PI3K/AKT signaling controls key signaling pathways involved in the maintenance of cellular viability and proliferation in many cells and tissues. Not surprisingly, activation of AKT is increased in many human malignancies and gain-of-function mutations are frequently found within PI3K/AKT axis, especially in solid tumors, making the PI3K/AKT signaling pathway an attractive target for molecular therapeutics.
In acute leukemia, activating mutations in the PI3K/AKT signaling cascade are rare – but nevertheless, we and others have reported frequent activation of AKT (i.e. phosphorylation of Thr308 and Ser473): In this study, we demonstrate global phosphorylation of AKT in native acute leukemia samples. Average expression levels are thereby statistically significantly elevated compared to physiologic hematopoietic mononuclear cells derived from healthy donors. Moreover, augmented expression levels are exclusively found in the leukemia cohort.
The mechanisms of AKT activation in acute leukemia are only partially understood. One mechanism of constitutive phosphorylation of AKT can be explained by the presence of gain-of-function mutant tyrosine kinases, which are found in approximately 30-40% of adult AML and ALL. However, we did not find an exclusive correlation of phospho-AKT expression levels and the presence of TK mutations, suggesting other mechanisms, which render AKT autoactivated in leukemia cells. Evaluation of the triggering mechanisms are topic of ongoing research.
Globally targeting the AKT signaling pathways may be a promising approach to treat acute leukemia. We herein evaluated the antileukemic efficacy of the novel dual PI3K/MTOR inhibitor NVP-BGT226, a pan-PI3Kinase inhibitor also targeting the rapamycin-sensitive MTOR complex 1 as well as the rapamycin-insensitive MTOR complex 2.
Using defined cell line models, and primary leukemia patient as well as donor samples we studied the distinct effects of NVP-BGT226 on cellular proliferation, cell cycle progression and induction of apoptosis. Thereby we compared NVP-BGT226 to a second dual inhibitor, NVP-BEZ235, which is currently under investigation in a phase I study for relapsed/refractory ALL or AML (European Clinical Trials Database number: EUDRACT2011-005050-61).
Our cell models included cell lines with defined genomic alterations rendering the AKT signaling pathway autoactivated, i.e. (i) a PTEN-deficient acute T-lymphoblastic leukemia cell line (Jurkat), (ii) patient-derived leukemia cell lines with well described TK-mutations (MOLM14 harboring a FLT3 ITD mutation and K562 harboring a BCR-ABL1 fusion transcript mutation), (iii) engineered Ba/F3 cell lines transfected with mutant tyrosine kinases expressed in an otherwise isogenic cellular background and (iiii) native ex vivo acute leukemia cells, with or without a defined TK-mutation, derived from consented patients with newly diagnosed acute leukemia. In addition, we comparatively studied native physiologic mononuclear cells derived from bone marrow donors.
In PTEN-deficient Jurkat cells, NVP-BGT226 proved to potently inhibit cellular proliferation in the low nanomolar range. The sensitivity profile is thereby in the same range compared to the additionally tested dual PI3K/MTOR inhibitor, NVP-BEZ235.
It was previously noted, that the predominant antitumor effect of inhibitors of PI3K/AKT/MTOR signaling cascades is mediated via inhibition of cellular proliferation rather than induction of apoptosis [32, 38, 39]. Surprisingly however, NVP-BGT226 proved to have genuine proapoptotic efficacy – whilst the proapoptotic effect achieved by NVP-BEZ235 was, as expected by previous reports, at most moderate.
To model the effects of NVP-BGT226 and NVP-BEZ235 on mutant-TK triggered AKT activation, we chose two well established acute leukemia cell lines harboring a FLT3 ITD mutation (MOLM14) or a BCR-ABL1 mutation (K562). Similar to the findings for Jurkat cells, both inhibitors, proved to be highly potent in inhibiting cellular proliferation. However again, NVP-BEZ235 only moderately induced a meaningful proapoptotic effect, whereas NVP-BGT226 was a strong inducer of the programmed cell death machinery.
As the AKT pathway controls cell cycle checkpoints, we speculated that the discrepancy may be due to differential activity on the cell cycle compartment. And indeed, a strong and sustained G0/G1 arrest was observed for NVP-BEZ235 preventing cells to undergo apoptosis.
On the protein level, where both agents were similarly targeting downstream proteins controlling cell cycle progression (such as S6Kinases and RB) or ULK1-induced autophagy, only NVP-BGT226 was capable to override cell protective mechanisms to potently induce apoptosis.
We speculated that the cell cycle arrest induced by NVP-BEZ235 might be overcome by combination approaches: TKI, for which we demonstrated insufficient global suppression of AKT signaling pathways – but additional effects on alternative survival pathways such as MAPK and STAT signaling, may be an attractive molecularly defined partner to combine with dual PI3K/MTOR inhibitors. Indeed on the protein level, combination of TKI with either of the tested dual PI3K/AKT inhibitors efficiently and globally shut down AKT signaling pathways - as well as additional targets (ERK1/2, STAT5) triggered by mutant-TKs.
In an attempt to mathematically define the extend of combination efficacy, we established isobologram assays to compute combination indices (CI). Together, calculated CIs for TKI plus dual PI3K/MTOR inhibitor treatment were close to or smaller than 1, indicating an additive to superadditive (synergistic) effect for all tested endpoints.
Notably, combination of TKI with NVP-BEZ235 was capable to override cell cycle arrest seen for NVP-BEZ235 monotherapy to potently induce apoptosis in leukemia cells.
One might speculate that cell-type specific off-target effects may have prevented cells to undergo apoptosis. To confirm our findings, we established an isogenic Ba/F3 cell line model transfected with FLT3 ITD (corresponding to MOLM14 cells) or BCR-ABL1 (corresponding to K562 cells) mutations. NVP-BGT226 revealed high potency to inhibit cellular proliferation in the same range as NVP-BEZ235.
As expected, while meaningful proapoptotic effects were achieved by NVP-BGT226 in all cell strains, FLT3 ITD and BCR-ABL1 transfected Ba/F3 cells were only moderately sensitive towards NVP-BEZ235.
We additionally created several more Ba/F3 cell lines transfected with tyrosine kinases harboring known leukemia-driving gain-of-function mutations and tested for NVP-BGT226 and NVP-BEZ235 sensitivities. While NVP-BGT226 again displayed a beneficial pro-apoptotic profile for all tested transfectants, NVP-BEZ235 surprisingly retained meaningful proapoptotic activitiy in some cell strains. Two sensitive transfectants (harboring a FLT3 D835V or KIT D816Y mutation) were immunoblotted – and showed higher elevated threonine 308 phosphorylation levels compared to FLT3 ITD or BCR-ABL1 transfected cells.
This observation may have far-reaching consequences: It is tempting to speculate that activation of the PI3K/AKT pathway is at least in part dependent on the specific type of TK gain-of-function mutation and that different gain-of-function mutations may display a very distinct pattern of activated PI3K/AKT signaling cascades. This again might influence the susceptibility of cells towards PI3K/AKT-targeted inhibitors. In this context, it is well described for TKI therapy of CML and GIST and has recently been shown for TKI therapy in acute leukemia as well, that resistance towards TK-inhibitors is often caused by secondary mutations within the tyrosine kinase domain (such as point mutations at FLT3 D835) of the respective tyrosine kinase . Such mutations may activate AKT signaling, as previously demonstrated for imatinib-resistant GIST tumors , and sensitize cells towards targeted therapies.
We tested this theory using two cell models comparing primary TK-sensitive mutations with secondary TK-insensitive mutations: The first model consists of a mast cell leukemia cell line (HMC1.1), which harbors an imatinib-sensitive KIT V560G mutation – and a derivative sister cell line (HMC1.2), which is characterized by a secondary activation loop KIT D816V mutation, rendering the cells insensitive towards imatinib [42, 43]. Additionally we tested the GIST solid tumor cell line GIST882 (harboring an imatinib-sensitive KIT K642E mutation)  with a second cell line, which was established from a patient with relapsing GIST under imatinib therapy (GIST48) . This cell line harbors a primary homozygous juxtamembrane KIT mutation (V560D) plus a secondary heterozygous imatinib-insensitive activation loop mutation (D820A).
Indeed, in our experiments, NVP-BEZ235 as well as NVP-BGT226 potently induced apoptosis irrespective of the sensitivity profile towards TKI – with NVP-BGT226 again being the more potent inhibitor (data provided as Additional file 3: Figure S2 with the online version of the article). Together, dual PI3K/MTOR inhibitors such as NVP-BGT226 or NVP-BEZ235 may be of special clinical value in the desperate case of tumor progress due to TKI-resistance, which is an ever increasing problem in the treatment of relapsed acute leukemia. The underlying molecular mechanisms determining the susceptibility of cells towards induction of apoptosis as well as sensitivity towards NVP-BGT226 or NVP-BEZ235 (e.g. higher binding affinities and alternative (unknown) targets) is elusive and will need to be answered in future studies.
Most importantly however, we did show that dual inhibition of pan class I PI3Kinases plus MTOR1/2 complexes does translate into a genuine antiproliferative but also proapoptotic effect in native leukemia cells treated ex vivo – with NVP-BGT226 being the more potent drug with regard to induction of apoptosis. Augmented phosphorylation of AKT rather than mere expression of AKT protein levels seemed to be a prerequisite for treatment response. However, this observation will need prospective validation. Furthermore, efficacy was not restricted to leukemia samples with identified genomic mechanisms of AKT activation (such as tyrosine kinase mutations), suggesting alternative mechanisms of activation yet to be identified.
Of note, among the native leukemia samples treated successfully ex vivo with either agent were cases from patients with poor prognostic features lacking effective therapeutic options. For example, both agents were effective in AML with mutant FLT3, including a patient with TKI-resistant FLT3 ITD (B1 sheet)-positive AML  who had relapsed after allogeneic stem cell transplantation.
Other refractory AML cases with ex vivo sensitivity of cells to PI3K/MTOR inhibition included a relapsed elderly patient with MLL-rearranged AML. In this context, it has been shown that MLL rearrangements associate with high EVI1 expression, which predicts for dismal prognosis . Further, Yoshimi and colleagues recently have demonstrated that EVI1 activates AKT signaling due to loss of PTEN activity . As there are currently no effective therapy options for treatment of EVI1-associated AML, targeting the PI3K/AKT/MTOR pathway may be particularly of interest.
Preliminary data of an early phase I trial of NVP-BEZ235 in the treatment of advanced unresectable solid tumors demonstrated good tolerability with no dose-limiting toxicities. Notably, hematologic side effects were seen – but were mild to moderate with reversible anemia after treatment discontinuation . Currently, a study evaluating efficacy of NVP-BEZ235 in acute leukemia is recruiting (European Clinical Trials Database number EUDRACT2011-005050-61).
In our studies, NVP-BGT226 proved to be the more effective agent with regard to antileukemic efficacy. Ex vivo treatment revealed IC50s in the nanomolar or lower micromolar range and thus NVP-BGT226 may be an attractive agent for targeted treatment of acute leukemias.
A very recent phase I study evaluating NVP-BGT226 in advanced solid tumors demonstrated variable antitumor activity . In this context, another recent report demonstrated that NVP-BGT226 results in cell cycle arrest in pancreatic cancer cell lines , which is in clear contrast to our findings. This may argue for the rather low antitumor efficacy reported in the above mentioned phase I trial in advanced solid tumors. Our data clearly states a differential biological behavior of acute leukemia cells with regard to regulation of cell growth, cell cycle progression and induction of apoptosis, which may still support specific clinical testing of NVP-BGT226 in acute leukemia.
Moreover, in our studies, normal mononuclear cells were significantly less inhibited by dual PI3K/MTOR inhibition than leukemia cells, indicating a therapeutic gap of these agents in the treatment of acute leukemia without significant suppression of normal hematopoiesis. Nevertheless, as NVP-BGT226 targets physiologic cells in the highest tested doses, clinical evaluation will need to address potential side effects on the hematopoietic progenitor/stem cell pool. However, even in the case of significant stem cell suppression, NVP-BGT226 may still serve as an attractive agent for bridging-to-transplant strategies or allogeneic transplant conditioning regimens – especially for high-risk or elderly patients lacking other options.