Sak/Plk4 and Mitotic Fidelity

Sak/Plk4 and mitotic ﬁdelity

Carol J Swallow1,2, Michael A Ko1,2, Najeeb U Siddiqui1,3,4, John W Hudson5 and James W Dennis*,1,3,4

1Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Ave. R988, Toronto, Ontario, Canada M5G 1X5; 2Department of Surgery, University of Toronto, Ontario, Canada; 3Department of Microbiology and Medical Genetics, University of Toronto, Ontario, Canada; 4Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada; 5Department of Biological Sciences, University of Windsor, Ontario, Canada

Sak/Plk4 differs from other polo-like kinases in having (Fernebro et al., 2002). Mutation of classical tumor only a single polo box, which assumes a novel dimer fold suppressor genes follows the Knudsen 2-hit model, that localizes to the nucleolus, centrosomes and the whereby loss of the wild-type allele as a ‘second hit’ cleavage furrow.Sak expression increases gradually in S frequently results in relaxed cellular growth controls through M phase, and Sak is destroyed by APC/C (Knudson, 1971). Inherited cancer syndromes such as dependent proteolysis.Sak-deficient mouse embryos Li-Fraumeni (p53, Chk2) and ataxia telangiectasia arrest at E7.5 and display an increased incidence of (ATM) are examples of autosomal recessive mutations apoptosis and anaphase arrest.Sak þ /À mice are haploin- in checkpoint proteins that normally delay the cell cycle sufficient for tumor suppression, with spontaneous tumors in response to DNA damage and environmental stresses developing primarily in the liver with advanced age. (Malkin et al., 1990; Bell et al., 1999). Partial or During liver regeneration following partial hepatectomy, complete loss of their activities leads to genomic Sak þ /À hepatocytes display a delay in reaching the first M instability, further mutations and cancer (Sherr, 2004). phase, multipolar spindles, disorganized tissue morphol- A weakening of checkpoint controls increases the ogy and loss of acuity for cyclin B1 expression.Similarly, probability of mitotic errors over multiple cell divisions. Sak þ /À MEF cells proliferate slowly, and show a high Thereafter, progression of benign lesions to malignancy incidence of centrosome hyper-amplification.We suggest becomes a form of natural selection for cells bearing that Sak provides feedback to cell cycle regulators, and additional activating mutations in proto-oncogenes and thereby precision to the switch-like transitions of centro- loss of tumor suppressors. Predisposition to CIN also some duplication and exit-from-mitosis.Sak binds to p53, occurs in more subtle forms through partial deficiency in and studies are underway to provide a molecular context genes controlling the cell cycle. Notably, proteins that for the Sak-p53 interaction.Animal models of haploin- affect the sequential expression of cyclins D, E, A, B and sufficiency and more comprehensive models of cell cycle activation of their partners Cdks 6, 4, 2, 1 can alter the regulation should contribute to improvements in cancer timing and fidelity of sequential events, including risk assessment and novel therapies. centrosome duplication, spindle assembly, anaphase Oncogene (2005) 24, 306–312. doi:10.1038/sj.onc.1208275 and cytokinesis (Elledge, 1996). The Cdk family of Ser/Thr kinases determines the sequential order of cell Keywords: mitosis; Sak/Plk4; chromosomal instability; cycle events, but they are also responsive to inputs colorectal cancer; hepatoma from the physical and biochemical environment that can impose delay or arrest (Sherr, 2004). Polo-like kinases (plks) are components of this complex circuitry and impact multiple steps in cell cycle progression, although much remains to be learned concerning Introduction: Chromosomal instability and their interacting partners (Blagden and Glover, 2003; haploinsufficient tumor suppressors Barr et al., 2004). Cyclin-dependent kinase inhibitors (CKI) negatively DNA replication and chromosomal segregation occur regulate Cdks, generally in response to extrinsic condi- with remarkable fidelity at all levels of complexity, from tions, and thereby delay or suppress cell cycle progres- a fertilized egg through to the billions of cells in adult sion. Mice with a single functional allele of the mammals. However, errors do occur and can lead to p21(WAF1/CIP1) or INK4 family members display aneuploidy, chromosome breaks and translocations. early cancer development (Jackson et al., 2003). Chromosomal instability (CIN) is very common in Similarly, mice heterozygous for the CKI p27Kip display malignant tumors, occurring in B80% of colon cancers hypersensitivity to a spectrum of mutagenic agents, producing tumor phenotypes intermediate between that *Correspondence: JW Dennis; of wild type and null mice (Philipp-Staheli et al., 2001). E-mail: [email protected] However, tumors in the heterozygous mice retain the Plk4 and mitotic fidelity CJ Swallow et al 307 wild-type allele and p27Kip protein expression is not a silenced, indicating that reduced dosage is sufficient to Kinase domaincry-pb pb both enhance initiation and progression to malignancy Sak (Philipp-Staheli et al., 2001; Muraoka et al., 2002). 12 265 596 836 847 911 Compound mutant mice heterozygous for CKIs or aa lacking other negative regulators of G1/S progression, for example p21Cip1 þ /À plus p18INK4cÀ/À or Rb þ /À, display further acceleration and increased penetrance of tumor ATP T-loop PEST1 PEST2 PEST3 binding development (Tsai et al., 2002; Bai et al., 2003; Jackson Thr170 272-300 808-832 840-878 et al., 2003). Haploinsufficiency for cancer suppression Lys41 has been observed for the spindle assembly checkpoint n k protein Mad2, with elderly heterozygote mice showing io b t c a pb c S k ∆ D e one k an increased incidence of cancer, as seems typical for 0 M f l T a s a 7 1 W Moc S k 1 4 n a a n these spindle assembly checkpoint insufficiencies T K r i S t e k k s T IP: Flag (Michel et al., 2001). The majority of human cancers a a o a W S S N C show no obvious inheritance pattern, but it appears Probe: Flag rather that mildly hypomorphic alleles of genes that encode mitotic regulators may act in combination to IP: p53 increase the probability of chromosomal pathologies and 12345 Probe: Flag cancer. Furthermore, the aging process involves cumula- 123 tive oxidative damage to gene promoters (Lu et al., 2004), Figure 1 (a) Structural features of Sak. Sak has a Ser/Thr kinase and a selective reduction in expression of mitotic domain at the amino-terminus, a single polo-box domain (pb) at regulators and centrosome proteins including cyclins A, the carboxy-terminus, and three PEST destruction motifs. The B, F and Plk1 (Ly et al., 2000). In combination, conserved ATP-binding domain with the required Lys41 residue and the T-loop activation domain with the required Thr170 are incremental losses in multiple mitotic regulators may indicated. The cryptic polo-box (cry-pb) interacts with the Sak dampen the switch-like behavior of cyclin/cdk activation kinase domain and with Tec kinase . Both pb and cry-pb are self- and destruction, thereby increasing the likelihood of association domains, suggesting that Sak may form a homodimer. mitotic errors and cancers over time. Sak/Plk4 is required (b) Sak kinase activity. In vitro phosphorylation of casein by for mitotic fidelity, and mice with a single Sak gene FLAG-Sak constructs immunoprecipitated from NIH 3T3 cells. Lane 1, Wild-type FLAG-Sak; lane 2, FLAG-SakT170D (activat- display an increased incidence of cancer with advancing ing); lane 3, FLAG-SakK41M (inactivating); lane 4, no transfec- age (Ko et al., submitted). Herein we review the current tion; lane 5, casein only. (c) Sak associates with p53. Proteins information on Sak kinase structure, function and role in immunoprecipitated from NIH-3T3 cells transfected with 3X- cell cycle control and carcinogenesis. FLAG-Sak constructs using anti-FLAG (upper panel) or anti-p53 (lower panel) antibodies, probed with anti-FLAG. Lane 1, Wild- type FLAG-Sak; lane 2, mock transfection; lane 3, FLAG-Sak pb deleted (SakDpb) Sak gene structure

The Sak gene is found on mouse chromosome 3 and additional complexities of coordinating cell division, human chromosome 4q28, a region that frequently polarity and checkpoints in animals. Curiously, plk undergoes rearrangement or loss in human cancers, and homologs are absent in Arabidopsis (Initiative, 2000), at a particularly high rate in hepatocellular carcinomas suggesting a possible connection with basic structural (E70%) (Hammond et al., 1999). The kinase domain of differences, such as the absence of centrosomes and the murine Sak is followed by 660 amino acids containing a presence of a preprophase band that can contribute to single 64 amino-acid polo box (pb) domain at the spindle and phragmoplast orientation in plants, while COOH end (Figure 1a). The Sak-pb is equally related by yeast and animal cells require additional inputs from phylogenetic comparison to the pb1 and pb2 domains of spindle poles, the membrane cortex and other extrinsic other plks. The Sak-pb domain (Sak839–925) and the sources. Mammalian Plk1 and Plk3 functions overlap upstream region designated the ‘cryptic polo box’ and are conserved, as expression of either Plk1 or Plk3 (cry-pb) (Sak596–836) are both self-association domains in ts cdc5-1 mutant yeast cells rescues the mitotic defects (Leung et al., 2002). As might be expected, Sak kinase (Ouyang et al., 1997; Lee et al., 1998). These and other activity is markedly increased by a SakT170D muta- observations suggest that the mitotic functions of Cdc5 tion in the T loop, while SakK41M, an inactiva- may be distributed among the various mammalian plks, ting mutation in the ATP-binding domain, eliminates thereby affording additional levels of cell cycle control activity (Figure 1b). during development. However, Sak expression in ts The kinase domain of Sak is most closely related to cdc5-1 cells did not rescue the growth defect, suggesting that of plks, but Sak appears to have diverged from a a nonredundant role for Sak kinase. Furthermore, pb primordial plk early in the radiation of metazoans (Fode domain swaps between Cdc5or Plk1 and Sak did not et al., 1994). The plk-related genes number 1 in single permit rescue of cdc5-1 mutant cells (Hudson and cell eukaryotes, 3 in C. elegans,2inDrosophila and 4 in Dennis, unpublished results). vertebrates (Hudson et al., 2001). This modest expan- The C-terminus of Sak contains 3 PEST sequences, sion of the family may have been driven by the which are commonly associated with reduced protein

Oncogene Plk4 and mitotic fidelity CJ Swallow et al 308 stability, and indeed Sak protein displayed a short half- different between the two proteins (Leung et al., 2002; life (2–3 h) in nonsynchronized cells (Fode et al., 1996). Elia et al., 2003). The Sak-pb (Sak870–960) forms an Sak protein is ubiquitinated and destroyed by the intermolecular homodimer, in which b-strands 6, 1, 2, anaphase-promoting complex (APC/C) following mito- and 3 from one monomer form a contiguous antiparallel tis, similar in this regard to Cdc5(Shirayama et al., sheet with b-strands 4 and 5from the second monomer 1998), and Plk1 (Lindon and Pines, 2004). The removal (Leung et al., 2002). In contrast, the two pb domains of of PEST sequences from Sak enhances the stability of Plk1 form an intramolecular heterodimer. These distinct the protein (Yamashita et al., 2001). Tec tyrosine kinase Sak and Plk1 pb dimer structures place different binds to the cry-pb, stabilizes the protein by protecting it sequences in the interior and at the dimer clefts (Elia from PEST-dependent proteolysis, and enhances phos- et al., 2003). The structure is flipped around such that phorylation of the Sak kinase domain (Yamashita et al., the residues forming the interface between the pbs in the 2001). Tec tyrosine kinases are activated downstream of Sak homodimer are largely at the exterior of the Plk1 pb many cell-surface receptors, and influence a range of heterodimer (Elia et al., 2003). This argues against signaling, including pathways that regulate cytoskeletal conservation of a binding site at the dimer interfaces. reorganization and cell motility (Lucas et al., 2003). Tec The positively charged cleft at the pb1–pb2 heterodimer and Sak kinases are both highly expressed in lymphoma interface of Plk1 binds to phosphopeptides containing S- cells where they may interact to regulate substratum- pS/pT-P/X, and thereby docks the protein to specific independent growth (Fode et al., 1994). phosphoproteins on centrioles. The pb dimer of Plk1 binds to the kinase domain in a phospho-independent manner, which blocks the catalytic cleft. Phosphopep- tide binding may alter the pb conformation to release Sak expression the kinase domain, and with the resultant T loop phosphorylation, the kinase becomes activated (Elia Expression of Sak at the mRNA and protein levels is cell et al., 2003). Attempts to identify synthetic phosphopep- cycle regulated, in a pattern similar to other plk family tides that bind to recombinant Sak-pb have thus far members (Fode et al., 1996). Sak expression in cells been unsuccessful (F Sicheri, unpublished observations). released from serum starvation is low in G , increases in 1 Sak-pb homodimerizes when co-expressed in NIH- S through G , and peaks in M phase (Fode et al., 1996). 2 3T3 cells as myc- and flag-tagged forms of the protein Similarly, after 70% hepatectomy, Sak transcripts in the (Leung et al., 2002). Both full-length GFP-Sak and regenerating liver increase incrementally from late G 1 GFP-tagged Sak-pb localize to the nucleolus in G through S, peaking at G /M (Ko et al., submitted). 2 2 phase, to centrosomes in early M, and to the cleavage Early termination of transcription of the mouse gene, furrow during telophase (Hudson et al., 2001). Similarly, and alternative splicing at exons 5and 6, results in the localization of Plk1 to mitotic structures is dependent on Sak-b transcript, which encodes only the kinase domain its two pb domains (Lee et al., 1998). The subcellular and a carboxyl end translated from 147-bp of sequence localizations of Plk1 and Sak overlap with the exception contiguous with exon 5(Fode et al., 1996). Sak-b that Plk1 is not observed in the nucleolus, while Sak was transcripts account for only 5% of the total Sak not found at the centromeres (Hudson et al., 2001). transcripts in mouse tissues, and have not been detected Sak (cry-pb), a region upstream of the Sak-pb in human cells or tissues. Human Sak cDNA has an 596–836 domain, has secondary structure motifs with similarity to insertion corresponding to an extension of exon 5, which Sak-pb. Sak cry-pb also functions as a self-association adds the Sak-b carboxy tail into the full-length Sak domain as well as binding to the Sak kinase domain,but sequence to produce an insertion of 34 amino acids not to the Sak-pb (Leung et al., 2002). GFP-Sak cry-pb relative to the mouse Sak-a protein sequence (Hudson localized to centrosomes, and current efforts are directed et al., 2000). Exons 5, 6 and 7 of human and mouse Sak at solving the structure of this domain (G Leung and F are less conserved in cDNA sequence, and appear to Sicheri, personal communication). It is possible that full- encode a peptide sequence that joins functional domains length Sak dimerizes through both the Sak cry-pb and the in the protein. Analysis of the murine Sak 50 upstream Sak-pb domains, thereby forming two unique interfaces region revealed one major transcription start site at that may function as interaction domains. Dimerization position À303 bp relative to the start of translation, a of each domain could be independently regulated to proximal region for positive regulation at À1toÀ454 allow multiple upstream inputs to Sak kinase activity and and a more distal domain effecting negative regulation localization. We have found that Sak-pb is necessary but at À455 to À991 (Hudson et al., 2000). The factors that not sufficient for Sak binding to p53, suggesting that directly control Sak transcription during the cell cycle interactions between the subdomains of Sak are impor- remain to be determined. tant for binding (Figure 1c).

Sak protein structure Sak-deﬁcient embryos The X-ray structures of Sak and Plk1 pb domains in isolation reveal six b-strands and an a-strand, but We generated a targeted mutation in ES cells that surprisingly, the organization of the strands is markedly deleted the start of translation as well as part of the Sak

Oncogene Plk4 and mitotic fidelity CJ Swallow et al 309 kinase domain (Hudson et al., 2001). SakÀ/À embryos Cdc20 activity (Bembenek and Yu, 2001). In yeast, Cdc5 lack Sak protein and arrest at BE7.5, shortly after phosphorylates Cdc14 leading to release of Cdc14 from gastrulation and prior to somite and neural tube the nucleolus, which stimulates the mitotic exit network formation. SakÀ/À embryos showed a high fraction of during late anaphase (Visintin et al., 2003). cells in mitosis, and many of the mitotic figures The requirement for Sak in the postgastrula mamma- displayed a morphology similar to that described for lian embryo may involve placement of the spindle poles Cdc5-1 mutant yeast cells, with partially segregated and cleavage furrow in relation to cortical cues, and a chromosomes (Hartwell et al., 1973; Sharon and resultant impact on tissue morphogenesis. Sak is highly Simchen, 1990). SakÀ/À blastocyst explants also arrested expressed in ciliated epithelium of the nasal cavity and with a marked accumulation of phosphorylated histone large airways of the late stage embryo, which unlike H3 and cyclin B1 positive cells, indicating an anaphase other sites of Sak expression, are not proliferating cells block. In addition, SakÀ/À blastocyst cultures displayed a (Fode et al., 1994). A common feature of ciliated and marked increase in dumbbell-shaped cells blocked in dividing cells is the motile nature of microtubules, which telophase. Apoptotic cells in the SakÀ/À embryos were radiate from centrioles and mark the polarity of cell. increased 20-fold, a phenotype also observed in E6.5 Transient overexpression of Sak in fibroblasts resulted Mad2À/À embryos, which are defective in the spindle in an enlarged, flattened morphology, and rapid loss of assembly checkpoint (Michel et al., 2001). The cellular viability after 48 h (Fode et al., 1996). phenotype in SakÀ/À embryos can be summarized as a late mitotic delay causing a high rate of apoptosis, and ultimately embryo death. Sak and cancer in mice Sak was the first mammalian plk to be mutated at the whole organism level. Recently, Plk2/Snk-deficient mice Sak mRNA and protein are present in Sak þ /À MEFs at were reported to be normal for postnatal growth, but B50% the level found in Sak þ / þ MEFs. Sak þ /À MEFs skeletal development and placental growth was slower display a phenotype of slower growth as well as an during gestation (Yarm, 2002). Plk2/Snk is an early- increased incidence of centrosome amplification and þ / þ response gene expressed at G1 in cultured cells, and abnormal spindles, compared to Sak MEFs (Ko À/À Plk2 MEFs are delayed at G1/S and grow slowly. et al., submitted). Furthermore, heterozygous Sak mice Plk2/Snk transcripts are found mainly in the hippocam- on a CD1 background developed cancer at a variety of pal neurons of adult rodent brains, a more restricted primary sites with age, most strikingly the liver and the profile than other plks. Nonetheless, the Plk2À/À MEF lung (Ko et al., submitted). The liver tumors are slow-growth phenotype suggests a role in cell cycle multifocal, high grade and display prominent mitotic regulation, which is likely more context-dependent than irregularities. An analysis of four extragenic poly- that of Sak. morphic microsatellite markers in the region immedi- SakÀ/À mouse embryos proceed through many cell ately flanking the Sak gene revealed no loss of divisions before arresting, which suggests that Sak may heterozygosity in liver tumor tissue compared to be redundant or maternally supplied at early stages of adjacent noncancerous liver in 10 mice, and Sak embryogenesis. The former explanation seems more expression was preserved at both the mRNA and likely, as Sak turns over completely with each cell cycle protein levels in the tumors. By applying the mitogenic in cultured fibroblasts. Alternatively, Sak may not be stimulus of partial hepatectomy (PH), we observed an required until morphogenesis begins in the embryo at increased incidence and reduced latency of hepato- BE6.5. Another possibility is that defective mitosis does cellular carcinogenesis in Sak þ /À mice (Ko et al., occur in early SakÀ/À embryos, but checkpoints are submitted). An examination of the regenerating liver absent and apoptosis is delayed until after gastrulation. revealed a B4 h delay in the first mitotic division and a The cell cycle and notably the mitotic exit network has marked increase in multipolar spindles at 40–44 h. certain unique features in the early embryo. For Strikingly, the sharpness of cyclin B1 accumulation example, in the Drosophila syncytium, only partial and destruction in the first cell cycle after PH was destruction of cyclin B at spindle poles and centromeres diminished in Sak þ /À mice (Figure 2a). Similarly, in occurs between mitoses, but with cellularization of the Sak þ /À MEFs, the rise and fall of cyclin B1 after release embryo, destruction of cyclin B is nearly complete from serum starvation was flattened and extended between cell cycles (Su et al., 1998). In early Drosophila (Figure 2b). Taken together, our data demonstrate that and Xenopus embryogenesis, mitotic exit depends only Sak is haploinsufficient for tumor suppression, compar- on fizzy (Cdc20), while fizzy-related (Cdh1) is not able in this regard to Mad2, BubR1 and p27Kip (Michel expressed until the larval stage of development, when et al., 2001; Philipp-Staheli et al., 2001; Dai et al., 2004). APC-Cdh1 is required for complete destruction of The Sak þ /À MEFs also display a weakened mitotic mitotic cyclins (Sigrist and Lehner, 1997; Brassac et al., checkpoint when treated with nocodazole, as well as 2000). In this regard, cell division occurs until the post- centrosome hyperamplification, similar in this regard to gastrula stage in SakÀ/À embryos, at which point cyclin p53 and p21Cip1 deficient MEFs (Tarapore and Fukasa- B1 persists and late anaphase arrest is evident, suggest- wa, 2002). Prolonged activation of cyclinE/Cdk2 also ing that APC-Cdh1 activity may be insufficient. causes centrosome amplification and multipolar spin- Mammalian Cdc14a is reported to dephosphorylate dles, thereby disrupting chromosome segregation. In Cdh1, and thereby increase APC-Cdh1 but not APC- late G1/S, Cdk2/cyclinE phosphorylates nucleophosmin,

Oncogene Plk4 and mitotic ﬁdelity CJ Swallow et al 310 a +/+ +/- PH liver Sak Sak Time (h):0 4 24 36 40 44 48

Cyclin B1 Sak+/+ 44 h γ-tubulin

Cyclin B1 Sak+/- γ-tubulin

b MEFs 7 d

Time (h) : 0 4 8 1216 20 24 28 32

Cyclin B1 Figure 3 Mitotic abnormalities and hepatocellular disorganiza- Sak+/+ tion following partial hepatectomy in Sak þ /À mice. Regenerating γ-tubulin liver from Sak þ / þ and Sak þ /À mice subjected to 2/3 partial hepatectomy (PH), stained with H&E. At 44 h (top panels), Sak þ /À hepatocytes in anaphase were larger and displayed frequent Cyclin B1 aberrant mitotic spindles. Examples of mitotic ﬁgures are indicated Sak+/- (arrows). At 7 days PH (lower panels), sinusoidal architecture and þ /À γ-tubulin hepatocyte organization was distorted in Sak mice, although liver volume was comparable to that of wild-type littermates Figure 2 A mitotic delay in Sak heterozygous MEF cells and mice. Cyclin B1 levels in (a) regenerating liver of Sak þ / þ and Sak þ /À mice elegans, Plx1 has been shown to increase the micro- killed at the indicated times after 2/3 partial hepatectomy tubule-severing activity of katanin in vitro (McNally (PH), and (b) murine embryonic ﬁbroblasts (MEFs) derived from et al., 2002). Therefore, we are investigating the Sak þ / þ and Sak þ /À mice, synchronized by serum starvation and released into serum for the indicated times. Loading control: possibility that Sak plays a role in microtubule g-tubulin remodeling and spindle placement in animal tissues.

a negative regulator of centrosome duplication, thus Sak expression in human cancer causing its release from centrosomes and stimulation of centrosome duplication (Tarapore et al., 2002). Inter- Sak expression is apparent only in proliferating cells, estingly, centrosome duplication is also stimulated by and correlates with proliferative index (Fode et al., 1994; phosphorylation of nucleophosmin at Ser-4 by Plk1 Macmillan et al., 2001). However, Sak transcripts were (Zhang et al., 2004). We are currently examining the only modestly increased in human colorectal cancer interaction between Sak and p53 as well as other tissue (T), as compared with adjacent normal mucosa effectors to determine how Sak influences multiple (NM) (T : NM ratio of 1.3170.10, mean7s.e.m., events during cell cycle progression. n ¼ 71) (Macmillan et al., 2001). Analysis of Plk1 The Sak þ /À liver architecture was poorly organized transcripts in the same specimens showed a mean compared to that of Sak þ / þ animals 7 days after PH, by T : NM ratio of 1.6270.13, consistent with reports of which time normal liver mass had been restored in both Plk1 overexpression in colorectal and other primary genotypes (Figure 3). Three-dimensional organization of cancers (Wolf et al., 1997; Macmillan et al., 2001; hepatocytes following PH may be dependent on spindle Takahashi et al., 2003). Interestingly, a small group of placement and orientation during mitosis. A certain colorectal cancer patients showed Sak transcript levels dynamic instability of microtubule asters radiating from more than 2SD below the mean, without an associated spindle poles may allow precise positioning of the reduction in Plk1 expression. The latter finding is more mitotic apparatus and the cleavage furrow. Microtu- analogous to Plk3 expression in human cancer, which is bules are oriented with their (À) end at the centrosomes reduced compared to normal tissues, particularly in lung and ( þ ) end pointing towards the cortex. The fibers and head and neck cancers (Wolf et al., 1997; Dai et al., undergo phases of shortening and growth, a dynamic 2000). We have recently identified an even higher rate of instability that allows an iterative process of pattern reduced Sak expression in human hepatoma, in which formation at the cellular level during cell movement, cell preliminary analysis also indicates frequent loss of division and morphogenesis. Plk1 is known to interact heterozygosity (E70%) close to the Sak gene locus with katanin, the microtubule-severing protein (Ko et al., submitted). This is consistent with the high (McNally et al., 2002), as well as with microtubule rates of 4q loss reported in hepatoma, and also in head stabilizing and destabilizing proteins (TCTP and Stath- and neck squamous cell carcinomas. Since Sak is min, respectively) (Budde et al., 2001; Yarm 2002). In C. haploinsufficient to suppress spontaneous hepatoma

Oncogene Plk4 and mitotic fidelity CJ Swallow et al 311 development in mice, the reduced Sak expression we Table 1 SAK expression in human colorectal cancer cell lines have observed in human tumors could contribute to Instability type Cell line SAK/PBGDa s.e.m.b carcinogenesis. Reduced Sak function could also arise through sequence changes that render the protein Chromosomal SW 480 1.033 0.033 unstable, as with Plk1 (Simizu and Osada, 2000), SW 837 1.067 0.125 COLO 320 HSR 1.008 0.074 decrease kinase activity, weaken interaction with bind- LoVo 0.817 0.047 ing partners, or cause mislocalization. Further analysis HT29 0.781 0.043 of the Sak gene in human cancer patients may reveal SW 620 0.629c 0.052 such changes. LS 1034 0.387c 0.053 Sak transcript levels in colorectal tumors relative to LS 513 0.656 0.041 paired normal mucosa were significantly higher in Microsatellite DLD-1 1.117 0.197 patients over 60 compared to less than 60 years of age HCT 116 0.978 0.097 (Macmillan et al., 2001). The difference was due to both LS 174T 0.920 0.071 a significant decrease in Sak expression in normal LS 411N 1.018 0.109 mucosa as well as an increase in expression in colorectal SW 48 1.153 0.114 tumors with age. A decline in Sak expression with age is aRatio of expression of Sak to that of the control gene porphobilinogen not unexpected in light of the decrease in other cell cycle deaminase (PBGD), as determined by semiquantitative RT–PCR, genes previously shown in fibroblast cell lines derived mean of four separate experiments. bs.e. of the mean, n ¼ 3–4 from donors of increasing age, in particular Plk1 (Ly repetitions. cOne copy of Sak on FISH analysis. et al., 2000). The resultant loss of precision in timing of cell cycle events in all probability contributes to the progression provide robust properties (Xiong and increased incidence of cancer with age. Sak expression Ferrell, 2003). These are regulatory features of the did not correlate with stage of colorectal cancer, system that enhance cell survival under different suggesting that loss of Sak is not simply a result of conditions. Robustness in cell cycle regulation also cancer progression. implies that incremental changes in the activity of one or more regulatory proteins can be tolerated, possibly causing delays, yet the cell cycle still proceeds with Sak and CIN sufficient accuracy for survival of populations. Positive and double-negative feedback loops, for example Our studies in Sak þ /À mice and MEFs indicate that Sak regulation of/by Cdc2 of/by Cdc25, Cdc2 of/by Wee1 gene dosage is critical to chromosomal stability. We and Plk1 of/by Cdc2, as described by Ferrell (2002), examined the association between Sak mRNA expres- make the kinase and APC/C networks switch-like, and sion and genomic instability in 13 human colorectal ensure discrete biochemical states and irreversible cancer cell lines previously characterized as exhibiting transitions. Many mitotic regulators appear to be CIN (n 8 lines) or microsatellite instability (MIN, ¼ haploinsufficient for tumor suppression, suggesting that n 5lines) (Table 1). Mean SAK expression was ¼ various combinations of naturally occurring hypo- significantly lower in the CIN lines (0.79770.083) than morphic alleles and age-related losses in gene expression in the MIN lines (1.03770.043) (P 0.029) (Macmillan ¼ may be a common route to cancer. Reduced activity of and Swallow, unpublished results). Furthermore, two of proteins that regulate the Cdks may reduce biochemical the three CIN lines with the lowest Sak expression cooperativity, leading to inappropriate Cdk activation demonstrated only one copy of Sak on FISH analysis and/or destruction, and thereby increase the probability (Table 1). In contrast to Sak, there was no significant of mitotic errors. For example, the failure in Sak þ /À difference in Plk1 expression between the CIN and MIN mice of cyclinE/Cdk2 and/or cyclin B1/Cdk1 to attain lines (1.44670.140 versus 1.62870.211) (P 0.49). We ¼ the sharp peak levels required for proper centrosome pursued this further by classifying the colorectal cancers duplication and/or chromosome segregation may lead to we had previously tested as either MIN or CIN, based heritable errors. on microsatellite instability at the BAT-26 locus. Our results suggest that tissues in the embryo and Mirroring the cell line data, Sak expression was lower adult Sak þ /À mice may tolerate a certain amount of in the CIN than the MIN tumors (P 0.06), while Plk1 ¼ variance in mitotic regulation, but the lifetime prob- was the same between the two groups (P 0.69). ¼ ability that a rare mitotic error becomes a significant Although retrospective studies of human cancer patients event causing cancer increases exponentially with age, are correlative at best, the cancer phenotype in Sak þ /À particularly when coupled with the highly selective mice strongly supports the concept that insufficient Sak, cellular processes that enhance growth autonomy in in the absence of other abnormalities, can cause CIN. cancers. Therefore, incremental differences in biochemical activities causing small increases in the probability Future directions of mitotic error can in theory markedly decrease the thresholds for pathology when magnified by clonal The macroevents marking the cell cycle, such as expansion and selection of cells that gain growth equitable segregation of chromosomes, are discrete, autonomy. To understand the contribution of mitotic and errors can become heritable if the cells survive. The regulators to cancer development, we require detailed highly cooperative feedback controls on cell cycle models of the molecular circuitry, and their effects either

Oncogene Plk4 and mitotic ﬁdelity CJ Swallow et al 312 direct or indirect on the binary outcomes of the cell Acknowledgements cycle. This should lead to new therapies designed to This research was supported by a grant from the National induce proapoptotic function or growth arrest in tumor Cancer Institute of Canada to JWD and from the National cells, customized for the altered molecular circuitry of Colorectal Cancer Campaign to CJS. We thank Jennifer C each cancer patient. Macmillan and Carla Rosario for technical contributions.

References

Bai F, Pei XH, Godfrey VL and Xiong Y. (2003). Mol. Cell Lucas JA, Miller AT, Atherly LO and Berg LJ. (2003). Biol., 23, 1269–1277. Immunol. Rev., 191, 119–138. Barr FA, Sillje HH and Nigg EA. (2004). Nat. Rev. Mol. Cell Ly DH, Lockhart DJ, Lerner RA and Schultz PG. (2000). Biol., 5, 429–440. Science, 287, 2486–2492. Bell DW, Varley JM, Szydlo TE, Kang DH, Wahrer DC, Macmillan JC, Hudson JW, Bull S, Dennis JW and Swallow Shannon KE, Lubratovich M, Verselis SJ, Isselbacher KJ, CJ. (2001). Ann. Surg. Oncol., 8, 729–740. Fraumeni JF, Birch JM, Li FP, Garber JE and Haber DA. Malkin D, Li FP, Strong LC, Fraumeni Jr JF, Nelson CE, (1999). Science, 286, 2433–2434. Kim DH, Kassel J, Gryka MA, Bischoff FZ and Tainsky Bembenek J and Yu H. (2001). J. Biol. Chem., 276, MA. (1990). Science, 250, 1233–1238. 48237–48242. McNally KP, Buster D and McNally FJ. (2002). Cell Motil. Blagden SP and Glover DM. (2003). Nat. Cell Biol., 5, 505–511. Cytoskeleton, 53, 337–349. Brassac T, Castro A, Lorca T, Le Peuch C, Doree M, Labbe J- Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, C and Galas S. (2000). Oncogene, 19, 3782–3790. Gerald W, Dobles M, Sorger PK, Murty VVVS and Benezra Budde PP, Kumagai A, Dunphy WG and Heald R. (2001). R. (2001). Nature, 409, 355–359. J. Cell Biol., 153, 149–158. Muraoka RS, Lenferink AE, Law B, Hamilton E, Brantley Dai W, Li Y, Ouyang B, Pan H, Reissmann P, Li J, Wiest J, DM, Roebuck LR and Arteaga CL. (2002). Mol. Cell. Biol., Stambrook P, Gluckman JL, Noffsinger A and Bejarano P. 22, 2204–2219. (2000). Genes Chromosomes Cancer, 27, 332–336. Ouyang B, Pan H, Lu L, Li J, Stambrook P, Li B and Dai W. Dai W, Wang Q, Liu T, Swamy M, Fang Y, Xie S, Mahmood (1997). J. Biol. Chem, 272, 28646–28651. R, Yang YM, Xu M and Rao CV. (2004). Cancer Res., 64, Philipp-Staheli J, Payne SR and Kemp CJ. (2001). Exp. Cell 440–445. Res., 264, 148–168. Elia AE, Rellos P, Haire LF, Chao JW, Ivins FJ, Hoepker K, Sharon G and Simchen G. (1990). Genetics, 125, 475–485. Mohammad D, Cantley LC, Smerdon SJ and Yaffe MB. Sherr CJ. (2004). Cell, 116, 235–246. (2003). Cell, 115, 83–95. Shirayama M, Zachariae W, Ciosk R and Nasmyth K. (1998). Elledge SJ. (1996). Science, 274, 1664–1672. EMBO J., 17, 1336–1349. Fernebro E, Halvarsson B, Baldetorp B and Nilbert M. (2002). Sigrist SJ and Lehner CF. (1997). Cell, 90, 671–681. BMC Cancer, 2, 25. Simizu S and Osada H. (2000). Nat. Cell Biol., 2, 852–854. Ferrell Jr JE. (2002). Curr. Opin. Cell Biol., 14, 140–148. Su TT, Sprenger F, DiGregorio PJ, Campbell SD and Fode C, Binkert C and Dennis JW. (1996). Mol. Cell. Biol., 16, O’Farrell PH. (1998). Genes Dev., 12, 1495–1503. 4665–4672. Takahashi T, Sano B, Nagata T, Kato H, Sugiyama Y, Fode C, Motro B, Youseﬁ S, Heffernan M and Dennis JW. Kunieda K, Kimura M, Okano Y and Saji S. (2003). Cancer (1994). Proc. Natl. Acad. Sci.USA, 91, 6388–6392. Sci., 94, 148–152. Hammond C, Jeffers L, Carr BI and Simon D. (1999). Tarapore P and Fukasawa K. (2002). Oncogene, 21, 6234– Hepatology, 29, 1479–1485. 6240. Hartwell LH, Mortimer RK, Culotti J and Culotti M. (1973). Tarapore P, Okuda M and Fukasawa K. (2002). Cell Cycle, 1, Genetics, 74, 267–286. 75–81. Hudson JW, Chen L, Fode C, Binkert C and Dennis JW. Tsai KY, MacPherson D, Rubinson DA, Nikitin AY, Bronson (2000). Gene, 241, 65–73. R, Mercer KL, Crowley D and Jacks T. (2002). Proc. Natl. Hudson JW, Kozarova A, Cheung P, Macmillan JC, Swallow Acad. Sci. USA, 99, 16865–16870. CJ, Cross JC and Dennis JW. (2001). Curr. Biol., 11, 441–446. Visintin R, Stegmeier F and Amon A. (2003). Mol. Biol. Cell, Initiative TA. (2000). Nature, 408, 796–815. 14, 4486–4498. Jackson RJ, Engelman RW, Coppola D, Cantor AB, Wharton Wolf G, Elez R, Doermer A, Holtrich U, Ackermann H, W and Pledger WJ. (2003). Cancer Res., 63, 3021–3025. Stutte HJ, Altmannsberger H-M, Rubsamen-Waigmann H Knudson Jr AG. (1971). Proc.Natl.Acad.Sci.USA, 68, 820–823. and Strebhardt K. (1997). Oncogene, 14, 543–549. Lee KS, Grenfell TZ, Yarm FR and Erikson RL. (1998). Proc. Xiong W and Ferrell Jr JE. (2003). Nature, 426, 460–465. Natl. Acad. Sci. USA, 95, 9301–9306. Yamashita Y, Kajigaya S, Yoshida K, Ueno S, Ota J, Ohmine Leung GC, Hudson JW, Kozarova A, Davidson A, Dennis JW K, Ueda M, Miyazato A, Ohya K, Kitamura T, Ozawa K and Sicheri F. (2002). Nat. Struct. Biol., 9, 719–724. and Mano H. (2001). J. Biol. Chem., 276, 39012–39020. Lindon C and Pines J. (2004). J. Cell Biol., 164, 233–241. Yarm FR. (2002). Mol. Cell. Biol., 22, 6209–6221. Lu T, Pan Y, Kao SY, Li C, Kohane I, Chan J and Yankner Zhang H, Shi X, Paddon H, Hampong M, Dai W and Pelech BA. (2004). Nature, 429, 883–891. S. (2004). J. Biol. Chem, 279, 35726–35734.

Oncogene