Cell cycle-dependent activity of the novel dual PI3K-MTORC1/2 inhibitor NVP-BGT226 in acute leukemia

The PI3K-AKT signal transduction pathway is frequently activated in acute leukemias (recently reviewed by Polak and Buitenhuis [29]). Moreover, mice transplanted with AKT-activated hematopoietic stem cells develop acute leukemia, indicating the leukemogenic potential of an activated PI3K/AKT pathway [9].

Maximal activation of AKT results from the phosphorylation of threonine and serine residues at positions 308 (Thr) and 473 (Ser). We addressed whether AKT is activated in acute leukemia and evaluated phospho-AKT expression levels of native acute leukemia blood and/or bone marrow samples (total n=62) collected from adult patients with newly diagnosed AML or mixed phenotype and lymphoblastic leukemia.

A flow cytometry-based intracellular immunostain was set up to assay for Thr308 and Ser473 phosphorylation patterns in native leukemia blasts. In addition, phospho-AKT expression levels of physiologic hematopoietic blasts derived from healthy blood and bone marrow donors (n=12) were determined. Relative ratios compared to unspecific IgG-staining were calculated and normalized to the median expression level of the healthy donor cohort as shown in Figure 1.

In contrast to the healthy donor cohort, where phospho-AKT expression levels clustered around 1 (1.0 for Ser472 and 0.97 for Thr308) on a normalized relative expression level scale (standard deviation 0.3 each), acute leukemia specimens were frequently found to have augmented phosphorylation patterns of AKT. Phosphorylation levels for both Ser473 as well as Thr308 thereby revealed wide expression variance ranging from sheer absence to ~17-fold increase of phosphorylation levels in leukemia samples compared to the donor cohort. Mean expression levels in the leukemia cohort were statistically significantly higher, with an approximately 2-fold elevation of both Ser473- (p = 0.007) as well as Thr308-phosphorylation (p = 0.005) compared to the healthy donor controls in a student’s t-test.

Notably, strongly phosphorylated specimens were exclusively found in the acute leukemia cohort (≥ two-fold expression above normal was found in ~30% of all tested samples compared to 0% in the donor cohort).

Subanalysis of leukemia blasts derived from bone marrow aspirates (n=23) versus peripheral blood specimens (n=39 (Ser473) or n=38 (Thr308)) revealed no significant difference of phospho-AKT expression at codon Thr308 (p = 0.06) as well as Ser473 (p = 0.09).

Comparative analysis of expression levels with leukemia subclassifications, chromosomal or gene mutation status, leukocyte count, age or gender did not reveal a strong correlation between AKT phosphorylation levels and clincial parameters. This is in contrast to previous reports demonstrating a positive association of Thr308 phosphorylation with high-risk cytogenetics and poor prognosis [30] (patient characteristics are provided with Additional file 1: Table S1 with the online version of the article).

Our findings of frequent and augmented phosphorylation of AKT in acute leukemia samples suggest that the AKT pathway is (auto) activated and may provide a promising target for directed therapeutics:

Using Jurkat cells, a PTEN-deficient acute lymphoblastic leukemia cell line rendering AKT signaling pathways autoactivated [31], we now provide evidence that NVP-BGT226 is capable of inhibiting oncogene-driven PI3K/AKT/MTOR signal transduction pathways in acute leukemia.

To better compare efficacy in the context of established compounds, we co-investigated the dual PI3K/MTOR inhibitor NVP-BEZ235. This compound has recently been tested to have significant activity against native leukemia cells [32].

Cell lysates extracted from Jurkat cells treated with NVP-BGT226 or NVP-BEZ235 were immunoblotted together with various phospho-AKT control lysates (treated with the pan-PI3K inhibitor LY294002 or the specific MTORC1 inhibitor rapamycin).

The western blot experiment provided with Figure 2A reveals, that dual inhibition of PI3Kinases and MTOR1/2 complexes by NVP-BGT226 consecutively inhibits serine (S473) as well as threonine (T308) phosphorylation of AKT. Moreover, inhibition of AKT activity leads to potent dephosphorylation of known downstream targets such as p70S6K and retinoblastoma protein (RB) (required for cell growth and G1 cell cycle progression) (reviewed in Panwalkar et al. [33]), ULK1 (a key initiator of MTOR-mediated autophagy when dephosphorylated) [34] and increased cleavage of caspase 3 (a global marker for activated apoptosis cascades).

While similar potency to inhibit S473-AKT and p70S6Kinases was observed for NVP-BGT226 as well as NVP-BEZ235 – the capacity to mediate T308-AKT and RB dephosphorylation as well as cleavage of caspase 3 was more pronounced for NVP-BGT226 compared to NVP-BEZ235.

Suppression of PI3K-AKT-MTORC1/2 signal transduction did translate into a potent antiproliferative effect for both dual PI3K/MTOR inhibitors (Figure 2B) – with similar potency in the lower nanomolar range (IC50s were calculated by linear regression analysis using dose-effect plots).

Surprisingly, a strong discrepancy was noticed for the proapoptotic potential of these two inhibitors. Potent induction of apoptosis was observed for NVP-BGT226, while in contrast, virtually any meaningful proapoptotic effect was measured for NVP-BEZ235 in an annexin V-based assay (Figure 2C). This observation is consistent with immunoblot findings of reduced cleavage intensity of caspase 3 in NVP-BEZ235 treated cells.

To expand our studies to other oncogene-driven AKT-activated leukemia cell models, we chose leukemia cell lines with known gain-of-function tyrosine kinase mutations, which are prevalent in 30-40% of patients with AML (FLT3 and KIT) or ALL (BCR-ABL1 and FLT3) [2]: The acute monocytic leukemia cell line MOLM14 (harboring a FLT3 ITD mutation) and the CML blast crisis cell line K562 (harboring a BCR-ABL1 fusion transcript mutation) were exposed to NVP-BGT226 in a dose dependent manner and inhibition of cellular proliferation was determined. In addition, efficacy of NVP-BGT226 was directly compared to NVP-BEZ235. Both inhibitors proved to be highly sensitive with estimated IC50s in the lower nanomolar ranges (<100 nM) for both cell lines (Figure 3A).

When looking at the capacity to induce apoptosis in these leukemia cells, NVP-BGT226 proved to be a strong inducer of programmed cell death in both cell lines. However, estimated IC50s were considerably higher compared to the antiproliferative capacity (Figure 3B).

Interestingly, when treating cells with NVP-BEZ235 only a minor proportion of cells underwent apoptosis with IC50s that were not reached up to doses of 10 000 nM.

The obvious discrepancy of NVP-BGT226 and NVP-BEZ235 to induce apoptosis – while both agents are highly sensitive with regard to inhibition of cellular proliferation, lead us to hypothesize that divergent cell cycle effects may be the reason for this observation.

We treated MOLM14 and K562 cells with ~IC50 doses of NVP-BGT225 (500 nM) or the 2-fold dose of NVP-BEZ235 (1000 nM) and set up time-dependent cell cycle analysis by PI-stain flow cytometry. Accumulation of cells in the G1/G0, S or G2/M phases was monitored 6, 24 and 72 hours after application of either agent.

Of interest, NVP-BGT226 produced a shift of cells from G2/M and S-phase to the G1/G0 phase – but also markedly increased the proportion of a sub-G0/G1 fraction, indicating dead/apoptotic cells, with a proportion of 50% (MOLM14, Figure 3C) and 41% (K562, Figure 3D) 72 hours after treatment. In contrast, NVP-BEZ235 lead to profound und sustained accumulation of cells in the G0/G1 phase – with only 19% (MOLM14) and respectively 13% (K562) of cells rendering into the sub-G0/G1 fraction after 72 hours of incubation.

Even more, when using high doses (i.e. 10 000 nM), which kill virtually all cells exposed to NVP-BGT226, strong accumulation of MOLM14 as well as K562 cells within the G1/G0 fraction was observed for NVP-BEZ235-treated cells (sub-G1/G0 fractions of only 35% (MOLM14) and 17% (K562)). This observation argues for a potent and sustained cell cycle arrest caused by NVP-BEZ235 in these cell lines.

For validation purposes, we set up immunoblotting experiments using whole cell lysates extracted from MOLM14 or K562 cells treated with either NVP-BGT226 or NVP-BEZ235 (Figure 4). For comparative analysis, additional lysates from cells treated with an ABL1 or FLT3 tyrosine kinase inhibitor (imatinib for K562 BCR-ABL1 cells, sunitinib for MOLM14 FLT3 ITD cells) as well as rapamycin were used.

NVP-BGT226 as well as NVP-BEZ235 potently suppressed phosphorylation of AKT at Ser473 as well as Thr308. As expected, these compounds did not affect phosphorylation of FLT3 or ABL1 tyrosine kinases, nor did they affect phosphorylation patterns of MAPkinases (ERK1/2) or STAT5, which are known downstream signaling targets activated by oncogeneic TK mutations such as FLT3 ITD or BCR-ABL1.

It has to be noted, that basal phosphorylation levels of T308-AKT in MOLM14 and K562 cells were relatively weak to absent – which will be discussed later in more detail using an isogenic Ba/F3 mutant-TK model.

We furthermore probed for downstream signaling targets of AKT: Activation of autophagy cascades (via ULK1) and decreased cell cycle progression in G1 (via dephosphorylation of p70S6K and RB) was similarly seen for both agents – and correlated best with dephosphorylation of AKT at Ser473. In contrast, only NVP-BGT226 treated cells managed to override halt of cell growth and induction of autophagy to induce apoptosis in a cell cycle independent manner as indicated by increased cleavage activity at caspase 3 in both tested cell lines.

The western blot experiments hereby support the findings taken from the cell-based assays for cellular proliferation and induction of apoptosis for both agents.

On a side note, comparative analysis of a specific MTORC1 inhibitor (rapamycin) revealed consecutive dephosphorylation of p70S6K – but no concomitant meaningful inhibition of ULK1 or RB phosphorylation, no cleavage of caspase 3 and no effect on FLT3 or ABL1 signaling in the tested dose. Importantly, rapamycin did not suppress AKT phosphorylation – but activates AKT via a negative feed back loop mechanisms as previously reported [24, 26]. This may counteract clinical efficacy of single MTORC1 inhibition.

For TKI-treated cells we confirmed potent inhibition of the corresponding tyrosine kinase, as well as downstream signaling pathways including MAPKinases, STATs as well as AKT [35, 36]. However, dephosphorylation of the AKT pathway was less pronounced compared to STAT5 or ERK1/2 inhibition, leaving downstream signals (partially) phosphorylated. This observation argues for a potential rescue mechanism of TKI monotherapy, which may be overridden by combination approaches: As indicated in our immunoblot panel, a combination of TKI with PI3K/AKT signaling inhibitors, such as rapamycin or dual PI3K/MTOR inhibitors, potently and globally suppresses AKT signaling pathways as well as mutant-TK mediated pathways including MAPKinases and STAT5 signaling.

To provide a mathematical tool to describe the combination effect of two agents, we performed fixed ratio dilution experiments to create isobolograms using a method of Chou and Talalay [37]. Cells were treated with the single agents and fixed ratios of NVP-BGT226 or NVP-BEZ235 plus sunitinib (MOLM14 cells) or imatinib (K562 cells) to assess for induction of apoptosis. This was used to create isobolograms (Figure 5). Combination of NVP-BGT226 with sunitinib in MOLM14 cells (Figure 5A) resulted in an experiment point that falls to the left of the predicted line of additive effect when taking ED90 (i.e. 90% apoptotic cells) as the experimental end point (indicating a superadditive effect). Similar results were achieved for NVP-BGT226 combined with imatinib in K562 cells with an experiment point lying on (ED50) or falling to the left (ED90) of the predicted line of additive effect (Figure 5B).

Calculation of combination indices (CI) revealed a CI close to 1 (i.e. additive effect) for ED50s in both cell lines and a CI < 1 for ED90 – indicating synergy (i.e. superadditive effects).

Due to the moderate proapoptotic effect of NVP-BEZ235 when administered as single agent, calculation of isobolograms and resultant CIs were restricted to ED25-50 concentrations for NVP-BEZ235+TKI combinations. Nevertheless, a strong synergistic effect was revealed for both combinations of NVP-BEZ235 plus sunitinib in FLT3 ITD positive MOLM14 cells (Figure 5C), or NVP-BE235 plus imatinib in BCR-ABL1 positive K562 cells (Figure 5D) with CIs well smaller than 1. Additionally, estimated ED90s are provided along with each figure as well. These findings indicate that a combination approach may override the G1/G0 arrest observed for NVP-BEZ235 monotherapy – which is supported by increased cleavage of caspase 3 in the western immunoblot experiments when combined with TKI (Figure 4).

In order to minimize cell type specific off-target effects to validate our findings for the mutant FLT3 ITD cell line MOLM14 and the BCR-ABL1 positive cell line K562, we established an isogenic Ba/F3 cell line model transfected with AKT-autoactivating FLT3 ITD or BCR-ABL1 mutations. We further comparatively extended our studies to additional leukemia-associated mutant-TK (a full list of tested mutant-TK Ba/F3 cell transfectants are provided with Table 1).

Immunoblotting for phospho-AKT was performed after successful transfection and weaning of IL3-dependent growth (true autoactivating TK mutations result in IL3-independent growth) and found that AKT activation increases after transfection of plasmid vectors encoding for a FLT3 ITD, FLT3 D835V, KIT D816Y or BCR-ABL1 isoform.

While cytokine-starved parental BaF3 cells did only reveal moderate, if any, phosphorylation levels of AKT, IL3-stimulated or oncogene-transfected Ba/F3 cells did globally activate AKT on codons Thr308 as well as Ser473. Notably, TK-mediated activation of AKT was by far more pronounced compared to physiologic, cytokine (IL3)-mediated activation of AKT (Figure 6A).

We tested our model by treating Ba/F3 cells transfected with the gain-of-function FLT3 D835V mutation with either NVP-BGT226 or NVP-BEZ235 and probed for T308- or S473-phosphorylated AKT isoforms in a western immunoblot using whole cell lysates. Both inhibitors potently and globally suppressed AKT phosphorylation of initially maximally activated AKT (Figure 6B). Jurkat cells treated with well established PI3K inhibitors (Wortmannin, LY294002) served as controls.

We next utilized our Ba/F3 model to evaluate the mutant-TK specific antiproliferative effect of either NVP-BGT226 or NVP-BEZ235 in an isogenic cellular background. Both agents revealed compound-specific – but also distinct mutation-specific activity, with the parental cell line (stimulated with IL3) being the least sensitive for both tested agents (estimated IC50s by regression dose effect analysis ~2400 nM for NVP-BEZ235 and ~800 nM for NVP-BGT226). BCR-ABL1, FLT3 D835V and KIT D816Y transfectants displayed an intermediate sensitivity pattern (estimated IC50s for NVP-BEZ235 ~10-130 nM and ~65-180 nM for NVP-BGT226) whereas FLT3 ITD demonstrated high sensitivity for both agents with IC50s below 10 nM. Representative dose vs. effect graphs are shown in Figure 7A/B. A summary of achieved IC50s is provided in Table 1 - together with additional (mutant-) TK isoforms tested.

When testing for induction of apoptosis, NVP-BGT226 proved to be highly potent in virtually all tested cell lines, with transfectant-specific IC50s raging from ~120-1800 nM (again, the parental IL3-stimulated cell line being the least sensitive). In contrast, the high capacity to inhibit cellular proliferation for NVP-BEZ235 did not similarly translate into potency to induce apoptosis for all tested transfectant cell lines. Importantly, Ba/F3 FLT3 ITD cells, which were highly inhibited with regard to cellular proliferation, did only show moderate induction of apoptosis towards NVP-BEZ235 (IC50 not reached up to tested doses of 10 000 nM, proportion of apoptotic cells at 5000 nM: 27%). In analogy, BCR-ABL1 transfected cells failed to achieve IC50 as well, with a proportion of 39% apoptotic cells at 5000 nM). These findings are in line with our results for the corresponding tested human leukemia cell lines. Notably, other transfectants (e.g. FLT3 D835V and KIT D816Y) retained some level of sensitivity with regard to induction of apoptosis. Representative dose vs. effect graphs are shown in Figure 7C/D. A full list of IC50s for both agents and additionally tested mutant-TK Ba/F3 cells are provided with Table 1.

We confirmed our observations at the protein level and treated Ba/F3 parental (+/− IL3), FLT3 ITD, FLT3 D835V, KIT D816Y or BCR-ABL1 transfected cells with NVP-BGT226 or NVP-BEZ235 to probe whole cell lysates for AKT phosphorylation in an immunoblot. Dual inhibition of PI3Kinases and MTOR1/2 lead to potent AKT dephosphorylation of initially activated AKT in IL3-stimulated or mutant-TK activated cells in the low nanomolar range (Figure 7E). This went along with the observed antiproliferative effects for both agents on the cellular level. In line with our cellular apoptosis assays, immunoblotting for cleaved caspase 3 as an indicator for induced apoptosis again revealed that higher doses are needed to induce programmed cell death in these cell lines (if any, when looking at NVP-BEZ235). These findings argue for a complex regulation of programmed cell death, which will need to be studied in more detail in future studies. One hypothesis may state that induction of apoptosis is mediated via Thr308: We observed a particular high phosphorylation pattern of Thr308 in cells transfected with the tyrosine kinase domain (TKD)-mutated FLT3 D835V and KIT D816Y isoforms in our assays (see Figure 6). Interestingly these were the cell lines to display the highest rates of apoptosis after treatment. In contrast BCR-ABL1 or FLT3 ITD transfectants, presenting with comparably lower p-T308 AKT levels, were by far less sensitive towards NVP-BEZ235 with regard to induction of apoptosis (compare Figure 7C). These observations are in line with Thr308 phosphorylation levels seen in MOLM14 and K562 cell lines, which were relatively weak to absent (see also Figure 4).

To evaluate, whether our in vitro data derived from leukemia cell lines and mutant-TK cell line models translate into a clinically meaningful antiproliferative effect in native leukemia cells, we treated an acute leukemia sample taken from a patient suffering from FLT3 mutant-TKD2 positive AML and a sample from a patient with AML tested negative for FLT3 or KIT mutations with varying concentrations of NVP-BGT226 or NVP-BEZ235 and tested for the capacity to inhibit cellular proliferation ex vivo using an XTT-based assay. The FLT3 TKD2 positive leukemia sample (pat. 556) revealed high sensitivity towards NVP-BGT226 as well as NVP-BEZ235 with calculated IC50s in the low nanomolar range in a dose-effect plot. In contrast the AML sample lacking mutant-TK isoforms (pat. 303) was virtual insensitive towards both agents with IC50s well above 5000 nM. Importantly, mononuclear cells extracted from an aspirate of a bone marrow donor (bm-donor 552) revealed a sensitivity profile of IC50s > 1000 nM for both compounds. Dose-effect plots were created for tested patient samples to calculate IC50s, which are provided in Table 2 along with AKT expression patterns (exemplary dose-effect plots to calculate IC50s is provided in Additional file 2: Figure S1A; patient characteristics are provided in Additional file 1: Table S1).

The findings of equipotent sensitivity profiles of NVP-BGT226 and NVP-BEZ235 with regard to inhibition of cellular proliferation in native AKT-activated leukemia cells are in line with our in vitro data provided above. Notably, the PI3K/AKT/MTOR pathway is a target of NVP-BGT226 as well as NVP-BEZ235 in native acute leukemia cells as verified in an immunoblot experiment for two patient samples with newly diagnosed acute leukemia (Additional file 2: Figure S1B, provided with the online version of the article). This further underlines and validates the herein described in vitro and ex vivo data rather than arguing for off-target effects.

Correlation of ex vivo responses to NVP-BGT226 and NVP-BEZ235 with AKT expression levels suggests that augmented activation of AKT (compared to healthy bone marrow donors), i.e. phosphorylation of Thr308 as well as Ser473 but not mere AKT protein levels, may be a requisite for inhibition of cellular proliferation in response towards dual PI3K/MTOR inhibition. Clearly, analysis of pan-AKT protein levels may not predict for response, as AKT expression was highest in the AML sample refractory towards both inhibitors (Table 2).

Next, we studied, whether NVP-BGT226 and NVP-BEZ235 are capable of inducing apoptosis in native leukemia samples. Leukemia blasts extracted from acute myeloid, promyelocytic or lymphoid leukemia with or without detectable TK mutations were treated with NVP-BGT226 or NVP-BEZ235 in dose dilution series and apoptosis was assessed by an Annexin V/PI stain.

In analogy to our in vitro data described before, both agents demonstrated variable apoptosis induction. Notably, NVP-BGT226 proved to be the more potent drug with high effectivity and IC50s in the lower nanomolar range in some patient samples (Table 2). Of note, native mononuclear cells derived from bone marrow donors revealed much higher IC50s for both agents. Analysis of AKT expression levels suggest that global activation of AKT with augmented phosphorylation of Ser473 as well as Thr308 beyond a baseline set as 1 on a normalised AKT expression scale is a prerequisite to predict response towards the dual PI3K/MTOR inhibition. However, this observation will need prospective verification on a larger patient cohort.

Ba/F3 cell lines were obtained through the American Type Culture Collection (ATCC, Manassas, VA). The MOLM14 cell line was acquired through the Fujisaki Cell Center (Okayama, Japan). The MLL-AF9 fusion positive acute monocytic leukemia cell line MOLM-14 harbors a heterozygous FLT3 ITD mutation [52].

The T-lymphoblastic cell line Jurkat and the CML blast crisis cell line K562 were obtained from the Deutsche Sammlung für Mikroorganismen und Zellkulturen (DSMZ, Braunschweig, Germany).

The human mast cell leukemia cell line HMC1.1, harboring an imatinib-sensitive KIT V560G mutation [42], and the sister cell line HMC1.2, harboring an additional imatinib-insensitive KIT D816V mutation [43] were provided by Prof. Heinrich, OHSU, Oregon. The GIST tumor cell lines GIST48 and GIST882 were kindly provided by Dr. Kopp, (University of Tübingen, Germany). GIST882 is harboring a KIT K642E mutation [44]; GIST48 was established from a patient with relapsing GIST under imatinib therapy (GIST48). This cell line harbors a primary juxtamembrane KIT mutation plus a secondary imatinib-insensitive mutation in the kinase domain [45].

Cells were cultured in RPMI 1640, supplemented with 10% fetal bovine serum (GIBCO/Invitrogen, Darmstadt, Germany or Biochrom AG, Berlin, Germany), 1% penicillin G (10,000 units/mL), and streptomycin (10,000 μg/mg) and 2 mmol/L L-glutamine. In addition, parental Ba/F3 cells were supplemented with 10 ng/ml of mouse-IL3. Negativity for mycoplasma contamination was confirmed using the pluripotent PCR Mycoplasma test Kit (AppliChem, Darmstadt, Germany). Cell lines harboring a mutant KIT, FLT3 or BCR-ABL1 were sequence confirmed.

Bone marrow aspirate and peripheral blood samples from consented patients with acute leukemia as well as samples from healthy blood and bone marrow donors were collected in 5000 U heparin with the approval of the ethics committee of the Medical Faculties of the University of Tübingen or the University of Ulm. Mononuclear cells were isolated by Ficoll-Hypaque density gradient fractionation. Cells were cultured in DMEM medium, supplemented with 20% fetal bovine serum (GIBCO/Invitrogen, Darmstadt, Germany or Biochrom AG, Berlin, Germany), 1% penicillin G (10,000 units/mL), and streptomycin (10,000 μg/mg) and 2 mmol/L L-glutamine.

The dual pan class I PI3K AND MTOR complex 1 and 2 inhibitors NVP-BEZ235 [28] and NVP-BGT226 [27], two imidazo[4,5-c]quinoline derivatives competitively binding to the ATP-binding cleft of these enzymes were provided by Novartis (Basel, Switzerland). Stock solutions were created according to the manufacturer’s instructions. Rapamycin and the PI3K inhibitors LY294002 and Wortmannin were obtained from Cell Signaling (Danvers, MA). The TKI dasatinib (formerly BMS-354825) and sunitinib (formerly SU11248) were obtained from the University of Tübingen Hospital Pharmacy and dissolved in DMSO to create 10 mmol/L stock solutions and stored at −20°C.

Rabbit anti-panAKT, panFLT3, panABL1 or anti-cleaved caspase 3 antibodies were used at a 1:500 to 1:1000 dilution. Rabbit anti-phospho-AKT antibodies detecting phosphorylated isoforms of T308-AKT, S473-AKT, S807/811-RB, S575-ULK1, T389-p70S6K, Y694-STAT5, FLT3, ABL1 or T202/Y204-ERK1/2 were used at a 1:250 to 1:1000 dilution. An anti-actin mouse monoclonal antibody was used as a loading control. All antibodies, if not otherwise indicated, were purchased from Cell Signaling Technology.

As controls for AKT Thr308- and Ser473-phosphorylation we used Jurkat cells untreated (phosphorylated positive control) or treated with LY294002 or Wortmannin (negative controls lacking T308/S473 phosphorylation).

Infrared dye-conjugated secondary goat anti-rabbit and anti-mouse antibodies to use in a LI-COR® imaging detection system were used according to standard protocols (LI-COR Biosciences, Lincoln, NE). For flow cytometry studies, fluorescent dye-conjugated secondary goat anti-rabbit or anti-mouse antibodies were used according to standard protocols. Cell Signaling anti-rabbit IgG(H+L),F(ab’)2Fragment Alexa Fluor conjugate antibodies were used to assess expression levels provided in Table 2. The Invitrogen Zenon Alexa Fluor labeling kit was used for expression levels provided in Additional file 1: Table S1.

Cell pellets were lysed with 100 to 150 μl of protein lysis buffer (50 mmol/L Tris, 150 mmol/L NaCl, 1% NP40, 0.25% deoxycholate with added inhibitors aprotinin, AEBSF, leupeptin, pepstatin, sodium orthovanadate, and sodium pyruvate, respectively phosphatase inhibitor cocktails „2“ and „1“ or „3“ (Sigma, St. Louis, MO). Protein from cell lysates (75 to 200 μg protein) was used for whole cell protein analysis after denaturing by Western immunoblot assays using a BioRad Criterion system. Nonspecific binding was blocked by incubating the blots in nonfat dry milk or BSA. Primary antibodies were incubated for one hour or over night, followed by several washes of Tris-buffered saline (TBS) containing 0.005% Tween 20. The appropriate secondary antibody was applied for 30‘, followed by several washes. Antibody-reactive proteins were detected using a LI-COR Odyssey® fluorescence optical system.

Intracellular (phospho-)AKT protein expression levels were assayed as follows: Cells were fixed and permeabilized using the Fix & Perm® Fixation and Permeabilization kit (ADG-An der Grub Bioresearch, Kaumberg, Austria). Unlabeled primary AKT antibodies were added in a 1:1000 dilution to the cell suspension and incubated for 1 hour at room temperature followed by PBS washing and resuspension. Fluorescent dye-conjugated secondary antibodies were added in a 1:10 000 dilution and cells were incubated for 30 min at room temperature. After rinsing and resuspension, (phospho-)AKT protein expression levels were assayed using a FACScalibur® flow cytometer loaded with CellQuest® analysis software (BD, Heidelberg, Germany).

To compare constitutive activation of AKT mediated by autoactive tyrosine kinase signaling in a homologous cellular background, an isogenic cell model (Ba/F3) expressing different human tyrosine kinase mutations was established. An IL3-dependent murine pro-B cell line (Ba/F3) was transfected with plasmid vectors containing cDNA of human (mutant) FLT3 and KIT isoforms, as well as the BCR/ABL1 fusion mutation isoform. Gain-of function tyrosine kinase mutations lead to factor-independency.

Site-directed mutagenesis and generation of a Ba/F3 cell lines stably expressing mutant KIT D816V, D816Y, FLT3 ITD, D835V, D835Y, K663Q, BCR/ABL1 and FLT3 wildtype was previously performed as described before [36, 53‐55].

FLT3 S451F cDNA cloned into a pCMVneo plasmid vector [53] was generously provided by Dr. Fröhling, University of Ulm, Germany. KIT wildtype cDNA cloned into a pJP1563 plasmid vector was obtained from the DNASU Plasmid Repository at the Biodesign Institute of the Arizona State University (ASU). Lipofection transfection into the parental Ba/F3 cell line was performed to stably express KIT wildtype or mutant FLT3 S451F by double selection for neomycin (pCMVneo plasmid), blasticidin (pJP1563 plasmid) or gentamicin (G418; all other plasmids) resistance and IL-3-independent growth. The Ba/F3 KIT wildtype cell line was cultured using recombinant human stem cell factor (SCF/KIT Ligand, R&D, Minneapolis, MN) as a growth supplement.

Cells were treated in dilution series with the respective small molecule inhibitor.

Translocation of phosphatidylserine from the inner to the outer leaflet of the plasma membrane as an early indicator of apoptosis was analyzed using an Annexin V-based assay (Immunotech, Marseilles, France) and a FACScalibur® flow cytometer loaded with CellQuest® analysis software (BD, Heidelberg, Germany) [35].

Cellular proliferation was measured using an 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide inner salt (XTT)–based assay (Sigma) as described previously [35].

A propidium iodide-based flow cytometry assay was assessed as described previously [56]. In short, a propidium iodide stain assay is used to segregate cells according to the DNA content, which is graphically shown in a histogram plot (high content in G2/M, intermediate content in S-phase, low content in G1/G0 and lowest content in dead/apoptotic cells, which defines a sub-G1/G0 fraction),

Linear regression dose effect plots to calculate IC50s were computed with values in between upper and lower threshold doses of minimal/maximal dose effects using Calcusyn Software (Biosoft, Cambridge, UK), which is based on equations provided by Chou and Talaly [37].

Isobologram analyses were performed as we have previously described [54, 55]. In short, cells were treated with fixed ratios in relationship to the individual agent ED and data was analyzed using the method of Chou and Talalay to produce isobolograms. This allowed calculation of combination indices (CI). The CI provide a numerical description of the effects of a combination treatment. Specifically, a CI < 1 indicates synergy, a CI = 1 indicates an additive effect, and a CI > 1 indicates antagonism of the two agents.