Abstract
NOTCH activation plays oncogenic roles in acute T-lymphoblastic leukaemia (T-ALL). However, whether NOTCH is oncogenic or tumor-suppressive in acute myeloid leukaemia (AML) is still controversial. Herein, the roles of NOTCH in AML are reviewed. AML cells express NOTCH and NOTCH ligands; however, cell-autonomous activation is not observed. Activating NOTCH1 mutations are rare in AML, unlike in T-ALL. NOTCH ligand stimulation generally suppresses the in vitro growth of AML cells but promotes transient growth of some samples. Conversely, knockdown of NOTCH1 and NOTCH2 does not affect the growth of AML cells, whereas it suppresses the growth of T-ALL cells. These findings suggest that NOTCH is dispensable or suppressive for AML cell growth. However, the effects of NOTCH differ depending on cell conditions, and various stemness-related signals modify these effects; hence, forced NOTCH activation in vitro may not exhibit effects in bone marrow. Thus, further understanding is required for the development of AML therapies targeting NOTCH signalling.
NOTCH signalling plays crucial roles in cell fate decisions during development, stem cell self-renewal, and differentiation in various systems such as neurogenesis and haematopoiesis (1). Dysregulation of NOTCH signaling has been reported in human haematological malignancies and various solid tumors (2). NOTCH activation plays oncogenic roles in many cancers, while it acts as a tumor suppressor in certain malignancies. Therefore, the NOTCH pathway is a potential, context-dependent therapeutic target (3). Notably, more than half patients with acute T-lymphoblastic leukaemia (T-ALL) have activating NOTCH1 mutations (4). However, the roles of NOTCH signalling in acute myeloid leukaemia (AML) are still controversial. We first reported that NOTCH1 and JAGGED1 proteins are expressed in AML cell lines and primary AML cells (5). Thus far, we have investigated the roles of NOTCH signaling in AML cells compared with T-ALL cells. We have also investigated the crosstalk between NOTCH and other stemness-related signalling pathways such as HEDGEHOG and wingless-type mouse mammary tumor virus integration site family (WNT).
NOTCH is a crucial regulator of normal stem cells and, therefore, may play important roles in leukaemia stem cells. To develop NOTCH-targeting therapeutics for leukaemia stem cells, the mechanisms of NOTCH regulation must be clearly understood in a context-dependent manner. In this review, present our findings in addition to outstanding work from other investigators, focusing on the role of NOTCH signalling in AML cells.
NOTCH Signalling Pathway
NOTCH is a transmembrane receptor that forms a heterodimer consisting of extracellular (NEC) and transmembrane (N™) subunits (Figure 1). NOTCH is expressed on both haematopoietic cells and leukaemia cells. NOTCH ligands are also transmembrane proteins expressed on bone marrow stromal cells. Mammalian cells express four kinds of NOTCH receptors, NOTCH1-4, and five NOTCH ligands, delta-like protein (DLL) 1, DLL3, DLL4, JAGGED1, and JAGGED2. Ligand binding to NOTCH receptors causes cleavage of N™ by γ-secretase and releases an intracellular fragment of NOTCH (ICN). ICN is translocated to the nucleus, where it induces expression of its target genes such as hairy and enhancer of split 1 (HES1) and v-Myc avian myelocytomatosis viral oncogene homolog (MYC). Although NOTCH activation generally leads to maintenance of stem cells, its roles differ in cell type- and context-dependent manners. In lymphopoiesis, NOTCH1 and NOTCH2 affect T-cell differentiation and marginal zone B-cell differentiation, respectively, while their roles in myeloid lineages are not fully clarified.
NOTCH Expression and Mutation in AML Cells
Activating NOTCH1 mutations are involved in the pathogenesis of T-ALL. Weng et al. reported that 56% of patients with T-ALL have mutations in the heterodimerization domain that cause ligand-independent ICN cleavage and/or mutations in polypeptide enriched in proline, glutamine, serine, and threonine (PEST) domain that prolong the half-life of ICN (4). Previously, we demonstrated the presence of NOTCH1 and JAGGED1 proteins in AML cells by immunoblotting (5). Among 23 cell lines and patient samples, NOTCH1 or JAGGED1 were expressed in 16 samples. Eight samples expressed both NOTCH1 and JAGGED1, which has the potential to cause autonomous activation of NOTCH. Flow cytometric analysis also showed that NOTCH1 or JAGGED1 were expressed in more than half of the AML patient samples and all AML cell lines examined. NOTCH1 expression on AML cell lines was more intense than on Jurkat, a T-ALL cell line (6).
Next, we searched for NOTCH1 mutations in 20 AML cell lines and patient samples. We detected a missense mutation in the PEST domain in an AML sample (FAB M4) expressing CD2 and CD4 (7). Wouters et al. reported that NOTCH1 mutations were detected in acute myeloid/T-lymphoid leukaemia cases (8). Our case, which was negative for CD7, did not fit this category. We also reported that NOTCH1 mutations were not detected in 20 cases of myelodysplastic syndrome (9), which is in agreement with the report by Palomero et al., which stated that an activating NOTCH1 mutation was detected in the heterodimerization domain in only one AML case out of 121 patients with AML or myelodysplastic syndrome (10). Thus, NOTCH1 mutations are very rare in AML, unlike in T-ALL.
Effects of NOTCH Activation in AML Cells
To examine the function of NOTCH in AML cells, we tested the effects of recombinant NOTCH ligands on the in vitro growth of AML cell lines. We established a novel cell line (TMD7) from blast cells of a patient with AML with trilineage myelodysplasia, for which NOTCH ligand stimulation caused a growth response (11). Stimulation with DLL1 and DLL4 bound on culture plates promoted short-term growth, while it suppressed long-term growth. The number of clonogenic cells recovered after suspension culture (CCR) in the presence of the ligands, which reflects the self-renewal capacity of leukaemia stem cells, was also decreased in comparison with CCR without the ligands. We also found that soluble ligands in the culture medium did not affect growth; the ligands must be immobilized on culture plates to affect the cells (12). For a monoblastic leukaemia cell line (THP-1), JAGGED1, DLL1, and DLL4 suppressed both short- and long-term growth and induced differentiation into macrophage-like cells (13).
Next, we examined the effects on the in vitro growth of cells obtained from 12 AML patient samples. Ligand stimulation caused three types of short-term growth responses: growth promotion, growth suppression, or no significant effect. Ligand stimulation suppressed CCR and induced differentiation in the cells of some samples; promotion of CCR was not observed in the cells of any of the samples examined (14). Thus, we found that NOTCH activation acts as a tumor suppressor in AML cells. Recently, two studies supported this view by demonstrating that induction of constitutively active NOTCH in AML cells inhibited growth (15, 16). In contrast, Kode et al. reported that activating mutations of β-catenin in osteoblasts stimulates JAGGED1 expression, which subsequently activates NOTCH in haematopoietic stem cells and causes AML (17). This indicates that NOTCH activation plays oncogenic roles in the pathogenesis of AML.
Acute promyelocytic leukaemia (APL) is a sub-type of AML characterized by expression of the promyelocytic leukemia (PML)-retinoic acid receptor alpha (RARA) fusion gene. Treatment with all-trans retinoic acid (ATRA) differentiates APL cells into neutrophil-like cells and induces apoptosis. We found that treatment with ATRA plus DLL1 differentiated the APL-derived cell line NB4 and primary APL cells into monocyte-like cells and reduced apoptosis compared with ATRA-alone, although DLL1 stimulation itself did not affect growth (18). Thus, NOTCH signalling affects the direction of differentiation. Recently, Grieselhuber et al. reported that primary APL samples overexpress JAGGED1 mRNA in comparison with other AML subtypes. Induction of PML–RARA expression in a cell line increased JAGGED1 protein levels and cleaved the NOTCH1 fragment. NOTCH inhibition reduced growth of APL cells (19). These findings indicate that NOTCH activation plays oncogenic roles in APL.
In a monocytic leukaemia cell line (U937), we showed that DLL1 stimulation reduced tumor necrosis factor-α-induced apoptosis by decreasing activation of caspases, although DLL1 stimulation itself did not affect growth (20). Thus, NOTCH signalling suppresses the apoptosis induced by stimulation in a subset of myeloid leukaemia cells.
Recently, Adamia et al. reported that the majority of AML cases express aberrantly spliced NOTCH2 mRNA and that the cases expressing splice variant NOTCH2 showed lower expression of NOTCH target genes such as HES1. This suggests that splice variant NOTCH2 might act as dominant-negative. AML cases expressing splice variant NOTCH2 had a poor prognosis (21).
Downstream Molecules in AML Cells Stimulated with NOTCH Ligands
As mentioned above, DLL1 stimulation promoted short-term growth of TMD7 cells while it suppressed that of THP-1 cells. To clarify the difference in mechanisms, downstream molecules of NOTCH were examined. When stimulated with DLL1, ICN bands emerged in immunoblots for both cell lines, and HES1 expression was up-regulated, as determined by quantitative reverse transcription-polymerase chain reaction (qRT-PCR). In THP-1 cells, DLL1 stimulation down-regulated MYC expression, while it up-regulated expression of FBJ murine osteosarcoma viral oncogene homolog (FOS), cyclin D1 (CCND1), and cyclin-dependent kinase inhibitor-1A (CDKN1A). Comparatively, DLL1 stimulation in TMD7 cells down-regulated expression of FOS, while it did not significantly affect MYC, CCND1, or CDKN1A expression levels (22). Microarray analysis also showed that gene expression profiles accompanying NOTCH activation differ between THP-1 and TMD7 cells (23). Additionally, DLL1 stimulation activated the nuclear factor-kappa B (NF-κB) pathway in THP-1 cells by increasing the expression of certain components of the pathway and inducing phosphorylation of NF-κB inhibitor (IκB), IκB kinase (IKK), and v-rel avian reticuloendotheliosis viral oncogene homolog A (RELA). In contrast, DLL1 stimulation did not affect NF-κB activity in TMD7 cells (23). Although these results cannot clearly account for the differences in growth, it is apparent that the effects of NOTCH activation on the downstream molecules depend on the cell type and cause different cellular responses.
Effects of NOTCH Inhibition in AML Cells
NOTCH activity can be inhibited by γ-secretase inhibitors (GSI) or antibodies to NOTCH receptors or NOTCH ligands (3). NOTCH inhibition is a promising therapy for various types of cancer such as T-ALL in which NOTCH plays oncogenic roles. Although GSI treatment suppresses in vitro growth and induces apoptosis of T-ALL cell lines (4, 24), successful clinical trials of GSI for T-ALL have not been reported.
We examined the effects of GSI on the in vitro growth of AML cell lines. GSI-XXI (compound E) did not significantly suppress the growth of AML cells, while it slightly promoted the growth of THP-1 cells. GSI-XII (Z-IL-CHO) suppressed the growth of AML cell lines examined; however, this effect might be due to off-target effects of GSI because GSI inhibits various substrates such as cadherin, CD44, and v-erb-b2 avian erythroblastic leukemia viral oncogene homolog 4 (ERBB4) (25). We also found that GSI treatment induces erythroid differentiation of erythroblastic leukaemia cell lines (26). To our knowledge, this is the only report to show differentiation-induction of leukaemia cells by GSI.
To examine the effects of NOTCH inhibition excluding off-target effects, we used small-interfering RNA (siRNA) to knock-down NOTCH1 and NOTCH2 expression. Neither NOTCH1 nor NOTCH2 knockdown affected the growth of TMD7 and THP-1 cells, while both suppressed the growth of T-ALL cell lines (27). This suggests that AML cells, or at least the two AML cell lines examined, do not require NOTCH signalling for their growth. In these experiments, we found that NOTCH2 knockdown increased the level of the cleaved NOTCH1 fragment without increasing NOTCH1 expression in TMD7 and THP-1 cells. The mechanism and biological meaning behind this interesting phenomenon remain to be determined.
Interaction Between NOTCH and WNT or Hedgehog Signaling
Haematopoietic stem cells (HSCs) are regulated by various stemness-related signaling pathways such as NOTCH, WNT, and HEDGEHOG. Therefore, we experimentally examined the crosstalk between NOTCH and WNT or HEDGEHOG signalling in AML and T-ALL cells by administering their ligands and inhibitors.
WNT signaling regulates self-renewal of HSCs (28). We examined the effects of recombinant WNT3a protein on the in vitro growth of AML and T-ALL cell lines. WNT3a treatment did not affect the short-term growth, while it slightly promoted CCR for some AML and T-ALL cell lines, which suggests the promotion of self-renewal capacity. WNT3a treatment did not affect expression and activity of NOTCH (29). Treatment with a WNT inhibitor, quercetin, suppressed the growth of AML and T-ALL cell lines and reduced NOTCH1 expression in a T-ALL cell line, DND-41 (30). The combinatorial administration of quercetin and GSI synergistically suppressed the growth of DND-41 cells (31). Treatment with a combination of these inhibitors might overcome the failure of GSI-alone in treating T-ALL.
Hedgehog signalling also regulates HSC self-renewal (32). Treatment with recombinant sonic Hedgehog protein did not affect short-term growth, while it slightly promoted CCR of some AML cell lines, but no effect on NOTCH expression was observed (33). Treatment with a Hedgehog inhibitor, cyclopamine, suppressed the growth of some AML and T-ALL cell lines (30, 31). Cyclopamine treatment reduced NOTCH1 expression in DND-41 cells, and combinatorial administration of cyclopamine and GSI additively suppressed the growth of DND-41 cells (31). Thus, WNT and Hedgehog signalling might be related to the NOTCH pathway.
Interaction Between NOTCH and Mechanistic Target of Rapamycin (mTOR), Hypoxia-inducible Factor (HIF), or Bone Morphogenetic Protein (BMP) Signalling
Abnormal activation of mTOR signalling is also involved in the growth of leukaemia cells (34). NOTCH activation is known to activate mTOR signaling by down-regulating the transcription of phosphatase and tensin homolog (PTEN), a suppressor of mTOR signaling, through induction of HES1 expression (35). We, therefore, examined the effects of NOTCH knockdown by siRNA on mTOR signaling. NOTCH1 and NOTCH2 knockdown reduced the level of mTOR protein in THP-1 cells. Contrastingly, NOTCH activation by stimulation with NOTCH ligands increased the expression of mTOR in THP-1 cells (27). Thus, we identified a novel link from NOTCH to mTOR signalling in THP-1 cells. Moreover, we examined the effects of an mTOR inhibitor (PP242) on NOTCH signaling in AML and T-ALL cell lines. PP242 treatment up-regulated NOTCH1 and cleaved NOTCH1 in Jurkat cells but suppressed expression and cleavage in DND-41 cells (36). Thus, mTOR activity differentially affect NOTCH signaling in a cell-dependent manner.
HIF-mediated signaling also plays roles in maintenance of both HSCs and leukaemia stem cells under hypoxic environments such as bone marrow niches (37). Two patterns of interaction between HIF and NOTCH signaling were previously reported: i) HIF1α binds to cleaved NOTCH1, which results in its stabilization and the activation of NOTCH signalling (38); and ii) HIF1α represses a negative feedback loop of the NOTCH1–HES1 system by inhibiting HES1 binding to the HES1 promoter, resulting in enhancement of NOTCH signalling (37). We found that treatment with an HIF1 inhibitor, echinomycin, suppressed in vitro growth and induced apoptosis of AML and T-ALL cell lines. Interestingly, echinomycin treatment reduced levels of NOTCH1 and cleaved NOTCH1 in Jurkat and NB4 cells (39). Thus, we found a novel interaction wherein HIF1 inhibition suppresses NOTCH1 expression and activation.
BMP4 signalling also regulates HSCs (40). We showed that stimulation with recombinant BMP4 protein had diverse effects on the growth of AML and T-ALL cell lines but did not affect NOTCH activity (41).
Why Is NOTCH Signaling Not Autonomously Activated in AML Cells?
Our in vitro experiments showed that half of all AML samples expressed NOTCH protein or NOTCH ligand protein. Cells expressing both NOTCH and NOTCH ligands are assumed to be autonomously activated by their own NOTCH ligands; however, cleaved NOTCH1 was not detected by immunoblot analysis in most AML samples. HES1 mRNA expression levels in AML cells are much lower than those in T-ALL cells, as measured by qRT-PCR (42). Moreover, NOTCH knockdown by siRNA did not affect the growth of AML cells (27). It seems that NOTCH signalling is dispensable for the growth of AML cells, at least in vitro, or that any effects must be weak. Potential explanations for lack of autonomous activation are as follows: i) NOTCH ligands expressed on AML cells might be of insufficient quantity or under inappropriate conditions for NOTCH activation; ii) NOTCH and its ligands within the same cells might experience cis-inhibitory interactions (43); iii) there might be defects in NOTCH receptor glycosylation, which is essential for ligand binding (44); and iv) NOTCH-suppressive molecules such as NUMB, F-box and WD repeat domain containing 7 (FBXW7), or some microRNAs might be overexpressed. These possibilities must be examined in future studies.
NOTCH works through a non-canonical pathway in addition to the canonical pathway. The non-canonical signalling pathway antagonizes WNT/β-catenin signalling and is independent of NOTCH ligand-induced cleavage and transcription (45). Although cleaved NOTCH1 and HES1 mRNA are not observed, it is possible that non-canonical NOTCH signalling plays an important role in AML cells.
Is NOTCH Activation Oncogenic or Tumor-Suppressive in AML Cells?
The next step is to determine the roles of NOTCH signalling in AML cells, especially in vivo. As AML cells express considerably high levels of NOTCH, it is probable that NOTCH signalling plays some role, especially in bone marrow, where stromal cells express NOTCH ligands. Whether NOTCH plays an oncogenic or tumour-suppressive role in AML is still controversial, as mentioned above. The diverse views arise, in part, from the varied roles of NOTCH that depend on cell type, condition, and the environment. Notably, the observed effects can be different due to the conditions used in each experiment, especially when comparing in vitro and in vivo studies.
Our in vitro experiments showed that exogenous stimulation with recombinant NOTCH ligands suppresses the long-term growth of AML cells despite transiently promoting the short-term growth of a subset of AML samples. Because NOTCH ligands are expressed on bone marrow stromal cells, NOTCH in AML cells can be activated in vivo. If NOTCH activation suppresses the growth and self-renewal of leukaemia stem cells, cells expressing NOTCH will disappear. So, why do AML cells express NOTCH? Possible explanations for these apparently contradictory observations are as follows: i) although NOTCH activation itself suppresses growth, its effects on other stemness-related signalling might potently stimulate growth and self-renewal in vivo where various signalling pathways and cytokines interact, and ii) the in vitro effects of NOTCH activation might differ from in vivo behaviour because the ligand density on culture plates is different from that expressed on stromal cells. Supporting the latter model, NOTCH ligand density has been reported to affect the cell fate decision (46), and supraphysiological NOTCH signals have deleterious effects on self-renewal of HSC (47).
The in vivo experiments showed that induction of active NOTCH expression in AML cells causes growth arrest and apoptosis in mice (15), which demonstrates that NOTCH activation acts as tumor suppressor. However, genetically-forced NOTCH activation does not necessarily correspond to the true effects of NOTCH activation in bone marrow of patients with AML. It is certain that NOTCH works as a tumor suppressor in chronic myelomonocytic leukaemia because inactivating NOTCH pathway mutations are found in cells from patients with such disease (48). However, such inactivating mutations have not been found in AML cells.
Conclusion
The present review presented various findings regarding the roles of NOTCH signaling in leukaemia cells, especially in AML cells. Although NOTCH is expressed in AML cells, the role of NOTCH, including whether it is oncogenic or tumor suppressive, has not been elucidated. The problems and potential reasons for the diverse interpretations of the role of NOTCH signaling were also presented. Evidence strongly suggests that the roles are diverse in cell type- and case-dependent manners. To develop NOTCH-targeting therapies for AML, further understanding of the roles of NOTCH is required.
Footnotes
-
This article is freely accessible online.
- Received July 16, 2014.
- Revision received September 1, 2014.
- Accepted September 4, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved