The bone marrow niche, stem cells, and leukemia: impact of drugs, chemicals, and the environment

Abstract

Hematopoietic stem cells (HSCs) are a unique population of somatic stem cells that can both self-renew for long-term reconstitution of HSCs and differentiate into hematopoietic progenitor cells, which in turn give rise, in a hierarchical manner, to the entire myeloid and lymphoid lineages. The differentiation and maturation of these lineages occurs in the bone marrow niche, a microenvironment that regulates self-renewal, survival, differentiation, and proliferation, with interactions among signaling pathways in the HSCs and the niche required to establish and maintain homeostasis. The accumulation of genetic mutations and cytogenetic abnormalities within cells of the partially differentiated myeloid lineage, particularly as a result of exposure to benzene or cytotoxic anticancer drugs, can give rise to malignancies like acute myeloid leukemia and myelodysplastic syndrome. Better understanding of the mechanisms driving these malignancies and susceptibility factors, both within hematopoietic progenitor cells and cells within the bone marrow niche, may lead to the development of strategies for prevention of occupational and cancer therapy–induced disease.

Introduction^a

All mature blood cells are derived from a common cellular ancestor, the hematopoietic stem cell (HSC), which divides to produce another HSC and a hematopoietic progenitor cell (HPC). HPCs, in turn, divide to produce the progenitor cells of either the myeloid or lymphoid lineages (Fig. 1). This process of blood cell differentiation and maturation occurs not in isolation but in the specialized and complex environment of the bone marrow (BM) niche (Fig. 2).² It is almost always the immature, partially differentiated progenitor cells of the myeloid lineage that acquire genetic mutations and cytogenetic abnormalities, and, consequently, multiply to give rise to neoplastic diseases such as myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), the most common malignancies that result from exposures either to benzene or to cytotoxic antitumor therapeutic agents. Lymphoid malignancies also occur both in cancer survivors who were treated with cytotoxic drugs and in individuals exposed to benzene, but much less commonly than neoplasms of the myeloid lineage. True stem cell leukemias, without cell surface markers of either myeloid or lymphoid lineages, are so rare that little can be said about their possible causes.³

Figure 1.

Differentiation pathways for myeloid and lymphoid cells, which are all ultimately derived from the hematopoietic stem cell (HSC). HSCs divide to produce another HSC and a hematopoietic progenitor cell (HPC), or multipotent stem cell, from which cells of both myeloid and lymphoid lineages evolve. Treatment-related and benzene-related leukemias most commonly are of the myeloid lineage (boxed) and result from accumulation of mutations in specific genes, some of which confer a proliferation or survival advantage while others result in impaired hematopoietic differentiation. Figure courtesy of Clayton Smith

Figure 2.

The bone marrow niche, showing hematopoietic and stromal cell differentiation. Figure courtesy of Richard Irons.

For nearly two centuries the group of diseases that we term the leukemias has been under intense investigation at both the lab bench and the bedside by experimental hematologists, pathologists, epidemiologists, and clinicians, all with their own ways of thinking, their own methods and techniques, and their own vocabularies. For decades, toxicologists have been intent on understanding how exposure to a variety of chemical scan, like exposure to ionizing radiation, result in impaired BM function leading to inhibition of blood cell formation or to malignancy, characterized by uncontrolled proliferation of immature blood cells. Much of this effort has focused on the toxic effects of benzene. Epidemiologists have devoted much time and effort to establishing exactly which clinical entities result from prolonged exposures to substances as diverse as chemotherapeutic alkylating agents in cancer therapy and industrial solvents, principally benzene, in the context of a continuously evolving and extremely complex histopathologic and genetic classification process³ for neoplastic diseases of the hematopoietic and lymphoid systems.

With their increasing success in treating various cancers with cytotoxic drugs, clinicians have increasingly had to confront acute leukemia (AL), especially AML, as a significant and often fatal complication of aggressive cancer chemotherapy. Therapy-associated myeloid neoplasms are currently estimated to account for 10–20 percent; of all cases of AML and MDS.³ Although AML resulting from benzene intoxication has been observed for a much longer time, it is the therapy-associated neoplasms that have provided the clinical source material for application of high-resolution cytogenetics, specific gene sequencing, and immunochemical identification of cell surface markers, all of which form the basis of the modern classification of leukemias and related diseases and provide insights into their pathogenesis and indications for treatment.³ Restrictions on the use of benzene in the workplace in most countries during the 20th century have led to the virtual disappearance of benzene-associated diseases from clinics in most of the world. However the continued use of benzene for many purposes in developing countries, especially in Asia, has continued to produce new cases of benzene-associated hematological disease in those places, including MDS and AML; research continues on these occupational diseases in Asian populations, but at a modest pace.⁴ The modern (2008) WHO classification of diseases of hematopoietic and lymphoid tissues³ has a section devoted to therapy-related myeloid neoplasms. It is not certain that benzene-related myeloid neoplasms are identical on a molecular genetic level to the homologous diseases that complicate anti-tumor therapies or that arise in the absence of prior exposure to any recognized causative agents.

The recognition that a meeting to bring together clinical and laboratory investigators with a common interest in BM function was long overdue led us to form an international organizing committee that set about planning an interdisciplinary program that emphasized the most recent developments in understanding BM and stem cell function and the adverse effects of environmental factors including anticancer therapeutic agents on BM in general and on Husks in particular. Sponsored by the New York Academy of Sciences and Rutgers, The State University of New Jersey, the meeting “The Bone Marrow Niche, Stem Cells, and Leukemia: Impact of Drugs, Chemicals, and the Environment,” was held May 29–31, 2013, at the New York Academy of Sciences in New York City. The meeting, which included individual talks, panel discussions, and a large poster session, was designed to foster interdisciplinary interactions among experimental hematologists, clinicians, and toxicologists with interests in hematopoietic stem cells, the bone marrow environment in which they reside, and the diseases that result from their neoplastic transformation. The conference included an introduction chaired by Helmut Greim and Robert Snyder and sessions on (1) stem cells, the niche, and myeloid neoplasms, chaired by Richard D. Irons and Dorothy A. Sipkins; (2) stem cell signaling and the niche, chaired by David Ross and Emmanuelle Passegué; (3) dysregulation of gene expression in myeloid neoplasms, chaired by Michelle M. Le Beau and Martyn T. Smith; (4) therapy-related myeloid neoplasms, chaired by David Eastmond and Richard A. Larson; and (5) models and tools, chaired by Robert Oostendorp and Lucy A. Godley.

Introducing the conference, Robert Snyder (Rutgers, the State University of New Jersey) presented a concise historical summary of the evolving recognition during the 19th century of leukemia as a distinct kind of disease of the blood, different from infections. In parallel, he sketched the isolation of benzene from coal tar early in the 19th century, the discovery of its powerful solvent effect on natural rubber, and the widespread use of vast amounts of benzene in the developing rubber industry, which employed large numbers of workers who were occupationally exposed to benzene by contact and inhalation. The first report of benzene-induced a plastic anemia in rubber workers appeared in 1897 (Ref. 5), and the identification of AML as another occupational disease of benzene workers followed during subsequent decades.^6,7 Snyder also described the series of international scientific meetings that have been organized at approximately 5-year intervals since 1988 to serve as a focus for laboratory and epidemiological research on the adverse health effects of benzene. The most recent of these,⁸ in 2009, was the first in this series to feature an expanded program that included clinical research on therapy-related myeloid neoplasms as well as occupational studies on benzene. It was also the forum in which many participating toxicologists were first introduced to the concept of the BM niche, and the complex and vital roles it plays in normal hematopoiesis and in the pathogenesis of myeloid neoplasia. Snyder concluded by noting that many other drugs and chemicals have adverse effects on BM in susceptible individuals, and that understanding susceptibility factors could lead to more effective strategies for prevention of both occupational and therapy-related disease.

Stem cells, the niche, and myeloid neoplasms

Normal and neoplastic stem cells

Irving Weissman (Stanford University), opened the session with a discussion of leukemic stem cells (LSCs) and recent advances in defining them.^9–14 As described by Weissman, the key characteristic distinguishing Husks from their daughter cells is their ability to self-renew; that is, when Husks divide, they give rise to Husks through self-renewal and progenitors through differentiation. When regulation of self-renewal goes awry, the genesis of LSCs ensues. LSCs represent the only self-renewing cells in the cancerous tumor. Following the progression from Husks to myelogenous leukemias for decades, Weissman and his colleagues have found that the development of LSCs begins at the stage of Husks. Upon progressing to malignancy, the LSC resembles a downstream oligolineage or multilineage progenitor, with self-renewing capabilities and mechanisms in place to evade programmed cell death and programmed cell removal. While there are many mechanisms for avoiding cell death, Weissman emphasized that there is one dominant method for avoiding programmed cell clearance, a “don't eat me” mechanism that involves the expression of the cell surface protein CD47, the ligand for macrophage SIRPα. Weissman stated that all cancers tested to date express CD47; he also presented data demonstrating that blocking the CD47–SIRPα interaction with antibodies leads to phagocytosis and tumor cell death. In addition, Weissmann highlighted an example where anti-CD47 antibodies led to phagocytosis of AML and depletion of AML in the BM of mice.

Throughout his presentation, Weissman highlighted some of the experimental limitations associated with identifying LSCs and proving that the identified cell, or mutation, is the repository of progression to leukemia. He commented that if we really want to understand what initiates the leukemia it is important that “we ask the leukemia.” To do so requires acquisition of viable frozen cell suspensions of patients with leukemia and access to powerful techniques that support single-cell analysis. It is important to understand the role of the mutations at various stages of disease progression and to be able to distinguish those mutations that are necessary from those that are sufficient, those that initiate the disease from those that are responsible for maintaining it, and those that simply become fixed in the cell population with potentially little consequence. He emphasized that our understanding of leukemia development and progression stands to benefit immensely from focused analysis of those cells that have acquired the stem cell capability for self-renewal.

Myeloproliferative neoplasm development remodels the osteoblastic bone marrow niche and promotes myelofibrosis

Signals from the BM niche in which HSCs reside have been shown to play an integral role in balancing HSC quiescence, proliferation, and differentiation in the maintenance of blood homeostasis.¹⁵ Emmanuelle Passegué (University of California, San Francisco) discussed how leukemic hematopoiesis could affect the BM microenvironment to create a self-reinforcing leukemic niche that promotes leukemia development, while negatively affecting normal HSC function.¹⁶ Using an inducible SCL-tTA∷TRE-BCR/ABL double-transgenic mouse model of human chronic-phase chronic myeloid leukemia (CML)¹⁷ together with in vitro co-culture/imaging approaches, Passegué and colleagues examined the effect of myeloproliferative neoplasia (MPN) development on the endosteal BM niche. She described a process wherein leukemic myeloid cells remodel the endosteal BM niche by stimulating mesenchymal stem cells (MSCs) to proliferate and overproduce functionally altered osteoblastic lineage cell (OBC) derivatives, which accumulate in the BM cavity as inflammatory myelofibrotic cells. Passegué highlighted the importance of cell–cell interactions and close proximity signals between leukemic myeloid cells and expanding MSCs in this process. She described roles for soluble factors in the BM cavity, specifically the important synergistic interaction between thrombopoietin and CCL3 in driving OBC expansion. A role for IL-6, TNF-α, or TGF-β, was excluded, but Passegué emphasized that other still unknown soluble or membrane-bound factors expressed by leukemic myeloid cells are likely to contribute to MSC stimulation. She then described how pathway-directed gene-expression analysis implicated changes in signaling activity as a potential contributor to the remolding of MPN-expanded OBCs into inflammatory myelofibrotic cells, specifically changes to TGF-β, Notch pathways, and inflammatory signaling. The remodeled OBCs were also demonstrated to have altered gene expression of many HSC regulatory genes, which correlated to their reduced ability to maintain normal Husks. Passegué noted that LSC maintenance remained unaffected by these BM niche changes, likely because of the different requirements in adhesion molecules for homing and retention between transformed LSCs and normal Husks. Overall, these results identified additional features of MPN pathogenesis and uncovered how leukemic myeloid cells create a self-reinforcing leukemic niche that promotes MPN development and compromises maintenance of normal hematopoiesis. Passegué concluded her presentation by highlighting the importance of exploiting this interrelationship between leukemic hematopoiesis and the leukemic microenvironment for the development of new therapeutic strategies in the treatment of myeloid malignancies.

Niche targeting of leukemia stem cells

Catriona Jamieson (University of California, San Diego) began by describing her interest in understanding the molecular events involved in initiating self-renewing LSCs, as well as those responsible for mediating therapeutic resistance.^18,19 Cell-cycle changes, aberrant microenvironmental cues from the BM niche, activation of key signaling pathways, and deregulation of RNA processing are all possible mechanisms involved in the malignant reprogramming of LSCs. Jamieson presented recent work on blast-crisis progenitors in CML that adopt more primitive yet deregulated stem cell characteristics and harbor resistance to therapies. She described a process of malignant progenitor transformation and LSC maintenance that occurs in inflammatory niches through a mechanism involving alternative splicing of GSK-3β and BCL2 family gene RNAs. Using lentiviral-enforced expression of BCR-ABL and JAK2 in CD34⁺ cord blood, TNF-α– and STAT-mediated activation of adenosine deaminase RNA associated protein (ADAR1) was observed. Jamieson explained that this was an exciting observation, as the ADAR family of RNA editases has been implicated as having important roles in HSC maintenance, and differential RNA editing has been observed in CML progression. Jamieson presented data in blast-crisis CML humanized mouse models demonstrating that shRNA knockdown of ADAR1 prevented LSC self-renewal while its overexpression resulted in GSK-3β mis-splicing and myeloid progenitor differentiation. This mis-splicing of GSK-3β prevented phosphorylation and degradation of both β-catenin and GLI sonic hedgehog pathway transcriptional activators, with the ultimate outcome of increased LSC self-renewal, survival, and dormancy. These data suggest that ADAR1 activity may contribute to blastic transformation of CML. She highlighted that ADAR-mediated RNA editing occurs primarily in primate-specific Alu repeat sequences, noting that more study is necessary to understand if the activation of RNA editases may be important for species-specific malignant reprogramming of progenitors.

Jamieson concluded her talk with a discussion on selective therapeutic strategies for CML that target regulators of cell survival, self-renewal, or RNA processing pathways in leukemic cells while sparing normal HSC populations. Understanding that aberrant hedgehog gene expression patterns distinguish blast crisis from chronic-phase CML and disease states from normal progenitors, she highlighted an example of potential therapeutic importance involving a sonic hedgehog inhibitor that pushes cells out of the BM and into the peripheral blood, where they undergo terminal proliferation.

Niche and signaling regulation of the state and fate of stem cells

Linheng Li (Stowers Institute for Medical Research) is interested in understanding how the different states of HSCs are regulated, and specifically how underlying signaling regulates their quiescence and activation in different BM niches.^20,21 He explained that while proliferating HSCs play an important role in supporting routine blood production, quiescent HSCs are critical for long-term maintenance and for replenishing the active HSC pool. Quiescent long-term HSCs are primarily located in the endosteum of the bone, where they are directly attached to N-cadherin⁺ stromal cells. Li noted that a role for Wnt signaling in self-renewal and maintenance of HSCs has been well documented; however, the specific involvement of noncanonical Wnt signaling in regulating HSCs has been largely unstudied. Therefore, Li and colleagues set out to understand the expression profiles of the Wnt receptors Frizzleds in HSCs. He described work supporting the notion that noncanonical Wnt signaling maintains quiescent long-term HSCs through Flamingo and Frizzled8 (Fz8) interaction in the endosteal niche. In contrast, another noncanonical Wnt receptor, Frizzled5 (Fz5), was expressed in less quiescent (or primed) HSCs residing in Nestin-GFP+ stromal cells located in the perivascular zone of central marrow. Fz5 was not expressed in H2B-GFP label–retaining quiescent HSCs, in endosteal cells, or in sinusoidal cells. In an Mx1-Cre:Fz5 or Nestin-Cre:Fz5 knockout mouse model, migration of HSCs out of the perivascular zone was observed, but there was no change in HSCs isolated from the endosteum. Functionally hematopoietic reconstitution was only affected in the Fz5⁺ HSCs. Li indicated that these data support a role for Fz5 in maintaining HSCs in the perivascular zone. He proposed that noncanonical Wnt signaling maintains deep-quiescent and primed HSCs residing, respectively, in the endosteal and perivascular zones. In these zones, Fz8 and Fz5 are differentially expressed, withFz8 mediating noncanonical Wnt signaling in the endosteal niche and Fz5 maintaining noncanonical Wnt signaling in the perivascular niche.

Niche-initiated oncogenesis

As highlighted in preceding talks in this session, HSCs live in a complex microenvironment, composed of heterotypic cells important for influencing their behavior. Although it is recognized that the microenvironment in which leukemic cells reside also modulates their behavior, whether or not the microenvironment is involved in the induction of cancer is less clear. David Scadden (Massachusetts General Hospital and Harvard University) began his presentation by emphasizing that investigating the role of the microenvironment in cancer induction presents a challenge in that current approaches are restricted by a limited understanding of the specific cell types comprising the BM stroma. He noted that we are still in the early days of understanding exactly what the niche is, currently are building a better understanding of the different component parts, and have yet to piece together a complete picture of the real influence it can have on HSC function. Scadden and colleagues are working to better define the heterologous cells in the niche microenvironment and understand their influence on the initiation of leukemogenesis.^22,23

Existing research supports the view that specific mesenchymal cells in bone are niche components for HSCs in the BM stroma. By genetically modifying the function of osteolineage cells, Scadden and colleagues were able to disrupt the integrity of the hematopoietic system. Targeted deletion of the miRNA-processing endonuclease Dicer1 from osterix-expressing mesenchymal osteoprogenitor cells led to altered survival, proliferation, and differentiation of HSCs and progenitor cells. These hematopoietic defects recapitulate key features of human MDS and are associated with the emergence of frank leukemia with distinctive secondary genetic abnormalities in HSC and progenitor cells. Scadden highlighted, importantly, that the original deletion of Dicer1 in the mesenchymal cells of the microenvironment was not present in the leukemic cell secondary genetic abnormalities. To investigate whether the leukemic phenotype depended on the stroma, Scadden and colleagues transplanted the leukemic cells and found the leukemia could only engraft in recipients who had the genetically altered osteolineage cells. These findings suggest that the interaction between the cells in the microenvironment and the hematopoietic cells can initiate malignancy and is necessary for its maintenance.

Although still in the early days of understanding the role of the niche, Scadden emphasized that these data make it possible to imagine a niche-based model of oncogenesisin which a multi-step process includes an initiating step in the heterologous cells that make up the stroma which then leads to secondary genetic changes in other cells. This is an exciting postulation, as it challenges the existing premise that the initiating events in cancer are completely, or predominantly, cancer cell autonomous.

Contribution of the reprogrammed vascular niche to stem cell self-renewal and organ regeneration

Shahin Rafii (Weill Cornell Medical College and Howard Hughes Medical Institute) discussed the potential role of the vascular niche in stem cell self-renewal.²⁴ Rafii described how endothelial cells (ECs) are not just passive conduits to deliver oxygen and nutrients; they also provide a vascular niche in which specific angiocrine factors influence HSC self-renewal and differentiation. He highlighted the fact that microvascular ECs within different tissues are known to be distinct structurally, phenotypically, and functionally; however, the molecular signatures and microenvironmental cues that help sustain these tissue-specific properties are poorly understood. Rafii discussed the challenges associated with studying vascular ECs and described new strategies for overcoming these challenges and investigating the vascular niche in vitro.

Using angiogenic models, Rafii and colleagues have been able to demonstrate that EC-derived angiocrine growth factors expressed in unique combinations support in vitro self-renewal and in vivo repopulation of HSCs. Tangential to these findings, Rafii hypothesized a similar role of the vascular niche in leukemia, in which the tumor vascular niche is comprised of a unique population of maladaptively activated ECs that produce defined growth factors which selectively induce growth of leukemogenic cells. Rafii commented that a current limitation of the existing studies is they tend to investigate the role of the vascular niche in a model already harboring a tumor (i.e., final-stage model). In this final stage, an inflammatory, hypoxic environment already exists in the niche, which triggers a proliferation of vessels to improve oxygen supply. Hypoxia also triggers cytokine signaling and gene transcription, both of which influence cell proliferation. This limitation makes it challenging to sort out if the vascular niche contributes to initiation or progression in the leukemogenic process. Nonetheless, it is clear that the tumor microenvironment contributes to the uniqueness of the endothelial microenvironment.

Rafii ended his presentation by discussing the potential for exploiting the interplay between the vascular niche and leukemogenic cells for the development of novel treatments. If an environment exists in which each tumor is vascularized by maladaptive ECs with angiocrine attributes responsible for supporting the expansion of tumor-initiating cells, then selective targeting of these angiocrine factors should diminish the growth of the tumor.

Stem cell signaling and the niche

Niche-secreted factors regulate stem cell behavior

Robert Oostendorp (Technical University of Munich) presented the available information on the function of the BM niche—the surrounding micro-environment of the HSCs—to control the balance between HSC self-renewal and differentiation, and possibly HSC dormancy and proliferation.²⁵ To elucidate the molecular mechanisms involved in this regulation, most investigators—including Oostendorp's group—have begun dissecting these mechanisms using stromal cell lines as in vitro models of the niche; the results obtained thus far are increasingly being validated in vivo.

Factors secreted by the niche, such as secreted frizzled-related protein 1 (Sfrp1) and pleiotrophin (Ptn) play critical roles in maintaining HSC self-renewal and in the balance between myeloid and lymphoid regenerative potential. Loss of Sfrp1 causes slow exhaustion of quiescent HSCs,²⁶ whereas Ptn regulates myeloid/lymphoid engraftment.²⁷ One of the main HSC pathways affected is the Wnt signaling pathway, as well as associated Smad signaling modulators. In particular, the niche seems to have a critical role in maintaining the balance between canonical and non-canonical Wnt signaling in HSCs. Disturbances in the secretion of niche factors may thus deregulate HSC pathways, facilitating disturbances in self-renewal and differentiation. In severe cases, this may lead to malignant transformation. A large amount of gene expression data has been generated by the Oostendorp lab and other investigators, which will facilitate a better understanding of the regulatory network of the niche. However, the study of the functional interactions between different signaling modules identified in vitro and in vivo has only just begun. Considering the role of the niche in maintaining quiescence, the generated insights may aid in developing new tools for early detection of malignant transformation, and may also aid in improving therapies for eradicating therapy-resistant malignant disease caused by the existence of quiescent malignant stem cell–like cells.

The Ah receptor in stem cell cycling, regulation, and quiescence

Thomas A. Gasiewicz (University of Rochester Medical Center) discussed the possible role of the aryl hydrocarbon receptor (AhR) in the etiology and/or progression of certain hematopoietic diseases.²⁸ Since the AhR is deregulated by many toxic chemicals to which humans are exposed, it may be one mechanism by which extrinsic factors affect the regulation of HSCs.

Persistent activation of AhR by the potent agonist dioxin results in altered numbers and function of HSCs in mice. HSCs from AhR knockout (KO) mice are hyperproliferative from an altered cell cycle. In addition, aging AhR KO mice show characteristics consistent with premature BM exhaustion and are prone to hematopoietic disease development. Furthermore, the AhR gene appears to be regulated under conditions that control HSC proliferation. Specifically, AhR activation by TCDD alters HSC/progenitor cell function—which affects HSC migration and trafficking and leads to compromised HSC function—and induces expression of a series of genes that are involved in sending niche-derived signals. In AhR KO mice, excessive HSC cycling and loss of quiescence is observed.

These data and others present evidence for a function of the AhR in HSC regulation. Dr. Gasiewicz proposes that the increased proliferation of HSCs lacking AhR expression or activity is a result of loss of quiescence, and, as such, AhR normally acts as a negative regulator to curb excessive or unnecessary proliferation of HSCs. Similarly, prolonged and/or inappropriate stimulation of AhR activity may compromise the ability of HSCs to sense environmental signals that allow these cells to balance quiescence, proliferation, migration, and differentiation. These data also support a hypothesis that deregulation of AhR function has an important role in the etiology and/or progression of certain hematopoietic diseases, many of which are associated with aging.

Stress-induced activation of hematopoietic stem cells in vivo

Marieke A. G. Essers (Heidelberg Institute for Stem Cell Technology and Experimental Medicine and German Cancer Research Center), like Dr. Oostendorp, stressed that tissue stem cells are ubiquitous in the mammalian organism and are responsible for the maintenance and repair of most organs and tissues. In the hierarchically organized blood system, HSCs with lifelong self-renewal capacity are at the top of the hierarchy, giving rise to active HSCs that typically control blood cell production during healthy homeostasis. However, under stress conditions such as viral infections or after blood loss, where large amounts of mature blood cells are lost, feedback signals are thought to signal back to the dormant HSCs, leading to their activation and production of new mature blood cells. The molecular and cellular mechanisms, including which cytokines are a part of these feedback loops, remain largely unexplored.

When studying the activation of HSCs under stress conditions, the group observed that the cytokine IFN-α is able to activate the entire HSC pool, including dormant HSCs.²⁹ Activated HSCs start to proliferate in vivo and upregulate stem cell antigen 1 (Sca-1). Since IFN-α is produced in virally infected immune cells to block the infection of more mature blood cells, it may have a role in the regulation of HSCs. To further elaborate this effect on HSCs, Esser's group explored whether, during infections, IFN-α might be part of a feedback loop leading to the activation of HSCs. Using a reporter mouse to monitor IFN-α production in the BM, several forms of BM stress were tested. Injection of mice with lipopolysaccharide (LPS), which triggers inflammatory reactions, led to increased IFN-α production, followed by a TLR4-dependent activation of quiescent HSCs. Similar to IFN-α, LPS-induced activation was accompanied by and dependent upon upregulation of Sca-1 on the surface of HSCs. Whereas IFN-α has a direct effect on HSCs, both in vivo and in vitro experiments showed that LPS acts indirectly on HSCs, possibly by triggering TNF-α activation. This indirect mechanism of HSC activation in response to LPS and the potential role the BM niche plays in this process are presently being investigated.

Aldehyde dehydrogenases in normal and malignant hematopoietic stem cells

Clayton Smith (University of Colorado, Denver) presented evidence that the aldehyde Dehydrogenase (ALDH) gene family, which consists of at least 19 members, may play a role in regulating HSC cell-fate decisions. ALDH1A1 is highly expressed in quiescent stem cells, compared to more committed progenitors.³⁰ These enzymes metabolize reactive oxygen species (ROS) and reactive aldehydes (RAld). Data from Smith's group and others suggest that the intracellular ROS and RAld provide fine-level control over a variety of normal cellular processes or cause protein and DNA damage leading to cellular dysfunction. Since ALDHs are expressed at high levels in HSCs, metabolism of RAld may play an important role in cellular homeostasis.

Loss of two ALDH family isoforms in HSCs, ALDH1A1 and ALDH3A1, led to increases in intracellular ROS and RAld and widespread perturbations in cell signaling, gene expression, and cell cycle progression, as well as a predisposition to leukemia formation.^31,32 Approximately one-third of human leukemias also failed to express ALDH1A1 and ALDH3A1 and were exquisitely sensitive to toxic ALDH substrates. These observations set the stage for future studies designed to better understand the role of ROS and RAld in normal and malignant stem cells and to develop new therapies for AML, MDS, and other disorders.

Niche regulation for hematopoietic stem cells

Toshio Suda (School of Medicine, Keio University) began by noting that in the microenvironment of the stem cell niche, adult HSCs are kept quiescent, which is thought to be a characteristic property for the maintenance of HSCs. Normal HSCs maintain intracellular hypoxia, stabilize the hypoxia-inducible factor-1α (HIF-1α) protein, and generate ATP by anaerobic metabolism.³³ In HIF-1α deficiency, HSCs become metabolically aerobic and overcome cell cycle quiescence, and finally become exhausted. Accordingly, an increased presence of HIF-1α protein in VHL-mutated HSCs and their progenitors induces cell cycle quiescence and accumulation of HSCs in the BM. Restored glycolysis by pyruvate dehydrogenase kinases ameliorates cell cycle quiescence and stem cell capacity. Angioprotein 1 (Angpt1), which triggers vascularization, and by that increases oxygen supply, increased stem–stem division. This clearly indicates that the HIF-1α–dependent hypoxic microenvironment is one factor that maintains HSC quiescence. On the basis of the physiological nature of HSCs, the functional alterations in HSCs and niches in hematological malignancies, multiple myeloma, and CML were discussed. Suda concluded that comparison between normal and abnormal HSCs and niches will be important to development of new treatments for cancer.

Apoptosis-related gene expression profiling of hematopoietic stem/progenitor cells after radiation exposure

In order to clarify the effects of irradiation on the mechanisms of aging, Yoko Hirabayashi (National Institute of Health Sciences, Japan) exposed mice to single whole-body irradiation and investigated the response of HSCs/HPCs.^34,35 As a result, the lineage-negative, c-kit–positive, stem cell antigen–positive (LKS) fraction did not recover in 2Gy whole-body irradiated mice. It remained at levels approximately 50–80% of those in age-matched non-irradiated controls until 18 months of age. The expression of genes, specifically those related to apoptosis, was quantified by real-time PCR analysis. Gene expression was investigated in BM cells and cells in the LKS fraction from 21-month-old mice with or without radiation exposure at six weeks of age in comparison to two-month-old non-irradiated control mice. In mice more than 1 year after radiation exposure, five out of 11 selected genes showed significantly altered expression patterns. Whereas most genes are downregulated, Ccnd1, Fyn, and Pik3r1 were upregulated in the LKS fraction after radiation, compared to the fraction without radiation exposure. Moreover, in the LKS fraction there was an increased number of cells, increased cell cycling, and prolonged oxidative stress. These findings indicate a possible prolonged proliferation of cells in the LKS fraction after single-dose irradiation. Interestingly, increased expression of Ccnd1 was observed, specifically in mice with radiation exposure, in addition to the upregulation of the gene in the LKS fraction of aged mice. These findings indicate that the aging phenotype may be enhanced by radiation exposure.

Endothelial cells expressing constitutively active AKT1 combined with mesenchymal stem cells are capable of reconstituting bone marrow niche in vitro

Irina Meln (Almazov Federal Heart, Blood and Endocrinology Centre discussed the equilibrium between new blood cell formation and HSC and HPC quiescence in the BM, which depends on the crosstalk between ECs, mesenchymal stem/stromal cells (MSCs) and HSCs/HPCs. In order to study the mechanisms of interaction between those three cell types and the molecular signals that regulate the fate of HSCs/HPCs, the group reconstituted the BM niche in vitro by co-culturing HSCs/HPCs with ECs and MSCs. She described three particular research directions: generation of an in vitro model of the vascular niche by co-culturing HSCs/HPCs with ECs expressing constitutively active AKT1; establishing an in vitro model of the osteogenic niche by co-culturing HSCs/HPCs with stromal cells containing constitutively active full-length Notch ligands; and studying the crosstalk between ECs and stromal cells from different sources.

The specific purpose of the vascular part of the project is to establish EC cultures that are viable in serum as well as in endothelial cytokine-free media that is tailored for HSC/HPC maintenance. It has been shown that the activation of AKT1 in ECs increases their viability and has effects on HSC/HPC expansion.³⁶ EC cultures that overexpress a constitutively active (myristylated) form of AKT1 (AKT1-ECs) have significantly higher levels of Akt1 mRNA expression and protein accumulation compared to the basal levels. She noted they have characterized endothelial marker expression in AKT1-ECs and their viability in HSC/HPC media. Activation of AKT1 improves the viability of ECs in HSC/HPC media and supports the expansion of CD34⁺ CD38⁺ HSCs/HPCs in vitro.

There is a well-known paradigm that suggests that MSCs have the ability to support cell growth owing to their release of trophic factors.³⁷ To increase the viability of ECs in co-culture systems, the group decided to add MSCs to the co-culturing system. In the next part of the project, they identified what type of stromal cells support the viability of ECs at the highest level. Using conditioned media (CM) from BM-derived MSCs and subcutaneous adipose tissue–derived stem cells (qASCs), they showed that CM significantly stimulates the proliferative activity of ECs compared to unconditioned media. Additionally, they demonstrated that BM-derived MSC-CM enhances proliferation of ECs better than sqASC-CM. To investigate the factors that could be involved in vasculotrophic action of stromal cells, the group analyzed angiogenic factors that were secreted into the media. Some factors that play essential roles in angiogenesis, including angiopoetin 2, follistatin, hepatocyte growth factor (HGF), and vascular endothelial growth factor (VEGF), are increased significantly in BM-derived MSC-CM compared to sqASC-CM and unconditioned media. This suggests that using BM-derived MSCs for co-culture systems would support ECs. However the use of BM-derived MSCs could have disadvantages such as high variation in properties between donors, low accessibility, and relatively low proliferative potential in vitro. As an alternative source of MSCs, the group used a more available material—umbilical cord (UC)–derived MSCs. Unfortunately, the ability of UC-derived MSCs to support HSCs/HPCs was rather low when compared to BM-derived MSCs. To overcome such a flaw, UC-derived MSCs were transduced with full-length Notch ligands (DLL1 and JAG). After 2 weeks of culturing UC-derived MSCs with HSCs/HPCs, UC-derived MSCs modified with DLL1 demonstrated a two-fold increase in their ability to support CD34⁺ expansion. Additionally, UCMSCs transduced with DLL1 were the only type of cells leading to accumulation of CD34⁺CD38^- cells, which is a less differentiated population of HSCs/HPCs than CD34⁺CD38⁺.

Reconstructing the human hematopoietic niche: opportunities for studying normal and malignant hematopoiesis

Richard W. J. Groen (University Medical Center Utrecht and the Dana-Farber Cancer Institute), like the preceding speaker, reported on efforts to develop a humanized model, which more closely resembles the human niche than the currently available conventional xenografts or genetically engineered murine models.³⁸ The goal is to investigate essential interactions within the human BM microenvironment to better understand the development of normal and malignant hematopoiesis and to evaluate tools for therapeutic interference. In this humanized model, a biocompatible scaffold seeded with human BM stromal cells allows the establishment of a local microenvironment resembling human bone. Inoculation of these mice with human CD34⁺ progenitor cells resulted in homing to the human bone environment and the generation of hematopoietic cells. The functionality of these humanized niches was extended to primary samples obtained from patients with several malignancies (e.g., multiple myeloma (MM), AML, other leukemias, MDS) that are highly dependent on the BM microenvironment for survival and expansion. In addition, gene-marking MM and T cell acute lymphoblastic leukemia cells with luciferase enabled whole body bioluminescence imaging to follow tumor outgrowth in time. Importantly, in this model, the response of primary MM cells to established anti-MM agents correlates with clinical responses in the respective patients.

Dysregulation of gene expression in myeloid neoplasms

Epigenetic mechanisms and therapeutics

Lucy A. Godley (University of Chicago) discussed the process of epigenetic modification of DNA and its influence on gene expression. Methylation of DNA results in a decrease in expression of the genes encoded in that region. Conversely, when the base cytosine is metabolized to 5-methylcytosine, and then is altered to 5-hydroxymethylcytosine, the expression of genes in that region of DNA is increased. This latter modification can be repaired by cellular enzymes, bringing gene expression back to normal levels. Godley explored the effect on methylation and hydroxymethylation of mutations commonly found in myeloid malignancies: DNMT3A (encodes a DNA methyltransferases), TET2 (encodes an enzyme that converts 5-methylcytosine to 5-hydroxymethylcytosine), IDH1/2 (encodes metabolic enzymes that produce an oncometabolite, 2-hydroxyglutarate, which inhibits α-ketoglutarate–dependent enzymes such as TET2 and the jumonji histone demethylases), EZH2 (encodes a catalytic subunit of a histone methylase), and ASXL1 (encodes another modulator of histone methylation). She noted that throughout erythropoiesis, hydroxymethylcytosine decreases and certain active DNA regions become hydroxymethylated.³⁹ Patients with myeloid malignancies have many mutations that possibly disrupt the balance of methylation/hydroxymethylation, which may play a pivotal role in the formation of hematopoietic malignancies. Understanding this process better may lead to strategies for clinically effective drugs to slow or reverse this process.

Gene expression in myelodysplastic syndromes and acute myeloid leukemia

Stephen Nimer (University of Miami) presented evidence that both genetic and epigenetic abnormalities underlie myeloid neoplasms. By examining gene expression in patients, insight may be gained toward diagnosis and classification, prognosis, identification of pathogenic mechanisms, and the choice of therapy. Examples for AML patients were presented, including the AML M7 pediatric cohort, which includes patients with a fusion protein (CBFA2T3–GLIS2) that leads to self-renewal and activates BMP signaling; these patients have a poor prognosis.⁴⁰ In contrast, AML patients with the t(8;21) AML1–ETO fusion protein, associated with the M2 FAB subtype of AML, appear to activate self-renewal genes, and these patients have a better-than-average prognosis.⁴¹ By understanding the recurring genetic modifications and their effects on gene expression leading to cancer, researchers can strive to develop treatments that target the altered gene expression and thus treat the leukemia. However, the complex upregulation and downregulation in gene expression in these cancer patients requires an understanding (and subsequent control) of this process on a gene-specific basis.

Role of mutations in epigenetic regulators in pathogenesis of myeloid malignancies

Ross L. Levine (Memorial Sloan Kettering Cancer Center) discussed mutational profiling of genotypes in sub-types of AML, myelodysplasia, and MPNs to allow new therapeutic and mechanistic insight.⁴² Mutations may be of prognostic relevance and may improve prediction of risk in AML patients. The observed genotypes in leukemia patients are complex, suggesting that combinations of mutations including those in TET2, ASXL1, DNMT3A, and EZH2 may allow segregation of patients into subgroups with prognostic relevance. Insight into the functional consequences of these mutations, such as TET2 mutations leading to increased methylation and decreased demethylation, impaired hematopoietic differentiation, increased stem cell self-renewal, and myeloid transformation, may provide insight into the mechanism of hematopoiesis.

MicroRNAs in myeloid leukemia

Guido Marcucci (Wexner Medical Center at The Ohio State University) discussed microRNA and its possible role in growth control of both normal and malignant cells. Unlike coding RNAs (messenger RNAs, or mRNAs), microRNAs do not encode proteins, but rather bind mRNAs to regulate their stability and translation into proteins. A single microRNA can influence hundreds of mRNAs, and thus has a strong influence on homeostasis. Loss and gains in microRNAs have been found to be associated with various types of cancer, including myeloid leukemias. This talk presented evidence for microRNAs as prognosticators, treatment-response predictors, molecular targets and novel therapeutic compounds for AML.⁴³ Using the example of mir-181a, mice provided with synthetic miR181a had a lower tumor burden and higher survival. The possible treatment of AML patients with a liposomal preparation and a specific ligand was raised as a possible means to allow target delivery to the cells.

Application of genome-wide profiling to evaluate effects of benzene and its metabolites from yeast to humans

Luoping Zhang (University of California) discussed functional profiling of susceptibility genes using a cell-survival bioassay and high-throughput screening in yeast to identify novel human susceptibility genes and pathways. Upon probing of functional classes of genes whose expression was modified by exposure, DNA damage and repair pathways appear to be altered following exposure to a series of benzene metabolites.⁴⁴ Similarly, expression of genes in DNA repair pathways were again found to be affected in yeast cells following exposure to formaldehyde. Parallel studies in a near-haploid human cell line derived from a CML patient showed alterations in expression of a number of genes; however the functions of these genes had not been identified and the functions of these genes need to be tested using techniques such as RNAi knock-out in wild-type KBM7 or other cells. Zhang concluded that functional toxicogenomics can be used to explore the mechanisms of chemically induced malignancies.⁴⁵ However, it is important to note that the bioavailability of these chemicals at critical target sites in people may not be similar to that of cells treated in isolation. Therefore, the susceptible genes selected from these functional screens need to be confirmed in human studies.

Benzene cytogenetics, myelodysplasia, and acute myeloid leukemia: new insights into a disease continuum

Richard D. Irons (Fudan University) reported results from hospital-based clinical epidemiology studies in Shanghai, China that call into question long-standing assumptions regarding the mechanism of benzene leukemogenesis. Detailed characterization of structural and molecular cytogenetic lesions was performed in over 1350 subjects prospectively diagnosed with MDS or AML.⁴⁶ A total of 80/649 cases of MDS and 78/722 cases of AML were determined to have some benzene exposure, with 29 MDS and 54 AML cases having experienced moderate to heavy chronic occupational exposure to the solvent.

Common wisdom holds that MDS and AML following benzene exposure are similar, if not identical, to therapy-related MDS/AML. However, Irons and colleagues observed that the pattern of cytogenetic abnormalities in benzene-exposed cases more closely resembled de novo disease.⁴⁷ When benzene-exposed AML cases in the Shanghai series were compared against four other large series of therapy-related AML, deletions of chromosomes 5 or 7, which are common in therapy-related AML, were significantly decreased (P = 0.0003 and 0.0011, respectively), while recurring abnormalities involving the balanced translocations, t(8;21) or t(15;17), were increased in benzene exposed cases (P = 0.0512 and 0.0000, respectively). Further, aneuploidy was not elevated in benzene-exposed AML cases relative to de novo cases in the Shanghai series.

Benzene myelotoxicity is known to be dependent on cell-specific bioactivation by myeloperoxidase, which is uniquely expressed in committed myeloid progenitor cells (MPCs). However, this is inconsistent with direct targeting of HSCs, which are the proximate cell of origin for AML with t(8;21) or t(15;17). In light of immune-mediated inflammatory changes in BM that occur in benzenemyelotoxicity, these investigators propose an indirect mechanism of leukemogenesis in which disruption of extrinsic signaling by MPCs in the BM niche leads to HSC dysfunction and BM failure or leukemic transformation.

Robert Hromas (University of Florida) discussed the role of chromosomal translocations in the malignancy process. The emergence of AML in cancer patients whose primary malignancy was treated with high-dose radiation or chemotherapy is a devastating clinical issue. These secondary leukemias are characterized by chromosomal translocations, and sequencing across the translocation junctions has suggested that non-homologous end-joining (NHEJ) DNA repair is responsible for many of these translocations. There are two pathways of NHEJ: a classical pathway (cNHEJ) involving the Ku complex, which appears to repress translocations; and an alternative pathway (aNHEJ) that involves CtIP and DNA ligase III, along with the regulatory role of PARP1, which appears to be required for chromosomal translocations.⁴⁸

Hromas and colleagues proposed that inhibition of PARP1, the rate-limiting step of aNHEJ, would be an appropriate target to inhibit the occurrence of chromosomal translocations. PARP1 inhibitors are currently being tested in clinical trials and were utilized by Hromas and colleagues. Using model systems as well as cell lines treated with high-dose radiation or a clinically relevant chemotherapy, Hromas and colleagues showed that modulation of PARP1 activity decreased the occurrence of chromosomal translocations.

The presenter suggested that cancer patients might benefit from screening procedures that identify those with high PARP1 expression or activity. This would allow for identifying those patients most at risk for the occurrence of a secondary cancer after high-risk treatments, and either those patients could be monitored long term for the emergence of AML, or they might benefit from treatment with a PARP1 inhibitor such as olaparib to prevent translocations leading to subsequent cancers.

Therapy-related myeloid neoplasms

Genetics of therapy-related myelodysplastic syndromes and acute myeloid leukemia

Mette K. Andersen (University of Copenhagen) described the genetics of therapy-related MDS and AML.⁴⁹ The majority of these patients has acquired cytogenetic and molecular characteristics of diagnostic and prognostic importance. A subset of cases, approximately 10–20%, has a history of chemotherapy, radiotherapy, or combined modality therapy for a prior malignant or non-malignant disease. Molecular, cytogenetic, and clinical studies of such cases of therapy-related myeloid neoplasms (t-MN) are important for a couple of reasons; they represent the most serious long-term complication of cancer therapy and they offer insight into the etiology of MDS and AML, since in many cases they are directly related to previous cytotoxic exposure. Alkylating agents characteristically induce therapy-related MDS with chromosome 5 and 7 abnormalities, as well as complex karyotypes, 2–7 years after therapy, whereas topoisomerase II inhibitors more often induce overt therapy-related AML with recurrent balanced translocations often within 24 months of initial therapy.

In order to gain further insight into leukemogenesis, Andersen and colleagues have screened 140 t-MN patients (the Copenhagen series) for mutations in 17 genes involved in tumor suppression, differentiation, signal transduction, and epigenetic regulation. The results were correlated with type of previous therapy, cytogenetic aberrations, and clinical characteristics. Mutation of the TP53 gene was the most common genetic abnormality in their cohort and was observed in 34% of patients. It was significantly related to previous therapy with alkylating agents, older age, and a poor prognosis. Cytogenetically, TP53 mutations were associated with a complex karyotype and deletion or loss of chromosomes 5 and 17, but were not associated with chromosome 7 abnormalities. Among cases with a complex karyotype, 72% had TP53 mutations, which is similar to what has been observed in de novo MDS/AML with a complex karyotype. In contrast, mutations of the RUNX1 gene were observed in 16% of patients and were significantly associated with deletion or loss of chromosome 7, but were unrelated to chromosome 5 abnormalities, underscoring that t-MN with chromosome 5 and 7 abnormalities are different at the molecular genetic level. Mutations of tyrosine kinase genes (FLT3, KIT, JAK2), genes in the BRAF/RAS pathway (NRAS/KRAS, BRAF, PTPN11), and NPM1 were each observed in 1–10% of cases. Mutations of FLT3 and NPM1 were associated with presentation as overt AML with a normal karyotype. Finally, mutations of IDH1/2 were observed in 9% of t-MN and were associated with a normal karyotype and with der(1;7), but were inversely correlated with other chromosome 7 aberrations.⁵⁰ So far, der(1;7) and other chromosome 7 abnormalities have been regarded as similar cytogenetic entities, because they all result in loss of 7q material. However, these results suggest that they are biologically different.

In the Copenhagen series, a total of 143 gene mutations including fusion genes were detected in 140 patients with t-MN; 36% had no detectable mutations, 37% had one mutation, 25% had two mutations (most often activating mutations of signal transduction cooperating with mutations resulting in impaired differentiation), and 1% had three mutations. Recently, other studies have identified mutations in epigenetic regulator genes (TET2, DNMT3A, ASXL, EZH2) and in splicing genes (SRSF2, SF3B1), primarily in t-MN patients with a normal karyotype.

In conclusion, these studies show that in t-MN, different genetic pathways can be defined by their specific cytogenetic and molecular aberrations. However, many patients with t-MN still have unidentified genetic abnormalities. Further studies are needed to understand the pathogenesis of therapy-related disease, to identify patients at risk, and to ultimately be able to prevent this life-threatening complication of chemotherapy.

Genetic pathways leading to alkylating agent–induced therapy-related myeloid neoplasms

Michelle M. Le Beau (University of Chicago Comprehensive Cancer Center) began by describing the karyotypes of 380 patients with t-MN studied at the University of Chicago. Most cases were characterized by the loss of genetic material. The most common subtype of t-MN (∼75%) develops after exposure to alkylating agents. This subtype is characterized by deletion of the long arm of chromosome 5 (del(5q)) and/or loss or deletion of chromosome 7 (−7/del(7q)), and arises in a HSC/HPC. Chromosome 5 abnormalities are often associated with complex karyotypes, chromosomal instability, and loss or mutation of the TP53 gene on 17p. Genome-wide studies by Downing and Nakitandwe indicate, on average, 7.2-copy-number changes per case of t-MN (deletions/amplifications =1.7/1). Cases with del(5q) had on average 14.1 abnormalities compared to only 2.5 lesions per case among those with −7/del(7q).

The long arm of chromosome 5 is a rich locus of genes involved in hematopoiesis. The most commonly deleted segment is 5q14–q33. Two minimally deleted segments of 5q have been identified—the segment within 5q31.2 is associated with de novo AML and t-MNs, whereas the second is more distal and spans 5q33.1, includes RPS14, and is associated with MDS with an isolated del(5q).

Current studies support a haploinsufficiency model, in which loss of a single allele of more than one gene on 5q contributes to the development of myeloid neoplasms. Using mouse models, Le Beau and co-workers previously showed that haploinsufficiency of Egr1 (5q31.2) or Apc (5q22) independently recapitulates some features of human MDS. EGR1 is a transcriptional regulator of TP53, CDKN1A (p21), PTEN, and many other genes, and plays a role in maintaining the quiescence of HSCs and their retention in the BM niche. Egr1−/− mice treated with ethylnitrosourea (ENU) develop a myeloproliferative disorder characterized by ineffective erythropoiesis. APC is part of the Wnt signaling cascade that regulates mitosis and cell migration. APC is essential for HSC maintenance and survival; haploinsufficiency impairs hematopoiesis. TP53 at chromosome 17p13.1 mediates cell cycle arrest, DNA repair and apoptosis, and maintenance of stem cell self-renewal. Loss of TP53 activity is significantly associated with t-MN with a del(5q).

Le Beau discussed how Apc haploinsufficient mice develop severe macrocytic anemia caused by a block in erythropoiesis; dysplasia and monocytosis are prominent. Mice with Apc^del/+ have increased erythroidcolony-forming units in the spleen. The block in erythroid differentiation is not due to increased apoptosis. Loss of one allele of Egr1 or Tp53, or both, equally accelerates Apc^del/+-induced disease. Transplantation of BM cells from Apc^del/+ mice with macrocytic anemia does not result in disease in the recipients, raising the possibility that the disease is cell extrinsic. The investigators found that transplantation of either wild-type, Egr1^+/−, Tp53^+/−, or Apc^del/+ BM cells into Apc^del/+ recipients results in anemia, whereas control recipients remain healthy. Thus, Apc^del/+-induced disease is induced by the Apc haploinsufficient marrow microenvironment, implicating aberrant Wnt signaling in the BM microenvironment in the pathogenesis of macrocytic anemia.

Mice have now been generated expressing a single allele each of Egr1 and Apc. Le Beau described data demonstrating that Egr1 and Apc haploinsufficiency cooperate in the development of a fatal macrocytic anemia with a block in erythroid development at the late basophilic erythroblast stage.⁵¹ Inactivation of TP53 may be required for progression to AML in the setting of a del(5q). Mice transplanted with Egr1⁺⁻, Apc^del/+ BM cells expressing a Tp53-specific shRNA (∼90% knockdown) develop an aggressive AML in about 15% of animals. Altered Wnt signaling can recapitulate a macrocytic anemia with monocytosis, and a loss of Tp53 in the context of Egr1 and Apc haploinsufficiency promotes leukemia development in mice. Patients with deletions of chromosome 5 have downregulation of specific Wnt signaling genes (negative regulators) and upregulation of other Wnt signaling genes (positive regulators).

These results suggest that EGR1 and APC haploinsufficiency cooperate in the development of myeloid disorders, and that further mutations, such as that achieved by complete inactivation of TP53, are required for progression to AML. Le Beau concluded with a model proposing that MDS and t-MN represent diseases of the tissue, essentially involving both the microenvironment and hematopoietic cells. In t-MN, alkylating agents may act on both HSCs and the marrow stroma, leading to abnormal Wnt signaling in the marrow stroma, and creating a permissive environment for the acquisition of cooperating mutations in HSCs, such as a del(5q). Ultimately, cytopenias and dysplasia emerge with loss of TP53 in the context of EGR1 and APC haploinsufficiency; decreased apoptosis and AML result.

Topoisomerase II and leukemia

Neil Osheroff (Vanderbilt University School of Medicine) reviewed the critical biochemistry of type II topoisomerases and their role as essential enzymes that remove knots, tangles, and torsional stress from DNA.⁵² These enzymes function by generating transient double-stranded breaks in the genetic material. To maintain genomic integrity while the DNA is cleaved, topoisomerases form a covalent phosphotyrosyl bond between the newly generated DNA termini and an active-site tyrosyl residue, generating the cleavage complex. Humans encode two isoforms, topoisomerase IIα and IIβ. Beyond their essential physiological functions, type II topoisomerases are targets for several widely prescribed anticancer drugs, including etoposide, doxorubicin, and mitoxantrone. These drugs kill cells by stabilizing topoisomerase II–generated DNA strand breaks, thereby preventing replication.

Although topoisomerase II–targeted drugs play pivotal roles in treating numerous human cancers, these agents are associated with the generation of specific leukemias in ∼ 2–3% of patients. There is strong evidence that the chromosomal breaks mediated by human type II topoisomerases generate translocation breakpoints. Etoposide treatment is associated with therapy-related myeloid leukemia (t-AML), which features translocations in the MLL (mixed lineage leukemia) gene at chromosomal band 11q23. Furthermore, treatment with mitoxantrone, another topoisomerase II inhibitor, has been implicated in the generation of acute promyelocytic leukemia and translocations involving the PML and RARA genes.

Genetic evidence implicates a role for the quinone metabolite of etoposide in the generation of t(11q23)-associated AML. Etoposide quinone induces higher levels of topoisomerase II–mediated DNA breaks than does the parent compound and appears to function by an alternative mechanism that involves covalent adduction of the enzyme. 1,4-benzoquinone is also known to bind covalently to topoisomerase II to poison the enzyme. Infant AML that includes chromosome 11q23 rearrangements is associated with maternal diets (during pregnancy) that are high in naturally occurring topoisomerase II–active agents. Soy and green tea bioflavonoids as well as curcumin (the flavor component of turmeric) all display activity against human type II topoisomerases. Thus, etoposide-related t-AML may be a model for certain environmentally induced leukemias.

Familial myelodysplastic syndrome/acute myeloid leukemia and germline RUNX1 mutations

Jane E. Churpek (University of Chicago) described several of the known genetic cancer predisposition syndromes and emphasized that an inherited susceptibility to MDS or AL is likely to be more common than previously appreciated.⁵³ The use of next-generation sequencing technologies is allowing the identification of germline variants involved in MDS and AL susceptibility. Merging these methods with detailed family histories provides a unique opportunity to understand the multi-step process of leukemogenesis.

A number of monogenic disorders with predisposition to MDS/AL are already known, and some of these also predispose to solid tumors. These syndromes provide insight into the involved pathways, which appear quite diverse. These include inherited BM-failure syndromes such as Fanconi anemia, Down syndrome, telomere disorders, Shwachman-Diamond and Diamond-Blackfan syndromes, Li Fraumeni syndrome, and familial AML with mutated CEBPA or mutated GATA2. These account for perhaps 1% of sporadic cases of AML. Patients with familial platelet disorder (FPD) with propensity to myeloid malignancy often present with a bleeding disorder; 40% subsequently develop MDS/AML. FPD is caused by a monoallelic mutation in RUNX1. Of note, RUNX1 is somatically mutated in about 10–15% of de novo AML cases. Thus, careful collaborative studies offer the potential to understand the mechanisms of leukemogenesis, both in patients with inherited MDS/AL syndromes and also in the much larger number with sporadic disease.

The genetics of therapy-induced second cancer risk

Kenan Onel (University of Chicago) reported on the use of genome-wide association studies (GWAS) in cancer.⁵⁴ Although these studies have succeeded in identifying numerous single nucleotide polymorphisms (SNPs) associated with cancer risk, the contribution to risk of these variants is small, rendering them of little use clinically. Perhaps a third of cases of t-MN may result from germline cancer-predisposing mutations rather than exposure to therapy. Whereas GWAS primarily focus on main effects, most complex diseases are also influenced by myriad environmental risk factors. These risk factors expose genetic liabilities to disease, and thus the influence on disease of a specific SNP is often context dependent. Therefore, Onel and colleagues hypothesized that a variant identified in a GWAS that has only a small effect overall may actually exert a large effect in a specific patient subset sharing a common environmental context. To test this, he performed GWAS on two different therapy-induced cancers: radiation therapy–induced second cancers after pediatric Hodgkin lymphoma and therapy-related AML.

The risk of second malignant neoplasms in children with lymphoma is associated with radiation exposure and correlated positively with dose and inversely with age. The most common second cancers are breast and thyroid carcinomas. In collaboration with the Childhood Cancer Survivor Study, Onel analyzed 96 cases in the discovery set and 82 controls (i.e., children treated for Hodgkin lymphoma who did not develop secondary cancer). After genotyping roughly 600,000 SNPs in each genome, two SNPs were significant: one on chromosome 6 (PRDM1) and one on chromosome 18 (ATG5).⁵⁵ He went on to show that transcript levels were lower in those with SNPs in PRDM1 and ATG5, and the protein level of PRDM1 was also lower. PRDM1 protein is induced in the cells with both SNPs (the risk background) but not in the cases with neither SNP (protective haplotype). Thus, there are differences in cellular response depending on the presence of the SNPs. For example, slower and less significant responses of MYC were observed in the context of the risk haplotype than in the protective haplotype. They concluded that PRDM1 is a radiation-responsive tumor suppressor gene.

For the t-MN study, Onel performed a GWAS on 78 cases from the University of Chicago and 148 population controls. However, the top 10 candidates did not reach statistical significance, and none of the top hits replicated in a second set of 70 t-MN cases provided by collaborators at Washington University. However, the discovery cases and the replication controls had different frequencies of chromosomal abnormalities, reflecting different exposures. When evaluating only cases with abnormalities of chromosomes 5 and 7, the SNPs did reach statistical significance in both the discovery cases and the replication cases. Therefore, by conditioning on exposures common to cases and controls in GWAS, one can reduce genetic noise due to exposure heterogeneity and enhance power even as the sample size is reduced. Similar GWAS were done in a larger cohort of de novo AML, but again, no SNPs reached genome-wide significance. These data suggested that common variants can have large effects in the context of specific etiological exposures and that in studies of cancer, the presence of shared recurrent somatic mutations could be surrogates for a shared mechanism of carcinogenesis. Thus, future studies incorporating exposures into genomic investigations of complex diseases may reveal patient subsets for which specific SNPs contribute meaningfully to disease.

Investigation into the crosstalk between acute myeloid leukemia cells and the bone marrow microenvironment

Ashley Hamilton (Cancer Research U.K., London Research Institute) described elements of the complex interplay between cell-intrinsic and cell-extrinsic regulatory signals generated by the BM microenvironment that influence the cell fate of HSCs to either self-renew or differentiate. As has been highlighted many times throughout the meeting, a balance exists within the crosstalk between HSCs and the niche, which allows HSC dormancy, activation, and differentiation. Alterations of this balance may lead to uncontrolled cellular proliferation and ultimately to the promotion of leukemia.⁵⁶ However, it remains to be determined exactly how the hematopoietic microenvironment contributes to the deregulation of normal hematopoiesis and/or promotes the maintenance of leukemia cells as a leukemic niche. Hamilton and colleagues performed microarray analysis of MS5 stromal cells that were co-cultured with a panel of leukemia cell lines as well as patient samples of AML. The most significantly upregulated pathways, as compared to MS5 cells cultured alone, involved cytoskeleton remodeling, cell cycle, cell adhesion, and development through cytokine signaling. Since a number of effectors of the TGF-β signaling pathway were upregulated in the stroma co-cultured with leukemic cells, they then investigated inhibition of this pathway using a specific inhibitor of TGF-β receptor kinase, SB-431542. Treatment with the inhibitor significantly increased the levels of apoptosis in the AML cells co-cultured with stromal cells, but had no effect on normal HSCs, and also reduced the level of AML cell engraftment in mice on treatment in vivo. These data highlight the potential for the development of leukemia therapies directed at modifying the BM microenvironment.

Functional analysis of the bone marrow microenvironment in myelodysplastic syndrome: targeting the disease niche

Ruben A. Ferrer (University Hospital Carl Gustav Carus at Dresden University of Technology) discussed the controversy regarding the contribution of the BM microenvironment in MDS.⁵⁷ Multipotent mesenchymal stromal cells (MSC) and monocyte/osteoclasts (MON-OC) regulate the hematopoietic niche, and their functionality is different in the various MDS disease stages. MSCs themselves appear dysplastic in most of the cases. Ferrer and colleagues studied the properties of primary cells from patients with lower-risk MDS, with and without del(5q), or higher-risk MDS. Compared to age-matched healthy controls, MDS-MSC showed impaired clonal potential and growth with distinct differentiation defects, particular expression of adhesion and cell surface molecules important for intercellular communication, and altered secretion of niche cytokines, among them lower stromal-derived factor 1α (SDS1) levels and KIT-1 expression. Healthy CD34⁺ HSC and HPCs migrated less towards MDS-MSC culture supernatants, produced reduced numbers of cobblestone area–forming cells and fewer colony-forming units compared to controls when cultured with MDS stroma. The hematopoietic cells presented increased proliferation when co-cultured with high-risk MDS-MSC. Stroma from high-risk MDS provoked increased apoptosis of leukemia cells compared to stroma from low-risk MDS and healthy controls. They also demonstrated that the numbers of marrow MON-OC progenitors were reduced and the ability of CD14⁺ cells to generate terminally differentiated bone reabsorbing osteoclasts was impaired. MDS progenitors in the presence of high-risk MDS-stroma showed higher osteoclast differentiation. Thus, the BM microenvironment is involved in the pathophysiology of MDS and may present a potential target for therapeutic approaches.

Models and tools

Regulation of leukemia cell dormancy by the bone marrow niche

Dorothy A. Sipkins (University of Chicago) discussed the characteristics of the microenvironment facilitating leukemia metastasis to the BM and the signaling pathways favoring homing and survival of LSCs at the expense of normal HPCs. Innovative imaging systems were described that facilitate real-time visualization of leukemia cell trafficking to the niche in mouse xenograft models.⁵⁸ Models using ALL cells in mice showed that cells migrate to specific areas of the BM niche and that CXCL12 or SDF-1 is a major signal regulating homing to the niche. The subsequent growth of leukemic cells within BM niches relies on hijacking of normal HPC signaling pathways. In addition, leukemic cells remodel the normal BM microenvironment at the expense of normal HPCs by overexpressing niche regulatory molecules including SCF.⁵⁹ SCF overproduction leads to loss of normal HPCs in the BM over time. Osteopontin (OPN) is another key signaling molecule produced by the normal HPC niche that is abnormally secreted by both blasts and stroma in leukemic BM. OPN's normal function is to maintain HSCs in a quiescent state. Recent work has shown that it also co-localizes with quiescent leukemic cells in primary human samples, suggesting that OPN promotes leukemic cell dormancy. Dormant leukemic cells residing in the niche exit the cell cycle and are more difficult to target with drugs and may also precipitate disease relapse. In a mouse model of ALL, blocking leukemia cell adhesion to OPN forced ALL cells into active phases of the cell cycle and rendered them more sensitive to chemotherapy. Targeting signaling molecules in the niche such as OPN offers an opportunity to control leukemia cell behaviors such as migration and proliferation so that they can be better targeted with cytotoxic therapies.⁶⁰

Methods to analyze homing of stem cells in none marrow

Susie K. Nilsson (CSIRO) presented novel techniques to analyze homing of cells within the marrow.⁶¹ Isolation methodology is important to accurately depict niche populations, and interestingly, commonly used marrow-flushing techniques only capture approximately 70% of HSCs, failing to remove those residing in the endosteal region. Isolation and characterization of endosteal HSCs using mechanical dissociation, enzymatic digestion, and the same panel of cell surface markers demonstrated that cells in this region of the BM have a significantly higher hematopoietic potential than their centrally located counterparts.⁶² Immunofluorescent tracking of cells allows easy identification of cells, quantification of the number that home to the marrow post-transplant, as well as the location of individual cells within the marrow microenvironment. Consequently, these techniques allow calculation of the homing efficiency as well as the spatial distribution of HSCs. A key aspect of stem cell homing is their interaction with the vasculature, followed by trans-marrow migration and, finally, lodgment into the niche. Sinusoidal ECs are an important BM cell type since they pose the initial barrier for entry into the marrow, and all BM cells are within four cells of the vasculature.⁶³ The scavenger capability of ECs was utilized to devise a method, involving ingestion of colloids and fluorescent labels, to functionally label vascular cells. She described that cells from various parts of the BM tend to migrate back to that specific area; endosteal cells will preferentially migrate back to the endosteum and, similarly, cells isolated from the central marrow will home to the central region. This suggests that cell surface markers on specific subtypes of cells within the niche are critical in determining the homing site of transplanted cells.

Modeling exposure-induced leukemogenesis in the mouse

Michael J. Thirman (University of Chicago) discussed MLL fusion genes and their role in conferring a growth advantage to leukemic cells in mouse systems.⁶⁴ Translocations involving 11q23 involve more than 50 different loci within the genome. The critical consequence of these translocations is the formation of MLL fusion genes that have the potential to drive leukemic transformation of hematopoietic cells. The mechanisms underlying the generation of these MLL fusions were also investigated, and a propensity for breaks within an 8.3kb breakpoint cluster region could reflect topoisomerase II binding sites or regions of scaffold attachment. The fusion of MLL to ELL occurs in one of the more frequent 11q23 translocations observed in patients with AML. Two alternative strategies to develop mouse models of MLL-ELL leukemia were investigated. After retroviral infection and transplantation, all mice developed AL. However, a knock-in model of MLL-ELL leukemia did not result in AL, indicating that the Mll-Ell fusion gene is insufficient by itself to trigger leukemia and demonstrating the importance of a second hit to develop leukemia in this system. However, treatment of the Mll-Ell knock-in mice with the carcinogen ENU or ionizing radiation induced AL similar to that observed in human AML cases that exhibit 11q23 translocations. The Mll-Ell knock-in mice could therefore be utilized as a screening system to evaluate the leukemic potential of environmental exposures.

Redox proteomics to measure oxidative stress

Dean P Jones (Emory University) described new redox proteomic methods to measure oxidative stress.⁶⁵ Oxidative stress can cause both covalent and non-covalent modifications of cysteines with peroxides, often generating reversible modifications and reactive aldehydes and radicals causing irreversible modifications. It is well known that reduced and oxidized forms of glutathione, thioredoxin, and glutaredoxin participate as redox sensors, but there are approximately 200,000 cysteines in the mammalian proteome and approximately 80% of these are conserved. Redox-sensitive cysteine residues provide a means to control cellular function without altering underlying mechanisms, and redox-sensitive cysteines in proteins are dynamically responsive.⁶⁶ Metabolic systems are organized in a hierarchical network structure involving ionic signaling, redox signaling, and kinase signaling, which can be involved in crosstalk. High-resolution mass spectrometry can assess the state of the redox proteome in a cell, and when coupled with genomics and metabolomics can give a global picture of system response. Redox signaling is important to stem cell behavior in the niche, and these techniques may offer new insights into niche function.

Long-term quantitative single-cell imaging: new tools for old questions

Timm Schroeder (Helmholz Center Munich) discussed quantitative single-cell imaging techniques and their application to long-established questions in hematopoiesis, as well as their future potential.⁶⁷ Data based on average cell readouts is difficult to interpret, since readouts are not specific to a single cell and population-based measurements are not informative. In a dynamic system such as stem cell behavior in the niche, continuous quantitative single-cell imaging is required to track a cell's past, present, and future.⁶⁸ New tracking and continuous imaging tools were developed in cell culture to enable this analysis, and cells could be monitored using fluorescent tags over multiple generations and for long periods of time. Methods were also developed to monitor the level and activity of individual proteins in single cells across multiple generations.⁶⁹ These new techniques allow tracking of stem cell evolution, definition of cellular events occurring in switching of cell lineages, and testing of long-standing models of hematopoietic development.Time-lapse imaging and single-cell tracking, for example, have demonstrated that ECs can be hemogenic and give rise to other blood cells, which has been difficult to address using conventional techniques. The technical and data-handling challenges are considerable, but use of single-cell monitoring methods will be important to a comprehensive understanding of stem cell behavior.

Genome–exposome interactions in leukemia etiology

Martyn T. Smith (University of California) discussed the exposome⁷⁰ and genome–exposome interactions in leukemia etiology. The need for a comprehensive analysis of the causes of leukemia is illustrated by the fact that of the 13,000 new cases of AML each year, we can explain perhaps only 15% in terms of the causative agent ⁷¹ To define the cause of the majority of de novo leukemias, a non-targeted approach is needed. The exposome represents all environmental exposures, including those during development, and incorporates lifestyle and other exposures over a lifetime. Novel phenotyping methods have been developed to perform these analyses. When coupled with genomic, epigenomic, and microbiomic analysis, the exposome, an untargeted and agnostic approach, will assist in generating an integrated picture of disease causality and investigating the specific role of environmental exposures in disease.⁷²

Future directions

As a rule, human diseases have been described in increasingly greater detail as concepts and methods in pathology have developed. The etiology of traumatic injuries due to accidents or battle was readily understood early in human history. Only after the roles of various organs or specific cells in the body were appreciated was it possible to understand the more subtle relationships between the presence of pathological agents such as bacteria or viruses to the biological responses they provoked, and the ultimate expression of clinical disease. Exposure to chemicals in the environments in which we live and work may also cause tissue damage, resulting in toxic or carcinogenic responses. Carcinogenesis bioassays have been developed to identify cancer-causing agents, in which animals are dosed for most of their lifetimes with the highest-tolerated doses of specific chemicals and data are recorded on whether tumors are formed, what types of tumors, and how many of each kind. This empirical procedure has saved many lives by identifying carcinogens, with subsequent reduction of human exposures to these agents. It has become clear that if we are to thoroughly understand neoplastic diseases we must evaluate the mechanisms by which these chemicals damage cells as well as the sequence of events in tissues that the chemicals initiate. Indeed, exposure of the target organ to these chemicals can often be used as a tool to decipher the mechanisms by which they cause tissue damage.

As toxicology has matured as a biological science, methods have been developed that permit detailed descriptions of the mechanisms by which carcinogenic events are induced and have thereby enabled a comparison of various chemical structures and the likelihood that they can cause cancer. Thus, various procedures have been developed to examine the roles of chemicals in cancer development. These include pharmacokinetics, metabolic activation to reactive metabolites, intracellular interactions and localization, effects of genotoxic agents on progenitor cells to induce chromosomal aberrations (CA) or micronuclei (MN), etc. These methods have often proved valuable when the underlying morphology and cell biology of the target organs are well established. Carcinogenesis and mutagenesis in the BM has long presented difficult problems to overcome because, although it was known that the processes of cellular differentiation and proliferation were the prime functions of the BM, it was not realized that these processes were related to specific locations within the marrow. There appear to be distinct differences in BM niches responsible for the day-to-day stem cell control of normal blood cell production and those places where quiescent stem cells provide a reserve capacity that may be called upon when needed to supplement or enhance initiation of new cell production. This conference featured investigators who have made significant progress in identifying, characterizing, and localizing stem cell niches. It also attracted a number of toxicologists interested in the mechanisms by which the functions of these niches can be influenced by environmental chemicals.

The involvement of the BM microenvironment in the induction, promotion, and progression of hematopoietic malignancy or the induction of MN or CA as indicators of genotoxicity has not received a great deal of attention in toxicology research. Granted, toxicologists are well aware of the importance of cellular interactions, but the possibility that cancer initiation can occur in a cell type which is distinct from that which actually gives rise to the cancer has, for the most part, been overlooked. Perhaps this is one reason why the mechanism of benzene-mediated carcinogenesis has remained elusive. The concepts and technologies highlighted throughout this workshop could have profound implications for furthering our understanding of hematopoietic toxicity. However, this requires toxicologists to incorporate them into experimental designs and consider them during data interpretation. For example, when interpreting standardized in vivo genotoxicity assays, it is important to consider that the micronuclei or chromosomal aberrations that are scored may be the result of either a direct interaction of the genotoxic agent with HSCs/HPCs or a secondary event resulting from a direct interaction of a chemical with the cells of the microenvironment. This distinction has obvious implications for understanding the mechanism of action, and also great implications for evaluating the risk posed by chemical exposure.

This conference succeeded in emphasizing the need to further pursue the location and roles of various BM niches in normal and aberrant hematopoiesis. It suggested that the technologies that have led to the identification of the niches could be used to study the roles of environmental chemicals and drugs in leukemogenesis. Perhaps most importantly, it provided an environment in which hematologists, clinicians, and toxicologists could engage in dialogues which can result in collaborative research aimed at describing the entire process of leukemia development, from environmental exposures to the clinical expression of specific neoplastic diseases. It is our hope that joint research efforts will emerge based on these conversations, and that a robust body of data and concepts can be developed over the next several years that can be exploited to advance the understanding of how the BM responds to toxic agents, culminating in a future need for a subsequent conference on the niche, stem cells, and leukemia.

Figure 3.

Model for the leukemic bone marrow (BM) niche. During myeloproliferative neoplasia (MPN) development, the endosteal BM niche is remodeled into a self-reinforcing leukemic niche that promotes leukemia development while negatively affecting normal hematopoietic stem cell (HSC) maintenance. Leukemic myeloid cells stimulate multipotent stromal cells (MSCs) to proliferate and overproduce functionally altered osteoblastic lineage cells (OBCs), which accumulate as inflammatory myelofibrotic cells in the BM cavity. These MPN-expanded OBCs are compromised in their ability to maintain normal HSCs and promote MPN development by favoring myeloid differentiation. Reprinted from Figure courtesy of Emanuelle Passegué.

Figure 4.

Adverse outcome in AML patients with mutations in specific epigenetic modifiers. Reprinted from Ref. 73.

Figure 5.

Haploinsufficiency of Egr1 and Apc in a mouse model, together with loss of Tp53, promotes development of macrocytic anemia and acute myeloid leukemia. A similar sequence of genetic events resulting from DNA damage to the bone marrow microenvironment and/or HSCs due to alkylating agents and/or radiation therapy may lead to therapy-related myeloid neoplasms in humans. Figure courtesy of Michelle M. Le Beau

Figure 6. Topoisomerase II: Life and Death.

Topoisomerase II is a critical enzyme for normal cell growth. Stabilization of the topoisomerase II–DNA cleavage complex by anticancer drugs creates double-strand DNA breaks and leads to cell death. However, these same DNA breaks can generate translocation breakpoints and are implicated in the generation of human acute leukemia. Figure courtesy of Neil Osheroff.

Figure 7.

Pushing the cancer stem cell (CSC) out of hiding. If tumor cell dormancy versus proliferation can be controlled, the effect of cytotoxic chemotherapies can be optimized. Reprinted from Ref. 60.