The chk1 pathway is required to prevent mitosis following cell-cycle arrest at ‘start’

The chkl pathway is required to prevent mitosis following cell-cycle arrest at 'start' Antony M. Carr*, Mohammed Moudjout, Nicola J. Bentley* and lain M. Hagan t *MRC Cell Mutation Unit, Sussex University, Falmer, Sussex BN1 9RR, UK. tDepartment of Biochemistry and Molecular Biology, University of Manchester, Stopford Building, Oxford Road, Manchester M13 9PT, UK. Background: The G2-M-phase transition is controlled by cell-cycle checkpoint pathways which inhibit mitosis if previous events are incomplete or if the DNA is damaged. Genetic analyses in yeast have defined two related, but distinct, pathways which prevent mitosis - one which acts when S phase is inhibited, and one which acts when the DNA is damaged. In the fission yeast Schizosacclaromyces pombe, many of the gene products involved have been identified. Six 'radiation checkpoint' (rad) gene products are required for both the S-M and DNAdamage checkpoints, whereas Chkl, a putative protein kinase, is required only for the DNA-damage checkpoint and not for the S-M checkpoint following the inhibition of DNA synthesis. Results: We have genetically defined a third mitotic control checkpoint pathway in fission yeast which prevents

mitosis when passage through 'start' (the commitment point in G1) is compromized. In cycling cells arrested at start, mitosis is prevented by a Chkl-dependent pathway. In the absence of Chkl, G cells attempt an abortive mitosis with a IC DNA content without entering S phase. Similar results are seen in the absence of Rad17, a typical example of a rad gene product. Conclusions: Genetic dissection of checkpoints in logarithmically growing fission yeast has identified a pathway that couples mitosis to correct passage through start. This pathway is related to the DNA-structure checkpoints which ensure that mitosis is dependent on the completion of replication and the integrity of the DNA. We propose that all three mitotic control checkpoints monitor distinct DNA or protein structures at different stages in the cell cycle.

Current Biology 1995, 5:1179-1190

Background The coordination of cell-cycle progression is maintained by checkpoints that prevent passage through the cell cycle if key events are not completed [1-4]. There are several well documented dependency relationships in the cell cycle; the main examples are: mitosis is prevented until DNA replication is complete [5]; DNA synthesis cannot be initiated until completion of the previous mitosis [6]; mitosis does not progress if spindle formation is blocked [7-9]; and mitosis is prevented during G1 when cells are not committed to the mitotic cycle [10,11]. Checkpoints that control mitosis in response to the DNA status have been genetically defined in both budding and fission yeast [12-18]. Two distinct checkpoints have been defined: the S-phase-mitosis (S-M) checkpoint, which responds to changes in the replicative state of the DNA; and the DNA-damage (or 'radiation') checkpoint, which monitors the status of the DNA for both spontaneous and induced DNA damage, and for lesions in the newly replicated DNA (such as incomplete ligation). Although many proteins that act in these checkpoints have been identified, little is known of the biochemistry of the mechanisms that underlie the checkpoints, or of the pathways which transduce the signal from the lesion to the cell-cycle machinery. Recent

genetic evidence has potentially linked the S-M checkpoint to DNA polymerase E [19], strengthening the notion that the replication machinery is directly responsible for initiating the S-M checkpoint signal. In the fission yeast Schlizosaccharomyces pombe, the S-M and radiation checkpoints require many of the same gene products, suggesting that they are closely related in mechanism. The radiation checkpoint requires the products of at least seven non-essential genes - radl, rad3, rad9, rad17, rad26, hutsl and chkl - and also involves the essential and functionally redundant genes encoding 14-3-3 protein homologues, rad24 and rad25 [20]. Null mutations in the 'checkpoint rad' genes (radl, rad3, rad9, radl7, rad26 and blusl) have very similar phenotypes these mutants are defective in both the S-M and DNAdamage checkpoints [13,16]. In contrast, the Chkl protein and the 14-3-3 homologues are specific to the DNA-damage checkpoint, and are not required for the S-M checkpoint, which arrests mitosis when DNA synthesis is inhibited (either genetically or biochemically) by the absence of substrate nucleotide subunits [16,21]. The Chkl protein is a putative protein kinase which, when overexpressed, prevents cell-cycle progression [20]. Although the substrate(s) for the Chkl kinase has not been identified, and many of the components of the relevant signal-transduction pathway await identification,

Correspondence to: Antony M. Carr. E-mail address: [email protected]

© Current Biology 1995, Vol 5 No 10

1179

1180

Current Biology 1995, Vol 5 No 10 Chkl has been implicated in the regulation of p3 4 cdc2 activity [21]. It acts downstream of the checkpoint ad gene products and may be upstream of the rad24/rad25 function [20].

checkpoint pathways in fission yeast, and provides a conceptual framework for examining the origin of the checkpoint signals at different stages of the cell cycle.

In this report, we demonstrate that the cell-cycle arrest pathway mediated by Chkl is required to prevent mitosis in cells blocked in passage through 'start' - the point in G1 phase when cells commit to the mitotic cycle. This observation further subdivides the mitotic-control

Results cdc10 arrest is 'bypassed' in chkl and radl7 mutants The chk I gene has previously been shown to be required for mitotic arrest in response to DNA damage and

chkl prevents haploid mitosis Carr et al. _

.

..

.

.

.

Fig. 1. Cell-cycle progression in synchronous cdclO, cdclO chkl and cdclO radl7 single- and double-mutant cultures. In exponentially growing G2 cells from (a,d) cdc10-V50, (b,e) cdcl0V50 chkl-r27d and (c,f) cdcl0V50 radl7-d cultures synchronized using lactose gradients, the progress through the cell cycle was followed by (a-c) scoring for septation index at the permissive temperature (27 C) and (d-f) scoring for both septation index and microtubule distribution at the restrictive temperature (36 °C). At the permissive temperatures, all cultures divided synchronously for several divisions as expected. At the restrictive temperature, cdc10 cells underwent a single normal division and then arrested as mononucleate cells with interphase microtubules. In contrast, both cdcl0 chkl and cdclO rad17 cells continued into a second, catastrophic, mitosis without appreciable delay compared to the permissive-temperature control culture. Photomicrographs of the cells after staining for microtubules, SPBs and .fWNA

r ............ chh*n. in FI;r .

RESEARCH PAPER

cdc1( O-v50

80 (a)

cdclO-v50 chkl-r27d (b)

cdc10-v50 rad17-d (c)

60 40

270 C

20

(d)

CT

(e)

(f)

360C 25

0 100200 31jO 400 500 0 10 200300 40 500 Time (mins) -Septa -Spindlles

100200 300 400 500

)I A

unligated DNA, but not for the response to the presence of unreplicated DNA [16,21]. To identify further events that may be monitored by the checkpoint pathways which control mitosis, we tested the ability of the chkl1-r27d null mutant and of a representative of the checkpoint rad class of mutant (rad I 7-d) to prevent mitosis in a range of cdc mutant backgrounds. All of the cdc mutants used usually cause cell-cycle arrest with a single undivided nucleus at the restrictive temperature (Table 1). As expected, neither Chkl nor Radl7 function was required to prevent mitosis when activation of the p34cdc2-p56cdc13 mitotic kinase was prevented by temperature-sensitive cdc2, cdc13 or cdc25 mutations. Similarly, the cell-cycle arrest caused by the cdc5 and cdc28 mutations did not require Chkl or Rad17 functions (see also [22]). When bulk DNA synthesis was prevented by a cdc22 mutation, cell-cycle arrest was dependent on the presence of the radl 7+, but not of the chkl + gene product. We therefore confirm the previous observations that the S-M checkpoint is independent of Chkl, but dependent on the checkpoint rad gene products [16,21]. The rad17-d mutant has a profile of mitotic arrest defects in S and G2 phases similar to that seen in cdc2-3w mutant cells (Table 1; [23]). However, mitosis does not occur in a cdc2-3uw mutant background in cells arrested at start by temperature-sensitive cdclO mutants [5]. In sharp contrast, we found that Chkl and Radl7 functions were both required in order to prevent mitosis when the cell cycle was arrested prior to commitment at start. To confirm the cells undergo a true mitosis, it is important to establish the kinetics of mitotic and post-mitotic events. We therefore quantified, by fluorescence microscopy, spindle formation, spindle function, cytokinesis and the correct re-establishment of an interphase cytoskeleton following a temperature shift. Synchronous cultures of cdclO single mutants and cdclOcchkl and

cdcl0 rad17 double mutants were examined. The cdcl0 single-mutant cells maintained an interphase distribution of microtubules, contained a single spindle pole body (SPB) for at least 6 hours after cell-cycle arrest, and did not septate (Figs 1,2). This indicates that mitosis does not occur in cells arrested by the loss of Cdcl0 function. In contrast, spindle formation and the appearance of septa in cdclO chkl (Figs 1,3) and cdclO radl7 (Figs 1,4) double mutants followed similar kinetics whether Cdc10 function was lost or normal. Observation of nuclei by DAPI (4',6'-diamidino-2-phenylindole dihydrochloride) staining indicated that chromosome segregation was attempted but not completed correctly, and many cells exhibited a 'cut' phenotype with lumps of chromatin strung out along the spindle [24]. We also followed the appearance of post-anaphase arrays of cytoplasmic microtubules (data not shown). In all cases, the post-anaphase array of microtubules was established normally and the SPBs were inactivated. This indicates that all aspects of cell-cycle progression that normally follow commitment to mitosis (from spindle formation, through cytokinesis, to the correct establishment of the interphase cytoskeleton in daughter cells) are present. These results demonstrate normal entry into, and exit from, mitosis by chk I and radl7 cells, even in the absence of Cdcl0 function. Further examination of the photomicrographs revealed an interesting phenomena (Figs 3,4). Following the first abortive attempt at mitosis after loss of CdclO function, many cells attempted a further round of spindle formation within a very short time (30 minutes to 1 hour). The SPB staining at either end of the spindle (for example, see Fig. 3p-u) demonstrated that the microtubule staining was a genuine spindle rather than a short cytoplasmic microtubule bundle. These observations show that, following the first abortive mitosis and missegregation of the DNA, all checkpoint control regulating the

1181

1182

Current Biology 1995, Vol 5 No 10 timing of M-phase onset is lost in these cells. This situation is reminiscent of the multiple abortive mitoses that have been reported in mammalian cells when checkpoints are compromised with caffeine [25].

Cell-cycle arrest loss not due to cdclO block leakage In all the above experiments, we obtained identical results with two separate temperature-sensitive mutants, cdclO-V50 (Figs 1-4) and cdc10-129 (data not shown).

Fig. 2. Photomicrographs of synchronized cdclO-V50 cells at 36 °C. (a-c) Time zero. Initially, cells are in interphase, with typical interphase microtubules, a single nucleus and a single SPB. (d-f) At the peak of the first mitosis, normal spindle structures are present and the DNA segregates equally to either pole. The SPBs have separated and a single one is present at each end of the spindle. (g-i) Interphase (240 min). Following the first mitosis, cells remain in interphase with typical interphase arrays on microtubules and no evidence of either SPB separation, spindle formation or attempted chromatin segregation. (j-o) Interphase. (j-l), 380 min; (m-o), 440 min. Cells remain in this state, while elongating as mass continues to accumulate. This is a typical cdc morphology. (a,d,g,j,m) Superimposed images of TAT1 (tubulin; red) and Sad1 (SPB; bright dots) immunofluorescence with DAPI staining (DNA; blue) and phase contrast. (b,e,h,k,n) TAT1 and Sad1 immunofluorescence with DAPI staining. (c,f,i,l,o) Cartoon representation of morphology.

chkl prevents haploid mitosis Carr et al. However, it remains possible to interpret the data shown in Figures 1-4 as resulting from incomplete or 'leaky' cell-cycle arrest in these mutants. It could be argued that these cells are merely arrested transiently at start before 'leaking through' the block to S phase. Although the cdcl0 chkl cells at the restrictive temperature did not pass through S phase, as judged by fluorescence-activated cell sorting (FACS) analysis (data not shown), this technique is not sufficiently sensitive to detect entry into early S phase. Therefore, we have addressed this issue by examining the sensitivity of the abortive mitosis to inhibition by the S-phase checkpoint Genetic analysis (Table 1 and [4,16,21]) has demonstrated that the S-M checkpoint (which is independent of Chkl function) is activated when S phase is prevented by the absence of nucleotide precursor subunits and that the DNA-damage checkpoint that is dependent on Chkl can be activated during late S phase and G2 by lesions such as incompletely ligated DNA. The S-M checkpoint therefore defines a 'window' during S phase where mitosis can be prevented in both wild-type and chkl null mutant cells by inhibiting DNA synthesis. It is possible to use this window to test whether cdcl0 arrested cells have leaked through and entered into S phase. If they have, they will become subject to the S-M checkpoint when in the presence of hydroxyurea, and mitosis will be arrested. If they are not, they will not engage the S-M checkpoint (as they are not in S phase), and so hydroxyurea will not affect entry into abortive mitosis. We determined that both the chkl cdc10-129 and chkl cdclO-V50 double mutants execute abortive mitosis in the presence of 10 mM hydroxyurea (Fig. 5), showing that cdclO chkl double-mutant cells were executing mitosis before S phase was initiated - from G1 during cdcl0 arrest and not after leaking through into S phase. As hydroxyurea will arrest cell-cycle progression normally in synchronized chkl mutants when CdclO function is intact, this observation identifies three distinct circumstances which can activate the mitotic control checkpoints. Firstly, during G1 and before the onset of S phase, a checkpoint is required to prevent mitosis. This checkpoint is Chkl-dependent. Secondly, during S phase, a distinct Chkl-independent checkpoint is required to prevent mitosis. Thirdly, during the late stages of S phase and during G2, the Chkl-dependent DNA-damage checkpoint is required to prevent mitosis. Abortive mitosis is not an artifact of the synchronization procedure In order to eliminate the possibility that the procedure we have used to synchronize the cells causes artifactual loss of the G1-M checkpoint in chkl and radl7 null mutant cells, we performed two additional experiments. In the first, asynchronous cultures of cdcl0, cdclO chkl and cdcl0 rad17 cells were shifted to the restrictive temperature for CdclO function and cell-cycle progression was monitored by following the septation index (the data in Figs 1-4 demonstrate that this is a suitable measure

RESEARCH PAPER

of mitotic progression). The loss of CdclO function in a chkl + background affected mitosis after approximately 150 minutes. At this time, the cdcl0 single-mutant culture showed a decrease in septation index, which remained low for several hours. This trough of septation was not seen in cdc 10 chkl or cdc 10 rad 17 cultures (Fig. 5e). In the second experiment, the asynchronous shift experiments were repeated after incubating the cells in lactose (Fig. 5g,h) in order to eliminate the possibility that the brief exposure to lactose during synchronization affected cell-cycle kinetics. Cells behaved in an identical fashion whether or not they had been incubated in lactose prior to the temperature shift. Together, these results confirm that the bypass of the G1-M checkpoint seen in Figures 1-4 is not an artifact of the synchronization procedure. The G1-M checkpoint deficiency is not confined to cdc10 arrest In order to determine whether the loss of the G1-M checkpoint in Chkl-deficient cells is unique to cdclO mutant arrest, we sought an alternative way of arresting cells at start. A partner of CdclO in the transcription factor complex that initiates S phase is the product of the sctl+/resl+ gene [26,27]. We therefore tested whether sctl+ /resl+-defective cells arrested cell-cycle progression in the presence of chkl-r27d or radl7-d mutations. As the available mutation in the sctl/resl gene (a partial deletion) is already defective in start transit at the permissive temperature of 30 C, leading to aberrant cell size at division, it is not possible to generate physiologically normal synchronous cultures [26]. However, upon shift of an asynchronous culture to the restrictive temperature of 36 C, there is an increase in cell length at division in the res 1- mutant, which reflects a delay to mitotic execution arising from a leaky G1 arrest [26,27]. This delay was not seen in the presence of deficiencies of either Chkl or Radl7 (Fig. 6), where many cut and anucleate cells accumulated. Indeed, even at 30 °C (the permissive temperature), cell-length distribution of resl chklI and res1 rad17 double mutants, unlike the res 1 single mutant, resembled that of the wild-type population. This suggests that mitotic commitment was being limited by the pl07wecl-dependent cell-size control in the double mutants and was no longer being influenced by the Chkl- and Radl7-dependent checkpoints. As the chklr27d mutant retains the S-M checkpoint, it is unlikely that the Chkl-dependent delay seen in res 1 cells is due to a delay to S phase. The results seen with resl cultures are therefore fully consistent with the inability of radl7 and chkl strains to prevent mitosis occurring during pre-Sphase G1 arrest. ruml + and chkl + genes do not interact genetically We have demonstrated that Chkl function is required to prevent mitosis in cells which cannot progress through start. This has interesting parallels with the observation that the product of the rum ml + gene is required to establish the G1 phase of the cycle [10]. The deletion of the nnl gene causes mitosis to occur in the absence of

1183

1184

Current Biology 1995, Vol 5 No 10

chkl prevents haploid mitosis Carr et al.

RESEARCH PAPER

4 Fig. 3. (Facing page). Photomicrographs of synchronized cdclO-V50 chkl-r27d double-mutant cells at 36 C. (a-c) Time zero. Initially cells are in interphase with typical interphase microtubules, a single nucleus and a single SPB. (d-f) The first peak of mitosis. Normal spindle structures are present and the DNA segregates equally to either pole. The SPB has duplicated and a single SPB is present at each end of the spindle. (g-i) Interphase. Following the first mitosis, cells enter interphase with typical interphase arrays on microtubules and no evidence of either SPB separation, spindle formation or attempted chromatin segregation. (j-o) The second peak of mitosis. Cells then enter the second mitosis - the SPB separates, the interphase microtubules disappear and spindles are formed with an SPB at either end. The DNA, however, does not segregate properly and is often fragmented and stretched over the length of the spindle. The cells divide, with the septum often bisecting the nuclear material (data not shown). (p-u) The third attempt at mitosis. After cell division (cytokinesis), many cells rapidly form a spindle, suggesting that control of mitosis has completely broken down. (a,d,g,j,m,p,s)Superimposed images of TAT1 (tubulin; red) and Sad] (SPB; bright dots) immunofluorescence with DAPI staining (DNA; blue) and phase contrast. (b,e,h,k,n,q,t) TAT1 and Sad1 immunofluorescence with DAPI staining. (c,f,i,l,o,r,u) Cartoon representation of morphology.

Fig. 4. Photomicrographs of synchronized cdclO-V50 rad17-d double-mutant cells at 36 C. (a-c) Time zero. Initially, cells are in interphase with typical interphase microtubules, a single nucleus and a single SPB. (d-f) The first peak of mitosis. Normal spindle structures are present and the DNA segregates equally to either pole. The SPB has duplicated and a single one is present at each end of the spindle. Following the first mitosis, cells enter interphase with typical interphase arrays of microtubules and no evidence of SPB separation (data not shown). (g-i) The second peak of mitosis. Cells then enter the second mitosis. The two SPBs separate, the interphase microtubules disappear and spindles are formed with a pole body at either end. The DNA, however, does not segregate properly and is often fragmented and stretched over the length of the spindle. The cells divide, with the septum often bisecting the nuclear material (data not shown). (j-I) The third peak of attempted mitosis. After cell division (cytokinesis), many cells rapidly form a spindle, suggesting that control of mitosis has completely broken down. (a,d,g,j) Superimposed images of TAT1 (tubulin; red) and Sadl (SPB; bright dots) immunofluorescence with DAPI staining (DNA; blue) and phase contrast. (b,e,h,k) TAT1 and Sad1 immunofluorescence with DAPI staining. (c,f,i,l) Cartoon representation of morphology.

1185

1186

Current Biology 1995, Vol 5 No 10

rn

(a) 27 °C, - hydroxyurea

(b) 27 C, +hydroxyurea

(c) 36 C, - hydroxyurea

(d) 36 C, +hydroxyurea

30 C 0

a

20

V) o

10

I

30 0 C) o

20

V)

10

100

200

300

400

100

200

300

40

Time (min)

Time (min)

s5 . (0

(e)

40 C

0

ca

20.

6

10, -0

--C

cdc10

- -

cdcl10chk1

-

_-cdc0rad17

_-

cdc10chk, - hydroxyurea cdclOchkl, + hydroxyurea

(h)

(g) 40. C

0

1

30

)

I

a-

20. ... -U-cdclO chki, lac chkl, + +lac -

1n I-

.

O

100

.

.

300 200 Time (min)

.

400

._

10

cdclO rad17, - lac -U- cdc 0 rad17, +lac

200 300 Time (min)

CdclO function. Overexpression of rnm 1+ prevents mitosis, but does not prevent passage through the G1 and S phases of the cell cycle - continued passage through multiple discrete S phases are observed without intervening mitoses. The G2-arrest profile of rml- null cells following exposure to ionizing radiation is normal (Fig. 7), and rum 1- cells are not sensitive to radiation (data not shown).

400

500

Fig. 5. Inappropriate mitosis in chkl strains is not due to incomplete arrest of cdclO cultures or the synchronization method. (a-d) cdclO arrest of synchronous cultures of cdc10-129chkl-r27d double mutants in the presence of hydroxyurea. 'Cells were synchronized over lactose gradients and the culture split into four samples. (a,b) Two samples were incubated at 27 C either (a) without or (b) with 10 mM hydroxyurea, which was added at time 60 min immediately preceding the first mitosis. Cells were monitored for septation index. Hydroxyurea prevents the second attempt at mitosis (and thus septation) by blocking S phase and activating the DNA replication checkpoint, which remains intact in chkl mutant cells. This is consistent with previous work which demonstrates that, under identical conditions, chkl- cells can arrest mitosis for over 8 h following the addition of hydroxyurea [16]. (c,d) The remaining two samples were incubated at 36 C, one (c) without and one (d) with 10 mM hydroxyurea and scored in a similar manner. In this case, hydroxyurea did not prevent the second mitosis. This demonstrates that the second mitosis in chk-r27d cells which have lost CdclO0 function at the restrictive temperature is not due to cells leaking through the cdclO block, passing through S phase and attempting mitosis with replicated DNA, but represents an attempt at mitosis before the cells initiate DNA synthesis. (e) Asynchronous cultures of cdclO-V50, cdclO-V50chklr27d and cdclO-V50 rad17-d mutants at the restrictive temperature, monitored for septation index following a temperature shift from 27 C to 36 C at time zero. cdclO-VSO0 cells show a trough of septation starting at approximately 150 min which is not seen in the other cultures. This is consistent with no mitotic arrest in the double mutants following loss of Cdc10 function. (f) The addition of 10 mM hydroxyurea to an asynchronous culture of cdclO-V50 chkl-r27d cells at the same time as the temperature shift to the restrictive temperature (time zero) does not affect the kinetics of cell-cycle progression, confirming that these cells behave in an analogous manner to the synchronous cultures of cdclO129 chkl-r27d cells shown in (a-d). (g,h) Pre-incubation of either (g) cdcl0V50 chkl-r27dor (h) cdc10-V50 rad17-d in 20 % lactose for 15 min prior to temperature shift does not affect the kinetics of cell-cycle progression.

This rules out the possibility that Ruml acts as a specific inhibitor of mitotic p3 4 cdc2 activity after DNA damage. This is also supported by the results of overexpression experiments: the mitotic arrest caused by chkl overexpression [20] is not dependent on the presence of the ruml + gene, and the mitotic arrest caused by the overexpression of rum 1+ [10] is not dependent on the presence of the clk 1+ gene (data not shown; summarized in Table 1). Thus, Ruml is not activated by a Chkl-dependent

chkl prevents haploid mitosis Carr et al.

RESEARCH PAPER

Fig. 6. Mitotic execution in cells arrested at start when Chkl is missing is not specific to cdcl0 mutants. Asynchronous cultures of resl, resl radl7or resi chkl cells

were grown at 30 C prior to shift to 37 °C for 6 h. Cell lengths were measured using an adaptation of NIH image software (Improvision, Coventry, UK). The celllength distribution is plotted as length against frequency. Although the cell-size profiles of resl single-mutant cells at the permissive temperature of 30 °C are skewed, with many longer cells visible, the profiles of the resl chkl and resl rad17

double mutants resemble that of a wildtype population. In contrast to the single mutant, which clearly exhibits a cdc phenotype after shift to 37 °C (characterized by a much broader cell-size profile) the resl chkl and resl radl7 double mutants

show no change in their size profiles following a shift from 30 C to 37 C. Following the shift, cells were fixed and processed for immunofluorescence staining for microtubules and DNA as described in Materials and methods. Cut and anucleate cells accumulated in double-mutant populations (data not shown), indicating that abortive mitoses were occurring at a high frequency. The skewed variable cell-size distribution profile of the resl mutant is a typical profile for a leaky cdc mutant as division size is random.

pathway after DNA damage, or by the overexpression of Chkl. These experiments also rule out the possibility that the mitotic delay caused by the overexpression of rum i + is indirect, acting through the chk i +gene product. From these observations, we conclude that chk I and rum 1 show no evidence of genetic interaction, and that rum 1+ is not involved in the radiation checkpoint. The data of Moreno and Nurse [10], and the data we report here, indicate that both Ruml and Chkl are involved in preventing mitosis during the G1 period of the cell cycle. However, we find no evidence that they operate in the same checkpoint pathway. Our interpretation of this is as follows. Ruml is involved in establishing the G1 period of the cycle, but is not part of the checkpoint pathway which is engaged during this G1 period to prevent mitosis. When cells cannot exist in G1 - when nrum + is deleted - the Chkl-dependent G1-M checkpoint cannot be engaged, although it remains intact. Investigation of this hypothesis must await the creation of biochemical frameworks for the DNA-structure checkpoint pathways and the mechanism of Ruml action.

Discussion DNA-structure checkpoints have previously been shown to arrest mitosis after a block to the cell cycle imposed in either S phase or the G2 period (reviewed in [28]). They also arrest mitosis in response to DNA damage. In addition to preventing mitosis, DNA damage also prevents DNA synthesis. In Saccliaronlyces cerevisiae, the inhibition of DNA synthesis is dependent on some of the same

proteins that are involved in mitotic arrest [18,29]. The control of DNA synthesis following DNA damage has not been addressed in S. pombe, but it is clear from previous work that the effects of Chkl and Radl7 function on mitosis can be addressed independently of any effects that they may have at the G1-S border [16,20]. At the restrictive temperature, cdc10 mutant cells are blocked in progress through start - they arrest before the initiation of S phase with a 1C DNA content and do not attempt mitosis, but remain capable of mating. This phenotype defines 'start' in fission yeast, and cdcl0 mutant cells blocked at start have been reported to arrest mitosis normally when the S-M checkpoint is compromised by the cdc2-3w mutation [5,23]. We have demonstrated here that, in rapidly proliferating cells compromised in passage through start, the chkl gene product is required to prevent mitosis before DNA synthesis is initiated. Loss of this novel checkpoint causes cells that'are prevented from passing through start to enter a catastrophic mitosis with unreplicated DNA. As this catastrophic mitosis is not prevented by blocking DNA synthesis with hydroxyurea (which causes mitotic arrest in chkl cells with functional Cdc10), we are confident that our results are not due to cdcl0-arrested cells leaking through start arrest and passing into S phase. If such leak-through did occur, it would result in the onset of DNA synthesis and cells would become sensitive to hydroxyurea arrest, thus activating the Chkl-independent S-M checkpoint which would prevent mitosis. Furthermore, a hallmark of leak-through is cell-to-cell variation in a population. This is demonstrated by the

1187

1188

Current Biology 1995, Vol 5 No 10

growing cells which have not yet entered S phase. Other explanations cannot, of course, be precluded. For example, the phenotype of the chkl cdclO double mutant is similar to that of the cdc18 deletion mutant, where cells pass through start normally, but S phase'is not established and thus the S-M checkpoint cannot be engaged [30,31]. It is formally possible (but we consider it unlikely) that, in the absence of Chkl function, cdclO arrested cells can pass start, but do not activate Cdcl8 function and therefore execute mitosis.

.LI

0

E ._ It

v) v

0

50

100 150 Time (min)

200

._

E 0. 0.

IJ

0

--

50

100 150 Time (min)

200

OGy

-0-50 Gy --I-

1 0 Gy 250 Gy

Fig. 7. Radiation checkpoint measurements in ruml- cells. The duration of the radiation checkpoint-induced mitotic delay was measured in G2-synchronized ruml::ura4 and wild-type cells. Similar results were obtained for both strains, indicating that ruml null cells are not defective in the radiation checkpoint.

data in Figure 6 on the res l single mutant, where partial loss of Resl function leads to a somewhat arbitrary cell size at mitosis. This contrasts sharply with the regimented conformity of cell size and mitotic timing seen in the cdclOchkl double mutants executing mitosis before S phase. Arrested cdclO cells are in a state prior to the onset of hydroxyurea-sensitive DNA synthesis that is consistent with their arrest in G1 phase. We can see no evidence that leads us to redefine the previously defined arrest point of cdcl0 mutants, and we conclude that the most likely explanation for our observations is that Chkl function is required for a checkpoint that prevents mitosis in

It has recently been reported that chkl cdclO double mutants arrest mitosis normally when exiting from stationary phase [32]. This contrasts with our results, and demonstrates that different checkpoint controls operate in rapidly cycling cells than in cells re-entering the cell cycle from GO phase following nitrogen starvation. There are likely to be significant physiological differences between nitrogen-starved cells re-entering the cell cycle and rapidly proliferating G1 cells. For example, nitrogenstarved cells are known to condense their chromosomes, and it takes approximately 2-4 hours following a shift to rich media before interphase chromatin is re-established [33]. In view of this, it would be surprising if there were not differences in the pathways controlling cell-cycle progression between cycling cells and cells released from nitrogen starvation. A possible model for the checkpoints controlling mitosis in fission yeast The S-phase and DNA-damage checkpoints, which have been shown previously to be deficient in the checkpoint rad class of mutants of fission yeast, monitor the status and integrity of the DNA. It is not known if these pathways monitor the DNA directly, or through protein structures associated with the DNA. In the case of the S-M checkpoint, the available genetic evidence suggests that the replication complex is the source of the S-M checkpoint signal (reviewed in [28]), but this has not been demonstrated biochemically. During progression through the cell cycle, it is known that specific protein structures are associated with the DNA during the different periods (Fig. 8). During G1 phase, pre-replication protein complexes are associated with the origins of replication, and these undergo unknown modifications as cells pass through start [34] and assemble replication complexes. During DNA synthesis, the replication complexes are associated intimately with the DNA and are presumed to dissociate only when replication is complete. By this time, post-replication complexes are associated with the DNA at the origins of replication [34]. We suggest that the DNA-protein structures that are particular to the different phases of the cell cycle could provide targets for the different DNA-structure checkpoints to monitor as the sources of the various cell-cycle checkpoint signals which inhibit mitosis. In addition, the repair complexes, which may also contain a number of

RESEARCH PAPER

chkl prevents haploid mitosis Carr et al. Fig. 8. Model for mitotic entry checkpoints in S. pombe. Prior to the commitment to DNA synthesis, pre-initiation complexes could generate a signal via a which pathway Chkl-dependent inhibits mitosis. Following the formation of replication complexes, a Chkl-independent pathway can generate a signal that prevents mitosis from occurring. At any point in the cycle, the Chkl-dependent DNA-damage checkpoint can respond to DNA damage/repair structures to prevent mitosis. Work reported by Hayles and Nurse [32] suggests that nitrogen-starved GO cells prevent mitosis via a Chkl-independent pathway.

replication proteins, are transiently associated with the DNA during the act of repair. These may similarly provide suitable DNA-protein complexes which could act as the source of the radiation-checkpoint signal, from any point within the mitotic cycle. Interestingly, upon entering stationary phase, the structure of the protein complex associated with the origin undergoes a seemingly distinct modification [34]. This could account for the apparent requirement for different mitotic control checkpoint pathways in cells exiting from stationary phase [32] when compared to those in logarithmic growth.

Conclusions We have demonstrated that fission yeast cells arrested in passage through start are prevented from entering mitosis by a checkpoint pathway which requires the function of Chkl and Radl7. This pathway is closely related to, but distinct from, the previously described DNA-structure checkpoints which prevent mitosis following a block to DNA synthesis or when the DNA sustains damage. The identification of three distinct mitotic-control checkpoints in fission yeast provides a framework within which the biochemical nature of these checkpoint pathways can be investigated.

Materials and methods Genetic techniques A list of strains created and used in this study is given in Table 1. Double mutants were constructed using standard genetic techniques as described in Gutz et al. [35] and monitored for the cut phenotype (ascertained microscopically following DAPI and calcofluor staining) [13] and the rapid-death phenotype (ascertained by testing the ability of cells to recover from a brief temperature shock - 4 h at 36 °C for temperature-sensitive cdc mutants) [13]. The radl 7 mutant has a very similar phenotype to the other checkpoint rad mutants - radI, rad3, rad9,

rad26 and husl. Double mutants of all of these with cdc22, cdcl7 and cdclO were constructed and checked following a temperature shift. All displayed defects similar to the respective radl 7 double mutants, suggesting that radl7 can be considered as a representative of the checkpoint rad class of mutants for this analysis.

Radiation checkpoint measurement Cultures of synchronous cells were prepared on a 7.5-30 % lactose gradient. G2 cells were recovered from the top of the gradient, washed in supplemented yeast extract (YES) media [35], and inoculated into fresh YES media. Samples were subjected to either 0, 50, 100 or 250 Gy ionizing radiation using a Gammacell 1000 137 Cs source (12 Gy min -1) and incubated at 29 °C. Aliquots were removed at 15 min intervals and fixed in methanol for estimation of passage through mitosis by DAPI

and calcofluor staining [13]. Overexpression studies

The chk 1+ and rum 1+ genes were overexpressed from the nmt I promoter [36]. The cihkl + cDNA is cloned as an NdeI-BamHI fragment in the pREP1 vector [16]. The ruml+ construct (a gift from P. Nurse and B. Stern) was in pREP3 [10]. The nmtl promoter takes approximately 16 h to be fully expressed following a shift to thiamine-free media [36]. Cells were observed microscopically 24 h after the shift and tested for colony-forming ability (alongside 'empty vector' controls) under the same conditions. Cells did not form colonies when extensive elongation was observed.

Immunofluorescence Synchronous cultures for immunofluorescence were prepared from asynchronous cultures growing exponentially at 25 °C [37]. A two-step synchronization procedure was used: cells were initially synchronized on a 40 ml 10-40 % lactose gradient and were then immediately passed over a second (12 ml) 10-40 % lactose gradient in order to achieve optimal uniformity of the cell size. Cells were then washed, inoculated into fresh medium, split into two and incubated at either 25 °C or 36 °C. Cell plate indices were monitored using calcofluor staining and scoring only those cells with a equatorial brightly staining bar as having septa. Cells (107) were harvested every 20 min from the 36 °C culture and prepared for glutaraldehyde/formaldehyde immunofluorescence according to Hagan and Hyams [38]. Samples were stained using TAT1 anti-tubu-

1189

1190

Current Biology 1995, Vol 5 No 10 lin [39] and rabbit polyclonal anti-Sadl [40] antibodies with CY3 conjugated anti-mouse ackson Lab) and FITC anti-rabbit (Cappel) antibodies at appropriate dilutions. Cells were scored for the presence of mitotic spindles, as judged by a strong tubulin staining bar between two Sadl dots, and for the post-anaphase array [38]. Color images were produced by image capture using a Hamamatsu SIT Camera and C2400 processor. Immunofluorescence of induced cells was performed 24 h after shift to thiamine-free media. Acknowledgements: We would like to thank K. Gull for the gift of TAT1 and use of the microscope imaging system, and H. Okayama, P. Nurse, B. Stern, V. Simanis and S. MacNeill for strains. A.C. and N.B. are supported by the MRC. I.H. and M.M are supported by the CRC. M.M. was a European Exchange Fellow supported by the Royal Society/CNRS and the Association pour la Researche sur Cancer. This work is supported, in part, by European Community contract EV5V-CT94-0396.

21. 22. 23. 24.

25. 26.

27. 28.

References 1. Hartwell LH, Weinert TA: Checkpoints, controls that ensure the order of cell-cycle events. Science 1989, 246:629-634. 2. Murray AW: Creative blocks, cell-cycle checkpoints and feedback controls. Nature 1992, 359:599-604. 3. Weinert T, Lydall D: Cell-cycle checkpoints, genetic instability and cancer. Cancer Biol 1993, 4:129-140. 4. Sheldrick KS, Carr AM: Feedback controls and G2 checkpoints, fission yeast as a model system. Bioessays 1993, 15:775-782. 5. Enoch T, Nurse P: Coupling M phase and S phase, control maintaining the dependence of mitosis on chromosome replication. Cell 1991, 65:921-923. 2 6. Broek D, Bartlett R, Crawford K, Nurse, P: Involvement of p34cdc in establishing the dependency of S-phase on mitosis. Nature 1991, 349:388-394. 7. Hiraoka Y, Toda T, Yanagida M: The NDA3 gene of fission yeast encodes beta-tubulin - a cold-sensitive nda3 mutation reversibly blocks spindle formation and chromosome movement in mitosis. Cell 1984, 39:349-358. 8. Hoyt MA, Totis L, Botstein D: S. cerevisiae genes required for cellcycle arrest in response to loss of microtubule function. Cell 1991, 66:507-517. 9. Li R,Murray AW: Feedback control of mitosis in budding yeast. Cell 1991, 66:519-531. 10. Moreno S, Nurse, P: Regulation of progression through the G1 phase of the cell cycle by the ruml+ gene. Nature 1994, 367:236-242. 11. Carr AM: Cell-cycle co-ordination after too much rum. Bioessays 1994, 16:309-312. 12. Weinert TA, Hartwell LH: The RAD9 gene controls the cell-cycle response to DNA damage in Saccharomyces cerevisiae. Science 1988, 241:317-322. 13. Al-Khodairy F, Carr AM: DNA-repair mutants defining G2 check1992, point pathways in Schizosaccharomyces pombe. EMBO 11:1343-1350. 14. Rowley R, Subramani S, Young PG: Checkpoint controls in Schizosaccharomycespombe, radl. EMBO 1992, 11:1335-1342. 15. Enoch T, Carr AM, Nurse P: Fission yeast genes involved in coupling mitosis to completion of DNA replication. Genes Dev 1992, 6:2035-2046. 16. Al-Khodairy F, Fotou E, Sheldrick KS, Griffiths DJF, Lehmann AR, Carr AM: Identification and characterization of new elements involved in checkpoints and feedback controls in fission yeast. Mol Biol Cell 1994, 5:147-160. 17. Weinert TA, Kiser GL, Hartwell LH: Mitotic checkpoint genes in budding yeast and the dependence of mitosis on DNA replication and repair. Genes Dev 1994, 8:652-665. 18. Allen JB, Zhou Z, Siede W, Friedberg EC, Elledge SJ:The Sad1/Rad53 protein kinase controls multiple checkpoints and DNA-damageinduced transcription in yeast. Genes Dev 1994, 8:2401-2415. 19. Navas TA, Zhou Z, Elledge SJ:DNA polymerase e links the DNA replication machinery to the S phase checkpoint. Cell 1995, 80:29-39. 20. Ford JC, Al-Khodairy F, Fotou E,Sheldrick KS, Griffiths DJF, Carr AM: S.pombe 14-3-3 homologues encode an essential function required for the DNA-damage checkpoint. Science 1994, 265:533-535.

29.

30. 31. 32. 33. 34. 35. 36. 37. 38. 39.

40. 41. 42. 43. 44. 45.

46.

Walworth N, Davey S, Beach D: Fission yeast chkl protein kinase links the rad checkpoint pathway to cdc2. Nature 1993, 363:368-371. Ohi R,McCollum D, Hirani B, Denhaese+ GJ, Zang X, Burke ID, et al.: The Schizosaccharomyces pombe cdc5 gene encodes an essential protein with homology to c-myb. EMBO 1 1994, 13:471-483. Enoch T, Gould KL, Nurse P: Mitotic checkpoint control in fission yeast. Cold Spring Harb Symp Quant Biol 1991, 56:409-416. Hirano T, Funahashi 5, Uemura T, Yanagida M: Isolation and characterization of Schizosaccharomyces pombe cut mutants that 1986, 5: block nuclear division but not cytokinesis. EMBO 2973-2979. Schlegel R, Pardee AB: Periodic mitotic events induced in the absence of DNA replication. Proc Natl Acad Sci USA 1987, 84: 9025-9029. Tanaka K, Okazaki K, Okazaki N, Ueda T, Sugiyama A, Nojima H, Okayama H: A new cdc gene required for S phase entry of Schizosaccharomyces pombe encodes a protein similar to cdc10 and SW14 gene products. EMBO 1992, 11:4923-4932. Caligiuri M, Beach D: sctt functions in partnership with cdc10 in a transcription complex that activates cell-cycle start and inhibits differentiation. Cell 1993, 72:607-619. Carr AM: DNA-structure checkpoints in fission yeast. Semin Cell Biol 1995, 6:65-72. Siede W, Friedberg AS, Friedberg EC: RAD9-dependent G1 arrest defines a second checkpoint for damaged DNA in the cell cycle of Saccharomyces cerevisiae. Proc Nati Acad Sci USA 1993, 90:7985-7989. Kelly TJ, Martin S, Foresburg SL, Stephen RJ,Russo A, Nurse P: The fission yeast cdc18+ gene product couples Sphase to start and mitosis. Cell 1993, 74:371-382. Li JJ, Deshaies RI: Exercising self-restraint - discouraging illicit acts of Sand M in eukaryotes. Cell 1993, 74:223-226. Hayles J, Nurse P: A pre-start checkpoint preventing mitosis in fission yeast acts independently of p3 4cdc2 tyrosine phosphorylation. EMBOJ 1995, 14:2760-2771. Sazer S, Nurse P: A fission yeast RCC1-related protein is required for the mitosis to interphase transition. EMBO J 1994, 13:606-615 Diffley JFX, Cocker JH, Dowell SI, Rowley A: Two steps in the assembly of complexes at yeast replication origins in vivo. Cell 1994, 78:303-316. Gutz H, Heslot H, Leupold U, Loprieno N: Schizosaccharomyces pombe. In Handbook of Genetics, vol. 2. Edited by King RC. New York: Plenum Press; 1974:395-446. Maundrell K: nmtl+ of fission yeast. A highly transcribed gene completely repressed by thiamine. J Biol Chem 1990, 265:10857-10864. Barbet NC, Carr AM: Fission yeast weel protein kinase is not required for DNA-damage-dependent mitotic arrest. Nature 1993, 364:824-827. Hagan IM, Hyams IS: The use of cell division cycle mutants to investigate the role of microtubule distribution in the fission yeast Schizosaccharomycespombe. I Cell Sci 1988, 89:343-357. Woods A, Sherwin T, Sasse R,Macrae TH, Bains AJ, Gull K: Definition of individual components within the cytoskeleton of Trypanosoma-brucei by a library of monoclonal antibodies. J Cell Sci 1989, 93:491-500. Hagan IM, Yanagida M: The product of the spindle formation gene sadl+ associates with the fission yeast spindle pole body and is essential for viability. J Cell Biol 1995, 129:1033-1047. Nurse P,Thuriaux P, Nasmyth K: Genetic control of the cell division cycle in the fission yeast Schizosaccharomyces pombe. Mol Gen Genet 1976, 146:167-178. Hagan IM, Hayles J, Nurse P: Cloning and sequencing of the cyclinrelated cdc13 gene and a cytological study of its role in fission yeast mitosis. J Cell Sci 1988, 91:587-595. Nasmyth KA, Nurse P: Cell division cycle mutants altered in DNA replication and mitosis in the fission yeast Schizosaccharomyces pombe. Mol Gen Genet 1981, 182:119-124. Nasmyth KA: Temperature-sensitive lethal mutants in the structural gene for DNA ligase in the yeast Schizosaccharomyces pombe. Cell 1977, 12:1109-1120. Sarabia MJF, Mclnery C, Harris P, Gordon C, Fantes P: The cell-cycle genes cdc22 and suc22 of fission yeast Schizosaccharomyces pombe encode the large and small subunits of ribonucleotide reductase. Mol Gen Genet 1994, 238:241-251. Russell P, Nurse P: cdc25 functions as an inducer in the mitotic control of fission yeast. Cell 1986, 45:145-153. Received: 16 June 1995; revised: 28 July 1995. Accepted: 10 August 1995.