CIP2A Modulates Cell-Cycle Progression in Human Cancer Cells by Regulating the Stability and Activity of Plk1

Proper cell-cycle progression involves coordination of multiple events, such as chromosome condensation, spindle formation, chromosome segregation, and cytokinesis, and is tightly controlled by posttranslational modifications such as phosphorylation and ubiquitination (1, 2). For example, mitotic kinases, including cyclin-dependent kinase 1, polo-like kinases (Plk), and Aurora kinases, phosphorylate their substrates during mitotic progression. In contrast, components of the ubiquitin–proteasome system, such as the anaphase-promoting complex/cyclosome (APC/C), in turn directs the ordered destruction of critical mitotic substrates (i.e., spindle assembly checkpoint proteins) and mitotic kinases (1, 2). Thus, the concerted effort of mitotic kinases and the APC/C is required for fine tuning of cell-cycle progression.

Plk1 plays a crucial role in multiple steps of mitosis, including the G₂–M transition, centrosome maturation, bipolar spindle formation, chromosome segregation, and cytokinesis (3, 4). Plk1 contains an N-terminal kinase domain and a C-terminal polo-box domain (PBD) that have been implicated in regulating kinase activity and subcellular localization (3, 4). Plk1 activity begins to increase during G₂-phase and peaks in mitosis. This temporal control is tightly regulated by phosphorylation and ubiquitin-dependent proteolysis (4). In the late G₂-phase, Plk1 is activated by phosphorylation of Thr210 in the T-loop by Aurora A kinase/bora (4). In contrast, downregulation of Plk1 is mediated by the ubiquitin–proteasome system, involving APC/C-Cdh1 or -Cdc20, during the late M and G₁ phases (2, 4). Consistent with the diverse roles of Plk1 in mitosis, inhibition of Plk1 leads to multiple mitotic defects, including aberrant spindle formation, misaligned chromosomes, and improper chromosome condensation (5–7). In addition, Plk1 is highly upregulated in tumors and inhibition of its activity induces apoptosis in cancer cells, but not in normal cells (4, 8). Thus, Plk1 is a promising drug target in cancer therapy (8). However, the regulatory mechanisms of tumor-associated Plk1 are poorly understood.

Cancerous inhibitor of protein phosphatase 2A (PP2A; CIP2A), also named KIAA1524 or p90 tumor-associated antigen, is a novel oncogene that is known to inhibit c-Myc–associated PP2A activity and thereby stabilize the oncogenic c-Myc in human malignancies (9). Downregulation of CIP2A reduces malignant cellular growth and in vivo tumor formation (9–11). In addition, overexpression of CIP2A has been found in several common cancers and is associated with poor prognosis (9–14). Furthermore, recent reports have suggested that CIP2A is a potential target for anticancer drugs in several cancers (11, 15–18). Although the oncogenic role of CIP2A in human malignancies has been suggested, the mechanisms through which it exerts its oncogenic properties are still unclear.

In this study, we investigated the novel functions of CIP2A in cell division. We showed that CIP2A binds with Plk1 to regulate Plk1 stability and activity during mitosis. Furthermore, we provided evidence of the clinical relevance of this regulatory mechanism in human cancers.

HeLa, H1299, H460, Hs68, 293T, and 293 cells (American Type Culture Collection, ATCC) were cultured in Dulbecco's Modified Eagle Medium (Gibco-BRL) supplemented with 10% FBS and penicillin/streptomycin. Fluorescent ubiquitination-based cell-cycle indicator (FUCCI)-expressing HeLa cells were obtained from the RIKEN Cell Bank. Cells were preserved and passaged according to the ATCC protocol for no longer than 2 months and tested monthly for Mycoplasma infection by Hoechst 33258 staining. Cells were synchronized at G₁–S or G₂–M phase by a double thymidine block/release or nocodazole block protocol (19). Briefly, cells were treated with 2 mmol/L thymidine (Sigma) for 16 hours. After releasing cells for 2 hours, cells were transfected with siRNAs. Eight hours after release from the first thymidine block, thymidine was re-added for an additional 14 hours. Cells were released from the second thymidine block into medium with or without 100 ng/mL nocodazole (Sigma), dependent on the experimental design. To measure endogenous Plk1 stability, cells were incubated with 100 μg/mL cycloheximide (Sigma) for the indicated times. For proteosome inhibition, cells were incubated with 10 μmol/L MG132 (Sigma) for 3 hours. Okadaic acid (50 nmol/L; Sigma) or forskolin (40 μmol/L; Sigma) was used to inhibit or activate PP2A activity, respectively.

Human CIP2A cDNA purchased from Origene was PCR-amplified and cloned into pEXPR-IBA105 (Strep fusion) and pcDNA3.1-Myc (Myc fusion) vectors. CIP2A.1 siRNA–resistant cDNA, generated by inserting four silent mutations, was cloned into the pcDNA3.1-Myc vector. The following mutagenizing oligonucleotides were used: 5′-tcaccttgttggcccatagtagtttaactgtcgtcgtcttcgcactttcaatattatcc-3′ (sense), and 5′-ggataatattgaaagtgcgaagacgacgacagttaaactactatgggccaacaaggtgat-3′ (antisense). PTEN, Flag-tagged CHFR, and Plk1 constructs (WT, FAA, N-terminus, and C-terminus) were kindly provided by Dr. J.K. Park (Korea Institute of Radiological and Medical Sciences, Seoul, South Korea), Dr. J. H. Seol (Seoul National University, Seoul, Korea), and Dr. H. S. Yim (Hanyang University, Seongdong-gu, Korea), respectively.

The following sequences were used for RNA interference: CIP2A.1, 5′-cugugguuguguuugcacutt-3′; CIP2A.2, 5′-ggagugguuugucggagca-3′; CIP2A.3, 5′-caaguguaccacucuuaua-3′; Plk1, 5′-tgaagaagaucacccuccu-3′; Cdh1, 5′-ugagaagucucccagucag-3′; Cdc20, 5′-gaagaccugccguuacauu-3′; PP2A-Aα, 5′- gcaucaaugugcugucauatt-3′; PP2A-Aβ, 5′-cgacucaacaguauuaagatt-3′; and MAD2, 5′- aaagtggtgaggtcctggaaa-3′. Nonsilencing siRNA (Bioneer) was used as a negative control. Transfection of siRNAs (50 nmol/L CIP2A.1, CIP2A.2, Plk1, Cdc20, and MAD2; 100 nmol/L CIP2A.3; and 20 nmol/L PP2A-Aα and PP2A-Aβ) was conducted using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. For stable depletion of CIP2A, FUCCI–HeLa cells were transduced with control or CIP2A short hairpin RNA (shRNA) lentivirus (Santa Cruz Biotechnology, Inc.) according to the manufacturer's protocol.

The following antibodies were used: mouse monoclonal antibodies against CIP2A, cyclin B1 (GNS1), and p27 (Santa Cruz Biotechnology, Inc.); Cdc20, Cdh1 (DH01), and Plk1 (Abcam); Flag-tag and β-actin (Sigma); hemagglutinin (HA)-tag (Roche); phospho-MPM2 (Millipore); PP2Ac (BD Transduction Laboratories); and Strep-tag (Qiagen); rabbit monoclonal antibodies against Plk1 (208G4; Cell Signaling Technology, Inc.) and phospho-Akt (Epitomics); rabbit polyclonal antibodies against c-Myc, phospho-Plk1 (Thr210), phospho-histone H3 (phospho-H3), PTEN, and Myc-tag (Cell Signaling Technology, Inc.); pericentrin and MAD2 (Abcam); cyclin B1, cyclin A, and PP2A-Aα/β (Santa Cruz Biotechnology, Inc.); and CIP2A (Novus Biologicals).

Cells grown on coverslips were fixed with 4% paraformaldehyde in PBS. The cells were permeabilized and blocked with 0.2% Triton X-100 and 5% fetal calf serum in PBS. The fixed cells were consecutively incubated with primary antibodies against phospho-H3 (Millipore; 1:500), cyclin B1 (Santa Cruz Biotechnology, Inc.; 1:200), pericentrin (Abcam; 1:500), Plk1 (208G4; Cell Signaling Technology, Inc.; 1:200), or CIP2A (Santa Cruz Biotechnology, Inc.; 1:200) and secondary antibodies such as anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 594 (Molecular Probes). Slides were mounted in medium containing 4′,6-diamidino-2-phenylindole (DAPI), and images were then obtained using a confocal laser-scanning microscope (LSM 710; Carl Zeiss, Inc.). All images were processed by using ZEN 2009 Light Edition (Carl Zeiss) and Adobe Photoshop CS4.

Human tissue microarrays were purchased from two commercially available tissue arrays [SuperBioChips (cat. numbers: CC, CCN, CD, CDN, CQ, CQN, CZA, and BB) and AccuMax (cat. numbers: A206III and A206V)]. Tissue arrays were able to collect up to 142 lung cancer, 59 gastric cancer, 59 colon cancer, and 55 cervical cancer specimens, along with several matched normal tissues. Immunohistochemical staining was conducted with anti-CIP2A rabbit polyclonal antibody (Novus Biologicals; 1:100), anti-Plk1 rabbit polyclonal antibody (Abcam; 1:100), or anti-phospho-Plk1 rabbit monoclonal antibody (Abcam; 1:100). Immunostaining was detected by the avidin–biotin–peroxidase method according to the manufacturer's instructions (Invitrogen). The staining scoring was evaluated on a scale of 0–3. Staining intensity was scored as follows: 0 (no visible staining), 1+ (faint staining), 2+ (moderate staining), and 3+ (strong staining; Supplementary Fig. S1A).

Immunoprecipitation was conducted as previously described (20). Briefly, cells were lysed by NP-40 lysis buffer and the lysates were then precipitated with negative control mouse antibody (Santa Cruz Biotechnology, Inc.) or mouse monoclonal antibody against either CIP2A or Plk1 (Santa Cruz Biotechnology, Inc.). Immune complexes were collected using protein G-Sepharose and washed three times and SDS sample buffer was then added. Strep-tagged CIP2A proteins were pulled down by using Strep-Tactin beads (IBA TAGnologies), washed five times, and eluted with desthiobiotin (IBA TAGnologies) according to the manufacturer's protocol. The eluted samples were size-fractionated by electrophoresis and detected using immunoblotting.

HeLa cells released from double thymidine block/release or nocodazole block were trypsinized, washed twice in PBS, and fixed with ice-cold 70% ethanol. Fixed cells were incubated with 50 μg/mL propidium iodide and 100 μg/mL RNase at 37°C for 30 minutes and then analyzed with a FACScan flow cytometer (Becton Dickson).

FUCCI-expressing HeLa cells (21) were transduced with control or CIP2A shRNA lentivirus (Santa Cruz Biotechnology, Inc.) according to the manufacturer's protocol. Cells synchronized by a double thymidine block were released with fresh medium and time-lapse images were acquired at 30-minute intervals using confocal and phase-contrast microscopy (Carl Zeiss, Inc.).

Cells transfected with HA-tagged ubiquitin and empty or CIP2A-Myc vector were synchronized at the G₁–S border by a double thymidine block and then released. Eight hours after release, the cells were treated with MG132 for 3 hours and then analyzed for ubiquitination in vivo, as previously described (22). Briefly, the cells were lysed by incubation with 2 volumes of TBS containing 2% SDS at 95°C for 10 minutes. After adding eight volumes of 1% Triton X-100 in TBS, lysates were sonicated for 2 minutes and then precleared with protein G-Sepharose. The lysates were then immunoprecipitated with anti-Plk1 antibody (Abcam) coupled to protein G-Sepharose. The beads were washed with 0.5 mol/L LiCl in TBS, washed twice with TBS, and immunoblotted using anti-HA antibody.

Recombinant active His-Plk1 (produced in Sf-9 cells) was purchased from R&D Systems. Strep-tagged CIP2A from 293 cells expressing Strep-CIP2A was purified by the Strep-tag purification method. Various concentration ratios of purified Strep-CIP2A protein (0.25, 0.5, or 1 μg) were subjected to in vitro kinase assay with active His-Plk1 (0.1 μg). In vitro kinase assays were conducted in kinase buffer (20 mmol/L HEPES, pH 7.5, 150 mmol/L KCl, 10 mmol/L MgCl₂, 1 mmol/L EDTA, 2 mmol/L dithiothreitol, 5 mmol/L NaF, and 0.2 mmol/L Na₃VO₄) supplemented with 50 μmol/L ATP and 2.5 μCi of [γ-³²P] ATP at 30°C for 30 minutes in the presence of 2 μg dephosphorylated α-casein (Sigma). The reaction mixture was analyzed by SDS-PAGE and autoradiography. BI2536 (Selleck Chemicals) was used as a positive control to inhibit Plk1 activity.

In situ proximity ligation assay (PLA) was conducted as previously described (23). Paraformaldehyde-fixed HeLa cells were permeabilized with 0.2% Triton X-100, washed, and blocked with blocking solution (Olink Bioscience). Antigen-retrieved normal and cancer tissues (SuperBioChips) were incubated with 3% hydrogen peroxide, washed, and blocked with blocking solution. Mouse monoclonal anti-CIP2A antibody (Santa Cruz Biotechnology, Inc.; 1:200) together with rabbit polyclonal anti–phospho-Plk1 (Thr210) antibody (Cell Signaling Technology, Inc.; 1:100) or rabbit polyclonal anti-CIP2A antibody (Novus Biologicals; 1:200) together with mouse monoclonal anti-Plk1 antibody (Abcam; 1:300) were used for the proximity ligation reaction. The assay was conducted according to the manufacturer's protocol using the Duolink Detection Kit with a pair of nucleotide-labeled secondary antibodies (rabbit PLA probe MINUS and mouse PLA probe PLUS; Olink Bioscience). Nuclei were stained with Hoechst, which was included in the PLA reagent kit. Amplified PLA signals, represented as red fluorescent dots, were analyzed using confocal microscopy and quantified using CellProfiler software (available at www.cellprofiler.org).

The correlation between CIP2A and Plk1 immunointensity was analyzed using Spearman rank correlation test. The two-tailed Student t test was conducted to analyze statistical differences between groups. P values of less than 0.05 were considered statistically significant. Statistical analyses were conducted using Excel and XLSTAT software.

To investigate the role of CIP2A in cell division, we first examined CIP2A protein expression during different phases of the cell cycle in HeLa cells synchronized by using a double thymidine and thymidine–nocodazole block protocol (Fig. 1A). The expression of CIP2A increased during the G₂–M phase (Fig. 1A, left) and decreased during the exit from mitosis (Fig. 1A, right). Moreover, most of the CIP2A protein expressed was localized in the cytoplasm during interphase, and, as cells passed from the S-phase to the G₂–M phase, CIP2A was initially concentrated in the pericentrosomal region and then clearly localized at spindle poles during mitosis, as shown by immunofluorescence analysis (Fig. 1B). Notably, CIP2A was enriched in the nucleus during entry into mitosis (Fig. 1B and C). These cell cycle–dependent expression and localization patterns of CIP2A were also observed in two other cell lines (Supplementary Fig. S2A and S2B).

To further understand the cell cycle–dependent regulation of CIP2A, we used cells that expressed high or very low amounts of CIP2A, that is, H1299 lung cancer cells or Hs68 normal fibroblasts, respectively (Supplementary Fig. S2C). Because it has been well established that PTEN induces G₁ arrest in most types of cancer cells (24, 25) and that Plk1 depletion causes G₂–M arrest but not apoptosis in Hs68 cells (8), we overexpressed a WT PTEN construct in H1299 cells (PTEN null) or depleted Plk1 with siRNA in Hs68 cells. Ectopic overexpression of WT PTEN suppressed CIP2A expression and caused G₁ arrest in H1299 cells (Fig. 1D). On the other hand, CIP2A expression was obviously increased in Plk1-depleted and mitotic Hs68 cells (Fig. 1E and F). PTEN overexpression or Plk1 depletion did not affect cell viability (Supplementary Fig. S2D). Likewise, we also observed that irradiation, which induces G₁ or G₂–M arrest in a p53-dependent manner (26), triggered a reduction in CIP2A expression in H460 cells (p53 WT), but not in H1299 cells (p53 null; Supplementary Fig. S2E). Taken together, these results indicated that the expression and localization of CIP2A were associated with cell-cycle progression.

Next, we designed three siRNAs against different regions of CIP2A and found that two of them (#1 and #2 siRNAs) were highly efficient in silencing the CIP2A gene, but #3 siRNA exhibited poor efficiency (Fig. 2A, top). In addition, we constructed a CIP2A siRNA-resistant rescue vector (CIP2Arv) against the #1 siRNA to use in rescue experiments (Fig. 2A, bottom).

To examine the role of CIP2A during mitosis, cells were arrested at mitosis using nocodazole (Fig. 2B). Interestingly, mitotic arrest was clearly reduced in CIP2A-depleted cells, similar to MAD2 depletion or CHFR overexpression, which are known to cause reduction of the mitotic index by inducing defects in the spindle checkpoint (1, 27) or the activation of the prophase checkpoint (28) under nocodazole treatment, respectively (Fig. 2B). Consistent with this, mitotic phosphoproteins, assessed by Western blotting with anti–phospho-MPM2 antibody, were downregulated in CIP2A-depleted (#1 siRNA), MAD2-depleted, or CHFR-overexpressing cells compared with cells transfected with control siRNA or poor-efficiency CIP2A siRNA (#3; Fig. 2C). Importantly, the reduced mitotic index was restored by overexpression of CIP2Arv, excluding off-target effects of siRNA (Fig. 2D and E). A high percentage of MAD2-depleted cells contained multilobed nuclei, whereas CHFR-induced premitotic arrest did not affect nuclear morphology (Fig. 2F and G); these results are consistent with those of previous studies (27, 28). Interestingly, the presence of normal interphase nuclei in CIP2A-depleted cells suggested that, similar to CHFR overexpression, CIP2A depletion induced premitotic arrest (Fig. 2F and G).

Because CIP2A has been shown to inhibit c-Myc–associated PP2A (9), we speculated that CIP2A-mediated cell-cycle regulation may be dependent on PP2A. Unexpectedly, both inhibition of PP2A either by PP2A-specific siRNAs or okadaic acid treatment and activation of PP2A by forskolin treatment were not able to rescue the CIP2A-depletion phenotype (Supplementary Fig. S3A and S3C). Consistent with this, depletion or activation of PP2A did not rescue reduced mitotic phosphorylation due to CIP2A depletion (Supplementary Fig. S3B and S3D). Along with the observation that CIP2A depletion induced premitotic arrest, these results indicated that CIP2A was required for mitotic progression independent of PP2A.

To further investigate the role of CIP2A during mitotic progression, cells were transfected with CIP2A siRNAs in the interval between the two thymidine blocks and then released into the cell cycle with or without nocodazole (Fig. 3A, top). CIP2A depletion retarded the entry of synchronized cells into mitosis compared with the control cells, regardless of nocodazole treatment (Fig. 3A, bottom). Moreover, the accumulation of mitotic phosphoproteins and mitotic regulators, such as cyclin B1 and Plk1, was delayed in CIP2A-depleted cells, whereas the expression of cyclin A, which ordinarily accumulates in S and G₂ phases, was elevated (Fig. 3B, bottom). However, CIP2A depletion did not significantly affect the expression of c-Myc protein (Fig. 3B, bottom) or the kinetics of DNA replication during cell-cycle progression (Fig. 3B, top). To distinguish cells in the S–G₂–M (green) phase from the G₁-phase (red), we used a FUCCI. FUCCI can help visualize the dynamics of cell-cycle progression by harnessing the ubiquitination oscillators that control cell-cycle transitions (21). CIP2A was stably depleted by transducing FUCCI-expressing HeLa cells with a lentiviral vector expressing CIP2A shRNA (Fig. 3C, right). Knockdown of CIP2A delayed entry of the cells into mitosis and the G₁-phase by approximately 2 to 3 hours and resulted in a much more irregular entry into mitosis compared with that in cells expressing control shRNA (Fig. 3C, left, and D). Although CIP2A depletion primarily prevented mitotic entry, prolonged culture allowed delayed entry into mitosis. Therefore, we next asked whether CIP2A depletion caused mitotic defects. After 72 hours, CIP2A depletion significantly increased aberrant micronuclei, and this phenotype was rescued by cotransfection with CIP2Arv (Fig. 3E and F). In addition, CIP2A depletion caused multiple mitotic abnormalities (Fig. 3G and H), including the formation of multipolar spindles (Fig. 3H, ii), monopolar spindles (Fig. 3H, iii), misaligned chromosomes (Fig. 3H, iv and v), and lagging chromosomes (Fig. 3H, vi). Taken together, these results suggested that CIP2A was required for timely entry into mitosis and the proper completion of mitosis.

Because the stage-specific activity of mitotic kinases is crucial for proper cell-cycle progression (1) and because CIP2A is required for mitotic entry, we screened for kinases capable of rescuing CIP2A depletion mitotic delay. We overexpressed several mitotic kinases, including Aurora A, Aurora B, Plk1, and Nek2A, and among all of these kinases, only Plk1 rescued the CIP2A-depletion phenotype (Fig. 4A and B). Thus, we hypothesized that CIP2A regulated Plk1 activity during cell-cycle progression. To test this, we conducted coimmunoprecipitation assays with CIP2A and Plk1. Interestingly, CIP2A interacted with endogenous Plk1 (Fig. 4C), and this interaction was particularly apparent during mitosis (Fig. 4D). To further dissect the interaction between CIP2A and Plk1, cells were transfected with Strep-tagged CIP2A with various Flag-tagged Plk1 constructs, including WT, polo-box–mutant (three point mutations at W414F, V415A, and L427A, which are known to disrupt the PDB function (FAA; ref. 29), N-terminal–mutant (amino acids 1–305; N), and C-terminal–mutant (amino acids 306–603; C) Plk1, followed by a pull-down assay using Strep-CIP2A. As shown in Fig. 4E, Strep-CIP2A associated with WT Plk1, FAA Plk1, and C-terminal Plk1, but not with N-terminal Plk1. In addition, the interaction was reduced in FAA Plk1 compared with WT or C-terminal Plk1 (Fig. 4E), indicating that the CIP2A–Plk1 interaction required the intact PBD of Plk1. To further confirm this interaction, in situ PLAs were conducted to visualize the in vivo interactions between two proteins. Notably, positive signals indicating interactions between CIP2A and Plk1 were clearly observed in mitotic cells compared with interphase cells (Fig. 4F and G). Moreover, these signals began to increase from prophase, peaked at prometaphase, and then gradually decreased in anaphase (Fig. 4H and I). Taken together, these results showed that CIP2A interacted directly with Plk1 during mitosis.

Because Plk1 directly interacted with CIP2A (Fig. 4) and its expression was decreased in CIP2A-depleted cells (Fig. 3B; Supplementary Fig. S3B), we hypothesized that Plk1 stability may be affected by CIP2A during mitosis. To investigate this possibility, we examined the half-life of mitotic Plk1 in the presence of protein synthesis (cycloheximide) or proteasome (MG132) inhibitors. CIP2A depletion significantly reduced the stability of Plk1 protein after cycloheximide treatment (Fig. 5A and B). Interestingly, inhibition of the proteasome with MG132 blocked Plk1 degradation in CIP2A-depleted cells (Fig. 5C), supporting that CIP2A has a role in the stabilization of Plk1. Indeed, ectopic overexpression of CIP2A reduced the ubiquitination of Plk1 in mitotic cells (Fig. 5D). Because Plk1 degradation is mediated by APC/C-Cdh1 or -Cdc20, we determined which of these cofactors were involved in the regulation of Plk1 stability in CIP2A-depleted cells. Notably, Plk1 levels in CIP2A-depleted cells were restored by Cdh1 depletion, but not by Cdc20 depletion (Fig. 5E), indicating that CIP2A stabilized Plk1 by interfering with APC/C-Cdh1–mediated degradation during mitosis.

Unexpectedly, we also observed that CIP2A depletion significantly reduced γ-tubulin recruitment and microtubule organization during mitosis (data not shown). Because Plk1 activity is critical for centrosome maturation during mitosis (6, 30), we considered the possibility that CIP2A depletion may affect Plk1 activity and stability. To test this possibility, we measured the level of Plk1 phosphorylation at Thr210, indicative of Plk1 activation, and Plk1 kinase activity. Notably, Plk1 phosphorylation and kinase activity were dramatically decreased during mitosis in CIP2A-depleted cells compared with control cells, despite the increase in total Plk1 (Fig. 5F and Supplementary Fig. S4A and S4B). To further confirm this, we conducted in vitro kinase assays using Strep-CIP2A (purified from 293 cells) and His-Plk1 (obtained from Sf9 cells) protein. Consistent with the decrease in Plk1 activity by CIP2A depletion, CIP2A protein noticeably enhanced Plk1 activity and autophosphorylation (Fig. 5G). Furthermore, we observed that CIP2A-depleted cells were less sensitive to BI2536, a pharmacologic inhibitor of Plk1, than control cells (Supplementary Fig. S5), supporting that Plk1 possessed CIP2A-dependent activity. Taken together, these results suggested that the CIP2A–Plk1 interaction increased both the activity and stability of Plk1 during mitosis.

To assess the physiologic relevance of CIP2A-mediated Plk1 regulation in human cancers, we evaluated CIP2A and Plk1 expression using tissue microarrays derived from distinct human cancer tissues (see the Materials and Methods for details). Consistent with other reports, CIP2A and Plk1 were significantly increased in lung, gastric (stomach), colon, and cervical cancers (Fig. 6A). Importantly, the staining regions of these two proteins in serial sections of the same tissues were strikingly similar (Fig. 6B–D). Similar staining patterns were also observed in various cancer tissues (Supplementary Fig. S1B). Furthermore, we found that CIP2A expression was strongly correlated with Plk1 expression in four common types of cancer tissues (Supplementary Table S1; lung cancer, ρ = 0.682; cervical cancer, ρ = 0.77; gastric cancer, ρ = 0.787; and colon cancer, ρ = 0.735). Finally, we verified the in vivo interactions of these two proteins in cancer and normal tissues by in situ PLA. Remarkably, positive signals were predominantly detected in cancer tissues compared with their normal counterparts (Fig. 6E and F). Of the normal tissues tested, positive signals were also detected in stomach and colon tissues, both highly proliferative tissues, but to a lesser extent than in corresponding cancer tissues (Fig. 6E). Collectively, our data provided strong evidence for the physiologic relevance of CIP2A-mediated Plk1 regulation in a wide variety of human cancers.

In this study, we showed a novel regulatory function for CIP2A during cell-cycle progression. Although CIP2A is a well-known tumor marker and potent oncogene in a variety of tumors, the regulatory functions of CIP2A during cell-cycle progression have not been well studied. We showed for the first time that CIP2A controlled mitotic progression by directly interacting with the mitotic kinase Plk1, thereby stabilizing and promoting Plk1 activity. We also showed that CIP2A and Plk1 expression was tightly correlated in a large variety of cancers. Therefore, these findings suggested that CIP2A plays an important role in fine-tuning Plk1 expression and ultimately contributes to malignant cell growth and tumor formation.

Our results revealed that CIP2A was associated with cell-cycle regulation. Similar to our observations, several lines of evidence have indicated that cell cycle–arresting agents/conditions modulate CIP2A expression in various cancer cells. For example, serum starvation reduces CIP2A expression in gastric cancer cells and doxorubicin treatment reduces it in p53-proficient cells but not in p53-deficient cells (31, 32). Moreover, several cell cycle–promoting factors, such as c-Myc and E2F1, have been reported to stimulate CIP2A expression (12, 33). In addition to the cell cycle–dependent expression of CIP2A, we also found that CIP2A exhibited nuclear and centrosomal localization during mitosis. Consistent with this observation, a recent report has shown that cytoplasmic CIP2A shuttles into the nucleus upon treatment with leptomycin B, an inhibitor of the nuclear export factor (34), suggesting that the nuclear-cytoplasmic shuttling of CIP2A may be important for mitotic progression. Furthermore, we showed that CIP2A is required for mitotic progression. Depletion of CIP2A and overexpression of CHFR, which regulates entry into mitosis by ubiquitinating Plk1 upon microtubule poisoning (28), caused the same phenotype of Plk1-dependent premitotic arrest, albeit via different mechanisms, thus supporting the occurrence of CIP2A-mediated Plk1 regulation during mitotic progression. Similarly, a recent report showed that depletion of CIP2A disrupted the cell cycle in human ovarian cancer cells (35). In terms of c-Myc regulation, many reports have observed reduced levels of c-Myc in CIP2A-depleted cells (9, 12, 32, 35, 36). These studies were inconsistent with our current observation that CIP2A depletion did not affect c-Myc levels or the kinetics of DNA replication; however, it should be noted that these previous reports examined c-Myc levels at 72 hours after CIP2A depletion, while we observed c-Myc expression after relatively short-term depletion. In addition, a previous report showed that S-phase synchronization in thymidine/aphidicolin-treated gastric cancer cells did not affect CIP2A expression compared with unsynchronized cells, suggesting that cell-cycle activity is not associated with c-Myc–mediated regulation of CIP2A expression in gastric cancer cells (12). Therefore, it is likely that CIP2A-dependent cell-cycle regulation is independent of c-Myc.

Our data indicated that CIP2A blocks ubiquitin-dependent proteolysis of mitotic Plk1, similar to previous studies that have shown the CIP2A-dependent proteolytic degradation of c-Myc (9). On a molecular level, R337, located near the C-terminal region of Plk1, acts as a functional destruction box (4). Given that CIP2A interacted with the C-terminal region of Plk1 and that ectopic overexpression of CIP2A inhibited Plk1 ubiquitination, it is possible that the CIP2A–Plk1 interaction could hinder the reaction between APC/C-Cdh1 and the destruction box on Plk1, thereby decreasing Plk1 ubiquitination. APC/C-Cdh1 is thought to be active during late M and G₁-phase (2). However, recent observations have indicated that APC/C-Cdh1 is also active during early mitosis (37). Thus, the CIP2A–Plk1 interaction could also protect Plk1 from APC/C-Cdh1–dependent proteolysis throughout mitosis.

In addition to the regulation of Plk1 stability, we also found that CIP2A regulated Plk1 activity during mitosis. The increased autophosphorylation and activity of Plk1 induced by the addition of CIP2A protein may suggest that the interaction of CIP2A–Plk1 via PBD could relieve the autoinhibitory interaction of PBD with the kinase domain of Plk1 (3, 4). The Plk1 PBD plays a pivotal role in mediating both the kinase activity and localization of Plk1 (3, 4). Interestingly, we also found that CIP2A depletion increased Plk1 mislocalization during mitosis (Supplementary Fig. S6). Proper localization of Plk1 is required for the phosphorylation and activation of Plk1 during mitosis (3, 4). Thus, it is also likely that decreased activity of Plk1 in CIP2A-depleted cells may be caused by mislocalization of Plk1.

CIP2A is a clinically relevant prognostic marker in many types of cancers. For example, high CIP2A expression predicts poor survival in several types of cancers (10, 13, 14, 36, 38). Furthermore, recent reports suggest that CIP2A expression is associated with resistance to several anticancer drugs (14, 15, 31). Likewise, it has been well established that Plk1 is upregulated in tumors and has been validated as a therapeutic target in many types of cancers (8). In addition, Plk1 plays a key role to mitotic reentry following DNA damage-induced arrest, a mechanism contributing to tumor resistance against anticancer drugs (39–41). Our immunohistochemical and PLA analyses indicated that CIP2A tightly cooperated with mitotic Plk1 in a wide variety of human cancers. Thus, our findings suggested that the CIP2A–Plk1 complex may serve as a potential prognostic marker for poor survival and tumor resistance against anticancer drugs. In addition, small molecules interfering with CIP2A–Plk1 binding or triggering CIP2A downregulation could be effective as antimitotic drugs for cancer therapy.

How can CIP2A modulate mitotic Plk1 through multiple mechanisms? CIP2A has multiple predicated domains, including armadillo-like, transmembrane, leucine zipper, and coiled-coil domains (Supplementary Fig. S7A). Interestingly, under nonreducing conditions, the molecular mass of endogenous CIP2A was approximately 360 kDa (Supplementary Fig. S7B), suggesting that endogenous CIP2A forms a multimeric complex. Scaffold proteins, which usually form large complexes, are crucial regulator of particular signaling pathways, acting by tethering signal components, localizing these components to specific areas, coordinating positive or negative feedback signals, and protecting the signaling components (42). This study suggested that CIP2A regulated mitotic Plk1 via multiple mechanisms, including the regulation of stability, activity, and localization. On the basis of these aspects, one reasonable answer to the question stated earlier is that CIP2A is a transient mitotic scaffold for Plk1 during mitosis.

In conclusion, our results collectively indicated that CIP2A acted as a mitotic regulator by directly regulating Plk1 and was required for mitotic progression in human cancer cells. Our work provides novel molecular insights into CIP2A-mediated Plk1 regulation in proliferating cancer cells.

No potential conflicts of interest were disclosed.

Conception and design: J.-S. Kim

Development of methodology: J.-S. Kim, E.J. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.-S. Kim, E.J. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-S. Kim, E.J. Kim, J.S. Oh, I.-C. Park, S.-G. Hwang

Writing, review, and/or revision of the manuscript: J.-S. Kim, J.S. Oh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-S. Kim

Study supervision: J.-S. Kim

The authors thank A. Miyawaki (RIKEN) for providing the HeLa cells expressing the FUCCI probes.

This work was supported by Basic Science research Program (Grant no. 2012R1A1A2002955) and the Nuclear Research and Development Program through a National Research Foundation of Korea (NRF; Daejeon, Korea) grant funded by the Korean government.

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.