Circadian biological clocks are self-sustained biochemical oscillators. Their properties include an intrinsic time constant of approximately 24 hours, temperature compensation (so that they run at a period of 24 hours irrespective of temperature), and entrainment to daily environmental cycles 1. Many biological processes are controlled by these clocks, including gene expression, neuronal activity, photosynthesis, sleeping and waking, and development. Microarray analyses of mRNA expression patterns in eukaryotes have demonstrated that 5-10% of genes exhibit daily rhythms of mRNA abundance. But mRNA abundance is not necessarily comparable with transcriptional activity. For example, microarray and promoter-trap experiments in the eukaryote Arabidopsis have demonstrated that only 6% of genes showed rhythms of mRNA abundance 2, whereas about 35% of promoters were rhythmically controlled 3. These results imply that the promoters of many eukaryotic genes are controlled by the biological clock, but that post-transcriptional control mechanisms counterbalance the rhythmic transcriptional activity of some genes so that their mRNA abundances are constant.

In prokaryotic cyanobacteria, it is not a mere fraction of the total entourage of promoters that is regulated by the daily biological clock; rather, there is global control of promoter activity by the daily timekeeper 4. This remarkable property was demonstrated by a promoter-trap experiment using random insertion of a promoterless luciferase gene throughout the genome of Synechococcus elongatus. Of the more than 800 insertion-line colonies analyzed, all displayed circadian rhythms of glowing luciferase function with the same period 4. The pattern of rhythmic expression differed between the promoters, in terms of both phasing and waveform (Figure 1a-d). Heterologous promoters, such as an Escherichia coli promoter (conIIp) were also transcribed rhythmically when inserted into the cyanobacterial chromosome 5. Apparently the cyanobacterial clock controls gene expression globally - by regulating the activity of all promoters. No microarray analysis of mRNA abundances in cyanobacteria has yet been reported, but it is likely - as in the case of Arabidopsis - that some of the genes whose promoter activities are rhythmic may exhibit 'de-regulated', that is non-rhythmic, patterns of mRNA abundance.

In cyanobacteria, there are at least three essential clock-specific genes, kaiA, kaiB, and kaiC, that form a cluster on the chromosome 6. Some features of kai gene regulation appear reminiscent of the regulation of eukaryotic clock genes. For example, there are rhythms in the abundance of the kaiA and kaiBC transcripts 6 and of the KaiB and KaiC proteins 78. KaiA, KaiB and KaiC interact with each other 910 and with a histidine kinase, SasA 11; these interactions appear to lead to the formation of protein complexes in vivo 12. KaiC exists in phosphorylated forms in vivo 8, suggesting another similarity to the post-translational control of eukaryotic clock proteins. KaiA stabilizes KaiC in its phosphorylated form, and KaiB antagonizes the effect of KaiA 8131415. The ratio of phosphorylated to non-phosphorylated KaiC is correlated with the period at which the clock runs 15.

Continuous overexpression of KaiC was found to repress the kaiBC promoter (kaiBCp), suggesting negative feedback of KaiC on its own promoter in an analogous fashion to the situation for eukaryotic clock proteins 6. The kaiBC promoter is not the only target of KaiC, however; the recent paper by Nakahira and coworkers 16 reports the unexpected result that KaiC overexpression represses the rhythms of all promoters in the S. elongatus genome. Intriguingly, this study identified two classes of response to KaiC repression. The first class, termed 'high amplitude' by Nakahira and coworkers 16, was exhibited by 5-10 % of the promoters, including kaiBCp; these promoters normally show a high-amplitude oscillation that is obliterated by KaiC overexpression (Figure 1e). This pattern reflects promoters whose expression is 'clock-dominated', with practically no basal activity at trough phases or during KaiC overexpression. The second response - exhibited by 90-95% of promoters - is a 'clock-modulated' response, termed 'low amplitude' by Nakahira and coworkers 16 (Figure 1f). This is a lower amplitude oscillation, in which the rhythmic component is abolished by KaiC overexpression, but a significant non-rhythmic basal level remains. These results indicate that KaiC (probably as part of a complex) coordinates genome-wide gene expression; the majority of genes have significant basal activity and are rhythmically modulated by the KaiABC oscillator, while in a smaller subset of genes, the oscillator dominates transcriptional activity (Figure 1e,1f) 16. This latter class might turn out to contain genes that encode proteins intrinsically involved in the cell's circadian-clock system.

Considerable evidence indicates that circadian feedback loops in eukaryotes are autoregulatory, whereby clock proteins directly or indirectly regulate the activity of their own genes' promoters 17. It was therefore a surprise to discover that the kai promoters are dispensable; Kai proteins can be expressed from a heterologous promoter and the cyanobacterial clock ticks along unperturbed 1516. The cyanobacterial transcriptional apparatus recognizes the heterologous promoter (in this case trcp from E. coli), but trcp is obviously not a promoter that evolved in conjunction with cyanobacterial clock genes. We first reported the functional replacement of kaiBCp 15, and now Nakahira and coworkers 16 report the functional replacement of both kaiAp and kaiBCp. Both studies found that expression of the Kai proteins needs to be within a permissive window of intracellular concentration to permit rhythmicity 1516. Thus, the circadian feedback loop in cyanobacteria does not require negative feedback of clock proteins upon specific clock promoters; apparently all that is required is the expression of an appropriate level of Kai proteins. Even temperature compensation - a defining characteristic of circadian clocks - is preserved when trcp replaces kaiBCp 16.

The pervasiveness of rhythmic transcriptional activity, and the fact that the clockwork does not require specific clock-gene promoters, suggests a broadly global mechanism for the cyanobacterial clock system. But what is the basis of this global regulation? One possibility could be rhythmic control by RNA polymerase sigma subunits, which often determine the promoter specificity of the polymerase. But studies of sigma subunits in cyanobacteria have not yielded explanations for global regulation 18. An alternative is the possibility that chromosomal topology is involved. The chromosome in most bacteria is organized into a 'nucleoid', which has a highly organized architecture based on condensation and coiling of DNA 19. It is well known that changes in the local supercoiling status of DNA can affect the transcriptional rate of genes 20, and our findings concerning the behavior of promoters in cyanobacteria support those observations 21. We proposed in 2001 that KaiC might mediate both its own negative feedback regulation and global regulation of the cyanobacterial genome by orchestrating oscillations in the condensation and/or supercoiling status of the entire cyanobacterial chromosome 22.

The most recent findings from our lab 15 and the lab of Susan Golden 21, in addition to the study of Nakahira and coworkers 16, are consistent with this hypothesis, namely that the condensation or supercoiling status of the cyanobacterial chromosome rhythmically changes such that it becomes an oscillating nucleoid, or 'oscilloid' (Figure 2). There is already a precedent for daily rhythms of topology in the chloroplast chromosome of the eukaryotic alga Chlamydomonas 23. In cyanobacteria, we postulated that these topological oscillations promote rhythmic modulation of the transcription rates of all genes, accounting for the global regulation of gene expression 22. Gene-specific cis-regulatory elements that mediate rhythmic gene expression might therefore be (at least partially) responsive to chromosomal status rather than exclusively to trans factors, leading to clock-dominated and clock-modulated expression patterns (Figures 1 and 2). In addition, heterologous promoters (for example E. coli trcp) that are integrated into the chromosome are driven rhythmically because they are also subjected to the oscillating chromosomal status 1516. Finally, KaiC (or, most likely, a KaiC-containing protein complex) is a key player in regulating these changes of chromosomal status 1516, and the phosphorylation status of KaiC is important in the regulation of this complex's activity (Figure 2) 815.

At present, it appears that the clock system in cyanobacteria is different from that in eukaryotes, and that changes in chromosomal topology could be a key element. In the fullness of time, however, we might find rhythmic modulation of chromosomal structure to be important in eukaryotic clock regulation - indeed, suggestive evidence for that hypothesis already exists for the mammalian clock 24. If this proves to be the case, the investigations of the cyanobacterial clock may lead to fundamental insights that are broadly applicable to all organisms.