Targeting PIM Kinases Impairs Survival of Hematopoietic Cells Transformed by Kinase Inhibitor–Sensitive and Kinase Inhibitor–Resistant Forms of Fms-Like Tyrosine Kinase 3 and BCR/ABL

A high percentage of acute leukemias express functional class III receptor protein tyrosine kinase (PTK) Fms-like tyrosine kinase 3 (FLT3). Approximately 25% of acute myeloid leukemia (AML) patients have activating FLT3 mutations, either internal tandem duplications (ITD) or mutations in the activation loop of the tyrosine kinase domain (1). Overexpression and/or activating point mutations have also been found in acute lymphoblastic leukemia (ALL) from infants associated with rearrangement of the MLL gene. Activating mutations of the FLT3 tyrosine kinase are the most common somatic genetic alterations in acute leukemia, and there is evidence that patients with FLT3-ITD may be associated with poorer prognosis. The biological effects of activating FLT3 mutations have been tested in several in vitro and in vivo experimental systems, indicating that activating FLT3 mutants provide growth factor–independent proliferation and survival to hematopoietic cells in vitro and in vivo (2).

Several new compounds with activity against FLT3 by competing with ATP binding have been recently characterized (3). These drugs were shown to be selectively cytotoxic to cells either transfected with FLT3-ITD or to primary AML samples carrying FLT3-ITD. In vivo application of three different compounds (CEP-701, PKC-412, and MLN-518) has been shown to prolong survival of mice transplanted either with FLT3-ITD–expressing marrow or FLT3-ITD–transfected Ba/F3 cells (4). However, these compounds show limited activity against activation loop FLT3 mutations (5, 6), and resistance to these inhibitors can develop rapidly (7, 8). The molecular mechanisms of hematopoietic transformation by mutated FLT3 are not well understood. Significant differences have been observed in downstream signaling either induced by wild type (WT) or by constitutive active mutated FLT3. First, restoration of the apoptosis inhibitor Bcl-XL was critical for survival of hematopoietic cells by FLT3-ITD but not WT-FLT3. Second, hematopoietic cells transformed by activating FLT3 mutants showed constitutive activation of the signal transducer and activation of transcription 5 (STAT5), whereas transfectants with WT-FLT3 showed only weak or transient activation of STAT5 when exposed to FLT3 ligand (9). Moreover, STAT5 phosphorylation and activation is inhibited by selective FLT3 PTK inhibitors. These findings suggest that regulation of antiapoptotic proteins like Bcl-XL in conjunction with activation of STAT5 may be critical for the leukemogenic effects of oncogenic FLT3 mutants.

Constitutive activation of STAT5 has been shown to be a critical mediator for transformation by several PTK fusion genes, including BCR/ABL, FLT3, or TEL/JAK2 associated with chronic myeloproliferative disorders (10). Several downstream STAT5 target genes (such as SOCS1, BCL-XL, and PIMs) were shown to play a role in cancer. The PIM family of serine/threonine kinases (PIM-1 to PIM-3) has been originally identified as proviral integration sites involved in lymphomagenesis induced by murine leukemia virus (11). PIM kinases are highly homologous to each other sharing a unique consensus hinge region sequence, but they have <30% sequence identity to other kinases. Determination of the crystal structure of PIM-1 revealed that PIMs adopt a constitutive active confirmation based on extensive hydrophobic and hydrogen bond interactions (12). PIM-1 has been shown to protect hematopoietic cells from apoptosis but to be dispensable for BCR/ABL–mediated transformation (13, 14). In addition, a recent study has suggested that PIM-1 up-regulation by constitutive activated FLT3 may play a role in FLT-3–mediated survival (15). PIM-2 has been shown to be transcriptionally regulated and to maintain cell size and mitochondrial potential independent of activation of the phosphatidylinositol 3-kinase/AKT/target of rapamycin (TOR) pathway. Its cytoprotective activity was dependent on phosphorylation of the translational repressor 4E-PB1 and the BH3 apoptotic regulator BAD (16). Increased PIM-2 expression has been found in a wide range of malignancies, including leukemia, prostate cancer, mantle cell lymphoma, and multiple myeloma (17, 18). Interestingly, gene expression profiling showed strong up-regulation of PIM-2 expression in hematopoietic cells transformed by FLT3-ITD (19). Moreover, promotion of survival of lymphoid cells by PIM-2 has been shown to be dependent on nuclear factor-κB activation (20).

Here, we have analyzed the role of PIMs in transformation of hematopoietic cells by various oncogenic PTKs and their potential as therapeutic targets in leukemia.

Cells. WT interleukin-3 (IL-3)–dependent murine Ba/F3 cells and generation of the stably factor-independent growing Ba/F3 cells transformed by PTK fusion genes (BCR/ABL, TEL/JAK2, TEL/TRKC, FLT3-ITD, and H4/PDGFβR) used in this study have been previously described (2, 21, 22). Cells were kept in RPMI 1640 with 10% FCS (Invitrogen, Carlsbad, CA) and 1 ng/mL recombinant murine IL-3 (WT-Ba/F3 cells, Endogen, Pierce Biotechnology, Woburn, MA) at 37°C in 5% CO₂. Ba/F3 cells stably expressing FLT3-D835Y, FLT3-I836del, FLT3-ITD-G697R, and FLT3 WT and the human leukemia cell lines MV4-11, K-562, REH, PL-21, and MOLM-13 have been previously characterized (5–7, 23). In addition, Ba/F3 cells transformed by BCR/ABL (p185) WT and various mutants detected in patients under imatinib therapy have been previously described (24). Cellular growth curves were established as followed: 10⁵ live cells were plated in triplicates and counted daily using a hemacytometer. Cells were diluted appropriately to be <0.5 × 10⁶/mL. Viability was analyzed by trypan blue dye exclusion. Apoptotic cell death was determined by using 7-AAD (BD Biosciences, San Diego, CA) according to manufacturer's protocol.

Gene transfer. WT and kinase-dead mutants (PIM-KD) murine PIM-1(K67M) and PIM-2(K61M) cDNAs were cloned into the retroviral expression vector pMSCV-IRES-EGFP/EYFP (19, 25). Retrovirally delivered small interfering RNAs (siRNA) targeting PIM-2 have been characterized previously (16). Retroviral stocks and transduction of Ba/F3 cells were done as described (21). Synthetic siRNAs targeting mouse PIM-2 (Santa Cruz Biotechnology, Santa Cruz, CA) and human PIM-2 (silencer validated siRNAs, Ambion, Inc., Austin, TX and Qiagen, Hilden, Germany) and synthetic siRNA targeting mouse PIM-1 (Santa Cruz Biotechnology) or human PIM-1 (Qiagen) as well as control siRNAs (Santa Cruz Biotechnology and Qiagen) were transfected using the AMAXA nucleofector following the manufacturer's instruction (AMAXA Biosystems, Cologne, Germany).

Gene expression analysis. PIM mRNA expression was quantified by SYBRGreen real-time reverse transcription-PCR (RT-PCR). 1 μg of RNase-free DNase (Qiagen)–treated total RNA (isolated with the RNeasy kit, Qiagen) was used for cDNA synthesis (AMV first-strand cDNA synthesis kit, Roche Diagnostics, Rotkreuz, Switzerland), and 50 ng cDNA was used for RT-PCR in iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) using an icycler (Bio-Rad). Expression levels were normalized to glyceraldehyde-3-phosphate dehydrogenase mRNA (ΔC_T method). Primer sequences and PCR conditions are available upon request. PIM protein expression was analyzed by Western blot analysis. Lysates (50 μg) were separated on 8% to 12% SDS-PAGE and blotted using conventional methods. Antibodies recognizing PIM-1 (clone 19F7), PIM-2 (clone 1D12), and actin (clone C-11) were purchased from Santa Cruz Biotechnology.

Previous studies have proposed PIM-1 serine/threonine kinase being a STAT5 downstream target gene (26). Based on our previous work showing that STAT5 is an important mediator of malignant transformation by TEL/JAK2 and being able to induce a lethal myeloproliferative disorder itself, we aimed to learn about the role of its putative targets, the PIM kinases, in PTK-mediated transformation (27).

First, expression levels of the three known PIM serine/threonine kinases (PIM-1, PIM-2, and PIM-3) were determined in Ba/F3 cells stably transformed to IL-3 independent growth by six constitutively active PTKs. High levels of PIM-1 and PIM-2 mRNA levels were detected in cells transformed by TEL/JAK2, TEL/TRKC, TEL/ABL, BCR/ABL, H4/PDGFRβ, and FLT3-ITD (Fig. 1A). In parental Ba/F3 cells, PIM-1 mRNA expression was strictly dependent on IL-3 stimulation and decreased rapidly after IL-3 depletion. In contrast, in several experiments using different Ba/F3 cell batches, PIM-2 mRNA expression unchanged upon removal of IL-3 (Fig. 1A). No significant PIM-3 expression levels could be detected in all cell lines tested (data not shown). Western blot analysis confirmed abundant levels of PIM-1 and low but detectable levels of PIM-2 in all cell lines tested (Fig. 1B). In parental Ba/F3 cells, depletion of IL-3 lead to a decrease in PIM-1 and PIM-2 protein expression. These data show abundant expression of PIM-1 and PIM-2 but not PIM-3 in hematopoietic cells transformed by oncogenic PTKs. In addition, it suggests that in Ba/F3 cells, PIM-2 expression, unlike PIM-1 expression, is mainly regulated at the post-transcriptional level.

Overexpression of PIM-1 is sufficient to transform hematopoietic cell lines to growth factor independence (13). Likewise, in our series of experiments, overexpression of PIM-1 or PIM-2 or both together was capable to induce growth factor–independent growth of Ba/F3 (data not shown). We also investigated if overexpression of PIM-1, PIM-2, or coexpression of both in bone marrow of mice could induce a hematologic disorder, but no disease was observed, after 10 months (data not shown). These data suggest that PIM kinases may exert their oncogenic properties in vivo only in combination with other genetic alterations.

To address whether expression of PIMs is required for transformation of Ba/F3 cells by oncogenic PTKs, PIM-KD variants were coexpressed in PTK-transformed Ba/F3 cells. PIM-KD mutants have been previously tested in vitro to lack any kinase activity and to be able when overexpressed to reduce kinase activity of the wild-type form in a dominant-negative manner (19, 25, 28). In analogy with previous reports, expression of PIM-1KD alone had no significant effect on the growth of Ba/F3 cells transformed by BCR/ABL or FLT3-ITD (Fig. 2; refs. 13, 14). Expression of PIM-2KD resulted in a decreased growth rate of Ba/F3 cells transformed by TEL/JAK2 and to lesser extent in BCR/ABL cells. In sharp contrast, expression of PIM-2KD in FLT3-ITD cells rapidly induced apoptotic cell death in several experimental settings (Fig. 2C). Expression of PIM-1KD and PIM-2KD variants was checked by Western blot analysis and by expression of the green fluorescent protein (GFP) selection marker (EGFP and EYFP) driven from an internal ribosomal entry site (IRES) in the retroviral construct (data not shown). This result confirms and extends previous observations showing that expression of PIM-2KD significantly inhibited growth of murine hematopoietic cells (32-D) transformed to growth factor independence by the FLT3-ITD (19). Expression of PIM-2KD in parental Ba/F3 cells growing in IL-3, or in Ba/F3 expressing a wild-type FLT3, stimulated with FLT3 ligand impaired cellular proliferation but did not show any significant impairment of cell survival (data not shown). This observation could reflect the fact that mice lacking all PIM kinases are viable and show only mildly impaired responses to hematopoietic growth factors (29). Surprisingly, coexpression of PIM-1KD and PIM-2KD significantly impaired growth of Ba/F3 cells transformed by TEL/JAK2 or BCR/ABL and rapidly induced cell death in FLT3-transformed cells (Fig. 2). These experiments suggest that PIM-2 and not PIM-1 may play a critical role in mediating growth factor–independent growth and survival of Ba/F3 cells by FLT3-ITD. Moreover, it suggests that combined activation of PIM-1 and PIM-2 may be critical mediators of survival of Ba/F3 cells by oncogenic PTKs.

Expression of kinase-dead mutants harbors the possibility that the observed effect is the consequence of nonspecific interaction with unknown molecules. In addition, because no isoform-specific PIM substrates have been identified, cross-reactivity with both wild-type PIMs cannot be ruled out. Therefore, we also targeted PIM-2 by using RNA interference (RNAi). In our previous work, we have identified a highly specific siRNA targeting murine PIM-2. Expression of this siRNA from the pBABE retrovirus significantly reduced PIM-2 mRNA and protein expression (16). In analogy to expression of a PIM-2KD mutant, retroviral expression of siRNAs targeting PIM-2 rapidly induced apoptotic cell death of Ba/F3 cells transformed by FLT3-ITD. In contrast, Ba/F3 cells transformed by BCR/ABL or TEL/JAK2, or overexpressing FLT3 stimulated with FLT3L, as well as parental Ba/F3 growing in IL-3 remained unaffected (Fig. 3A; data not shown). Our observation made in Ba/F3 mouse myelolymphoid cells transformed through overexpression of activating FLT3 mutations were confirmed in three human acute leukemia cell lines (Fig. 3B): MV4-11, expressing exclusively the mutated allele of FLT3 (FLT3-ITD, homozygous), MOLM-13, and PL-21 (FLT3-ITD, heterozygous; ref. 23). Although attempts to generate MV4-11 cells stably expressing PIM-2 siRNAs by using a viral-based expression system failed (data not shown), nucleofection of synthetic siRNAs targeting human PIM-2 led to induction of apoptosis of a significant fraction of MV4-11 cells. In contrast, the respective differences in cell death were still significant in MOLM-13 and PL-21 cells, consistent with their heterozygous genotype (Fig. 3B). No induction of significant apoptosis was detected by expression of PIM-2 siRNAs in REH (pre-B-ALL) and in K-562 cells [BCR/ABL-positive, chronic myeloid leukemia (CML)]. Efficacy of siRNA-mediated PIM-2 down-regulation was in the absence of a specific anti-human PIM-2 antibody only determined by real-time PCR (Fig. 3B).

Our observations made by overexpression of dominant-negative acting kinase-dead mutants as well as targeting expression by siRNAs strongly suggests that PIM-2 is a critical target for transformation of hematopoietic cells by FLT3-ITD. These results may somehow contrast a recent report that analyzed the role of PIM kinases in inhibition of FLT3-ITD by a small-molecule inhibitor (15). In their work, treatment of hematopoietic cells harboring FLT3-ITD with the CEP-701 small-molecule tyrosine kinase inhibitor lead to up-regulation of PIM-1 expression. In addition, overexpression of PIM-1 rendered Ba/F3 cells expressing FLT3-ITD more resistant to cytotoxicity induced by the CEP-701 small-molecule tyrosine kinase inhibitor. Although not analyzed, one explanation of their finding could be that CEP-701 may also inhibit one of the PIMs, presumably PIM-2, leading to a compensatory up-regulation of PIM-1. Based on this hypothesis, we were analyzing the expression level of PIMs upon down-regulation by siRNA and overexpression of the KD-mutants. Interestingly, as shown in Fig. 4A, down-regulation of PIM-1 by nucleofection of synthetic siRNAs in Ba/F3 transformed by BCR/ABL or FLT3-ITD led to a significant increase of PIM-2 mRNA expression. In contrast, we could not observe a significant change of PIM-2 mRNA expression levels upon down-regulation of PIM-1. Similar results have been obtained in Ba/F3 cells overexpressing the dominant-negative acting PIM-KD mutants (data not shown). Compensatory up-regulation of PIM-2 expression upon down-regulation of PIM-1 may explain why we do not observe a significant effect upon inhibition of PIM-1 in the PTK-transformed cells analyzed here. Upon cotransfection of different siRNAs targeting PIM-1 and PIM-2, a significant reduction in survival of Ba/F3 cells transformed by FLT3-ITD and BCR/ABL was observed confirming the results obtained by coexpression of PIM-KD mutants (Fig. 4B; data not shown). Because expression of synthetic siRNAs targeting PIM-2 lead to no significant effect in MV4-11 but BCR/ABL–positive K-562 cells we now asked whether like in the Ba/F3 cell model, interference with both PIMs would be necessary. In several experimental setups using two different siRNA targeting each human PIM, a significant decrease in viability was observed upon conucleofection with siRNAs targeting PIM-1 and PIM-2 in both cell lines (Fig. 4C). Keeping in mind that gene transfer rate by nucleofection of human cell lines is limited, this observation suggests that like in the Ba/F3 cell model, combined interference with both PIMs can significantly reduce cell growth of human BCR/ABL–expressing cells. In contrast to the experiments with Ba/F3 cells, we have observed some cross-reactivity with the applied siRNAs (Fig. 4D). This is not surprising, because PIM-1 and PIM-2 show >70% sequence homology at the nucleotide level. However, independently of any cross-reactivity, these experiments show that both PIM-1 and PIM-2 have to be targeted to obtain a significant growth reduction.

Two types of mutations lead to constitutive activation of FLT3 in acute leukemia: ITDs in the juxtamembrane region (length mutations) and point mutations in the activation loop of the PTK domain (1). Whereas FLT3 length mutations can be successfully targeted by several small-molecule inhibitors, the response of FLT3 kinase domain mutants to these compounds is variable. However, similar to cells transformed by FLT3-ITD, Ba/F3 cells transformed by the kinase domain mutations FLT3-D835Y or FLT3-I836del underwent rapidly apoptotic cell death upon expression of the siRNA targeting PIM-2 (Fig. 5A). These data further support the idea that PIM-2 is a critical downstream target for cellular transformation by mutant FLT3 either activated by length mutations (ITD) or activation loop mutations.

PTK-transformed hematopoietic cells treated by small-molecule inhibitors targeting the kinase domain can develop resistance by acquiring additional mutations in the kinase domain (30). We and others have recently identified dual mutations in the kinase domain of FLT3 (G697R) combination with FLT3 length (ITD) mutation, which confers resistance to several small-molecule kinase inhibitors, including PKC-412 and SU-5614 (7, 31). Interestingly, down-regulation of PIM-2 by RNAi also induced rapid apoptotic cell death in Ba/F3 cells expressing the dual FLT3 mutation (FLT3-ITD-G697R; Fig. 5A). These results show that down-regulation of PIM-2 by RNAi specifically induces cell death in Ba/F3 cells transformed by activating length (FLT3-ITD) or activation loop mutations (FLT3-D835Y and FLT3-I386del), as well as activated FLT3 mutants with additional mutations (FLT3-ITD-G697R) resistant to small-molecule tyrosine kinase inhibitors.

Although PIM-1 and PIM-2 were expressed at high levels in PTK-transformed Ba/F3 cells, only interference of PIM-2 function by either expression of a kinase-dead mutant or by siRNA mediated down-regulation of expression selectively abrogated survival of cells transformed with various activating mutations of FLT3. The correlate of this specificity is currently not understood. Previous studies have shown that activation of the AKT signaling pathway as determined by phosphorylation of BAD was critical for mediating survival by WT-FLT3 but not by mutant FLT3 (9). In addition, deletion of AKT or treatment with rapamycin targeting the downstream target mammalian TOR, provided a critical role of PIM-2 (and not PIM-1) for hematopoietic cell growth and survival (32). Based on these data, one could speculate that activation of AKT may not be critical for transformation by FLT3 mutant and therefore leaving PIM-2 for compensatory regulation of critical cellular functions necessary for survival.

Coexpression of PIM-1KD and PIM-2KD mutants or cotransfection of siRNAs targeting PIM-1 and PIM-2 leads to a significant decrease in survival of Ba/F3 cells transformed by (p210)BCR/ABL, the hallmark of CML. Whereas the small-molecule ABL inhibitors like imatinib are highly effective in treatment of chronic-phase CML, patients with pre-B-ALL containing the p185 BCR/ABL isoform seem to develop resistance more rapidly (33). We have therefore analyzed whether combined down-regulation of PIM-1 and PIM-2 through siRNA targeting may be able also to impair survival of hematopoietic cells transformed by BCR/ABL p185 (p185 WT) and several point mutations detected in patients that developed clinical resistance against imatinib-mesylate. As shown in Fig. 5B, cotransfection of siRNAs targeting human PIM-1 and PIM-2 led to significant decrease in viability in cells expressing p185-WT, E255Y, Y253H, and E255K, whereas cells expressing p185-H396P showed only a limited response rate. These results suggest that targeting PIMs could be of benefit to overcome therapeutic resistance against small-molecule tyrosine kinase inhibitors not only in cells transformed by activating FLT3 mutations but also by BCR/ABL fusion gene variants.

Taken together, our data suggest that inhibition by either blocking PIM-2 kinase activity through a small-molecule inhibitor or through genetic targeting by siRNA could be a new avenue to treat FLT3-mediated leukemia and/or to overcome resistance to small-molecule FLT3 inhibitors. In addition, application of a broad PIM inhibitor targeting PIM-1 and PIM-2 may have a general benefit in the treatment regime against PTK-mediated leukemias resistant to small-molecule tyrosine kinase inhibitors. Although recent determination of the crystal structure of PIM-1 could facilitate the identification of small-molecule inhibitors, it remains to be elucidated whether PIMs may exert any nonenzymatic (e.g., scaffold) function that could support PTK-transformed growth (12, 34). Alternatively, improved delivery technologies like covalent attachment of cholesterol or by protamine-antibody fragment fusions may even allow therapeutic targeting of PIMs by siRNAs (35, 36). Because animals lacking all PIM genes are viable with a mild phenotype, one could expect that this therapeutic approach targeting PIM-1 and PIM-2 could be followed with tolerable side effects (29, 37).

Grant support: Swiss National Science Foundation grant 3100A0-100758, Swiss Cancer League grant KFS-01077-09-2000, the NOVARTIS Research Foundation, and Gertrude Von Meissner Foundation (J. Schwaller).

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.

We thank Drs. A. Berns (Netherlands Cancer Center, Amsterdam), S. Baker (St. Jude Children's Research Hospital), and P. Rothman (Columbia University, New York) for providing valuable reagents and H. Kohler, M. Berczy, L. Ho, R. Kralovics, and A. Biondi for their input for the presented work.