Etd1p is a novel protein that links the SIN cascade with cytokinesis

Many genes required for cytokinesis in yeasts have been identified in screenings for temperature‐sensitive mutants, but not all genes yield thermo‐sensitive mutant alleles. To identify new genes, we used ethanol as a conditional system that operates independently of temperature. Two large groups of mutants were isolated as ethanol‐sensitive mutants (ets) and ethanol‐dependent mutants (etd) (Jimenez and Oballe, 1994). One of the mutants belonging to the latter group, etd1‐1, produced elongated and multinucleate cells without a septum under restrictive conditions (absence of ethanol), a characteristic phenotype of S. pombe cells defective in cytokinesis (Figure 1A and B).

The etd1 gene encodes an uncharacterised 391‐amino‐acid polypeptide (Jimenez and Oballe, 1994). Etd1p sequence is not conserved in other eukaryotes and conventional sequence analysis failed to provide any information about its putative function. To determine whether etd1 was essential, one copy of the etd1 gene was replaced with the ura4 gene in a diploid S. pombe strain. Spores deleted for etd1 (etd1Δ) germinated and accumulated multiple nuclei without septation, an identical phenotype to that of etd1‐1 mutant cells under restrictive conditions (Figure 1C). Thus, etd1 encodes an essential new protein required for cytokinesis in fission yeast.

To further analyse the function of etd1, we examined the effect of increased expression of the etd1 gene. A cDNA encoding etd1 was cloned in the pREP3X vector under the control of the thiamine‐repressible nmt1 promoter (Maundrell, 1993). The pREP3X‐etd1 plasmid was transformed into etd1‐1 mutant cells and in wild‐type cells. Under repressed conditions (+thiamine), the nmt1:etd1 construction provided sufficient Etd1p expression to rescue the lethal phenotype of the etd1‐1 mutant (see Figure 1D). Under derepressed conditions (−thiamine), Etd1p overproduction generated elongated and multinucleate cells in both etd1‐1 mutant and wild‐type backgrounds (Figure 1D and data not shown). Thus, the phenotypic defect caused by an excess of Etd1p was identical to that produced by a deficiency of this protein, suggesting that Etd1p functions in a stoichiometric protein complex.

To better understand the role of Etd1p in cytokinesis, we examined the localisation of the protein in S. pombe cells. A strain was constructed that expressed an Etd1p‐GFP fusion from the thiamine‐repressible nmt41x promoter. A single copy of the construction was integrated at the leu1 locus. Under expression conditions, the resulting Etd1p‐GFP fusion protein was functional and able to complement the lethality of the etd1‐1 mutant and the etd1 null allele. In interphase cells, Etd1p‐GFP was located at the cell cortex and was more concentrated at the cell tips (Figure 2A, cell 1). In early anaphase, Etd1p‐GFP became concentrated in the medial region of the cell cortex as a broad band (Figure 2A, cell 2) and finally as a ring late in anaphase before septation (Figure 2A, cell 3). At the time of septum formation, Etd1p‐GFP spread into the cell (Figure 2A, cell 4) and once the primary septum is formed, it appears as a double layer at the cell equator (Figure 2A, cell 5). Finally, during cell separation, Etd1p‐GFP signal disappeared from the middle of the cell (Figure 2A, cell 6). An identical localisation was observed when the Etd1p‐GFP fusion was expressed under its own promoter on a multicopy plasmid or from its normal chromosomal locus, although fluorescence was almost undetectable in this latter case (data not shown). In spheroplasts lacking the cell wall, Etd1p remained associated with the cell periphery, suggesting that this protein is associated with the cell cortex or anchored to the cell membrane (Figure 2B).

To better characterise the cell cycle dynamics of Etd1p localisation, cdc25‐22 mutant cells expressing Etd1p‐GFP were synchronised by a temperature block–release protocol. This procedure causes a G2‐phase cell cycle block at the restrictive temperature after which entry into mitosis is synchronously induced by shifting the culture to the permissive temperature. In G2 cells, Etd1p‐GFP was observed at the cell tips (Figure 2C and D, time 0) and prior to anaphase began to relocate to the cell centre, about 15–20 min after the release (Figure 2C). Between 20 and 40 min after the release, Etd1p‐GFP was observed as a broad band at the cell cortex in the middle region of the cell (Figure 2C, time 20–40 min; Figure 2D, time 35 min). At late anaphase, Etd1p compacted into a ring at the site of cytokinesis (Figure 2C and D, time 55 min). After this, Etd1p was observed in the region of septum formation, spreading in a centripetal manner into the cell as the actomyosin ring contracted (Figure 2D, time 75 min). This observation suggests that it is probably associated with the ingressing plasma of the cleavage furrow. Between 75 and 95 min after the release, coinciding with septum formation, Etd1p‐GFP was observed along both sides of the septum (Figure 2C and D, time 80–95 min). Finally, after degradation of the primary septum, and simultaneous to cell separation, the Etd1p‐GFP signal decreased strikingly (Figure 2C, time 100–115 min). Overall, localisation and dynamics of Etd1p agree with a role of this protein in cytokinesis.

Dynamics of SPBs can be used to determine more precisely the cell cycle‐dependent localisation of Etd1p. Cdc7p is particularly useful to this end because this protein binds to the SPB at the G2–M transition and, after SPB duplication, it remains associated with only one of the two poles of the spindle during anaphase B (Sohrmann et al, 1998). Using Cdc7p‐GFP, we found that Etd1p‐GFP initiated its localisation at the cell centre after association of Cdc7p to the SPB and before SPB duplication (metaphase), being more concentrated as the SPB duplicated and initiated separation (anaphase A) (Figure 2E, cells 2 and 3 respectively). In agreement with this observation, only a reduced fraction of cells showed Etd1p in the medial cortex region in metaphase‐arrested cells by using the nda3‐KM311 mutation (Figure 2F). In these nda3‐KM311‐arrested cells, the actomyosin ring was already formed, as revealed by the medial ring component Cdc15p‐GFP (Figure 2F), indicating that Etd1p arrives to the middle cortex after medial ring assembly. At the time that Cdc7 is seen on only one SPB, in anaphase B, Etd1p compacted to a tight ring (Figure 2G, cell 2).

Thus, Etd1p localises to the cell tips during interphase, relocates to the cell centre shortly coincident with the metaphase to anaphase transition, after actomyosin ring assembly, and compacts to a medial ring during anaphase B (see Supplementary data, Movie 1).

To determine whether Etd1p associates to the actomyosin ring, we examined its location in different types of actomyosin ring mutants. Mid1p is a key factor for the central positioning of the cytokinetic ring (Sohrmann et al, 1996). The location and dynamics of Etd1p and Mid1p at the cell centre are very similar (Celton‐Morizur et al, 2004); however, in mid1‐deleted cells, Etd1p‐GFP localised at the randomly positioned actomyosin rings found in these mutant cells (Figure 3A). Therefore, medial ring components rather than Mid1p could be involved in the medial ring localisation of Etd1p. Cdc8p tropomyosin is an essential protein required to form F‐actin rings (Balasubramanian et al, 1992; Arai et al, 1998). We therefore analysed the localisation of Etd1p‐GFP in cdc8‐110 mutant cells and found that, at the restrictive temperature of 36°C, Etd1p never formed a ring (Figure 3B, upper panels). The S. pombe Cdc15p is also required for medial ring formation during cytokinesis (Fankhauser et al, 1995; Carnahan and Gould, 2003). Overexpression of cdc15 is sufficient to drive medial actin recruitment in G2‐arrested cells, indicating that Cdc15p plays a key role in the establishment of the medial actomyosin ring. In cdc15‐140 mutant cells under restrictive conditions (Fankhauser et al, 1995), we also failed to detect Etd1p‐GFP localised as a ring (Figure 3B, lower panels). In both cdc8‐110 or cdc15‐140 mutants at the restrictive conditions, Etd1p‐GFP remained at the cell tips or as a diffuse central band in mitotic cells but it was not assembled into the medial ring. Therefore, we conclude that Etd1p requires the actin ring for its proper localisation to the division site.

Recruitment of Etd1p into the medial ring could take place through an interaction with Cdc15p. This is based on the fact that cells expressing Cdc15p tagged with HA or GFP in combination with HA‐tagged Etd1p showed a synthetic lethal cdc phenotype (data not shown). To investigate a possible physical interaction between Etd1p and Cdc15p, strains carrying a plasmid expressing Etd1p‐GFP (tagged at the N‐terminus), Cdc15p‐Myc (tagged at the N‐terminus) or both were constructed. Protein extracts were prepared from these strains and the association between these proteins was determined in co‐immunoprecipitation experiments. As shown in Figure 3C, anti‐Myc immunoprecipitates contained Etd1p‐GFP, demonstrating that Etd1p interacts physically with Cdc15p. Similarly, Cdc15p was detected in anti‐GFP immune complexes (data not shown). Thus, Etd1p may localise to the actomyosin ring by association with Cdc15p.

Cytokinesis in S. pombe cells requires actomyosin ring assembly and F‐actin patch rearrangement from the cell tips to the division site. Cdc15p is involved in both processes (Fankhauser et al, 1995). F‐actin cables are also involved in the formation of the actomyosin ring (Arai and Mobuchi, 2002). To determine whether Etd1p has a role in any of these events, we analysed the assembly of Cdc15p‐GFP, the formation of F‐actin cables and dynamics of Crn1p‐GFP (coronin), a marker for actin patches (Pelham and Chang, 2001). As shown in Figure 4A–C, neither of these processes was affected in etd1‐1 mutant cells, suggesting that Etd1p functions downstream of Cdc15p and F‐actin patch recruitment in cytokinesis. However, in etd1‐1 mutant cells, the medial ring marked with Cdc15p‐GFP seems to fail constriction.

To better determine a role of Etd1p in actomyosin ring constriction, we used the myosin regulatory light chain (encoded by the rlc1 gene) tagged with GFP as a ring marker (Le Goff et al, 2000). Time‐lapse images of rlc1‐GFP in wild‐type and etd1‐1 living cells progressing from G2 to cytokinesis were obtained. In wild‐type cells, the actomyosin ring assembled early during mitosis (in metaphase) and initiated constriction late in anaphase, between 30 and 35 min after assembly (Figure 4D, upper panels). In etd1‐1 mutant cells, actomyosin ring assembled as in wild type, but the ring failed to constrict and finally collapsed (Figure 4D, lower panels, and Supplementary data, Movie 2A and B). Thus, Etd1p is not required for actomyosin ring assembly, but is essential for ring contraction.

Actomyosin ring contraction requires proper ring assembly and the activation of the SIN. Since etd1‐1 mutant cells assembled a normal medial ring, we wondered whether Etd1p might be required for SIN signalling. In fact, Etd1p‐defficient cells resembled sin mutants (see Figure 1). The protein kinase complex Sid2p–Mob1p functions at a late stage of the SIN pathway by transmitting the signal from the SPB to the medial ring to initiate cytokinesis (Salimova et al, 2000; Hou et al, 2004). We produced Mob1‐GFP and Sid2p‐GFP constructions and determined that, as previously described (Salimova et al, 2000; Hou et al, 2004), Mob1p‐GFP localised to both SPBs during mitosis and at the division site during septation in wild‐type cells (Figure 5A, upper panels). In Etd1p‐deficient cells, the SPB localisation of Mob1p‐GFP remained unaffected, but notably, Mob1p‐GFP did not relocalise to the division site (Figure 5A, lower panels). We were unable to construct a sid2‐GFP etd1‐1 double mutant due to negative genetic interaction between these two alleles (data not shown). However, in a strain deleted for etd1 (etd1Δ) kept alive by expressing etd1 from the weak nmt81x promoter (etd1Δnmt81x:etd1), we observed that the localisation of Sid2p‐GFP to the cleavage site also required Etd1p (see below). We thus conclude that Etd1p is required for the transduction of the Sid2p–Mob1p signal from the SPB to the division site.

Since Etd1p associates to the actomyosin ring and is required for SIN signalling, this new protein might be a ring component required for the recruitment of Sid2–Mob1 complexes to this structure, that is, a downstream element of the SIN cascade localised at the medial ring. If this were the case, activation of upstream elements of SIN should take place normally in etd1‐mutant cells, and similarly, ectopic activation of the SIN alone should not bypass the requirement of Etd1p to relocate Sid2p or Mob1p from the SPB to the actomyosin ring.

To analyse upstream activation of SIN, we studied the localisation of Cdc7p‐GFP in etd1‐1 mutant cells by time‐lapse microscopy. The Spg1p GTPase localises to the SPBs throughout the cell cycle. In interphase cells, Spg1p is GDP‐bound, but upon entry into mitosis, it converts into the GTP‐bound form. Spg1p is then active at both SPBs until anaphase B, when it converts back into the inactive GDP‐bound form at one of the two SPBs. Cdc7p only binds the active (GTP‐bound) form of Spg1p (Sohrmann et al, 1998). Thus, Cdc7p is an excellent marker for monitoring upstream activation of the SIN cascade. As shown in Figure 5B (upper panels), under permissive conditions for etd1‐1, Cdc7p‐GFP appeared at both SPBs at the initiation of mitosis and only at one SPB as cells progressed through anaphase until the completion of cell division. Cdc7p‐GFP also localised to both SPBs in early mitosis in etd1‐1 cells under restrictive conditions, indicating that the initial activation of Spg1p does not require Etd1p. However, in these Etd1p‐deficient cells, Cdc7p‐GFP signal rapidly decayed early in anaphase (Figure 5B, lower panels), suggesting that Etd1p is somehow necessary to maintain Spg1p activity during anaphase until the completion of cytokinesis. Quantification of the fluorescence intensity at the SPBs reinforces this observation (see Figure 5B).

To determine whether this premature Spg1p inactivation in cells lacking Etd1p is the only reason for the failure of SIN signalling through Mob1p–Sid2p, we maintained the SIN active with a cdc16‐116 mutation and investigated the localisation of Mob1p‐GFP and Sid2p‐GFP in an etd1Δ nmt81x:etd1 background. Under repressed conditions at 25°C (+T), these cells mimicked the sin defect of a deficiency in Etd1p and were unable to localise Sid2p–Mob1p proteins to the cell division site (see Figure 5C, upper panels, for the case of Sid2p‐GFP). Upon inactivation of Cdc16p in these cells (at 34°C), frequent unseptated postmitotic cells were found (Figure 5C, lower panels, for Sid2p‐GFP). Similar results were obtained by activation of the SIN by overexpressing plo1 or spg1. These results indicate that Etd1p functions downstream of Spg1p in SIN signalling.

To elucidate the role of Etd1p in SIN signal transduction, we analysed the localisation of Etd1p‐GFP in different mutants of the SIN pathway. Interestingly, we found that in cdc7^ts and sid2^ts mutant cells, Etd1p‐GFP failed to localise to the medial ring at the restrictive temperature. Instead, these mutant cells accumulated Etd1‐GFP in a broad band at the plasma membrane overlying the site of cytokinesis (Figure 6A, upper panel, and data not shown). These observations indicate that relocation of Etd1p from the cell cortex to the actomyosin ring requires Spg1p activation. In agreement with this observation, Etd1p‐GFP localised to the multiple septa produced by the ectopic activation of SIN (Figure 6A, lower panels, for cdc16‐116 mutants). Hence, relocation of Etd1p from the cell cortex to the ring is a downstream step of the SIN pathway required for signalling the initiation of cytokinesis.

In S. pombe, sterol‐rich membrane domains and Etd1p follow similar cell cycle dynamics. In sin mutants, these sterol‐rich membrane domains accumulate in a specific pattern, which is different from that observed in medial ring mutants (Takeda et al, 2004; Figure 6B). Interestingly, patterns in etd1‐1 and sid2‐250 mutant cells are identical (Figure 6B). Thus, although Etd1p is located at the cell cortex and the medial ring, this protein behaves more like a SIN‐signalling protein than a structural component of the cytokinetic machinery.

As described in Figure 1D, overproduction of Etd1p caused a SIN‐defective phenotype similar to that of etd1‐1 mutants. The above properties described for Etd1p‐deficient cells also occurred in Etd1p‐overproducing cells (see Supplementary Figure 1 for Crn1p‐GFP, Rlc1p‐GFP and Cdc7p‐GFP), suggesting that the amount of Etd1p is critical in the control of cytokinesis in fission yeasts.

As shown in Figure 2, the cellular localisation of Etd1p is regulated in a cell cycle‐dependent manner. To study whether its expression was also regulated in a cell cycle‐dependent manner, we determined the levels of Etd1p and etd1 mRNA in a synchronous culture, using a strain in which the single chromosomal copy of etd1 had been tagged with three copies of HA (see Materials and methods). Cells were synchronised using the cdc25‐22 block–release protocol, as described above. Etd1p‐HA levels were sharply periodic, rising to a peak at the end of anaphase, shortly after the peak of binucleated cells (Figure 7A and B, 60 and 60–75 min), and decreasing at 90–105 min when most of the cells had completed septation (Figure 7A and B). The kinetics of etd1 mRNA accumulation was similar to that described for Etd1p, increasing briefly before accumulation of the protein (see Figure 7B). The results of this experiment show that etd1 mRNA and protein levels fluctuate during the cell cycle, peaking during actomyosin constriction and septation.

To investigate whether the decay in Etd1p levels after septation was due to rapid degradation by the proteasome pathway, we determined Etd1p levels in the mts3‐1 temperature‐sensitive mutant, which is defective in subunit 14 of the 26S proteasome (Gordon et al, 1996). The amount of Etd1p was higher in the mts3‐1 strain at the restrictive temperature than in the wild‐type strain or mts3‐1 at 25°C (Figure 7C, upper panels). This result suggests that the proteasome degradation pathway might be involved in the regulation of Etd1p levels.

In the mts3‐1 mutant at 36°C, Etd1p high molecular weight bands were observed after a longer exposure of the films (data not shown), which could represent ubiquitin conjugates. To test this possibility, we expressed a His₆‐tagged version of ubiquitin in mts3‐1 cells and determined the presence of Etd1–ubiquitin conjugates. Extracts were purified using Ni²⁺‐NTA resin (Treier et al, 1994), separated on a polyacrylamide gel and then Western blotted with anti‐HA antibodies. Etd1p high molecular bands were detected in mts3‐1 mutants at the restrictive temperature (Figure 7C, lower panels). These results indicate that Etd1p is polyubiquitinated and degraded through the ubiquitin‐dependent 26S‐proteasome pathway.

In this study, we identified a novel protein, Etd1p, required for constriction of the actomyosin ring in fission yeast (see Figure 4). This protein resides associated with the cell membrane (Figure 2B). According to its hydrophilic amino‐acid profile, Etd1p is a soluble protein (data not shown), suggesting that this plasma membrane association may be mediated by interactions with other membrane proteins or by post‐translational modifications.

During the cell cycle, Etd1p relocalises from the cell tips to the medial cortex region in the metaphase–anaphase transition, and it translocates to the actomyosin ring as the cells initiate cytokinesis in anaphase B (Figure 2). In mutant cells defective in medial ring assembly or SIN activity, Etd1p fails to assemble into a ring (Figures 3 and 6), indicating that recruitment of Etd1p to the actomyosin ring requires ring components and SIN signalling. In these mutants, Etd1p relocalised normally from the cell tips to the centre, indicating that this relocation of Etd1p occurs by an independent mechanism to that involved in its assembly into the actomyosin ring. The localisation pattern of sterol‐rich membrane domains is regulated in a cell cycle‐dependent manner (Wachtler et al, 2003), following a similar dynamics to that described for Etd1p (Figures 2 and 3). It is possible that these membrane domains could play a role in the mechanism regulating cell cortex association of Etd1p and its relocation from the tips to the cell centre. As shown in this report, Etd1p associates to Cdc15p (Figure 3C), a key element of the actomyosin ring, which has been recently shown to be required for the proper localisation of the sterol‐rich domains at the cell centre (Takeda et al, 2004).

Once in the medial cell cortex, the recruitment of Etd1p into the ring requires the actomyosin ring and is triggered by SIN activation (Figures 3 and 6). The fact that signalling of the Sid2p–Mob1p protein kinase complex from the SPB to the division site requires Etd1p (Figure 5), the association of Etd1p with Cdc15 (Figure 3C) and the synthetic lethality found in double etd1‐1 sid2‐250 mutants suggests that Etd1p, Cdc15p and Sid2p–Mob1p are closely regulating the initiation of septum formation and ring constriction. In a simple model, Etd1p could be a scaffold protein that associate to Cdc15p to facilitate the recruitment of Sid2p–Mob1p to the division site; however, the requirement of Etd1p is suppressed by certain morphogenetic mutants (Jimenez and Oballe, 1994), suggesting that Etd1p could play a regulatory role in ring constriction rather than being an integral scaffold component of the contractile ring. The role of the S. pombe Cdc42p GTPase and other cell polarity proteins in Etd1p localisation and function is currently under study.

The SIN cascade coordinates mitosis and cytokinesis. While mechanisms coordinating karyokinesis (mitosis) with SIN activity are well established (Bardin and Amon, 2001; Guertin et al, 2002b), little is known about how cytokinetic events (initiation, progression and exit) are connected with this regulatory cascade. Wild‐type cells septate only once per cell cycle and this is controlled by SIN regulation. Therefore, when cytokinesis is initiated in response to SIN signalling, a feedback mechanism must operate in wild‐type cells to maintain SIN activity during ring constriction and to halt it once septation is completed. Etd1p is a good candidate for the coordination of ring constriction with SIN activity: recruitment of this protein to the actomyosin ring is triggered by SIN activity, it accumulates at the medial ring during cytokinesis, is required for the execution and maintenance of SIN signalling and is degraded at the end of septation. A model for this regulatory circuit is schematised in Figure 8 and discussed below.

The initiation of actomyosin ring constriction requires relocation of Etd1p to the division site. This process only occurs when actomyosin ring has been assembled and mitosis triggers initial SIN activity. Thus, Etd1p integration into the actin ring may ensure that the ring components and mitotic events take place normally before ring constriction is initiated. In Saccharomyces cerevisiae, it has been proposed that Tem1p becomes active when it is exposed to Lte1 during the migration of the nucleus to the bud at anaphase. This mechanism ensures that one end of the spindle, and hence one set of chromosomes, has traversed the bud neck and entered the daughter cell before triggering exit from mitosis by activating the MEN (Bardin et al, 2000). In S. pombe, a rod‐shaped cell that divides by fission, constriction of the medial ring triggered by the SIN at anaphase ensures that each cell will receive a nucleus in the course of a normal division. Like Lte1 in the bud, Etd1p localisation at the medial ring provides an excellent means by which to coordinate ring constriction and SIN activity.

Once at the division site, progression of ring contraction is accompanied by increasing levels of Etd1p (see Figure 7), which colocalises with new membrane surface behind the ring as it constricts (see Figure 2). Since Spg1p becomes inactive prematurely in the absence of Etd1p (Figure 5B), the accumulating Etd1p at the contracting ring could have a role in a feedback loop maintaining SIN activity during the process.

Protein degradation by the proteasome pathway has been shown to regulate processes such as chromosome segregation by degrading securin (Funabiki et al, 1996), or exit from mitosis by degrading mitotic cyclins (Glotzer et al, 1991). Etd1p is the first component of the SIN known to be ubiquitinated and degraded via the proteasome during each cell cycle (Figure 7C). Since the lack of Etd1p impedes SIN signalling (Figure 5), we suggest that the proteolytic degradation of Etd1p at the end of cytokinesis ensures SIN inactivation once ring constriction and septation are completed.

S. pombe divides by medial fission through the use of an actomyosin contractile ring analogous to the cleavage furrow in animal cells (Balasubramanian et al, 2004). The protein described in this study provides a novel function that sheds light on a new mechanism coordinating initiation, progression and exit of cytokinesis with SIN activation in the fission yeast. It is clear that some of the components of the SIN have homologues in higher eukaryotes (Guertin et al, 2002a). Etd1p does not appear to have sequence homologues in other eukaryotes; however, the identification of etd1 functional homologues may be helpful in understanding how mammalian cells connect the SIN cascade and cytokinesis.

Strains used in this study are listed in Supplementary Table I. Growth conditions and strain manipulations were as described previously (Moreno et al, 1991). Medium containing 6% ethanol was made up as described by Jimenez and Oballe (1994). Yeast transformation was carried out using the lithium acetate transformation protocol (Norbury and Moreno, 1997). Experiments in liquid culture were carried out in minimal medium (EMM) supplemented as required, starting with a cell density of 2–4 × 10⁶ cells/ml, corresponding to the mid‐exponential phase growth. To prepare spheroplasts, growing cells were washed with water, resuspended in 0.65 M KCl containing 1 mg/ml novozyme (at about 10⁷ cells/ml) and incubated at 30°C until spheroplasts were formed (for 20–30 min).

Induction of synchrony was accomplished by arrest–release of a cdc25‐22 mutant. Cells were grown to mid‐exponential phase at 25°C and then shifted to 36°C for 4 h. The culture was cooled rapidly to 25°C by introducing it into an ice‐cold water bath. Samples were taken every 15 min for analysis of RNA and protein. For regulated nmt expression, described methods were used (Maundrell, 1993). Repression was achieved by adding thiamine, and nmt‐repressed cells analysed 12–15 h after the addition. Thiamine was removed for nmt‐driven expression and cells analysed 18–20 h after induction.

The etd1Δ null allele was generated by a PCR‐based approach as described previously (Bahler et al, 1998). The S. pombe ura4⁺ gene was amplified by PCR using the KS‐ ura4⁺ plasmid as template and the primers 5′‐ ATGGGTTGACGCGTACAGTCTTTTCTCTACGGCAATAATTCTTTCTTTAATTTAGCTTAAGAGAAATATAGATTTACGAGTCGCCAGGGTTTCCCAGTCACGAC‐3′ and 5′‐ATTGAAATTATCCTAAGCGAAATTTTGACACTTTGGATAACAAATAGATACTTCCCATTCCATTAATGGATTAAATAGCCAGCGGATAACAATTTCACACAGGA‐3′ (the 5′ ends are targeting sequences corresponding to 80 nucleotides immediately upstream and downstream of etd1 ORF and the 3′ ends are 24 nucleotides corresponding to sequences on either side of pBlueScript). A diploid was constructed by mating the h⁺leu1‐32 ura4‐d18 ade6‐M210 and h⁻leu1‐32 ura4‐d18 ade6‐M216 strains. This diploid was transformed with the PCR‐amplified fragment, and the ura4⁺ transformants were checked for the correct integration of the DNA fragment in the etd1 locus by Southern blot analysis. A heterozygous diploid was sporulated and the spores were germinated in YE or YE+6% ethanol at different temperatures between 25 and 36°C; after tetrad dissection, two spores were able to grow and two spores died under all conditions analysed.

In order to obtain a conditional deficiency for etd1, an etd1Δ nmt81x:etd1 strain was constructed, where etd1 expression is regulated under the weak repressible nmt81X promoter. The PstI–EcoRI fragment of nmt81x:etd1 was subcloned into Pst1–SacI sites of pJK148. The resulting plasmid was linearised with NruI and integrated at the leu1 locus in the etd1::ura4/etd1 ade6‐M210/ade6‐M216 ura4‐d18/ura4‐d18 leu1‐32/leu1‐32 strain. Diploid cells were selected in EMM with adenine and thiamine. The sporulated cells were digested with β‐glucuronidase and selected in EMM supplemented with adenine.

Etd1‐HA was generated by a PCR‐based approach as described by Bahler et al (1998). Primers 5′‐TAGATTTCGATGCGTCCATGCTACCAGGGCTAAACGATCATGTGCATTATTTAATGTCCCAACTCCAACTATTATTACACCGGATCCCCGGGTTAATTAA‐3′ and 5′‐AAGCTTCTTAAGATGCAGAAGTATAAAAAGGTTGTTAATGGCCAGTTTCACCTTTACTTGCTATCCACCAATGCGAAGAATTCGAGCTCGTTTAAAC‐3′ were used to amplify 3HA‐Kan from pFA6a‐3‐HAkanMX6. The PCR‐amplified fragment was transformed into a wild‐type h⁻ 972 haploid strain. Stable single‐copy integrants were selected and checked for the correct integration of the DNA fragment at the etd1 locus by PCR and Southern blot.

The pREP41‐EGFP‐Etd1 plasmid was created by cloning the NdeI–SmaI fragment containing the etd1 ORF generated by PCR from genomic DNA using the primers 5′‐ATTTCGCATATGGGAGTTGTATACGG‐3′ and 5′‐TAGATACTTCCCGGGCCATTAATG‐3′ in the vector pREP41‐EGFP (Craven et al, 1998) digested with NdeI and SmaI. A pJK148nmt41x.EGFP‐etd1 strain was created by PCR amplification from pREP41‐EGFP‐etd1 and subcloning into PstI–EcoRI sites of pJK148. The resulting plasmid was linearised with NruI and integrated in the S. pombe leu1 locus by homologous recombination.

Total protein extracts were prepared from 3 × 10⁸ cells as described previously (Daga et al, 2003). For Western blot analysis, 100 μg of total protein extract was run on a 12% SDS–PAGE gel, transferred to nitrocellulose and probed with anti‐HA 12CA5 (0.15 μg/ml) monoclonal antibodies. Goat anti‐mouse antibodies conjugated to horseradish peroxidase (Amersham) (1:3500) were used as secondary antibodies. As loading controls, rabbit affinity‐purified anti‐Cdc2p C2 antibodies (1:500) were used. Immunoblots were developed using the Super Signal kit (Pierce).

Total protein extracts were prepared from 3 × 10⁸ cells using HB buffer (Moreno et al, 1991). Cell extracts were spun at 4°C in a microfuge for 15 min, and protein concentrations were determined. A 2–3 mg portion of total soluble protein was subjected to immunoprecipitation by consecutive incubation with the monoclonal anti‐Myc 9E10 (1 μg) for 1 h on ice, and protein G‐Sepharose (Pharmacia‐Biotech) for 30 min at 4°C in a rotating wheel. Immunoprecipitates were washed six times with 1 ml of HB buffer. Immunoprecipitates were resolved on 10% SDS–polyacrylamide gels, blotted and probed with anti‐Myc 9E10 (1:1000) or anti‐GFP (1:1000) antibody.

RNA from cells was prepared by lysis with glass beads in the presence of phenol (Moreno et al, 1991). RNA gels were run in the presence of formaldehyde, transferred to GeneScreen Plus (NEN, Dupont) and probed with the etd1 ORF according to the manufacturer's instructions.

DNA, septum material and F‐actin were visualised using DAPI, calcofluor and rhodamine‐phalloidin respectively. Staining with rhodamine‐conjugated phalloidin was performed as described (Marks et al, 1986). Sterol‐rich membrane domains were localised in live cells by filipin staining (Takeda et al, 2004). Filipin (Sigma F9765) was used at a final concentration of 5 μg/ml (in 1% DMSO). Cells were examined using a Leica fluorescence microscope equipped with a plan Apo × 100 lens. Three‐dimensional confocal images were obtained with a confocal Zeiss microscope. For time‐lapse experiments, cells were imaged in Z‐series every 5 min with a step size of 0.3 μm between focal planes.

Supplementary data are available at The EMBO Journal Online.