Circadian rhythms in the synthesis and degradation of a master clock protein KaiC in cyanobacteria.

A circadian rhythm in the accumulation of the core clock protein KaiC has been proposed to be important for proper circadian timing in the cyanobacterium Synechococcus elongatus PCC 7942 under continuous light conditions. Cycling in the abundance of the KaiC protein is delayed to the rhythm of its mRNA by approximately 8 h, consistent with the proposed function of KaiC as a negative feedback regulator of kaiBC transcription. Here, we present temporal profiles of the synthesis and degradation of KaiC protein that determine the rhythm of its accumulation. The rate of KaiC synthesis shows a robust circadian oscillation, which is delayed to the mRNA rhythm slightly and advances the rhythm of KaiC accumulation by approximately 6 h. The stability of KaiC protein also shows circadian fluctuations, such that KaiC degradation is suppressed during the mid-subjective night. These results suggest that transcriptional, translational, and posttranslational processes are important for the proper circadian changes in KaiC accumulation. Moreover, the turnovers of the phosphorylated and non-phosphorylated forms of KaiC show robust circadian rhythms with an anti-phase relationship to each other. Interestingly, when translation was inhibited, KaiC degradation and phosphorylation proceeded within at least 4 h in a circadian phase-dependent manner. Thus, the circadian timing seems flexible even when any perturbation in protein synthesis occurs.

ATP-binding autokinase protein (6) that negatively regulates its own expression (7) in a KaiA-dependent manner (8). Interestingly, the overexpression of kaiC represses not only the kaiBC promoter but also most gene promoters (9). Therefore, KaiC is thought to be a promoter-nonspecific, genome-wide transcriptional modifier, possibly acting via its effect on the basic transcription machinery or on the state of chromosome compaction (9 -11).
Although the biochemical activities of the Kai proteins remain unclear, they show dynamic circadian patterns in their accumulation, protein complex formation, and posttranslational modification. The KaiB and KaiC proteins accumulate in a circadian fashion, peaking at circadian time (CT) 1 15-18. 2 under continuous light conditions, which delays the rhythm of kaiBC mRNA accumulation by ϳ8 h (12). This type of time lag between mRNA and protein rhythms has also been observed for some negative regulators in eukaryotic clock systems and is thought to be important in causing feedback loops to oscillate (5,13).
KaiC forms a hexamer in an ATP-dependent manner in vitro (10,14). The KaiC hexamer forms larger protein complexes with KaiA, KaiB, and a sensory histidine kinase, SasA, during the subjective night (15)(16)(17)(18). KaiC undergoes phosphorylation at Ser and Thr residues, most probably through its own autokinase activity. Phosphorylation of KaiC shows a robust circadian rhythm peaking at CT16. KaiA enhances the (auto)phosphorylation of KaiC both in vitro and in vivo (8), and this effect is inhibited by KaiB, which possibly enhances the auto-dephosphorylation activity of KaiC (11,18,19). The kaiA2 mutation in the KaiA protein lengthens the circadian period by ϳ8 h (7) and reduces KaiC phosphorylation (8). The kaiC15 mutation, which was mapped to one of two KaiA-binding domains of KaiC, suppresses the effects of the kaiA2 mutation, restoring the wild-type magnitude of KaiC phosphorylation and circadian period length (8). Thus, KaiA-mediated KaiC phosphorylation seems important for circadian pacemaking.
Importantly, a temporal increase in the amount of KaiC shifts the phase of the Synechococcus clock (7,12). Therefore, the rhythmic accumulation of KaiC is thought to be important for circadian timing. However, no quantitative information on KaiC protein turnover that determine its accumulation rhythm has been available. Here, we present temporal profiles of KaiC protein synthesis and degradation. Our results suggest that rhythmic protein synthesis and degradation are both important for the proper circadian accumulation of KaiC protein.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture-NUC42 (20) and NUC301 3 were used as wild-type reporter strains. NUC301 carries a P kaiBC ::luxAB reporter unit with a kanamycin-resistance gene at a specific targeting site designated NS1. kaiC13 (also named CLAb) (21) was used as an arrhythmic mutant strain. The PpsbAI::luxAB reporter unit of this strain was replaced with a P kaiBC ::luxAB cassette (with a selectable marker gene at the specific target site, NS1) to evaluate kaiBC promoter activity. Synechococcus cells were grown in modified BG-11 medium (22) under continuous light (LL) conditions (50 mol Ϫ2 s Ϫ1 from white fluorescent lamps) at 30°C.
Methionine Uptake Analysis-Methionine uptake was analyzed as described by Chen et al. (23) with some modifications. Synechococcus cells were grown in a continuous culture system to an optical density of 0.25 at 730 nm (A 730 ), exposed to two cycles of alternating 12-h light/ 12-h dark, and then placed under LL conditions. Cells were collected at the appropriate times, resuspended in 800 l of fresh BG-11 medium containing 1.48 MBq of [ 35 S]methionine (1000 Ci/mmol; Amersham Biosciences), and incubated at 30°C under LL for 30 min. After centrifugation (20,000 ϫ g, 2 min), cell pellets were washed with 1 ml of 1% casamino acid and then filtered through prewashed 0.45-m filters (Millipore). The filters were rinsed twice with 5 ml of cold fresh medium, and the radioactivity was measured with a liquid scintillation counter (Aloka LSC-5100). To correct for background counts, 800 l of cell suspension was denatured at 90°C for 20 min, cooled on ice, and then mixed with 1.48 MBq of [ 35 S]methionine. The cells were immediately filtered and processed as described above.
KaiC Protein Synthesis Rate Analysis-Cells were collected from the continuous culture system at the appropriate times, resuspended in 600 l of fresh BG-11 medium containing 1.11 MBq of [ 35 S]methionine (1000 Ci/mmol), and incubated at 30°C under LL for 30 min. After centrifugation, cell pellets were washed once with 1 ml of 1% casamino acid and then twice with 1 ml of fresh BG-11 medium, immediately frozen, and stored at Ϫ80°C. Cells were resuspended in 500 l of lysis buffer (20 mM HEPES, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, 1% SDS, 1 g/ml leupeptin, 1 g/ml pepstatin A, 1 mM phenylmethylsulfonyl fluoride, pH 8.0) and disrupted using a Multi-Beads Shocker (Yasui Kikai, Osaka, Japan) with zirconium beads (0.1 mm diameter; Biospec Products) at 2°C. After centrifugation, the supernatants were used as total cell extracts. Immunoprecipitation assays were performed as described by Kageyama et al. (17) with some modifications. Cell extracts (500 l with a total protein concentration of 1 mg/ml) were incubated at 4°C for 2 h with a 10-l bed volume of protein G-Sepharose beads (Amersham Biosciences) coupled with purified rabbit anti-KaiC IgG. The beads were washed six times with 1 ml of lysis buffer and shaken with 20 l of lysis buffer plus 30 l of SDS sample buffer (62.5 M Tris-Cl, 2% SDS, 10% glycerol, pH 6.8) for 5 min. Eluted proteins were size-fractionated by SDS-PAGE and then subjected to immunoblotting analysis using anti-KaiC antiserum as described previously (8) and autoradiography using a BAS 2000 image analyzer (Fuji Film).
RNA blot analysis was performed as described previously (16). KaiC Stability Analysis-Cells were collected from the continuous culture system at the appropriate times and incubated with chloramphenicol (Chm; 400 g/ml) at 30°C under LL for 4 or 8 h. Control samples were also incubated under the same conditions without Chm. After centrifugation, cell pellets were immediately frozen and stored at Ϫ80°C. The cells were resuspended in 500 l of SDS sample buffer, disrupted using a Multi-Beads Shocker, and used for Western blotting analysis. Densitometric analysis of the blots was performed with NIH Image software, version 1.61. In each experiment, the linearity of the densitometric values was verified with serial dilutions of extracts.

Circadian Oscillations in the Rate of KaiC Synthesis-To
analyze the temporal dynamics of KaiC protein synthesis, we used a combinatorial approach with a pulse-chase labeling method with [ 35 S]methionine and immunoprecipitation. Chen et al. (23) demonstrate the circadian rhythms of Synechococcus sp. RF-1 using the uptake of amino acids. Therefore, we initially examined a temporal profile of [ 35 S]methionine uptake activity in S. elongatus. Cells carrying a bacterial luciferase reporter gene used to monitor kaiBC expression (7) were grown in a continuous culture system under standard LL conditions after a 12-h dark treatment to reset the clock. Every 6 h, the cells were removed and incubated with liquid medium containing [ 35 S]methionine for 30 min under LL. The radioactivity of the incorporated [ 35 S]methionine was then measured to estimate the temporal profile of methionine uptake (Fig. 1B). Methionine uptake activity remained constant throughout the circadian cycle, whereas kaiBC promoter activity showed robust circadian fluctuations (Fig. 1A) (7).
For the protein synthesis assay, cells were incubated with [ 35 S]methionine for 30 min at 2-or 4-h intervals under LL. Proteins were extracted and subjected to immunoprecipitation with anti-KaiC antisera. The immunoprecipitated materials were subjected to SDS-PAGE, immunoblotting, and autoradiography. As shown in Fig. 1C (middle panel), a robust circadian oscillation in the rate of KaiC protein synthesis was demonstrated, which peaked at CT12. From the same continuous culture, we simultaneously collected other cell samples and processed them for Northern and Western blot analyses to compare kaiBC mRNA and KaiC accumulation profiles. As reported previously, the levels of kaiBC mRNA and KaiC protein show circadian rhythms peaking at ϳCT10 and ϳCT18, respectively (Figs. 1, C and D, upper and lower panels) (7,12,16). Thus, the rate of KaiC synthesis changes in a circadian manner in parallel with the accumulation of kaiBC mRNA with a slight delay, whereas it advances to the rhythm of KaiC protein accumulation by ϳ6 h. To our knowledge, this is the first demonstration of a circadian protein synthesis profile of clock-related proteins in any organisms. Under LL, KaiC undergoes robust circadian phosphorylation, which peaks at CT16 (8). KaiC protein was detected as doublet bands on immunoblots after SDS-PAGE on 10% gels as shown in Fig. 1C. The upper band with lower mobility corresponds to the phosphorylated form of KaiC, and the lower band corresponds to the non-phosphorylated form (8). Our autoradiography results show that the 35 S-labeled KaiC protein also appeared as doublet bands, indicating that the newly synthesized KaiC was phosphorylated rapidly within 30 min.
Circadian Rhythm in the Rate of KaiC Degradation-The accumulation profile of KaiC protein is determined by the balance between its synthesis and degradation. Xu et al. (11) recently have estimated the half-lives of the non-phosphorylated and phosphorylated forms of KaiC to be 8.8 and 2.0 h, respectively. This is based on the stability of transiently overexpressed KaiC in a kaiC-null background strain under the control of the Escherichia coli trc promoter that was controlled by pulsed administration of the inducer isopropyl-␤-D-thiogalactopyranoside. However, that experiment did not examine the temporal profile of endogenously expressed KaiC. To estimate a temporal profile of the rate of KaiC degradation, we used a different approach and monitored the stability of KaiC in the presence of the protein synthesis inhibitor Chm during the course of a circadian cycle. Fig. 2A shows that total protein synthesis was inhibited by Chm at 400 and 800 g/ml. 4 Cells were removed from a continuous culture grown under LL after a 12-h dark treatment. They were incubated with or without Chm (400 g/ml) for 4 h at 4-h intervals under LL, processed, fractionated by SDS-PAGE on 12% gels, and then subjected to immunoblot analysis with anti-KaiC antisera (Fig. 2B). As shown in Fig. 2, B-D, from late subjective night to early subjective day (CT20, CT24, and CT4), KaiC was reduced to ϳ70% within 4 h in the presence of Chm (indicating that a half-life was ϳ6 h), whereas around the mid-subjective night (CT16), KaiC was more stable (ϳ15% reduction, indicating an estimated half-life of ϳ20 h). In our experiment, the average halflife of KaiC protein was ϳ10 h. The fact that KaiC is most stable when its accumulation rhythm peaks suggests that a circadian change in the stability of KaiC contributes to the proper phasing of the rhythm of its amplitude or accumulation (see "Discussion").
We also examined the stability of KaiC in an arrhythmic mutant strain, kaiC13 (7,9). This mutant encodes the amino acid substitution G460E in KaiC. The level of KaiC accumulation in this mutant was continuously low, and the phosphorylated form of KaiC was more abundant than the non-phosphorylated form throughout the circadian cycle (Fig. 3B). No difference in the rate of degradation of KaiC was found at 8 and 16 h under LL in the mutant as shown in Fig. 3, C and D, confirming that the stability of KaiC is modified in a circadian fashion in wild-type cells. Moreover, the stability of KaiC protein is lower in the kaiC13 mutant than in the wild-type strain. The lower level of KaiC accumulation in the mutant strain is due to both a reduction in the level of kaiBC expression and a decrease in the stability of KaiC. Moreover, the phosphorylated form of KaiC in the kaiC13 mutant appeared more stable than the non-phosphorylated form (Fig. 3C).
Differential Turnover of Phosphorylated and Non-phosphorylated Forms of KaiC-Finally, we analyzed the temporal profiles of the phosphorylated and non-phosphorylated forms of KaiC in the presence of Chm. Fig. 4A presents a schematic diagram of phosphorylated and non-phosphorylated KaiC pro-tein turnover. We analyzed the same protein samples as were used to collect the data shown in Fig. 2 by separating both forms of KaiC on 10% gels (Fig. 4B). We calculated the relative changes in phosphorylated KaiC (upper band) and non-phosphorylated KaiC (lower band) during a 4-h treatment with Chm. When the translation of KaiC was inhibited by Chm at CT4 or CT8, the level of phosphorylated KaiC increased, whereas the level of non-phosphorylated KaiC dramatically decreased because of both degradation and phosphorylation (Fig. 4, A and D). These results are consistent with the obser- vation that both total KaiC and phosphorylated KaiC increased from CT4 to CT12 in the absence of Chm, whereas non-phosphorylated KaiC decreased (Fig. 4C) (8). These results also indicate that KaiC phosphorylation does not require de novo protein synthesis during the subjective day. In the presence of Chm during the late subjective night (CT20 -24), the level of phosphorylated KaiC was reduced dramatically because of both degradation and dephosphorylation (Fig. 4A), whereas the level of non-phosphorylated KaiC was less affected. Thus, the turnovers of the phosphorylated and non-phosphorylated forms of KaiC show robust circadian rhythms and the non-phosphorylated and phosphorylated forms have an anti-phase relationship to each other, even in the presence of Chm.

DISCUSSION
The levels of kaiBC mRNA and KaiC protein oscillate under LL, peaking at ϳCT8 and ϳCT16, respectively (Fig. 1D) (7,8,12). In this study, we revealed temporal changes in the rates of synthesis and degradation of KaiC. Rhythmic KaiC synthesis peaked at CT12 with a slight delay in the rhythm of mRNA accumulation (Fig. 1, C and D). Thus, the rate of KaiC protein synthesis seems primarily dependent on the level of kaiBC mRNA accumulation. Consistent with this finding, Xu et al. (11) have demonstrated that transiently induced kaiBC mRNA is degraded rapidly. We found that the rate of KaiC protein degradation is modified in a circadian phase-dependent manner (Fig. 2D). KaiC was most stable at CT16 (the calculated half-life at this time point was ϳ20 h), at which time its accumulation and phosphorylation levels were maximal (8,12). From the late subjective night to early subjective day, the stability of KaiC protein decreased (with a half-life of ϳ6 h) and the level of its accumulation decreased.
We previously estimated the circadian profile of the absolute level of cellular KaiC (18). The levels of KaiC protein oscillate with an average content of ϳ10,000 molecules/cell. Based on previous results, we estimated a quantitative profile of KaiC protein turnover (synthesis and degradation) per single cell (Fig. 5). The amount of KaiC protein degraded within 4 h (Fig.  5B) was calculated on the basis of the amount accumulated (Fig. 5A) and the mean value of its degradation rate at each time point (Fig. 2D). Furthermore, to address the relevance of circadian changes in KaiC synthesis and stability to the accumulation rhythm profile, we simulated KaiC accumulation profiles with constant rates of KaiC synthesis or degradation (Fig.  5C). We confirmed that the accumulation profile (Fig. 5C, black solid line), which was generated from the calculated amounts of synthesized and degraded KaiC protein shown in Fig. 5A, correlates well with the experimental data (Fig. 5C, gray solid  line). If the KaiC synthesis rate is constant at the average level, the resulting KaiC accumulation rhythm is largely dampened with a slight phase delay. On the other hand, if the degradation rate is constant, the amplitude of the rhythm of KaiC accumulation is also reduced with a slight phase advance. These results suggest that oscillations in both KaiC protein synthesis and degradation are important in generating a KaiC accumulation profile with proper amplitude and phasing and thereby determine the ϳ8-h delay between the mRNA and protein rhythms.
When Chm was added during the subjective day (CT4 -12), the level of phosphorylated KaiC increased and the level of non-phosphorylated KaiC decreased. The level of non-phosphorylated KaiC decreased greatly when Chm was added compared with the levels in the absence of Chm (Fig. 4B). This might occur, because the degradation rate of non-phosphorylated KaiC is rapid during this time or because the non-phosphorylated form of KaiC is phosphorylated by (auto)phosphorylation activity. The increase in the phosphorylated form of KaiC, regardless of the presence of Chm (Fig. 4, B and D), supports the latter possibility. During the subjective night, the rate of KaiC protein synthesis becomes lower, probably because  (Fig. 2D). To calculate the protein synthesis rate, we inferred that the integrated amount of KaiC synthesized over 24 h is equal to the amount of KaiC degraded over 24 h. The amount of KaiC synthesized over 4 h at the indicated times was calculated as the rate of KaiC synthesis multiplied by the average KaiC degradation amount within 4 h. C, simulated KaiC protein accumulation profiles when either the rate of protein synthesis or degradation was constant. Experimental values for KaiC accumulation are taken from the same values in Fig. 4A (gray solid  line). We obtained the following equation for KaiC accumulation after 4 h:  Fig. 2D). of a reduced kaiBC mRNA pool. At ϳCT16, KaiC protein becomes relatively stable. Previous gel-filtration and co-immunoprecipitation studies have demonstrated that the KaiC hexamer forms a tight complex with KaiA, KaiB, and SasA during mid-subjective night (16 -18). Therefore, it is plausible that such posttranslational modification or complex formation transiently changes the stability of KaiC. The increased KaiC stability would further promote the accumulation of KaiC while its synthesis rate decreased. The levels of the non-phosphorylated and phosphorylated forms of KaiC are also stable in the presence of Chm from CT16 to CT20 (Fig. 4). From the late subjective night to early subjective day (CT20 to CT4), the rate of KaiC synthesis is maintained at a low level and the rate of KaiC degradation increases. Therefore, the net amount of KaiC decreases. Moreover, the level of phosphorylated KaiC is also reduced during the subjective night (regardless of the presence of Chm) because of degradation or dephosphorylation. This dramatic change seems to be attributed to the formation of a KaiC-containing complex with KaiB, which accelerates KaiC auto-dephosphorylation activity at around CT20 (11,(17)(18)(19). Moreover, these results indicate that, even when translation is inhibited, KaiC degradation and phosphorylation proceed within at least 4 h in a circadian phase-dependent manner, suggesting a flexibility in the circadian timing that accommodates any perturbation in protein synthesis.
In Neurospora and Drosophila, the stability of the clock proteins, Period and Frequency, is modulated in a circadian fashion because of their rhythmic phosphorylation (5). In S. elongatus, KaiC is also rhythmically phosphorylated, most probably through autophosphorylation (Fig. 4B) (8). However, the effect of KaiC phosphorylation on its stability remains unclear. Using a transient KaiC overinducer in a kaiC-depleted strain, Xu et al. (11) suggest that the stability of phosphorylated KaiC is lower than that of non-phosphorylated KaiC. However, in this study, we demonstrated that KaiC is most stable when its phosphorylation peaks at around CT16 (Fig.  4D). Thus, there appears no simple correlation between the phosphorylation status and the stability of KaiC.