Abstract
Arrest of the cell cycle at the G2 checkpoint, induced by DNA damage, requires inhibitory phosphorylation of the kinase Cdc2 in both fission yeast and human cells. The kinase Wee1 and the phosphatase Cdc25, which regulate Cdc2 phosphorylation, were evaluated as targets of Chk1, a kinase essential for the checkpoint. Fission yeast cdc2-3w Δcdc25 cells, which express activated Cdc2 and lack Cdc25, were responsive to Wee1 but insensitive to Chk1 and irradiation. Expression of large amounts of Chk1 produced the same phenotype as did loss of the cdc25gene in cdc2-3w cells. Cdc25 associated with Chk1 in vivo and was phosphorylated when copurified in Chk1 complexes. These findings identify Cdc25, but not Wee1, as a target of the DNA damage checkpoint.
Eukaryotic cells have cell cycle checkpoints that arrest division in response to DNA damage (1, 2). In the case of damage inflicted during the G2 phase of the cell cycle (gap of time after DNA synthesis, but before mitosis), arrest occurs before the onset of mitosis. In metazoans, G2 checkpoints are important for the maintenance of genome integrity, allowing time for the repair of damaged DNA, or, in the case of severe damage, for activation of programmed cell death. In haploid yeast the G2 checkpoint is required for viability. Understanding how the checkpoint signal intersects with the central machinery controlling progression from the G2 to the M phase is a major goal of current cell cycle studies.
In the fission yeast Schizosaccharomyces pombe, checkpoint arrest in the G2 phase requires a large number of proteins, many of which are believed to play a direct role in DNA repair. These proteins include Rad3, a kinase related to the ATM protein that is defective in ataxia telangiectasia patients (3). Damaged DNA is presumed to activate Chk1, a protein kinase that is essential for the checkpoint arrest (4-6). The ultimate target of the checkpoint signal is believed to be Cdc2, the cyclin-dependent kinase that induces mitosis. In the normal cell cycle of fission yeast and mammalian cells, the timing of mitosis is determined by the inhibitory phosphorylation of Cdc2 (7). In fission yeast this phosphorylation occurs on Tyr15 and is catalyzed by the kinases Wee1 and Mik1, with Wee1 being the most active. Dephosphorylation of Tyr15 and consequent induction of mitosis is catalyzed by the phosphatase Cdc25. The induction of mitosis is thought to be facilitated by activation of Cdc25 and inhibition of Wee1 activity during the G2-M transition. Initial studies suggested that the DNA damage checkpoint operated independently of tyrosine phosphorylation of Cdc2 (8, 9), but a more recent analysis established that inhibitory phosphorylation of Tyr15 is essential for the checkpoint (10). These findings suggest several mechanisms for Chk1-mediated regulation of Cdc2, including the Chk1-dependent activation of Wee1 or inhibition of Cdc25 (10, 11). We designed genetic and biochemical experiments to test these possibilities.
Cells of the genotype wee1-50 Δmik1, which lack Mik1 and express temperature-sensitive Wee1, undergo rapid dephosphorylation of Cdc2 on Tyr15 and induction of mitosis when shifted to the restrictive temperature (12). Irradiation of these cells before the temperature shift causes delay of dephosphorylation of Cdc2 and of entry into mitosis (10). Thus, irradiation induces a transient cell cycle arrest after inactivation of Tyr15kinases, potentially by inhibition of Cdc25 activity. An experiment was performed to determine whether this delay was due to an authentic checkpoint mediated by Chk1 kinase. For this experiment we usedchk1 + and Δchk1 alleles in awee1-50 Δmik1 background. Cells synchronized in early G2 phase were exposed to gamma irradiation or mock irradiated and then shifted from 25° to 35°C. Irradiation caused a ∼60-min delay of mitosis in the chk1 + cells (Fig. 1). This delay is attributable to a reduction in the rate of dephosphorylation of Cdc2 on Tyr15 (10). In contrast, gamma irradiation did not delay mitosis in Δchk1 cells (Fig. 1). Therefore, Chk1 is required for the irradiation-induced delay of mitosis observed after inactivation of Wee1 and Mik1 kinases.
Requirement of Chk1 for the radiation-induced delay of mitosis that occurs after inactivation of Wee1 and Mik1 kinases. Synchronous cultures of wee1-50 Δmik1 (PR754) orwee1-50 Δmik1 Δchk1 (NR1604) cells in early G2 phase were split and irradiated with 100 Gy of gamma radiation (+γ) or mock irradiated (−γ). The cultures were then shifted to 35°C to inactivate Wee1-50 protein. Cell cycle progression was monitored by counting the percent of cells undergoing septation (24).
These findings implicate Cdc25 as a potential target of Chk1 regulation, but they do not indicate whether Wee1 may also be regulated by Chk1. We used a cdc2-3w Δcdc25 strain to address this question. The cdc2-3w mutation is a dominant active allele that relieves the requirement for Cdc25 but leaves Cdc2 responsive to Wee1. The latter point is evident from the observations thatcdc2-3w wee1-50 cells undergo lethal premature mitosis when incubated at 35°C and that cdc2-3w cells are sensitive to increased wee1 + gene dosage (13, 14). In addition, elimination of the kinase Nim1, an inhibitor of Wee1 (15-18), causes a large increase in the size ofcdc2-3w Δcdc25 cells, indicative of a G2 delay resulting from increased Wee1 activity (19).
If Chk1 activates Wee1, then cdc2-3w Δcdc25 cells should undergo an irradiation-induced checkpoint delay or arrest. However, whereas wild-type cells underwent cell cycle arrest in response to gamma irradiation, cdc2-3w Δcdc25 cells did not (Fig. 2A). The irradiatedcdc2-3w Δcdc25 cells underwent division at a size equivalent to the mock-irradiated cdc2-3w Δcdc25 cells. Expression of large amounts of a fusion protein of glutathione-S-transferase (GST) with Chk1 induces G2arrest, presumably by inappropriate activation of the checkpoint signal (5, 10, 20). Whereas wild-type cells underwent cell cycle arrest in response to GST-Chk1 expression, cdc2-3w Δcdc25cells were insensitive (Fig. 2B). The cdc2-3w Δcdc25 cells that expressed GST-Chk1 continued to divide at a rate and cell size that was equivalent to that of cdc2-3w Δcdc25 cells in which GST-Chk1 expression was repressed.
Responses of fission yeast strains to gamma radiation and expression of GST-Chk1. (A) Asynchronous wild-type (PR109) and cdc2-3w Δcdc25 (GL192) cells growing in YES liquid medium at 25°C were irradiated at 3 Gy min−1 or mock irradiated and photographed after 6 hours. Wild-type cells ceased division and became elongated in response to irradiation, exhibiting a typical checkpoint response. Irradiatedcdc2-3w Δcdc25 cells continued division at a cell length that was not significantly different from the mock-irradiated control. (B) Wild-type (BF1921), cdc2-3w Δcdc25(BF1910), cdc2-3w (BF1911), and Δwee1(NR1970) strains having an integratednmt1:GST-chk1 + construct were grown in media that derepressed (EMM2, nmt1:GST-chk1 on) or repressed (EMM2+B1, nmt1:GST-chk1 off) expression of GST-Chk1 from the thiamine-repressible nmt1 promoter (25). Cells were photographed after 16 hours (27 hours in the case of NR1970) at 30°C. Expression of GST-Chk1 caused cell cycle arrest in wild-type cells, whereas cdc2-3w Δcdc25 cells continued to divide. Cells of the genotype cdc2-3w became moderately elongated in response to GST-Chk1 expression, dividing at a size similar to that of cdc2-3w Δcdc25 cells. Elongated Δwee1cells first appeared at ∼22 hours after incubation in derepressing medium. Numbers in panels refer to cell length at division. Scale bars represent 22 μm.
These studies indicate that the DNA damage checkpoint does not activate Wee1 because cdc2-3w Δcdc25 cells respond to increased Wee1 activity but are insensitive to irradiation and expression of GST-Chk1. This conclusion is consistent with the fact thatwee1 − cells undergo cell cycle arrest in response to irradiation (8). Moreover, GST-Chk1 expression in Δwee1 cells led to the appearance of elongated cells, although the response of Δwee1 cells was delayed relative to that of wild-type cells (Fig. 2B). This delay may account for the previous failure to observe an effect of Chk1 overexpression inΔwee1 cells (11).
Although the cdc2-3w allele bypasses the requirement for Cdc25, cdc2-3w cells remain responsive to changes in Cdc25 activity. This conclusion is derived from the observation thatcdc2-3w cells divide at a length of ∼9 μm, whereascdc2-3w Δcdc25 cells divide at ∼17 μm (13). If Chk1 primarily acts by inhibiting Cdc25, then expression of large amounts of GST-Chk1 in a cdc2-3w background should result in the same phenotype as mutational inactivation of Cdc25. As predicted by this hypothesis, cdc2-3w cells that overexpressed Chk1 underwent division at a size that was very similar to that ofcdc2-3w Δcdc25 cells (Fig. 2B).
We tested whether the checkpoint mechanism might involve a close interaction between Cdc25 and Chk1. Lysates from cells expressing functional GST-Chk1 fusion protein or unfused GST were analyzed by glutathione (GSH)-Sepharose precipitation followed by immunoblotting. GST-Chk1 precipitated with Cdc25, whereas no Cdc25 was detected in association with GST (Fig.3A). Incubation of GST-Chk1 with associated Cdc25 in the presence of [γ-32P]adenosine triphosphate (ATP) resulted in phosphorylation of Cdc25 (Fig. 3B), suggesting that Cdc25 may be a direct substrate of Chk1 kinase. Chk1 protein purified from an insect cell expression system phosphorylated Cdc25 in vitro (21).
Association of Cdc25 with GST-Chk1 in vivo and phosphorylation of Cdc25 in GST-Chk1 complexes in vitro. (A) GST or GST-Chk1 were expressed in cells in which genomic cdc25 + was replaced withcdc25:6HA, a functional cdc25construct containing a sequence encoding six tandem copies of the hemagglutinin (HA) epitope (26). GST and GST-Chk1 were purified by GSH-Sepharose chromatography. Immunoblotting with antibody to GST (bottom panel) detected GST as a ∼27-kD protein (lane 1), whereas GST-Chk1 was detected as a full-length ∼80-kD fusion protein together with ∼65-, ∼63-, and ∼27-kD presumptive degradation products (lane 2). Immunoblotting with antibody to HA (top panel) detected HaCdc25 (gene product of cdc25:6HA) in association with GST-Chk1. (B) Incubation of GST and GST-Chk1 complexes with [γ-32P]ATP followed by immunoprecipitation with antibody to Cdc25 showed that Ha-Cdc25 became phosphorylated in the GST-Chk1 complexes (lane 2), whereas no Ha-Cdc25 was detected in association with GST (lane 1) (27).
Activation of the DNA damage checkpoint requires Rad3, a kinase related to the ATM protein that is defective in ataxia telangiectasia patients (3). DNA damage leads to increased phosphorylation of Chk1 by a Rad3-dependent process, suggesting that Chk1 may be activated by phosphorylation (5). Our studies identify Cdc25 as a key, possibly direct, target of Chk1. In addition, these findings exclude Wee1 as an important Chk1 substrate. Therefore, we propose that Rad3-dependent activation of Chk1 leads to negative regulation of Cdc25 (Fig. 4). This negative regulation may occur by direct inhibition of Cdc25 activity, prevention of the activation of Cdc25 that occurs at the G2-M transition, or interference in the interaction between Cdc25 and Cdc2. Inhibitory phosphorylation of Cdc2 is crucial for G2 DNA damage arrest in mammalian cells (22, 23). In these cells it is not known whether this arrest is brought about by inhibition of Cdc2 dephosphorylation, nor is it known if mammals have a Chk1 homolog. However, in view of the striking degree of homology of mitotic control mechanisms in fission yeast and mammals, we expect that the S. pombe checkpoint control will serve as a useful paradigm for investigating the DNA damage checkpoint mechanism in more complex organisms.
Model of the DNA damage checkpoint mechanism in fission yeast. Rad3 and Chk1 kinases are required for the checkpoint. Chk1 undergoes a Rad3-dependent phosphorylation in irradiated cells, and Chk1 overexpression induces cell cycle arrest by a Rad3-independent mechanism (20), indicating that Chk1 activation is regulated by Rad3, perhaps by direct phosphorylation. Chk1 inhibits Cdc25 and thereby prevents Cdc2 Tyr15 dephosphorylation.
↵* To whom correspondence should be addressed. E-mail: prussell{at}scripps.edu.
