Abstract
The serine/threonine kinase Pim is known to promote cell cycle progression and to inhibit apoptosis leading to tumorigenesis. However, the precise mechanisms remain unclear. We show, herein, that all the Pim family members (Pim1, Pim2, and Pim3) bind to and directly phosphorylate the cyclin-dependent kinase inhibitor p27Kip1 at threonine-157 and threonine-198 residues in cells and in vitro. The Pim-mediated phosphorylation induced p27Kip1 binding to 14-3-3 protein, resulting in its nuclear export and proteasome-dependent degradation. Ectopic expression of Pim kinases overcome the G1 arrest mediated by wild-type p27Kip1 but not by phosphorylation-resistant T157A-p27Kip1 or T198A-p27Kip1. In addition to the posttranslational regulations, p27Kip1 promoter assay revealed that Pim kinases also had the ability to suppress p27Kip1 transcription. Pim-mediated phosphorylation and inactivation of forkhead transcription factors, FoxO1a and FoxO3a, was involved in the transcriptional repression of the p27Kip1 gene. In contrast, inhibition of Pim signaling by expressing the dominant-negative form of Pim1 increased nuclear p27Kip1 level and attenuated cell proliferation. Because the CDK inhibitor p27Kip1 plays a crucial role in tumor suppression by inhibiting abnormal cell cycle progression, Pim kinases promote cell cycle progression and tumorigenesis by down-regulating p27Kip1 expression at both transcriptional and posttranslational levels. [Cancer Res 2008;68(13):5076–85]
- Pim kinase
- p27Kip1
- FoxO transcription factor
- cell cycle
- cancer chemotherapy
Introduction
Human cancer remains a serious disease, and at present, there is still no fully critical chemotherapeutic strategy against it. Tumorigenesis occurs when proto-oncogenes are activated and tumor suppressor genes are, frequently, mutated and inactivated. Most of the genes are associated with cell cycle progression and cell growth. Understanding how proto-oncogenes and tumor suppressor genes contribute to cancer cell proliferation and tumorigenesis could be promising for anticancer drug development.
The serine (Ser)/threonine (Thr) kinase pim gene is a proto-oncogene ( 1). The pim gene has been identified as a common integration site of the Moloney murine leukemia virus ( 2). Pim kinase has three family gene products (Pim1, Pim2, and Pim3; ref. 3). They are composed of at least six translational variants: Pim1 short isoform (Pim1S), Pim1 long isoform (Pim1L), Pim2 short isoform (Pim2S), Pim2 medium isoform (Pim2M), Pim2 long isoform (Pim2L), and Pim3 in human. Pim kinases have been implicated in the control of tumorigenesis. In pim1 or pim2 transgenic mice, Pim has been shown to enhance the development of lymphoma and leukemia ( 2, 4– 6). The expression level of Pim kinases is frequently elevated in patients with lymphoma, leukemia, and prostate cancer ( 4, 6, 7). In particular, the expression of Pim1 has been correlated with measures of clinical outcome ( 8). Moreover, pim1 was recently identified as a target of aberrant somatic hypermutation in non–Hodgkin's lymphoma and B-cell lymphoma ( 9– 11), and some of the mutations significantly increase Pim1 enzymatic activity ( 12). Thus, it is suggested that Pim kinase overexpression and activation induce tumorigenesis.
Pim kinases have also been implicated to induce cell cycle progression and cell growth. Overexpression of Pim kinases is reported to promote cell cycle and cell growth, whereas pim knockdown delays them ( 7, 13, 14). In pim1, pim2, and pim3 triple knockout mice, the cell number and body size was decreased ( 15). Pim1 promotes cell cycle progression at the G1-S and G2-M transitions. The G1-S–stimulating phosphatase CDC25A was found to be a substrate of Pim1, and it has been shown that it could be activated through phosphorylation by Pim1 ( 16). Furthermore, Pim1 phosphorylated G2-M–stimulating phosphatase CDC25C and increased its phosphatase activity ( 17). Moreover, Pim1 was reported to phosphorylate C-TAK1 and inactivate its kinase activity, resulting in CDC25C activation ( 18). However, the mechanisms by which Pim kinases stimulate cell cycle progression and cell growth are not fully understood. Therefore, we tried to identify novel substrates of Pim1 and to elucidate the mechanism by which Pim1 induced cell cycle progression and cell growth.
We identified the CDK inhibitor p27Kip1 as a novel substrate of Pim kinases. p27Kip1 is known to be a key molecule that modulates cell cycle progression at the G1-S transition ( 19– 24). The expression of p27Kip1 is regulated at both transcriptional and posttranslational levels ( 25– 28). From the observation of p27Kip1 knockout mice, p27Kip1 behaved like a tumor suppressor ( 29). Furthermore, reduced p27Kip1 expression is frequently observed in most human tumors. The reduced expression of p27Kip1 was reported to correlate with tumor progression and poor patient survival ( 30). Here, we showed that Pim kinases phosphorylated and down-regulated p27Kip1 at the posttranslational level. We also showed that Pim kinases repressed p27Kip1 expression at the transcriptional level by inactivating Forkhead transcription factors. Moreover, we found the inverse correlation between pim1 and p27Kip1 mRNA expression in patients with cancer. These results indicate that down-regulation of p27Kip1 expression is one of the important mechanisms of Pim-mediated cell cycle progression and tumorigenesis.
Materials and Methods
Reagents and cell culture conditions. Imatinib (Glivec or Gleevec, formerly STI571) was kindly provided by Novartis (Basel, Switzerland). LY294002 and G418 were purchased from Sigma. MG132 was purchased from Wako. Human embryonic kidney 293T cells were cultured in DMEM supplemented with 10% FBS. Human fibrosarcoma HT1080, human prostate carcinoma 22Rv1 and human myelogenous leukemia K562 cells were cultured in RPMI 1640 supplemented with 10% FBS. To assess cell viability, the MTS assay was employed (Promega). The absorbance was measured at 490 nm with a reference at 690 nm, using a microplate-spectrophotometer (Benchmark Plus; Bio-Rad).
Plasmid construction. Information on the plasmids used can be found in the Supplementary Materials and Methods.
Transient transfection, immunoprecipitation, and Western blot analysis. HEK293T cells were transfected with the appropriate plasmids using LipofectAMINE 2000 (Invitrogen). HT1080 cells were transfected using FuGENE6 (Roche). K562 cells were transfected using Cell Line Nucleofector Kit V and Amaxa-Nucleofector (Amaxa).
The preparation of cell lysates for immunoprecipitation and Western blot analysis and of the whole-cell lysates were done as previously described ( 19, 20). The nuclear and cytoplasmic fractions were separated using an NE-PER kit (Pierce). For immunoprecipitation, we used antibodies to p27Kip1 (C-19; sc-528; Santa Cruz Biotechnology), FoxO3a (Cell Signaling Technology), or FLAG-tag (clone M2; Sigma). In some experiments, cell lysates were incubated with normal mouse IgG-conjugated agarose (Santa Cruz Biotechnology) or protein A agarose that had been conjugated with normal rabbit IgG. Then, the immunoprecipitated proteins or the cell lysates were electrophoresed and blotted onto a nitrocellulose membrane. The membranes were incubated with antibodies to Pim1 (12H8; sc-13513), p27Kip1 (C-19; sc-528), p-p27Kip1 [(Ser10)-R; sc-12939-R], p-p27Kip1 [(Thr187)-R; sc-16324-R], GFP (B-2; sc-9996), actin (C-2-HRP; sc-8432 HRP), 14-3-3β (H-8; sc-1657), 14-3-3𝛉 (C-18; sc-7683; Santa Cruz Biotechnology), FoxO3a, phospho-FoxO1 (Thr24)/FoxO3a (Thr32), phospho-FoxO3a (Ser253), phospho-(Ser/Thr) Akt substrate (Cell Signaling Technology), human Topo IIβ (BD Transduction Laboratories), and α-tubulin (Serotec). The membranes were incubated with HRP-conjugated secondary antibody. After washing, the membranes were developed with an enhanced chemiluminescence system (GE Healthcare). Blots were scanned using an Image Reader LAS-3000 mini (Fuji Film) and were quantified using Multi Gauge software.
Small interfering RNA design and transfection. The pim1 small interfering RNA (siRNA) sequence can be found in the Supplementary Materials and Methods. Cells were transfected with the siRNAs using the LipofectAMINE RNAi MAX (Invitrogen).
Purification of recombinant GST-Pim1S and p27Kip1 proteins. The details of the purification method used can be found in the Supplementary Materials and Methods.
In vitro kinase assay. The details of the method used can be found in Supplementary Materials and Methods. The reactions were electrophoresed and stained with CBB. After staining, the levels of incorporated radioactivity were visualized and quantified with a Typhoon 9410 (GE Healthcare).
For the CDK2 assay, cells were solubilized with lysis buffer [0.2% NP40, 150 mmol/L sodium chloride, 50 mmol/L Tris-HCl (pH 7.6), 1 mmol/L phenylmethylsulfonylfluoride, and 1 mmol/L aprotinin] and CDK2 was immunoprecipitated using an anti-CDK2 antibody (Upstate Biotechnology). The kinase activity in the immunoprecipitated CDK2 was estimated using cdk1/cdc2 kinase assay kit (Upstate Biotechnology). The reactions were electrophoresed and stained with CBB. The levels of incorporated radioactivity were visualized and quantified using a Typhoon 9410.
Immunostaining. HT1080 cells were transfected with the appropriate plasmids. After transfection for 24 h, cells were fixed in 4% formaldehyde in PBS. After washings and permeabilization, the HA-Pim1S was detected by staining with rat anti-HA antibody, following incubation with Oregon green–conjugated anti-rat antibody (Molecular Probes). The FLAG-p27Kip1 were detected by staining with rabbit anti-FLAG antibody (Sigma), following treatment with Alexa Flour 568–conjugated anti-rabbit antibody (Molecular Probes). Nuclei were also detected by staining with Hoechst 33342. The details for immunostaining have been previously described ( 19, 20).
RNA preparation and real-time PCR. Total RNAs from HEK293T or K562 cells were extracted using RNeasy Mini Kit (Qiagen). RNA was reverse-transcribed by First-Strand Synthesis SuperMix (Invitrogen). Then, the amount of p27Kip1 mRNA was quantified by using PCR LightCycler 480 (Roche) and normalized by the amount of glyceraldehyde-3-phosphate dehydrogenase mRNA. The sequence of the primers for p27Kip1 and glyceraldehyde-3-phosphate dehydrogenase can be found in the Supplementary Materials and Methods.
Luciferase assay. HEK293T cells were transfected with the appropriate plasmids and a pGL4.10 vector encoding p27Kip1 promoter (bases 2873-3552, GenBank accession no. AB003688; ref. 26) and pGL4.74 (Promega) as internal controls. After transfection, the cells were lysed. Luciferase activity was measured using a PicaGene Luciferase assay system (Toyo-Inc. Co., Ltd.) and LB 960 Microplate Luminometer Centro (Berthold).
Flow-cytometric analysis of transfected cells. HEK293T cells were transfected with the appropriate plasmids. After transfection for 36 h, the cells were stained with propidium iodide. Analyses were performed using a Cytomics 500 flow cytometer (Beckman Coulter). The details of the method used have been previously described ( 31).
Expression analysis on cDNA filter array. Cancer Profiling Array II was purchased from Clontech. The details of the method can be found in Supplementary Materials and Methods.
Results
Relationship between Pim1 expression and cell cycle progression. A previous study reported that the amount of Pim1S increased at the G1-S transition in chronic myelogenous leukemia (CML) BV173 cells ( 32). We confirmed the increase of endogenous Pim1S at the G1-S transition in another CML K562 cells after cell synchronization (data not shown). When HEK293T cells were transfected with a wt-Pim1S–expressing plasmid, decrease in the cell number at the G1 phase and increase in the cell number at S phase were observed ( Fig. 1A ). These results suggested that Pim could promote the G1-S transition.
Pim kinases promote the G1-S transition by up-regulating CDK2 activity. A, HEK293T cells were transfected with empty pc5FLAG (Mock) or pc5FLAG-encoding wt-Pim1S (Pim1S). After transfection for 36 h, cells were stained with propidium iodide and analyzed using a flow cytometer. B, the polyclonal K562 cells that had been transfected with empty pc5FLAG (Mock) or pc5FLAG encoding wt-Pim1S (wt-Pim1S) were cultured in normal growth medium. CDK2 proteins were immunoprecipitated from polyclonal K562/Mock or K562/wt-Pim1S cells and were subjected to CDK2 kinase assay, as described in Materials and Methods. The reaction samples were electrophoresed and stained with CBB (right, middle). After staining, the levels of incorporated radioactivity were visualized (right, top) and quantified (left) with Typhoon 9410. The amount of immunoprecipitated CDK2 was also shown (right, bottom). Columns, mean of three independent experiments; bars, SD (left). C, the expression level of endogenous (endo) and exogenous (exo) Pim1S, p27Kip1, and β-actin proteins in polyclonal K562/Mock (Mock) or K562/wt-Pim1S (wt-Pim1S) cells was confirmed by immunoblot analysis.
To clarify the mechanisms of G1-S transition, we estimated CDK2 kinase activity in the polyclonal K562/wt-Pim1S and K562/Mock cells because CDK2 activity is the key regulator for G1-S transition. We found that CDK2 activity was up-regulated in the Pim1S transfectant but that the amount of CDK2 was unaffected ( Fig. 1B). It has been well-known that CDK2 activity is regulated by other regulatory factors. Therefore, we examined whether Pim1S down-regulated the CDK inhibitor p27Kip1 expression or not. As shown in Fig. 1C, Pim1S overexpression down-regulated the expression level of p27Kip1. These results suggest that Pim1S stimulates cell cycle progression at the G1-S phase, possibly by down-regulating p27Kip1 and activating CDK2.
Pim binds to and phosphorylates p27Kip1 in vitro and in cells. The concentration of p27Kip1 is known to be transcriptionally and posttranslationally regulated ( 25– 28). To clarify the mechanisms of Pim-mediated down-regulation of p27Kip1, we sought to check whether p27Kip1 was directly phosphorylated by Pim kinases because overexpression of Pim1S led to a decreased p27Kip1 level ( Fig. 1C).
Putative substrate target sequences of Pim have been identified using a chemically synthesized peptide library ( 33, 34). Using an in silico search, we identified two high homologues to Pim consensus sequences around threonine-157 (T157) and threonine-198 (T198) residues in p27Kip1 ( Fig. 2A, left ). Then, we investigated the interaction between Pim and p27Kip1 by immunoprecipitating analysis. Figure 2B (left) shows that HA-tagged p27Kip1 was coimmunoprecipitated with FLAG-tagged Pim1S, suggesting that Pim kinases interact with p27Kip1 in cells. To confirm that Pim1S forms a complex with p27Kip1 at endogenous expression levels, we performed the immunoprecipitation following Western blot analysis using K562 and 22Rv1 cells ( Fig. 2B, right) because it has already been shown that Pim1S was up-regulated and potentially had a strong role in the progression of these cancers ( 7, 8, 35, 36). Endogenous Pim1S could be detected in endogenous p27Kip1 immunoprecipitants ( Fig. 2B, right, lanes 2 and 5). These results indicate that Pim1S physiologically interacts with p27Kip1 in cells.
Pim kinases bind to and directly phosphorylate p27Kip1. A, alignment of the amino acid sequence of the Pim consensus sequence with the primary sequences of human p27Kip1 around T157 and T198 residues. Identical residues are denoted by white letters on black background (left). HEK293T, K562, and 22Rv1 cells were transfected with empty pHM6 alone (−) or pHM6 encoding wt-Pim1S (wt) or KD-Pim1S (KD) together with pFLAG-CMV-2 encoding p27Kip1. After transfection for 24 h, the cells were harvested and lysed. The whole cell lysates were electrophoresed and immunoblotted with the indicated antibodies. B, HEK293T cells were transfected with empty pc5FLAG alone (−) or pc5FLAG encoding PIm1S (+) together with pHM6-p27Kip1. After transfection for 24 h, the cells were lysed. The FLAG-tag protein was immunoprecipitated from the cell lysates. The immunoprecipitated proteins (IP) and the cell lysates (Input) were subjected to immunoblot analysis with the indicated antibodies (left). The nuclear fraction of K562 and 22Rv1 cells were incubated with protein A agarose that had been conjugated with normal rabbit IgG (IgG), rabbit anti-p27Kip1 antibody (p27Kip1), or rabbit anti-FoxO3a antibody (FoxO3a). The immunoprecipitated proteins were subjected to immunoblot analysis with the indicated antibodies (right). C, HEK293T cells were transfected with empty pHM6 alone (−) or pHM6-Pim1S (+) together with empty pFLAG-CMV-2 (−) or pFLAG-CMV-2 encoding wt-p27kip1 or the indicated p27Kip1 mutants. After transfection for 24 h, the cell lysates were electrophoresed and immunoblotted with the indicated antibodies. D, purified recombinant wt-p27Kip1 (lane 1), S10A-p27Kip1 (lane 2), T157A-p27Kip1 (lane 3), T187A-p27Kip1 (lane 4), T198A-p27Kip1 (lane 5), and GST (lane 6) proteins were incubated with GST-tagged recombinant Pim1S for 30 min at 30°C in the presence of [γ-32P]ATP. The reaction samples were electrophoresed and stained with CBB (bottom). After staining, the levels of incorporated radioactivity were visualized with Typhoon 9410 (top).
We then examined whether Pim kinases directly phosphorylated p27Kip1. The consensus sequence of Pim kinase is very similar to that of Ser/Thr kinase Akt. The commercially available anti–phospho-Ser/Thr Akt substrate antibody could preferentially recognize the conserved Akt phosphorylation motif ( 19, 20, 22). In addition, the predicted Pim phosphorylation sites around T157 and T198 residues in p27Kip1 have been reported to be phosphorylated by Akt ( 19, 20, 22). We thus estimated the Pim-mediated phosphorylation at T157 and T198 residues using the Akt substrate antibody. As shown in Fig. 2A (right), the Akt substrate antibody recognized the phosphorylated form of p27Kip1 only when HEK293T, K562, and 22Rv1 cells were cotransfected with a wt-Pim1S–expressing plasmid but not with a kinase-inactive Pim1S (KD-Pim1S), indicating that Pim1S kinase activity is required for p27Kip1 phosphorylation. As shown in Supplementary Fig. S1, recombinant GST-tagged Pim1S was incubated in vitro with recombinant p27Kip1 and the band of phosphorylated p27Kip1 could be detected. The phosphorylation was inhibited by adding the Pim1 inhibitor LY294002 ( 3). Pim1S is known to be phosphorylated by itself at Ser8 ( 33). Consistent with this report, Pim1S was phosphorylated by itself, and the phosphorylation was also inhibited by LY294002 (Supplementary Fig. S1). These results suggest that Pim1S directly phosphorylates p27Kip1.
Pim kinases were composed of Pim1S/1L, Pim2S/2M/2L, and Pim3 in human ( 3, 14, 37). As shown in Supplementary Fig. S2, all variants were able to phosphorylate p27Kip1. In addition, all recombinant Pim variants could phosphorylate p27Kip1 in vitro (data not shown). These results strongly suggest that p27Kip1 is a substrate of Pim kinases in vivo.
Identification of T157 and T198 as Pim phosphorylation sites in p27Kip1. p27Kip1 is known to be phosphorylated at serine-10 (S10) and threonine-187 (T187) residues, in addition to T157 and T198 ( 19). To identify the phosphorylation sites, we first tested whether Pim1S could phosphorylate NH2-terminal–deleted p27Kip1 mutants ΔN27 (amino acids 27–198) and ΔN52 (amino acids 52–198) and COOH-terminal–deleted p27Kip1 mutants 136STOP (amino acids 1–136), 156STOP (amino acids 1–156), and 185STOP (amino acids 1–185). Although Pim1S phosphorylated wild-type, ΔN27, ΔN52, and weakly, 185STOP, Pim1S could not phosphorylate 136STOP or 156STOP, lacking both T157 and T198 residues (Supplementary Fig. S3). We then prepared four point mutants (S10A, T157A, T187A, and T198A-p27Kip1) in which predicted phosphorylation sites were converted to alanine. As shown in Fig. 2C, the anti-Akt substrate antibody slightly recognized T157A-p27Kip1 or T198A-p27Kip1 even when cells were cotransfected with Pim1S. Point mutations at S10 or T187 did not affect the band intensities recognized by the anti-Akt substrate antibody. Immunoblot analysis using anti–phospho-S10 or anti–phospho-T187 p27Kip1 revealed that Pim1 was not associated with p27Kip1 phosphorylation at S10 or T187 residues ( Fig. 2C). These results suggest that both T157 and T198 residues are the Pim phosphorylation sites. Consistent with these results, mutation at T157 or T198 residues decreased Pim1S-mediated p27Kip1 phosphorylation in vitro ( Fig. 2D). These results strongly indicate that Pim kinases phosphorylate p27Kip1 at both T157 and T198 residues.
Pim kinases stimulate nuclear export of p27Kip1. It has been reported that Akt-mediated p27Kip1 phosphorylation at T157 and T198 residues induced the binding to 14-3-3 protein ( 19, 20, 22, 38) and promoted nuclear export of p27Kip1 ( 19– 24, 38). Because Pim kinases also promoted p27Kip1 phosphorylation at T157 and T198 residues ( Fig. 2), we analyzed the interaction between p27Kip1 and 14-3-3 after Pim1-mediated phosphorylation. When HEK293T cells were transfected with FLAG-tagged wt-p27Kip1, p27Kip1 binding to endogenous total 14-3-3 ( Fig. 3A, top ) or endogenous 14-3-3𝛉 ( Fig. 3A, second panel) was promoted only by HA-tagged wt-Pim1S expression. As the 14-3-3𝛉 mutant (RA), which lost its ligand-binding ability ( 19), failed to bind to p27Kip1 even in the presence of wt-Pim1S (Supplementary Fig. S4), the binding to 14-3-3𝛉 became specific. Consistent with the previous reports ( 19, 20, 22), the interaction between 14-3-3 and p27Kip1 was weakened when the T157 or T198 residue in p27Kip1 was converted to Ala (T157A or T198A, respectively). Moreover, mutation at both residues (T157A/T198A) could no longer bind to 14-3-3, even when the cells were cotransfected with wt-Pim1S ( Fig. 3A, lane 8). We also checked the p27Kip1 binding to other 14-3-3 isoforms (β, ε, σ, ζ, and γ) and found that the exogenously expressed 14-3-3 isoforms bound to p27Kip1-like endogenous 14-3-3 protein did (Supplementary Fig. S4). These results suggest that Pim kinases promote 14-3-3 binding to p27Kip1 by phosphorylating at both T157 and T198 residues.
Pim-mediated p27Kip1 phosphorylation promotes its binding to 14-3-3 and nuclear export. A, HEK293T cells were transfected with empty pHM6 (−, lanes 1 and 2) or pHM6-wt-Pim1S (+, lanes 3–8) together with empty pFLAG-CMV-2 (−, lane 1) or pFLAG-CMV-2 encoding wt-p27kip1 (wt, lanes 2 and 3) or the indicated p27Kip1 mutants. After transfection for 24 h, the cells were harvested, lysed, and immunoprecipitated with an anti-FLAG agarose, as described in Materials and Methods. The immunoprecipitants (IP) and the cell lysates (Input) were subjected to immunoblot analysis with the indicated antibodies. B, HT1080 cells were transfected with pHM6-wt-Pim1S together with pFLAG-CMV-2 encoding wt-p27Kip1 (wt), T157A-p27Kip1 (T157A), T198A-p27Kip1 (T198A), or T157A/T198A-p27Kip1 (T157A/T198A). After transfection for 24 h, HA-tagged Pim1S proteins were detected by staining with an anti-HA antibody following Oregon green–conjugated anti-rat antibody incubation. The FLAG-tagged p27Kip1 proteins were detected by staining with an anti-FLAG antibody, following treatment with Alexa Flour 568–conjugated anti-rabbit antibody. Nuclei were detected by staining with Hoechst 33342. The cells were washed and then visualized using a fluorescence microscope equipped with a CCD camera (top). The percentage of the cells that exhibited cytoplasmic p27Kip1 was determined by counting 200 transfectants that expressed both Pim1 and p27Kip1 protein in each cell (bottom). C, HEK293T cells were transfected with empty pc5FLAG alone (−) or pc5FLAG encoding PIm1S (+) together with pFLAG-CMV-2 encoding p27Kip1 (+). After transfection for 24 h, the cells were harvested. The cytoplasmic and nuclear fractions were separated, electrophoresed, and immunoblotted with the indicated antibodies. D, HEK293T cells were cotransfected with empty pHM6 alone (−) or pHM6 encoding wt-, T157A- or T198A-p27Kip1 together with pc5FLAG encoding none (−) or wt-Pim1S (+). After transfection for 36 h, the cells were stained with propidium iodide and analyzed with a flow cytometer.
Next, we investigated the subcellular localization of p27Kip1. When HT1080 cells were cotransfected with wt-Pim1S and wt-p27Kip1, we could observe the nuclear export of wt-p27Kip1 ( Fig. 3B, top) but not that of T157A, T198A, and T157A/T198A-p27Kip1. By counting, we found that the ratio of cells with cytoplasmic wt-p27Kip1 was ∼80% ( Fig. 3B, bottom). However, the percentage of cells that contained cytoplasmic T157A, T198A, and T157A/T198A-p27Kip1 was <20%. Immunoblot analysis confirmed that the decrease in nuclear p27Kip1 and the increased cytoplasmic p27Kip1 were also observed in Pim1S transfectants ( Fig. 3C), indicating that Pim-mediated p27Kip1 phosphorylation promoted 14-3-3 binding and nuclear export. To confirm the direct link between p27Kip1 phosphorylation and cell cycle progression, we transfected wt-Pim1S together with wt-p27Kip1 or phosphorylation-resistant p27Kip1 into HEK293T cells because their transfection efficiency was >90% (data not shown). Ectopic expression of wild-type or mutant p27Kip1 in the absence of wt-Pim1S induced cell cycle arrest at the G1 phase ( Fig. 3D; data not shown). Cotransfection of wt-Pim1S rescued cells from wt-p27Kip1–mediated, but not T157A- or T198A-p27Kip1–mediated cell cycle arrest at the G1 phase. The results clearly indicate that Pim kinases promote cell cycle progression by phosphorylating and inactivating p27Kip1 in cells.
Promotion of p27Kip1 degradation by Pim kinases. The cytosolic p27Kip1 was ubiquitinated by ubiquitin ligase KPC1/2 following degradation by proteasome at the G1-S phase ( 25). We first examined the change in p27Kip1 half-life. Overexpressed Pim1S stimulates the decrease in p27Kip1 levels (Supplementary Fig. S5). We next investigated the role of proteasome in Pim1S-mediated down-regulation of endogenous p27Kip1. The decrease in p27Kip1 expression was attenuated by treatment with proteasome inhibitor MG132 ( Fig. 4A ). However, MG132 could not fully rescue the Pim-mediated down-regulation of p27Kip1. These results suggest that Pim kinases down-regulate p27Kip1 level by promoting proteasome-dependent degradation and by other mechanisms.
Proteasome-dependent degradation and transcriptional repression are both associated with Pim-mediated decrease in p27Kip1 expression. A, HEK293T cells were transfected with empty pc5FLAG alone (−) or pc5FLAG encoding wt-Pim1S (+). After transfection for 24 h, medium was replaced with normal growth medium containing 10 μmol/L of MG132, and the cells were incubated for an additional 3 h. The whole cell lysates were electrophoresed and immunoblotted with the indicated antibodies. B, HEK293T cells were transfected with empty pc5FLAG alone (Mock) or pc5FLAG vector encoding wt-Pim1S (Pim1S). After transfection for 24 h, RNA was extracted from each sample. The amounts of p27Kip1 mRNA were quantified by real-time PCR, as described in Materials and Methods. C, K562 cells were transfected with empty pc5FLAG alone (−) or pc5FLAG encoding wt-Pim1S (+). After transfection for 6 h, 3 μmol/L of imatinib was added to some samples (+). After incubation for an additional 24 h, RNA was extracted from each sample. The amounts of p27Kip1 mRNA were quantified by real-time PCR, as described in Materials and Methods. Columns, mean of experiments done in triplicate; bars, SD.
The concentration of p27Kip1 is transcriptionally and posttranslationally regulated ( 25– 28). Therefore, we quantitatively evaluated the change in p27Kip1 mRNA level by real-time PCR. As shown in Fig. 4B, overexpression of Pim1S in HEK293T cells down-regulated the amount of p27Kip1 mRNA to 50% of controls. Consistent with the previous report ( 39), imatinib treatment down-regulated Pim1S level and induced p27Kip1 expression in K562 cells in a dose-dependent fashion (Supplementary Fig. S6). Under these conditions, imatinib increased the level of p27Kip1 mRNA ( Fig. 4C). The imatinib-mediated increase in p27Kip1 mRNA expression was displaced by ectopic expression of Pim1S ( Fig. 4C). These results suggest that Pim1S down-regulates p27Kip1 level by inducing proteasome-dependent degradation and by repressing its transcription.
Pim kinases repress p27Kip1 transcription by phosphorylating and inactivating FoxO transcription factors. It is well known that p27Kip1 transcription is regulated by FoxO transcription factors such as FoxO1a, FoxO3a, and FoxO4 ( 40, 41). In this family, FoxO3a is reported to contribute mainly to the transcription of the p27Kip1 gene ( 27, 40). Using an in silico search, we found that three residues, threonine-32 (T32), serine-253 (S253), and serine-315 (S315) partially matched the Pim consensus sequence in human FoxO3a ( Fig. 5A, top ). Phosphorylation of the residues was reported to reduce FoxO3a's transcriptional activity by inducing its nuclear export ( 40– 42). Because we had discovered that endogenous Pim1S interacted with endogenous FoxO3a in cells ( Fig. 2B, right, lanes 3 and 6), we examined whether Pim could phosphorylate FoxO3a in cells. When HEK293T cells were transfected with Pim1S and wt-FoxO3a, we observed the clear bands of phospho-FoxO3a at T32 and S253 residues ( Fig. 5A, bottom). The Pim-mediated phosphorylation was not observed in AAA-FoxO3a (T32A/S253A/S315A), even when Pim1S was overexpressed. Pim1S might regulate FoxO levels by phosphorylation because the expression level of wt-FoxO3a, but not AAA-FoxO3a, was diminished by Pim1S expression. Moreover, a decrease in endogenous Foxo3a expression and an increase in endogenous FoxO3a phosphorylation at T32 and S253 residues was observed in HEK293T cells ( Fig. 5B). We also confirmed that the Pim1S-mediated phosphorylation of another FoxO family member, FoxO1a at T24, S256, and S319, resulted in p27Kip1 transcription inhibition (data not shown). These results indicate that Pim1S phosphorylates and down-regulates FoxO transcription factors in cells.
Pim kinases down-regulate p27Kip1 transcription by phosphorylating and inactivating FoxO3a. A, alignment of the amino acid sequence of the Pim consensus sequence with the primary sequences of human FoxO3a around T32 and S253 residues. Identical residues are denoted by white letters on black background (top). HEK293T cells were transfected with empty pHM6 alone (−) or pHM6 encoding wt-Pim1S (+) together with empty pc5FLAG alone (−) or pc5FLAG encoding wt-FoxO3a (wt) or AAA-FoxO3a (AAA). After transfection for 24 h, the whole cell lysates were electrophoresed and immunoblotted with the indicated antibodies. Asterisks, the background bands (bottom). B, HEK293T cells were transfected with empty pc5FLAG alone (−) or pc5FLAG encoding wt-Pim1S (+). After transfection for 24 h, the whole cell lysates were electrophoresed and immunoblotted with the indicated antibodies. C, HEK293T cells were transfected with empty pHM6 alone (−) or pHM6 encoding wt-Pim1S (+) together with pc5FLAG encoding none (Mock), wt-FoxO3a (wt), AAA-FoxO3a (AAA), or H212R-FoxO3a (H212R). Cells were also transfected with pGL4.10 vector containing p27Kip1 promoter and pGL4.74 as an internal control. After transfection for 24 h, luciferase activities were calculated as described in Materials and Methods. Columns, mean of experiments done in triplicate; bars, SD. D, cancer profiling Array II containing normalized cDNA from normal tissue (N) or tumor tissue (T) of four patients with prostate cancer was probed with 32P-labeled pim1 and p27Kip1 probes, as described in Materials and Methods. Images were acquired and subsequent data analysis was done using Typhoon 9410. The densitometry units for the pim1 and p27Kip1 probed array were normalized to the densitometry units for a control ubiquitin-probed array.
To confirm the involvement of Pim in FoxO-mediated p27Kip1 transcription, we cloned a fragment of the human p27Kip1 promoter containing the putative FoxO-responsive elements ( 26) into pGL4.10 to generate the p27Kip1 reporter vector. In the absence of Pim1S, transfection of wt-FoxO3a and AAA-FoxO3a, but not mock or H212R-FoxO3a, increased p27Kip1 promoter activity ( Fig. 5C). The increase in luciferase activity by transfecting wt-FoxO3a was suppressed by wt-Pim1S expression. Ectopic expression of wt-Pim1S could not attenuate the AAA-FoxO3a–mediated increase in p27Kip1 reporter activity ( Fig. 5C). These results indicate that Pim phosphorylate and inactivate FoxO transcription factors resulting in the suppression of p27Kip1 transcription.
To confirm the negative correlation between pim1S and p27Kip1 mRNA expression in a pathologic condition, we examined their expression in some tumor tissue (T) compared with corresponding normal tissue (N) from patients with prostate cancer. It has already been suggested that Pim1 has a potentially strong role in these cancer progressions ( 8). In tumor tissue from patients with prostate cancer, pim1S expression levels were relatively higher than in individual normal tissue; however, p27Kip1 expressions were lower ( Fig. 5D). These results suggest that Pim kinases are associated with tumorigenesis by negatively regulating p27Kip1 expression.
Inhibition of endogenous Pim signaling increased the amount of p27Kip1 and attenuated cell proliferation. To assess the role of endogenous Pim signaling in p27Kip1 expression and cell growth, we generated a dominant-negative form of Pim1 (ΔN-Pim1; ref. 13). We then established polyclonal K562 cells that had been transfected with ΔN-Pim1 ( Fig. 6A, top ). The levels of p27Kip1 protein ( Fig. 6A, bottom, left) and p27Kip1 mRNA ( Fig. 6A, bottom, right) were both down-regulated in K562/wt-Pim1S cells although they increased in K562/ΔN-Pim1 cells. In addition, we confirmed that ΔN-Pim1 had no kinase activity (Supplementary Fig. S7, left). We observed the increase in p27Kip1 expression in nucleus (N) and the decrease in p27Kip1 expression in cytoplasm (C; Supplementary Fig. S7, right). Moreover, as shown in Fig. 6B, pim1 gene silencing by two independent pim1 siRNA resulted in the increase of p27Kip1 expression in 22Rv1 cells. These results confirmed that endogenous Pim signaling negatively regulated p27Kip1 expression in cells. Consistent with this increase, inhibition of Pim signaling by dominant-negative Pim1 (ΔN-Pim1) diminished the growth of K562 cells, whereas wt-Pim1S expression promoted cell proliferation ( Fig. 6C). In 22Rv1 cells depleted of Pim1 by siRNA, accumulation of p27Kip1 was also observed and the growth of the cells was diminished ( Fig. 6D). These results suggest that Pim kinases play an important role in cell cycle progression and cell proliferation by regulating p27Kip1 expression at transcriptional and posttranscriptional levels.
Induction of nuclear p27Kip1 expression and inhibition of cell growth by expressing a dominant-negative form of Pim1 or knockdown of Pim1 by siRNA. A, the polyclonal K562 cells that had been transfected with an empty pc5FLAG (K562/Mock), pc5FLAG encoding wt-Pim1S (K562/wt-Pim1S), or dominant-negative form of Pim1 (K562/ΔN-Pim1) were cultured in normal growth medium. The total cell lysates were electrophoresed and immunoblotted with the indicated antibodies (top). Expression level of p27Kip1 protein in A was quantified (bottom, left). p27Kip1 mRNA levels in K562/Mock, K562/wt-Pim1S, and K562/ΔN-Pim1 cells were quantified by real-time PCR method (bottom, right). Columns, mean of experiments done in triplicate; bars, SD. B, 22Rv1 cells were transfected with siRNAs targeting pim1. After transfection for 48 h, the whole cell lysates were electrophoresed and immunoblotted with the indicated antibodies. C, K562/Mock, K562/wt-Pim1S, and K562/ΔN-Pim1 cells were plated at a low density in triplicate wells. MTS assay was time-dependently performed, as described in Materials and Methods. Points, mean of three independent experiments; bars, SD. D, 22Rv1 cells which transfected with each siRNA were plated at a low density in triplicate wells. Cell number counting was time-dependently performed. Points, mean of three independent experiments; bars, SD.
Discussion
Pim kinases are overexpressed in patients with lymphoma, leukemia, and prostate cancers ( 4, 6– 8, 26). The ability of Pim to stimulate cell growth and inhibit apoptosis may contribute to the promotion of tumorigenesis ( 3, 5, 13, 31). Previous studies have reported that Pim1 stimulates cell cycle progression at the G1-S and G2-M transitions by phosphorylation of p21Waf1/Cip1 and CDC25A ( 16, 43) and CDC25C ( 17, 18). We confirmed that Pim1S overexpression promoted cell cycle progression at both the G1-S and G2-M transitions in HEK293T and K562 cells ( Fig. 1A; data not shown). Under these conditions, we found an increase in CDK2 activity and down-regulation of CDK inhibitor p27Kip1 expression ( Fig. 1B and C). Pim might accelerate cell cycle progression at the G1-S transition in part by down-regulating p27Kip1 expression levels.
There are three genes encoding Pim kinases, pim1, pim2, and pim3 ( 3). It is known that the consensus sequences of Pim1, Pim2, and Pim3 are very similar and that they phosphorylate the same substrate, such as Bad and 4EBP-1 ( 14, 33, 34, 44). The depletion of all pim1, pim2, and pim3 genes is required for the drastic phenotype of mice ( 15). In addition, up-regulation of pim2 is observed in pim1 knockout mice ( 45). Thus, Pim1, Pim2, and Pim3 could compensate for each function by sharing functional similarity. Consistent with these reports, we showed that all the Pim variants could phosphorylate p27Kip1 at T157 and T198 residues (Supplementary Fig. S2). Therefore, p27Kip1 might be an important substrate of Pim kinases.
We previously reported that T198 residue in p27Kip1 is phosphorylated by Akt and RSK1/2 ( 19, 20, 22). These kinases are regulated through different pathways. Akt is regulated by the phosphoinositide-3-kinase pathway and RSK1/2 through the Ras/Raf/MAPK pathway, whereas Pim is regulated by the JAK/STAT pathway ( 3). It is still unclear why three kinases phosphorylate the same residues in p27Kip1. To explain how these kinases coordinate their contributions to cell cycle progression at the G1-S transition by down-regulating p27Kip1, we checked the changes in the amounts of Akt and phospho-Akt (T308) after cell synchronization in K562 cells. The amounts were not changed throughout the cell cycle, although the amount of Pim1 was drastically increased at the G1-S transition (data not shown). Therefore, Pim1S mainly contributes to p27Kip1 down-regulation at the G1-S phase in K562 cells.
Phosphorylation of p27Kip1 has been thought to regulate its concentration and subcellular localization ( 19). CDK2-dependent phosphorylation at T187 has been well characterized and has been shown to be associated with proteasome-dependent degradation ( 28, 46). The S10 residue was identified as a major phosphorylation site of p27Kip1 and might regulate the nuclear-cytoplasmic shuttling. However, S10 phosphorylation was not sufficient for its nuclear export ( 47). Rodier and colleagues ( 48) also suggested that another signal might be necessary to direct p27Kip1 to the cytosol. Previous reports have shown that phosphorylation of p27Kip1 at T157 or T198 residues was associated with its binding to 14-3-3 proteins, which drives cytoplasmic localization ( 19, 20, 22, 38). Consistent with these reports, 14-3-3 protein binding to p27Kip1 did not require its phosphorylation at both T157 and T198 residues at the same time ( Fig. 3A; Supplementary Fig. S4). 14-3-3 proteins preferentially bound to p27Kip1 that was phosphorylated at the T198 residue rather than at the T157 residue. Regarding nuclear export of p27Kip1, phosphorylation at both T157 and T198 residues at the same time might be required for its nuclear export because single mutation at either T157 or T198 was enough to abrogate the nuclear export ( Fig. 3B). The 14-3-3 proteins were reported to form homodimers or heterodimers between different isotypes. Binding the dimer to both phosphorylated T157 and phosphorylated T198 might be required for promoting nuclear export. The nuclear export would be essential for the G1-S transition because Pim1S expression could not overcome the G1 cell cycle arrest induced by T157A-p27Kip1 or T198A-p27Kip1 ( Fig. 3D). Therefore, phosphorylation of p27Kip1 at T157 and T198 residues by Pim1 critically contribute to p27Kip1nuclear export and cell proliferation.
Overexpression of Pim1S decreased p27Kip1 expression ( Fig. 1C) and attenuated the half-life of p27Kip1 (Supplementary Fig. S5). It is well known that p27Kip1 is degraded in cytosol by ubiquitin ligase KPC1/2 at the G1-S transition ( 25). Thus, nuclear-cytoplasmic shuttling of p27Kip1 by Pim at the G1-S transition may promote its degradation by KPC1/2. Pim-mediated p27Kip1 down-regulation was mediated not only by proteasome-dependent degradation ( Fig. 4A) but also by transcriptional repression ( Fig. 4B and C). We identified FoxO transcription factors as new substrates of Pim ( Fig. 5A). Pim1S expression induced the phosphorylation of FoxO3a ( Fig. 5A and B) and inactivated its transcriptional activity ( Fig. 5C). We also confirmed that Pim1S phosphorylated FoxO1a and inactivated its transcriptional activity as well (data not shown). A previous report showed that phosphorylation at T32, S253, and S315 residues in FoxO3a induced 14-3-3 binding, nuclear export, and proteasome-mediated degradation ( 42). Consistent with that report, we found that Pim1S overexpression decreased the exogenous and endogenous FoxO3a expression level ( Fig. 5A and B).
In a physiologic condition, Pim expression was regulated by JAK/STAT signaling ( 3). In a pathologic condition, acute myelogenous leukemia (AML), FMS-like tyrosine kinase-3 (FLT3) activation by in-frame internal tandem duplications (ITD) was reported to induce the FoxO3a inactivation ( 49), leading to p27Kip1 down-regulation ( 36). ITD-FLT3 was shown to stimulate Pim1 expression by activating STAT5 ( 36). Therefore, Pim1 might down-regulate p27Kip1 expression by negatively regulating FoxO3a downstream of ITD-FLT3. In CML, BCR-ABL activates STAT3 and STAT5, leading to Pim1 expression ( 39). We detected the increase in p27Kip1 protein and mRNA expression in K562 treated with the BCR-ABL inhibitor imatinib. We also found that Pim1S transfection repressed p27Kip1 mRNA expression induced by imatinib ( Fig. 4C). These results suggested that Pim stimulated cell growth in AML and CML by down-regulating p27Kip1.
In patients with cancer, p27Kip1 is decreased, and its decrease is strongly correlated with a poor prognosis ( 30). We detected the inverse correlation between Pim1S and p27Kip1 mRNA in prostate cancer ( Fig. 5D). A negative correlation between Pim1S and p27Kip1 expression was also observed in various cell lines (Supplementary Fig. S8). In these tumors, Pim may contribute to p27Kip1 down-regulation and poor prognosis. Moreover, targeting Pim1 was shown to impair the survival of hematopoietic cells that express inhibitor-resistant forms of FLT3 and BCR-ABL ( 35). Thus, inhibiting Pim kinases may contribute to the cure of AML and CML in patients resistant to the inhibitors of FLT3 and BCR-ABL. Because pim knockout mice do not show serious defects, Pim kinases are druggable targets, and the small molecule targeting Pim kinases could be a new anticancer drug.
Disclosure of Potential Conflicts of Interest
None of the authors have conflicts of interest.
Acknowledgments
Grant support: Ministry of Education, Culture, Sports, Science, and Technology of Japan 17016012 and 18015008 (T. Tsuruo and N. Fujita), in part by the Kobayashi Institute for Innovative Cancer Chemotherapy (N. Fujita), and the Vehicle Racing Commemorative Foundation (N. Fujita).
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.
Footnotes
-
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received February 20, 2008.
- Revision received April 19, 2008.
- Accepted April 21, 2008.
- ©2008 American Association for Cancer Research.