Polo-like kinase 1 (Plk1) overexpression enhances ionizing radiation-induced cancer formation in mice

Polo-like kinase 1 (Plk1), a serine/threonine protein kinase normally expressed in mitosis, is frequently up-regulated in multiple types of human tumors regardless of the cell cycle stage. However, the causal relationship between Plk1 up-regulation and tumorigenesis is incompletely investigated. To this end, using a conditional expression system, here we generated Plk1 transgenic mouse lines to examine the role of Plk1 in tumorigenesis. Plk1 overexpression in mouse embryonic fibroblasts prepared from the transgenic mice led to aberrant mitosis followed by aneuploidy and apoptosis. Surprisingly, Plk1 overexpression had no apparent phenotypes in the mice. Given that no malignant tumor formation was observed even after a long period of Plk1 overexpression, we reasoned that additional factors are required for tumorigenesis in Plk1-overexpressing mice. Because Plk1 can directly participate in the regulation of the DNA damage response (DDR) pathway, we challenged Plk1-overexpressing mice with ionizing radiation (IR) and found that Plk1-overexpressing mice are much more sensitive to IR than their wild-type littermates. Analysis of tumor development in the Plk1-overexpressing mice indicated a marked decrease in the time required for tumor emergence after IR. At the molecular level, Plk1 overexpression led to reduced phosphorylation of the serine/threonine kinases ATM and Chk2 and of histone H2AX after IR treatment both in vivo and in vitro. Furthermore, RNA-Seq analysis suggested that Plk1 elevation decreases the expression of several DDR genes. We conclude that Plk1 overexpression may contribute to tumor formation by both inducing chromosomal instability and suppressing the DDR pathway.

Chromosome instability and aneuploidy are hallmarks of human cancers (1). Most human cancers contain aneuploid cells, and a significant number of studies have pointed to failure in various critical mitotic events as a cause of aneuploidy in tumors (2)(3)(4). Defects in mitotic events, including centrosome maturation, microtubule-kinetochore attachment, chromosome alignment, and completion of cytokinesis, are the causes of aberrant cell division, eventually resulting in aneuploidy in daughter cells (5,6). The regulation of proper mitotic progression that is important to genomic integrity is predominantly controlled by protein phosphorylation driven by several evolutionarily conserved serine/threonine kinases, known as mitotic kinases. The most prominent mitotic kinases include Polo-like kinase 1 (Plk1), cyclin-dependent kinase 1 (Cdk1), aurora family, and NIMA (never in mitosis A) family kinases (7).
The Polo gene was first cloned from Drosophila melanogaster, with the observation that mutations in Polo induced abnormal spindle poles during mitosis (8). Five mammalian homologues for Polo, named Plk1 to Plk15, were soon identified (9). Among them, the best characterized member of human Plk family is Plk1. It has been well-documented that Plk1 is involved in almost every step of mitosis (10). Thus, it is probably not surprising that Plk1 is overexpressed in many cancer types, including melanoma, breast, non-small cell lung, colorectal, prostate, pancreatic, ovarian, and head and neck cancers, as well as non-Hodgkin's lymphomas and acute myeloid leukemia (9,11). More importantly, recent studies have also linked Plk1 with other cancer-associated pathways (12). For example, cross-talk between Plk1 and the p53 tumor suppressor has been described, because Plk1 negatively regulates both p53 protein stability and nuclear localization (13,14). In another study, Plk1 elevation was shown to cause PTEN inactivation, resulting in a tumor-promoting metabolic state in prostate cancer cells (15). In line with this observation, Plk1-associated kinase activity was demonstrated to contribute to the low-dose arsenic-mediated metabolic shift from oxidative phosphorylation to glycolysis via activation of the PI3K/AKT/mTOR pathway (16). In addition, elevation of Plk1 leads to acquisition of resistance to various therapies, including radiotherapy (17), taxol (18), metformin (19), gemcitabine (20,21), and androgen signaling inhibitors (22). All these studies strongly suggest that Plk1 likely plays a critical role during carcinogenesis.
Inhibition of cell proliferation and induction of apoptosis are two basic principles of anticancer therapy. Antimitotic therapy is a standard of care for many cancer types. Overexpression of Plk1, observed in a broad spectrum of cancers, has often been correlated with disease stage, histologic grade, poor prognosis, metastatic potential, and survival (9). The key role of Plk1 in oncogenic events gave impetus to the development of potent and specific small molecule Plk1 inhibitors (10). Consequently, several Plk1 inhibitors have been developed, with some agents showing encouraging results in various cancer cell lines and xenograft models of human cancer (23,24). Preclinical studies have demonstrated a particular sensitivity to Plk1 inhibition in human cancer cells harboring specific genetic abnormalities, including mutations in p53, Ras, and PTEN (10). In contrast to marked antitumor activity in leukemia, overall antitumor activity of Plk1 inhibitors in patients with solid tumor has been modest in trials (23). Thus, there is a critical need for understanding the physiological functions of Plk1 in vivo. Lack of Plk1 transgenic mice is one major hurdle to achieve that.
Herein, we generated a transgenic mouse model to investigate the involvement of Plk1 in tumorigenesis. Using a Cre-loxP system, we achieved the conditional expression of Plk1. Elevated Plk1 expression resulted in mitotic failure, leading to apoptosis and aneuploidy in mouse embryonic fibroblasts (MEFs). 2 In addition, Plk1 overexpression caused a defective DNA damage response (DDR) pathway. Our findings indicated that this mouse model is a useful system to study the physiological roles of Plk1 in cancer etiology and therapeutics.

Strategy and design for the conditional Plk1 transgenic mouse
Considering that overexpression of Plk1 induces abnormalities in cell division in vitro, we selected a conditional transgenic system to prevent the possible embryonic lethality of founder animals (Fig. 1A). The configuration of this transgenic cassette was chosen so that Plk1 would be overexpressed in the transgenic mice after expression of Cre recombinase. To prevent the chromosomal positional effects associated with the random nature of transgenic integration, we inserted the cassette into the well-characterized euchromatic locus Rosa26 (Fig. 1A). The standard transfection procedures using this exchange vector were followed by the recombinase-mediated integration, which resulted in more than 90% positive embryonic stem (ES) cell clones ( Fig. 1, B-H). Plk1-KI mice were subsequently obtained from the recombinant ES cells.

Establishment and characterization of transgenic mouse founders for Plk1 expression
We successfully obtained four conditional transgenic founders by pronuclear injection to mouse embryos. Female conditional Plk1-KI mice were mated with male CMV-Cre transgenic mice that have a wide tissue distribution for expression of Cre recombinase (25). Expression of Plk1 in CMV-Cre/Plk1-KI double transgenic mice was detected in several organs with quantitative RT-PCR (qRT-PCR) ( Fig. 2A), immunoblotting (IB) analysis (Fig. 2B), and immunohistochemistry (IHC) staining (Fig. 2C). Increase of Plk1 expression was also confirmed in CMV-Cre/Plk1-KI MEFs (Fig. 2D). To further validate the Plk1 overexpression system, Plk1-KI mice were crossed with Myod-Cre mice, which express Cre in a muscle-specific manner. As indicated, Plk1 expression was detected only in muscle in Myod-Cre/Plk1-KI mice (Fig. 2, E and F). These data indicate that the Plk1 conditional transgenic system works properly in vivo.

Elevated Plk1 expression leads to mitotic abnormalities and apoptosis in late-passage mouse embryonic fibroblasts
Considering that Plk1 has multiple functions during the cell cycle, we next examined the effects of Plk1 overexpression on the proliferation and cell-cycle progression of MEFs. Overexpression of Plk1 was confirmed in MEFs isolated from two different CMV-Cre/Plk1-KI mouse lines (supplemental Fig. S1A). Surprisingly, massive Plk1 overexpression had no apparent impact on the proliferative activity of early-passage MEF lines compared with control MEFs (supplemental Fig. S1B). Furthermore, we assessed the potential correlation between Plk1 overexpression and cell cycle distribution. Despite much higher levels of the Plk1 expression in CMV-Cre/Plk1-KI lines, FACS analyses showed no apparent increase in G 2 /M populations compared with control MEF cells (supplemental Fig. S1iC). Given that MEFs undergo abnormal mitosis and senescence at late passages, we tested whether Plk1 overexpression affects this process. Interestingly, induction of Plk1 overexpression resulted in significantly increased abnormal mitotic figures in late-passage MEFs (Fig. 3A). Staining by propidium iodide showed various mitotic defects, including chromosome misalignment, chromosome missegregation, and chromosome bridges, observed at higher rates in late-passage Plk1-overexpressing MEFs (Fig. 3A). To further assess the chromosome instability induced by Plk1 overexpression, we analyzed chromosome spreads generated from serial passages of MEFs according to the 3T3 protocol (Fig. 3B). The total number of chromosomes in each cell was determined, and cells were scored as diploid (2n ϭ 40) or aneuploid. Primary control MEFs, under normal culture conditions, spontaneously become aneuploidy. However, we found that Plk1-overexpressing MEFs were significantly more aneuploidy than control MEFs at later passages (Fig. 3B). Examination of the absolute number of chromosomes per cell revealed a highly significant increase in the frequency of di-and multinucleated cells compared with control MEFs.
The proliferation of control and Plk1-overexpressing MEFs at later passages was investigated using MTT assay. Control MEF cells was observed to proliferate significantly faster than the Plk1-KI cells at passage 5 (Fig. 3C). To determine whether the growth arrest in the Plk1-KI MEFs was associated with cell cycle defects, DNA content was analyzed. We found that Plk1-KI cells at passage 5 caused a significant increase in the cell number of cells in G 2 /M phase, concomitant with a significant decrease in the percentage of cells in G 0 /G 1 phase (Fig. 3D). Furthermore, MEFs with overexpression of Plk1 exhibited a substantial increase in the sub-G 1 population, suggesting that a large number of cells at G 2 /M phase underwent cell death (Fig.  3D). Using senescence-associated ␤-galactosidase assay, we found that Plk1-overexpressing MEFs displayed an increased senescence phenotype (Fig. 3E). 2

Plk1 contributes to cancer formation Phenotypic analysis of Plk1-KI mice
The CMV-Cre/Plk1-KI double transgenic mice displayed no apparent developmental abnormalities during mouse development. They grew and gained weight normally during the first 9 months of their life. However, by 18 months of age CMV-Cre/ Plk1-KI mice exhibited slightly reduced weight compared with WT littermates (Fig. 4A). Litter sizes were similar for CMV-Cre/Plk1-KI and WT mice, ranging from 7 to 9 pups per litter. The pups were born healthy and developed normally at a normal growth rate, maturing at the age of 3-4 weeks. Further, we did not observe any obvious behavioral difference between CMV-Cre/Plk1-KI and WT animals. However, a minor reduced survival rate in Plk1-KI mice was observed, when the mice were monitored up to 25 months of age (Fig. 4B). The complete morphological analyses did not reveal significant histological differences between the organs of CMV-Cre/Plk1-KI mice and WT animals at the ages of both 1 and 11 months (Fig.  4, C and D). To determine the consequences of Plk1 overexpression on whole-body energy balance, we performed indirect calorimetry analyses. On a chow diet, CMV-Cre/Plk1-KI mice showed a slightly higher energy expenditure than control mice, as indicated by both slightly increased O 2 consumption (Fig.  4E) and slightly increased CO 2 production (Fig. 4F). Considering that Plk1 is critical for energy metabolism in cultured cells (16), we examined the glucose uptake in a glucose tolerance test (GTT). Glucose uptake from CMV-Cre/Plk1-KI mice was slightly increased in comparison to WT littermates (Fig. 4G). Given that Plk1 is essential for many mitotic events, we exam- A, relative Plk1 mRNA levels in the tissues of CMV-Cre/Plk1-KI animals were compared with the levels in Plk1-KI but without Cre expression (Control, Ctrl) animals (mean Ϯ S.D., n ϭ 3 for each tissue). B, total cellular protein was prepared from the indicated tissues of adult CMV-Cre/Plk1-KI or control mice and subjected to IB. C, animals were sacrificed at the age of 1 month, followed by anti-Plk1 IHC staining of indicated tissues. Sections representing the organs of control (Plk1-KI only) and CMV-Cre/Plk1-KI mice. Bar, 200 m. D, top panel, whole cell extracts from MEFs isolated from CMV-Cre/Plk1-KI and control animals were subjected to IB. Bottom panel, PCR of mouse genomic DNA (tail clips) are presented. E, relative Plk1 mRNA levels in the tissues of Myod-Cre/Plk1-KI animals were compared with the levels in control animals (mean Ϯ S.D., n ϭ 3 for each tissue). F, total cellular protein was prepared from the indicated tissues of Myod-Cre/Plk1-KI and control mice. **, p Ͻ 0.05; ***, p Ͻ 0.01.

Plk1 contributes to cancer formation
ined the proliferative activity in CMV-Cre/Plk1-KI mice by staining Ki67 of various tissues. The proliferation index was moderately increased only in the liver (see Fig. 6A).

Plk1-KI mice and MEFs are hypersensitive to DNA damage
Because no malignant tumor formation was detected in the Plk1-overexpressing mice after a long latency, we then challenged Plk1-overexpressing mice with carcinogens to see whether Plk1 overexpression affects carcinogenesis under stimulation condition. Mounting evidence suggests that Plk1 directly participates in the regulation of DNA double-strand break repair (26,27). We therefore explored whether Plk1 overexpression affects cellular response to DNA damage in vivo.
Cohorts of CMV-Cre/Plk1-KI and WT mice were irradiated with 10 Gy of ionizing radiation (IR) and closely monitored for radiation toxicity. Although ϳ60% of WT littermates (n ϭ 14) survived beyond 50 days, over 80% of CMV-Cre/Plk1-KI ani-mals (n ϭ 15) died within 50 days (Fig. 5A), indicating that CMV-Cre/Plk1-KI mice were more sensitive to IR) than WT littermates. To confirm this observation, we plated MEFs (WT or CMV-Cre/Plk1-KI) at clonal density and exposed to increasing doses of IR, followed by colony formation assay. In agreement with the in vivo results, the viability of CMV-Cre/Plk1-KI MEFs after IR was significantly impaired compared with WT MEFs, indicating that Plk1-overexpressing MEFs are more sensitive to DNA damage induced by IR (Fig. 5B). In a separate experiment, CMV-Cre/Plk1-KI and WT mice at the age of 10 weeks were exposed to a single dose of 5 Gy whole body ␥-ray irradiation and monitored for additional 4 months. CMV-Cre/ Plk1-KI, but not WT, mice showed severe dermatitis, scratching bloody areas in head, neck, and ears (Fig. 5C). Although none of WT mice had tumors after 4 months, 10.3% of CMV-Cre/Plk1-KI developed life-threatening spleen lymphomas, a difference that was statistically significant (p ϭ 0.0386; Fig. 5D,

Plk1 contributes to cancer formation
bottom panel). Upon IR, CMV-Cre/Plk1-KI mice showed apparent splenic enlargement and disruption of normal splenic architecture, forming diffused lymphoma (Fig. 5D, top panel). Increased radiation sensitivity is one of the hallmarks of defective DDR. The results described above suggested that Plk1 elevation leads to inhibition of DDR. To directly test this hypothesis, we analyzed DDR of mice. Spleens were prepared from three different mice that were exposed to IR and subjected to IB to follow the DDR pathway. IR led to activation of DDR, indicated by elevated phosphorylation of ATM, Chk2, and H2AX in WT, but not in CMV-Cre/Plk1-KI mice (Fig. 5E). Finally, IR also led to activation of the ATM/Chk2 pathway in WT but not in CMV-Cre/Plk1-KI MEFs (Fig. 5F).

Overexpression of Plk1 leads to increased proliferation and hypersensitivity to DNA damage in liver
Because liver is the only tissue that showed increased proliferation upon Plk1 overexpression in the absence of IR (Fig. 6A), we further analyzed liver-associated phenotypes after IR. A 4-fold increase in liver tumors in CMV-Cre/Plk1-KI versus normal littermates was observed (p ϭ 0.0472; Fig. 5D, bottom  panel). Upon IR, the livers of CMV-Cre/Plk1-KI mice were apparently enlarged and carrying small tumors compared with WT controls (Fig. 6B). IR of WT mice led to multifocal lymphoid hyperplasia, seen in a perivascular location (Fig. 6C, top  panel). For IR-treated CMV-Cre/Plk1-KI mice, we observed multiple lymphocytic overgrowth in the parenchyma and mitotic figures, suggesting the formation of mild lymphoma (Fig. 6C, middle panel) or severe lymphoma (Fig. 6C, bottom  panel). To determine whether lymphoma in IR-treated CMV-Cre/Plk1-KI mice is of T-cell or B-cell origin, we stained a T cell marker with CD3. Approximately 80 -90% of the neoplastic round cells demonstrated moderate to strong positive staining with CD3 antibody, indicative of T lymphocyte origin (Fig. 6D). Further, Ki67 IHC staining was performed to the same samples. Ki67 signal was virtually undetectable in WT samples but significantly increased in lymphoma areas of CMV-Cre/Plk1-KI mice (Fig. 6E). Finally, we also observed apparent diffuse and sever fatty change in the parenchyma of liver of one CMV-Cre/ Plk1-KI mouse (Fig. 6F).
To understand how Plk1 overexpression accelerates IR-induced cancer formation, we performed an RNA-Seq analysis of MEFs under different conditions (Ϯ Plk1-KI Ϯ IR irradiation). Expression levels of 648 genes were affected upon Plk1 overex-

Plk1 contributes to cancer formation
pression, whereas most of the affected genes were down-regulated. In the presence of IR irradiation, Plk1 overexpression led to modulation of 227 genes (Fig. 7, A and B). This observation is expected because it has been documented that Plk1 is a target of DNA damage. Plk1 KI affects the expression of 97 genes in both the presence and the absence of IR irradiation. Detailed description and analysis of RNA-Seq data will be reported separately. Herein, we only focus on genes related to DNA damage repair pathways. As indicated in Fig. 7C, Plk1 overexpression clearly affects the expression of a series of DNA damage repair genes. In addition, RT-PCR was performed to confirm this observation (Fig. 7D). Thus, deregulation of DNA damage repair pathways upon Plk1 overexpression is the likely reason to explain the enhanced IR-induced cancer formation.

Discussion
Mitotic kinases are critical regulators of cell cycle progression, and the timing of expression and activities of those kinases are tightly regulated during the cell cycle. Plk1 is associated with mitotic spindle poles and centromere/kinetochore regions and the expression and activity of Plk1 peak during late G 2 to M phase (9, 10). Although the majority of studies highlights the

Plk1 contributes to cancer formation
essential role of Plk1 for mitosis, overexpression of Plk1 is often observed in interphase of many cancer cell lines and tumor samples. In addition to these pathological observations, evidence that Plk1 overexpression in NIH3T3 cells induces tumor formation in nude mice suggests that dysregulation of Plk1 expression is a potential cause of sporadic malignant tumors (28). These observations have prompted research into the potential therapeutic application of Plk1 inhibition in cancer. Mitotic catastrophe and tumor growth inhibition have been observed in preclinical studies using Plk1 inhibitors or siRNAs (12,29). However, with all the evidence pointing to Plk1 as a worthwhile cancer target, it should be noted that the molecular evidence of Plk1 being a bona fide oncogene is still lacking. Accordingly, we have established a Plk1 transgenic mouse line to achieve overexpression of Plk1 with a new conditional trans-genic system. In this system, the expression of Plk1 is regulated by the recombination with the transgene mediated with Cre protein and its target sequences.
Previous cellular studies indicate that sustained overexpression of Plk1 increases the oncogenic potential of cultured cells (28). Herein, we have taken advantage of a conditional expression model to modulate the expression of Plk1. Our data based on MEFs confirmed the effect of Plk1 overexpression on the generation of misaligned chromosome and aneuploidy. We have also observed that Plk1 overexpression induces apoptosis in cultured MEFs at later passages. Several studies have indicated that expression of activated oncogenes, such as Aurora-A or Myc, generally induces apoptotic cell death in MEFs. Further, apoptotic cell death induced by Aurora-A or Myc has been demonstrated to be p53-dependent (30 -32). Because studies

Plk1 contributes to cancer formation
have also revealed a connection between Plk1 and p53, it is possible that p53 might be involved in apoptosis of MEFs with deregulated Plk1. It was shown that Aurora-A overexpression in mammary epithelial cells results in increased apoptosis, with a low frequency of tumor formation that is accelerated in a p53-null background (32). It should be interesting to test whether Plk1 overexpression accelerates tumor formation in a p53-null background.
To our surprise, Plk1 overexpression alone is not enough to drive cancer initiation, indicating that additional genetic modifications are required for tumor formation. Of note, inducible Plk1 knockdown mice described previously did not show any significant phenotype either (33). Increasing evidence suggests that Plk1 is directly involved in overall DNA damage response, including DNA damage checkpoint activation, checkpoint maintenance, damage recovery, and DNA repair (34). Accordingly, we challenged Plk1-overexpressing mice with IR to ask whether Plk1 overexpression affects DDR pathway. Indeed, Plk1 overexpression clearly showed increased radiation sensitivity and accelerated IR-induced carcinogenesis in mice. Based on our observation, we postulate that Plk1 is a critical regulator of IR-induced carcinogenesis, likely because of premature

Plk1 contributes to cancer formation
checkpoint termination and reduced DNA repair in Plk1-KI mice.
We understand that most of previous studies suggested that Plk1 plays an important role in turning off the later stages of DDR. Interestingly, we recently found that Plk1 also phosphorylates key factors upstream of ATM/ATR and regulates their DDR-related functions in human cells. Specifically, Plk1 directly phosphorylates Mre11, a component of the Mre11-Rad50 -Nbs1 complex, at serine 649 during DDR. Phosphorylation of Mre11 inhibited loading of the Mre11-Rad50 -Nbs1 complex to damaged DNA, leading to checkpoint termination and inhibition of DNA repair (35). This is one likely mechanism to explain why Plk1 KI mice are hypersensitive to ionizing radiation, because Plk1-overexpressed cells will continue to cycle even in the presence of DNA damage, eventually resulting in a much higher chance for tumorigenesis. In addition, Plk1-KI leads to inactivation of p53 and PTEN, two most important tumor suppressors. This statement has been supported by many of our previous publications. For example, we once showed that Plk1 depletion leads to p53 stabilization (36). Mechanistically, we later showed that Plk1 phosphorylation of GTSE1 and Topors, two negative regulators of p53, results in p53 inactivation (13,14). Directly related to Plk1-KI mouse work, we recently showed that Plk1 KI causes p53 inactivation in MEFs (37). Finally, our RNA-Seq analysis provided in Fig. 7 of this manuscript clearly showed that additional DNA damage repair genes are down-regulated upon Plk1 overexpression. Thus, Plk1-KI-associated increased cancer formation upon IR is likely due to multiple factors.

Animal experiments
All procedures involving mice were guided by Purdue University Animal Care and Use Committee (Protocol no. 1111000133E001). Mice housed in the animal facility with free access to standard rodent chow and water were under pathogen-free conditions and maintained in a 12-h light/12-h dark cycle.

Isolation of total RNA from murine tissues
After Plk1-KI and WT mice were quickly killed, different organs were prepared and stored in RNAlater stabilization reagent (Qiagen). For the RNA preparation, tissue pieces of 20 mg were dispersed with a PCR tissue homogenizing kit, followed by total RNA isolation using an RNeasy mini kit (Qiagen).

Establishment of MEFs derived from WT and Plk1-KI embryos
After removing the head and organs of embryos at day 13.5, the tissue was minced and rinsed with PBS, followed by incubation with 0.5 ml of 0.1% trypsin for 20 min at 37°C. Trypsin was inactivated by addition of 5 ml of DMEM supplemented with 10% FBS. After centrifugation, the cell pellets were resuspended with DMEM and plated onto a 10-cm dish. Medium was changed to remove the large clumps after 24 h. For Plk1-KI MEFs, 5 ϫ 10 6 cells were treated with 1 mM 4-hydroxytamoxifen for 48 h to induce Plk1 expression. The genotypes of embryos were detected by PCR with primers that are specific to the Plk1 sequence. The cells were cultured at 37°C with 8% CO 2 in DMEM supplemented with 10% (v/v) fetal calf serum, penicillin, streptomycin, and 50 M 2-mercaptoethanol. The MEFs were frozen as stocks at the second passage and used for subsequent studies.

Plk1 expression analysis
Real-time RT-PCR was carried out using the primer pairs 5Ј-TAATGACTCAACACGCCTGATT-3Ј and 5Ј-AGCTCA-GCAGCTTGTCTACCAT-3Ј. The ␤-actin-specific primers were used as controls: 5Ј-GAGGAGCACCCCGTGCTGC-3Ј and 5Ј-CCTGCTTGCTGCTGATCCACA-3Ј. To quantify Plk1 transcripts, 1 g of RNA was subjected to qRT-PCR. The qRT-PCR was conducted using a TaqMan gene expression assay for murine Plk1. As an internal control in the TaqMan assay, murine ␤-actin was used (Applied Biosystems). In brief, the 10-l PCR included 1 l of RT product, 1 ϫ TaqMan Universal PCR Master Mix, 0.2 l of TaqMan probe, 1 l of forward primer, and 1 l of reverse primer. The reactions were incubated in a 96-well plate at 95°C (10 min), followed by 40 cycles of 95°C (15 s) and 60°C (1 min). All reactions were run in triplicate.

Analysis of mouse genotypes
To test the genotype of WT or Plk1-KI mice, genomic DNA was prepared from tail clips 0.5-0.8 mm in length with Viagen Direct PCR-Tail reagent (Peqlab Biotechnologie, Erlangen, Germany) according to the manufacturer's protocol. For the standard PCR, 10 ng of genomic DNA was amplified using the sense primer 5Ј-ACTTCGTATAGCATACATTATACGAAG-TTATC-3Ј and the antisense primer 5Ј-TCCTTTACCC-AGAAAGCGC-3Ј.

FACS analysis
The cells in culture dishes were trypsinized, washed with PBS, and fixed in 95% ice-cold ethanol overnight. After centrifugation, the cells were resuspended in PBS and incubated with 50 mg/ml propidium iodide in the presence of 100 units/ml RNase A, followed by FACS analysis.

Measurement of energy expenditure
Oxygen consumption (VO 2 ) and carbon dioxide production (VCO 2 ) were measured under a consistent environmental temperature and light cycle using an indirect calorimetry system (Oxymax; Columbus Instruments). After the male mice were acclimated to the metabolic chamber for 2 days, VO 2 and VCO 2 were measured in each individual mouse at 15-min intervals during a 48-h period.

Glucose tolerance test
GTT was performed by intraperitoneal injection of D-glucose (Sigma) at a dose of 2 mg/g body weight with an Onetouch Ultra glucometer (Lifescan) into male mice after an overnight fast. Blood glucose levels were then measured at different times (15,30,45,60,75, and 90 min).

Chromosome aneuploidy
After cells were treated with 50 ng/ml colchicine for 4 h, the cells were collected and hypotonically swollen in 75 nM KCl for 30 min at 37°C. The cells were fixed in freshly made Carnoys fixative solution (75% methanol, 25% acetic acid) with three changes of fixative. The cells were then dropped onto glass slides and dried at room temperature. Chromosomes were stained with DAPI for 30 m, rinsed with PBS three times, and mounted.

Tumor analysis and histopathology
Aged mice were sacrificed by cervical dislocation following anesthetization with isofluorine. Necropsies were performed, and tissues as well as tumors observed by gross inspection were fixed in 10% formalin 2 days and then embedded in paraffin. For IHC analysis, representative sections were deparaffinized, rehydrated in graded alcohols, and processed using the avidinbiotin immunoperoxidase method. Sections were subjected to antigen retrieval by microwave oven treatment using standard procedures.

RNA-Seq analysis
MEFs were irradiated with 2 Gy ␥-ray, incubated for 12 h, and harvested for RNA extraction. Total RNA was extracted from 1 ϫ 10 7 cells using TRIzol reagent (Life Technologies), followed by purification with the RNeasy mini kit (Qiagen) based on the manufacturer's instructions. Quality of input total RNA was assessed using an Agilent Bioanalyzer RNA nano chip. Libraries were constructed largely as directed by the Illumina TruSeq Stranded mRNA sample preparation guide (catalog no. RS-122-9004DOC), using reagents from the Illumina TruSeq RNA library preparation kit (catalog nos. RS-122-2001 and RS-122-2002). Briefly, poly(A) ϩ RNA was isolated by annealing biotinylated oligo(dT) to total RNA, followed by capture with streptavidin-conjugated magnetic beads. Non-bound RNAs were then largely removed by discarding the supernatant after magnetic pelleting of Ampure beads. More binding buffer was added followed by an 80°C incubation to strand denature all RNA from the oligo(dT) captured poly(A) ϩ RNA, followed by another cycle of binding and supernatant removal. The poly(A) ϩ RNA was eluted in a divalent cation and hexamer first-strand synthesis primer containing buffer and fragmented with a 4-min incubation at 95°C. Reverse transcriptase in the presence of actinomycin D to repress any DNA-templated polymerization was used to synthesize first-strand synthesis of cDNA. Second-strand synthesis was performed, replacing dTTP with dUTP, which largely eliminates second-strand amplicons during subsequent amplification. cDNA ends were repaired, 3Ј-adenylated, and ligated to Illumina adapters. Products were subjected to a 0.8:1 Ampure:sample purification to reduce lower molecular weight amplicons, and the resulting libraries were assessed with an Agilent DNA high sensitivity chip for yield. Based on the overall yields, it was surmised that 8 cycles of amplification, rather than the 15 cycles recommended by the Illumina protocol, would be used. Amplified final libraries were titered for clustering using a KAPA library quantification kit Illumina (KAPA catalog no. KR045), pooled, and clustered in a single HiSeq 2500 v2 Rapid Chemistry lane to generate 60 base single reads. To analyze the data, all sequenced libraries were mapped to the mouse genome (UCSC mm10) using STAR RNA-Seq aligner (37). The uniquely mapped reads were assigned to mm10 refGene genes using featureCounts (from subread) (38). The data were normalized using TMM (trimmed mean of M values) method. The gene expression was finally summarized on base-2 logarithmic scale. We excluded genes with average expression level lower than 1 for all phenotypes. Differential expression analysis was performed using edgeR with paired samples setting (39,40). The false discovery rate was computed from p values using the Benjamini-Hochberg procedure. Genes with false discovery rates of Ͻ0.05 and absolute values of fold change (FC) larger than 1.5 were considered as differentially expressed genes (DEGs).

Statistical analysis
All data are presented as means Ϯ S.D. Statistical calculations were performed with Microsoft Excel analysis tools. Although a two-tailed, unpaired Student's t test was used to assess the difference between the effects of treatment in cell lines, one-way analysis of variance was used to determine statistically significant differences from the means in the animal study. p values of Ͻ0.05 were considered statistically significant. *, p Ͻ 0.05; **, p Ͻ 0.01.
Author contributions-Z. L. conducted most of the experiments, analyzed the results, and wrote most of the paper. J. Liu, J. Li, Y. K., and G. S. also conducted some of the experiments. X. R., Y. L., and J. W. analyzed RNA-Seq data. X. L. analyzed the data and wrote the paper with Z. L.