At first, the saprophytic eukaryote Neurospora crassa and the photosynthetic prokaryote Synechococcus elongatus may seem to have little in common. However, in both organisms a circadian clock organizes cellular biochemistry, and each organism lends itself to classical and molecular genetic investigations that have revealed a detailed picture of the molecular basis of circadian rhythmicity. In the present chapter, an overview of the molecular clockwork in each organism will be described, highlighting similarities, differences and some as yet unexplained phenomena.
Circadian clocks are internal timekeepers that integrate and respond to environmental time cues, such as light and temperature, to provide an accurate depiction of external time. Time signals sent from circadian clocks ensure that clock-controlled processes occur at a favourable and predictable time of day at both the cellular and organismal level . Circadian clocks seem to have arisen several times in the course of evolution ; nevertheless, all clocks are built from the interactions between pathways that sense and evoke a response to environmental stimuli. Thus several clock and clock-associated molecules are also integral components of light-input pathways [3–6]. Others are molecules that sense changes in cellular redox potential  or even magnetic fields . Moreover, proteins that regulate the activity and turnover of these clock gene products, for example, kinases, phosphatases and chromatin-remodelling enzymes, influence clock time .
Circadian clocks have a periodicity of about (circa) a day (diem). However, when organisms are exposed to the day–night cycle, the periodicity is entrained by environmental time cues to exactly 24 h. The situation is different in the laboratory when organisms are kept in constant darkness (saprophytes) or constant low light (plants and photosynthetic bacteria). Under these conditions, their circadian oscillators free-run with a periodicity that is usually slightly greater or less than 24 h. The features of a good circadian clock are the same as those for a mechanical timepiece: accuracy, persistence, temperature compensation of period, and the ability to be reset . Historically, rhythmic biochemistry, physiology and behaviour were assayed without knowing the identity of the endogenous timekeepers. However, once the phenomenology of circadian oscillators was described and analysed, the remaining questions surrounded the exact biochemical nature of the clockwork.
In the early days of circadian research, there were two opposing schools of thought as to the identity of the clock: one professed that circadian rhythmicity emerged as a network property of the whole cell that would be impossible to dissect genetically; and the second was convinced that genes that function to build a molecular clock could be identified by forward genetic approaches. The latter hypothesis was proved correct after clock mutants were identified in Neurospora and Drosophila, whose properties segregated according to simple Mendelian genetics. Central clock genes that when mutated give rise to altered rhythmicity have since been mapped and isolated in many other organisms, including Synechococcus, Chlamydomonas, Arabidopsis, zebrafish, mice and humans [2,11]. The picture that has emerged from data gathered over the years in a number of laboratories, but initially revealed from work carried out in Neurospora and Drosophila, is that many clock components are transcription factors functioning in positive and negative feedback loops that take approx. 24 h to complete .
This is not to say that the entire circadian system can be understood on the basis of the action of these genes alone. It has long been appreciated that metabolism and/or the energy state of the cell play important roles in rhythm generation. For example, mutations in mitochondrial genes and components of lipid biosynthesis affect circadian rhythmicity in Neurospora , and the more recent discoveries that many clock genes respond to the metabolic state of the cell has rekindled interest in the role of metabolism in influencing the time on the circadian clock .
Circadian clocks in eukaryotes and prokaryotes
Neurospora was identified in the 1950s as a good model organism for the study of a eukaryotic circadian clock . Neurospora belongs to the class of Ascomycetes that are characterized by the formation of a sac-like structure (ascus) in which their sexual spores develop. In its vegetative form, it produces long branching tube-like structures called hyphae that can differentiate into aerial hyphae that typically bud-off asexual spores called conidia. Neurospora can be grown on defined medium, and the induction of both sexual and asexual reproduction is straightforward, thus facilitating genetic analysis of traits. These traits were famously exploited by Beadle and Tatum in work that led to the ‘one gene–one enzyme hypothesis' . The development of so called ‘race tubes' (long glass tubes containing a thin layer of growth medium with upturned ends plugged with cotton wool) enabled researchers to easily measure growth characteristics of mutant strains . Pittendrigh et al.  used this assay to good effect to demonstrate the existence of a Neurospora circadian clock. When cultures of Neurospora were placed in constant conditions of darkness and temperature, bands of asexual spores were seen to form once every 22 h in these race tubes. Other circadian characteristics, such as temperature compensation and resetting of the rhythm were later confirmed . Subsequently, mutagenized asexual spores were germinated and grown along these tubes and mutants with altered periodicities of sporulation isolated . As would be expected, many genes that are involved in the development of conidia are under clock control. However, conidiation is not the only process under clock control; microarray analysis suggests that at the level of transcription, 10–15% of Neurospora genes are clock-regulated, many of which are not linked to asexual reproduction . Given the strong likelihood that post-transcriptional events are also under circadian control, one can predict that clock control of the cellular biochemistry and physiology of Neurospora is extensive.
Although numerous examples of circadian rhythms had been described in eukaryotes by the 1950s, at this time there was no evidence for the existence of circadian clocks in prokaryotes. Evidence for prokaryotic circadian clocks only emerged in the mid to late 1980s, when it was shown that certain species of cyanobacteria displayed daily cycles of nitrogen fixation, photosynthesis, carbohydrate synthesis, amino acid uptake and cell division [20,21], some of which were confirmed to be under clock control . Further research identified Synechococcus elongatus to be a good choice of organism to work out the molecular details of the underlying cyanobacterial circadian system . Despite the fact that molecular investigations of the cyanobacterial clock were initiated at least 20 years later than similar studies of eukaryotic clocks, the cyanobacterial clock quickly became one of the best understood. Uniquely in circadian clock research, the temporal changes and structures of the key clock molecules have been analysed in great detail in vitro .
The Neurospora clockwork
In Neurospora, screens for rhythm mutants led to the identification of several putative clock genes, including chrono, frq (frequency) and prd (period 1-4) . The first of these genes to be cloned was frq, which soon emerged as a key component of the molecular clock . Other clock mutant strains, wc-1 (white collar-1) and wc-2, were isolated in screens for factors involved in light sensing and light signal transduction . The latter encode zinc-finger transcription factors  that form a heterodimeric complex, the WCC (WHITE COLLAR complex). The WCC regulates rhythmic transcription of frq through its binding to specific elements in the frq promoter [3,27] (Figure 1).
Following its translation, FRQ forms a homodimer that interacts with FRH [FRQ-interacting helicase  (forming the FFC (FRQ/FRH complex)] and CK1 (casein kinase 1) . On translocation of FRQ to the nucleus, the WCC complex is phosphorylated. Hyperphosphorylation of WCC reduces its binding to the frq promoter, and thus transcription of frq decreases  (Figure 1). FRQ also affects the transcription of wc-2  and post-transcriptionally regulates levels of WC-1 protein in the cytosol. Hypophosphorylated FRQ removes WC-1 from the nucleus, possibly by shuttling it into the cytoplasm , whereas hyperphosphorylated FRQ in the cytoplasm promotes accumulation of cytoplasmic WC-1 [30,33]. Both of these roles of FRQ are important for the generation of a robust circadian oscillation, but only FRQ's role in negative feedback is essential for generating circadian rhythmicity .
Rhythmic transcription in Neurospora relies not only on the availability of active WCC, but also on local changes in the structure and accessibility of the DNA. These changes are brought about by the chromodomain protein CLOCKSWITCH and at least one other, as yet unidentified, remodelling protein . Their action contributes to the production of peak levels of frq RNA synthesis during the morning and, consequently, a peak in FRQ protein by midday . Cycling levels of frq RNA are additionally enhanced through the action of FRH. FRH is a cofactor of the Neurospora exosome, and the complex of FRQ and FRH binds to the frq transcript, targeting it for degradation. Thus when FRQ levels are low, frq RNA is more stable and vice versa . Because FRH affects the stability of many transcripts, knocking down frh expression has, besides its effects on the clock, dramatic effects on viability. However, a mutation in FRH has recently been reported that seems only to affect its role in the clock and leaves all other activities intact . The mutation disrupts the interaction between the FFC and the WCC. This leads to a breakdown of the negative feedback loop, because it results in a reduction of WC-1 phosphorylation and thus elevated WCC activity. In addition, WCC levels are reduced in the mutant strain, because the positive feedback of FRQ on WC levels is also disrupted. Owing to its association with the exosome, the FRQ–FRH complex may regulate the half-life of many different RNAs , reinforcing the cyclical abundance of clock-controlled gene transcripts and, perhaps, imposing rhythms in the abundance of constitutively transcribed RNAs.
Phosphorylation has an essential role in the Neurospora clock, and the phosphorylation state of key clock components is an important determinant of both phase and period length. FRQ recruits CK1 and CK2 to phosphorylate the WCC in the nucleus, thus down-regulating its own transcription. The same kinases phosphorylate FRQ and regulate its half-life . As soon as it is made, FRQ protein is phosphorylated and continues to be phosphorylated by PKA (protein kinase A), CK1, CK2 and a calcium/calmodulin kinase until hyperphosphorylated FRQ is targeted for degradation via the ubiquitin–proteosome pathway . Hyperphosphorylated FRQ attracts FWD-1 (F-box/WD-40-repeat-containing protein), and this complex is the substrate of the E3 ubiquitin ligase SCF (Skp1/cullin/F-box). Components of this degradation pathway are highly conserved, with homologues found in flies and mammals, and it transpires that fly and mouse clocks use similar pathways to ensure clock protein turnover .
The activity of the protein phosphatases PP1, PPA2 and PP4 also has an impact on the negative feedback regulating the phosphorylation state of FRQ and the WC proteins . Degradation of hyperphosphorylated FRQ releases repression of the WCC and allows the cycle of transcription and translation to begin once again. Recent data garnered from MS has allowed the majority of FRQ phospho-sites to be mapped. The importance of phosphorylation at different sites for fine-tuning periodicity was revealed through mutation analysis. Sites located in the N-terminal half of FRQ result in period lengthening, whereas mutation of sites at the C-terminus result in period shortening [40,41].
Phosphorylation of specific sites in FRQ also plays a role in temperature compensation of period, a characteristic of circadian clocks that is still poorly understood. When the chrono and period-3 genes were cloned, they were found to encode subunits of CK2. Mutant forms of chrono and period-3 were already known to affect temperature compensation, but the latest study by Mehra et al.  shows that as temperature changes, different sites on FRQ are phosphorylated by CK2. Importantly, when these sites are mutated, temperature compensation is affected. Hence, temperature compensation depends upon changes in FRQ activity brought about through the phosphorylation of specific sites. These sites may become accessible to the kinase via temperature-induced changes in the conformation of FRQ, or be phosphorylated owing to temperature-induced changes in the binding affinity of CK2 .
Although a great deal is now known about the molecules that make up Neurospora's circadian clock, the exact role of several clock genes and proteins has yet to be fully worked out. For instance, although two forms of the FRQ protein, sFRQ (small FRQ) and lFRQ (large FRQ) are made, expressing either form of FRQ alone can rescue rhythmicity in a frq-null strain . Indeed, the only observable phenotypic difference between these strains is a 1 h change in period length . This is at odds with the expectation of a stronger phenotype because of the complex regulation associated with sFRQ and lFRQ expression. The ratio of the two forms is regulated both by upstream open reading frames and temperature-sensitive alternative splicing of the frq mRNA . Additionally, frq encodes an overlapping and almost completely complementary transcript named qrf, with no large open reading frame. Expression of qrf affects the response of the clock to light , but its mode of action is unknown. Moreover, as the wc-1 and wc-2 genes are studied in more detail, it appears that they too produce different transcripts and protein isoforms [46,47] the significance of which is not fully understood.
In Neurospora, light is sensed directly by the clock. WHITE COLLAR-1 is both a transcriptional activator of frq in the dark and light and the primary blue-light photoreceptor [3,48]. Light perceived by WHITE COLLAR-1-associated FAD induces a conformational change in the WCC, allowing it to bind to a light element in the frq promoter  and evade FRQ repression. Without desensitization to light, the resulting elevated level of frq, and consequently FRQ, would result in arhythmicity. However, VIVID, a repressor of the WCC complex and a second blue-light photoreceptor [5,50], suppresses light responses at dawn and allows the clock to run during the day .
With the exception of osmotic stress, which is relayed to the clock via MAPKs (mitogen-activated protein kinases) , the impact of signals from the environment on the clock, other than light and temperature, are less well studied. The exact nature of clock output is also currently under investigation. The WCC is known to bind to many different promoters , but how time signals are relayed from the clock to activate and repress clock-controlled processes is not known in detail.
Clock components in cyanobacteria
In the absence of a circadian phenotype that could be easily exploited in a large-scale forward genetic screen, the bacterial luciferase gene was fused to the circadian-controlled Synechococcus photosystem II promoter. Cyclical expression of the transgene in the presence of the substrate luciferin resulted in the transformed bacteria emitting light with a 24 h periodicity. This strain was then mutated and colonies of bacteria monitored for altered periodicity of bioluminescence. Mutants with periods ranging from 16 h to 60 h and several arrhythmic strains were identified . Surprisingly, it turned out that only three genes were responsible for the altered rhythmicity, and these were named kai A, B and C (from the Japanese word ‘kaiten', meaning ‘cycle'). The gene sequences gave few clues as to the nature of the encoded proteins. Neither KaiA nor KaiB contain any recognizable domain that could shed light on their function. On the other hand, KaiC shares some similarity to the bacterial superfamily of DNA helicases/recombinases and contains two Walker domains that bind ATP .
KaiC forms an ATP-dependent homo-hexameric complex that, via its autokinase and autophosphatase activity, generates circadian oscillations in its phosphorylation status . KaiA promotes phosphorylation of KaiC, whereas KaiB inhibits the action of KaiA [57,58] (Figure 2). During the late daytime and early evening, KaiC becomes increasingly phosphorylated, stimulated by binding of KaiA dimers. Late in the subjective night, KaiB is released from the plasma membrane, and its interaction with the KaiC complex results in inactivation of KaiA and the subsequent dephosphorylation of KaiC. At this time, the complex dissociates and the cycle begins again (reviewed in ).
As is the case in eukaryotes, Synechococcus clock gene transcripts are rhythmically expressed. Continuous overexpression of kaiC represses expression of the dicistronic kaiBC transcript, suggesting negative feedback of kaiC on its own promoter . A surprising finding was that the overexpression of kaiC results not just in the repression of clock genes, but repression of all promoters in the Synechococcus genome . The pervasiveness of rhythmical transcription in Synechococcus and the fact that simple expression of the three kai genes is sufficient to drive rhythmicity suggested that a global mechanism was at work. One idea was that rhythmic gene expression was controlled by global changes in chromosome topology . The bacterial chromosome is organized into a nucleoid, and changes in the local supercoiling status of DNA regulate transcription. Thus, from the known similarity between KaiC and DNA helicases, the proposal arose that KaiC itself might change the degree of chromosome condensation. However, KaiC displays only weak binding to DNA and may effect chromosome compaction indirectly [61,62]. In a recent study in which circadian gene expression was correlated with the superhelicity of an endogenous plasmid, it was found that the AT-richness of genes dictated when they were expressed . A candidate pathway from KaiC to the chromosome is via the KaiC-binding histidine kinase SasA (Synechococcus adaptive sensor protein). The response-regulator partner of SasA is RpaA (phycobilisome-associated protein), a putative DNA-binding protein without which circadian transcription is severely attenuated. The activity of this phospho-relay alters depending on the phosphorylation status of KaiC , with activity peaking shortly before peak levels of maximally phosphorylated KaiC. However, in the absence of SasA, chromosome compaction is still rhythmic, therefore the direct regulators of chromosome state have yet to be identified.
Because kaiB and kaiC are rhythmically expressed, it was initially thought that this must be important for the generation of rhythmicity, but intriguingly, in prolonged darkness in which neither transcription nor translation of kaiB and kaiC occurs, a 24 h rhythm in KaiC phosphorylation persists . This injected some uncertainty regarding the role of the transcription and translation of clock genes for circadian timekeeping, and indicated that perhaps the cycle of KaiC phosphorylation and dephosphorylation alone constituted the oscillator. Indeed, Nakajima et al.  were able to reconstitute the KaiC phosphorylation cycle in vitro with only purified KaiA, B and C and ATP present. Their first astounding observation was that the phosphorylation status of KaiC oscillated with a period of 24 h. Secondly, they observed that this oscillation is temperature compensated. Thirdly, the in vitro periodicity of KaiC mutant proteins was similar to that seen in vivo in the corresponding mutant strains. Thus the phosphorylation state of KaiC generates temperature-compensated time-of-day information that correlates with, but has not been directly linked to, changes in chromosome compaction .
After almost dismissing the importance of rhythmic kai gene transcription, recent work has invested it with a key clock role. In kaiA-overexpressing strains, KaiC is constitutively hyperphosphorylated yet gene expression is circadian. Therefore circadian rhythms in transcription and translation can be generated independently of circadian cycles of KaiC phosphorylation. However, when there is no oscillation in KaiC phosphorylation status, the period is shorter and the amplitude of gene expression smaller than in a wild-type strain. In addition, it transpires that at low temperatures, circadian rhythmicity persists only when there is both rhythmic gene expression and rhythmic phosphorylation of KaiC . These results led Kitayama et al.  to suggest that, in certain environments, both transcription and post-transcriptional circadian oscillations are required for a functional and robust circadian clock.
Because the KaiC phosphorylation cycle takes place in vitro, it has been possible to monitor the interactions between clock proteins and their structure at different times of the day [66,67]. In combination with biochemical analysis of wild-type and mutant KaiC, it has been shown that sequential phosphorylation and then dephosphorylation of KaiC amino acid residues Ser431 and Thr432 is required for rhythmicity in vivo [68,69]. When both sites are phosphorylated KaiC has autophosphatase activity, whereas when the sites are dephosphorylated KaiC acts as a kinase . A temperature-compensated circadian rhythm of KaiC ATPase activity is essential for this phosphorylation rhythm, and ATPase activity is stimulated by KaiA and reduced by KaiB. Although KaiC contains two ATP-binding domains, ATPase activity is extremely low. Over a wide range of temperatures, 15 molecules of ATP are hydrolysed per circadian day . It is believed that the very low ATPase activity of KaiC results from an immediate autoregulatory inhibition of ATPase activity by its own action, presumably through a conformational change of KaiC. Moreover, this autoregulation is hypothesized to be the basis of temperature compensation of the KaiC ATPase activity.
Compared with the detailed understanding of the molecular clockwork, information about the proteins that sense the light environment and relay this information to the cyanobacterial clock is less plentiful. Nevertheless, mutant screens have revealed the identity of several signalling proteins that affect the clock's response to light, for example those encoded by lpdA (light-dependent period) and cikA (circadian input kinase). Interestingly, neither of these proteins is a photoreceptor, but rather they sense the redox state of the cell, which changes as a result of an increase or decrease in photosynthesis. Both LpdA and CikA bind to the complex of Kai proteins and affect their activity, thus influencing clock time . Response to the light environment also occurs via Pex protein (encoded by period extender) . Pex accumulates in the dark and binds to the kaiA promoter, delaying its expression .
Classical and molecular genetic approaches have revealed a detailed picture of the molecular basis of circadian rhythmicity in both Neurospora and Synechococcus. The remarkable efficiency and ease with which these organisms can be manipulated betrays the astounding fact that in many respects the fundamental mechanisms are similar to those in complex eukaryotes, including humans . However, although the basic principles of circadian rhythmicity have emerged there is much to learn about how clocks interact with the real world. For example, important details of entrainment, and temperature and nutritional compensation, which will allow us to understand in full how the circadian system bestows its selective advantage on organisms, have yet to be discovered.
The extent to which the circadian clockwork of Synechococcus and Neurospora is representative of the clockwork of other bacteria and fungi is not yet known, but orthologues of Neurospora clock genes have been identified in other fungi , and the kai genes are present in other species of cyanobacteria . Although clock proteins in Neurospora and Synechococcus are different, some parallels may be drawn. Both the Synechococcus and Neurospora clocks feature clock genes that are rhythmically expressed and are integral components of interlocking autoregulatory negative feedback loops. Both express pivotal clock proteins whose activities are regulated rhythmically via phosphorylation. These latter features are important in all clocks studied to date. In Neurospora and Synechococcus, phosphorylation also induces structural changes in key clock proteins required for temperature compensation of period.
Initially, the discovery that circadian phosphorylation of KaiC could be reconstituted in vitro indicated that the rhythm generator had been found. However, now that the importance of circadian clock gene transcription has once again been demonstrated, it seems likely that, at more than one level of biochemistry, rhythmic gene expression is important if circadian rhythmicity is to persist in defiance of competition from cellular and environmental noise. Similar to a food web where the fate of each organism depends on many other organisms, in the cell it is to be expected that a variety of factors link directly and indirectly to the clock, and small changes in their abundance or activity reverberate through the circadian clockwork. Thus each new connection to the clock provides an increasingly holistic view of the system.
• Circadian clocks are molecular timekeepers found in both eukaryotes and prokaryotes that provide organisms with a means to predict and prepare for environmental change.
• Forward and reverse genetic approaches in Neurospora and Synechococcus have identified key molecular components of the circadian clockwork and the network logic of their interactions.
• Circadian clocks are transcription/translation-based oscillators that generate and drive rhythmicity on a number of different levels through positive and negative feedback circuits that operate on transcriptional and post-transcriptional processes.
• Kinases and phosphatases regulate post-translational control of clock-protein activity and half-life. In both Synechococcus and Neurospora, the structure and activity of key clock components change depending on their phosphorylation state, and this regulates the phase, amplitude and temperature-compensation properties of the clock.
• To align their clocks with the environment, molecules that sense the daily changes in light and temperatures relay this information to clock proteins and, via changes in phosphorylation, alter clock time.
I thank Christian Heintzen, Seona Thompson and Suzanne Hunt for careful reading of the manuscript and helpful suggestions.
- © The Authors Journal compilation © 2011 Biochemical Society