The fission yeast Chs2 protein interacts with the type-II myosin Myo3p and is required for the integrity of the actomyosin ring

Cytokinesis is a crucial event, regulated temporally and spatially, through which a mother cell divides into two daughters. In most eukaryotes, cytokinesis is carried out by the formation and ingression of a cleavage furrow. The plasma membrane invaginates owing to its association with an underlying actomyosin-based structure called the contractile actomyosin ring (CAR), which assembles at the onset of cytokinesis along a plane perpendicular to the elongating mitotic spindle (reviewed in Balasubramanian et al., 2004; Glotzer, 2005). The fission yeast Schizosaccharomyces pombe has emerged as a powerful model organism for studying cytokinesis because it is amenable to genetic manipulation, favourable for microscopic analysis, and it carries out cytokinesis similarly to animal and protozoan cells.

One of the fields that has attracted most attention in recent years is how cells assemble and constrict the medial ring. In S. pombe, as in animal cells, the presence of a functional CAR is essential for survival (Balasubramanian et al., 2004; Krapp et al., 2004; Wolfe and Gould, 2005). The assembly of the contractile ring occurs in several steps, starting at the onset of cytokinesis, when components of the CAR are recruited to the cell cortex overlying the nucleus in a Mid1p-dependent manner. Then, the type-II myosin Myo2p, its regulatory light chains (Rlc1p and Cdc4p) and Rng2p reach the midzone of the cell. They are followed by the arrival of Cdc12p and Cdc15, which recruit other proteins that promote actin polymerisation and thereby contribute to ring assembly (Balasubramanian et al., 2004; Wu et al., 2003). This primary actomyosin ring matures through the assembly of other proteins into the ring, as is the case of the other type-II myosin present in S. pombe, Myo3p/Myp2 (henceforth called Myo3p).

The next steps of cytokinesis – ring contraction and septum formation – require a conserved signalling pathway called the septation initiation network (SIN), which is organized at the spindle pole body (reviewed by Krapp et al., 2004). Unlike animal cells, yeasts synthesize a division septum behind the ring as it constricts, generating cell wall material between daughter cells (Bi, 2001; Feierbach and Chang, 2001; Guertin et al., 2002).

Over the past few years, a considerable body of information has been obtained about contractile ring assembly and function. However, the membrane-trafficking events required to deliver new membrane to the cell surface during ring constriction are less well characterized. In addition to the delivery of the membrane to compensate for the expanding plasma membrane surface, membrane traffic during cytokinesis could also mediate the delivery of proteins that control the ingression of the cleavage furrow and/or cell separation. We were therefore prompted to gain more information about the relevance of membrane proteins present at the midzone during cytokinesis by studying the role of the S. pombe Chs2p in this process. chs2⁺ codes for a transmembrane protein with significant similarity to the S. cerevisiae Chs2p, but that lacks most of the residues considered to be essential for chitin synthase activity (Martin-Garcia et al., 2003). In S. cerevisiae, Chs2p is the chitin synthase enzyme responsible for the synthesis of the primary septum (Shaw, 1991). It has not been possible to establish the presence of chitin in S. pombe (Arellano et al., 2000). In a previous work we showed that S. pombe Chs2p localises to the growing edge of the septum, that it lacks chitin synthase activity, and that it leads to abnormal cytokinesis when overexpressed (Martin-Garcia et al., 2003). Here, we show that Chs2p is functionally related to proteins present at the contractile ring and that it collaborates with them in maintaining CAR integrity during the late stages of contraction. Additionally, we show that Chs2p constitutes a physical link between the membrane and the CAR.

chs2⁺overexpression from the pREP41Xchs2⁺ plasmid leads to abnormal septation, with about 50% of the cells showing several septa that define cell compartments in which 0, 1 or 2 nuclei can be observed (Martin-Garcia et al., 2003). To better understand the effect of chs2⁺ overexpression in cytokinesis, we analysed, by direct fluorescence microscopy, several GFP-tagged versions of proteins that are present at the CAR. In the wild type (WT), Cdc15p formed a ring at the midzone (Fig. 1A). When chs2⁺ was overexpressed, Cdc15-GFP was observed at the septal region in 88% of the cells under division (n=120), although in 53% of them the protein did not seem to assemble properly into a ring, showing an asymmetric distribution (24% of the cells; see asterisk in Fig. 1A) or a very strong fluorescence signal that indicated an abnormal accumulation of the protein (12% of the cells; arrow in Fig. 1A). It was even possible to observe delocalised spots in the cytoplasm or near the cell cortex (17% of the cells; not shown). When Cdc4-GFP was used to monitor the CAR in chs2⁺-overexpressing cells, it was observed that 26% of the mitotic cells showed two rings that were very close together (arrowhead in Fig. 1B); 16% of the cells showed an abnormal localisation of the protein after septum completion (diamond in Fig. 1B), and 9% of the cells exhibited a strong fluorescent signal at the equatorial area (not shown). In 49% of the cells undergoing division (n=131), Cdc4-GFP was properly localised, in agreement with the number of cells in the culture that showed a normal morphology. We confirmed these findings by confocal microscopy. Using three dimensional (3D) reconstructions, in a significant number of cells (23%; n=30) we observed a structure consisting of a normally shaped Cdc4-GFP ring with a more diffuse fluorescent signal inside. Further analysis of each plane of the z-series indicated that these structures corresponded to two parallel rings located very close to each other (one contracting and the other still located at the cell periphery, see Fig. 1C). In some cases, one of the rings seemed to be slightly tilted with respect to the other, giving the appearance of an un-zipped ring (Fig. 1C). It can be concluded that chs2⁺ overexpression interferes with normal CAR assembly and/or function.

In a WT strain, Chs2p localises to the midzone of the cells and forms a ring that contracts preceding septum synthesis (Martin-Garcia et al., 2003) (Fig. 2). In order to gain further information about the step of cytokinesis in which Chs2p performs its function, we investigated which proteins are required for Chs2p to localise properly. First, we wondered whether Chs2p localisation might depend on actin. To address this question, we synchronised cells carrying an integrated Chs2-GFP protein and the cdc25-22 mutation (see Materials and Methods). Cells were released from the block in the presence of 100 μM latrunculin A or the solvent DMSO, and samples were taken every 15 minutes over a total of 2 hours. Rhodamine-phalloidin staining was used to assess the distribution of actin in these samples. After 75 minutes at the permissive temperature in the presence of DMSO, the cells were able to form Chs2p rings that co-localised with actin (Fig. 2A). However, in the presence of latrunculin A Chs2p could not be detected at the equatorial zone either at this time-point or later. We only observed some aberrant fluorescence, caused by formaldehyde fixation, which gave the appearance of clusters of Chs2p in the cytoplasm. When latrunculin A was added to an asynchronous culture, we observed that the cells in which Chs2-GFP had reached the equatorial area before treatment showed a properly localised protein even after 20 minutes of incubation in the presence of the drug (Fig. 2B). It may be concluded that Chs2p localisation to the CAR, but not its maintenance there, depends on F-actin.

We then wondered which proteins might be involved in the delivery of Chs2p. Some cell wall enzymes are delocalised in myo52 mutants, which are defective in a type-V myosin involved in the polarised delivery of cell wall components (Motegi et al., 2001; Mulvihill et al., 2006; Win et al., 2001). Since Chs2p is a chitin synthase-like protein, we wished to know whether it was delivered to the septal region through this mechanism. The myo52Δ cells show growth and morphology defects at 37°C, although at 25°C polarity and septation defects are observed in a number of cells (Win et al., 2001). When a myo52Δ strain carrying the Chs2-GFP protein mutant was incubated at 25°C, Chs2p was delocalised in 19% of the septating cells (n=110), most of them exhibiting morphological defects (not shown). At 37°C, Chs2p was delocalised in 41% of cells under division. The asterisk in Fig. 2D points to the septal region of a cell in which a new septum had started to form but the Chs2-GFP could not be observed. Additionally, in 32% of the cells where the protein was localised to the equatorial area fluorescence seemed to be weaker than in the WT, or did not seem to be properly incorporated into a ring (arrowhead in Fig. 2D). We confirmed these results by confocal microscopy (not shown). Thus, Myo52p seems to be important for the delivery and even for the assembly and/or stability of the Chs2p ring.

The exocyst is a multiprotein complex, important for the targeting and fusion of Golgi-derived vesicles at the plasma membrane, which has been implicated in cell separation and endocytosis in S. pombe (Gachet and Hyams, 2005; Martin-Cuadrado et al., 2005; Wang et al., 2002). To determine whether Chs2p targeting to the septum was dependent on this complex, we observed the distribution of the Chs2-GFP fusion protein in an exo70Δ strain. At 37°C the protein was located throughout the cytoplasm in 56% of the septating cells (n=114; Fig. 2E). Additionally, in cells with a proper Chs2-GFP localisation (44%), the intensity of the signal seemed weaker than in the WT strain (arrowhead in Fig. 2E). Even at 25°C Chs2p was not properly localised in 43% of the septating cells (not shown). Finally, in a percentage of cells that had completed septum formation, Chs2p was still observed in the medial region, a phenomenon that was never detected in the WT strain. We observed this either at 25°C or at 37°C, being more frequent at the lower temperature (18% of septated cells; n=50; see diamond in Fig. 2E). Similar results were obtained using a sec8-1 strain (not shown). Therefore, the exocyst is involved in the delivery of Chs2p to the equatorial region and might be also involved in the turnover of this protein at this location.

We then analysed the dependence of Chs2p localisation on several proteins involved in CAR assembly and contraction. We first studied Chs2-GFP localisation in mutants affected in the myosin type II components of the CAR (light chains Cdc4p and Rlc1p and heavy chains Myo2p and/or Myo3p). In cdc4-8 and rlc1Δ mutants at the restrictive temperature (37°C and 20°C, respectively) Chs2p was not localised to the cell division site in 92% and 86% of the cells, respectively (n=150; Fig. 2F). In the myo3Δ strain incubated at 25°C (Fig. 2F) and myo2-E1 strain incubated at 37°C (not shown) Chs2p was observed in some forming rings (82% and 70% respectively; n=160), but a strong background of fluorescence and some fluorescent dots were also observed in the cytoplasm (see arrowhead in the corresponding panel of Fig. 2F). In a double myo2-E1 myo3Δ mutant at 37°C Chs2p was never observed at the medial region (Fig. 2F). A cdc15-140 mutant strain was used to assess the dependence of Chs2p localisation on Cdc15p, a protein that plays a critical role in the assembly and maintenance of the CAR. Fig. 2G shows that when this mutant was arrested at 37°C Chs2p was dispersed throughout the cytoplasm. All these results show that the existence of an assembled and functional CAR is required for Chs2p to reach the equator of the cell and to assemble into a ring.

We then analysed Chs2p localisation in mutants affected in the SIN pathway. As shown in Fig. 2H, when the SIN mutant cdc11-119 was arrested at the restrictive temperature, Chs2-GFP was dispersed throughout the cytoplasm. The same result was obtained using a cdc14-118 strain (not shown). In contrast, in a cdc16-116 mutant, in which the SIN pathway is hyperactive, Chs2-GFP was observed at the leading edge of the growing septa (asterisks in Fig. 2H). Interestingly, in this strain Chs2p was abnormally maintained in all mature septa (arrowheads in Fig. 2H). These results show that the presence of Chs2p at the medial ring depends on the activity of the SIN pathway.

In order to uncover the function that the Chs2p might be performing in cells under physiological conditions, we carried out a study of genetic interactions between chs2Δ and mutants affected in several steps of cytokinesis. We used mutants in essential proteins of the CAR such as profilin (cdc3-6), formin (cdc12-112), myosin light chain Cdc4p (cdc4-8), myosin heavy chain Myo2p (myo2-E1), IQGAP-related protein Rng2 (rng2-D5), PCH domain protein Cdc15p (cdc15-140) and β(1,3)-glucan synthase (cps1-191). We also used mutants in several SIN proteins (cdc11-119 and cdc16-116) and mutants in non-essential components of the CAR, such as myosin heavy chain Myo3p (myo3Δ), myosin regulatory light chain (rlc1Δ) and PCH domain protein Imp2p (imp2Δ). Finally, we analysed a triple mutant myo2-E1 myo3Δchs2Δ.

Significant genetic interaction was detected with strains cdc15-140 and cdc4-8, since the double cdc15-140 chs2Δ and cdc4-8 chs2Δ mutants were more thermosensitive than the parental strains and did not grow at 32°C (Fig. 3A). We also found an interaction with the imp2Δ mutant, but in this case the double imp2Δchs2Δ strain was more thermoresistant than the original imp2Δ strain and was able to grow at 32°C (Fig. 3A). To study these interactions closely, we analysed the morphology of the mutants by staining the cells with DAPI and Calcofluor. Cells of the cdc15-140 chs2Δ and cdc4-8 chs2Δ double mutants were similar to those of the single cdc15-140 or cdc4-8 strains (not shown). Single imp2Δ mutants underwent an aberrant cytokinesis that led to the production of numerous multiseptated and branched cells (73% of the culture; n=200; see Fig. 3A). Rhodamine-phalloidin staining revealed that in 16% of septating cells (n=40) the medial ring did not seem to fully disassemble after septum formation and was observed as an extra misplaced ring (Fig. 3A), as has been described previously (Demeter and Sazer, 1998). As detected by Calcofluor staining, in the double imp2Δchs2Δ mutant, cells were elongated but only a small number of cells underwent aberrant cytokinesis (21%; n=200; Fig. 3A). Rhodamine-phalloidin staining revealed actin patches at the septal region of imp2Δchs2Δ septated cells (Fig. 3A), although they formed misplaced rings in less than 1% of the septating cells (n=40).

We also analysed genetic interactions between chs2Δ and mutants in the type-II myosins. We found no significant difference in the growth of the single myo2-E1 or myo3Δ mutants and the double myo2-E1 chs2Δ or myo3Δchs2Δ strains at 25°C, 28°C, 32°C or 37°C. However, the triple myo2-E1 myo3Δchs2Δ mutant was not able to grow at 32°C (Fig. 3B) while the double myo2-E1 myo3Δ did grow at this temperature. Since Myo3p has been shown to be important for cytokinesis at low temperatures or in the presence of KCl, we looked for possible genetic interactions in these conditions. We analysed the growth of WT, myo3Δ, chs2Δ, and myo3Δchs2Δ strains on minimal medium supplemented with 1 M KCl incubated at 25°C. As shown in Fig. 3B, the sensitivity of the myo3Δ strain to KCl was exacerbated by the chs2Δ null mutation, pointing to a functional relationship between the two proteins. By Calcofluor staining we observed that the double myo2-E1chs2Δ mutant had a stronger defect in cytokinesis than the single myo2-E1 mutant, although it is also possible that these cells had a defect in cell separation after division. In the myo2-E1 strain 43% of the cells were septate. Of those, 38% of cells had more than one septum. In the myo2-E1chs2Δ mutant, 58% of the cells were septate, 46% had several septa. The enhancement of the defect in cytokinesis was stronger when the myo3Δ mutation was combined with the chs2Δ mutation. Cells having septa were present at 61% in the myo3Δ strain and at 93% in the myo3Δchs2Δ double mutant. The number of cells having more than one septum was 27% in the case of the single mutant and as high as 82% for the double mutant. According to the Calcofluor staining of cells grown at 28°C, over 50% of the septa in the myo3Δchs2Δ cells were wide and aberrant (Fig. 3B), whereas this kind of septum represented less than 10% in the myo3Δ cells.

We next wished to know whether the cytokinesis defect in the myo3Δchs2Δ mutant was a consequence of an altered CAR. To check this, we introduced an Rlc1-GFP fusion protein into these strains. Asynchronous myo3Δ cultures analysed by conventional fluorescence microscopy showed 3% of the septate cells (n=78) in which the green fluorescence did not completely overlap with Calcofluor staining (arrows in Fig. 4A), suggesting the appearance of a second ring very close to the first one. This phenotype was absent in the WT (Fig. 4A) and the chs2Δ strains (not shown) but was more frequent in the double myo3Δchs2Δ mutant, reaching a value of 15%. When observed under the confocal microscope, the Rlc1-GFP ring was slightly defective in a myo3Δ mutant, i.e. not completely circular as it is in the WT or the chs2Δ strains (see Fig. 4A). This defect was aggravated in the myo3Δchs2Δ strain, where the rings were misshapen and distorted, as if they were elastic or tensionless (Fig. 4A). Additionally, we confirmed that four out of 20 analysed cells exhibited an extra ring very close to the pre-existing one (see Movie 3 in supplementary material). These structures were not detected in the single myo3Δ mutant (n=20 cells).

It has been suggested that the Rlc1p ring does not contract in a myo3Δ mutant (Le Goff et al., 2000). However, when we performed time-lapse analysis using a conventional fluorescence microscope we found that this ring did contract in the myo3Δ mutant (see Fig. 4C), although more slowly than in the WT, taking into account the time required until the ring disassembles (compare the kinetics of ring contraction in the WT, in Fig. 4B, with that of cells 1 and 2, marked with arrowheads in Fig. 4C). Possibly, the different myo3Δ strain used in both laboratories could account for this difference. The chs2Δ mutant showed a similar rate of ring contraction and disassembly to that of the single myo3Δ mutant (Fig. 4E and Fig. S1 in supplementary material). In the double myo3Δchs2Δ strain, contraction and disassembly of the ring proceeded more slowly than in the WT and the myo3Δ strain (see cells 1 and 5, marked with arrows in Fig. 4D). This delay was exacerbated in cells that had several septa (see cell 3, marked with asterisks in Fig. 4D). Since cells with different rates of contraction in the same strain were observed, we calculated the rate of contraction for 30 cells in each strain and obtained an average rate of contraction for each mutant. The cells that did not show contraction of the F-actin rings were not included in the average. The result is shown in Fig. 4E (see also Movies 1 and 2 in supplementary material).

Apart from the delay in ring contraction, we detected some myo3Δchs2Δ cells where a ring disappeared without contracting (see cell 4, marked with a diamond in Fig. 4D) and cells where the Rlc1 ring seemed to contract and then expand again before disappearing in later frames (see cell 2, marked with a dot in Fig. 4D). The last phenomenon was also detected in some myo3Δ cells (see cell 1, Fig. 4C) and in a small percentage of chs2Δ cells (Fig. S1 in supplementary material). To study this phenotype in detail we performed confocal time-lapse experiments and observed that in a certain percentage of myo3Δ, chs2Δ and myo3Δchs2Δ cells, the rings did not seem to disassemble as in the WT strain – they did not form punctual dots – giving the appearance of expanding back to the cell periphery. In addition, in 26% of the myo3Δchs2Δ cells (n=70), a secondary ring was formed very close to the previous one, which was in contraction (Movies 2 and 3 in supplementary material). This was only observed in 2% of cells (n=70) from the single myo3Δ mutant. In most cases this new ring disappeared without contracting. All these results suggest that Chs2p collaborates with the type-II myosins in cytokinesis.

Since all the above results strongly suggest a functional relationship between Chs2p and the actomyosin ring during cytokinesis, we analysed whether there was a physical interaction between Chs2p and any of the CAR proteins. To analyse whether Chs2p interacted with the myosins, Myo3p or Myo2p, co-immunoprecipitation assays were performed in strains bearing the HA-tagged chs2⁺ gene under the control of the 41X version of the nmt1⁺ promoter and/or the GFP-tagged myo3⁺ or the myo2⁺ genes cloned in the pREP81X plasmid. We were unable to perform this experiment using native levels of the proteins because Chs2p is a protein present in the cells at an extremely low abundance (Wu and Pollard, 2005) and we were unable to detect it when the chs2⁺ gene was controlled by its own promoter even in a multicopy plasmid. However, when we expressed this gene from the 41Xnmt1⁺ promoter for 18 hours the cells did not show the phenotype of overexpression, indicating that the level of the protein was not overly high. First, we incubated cell extracts from the WT and the strains carrying the pREP41Xchs2⁺-HA plasmid, the 81Xmyo3⁺-GFP plasmid, or both, in the presence of monoclonal anti-GFP antibody. Then, western blot analyses were performed using anti-GFP or anti-HA antibodies. In parallel, total cell extracts from the same strains were analysed by western blot. As seen in Fig. 5A, Chs2-HA was detected in anti-GFP immunoprecipitates from the strain bearing both 41Xchs2-HA and 81Xmyo3⁺-GFP, but not from the control strains, pointing to a physical interaction between both proteins. The same result was obtained when the cell extracts were immunoprecipitated with the anti-HA antibody and developed with the anti-GFP (Fig. 5B). By contrast, we did not detect any Myo2-GFP protein associated with Chs2p-HA, although the Chs2-HA protein had been properly immunoprecipitated (Fig. 5C). The same result was obtained when immunoprecipitation was performed using anti-GFP antibody (not shown).

Since Myo3p and Chs2p interact physically, we wondered whether Chs2p might be required for the correct localisation of Myo3p. We used a strain in which the Myo3-GFP fusion protein, under the control of its own promoter, was integrated in the chromosome. It has already been described that Myo3p is asymmetrically distributed in the medial ring, giving a stronger signal at one side of the ring (Wu et al., 2003). We observed that this asymmetric distribution of the fluorescence in the medial ring was more evident in the chs2Δ mutant; in this strain, a strong spot of fluorescence was detected on one side of the ring but fluorescence on the other side of the ring had faded away (Fig. 5D). In order to estimate this different distribution in fluorescence of Myo3-GFP in WT and chs2Δ strains, we plotted the intensity of relative fluorescence against the distance along a line drawn along the ring over the average projection of a z-stack of images obtained with a confocal microscope. The result confirmed that in a chs2Δ strain the asymmetry in the fluorescence intensity of Myo3-GFP was stronger than in the WT (Fig. 5D).

In order to uncover the function that Chs2p might be playing in cytokinesis, we analysed some CAR proteins in the WT or chs2Δ strains by time-lapse microscopy. A Cdc15-GFP fusion protein was used to study the kinetics of ring contraction in both strains. A GFP-tagged Hht2p histone was used to visualize the progression of nuclear division so that the photographs could be compared at the same time points. First, we used a conventional microscope, observing that Cdc15-GFP remained longer in the contracted state in the chs2Δ strain than in the WT (Fig. 6A). A similar defect was observed using a Cdc4-GFP fusion protein (not shown).

We then analysed these strains using confocal microscopy and observed that in six out of ten imaged chs2Δ cells the Cdc15-GFP ring was a discontinuous structure in the very late stages of ring contraction, exhibiting a two-dot structure (not shown). The ring seemed to persist in this state until the time at which, in the WT, the ring had disassembled completely. To study what happened in the late stages of CAR contraction in the chs2Δ strain in more detail, we performed tighter time-lapse experiments in order to follow, by confocal microscopy, Cdc15-GFP contraction in the chs2Δ mutant and the WT strains. Since at this point we were more interested in studying ring morphology than the rate of contraction, strains carrying the Cdc15-GFP but not the Hht2-GFP fusion protein were used. We took z-series every 3 minutes; Fig. 6B shows lateral and frontal views of the 3D reconstruction of the cells or the rings, at specified time-points after the CAR had started its constriction. This allowed us to observe that a portion of the ring appeared to be separated from the main structure (Fig. 6B and Movie 4 in supplementary material); in the following steps the ring had adopted the appearance of a two-dot structure (confirming our previous observation, see above). Later, this double dot appeared as a unique dot of fluorescence. Finally, the ring took longer to disassemble completely than in a WT. We observed this behaviour in four out of 15 chs2Δ cells visualized by this method. Additionally, in another three out of those 15 chs2Δ cells, we detected the two-dot structure in contracting rings that showed a normal morphology during the initial steps of constriction. The two-dot structure was never observed in the WT.

All the above results suggested that Chs2p is important to strengthen or stabilise the actomyosin ring at the end of the contraction. We wondered whether the alteration in the CAR caused by chs2Δ deletion had any effect on septation kinetics. To explore this, we constructed a double cdc25-22 chs2Δ mutant and synchronised cells in G2/M (see Materials and Methods). Aliquots were taken every 15 minutes to count nuclei and complete septa. As shown in Fig. 6C, in the chs2Δ mutant septation was delayed by 15 minutes with respect to the control strain. According to the Calcofluor staining, the morphology of the septa was similar to that of WT septa. The delay in septation was not a consequence of a delay in resuming the cell cycle after the arrest, since both strains duplicated their nuclei with the same kinetics during the first nuclear cycle after the release (Fig. 6C). The same result was obtained when WT and chs2Δ mutants were synchronized by lactose gradients (not shown).

In this work we studied the relationship between the chitin synthase-like protein Chs2p and the actomyosin ring in fission yeast. Analysis of the proteins required for Chs2p localisation at the leading edge of the growing septa (Martin-Garcia et al., 2003) revealed that actin is required for Chs2p to reach the CAR, as it is for tropomyosin Cdc8p, UCS protein Rng3p and Myo3p (Wu et al., 2003). Myo3p also requires actin in order for it to remain at the CAR; however, Chs2p remains at the equatorial ring for more than 45 minutes after latrunculin A had been added to the medium, showing that it must be anchored to the ring through proteins other than actin and Myo3p. Myo2p, Cdc4p, Rlc1p and Cdc15p are able to reach the equatorial zone of the cell and to remain there in the presence of latrunculin A (Wu et al., 2003). Thus, the behaviour of Chs2p is different from that of all these proteins.

Myo52p is important for the proper delivery and assembly of Chs2p at the contractile ring, although there must be another mechanism involved in its delivery, since the protein was able to reach the cell division site in a percentage of myo52Δ cells. The exocyst complex must also be important for the targeting of Chs2p and/or the fusion of Chs2p-carrying vesicles to the equatorial membrane because this protein was delocalised in a significant proportion of cells in the exo70Δ mutant. Chs2p was still partially localized in a certain percentage of cells, probably because other subunits of this complex can compensate for its absence. In addition, Chs2p was abnormally maintained in the equatorial region of some cells that had already completed septum formation. The exocyst is involved in endocytosis (Gachet and Hyams, 2005). Thus, Chs2p disappearance from the medial ring by endocytosis could be impaired in the exocyst mutants. It can be concluded that the general system for polarised secretion is required for Chs2p protein to reach the equatorial ring.

Additionally, proper localisation at the ring requires the myosin component of the CAR. In S. pombe, there are two type-II myosin heavy chains: Myo2p and Myo3p. Deletion of myo2⁺ is lethal whereas the absence of myo3⁺ leads to alterations in cytokinesis under stress conditions. However, the fact that the double myo2E1 myo3Δ mutant is more defective for cytokinesis than the single myo2E1 indicates that the Myo3p protein must participate in cytokinesis under normal conditions (Bezanilla et al., 1997; Bezanilla and Pollard, 2000; Kitayama et al., 1997; May et al., 1997; Motegi et al., 1997; Motegi et al., 2000). Chs2p was partially localised in myo2-E1 or myo3Δ mutants, but completely delocalised in a myo2-E1 myo3Δ strain. Chs2p was also delocalised in the cdc4-8 and rlc1Δ mutants, affected in the myosin light chains, which are shared by both heavy chains. Chs2p localisation also depends on Cdc15p, a protein involved in actomyosin ring assembly and maintenance, which is also related to proteins such as PACSIN (Lippincott and Li, 2000), with functions in vesicular trafficking events. Finally, SIN signalling is also needed for proper Chs2p localisation. Furthermore, analysis of Chs2p localisation in a cdc16-116 mutant revealed that the disappearance of Chs2p from the midzone after septum completion requires the SIN signal to be turned off.

Genetic interactions with cdc15-140, cdc4-8, imp2Δ and double myo2-E1myo3Δ mutants, suggest a functional relationship between Chs2p and the CAR. Deletion of chs2⁺ in the myo3Δ background enhanced sensitivity to 1 M KCl, and deletion of chs2⁺ in a myo2-E1 myo3Δ mutant exacerbated thermosensitivity. The genetic interactions and the fact that Chs2p requires Myo2p and Myo3p for proper localisation and that in a chs2Δ strain Myo3p distribution at the ring is altered, suggest that Chs2p and the type-II myosins could be part of a multiprotein complex. In fact, we observed physical interaction of Chs2p with Myo3p.

Study of the ring in the myo3Δchs2Δ strain revealed that this structure had an abnormal shape and seemed to be tensionless. Probably, the lack of Chs2p, which provides a link between the plasma membrane and the ring, together with the lack of the myosin Myo3p, which collaborates with Myo2p to pull the membrane inward, would result in a weakened ring that would be unable to contract properly. Additionally, we have observed that the ring contracted more slowly in the myo3Δchs2Δ mutant than in the WT strain and that in some myo3Δchs2Δ cells there was a spurious secondary ring very close to the one driving septum formation. All these results would explain the increase in septation and the aberrant septa observed in this double-mutant strain. Our results support the idea that Chs2p and Myo3p must collaborate in the CAR contraction process. The fact that the double myo3Δchs2Δ mutant was more sensitive than the single mutants to KCl (Fig. 3) or to the calcineurin inhibitor FK506 (our unpublished results) suggests that both proteins could also collaborate under stress conditions. We also obtained some evidence to suggest that Chs2p must interact and/or collaborate with other proteins that function in septation, such as Myo2p, Cdc4p or Cdc15p.

To analyse the cytokinesis process in chs2Δ cells more closely, we followed the kinetics of CAR contraction by time-lapse experiments in this mutant and in the WT strain. In cells lacking Chs2p, the integrity of the ring was compromised during the last stages of contraction so it appeared to be partially disassembled (Fig. 6). Similar abnormal rings were also observed in the myo3Δchs2Δ strain (compare Movies 3 and 4 in supplementary material). One possible explanation for this phenotype could be that during late contraction the ring has to exert a considerable force to pull the membrane inwards and maintain the tension required for membrane ingression. The lack of an attachment point of the ring to the membrane could cause the observed partial breakages in the ring. In the chs2Δ cells, the ring remained in the contracted state longer than in the WT before its disassembly. This delay probably allows chs2Δ cells to repair the damaged ring through a mechanism (known as the cytokinesis checkpoint) that ensures completion of cytokinesis after minor perturbations. As a consequence, cells septate more slowly in the mutant (Fig. 6C).

Imp2p participates in the last stages of contraction, mediating disassembly of the ring (Demeter and Sazer, 1998). Overexpression of imp2⁺ leads to the appearance of secondary rings that are slightly displaced from the forming septum. We observed this phenomenon in the double myo3Δchs2Δ cells (Fig. 4A). As occurs with the chs2Δ mutant, the imp2Δ strain shows the strongest genetic interaction with cdc15 and cdc4 mutants, although the interaction is stronger for the imp2Δ mutant. In an imp2Δ mutant, actomyosin rings are more stable and do not disassemble properly. Therefore, suppression of the imp2Δ phenotype by deletion of the chs2⁺ gene would be in agreement with a role for Chs2p in the stabilisation of the CAR. Additionally, chs2⁺ overexpression led to the accumulation of some CAR proteins at the midzone (Fig. 1), as occurs in the imp2Δ mutant. All these data support the idea of the participation of Chs2p in CAR stability.

How does Chs2p contribute to maintaining the integrity of the CAR? Chs2p is predicted to have seven potential transmembrane regions. It is possible that this protein might play a passive role, anchoring some components of the CAR to the plasma membrane so that this structure is strengthened during contraction. However, it is also possible that Chs2p might play an active role in cytokinesis. Myo3p has been shown to be essential for the regulation of chloride ion homeostasis and cytokinesis by calcineurin (Fujita et al., 2002). We observed that growth of the double myo3Δchs2Δ mutants was slower than that of the single mutants on YEPD plates supplemented with FK506 or on minimal medium without calcium (our unpublished results). Myo3p has been proposed to perform some signalling for the septation machinery in response to environmental conditions (Bezanilla and Pollard, 2000; Mulvihill and Hyams, 2003). It is therefore possible that Chs2p might sense some external signal and act on the calcium-signalling pathway, which in turn would regulate cytokinesis through Myo3p and other CAR proteins.

Comparative analyses of the expression patterns of orthologue genes in budding and fission yeast have shown that cdc15⁺, imp2⁺, myo3⁺ and chs2⁺ belong to the same cluster as HOF1 (related to cdc15⁺ and imp2⁺), MYO1 (the type-II myosin in S. cerevisiae) and CHS2, pointing to a conserved role of these proteins in both organisms (Peng et al., 2005). Recently, a role for ScChs2p in actomyosin ring stability has been reported (VerPlank and Li, 2005). The question thus arises as to whether the Chs2 proteins from both yeasts might have similar structural roles in the cytokinesis process regardless of the catalytic activity. Interestingly, overproduction of ScChs2p in the fission yeast, which lacks chitin in the vegetative cell wall (Arellano et al., 2000), produces an alteration of cytokinesis similar to that produced by overproduction of the fission yeast Chs2p (our unpublished results). This suggests that both proteins might be able to interact with the same components of the septation machinery. By using conditional-expression plasmids, we have seen that neither the S. pombe chs2⁺ gene nor a catalytically inactive ScCHS2 mutant gene is able to support closure of the bud neck in S. cerevisiae (unpublished results), which demonstrates that this process requires active chitin synthesis and not just the presence of a Chs2 protein. This observation confirms the notion that in this organism ring contraction and septum formation are interdependent, as suggested elsewhere (Schmidt et al., 2002; VerPlank and Li, 2005), and that ScChs2p-dependent chitin synthesis provides the strength required to overcome the internal pressure developed during contraction (Cabib, 2004). In S. cerevisiae, lack of ScChs2p protein leads to a strong defect in cytokinesis, whereas in S. pombe the lack of Chs2p only elicits subtle defects in this process. In the fission yeast, chitin has been replaced by β(1,3) glucan at the primary septum (Humbel et al., 2001) and, unlike the situation in the budding yeast, the contractile ring is essential for survival. Thus, it seems that the S. pombe Chs2p would have lost its catalytic activity during evolution, but would have maintained a role in collaborating with other CAR proteins in a finely tuned mechanism that ensures the integrity of this structure.

All general techniques for S. pombe culture have been described previously (Moreno et al., 1991). The strains used are listed in Table S1 in supplementary material. For cell synchronisation at the G2-M phase, cells from strains carrying a cdc25-22 mutation were grown at 25°C to early log phase, arrested at 37°C for 3 hours, and released at 25°C by cooling down the culture in ice-cold water. The actin depolymerising drug latrunculin A (Sigma) was used at 100 μM from a stock at 5 mM in DMSO.

All general techniques have been described previously (Sambrook et al., 1989). Double mutants were obtained either by gene replacement of the chs2⁺ gene in the corresponding background using the chs2::ura4⁺ deletion module (Martin-Garcia et al., 2003), by tetrad analysis or, occasionally, by `random spore' selection from crosses (Moreno et al., 1991).

Three copies of the HA epitope were cloned as a NotI-NotI DNA fragment in the NotI site of the chs2⁺-GFP fusion protein (Martin-Garcia et al., 2003) such that the HA replaced the GFP. The localisation of Chs2-GFP was assessed using the fusion protein integrated at the chromosome under the control of its own promoter. However, in cases where the signal was very low an episomal plasmid (pAL-Chs2-GFP) was used to improve the visibility in photomicrographs.

Cell extracts were obtained from 150-300 ml cultures by breaking the cells in lysis buffer (50 mM Tris-HCl pH 7.2, 350 mM KCl, 25 mM EDTA, 1% Triton X-100, 1 mM (PMSF, 1 μg/ml aprotinin, leupeptin and pepstatin), using a Fastprep procedure (Savant; BIO 101). For western blots, a volume corresponding to 100 μg of protein was combined with 2 × sample buffer (100 mM Tris-HCl, pH 6.8, 2% SDS, 286 mM β-mercaptoethanol, 20% glycerol) and heated for 3 minutes at 65°C. Proteins were separated in 5% polyacrylamide gels, transferred to Immobilon-P membranes (Bio-Rad) in Tris-glycine buffer and decorated with monoclonal anti-HA (12C5A, Roche; 1:4000) or monoclonal anti-GFP (JL8, BD Biosciences; 1:1000) antibodies. ECL Advanced (Amersham) was used to develop the blots. For co-immunoprecipitation, 5 mg total protein were incubated for 2 hours at 4°C in the presence of anti-HA (1:50) or anti-GFP (1:100) antibodies. After the addition of Protein A-Sepharose beads (CL4B, Amersham) the samples were incubated overnight. The beads were washed three times at 4°C at 3000 rpm with lysis buffer and once with PBS. For elution, the beads were resuspended in 100 μl of 2 × sample buffer with 2 M urea, heated to 65°C for 3 minutes, and spun at 3000 rpm in a microfuge. 40 μl of the supernatant was loaded onto gels to perform anti-HA or anti-GFP western analyses.

Calcofluor and DAPI staining were performed as described previously (Arellano et al., 2000) on cells that had been fixed in cold 70% ethanol. Rhodamine-phalloidin staining of actin was performed as described previously (Marks and Hyams, 1985). For conventional fluorescence microscopy, images were captured with a Leica DM RXA microscope equipped with a Photometrics Sensys CCD camera, using the Qfish 2.3 program. Confocal microscopy was performed on a Leica TCS SL spectral confocal microscope with a 63 × 1.4 NA oil objective. The cells were then imaged in z-series consisting of 25-30 sections with a step size of 0.2 μm between each focal plane. The images were processed to obtain three-dimensional reconstructions from the cells. Images were processed with Adobe Photoshop, Image Ready or Leica Confocal Software (LCS).

We thank P. Pérez and D. Mulvihill for a critical reading of the manuscript and E. Cabib for helpful discussions and encouragement. We are indebted to M. Balasubramanian, A. Bueno, J. Hyams, I. Mabuchi, S. Moreno, D. P. Mulvihill, P. Pérez, T. D. Pollard, J. C. Ribas, Y. Sánchez, S. Sazer, V. Simanis, C. R. Vázquez and J. Q. Wu for strains and plasmids. We thank C. Castro for help with confocal microscopy. This work has been supported by grants SA128/04 and CSI02C05 from the Junta de Castilla y León and grant BIO2001-2048 from the CICYT. R.M.G. was supported by a FPU fellowship from Ministerio de Educación y Ciencia.