p53-Dependent Induction of Prostate Cancer Cell Senescence by the PIM1 Protein Kinase

The PIM family of serine/threonine protein kinases has been implicated in the initiation or progression of multiple cancer types. The PIMs were initially cloned as proviral integration sites in Moloney murine leukemia virus–induced murine T-cell lymphomagenesis (1-3). In transgenic mouse models, the PIM family of protein kinases function as weak oncogenes stimulating T-cell lymphomas (1, 4) and can complement the activity of both c-MYC and AKT to enhance tumorigenesis (5-8). The induction of lymphomas in PIM-containing mice is markedly enhanced by treatment of these animals with chemical carcinogens (9) or γ-irradiation (10). The PIM protein kinases are overexpressed in many human cancers, including prostate cancer (11, 12), lymphoma (13, 14), leukemia (15), head and neck squamous cell carcinomas (16, 17), and pancreatic and colon cancers (18, 19). In human prostate, expression of PIM1 is low in benign prostatic hypertrophy, moderate in high grant intraepithelial neoplasia, and increased in frank cancer (11, 20). Both high Gleason grade and progression to aggressive metastatic prostate cancer has been associated with increased PIM levels (12, 21). Overexpression of PIM1 in human prostate cancer cells markedly increases their growth as tumors in animals (22).

Recent experiments have shown in normal fibroblasts that the overexpression of PIM1 can induce senescence rather than enhance growth (23). Oncogene-induced senescence (OIS) is well known and caused by multiple genes (24-27), including mutant RAS, RAF, and ERB-2 (for review, see refs. 28-30). Like other inducers of senescence, OIS is associated with a flattened cellular morphology; cell cycle arrest; stimulation of secretion of multiple cytokines, including interleukin (IL)-6 and IL-8; and activation of signaling networks driven by marked changes in the levels of specific transcription factors, for example, CEBP/β (31-33). Although the mechanism of OIS is complex, it seems to involve DNA replicative stress leading to the production of double-strand breaks and the recruitment of the serine-threonine kinase ataxia-telangiectasia mutated (ATM; refs. 34, 35). ATM enhances the activity of p53 directly and, through the phosphorylation of the CHK2 protein kinase, modifies the phosphorylation of the p53 protein (36, 37). The induction of OIS is associated with downstream activation of the p53 and pRb (38, 39) pathways with increases in inhibitors of cyclin-dependent protein kinases, including p16^INK4a and p21 (40-43), although OIS seems to proceed in p16 knockout mice (44). Activation of both c-MYC and RAS had been associated with the induction of a DNA damage response followed by OIS (45, 46). The induction of senescence as a result of the DNA damage response seems to depend on the extent of DNA damage, with minor changes being repaired (47). In precancerous lesions in animals and humans, OIS is a part of a tumorigenesis barrier imposed by DNA damage checkpoints (34). Because oncogenes are well known to induce cell death through the activation of the caspase cascade followed by apoptosis, there seems to be a critical balance between senescence and apoptosis, mediated in part by proteins that block cell death (i.e., BCL-2) and transcription factors (i.e., BRN-3a) that cooperate with p53 to induce growth inhibition rather than apoptosis (48).

Based on our observation that PIM1 overexpression in the human prostate cancer cell line 22Rv1, but not Du145, inhibits the growth of these cells, we have evaluated the ability of PIM1 overexpression to induce senescence and apoptosis in these human prostate cancer cells and studied the role of p53 and p21 in this process.

The following mouse monoclonal antibodies were used in these studies: anti-PIM1 and anti-human IL-6 (Santa Cruz Biotechnology), anti-human p21^CIP1/WAF1 (Transduction Laboratories, BD Biosciences), anti-cyclin B1 (Pharmigen, BD Biosciences), anti–γ-tubulin, and anti–glyceraldehyde-3-phosphate dehydrogenase (GAPDH)–peroxidase conjugated (Sigma). Additionally, rabbit monoclonal anti–Ki-67 and rabbit polyclonal anti-γH2AX antibodies were purchased from Abcam. Rabbit polyclonal antibodies used in these experiments include anti-p27 (Santa Cruz Biotechnology) and anti–dimethyl-histone H3 (Lys9; Millipore). Antibodies to p53 and phospho-CHK2 (Thr68) were purchased from Cell Signaling Technology, Inc. Fluorescein isothiocyanate–conjugated anti-mouse and Cy3-conjugated anti-rabbit antibodies were from Sigma. Doxycycline was purchased from Sigma and prepared as a stock solution at 2 mg/mL in PBS.

A p53-green fluorescent protein (p53-GFP) plasmid (plasmid 11770) was purchased from Addgene, and a retroviral vector encoding the dominant-negative variant of p53 (pBabe/p53DN) was kindly provided by Dr. Carola Neumann (Medical University of South Carolina, Charleston, SC). Lentiviral pTRIPz vectors encoding Tet-inducible control shRNA (RHS4743) and p21 shRNA (V2THS_202469) were purchased from Open Biosystems (Thermo Fisher Scientific, Inc.).

Retroviral vectors encoding PIM1 or a kinase-dead variant of PIM1 (KD PIM1) were designed as described previously (49). To generate inducible PIM1 expression constructs, the coding region for PIM1 gene was inserted into the AgeI-MluI sites of the pTRIPz/shRNA lentiviral vector (Open Biosystems). This procedure resulted to substitution of the RFP-shRNA cassette downstream of the Tet/ON promoter with the PIM1 cDNA. The pTRIPz vector with excision of RFP-shRNA cassette and without insertion of PIM1 was used as a vector control (pTRIPz). All of these constructs were verified by sequence analysis.

Human prostate 22Rv1, Du145, RWPE-1, RWPE-2, and Cos7 cells were obtained from the American Type Culture Collection. 22Rv1 and DU145 cells were grown in RPMI 1640, and Cos7 cells were grown in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific) supplemented with 10% fetal calf serum. RWPE-1 and RWPE-2 cells were maintained in keratinocyte medium supplemented with 5 ng/mL human recombinant epidermal growth factor and 0.05 mg/mL bovine pituitary extract (Invitrogen). Murine embryonic fibroblasts triple knockout for PIM1, PIM2, and PIM3 kinases (TKO MEF) were derived from 14.5-day-old embryos and were genotyped as described (50). The tissue culture medium was supplemented with 100 units/mL penicillin and 100 μg/mL streptomycin (Gibco).

The production of infectious retroviruses encoding PIM1, KD PIM1, dominant-negative p53 (DN p53), and vector control (pLNCX) was carried out as previously described (49). 293FT cells and the Trans-Lentiviral packaging system (Open Biosystems) were used to produce infectious lentiviruses according to the manufacturer's protocol.

To generate 22Rv1, Du145, RWPE-1, and RWPE-2 stable pools that constitutively express PIM1, cells were plated 16 to 18 hours before transduction at 1 × 10⁵ cells/60-cm plates and then infected with 5 × 10⁴ retroviral particles in the presence of 8 μg/mL Polybrene. After 6 hours of incubation, the virus-containing medium was replaced with fresh medium. On the next day, 400 μg/mL G418 was added to select the stably infected cell population. After 10 days of selection, stable cell pools were established, and the expression of PIM1 transgenes was verified by Western blot analysis.

For inducible PIM1 expression assays, 22Rv1, Cos7, or TKO MEF cells were infected with lentiviruses encoding vector control (pTRIPz) or PIM1 (pTRIPz/PIM1) using ViraDuctin Lentivirus Transduction Kit (Cell Biolabs, Inc.). Cells were selected with 5 μg/mL puromycin for 12 days. Individual clones were isolated, and cells were incubated in the absence or presence of doxycycline and subjected to Western blotting to analyze PIM1 expression.

Additional pools of 22Rv1 cells that overexpressed DN p53 were established through retroviral transduction. Stable cell lines derived from two individual clones expressing inducible PIM1 were infected with retroviruses encoding vector control (pLNCX) or DN p53 (pBabe/DN p53) using the protocol described above. G418 (400 μg/mL) or hygromycin (500 μg/mL) were added to select stably infected cell populations expressing PIM1 and vector control or DN p53. After 10 days of selection, stable cell pools were established, cells were incubated in the absence or presence of doxycycline, and expression of the transgenes was verified by Western blot analysis.

To generate cells expressing PIM1 and p53 wild-type proteins, 22Rv1 stable cell lines with inducible PIM1 expression or a vector control were transfected with p53-GFP plasmid using TransIT-Prostate transfection kit (Mirus) according to the manufacturer's protocol. Cells were selected with G418 (400 μg/mL) for 14 days and individual clones were isolated. Cells demonstrating GFP fluorescence were incubated with doxycycline (20 ng/mL) to induce PIM1 expression and analyzed by Western blotting for the existence of p53-GFP fusion and PIM1 proteins.

For p21 knockdown experiments, 22Rv1 stable pools with constitutive expression of PIM1 (pLNCX/PIM1) or vector only (pLNCX) in late passage were transduced with lentiviruses encoding Tet/ON inducible control shRNA or p21 shRNA as described. Doxycycline (1 μg/mL) was added 1 day after transduction to induce the expression of shRNAs. After selection with puromycin (4 μg/mL) for 10 days, cells were pooled and analyzed.

To carry out the MTT assay, cells were seeded in triplicates into 96-well plates (5 × 10³ per well) and allowed to adhere overnight. At the indicated times, 5 mg/mL MTT (Sigma) were added to the medium at 1:5 dilution and incubated for 3 hours; then, supernatants were removed and 100 μL DMSO per well were added to dissolve intracellular purple formazan. Absorbance was then measured at 570 nm. To measure cell growth with crystal violet (51), cells were plated in triplicates at 5 × 10⁴ per well in 12-well plates and incubated for 10 days, fixed with 4% paraformaldehyde in PBS, and stained with 0.1% crystal violet for 20 minutes (1 mL/well). Cells were then washed three times with water and allowed to dry. The dye was then extracted from cells using 10% acetic acid, and the absorbance of this solution was measured at 590 nm.

Total RNA was extracted with the Trizol reagent (Invitrogen) and purified with acid phenol extraction. Single-stranded cDNA was constructed by Superscript III polymerase (Invitrogen) and oligo(dT) primers. Real-time PCR was performed using iCycler (Bio-Rad) and SYBR Green PCR master mix reagents (Bio-Rad). The following primers were used: p21 forward, 5′-TGGAGACTCTCAGGGTCGAAA-3′; p21 reverse, 5′-CGGCGTTTGGAGTGGTAGAA-3′; TP53INP forward, 5′-CTCATTGAACATCCCAGCATG-3′; TP53INP reverse, 5′-ATTTCATTTGCTTCCACTCTG-3′; DDIT4 forward, 5′-GCTCAGGATTTCGACTTGTTAAG-3′; DDIT4 reverse, 5′-AAACTACAAATGACTTTAGCTGACTAG-3′; IL-6 forward, 5′-AAAGAGGCACTGGCAGAAAA-3′; IL-6 reverse, 5′-TTTCACCAGGCAAGTCTCCT-3′; IL-8 forward, 5′-CTGCGCCAACACAGAAATTA-3′; IL-8 reverse, 5′-ACTTCTCCACAACCCTCTGC-3′; GAPDH forward, 5′-CAGCCTCAAGATCATCAGCA-3′; GAPDH reverse, 5′-GTCTTCTGGGTGGCAGTGAT-3′.

Cells were seeded in six-well tissue culture plates (1 × 10⁵ per well), cultured for 2 days, washed with PBS, fixed, and stained using the Senescence β-Galactosidase Staining Kit (Cell Signaling). Stained cells were visualized using a Nikon Eclipse TE 2000-S microscope under brightfield, and images were captured with Nikon Digital Camera DXM 1200F using the Q Capture software.

Floating and adherent cells were harvested and combined, washed with PBS, and fixed with cold 70% ethanol and stored at −20°C. The cells were then washed with PBS and resuspended in 1 mL buffer containing 100 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1 mmol/L CaCl₂, 0.5 mmol/L MgCl₂, 0.1% NP40, 40 μg/mL RNase H, and 25 μg/mL propidium iodide. After incubation for a minimum of 30 minutes at room temperature, the cells were then analyzed with a FACScalibur flow cytometer using channel FL3.

To generate tumor xenografts, 22Rv1 early-passage cells expressing the indicated transgenes were washed twice with PBS, and the cell density was adjusted to 0.5 × 10⁶ cells/50 μL in serum-free Dulbecco's modified Eagle's medium. An equal volume of Matrigel (BD Biosciences) was then added, and the cell suspension was injected into both the left and right dorsal flank of male Nu/nu nude mice (Charles River). On day 2 after cell implantation, doxycycline (1 μg/mL) was added into the drinking water supplemented with 5% glucose, and this solution was changed every 2nd day. Four mice carrying eight tumors were evaluated in each group. Tumor size was determined by caliper measurements, and tumor volume was calculated using the equation (L × W²)/2.

Cultured 22Rv1, Cos7, or MEF cells were plated on glass coverslips into six-well plates with a density of 2 × 10⁵ per well. The following day, cells were washed two times with PBS; fixed with 4% paraformaldehyde for 15 minutes; premeabilized with 0.1% Triton X-100 solution; and stained with IL-6, IL-8, histone H3 (Lys9), or Ki67 antibodies. For detection of centrosomes, the cells were fixed in 10% methanol. Fixed cells were incubated in blocking solution and PBS with 2% bovine serum albumin containing 2% goat serum, and immunostained with primary antibodies overnight at 4°C, followed by treatment with fluorescein isothiocyanate– or Cy3-conjugated secondary antibodies for 1 hour.

To stain tumor xenografts, these tissues were first embedded in Optimum Cutting Temperature (OCT; VWR) and frozen cryostat sections (8–10 μm) were prepared. Tissues were fixed for 7 minutes with freshly prepared 4% paraformaldehyde in PBS, permeabilized with 0.1% Triton X-100 for 30 minutes, and blocked for 30 minutes in PBS/0.1% Triton X-100/5% heat-inactivated goat serum. Sections were immunostained with anti-Ki67 antibody (1:1,000 dilution) for 1 hour, and then incubated with Cy3-conjugated secondary antibody (1:1,000 dilution) for an additional 1 hour. Fluorescence was visualized with Nikon Eclipse E800 microscope, and images were taken by Nikon digital sight DS 2 Mv camera using the NIS element BR 2.30 software.

All experiments were repeated a minimum of two times, and statistical analysis was carried out using Student's t test unless otherwise stated.

To investigate the growth-regulatory activity of PIM1 in prostate cancer cells, we infected the human prostate cancer cell line 22Rv1 with a retrovirus expressing this cDNA, selected with G418, and collected pools of transfected cells. At increasing passage number, these cells were frozen, stored, and evaluated for their ability to grow in tissue culture (Fig. 1A). When compared with early-passage (numbers 7–10) cells, late-passage (numbers 15–17) PIM1-containing cells seemed to slow their growth. By the final passage (numbers 20–25), PIM1-expressing cells barely divided (Fig. 1B). The final-passage cells were found to have a flat and spread morphology and to be β-galactosidase (β-Gal) positive, consistent with the suggestion that they had undergone senescence (Fig. 1B). To further document the induction of the senescent phenotype by PIM1, we compared these cells to 22Rv1 cells treated with the known senescence inducer (Supplementary Fig. S1), hydrogen peroxide (H₂O₂). Increases in the protein inhibitors of cyclin-dependent kinases (31, 52, 53) are associated with the induction of senescence. Western blots show that there was no increase in the expression of the cyclin-dependent kinase inhibitor p27^Kip1 and only a moderate elevation of p16^INK4 in PIM1-containing cells compared with vector control and H₂O₂-treated cells (Fig. 1C). The level of cyclin-dependent kinase inhibitor p21^CIP was markedly increased in both H₂O₂-treated and PIM1-expressing senescent cells, suggesting the possibility that p21 was potentially an important factor in the development of PIM1-induced senescence. The increases in p21 seemed to be transcriptionally driven as the level of mRNA for this protein was also markedly elevated (Fig. 1D). The induction of senescence by H₂O₂ also increased the amount of PIM1 protein in these tumor cells.

Cellular senescence has previously been shown to be associated with activation of a network cytokine secretion, including IL-6 and IL-8, and many proinflammatory proteins (31, 54, 55). We find that the levels of both IL-6 and IL-8 mRNA and protein are elevated in late-passage PIM1-containing cells (Fig. 2A and B and data not shown), consistent with the induction of senescence. Senescent cells accumulate stress-associated heterochromatin foci (SAHF). These heterochromatin foci are localized in the nucleus and are produced upon cellular exposure to varied stresses (31, 56). To examine whether PIM1-induced senescence correlates with SAHF formation, we performed fluorescence staining for the SAHF marker trimethyl-histone H3 (Lys9m). Our results show that late-passage cells with forced expression of PIM1 accumulated SAHF compared with vector control cells (Fig. 2C). These results show that overexpression of PIM1 kinase is capable of inducing senescence in 22Rv1 cells. In comparison, expression of PIM1 in three additional human prostate cancer cell lines, RWPE-1, RWPE-2, and Du145, neither inhibited cell growth nor induced the senescent phenotype (Supplementary Fig. S2 and data not shown). These cell lines differ from 22Rv1 cells in that they have inactive p53, whereas 22Rv1 expresses wild-type p53 protein, suggesting the possibility that p53 plays a role in regulating cellular senescence induced by PIM1 in 22Rv1 cells.

To evaluate the ability of PIM1 protein to regulate these processes, we created 22Rv1 cell pools that contained a doxycycline-inducible PIM1 protein. These cells show that the level of PIM1 protein induced by varied doses of doxycycline correlated with the level of induction of the p21 protein (Fig. 3A). We find that the application of 20 ng/mL doxycycline to PIM1-containing cells induced a progressive time-dependent increase in both the number of cells with 4N and 8N DNA (polyploid) content as measured by DNA histogram analysis (Fig. 3B). However, final-passage cells treated with 20 ng/mL doxycycline also showed a marked increase to ∼25% in cells demonstrating <2N DNA content, consistent with the induction of apoptosis (Fig. 3B; Table 1). Treatment with higher doses of doxycycline (2 μg/mL) caused high levels of PIM1 expression and early apoptosis, and death at low passage numbers (data not shown). Additionally, to determine whether the 22Rv1 cells that constitutively expressed PIM1 studied in Figs. 1 and 2 were undergoing apoptosis, we measured the number of cells with <2N DNA content. PIM1-containing cells, but not those expressing a kinase-dead PIM1, had elevated levels of <2N cells, increased p21 expression, and showed cleavage of both poly(ADP-ribose) polymerase and β-catenin, two markers of apoptosis induction (Supplementary Fig. S4). Thus, 22Rv1 prostate cancer cells expressing elevated levels of PIM1 developed both a senescent and apoptotic phenotype.

Because 22Rv1 cells grow in clusters, making analysis of chromosome structure difficult, to document the induction of polyploidy and the potential changes in centrosome number by PIM1 overexpression, we used Cos7 cells whose morphology can be easily examined. We infected these cells with lentiviruses encoding vector only (TRIPZ) or PIM1 under the control of the doxycycline-inducible promoter. Then, we selected stable cell pools overexpressing PIM1 (Fig. 3D). Using antibodies to γ-tubulin, which stains centrosomes, passage 8 Cos7-overexpressing PIM1 cells were examined for changes in centrosome number. By counting multiple cell fields, the fraction of Cos7 cells containing more than two centrosomes in PIM1-expressing cells was found to be greater than 20% compared with 5% for control cells expressing vector alone (Fig. 3E). In this culture, cells with multiple centrosomes (row 1, arrows), tetraploid cells with four mitotic spindles (row 2, arrows), and polyploid cells with multiple centrosomes and mitotic spindles (row 3, arrows) are all visible. Although it is more difficult to identify centrosomes in fibroblasts, we repeated this experiment in embryonic fibroblasts that were isolated from mice that were genetically engineered to knock out all of the three PIM isoforms (57) and then transfected to overexpress only PIM1. γ-Tubulin staining of these cells showed the induction of multiple centrosomes (Supplementary Fig. S5).

To evaluate whether the p53 pathway was activated in these PIM1-containing prostate cancer cells, we carried out quantitative reverse transcriptase-PCR (qRT-PCR) analysis of mRNA from cells expressing PIM1 at early and late passages (Fig. 1A), and examined the levels of three p53-activated genes, p53-inducible nuclear protein 1 (TP53INP1; ref. 58), DNA-damage inducible transcript 4 (DDIT4, also known as REDD1; ref. 59), and p21. We found that PIM1-expressing cells at late, but not at early, passage showed marked increases in the p53-inducible genes, including TP53INP1 and DDIT4. The increase in p21 transcripts is statistically significant even at early passages but with much more dramatic effects in late-passage cells (Fig. 4A). p53 is often induced as a result of DNA damage. To document that DNA damage is occurring in PIM1-overexpressing cells, we treated the inducible cells with varying amounts of doxycycline (Fig. 4B) and then measured the levels of phosphorylated CHK2 and γH2AX, both of which are modified by double-strand DNA breaks. CHK2 is phosphorylated on Thr68 by the ATM/ATR protein kinases that are activated by double-strand DNA damage. Phosphorylation of CHK2 at Thr68 is required for subsequent activation of p53 protein in response to DNA damage (37, 60). Likewise, histone H2AX is phosphorylated on Ser139 when double-strand DNA damage occurs, yielding the γH2AX form of this histone (61, 62). A marked change in the level of phosphorylation of these proteins occurs in late-passage PIM1-containing 22Rv1 cells and mirrors increases in the levels of the p21 protein (Fig. 4B). Thus, in 22Rv1 prostate cancer cells, long-term PIM1 overexpression seems to induce polyploidy and DNA damage that is associated with the activation of p53-inducible genes, and ultimately leads to both cellular senescence and apoptosis.

To evaluate whether the activation of the p53 pathway is essential for the induction of cellular senescence by PIM1, we overexpressed DN p53 in the 22Rv1 cells that contained doxycycline-inducible PIM1 and treated these cells with doxycycline (20 ng/mL). We found that in early-passage cells, DN p53 is capable of decreasing the level of the p53-inducible TP53INP-1 transcript and suppressing the induction of p21 RNA (Fig. 5A). These results are consistent with reduction of p21 protein seen by Western blotting in PIM1/DN cells (Fig. 5B). Increased levels of PIM1 kinase are known to elevate the cellular levels of cyclin B1, which can contribute to deregulation of cell division and genomic instability induced by PIM1 (63). The expression of cyclin B was gradually increased in early-passage PIM1-containing cells, but no marked changes in cyclin B1 levels was seen in PIM1/DN p53–coexpressing cells (Fig. 5B). To examine the ability of DN p53 to suppress PIM1-induced polyploidy, 22Rv1-expressing inducible PIM1 with or without DN p53 were grown in 20 ng/mL doxycycline. Early-passage cells were analyzed by DNA histogram analysis to determine the fraction expressing 8N DNA content. DN p53 suppressed the induction of the 8N cell number (Fig. 5C). To determine the effect of DN p53 on apoptosis (<2N DNA content), the same cells were grown in the presence of 20 ng/mL doxycycline and late-passage cells were studied. As shown in Fig. 5C, DN p53 was able to suppress the ability of PIM1 to induce both polyploidy and apoptosis in these prostate tumor cells. Western blots carried out on extracts of these cells showed that DN p53 suppressed both the induction of phosphorylation of the CHK2 protein kinase and γH2AX (Fig. 5D). As shown in Fig. 5D, the cyclin B1 level is markedly decreased in cells expressing high levels of PIM, and this change is reversed by the overexpression of DN p53. Finally, staining of these late-passage PIM1–overexpressing cells with β-Gal shows that blocking p53 activity also prevents the induction of this marker of cell senescence by PIM1 (Fig. 5E).

To examine the ability of DN p53 to regulate 22Rv1 tumor growth driven by PIM1, these cells were injected into the flanks of nude mice and the animals were treated with doxycycline in water. On day 27, tumor size was measured and the animals were sacrificed. In comparison to vector, PIM1-containing 22Rv1 tumors grew poorly, and this decreased growth was completely reversed by the overexpression of DN p53. In addition, the tumors coexpressing PIM1 and DN p53 showed a slight statistical increase in growth compared with vector control. Western blots carried out on tumor extracts showed a lower level of p21 protein and a decreased phosphorylation of CHK2 and γH2AX in DN p53–expressing tumors when compared with those expressing PIM1 (Fig. 6). Consistent with these results and the slow growth of the tumors, staining of PIM1-containing tumors showed low levels of cell division, as measured by the proliferation marker Ki67, and stained positively for senescence-associated β-galactosidase (SA-β-Gal). Both of these changes were reversed by DN p53 overexpression. Cells isolated from these tumors and placed back into culture behaved similarly to the tumors (Supplementary Fig. S5) with predictable changes in protein levels (Supplementary Fig. S5A), Ki-67 (Supplementary Fig. S5B), and SA-β-Gal (Supplementary Fig. S5C).

The observation that DN p53 blocks PIM1-induced senescence suggested the possibility that activation or overexpression of p53 would enhance the induction of these cellular changes. To further define the role of p53 in PIM1-induced senescence, 22Rv1 cells containing inducible PIM1 were transfected with p53 and doubly expressing clones were selected with G418. Comparison of the growth of 22Rv1 cells expressing p53, PIM1, or both proteins showed that PIM1 expression alone did not inhibit the growth of cells on tissue culture–coated plates, whereas p53 had a modest inhibitory effect, and the simultaneous overexpression of both proteins causes significant growth inhibition (Fig. 7A). This result is clearly seen when p53 and p53/PIM1 22Rv1 cells are allowed to grow for 10 days in tissue culture, and then fixed and stained with crystal violet (Fig. 7B). This inhibition in cell growth could result from the effect of PIM1 and p53 on elevating the levels of p21 (Fig. 7C). Also, when PIM1 and p53 were expressed simultaneously, β-Gal staining of these cell lines revealed significantly increased levels of cell senescence (Fig. 7D).

The Cdk inhibitor p21 was first identified as an overexpressed marker in senescent cells and later found to be capable of regulating senescence in both normal and cancer cells. p21 mediates, in part, the effects of p53, although other genes are also involved (for review, see ref. 64) in the p53-induced biological changes. To examine the role of p21 in these cells, lentiviral vectors were used to transduce a Tet/ON-inducible shRNA directed at p21 (sh-p21) or a control sequence (sh-c) into 22Rv1 cells constitutively expressing PIM1 or a control vector. This shRNA was successful in decreasing p21 levels in control and PIM1-containing cells (Fig. 8B), but had no effect on the growth of 22Rv1 cells (Fig. 8A). In contrast, the growth of PIM1-containing 22Rv1 cells was markedly stimulated (Fig. 8A) by decreasing the p21 levels. In these cells, decreased p21 levels correlated with a markedly lower level of γ-H2AX (Fig. 8B), suggesting that cells containing lower p21 levels showed a decreased level of DNA damage. In addition, lower levels of p21 protein in PIM1-containing cells correlated with a marked decrease in SA-β-Gal staining and an inhibition in the induction of the senescence phenotype.

The current study shows that PIM1 kinase can induce cellular senescence in prostate carcinoma cells 22Rv1 based on (a) irreversible inhibition of cell proliferation accompanied by decreasing Ki-67 staining, (b) formation of β-Gal–positive cells both in culture and in xenograft models, (c) elevation in the mRNA and protein levels of IL-6 and IL-8, and (d) SAHP-positive staining. PIM1-induced senescence has been previously reported in primary normal MEFs (23). The induction of senescence by PIM1 seems to depend on an intact p53-p21 pathway. PIM1 induced senescence in 22Rv1 cells that are p53 wild type, whereas prostate cell lines, including PC3, Du145, RWPE-1, and RWPE-2, which contain a mutant, deleted, or inactivated p53 gene grow with high levels of PIM1 and avoid senescence. Additional evidence pointing to the importance of p53 in PIM-induced senescence includes the observation that DN p53 protein blocks the PIM1-induced senescence in culture, and that in mouse xenograft models (Figs. 5 and 6; Supplementary Fig. S5), the knockdown of p21 mRNA markedly enhances cell growth and decreases the number of β-Gal–positive cells (Fig. 8). Our observation that ectopically expressed p53 induces cell growth arrest and increases the number of β-Gal–positive cells in 22Rv1/PIM1 cells compared with vector control (Fig. 7) also suggests that the activation of p53 pathway is capable of enhancing the PIM1-induced senescence response. Because PIM1 does not upregulate p27 (Fig. 1C), a protein critical to the induction of senescence by AKT1 in prostate cells (52), it is likely that the mechanism by which these protein kinases function to induce senescence is markedly different.

The control of PIM1 protein expression levels in tumors is complex and is regulated at the transcriptional (65, 66) and translational levels by specific microRNAs (67). Pim expression is controlled posttranslationally by modifications such as ubiquitination and protein dephosphorylation (68, 69). Quantitative expression of PIM1 measured by qRT-PCR reveals 2.75-fold and 4.45-fold increase of PIM1 mRNA in benign prostatic hyperplasia and in prostate cancer, respectively, when compared with PIM1 mRNA levels in normal prostate (21, 70). These data suggest that PIM1 overexpression is likely an early event in prostate carcinogenesis. We find that PIM1 expression is also capable of inducing apoptosis in a p53-dependent manner, and the time course of induction of cell death seems related to the level of PIM1 in these tumor cells. In a breast cancer cell model driven by oncogenes, apoptosis and senescence have been shown to occur simultaneously in one culture (71). The reason why a cell chooses to undergo apoptosis versus senescence is complex and not fully understood. The extent and persistence of DNA damage may play a major role in regulating both of these phenotypes (35). Alternatively, overexpression of antiapoptotic molecules, such as BCL-2 or transcription factors CREB and SLUG (28, 72-74), may enhance the induction of senescence. Factors that modulate growth arrest versus senescence include cofactors for p53-mediated transcriptional activation, including the BRN-3a protein that induces senescence over apoptosis. The DAXX protein can function to inhibit p21 without regulating p53 to influence the induction of proapoptotic genes (75-77).

Consistent with its activation of p53, our results indicate that PIM1 induces 3-fold increase in p21 mRNA level in early-passage cells and a marked 12- to 16-fold increase in the levels of the p21 transcript when PIM1-expressing prostate cancer cells undergo senescence (Figs. 1D and 4A). Previous data showed that PIM1 phosphorylates p21 protein on Thr145 and on Ser146, thus promoting p21 protein stabilization in p53-null H1299 lung carcinoma cells (78). Because the expression of DN p53 decreased the level of p21 protein in this model, we suggest that the elevation of p21 protein is largely driven by increased transcription of the p21 mRNA and not by changes in protein stability (Fig. 5). It cannot be ruled out that in early-passage cells where major DNA damage has not taken place, small changes in p21 levels could be regulated by protein stability. We find that in early-passage 22Rv1 cells that express relatively large amounts of PIM1, there was no highly significant change in p53 protein levels and no induction of the p53 target genes TP53INP-1 and DDIT-4 (Fig. 4A). It is possible that small changes in p53 levels activated by PIM1 stabilization of the Mdm2-ARF complex (23) could be driving increases in transcription of p21, or, as our earlier work suggested (79), that other signaling pathways (i.e., PKC) possibly regulated by PIM1 could be playing a role in modulating p21 mRNA levels.

The DNA damage signaling pathway connects cellular stresses to cellular senescence (80). We analyzed the DNA damage response in 22Rv1 PIM1-expressing cells by measuring phosphorylation of CHK2 and the minor histone H2AX (Fig. 4C). In response to DNA damage, H2AX becomes phosphorylated on Ser139 and CHK2 on Thr68 by ATM (37, 60, 62). Early-passage cells do not induce changes in these two proteins; however, in late-passage cells, both CHK2 phosphorylation and γH2AX expression are markedly increased (Fig. 4B). Because CHK2 can phosphorylate and activate p53, these changes explain in part the temporal relationship between PIM1-dependent DNA damage and the activation of the p53 pathway leading to transcription of downstream genes (e.g., DDIT4; ref. 59). OIS has been linked to DNA replicative stress, possibly through the induction of reactive oxygen species (81, 82) marked by aberrant premature termination of replication fork progression, leading DNA breakage, DNA damage response, and induction of senescence (34, 83). Genomic instability and DNA damage have been observed in both early human precancerous lesions and tumors, and this instability is closely related with induction of DNA damage response (84). Recent data also showed that PIM1 kinase plays a central role in DNA damage–evoked neuronal death by regulating aberrant neuronal cell cycle activation (85). Our results (Fig. 3; Supplementary Fig. S4), and data from others (86, 87), identify defects in mitotic spindle checkpoint, centrosome amplification, and chromosome misaggregation, resulting in the appearance of aneuploidy and polyploidy in PIM1-overexpressing cells. The mechanism by which PIM1 induces these changes is unclear. Increases in cyclin B1 expression in the early passage of 22Rv1/PIM1 cells could possibly contribute and have been identified in other cell lines (Fig. 5; ref. 63). During mitosis, PIM1 has been shown to be located in the spindle poles in complex with NuMA, HP1β, dinein, and dynactin, suggesting that the PIM1 kinase might regulate the mitotic apparatus (88).

A model describing the findings of the present study can be summarized as follows: PIM1 induces chromosomal and genomic instability followed by DNA damage and p53 activation, which in turn enhances p21 expression; these cellular changes then lead to senescence. The 22Rv1 cell line originated from primary human carcinoma that shows significant genetic drift caused by microsatellite instability compared with other human prostate carcinoma cell lines (89). The additional genomic instability induced by PIM1 expression enhances DNA damage response and p53 activation, leading to senescence in 22Rv1 cancer cells. Thus, for PIM to enhance tumor growth and progression, two cellular changes are critical (a) baseline genomic instability and (b) inactivation of the p53 pathway, allowing these tumor cells to bypass the induction of cellular senescence and apoptosis.

Small-molecule inhibitors of the PIM protein kinases have been described (for review, see ref. 90) and are targeted for the treatment of prostate cancer, leukemia, and lymphoma. Our studies suggest that tumors that develop elevated PIM kinase levels will most likely have a defective p53 pathway. Thus, the ability to induce cellular apoptosis by targeting PIM1 alone may be compromised by the absence of a robust p53 response.

No potential conflicts of interest were disclosed.

We thank Margaret Romano [Pathology and Laboratory Medicine at the Medical University of South Carolina (MUSC)] for preparation of frozen sections from tumor xenografts, Richard M. Peppler (Hollings Cancer Center at MUSC) for flow cytometry analysis, and Pamela A. Knox (Hollings Cancer Center) for technical assistance in the preparation of the manuscript. Dr. Carola Neumann (MUSC) kindly provided the DN p53 plasmid, and Dzmitry Fedorovich (MUSC) helped in the creation of the 22Rv1 cells coexpressing PIM1 and wild-type p53 proteins. We thank Dr. Anton Berns (The Netherlands Cancer Institute) for the PIM knockout mice.

Grant Support: Department of Defense grant W8IXWH-08. The flow cytometry shared resource was funded by 2P30-CA138313.

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.