Interaction of cdc2 and rum1 regulates Start and S-phase in fission yeast

Progression through the eukaryotic cell cycle is controlled at two major points -in G₁ before the initiation of S-phase, and in G₂ before entry into mitosis. In yeast the G₁ control is called Start, and represents the point of commitment to the mitotic cell cycle, so that cells are only able to initiate alternative developmental programs such as conjugation and meiosis from pre-Start G₁ (Hartwell, 1974). In fission yeast the products of the cdc2, cdc10 and res1/sct1 genes are essential for progression past Start (Nurse and Bissett, 1981; Tanaka et al., 1992; Caligiuri and Beach, 1993). The cdc10 and res1/sct1 gene products form a transcriptional complex essential for the expression of genes required for S-phase, suggesting that the activation of such cell cycle regulated transcription is an important aspect of Start (Lowndes et al., 1992; Tanaka et al., 1992; Caligiuri and Beach, 1993). The role of the p34^cdc2 kinase at Start remains unclear, although it appears that the p85^cdc10 containing transcriptional complex is unable to bind DNA in the absence of cdc2 function (Reymond et al., 1993).

To understand how progression through G₁ is regulated, it is necessary to identify those gene functions that determine the timing of Start, in order to identify the rate limiting step. The rum1 gene product has recently been shown to play a critical role in determining the timing of Start in the fission yeast cell cycle (Moreno and Nurse, 1994). Whilst a wild-type cell is able to arrest or delay progression past Start, in conditions such as nutrient limitation where the cell must remain in G₁ until it has grown sufficiently to achieve the critical cell mass required to pass Start (Nurse and Thuriaux, 1977; Nasmyth et al., 1979), cells lacking the rum1 gene undergo Start immediately after completing mitosis (Moreno and Nurse, 1994). rum1 function is therefore required to maintain the cell in the pre-Start G₁ interval, and overexpression of the rum1 gene product resets a G₂ cell back to pre-Start G₁, suggesting that rum1 may define a cell as being before Start (Moreno and Nurse, 1994).

A model for rum1 function has been proposed, in which rum1 acts by maintaining p34^cdc2 in a pre-Start G₁ form, inhibiting its activity until the critical mass required for Start is reached, at which point rum1 inhibition is released, and the cell passes Start (Moreno and Nurse, 1994). It is not yet possible to test this model directly in vitro, since we still do not know in what form or complexes p34^cdc2 is active at Start in fission yeast, whilst the detectable p34^cdc2 kinase activity against the standard histone H1 substrate is very low in G₁ (Booher et al., 1989; Moreno et al., 1989; Creanor and Mitchison, 1994). However, it has recently been shown that bacterially produced rum1 protein does potently inhibit mitotic p34^cdc2 kinase activity in vitro, associated with the B-type cyclin p56^cdc13 (J. Correa and P. Nurse, unpublished). Here we describe a physiological approach to studying cdc2 Start activity in vivo, and provide genetic evidence suggesting that rum1 acts as an inhibitor of p34^cdc2 at Start, and is necessary to maintain p34^cdc2 in a pre-Start G₁ form.

Both cdc2 and rum1 have also been implicated in the controls restricting S-phase to one round per cell cycle, since subjection of cdc2^ts mutants to a brief heat shock treatment causes G₂ cells to undergo an extra round of S-phase in the absence of mitosis, whilst overexpression of rum1 causes multiple rounds of S-phase without mitosis (Broek et al., 1991; Moreno and Nurse, 1994). In both cases rereplication requires passage of Start, suggesting that regulation of Start may form the basis of the dependency of S-phase upon the completion of mitosis. We provide genetic data showing that rum1 and cdc2 interact in this aspect of cell cycle control, and describe new dominant cdc2 mutants that cause overreplication when overexpressed in a wild-type cell, resembling overexpression of rum1.

The following strains were used: ade6-704 ura4-D18 leu1-32 h⁻, cdc2-33 h⁻, cdc2-33 rum1Δ h⁻, cdc2-M26h⁻, cdc2-M26 rum1Δ h⁻, cdc2-56h⁻, cdc2-56 rum1Δ h⁻, cdc10-129 ade6-704 h⁻, rum1Δ::ura4⁺ ura4D-18 leu1-32 ade6-M210 h⁺. Basic fission yeast methods were as described by Moreno et al. (1991).

Integrants were obtained in the wild-type strain ade6-704 ura4-D18 leu1-32 h−, using the ade6-704 / sup3-5 system (Hofer et al., 1979; Carr et al., 1989) and the lithium acetate transformation method (Okazaki et al., 1990). Southern blotting confirmed that integration occurred at some site other than the cdc2 locus, allowing integrants to be crossed to cdc2^ts alleles, as described below (see ‘N152/I172 induced overreplication’). Crosses involving integrants of nmt1 constructs were performed on malt extract plates containing 5 μg/ml thiamine, to repress expression from nmt1. Since deletion of rum1 causes sterility, double mutants involving rum1Δ were made by crossing rum1Δ::ura4⁺ura4D-18 leu1-32 ade6-M210 h⁺ transformed with pREP3X rum1⁺, so that rum1 is expressed from the plasmid, and the double mutants were subsequently checked to ensure that the plasmid had been lost.

All experiments in liquid culture were carried out in minimal medium, starting with a cell density of 2-8×10⁶ cells/ml, corresponding to mid-exponential phase growth. Temperature shift experiments were carried out using a water bath at 36.5°C, and the temperature was checked carefully, since the G₁ arrest of cdc2^ts mutants is weaker at lower temperatures.

To induce expression from the nmt1 promoter, cells were grown in minimal medium containing 5 μg/ml thiamine to mid-exponential phase, then spun down and washed three times with water, before resuspending in fresh medium lacking thiamine at a density calculated to produce 4×10⁶ cells/ml at the time of peak expression from the nmt1 promoter.

To measure cell number, cell were fixed in 0.9% saline/3.7% formaldehyde and sonicated before analysis with a ZM Coulter counter (Coulter Electronics Ltd).

cdc2-N152 and cdc2-I172 were originally analysed using pREP41, a multicopy vector in which expression is driven by the medium strength version of the nmt1 promoter (Basi et al., 1993). To obtain integrants, the 0.9 kb cdc2 cDNA was cloned as an NdeI-BamHI fragment into the integrative plasmids pRIP45 and pREP5, which are identical except that the former contains the medium strength nmt1 promoter, and the latter wild-type nmt1. pRIP45 and pREP5 were made by eliminating the 2.2 kb LEU2 containing HindIII fragment from pREP41 and pREP1, respectively (Maundrell, 1993), and then inserting a 0.5 kb PstI fragment containing the nonsense suppressor sup3-5, which suppresses the ade6-704 mutation and so can be used to isolate integrants as described above. pREP5 was made by Jacqueline Hayles, and pRIP45 by K.L.

About 10⁷ cells were spun down, washed once with water, then fixed in 70% ethanol and processed for flow cytometry or DAPI staining, as detailed previously (Sazer and Sherwood, 1990; Moreno et al., 1991). A Becton-Dickinson FACScan was used for flow cytometry. Overreplicating cells expressing cdc2-I172 are osmotically sensitive, and so were washed with 1 M sorbitol instead of water. DNA contents were assigned by reference to 1C, 2C and 4C controls, prepared using nitrogen starved haploid cells, exponentially growing haploid cells and exponentially growing diploid cells, respectively.

For anti-α-tubulin immunofluorescence, cells were fixed with paraformaldehyde and glutaraldehyde and stained with TAT1 primary antibody (Woods et al., 1989), followed by Goat α-mouse Texas Red secondary antibody (Jackson Immunoresearch Laboratories), essentially as described (Moreno et al., 1991). Cells were viewed with a Zeiss Axioskop microscope.

I172 induced rereplication was difficult to analyse genetically, since although p34cdc2-I172 overexpression causes overreplication at 32°C, rereplication is largely inhibited at the higher temperatures needed to arrest ts mutants, even in the presence of 1 M sorbitol. In addition, crossing the pREP5 cdc2-I172 integrant to cell cycle mutants such as cdc25 and cdc2 produced strains that were too sick to work with even at the permissive temperature, whilst the need to include sorbitol in the medium to observe I172 induced rereplication further complicated such experiments. For these reasons we concentrated on N152.

N152 is more amenable to study, since lower levels of expression produce rereplication, and crossing the pRIP45cdc2-N152 integrant to cdc10-129 produced a double mutant that grows well at 25°C and so could be used to show N152 induced rereplication requires cdc10 function. However, combining the pRIP45cdc2-N152 integrant with cdc2^ts or cdc25^ts mutants produced strains that were very sick even at 25°C, and could not be analysed.

Cell extracts were made as described using HB15 buffer (Moreno et al., 1989, 1991) and spun at 4°C in a microfuge for 15 minutes, before assaying the protein concentration of the supernatant using the BCA assay kit (Pierce), and adjusting the sample volumes to give a uniform concentration in each case. p34^cdc2 or p56^cdc13 were immunoprecipitated from 1.3 mg of protein extract, using affinity purified anti-C-terminal antibody C2, or anti-p56^cdc13 antibody SP4, respectively, at 0°C for 1.5 hours. Protein A-Sepharose was then added for 30 minutes at 4°C and the immuneprecipitates washed 5 times with HB15 buffer, before splitting in two. One half was heated to 99°C for 2 minutes in 1× SDS-PAGE sample buffer and used for immunoblotting with ‘Immobilon-P’ transfer membrane (Millipore), according to the manufacturer’s instructions. The other half was resuspended in 10 μl of HB15 containing 200 μM ATP, 1 mg/ml histone H1 (Boehringer Mannheim) and 40 μCi/ml [γ-³²P] ATP, and incubated at 32°C for 20 minutes. The reaction was stopped with 10 μl of 2× sample buffer, denatured at 99°C for 2 minutes and the samples run on an 11% SDS-PAGE gel. Phosphorylated histone H1 was detected by autoradiography.

If rum1 functions specifically at Start as an inhibitor of p34^cdc2, cells lacking the rum1 gene should have increased p34^cdc2 Start activity, whilst p34^cdc2 activity in G₂ should remain unchanged. Since we do not yet know in what complexes p34^cdc2 functions at Start in fission yeast, and in vitro can only detect an increase in p34^cdc2 kinase activity once cells have entered G₂ (Moreno et al., 1989; Creanor and Mitchison, 1994), we developed an in vivo approach to test this idea. When a cdc2^ts mutant is placed at the restrictive temperature, cells become arrested in G₁ and G₂, and then either remain arrested upon continued incubation, or leak past one or both of the two block points. The latter depends on the particular allele chosen, and probably reflects the amount of residual activity at the restrictive temperature. If cells lacking rum1 have increased p34^cdc2 Start activity, shifting a cdc2^tsrum1Δ double mutant to the restrictive temperature should result in fewer cells becoming arrested in G₁, and a more transient arrest for those cells that do arrest before Start. In contrast, the ability of such cdc2^ts mutants to arrest in G₂ should remain unchanged. cdc2-33 is typical of many cdc2^ts mutants in causing a very tight G₂ arrest, such that cells never enter mitosis upon prolongued incubation at 36.5°C, whilst the G₁ arrest is transient, and cells undergo S-phase after several hours at the restrictive temperature. As shown in Fig. 1A, incubation of cdc2-33 rum1Δ at the restrictive temperature causes fewer cells to become arrested before S-phase, and those cells that do arrest initially leak past the G₁ block point a generation time earlier than the cdc2-33 rum1⁺ control. Cell number increase is identical for both strains, and has ceased after 1 hour at 36.5°C, whilst DAPI staining shows that both strains remain arrested and elongate with a single interphase nucleus, demonstrating that deletion of rum1 has no effect on G₂ arrest (not shown).

Fig. 1B shows a quantification of the percentage of G₁ arrested cells, based on the data in Fig. 1A, together with data from equivalent experiments with other alleles of cdc2. cdc2-M26 causes an extremely tight arrest in G₁ and G₂ at 36.5°C, and the arrested cells never leak past Start or undergo S-phase. However, deletion of rum1 in this strain again causes fewer cells to arrest in G₁, and those cells that do arrest leak through and undergo S-phase by 4 hours (Fig. 1B, left hand panel). We also examined cdc2-56, a rather leaky cdc2^ts mutant, where even the G₂ block is weak, and arrest at Start is very transient.

Deletion of rum1 dramatically reduces the ability of this strain to arrest in G₁ (Fig. 1B, right hand panel). These results show that deletion of rum1 specifically affects the ability of cdc2^ts mutants to become arrested at Start, without affecting the G₂ arrest of such mutants, suggesting that in the absence of rum1 p34^cdc2 has increased Start activity. The data indicate a close interaction between cdc2 and rum1, and are consistent with a function for rum1 as an inhibitor of p34^cdc2 at Start.

Exposure of cdc2^ts or cdc13^ts mutants to a transient heat shock treatment causes resetting of G₂ cells back to pre-Start G₁, resulting in a single extra round of DNA replication in the absence of mitosis upon return to permissive conditions (Broek et al., 1991; Hayles et al., 1994). This suggests that the cdc2/cdc13 complex defines a cell as being in G₂, and so is important for the regulation of S-phase within the cell cycle (Hayles et al., 1994). Deletion of the cdc13 gene causes multiple rounds of S-phase in the absence of mitosis, and overexpression of rum1 in G₂ cells has the same effect (Hayles et al., 1994; Moreno and Nurse, 1994). It is possible that rum1 may inhibit cdc2/cdc13 complex formation, potentially explaining why pre-Start G₁ cells have very low levels of this complex (Hayles and Nurse, 1995). If rum1 maintains p34^cdc2 in a pre-Start G₁ form, preventing the formation of an active cdc2/cdc13 complex, deletion of the rum1 gene might be expected to reduce the ability of cdc2^ts mutants to undergo rereplication when exposed to a transient heat shock.

As shown in Fig. 2, deletion of rum1 completely abolishes the diploidization of cdc2^ts mutants upon heat shock, and the rum1Δ cdc2^ts double mutants remain viable haploids at the end of the experiment. The results again suggest a close interaction of rum1 and cdc2 in the regulation of Start and S-phase within the cell cycle, and are consistent with the proposed model, where rum1 is not simply a G₁ inhibitor of p34^cdc2, but is also necessary to stabilise a pre-Start form of the kinase (Moreno and Nurse, 1994). Possible explanations for these contrasting aspects of rum1 function are discussed in more detail below (see Discussion).

A role for p34^cdc2 in the controls regulating S-phase within the cell cycle, and thus in determining cell cycle ‘memory’, was inferred from the diploidization of cdc2^ts mutants after heat shock. Given the unphysiological nature of this treatment, together with the fact that in the absence of heat shock such cdc2 mutants cause arrest in G₁ or G₂ at the restrictive temperature, we have searched for new cdc2 mutants specifically defective in the regulation of S-phase, to facilitate study of this aspect of cdc2 function. Our approach was to look for dominant mutants causing overreplication when overexpressed in a wild-type cell, to levels at which the wild-type protein has no effect on cell cycle progression. One such mutant, cdc2-I172, was found in a screen of dominant lethal cdc2 mutants produced by chemical mutagenesis of a plasmid expressing the cdc2 cDNA from a mutated version of the regulatable nmt1 promoter (Labib et al., 1995). A second, cdc2-N152, had previously been shown to cause a dominant lethal phenotype when expressed in a wild-type strain either from cdc2’s own promoter, or from the thiamine repressible nmt1 promoter, but even under repressed conditions was too lethal to allow analysis of the mutant phenotype (MacNeill and Nurse, 1993). We expressed N152 from a mutated version of the nmt1 promoter (Basi et al., 1993), and found that cells grow well in repressed conditions, but undergo rereplication upon induction of p34_cdc2-N152.

As shown in Fig. 3, overexpression of I172 or N152 in a wild-type strain prevents cell division and causes a large increase in DNA content, from 2C typical of a normal G₂ cell up to 16C, resulting in highly elongated cdc⁻ cells with very large nuclei. The phenotype resembles overexpression of rum1, though cells are unable to recover from expression of N152/172, presumably because p34^cdc2 has essential functions in other parts of the cell cycle which are inhibited by these two mutants. The two mutants produce a very similiar phenotype, although discrete rounds of DNA synthesis are more apparent for I172 than for N152, whilst N152 causes rereplication at lower levels of expression than I172.

No signs of chromosome condensation are seen in arrested cells during a timecourse of induction, whilst anti-α-tubulin immunofluorescence shows that rereplicating cells have interphase microtubules, indicating that repeated rounds of S-phase occur in the absence of mitosis (Fig. 4A, 200 cells were examined after 0 hours, 16 hours, 18 hours and 20 hours of N152 induction). This is further shown by the fact that N152 expression results in loss of endogenous mitotic p34^cdc2/p56^cdc13 kinase activity in overreplicating cells (Fig. 4B). N152 has proven much more amenable to further analysis (for technical reasons detailed in Materials and Methods), and the following analysis concentrates on this mutant.

Consideration of the mutated residues corresponding to N152 and I172 suggests that they are dominant negative mutants, inactive forms of p34^cdc2 causing rereplication when overexpressed by titrating some regulator of the wild-type protein. cdc2-I172 represents mutation of Thr172 of p34^cdc2 to Ile172. Thr172 is conserved in domain VIII of almost all Ser/Thr specific protein kinases, incuding all p34^cdc2 homologues in other organisms, but is absent from protein tyrosine kinases, and is likely to have a role in accomodating Ser/Thr substrate sites for catalysis (Hanks et al., 1988). This is indicated by the fact that Thr172’s side chain comes very close to that of the catalytic base in a proposed model for p34^cdc2 structure, based upon the crystal structure of cyclic AMP-dependent protein kinase (Knighton et al., 1991; Marcote et al., 1993; Endicott et al., 1994). cdc2-N152 represents mutation of Asp152 of p34^cdc2 to Asn152. Asp152 is conserved in domain VII of all protein kinases, within the central catalytic domain, and the equivalent residue in cyclic AMP-dependent protein kinase is involved in binding MgATP and is thought to play a direct role in catalysis (Hanks et al., 1988; Knighton et al., 1991). Both N152 and I172 are therefore very likely to represent inactive forms of p34^cdc2, and Fig. 4B shows that indeed p34^cdc2-N152 lacks any detectable H1 kinase activity in vitro.

Rereplication induced by N152 overexpression requires cdc10 function, indicating that overreplication involves resetting of G₂ cells back to pre-Start G₁ (Fig. 5). Overexpression of the inactive p34^cdc2-N152 protein therefore appears to titrate some regulator of wild-type p34^cdc2+ in G₂, causing cells to pass Start once again and reenter S-phase, without undergoing mitosis. Since rum1 is required for rereplication induced by heat shock of cdc2^ts mutants, we investigated the effect of overexpressing N152 or I172 in a cell lacking the rum1 gene. As shown in Fig. 6, rum1 deletion greatly reduces overreplication produced by N152 overexpression, and has a more modest effect on I172 induced rereplication. In both cases, cell cycle arrest still occurs as in the presence of rum1, but the arrested cells are unable to rereplicate to the same degree, again suggesting that rum1 may be required to maintain p34^cdc2 in a pre-Start G₁ form. Whilst deletion of rum1 completely suppresses the diploidization of cdc2^ts mutants, where transient heat shock provides only one opportunity to rereplicate, it is perhaps not surprising that rum1 deletion does not totally suppress the rereplication caused by N152/I172, since the mutants are continually expressed and so provide a constant signal for rereplication.

Overreplication caused by N152/I172 overexpression is completely suppressed by co-overexpression of wild-type p34^cdc2+ from the same nmt1 promoter, restoring normal cell growth and division, confirming that the mutants produce rereplication by specifically interfering with p34^cdc2 function, rather than by affecting a control in which the wild-type protein plays no part. We also tried to suppress N152 by co-overexpressing other genes involved in cell cycle control, and found that whilst multiple copies of the suc1 gene, or co-induction of suc1 or cdc25 from the nmt1 promoter, have no effect on overreplication, cells with multiple copies of the cdc13 gene are still able to enter mitosis and cell division 24 hours after inducing N152 expression. This suppression could simply mean that N152 cannot prevent mitotic entry in the presence of increased levels of p56^cdc13, or could also imply that p56^cdc13 specifically antagonizes resetting pre-Start and overreplication. We therefore used a lower level of cdc13 expression, from the same mutated version of the nmt1 promoter used to express N152, and found that cells inducing both N152 and p56^cdc13 still undergo cell cycle arrest upon induction, but overreplication is dramatically reduced (Fig. 7). This shows that p56^cdc13 can specifically antagonize rereplication induced by N152, and it is possible that titration of p56^cdc13 by overexpression of N152/I172 may in part explain how these mutants cause rereplication, since deletion of the cdc13 gene also leads to multiple rounds of S-phase in the absence of mitosis (Hayles et al., 1994).

In both fission yeast and budding yeast Start is known to involve the activation of a transcriptional program required for the initiation of S-phase, and in vertebrate cells transcription is also needed to pass the restriction point (Adolph et al., 1993). Both yeasts also require p34^cdc2 activity at Start, although the relevant substrates remain unknown, whilst vertebrate cells require related CDKs (cyclin dependent kinases) for G₁ progression (Sherr, 1993, provides a review).

Whilst budding yeast G₁ cyclins that act in association with p34^CDC28 are well characterised, no fission yeast cyclins have yet been shown to have a role in the G₁ phase of cycling cells. Apart from the mitotic cyclin cdc13, other cyclins have been identified such as puc1 and cig1, but there is no evidence that these proteins act in the cell cycle at Start (Forsburg and Nurse, 1991; Bueno et al., 1991; Forsburg and Nurse, 1994, erratum in Cell 73 no. 5). The cig2 gene is periodically expressed at the G₁/S boundary, but a role for this B-type cyclin in G₁ progression has yet to be demonstrated (Bueno and Russell, 1993; Connolly and Beach, 1994; Obara-Ishihara and Okayama, 1994).

rum1 encodes a protein with a critical role in determining the cell cycle timing of Start that is proposed to maintain p34^cdc2 in a pre-Start G₁ form, inhibiting its action until the cell has achieved the critical size required for progression past Start (Moreno and Nurse, 1994). Given the poorly characterized nature of fission yeast p34^cdc2 Start activity, we used a physiological approach to show that in the absence of rum1, cdc2^ts mutants are less able to arrest in G₁ at the restrictive temperature, implying that p34^cdc2 has increased Start activity in these cells, consistent with rum1 being an inhibitor of a pre-Start G₁ form of p34^cdc2. Recent work has shown that rum1 is indeed capable of inhibiting p34^cdc2 activity, since bacterially produced rum1 protein inhibits mitotic p34^cdc2/p56^cdc13 H1 kinase activity in vitro (J. Correa and P. Nurse, unpublished). This may explain why rum1 is also required to prevent entry into mitosis from pre-Start G₁ (Moreno and Nurse, 1994). It is therefore possible that our genetic data may reflect direct action of rum1 on p34^cdc2 complexes in G₁, although the nature of such complexes remains to be determined.

We also describe new dominant cdc2 mutants that cause overreplication when overexpressed, resetting a G₂ cell back before Start, and confirming the role of p34^cdc2 in the regulation of S-phase within the cell cycle. We suggest that cdc2 and rum1 interact in such controls, since deletion of rum1 abolishes the diploidization of cdc2^ts mutants and reduces the ability of p34^cdc2-N152 and p34^cdc2-I172 to cause overreplication when overexpressed in a wild-type cell. Our results indicate that rum1 is necessary to stabilize p34^cdc2 in a pre-Start G₁ form, as opposed to forming the G₂ specific complex with p56^cdc13. It is possible that N152/I172 cause rereplication by efficient titration of p56^cdc13, since co-overexpression of cdc13 suppresses N152 induced rereplication, whilst it has recently been shown that either deletion of cdc13 or heat shock of cdc13^ts mutants produces overreplication (Hayles et al., 1994). Other cdc2 dominant lethals have previously been described, one of which has been shown to be suppressed by cdc13, yet these mutants only cause G₂ arrest on overexpression (Fleig and Nurse, 1991; Fleig et al., 1992; MacNeill and Nurse, 1993). Either N152/I172 are particularly efficient at titrating p56^cdc13, or they titrate an additional factor that regulates p34^cdc2 activity. It is of interest that in the predicted three-dimensional structure of p34^cdc2, the side chains of both Asp152 and Thr172 come very close to the catalytic base (Endicott et al., 1994), so that mutation to bulky residues such as Asn152 and Ile172 is likely to interfere directly with catalysis, and N152 and I172 probably act in a similiar manner. If N152/I172 cause rereplication by titrating p56^cdc13, allowing resetting of p34^cdc2+ back to a pre-Start G₁ form, then it is interesting that rum1 is still required for efficient overreplication in these conditions.

Overexpression of p34^CDC28-N154 in budding yeast, or of p34^CDC2Hs-N146 in human cells, corresponding to p34^cdc2-N152 in fission yeast, has not been reported to cause rereplication in these systems (Mendenhall et al., 1988; Van den Heuvel and Harlow, 1993). This may reflect differences in p34^cdc2 regulators in these organisms. For example, budding yeast B-type cyclins are required after Start for the onset of S-phase as well as for mitosis (Schwob et al., 1994). Titration of these proteins by a dominant negative p34^cdc2 protein would therefore prevent overreplication. Overexpression of p34^cdc2-N152 in budding yeast causes cell cycle arrest with multiple elongated buds, equivalent to deletion of CLBs1-6, and overreplication does not occur (K.L., unpublished data).

Our data supports the proposed model for rum1 function, in which rum1 maintains p34^cdc2 in a pre-Start G₁ form, inhibiting its activity until the critical mass needed for Start is reached, and rum1 inhibition is released. In this case, rum1 action at Start may be analogous to an archer’s hand restraining an arrow in a bow -just as the archer’s hand inhibits progression of the arrow, but also acts to stably maintain it in a potentially active form, so too rum1 may inhibit p34^cdc2 at Start, but also be required to stably maintain the G₁ form in conditions where Start progression is delayed.

In order to directly test this model we need to know the nature of fission yeast cdc2/G₁ cyclin complexes. Until these are characterised various scenarios can be imagined for the role of rum1 in G₁ progression. rum1 may directly inhibit cdc2/G₁ cyclin complexes until the critical size required for Start is reached. Alternatively rum1 could have a more indirect role, preventing formation of an active cdc2/cdc13 complex in pre-Start G₁ cells and thereby allowing cdc2/G₁ cyclin complexes to form instead. These areas are currently under investigation. A variety of inhibitors of cyclin dependent kinases have now been described in budding yeast and vertebrate cells (see reviews by Hunter, 1993; Nasmyth and Hunt, 1993; Pines, 1994). These proteins are required to inhibit cell cycle progression in response to diverse signals such as DNA damage, contact inhibition, lack of nutrients, mating pheromones etc., and in mammalian cells their loss may be an important factor in the generation of tumours. p21 is one such inhibitor which appears to mediate p53 induced G₁ arrest in response to DNA damage (El-Deiry et al., 1993; Dulic et al., 1994), whilst it has been reported that the p16 inhibitor is mutated in many human cancers (Lamb et al., 1994; Nobori et al., 1994). In addition to inhibiting CDK/cyclin complexes, p21 also binds to and inhibits the DNA replication protein PCNA, showing that these small inhibitory proteins may have more than one function (Waga et al., 1994). It will be of interest to see how the functions of such proteins relate to rum1, and whether any is capable of complementing deletion of the rum1 gene in fission yeast.

There is some evidence that p34^CDC2 may also be involved in determining the dependency of S-phase upon mitosis in mammalian cells (Yoshida et al., 1990; Usui et al., 1991), just as p34^cdc2 acts in fission yeast to determine cell cycle memory, and so it is possible that a rum1 like CDK-inhibitor may interact in this control in higher eukaryotic cell cycles.

We thank the members of our groups for helpful discussions, and are particularly grateful to Rachel Craven for much help in early stages of this work. We thank Kinsey Maundrell and Gabrieli Basi for generously providing us with mutated versions of the nmt1 promoter prior to their publication, Stuart MacNeill for the N152 mutant, Ken Johnson for photographic assistance, Kevin Crawford for confirming the sequence of N152 and I172, Alberto Orfao and colleagues for access to their FACScan machine, Jane Endicott for advice and ideas based upon her model for p34^cdc2 3-D structure and Jacky Hayles for pREP5 and SP4 antibody. This work was funded by the Science and Education Research Council, The Imperial Cancer Research Fund, The Worshipful Company of Bakers, the Dirección General de Investigación científica y Técnica, and the European Molecular Biology Organization, from whom K.L. now receives a post-doctoral fellowship.