Overexpression of the Oncogenic Kinase Pim-1 Leads to Genomic Instability

Introduction

Faithful segregation of the genome during cell division in human cells is dependent on the normal function of the mitotic spindle. Defects in the organization, constitution, or regulation of the mitotic spindle apparatus are thought to be important causes of chromosome missegregation and aneuploidy in human cancer (1) . The molecular mechanisms underlying the development of aneuploidy in the majority of human cancers are not fully defined, although mutations in mitotic checkpoint genes such as hBUB1 and CHFR have been identified in a subset of human cancers and cell lines (2 , 3) . The development of aneuploidy has also been associated with mutations/overexpression of several tumor suppressor genes and oncogenes, including BRCA1/2, p53, pRb, Myc, Ras, Mos, and AURORA-A (4, 5, 6) .

Pim-1 is an oncogenic serine-threonine kinase of ill-defined function that has been implicated in lymphomagenesis (7) . Pim-1 is widely expressed in tissues, with the highest expression found in hematopoietic tissues and testes (7) , and can be induced by a number of cytokines in hematopoietic cells where it enhances cellular survival (8 , 9) . Although several substrates of Pim-1 kinase have been identified, the precise roles of Pim-1 in normal cellular physiology or tumorigenesis remain obscure. Recently, microarray expression profiling identified PIM-1 overexpression in a significant proportion of human prostate tumors (10) . To explore the roles of Pim-1 in prostate carcinogenesis, we established nontumorigenic and tumorigenic prostate epithelial cell lines overexpressing Pim-1. Our analyses indicate that overexpression of Pim-1 interferes with the mitotic spindle checkpoint, leading to polyploidy and chromosome missegregation.

Materials and Methods

Cell Lines and Retroviral Constructs.

RWPE-1 and LNCaP cell lines were obtained from American Type Culture Collection and maintained as recommended. MSCV-Pim-1 retrovirus vector was made by inserting FLAG-tagged M_r 33,000 murine Pim-1 cDNA into the MSCVneoEB. Virions were produced in ecotropic Phoenix cells and used to infect cell lines as described (9) . After selection in 200–500 μg/ml G418, the resistant clones were pooled.

Western Blot Analysis.

Total cell lysates were resolved by SDS-PAGE and processed for Western blot analysis. Antibodies used are FLAG M2 (Sigma), Pim-1 (Santa Cruz Biotechnology), and actin (Santa Cruz Biotechnology).

Flow Cytometric Analysis.

Cells were fixed with ethanol and stained with propidium iodide for DNA content analysis. For BrdUrd 5 analysis, cells were pulsed with 10 μm BrdUrd (Sigma) for 30 min before harvesting, then stained with anti-BrdUrd-FITC-conjugated antibody (Becton Dickinson). For MPM2 and phosphohistone H3 staining, the cells were incubated with either MPM2 or antiphophohistone H3 antibodies (Upstate) for 30 min, washed in PBS, and then stained with antimouse or antirabbit FITC-conjugated antibodies (Sigma).

Immunofluorescence.

Cells on cover slides were incubated with the following primary antibodies overnight at 4°C: anti-α-tubulin and anti-γ-tubulin (Sigma). Secondary antibodies were Alexa Fluor 594 antimouse- or antirabbit-FITC conjugated IgG (Molecular Probes). After counterstaining with DAPI (Sigma), images were captured with a Zeiss Axioskop 40 microscope.

Quantitative RT-PCR Analysis.

RNA isolation and quantitative RT-PCR (TaqMan; Applied Biosystems) using SYBR-GREEN dye were performed as described (11) . PCR reactions were performed in triplicate. Primers used are: Pim-1 Forward, 5′-CGAGTGCCCATGGAAGTGGT-3′; Pim-1 Reverse, 5′-CGGGCCTCTCGAACCAGT-3′; 18S Forward, 5′-CGCCGCTAGAGGTGAAATTCT-3′; and 18S Reverse, 5′-CGAACCTCCGACTTTCGTTCT-3′.

Human Prostate Tumor Specimens.

Radical prostatectomy specimens were obtained from the University of Alabama Tissue Procurement Center. Glandular tissue was grossly dissected from samples with no evidence of carcinoma (benign) or samples with 20–90% carcinoma (malignant) as determined by examination of H&E-stained frozen sections. Total RNA was prepared from the tissue using TRIzol (Life Technologies, Inc.). Due to the infiltrative nature of prostate cancer, the determined PIM-1 levels in these samples should be considered a minimum estimate.

Time-Lapse Imaging and Analysis.

Cells were plated on bottom glass dishes (Warner Instrument). The cells were stained with Hoechst 33258 (Sigma) and maintained at 37°C. Differential interference contrast (DIC) and fluorescence images were captured every 15 min using a ×40 objective on a Zeiss Axiovert 200 M microscope.

Cell Division Tracking.

Cells were labeled with the dye PHK67 (Sigma) following the manufacturer’s instruction. After labeling, cells were replated and harvested every 2 days, and PKH intensity was measured by flow cytometry.

Results

Abnormal Cell Cycle and Polyploidy in Pim-1 Overexpressing Cells.

To evaluate the role of PIM-1 overexpression in prostate carcinogenesis, we established prostate epithelial cell lines that stably overexpress Pim-1. The immortalized, nontumorigenic prostate epithelial cell line RWPE-1 (12) and the prostate carcinoma cell line LNCaP were transduced with control retrovirus or retrovirus expressing a FLAG-tagged murine Pim-1 gene. After selection in G418, pools of stable Pim-1 expressing clones were obtained for further study. Western blot analysis for the FLAG epitope and the Pim-1 protein confirmed protein overexpression in the cell lines (Fig. 1A) ⇓ . To assess the relative overexpression of Pim-1 in transduced cells, we used quantitative RT-PCR using primers that recognize both the human and murine Pim-1 mRNAs. Pim-1 expression levels in RWPE-Pim-1 and LNCaP-Pim-1 cells were 10.5- and 14.5-fold higher, respectively, than those in control cells (Fig. 1B) ⇓ . We also determined levels of PIM-1 expression in benign and malignant human prostate specimens. We identified a tumor with PIM-1 expression that is ∼30-fold higher than that in normal prostate (Fig. 1B) ⇓ , indicating that the level of overexpression in our stable cell lines is within the range seen in human tumors. Initial characterization of the Pim-1-overexpressing cell lines indicate that elevated Pim-1 expression did not significantly alter the growth rates of the cell lines as determined by cell counting, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, or colony forming assay (data not shown).

Recent studies indicating that Pim-1 localizes to the spindle poles during mitosis and interacts with the nuclear mitotic apparatus protein, NuMA (13) , prompted us to examine whether Pim-1 plays a role in regulating the cell cycle. Flow cytometric analysis of propidium iodide-stained, asynchronously growing cells showed a significant accumulation of cells with 4N DNA content in Pim-1-overexpressing cells derived from both the nontumorigenic RWPE-1 and the tumorigenic LNCaP cell lines (Fig. 1, C and D) ⇓ . Furthermore, Pim-1 overexpression led to polyploidy (DNA content >4N). Notably, polyploidy in Pim-1-overexpressing cells occurred without induction of microtubule stress, but can be additionally increased by treating the cells with the antimicrotubule agent Taxol (Fig. 1, C–E) ⇓ . The polyploidy observed in Pim-1-overexpressing cells suggests that cells with a 4N DNA content are able to undergo an additional round of DNA replication without prior cell division, resulting in cells with 8N DNA content. To demonstrate this directly, we performed BrdUrd incorporation experiments. In RWPE-Neo cells, BrdUrd incorporation is present mostly in the S phase fraction between 2N and 4N (Fig. 1F) ⇓ . By contrast, in RWPE-Pim-1 cells the most active BrdUrd incorporation was seen in the population of cells with DNA content between 4N and 8N (Fig. 1F) ⇓ , confirming that the 8N cells arose by rereplication of tetraploid cells. Thus, elevated expression of Pim-1 leads to polyploidy by subverting the checkpoints that prevent DNA rereplication without prior cell division.

Mitotic Checkpoint Defect in Pim-1-Overexpressing Cells.

The integrity of the mitotic spindle checkpoint in Pim-1-overexpressing cells was assessed by monitoring the mitotic index in cells treated with Taxol for various time points. The mitotic index was determined by staining cells with MPM-2 antibody, which recognizes mitosis-specific phosphoepitopes (14) . Control RWPE-Neo cells showed a robust mitotic arrest, with 40% of the cells arrested by the 18 h time point, compared with only 8% of RWPE-Pim-1 cells (Fig. 2A) ⇓ . We also monitored phosphorylation of histone H3 as an additional mitotic marker. Histone H3 phosphorylation begins on entry into mitosis, and appears to be important for chromosome condensation and segregation (15) . After 24-h Taxol treatment, 52% of RWPE-Neo cells were phospohistone H3 positive, compared with 18% of RWPE-Pim-1 cells (Fig. 2B) ⇓ . A similar mitotic checkpoint defect was observed in LNCaP cells, where 30% of LNCaP-Pim-1 cells were phosphohistone H3 positive after 18 h Taxol treatment compared with 9% of LNCaP-Neo cells (data not shown). In sum, these results indicate that Pim-1-overexpressing cells have a mitotic checkpoint defect.

Spindle Defects and Chromosome Missegregation in Pim-1 Overexpressing Cells.

Loss of the mitotic spindle checkpoint may lead to chromosomal instability and polyploidy by affecting chromosome segregation. Therefore, we examined the microtubule network, mitotic spindle, and centrosomes in Pim-1-overexpressing cells. Cells were stained with α-tubulin antibody to highlight the microtubules. Control RWPE-Neo and LNCaP-Neo cells presented a flattened morphology, with dendritic extensions and a fine network of cytoplasmic microtubules (Fig. 2, C and E) ⇓ . By contrast, the Pim-1-overexpressing cells (particularly RWPE-Pim-1 cells) were more rounded-up with uneven microtubule staining (Fig. 2, D and F) ⇓ . An increase in the number of multinucleated cells was also apparent in both RWPE-Pim-1 and LNCaP-Pim-1 cells (Fig. 2G) ⇓ consistent with our flow cytometry data showing polyploidy in Pim-1-overexpressing cells (Fig. 1) ⇓ . Another notable feature of Pim-1-overexpressing cells is an increase in the number of micronuclei (Fig. 2H) ⇓ . Micronuclei arise by chromosome missegregation during mitosis and appear in the cytoplasm as DNA-containing spheres surrounded by a nuclear membrane. The extent of micronucleus formation is an indication of the frequency of cells in a population that is losing chromosomes and is a marker of aneuploidy (16) .

We next examined the status of the mitotic spindles and centrosomes in Pim-1-overexpressing cells. Cells were costained with antibodies to α-tubulin (to mark spindles) and γ-tubulin (to mark centrosomes). Examination of mitotic cells revealed spindle abnormalities in a majority of Pim-1-overexpressing cells. Whereas most RWPE-Neo cells in mitosis establish strong bipolar spindles with normal chromosomal congression (Fig. 3A) ⇓ , a significant fraction of RWPE-Pim-1 cells in mitosis showed a range of abnormalities (Fig. 3, B–D) ⇓ . These abnormalities include disorganized, weakly stained, or multipolar spindles, and misaligned or missegregated chromosomes. These abnormalities were seen in 53% of RWPE-Pim-1 cells compared with 21% of RWPE-Neo cells (>200 mitoses counted from three independent experiments).

An increase in the number of cells with multiple (>2) centrosomes was also apparent in Pim-1-overexpressing cells compared with controls (15.7% versus 5.8% of >300 cells counted from three experiments). Centrosome amplification may result from defective cell division and multinucleation, leading to amplification of centrosome number during subsequent cell cycles (5 , 17) . Alternatively, direct dysregulation of the centrosome duplication cycle can lead to centrosome amplification (5 , 17) . We favor the former possibility in Pim-1-overexpressing cells, because these cells have a checkpoint defect that leads to polyploidization and multinucleation.

Delayed Cytokinesis in Pim-1-Overexpressing Cells.

The existence of a cytokinesis checkpoint was described recently in yeast (18) . In a cell with a misaligned spindle in which the cell fails to correct the defect, polyploidy ensues because of a failure to complete cytokinesis (18) . The results presented thus far indicating abnormal spindle assembly, loss of checkpoint control, polyploidy, and multinucleation in Pim-1-overexpressing cells all suggest that these cells have a cytokinesis delay. To observe this directly, we used time-lapse video microscopy. We followed cells by both differential interference contrast (DIC) and fluorescence microscopy after staining the DNA with Hoechst 33258 dye. Single cells about to enter mitosis were photographed every 15 min (Fig. 4) ⇓ . Under the experimental conditions we used, a control cell completed cell division in 180 min (Fig. 4) ⇓ . The Pim-1-overexpressing cell, on the other hand, is still in anaphase at this time, and took an additional 180 min to complete division (Fig. 4) ⇓ . Pim-1-overexpressing cells also had less compacted chromosomes and lagging chromosomes (Fig. 4 ⇓ , arrows).

To demonstrate cytokinesis delay at a cell population level, we followed the cell division rate using the fluorescent cell linker PKH67 dye. The decrease in fluorescent intensity as the dye is dispersed between the daughter cells during cell division has been used for cell division tracking (6) . It took 1.5 days to achieve 50% decrease of median fluorescent dye intensity in RWPE-Neo cells, whereas this took 3 days in RWPE-1-Pim-1 cells (n = 2). Together these data indicate that Pim-1-overexpressing cells have defective cytokinesis.

Discussion

Chromosome alignment and segregation during mitosis are critically dependent on the normal formation of a bipolar mitotic spindle. In the present work, we have identified a novel role for Pim-1-overexpression in overriding the mitotic spindle checkpoint, resulting in genomic instability. Pim-1 was shown recently to interact with and phosphorylate the nuclear mitotic apparatus protein, NuMA (13) . It was proposed that Pim-1 is important for maintaining a stable complex linking the chromosomal kinetochores to the spindle microtubules consisting of NuMA, dynein/dynactin, and HP1β (13) . Our results suggest that overexpression of Pim-1 could affect spindle dynamics by disrupting the normal function of this complex.

Due to loss of checkpoint control, Pim-1-overexpressing cells with spindle abnormalities are not arrested in mitosis, but they fail to efficiently complete cytokinesis. This results in polyploidy and multinucleation. Therefore, Pim-1 overexpression represents a novel mechanism that can be used by tumors to override the mitotic spindle checkpoint. At present, the mechanism of Pim-1 subversion of the mitotic checkpoint is not clear; however, we have observed that Pim-1-induced polyploidy is passage dependent, implying the need for accumulation of additional changes. 6

Expression studies of PIM-1 in human prostate cancer present a paradox in that although PIM-1 expression is increased in a majority of prostate carcinomas compared with benign prostate tissue, poor clinical outcome is correlated with decreased expression of PIM-1 (10) . Our results linking PIM-1 expression to genomic instability may help explain this paradox. Early in the process of tumorigenesis, genomic instability is thought to promote the acquisition of protumorigenic mutations, which are selected for during the course of tumor progression. However, when a tumor has acquired an abnormal genome that confers growth advantage, it is advantageous for these genetic changes to be “fixed.” This has been referred to as genetic convergence (19) . Thus, PIM-1 overexpression may be important for driving genomic instability in early tumors, whereas advanced tumors down-regulate PIM-1 to stabilize any acquired abnormalities.

In summary, our results identify a novel role for Pim-1 overexpression in inducing genomic instability through effects on the mitotic spindle checkpoint. These findings generate several hypotheses regarding the value of Pim-1 as a biomarker and as a therapeutic target in prostate tumors.