Events at the end of mitosis in the budding and fission yeasts

Coordination of mitosis and cytokinesis is a problem that all eukaryotes must resolve if they are to assure stable transmission of the genome during cell division. The S. pombe septation initiation network (SIN) and the S. cerevisiae mitotic exit network (MEN) are signal transduction pathways that contribute to achieving this. The role of the SIN is to initiate contraction of the actin ring and synthesis of the division septum, thereby bringing about cytokinesis. The MEN also plays a role in regulating cytokinesis, and has a second role as part of a checkpoint that monitors orientation of the mitotic spindle during anaphase. It controls inactivation of mitotic cyclin-dependent kinases (CDKs) at the end of mitosis and thereby regulates exit from mitosis. Components of both the SIN and the MEN are localised to the spindle pole body and the division site, and many of the proteins segregate asymmetrically on the spindle pole bodies during anaphase, emphasising the essential role played by changes in protein localisation in the function of these networks. In this Commentary, I discuss the importance of these topics in SIN and MEN regulation. Here, and the accompanying Cell Science at a Glance poster, I have tried to paint a `broad brush strokes' picture, rather than providing a comprehensive review of the subtleties of these networks, and highlight differences and similarities between the SIN and the MEN. Nomenclature in this field is confusing: a list of gene name-equivalents in S. cerevisiae MEN and S. pombe SIN is given in Table 1.

S. pombe is a rod-shaped organism that grows mainly by elongation at its tips. It divides by formation of a centrally placed division septum, producing two daughter cells of equal size. The position of the future division site, where the septum will form at the end of mitosis, is defined early in mitosis by the assembly of a contractile actomyosin ring (CAR) at the cell cortex. Contraction of the ring, which is thought to guide deposition of the division septum (reviewed by Ishiguro, 1998; Rajagopalan et al., 2003), is triggered by the SIN (reviewed by Bardin and Amon, 2001; McCollum and Gould, 2001; Yeong et al., 2002).

Signaling by the SIN is mediated through the GTPase Spg1p (Schmidt et al., 1997), which, in its GTP-bound form, binds the protein kinase Cdc7p (Fankhauser and Simanis, 1994). Signal transduction requires two other kinases, Sid1p and Sid2p, and their associated subunits, Cdc14p and Mob1p, respectively (Fankhauser and Simanis, 1993; Guertin et al., 2000; Hou et al., 2000; Sparks et al., 1999). Note that S. pombe Cdc14p is unrelated to S. cerevisiae Cdc14p, which encodes a phosphoprotein phosphatase (see below). The GTPase-activating protein (GAP) for Spg1p comprises two proteins: Cdc16p (Fankhauser et al., 1993), which has homology to other GAP proteins, and Byr4p (Song et al., 1996), a scaffold protein that interacts with both Spg1p and Cdc16p, and promotes the GAP activity of Cdc16p in vitro (Furge et al., 1998). In the absence of Cdc16p, Byr4p inhibits both GTP release and hydrolysis by Spg1p in vitro, thereby locking it into the active state (Furge et al., 1998). It is assumed that this also applies in vivo. Activation of the SIN is one of the mitotic roles of the Polo-family kinase Plo1p (Ohkura et al., 1995; Tanaka et al., 2001), which is recruited to the spindle pole body at the onset of mitosis (Bahler et al., 1998; Mulvihill et al., 1999). To date, no guanine-nucleotide-exchange factor (GEF) specific for Spg1p has been identified in screens for SIN mutants. However, analysis of the localization of the GAP components during mitosis (Cerutti and Simanis, 1999; Li et al., 2000) is consistent with a mechanism in which modulation of GAP activity, together with spontaneous exchange of GTP for GDP on Spg1p, is sufficient for regulation of the signaling status of Spg1p, without the requirement for a specific GEF. It must be emphasized that this concept has not yet been validated. SIN signaling is mediated from a scaffold composed of Cdc11p and Sid4p (Chang and Gould, 2000; Krapp et al., 2001; Tomlin et al., 2002).

Finally, two additional putative SIN components have been identified genetically: pld6 (Cullen et al., 2000) has not been cloned, whereas etd1 has been cloned but not characterized (Jimenez and Oballe, 1994). It is noteworthy that these genes, and mutations in mob1 (Salimova et al., 2000), were isolated by either novel selection methods for conditional mutants or screens using specifically sensitized genetic backgrounds, which suggests that further `targeted' screens may reveal additional components and regulators of the SIN. Whether this represents the full set of genes required for SIN signaling is not known. However, the fact that many have been isolated only once in genetic screens, indicates that the screens are not yet saturated and that additional components remain to be found.

Increased expression of spg1, plo1 or cdc7, or loss-of-function mutants in either subunit of the GAP, causes multiple rounds of septum formation without cell cleavage (Fankhauser and Simanis, 1994; Ohkura et al., 1995; Schmidt et al., 1997). Ectopic activation of the SIN in interphase cells–for example, by overexpression of spg1–promotes formation of the contractile ring and septum independently of entry into mitosis (Schmidt et al., 1997). In contrast, loss-of-function mutations in any of the SIN kinases, their associated subunits, Spg1p or the Sid4p-Cdc11p scaffold leads to formation of elongated, multinucleated cells, because cell growth and elongation, S phase and mitosis all continue in the absence of cytokinesis (Balasubramanian et al., 1998; Nurse et al., 1976) (Fig. 1). Cells possessing null mutations of these genes form contractile rings that either do not contract or do so only to a limited extent (Chang and Gould, 2000; Fankhauser and Simanis, 1993; Fankhauser and Simanis, 1994; Guertin et al., 2000; Hou et al., 2000; Krapp et al., 2001; Salimova et al., 2000; Schmidt et al., 1997; Sparks et al., 1999; Tomlin et al., 2002). Increased expression of byr4 also inhibits SIN signaling (Song et al., 1996). It is not clear whether all the required ring components are present in a CAR assembled in the absence of SIN signaling; indeed there is some suggestion that they may not be (Mulvihill et al., 2001). Plo1p is involved in multiple events during mitosis and plo1 null mutants have a complex phenotype (Ohkura et al., 1995). However, shut-off of plo1 expression (Ohkura et al., 1995), and some alleles of plo1 (Tanaka et al., 2001), produce a phenotype characteristic of SIN mutants.

All the SIN proteins and most of their regulators localize to the spindle pole body at some point in the cell cycle (Fig. 2). This is mediated by a scaffold composed of Cdc11p and Sid4p (Krapp et al., 2001; Tomlin et al., 2002). Localization studies suggest that Sid4p is the primary link between this scaffold and structural components of the spindle pole body, whereas the binding of all other SIN components except Plo1p requires Cdc11p; this implies that Cdc11p provides the primary link to the SIN signaling module. Plo1p interacts with the core spindle pole body component Cut12p (Grallert and Hagan, 2002; MacIver et al., 2003), which may mediate its activity and regulate its association with the spindle pole body; however, Plo1p can still associate with the spindle pole body in a cut12 mutant, so it is likely that it has other partners on the spindle pole body. Whether there are any interactions between Cut12p and other components of the SIN remains to be investigated.

Sid4p and Cdc11p are present at the spindle pole body throughout the cell cycle (Chang and Gould, 2000; Krapp et al., 2001). Both Byr4p and Cdc16p are found at the spindle pole body in interphase (Cerutti and Simanis, 1999; Li et al., 2000), when Spg1p is predominantly in the GDP-bound form (Sohrmann et al., 1998). After entry into mitosis, Byr4p remains on the spindle pole body whereas Cdc16p is not detectable (Cerutti and Simanis, 1999). At this time Spg1p is in the GTP-bound form and associates with Cdc7p on both poles of the mitotic spindle (Sohrmann et al., 1998). Concurrently with the onset of anaphase B and inactivation of mitotic cyclin activity, one pole of the spindle pole body reassembles the Byr4p-Cdc16p GAP, and inactivates Spg1p signaling, while the other retains Spg1p-GTP and Cdc7p (Sohrmann et al., 1998). This `active' spindle pole also now recruits the Sid1p-Cdc14p protein kinase. This transition is inhibited by mitotic Cdc2p activity (Guertin et al., 2000). The protein kinase Sid2p is also on the spindle pole body throughout interphase (Sparks et al., 1999), but its partner Mob1p is not (Hou et al., 2000; Salimova et al., 2000). Upon entry into mitosis, Mob1p appears on the spindle pole body, and is then seen on both spindle poles throughout mitosis (Hou et al., 2000; Salimova et al., 2000). At the end of anaphase, as the spindle breaks down, Sid2p-Mob1p is also localized at the site of the CAR, and remains there during ring contraction, flanking the developing septum (Hou et al., 2000; Salimova et al., 2000; Sparks et al., 1999). It is presumed, but not proven, that Sid2p-Mob1p supplies the signal for ring contraction, and that the population of the protein that appears at the ring must first pass through the spindle pole body. Once septation has been completed, the GAP reappears on the `active' pole and signaling is shut down. The γ-tubulin-ring complex, which is essential for nucleation of microtubules at the spindle pole body, also affects the SIN, perhaps at the level of Sid1p (Vardy et al., 2002); the molecular basis for this is unclear.

A recent study (Lu et al., 2002) has described a new thermosensitive mutation in cdc7, which raises the possibility that Cdc7p can function effectively in the cytoplasm. Although cells expressing a fusion of GFP to this mutant Cdc7p are viable, the protein does not appear to localize on the spindle pole body during mitosis. Note, however, that this protein might nevertheless still associate with the spindle pole body, but transiently.

Asymmetry seems to be achieved at the level of the GAP, although the mechanism remains unclear. Nonetheless, it is clear that the anaphase-promoting complex (APC/C; also known as the cyclosome) must act for SIN components to become asymmetric (Chang et al., 2001; Guertin et al., 2000); whether this is due to inactivation of Cdc2p, degradation of another substrate, or both, is unclear. The two spindle poles are clearly not equivalent (Bridge et al., 1998; Cerutti and Simanis, 1999), and differences between them may be generated during the conservative division of the spindle pole body. Resolution of which of the two spindle pole bodies is active in SIN signaling awaits experiments using slow-folding red-fluorescent-protein-tagged spindle pole body components to distinguish old and new spindle pole bodies, as described in an elegant study of spindle pole body inheritance in S. cerevisiae (Pereira et al., 2001).

SIN signaling is negatively regulated by Dma1p (Guertin et al., 2002; Murone and Simanis, 1996), and strongly increased expression of dma1 inhibits septum formation. It has been suggested that Dma1p prevents inappropriate localization of Plo1p (Guertin et al., 2002). It interacts with the scaffold protein Sid4p, which is thought to anchor it at the spindle pole body (Guertin et al., 2002). Dma1p also appears at the CAR during anaphase; its role and anchors there are unknown. Two other putative negative regulators of SIN signaling, zfs1 (Beltraminelli et al., 1999) and scw1 (Jin and McCollum, 2003; Karagiannis et al., 2002), have also been identified. Zfs1p is a zinc-finger protein, whereas Scw1p is an RNA-binding protein. Loss of either protein reduces, but does not eliminate, the requirement for SIN signaling. Their mode of action is unclear. However, mutation of scw1 does not restore activity to SIN mutants and may stabilize microtubules (Jin and McCollum, 2003), which suggests that the rescue is indirect.

Phosphoprotein phosphatases have been implicated in signaling by the SIN, although in no case is it clear what the relevant substrate is. Loss of Flp1p, the S. pombe orthologue of the phosphatase Cdc14p, which is the main effector of the MEN (see below), produces cytokinesis defects, and a flp1-null mutant exhibits strong genetic interactions with SIN mutants (Cueille et al., 2001; Trautmann et al., 2001). Mutations in regulatory subunits of protein phosphatase 2A (PP2A), such as Par1p, can rescue some SIN mutants, which suggests that PP2A is a negative regulator of the SIN (Jiang and Hallberg, 2001; Le Goff et al., 2001), although additional genetic interactions suggest that PP2A functions at multiple levels in the SIN (Kinoshita et al., 1996; Le Goff et al., 2001; Tanabe et al., 2001). Par1p-PP2A localizes first to the spindle pole body and then to the contractile ring during mitosis, showing a pattern similar to that of Mob1p (Le Goff et al., 2001). The calcium-dependent phosphatase calcineurin (Yoshida et al., 1994) also interacts genetically with SIN components (Lu et al., 2002).

The SIN may play a role in interphase, since Plo1p is recruited to the spindle pole body prematurely in SIN mutants (Mulvihill et al., 1999), and Plo1p kinase activity is increased in cdc7 mutants (Tanaka et al., 2001), which implies that the SIN negatively regulates Plo1p. The SIN may also respond to the intra-mitotic control on the duration of anaphase B uncovered by Hagan et al. (Hagan et al., 1990). It would clearly be of interest to study the localization of SIN components in these cells. Increased expression of cdc14 arrests cells in G2 phase in a wee1-dependent fashion (Fankhauser and Simanis, 1993). The mitotic inhibitor wee1 is also required for the function of the actin ring checkpoint (Le Goff et al., 1999; Liu et al., 2000). The role of Wee1p is to phosphorylate and inhibit Cdc2p and to control the timing of mitosis (Nurse, 1990). These observations suggest there is a link between the SIN and the mitotic control, although its nature is unclear.

It has been suggested that the SIN kinases act in a linear order: Cdc7p, Sid1p and Sid2p (Guertin et al., 2000). This idea is based primarily upon localization studies, and assays of Sid2p kinase activity, and to date no biochemical evidence supports the existence of a linear MAP-kinase-like module. Therefore, a more complicated arrangement of the elements of the SIN cannot be excluded (Fig. 3). Cdc11p is a phosphoprotein and becomes hyperphosphorylated when the SIN is activated during anaphase. This depends upon Cdc7p (Krapp et al., 2003) and Plo1p (A. Krapp, E. Cano and V. Simanis, unpublished), and may contribute to activation of the SIN. The effectors through which the SIN brings about ring contraction are unknown, although it does regulate the appearance of glucan synthases at the division site (Liu et al., 2002).

The SIN signal transduction network has counterparts in other organisms. The best characterized of these is the MEN in S. cerevisiae (Fig. 4). Note that, in S. cerevisiae, the Cdc2p equivalent is named Cdc28p, and mitotic cyclins are named CLBs. In addition to using the MEN to regulate formation of the actin ring and cytokinesis at the end of mitosis, budding yeast also uses it to control the degradation of CLBs to bring about mitotic exit. The MEN is part of a checkpoint that monitors spindle position in anaphase. MEN mutants arrest with separated chromosomes, elevated mitotic CDK activity and an elongated spindle (Fig. 1D). Since a decrease in CDK activity is required for mitotic exit and cytokinesis, involvement of the MEN in cytokinesis was not recognised initially because the mitotic exit defect predominates (see below). The following section is simply an overview of events at the end of mitosis in S. cerevisiae and the role played by the MEN in them; exhaustive reviews are available elsewhere (Bardin and Amon, 2001; Geymonat et al., 2002; Jensen and Johnston, 2002; Yeong et al., 2002).

When all the sister chromatids are attached properly to the mitotic spindle, and the decision to commit to anaphase has been made, a number of proteins are degraded after ubiquitylation by the APC/C, its specificity being conferred by targeting subunits called Cdc20p (APC/C^Cdc20p) and Cdh1p (APC/C^Cdh1p). Recent reviews describe proteolysis during mitosis (Irniger, 2002), and the spindle assembly checkpoint, which governs the metaphase-anaphase transition (Zhou et al., 2002). The initial wave of degradation is APC/C^Cdc20p dependent and targets the securin Pds1p, and mitotic and S-phase B-type cyclins (Wasch and Cross, 2002). Elimination of securin liberates separase (Esp1p) to cleave the cohesin complex and permit sister chromatid separation (Nasmyth, 2002; Petronczki et al., 2003). Esp1p also cleaves other proteins during mitosis that may contribute to spindle stability and orderly anaphase progression (Sullivan et al., 2001). As anaphase proceeds, the APC/C switches specificity factors, and APC/C^Cdh1p completes ubiquitylation of the remaining B-type cyclin pools, and also targets other proteins for destruction. In the context of mitotic exit, the most notable are Spo12p and Cdc5p.

Complete inactivation of mitotic CDK activity is not essential for sister chromatid separation and anaphase onset, but it is required for normal exit from mitosis (Irniger, 2002). Once B-type cyclin CDK activity is reduced to low levels, cytokinesis can occur, origins of replication can be reset by formation of pre-replication complexes, and cdk inhibitors (CKIs) such as Sic1p can accumulate to establish and maintain a G1 state. The phosphoprotein phosphatase Cdc14p is a key element in bringing about mitotic exit in S. cerevisiae. Its main job seems to be to reverse phosphorylation events by proline-directed protein kinases, including Cdc28p (Gray et al., 2003; Visintin et al., 1998). The known substrates of Cdc14p that contribute to mitotic exit include Cdh1p, which it dephosphorylates to permit it to interact with the APC/C, thereby promoting B-type cyclin degradation during late mitosis and G1 phase. Cdc14p also dephosphorylates the transcription factor Swi5p, which permits it to enter the nucleus and activate expression of the CKI Sic1p, which is then dephosphorylated by Cdc14p and protected from ubiquitylation by the Skp1–Cullin–F-box (SCF) ubiquitin ligase complex and therefore protected from degradation. These two events promote Sic1p accumulation and lower the activity of the Cdc28p-CLB complex. Cdc14p also targets components of the MEN (see below).

Changes in subcellular localization play a major role in regulating the access of Cdc14p to its substrates (Fig. 5). During interphase Cdc14p is sequestered in the nucleolus, tethered to the RENT complex by Net1p. Early in mitosis four proteins, Spo12p, Cdc5p, Esp1p and Slk19p, collectively referred to as the FEAR network, bring about the release of Cdc14p from the nucleolus (Dumitrescu and Saunders, 2002; Stegmeier et al., 2002). How this is achieved remains unclear, but the role of Esp1p in FEAR function may not require its proteolytic activity (Sullivan and Uhlmann, 2003). The polo-like kinase Cdc5p can promote release of Cdc14p from the RENT complex in vitro, but whether this is the in vivo mechanism is unclear (Shah et al., 2001; Yoshida and Toh-e, 2002).

The activity of the MEN is then required to keep Cdc14p out of the nucleolus (Stegmeier et al., 2002) and to allow it to associate with the spindle pole body during mitosis (Yoshida et al., 2002). In MEN mutants, Cdc14p remains in the nucleolus (Fig. 1E). Signaling by the MEN is mediated by the GTPase Tem1p, which binds to a protein kinase, Cdc15p (Asakawa et al., 2001; Bardin et al., 2000; Lee et al., 2001c). Cdc15p then directly activates the Dbf2p-Mob1p complex (Mah et al., 2001). Tem1p is negatively regulated by the two-component GAP Bfa1p-Bub2p (Geymonat et al., 2003; Ro et al., 2002). Mitotic exit requires the GEF Lte1p, but only at low temperatures (Shirayama et al., 1994). Bfa1p is a substrate of the FEAR component Cdc5p, which inactivates the Bfa1-Bub2p GAP (Geymonat et al., 2003; Hu et al., 2001). Bub2p is also phosphorylated during mitosis, but the role of this remains unclear (Hu and Elledge, 2002). Cdc5p is also required downstream of Tem1p in the MEN, to provide full activation of Dbf2p (Lee et al., 2001a). Cdc14p activates Cdc15p, which provides a potential positive feedback loop to maintain Cdc14p outside the nucleolus, and acts on Lte1p to release it from the cell cortex (Jensen et al., 2002); its effect on Lte1p activity is unclear.

This system of CDK inactivation also contains the seeds of its own inactivation: as already mentioned, Bfa1p is dephosphorylated and reactivated by Cdc14p, thereby limiting signaling by the MEN (Hu et al., 2001; Pereira et al., 2002). Likewise, activation of APC/C^Cdh1p promotes destruction of Spo12p and Cdc5p, both components of the FEAR network, thus curtailing FEAR activity. Furthermore, Cdc14p also promotes expression of AMN1, whose product interferes with formation of the Tem1p-Cdc15p complex (Wang, Y. et al., 2003), thus blocking MEN signaling (the so-called AMEN network). Amn1p is synthesized only after the MEN has acted, and its activity is in turn limited by its being a Cdc28p-CLB substrate, which promotes its SCF-mediated destruction (Wang, Y. et al., 2003). What causes Cdc14p to return to the nucleolus is unknown, although inactivation of the MEN and FEAR and activation of AMEN might all contribute. Mutations in nuclear transport machinery have been shown to cause inappropriate release of Cdc14p from the nucleus (Asakawa and Toh-e, 2002), but the identities of the importins and exportins that control its transport to and from the nucleus are unknown.

Like its fission yeast counterpart, the MEN is also associated with the spindle pole body (reviewed by Bardin and Amon, 2001; Pereira and Schiebel, 2001), where it is anchored to a scaffold protein, Nud1p (Gruneberg et al., 2000): see Fig. 4. Although there are subtle differences in how the conserved components of these networks behave, a common theme is asymmetry, the signaling components being associated with the spindle pole body that is directed into the daughter cell. The GEF Lte1p is spatially restricted to the daughter cell, where it is associated with the cortex. The current model is that conversion of Tem1p to the GTP-bound form, and therefore activation of the MEN, only occurs when one spindle pole body enters the daughter cell and encounters Lte1p, although this view has recently been challenged (see below). However, since Lte1p is essential for mitotic exit only at low temperatures, other mechanisms must also play a role (Adames et al., 2001).

Cell polarity proteins such as Kel1p and Kel2p also regulate mitotic exit. Their role is unclear, but studies indicate that they act as inhibitors of mitotic exit (Hofken and Schiebel, 2002), and it has also been suggested that they anchor Lte1p in the bud (Seshan et al., 2002). Cells carrying mutations in both S. cerevisiae RAS genes are unable to exit from mitosis (Morishita et al., 1995). Structure-function analysis of Lte1p has led to the proposition that it is an effector of RAS2, and that its GEF domain is essential for binding to Ras2p to anchor it at the cortex, rather than acting as a GEF for Tem1p (Yoshida et al., 2003). The role of Lte1p is therefore unclear, although it has been suggested that Lte1p inhibits Kel1p and Kel2p (Yoshida et al., 2003).

The cell polarity proteins Ste20p and Cla4p also play an important role in regulating mitotic exit. The GTPase Cdc42p, its activator Cdc24p and its effector, the protein kinase Cla4p, are all required for association of Lte1p with the cell cortex but not its subsequent retention there (Hofken and Schiebel, 2002; Jensen et al., 2002; Seshan et al., 2002). This appears to require its phosphorylation, which is mediated by a combination of Cdc28p-CLN activity and Cla4p. Cla4-null cells have a defect in mitotic exit at low temperatures (Hofken and Schiebel, 2002). Ste20p, another Cdc42p effector, is also implicated in mitotic exit. It may activate the FEAR component Cdc5p (Hofken and Schiebel, 2002), or contribute to activation of APC/C^Cdc20p (Chiroli et al., 2003). Whether S. pombe cell polarity proteins control the SIN is unknown.

The MEN controls not only cyclin degradation but also cytokinesis. Classical MEN mutants arrest with extended spindles, separated chromosomes, and elevated Cdc28p-CLB activity. However, some alleles of MEN genes permit mitotic exit, but not cytokinesis (Jimenez et al., 1998). Furthermore, mutations in Net1p that bypass the requirement for MEN in cyclin degradation still leave cells with a defect in cytokinesis (Lippincott et al., 2001). The MEN has been implicated in regulation of septin and actomyosin ring dynamics (Lee et al., 2001b; Lippincott et al., 2001), although the relevant substrates are unknown. Cells possessing null mutations in the PP2A B′ regulatory subunit RTS1 (the orthologue of Par1) are stress sensitive and arrest in G2 phase with an undivided nucleus. It is not clear whether this protein has any role in regulating the MEN (Shu et al., 1997). Interestingly, PP2A-RTS1 acts downstream of the MEN to regulate septin dynamics (Dobbelaere et al., 2003). Increased expression of CDC5 can promote formation of septal structures in S. cerevisiae (Song et al., 2000), and increased expression of BFA1 arrests cells in late anaphase (Li, 1999; Ro et al., 2002). Thus the effects produced by increased expression of these MEN proteins are similar to that of their S. pombe counterparts (Fig. 1).

In contrast to its S. cerevisiae counterpart (Cdc14p), S. pombe Flp1p (also referred to as Clp1p) (Cueille et al., 2001; Trautmann et al., 2001) is not essential for mitotic exit, cyclin degradation, CKI stabilization in G1 phase, or cell proliferation. However, flp1 mutants show defects in cytokinesis (Cueille et al., 2001; Trautmann et al., 2001), and signal transduction by the SIN is less effective (Cueille et al., 2001). Flp1-null cells divide at a reduced size, indicating that mitosis is advanced, since in rapid exponential growth the G2-M transition is limiting for cell cycle progression in S. pombe. Furthermore, strong overexpression of flp1 causes a wee1-dependent G2 arrest. The biological relevance of the latter result remains to be established, but it is clear that Flp1p can influence the activity of mitotic regulators in fission yeast.

Flp1p is present at multiple locations in the cell. During interphase it is associated with the spindle pole body and also in the nucleolus. Upon entry into mitosis it is released from the nucleus and associates with the spindle and contractile ring. Its substrates and anchors are unknown. By analogy with Cdc14p of S. cerevisiae, it is reasonable to assume that changes in localization are important for its regulation, although this remains to be proven. SIN activity is not required for release of Flp1p from the nucleolus, but inactivation of SIN signaling seems to be needed for it to return to the nucleolus at the end of mitosis. By analogy with S. cerevisiae, one would predict that Flp1p should be required for dephosphorylation of Ste9p, the orthologue of Cdh1p: this is not the case (Cueille et al., 2001). Equally, one would predict that both Byr4p and Cdc7p are substrates of Flp1p. Although there is a very strong genetic interaction between cdc7 and flp1 mutants, neither protein has yet been shown to be regulated by Flp1p. The role of the S. pombe equivalents of the FEAR components in release of Flp1p from the nucleolus remains to be assessed.

The SIN and Flp1p are also important for an “actin ring checkpoint” in fission yeast, which prevents ring contraction and mitosis in the subsequent cell cycle if the CAR is defective (Le Goff et al., 1999; Liu et al., 2000; Liu et al., 1999). Their exact role in this checkpoint is unclear: they may participate in the transduction process; alternatively, if the SIN never signals that the ring should contract, then this checkpoint might not be activated.

In S. cerevisiae, the MEN forms one branch of the spindle assembly checkpoint, preventing mitotic exit if the spindle is misorientated (Bardin and Amon, 2001; Pereira and Schiebel, 2001; Yeong et al., 2002). In S. pombe the SIN and its regulators Dma1p and Zfs1p have also been implicated in preventing cytokinesis if the mitotic spindle is defective (Beltraminelli et al., 1999; Fankhauser et al., 1993; Guertin et al., 2002; Murone and Simanis, 1996). In both S. pombe and S. cerevisiae, the spindle assembly checkpoint prevents sister chromatid separation, anaphase onset and B-type cyclin degradation until all sister chromatids are attached to the spindle in a bipolar fashion, and the kinetochores are under tension (Zhou et al., 2002). How the spindle checkpoint impinges on the SIN is unclear, but inactivation of Cdc2p in spindle-assembly-checkpoint-arrested cells promotes septum formation (He et al., 1997), which suggests that Cdc2p prevents septation. The APC/C must also act to bring about septum formation (Chang et al., 2001), although it is not known whether this is by inactivation of Cdc2p through promoting ubiquitylation of Cdc13p (the major mitotic B-type cyclin), some other substrate, or both. Inactivation of Cdc28p in mitotically arrested cells also promotes mitotic exit and cytokinesis in S. cerevisiae (Ghiara et al., 1991).

In S. pombe, degradation of Cdc13p after ubiquitylation by the APC/C at the metaphase-anaphase transition reduces Cdc2p activity to low levels, permitting origin resetting and the initiation of cytokinesis. Its continued degradation during the M-G1 transition requires the Cdh1p orthologue Ste9p (Blanco et al., 2000; Yamaguchi et al., 2000). In contrast to S. cerevisiae, the SIN does not regulate the degradation of the mitotic cyclin Cdc13p (Guertin et al., 2000). Indeed, it seems that elevated Cdc2p activity inhibits the SIN (see above).

In addition to controlling cytokinesis, S. cerevisiae has adapted the MEN to a second role, namely regulation of mitotic exit, in part through regulating degradation of mitotic cyclins, in response to a spindle orientation checkpoint. Since in the budding mode of division, the cells choose the division site at the onset of the cell cycle, it is essential to ensure that the spindle is correctly positioned with one end in the mother and the other in the daughter cell, before permitting inactivation of mitotic cyclins, mitotic exit and cytokinesis. Thus, while a large part of the B-type cyclins are degraded at anaphase onset, the rest are not degraded unless the anaphase spindle is correctly positioned. A signal transduction system, such as the MEN, that is located upon the spindle pole body is an ideal way to monitor the position of the anaphase spindle. In S. pombe, where the division site is fixed early in mitosis by the position of the interphase nucleus (Chang and Nurse, 1996), the separated chromosomes need only be moved apart by extension of the spindle to ensure that each cell will receive one nucleus. What delays septation until the time of spindle breakdown is not clear. It is interesting to note that S. pombe cells also monitor orientation of the mitotic spindle early in mitosis (Gachet et al., 2001) (reviewed by Wang, H. et al., 2003). Whether the SIN participates in this checkpoint is unknown.

Both transduction networks show asymmetric distribution of proteins on the spindle pole bodies (Figs 2 and 5). In S. cerevisiae, the logic for this is obvious, for it assures that the mitotic spindle has traversed the bud neck and entered the daughter cell before the MEN is activated. Although the asymmetry of protein segregation in these networks was first discovered in the fission yeast (Sohrmann et al., 1998), it is less clear what purpose it serves.

In the absence of GAP regulation of the SIN, its components segregate symmetrically during mitosis (Sohrmann et al., 1998) but, if SIN signaling is attenuated, the cells are viable in spite of this (Fournier et al., 2001). Thus asymmetry is not essential, the caveat being that we do not know whether both spindle pole bodies can signal. Assuming that the amount of SIN protein at the spindle pole body is reflected by the immunofluorescence/GFP signals, a common theme for all the asymmetric components (Byr4p, Cdc16p, Cdc7p, Sid1p and Cdc14p) is that the signal is weak at the onset of anaphase B and strong at the end of anaphase. It is tempting to draw an analogy with the control of mitosis, where Cdc25p activity eventually overwhelms the inhibitory effects of Wee1p on Cdc2p, to suggest that the gradual accumulation of SIN proteins at this pole titrates an inhibitor of SIN signaling, thereby delaying septum formation until the end of mitosis. Asymmetry of rate-limiting components would permit cells to control the signaling machinery at one place. In S. cerevisiae, elevated Cdc28p-CLB activity prevents cytokinesis until the end of mitosis. Since S. pombe B-type cyclins are destroyed at the onset of anaphase B (Alfa et al., 1990; Decottignies et al., 2001), some other timing mechanism must operate to prevent septum formation until nuclear separation has been completed. Proteins such as Dma1p may fulfil this role. It is also possible that asymmetry is not necessary for regulation of septum formation and simply reflects the fact that spindle pole bodies are different.

S. pombe is not dimorphic, although some related strains are (reviewed by Sipiczki, 2000). It is possible that the asymmetry played a role in ancestral yeasts that grew as filaments, perhaps ensuring that the growth machinery was correctly orientated to the growing end of the cell at the end of mitosis.

Orthologues of the SIN proteins are found in filamentous fungi, where they perform a similar function to the SIN in controlling cytokinesis (reviewed by Harris, 2001). Some of these genes have been identified in multicellular organisms, such as C. elegans, and functional inactivation has demonstrated that some are essential for life, although their roles remain to be elucidated (Gruneberg et al., 2002). Orthologues of the SIN components Plo1p, Mob1p and Sid2p are found in all eukaryotes. The counterparts of Cdc5p/Plo1p are known as Polo-like kinases and are required for formation of a bipolar spindle, exit from mitosis, and cytokinesis (reviewed by Donaldson et al., 2001; Glover et al., 1998). The counterpart of the Dbf2p/Sid2p-like protein kinases in Drosophila is called Warts and is a tumor suppressor (Justice et al., 1995; Xu et al., 1995). Its equivalent in human cells, LATS1, interacts with the mitosis inducer CDK1 (Tao et al., 1999). Both Polo and Warts proteins associate with the centrosome, then the spindle (which determines the position of the cleavage furrow) and finally the midbody. The function of Mob1p-like proteins is less well understood, at present, although they are found as components of multiple protein complexes in human cells (Moreno et al., 2001). Proteins containing GAP domains that have significant homology to Bub2p/Cdc16p exist in higher eukaryotes and are present at the centrosome (Cuif et al., 1999). Whether these function in mitotic exit is unknown. Cloning of the centrosomal protein centriolin reveals the presence of a small region that has sequence similarity to Cdc11p and Nud1p (Gromley et al., 2003). Increased expression of this domain, or siRNA-mediated knock-down of expression reveals roles for centriolin in cytokinesis and G1 progression. The domain homologous to Cdc11p/Nud1p can also interact with Bub2p, which raises the possibility that functional counterparts of all the SIN and MEN regulators exist in higher eukaryotes.

Higher eukaryotes have orthologues of the Cdc14p phosphoprotein phosphatase. In C. elegans, it is involved in cytokinesis (Gruneberg et al., 2002). In human cells, there are two proteins. One of these is located in the nucleolus, whereas the other is associated with centrosomes and may control their duplication. As is the case in C. elegans, these proteins play a role in cytokinesis (Kaiser et al., 2002; Mailand et al., 2002).

A protein with a similar arrangement of domains to Dma1p, called Chfr, has been identified in mammalian cells (Scolnick et al., 2000). This seems to be part of a checkpoint that monitors microtubule integrity in prophase (reviewed by Cortez and Elledge, 2000). Chfrp can act as a ubiquitin ligase in vitro, and Polo-like kinases may be among its targets in some systems (Chaturvedi et al., 2002; Kang et al., 2002). Its expression is also altered in some tumours (Shibata et al., 2002; Toyota et al., 2003).

Further advances in our understanding of SIN and MEN function will come from several directions. Genome-wide and proteomic analyses in both yeasts will undoubtedly reveal new components of the SIN and MEN signaling networks, and this will surely be complemented by screens for new mutants, perhaps using strategies other than thermosensitivity. Detailed structure-function analysis of SIN and MEN proteins has already begun to provide insights into their regulation (Bardin et al., 2003; Guertin and McCollum, 2001), and 3D structures will add further insight (Gray et al., 2003). An area that is only beginning to be explored at present is roles for these proteins in meiosis. In S. pombe, cells do not septate during meiotic divisions, but nothing is known about how the SIN and its regulators behave during meiosis. In S. cerevisiae, recent studies have implicated components of the FEAR network in meiotic chromosome segregation (Buonomo et al., 2003; Clyne et al., 2003; Lee and Amon, 2003; Marston et al., 2003). Analysis of the human orthologues of these proteins may well reveal roles for some of them in tumour development and cell cycle progression.

Work in my lab is funded by the Swiss Cancer League, The Swiss National Science Foundation, The Fondation Forez and ISREC. I am grateful to Rosella Visintin and Angelika Amon (Center for Cancer Research, MIT, Cambridge USA) for providing the image of an S. cerevisiae MEN mutant shown in Fig. 1D, and to Elmar Schiebel (Patterson Institute for Cancer Research, Christie Hospital, Manchester, UK), for that shown in 1E. I apologise for citing reviews at the expense of all the relevant original literature. I am also grateful to members of my lab for many discussions about SIN and in particular to Andrea Krapp, Iain Hagan, and Richard Sever for critical reading of the text.