Cdc28-Clb mitotic kinase negatively regulates bud site assembly in the budding yeast

Giving rise to a healthy progeny is an intricate process even for a lower eukaryote such as the budding yeast Saccharomyces cerevisiae. The construction of a bud in S. cerevisiae involves complex space-time coordination of various cellular events. The process of budding begins with the establishment of spatial polarity in the mother cell, the first indication of which is the assembly of an incipient bud site in the cortex of the mother. The bud site can be positioned in two distinct patterns; in haploid cells, it is located adjacent to the previous bud site (axial), whereas in diploids, it can also form at a distal position (bipolar) (Chant, 1996; Roemer et al., 1996). Subsequently, the bud site is used as a spatial marker for the reorganization of the actin cytoskeleton along which the raw material for the construction of a bud is transported (Chant, 1996; Roemer et al., 1996; Madden and Snyder, 1998). The new bud finally emerges as a small protrusion and continues to grow through most of the division cycle.

As the nature of the bud site assembly remains largely unknown, the translocation of various bud site components to a specific position in the cortex is generally referred to as the bud site assembly. However, not all bud site components get translocated to the cell’s periphery at the same time; some are localised to the site earlier then others. For example, Spa2 is one of the earliest to arrive at the bud site while Sec3 is localised somewhat later (Snyder, 1989; Gehrung and Snyder, 1990; Finger et al., 1998). Although they first appear at the bud site in G₁, some bud site components display intriguing spatial dynamics. Spa2, which first localises at the cortex, is found at the tip of the newly formed bud as soon as the bud emerges from the cortex. It remains there until late S phase, and then it relocates to the mother-bud neck (Snyder, 1989; Gehrung and Snyder, 1990). The physiological significance of these complex dynamics is not clear. It has been proposed that the presence of some of the axial bud site selection components and the 10 nm filaments at the neck may serve as a spatial cue for the positioning of the subsequent bud site in haploid cells (Madden and Snyder, 1998).

The morphogenesis of a bud involves controls at various levels: (1) bud site selection, (2) signal transduction between the selected site of growth and the cytoskeleton, (3) polarized secretion and (4) the molding of the bud in an appropriate shape. As would be expected, the process of budding requires a variety of proteins, many of which are themselves components of the bud site. For instance, Bud3 and Bud4 are retained at the bud scar (old bud site), presumably to serve as a spatial cue for the establishment of a new bud site in an axial position during the next cell cycle (Chant et al., 1995; Chant, 1996). Bud7, Bud8 and Bud9 are the key executors of the bipolar pathway (Zahner et al., 1996). Similarly, Bud1, Bud2 and Bud5 comprise a GTPase module that reads the pattern specified by the axial- or bipolar-specific genes and communicates this signal to a second GTPase module with the Rho family GTPase Cdc42 at its core (Bender and Pringle, 1989; Chant et al., 1991; Zheng et al., 1995). The activated Cdc42 plays a central role in signal transduction via various effectors including Gic1, Gic2 (Chen et al., 1997; Brown et al., 1997), Ste20 (Peter et al., 1996; Eby et al., 1998) and Bni1 (Imamura et al., 1997; Evangelista et al., 1997) that eventually leads to actin nucleation and reorganization of the actin cytoskeleton. Both actin-dependent and actin-independent pathways are deployed in the morphogenesis of a bud. While the actin-dependent pathway is responsible for the vectorial transport of certain secretory vesicles and a putative vesicle- docking complex to the sites of cell surface growth (Finger and Novick, 1998), proteins like Sec3 use an actin-independent pathway for their polarized distribution (Finger et al., 1998).

In addition to the spatial controls, construction of a bud must also be temporally integrated with the other events of the cell division cycle. Indeed, the initiation of bud formation is restricted to the G₁ phase and is therefore bounded by two major cell cycle transitions: (1) completion of the preceding M phase (M to G₁ transition) (2) commitment to a new division cycle (START). Traditionally, emergence of a bud has been considered as an indicator of a cell’s commitment and entry to a new cycle since it is the most visible landmark. Therefore, the START-dependent aspects of bud emergence have been studied in some detail. The critical role of the START-kinase (Cdc28-Cln) in the emergence of a bud is most obviously underlined by the fact that the cells lacking Cdc28- Cln kinase activity arrest in G₁ without a bud (Reed, 1992; Nasmyth, 1993). The Cln-kinase is also believed to activate Cdc42, which in turn triggers the reorganization of the actin cytoskeleton (Lew and Reed, 1995a; Shimada et al., 2000). Among the post-START regulators that affect the budding process is the Cdc28-Clb kinase (the mitotic kinase), which helps to switch bud-growth from apical to an isotropic mode. Bud formation is also intertwined with the regulation of nuclear division in that a morphogenetic checkpoint mechanism, involving Cdc28 and Swe1, delays the onset of nuclear division in the event that a cell fails to efficiently execute the bud emergence program (Lew and Reed, 1995b). A host of such elegant studies have identified a variety of effectors that affect the process of budding. They have also led to a general notion that most cellular events relevant to the bud formation may be triggered upon passage through START (Reed, 1992; Nasmyth, 1993).

Although the temporal dependence of bud formation on a successful completion of mitosis has been obvious, the regulatory relationship between the exit from mitosis and the process of budding has not attracted as much attention thus far. The final departure from mitosis, facilitated by a dramatic inactivation of the Cdc28-Clb kinase via proteolytic destruction of the mitotic cyclins, is a prerequisite for the entry into subsequent G₁ phase. The ubiquitin-dependent degradation of cyclins in late telophase requires a multisubunit ubiquitin ligase (E3) called the Anaphase-Promoting Complex (APC), which also catalyzes chromosome segregation during the metaphase to anaphase transition (Townsley and Ruderman, 1998; Zachariae and Nasmyth, 1999). The regulation of APC is complex and is now being extensively investigated. However, other important questions connected to mitotic exit remain. Why must a cell inactivate the mitotic kinase in order to gain entry into the subsequent division cycle? Does mitotic kinase inhibit the cellular events that mark the entry into a new cycle? How do cells restrict the initiation of bud formation to the G₁ phase? In what follows, we address some of these issues in the context of a functional relationship between the mitotic kinase and the process of budding.

All strains used in this study were haploids and were derived from the wild-type W303 genetic background. Cells were grown in standard yeast-extract peptone supplemented with 2% glucose or raffinose- galactose.

In experiments involving the cdc28-4 mutant, cells were released from stationary phase into medium prewarmed at restrictive temperature (36°C) so that cells arrested predominantly prior to START (since this mutant is also defective in entry into mitosis) and also to eliminate any spatial cues that may be remnants from the previous cycle. To obtain stationary phase cultures, cells were plated on YEP-glucose and incubated at 24°C for 60-72 hours. Cells were scraped and inoculated directly into YEP-glucose at a starting optical density at 600 nm (OD₆₀₀) of 0.2-0.3 and subsequently incubated at 36°C for 3.5 hours. Samples were collected for confocal microscopy, immunofluorescence and flow cytometric analysis at various times. In other experiments where synchronization at the prenuclear division stage was required, log phase cultures were treated with nocodazole to a final concentration of 15 μg/ml at 24°C for 3 hours. In experiments involving the repression of expression from a MET3- promoter, methionine was added to the medium to a final concentration of 24 μg/ml. For disruption of actin cytoskeleton, Latrunculin-A (Molecular Probes, Inc.) was added to a final concentration of 100 μM from a 10 mM stock solution in DMSO.

For preparation of cell extracts to determine Cdc28/Clb2 kinase activity, cells were lysed by vortexing vigorously with acid-washed beads and an appropriate volume of lysis buffer. Extracts were cleared by centrifugation and the resulting supernatant was used to assay mitotic kinase activity. Immunoprecipitation, using polyclonal antibodies against Clb2, and kinase assays were performed as described (Surana et al., 1993). The kinase activity was quantitated using the PhosphorImager (Molecular Dynamics). For western blot analysis, 50 μg of cell extracts were resolved by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membrane. The membrane was probed with anti-Cdc28 antibodies (1:500 dilution) and then with anti-Clb2 antibodies (1:500 dilution). All western blot analyses were performed using the enhanced chemiluminescence kit (Amersham) according to the manufacturer’s instructions.

RNA extraction was done according to the method of Cross and Tinkelenberg (Cross and Tinkelenberg, 1991) and northern blot analysis was performed as described (Price et al., 1991). Cells were prepared for FACScan analysis (flow cytometry) (Lim et al., 1996). For immunofluorescence, samples were prepared and stained as described (Kilmartin and Adams, 1984), and nuclei were visualized by staining with diamidinophenylindole (DAPI).

In experiments involving elutriation of daughter cells, the elutriated population was resuspended in Calcofluor solution (1 mg/ml) (Fluorescent Brightener 28, Sigma), and washed two times with distilled water. Cells with bud scars were quantitated under UV excitatory illumination. For visualization of the actin cytoskeleton, cells were first fixed in 3.7% formaldehyde solution and then stained with rhodamine-tagged phalloidin (Molecular Probes, Inc.).

To observe localisation of Spa2-GFP, cells were washed and resuspended in a solution containing 1.2 M sorbitol, 0.1 M phosphocitrate (pH 5.9) before snap-freezing in liquid nitrogen. Spa2- GFP was visualised by confocal microscopy using the Bio-Rad MRC- 1024 scanhead (Bio-Rad, Richmond, CA, USA) connected to an axiophot microscope (Zeiss, Thornwood, NY, USA) equipped with epifluorescence optics. For time-lapse microscopy, cells were spotted onto a glass slide coated with a very thin layer of minimal medium containing 25% gelatin (300 Bloom, Sigma) and were observed by confocal microscopy at various time points as they progressed through the cell cycle.

Since the assembly of a bud site precedes the actual emergence of a bud, we deemed it a more appropriate indicator of a cell’s decision to initiate the budding process. Therefore, we used asymmetric localization of Spa2 to serve as a marker for the initiation of bud site assembly as it is perhaps the one of the bud site components known to arrive earliest at the cortex (Snyder, 1989; Gehrung and Snyder, 1990). For the purpose of visualization, we used yeast strains in which Spa2 had been tagged with the Green Fluorescent Protein (GFP). That this SPA2-GFP construct was physiologically functional was indicated by its ability to complement the shmoo formation defect of spa2Δ strain (M. Snyder, personal communication).

First, we asked when during the cell cycle progression is the bud site assembled. The temperature-sensitive cdc15-2 cells carrying native promoter-driven SPA2-GFP on a CEN vector were synchronized in late telophase by growth at 36°C and then allowed to resume progression through the division cycle at 16°C. We used 16°C temperature to slow down the progression so that the temporal sequence of the post-telophase events could be discerned more clearly. Samples were withdrawn at various times and were analyzed for the levels of the Cdc28- Clb2 kinase activity, CLN1 RNA, budding index and the localization of Spa2-GFP. The kinase activity begins to decline at 30 minutes after the release from telophase arrest and reaches its minimum at around 60 minutes (Fig. 1A). Spa2- GFP, seen as dispersed patches in the telophase-arrested cells (0 minutes), first appears at the neck region after the release (30 minutes) and then at the cortex. The proportion of cells that show Spa2 localization at the cortex rises almost immediately after the inactivation of the mitotic kinase followed by the appearance of CLN1 RNA and the nascent buds. However, the expression of CLN1, which is normally transcribed as cells undergo START, is seen 15 minutes after approximately 60% of the cells have already translocated Spa2 to the incipient bud site. This suggests that Spa2 is localized to the cortex considerably before cells transit through START. The bud sites were easily distinguishable from the site of cytokinesis since in this strain, cell separation is delayed with respect to bud emergence.

To determine more directly if the bud site assembly requires passage through START, we tested cdc28-4 mutant cells for their ability to localize Spa2 to the cortex at the restrictive temperature. The cdc28-4 cells are unable to traverse START at 36°C and therefore arrest in G₁ without buds. Stationary phase cells, obtained by growth for 3 days, were reinoculated into fresh growth medium at 36°C and were analyzed for their ability to assemble a bud site. As expected, at the end of 3.5 hours, all cells had grown in size but remained unbudded. Approximately 90% of the cells showed a bright crescent at the cortex (Fig. 1B, top row) indicating that the translocation of Spa2 can occur in the absence of Cdc28 function. This observation is identical to what has been reported previously (Snyder et al., 1991). When these cells were subjected to immunoelectron-microscopy using anti-GFP antibodies, Spa2- GFP was seen concentrated at the sub-cortical region near the cell periphery (data not included). The localization of Spa2 in cdc28-4 cells was found to be actin-independent since it was insensitive to Latrunculin-A (LAT-A) treatment. The fate of Spa2-GFP protein was also monitored in another START- defective mutant cln1Δcln2Δcln3Δ (US1624), kept alive by GAL-CLN3 on a CEN vector. When the stationary phase cells were reinoculated into glucose medium at 24°C, they grew in size, and as expected, remained unbudded. The majority of these cells translocated Spa2 to the cortex within 4 hours (Fig. 1B, 2^nd row). As in cdc28-4 cells, Spa2 localization in this strain remained unperturbed by LAT-A treatment. Surprisingly, a proportion (25%) of these cells exhibited two crescents on opposite sides of the cortex. The reason behind this anomalous behavior in the absence of CLN functions is not clear at present. That the cortical site at which Spa2 localizes in these experiments is indeed a prospective bud site was confirmed using time-lapse microscopy (see Fig. 5).

To further test the validity of our observations, the localization of other bud site components was monitored in cdc28-4 cells. The stationary phase cells harboring SEC3-GFP, GFP-CDC42 or BNI-GFP were treated as described above. Consistent with the earlier reports that translocation of Sec3 to bud site requires Cdc28-Cln kinase activity (Finger et al., 1998), none of the cells showed any cortical localization of Sec3-GFP (Fig. 1B; 3^rd row). Similarly, the majority of cells also did not exhibit cortical staining of GFP-Cdc42, though a small proportion (29%) did contain GFP-Cdc42 patches at the cell periphery (Fig. 1B, 4^th row). In contrast, Bni1-GFP showed clear cortical localization in these cells (>75%). Since the arrival of Bni1 to the cortex is reported to be dependent on Spa2 localization (Madden and Snyder, 1998), this observation implies that Spa2 localization seen in our experiments is physiologically consistent. Moreover, the kinetics of bud site assembly in cdc28-4 and triple-clnΔ strains are very similar to those seen in the wild-type strain when cells are released from stationary phase (Fig. 1C). In a separate experiment, we have found that Spa2 localization in cdc28-4 mutant is also dependent on Cdc42 (data not shown). This, together with our finding that cdc28-4 cells can localize Spa2 to the cortex even though they are largely unable to translocate Cdc42 (Fig. 1B), suggests that while Cdc42 function is required for the translocation of Spa2 at the cell periphery, its presence at the cortex may not be necessary for this function.

There is another concern. In haploid cells the new bud site is laid in juxtaposition to the old site called bud scar. The old site with any residual Spa2 could be misconstrued as the newly assembled site. To alleviate this suspicion, freshly elutriated daughter cells (devoid of any bud scars) from cdc28-4 strain carrying SPA2-GFP construct were released into growth medium prewarmed at 36°C. No uniquely localized Spa2-GFP fluorescence was seen in the daughter cells immediately after elutriation, indicating the absence of residual Spa2 remaining at any specific location in the cortex (Fig. 1D, left panel). After 90 minutes at 36°C, however, the majority of the cells contained clearly visible Spa2-GFP patches at the cell periphery (Fig. 1C, right panel). Taken together, these results suggest that the assembly of a bud site does not require the activity of Cdc28-Cln kinase and that most likely, the bud site assembly commences prior to passage through START.

If the initiation of bud site assembly is not dependent on the passage through START, what restricts its occurrence to the G₁ phase? The correlation between the sharp decline in the Cdc28- Clb2 kinase activity during the mitotic exit and the pre-START appearance of Spa2 (Fig. 1) at the cortex raises the possibility that inactivation of the mitotic kinase may be a prerequisite for the initiation of bud site assembly. We reasoned that if Cdc28- Clb kinase is a critical inhibitor of the budding process, then its inactivation at an inappropriate time in the cell cycle could prematurely trigger the onset of bud formation. To test this, wild-type cells carrying four copies of GAL-SIC1 integrated at the URA3 locus and native promoter-driven SPA2-GFP on a CEN vector were treated with nocodazole (Noc) in raffinose medium at 24°C to synchronize them in the pre-nuclear division state. At this stage of the cell cycle, Spa2 is seen clearly at the mother bud neck (Fig. 2A, middle panel). When Sic1 expression is induced by the addition of galactose, the Cdc28-Clb2 kinase activity declines rapidly without any dramatic loss of Clb2 (Fig. 2A, upper panel). Spa2-GFP fluorescence diminishes at the neck region and appears at the cortex; after 1 hour buds begin to emerge and Spa2 is seen at the tips of the new buds (middle panel). The nucleus in these cells remained undivided throughout the experiment (graph in the upper panel). As expected, the newly formed buds are elongated due to hyper-polarization caused by the lack of Cdc28-Clb kinase activity (Lew and Reed, 1995a). The new buds show conspicuous accumulation of actin patches, whereas cells maintained in raffinose do not show any discernable actin polarization (Fig. 2A). These results argue that the untimely inactivation of the mitotic kinase alone is sufficient to drive cells to establish new bud sites at an inappropriate stage of the division cycle. Moreover, the observation that there was no decrease in Clb2 levels during the course of the experiment implies that it is not the level of the Clbs per se but rather the activity of the mitotic kinase that has to be reduced in order to permit bud site assembly.

Since the mitotic kinase can also be inactivated by phosphorylation of Cdc28 on Tyr19 by the tyrosine kinase Swe1, we tested if overexpression of Swe1 can also lead to premature establishment of new bud sites. The wild-type cells carrying SPA2-GFP and GAL-SWE1 on CEN vectors were synchronized in the prenuclear division stage by nocodazole treatment in raffinose medium and then induced to express Swe1 by the addition of galactose. As before, Spa2 remained near the neck region in raffinose medium but was seen at the cortex within 45 minutes of Swe1 induction (Fig. 2B, middle panel, arrows). As the new buds emerged from the freshly assembled sites, Spa2 moved to the bud tips (middle and lower panels). Swe1 expression caused only moderate reduction in the capacity of Cdc28-Clb2 kinase to phosphorylate histone H1 and did not result in a decline in the abundance of the Clb2 protein (Fig. 2B, upper panel). This is consistent with a previously published report that overexpression of Swe1 in wild-type cells completely abolishes the in vivo function of the mitotic kinase, though only moderately diminishes its ability to phosphorylate histone H1 (Lim et al., 1996). The induction of new buds due to SWE1 overexpression has been noted previously (Booher et al., 1993) in a different context. These observations imply that the mitotic kinase is a critical negative regulator of the bud site assembly such that in its absence the assembly process is triggered prematurely. To confirm that the induction of budding due to overexpression of Swe1 is indeed mediated via Cdc28, we carried out the same experiment in a strain where the endogenous CDC28 gene had been replaced with constitutively active CDC28Y19F allele. Expression of Swe1 in this strain failed to induce budding whereas more than 70% of cells formed new buds in the wild-type controls (Fig. 2C). Moreover, there was no translocation of Spa2 to the prospective bud site (data not shown). These data imply that the effect of Swe1 is mediated directly through Cdc28. They also suggest that in cells that have entered mitosis, the dephosphorylation of Tyr19 residue is specifically required to prevent untimely assembly of a new bud site.

To verify that the establishment of bud site in nocodazole-arrested cells is independent of Cln-kinase activity, a strain deleted for all three G₁ cyclins carrying GAL-CLN3 on a CEN plasmid and MET-SWE1 on a 2 μm plasmid was treated with nocodazole in galactose medium for 3 hours. Subsequently, the cells were transferred to glucose medium for 1 hour to terminate CLN3 expression (Fig. 2D, northern blot). SWE1 expression was induced from MET3 promoter in one half of the culture by transferring the cells to medium lacking methionine. Though all cells remained unbudded due to a lack of G₁ cyclins, approx. 80% of them were able to localize Spa2 to the cortex within 3 hours of Swe1 induction (Fig. 2D, −Met, red arrows). Moreover, these cells showed no visible polarization of actin patches. In cells where SWE1 was not induced, Spa2 remained at the mother-bud neck and did not translocate to the cortex (+Met). This experiment reconfirms our earlier assertion that Cln-kinase activity is not required for bud site establishment.

We wished to determine the stage of the division cycle during which cells first become ‘sensitized’ to the loss of mitotic kinase activity and respond by initiating bud formation. The three cell cycle stages we selected were early S phase and G₂/M (or prenuclear division stage), as defined by the arrest points of the cdc8, cdc13 and cdc23 mutants, respectively, at the minimal restrictive temperature (30°C). Exponentially growing cdc8 and cdc13 mutant cells carrying both GAL-SWE1 and the native promoter-driven SPA2-GFP on CEN vectors, and cdc23 cells with GAL-SWE1 integrated at the TRP1 locus, were shifted to 30°C. As a control, a wild-type strain carrying the same two constructs was treated with nocodazole at 24°C. After >95% of the cells had arrested at their respective positions in the cell cycle, the cultures were induced to express Swe1 by the addition of galactose and were monitored for both the localization of Spa2 at the cortex and the emergence of new buds. While 40-50% of the cdc13-1 and cdc23-1 cells formed buds within 2-3 hours, the cdc8-1 mutant showed no significant rebudding (Fig. 3). The nuclei in cdc13 and cdc23 cells remained undivided and the new buds showed normal Cdc11 staining at their bases, normal actin polarization and chitin distribution (data not shown). However, the proportions of cdc13 and cdc23 mutant cells that budded in response to Swe1 expression are lower (40-50%) compared to the Noc-treated wild-type cells (80%). The reason for this difference is not clear at present. Nevertheless, these results suggest that cells become capable of producing new buds in response to Swe1- mediated inactivation of Cdc28-Clb kinase some time in G₂ or M phase (also see Discussion). These observations also suggest that premature bud formation in response to the inactivation of mitotic kinase is not specific to cells treated with nocodazole.

The fact that untimely inactivation of the mitotic kinase can lead to premature assembly of a bud site strongly argues that the mitotic kinase is a potent inhibitor of the bud site assembly. If so, then the presence of the mitotic kinase in G₁ should prevent cells from assembling a new site. To test this, we used two cln1Δcln2Δcln3Δ strains, both containing GAL-CLN3 and SPA2-GFP, but only one of them harboring a multicopy vector carrying clb2dbΔ (CLB2 lacking the destruction-box-sequence) under the control of the MET3 promoter. Stationary phase cells were resuspended in fresh glucose medium with or without methionine at 24°C. After 4 hours, cell samples were withdrawn and analyzed for the level of Cdc28-Clb2 kinase activity and for the localization of Spa2. As expected, cells in both cultures remained unbudded. The cells without the clb2dbΔ plasmid showed little kinase activity (Fig. 4, upper panel but >70% of them exhibited Spa2-GFP fluorescence at the cortex (lower panel). The cells expressing clb2dbΔ, on the other hand, contained substantially high level of the corresponding kinase activity (Fig. 4, lower panel). However, only approx. 28% of these cells showed cortical localization of Spa2, suggesting that they were diminished in their capacity to initiate the assembly of a new bud site. These results are consistent with our notion that the mitotic kinase acts as a negative regulator of the incipient bud site establishment and explain why inactivation of the mitotic kinase at the completion of the preceding mitosis is a prerequisite for the G₁-specific events such as bud formation.

Thus far we have referred to the fluorescent patch (or crescent) of Spa2-GFP seen at the cortex as the bud site. To ascertain that these patches are indeed the future bud sites, stationary phase cells of cdc28-4 strain carrying SPA2-GFP on a CEN plasmid were resuspended in fresh growth medium at 36°C. After 3.5 hours, when Spa2 had been translocated to the cortex in approx. 80% of the cells (Fig. 5A), the culture was shifted to 24°C. Cell samples were withdrawn every 5 minutes and were monitored for both Spa2 localization and bud emergence. As the nascent buds began to appear at 50 minutes after the release, the Spa2 patches, which were first observed at the cortex at 36°C, appeared at the tips of the newly formed buds (Fig. 5A). As expected, actin patches accumulated in the emerging buds. The decline in the proportion of cells with a cortical Spa2 patch was found to be inversely proportional to the rise in the number of budded cells with Spa2 at the bud tips. We observed neither any dispersal of Spa2 from the cortex nor appearance of an extra patch at any time during the release at 24°C. This implies that the Spa2 patches eventually seen at the bud-tips are most likely the ones that first appeared at the cortex during the incubation at 36°C. It should be noted that in Fig. 5A, cells showing GFP fluorescence and the corresponding actin polarization are two different sets of cells from the same sample.

To confirm further that the site where Spa2 localizes at the cortex is the prospective bud site, stationary phase cells of cdc28-4 strain carrying either SPA2-GFP or BNI1-GFP on a CEN plasmid were inoculated in fresh medium at 36°C. After 3 hours, when most cells showed a green crescent at the cortex, they were shifted to 24°C and subjected to time-lapse microscopy to observe the emergence of new buds. In most cells, which budded in a given microscopic field, tiny buds with a GFP patch at the tip could be seen after 2-2.5 hours at the same position where the GFP crescent was initially located. As expected of cdc28-4 cells released from G₁ arrest, many newly formed buds exhibited distorted morphology. Fig. 5B shows two representative cells, one carrying Spa2-GFP and the other Bni1-GFP, photographed at 0, 2.5 and 3.5 hours after the release at 24°C. Thus, collectively our data suggest that the cortical sites where Spa2 localizes prior to START are most likely the future sites for bud emergence.

Our assertion that the Cdc28-Clb mitotic kinase negatively regulates the establishment of the incipient bud site raises an important question. What is the mechanism by which mitotic kinase prevents bud site assembly? One possibility is that the kinase prevents localization of bud site components by inhibiting the transport machinery that translocates them to the cortex. Since such a transport mechanism has so far not been described, we first sought to identify genes whose products may participate in translocation of bud site components, such as Spa2. Hence, we wished to isolate temperature-sensitive mutants that were unable to bud, not because of their failure to undergo START, but due to their inability to assemble a bud site. Furthermore, such mutants, although unbudded, would be competent to enter S phase at the nonpermissive temperature. We sought such mutants from a bank of 1100 temperature- sensitive mutants, originally selected only for their inability to grow at 36°C (kindly provided by Usha Vijayraghavan). Using microscopic examination and flow cytometry analyses, we identified 30 mutants that were unable to bud at 36°C but underwent DNA replication. These mutants were then transformed with a CEN plasmid carrying Spa2-GFP in order to further select those that failed to assemble a bud site, i.e. to localize Spa2 to the cortex. Of the 30 mutants, five were found unable to localize Spa2-GFP to the cortex at 36°C. The terminal phenotype of each of the five mutants was due to a single recessive mutation. Upon further characterization, two of these mutants were found to carry mutations in SEC2 and SEC27 genes, respectively. SEC27, which we chose to study further, is thought to be primarily involved in retrograde transport from ER (Endoplasmic Reticulum) to Golgi (Letourneur et al., 1994) and is a member of the Coatomer COP1 complex.

When incubated at the restrictive temperature (31°C or 37°C), 70% of the sec27 mutant cells (designated as sec27-b1) remained unbudded but showed 2N DNA content; approximately 10% of these cells had undergone nuclear division. The remaining 30% had barely visible buds, which did not grow in size even after prolonged incubation (data not shown). In cells that remained unbudded, Spa2-GFP did not localize to a specific cortical position but instead showed a punctate distribution (Fig. 6A). It is noteworthy that Spa2-GFP was significantly mislocalized in sec27-b1 even at the permissive temperature (24°C) (Fig. 6A).

To test if Sec27 function is indeed required for Spa2 localization in cdc28-4 mutant, we constructed a cdc28-4 sec27-b1 double mutant carrying a SPA2-GFP construct on a CEN plasmid. The stationary phase cells were allowed to resume cell cycle progression at 36°C as described before. While the cortical localization of Spa2 was seen in >80% of the control cdc28-4 cells, only 30% of the cdc28-4 sec27-b1 mutant cells showed clear Spa2 localization at the cortex (Fig. 6B), implying that Sec27 may play a significant role in translocation of Spa2 to the bud site. We have tested cortical localization of Spa2 in strains carrying mutant alleles of other COPI subunits (ret1-1, ret2-1, ret3-1 and sec21-1) in combination with cdc28-4. None of the double mutant cells were able to localize Spa2 to the cell periphery at the restrictive temperature (data not shown).

If Sec27 is indeed important in translocation of the bud site component to the cell cortex, then the premature bud site assembly in response to the inactivation of the mitotic kinase (Fig. 2) should be suppressed in these mutants. To test this, the wild-type and sec27-b1 cells carrying both SPA2-GFP and GAL-SWE1 on CEN vectors were synchronized in the prenuclear division state by treatment with nocodazole at 24°C for 3 hours and then shifted to 31°C for 1 hour to inactivate the mutant Sec27 protein before Swe1 expression was induced by the addition of galactose. As expected, within 2 hours of Swe1 induction, most wild-type cells had translocated Spa2 to the cortex (data not shown) and approx. 60% of them eventually formed buds (Fig. 6C). However, only about 20% of the sec27-b1 cells exhibited Spa2 patches at the cell cortex and formed new buds. Hence, the lack of Sec27 function significantly suppresses premature bud site assembly and bud formation in response to Swe1 expression. The requirement of a functional Sec27 for premature bud formation underscores the central importance of vesicle transport mechanisms in bud site establishment irrespective of the cell cycle stage. These results also raise the possibility that the mitotic kinase may prevent untimely assembly of a bud site, by inhibiting, directly or indirectly, the premature activation of the transport machinery. We are currently investigating this issue.

The activation of Cdc28-Cln kinase, induction of START-dependent transcription and bud emergence are some of the major events that mark a cell’s entry into a new division cycle. However, these activities are not triggered until cells have completed the preceding mitosis by inactivating the mitotic kinase Cdc28-Clb. Why is the inactivation of this kinase a prerequisite for the initiation of these events? In this report we have addressed this question with a specific focus on the functional relationship between bud formation and the mitotic kinase. We have shown that the bud site assembly is critically dependent on the inactivation of the Cdc28-Clb kinase. This is most dramatically illustrated by the observation that the kinase inactivation by either Sic1 or Swe1 overexpression is sufficient to prematurely establish a new bud site and form a new bud even prior to nuclear division (Fig. 2). The fact that the presence of the Cdc28-Clb2dbΔ kinase prevents pre-START localization of Spa2 in G₁ (Fig. 4) further argues that the mitotic kinase is a crucial negative regulator of the bud site assembly. We note that some of the observations presented here, i.e. cortical localization of Spa2 in cdc28-4 cells and emergence of buds in response to Swe1 overexpression, have been reported earlier in different contexts. However, our study explores extensively these disparate preliminary observations and provides a new regulatory framework for the coordination of morphogenesis with cell cycle.

The premature establishment of a bud site in response to the kinase inactivation by Swe1 was not restricted to the nocodazole-treated cells but was also observed in cdc13-1 and cdc23-1 mutants that arrest in late S or G₂ phase of the cell cycle, respectively (Fig. 3). The cdc8 mutant, however, did not initiate bud formation in response to Swe1 overexpression. Recently, it has been shown that the overexpression of proteolysis-resistant Sic1 causes bud formation in cdc8 mutant at the restrictive temperature (Haase and Reed, 1999). The explanation for this difference may lie in the possibility that the premature formation of a bud may be inhibited not only by the mitotic kinase but all Cdc28-Clb complexes, which include the S phase kinases Cdc28-Clb5/Clb6. It is known that while Sic1 inhibits the activity of both mitotic and S phase kinase complexes (Schwob et al., 1994), phosphorylation of Cdc28 by Swe1 affects mainly the mitotic kinase activity but does not prevent progression through S phase (Lim et al., 1996). The cdc8 mutant is likely to contain a higher amount of S phase kinase as compared to cdc13-1 and cdc23-1 mutants, which are known to exhibit substantial level of mitotic kinase activity. Thus, the ability of cdc8 mutant to suppress bud formation despite overexpression of Swe1 may be due to the presence of the high S phase kinase activity, which is effectively inhibited by Sic1 but not by the Swe1 kinase.

Thus, budding yeast cells acquire the ability to establish a new bud site, perhaps as early as the initiation of S phase. However, this capability is suppressed first by the S phase kinases, whose activities begin to rise shortly after passage through START. Subsequently, during the late S phase, G₂ and mitosis, it continues to be suppressed by the high level of the Clb-kinase activity. This inhibition is lifted only when the mitotic kinase is inactivated via catastrophic destruction of the associated cyclins during cells’ exit from mitosis. Hence, it is perhaps the collective ability of the Cdc28-Clb kinase complexes to inhibit bud site assembly that restricts the initiation of bud formation to the G₁ phase. That the mitotic kinase is a negative regulator of the bud site assembly is also of physiological significance to the persistent proteolysis of mitotic cyclins throughout G₁.

One serious concern is whether the Spa2 (and Bni1) crescents or patches, which we refer to as bud sites, are the prospective bud sites. Both the kinetic and time-lapse experiments (Fig. 5) suggest that buds indeed emerge from the cortical sites where Spa2 is localized. It must be noted that in some of our experiments (Fig. 2) the buds do not always form at an axial position as would normally be expected of haploid cells. We believe it may be due to the peculiarity of the genetic background (W303) of our strains since we have consistently noticed that in this background bud patterning is somewhat relaxed.

How does the mitotic kinase prevent bud site assembly? The isolation of sec2 and sec27 mutants in our genetic screen indicates that vesicular transport is important for the translocation of Spa2, and most likely of other bud site components, to the cell cortex. The suppression of premature budding in sec27-b1 mutant (Fig. 6) only suggests that the transport mechanism requiring Sec27 function may be responsible for the translocation of Spa2 from the mother-bud neck to the cell periphery when the kinase is untimely inactivated. It follows then that the ‘Spa2 transport machinery’ remains normally inactive during M phase as long as the mitotic kinase activity is high. Although further investigations will be required to ascertain whether Cdc28-Clb kinase directly inhibits any of the components of this secretory system, our data raise the possibility that this system could be a possible target of the mitotic kinase. However, it is equally likely that the mitotic kinase does not regulate the secretory system per se but it prevents Spa2 (and other bud site components) from localizing to the cortex. In mammalian cells, cdc2 mitotic kinase is implicated in the fragmentation of Golgi apparatus by inhibiting COPI vesicle docking (Lowe et al., 1998; Warren and Malhotra, 1998). One perplexing aspect of the translocation of bud site components like Spa2 (this study), Sec3 (Finger et al., 1998), Aip3/Bud6p (Jin and Amberg, 2000) and Myo2 (Ayscough et al., 1997) to the cortex is that they reach the cell periphery in an actin cytoskeleton-independent manner. It will be interesting to explore the nature of the mechanism by which the bud site components are transported to the prospective bud site and how this transport machinery achieves polarity in its cargo-delivery independently of the actin cytoskeleton.

Traditionally, the emergence of a bud is considered a prominent hallmark of a cell’s commitment to mitosis since it is the most visible morphological landmark. Therefore, it has been generally thought that the events relevant to bud emergence are triggered once cells have traversed START. Our finding that the bud site assembly is initiated prior to START (Fig. 1) seems contradictory to this belief. It implies that cells commence at least one aspect of the next division cycle (i.e. bud site assembly) soon after they exit the preceding mitosis, without having made what would traditionally be considered the commitment to the next cycle. As we have shown, the bud site assembly clearly occurs in both cdc28-4 (at the restrictive temperature) and in G₁ cyclin-deficient cells. However, if the site to which Spa2 is translocated in cdc28-4 mutant at 37°C is exclusively a bud site, what would its fate be if cells were challenged with mating pheromones; it is known that cdc28-4 cells can form mating projections at the restrictive temperature in response to pheromones. It is tempting to suggest that the bud site assembled immediately after the completion of mitosis may have dual potential. It would support the emergence of a bud if cells were to proceed to the division cycle; but it could also be used as a site for construction of a mating projection (shmoo) if cells encountered pheromones. Careful observations using time-lapse microscopy will be needed to test this notion. Finally, we suggest that bud formation is broadly a three- tiered process. It involves (1) inactivation of the mitotic kinase to make way for the bud site assembly (pre-START), (2) determination of the pattern in which the site is positioned (axial or bipolar) and (3) the reorganization of the cytoskeleton leading to bud emergence (START-dependent). It remains to be seen how the cellular activities at these levels are coordinated to produce a normal bud at a specific position at the right time.

We are grateful to Mike Snyder and Yves Barral for SPA2-GFP constructs, and Peter Novick for SEC3-GFP plasmid. We thank Charlie Boone and Marie Evangelista for the ADH-BNI1-GFP construct and Liu Jun for the GFP-CDC42 plasmid. We also thank Shannon Allan for his much-needed help in the mutant hunt, Hong Hwa Lim and Foong May Yeong for their help with some of the experiments and figures, Chong Jin Loy and Phuay-Yee Goh for the critical reading of the manuscript, and Rafael Tham Kim Leong and Richard Chng Eng Hee for their assistance in the photographic work. This work was supported by the Science and Technology Board of Singapore.