Cytokinetic actomyosin ring formation and septation in fission yeast are dependent on the full recruitment of the polo-like kinase Plo1 to the spindle pole body and a functional spindle assembly checkpoint

Cytokinesis, the process by which a cell with newly divided nuclei cleaves to form two juxtaposed daughter cells, is essential for cells to propagate. Underlying cytokinesis is the contraction of the cytokinetic actomyosin ring(CAR). As its name implies, the constriction of the CAR is based on the sliding interaction of actin filaments powered by the motor activity of a type II myosin (Schroeder, 1973). Numerous other proteins are required for ring assembly and contraction and to link the CAR to the cell membrane, allowing actin filament sliding to constrict the cell surface in the space vacated by the retreating chromosomes(reviewed in Rappaport, 1996;Shuster and Burgess, 1999;Glotzer, 2001).

In the fission yeast Schizosaccharomyces pombe, the CAR contains two type II myosins, Myo2 and Myp2 (Marks and Hyams, 1985; Bezanilla et al., 1997; Kitayama et al.,1997; May et al.,1997; Mulvihill et al.,2000) (reviewed in Win et al.,2002), as well as a number of other proteins required for its correct placement and function (reviewed inLe Goff et al., 1999a;Balasubramanian et al., 2000). Although other classes of myosins localise to the division plane in S. pombe (Win et al., 2001;Win et al., 2002), only the type II myosins appear to have a role in CAR contraction. Myo2 is essential for cytokinesis (Kitayama et al.,1997; May et al.,1997), whereas Myp2 is dispensable under normal growth conditions(Bezanilla et al., 1997;Motegi et al., 1997;Mulvihill et al., 2000). Why fission yeast should have two type II myosins for cytokinesis (their only known function) when other organisms get by perfectly well with just one(Watts et al., 1987;De Lozanne and Spudich, 1987;Knecht and Loomis, 1987)remains to be determined. This is particularly puzzling given that Dictyostelium cells lacking myosin II can undergo cytokinesis in the complete absence of the motor protein(Gerisch and Weber, 2000), as can certain strains of the budding yeast, Saccharomyces cerevisiae(Bi et al., 1998).

Myo2 localises to the CAR early in mitosis, but CAR contraction only begins at the end of anaphase B/telophase when the two daughter nuclei have been segregated to opposite ends of the cell and then returned to what will become midpoints of the two daughter cells(Kitayama et al., 1997). However, the exact timing of the recruitment of Myo2 to the division plane has yet to be determined. This is due, at least in part, to the fact that metaphase is extremely brief in S. pombe(Nabeshima et al., 1998) and,hence, the precise relationship between chromosome alignment and segregation and CAR formation and function is difficult to establish. The onset of anaphase chromosome separation is regulated by the anaphase-promoting complex(APC), which determines both the loss of sister chromatid cohesion and cyclin B degradation (reviewed in Morgan,1999; Zacchariae and Nasmyth,1999). The APC in turn receives signals that can inhibit its activity from the spindle assembly checkpoint (SAC). This provides a mechanism to inhibit anaphase until all sister chromatids are attached to the spindle and properly aligned at the metaphase plate. In S. pombe, microtubule depolymerisation also delays the onset of cytokinesis(Alfa et al., 1990), as well as inhibiting the activation of MPF, an event generally considered to be essential for cytokinesis to occur(Satterwhite and Pollard,1992). The relationship between anaphase onset and CAR formation remains to be established, as does the trigger for CAR contraction. However,several molecules have roles in both processes. One such protein is Plo1, the S. pombe polo-like kinase (Ohkura et al., 1995). Perturbing the cellular level of Plo1 can dramatically effect both mitosis and cytokinesis, even driving CAR and septum formation in interphase cells (Ohkura et al., 1995). In the budding yeast, S. cerevisiae, the polo-like kinase Cdc5p has an essential role in activating the onset of anaphase by both activating the APC and phosphorylating APC substrates(Shirayama et al., 1998;Alexandru et al., 2001). It also provides the signal to activate the mitotic exit network (MEN) and thus the eventual onset of cytokinesis (Hu et al., 2001; Stegmeier et al.,2002). The MEN equivalent in the fission yeast, the septation initiation network (SIN), is a signal transduction pathway that lies downstream of Plo1 (Mulvihill et al.,1999; Tanaka et al.,2001). Both Plo1 and all the components of the SIN localise to the spindle pole body (SPB), and this association is essential for the correct regulation of cytokinesis (reviewed inBalasubramanian et al., 2000;Bardin and Amon, 2001). The SIN is activated by a G protein, Spg1 (Schmidt et al., 1997), which in its GTP-bound form activates a downstream protein kinase, Cdc7 (Fankauser and Simanis, 1994;Sohrmann et al., 1998). The GAP complex, which restores the GDP form of Spg1 is made up of two proteins,Byr4 and Cdc16, the latter being a homologue of the budding yeast spindle checkpoint protein, Bub2 (Fankhauser et al., 1993; Furge et al.,1998). Cdc7 function is required for the stable recruitment of Myo2 to the CAR, but whether it directly phosphorylates Myo2 heavy chain remains to be demonstrated (Mulvihill et al., 2001). Cdc7 also activates downstream kinases in the SIN,leading to cytokinesis and the eventual formation of a septum. Previous studies have demonstrated the existence of checkpoint mechanisms that act through the SIN proteins to prevent septation from occurring in mitotic cells if its progression is inhibited (Murone and Simanis, 1996;Beltraminelli et al., 1999;Le Goff et al., 1999b). Thus,the SIN provides a mechanism in which communication can occur between spindle formation, CAR formation and cytokinesis. Here we present evidence to show that the formation of a stable CAR in S. pombe is coincident with the onset of anaphase A. Activation of the SAC delays CAR formation by inhibition of the localisation of Plo1 to the SPB. This in turn prevents downstream events such as the Cdc7 localisation to the SPB and the recruitment of Myo2 to the CAR.

The strains used in this study are listed inTable 1. Cell culture and maintenance were carried out according to Moreno et al.(Moreno et al., 1991). Cells were grown in rich medium (YES), except when promoter repression or derepression was required in which case cells were grown in minimal medium(EMM2). Repression of the nmt1 promoter(Maundrell, 1993) was carried out by the addition of 4 μM thiamine to the growth medium. Genetic crosses were carried out on MSA plates (Egel et al., 1994) and tetrads dissected using a Singer MSM Micromanipulator (Singer Instruments, UK). To follow CAR formation and the separation of centromeric DNA simultaneously, a strain containing the GFP-LacI-NLS fusion gene integrated at the his7⁺ locus and the LacO array integrated on the lys1⁺ locus (for details,see Nabeshima et al., 1998)was crossed with the myo2-gc cdc25-22 strain, and subsequent gfp-lacI-NLS:his7 lacO:lys1 myo2-gc cdc25-22 cells were identified by visual screening. Synchronous cultures were generated by size selection(Carr et al., 1995), by transient arrest using the cdc25-22 mutation(Booher et al., 1989) or by transient arrest with hydroxyurea (HU)(Mitchison and Creanor, 1971). In all cases, samples were removed every 20 minutes and fixed for microscopy. Microtubule depolymerisation was achieved by the addition of either 100μg/ml thiabendazole (TBZ) or 25 μg/ml methyl 2-benzimidazolecarbamate(MBC). The actin cytoskeleton was depolymerised by the addition of 10 μM latrunculin B (Gachet et al.,2001). Depolymerisation drugs were added to cultures immediately prior to the cells' release into growth conditions after synchronisation, and their efficacy was checked by immunofluorescence with appropriate fixation controls.

For GFP autofluorescence microscopy, cells were fixed in 3.7% formalin for 10 minutes. DAPI staining of DNA and calcofluor staining to visualise septa were performed according to Moreno et al.(Moreno et al., 1991). Actin localisation was carried out using rhodamine-conjugated phalloidin according to Marks and Hyams (Marks and Hyams,1985). Cells were visualised using a Zeiss Axiophot microscope,and images were captured via a Hamamatsu C2400-08 digital camera with C2400 controller using Openlab software (Improvision, Coventry, UK). Plo1-GFP fluorescence intensity measurements were determined using Openlab software by comparing GFP fluorescence at the SPB with the cytoplasmic background signal.

To follow the dynamics of the S. pombe cytokinetic actomyosin ring(CAR), we took advantage of a strain in which the gene encoding the essential type II myosin, Myo2, was tagged at its genomic locus with a cDNA encoding green fluorescent protein (Mulvihill et al., 2001). This strain, myo2-gc, grew normally over a range of temperatures (25°C to 36°C), enabling us to follow the dynamics of CAR assembly and contraction in living cells in real time(Fig. 1A). To determine the precise timing of mitotic events with respect to CAR formation and function, myo2-gc was crossed with a strain in which the essential SPB component Cut12 was also tagged with GFP(Bridge et al., 1998). myo2-gc cut12-gfp cells were synchronised using either of two distinct methods. The first made use of cell-size selection of early G2 cells on a lactose gradient (Carr et al.,1995). The second was to make a myo2-gc cut12-gfp strain carrying the cdc25-22 temperature-sensitive allele and to use the ability of this mutant allele to transiently arrest cells immediately prior to the onset of mitosis (Booher et al.,1989). Similar results were obtained with both methods. When G2 cells were allowed to synchronously enter mitosis, SPB duplication occurred prior to the formation of Myo2 rings, which in turn occurred prior to the appearance of binucleate cells and septa(Fig. 1B-D). Thus the Myo2 ring forms after SPB separation but before anaphase nuclear separation.

These findings do not clarify whether the CAR forms prior to, or subsequent to, the onset of anaphase A. To address this question the myo2-gcallele was introduced into a genetic background in which the cen1⁺ locus was tagged with GFP(Nabeshima et al., 1998). In asynchronous cultures, Myo2 rings were only seen in cells in which it was possible to see two Cen1-GFP dots (data not shown). When cells were synchronised using the cdc25-22 allele, separated centromeres just preceded the appearance of Myo2 rings (Fig. 2A), and cells were often seen with separated Cen1 loci but no CAR(Fig. 2B). We therefore conclude that stable Myo2 rings form immediately following the onset of anaphase A.

To confirm that Myo2 ring formation was an anaphase event and to explore the possibility that ring formation was dependent on APC activity, cells were arrested in mitosis by overexpressing the mad2⁺ gene. Overproduction of Mad2 results in the accumulation of cells with short metaphase spindles but which do not initiate cytokinesis(He et al., 1997). mad2⁺ was overexpressed in myo2-gc cells using the full-strength nmt1⁺ promoter for 16 hours. Aseptate cells with condensed chromatin, typical of a metaphase arrest(Fig. 2C arrow heads), did not possess a ring of Myo2 (Fig. 2C, left panel), demonstrating that a stable CAR forms post metaphase. We confirmed these findings using a different method of removing functional APC from myo2-gc cells by shut off of the essential APC component, Lid1 (Chang et al.,2001). Cells lacking Lid1 arrested without a CAR (data not shown).

We next addressed the dependency of Myo2 ring formation on the actin and microtubule cytoskeletons. First, we examined the requirement for actin by using the actin depolymerising drug latrunculin B(Gachet et al., 2001). myo2-gc cdc25-22 cells were synchronised by temperature block and released in the presence or absence of 10 μM latrunculin B, a concentration that completely depolymerises actin (Fig. 3A,C). Whereas Myo2 was recruited to the cell equator in the presence of latrunculin B at the same time as in control cells, it accumulated as a diffuse punctate band at the medial cell cortex rather than forming a distinct ring (Fig. 3D),consistent with previous reports (Naqvi et al., 1999; Motegi et al.,2000). Interestingly, the deposition of septal material continued in latrunculin-B-treated cells, also in a punctate manner, with the spots of septal material corresponding to foci of Myo2(Fig. 3E). Thus, the timing of Myo2 recruitment to the cell equator and subsequent deposition of septal material is actin independent but CAR assembly is actin dependent. These results also further emphasise the intimate relationship between the CAR and the positioning of the cytokinetic septum.

We next examined the effect of depolymerising microtubules on Myo2 ring formation. Synchronised myo2-gc cdc25-22 cells were arrested and released in either latrunculin B, TBZ, MBC or DMSO as a solvent control. As described above, Myo2 was recruited to the cell equator after 40 minutes in both the control and latrunculin B cultures (albeit the rings were not formed in latrunculin B) (Fig. 4A,B). By contrast, microtubule depolymerisation resulted in a delay of 90 minutes in the appearance of Myo2 rings (Fig. 4C,D). The fact that no such delay was observed in latrunculin,together with the fact that a similar delay was observed in both MBC and TBZ,demonstrate that this is a microtubule-specific phenomenon. The delay in the appearance of binucleate cells in Fig. 4D is caused by the activation of the spindle orientation checkpoint (Gachet et al.,2001).

To eliminate the possibility that the delay in Myo2 ring formation was a consequence of the cdc25-22 mutation, we attempted to repeat the experiment in a strain possessing the wild-type copy of the cdc25⁺ gene. However, size-selected wild-type G2 cells released into TBZ do not attain the critical mass required for entry into mitosis. As an alternative, myo2-gc cells were transiently arrested in S phase using the DNA synthesis inhibitor hydroxyurea (HU) and then washed into fresh medium in the presence or absence of TBZ(Fig. 6A). Cells continue to grow in HU and therefore attain the critical mass for mitotic entry prior to release into TBZ. HU-blocked cells released into TBZ exhibited a 40 minute delay in CAR formation, as visualised by the accumulation of Myo2-GFP rings. Thus the delay in CAR formation and subsequent septation is not an effect of the cdc25-22 mutation but, rather, indicates the existence of a checkpoint mechanism acting at an early stage of mitosis.

There are two possible reasons why a delay in CAR formation is induced when microtubules are depolymerised. The first is that components of the CAR are delivered to the cell equator along cytoplasmic microtubules(Bezanilla et al., 2000). Thus,in the absence of microtubules, components of the CAR are localised to the cell equator by a less efficient mechanism, such as diffusion. Alternatively,the delay could be caused by a checkpoint mechanism that prevents CAR formation from occurring if spindle formation is compromised. One way to distinguish between these possibilities would be to identify proteins required to maintain the delay. As CAR formation occurs early in anaphase, obvious candidates are proteins involved in the spindle assembly checkpoint. We therefore created a myo2-gc cdc25-22 strain in which the gene encoding for the spindle assembly checkpoint protein Mad2 was replaced with the ura4⁺ gene (myo2-gc cdc25-22 mad2Δ). When these cells were synchronously released into mitosis in the presence of DMSO, cytological events such as CAR formation, nuclear separation and septum formation occurred with the same timing as in mad2⁺strains (Fig. 5A-C). However,when cells were released into TBZ (Fig. 5D) or MBC (Fig. 6E), the delay was alleviated, thus demonstrating the existence of a Mad2-dependent checkpoint. Synchronised mad2Δ strains showed a marked reduction in viability in TBZ compared with the equivalent mad2⁺ strain (Fig. 5B,D). This is explained by the earlier appearance of cells with septa and unseparated nuclei, the typical `cut' phenotype(Hirano et al., 1986;Yanagida, 1998).

Having established that the CAR assembly checkpoint is Mad2 dependent, we next determined whether the checkpoint was also disrupted in cells lacking other spindle checkpoint components. The experiment was therefore repeated in a strain bearing a mutation in the cdc16 gene, which encodes an essential component of the GAP complex that regulates the onset of cytokinesis and subsequent septum formation. In the absence of functional Cdc16, cells go through multiple rounds of unregulated septum formation(Minet et al., 1979;Fankhauser et al., 1993). Cells bearing the cdc16.116 temperature-sensitive mutation were arrested in interphase using hydroxyurea for 3 hours, then washed in fresh medium lacking hydroxyurea and raised to the restrictive temperature of 36°C in the presence or absence of TBZ. In contrast to the delay observed in cells incubated at the permissive temperature (data not shown), a reduced delay was seen at 36°C, and cells went on to accumulate multiple septa with similar timing to that observed in controls(Fig. 6B). Thus, both Mad2 and Cdc16 contribute to the observed delay in both CAR and septum formation upon microtubule depolymerisation. Using similar strategies we determined that the CAR formation checkpoint is also dependent on Zfs1, which is involved in the fission yeast septation checkpoint(Beltraminelli et al., 1999)(Fig. 6D). To determine whether these proteins were functioning in a single pathway, we created a myo2-gc strain lacking both zfs1⁺ and mad2⁺. This strain demonstrated cytokinetic defects even at 25°C and failed to synchronise normally using the cdc25-22allele and therefore the delay in CAR formation could not be compared with other strains. Further examination revealed synthetic lethality between the mad2Δ and zfs1Δ alleles, suggesting that Mad2 and Zfs1 function on separate pathways (data not shown).

We next investigated whether the relocalisation of actin from the cell ends to the cell equator occured normally in TBZ. myo2-gc cdc25-22 cells were synchronised and released in the presence of DMSO or TBZ, and actin imaged using rhodamine-phalloidin. Medial recruitment of actin showed a delay similar to that seen with Myo2 - actin remaining at the cell tips longer when microtubules were absent - with no observable Myo2 localisation(Fig. 7A).

We were interested to examine whether components of the medial ring, which recruit to the cell equator prior to Myo2, were also effected by microtubule depolymerisation. One such protein is Dmf1/Mid1, which precedes actin at the division site and is required for the correct position and orientation of the CAR (Chang et al., 1996;Sohrmann et al., 1996). A strain was created bearing a gfp⁺-tagged copy of the mid1 gene in combination with the cdc25-22 allele. Cells were synchronised by temperature arrest and release and allowed to enter mitosis in the presence or absence of TBZ(Fig. 7B) or MBC (data not shown). In control cells, Dmf1/Mid1 was exported from the nucleus to the medial ring early in mitosis and persisted there until septation(Fig. 7C)(Sohrmann et al., 1996). In the presence of the anti-microtubule drugs, however, Mid1-GFP remained in the nucleus in ∼81% of cells (Fig. 7B). In those minority of cells in which Mid1-GFP did exit the nucleus, the protein failed to form a distinct ring but rather remained as a diffuse band at the cell cortex (Fig. 7D).

Having established that the recruitment of not only Myo2 but also actin and Dmf1/Mid1 to the medial cortex is subject to a microtubule checkpoint, we examined the effect of microtubule drugs on the localisation of possible regulators of these proteins. One such protein is the polo protein kinase Plo1, which controls both actin and Dmf1/Mid1 ring formation(Ohkura et al., 1995;Bähler et al., 1998). Plo1 localises to the SPB in a cell-cycle-specific manner. SPB staining is first visible at the start of mitosis, the intensity reducing as anaphase progresses(Bähler et al., 1998;Mulvihill et al., 1999). plo1-gfp cdc25-22 cells(Bähler et al., 1998) were synchronised by temperature shift and allowed to enter mitosis either in the presence or absence of TBZ (Fig. 8A). Plo1 recruited to the SPB with similar timing in both conditions. However, in TBZ a reduced amount of Plo1 was recruited onto the SPB's, the Plo1-GFP signal being ∼20% of that seen in control cells(Fig. 8B,C). Whereas in wild-type cells Plo1 localised transiently to the SPBs, with the protein becoming less visible as anaphase progressed, in TBZ the reduced Plo1-GFP signal persisted long after the control signal had disappeared(Fig. 8A). As cells began to leak through the microtubule checkpoint and septate, this reduced level of Plo1 persisted at the SPBs.

As Plo1 is a true SPB component and binds directly to SPBs rather than the minus ends of microtubules (Mulvihill et al., 1999), we next determined whether the recruitment of the remaining population of Plo1 to the SPB was influenced by the spindle checkpoint mechanism. The above experiment was therefore repeated in a plo1-gfp mad2Δ cdc25-22 strain. In contrast to the situation in a mad2⁺ background, Plo1 intensity in cells lacking microtubules increased to a similar amount to the control(Fig. 8E,F). Interestingly,Plo1 seemed to be associated with the SPB for longer in the absence of Mad2 than in its presence and was seen to persist for longer at the SPB in DMSO control cells, even those which had completed mitosis(Fig. 8E), suggesting that Mad2 may have a role in regulating Plo1's association with the SPB. Consistent with this, in plo1-gfp cells overexpressing mad2⁺,Plo1 failed to recruit fully to the SPB(Fig. 8G), with only ∼10%of the normal SPB associated Plo1 intensity being observed. Plo1 is recruited prematurely to the SPB in strains bearing the stf1.1 allele, a dominant mutation in the gene encoding for the essential SPB component Cut12 and which suppresses the cdc25-22 mutation(Bridge et al., 1998;Mulvihill et al., 1999). No effect on the length of the delay of CAR formation was observed in stf1-1 (data not shown).

Finally, we investigated whether the localisation of components of the septation initiation network, which act downstream of Plo1(Mulvihill et al., 1999;Tanaka et al., 2001), were effected by microtubule disruption. A cdc25-22 strain containing a GFP-tagged copy of the gene encoding the Cdc7 protein kinase was synchronised in G2 and released either in the presence of absence of TBZ. As originally described by Sohrmann et al., Cdc7 localised to both SPBs early in mitosis but was associated with only one pole during anaphase(Sohrmann et al., 1998). At the end of mitosis, no SPB-associated Cdc7 was visible. Interestingly,Cdc7-GFP was seen to recruit to the SPB coincident with Myo2 ring formation(Fig. 9A). However, in TBZ(Fig. 9B) or MBC (data not shown), Cdc7 recruitment to the SPB was delayed to the same extent as Myo2 ring formation (i.e. SPB bound Cdc7 became visible coincident with the appearance of Myo2 rings). These results demonstrate that the SIN components,which act downstream of Plo1 and upstream of Myo2 activity, are also affected by the CAR formation checkpoint.

The G2/M transition in fission yeast is associated with a major reorganisation of the cytoskeleton. Interphase cytoplasmic microtubules depolymerise at the onset of mitosis and the mitotic spindle begins to form between two juxtaposed duplicated SPBs(Hagan and Hyams, 1988). Actin patches and cables break down during early mitosis to be reorganised with myosin to form a ring at the cell equator, where they associate with many other proteins (Marks and Hyams,1985; Le Goff et al.,1999a). Mis-positioning of the nucleus by disrupting cytoplasmic microtubules results in the mis-orientation of the CAR, demonstrating that an intact microtubule cytoskeleton is required to coordinate the position of the spindle with the CAR and hence the division plane(Chang et al., 1996).

In higher eukaryotes, regulation of type II myosins is by post-translational modification of their light chains(Satterwhite et al., 1992). This does not appear to be the case in S. pombe. Both myosin II light chains are essential for cytokinesis. Cdc4, the essential light chain(Naqvi et al., 1999), is a phosphoprotein but the timing of cytokinesis unchanged in cdc4mutants in which the phosphorylation sites are mutated(McCollum et al., 1999). Cells lacking the regulatory light chain, Rlc1, grow normally at higher temperatures but display cytokinetic defects at lower temperatures. These can be almost totally suppressed by removing the second IQ domain within the Myo2 neck(Naqvi et al., 2000;Le Goff et al., 2000). Whether post-translational modifications of Rlc1 are important for Myo2 function (as is the case of myosin IIs in most non-muscle cells and in smooth muscle)remains an open question. However, S. pombe contains no obvious myosin light chain kinase (MLCK), and cells are insensitive to the MLCK inhibitor ML-7 (D.P.M. and J.S.H., unpublished). Rather, recruitment of Myo2 to the CAR depends upon phosphorylation of residues within the tail domain(Mulvihill et al., 2001).

Proteins essential for the correct timing of CAR formation in S. pombe are components of the SIN pathway(McCollum and Gould, 2001). Spg1 and Plo1 inappropriately drive ring formation when overproduced in interphase cells (Ohkura et al.,1995; Schmidt et al.,1997). Recruitment of Myo2 to the CAR is dependent on functional Cdc7 protein kinase (Mulvihill et al.,2000; Mulvihill et al.,2001), a key component of the SIN, which acts downstream of both Plo1 and Spg1 (Sohrmann et al.,1998; Mulvihill et al.,1999; Tanaka et al.,2001). Another component of the SIN is Cdc16, the fission yeast homologue of the spindle assembly checkpoint protein Bub2. Inactivation of Cdc16, which forms part of the Spg1 GAP, drives CAR formation and septation(Fankhauser et al., 1993;Furge et al., 1998;Sohrmann et al., 1998;Cerutti and Simanis, 1999;Mulvihill et al., 2000).

In this report we have carried out a detailed analysis of the timing of Myo2 ring formation and the factors that regulate it. Previous studies of the timing of Myo2 ring formation (Kitayama et al., 1997; Bezanilla et al.,2000; Motegi et al.,2000) were carried out using Myo2 driven from a heterologous promoter, in the presence of the native protein. Using the myo2-gcstrain in which Myo2-GFP is under the control of its own promoter and is the sole source of Myo2 in the cell we have established that Myo2 forms a stable ring at the cell equator at the onset of anaphase. CAR formation is delayed in response to microtubule depolymerisation, and this delay is dependent on a functional spindle assembly checkpoint. Finally, we provide evidence to suggest that the recruitment of regulators of CAR formation to the SPB may provide the signal for Myo2 ring formation.

The timing of Myo2 ring formation was examined by introducing GFP-tagged SPB and centromere markers into myo2-gc. An intact Myo2 ring was only observed in cells that had a mitotic spindle and separated centromeres. Thus CAR formation is coupled to anaphase onset. This finding was further supported by the fact that myo2-gc cells arrested in metaphase by overexpressing the SAC component Mad2 accumulated as aseptate cells lacking a Myo2 ring, further linking CAR formation to anaphase commitment. The CAR persists in an uncontracted state throughout the remainder of mitosis and only contracts at the end of telophase, marking the onset of cytokinesis.

The timing of Myo2 ring formation was dependent upon the presence of a functional microtubule cytoskeleton. Cells allowed to enter mitosis in the presence of the antimicrotubule drugs TBZ or MBC showed a significant delay in CAR formation. No such delay was observed in latrunculin B, although Myo2 was not incorporated into a ring in the absence of actin. It has previously been demonstrated that execution of the G2/M transition in the presence of TBZ has no effect on the recruitment of Cdc2 and Cdc13 (which together form MPF in fission yeast) to the SPB nor on chromosomes condensation(Alfa et al., 1990). The same study also showed that TBZ delayed the activation of MPF activity, normally associated with mitotic entry, by 2 hours. Myo2 ring formation is also delayed by TBZ but to a lesser extent, suggesting that CAR formation is not dependent upon Cdc2 activity. This is not unexpected as overexpressing the polo-like kinase, Plo1, induces CAR formation and eventual septation without an associated increase in Cdc2 activity(Ohkura et al., 1995).

Using cold-sensitive tubulin mutants, CAR formation has been demonstrated to occur independently of the presence of a mitotic spindle(Chang et al., 1996). However,these experiments were carried out in asynchronous cultures, so the delay in CAR formation described in this study would not have been observed. Beltraminelli et al. have shown that the appearance of actin rings and septa were delayed when synchronous populations of nda2-KM52 and nda2-KM52 zfs1Δ cells were incubated at the restrictive temperature, thus demonstrating that the CAR checkpoint exists in cold-sensitive tubulin mutants(Beltraminelli et al., 1999). Thus, CAR formation is independent of Cdc2 activity and can occur in the absence of spindle formation.

What provides the signal for CAR formation and subsequent septation? Data presented here suggest that the localisation of the polo kinase Plo1 as well as Cdc7 (an essential component of the SIN pathway) to the SPBs are key events. This is consistent with the findings that Plo1 has an essential role in promoting actin ring formation (Ohkura et al., 1995). In a normally dividing culture, Plo1 is recruited to the SPB at the onset of mitosis and is, indeed, the earliest mitotic event in fission yeast demonstrated to date(Mulvihill et al., 1999). However, when cells enter M phase in TBZ, Plo1 is recruited to the SPB but to only ∼20% of its normal level. Hence, a subfraction of Plo1 localises to the SPB in the absence of MPF activity, further suggesting that this may be one of the earlier steps in the commitment to mitosis. As with Plo1, Cdc2 and Cyclin B localise to the SPB at the normal time in the presence of TBZ but fail to delocalise until MPF activity increases(Alfa et al., 1990). Thus, MPF may provide the signal to delocalise regulators of CAR formation from the SPB.

Intriguingly, recruitment of Cdc7 to the SPB always precedes Myo2 ring formation. This occurs in normally dividing cells, as well as cells in which CAR formation has been delayed by microtubule depolymerisation. Thus,localisation of Cdc7 to the SPB may also be an essential prerequisite for CAR formation. We have previously shown that Cdc7 activity is required for the recruitment of Myo2 to the CAR but whether SPB-bound Cdc7 is the active protein remains to be discovered(Mulvihill et al., 2001). Importantly, Cdc7 localisation is downstream of Plo1 activity(Mulvihill et al., 1999).

Microtubule depolymerisation in pre-mitotic cells caused a delay in the appearance of not only Myo2 but also actin and Mid1/Dmf1. This is consistent with the fact that, like Myo2, Mid1 rings form after centromere separation. No Mid1 rings were observed in cells overexpressing Mad2, rather Mid1 formed a diffuse band at the cell cortex adjacent to the nucleus (data not shown). We were initially concerned that microtubule depolymerisation with TBZ or MBC might inhibit CAR formation indirectly by affecting the actin cytoskeleton. However, parallel experiments with latrunculin B showed that CAR components appeared at the cell equator with normal timing, albeit that the CAR was not assembled in the absence of actin. Further, actin is not depolymerised by TBZ under our experimental conditions but remains at the cell tips. Thus, the observed delay in CAR formation is a microtubule-specific phenomenon. An interesting aside of the latrunculin experiments was the observation in latrunculin-treated cells that patches of septal material formed on the outside of the cell membrane coincident with the location of patches of Myo2 on the inside of the cell membrane. Thus, independent of actin, Myo2 appears to link to the cell membrane and, more specifically, to the intacellular domain of the glucan synthases that lay down the division septum.

To further characterise the TBZ-induced delay in CAR formation we attempted to identify proteins required to maintain the delay. Because of the intimate relationship between mitosis and cytokinesis, we looked initially at components of the spindle assembly checkpoint. The TBZ delay was reduced in cells lacking Mad2. However, since Mad2 is involved in microtubule-based processes other than spindle assembly in S. pombe(Petersen et al., 1998), we also investigated the situation in mutants of Cdc16 which provides the link between spindle integrity and the SIN(Fankhauser et al., 1993). As with Mad2, functional Cdc16 was required to maintain the full TBZ delay. This suggests that the delay is not caused by problems in the recruitment of CAR components to the cell equator along microtubule tracks(Bezanilla et al., 2000) but,rather, that CAR formation is downstream of the spindle assembly checkpoint. Consistent with these findings, PtK cells injected with anti-Mad2 antibodies are capable of forming a cytokinetic furrow independently of a pre-anaphase spindle (Canman et al.,2000).

To better understand the relationship between the SAC and CAR formation, we examined the effect of TBZ on Plo1 recruitment in the presence and the absence of Mad2. Strikingly, the reduction in recruitment of Plo1 to the SPBs in response to TBZ was abolished in mad2Δ cells. Increasing the cellular level of Mad2 also reduced the SPB recruitment of Plo1. The localisation of Plo1 to the SPB is influenced by more than one SPB protein(Mulvihill et al., 1999), and our findings suggest that checkpoint proteins play a direct role in this process. Whether the TBZ-sensitive/Mad2-dependent and TBZ-insensitive/Mad2-independent Plo1 subpopulations serve different roles remains to be explored. However, it is interesting to note that two distinct peaks of Plo1 activity, the first being ∼20% of the activity of the later,major peak, are observed when highly synchronised cells enter M phase(Tanaka et al., 2001). From the results presented here, we conclude that the Plo1 subpopulation is involved in signalling to the CAR, the latter Plo1 being required for other functions. The TBZ-induced delay in the association of Cdc7 with the SPB is consistent with such a view. Cdc7 lies downstream of Plo1(Mulvihill et al., 1999) and upstream of Myo2 assembly into the CAR(Mulvihill et al., 2001). A diagrammatic representation of the relationship between the various elements discussed here is shown in Fig. 10. Although there is no direct evidence to support the idea that SPB-association equates to Plo1 activity, we see localisation as necessary to initiate the pathway leading to both CAR formation and septation. A feedback loop from the SIN proteins back to Plo1 as suggested by Tanaka et al. provides a way of downregulating the pathway(Tanaka et al., 2001). Hence,TBZ only induces a delay and not a true cell cycle arrest. Evidence of a link between the Pololike kinases, the spindle assembly checkpoint and cytokinesis was recently forthcoming from studies in budding yeast showing that the polo-like kinase Cdc5p phosphorylates and activates Bfa1, the partner of Bub2(Hu et al., 2001). An as yet unexplained feature of our model is that changes in the localisation of both Plo1 and Cdc7 occur early in mitosis, as indeed does CAR formation (this report). CAR contraction and septum formation on the other hand occur at the end of mitosis, a gap of some 30 minutes in cells growing at 25°C. Further studies are required to determine what other events must occur during this window to terminate the sequence of events set in train by Plo1's redeployment.

We thank Iain Hagan, Dannel McCollum, Tomohiro Matsumoto, Shelley Sazer,Viesturs Simanis and Mitsuhiro Yanagida for strains and plasmids. We also thank Yannick Gachet and Jeremy Bentham for helpful comments on the manuscript. This study was supported by the Wellcome trust, grant number 062095.