Skip to main content
  • Loading metrics

Evolution of KaiC-Dependent Timekeepers: A Proto-circadian Timing Mechanism Confers Adaptive Fitness in the Purple Bacterium Rhodopseudomonas palustris

  • Peijun Ma,

    Affiliation Department of Biological Sciences, Vanderbilt University, Nashville, Tennessee, United States of America

  • Tetsuya Mori,

    Affiliation Department of Biological Sciences, Vanderbilt University, Nashville, Tennessee, United States of America

  • Chi Zhao,

    Affiliation Department of Biological Sciences, Vanderbilt University, Nashville, Tennessee, United States of America

  • Teresa Thiel,

    Affiliation Department of Biology, University of Missouri-St. Louis, St. Louis, Missouri, United States of America

  • Carl Hirschie Johnson

    Affiliation Department of Biological Sciences, Vanderbilt University, Nashville, Tennessee, United States of America


Circadian (daily) rhythms are a fundamental and ubiquitous property of eukaryotic organisms. However, cyanobacteria are the only prokaryotic group for which bona fide circadian properties have been persuasively documented, even though homologs of the cyanobacterial kaiABC central clock genes are distributed widely among Eubacteria and Archaea. We report the purple non-sulfur bacterium Rhodopseudomonas palustris (that harbors homologs of kaiB and kaiC) only poorly sustains rhythmicity in constant conditions–a defining characteristic of circadian rhythms. Moreover, the biochemical characteristics of the Rhodopseudomonas homolog of the KaiC protein in vivo and in vitro are different from those of cyanobacterial KaiC. Nevertheless, R. palustris cells exhibit adaptive kaiC-dependent growth enhancement in 24-h cyclic environments, but not under non-natural constant conditions. Therefore, our data indicate that Rhodopseudomonas does not have a classical circadian rhythm, but a novel timekeeping mechanism that does not sustain itself in constant conditions. These results question the adaptive value of self-sustained oscillatory capability for daily timekeepers and establish new criteria for circadian-like systems that are based on adaptive properties (i.e., fitness enhancement in rhythmic environments), rather than upon observations of persisting rhythms in constant conditions. We propose that the Rhodopseudomonas system is a "proto" circadian timekeeper, as in an ancestral system that is based on KaiC and KaiB proteins and includes some, but not necessarily all, of the canonical properties of circadian clocks. These data indicate reasonable intermediate steps by which bona fide circadian systems evolved in simple organisms.

Author Summary

Circadian (daily) rhythms are an adaptation of organisms to the dramatic changes in the environment over day and night, and they are defined by properties that include rhythmic persistence in constant conditions and temperature compensation. These "biological clocks" are practically ubiquitous in eukaryotes, but in prokaryotes they have been persuasively documented only among cyanobacterial species, where the kaiABC genes regulate a biochemical timekeeping mechanism. The kaiB and kaiC genes are found widely among other prokaryotic species, but non-cyanobacterial prokaryotes usually have poor persistence of rhythmic function in constant conditions, and are therefore considered to not have circadian systems. We have examined a purple non-sulfur bacterium (Rhodopseudomonas palustris) that harbors kaiB and kaiC homologs. Its daily rhythm of nitrogen fixation does not continue in constant conditions, and biochemical properties of the KaiC protein that are known to be crucial for circadian rhythmicity in cyanobacteria are altered in Rhodopseudomonas' KaiC. Nevertheless, we find that adaptive growth of the cells is dependent upon kaiC in rhythmic environments, but not in non-rhythmic environments. We propose that the Rhodopseudomonas system is an ancestral "proto-circadian" timekeeper that is based on KaiC and KaiB, and that elucidates intermediate steps by which biological clocks evolved in simple organisms.


Circadian (daily) rhythms are a fundamental property of cellular and organismal organization that regulate gene expression, metabolic activity, sleep, behavior, and many other biological processes to provide a fitness advantage [14]. These rhythms have three defining and "canonical" characteristics. The first is sustained persistence in constant conditions with a period of ~ 24 h. The second is the establishment of a particular phase relationship (aka “phase angle”) between the endogenous biological clock and the daily environmental cycle by entrainment. The final characteristic is that the period of the free-running rhythm is “temperature compensated,” i.e. that the period is almost the same at different constant ambient temperatures (Q10~1.0). These three properties define circadian rhythms, not any particular biochemical mechanism. The fascination of this phenomenon is to explain how a biochemical oscillator could have evolved that has such a long time constant (~24 h) which is both precise and temperature compensated.

Biologists have known for decades that circadian systems are practically ubiquitous among eukaryotic organisms, but these rhythms were originally thought to be absent from prokaryotes [5]. Beginning about 25 years ago, however, this dogma was overturned as research on prokaryotic cyanobacteria persuasively demonstrated that nitrogen fixation and gene expression exhibited bona fide circadian characteristics [57]. Once the floodgates of “eukaryotic-centric thinking” were opened, a torrent of information about circadian timekeeping in cyanobacteria was released, including that gene expression and chromosomal topology is globally regulated [813] in addition to circadian rhythms of nitrogen fixation, metabolism, and cell division [57,1417]. Moreover, cyanobacteria have become the model system that most persuasively illustrates the adaptive value of circadian organization in 24-h rhythmic environments [24]. Three essential core clock genes have been identified in the cyanobacterium Synechococcus elongatus (kaiA, kaiB, and kaiC [18]), and the proteins they encode can reconstitute a circadian rhythm in vitro [19]. Salient properties of the central KaiABC pacemaker are a sustained rhythm of KaiC phosphorylation, a temperature-compensated ATPase activity of KaiC, and rhythmic interactions among the KaiA/KaiB/KaiC proteins [1923].

Despite the findings in cyanobacteria, there are essentially no other prokaryotes for which bona fide circadian properties have been documented. Homologs of the essential kaiB and kaiC genes are distributed widely among eubacterial and Archaeal species [24,25], but a third essential gene–kaiA–is found only in the cyanobacteria. Perhaps a KaiB/KaiC “hourglass timer” is the more common timekeeping mechanism among prokaryotes, and the cyanobacterial KaiA/KaiB/KaiC circadian trio is an exceptional case? Even among the cyanobacteria, the kaiA gene is missing or truncated in the important cyanobacterial genus Prochlorococcus, which apparently has daily rhythms but not a sustained circadian rhythmicity in constant conditions [26,27]. One group has speculated that the cyanobacterial circadian clock has "de-evolved" to an hourglass timer in kaiA-free Prochlorococcus [28]. Despite the lack of documented circadian phenomenon among non-cyanobacterial prokaryotes, bacteria that live in daily light/dark cycles–especially those performing photosynthesis–would also be expected to benefit from a timing mechanism to adapt to these daily changes. Therefore, it is reasonable to predict that there are other bacterial species that possess a daily timekeeper, whether it be a circadian clock or an hourglass timer.

The purple non-sulfur Rhodopseudomonas palustris is a gram-negative bacterium that is unrelated to cyanobacteria (R. palustris is an alpha-proteobacterium) but harbors homologs of kaiB and kaiC (herein, kaiBRp & kaiCRp). R. palustris is a metabolic "acrobat," as it can vault between four different modes of metabolism: photo-heterotrophic, photo-autotrophic, chemo-heterotrophic, and chemo-autotrophic [29]. Therefore it displays considerable metabolic versatility as it can grow aerobically or anaerobically, heterotropically or autotropically. Under anaerobic or microaerobic conditions in which R. palustris uses light to photosynthesize, it can also fix nitrogen. When it is in its photoautotrophic or photoheterotrophic modes, R. palustris relies upon sunlight for energy and therefore optimal adaptation to the daily light/dark cycle could involve the evolution of a daily timekeeper (as in the case of cyanobacteria), potentially to anticipate dawn and dusk so as to maximize the regulation of photosynthetic capacity. We therefore tested the possibility that R. palustris exhibits daily timekeeping properties that may or may not be circadian. We found that a kaiC-dependent system allows R. palustris to adapt optimally to daily light/dark cycles under photoheterotrophic conditions, but the characteristics of the daily patterns observed are not the same as those found in cyanobacteria. These data suggest that the timekeeping mechanism in R. palustris constitutes a “proto-circadian” system. By "proto-circadian," we mean a phenomenon that shows some–but perhaps not all–canonical properties of a bona fide circadian rhythm. This could be because the phenomenon is an evolutionary precursor under selection on a trajectory towards a complete circadian system, or because the ecological niche of the organism does not provide a selective advantage for some of the defining circadian properties and therefore they have not arisen in the organism's daily timekeeper. In the case of the R. palustris KaiB/C system, we suggest that it is an evolutionary precursor to the more robust and resilient KaiA/B/C system found in cyanobacteria.


Purple non-sulfur bacteria such as R. palustris are globally distributed in soil and water environments and they frequently cohabit niches with cyanobacteria [29,30,31,32], with the proviso that R. palustris prospers in multiple metabolic modes [29] whereas cyanobacteria prefer aerobic conditions. Therefore, the daily environmental conditions experienced by purple non-sulfur bacteria and cyanobacteria are similar, which might lead to similar selective pressures to evolve a daily timekeeping mechanism. As mentioned above, R. palustris also shares homologs of the clock genes kaiB and kaiC with cyanobacteria, although the order of kaiBRp and kaiCRp is reversed as compared with the S. elongatus kaiABCSe trio, as shown in S1A Fig (however, the order is the same as for the Synechocystis kaiCB locus 3, S2 Fig). Moreover, the motif structure of kaiCRp is practically the same as for kaiCSe; both have a double domain CI/CII sequence with Walker A/B motifs, DXXG motifs, and catalytic EE residues in essentially the same locations (S1B Fig). The major differences between kaiCRp vs. kaiCSe are that (1) the demonstrated phosphorylation sites in kaiCSe are TST, whereas those same residues in kaiCRp are TSS, and (2) the C-terminal region in kaiCRp is longer, which may relate to the fact that this is the region of KaiCSe that KaiA binds to stimulate KaiCSe autophosphorylation [21,22]–since there is no KaiA in R. palustris, it is logical that this region may differ between kaiCRp and kaiCSe.

Temporal patterns of nitrogen fixation in R. palustris

These ecological and genetic similarities with cyanobacteria encouraged us to investigate whether R. palustris cells exhibit circadian properties. Under phototrophic conditions, the energy metabolism of R. palustris is at least partially dependent upon photosynthesis and therefore we expected that a circadian oscillator that allows the anticipation of the daily light cycle would be most likely to be expressed under phototrophic conditions. Inspired by the pioneers who first discovered circadian rhythms in cyanobacteria by monitoring nitrogen fixation [5,6,33], we measured the nitrogen fixation rate of R. palustris (strain TIE-1 [13]) under photoheterotrophic conditions in daily light/dark cycles (LD 12:12 = 12 h light/12 h dark). In R. palustris, nitrogen fixation is light-dependent and inhibited by oxygen. As shown in Fig 1A and 1C, under anaerobic conditions, nitrogen fixation peaked in the middle of the day in wild-type (WT) R. palustris at 23°C and 30°C. In WT, replicate cultures reproducibly peaked at the same phase in LD at both 23°C and 30°C, indicating at least some degree of entrainment and temperature compensation (the data of Fig 1 are replotted in S3 Fig with all three replicate cultures averaged together). We compared this daily pattern in WT with that of a strain in which the kaiCRp gene was knocked out (the RCKO strain) and of a kaiCRp-knockout strain in which the kaiCRp gene was replaced at an ectopic site (the RCKO+kaiCRp strain). In RCKO, there is no expression of kaiCRp transcripts, but the level of kaiBRp transcript is equivalent to that of WT (Fig 2A), and the daily profile of nitrogen fixation is altered (Fig 1B and 1D). The nitrogen fixation of RCKO peaked in the day phase, but the particular phase adopted by three replicate cultures is highly variable (Fig 1B and 1D), unlike the stable phase relationship for WT replicates (Fig 1A and 1C, also see S3 Fig). In the RCKO+kaiCRp strain, expression of KaiCRp was rescued (Fig 2B) as well as a consistently phased nitrogen fixation pattern in LD12:12 (Fig 2C).

Fig 1. Daily patterns of nitrogen fixation in WT vs. RCKO strains.

A and C, nitrogen fixation activities of the wild-type R. palustris (WT) at 30°C and 23°C. B and D, nitrogen fixation activities of the kaiCRp-deletion strain (RCKO) at 30°C. The three traces represent three individual cultures under anaerobic conditions that were tested at different phases of LD 12:12 after growth for ~2 weeks in LD 12:12. The black and white bars underneath represent the light conditions. Nitrogen fixation rates were calculated based on the amount of C2H4 (nmol) produced by 1010 cells per hour. These two-cycle LD experiments were repeated twice, each time with 3 replicate cultures; one representative experiment is shown in this figure (one-cycle LD assays were conducted in five independent experiments, each time with phasing data equivalent to those shown in this figure). The data of Fig 1 are replotted in S3 Fig with all three replicate cultures averaged together (complete time series data appear in S4 Table).

Fig 2. Characterizations of the RCKO and RCKO+kaiCRp strains.

A. The mRNA levels of kaiBRp and kaiCRp were quantified by quantitative PCR from the WT (red) and RCKO (black) strains. The mRNA levels of clpX were included as an internal control. B. immunoblot by anti-FLAG antibody to confirm the expression of FLAG-KaiCRp in the rescued strain (RCKO+kaiCRp). C. nitrogen fixation of the RCKO+kaiCRp strain under LD12:12 cycles at 30°C. Data using the HA-KaiCRp rescued strain are equivalent. The three traces represent three individual cultures. These experiments were repeated twice, each time with 3 replicate cultures; one representative experiment is shown in this figure (complete time series data for panel C appear in S5 Table).

As explained above, a defining characteristic of bona fide circadian rhythms is their persistence in constant conditions, and the nitrogen fixation rhythm of cyanobacteria exhibits a robust rhythm in constant light (LL [5,6,33]). However, R. palustris did not exhibit a sustained rhythm when transferred from LD to LL. As shown in Fig 3, WT R. palustris cultures exhibited a robust rhythm in LD, but upon transfer to LL, high-amplitude rhythmicity was not sustained at 23°C or 30°C. Fig 3 illustrates the averaged results of multiple cultures, but the same effect is clear when individual cultures are plotted separately (S4 Fig). There was some suggestion of a low-amplitude rhythmicity in LL at both 23°C and 30°C with a free-running period (FRP) of approximately 22 h, but Cosinor analyses indicated that these periodicities were not significant (R2 is only 0.20 at 30°C, and a clearly non-significant 0.03 at 23°C; S5 Fig). Therefore, the daily rhythm of nitrogen fixation in WT R. palustris was clear in LD, but was either non-existent or highly damped in constant LL conditions, and therefore did not conform to the first defining property of circadian rhythmicity, namely robust persistence in constant conditions. In the RCKO strain, nitrogen fixation was not rhythmic by eye (Fig 3 and S4 Fig) or by Cosinor analysis (S5 Fig).

Fig 3. Lack of robust persistence of the nitrogen fixation rhythm in LL after transfer from LD.

A, nitrogen fixation activities of the WT (red) and RCKO (green) strains at 30°C. B, nitrogen fixation activities of the WT (red) and RCKO (green) strains at 23°C. Data is plotted as the mean of three individual cultures. Data points are mean +/- S.D. Black-white bars indicate the light/dark conditions: black is dark, white is light. These experiments were repeated five times (five times for WT, and five times for RCKO), each time with three replicate cultures; one representative experiment is shown in this figure (the three replicate cultures of this experiment are plotted separately in S4 Fig, and complete time series data appear in S6 Table).

Molecular correlates: KaiCRp ATPase activity and phosphorylation

In the cyanobacterium S. elongatus, KaiCSe undergoes robust rhythms in LL and LD of phosphorylation [21,3436] and often rhythms of abundance as well [37]. Using the RCKO+kaiCR strain (Fig 2), we assessed KaiCRp phosphorylation and abundance in vivo using an anti-HA antibody to the HA-tagged KaiCRp (Fig 4) after electrophoretic separation by Phos-Tag PAGE into different phosphoforms of KaiCRp (Fig 4). We generated different phosphoforms of KaiCRp by incubating the protein with P-32 labeled ATP overnight at 4°C and at 30°C (Fig 4A and 4B, lanes 2–5). Although SDS-PAGE separates the different phosphoforms of KaiCSe [19,20,23,34,36](Fig 4A, lane 1), it does not separate the different phosphoforms of KaiCRp (Fig 4A, lanes 2–5). However, SDS-PAGE including Phos-Tag reagent can separate phosphorylated KaiCRp into at least three differentially migrating variants, labeled P1, P2, and P3 in Fig 4B (non-phosphorylated KaiCRp is labeled NP). When the Phos-Tag PAGE is used to analyze extracts from R. palustris cells in LD and LL, we found a clear daily rhythm of KaiCRp phosphorylation in vivo in LD, with a trough of KaiCRp phosphorylation at the end of the night phase. However, this daily rhythm either disappears or is highly damped in LL, which is the same pattern seen with nitrogen fixation (Fig 3). Moreover, there is not an obvious 24-h cycle of KaiC abundance in LL. Therefore, KaiCRp exhibits a daily–but not a circadian–rhythm of phosphorylation status in vivo.

Fig 4. Phosphorylation patterns of KaiCRp in LD and LL.

A & B. Electrophoresis analyses of 32P-labeled KaiCSe and KaiCRp. Lane 1 is 32P-labeled KaiCSe. Lanes 2–5 are purified KaiCRp that was mixed with [γ-32P]ATP. Samples for lanes 2 and 4 were immediately denatured after the addition of [γ-32P]ATP by mixing with SDS-PAGE sample buffer (time zero samples, lanes 2 and 4). Samples for lanes 3 and 5 were incubated with [γ-32P]ATP for 24 h at either 4°C (lane 3) or 30°C (lane 5) prior to SDS denaturation and inactivation. The samples were subjected to either regular SDS-PAGE (A) or phosphate-affinity SDS-PAGE with Phos-Tag (B). In both panels, the left portion is a Coomassie-Blue stained gel (CBB) and the right portion is the autoradiogram of P-32 radioactivity. In panel B, three bands of 32P-labeled KaiCRp are indicated as P1, P2, P3 and an unlabeled KaiCRp band is indicated as NP (non-phosphorylated). C & D. The strain in which KaiCRp has been tagged with HA (RCKO + HA-kaiCRp) was grown under LD 12:12 and then transferred to LL. Cells were collected every 6 h. Total protein extracts were separated on SDS-polyacrylamide gels with Phos-tag and analyzed by immunoblotting using an anti-HA antibody. The specificity of the anti-HA antibody against HA-tagged KaiCRp was confirmed with extracts from the RCKO cells. C. immunoblot in which the phospho- (P1, P2, P3) and nonphospho- (NP) forms of KaiCRp are indicated. D. quantification of phosphorylation states of HA-KaiCRp from the immunoblot in panel C. “ZT” = Zeitgeber Time in LD, where ZT0 is lights-on and ZT12 is lights-off (complete time series data for panel D appear in S7 Table).

Another key property of KaiCSe is its ability to hydrolyze ATP by a reaction that is very slow (15 ATP molecules hydrolyzed per day per KaiCSe monomer at 30°C) and temperature-compensated (Q10 ~ 1.0 [38,39]). This ATPase reaction has been proposed to be the core biochemical reaction underlying circadian periodicity in cyanobacteria [38]. The ATPase activity of KaiCSe is significantly reduced when the four catalytic glutamate (E) residues (S1B Fig) are replaced with glutamine (Q) residues [39]. While KaiCRp hydrolyzes ATP in vitro and this activity is decreased when the four catalytic EE residues are replaced with glutamine residues (EQ1EQ2), the rate of ATP hydrolysis by native KaiCRp was >100X higher than that for KaiCSe at 30°C (Fig 5A). The ATPase activity of KaiCRp co-migrates with the abundance profile of the purified protein upon gel filtration chromatography, indicating that the measured activities are those of the KaiC proteins and not of a putative contaminant (S6 Fig). Unlike the case for KaiCSe, the rate of ATP hydrolysis by KaiCRp was not temperature compensated (Q10 = 1.9, Fig 5B).

Fig 5. ATPase activity of KaiCRp and KaiCRp-EQ1EQ2 in vitro at 30°C.

A. Purified native KaiCRp and the KaiCRp-EQ1EQ2 mutant (EQ1EQ2) were incubated with1 mM ATP for 0.5, 1, 2, 4, 12, and 24 h at 30°C. The amount of ATP hydrolyzed was quantified by HPLC. Dark blue, buffer only; red, KaiCRp; light blue, KaiCRp-EQ1EQ2; green, KaiCSe. B. ATPase hydrolysis rates of native KaiCRp at different temperatures. Q10 was calculated based on these rates. These experiments were repeated twice; one representative experiment is shown in this figure.

KaiC-dependent growth in LD versus LL

Despite the lack of a robust free-running rhythm in LL (Fig 3), KaiCRp clearly affected the timing of temporal events in LD (Fig 1, S3 Fig). We reasoned that consistent temporal patterning in LD (the typical environmental condition) might enhance fitness even if the rhythmicity does not persist robustly in LL (which is, after all, a special or non-environmental condition). For microbial organisms, growth rates are an excellent gauge of fitness. The growth rates of all three strains (WT, RCKO, and RCKO+kaiCRp) were equivalent in the non-rhythmic LL condition under which the cells grew photoheterotrophically (Fig 6A and 6C). Therefore, knocking out the kaiCRp gene had no detectable effect upon the growth of cells under noncyclic photoheterotrophic conditions.

Fig 6. Growth rate of R. palustris is kaiCRp-dependent in cyclic 24-h conditions.

Cell densities (OD600) of the WT (red), RCKO (black) and RCKO-kaiCRp (green) strains were measured every two days under (A) LL conditions at 30°C; (B) LD 12:12 cycles at 30°C; (C) LL conditions at 23°C; (D) LD 12:12 cycles at 23°C; (E) LD 1:1 cycles at 30°C; (F) LD 12:12 cycles at 30°C in YPA medium. All cultures were grown anaerobically. These experiments were repeated three times, and each time there were three replicate cultures. In this figure, the data from the three separate experiments were pooled and plotted as the mean +/- S.D. (therefore, each datum is the average of n = 9).

However, the result was completely different under temporally cycling photoheterotrophic conditions. In a light/dark cycle of 12 h light, 12 h darkness (LD 12:12), the RCKO strain grew significantly slower at either 23°C or 30°C (Fig 6B and 6D). This diminution of growth rate was clearly kaiCRp–dependent because the replacement of kaiCRp to the RCKO strain rescued the growth defect in LD 12:12 (RCKO-kaiCRp strain, Fig 6B and 6D), but had no detectable impact under LL conditions (Fig 6A and 6C). Under chemoheterotrophic growth conditions in YPA medium, however, WT and RCKO grew at the same rate in LD (Fig 6F). Consequently, knocking out the kaiCRp gene did not affect growth under chemoheterotrophic conditions in LD.

It might be argued that the kaiCRp–dependent growth effect under photoheterotrophic conditions in LD may be due to the reduced total amount of light in LD vs. LL, because LD 12:12 has half the amount of light as LL over time. We therefore tested LD 1:1 (1 h light/1h dark) cycles that also have half the amount of light as LL, but in which the cycle has no 24-h information. Under this condition, there was no difference in the growth rate between WT and RCKO (Fig 6E). Consequently, functionally KaiCRp had an important effect on photoheterotrophic growth/fitness in 24-h cyclic environments (LD 12:12), but not in environments without 24-h cycles (LL or LD 1:1) or under chemoheterotrophic conditions when energy metabolism is not dependent upon light (Fig 6).


A proto-circadian system in Rhodopseudomonas that is adaptive

Numerous bioinformatic studies have noted the wide distribution of kaiB and kaiC homologs among many prokaryotic groups, including Archaea [24,25]. Based upon the discovery of circadian rhythms in S. elongatus and its essential kaiABC clock gene cluster [7,18], these bioinformatic studies speculated that many prokaryotic species might harbor circadian systems [24,25]. Nevertheless, attempts to discover bona fide circadian rhythms in prokaryotes outside the cyanobacteria have been frustratingly unsuccessful [5]. An example is a study of “rhythmic” gene expression in the purple bacterium Rhodobacter sphaeroides, where noisy circa-20 h oscillations of a reporter were observed under aerobic conditions, but neither temperature compensation nor entrainment was reported [40]. Therefore, even though that genome includes a kaiC-B-PAS/HK operon (as in R. palustris, S1B Fig), it is impossible to conclude whether Rhodobacter has a circadian system or not. In many other prokaryotic species that have been tested, observing consistent persisting circa-24-h rhythms has been elusive, let alone temperature compensation or a rigorous demonstration of entrainment [5].

As reported herein, R. palustris exhibits some kaiC-dependent properties that are consistent with a circadian-type system, including a phase angle in LD that is consistent at different temperatures (implying temperature compensation), and slowing and suppressing of circadian rhythms in cyanobacteria when KaiCRp is expressed in the heterologous S. elongatus (S7 Fig). Nevertheless, other observations from R. palustris are unlike those found in S. elongatus, such as (1) the nitrogen fixation rhythm in constant conditions is either not consistently rhythmic or is highly damped (not well sustained), (2) there is no 24-h rhythm of KaiCRp phosphorylation status or abundance in constant LL conditions, and (3) the ATPase activity of KaiCRp is significantly higher than that of KaiCSe and not temperature compensated. It might be argued that the nitrogen fixation rhythm is not well coupled to a putative central pacemaker in R. palustris (unlike the case in cyanobacteria [5,6,33]) and a different parameter would be a better indicator, but this suggestion misses the point that monitoring one output rhythm after another [5] may be a fruitless endeavor and that the standards established for circadian-type systems in eukaryotes may not be the optimal criteria when studying prokaryotes.

We have taken a different approach with R. palustris that focused upon criteria related to adaptive significance rather than canonical properties. The circadian system in S. elongatus has been rigorously established to provide a fitness advantage in 24-h rhythmic environments, but not in constant non-selective environments [24]. We found that growth of R. palustris under photoheterotrophic conditions is clearly dependent upon kaiC in 24-h cyclic environments, but not in constant (non-24-h cycles) environments (Fig 6). Therefore, despite the lack of a robustly sustained nitrogen fixation rhythm (Fig 3, S4 Fig) or rhythms of KaiC phosphorylation (Fig 4) in constant conditions, R. palustris has a kaiC-dependent timekeeping system that has at least some circadian-like properties. Moreover, the anticipation of dawn by the nitrogen fixation pattern in LD (i.e., the fact that nitrogen fixation activity is rising at the end of the night before lights-on) that is obvious in Fig 1C and Fig 3A and 3B implies a timekeeper rather than a direct light/dark response. It is tempting to conclude that the daily phasing of metabolic events (exemplified by nitrogen fixation, Fig 1) among the WT, RCKO, and RCKO+kaiCRp strains is responsible for the differential growth rates (Fig 6). However, there may be other kaiCRp-dependent factors that are responsible for the differential growth rates–but if so, these factors must be dependent upon an interaction between KaiCRp and a rhythmic environment with a 24-h periodicity. A fully operational circadian system may underlie the phenomena we observed in R. palustris. However, the lack of a robustly persisting rhythm in constant conditions coupled with properties of KaiCRp that clearly differ from those of KaiCSe persuade us to be more cautious. We propose instead that the timekeeping system in R. palustris is more likely to be a “proto-circadian oscillator” in which natural selection has not yet selected for a sufficiently resilient clock that it sustains itself in constant conditions.

Evolution of clocks and new criteria for circadian-type timekeepers in prokaryotes

If the proto-clock in R. palustris is adaptive despite the absence of a persisting free-running rhythm, why has evolution repeatedly selected for circadian rhythms in eukaryotes that persist robustly and sustain free-running rhythms in constant conditions? There is presently no clear answer why biological oscillators that persist in constant conditions (a condition that is rare in nature–restricted to caves or to polar regions in summer or winter) were selected over hourglass timers to be the biological timekeepers for organisms in natural light/dark (LD) cycles. Roenneberg and Merrow have stated, “Evolution has shaped circadian clocks in a cyclic world; temporal constancy of environmental qualities must have been an extremely rare exception. It is therefore the mechanism of entrainment that has evolved and not sustained rhythmicity in constant conditions. The latter is theoretically not even essential for a functional entrained circadian system” [41]. The case of R. palustris is an excellent example that supports the foregoing statement. What might be the selective forces that would encourage a self-sustained circadian clock? Some modelers have suggested that environmental noise (e.g., light and temperature fluctuations) strongly select for the complexity of a circadian timekeeper that sustains itself after being transferred to unnatural constant conditions [42]. The environmental noise used in these simulations included daily fluctuations in light intensity & temperature as well as annual changes in photoperiod (however, organisms that live near the equator and experience practically no annual cycle of photoperiod nevertheless evolved self-sustained circadian oscillators).

The comparison among Rhodopseudomonas, Synechococcus, and Prochlorococcus might be enlightening in this regard. Cyanobacteria such as S. elongatus are the only prokaryotic group to have the kaiA gene. The known function of the KaiA protein in cyanobacteria is to bind to the C-terminal tentacles of KaiC and stimulate the phosphorylation of KaiCSe [21,22,34,36]. Consistent with the absence of KaiA from R. palustris, the KaiA-binding C-terminal tentacles of KaiCSe are significantly different from the corresponding regions of KaiCRp (an obvious difference from S1B Fig is that the C-terminal regions of KaiCRp are longer than those of KaiCSe, but the sequence of residues is also different). In addition, KaiA is sequestered during the KaiCSe dephosphorylation phase, thereby inactivating its activity; this feature of KaiA regulation assists the synchronization and robustness of the cyanobacterial KaiABC oscillation [21,22,36]. Perhaps the inclusion of KaiA is an evolutionary innovation in cyanobacteria that enhances the sustainment of kaiB/kaiC-based oscillations, resulting in the exquisite rhythms of cyanobacteria in LL [7,21,22]. We can infer this conclusion both because the kaiB/kaiC-based system without kaiA of R. palustris is not well sustained (this report), but also because the widespread marine cyanobacterium Prochlorococcus “lost” the kaiA gene in its evolution (non-functional vestiges of kaiA remain in the Prochlorococcus genome [24,25,28]; Prochlorococcus also does not sustain robust rhythms in constant conditions [2628]. Given that the complete kaiABC system is adaptive in the soil/freshwater cyanobacterium S. elongatus [24], why might selection have relaxed for maintaining kaiA in Prochlorococcus? The explanation may lie in the relatively consistent marine environment of Prochlorococcus. As suggested by other researchers [2628], in the interests of genomic streamlining Prochlorococcus could afford to lose the sustained kaiABC system in favor of a reduced kaiBC system because the daily rhythmicity of its marine environment is less noisy than that of most terrestrial/aquatic environments.

Whether those evolutionary speculations be true or not, our study establishes a new set of criteria for circadian-type timekeeping systems in prokaryotes that harbor kaiC and/or kaiB homologs based on adaptiveness rather than upon expression of rhythmic parametrics. In particular, the experiments depicted in Fig 6 demonstrate kai-dependent growth under selective (24-h cycles) versus non-selective conditions. This approach can be applied to many prokaryotic species. A particularly interesting example would be the application of this criterion to prokaryotic species that harbor only kaiC homologs to address the question of whether the ancestral role of kaiC (a RecA/DnaB-related gene [43]) has no rhythmicity function, and the evolutionary addition of kaiB conferred the adaptive timekeeping properties observed in R. palustris and Prochlorococcus.

Materials and Methods

Bacterial strains and culture conditions

All bacterial strains used in this study are listed in S1 Table. E.coli was grown in Luria-Bertani (LB) broth at 37°C with shaking, and when relevant, gentamicin (50 μg/ml) was added. For photoheterotrophic growth, R. palustris strains were grown in Freshwater-Base (FW) medium [30] supplied with 20 mM sodium acetate, 50 mM sodium bicarbonate, 20 mM MOPS (pH 7.2), 1 mM potassium phosphate, 1 mM sodium sulfate, multivitamin solution and trace elements solution. The medium was aliquoted into sealed serum bottles and the headspace was flushed with N2 gas for 20 min before inoculating cells. N2 gas was used as the sole nitrogen source to ensure nitrogenase expression. The cultures were maintained at 30°C (or 23°C as specified) and illuminated with cool-white fluorescent lamps (40–50 μE m−2s−1) with gentle shaking. For aerobic chemoheterotrophic growth, R. palustris strains were grown in YPA medium containing 0.3% yeast extract and 0.3% peptone (= YPA medium, [30]) at 30°C with gentle shaking, and when relevant, gentamicin (400 μg/ml) was added. Cyanobacteria (S. elongatus reporter strain AMC149 [7]) was grown in BG-11 medium [44] at 30°C, and illuminated by cool-white fluorescence lamps (40–50 μE m−2s−1) with air bubbling. The medium of AMC149 was supplemented with spectinomycin (25 μg/ml). For growth on solid media, 1.5% agar was included with the LB, FW or BG-11 medium and appropriate antibiotics.

Construction of the kaiCRp deletion strain

All plasmids and primers used in constructing strains are listed in S2 Table and S3 Table. The kaiCRp deletion strain (RCKO) was constructed in R. palustris TIE-1 strain by overlap extension PCR and conjugation [45,46]. To delete the kaiCRp gene, the 1 kb upstream region and the 1 kb downstream region of the kaiCRp ORF were cloned from the genomic DNA of R. palustris TIE-1 strain and fused by overlap extension PCR. The 2-kb DNA fragment was ligated with the suicide vector pJQ-200KS [47]. The resulting plasmid, pJQ-200KS-RCKO, was transformed into R. palustris TIE-1 strain by conjugation with E. coli S17-1 [45,46,48]. The integration of the plasmid in either the upstream or the downstream region of the kaiCRp locus was selected by gentamicin resistance and screened by PCR. Following the selection, the integrants were grown in non-selective YPA medium for several generations and then plated on YPA agar medium with 10% sucrose [45] to induce double recombination. Among the survivors of the sucrose-YP medium, the double recombinants were selected by PCR screening. The deletion of the kaiCRp gene was confirmed by sequencing and Q-RT-PCR.

Construction of the FLAG-kaiCRp strain and HA- kaiCRp strain

To complement the kaiCRp gene deletion, a FLAG-tagged or HA-tagged kaiCRp gene was restored to the genome of RCKO strain in the region surrounding the glmUSX-recG locus [45]. These tags allow us to confirm the expression of KaiCRp in the rescued strain with the appropriate anti-tag antibody. To construct the insertion plasmid pJQ200KS-insert, the 1 kb upstream and the 1 kb downstream regions of the glmUSX-recG locus of R. palustris TIE-1 were cloned and fused by overlap extension PCR. A NcoI site was incorporated in the middle to allow the insertion of genes of interest [45]. The resulting DNA fragment was ligated with pJQ200KS by SphI and SmaI, as described by Bose and Newman [45,47]. To include the native promoter region of the kaiCRp gene in the insertion plasmid, a 469-bp region upstream of the kaiCRp gene was cloned and ligated to the NcoI site of pJQ-200KS-Insert, resulting in pJQ200KS-Insert-PkaiCRp. A NdeI-XbaI site was incorporated downstream of the promoter region to allow the insertion of genes of interest. The KaiCRp gene was cloned from genomic DNA of the R. palustris TIE-1 strain, while a FLAG tag or HA tag was fused to its N-terminus. This DNA fragment was ligated with the plasmid pJQ200KS-Insert-PkaiCRp by NdeI and XbaI where the FLAG- kaiCRp or HA- kaiCRp gene is under the control of the kaiCRp promoter. The resultant plasmid, pJQ200KS-Insert-PkaiCRp-FLAGkaiCRp, or pJQ200KS-Insert-PkaiCRp-HAkaiCRp, was then transformed into RCKO strain by conjugation with E.coli S17-1 [45,46,48]. The selection procedures were similar to those used to construct the kaiCRp deletion strain. The integration of FLAG-kaiCRp and HA- kaiCRp was confirmed by sequencing and immunoblotting.

Growth experiments

Batch liquid cultures of R. palustris strains including the wild-type R. palustris TIE-1, the kaiCRp knockout strain RCKO, and the kaiCRp complemented strain RCKO-FLAGkaiCRp were grown anaerobically in FW or YPA medium. For growth curve experiments and doubling time calculations, cultures were grown at 30°C or 23°C with gentle shaking under either LL conditions or LD (LD 12:12 or LD 1:1) conditions to test fitness by growth rate (as done for cyanobacteria in [49]). Seed cultures of these strains were grown anaerobically in FW medium under LL before inoculation. Growth was monitored by measuring the optical density (OD) at 600 nm.

Nitrogenase activity measurement

Nitrogenase activity of R. palustris was measured by the acetylene reduction assay [50]. Assays were carried out in sealed serum bottles containing the anaerobic cultures under LL or LD conditions. At each time point, 3 mL of the culture was withdrawn anaerobically, injected into a 7-mL blood collection tube (Becton Dickinson Vacutainer, 8020128) containing 10% acetylene gas (final concentration) and 90% nitrogen gas (final concentration) and incubated under light (40–50 μE m−2s−1) for 3 h. The reaction was stopped by injecting 100 μL of 0.5 M EDTA and stored at room temperature until analyzed by gas chromatography. For time-course experiments, samples were taken every 3 hours. A 500 μl aliquot of the gas phase from the headspace was analyzed by a gas chromatograph (Shimadzu GC-2010 Plus) fitted with a flame ionization detector and a Rt-Alumina BOND/MAPD PLOT column (Restech, PA). The temperatures of the injector, detector, and oven were 200°C, 200°C, and 130°C, respectively.

Phylogenetic analysis

The amino acid sequences of KaiC proteins from selected cyanobacterial species and purple non-sulfur bacterial species were retrieved from NCBI GenBank. The phylogenetic tree for KaiC was constructed using web service integrated MUSCLE alignment, Gblocks curation and PhyML method [5157]. The KaiC sequences were aligned and compared using PRALINE multiple sequence alignment program with the homology-extended alignment strategy [5861].

Quantitative reverse-transcription PCR

RNA was isolated from 3–5 ml of exponentially growing R. palustris cells with a NucleoSpin RNA II kit (Clontech). 300 ng of RNA was then used to synthesize cDNA by using the iScript cDNA synthesis kit (Bio-Rad). With 1 μl of the synthesized cDNA as template, quantitative reverse-transcription PRC was conducted on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) by using iTaq SYBR green supermix (Bio-Rad). The program was run at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. A final melting curve was performed for each reaction to ensure that only a single peak was amplified. The primers were designed by using Primer3 web service ( and shown in S3 Table.

Phosphorylation assay for KaiCRp

KaiCSe was labeled with P-32 by incorporation of γ-phosphate from 32P-labeled ATP as previously described [62]. To label KaiCRp, purified KaiCRp (0.2 μg/μL) was incubated at either 4°C or 30°C in 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl2, 1 mM DTT, 1 mM ATP and 0.22 μM [γ-32P]ATP (3000 Ci/mmol, PerkinElmer) for 24 hours. The phosphorylation reactions were terminated by mixing with SDS-PAGE sample buffer, heated at 96°C for 10 min, and subjected to either regular SDS-PAGE (10% acrylamide) or phosphate affinity SDS-PAGE (7.5% acrylamide with 50 μM Phos-Tag and 100 μM MnCl2)[63]. Gels were fixed, stained with colloidal Coomassie Brilliant Blue G-250, dried and imaged by a CCD camera (for CBB staining) or storage phosphor imaging technique (for 32P-autoradiography).

Immunoblot assay for HA-KaiCRp

The HA-kaiCRp strain was cultured in LD 12:12 cycles under anaerobic conditions. When the cell density (OD600) reached 0.4 in the exponential growth phase, 5 ml cells were collected every 6 h for 24 h in LD and then the culture was released to LL, under which conditions the cell collection was continued for another 48 h. After each collection, cells were centrifuged at 4°C, and cell pellets were immediately flash-frozen with liquid nitrogen. Total protein was extracted from cell pellets that had been resuspended in KaiC extraction buffer [37] by sonication, and the amount of total protein was quantified with the Lowry protein assay [64]. Equal amounts of total protein (10 μg) were placed into each well, and the proteins in the extracts were separated by phosphate affinity SDS-PAGE (7.5% acrylamide with 50 μM Phos-tag and 100 μM MnCl2) [63] and transferred onto polyvinylidene difluoride (PVDF) membranes. The immunoblots were treated with 0.5 μg/mL anti-HA monoclonal antibody (HA.11 clone 16B12, Covance, 1:2000 dilution) and detected with Pierce ECL Western Blotting Substrate (Thermo Scientific). The signal was captured by an Alpha Innotech gel image system (Cell Biosciences).

Purification of KaiC proteins and ATPase activity assay

The kaiCRp gene was cloned from the genomic DNA of R. palustris. A GST tag was added into the N-terminus for protein purification. Using the QuickChange II kit (Agilent), site-directed mutagenesis was conducted to generate mutated KaiCRp in which glutamate (E) residues were replaced with glutamine (Q) residues. The KaiCRp-EQ1EQ2 mutant had the four catalytic E residues (two in the CI domain, two in the CII domain) replaced by Q residues. GST-tagged KaiCRp and KaiCRp-EQ1EQ2 was expressed and purified in E. coli DH5α cells following the protocol of expressing and purifying KaiCSe [19,23]. GST fusion proteins were purified by affinity chromatography on glutathione-agarose resin (Pierce/Thermo Scientific) in buffer containing 20 mM Tris-HCl (pH 8.0), 300 mM NaCl, 5 mM MgCl2 and 1 mM ATP, and cleaved from GST using human rhinovirus 3C protease. The proteins were further purified by ion-exchange chromatography on Q Sepharose (GE Healthcare Life Sciences, 17-1014-01) with a gradient of NaCl. To confirm the purity of the KaiCRp proteins, some preparations were further subjected to gel filtration chromatography (Superdex 200 HR 10/30, GE Healthcare Life Sciences) to verify co-migration of the ATPase activity with KaiCRp (S6 Fig).

To assay ATPase activity, purified KaiCRp and KaiCRp-EQ1EQ2 proteins were incubated in 20 mM Tris-HCl (pH 8.0), 165 mM NaCl, 5 mM MgCl2, 1 mM ATP, 0.5 mM EDTA at 0, 23, 30 and 37°C for 0.5, 1, 2, 4, 12 and 24 hours. At each time point, 15 μL of the reaction was withdrawn and mixed with 75 μL of 260 mM sodium phosphate, 10 mM EDTA, pH 5.0 (= 6-fold dilution), snap-frozen in liquid nitrogen and stored at -80°C until the day of measurement. The hydrolysis of ATP by the KaiC proteins were quantified using high-performance liquid chromatography (HPLC) as described [65] by using a Synergi Polar-RP column (250 × 4.6 mm, Phenomenex) with an isocratic mobile phase of 260 mM sodium phosphate, pH 5.0 at a flow rate of 0.8 mL/min. Ten μL of the sample was injected and the eluents were monitored at a wavelength of 259 nm. Peak areas of analytes were obtained from each chromatogram by the software and inspected manually. The linearity of the detector response was assessed by injecting a wide range of known concentrations (1.66–166 μM) of the nucleotide, and a standard curve was generated by plotting adenine nucleotide concentrations versus peak area. The analyte concentrations of the samples were calculated from the standard curve.

Statistical analyses

Statistical analyses were performed in R. The modified cosinor method [66] was used to determine if the data were rhythmic or not on a 12-24-h time basis. The nitrogen fixation data were detrended linearly [65], and the detrended data were fitted to a series of cosine curves with different periods ranging from 12 h to 24 h. Along with curve fitting, the R2 value was calculated. Sample and replicate numbers are indicated in the legend to S5 Fig.

Supporting Information

S1 Text. Supplemental Methods.

Including: Construction of kaiCRp expression strains in S. elongatus, Measuring luminescence rhythms in cyanobacterial strains.


S1 Fig. Similar kaiBC genes of R. palustris and S. elongatus.

(A) Arrangement of kaiASe, kaiBSe, and kaiCSe genes in the genome of S. elongatus (separate promoters drive expression of kaiA and kaiBC transcripts) as compared with kaiCRp, kaiBRp, and a histidine kinase gene with a PAS domain in R. palustris (bioinformatic analyses suggest that a single promoter drives expression of these three genes). (B) Comparison of motifs found in kaiCRp versus kaiCSe shows the similarity of these genes. Both of the kaiC genes contain two RecA-like NTPase superfamily domains that include Walker A and Walker B motifs, catalytic EE residues, and DXXG motifs. The known phosphorylation sites of KaiCSe are located in the second domain (CII) as TST, while in KaiCRp they are TSS. The C-terminus of KaiCRp is about 50 amino acid longer than that of KaiCSe.


S2 Fig. Phylogenetic tree of kaiC genes in four cyanobacteria species and two purple non-sulfur bacteria strains.

Two strains of R. palustris (purple non-sulfur bacteria) are shown in comparison with four species of cyanobacteria. Note that Synechocystis sp. PCC 6803 harbors three copies of kaiC, two copies of kaiB clustered with kaiC genes, and one copy of kaiA clustered with kaiC, while Cyanothece sp. ATCC 51142 harbors two copies of kaiC, two copies of kaiB, and one copy of kaiA. The kaiA gene is only present among cyanobacteria. The red numbers are the bootstrap values signifying the confidence of each node.


S3 Fig. Daily patterns of nitrogen fixation in WT vs. RCKO strains.

These data are replotted from Fig 1 with all three replicate cultures averaged together. Data are mean +/- S.D. (n = 6).


S4 Fig. Lack of robust persistence of the nitrogen fixation rhythm in LL in individual cultures.

Each trace represents the nitrogen fixation activity of each individual culture from the three replicates of the experiment depicted in Fig 3. A, WT at 30°C; B, RCKO at 30°C; C, WT at 23°C; D, RCKO at 23°C.


S5 Fig. Statistical cosinor analyses of rhythmicity in LL.

Modified Cosinor analyses [66] were performed to assess the rhythmicity of nitrogen fixation activities under LL conditions. The free running period (FRP) was estimated by picking the highest R2 value. Upper Left Panel: nitrogen fixation activity of the WT strain at 30°C; the highest R2 value is 0.2 with the corresponding FRP of about 22 h. Data were from five independent experiments. Each experiment included at least three individual cultures. Upper Right Panel: the nitrogen fixation activity of the WT strain at 23°C; the highest R2 is 0.03 with the corresponding FRP of about 22 h. Data were from 2 independent experiments, and each experiment included three individual cultures. Lower Left Panel: the RCKO strain at 30°C; the R2 was 0.0008. Data were from 2 independent experiments, and each experiment included three separate cultures.


S6 Fig. Co-migration of native KaiCRp abundance and ATPase activity by gel filtration chromatography.

After GST affinity chromatography (on glutathione-agarose resin), cleavage from GST by protease, and ion-exchange chromatography on Q Sepharose as described in Experimental Procedures, the purified native KaiCRp was chromatographed on a gel filtration column (Superdex 200 HR 10/30) and fractions were collected. The ATPase activity of each fraction was determined as described in the Experimental Procedures. Upper Panel: The ATPase activity co-migrates on the gel filtration column with KaiCRp abundance, indicating that the ATPase activity is attributable to the KaiCRp rather than a contaminating protein of dissimilar molecular weight. Lower Panel: SDS-PAGE electrophoresis of fractions 1–30 showing the KaiCRp band in fractions 17–22. Similar results were obtained with the KaiCRp-EQ1EQ2 mutant protein.


S7 Fig. Expression of kaiCRp in S. elongatus cells affects the luminescence rhythm that reports circadian gene expression.

A, kaiCRp was overexpressed in the WT S. elongatus luxAB reporter strain AMC149 {in AMC149, the luminescence reporter luxAB indicates the activity of the psbAI promoter (Kondo et al., 1993)}. Upper panel, no IPTG induction; lower panel, IPTG (500 μM) was applied to the cultures after entrainment. Compared to the rhythm of AMC149 (blue dots), the luminescence rhythm was suppressed and the FRP was lengthened when kaiCRp was overexpressed (red dots). The Ptrc promoter that drives kaiC expression is slightly leaky (Xu et al., 2003), and therefore some KaiCRp is expressed even without IPTG treatment. The traces shown are representative examples of at least six replicates. B, Quantification of free-running periods in AMC149 (blue) and AMC149oxkaiCRp (red) strains with and without IPTG induction. Data are mean +/- S.D. (n = 6). See S1 Text for Supplemental Methods.


S1 Table. Bacterial strains used in this study.



These experiments were begun during the 2011 Microbial Diversity summer course at the Marine Biological Laboratory (Woods Hole, MA), and we thank Drs. Stephen Zinder, Daniel Buckley and Dr. Dianne Newman for inspiring these experiments. We are grateful to the following for materials: Dr. Dianne Newman (R. palustris TIE-1 strain), and Drs. Caroline Harwood & Peter Greenberg (plasmids and the E.coli strain for making gene knockouts). We thank the following for assistance and/or facilitating experiments: Drs. Jamey Young & Kenneth Debelak for assistance with gas chromatography, Philip Kingsley & Drs. Lawrence Marnett & James Weeks for their help with the HPLC analysis of ATPase activity, Drs. Karen Gamble & Andrew Millar for statistical analyses advice, and Brenda Pratte, Adeola Adebiyi & Lara Jazmin for technical assistance with the nitrogen fixation assay.

Author Contributions

Conceived and designed the experiments: PM CHJ. Performed the experiments: PM TM CZ. Analyzed the data: PM TM CZ CHJ. Contributed reagents/materials/analysis tools: TM TT. Wrote the paper: PM TM TT CHJ.


  1. 1. Dunlap JC Chronobiology: Biological timekeeping. Sunderland, Mass.: Sinauer 2004.
  2. 2. Ouyang Y, Andersson CR, Kondo T, Golden SS, Johnson CH. Resonating circadian clocks enhance fitness in cyanobacteria. Proc Natl Acad Sci U S A. 1998; 95: 8660–8664. pmid:9671734
  3. 3. Woelfle MA, Ouyang Y, Phanvijhitsiri K, Johnson CH. The adaptive value of circadian clocks: an experimental assessment in cyanobacteria. Curr Biol. 2004;14: 1481–1486. pmid:15324665
  4. 4. Ma P, Woelfle MA, Johnson CH. An Evolutionary Fitness Enhancement Conferred by the Circadian System in Cyanobacteria. Chaos Solitons Fractals. 2013; 50: 65–74. pmid:23626410
  5. 5. Johnson CH, Golden SS, Ishiura M, Kondo T. Circadian clocks in prokaryotes. Mol Microbiol. 1996; 21: 5–11. pmid:8843429
  6. 6. Grobbelaar NH, Lin HY, Chow TJ. Dinitrogen-fixing endogenous rhythm in Synechococcus RF-1. FEMS Microbiology Letters. 1986; 37: 173–177.
  7. 7. Kondo T, Strayer CA, Kulkarni RD, Taylor W, Ishiura M, et al. Circadian rhythms in prokaryotes: luciferase as a reporter of circadian gene expression in cyanobacteria. Proc Natl Acad Sci U S A. 1993; 90: 5672–5676. pmid:8516317
  8. 8. Ito H, Mutsuda M, Murayama Y, Tomita J, Hosokawa N, et al. Cyanobacterial daily life with Kai-based circadian and diurnal genome-wide transcriptional control in Synechococcus elongatus. Proc Natl Acad Sci U S A. 2009; 106: 14168–14173. pmid:19666549
  9. 9. Smith RM, Williams SB. Circadian rhythms in gene transcription imparted by chromosome compaction in the cyanobacterium Synechococcus elongatus. Proc Natl Acad Sci U S A. 2006; 103: 8564–8569. pmid:16707582
  10. 10. Taniguchi Y, Takai N, Katayama M, Kondo T, Oyama T. Three major output pathways from the KaiABC-based oscillator cooperate to generate robust circadian kaiBC expression in cyanobacteria. Proc Natl Acad Sci U S A. 2010; 107: 3263–3268. pmid:20133618
  11. 11. Vijayan V, Zuzow R, O'Shea EK. Oscillations in supercoiling drive circadian gene expression in cyanobacteria. Proc Natl Acad Sci U S A. 2009; 106: 22564–22568. pmid:20018699
  12. 12. Vijayan V, Jain IH, O'Shea EK. A high resolution map of a cyanobacterial transcriptome. Genome Biol. 2011; 12: R47. pmid:21612627
  13. 13. Woelfle MA, Xu Y, Qin X, Johnson CH. Circadian rhythms of superhelical status of DNA in cyanobacteria. Proc Natl Acad Sci U S A. 2007; 104: 18819–18824. pmid:18000054
  14. 14. Dong G, Yang Q, Wang Q, Kim YI, Wood TL, et al. Elevated ATPase activity of KaiC applies a circadian checkpoint on cell division in Synechococcus elongatus. Cell. 2010; 140: 529–539. pmid:20178745
  15. 15. Liu Y, Tsinoremas NF, Golden SS, Kondo T, Johnson CH. Circadian expression of genes involved in the purine biosynthetic pathway of the cyanobacterium Synechococcus sp. strain PCC 7942. Mol Microbiol. 1996; 20: 1071–1081. pmid:8809759
  16. 16. Mori T, Binder B, Johnson CH. Circadian gating of cell division in cyanobacteria growing with average doubling times of less than 24 hours. Proc Natl Acad Sci U S A. 1996; 93: 10183–10188. pmid:8816773
  17. 17. Diamond S, Jun D, Rubin BE, Golden SS. The circadian oscillator in Synechococcus elongatus controls metabolite partitioning during diurnal growth. Proc Natl Acad Sci U S A. 2015; 112: E1916–1925. pmid:25825710
  18. 18. Ishiura M, Kutsuna S, Aoki S, Iwasaki H, Andersson CR, et al. Expression of a gene cluster kaiABC as a circadian feedback process in cyanobacteria. Science. 1998; 281: 1519–1523. pmid:9727980
  19. 19. Nakajima M, Imai K, Ito H, Nishiwaki T, Murayama Y, et al. Reconstitution of circadian oscillation of cyanobacterial KaiC phosphorylation in vitro. Science. 2005; 308: 414–415. pmid:15831759
  20. 20. Tomita J, Nakajima M, Kondo T, Iwasaki H. No transcription-translation feedback in circadian rhythm of KaiC phosphorylation. Science. 2005; 307: 251–254. pmid:15550625
  21. 21. Johnson CH, Stewart PL, Egli M. The cyanobacterial circadian system: from biophysics to bioevolution. Annu Rev Biophys. 2011; 40: 143–167. pmid:21332358
  22. 22. Egli M, Johnson CH. A circadian clock nanomachine that runs without transcription or translation. Curr Opin Neurobiol. 2013; 23: 732–740. pmid:23571120
  23. 23. Mori T, Williams DR, Byrne MO, Qin X, Egli M, et al. Elucidating the ticking of an in vitro circadian clockwork. PLoS Biol. 2007; 5: e93. pmid:17388688
  24. 24. Loza-Correa M, Gomez-Valero L, Buchrieser C. Circadian clock proteins in prokaryotes: hidden rhythms? Front Microbiol. 2010; 1: 130. pmid:21687756
  25. 25. Dvornyk V, Vinogradova O, Nevo E. Origin and evolution of circadian clock genes in prokaryotes. Proc Natl Acad Sci U S A. 2003; 100: 2495–2500. pmid:12604787
  26. 26. Holtzendorff J, Partensky F, Mella D, Lennon JF, Hess WR, et al. Genome streamlining results in loss of robustness of the circadian clock in the marine cyanobacterium Prochlorococcus marinus PCC 9511. J Biol Rhythms. 2008; 23: 187–199. pmid:18487411
  27. 27. Zinser ER, Lindell D, Johnson ZI, Futschik ME, Steglich C, et al. Choreography of the transcriptome, photophysiology, and cell cycle of a minimal photoautotroph, prochlorococcus. PLoS One. 2009; 4: e5135. pmid:19352512
  28. 28. Mullineaux CW, Stanewsky R. The rolex and the hourglass: a simplified circadian clock in prochlorococcus? J Bacteriol. 2009; 191: 5333–5335. pmid:19561127
  29. 29. Larimer FW, Chain P, Hauser L, Lamerdin J, Malfatti S, et al. Complete genome sequence of the metabolically versatile photosynthetic bacterium Rhodopseudomonas palustris. Nat Biotechnol. 2004; 22: 55–61. pmid:14704707
  30. 30. Jiao Y, Kappler A, Croal LR, Newman DK. Isolation and characterization of a genetically tractable photoautotrophic Fe(II)-oxidizing bacterium, Rhodopseudomonas palustris strain TIE-1. Appl Environ Microbiol. 2005; 71: 4487–4496. pmid:16085840
  31. 31. Proctor LM. Nitrogen-fixing, photosynthetic, anaerobic bacteria associated with pelagic copepods. Aquat Microb Ecol. 1997; 12: 105–113.
  32. 32. Stal LJ. Physiological ecology of cyanobacteria in microbial mats and other communities. New Phytol. 1995: 1–32.
  33. 33. Schneegurt MA, Sherman DM, Nayar S, Sherman LA. Oscillating behavior of carbohydrate granule formation and dinitrogen fixation in the cyanobacterium Cyanothece sp. strain ATCC 51142. J Bacteriol. 1994; 176: 1586–1597. pmid:8132452
  34. 34. Iwasaki H, Nishiwaki T, Kitayama Y, Nakajima M, Kondo T. KaiA-stimulated KaiC phosphorylation in circadian timing loops in cyanobacteria. Proc Natl Acad Sci U S A. 2002; 99: 15788–15793. pmid:12391300
  35. 35. Xu Y, Mori T, Johnson CH. Cyanobacterial circadian clockwork: roles of KaiA, KaiB and the kaiBC promoter in regulating KaiC. Embo J. 2003; 22: 2117–2126. pmid:12727878
  36. 36. Qin X, Byrne M, Mori T, Zou P, Williams DR, et al. Intermolecular associations determine the dynamics of the circadian KaiABC oscillator. Proc Natl Acad Sci U S A. 2010; 107: 14805–14810. pmid:20679240
  37. 37. Xu Y, Mori T, Johnson CH. Circadian clock-protein expression in cyanobacteria: rhythms and phase setting. EMBO J. 2000; 19: 3349–3357. pmid:10880447
  38. 38. Terauchi K, Kitayama Y, Nishiwaki T, Miwa K, Murayama Y, et al. ATPase activity of KaiC determines the basic timing for circadian clock of cyanobacteria. Proc Natl Acad Sci U S A. 2007; 104: 16377–16381. pmid:17901204
  39. 39. Murakami R, Miyake A, Iwase R, Hayashi F, Uzumaki T, et al. ATPase activity and its temperature compensation of the cyanobacterial clock protein KaiC. Genes Cells. 2008; 13: 387–395. pmid:18363969
  40. 40. Min H, Guo H, Xiong J. Rhythmic gene expression in a purple photosynthetic bacterium Rhodobacter sphaeroides. FEBS Lett. 2005; 579: 808–812. pmid:15670851
  41. 41. Roenneberg T, Merrow M. Life before the clock: modeling circadian evolution. J Biol Rhythms. 2002; 17: 495–505. pmid:12465882
  42. 42. Troein C, Locke JC, Turner MS, Millar AJ. Weather and seasons together demand complex biological clocks. Curr Biol. 2009; 19: 1961–1964. pmid:19818616
  43. 43. Leipe DD, Aravind L, Grishin NV, Koonin EV. The bacterial replicative helicase DnaB evolved from a RecA duplication. Genome Res. 2000; 10: 5–16. pmid:10645945
  44. 44. Bustos SA, Golden SS. Expression of the psbDII gene in Synechococcus sp. strain PCC 7942 requires sequences downstream of the transcription start site. J Bacteriol. 1991; 173: 7525–7533. pmid:1938947
  45. 45. Hirakawa H, Oda Y, Phattarasukol S, Armour CD, Castle JC, et al. Activity of the Rhodopseudomonas palustris p-coumaroyl-homoserine lactone-responsive transcription factor RpaR. J Bacteriol. 2011; 193: 2598–2607. pmid:21378182
  46. 46. Bose A, Newman DK. Regulation of the phototrophic iron oxidation (pio) genes in Rhodopseudomonas palustris TIE-1 is mediated by the global regulator, FixK. Mol Microbiol. 2011; 79: 63–75. pmid:21166894
  47. 47. Quandt J, Hynes MF. Versatile suicide vectors which allow direct selection for gene replacement in gram-negative bacteria. Gene. 1993; 127: 15–21. pmid:8486283
  48. 48. Simon LD, Randolph B, Irwin N, Binkowski G. Stabilization of proteins by a bacteriophage T4 gene cloned in Escherichia coli. Proc Natl Acad Sci U S A. 1983; 80: 2059–2062. pmid:6340113
  49. 49. Xu Y, Ma P, Shah P, Rokas A, Liu Y. Non-optimal codon usage is a mechanism to achieve circadian clock conditionality. Nature. 2013; 495: 116–120. pmid:23417065
  50. 50. Hardy RW, Holsten RD, Jackson EK, Burns RC. The acetylene-ethylene assay for N2 fixation: laboratory and field evaluation. Plant Physiol. 1986; 43: 1185–1207.
  51. 51. Dereeper A, Audic S, Claverie JM, Blanc G. BLAST-EXPLORER helps you building datasets for phylogenetic analysis. BMC Evol Biol. 2010; 10: 8. pmid:20067610
  52. 52. Dereeper A, Guignon V, Blanc G, Audic S, Buffet S, et al. robust phylogenetic analysis for the non-specialist. Nucleic Acids Res. 2008; 36: W465–469. pmid:18424797
  53. 53. Edgar RC. MUSCLE: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinformatics. 2004; 5: 113. pmid:15318951
  54. 54. Castresana J. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol Biol Evol. 2000; 17: 540–552. pmid:10742046
  55. 55. Guindon S, Gascuel O. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol. 2003; 52: 696–704. pmid:14530136
  56. 56. Chevenet F, Brun C, Banuls AL, Jacq B, Christen R. TreeDyn: towards dynamic graphics and annotations for analyses of trees. BMC Bioinformatics. 2006; 7: 439. pmid:17032440
  57. 57. Anisimova M, Gascuel O. Approximate likelihood-ratio test for branches: A fast, accurate, and powerful alternative. Syst Biol. 2006; 55: 539–552. pmid:16785212
  58. 58. Simossis VA, Heringa J. PRALINE: a multiple sequence alignment toolbox that integrates homology-extended and secondary structure information. Nucleic Acids Res. 2005; 33: W289–294. pmid:15980472
  59. 59. Simossis VA, Kleinjung J, Heringa J. Homology-extended sequence alignment. Nucleic Acids Res. 2005; 33: 816–824. pmid:15699183
  60. 60. Heringa J. Local weighting schemes for protein multiple sequence alignment. Comput Chem. 2002; 26: 459–477. pmid:12144176
  61. 61. Heringa J. Computational methods for protein secondary structure prediction using multiple sequence alignments. Curr Protein Pept Sci. 2000; 1: 273–301. pmid:12369910
  62. 62. Xu Y, Mori T, Qin X, Yan H, Egli M, Johnson CH. Intramolecular regulation of phosphorylation status of the circadian clock protein KaiC. PLoS ONE. 2009; 4: e7509. pmid:19946629
  63. 63. Kinoshita E, Kinoshita-Kikuta E, Takiyama K, Koike T. Phosphate-binding tag, new tool to visualize phosphorylated proteins. Mol Cell Proteomics. 2006; 5, 749–757. pmid:16340016
  64. 64. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem. 1951; 193: 265–275. pmid:14907713
  65. 65. Sudo J, Terui J, Iwase H, Kakuno K. Assay of ATPase and Na,K-ATPase activity using high-performance liquid chromatographic determination of ADP derived from ATP. J Chromatogr B Biomed Sci Appl. 2000; 744: 19–23. pmid:10985562
  66. 66. Kucho K, Okamoto K, Tsuchiya Y, Nomura S, Nango M, et al. Global analysis of circadian expression in the cyanobacterium Synechocystis sp. strain PCC 6803. J Bacteriol. 2005; 187: 2190–2199. pmid:15743968