The Cdc4/34/53 pathway targets Cdc6p for proteolysis in budding yeast

Lucy S. Drury, Gordon Perkins, John F.X. Diffley

Author Affiliations

  1. Lucy S. Drury1,
  2. Gordon Perkins1 and
  3. John F.X. Diffley1
  1. 1 Imperial Cancer Research Fund, Clare Hall Laboratories, South Mimms, EN6 3LD, UK


The budding yeast Cdc6 protein (Cdc6p) is essential for formation of pre‐replicative complexes (pre‐RCs) at origins of DNA replication. Regulation of pre‐RC assembly plays a key role in making initiation of DNA synthesis dependent upon passage through mitosis and in limiting DNA replication to once per cell cycle. Cdc6p is normally only present at high levels during the G1 phase of the cell cycle. This is partly because the CDC6 gene is only transcribed during G1. In this article we show that rapid degradation of Cdc6p also contributes to this periodicity. Cdc6p degradation rates are regulated during the cell cycle, reaching a peak during late G1/early S phase. Removal of a 47‐amino‐acid domain near the N‐terminus of Cdc6p prevents degradation of Cdc6p. Likewise, mutations in the Cdc4/34/53 pathway involved in ubiquitin‐mediated degradation block proteolysis and genetic evidence is presented indicating that the N‐terminus of Cdc6p interacts with the Cdc4/34/53 pathway, probably through Cdc4p. A stable Cdc6p mutant which is no longer degraded by the Cdc4/34/53 pathway is, none the less, fully functional. Constitutive overexpression of either wild‐type or stable Cdc6p does not induce re‐replication and does not induce assembly of pre‐replicative complexes after DNA replication is complete.


The Saccharomyces cerevisiae Cdc6 protein and its Schizosaccharomyces pombe homologue cdc18+ play central roles in DNA replication (Diffley, 1996; Nasmyth, 1996; Stillman, 1996; Wuarin and Nurse, 1996). These proteins are essential for initiating DNA replication and must be synthesized during the G1 phase of each cell cycle. In the absence of de novo synthesis of Cdc6p or cdc18+ during G1, cells are unable to enter S phase and, instead, rapidly lose viability and undergo chromosome segregation without replication (Kelly et al., 1993; Piatti et al., 1995).

Budding yeast replication origins exist in two chromatin states during the cell cycle, both on high copy plasmids and in their normal chromosomal location (Brown et al., 1991; Diffley et al., 1994; Santocanale and Diffley, 1996). After replication origins fire they are in the ‘post‐replicative state’, which is characterized primarily by binding of the six subunit origin recognition complex (ORC) (Bell and Stillman, 1992; Diffley and Cocker, 1992; Rao and Stillman, 1995; Rowley et al., 1995; Santocanale and Diffley, 1996). At the end of mitosis a ‘pre‐replicative complex’ (pre‐RC) assembles, which is characterized by an additional region of protection adjacent to the ORC binding site (Diffley et al., 1994). This is approximately the time when Cdc6p first appears (Piatti et al., 1995). In the absence of de novo Cdc6p synthesis, pre‐RCs do not form at replication origins (Cocker et al., 1996) and DNA replication cannot occur (Piatti et al., 1995; Detweiler and Li, 1997). In addition, Cdc6p is bound to chromatin during G1 (Donovan et al., 1997) and pre‐RCs are thermolabile in vivo in a cdc6 temperature‐sensitive mutant (Cocker et al., 1996; Detweiler and Li, 1997). CDC6 interacts genetically with the origin recognition complex (ORC) (Li and Herskowitz, 1993; Liang et al., 1995; Loo et al., 1995) and recombinant Cdc6p can interact in cell extracts with ORC (Liang et al., 1995). Cdc6p is required, at least in part, to load the Mcm proteins onto pre‐replicative chromatin, a reaction which has been conserved from yeast to Xenopus (Coleman et al., 1996; Donovan et al., 1997). Together these results indicate that Cdc6p plays a direct role as a component of functional pre‐initiation complexes at replication origins in budding yeast.

In both fission and budding yeasts cyclin‐dependent protein kinases (cdks) play a key role in blocking re‐replication (Broek et al., 1991; Hayles et al., 1994; Moreno and Nurse, 1994; Dahmann et al., 1995; Piatti et al., 1996). This occurs in budding yeast, at least in part, because Cdc28p (cdk1), together with the B‐type cyclins, blocks formation of pre‐RCs. Expression of Cdc6p during G1 can promote efficient formation of pre‐RCs and subsequent DNA replication. However, Cdc6p expressed after a ‘point of no return’ in late G1 can no longer promote pre‐RC formation and DNA replication (Piatti et al., 1996). This point of no return occurs concomitantly with the activation of Cdc28p kinase activity by the S phase‐promoting B‐type cyclins Clb5 and Clb6 and can be delayed by deletion of the CLB5 and CLB6 genes. Furthermore, overexpression of the Clb kinase inhibitor Sic1p in G2/M is sufficient to drive pre‐RC formation (Dahmann et al., 1995). Because entry into S phase requires the presence of at least one Clb (Schwob et al., 1994) and because Clb kinases are present from the beginning of S phase until the end of mitosis, new pre‐RCs cannot assemble until after mitosis. Therefore, inhibition of pre‐RC formation by Clb kinase appears to play a critical role in limiting DNA replication to once per cell cycle and in making entry into S phase dependent upon passage through mitosis. Understanding how Clb kinases block Cdc6p‐dependent pre‐RC formation is an important goal in understanding how replication is regulated.

Both CDC6 and cdc18+ are transcribed during a narrow window in the G1 phase of the cell cycle (Zhou and Jong, 1990; Kelly et al., 1993; Zwerschke et al., 1994; Piatti et al., 1995) and the levels of Cdc6 and cdc18+ proteins closely mirror the levels of transcript, indicating that they are unstable proteins (Nishitani and Nurse, 1995; Piatti et al., 1995; Muzi‐Falconi et al., 1996). An important issue at present is whether regulation of levels and stability of these proteins is important for orderly cell cycle regulation. Overexpression of either cdc18+ or the cdk inhibitor rum1+ causes re‐replication in fission yeast. Overexpression of rum1+ leads to accumulation of cdc18+ and it has been suggested that rum1+‐induced endoreduplication might occur by stabilization of cdc18+ (Jallepalli and Kelly, 1996). On the other hand, cdc4 mutants (see below) stabilize Cdc6 protein and allow its accumulation in G2/M nuclei without causing re‐replication, suggesting that proteolysis of Cdc6p is not the only mechanism by which re‐replication is blocked in budding yeast (Piatti et al., 1996). In neither case, however, is it possible to rule out secondary effects; rum1+ overproduction might have an effect unrelated to cdc18+ stability which is responsible for inducing re‐replication, while the cdc4 mutant might also be defective in degradation of an additional factor, such as Sic1p, which might prevent re‐replication. Therefore, a better understanding of the mechanisms regulating degradation of these proteins is required.

In this paper we examine degradation of Cdc6p during the cell cycle. We show that a 47‐amino‐acid domain in the N‐terminus of Cdc6p is essential for targeting it for degradation via the Cdc4/34/53 pathway. We also show that stable mutants of Cdc6p support normal, regulated formation of pre‐replicative complexes and do not exhibit any evidence of re‐replication. These results are discussed with respect to regulation of pre‐replicative complex formation in budding and fission yeast.


Identification of Cdc4p as a Cdc6p‐interacting protein

In order to better understand the function and regulation of Cdc6p, we searched for Cdc6p‐interacting proteins. We have employed a two‐hybrid screen (Fields and Song, 1989) using the entire Cdc6p fused to the LexA DNA binding domain as ‘bait’ and a library of yeast cDNAs fused to the Gal4 transcriptional activation domain (Durfee et al., 1993). A single positive clone that interacted with the Cdc6p–LexA fusion but not LexA alone was identified in this analysis. Sequencing revealed that this clone contained a truncated copy of the CDC4 gene fused in‐frame to the Gal4p activation domain. Genetic experiments indicate that CDC4 interacts with CDC34, CDC53 and SKP1 (Mathias et al., 1996). Cdc34p is an E2‐type ubiquitin conjugating enzyme (Goebl et al., 1988) and, together, Cdc4p, Cdc34p, Cdc53p and Skp1p target the Cdk inhibitor Sic1p for degradation at the end of the G1 phase (Schwob et al., 1994). The Cdc4 protein plays an unknown but essential role in this complex. It is composed of an ‘F box’, required for interaction with Skp1p (Bai et al., 1996), and a series of ‘WD‐40’ repeats (Fong et al., 1986). All of the WD‐40 repeats of Cdc4p are present in the clone we have isolated, while the F‐box is absent (Figure 1A).

Figure 1.

Two‐hybrid interaction between Cdc6p and Cdc4p. (A) Features of the truncated Cdc4p clone isolated in a two‐hybrid screen with LexACDC6. The relative positions of the F box and the seven WD40 repeats in full‐length Cdc4p (amino acids 1–779) are shown above. Below is an indication of the Cdc4p sequences found in the rescued clone (amino acids 339–779), which contains the WD‐40 region but not the F box. (B) Cells containing the indicated Gal4 activation domain (GAD) and LexA two‐hybrid vectors were assayed for β‐galactosidase activity. The column on the left shows cells expressing GAD alone. The column on the right shows cells expressing the GAD–Cdc4 fusion protein. Cells also expressed LexA fused to the indicated regions of Cdc6p.

We next used this CDC4‐containing clone to identify regions of Cdc6p that are necessary for these interactions in the two‐hybrid system (Figure 1B). Initial analysis indicated that the first 47 amino acids of Cdc6p were sufficient to interact with Cdc4p (Figure 1B), while the C‐terminal 466 amino acids did not (data not shown). Subdivision of the N‐terminal 47‐amino‐acid domain showed that Cdc4p can interact with amino acids 1–18 as well as amino acids 17–47, but the second element (17–47) induced considerably more β‐galactosidase activity. This may reflect a higher affinity of Cdc4p for this region, although other explanations are also possible.

The Cdc4p‐interacting domain of Cdc6p is required for its degradation

In Figure 2A wild‐type (W303‐1a) cells were first synchronized in G1 with α‐factor mating pheromone and then released. Levels of Cdc6p, expressed from its own promoter, were then examined by immunoblotting with a monoclonal antibody to Cdc6p. This experiment demonstrates that Cdc6p levels are sharply periodic, appearing in the second and third cycles at the end of mitosis and disappearing as buds emerge, which is approximately the time that DNA replication begins. This confirms previous results reported by Piatti et al. (1995) and demonstrates that the epitope tag used by these investigators did not significantly alter Cdc6p degradation.

Figure 2.

Removal of the Cdc4p interaction domain stabilizes Cdc6p. (A) A culture of W303‐1a cells was grown in YPD until the cells reached a density of 1×107 cells/ml, whereupon they were blocked with α‐factor. After the population was uniformly blocked (∼3 h), α‐factor was removed by centrifugation and washing. Cells were resuspended in fresh YPD at 1×107 cells/ml and returned to 30°C. At the indicated times samples were taken to assess the budding index of the cells and for determination of protein levels by immunoblotting. (B) YLD24 (GAL1‐10–CDC6) and YLD25 (GAL1‐10–CDC6ΔNT) were grown in YP‐GAL plus 2 mM methionine until they reached a density of 5×106 cells/ml, whereupon they were blocked with nocodazole or α‐factor. When the cells were blocked glucose was added (T = 0 min) to repress the GAL1‐10 promoter. Samples were processed and immunoblotted to detect both Cdc6p and Cdc6ΔNTp.

If the two‐hybrid interactions described above are of biological significance, we can make two predictions. Firstly, deletion of the N‐terminal 47 amino acids should render Cdc6p more stable in vivo and, secondly, Cdc6p should be rendered more stable in cdc4 mutants

To test the first prediction we used a strain containing a single copy of the intact CDC6 gene under the control of the MET3 promoter (Piatti et al., 1995, 1996; Cocker et al., 1996; Donovan et al., 1997). In the presence of methionine, which represses this promoter (Cherest et al., 1987), this strain rapidly ceases to proliferate (Piatti et al., 1995). Into this strain we have integrated single copies of either full‐length CDC6 or a 5′ deletion removing amino acids 2–47 (Cdc6ΔNT), each under the control of the GAL1‐10 promoter. In the presence of methionine and galactose only CDC6 genes under the control of the GAL1‐10 promoter are expressed. In Figure 2B cells were first blocked in either G2/M with nocodazole or in G1 with α‐factor. The full‐length and truncated CDC6 genes were then switched off by addition of glucose and the rate of Cdc6p or Cdc6ΔNTp disappearance was followed by immunoblotting. Lanes 1–10 indicate that the full‐length protein disappears rapidly after transcriptional repression in both G1 and G2/M blocked cells (t1/2 = 13.5 and 14 min respectively). Figure 2B, lanes 11–20 demonstrate that, unlike the full‐length protein, truncated Cdc6ΔNTp disappears only very slowly after transcriptional repression in both G1 and G2/M (t1/2 = 76 and 46 min respectively). Therefore, in both G1 and G2/M the rate of Cdc6p degradation is significantly enhanced by the presence of amino acids 2–47.

Cdc6p degradation requires the CDC4, CDC34 and CDC53 gene products

Stabilization of Cdc6p in cdc4 mutants has previously been shown in cells blocked in G2/M with the microtubule inhibitor nocodazole (Piatti et al., 1996). We have extended this to examine whether two other members of this epistasis group, Cdc34p and Cdc53p, were also required for Cdc6p destruction during G2/M. To address this we have constructed strains in which the full‐length CDC6 gene was placed under the control of the GAL1‐10 promoter. The rate of Cdc6p disappearance in wild‐type cells as well as cdc4, cdc34 and cdc53 mutant cells was examined as follows. After blocking in G2/M with nocodazole, cultures were shifted to the restrictive temperature (37°C) in galactose‐containing medium. After 30 min transcription was repressed by addition of glucose at the restrictive temperature and the rate of Cdc6p disappearence was measured by immunoblotting. In the parental wild‐type strain Cdc6p rapidly disappears (Figure 3A, lanes 1–4), so that by 15 min >50% of the protein is gone. In the cdc4 (lanes 5–8), cdc34 (lanes 9–12) and cdc53 (lanes 13–16) mutants, however, Cdc6p levels remain constant for 60 min. Therefore, Cdc6p destruction requires all three gene products in cells blocked in G2/M. When similar experiments were performed on strains harbouring mutations in genes encoding components of the anaphase‐promoting complex (APC) (Hershko et al., 1994; Lamb et al., 1994; Irniger et al., 1995; King et al., 1995, 1996; Sudakin et al., 1995; Tugendreich et al., 1995; Nasmyth, 1996; Peters et al., 1996; Zachariae et al., 1996) (cdc16, Figure 3B, lanes 1–5; cdc23, Figure 3B, lanes 6–10) Cdc6p was not stabilized, suggesting that the APC is probably not required for Cdc6p proteolysis during G2/M.

Figure 3.Figure 3.
Figure 3.

Cdc6p is stabilized in cdc4/34/53 temperature‐sensitive mutant cells. (A) Logarithmically growing cultures of YLD15 (wild‐type), YLD16 (cdc4), YLD17 (cdc34) and YLD18 (cdc53) were blocked with nocodazole. When 90% of the cells were blocked the cultures were resuspended in YP‐GAL containing nocodazole at 37°C to induce the GAL1‐10 promoter. At time T = 0 min GAL1‐10–CDC6 expression was repressed by addition of glucose. Samples were taken and processed for immunodetection of Cdc6p. (B) An experiment similar to that shown in (A) was performed using YLD13 (cdc16) and YLD14 (cdc23), which also harbour an integrated copy of GAL1‐10–CDC6. (C) and (D) MTY668 and its congenic parental strain were blocked in G2 with nocodazole. Cells were released from the block into YPD containing α‐factor at 37°C. Samples were taken every 10 min to assess budding index (D) and to monitor the levels of endogenous Cdc6p (C) as described in Materials and methods.

Cdc6p is not normally present in cells during G2/M however (Piatti et al., 1995) and, moreover, the experiments described above involve expression of Cdc6p from the strong GAL1‐10 promoter. We therefore sought to examine the genetic requirements for Cdc6p degradation under more physiological conditions. More specifically, we sought to determine if Cdc4p is essential for degradation of Cdc6p when Cdc6p is expressed from its own promoter during G1. To address this we arrested wild‐type or cdc4 mutant cells in nocodazole at 25°C and then released them into α‐factor mating pheromone at the cdc4 restrictive temperature. Consistent with what happens during a normal cell cycle, in wild‐type cells Cdc6p first accumulates after mitosis and is then largely destroyed during G1 (Figure 3C, lanes 1–10).

A residual population of Cdc6p remains for an extended period in these α‐factor‐blocked cells (lanes 7–10). A significant fraction of the Cdc6p present in α‐factor‐blocked cells is found bound to chromatin (Donovan et al., 1997) which, taken with the result described above, suggests that Cdc6p in pre‐RCs may be more stable than free Cdc6p. Note that expression of Cdc6p from the GAL1‐10 promoter, even for short periods of time (e.g. Figure 2B), results in levels of Cdc6p which are very much higher than is normally seen during the cell cycle (data not shown). Thus the fraction of Cdc6p in pre‐RCs is likely to be exceedingly small. Longer exposures of the immunoblot in Figure 2B, lanes 1–10 show that significant amounts of Cdc6p are still present after 60 min (data not shown).

In the cdc4 mutant (Figure 3C, lanes 11–20) Cdc6p accumulates to significantly higher levels and is not destroyed during the course of the experiment. Therefore, the dominant pathway for Cdc6p degradation when expressed at normal levels from its own promoter involves the Cdc4/34/53 pathway. Previous experiments have indicated that Cdc4p has an additional role late in the cell cycle (Schwob et al., 1994). In the experiment shown in Figure 3C and D cdc4 mutant cells proceed through mitosis and enter G1 with normal kinetics (Figure 3D), arguing that Cdc4p is not required after the nocodazole block.

Finally, additional evidence for interaction between the N‐terminus of Cdc6p and the Cdc4/34/53 pathway is shown in Figure 4. Overproduction of Cdc6p does not alter the morphology of wild‐type W303‐1a cells (Figure 4A). Overproduction of Cdc6p does not block cell growth but does cause a dramatic change in cell morphology at the permissive temperature in a congenic strain harbouring the cdc4 mutant (Figure 4C). Similar results are seen when full‐length Cdc6p is overproduced in cdc34 and cdc53 mutants (data not shown). This phenotype is very similar to the terminal phenotype these mutants exhibit at the restrictive temperature; in each case the cells become multiply budded and elongated. The N‐terminus of Cdc6p is critical for generating this phenotype, since overproduction of Cdc6ΔNTp did not effect cell morphology (Figure 4D). Taken together, these results indicate that the N‐terminal 47 amino acids of Cdc6p target it for destruction via the Cdc4/34/53 pathway, probably via interactions with Cdc4p.

Figure 4.

Genetic evidence for interaction between the N‐terminus of Cdc6p and Cdc4p. W303‐1a cells overexpressing Cdc6p or Cdc6ΔNTp from the GAL1‐10 promoter at 25°C do not exhibit any change in cell morphology (A and B respectively). Cdc6p overproduction in a congenic cdc4 mutant growing at 25°C causes the cells to become elongated and multiply budded (C). This effect is mediated by the N–terminus of Cdc6p, since overproduction of Cdc6ΔNTp in the cdc4 strain (D) had no observable effect.

Post‐transcriptional regulation of Cdc6p levels in the cell cycle

Since the CDC6 gene is periodically transcribed roughly coincident with the appearance of Cdc6p (Zhou and Jong, 1990; Zwerschke et al., 1994; Piatti et al., 1995), the periodic appearance of Cdc6p might simply be explained by periodic transcription followed by constitutive rapid degradation. Alternatively, the stability of Cdc6p may also vary during the cell cycle. To begin to address this, logarithmically growing cells expressing Cdc6p constitutively from the GAL1‐10 promoter were synchronized by arrest in G1 with α‐factor followed by release from the block. Cdc6p levels were monitored after release from the block and throughout the following cell cycles. As can be seen in Figure 5A, even with constitutive CDC6 expression, levels of Cdc6p oscillate dramatically during the cell cycle. The rapid drop in Cdc6p level begins with the emergence of buds (30 min, lane 4) and by the time all of the cells are budded, Cdc6p is barely detectable (40 and 50 min, lanes 5 and 6). Cdc6p levels start to rise again later in the cell cycle, well before cell division (lanes 7–11), and reach a maximum again during the next G1 period (lanes 12–15). A second period of Cdc6p disappearance can be seen in the next cell cycle, again concomitant with bud emergence (lanes 16–17). This result suggests that Cdc6p is unable to accumulate during late G1 and/or early S phase. Additional block and release experiments have indicated that the inability to accumulate Cdc6p during this period requires the N‐terminal 47 amino acids of Cdc6p (data not shown).

Figure 5.

Post‐transcriptional regulation of Cdc6p levels during the cell cycle. (A) A culture of YLD23 (GAL1‐10–CDC6) grown in YP‐GAL was first synchronized with α‐factor and then released into fresh medium. After release from the cell cycle block samples were taken at 10 min intervals for 180 min. The budding index was determined to assess synchrony and immunoblotting was performed to observe levels of Cdc6p at each time point. Lanes 1–18 in the immunoblot correspond to the time points above each lane. (B) YLD29 (cdc7, GAL1‐10–CDC6) was grown in YP‐Raf at 25°C to mid‐log phase and synchronized with α‐factor. Once a 90% block was achieved the cells were harvested, washed and resuspended in YP‐GAL at 37°C. After bud emergence (80 min later) expression of CDC6 was repressed by addition of glucose. Samples were taken and processed for immunoblotting to detect Cdc6p. (C) YLD15 and YLD16 were grown in YP‐Raf at 25°C to mid‐log phase and then blocked with HU (0.2 M). The cells were left to block for ∼5 h, at which time >95% cells were arrested with large buds. The cells were harvested and resuspended in YP‐GAL plus HU. After 30 min glucose was added at 2% to repress CDC6 expression from the GAL1‐10 promoter. Samples were taken at the times indicated and processed for immunoblotting.

Figure 5B shows that the inability to accumulate Cdc6p is due, at least in part, to an increase in the rate of Cdc6p degradation at the end of G1. Cells were blocked at the G1/S boundary using a cdc7 temperature‐sensitive mutant and Cdc6p degradation rates were measured as above. In this experiment the degradation of Cdc6p was so rapid that we were unable to estimate its half‐life, since the protein had largely disappeared by the first time point after transcriptional repression (t1/2 < 5 min). In longer exposures of this immunoblot (data not shown) a low residual level of Cdc6p which does not disappear can be seen, consistent with the idea that Cdc6p in pre‐RCs is more stable than free Cdc6p. These experiments show that Cdc6p degradation rates are not constant over the cell cycle but, instead, peak as cells pass from G1 into S phase.

Figure 5C shows that wild‐type cells blocked in S phase with the ribonucleotide reductase inhibitor hydroxyurea (HU) are unable to accumulate high levels of Cdc6p. cdc4 mutant cells, however, are capable of accumulating high levels of Cdc6p when blocked in S phase with HU. Moreover, this Cdc6p is completely stable after transcriptional repression. From these experiments we conclude that Cdc6p becomes hyper‐unstable during late G1 and S phase and is degraded during this period by a Cdc4‐dependent pathway.

Stable Cdc6p supports normal cell cycle progression and regulated pre‐RC assembly and does not induce re‐replication

The experiments described so far reveal rich post‐transcriptional regulation of Cdc6p levels. Because of this and because of the effects of cdc18+ overexpression in fission yeast, it was of particular interest to investigate the effects of stabilization of Cdc6p. Identification of the Cdc4p‐interacting domain and generation of the stable Cdc6pΔNT mutant has allowed us to do this. Strains containing both MET3–CDC6 and either GAL1‐10–CDC6 or GAL1‐10–CDC6ΔNT were first plated under conditions where Met–CDC6 was repressed and either GAL1‐10–CDC6 or GAL1‐10–CDC6ΔNT were expressed. As shown in Figure 6A, both full‐length and truncated, stable Cdc6p are capable of supporting growth to a similar extent.

Figure 6.Figure 6.
Figure 6.

Cdc6ΔNTp supports normal growth and regulated pre‐replicative complex formation and does not induce re‐replication. (A) YLD23 (vector alone), YLD24 (GAL1‐10–CDC6) and YLD25 (GAL1‐10–CDC6ΔNT) were streaked out on agar plates to test whether Cdc6ΔNTp could support growth. (Panel i) plates contained 2% glucose and 2% galactose as well as 2 mM methionine (repressing both the GAL1‐10 and MET3 promoters). (Panel ii) plates contained 2% glucose and 2% galactose but without methionine (repressing the GAL1‐10 promoter while inducing the MET3 promoter). (Panel iii) plates contained 2% galactose plus methionine (inducing the GAL1‐10 promoter, repressing the MET3 promoter). (Panel iv) plates contained galactose, without methionine (inducing both the GAL1‐10 and MET3 promoters). (B) YLD24 (GAL1‐10–CDC6) and YLD25 (GAL1‐10–CDC6ΔNT) were grown in YP‐GAL to mid‐log phase and blocked with either nocodazole or α‐factor. Samples were then processed for footprinting of the 2 μ origin of DNA replication. Two DNase I concentrations are shown for each sample. The position of the ORC‐induced hypersensitive site is marked with an asterisk in G2 blocked cells. The extended region of protection corresponding to the pre‐RC is marked with a bracket. (C) Cultures of W303‐1a, YLD24 (CDC6), YLD25 (CDC6ΔNT) and YLD26 (CDC6NLSΔNT) were grown overnight in YP‐GAL, diluted the next morning and grown to mid‐log phase in YP‐GAL. Samples were taken and prepared for FACS.

Recently, Elsasser et al. (1996) have tested a very similar N‐terminal Cdc6p deletion and have concluded that their truncation was less functional than the full‐length protein. This conclusion was based on the fact that while both full‐length and truncated Cdc6p could support growth on galactose, only the full‐length protein could support growth on medium containing both glucose and galactose, conditions that should severely repress this promoter. Although we agree that truncated Cdc6p does not support growth on glucose and galactose, we also find that even the full‐length protein is unable to support growth under these conditions (Figure 6A), thus we see no differences between the full‐length and truncated Cdc6p. Although the reasons for this discrepancy are unresolved, one possibility is that the full‐length construct used by Elsasser et al. contained 52 bp of Cdc6 promoter sequence absent from their truncation and absent from both of our constructs. It is possible that the growth supported by full‐length Cdc6p is due to residual basal transcription from the remaining Cdc6 promoter. In our hands both full‐length and truncated Cdc6p support very similar growth rates on galactose and lose ARS‐containing plasmids at similar rates (data not shown) under these conditions. Although the two proteins support very similar levels of growth and DNA replication, we cannot conclude unequivocally that Cdc6p and Cdc6ΔNTp are equally functional, since comparison is inevitably complicated by the fact that Cdc6ΔNTp is more stable and, therefore, always more abundant than full‐length Cdc6p.

The N‐terminal 47 amino acids contain four of the eight S/TP motifs present in Cdc6p, which represent relaxed consensus sequences for cdk1 kinases (Langan et al., 1989; Shenoy et al., 1989). Furthermore, Elsasser et al. have shown that Cdc28–Clb kinase interacts strongly with Cdc6p through the N‐terminal 47 amino acids (Elsasser et al., 1996). Since Clb kinases block formation of Cdc6p‐dependent pre‐RCs (Dahmann et al., 1995), it was of considerable interest to examine whether Cdc6ΔNTp‐dependent pre‐RC formation was still under cell cycle regulation. To address this we used genomic footprinting to examine the 2 μ origin in α‐factor‐ (Figure 6B, lanes 1, 2, 5 and 6) and nocodazole‐arrested (Figure 6B, lanes 3, 4, 7 and 8) cells expressing either full‐length (lanes 1–4) or truncated (lanes 5–8) Cdc6p constitutively from the GAL1‐10 promoter. In both strains the ORC‐induced hypersensitive site (asterisk) is clearly visible in nocodazole‐arrested cells. In G1 blocked cells, a strong pre‐RC is seen. This is manifested as suppression of the ORC‐induced hypersensitive site. Additionally, DNase I cleavage sites within the bracketed region are relatively less intense in the pre‐RC. From this we conclude that the truncated Cdc6 protein supports efficient pre‐RC formation during G1. Moreover, neither the full‐length nor the stable Cdc6p support any detectable pre‐RC formation when overproduced during G2.

Finally, we have tested whether overexpression of stable or unstable Cdc6p could cause re‐replication. To address this we examined the DNA content of cells overproducing either full‐length or stable Cdc6p by fluorescence‐activated cell sorting (FACS) analysis. The N‐terminal 47 amino acids of Cdc6p contain a sequence which resembles a nuclear localization sequence and which can promote nuclear localization when fused to β‐galactosidase (Jong et al., 1996). Therefore, to rule out effects caused by differences in subcellular localization, we have also tested a construct in which this N‐terminal domain of Cdc6p was replaced with the nuclear localization signal sequence from SV40 large T antigen. This protein, Cdc6ΔNT–NLSp, is considerably more stable than the full‐length protein and is not degraded by the Cdc4/34/53 pathway (data not shown). Cdc6p, Cdc6ΔNTp or Cdc6ΔNT–NLSp were expressed for 48 h from the GAL1‐10 promoter and DNA content was analysed by FACS analysis. As a control for 1C and 2C DNA contents, W303‐1a cells were treated in an identical manner and subjected to FACS analysis. As shown in Figure 6C, overexpression of full‐length Cdc6p, Cdc6ΔNTp or Cdc6ΔNT–NLSp does not result in accumulation of significant numbers of cells with a DNA content measurably greater than 2C. Thus even high levels of overexpression of stable Cdc6 protein do not induce endoreduplication in budding yeast.


Mechanism of Cdc6p degradation

The results described in this paper indicate that a short sequence in the N‐terminus of Cdc6p targets it for rapid degradation via the Cdc4/34/53 pathway. This sequence interacts with Cdc4p in a two‐hybrid assay, suggesting that Cdc4p may play a role in substrate recognition. This interaction may be direct or may occur through some other component of the pathway that in turn interacts with Cdc4p. The Cdc4p‐containing clone we isolated does not contain the ‘F box’ and, therefore, cannot interact with Skp1 (Bai et al., 1996). Consequently, if the Cdc4p–Cdc6p interaction is indirect, it is unlikely to occur through Skp1. Further biochemical analysis is required to analyse these interactions and the availability of a short Cdc4p‐interacting sequence should facilitate this analysis.

PEST sequences have been postulated to play a role in directing unstable proteins for degradation (Rechsteiner and Rogers, 1996) and a single PEST sequence (score 5.15, where scores >5 are considered significant) is found between amino acids 32 and 46. Consistent with the idea that the PEST sequence may be important for interaction, a deletion that removes it (compare Figure 1, amino acids 17‐47–17‐33) has the consequence that the protein no longer interacts with Cdc4p. While this PEST sequence may be important for Cdc4p interaction, we note that amino acids 1–18, which interact weakly with Cdc4p, do not contain a near match to the PEST sequence.

Two lines of evidence suggest that additional sequences within Cdc6p are also essential for targeting it for degradation. Firstly, the N‐terminus of Cdc6p does not confer instability when fused to a heterologous protein, β‐galactosidase (data not shown). Since this domain is sufficient for Cdc4p interaction in the two‐hybrid system (see Figure 1), this suggests that Cdc4p interaction, though essential, is not sufficient to target a protein for degradation. Secondly, we have recently identified a point mutation within the C‐terminal half of Cdc6p which results in significant Cdc6p stabilization (data not shown). Further experiments are underway to characterize this second domain of Cdc6p.

Many important details regarding the mechanism of Cdc6p degradation remain to be worked out. For example, the requirement of Cdc34p for degradation of Cdc6p suggests that degradation might occur via ubiquitination of Cdc6p, however, at present we do not know if Cdc6p is ubiquitinated in vivo or in vitro by Cdc34p. In addition, we do not know if proteolysis is catalysed by the proteosome. However, given the interactions between Cdc6p and the Cdc4/34/53 pathway, we think this is most likely to be the mechanism.

Regulation of Cdc6p degradation

Phosphorylation of substrate proteins is thought to play a critical role in targeting them for degradation via the Cdc4/34/53 pathway. Thus Sic1p is stable during G1 but becomes destabilized upon activation of G1 cyclins (Schwob et al., 1994). Likewise, Cdc53p has been shown to bind to Cln2 only when Cln2 has been phosphorylated (Willems et al., 1996). At present we do not know if Cdc6p phosphorylation plays a role in regulating its degradation, however, the fact that the rate of Cdc6p degradation varies during the cell cycle (Figure 5) suggests that this may be the case. Interestingly, both Cdc6p and Sic1p become unstable late in G1, suggesting that they may be regulated in a similar manner. Cdc6p and Sic1p differ, however, in the fact that Sic1 is extremely stable in α‐factor‐blocked cells while Cdc6p is unstable.

Why is Cdc6p an unstable protein?

The experiments presented in this paper demonstrate that free Cdc6p is rapidly degraded, especially around the time that DNA replication initiates. Yet deletion of the Cdc4p interaction domain, which greatly stabilizes the protein without affecting its ability to promote DNA synthesis, causes no obvious phenotype. We estimate that, because of the increased stability of Cdc6ΔNTp and its expression from the strong GAL1‐10 promoter, this protein accumulates to levels during G2 that are at least 1000‐fold higher than normal (data not shown). Despite this vast overexpression, re‐replication does not occur and pre‐RCs are not formed during G2. Thus it is reasonable to ask ‘why is Cdc6p unstable?’. We suggest two possibilities.

First, we note that the results described here are in sharp contrast to those obtained in fission yeast, where more modest overexpression of cdc18+ is sufficient to induce rampant re‐replication (Nishitani and Nurse, 1995; Muzi‐Falconi et al., 1996). Thus one possibility is that in budding yeast redundant mechanisms may act to ensure once per cell cycle regulation of DNA replication, while in fission yeast regulation of cdc18+ proteolysis may be the primary mechanism by which re‐replication is blocked. Since inhibition of Clb–Cdc28 kinase activity is sufficient to promote pre‐RC formation in G2 (Dahmann et al., 1995), any additional mechanisms blocking pre‐RC formation during G2 must also be blocked by Clb–Cdc28 kinase. Candidates for such additional targets might include the Mcm proteins, which are components of Cdc6p‐dependent pre‐replicative chromatin (Coleman et al., 1996; Donovan et al., 1997). In budding yeast the Mcm proteins are only present in the nucleus during G1 (Hennessy et al., 1990; Yan et al., 1993; Dalton and Whitbread, 1995; Whitbread and Dalton, 1995), while in fission yeast at least one of the Mcm proteins (cdc21+) is constitutively nuclear (Maiorano et al., 1996). Thus perhaps nuclear localization of Mcm proteins in budding yeast is inhibited by Clb kinase, as is the case for nuclear localization of the Swi5p transcription factor (Moll et al., 1991), and this pathway may not exist in fission yeast.

A second possible reason for Cdc6p degradation might be that the N‐terminus of Cdc6p has dual functions: it may target Cdc6p for degradation and it may have an additional inhibitory affect on cell cycle progression. Elsasser et al. (1996) have shown that this region of Cdc6p interacts very tightly with Clb–Cdc28 and can inhibit its kinase activity. Thus degradation of full‐length Cdc6p may be important to allow full activation of Clb–Cdc28 kinase. This might help to explain why overexpression of full‐length Cdc6p in certain strain backgrounds causes a G2/M delay (Bueno and Russell, 1992; Elsasser et al., 1996; data not shown). Why the N‐terminus of Cdc6p should interact so tightly with Clb–Cdc28 kinase is unclear since, as shown here, it is not required for blocking pre‐RC formation or for firing of replication origins.

Materials and methods

Strains and media

Strains (listed in Table I) were grown in YP containing either glucose (YPD), raffinose (YP‐Raf) or galactose (YP‐Gal) at 2% unless otherwise stated (Rose et al., 1989).

View this table:
Table 1. Strain list

Plasmid constructs

pRS303G is an integrating vector with the GAL1‐10 promoter. It was constructed by inserting an 813 bp fragment containing the GAL1‐10 promoter between the EcoRI and BamHI sites of the pRS303 polylinker (Sikorski and Hieter, 1989). pLD3 and pLD4 have full‐length CDC6 and truncated CDC6 (termed CDCΔNT, encoding amino acids 48–513) respectively subcloned into the BamHI site of pRS303G so that the 5′‐end of the gene is at the promoter‐proximal side. In pLD5 the BamHI–SpeI fragment from pLD3, containing the intact CDC6 gene, was inserted between the BamHI and SpeI sites of pRS303G, thus removing the BamHI site at the 3′‐end of the gene. The N‐terminal 139 nt fragment (corresponding to amino acids 1–47) was replaced with a double‐stranded oligonucleotide coding for the SV40 nuclear localization signal (oligonucleotide numbers 74507 and 74508 in Table II). pLD6 and 7 are based on pRS306 (Sikorski and Hieter, 1989). pLD6 contains the entire GAL1‐10–CDC6 sequence subcloned from pLD3 using ClaI and SacI, and pLD7 has the entire GAL1‐10–CDC6ΔNT fragment subcloned from pLD4 in a similar manner.

View this table:
Table 2. Oligonucleotide primers

Oligonucleotide primers

The oligonucleotide primers are listed in Table II.

To construct a LexA–Cdc6p fusion in pBTM116, CDC6 was generated by PCR amplification using primer 49628 for the 5′‐end of CDC6 and primer 49362, which anneals to a sequence 200 bp 3′ of the CDC6 stop codon. Cdc4p interactions with the N‐terminus of Cdc6p were investigated using PCR‐generated fragments cloned into the BamHI and SalI sites of pBTM116. The following pairs of primers were used (5′ and 3′ respectively); 49628 and 67393 for amino acids 1–47, 49628 and 67394 for amino acids 1–18, 67396 and 67393 for amino acids 17–47, 67396 and 70589 for amino acids 17–33, and 70588 and 67393 for amino acids 34–47.

The CDC6 gene in pLD3 was amplified by PCR from W303‐1a genomic DNA using the 5′ primer 49627 and the 3′ primer 49362. pLD4 contains a copy of CDC6ΔNT generated by PCR using the 5′ primer 67395 and the 3′ primer 49362.

Two‐hybrid experiments

A two‐hybrid screen was performed with pBTM116 containing a full‐length CDC6 gene. A yeast cDNA library fused to the Gal4 transcriptional activation domain (a gift from Steve Elledge) was transformed into strain L40 containing the LexA–Cdc6 plasmid. Transformants were first selected for histidine prototrophy and then assayed for induction of β‐galactosidase activity. Plasmids from positive clones were rescued into Escherichia coli and re‐transformed into L40 containing either the LexA–Cdc6 plasmid or LexA alone (pBTM116).

Initial experiments to define the Cdc4p‐interacting region of Cdc6p utilized three restriction digest fragments from CDC6 sub‐cloned into BTM116 in‐frame with lexA. The first fragment derived from a PstI digest of the BTM116/CDC6 plasmid followed by re‐ligation of the plasmid. The resulting plasmid contains the first 139 nt of the CDC6 coding sequence. The second clone contains the first 700 nt of the CDC6 coding sequence and was also derived from the BTM116/CDC6 plasmid. This plasmid was generated by a BglII/BamHI double digest and re‐ligation of the compatible cohesive ends. The third construct was made by removing the CDC6 gene from the BTM116 plasmid using HincII (nt 205 in CDC6) and SalI (in the polylinker downstream of CDC6) and subcloning this fragment into the SmaI and SalI sites of pBTM116.

Cell cycle experiments and footprinting

Cell cycle synchronization was performed as previously described (Diffley et al., 1994). α‐factor block and release or nocodazole block and release experiments entailed washing the blocked cells with 2×2 vol. fresh medium and returning them to fresh medium at the same cell density.

To analyse protein binding at the 2 μ origin, DNase footprinting was performed as described previously (Diffley et al., 1994; Santocanale and Diffley, 1996).

Cells extracts and immunoblotting

Samples of 10 ml (107 cells/ml) were harvested and resuspended in 200 μl ice‐cold lysis buffer (100 mM NaCl, 25 mM Tris, pH 7.6, 10% w/v glycerol, 0.1% w/v Tween) containing protease inhibitors (1 μg/ml pepstatin A, 10 mM benzamidine, 2 μg/ml leupeptin and 1 mM phenylmethylsulfonyl fluoride). An equal volume of acid‐washed glass beads was added and the samples were vortexed to achieve lysis. One half volume of 3× Laemmli SDS sample buffer was added to the lysates and they were boiled for 3 min. Samples were run on 10% SDS–polyacrylamide gels, immunoblotted and detected using ECL (Amersham) reagents as per the manufacturer's instructions. Detection of Cdc6p was performed using monoclonal antibody 9H85 (Donovan et al., 1997).

Estimation of protein degradation rates

Autoradiographs were scanned using a Molecular Dynamics densitometer and amounts of Cdc6p were determined using ImageQuant software. Plots of time against ln[Cdc6p] were used to estimate half‐lives using the formula t1/2 = −0.69/m, where m is the slope of the line.


We are grateful to Miguel Godinho Ferreira, Kim Nasmyth and Mike Tyers for yeast strains and Steve Elledge for the yeast cDNA library. Thanks are due also to Miguel Godinho Ferreira for help with the cell cycle experiment in Figure 2A. We thank Ian Goldsmith and colleagues for oligonucleotide synthesis and Nicola O'Reilly for synthesis of α‐factor. We also thank Tim Hunt for critical reading of this manuscript.