pawliuk and Humphries, Unpublished data; Sauvageau et al., 1995). lntegrity o the HOXAlO retrovirus was verified by Northern and Western blot analysis whicl detected the two expected HOXA 10-containing viral mRNAs (due to the splice dono and acceptor sites in the MSCV 2.1 vector) (Hawley et al,, 1994) and the HOXA I( protein in the viral producer cells, respectively (Figure 3.1 B and 3.1 C). For some part of this study I used in addition an MSCV retroviral vector containing the HOXB4 cDNl under the control of the viral LTR, which is desribed in Chapters 2 and 5.

To investigate the effects of HOXA 70 overexpression in primary hematopoietic cells, bone marrow cells from mice injected 4 days previously with 5-fluorouracil (5 FU) were CO-cultivatedwith neo or HOXA10 viral producer cells for 48 hours, and thi subsequent effects of deregulated expression of HOXAlO on the behavior O hematopoietic cells was analysed both in vitro and in vivo. A

HOXAlO

- 4.6 kb 3.9 kb neo -

Figure 3.1. Structure and expression of the HOXAlO retrovirus used in this study.

(A) Diagrammatic representation of the integrated HOXA 10 and ne~proviruses.Expected size of tl full-length viral transcripts and those initiated from the intemal PGK promoter are shown. Two vir transcripts are generated from each vifus due to the splice donor and acceptor sites in the MSCV 2 vector (6)Northem blot analysis of total RNA isolated from the HOXA 10-viral producer cells. Tt membrane was hybridixed with a probe specific for neo that detects both viral transcripts (a 4.6 kb fu length and 3.9 kb spliced RNA) and the 1.3 kb neo transcript initiated from the PGK promoter (left) ar subsequently with a full-length HOM 10 cDNA probe that detects both viral transcripts (right). (C) Weste blot analysis of whole cell lysate from ne* and HOXAIO-viral producer cells. The blot was probed wi polyclonal antisera directed against a HOM10 synthetic oligopeptide. The arrow indicates the band fi the HOXA10 protein rnigrating at the expected size of 55 kDa.

3.2.2 Altered colony formation in vitro of myeloid progenitor cell overexpressing HOXA 10

To test whether overexpression of HOXA 10 would affect the ability of cornmitte rnyeloid progenitor cells to complete their differentiation in vitro, ne* and HOXAl l 69 transduced bone marrow cells were plated, immediately after retroviral infection, in methylcelIulose cultures for rnyeloid colony formation. The efficiency of retroviral infection, as assessed by the proportion of G418 resistant clonogenic cells, varied between experiments and viral producer cells, and was 50e8% and 3W3% for neo- and HOXA IO-retrovirus, respectively (meanf SD, from 2 independent experiments).

80 4 neo 70 - HûXB4

Figure 3.2 Effects of HOXAlO and HOXB4 overexpression on the relative frequency of various colony types generated in vitro, immediately following retroviral infection of bone marrow cells.

On day 12-13, well isolated G418 resistant colonies were randomIy picked (n=74 for neo, n=76 for HOX84 and n=77 for HOM10) and examined after Wright staining. Results are expressed as meartkSD from two independent experiments. The various colony types (GEMM, GIGM and MD) generated from neo- and HOXB4-transduced cells were not significantly different, p>0.5. Megakaryocytehlast colonies were significantly increased in HOXA10 cultures compared to neo cultures, pc0.01; and G/GM and GEMM colonies were significantly decreased in HOXA10 cultures compared to neo cultures, pc0.05. Although no M0 colonies were detected in HOXAIO cultures they were not significantly decreased compared to neo cultures as those numbers were both low and varied greatly between experiments (student t-test).

The total number of colonies generated was similar for both types of infected cells (-80 colonies/1000 cells). However, there was a striking difference between the cellular constituents of neo- and HOXA 1 O-transduced colonies as revealed by cytological examinations of Wright-Geimsa stained G418 resistant colonies (Figure 3.2). About 45% of the progenitors transduced with HOXA 10 generated large colonies containing megakaryocytes and blast cells (Figure 3.7A ), a colony type not detected among neo-transduced colonies. The generation of this unique cdony type ii HOXAlO cultures was accompanied by a proportional reduction in multilineagi GEMM, granulocyte-macrophage and granulocyte colonies; moreover, no unilineagt macrophage colonies could be detected among (3418 resistant (i.e.HOXAl0, transduced) colonies. In contrast to this altered myeloid differentiation mediated b! HOXA IO, overexpression of HOXB4 did not alter the proportion of various colon! types generated in vitro (Figure 3.2).

3.2.3 HOXAlO overexpression increases the maintenance of day fi CFU-S in vitro.

In Our previous study on the effects of overexpression of HOXB4 in murine bon€ marrow cells, we observed a more than 2 log enhancernent in the recovery O multipotent myeloid progenitor cells, day 12 CFU-S, after a 7 day culture perioc following retroviral infection with HOXB4 (Sauvageau et al., 1995). To assess whethei HOXA 10 overexpression had a similar effect, the day 12 CFU-S content of neo- anc HOXAlO-infected bone marrow cells were measured irnmediately after retrovira infection and again after 7 days in liquid culture supplemented with growth factors anc 1.4 mg/ml G418 for selection of transduced cells. The day 12 CFU-S frequencies measured immediately after retroviral infection were similar for neo-and HOXA 10- transduced bone marrow cells (Figure 3.3, day 0). However after maintaining these cells for 7 days in liquid cultures the day 12 CFU-S content of cultures initiated with HOXA 10-transduced cells showed a net increase to 200% of input values, whereas levels in neo control cultures had decreased to -5% of input (Fig 3.3, day 7). Thus HOXA 10 like HOXB4 can reverse the decline in day 12 CFU-S numbers normally observed under these culture conditions, suggesting that overexpression of HOXAIC can affect processes involved in the generation or maintenance of cells with day 12 CFU-S ability. Starting cells post in vitro culture (day 0) (day 7)

Figure 3.3. Effect of overexpression of HOXA 10 on the recovery of day 12 CFU-S in vitro.

The CFU-S content was assessed immediately after CO-cultivation(day 0) and also afier 7 days in liquid culture. Resuits (meaeSD) are expressed as day 12 CFU-S nurnbers per 105 starting day O cells from 2 independent experiments, except for day O values for HOM10 which represent 1 experirnent (ne0 white and HOXA 10 black bars). Cytological examination of Wright-Geimsa stained cell preparations from day 12 spleen colonies, showed the same proportion of differentiated erythroid (>go%) and myeloid elements for both HOXA 10- and neo-transduced colonies (data not shown), indicating that overexpression of HOXA 10 did not overtly alter the pattern of terminal differentiation of day 12 CÇU-S during spleen colony formation in vivo.

3.2.4 Expansion in vivo of myeloid progenitor cells overexpressing HOXA 10 and their aItered colony formation in vitro

To delineate further the effects of HOXA 10 overexpression on hematopoiesis, we analysed long-term myeloid and lymphoid reconstitution in lethally irradiated mice transplanted with HOXA 10- or neo-transduced bone marrow cells (hereafter called HOXA 1 O and neo mice, respectively). At 8-1 5 weeks post transplantation, hematopoietic regeneration in either neo or HOXA 10 mice was essentially completely donor derived, as >85% of bone marrow, thymic and peripheral blood leukocytes were 7 2 of transplant origin (Ly5.1+). Moreover, the intensities and patterns of proviral signals seen upon Southern blot analysis of DNA from bone marrow and thymus of these mice, were consistent with high level, poiyclonal reconstitution by transduced cells in both neo and HOXA10 mice (Figure 3.4A and 3.48). Expression of the transduced HOXA10 cDNA was readily detected in total bone marrow cells of HOXA 10 mice, at both the RNA and protein levels by Northem and Western blot analyses, respectively (Figure 3.5A and 3.58). Endogenous HoxalO expression in contrast, was below the detection level, using these same methods, in either neo, HOXA 10 or normal untransplanted mice (Figure 3.5A and 3.56). -5m neo Ju 1 ajqt6s a (19

Figure 3.4. Southern blot analysis of DNA isolated from hematopoietic tissue a HOXAIO and ne0 mica.

(A) Dernonstration of the presence of intact integrated HOM 10 and neo proviruses. DNA from boni marrow of HOXA 10 mice was digested with Kpnl and that of neo rnouse with SsB to release the integrata HOM10 (4.6 kb) and neo (2.7 kb) proviral fragments, respectively. The blots were hybridized with a nec specific probe to detect the proviruses and then subsequently with a probe specific for murine EpoR tc provide a single gene copy control for loading. Exposure times were equivalent for both probes (-3! hour). (6) Anaiysis of proviral integration patterns in DNA isolated from bone marrow and thymic celfs fron these same HOM10 and neo mice. DNA was digested with Hindll which cuts the integrated proviru: once, generating DNA fragments unique to each integration site. The membranes were hybridized with ; probe specific for neo to detect proviral fragments, and subsequently with a probe specific for murint €@OR to provide a single gene copy control for loading. Exposure times were 72 hours for the neo probt and 24 hours for the EpoR probe. The numbers assigned to various lanes identifies a specific rnouse The HOM 10 mouse number 1 was sacrificed 8 weeks post transplantation whereas other HOXA 10 micc and the neo mouse, were sacrificed 15 weeks post transplantation. Abbrev. BM, bone marrow; T, thymus. un-lx neo HOXAIO

HoxalO

Figure 3.5. Northern and Western blot analyses to demonstrate high levels of vil derived HOXAlO messages and protein in reconstituted HOXAiO mice. (A) Northem blot analysis of total RNA (10pg) isolated from the bone marrow of normal-. neo-ai HOXA IO-mice. The membranes were hybridized with a probe specific for neo that detects both vil transcripts in neo (2.7 kb) and HOW 10 mice (4.6 kb full-length and 3.9 kb spliced RNA) and the 1.3 neo transcnpt initiated from the PGK prornoter, and subsequentiy with a probe specific for the muni Hoxa 10 for detection of both the endogenous Hoxa 10 messages and the HOM10 viral transcripts. TI endogenous 2.0 or 2.5 kb murine HoxalO messages were not detected in any of the mice. As a conti for loading the membranes were also hybridixed with a probe specific for actin. Exposure times we equivalent for both Hoxa10 and actin probes (-48 hours). (B) Western blot aanalysis of whole cell lysati from normal-,ne& and HOXAI&mice. The blot was probed with a polyclonal antisera directed against HOXA 10 synthetic oligopeptide. The arrow shows the band for the HOM 10 protein migrating at tl expected size of 55 kDa. Each number assigned to vanous lanes identifies a specific mouse (not tt same as in figure 5). HOM 10 mice number 1 and 2 were leukemic whenas 3 and 4 were not. Abbre BM, bone marrow; un-Tx, normal untransplanted mouse. When analysed 8-15 weeks post transplantaton in contrast to neo control mice, HOXA 10 mice showed weight loss and signs of hematological abnormalities (Table 3.1). Although their peripheral blood and bone marrow nucleated cell numbers were within normal range, HOXA 10 mice had moderate splenomegaly and mild anemia (Table 3.1).

Table 3.1 Body weight and hematological parameters in neo and HOXAIO mice

- -- -

Peripheral blood parametersa

Mice Body weightb Spleen weighta WBC/femura RBC Hb WBC (9) (9) (XI 07) (XI 09/mi) (gfdl) (xi 06fml) ------

neo 31 -5k3.0 0.1 1+0.01 1.9k0.5 8.6+1 .O 15.8f 1.5 8.3I2.3

Results shown represents meankSD. an=7 for neo mice and n=9 for HOM10 mice, 8-1 5 weeks post transplantation bn=l 1 mice sex and age matched, 15 weeks post transplantation CSignificantIyless than neo control. pc0.03 d~ignificantlygreater than neo control, pc0.03 eSignificantly less than neo control, pc0.05

The numbers of myeloid progenitor cells were significantly increased in HOXA 10 mice when compared to neo control mice 15 weeks post transplantation (Table 3.2). This was most pronounced in the spleen, with on average a 9-fold increase over that found in neo mice (Table 3.2). Preferential expansion of HOXA 10-transduced progenitor ceils over that of non-transduced was indicated by a higher proportion of G418-resistant rnyeloid colonies detected from the HOXA 10 mice compared to that of neo control mice (Table 3.2), despite lower gene transfer in the initial transplanted bone marrow inoculum for the HOXA 10 (67.5&4% and 42.5I18% for neo and HOXA 10 respectively). For some of these HOXA 10 mice their splenic nucleated cell counts were also found to be elevated, which in al1 cases was due to an increase in cells expressing both the Mac-1 and Gr-1 antigens as determined by flow cytometric analyses.

Table 3.2. HOXAlO mice have increased numbers of myeloid clonogenic progenitor cells when compared to neo control mice.

Mice Number of Number of myeloid CFC (16)fiemur myeloid CFC (103)/spleen (% G4183 (% G4183

neo

HOXA 1 O 75.0k37a (62.0I13)

Results are shown as meaniSD of the number of myeloid clonogenic progenitor cells in bone marrow and spleen of n=6 neo and n=8 HOXA 10 mice, 15 weeks post transplantation. Abbrev. CFC, colony forming cell aSignificantly higher than in neo control mice, p<0.05 b~ignificantlyhigher than in neo control mice, pc0.005

Myeloid clonogenic progenitors of HOXAlO mice remained growth factor dependent for in vitro colony formation when analysed 8 weeks post transplantation. However, in the presence of added growth factors, -45% of the HOXAl0-transduced progenitor cells from bone marrow of these mice generated a unique colony type containing megakaryocytes and blast cells (Figure 3.76, Table 3.3), representing at least a 35-fold increase in the absolute numbers of this progenitor type compared to ne0 control mice. Curiously, despite this high frequency in bone marrow of HOXA IO rnice of a progenitor cell with potential to differentiate in vitro into megakaryocytes, inspection of bone marrow and peripheral blood smears from HOXA 10 mice revealed no gross increase in megakaryocyte or platelet counts. In addition to a high proportion of megakaryocyte and blast cell colonies, another -20% of colonies from HOXA 10 mice were small (-100 cells), and contained mainly highly granular blast cells and very few differentiated myeloid elements, a colony type not detected among the neo- transduced colonies (Figure 3.7C, Table 3.3). Furthermore, no unilineage 7 7 macrophage colonies could be detected among the HOXA 10-transduced colonies, which in contrast represented about 30% of colonies generated from bone marrow 01 neo mice.

Table 3.3 HOXAIO mice have greatly increased numbers of bone marrow megakaryocytelblast and blast colony forming cells (CFC) and decreased numbers of unilineage macrophage CFC.

Number of G418'CFC / femur.103

Mice Total CFC GEMMBFU-Ea GlGM M MegaJBlast Biast

neo 16.9f0.8 5.5f 1.5 6.5I2 4.7I3 ~0.33 ~0.33

HOXA 10 26.2I1.3 3.812 5.0I2.5 ~0.46 11.541 5.5k1

Fold difference 1.6 0.7 0.7 0.1 36.5 16.0 HOXA 1 Olneo

On day 12-13well isolated colonies were randomly picked and analysed by Wright- Geimsa staining (n=50for neo and n=58 for HOXA10) . Results shown represent rneanfSI) of the number of various G418 resistant colony types generated from 2 mice in each group. =BFU-E colonies represented only 0.5% of analysed colonies and therefore were combined with GEMM Abbrev. GEMM, granulocyte-erythrocyte-macrophage-megakaryocyte; G/GM, granulocyte/granulocyte-macrophage; M, macrophage; Mega/Blast, rnegakaryocyte-blastcells.

Interestingly, macrophage colonies could be detected in the absence of G418 in cultures initiated with bone marrow cells from HOXA 10 mice (data not shown), indicating the presence of non-transduced macrophage progenitor cells capable of normal differentiation in these mice. Thus the block in macrophage colony formation due to overexpression of HOXA 10 appears to be intrinsic to transduced cells rather than attributable to accessory cells. Table 3.4 Scal +LinoWGA+ cells overexpressing HOXA 1O have increased potential to generate megakaryocyte-containing colonies in vitro

Mice Scal +Lin'WGA+ % G418' G418' Scal +LineWGA+cells (103) celk (103) ffemur, Scal +Lin'WGA+ cells Ifernur generating megakaryocyte containing colonies, (% of total G418' colonies)

neo 29.019.9 30.0I 10

Scal+Lin-WGA+ cells purified from bone marrow cells of neo (n=2) and HOM 10 (n=2) mice 8 weeks post transplantation, were seeded as single cells/well and foliowing G418 selection, 9-1 5 days later, those wells containing 21 0 cells (n=296 for HOXA IOand nd86 for neo mice) were scored for the presence of megakaryocytes. Results shown represents meanriSD. aSignificantly different from neo control, pcO.05

Similarly, the induction of megakaryocytic differentiation also appears to b~ intrinsic to HOXA 1 O-transduced progenitor cells. When cells with the primitiw phenotype Scal +Lin-WGA+ (purified from bone marrow of HOXA 1 O ,ne0 anc untransplanted mice) were cultured as single cells in serurn free liquid cultures suppiemented with various growth factors, HOXA IGtransduced Scal +Lin-WGA+cells showed increased potential to generate megakaryocyte containing colonies, bott compared to neotransduced and untransduced Scal+Lin-WGA+cells (Table 3.4). Thus, overexpression of HOXA IO in vivo, like that of HOXB4 (Sauvageau et al. t 995) induces expansion of myeloid progenitor cells, but in contrast to HOXB4, alters their normal differentiation.

3.2.5 Overexpression of HOXA 10 in vivo is not permissive for pre-8 lymphoid colony formation.

While transduced myeloid progenitor cells were increased in numbers in HOXA IC mice, total pre-B colony forming cells were slightly reduced in number (-2-fold) and virtually none were derived from transduced cells (G418-resistant). Overall, 79 transduced pre-B progenitors in HOXA10 mice were some 22-fold lower in absolutt number compared to those in neo control mice (Table 3.5).

Table 3.5. Overexpression of HOXA IO is not permissive for lymphoid pre-B colony formation.

Mice Number of pre-8 % G418' progenitors(103)/femur pre-6 progenitors

neo 1Of5.5 4321 0

HûXA 1O 5.5+4 4*4.5a

Results are expressed as meanSD of the number and the % G418 resistant colonies generated from bone marrow of neo (n=5) and HOXAlO (n=5) mice, 8 and 14 weeks post transplantation. asignificantly less than in neo control mice, p<0.0005

To further demonstrate the inability of HOM10-transduced cells to contribute to B. lymphopoiesis, the intensity of the proviral signal in DNA isolated from spleen, ir which mature 6-cells normally constitute the majority of cells (-2/3), was compared tc that from bone marrow and thymus. For that purpose only those HOXA10 mice werc analysed (n=4) in which the proportion of splenic B- and myeloid cells were withir normal range, as determined by flow cytometric analyses (B220+ cells, 60-62% anc Mael+, 3-4%). In contrast to neo control mice, where equal proviral signals were detected in al1 three tissues, the strength of the proviral signal from spleen cells ir HOXA 10 mice was on average only 113 of that in bone marrow and thymus, consisteni with a minimal contribution of marked cells from the B cell fraction (Figure 3.8A). Flow cytometric analysis of various B ce11 populations in bone marrow and spleen of nec and HOXA10 mice showed that the absolute numbers of pro-B, pre-B and mature B cells were within normal range in HOXAlO mice (data not shown). Thus the regeneration and differentiation of B-cells in HOXA 10 mice frorn untransduced cells (estimated to be 250% of bone marrow cells in the transplant inoculum, see above appeared unaffected and compensated for impaired B-ce11 differentiation b! transduced cells .

HOXA10 mice when analysed 8-15 weeks post transplantation had normal thymic size, and by using antibodies against CD4 and CD8, were found to have norma numbers and relative frequencies of thymocyte subpopulations (data not shown). Th6 contribution of transduced cells to thymic regeneration in these HOXAIO mice wa: shown by Southern blot analysis that detected proviral signais, which in most case: was comparable to levels seen in bone marrow (Figure 3.5B and 3.8A) Overexpression of HOXA 10 thus appears to have no gross effect on T lymphoic development.

3.2.6 Acute myeloid leukemia arises in recipients of HOXA 7 O-t ransducec bone marrow cells.

In two independent experiments, a total of 23 mice were transplanted with HOXAlO transduced bone marrow cells. Of these, 11 animais were sacrificed 8-15 weeks post- transplantation but the remaining 12 were monitored for a prolonged period of time tc test whether overexpression of HOXA 10 would eventually lead to a disease state, Nine of these HOXA 1O mice (from both transplantation experiments) became terminally il1 and died 19-48 weeks post-transplantation; 3 remained alive at 52 weeks post transplantation (Figure 3.6). The diminished survival of HOXA 10 mice contrasts with mice transplanted with HOXB4-transduced bone marrow cells, for which survival was not different from that of the neo control mice (Figure 3.6). Weeks post-transplantation Figure 3.6. Survival of HOXAlO mice compared to that of neo control mice and mice transplanted with HOXB4-transduced bone marrow ceils using Kaplan-Meier estimates.

Survival at various time points post transplantation was calculated based on the size of each group of mice when an animal died or becarne terrninally ill. The size of each group at various tirne points post transplantation is shown in the upper part of the graph where a=HOXAIO, b=HOXB4 and c=neo mice. Survival of HOM10 mice was statistically different from that of neo control mice and HOXB4 mice, pe0.05.

In al1 cases where documentation was possible (n=7), acute myeloid leukemia (AML) was the cause of death of the HOXA10 mice. These mice had elevated white blood ceIl counts (>50,0OO/pl), were profoundly anemic and some had hind limb paralysis, likely caused by leukemic infiltration of lumbo-sacral roots. Cytological examination revealed large numbers of blast cells in bone marrow, spleen and lymph nodes (Figure 3.7 D-F). For the two mice tested, the leukemia was readily transplantable to both irradiated (n=5) and non-irradiated (n=5) recipients, which developed a fulminant leukemia 3-5 weeks after intravenous injection of 2 x 106 bone marrow cells. Figure 3.7 Wright-Geimsa staining of cytospin preparations from HOXAlO-transducec myeloid colonies (A-C) and from hematopoietic tissues from a representative leukemic HOXA10 mouse (D-F) (A and B) Representative megakaryocyte (a) and blast ceIl (b) colonies picked from methylcellulosf cultures initiated with cells obtained either immediately following retroviral infection of 5-FU bone marrou cells (A) or with cells recovered from bone marrow of HOM 10 mice (B). (C)Smaller colony observed onIl in methylcellulose cultures initiated with bone marrow tells from HOM10 mice containing immature blast like myeloid cells (c) together with few differentiated granulocytic elements (d). D to F show periphera blood, bone marrow and lymph nodes, respectively, from a representative HOXA 10 mouse tha developed acute myeloid leukemia. Note the presence of leukemic blasts together with a few mature granulocytic elements in each of these three organs. Magnification: x600. Southem blot analysis of hematopoietic tissue from leukemic HOXA 10 mice showed that the leukemic cells contained the HOXA IO-provirus, and that the leukemia appeared mono-or bi-clonal (Figure 3.88). Similarly, Northern and Western blot analysis of bone marrow cells from two leukemic mice detected high levels of the full length retrovirally driven HOXA 10 messages (4.6 kb and 3.9 kb) and the HOXA 10 protein (Figure 3.5A and 3SB, respectively). The mono-or bi-clonality of the leukemias, together with their delayed onset, suggests that overexpression of HOXA 10 is in itself not fully transforming and that additional mutation(s) is (are) required for leukernic transformation. Further evidence for the leukemogenic nature of overexpression of HOXA10 was obtained by transplanting bone marrow cells from an apparently healthy primary HOXA 7 0 mouse (1 5 weeks post transplantation) into lethally irradiated secondary recipients. By 8 to 12 weeks post-transplantation al1 of the secondary recipients (n=17) developed AML which in al1 cases was found to be caused by the same HOXA 7Gtransduced leukemic clone, as evident by Southern blot analysis (Figure 3.88). This clone was also found to be the dominant HOXA IO- transduced clone in the bone marrow of the primary donor mouse (Figure 3.86) indicating the existence of a preleukemic state in this rriouse. Interestingly, the primitive Scal+Lin-WGA+ bone marrow fraction of that same prirnary HOXA 10 mouse was not as leukemogenic. Of the 2 secondary recipients which received HOXAIO- transduced Scal+Lin-WGA+cells, the AML developed after a longer latency period (34 and 45 weeks) and interestingly in the one mouse analysed, the leukemia originated from a different HOXA IO-transduced clone (Figure 3.88).

Although overexpression of HOXA 10 in non leukemic primary recipients did not render clonogenic cells growth factor independent, al1 three leukernic clones tested showed growth factor independent growth, in cultures containing fetal calf serum. Figure 3.8. Southern blot analysis to determine: (A) the intensity of the proviral signa in bone marrow, spleen and thymus of HOXAlO and neo mica; and (8) the cionality a the leukemias ln HOXAlO mice (A) Southern blot analysis of the intensity of the proviral signal in bone marrow, spleen and thymus a HOX410 and neo mice. DNA was digested with Kpnl, which releases the integrated HOXA10 (4.6 kb) anc neo (2.7 kb) proviral fragments. The blot was hybridized to a probe specific for neo to detect the proviruses and subsequently to a probe specific for the murinesHP gene as a single copy gene control for loading. Digitized images of the autoradiograms were obtained by densitometric scanning with a Computing Densitometer (Molecular Dynamics) and the signal intensity for each lane then analysed by ImageQuaNT (4.1). Each number assigned to individual lanes identifies a specific mouse. HOXA10 and neo mice were sacrificed 8 (number 6 and 3) or 14 weeks post transplantation. (8) Southem blot analysis of proviral integration sites in DNA isolated from leukemic HOXA 10 mice. DNA was digested with either Hindlll or EcoRl that each cut the integrated provirus once, generating DNA fragments unique to each integration site. The membranes were hybridized with a probe specific for neo to identify proviral fragments. Abbrev. BM, bone marrow; S, spleen; Tl thymus; LN, lymph node; 1-, primat'y mouse; 201 secondary mouse; SLW, Scal+Lin'WGA+ bon8 marrow cells; TBM, total bone mamw. 3.3 Discussion

In this study I have shown that retroviral overexpression of one of the 5' located Hoi genes, HOXA IO, perturbs the differentiation of both rnyeloid and B-lymphoic progenitor cells and eventually leads to th8 generation of acute myeloid leukemia These effects are quite distinct from those previosly observed for the 3' locatec HOXB4 gene, whose overexpression enhanced the expansion of primitive hematopoietic cells but neither altered hernatopoietic differentiation nor predisposec to leukemia. Taken together, these data add further functional evidence foi involvement of Hox genes in key hematopoietic developmental processes in a Hor gene-specific manner.

A striking effect of overexpression of HOXA10 was the enhanced generation in vifrc of colonies containing megakaryocytes and blast cells and the suppression 01 macrophage colony formation. This effect was seen both for the transduced progenitors recovered from the regenerated bone marrow of HOXA 10 mice and immediately following HOXA 10 infection of committed progenitor cells in the 5-FU bone marrow, thus arguing for a direct effect of HOXAIO on differentiation of cornmitted myeloid progenitor cetls. The detection of non-transduced macrophage progenitors in bone marrow of HOXA 10 mice further supports the idea that the obse~edblock in macrophage colony formation is an intrinsic property of HOXA IO- transduced cells rather than attributed to accessory cells. The induction of megakaryocytic differentiation also appears to be intrinsic to HOXA 10-transduced progenitor cells, since HOXA 70-transduced Scal +Lin-WGA+cells, when cultured as single cells in liquid culture showed, a similar increase in megakaryocyte generation. Interestingly, alt hough colonies containing megakaryocytes were generated in high frequency frorn HOXAlO-transduced progenitor cells in vitro, t did not detect any gross increase in megakaryocyte or platelet numbers in HOXA 10 reconstituted mice. This suggests that the probability of HOXA IO-mediated megakaryocytic induction is increased when growth factor concentrations are above normal physiological levels as they are in vitro, or alternatively, that HOXAlO induced megakaryocyti differentiation could be inhibited in vivo by some unknown negative regulator mechanisms either not present or functional in vitro.

The transcriptional control of megakaryocytic development remains large11 unknown. Of the lineage-specific transcription factors known to be expressed in thc megakaryocytic lineage e.g. GATA-1 and -2, Tal-l/SCL and NF-E2, only NF-E2 ha been shown to be essential for normal megakaryocytic differentiation, where it ii needed for completion of megakaryocytic maturation and platelet generation (Pevn: et al., 1991; Shivdasani et al., 1995a; Shivdasani et al., 1995b; Tsai et al., 1994). 1 functional role for GATA-1 during early stages of megakaryocytic development ha: been inferred, since forced overexpression of GATA-1 is able to reprogram bot1 murine and avian myeloid cell iines to differentiate into the megakaryocytic lineagt (Kulessa et al., 1995; Visvader et al., 1995). Although HOXA 10 expression has no been detected in the Iimited studies of cell lines which have both megakaryocytic anc erythroid differentiation potential (Lowney et al., 1991; Magli et al., 1991; Vieille Grosjean et al., l992), the results presented in this study implicate Hox genes ai potentially important regulators of megakaryocytic differentiation. A more detailet assessrnent of Nox gene expression patterns, particularly HOXA 1O, durins megakaryocytic differentiation will now be of considerable interest as will attempts tc identify effects subsequent to Hox gene knockout.

Although HOXAlO expression during megakaryocytic differentiation has not beer assessed, its expression during other stages of myeIoid differentiation has beer analysed in sorne detail (Lawrence et al., 1995; Sauvageau et al., 1994). HOXA II expression was found to be highest in subpopulations of human bone marrov enriched for LTC-IC, CFU-GM and BFU-E cells, then down-regulated -6-fold witt progression to bone rnarrow CD34 cells (Sauvageau et al., 1994) and undetectablc in mature circulating monocytes and granulocytes (Lawrence et al., 1995). Thus th€ 87 suppression of macrophage colony formation and the appearance of colonie: containing immature myeloid blast cells when HOXA 10 is overexpressed suggest tha the observed down-regulation of HOXA 10 as myeloid cells mature is a critical even for normal myeloid differentiation.

The second major effect of overexpression of HOXA 10 was an impairment in earlg B cell development, as reflected by virtual absence of HOXA IO-transduced pre-E clonogenic progenitor cells in bone marrow of HOM10 mice. Non-transduced B ceIl5 in HOXA 10 mice appeared to develop normally and compensated for the absence oi transduced B cells; thus the effect of HOXA 10 again appears to be intrinsic tc transduced cells. This finding for HOXAlO contrasts with the increase in pre-E progenitor cells induced by overexpression of HOXB4, in mice transplanted witt HOXB4-transduced bone marrow cells shown previously (Sauvageau et al., 1995: and presented in Chapter 5, thus indicating differentiai effects of Hox proteins on 6. lymphopoiesis. Using an RT/PCR based approach, the murine Hoxa 70 message, car not be detected in FACS-purified B ceii subpopulations from normal (B6C3)FI mice, representing both early and late stages of B cell development (Sauvageau et al. Unpublished data). These results are in agreement with earlier studies, wherc HOXA10 expression could not be detected in cell lines of pre-B or mature B cell origir nor in leukemic cells from patients with pre-B cell acute lymphoid leukemia (ALL), 6. cell ALL or B-cell chronic lymphoid leukemia (CLL) (Celetti et al., 1993; Lawrence ei al., 1995; Vieille-Grosjean et al., 1992). The observed block in 6-cell developmeni when HOXA 10 is retrovirally overexpressed suggests that HOXA 10 target genes ma) interfere with normal 0-cell development, and thus that downregulation of HOXA10 is important following cornmitment to the B-cell lineage. Since many Hox genes are expressed during hematopoiesis, it is also possible that HOXA 10 might be rnimickinç the overexpression of or blocking the action of another Hox gene normally involved in B cell development. A major consequence of HOXA 10 overexpression in transplanted mice was a high frequency of acute myeloid leukemia. The relatively long latency period (>19 weeks] and the mono- or bi-clonality of the leukemias suggest that other complementing mutation(s) is(are) needed for dominant outgrowth of a single ceIl. The hallmark 01 leukemia is a block in the normal differentiation program coupled with unrestrained growth which leads to clonal expansion of immature blast cells. Like HOXA IO, overexpression of HOXB4 clearly leads to expansion of primitive hematopoietic cells as evident by increased progenitor numbers in transplanted mice and enhanced recovery of day 12 CFU-S following in vitro culture ((Sauvageau et al., 1995) and Chapter 5). Overexpression of HOXB4 however, unlike HOXAlO does not predispose to leukemia. This difference may relate to the absence of HOXB4 induced differentiation changes in contrast to profound effects of HOXA 10 on differentiation. The effects of HOXA 10 are thus reminiscent of that reported for the nuclear oncoprotein c-myb which like HOXA 10 is normaily predominantly expressed in immature hematopoietic cells and is thought to contribute to leukemogenesis by blocking differentiation but maintaining proliferation (Thompson and Ramsay, 1995). Perkins et. al., using experimental strategies similar to those applied here for HOXA 10, have shown that Hoxb8 also has leukemogenic potential, and at least in sorne cases, the onset of the leukemias may have been triggered by autocrine growth factor production (Perkins arid Cory, 1993). Interestingly, some of the HOXA 10 leukemic clones were also found to be capable of factor-independent growth. The recently observed CO-activationof the PBX- 1 related Meis- 7 gene with Hoxa 7 or Hoxa9 in myeloid leukemias in BXH-2 mice (Nakamura et al., l9Q6b) raises the possibility that the leukemogenic effect of HOXA 10 may also involve Hox protein co- factors.

The results presented in this study, together with the extensive literature showing that HOXA 10 expression is largely confined to primitive cells of the myeloid lineage, suggest that under normal physiological conditions HOXA 10 is likely involved in processes of hematopoietic Iineage cornmitment and differentiation, playing a positiv~ role in megakaryopoiesis but negatively regulating monocytic and B-cell developmen These results add to the recognition of Hox genes as important regulators c hematopoiesis and point to Hox gene-specific effects that likely reflect their regulatioi of different target genes during hematopoietic development. Further resoIution of th1 Hox gene "code" and the molecular processes that they affect during hematopoietil development remains an important challenge. Chapter 4

Overexpression of HOXB3 ln hemopoietic cells causes defective lymphoid development and progressive myeloproliferation*

2~hematerial presented in this Chapter is essentially as described in: G. Sauvageau, U. Thorsteinsdottir, M.R. Hough, P. Hugo, H.J. Lawrence, C. Largrnan, and R-K Humphries (1997) Overexpression of HOXB3 In Hernopoietic Cells Causes Defective Lymphoic Development And Progressive Myeloproliferation, Immunity. 6: 13-22 4.1 Introduction

In the last Chapter it was shown that retroviral overexpression of one of the 5' locatec Hox genes, HOXAIO, perturbs the differentiation of both myeloid and B-lymphoic progenitor cells and eventually leads to the generation of acute myeloid leukemia These effects are qi le distinct from those previously reported for the 3' locatec HOXB4 gene, whose overexpression enhanced the expansion of primitivt hematopoietic cells, but neither altered hematopoietic differentiation nor predisposec to leukemia (Sauvageau et al., 1995). Together these results suggest that during hematopoiesis Hox genes can regulate different target genes, in both lineage- anc stage-specific manners, which might be reflected in their normal expression pattern.

To explore this concept further, the effects of overexpression of a third Hox gene the 3' located HOXB3 gene, were analyzed in a murine transplantation model. Thc HOXB3 gene was selected based on its distinct expression pattern in human bon€

marrow cells, where its expression is strictly restricted to LTC-IC enriched fraction 0 CD34+, which represents less than 1% of al1 nucleated cells found in the bone marro\ii (Giampaolo et al., 1994; Sauvageau et al., 1994). ln contrast to findings for HOXB4, overexpression of HOXB3 did not cause expansion of the most primitive hematopoietic cells (i.e. HSCs or multipotential progenitor cells). However, HOXB3 severely impairec both B and T lymphoid development, and enhanced proliferation of bi-potential CFU- GM progenitor cells. 4.2 Results

4.2.1 cDNA cloning of a HOXB3 transcript from hemopoietic cells anc generation of a HOXB3 retroviral vector

As is observed with many horneobox genes (Magli et al., 1991), HOXB3 has severa transcripts, some of which may be tissue specific (Sham et al., 1992). In an effort tc identify a HOXB3 transcript that is norrnally expressed in primitive blood cells, a 3.2 kt human HOXB3 cDNA was isolated from a CD34+ bone marrow expression library, The predicted coding region of this HOXB3 cDNA (Geneûank accession # U59298: shares > 97% identity at the protein level over the coding region to the mouse Hoxb-L previously isolated from ernbryonic tissue (Sham et al., 1992). The HOXB3 codinc sequence was subcloned into the MSCV 2.1 retroviral vector 5' to a phosphoglycerat~ kinase promoter (pgk)-driven neo gene such that HOXB3 expression was under the control of the regdatory elernents in the viral long-terminal repeat (LTR) (Figure 4. t A), lntegrity of this virus was shown both by the expression of the expected 4.1 kb LTR- derived full-length HOXB3 message in reconstituted hemopoietic tissue of mice transplanted with HOXB3-transduced bone marrow (Figure 4.1 0) and by the presence of an unrearranged 4.1 kb proviral fragment detected by Southern blot analysis ol genomic DNA isolated from bone marrow and thymus of such mice (Figure 4.35).

4.2.2 Hemopoietic reconstitution of recipients of HOXB3-transduced bone marrow cells

High-titer, helper-free viral producer cell lines for either the HOXB3 or a neo-control retrovirus were used to infect murine bone marrow cells as summarized in Figure 4.2. lmmediately following retroviral infection, a proportion of the bone marrow cells was plated under G418 selection in methylcellulose cultures supplemented with hemopoietic growth factors that support proliferation and differentiation of myeloid and erythroid progenitors (Figure 4.2). The gene transfer efficiencies as assessed by recovery of G418 resistant clonogenic progenitors were from 15 to 50% in the experiments. Analyses of the cellular content of HOXB3-transduced colonies (i.1 G418 resistant) showed al1 types of myeloid andfor erythroid colonies, suggesting th; HOXB3 overexpression was permissive for in vitro differentiation of already cornmitte myeloid progenitor cells (data not shown).

4.1 kb 1.3 kb m LTR I

LTR LTR I

Figure 4.1 Structure and expression of the HOXB3 and neo control retroviruses use in this study

(A) Diagrammatic representation of the integrated HOXB3 (upper) and neo (lower) proviruses. (E Northern blot analysis of 5 pg of total RNA extracted from splenocytes of mice 18 weeks afté transplantation with neo- (left lane) or HOXB3- (right lane) transduced bone mamw cells. The membran was sequentially hybridized to a probe specific for neo and $-actin and exposed for 48 hours. Two signal were obtained with the neo probe. The upper band represent long terminal repeat (LTR)-derive message of 2.7 and 4.1 kb for the neo and HOXB3 provirus respectively and the lower band cornmon t both constructs represents a 1.3 kb signal originating from the intemal pgk promoter. Evalulation of gene transfer to donogenic progeniton

5-FU Pm-stimulation (11-3 + 11-6 + Steel ) Methyiçellulose 2d /A= Determination of colony bnor (L~5-1+) B~~~ manow Viral infection < phenotype (CO-cuitivation) harvest 2d - \B 95QcGy - Evaiuation of various fi = Transplantation in hematopoietic cell recipients (Ly52+) cornpartment

Figure 4.2 Overview of experimental design

To explore in more detail the possibility that HOXB3 might affect hemopoiesis, reconstitution of lymphoid and rnyeioid populations in lethally irradiated mice transplanted with HOXB3-transduced bone marrow cells was examined. In 3 separate transplantation experiments, recipient mice (Ly5.2+) were injected with a life-sparing dose (2 x 105 cells) of congenic LyS.1+ bone rnarrow cells immediately following their infection with HOXB3 or neo retroviruses, as detailed in Figure 4.2. Based on the CRU quantitation in a similar population of cells (Sauvageau et al., 1995), it was estimated that each recipient received between 40 to 100 CRU cells, of which approximately 1/51 to 1/2 were estimated to be transduced, given the gene transfer efficiency to cionogenic progenitor cells in the transplant inoculum (see above). Donor-derived (Le., Lys. 1+) reconstitution of hemopoietic tissues was assessed 14 to 30 weeks after transplantation by FACS analysis (Figure 4.3A). Hemopoietic regeneration in either HOXB3 or neo mice was essentially completely donor-derived, as 180% of splenic, thymic and peripheral blood leukocytes were of donor origin (Ly5.1+, Figure 4.3A). The slightly lower levels that were detected in bone marrow of neo mice presumably reflect the presence of cells of the erythroid lineage which are negative for the expression of Ly5.1, rather than incomplete chimerism (Figure 4.3A). Furthermore, Southern blot analysis of DNA isolated from bone marrow and thymuses of these mice indicated a significant contribution by transduced cells to this regeneration, as evidenced by detection of the expected ne0 or HOXB3 provira signals at levels comparable to that of the endogenous HOXB3 gene (Figure 4.36).

PBL BW SPL THY

Figure 4.3 Donor-derived hemopoietic reconstitution of HOXBS and neo rnice (A) FACS analysis of proportion of donor-derived (Ly5.1+) cells in various hemopoietic tissues of HOXB3 (white bars) and neo (black bars) mice. Results shown are meanfSD for n r 6 mice (0)Southern blot analysis of DNA isolated from bone marrow and thymuses of HOXB3 (83-1 to 63-6)or neo (nml and nee2) rnice, sacrificed 18 to 25 week after transplantation. DNA was digested with Kpnl which cuts once in each long-terminal repeat (LTR) of the integrated neo or HOXB3 proviruses to release proviral fragments of 2.7 kb (neo) or 4.1 kb (HOXB3). Donor-derived reconstitution (i.e., %LyS.1) of the mice shown in this Figure varied between 86% to 98% for HOXB3 mice and between 78 to 94% for neo controls. The membrane was sequentially hybridized to probes specific for neo (specific activity -30 x la6 DPM, 70 hours exposure time) and HOXB3 (specific activity not measured, enposure tirne: 24 hours). The endogenous HOXB3 signal is detected at 1.4 kb. Abbrev. PBL. peripheral blood leukocytes; BM or 6, bone marrow; SPL, spleen; THY or T, thymus. 4.2.3 Recipients of H 0x83-transduced cells show increased granulopoiesis in their bone marrow and spleen

When analyzed 18-27 weeks after transplantation, age and sex-matched neo or HOXB3 mice had similar peripheral blood values (hemoglobin and white blood cells, data not shown) and bone marrow cellularity (Figure 4.4A). HOXB3 mice however showed a significant reduction in thymic cellularity and a gradua1 increase in spleen size over time (Figure 4.4A and 4-48). FACS analysis (Figure 4.5) and cytological examination (not shown) of spleen and bone marrow cells documented a consistent increase in mature granulocytes in these tissues. lncreased numbers of peripheral blood granulocytes were also observed in some HOXB3 mice (Figure 4.50, e.g., HOXB3#1 versus HOXB3#2). The absolute numbers of in vitro myeloid colony forming cells in bone marrow and spleen of HOXB3 mice were increased 3-fold (Figure 4.4C), consistent with the increase in bone marrow and spleen mature granulocytes in these two tissues. Significantly, 50-100% of these progenitors were G418 resistant, indicating that retrovirally transduced cells contributed substantially to these cell pools (Figure 4.4C).

Of the G418-resistant myeloid progenitors present in the bone marrow of these mice, between 10 to 100% generated small colonies (400-200 cells) containing a mixture of greater than 95% mature and immature granulocytic cells, a colony type rarely detected in cultures initiated with cells from neo mice. The cellular content of larger G418-resistant colonies generated from bone marrow of HOXB3 mice was similarly evaluated. Although al1 expected colony types were found (Le., CFU-GEMM, CFU-GM, CFU-M, CFU-G, CFU-Basophils-Eosinophil, BFU-E), the majority of these colonies were either restricted to the granulocytic andor macrophage lineages (46 of 54 large colonies examined from 3 mice). Thus only one colony out of 54 was of mixed lineage (CFU-GEMM), in contrast to 8 of 21 such colonies from neo mice. BM SPL THY C weeks post-transplantation T G418 resistant c 1 6a 100 not G418 resistant L EX- B CIO -O (P C) O k 1 83 ne0 83 neo 83 neo BM myeloid SPL myeloid BM pre-6 ceils Figure 4.4 Hemopoietic parameters of neo and HOXB3 mice

(A) Bone marrow, spleen and thymic cellularity of HOXB3 (white bars, n=13) and neo mice (black bars n=8) 18-27 weeks after transplantation. (6) lncrements in spleen weight in HOXB3 mice (black bars) witt time. (C) Myeloid and pre-B lymphoid clonogenic progenitor content in bone marrow and spleen of nec: (n=5) and HOXB3 (n=8) mice. Al1 results shown (A,B and C) represent mean values SD. ", significantl] decreased compared to neo control, p c 0.005; ', significantly increased compared to neo control, pcO.OE (two-tailed student T test with unequal variance) Abbrev.: BM, bone marrow; SPL, spleen; THY, thymus. A BonY-

hl," LX. KMac- 1

C mLymph Nodes

Figure 4.5 Flow cytometric analysis of various hematopoietic populations

Results from flow cytometric analysis of cells isolated fom bone manow (A) spleen (B) lymph nodl (C) and peripheral blood leukocytes (D) of mice transplanted 18 weeks earlier with HOXBS or ne transduced cells. HOXB3 #1 and HOXB3 #2 are 2 representative mice with moderate and seve phenotypes respectively. Numbers in boxes represent percentages of live cells found in this regio Lower number of cells were collected for analysis of lymph node populations (C). Together, these data suggest that HM3caused the expansion of more mature uni or bi-lineage myeloid progenitors but had minimal effect on the proliferation of earlier, multipotent cell types. This possibility was further assessed by limit dilution analysis to quantitate the number of long-term repopulating cells present in HOXB3 mice using the competitive repopulating unit (CRU) assay (Szilvassy et al., 1990). By 18 weeks post-transplantation, CRU frequencies in recipients of neo or HOXB3 transduced marrow were similar and, as expected for experiments using transduced rnarrow, had only regenerated to approximately 508% of normal levels (Table 4.1). Thus HOXB3 overexpression, in contrast to that of HOXB4 described previously (Sauvageau et al., 1995) and in Chapter 5, does not appear to enhance the regenerative potential of primitive long terni repopulating cells.

Table 4.1 Evaluation by limiting dilution analysis of competitive long-term repopulating cells (CRU) in primary mice tranrplented with HOXB3 or neo controi- transduced bone marrow cells, 18 weeks after transplantationa.

Number of cells injected into neo-transduced bone marrow HOXB3transduced bone marrow secondary recipients recipients recipients

CRU frequency per 105 cells O. 43 (95% CI) (0.28-0.99) Relative to normal (%lb 5.1 % 8.5% aResults are expressed as number of mice repopulated with donor-derived cells (Ly5.1+) over control. b~orrnalpretransplantation CRU values from n=2, 16 week old (PebC3)Fl mice = 8.5 (5.5-1 3.3) CRU/I~~bone marrow cells. Abbrev. CI, confident interval 4.2.4 Recipients of HOX83-transduced cells have defective T cell development

HOXB3 mice had significantly smaller thymuses than age and sex matched neo control recipients (Figure 4.4A) with corresponding abnomalities in the numbers and proportions of various subpopulations detected by flow cytometry (Figure 4.6). All recipients of HOXBSinfected cells had abnormal CD4/CD8 profiles as evident by a 24-fold decrease in the proportion of CD4+CD8+(double positive) cells and a 6-fold decrease in a single positive cell population (CD4-CD8+, Figure 4.6). In contrast, the proportion of CD4-CD8- (double-negative) and CD4loCD8- thymocytes were increased 3 and 2 fold, respectively (Figure 4.6 and data not shown). Although there was some heterogeneity among mice analyzed, in most mice (1 1 of 14) the majority of thymocytes were either CD4-CD8- or CD4loCD8-. These cells expressed little or no 8220 or Mac1 (in generai c 3% each) and, rnorphologicaily, were free of contaminating myeloid cells as assessed by cytospin preparations (data not shown). The dominant thymic CD4-CD8- and CD4loCD8- populations in these mice were heat- stable antigen-positive (HÇA~~)and expressed variable levels of CD25 (IL-2Ra, Figure 4.7 and data not shown). Curiously, the majority of these cells were y6 TCR+, whereas few were of the ap iineage (Figure 4.7). There was a tendency for mice analyzed at later times (i.e., >25 weeks post transplantation) to have a more severe thymic phenotype with one mouse analyzed at 14 weeks post transplantation having over 90% of CD4-C D8- or CD41oCD8- thymocytes (data not shown). HOXB3 overexpression was thus associated with an apparent block in thymocyte maturation from the double negative to the double positive stages coupled with expansion of the CD~~OCDB-~GTCR+ and CD4-CDB-yGTCR+ thymocytes (Figure 4.7). Ratio neo / HOXB3

Figure 4.6 Cellularity of thymuses and their subpopulations in individual neo and HOXB3 mice Results shown are cellularity of thymuses and their subpopulations for individual neo (cross) and HOXB3 (circles) mice sacrificed between 18 to 27 weeks after transplantation. Mean values are shown as horizontal lines. Ratios of mean values for each thymic population in neo and HOXB3 mice are indicated below with p values (two-tailed student T test with unequal variance). One HOXB3 recipient was omitted from this Figure because it had excessive values including 5.5 x 107 y6 T cells (see discussion). Three mice analyzed at 25-27 weeks post transplantation had >80% yS T cells in their thymus but as their exact thymic cellularity is not availabIe, these mice were not included in this Figure. Contaminant Mac-1 or 8220- positive cells were below 5Oh in al1 thymic samples used for these analyses. HOX63 Moderaie Phenotype

HOXB3 Severe Phenotype

CD4 Ly5.1 ap-TCR $-TCR IL-2R HSA

Figure 4.7 Cytofluorometric analysis of thymocytes subpopulations Thymocytes analysed were isolated from 1 representative neo mouse and 2 representative HOXf mice with modetate and severe thymic phenotype. CD4 and CO8 profiles of total thymocytes are shown the scatterplot (left). Each of the histograms (right) illustrates a FACS profile of 5.000 C04' CDB' (dou5 negative) cells analyzed for their expression of: ap and @-TCR, IL-2 receptor (CD25) and the heat stab antigen (HSA). Results for al1 neo and HOXB3 mice analyzed (imluding these) are presented in Figure 6. Although the intensity of the proviral signal seen on Southem blot analysis of DNA extracted from whole thymuses of HOXB3 mice was consistent with high level reconstitution by transduced cells (Figure 4.38, al1 thymuses > 80% CD4lo/-CD8- T cells), RTIPCR analysis was also performed to assess proviral expression in various thymic subpopulations of neo and HOXB3 mice. In contrast to neo control mice in which strong to moderate retroviral derived signals were detected in al1 subpopulations, for HOXB3 mice, retroviral-derived HOX83 mRNA were only detected at high levels in double-negative thymocytes, at low to moderate levels in CD4+CD8- (mostly CD4ioC~8+)and was not detectable in CD8+CD4+ and CD8+CD4- cells (Figure 4.8). These results thus suggest that HOXB3 overexpression is not compatible with progression from double negative to the double positive stage of thymic development and further indicate that the few double-positive thymocytes and their progeny found in some HOX83 mice were either derived from untransduced cells or HOXB3 transduced cells in which proviral expression was down regulated.

To assess whether the perturbations in thymic development in HOXB3 mice extended into the periphery, T cells in peripheral blood, spleen and lymph nodes were analyzed by flow cytometry (Figure 4.5). In al1 HOXB3 mice analyzed (n=6) there were normal distributions of CD4+ and CD8+ single positive cells and essentially no CD4+CD8+ T cells. Although these single positive cells were donor-derived, whether they arose from transduced or non-transduced cells was not ascertained. -Neo HOXB3

Neo

Figure 4.8 AT-PCR analysis purified thymic subpopulations Total RNA was isolated from different thymic subpopulations,l0,000 ceJls/sample, that were previously purified by FACS from thymuses ofHOX83 or neo mice (n=4 separate mice. only 1 representative of each is shown). The blot was sequentially hybridized with a probe specific for B-actin (lower panel) and neo (upper panel). Hybridization to a probe specific to HOXB3 gave results superimposable to those obtained with the neo probe (data not shown). Note that doublepositive (CD4+CD8+) and single positive CD4-CDBC cells from HOXB3-recipients do not express the retroviral- derived transcripts. Exposure times are 30 minutes for neo and 1 hour for actin.

4.2.5 Altered 8 cell development by overexpression of HOXB3

To detect possible effects of HOXB3 overexpression on B cell development, bone marrow IL-7-responsive B lymphoid clonogenic progenitor numben were measured. While the absolute numbers of these progenitors in HOXB3 mice were within the normal range ( -7000 per femur, Figure 4.4C) only a small proportion were G418 resistant. This contrasts with both the high proportion of transduced myeloid progenitors in these same HOXB3 mice and that of pre-6 progenitors in neo-control mice (Figure 4.4C). Furthemore, the rare G418 resistant colonies derived from the HOXB3 mice were only approximately one tenth of normal sire. Together these results suggest that HOXB3 overexpression inhibits early 6 ce11 development.

To characterize further possible B cell alterations in HOXB3 mice, phenotypically distinct 6 cell populations were evaluated by flow cytornetry using a combination of monoclonal antibodies (anti- CD45R, CD43, IgM and IgD) (Figure 4.5). Normal proportions of the B cell subpopulations were seen in the peripheral blood and lymph nodes of HOXB3 mice (Figure 4.5, profiles from 2 of 5 HOXB3 recipients analyzed 18 weeks after transplantation); their proportions were somewhat reduced in the spleen but taking into consideration the increase in splenic cellularity, the absolute number of B cells in this organ was within normal range. Abnomalities were apparent however in the bone marrow where total B cells were consistently reduced (Figure 4.5A) with marked variations (moderate phenotype, HOXB3#1 and severe phenotype, HOXB3#2). HOXB3 mice manifested a 2-1 0 fold reduction in total bone marrow 6 cells (6220+) with al1 B cell populations being affected in those mice showing a more severe phenotype (HOXB3#2, Figure 4.5). 4.3 Discussion

Overexpression of HOXB3 had multiple effects on hemopoiesis as assessed in a bone marrow transplantation model. Specifically, overexpression of HOXB3 leads tc an almost complete block in th8 thymic production of CD4+CD8+ T lymphocytes accompanied by an expansion of YS-TCR+ thymocytes, impaired B lymphoid development and enhanced rnyelopoiesis leading to a myeloproliferative disorder.

HOXB3-induced alterations were particularly prominent in thymocyte developrnent. The generation of mature ap-T cells were blocked by overexpression of HOXB3, as evidenced by reductions in double positive and single positive thymocytes. Furthermore, the few double positive thymocytes and their progeny found in thymuses of HOXB3 mice were negative for retroviral-derived HOXB3 expression representing either selection for transduced cells in which there was spontaneous downregulation of the LTR driven expression or contributions from untransduced cells. In contrast ta the block in ap-T cell maturation, the y5+ thymocytes in HOXB3 mice were expanded (on average -4 fold). This reached the extreme in one HOXB3 mouse which had a large thymus containing >97% HSA+y6 T cells. As we did not rule out the possibility that these cells had acquired pre-neoplastic properties as a result of additional somatic mutations, this mouse was not included in Figure 4.6. The expanded yG-TCR+ thymocytes in HOXB3 mice were distributed equally between CD4-CD8- and CD4ioCD8- populations and consisted predominantly of cells expressing high levels of

HSA, suggesting that.they represent an immature subset of y6 T cells (Suda and Zlotnik, 1993).

Together these results indicate that overexpression of HOXB3 has different effects on precursors of the ap TCR lineage than on those of the y6 TCR lineage, blocking the maturation of the former whereas inducing expansion of the latter. Whether these different effects are reflecting normal HOXB3 expression patterns in these cells is currently unknown. While we have shown using an RTIPCR approach that the murine Hoxb-3 is norrnally expressed at very Iow levels in FACS purified double negative, double positive and mature single positive thymocytes (Sauvageau et al., Unpublished data), its expression pattern in subsets of double negative cells is currently unknown. Further characterization of HOXB3 expression in these subsets will be of interest since this population contains both mature y6 T cells and precursors of both y6 and ap iineages (Dudley et al., 1995).

It is curious that the defect in thymocyte development is not compensated by the non-transduced donor cells. One possible explanation for this is that HOXB3- transduced progenitors have a superior thymus-repopulating ability and therefore occupy the majority of the limited nurnber of thymic niches (Spangrude and Scollay, 1990). Preferential niche occupation could also occur as a result of other mechanisms such as self-renewal advantages of HOXB3-transduced cells or active suppression of normal thymocyte development by HOXf33 transduced cells. This thymic defect does not appear to be due to replacement by myeloid cells, since no myeloid infiltration is seen cytologically and fewer than 5% Mac-l+ cells are seen on FACS analysis of the thymuses isolated from HOXB3 mice.

B lymphoid development was also irnpaired by HOXB3 overexpression. This was apparent from the virtual absence of transduced pre-B colony forming cells in bone marrow of HOXB3 mice and small colony size from the few such progenitors detected. However, in contrast to the findings in the thymus, the absolute number of pre-B colony forming cells in HOXB3 mice were within normal range presumably reflecting competitive repopulation from non-transduced cells. Some overall reduction in B cell content of the bone marrow was however seen, likely reflecting infiltration and displacement by the expanding granulocytic populations (Figure 4.5).

The defects in B and T cell development described in this manuscript have not been reported with any other molecules, including overexpression of another homeobox gene Hlx, which caused a release of immature CD4+CD8+ T cells in the periphery (Allen et al., 1995).

HOXB3 mice also developed a myeloproliferat ive disorder as evidenced by splenomegaly, a marked accumulation of myeloid clonogenic progenitors with granulocyte andhr macrophage differentiation potential and by granulocytic infiltration of bone marrow and other hemopoietic organs. For ail HOXB3 mice analyzed (n=4) this myeloproliferative disorder was readily transplantable to secondary recipients (data not shown). Most HOXB3 mice were sacrificed before 30 weeks and, in this tirne frame, progression of the myeloproliferative disorder to frank leukemia was not observed except for one mouse. Bone marrow clonogenic progenitor cells from this mouse were al1 G418 resistant and showed growth factor independence in the presence of fetal calf serum. Studies to assess characterization of this leukemia and further leukemic transformations that might develop in HOXBS mice, are currently underway. Although incornplete, these data nonetheless support the accumulating evidence of Hox gene involvement in both murine and human leukemic transformation (Borrow et al., 1996; Nakamura et al., l9Q6a;Nakamura et al., 1996b; Perkins et al., 1990; Thorsteinsdottir et al., 1996).

The results presented in this chapter stand in contrast to the findings previously described (Sauvageau et al., 1995) and presented in Chapter 5, that overexpression of HOXB4 did not detectably perturb myeloid, 8 or T cell differentiation but however induced expansion of myeloid and lymphoid progenitor cells and enhanced up to 47 fold the regeneration of the most primitive hemopoietic repopulating cell. These results also differ from findings presented in Chapter 3, that overexpression of HOXA 10 in murine bone marrow cells leads to enhanced formation of megakaryotic progenitors in vivo and in vitro, diminished nurnbers of macrophage and 8 lymphoid progenitors and generation of myeloid leukemias in a significant proportion of mice 5 to 8 months after transplantation. These findings thus demonstrate very distinctive effects that results from enforced expression of different Hox genes in marrow cells. These observations 1 O9 cornbined with previous study by Sauvageau et. al, which showed that Hox rnRNA levels differ significantly between various purified subpopulations of bone marrow cells suggest that, as in ernbryonic developrnent, Hox genes are important regulators of early hemopoietic developmental processes. Chapter 5

Enhanced polyclonal regeneration of hematopoietic stem cells overexpressing HOXB4 following bone marrow transplantations

3~hematerial presented in this Chapter is essentially as described in: U. Thorsteinsdottir, G. Sauvageau, M. R. Hough and R. Keith Humphries. (1997) Enhanced polyclona~ regeneration of hernatopoietic stem cells overexpressing HOXB4 following bone rnarrow transplantation. Manuscript in preparation. 5.1 Introduction

As discussed in detail in Chapter 1, overexpression of HOXB4 in murine bone marrow cells was found to greatly enhance the regeneration of the HSC compartment following bone marrow transplantation, without detectable effects on hematopoietic differentiation, or on mature end cell output (Sauvageau et al., 1995). These results thus implicated HOXB4 as a potential regulator of self-renewal of very early but not late hematopoietic cells. In these studies, the assessrnent of the effects of overexpression of HOXB4 were done at 20 weeks post transplantation, and at that time point the HSC compartment was regenerated slightly above normal pretransplantation levels. By serialiy transplanting these bone marrow cells, HOXB4 transduced HSC were demonstrated to still harbor extensive potential for expansion, as evidenced by their -1000-fold expansion in secondary recipients, which again brought the HSC pool in those mice near to pretransplantation levels. Based on these results, I hypothesized that this greatly enhanced expansion potential of HOXB4- transduced HSC would only be expressed in a highly proliferative environment, such as in the early phase of hematopoietic regeneration, whereas in steady state hematopoiesis their expansion would be constrained. The HSC pool in recipients of HOXB4-transduced bone marrow cells might thus plateau, presumably at pretransplantation levels, but would not become exhausted nor necessarily continue to expand. To test this hypothesis, I designed experiments to analyze the size of the regenerated pool of HSC in mice transplanted with HOXB4transduced bone marrow cell, as a function of time after transplantation for a period extending up to a year after transplantation. Another unresolved issue from the initial studies on the effects of HOXB4 was the actual contribution to this expansion by individual HSC, i.e was HOXB4 primarily acting on one or few HSC, or on a broad spectrum of HSCs. In this study I have addressed that issue by analyzing the degree of polyclonality in the regenerated pool of HSC concurrently with the quantitation of the size of the pool. The results presented in this chapter confirm Our previous finding tha overexpression of HOXB4 enhances the expansion of HSC following bone rnarrol transplantation and, moreover, show that in vivo expansion of HOXB4-transducec HSCs is subjected to environmental control rnechanisrns, as the pool size of HOXB4 transduced HSCs eventually stabilizes when it reaches that typical of normal (pre transplant) mice. lncreased HOX54 expression also appeared to have a pronouncec effect on many HSCs, as the regenerated pool of HOXB4-transduced HSC ir transplanted mice was highly polyclonal. Furthermore, detailed analysis of various later cell populations in these mice strongly suggests that overexpression of HOX& does not overtly alter myeloid or lymphoid differentiation, nor lead to dominans outgrowth of any type of hematopoietic cell. 5.2 Results

5.2.1 Experimental model

To study the effects of increased and extended expression of HOXB4 on hematopoiesis, the HOXB4 cDNA isolated from human fetal liver cells (Piverali et al., 1990) was introduced into murine bone marrow cells using retroviral-mediated gene transfer. The murine stem cell virus (MSCV) 2.1 retroviral vector was used (Hawley et al., 1992) in which the HOXB4 cDNA was inserted 5' to the phosphoglycerate kinase promoter (PGK)-driven neo gene such that the HOXB4 expression was under the control of the regulatory elements within the MSCV 2.1 long terminal repeat (LTR) (Figure 5.1A). The LTR sequences in this MSCV vector have been shown previously to give high and long-term expression both in primitive murine hematopoietic cells and their mature progeny (Hawley et al., 1994; Pawliuk and Humphries, Unpublished data; Sauvageau et al., 1995). The MSCV PGK-neo vector lacking the HOXB4 cDNA was used as a control (Hawley et al., 1992). The generation of high titer, polyclonal and helper-virus free HOXB4-viral producer cells, was carried out as described in Chapter 2. lntegrity of the HOXB4-retrovirus was verified both by the detection of unrearranged 3.9 kb proviral fragment by Southem blot analysis of DNA isolated from bone marrow of HOXB4 mice (Figure 5.1C), and by Norihem blot analysis of RNA isolated from the same tissue, which detected the full length HOXB4 LTR-driven mRNA (Figure 5.3).

To assess the possible effects of HOXB4 overexpression on hematopoietic regeneration, lethally irradiated mice were transplanted with HOXB4- or neo- transduced bone marrow cells. Cohorts of mice from three independent transplantation experiments (hereafter called HOXB4 and neo mice, respective!y), were assessed for regeneration of various hematopoietic compartments at different times after transplantation, beginning as soon as 16 weeks and as late as 52 weeks after transplantation (Figure 5.18). In an effort to achieve high retroviral transduction to primitive hematopoietic cells, bone marrow cells from mice treated 4 days previously with 5-fluorouracil (5-FU) were CO-cultivatedwith HOXW or neo- viral producer ceIl' for 48 hours prior to transplantation. Recovered cells were transplanted without pre selection into lethally irradiated recipients at a dose of 2x1 O5 cells/recipient, estirnatec to contain -30-40 HSCs (Sauvageau et al., 1995). The gene transfer efficiencies to thi transplanted bone marrow as assessed by the proportion of G418 resistant clonogenic cells varied between experiments, and were 30 to 58% and 70 to 74% for HOXB4- an( neo-transduced cells, respectively (Figure 5.1B). Assuming a retroviral infectioi efficiency of HSCs no greater than that of clonogenic progenitor cells, each recipien would have received an estimated maximum of 15-30 transduced (neo or HOXB4 HSCs, plus an approximately equal number of non-transduced HSCs. LTR PGK-im, ni0

insinant CFC tlme port TI an0 numbir al mka

neo HOXB4 ------32-A 32-8 32-A 32.8 42-A 52-A 52-8 BTBTB TBTBf 8TBT

Figure 5.1. Structure of the HOXB4 and control neo retroviruses and the exparimental outline

(A) Diagrammatic representation of the integrated HOXB4 and neo provinises. Expected size of the full-length viral transcripts and also those initiated from the PGK promoter are shown, as are the sites for the various restriction enzymes used in this study. (B) Experimental outline showing the number of HOXB4 and neo mice from 3 transplantation experiments that were used in this study, and the time post transplantation when they were analysed. Also shown is the initial gene transfer to the transplanted bone marrow inoculum received by neo and HOXB4 mice in these transplantations. (C) Southern blot analysis of ONA isolated from bone marrow and thymus of some of the neo (sacrificed 32 weeks post -transplantation) and HOXM mice (sacrificed 32,41 and 52 weeks post transplantation) used in this study ,ta demonstrate the presence of the integrated provirus. DNA was cut with Kpnl, which releases the neo (2.7 kb) and the HOXB4 (3.9 kb) proviruses, and the blot was successively hybridized to probes specif~ for the nea and HOXB4 genes (full length HOXBQ cDNA was used as a probe). The endogenous murine HOXB4 is detected at 1.3 kb by the HOXB probe and pmvides a single gene copy control of loading. In some of the HOXB4 mice, in addition to the full length HOXB4 provirus, a weaker 2.7 kb proviral signal is detected with the neo probe but not with the HOXB4 probe. This probably represents the loss of the HOXB.4 gene from a few of the integrated provinises. Abbrev. CFC, colony forrning cells; 6, bone marrow: T, thymus. 5.2.2 Enhanced regeneration of CRU cells in mice transplanted with HOXB44ransduced bone marrow cells

Hematopoietic regeneration in both neo and HOXB4 mice, from al1 3 transplantatior experiments, was essentially completely donor-derived as >85% of bone marrow spleen, thymic and peripheral blood leukocytes were of transplant origin (Ly5.V) at al times analyzed. A major contribution by transduced cells to this reconstitution wac evident by Southern blot analysis that teadily detected the neo- or the HOXB4- proviruses in the bone marrow and thymuses of these mice (Figure 5.1C).

To determine the effects of HOXB4 overexpression on the size of the regeneratec pool of HSCs as a function of tirne, HSC numbers in bone marrow of neo and HOXB4 mice were quantitated using the CRU assay (described in detail in Chapter 1) at the various time points outlined in Figure 5.16. The CRU numbers in bone marrow of nec control mice were quantitated 16, 20 and 32 weeks post transplantation. At al1 of thess time points the CRU pool was reconstituted to levels ~10%of normal pretransplantation CRU values (Table 5.1). This Iow regeneration of CRU numbers aftei transplantation is consistent with previous studies by others following transplantation of non-retrovirally infected normal or 5-FU-treated adult bone marrow cells, where HSC levels did not regenerate above 10% of normal pre-transplantation values, as assessed either by the CRU assay or Harrison's Cornpetitive repopulating assay (described in Chapter 1) (Harrison and Astle, 1982; Harrison et al., 1990; Pawliuk el al., 1996). Table 5.1. CRU numbers regenerated in primary recipients of neo- or HOXB4- transduced bone marrow cells

neo -transduced bone marrow HOXB4-transduced bone marrow recipients recipients Weeks Total CRUIfemur % of normala Total CRUffemur % of normal a post Tx (95% CI) levels (95% CI) levels

Experiment 1 20"

Experiment 2 7eb

Experiment 3 41%

aNormal pretransplantation CRU values from n=2. 16 week old (PebC3)Fl mice=1667(1042- 2667) CRUIfemur b~tboth these time points CRU numbers where evaluated in bone marrow cells pooled from three neo and threeHOXB4 mice CCRUnumbers were evaluated in n=t HOXB4 mouse d~~~ numbers were evaluated in n=l neo and in individually in n=2 HOXB4 mice, and their values were then averaged eCRU numbers were evaluated in individual n=2 HOXB4 mice, and their values were then averaged Abbrev. Tx, transplantation; CI, confident interval

CRU numbers in HOXB4 mice from al1 three transplantation experiments, ir contrast, were regenerated to levels near to or slightly above norma pretransplantation values, or with on average some 14-fold greater numbers of CRL than in neo control mice (Table 5.1). In both experiments 1 and 2 the frequency of CRL cells in HOXB4 mice were similar at the early ( 20 or 16 weeks) and late time points (32 or 52 weeks) after transplantation (Table 5.1). This suggests that the enhancec expansion of CRU cells occurs relatively early during hematopoietic regeneration, witi- stabilization in pool size at normal levels. Similarly, in HOXB4 mice frorr transplantation experiment 3, which were anatyzed 41 weeks post transplantation, tht CRU frequency reached that of normal unmanipulated mice Fable 5.1).

5.2.3 Polyclonal hematopoietic regeneration by HOXB4-transduced cell! both at the levels of CRU and mature end cells

Hematopoietic regeneration by transduced cells in neo and HOXB4 recipients wa! further assessed by Southern btot analyses of proviral integration sites (clonality) ir DNA isolated from hematopoietic tissues of neo and HOXB4 recipients initiall! transplanted with neo or HOXB4-transduced bone marrow cells (primary recipients) and those which were transplanted with various dilutions of bone marrow from thest pflmary recipients for the CRU quantitation (secondary recipients). In the bone marrow spleen and thymus of primary HOXB4 recipients, sacrificed 32 and 52 weeks pos transplantation, multiple proviral integrations could be detected (Figure 5.2A and B) Since proviral signais of different intensities couid be detected in the same tissues these multiple integrations most likely reflect polyclonal regeneration by HOXB4 transduced HSCs, rather than multiple proviral integrations into the same HSC. Thi: interpretation was further supported by clonal analysis of the secondary transplan recipients (see below). Thus, even as late as 52 weeks post transplantation, th€ hematopoietic regeneration in the primary HOXB4 recipients was polyclonal an( without apparent dominance of any HOXB4-transduced clone. As the quantitation of CRU cells in primary mice was based upon injecting CR1 cells at a limiting dilution into secondary recipients, Southem blot analysis of provira integrations in thymus (lymphoid) and bone marrow (myeloid) of these secondaq recipients couid be used to analyze further the clonality of the regenerated pool O transduced lympho-myeloid repopulating (CRU) cells in primary HOXB4 and nec mice. Of the 10 secondary recipients (32-1 to 32-1 0) that were positive for lympho, myeloid repopulation by donor cells from one of the primary HOXB4 mice sacrificed 31 weeks post transplantation, 8 had detectable HOXB4 proviral integration in their bon< marrow and thymus (Figure 5.2A). In the case of the two "negative" mice (32-5 and 32s 119 IO), which both had low donor repopulation (Ly5.1+ PBL, 4 and 5 %), the expression of HOXB4 could however be detected by Northem blot analysis of their bone marrow (Figure 5.3), suggesting that this provided a more sensitive measure of the presence of HOXB4 than the Southem blot analysis. The one mouse analyzed (32-1 1) that was negative for donor derived repopulation was also negative for proviral integration by Southern blot analysis (Figure 5.2A). There was thus complete concordance between detection of donor derived hematopoietic regeneration and contribution to regeneration by HOXB4 transduced cells in secondary mice. In essence, al1 donor- derived regeneration in the primary mouse must thus have been derived from HOXW transduced stem cells. Of the secondary mice analyzed, several, including those which were estimated to receive between 1 and 2 CRU cells (mice 32-6, 32-8 and 32-9), had a common proviral integration in their bone marrow and thymus, thus confirming the lympho-myeloid repopulating potential of the regenerated HOXB4-transduced CRU cells (Figure 5.2A). Self-renewal of HOXB4-transduced CRU celis was ais0 demonstrated by detecting a common lympho-myeloid repopulating clone in al1 of the secondary recipients (32-1 to 32-4) receiving high cell dose (43 CRU/mouse) and in one mouse (32-6) receiving fewer CRU cells (2 CRU/mouse). Interestingly, this clone could not be detected in bone marrow, spleen or thymus of the primary HOXB4 mouse (Figure 5.2A), indicating that this cell, despite extensive self-renewal division, did not contribute significantly to bone marrow, spleen or thymic repopulation in this primary HOXB4 recipient. The detection of 3 different HOXB4-transuced CRU clones, and at least 7 others with either lymphoid or myeloid potential in these secondary HOXB4 mice, together with detection of other HOXB4-transduced CRU clones in the primary recipient, strongly suggest that the enhanced CRU regeneration in the primary HOXB4 mouse sacrificed 32 weeks post transplantation was a polyclonal event. Figure 5.2 Southern btot analysis of proviral integration patterns in primary nso and HOXB4 mice sacrlficeâ 32 weeks (A) or 52 weeks (6) past transplantation and theii secondary reciplenti from the CRU assay. DNA was digested with BarriHI (A) or EcoFll (B), that cuts the integrated provirus once, generating a DNA fragment specific for each proviral integration site. The membranes were first hybridized to a nea specific probe for detectiun of proviral fragments and subsequently with a probe specific for the SHlP gene Io provide a single wpy control of loading. Exposure times were 48 hours for the neo and SHlP probes. To demonstrate that the proviral fragments contained the HOXB4 cDNA the blots were also hybridized with full-length H-4 cDNA probe, which generated the same proviral banding pattern as the neo probe (data not shown). Each mouse is identified with a specific number derived from the time that the pnrnary recipient was sacrificeci and above that number are indicated the number of bone mamw cells received by each secondary recipient and the estimated CRU content of each inoculum. Percentage of donor derived leukocytes (Ly5.1+) in the peripheral blood of the secondary neo mice was: 32-1 (44% ), 32-1(1 8%) and that of secondary HOXB4 mice: 32-1(6O%), 32-2(76%), 32-3(83%), 32-4(52%), 32-5(4%), 32-6(23%), 32-7(18%), 32-8(9%), 32-9(51%), 32-10(5%), 32-1 1(O%), 52-1 (5%), 52-2(66%), 52-3(28%), 52-4(47%), 52-5( 5%), 52-6(36%), 52-7(47%), 52-8(20%) and 52-9(50%). Abbrev. B, bone mamw; T, thymus. To determine if the polyclonality of the CRU pool was maintained with time in primary HOXB4 mice, proviral integration was also assessed in DNA isolated from bone marrow and thymus of those secondary HOXB4 mice regenerated with bone marrow cells from the primary HOXB4 mouse sacrificed 52 weeks post transplantation (Figure 5.28). In al1 of the 9 secondary HOXB4 recipients that were scored positive for donor derived regeneration, provirai integration(s) were detected in their bone marrow andor thymus (Figure 5-28). Of those, 5 had a common proviral pattern in their bone marrow and thymus (mice 52-2, 52-4, 52-6, 52-8, and 52-9 (bone marrow signal very faint)), indicating again the lympho-myeloid repopulating potential of the regenerated CRU cells. Interestingly, the proviral integrations in the regenerated pool of HOXB4- transduced repopulating cells in this primary HOXB4 mouse were more complex than in the HOXB4 mouse analyzed 32 weeks post transplantation (Figure 5.2A and 5.28). This was evident both by the numbers of transduced clones detected, where a common HOXB4-transduced clone was only detected in thymuses of two secondary mice (Le. mice 52-2 and 52-3) and by the complexity of proviral banding patterns of individual clones (Figure 5.2B). This difference is likely a reflection of higher gene transfer to the transplant inoculum received initially by the HOXB4 mice analyzed at 52 weeks than by those analysed 32 weeks after transplantation (Figure 5.18). Furthermore, these results, together with polyclonal regeneration of bone marrow, spleen and thymus of the primary HOXB4 mouse by transduced totipotent cells, indicate that even as late as 52 weeks after transplantation, the CRU pool is highly polyclonal.

By Northern blot analysis, the HOXB4 message could readily be detected in bone marrow of both primary HOXB4 recipients at 41 weeks post transplantation, and in secondary recipients 13 weeks after transplantation of bone marrow cells from primary HOXB4 recipients sacrificed 32 or 52 weeks after transplantation (Figure 5.3). Thus, the HOXB4 message was clearly expressed at high levels in HOXB4 recipients, without any evidence for promoter shutdown, for more than a year after transplantation. As seen in Figure 5.3, the strength of the HOXB4 message variel between the mice analyzed, and, as expected roughly correlated with their levels c donor derived repopulation (%Ly5.1).

l0 H0)W 2O HOXB132 W. post Tx 20 HOXB452w.~Tx 41 W.- post lx

Figure 5.3 Northern blot analyses to detect expression of the transduced HOXB4 gene in bone manow of HOXB4 mice

Total RNA (-5pg) was isolated from bone marrow of primary HOXB4 mice sacrificed 41 weeks post transplantation and those secondary recipients that received varying numbers of bone marrow cdls from the primary HOXB4 mice sacrificed 32 and 52 weeks post transplantation. The blots were sequentially hybridized to HOXB4specific probes to detect the viral transcript (3.9 kb) and to probe specific for actin to assess for loading. Each number assigned to individual lanes identifies a specific mouse which cm also be identified in Figure 2 on the Southern blot analysis. The percentage donor derived repopulation in the secondary HOXB4 mice analysed were: 32-1 (60%), 32-2(76%), 32-3(83%), 32-4(52%), 32-5(4%), 32- 6(23%), 32-7(18%), 32-8(9%), 32-9(51%), 32-1 0(5%), 52-1 (5%), 52-2(66%), 52-3(28%), 52-4(47%), 52- 5( 5%). 52-6(36%) and 52-7(47%).

5.2.5 HOXB4-induced expansion in vivo of clonogenic progenitor cells

Elevated numbers of rnyeloid clonogenic progenitors were also detected in HOXB4 mice compared to contrd neo mice (Figure 5.48). This was most prominent in the spleen. At al1 time points, myeloid progenitors were significantly increased in spleens of HOXB4 mice, compared to those of neo control mice, with the greatest expansion documented being -40-fold over control (Figure 5.4). In mice from al1 three transplantations, the splenic progenitor nurnbers increased with time after transplantation, and the magnitude of this expansion appeared to correlate with the initial HOXB4 retroviral transduction efficiency (Figure 5.4 and Figure 5.1 8).The bone marrow rnyeloid progenitor numbers were also increased in HOXB4 mice compared to neo control mice, but this increase was not statistically significant except in mice analyzed 20 weeks post transplantation (Figure 5.4). In both the bone marrow and spleen of HOXB4 mice, the majority of myeloid progenitors were G418 resistant (68+18% and 75+30%, respectively), likely reflecting preferential expansion of HOX54-transduced cells over untransduced cells. Overall, the total body increase in myeloid progenitor numbers was less than 2-fold in the majority of the mice analyzed, with the exception of those analyzed 20 weeks post transplantation, where the increase was -5-fold. Overexpression of HOXB4 was not found to cause preferential expansion of any one of the different types of myeloid progenitor (i.e. CFU-M, CFU- GM, BFU-E, and CFU-GEMM), as transduced progenitors from HOXB4 mice generated the various in vitro colony types with the same relative frequencies as those derived from neo control mice (data not shown). Evaluation of bone marrow pre-B lymphoid colony forming cells at various times after transplantation showed equivalent numbers in HOXB4 and neo mice at an early time point (16 weeks after transplantation), but elevated numbers (2 to 3-fold) in HOXB4 mice at later time points (32 and 41 weeks post transplantation) (Figure 5.4). As seen for the myeloid progenitor cells, a major proportion of pre-6 progenitor cells in HOXB4 mice were 6418 resistant (HOXB4 rnice=59+9% vs. neo mice=37+10%), consistent with their preferential derivation from HOXBetransduced cells. 1000000 Myeloid CFCJfernur neo

- - - 20 52 16 32 41 Experiment 1 Experiment 2 Experirnent 3 I Myeloid CFCIspleen

20 52 16 32 41 Experiment 1 Experiment 2 Experiment 3 100000 , 1

20 52 16 32 41 Experiment 1 Experiment 2 Experiment 3 Weeks post transplantation Figure 5.4 Effects of HOXB4 overexpression on the expansion of myeloid and pre-B colony forming cells following bone marrow transplantation Results shown are means kSD of the numbers of in vitro myeloid colony forming cells in bone marrow (top) and spleen (middle) and of pre-B colony forming cells in bone marrow (bottom) of individual neo and HOXM mice, at various tirne points after transplantation. The number of neo and HOXB4 mice analyzed at each time point are shown in Figure 1B. Bone marrow myeIoid colony forming cells in HOXB4 mice were only significantly increased over that of neo control mice in experiment 1 at 20 weeks post transplantation (pe0.05). Number of splenic myeloid colony forming cells were significantly increased at al1 time points in HOXB4 mice (p<0.05). Pre-B colony forming cells were not significantly increased in HOXB4 mice at any time point. Abbrev. CFC, colony forming cell. 5.2.5 Enhanced regeneration of HOXB4-transduced CRU cells is nat reflected in the number of the regenerated Scal +Lin-WGA+ cells

Murine bone marrow cells that express the Ly6AiE antigen (Scal+), but are negative for various lineage markers (8220, Ly-1, Gr-1 and Mac-1) and bind to lectin wheat germ agglutinin (WGA+), have previously been shown to be enriched -400-fold for cells with long term lympho-myeloid repopulating potential when compared to unfractionated bone marrow (Rebel et al., 1994). To analyze whether the enhanced regeneration of CRU cells in HOXB4 mice was reflected in the size of the regenerated pool of Scal+Lin-WGA+ cells, the numbers of these cells were measured in bone marrow of HOXB4 and neo rnice 32 and 52 weeks post-transplantation. In both neo and HOXB4 mice, >90% of Scal+Lin- cells were donor-derived (LyS.l+). Interestingly, the proportion and absolute number of Scal+Lin-WGA+cells were higher in the bone marrow of neo mice than in HOXB4 mice and thus in contrast to the CRU nurnbers (Table 5.2). This suggests that HOXB4 expansion of CRU is not tightly coupled to expansion of phenotype.

Table 5.2 Regeneration of ScaVLin-WGA+ cells in neo and HOXB4 micea

neo HOXB4

32 weeks post Tx 32 weeks post Tx 52 weeks post Tx

% Scat +Lin-WGA+ cells 0.073+0.0042 0.01 5+0.0014 0.041 f0.024 in total bone marrow

Number of Scal +LinSWGA+ 15450f5020 281 0f580b 72303c2079 celldfemur

CRU numbers/femur 128 1300 1062

aResults shown are means f SD for n=2 neo mice and n=2 HOX84 mice 32 and 52 weeks pOSt transplantation b~ignificantlylower than in neo control mice, pc0.05,students t-test Abbrev. Tx, transplantation; #, number 5.2.6 Lack of HOXB4 effects on mature end cell output In contrast to increases in both CRU and clonogenic progenitor numbers, the peripheral blood white and red blood cell counts in HOX84 mice analyzed at early and late time points were within the normal range (Table 5.3). The same was true for their femoral, thymic and splenic nucleated cell counts, with the exception of the two mice analyzed 52 weeks post transplantation, which had moderate splenornegaly (Table 5.3).

Table 5.3 Hematological parameters in HOXB4 and neo mice 16-52 weeks post transplantationa

Peripheral blood Weeks Mice TNClfemur TNC/thymus TNC/spleen RBC WBC POS~TX (XI 07) (XI 07) (XI 08) (XI 09/m1) (XI 06/m1)

Experiment 1 20 neo (n=3) HOXB4 (n=3)

52 HOXB4 (n=2)

Experiment 2 16 neo (n=3) HOXB4 (n=3)

32 neo (n=2) HOX84 (n=2)

Experiment 3 4 1 neo (n=2) HOX84 (n=2)

aResults shown are meanISD for the indicated number of neo and HOXB4 mice. Abbrev. Tx, transplantation; TNC, total nucleated cell counts; RBC, red blood cells; WBC, white blood cells

To analyze further the possible effects of HOXB4 overexpression on the cellular constitution of hematopoietic organs, the absoiute numbers and proportions of various Iyrnphoid and myeloid cell populations in bone marrow, spleen and thymus of HOX84 and neo mice were assesed by flow cytometric analysis. As shown in Table 5.4, thc absotute numbers of bone marrow B and myeloid celt subpopulations, and of splenic B, T and myeloid cell subpopulations in HOXB4 mice analyzed 32 weeks posi transplantation, were al1 comparable to that of neo control mice, as were their thymic CD4 and CD8 subpopulations. Similar results were also obtained for the HOXB4 mice analyzed 16 and 52 weeks post transplantation except for the two HOXB4 mice sacrificed 52 weeks after transplantation with splenomegaly. These mice had an elevation in al1 splenic B,T and myeloid cell subpopulations assessed, consistent with the increase in their spleen cellularity (i.e. -2- and -5-fold).

Table 5.4 Absolute numbers of various phenotypically defined hematopoietic populations in neo and HOXB4 mice 32 weeks after transplantationa

Mice Bone marrow

B220+ B220+/CD43+ B220+11gM+ Mac1+ (XI 06) (x1 06) (x1 06) (XI 06)

neo HOXB4

Spleen

neo HOXB4

Thymus

CD4'CDB' CD4+CD8+ CD4+CD8' CD4'CDB+ (XI 07) (xi 07) (XI 07) (XI 07) ne0 HOXB4

aResults shown are meartkSD for n=2neo and n=2 HOX84 mice. 5.3 Discussion

This study confirmes our earlier findings that overexpression of HOXB4 enhances the regeneration of the HSC compartment following transplantation. Moreover, measure of the numbers of regenerated HSC at various times after transplantation, revealed stabilization at normal pre-transplantation levels, indicating that in vivo the expansion of HOXB4-transduced HSC can bs regulated by in vivo control rnechanisrns. HOXB4 was also demonstrated to act on multiple HSCs, as the regenerated pool of HOXBC transduced HSCs in mice transplanted with HOXB4-transduced bone marrow cells was highly polyclonal, even as late as one year after transplantation. Furtherrnore, detailed analysis of various later cell populations in these mice strongiy suggests that overexpression of HOXB4 does not overtly alter myeloid or Iymphoid differentiation, nor lead to dominant outgrowth of any type of hematopoietic cells.

Previous studies have shown that there is only limited recovery of the HSC pool in mice transplanted with normal bone marrow cells, in contrast to complete regeneration of progenitor cells and more mature cells to normal levels (Harrison and Astle, 1982; Harrison et al., 1990; Pawliuk et al., 1996). This has been demonstrated both by competitive repopulating studies, where the competitive repopulating potential of transplanted bone marrow was found to be reduced -10-fold relative to normal bone marrow (Harrison and Astle, 1982; Harrison et al., 1990), and more recently by limit dilution analysis, which showed that the CRU numbers following single transplantation only recover to -10% of pretransplantation levels (Pawliuk et al., 1996). It has been suggested that the transplantable HSCs may fail to fully regenerate the HSC compartrnent because of their inherently limited capacity for self-renewal andor the sustained proliferative stress imposed on these cells during the early phase of hematopoietic regeneration, resulting in stimuli that could favor differentiation rather than self-renewal responses andor as a result of negative feedback mechanisms that prematurely inhibit their expansion (Harrison et al., 1990; Pawliuk et al., 1996). In mice transplanted with HOXB44ransduced bone marrow cells, there is in contrast, regeneration of HSC (CRU) numbers to normal pre-transplantation levels, when analyzed over periods extending from 16 to 52 weeks after transplantation. Although mice in these experiments received a transplant inocula of which likely 40% of total CRU were transduced, analysis of the regenerated pool of CRU cells indicated that the vast majority of the CRU cells in these HOXB4 mice were positive for the presence of the HOXB4 provirus. This was evidenced by detecting the HOXB4 provirus in ail of the secondary recipients of bone rnarrow cells frorn these primary transplant recipients, including those that were estimated to have received 1 to 2 CRU cells. Thus, an enhanced regeneration of CRU in HOXB4 rnice appeared to be an intrinsic property of HOXB4-transduced CRU, conferring on thern the capacity to outcompete the regeneration of untransduced CRU in the post-transplant period. Moreover, clonal analysis indicated that this property was not limited to selected CRUs, as hematopoiesis in HOXB4 recipients was highly polyclonal.

One explanation for the enhanced regeneration HOXB4transduced CRU is that the initial CRU pool, prior to their expansion in the mice, was larger in the HOXB4 recipient than in the neo control mice, due to overexpression of HOXB4 enhancing either the recruitment of CRUs in the transplanted inocula or the seeding efficiency of CRU cells to appropriate locations following transplantation. To evaluate the contribution of such factors requires knowledge about the exact numbers of CRU cells received by each recipient, so the expansion per CRU car be estimated. Although the number of CRU received by primary recipients was not evaluated either in this study or in Our previous study when HOXB4-transduced bone marrow was serially transplanted (Sauvageau et al., 1995), the numbers received by secondary recipients in that study were known. There secondary neo and HOXB4 recipients were transplanted with cell doses estimated to contain equivalent nurnbers of cells detectable as CRU (2-5 CRU per mouse), thus eliminating differences due to HOXB4 enhancing recruitment of CRU. When analyzed 16 weeks after transplantation, the expansion per CRU in secondary HOXB4 recipients was estimated to be -50-fold higher than in neo control recipient, which is much greater expansion than could be accounted for by increased seeding efficiency alone (Sauvageau et al., 1995). Other possibilities are that HOXB4 enhances the probability of HSCs to execute self-renewal divisions or increases the self-renewal potential (Le. atlows for more self-renewal divisions) and/or proliferation (Le. increased number of cell divisions) of HSCs. One way to deterrnine whether HOXB4 enhances the probability of CRU cells to undergo self-renewal divisions, would be to analyze the size of the CRU pool at very early tirne points after transplantation, when the CRU regeneration is in log phase. If HOXB4 increases the self-renewal probability of CRU cells, HOXB4-transduced CRU woutd expand their numbers faster than control CRU cells. Another explanation for the enhanced regeneration of HOXB4transduced HSC is that overexpression of HOXB4 reduces the sensitivity of HSC to negative feedback mechanisms that might normally prematurely inhibit the expansion of HSC. However, the enhanced ability of HOXB4-transduced progenitor cells to generate secondary colonies in vitro upon replating as we previously reported (Sauvageau et al., 1995), suggest that more likely explanations are those discussed above, involving enhanced HSC self-renewal.

In Our previous study on the effects of overexpression of HOXB4 we showed by serial transplantation that the CRU regenerated in primary recipients retained the capacity for extensive regeneration upon further transplantation into secondary recipients. In both primary and secondary HOXB4 recipients the CRU levels reached near normal values or with an overall -2500-fold greater net expansion over serially transplanted CRU in neo control recipients (Sauvageau et al., 1995). These serial transplantation studies thus demonstrated that the HOXB4-transduced CRU cells in the primary HOXB4 recipients still harbored extensive expansion potential. To assess how this enhanced expansion potential of HOXB4-transduced CRU cells would affect their behavior in steady state hematopoiesis, the CRU frequency in two cohorts of primary HOXB4 recipients were measured at early (16 or 20 week) and late (32 or 52 weeks) time points after transplantation. Interestingly, the CRU frequency was similar in both cohorts of mice and at both time points, with apparent stabilization in the size of the CRU pool around normal adult mouse marrow levels. Thus, the population of HOXB4-transduced CRU cells did not continue to expand nor become exhausted in the primary HOXB4 recipients. These data point to the existence of regulatory mechanisms that can control the size of the CRU pool and which overexpression of HOXB4 does not override. Shutdown of HOXB4 expression in CRU cells could explain the observed stabilization in the size of the CRU pool. The detection of the HOXB4 message in bone marrow of secondary recipients repopulated with as few as -1 CRU cells originating from primary HOXB4 recipients sacrificed 32 or 52 weeks after transplantation, strongly indicates that the HOXB4 message was being expressed at high levels in CRU cells as late as 52 weeks post transplantation, and promoter shutdown is thus an unlikely explanation for the stabilization in the CRU pool. In mice, the number of HSCs remains constant throughout their adult live, indicating that during steady state in vivo hematopoiesis the proliferation of HSCs is controlled (Harrison and Astle, 1982; Harrison and Lerner, 1991 ; Ogden and Micklem, 1976). These likely involve mechanisms that keep the majority of HSC out of cycle, as many of these cells are highly resistant to cytotoxic drugs such as 5-FU (Harrison and Lerner, 1991). The stabilization of the CRU pool in HOXB4 mice at normal levels, suggests that although CRU cells overexpressing HOXB4 have enhanced regenerative potential, their ability to respond to this regulatory mechanism may not be altered. However, as the cycling status of CRU cells in HOXB4 recipients is currently unknown, other regulatory mechanisms acting to maintain stable levels of HOXB4-transduced CRU cells cannot be ruled out.

Overexpression of HOXB4 also promoted expansion of intermediate types of progenitors in HOXB4 mice to levels above normal values. Whether this expansion was due to an increased input from more primitive cells or increased division of progenitor cells prior to reaching mature end cell stage, is not clear. However, the latter possibility is suggested by our previous observation that progenitor cellz overexpressing HOXB4 replate better in vitro than control neo infected progenitor: (Sauvageau et al., 1995).

Despite profound effects on the expansion of primitive hematopoietic cells overexpression of HOXB4 did not alter the ability of these cells to complete norrna differentiation programs in vitro or in vivo to produce mature cells of both lymphoic and myeloid lineages, nor did HOXB4 appear to promote preferential expansion alonc any hematopoietic lineage or lead to leukemia despite evidences of persistent HOXBd expression for at least 52 weeks.

Limited information exists about the downstream targets regulated byHox genes. 0i those identified to date, the majority appear to be either growth factors (e.g, decapentaplegic (dpp), a member of the TGFP superfamily, and basic fibroblast growth factor) (Capovilla et al., 1994; Care et al., 1996; ller et al., 1995) or cell adhesion molecules (e.g. connectin, N-CAM and L-CAM) (Goomer et al., 1994; Gould and White, 1992; Jones et al., 1992). Although Hox proteins have not yet been shown to direct11 regulate hematopoietic adhesion molecules or growth factors, the evidence cited above and the important roles that both adhesion molecules and growth factors play in hematopoiesis makes them interesting candidates for further study.

The effects of HOXB4 overexpression during hernatopoietic regeneration can be viewed as more "restricted" than those generated by overexpression of either HOXA 1C or HOXB3 that were presented in Chapters 3 and 4. The effects of HOXB4 are predorninantly seen on the expansion of HSCs during the early phase 01 hematopoietic regeneration, despite LTR-driven expression in most other hematopoietic cells and for a more extended period (>1 year post transplantation). It could thus be hypothesized that targets "availabfe" to HOXB4 were restricted to primitive hematopoietic cells and that these targets might increase the sensitivity of these cells to some form of "stem cell expanding factor(s)" present only during the early phase of hematopoietic regeneration. The ability of HOXB4 to enhance th( regeneration of HSC, without altering their ability to differentiate norrnally, maket HOXB4 a useful tool to gain insight into the molecular processes that regulate proliferation of early hematopoietic cells. Chapter 6

General Discussion and Conclusions

6.1 Overexpression of Hox genes perturbs hematopoietic development at multiple stages and involving multiple lineages

The transcriptional control of hematopoiesis is not well characterized. To identib transcription factors that might play an instructive role in regulation of hematopoiesis, various approaches have been applied. Many of the factors now known to be active in hematopoietic cells have been identified either by their aberrant expression in hernatological malignancies (Nichols and Nimer, 1992) or by detection of their ability to bind to cis-regulatory sequences of various lineage specific genes (Georgopoulos et al., 1992; Tsai et al., 1989). By analogy of hematopoiesis to other developmental programs, another fruitful approach has been to focus on factors highly conserved in evolution and that play central roles in tissue specification in other developmental systems. Taking this approach, it has been demonstrated that at least 20 of the 39 members of the Hox homeobox gene family, which are best known for their roles in specifying cell identity and pattern formation in a number of embryonic tissues, are expressed in normal hematopoietic cells. In hematopoietic cells, Hox gene expression appears to be largely confined to early cells (Giampaolo et al., 1994; Sauvageau et al., 1994), although a reactivation in expression of some of these genes has been described upon mitogenic activation of mature NK and T cells (Care et al., 1994; Quaranta et al., 1996). In early hematopoietic cells, Hox genes display distinctive expression patterns, with genes located at the 3' side of the Hox clusters being preferentially expressed in the most primitive subpopulation of bone marrow cells, while others, primarily located at the 5' end show a broader range of expression with downregulation at later stages of hematopoietic differentiation (Sauvageau et al., 1994). Functional roles for Hox genes in hematopoiesis is further strengthened by studies where their normal expression has been altered. Thus, for example overexpression of Hoxb-8 in murine hematopoietic cells enhances the generation O non-leukemic myeloid cell lines (Perkins and Cory, 1993), and our group has recentl! shown that overexpression of the 3' located HOXB4 selectively enhances tht expansion of primitive hematopoietic cells, most profoundly the HSC (Sauvageau e al., 1995).

The main goals pursued in this thesis were to provide further insight into thi possible functional roles of Hox genes in hematopoiesis, and to determine whethe there were Hox gene-specific roles. The initial strategy taken was to retrovirall! overexpress, in murine bone marrow cells, two selected Hox genes, HOXB3 an( HOXAIO, that were chosen based on their divergent expression patterns ir hematopoietic cells. The subsequent effects of these manipulations on the behavior O hematopoietic cells were then analyzed both in a transplantation model and in variou: culture systems. Manipulations of both these genes resulted in demonstrabk perturbations in hematopoiesis, and each gene generated a unique phenotype. Thesi effects are summarized in figure 6.1, in relation to those reported for thr overexpression of HOXB4. In sharp contrast to HOXB4, both HOXB3 and HOXA 1( failed to enhance the regeneration of HSC, but rather induced expansion of mort restricted progenitor cells and altered myeloid and lymphoid differentiation, albeit ir different manners. Specifically, overexpression of HOXA 1 O lead to enhancec formation of progenitor cells with megakaryocytic potential, was non-permissive fo macrophage and B-lymphoid differentiation and predisposed to leukemic transformation (Thorsteinsdottir et al., 1996). Overexpression of HOXB3, in contrast blocked thymic production of CD4+CD8+ T lymphocytes and enhanced expansion O y6-thymocytes, suppressed early B-lymphoid development, and led to é myeloproliferative disorder. (Sauvageau et al., 1996). Taken together, these results combined with both the effects of overexpression of HOXB4 and the extensive literature showing that Hox gene expression is largely confined to early hematopoietic cells, strongly indicate that Hox genes are involved in regulation of both proliferation and differentiation of early hematopoietic cells of multiple hematopoietic lineages. Furthermore, they point to Hox gene-specific effects, which Iikely reflects their regulation of different target genes during hematopoietic development.

Figure 6.1. Schematic representation of the effects observed on various hematopoietic lineages with overexpression of HOXB3, HOXB4 or HOXAlO

The second aim of the work presented in this thesis was to delineate further the nature of effects of HOXB4 overexpression on the regenerative potential of HSC. For that purpose, the size and the clonal composition of the regerierated pool of HSCs in mice transplanted with bone marrow cells overexpressing HOXB4 was analyzed at various time points (16 to 52 weeks) after transplantation. The results from these studies confirmed Our previous observation that HOXB4 overexpression enhances the regeneration of HSC following transplantation, and demonstrated for the first time that this effect involves multiple HOXB4-transduced HSCs, and that in vivo expansion of HOXB4-transduced HSCs is ultimately subject to endogenous control mechanisms that limits their further expansion. Furthermore, detailed analysis of various later cell populations in these mice (as late as 52 weeks after transplantation) strongly suggestç that overexpression of HOXBI does not overtly alter myeloid or lymphoid differentiation, nor lead to dominant outgrowth of any type of hematopoietic cell.

6.2 Overexpression of Hox genes: phenotype vs function

The striking differences in the phenotypes generated by overexpression of HOXB4, HOXB3 and HOXA 10, strongly suggest that under normal physiological conditions these genes play different functional roles in hematopoiesis. However, it becomes more problematic to infer from the phenotype generated by their overexpression, the exact stage(s) of hematopoiesis where these genes, under normal physiological conditions, have functional roles. This is due in part to limitations that are inherent in the forced overexpression approach taken. The LTR in the MSCV 2.1 retrovirus used to drive the forced expression of HOXB3, HOXB4 and HOXA 10 has been shown previously to direct high and long-term expression in primitive murine hematopoietic cells, as well as in their mature progeny (Pawliuk and Humphries, Unpublished data). Because Hox genes are expressed at relatively low levels under normal physiological conditions, their forced expression directed from the LTR thus both increases their levels in cells where they are normally expressed and extends their persistent expression into ceIl types were they are nomally not expressed. Thus in order to draw conclusions from the perturbations generated by their overexpression about the stage(s) of hematopoiesis where these Nox genes have physiological roles, the knowledge of their normal expression patterns becomes essential. Although Hox gene expression in hematopoietic cells has been analyzed in some details, the picture is far from complete.

In the case of HOXB4, the picture is perhaps the clearest. Its overexpression causes selective expansion of primitive hematopoietic cells, and thus appears to primarily affect the behavior of very primitive ce11 types where HOXBI is normally expressed, whereas later cell types appear unaffected. From these data, a role for HOXB4 in the regulation of differentiation of the earliest hematopoietic cells can thus be implied. In contrast to HOXB4, both HOXB3 and HOXA 10 overexpression affect multiple stages and lineages of hematopoiesis, some of which have not been analyzed for Hox gene expression. Thus, for example, although overexpression of HOXA 10 greatly en hances the generation of progenitor cells with megakaryocytic potential, a role for HOXA 10 in normal rnegakaryopoiesis is not clear, because analysis of Hox gene expression in these progenitor cells has been hampered by the difficulties in obtaining relatively pure populations of these cells. In contrast to the megakaryocytic lineage, expression of HOXA 10 in the granulocytic and monocytic lineages is better characterized, where HOXA 10 expression is confined to progenitor cells of these lineages and then sharply downregulated as they differentiate. Thus both the apparent block in monocyte differentiation and the high frequency generation of myeloid leukemias in recipients of HOXA 10 transduced bone marrow cells, suggest that downregulation of HOXA 70 expression is critical for progenitor cells of granulocytic and monocytic lineages to go through the final stages of differentiation. Overexpression of HOXA 10 also appears to block B lymphopoiesis, from at least the stage of pre-8 progenitor cells, without detectable concornmitant expansion of any B cell progenitor. These data, combined with the lack of Hoxa-10 expression both in purified murine 6 cell subpopulations representing early and late stages of B cell development, and in B lymphoid cell lines, suggest that HOXA 10 might negatively regulate cornmitment of progenitor cells to the B cell lineage. However, this assumption needs further clarification by using different experimental models (e.g. transgenic mice or cell line with lympho-myeloid potential), where both the earliest stage of B cell development affected by overexpression of HOXAIO and its effect on molecules known to be essential for B lymphopoiesis (e.g. E2A, Pax5 and Rag-1) could be assessed. Although HOXB3 expression in various subpopulations of CD34+ human bone marrow cells is similar to that of HOXB4, the results presented in this thesis suggest different roles for these two genes in primitive hematopoietic cells. However, due to limitations of the experimental system used in 139 these experiments, this difference was not characterized further. The most striking effect of overexpression of HOX83 was an almost complete block in the thyrnic production of ai3 CD4+CD8+ T lymphocytes accompanied by an expansion of $5- TCR+ thymocytes. However, a more detailed analysis of HOXB3 expression in various thymocyte populations is needed, to evaluate the possible physiological role of HOXB3 in the development of ap and y6 T cells. Similarly, a possible physiological role for HOXB3 in B cell development needs further clarification. As for HOXA 10, the earliest stage of B cell development affected by HOXB3 overexpression needs to be assessed, as well as normal HOXBS expression at various stages of B cell development.

Some properties of Hox genes add a level of complexity that needs to be considered when interpreting these overexpression data. As discussed in Chapter 1, a functional redundancy has been suggested between Hox genes, particularly those that belong to the same paralogous group. In hematopoietic cells many members of the Hox family are active, including some paralogs of HOXB3 and HOXB4, however expression of HOXA 10 paralogs has not been detected. It can thus not be ruled out that some of the effects observed when Hox genes are overexpressed could be caused by them either mimicking or blocking the action of another Hox gene. However, the strikingly different phenotypes generated when HOXB3, HOXB4 or HOXA 10 are overexpressed at least argues against the possibility that redundancies might involve Hox genes from different paralogous groups. It is believed that Hox gene expression can, at least in some cases, be auto- andlor cross regulated by other Hox genes (Zappavigna et al., 1994 ).Thus, for example, it has been shown that Hox genes located at the 3' end of the Hox clusters may regulate the expression of more 5' Iocated genes (Carè et al., 1994; Quaranta et al., 1996). The effects of overexpression of a particular Hox gene could thus also be attributed to its activation of more 5' located genes. The different phenotypes generated by overexpression of HOXB3 and its next neighbor HOXB4, however, would argue against this possibility. Furthermore, it has recently been reported that embryonic stem cells overexpressing HOXB4 do nc

up-regulate the expression of a more 5' Iocated Hox gene, Hoxb-8, when grown ii conditions that allow differentiation along hematopoietic lineages (Helgason et al, 1996). Another property of Hox genes that was discussed in Chapter 1 is thei interaction with CO-factors,such as the pbx proteins, which have at least in somc instances been shown to be important for their functional specificity. It has also beei proposed that these CO-factorscould be shared by different Hox genes, although thc extent of this sharing is far from clear. In that scenario, overexpression of a particula Hox gene thus could have a dominant negative effect by binding al1 available CO-facto molecules and therefore preventing other Hox genes, sharing that same CO-factor from activating their target genes. Although this possibility cannot be ruled out, tht different effects of overexpression of HOXB3, HOXB4 and HOM10 genes would favo other mechanisms.

6.3 Hox genes and hematopoiesis

Together, the results presented in this thesis, combined with earlier functional studiei where altered Hox gene expressions were also found to produce effects or differentiation or proliferation of hematopoietic cells (Carè et al., 1994; Giampaolo e. al., 1994; Lill et al., 1995; Perkins and Cory, 1993; Sauvageau et al., 1994; Shen et al. 1992; Wu et al., 1992), and the demonstration that at least 20 of the 39 members of the Hox gene family are expressed in hematopoietic cells in an apparent stage-dependeni manner (Giampaolo et al., 1994; Sauvageau et al., 1994), point to the existence of a complex "Hox code" in hematopoietic cells that may be deterministic for their growth and differentiation. The big question thus arises: What is the biological function of HOA genes in hematopoiesis? This question may be addressed by inferring their biological function in hematopoiesis from what is known about their functions in othei developmental systems. Both in Drosophila and vertebrate embryonic development, Hox genes are thought to confer positional information dong various body axes thereby directing morphogenesis (Krumlauf, 1994; Lawrence and Struhl, 1996). 141 However, the nature of their mechanisms of action, that is how they can 'identify" and "transfominthe morphology of sets of cells, is currently Iargely unknown. In Drosophila, the prevalent view is that Hox genes can select for different developmental prograrns, for example by activating or repressing incompletely overlapping sets of genes, thereby leading to the formation of different structures, such as legs or antennae (Lawrence and Stnihl, 1996). Based on the evolutionary conservation of both Hox gene organization and at least sorne of their properties (see Chapter 1 for more detail), similar mechanisms of action would be suggested for Drosophila and vertebrate Hox genes. Homeotic transformations in vertebrates have been best documented within the vertebral column, where they result either in the appearance of additional vertebra or in a transition from one morphological type to another (Krumlauf, 1994). It has been proposed that these transformations are unlikely the results of selection of alternative developmental pathways but rather result from different control of growth rates within the developing vertebra. Based on these absemations, a model has been proposed whereby sequential activation of Hox genes along the anterior-posterior axis with the generation of the Hox gene "rainbow", determines the growth rates of cells with in each segment and thus the patterning of that structure (Duboule, 1995). ln this scenario, Hox genes can be viewed as having a general function as biological docks in developrnental programs, where ordered and time dependent series of sequential transcriptional events need to be carefully orchestrated.

It is not immediately obvious that the hematopoietic system requires positional information or that it has any "axes". However, some anaiogy can be drawn between hematopoiesis and embryogenesis. First, the regulation of Hox gene expression in hematopoietic cells has striking parallels with the colinear Hox gene expression du ring embryogenesis. Second, the adult bone marrow has a three-dimensional structure and organization, within which hematopoietic cells interact with stroma1 cells and migrate as they differentiate. Third, hematopoiesis is an ordered and time- dependent process of differentiation and proliferation leading to the generation of mature blood cells. It is thus possible to imagine that Hox genes might confei positional information to hernatopoietic cells, mediated by interactions with stroma cells or regulatory molecules secreted by these cells, which would coordinate theii subsequent growth and migration as they differentiate. The pattern of Hox gene expression in a given cell type could thus be viewed as some sort of barometer which "informs" the cell where it is exactly positioned with respect to state of development, and thus insure that the subsequent behavior of that cell would be according to thal "location".

As discussed in Chapter 1, Iimited information exists about the downstream targets regulated by Hox genes. lnterestingly, however, of those few identified the majority appear to be either growth factors (e.g. decapentaplegic (dpp), a member of the TGFP superfamily, and basic fibroblast growth factor) (Capovilla et al., 1994; Care et al., 1996; ller et al., 1995) or cell adhesion rndecules (e.g. connectin, N-CAM and L-CAM) (Goomer et al., 1994; Gould and White, 1992; Jones et al., 1992). Given the importani roles that both growth factors and ceIl-adhesion molecules play in hematopoiesis, it is tempting to speculate that Hox gene targets in hematopoietic cells could be either growth factors or cell adhesion molecules (previously identified or unidentified).

6.4. Future directions

In hematopoietic cells expression of Hox genes has mainly been analyzed in leukemic ceIl lines and CD34+ subpopulations of normal human bone marrow, whereas limited data exists on their expression in murine hematopoietic cells. An important next step would thus be to systematically analyze Hox gene expression in adult murine hematopoietic cells at various stages of both myeloid and lymphoid development. Furthermore, their expression in similar subpopulations at various stages during ontogeny, needs to be resolved as an initial step in assessing the role of Hox gene in embryonic hematopoiesis. The functional roles of Hox genes in hematopoiesis have been analyzed either b overexpression or inactivation of specific Hox genes. Of these approches overexpression studies have been more "fruitfulu with dramatic perturbation i hematopoiesis observed (as described in this thesis). In contrast, "knockout studies have revealed more subtle perturbations (Lawrence et al., 1996; Lawrence et al 1996). This discrepancy can be explained at least in part by the functional redundanc amongst Hox genes of the same paralogous group, which would thus complement fc the absence of one of its members. Recently, with the growing list of single Hox genl knockouts, a generation of double and even triple Hox gene knockout, by mating c single knockouts, is now possible. Detailed analyses of possible hematologic; abnormalities in these double or triple knockout mice are likely to be informative ii resolving the role of Hox genes in hematopoiesis.

Another interesting avenue to explore, is the role of individual Hox gene clusters ii hematopoiesis. Expression of Hox genes from three clusters i.e. HoxA, HoxB an[ HoxC, has been detected in hematopoietic cells, or specifically, 9 of the 11 Hox, genes, 8 of the 10 HoxB genes and 3 of the 9 HoxC genes. An important issue tc resolve is whether any one of these clusters are essential for hematopoiesis in genera or in any of the hematopoietic lineages. With the use of the IoxPICre recombinasc system it is now possible to "knockout" an entire Hox gene cluster in a tissue specific rnanner (Gu et al., 1994). For that purpose loxP sites could be inserted on both ends o a Hox cluster (A, B or C cluster) in embryonic stem cells with the subsequen generation of mice homozygous for these loxP insertions. These mice would then br mated with transgenic mice which direct expression of the Cre recombinase if hematopoietic cells. The result would be deletion of the Hox cluster between the loxF sites only in cells expressing the Cre recombinase (Le. hematopoietic populations and thus eliminate problems due to early embryonic lethality. Transgenic mice whict express Cre recombinase at various stages of 6 and T Iymphoid development havi been generated. However, mice directing Cre expression in myeloid cells would neec to be generated.

As discussed in Chapter 1 there is now ample evidence suggesting that, in additior to Hox genes, many of the molecules and principles applied during morphogenesis a various embryonic structures in diverse animals, are highly conserved. Central to this machinery are secreted molecules of the hedgehog, Wnt and BMPs families of genes some of which have been found to act in pathways downstream or upstream of Hor genes. In a recent report (Van Den Berg et al., 1996), expression of some members a the Wnt family in human fetal liver stromal cells was described. This report alsc described the expression of selected frizz/ed family members, the receptors for Wn proteins, on CD34+ human bone marrow cells. Interestingly, co-cultivation of humar bone marrow cells on stromal cells overexpressing some of these Wnt genes resultec in greatly increased clonogenic progenitor output (20-30-foid increase in CFU-GEMM)' compared to control. Based on this data it would thus be very interesting to determine whether Hox genes could be acting in pathways either upstream or downstream 01 Wnt genes or their receptors. It will also be of interest to determine whether members of the hedgehog or BMP families are active in hematopoietic cells and if so, whethei they might act in pathways upstream or downstream of Hox genes. Chapter 7

References

Alkema, M. J., van der Lugt, N. M. T., Bobeldijk, R. C., Berns, A., and van Lohuizen, M (1995). Transformation of axial skeleton due to overexpression of bmi- 1 in transgenic mice. Nature 374, 724-727.

Allen, J. D., and Adams, J. M. (1993). Enforced expression of Hlx homeobox genc prompts myeloid cell maturation and altered adherence properties of T cells. Blood 81 3242-325 1. Allen, J. D., Harris, A. W., Bath, M. L., Strasser, A., Scollay, R., and Adams, J. M. (1995) Perturbed development of T and B cells in mice expressing an Hlx homeobo) transgene. J lmmunol 154, 1531-1542.

Antica, M., Wu, L., Shortman, K., and Scollay, R. (1994). Thymic stem cells in muusc bone marrow. Blood 84, 1 11-1 17.

Bachiller, D., Macias, A., Duboule, D., and Morata, G. (1994). Conservation a, functional hierarchy between rnamrnalian and insect HOXIHOM genes. EMBO J. 13 1930-1 941.

Barba, P., Magli, M. C., Tiberio, C., and Cillo, C. (1993). HOX gene expression ir human cancers. Advances in Experimental Medicine & Biology 348, 45-57.

Bateson, W. (1894). Materials for the study of variation (New York: Macmillan).

Becker, A., McCulloch, E. A., and Till, J. E. (1963). Cytological demonstration of the clonal nature of spleen colonies derived from transplanted mouse marrow cells' Nature 197, 452.

Bedford, F. K., Ashworth, A., Enver, T., and Wiedemann, L. M. (1993). HEX: a nove homeobox gene expressed during haematopoiesis and consewed between mouse and human. Nucleic Acids Res. 21, 1245-1249.

Behringer, R. R., Crotty, D. A., Tennyson, V. M., Brinster, R. L., Palmiter, R. D., anc Wolgemuth, D. J. (1993). Sequences 5' of the homeobox of the Hox-1.4 gene direci tissue-specific expression of lac2 during mouse development. Development 117, 823- 833.

Bieberich, C. J., Ruddle, F. H., and Stenn, K. S. (1992). Differential expression of the 3.1 gene in adult adult mouse skin. Ann. N. Y. Acad. Sci. 642, 346-354.

Bienz, M. (1994). Homeotic genes and positional signalling in the Drosophila viscera, Trends Genet 10, 22-26. Billeter, M., Qian, Y. Q., Otting, G., Muller, M., Gehring, W., and Wuthrich, K. (1993). Determination of the nuclear magnetic resonance solution structure of an Antennapedia homeodomain-DNA complex. J. Mol. Biol. 234, 1084-1093.

Bodine, D. M., Crosier, P. S., and Clark, S. C. (1992). Effects of hematopoietic growth factors on the survival of primitive stem cells in liquid suspension culture. Blood 78, 914-920.

Boncinelli, E.1 Acampora, D., Pannese, M., D'Esposito, M., Somna, R., Gaudino, G., Stornaiuolo, A., Cafiero, M., Faiella, A., and Simeone, A. (1989). Organization of human class I homeobox genes. Genome 31,745-756.

Borrow, J., Shearman, A. M., Stanton Jr, V. P., Becher, R., Collins, T., Williams, A. J., Dube, I., Katz, F., Kwong, Y. L., Morris, C., Ohyashiki, K., Toyama, K., Rowley, J., and Housman, D. E. (1996). The t(7;11)(p15;p15) translocation in acute myeloid leukaemia fuses the genes for nucleoporin NUP98 and class 1 homeoprotein HOXA9. Nature Genetics 12, 159-167.

Bradley, T. R., and Hodgson, G. S. (1979). Detection of primitive macrophage progenitor cells in mouse bone marrow. Blood 54, 1446.

Brady, G., Barbara, M., and Iscove, N. (1990). Representative in vitro cDNA amplification from individual hemopoietic cells and colonies. Methods Mol. Cell. Biol. 2, 1-9.

Briegel, K., Lim, K. C., Plank, C., Beug, H., Engel, J. D., and Zenke, M. (1993). Ectopic expression of a conditional GATA-Slestrogen receptor chimera arrests erythroid differentiation in a hormone-dependent manner. Genes Dev 7, lO97-IlO9.

Burke, A. C., Nelson, C. E., Morgan, B. A., and Tabin, C. (1995). Hox genes and the evolution of vertebrate axial morphology. Development 121, 333-346.

Capel, B., Hawley, R., Covarrubias, L., Hawley, T., and Mintz, B. (1989). Clonal contributions of small numbers of retrovirally marked hematopoietic stem cells engrafted in unirradiated neonatal WNVv rnice. Proc. Natl. Acad. Sci. USA 86,4564.

Capovilla, M., Brant, M., and Botas, J. (1994). Direct regulation of decapentaplegic by Ultrabithorax and its role in Drosophila midgut morphogenesis. Cell 76, 461 -475.

Care, A., Silvani, A., Meccia, E., Mattia, G., Stoppacciaro, A., Parmiani, G., Peschle, C., and Cdlombo, M. P. (1996). HOXB7 constitutively activates basic fibroblast growth factor in melanomas. Mol. Cell. Biol. 16, 4842-4851.

Care, A., Testa, U., Bassani, A., Tritarelli, E., Montesoro, E., Samoggia, P., Cianetti, L., and Peschle, C. (1994). Coordinate expression and proliferative role of HOXB genes in activated adult T lymphocytes. Mol Cell Biol 14, 4872-4877.

Cashman, J. D., Eaves, A. C., and Eavas, C. J. (1992). Granulocyte-macrophage colony-stimulating factor modulation of the inhibitory effect of transforming growth factor$ on normal and leukemic human hernatopoietic progenitor cells. Leukemia 6, 886. Celetti, A., Barba, P., Cillo, C., Rotoli, B., Boncinelli, E., and Magli, M. C. (1993) Characteristic patterns of HOX gene expression in different types of human leukemia Int. J. Cancer 53,237-244.

Chan, S.-K., Jaffe, L., Capovilla, M., Botas, J., and Mann, R. S. (1994). The DNP binding specificity of ultrabithorax is modulated by cooperative interactions witt extradenticle, another homeoprotein. Cell 78, 603-615.

Chan, S.-K., and Mann, R. S. (1993). The segment identity functions of Ultrabithorm are contained within its homeo domain and carboxyl-terminal sequences. Genes Dev 7, 796-811 . Chan, S.-K., and Mann, R. S. (1996). A structural model for a homeotic protein. extradenticle-DNA complex accounts for the choice of HOX protein in the heterodimer Proc. Natl. Acad. Sci. USA 93, 5223-5228.

Chan, S.-K., Pôpperl, H., Krumlauf, R., and Mann, R. S. (1996). An extradenticle. induced conformational change in a HOX protein overcomes an inhibtory function oi the conserved hexapeptide motif. EMBO J. 15, 2476-2487.

Chang, C. P., Brocchieri, L., Shen, W. F., Largman, C., and Cleary, M. L. (1996). Pbr modulation of Hox homeodomain amino-terminal arms establishes different DNA- binding specificities across the Hox locus. Mol. Cell. Biol. 16, 1734-1745.

Chang, C. P., Shen, W. F., Rozenfeld, S., Lawrence, H. J., Largman, C., and Cleary, M, L. (1995). Pbx proteins display hexapeptide-dependent cooperative DN.4 binding with a subset of Hox proteins. Genes Dev. 9, 663-674.

Charité, J., De Graaff, W., Shen, S., and Deschamps, J. (1994). Ectopic expression 01 Hoxb-8 causes duplication of the ZPA in the forelimb and homeotic transformation 01 axial structures. Cell 78, 589-601.

Chiang, C., Litingtung, Y., Lee, E., Young, K. E., Corden, J. L., Westphal, H., and Beachy, P. A. (1996). Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function. Nature 383, 407-413.

Chouinard, S., and Kaufman, T. C. (1991). Control of expression of the homeotic labial ([ab) Iocous in Drosophila melanogaster evidence for both positive and negative autogenous regliiation. Development 113, 1267-1280.

Cohen, K. J., Hanna, J. S., Prescott, J. E., and Dang, C. V. (1996). Transformation by the bmi-1 oncoprotein correlates with its subnuclear localization but not its transcriptional suppression activity. Mol. Cell. Biol. 16, 5527-5535.

Cohn, M. J., and Tickle, C. (1996). Limbs: a model for pattern formation within the vertebrate body plan. Trends Genet. 12, 253-257.

Concordet, J.-P., and Ingham, P. (1995). Patterning goes Sonic. Nature 375, 279-280.

Condie, B. G., and Capecchi, M. R. (1994). Mice with targeted disruptions in the paralogous genes hoxa-3 and hoxd-3 reveal synergistic interactions. Nature 370, 304- 307. Cone, R. O., and Mulligan, R. C. (1984). High-efficiency gene transfer into mammalian cells: Generation of helper-free recombinant retrovirus with broad mammalian host range. Proc. Natl. Acad. Sci. USA 81, 6349-6353.

Conneally, E., Bardy, P., Eaves, C. J., Thomas, R., Chappel, S., Shpall, E. J., and Humphries, R. K. (1996). Rapid and efficient selection of human hematopoietic cells expressing murine heat-stable antigen (HSA) as an indicator of retroviral-mediated gene transfer. Blood 87, 456-464.

Cowing, D., and Kenyon, C. (1996). Correct Hox gene expression established independently of position in Caenorhabditis elegans. Nature 382, 353-356.

Cumano, A., Dieterien-Lievre, F., and Godin, 1. (1996). Lymphoid potential, probed before circulation in mouse, is restricted to caudai intraembryonic splanchnopleura. Cell 86, 907-916.

Cumano, A., Paige, C. J., Iscove, N. N., and Brady, G. (1992). Bipotential precursors of B cells and macrophages in murine fetai liver. Nature 356, 612.

Damen, J. E., Liu, L., Rosten, P., Humphries, R. K., Jefferson, A. B., Majerus, P. W., and Krystal, G. (1996). The 145-kDa protein induced to associate with Shc by multiple cytokines is an inositol tetraphoshate and phosphatidylinositol 3,4,5-trisphosphate 5- phosphatase. Proc. Natl. Acad. Sci. USA 93, 1689-1 693.

Davidson, D. (1995). The function and evolution of Msx genes: pointers and paradoxes. Trends Genet. 11,405-411.

Davis, L., Kuehl, M., and Battey. (1994). Subcloning fragments into plasmid vectors. In Basic rnethodes in molecular biology, pp.277-302. Appleton and Lange, Nowalk, CT.

De Vita, G., Barba, P., Odartchenko, N., Givel, J. C., Ferschi, G., Buccuarelli, G., Magli, M. C., Boncinelli, E., and Cillo, C. (1 993). Expression of homeobox-containing genes in primary and metastatic colorectal cancer. Eur. J. Cancer 29A, 887-893.

Dear, T. N., Colledge, W. H., Carlton, M., Lavenir, I., Larson, T., Smith, A., Warren, A. J., Evans, M. J., Sofroniew, M. V., and Rabbitts, T. H. (1995). The Hoxll gene is essential for cell survival during spleen development. Development 121, 2909-291 5.

Deguchi, Y., Agus, D., and Kehrl, J. H. (1993). A human homeobox gene, HB24, inhibits development of CD4+ T cells and impairs thyrnic involution in transgenic mice. J. Biol. Chem. 268, 3646-3653.

Deguchi, Y., and Kehrl, J. H. (1991). Selective expression of two homeobox genes in CD34-positive cells from human bone marrow. Blood 78, 323-328.

Deguchi, Y., Moroney, J. F., Wilson, G. L., Fox, C. H., Winter, H. S., and Kehrl, J. H. (1991). Cloning and expression of a human horneobox gene which resembles a diverged Drosophila homeobox gene and is expressed in activated lymphocytes. New Biologist 3, 353-363. Deguchi, Y., Thevenin, C., and Kehrl, J. H. (1992). Stable expression of HB24, a diverged human homeobox gene, in T lymphocytes induces genes involved in T cell activation and growth. J Biol Chem 267, 822298229.

Dekker, E. J., Pannese, M., Houtzager, E., Boncinelli, E., and Durston, A. (1993). Colinearity in the Xenopus laevis Hox-2 complex. Mech Dev 40, 3-12.

DeRobettis, E. M. (1994). The horneobox in cell differentiation and evolution. In Guiedook to the horneobox genes, D. Duboule, ed. (London: Sambrook and Tooze), pp. 13-26.

Desplan, C., Theis, J., and O'Farrell, P. H. (1988). The sequence specificity of homeodomain-DNA ineraction. Cell 54, 1O81 -1090.

Dick, J. E., Magli, M. C., Huszar, C., Phillips, R. A., and Bernstein, A. (1985). Introduction of a selectable gene into primitive stem cells capable of long term reconstitution of the hemopoietic system of WMlv mice. Cell 42, 71-79.

Duboule, D. (1995). Vertebraie Hox genes and proliferation: an alternative pathway to homeosis. Curr. Biol. 5, 525-528.

Duboule, D., and Morata, G. (1994). Colinearity and functional hierarchy among genes of the homeotic complexs. Trends Genet. 10, 358-364.

Dudley, E. C., Girardi, M., Owen, M. J., and Hayday, A. C. (1995). cwp and y8 T cells can share a late common precursor. Curr. Biol. 5, 659-669.

Eaves, C. J. (1995). ln vitro assays for quantitating hemopoietic progenitor cells. In Williams Hernatologt, W. J. William, ed.: McGraw Hill Inc.).

Eaves, C. J., Cashman, J. D., and Kay, R. J. (1991). Mechanisms that regulate the cell cycle status of very primitive hematopoietic cells in long-term human marrow cultures II. Analysis of positive and negative regulators produced by stroma1 cells within the adherent layer. Blood 78, 110.

Eid, R., Koseki, H., and Schughart, K. (1993). Analysis of IacZ reporter genes in transgenic embryos suggests the presence of several cis-acting regulatory elements in the murine Hoxb-6 gene. Dev. Dyn. 196, 205-216.

Ekker, S. C., Jackson, D. G., von Kessler, D. P., Sun, B. I., Young, K. E., and Beachy, P. A. (1994). The degree of variation in DNA sequence recognition among four Drosophila homeotic proteins. EMBO J. 13, 3551-3560.

Ekker, S. C., Young, K. E., von Kessler, D. P., and Beachy, P. A. (1991). Optimal DNA sequence recognition by the Ultrabithorax homeodomain of Drosophila. EMBO J. 10, 1179-1186.

Ernst, P., and Srnale, S. T. (1995). Combinatorial regulation of transcription 1: general aspects of transcriptional control. lmmunity 2, 31 1-319. Fairbairn, L. J., Cowling, G. J., Reipert, B. M., and Dexter, T. M. (1993). Suppression of apoptosis allows differentiation and development of multipotent hemopoietic cell line in the absence of added growth factors. Cell 74, 823-832.

Felsenfeld, G. (1992). Chromatin as an essential part of the transcriptional mechanism. Nature 355, 219-223.

Fietz, M., Concordet, J.-P., Barbosa, R., Johnson, R., and Krauss, S. (1994). The hedgehog gene family in Drosophila and vertebrate development. Development SU/@., 43-51.

Flegel, W. A., Singson, A. W., Margolis, J. S., Bang, A. G., Posakony, J. W., and Murre, C. (1993). Dpbx, a new homeobox gene closely related to the human proto-oncogene pbxl : molecular structure and developmental expression. Mech. Dev. 41, 155-161.

Fraser, C. C., Szilvassy, S. J., Eaves, C. J., and Humphries, R. K. (1992). Proliferation of totipotent hematopoietic stem cells in vitro with retention of long-term competitive in vivo reconstituting ability. Proc Natl Acad Sci USA 89, 1968-1972.

Friedmann, Y., Daniel, C. A., Strickland, P., and Daniel, C. W. (1994). Hox genes in normal and neoplastic mouse mammary gland. Cancer Res 54, 5981-5985.

Frohman, M. A., Martin, G. R., Cordes, S., Halamek, L. P., and Barsh, G. S. (1993). Altered rhombomere-specific gene expression and hyoid bone differentiation in the mouse segmentation mutant kreisler (krj. Development 117, 925-936.

Furukubo-Tokunaga, K., Flister, S., and Gehring, W. J. (1993). Functional specificity of the Antennapedia homeodomain. Proc. Natl. Acad. Sci. USA 90, 6360-6364.

Garber, R. L., Kuroiwa, A., and Gehring, W. J. (1983). Genomic and cDNA clones of the homeotic locus Antennapedia in Drosophila. EMBO J 2, 2027-2036.

Gaunt, S. J. (1991). Expression patterns of mouse Hox genes: ches to an understanding of developmental and evolutionary strategies. Bioessays 13, 505-513.

Georgopoulos, K., Bigby, M., Wang, J.-H., Molnar, A., Wu, P., Winandy, S., and Sharpe, A. (1 994). The lkaros gene is required for development of al1 lymphoid lineages. Cell 79, 143.

Georgopoulos, K., Moore, D. D., and Derfler, B. (1992). Ikaros an earIy lymphoid restricted transcription factor, a putative mediator for T cell commitrnent. Science 258, 808-81 2.

Gerard, M., Chen, J.-Y., Gronemeyer, H., Chambon, P., Duboule, D., and Zakany, J. (1996). In vivo targeted mutagenesis of a regulatory element required for positioning the Hoxd- 7 1 and Hoxd- 1 O expression boundaries. Genes Dev. 10, 2326-2334.

Giampaolo, A., Sterpetti, P., Bulgarini, O., Samoggia, P., Pelosi, E., Valtieri, M., and Peschle, C. (1994). Key functional role and lineage-specific expression of selected HOXB genes in purified hematopoietic progenitor differentiation. Blood 84, 3637-3647. Gibson, G., Schier, A., LeMotte, P., and Gehring, W. J. (1990). The specificities of Se combs reduced and Antennapedia are defined by a distinct portion of each protein tha includes the homeodomain. Cell 62, 1087-1103.

Godfrey, D. l., and Zlotnik, A. (1993). Control points in early T-cell developmeni Immunol. Today 14, 547-553.

Godin, 1. E., Garcia-Porrero, J. A., Coutinho, A., Dieterien-Lievre, F., and Marcos, M. P (1993). Paraaortic splanchnopleura from early mouse embryo contains B1i lymphocyte precursors. Nature 364, 67-70.

Gonzales-Reyes, A., and Morata, G. (1990). The developmental effect a overexpressing a Ubx product in Drosophila embryos is dependent on its interaction: with other homeotic products. Cell 61, 51 1-522.

Goomer, R. S., Holst, B. D., Wood, 1. C., Jones, F. S., and Edelman, G. M. (1994) Regulation in vitro of an L-CAM enhancer by homeobox genes HoxD9 and HNF-1 Proc Nat1 Acad Sci Usa 91, 7985-7989.

Gorski, D. H., LePage, D. F., Patel, C. V., Copeland, N. G., Jenkins, N. A., and Walsh, K (1993). Molecular cloning of a diverged homeobox gene that is rapidly down regulated during the GO/G1 transition in vascular smooth muscle cells. Mol. Cell. Biol 13, 3722-3733.

Gould, A. P., and White, R. A. H. (1992). Connectin, a target of homeotic gene contrc in Drosophila. Development 116, 1163-1 174.

Graba, Y., Aragnol, D., Laurenti, P., Garzino, V., Charmont, D., Berenger, H., ani Pradel, J. (1992). Homeotic control in Drosophila; the scabrous gene is an in vivi target of ulfrabithorax proteins. EMBO J. 11, 3375-3384.

Graham, A., Papalopulu, N., and Krumlauf, R. (1989). The murine and Drosophili homeobox clusters have common features of organization and expression. Cell 57 367-378.

Graham, G. J., Wright, E. G., Hewick, R., Wolpe, S. D., Wilkie, N. M., Donaldson, D. Lorimore, S., and Pragnell, 1. B. (1990). Identification and characterization of ai inhibitor of haemopoietic stem cell proliferation. Nature 344, 442-444.

Grunstein, M. (1990). Histone function in transcription. Annu. Rev. Cell. Biol. 6, 643 678.

GU, H., Marth, J. D., Orban, P. C., Mossmann, H., and Rajewsky, K. (1994). Deletion o DNA polymerase P gene segment in T cells using cell type-specific gene targeting Science 265, 103-1 06.

GU,Y., Nakamura, T., Alder, H., Prasad, R., Canaani, O., Cimino, G., Croce, C. M., anc Cananni, E. (1992). The t(4:11) chromosome translocation of human acute leukemia! fuses ALL-1 gene, related to Drosophila trithorax, to the the AF4 gene. Cell 71, 701 708. Haas, W., and Tonegawa, S. (1992). Development and selection of gamma delta ' cells. Curr. Opin. Immunol. 4, 147-155.

Hanna-Rose, W., and Hansen, U. (1996). Active repression mechanisms of eukariotil transcription repressors. Trends Genet. 12, 229-234.

Harrison, O. E., and Astle, C. M. (1982). Loss of stem cell repopulating ability witl transplantation. Effects of donor age, cell number and transplant procedure. J. EXF Med. 156, 1767.

Harrison, D. E., Astle, C. M., and Delaittre, J. A. (1978). Loss of proliferative capacity ii immunohemopoietic stem cells caused by serial transplantation rather than aging Journal of Experimental Medicine 147, 1526.

Harrison, O. E., Jordan, C. T., Zhong, R. K., and Astle, C. M. (1993). Primitivt hemopoietic stem cells: Direct assay of most productive populations by competitivc repopulation with simple binomial, correlation and covariance calculations. Exp Hematol. 21, 206-219.

Harrison, D. E., and Lemer, C. P. (1991). Most primitive hematopoietic stem cells arc stimulated to cycle rapidly after treatment with 5-fluorouracil. Blood 78, 1237-40.

Harrison, D. E., Stone, M., and Astle, C. M. (1990). Effects of transplantation on thi primitive immunohematopoietic stem cell. J. Exp. Med. 172, 431 -437.

Harrison, S. C. (1991). A structural taxonomy of DNA-binding domains. Nature 353 715-71 9.

Hatano, M. C., Roberts, C. W. M., Minden, M., Crist, W. M., and Krosmeyer, S. J. (1991) Deregulation of the homeobox gene, HOX11, by the t(10; 14) in T cell leukemia Science 253, 79-82.

Hatzfeld, J., Li, M. L., Brown, E. L., Sookdeo, H., Levesque, J. P., O'Toole, T., Gurney C., Clark, S. C., and Hatzfeld, A. (1991). Release of early human hematopoietic progenitors from quiescence by antisense transforming growth factor betal or Rt oligonucleotides. J. Exp. Med. 174, 925-929.

Haupt, Y., Alexander, W. S., Barri, G., Klinken, S. P., and Adams, J. M. (1991). Nove zinc finger gene implicated as myc collaborator by retrovirally acceleratec lymphomagenesis in Ep-myc transgenic mice. Cell 65, 753-763.

Hawley, R. G., Fong, A., Lu, M., and Hawley, T. S. (1994). The HOXI1 homeobox containing gene of human Ieukemia immortalizes murine hematopoietic precursors Oncogene 9, 1-1 2.

Hawley, R. G., Fong, A. 2. C., Burns, B. F., and Hawley, T. S. (1992). TransplantabIf myeloproliferative disease induced in mice by interleukin 6 retrovirus. J. Exp. Med 176, 1149-1163.

Helgason, C. O., Sauvageau, G., Lawrence, H. J., Largman, C., and Humphries, R. K (1996). Overexpression of HOXB4 enhances the hematopoietic potential of embryonic stem cells differentiated in vitro. Blood 87, 2740-2749. Hentsch, B., Lyons, I., Li, R., Hartley, L., Lints, T. J., Adams, J. M., and Harvey, R. P. (1996). Hlx homeo box gene is essential for an inductive tissue interaction that drives expansion of embryonic liver and gut. Genes Dev. 10, 70-79.

Hirsch, M. R., Gaugler, L., Deagostini-Bazin, H., Bally-Cuif, L., and Goridis, C. (1990). Identification of positive and negative regulatory elements governing cell-type-specific expression of the neural cell adhesion molecule gene. Mol. Cell. Biol. 10, 1959-1968.

Hirsch, M. R., Valarche, I., Deagostini, B. H., Pernelle, C., Joliot, A., and Goridis, C. (1 991). An upstream regulatory element of the NCAM promoter contains a binding site for homeodomains. Febs Letter 287, 197-202.

Hodgson, G. S., and Bradley, T. R. (1979). Properties of haematopoietic stem cells surviving 5-fluorouracil treatment: evidence for a pre-CFU-S cell. Nature 281, 381-383.

Hoey, T., and Levine, M. (1988). Divergent homeo box proteins recognize similar DNA sequences in Drosophila. Nature 332, 858-861.

Hogan, B. L. M. (1996). Bone morphogenetic proteins in development. Curr. Opin. Genet. Dev. 6, 432-438.

Horan, G. S. B., Ramirez-Solis, R., Featherstone, M. S., Wolgemuth, D. J., Bradley, A., and Behringer, R. R. (1995). Compound mutants for the paralogous hoxa-4, hoxb-4 and hoxd-4 genes show more complete homeotic transformations and dose- dependent increase in the number of vertebrae transformed. Genes Dev. 9, 1667- 1677.

Hough, M. R., Chappel, M. S., Sauvageau, G., Takei, F., Kay, R., and Humphries, R. K. (1996). Redution of early B lymphocyte precursors in transgenic mice overexpressing the murine heat-stable antigen. Journal of lmmunology 156, 479-488.

Hough, M. R., Takei, F., Humphries, R. K., and Kay, R. (1994). Defective development of thymocytes overexpressing the costirnulatory molecule, heat-stable antigen. Journal of Experimental Medicine 179, 177-84.

Hugo, P., Kappler, J. W., McCormack, J., and Marrack, P. (1993). Fibroblasts can induce thymocyte positive selection in vivo. Proc. Natl. Acad. Sci. USA 90, 10335- 10339.

Humphries, R. K., Eaves, A. C., and Eaves, C. J. (1981). Self-renewal of hemopoietic stem cells during mixed colony formation in vitro. Proc. Natl. Acad. Sci. USA 78, 3629.

Huyhn, A., Dommerques, M., Izac, B., Croisillel, L., Katz, A., Vainchenker, W., and Coulombel, L. (1995). Characteriration of hematopopietic progenitors from human yolk sacs and embryoos. Bfood 86, 4474-4485.

Iler, N., Rowitch, D. H., Echelard, Y., McMahon, A. P., and Abate-Shen, C. (1995). A single homeodomain binding site restricts spatial expression of Wnt-1 in the developing brain. Mech. Dev. 53, 87-96. tnamori, K., Takeshita, K., Chiba, S., Yazaki, Y., and Hirai, H. (1993). Identification O homeobox genes expressed in human T-lymphocytes. Biochem. Biophy. Res Commun. 196, 203-208.

Ingham, P. W. (1995). Signalling by hedgehog family proteins in Drosophila anc vertebrate development. Curr Opin Genet Dev 5, 492-498.

lzpisua-Belmonte, J., Falkenstein, H., Dolle, P., Renucci, A., and Duboule, O. (1991) Murine genes related to the Drosophila Abd6 homeotic gene are sequentiall) expressed during development of the- posterior part of the body. EMBO J. 70, 2279 2289.

Jacobsen, S. E., Ruscetti, F. W., Ortiz, M., Gooya, J. M., and Keller, R. (1994). The growth response of Lin-Thy-l+ hematopoietic progenitors to cytokines is determinec by the balance between synergy of multiple stimulators and negative cooperation O multible inhibitors. Exp. Hematol. 22, 985-989.

Johnson, G. R., and Metcalf, 0. (1977). Pure and mixed erythroid colony formation if vitro stimulated by spleen conditioned medium with no detectable erythropoietin. Proc Natl. Acad. Sci. USA 74, 3879-3882.

Jones, F. S., Holst, B. D., Minowa, O., DeRobertis, E. M., and Edelman, G. M. (1993) Binding and transcriptional activation of the promoter for the neural cell adhesior molecule gene by HOXC6 (HOX-3.3). Proc. Natl. Acad. Sci. USA 90, 6557-6561.

Jones, F. S., Prediger, E. A., Bittner, D. A., De Robertis, E. M., and Edelman, G. M (1992). Cell adhesion molecules as targets for Hox genes: Neural cell adhesior molecule promoter activity is modulated by cotransfection with Hox-2.5 and -2.4. Pro Natl. Acad. Sci. USA 89, 2086-2090.

Jones, R. J., Wagner, J. E., Celano, P., Zicha, M. S., and Sharkis, S. J. (1990) Separation of pluripotent haematopoietic stem cells from spleen colony-forming cells Nature 347, 188-189.

Jordan, C. T., and Lemischka, 1. R. (1990). Clonal and systemic analysis of long-tenr hematopoiesis in the mouse. Genes Dev. 4, 220-232.

Jurgens, G. (1985). A group of genes controlling the spatial expression of the bithorar complex in Drosophila. Nature 316, 153-155.

Kalionis, B., and O'Farrell, P. H. (1993). A universal target sequence is bound in vitrc by diverse homeodomains. Mech. Dev. 43, 57-70.

Kamps, M. P., Murre, C., Sun, X.-H., 2nd Baltimore, D. (1990). A new homeobox gens contributes the DNA binding domain of the t(1;19) translocation protein in pre-B ALL, Cell 60, 547-555.

Kanno, M., Hasegawa, M., Ishida, A., Isono, K., and Taniguchi, M. (1995). me/-18, a Polycomb group-related mammalian gene, encodes a transcriptional negative regulator with tumor suppressive activity. EMBO J. 14, 5672-5678. Kaplan. E.L., Meier,P. (1958). Nonparametric estimation from incomplete observation J Am Stat Assoc 53,457

Katayama, N., Shib, J., Nishikawa, S., Kina, T., Clark, S. C., and Ogawa, M. (1993) Stage-specific expression of c-kit protein by murine hematopoietic progenitors. Blooc 82, 2353.

Kaufman, T. C. (1983). The genetic regulation of segmentation in Drosophili melanogaster. In Time, space, and pattern in embryonic development, W. R. Jeffre! and R. A. Raff, eds. (New York: Alan Liss), pp. 365-383.

Keller, G., Paige, C., Gilboa, E., and Wagner, E. F. (1985). Expression of a foreigr gene in myeloid and lymphoid cells derived from multipotent haematopoietii precursors. Nature 318, 149.

Keller, G., and Snodgrass, R. (1990). Life span of multipotential hematopoietic sten cells in vivo. J. Exp. Med. 171, 1407-1418.

Kenyon, C. (1994). If birds can fly, why can't we? Homeotic genes and evolution. Cel 78, 175-180.

Kessel, M., and Gruss, P. (1991). Homeotic transformations of mu rine prevertebrat and concornmitant alteration of Hox codes induced by retinoic acid. Cell 67, 89-104.

Kieran, M. W., Perkins, A. C., Orkin, S. H., and Zon, L. 1. (1996). Thrombopoietir rescues erythroid colonies from erythropoietin receptor deficient embryos. Proc. Natl Acad. Sci. USA 93, 9 126-91 31 . Kissinger, C. R., Liu, B. S., Martin-Blabco, E., Komberg, T. B., and Pabo C. 0.(1990) Crystal structure of an engrailed homodomain-DNA complex at 2.8 A resolution: < framework for understanding homeodomain-DNA interactions. Cell 63, 579-590.

Kongsuwan, K., Webb, E., Housiaux, P., and Adams, J. M. (1988). Expression O multible homeobox genes within diverse mammalian haemopoietic linages. EMBO J 7, 21 31 -2138.

Krumlauf, R. (1993). Hox genes and pattern formation in branchial region of thc vertebrate head. Trends Genet. 9, 106-112.

Krumlauf, R. (1994). Hox genes in vertebrate development. Cell 78, 191-201.

Krystal, G.,Lam, V., Dragowska, W., Takahashi, C., Appel, J., Gontier, A., Jenkins, A. Lam, H., Quon, L., and Lansdorp, P. (1994). Transforming growth factor Pl is ar inducer of erythroid differentiation. J. Exp. Med. 180, 851-860.

Kubo, R. T., Born, W., Kappler, J. W., Marrack, P., and Pigeon, M. (1989) Characterization of a monoclonal antibody which detects al1 murine T cell receptors. J Immunol. 342, 2736-2742.

Kulessa, H., Frampton, J., and Graf, T. (1995). GATA-1 reprograms aviar myelomonocytic cells into eosinophils, thromboblasts and erythroblast. Genes & Dev 9, 1250-1262. Landecker, H. L., Sinclair, D. A. R., and Brock, H. W. (1994). Screen for enhancers c Polycomb and Polycomblike in Drosophila melanogaster. Dev. Genet. 15, 425-434.

Langston, A. W., and Gudas, L. J. (1992). Identification ofa retinoic acid responsivc enhancer 3' of the murine homeobox gene Hox-1.6. Mech. Dev. 38, 217-228.

Lansdorp, P. M., and Dragowska, W. (1992). Long-term erythropoiesis from constan numbers of CD34+ cells in serum-free cultures initiated with highly purified progenito cells from human boria manow. J. Exp. Med. 175, t 501-1509.

Lawrence, H. J., Helgasan, C. D., Sauvageau, G., Fong, S., Izon, D. J., Humphries, R K., and Largman, C. (1996). Mice bearing a targeted interruption of the homeobo: gene HOXA9 have deffects in myeloid, erythroid and lymphoid hematopoiesis. Blood ln press.

Lawrence, H. J., Sauvageau, G., Ahmadi, N., Lopez, A. R., LeBeau, M. M., Link, M. Humphries, R. K., and Largrnan, C. (1995). Stage- and lineage-specific expression O the HOXA10 homeobox gene in normal and leukemic hematopoietic cells. Exp Hematol. 23, 1159-1 165.

Lawrence, H. J., Sauvageau, G., Humphries, R. K., and Largman, C. (1996). The rolt of HOX homeobox genes in normal and leukemic hematopoiesis. Stem Cells 14, 281 291.

Lawrence, H. J., Stage, K. M., Mathews, C. H. E., and Largman, C. (1993). Expressior of HOXC homeobox genes in lymphoid cells. Cell Growth Differen. 4, 665-669.

Lawrence, P. A., and Morata, G. (1994). Homeobox genes: their function in Drosophil; segmentation and pattern formation. Cell 78, 181-1 89.

Lawrence, P. A., and Struhl, G. (1996). Morphogens, compartments, and pattern lesson from Drosophila? Cell 85, 951-961 . Lemischka, 1. R., Raulet, D. H., and Mulligan, R. C. (1986). Developmental potentia and dynamic behaviour of hematopoietic stem cells. Cell 45, 91 7-927.

Lepault, F., Ezine, S., and Gagnerault, MA. (1993). T- and B-lymphocyte differentiation potentials of spleen colony-forming celIls. Blood 81, 950-955.

Lerner, C. P., and Harrison, 0. E. (1990). 5-fluorouracil spares hemopoietic stem celk responsible for long-term repopulation. Exp. Hematol. 18, 114.

Levine, M., and Hoey, 7. (1988). Homeobox proteins as sequence specific transcription factors. Cell 55, 537-540.

Lewis, E. (1978). A gene complex controlling segmentation in Drosophila. Nature 276 565-570.

Li, C. L., and Johnson, G. R. (1995). Murine hematopoietic stem and progenitor cells: 1, Enrichment and biological characterization. Blood 85, 1472. Li, C. L., and Johnson, G. R. (1994). Stem cell factor enhances the survival but not the self-renewal of murine hematopoietic long-term repopulating cells. Blood 84, 408-414.

Lichtman, M. A. (1981). The ultrastructure of the hematopoietic environment of the marrow. Exp. Hematol. 9, 391.

LilI, M. C., Fuller, J. F., Herzig, R., Crooks, G. M., and Gasson, J. C. (1995). The role oi the homeobox gene, HOX 87, in human myelomonocytic differentiation. Blood 85, 692-7.

Lin, L., and McGinnis, W. (1992). Mapping functional specificity in the Dfd and Ub% homeo domains. Genes Dev. 6, 1071 -1 081.

Lonai, P., Arman, E., Czosnek, H., Ruddle, F. H., and Blatt, C. (1987). New murine homeoboxes: structure, chromosomal assignment, and differential expression in aduli erythropoiesis. DNA 6, 409-418.

Lord, B. I., Dexter, T. M., Clements, J. M., Hunter, M. A., and Gearing, A. J. (1992). Macrophage-inflammatory protein protects multipotent hematopoietic cells from the cytotoxic effects of hydroxyurea in vivo. BIood 79, 2605-2609.

Lowney, P., Corral, J., LeBeau, M. M., Deaven, L., Lawrence, H. J., and Largman, C. (1991). A human Hox 1 homeobox gene exhibits myeloid-specific expression ol alternative transcripts in human hernatopoietic cells. Nucleic Acids Res. 19, 3443- 3449.

Lu, Q., and Kamps, M. P. (1996). Structural deterrninants within Pbxl that mediate cooperative DNA binding with pentapeptide-containing Hox proteins: proposal for a model of a Pbxl-Hox-DNA cornplex. Mol. Cell. Biol. 16, 1632-1640.

Lu, Q., Knoepfler, P. S., Scheele, J., Wright, D. D., and Kamps, M. P. (1995). 80th Pbxl and E2A-Pbxl bind the DNA motif ATCAATCAA cooperatively with the products of multiple murine Hox genes, some of which are themselves oncogenes. Mol. Cell. BioI. 15, 3786-3795.

Lumsden, A., and Krumlauf, R. (1996). Patterning the vertebrate neuraxis. Science 274, 1109-1115.

Lutz, B., Lu, H.-C., Eichele, G., Miller, D., and Kaufman, T. C. (1996). Rescue of Drosophila labial nul1 mutant by the chicken ortholog Hoxb-1 demonstrates that the function of Hox genes is phylogenetically conserved. Genes & Dev. 10, 176-184.

Mackarehtschian, K., Hardin, J. D., Moore, K. A., Boast, S., Goff, S. P., and Lemischka, 1. R. (1 995). Trageted disruption of the flk2lflt3 gene leads to deficiencies in primitive hematopoietic progenitors. lmrnunity 3, 147-161.

Magli, M. C., Barba, P., Celetti, A., De Vita, G., Cillo, C., and Boncinelli, E. (1991). Coordinated regulation of HOX genes in human hematopoietic cells. Proc. Natl. Acad. Sci. 88, 6348-6352.

Magli, M. C., Iscove, N. N., and Odartchenko, N. (1982). Transient nature of early haematopoietic spleen colonies. Nature 295, 572-529. Malicki, J., Schughart, K., and McGinnis, W. (1990). Mouse Hox-2.2 specifies thoracic segmental identity in Drosophila embryos and lavae. Cell 63, 961-967.

Mann, R. S., and Abu-Shaar, M. (1996). Nuclear import of the homeodomain proteir Extradenticle in response to Wg and Dpp signalling. Nature 383, 630-633.

Mann, R. S., and Hogness, D. S. (1990). Functional dissection of Ultrabithorax protein: in D. melanogaster. Cell 60, 597-610.

Markowitz, D., Goff, S., and Bank, A. (1988). Construction and use of a safe anc efficient amphotropic packaging cell line. Virol. 167, 400-406.

Markowitz, D., Goff, S., and Bank, A. (1988). A safe packaging line for gene transfer separarting viral genes on two different plasmids. J. Virol. 62, 1120-1 124.

Marshall, H., Studer, M., Popperl, H., Aparicio, S., Kuroiwa, A., Brenner, S., anc Krumlauf, R. (1994). A conserved retinoic acid respons element required for earlg expression of the homeobox gene Hoxb- 1. Nature 370, 567-571.

Mathews, C. H. E., Detmer, K., Boncinelli, E., Lawrence, H. J., and Largman, C. (1991) Erythroid-restricted expression of homeobox genes of the human HOX 2 locus. Blooc 78, 2248-2252.

Matsuzaki, K., Xu, J., Wang, F., McKeehan, W. L., Krummen, L., and Kan, M. (1993). P widely expressed transmembrane serinehhreonine kinase that does not bind activin, inhibin, transforming growth factor beta, or bone morphogenic factor. J Biol Chem 268, 12719-1 2723.

Matthews, W., Jordan, C. T.,Wiegand, G. W., Pardoll, D., and Lemischka, 1. R. (1991). P receptor tyrosine kinase specific to hematopoietic stem and progenitor cell-enrichec populations. Cell 65, 1143-1 152.

Mauch, P., and Hellman, S. (1989). Loss of hematopoietic stem cell self-renewal aftei bone marrow transplantation. BIood 74, 872-875.

McGinnis, N., Kuziora, M. A., and McGinnis, W. (1990). Human Hox-4.2 and drosophila deformed encode similar regulatory specificities in Drosophila embryo and larvae. Cell 63, 969-976.

McGinnis, W., Garber, R. L., Win, J., Kuroiwa, A., and Gehring, W. J. (1984). A homologous protein-coding sequence in Drosophila homeotic genes and its conservation in other metaroans. Cell 37, 403-408.

McGinnis, W., Hart, C. P., Gehring, W. J., and Ruddle, F. H. (1984). Molecular cloning and chromosome mapping of a mouse DNA sequence homologous to homeotic genes of Drosophila. CelI 38, 675-680.

McGinnis, W., Levine, M. S., Hafen, E., Kuroiwa, A., and Gehring, W. J. (1984). A consewed DNA sequence in homoeotic genes of the Drosophila Antennapedia and bithorax complexes. Nature 308, 428433. McKay, 1. J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A., and Lewis, , (1994). The kreisler mouse: a hindbrain segmentation mutant that lacks tw rhornbomeres. Development 120, 21 99-221 1.

McKercher, S. FI., Torbett, B. E., Anderson, K. L., Henkel, G. W., Vestal, D. J., Baribaul H., KIernsz, M., Feeney, A. J., Wu, G. E., Paige, C. J., and Maki, R. A. (1996). Targete distribution of PU.l gene results in multible hematopoietic abnomalities. EMBO J l! 5647-5658.

McMahon, A. P., and Bradley, A. (1990). The Wnt-1 (int-1) proto-oncogene is require for development of a large region of the mouse brain. CeII 62, 1073-1085.

Medvinsky, A., and Dzierzak, E. (1996). Definitive hematopoiesis is autonomousl initiated by the AGM region. Cell 86, 897-906.

Metcalf, D. (1993). Hematopoietic regulators: redundancy or subtlety. Blood 82, 3515 3523.

Metcalf, 0.(1 991). Lineage commitment of hemopoietic progenitor cells in developin blast cell colonies: lnfluence of colony-stimulating factors. Roc. Natl. Acad. Sci. US, 88, 11310-1 1314.

Monica, K., Galili, N., Nourse, J., Saltman, D., and Cleary, M. (1991). PBX2 and PBX: new homeobox genes with extensive homology to the human proto-oncogene PBX; Mol. Cell. Biol. 11, 6149-61 57.

Moretti, P., Simmons, P., Thomas, P., Haylock, D., Rathjen, P., Vadas, M., an' D'Andrea, R. (1994). ldentification of homeobox genes expressed in huma haemopoietic progenitor cells. Gene 144, 213-21 9.

Morrison, S. J., and Weissman, 1. L. (1994). The long-term repopulating subset c hematopoietic stem cells is deterministic and isolable by phenotype. lmmunity 1, 661 673.

Mucenski, M. L., McLain, K., Kier, A. B., Swerdlow, S. H., Schreiner, C. M., Miller, T. A Pietryga, D. W., Scott, J. W., and Potter, S. S. (1991). A functional c-myb gene i requered for normal fetal hepatic hematopoiesis. Cell 65, 677. Müller-Sieburg, C. E., and Riblet, R. (1996). Genetic control of the frequency c hematopoietic stem cells in mice: rnapping of a candidate locus to chromosome 1. . Exp. Med. 183, 1141-1 150.

Nakahata, T., and Ogawa, M. (1982). ldentification in culture of a class of hemopoieti~ colony-forming units with extensive capability to self-renew and generatc muitipotential hemopoietic colonies. Proc. Natl. Acad. Sci. USA 79, 3843-3847.

Nakamura, T., Largaespada, D. A., Lee, M. P., Johnson, L. A., Ohyashiki, K., Toyama K., Chen, S. J., Willman, C. L., Chen, 1.-M., Feinberg, A. P., Jenkins, N. A., Copeland, N G., and Shaughnessy Jr., J. D. (1996a). Fusion of the nucleoporin gene NUP98 tc HOXA9 by the chromosome translocation t(7;11)(p15;p15) in human myeloic leukaemia. Nature Genetics 12, 154-158. Nakamura, T., Largaespada, D. A., Shaughnessy Jr., J. D., Jenkins, N. A., and Copeland, N. G. (1996b). Cooperative activation of Hoxa and Pbxl-related genes in murine myeloid leukaemias. Nature Genetics 12, 149-153.

Nardelli-Haefliger, D., Bruce, A. E. E., and Shankland, M. (1 994). An axial domain of HodHox gene expression is formed by morphogenetic alignment of independently specified cell lineages in the leech Helobdella. Development 120, 1839-1849.

Nichols, J., and Nimer, S. D. (1992). Transcription factors, translocations, and leukemia. Blood 80, 2953-2963.

Niswander, L., Jeffrey, S., Martin, G. R., and Tickle, C. (1994). A positive feedback loop coordinates growth and patterning in the vertebrate Iimb. Nature 371, 609-612.

Nonchev, S., Vesque, G., Maconochie, M., Seitanidou, T., Ariza-McNaughton, L., Frain, M., Marshall, H., Sham, M. H., Krumlauf, R., and Chamay, P. (1996). Segmentai expression of Hoxa-2 in the hindbrain is directly regulated by Krox-20. Development 122, 543-554.

Nose, A., Mahajan, V., and Goodman, C. S. (1992). Connectin: a homophilic ce11 adhesion molecule expressed on subset of muscles and the motoneurons that innervate them in Drosophila. Cell 70, 553-567.

Nourse, J., Mellentin, J. D., Galili, N., Wilkinson, J., Stanbridge, E., Smith, S. D., and Cleary, M. (1990). Chromssomal translocation t(1;19) results in synthesis of a homeobox fusion mRNA that codes for a potential chimeric transcription factor. Cell 60, 535-545.

Nuez, B., Michalovich, D., Bygrave, A., Ploemacher, R., and Grosveld, F. (1995). Defective hematopoiesis in fetal liver resulting from inactivation of the EKLF gene. Nature 375, 316.

Nusse, R., and Varmus, H. E. (1992). Wnt genes. Cell 69, 1073-1087.

Ogawa, M. (1993). Differentiation and proliferation of hematopoietic stem cells. Blood 8 1, 2844-2853.

Ogden, D. A., and Micklem, H. S. (1976). The fate of serially transplanted bone marrow cell populations from young and old donors. Transplantation 22, 287-293.

Ogura, T., and Evans, R. M. (1995). Evidence for two distinct retinoic acid response pathways for HOXB1 gene regulation. Proc Natl Acad Sci Usa 92, 392-396.

Ogura, T., and Evans, R. M. (1995). A retinoic acid-triggered cascade of HOXB1 gene activation. Proc Natl Acad Sci Usa 92, 387-391.

Ohara, A., Suda, T., Tokuyama, N., Suda, J., Nakayama, K., Miura, Y., Nishikawa, S., and Nakauchi, H. (1991). Generation of B lymphocytes from a single hemopoietic progenitor cell in vitro. Int. Immunol. 3, 703. Okuda, T., van Deursen, J., Hiebert, S. W., Grosveld, G., and Downing, J. R. (1996). AML1, the target of multible chromosomal translocations in human leukemia, is essential for normal fetal liver hematopoiesis. Cell 84, 321-330.

Olson, M. C., scott, E. W., Hack, A. A., Su, G. H., Tenen, D. G., Singh, H., and Simon, M. C. (1995). PU.1 is not essential for early myeloid gene expression but is required for terminal myeloid differentiation. lmmunity 3, 703-714.

Orkin, S. H. (1995). Transcription factors and hematopoietic development. J. Biol. Chem. 270,4955-4958.

Orkin, S. T. (1 996). Development of the hematopoietic system. Curr. Opin. Genet. Dev. 6, 597-602.

Orlando, V., and Paro, R. (1 995). Chromatin multiprotein complexes involved in the maintenance of transcription patterns. Curr. Opin. Genet. Dev. 5, 174-179.

Orlic, D., Fischer, R., Nishikawa, S. I., Nienhuis, A. W., and Bodine, D. M. (1993). Purification and characterization of heterogeneous pluripotent hematopoietic stem celi populations expressing high levels of c-kit receptor. Blood 82, 762-770.

Palacios, R., and Nishikawa, S. 1. (1992). Developmentally regulated cell surface expression and function of c-kit receptor during lymphocyte ontogeny in the embryo and adult mice. Development 175, 1 133-1 147.

Palis, J., Kingsley, P. D., McGrath, K. E., and Silver-Morse, L. (1994). Gene expression during early mouse yolk sac development. Blood 84, 75a.

Parr, B. A., and McMahon, A. P. (1995). Dorsalizing signal Wnf-7a required for normal polarity of D-V and A-P axes of mouse Iimb. Nature 374, 350-353.

Patel, N. H., Kornberg, T. B., and Goodman, C. S. (1989). Expression of engrailed during segmentation in grasshooper and crayfish. Development 107, 201-21 2.

Pawliuk, R., Eaves, C. J., and Humphries, R. K. (1996). Evidence of both ontogeny and transplant dose regulated expansion of hematopoietic stem cells in vivo. Blood 8, 2852-2858. Pawliuk, R., and Humphries, R. K. (Unpublished data). . Peifer, M., and Wieschaus, E. (1990). Mutations in the Drosophila gene extradenticle affects the way specific horneo domain proteins regulate segmental identity. Genes Dev, 4, 1209-1223.

Perkins, A., Kongsuwan, K., Visvader, J., Adams, J. M., and Cory, S. (1990). Homeobox gene expression plus autocrine growth factor production elicits myeloid leukemia. Proc. Natl. Acad. Sci, USA 87, 8398-8402.

Perkins, A. C,, and Cory, S. (1993). Conditional immortalization of rnouse myelomonocytic, megakaryocytic and mast cell progenitors by the Hox-2.4 homebox gene. EMBO J. 12, 3835-3846. Perkins, A. C., Sharpe, A. H., and Orkin, S. H. (1995). Lethal B-thalassemia in mici lacking the erythroid CACCC-transcription factor EKLF. Nature 375, 31 8.

Perrimon, N. (1994). The genetic basis of pattemed baldness in Drosophila. Cell 76 78 1-784.

Peterson, C. J., and Tamkun, J. W. (1995). The SWI-SNF complex: a chromatir remodeling machine. Trends Biochem. Sci., 143-146.

Petrini, M., Quaranta, M. T., Testa, U., Samoggia, P., Tritarelli, E., Carè, A., Cianetti, L. Valtieri, M., Barletta, C., and Peschle, C. (1992). Expression of selected human HOX-2 genes in BK acute lymphoid leukemia and interleukin-2linterleukin-1 beta-stimulatec natural killer lymphocytes. Blood 80, 185-193.

Pevny, L., Simon, M. C., Robertson, E., Klein, W. H., Tsai, S. F., D'Agati, V., Orkin, S. H. and Costantini, F. (1991). Erythroid differentiation in chimaeric mice blocked by ; targeted mutation in the gene for transcription factor GATA-1. Nature 349, 257-260.

Phelan, M. L., Rambaldi, I., and Featherstone, M. (1995). Cooperative interaction: between HOX and PBX proteins mediated by a conserved peptide motif. Mol. Cell Biol. 15, 3989-3997.

Phelan, M. L., Sadoul, R., and Featherstone, M. S. (1994). Functional differencei between HOX proteins conferred by two residues in the homeodomain N-terminal am Mol. Cell. Biol. 14, 5066-5075.

Phillips, R. L., Reinhart, A. J., and Van Zant, G. (1992). Genetic control of murinf hematopoietic stem cell pool sire and cycling kinetics. Proc. Natl. Acad. Sci. USA 89 11607-11611. Pinneault, N., and Humphries, R. K. (unpublished). . Piverali, A. F., D'Esposito, M., Acampora, O., Bunone, G., Negri, M., Faiella, A. Stornaiuolo, A., Pannese, M., Migliaccio, E., Simeone, A., Della Valle, G., anc Boncinelli, E. (1 990). Expression of HOX homeogenes in human neuroblastoma cel culture lines. Differentiation 45, 61 -69.

Ploemacher, R. E., and Brons, N. H. C. (1988). Cells with marrow and spleer repopulating ability and forming spleen colonies on day 16, 12 and 8 are sequentiall) ordered on the basis of increasing rhodamine retention. J. Cell Physiol. 136, 230.

Ploemacher, R. E., Van der Loo, J. C. M., Van Beurden, C. A. J., and Baert, M. R. M, (1993). Wheat germ agglutinin affinity of murine hemopoietic stem cell subpopulations is an inverse function of their long-term repopulating ability in vitro and in vivo. Leukemia 7, 120.

Popperl, H., Bienz, M., Studer, M., Chan, S.-K., Aparicio, S., Brenner, S., Mann, R. S., and Krumlauf, R. (1995). Segmental expression of Hoxb-1 is controlled by a highly conserved autoregulatory loop dependent upon exdlpbx. Cell 81, 1031-1 042. Popperl, H., and Featherstone, M. S. (1993). Identification of a retinoic acid response element upstream of the murine Hox-4.2 gene. Mol. Cell. Biol. 13, 257-265. Porcher, C., Swat, W., Rockweli, K., Fujiwara, Y., AIt, F. W., and Orkin, S. (1996). The T cell leukemia oncoprotein SCUtal-1 is essential for develpment of al1 hematopoietic lineages. Cell 86, 47-57.

Puschel, A., Balling, R., and Gruss, P. (1990). Position-specific activity of the Hoxl. 7 promoter in transgenic mice. Development 108, 435-442.

Quaranta, M. T., Petrini, M., Tritarelli, E., Samoggia, P., Care, A., Bottero, L., Testa, U., and Peschle, C. (1996). HOXB cluster genes in activated natural killer lymphocytes. J. Immunol. 156, 2462-2469.

Rauskolb, C., Peifer, M., and Wieschaus, E. (1993). extradenticle, a regulator of homeotic gene activity, is a homolog of the homeobox-containing human proto- oncogene pbxl . Cell 74, 1-20. RebeI, V. I., Dragowska, W., Eaves, C. J., Humphries, R. K., and Lansdorp, P. M. (1994). Amplification of Sca-l+ Lin- WGA+ cells in serum-free cultures containing steel factor, interleukin-6, and erythropoietin with maintenance of cells with long-term in vivo reconstituting potential. Blood 83, 128-136.

Rebel, V. I., Miller, C. L., Eaves, C. J., and Humphries, R. K. (1996). The repopulating potential of fetal liver hematopoietic stem cells in mice exceeds that of their adult bone marrow coünterparts. Blood 87, 3500-3507.

Rich, 1. N., and Zimmermann, F. (1995). The homeobox (HOX) 6 gene is expressed in erythropoietin-producing locations and in erythropoietic cells but not in stem cells during murine development. Blood 86, 301 a. Riddle, R. D. (1995). Induction of the LIM homeobox gene Lmxl by Wnta establishes dorsoventral pattern in the vertebrate limb. CeIl 83, 631-640.

Riddle, R. O., Johnson, R. L., Laufer, E., and Tabin, C. (1993). Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75, 1401-141 6.

Roberts, C.,Shutter, J. R., and Korsmeyer, S. J. (1994). Hox11 controls the genesis of the spleen. Nature 368, 747-749.

Rosenfeld, M. G. (1991). POU-domain transcription factors: powerfull developmental regulators. Genes Dev. 5, 897-907.

Ruscetti, F. W., Jacobsen, S. E., Birchenall, R. M., Broxmeyer, H. E., Engelmann, G. L., Dubois, C., and Keller, J. R. (1991). Role of transforming growth factor-betal in regulation of hematopoiesis. Ann. Ny. Acad. Sci. 628, 31-43.

Saffran, D. C., Faust, E. A., and Witte, O. N. (1992). Establishment of a reproducible culture technique for the selective growth of B-cell progenitors. Curr. Top. Microbiol. Immunol. 182, 37-43.

Salser, S. J., and Kenyon, C. (1994). Patterning C. elegans: Horneotic cluster genes, cell fates and cell migrations. Trends Genet 10, 159-164. Sanchez-Garcia, I., and Rabbits, T. H. (1994). The LIM domain: a new structural moti found in zinc-finger-like proteins. Trends Genet. 10, 315-320.

Sauvageau, G., Lansdorp, P. M., Eaves, C. J., Hogge, D. E., Dragowska, W. H., Reid D. S., Largman, C., Lawrence, H. J., and Humphries, R. K. (1994). Differentia expression of homeobox genes in functionally distinct CD34+ subpopulations O, human bone marrow cells. Proc. Natl. Acad. Sci. 91, 12223-12227.

Sauvageau, G., Rosten, P., Pineault, N., and Humphries, R. K. (1997). Manuscript ir preparation. .

Sauvageau, G., Thorsteindottir, U., Eaves, C. J., Lawrence, H. T., Largman, C. Lansdorp, P. M., and Humphries, R. K. (1995). Overexpression of HOXB4 ir hematopoietic cells causes the selective expansion of more primitive populations ir vitro and in vivo. Genes and Dev. 9, 1753-1766.

Sauvageau, G., Thorsteinsdottir, U., Hough, M. R., Hugo, P., Lawrence, H. J., Largman C., and Humpries, R. K. (1996). Deregulated expression of HOXB3 in hematopoietic cells causes defective lymphoid development and progressive myeloproliferation Manuscript in preparation. Sauvageau, G., Thorsteinsdottir, U., and Humphries, R. K. (Unpublished data). . Sawada, S., Scarborough, J. D., Killeen, N., and Littman, 0. R. (1994). A lineage. specfic transcriptionai silencer regulates CD4 gene expression during T lymphocyt~ development. Cell 77, 917-929.

Scott, E. W., Simon, M. C., Anastasi, J., and Singh, H. (1994). Requirement oi transcription factor PU.l in the development of multiple hematopoietic lineages, Science 265, 15731577.

Scott, M. P. (1992). Vertebrate homeobox gene nomenclature. Cell 71, 551-553.

Scott, M. P., and Weiner, A. J. (1984). Structural relationships among genes thai control development: sequence homology between ths Antennapedia, Ultrabithorar and fushi tarazu loci of Drosophila. Proc. Natl. Acad. Sci. USA 81, 41 15-4119.

Sham, M. H., Hunt, P., Nonchev, S., Papalopulu, N., Graham, A., Boncinelli, E., anc Krumlauf, R. (1992). Analysis of the murine Hox-2.7 gene: Conserved alternative transcripts with differential distributions in the nervous system and the potential foi shared regulatory regions. EMBO J. 7 1, 1825-1836.

Sham, M. H., Vesque, C., Nonchev, S., Marshall, H., Frain, M., Das, G. R., Whiting, J., Wilkinson, D., Charnay, P., and Krumlauf, R. (1993). The zinc finger gene Krox2C regulates HoxB2 (Hox2.8) during hindbrain segmentation. Cell 72, 183-196. Shankland, M. (1994). Leech segmentation: a genetic perspective. Bioessays 76, 801- 808.

Shen, W.-F., Detmer, K., Mathews, C. H. E., Hack, F. M., Morgan, D. A., Largman, C., and Lawrence, H. J. (1992). Modulation of homeobox gene expression alters the phenotype of a human hematopoietic cell line. EMBO J. 7 1, 983-989. Shen, W.-F., Largman, C., Lowney, P., Hauser, C. A., Simonitch, T. A., Hack, F. M., ani Lawrence, H. J. (1989). Lineage-restricted expression of homeobox-containing gene: in human hematopoietic cell lines. Proc. Natl. Acad. Sci. USA 86, 8536-8540.

Shivdasani, R. A., Mayer, E. L., and Orkin, S. H. (1995a). Absence of blood forrnatioi in mice lacking the T-cell leukaemia oncoprotein tal-11SCL. Nature 373, 432-434.

Shivdasani, R. A., and Orkin, S. H. (1996). The transcriptional control O hematopoiesis. Wood 87, 4025-4039.

Shivdasani, R. A., Rosenblatt, M. F., Zucker-Franklin, D., Jackson, C. W., Hunt, P. Saris, C. J. M., and Orkin, S. H. (1995b). Transcription factor NF-E2 is required fo platelet formation independent of the action of thrombopoietin/MGDF ir megakaryocyte development. Cell 81, 695-704.

Sieweke, M. H., Tekotte, H., Frampton, J., and Graf, T. (1996). MafB is an interaction partner and repressor of Ets-1 that inhibits erythroid differentiation. Ce11 85, 49-60.

Simeone, A., Pannese, M., Acampora, D., DIEsposito, M., and Boncinelli, E. (1988). A hast three human homeoboxes on chromosome 12 belong to the same transcriptior unit. Nucl. Acids Res. 16, 5379-5390.

Siminovitch, L., McCulioch, E. A., and Till, J. E. (1963). The distribution of colony forming cells among spleen colonies. J. Cell. Comp. Physiol. 62, 327-336.

Simon, J. (1995). Locking in stable states of gene expression: transcriptional contro during Drosophiia development. Curr. Opin. Cell Biol. 7, 376-385.

Siu, G., Wurster, A. L., Duncan, D. D., Soliman, T. M., and Hedrick, S. M. (1994). P transcriptional silencer controls the developmental expression of the CD4 gene EMBO J. 13, 3570-3579.

Spangrude, G. J., and Brooks, D. M. (1993). Mouse strain variability in the expressior of the hematopoietic stem cell antigen Ly-GAIE by bone marrow cells. Blood 82, 3327. 3332.

Spangrude, G. J., Brooks, O. M., and Tumas, D. B. (1995). Long-terrn repopulation 01 irradiated mice with limiting numbers of purified hematopoietic stem cells: In vivc expansion of stem celi phenotype but not function. Blood 85, 1006-101 6.

Spangrude, G. J., Heimfeld, S., and Weissman, 1. L. (1988). Purification anc characterization of mouse hematopietic stem cells. Science 241, 58-62.

Spangrude, G. J., and Johnson, G. R. (1990). Resting and activated subsets of mouse multipotent hematopietic stem cells. Proc. Natl. Acad. Sci. USA 87, 7433-7437.

Spangrude, G. J., and Scollay, R. (1990). Differentiation of hematopoietic stem cells in irradiated mouse thymic lobes. Kinetics and phenotype of progeny. J. Immunol. 145, 3661-3668. St Johnston, D., and Nusslein-Volhard, C. (1992). The origin of pattern and polarity in the Drosophila embryo. Cell 68, 201 -21 9.

Stark, K., Vainio, S., Vassileva, G., and McMahon, A. P. (1994). Epithelial transformation of metanephric mesenchyme in the developing kidney regulated by Wnt-4. Nature 372, 679-683.

Struhl, G., and White, R. (1985). Regulation of the Ultrabithorax gene by other bithorm complex genes. Cell 43, 507-519.

Stuart, E. T., and Gruss, P. (1996). PAX: developmental control genes in cell growth and differentiation. Cell Growth & Differ. 7, 405-412.

Suda, T., Okada, S., Suda, J., Miura, Y., Ito, M., Sudo, T., Hayashi, S., Nishikawa, S., and Nakauchi, H. (1989). A stimulatory effect of recombinant murine interleukin-7 (IL- 7) on %-cellcolony formation and an inhibitory effect of IL-1a. Blood 74, 1936-1941.

Suda, T., and Zlotnik, A. (1993). Origin, differentiation, and repertoire selection 01 CD3+CD4-CD8- thymocytes bearing either alpha beta or gamma delta T ceIl receptors. J. Immunol. 150, 447-455.

Sun, B., Hursh, D. A., Jackson, D., and Beachy, P. A. (1995). Ultrabithorax protein is necessary but not sufficient for full activation of decapentaplegic expression in the viseral rnesoderm. EMBO J. 14, 520-535.

Sutherland, H. J., Lansdorp, P. M., Henkelman, D. H., Eaves, A. C., and Eaves, C. J. (1990). Functional characterization of individual human hematopoietic stem cells cultured at limiting dilution on supportive marrow. Proc. Natl. Acad. Sci. USA 87, 3584- 3588. Szilvassy, S. J., and Cory, S. (1993). Phenotypic and functional characterization of competitive long-term repopulating hematopoietic stem cells enriched from 5- fluorouracil-treated murine marrow. Blood 81, 2310-2320.

Szilvassy, S. J., Humphries, R. K., Lansdorp, P. M., Eaves, A. C., and Eaves, C. J. (1990). Quantitative assay for totipotent reconstituting hematopoietic stem cells by a competitive repopulation strategy. Proc. Natl. Acad. Sci. 87, 8736-8740. Tabin, C. (1995). The initiation of the limb bud: growth factors, Hox genes, and retinoids. Cell 80, 671 -674.

Takeshita, K., Bollekens, J. A., Hijiya, N., Ratajczak, M., Ruddle, F. H., and Gewirtz, A. M. (1993). A homeobox gene of the Antennapedia class is required for human adult erythropoiesis. Proc. Natl. Acad. Sci. 90, 3535-3538.

Tamkun, J. W. (1995). The role of brahma and related proteins in transcription and development. Curr. Opin. Genet. Dev. 5, 473-477.

Taniguchi, Y., Komatsu, N., and Moriuchi, T. (1995). Overexpression of HOX4A (HOXD3) homeobox gene in human erythroleukemia HEL cells results in altered adhesive properties. Blood 85, 2786-2794. Tear, G., Akam, M., and Martinez-Arias, A. (1990). Isolation of an abdominal-A gene from the locust Schistocerca gregaria and its expression during early embryogenesis. Development 110, 91 5-925.

Thali, M., Muller, M. M., DeLorenzi, m., mathias, P., and Bienze, M. (1988). Drosophila homeotic genes encode transcriptional activators similar to OTF-2. Nature 336, 598- 600.

Thomas, K. R., and Capecchi, M. R. (1987). Site-directed mutagenesis by gene targetting in mouse embryo-derived stem cells. Cell 51, 503.

Thompson, M. A., and Ramsay, R. G. (1995). Myb: an old oncoprotein with new roles. BioEssays 17,341 -350.

Thorsteinsdottir, U., Sauvageau, G., Hough, M. R., Dragowska, W., Lansdorp, P., Lawrence, H. J., Largman, C., and Humphries, R. K. (1996). Overexpression of HOXAIO in murine hematopoietic cells perturbs both myeloid and lymphoid differentiation and leads to acute myeloid leukemia. Mol. Cell. Biol. In press.

Till, J. E., and McCulloch, E. A. (1961). A direct measurement of the radiation sensitivity of normal mouse bone marrow cells. Radiat. Res. 14, 213-222.

Tkachuk, D. C., Kohler, S., and Cleary, M. L. (1992). Involvement of a homolog of Drosophila trithorax &y 1 lq23 chromosomal translocation in acute ieukemias. Celi 71, 69 1-700.

Tomotsune, D., Shoji, H., Wakamatsu, Y., Kondoh, H., and Takahashi, N. (1993). A mouse homologue of the Drosophila tumor-suppressor gene l(2)gl controlled by Hox- C8 in vivo. Nature 365, 69-72.

Tsai, F. Y., Keller, G., Kuo, F. C., Weiss, M., Chen, J., Rosenblatt, M., Alt, F. W., and Orkin, S. H. (1994). An early haematopoietic defect in mice lacking the transcriptional factor GATA-2. Nature 371, 221 -226.

Tsai, S.-F., Martin, D. 1. K., fon, L., D'Andrea, A. D., Wong, G. G., and Orkin, S. H. (1989). Cloning of cDNA for the major DNA-binding protein of the erythroid lineage through expression in mammalian cells. Nature 339, 446-451.

Unkeless, J. C. (1979). Characterization of a monoclonal antibody directed against mouse macrophage and lymphocyte Fc receptors. J. Exp. Med. 150, 580-596.

Urbanek, P., Wang, Z .-Q., Fetka, I., Wagner, E., and Busslinger, M. (1994). Complete block in early B cell differentiation and altered patterning of the posterior midbrain in mice lacking PaxWBSAP. Cell 79, 901.

Valge-Archer, V. E., Osada, H., Warren, A. J., Forster, A., Li, J., Baer, R., and Rabbits, T. H. (1994). The LIM protein RBTN2 and the basic helix-loop-helix protein TA1 are present in a complex in erythroid cells. Proc. Natl. Acad. Sci. USA 91, 8617.

Van Den Berg, O. J., Sharma, A., Bruno, E., Altenhofen, J., Brant, J., and Hoffman, R. (1 996). A role for members of the Wnt gene family in human hematopoiesis. Blood 88, Supplement 1, 141a. van der Hoeven, F., Zakany, J., and Duboule, D. (1996). Gene transpositions in th€ HoxD complex reveals a hierarchy of regulatory controls. Cell 85, 1025-1035. van der Lugt, N. M. T., Domen, J., Linders, K., van Roon, M., Robanus-Maandag, E., te Riele, H., van der Valk, M., Deschamps, J., Sofroniew, M., van Lohuizen, M., anc Berns, A. (1994). Posterior transformation, neurological abnorrnalities, and sever€ hematopoietic defects in mice with a targeted deletion of the bmi-1 proto-oncogene, Genes Dev. 8, 757-769. van Dijk, M., and Murre, C. (1994). extradenticle raises the DNA binding specificity 01 homeotic selector gene products. Cell 78, 617-624. van Lohuizen, M., Verbeek, S., Scheijen, B., Wientjens, E., van der Gulden, H., and Berns, A. (1991). Identification of cooperating oncogenes in Ep-mye transgenic mice by provirus tagging. Cell 65, 737-752.

Van Zant, G. (1984). Studies of hematopoietic stem cells spared by 5-fluorouracil. J. Exp. Med. 159, 679.

Van Zant, G., Scott-Micus, K., Thompson, B. P., Fleischman, R. A., and Perkins, S, (1992). Stem cell auiescence/activation is reversible bv serial trans~lantationand is hdependent of stroma1 cell genotype in mouse aggregation chimeras. Exp. Hematol. 20, 470-475.

Van Zant, G., Thompson, B. P., and Chen, J. (1991). Differentiation of chimeric bone marrow in vivo reveals genotype-restricted contributions to hematopoiesis. Exp. Hematol. 19, 941-949.

Vieille-Grosjean, I., and Huber, P. (1995). Transcription factor GATA-1 regufates human HOXB2 gene expression in erythroid cells. J Biol Chem 270, 4544-4550.

Vieille-Grosjean, I., Roullot, V., and Courtois, G. (1992). Lineage and stage specific expression of HOX 1 genes in the human hematopoietic system. Biochem. Biophys. Res. Commun. 183, 1124-1 130.

Visser, J. W. M., Bauman, J. G. J., Mulder, A. H., Eliason, J. F., and deleeuw, A. M. (1984). Isolation of murine pluripotent hemopoietic stem cells. J. Exp. Med. 59, 1576- 1589.

Visvader, J. E., Crossley, M., Hill, J., Orkin, S. H., and Adams, J. M. (1995). The C- terminal zinc finger of GATA-1 and GATA-2 is sufficient to induce megakaryocytic differentiation of an early myeloid cell line. Mol. Cell. Biol. 15, 634-641.

Visvader, J. E., Elefanty, A. G., Strasser, A., and Adams, J. M. (1992). GATA-1 but not SCL induces megakaryocytic differentiation in an early myeloid line. Embo J 17, 4557- 4564.

Vogel, A., Rodrigues, C., Warnken, W., and Izpisua-Belmonte, J. C. (1995). Dorsal fate specified by chick Lmxl during vertebrate limb development. Nature 378, 716-720. Vogels, R., Charite, Je, de Graaff, W., and Deschamps, J. (1993). Proximal cis-acting elements cooperate to set Hox-b7 (Hox-2.3) expression boundaries in transgenic mice. Development 7 17, 71-82.

Wadman, I., Li, J., Bash, R. O., Forster, A., Osada, H., Rabbits, T. H., and Baer, R. (1994). Specific in vivo association between the bHLH and LIM proteins implicated in T cell leukemia. EMBO J., 4831.

Warren, A. J., Colledge, W. He,Carlton, M. B. L., Evans, M. J., Smith, A. J. H., and Rabbits, T. H. (1994). The oncogenic cysteine-rich LIM domain protein Rbtnê is essential for erthroid development. Cell 78, 45.

Watrin, F., and Wolgemuth, D. J. (1993). Conservation and divergence of patterns of expression and lineage-specific transcripts in orthologoues and paralogoues of mouse Hox-1.4 gene. Dev. Biol. 156, 136-145.

Weiss, M. J., Keller, G., and Orkin, S. H. (1994). Novel insights into erythroid development revealed through in vitro differentiation of GATA-1- em bryonic stem cells. Genes Dev. 8, 1184.

Weiss, M. J., and Orkin, S. H. (1995). Transcription factor GATA-1 permits survival and maturation of erythroid precursors by preventing apoptosis. Proc. Natl. Acad. Sci. USA 92, 9623.

Whiting, J., Marshall, H., Cook, M., Krumlauf, R., Rigby, P., Stott, D., and Allemann, R. (1991). Multible spatially-specific enhancers are required to reconstruct the pattern of Hox-2.6 gene expression. Genes Dev. 5, 2048-2059.

Wilkinson, D., Bhatt, S., Cook, M., Boncinelli, E., and Krumlauf, R. (1989). Segmental expression of hox2 homeobox-containing genes in the developing mouse hindbrain. Nature 34 7, 405-409.

Wolberger, C., Vershon, A. K., Liu, B. S., Johnson, A. D., and Pabo, C. 0. (1991). Crystal structure of a MATa2 homeodcmain-operator cornplex suggests a general model for homeodomain-DNA interactions. Cell 67, 51 7-528.

Wong, P. M., Chung, S.-W., Dunbar, C. E., Bodine, D. M., Ruscetti, C., and Nienhuis, A. W. (1 989). Retrovirus-mediated transfer and expression of interleukin-3 gene in mouse hematopoietic cells results in a myeloproliferative disorder. Mol. Cell. Biol. 9, 789-808.

Wu, A. M., Siminovitch, L., Tiii, J. E., and McCulloch, E. A. (1968). Evidence for a relationship between mouse hematopoietic stem cells and cells forming colonies in cultures. Proc. Natl. Acad. Sci. USA 59, 1209.

Wu, A. M., Till, J. E., Siminovitch, L., and McCulloch, E. A. (1968). Cytological evidence for a relationship between normal hematopoietic colony-forming cells and cells of the lymphoid system. J. Exp. Med. 127, 455-464.

Wu, H., Liu, X., Jaenisch, R., and Lodish, H. F. (1995). Generation of committed erythroid BFU-E and CFU-E progenitors does nor require the erythropoietin or the eryt h ropoietin receptor. Cell 83, 59-67. Wu, J., Zhu, J.-Q., Zhu, D.-X., Scharfman, G., Lamblin, G., and Han, K.-K. (1992). Selective inhibition of normal murine myelopoiesis in vitro by a Hox 2.3 antisense oligodeoxynucleotide. Cell. Mol. Biot. 38, 367-376.

Wu, L., Antica, M., Johnson, G. R., Scollay, R., and Shortman, K. (1991). Developmental potential of the earliest precursor cells from the adult thymus. J. Exp. Med. 174, 1617.

Wu, L., Scollay, R., Egerton, M., Pearse, M., Spangrude, G. J., and Shortman, K. (1991). CD4 expressed on earliest T-lineage precursor cells in the adult murine thymus. Nature 349, 71-74.

Yamomoto, M., Ko, L. J., Leonard, M. W., Beug, H., Orkin, S. H., and Engel, J. O. (1990). Activity and tissue-specific expression of the transcription factor NF-€1 multigene family. Genes Dev. 4, 1650.

Yu, B. D., Hess, J. L., Horning, S. E., Brown, G. A. J., and Krosmeyer, S. J. (1995). Altered Hox expression and segmenta1 idenity in Ml/-mutant mice. Nature 378, 505- 508.

Zappavigna, V., Sartori, D., and Mavilio, F. (1994). Specificity of HOX protein function depends on DNA-protein and protein-protein interactions, both mediated by the homeo domain. Genes Dev. 8, 732-744.

Zeng, W., Andrew, D. J., Mathies, L. D., Horner, M. A., and Scott, M. P. (1993). Ectopic expression and function of the Antp and Scr homeotic genes: the N terminus of the homeodomain is critical to functional specificity. Development 118, 339-352.

Zhao, J. J., Lazzarini, R. A., and Pick, L. (1996). Functional dissection of the mouse Mx-a5gene. EMBO J. 15, 1313-1322.

Zhao, J. J., Lazzarini, R. A., and Pick, L. (1993). The mouse Hox-1.3 gene is functionally equivalent to the Drosophila Sex combs reduced gene. Genes Dev. 7, 343-354.

Zheng, C., Pinsonneault, J., Gellon, G., McGinnis, N., and McGinnis, W. (1994). Deformed protein binding sites and cofactor binding sites are required for the function of a small segment-specific regulatory element in Drosophila embryos. EMBO J. 13, 2362-2377.

Zhuang, Y., Soriano, P., and Weintraub, H. (1994). The helix-loop-helix gene E2A is required for B ce11 formation. Cell 79, 875.

Zlotnik, A., and Moore, T. A. (1995). Cytokine production and requirements during T- ceIl development. Curr. Opin. Immunol. 7, 206-213.

Zon, L. 1. (1995). Developmental biology of hematopoiesis. Slood 86, 2876-2891. l MAGE EVALUATI ON TEST TARGET (QA-3)

APPLIEO k IWGE . lnc 1653 East Main Street -.- ~och&ter.fi 14609 USA I -,-- Phone: 716/482-0300 SB-- Faic 7161288-5989

O 1993,Wied Image, Inr. Ni Rights Rase-