Recovery of the Yeast Cell Cycle from Heat Shock-induced G1 Arrest Involves a Positive Regulation of G1Cyclin Expression by the S Phase Cyclin Clb5*

In the yeast Saccharomyces cerevisiae, heat shock stress induces a variety of cellular responses including a transient cell cycle arrest before G1/S transition. Previous studies have suggested that this G1 delay is probably attributable to a reduced level of the G1 cyclin gene (CLN1 and CLN2) transcripts. Here we report our finding that the G1 cyclin Cln3 and the S cyclin Clb5 are the key factors required for recovery from heat shock-induced G1 arrest. Heat shock treatment of G1 cells lacking either CLN3 orCLB5/CLB6 functions leads to prolonged cell cycle arrest before the initiation of DNA synthesis, concomitant with a severe deficiency in bud formation. The inability of the clb5 clb6 mutant to resume normal budding after heat shock treatment is unanticipated, since the S phase cyclins are generally thought to be required mainly for initiation of DNA synthesis and have no significant roles in bud formation in the presence of functional G1cyclins. Further studies reveal that the accumulation of G1cyclin transcripts is markedly delayed in the clb5 clb6mutant following heat shock treatment, indicating that theCLN gene expression may require Clb5/Clb6 to attain a threshold level for driving the cell cycle through G1/S transition. Consistent with this assumption, overproduction of Clb5 greatly enhances the transcription of at least two G1cyclin genes (CLN1 and CLN2) in heat-shocked G1 cells. These results suggest that Clb5 may positively regulate the expression of G1 cyclins during cellular recovery from heat shock-induced G1 arrest. Additional evidence is presented to support a role for Clb5 in maintaining the synchrony between budding and DNA synthesis during normal cell division as well.

Cells from various organisms are equipped through evolution with the ability to react to an abrupt elevation of temperature in their environment with what has been collectively termed heat shock response to increase their chance of survival in such an environment (1)(2)(3)(4). Heat shock treatment of yeast cells, for instance, induces several observable responses including a reprogramming of gene expression, acquisition of thermotolerance, and a transient cell cycle arrest at G 1 (1,(5)(6)(7). The reprogramming of gene expression allows biased synthesis of proteins and metabolites with protective functions against the heat insult. An important group of proteins that are induced to high levels immediately after heat exposure in yeast as well as other organisms is known to be the heat shock proteins, which act as molecular chaperones to help minimize the detrimental effects of protein denaturation caused by heat shock treatment (3, 8 -11). The heat shock-induced reprogramming of gene expression also results in a temporal reduction in the expression of proteins whose functions are not in immediate need for thermotolerance acquisition. This down-regulation of gene expression may be the reason behind the transient cell cycle arrest at G 1 in yeast cells, since two G 1 cyclin genes required for G 1 /S transition, CLN1 and CLN2, are found to be repressed after heat shock treatment (12). A stress-inducible protein, Xbp1, has been proposed to function in the repression of gene expression following heat shock treatment (13). Apart from down-regulating the expression of many genes, heat shock stress is also believed to result in a down-regulation of the cyclic AMP (cAMP)-Ras pathway (7,14), which may affect cell cycle progression through the G 1 cyclin Cln3 (15).
The heat shock-induced G 1 arrest in yeast lasts for a period of approximately 1 h (7). Once the heat shock proteins are induced and thermotolerance is acquired, the normal cell cycle resumes. Normal cell cycle progression in yeast relies on sequential activation of the cyclin-dependent kinase Cdc28 by the cell cycle stage-specific cyclins (16 -19). G 1 /S transition, often referred to as START, is dependent on three G 1 cyclins: Cln1, Cln2, and Cln3 (16 -19). Cln3 is particularly important, since it controls the expression of other G 1 cyclins, as well as Clb5 and Clb6, two S phase cyclins required for initiation of DNA synthesis (20,21). By a yet unknown mechanism, Cln3 activates the transcription factor SBF, which is composed of the Swi4 and Swi6 proteins (22)(23)(24). SBF in turn drives the transcription of a set of genes including CLN1 and CLN2, leading to execution of START and budding (22)(23)(24). In a parallel fashion, Cln3 is also thought to activate another transcription factor MBF, consisting of Mbp1 and Swi6, which then stimulates expression of the genes involved in DNA synthesis including CLB5 and CLB6 (22,23).
Thus, the cyclin cascade initiated by Cln3 sets in motion two parallel cell cycle events downstream of START, i.e. budding and DNA replication, each controlled by a pair of functionally overlapping cyclins. Although G 1 cyclin functions are sufficient for passage of START and bud formation, the initiation of DNA replication cannot properly take place without the functions of Clb5/Clb6. Loss of Clb5/Clb6 functions results in a delay in initiation of DNA synthesis but has no effects on the timing of bud emergence (25), suggesting that Clb5/Clb6 are principally required for DNA synthesis control. Under special conditions, however, Clb5 has been shown to be more versatile, capable of performing some overlapping functions with the G 1 cyclins. Overproduction of Clb5, for example, can rescue the lethality of the cln1 cln2 cln3 triple mutant (26). Similarly, elevation of the Clb5 activity through inactivation of Sic1, an inhibitor of the Clb5-Cdc28 kinase, also suppresses the START deficiency in the cln1 cln2 cln3 triple mutant (27,28). These observations suggest that, although Clb5 normally is not required for execution of START and budding, it is able to provide these functions if its activity is sufficiently increased.
In the present study, the roles of various cyclins in the recovery of cell cycle from heat shock-induced G 1 arrest have been examined. Our results suggest that Clb5 is one of the key factors required for this process. Clb5 facilitates the cell cycle recovery following heat shock by promoting initiation of DNA replication on one hand and positively regulating the expression of the G 1 cyclin genes to promote budding on the other.

EXPERIMENTAL PROCEDURES
Strains, Media, and General Methods-All yeast strains used in this study were derived from the wild type strain W303 and are listed in Table I. The rad24 and cln1 cln3 mutants were originally obtained from U. Surana and I. Herskowitz, respectively. YUS454 (cln1 cln2 cln3) contains the CLN3 gene under control of the methionine-repressible MET3 promoter for viability. Yeast extract-peptone, synthetic complete, and dropout media were prepared as described by Rose et al. (29). Recombinant DNA methodology was performed as described by Sambrook et al. (30). PCR was performed with Vent polymerase (New England Biolabs) as recommended by the manufacturer. Genetic manipulations were performed according to standard methods (29). Cell morphology and DNA content analysis by FACScan were performed as described previously (31).
Plasmid Construction-Plasmids used in this study are listed in Table II. To generate the plasmids pMC229 -233, polymerase chain reaction-amplified coding regions of CLN2, CLN3, CLB3, CLB5, and CLB6 were cloned individually into a centromere vector under control of the GAL1 promoter containing URA3 as a selectable marker. The plasmids pMC234 and pMC235 were generated by fusing the coding regions of CLN2 and CLN3 with GAL1 in a centromere vector containing the LEU2 gene.
Synchronization and Heat Shock Treatment-Overnight cultures were diluted to an A 600 of 0.1 and were allowed to grow at 25°C to an A 600 of 0.3. At this point, ␣-factor was added to a final concentration of 5 g/ml. When greater than 95% cells had been arrested in G 1 , the cultures were divided into two halves, with one shifted to 42°C for 30 min and another remaining at 25°C. The cells were then washed by filtration and resuspended in fresh medium at 25°C. Samples taken at intervals were used for analysis of cell morphology, DNA content, and Northern blotting.
Gene Overexpression-In experiments where the expression of a gene was driven by the GAL1 promoter, the cells were first grown in medium containing raffinose as the sole carbon source. The cells were synchronized in G 1 as described above. Galactose was then added to 2% to induce the gene expression for 15 min at 25°C before being subjected to heat shock treatment. After ␣-factor was removed by filtration, the cells were resuspended in fresh medium containing raffinose and galactose and incubated at 25°C. Samples were collected at intervals for further analysis.
In experiments showing that CLB5 overexpression overrides ␣-factor-induced G 1 arrest, the G 1 -arrested cells were split into two parts with galactose added to one of them, and both cultures were continuously incubated with ␣-factor. 1 The abbreviation used is: kb, kilobase pair(s).   1. Heat shock induces a transient G 1 arrest in wild type and DNA checkpoint mutants. Cells were synchronized in G 1 by ␣-factor and divided into two halves with one shifted to 42°C for 30 min and another remaining at 25°C. After washing, the cells were resuspended in fresh medium and incubated at 25°C. Samples taken at intervals were analyzed for DNA content by FACScan and budding profiles. The heat-treated cultures are shown on the right and the untreated ones on the left. The strains were W303 (wild type), YMC309a (rad9), and YMC431 (rad24). The percentage of unbudded cells is presented with S.D. values. At least 200 cells were counted for each sample.
Northern Analysis-Total RNA was isolated as described by Cross and Tinkelenberg (33), and Northern blot analyses were performed as described by Price et al. (34). To show RNA signals of various genes on the same blot, the DNA probes used earlier on RNA blot were stripped by heating at 80°C for 30 min in a solution containing 0.1% SDS, 1 M EDTA, and 10 M Tris-HCl (pH 7.2). After washing briefly in H 2 O, the blot was hybridized with another 32 P-labeled probe. The RNA levels were quantitated with a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).

Heat Shock Stress Causes a Transient Cell Cycle
Arrest at G 1 -Previous studies on yeast cell cycle response to heat shock treatment were performed with asynchronous cell populations (12). To better assess the duration of the G 1 arrest period and the kinetics of the recovery, we used mating pheromone (␣factor)-synchronized cells for examination of their response to heat shock treatment. Wild type cells were synchronized by ␣-factor and divided into two halves with one shifted to 42°C for 30 min, while another remained at 25°C, followed by washing and resuspension in fresh medium devoid of the mating pheromone. As shown in Fig. 1, cells that had been exposed to high temperature initiated DNA replication and budding at a later time point than those that were not treated with heat shock. The delay was about 60 min ( Fig. 1). This result confirms the previous finding that the cell cycle arrest induced by heat shock is prior to START, since budding and DNA replication were both delayed (12).
It is well known that another type of stress, DNA damage, also induces cell cycle arrest in yeast, which is mediated through a control mechanism termed "checkpoint" (35,36). The checkpoint genes such as RAD9 and RAD24 have been shown to be required for the transient cell cycle arrest at G 1 caused by DNA damage (37). To test whether the DNA damage checkpoint also functions in the G 1 arrest caused by heat shock treatment, we examined the cell cycle response to heat shock stress in the rad9 and rad24 mutants. As shown in Fig. 1, both mutants exhibited a G 1 arrest that lasted about 60 min, similar to the wild type cells, suggesting that the cell cycle arrest after heat shock is not mediated through DNA damage checkpoint genes.
Cln3 and Clb5/Clb6 Are Essential for Recovery from Heat Shock-induced G 1 Arrest-Decrease in the CLN1 and CLN2 transcript levels has been suggested as a possible cause of the cell cycle arrest in heat shock-treated cells (12). We therefore examined various cyclin mutants for their response to heat shock treatment. Cells lacking CLN2 responded to heat shock treatment with a G 1 arrest and recovered 60 min later, as did wild type cells (Fig. 2). The cln1 mutant behaved similarly (data not shown). Deletion of both CLN1 and CLN2 genes caused a further delay of 30 min (Fig. 2). Nevertheless, the double mutant resumed cell cycle and divided normally after the 90-min delay ( Fig. 2 and data not shown).
Remarkably, the cln3 mutant failed to recover from the heat shock-induced G 1 arrest for at least 5 h (Fig. 2). The mutant cells showed no sign of DNA replication and virtually no bud formation during the entire period of the experiment (Fig. 2). The failure of the cln3 mutant to initiate DNA replication and budding was not because the mutant was particularly sensitive to the heat treatment, since the cln3 cell viability was not significantly affected by the heat shock treatment (data not shown). This result indicates that CLN3 is essential for cell recovery from heat shock-induced G 1 arrest. A similar defect in recovery from the arrest was also observed in the clb5 clb6 double mutant except, in this case, the budding defect was less pronounced than that of the cln3 cells (Fig. 2). Nevertheless, the clb5 clb6 double mutant displayed a conspicuous budding defect after heat shock. 2 h after release from synchronization, for example, only 26% of the double mutant cells were budded, compared with over 50% in the populations of either the wild type or any of the cln1, cln2, clb5, and clb6 single mutants (Fig.  2, and data not shown). Furthermore, the budded cell population of the double mutant was limited to about 50% for at least 5 h (Fig. 2). It is evident, therefore, that the clb5 clb6 mutant was unable to recover from the heat shock-induced G 1 arrest for a rather prolonged period. Both cln3 and clb5 clb6 mutants eventually resumed normal division without showing significant reductions in their viability (data not shown). In comparison, the clb3 clb4 double mutant, defective in another pair of S phase cyclins, Clb3/Clb4, (25,38,39), exhibited no defect in recovery from the G 1 arrest (Fig. 2).
Transcription of the CLN Genes Is Impaired in the clb5 clb6 Mutant after Heat Shock-The severe defect in recovery from heat shock-induced G 1 arrest exhibited by the cln3 mutant is, by and large, within expectations, since this cyclin is the crucial factor in promoting the initiation of bud formation and DNA replication. However, the inability of the clb5 clb6 double mutant to resume normal budding was unexpected, because the Clb5/Clb6 cyclins are thought to be required mainly for DNA synthesis and not for bud formation when functional G 1 cyclins are present (25). To seek an explanation for this finding, we tested whether the CLN transcript levels were affected in the double mutant. As shown in Fig. 3a, the CLN1 and CLN2 transcripts in the untreated wild type cells accumulated to a peak level 30 min after release from synchronization, followed by periodic oscillations. In the heat shock-treated wild type cells, the CLN1 and CLN2 transcripts lagged 30 -60 min before reaching the peak levels (Fig. 3a). Similarly, the accumulation of the CLB5 transcript was also delayed, albeit to a lesser extent compared with that of CLN1 and CLN2 (Fig. 3a). The CLN3 transcript was present in ␣-factor-treated cells and appeared to be the least affected by heat shock treatment (Fig.  3a). In contrast, the clb5 clb6 mutant displayed a marked reduction in all G 1 cyclin transcripts after heat shock treatment (Fig. 3b). The CLN1 and CLN2 transcripts in clb5 clb6 cells did not reach levels comparable with those in the untreated cells until 180 min after release from G 1 synchronization (Fig. 3b). The level of the CLN3 transcript was also decreased significantly in the mutant after heat exposure (Fig.  3b). The clb5 clb6 mutant, nevertheless, produced the ACT1 FIG. 3. Accumulation of CLN and CLB5 transcripts was delayed after heat shock treatment. Cells were processed by the procedure described in the legend to Fig. 1. Samples were taken at intervals for measuring mRNA abundance. The transcript levels are shown in the upper panels. The same blot was probed and reprobed (after stripping) by labeled DNA fragments of the indicated genes. ACT1 is the gene for yeast actin and serves as a loading control. SSA4 is an HSP70 heat shock-inducible gene and serves as a heat shock control (52). Quantitative presentations of the RNA signals derived from PhosphorImager analysis are shown in the lower panels. a, the wild type strain (W303). b, the clb5 clb6 double mutant (YMC406). HS, heat shock. and SSA4 transcripts in a manner similar to wild type cells after heat exposure, indicating that the mutant was not defective in the general and heat shock-induced transcriptions (Fig.  3). These results together suggest that the Clb5/Clb6 cyclins are involved in the recovery of CLN transcript abundance in cells that have undergone heat shock treatment.
Overexpression of CLN and CLB5 Eliminates the Heat Shock-induced G 1 Arrest-The observation that the CLN transcript accumulation was delayed by heat shock treatment supports the notion that the cell cycle arrest following heat shock is attributable to the suboptimal level of G 1 cyclin expression. Indeed, overexpression of CLN2 is able to abolish the G 1 arrest induced by heat shock treatment in asynchronous cells (12). To better understand the mechanisms of the heat shock-induced G 1 arrest, we examined the effects of overexpression of other cyclin genes on cell cycle response to heat shock. The CLN2, CLN3, CLB5, and CLB3 genes were each placed under control of pGAL1, a strong inducible promoter, and introduced into wild type cells. After synchronization in G 1 by ␣-factor in a raffinose-containing medium, galactose was added for 15 min to induce the cyclin gene expression, followed by heat shock and release to fresh medium containing galactose. As shown in Fig. 4, overexpression of each CLN2, CLN3, and CLB5 completely eliminated the G 1 arrest. The cyclin-overproducing cells displayed cell cycle kinetics after heat shock treatment comparable with those of untreated wild type cells, as judged by the FACScan pattern and budding profile (Fig. 4). In contrast, overexpression of two other S phase cyclin genes, CLB3 and CLB4, generated no effect on the G 1 arrest after heat shock (Fig. 4, and data not shown), an observation in agreement with the results presented in Fig. 2.
Overexpression of CLB5 Enhances CLN Transcriptions-It has been suggested that the G 1 cyclins are the rate-limiting factors in G 1 cells for passage of START (16). The finding that heat shock stress failed to result in cell cycle arrest in CLN overexpression cells suggests the same for the heat shock- FIG. 4. Overexpression of CLN and  CLB5 eliminates heat-induced G 1 arrest. Wild type cells containing each of the GAL1-CLN2, GAL1-CLN3, GAL1-CLB5, or GAL1-CLB3 constructs were grown in raffinose and synchronized by ␣-factor. The cells were proportioned equally into three parts with galactose added to one of them. The three cultures were incubated for another 15 min before subjecting two of them including the one in galactose to heat shock treatment. After washing, the cells were resuspended in fresh medium containing raffinose (for the ϪGAL cultures) or raffinose plus galactose (for the ϩGAL culture) and incubated at 25°C. Samples taken at intervals were analyzed for DNA content and budding profiles. The percentage of unbudded cells is presented with the S.D. values. HS, heat shock. GAL, galactose.
induced G 1 arrest. Since the Clb5/Clb6 cyclins are not ratelimiting factors for START (25), the ability of overexpressed CLB5 to negate the heat shock-induced G 1 arrest is indicative of an indirect role for Clb5/Clb6 in START promotion, possibly as stimulators of the CLN gene transcription. This possibility was investigated by analyzing the CLN transcript abundance in cells overexpressing CLB5. As shown in Fig. 5a, CLB5 overexpression reversed the delay of CLN1 and CLN2 transcripts caused by heat shock, making them appear at the same time as in the wild type cells that experienced no heat shock stress (Fig.  5a). CLN3 transcript levels were not significantly affected by CLB5 overexpression (Fig. 5a). Overexpression of another Btype cyclin gene (CLB3) had no effect on the CLN1 transcription at all (Fig. 5b), consistent with the results described above, namely that neither the mutations nor the overexpression of the CLB3/CLB4 genes conferred any impact on heat shockinduced G 1 arrest. It is noteworthy that both CLN1 and CLN2 transcripts retained a periodic expression pattern throughout the cell cycle in the CLB5 overexpression cells.
CLB5 Overexpression Stimulates CLN1/CLN2 Transcription in the cln3 and swi4 Mutants-The finding that CLB5 overexpression stimulated CLN1/CLN2 gene transcription led us to test whether an increase in Clb5 activity will bring about G 1 /S transition in cln3 cells that had endured heat shock stress. In the absence of heat shock treatment, CLB5 overexpression resulted in slight advancement in cell cycle progression in cln3 cells (Fig. 6a). Heat shock treatment of cln3 cells led to persistent cell cycle arrest at G 1 , which could be effectively negated by CLB5 overexpression (Fig. 6a), suggesting that the defect of heat-treated cln3 mutant in carrying out G 1 /S transition is likely to be derived from insufficient CLN and CLB5 transcriptions. This was ascertained by examining the levels of CLN1, CLN2, and CLB5 transcripts in heat-treated cln3 cells. As shown in Fig. 6b, heat shock treatment strongly prevented the accumulation of CLN1/CLN2/CLB5 transcripts, and no transcripts were detected for these three cyclins until 180 min after release from synchronization. Induction of CLB5 overexpression allowed the appearance of the CLN1/CLN2 transcripts to be advanced by 90 min in the heat-treated cln3 cells (Fig. 6b). This suggests that CLB5 overexpression can stimulate CLN1/CLN2 transcription in the absence of the CLN3 function.
Since CLN1/CLN2 expression is mainly dependent on the transcription factor SBF (22, 24), we therefore examined the CLN1/CLN2 levels in the swi4 mutant, which is defective in the DNA binding component of SBF (24). The swi4 cells exhibited no defect in recovery from heat shock treatment, since they responded to heat shock with a G 1 arrest that lasted for about 60 min (Fig. 6c). CLB5 overexpression in swi4 cells completely eliminated this G 1 arrest (Fig. 6c). The appearance of CLN1/ CLN2 transcripts in swi4 cells was delayed by heat shock treatment until 90 min after release from synchronization, which was, again, greatly advanced by the overexpression of CLB5 (Fig. 6d). These results indicate that the stimulation of CLN1/CLN2 transcription by Clb5 was achieved independently of SBF functions.
Overexpression of Clb5 Stimulates CLN1/CLN2 Transcription in the Presence of Mating Factor-The CLN1/CLN2 gene transcription is known to be inhibited by ␣-factor (40). The data presented in Fig. 6 already showed that CLB5 overexpression could override the ␣-factor-dependent inhibition of CLN1/ CLN2 transcription. To further demonstrate this finding, the CLN1/CLN2 transcription was analyzed using samples derived from the cultures in which ␣-factor was constantly present. In agreement with the previous report (40), the transcription of CLN1 and CLN2, but not CLN3, was strongly repressed in the presence of ␣-factor in wild type cells (Fig. 7a). No significant accumulation of the CLN1/CLN2 transcripts was observed until the end of the time course (240 min, Fig. 7a). In a sharp contrast, the CLN1/CLN2 transcription was markedly induced in the CLB5 overexpression cells shortly after the addition of galactose (Fig. 7a). This result demonstrates that CLB5 overexpression can strongly stimulate the CLN1/CLN2 transcription despite the presence of ␣-factor.
CLB5 overexpression in wild type cells not only stimulated the CLN1/CLN2 expression, but also promoted budding and DNA replication, in the presence of ␣-factor (Fig. 7b). Similar results were also observed in ␣-factor-treated cells of the cln1 cln2, cln1 cln3, and swi4 mutants (Fig. 7b).
CLB5 May Coordinate between Budding and DNA Replica- tion-It is important to note that the CLB5-promoted cell cycle progression in the presence of ␣-factor retained a good synchrony between budding and DNA replication (Fig. 7b). This raised an interesting possibility that the Clb5 regulation of CLN gene expression may not be limited to cell cycle response to heat shock stress and could conceivably be important in coordination of budding and DNA synthesis. To examine this hypothesis, we followed the timing of budding relative to DNA synthesis in the Clb5-overproducing cells in the absence of functional Cln cyclins. If Clb5 is important for coordination between DNA synthesis and budding by positively regulating G 1 cyclin gene expression, an increase in Clb5 levels will promote the two processes simultaneously only in the presence of at least one CLN gene, as is the case in the experiment presented in Fig. 7b. Wild type cells, as shown in Fig. 8, exhibited good coordination of budding and DNA synthesis during the cell cycle after release from ␣-factor synchronization. At the time, for example, when nearly half of the population had replicated DNA, as judged by the size of the 2N peak in FAC-Scan (93 min after release), 52% of the cells remained unbudded. Wild type cells overexpressing CLB5 initiated DNA synthesis earlier, generating half of the population in S phase at 79 min after release. Importantly, the budding process was similarly advanced in these cells, showing 53% of the unbudded population at this time point. This suggests, therefore, that, although Clb5 is thought to be much more efficient in promoting DNA synthesis than in promoting bud formation (16,17,25,41,42), its overproduction in the presence of one or more functional G 1 cyclin genes maintains the coordination between budding and DNA synthesis. In contrast, in cells lacking all three G 1 cyclin genes (the cln1 cln2 cln3 triple mutant), an uncoupling of the two processes was observed. While CLB5 overexpression allowed the triple mutant to bud as well as to replicate DNA after release from ␣-factor synchronization, the budding process was significantly delayed in relation to DNA synthesis. At the time when half of the population had replicated DNA (79 min), 94% of the population still remained unbudded (Fig. 8). These results confirm the anticipation that an increase in Clb5 activity will advance both processes of DNA synthesis and budding simultaneously only in the presence of functional G 1 cyclin genes.

Heat Shock-induced G 1 Arrest Is Unlikely to Be Caused by
Checkpoint-like Mechanisms-Like many other organisms, yeast cells react to heat shock stress with a reprogramming of cellular activities to increase their ability to survive in the stressful environment. One of the consequences of this reaction is the transient cell cycle arrest at G 1 . Since another type of stress signal, DNA damage, causes a similar G 1 delay, which is  (37), it is reasonable to ask whether the heat shock-induced G 1 delay is also a result of active inhibition of the cell cycle by a mechanism similar to the DNA damage checkpoint. It has been reported that a putative transcriptional repressor, Xbp1, is induced by various stress signals including heat shock (13). Overexpression of XBP1 leads to down-regulation of transcription of an extensive array of genes including CLN1/CLN2/CLN3 (13). However, the role of Xbp1 in the repression of CLN gene expression and the cell cycle delay in response to heat shock treatment has remained unclear, as the xbp1 null mutation does not affect the CLN gene repression after stress treatment (13). Our present study shows that the heat shock-induced G 1 delay does not involve the functions of some of the DNA damage checkpoint genes. In addition, we also found that elimination of the G 1 delay in heat-treated cells by artificially forcing the passage of START has no discernible effect on cell viability (data not shown). Together with the earlier observation that the heat stress-induced cell cycle arrest is not required for acquisition of thermotolerance (5), these results suggest that the G 1 arrest after heat shock is unlikely to be a manifestation of a checkpoint-like control mechanism. Instead, it may be a result of insufficient START-promoting G 1 cyclins caused by the selective gene expression employed by the cell to gain thermotolerance.
The Roles of G 1 Cyclins in Recovery from Heat Shock-induced G 1 Arrest-The G 1 cyclins are probably the only proteins limiting for START execution, since cells overproducing G 1 cyclins can pass through START immediately after cytokinesis (16). Rowley et al. (12) have attributed the heat shock-induced G 1 arrest to low levels of CLN1 and CLN2, but not CLN3, transcripts. We have confirmed their observation using G 1 -synchronized cells. We show that, in wild type cells, the accumulation of the CLN1/CLN2 transcripts is strongly suppressed by heat shock for at least 30 min, while that of CLN3 is less affected. However, a significant reduction in the CLN3 transcript level is observed in the clb5 clb6 mutant, suggesting that heat shock stress exerts an adverse impact on the expression of all three G 1 cyclin genes, particularly in the absence of CLB5/CLB6 functions. FIG. 7. Overexpression of CLB5 promotes CLN1/2 transcription and cell cycle progression in the presence of ␣-factor. Wild type as well as various mutant cells carrying GAL1-CLB5 were grown in raffinose and arrested in G 1 by ␣-factor at 25°C for 2 h. The cultures were then divided into two halves with galactose added to one of them and were further incubated at 25°C with ␣-factor. Samples taken at intervals were analyzed for DNA content, budding profiles, and mRNA abundance. a, the CLN transcript levels in ␣-factor-treated wild type cells with or without CLB5 overexpression. b, FACScan analyses and budding profiles of wild type (W303), cln1 cln2 (YMC434), cln1 cln3 (YMC436), and swi4 (YMC435) cells with or without CLB5 overexpression. GAL, galactose.
Of the three G 1 cyclins, Cln3 is the most important in that it determines the expression of the other two, as well as the expression of the Clb5 and Clb6 cyclins (16,17,19). Cln3 as such, however, is not believed to be directly involved in DNA replication and budding (25). Our studies indicate that Cln3 is required for cellular recovery from heat shock-induced G 1 arrest. Neither the cln1 nor cln2 mutants exhibited a discernible defect in the recovery, and the cln1 cln2 double mutant only showed a 30-min delay in the recovery compared with the wild type (and either of the single mutants). In contrast, the cln3 mutant could not recover from the arrest during the entire 5-h observation. Therefore, although CLN1/CLN2 transcripts are most apparently inhibited from accumulation by heat shock stress among all the cyclin genes investigated, they are nonessential for the process of cell cycle recovery. This is probably due to the presence of Clb5 function, which is capable of carrying out START execution in place of G 1 cyclins. In the absence of Cln3, the expression of all three cyclin genes (CLN1/ CLN2 and CLB5) is greatly reduced after heat shock. This explains why cln3 mutant is unable to recover from the heat shock-induced G 1 arrest for a prolonged time.
The Role of Clb5/Clb6 in Cell Cycle Recovery from Heat Shock-induced G 1 Arrest-The Clb5/Clb6 cyclins are thought to be required mainly for normal S phase promotion. In the clb5 clb6 double mutant, initiation of S phase is greatly delayed (25). The double mutant, however, buds properly under normal growth conditions (25). This has led to the conclusion that the Clb5/Clb6 cyclins exercise no functions in bud formation (25). This view may now need modification, since we have found that the clb5 clb6 double mutant displayed a severe defect in resumption of normal budding after heat shock treatment for a prolonged period of time. In fact, the clb5 clb6 mutant is even less competent in bud formation than the cln1 cln2 mutant after heat shock treatment (Fig. 2).
Clb5, and possibly Clb6, may exert a double effect on the cell cycle recovery from heat shock-induced G 1 arrest. First, Clb5/ Clb6 function as promoters of DNA replication. In the absence of Clb5/Clb6, the DNA replication is impaired but not blocked, due to the activities of G 1 cyclins, which can take over parts of Clb5/Clb6 functions. This has been the explanation for the lethality and permanent arrest before S phase of the cln1 cln2 clb5 clb6 quadruple mutant (25). Heat shock stress leads to a decline in CLN gene transcript and hence a frailty of G 1 cyclins to compensate for the Clb5/Clb6 function in initiation of DNA replication. Second, Clb5/Clb6 indirectly participate in G 1 /S transition and bud formation through the positive regulation of the G 1 cyclin gene expression. The observation that clb5 clb6 cells are more deficient than cln1 cln2 cells in bud formation following heat shock suggests, once again, that all three CLN genes require Clb5/Clb6 for optimal expression in heat-shocked cells. Therefore, the substantial delay of CLN1/CLN2/CLN3 transcript accumulation in clb5 clb6 cells after heat shock treatment explains the budding defect of the double mutant.
Clb5 is also thought to perform some overlapping functions later in the cell cycle with another two S phase cyclins, Clb3/ Clb4, since it is required for spindle formation in the absence of CLB3/CLB4 genes (25). We have shown that the Clb3/Clb4 cyclins are not involved in cell cycle recovery from heat shockinduced G 1 arrest, and overexpression of CLB3 genes has no effect on CLN gene transcription.
Positive Regulation of CLN Gene Expression by Clb5-It is clear that the transcription of at least two G 1 cyclin genes, CLN1 and CLN2, can be strongly stimulated by Clb5 overproduction. In heat shock-treated cells, for instance, galactose induction of CLB5 completely negated the delay of CLN1/CLN2 transcript accumulation. The fact that the transcription of CLN3 is not as dramatically enhanced as CLN1/CLN2 is probably due to the fact that it is already near an optimal level. Overexpression of CLB3 in the same way as CLB5 made no difference to CLN transcription, indicating that CLB3 plays no roles in cell cycle recovery from the G 1 arrest. CLB5 overexpression also remarkably stimulates the CLN1/CLN2 transcription in the heat shock-treated cln3 and swi4 mutant cells. In the cln3 mutant, the appearance of CLN1/CLN2 transcripts is advanced by 90 min by Clb5 after heat shock treatment. There is at least 60-min advancement of CLN transcription in the heat shock-treated swi4 mutant as well. In addition, the transcription of the CLN1 and CLN2 genes, which is normally repressed by the mating pheromone ␣-factor, can be greatly stimulated by CLB5 overexpression in the presence of ␣-factor. Therefore, Clb5 is able to stimulate CLN1/CLN2 gene transcription when its activity is sufficiently elevated.
In fact, the effect of artificially elevated Clb5 activity on CLN gene expression has been investigated previously using small daughter cells isolated by centrifugal elutriation (25). Although the accumulation of the CLN1 transcript was noticeably advanced by Clb5 overproduction, it was not sufficient to prompt the authors to draw the conclusion that Clb5 is able to stimulate the CLN gene expression (25). The discrepancy in the degree of CLN activation by Clb5 in their observation and ours can be attributed to the fact that these authors used small early G 1 cells, whereas we employed late G 1 cells. It is likely that the accumulation of certain factors, such as those for gene transcription and mRNA translation, only reach a functional level FIG. 8. Coordination of budding and DNA replication by CLB5. Wild type cells containing GAL-CLB5 were synchronized in G 1 by ␣-factor in raffinose. The culture was divided into two parts, galactose was added to one part, and the culture incubated for another 15 min. After washing, the cells were released into fresh medium containing raffinose (left) or raffinose plus galactose (middle). The cln1 cln2 cln3 mutant (YUS454) containing GAL-CLB5 (right) was grown in raffinose in the absence of methionine and synchronized in G 1 by ␣-factor in medium containing methionine. The cells were then released into fresh raffinose medium containing galactose and methionine. Aliquots of cells were taken at intervals for analysis of DNA content and budding profiles.
for Clb5-stimulated CLN gene expression around late G 1 . Thus, overproduction of Clb5 alone in the early G 1 cells is not expected to accomplish the task of activating CLN gene expression. Indeed, using late G 1 cells, these authors observed strong induction of PCL1 and RNR1, two genes regulated by SBF and MBF, respectively, by Clb5 overproduction in the absence of three G 1 cyclin genes (25).
In view of the previous findings that Clb5 overproduction can perform START functions and promote transcription of the SBF-and MBF-dependent genes in the cln1 cln2 cln3 triple mutant (25,26), our new observation of CLN gene activation by Clb5 may be, after all, not completely unexpected. However, the important question remaining to be answered is whether Clb5 actually performs these functions under normal physiological conditions, as it did under those artificial ones. We now provide evidence to suggest that Clb5 is indeed required for optimal expression of CLN genes in vivo. First, the clb5 clb6 mutant is defective in resumption of normal budding after heat shock treatment despite the presence of all three CLN genes. Wild type G 1 cells recover from heat shock-induced arrest after about 1 h. At this point, both budding and DNA replication have progressed approximately halfway (Fig. 1). Under the same conditions, however, the clb5 clb6 mutant is completely defective in initiation of DNA replication, and severely defective in bud formation, during the entire time course of 5 h (Fig.  2). Therefore, a physiological situation has been identified under which Clb5 and Clb6 are essential for proper bud formation. Second, transcription of all three CLN genes becomes impaired in the clb5 clb6 mutant after heat shock. Heat shock treatment of wild type cells delays appearance of the peak level CLN1/CLN2 transcripts for 30 -60 min. However, attainment of peak levels of the CLN1/CLN2 transcripts is delayed for at least 150 min by heat shock stress in the double mutant. The accumulation of the CLN3 transcript is also significantly delayed in the double mutant. This indicates that the accumulation of the CLN transcripts to peak levels, crucial for promoting G 1 /S transition, requires the activity of the Clb5/Clb6 cyclins in heat shock-treated cells. These results obtained under physiological conditions clearly advocate a role for Clb5/Clb6 in positive regulation of CLN gene expression in vivo.
Clb5 is normally under the negative control by Sic1, which binds to, and inhibits the activity of, the Clb5-Cdc28 kinase (42). Phosphorylation of Sic1 by Cln-associated Cdc28 kinase and its subsequent ubiquitination by the Cdc4-Skp1-Cdc53 complex triggers the proteolytic inactivation of Sic1 (43)(44)(45)(46). The ability of Clb5 to promote expression of Cln1/Cln2 suggests the existence of a regulatory circuit that can rapidly disable the inhibitory effect of Sic1 when Clb5 levels are elevated.
How Clb5 regulates the expression of the CLN genes in vivo is still open to speculation. One possibility is that Clb5 may mimic the function of Cln3, since it can strongly activate CLN1/ CLN2, as well as other genes regulated by SBF and MBF (25), in the absence of Cln3. This possibility is made unlikely by the fact that the CLN3 gene itself, whose transcription is independent of SBF and MBF (24), may require Clb5 for activation in heat shock-treated cells (Fig. 3). In addition, Clb5 can also stimulate CLN gene expression in the absence of Swi4, a component of SBF. It is therefore possible that Clb5 may promote CLN expression independently of Cln3 and SBF. This proposal is also consistent with the previous finding that persistent overexpression of SWI4 did not prevent the heat shock-induced decrease in CLN1/CLN2 transcript abundance (12). Cln3 and SBF are not the only activators of CLN1/CLN2 transcription, since CLN/ CLN2 transcription still occurs, albeit less efficiently, in either the cln3 or the swi4 null mutants. A number of other factors, such as Bck2, Rem1, Bry1, and Sit4, have been implicated in the regulation of transcription of the CLN1/CLN2 genes independent of the Cln3/SBF pathway (47)(48)(49)(50). Interestingly, overproduction of Bck2, like that of Clb5, can strongly induce CLN1/CLN2 transcription (47). Whether the CLN1/CLN2 gene regulation by Clb5 involves these factors has not been investigated.
The Potential Role for Clb5 in Coordination between DNA Replication and Budding-The positive regulation of CLN gene transcription by Clb5 has been shown to be important in cell recovery from heat shock-induced G 1 arrest. However, it is unlikely to be the only function of this regulation. The hypothesis that Clb5 may also play an important role in maintaining the synchrony between budding and DNA replication has been investigated by analyzing the effect of CLB5 overexpression on the timing of bud emergence relative to that of DNA replication in the presence or absence of functional Cln cyclins. The ability of CLB5 to promote both DNA replication and bud formation in a coordinated manner is evident in the cells carrying at least one G 1 cyclin gene. First, CLB5 overexpression is found to be able to promote G 1 /S transition in ␣-factor-arrested cells. This phenomenon is probably independent of the Clb5-activated CLN1/CLN2 transcription, because it occurs even in cells lacking all G 1 cyclin genes (51). Nonetheless, the important point is that DNA replication and budding are promoted simultaneously by Clb5 in cells carrying one or more G 1 cyclin genes (Fig. 7b). Examination of synchronized cell cycle progression further supports this notion. In the presence of CLN genes, initiation of DNA replication and bud formation are equally advanced by CLB5 overexpression (Fig. 8). However, in the absence of CLN genes, the synchrony between DNA replication and bud formation is lost (Fig. 8). Therefore, although Clb5 is intrinsically a poor promoter for bud emergence (16,17,25,41,42), its overproduction maintains the coordination between the two processes as long as at least one CLN gene is present. The dependence of Clb5-driven synchronous progression of cell cycle on the presence of CLN suggests that Clb5 promotes bud formation indirectly through increasing the activity of Cln cyclins, or the expression of the CLN genes.