The cyanobacterial circadian clock is based on the intrinsic ATPase activity of KaiC

The earth rotates on its axis with a period of 24 h, imposing a dramatic diurnal oscillation on the biosphere. Endogenous biological clocks with circadian periods of ≈24 h are nearly ubiquitous, from bacteria to humans, and serve to coordinate organisms with these daily environmental changes. These clocks make important contributions to fitness, and, in competitive environments, disruption of the circadian clock has dire consequences in cyanobacteria, plants, and mammals (1). Enormous progress has been made in elucidating the mechanism of circadian timekeeping in the cyanobacterium Synechococcus elongatus PCC 7942. The kaiABC locus is essential for rhythmicity (2), and recently a self-sustaining and temperature-compensated circadian rhythm in the phosphorylation of KaiC has been reconstituted in vitro in a minimal system containing the three Kai (Japanese for “cycle”) proteins plus ATP (3, 4). In a recent issue of PNAS, Terauchi et al. (5) established that the ATPase activity of KaiC is the fundamental reaction underlying the cyanobacterial circadian oscillation (Fig. 1).

Fig. 1.

The circadian cycle in KaiC phosphorylation and ATPase activity. Dephosphorylated KaiC (C) hexamers interact with KaiA (A), which stimulates autokinase activity. Initially, T432 is phosphorylated (magenta circles), and then S431 is phosphorylated (yellow circles). S431 phosphorylation permits KaiB (B) binding, which blocks the stimulation of KaiC autokinase activity; thereafter, KaiC dephosphorylates at T432 and then at S431. Throughout the cycle, the ATPase activity of KaiC is evident, although ATPase activity is higher (indicated by thicker lines) during the phase of autokinase activity. This ATPase activity is hypothesized to be translated into conformational changes in the KaiC hexamer (indicated by the progressive changes in color) that, in turn, affect ATPase activity. Not all KaiC monomers undergo phosphorylation or dephosphorylation in each cycle, although for simplicity here the cycle is diagrammed as if they do. However, monomer exchange with a free KaiC pool or among hexamers allows for synchronization among the population of KaiC hexamers and the maintenance of robust high-amplitude rhythms in ATPase activity and phosphorylation status. Input components (e.g., CikA and LdpA) and output components (e.g., SasA) also interact with the Kai complexes but are omitted for simplicity. (Modified from ref. 2.)

Genetic screens in S. elongatus for mutations that affect clock function have identified three kai genes. Loss of function of any of these three genes results in arrhythmicity, whereas less severe mutations alter period or phase (2). The kai gene products apparently function only in circadian timekeeping, given that kai mutants show no growth defects relative to wild type when grown in pure culture. Although kaiB and kaiC are cotranscribed in clock-regulated fashion, and kaiA transcription also cycles, rhythmic transcription of the kai genes is not required for clock function, indicating that posttranscriptional events are sufficient for the generation of circadian oscillations. KaiB and KaiC protein levels oscillate, whereas KaiA accumulates constitutively. KaiC exhibits a rhythm in phosphorylation in vivo (6) and possesses both autokinase and autophosphatase activity (7, 8). KaiC autophosphorylation is enhanced by KaiA and is negatively regulated by KaiB (6, 9). The net phosphorylation state of KaiC oscillates with circadian period, and phosphorylation is necessary for sustained rhythmicity (10, 11).

In the presence of ATP, KaiC forms a hexamer, similar to the RecA protein to which it is related (12, 13). The in vitro reconstitution of the circadian rhythm in KaiC phosphorylation has enabled the examination of associations among the Kai proteins both biochemically (14, 15) and microscopically (15). KaiA associates with the KaiC hexamer to promote KaiC phosphorylation; KaiB then associates with the KaiC–KaiA complex and inactivates KaiA, initiating dephosphorylation. A sequential program of phosphorylation of KaiC on two sites, S431 and T432, has been characterized (16). The phosphorylation state of each of these two residues regulates the phosphorylation/dephosphorylation of the other, and the phosphorylation state of S431 regulates KaiB association. Dephosphorylated KaiC is an autokinase and autophosphorylates first on T432 and then on S431. Double phosphorylation converts KaiC to an autophosphatase. T432 dephosphorylation occurs first and is independent of KaiB binding. S431 phosphorylation is necessary for KaiB interaction with KaiC, and this interaction promotes dephosphorylation of S431. As KaiC becomes dephosphorylated, KaiB dissociates, permitting KaiA reactivation and initiating a new round of phosphorylation.

RecA superfamily members exhibit ATPase activity, converting the energy released by ATP hydrolysis into mechanical force (17). Terauchi et al. (5) show that KaiC possesses ATPase activity both on its own and in mixtures with KaiA and KaiB. The daily decrease in ATP concentration in these mixtures is ≈30-fold more than is necessary to account for the uptake of phosphate by KaiC in its autokinase activity, suggesting that the primary consumption of ATP results from ATPase activity. Consistent with this observation, 70% of the basal ATPase activity persists in a truncated form of KaiC that lacks autophosphorylation sites. However, phosphorylation modulates ATPase activity, and experiments with KaiC mutants that mimic the dephosphorylated (S431A/T432A) and doubly phosphorylated (S431D/T432E) forms suggest that phosphorylation decreases ATPase activity. KaiA and KaiB modulate the phosphorylation of KaiC, with KaiA stimulating autophosphorylation and KaiB blocking that stimulation. Additionally, KaiA and KaiB stimulate and depress, respectively, the ATPase activity of KaiC. Unlike the indirect effect of KaiB on autophosphorylation, the decrease of KaiC ATPase activity due to KaiB is independent of KaiA. This modulation by KaiA and KaiB confers a circadian rhythm on KaiC ATPase activity. ATPase and autokinase activities are in phase, and they peak ≈4 h before the peak in KaiC phosphorylation.

Not all KaiC molecules in the in vitro reaction mixture undergo phosphorylation/dephosphorylation in each daily cycle. Moreover, the in vitro reaction is not a homogeneous mixture of one complex that changes sequentially over time but rather is a mixture of complexes in which the proportion of each species oscillates. KaiC is present at all times in free hexamers as well as in hexamers bound to KaiA, and KaiB associates with both types of complexes during the dephosphorylation phase (15). During this phase, monomeric KaiC can also be detected. FRET analysis has shown monomer exchange among KaiC hexamers, which may be critical to maintain the synchrony among hexamers that is needed to sustain high-amplitude oscillations (15).

Temperature compensation is a critical attribute of circadian rhythms and has been established for the in vitro

The kai gene products apparently function only in circadian timekeeping.

KaiC phosphorylation cycle (4). Terauchi et al. (5) show that the ATPase activity of KaiC is only slightly affected by temperature (Q₁₀ ≈ 1.2) in mixtures with KaiA and KaiB, which is consistent with the in vivo rhythm in cyanobacterial gene expression. Interestingly, the ATPase activity of KaiC when incubated without KaiA and KaiB is almost constant across temperatures. Moreover, the ATPase activities of KaiC variants that mimic the dephosphorylated and doubly phosphorylated forms of KaiC are also unaltered by temperature. Thus, temperature compensation of the KaiC ATPase activity is intrinsic to KaiC itself and is independent of its phosphorylation state.

Is the ATPase activity of KaiC the fundamental biochemical activity underlying the cyanobacterial circadian oscillator? If so, one would expect that mutations of KaiC that alter period length would similarly alter ATPase activity. Terauchi et al. (5) tested this prediction by comparing wild type with three short-period and two long-period KaiC variants and confirmed that clock frequencies (the inverse of period length) are directly proportional to ATPase activity of both mutant and wild-type KaiC.

This is the first demonstration of a simple biochemical reaction serving as the basis for a circadian timekeeper. However, “simple” is an inappropriate descriptor. KaiC possesses at least three enzymatic activities—autokinase, autophosphatase, and ATPase—and these three activities mutually influence each other. KaiC assembles into homohexamers and higher-order complexes with KaiA and KaiB, and the latter complexes influence KaiC enzymatic activities. The activity of KaiC (15 ATP per day) is unusually low. Terauchi et al. (5) hypothesize that ATP hydrolysis by KaiC induces conformational change in the KaiC hexamer, which may slow the rate of ATP hydrolysis, and they further speculate that this autoregulation may contribute to temperature compensation. Additional biophysical and structural analyses should enhance understanding of how intra- and intermolecular interactions affect ATPase activity.

The Kai proteins form an oscillator, but for this oscillator to serve as a useful clock, it must (i) accept environmental signals that provide the temporal cues to allow entrainment to local time and (ii) generate useful output signals to regulate physiological processes. In vivo, the Kai protein complexes form higher-order complexes (collectively termed the “periodosome”) with input pathway members such as the His kinases CikA and LdpA and with output pathway members such as the His kinase SasA (2). Rhythmic gene expression in S. elongatus is global and may stem from rhythmic changes in chromosome compaction (18). Such changes occur in the absence of KaiA and KaiC, suggesting that the Kai proteins are not direct effectors. Nonetheless, the rhythmicity of these changes requires both KaiA and KaiC, although the molecular details remain incompletely defined (18).