Research Article

The Neurospora Checkpoint Kinase 2: A Regulatory Link Between the Circadian and Cell Cycles

See allHide authors and affiliations

Science  04 Aug 2006:
Vol. 313, Issue 5787, pp. 644-649
DOI: 10.1126/science.1121716


The clock gene period-4 (prd-4) in Neurospora was identified by a single allele displaying shortened circadian period and altered temperature compensation. Positional cloning followed by functional tests show that PRD-4 is an ortholog of mammalian checkpoint kinase 2 (Chk2). Expression of prd-4 is regulated by the circadian clock and, reciprocally, PRD-4 physically interacts with the clock component FRQ, promoting its phosphorylation. DNA-damaging agents can reset the clock in a manner that depends on time of day, and this resetting is dependent on PRD-4. Thus, prd-4, the Neurospora Chk2, identifies a molecular link that feeds back conditionally from circadian output to input and the cell cycle.

Many core components of cellular circadian clocks constitute molecular feedback loops (14). Additional circuits, although not required for circadian function, can connect the clock to its outputs or even feed back to affect the core loops or input to them (57). In mammals, one prominent circadian output leads to regulation of the cell cycle (810).

The period-4 (prd-4) strain, a circadian clock mutant in Neurospora, was isolated more than 25 years ago in a genetic screen for mutants that failed to show an appropriate circadian phase shift in response to a light pulse (11, 12). The mutant displayed a period 3 hours shorter than wild type at 25°C that, unlike other short-period clock mutants, became even shorter as the ambient temperature rose. Heterokaryons bearing wild-type and mutant prd-4 nuclei have an intermediate period, indicating semidominance of this allele (13). Genetic mapping placed prd-4 on the right arm of linkage group I between os-1 and arg-13 (Fig. 1A) (13, 14).

Fig. 1.

Cloning and identification of prd-4. (A) Genetic map of part of the right arm of linkage group I showing the location of prd-4 with respect to mapped markers (37). (B) Genetic and physical map between os-1 and arg-13 (13, 14) showing predicted genes (15). (C) Sequencing of the region shown in (B) (light pink box) from diverse prd-4 mutant (prd-4m) strains revealed a single C to T transition associated with the prd-4m phenotype. (D) Western blotting with antisera directed against PRD-4 (see Methods in SOM) identifies an ∼80-kD protein (arrow) in prd-4m but not in the prd-4 deletion strain (Δprd-4). Asterisk, nonspecific band. (E and F) Expression of the allele bearing the single base change in the predicted prd-4 gene phenocopies the characteristic shortening of period length seen in prd-4m (E) and the partial loss of temperature compensation (F). Period lengths are average ±SD. Each data point in (F) corresponds to three or more race tubes (>20 individual circadian period estimates; see SOM).

Cloning and identification of the period-4 gene. A chromosomal walk from arg-13 to os-1 identified cosmids spanning the interval (13, 14), but the semidominance of the mutation, combined with inherent variability among heterokaryotic primary transformants, made it impossible to use transformation to identify prd-4 (13). The DNA sequence spanning the interval between arg-13 and os-1 (15) revealed about 39 thousand base pairs (kbp) containing 12 predicted coding sequences (Fig. 1B). DNA sequencing of this interval from strains bearing the prd-4 mutation consistently identified a single base-pair change in an open reading frame (ORF) close to os-1 (Fig. 1, B and C). The ORF was deleted by homologous gene replacement [(16), see Methods in Supporting Online Material (SOM)], and this strain was used as the recipient in DNA transformation experiments to confirm that the point mutation was indeed responsible for the prd-4 phenotype: We cloned the genomic fragment containing the gene and its promoter from the canonical prd-4 mutant strain and transformed this fragment into a neutral chromosomal site in the knockout strain (Fig. 1D and SOM). Because primary transformants are heterokaryons having both transformed and nontransformed nuclei, we backcrossed a primary transformant to the deletion mutant and isolated homokaryotic progeny, some of which expressed the mutant PRD-4 protein (for example, strain 356-16) and thus carried the transgene (Fig. 1D). Such strains recapitulated both the short period length and altered temperature compensation phenotypes of prd-4 (Fig. 1, E and F), which indicated that the transforming DNA contained the gene.

The semidominance of the original prd-4 mutation is consistent with altered function or gain of function, as was the appearance of PRD-4 protein in the mutant. However, to our surprise, the prd-4 deletion mutant displayed a period length similar to wild type's (Fig. 1E). This suggested the existence of functionally redundant gene(s), whose actions could compensate for the loss of PRD-4, or that the gain-of-function prd-4 mutation might cause PRD-4 to acquire a novel function, specificity, or regulation with respect to the clock. Because prd-4 null mutants are fully rhythmic, the gene is clearly not essential for basic clock function, but PRD-4 might connect other cellular processes to the circadian pacemaker.

Neurospora crassa PRD-4 protein is an ortholog of Chk2. After processing to remove three introns, prd-4 mRNA encodes a 702–amino acid protein bearing a forkhead-associated (FHA) domain followed by a serine-threonine protein kinase domain (Fig. 2A); this is an arrangement uniquely characteristic of a class of checkpoint kinases (17). BLAST searches identified PRD-4 as a checkpoint kinase 2 (Chk2), with human checkpoint kinase 2 (hChk2) being the closest homolog (BLAST score of e–72). The sequences show 38% identity and 51% similarity over the FHA domain and 43% identity and 62% similarity over the kinase domain (fig. S1). A reciprocal search of the Neurospora crassa genome with different Chk2 sequences also identifies PRD-4 as the highest-scoring homolog. The mutation responsible for the prd-4 phenotype results in a change from serine to leucine at position 493 (S493L) within a highly conserved portion of the kinase domain (Fig. 2A and fig. S1).

Fig. 2.

prd-4 is the Neurospora crassa Chk2. (A) Intronexon structure of the prd-4 transcript (top) reveals untranslated regions (light blocks), the ORF (dark blocks), and introns (diagonal lines). Variation in polyadenylation sites is depicted by differences in light block height. Locations of the FHA and kinase domains in PRD-4 (middle panel) and a portion of the alignment of the kinase domain of eight representative Chk2s (bottom panel) are shown. (Full alignments of FHA and kinase domains are in fig. S1.) Arrow, serine 493, mutated to leucine in the canonical prd-4m strain. (B) PRD-4 is phosphorylated in response to treatment with the DNA-damaging agent MMS. Protein extracts made from Neurospora treated with 0.1% MMS were used for Western blotting (see SOM for Methods). Arrows, PRD-4 and phosphorylated PRD-4 (PRD-4P); asterisk, nonspecific band. Treatment with lambda phosphatase (2PP) restores the mobility of PRD-4 to that of the untreated control and confirms that mobility shifts caused by MMS are due to phosphorylation of PRD-4. (C) prd-4, like hChk2, complements loss of RAD53 in yeast. S. cerevisiae strain W2105-7b (MATa leu2 sml1Δ::URA3 rad53Δ::HIS3 RAD5 W303) (25) was transformed with empty plasmid vector (pBJ245), human Chk2 (pMH267) (18), or pBJ245 containing S. cerevisiae RAD53 or Neurospora prd-4, each under a galactose-inducible promoter.

In mammals, Chk2 links the DNA damage–activated kinases ATM (ataxia telangiectasia mutated) and ATR (ATM and Rad3-related) with checkpoint effectors in the cell cycle and DNA-repair machinery (18, 19). In addition, mutations in Chk2 have been associated with human tumors, which qualifies the gene as a tumor suppressor (20). In many systems, Chk2 acts by phosphorylating the phosphatases Cdc25A and Cdc25C (17). The phosphorylation of human Cdc25C on Ser216 prevents it from dephosphorylating and thereby activating the Cdc2–cyclin B complexes required for entry into mitosis. Chk2 can also phosphorylate, and thereby stabilize, p53 (21, 22), which allows it to transactivate p53 target genes and leads to sustained G1 and G2 arrest and/or apoptosis (17). To bring about these actions, Chk2 is phosphorylated in response to DNA damage caused by ionizing radiation and by radiomimetic drugs like methylmethane sulfonate (MMS) (18, 23). Therefore, we treated Neurospora with 0.1% MMS and confirmed that, like other Chk2s, PRD-4 is phosphorylated in response to DNA damage (Fig. 2B). In Saccharomyces cerevisiae, the sensitivity to MMS seen in rad53Δ sml1Δ double mutants (24, 25) is rescued by transformation with human Chk2, and this sensitivity has been used to show that RAD53 and hChk2 are homologs (18). We applied this functional test by expressing prd-4 in a rad53Δ sml1Δ S. cerevisiae strain grown on media with and without MMS. As expected (18), the rad53Δ sml1Δ strain containing the plasmid vector alone is severely impaired by the presence of MMS in the medium (Fig. 2C, lane 2), but Neurospora PRD-4, like yeast Rad53 and human Chk2, can complement the rad53 loss of function (Fig. 2C, lanes 1, 3, and 4), which demonstrates that PRD-4 can functionally substitute for Rad53. Thus, on the basis of strong sequence similarity, regulation, and both functional and physiological tests, prd-4 encodes an ortholog of the cell cycle regulator Chk2.

PRD-4 signals DNA-damage information to the circadian clock. The semidominance of the original prd-4S493L allele suggests that it could encode an inappropriately active kinase, and an obvious candidate for a clock-relevant substrate is the clock protein FRQ (1, 4). The phosphorylation profile of FRQ changes in a well-defined and characteristic manner through the day (26), but this characteristic pattern is altered in the prd-4S493L mutant: FRQ is precociously phosphorylated during its daily cycle after a light-to-dark transfer (Fig. 3A), as well as after cycling is initiated by a temperature step (Fig. 3B and fig. S2A). Because the kinetics of FRQ phosphorylation and turnover determine period length (27), this early phosphorylation could account for the shortening of the FRQ cycle and the resulting short circadian cycle. A working model—activation of the checkpoint pathway via DNA damaging agents such as MMS might transmit a signal to the clock via FRQ—allows several predictions. First, treatment of Neurospora with γ radiation or a radiomimetic like MMS should result in activation of PRD-4, causing enhanced phosphorylation of FRQ; second, MMS-induced FRQ phosphorylation should be PRD-4–dependent and perhaps time-of-day dependent, because it would be superimposed on the daily circadian cycle of FRQ phosphorylation; and third, and most salient, activation of PRD-4/Chk2 should reset the clock.

Fig. 3.

PRD-4 participates in signaling DNA damage to the circadian clock via modification of FRQ. (A) Mutation of prd-4 results in early phosphorylation of FRQ. Western blots of FRQ over 24 hours in the dark in prd-4+ and prd-4S493L backgrounds. Electrophoresis was extended to accentuate the normal cycle in FRQ phosphorylation-induced mobility shifts (26). Light gray boxes highlight the largest difference in FRQ mobility between wt and prd-4S493L. (B) Time course of FRQ phosphorylation after transfer in darkness from 4°C to 25°C to initiate synthesis of FRQ. FRQ mobility is decreased earlier in samples harvested from strains expressing PRD-4S493L, which indicates premature phosphorylation of FRQ. Coom, Coomassie blue–stained gel (loading control). (See also fig. S2A.) (C) MMS-induced phosphorylation of FRQ requires PRD-4 and varies with circadian phase. prd-4+ (wt) and Δprd-4 (ko) strains were grown in liquid medium, in constant light at 25°C for 24 hours, before addition (+) or mock addition (–) of 0.1% MMS for 2 hours. In early evening (DD2) and subjective midday (DD16), FRQ is phosphorylated in response to MMS, which results in a mobility shift, but this is not seen in the Δprd-4 (ko) background. Note that after 6 hours in the dark (DD6, mid–subjective night), FRQ is normally highly phosphorylated, and MMS results in no additional shift.

To test the first prediction, cultures were treated with MMS at three different times (Fig. 3C and fig. S2B). After 2 hours in darkness (DD2), FRQ is abundant but not fully phosphorylated, and treatment with MMS results in clearly enhanced FRQ phosphorylation; this is not seen in the Δprd-4 background which indicates that PRD-4 is required to mediate this effect. In the subjective night at DD6, FRQ is maximally phosphorylated just before its degradation (26), and addition of MMS has no further apparent effect, presumably because FRQ is already maximally phosphorylated. In the mid–subjective day (DD16), when FRQ is newly synthesized but has yet to be significantly phosphorylated, addition of MMS again results in a significant increase in FRQ phosphorylation that is dependent on PRD-4 function (Fig. 3C).

Immunoprecipitation was used to test whether PRD-4 physically interacted with FRQ (Fig. 4A). FRQ weakly associates with PRD-4 in clock wild-type vegetative cultures before and after MMS treatment (lanes 2 and 3), and association of FRQ with PRD-4S493L is increased compared with wild type before DNA damage (lane 6). The interaction between FRQ and PRD-4 is attenuated after addition of MMS in the mutant, although it is difficult to determine whether this change is the same for the wild-type because of the weak interaction. A similar pattern of association before activation and diminution of binding following DNA damage has been reported for PML and BRCA1 in a mammalian system (28). These data, together with the short-period phenotype and precocious FRQ phosphorylation, suggest that the S493L mutation results in basal kinase activity that can be further stimulated by DNA damage resulting in hyperphosphorylation of its substrate (FRQ) and subsequent reduced association. Consistent with this, the association of PRD-4 with FRQ after MMS treatment is stronger when the D414A mutation (Asp414 replaced by Ala, which eliminates kinase activity) is introduced into the S493L background (lanes 7 and 9), although it is clear that MMS treatment still results in reduced association between PRD-4 and FRQ. Effects on circadian timing (Fig. 4B) are also consistent with the molecular data and the working model: Strains bearing the PRD-4S493L, D414A double mutations (increased association but catalytically inactive) have a wild-type clock, which strongly suggests that it is the precocious phosphorylation of FRQ (not just the increased interaction) that results in the characteristic short period of PRD-4S493L. Additionally, these data establish a direct association between FRQ and PRD-4 (Chk2), which confirms the first and second predictions and leads to the third. Because changes in FRQ phosphorylation are known to affect the clock (4, 27), these data indicate that MMS might also do so.

Fig. 4.

PRD-4 interacts with FRQ in vivo. (A) Antisera recognizing PRD-4 specifically immunoprecipitate PRD-4 complexes containing FRQ from extracts made from clock–wild type (prd-4+) cultures following 2 hours treatment with and without 0.1% MMS. The FRQ–PRD-4 interaction is increased in prd-4S493L and in the kinase-dead double mutant (prd-4S493L, D414A). Immunoprecipitations from Δprd-4 and frq null (frq10, prd-4S493L) provide negative controls. Input; total extract from wild type representing 3% of that used for immunoprecipitation (IP); PI, preimmune serum. (B) Race tubes of relevant strains. Although FRQ–PRD-4 interaction is preserved in the kinase-dead strain (A, lane 8), the period shortening effect on the clock is lost. Period estimates (±SD) are from three to six race tubes, each with five or six cycles.

To directly test the effect of a DNA-damaging agent on the circadian clock, Neurospora cultures were exposed to MMS for 2 hours immediately before the light to dark (L/D) transfer that normally sets the clock to subjective dusk and initiates circadian cycling. These cultures exhibited a 4-hour phase advance of the rhythm (with no effect on period) (Fig. 5A), consistent with an MMS-derived premature phosphorylation of FRQ; this clock effect was not seen in the Δprd-4 strain (Fig. 5A). Next, strains were systematically exposed to MMS for short intervals at times scanning a full circadian cycle (Fig. 5B). MMS yielded large phase advances during the subjective day, beginning at the time FRQ first appears (26) and decreasing over the course of the day, in line with the normal course of FRQ phosphorylation as predicted from the previous data (Fig. 3C); that is, clock effects are seen only when MMS can result in enhanced FRQ phosphorylation. The phase of the rhythm was unaffected by MMS in the Δprd-4 strain. Taken together, these data establish that DNA-damaging agents can act as strong clock-resetting cues and that PRD-4 (Chk2) is essential for mediating the effect of one of these, MMS, on the clock.

Fig. 5.

PRD-4 establishes a link between clock and cell cycle checkpoints. (A) MMS interferes with clock phasing in a PRD-4–dependent manner. Strains were treated (+0.1% MMS) or mock treated (no MMS) for 2 hours, allowed 1 hour to recover, and inoculated onto race tubes at 25°C in the dark. Strains never grown in liquid (no liquid) controlled for effects caused by manipulations (six race tubes per treatment). Densitometric scans are aligned by the 24-hour marks (solid vertical black lines). A dashed vertical line marks the peaks of the rhythm in controls on day 4, and a red line, the corresponding peak in the MMS-treated sample. Red bar shows the time of MMS-treatment recovery and dashed curve, growth before the first growth mark. (B) MMS resets the clock in a PRD-4–dependent and circadian phase–specific manner. The treatment described in (A) was carried out at different times of the subjective day for prd-4+ (top) and Δprd-4 (bottom) strains. Darkness from 0 to 11 hours corresponds to subjective night and from 11 to 22 hours to subjective day. Circadian times (CT) are indicated above. Data points represent phase ±1 SD, estimated from three to six race tubes compiled from four independent experiments. (C) prd-4 is regulated by the circadian clock. Average relative level (± 2 SEM; for three to five independent time series) of prd-4 (black) and frq mRNA (internal control, gray) determined by reverse transcriptase–polymerase chain reaction (RT-PCR) from samples collected over 2 days from both frq+ (top) (period length, 22 hours) and frq7 (bottom) (period length, 29 hours) cultures. (D) PRD-4 (as Chk2) is assumed to act through Cdc25 in Neurospora as part of the well-established eukaryotic checkpoint pathway that senses DNA damage and signals arrest to the cell cycle. PRD-4 also signals DNA damage information to the clock, in part by modifying FRQ and shifting the rhythm. The circadian clock established by the FRQ negative-feedback loop regulates the circadian cycling of prd-4 products, which establishes prd-4 as an output or clock-controlled gene and describes a nested feedback loop closing around the circadian clock.

These results suggest that PRD-4 acts only when there is DNA damage and that it is an essential mediator of the effects of DNA damage on the clock. Although there is no precedent for chemical mutagens resetting the clock, a growing body of knowledge supports a role for the clock in DNA-damage responses and tumor suppression (29) and for interrelationships between checkpoint kinases and clock components. For example, the human Timeless protein interacts with the cell cycle checkpoint protein Chk1 and the ATR-ATRIP complex (30), so this type of regulation may exist broadly in circadian systems.

In mammals, the clock protein Per2 regulates expression of tumor suppressor genes including p53; thus it may modulate sensitivity to DNA-damaging radiation and apoptosis in a circadian fashion (29). In addition, the gene encoding the cyclin B1–Cdc2 kinase regulator Wee1 is directly under circadian control in regenerating mouse liver (31). In fact, Chk2 in mammals controls cell cycle progression through regulation of Cdc25C, the same protein whose phosphorylation status and activity are modulated by the clock-controlled protein Wee1 (31). This parallel suggests that prd-4 might be similarly clock-regulated: In fact, prd-4 expression is regulated by the circadian clock, peaking in the subjective morning in parallel with frq (Fig. 5C).

This regulation of prd-4 by the clock closes a loop: Circadian regulation of prd-4 expression as an output of the clock could provide a time-of-day specific gate, albeit a weak one, for the ability of DNA-damaging agents to affect the clock (Fig. 5D). Strains bearing PRD-4S493L, however, have partially bypassed the need for activation of the kinase and have lost the loop from output to input closing around the core FRQ/WCC circadian feedback loop. This explains the circadian regulation of prd-4 expression, the PRD-4–dependent phase shift in response to MMS, and the effects of prd-4S493L on the period length of the clock. The possibility of a cell cycle–related circadian output feeding back onto the clock can perhaps also be seen in hindsight in studies where inhibitors suggested a role for cyclin-dependent kinases in the Bulla gouldiana circadian system (32). Regulation through Chk2 is consistent with the increased incidence of tumors in Per2 mutant mice, because this connection seems to be mediated through effectors like p53, another protein regulated by Chk2. Altogether, a variety of data implicate Chk2 as an important upstream link between the clock and cell cycle checkpoints. Thus, as previously suggested (9, 29), the clock has a protective role in addition to its pacemaker functions: Corroborative experimental and medical evidence show a higher incidence of mammary tumors in rats subjected to constant light (33) and of breast cancer among shift workers (34), as well as circadian variations in effectiveness of anticancer treatments (35, 36). The suggestive coordination of the clock and cell division through cell cycle checkpoints might further support the presumed ancient origin of clocks and their integral role in basic cell biology.

Note added in proof: Gery et al. (38) now report that, reminiscent of FRQ and PRD-4, the mammalian clock protein Per1 interacts with Chk2 and ATM both prior to and after DNA damage-induced activation, and that Per1 is needed for and promotes Chk2 activation. Although not therin examined, this result would be consistent with an effect on the mammalian clock initiated by DNA damage and effected via CHK2-mediated modification of clock proteins. Each study suggests that core clock proteins could act in tumor suppression, both through transcriptional control of cell cycle–related genes and through apparently conserved direct interactions with cell cycle checkpoint pathway proteins. All these data are consistent with an emerging role of circadian regulation in cancer.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article