Organization is an underlying theme in the study of living systems. This truth is evident in any number of basic cellular functions. By necessity, the execution of all biological functions falls within the persistent influence of an oscillatory external environment. It is of little surprise then that the organization of living systems extends into temporal perception and even anticipation. This review focuses on the circadian system of internal daily organization, introducing the core biological clock and the rhythms that it drives. Because excellent reviews have recently appeared detailing a variety of fine points on circadian regulation (de Paula et al., 2007; Dunlap et al., 2007; Loros et al., 2007; Brunner & Kaldi, 2008; Jinhu & Yi, 2010), this review is aimed more towards providing a historical context and introducing the field, the key molecular questions and players, and focuses on the regulation of these players through protein phosphorylation.
Circadian rhythms have been reported in almost all branches in the universal tree of life (Dunlap, 1999). At its most basic, the circadian system can be thought of in three parts as input, a central oscillator, and output (Fig. 1a, Pittendrigh & Bruce, 1957). The central oscillator functions as an endogenous self-sustaining rhythm generator with a period approximating that of the 24-h cycle of Earth's rotation. Circadian rhythms manifest at all levels of tissue organization; yet, they are generated at the level of individual cells and persist in the absence of external stimuli. Furthermore, the period length of the rhythms controlled by circadian clocks is largely the same at different ambient temperatures, reflecting a mechanism for compensation. This last feature is central to the concept of a ‘clock’ and sets circadian clocks apart from other kinds of observable biochemical rhythms such as the cell cycle. However, clocks are not temperature insensitive; they are universally associated with the perception of light and temperature cues that provide time information about the environment. This allows internal time to be appropriately phased with respect to external time so that appropriate biochemical, physiological or behavioral activities, regulated clock output, occur at appropriate times of day. These features – c. 24-h period length, persistence in the absence of environmental cues, a compensation mechanism to keep period length similar under different conditions of temperature and nutrition and ability to use changes in the environment as time signals – are the cardinal characteristics that make a biological rhythm a circadian rhythm.
This review will focus on contributions to the field of chronobiology gleaned from the study of the circadian system in Neurospora crassa, with a detailed look at the central clock protein FREQUENCY (FRQ), its protein–protein interactions, and its regulation through post-translational modification (PTM). While oscillatory behavior has been observed in strains lacking FRQ (see, de Paula et al., 2006; Shi et al., 2007; Schneider et al., 2009), these noncircadian rhythms will not be covered here. Finally, a brief comparison between the fungal and animal clocks will be made to highlight the many conserved features between these two kingdoms.
Introduction to N. crassa
Neurospora crassa has a long history as a genetic model dating from when Bernard O. Dodge recognized the potential of Neurospora genetics and encouraged Thomas H. Morgan and Carl Lindegren to pursue further work (Tatum, 1961). It was through Dodge's enthusiasm that George W. Beadle, along with research associate Edward Tatum, began their biochemical genetic studies by isolating metabolic mutants of Neurospora (Beadle & Tatum, 1941), fully establishing the organism as a modern genetic model. Neurospora is a filamentous fungus classified under the phylum Ascomycota and is widely distributed in nature (Davis, 2000). Neurospora is commonly found in its vegetative state growing as a mycelium of haploid hyphae. The tubular hyphae form incomplete septae, undergo cytoplasmic streaming, and contain multiple nuclei that can be of different genotypes, comprising a heterokaryon. Neurospora crassa undergoes both sexual and asexual reproductive cycles. Upon certain cues, mycelia can undergo a developmental switch and begin to form aerial hyphae that segment into the asexual spores called macroconidia. Macroconidia, also known simply as conidia, are characteristically bright orange due to carotenoid pigments and are strongly hydrophobic. This process of conidia production is controlled in a daily manner and peaks just before dawn.
The physical manifestation of the internal clock through conidial production is perhaps the primary reason why Neurospora helped to found the molecular era of clock research and continues to contribute to our understanding of clock biology. These rhythms manifest as alternating zones of mycelia and conidia production that form a linear pattern down a simple glass tube (so-called race tubes, Fig. 2a). Briefly, a small amount of an agar-based defined growth medium is injected into a hollow tube, and after the medium is solidified, a culture is inoculated with conidia at one end. This culture is allowed to grow under constant light (LL) before transfer to constant darkness (DD) in order to synchronize the clocks in the growing hyphae and set them to subjective dusk. Under these conditions, Neurospora grows with a quasilinear growth rate down the tube where, under control of the circadian oscillator, a developmental switch is thrown into one of two positions. In the late subjective night, the mycelia that are laid down are determined to differentiate and they do so by forming patches of aerial hyphae leading to conidial production. As time and growth proceed, mycelia laid down later in the subjective day and early evening are determined never to differentiate. This growth pattern is determined by the circadian time of day when the mycelia at the growth front are laid down and is fixed once the growth front has passed. Because distance along the tube correlates with time since the light to dark transition, this time history of asexual differentiation potential easily allows the conversion of spatial information into temporal information for the calculation of period and phase (Fig. 2a). This self-reporting assay has facilitated mutant screens that identified many central circadian clock components and thereby contributed to our understanding of the basic mechanism behind all circadian rhythms. More recently, other molecular reporter systems have been developed to visualize and automate the high-throughput recording of rhythms, and even individual molecules, within the clock (Fig. 2b, Morgan et al., 2003; Gooch et al., 2008; Schafmeier et al., 2008; Castro-Longoria et al., 2010; Chen et al., 2010a, b).
The circadian clock in N. crassa
In Neurospora, circadian rhythms were first documented in 1959 (Pittendrigh et al., 1959). These developmental rhythms fulfilled the criteria discussed above for defining them as circadian; they had a sustained period of ∼22.5 h under constant conditions; the phase was set by a single transition from light to dark; they could be entrained by a 24-h light cycle; and the period was not very dependent on ambient temperature. A variety of Neurospora strains that helped to visualize rhythms appeared contemporaneously in the literature, all with rhythmic-inspired nomenclature including patch (Stadler, 1959), clock and wrist watch (Sussman et al., 1965), timex and band (Sargent et al., 1966), all revealing underlying metabolic rhythms through the production of conidia. Interestingly, though, not all of the rhythms exposed in these strains were subsequently shown to be truly circadian in nature and so most of the strains have fallen out of use. However, band proved very useful, and the band phenotype has subsequently allowed the isolation of mutant strains with altered clock properties including period and temperature compensation (Feldman & Hoyle, 1973; Gardner & Feldman, 1980, 1981; Loros et al., 1986). Complementation analysis identified a single locus identified early on by Feldman, called frequency (frq), that displayed long, short, and arrhythmic periods, some of which also altered or disrupted temperature compensation (Gardner & Feldman, 1980; Loros et al., 1986). These early analyses demonstrated the genetic basis of circadian rhythms and showed that a single allele could control many of the core circadian properties in Neurospora.
The cloning of frq in 1986 and its subsequent publication (McClung et al., 1989) opened up the clock in Neurospora to molecular dissection, leading to a concrete understanding of the mechanism behind circadian rhythms (Dunlap, 2008). The molecular mechanism underlying circadian rhythms in this system revolves around a central time-delayed negative feedback loop (Fig. 1b, Aronson et al., 1994). This negative feedback, first demonstrated in Neurospora to be essential for rhythms, is now known to be a conserved feature of all circadian rhythms (Dunlap, 1999). Within the circadian system, positive elements drive the activation of negative elements, which feed back to limit their own activation. Key to this autoregulatory feedback is a built-in system for time delay, thus extending the period to approximately 24 h. Both frq message and a long and a short form of the FRQ protein are rhythmically expressed in a 22.5-h cycle under constant conditions with an approximately 4-h phase difference (Fig. 3, Aronson et al., 1994; Garceau et al., 1997). FRQ is also phosphorylated in a time-dependent manner as more fully described below (Garceau et al., 1997; Baker et al., 2009; Tang et al., 2009). The period of the molecular rhythms seen in frq/FRQ matches that of the conidiation rhythm, frq mutants that alter the banding period also alter the molecular period (Garceau et al., 1997), and conditions that stop FRQ cycles also result in arrhythmic strains. frq expression is regulated by white collar-1 and white collar-2 (Crosthwaite et al., 1997). These GATA-family transcription factors form a heterodimeric complex through interaction with their PAS domains, named the white collar complex (WCC, Linden & Macino, 1997; Ballario et al., 1998; Cheng et al., 2002).
A good way to understand the operation of the clock is to follow the events in time (Fig. 4). In Neurospora, the core oscillation occurs in two major steps (Merrow et al., 1997). Starting in the late subjective night, the WCC binds to the frq promoter to induce frq expression that peaks in the early subjective morning; this is the positive arm in the feedback process (Crosthwaite et al., 1997; He et al., 2002; Belden et al., 2007). FRQ synthesis begins, and within the first 3–6 h of expression, FRQ forms functional homodimers through a coiled-coil domain (Cheng et al., 2001a, b) and enters the nucleus (Luo et al., 1998) to rapidly repress frq transcription via a time-of-day-specific increase in its interaction with the WCC (Denault et al., 2001; Hong et al., 2008; Baker et al., 2009). The interaction of FRQ with the WCC facilitates phosphorylation of the WCC that inactivates it, removing the WCC from the frq promoter, and this negative feedback is further helped by the eventual clearance of WCC from the nucleus (Schafmeier et al., 2005, 2008; He et al., 2006; Hong et al., 2008). As negative feedback reaches completion in the mid to late day, frq expression declines as does the rate of FRQ synthesis; FRQ levels peak in the late subjective day, and the cycle is completed by the delayed release of FRQ-mediated repression, a process that requires the extensive phosphorylation of FRQ over the next 14–18 h (Merrow et al., 1997; Schafmeier et al., 2006). Eventually, FRQ phosphorylation makes it attractive to its ubiquitin ligase FWD-1 (F-box/WD-40 repeat-containing protein, He et al., 2003); FRQ is ubiquitylated and targeted to the proteasome, where it is turned over. The cycle can then restart.
In addition to the negative feedback loop, there is a nested positive feedback loop. Positive feedback was first observed when loss of FRQ was seen to result in reduced WC-1 levels through a post-translational mechanism (Lee et al., 2000; Cheng et al., 2001a, b). This occurs independent of the transcriptional autoregulation of wc-1 and wc-2 (Kaldi et al., 2006; Neiss et al., 2008). The basis for this is that WC-1 is unstable when it is transcriptionally active; that is, WC-1 appears to be turned over as a natural part of its activity cycle in a manner similar to that known for transcription factors in yeast (Tansey, 2001). Thus, early in the cycle when WC-1 can bind to DNA and activate transcription, it is unstable because it is active, but later in the cycle when WCC has been inactivated via phosphorylation and can no longer bind to DNA, WC-1 is stable; hence, its levels increase (He et al., 2005; Schafmeier et al., 2005; Shi et al., 2010). This stabilization is thus observed late in the negative cycle when FRQ levels are high and hyperphosphorylated (Schafmeier et al., 2006) and is an indirect result of FRQ's repressive activity on the WCC. This continued inhibition is in part mediated through subcellular location by maintaining inactive WCC in the cytosol and also keeping chromatin in a repressed state at the frq promoter (Belden et al., 2007; Cha et al., 2008; Schafmeier et al., 2008). Interestingly, WC-1 can be found with the promoter throughout the day, even when it is inactivated, whereas WC-2 binds cyclically (Belden et al., 2007).
Throughout this cycle, FRQ interacts with several proteins other than the WCC, many in a phase-dependent manner (Hong et al., 2008; Baker et al., 2009). Upon translation, FRQ binds stoichiometrically to FRH (for FRQ-interacting RNA helicase, Cheng et al., 2005). FRH is essential for negative feedback in the Neurospora clock as strains with disrupted functions are arrhythmic and express high levels of frq as a result of uninterrupted WCC binding to the frq promoter (Guo et al., 2009; Shi et al., 2010). FRH was first identified biochemically, as the name suggests, through its interaction with FRQ (Cheng et al., 2005) and its importance later was confirmed and clarified through a genetic screen (Shi et al., 2010). Together, FRQ and FRH form the FRQ–FRH complex, FFC, which acts as part of the negative arm of the clock. Formation of the FFC is independent of FRQ dimerization, but is crucial for the stability of FRQ as knockdown of FRH decreases the levels of FRQ (Cheng et al., 2005; Guo et al., 2010). Interestingly, FRQ's degradation in the absence of FRH does not require FWD-1 (Guo et al., 2010), potentially suggesting a structural component to FRQ stability. FRH also interacts with the WCC in the absence of FRQ (Cheng et al., 2005), but, cryptically, FRQ mutants that lack FRQ–FRH interaction also block FRH–WCC interaction (Guo et al., 2010). In Neurospora, FRH is essential; however, one nonlethal allele frhR806H results in arrhythmicity while maintaining normal growth (Shi et al., 2010). This shows that the essential role of FRH for survival can be functionally separated from its role in the core circadian clock. This point mutation results in a protein that maintains interactions with FRQ, but yields a FRQ/FRH complex that can no longer meaningfully interact with the WCC to bring about its inhibition. WCC levels are thus very low in strains bearing frhR806H while frq levels are high, consistent with the model where transcriptionally active WC-1 is turned over rapidly (Guo et al., 2010; Shi et al., 2010).
FRH is homologous to the yeast RNA-binding protein Dob1p/Mtr4p. Mtr4p is a component of the TRAMP complex involved in RNA metabolism and turnover and interacts with the exosome complex. While no clear evidence exists that FRH binds the Neurospora homologs of the TRAMP complex, it does bind the exosome components RRP6 and RRP44 and is proposed to play a conserved role in the Neurospora exosome (Guo et al., 2009). Knockdown of the exosome catalytic subunit RRP44 also results in loss of the overt conidiation rhythm and results in long-period rhythms in FRQ protein. FRH can also bind frq mRNA and influence its message stability and poly(A) tail length (Guo et al., 2009). These data suggest an added layer to the mechanism of negative feedback in addition to simple inhibition of the WCC through post-transcriptional activity directly on frq.
Post-translational modifications in circadian rhythms
FRQ has no known enzymatic activity. This leads to the prediction that, in agreement with the protein interactions described above, the primary function of the post-translational modifications (PTMs) of FRQ may be to regulate its use as a platform for recruiting, in a temporal fashion, other factors involved in the negative feedback process required to sustain rhythms (He et al., 2006; Baker et al., 2009). The circadian circuitries for both the positive and the negative complexes are broadly controlled by PTMs (Mehra et al., 2009a, b). As mentioned previously, phosphorylation of the negative factors is typical, rhythmic, and coincident with their degradation, and it appears to be the major contribution determining the time constant of the feedback loop. Phosphorylation of WCC components also plays a role in the positive loop, where it helps to regulate DNA binding and transcriptional activation. While other PTMs such as ubiquitination, sumoylation, and acetylation are involved in the circadian clock in other systems (Mehra et al., 2009a, b), phosphorylation and the kinases and phosphatases that regulate this are the primary controllers of circadian period length in all systems. Indeed, even the prokaryotic circadian system in Synechococcus is based on phosphorylation and can cycle without the aid of transcription or translation (Nakajima et al., 2005). Our understanding of phosphorylation has expanded to include both degradative and nondegradative consequences, which are often dependent on where and when the protein is phosphorylated.
Phosphorylation of FRQ
FRQ is phosphorylated immediately after synthesis and becomes increasingly so throughout the circadian day in a remarkably choreographed manner (Garceau et al., 1997; Baker et al., 2009). This phosphorylation is regulated, as the shifts in the mobility of FRQ on SDS gels caused by phosphorylation are limited in number, are phase specific, and are reproducible in magnitude across experiments (Fig. 3a). FRQ is also phosphorylated in constant light, although under these conditions, FRQ is heterogeneously phosphorylated across a range of >85 individual locations (Baker et al., 2009; Tang et al., 2009). Mechanistically, phosphorylation primarily influences FRQ stability and period length, as most mutations that reduce the level of phosphorylated FRQ also tend to increase its stability and lead to long periods (Liu et al., 2000; Gorl et al., 2001; Ruoff et al., 2005; He et al., 2006). By the end of the circadian day, the longest unmodified region of the 998 amino acid full-length FRQ is <150 amino acids. Thus, it is probable that by the end of the FRQ phosphorylation cycle, the protein is in a very different conformation than when it was first translated, consistent with a model where FRQ serves as a time-dependent platform to facilitate interactions among clock components.
Experiments designed to understand the roles of different domains are consistent with this view. FRQ contains two PEST domains characterized by their amino acid content (Fig. 3c, Merrow & Dunlap, 1994). Deletion or site-directed mutation of PEST-1 leads to a reduction in FRQ phosphorylation and degradation as well as loss of conidial banding (Gorl et al., 2001; Baker et al., 2009; Tang et al., 2009). Among sites in PEST-2 phosphorylated in vitro are S885 and S887 (Schafmeier et al., 2006). Mutation of these sites to asparagine to prevent phosphorylation resulted in loss of circadian rhythmicity on race tubes and had no effect on FRQ turnover or negative feedback. Instead, it was proposed that these sites help to stabilize the positive arm through a nested positive feedback loop. However, this nondegradative function is not entirely clear as the mutation of S885 and S887 to alanine resulted in a no-clock phenotype (Tang et al., 2009). The phosphorylation status of FRQ is also related to the subcellular steady-state distribution of FRQ (Diernfellner et al., 2009). As FRQ progresses through the cycle and picks up increasing phosphate, it switches from being predominantly nuclear, where it represses WCC activity, to predominantly cytoplasmic (Luo et al., 1998).
Recently, the time course of FRQ phosphorylation has been followed through the use of heavy isotope labeling in conjunction with tandem MS (Baker et al., 2009). In general, many individual phosphorylation events seem to lack unique roles. Instead, most phosphorylations occur in spatial and temporal clusters, with the result that they probably form charged regions on the FRQ protein, and these lead to changes in the activity of FRQ. FRQ is of course unmodified immediately after its synthesis, but phosphorylation events occur rapidly in the middle part of the proteins between the PEST-1 and the FFD domain (Fig. 3c). The roles of these events are not well understood as mutational analysis did not alter circadian rhythms at steady state. Soon after this, phosphorylations are clustered near to the C-terminus, and serine to alanine mutations of these residues, or deletion of the entire C-terminus (FRQSHORT), results in FRQ protein that is less stable and that supports a short-period rhythm (Baker et al., 2009; Tang et al., 2009). This indicates that phosphorylations act here to stabilize FRQ. During the middle of the cycle, the PEST-1 domain shows a drastic increase in phosphorylation. Functional studies into the role of phosphorylation at this region showed a period length that is longer than normal and more stable FRQ, suggesting that phosphorylation of these residues is needed to promote turnover of FRQ. Still later other domains of FRQ, predominantly residues specific to the long FRQ isoform (Fig. 3), become modified and serine to alanine mutations in this region result in long-period clocks, suggesting again a role in promoting turnover. In all, the sequential modifications appear to be acting to modulate the structure of FRQ with the goal of regulating interactions with other proteins as further described below, among these the ubiquitin ligase FWD-1 that will lead to the turnover of FRQ.
Phosphorylation of the WCC
WC-1 and WC-2 are both regulated by phosphorylation and several sites on these proteins have been identified using methods in MS (He et al., 2005; Huang et al., 2007; Baker et al., 2009; Sancar et al., 2009). In the dark, WC-2 is phosphorylated rhythmically throughout the circadian day, with maximal phosphorylation around late subjective day (CT8–12, Schafmeier et al., 2005). Because WC-2 levels are nearly constant, the cycle in phosphorylation must be mediated by kinases and phosphatases. This light-independent phosphorylation is dependent on FRQ and dephosphorylation of the WCC increases its binding to DNA as noted above (He et al., 2005; Schafmeier et al., 2005). Although to date MS/MS analysis of WC-2 has only identified one phosphorylated site at S433 (Baker et al., 2009; Sancar et al., 2009), two-dimensional gels of WC-2 suggest at least eight possible phosphorylation states (Schafmeier et al., 2005). Mutation of S433 to alanine results in a moderate decrease in circadian period length and a slight increase in transcriptional activity (Sancar et al., 2009). WC-1 is also phosphorylated in the dark, but it is not clear whether this is time dependent (He et al., 2005). Upon exposure to light, WC-1 becomes hyperphosphorylated after 15 min, a time corresponding to a reduction in light-induced transcription and degradation of WC-1 (Talora et al., 1999). To date, at least 10 phosphorylation sites have been identified on WC-1 (He et al., 2005; Baker et al., 2009; Sancar et al., 2009), with at least six clustered in the C-terminus of WC-1 near its DNA-binding domain (He et al., 2005). Phosphorylation of this region is sequential, FRQ dependent, and required for negative feedback (Huang et al., 2007). This shows that the core negative feedback loop of FRQ inhibiting its own transcription is closed around FRQ, promoting phosphorylation of the WCC.
Regulatory enzymes in the clock
All of the phosphorylation events that regulate the ticking of the clock require the activity of kinases and phosphatases. Molecular dissection has revealed several important players in these processes. Kinases included on this list are the acidic-directed casein kinases 1 and 2 (CK1a and CK2), the Neurospora homolog of checkpoint kinase-2 (PERIOD-4, PRD-4), as well as CAMK-1, and the basophilic protein kinase A (PKA, Gorl et al., 2001; Yang et al., 2001, 2002, 2003; Pregueiro et al., 2006; Huang et al., 2007; Mehra et al., 2009a, b). FRQ physically interacts with several of these kinases including CK1a, CK2, and PRD-4 (Gorl et al., 2001; Pregueiro et al., 2006; Baker et al., 2009). This group of kinases is joined by several known phosphatases including protein phosphatase-1 (PP1), PP2a, and PP4 (Yang et al., 2004; Cha et al., 2008). Many of these same enzymes play similar regulatory roles in animal circadian clocks (Table 1).
|Negative arm||Phosphoprotein scaffold||FRQ|
|Positive arm||PAS-domain transcription factor||WC-1||CYC||BMAL|
|Regulatory enzymes||Protein kinase||CK1a||DBT||CK1δ/ɛ|
|Degradation (F-box domains)|
Out of the kinases, several lines of data suggest that FRQ has the most intimate physical interaction with casein kinase 1 (Gorl et al., 2001; He et al., 2006; Querfurth et al., 2007; Baker et al., 2009). Neurospora contains two casein kinase 1 orthologs: ck1a and ck1b. Of these, ck1a is more similar to the mammalian CK1ɛ, functionally conserved in mammalian circadian rhythms. CK1a participates in the regulation of both FRQ and the WCC, whereas CK1b appears to play no role in the clock (Gorl et al., 2001; He et al., 2006). FRQ interacts with CK1a through the FRQ–CK1a interacting domain (FCD, He et al., 2006) that is positioned on FRQ just N-terminal to the heavily phosphorylated PEST-1 domain responsible for FRQ degradation (Fig. 3c, Gorl et al., 2001; Baker et al., 2009). FRQ acts as a substrate recruiting scaffold by bringing CK1a to the WCC (He et al., 2006; Baker et al., 2009). FRQ–CK1a interaction is continuous throughout the circadian day, but increases specifically early in the cycle when hypophosphorylated FRQ is in complex with the WCC (Baker et al., 2009). This timing is consistent with the role of CK1a in phosphorylating the WCC, leading to its inactivation and repression (Schafmeier et al., 2005; He et al., 2006). FRQ is also a target of CK1a phosphorylation. FRQ can be phosphorylated at a minimum of 41 serine/threonine by CK1a in vitro (Tang et al., 2009) and many of the same sites are found to be phosphorylated in vivo (Baker et al., 2009; Tang et al., 2009). Introduction into the Neurospora ck1a of a mutation shown in Drosophila to increase period in flies also increases period length in Neurospora. Because circadian period length is tied to the FRQ degradation rate, this result suggests that FRQ is a direct target of ck1a and reveals a conserved mechanism across widely divergent phyla (He et al., 2006). Furthermore, placing just the FCD and PEST-1 regions of FRQ into green fluorescent protein is sufficient to increase the degradation rate of the recipient protein (Querfurth et al., 2007), consistent with the idea that CK1a binds to and phosphorylates FRQ, leading to its turnover through these domains. CK1a describes several different protein isoforms that vary in their C-terminal tail length as a result of mRNA splicing. At least two major isoforms interact with FRQ, but are not found in complex with each other (Querfurth et al., 2007). Because FRQ functions as a dimer, with both molecules containing an FCD, it becomes an interesting question as to how these protein complexes are assembled.
Casein kinase 2 was biochemically purified from Neurospora extracts based on its ability to phosphorylate FRQ (Yang et al., 2002). CK2 is functions as a heterodimer, with one subunit encoded by a single catalytic domain, cka, and one of two regulatory domains encoded by ckb-1 and ckb-2. Of the two regulatory subunits, only ckb-1 is known to be involved in the circadian clock (Yang et al., 2003), whereas ckb-2 appears to play no role in the clock. CK2 physically associates with FRQ, although this interaction appears to be weak and possibly transient, suggesting that FRQ is primarily a substrate of CK2 (Baker et al., 2009). Mutation of cka abolishes rhythms in frq and several ccgs in constant darkness (Yang et al., 2003). Similar to CK1a, many of the phosphorylation sites identified on FRQ in vivo can also be phosphorylated by CK2 in vitro (Tang et al., 2009). Mutation of in vivo identified phosphorylation sites on FRQ that conform to the predicted motif of CK2 also resulted in lengthening of the circadian period (Yang et al., 2003; Baker et al., 2009; Tang et al., 2009). CK2 is unique in its regulation of FRQ stability, differing from CK1 in one major functional role. Cloning of two classic circadian mutants defective in temperature compensation, period-3 and chrono, identified separate point mutations in both cka and ckb-1, respectively (Mehra et al., 2009a, b). Mutation of specific residues on FRQ phosphorylated by CK2 in vitro phenocopies the chrono circadian behavior across a range of temperatures, thus identifying the first molecular process involved in regulating temperature compensation.
Finally, whenever there is kinase activity, phosphatases are readily available to counterbalance the equation. PP1 and PP2a represent two major families of protein phosphatases widely conserved in eukaryotes (Gallego & Virshup, 2005), and both are important for circadian rhythms. The mutation of ppp-1 (the catalytic subunit of PP1) destabilizes FRQ and results in a slightly reduced period, while PP2a activity influences frq transcription (Yang et al., 2004). As mentioned, WC-2 undergoes rhythmic phosphorylation while the levels of the protein do not cycle, and the mutation of rgb-1, the catalytic subunit of PP2a, causes constitutive hyperphosphorylation of WC-2 (Schafmeier et al., 2005). A recent report showed that WC-2 rapidly shuttles between the nucleus and the cytoplasm and that this is required for dephosphorylation by rgb-1 of inactive, hyperphosphorylated species in the cytoplasm (Schafmeier et al., 2008). One last PP2a-related phosphatase, PP4, plays a similar role in regulating WCC localization and transcriptional control. Knockout of PP4 also results in a short rhythm and hyperphosphorylated FRQ (Cha et al., 2008). Which of these proteins plays the major role in WCC localization is still under debate and it is possible that both contribute equally because loss of function of neither rgb-1 nor pp4 completely abolishes circadian rhythmicity.
Input and output in the clock
The central clock components described above are also sufficient to explain the molecular bases of other features of the oscillator such as input and output. WC-1 is also a blue-light photoreceptor and responsible directly or indirectly for all known light responses in Neurospora (Ballario et al., 1996; Froehlich et al., 2002; He et al., 2002). Light activation results in the WCC directly binding to and inducing transcription of hundreds of genes (Chen et al., 2009; Smith et al., 2010). WC-1 light activity is due to its association with flavin adenine dinucleotide, which becomes covalently linked to the conserved light-sensing domain (LOV domain for light oxygen voltage. For a comprehensive review of photosensing in Neurospora, see Dunlap & Loros, 2005; Chen et al., 2010b). Light induction by the WCC drives an increase in frq levels. Additionally, because frq cycles, the impact of light on the oscillator is dependent on time. When frq levels are low and on the rise (before subjective dawn), brief light pulses will drive frq mRNA to peak levels and thereby advance the clock to the time corresponding to the highest frq expression, mid to late morning. Conversely, when frq levels are decreasing, light exposure will again induce frq mRNA to peak levels, this time setting the clock back, leading to phase delays. This provides a molecular explanation for light entrainment and phase resetting (Crosthwaite et al., 1997).
Input to the clock can be modified by VVD, a small (21 kDa) flavin-binding blue-light photoreceptor consisting of a PAS domain variant called a LOV domain and an N-terminal cap (Heintzen et al., 2001; Zoltowski et al., 2007) that is responsible for conferring photoadaptation in Neurospora (Heintzen et al., 2001; Schwerdtfeger & Linden, 2001, 2003; Shrode et al., 2001). Photoadaptation refers to the phenomenon in which, following light exposure in wild-type Neurospora, elevated levels of light-induced gene transcription are transient and generally revert to preinduction levels within 2–4 h, and Neurospora can detect changes in light intensity rather than simply seeing ‘on vs. off’ (Schwerdtfeger & Linden, 2001, 2003). The molecular basis of photoadaptation is that VVD is rapidly and highly light induced; light also causes the N-terminal cap to come off the LOV domain, with the result that, in solution, VVD forms a rapidly exchanging dimer in light (Zoltowski et al., 2007). This suggested that the VVD LOV domain might interact in vivo with PAS domains from other proteins like those in WC-1 or WC-2 (Zoltowski & Crane, 2008; Lamb et al., 2009), and recent studies have shown this to be the case: light-induced VVD rapidly moves to the nucleus, where it physically interacts with the WC-1 in the WCC to reduce its ability to activate transcription, and because VVD induction is graded with increasing light, incremental light exposures yield more VVD to inactivate the newly activated WCC, in all cases leading to photoadaptation (Chen et al., 2010a, b; Hunt et al., 2010; Malzahn et al., 2010). VVD is itself clock regulated, which leads to circadian gating of light response (Heintzen et al., 2001), and along with repressing the WCC activity in light, VVD modulates various WCC-mediated circadian clock properties to keep the light response regulated and the clock running, appropriately entrained. Specifically, VVD allows the clock to run through the dawn transition and take its principal phase cues from dusk (Elvin et al., 2005), and it contributes to temperature compensation of the circadian clock phase (Hunt et al., 2007). A nice added touch is that although VVD decays with darkness, the elevated production of VVD during the day ensures that some remains at night to inactivate any WCC that is light induced due to moonlight, thereby ensuring that the clock can run (Malzahn et al., 2010).
Temperature can also act as an environment input to entrain the clock; however, it seems to have an influence primarily on FRQ (Liu et al., 1997; Pregueiro et al., 2005). At different ambient temperatures, the average level of frq expression and amplitude of frq cycling is quite similar, whereas FRQ levels are seen to cycle around a higher mean level at higher temperatures. This means of course that the absolute number of molecules of FRQ corresponds to one time of day at one temperature, but corresponds to a different time of day at another temperature: if a culture is rapidly shifted between temperatures, FRQ levels will not change as quickly as the temperature; thus, the new time of day after the shift will be the time corresponding to that level of FRQ at that temperature. As we saw before, the low point of FRQ is near to dawn, and in practice, the high point of FRQ when the clock is running at about 20–22 °C is still lower than the low point when the clock is running at 28 °C. Thus, explicitly then, after a discreet shift up in temperature from 21 to 28 °C at any time of day, the clock will be reset to subjective dawn, the new phase representing the low point in the FRQ cycle at the new temperature (Liu et al., 1998).
Another influence of temperature on FRQ and the clock is seen at the post-translational level. frq pre-mRNA undergoes complex temperature-sensitive alternative splicing in the 5′UTR that produces either a short (sFRQ) or a long (lFRQ) form of the protein differing by 100 amino acids (Fig. 3, Liu et al., 1997; Colot et al., 2005; Diernfellner et al., 2007). Under standard laboratory conditions, both of these proteins are expressed together, but at low temperatures (∼20 °C), there are roughly comparable levels of sFRQ and lFRQ, while high temperatures (∼28 °C) result in relatively more lFRQ as well as more FRQ altogether. Either FRQ form can support circadian rhythmicity, although both are required for robust rhythms across the typical temperature range of 20–30 °C (Liu et al., 1997). The mutation of key phosphorylation sites that are found solely in lFRQ can decrease period length, even in the presence of sFRQ, to that found in strains only expressing sFRQ (Diernfellner et al., 2007; Baker et al., 2009). However, the function of these lFRQ-only phosphorylation sites is currently not well understood.
Output refers to the regulatory pathways and mechanisms whereby time information generated by the clock is used to regulate the time-of-day-specific expression of the overt rhythms in the cell. It is thus of singular importance because output is the true source of the biology that makes rhythms themselves important, and yet it remains poorly understood at the molecular level (Vitalini et al., 2006). Recalling from above that the operation of the core clock results in cyclical activation and deactivation of the WCC transcription factor, circadian output is best understood largely as a consequence of the daily cycle of gene expression brought about by this negative feedback in the central clock, where it impacts the expression of genes that do not participate in the feedback loop. In fact, in addition to frq, a large number of transcripts have been identified that undergo circadian regulation. These are referred to as clock-controlled genes (ccgs, Loros et al., 1989) and they can loosely be thought of as primary ccgs that are acted upon directly by the WCC and secondary ccgs that are directly regulated by factors regulated directly or indirectly by the WCC and are thus further downstream from the clock. Primary ccgs share the conserved DNA-binding motif for the WCC and are regulated directly through the WCC binding to their promoter. An additional level of regulation can occur when the FRQ–FRH complex interacts with mRNA and the exosome through FRH (Guo et al., 2009). This allows control at the mRNA level through the regulation of message stability rather than production.
Recently, a genetic screen was undertaken with the objective of identifying components of the output pathway. This effort uncovered several new alleles of rrg-1, a member of the p38-type MAP kinase involved in the osmotic stress (Vitalini et al., 2004, 2007). Osmotic sensing-2, OS-2, is a downstream kinase in this pathway and undergoes circadian-regulated phosphorylation leading to its activation that is dependent on rrg-1. The circadian rhythm in the phosphorylation of OS-2 is also dependent on FRQ (Vitalini et al., 2007). This system of daily activation of a kinase cascade sets up an output pathway connected to the core clock that may allow anticipation of osmotic stress associated with changes in the environment. This type of output regulation through phosphorylation also raises the possibility that the FFC in association with CK1a might have targets other than just the WCC. These examples provide a direct transcriptional and a potential post-transcriptional link between the core clock and output.
Conserved mechanisms across phyla
Concurrent with the effort to unravel the N. crassa molecular clock was a concerted effort focused on Drosophila, and in the mid 1990s, genetic screens and homology-based searches identified potential components and drove work on vertebrate clocks. In Drosophila, the first genetic screen for circadian mutants revealed a single locus, named period, having short, long, and arrhythmic phenotypes (Konopka & Benzer, 1971). Subsequent cloning and characterization of per (Bargiello & Young, 1984; Reddy et al., 1984) found that, similar to frq, the mRNA and protein are expressed in a rhythmic fashion (Siwicki et al., 1988; Hardin et al., 1990). PER also undergoes time-dependent multisite phosphorylation (Edery et al., 1994). Vertebrates have three homologous per genes, all of which are involved in the negative complex and are the target of timed phosphorylation (Tei et al., 1997; Bae et al., 2001; Lee et al., 2001). Similar to FRQ, the PER phosphoproteins in both flies and vertebrates are found in complex with other proteins required to execute negative feedback (Table 1). These binding partners are divergent among animal models (Gallego & Virshup, 2007). Similar to the Neurospora negative complex, animal PER proteins are also found associated with CK1 (Gallego & Virshup, 2007). Therefore, in all animal and fungal systems, a primary phosphoprotein mediates circadian period length, is found in complex with a divergent binding partner, and acts as a scaffold for interaction with CK1.
In addition to similarities in the organization of the negative arm repressor complex, animal circadian systems were found to have striking similarities in the positive arm of the feedback loop (Table 1). per expression in both Drosophila and mammalian systems is induced by a PAS domain containing heterodimeric transcription factor complex (Dunlap, 1999). One of these transcription factors, CLOCK, was first identified in mice and is part of the positive arm in both flies and mammals (King et al., 1997; Allada et al., 1998). In vertebrates, CLOCK binds BMAL1 (Hogenesch et al., 1998), a protein that shares sequence homology to WC-1 and that occupies the same place and plays the same role in the mammalian clock that WC-1 plays in Neurospora (Lee et al., 2003). In flies, CLOCK is found in association with CYCLE, an ortholog to BMAL1 (Rutila et al., 1998). These transcription factors are inhibited by the negative complex containing the PER proteins and their binding partners in the manner just described for the Neurospora feedback loop (Dunlap, 1999). Therefore, while some of the exact constituents of the positive and negative complexes are not orthologous proteins, the functional logic of the clock architecture is conserved.
While the logic is conserved, it is interesting to see the degree of variation among the components. Mammals use PER proteins in the negative complex, but the principal repressors in this complex are the CRY proteins that, in Drosophila, serve only as photoreceptors, with no role in the oscillator itself (Dunlap, 1999). Among the fungi, WC-1 and WC-2 are broadly and highly conserved in the fungi, but FRQ is much less conserved (Dunlap & Loros, 2006; Salichos & Rokas, 2010); thus, it is likely that other protein(s) assume the role of a negative element. However, the conserved nature of the regulatory logic suggests that the negative element(s) will work in a manner similar to that of FRQ and PER.