Dual Inhibition of PI3K and mTORC1/2 Signaling by NVP-BEZ235 as a New Therapeutic Strategy for Acute Myeloid Leukemia

In acute myeloid leukemia (AML), the constitutive activation of signal transduction pathways enhances the survival and proliferation of the leukemic cells (1). The increased understanding of AML biology in recent years, however, contrasts with the remaining poor prognosis for patients with this disease (2). The development of new therapies is thus highly desirable, and the targeting of signaling pathways that are preferentially activated in transformed AML cells represents a rapidly expanding research field in this regard (1).

The class IA phosphoinositide 3-kinase (PI3K) is a lipid kinase that is mainly induced following growth factor receptor activation. Upon PI3K stimulation, the generation of phosphatidylinositol (3,4,5)-trisphosphate (PIP3) induces the activating phosphorylation by PDK1 of the oncogenic kinase Akt on Thr³⁰⁸. The PI3K signaling pathway is frequently deregulated in cancer cells, and its constitutive activity strongly contributes to the oncogenic process (3). In AML, constitutive PI3K/Akt activation, mainly due to the activity of the PI3K p110δ isoform (4, 5), is detected in 50% to 80% of samples at diagnosis (6–8) and has recently been linked to an insulin-like growth factor-1 (IGF-1)/IGF-1 receptor (IGF-1R) autocrine loop (9). Although specific inhibition of PI3K p110δ activity barely induces apoptosis in AML (9, 10), the PI3K/Akt pathway markedly contributes to AML cell proliferation (4, 5, 9), therefore representing a potential therapeutic target for this disease.

The serine/threonine kinase mTOR (mammalian target of rapamycin) is a component of two exclusive complexes, mTORC1 and mTORC2, which are defined by both their molecular composition and substrate specificity. The interaction of mTOR with raptor (regulatory-associated protein of mTOR) defines mTORC1, whose substrates include the ribosomal protein S6 kinase (S6K) and the eukaryotic initiation factor 4E (eIF4E)–binding proteins (4E-BP; refs. 11, 12). The mTORC1 pathway generally controls the cap-dependent translation of mRNAs through the phosphorylation of the key translation regulator 4E-BP1 (13). Briefly, hypophosphorylated 4E-BP1 molecules block the formation of the translation-initiating complexes (eIF4F) through their interaction with the eIF4E proteins that bind the 7-methyl-GTP (⁷mGTP) cap structure located at the 5′ end of most mRNA molecules (14). Following a complex sequence of phosphorylation events, the release of 4E-BP1 from eIF4E enables the interaction between the eIF4E and eIF4G proteins that in turn initiates eIF4F assembly (15, 16). In contrast, the role of the mTORC2 complex, which is based on the interaction between rictor (rapamycin-insensitive companion of mTOR) and mTOR (17, 18), has only recently emerged in cancer cell biology and is mainly related to the control of Akt Ser⁴⁷³ phosphorylation (19).

Rapamycin and its derivates (RAD001, CCI-779), referred to as rapalogs, are highly specific mTORC1 inhibitors that have been developed as anticancer drugs due to the frequent activation of the mTORC1 pathway in different cancers including AML (20). Rapamycin, through its association with the FK506 binding protein 12 (FKBP-12; ref. 21), induces the disassembly of the mTORC1 complex (22, 23), resulting in the repression of mTORC1 activity. However, although an interesting antitumor activity of rapamycin as a monotherapy has been reported in some AML patients (24), mechanisms underlying resistance to rapamycin have now been described. First, the rapamycin-induced inhibition of mTORC1 activity in primary AML cells increases PI3K activity via the stimulation of an IGF-1/IGF-1R autocrine loop, which enhances AML cell proliferation (10). Furthermore, allosteric inhibition of mTORC1 does not block protein translation, mainly due to the sustained high level of 4E-BP1 phosphorylation in AML cells treated with rapamycin (25). Moreover, rapamycin generally fails to inhibit mTORC2 activity (17, 18), although evidence has suggested that prolonged treatment with rapamycin may inhibit mTORC2 activity against its substrate Akt Ser⁴⁷³, depending on the cell type (26, 27). Taken together, these data led to the development of new strategies to efficiently target mTOR signaling, such as direct ATP-competitive “active-site” mTOR inhibitors. Recently, different groups have reported the identification and characterization of those molecules (28–31) and observed promising anticancer activities both in vitro and in vivo (29, 32). However, in AML, the implication of the PI3K/Akt signaling pathway in blast cell proliferation (4, 5, 9) and the oncogenic cross-talk between mTOR and PI3K (10) clearly suggest an additional benefit resulting from the simultaneous blockade of PI3K and mTOR catalytic activities.

In our current study, we report for the first time the effects of the dual PI3K and mTOR inhibitor NVP-BEZ235 (Novartis; ref. 33) in AML. Our data show that this molecule inhibits PI3K, mTORC1, and mTORC2 activities in both human AML cell lines and primary AML samples. Moreover, in contrast to rapamycin, NVP-BEZ235 causes a complete blockade of the phosphorylation of the translation repressor 4E-BP1. This results in a marked inhibition of protein translation in AML cells. NVP-BEZ235 also shows potent antileukemic activities in AML without affecting normal hematopoiesis ex vivo. Given these results, NVP-BEZ235 shows great promise for testing in clinical trials as a novel AML therapeutic agent.

Bone marrow samples were obtained from 21 patients with newly diagnosed AML (at the exclusion of AML3, AML6, and AML7 phenotypes), all included in various chemotherapy trials initiated by the French Multicenter Group, Groupe Ouest Est des Leucémies et Autres Maladies du Sang (GOELAMS). Normal peripheral CD34⁺ cells were purified from healthy allogenic donors after informed consent. All biological studies were approved by the GOELAMS Institutional Review Board, and signed informed consent was obtained in accordance with the Declaration of Helsinki.

Blast cells were isolated from bone marrow aspirates from AML patients at diagnosis by a Ficoll-Hypaque density gradient separation. Only bone marrow samples with >80% of blast cells were used. Cell viability was assessed by trypan blue assay, and only samples with <5% trypan blue–positive cells were further processed. The human leukemic cell lines were purchased from the German Resource Centre for Biological Material for MV4-11 and from American Type Culture Collection for MOLM-14 and OCI-AML3. We used from 1 to 1,000 nmol/L NVP-BEZ235 (Novartis), 10 nmol/L rapamycin (Sigma), 10 μmol/L UO126 (Cell Signaling Technology), and 25 μmol/L LY294002 (Sigma).

The global protein synthesis and cell proliferation were assessed by [³H]leucine and [³H]thymidine pulse assays, respectively, as reported (25).

⁷mGTP pull-down experiments were performed as reported (25). Briefly, cell lysates were incubated at 4°C for 2 hours with ⁷mGTP-Sepharose beads (Amersham) and then washed and resuspended in boiling Laemmli sample buffer.

Polysome analyses were done as reported (25) in the MV4-11 leukemic cell line. After separation through 10% to 50% sucrose density gradients, the polysome profile was generated by the continuous measure of the absorbance at 254 nm.

Small interfering RNA (siRNA) targeting human rictor (L-016984-00, Dharmacon) and control nontargeted siRNAs (Dharmacon) were transfected using the Amaxa nucleofector (Amaxa Biosystems) as previously described (25).

Whole-cell extracts and Western blots were performed as previously described (25) using antibodies listed in Supplementary Table S1. The images were captured using a charge-coupled device camera (LAS3000 from Fujifilm). The signal intensity was quantified using Multigauge software from Fujifilm.

Apoptosis was quantified by Annexin V–PE staining (Becton Dickinson) as previously described (34), and cell cycle was assessed using DNA staining with Draq5 (Alexis Biochemicals). Detailed cycle analysis of MV4-11 cells was performed by quantifying G₀-G₁, S, and G₂-M phases by propidium iodide staining using CycleTEST PLUS kit (Becton Dickinson) according to the manufacturer's recommendations. For the quantification of G₀ and M phases, 10⁶ cells were permeabilized with 1 mL of ice-cold ethanol (1 hour, 4°C). Following two washes with PBS, 1% fetal bovine serum, 0.25% Triton X-100 (PFT), the cells were stained in 200 μL PFT for 30 minutes at room temperature in the dark, either with 1 μg of 7-aminoactinomycin D (7-AAD; Sigma) and 5 μL of Alexa Fluor 488–conjugated anti-human Ki67 mAb (B56, Becton Dickinson) or with 2 μL of Alexa Fluor 488–conjugated anti-phospho(Ser¹⁰)-histone H3 polyclonal antibody (Cell Signaling Technology), respectively.

The formation erythroid (BFU-E), granulo-macrophage (CFU-GM), and granulo-erythroid-megakaryocytic-monocytic (CFU-GEMM) colony-forming units from normal CD34⁺ hematopoietic cells, exposed or not to NVP-BEZ235 during 12 hours, were assessed as reported (34).

CFU-leukemia (CFU-L) assays from primary AML cells exposed or not to NVP-BEZ235 during 12 hours were performed, as reported (24, 34).

Data are expressed as mean values and SD. Statistical significance of differences observed between experimental groups was determined using Student's t-test. *, **, and *** mean P < 0.05, P < 0.01, and P < 0.001, respectively. The experiments performed in AML cell lines were done in triplicate. In each case using primary AML cells, the term “independent experiments” means (a) that the experiments were performed on different days and (b) that cells from different patients were used.

The dual PI3K/mTOR inhibitory activity of NVP-BEZ235 was assessed both in human AML cell lines (MV4-11, MOLM-14, and OCI-AML3) and in primary leukemic cells sampled from AML patients at diagnosis. Dose-dependent experiments (from 1 to 1,000 nmol/L) revealed that PI3K activity is strongly inhibited by NVP-BEZ235 treatment in all three AML cell lines tested. Indeed, Akt phosphorylation on Thr³⁰⁸, which directly reflects the inhibition of PI3K activity, was suppressed by a 10 nmol/L dose of NVP-BEZ235 (Fig. 1A). Moreover, NVP-BEZ235 was found to inhibit P70S6K Thr³⁸⁹ phosphorylation from 25 nmol/L in a similar manner to 10 nmol/L rapamycin, reflecting an inhibition of mTORC1 activity (Fig. 1A). In these experiments, the broad-spectrum kinase inhibitor LY294002 was used as a control for the inhibition of Akt Thr³⁰⁸ and P70S6K Thr³⁸⁹ phosphorylation (Fig. 1A). Finally, NVP-BEZ235 did not suppress extracellular signal-regulated kinase (ERK) Thr²⁰²/Tyr²⁰⁴ phosphorylation, in contrast to the specific mitogen-activated protein kinase (MAPK)/ERK kinase (MEK) 1/2 inhibitor UO126, and did not decrease signal transducer and activator of transcription 5 (STAT5) Tyr⁶⁹⁴ phosphorylation in the MV4-11 and MOLM-14 cell lines (the OCI-AML3 cell line is deficient in STAT5 Tyr⁶⁹⁴ phosphorylation and was used as a negative control), even at the maximal concentration of 1,000 nmol/L (Fig. 1A). These results emphasize the specificity of NVP-BEZ235 against class IA PI3K and mTOR activities in AML cells. Based on these results, all subsequent experiments were performed using 10, 100, and 1,000 nmol/L concentrations of NVP-BEZ235.

In a cohort of 10 primary AML samples, treatment with NVP-BEZ235 was also found to induce a dose-dependent decrease in Akt Thr³⁰⁸ and P70S6K Thr³⁸⁹ phosphorylation without affecting the ERK/MAPK pathway (Fig. 1B, left; Supplementary Fig. S1). Quantification by Western blotting of these 10 AML samples (Fig. 1B, right) indicated a significantly decreased phosphorylation of P70S6K Thr³⁸⁹ at all concentrations of NVP-BEZ235 (decreases from 54% to 90% using 10 to 1,000 nmol/L NVP-BEZ235). We also determined that NVP-BEZ235 still markedly inhibited P70S6K Thr³⁸⁹ phosphorylation after a 24-hour incubation in both primary AML cells and the MV4-11 cell line (Supplementary Fig. S2). NVP-BEZ235 thus behaves as a potent cell-permeable PI3K and mTORC1 inhibitor in leukemic cells.

We further wished to examine mTORC2 activity in AML cells. We first assessed the mTOR autophosphorylation site on Ser²⁴⁸¹ (35, 36), as this closely reflects its catalytic activity in both mTORC1 and mTORC2 complexes (29). Interestingly, we found that in the AML cell lines, NVP-BEZ235 induced a dose-dependent inhibition of mTOR Ser²⁴⁸¹ phosphorylation, in contrast to rapamycin (Fig. 2A), which suggested that not only mTORC1 but also mTORC2 could be targeted by this compound. The Akt and SGK serine/threonine kinases are the best-characterized mTORC2 substrates (27, 29, 37). As expected, NVP-BEZ235 suppressed Akt Ser⁴⁷³ phosphorylation in a dose-dependent manner (Fig. 2A). In addition, the phosphorylation of FOXO3a at Ser²⁵³, which directly reflects the activation state of Akt on Ser⁴⁷³, was also inhibited to a similar extent by NVP-BEZ235 (Fig. 2A). The activity of SGK was then assessed by testing the phosphorylation of its substrate NDRG1 on Thr³⁴⁶ (38). Accordingly, NVP-BEZ235 strongly inhibited NDRG1 Thr³⁴⁶ phosphorylation (Fig. 2A). Not surprisingly, rapamycin did not inhibit Akt (Ser⁴⁷³), FOXO3a (Ser²⁵³), or NDRG1 (Thr³⁴⁶) phosphorylation under the same conditions, which confirms that the allosteric inhibition of mTORC1 does not repress mTORC2 activity in short-term experiments (Fig. 2A).

NVP-BEZ235 also suppressed Akt Ser⁴⁷³, FOXO3a Ser²⁵³, and NDRG1 Thr³⁴⁶ phosphorylation in primary AML samples at a dose of 10 nmol/L (Fig. 2B, left; Supplementary Fig. S3A). Western blotting quantification of the signals from the AML samples tested indicated a mean decrease from 56% to 83% and from 71% to 89% for the bands detected with the anti-Akt Ser⁴⁷³ (n = 8) and anti-NDRG1 Thr³⁴⁶ antibodies (n = 5), using 10 to 1,000 nmol/L NVP-BEZ235, respectively (Fig. 2B, right).

It has been established that full Akt and SGK activation also requires PI3K activity (39–41), and to therefore more precisely assess the biochemical activity of the mTORC2 complex in AML cells, we tested the phosphorylation status of paxillin on Tyr¹¹⁸, reported as an indirect substrate for mTORC2 activity in nontransformed cells (17). We verified that this phosphorylation was indeed dependent on mTORC2 activity in primary AML cells by disrupting the mTORC2 complex using rictor siRNA. As shown in Fig. 2C, paxillin phosphorylation on Tyr¹¹⁸ decreased when rictor expression was knocked down by siRNA, thus indicating that paxillin Tyr¹¹⁸ phosphorylation reflects mTORC2 activity in primary AML cells. Interestingly, NVP-BEZ235 induced a dose-dependent decrease in paxillin Tyr¹¹⁸ phosphorylation in primary AML cells (Fig. 2D, left; Supplementary Fig. S3B), and the quantified immunoblotting signals from four AML samples indicated a mean decrease of paxillin Tyr¹¹⁸ phosphorylation of 43% and 73% using 100 and 1,000 nmol/L NVP-BEZ235, respectively (Fig. 2D, right). Taken together, these data clearly show that NVP-BEZ235 represses mTORC2 activity in AML cells.

We have recently shown that the limitations in the antileukemic activity of the rapalogs are mainly due to the rapamycin-resistant phosphorylation of the translation regulator 4E-BP1, leading to a deregulation of cap-dependent mRNA translation (25). We thus hypothesized that a direct targeting of mTOR catalytic activity in an FKBP-12–independent manner would induce the inhibition of the oncogenic translation process, resulting in a more potent antileukemic effect. We therefore compared the effects of rapamycin and NVP-BEZ235 on 4E-BP1 phosphorylation in AML cells. Interestingly, and in contrast to rapamycin, NVP-BEZ235 induced a strong dephosphorylation of 4E-BP1 on the Thr^37/46, Thr⁷⁰, and Ser⁶⁵ residues in the AML cell lines (Fig. 3A). Similar results were obtained in primary AML samples (Fig. 3B, left; Supplementary Fig. S4). Quantification of the Western blotting signals obtained from 14 AML samples indicated a mean decrease from 46% to 97% using 10 to 1,000 nmol/L NVP-BEZ235, respectively (Fig. 3B, right). Furthermore, this blockade of 4E-BP1 phosphorylation was sustained after 24 hours of exposure of NVP-BEZ235 (Supplementary Fig. S1). NVP-BEZ235 is thus a potential inhibitor of protein translation in AML.

The phosphorylation of 4E-BP1 is the limiting step in the assembly of the translation-initiating complex eIF4F, and is initiated by the interaction between the eIF4E and eIF4G proteins (25). We thus performed ⁷mGTP pull-down assays using the MV4-11 cell line to examine whether a 24-hour exposure to NVP-BEZ235 modulates the interaction between eIF4E and eIF4G (active translation) or 4E-BP1 (inactive translation). As shown in Fig. 4A, NVP-BEZ235 decreased the levels of eIF4G but increased the amounts of 4E-BP1 associated with eIF4E, indicating an inhibition of eIF4F assembly.

We additionally performed polysome analysis in MV4-11 cells from which we extracted a nonpolysomal fraction corresponding to ribosomes not engaged in translation and a polysomal fraction containing the ribosomes undergoing active translation (25). After centrifugation through a sucrose gradient, continuous measurements at a 254-nm absorbance revealed three peaks corresponding to 40S and 60S ribosome subunits and 80S ribosomes, respectively (Fig. 4B, arrows). The polysomes had sedimented below (Fig. 4B, horizontal two-point arrow). Treatment of the MV4-11 cell line with NVP-BEZ235 clearly blocked translation initiation, as indicated by the large to small polysome shift and a concomitant increase in the free ribosome concentration (Fig. 4B). We further performed [³H]leucine assays, in which the radioactivity levels are proportional to the neosynthesized protein concentrations. NVP-BEZ235 (from 100 nmol/L) markedly decreased the global protein synthesis levels in the MV4-11 cell line, and this closely correlated with the suppression of phosphorylation of the translation regulator 4E-BP1 (Fig. 4C). Moreover, the expression of three highly oncogenic proteins known to be regulated at the translation initiation level, c-Myc, cyclin D1, and Bcl-xL (42), was markedly reduced in these cells after 24 hours of exposure to NVP-BEZ235. These data further show that NVP-BEZ235 is a potent inhibitor of the translation process in AML cells via the inhibition of the rapamycin-resistant phosphorylation of 4E-BP1.

We investigated the antileukemic activity of the NVP-BEZ235 compound and assayed the cell proliferation rates using [³H]thymidine pulse assays. We consistently observed a dose-dependent decrease in cell proliferation, with an IC₅₀ of 59 nmol/L in primary AML cells (n = 4), and of 25 nmol/L in MV4-11, 45 nmol/L in MOLM-14, and 60 nmol/L in OCI-AML3 cells (Fig. 5A).

Consistent with its antiproliferative effects, the cell cycle analysis using propidium iodide staining showed a significant increase of the proportion of MV4-11 cells in the G₀-G₁ phase of the cell cycle and a decrease of the proportion of cells in the S and G₂-M phases of the cell cycle on NVP-BEZ235 treatment (Fig. 5B, first column). Moreover, using a more detailed cell cycle analysis by 7-AAD/Ki67 and phospho(Ser¹⁰)-histone H3 staining, we observed that NVP-BEZ235 led to the accumulation of cells in the G₀ phase (16-fold increase; Fig. 5B, second column) and to a strong reduction of cells in the M phase (7-fold decrease; Fig. 5B, third column). Similar results were obtained from the same experiments in MOLM-14 cells (data not shown). We further found by Western blotting that NVP-BEZ235 also downregulates the expression of the cell cycle activators CDK2 and SCF^SKP2 and increases the expression of the cell cycle inhibitor p27^kip1 (Fig. 5C).

We generated clonogenic cultures of primary AML cells and incubated these with 10 to 1,000 nmol/L NVP-BEZ235. The subsequent levels of clonogenic leukemic colony formation (CFU-L) from three different AML samples were dramatically reduced with a mean decrease from 69% to 96% on treatment with 10 to 1,000 nmol/L NVP-BEZ235, respectively (Fig. 6A). In contrast, NVP-BEZ235 barely affected the clonogenic growth and differentiation of normal CD34⁺ hematopoietic progenitors, even at the maximal concentration of 1,000 nmol/L. The number of mixed (CFU-GEMM), erythroid (BFU-E), or granulo-monocytic (CFU-GM) colonies was not significantly decreased by NVP-BEZ235 (Fig. 6A).

We next determined the effect of NVP-BEZ235 on AML cell survival. NVP-BEZ235 was found to induce from a 100 nmol/L dose a significant apoptosis response in primary AML samples and in MV4-11 cells, as evidenced by an increase in Annexin V staining (Fig. 6B). Conversely, NVP-BEZ235 barely affected the survival of normal immature CD34⁺ cells, even at the higher concentration of 1,000 nmol/L (Fig. 6B). The proapoptotic effects of NVP-BEZ235 in AML cells correlated with a decreased expression of the antiapoptotic protein MCL-1, suggesting an induction of the mitochondrial apoptotic pathway. We also detected a cleavage of caspase-3 and of its substrate poly(ADP-ribose) polymerase (PARP) by Western blotting in primary AML samples and in MV4-11 leukemic cells (Fig. 6C). Taken together, these results indicate that NVP-BEZ235 preferentially impairs the survival of leukemic cells when compared with normal hematopoietic cells, suggesting a favorable therapeutic index within the hematopoietic system in vivo for this compound.

Despite the considerable improvements in our knowledge of AML biology in recent years, the treatment of this disease remains a challenge for clinicians. Major efforts have been made to develop new compounds that target the preferentially activated signaling pathways implicated in AML cell proliferation and survival. The PI3K/Akt/mTOR axis represents a promising therapeutic target in several cancers, as many components of these signaling pathways are frequently deregulated in tumor cells. We recently showed that a constitutive activation of PI3K is detected in ∼50% of primary AML samples at diagnosis(8), mostly due to an autocrine stimulation of the IGF-1 receptor (9). Accordingly, either the blockade of this IGF-1 autocrine production or the specific inhibition of PI3K p110δ activity leads to a strong decrease of AML cell proliferation (4, 5, 9). The constitutive activation of the mTORC1 pathway is detectable in virtually almost all AML patient samples, which emphasized the potential of using mTOR inhibitors as therapeutic agents in these tumors. However, first-generation mTOR inhibitors of the rapalog family, which are highly specific mTORC1 allosteric inhibitors, produced heterogeneous results in clinical trials (20). In AML, the therapeutic value of rapalogs seems weak, at least in monotherapy settings (24). One possible strategy to enhance their effectiveness consists of dual inhibition of both the PI3K and mTOR signaling pathways (10, 34). Indeed, although the constitutive activation events for PI3K and mTORC1 are independent (25), the PI3K and mTOR signaling networks are closely related and are subject to complex cross-talk and feedback interactions. In this regard, mTORC1, through the phosphorylation of P70S6K on Thr³⁸⁹, decreases PI3K activity by inducing the degradation of molecular adaptors of the IRS family (43). Accordingly, the RAD001-mediated inhibition of mTORC1 increases PI3K activity, which contributes to a reduction in the antileukemic activity of RAD001 (10). Moreover, both PI3K and mTORC2 regulate the full activity of the oncogenic kinase Akt via the phosphorylation of Thr³⁰⁸ and Ser⁴⁷³, respectively (29, 37). These observations underscore the rationale for a dual PI3K and mTORC1 inhibition strategy to treat AML.

Accordingly, in our current study, we tested the antileukemic activity of the NVP-BEZ235 compound, an ATP-competitive inhibitor that directly targets the PI3K and mTOR catalytic domains. This compound has already been tested in other cancer models (44, 45) or other hematologic malignancies (46–48), and promising preclinical data were reported. Interestingly, in AML, NVP-BEZ235 showed marked antileukemic effects in vitro, as it induced a dose-dependent decrease in AML cell proliferation, inhibited the clonogenic growth of AML progenitors, and significantly induced apoptosis in AML cells. Furthermore, these effects were highly specific to primary AML cells, as NVP-BEZ235 did not affect normal hematopoiesis.

As expected, NVP-BEZ235 strongly inhibited the PI3K and mTORC1 signaling pathways, which was evidenced by the decrease of Akt Thr³⁰⁸ and P70S6K Thr³⁸⁹ phosphorylation, respectively, in AML cells. Moreover, we did not observe an S6K-dependent increase in Akt Thr³⁰⁸ phosphorylation in NVP-BEZ235–treated AML cells due to the concomitant inhibition of PI3K and mTOR kinases. Furthermore, NVP-BEZ235 did not induce off-target effects against other signaling pathways that are frequently deregulated in primary AML cells, such as ERK/MAPK and STAT5, emphasizing the specificity of this compound against class IA PI3K and mTOR activities. We also show from our present data that NVP-BEZ235 inhibits mTORC2 activity in AML. This was first suggested by the fact that NVP-BEZ235 fully inhibits mTOR Ser²⁴⁸¹ phosphorylation, which reflects the mTOR catalytic activity regardless of its association to mTORC1 or mTORC2 complexes. Second, NVP-BEZ235 strongly suppresses the phosphorylation of the well-characterized mTORC2 substrates Akt Ser⁴⁷³ and NDRG1 Thr³⁴⁶ in AML cells. However, Akt and NDRG1 phosphorylation are also dependent on PI3K activity (39–41), and accordingly, the blockade of PI3K by the specific p110δ inhibitor IC87114 in primary AML cells also suppressed Akt Ser⁴⁷³ phosphorylation (4, 5). To more precisely elucidate the activity of NVP-BEZ235 against mTORC2, we used paxillin Tyr¹¹⁸ phosphorylation as a marker of mTORC2 activity, as the siRNA-mediated decrease of rictor expression markedly inhibits this phosphorylation event in primary AML samples. Interestingly, a parallel was observed between decreased Akt Ser⁴⁷³ and paxillin Tyr¹¹⁸ phosphorylation in AML cells treated with NVP-BEZ235, which confirmed the inhibition of the mTORC2 activity. Although only recently suggested, the role of mTORC2 in oncogenesis is now established (19). In AML, Zeng and colleagues (27) have shown that mTORC2 controls Akt activation by showing that the disruption of mTORC2 following prolonged exposure to rapalogs inhibits Akt Ser⁴⁷³ phosphorylation. This suggests that mTORC2 inhibition plays a role in the antileukemic activity of these molecules (27). We thus speculate that the targeting of mTORC2 may contribute to the antileukemic activity of NVP-BEZ235.

The critical mechanism underlying NVP-BEZ235 action, however, may also result from the inhibition of protein translation. We recently emphasized that the weak antileukemic activity of the rapalogs was mainly due to the sustained high level of 4E-BP1 phosphorylation in AML cells treated with these compounds (25). Conversely, we show herein that in contrast to rapamycin, NVP-BEZ235 strongly blocks mTOR catalytic activity. This results in the inhibition of 4E-BP1 Thr^37/46 phosphorylation residues, which have been reported to be direct substrates for mTOR catalytic activity (49). As 4E-BP1 phosphorylation follows a hierarchical process in which the priming phosphorylation on Thr^37/46 residues is rate limiting (16, 50), the inhibition of mTOR catalytic activity results in a complete inhibition of 4E-BP1 phosphorylation on Ser⁶⁵ and Thr⁷⁰. These results thus provide a basis for understanding the differential effects between rapalogs and mTOR catalytic inhibitors toward 4E-BP1 phosphorylation and translation inhibition in AML. Indeed, phosphorylation of the translation regulator 4E-BP1 is the limiting step for the assembly of the translation-initiating complexes. As a consequence of 4E-BP1 dephosphorylation, NVP-BEZ235 markedly inhibits the assembly of eIF4F-initiating complexes and reduces the recruitment of mRNA molecules to polysomes. This results in a global inhibition of protein synthesis and thus a decreased expression of oncogenic proteins regulated at the translation initiation level (e.g., c-Myc, cyclin D1, and Bcl-xL). As we have recently shown that the deregulation of oncogenic mRNA translation is of major importance in the oncogenic process in AML (25), we now speculate that an important component of the antileukemic activity of NVP-BEZ235 in AML is the inhibition of oncogenic mRNA translation.

Overall, our present findings firmly establish that NVP-BEZ235 fulfills all of the biological criteria for rapid clinical testing. Indeed, NVP-BEZ235 inhibits the PI3K, mTORC1, and mTORC2 signaling pathways, leading to the death of AML cells. Moreover, from the perspective of clinical development, NVP-BEZ235 barely affects the differentiation or survival of normal CD34⁺ hematopoietic progenitors, strongly indicative of a favorable therapeutic index for this molecule in vivo. A phase I/II clinical trial with NVP-BEZ235 is now in progress for advanced solid tumors (ClinicalTrials.gov NCT00620594), and the preliminary data from these trials suggest that the general tolerance of this orally bioavailable molecule is good. We provide here a strong rationale for undertaking clinical trials of NVP-BEZ235 in AML patients also in the very near future.

S-M. Maira: Novartis Pharma employee and shareholder. The other authors disclosed no potential conflicts of interest.

We thank all participating investigators from the GOELAMS.

Grant Support: Ligue Nationale Contre le Cancer (Comité de Paris), Institut National du Cancer, the Fondation de France, and Association Laurette Fugain. J. Tamburini and A.S. Green are recipients of grants from the Fondation pour la Recherche Medicale, S. Park from Assistance Publique des Hôpitaux de Paris/La Caisse Nationale d'Assurance Maladie, and N. Chapuis from Institut National de la Santé Et de la Recherche Médicale.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.