Oncogene (2011) 30, 1009–1019 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc REVIEW therapy directed against human acute myeloid leukemia stem cells

R Majeti

Division of Hematology, Department of Internal Medicine, Cancer Center, and Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Palo Alto, CA, USA

Accumulating evidence indicates that many human this model (Kelly et al., 2007; Quintana et al., 2008), cancers are organized as a cellular hierarchy initiated there is strong evidence for the existence of cancer stem and maintained by self-renewing cancer stem cells. This cells in several human malignancies, including acute cancer stem cell model has been most conclusively myeloid leukemia (AML) (Bonnet and Dick, 1997; established for human acute myeloid leukemia (AML), Dick, 2008). Within a tumor, it is only these cancer stem although controversies still exist regarding the identity cells that are able to transplant the disease and of human AML stem cells (leukemia stem cell (LSC)). differentiate into the heterogeneous cells composing A major implication of this model is that, in order to the tumor. This model postulates that one reason eradicate the cancer and cure the patient, the cancer stem for the relative ineffectiveness of current cancer che- cells must be eliminated. Monoclonal antibodies have motherapy regimens is that these therapies target the emerged as effective targeted therapies for the treatment rapidly proliferating tumor cells, which do not include of a number of human malignancies and, given their target all of the cancer stem cells (Reya et al., 2001). Thus, specificity and generally minimal toxicity, are well tumors often shrink in response to chemotherapy, but positioned as cancer stem cell-targeting therapies. One almost universally recur due to sparing of some cancer strategy for the development of monoclonal antibodies stem cells. A key implication of this cancer stem cell targeting human AML stem cells involves first identifying model is that in order to eradicate the tumor and cure cell surface preferentially expressed on AML the patient, therapies must target and eliminate the LSC compared with normal hematopoietic stem cells. cancer stem cells. In recent years, a number of such antigens have been Monoclonal antibodies have emerged as a potent and identified, including CD123, CD44, CLL-1, CD96, CD47, effective molecularly targeted therapy for human cancer, CD32, and CD25. Moreover, monoclonal antibodies usually in combination with chemotherapy (Adams and targeting CD44, CD123, and CD47 have demonstrated Weiner, 2005; Finn, 2008). Such antibodies have been efficacy against AML LSC in demonstrated to be effective in the treatment of a models. Hopefully, these antibodies will ultimately prove number of human cancers, including rituximab for to be effective in the treatment of human AML. B-cell non-Hodgkin lymphoma, trastuzumab for HER2- Oncogene (2011) 30, 1009–1019; doi:10.1038/onc.2010.511; positive breast carcinomas, cetuximab for colorectal published online 15 November 2010 carcinomas and head and neck squamous cell carcino- mas, alemtuzumab for chronic lymphocytic leukemia Keywords: acute myeloid leukemia; cancer stem cells; (CLL) and T-cell lymphoma, and others (Adams and monoclonal antibodies Weiner, 2005; Finn, 2008). Antibodies effective against cancer are believed to function by several mechanisms including, antibody-dependent cellular cytotoxicity, stimulation of complement-dependent cytotoxicity, Introduction inhibition of signal transduction, or direct induction of (Cheson and Leonard, 2008). In some cases, A growing body of evidence indicates that many human an individual therapeutic antibody may function cancers are organized as cellular hierarchies initiated through several of these mechanisms simultaneously. and maintained by a small pool of self-renewing cancer Because of their target antigen specificity and mechan- stem cells (Reya et al., 2001; Jordan et al., 2006; Dalerba isms of action, monoclonal antibodies are able to deliver et al., 2007). Although some controversy still surrounds their therapeutic effects with minimal toxicity. Given the need to develop agents that can target and eliminate Correspondence: Dr R Majeti, Division of Hematology, Department cancer stem cells, and the antigen specificity and efficacy of Internal Medicine, Cancer Center, Stanford Institute for Stem Cell of monoclonal antibodies, recent investigation has Biology and Regenerative Medicine, Stanford University School of focused on the development of monoclonal antibodies Medicine, 265 Campus Drive, G3021B, Stanford, CA 94305, USA. E-mail: [email protected] targeting cancer stem cell antigens. This review will Received 11 May 2010; revised 26 August 2010; accepted 29 September focus on the pre-clinical investigation of monoclonal 2010; published online 15 November 2010 antibodies targeting human AML stem cells. Antibody therapy for AML LSC R Majeti 1010 Human AML stem cells that the Lin–CD34 þ CD38ÀCD90À fraction of human cord blood contains a non-HSC multipotent progenitor, Human AML is a clonal malignancy characterized by and hypothesize that this multipotent progenitor is the the accumulation of immature myeloid cells defective in cell of origin for human AML (Majeti et al., 2007), their maturation and function. Much is now known although it is possible that loss of CD90 expression about the biology of AML, including chromosomal occurs as a consequence of leukemic transformation of translocations resulting in fusions, chromosomal HSC. Similar to the normal hematopoietic differentia- and subchromosomal deletions of putative tumor tion pathway, these LSC in turn give rise to leukemia suppressor , dysregulation of programmed cell progenitor cells with the surface immunophenotype of death, single-gene mutations correlating with leukemia, CD34 þ CD38 þ , which further differentiate into the and derangements of the telomere/telomerase system bulk leukemic blast population defined as CD34À. (Gilliland et al., 2004). A model has been proposed AML affects 13,000 adults annually in the United that AML results from the combination of mutations States, most of them over the age of 65 years (Low- that fall into two broad categories: (1) stimulators of enberg et al., 1999; Estey and Dohner, 2006). Current proliferation (for example, Ras, FLT3 and c-) and (2) standard of care includes treatment with several cycles inhibitors of differentiation (for example, CBF, CEBPA of high-dose chemotherapy and often includes allo- and MLL) (Gilliland and Griffin, 2002; Gilliland et al., geneic hematopoietic cell transplantation. Even with 2004). In order for AML to develop, cells must acquire these aggressive treatments, which have significant side at least one mutation from each category; however, in effects, 5-year overall survival is between 30–40%, and most cases of human AML, the molecular identity of much lower for those over age of 65 years (Lowenberg such mutations is unknown. et al., 1999; Estey and Dohner, 2006). AML is a Recent experiments have added to our fundamental heterogeneous disorder with several different patholo- knowledge by demonstrating that AML, similar to gical classification schemes, including the French– normal hematopoiesis, is composed of a hierarchy of American–British and the more recent World Health cells that is maintained by a small pool of leukemia stem Organization classification (Bennett et al., 1976; Vardi- cells (LSC) (Tan et al., 2006; Dick, 2008). Stem cells in man et al., 2009). The World Health Organization any biological process are defined by the properties of scheme was developed to reflect the molecular under- self-renewal, which is the ability to give rise to progeny standing of several subtypes of AML and to take into identical to the parent through cell division, and account the prognostic importance of cytogenetic multipotency, which is the ability to give rise to all the abnormalities. Immunophenotyping by flow cytometry differentiated cells of that biological process (Reya et al., is routinely performed on clinical specimens and shows 2001). Most of the cells within AML do not possess that most, but not all, human AML samples are these properties, which are restricted to the infrequent predominantly CD34 positive (Vardiman et al., 2009). LSC that are able to maintain the leukemia. Both In published reports assaying a variety of subtypes of in vitro and in vivo experimental approaches using CD34-positive AML samples, LSC have been demon- fluorescence-activated cell sorting-purified subpopula- strated to reside in the Lin–CD34 þ CD38À subpopula- tions of leukemia cells were utilized to identify human tion of cells. The clinical significance of the cancer stem AML LSC. Culture of AML subpopulations on bone cell model for AML is suggested by a study that marrow feeder cells identified long-term culture-initiat- correlated the percentage of CD34 þ CD38À LSC at the ing activity in the CD34 þ CD38À leukemia fraction of time of diagnosis with the duration of relapse free most samples assayed (Ailles et al., 1997; Blair et al., survival in 92 AML patients (van Rhenen et al., 2005). 1998). Dick and colleagues were the first to establish Those patients with 43.5% AML LSC had a median xenotransplantation models for human AML, and used relapse free survival of 5.6 months compared to those this assay to identify long-term engrafting AML with o3.5% AML LSC who had a median relapse subpopulations (Lapidot et al., 1994; Bonnet and Dick, free survival of 16.0 months (P ¼ 0.02), suggesting 1997). In published reports assaying a variety of clinical the importance of LSC in the clinical outcome of and biological subtypes of AML, LSC were found to be human AML. negative for expression of lineage markers (Lin–), positive for expression of CD34, and negative for New controversies over the identification of human expression of CD38 (Bonnet and Dick, 1997; Blair AML LSC et al., 1998; Ailles et al., 1999; Levis et al., 2005). The initial identification of human AML LSC was In normal human hematopoietic development, this accomplished through xenotransplantation of fluores- Lin–CD34 þ CD38À fraction is known to be enriched cence-activated cell sorting-purified subpopulations into for hematopoietic stem cells (HSC), leading to a immunodeficient NOD/SCID mice that lack B and T proposal that AML LSC arise from mutational trans- lymphocytes (Bonnet and Dick, 1997). Although these formation of HSC (Bonnet and Dick, 1997). However, mice exhibited bone marrow engraftment of a subset of human HSC have been shown to be positive for AML samples, the levels of human chimerism were expression of CD90, whereas AML LSC have been often quite low and required transplantation of found to lack expression of this marker (Baum et al., thousands of cells. This prompted the development of 1992; Craig et al., 1993; Murray et al., 1995; Blair et al., more permissive mouse models, most recently the 1997; Miyamoto et al., 2000). We have recently shown NOD/SCID/IL2RgÀ/À (NSG) strain, which is additionally

Oncogene Antibody therapy for AML LSC R Majeti 1011 deficient in natural killer cells (Ishikawa et al., 2005, (Taussig et al., 2010). These results indicate that the 2007; Shultz et al., 2005, 2007). The first AML NSG surface immunophenotype of AML LSC remains an xenotransplantation experiments supported the open question for investigation, particularly with the finding that LSC are exclusively contained in the more permissive NSG xenotransplantation assay. CD34 þ CD38À subpopulation (Ishikawa et al., 2007). In this report, as few as 1000 CD34 þ CD38À AML cells serially transplanted the disease. In contrast, Cell surface antigens preferentially expressed on AML CD34 þ CD38 þ or CD34À cells from either the LSC compared with HSC primary human AML samples or subsequent xenografts failed to engraft, despite the transplantation of greater One strategy for the development of monoclonal than 106 cells. However, recently the concept that all antibodies targeting human AML stem cells, involves AML LSC are contained within the CD34 þ CD38À first identifying cell surface antigens preferentially population has been challenged by the finding that expressed on AML LSC compared with normal many anti-CD38 antibodies, including those utilized in HSC. A number of antigens expressed on AML blasts the original isolation of LSC, inhibit engraftment of have been reported, and monoclonal antibodies CD34 þ CD38 þ leukemia cells through an Fc-depen- targeting several of these antigens are in pre-clinical dent mechanism (Taussig et al., 2008). When this development. However, this discussion will be restricted mechanism was inhibited either by blockade of Fc to those antigens demonstrated to be expressed on receptors with immunoglobulin or through the use of functional LSC defined by expression on engrafting immunosuppressive antibodies, AML repopulating ac- leukemia-initiating cells in xenotransplantation assays tivity was detected in the CD34 þ CD38 þ population. (Table 1). Moreover, CD34 þ CD38 þ cells engrafted in NSG One of the first antigens reported to be preferentially mice, contrary to the results of the first report with these expressed on human AML LSC is the high-affinity mice (Ishikawa et al., 2007). A second report from the interleukin-3 (IL-3) receptor a-chain, also known as same group investigated the surface immunophenotype CD123 (Jordan et al., 2000). By flow cytometry, CD123 of AML LSC from patient samples with nucleophosmin was found to be expressed on nearly all CD34 þ CD38À mutations (NPM1c), and determined that in approxi- cells from 18 primary AML samples, whereas no mately half of the cases, LSC were exclusively contained expression was detected on CD34 þ CD38À cells from in the CD34À fraction, whereas the other half exhibited five normal bone marrow specimens. Transplantation LSC-activity in both the CD34 þ and CD34À fractions assays in NOD/SCID mice found that in three out of

Table 1 Cell surface antigens preferentially expressed on AML LSC compared with HSC Antigen Positive AML Engraftment summary Function Normal tissue expression Comment Reference samples

CD123 18/18 (100%) CD34 þ CD123 þ : 3/3 High-affinity IL-3 Restricted to hemato- Conflicting data on Jordan et al. CD34 þ CD123–: 0/3 receptor (IL3RA) poietic cells of myeloid expression on (2000) lineage normal HSC CD44 8/8 (100%) CD44 þ and CD44À Mediates cell Ubiquitously expressed Cancer stem cell Jin et al. not assessed adhesion and can with many alternatively marker on a number (2006) transduce signals spliced isoforms of solid tumors CLL-1 77/89 (87%) CD34 þ CLL-1 þ : 3/3 Unknown Restricted to hemato- Expression may van Rhenen poietic cells of myeloid identify minimal et al. (2007) lineage residual disease and predict relapse CD96 19/29 (66%) CD96 þ : 4/5 May have a function Resting and activated Expressed on only Hosen et al. CD96–: 1/5 in NK T cells and NK cells, 5% of (2007) and/or activation possibly intestinal CD34 þ CD38– epithelium CD90 þ cells in bone marrow CD47 13/13 (100%) CD34 þ CD38– Binds SIRPa and Widely expressed at Differential expres- Majeti et al. CD47 þ : 3/3 inhibits low levels sion facilitated pro- (2009) spective separation of residual normal HSC from LSC CD32 21/61 (34%) CD32a: CD34 þ CD38– Fc-g receptor 2 Restricted to hemato- Not expressed on Saito et al. CD32 þ engraft CD32b: (FCGR2) poietic cells functional HSC (2010) CD34 þ CD38–CD32 þ no engraftment CD34 þ CD38–CD32À engraft CD25 15/61 (25%) CD34 þ CD38– High-affinity IL-2 Restricted to hemato- Not expressed on Saito et al. CD25 þ : 4/4 receptor (IL2RA) poietic cells functional HSC (2010)

Abbreviations: HSC, ; LSC, leukemia stem cell; NK, natural killer; SIRPa, signal regulatory -a.

Oncogene Antibody therapy for AML LSC R Majeti 1012 three samples investigated, all mice transplanted with expressed in CD34 þ CD38- AML cells that possess a CD34 þ CD123 þ cells engrafted, whereas no mice signal sequence, and are, therefore, likely cell surface transplanted with CD34 þ CD123À cells engrafted. No molecules (Hosen et al., 2007). Analysis of CD96 functional assays were conducted with normal hemato- expression on a panel of human AML samples and poietic cells to validate the lack of expression of CD123 normal bone marrow demonstrated expression on the on HSC. However, a separate report demonstrated that majority of LSC assayed, but only on approximately 5% CD34 þ CD38–CD123 þ cord blood cells were capable of normal HSC. CD96-positive and CD96-negative cells of initiating multilineage hematopoiesis in NOD/SCID were purified by fluorescence-activated cell sorting from mice, suggesting that normal HSC may express CD123 five human AML samples and transplanted into new- (Taussig et al., 2005). Since this first description born immunodeficient mice. In four out of five cases, (Jordan et al., 2000), a number of other investigators only the CD96-positive cells yielded long-term leukemic have reported increased expression of CD123 on engraftment, indicating that AML LSC express CD96. CD34 þ CD38À AML LSC (Testa et al., 2004; Florian Initially, CD96 was identified as a T-cell surface et al., 2006; Yalcintepe et al., 2006). molecule that is highly upregulated upon T-cell activa- The Dick lab reported the identification of increased tion (Wang et al., 1992). CD96 is expressed at low expression of CD44 on engrafting CD34 þ CD38À cells levels on resting T and natural killer cells and is from multiple AML samples compared with normal strongly upregulated upon stimulation in both cell human cord blood (Jin et al., 2006). Furthermore, they types (Wang et al., 1992). It is not expressed on other showed that the CD44-6v isoform was differentially hematopoietic cells, and examination of its expression expressed using an isoform-specific antibody. The pattern showed that it is only otherwise present on percentage of CD34 þ CD38À cells expressing CD44 some intestinal epithelia (Wang et al., 1992; Gramatzki was neither reported nor were engraftment studies et al., 1998). comparing the leukemia-initiating activity of CD44 þ In our own work, we identified increased expression and CD44À subpopulations. In this report, an activat- of CD47 on primary human AML stem cells (Jaiswal ing CD44-specific antibody was demonstrated to eradi- et al., 2009; Majeti et al., 2009). By flow cytometry, cate AML LSC, but not normal cord blood or bone CD47 was more highly expressed on multiple specimens marrow CD34 þ CD38À engraftment (discussed in de- of primary AML compared with normal bone marrow tail in the next section). This result indicates the HSC (Lin–CD34 þ CD38–CD90 þ ), which expressed preferential nature of CD44 expression on AML LSC. low but detectable CD47. Moreover, CD47 expression CD44 is a widely expressed protein with a number of did not discriminate subpopulations within a given alternatively spliced isoforms. The specific isoform(s) AML sample, as the bulk AML blasts expressed similar differentially expressed between AML LSC and normal levels of CD47 as the CD34 þ CD38– fraction. In one HSC, apart from CD44-6v, have neither been reported sample, we determined that the CD34 þ CD38– fraction nor has any functional difference in AML LSC been contained a majority of CD47hi-expressing cells reported for these isoforms. (99%) with a rare CD47lo-expressing population (1%). CLL-1, C-type lectin-like molecule-1, was identified by When this CD47lo-expressing population was isolated a Dutch group of investigators as an AML LSC-specific and functionally assayed by xenotransplantation, it surface molecule (van Rhenen et al., 2007). CLL-1 was engrafted normal multilineage hematopoiesis. Moreover, found to be expressed on a fraction of CD34 þ CD38À this particular AML sample harbored a FLT3-ITD AML cells, ranging from 0–100% with a median of 33% mutation that was not present in the CD47lo-expressing of cells, from 77 out of 89 patient samples evaluated. In population or in the engrafted cells, suggesting that the the majority of the 10 normal bone marrow samples CD34 þ CD38-CD47lo population contained residual evaluated, CLL-1 was not expressed on CD34 þ CD38À normal HSC (Majeti et al., 2009). This prospective sepa- cells. Notably, CLL-1 expression was also neither ration of normal HSC from AML LSC based on detected on CD34 þ CD38À cells from regenerating bone differential expression of CD47 demonstrates the signifi- marrow post chemotherapy nor on granulocyte colony- cance of CD47 as an AML LSC marker. Further in- stimulating factor-mobilized peripheral blood. A total of vestigation of CD47 determined that it is expressed widely three NOD/SCID-engrafting patient samples were identi- at low levels in most tissues. This finding has significant fied, and in all three 100% of the CD34 þ cells were also implications for monoclonal antibody therapies targeting CLL-1 positive. Hence, it was not surprising that in these CD47 and is discussed in more detail below. three cases, isolated CD34 þ CLL-1 þ cells engrafted A recent bioinformatic analysis of genes differentially AML as well. Finally, examination of serial bone marrow expressed between purified CD34 þ CD38– AML cells samples from several patients determined that CLL-1 and normal bone marrow CD34 þ CD38À cells identi- expression on CD34 þ CD38À cells tracked with clinical fied CD32 (FCGR2) and CD25 (IL2RA) as LSC- remission or relapse status. The same group had specific surface molecules (Saito et al., 2010). Examina- previously investigated CLL-1 expression in normal tion of 61 AML cases by flow cytometry determined tissues and found no expression outside the hemato- that CD32 was expressed on 21 (34%) and CD25 was poietic system, with expression restricted to granulocytes expressed on 15 (25%) of these samples. Further and (Bakker et al.,2004). investigation of CD32 expression revealed three patterns CD96 expression on AML LSC was identified using a of expression: CD32-a with a single positive peak, signal sequence trap strategy to identify CD32-b with a bimodal expression showing a minor

Oncogene Antibody therapy for AML LSC R Majeti 1013 CD32-positive population and CD32-c with a single cellular cytotoxicity of primary AML blasts, although negative peak. The relative proportion of patterns efficacy against LSC was not reported (Zhao et al., CD32-a and CD32-b among the 21 CD32-positive 2010). Additional antibodies targeting CD44, CD123 patient samples was not reported. Upon xenotrans- and CD47 have been investigated in pre-clinical models plantation into NSG mice, CD34 þ CD38–CD32 þ cells for the ability to target and eliminate primary AML from CD32-a patient samples engrafted with AML. On LSC (Table 2). the contrary, CD34 þ CD38–CD32 þ cells did not exhibit leukemia-initiating activity from CD32-b patient CD44 samples, which was restricted to the CD34 þ CD38– Monoclonal antibody targeting of AML LSC was first CD32– cells. These data indicate that CD32 is expressed reported with the activating anti-CD44 antibody H90 on AML LSC only from a minority of patient samples. (Jin et al., 2006). Several different treatment models Regarding functional investigation of CD25, the were established in NOD/SCID mice and used to assess authors reported that CD34 þ CD38–CD25 þ AML antibody efficacy. In the first model, antibody treat- cells engrafted in all four recipients evaluated, but did ment was initiated 10 days after transplantation of not mention results of engraftment studies conducted CD34 þ CD38À AML LSC with either control IgG or with CD34 þ CD38–CD25À cells. Notably, CD25 ex- H90 administered three times per week. After 3 weeks, pression has recently been reported to be an adverse mice were sacrificed and analyzed for bone marrow prognostic factor when expressed on greater than 10% AML engraftment, which identified a marked decrease of AML blasts, and its expression correlated with in all four samples reported. In the second model, minimal residual disease (Terwijn et al., 2009). In the antibody treatment was initiated 3–6 weeks after final experiments investigating CD32 and CD25, ex- transplantation in the setting of more established pression of these antigens on normal cells and tissues disease. Treatment starting at 3 weeks post transplanta- was reported (Saito et al., 2010). Expression of both tion resulted in significant elimination of leukemia in antigens was found to be restricted to cells of the two out of two samples; however, initiation of therapy hematopoietic system, although in the case of CD32, a at 4–6 weeks post transplantation did not reduce bone large number of alveolar in the lung were marrow leukemic engraftment in two additional sam- identified with high levels of expression. Functional ples. In contrast, anti-CD44 antibody treatment of mice xenotransplantation experiments demonstrated that engrafted with cord blood or bone marrow CD34 þ HSC function could be isolated in the CD34 þ CD38– cells had no effect on normal hematopoietic engraft- CD32– or CD34 þ CD38ÀCD25À fraction of human ment, indicating the specificity of the antibody effect for cord blood, suggesting that elimination of CD32À or AML. To investigate targeting of the LSC-population in CD25-expressing cells will not target normal HSC. these models, two strategies were employed. First, the percentage and absolute number of CD34 þ CD38À cells remaining in the bone marrow after treatment were Monoclonal antibodies targeting AML LSC-specific determined and found to be significantly reduced with surface antigens H90 treatment compared with control for AML, but not cord blood. Second, bone marrow from control IgG or According to the cancer stem cell model, curative H90-treated mice was assayed for functional LSC by therapies must target and eliminate all cancer stem secondary transplantation; unlike control, no engraft- cells. Monoclonal antibodies are well positioned as ment was detected in mice transplanted with cells from cancer stem cell-targeting therapies given their antigen H90-treated mice. Collectively, these experiments specificity and generally minimal toxicity. In the case of demonstrate that the anti-CD44 antibody H90 is able AML, the anti-CD33 monoclonal antibody conjugate to target and eliminate AML LSC in vivo, but that the gemtuzumab ozogamicin (Mylotarg) was originally effect is limited to situations with a lower burden of given accelerated approval by the FDA for the treat- disease. ment of relapsed AML in elderly patients, in whom it The mechanism by which H90 mediated the elimina- induced some complete remissions (Burnett and Mohite, tion of AML LSC was investigated. This antibody had 2006; Stasi et al., 2008). Moreover, a new report suggests been previously demonstrated to induce the differentia- that CD33 is expressed on a subset of CD34 þ CD38À tion of primary AML blasts in vitro (Charrad et al., LSC that can be killed by Mylotarg in vitro (Jawad 1999). Investigation of its effect on AML in vivo et al., 2010). However, post-approval clinical trials demonstrated that leukemic blasts increased expression demonstrated no benefit from the addition of Mylotarg of markers of mature myeloid cells (CD14, CD15 or to induction or maintenance therapy, no improvement CD11b) in bone marrow from mice treated with H90, in survival outcomes, and increased hepatic toxicity. For suggesting that induction of differentiation also occurs these reasons, in June 2010, Mylotarg was voluntarily in vivo (Jin et al., 2006). Additional experiments removed from the US market and is now only available indicated that cells from H90-treated mice exhibited as an investigational agent in clinical trials. decreased short-term homing to the bone marrow and Recently, novel antibodies targeting several of the spleen in secondary recipients, and also exhibited altered AML LSC-specific antigens outlined above have been LSC fates with localized but not disseminated growth. developed. A monoclonal antibody directed against Thus, it appears the mechanism of H90 action against CLL-1 has been reported to stimulate antibody-dependent AML LSC is multifactorial.

Oncogene Antibody therapy for AML LSC R Majeti 1014 Table 2 Monoclonal antibodies targeting AML LSC-specific surface antigens Antigen Antibody clone Efficacy in treatment Targeting of LSC Effect on normal Postulated Comment Reference model mechanism

CD44 H90 Treatment initiated Decrease in No effect on cord Induces differentia- Limited efficacy Jin et al. (mouse IgG1) 10 days post trans- CD34 þ CD38– cells blood or bone tion of AML LSC in setting of (2006) plantation: decrease in bone marrow after marrow en- and inhibits homing highly engrafted AML in 4/4 samples; treatment with H90. grafted mice to the bone marrow bone marrow. treatment initiated No engraftment in 3–6 weeks post secondary trans- transplantation: plants. decrease in AML in 2/2 (3 weeks) and 0/2 (4–6 weeks) samples CD123 7G3 Treatment initiated Decrease in No effect on Inhibits homing to Limited efficacy Jin et al. (mouse IgG2a) 28 days post CD34 þ CD38– cells bone marrow the bone marrow in established (2009) transplantation: in the bone marrow engrafted mice and can prevent ac- disease treatment decrease AML in after treatment with tivation of human models. May be 2/5 samples 7G3. Reduced en- CD123 by human more effective in graftment, but not IL-3 humans due to complete elimina- lack of mouse tion, in secondary IL-3 activation recipients. of human CD123. CD47 B6H12 Treatment initiated Decrease in CD34 þ No effect on Blocks an inhibitory Limited efficacy Majeti et al. (mouse IgG1) 8–12 weeks post– cells after anti-CD47 in vitro phagocy- CD47-SIRPa inter- in setting of (2009) transplantation: treatment. No tosis of CD34 þ action, thereby en- highly engrafted decrease AML in 3/3 engraftment in normal bone abling phagocytosis bone marrow. samples (8/8 mice) secondary marrow Wide normal with clearance of recipients. progenitors tissue expression the bone marrow a potential in 3/8 mice factor for clinical translation.

Abbreviations: AML, acute myeloid leukemia; IL-3, interleukin-3; LSC, leukemia stem cell.

CD123 bone marrow cells. Secondary transplantation studies The AML LSC-specific antibody furthest along in conducted with two of these samples demonstrated clinical development is an antibody targeting CD123. reduced bulk AML and CD34 þ CD38À engraftment in Studies investigating the efficacy of the anti-CD123 the bone marrow, but not complete elimination. These antibody 7G3 in xenotransplantation models were data suggest that AML LSC were targeted in 7G3- recently reported (Jin et al., 2009). As with CD44, treated mice, but not eradicated. Collectively, these data several different assays were used to investigate target- indicate that in NOD/SCID models 7G3 is able to ing of CD123 in NOD/SCID models. In the first model, inhibit engraftment of AML LSC, but has limited bulk AML or isolated CD34 þ CD38À cells were efficacy against established disease. incubated ex vivo with the 7G3 antibody before Several different mechanisms were investigated for the transplantation into NOD/SCID mice. Such ex vivo activity of 7G3 against AML LSC. Short-term homing incubation markedly reduced the engraftment of 10 out and direct intrafemoral injection studies determined that of 11 primary AML samples, but only 2 out of 5 samples 7G3 markedly inhibited AML LSC homing to the bone of normal bone marrow or cord blood. The reduction in marrow and that this effect was only partially mediated engraftment was sustained in five out of seven samples by immune effector cells (natural killer and/or other assessed at 8–10 weeks post transplantation. LSC CD122-dependent cells). The 7G3 antibody is known targeting was suggested by the finding that, ex vivo to inhibit the ability of IL-3 to bind and activate 7G3 treatment reduced the number of CD34 þ CD38– CD123. In vitro studies demonstrated that 7G3 inhibited AML cells in the bone marrow of xenotransplanted mice spontaneous and IL-3-induced proliferation of CD34 þ from three samples. A second model involved the CD38– AML cells. This proliferation could not be administration of 7G3 to NOD/SCID mice 24 h post assessed in NOD/SCID mice as mouse IL-3 is not able transplantation of AML cells, and resulted in statisti- to activate human CD123. It is possible that in human cally significant reduction in engraftment in only one patients, 7G3 will have a more profound impact on out of three samples. A final established disease model AML LSC than in mouse models due to this additional initiated antibody treatment 28 days post transplanta- mechanism. tion resulting in a significant reduction in bone marrow A human IgG1 chimeric variant of 7G3 has been engraftment in only two out of five mice. No inhibitory generated for clinical development. This chimeric anti- effects were observed in mice transplanted with normal body retains human CD123 binding and neutralization,

Oncogene Antibody therapy for AML LSC R Majeti 1015 but is likely to be less immunogenic when administered CD47 in a CD47-deficient human AML cell line to human patients. Pre-clinical toxicity studies con- inhibited phagocytosis and facilitated engraftment in ducted in cynomolgus monkeys yielded no antibody- immunodeficient mice (Jaiswal et al., 2009). Moreover, related toxic effects, including no abnormalities of the upregulation of CD47 in AML appears to result hematological parameters. This same chimeric variant from co-option of a normal protective mechanism as we has been investigated in a phase I clinical trial of observed that CD47 expression on mouse HSC and relapsed or refractory high-risk AML (ClinicalTrials. progenitors increased upon mobilization from the bone gov ID: NCT00401739), and an interim update reported marrow into the peripheral blood in which these cells no treatment-related hematologic or metabolic toxicity, pass through phagocyte-lined sinusoids in the bone with two mild infusion reactions and one infection marrow, liver, and spleen (Jaiswal et al., 2009). considered possibly related to the treatment (Roberts On the basis of these results, we predicted that et al., 2008). No grade 3–4 adverse events were observed disruption of the CD47–SIRPa interaction with a in the 26 patients enrolled at the time of this report. Of monoclonal antibody directed against CD47 would the eight patients enrolled in the two highest dose preferentially enable the phagocytosis of AML LSC cohorts, one complete response has been observed. (Figure 2). Many monoclonal antibodies have been Clearly, further clinical studies are needed to determine raised against human CD47, including several that block the efficacy of this antibody in larger numbers of AML the CD47–SIRPa interaction and some that bind a patients. different region of CD47 and do not block this interaction (Subramanian et al., 2006). We first inves- tigated the ability of these anti-CD47 antibodies to CD47 enable the phagocytosis of AML LSC in vitro with As detailed above, we identified increased expression of human and mouse effector cells (Majeti CD47 on AML LSC compared with normal HSC from et al., 2009). In both cases, blocking, but not non- multiple primary specimens of AML. CD47 is a widely blocking, anti-CD47 monoclonal antibodies enabled expressed immunoglobulin superfamily member trans- phagocytosis of human AML LSC. Moreover, no membrane protein implicated in multiple cellular phagocytosis was detected with normal bone marrow processes and a number of protein–protein interactions CD34 þ cells, indicating the preferential effect for AML (Brown and Frazier, 2001). In one such interaction, LSC. Using mouse macrophages, we determined that an CD47 serves as the ligand for signal regulatory protein-a anti-mouse SIRPa antibody also enabled phagocytosis (SIRPa), which is expressed on phagocytic cells, of AML LSC, consistent with our proposed mechanism. including macrophages and dendritic cells, that when With these in vitro results demonstrating a potent activated initiates a signal transduction cascade resulting effect of anti-CD47 antibodies against AML LSC, we in the inhibition of phagocytosis (Oldenborg et al., 2000, conducted two different in vivo experimental models to 2001; Blazar et al., 2001; Okazawa et al., 2005; Barclay establish CD47 targeting as a viable therapeutic strategy and Brown, 2006). In this way, CD47 functions as a ‘do (Majeti et al., 2009). First, we utilized an ex vivo coating not eat me’ signal by binding to SIRPa and delivering assay in which AML LSC were incubated with IgG1 a dominant inhibitory signal. This function led us isotype control, anti-CD45, or blocking anti-CD47 to hypothesize that increased expression of CD47 on antibody. Unlike controls, cells coated with anti-CD47 human AML LSC contributes to pathogenesis by antibody failed to engraft when transplanted into NSG inhibiting phagocytosis of these cells through the mice. Second, we established a treatment model in which interaction of CD47 with SIRPa (Figure 1). Consistent NSG mice were first transplanted with AML LSC from with this hypothesis, we found that expression of mouse three independent samples, then 8–12 weeks later once

α α --- SIRP CD47 --- SIRP AML HSC LSC CD47

Pro-Phagocytic Stimulus Ø + Receptor Phagocyte Phagocyte

HSC: No Phagocytosis AML LSC: No Phagocytosis Figure 1 Model of CD47-mediated inhibition of AML LSC phagocytosis. Compared with normal HSC (left), AML LSC (right) express increased levels of CD47 that contribute to pathogenesis by binding phagocyte SIRPa, which delivers a potent inhibitory ‘do not eat me’ signal that predominates over a pro-phagocytic ‘eat me’ stimulus.

Oncogene Antibody therapy for AML LSC R Majeti 1016

Anti-CD47 Anti-CD47

Ø α Ø SIRPα SIRP CD47 AML CD47 HSC LSC

Pro-Phagocytic Stimulus Ø + Receptor

Phagocyte Phagocyte

HSC: No Phagocytosis AML LSC: Phagocytosis Figure 2 Model of a blocking anti-CD47 monoclonal antibody preferentially enabling phagocytosis of AML LSC compared with normal HSC. A blocking monoclonal antibody directed against CD47 disrupts the dominant inhibitory SIRPa signal in phagocytes. This unmasks the pro-phagocytic stimulus on AML LSC resulting in their phagocytic elimination (right), but has no effect on normal HSC as they lack pro-phagocytic stimuli (left).

robust bone marrow engraftment had been established, pro-phagocytic signal. The molecular identification of the mice were treated with either control IgG or a prophagocytic ‘eat me’ signals on AML LSC is an active blocking anti-CD47 antibody. In these mice, anti-CD47 area of our investigation. antibody treatment resulted in the rapid clearance of Our experiments suggest that a blocking anti-CD47 circulating AML in the peripheral blood. Furthermore, antibody will not be toxic to HSC; however, one report eight out of eight mice had marked decreases in the level demonstrated a role for the CD47–SIRPa interaction in of AML in the bone marrow, with three of these eight the regulation of HSC engraftment. These investigators exhibiting complete clearance. Investigation of targeting used positional genetics to characterize the molecular of LSC showed a marked reduction of CD34 þ LSC- basis for superior human HSC engraftment in mice of enriched cells in the bone marrow of anti-CD47-treated the NOD background, and identified a polymorphism in mice, and more significantly, a complete absence of NOD SIRPa that conferred enhanced binding to human engraftment in secondary recipients. These results CD47 (Takenaka et al., 2007). Although this result indicate that in a pre-clinical in vivo treatment model leaves open the possibility that anti-CD47 antibody may with high levels of engraftment, a blocking anti-CD47 also target HSC, it is not clear whether such targeting antibody depleted AML and targeted LSC. would occur in bone marrow stem cell niches in the non- CD47 is widely expressed at low levels in many transplantation setting. Ultimately, the best way to tissues, which could have a significant impact on its determine this is through non-human primate studies, potential as a therapeutic antibody target. To explore and eventually human clinical trials, which we are this issue further, we utilized a rat anti-mouse CD47 actively pursuing. antibody known to block the CD47-SIRPa interaction Treatment of human AML LSC-engrafted mice with in several mouse experiments (Majeti et al., 2009). First, anti-CD47 antibody resulted in undetectable levels of we administered this anti-mouse CD47 antibody to wild- human AML in only three out of eight mice, raising type mice in doses able to coat 100% of the cells in the the important question of how this therapy can be bone marrow. There was no effect on hematopoietic optimized into a curative approach. Review of the eight stem and progenitor cells, but a severe neutropenia did leukemia-engrafted mice treated with anti-CD47 anti- develop. In addition, there was no toxicity to other body suggested that those mice with starting bone organs as determined by metabolic blood parameters marrow engraftment higher than 60% failed to com- and histological evaluation. Second, we used anti-mouse pletely clear human leukemia. This suggests that with CD47 against a mouse model of AML with high CD47 higher levels of engraftment, the bone marrow may not expression. In vitro, this antibody enabled phagocytosis contain sufficient phagocytic activity to completely of mouse AML cells, and in vivo, anti-CD47 treatment eliminate AML. Given this consideration, there are resulted in a survival benefit. These results suggest that several strategies that could potentially improve the regardless of its wide expression, it is possible to target efficacy of anti-CD47 therapy: (1) the overall level of CD47 on AML cells without significant toxicity. leukemic burden could first be reduced by treatment What are the mechanisms responsible for this selective with chemotherapy before administration of anti-CD47 targeting? One possible explanation for this lack of antibody; (2) granulocyte colony-stimulating factor or significant toxicity is that in order for the anti-CD47 granulocyte colony-stimulating factor could antibody to enable phagocytosis of a cell, that cell must be administered to increase endogenous macrophage express a pro-phagocytic ‘eat me’ signal on its surface production and differentiation before administration of (Figure 2). Thus, blockade of the CD47–SIRPa inter- anti-CD47 antibody; (3) granulocyte colony-stimulating action on cancer cells is effective because these cancer factor or CXCR4 antagonists could be used to mobilize cells express an ‘eat me’ signal, whereas normal cells the leukemia cells into the periphery and then treatment are unaffected because they lack expression of such a with anti-CD47 antibody would facilitate phagocytic

Oncogene Antibody therapy for AML LSC R Majeti 1017 have tested this hypothesis in human non-Hodgkin Anti-CD47 lymphoma, in which we observed the synergistic Ø SIRPα eradication of established non-Hodgkin lymphoma xenografts with combination anti-CD47 and rituximab CD47 AML treatment (Chao et al., 2010). We are actively working Anti-LSC LSC +++ FcR LSC Antigen to develop a similar combination antibody therapy targeting human AML LSC. Pro-Phagocytic + Receptor Stimulus

Phagocyte Summary

The cancer stem cell model for AML proposes that in AML LSC: Potent Phagocytosis order to achieve cure, all LSC must be eliminated. Figure 3 Model of potential synergy between a blocking anti- Monoclonal antibodies targeting AML LSC-specific CD47 monoclonal antibody and a second AML LSC-targeting surface antigens represent a potentially powerful strat- antibody. The combination of a blocking anti-CD47 antibody egy to target and eliminate LSC. In recent years, a with an Fc receptor-activating anti-AML LSC antibody should number of antigens preferentially expressed on AML synergize in stimulating the phagocytosis and elimination of AML LSC, by simultaneously blocking the ‘do not eat me’ signal (anti- LSC compared with normal HSC have been identified, CD47) and delivering a potent ‘eat me’ signal (Fc receptor- including CD123, CD44, CLL-1, CD96, CD47, CD32 activating antibody). and CD25. Moreover, monoclonal antibodies targeting CD44, CD123 and CD47 have demonstrated efficacy against AML LSC in xenotransplantation models. elimination of leukemia cells by the more numerous Hopefully, these antibodies will ultimately prove to be macrophages of the spleen and liver. We are currently effective in the treatment of human AML. investigating these approaches for optimizing anti-CD47 antibody therapy. Ultimately, it is unlikely that a single antibody will be Conflict of interest curative for AML, as has been shown by experience with currently approved anti-cancer antibodies (Adams and The author declares no conflict of interest. Weiner, 2005; Finn, 2008). It is more likely that curative therapies will incorporate a combination of antibodies targeting several different antigens simultaneously. In this regard, anti-CD47 antibodies offer the possibility of Acknowledgements a novel mechanism of antibody synergy in the treatment of AML. Specifically, the combination of a blocking I would like to acknowledge Mark Chao and Max Jan for anti-CD47 antibody with an Fc receptor-activating critical review of the manuscript and Irv Weissman for support and mentorship. This work was supported in part by a anti-AML LSC antibody should synergize in stimulating grant from the American Association for Cancer Research. the phagocytosis and elimination of AML LSC, by RM holds a Career Award for Medical Scientists from the simultaneously blocking the ‘do not eat me’ signal Burroughs Wellcome Fund. RM has filed the US Patent (anti-CD47) and delivering a potent ‘eat me’ signal Application Serial No. 12/321,215 entitled ‘Methods For (Fc receptor-activating antibody) (Figure 3). In fact, we Manipulating Phagocytosis Mediated by CD47’.

References

Adams GP, Weiner LM. (2005). Monoclonal antibody therapy of Baum CM, Weissman IL, Tsukamoto AS, Buckle AM, Peault B. cancer. Nat Biotechnol 23: 1147–1157. (1992). Isolation of a candidate human hematopoietic stem-cell Ailles LE, Gerhard B, Hogge DE. (1997). Detection and population. Proc Natl Acad Sci USA 89: 2804–2808. characterization of primitive malignant and normal progenitors Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, in patients with acute myelogenous leukemia using long-term Gralnick HR et al. (1976). Proposals for the classification of the coculture with supportive feeder layers and cytokines. Blood 90: acute leukaemias. French-American-British (FAB) co-operative 2555–2564. group. Br J Haematol 33: 451–458. Ailles LE, Gerhard B, Kawagoe H, Hogge DE. (1999). Growth Blair A, Hogge DE, Ailles LE, Lansdorp PM, Sutherland HJ. (1997). characteristics of acute myelogenous leukemia progenitors that Lack of expression of Thy-1 (CD90) on acute myeloid leukemia cells initiate malignant hematopoiesis in nonobese diabetic/severe com- with long-term proliferative ability in vitro and in vivo. Blood 89: bined immunodeficient mice. Blood 94: 1761–1772. 3104–3112. Bakker AB, van den Oudenrijn S, Bakker AQ, Feller N, van Meijer M, Blair A, Hogge DE, Sutherland HJ. (1998). Most acute myeloid Bia JA et al. (2004). C-type lectin-like molecule-1: a novel myeloid leukemia progenitor cells with long-term proliferative ability in vitro cell surface marker associated with acute myeloid leukemia. Cancer and in vivo have the phenotype CD34(+)/CD71(-)/HLA-DR. Blood Res 64: 8443–8450. 92: 4325–4335. Barclay AN, Brown MH. (2006). The SIRP family of receptors and Blazar BR, Lindberg FP, Ingulli E, Panoskaltsis-Mortari A, Old- immune regulation. Nat Rev Immunol 6: 457–464. enborg PA, Iizuka K et al. (2001). CD47 (-associated

Oncogene Antibody therapy for AML LSC R Majeti 1018 protein) engagement of dendritic cell and macrophage counter- Jordan CT, Upchurch D, Szilvassy SJ, Guzman ML, Howard DS, receptors is required to prevent the clearance of donor lymphohe- Pettigrew AL et al. (2000). The interleukin-3 receptor alpha chain is matopoietic cells. J Exp Med 194: 541–549. a unique marker for human acute myelogenous leukemia stem cells. Bonnet D, Dick JE. (1997). Human acute myeloid leukemia is Leukemia 14: 1777–1784. organized as a hierarchy that originates from a primitive hemato- Kelly PN, Dakic A, Adams JM, Nutt SL, Strasser A. (2007). poietic cell. Nat Med 3: 730–737. Tumor growth need not be driven by rare cancer stem cells. Science Brown EJ, Frazier WA. (2001). Integrin-associated protein (CD47) 317: 337. and its ligands. Trends Cell Biol 11: 130–135. Lapidot T, Sirard C, Vormoor J, Murdoch B, Hoang T, Burnett AK, Mohite U. (2006). Treatment of older patients with acute Caceres-Cortes J et al. (1994). A cell initiating human acute myeloid leukemia—new agents. Semin Hematol 43: 96–106. myeloid leukaemia after transplantation into SCID mice. Nature Chao MP, Alizadeh AA, Tang C, Myklebust JH, Varghese B, Gill S 367: 645–648. et al. (2010). Anti-CD47 antibody synergizes with rituximab to Levis M, Murphy KM, Pham R, Kim KT, Stine A, Li L et al. (2005). promote phagocytosis and eradicate non-Hodgkin lymphoma. Cell Internal tandem duplications of the FLT3 gene are present in 142: 699–713. leukemia stem cells. Blood 106: 673–680. Charrad RS, Li Y, Delpech B, Balitrand N, Clay D, Jasmin C et al. Lowenberg B, Downing JR, Burnett A. (1999). Acute myeloid (1999). Ligation of the CD44 adhesion molecule reverses blockage leukemia. N Engl J Med 341: 1051–1062. of differentiation in human acute myeloid leukemia. Nat Med 5: Majeti R, Chao MP, Alizadeh AA, Pang WW, Jaiswal S, Gibbs Jr KD 669–676. et al. (2009). CD47 is an adverse prognostic factor and therapeutic Cheson BD, Leonard JP. (2008). Monoclonal antibody therapy for antibody target on human acute myeloid leukemia stem cells. Cell B-cell non-Hodgkin’s lymphoma. N Engl J Med 359: 613–626. 138: 286–299. Craig W, Kay R, Cutler RL, Lansdorp PM. (1993). Expression Majeti R, Park CY, Weissman IL. (2007). Identification of a hierarchy of Thy-1 on human hematopoietic progenitor cells. J Exp Med 177: of multipotent hematopoietic progenitors in human cord blood. 1331–1342. Cell Stem Cell 1: 635–645. Dalerba P, Cho RW, Clarke MF. (2007). Cancer stem cells: models Miyamoto T, Weissman IL, Akashi K. (2000). AML1/ETO-expressing and concepts. Annu Rev Med 58: 267–284. nonleukemic stem cells in acute myelogenous leukemia with Dick JE. (2008). Stem cell concepts renew cancer research. Blood 112: 8;21 chromosomal translocation. Proc Natl Acad Sci USA 97: 4793–4807. 7521–7526. Estey E, Dohner H. (2006). Acute myeloid leukaemia. Lancet 368: Murray L, Chen B, Galy A, Chen S, Tushinski R, Uchida N et al. 1894–1907. (1995). Enrichment of human hematopoietic stem cell activity in the Finn OJ. (2008). Cancer immunology. N Engl J Med 358: 2704–2715. CD34+Thy-1+Lin- subpopulation from mobilized peripheral Florian S, Sonneck K, Hauswirth AW, Krauth MT, Schernthaner GH, blood. Blood 85: 368–378. Sperr WR et al. (2006). Detection of molecular targets on the Okazawa H, Motegi S, Ohyama N, Ohnishi H, Tomizawa T, surface of CD34+/CD38—stem cells in various myeloid malig- Kaneko Y et al. (2005). Negative regulation of phagocytosis nancies. Leuk Lymphoma 47: 207–222. in macrophages by the CD47-SHPS-1 system. J Immunol 174: Gilliland DG, Griffin JD. (2002). The roles of FLT3 in hematopoiesis 2004–2011. and leukemia. Blood 100: 1532–1542. Oldenborg PA, Gresham HD, Lindberg FP. (2001). CD47-signal Gilliland DG, Jordan CT, Felix CA. (2004). The molecular basis of regulatory protein alpha (SIRPalpha) regulates Fcgamma and leukemia. Hematology Am Soc Hematol Educ Program, 80–97. complement receptor-mediated phagocytosis. J Exp Med 193: Gramatzki M, Ludwig WD, Burger R, Moos P, Rohwer P, Grunert C 855–862. et al. (1998). Antibodies TC-12 (‘unique’) and TH-111 (CD96) Oldenborg PA, Zheleznyak A, Fang YF, Lagenaur CF, Gresham HD, characterize T-cell acute lymphoblastic leukemia and a subgroup of Lindberg FP. (2000). Role of CD47 as a marker of self on red blood acute myeloid leukemia. Exp Hematol 26: 1209–1214. cells. Science 288: 2051–2054. Hosen N, Park CY, Tatsumi N, Oji Y, Sugiyama H, Gramatzki M Quintana E, Shackleton M, Sabel MS, Fullen DR, Johnson TM, et al. (2007). CD96 is a leukemic stem cell-specific marker in human Morrison SJ. (2008). Efficient tumour formation by single human acute myeloid leukemia. Proc Natl Acad Sci USA 104: 11008–11013. cells. Nature 456: 593–598. Ishikawa F, Yasukawa M, Lyons B, Yoshida S, Miyamoto T, Reya T, Morrison SJ, Clarke MF, Weissman IL. (2001). Stem cells, Yoshimoto G et al. (2005). Development of functional human cancer, and cancer stem cells. Nature 414: 105–111. blood and immune systems in NOD/SCID/IL2 rec. Blood 106: Roberts AW, He S, Bradstock KF, Hertzberg MS, Durrant STS, 1565–1573. Ritchie D et al. (2008). A phase 1 and correlative biological study of Ishikawa F, Yoshida S, Saito Y, Hijikata A, Kitamura H, Tanaka S CSL360 (anti-CD123 mAb) in AML. Blood 112: 2956a. et al. (2007). Chemotherapy-resistant human AML stem cells home Saito Y, Kitamura H, Hijikata A, Tomizawa-Murasawa M, Tanaka S, to and engraft within the bone-marrow endosteal region. Nat Takagi S et al. (2010). Identification of therapeutic targets for Biotechnol 25: 1315–1321. quiescent, chemotherapy-resistant human leukemia stem cells. Jaiswal S, Jamieson CH, Pang WW, Park CY, Chao MP, Sci Transl Med 2: 17ra19. Majeti R et al. (2009). CD47 is upregulated on circulating Shultz LD, Lyons BL, Burzenski LM, Gott B, Chen X, Chaleff S et al. hematopoietic stem cells and leukemia cells to avoid phagocytosis. (2005). Human lymphoid and myeloid cell development in NOD/ Cell 138: 271–285. LtSz-scid IL2R gamma null mice engrafted with mobilized human Jawad M, Seedhouse C, Mony U, Grundy M, Russell NH, Pallis M. hemopoietic stem cells. J Immunol 174: 6477–6489. (2010). Analysis of factors that affect in vitro chemosensitivity of Shultz LD, Ishikawa F, Greiner DL. (2007). Humanized mice in leukaemic stem and progenitor cells to gemtuzumab ozogamicin translational biomedical research. Nat Rev Immunol 7: 118–130. (Mylotarg) in acute myeloid leukaemia. Leukemia 24: 74–80. Stasi R, Evangelista ML, Buccisano F, Venditti A, Amadori S. (2008). Jin L, Hope KJ, Zhai Q, Smadja-Joffe F, Dick JE. (2006). Targeting of Gemtuzumab ozogamicin in the treatment of acute myeloid CD44 eradicates human acute myeloid leukemic stem cells. Nat Med leukemia. Cancer Treat Rev 34: 49–60. 12: 1167–1174. Subramanian S, Parthasarathy R, Sen S, Boder ET, Discher DE. Jin L, Lee EM, Ramshaw HS, Busfield SJ, Peoppl AG, Wilkinson L (2006). Species- and cell type-specific interactions between CD47 et al. (2009). Monoclonal antibody-mediated targeting of CD123, and human SIRPalpha. Blood 107: 2548–2556. IL-3 receptor alpha chain, eliminates human acute myeloid leukemic Takenaka K, Prasolava TK, Wang JC, Mortin-Toth SM, stem cells. Cell Stem Cell 5: 31–42. Khalouei S, Gan OI et al. (2007). Polymorphism in Sirpa modulates Jordan CT, Guzman ML, Noble M. (2006). Cancer stem cells. N Engl engraftment of human hematopoietic stem cells. Nat Immunol 8: JMed355: 1253–1261. 1313–1323.

Oncogene Antibody therapy for AML LSC R Majeti 1019 Tan BT, Park CY, Ailles LE, Weissman IL. (2006). The cancer stem leukemia at diagnosis predicts high minimal residual disease and cell hypothesis: a work in progress. Lab Invest 86: 1203–1207. poor survival. Clin Cancer Res 11: 6520–6527. Taussig DC, Miraki-Moud F, Anjos-Afonso F, Pearce DJ, Allen K, van Rhenen A, van Dongen GA, Kelder A, Rombouts EJ, Feller N, Ridler C et al. (2008). Anti-CD38 antibody-mediated clearance of Moshaver B et al. (2007). The novel AML stem cell associated human repopulating cells masks the heterogeneity of leukemia- antigen CLL-1 aids in discrimination between normal and leukemic initiating cells. Blood 112: 568–575. stem cells. Blood 110: 2659–2666. Taussig DC, Pearce DJ, Simpson C, Rohatiner AZ, Lister TA, Kelly G Vardiman JW, Thiele J, Arber DA, Brunning RD, Borowitz MJ, et al. (2005). Hematopoietic stem cells express multiple myeloid Porwit A et al. (2009). The 2008 revision of the World Health markers: implications for the origin and targeted therapy of acute Organization (WHO) classification of myeloid neoplasms and acute myeloid leukemia. Blood 106: 4086–4092. leukemia: rationale and important changes. Blood 114: 937–951. Taussig DC, Vargaftig J, Miraki-Moud F, Griessinger E, Sharrock K, Wang PL, O’Farrell S, Clayberger C, Krensky AM. (1992). Luke T et al. (2010). Leukemia initiating cells from some acute Identification and molecular cloning of tactile. A novel human T myeloid leukemia patients with mutated nucleophosmin reside in cell activation antigen that is a member of the Ig gene superfamily. the CD34- fraction. Blood 115: 1976–1984. J Immunol 148: 2600–2608. Terwijn M, Feller N, van Rhenen A, Kelder A, Westra G, Zweegman Yalcintepe L, Frankel AE, Hogge DE. (2006). Expression of S et al. (2009). Interleukin-2 receptor alpha-chain (CD25) expression interleukin-3 receptor subunits on defined subpopulations of acute on leukaemic blasts is predictive for outcome and level of residual myeloid leukemia blasts predicts the cytotoxicity of diphtheria toxin disease in AML. Eur J Cancer 45: 1692–1699. interleukin-3 fusion protein against malignant progenitors that Testa U, Riccioni R, Diverio D, Rossini A, Lo Coco F, Peschle C. engraft in immunodeficient mice. Blood 108: 3530–3537. (2004). Interleukin-3 receptor in acute leukemia. Leukemia 18: Zhao X, Singh S, Pardoux C, Zhao J, Hsi ED, Abo A et al. (2010). 219–226. Targeting C-type lectin-like molecule-1 for antibody-mediated van Rhenen A, Feller N, Kelder A, Westra AH, Rombouts E, immunotherapy in acute myeloid leukemia. Haematologica 95: Zweegman S et al. (2005). High stem cell frequency in acute myeloid 71–78.

Oncogene