¹Department of Biological Sciences, and the Walther Cancer Research Institute, University of Notre Dame, Notre Dame, Indiana, IN 46556, USA

²Department of Cell Biology and the Cancer Center, University of Massachusetts Medical School, Worcester, Massachusetts, MA 01605, USA

Correspondence to: G Sluder, University of Massachusetts Medical School, Department of Cell Biology, 377 Plantation Street, Worcester, MA 01605, USA;E-mail: Greenfield.sluder@umassmed.edu

Oncogene (2002) 21, 6154-6160. doi:10.1038/sj.onc.1205826

cancer; Cdk2; centrosome; cyclin; mitosis

Introduction

The magic number in cell division is two: during S phase the cell makes two copies of its genome, and during mitosis the equal segregation of all sister chromosomes to the two daughter cells critically depends upon the formation of a bipolar mitotic spindle. Since spindle polarity in higher animal cells is usually dependent upon the number of centrosomes present, the cell must have exactly two centrosomes by the onset of mitosis or else mitotic defects will occur that can have disastrous consequences for the organism (reviewed in Brinkley, 2001; Rieder et al., 2001). For example, failure of the interphase centrosome to duplicate before mitosis leads to a monopolar spindle and the formation of a single polyploid daughter cell. Conversely, if a cell contains more than the normal two centrosomes, it is apt to assemble a multipolar spindle at mitosis, which randomly distributes chromosomes to multiple daughter cells. Finally, a complete lack of a centrosome does not prevent the formation of a bipolar spindle (Khodjakov et al., 2000), but does lead to an increased incidence of cleavage failure (Hinchcliffe et al., 2001; Khodjakov and Rieder, 2001; Piel et al., 2001). Thus, mis-regulation of centrosome number produces genetic imbalances that can contribute to the loss of growth regulation and the genesis of the transformed phenotype. Indeed, the multiple centrosomes found in the cells of many high-grade human tumors are thought to cause the genomic instability that allows the evolution of aggressive growth characteristics (Lingle and Salisbury, 2000). Thus, it is of obvious importance for the cell to exercise tight control over centrosome duplication. In this review we discuss the development of our understanding of how Cdk2 - cyclin E participates in the control of centrosome duplication.

Basics of centrosome duplication

At the end of mitosis each daughter cell inherits a single centrosome and by the start of the next mitosis each contains just two centrosomes that will establish the bipolar spindle. This precise doubling of the interphase centrosome is known as centrosome duplication or centrosome reproduction (see Hinchcliffe and Sluder, 2001; Stearns, 2001). In mammalian cells this process is thought to begin in late G₁ with the loss of the orthogonal relationship between the mother and daughter centrioles (Kuriyama and Borisy, 1981; however, see Mack and Rattner, 1993; Piel et al., 2000, 2001 for examples of centriole separation earlier in the cell cycle). Centriole duplication starts at the beginning of S phase or during S phase with the appearance of short daughter centrioles, or procentrioles, at right angles to the two original centrioles. These procentrioles elongate during S and G₂, reaching mature length in mitosis or the following G₁ (Kuriyama and Borisy, 1981; Lange et al., 2000). The completion of centrosome reproduction occurs with daughter centrosome disjunction and separation at a variable time in G₂, with pairs of mother-daughter centrioles in each centrosome (Aubin et al., 1980; Kochanski and Borisy, 1990; Sharp et al., 2000).

The cell must not only control centrosome copy number but also coordinate the events of centrosome reproduction with nuclear events in the cell cycle. Control of centrosome duplication is exercised by limits that are intrinsic to the centrosome itself and by extrinsic controls imposed by changing cytoplasmic conditions during cell cycle progression. Limits intrinsic to the centrosome determine the number of daughter centrosomes that arise from the parent centrosome at each round of duplication (reviewed in Sluder and Hinchcliffe, 1999). This is based in the cycle of centriole splitting and centriole duplication (Sluder and Rieder, 1985). Extrinsic cytoplasmic controls determine when the centrosome duplicates in relation to the progression of nuclear events such as DNA synthesis and mitosis. Evidence for cytoplasmic regulation of centrosomal events came from studies on zygotes demonstrating that repeated centrosome duplication occurs in enucleated cells and cells in which protein synthesis is completely blocked (Sluder et al., 1986, 1990; Gard et al., 1990). These studies indicated that centrosome duplication is not dependent upon the completion of nuclear events such as DNA synthesis per se and can take place in an orderly fashion from pre-existing subunit pools that are not limiting. Given that centrosome reproduction is driven by cytoplasmic mechanisms, it was logical to investigate whether centrosome duplication is regulated by the activities of the cyclin dependent kinases (Cdk) that control cell cycle progression. This was an attractive possibility because it provided a logical way for the cell to coordinate centrosome duplication with nuclear events in the cell cycle.

CDK activity and centrosome reproduction

The involvement Cdk1-cyclin B (also known as p34^cdc2-cyclin B or MPF: maturation promoting factor) in centrosome duplication was first examined. The synthesis of cyclin B starting in S phase and the precipitous rise in the activity of this kinase complex in late G2 was known to drive the cell cycle into mitosis (Murray and Kirschner, 1989). Findings that complete inhibition of protein synthesis does not stop repeated centrosome duplication in zygotes provided compelling evidence that the cycle of Cdk1-B activity does not control if and when centrosomes reproduce (Sluder et al., 1990; Gard et al., 1990). Nevertheless, the possibility still remained that the absolute value of Cdk1-B activity might provide conditions that allowed centrosome duplication in early interphase and later, as cyclin B was synthesized, blocked it from late S phase through the end of mitosis. However, a direct test of this possibility, again using sea urchin zygotes, revealed that centrosome reproduction is not dependent on the absolute value of Cdk1-B activity but rather dependent upon cell cycle stage (Hinchcliffe et al., 1998). When these zygotes are arrested in S phase with aphidicolin, a specific inhibitor of the alpha DNA polymerase, Cdk1-B activity rises to supra-mitotic levels yet centrosomes repeatedly double. Complete inhibition of protein synthesis in the first cell cycle arrests the cycle in S phase with low pre-fertilization Cdk1-B activity yet the centrosomes continue to reproduce. However, when the zygotes are arrested in mitosis by the expression of 90 cyclin B, a non-degradable form of this cyclin (Glotzer et al., 1991), centrosomes split into half centrosomes but fail to fully reproduce under conditions of high Cdk1-B activity. This work demonstrates that centrosomes can repeatedly reproduce during S phase regardless of whether Cdk1-B activity is high or low and that mitosis is not a phase of the cell cycle that supports the complete duplication of the centrosome.

The involvement of CDK2-cyclin E in centrosomeduplication

Attention then turned to Cdk2-E in large part due to the temporal correlation between the rise in its activity and the first morphological events of centrosome duplication. Cdk2-E activity rises shortly before the onset of S phase, which is the time when centrioles duplicate as seen by the appearance of procentrioles. Also, since cell cycle progression into S phase depends upon the activation of Cdk2-E (Dulic et al., 1992), this kinase complex could in principle provide the needed coordination between centrosome and DNA replication. In the late 1990s a number of laboratories independently started to investigate this possible association of Cdk2-E activation with centrosome reproduction, and a series of papers appeared in rapid succession. To simplify our discussion we will categorize these studies by experimental system.

Early embryonic cells and extracts

Hinchcliffe et al. (1999) developed an S phase-arrested Xenopus egg extract that supports multiple rounds of sperm centrosome duplication in vitro. The observation that the asters organized by the sperm centrosomes routinely doubled three and sometimes four times demonstrated that centrosomes were completely reproducing not just splitting into half centrosomes, each with a single centriole. Cdk2-E activity was then specifically inhibited in these extracts by the addition of recombinant 34Xic-1, a NH₃-terminal truncated form of the Xenopus cyclin dependent kinase inhibitor Xic-1^p27. At the concentration used 34Xic-1 specifically inhibits the activity of Cdk2-E but not Cdk1-A or Cdk1-B (Su et al., 1995). Cdk2-cyclin A activity was not a factor in these experiments, because Cdk2 does not complex with cyclin A until after the mid-blastula transition in Xenopus (Rempel et al., 1995). Inhibiting Cdk2-E activity prevented the repeated duplication of the centrosome in these extracts (Figure 1). When a 1.4 molar excess of purified baculovirus-expressed Cdk2-E was added to the 34Xic-1 treated extracts, multiple rounds of aster duplication were restored (Figure 1). Interestingly, centrosomes doubled once when Cdk2-E activity was blocked with 34Xic-1, but it is not known if this represents a complete duplication or the splitting and separation of the two sperm centrioles each of which organizes an aster (see Sluder and Rieder, 1985).

At about the same Lacey et al. (1999) published a different Xenopus egg extract assay system in which isolated mammalian centrosomes were used to examine centriole disjunction as a function of Cdk2-E activity. Mother-daughter centriole splitting (also known as disorientation) was used as a measure of centrosome duplication because this is reported to be the leading morphological event in centrosome reproduction (Kuriyama and Borisy, 1981). In control extracts, with normal levels of Cdk2-E activity, they found that mother-daughter centriole pairs would disjoin. Extracts containing the cyclin dependent kinase inhibitors (CKIs) p21 or p27 (Jackson et al., 1995) to block Cdk2-E activity did not show centriole disjunction. To confirm these results, they used Xenopus embryos arrested in interphase with protein synthesis inhibitors, which allow repeated centrosome duplication without cell cycle progression (Gard et al., 1990). When individual blastomeres were microinjected with the CKIs p21 or p27 to inhibit the activity of Cdk2-E, the centrosomes did not repeatedly duplicate, as did those in the uninjected cells of the same zygote. Together, these in vitro and in vivo studies indicate that Cdk2-E activity is needed for centrosome duplication in early embryo model systems.

Mammalian somatic cells

The importance of Cdk2-E in the initiation of centrosome duplication is not restricted to early zygotes; Cdk2 complexed with cyclins E and A appear to similarly link centrosome duplication with S phase in mammalian somatic cells. An early indication that Cdk2 activity might play a regulatory role in the control of centrosome duplication in mammalian cells came from an examination of the effects of inhibiting p21^cip1/waf-1 in human hematopoetic cells (Mantel et al., 1999). p21^cip1/waf-1 is an inhibitor of Cdk2 activity that normally participates in the regulation of the G₁/S transition. These workers found that inhibition of p21^cip1/waf-1 activity increased Cdk2 activity and allowed the cells to accumulate multiple centrosomes. Whether Cdk2 was coupled to cyclin A or E was not examined.

An important study (Balczon et al., 1995) demonstrated that when Chinese hamster ovary (CHO) cells are arrested in S phase for prolonged periods of time with hydroxyurea, the centrosome duplicates multiple times (Figure 2). Matsumoto et al. (1999) used this as an experimental system to demonstrate that centrosome re-duplication is inhibited when the activity of Cdk2 is blocked by drugs or the over-expression of p21^cip1/waf-1. This study, however, did not differentiate between the relative roles of Cdk2-E and Cdk2-A in the initiation of centrosome reproduction.

A subsequent study, also using CHO cells, provided evidence that Cdk2 coupled with cyclin A is significantly more effective than Cdk2 coupled with cyclin E in restoring centrosome re-duplication when cells are arrested at the G1/S boundary by the expression of a mutant Rb construct that lacks phosphorylation sites (Meraldi et al., 1999). These findings raise the possibility that embryos and somatic cells use different Cdk2 cyclin combinations to initiate centrosome duplication. However, the possibility that centrosome reproduction in somatic cells is to some extent subject to Cdk2-cyclin E activity is indicated by reports that nucleophosmin NO38/B23 phosphorylation by Cdk2-E is directly involved in centrosome duplication (Okuda et al., 2000). Also, Cdk2-E activity stabilizes the levels of mMps1p, a kinase involved in centrosome reproduction (Fisk and Winey, 2001). Despite uncertainty over the relative roles of Cdk2-E and Cdk2-A, the important theme that has emerged is that the cell's entry into S phase and centrosome duplication are linked through a rise in Cdk2 activity.

Yeast

In budding yeast, the function of the centrosome is performed by an organelle called the spindle pole body (SPB). Like the centrosome, the SPB exists as a single copy during interphase, which must reproduce exactly once prior to the onset of mitosis (SPB reproduction cycle reviewed in Adams and Kilmartin, 2000). It appears that the SPB requires Cdk activity to both drive its duplication cycle, and to prevent re-duplication without coordinate cell cycle progression (Haase et al., 2001).

To uncover the role played by Cdks in the duplication cycle of the SPB, Haase et al. (2001) used budding yeast deleted for various combinations of cyclins. To monitor SPB number, they used yeast expressing the SPB component Spc42 coupled to green fluorescent protein (Spc42 - GFP). S. cerevisiae have a single Cdk called p34^cdc28, which is coupled to one of nine cyclins to give p34^cdc28 its cell cycle stage/substrate specificity (Andrews and Measday, 1998). These cyclins are classified as Clb 1-4 (mitotic B-type cyclins); Clb 5 and 6 (S phase B-type cyclins); and Cln 1, 2 and 3 (the so-called G₁ cyclins). When they deleted the all B-type cyclins (Clb 1-6), they found that the SPB could reproduce; confirming the work from urchin that centrosome reproduction does not require cyclin B mediated Cdk activity (Sluder et al., 1990; Hinchcliffe et al., 1998). However, unlike sea urchin centrosomes, yeast SPBs cannot duplicate again in the complete absence of the Clbs. When the two S phase Clbs (5 and 6) were restored, a small but significant portion of SPBs could undergo re-duplication to give the multi-SPB phenotype. This finding is also consistent with findings of centrosome reduplication in zygotes and mammalian somatic cells arrested in S phase (see Balczon et al., 1995; Hinchcliffe and Sluder, 1998). The authors reasoned that Cln 1, 2 and 3 drive SPB duplication in budding yeast. However, loss of Cln 1-3 leads to cell cycle arrest, so the extent to which these kinase complexes play a direct role in the SPB duplication cycle cannot be addressed.

Haase et al. (2001) also investigated how SPB re-duplication is normally prevented. To test their hypothesis that the mitotic cyclins function to prevent re-duplication of the SPB, they arrested yeast in S phase with hydroxyurea. In cells not expressing Clb1, the SPBs underwent repeated duplication. In contrast, S phase arrested cells that expressed Clb1 failed to re-duplicate their spindle pole bodies. This indicates that in yeast, the activity of mitotic cyclin dependent kinases prevents the repeated duplication of the spindle pole body. Interestingly, this control strategy is not used by sea urchin zygotes whose centrosome undergo repeated rounds of duplication in the presence of supra-mitotic levels of Cdk1-B activity (Hinchcliffe et al., 1998).

The targets of CDK2 activity

Once the importance of Cdk2 activity in centrosome reproduction was established, research focused on identifying targets of Cdk2 activity that are involved in the centrosome duplication cycle. A priori, Cdk2 could directly phosphorylate centrosomal components and/or it could modulate pathways that in turn influence centrosome reproduction. The results of two recent studies suggest that Cdk2-E operates in both fashions.

To identify direct centrosomal targets of Cdk2 activity Okuda et al. (2000) used an in vitro kinase assay using unreplicated centrosomes isolated from G₀ cells as a substrate. They found that Cdk2-cyclin E phosphorylated only a single centrosome polypeptide. This polypeptide was identified as NO38/B23 (also known as nucleophosmin), a previously identified component of the nucleolus (Schmidt-Zachman et al., 1987). Immunofluorescence analysis of Swiss 3T3 cells revealed that nucleophosmin appeared on centrosomes during mitosis and remained localized to the unduplicated interphase centrosome during early interphase. However, once the centrosome duplicated its nucleophosmin immunoreactivity was lost. Functional evidence that nucleophosmin is involved in centrosome reproduction was provided by the finding that the expression of a non-phosphorylatable mutant nucleophosmin blocked centrosome reproduction. Electron microscopy of such cells revealed that the centrioles had not separated or assembled procentrioles. In addition, microinjection of antibodies to rat nucleophosmin also blocked the duplication of centrosomes. Together these results indicate that nucleophosmin associates with the centrosomes during mitosis thereby preventing their splitting and duplication until late G₁, when Cdk2-E activity rises in preparation for S phase. Phosphorylation of nucleophosmin causes it to come off of the centrosome thereby allowing centriole disorientation and splitting to occur which could be a prerequisite for the assembly of procentrioles and the completion of centrosome reproduction.

Going forward, it will be important to investigate if Cdk2-cyclin A also phosphorylates nucleophosmin at sites that cause it to dissociate from the centrosome, as would be predicted by the results of Meraldi et al. (1999). If so, it will be interesting to know whether Cdk2-A is indeed more active than Cdk2-E in phosphorylating nucleophosmin in mammalian cells. Also, it will be of interest to determine why centrioles can split apart during prolonged mitosis (Sluder and Begg, 1985; Gallant and Nigg, 1992), at point in the cell cycle when nucleophosmin should be associated with the centrosome. Perhaps the Cdk2-E phosphorylation of nulceophosmin does not control the splitting of the mother-daughter centrioles but rather limits the assembly of daughter centrioles. For a more detailed description of nucleophosmin and centrosome reproduction, see the article by Okuda in the present issue of Oncogene.

Cdk2-E also indirectly participates in the control of centrosome duplication through the stabilization of mMps-1p kinase levels during S phase when the centrosome duplicates (Fisk and Winey, 2001). This kinase (also known as mouse esk protein kinase) is the mouse ortholog of the yeast kinase Mps-1p, originally identified as essential for the duplication of the spindle pole body (Winey et al., 1991). Fisk and Winey found that mMps-1p localized to centrosomes throughout the cell cycle, by both immunofluorescence and in vivo characterizations of mouse cell lines stably expressing mMps-1p-GFP. Functional evidence that mMps-1p kinase activity is required for centrosome duplication comes from the finding that a kinase-dead mMps-1p localizes to the centrosomes yet prevents their duplication. Conversely, when mMps-1p is over-expressed during S phase arrest, correlative light and serial section electron microscopy showed that the centrosomes/centrioles reduplicated in a cell line that does not show centrosome reduplication when arrested in S phase. That is, extra mMps-1p drives centrosome reduplication when it otherwise would not occur. In addition, these workers found that Cdk2-E activity stabilizes mMps-1 protein levels. When Cdk2-E activity is blocked by drug treatments or by over-expression of p21 or p27, the cellular level of mMps-1p dramatically drops and its localization to the centrosome is lost. However, the importance of mMps-1p stabilization by Cdk2-E in centrosome reproduction has been recently brought into question (Stucke et al., 2002). siRNA mediated reduction of mMps-1p protein levels led to the apparent loss of this protein from the centrosomes yet their duplication was not compromised. Whether this means that mMps-1p is not involved in centrosome duplication or alternatively that small amounts of this kinase are sufficient to facilitate centrosome reproduction remains to be determined (for a more comprehensive review of Mps-1 see article by Winey and Huneycutt in this issue of Oncogene).

Blocking centrosome re-replication

Once the centrosome has reproduced in early S phase, it is of obvious importance for the cell to prevent the daughter centrosomes from reduplicating before the cell reaches mitosis. Indeed, the disregulation of centrosome reproduction, called centrosome amplification, has been said to be an early event in the neoplastic transformation of cells (Lingle and Salisbury, 2000; Brinkley, 2001). Thus far, all evidence points to Cdk2-E playing an important role in providing permissive conditions for centrosome reproduction by the phosphorylation of nucleophosmin, the stabilization of Mps-1, and possibly other targets yet to be discovered. The observation that many aggressive human tumors show elevated levels of cyclin E (reviewed in Keyomarsi and Herliczek, 1997; Mussman et al., 2000) raises the important question of whether centrosome reduplication is normally prevented by the cell passing through S phase before reduplication can occur, or by additional mechanisms that specifically prevent the daughter centrosomes from reproducing again regardless of Cdk2-E activity. Several recent studies provide evidence that there are in fact one or more mechanisms that block the reduplication of the centrosomes. An early indication for the existence of a re-replication block to centrosome re-duplication came from a study on sea urchin zygotes (Hinchcliffe et al., 1998). These workers found that the centrosomes in zygotes arrested just after first mitosis in G₁ of the second cell cycle would duplicate completely but only once. However, if the zygotes were arrested in S phase of that second cycle, the centrosomes would reduplicate. These observations led to the proposal that centrosome duplication is a two-step process involving the completion of the morphological events of reproduction followed by a 'licensing' event during S phase that prepares the daughter centrosomes for duplication during the next cell cycle (Hinchcliffe et al., 1998). The nature of this licensing event and the role that Cdk2-E might play in it are not known. At a minimum, this putative licensing event must occur downstream of nucleophosmin phosphorylation by Cdk2-E.

Direct tests of the role of elevated Cdk2-E activity in centrosome reduplication revealed that this leads to the precocious duplication of the centrosome before S phase but does not lead to a significant incidence of centrosome reduplication despite continuing high levels of Cdk2-E activity (Spruck et al., 1999; Mussman et al., 2000). These findings are important because they directly point to the existence of a mechanism, distinct from Cdk2-E, that limits centrosome reproduction to just once per cell cycle. Although the identity of this mechanism is not presently known, evidence is accumulating that the p53 pathway plays an important role in blocking centrosome reduplication (Carroll et al., 1999; Tarapore et al., 2001a,b; reviewed by Fukasawa et al. in this issue of Oncogene). However, the p53 pathway may not be the only block to centrosome reduplication. Recent reports have provided evidence for the importance of proteosome mediated proteolysis in blocking centrosome reduplication. Nakayama et al. (2000) discovered that 38% of mouse fibroblasts deficient for Skp2 (an F-box adapter protein that targets specific proteins for degradation by ubiquitin-mediated proteolysis) contain supernumerary centrosomes. The finding that these cells did not have elevated Cdk2-E activity (also see Spruck et al., 1999; Mussman et al., 2000) raises the intriguing possibility that there are proteins outside the Cdk2 pathway that need to be degraded as soon as the centrosome reproduces. Consistent with this notion, Wojcik et al. (2000) found a high incidence of extra centrosomes in Drosophila neuroblasts from embryos with mutant slimb (supernumerary limbs). Slimb is also an F-box protein involved in ubiqutin-proteolysis (see Hinchcliffe and Sluder, 2001).

Another level of complexity in the mechanisms that serve to initiate or limit centrosome duplication comes from the finding that activation of a Calcium/calmodulin-dependent kinase II pathway is needed for the initiation of centrosome reproduction in Xenopus egg extracts (Matsumoto and Maller, 2002). It remains to be explored how this Ca²⁺ pathway interacts with Cdk2 activity and if this could provide a limit to centrosome reduplication.

Conclusions

It has become clear that a rise in Cdk2 activity initiates both centrosome duplication and DNA synthesis. In this way the essential temporal coordination between two key preparatory events for mitosis is assured. Equally clear should be an appreciation that the role of Cdk2 activity in centrosome reproduction is but one piece of a complicated puzzle. This complexity is understandable, because the penalty for errors in controlling centrosome number is great - as evidenced by the genomic instability observed in cancer cells with multiple centrosomes. The tight control of centrosome number during mitosis is also important because cells do not appear to have a checkpoint that monitors mitotic spindle polarity; consequently mistakes are not corrected (Sluder et al., 1997). Finally, problems stemming from defective control of centrosome duplication are not limited to mitosis; inappropriate centrosome number can cause problems in cytoplasmic organization during interphase.

Acknowledgements

We thank Mark Winey, Michel Bornens, Tim Stearns, Peter Jackson, Jim Maller, Conly Rieder and Alexey Khodjakov for their continued conversations on cell cycle regulation of centrosome reproduction and function. Thanks also to Kenji Fukasawa for editorial input and patience. Work from the Sluder Laboratory is funded by a grant from the National Institutes of Health (GM 30758). EH Hinchcliffe is supported by funds from the Walther Institute for Cancer Research, and the University of Notre Dame.

Adams IR, Kilmartin JV. (2000). Trends Cell Biol., 10: 329-335. Article MEDLINE

Andrews B, Measday V. (1998). Trends Genet., 14: 66-72. Article MEDLINE

Aubin JE, Osborn M, Weber K. (1980). J. Cell Sci., 43: 177-194. MEDLINE

Balczon R, Bao L, Zimmer WE, Brown K, Zinkowski RP, Brinkley BR. (1995). J. Cell Biol., 130: 105-115. MEDLINE

Brinkley BR. (2001). Trends Cell Biol., 11: 18-21. MEDLINE

Carroll PE, Okuda M, Horn HF, Biddinger P, Stambrook PJ, Gleich LL, Li YQ, Tarapore P, Fukasawa K. (1999). Oncogene, 18: 1935-1944. MEDLINE

Dulic V, Lees E, Reed SI. (1992). Science, 257: 1958-1961. MEDLINE

Fisk HA, Winey M. (2001). Cell, 106: 95-104. MEDLINE

Gard DL, Hafezi S, Zhang T, Doxsey SJ. (1990). J. Cell Biol., 110: 2033-2042. MEDLINE

Gallant P, Nigg EA. (1992). J. Cell Biol., 117: 213-224. MEDLINE

Glotzer M, Murray AW, Kirschner MW. (1991). Nature, 349: 132-138. MEDLINE

Haase SB, Winey M, Reed SI. (2001). Nat. Cell. Biol., 3: 38-42. Article MEDLINE

Hinchcliffe EH, Cassels GO, Rieder CL, Sluder G. (1998). J. Cell Biol., 140: 1417-1426. MEDLINE

Hinchcliffe EH, Sluder G. (1998). Cell Biol. Int., 22: 3-5. MEDLINE

Hinchcliffe EH, Sluder G. (2001). Genes Dev., 15: 1167-1181. Article MEDLINE

Hinchcliffe EH, Li C, Thompson EA, Maller JL, Sluder G. (1999). Science, 283: 851-854. Article MEDLINE

Hinchcliffe EH, Miller FJ, Cham M, Khodjakov A, Sluder G. (2001). Science, 291: 1547-1550. Article MEDLINE

Jackson PK, Chevalier S, Philippe M, Kirschner MW. (1995). J. Cell Biol., 130: 755-769. MEDLINE

Keyomarsi K, Herliczek TW. (1997). Prog. Cell Cycle Res., 3: 171-191. MEDLINE

Khodjakov A, Cole RW, Oakley BR, Rieder CL. (2000). Curr. Biol., 10: 59-67. Article MEDLINE

Khodjakov A, Rieder CL. (2001). J. Cell Biol., 153: 247-252.

Kochanski RS, Borisy GG. (1990). J. Cell Biol., 110: 1599-1605. MEDLINE

Kuriyama R, Borisy GG. (1981). J. Cell Biol., 91: 814-821. MEDLINE

Lacey KR, Jackson PK, Stearns T. (1999). Proc. Natl Acad. Sci. USA, 96: 2817-2822. Article MEDLINE

Lange BM, Faragher AJ, March P, Gull K. (2000). Curr. Top. Dev. Biol., 49: 235-249. MEDLINE

Lingle WL, Salisbury JL. (2000). Curr. Top. Dev. Biol., 49: 313-329. MEDLINE

Mack G, Rattner JB. (1993). Cell Motil. Cytoskeleton, 26: 239-247. MEDLINE

Mantel C, Braun SE, Reid S, Henegariu O, Liu L, Hangoc G, Broxmeyer HE. (1999). Blood, 93: 1390-1398. MEDLINE

Matsumoto Y, Hayashi K, Nishida E. (1999). Curr. Biol., 9: 429-432. Article MEDLINE

Matsumoto Y, Maller JL. (2002). Science, 295: 499-502. MEDLINE

Meraldi P, Lukas J, Fry AM, Bartek J, Nigg EA. (1999). Nat. Cell Biol., 1: 88-93. Article MEDLINE

Murray AW, Kirschner MW. (1989). Nature, 339: 275-280. MEDLINE

Mussman JG, Horn HF, Carroll PE, Okuda M, Tarapore P, Donehower LA, Fukasawa K. (2000). Oncogene, 19: 1635-1646. MEDLINE

Nakayama K, Nagahama H, Minamishima YA, Matsumoto M, Nakamichi I, Kitagawa K, Shirane M, Tsunematsu R, Tsukiyama T, Ishida N, Kitagawa M, Nakayama K, Hatakeyama S. (2000). EMBO J., 19: 2069-2081. Article MEDLINE

Okuda M, Horn HF, Tarapore P, Tokuyama Y, Smulian AG, Chan PK, Knudsen ES, Hofmann IA, Snyder JD, Bove KE, Fukasawa K. (2000). Cell, 103: 127-140. MEDLINE

Piel M, Meyer P, Khodjakov A, Rieder CL, Bornens M. (2000). J. Cell Biol., 149: 317-330. MEDLINE

Piel M, Nordberg J, Euteneuer U, Bornens M. (2001). Science, 291: 1550-1553. Article MEDLINE

Rempel RE, Sleight SB, Maller JL. (1995). J. Biol. Chem., 270: 6843-6855. MEDLINE

Rieder CL, Faruki S, Khodjakov A. (2001). Trends Cell. Biol., 11: 413-419. MEDLINE

Schmidt-Zachmann MS, Hugle-Dorr B, Franke WW. (1987). EMBO J., 6: 1881-1890. MEDLINE

Sharp DJ, Rogers GC, Scholey JM. (2000). Nature, 407: 41-47. Article MEDLINE

Sluder G, Begg DA. (1985). J. Cell Sci., 76: 35-51. MEDLINE

Sluder G, Hinchcliffe EH. (1999). Biol. Cell, 91: 413-427. MEDLINE

Sluder G, Miller FJ, Cole R, Rieder CL. (1990). J. Cell Biol., 110: 2025-2032. MEDLINE

Sluder G, Miller FJ, Rieder CL. (1986). J. Cell Biol., 103: 1873-1881. MEDLINE

Sluder G, Rieder CL. (1985). J. Cell Biol., 100: 887-896. MEDLINE

Sluder G, Thompson EA, Miller FJ, Hayes J, Rieder CL. (1997). J. Cell. Sci., 110: 421-429. MEDLINE

Spruck CH, Won KA, Reed SI. (1999). Nature, 401: 297-300. Article MEDLINE

Stearns T. (2001). Cell, 105: 417-420. MEDLINE

Stucke VM, Sillje HH, Arnaud L, Nigg EA. (2002). EMBO J., 21: 1723-1732. MEDLINE

Su J-Y, Rempel RE, Erikson E, Maller JL. (1995). Proc. Natl. Acad. Sci. USA, 92: 10187-10191. MEDLINE

Tarapore P, Tokuyama Y, Horn HF, Fukasawa K. (2001a). Oncogene, 20: 6851-6863.

Tarapore P, Horn HF, Tokuyama Y, Fukasawa K. (2001b). Oncogene, 20: 3173-3184. MEDLINE

Winey M, Goetsch L, Baum P, Byers B. (1991). J. Cell Biol., 114: 745-754. MEDLINE

Wojcik EJ, Glover DM, Hays TS. (2000). Curr. Biol., 10: 1131-1134. Article MEDLINE

Figure 1 Inhibiting Cdk2-cyclin E activity prevents the repeated duplication of the centrosome in Xenopus egg extracts. (a) Frames from a time-lapse video sequence, showing asters doubling just once over 5 h in an aphidicolin-treated extract containing 175 nM 34Xic1. Minutes after addition of sperm nuclei are seen in the lower right of each frame. Polarization optics. 10 m per scale division. (b) Frames from a time-lapse movie, showing the restoration of multiple rounds of aster doubling in an aphidicolin-treated extract containing 175 nM 34Xic1 plus 245 nM active Cdk2-cyclin E complex. Minutes after addition of sperm nuclei are seen in the lower right of each frame. Polarization optics. 10 microns per scale division. (c) Comparision of total rounds of doubling for asters in cytoplasmic extracts treated with Xic1-C, 34Xic1 alone, or 34Xic1 plus Cdk2-cyclin E. Each bar represents the doubling of an individual centrosome. Xic1-C is the C-terminal portion of Xic1; because it does not inhibit Cdk2-cyclin E activity, it is added as a control. All the asters in the Xic1-C treated extracts undergo at least two rounds of doubling. Note that in the 34Xic1 treated extracts 79% of asters double only once, whereas adding back Cdk2-cyclin E restores the ability of 92% of asters to double at least twice

Figure 2 Reduplication of the centrosome during prolonged S phase in CHO cells arrested in S phase for 60 h by aphidicolin. Microtubules (anti-a tubulin: red) and centrosomes (anti- tubulin: green). Fluorescence microscopy. Bar=10 m