No Transcription-Translation Feedback in Circadian Rhythm of KaiC Phosphorylation

See allHide authors and affiliations

Science  14 Jan 2005:
Vol. 307, Issue 5707, pp. 251-254
DOI: 10.1126/science.1102540


An autoregulatory transcription-translation feedback loop is thought to be essential in generating circadian rhythms in any model organism. In the cyanobacterium Synechococcus elongatus, the essential clock protein KaiC is proposed to form this type of transcriptional negative feedback. Nevertheless, we demonstrate here temperature-compensated, robust circadian cycling of KaiC phosphorylation even without kaiBC messenger RNA accumulation under continuous dark conditions. This rhythm persisted in the presence of a transcription or translation inhibitor. Moreover, kinetic profiles in the ratio of KaiC autophosphorylation-dephosphorylation were also temperature compensated in vitro. Thus, the cyanobacterial clock can keep time independent of de novo transcription and translation processes.

Circadian clocks are ubiquitous endogenous biological timing processes that adapt to daily alterations in environmental conditions in bacteria, fungi, plants, and animals. In most model organisms, the core process that generates and maintains self-sustaining oscillations is thought to be a ∼24-hour-period transcriptional and translational oscillatory process (TTO) based on negative feedback regulation of clock genes (1, 2). Cyanobacteria are the simplest organisms known to exhibit circadian rhythms. In the cyanobacterium Synechococcus elongatus PCC 7942, three adjacent genes at the kai locus (kaiA, kaiB, and kaiC) are essential for circadian function (3). The amounts of accumulated kaiBC operon mRNA and KaiC protein oscillate in a circadian manner with a ∼6-hour delay under continuous light (LL), reminiscent of the temporal profiles of negative elements in eukaryotic clock systems (4). Continuous overexpression of kaiC nullifies circadian rhythms and abolishes kaiBC expression, whereas a transient increase in KaiC sets the phase of the rhythm (3, 4). These observations led to the cyanobacterial TTO model for explaining prokaryotic circadian rhythm generation, in which KaiC and KaiA negatively and positively regulate kaiBC transcription, respectively (3). In the cyanobacterial circadian system, essentially all the promoter activities on the genome show such rhythms (5). Moreover, the functional target of the Kai clock system may not be clock-gene–specific but could affect genome-wide gene expression, such as chromosomal DNA conformation (68). In this revised model, rhythmic kaiBC transcription is still thought to be essential for producing and maintaining a self-sustaining basic oscillation underlying circadian genome-wide global expression control, consistent with the widely accepted TTO model.

Because Synechococcus is an obligate photoautotroph, most of its metabolic activities, including total RNA and protein syntheses, are quickly suppressed when the cells are transferred to constant dark (DD) conditions (9). Nevertheless, dark treatments for 3 to 42 hours affected the phase of the rhythm less when the bacteria were returned to continuous light (LL) (4), which reveals that the clock can free-run during DD and thus contradicts the TTO model. Indeed, it has not yet been examined whether rhythmic kaiBC expression per se is truly essential for generating circadian rhythms. Thus, we examined kaiA and kaiBC mRNA accumulation profiles under DD conditions. Expression of kaiBC mRNA showed robust circadian rhythms during LL after a 12-hour dark treatment (Fig. 1, A and B). Conversely, when cells were transferred to DD from the light, the kaiBC mRNA was reduced within 4 hours to almost undetectable levels (less than 0.1% of the peak levels during LL) (Fig. 1, A and B). All kaiA mRNA also disappeared (Fig. 1B). We confirmed that general transcription activity, as measured by the incorporation of [32P]-UTP (uridine triphosphate) into RNA, was lowered to background levels, whereas active transcription was observed in the light (Fig. 1C). Therefore, de novo clock-gene expression and Kai protein syntheses, essential processes in the TTO model, must be shut off almost completely in DD.

Fig. 1.

kaiBC mRNA expression is immediately abolished in the dark. Temporal expression patterns of kaiBC (A and B) and kaiA (B) mRNAs during continuous light (LL) and dark (DD) conditions. Cyanobacterial cells were grown in a continuous culture system at 30°C (OD730 of ∼0.2). After three 12:12 hour light-dark (LD) cycles, cells were transferred to LL (40 μmol m–2 s–1; white bar). Cells were also transferred to DD (black bar) after two LD cycles and 12 hours light (hour 0 during DD is equivalent to hour 12 during LL). The kaiABC-deleted cells (Δ) were collected at hour 12 during LL (LL 12). Northern blotting was performed as described (3). (C) Total RNA synthesis was dramatically lowered in the dark. [32P]UTP uptake assay (23) was performed with cells that had been grown during LL or DD at circadian time (CT) 8 (LL 32, DD 20) or at CT 20 (LL 20, DD 8). Although transcription activity was lowered in the dark, rRNA is highly stable (A) because it is protected within ribosomes.

Under LL conditions, KaiC shows not only a circadian rhythm of accumulation but also circadian changes in phosphorylation at Ser and Thr residues (10). Phosphorylation of KaiC is essential for circadian timing, because mutation of the phosphorylation sites abolishes circadian oscillations in Synechococcus (11, 12).

To address how Kai proteins behave in the dark, we examined the accumulation profiles of the Kai proteins and the phosphorylation profile of KaiC under DD conditions. In contrast to the rhythmic KaiC accumulation in LL, KaiA, KaiB, and KaiC proteins were maintained at constant levels, with abolition of their rhythmic accumulation after transfer to DD (Fig. 2, A to C and E). Therefore, Kai proteins were more stable during DD than during LL (13). A KaiC-associating histidine kinase, SasA (14), also accumulated to a constant amount, as observed in LL (Fig. 2E). Even in the absence of the rhythmic accumulation of the Kai proteins and of detectable kaiA and kaiBC mRNAs, KaiC phosphorylation showed a robust circadian rhythm lasting at least 56 hours in DD, equivalent to that observed in LL (Fig. 2, A, F, and G). The KaiC phosphorylation rhythm in DD was less affected in the presence of excess amounts of a transcription or translation inhibitor (200 μg/ml rifampicin or 400 μg/ml chloramphenicol) (Fig. 2A). These inhibitors repressed >95% of total protein and RNA syntheses during LL at these concentrations (15, 16). Therefore, in contrast to the widely accepted TTO model, the self-sustainable rhythm during DD can persist at the posttranslational level without any accompanying de novo synthesis of central clock-gene mRNAs or their encoded proteins. The kaiC4 mutant strain that harbors a P236S substitution in KaiC lengthens the period of the kaiBC expression rhythm by 4 hours during LL (3). The phosphorylation rhythm in the dark was lengthened accordingly (Fig. 2, F and G). Therefore, the circadian period length appears precisely determined, even in the absence of transcriptional feedback. In addition, because the KaiC phosphorylation cycle in the presence of rifampicin was also observed during LL (Fig. 2A), kaiBC transcription is not necessary for generation of the posttranslational rhythm, regardless of LL or DD conditions.

Fig. 2.

KaiC phosphorylation rhythm without kaiBC mRNA accumulation. (A) Temporal profiles of KaiC accumulation and phosphorylation during LL and DD were examined by Western blot. Total protein (1 μg) prepared from cells was analyzed by immunoblotting with antiserum to KaiC on 10% SDS polyacrylamide gels (bisacrylamide:acrylamide = 1:149), as described previously (10). The upper and lower bands correspond to phosphorylated (P) and nonphosphorylated (NP) forms of KaiC, respectively. When Rifampicin (Rif) (200 μg/ml) and chloramphenicol (Cm) (400 μg/ml) were added to the cultures at hour 0 during DD (DD 0) or at hour 12 during LL (LL 12), the KaiC phosphorylation rhythm profiles were less affected. Representative data are shown. (B) The same protein samples of KaiC in LL or DD used in (A) were analyzed on 10% polyacrylamide gels (bisacrylamide:acrylamide = 2:73) to appear as single-band signals. (C) Quantification of KaiC by densitometry of the blots shown in (B). Values at times LL 12 or DD 0 were normalized to 1.0. (D) Percentages of the phosphorylated KaiC (P-KaiC) levels compared with the total KaiC levels in (A) are shown. (E) KaiA, KaiB, and SasA accumulation was also maintained at constant levels in the dark (Western blot). (F) The KaiC phosphorylation rhythm under DD in wild-type (WT) and kaiC4 mutant strains (Western blot). Asterisks indicate the peaks of KaiC phosphorylation. (G) Densitometric analysis of (F).

The basic features of circadian rhythms include temperature and nutrient compensation such that the periods are similar at different ambient temperatures or with different nutritional supplementation (1). Because the KaiC phosphorylation rhythm is maintained with a ∼24-hour period, even when metabolic activities are lowered during DD, the rhythm must be nutrient compensated. The period of KaiC phosphorylation cycling under DD was almost the same at different temperatures (25°C, 30°C, and 37°C) (Fig. 3, A and B). Thus, the circadian oscillatory process during DD is temperature compensated.

Fig. 3.

Analysis of KaiC phosphorylation at different temperatures. (A and B) Western blot analysis. KaiC phosphorylation rhythms were examined under DD at 25°C, 30°C, and 37°C. Densitometric analysis of (A) is shown in (B). (C and D) In vitro KaiC-autokinase/autophosphatase activity at different temperatures. Recombinant KaiA and KaiC proteins were produced in E. coli and purified as described (10). In the presence (0.05 μg/μl) or absence of KaiA, KaiC (0.2 μg/μl) was incubated in a reaction buffer (50 mM Tris, 150 mM NaCl, 5 mM MgCl2, and 1 mM ATP; pH 8.0) at 25°C, 30°C, and 35°C. Samples of the reaction mixtures were collected at the indicated times and subjected to SDS–polyacrylamide gel electrophoresis on 7.5% gels followed by Coomassie brilliant blue staining (C). Densitometric analysis of (C) is shown in (D), in which the ratio of P-KaiC to total KaiC is plotted.

KaiC phosphorylation in vivo is regulated by its own autokinase (10, 17) and autophosphatase (6, 18, 19) activities. KaiA activates KaiC autophosphorylation (or inhibits KaiC autodephosphorylation), whereas KaiB negates KaiA's effect on KaiC both in vitro and in vivo (6, 10, 18, 19). To characterize the autophosphorylation-dephosphorylation reactions of KaiC in vitro, recombinant KaiC protein was purified from Escherichia coli and incubated with adenosine triphosphate at different temperatures in the presence or absence of KaiA (25°C, 30°C, and 35°C). Prepared recombinant protein (time zero) appeared as phosphorylated and nonphosphorylated bands on SDS polyacrylamide gels (Fig. 3C). At 30°C, KaiC was almost completely dephosphorylated after 6 hours of incubation with ATP, whereas the addition of KaiA enhanced the phosphorylation of KaiC (Fig. 3C). The equilibrated ratio of phosphorylated and nonphosphorylated forms of KaiC, and the transition profiles to the equilibrium states, were less affected by different temperatures (25°C and 35°C) in the presence or absence of KaiA (Fig. 3, C and D) (20). These phenomena are probably caused by two antagonistic reactions—autophosphorylation and autodephosphorylation—each with a similar temperature coefficient regardless of the presence of KaiA. The temperature compensation of the net KaiC autophosphorylation and autodephosphorylation reactions in vitro is essentially due to the biochemical properties of the KaiC protein and does not require any other specific proteins or de novo synthesis of KaiC. Therefore, KaiC's temperature-insensitive autophosphorylation-dephosphorylation reaction rates may partly explain the temperature compensation of the KaiC phosphorylation cycle during DD and are critical processes for circadian timing.

The KaiC phosphorylation rhythm in vivo is accompanied by formation of a series of KaiC complexes with KaiA and/or KaiB in a circadian fashion during LL (18, 21). KaiA stimulates phosphorylation of KaiC hexamer progressively, and then the phosphorylated KaiC hexamer forms a complex with KaiA. This KaiA binding likely triggers the binding of phosphorylated KaiC-KaiA complexes to KaiB, an attenuator of KaiC autophosphorylation, to shift KaiC from a phosphorylation-dominating state to a dephosphorylation-dominating state. Reduction in the KaiC phosphorylation level would, in turn, accelerate phosphorylation reactions and/or inhibit dephosphorylation with as-yet-unknown nonlinear processes, whereby both states alternate periodically. Such autoregulatory posttranslational dynamics of the Kai proteins would be the core to generating the temperature-compensated, self-sustaining KaiC phosphorylation cycle as a minimal circadian circuit during DD (Fig. 4). The genome-wide transcriptional control exerted by KaiC that occurs during LL (6, 7) may connect to the minimal oscillator to form an extended feedback loop that further amplifies and stabilizes the posttranslational oscillatory process and serves as a major source of physiologically functional rhythm outputs (Fig. 4). Posttranslational biochemical oscillation would also be physiologically important as a “time memory” process to keep circadian rhythms ticking even in dark or metabolically limited (nonpermissive) conditions that severely perturb transcriptional or translational processes.

Fig. 4.

A model for the posttranslational oscillator coupled with TTO. The KaiC phosphorylation cycle can be maintained in the dark as a minimal timing loop without transcription or translation (gray area). During LL, gene expression activated by an energy supply from photosynthesis expands the oscillation to the TTO form (green area). Two histidine kinases (SasA and CikA) (24) might be required to connect KaiC function to a process in the general transcription mechanism, such as chromosome superhelicity (8), which feeds back to the kaiBC promoter (PkaiBC) activity and regulates output gene expression globally in a circadian manner. In the dark or under nutrition-limited conditions, the posttranslational oscillator may work as a “time memory” process to ensure robust circadian organization in Synechococcus. Pi and CCGs indicate phosphate and clock-controlled genes, respectively.

Robust oscillations of the phosphorylation of core clock proteins have been shown to be critical in these circadian systems in Neurospora, Arabidopsis, Drosophila, and mammals (1). These phosphorylations have been shown to affect stability, intracellular localization, and/or DNA-binding property of key clock components, and these cyclic changes are important to maintain robust cycling of the TTO processes (1). Moreover, constitutive induction of both period and timeless genes in Drosophila has been shown to be sufficient to produce cycling in the coding protein abundances (22), further supporting the critical role of posttranslational regulation in the eukaryotic clock system. We show here that the de novo synthesis of core clock elements is not an absolute requirement in allowing biochemical reactions to oscillate with a temperature-compensated circadian period.

Supporting Online Material

Materials and Methods


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article