Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Oncogene (2005) 24, 2589–2598 & 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc ORIGINAL PAPERS The CHFR mitotic checkpoint protein delays cell cycle progression by excluding Cyclin B1 from the nucleus Matthew K Summers1,2, John Bothos1,2 and Thanos D Halazonetis*,1,3 1 Molecular and Cellular Oncogenesis Program, The Wistar Institute, Philadelphia, PA 19104-4268, USA; 2Program in Cell and Molecular Biology, Biomedical Graduate Studies, University of Pennsylvania, Philadelphia, PA 19104, USA; 3Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA CHFR, a novel checkpoint gene inactivated in human cancer, delays chromosome condensation in cells treated with microtubule poisons. To understand the molecular mechanism for this delay, we characterized cells with inactivated CHFR and stably transfected derivatives expressing the wild-type gene. After exposure to microtubule poisons, the CHFR-expressing cells arrested transiently in early prophase with a characteristic ruffled morphology of the nuclear envelope and no signs of chromosome condensation. Several markers suggested that Cyclin A/Cdc2 had been activated, whereas Aurora-A and -B and Cyclin B1/Cdc2 were inactive. Further, Cyclin B1 was excluded from the nucleus. Ectopic expression of Cyclin B1 with a mutant nuclear export sequence induced chromosome condensation, and thus overcame the CHFR checkpoint. We conclude that the mechanism by which CHFR delays chromosome condensation involves inhibition of accumulation of Cyclin B1 in the nucleus. Oncogene (2005) 24, 2589–2598. doi:10.1038/sj.onc.1208428 Published online 24 January 2005 Keywords: CHFR; Cyclin B1; checkpoint; prophase; mitosis Introduction The process by which eucaryotic cells segregate their genetic material during cell division is referred to as mitosis. Mitosis has been divided into distinct phases based mostly on morphological criteria. In early prophase, the centrosomes separate along the nuclear periphery. In mid and late prophase, the chromosomes condense. This is then followed by breakdown of the nuclear envelope, which indicates entry into prometaphase. In metaphase, the chromosomes align on the spindle equator. Separation of the sister chromatids defines entry into anaphase and, finally, chomatid *Correspondence: TD Halazonetis, Wistar Institute, 3601 Spruce Street, Philadelphia, PA, USA; E-mail: [email protected] Received 30 August 2004; accepted 1 December 2004; published online 24 January 2005 decondensation occurs in telophase (McIntosh and Koonce, 1989). Since mitosis requires the ordered execution of several biochemically distinct processes, eucaryotic cells have developed checkpoint mechanisms to ensure the timely execution of these events. The best-characterized mitotic checkpoint is the spindle checkpoint, which inhibits sister chromatid separation until all chromosomes have aligned on the mitotic spindle (Amon, 1999; Shah and Cleveland, 2000). This checkpoint is required for cells to properly progress through mitosis. Thus, complete inactivation of the spindle checkpoint leads to embryonic lethality and has not been described in cells that retain the capacity to divide (Basu et al., 1999; Dobles et al., 2000; Kalitsis et al., 2000). However, partial inactivation of the spindle checkpoint, induced mostly by heterozygous mutations of spindle checkpoint genes, has been documented in some cancer cells, albeit infrequently, and may contribute to the genetic instability that is characteristic of human cancer (Cahill et al., 1998, 1999; Michel et al., 2001). In addition to the spindle checkpoint, a checkpoint that functions early in mitosis has also been described (Jha et al., 1994; Rieder and Cole, 2000; Scolnick and Halazonetis, 2000). The only gene known so far to participate in this checkpoint is CHFR (Checkpoint with FHA and RING domains). In CHFR-expressing cells, exposure to microtubule poisons leads to a delay in chromosome condensation (Scolnick and Halazonetis, 2000). Unlike the spindle checkpoint, the CHFR checkpoint is not essential for cell proliferation. Perhaps because of this, the CHFR gene can be inactivated in human cancer, either by promoter methylation or in a few cases by missense mutations. The overall frequency of CHFR inactivation in cancer is quite high, ranging from 15 to 50% in various tumor types (Mizuno et al., 2002; Shibata et al., 2002; Bertholon et al., 2003; Corn et al., 2003; Mariatos et al., 2003; Satoh et al., 2003; Toyota et al., 2003; Erson and Petty, 2004), which is much higher than the frequency of inactivation of all spindle checkpoint genes combined. Despite its potential importance in human cancer, very little is known about the function of the CHFR gene. The protein encoded by CHFR contains FHA and RING domains and a cysteine-rich C-terminus. The function of the C-terminal cysteine-rich region is CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2590 a Mitotic Index (%) Morphology of CHFR-expressing cells exposed to microtubule poisons Characterizing CHFR-expressing cells whose cell cycle progression is delayed in response to microtubule poisons should help elucidate the function of CHFR at the molecular level. Towards this goal, HCT116 cells, which do not express endogenous CHFR protein, were transfected with a bicistronic vector encoding hemaglutinin (HA)-tagged wild-type CHFR protein and a neoresistance marker or with the same vector expressing only the neo-resistance marker. Pools of stably transfected CHFR and neo clones were then isolated. CHFR protein was expressed in the CHFR clones at levels that were modestly higher than the levels of endogenous CHFR in SAOS2 cells (Figure 1a). Consistent with our previous results (Scolnick and Halazonetis, 2000), 13 h after exposure to nocodazole, two pools (#1 and #2) of neo clones had a much higher fraction of cells with condensed chromosomes compared to two pools of CHFR-expressing clones. This was evident both by monitoring chromosome condensation directly under the microscope (Figure 1b) or by flow cytometric analysis of cells staining positively for histone H3 phosphorylated on Ser10, which is a well-established mitotic marker (Figure 1c). These results indicate that the transfected CHFR gene was functioning in the stable clones. Further, the two CHFR pools shown in Figure 1 Oncogene R1 CHF 60 2 neo 1 40 1 20 2 CHFR 0 7 10 Nocodazole 13 h c 100 2 80 neo 1 60 1 40 2 20 CHFR 0 7 Results neo1 CHFR SAO S2 HCT 116 b H3 Ser10 P (%) unclear; however, there has been some progress in understanding the function of the FHA and RING domains. Crystallographic evidence suggests that the FHA domain of CHFR binds phosphorylated proteins (Stavridi et al., 2002), as previously reported for other FHA domains (Durocher et al., 1999). Biochemical assays further suggest that the RING domain confers ubiquitin ligase activity (Chaturvedi et al., 2002; Kang et al., 2002; Bothos et al., 2003), consistent with previous analyses of other proteins with RING domains (Joazeiro et al., 1999; Levkowitz et al., 1999; Lorick et al., 1999). Many ubiquitin ligases target their substrates for degradation in the proteasome. At least in Xenopus cellfree extracts, CHFR can mediate degradation of pololike kinase 1 (Plk1) leading to inhibition of Cdc25C and Cdc2 (Kang et al., 2002). However, other in vitro studies suggest that CHFR is a noncanonical ubiquitin ligase (Bothos et al., 2003). Noncanonical ubiquitin ligases do not target proteins for degradation, but rather utilize their ubiquitin ligase activity for signal transduction (Pickart, 2001). Since the original publication on CHFR, very little work has been carried out describing its effects on human cells. Here, we studied cell lines that do not express CHFR and stably transfected derivatives that express wild-type CHFR, and characterize the CHFRinduced delay. Our findings suggest that CHFR delays cells in early prophase by inhibiting entry of Cyclin B1 in the nucleus. 10 Nocodazole 13 h Figure 1 Characterization of pools of stably transfected HCT116 neo- and CHFR-expressing clones. (a) CHFR protein expression in HCT116 neo and CHFR stably transfected cells (pools #1). Extracts of SAOS2 osteosarcoma cells were included to show the levels of endogenous CHFR. (b) Mitotic index of stably transfected HCT116 neo- and CHFR-expressing cells (pools #1 and #2) at different times after exposure to nocodazole. DAPI-stained cells were scored for chromosome condensation. Means and standard errors of three independent experiments are shown. (c) Fraction of cells scoring positive for histone H3 Ser10 phosphorylation. Stably transfected HCT116 neo- and CHFR-expressing cells (pools #1 and #2) were exposed to nocodazole, and H3 Ser10 phosphorylation was monitored by flow cytometry behaved identically to each other, as did the two neo pools, prompting us to focus our characterization of the effects of CHFR expression on one neo (#1) and one CHFR (#1) pool. The morphology of the CHFR and neo clones after exposure to microtubule poisons was monitored by DAPI staining and immunofluorescent staining of the lamin cytoskeleton (Figure 2 and data not shown). Many of the neo cells had condensed chromosomes and diffuse lamin staining indicating nuclear envelope breakdown, whereas the few cells that had not yet entered metaphase showed a typical interphase morphology with intact lamin cytoskeleton. The majority of the CHFR-expressing cells did not have condensed chromosomes or nuclear envelope breakdown, and also did not exhibit a classical interphase morphology; instead, the outline of the nuclear envelope was ruffled CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2591 Nocodazole HCT116 − neo Paclitaxel HCT116 neo HCT116 CHFR IR 6h 7h 8h 9h HCT116 − CHFR :20 :40 6h 7h 8h 9h :20 :40 10h :20 :40 11h :20 :40 12h :20 :40 13h * * * * * * * Figure 2 Deformation of the nuclear envelope of CHFR-expressing HCT116 cells delayed in the cell cycle after exposure to microtubule poisons. Neo- and CHFR-expressing cells were exposed to ionizing radiation (IR), nocodazole or paclitaxel. After 13 h, the cells were fixed and stained for Lamins A/C. The cells marked with an asterisk are in prometaphase (Figure 2). To determine whether this morphology was specific to cell cycle delay induced in response to microtubule poisons, we also examined cells 13 h after exposure to ionizing radiation (IR). In response to IR, both the neo- and CHFR-expressing cells had a smooth nuclear envelope outline typical of cells arrested in G2 (Figure 2). To better visualize the ruffled nuclear envelope morphology of CHFR-expressing cells exposed to microtubule poisons, we generated cells that, in addition to CHFR or neo, also expressed histone H2B fused to green fluorescent protein (GFP) (Kanda et al., 1998). These cells were synchronized at the G1/S boundary with a single thymidine block, released into S phase and challenged with nocodazole 4 h later. Chromosome condensation was monitored by live cell time-lapse microscopy. The neo cells showed a smooth outline of the nuclear envelope during S and G2 and then, about 9–10 h after release from the G1/S block, condensed their chromosomes (Figure 3). By comparison, the CHFR-expressing cells did not condense their chromosomes until approximately 12–13 h after release from the G1/S block. For the first 9–10 h, these cells had a smooth nuclear envelope outline, but then adopted a ruffled outline, which persisted until the chromosomes condensed (Figure 3). In the absence of microtubule poisons, cells in G2 have a smooth outline of the nuclear envelope. However, as they progress through prophase, the nuclear envelope shows symmetrical indentations, which are thought to be initiated by microtubules penetrating a partially depolymerized lamin cytoskeleton (Georgatos et al., 1997; Beaudouin et al., 2002; Salina et al., 2002). The ruffled nuclear envelope morphology of the CHFRexpressing cells exposed to microtubule poisons cannot be due to penetrating microtubules, but could indicate partial depolymerization of the lamin cytoskeleton. In turn, this would indicate that these cells were past G2, Figure 3 Deformation of the nuclear envelope of CHFR-expressing HCT116 cells delayed in the cell cycle after exposure to microtubule poisons. Neo- and CHFR-expressing cells that also express a histone H2B-GFP fusion protein were synchronized in G1/S and then released into the cell cycle (0 h time point). Nocodazole was added 4 h later. Images were captured every 20 min starting 6 h after release from the cell cycle block. Representative cells are shown yet not past mid prophase, since the chromosomes were not condensed. Molecular markers in CHFR-expressing cells exposed to microtubule poisons Analysis of molecular markers can be used to define the phase of the cell cycle in which the CHFR-expressing cells are delayed in response to microtubule poisons, as well as to begin to elucidate the function of CHFR at the molecular level. We first examined the Cyclin A/Cdc2 and Cyclin B1/Cdc2 complexes in the neo- and CHFRexpressing cells. Synchronized cells were exposed to nocodazole and at various time points thereafter cell extracts were prepared for immunoblotting analysis and kinase activity assays or the cells were fixed for analysis by immunofluorescence. Since CHFR induces a transient mitotic delay, whose duration can vary from cell to cell, the population of synchronized cells we studied was not homogeneous: some cells were being delayed in the cell cycle with uncondensed chromosomes, while others had overcome the CHFR checkpoint and had condensed their chromosomes (late prophase and prometaphase cells). Thus, before preparing extracts the two populations of cells were separated by shake-off and extracts were prepared separately for each population. The neoexpressing cells were similarly separated into adherent and nonadherent populations. We first examined Cyclin B1/Cdc2 kinase activity using histone H1 as a substrate (Figure 4a). In the adherent CHFR-expressing cells, Cyclin B1/Cdc2 was catalytically inactive, whereas in the nonadherent cells Cyclin B1/Cdc2 activity was high. The low Cyclin B1/ Cdc2 activity in some adherent cell fractions could be explained by low level contamination of these fractions Oncogene CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2592 10 12 12 Ad NA c HCT116−CHFR 11.5 h Ad NA Ad NA Ad 10 81 0 12 12 h pTP 42 Kd Cdc25C 42 Kd M Cdc2 pTP 32 Cdc2 NM M NM M pTP MI 3 MI 11 15 NM 3 12 28 Kd 24 64 6 7 HCT116−CHFR 10 68 P−H1 8 NA b 55 14 82 % 11.5 h Ad − NA Cdc2 Ad HCT116−CHFR NM M CycA HCT116−neo d Cyclin B1 DAPI HCT116 CHFR a 16 67 % Figure 4 Analysis of Cyclin A and Cyclin B1/Cdc2 in HCT116 neo- and CHFR-expressing cells exposed to nocodazole. The cells were synchronized in G1/S, released into the cell cycle (0 h) and exposed to nocodazole (4 h later). Prior to preparing cell extracts, the G2/ early prophase (adherent, Ad) cells were separated from the prometaphase (nonadherent, NA) cells by mitotic shake-off. The efficiency of cell separation was monitored by measuring the mitotic index (MI) of a small aliquot of DAPI-stained cells. (a,b) Cyclin B1/Cdc2 activity was measured directly by monitoring phosphorylation of histone H1 (32P-H1). Cdc2 and Cdc25C electrophoretic migration was monitored by immunoblotting; M and NM indicate the predominant forms of Cdc2 and Cdc25C in mitotic and nonmitotic cells, respectively. The presence of phosphorylated Cdc2 substrates was monitored by immunoblotting with an antibody that recognizes phosphorylated threonine followed by a proline (pTP). (c) Electrophoretic migration of Cdc2 associated with Cyclin A in CHFRexpressing cells exposed to nocodazole. Cyclin A/Cdc2 complexes were immunoprecipitated with antibodies that recognize Cyclin A; the immunoprecipitates were blotted for Cyclin A and Cdc2. (d) Intracellular localization of Cyclin B1 in CHFR-expressing HCT116 cells exposed for 13 h to nocodazole with cells that had condensing or condensed chromosomes (as shown by the mitotic index (MI) in Figure 4a). Thus, these findings suggest that in CHFR-expressing cells, Cyclin B1/Cdc2 is inactive during the checkpointinduced cell cycle delay. This conclusion was further supported by examining the electrophoretic migration of Cdc2. A slowly migrating form of Cdc2, representing inactive Cdc2 doubly phosphorylated on Thr14 and Tyr15, was present in the adherent CHFR-expressing cells, which had not yet initiated chromosome condensation, but was absent in the nonadherent (prometaphase/metaphase) cells (Figure 4a). The low Cyclin B1/Cdc2 activity in CHFR-expressing cells during the checkpoint-induced cell cycle delay was somewhat inconsistent with the ruffled nuclear envelope morphology of these cells, since lamin depolymerization is thought to be mediated by Cdc2 (Ward and Kirschner, 1990; Peter et al., 1991). Thus, to further explore Cdc2 activity, we immunoblotted the cell extracts with an antibody that recognizes phosphorylated threonine followed by a proline (Figure 4a). This phosphospecific antibody reacts with Cdc2 substrates, and also with substrates of Cdk2 and other kinases. In the neo-expressing cells, at least two bands, corresponding to molecular weights of 28 and 42 kDa, respectively, were detected specifically in the prometaphase/metaphase cells (nonadherent population, Figure 4a), suggesting that these bands are Cdc2 substrates. Surprisingly, in the CHFR-expressing cells at the 12 h Oncogene timepoint the same proteins were phosphorylated in both the adherent and nonadherent cells (Figure 4a and b). The phosphorylation of these proteins in the adherent cells could not be attributed to gross contamination by prometaphase/metaphase cells, as indicated by the low MI and the electrophoretic migration of Cdc2 and Cdc25C (Figure 4a and b). Further, the phosphorylation was unlikely to be mediated by Cyclin B1/Cdc2, based on the kinase activity assays. This left us with the possibility that the phosphorylation events recognized by the phospho-Thr-Pro-specific antibody were mediated by Cyclin A/Cdc2. We could not test directly for Cyclin A/Cdc2 kinase activity by immunoprecipitating Cyclin A-associated kinases, because Cyclin A associates with both Cdc2 and Cdk2. However, we could study the electrophoretic migration of Cdc2 that coimmunoprecipitates with Cyclin A as an indirect indicator of Cyclin A/Cdc2 activity. At the 10 h time point, when CHFR-expressing cells show little reactivity with the phospho-Thr-Pro-specific antibody, the fraction of Cdc2 that coimmunoprecipitated with Cyclin A migrated as a doublet with the higher band representing Cdc2 phosphorylated on both Thr14 and Tyr15. At the 11.5 h time point, when the adherent CHFR-expressing cells showed reactivity with the phospho-Thr-Prospecific antibody, the fraction of Cdc2 that coimmunoprecipitated with Cyclin A migrated mostly as a single band representing Cdc2, in which Thr14 and Tyr15 were dephosphorylated (Figure 4c). This result, which corre- CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2593 a 8 11.5 10 13 h HCT116− CHFR neo Ad NA Ad NA Ad NA Plk1 Plk1 γ − tubulin DAPI Plk1 HCT116 − neo HCT116 − CHFR b c HCT116 − neo Ad 8 HCT116 − CHFR NA 10 12 12 Ad 8 NA 10 12 12 h AurA γ − tubulin AurA DAPI HCT116 − neo HCT116 − CHFR d e γ − tubulin AurA pT288 DAPI * HCT116 − CHFR lates with the appearance of phospho-Thr-Pro-reactive bands (Figure 4b), suggests that, unlike Cyclin B1/Cdc2, Cyclin A/Cdc2 is active in CHFR-expressing cells during the checkpoint-induced cell cycle delay. Interestingly, in the nonadherent cell fraction (prometaphase/metaphase cells), Cdc2 was not associated with Cyclin A (Figure 4c), indicating that after chromosome condensation Cdc2 associates exclusively with Cyclin B1. In addition to examining Cyclin/Cdc2 complexes in cell extracts, we also monitored the intracellular localization of Cyclins A and B1 by immunofluorescence. Cyclin A was predominantly nuclear in both the neo- and CHFR-expressing cells (data not shown). Cyclin B1, which is known to translocate from the cytoplasm to the nucleus in mid prophase and whose nuclear localization is required for chromosome condensation (Pines and Hunter, 1991), was cytoplasmic in the CHFR-expressing cells during the checkpointinduced cell cycle delay (Figure 4d). Entry of Cyclin B1 in the nucleus is regulated at least in part by phosphorylation of residues surrounding its nuclear export sequence. The polo-like kinase Plk1 phosphorylates some of these residues (Li et al., 1997; Hagting et al., 1998, 1999; Toyoshima et al., 1998; Yang et al., 1998, 2001; Toyoshima-Morimoto et al., 2001), raising the possibility that CHFR might sequester Cyclin B1 in the cytoplasm by inhibiting Plk1. In support of this model, it is known that CHFR encodes a ubiquitin protein ligase, which at least in Xenopus extracts can target Plk1 for ubiquitin-dependent proteolysis (Kang et al., 2002). We therefore examined whether Plk1 protein levels might decrease in CHFR-expressing cells exposed to nocodazole. The cell extracts prepared from the adherent and nonadherent populations of synchronized neo- and CHFR-expressing cells were immunoblotted with Plk1-specific antibodies. There was no difference in Plk1 protein levels in response to CHFR expression (Figure 5a), indicating that the CHFRmediated cell cycle delay was not associated with a decrease in overall Plk1 protein levels. We further considered the possibility that Plk1 might be degraded at specific intracellular compartments, in which case total Plk1 protein levels might not be significantly affected. To explore this possibility, we monitored by immunofluorescence the intracellular localization of Plk1 in synchronized neo- and CHFRexpressing cells exposed to microtubule poisons. Again we noted no difference between the two cell types. In the CHFR-expressing cells that were arrested in early prophase and in the neo-expressing cells that had not yet entered prometaphase, Plk1 was present at the centrosomes and in the nucleus in a kinetochore-like pattern (Figure 5b). A second kinase that regulates the intracellular localization of Cyclin B1 is Aurora-A (Hirota et al., 2003). We therefore monitored Aurora-A protein levels, intracellular localization and phosphorylation state in synchronized neo- and CHFR-expressing cells exposed to microtubule poisons. In extracts prepared from the adherent and nonadherent populations, there was no difference in Aurora-A protein levels as a function of Figure 5 Analysis of Plk1 and Aurora-A in HCT116 neo- and CHFR-expressing cells exposed to nocodazole. The cells were synchronized in G1/S, released into the cell cycle (0 h), exposed to nocodazole (4 h) and examined 8–13 h after release from the cell cycle block, as indicated. (a,c) Plk1 and Aurora-A (Aur A) protein levels. G2/early prophase (adherent, Ad) cells were separated from prometaphase (nonadherent, NA) cells by mitotic shake-off before the extracts were prepared. (b,d,e) Immunofluorescence analysis for Plk1, Aurora-A (AurA) and Aurora-A phosphorylated on Thr288 12 h after release from the G1/S block. The centrosomes (arrowheads) were marked by staining for g-tubulin. In (e),the cell marked with an asterisk is in prometaphase CHFR expression (Figure 5c). By immunofluorescence, Aurora-A was present at the centrosomes of practically every cell (the synchronized neo- and CHFR-expressing cells were in late G2, prophase and prometaphase), Oncogene CHFR excludes Cyclin B1 from the nucleus MK Summers et al suggesting that CHFR does not inhibit accumulation of Aurora-A at centrosomes (Figure 5d). We further examined whether centrosomal Aurora-A was phosphorylated on Thr288, a residue that resides in the activation loop and whose phosphorylation state parallels Aurora-A activation (Hirota et al., 2003). In the CHFR-expressing cells delayed prior to chromosome condensation, centrosomal Aurora-A was not phosphorylated on Thr288; but phosphorylation was evident in cells that had overcome the checkpoint and condensed their chromosomes (Figure 5e). Thus, CHFRinduced cell cycle delay is associated with the presence of inactive Aurora-A at the centrosomes. Aurora-B localizes to kinetochores in G2, prophase and metaphase and phosphorylates histone H3 on Ser10 in mitosis (Hsu et al., 2000; Adams et al., 2001; Giet and Glover, 2001). The absence of histone H3 phosphorylation on Ser10 in cells whose progression was delayed by CHFR (Figure 1c) implies that Aurora-B is inactive in these cells. By immunoblotting and immunofluorescence, we did not detect decreased Aurora-B protein levels in cells delayed in the cell cycle by CHFR (data not shown), suggesting that CHFR does not target Aurora-B for protein degradation. Forced expression of Cyclin B1 in the nucleus overcomes the CHFR checkpoint Inspection of more than a thousand synchronized CHFR-expressing cells exposed to nocodazole failed to identify any cells with cytoplasmic Cyclin B1 that were initiating chromosome condensation or stained positive for histone H3-Ser10 phosphorylation. However, we were able to identify several cells that had Cyclin B1 in the nucleus and were initiating chromosome condensation and/or exhibited phosphorylation of histone H3 on Ser10 (Figure 6a). From this we infer that nuclear accumulation of Cyclin B1 precedes chromosome condensation and histone H3-Ser10 phosphorylation and may be required for overcoming the cell cycle delay induced by CHFR in response to microtubule poisons. To test this hypothesis, we examined whether targeting Cyclin B1 to the nucleus could overcome the CHFRmediated cell cycle delay. The nuclear export sequence (NES) of Cyclin B1 can be inactivated by substituting Phe146 with Ala (Hagting et al., 1999). The Cyclin B1 A146 mutant or wild-type Cyclin B1, as a control, were stably introduced in SAOS2 cells, which have an intact CHFR checkpoint (Scolnick and Halazonetis, 2000). The Cyclin B1 A146 mutant overcame the CHFR-induced cell cycle delay, whereas expression of wild-type Cyclin B1 had a smaller effect (Figure 6b), even though its level of expression was similar to that of the A146 mutant (Figure 6c). A dominant negative CHFR mutant that lacks the FHA domain partially overcame the delay, consistent with our previous findings (Scolnick and Halazonetis, 2000). Since expression of the Cyclin B1 A146 mutant in other cell systems is not sufficient to drive premature entry into mitosis (Jin et al., 1998; Toyoshima et al., 1998; Hagting et al., 1999; Yang et al., 2001), we propose that Oncogene a Cyclin B1 H3 Ser10 P b DAPI 1 2 H3 Ser10 P (%) 2594 25 20 15 10 5 0 3 4 c neo 1 2 neo CHFR wt A146 ∆FHA Cyclin B1 Cyclin B1 CHFR wt A146 ∆FHA 1 2 1 2 1 2 CHFR 1 2 3 4 Cyclin B1 Figure 6 Entry of Cyclin B1 in the nucleus overcomes the CHFRinduced delay. (a) Nuclear localization of Cyclin B1 is always accompanied by histone H3 phosphorylation on Ser10. CHFRexpressing HCT116 cells were released from a G1/S cell cycle block (0 h), exposed to nocodazole (4 h) and processed for immunofluorescence (12 h). The cells numbered ‘1’ and ‘3’ have nuclear Cyclin B1 and stain positive for H3 phosphorylation on Ser10. DAPI staining further shows that cell ‘3’ has initiated chromosome condensation. (b,c) Expression of Cyclin B1 with a mutant nuclear export sequence overcomes the CHFR-induced cell cycle delay. (b) Stably transfected SAOS2 clones expressing a neo vector with no other insert or with inserts expressing a dominant negative CHFR mutant (CHFRDFHA), wild-type Cyclin B1 or Cyclin B1 A146, all with an HA tag, were released from a G1/S cell cycle block (0 h), exposed to nocodazole (8 h) and then examined for H3 Ser10 phosphorylation by flow cytometry (14 h). Each column represents the mean and standard error from two independent pools of stably transfected clones. (c) Levels of ectopically expressed HA-tagged proteins in the stably transfected SAOS2 pools (two per insert) assayed by immunoblotting with an antibody that recognizes the HA tag in our system the Cyclin B1 146 mutant overrides the CHFR-induced delay. Discussion CHFR delays chromosome condensation in cells exposed to microtubule poisons (Scolnick and Halazonetis, 2000). The original findings were consistent with cell cycle delay either in late G2 or in early prophase. In this study, we suggest that the CHFR-induced delay occurs within early prophase, rather than in late G2. This conclusion stems from analysis of synchronized nocodazole-treated live cells expressing neo or CHFR, which shows that at the time the neo-expressing cells condensed their chromosomes, the CHFR-expressing cells deformed their nuclear envelope (Figure 3). In the absence of microtubule poisons, deformation of the nuclear envelope occurs in prophase, is associated with penetrating microtubules from the maturing centrosomes and is quickly followed by nuclear envelope breakdown (Georgatos et al., 1997; Schneider et al., 1999; Salina et al., 2002). Under these conditions, deformation of the nuclear envelope is thought to reflect partial depolymerization of the lamin cytoskeleton, which in turn is dependent on Cdc2 activity (Ward and Kirschner, 1990; Peter et al., 1991). By analogy, we interpret the deformation of the nuclear envelope in CHFR-expressing cells exposed to nocodazole to CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2595 indicate that the lamin cytoskeleton was partially depolymerized and that the cells were delayed past G2 in prophase. Further, because the CHFR-delayed cells had not initiated chromosome condensation, the delay would have to be in early, rather than late, prophase. A recent study has also characterized CHFR-delayed cells and concluded that the cells were delayed in antephase (Matsusaka and Pines, 2004). Antephase is a rarely used term, originally introduced in 1951 (Bullough and Johnson, 1951), to define a phase of the cell cycle in late G2 just before chromosomes begin to condense, when normal mouse epidermal cells are reversibly delayed when subjected to various stresses. Those, who use the term antephase, consider prophase synonymous with initiation of chromosome condensation. We have used the term early prophase to define cells that are past G2, but have not yet initiated chromosome condensation. Cells in early prophase can be distinguished from cells in G2, because they have initiated centrosome movement along the periphery of the nuclear membrane, which is the earliest described mitotic event visible under the microscope (McIntosh and Koonce, 1989). From the definitions of antephase and early prophase listed above, it follows that these two terms may refer to the same phase of the cell cycle. We will continue to use the term early prophase here, but remain open to adjusting our terminology, as consensus among investigators in the field of mitosis develops. We believe that a key finding of our study is the identification of active Cyclin A/Cdc2 in CHFR-delayed cells (Figure 7). This finding was principally based on the electrophoretic mobility of Cdc2 associated with Cyclin A (Figure 4c). Even though Cdc2 electrophoretic mobility is an indirect marker of kinase activity, several other observations support the conclusion that Cyclin A/Cdc2 is active in CHFR-delayed cells. Specifically, the CHFR-delayed cells show deformation of the lamin cytoskeleton, Plk1 localization at kinetochores and phospho-Thr-Pro reactive proteins (Figure 7). The former two features are known to require Cdc2 activity (Ward and Kirschner, 1990; Peter et al., 1991; Elia et al., 2003) and the same is likely to be true for the latter. Since Cyclin B1/Cdc2 was inactive in CHFR-delayed cells, these results are consistent with Cyclin A/Cdc2 being active. Early Prophase Mid Prophase CHFR Plk1 at Kinetochores Active Cyclin A /Cdc2 Deformation of Lamin Cytoskeleton pThrPro Antibody Reactive Proteins Active Aurora B H3 Ser10 Phosphorylation Active Aurora A and Plk1 Nuclear Cyclin B1 Chromosome Condensation Active Plk1 Active Cdc25C Active Cyclin B1 /Cdc2 Nuclear Envelope Breakdown Figure 7 Model for role of CHFR in regulating early mitotic events. CHFR is shown to delay the transition from early to mid prophase by keeping Aurora-A, Aurora-B, Plk1 and Cyclin B1/ Cdc2 inactive. However, the direct molecular target of the CHFR protein remains unknown. See text for more details Interestingly, another group recently concluded that Cyclin A/Cdks are inactive in CHFR-delayed cells. This group did not examine Cyclin A/Cdk2 or Cyclin A/Cdc2 activity or post-post-translational modifications, but reached their conclusion by showing that injection of Cyclin A in complex with a constitutively active Cdk2 mutant overcame the CHFR-induced delay (Matsusaka and Pines, 2004). This result implies that even Cyclin A/ Cdk2 is inactive in CHFR-delayed cells. However, in our opinion, this result cannot substitute for direct analysis of endogenous Cyclin A/Cdks. While it would be possible for Cyclin A/Cdc2 to be inactive in CHFRdelayed cells, it is particularly hard to imagine that Cyclin A/Cdk2 would be inactive, given that this complex is activated in S phase (Woo and Poon, 2003). Unlike Cyclin A/Cdc2, Cyclin B1/Cdc2 was inactive in CHFR-delayed cells (Figure 7). This became apparent thanks to an important technical improvement from our previous study (Scolnick and Halazonetis, 2000) that allowed us to separate CHFR-delayed cells from cells that had entered prometaphase. The molecular basis for low Cyclin B1/Cdc2 activity in CHFR-delayed cells might be related to cytoplasmic sequestration of Cyclin B1; both in this study and in a previous study (Satoh et al., 2003), Cyclin B1 was shown to be in the cytoplasm of CHFR-expressing cells. We established a link between cytoplasmic sequestration of Cyclin B1 and CHFRinduced delay by showing that ectopic expression of a Cyclin B1 mutant that constitutively localizes to the nucleus overcame the CHFR-induced delay. High levels of Cyclin B1 in the nucleus could lead to activation of Cyclin B1/Cdc2 via a positive feedback loop that involves phosphorylation and activation of Cdc25C by Cyclin B1/Cdc2 and subsequent dephosphorylation and activation of Cyclin B1/Cdc2 by Cdc25C (Strausfeld et al., 1994; Karlsson et al., 1999). We therefore propose that the mechanism by which CHFR delays progression through prophase involves regulation of the intracellular localization of Cyclin B1. By immunoblot analysis, we did not observe any change in Cyclin B1 levels or electrophoretic migration in CHFR-delayed cells that would indicate that CHFR directly ubiquitinates Cyclin B1 (data not shown). Therefore, CHFR is more likely to regulate Cyclin B1 indirectly. Cyclin B1 has both nuclear localization and nuclear export sequences. Phosphorylation of serines 126, 128, 133 and 147 adjacent to the nuclear export sequence leads to accumulation of Cyclin B1 in the nucleus (Li et al., 1997; Hagting et al., 1998, 1999; Toyoshima et al., 1998; Yang et al., 1998, 2001; Toyoshima-Morimoto et al., 2001). The kinase or kinases that phosphorylate Ser126 and Ser128 of Cyclin B1 have not been identified, although Cdc2 appears to be a good candidate. Interestingly, phosphorylation of Ser147 and/or Ser133 is mediated by Plk1 (ToyoshimaMorimoto et al., 2001), which at least in vitro is a substrate of CHFR (Kang et al., 2002). This raises the possibility that CHFR inhibits accumulation of Cyclin B1 in the nucleus by targeting Plk1 for degradation. However, by immunoblotting, we did not observe a significant change in protein levels of endogenous Plk1 Oncogene CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2596 in the CHFR-delayed cells compared to the neo controls (Figure 5a) and similar results were recently reported by others (Erson and Petty, 2004; Matsusaka and Pines, 2004). CHFR expression also did not affect the Plk1 immunofluorescence signal at the centrosomes or kinetochores (Figure 5b), thus excluding the possibility that CHFR promotes degradation of Plk1 at specific cellular compartments. Aurora-A is another kinase that regulates the nuclear translocation of Cyclin B1 in mid prophase (Hirota et al., 2003). The mechanism is unclear, but requires activation of the pool of Aurora-A at the centrosomes. CHFR expression did not affect the overall protein levels of Aurora-A or its immunofluorescence signal at the centrosomes (Figure 5). Thus, like Plk1, Aurora-A was not targeted for degradation by CHFR. CHFR is the most frequently inactivated mitotic checkpoint gene in human cancer, but its molecular function is poorly understood. Our study establishes that CHFR delays cells in early prophase, provides morphological criteria to identify CHFR-delayed cells, and examines the status of key regulators of mitosis, such as Cyclin A/ Cdc2, Cyclin B1/Cdc2, Plk1, Aurora-A and Aurora-B (Figure 7). The identification of Cyclin B1 as an indirect target of CHFR is likely to help dissect the molecular function of this important mitotic checkpoint protein. Materials and methods Cell lines, antibodies and microtubule poisons SAOS2 osteosarcoma and HCT116 colon carcinoma cells express wild-type CHFR protein or no detectable CHFR protein, respectively (Scolnick and Halazonetis, 2000). Antibodies to cell cycle markers were obtained from commercial vendors, as follows: Plk1 and H3 phosphorylated on Ser 10 from Upstate Biothechnology; Cdc2, Aurora-A, Aurora-B, Cyclin A and Cyclin B1 from Transduction Labs; Lamins A/C and B from Santa Cruz; Aurora-A phosphorylated on Thr288 and phospho-Thr-Pro from Cell Signaling Technologies. Mitotic stress was induced by adding nocodazole (1 mM) or paclitaxel (5 mM) to the tissue culture media. DNA damage was induced by exposing the cells to 5 Gy IR using a 137Cs irradiator. Stably transfected cell clones HCT116 and SAOS2 cells were used to generate pools of stably transfected clones using a pIRESN2 bicistronic plasmid (Clontech Laboratories) containing various inserts, as indicated. Colonies of stably transfected cells were selected with G418, expanded and screened by immunoblotting for expression of the protein encoded by the insert. Cell cycle synchronization and mitotic shake-off For cell cycle synchronization, HCT116 and SAOS2 cells were incubated with 1.3 mM thymidine for 18 and 20 h, respectively, then washed 2 with PBS (Life Technologies) and released into media containing 250 mM thymidine and 250 mM cytidine. Nocodazole was added 4–8 h later, as indicated. For separation of G2/prophase from prometaphase cells, the tissue culture plates were pulsed on a vortexer. The supernatant was collected and combined with a PBS wash (prometaphase cells). The attached cells (G2/prophase) were collected by trypsinization. Oncogene Live cell analysis Pools of stably transfected HCT116 cells were generated, as described above, except that a plasmid expressing histone H2B fused to GFP was also cotransfected. The cells were synchronized at the G1/S boundary; 4 h after release from the cell cycle arrest, fresh media was added consisting of DMEM without phenol red, 10% fetal calf serum (FCS), 25 mM HEPES (pH 7.4) and 1 mM nocodazole. Then, the cells were placed on a microscope (Leica) equipped with water immersion optics and a Delta-T4 heating system for the stage and objective lens (Bioptechs). Images were acquired every 20 min starting 6 h after release from the cell cycle block using an ORCA ER digital camera (Hamamatsu) and processed using Imagevision tools (SGI). Mitotic index For determination of the mitotic index, the cells were fixed in 2% paraformaldehyde for 15 min at room temperature (RT), stained with DAPI and mounted on glass slides using a cytospin centrifuge (Shandon). The mitotic index, which represents the fraction of cells with fully condensed chromosomes, was determined by inspecting per data point more than 200 cells. Flow cytometry analysis for histone H3 Ser10 phosphorylation For determination of the fraction of cells with histone H3 Ser10 phosphorylation, the cells were fixed in 0.1% paraformaldehyde for 15 min at RT, washed and resuspended in icecold 70% EtOH for at least 10 min. The cells were then extracted on ice with 0.2% Triton X-100 for 5 min, washed with PBS, blocked with 10% heat-inactivated FCS in PBS for 30 min at RT and incubated overnight at 41C with an antibody that recognizes H3 phosphorylated on Ser10 diluted 1 : 200 in 1% BSA/PBS. The cells were then washed with PBS, incubated for 30 min at RT with FITC-conjugated goat-anti-rabbit antibody (Vector Laboratories) diluted 1 : 200 in 1% BSA/ PBS, washed again and resuspended in PBS supplemented with 0.1% Tween-20, 2% FCS, 5 ng/ml RNAse and 5 mg/ml propidium iodide. After 1 h incubation at 371C, the cells were analysed by flow cytometry (Becton Dickinson). Immunofluorescence The cells were seeded on 12-mm autoclaved glass coverslips and allowed to adhere for at least 2 days. After the indicated experimental treatments, the cells were washed once with PBS, fixed in 3.7% paraformaldehyde/2% sucrose in PBS for 15 min at RT or in methanol at 201C for 20 min and permeabilized with 0.5% Triton X-100 in PBS on ice for 20 min. After two washes in PBS, coverslips were blocked for 30 min at RT with 2.5% gelatin in PBS. Primary antibodies were diluted in 1% BSA/PBS and incubated at RT for 2 h or overnight at 41C. The coverslips were then washed with PBS supplemented with 0.1% Tween-20 and incubated for 30 min at RT with the appropriate secondary antibody conjugated to Alexafluor 488 or 568 (Molecular Probes) at a dilution of 1 : 500 in 1% BSA/ PBS. After washing, the cells were counterstained with DAPI and mounted with Fluorimount G (Southern Biotechnology Associates). Immunoblotting and immunoprecipitations Cells were lysed with buffer consisting of 50 mM Tris (pH 8.0), 120 mM NaCl, 0.5% NP40, 1 mM DTT, 0.4 mg/ml Pefabloc SC, 2 mg/ml pepstatin, 1 mM Calyculin A, 1 mM okadaic acid, 15 mM CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2597 NaF and 1 mM sodium vanadate. The lysates were cleared by centrifugation, resolved by gel electrophoresis and immunoblotted with antibodies that recognize a panel of cell cycle proteins. For immunoprecipitation experiments, whole-cell lysates were diluted to 0.5 ml with lysis buffer and precleared with 30 ml protein G beads (Pharmacia) for 30 min at 41C. The supernatants were incubated with primary antibodies for 1 h at 41C and the immunocomplexes were captured on protein G beads. After washing off the unbound proteins with lysis buffer, the bound proteins were immunoblotted, as described above. Cyclin B1/Cdc2 kinase assay Cyclin B1/Cdc2 was immunoprecipitated from 50 mg wholecell lysate as described above using a monoclonal antibody that recognizes Cyclin B1. The beads were washed 3x with lysis buffer and then 2x with kinase buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 5 mM MgCl2, 1 mM DTT). The beads were then resuspended in 30 ml kinase buffer supplemented with 5 mg histone H1, 1 mCi 32P-g-ATP and 5 mM MnCl2. After incubation at 301C for 30 min, the reaction products were resolved by SDS gel electrophoresis and kinase activity was monitored by fluorography. Acknowledgements We thank G Wahl for the gift of plasmid expressing GFP fused to histone H2B and Daniel Scolnick for helpful discussions. Financial support was provided by the National Cancer Institute (CA89630). MKS and JB were supported by the National Cancer Institute Training Grants CA09677 and CA09171, respectively. References Adams RR, Maiato H, Earnshaw WC and Carmena M. (2001). J. Cell Biol., 153, 865–880. Amon A. (1999). Curr. Opin. Genet. Dev., 9, 69–75. Basu J, Bousbaa H, Logarinho E, Li Z, Williams BC, Lopes C, Sunkel CE and Goldberg ML. (1999). J. Cell Biol., 146, 13–28. Beaudouin J, Gerlich D, Daigle N, Eils R and Ellenberg J. (2002). Cell, 108, 83–96. Bertholon J, Wang Q, Falette N, Verny C, Auclair J, Chassot C, Navarro C, Saurin JC and Puisieux A. (2003). Oncogene, 22, 8956–8960. Bothos J, Summers MK, Venere M, Scolnick D and Halazonetis TD. (2003). Oncogene, 22, 7101–7107. Bullough WS and Johnson M. (1951). Proc. R. Soc. Lond. B. Biol. Sci., 138, 562–575. Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Markowitz SD, Kinzler KW and Vogelstein B. (1998). Nature, 392, 300–303. Cahill DP, da Costa LT, Carson-Walter EB, Kinzler KW, Vogelstein B and Lengauer C. (1999). Genomics, 58, 181–187. Chaturvedi P, Sudaki V, Bobiak ML, Fisher PW, Mattern MR, Jablonski SA, Hurle MR, Zhu Y, Yen TJ and Zhou BB. (2002). Cancer Res., 62, 1797–1801. Corn PG, Summers MK, Fogt F, Virmani AK, Gazdar AF, Halazonetis TD and El-Deiry WS. (2003). Carcinogenesis, 24, 47–51. Dobles M, Liberal V, Scott ML, Benezra R and Sorger PK. (2000). Cell, 101, 635–645. Durocher D, Henckel J, Fersht AR and Jackson SP. (1999). Mol. Cell, 4, 387–394. Elia AE, Cantley LC and Yaffe MB. (2003). Science, 299, 1228–1231. Erson AE and Petty EM. (2004). Mol. Carcinog., 39, 26–33. Georgatos SD, Pyrpasopoulou A and Theodoropoulos PA. (1997). J. Cell Sci., 110, 2129–2140. Giet R and Glover DM. (2001). J. Cell Biol., 152, 669–682. Hagting A, Jackman M, Simpson K and Pines J. (1999). Curr. Biol., 9, 680–689. Hagting A, Karlsson C, Clute P, Jackman M and Pines J. (1998). EMBO J., 17, 4127–4138. Hirota T, Kunitoku N, Sasayama T, Marumoto T, Zhang D, Nitta M, Hatakeyama K and Saya H. (2003). Cell, 114, 585–598. Hsu JY, Sun ZW, Li X, Reuben M, Tatchell K, Bishop DK, Grushcow JM, Brame CJ, Caldwell JA, Hunt DF, Lin R, Smith MM and Allis CD. (2000). Cell, 102, 279–291. Jha MN, Bamburg JR and Bedford JS. (1994). Cancer Res., 54, 5011–5015. Jin P, Hardy S and Morgan DO. (1998). J. Cell Biol., 141, 875–885. Joazeiro CA, Wing SS, Huang H, Leverson JD, Hunter T and Liu YC. (1999). Science, 286, 309–312. Kalitsis P, Earle E, Fowler KJ and Choo KH. (2000). Genes Dev., 14, 2277–2282. Kanda T, Sullivan KF and Wahl GM. (1998). Curr. Biol., 8, 377–385. Kang D, Chen J, Wong J and Fang G. (2002). J. Cell Biol., 156, 249–259. Karlsson C, Katich S, Hagting A, Hoffmann I and Pines J. (1999). J. Cell Biol., 146, 573–584. Levkowitz G, Waterman H, Ettenberg SA, Katz M, Tsygankov AY, Alroy I, Lavi S, Iwai K, Reiss Y, Ciechanover A, Lipkowitz S and Yarden Y. (1999). Mol. Cell, 4, 1029–1040. Li J, Meyer AN and Donoghue DJ. (1997). Proc. Natl. Acad. Sci. USA, 94, 502–507. Lorick KL, Jensen JP, Fang S, Ong AM, Hatakeyam S and Weissman AM. (1999). Proc. Natl. Acad. Sci. USA, 96, 11364–11369. Mariatos G, Bothos J, Zacharatos P, Summers MK, Scolnick DM, Kittas C, Halazonetis TD and Gorgoulis VG. (2003). Cancer Res., 63, 7185–7189. Matsusaka T and Pines J. (2004). J. Cell Biol., 166, 507–516. McIntosh JR and Koonce MP. (1989). Science, 246, 622–628. Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, Gerald W, Dobles M, Sorger PK, Murty VV and Benezra R. (2001). Nature, 409, 355–359. Mizuno K, Osada H, Konishi H, Tatematsu Y, Yatabe Y, Mitsudomi T, Fujii Y and Takahashi T. (2002). Oncogene, 21, 2328–2333. Peter M, Heitlinger E, Haner M, Aebi U and Nigg EA. (1991). EMBO J., 10, 1535–1544. Pickart CM. (2001). Mol. Cell, 8, 499–504. Pines J and Hunter T. (1991). J. Cell Biol., 115, 1–17. Rieder CL and Cole R. (2000). Curr. Biol., 10, 1067–1070. Salina D, Bodoor K, Eckley DM, Schroer TA, Rattner JB and Burke B. (2002). Cell, 108, 97–107. Satoh A, Toyota M, Itoh F, Sasaki Y, Suzuki H, Ogi K, Kikuchi T, Mita H, Yamashita T, Kojima T, Kusano M, Fujita M, Hosokawa M, Endo T, Tokino T and Imai K. (2003). Cancer Res., 63, 8606–8613. Oncogene CHFR excludes Cyclin B1 from the nucleus MK Summers et al 2598 Schneider U, Mini T, Jeno P, Fisher PA and Stuurman N. (1999). Biochemistry, 38, 4620–4632. Scolnick DM and Halazonetis TD. (2000). Nature, 406, 430–435. Shah JV and Cleveland DW. (2000). Cell, 103, 997–1000. Shibata Y, Haruki N, Kuwabara Y, Ishiguro H, Shinoda N, Sato A, Kimura M, Koyama H, Toyama T, Nishiwaki T, Kudo J, Terashita Y, Konishi S, Sugiur H and Fujii Y. (2002). Carcinogenesis, 23, 1695–1699. Stavridi ES, Huyen Y, Loreto IR, Scolnick DM, Halazonetis TD, Pavletich NP and Jeffrey PD. (2002). Structure, 10, 891–899. Strausfeld U, Fernandez A, Capony JP, Girard F, Lautredou N, Derancourt J, Labbe JC and Lamb NJ. (1994). J. Biol. Chem., 269, 5989–6000. Oncogene Toyoshima F, Moriguchi T, Wada A, Fukuda M and Nishida E. (1998). EMBO J., 17, 2728–2735. Toyoshima-Morimoto F, Taniguchi E, Shinya N, Iwamatsu A and Nishida E. (2001). Nature, 410, 215–220. Toyota M, Sasaki Y, Satoh A, Ogi K, Kikuchi T, Suzuki H, Mita H, Tanaka N, Itoh F, Issa JP, Jair KW, Schuebel KE, Imai K and Tokino T. (2003). Proc. Natl. Acad. Sci. USA, 100, 7818–7823. Ward GE and Kirschner MW. (1990). Cell, 61, 561–577. Woo RA and Poon RY. (2003). Cell Cycle, 2, 316–324. Yang J, Bardes ES, Moore JD, Brennan J, Powers MA and Kornbluth S. (1998). Genes, Dev., 12, 2131–2143. Yang J, Song H, Walsh S, Bardes ES and Kornbluth S. (2001). J. Biol. Chem., 276, 3604–3609.