Evolutionary origins of self-sustained Kai protein circadian oscillators in cyanobacteria

Mukaiyama, Atsushi; Furuike, Yoshihiko; Ito-Miwa, Kumiko; Onoue, Yasuhiro; Horiuchi, Kota; Kondo, Kanta; Yamashita, Eiki; Akiyama, Shuji

doi:10.1038/s41467-025-59908-7

Download PDF

Article
Open access
Published: 15 May 2025

Evolutionary origins of self-sustained Kai protein circadian oscillators in cyanobacteria

Nature Communications volume 16, Article number: 4541 (2025) Cite this article

6185 Accesses
2 Citations
119 Altmetric
Metrics details

Subjects

Abstract

Light–dark cycles affect photosynthetic efficiency in autotrophic cyanobacteria; therefore, determining whether ancient cyanobacteria possessed a self-sustained circadian clock when oxygenic photosynthetic systems were established is an important issue in chronobiology. Here we examine the oscillation of the clock protein KaiC in modern cyanobacteria, as well as the function and structure of ancestral Kai proteins, to determine the evolutionary origin of the self-sustained Kai-protein oscillators. The results show that the oldest double-domain KaiC in ancestral bacteria lacks the factors functionally and structurally essential for rhythmicity. The ancestral Kai proteins have acquired these factors through molecular evolution that occurred around Global Oxidation and Snowball Earth events, and are eventually inherited as a self-sustained circadian oscillator by the most recent common ancestor of cyanobacteria capable of oxygenic photosynthesis. This autonomous Kai protein oscillator is further inherited by most freshwater and marine cyanobacteria present today as an autotrophic basis for time-optimal acquisition and consumption of energy from oxygenic photosynthesis.

Evolution of kaiA, a key circadian gene of cyanobacteria

Article Open access 11 May 2021

From primordial clocks to circadian oscillators

Article Open access 22 March 2023

The cyanobacterial circadian clock

Article Open access 30 June 2025

Introduction

The most recent common ancestor (MRCA) of cyanobacteria emerged approximately 3–2 billion years (Ga) ago^1,2,3,4,5,6 and evolved into the current ecosystem through the Great Oxidation event (GOE) 2.3 Ga ago⁷, at least two Snowball Earth events (SEEs) 2.4 and 0.7 Ga ago^8,9,10, and the Neoproterozoic Oxygenation event (NOE) 0.8–0.6 Ga ago^11,12. Studies of fossils and molecular evolution models suggest that the MRCA of cyanobacteria already possessed primitive oxygenic photosynthetic systems^1,2,4. Because the efficiency of photosynthesis is strongly influenced by light–dark environmental cycles, determining whether primitive cyanobacteria possessed a timekeeping system when photosynthesis became active in the GOE period is important to understanding the physiological origin of circadian clock systems¹³. Primitive biological clocks are also of interest from the perspective of geoscience, and the evolution of the Earth’s rotation period from ancient times to the present remains an issue of debate¹⁴.

Circadian clocks with autonomous 24-h oscillatory, temperature compensatory, and synchronous properties have been identified in various extant organisms¹⁵, including bacteria¹⁶, fungi¹⁷, plants¹⁸, and mammals¹⁹. The circadian clock of cyanobacteria has been studied intensively using the Synechococcus elongatus PCC7942 (Se7942) strain as a model system^20,21. The clock oscillator can be reconstructed in a test tube by incubating the core protein KaiC from the Se7942 strain (KaiC^Se) in the presence of KaiA^Se, KaiB^Se, and adenosine triphosphate (ATP)²². KaiC^Se exhibits rhythms in phosphorylation^23,24 and interactions with KaiA^Se and KaiB^Se^25,26,27,28; however, the frequency of these rhythms is highly correlated with the ATPase activity of KaiC^Se itself^29,30. For example, when the ATPase activity of KaiC^Se doubles as a result of amino acid substitutions, the frequencies of both the in vitro oscillator and the intracellular rhythm also double. This process mediated by KaiC^Se is known as cross-scale causality and is also observed in temperature compensation³¹.

One hypothesis worth testing is whether primitive cyanobacteria utilized a Kai-protein oscillator. However, the concept of a protein-based oscillator is derived almost exclusively from experiments and observations using one extant species, the Se7942 strain. Therefore, whether Kai proteins functioned as circadian oscillators in ancestral cyanobacteria, or even in extant species other than the Se7942 strain, remains unknown.

In this work, we study the function and structure of extant and ancestral Kai-proteins to present a scenario for the evolution of the self-sustained Kai-protein oscillators. According to our estimates, the earliest Kai-protein oscillator emerges in the MRCA of cyanobacteria at approximately 2.2 Ga ago and is able to synchronize with temperature cycles of 18 h, which is shorter than the current rotation period of Earth. The ancient Kai-protein oscillator has been inherited through a long evolutionary process by most extant freshwater cyanobacteria and some marine cyanobacteria living under 24-h environmental cycles.

Results

Feasibility study of reconstructing Kai-protein oscillators in extant species

Duplication of the predecessor of the kaiC gene and their subsequent fusion into a double-domain predecessor occurred between 3.8 and 3.5 Ga ago³². We used the result of large-scale genome analyses of cyanobacteria and other species³³ to construct a phylogenetic tree of extant double-domain KaiC homologs (Fig. 1a and Supplementary Fig. 1). The KaiC tree included four groups: group I was composed mainly of KaiCs from freshwater cyanobacteria, including Se7942 (KaiC^Se, ID: 001); group II consisted of marine cyanobacteria; group III included other bacteria; and group IV was formed by other bacteria including archaea. Most freshwater cyanobacteria have only one kaiC of group I, but others have additional group III or IV kaiC originated from secondary copies or lateral transfer. Thus, these group III and IV KaiCs from freshwater cyanobacteria (gray entries in Fig. 1a) in the protein-based tree can be excluded when identifying groups I–IV and nodes based on species and growing habitant (see Supplementary Discussion).

**Fig. 1: Feasibility studies of in vitro oscillators for extant and ancestral Kai proteins.**

Rhodobacter sphaeroides KaiC (KaiC^Rs, ID: 194), which functions as an hourglass-like timekeeping system³⁴, and Rhodopseudomonas palustris TIE 1 KaiC (KaiC^Rp, ID: 184), which is not self-sustained but enhances fitness in the rhythmic environment³⁵, were classified into group IV in the present KaiC tree. The overall topology and grouping of the protein-based tree of double-domain KaiCs are consistent with those of the 16S rRNA-based phylogenic tree (Supplementary Fig. 2, Supplementary Discussion) as well as with a previous report³².

We used the in vitro rhythm assay²² to test the feasibility of establishing the in vitro oscillator for 41 representative extant species including the Se7942 strain at pH 8.0 and 30 °C (Supplementary Fig. 3). In most KaiCs from extant species belonging to group I, circadian rhythms were observed in the presence of the corresponding species-specific KaiA and KaiB. For example, Cyanothece sp. PCC7822 KaiC (KaiC^CP, ID: 007) cycled between phosphorylated and de-phosphorylated forms with a period of 27.2 h (Fig. 1b) in the presence of KaiA^CP and KaiB^CP, as observed in the Se7942 system (Supplementary Fig. 3a). The same rhythm was observed in Oscillatoriales cyanobacterium JSC 12 KaiC (KaiC^Os, ID: 002) (Fig. 1c). Cases in which an obvious rhythm was detected in a set (A^*B^*C^*) corresponding to species-specific KaiA (KaiA^*), KaiB^*, and KaiC^* are shown in filled circles in Fig. 1a. Regardless of the behaviors of the A^*B^*C^* set, cases in which oscillation was detected when KaiA^* and KaiB^* were replaced by KaiA^Se and KaiB^Se, respectively (A^SeB^SeC^*), are also marked with filled circles in a different column (Fig. 1a). In approximately 80% of group I A^*B^*C^* trials, the rhythm showed a period of 25.2 ± 4.9 h at 30 °C, whereas in approximately 70% of the examined group I A^SeB^SeC^* sets, the period was 23.4 ± 4.5 h (Supplementary Fig. 3a–c).

In species with KaiCs classified into groups II–IV that did not have a homologous gene for kaiA, KaiA^Se was hypothetically used for the in vitro rhythm assay (A^SeB^*C^*). Robust circadian rhythms were observed for most of the examined A^*B^*C^* sets in the first half of the group II entries (approximately 70%, Supplementary Fig. 3d), such as Synechococcus sp. RCC307 (KaiC^SR, ID: 046) (Fig. 1d) and Synechococcus sp. WH 7803 (KaiC^WH, ID: 056) (Fig. 1e). In the other half of group II, which consisted of Prochlorococcus strains lacking species-specific KaiA, most KaiCs could not be expressed or purified. We were able to purify KaiC from Prochlorococcus marinus str. MIT 9211 (ID: 073); however, it showed no rhythm, even in the corresponding A^SeB^SeC^* set (Fig. 1a and Supplementary Fig. 3d).

Entries in Group III are KaiCs from proteobacteria (cyan entries in Fig. 1a) and bacteria (yellow entries in Fig. 1a), and KaiCs as secondary copies of group I kaiC gene (gray entries in Fig. 1a). Proteobacterial KaiC from Rhodospirillum rubrum F11 (KaiC^Rr, ID: 104) in group III (Fig. 1f) failed to show self-sustained oscillation in their A^SeB^*C^* and A^SeB^SeC^* sets. Two gray entries in Group III, KaiCs from Cyanothece sp. ATCC 51142 (KaiC^Cy2, ID: 085) and Synechocystis sp. PCC 6803 (KaiC^Sy3, ID: 093), exhibited self-sustained oscillation in the presence of species-specific KaiB and chimeric KaiA homolog often called KaiA3³⁶ (Fig. 1g and Supplementary Fig. 3e).

The rhythmic fraction of the group IV entries was relatively low (Fig. 1a): 0% for both the A^SeB^*C^* and A^SeB^SeC^* sets of group IV (Supplementary Fig. 3f). For example, KaiC from Flavobacterium johnsoniae UW101 (KaiC^Fj, ID: 212) in group IV (Fig. 1h) did not show self-sustained oscillation in either the A^SeB^*C^* and A^SeB^SeC^* sets.

These results suggest that not all extant KaiC homologs with a double-domain architecture have a circadian clock function, which raises the question of how and when KaiC acquired its function as a self-sustained circadian clock.

In vitro reconstruction of ancestral Kai protein oscillators

We restored the possible amino acid sequences of ancestral KaiCs corresponding to seven evolutionary branching points (α–η) (Fig. 1a and Supplementary Fig. 1, Supplementary Discussion). Ancestral KaiCs consist of 509–560 amino acids and contain an N-terminal CI domain and a C-terminal CII domain (Fig. 2a and Supplementary Fig. 4a). Similar to KaiC^Se, all ancestral KaiCs retain catalytic glutamates essential for ATPase activity in the CI domain and kinase/phosphatase activity in the CII domain, as well as a potential dual phosphorylation site (S431 and T432 in KaiC^Se) in the CII domain (Fig. 2a). A phylogenetic tree of extant KaiB homologs was constructed (Supplementary Fig. 5) and compared with that of the double-domain KaiC homologs (Fig. 1a). According to the similarities in the taxonomic patterns of the belonging species (see coloring by growing habitat), the branches corresponding to the γ, δ, ε, and η points were assigned to the KaiB tree (Supplementary Figs. 5 and 4b), whereas α and β were identified as a point (α/β) with no clear distinction between them. Consistent with a previous study showing that ancestral KaiA appeared at a later time than ancestral KaiB and KaiC^32,37, we were only able to assign three later branches for the γ, δ, and ζ points in our tree of double-domain KaiA homologs (Supplementary Figs. 6 and 4c). Because ancient seawater was neutral (pH ~ 7)³⁸, whereas the current seawater is slightly basic (pH ~ 8), and because some cyanobacteria undergo a change in intracellular pH in response to the pH of their environment³⁹, these ancestral forms of Kai proteins were expressed and purified for an in vitro rhythm assay at pH 7.0 and a temperature of 30 °C.

**Fig. 2: Domain architecture and biochemical properties of ancestral KaiCs.**

A hypothetical oldest set (A^γB^α/βC^α) consisting of KaiA^γ, KaiB^α/β, and KaiC^α did not show any signs of oscillation, and this was also observed in the A^SeB^SeC^α set (Fig. 1i). Similar results were obtained for the second oldest A^γB^α/βC^β and A^SeB^SeC^β sets (Fig. 1j). By contrast, the A^γB^γC^γ and A^SeB^SeC^γ sets exhibited robust oscillations with periods of 23.4 ± 1.5 and 25.8 ± 0.3 h (Fig. 1k), respectively. Similarly, both the ζ sets (A^ζB^γC^ζ: 28.4 ± 1.8 h, A^SeB^SeC^ζ: 28.3 ± 0.1 h, Fig. 1l) and the δ sets (A^δB^δC^δ: 36.8 ± 2.7 h, A^SeB^SeC^δ: 22.2 ± 0.4 h, Fig. 1m) showed rhythmic oscillation. Neither the ε sets nor the η sets (A^γB^εC^ε, A^SeB^SeC^ε, A^γB^ηC^η, nor A^SeB^SeC^η) exhibited any rhythmicity (Fig. 1n, o).

These results indicate that the oldest KaiC^α (62% identity to KaiC^Se, Fig. 2a) and the subsequent KaiC^β (74% identity) lacked oscillatory ability and that this ability was acquired during the transition from KaiC^β to KaiC^γ (89% identity). Consistent with this interpretation, the ATPase activities of KaiC^α and KaiC^β were temperature-dependent (Q₁₀: 2.1–2.2), whereas those of KaiC^γ and KaiC^δ were temperature-compensated (Q₁₀: 0.9–1.2) (Fig. 2b and Supplementary Table 1). Pre–steady-state analysis of the ATPase activity supported this interpretation: KaiC^Se exhibits damped oscillation of ATPase activity even in the absence KaiA^Se and KaiB^Se ²⁹, and similar characteristics were observed in KaiC^γ, KaiC^δ, and KaiC^ζ (Fig. 2c). In addition, an Se7942 reporter strain in which kaiC^Se was replaced by kaiC^γ showed a bioluminescent rhythm under constant-light illumination conditions at 30 °C (Fig. 2d, e), and its period length (32.9 ± 0.7 h) was consistent with that of the in vitro A^γB^γC^γ set at pH 7.6–7.8 and 30 °C (Fig. 2f). The acquisition of oscillatory abilities that occurred in the transformation from KaiC^β to KaiC^γ is not a wild hypothesis confined to the test tube but a possible evolutionary event that could have occurred in primitive cyanobacteria.

We found that the period length of the A^γB^γC^γ set was sensitive to ambient pH. The cycle length, which was 39 h at pH 8.0 and 30 °C, became shorter under acidic conditions, finally stabilizing at 23.4 ± 1.5 h at pH 7.0 and 30 °C (Fig. 2f). The A^SeB^SeC^Se and A^δB^δC^δ sets showed a similar tendency, although they were less pH-dependent than the A^γB^γC^γ set. The results for the A^SeB^SeC^γ set (Figs. 1k and 2f) suggest that the pronounced pH sensitivity of the period length is derived from KaiC^γ itself. Taking into account that the intracellular pH of cyanobacteria changes in response to the external pH³⁹, that the pH of ancient seawater was ~7³⁸, and that the period length is insensitive to ionic strength (Supplementary Fig. 7), the period length at pH 7.0 and 30 °C is reasonably concluded to be a better estimate for ancestral types than that at pH 8.0 and 30 °C. In fact, the A^γB^γC^γ set was entrained by external temperature cycles of 18 to 24 h at pH 7 (Fig. 3a). It worth noting that successful entrainment by the short 18-h cycle was confirmed only for the γ set at pH 7 (Fig. 3a), but not for the A^SeB^SeC^Se set at pH 8 (Fig. 3b)⁴⁰.

**Fig. 3: In vitro temperature entrainment of the Kai-protein oscillators.**

Structural evolution of ancestral KaiC

To explore the structural origin of the autonomous oscillatory properties acquired in the transition from β to γ, we solved the X-ray crystal structures for KaiC^β, KaiC^γ, and KaiC^δ (Supplementary Table 2). The overall structures of KaiC^γ and KaiC^δ were similar to that of KaiC^Se (Fig. 4a), and their main chains could be superimposed with a root-mean-squared deviation (RMSD) of 1.1 Å. The similarity between KaiC^γ and KaiC^Se was confirmed by the superimposition of other parts using one particular CII domain (star in Fig. 4b). By contrast, the structure of KaiC^β differed markedly from that of KaiC^γ (main-chain RMSD of 2.0 Å). When KaiC^β and KaiC^γ were superimposed using one of the CII domains (star in Fig. 4c), the other CI and CII domains did not overlap (main-chain RMSD of 5.3 Å). These findings suggest that ancestral KaiCs acquired the basic architecture for establishing the autonomous oscillator found in KaiC^Se during the transition from KaiC^β to KaiC^γ.

**Fig. 4: Crystal structures of ancestral KaiCs.**

Structural differences between KaiC^β and KaiC^γ were found in both the CI and CII domains. Three major differences were observed in the CI domain with the ATPase activity responsible for period-length determination^29,30. The first is the type of bound nucleotide. KaiC^γ and KaiC^δ were crystallized in the pre-hydrolysis state with a non-hydrolyzable ATP analog (AMP-PNP) bound to the CI domain, as described for KaiC^Se ^29,41 (left in Fig. 4d and Supplementary Fig. 8a); by contrast, KaiC^β was crystallized in the post-hydrolysis state with an adenosine diphosphate (ADP) bound to its CI active site even in the presence of 5 mM AMP–PNP (right in Fig. 4d and Supplementary Fig. 8b).

The second difference is the interface between two adjacent CI domains. In the pre-hydrolysis states of KaiC^γ, KaiC^δ, and KaiC^Se (left in Fig. 4d and Supplementary Fig. 8c), each CI–CI interface is stabilized by the hydrogen bond between the Ser residue side chain (S46^γ, S44^δ, and S48^Se: Fig. 2a) in one CI domain and the carboxyl oxygen atom of the Glu residue (E196^γ, E194^δ, and E198^Se) in the adjacent domain. However, in KaiC^β, the corresponding Ser residue is replaced by Pro (P42^β: Fig. 2a) and the packing of the CI–CI interface becomes loose because of the cleavage of the inter-domain hydrogen bond (right in Fig. 4d), as observed for the post-hydrolysis CI ring of KaiC^Se ^27,29. In addition, the replacement of the amino acid corresponding to S199 in KaiC^γ (S199^γ) by A195 in KaiC^β (A195^β) leads to further selective stabilization of the post-hydrolysis CI conformation of KaiC^β. In the pre-hydrolysis state of KaiC^γ (left in Fig. 4d), the α₈ helix is stabilized by the hydrogen bond between S199^γ and E195^γ (between S197^δ and E193^δ in KaiC^δ, and between S201^Se and E197^Se in KaiC^Se). In KaiC^β, however, cleavage of this hydrogen bond by A195^β enables the partial unfolding of the α₈ helix (right in Fig. 4d), as confirmed by the loose CI–CI interface of the post-hydrolysis CI of KaiC^Se^27,29. These observations support the preferential binding of ADP to the loose CI–CI interface of KaiC^β.

The third difference is the insertion of a residue into the loop enclosing the nucleotide bound in the active site of the CI domain. Compared with KaiC^Se and KaiC^γ (top in Fig. 4e), KaiC^β had a proline residue inserted at the 235th position (P235^β: Fig. 2a) and its structure showed that the residue was located in the turn of the nucleotide enclosure loop (bottom of Fig. 4e). The distorted structure of the ADP bound nearby suggests that, in KaiC^β, P235^β pushes its adenine ring toward the center of the CI ring and inhibits the release of bound ADP. In rhythmic KaiCs such as KaiC^γ, KaiC^δ, and KaiC^Se, no such inhibitory structures for releasing nucleotides exist. Consistent with this structural observation, biochemical experiments showed that ADP bound to KaiC^β was highly stabilized and, in contrast to ADP bound to KaiC^Se and KaiC^γ, was not easily exchanged by AMP–PNP (Fig. 4f).

These observations suggest that the loose CI–CI interface results in the preferential binding of ADP (right in Fig. 4d) and that the elongation of the nucleotide enclosure loop (bottom in Fig. 4e) further prevents the exchange of generated or bound ADP (Fig. 4f). These factors might have caused the consistent occupation of the CI site by ADP in KaiC^β, resulting in a reduction of the ATPase activity (1.8 ± 0.2 ATP d⁻¹ at 30 °C, Fig. 2b) and the loss of the damped oscillatory relaxation in KaiC^β (Fig. 2c). KaiC might have acquired its oscillatory ability by evolving its CI structure to enable nucleotides bound to the CI domain to be exchanged on a reasonable circadian time scale.

The structural differences between KaiC^β and KaiC^γ were not only observed in the CI–CI interface but also in the CII domain and the CI–CII interface. In rhythmic KaiC^γ, a region upstream of the C-terminal tail to which KaiA^γ binds⁴² was clearly resolved as a loop structure (A-loop) with a stabilizing hydrogen bond between Q439^γ and I487^γ (left in Fig. 4g), as observed between Q441^Se and I489^Se in KaiC^Se ^41,43. In KaiC^β, however, the A-loop was destabilized by R436^β, which breaks this hydrogen bond (right in Fig. 4g). The electron density of the phosphor-switch, which changes conformation upon phosphoryl modification of the dual phosphorylation site (S431^Se and T432^Se) in rhythmic KaiC^Se ⁴¹, was clearly identified in KaiC^γ but not in KaiC^β (right in Fig. 4g). In addition, the 389th residue located in the CI–CII interface is Glu in KaiC^β (E389^β), whereas the corresponding residue is Gln in the rhythmic KaiC^γ (Q392^γ) and KaiC^Se (Q394^Se). In KaiC^Se, Q394^Se is critical for establishing the rhythmic coupling between the CI and CII domains by switching the interaction with E214^Se and/or R217^Se (Fig. 2a and left in Fig. 4g), as evidenced by the arrhythmicity of the Q394E variant of KaiC^Se ⁴¹.

Discussion

Conservation of autonomous Kai-protein oscillators in group I

Since the successful in vitro reconstruction of the A^SeB^SeC^Se oscillator in 2005²², the Se7942 strain has become the major target of research on the circadian clock system in cyanobacteria^20,21. So far, there have been no reports of comprehensive feasibility study of establishing in vitro oscillators for bacteria other than the Se7942 strain. For this reason, it was possible to argue that many cyanobacteria actually require transcription-translation feedback⁴⁴ to maintain their rhythm, and that Se7942 is an exceptional case. The present study clearly demonstrates that this is not the case, and that the self-sustained Kai-protein oscillators are conserved in approximately 80% of group I cyanobacteria (Fig. 1a).

At the same time, careful interpretation may be needed for the other entries of group I, which cannot be tested due to unavailability of the Kai-protein samples or failed to show the self-sustained circadian oscillation in vitro (Fig. 1 and Supplementary Fig. 3a–c). For example, it is known that Nostoc (Anabaena) sp. PCC 7120 exhibits in vivo bioluminescence rhythm^45,46, but in this study, we could not conduct the in vitro test because its species-specific KaiC^No (ID: 024 in Fig. 1a) was not expressed even by 12 different trials using different expression hosts, fusion tags, and culturing conditions⁴⁷. Since KaiC from Anabaena sp. 90 (KaiC^An, ID: 035 in Fig. 1a) was rhythmic in vitro in the presence of KaiA^An and KaiB^An (Supplementary Fig. 3b), it is possible that the A^NoB^NoC^No set will also be judged as rhythmic if KaiC^No can be prepared. As another example, KaiC from Fischerella sp. NIES 3754 (KaiC^Fi, ID: 060 in Fig. 1a) showed circadian rhythm in the presence of KaiA^Se and KaiB^Se, but was determined to be arrhythmic in the presence of KaiA^Fi and KaiB^Fi (Supplementary Fig. 3c). One possible explanation for these examples is that a part of the A^*B^*C^* sets in group I becomes stabilized and self-sustained only in special intracellular circumstances that cannot always be fully reproduced in vitro. If we were allowed to take such an extended view of self-sustained protein-based oscillation, the proportion of group I entries with the autonomous Kai-protein oscillators would be even higher than 80%. In this context, although this study does not deny the self-sustainment in the extended sense for the α and β sets in ancient bacteria, but it is clear from our experiments that they are not self-sustained in vitro in the strict sense.

γ node as the time when self-sustained Kai protein oscillators were established

It is important to verify obtained results from the perspective of uncertainties associated with the sequence estimation. The uncertainty in sequence prediction may come from the arbitrary nature of how much gap is allowed when aligning sequences of existing species. We thus performed the same in vitro rhythm assay on another set of ancestral sequences, which was estimated from the alignment using different gap-opening penalty (Supplementary Fig. 9a–c). While both the α and β nodes were arrhythmic, the γ nodes were found to be rhythmic regardless of the way of the sequence prediction (Supplementary Fig. 9d–f). Furthermore, the ancestral lineage of the autonomous oscillations originating from the γ node provides a reasonable explanation for the distribution pattern of the Kai-protein autonomous oscillators in the extant species (groups I–IV), excluding entries derived from secondary copies or lateral transfer (gray entries in Fig. 1a). These evidences suggest the limited effects of the arbitrariness of the sequence estimation on the present view that the self-sustained Kai-protein oscillator appeared at the γ node.

On the other hand, the effects of the uncertainty on the estimation of the cycle length may not be negligible for the following three reasons. First, the length of the cycle is a characteristic of the system composed of three types of ancestral Kai proteins, so the uncertainty is at least potentially additive. Second, the period length of the A^SeB^SeC^Se set can be tuned in the range of 60% to 660% even by one amino acid substitution in a very critical region of KaiC^Se ⁴⁸. Third, the cycle length of the A^γB^γC^γ set was more pH-sensitive than that of the A^SeB^SeC^Se set (Fig. 2f). Taken together, it would be reasonable to conclude that there is uncertainty between the period length of the reconstructed ancestral systems and the true period length of ancient bacteria.

A possible scenario for the evolution of the Kai protein oscillator

The 16S rRNA-based phylogenic tree (Supplementary Fig. 2) shows the presence of a node (γ point) diverging into freshwater and marine cyanobacteria, which can be interpreted as corresponding to the MRCA of cyanobacteria. KaiC^γ is the topologically identical node that branches into the extant KaiCs of freshwater cyanobacteria (group I) and marine cyanobacteria (group II) (Fig. 1a and Supplementary Fig. 1), and the same applies to KaiB^γ (Supplementary Fig. 5) and KaiA^γ (Supplementary Fig. 6). These observations clearly indicate that the A^γB^γC^γ set functioned as the self-sustained circadian oscillator in the MRCA of cyanobacteria capable of oxygenic photosynthesis.

There are many estimates as to when the cyanobacterial MRCA first appeared², which range from 3.5 Ga to 1.5 Ga ago^1,3,4,5,6. Our estimate obtained using 16S rRNA sequences dated an ancestral bacterium at an α/β point at 3.1 (3.5–2.8: upper–lower confidence intervals) Ga ago and the MRCA of cyanobacteria at the γ point at 2.2 (2.8–1.7) Ga ago (Fig. 5 and Supplementary Table 3). An ancestral form of the double-domain KaiA, which is evolutionarily related to KaiA^γ, might have appeared approximately 1.0 Ga ago³²; however, its appearance was subsequently updated to an earlier period, as late as 1.5 Ga ago³⁷.

**Fig. 5: Evolutionary origins and diversification of self-sustained Kai protein circadian oscillators.**

Under our divergence time estimate, the MRCA of cyanobacteria would have emerged around the GOE and first SEE period (Fig. 5). This period is characterized by the enhancement of oxygenic photosynthesis due to changes in the properties of ancient cyanobacteria^1,2,3 or other ancient bacteria^4,5,6, which destroyed the methane greenhouse and triggered global glaciation⁴⁹. The resonance between the endogenous clock and environmental light–dark cycles is the basis for more efficient growth and survival of organisms, including cyanobacteria⁵⁰. Thus, it is reasonable to speculate that the acquisition of the autonomous γ set allowed the MRCA of cyanobacteria to expect the timing of the next sunrise and prepare in advance to initiate oxygenic photosynthesis accordingly. The Earth’s rotation period 2 Ga ago has been suggested to be approximately 4 h shorter than that at present¹⁴; however, depending on the pH of the growing environment, the γ set in the cyanobacterial MRCA might have been sufficiently functional as a circadian oscillator (Figs. 1k and 3).

Regardless of the accuracy of the divergence time estimation, another ancestral bacterium corresponding to the ε/η point appeared at approximately the same time as the MRCA of cyanobacteria (Fig. 5). The ε/η point is assumed to possess not–self-sustained (Fig. 1o) and temperature-dependent (Fig. 2b) KaiC^η, an ancestral form of the hourglass-like KaiC^Rs ³⁴ and not–self-sustained KaiC^Rp ³⁵. The self-sustained protein-based circadian oscillator established in the γ set was subsequently inherited by KaiC^ζ and KaiC^Se and, consequently, by approximately 70% of group I cyanobacteria in freshwater (Fig. 1a and Supplementary Fig. 1). The inheritance from KaiC^γ applied not only to KaiC^ζ but also to KaiC^δ, and the autonomous oscillatory property was eventually passed on to marine cyanobacteria located downstream of KaiC^δ. However, our results indicate that it was not inherited by the Prochlorococcus group located downstream of KaiC^δ or that the inherited traits have changed by today. Given that the molecular evolution of oxygenic photosynthesis pre-dates the cyanobacterial MRCA by >1 Ga^2,4, the unique evolution of the ancestral Kai proteins during the emergence of the cyanobacterial MRCA provided the autotrophic basis for the time-optimal acquisition and consumption of energy, later leading to their massive proliferation across various habitats on Earth.

Methods

Construction of protein-based phylogenic trees

The amino acid sequences of KaiC homologs were obtained from a dataset described previously³³. From a total of 1194 entries, duplicate entries that had identical sequences but were from different species/strains were removed, resulting in 151 entries of 480–570 amino acid residues with >30% sequence identity to KaiC^Se. These were extracted to form a dataset for analyzing KaiC homologs with double-domain structures (Supplementary Fig. 1).

The MEGA X software (version 10.1.7)⁵¹ was used for phylogenic analyses. The 151 sequences were arranged using pairwise/multiple alignments with a gap-opening penalty of 7.5. A preliminary phylogenetic tree was constructed using the maximum likelihood method, and 100 bootstrap replications were generated. To confirm the stability of the resultant phylogenetic pattern, different alignment results were analyzed using gap-opening penalties of 2.5, 5.0, 10.0, 15.0, and 20.0. In the preliminary phylogenetic trees, double-domain homologs of KaiC were classified into four groups as follows: cyanobacteria in freshwater (group I: 58 entries), cyanobacteria in seawater (group II: 26 entries), other bacteria (group III: 23 entries), other bacteria including archaea (group IV: 42 entries), and outgroup (2 entries). The final phylogenetic tree of double-domain KaiC homologs (Fig. 1a) was obtained using a gap-opening penalty of 7.5 through 500 bootstrap replications. A detailed description of the entries in each group is provided in Supplementary Discussion and Supplementary Fig. 1.

Phylogenetic analyses of KaiA and KaiB homologs were performed as described for KaiC using a dataset from a previous study³³. For KaiB, we extracted 193 entries of 80–120 amino acid residues with >30% sequence identity to KaiB^Se (Supplementary Fig. 5). For KaiA, extraction was performed with 50–350 amino acid residues with >30% sequence identity to KaiA^Se, resulting in 151 entries of double-domain homologs of KaiA (Supplementary Fig. 6). The species used to construct the phylogenetic trees for KaiA and KaiB were not necessarily identical to those used for the KaiC tree. The final phylogenetic trees for KaiB (Supplementary Fig. 5) and KaiA (Supplementary Fig. 6) were obtained using a gap-opening penalty of 7.5 through 500 bootstrap replications. See Supplementary Information for detailed descriptions of the entries in each group.

Construction of 16S rRNA-based phylogenic trees

The 16S rRNA sequences of extant species with double-domain KaiC homologs were obtained from KEGG⁵². The 16S rRNA-based tree shown in Supplementary Fig. 2 was constructed using MEGA X software (version 10.1.7) with 108 entries of the 151 KaiC homologs used for construction of the protein-based tree (Fig. 1a).

Restoration of amino acid sequences of ancestral KaiCs

The amino acid sequences corresponding to seven nodes on the phylogenic tree of KaiC (Fig. 1a) were inferred on the basis of the maximum likelihood estimation using the amino acid substitution model of LG with Freqs. (+F) and gamma-distributed substitution rates. The oldest ancestor (KaiC^α) was detected directly below a node branching into groups I–IV, Pyrococcus horikoshii OT3 KaiC (KaiC^Hori), and Pyrodictium delaneyi KaiC (KaiC^Dela).

There were two branches emerging from KaiC^α: one ended in groups I–III, and the other ended in group IV. A node connecting to groups I–III (KaiC^β) showed high bootstrap values (98%–99%) irrespective of the gap-opening penalties (7.5–15). The KaiC^β node branched into groups I–II and group III. A node branching into groups I and II (KaiC^γ) also emerged with high bootstrap values (99%) irrespective of the gap-opening penalties used. Although the common ancestor of group II (KaiC^δ) showed a high bootstrap value (31%–97%), those of group I (KaiC^ζ) and group III (KaiC^ε) showed relatively lower values (20%–51% and 53%–63%, respectively). The common ancestor of group IV (KaiC^η) was detected with the bootstrap vale (98%) as high as those of KaiC^β and KaiC^γ.

Restoration of amino acid sequences of ancestral KaiAs and KaiBs

The amino acid sequences corresponding to the nodes on the phylogenic trees of KaiA and KaiB were estimated as described for ancestral KaiCs. In the KaiB tree constructed with 193 entries (Supplementary Fig. 5), we assigned one node for each, KaiB^γ and KaiB^δ, on the basis of similarities in the taxonomic patterns of the corresponding entries (see coloring by growth habitat in Supplementary Figs. 1 and 5). The KaiB^δ node was located downstream of KaiB^γ. Because a node upstream of KaiB^γ was a common ancestor for all entries, it was termed KaiB^α/β. A node that did not contain downstream archaeal entries was termed KaiB^ε.

Only three later branching points (γ, δ, and ζ) could be assigned in the double-domain KaiA tree consisting of 151 entries (Supplementary Fig. 6), which is consistent with the later appearance of ancestral KaiA than ancestral KaiB³⁷. KaiA^γ was reasonably allocated to a node corresponding to a common ancestor of the double-domain KaiA from both freshwater cyanobacteria and seawater cyanobacteria. Similarly, KaiA^ζ and KaiA^δ were assigned to nodes corresponding to ancestors of cyanobacteria in freshwater and seawater, respectively.

Divergence time estimation

The divergence times of the α/β, γ, δ, ε/η and ζ points in the 16S rRNA-based tree (Supplementary Fig. 2) were estimated using MEGA X software (version 10.1.7)⁵¹. According to the TimeTree⁵³ estimate of the divergence time between Synechococcus and Flavobacterium, a calibration point was established at the α/β point as normal distribution with a median at 3.134 Ga and a deviation of 0.5 Ga.

Cloning of extant and ancestral kaiA, kaiB, and kaiC

The DNA sequences of extant and ancestral Kai-proteins were synthesized by Eurofins genomics with codon optimization for expression in Escherichia coli (E. coli). Unless otherwise stated, the synthesized genes were cloned into the pGEX6P-1 vector (Cytiva) with BamHI and NotI cleaving sites to overexpress GST-tagged Kai proteins. For ancestral KaiC^α, KaiC^β, KaiC^ε, and KaiC^η, the synthesized genes were cloned into the pET-3a vector (Novagen) with NdeI and BamHI cleaving sites using nucleotides encoding hexagonal histidine (His) at the 3′ end. For extant KaiCs from Flavobacterium johnsoniae UW101 and Rhodopseudomonas palustris TIE-1 strains, genes were cloned into the pASK-IBA5plus vector (IBA GmbH) with the BsaI cleaving site to overexpress Strep-tagged KaiCs because of the poor yields associated with the GST-tagged form.

Expression and purification of extant and ancestral Kai-proteins

Extant and ancestral KaiCs were expressed in E. coli BL21 cells and then purified using affinity chromatography, ion exchange chromatography, and size exclusion chromatography⁴⁷ according to the type of fusion tags used. Irrespective of the extant or ancestral types, KaiA and KaiB were purified following the methods used for KaiA^Se and KaiB^Se, respectively⁵⁴.

In vitro rhythm assay

For the in vitro rhythm assay of the extant species, KaiA, KaiB, and KaiC were mixed in Tris-buffer (20 mM Tris, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT, and 0.5 mM EDTA, pH 8.0) containing 1 mM ATP, and the time-evolution of the phosphorylation of KaiC was monitored. The potential oscillation of the respective KaiCs was assessed in the presence of various concentrations of KaiA (0.01, 0.02, 0.04, and 0.08 mg/mL) and a fixed concentration of KaiB (0.04 mg/mL) according to previous observations that the amount of KaiA in the system markedly affects the amplitude or stability of the rhythm^31,48,55. KaiCs that showed any slight sign of oscillation under either condition exhibited a stable rhythm in the presence of moderate concentrations of KaiA (0.02–0.04 mg/mL) (Supplementary Fig. 3). For the assay of KaiC^Sy3 from Synechocystis sp. PCC 6803 (ID: 093 in Fig. 1a), a chimeric KaiA homolog (protein ID: WP_010873803.1, ~22% identity to KaiA^Se) that is not listed in the present KaiA tree but has recently been identified was used together with KaiB^Sy3 (ID: 179 in Supplementary Fig. 5d)³⁶. Similarly, for the assay of KaiC^Cy2 from Cyanothece sp. ATCC 51142 (ID: 085 in Fig. 1a), a KaiA homolog (protein ID: WP_009543244.1, ~17% identity to KaiA^Se) and KaiB^Cy2 (ID: 086 in Supplementary Fig. 5d) were used³⁶.

For the in vitro rhythm assay of the ancestral forms, KaiA, KaiB, and KaiC were mixed in Tris-buffer (20 mM Tris, 150 mM NaCl, 5 mM MgCl₂, 1 mM DTT, and 0.5 mM EDTA, pH 7.0) containing 1 mM ATP at final Kai-protein concentrations listed in Table 1.

Table 1 Kai-protein concentrations for in vitro rhythm assays

Full size table

ATPase measurements

ATPase measurements were performed by monitoring the accumulation of ADP⁴⁷ using an ultrahigh-performance liquid chromatography (UHPLC) system (Chromaster, Hitachi). Ancestral KaiCs and KaiC^Se were incubated in a Tris-buffer containing 1 mM ATP at pH 7.0 and pH 8.0, respectively. The ATPase activity of KaiC was defined as the number of hydrolyzed ATP molecules into ADP molecules per KaiC monomer per unit time.

Pre-steady-state ATPase assay

For the pre-steady-state ATPase assay²⁹, an apo-KaiC monomer was first prepared using a phospho-buffer (50 mM NaH₂PO₄, 150 mM NaCl, 50 mM l-arginine, 50 mM l-glutamic acid, 5 mM MgCl₂, and 1 mM DTT, pH 7.8)⁵⁶. Reassembly of KaiC hexamers was induced by mixing the apo-KaiC monomer with the phospho-buffer containing ATP, and the resultant mixture containing 1 mM ATP was immediately subjected to UHPLC analysis for ATPase measurements.

Entrainment by external temperature cycles

Kai-protein oscillators were exposed to temperature cycles⁴⁰ of one half of the period length at 40 °C and the other half at 30 °C using an automated sampling device⁵⁷. For each of the A^γB^γC^γ (pH 7.0) and A^SeB^SeC^Se (pH 8.0) sets, the Kai-protein oscillators were reconstituted at 0, 6, and 12 h intervals and incubated at 30 °C for 36, 30, and 24 h, respectively. The pre-incubated samples were subjected to five temperature cycles of 18 h (9 h at 40 °C and 9 h at 30 °C: 9H9L), 20 h (10 h at 40 °C and 10 h at 30 °C: 10H10L), or 24 h (12 h at 40 °C and 12 h at 30 °C: 12H12L) to evaluate the entrainability.

pH and Q ₁₀ values

The pH of the buffers listed in the main text and in Supplementary Information was measured at 25 °C during buffer preparation. The pKa value of Tris is slightly temperature-dependent⁵⁸. We confirmed that Tris buffers with pH 7.0 when prepared at 25 °C change to pH 6.8 at 30 °C and pH 6.6 at 40 °C, and Tris buffers with pH 8.0 at 25 °C change to pH 7.8 at 30 °C and pH 7.6 at 40 °C, as actual measured values at the corresponding temperatures. Therefore, Q₁₀ values calculated from data obtained at 30 °C and 40 °C using the corresponding Tris buffers could include additional pH-dependent effects, if any, of approximately –0.2 pH units.

In vivo rhythm assays

Synechococcus elongatus PCC 7942 was used as the background strain for this assay. To conduct the in vivo bioluminescence assay and analysis¹⁶, cyanobacterial cells carrying the kaiBC-reporter cassette were cultured at 30 °C for 3 days on BG-11 solid medium under constant light (LL) at 100 μmol m⁻² s⁻¹. After one or three 17-h light/17-h dark treatments, the cells were cultured at 30 °C under LL (45 μmol m⁻² s⁻¹) at an LED daylight lamp, and bioluminescence profiles were monitored using a photomultiplier tube detector.

X-ray structure analyses of KaiC^β, KaiC^γ, and KaiC^δ

Crystals of KaiC^β, KaiC^γ, and KaiC^δ were obtained using the sitting-drop vapor diffusion method with 5 mM AMP-PNP. KaiC^β (3 mg/mL) was mixed with a solution of 18 mM acetic acid and 65 mM sodium malonate (pH 5.7). The cryo-protectant treatment was performed using 30% (w/v) PEG4K. KaiC^γ (3 mg/mL) was crystallized in a solution of 0.94 M sodium formate and 1.2 M sodium malonate (pH 8.5), which are both cryo-protectants. KaiC^δ (3 mg/mL) was added to 0.1 M Tris-HCl (pH 7.0), 1 M KCl, and 1.14 M sodium malonate.

KaiC^β, KaiC^γ, and KaiC^δ crystals were frozen using liquid nitrogen and placed under the cryo-stream of the SPring-8 BL44XU beamline (Harima, Japan). The crystals were exposed to X-ray irradiation at a wavelength of 0.9 Å, and diffraction images were collected by using an EIGER X 16 M detector (Dectris) and the semi-automatic data reduction software, KAMO⁵⁹. Diffraction intensity data were processed using XDS (version Jun 30, 2023)⁶⁰. The initial models for KaiC^β, KaiC^γ, and KaiC^δ were obtained using the molecular replacement method with MOLREP (version 11.9.02)⁶¹ and the coordinates of the KaiC^Se hexamer (PDB ID: 2GBL) as a template. Refinement was achieved using REFMAC5 (version 5.8.0415)⁶². The model was constructed by using COOT (version 0.9.8.7)⁶³. PyMOL (Schrödinger) (version 2.5.2) was used for the graphic presentation of the refined models.

The KaiC^β, KaiC^γ, and KaiC^δ crystals had lattices of P2₁ (a = 93 Å, b = 386 Å, c = 108 Å, β = 113°; two hexamers in an asymmetric unit), P1 (a = 91 Å, b = 110 Å, c = 167 Å, α = 78°, β = 87°, γ = 82°; single hexamer in an asymmetric unit), and P2₁ (a = 92 Å, b = 175 Å, c = 95 Å, β = 97°; single hexamer in an asymmetric unit), respectively.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data needed to evaluate the conclusions in the paper are provided in the paper, the Supplementary Information, and Source Data file. Source data for all the plots, 16S rRNA sequences, and extant Kai-protein sequences are provided with this paper. The atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession codes 8ZN5 (KaiC^β), 8ZN6 (KaiC^γ), and 8ZN7 (KaiC^δ). Accession codes of analyzed published sequence data are available in the Source data file. Source data are provided with this paper.

References

Schirrmeister, B. E., Gugger, M. & Donoghue, P. C. Cyanobacteria and the great oxidation event: evidence from genes and fossils. Palaeontology 58, 769–785 (2015).
Article PubMed PubMed Central Google Scholar
Oliver, T. et al. The evolution and evolvability of photosystem II. Annu. Rev. Plant Biol. 74, 225–257 (2023).
Article CAS PubMed Google Scholar
Garcia-Pichel, F. et al. Timing the evolutionary advent of cyanobacteria and the later great oxidation event using gene phylogenies of a sunscreen. Mbio 10, e00561-19 (2019).
Nishihara, A., Tsukatani, Y., Azai, C. & Nobu, M. K. Illuminating the coevolution of photosynthesis and bacteria. Proc. Natl. Acad. Sci. USA 121, e2322120121 (2024).
Article CAS PubMed PubMed Central Google Scholar
Shih, P. M., Hemp, J., Ward, L. M., Matzke, N. J. & Fischer, W. W. Crown group oxyphotobacteria postdate the rise of oxygen. Geobiology 15, 19–29 (2017).
Article CAS PubMed Google Scholar
Betts, H. C. et al. Integrated genomic and fossil evidence illuminates life’s early evolution and eukaryote origin. Nat. Ecol. Evol. 2, 1556–1562 (2018).
Article PubMed PubMed Central Google Scholar
Bekker, A. et al. Dating the rise of atmospheric oxygen. Nature 427, 117–120 (2004).
Article ADS CAS PubMed Google Scholar
Kirschvink, J. L. Late proterozoic low-latitude global glaciation: the snowball Earth. in The Proterozoic Biosphere: A Multidisciplinary Study (eds Schopf, J. W., Klein, C. & Maris, D.) (Cambridge University Press, 1992).
Hoffman, P. F. et al. Snowball Earth climate dynamics and cryogenian geology-geobiology. Sci. Adv. 3, e1600983 (2017).
Article ADS PubMed PubMed Central Google Scholar
Macdonald, F. A. et al. Calibrating the cryogenian. Science 327, 1241–1243 (2010).
Article ADS CAS PubMed Google Scholar
Scott, C. et al. Tracing the stepwise oxygenation of the Proterozoic ocean. Nature 452, 456–459 (2008).
Article ADS CAS PubMed Google Scholar
Och, L. M. & Shields-Zhou, G. A. The neoproterozoic oxygenation event: environmental perturbations and biogeochemical cycling. Earth Sci. Rev. 110, 26–57 (2012).
Article ADS CAS Google Scholar
Edgar, R. S. et al. Peroxiredoxins are conserved markers of circadian rhythms. Nature 485, 459–464 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Wu, H. B., Murray, N., Menou, K., Lee, C. & Leconte, J. Why the day is 24 hours long: the history of Earth’s atmospheric thermal tide, composition, and mean temperature. Sci. Adv. 9, eadd2499 (2023).
Article PubMed PubMed Central Google Scholar
Dunlap, J. C. Molecular bases for circadian clocks. Cell 96, 271–290 (1999).
Article CAS PubMed Google Scholar
Ishiura, M. et al. Expression of a gene cluster as a circadian feedback process in cyanobacteria. Science 281, 1519–1523 (1998).
Article ADS CAS PubMed Google Scholar
McClung, C. R., Fox, B. A. & Dunlap, J. C. The Neurospora clock gene frequency shares a sequence element with the Drosophila clock gene period. Nature 339, 558–562 (1989).
Article ADS CAS PubMed Google Scholar
Strayer, C. et al. Cloning of the Arabidopsis clock gene TOC1, an autoregulatory response regulator homolog. Science 289, 768–771 (2000).
Article ADS CAS PubMed Google Scholar
Gekakis, N. et al. Role of the CLOCK protein in the mammalian circadian mechanism. Science 280, 1564–1569 (1998).
Article ADS CAS PubMed Google Scholar
Akiyama, S. Structural and dynamic aspects of protein clocks: how can they be so slow and stable? Cell Mol. Life Sci. 69, 2147–2160 (2012).
Article CAS PubMed PubMed Central Google Scholar
Swan, J. A., Golden, S. S., LiWang, A. & Partch, C. L. Structure, function, and mechanism of the core circadian clock in cyanobacteria. J. Biol. Chem. 293, 5026–5034 (2018).
Article CAS PubMed PubMed Central Google Scholar
Nakajima, M. et al. Reconstitution of circadian oscillation of cyanobacterial KaiC phosphorylation in vitro. Science 308, 414–415 (2005).
Article ADS CAS PubMed Google Scholar
Nishiwaki, T. et al. A sequential program of dual phosphorylation of KaiC as a basis for circadian rhythm in cyanobacteria. EMBO J. 26, 4029–4037 (2007).
Article CAS PubMed PubMed Central Google Scholar
Rust, M. J., Markson, J. S., Lane, W. S., Fisher, D. S. & O’Shea, E. K. Ordered phosphorylation governs oscillation of a three-protein circadian clock. Science 318, 809–812 (2007).
Article ADS CAS PubMed PubMed Central Google Scholar
Akiyama, S., Nohara, A., Ito, K. & Maéda, Y. Assembly and disassembly dynamics of the cyanobacterial periodosome. Mol. Cell 29, 703–716 (2008).
Article CAS PubMed Google Scholar
Kageyama, H. et al. Cyanobacterial circadian pacemaker: Kai protein complex dynamics in the KaiC phosphorylation cycle in vitro. Mol. Cell 23, 161–171 (2006).
Article CAS PubMed Google Scholar
Tseng, R. et al. Structural basis of the day-night transition in a bacterial circadian clock. Science 355, 1174–1180 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Chavan, A. G. et al. Reconstitution of an intact clock reveals mechanisms of circadian timekeeping. Science 374, eabd4453 (2021).
Article CAS PubMed PubMed Central Google Scholar
Abe, J. et al. Atomic-scale origins of slowness in the cyanobacterial circadian clock. Science 349, 312–316 (2015).
Article ADS CAS PubMed Google Scholar
Terauchi, K. et al. ATPase activity of KaiC determines the basic timing for circadian clock of cyanobacteria. Proc. Natl. Acad. Sci. USA 104, 16377–16381 (2007).
Article ADS CAS PubMed PubMed Central Google Scholar
Furuike, Y. et al. Cross-scale analysis of temperature compensation in the cyanobacterial circadian clock system. Commun. Phys. 5, 75 (2022).
Dvornyk, V., Vinogradova, O. & Nevo, E. Origin and evolution of circadian clock genes in prokaryotes. Proc. Natl. Acad. Sci. USA 100, 2495–2500 (2003).
Article ADS CAS PubMed PubMed Central Google Scholar
Schmelling, N. M. et al. Minimal tool set for a prokaryotic circadian clock. BMC Evol. Biol. 17, 169 (2017).
Pitsawong, W. et al. From primordial clocks to circadian oscillators. Nature 616, 183–189 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Ma, P., Mori, T., Zhao, C., Thiel, T. & Johnson, C. H. Evolution of KaiC-dependent timekeepers: a proto-circadian timing mechanism confers adaptive fitness in the purple bacterium rhodopseudomonas palustris. PLoS Genet. 12, e1005922 (2016).
Article PubMed PubMed Central Google Scholar
Köbler, C. et al. Two KaiABC systems control circadian oscillations in one cyanobacterium. Nat. Commun. 15, 7674 (2024).
Article PubMed PubMed Central Google Scholar
Dvornyk, V. & Mei, Q. M. Evolution of kaiA, a key circadian gene of cyanobacteria. Sci. Rep. 11, 9995 (2021).
Halevy, I. & Bachan, A. The geologic history of seawater pH. Science 355, 1069–1071 (2017).
Article ADS CAS PubMed Google Scholar
Belkin, S., Mehlhorn, R. J. & Packer, L. Proton gradients in intact cyanobacteria. Plant Physiol. 84, 25–30 (1987).
Article CAS PubMed PubMed Central Google Scholar
Yoshida, T., Murayama, Y., Ito, H., Kageyama, H. & Kondo, T. Nonparametric entrainment of the in vitro circadian phosphorylation rhythm of cyanobacterial KaiC by temperature cycle. Proc. Natl. Acad. Sci. USA 106, 1648–1653 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Furuike, Y. et al. Elucidation of master allostery essential for circadian clock oscillation in cyanobacteria. Sci. Adv. 8, eabm8990 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kim, Y. I., Dong, G., Carruthers, C. W. Jr., Golden, S. S. & LiWang, A. The day/night switch in KaiC, a central oscillator component of the circadian clock of cyanobacteria. Proc. Natl. Acad. Sci. USA 105, 12825–12830 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Pattanayek, R. et al. Visualizing a circadian clock protein: crystal structure of KaiC and functional insights. Mol. Cell 15, 375–388 (2004).
Article CAS PubMed Google Scholar
Young, M. W. & Kay, S. A. Time zones: a comparative genetics of circadian clocks. Nat. Rev. Genet. 2, 702–715 (2001).
Article CAS PubMed Google Scholar
Kushige, H. et al. Genome-wide and heterocyst-specific circadian gene expression in the filamentous Cyanobacterium Anabaena sp. strain PCC 7120. J. Bacteriol. 195, 1276–1284 (2013).
Article CAS PubMed PubMed Central Google Scholar
Arbel-Goren, R. et al. Robust, coherent, and synchronized circadian clock-controlled oscillations along Anabaena filaments. Elife 10, e64348 (2021).
Article CAS PubMed PubMed Central Google Scholar
Mukaiyama, A., Ouyang, D. Y., Furuike, Y. & Akiyama, S. KaiC from a cyanobacterium sp. PCC 7428 retains functional and structural properties required as the core of circadian clock system. Int. J. Biol. Macromol. 131, 67–73 (2019).
Article CAS PubMed Google Scholar
Ito-Miwa, K., Furuike, Y., Akiyama, S. & Kondo, T. Tuning the circadian period of cyanobacteria up to 6.6 days by the single amino acid substitutions in KaiC. Proc. Natl. Acad. Sci. USA 117, 20926–20931 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Kopp, R. E., Kirschvink, J. L., Hilburn, I. A. & Nash, C. Z. The Paleoproterozoic snowball Earth: a climate disaster triggered by the evolution of oxygenic photosynthesis. Proc. Natl. Acad. Sci. USA 102, 11131–11136 (2005).
Article ADS CAS PubMed PubMed Central Google Scholar
Woelfle, M. A., Ouyang, Y., Phanvijhitsiri, K. & Johnson, C. H. The adaptive value of circadian clocks: an experimental assessment in cyanobacteria. Curr. Biol. 14, 1481–1486 (2004).
Article CAS PubMed Google Scholar
Kumar, S., Stecher, G., Li, M., Knyaz, C. & Tamura, K. MEGA X: molecular evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, 1547–1549 (2018).
Article CAS PubMed PubMed Central Google Scholar
Kanehisa, M. & Goto, S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28, 27–30 (2000).
Article CAS PubMed PubMed Central Google Scholar
Kumar, S. et al. TimeTree 5: an expanded resource for species divergence times. Mol. Biol. Evol. 39, msac174 (2022).
Nishiwaki, T. et al. Role of KaiC phosphorylation in the circadian clock system of PCC 7942. Proc. Natl. Acad. Sci. USA 101, 13927–13932 (2004).
Article ADS CAS PubMed PubMed Central Google Scholar
Nakajima, M., Ito, H. & Kondo, T. In vitro regulation of circadian phosphorylation rhythm of cyanobacterial clock protein KaiC by KaiA and KaiB. FEBS Lett. 584, 898–902 (2010).
Article CAS PubMed Google Scholar
Mukaiyama, A., Osako, M., Hikima, T., Kondo, T. & Akiyama, S. A protocol for preparing nucleotide-free KaiC monomer. Biophysics 11, 79–84 (2015).
Article CAS PubMed PubMed Central Google Scholar
Furuike, Y., Abe, J., Mukaiyama, A. & Akiyama, S. Accelerating in vitro studies on circadian clock systems using an automated sampling device. Biophys. Physicobiol. 13, 235–241 (2016).
Article CAS PubMed PubMed Central Google Scholar
Good, N. E. et al. Hydrogen ion buffers for biological research. Biochemistry 5, 467–477 (1966).
Article CAS PubMed Google Scholar
Yamashita, K., Hirata, K. & Yamamoto, M. KAMO: towards automated data processing for microcrystals. Acta Crystallogr. D Struct. Biol. 74, 441–449 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Vagin, A. & Teplyakov, A. MOLREP: an automated program for molecular replacement. J. Appl. Crystallogr. 30, 1022–1025 (1997).
Article ADS CAS Google Scholar
Murshudov, G. N. et al. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr. D Biol. Crystallogr. 67, 355–367 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Buick, R., Des Marais, D. J. & Knoll, A. H. Stable isotopic compositions of carbonates from the Mesoproterozoic Bangemall Group, northwestern Australia. Chem. Geol. 123, 153–171 (1995).
Article ADS CAS PubMed Google Scholar

Download references

Acknowledgements

Diffraction data were collected at beamline BL44XU at the SPring-8 facility under the proposals 2021A6700, 2021B6602, 2023A6500, 2023B6842, and 2024A6930. This research was partly supported by the Platform Project for Supporting Drug Discovery and Life Science Research (BINDS) from AMED under grant number JP21am0101072 (support number 0583). This study was supported by Grants-in-Aid for Scientific Research (22H04984 to S.A., A.M., and Y.F., 17H06165 to S.A. and Y.F., 24H02301 to S.A., 22K15051 to Y.F., 19K16061 to Y.F., and 22K06175 to A.M.), Takeda Science Foundation (to S.A.), and Toyoaki Scholarship Foundation (to S.A.).

Author information

These authors contributed equally: Atsushi Mukaiyama, Yoshihiko Furuike.

Authors and Affiliations

Department of Bioscience and Biotechnology, Fukui Prefectural University, Eiheiji, 910-1195, Japan
Atsushi Mukaiyama
Research Center of Integrative Molecular Systems (CIMoS), Institute for Molecular Science, National Institutes of Natural Sciences, Okazaki, 444−8585, Japan
Yoshihiko Furuike, Yasuhiro Onoue, Kota Horiuchi, Kanta Kondo & Shuji Akiyama
Molecular Science Program, Graduate University for Advanced Studies, SOKENDAI, Okazaki, 444-8585, Japan
Yoshihiko Furuike, Kota Horiuchi, Kanta Kondo & Shuji Akiyama
Graduate School of Science and Institute for Advanced Research, Nagoya University, Nagoya, 464-8602, Japan
Kumiko Ito-Miwa
Institute for Protein Research, The University of Osaka, Suita, 565-0871, Japan
Eiki Yamashita

Authors

Atsushi Mukaiyama
View author publications
Search author on:PubMed Google Scholar
Yoshihiko Furuike
View author publications
Search author on:PubMed Google Scholar
Kumiko Ito-Miwa
View author publications
Search author on:PubMed Google Scholar
Yasuhiro Onoue
View author publications
Search author on:PubMed Google Scholar
Kota Horiuchi
View author publications
Search author on:PubMed Google Scholar
Kanta Kondo
View author publications
Search author on:PubMed Google Scholar
Eiki Yamashita
View author publications
Search author on:PubMed Google Scholar
Shuji Akiyama
View author publications
Search author on:PubMed Google Scholar

Contributions

A.M., Y.F., and S.A. conceived the project. S.A., A.M., and Y.F. supervised the project. A.M. and Y.F. performed most of the biochemical experiments and analysis. A.M., Y.F., Y.O., and K.H. performed rhythm and ATPase assays. Y.O. performed the nucleotide-exchange experiments. K.K. and K.H. performed temperature-entrainment experiments. Y.F. collected the diffraction data and analyzed it with input from E.Y. K.I.-M. performed in vivo experiments. S.A., A.M., and Y.F. drafted the manuscript. All the authors read and commented on the manuscript.

Corresponding authors

Correspondence to Atsushi Mukaiyama, Yoshihiko Furuike or Shuji Akiyama.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Transparent Peer Review file

Reporting Summary

Source data

Source Data Files

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Mukaiyama, A., Furuike, Y., Ito-Miwa, K. et al. Evolutionary origins of self-sustained Kai protein circadian oscillators in cyanobacteria. Nat Commun 16, 4541 (2025). https://doi.org/10.1038/s41467-025-59908-7

Download citation

Received: 16 August 2024
Accepted: 02 May 2025
Published: 15 May 2025
DOI: https://doi.org/10.1038/s41467-025-59908-7

This article is cited by

Effect of pH on the cyanobacterial circadian oscillator in vitro
- Kumiko Ito-Miwa
- Yasuhiro Onoue
- Kazuki Terauchi
Communications Biology (2025)