Swe1 regulation and transcriptional control restrict the activity of mitotic cyclins toward replication proteins in Saccharomyces cerevisiae

Results

Swe1 Inhibits Replication Initiation by Clb2-Cdk1 During S and G₂. To determine whether Swe1 regulates the potential activity of mitotic cyclins in the early cell cycle, we deleted SWE1 in clb5Δ clb6Δ cells expressing CLB2 from the CLB5 promoter (clb5::CLB2 clb6Δ) and monitored replication by assaying DNA content of cells released synchronously from G₁ arrest (Fig. 1A). In clb5Δ clb6Δ cells, replication began between 60 and 72 min, much later than in clb6Δ cells, in which Clb5 initiates replication between 12 and 24 min. Early expression of CLB2 in clb5Δ clb6Δ cells (clb5::CLB2 clb6Δ) advanced S-phase entry by ≈12 min. Deletion of SWE1 significantly advanced replication initiation in cells expressing CLB2 early (Fig. 1A). S-phase began between 24 and 36 min in clb5::CLB2 clb6Δ swe1Δ cells, ≈36 min sooner than in the clb5Δ clb6Δ cells, and only ≈12 min behind clb6Δ cells. The duration of S-phase was comparable in all strains. There was no effect of SWE1 deletion in clb6Δ cells, indicating that Swe1 does not inhibit the function of Clb5 in replication initiation. Because early S-phase entry depends on early expressed Clb2 in addition to SWE1 deletion, the data suggest that S-phase entry fully depends on Clb2 and that Clb3 and Clb4 did not contribute. To strengthen this conclusion, we analyzed the timing of S-phase entry in cells lacking CLB3 and CLB4 (construction of this strain is described below). The time of S-phase entry of clb5::CLB2 clb3Δ clb4Δ clb6Δ swe1Δ cells was indistinguishable from that of clb5::CLB2 clb6Δ swe1Δ cells, indicating that Clb2 can efficiently stimulate the origin initiation independently of Clb3-Clb6 (Fig. 1A).

DNA content analysis of unsynchronized cultures was consistent with these results. Compared with clb6Δ cells, clb5Δ clb6Δ cells show an increased number of cells with unreplicated DNA content because of delayed S-phase entry (Fig. 1A). Whereas early expression of CLB2 or deletion of SWE1 did not significantly reduce the proportion of cells in G₁, the combination of both reduced the population of these cells significantly, consistent with the accelerated onset of S-phase (Fig. 1A). Thus, Clb2 can effectively activate replication origins but is restricted from doing so during the normal S-phase period by its normal transcription timing and the inhibitory action of Swe1.

Swe1 likely acts directly to inhibit early expressed Clb2-Cdk1 function through inhibitory phosphorylation of Cdk1-Y19. To examine directly the importance of Cdk1-Y19 phosphorylation, we expressed a nonphosphorylatable form of Cdk1 (Cdc28-y19F) (16, 17), which should phenocopy SWE1 mutation in cells expressing clb5::CLB2. In fact, early expression of Clb2 and Cdc28-y19F advanced S-phase entry of clb5Δ clb6Δ cells by at least 24 min (Fig. 1B). The slightly smaller effect of Cdc28-y19F than deletion of SWE1 is consistent with reports that part of the Cdk1 inhibitory effect of Swe1 is independent of Y19 phosphorylation (24).

Although early expression of Clb2 in clb5Δ clb6Δ swe1Δ cells drove much earlier S-phase entry, it still was delayed ≈12 min compared with clb6Δ and clb6Δ swe1Δ cells, in which Clb5 stimulates S-phase entry (Fig. 1A). This result does not appear to reflect a difference in the time of cell cycle entry (start) because the time of bud emergence in clb5::CLB2 clb6Δ swe1Δ cells was indistinguishable from clb6Δ swe1Δ cells (data not shown). It did not appear to reflect instability of Clb2 in the early cell cycle, perhaps due to Cdh1-dependent degradation, because early expression of stabilized Clb2 lacking its destruction box drove replication initiation in clb5Δ clb6Δ swe1Δ cells with similar kinetics as Clb2 (data not shown). Increasing the dosage of early expressed Clb2 (by using a low-copy, plasmid-borne clb5::CLB2 in addition to the integrated copy) did not further accelerate S-phase entry either (data not shown). The slight difference in the ability of Clb2 to stimulate replication versus Clb5 may reflect inherent differences in their substrate specificities.

Clb3 and Clb4 Can Stimulate Replication Initiation in the Presence of Swe1. We also tested whether Swe1 blocks the function of early expressed CLB3 and CLB4. In contradiction to a previous report that early expression of CLB4 in place of CLB5 did not effectively activate origins (12), we observed that early expression of Clb4 advanced S-phase by ≈24 min (Fig. 1C). Deletion of SWE1 further advanced S-phase in clb5::CLB4 clb6Δ cells by ≈12 min. Early expression of Clb3 in clb5Δ clb6Δ cells also advanced replication initiation, in this case by ≈36 min, which was similar to the effect of Clb2 or Clb4 early expression in the absence of Swe1 (Fig. 1 A and C). Deletion of SWE1 in clb5::CLB3 clb6Δ cells advanced S-phase by a small (≈6 min) but reproducible degree (Fig. 1C). The DNA content of unsynchronized cultures supported these findings, because early expression of Clb3 or Clb4 with and, to a greater degree, without Swe1 decreased the proportion of cells with unreplicated DNA content (Fig. 1C). Together, these findings demonstrate that mitotic cyclins can effectively target Cdk1 toward essential replication factors. Our results further demonstrate that Swe1 strongly inhibits Clb2, has intermediate activity toward Clb3 and Clb4, and does not inhibit Clb5 or Clb6, at least toward replication substrates in this in vivo assay (Fig. 1A; for effect on Clb5, compare clb6Δ and clb6Δ swe1Δ, and data is not shown for Clb6).

Late-Firing Replication Origins Can Use Any Clb-Cdk1 for Activation. Next, we examined in more detail the function of mitotic cyclins in origin initiation, particularly late-firing origins. Previous studies have shown that late origin firing is defective in cells lacking Clb5 and that early expression of Clb2 does not compensate for lack of Clb5 (12, 25). However, we suspected that regulation by Swe1 could explain the failure of Clb2 to activate late origins in the previous study. To test this hypothesis, we examined initiation at one early (ARS305) and two late-firing origins (ARS603 and ARS1011) by using 2D agarose gel electrophoresis analysis of replication structures. Efficient activation of all three origins occurred in wild-type cells, indicated by the presence of “bubble” and “large Y” structures (Fig. 2A, filled and open arrowheads, respectively). In clb5Δ cells, early origin firing was efficient; however, late origin firing was inefficient, indicated by an almost complete absence of a bubble arc and corresponding increase in smaller Y structures, which are particularly apparent at the apex of the Y arc (Fig. 2A, double arrowhead). Individually, early expression of Clb2 (clb5::CLB2) or elimination of Swe1 (clb5Δ swe1Δ) increased only very slightly the efficiency of late origin firing in clb5Δ cells (Fig. 2A). However, early expression of Clb2 along with deletion of Swe1 (clb5::CLB2 swe1Δ) significantly increased the frequency of late origin firing in clb5Δ cells, indicating that Clb2 is able to activate late-firing origins (Fig. 2A, compare the relative intensities of bubble and large Y structures to the ascending Y arc and apex of the Y arc).

The failure of late origin firing in clb5Δ cells lengthens S-phase, as demonstrated by the increased proportion of S-phase cells in an unsynchronized culture (Fig. 2B). Cultures of clb5::CLB2 and clb5Δ swe1Δ cells show a similar proportion of S-phase cells as clb5Δ cultures, suggesting no rescue of the replication defect by early expressed Clb2 or the deletion of SWE1 (Fig. 2B). However, the combination of early expressed Clb2 and SWE1 deletion in clb5Δ cells decreased the length of S-phase (Fig. 2B), consistent with the restoration of late origin firing in clb5::CLB2 swe1Δ cells (Fig. 2A). Thus, Clb2 is able to activate early and late firing replication origins and to complete chromosomal replication at a rate similar to wild-type cells.

Although Clb2 clearly was able to increase late origin firing in clb5::CLB2 swe1Δ cells, it was not as efficient as in wild-type cells (Fig. 2A). We suspected that this result was due to the presence of Clb6 (in clb5::CLB2 swe1Δ cells), which, like Clb5, initiates S-phase slightly earlier than Clb2 (data not shown). Thus, Clb6 generates replication forks that can replicate through any unfired origins, particularly late-firing origins, and interfere with our ability to observe their activation by Clb2. To address this possibility, we analyzed late origin efficiency in cells lacking Clb6. In clb5Δ clb6Δ and clb5Δ clb6Δ swe1Δ cells, ARS603 initiated with efficiency similar to wild-type cells; thus, late origin activation can be attributed to one or more of the mitotic cyclins present (Clb1–Clb4) (Fig. 2C) (25). In clb5::CLB2 clb6Δ swe1Δ cells, the early S-phase entry wholly depends on Clb2 (Fig. 1A); hence, late origin firing in this strain should be attributable to Clb2. Late origin activation in clb5::CLB2 clb6Δ swe1Δ cells was nearly as efficient as in clb5Δ clb6Δ and clb5Δ clb6Δ swe1Δ cells (Fig. 2C, data not shown for ARS1011), indicating that Clb2 could effectively activate late origins. Similarly, early expressed Clb3 can drive late origin activation, because ARS603 activation was efficient in clb5::CLB3 clb6Δ swe1Δ cells, in which S-phase entry fully depends on early expressed Clb3 (Fig. 2C). To rule out that Clb3 or Clb4 contributed to late origin firing by Clb2 in clb5::CLB2 clb6Δ swe1Δ cells, we analyzed cells also lacking CLB3 and CLB4. Clb2 stimulated initiation of ARS603 in clb3Δ clb4Δ clb5::CLB2 clb6Δswe1Δ cells as efficiently as in clb5Δ clb6Δ and clb5Δ clb6Δ swe1Δ cells (Fig. 2C). Together with recent data indicating that increased Clb6 dosage can drive late origin firing (22), these data demonstrate that early and late-firing replication origins can use any Clb-Cdk1 for activation (however, Clb1 remains untested).

Phosphorylation of a Critical Replication Factor by Clb2-Cdk1 Is Restricted by Swe1 and Transcriptional Control. To examine directly the ability of Clb2-Cdk1 to target a normal Clb5- and Clb6-Cdk1 substrate, we monitored in vivo phosphorylation of Sld2, whose phosphorylation is essential for replication initiation (26). Based on its mobility shift in SDS/PAGE, the timing of Sld2 phosphorylation corresponded with S-phase entry in all strains (Fig. 3). In particular, Sld2 phosphorylation occurred at ≈24 min in clb6Δ cells and at ≈36 min in clb5::CLB2 clb6Δ swe1Δ cells. Swe1 significantly delayed Sld2 phosphorylation in clb5::CLB2 clb6Δ cells (Fig. 3). Thus, early expressed Clb2-Cdk1 can phosphorylate an essential replication initiation factor, but Swe1 inhibits this activity of Clb2-Cdk1. Interestingly, there are differences in Sld2 phosphorylation by Clb5- and Clb2-Cdk1. Clb5-Cdk1 phosphorylated the majority of Sld2 present, whereas Clb2-Cdk1 phosphorylated only about half of the available Sld2 (Fig. 3). Furthermore, the mobility shift of Sld2 was greater when phosphorylated by Clb5 than by Clb2, suggesting that only a subset of Sld2 phosphorylation sites were targeted by Clb2. Nevertheless, Clb2-Cdk1 phosphorylation of Sld2 clearly is sufficient to drive S-phase entry and support a normal rate of S-phase progression (Fig. 1A). Hence, transcriptional control and Swe1 regulation are potent mechanisms that restrict Clb2 function in the early cell cycle.

Clb2 Can Perform All Essential Cell Cycle Functions. clb3–6Δ cells are inviable, suggesting that Clb1 and Clb2 cannot perform an essential function(s) of Clb3–6 (4). However, we have demonstrated that Clb2 is capable of performing Clb5 and Clb6 replication function(s). Thus, lethality of clb3–6Δ mutant cells may result from the delayed activity of Clb2 (under its normal promoter and Swe1 regulation) relative to Clb3–6, or because Clb2 is unable to perform an essential function(s) of Clb3–6 other than DNA synthesis. If Clb2 is able to perform all of the essential functions of Clb3–6, then early expression of Clb2 and/or elimination of Swe1 might restore the viability of clb3–6Δ cells.

To determine whether Clb2 can substitute for essential Clb3-Clb6 function(s), we first examined whether elimination of Swe1 allows native Clb1 and Clb2 expression to rescue the viability of clb3–6Δ cells. We dissected 100 tetrads of a diploid strain heterozygous for clb3–6Δ and swe1Δ but did not recover clb3–6Δ or clb3–6Δ swe1Δ cells, indicating that endogenous Clb1 and Clb2 expression, even without Swe1 inhibition, cannot replace essential Clb3–6 function(s) (Table 1). To determine whether early expression of Clb2, with or without Swe1, can rescue the essential function(s) of Clb3–6, we analyzed the viability of haploid segregants of a diploid heterozygous for clb3Δ, clb4Δ, clb5::CLB2, clb6Δ, and swe1Δ. Of 176 tetrads dissected, we did not recover clb3Δ clb4Δ clb5::CLB2 clb6Δ cells, indicating that early expression of Clb2 alone cannot replace essential Clb3–6 function(s) (Table 1). However, early expressed Clb2 did rescue the viability of clb3–6Δ swe1Δ cells; we isolated 13 viable clb3Δ clb4Δ clb5::CLB2 clb6Δ swe1Δ spores, which was similar to their expected proportion (Table 1). Clb1 was not required for rescue by early expressed Clb2, because dissection of a diploid additionally heterozygous for clb1Δ yielded 10 viable clb1Δ clb3Δ clb4Δ clb5::CLB2 clb6Δ swe1Δ spores, which was similar to their expected proportion (Table 1). The good viability of cells lacking all Clbs except Clb2 was further demonstrated by their growth on rich medium, which was similar to that of clb5Δ clb6Δ cells (Fig. 4). Thus, Clb2-Cdk1 is inherently able to perform all essential Clb-Cdk1 functions; however, transcriptional regulation and inhibition by Swe1 normally restrict its intrinsically broad target range.

Discussion

Swe1 and Transcription Differentially Regulate the Function of B-Type Cyclins. This study shows that transcriptional control of cyclin gene expression and inhibition of Clb–Cdk1 complexes by Swe1 define two independent mechanisms that differentially regulate the function of CDKs in the early cell cycle. Previous studies had reported that early expression of Clb2 or Clb4 did not efficiently stimulate S-phase entry, suggesting that one or more DNA replication factors were poor substrates for mitotic CDK (11, 12). However, our findings attribute the poor S-phase promoting activity of Clb2 and Clb4 observed in the previous studies primarily to inhibition by Swe1, rather than to an intrinsically poor ability to interact with replication substrates. Indeed, in the absence of Swe1, early expressed Clb2, Clb3, or Clb4 were each able to stimulate S-phase entry with only a slight delay compared with Clb5 (Fig. 1).

Our results argue that Clb2 is an effective activator of S-phase and able to phosphorylate all essential cell cycle substrates. However, the limited phosphorylation of Sld2 and the slight delay in S-phase entry compared with Clb5 suggest that Clb2 is somehow defective in its interaction with at least a subset of S-phase substrates. While this manuscript was in revision, Loog and Morgan reported that Clb5-Cdk1 exhibits greater specificity than Clb2-Cdk1 toward a subset of CDK substrates, including Sld2 (27). Although Clb5-Cdk1 exhibits lower activity than Clb2-Cdk1 toward the majority of CDK substrates, its interaction with certain substrates is enhanced through a “hydrophobic patch” motif on Clb5 and an RXL motif on the substrate (27). Although Clb2 appears to lack this specific targeting motif, higher intrinsic activity of Clb2-Cdk1 (unrestrained by Swe1) may compensate for its lower specificity.

The relatively minor difference in the ability of Clb5- and Clb2-Cdk1 to target replication substrates in vivo suggested by our study contrasts with the great differences in the rates of substrate phosphorylations (up to three orders of magnitude) measured by Loog and Morgan by using purified proteins and emphasizes the difficulty in relating in vitro kinetic differences with an actual biological consequence. For example, incomplete Sld2 phosphorylation is sufficient for S-phase entry (Fig. 3) (26). However, redundancy of phosphorylation sites and differences in the importance of specific sites are not evident in bulk phosphorylation measurements. On the other hand, the sufficiency for S-phase entry may fail to reflect an unrecognized function of complete phosphorylation of Sld2 or other substrate(s) (see below). It is also worth noting that the much more rapid phosphorylation of Clb5-specific substrates in vitro occurred under conditions where the amount of Clb5-Cdk1 greatly exceeded the amount of Clb2-Cdk1 present (27).

To compare the relative kinase activities of Clb2- and Clb5-Cdk1 under our experimental conditions, we measured in vitro phosphorylation of histone H1, which represents a nonspecific CDK substrate (Fig. 5, which is published as supporting information on the PNAS web site). As expected, Clb2-Cdk1 attained a higher activity in the absence of Swe1; however, Clb5-Cdk1 was unaffected. In contrast to the kinetics of Sld2 phosphorylation (Fig. 3), H1 phosphorylation was detected sooner in immunoprecipitates from early expressing Clb2 extracts than Clb5 extracts. Thus, Clb5-Cdk1 is better at phosphorylating Sld2 than Clb2-Cdk1, whereas Clb2-Cdk1 is better at phosphorylating histone H1, consistent with the results of Loog and Morgan. Comparison of H1 kinase levels with the timing of S phase entry strongly suggests that a higher level of Clb2-Cdk1 than Clb5-Cdk1 H1 kinase activity is required to stimulate S-phase entry. In particular, the level of Clb2-associated H1 kinase activity when clb5::CLB2 clb6Δ swe1Δ cells began S-phase (between 24 and 36 min) was greater than Clb5-associated H1 kinase when clb6Δ swe1Δ cells began S-phase (between 12 and 24 min) (Fig. 5; see Fig. 1 for time of S-phase entry). Thus, differences in substrate specificities are apparent in vivo (e.g.: the kinetics of Sld2 phosphorylation and the timing of S-phase entry); however, transcriptional control and specific inhibition of Clb2-Cdk1 by Swe1 have greater functional consequences.

Swe1 Exhibits Specificity Toward CDKs. We have found that Swe1 differentially inhibits the replication-promoting function of CDKs in relation to their normal expression timing, with no inhibition of earliest-expressed Clb5- and Clb6-Cdk1, intermediate inhibition of Clb3- and Clb4-Cdk1, and the strong inhibition of Clb2-Cdk1 (Fig. 1). Analysis of H1 kinase levels also indicates that Swe1 inhibits Clb2-Cdk1 but not Clb5-Cdk1 (Fig. 5). Thus, Swe1 seems capable of reinforcing the stepwise expression of cyclin pairs by modulating the activation of the succeeding cyclin pair. This function may help optimize the temporal execution and coordination of cell cycle events, particularly in relation to bud morphogenesis. In considering the insensitivity of Clb5 to Swe1, it seems important that Clb5 has relatively weak activity toward most substrates, because a broadly potent, Swe1-insensitive CDK would likely override the Swe1 checkpoint, even in the presence of a Swe1-regulated CDK such as Clb2-Cdk1.

The relative Swe1 sensitivity of the Clbs also correlates with their similarities to each other, with Clb5 and Clb6 being most divergent from Clb1 and Clb2 and with Clb3 and Clb4 lying in between them (28). Interestingly, each of these cyclin pairs differs from the others in the “hydrophobic patch” motif; however, a role for this motif in targeting cyclins other than Clb5 remains to be demonstrated. Further studies will be required to identify the exact features that distinguish the functions of Clb2, Clb3, and Clb4. It will also be quite interesting to identify the features responsible for their distinct susceptibilities to Swe1 inhibition.

Evolution of Distinct CDK Functions and the Importance of Checkpoints. It has been suggested that the relative timing of S- and M-phases in a primordial eukaryote could have been ensured by oscillation of the activity of a single CDK (i.e., without a need for cyclin-specific functions) if replication-promoting functions simply required a lower overall level of CDK activity than mitotic functions (2, 29, 30). The finding that Clb2-Cdk1 alone can sustain all cell cycle functions clearly supports this possibility. Similarly, in Schizo-saccharomyces pombe, Cdc13 (which is functionally homologous to Clb2 and regulated by Wee1) can support cellular viability in the absence of the S-phase cyclin Cig2 (30). On the other hand, the relatively high level of Clb2-Cdk1 required to drive S phase (Fig. 5) argues against the premise that replication substrates might generally require a lower level of CDK function for activation to ensure that replication precedes mitosis. This finding suggests that without a highly specific mechanism for targeting replication substrates, the oscillation of a single CDK might not provide sufficient differentiation between the relative timing of S- and M-phases. Although CDK activity provides the driving force for execution of cell cycle events, much of the coordination between cell cycle processes is regulated by checkpoint kinases.

The evolution of cell cycle regulation by oscillation of a single CDK might have required the prior existence of cell cycle checkpoints to ensure that genome duplication be completed before its segregation (31, 32). Early expression of Clb2 and deletion of Swe1 did not notably advance cell division (data not shown), consistent with activation of a DNA and/or spindle checkpoint preventing premature mitotic entry that might result from early activation of a Clb2 substrate. In fact, we have detected activation of the DNA damage and replication checkpoint kinase Rad53 in the clb5::CLB2 clb6Δ swe1Δ cells (data not shown), consistent with the idea that a checkpoint acts to maintain the proper order of chromosome replication and segregation in cells lacking more specialized mechanisms of CDK regulation (including normal Clb2 transcription, Swe1 inhibition, and the presence of Clb5 and Clb6). An alternative, but nonexclusive explanation for activation of Rad53 in clb5::CLB2 clb6Δ swe1Δ cells is that Clb2 may be defective in phosphorylation of a Clb5- or Clb6-specific substrate or a specific phosphorylation site(s), which is not essential for S-phase but required to avert DNA damage during replication. This possibility may be exemplified by Sld2, which appears to be incompletely phosphorylated by Clb2-Cdk1 (Fig. 3).

In addition to regulation of CDKs, the temporal presentation of substrates (such as prereplicative complexes, replisomes, and kinetochores) probably also is a major factor in determining which CDK–substrate interactions occur. Cellular localization also plays a role in restricting a specific cyclin-Cdk activity. For example, Xenopus cyclin B-Cdk1, which localizes to the cytoplasm during S-phase, was recently shown to be capable of activating replication origins when relocalized to the nucleus (33). Together with our results, these findings and others discussed above suggest that the evolutionary diversification of B-type cyclins involves multiple mechanisms, including: (i) the emergence of new protein–protein interactions with a cyclin- or Cdk-specific regulator (such as Swe1), which does not necessarily require alteration of existing enzyme active site–substrate interactions, (ii) changes in expression timing, (iii) regulation of subcellular protein localization, and (iv) the emergence of specific cyclin–substrate interactions. Duplicated gene families have been indispensable participants in the evolution of protein diversity and, hence, multicellular organisms. This study has provided insights into the functional specialization of a critically important protein family.