To verify molecular mechanisms by which leukemia stem cells (LSCs) maintain a dormant state, we explored the activity of the major prosurvival signal pathways in CD34+/CD38− compartment, supposed to contain LSCs, and CD34+/CD38+ counterparts from patients with acute myelogenous leukemia (AML, n = 11) by fluorescence-activated cell sorting (FACS). CD34+/CD38− cells expressed a greater amount of p-janus kinase 2 (JAK2) and p-signal transducer and activator of transcription 5 (STAT5) than CD34+/CD38+ counterparts in all patients except for one case. In addition, we found that CD34+/CD38− cells were relatively resistant to cytarabine- and the inhibitor of the fms-like tyrosine kinase 3 (FLT3)-mediated growth inhibition, as measured by the clonogenic assay. Interestingly, blockade of JAK2/STAT5 signaling by the specific JAK2 inhibitor AZ960 stimulated cell cycling in CD34+/CD38− cells in conjunction with downregulation of cyclin-dependent kinase inhibitor p21waf1 and sensitized these cells to the growth inhibition mediated by cytarabine and the FLT3 kinase inhibitor. Moreover, exposure of CD34+/CD38− cells to AZ960 potently induced apoptosis in parallel with downregulation of antiapoptotic protein Bcl-xL, as measured by Western blot analysis. Taken together, JAK2/STAT5 signaling may be a promising molecular target to eradicate CD34+/CD38− leukemia cells in individuals with AML.
Acute myelogenous leukemia (AML) is organized as a cellular hierarchy, initiated and maintained by a subset of self-renewing leukemia stem cells (LSCs).1 Intensification of chemotherapy has led to remissions in 70–85% of individuals with AML. Unfortunately, postremission relapses occur frequently.2, 3 Some young patients benefit from hematopoietic stem cell (HSC) transplantation from an human leukocyte antigen HLA-matched donor;4 however, regimen-related toxicities and relapse remain serious problems.
LSCs share some antigenic feature with normal HSCs. For example, both LSCs and HSCs express CD34 but not CD38. However, LSCs can be phenotypically distinguished from HSCs by several disparate markers including CD117− and CD123+.1, 5, 6 LSCs are in a quiescent state and are characterized by capability of self-renewal and differentiation. LSCs are able to perpetuate leukemic cell growth in long-term culture assays, as well as, in the murine nonobese diabetic/severe combined immunodeficiency model system.1, 5–7 In addition, LSCs are likely to be refractory to conventional chemotherapeutic agents such as cytarabine and anthracyclines which interfere with DNA replication and induce apoptosis primarily in replicating cells. It is essential to eradicate LSCs to avoid clinical relapse in patients with AML.
The prosurvival signal pathways such as mitogen-activated protein kinase (MEK)/extracellular signal-regulated kinase (ERK), phosphoinositide-3 kinase (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin (mTOR) and signal transducer and activator of transcriptions (STATs) are frequently activated in AML and are associated with a poorer prognosis.8–12 For example, activated AKT has been found in over 50% of primary AML samples, as measured by Western blot analysis, whose prognosis was worse than those with undetectable AKT.11 In addition, concomitant activation of MEK/ERK and PI3K/AKT pathways in AML was associated with a worse prognosis than those with the activated single pathway.13
STAT family locates downstream of janus kinase (JAK) and mediates cytokine signals. On phosphorylation at tyrosine residues, STATs form dimerization and translocate to nucleus where they bind to DNA and regulate expression of target genes.14, 15 Aberrant expression of inflammatory cytokines such as interleukin-6 resulted in constitutive activation of STATs in AML cells.16 The internal tandem duplication (ITD) of the juxtamembrane domain of fms-like tyrosine kinase (FLT3-ITD) occurs in approximately 30% of AML.17 This mutation constitutively activates FLT3 without ligand binding, resulting in the activation of downstream prosurvival signals including STAT5 and are associated with elevated blast counts, increased relapse rate and poor overall survival.18
The prosurvival signal pathways are thus intimately involved in leukemogenesis and proliferation of leukemia cells. However, activity of these signal pathways in LSCs remains to be elucidated. Our study examined the activity of the major prosurvival signal pathways in CD34+/CD38− cells and CD34+/CD38+ counterparts from individuals with AML.
Material and Methods
The JAK2 inhibitor AZ960 was provided by AstraZeneca (Macclesfield, United Kingdom). The FLT3 kinase inhibitor sunitinib was provided by Pfizer (Kalamazoo, MI). AraC (1-β-D-arabinofuranosylcytosine, cytarabine) was obtained from Sigma (Deisenhofen, Germany).
Leukemia cells were freshly isolated from AML patients with French-American-British (FAB) classification system subtype M0 (case # 5), M1 (case # 1), M2 (case #s 4 and 9), M3 (case #s 8 and 10), M4 (case # 2), and myelodysplastic syndrome was transformed to AML (case #s 3, 6, 7 and 11) after obtaining informed consent with Kochi University Institutional Review Board approval. The informed consent was obtained in accordance with the Declaration of Helsinki. Characteristics of patients are summarized in Table 1. AML cells from case #s 1 and 2 expressed the FLT3-ITD mutation (data not shown).
Table 1. Characteristics of AML patients (n = 11)
CD34+/CD38− cells and CD34+/CD38+ counterparts were purified by magnetic cell sorting using a CD34 MultiSort kit and a CD38 MicroBead kit as recommended by the manufacturer (Miltenyi Biotec GmbH, Germany). CD34+ HSCs were isolated from healthy volunteers (n = 3) by magnetic cell sorting using CD34 MicroBeads as recommended by the manufacturer (Miltenyi Biotec GmbH).
Cell cycle analysis
Cell cycle distribution of CD34+/CD38− cells and CD34+/CD38+ counterparts was measured as previously described.19 Briefly, the cells were stained with Ki-67-fluorescein isothiocyanate FITC (Sigma) and propidium iodide and subjected to fluorescence-activated cell sorting (FACS).
Methylcellulose colony-forming assay
The colony-forming assay was performed with methylcellulose medium H4534 (StemCell Technologies, Vancouver, BC, Canada) as previously described.20
Western blot analysis
Western blot analysis was performed as described previously.21 Anti-p21waf1 (Santa Cruz Biotechnology, sc-6246 Santa Cruz, CA), p-STAT5 (Tyr694) (Cell Signaling Technology, #9351 Beverly, MA), STAT5 (C-17) (Santa Cruz Biotechnology, sc-835), BAX (Santa Cruz Biotechnology, sc-7480), Bcl-2 (Santa Cruz Biotechnology, sc-509), Bcl-xL (Cell Signaling Technology, #2764), cleaved Poly (ADP-ribose) polymerase PARP (Cell Signaling Technology, #9541) and GAPDH (Abcam Tokyo, Japan) antibodies were used.
Phosphoprotein analysis by FACS
The activity of the prosurvival signal pathways was assessed by FACS, as previously described.22 Anti-p-JAK2 (Tyr1007/1008; Cell Signaling Technology, #3771), p-FLT3 (Tyr591; Cell Signaling Technology, #3461), p-p44/42 MAP kinase (T202/Y204; Cell Signaling Technology, #9101), p-AKT (Ser473; Cell Signaling Technology, #9271), STAT5 (C-17; Santa Cruz Biotechnology, sc-835), p-STAT5 (Tyr694; Cell Signaling Technology, #9351) antibodies were used.
Statistical analysis of difference between two groups was assessed by Mann–Whitney U test.
CD34+/CD38− cells were in a dormant state and aberrantly expressed p-JAK2/STAT5
Leukemia cells were isolated from 11 patients with AML (Table 1). CD34+/CD38− cells and CD34+/CD38+ counterparts were purified by magnetic cell sorting, yielding >97% purity (Supporting Information Fig. 1). A greater amount of CD34+/CD38− cells was in a dormant state (G0 phase of the cell cycle) compared to CD34+/CD38+ counterparts (68 ± 15% vs. 16 ± 13%, p < 0.01), as assessed by cell cycle analysis (Fig. 1a).
We hypothesized that differentially activated prosurvival signal pathways in CD34+/CD38− cells might contribute to maintenance of quiescence. To verify this hypothesis, we assessed activity of the major prosurvival signal pathways including JAK2/STATs, MEK/ERK and Akt/mTOR in CD34+/CD38− cells and CD34+/CD38+ counterparts by using the phosphor-specific antibodies against these pathways. There was a trend that CD34+/CD38− cells expressed a greater amount of the phosphorylated forms of prosurvival signal-related proteins than CD34+/CD38+ counterparts (Figure not shown). Especially, the percentage of cells expressing p-JAK2 and p-STAT5 was significantly higher in CD34+/CD38− cells than CD34+/CD38+ counterparts in all cases except for one case, suggesting that JAK2/STAT5 was differentially activated in CD34+/CD38− cells. Western blot analyses also showed that levels of p-STAT5 were higher in CD34+/CD38− cells than CD34+/CD38+ counterparts (Fig. 1c). On the other hand, the phosphorylated forms of STAT5 were not detectable in CD34+ bone marrow mononuclear cells isolated from a healthy volunteer, as assessed by Western blot analysis (Supporting Information Fig. 2). Interestingly, CD34+/CD38− cells highly expressed cell cycle-dependent kinase inhibitor p21waf1 than CD34+/CD38+ counterparts (Fig. 1c).
CD34+/CD38− cells were resistant to AraC and the inhibitor of FLT3 kinase
We compared the sensitivity of CD34+/CD38− cells and CD34+/CD38+ counterparts to antileukemia agents including AraC and the FLT3 kinase inhibitor sunitinib. CD34+/CD38− cells form AML with FLT3-ITD (case # 1) and two other cases with the wild-type FLT3 (case #s 3 and 4) were relatively resistant to sunitinib compared to CD34+/CD38+ counterparts, as measured by colony forming assay (Fig. 2). CD34+/CD38− cells were also resistant to AraC compared to the CD34+/CD38+ counterparts (Fig. 2).
Blockade of STAT5 by AZ960 inhibited the colony forming ability of CD34+/CD38− cells and stimulated cell cycling in association with downregulation of p21waf1
Nearly 90% of CD34+/CD38− cells from case # 1 expressed p-JAK2 (Fig. 3a). Exposure of these cells to AZ960 (0.01–0.1 μM, 3 hr), a novel and specific inhibitor of JAK2 kinase with the additional activity against JAK1 kinase,23 effectively dephosphorylated JAK2 in conjunction with dephosphorylation of STAT5 (Fig. 3a). Approximately 50% of CD34+/CD38− cells from case # 4 expressed p-STAT5, although these cells did not express p-JAK2 at high level (Fig. 3a). CD34+/CD38− cells from case # 4 highly expressed p-JAK1, which probably increased levels of p-STAT5 (data not shown). AZ960 effectively inhibited colony formation of CD34+/CD38− cells from case #s 1 and 4 in a dose-dependent manner with IC50 of 0.06 and 0.02 μM, respectively (Figure not shown). Interestingly, exposure of CD34+/CD38− cells (case #s 1 and 4) to AZ960 (0.01–0.1 μM, 24 hr) decreased the population of cells expressing p21waf1 (Fig. 3b). Moreover, cell cycle analysis found that AZ960 (0.03 or 0.1 μM, 24 hr) decreased the population of CD34+/CD38− cells (case #s 1 and 4) in the G0 phase of the cell cycle in conjunction with increasing the population of cells in the S phase of the cell cycle (Fig. 3c), suggesting that AZ960 stimulated the dormant CD34+/CD38− cells.
CD34+/CD38− cells from AML with FLT3-ITD (case # 1) were refractory to sunitinib-mediated growth inhibition; however, when LSCs were exposed to combination of sunitinib and AZ960, their colony formation was potently inhibited. For example, either sunitinib (30 nM) or AZ960 (0.03 μM) alone inhibited their colony formation by approximately 10% and 40%, respectively (Fig. 3d). Combination of both at the same concentrations inhibited their colony formation by nearly 70% (Fig. 3d). AZ960 more potently enhanced antileukemia effects of AraC in CD34+/CD38− cells. Either AraC (2.5 nM) or AZ960 (0.01μM) alone inhibited the clonogenic growth of CD34+/CD38− cells from case # 1by approximately 15% or 30%, respectively. When both agents were combined at the same concentration, their growth was inhibited by nearly 60% (Fig. 3d). Likewise, AraC-mediated growth inhibition of CD34+/CD38− cells from case #s 4–6 was potently enhanced in the presence of AZ960 (Fig. 3d).
AZ960 induced apoptosis of CD34+/CD38− cells in association with downregulation of Bcl-xL
Furthermore, we found that inhibition of STAT5 by AZ960 induced cleavage of PARP, a marker of apoptosis, in parallel with downregulation of Bcl-xL and upregulation of proapoptotic protein BAX in CD34+/CD38− cells (Fig. 4).
Our study compared the activity of prosurvival signal pathways in CD34+/CD38− leukemia cells and CD34+/CD38+ counterparts in the patients with AML (n = 11). Interestingly, CD34+/CD38− cells expressed a greater amount of p-JAK2 and p-STAT5 than CD34+/CD38+ counterparts (Fig. 1b,c), suggesting that JAK2/STAT5 signaling was differentially activated in CD34+/CD38− cells. Further studies are required to verify molecular mechanisms, which contribute to activation of this signal pathway.
STAT5 was constitutively activated in 18 out of 26 (69%) individuals with AML, mainly because of the gain-of-function mutation in the receptor tyrosine kinases such as FLT3 and c-KIT or autocrine growth factor production.24 STAT5 was also activated in CD34+AML cells from 7 out of 13 patients (53%),24 as measured by Western blot analysis. Knockdown of STAT5 in CD34+AML cells by a short hairpin RNA inhibited their proliferation, indicating the important role of STAT5 in the growth of CD34+AML cells.24
STAT5 was shown to be involved in maintenance of self-renewal capacity in murine HSCs; forced-expression of constitutively active STAT5A in murine HSCs (CD34−KSL) produced a drastic expansion of multipotential progenitors and promoted HSC self-renewal in ex vivo.25 In addition, the investigators demonstrated that activation of STAT5A in HSCs (CD34−KSL), but not in CD34+KSL progenitor cells caused myeloproliferative diseases in vivo.25 Very recently, other investigators have shown critical roles of STAT5 in maintenance of quiescence in murine HSCs. Conditional deletion of STAT5 stimulated cell cycling in HSCs and reduced the long-term HSC pool.26 Interestingly, host deletion of STAT5 in conditional knockout mice permitted efficient donor long-term engraftment of HSCs without ablative conditioning.26
Our study found that human CD34+/CD38− cells were relatively resistant to AraC as well as sunitinib compared to CD34+/CD38+ counterpart (Fig. 2). In addition, we, for the first time, demonstrated that blockade of STAT5 by the specific inhibitor of JAK2 kinase AZ960-stimulated cell cycling in CD34+/CD38− cells in association with downregulation of p21waf1 and sensitized them to AraC- or sunitinib-mediated growth inhibition (Fig. 3b–d).
Notably, CD34+/CD38− cells aberrantly expressed antiapoptotic protein Bcl-xL, a transcriptional target of STAT527 (Fig. 4). It is likely that Bcl-xL contributes to the drug-resistant character of CD34+/CD38− cells. Importantly, AZ960 potently downregulated levels of Bcl-xL and induced apoptosis in CD34+/CD38− cells (Fig. 4).
FLT3 is one of the attractive molecular targets for treatment of AML expressing FLT-ITD. Unfortunately, our study found that CD34+/CD38− cells were relatively resistant to sunitinib-mediated growth inhibition (Fig. 2), mainly because most of CD34+/CD38− cells were in a dormant state in association with upregulation of Bcl-xL and p21waf1. This might be one of the reasons why response of AML patients to the inhibitor of FLT3 kinase was partial and of short duration in a clinical trial.28
Recent studies showed that anti-CD38 antibodies inhibited the engraftment of cord blood as well as leukemia cells in immunodeficient mice.29 In addition, the CD34− compartment was shown to contain LSCs in the cases with AML possessing the mutated nucleophosmin gene.30 These observations suggested that LSCs were more heterogeneous than previously recognized, and LSCs might reside in CD34+/CD38+ or even in the CD34− compartment in some cases. However, a greater amount of CD34+/CD38− cells was in a dormant state than CD34+/CD38+ counterparts in all cases examined (data not shown). Thus, we believe that LSCs were enriched in CD34+/CD38− compartment in our study.
Taken together, STAT5 was differentially activated in CD34+/CD38− leukemia cells. The inhibitor of JAK2 kinase inhibited STAT5, stimulated cell cycling in CD34+/CD38− cells and sensitized them to growth inhibition mediated by AraC or the FLT-3 kinase inhibitor. In addition, inhibition of STAT5 by AZ960 potently induced apoptosis of CD34+/CD38− cells in association with downregulation of Bcl-xL. STAT5 may be an attractive molecular target to develop curative treatment strategy for AML, although the sample number was limited and background of each patient was heterogeneous in our study. Further studies are clearly required to confirm these results.
This work was supported in part by The Kochi University President's Discretionary Grant (to T.I.) and Setsuro Fujii Memorial, the Osaka Foundation for Promotion of Fundamental Medical Research (to T.I). C.N. is grateful for a JSPS Research Fellowship for Young Scientists from the Japan Society for the Promotion of Science.