The therapeutic strategies for eradicating LSCs are summarized in Table 20.1. Molecules that control self-renewal and proliferation, molecules that regulate the cell cycle and differentiation, molecules that enhance cell survival, molecules constituting the hematopoietic (leukemic) niche that exists in hypoxic environments, and cell surface molecules specific for leukemia are candidates for eradicating LSCs. However, many of these molecules are present even in normal HSCs. To specifically eradicate LSCs while sparing normal HSCs, the specific molecular mechanisms in LSCs need to be clarified.

The canonical Wnt/β-catenin signal is constitutively active in both AML and chronic myeloid leukemia (CML), in addition to many human cancers [16]. Analysis of patient samples demonstrated that LSCs in the chronic phase of CML develop from HSCs, while those in the accelerated and blastic phases of CML develop by transformation of CMPs or GMPs through activation of the Wnt/β-catenin pathway. β- and γ-catenin are not essential at least for the maintenance of murine normal HSCs, and they are therefore potential therapeutic targets in LSCs [17].

The hedgehog (Hh) pathway is abnormally activated in various types of cancer [18]. During embryonic pattern formation, the Hh signaling pathway regulates proliferation and differentiation. In cancer, the pathway increases tumor invasiveness and regular CSC proliferation. In some preclinical models, aberrant activation of the Hh pathway is necessary for the maintenance of some CD34⁺ LSCs. PF-04449913 is a novel oral small molecule inhibitor that binds to smoothened (SMO), a membrane protein in the Hh pathway [19]. In AML cell lines and human primary AML cells, PF-04449913 appears to improve sensitivity to cytarabine in dormant LSCs. Multiple phase 1 studies are ongoing with PF-04449913 as a monotherapy or in combination with low-dose cytarabine or hypomethylating agents to treat AML [20]. Vismodegib, a SMO antagonist approved for the treatment of basal-cell carcinoma, in combination with ribavirin with or without decitabine, is also under investigation in relapsed/refractory AML (NCT02073838) [20].

Identifying cell surface antigens that are differentially upregulated on LSCs but not on normal HSCs has been a major focus. However, it is difficult to identify such antigens, and thus far no unique antigen has been found that is specifically expressed on CD34⁺CD38⁻ LSCs but not on normal HSCs. Despite these difficulties, many cell surface antigens have been identified that are preferentially expressed on CD34⁺CD38⁻ LSCs compared with normal human HSCs. These include CD25 [21], CD32 [21], CD33 [22], CD44 [23], CD47 [24, 25], CD96 [26], CD123 [27–29], TIM-3 [30–32], interleukin (IL)-1 receptor accessory protein (IL1RAP) [37], and C-type lectin-like molecule-1 (CLL-1) [38)], as summarized in Table 20.2.

In normal human BM, CD25 is expressed in a portion of CD4⁺ T cells (regulatory T cells), and CD32 (Fc gamma receptor II) is expressed in B cells, T cells, and monocytes; however, neither antigen is expressed in CD34⁺CD38⁻CD133⁺ HSCs. By demonstrating the functional significance of distinctly expressed genes using analyses including xenotransplantation, it was reported that CD25 and CD32 were most frequently expressed on LSCs of AML patients, 85% of whom were poor risk, and it was suggested that therapeutic strategies targeting these molecules may be effective and improve patient prognosis [21].

CD44 is a cell adhesion molecule whose ligand is hyaluronic acid and which contributes to construction of the LSC niche. The engraftment of human AML into immunodeficient mice is completely inhibited by administration of an anti-CD44 antibody (H90), while that of normal HSCs is only minimally affected [23]. In this paper, Jin et al. reported that CD44 is a key regulator of AML-LSC function and is required for proper homing of AML-LSCs to microenvironments and to maintain AML-LSCs in a leukemic niche. The finding that AML-LSCs require interactions with a niche to maintain their stem cell properties provides a therapeutic strategy to eliminate dormant AML-LSCs and may be applicable to other types of CSCs.

CD47 belongs to the immunoglobulin (Ig) superfamily and is a ligand of SIRPα (signal regulatory protein alpha) expressed on macrophages. When the IgV region of SIRPα expressed on macrophages combines with CD47 on LSCs, a tyrosine residue of SIRPα is phosphorylated, and the inhibitory signals of phagocytosis (“don’t eat me” signal) are transmitted, resulting in the inhibition of phagocytosis by macrophages. On the other hand, when the target cell does not express CD47 or the binding of SIRPα and CD47 is inhibited, the signal of the inhibition is not transmitted, and consequently the target cells are phagocytosed by macrophages. When an anti-CD47 antibody blocks the binding of SIRPα and CD47 expressed on LSCs, the “don’t eat me” signal to macrophages is blocked, and LSCs are phagocytosed. Increased phagocytic action by macrophages is a mechanism of action of anti-CD47 antibodies. CD47 is strongly expressed on AML-LSCs compared with normal HSCs. CD47 binds to the ligands thrombospondin-1 and SIRPα, which are involved in apoptosis, proliferation, adhesion, and migration of myeloma cells [24]. In clinical settings, increased expression of CD47 is a poor prognostic factor in AML. Treatment of human AML-engrafted mice with an anti-CD47 antibody could target AML-LSCs [25].

CD96 belongs to the Ig superfamily. Hosen et al. revealed that CD96 is a promising candidate as a specific antigen for LSCs [26]. Applying a signal sequence trap strategy to identify cell surface molecules expressed on human AML-LSCs and fluorescence-activated cell sorting (FACS) analysis demonstrated that CD96 is expressed on CD34⁺CD38⁻ AML cells in the majority of patients. Transplantation of AML cells into immunodeficient mice demonstrated that only CD96⁺ fractions showed significant levels of engraftment. These results demonstrate that CD96 is a cell surface marker present on AML-LSCs and may be a LSC-specific therapeutic target. Furthermore, FACS analysis demonstrates that CD96 is expressed on the majority of CD34⁺CD38⁻ LSCs of AML, whereas only a few normal HSCs weakly express CD96. Therefore, the authors mentioned that this molecule may be a candidate LSC-specific therapeutic target.

CD123 is an IL-3 receptor α chain, is highly expressed on AML-LSCs compared with their normal counterparts, and is being used to develop an antibody therapy. An anti-CD123-neutralizing antibody (7G3) targeted AML-LSCs, impairing homing to BM and activating innate immunity in NOD/SCID mice. 7G3 treatment profoundly reduced AML-LSC engraftment and improved mouse survival [27]. Targeting LSCs using the 7G3 antibody against CD123 seems to be an attractive approach. A phase 1 clinical study was conducted of relapsed or refractory high-risk AML using CSL360, a chimeric variant of 7G3; however, the remission rate was quite low [28]. One reason for this was the high tumor load, and strategies such as enhancing the ADCC (antibody-dependent cellular cytotoxicity) activity of anti-CD123 antibodies have been investigated. Moreover, it is an immune cell therapy, and preclinical work with a dual-affinity retargeting molecule generated from antibodies against CD3 and CD123, which is designed to redirect T cells against AML-LSCs, has been reported [29].

T-cell immunoglobulin mucin (TIM)-3 is highly expressed in AML-LSCs, except the FAB classification M3, but is not expressed at all in normal HSCs. While TIM-3-positive AML cells can repopulate AML in mice with high efficiency, TIM-3-negative AML cells do not reconstitute leukemia at all. In immunodeficient mice engrafted with AML cells and treated with an anti-TIM-3 antibody (ATIK2a), AML cells were significantly reduced, and those remaining after treatment with ATIK2a could not succeed in serial passage transplantation in mice, indicating that this antibody can target AML-LSCs [30]. TIM-3 protein is not detectable in normal HSCs or in other myeloerythroid or lymphoid progenitors, although monocyte lineage-committed progenitors begin to upregulate TIM-3. Therefore, expression of TIM-3 is highly specific to human AML-LSCs [31]. TIM-3 was originally identified as being selectively expressed on IFN-γ-secreting Th1 and Tc1 cells [32]. Interaction of TIM-3 with its ligand, galectin-9, triggers cell death in TIM-3⁺ T cells [33]. Both TIM-3 and PD-1 can function as negative regulators of T-cell responses [34, 35, 36]. Like PD-1, TIM-3 is also an immune checkpoint-associated surface molecule involved in the exhaustion of T cells. Inhibition of TIM-3 functions with an anti-TIM-3 antibody should be an effective treatment strategy. From the perspective of the abovementioned immune checkpoint inhibition, TIM-3 is expressed in some types of hematopoietic cells and seems to have lineage- or cellular context-dependent signal transduction pathways or functions. The role of TIM-3 in AML-LSCs is quite intriguing, and inhibition of this molecule is a promising clinical strategy. Further studies, however, are needed to clarify the specific function of TIM-3 before performing clinical trials to eradicate AML-LSCs.

IL1RAP is a co-receptor of the type 1 IL-1 receptor. IL1RAP is expressed on the cell surface in the majority of AML patients. IL1RAP is upregulated on immature cells in high-risk AML patients with chromosome 7 abnormalities, and AML patients with high IL1RAP expression have a poor prognosis. Ågerstam et al. reported that IL1RAP may become a therapeutic target in AML and that rapid clinical development of an antibody-based IL1RAP therapy for AML is promising [37].

CLL-1 is highly expressed on the blast compartment in the majority of AML cases, while it is completely absent on CD34⁺CD38⁻ resting BM cells. CD34⁺CLL-1⁺ cells repopulate in sublethally irradiated immunodeficient NOD/SCID mice, indicating that they contain leukemia-initiating cells. Furthermore, culture of normal CD34⁺ cells in a long-term culture system in the presence of an anti-CLL-1 antibody has no effect on CFU-E and CFU-GM colony formation, suggesting that CLL-1 is a possible target for therapy [38].

The role of CCAAT/enhancer-binding protein β (C/EBPβ), a regulator of emergency granulopoiesis, in the pathogenesis of the chronic phase of CML was examined, and it has been suggested that C/EBPβ is involved in BCR-ABL-mediated myeloid expansion. We found that enforced expression of C/EBPβ in CML-LSCs might induce exhaustion of LSCs, leading to a complete cure of this disease [39]. C/EBPβ-mediated LSC loss might reveal a unique therapeutic strategy to eradicate CML stem cells. Agents that upregulate expression of C/EBPβ in LSCs should be promising (Yokota A, Hirai H, Maekawa T, et al., unpublished observation).

In addition to therapies using the antibodies mentioned above, therapies based on small molecules can target intracellular proteins and pathways. Nuclear factor-kappa B (NF-κB) stimulates the transcription of genes encoding Ig of the κ class in B lymphocytes. NF-κB, a dimeric transcription factor complex that controls various aspects of cellular responses to stimuli, is now recognized as a pivotal regulator of cell survival, proliferation, and differentiation. When unstimulated, the complex is sequestered in the cytoplasm by a family of inhibitors of NF-κB (IκBs). NF-κB is constitutively active in AML enriched with LSCs, while its activity was not detectable in unstimulated normal HSCs/progenitor cells. Proteasome inhibitor such as MG-132, which inhibits NF-κB activity by inhibiting IκB degradation, induces the rapid apoptosis of AML-LSCs but not of normal HSCs. Therefore, inhibition of NF-κB signaling may provide a novel therapeutic strategy specific to LSCs [40]. Another molecule with inhibitory activity against NF-κB, parthenolide (PTL), targets primitive AML cells. Agents such as PTL, which is a sesquiterpene lactone that is the major active component in feverfew (Tanacetum parthenium), a herbal medicine that has been used to treat migraines and rheumatoid arthritis for a long time, inhibit the ability of LSCs to respond to oxidative stress and make LSCs sensitive to cell death stimuli, while normal stem cells remain relatively unharmed by these agents. The main mechanism of action of these molecules appears to revolve around the altered glutathione metabolism pathway found in leukemia cells [41]. Niclosamide also inhibits the transcription and DNA binding of NF-κB. It blocked tumor necrosis factor-induced IκBα phosphorylation, translocation of p65, and expression of NF-κB-regulated genes [42].

A body of investigations has established that AML cells are dependent on antiapoptotic molecules for survival such as the B-cell CLL/lymphoma 2 (Bcl-2) family, which suppresses the intrinsic or mitochondrial apoptotic pathway and plays an important role in AML pathogenesis, prognosis, and responsiveness to chemotherapeutic agents [43]. In addition, the antiapoptotic Bcl-2 family protein myeloid cell leukemia sequence 1 (Mcl-1) is essential for cell survival during the development and maintenance of AML [44, 45].

Normally, the antiapoptotic proteins Bcl-2, Bcl-X_L, and Mcl-1 suppress apoptosis effector molecules such as Bak as well as the proapoptotic BH3-only protein Bim, thereby preventing apoptosis. The BH3-only proteins Bad and Noxa, as well as BH3-mimetic drugs such as navitoclax and obatoclax, untether the apoptosis effectors and Bim from the antiapoptotic proteins, leading to apoptosis [46]. Apoptosis is initiated when members of the third subfamily, the BH3-only proteins (e.g., Bim, Bad, and Noxa), are activated by various cytotoxic stimuli.