Introduction

After DNA replication in eukaryotes, pairs of sister chromatids are linked along their lengths by catenated (topologically entangled) double-stranded DNA fibres—inevitable products of the replication process—and by cohesin complexes that most likely form a ring around the sisters (see below). Before cell division and chromosome segregation, links between sister chromatids have to be removed to enable chromosomes to segregate to opposite poles. Most of the DNA catenations on chromosome arms are lost between replication and mitosis, whereas most cohesins are released from chromosome arms during prophase. Centromeric catenations and cohesin complexes remain until the metaphase–anaphase transition.

In single-celled eukaryotes such as the budding yeast Saccharomyces cerevisiae and the fission yeast Schizosaccharomyces pombe, sister chromatids are separated along their lengths at metaphase, but in multicellular eukaryotes such as vertebrates, release of arm cohesion precedes loss of centromere cohesion. In meiosis, these processes differ. During the first meiotic division (meiosis I), homologous pairs of chromosomes are segregated rather than sister chromatids, which are not segregated until the second round of meiotic division (meiosis II). The sequence of events that accompany loss of cohesion at mitosis and meiosis has long puzzled cell biologists, but the past year has seen an expansion in our knowledge of these processes. This review discusses this knowledge and focuses on the events that occur between the onset of mitosis and the onset of anaphase. Events occurring immediately before and after this period will be described briefly and are reviewed elsewhere (Ohi and Gould 1999; Stegmeier and Amon 2004). Where relevant, we will compare and contrast the pathways underpinning mitosis and meiosis. The understanding of the fidelity of these pathways is paramount because defects at any stage can lead to aneuploidy or apoptosis that both contribute to cancer and other genetic diseases.

We will begin with an outline of the key structural and protein components that are pertinent to the present discussion. Proteins will be listed by their human and budding yeast nomenclature where different, e.g. Rad21/Scc1. The names and functions of the proteins referred to in the text and figures of this review are also summarised in Table 1.

Table 1 Proteins involved in the passage from prophase to the onset of anaphase
Full size table

The kinetochore

The kinetochore is a multiprotein complex that is the structural and functional hub of the centromere (Cleveland et al. 2003; Maiato et al. 2004). It contains constitutive proteins that provide a foundation structure throughout the cell cycle (Amor et al. 2004) and transient proteins that contribute to the different stages of cell division (Vagnarelli and Earnshaw 2004). Basic kinetochore structure consists of three zones that roughly correspond to the three layers seen with electron microscopy: the inner, middle and outer layers. The inner layer comprises constitutive, structural proteins such as the centromere-specific histone H3 homologue CENP-A/Cse4. The outer layer consists of proteins involved in microtubule contact, dynamics and spindle checkpoint signalling. The middle layer contains multiprotein complexes, some relatively stable and some more dynamic, that connect the inner and outer layers in a structural and functional manner (Maiato et al. 2004; McAinsh et al. 2003). Many proteins in the inner layer are necessary for binding of proteins in the middle and outer layers, but the hierarchical order of kinetochore protein deposition both within and between the layers remains to be fully elucidated.

Pericentric heterochromatin

Heterochromatin is a multi-subunit complex that causes chromatin to remain condensed and transcriptionally inactive throughout the cell cycle (Craig 2005; Fukagawa et al. 2004). As a rule, multicellular eukaryotes have their kinetochores embedded in tandemly repetitive DNA and/or regions highly enriched in retrotransposons. These sequences can be transcribed and have the potential to disrupt centromere structure and function if they are not silenced by heterochromatin (Craig 2005; Fukagawa et al. 2004; Wong and Choo 2004). Although the simple “point” centromeres of budding yeast have no heterochromatin, domains of heterochromatin are seen in the centromeres of the fission yeast. Wherever it occurs, pericentric heterochromatin has been found to be essential for mitosis. It is responsible for establishing sister chromatid cohesion via the cohesin complex and DNA topoisomerase II (Topo II) (Craig 2005). Furthermore, the heterochromatin cross-linking protein HP1 (known in fission yeast as Swi6), in addition to binding the fission yeast cohesin subunit SA/Scc3, has also been found to bind other kinetochore proteins (see below).

The cohesin complex

Mitotic cohesin has four subunits: SMC1, SMC3, Rad21/Scc1 and SA/Scc3 (Haering and Nasmyth 2003). Humans have two mitotic paralogues of SA/Scc3 (SA1 and SA2), and meiotic paralogues have been identified for all cohesins except SMC3. SMC1 and SMC3 form a ring structure that is sealed by Rad21/Scc1 and SA/Scc3, and it has been proposed that each cohesin complex encircles one DNA strand from each sister chromatid (Haering and Nasmyth 2003). The mechanism of cohesin removal at the metaphase–anaphase transition involves cleavage of the Rad21/Scc1 cohesin subunit by the protease Separase/Esp1. The mechanism of cohesion loss on chromosome arms is discussed below.

The passenger protein kinase complex

Passenger proteins were originally defined as being nuclear in G2, associating with condensing chromosomes during prophase, concentrating at the centromere in metaphase and transferring to the spindle at the onset of anaphase (Earnshaw and Cooke 1991). Of the known passenger proteins, four form a stable complex with a protein kinase function: Aurora B/Ipl1 kinase, INCENP/Sli15, Survivin/Bir1 and Borealin. This complex influences a number of mitotic events including the removal of chromosome condensation, arm cohesion, correction of aberrant kinetochore–microtubule attachments, bipolar spindle assembly, targeting of checkpoint proteins to kinetochores and cytokinesis (Vagnarelli and Earnshaw 2004).

Topoisomerase II

Topo II catalyses the passing of one double-strand of DNA through another by breaking and rejoining one of the strands, and can thereby catenate or decatenate two strands of DNA. Topo II localises to chromosomes during interphase and through anaphase. Centromeric Topo II increases between prophase and metaphase and decreases during anaphase (reviewed by Porter and Farr 2004). Identifying specific functions for Topo II has proven to be challenging. From studies using Topo II inhibitors or RNAi-mediated knockdown in different metazoan species, it was found that Topo II contributes to chromatin compaction prior to metaphase (Porter and Farr 2004). The gene knockdown of the two isoforms of human Topo II or the single isoform of Drosophila Topo II resulted in extended mitotic chromosomes and chromosome mis-segregation (Chang et al. 2003; Sakaguchi and Kikuchi 2004). However, the precise contribution of Topo II to the separation of sister chromatids remains to be fully elucidated. As Topo II can catenate and decatenate DNA, it is unclear whether it remains at centromeres until signalling from the spindle checkpoint to decatenate and release sister centromeres, or whether it actively catenates sister centromeres until it is removed in response to checkpoint cues. Current evidence appears to favour the latter possibility because during the metaphase–anaphase transition, Topo II is modified by the addition of the ubiquitin-like molecule SUMO, causing its removal from chromatin without affecting its activity (Azuma et al. 2003; Bachant et al. 2002). Although the mechanisms for the subsequent loss of sister centromere catenation are unknown, this finding implicates the presence of a Topo II-independent pathway that decatenates DNA in its absence. In addition, the temporal and spatial relationships between Topo II and cohesins remain to be investigated.

The stages leading to anaphase

During G2 phase of the eukaryotic cell cycle, checkpoint mechanisms ensure that a cell has accurately replicated its DNA and repaired any DNA damage. Also during G2, the level of mitotic cyclins begins to rise and culminates in the assembly of M-phase-promoting factor (MPF), a complex of mitotic cyclins and cyclin-dependent kinases. MPF initiates assembly of the mitotic spindle, breakdown of the nuclear envelope and chromosome condensation (reviewed in Ohi and Gould 1999).

After the onset of mitosis, a number of stages have to be traversed before anaphase chromosome segregation is possible. These stages have traditionally been classified as prophase, prometaphase and metaphase. In prophase, the mitotic spindle starts to form as asters in the cytoplasm and spindle fibres start to radiate from these asters. In prometaphase, kinetochores capture microtubules first from one pole then from the other, correct aberrant microtubule capture and oscillate until they line up at the metaphase plate. To aid the discussion of the underlying mechanisms leading to the separation of sister chromatids during these early mitotic phases, we have defined the following six functional stages (Fig. 1) and discussed each below:

  1. 1.

    Condensation and removal of links between chromosome arms

  2. 2.

    Microtubule capture by one of a pair of kinetochores

  3. 3.

    Correction of erroneous microtubule attachments

  4. 4.

    Bi-orientation, stabilisation and generation of tension.

  5. 5.

    Passing the spindle checkpoint

  6. 6.

    Separation of sister chromatids by removal of centromeric cohesin and DNA decatenation

Fig. 1
figure 1

Mitotic events leading to the safe passage of prophase through anaphase. The stages (1–6) of mitosis, from the chromosomal point of view, from prophase through the onset of anaphase are numbered as in the text. Before stage 1, Topo II catenation (blue ovals) has been removed from most of the chromosome excluding the centromere and some sites on chromosome arms. During stage 1, chromosomes condense and cohesins (red ovals) are removed from most sites along chromosome arms and remain concentrated at centromeres (yellow circles). At stage 2, microtubules (purple rods) are captured by one kinetochore. At stage 3, incorrect spindle attachments, syntelic (top) or merotelic (bottom), are corrected, resulting in stable bi-orientation and generation of tension by spindle dynamics that include a "Pac-man" depolymerisation of kinetochore microtubules (stage 4). The dotted arrow between syntelic and merotelic configurations indicates the possibility that one may change into the other with the addition or removal of a kinetochore–microtubule interaction. At metaphase (stage 5), all chromosomes bi-orient and align on the metaphase plate. At the onset of anaphase (stage 6), cohesins and Topo II are removed from centromeric and residual sites along chromosome arms and the chromosome starts to migrate towards the spindle poles

Full size image

Stage 1. Condensation and removal of links between chromosome arms

Chromosome condensation occurs during prophase and prometaphase and utilises condensin protein complexes (Ono et al. 2004). Condensin distribution changes throughout chromosomes in early prophase to an axial distribution in early prometaphase (Kireeva et al. 2004). Condensin-mediated chromosome condensation is essential for resolution of sister chromatids and for the correct apposition of sister kinetochores towards opposite poles (Moore et al. 2005; Ono et al. 2004). Chromosomes lacking condensin can delay condensation, fail to resolve sister chromatids, fail to associate with nonhistone centromere proteins and/or exhibit aberrant chromosome alignment and segregation (Hudson et al. 2003; Moore et al. 2005; Ono et al. 2004). Furthermore, at least in Caenorhabditis elegans, Aurora B/Ipl1 kinase is likely responsible for recruiting centromeric condensins (Moore et al. 2005).

Topo II-mediated DNA catenation of chromosome arms is almost complete by the onset of mitosis. The situation is different with cohesin-mediated sister chromatid cohesion, which is not lost from arms until prophase. Removal of cohesin in prophase involves the phosphorylation of the cohesin subunit SA/Scc3, most likely by Polo-like kinase 1 (Plk1/Cdc5) and/or Aurora B/Ipl1 (Hauf et al. 2005; Kitajima et al. 2005; McGuinness et al. 2005; Sumara et al. 2002) (Fig. 2). The reason that cohesins remain at centromeres during prophase is that they are protected by the recently identified protein Shugoshin (Sgo1, Japanese for “guardian spirit”) (Kitajima et al. 2005; McGuinness et al. 2005; Tang et al. 2004). Sgo1 protects centromeric cohesion until the metaphase–anaphase transition (see below). It is targeted to the kinetochore by the spindle checkpoint protein Bub1 and is degraded at the metaphase–anaphase transition (Kitajima et al. 2005; Salic et al. 2004; Tang et al. 2004). It is also possible that Sgo1 protects a small subset of chromosome arm cohesin complex in prophase and may also be a structural link between the inner kinetochore and microtubules (see below and Figs. 2 and 4). Some cohesins have been observed at chromosome arms at metaphase, and chromosome arms not treated with spindle poisons usually stay together at metaphase and are resistant to physical separation (Gimenez-Abian et al. 2004; Paliulis and Nicklas 2004).

Fig. 2
figure 2

Interactions of some of the proteins involved in the steps between prophase cohesin removal and stable bi-orientation and generation of tension between kinetochores. In stage 1, Sgo1, deposited on chromosomes by Bub1, protects cohesin subunit SA/Scc3 against Aurora/Ipl1 and PLK1/Cdc5-mediated phosphorylation and loss from DNA. In stage 2, microtubules are captured by kinetochores via the Dam1 complex (Dam1c) in budding yeast and unknown proteins in other organisms. Microtubules are further anchored to centromere (cen) DNA via the Ndc80, Ctf19 and KNL complexes and by DNA-binding proteins such as CENP-A/Cse4 and CENP-C/Mif2. In stage 3, syntelic kinetochore–microtubule interactions are corrected by Aurora B/Ipl1-mediated phosphorylation of target proteins such as Ndc80 and MCAK and by Sgo1 by an unknown mechanism. In stage 4, stable bi-orientation is accompanied by inter-kinetochore tension induced by a number of different proteins including mitotic kinases, cohesins and Topo II. Tension-induced separation of Aurora B and MCAK has been proposed to make MCAK available for dephosphorylation and subsequent reinforcement of microtubule depolymerisation-induced inter-kinetochore tension, making ready for the onset of anaphase

Full size image

Stage 2. Microtubule capture by one of a pair of kinetochores

Even though more than 60 different centromere proteins have now been described, it has not been until very recently that the search for the protein that tethers microtubules to kinetochores has yielded results. Study of the budding yeast has shown that microtubules are attached to kinetochores via the Dam1 complex (Westermann et al. 2005). This ten-subunit complex oligomerises to form rings and/or spirals around microtubule plus ends thereby stabilising them (Westermann et al. 2005) (Fig. 2). These rings can slide along microtubules and thus tether microtubules to kinetochores even when they are in flux. However, Dam1 homologues have yet to be found in fission yeast or complex eukaryotes. Furthermore, even though a number of microtubule-associated proteins are known in complex eukaryotes, whether any of them are involved in a Dam-like function remains to be seen.

What are the proteins that have been shown to link the Dam complex to centromeres? Dam1 interacts with proteins from the Ctf19 and Ndc80 kinetochore protein complexes (Shang et al. 2003; Westermann et al. 2005) (Fig. 2). The Ctf19 complex interacts with all three budding yeast centromere DNA-binding complexes but has so far not been identified in other eukaryotes (Ortiz et al. 1999). The Ndc80 complex has homologues in all eukaryotes so far studied and is named after one of its four subunits, known in humans as Hec1 (Kline-Smith et al. 2005). Studies in C. elegans have found that the Ndc80 complex binds to kinetochores via the kinetochore null (KNL) protein complex that is localised to kinetochores via the DNA-binding inner kinetochore proteins CENP-A and CENP-C (Cheeseman et al. 2004) (Fig. 2) and to the cohesin complex via an interaction between Hec1 and SMC1 (Zheng et al. 1999). It has also been shown from work with C. elegans that the conserved nine-subunit Mis12/Mtw1 protein complex assists microtubule attachments most likely by regulating the rate at which they are formed (Cheeseman et al. 2004). Interestingly, the Mis12/Mtw1 complex interacts directly with mammalian HP1 paralogues HP1α and HP1γ, linking heterochromatin formation with microtubule capture (Obuse et al. 2004).

Stage 3. Correction of erroneous kinetochore attachments

Immediately after microtubule capture of one of a pair of sister kinetochores, chromosomes move towards the pole from which the microtubule arose. This movement is caused by depolymerisation of the polar (minus) end of the microtubule and by kinetochore-mediated depolymerisation of the plus end. The latter, so-called “Pac-man” action, happens because kinetochore proteins such as CENP-E and cytoplasmic dynein and its binding partner dynactin hold on to microtubules during the process of depolymerisation (Cleveland et al. 2003; Heald 2000). However, when these processes are occurring, many other microtubules are also radiating from the same pole. It is therefore understandable that both sister kinetochores may be captured by microtubules from the same pole (so-called syntelic attachment, Fig. 1). The spindle checkpoint monitors bipolar kinetochore attachment by not allowing cells to progress to anaphase if any kinetochores have not achieved bi-orientation (see below). However, if cells relied purely on this checkpoint, anaphase would never be allowed to proceed if even one kinetochore had syntelic microtubule attachments. Recently, light has been shed on this conundrum with the discovery that Aurora B/Ipl1 kinase can correct syntelic microtubule attachments. Aurora B/Ipl1 differentiates between bi-oriented and syntelic microtubule attachments because the latter lacks tension (Dewar et al. 2004). Sensing this lack of tension at a syntelic kinetochore, Aurora B/Ipl1 disassembles one of the two kinetochore–microtubule attachments probably by phosphorylating one or more of its many substrates. Aurora B/Ipl1 phosphorylation of microtubule centromere-associated kinesin (MCAK) localises it to kinetochores (Fig. 2), and MCAK phosphorylation decreases its microtubule-depolymerising activities. However, it seems that residual MCAK depolymerisation activity is enough to destabilise microtubules at one of the syntelic sister kinetochores, resulting in attachment of only one sister to one spindle pole (Andrews et al. 2004; Kline-Smith et al. 2005; Lampson et al. 2004). MCAK is probably aided in this task by Aurora B/Ipl1-mediated phosphorylation of Ndc80 and Dam1 thereby disrupting their interaction (Shang et al. 2003) (Fig. 2). Aurora B/Ipl1 may also be involved in this process through phosphorylation of the phosphatase Cdc14/Clp1 (Trautmann et al. 2004).

Merotelic kinetochore attachments, where one of a pair of bi-oriented sisters is attached to two spindle poles (Fig. 1), are more troublesome to deal with than syntelic attachments. This is because kinetochores are still under tension from bi-orientation and therefore unable to trigger the anaphase checkpoint. For this reason, merotely is a major mechanism for generating aneuploidy in mammalian cells (Cimini et al. 2001), and therefore, correction of merotely is paramount to prevent genetic instability and carcinogenesis. Merotely is more common in species such as C. elegans with holocentric chromosomes (centromeres span whole chromosomes). Studies of C. elegans have implicated two mechanisms in prevention and correction of merotelic attachments. Firstly, correct chromosome condensation is a prerequisite for correct apposition of sister kinetochores. Secondly, the KLP-19 protein ensures that bi-oriented chromosomes are under constant tension, maintaining their apposition to opposite poles (Powers et al. 2004). Other mammalian proteins shown to be important for correction of merotelic attachments include the passenger protein Borealin (Vagnarelli and Earnshaw 2004) and MCAK (Kline-Smith et al. 2005).

Stage 4. Bi-orientation, stabilisation and generation of tension

After avoiding or correcting abnormal spindle attachments, a centromere bi-orientates and its position oscillates until it congresses at the metaphase plate. Microtubule depolymerisation is minimised giving each kinetochore pair some stability (Fig. 3). Proteins involved in this stabilisation include the EB1 protein and its binding partner adenomatous polyposis coli (APC), the Ran GTPase activating protein RanGAP1 and its binding partner RanBP2, and the Rho GTPase CDC42 and its effector mDia3 (Maiato et al. 2004; Yasuda et al. 2004). Further illustrating the interactivity of the vast network of mitotic and kinetochore proteins, mDia3 binds to CENP-A and HP1 in human cells (Yasuda et al. 2004). Microtubule stabilisation helps promote tension between bi-oriented kinetochores. Tension between physically connected kinetochores is more important than cohesion per se to satisfy the spindle checkpoint because two kinetochores on the same circular minichromosome generated tension in the absence of functional cohesins and Topo II (Dewar et al. 2004). Dewar et al. (2004) also found that Aurora B/Ipl1 kinase is essential for the generation of tension between kinetochores. It has been proposed that tension between kinetochores pulls sister kinetochores apart enough to separate the Aurora B/Ipl1 at the inner centromere region from its substrate MCAK at the kinetochore (Andrews et al. 2004; Tanaka et al. 2002) (Fig. 3). It was proposed that MCAK would then be free to be dephosphorylated, possibly by the protein phosphatase 1 (PP1) protein, thus, activating its microtubule depolymerisation function (Fig. 2). This depolymerisation would presumably activate Pac-man-like movement of the kinetochore towards the pole. However, due to sister kinetochore links made by cohesins and Topo II, each sister kinetochore would not be able to travel far but would be primed for quick segregation at the onset of anaphase.

Fig. 3
figure 3

Models for generation of tension at kinetochores with Aurora B, MCAK and Shugoshin. The centromeric region of a pair of mitotic sister chromatids is illustrated in a and b, with spindles (purple rods) attaching to centromeres via outer kinetochore proteins (yellow ovals). Prior to stable bi-orientation, microtubules form unstable attachments to kinetochores (double-headed arrows). a Aurora B/Ipl1 (blue) and MCAK (red) co-localise to a large extent with the former localising more to the inner centromere region. (Andrews et al. 2004; Tanaka et al. 2002). Aurora B/Ipl1 can phosphorylate MCAK and weaken its microtubule depolymerisation activity. After the generation of tension through bipolar spindle attachment at metaphase, MCAK is pulled away from Aurora B/Ipl1 at the inner centromere region, freeing MCAK to be dephosphorylated and activating its microtubule depolymerisation function, therefore enabling it to pull the kinetochore and chromosome towards the spindle pole. At the onset of anaphase, sister chromatids separate and MCAK remains at kinetochores whereas Aurora B transfers to the spindle midzone. b Shugoshin (green), in addition to protecting centromeric cohesins cleavage during prophase, has been shown to interact with microtubules (Indjeian et al. 2005). When kinetochores are not under tension, Shugoshin is able to interact with microtubules and sends a signal to destabilise them. As centromeres achieve stable bi-orientation, microtubules are pulled away from Shugoshin, which is now unable to destabilise them. Shugoshin is degraded at the onset of anaphase. These two models are not mutually exclusive and were adapted from Andrews et al. (2004), Indjeian et al. (2005), Tanaka et al. (2002)

Full size image

In addition to Aurora B/Ipl1 kinase, a role in the creation of tension at kinetochores has been implicated for Plk1/Cdc5 kinase (Gruneberg et al. 2004; Sumara et al. 2004) (Fig. 2). However, it is not known which targets of Plk1/Cdc5 are phosphorylated or whether Plk1/Cdc5 acts at kinetochores or spindle poles. Sgo1 has also been recently implicated in tension sensing (Indjeian et al. 2005) (Fig. 2). Budding yeast cells with mutation in the SGO1 gene were unable to arrest in response to a lack of tension between sister kinetochores. As Sgo1 has been found to be directly involved in interactions between kinetochores and microtubules (Salic et al. 2004), it has been suggested that it acts as a tension sensor (Indjeian et al. 2005) (Fig 2). In addition, KNL proteins have been shown to be essential for kinetochores to sustain tension (Cheeseman et al. 2004) (Fig. 2). Current evidence has therefore unveiled a network of proteins that perform the tension generating and sensing function.

Stage 5. Passing the spindle checkpoint

The spindle checkpoint does not allow decatenation and loss of cohesion of any sister kinetochores until all kinetochores have achieved bi-orientation at the metaphase plate with kinetochores under tension (reviewed in Taylor et al. 2004). In brief, during prophase and prometaphase, a complex of at least nine proteins assembles at the kinetochore aided at least in part by Aurora B/Ipl1 kinase (Lampson et al. 2004; Taylor et al. 2004). These proteins include Mad2, CDC20, Bub3 and BubR1 that together form the soluble mitotic checkpoint complex (MCC) that most likely directly inhibits the onset of anaphase (Fig. 4). Kinetochore checkpoint proteins also include CENP-E, which binds microtubules, and other spindle attachment sensors such as Mad2 and Bub1. The anaphase inhibition signal is down-regulated in a stepwise fashion as mono- and bi-oriented microtubule attachments are made and tension is generated. Only at this final stage is the soluble MCC no longer produced and the kinetochores are free to segregate (Fig. 4).

Fig. 4
figure 4

Interactions of some of the proteins involved in release of the mitotic checkpoint and subsequent release of centromeric cohesion. Unattached kinetochores release a soluble complex of checkpoint proteins that inhibit Cdc20. When every kinetochore has been captured, Cdc20 is free to activate the APC/cyclosome, a ubiquitin ligase that targets mitotic cyclin B and Securin for degradation. Separase, previously bound and inactivated by Securin, is then free to cleave the cohesin subunit Rad21/Scc1, resulting in the breakdown of centromeric cohesin complexes. At the same time, Topo II is sumoylated and released from sister chromatids. At this stage, sister chromatids can be segregated to opposite poles (red arrows) through kinetochore (k)–microtubule (m) interactions

Full size image

At the same time as the anaphase inhibition signal is down-regulated, Cdc20 becomes liberated from the kinetochore checkpoint complex and is free to activate the anaphase promoting complex/cyclosome (APC/C). The APC/C activates Separase/Esp1 by ubiquitinating and targeting the Separase/Esp1 inhibitor Securin/Pds1 for degradation (Fig. 4). Separase/Esp1 is then free to cleave the cohesin complex. The APC/C also induces the degradation of mitotic cyclin B (see below) and Sgo1 (Salic et al. 2004), the loss of the latter removing the protection from centromeric cohesins.

Stage 6. Separation of sister chromatids by removal of centromeric cohesin and DNA decatenation

DNA in sister kinetochores and at residual sites on chromosome arms is freed from being bound by the cohesin complex after cleavage of its Rad21/Scc1 subunit (reviewed in Haering and Nasmyth 2003), and this cleavage is enhanced by prior phosphorylation of Rad21 (Hauf et al. 2005) (Fig. 4). During the period leading up to the successful passage through the spindle checkpoint, Topo II has been actively catenating DNA at sister centromeres and possibly at a small number of regions along chromosome arms. At this point, Topo II is modified by the addition of the ubiquitin-like molecule SUMO, which causes its removal from chromatin but does not affect its activity (reviewed in Porter and Farr 2004) (Fig. 4). This loss of Topo II is accompanied by a loss of catenation through a largely unknown mechanism.

The transition from metaphase to anaphase is aided by the APC/C-dependent, ubiquitin-mediated destruction of cyclin B (Fig. 4). Free Securin/Pds1 induces Cdc14-mediated inactivation of the cyclin-dependent kinase Cdk1 via the FEAR and MEN networks of proteins (reviewed in Stegmeier and Amon 2004). In budding yeast, Cdc14 also controls the segregation of repetitive DNA and telomeres in a cohesin-independent but Aurora B- and condensin-dependent manner (D'Amours et al. 2004; Sullivan et al. 2004).

Mitosis vs meiosis

In the light of the preceding discussion, it would be of interest to compare and contrast the relevant features of mitosis and meiosis. In meiosis I, chromosome homologues are segregated to opposite poles whereas sister chromatids remain joined until meiosis II where sister chromatids segregate to opposite poles (reviewed in Allshire 2004; Marston and Amon 2004). Current data indicate that whereas some meiosis-specific mechanisms exist, numerous functional components have a common counterpart in mitosis and meiosis, suggesting that functional adaptation of these components has occurred during the evolution of these processes. Firstly, maternal and paternal chromosomes pair via chiasmata produced during recombination, with this pairing being aided by chromosome condensation (Chan et al. 2004) and, at least in the budding yeast, by centromere pairing (Kemp et al. 2004). Secondly, some cohesin subunits have meiotic counterparts: Rec8 substituting for Rad21/Scc1, STAG3 for SA/Scc3 and SMC1β for SMC1 (Marston and Amon 2004; Revenkova et al. 2004). Interestingly, Rad21/Scc1 is also present at centromeres during mouse meiosis from prophase I to the start of anaphase II, prompting the suggestion that meiosis II is more similar to mitosis than first thought (Xu et al. 2004). Thirdly, at least in fission yeast, the Spo13 protein and the monopolin protein complex, both present only during meiosis I, ensure that sister chromatids are orientated towards the same pole and help in the protection of centromeric cohesins, possibly by maintaining high levels of Securin/Pds1 (Katis et al. 2004; Lee et al. 2004). Fourthly, Sgo1 protects Rec8 from cleavage by Separase/Esp1 during anaphase I but not anaphase II, where at least in Drosophila, the Mei-S332 protein, an earlier identified weakly conserved homologue of Sgo1, is delocalised from centromeres by phosphorylation by the kinase Plk1/Cdc5 (Clarke et al. 2005). These results, however, disagree with the implication of Plk1/Cdc5 in the removal of arm cohesion in mitotic prophase (see above). Possible reasons could be that either Sgo1 and Mei-S332 have overlapping but non-identical functions, and/or that phosphoregulation of STAG3 differs from that of SA/Scc3. Finally, it has recently been shown that in fission yeast, the Mes1 protein is expressed in mitosis and meiosis but alternatively spliced and highly expressed between late MI and late MII (Izawa et al. 2005). During this time, its role is to inhibit APC/C-mediated degradation of MPF during anaphase of meiosis I but not during anaphase of meiosis II.

Cancer

It is not surprising that problems with proteins involved in cell division can lead to oncogenesis. In a microarray meta-analysis of expression of genes in more than 3,700 separate cancer samples, Rhodes et al. (2004) identified a small subset of genes that are universally activated in most cancer types or activated in undifferentiated cancer (Rhodes et al. 2004). Present in this list were Topo II, Plk1, Rad21, Cdc20, Survivin and CENP-A. Overexpression of Aurora B (Meraldi et al. 2004), INCENP and Securin (Zou et al. 1999) has also been linked to oncogenesis, and chromosomal instability (CIN) has been attributed to mutations in mitotic checkpoint proteins (Rajagopalan and Lengauer 2004).

Conclusions

This review has focussed on recent findings related to the events that occur between the onset of mitosis and the onset of anaphase. These events include the removal of connections between chromosome arms, chromosome condensation, microtubule capture at kinetochores, correction of aberrant microtubule attachments, bi-orientation, stabilisation and generation of tension at kinetochores, overcoming the spindle checkpoint and separation of sister chromatids at the metaphase–anaphase transition. The reasons behind the differences between mitosis and meiosis I have also been discussed, revealing the interactions between meiosis-specific factors and proteins involved in both processes. We have discussed recent important discoveries including that of Shugoshin and its homologues that protect mitotic centromeres from cohesin cleavage, the mechanisms involved in the generation and sensing of tension between kinetochores and the mechanisms involved in protection and orientation of sister kinetochores during the first meiotic division. Protein phosphorylation has emerged as a major controller of mitotic events. It can influence loss of arm cohesion during prophase, correction of inappropriate kinetochore–microtubule interactions, generation of inter-kinetochore tension, Rad21/Scc1 cleavage, and removal of meiosis-specific protection proteins. The concept of disruption of protein interaction by spatial separation has also emerged as a mechanism of control of protein activity. At each stage of mitosis or meiosis, correct associations need to be made between different proteins and protein complexes if genomic stability is to be maintained. Most importantly, the events that occur at each stage have to be choreographed precisely if cancer or genetic disease is to be avoided.