Rot1 plays an antagonistic role to Clb2 in actin cytoskeleton dynamics throughout the cell cycle

Polarized cell growth is fundamental for numerous cellular processes including differentiation, proliferation and morphogenesis. All eukaryotic cells essentially follow the same general scheme in their control of cell polarity (Nelson, 2003) - in response to external or internal signals, a site of polarization is established at the cell surface, leading to the recruitment of specific proteins, which are then responsible for cytoskeleton organization necessary for polarized cell surface growth. Saccharomyces cerevisiae cells undergo polarized growth throughout their life cycle (Lew and Reed, 1995; Pruyne and Bretscher, 2000a; Pruyne and Bretscher, 2000b). First, cells polarize their growth toward a specific site on the cell surface to form a bud. Bud growth can be divided into two periods: cell growth is initially concentrated at the tip of the bud (apical growth period), whereas growth is redirected thorough the whole bud during the G2 phase (isotropic growth period). Growth is then reoriented toward the mother-bud neck during cytokinesis.

Cytoskeletal polarization and rearrangements in eukaryotic cells are regulated by the Rho family of small GTPases (Jaffe and Hall, 2005). Yeast contains six Rho-related GTPases (Cdc42 and Rho1-Rho5) that participate in multiple aspects of cell polarity, secretion and cell wall biosynthesis (Chant, 1999). Cdc42 has been implicated in the control of morphogenetic events during the cell cycle, specifically in the generation of cellular polarity and the development of normal cell shape, acting via various effector proteins (Etienne-Manneville, 2004; Pruyne and Bretscher, 2000b). Cdc42 is also involved in polarized secretion acting at a late step in exocytosis (Adamo et al., 2001). In the case of Rho1, it functions through the cell integrity pathway to regulate the actin cytoskeleton (Delley and Hall, 1999; Dong et al., 2003).

Polarized growth involves the spatial regulation of the actin cytoskeleton with the assembly of actin filaments in patches that cluster at specified cortical domains and in cables that extend along the growth axis. Many of the proteins involved in actin cytoskeleton and secretion are localized to regions of polarized growth, including Cdc42 and its guanine-nucleotide exchange factor Cdc24 (Richman et al., 2002; Toenjes et al., 1999; Ziman et al., 1993), signaling proteins from the PKC pathway (Andrews and Stark, 2000; Denis and Cyert, 2005; van Drogen and Peter, 2002; Yamochi et al., 1994) and Spa2 (Arkowitz and Lowe, 1997). The existence of many physical interactions between proteins involved in cell polarity has been described (Drees et al., 2001) and some of these proteins function together as a complex called the polarisome (Fujiwara et al., 1998; Sheu et al., 1998). Spa2 is considered to be a scaffold protein of the polarisome (Ozaki-Kuroda et al., 2001; Sheu et al., 1998; Shih et al., 2005; Tcheperegine et al., 2005; van Drogen and Peter, 2002). The polarisome regulates polarized growth by two coordinated mechanisms - the local activation of Cdc42 and Rho1, and the regulation of exocytosis (Lechler et al., 2001; Shih et al., 2005; Tcheperegine et al., 2005). Spa2, or the polarisome, is not essential. However, in its absence, cells manifest several polarity-related phenotypes, such as reduced apical growth and an altered budding pattern (Sheu et al., 2000) and defects in cytokinesis (Shih et al., 2005; Snyder et al., 1991), pheromone response (Gehrung and Snyder, 1990) and pseudohyphal growth (Mösch and Fink, 1997).

In addition to the spatial controls, periods of polarized growth must be temporally coordinated with the events of the cell division cycle. Cell cycle progression is governed by the periodic activation and inactivation of different cyclin-dependent kinases (CDKs) (Morgan, 1997; Roberts, 1999). Work on Saccharomyces cerevisiae has unveiled the dependence of cytoskeleton dynamics on cell cycle control machinery (Lew and Reed, 1993; Lew and Reed, 1995). However, in most cases it is still poorly understood how the CDKs control cellular morphogenesis at a molecular level. Actin cytoskeleton polarization at the site of bud emergence is triggered at the start by the Cln1,2-Cdc28 kinase activities. Cln2-Cdc28 controls local activation of Cdc42 at the budding site at least in part by the regulation of the subcellular localization of Cdc24 (Shimada et al., 2000). Successive changes in polarized growth throughout the cell cycle are regulated by the mitotic Clb-Cdc28 kinases. Clb-Cdc28 complexes promote actin depolarization during the switch from apical to isotropic growth. It remains unknown how the kinase controls this process, although we know that it is mediated by the Cdc42 GTPase (Tjandra et al., 1998). In addition, Clb2-Cdc28 restrains repolarization to the mother-bud neck. Only when the Clb2-Cdc28 kinase is inactivated at the end of mitosis is the actin cytoskeleton directed to the neck region to complete cytokinesis. The inactivation of mitotic cyclins will also allow the assembly of a new bud site in G1 cells in preparation for the next budding event (Padmashree and Surana, 2001). The interplay between cell cycle control and cell morphogenesis is further demonstrated by the implication of Cdc42 effectors in mitotic exit (Hofken and Schiebel, 2002; Hofken and Schiebel, 2004) and the involvement of mitotic exit network (MEN) proteins, which control the inactivation of CDK, in cytokinesis (Hwa Lim et al., 2003; Jimenez et al., 2005).

ROT1 is an essential gene that was identified in a search for suppressor mutations of the lethality of a tor2^ts mutant strain defective in actin cytoskeleton function (Bickle et al., 1998). Furthermore, Rot1 is required for normal levels of the cell wall components and normal cell wall structure (Bickle et al., 1998; Machi et al., 2004; Takeuchi et al., 2006). A comprehensive genetic-interaction analysis predicts that ROT1 is clustered in the O-glycosylation/GPI biosynthesis and cell wall formation functional category (Schuldiner et al., 2005). These findings suggest that the protein is involved in a morphogenetic process that will affect cell wall biosynthesis. More recently, it has been suggested that Rot1 could function with Kar2 in protein folding in the endoplasmic reticulum (Takeuchi et al., 2006). However, its detailed molecular function remains unknown. Rot1 is predicted to contain 256 amino acids. Sequence analysis identifies a transmembrane domain at its C-terminus. Indeed Rot1 is an integral membrane protein mainly located within the nuclear envelope and ER system (M.A.J. et al., unpublished results) (Takeuchi et al., 2006). Interestingly, the Rot1 sequence displayed a great similarity to the sequence of unknown proteins from other yeast like Schizosaccharomyces pombe, Candida albicans, Kluyveromyces lactis and Yarrowia lipolytica among others.

In this work we investigated the function of the essential ROT1 gene. Rot1 is involved in the control of the actin cytoskeleton during the cell cycle, acting in a reverse manner to the Clb2 cyclin. Inactivation of ROT1 caused a loss of apical growth and the inability to polarize the actin cytoskeleton to the neck region at the end of mitosis leading to a defect in septum formation and cell division. The functional antagonism between Rot1 and Clb2 is further supported by genetic and biochemical data, which indicate that Rot1 is implicated in Clb2 protein degradation.

ROT1 is an essential gene. In order to investigate its function, we constructed a conditional mutant strain that expresses ROT1 under the control of the doxycycline-regulated tetO₇ promoter (Fig. 1A). As expected, cells were unviable in YPD containing 5 μg/ml doxycycline (Fig. 1B). In a first attempt to characterize the function of Rot1, doxycycline was added to exponentially growing tetO₇:ROT1 cells and their morphology was analyzed after 8 hours. A change was observed in cell distribution throughout the cell cycle (Fig. 1C). There was a significant increase in the percentage of budded cells (87% in tetO₇:ROT1 strain compared with 67% in the wild-type strain) and in the number of cells with two segregated nuclei (34% in tetO₇:ROT1 strain compared with 17% in the wild-type strain). FACS analysis demonstrated that DNA replication was not blocked (data not shown). Long spindles were disassembled in the budded cells with two segregated nuclei (Fig. 1C). Interestingly, incubation in the presence of doxycycline led to the rebudding of cells without cytokinesis being completed. Remarkably, the control of bud site selection in rebudded cells was lost because a random rather than an axial budding pattern was observed for the new bud (Fig. 1C). All these results point to a defect in cell cycle progression in the rot1 mutant along with a partial arrest at the latest stages of cell cycle owing to a defective cytokinesis.

In addition to the defect in cell division, a bud-morphogenesis defect is also associated with the inactivation of Rot1. To quantify this defect, we measured the ratio of bud length to bud width in postanaphase cells, which was expected to be >1 for ellipsoidal cells and 1.0 for round cells. As expected, wild-type cells had an average length/width ratio of 1.2; nevertheless, the average length/width ratio for rot1 cells was close to 1.0 (Fig. 2). This was more clearly observed when Rot1 was inactivated in strains with a hyperpolarization defect, like the clb2 strain: the bud length/width ratio was 1.1 in clb2 rot1 cells compared with 1.8 in clb2 cells. Thus, inactivation of Rot1 resulted in rounder cells than seen in the wild-type. This ellipsoidal shape of yeast cells is due to a phase of apical bud growth, and the presence of rounder cells is associated with defects in apical growth (Sheu et al., 2000). Our results indicate a role of Rot1 in growth polarization, especially in sustaining apical growth after bud emergence.

Actin cytoskeleton polarization in S. cerevisiae is governed by the Rho/Rac family of GTPases, which includes the Cdc42 protein. Cdc42 is not only involved in the polarization of growth during bud emergence and cytokinesis, but it also regulates the Clb2-induced apical-isotropic switch during bud growth (Richman et al., 1999; Tjandra et al., 1998). Therefore, we investigated whether ROT1 genetically interacts with CDC42 by introducing the tetO₇:ROT1 construction in a strain containing the cdc42-1 mutant allele. Remarkably, the tetO₇:ROT1 cdc42-1 cells were able to grow at 37°C - a condition under which the single cdc42-1 mutant is unviable (Fig. 3A). Suppression of the cdc42-1 mutant lethality is dependent on the ectopic expression of ROT1 from the tetO₇ promoter, since growth at high temperature of the tetO₇:ROT1 cdc42-1 cells was abolished by adding doxycycline to the growth medium (in this case cells are transformed with a centromeric plasmid bearing the ROT1 gene to allow viability in the presence of doxycycline). Moreover, a multicopy plasmid containing the ROT1 gene also suppressed the thermosensitive growth defect of the cdc42-1 strain (Fig. 3B). These results strongly reinforce that ROT1 is involved in the regulation of actin cytoskeleton polarization.

The cell integrity pathway (Levin, 2005) controls actin cytoskeleton polarization under different conditions. ROT1 genetically interacts with TOR2, which is involved in the regulation of the cell integrity pathway, among other cellular functions (Bickle et al., 1998). We investigated the existence of a genetic interaction between the inactivation of ROT1 and mutations affecting the cell integrity pathway. As shown in Fig. 4, the partial inactivation of ROT1 caused lethality in cells compromised in Pkc1, the MAPK Slt2 or the GTPase Rho1. In addition, the inactivation of ROT1 leads to the activation of the MAPK Slt2 (Fig. 4B). All these results suggest a functional connection between the cell integrity pathway and the function of Rot1, and are consistent with a function of Rot1 in actin cytoskeleton control.

So far we have described that ROT1 is required for both proper cytokinesis and apical bud growth. We envisage the possibility that Rot1 could affect actin cytoskeleton polarization, not only during the period of apical growth, but also at the end of mitosis. To test this possibility, actin filaments were stained with labelled phalloidin in rot1 cells in asynchronous cultures. Polarization either at the bud emergence site or in small buds was clearly observed both in the wild-type and rot1 cells (Fig. 5A). Importantly, when late-anaphase/telophase cells (as deduced from DAPI staining of DNA) were analyzed, actin cytoskeleton polarization to the mother-bud neck could not be detected in the case of rot1 cells. This result is more clearly observed when α-factor-synchronized cultures are used (Fig. 5B). In small-budded cells, actin cortical patches are concentrated 30 minutes after the release of the α-factor-induced arrest at the bud in both wild-type and rot1 strains. Actin cytoskeleton depolarization occurs as cells progress through the cell cycle. When cells have completed mitosis, wild-type cells repolarize the actin cytoskeleton to the mother-bud neck; however, no polarization was observed in the rot1 cells, even after longer incubation times. Rebudding was observed in rot1 cells and, interestingly, actin patches localized at the new buds. In conclusion, ROT1 is not required for actin cytoskeleton polarization during bud emergence, although it is necessary for the repolarization occurring at the mother-bud junction at the end of the cell division cycle. This could certainly explain the cell division defect caused by ROT1 inactivation.

Polarized cell growth in S. cerevisiae is controlled by several GTPases and downstream effector proteins as well as the polarisome complex. The mutation of the polarisome component Spa2 causes defects in apical growth (Sheu et al., 2000) and cytokinesis (Shih et al., 2005; Snyder et al., 1991), and is synthetically lethal with mutations of the cell integrity pathway (Costigan et al., 1992), a similar finding to that described herein for the rot1 mutant. Spa2 functions as a scaffold-like protein at the sites of polarized growth. We wondered whether the defect in polarization to the bud neck could be due to a defect in the Spa2 protein localization. However, in rot1 cells as in wild-type cells (Arkowitz and Lowe, 1997), Spa2 localized at the bud tip during bud emergence and growth, where it remained until late S phase and finally relocated at the mother-daughter bud neck where it formed a ring structure which split into two rings (Fig. 6A). The latter is an important result, which suggests that the actin contractile ring is functional in rot1 mutant cells (see below). Without completing cell division, Spa2 in the mother cells disappears from the neck region and concentrates at the cortex in the presumptive next bud site and, as soon as the new bud is formed, at the new bud tip. A similar relocalization of Spa2 was observed in the daughter cell over longer times.

Next, we examined the localization of the GTPase Cdc42 and its GEF Cdc24, which are also relocated at the bud-neck region at the end of mitosis in order to trigger actin cytoskeleton polarization (Richman et al., 2002; Toenjes et al., 1999). As shown in Fig. 6B,C, both proteins were properly located in the absence of Rot1. To summarize, Rot1 does not affect the dynamic localization of polarity proteins like Spa2, Cdc42 or Cdc24, rather it could affects the ability of these proteins to direct actin cytoskeleton polarization to the bud neck region.

Cytokinesis in budding yeast is accomplished by the concerted action of the actomyosin contractile ring and the formation of the septum by polarized secretion to the bud neck. As noted above, the results with Spa2-GFP suggested that the actomyosin contractile ring performs cytoplasm division in rot1 cells. However, the lack of actin cytoskeleton repolarization to the mother-daughter neck would impede proper septum formation. To investigate this possibility, we first analyzed whether cytoplasm division indeed occurred in rot1 cells by digesting the cell wall with zymolyase. The dramatic change in cell morphology caused by cell wall digestion, with a strong reduction in the percentage of budded cells with two nuclei and rebudded cells (Fig. 7A), demonstrated the splitting of the dividing cells. Cytoplasm division is normally effected by the contraction of an actomyosin ring; however, it can still occur in its absence albeit in an abnormal way (Schmidt et al., 2002). To clarify whether the actomyosin contractile ring is functional, we analyzed the localization of a Myo1-GFP fusion protein in α-factor-synchronized rot1 cells (Fig. 7B). As described for the wild-type strain (Bi et al., 1998; Lippincott and Li, 1998), Myo1 localized at the neck immediately after budding. Later in the cell cycle, there was a reduction in the signal size, which reflects the contraction of the actomyosin ring, until the signal finally disappears form the neck. Interestingly, Myo1 reappears at the new neck regions of the rebudded cells. This result confirmed that mother-daughter division in rot1 cells is due to the action of the actomyosin contractile ring.

Next, the membrane continuity and chitin deposition in the neck region were analyzed by fluorescence microscopy (Fig. 7C,D). We focused on rebudded cells since the presence of a new bud indicates the initiation of a new round of the cell cycle without completing previous division properly. The staining of cell membranes with DiI revealed no cytoplasm connection between mother and daughter cell bodies in all the cells analyzed (n=190 cells from five independent cultures), which confirmed that cytoplasmic division had been performed. Calcofluor White (CFW) staining revealed a signal at the neck edges corresponding to the chitin ring; strikingly, a full septum failed to form between mother and daughter cell bodies in 51.1±4.0% of the 190 rebudded cells analyzed. When only re-budded cells with a tiny new bud were considered, the percentage of cells without a closed septum rose to 80.2±4.0% (n=110 from three independent cultures). In conclusion, rot1 mutant cells manifest a defect in septation, which impedes the proper completion of cell division.

Previous results have shown that rot1 cells accumulate at late stages of the cell division cycle and that they have defects in both the apical-isotropic growth switch and actin cytoskeleton polarization during cytokinesis. These processes are regulated by the Clb2 mitotic cyclin, which led us to analyze a possible connection between Rot1 and Clb2. Completion of cell division requires CDK inactivation. To investigate whether the accumulation of rot1 cells at the end of the cell cycle and the incorrect septation could be related to defects in the elimination of mitotic CDK, Clb2 protein level and associated kinase activity were assayed in tetO₇:ROT1 cells after incubation in the presence of doxycycline. Western analysis confirmed the presence of Clb2 in rot1 cells, and kinase assays revealed that the Clb2-associated kinase was active (Fig. 8A,B). Next, genetic interactions between CLB2 and ROT1 were analysed. It is known that mitotic exit mutants are hypersensitive to the overexpression of Clb2 (Jaspersen et al., 1998). Similarly, high levels of Clb2 are toxic when Rot1 is not fully functional (Fig. 8C). Remarkably, the deletion of the CLB2 gene suppressed the lethality of the tetO₇:ROT1 mutation, although it did not completely restore a wild-type growth rate (Fig. 8D). This result clearly indicates that the lethality of tetO₇:ROT1 strain is associated, at least partially, with the presence of the mitotic Clb2 cyclin and points to a functional antagonism between Rot1 and Clb2.

We wondered whether the inactivation of the CLB2 gene would also suppress the actin cytoskeleton polarization defects of the rot1 mutant. As described above, inactivation of ROT1 leads to an accumulation of cells in the late stages of cell division. However, no important changes in cell distribution other than a slight increase in the percentage of re-budded cells could be observed after inactivation of ROT1 in a clb2 background (Fig. 8E). This result suggests that the defect in actin cytoskeleton polarization to the neck region is at least partially suppressed; in fact, a certain degree of polarization to the neck could be detected in clb2 rot1 cells (Fig. 8E). Moreover, clb2 rot1 cells show some degree of apical growth compared with rot1 cells (see Fig. 2). In conclusion, the clb2 deletion alleviates the actin cytoskeleton defects associated to ROT1 inactivation.

The main mechanism involved in the inactivation of Clb2 is the degradation of the protein by the proteasome, which is triggered by the ubiquitylation of Clb2 by the APC ubiquitin ligase at the end of mitosis. For this reason, we introduced the tetO₇:ROT1 gene in mutant strains in the APC-dependent protein degradation machinery and searched for genetic interactions. As shown in Fig. 9A, growth was severely impeded when the tetO₇:ROT1 gene was partially repressed in mutant cells in either the proteasome (pre1 pre2) or APC (apc2 and cdc16). In relation to the APC activator protein Cdc20, the growth assay revealed that the ectopic expression of ROT1 from the tetO₇ promoter suppressed the lethality of the cdc20 strain at a high temperature (Fig. 9B). These genetic interactions between ROT1 and genes related to the proteasome or APC might reveal a function of Rot1 in protein degradation by the proteasome during the cell cycle.

Bearing in mind the previous results, we hypothesised that Rot1 could be involved in Clb2 protein degradation. Therefore, Clb2 stability was assayed in promoter shut-off experiments with a GAL1:CLB2-HA gene. Clb2 was unstable in wild-type cells and was almost undetectable within 30 minutes after the repression of the GAL promoter by glucose. By contrast, Clb2 protein levels were affected only slightly in tetO₇:ROT1 cells, persisting for more than 120 minutes after promoter repression (Fig. 10A). This result demonstrated that a loss of Rot1 results in the stabilization of Clb2.

Degradation of Clb2 cyclin at the end of mitosis is controlled by the Cdc14 phosphatase, which also induces the accumulation of the CDK-inhibitor Sic1 (Bosi and Li 2005). The stabilization of Clb2 could arise from the inability to activate Cdc14. However, the fact that rot1 cells are able to complete mitosis suggests that Cdc14 is functional in the absence of Rot1. To clarify this point, the level of Clb2 and Sic1 proteins was analyzed after release from a telophase arrest induced by a thermosensitive cdc15 allele (Fig. 10B). In ROT1 cells, most of the Clb2 protein was degraded and Sic1 protein accumulated after incubation at the permissive temperature, which reflects Cdc14 activation. Later, degradation of Sic1 and the apparition of new buds mark the initiation of a new round of cell division. In the case of rot1 cells, Clb2 protein remained after the release form the arrest, confirming that Rot1 protein is required for its degradation. Interestingly however Sic1 protein accumulated. This result strongly suggests that Cdc14 is activated in these cells. Remarkably, in spite of the defect in Clb2 degradation, the rot1 cells are able to initiate new rounds of cell division (as deduced from the degradation of Sic1 and rebudding of cells) although with a delayed kinetic compared with that of ROT1 cells. In conclusion, the defect in Clb2 degradation in rot1 mutant cells was not due to a defect in Cdc14 function and might reflect a specific role of the protein in the control of Clb2 stability.

Previous results showed that in the absence of Rot1 degradation of Clb2 is not switched on. To further characterize the connection between Rot1 and Clb2 degradation we investigated the effect of the inactivation of ROT1 in Clb2 stability in G1 cells, in which Clb2 is highly unstable. As expected, Clb2 protein was nearly undetectable in α-factor-arrested wild-type cells just 10 minutes after translation shut-off with cycloheximide (Fig. 10C). A similar result was observed in tetO₇:ROT1 cells in the absence of doxycycline. However, when ROT1 was inactivated by the addition of doxycycline, Clb2 protein level remained unaltered after the translation shut-off. Thus, inactivation of Rot1 is able to turn off the degradation of Clb2 in G1 cells, which points to a direct role for Rot1 in the control of Clb2 protein stability.

In this study we have characterized a tetO₇:ROT1 mutant strain to investigate the cellular function of the essential ROT1 gene. Our work has revealed a role for Rot1 in cell morphogenesis during cell cycle progression. The correct regulation of cell morphogenesis is essential for many cellular processes in eukaryotic cells, such as cell division. In S. cerevisiae, different morphogenetic processes involving the actin cytoskeleton occur throughout the cell cycle. They include bud emergence, control of bud shape through the switch from apical to isotropic growth, action of an actomyosin contractile ring, which clinches the plasma membrane during cytokinesis, and polarization of growth to the neck region to form the septum and complete cell division. We have determined that Rot1 affects the actin cytoskeleton at two levels. First, rot1 cells were unable to polarize actin cytoskeleton to the neck region at the end of mitosis. Although the cytoplasm was split between mother and daughter cells, loss of polarized growth led to a defect in septum formation and the inability to properly complete cell division. It is necessary to point out that septum formation is not completely abolished in rot1 cells. This could be due to the `leaky' activity of the tetO₇:ROT1 gene. However, the fact that actin cytoskeleton polarization to the neck region was not observed in any of the rot1 cells analyzed argues against this possibility. On the contrary, this observation suggests that Rot1 is not strictly required to form a septum, rather to form a septum on schedule: cellular components required to form the septum will be delivered to the neck region more slowly in the absence of polarized growth. This is strongly supported by the significant increase in the percentage of cells lacking a septum when only cells at the early stages of the rebudding process are considered.

A second connection between ROT1 and the actin cytoskeleton was revealed by the round shape of buds in rot1 cells. This indicates that Rot1 is required for the maintenance of apical growth during bud growth. Different genes have been involved in the maintenance of apical growth (Sheu et al., 2000). Mutant cells in these genes also manifest an altered budding pattern because they present a random rather than axial budding pattern. Interestingly, rebudded rot1 cells also showed a random budding pattern for the new bud. This suggests a close relationship between the function of Rot1 and polarity proteins.

The involvement of Rot1 in the actin cytoskeleton is strongly reinforced by the genetic interaction between ROT1 and CDC42. Cdc42 is a master regulator of the actin cytoskeleton and, thus, cdc42 mutants are defective in the establishment of cytoskeleton polarity or the apical-isotropic switch (Adams et al., 1990; Richman et al., 1999). Suppression of the cdc42-1 mutant by the overexpression of Rot1 clearly indicates that high doses of Rot1 could compensate the defect in Cdc42 to induce actin cytoskeleton polarization. The synthetic interaction between rot1 and mutations in Rho1, Pkc1 and Slt2, proteins that are also involved in actin cytoskeleton regulation, further supports a role of Rot1 in the actin cytoskeleton. Previously published results also indicated that Rot1 might be involved in cellular morphogenesis. ROT1 was identified in a screening for suppressor mutations of a tor2^ts mutant strain, which is defective in the control of the actin cytoskeleton (Bickle et al., 1998). Moreover, mutant strains in ROT1 manifest cell wall defects (Bickle et al., 1998; Machi et al., 2004). It is known that defects in the cell wall activate the cell integrity pathway (Levin, 2005), and we have consistently observed a strong increase in the phosphorylated state of the Slt2 MAP kinase in rot1 cells. Defects in the actin cytoskeleton function owing to loss of Rot1 could certainly lead to defects in the cell wall and the activation of the cell integrity pathway.

As commented above, different morphogenetic processes involving the actin cytoskeleton are temporally regulated during cell cycle progression in S. cerevisiae. Rot1 is specifically involved in the apical-isotropic growth switch and the polarization to the neck region at the end of cell division. Conversely, it is not required for growth polarization during bud emergence or for the function of the actomyosin contractile ring. The fact that rot1 cells are able to polarize growth during bud emergence indicates that Rot1 is not required for actin cytoskeleton polarization per se. Thus, Rot1 must impinge on functions or proteins specifically involved in actin cytoskeleton regulation during the apical-isotropic switch and repolarization to the bud neck. As far as we know, the basic machinery involved in actin cytoskeleton polarization during bud emergence and at the end of mitosis is similar and involves the recruitment to the polarized growth site of the polarisome and the Cdc24 and Cdc42 proteins, with the consequent local activation of Cdc42. However, it is noteworthy that the temporal regulation of these processes through the cell cycle is carried out by different CDK activities - whereas actin cytoskeleton polarization during bud emergence is governed by the Cln1,2-Cdc28 kinases, the switch to isotropic growth and the polarization of growth at the end of mitosis are regulated by the Clb2-Cdc28 kinase. Thus, ROT1 mutant cells are specifically defective in the actin cytoskeleton functions regulated by Clb2-Cdc28. In fact, ROT1 inactivation suppresses the hyperpolarization defect in clb2 cells but not in CLN2-overexpressing cells (M.A.J. et al., unpublished results). Phenotypic traits and genetic interactions clearly indicated that ROT1 and CLB2 genes function antagonistically in the regulation of actin cytoskeleton - Clb2 is required for the switch to isotropic growth (Tjandra et al., 1998) and represses polarization to the neck (Lew and Reed, 1993), whereas Rot1 is required for apical growth maintenance and for polarized growth at the neck. The molecular basis of this antagonism is beginning to be unravelled. Our results indicate that Rot1 is involved in Clb2 degradation, which raises the possibility that Rot1 could control the actin cytoskeleton, at least partially, through the regulation of the Clb2 protein level. The genetic interactions we detected between ROT1 and genes involved in ubiquitin-mediated protein degradation, and the reported interaction between Rot1 and Nas2 (Ito et al., 2001), which is the yeast p27 proteasome-modulator protein (Russell et al., 1999), support a connection between Rot1 and proteasomal protein degradation. However, the fact that inactivation of ROT1 deactivates apical growth in a clb2Δ strain and that overexpression of ROT1 suppresses the thermosensitive growth defect of a cdc42-1 clb2Δ mutant (M.A.J. et al., unpublished results) indicates that Rot1 is able to affect actin cytoskeleton irrespective of Clb2. It cannot be ruled out that in the absence of Clb2, Rot1 could affect the actin cytoskeleton through the control of the redundant Clb1 cyclin. Nevertheless, Rot1 could also acts on the actin cytoskeleton through proteins other than Clb cyclins (see below). In this case, the observed stabilization of Clb2 in rot1 cells might reflect a feedback mechanism, which would block or delay Clb2 degradation in response to a defect in actin cytoskeleton function.

Rot1 is located in the nuclear membrane and ER network (Takeuchi et al., 2006). In light of the results we present here, different scenarios by which Rot1 could act in cellular morphogenesis can be envisaged. For instance, Rot1 might affect the secretory pathway, which is required for sorting to the cell surface enzymatic activities involved in cell wall biosynthesis and, maybe, cortical cues directing actin cytoskeleton polarization. In fact, the targeting of carboxypeptidase Y to the Golgi is affected in a rot1 mutant (Takeuchi et al., 2006), which is consistent with a defect in protein trafficking. An alternative, although not mutually exclusive, hypothesis emerges from the above-discussed connection between Rot1 and ubiquitin-mediated proteolysis. One of the major pathways of ubiquitin-mediated protein degradation is located at the ER - the ERAD (endoplasmic-reticulum-associated degradation) pathway. There is a synthetic interaction between rot1 and kar2 mutations (Takeuchi et al., 2006), which is consistent with Rot1 acting on ERAD. Interestingly, some proteins related to the actin cytoskeleton are ubiquitylated membrane-associated proteins and some of them could be substrates of the ERAD pathway (Hitchcock et al., 2003). Moreover, the Rot1-interacting and proteasome-interacting protein Nas2 is genetically linked to the polarisome (Drees et al., 2001), which further supports a role for the proteasome system in actin cytoskeleton regulation. It is tempting to speculate that Rot1 could affect the function of the actin cytoskeleton through the ERAD pathway. Regarding the role of Rot1 in the control of Clb2 protein stability, it is necessary to point out that components of the ERAD pathway are involved in the degradation of not only ER-located proteins, but also soluble proteins of the cytoplasm and nucleus (Deng and Hochstrasser, 2006). Work is in progress to characterize the molecular function of Rot1 in actin cytoskeleton regulation during the cell cycle. Given the functional antagonism between Rot1 and the Clb2 mitotic cyclin, this work will help to elucidate the poorly understood link between the cell cycle machinery and the cytoskeleton functions.

The yeast strains used in this study are shown in supplementary material Table S1. The substitution of the ROT1 promoter by the tetO₇ promoter was obtained by integrating a DNA fragment amplified from plasmid pCM225 (from E. Herrero, Universitat de Lleida, Spain). Tagging of the Rot1, Clb2, Sic1 and Myo1 proteins at the C-terminus was achieved by integrating a DNA fragment amplified from pFA6a plasmids (from J. R. Pringle, University of North Carolina, Chapel Hill). A centromeric plasmid containing the ROT1 gene was constructed by cloning a PCR-amplified fragment expanding from -430 to +955 of the ROT1 gene in an EcoRI-BamHI-cleaved pRS314 vector. Subsequently, an EcoRI-BamHI restriction fragment from this plasmid was subcloned in YEplac112 to obtain plasmid pROT1-2 μ. Plasmids pRS426Spa2GFP, pYS47 (CYC1-CDC24-GFP), p416MET(GFP-CDC42) and pGAL:CLB2-3HA were gifts from R. Arkowitz (University of Nice, France), M. Peter [Swiss Institute for Experimental Cancer Research (ISREC), Epalinges], D. I. Johnnson (University of Vermont, Burlington) and L. H. Johnston (NIMR, London, UK), respectively.

Cells were grown on standard yeast extract-peptone-dextrose (YPD) or synthetic dextrose (SD), galactose (SGal) or raffinose (SRaf) medium supplemented as required. To fully repress the tetO₇ promoter, doxycycline was added to a concentration of 5 μg/ml; partial repression of the tetO₇ promoter was achieved by adjusting doxycycline concentration to 0.5 μg/ml. For cell cycle synchronization, MATa cells were arrested with 5 μg/ml α mating pheromone for 4 hours and then released into fresh growth medium. In the case of the tetO₇:ROT1 strains, 5 μg/ml doxycycline was added during the incubation with the α-factor and after the release from the arrest.

F-actin and microtubule staining were performed on cells fixed overnight at 4°C in 4% formaldehyde. F-actin was visualized with Alexa Fluor 488-labeled phalloidin (Molecular Probes) at a concentration of 0.1 mg/ml. Microtubules were detected as described previously (Queralt and Igual, 2005). When only DNA staining was required, cells were fixed for 5 minutes by addition of 70% ethanol, washed in H₂O and incubated with 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI, Sigma). For cell membrane staining, cells were resuspended in 95% ethanol containing 0.1 mg/ml 1,1-dioctadecyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate (DiI, Promega). After incubation for 20 minutes at 37°C, cells were washed twice in H₂O. For chitin staining, cells were resuspended in 0.1 mg/ml Calcofluor White (Sigma). GFP-tagged proteins were analyzed in living cells grown on selective medium. Samples were analyzed in an Axioskop 2 fluorescence microscope (Zeiss) and pictures were taken with a SPOT digital camera (Diagnostic Instruments). Confocal microscopy assays were performed in a Leica TCS spectrophotometer confocal microscope; serial sections were obtained with a 140-nm interval and images were analyzed with the Leica Confocal Software (LCS).

To evaluate the Clb2 protein decay, strains carrying pGAL1:CLB2-3HA plasmid were grown on SRaf and incubated in the presence of 5 μg/ml doxycycline for 6 hours. Then, galactose to a final concentration of 2% galactose was added. After 30 minutes, cells were collected and resuspended in SD. Samples were harvested at different times and processed for western blot analysis. To evaluate the Clb2 protein decay in G1-synchronized cells, cells were grown on SRaf and arrested by α-factor in the presence of 5 μg/ml doxycycline. After 4 hours, galactose was added to induce the GAL1 promoter, and after 30 minutes, cycloheximide was added to block protein synthesis.

Western blot analysis, immunoprecipitation, kinase activity assay and fluorescence-activated cell sorter (FACS) analysis, were carried out as described previously (Queralt and Igual, 2005). Antibodies used were: anti-HA 12C5A antibody (Roche), anti-Myc 9E10 antibody (Roche), anti-PSTAIRE antibody (Santa Cruz) and phospho-p44/42 MAPK antibody (Cell Signalling Technology) for western analysis, and anti-HA 3F10 antibody (Roche) for immunoprecipitation of Clb2-HA.

We are very grateful to A. Arkowitz, B. Futcher, L. Hartwell, E. Herrero, D. I. Johnson, L. H. Johnston, D. Levin, M. Peter, J. R. Pringle, Y. Takai, M. Tyers, F. Uhlmann and D. H. Wolf for kindly supplying plasmids and strains. This work was supported by Grants BFU2004-05763-CO2-02 from Ministerio de Ciencia y Tecnología of the Spanish Governement and GRUPOS 04/23 and ACOMP06/030 from Generalitat Valenciana. M.A.J. is recipient of a Predoctoral Fellowship from Generalitat Valenciana.