Oncogenic Kit Triggers Shp2/Erk1/2 Pathway to Down-Regulate the Pro-Apoptotic Protein Bim and to Promote Apoptosis Resistance in Leukemic Cells

Oncogenic mutations leading to persistent kinase activities are implicated in various human malignancies. Thereby, signaling pathway-targeted therapies are powerful customized treatment to eradicate cancer cells. In murine and human leukemia cells harboring mutations in Kit, we previously showed that distinct and independent pathways controlled resistance to apoptosis or cell cycle. A treatment with PI3Kinase inhibitors to reduce cell proliferation combined with inhibitors of Erk1/2 activity to promote apoptosis had synergistic effects allowing eradication of leukemia cell growth. We reported here that BimEL, a pro-apoptotic member of the Bcl2 family proteins, is the target of Erk1/2 signaling and that its down-regulation is responsible for the apoptosis resistance of murine and human leukemic cells. Downstream of Kit mutant, the tyrosine phosphatase Shp2 maintains BimEL expression at a low level, through Erk/2 activation and proteosomal BimEL degradation. This process is controlled by Shp2 independently of other signaling pathways activated downstream of oncogenic Kit, demonstrating that Shp2 is a key regulator of Bim expression in the context of an oncogenic signaling. The increase in BimEL expression is associated to an increased apoptosis. Moreover, the depletion of Bim overcomes apoptosis associated with Erk1/2 inactivation in UO126-treated leukemic cells, thereby establishing the contribution of Bim to drug-induced apoptosis. These data provide a molecular rationale for using BH3 mimetics in combination with PI3K inhibitors to treat leukemia, especially in the case of an oncogenic signaling refractory to Tyrosine Kinase inhibitors.


Introduction
Oncogenic mutations in the receptor tyrosine kinase for stem cell factor (Kit) underly the development of a variety of neoplasms including leukemia. Activating Kit mutations generate many perturbations in various signaling networks and, thus, understanding their contribution to the malignant phenotype may provide a molecular rationale to design pathway-targeted therapies. This might be a key issue for patients with Kit-driven neoplasms that have a high probability of developing resistance to Tyrosine Kinase inhibitors.
In the spi-1/PU.1-transgenic mouse model, activating Kit mutations drive malignant transformation during the erythroleukemic process. In this model, the founding oncogenic event is the inappropriate overexpression of Spi-1/PU.1, a master transcriptional regulator of B lymphopoiesis and myelopoiesis that leads to inhibition of terminal erythroid differentiation and an extensive proliferation of proerythroblastic cells (preleukemic cells) [1]. Later on, the blastic crisis is associated with occurrence of somatic mutations in Kit that confer growth factor autonomy and tumorigenicity to proerythroblastic cells (leukemic cells: so-called HS2 cells) [2,3]. The Kit mutations target residues D814 or D818 in the phosphotransferase domain, which are homologous to human residues D816 or D820 mutated in patients with mastocytosis or AML. Our studies on the biochemical networks connecting Kit mutant to leukemogenesis have demonstrated that independent signaling pathways contribute to malignancy: cell survival is driven by the Kit/Shp2/Ras/Mek/Erk1/2 pathway whereas G1/S transition during cell cycle is accelerated by both the Kit/Stat5 and Kit/PI3K/Akt pathways [4]. Importantly, eradication of leukemic cells was only achieved with the concomitant inhibition of cell cycle and survival. This allowed to define drugs combinations for leukemia therapy. Indeed, the combined use of two clinically relevant compounds NVP-BEZ235 [5], an inhibitor targeting all isoforms of the PI3-Kinase as well as mTOR and Obatoclax [6], an antagonist of pro-survival Bcl2 factors, had strong synergistic effects to eradicate the growth of leukemic cells [4].
The activity of the mitochondrial pathway that controls apoptosis results from the coordination of interacting proapoptotic and anti-apoptotic proteins of the Bcl2 family [7]. An imbalance in this interplay can lead to a survival advantage that promotes neoplasm development. For example, amplification of the Mcl-1 gene is a frequent somatic event in human cancer [8] and genetic alterations that activate anti-apoptotic proteins of the Bcl2 family such as Bcl2 or Mcl-1 occur in hematopoietic malignancies [9]. Otherwise, somatic mutations in the proapoptotic factor Bax confer resistance to apoptosis in solid and hematopoietic tumors [10]. In addition, homozygous deletions of the pro-apoptotic Bim gene are identified in lymphoma [11].
Bim is a BH3-only protein of the Bcl2 family that is essential for apoptosis induced by growth-factor deprivation in a broad range of cell types [7]. Bim expression is subjected to different modes of regulation at both transcriptional and post-translational levels [12], that are governed by various signaling pathways, including the Erk1/2 and PI3K/Akt pathways [13].
In the present study, we searched whether proteins of the Bcl2 family are mediators of the apoptosis resistance of leukemic cells. We show that Erk1/2 activation induces the phosphorylation and the down-regulation of Bim EL via proteasomal degradation. This process was controlled by the Tyrosine phosphatase Shp2 independently of other signaling activated by Kit mutant showing that Shp2 as an important regulator of Bim expression in the context of an oncogenic signaling. The increase in Bim EL expression was associated with an increased apoptosis and the depletion of Bim overcomes apoptosis associated with Erk1/2 inactivation demonstrating that Bim EL is a crucial mediator of apoptosis resistance in leukemic cells. These functions of Bim EL make relevant the association of BH3 mimetics with PI3K inhibitors to eradicate the leukemic growth and provide a molecular basis to improve the response obtained with the combined inhibition of signals controlling survival and cell cycle.

Cell Lines and Inhibitors
HS2 cell lines were established from the spleen of leukemic spi-1-transgenic mice as previously described [1]. The human mast cell leukemia subclone HMC-1.2 [2,3] harboring Kit V560G and Kit D816V was kindly provided by Dr JH Butterfield (Mayo Clinic, Rochester, MN). Murine and human cells were cultured in alpha minimun essential medium (aMEM, GibcoBRL) supplemented with 10% fetal calf serum (FCS, GibcoBRL).
Knocking-down of Stat5 in 606HS2 and 931HS2 cells was performed through cell infection with a lentiviral vector encoding GFP as reporter gene and transducing Stat5-shRNA17 or control ns-shRNA. 48 h after the start of infection, infected HS2 cells were sorted for GFP expression, using a FACSVantage (Becton Dickinson), as previously described [4]. A lentiviral vector encoding an anti-luciferase shRNA was used as control.
Overexpression of Shp2 in 606HS2 and 931HS2 cells was performed through infection with retroviruses transducing the wild-type Shp2-MT or the Shp2 C459S -MT mutant and then selected for puromycin resistance. The cDNA encoding murine Shp2 WT was cloned in frame with a tag-myc epitope in C-terminus into pMSCV retroviral vector (Shp2 WT -MT). The Shp2 C459S mutant (Shp2 C459S -MT) was generated by mutagenesis of the wild-type Shp2 cDNA using the QuichChange Site-Directed Mutagenesis System (Stratagene, LaJolla, CA). The empty vector (c) was used as control.

Cell Death and Cell Apoptosis
The percentage of dead cells was evaluated by trypan blue exclusion using a Vi-Cell analyzer (Beckman Coulter). The percentage of apoptotic cells was analyzed by labeling the cells with cleaved caspase3-antibody. Cells were fixed with PFA1%, permeabilized with 70% cold ethanol and stained with phycoerythrin(FITC)-conjugated anti-active caspase3 antibodies (Becton Dickinson). Flow cytometry was performed with FACSsort (Becton Dickinson) and data were analyzed using Flow Jo (Treestar).

Results and Discussion
Bim EL is the Pro-apoptotic Factor Down-regulated by Erk1/2 Activity in Leukemic Cells We have recently shown that apoptosis resistance of HS2 leukemic cells is induced by the activation of the Erk1/2 pathway [4]. To go further we investigated which mediators of the intrinsic apoptotic pathway were targeted by Erk1/2 activity. In diverse cell types, activation of Erk1/2 signaling can lead either to the upregulation of pro-survival proteins of the Bcl-2 family, notably Bcl-2, Bcl-x L and Mcl-1, or to the decrease or inactivation of proapoptotic proteins such as Bim EL to achieve cell survival [13]. We analyzed the expression of the pro-survival factors Bcl-2, Bcl-x L , Mcl-1, the apoptotic mediators Bax, Bak and Bim in HS2 leukemic cells in response to a treament with the Mek inhibitor UO126 as an apoptosis inducer. Immunoblot analyses showed that expression levels of Mcl-1, Bcl-x L , Bax and Bak were not changed in cells treated with UO126 compared with untreated cells ( Figure 1A). Bcl-2 expression was not detectable in HS2 cells (data not shown). Bim is expressed as three major isoforms generated by alternative splicing: a short, long and extra-long protein named Bim S , Bim L and Bim EL [16]. In HS2 cells, Bim EL was the most abundant isoform expressed as two bands, expression of Bim L was modest while Bim S was undetectable. The treatment of HS2 cells with UO126 was associated with an increased expression of the lower band of Bim EL . Erk1/2 phosphorylates some serine residues on Bim EL [17,18] and an antibody that is specific for Bim EL phosphorylated at Ser 65 recognizes the Bim EL upper band. This phosphorylated form was undetectable when HS2 cells were treated with UO126 whereas the total amount of Bim EL was increased ( Figure 1A). Thus, both phosphorylation and expression level of Bim EL were dependent on the activity of Erk1/2.
Given that Erk1/2 activation depends on Kit mutant activity in leukemic cells, we controlled that Bim EL expression was changed in response to the inhibition of Kit. The Kit D818Y kinase activity in 931HS2 cells is sensitive to IM (Imatinib Mesylate or GleevecH), PP1 and PP2 whereas Kit D814Y kinase activity in 606 HS2 cells is sensitive to PP1 and PP2 but resistant to IM [15]. Treatment of 931HS2 cells with the Kit inhibitors IM, PP1 and PP2 abolished Bim EL phosphorylation and promoted an increase in Bim EL unphosphorylated form, a result comparable to that observed in response to UO126 treatment ( Figure 1B). This treatment did not change Mcl-1 and Bcl-x L expression. Treatment of 606HS2 cells with PP1 and PP2 but not with IM abolished Bim EL phosphorylation as well as Erk1/2 activation and promoted an increase in Bim EL unphosphorylated form. Thus, both phosphorylation and expression levels of Bim EL depend on the activation of Erk1/2 downstream of Kit mutant in HS2 leukemic cells.
These results were confirmed with the human HMC-1.2 cell line derived from an AML patient with a mast cell leukemia. These cells express mutant Kit D816V and the mutant Kit D816 in human, like mutant Kit D814 in mouse, promotes survival of leukemic cells through activation of the Kit/Shp2/Mek/Erk pathway [2,4]. We analyzed in HMC-1.2 cells whether Bim phosphorylation was dependent on either Kit or Erk1/2 activities. In cells treated with PP1, PP2 or UO126, an increase in the expression of Bim EL was observed when the Erk1/2 activation was abolished, whereas no change was detected in the expression of Bcl-2, Bcl-x L and Mcl-1 ( Figure 1C). In agreement with the resistance of Kit D816V to IM [2,4], neither expression of Bim EL nor activation of Erk1/2 was modified by IM treatment. These results indicate that the ability of activated Erk1/2 to repress Bim expression downstream of Kit mutant is faithfull reproduced in human cell line.

The Tyrosine Phosphatase Shp2 Controls the Downregulation of Bim
Besides the Erk1/2 pathway, Stat5 and PI3K/Akt pathways are activated downstream of Kit mutant, all cooperating to the extensive proliferation of HS2 leukemic cells [4]. To clarify whether Bim expression was specifically dependent on Erk1/2 activity, we investigated whether Shp2 was a regulator of Bim expression. Indeed, our previous data have shown that Shp2 was the proximal effector of Kit mutant in activating Erk1/2 and mediating apoptosis resistance in leukemic cells [4]. We first examined Bim EL expression in response to Shp2 knock-down by RNA interference. HS2 cells were infected with two lentiviral vectors transducing independent Shp2 short hairpin RNAs (shRNAs 77 or 78). Both shRNAs induced a major decrease (90-95%) in Shp2 protein level compared to a non silencing shRNA. Shp2 depletion abolished Erk1/2 activation and induced apoptosis as deduced from the detection of cleaved caspase-3 on immunoblotting (Figure 2A). Notably, Shp2 depletion was associated with an increase in the unphosphorylated form of Bim EL .
Conversely, we examined Bim EL expression in response to Shp2 overexpression. HS2 cells were infected with a retrovirus transducing a Myc-tagged wild-type Shp2. The overexpression of wild-type Shp2 resulted in an increase in Erk1/2 activation (2fold for 931HS2 cells and 3-fold for 606HS2 cells) and was associated with major changes in the pattern of Bim EL expression ( Figure 2B). Phosphorylated Bim EL levels were increased in transduced HS2 cells compared to control HS2 cells. To further confirm the role of Shp2 activity in the phosphorylation of Bim EL , the catalytically-inactive Shp2 C459S mutant [19] was overexpressed in HS2 cells through infection with a retrovirus transducing Myc-tagged Shp2 C459S . Overexpression of this mutant induced changes resembling to those induced by Shp2 depletion i.e. a disappearance of phosphorylated Bim EL , an increase in unphosphorylated Bim EL and a major reduction in Erk1/2 activation associated with a marked cleavage of caspase 3 ( Figure 2B). These observations indicate that Shp2 phosphatase activity controls Bim EL expression in leukemic cells through its ability to induce Erk1/2 activation.
The PI3K/Akt pathway can also regulate Bim EL expression. In particular, a loss of Akt activation can lead in turn to activation of FoxO3A transcription factor, resulting in an increased transcription of Bim [20]. However, no change in Bim EL expression was detectable when HS2 cells were treated with the PI3K inhibitor NVP-BEZ235 ( Figure 2C). Similarly, no change in Bim EL expression was seen in HMC-1.2 cells following a treatment with the PI3K inhibitor NVP-BEZ235 ( Figure 2C), whereas Akt phosphorylation was abolished (Figures 1C and 2C). Thereby, the expression of Bim was not influenced by the PI3K/Akt activity, a finding consistent with the absence of cross-talk between PI3K/Akt and Shp2/Mek/Erk signaling pathways in these cells  [4]. Though Stat5 signaling did not contribute to apoptosis resistance of leukemic cells, we also examined whether Stat5 could modulate Bim expression. Bim expression was compared in HS2 cells infected either with a lentivirus encoding a short hairpin RNAs for Stat5 or a control non silencing shRNA. As shown in Figure 2D, Stat5 depletion did not affect Bim expression.
Altogether, these findings highlight the function of Shp2 as a major player in controlling Bim expression in the context of an oncogenic signaling.

Knockdown of Bim Rescues HS2 Leukemic Cells from UO126-induced Apoptosis
To go further in the relevance of the down-regulation of Bim EL in response to Erk1/2 activation, we studied whether extinction of Bim may overcome the mortality induced by Erk1/2 inactivation in HS2 cells. HS2 cells were infected with lentiviruses transducing independent Bim sh-RNAs (shRNAs 92 or 94). Both induced a major extinction in Bim expression (90 to 95%) without detectable alterations in the expression of Mcl-1, Bcl-x L , Erk1/2 and Akt ( Figure 3A). The Bim-shRNA94-transduced 606HS2 and 931HS2 cells and control cells were treated during 48hours with UO126 at 10 mM or 20 mM. We controlled that Bim depletion was maintained during UO126 treatment ( Figure 3B). At the dose of 10 mM UO126 corresponding to IC 50 (as previously published [21]), a residual activation of Erk1/2 was detected, which was totally abolished when cells were treated with 20 mM UO126 ( Figure 3B). Treatment of both ns-shRNA-HS2 cells and Bim-shRNA-94-HS2 cells with 20 mM UO126 resulted in a cell death that was more marked than after treatment with 10 mM UO126 ( Figure 3C). This observation was consistent with the residual Erk1/2 activation seen with 10 mM of UO126. Remarkably, when cells were transduced with Bim shRNA (Figure 3B), the number of dead cells associated with UO126 treatment was reduced: 25% or 54% decrease with 10 mM or 20 mM UO126 in Bim-shRNA-94-931HS2 cells and 52% or 50% decrease with 10 mM or 20 mM UO126 in Bim-shRNA-94-606HS2 cells compared with ns-shRNA-HS2 cells.
To go further, the number of apoptotic cells as detected by flow cytometry analysis of active caspase-3 was determined in Bim- shRNA-94-HS2 and ns-shRNA-HS2 cells treated with 10 mM or 20 mM UO126 ( Figure 3C). The percentage of apoptotic cells (cleaved caspase-3 positive) was significantly decreased compared to controls (40% or 65% decrease with 10 mM and 20 mM UO126 in 931HS2 cells; 43% or 34% decrease with 10 mM and 20 mM UO126 in 606HS2 cells). We observed similar results when experiments were conducted with Bim-shRNA-92-HS2 cells (data not shown). Thus, depletion of Bim did significantly overcome apoptosis induced by UO126 treatment indicating the contribution of Bim to drug-induced apoptosis.
Intriguingly, the level of both spontaneous apoptosis and UO126-induced apoptosis did not appear to be related to the expression level of Bim EL . Indeed, while 931 and 606 HS2 cells expressed different amount of Bim EL , the rate of spontaneous or UO126-induced apoptosis was quite similar ( Figure 3C). To rule out the possibility that the high level of Bim EL in 606HS2 cells expressing mutant Kit D814Y reflected a cell line particularity, we analyzed Bim EL expression in ten independent HS2 cell lines all carrying mutant Kit D814Y . Bim EL expression level was reproducibly higher in these cells than that observed in 931HS2 cells expressing mutant Kit D818Y (data not shown). Thus, our data do not rule out that only a fraction of Bim EL is involved in the apoptotic process in 606HS2 cells. However, the differences in Bim EL expression might reflect differences in signaling downstream of Kit D818Y in 931HS2 cells and Kit D814Y in 606HS2 cells, producing different requirements for Bim during apoptosis.

Bim Level is Regulated by a Post-translational Mechanism
The mechanisms for regulating Bim expression vary according to the apoptotic stimulus and the cell type. One mechanism initiated in response to Erk1/2 activation is the proteasomal degradation of the phosphorylated form of Bim EL [17,22]. To check this possibility, we treated HS2 cells with the proteasome inhibitor MG132 and monitored expression of Bim EL by immunoblotting. Exposure of HS2 cells to MG132 (2 or 4 hours) resulted in a time-dependent increase of both phosphorylated and unphosphoryated Bim EL forms ( Figure 3D). Notably, a major increase in cleaved caspase-3 was detected paralleling the increase in Bim EL amount in cells treated with MG132 for 4 hours. Altogether, these data show that proteosomal degradation is the main process leading to maintain a low level expression of Bim in the leukemic cells and that impeding Bim down-regulation is associated with apoptotic signals in the leukemic cells. This mode of Bim regulation is in agreement with the down-regulation of Bim caused by different oncogenic kinases in different malignant hemopathies, such as BCR-ABL in chronic myeloid leukemia [17,23], Kit mutant in systemic mastocytosis [24] and JAK2 V617F in myeloproliferative disorders [25].
Bim is a BH3-only protein discovered for its ability to interact via the BH3 domain with the prosurvival proteins like Bcl2 or Mcl1 [16,26], thereby being a direct antagonist of pro-survival proteins. This led to design BH3 mimetics compounds capable to neutralize the function of pro-survival proteins of Bcl2 family and usable in therapy. In hematological malignancies, BH3 mimetics are undergoing clinical trials [6]. HS2 cells are sensitive to Obatoclax [4], but resistant to ABT-737(data not shown). Obatoclax and ABT-737 are BH3 mimetics that differ in their ability to target pro-survival proteins. Obatoclax binds to all Bcl2 members (Bcl2, Bxl-x L , Bcl W , Mcl-1 and A1) [27] whereas ABT-737 fails to bind to Mcl-1 and A1 [28]. Since HS2 cells do not express A1, it is attractive to consider that Mcl-1 is the most likely interactor of Bim EL . Since the expression of Mcl-1 is not sensitive to inhibition of Erk1/2 and Kit in cells harboring mutant Kit D814Y (606 HS2 cells) or Kit D818Y (931 HS2 cells) ( Figure 1A, 1B), the regulation of Bim EL expression level appears as the key mechanism that creates the imbalance between prosurvival and proapoptotic activities in Kit-driven leukemia. This underscores that all strategies to mimic Bim activity are valuable therapeutic approaches to abolish the aberrant survival of leukemic cells. However, though Obatoclax is efficient to induce apoptosis in murine HS2 and human HMC-1.2 cells, Obatoclax alone is not sufficient for suppression of leukemic cells [4]. The association of Obatoclax with a PI3K inhibitor such as NVP-BEZ235 to reduce cell proliferation, is required for a complete eradication of leukemic growth. Our present data provide a rationale to this combination of drugs. They highlight that using BH3 mimetics in combination with PI3K inhibitors is a promising strategy to treat leukemia, especially in the case of an oncogenic signaling refractory to Tyrosine Kinase inhibitors.