Epigenetic and Posttranslational Modifications in Light Signal Transduction and the Circadian Clock in Neurospora crassa

Exposure to environmental light stimulates physiological responses in all living organisms, from prokaryotes to eukaryotes [1]. Studies in plants have identified photoreceptors that respond to two types of light signals: blue light and red light [2,3,4]. Although plants were the first organisms in which molecular mechanisms of light signal transduction were characterized, the Ascomyceta filamentous fungus Neurospora crassa represents the most used model system. N. crassa is sensitive only to blue light, although genes coding for other putative photoreceptors have been identified [5,6]. Nevertheless, it seems that the deletion of these genes does not cause aberration in response to blue light stimulation [7]. Blue light exposure induces many different physiological responses in N. crassa, including entrainment, biosynthesis of photo protective pigments, induction of asexual conidiospores formation, phototropism of peritecial beaks, development of sexual structures, and the direction of ascospore dispersal [8,9,10,11,12,13]. Several genes coding for proteins activated in response to light stimulation have been identified [14,15]. Genome-wide analysis shows that approximately 5.6% of the protein-coding genes in N. crassa are responsive to light [16]. Light-sensitive genes are grouped in two large classes: early light-responders (ELR), with average peaks in mRNA expression at approximately 15 to 30 min, and late light-responders (LLR) in which mRNA expression peaks between 60 and 90 min [17].

The light signaling system is regulated by the White Collar complex (WCC). The WCC consists of a heterodimer formed by the product of the white collar-1 (wc-1) and white collar-2 (wc-2) genes [18,19]. WC-1 and WC-2 contain Per-ARNT-Sim (PAS) domains, which serve as versatile sensor and interaction modules in signal transduction for protein-protein interaction [20]. The WC-1 first PAS is a modified domain known as the light-oxygen-voltage domain (LOV), which is a PAS domain variant initially found in phototropins, a family of blue-light sensitive plant photoreceptors (Figure 1a) [21]. This domain is able to bind chromophores such as Flavin Adenine Dinucleotide (FAD) makingWC-1 a “photoreceptor”: when a chromophore captures photons, it generates a photo-adduct with the LOV domain, which changes WC-1 conformation and promotes signal transmission [22]. Like its molecular partner WC-1, WC-2 shares the same functional domains except the LOV domain (Figure 1b).

WC-1 heterodimerizes with WC-2 to form the WCC, which then translocates into the nucleus where it mediates conformational changes when a light signal is perceived and activates the transcription of hundreds of target genes [23]. Both WC-1 and WC-2 proteins can bind promoter sequences called Light Responsive Elements (LREs) by their zinc-finger motifs. The LREs promoter sequences contain GATC/G consensus sequences. These regions were recognized by in vitro binding assays using recombinant WC-1 and WC-2 zinc-finger domains or activated WCC [24]. WCC is able to homodimerize via the activated LOV domain and binds preferentially to tandem GATC motifs in the presence or absence of light [24,25]. Nevertheless, within only 15 min after light pulse, WCC transiently activates early light-responsive genes such as albino-3, vivid, and frequency triggering a long-term mechanism [26,27]. The transcription factor SUB-1 supports recruitment of -light-activated WCC to some promoters of its own target genes [25]. Moreover, many light-regulated genes are under delayed transcriptional control mediated by WCC and are expressed several hours after light induction [28,29,30]. WCC also acts as an inhibitor of wc-2 transcription [26]. This inhibition is regulated via an unidentified repressor of wc-2 transcription, which is expressed under the control of WCC [31]. After exposure to a light pulse, another factor that can modify WCC activity is Vivid, which is rapidly transcribed and translated. VIVID (VVD) is a small flavin PAS/LOV protein that binds blue-light photoreceptors [32,33]. It is involved in photoadaptation in Neurospora. Vivid inhibits WCC in a light-dependent fashion. The light-activated VVD acts as a competitive inhibitor of WCC homodimerization via its own LOV domain. It dampens the level of oscillating light-dependent transcripts, reverting them to the pre-induction phase (constant darkness) within 2 to 4 h after light exposure [34,35].

The current model describing the molecular mechanism driving the circadian rhythm in N. crassa consists of interlocked negative and positive feedback loops involving the clock genes frq, wc-1, and wc-2 [46]. In constant darkness, WCC (D-WCC) acts on the positive loop and activates the transcription of the frq gene encoding the FRQ protein which binds its own C-box [50,51,52]. After frq gene transcription, frq mRNA reaches its peak during the subjective day, and FRQ protein peaks 4 to 6 h later. It was found that frq mRNA possesses another start codon located 100 codons downstream of the first ATG: alternative translation initiation produces a 135-kDa protein (889 amino acids) that lacks the normal N-terminal domain [53]. Short and long FRQ (SFRQ, LFRQ) are required for optimal clock function [54]. LFRQ is the core of the circadian negative loop [55]. FRQ contains some important domains among its sequence [56] (Figure 2a).

After importation into the nucleus, FRQ homodimerizes via the N-terminal coiled coil domain and forms a functional homodimer [56]. LFRQ is also involved in temperature compensation of the clock [57,58]. Throughout this cycle, FRQ interacts with several proteins including FRH homologous to the yeast RNA-binding protein Dob1p/Mtr4p [59]. FRH associates with and stabilizes FRQ in an ATP-independent manner and it is essential for correct FRQ folding [60] (Figure 2b). In Neurospora the entire pool of FRQ is in complex with FRH; down regulation of frh expression abolishes circadian rhythmicity, resulting in high frq RNA levels. FRH binds the FRQ dimer in a stoichiometric manner to form a complex known as FCC; this complex acts as a part of the negative loop of the clock, inhibiting activity of the D-WCC and frq gene expression [61]. During this phase, FCC acts as a platform that recruits numerous kinases that will phosphorylate WCC and lead it to the degradation pathway (see Section 3.8). When FRQ levels drop below a certain threshold in the subjective late night, D-WCC is no longer inhibited by FRQ, and frq transcription is reactivated to start a new cycle [62]. Moreover, FRH acts non-enzymatically to support the intrinsically disordered FRQ [63]. This succession of positive and negative loops results in the circadian rhythmic oscillation of frq mRNA and protein, which is critical for the normal circadian behavior of an organism [64]. Moreover, FRH overexpression promotes WC-1 accumulation, confirming that FRH, together with FRQ, plays a role in stabilizing WC-1 [60]. Indeed, FRH-FRQ interaction is essential for maintaining correct WC protein levels in the positive feedback loops. Besides its role in repressing its transcription, FRQ also positively regulates the levels of both WC-1 and WC-2 through different mechanisms [65]. In the cytoplasm, FRQ acts as a positive factor by supporting WC-1 protein accumulation, leading to the formation of the WCC via an unknown mechanism [66]. The C-terminal part of FRQ is fundamental for its cytosolic localization [67]. This suggests that there are two different kinds of FRQ-mediated regulation of the circadian clock [68]. As said before, another light pulse can induce light-dependent frq mRNA to reset the clock when the frq level decreases. The Light/Dark transition is a highly reliable method for resetting circadian rhythm. Although the WCC/FRQ complex represents the core of the circadian system in Neurospora, other factors are also involved in this regulatory process.

FRQ is progressively phosphorylated during the day. Several evidences support that FRQ phosphorylation is related to its daily turnover [52,90]. First, in vivo phosphorylation is blocked after treatment with 6-DMAP, a general kinase inhibitor, and the total FRQ amount is increased, thus demonstrating the role of phosphorylation in FRQ stability [49]. Moreover, phosphorylation is essential for maintaining “in phase” the Neurospora circadian rhythm, as observed in strains treated with 6-DMAP in which circadian conidiation was aberrant [49]. Second, by systematic mutagenesis of the FRQ open reading frame (ORF), S513 has been identified among different putative phosphorylation sites as the most important. Elimination of S513 leads to a dramatic reduction in the rate of FRQ degradation and a significantly increased period caused by an increased stability of FRQ [49,91,92,93,94,95,96]. Third, over 75 sites have been identified by quantitative proteomics using nanoscale micro capillary liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS or LC tandem MS) and analyzed by stable isotope labeling with amino acids in cell culture (SILAC) quantification [93]. These sites are mapped along the entire FRQ amino acid sequence. FRQ can be subdivided into three main regions: the acidic C-terminal domain, the central domain, and the N-terminal domain that shows basic properties [94]. FRQ shows two PEST-like domains containing proline (P), glutamic acid (E), serine (S) and threonine (T) rich domain: PEST-1 in the central domain and PEST-2 in the N-terminal domain (Figure 2a). These domains are highly phosphorylated. Interestingly, deletion of the PEST-1 domain in FRQ results in an arrhythmic conidiation phenotype. Deletions of PEST-1 or mutations of S513 generate a very stable form of the FRQ protein, suggesting that loss of oscillating FRQ phosphorylation affects Neurospora circadian behavior. Deletion of PEST-2 does not reduce FRQ expression, but it is required for WC-1 stability. Some experimental evidence suggests that WC-1 accumulation is regulated by phosphorylation of FRQ at S513, S885, and S887. Mutation S885, 887 can affect the positive limb of the interconnected feedback loops, while deletion of the entire PEST-2 domain causes a more severe functional defect. PEST domain is involved in protein turnover in several organisms and seems to have the same function in FRQ [95]. The kinetics of FRQ oscillation is spatially and temporally coordinated by phosphorylation [96]. After the light/dark shift that leads Neurospora circadian entrainment (time point 0), a high amount of FRQ is hypophosphorylated in constant darkness (DD) up to 14 h after light synchronization (DD14, corresponding to subjective morning). A low amount of FRQ protein is highly phosphorylated 24 h after light synchronization (DD24, corresponding to subjective night). This means that at DD14 the total amount of FRQ is higher than the total amount at DD24, suggesting that phosphorylation is associated with the degradation pathway [23]. Newly synthesized hypophosphorylated FRQ contains a nuclear localization signal (NLS) and is trapped in the nucleus more efficiently than hyperphosphorylated species [56,66]. Despite the presence of a NLS, only a small fraction of FRQ is localized in the nucleus and the majority of the protein accumulates in the cytosol [23,96]. FRQ phosphorylation inhibits nuclear import, and retains the proteins in the cytoplasm where it acts as a positive factor by supporting the WC-1 protein accumulation, as said before. Since phosphorylated FRQ exert diverse functions such as support of WCC accumulation in the cytosol and inactivation of WCC activity into the nucleus, it indirectly influences the daily oscillation of hundreds of genes that are under the control of the WCC as well.

Recent observations provided evidence that FRH is critical for the maintenance of FRQ phosphorylation profile and stability [67]. Loss of FRQ-FRH interactions, or down-regulation of FRH, promotes the hypophosphorylated form of the FRQ:FRQ altering the normal circadian pathway [97]. Many kinases are involved in FRQ regulation. In the following paragraph, we will describe the principal kinases involved in the full length FRQ phosphorylation.

Actually there are described at least 113 phosphorylated sites identified in FRQ [93,101]. MS analyses showed that the full-length FRQ phosphorylation is favored when FRQ is preliminarily phosphorylated in some specific sites, suggesting that it is regulated by a specific mechanism for phosphorylation. One explanation of this process is that newly synthesized, hypophosphorylated FRQ adopts preferentially the closed conformation supported by electrostatic interactions of the positively charged N-terminal domain with the negatively charged remainder portion of the protein [115]. Early phosphorylation of some sites in the C-Term region and middle portion could assist to strengthen the closed state that allows stabilization of FRQ.

Subsequently CKIa and CKII start to phosphorylate the N-terminal of FRQ, masking the basical properties of the N-terminal domain. As a result, N-terminal domain can no longer interact with the C-terminal acid domain. This event causes a conformational change in FRQ that exposes the S513 and the PEST-1 sequence of the central domain. Then CKIa, CKII, and CAMK-1 phosphorylate the PEST sequences, the S513 residue, and many other residues dislocated along the entire protein, except for the C-terminal domain, which is never phosphorylated. This means that the circadian oscillation of FRQ phosphorylation is due to multiple reactions throughout the FRQ structure and in the spatial and temporal combination of these kinases [116]. The FRQ phosphorylation pathway is illustrated in Figure 2. The combination of these phosphorylation processes leads FRQ to the degradation pathway. Some findings indicate that hyperphosphorylated FRQ is degraded by proteasome. First, ophiobolin A, a well-known calmoduline inhibitor tested to identify CAMK involvement in FRQ phosphorylation, promotes protein ubiquitination and the formation of higher molecular weight conjugates of FRQ in vivo [117]; Second, disruption of the fwd1 gene (homolog of the Drosophila Slimb protein), which encodes a protein containing F-box and VD4causes hyperphosphorylation of FRQ with a pattern different from that observed in wild-type strains [118]. FWD, a member of the SCF-type ubiquitin ligase complex, seems to directly ubiquinate FRQ. Hence, FWD mutation decreases the FRQ degradation rate because the protein appears to be more stable than in the wild-type strain; Third, the accumulation of FRQ in the fwd-1 mutant strain is associated with dysfunctions in circadian oscillation though the Dark/Light transition still occurs; Fourth, FRQ physically interacts with FWD-1 in vivo, driving FRQ to the degradation pathway, as observed in the fwd-1 mutant that was still able to bind FRQ but lacked the F-box domain involved in FRQ degradation. In summary, these data indicate that FRQ phosphorylation promotes FRQ degradation through the ubiquitin-proteasome pathway mediated by ubiquitin E3 ligase (Figure 2f).

Evidence suggests that the light-dependent WCC pathway is influenced by protein phosphorylation [119,120]. It has been demonstrated that WC-1 becomes highly phosphorylated 15 min after multiple light pulses and that the phosphorylation level decreases to the same levels as in dark after 2 h [121,122]. Functional WC-2 is not required for PTM of WC-1 but it could influence the degree of PTM and the stability of WC-1. Light-dependent WC-1 phosphorylation oscillation correlates with the expression of light-dependent gene (e.g., al-3). As observed for FRQ, WC-1 phosphorylation is also coupled with its own degradation. Pharmacological studies and in vitro phosphorylation assays have suggested that protein kinase C (PKC) may phosphorylate WC-1 [123] (Figure 3a-2).

In vivo and in vitro experiments have confirmed that PKC regulates light responses affecting light-dependent WC-1 protein stability [124,125]. PKC seems to be able to phosphorylate the WC-1 zinc-finger domain. The zinc-finger domain is important for the function of WC-1 in the dark [26]. PKC physically interacts with WC-1 in dark conditions, but this interaction is no longer detectable after 30 min of light exposure when WC-1 is hyperphosphorylated. This suggests that light-mediated WC-1 hyperphosphorylation affects PKC/WC-1 interaction. Phosphorylation of WC-2 appears to be more stable over time after light exposure [122] and shows strong oscillation when analyzed by two-dimensional electrophoresis [126]. The cytoplasmic WC-2 protein is not modified in response to light, suggesting that light-specific phosphorylation is limited to nuclear pool of WC-2 [122]. Finally, it has been demonstrated that WC-2 light-specific phosphorylation depends on a functional WC-1. The translocation of WC-1 and WC-2 into the nucleus is independent of light-specific phosphorylation, unlike that observed for FRQ phosphorylation [23]. MS analyses identified S443 as the phosphorylation target in WC-2 protein [127] (Figure 3b).

S433 plays a role in the regulation of WCC activity but it does not affect WCC turnover according to mutagenesis experiments. The activity and the stability of WCC were increased and the timing of photoadaptation was anticipated (probably due to more rapid accumulation of the blue-light photoreceptor VVD) in the s433a strain, a WC-2 variant in which S443 is mutated in A, thus preventing WC-2 phosphorylation. In contrast, WCC was less active in the s433d strain in which mutation of Sto Emimics the phosphorylated state of WC-2. The phosphorylation site S433 is located directly upstream of a prolyl residue, suggesting that it could be target of serine/threonine kinases such as MAPK-activated protein kinase (MAPK) and Cyclin-Dependent Kinase (CDK) [128]. Table 2 lists the kinases that target the White Collar Complex.

Recent evidence suggests that WC-1 can be acetylated in vitro and in vivo along the entire sequence [129]. Though the role of WC-1 acetylation is not clear, this PTM seems respond to light signals. Neurospora Gcn Five-1, (the histone acetyl transferase homologous to the Yeast Gcn5p and Human GCN5) could be involved in acetylation of WC-1 as well as histone H3 of light induced gene promoters [130]. However, analysis of the nfg1^RIP mutant did not confirm this hypothesis because WC-1 was still acetylated in this mutant strain.

MS/MS analyses identified five phosphorylated Sin WC-1 (S990, S992, S994, S995, S998), which are phosphorylated in vivo and localize in a region immediately downstream of the zinc-finger domain [131] (Figure 3a-1). Biochemical analyses have demonstrated that these five phosphorylation sites are not involved in the light-induced hyperphosphorylation of WC-1. However, phosphorylation of these sites appears to be necessary for maintaining WC-1 steady state levels, as observed in strains in which these serine were mutated. Mutations at these sites affect normal circadian oscillation in N. crassa though they do not affect light-dependent responses. They might be involved in the positive limb of the circadian negative feedback loop [111]. The amount of WC-1 was found to determine the expression of frq in the dark in a wild-type strain; in the mutant with all five sites mutated, the frq mRNA level was slightly higher, suggesting that these phosphorylation events negatively regulate the transcriptional activation of WC-1 in the dark. S990 was the first of the five phosphorylated Serine mentioned above to be phosphorylated. Several experiments indicate protein kinase A (PKA) for this role (Figure 3a-1): both WC proteins are hypophosphorylated in the pka^C mutant strain, they physically interact with PKA and are phosphorylated by this kinase; PKA associates with WCC but not with FRQ, suggesting that phosphorylation of WCs by PKA is independent of FRQ; in vivo experiments demonstrated that PKC phosphorylates WC-1 in the region near the DNA binding domain and that this phosphorylation stabilizes the WC-1 protein; phosphorylation of WCs by PKA strongly inhibits WCC activity, resulting in the inhibition of frq transcription. Phosphorylation occurs upstream from CK-1a and CKII during WCC inhibition. Finally, PKA and its regulatory subunit PKAR are essential for normal circadian rhythm because WC-1 is less stable in a PKA mutant strain and the circadian transcription of clock-controlled genes is compromised. These data suggest that PKA can act as a priming kinase on WC-1. Bioinformatics analyses have shown that the sequence context around residues 990 (KSNSP) might be a direct target of PKA, whereas S995 (SHSSP) is not a typical PKA site (R-X-S/T or R-R/K-X-S) [132].

The FRQ-independent phosphorylation of S990 and S995 can convert the other sites into CKIa and CKII targets [107]. Phosphorylation of S990 significantly enhanced the ability of CK-1a to phosphorylate this region. This means that phosphorylation of S990 mediates the recruitment of the other kinases and that the subsequent phosphorylation events are mediated by a priming kinase. Recently, GSK entered in the group of the circadian clock regulators of N. crassa; GSK physically interacts with the WCC, phosphorylating both subunits with high specificity in a complex protein environment in vitro [133]. GSK also influences WC-1 protein stability, as observed in strains in which GSK (Glycogen Synthetase Kinase) activity was reduced, and it was coupled with an increased amount of WC-1. Since GSK affects the free-running period of Neurospora, the kinase must act on the dark form of the WCC.

In the nucleus, the WCC is present in excess over FRQ. Consequently, FRQ can neutralize only a very small fraction of WCC by binding stoichiometrically the complex and by regulating WCC phosphorylation [126]. An analysis carried out in different frq mutants (frq¹⁰, frq⁹) showed the typical loss of electrophoretic mobility of phosphorylated proteins for WC-1 and WC-2. Since the frq⁹ mutant produces a truncated FRQ protein lacking the C-terminal domain, this domain seems to be essential for WC-1 phosphorylation, as well as for shuttling of FRQ into the nucleus. FRQ-dependent phosphorylation leads to WCC inactivation [62]. This means that the phosphorylation level of the WCC during the circadian cycle correlates with the function of WCC as a transcription activator. Hypophosphorylated WCC efficiently binds the C-box of frq promoter, while hyperphosphorylated WCC binds the C-box with reduced affinity and does not efficiently activate the transcription of target genes.

How does FRQ influence WCC phosphorylation during daily circadian oscillation? FRQ physically interacts with CKIa and it is cyclically phosphorylated by this kinase (Figure 2c,d). It also forms a complex with WCC in the nucleus, suggesting that WCC could be target of CKIa. Immunoprecipitation experiments demonstrated the physical interaction between WC-1 and CKIa and this binding is FRQ-mediated. Disruption of the interaction between FRQ and CKIa also affects the binding between WC-1 and CKIa [98]. Functional mutation of CKIa (ck-1a^L strain) confirmed the involvement of this kinase in WCC phosphorylation.

FRQ/CKIa interaction is regulated by FRH [106] (Figure 2b,c); FRH with FRQ forms the FFC complex, which inhibits WCC activity [100]. This means that FRH is also important for WCC circadian phosphorylation. Moreover, WCC is also phosphorylated by CKII [98]. These results suggest that FRQ acts as a kinase substrate-recruiting subunit to mediate the phosphorylation of WCC by recruitment of CKIa and CKII. In turn, WCC phosphorylation affects its ability to be recruited at the C-box and LRE elements of the promoter sequences of regulated genes. WC-1 rhythmically binds the frq C-box, but this binding is enhanced in the ck-1a^L strain, confirming that a hypophosphorylated form of the WCC is required for its transcriptional activity [98].

The daily increase in WC-1 protein abundance seems to be correlated with the hyperphosphorylated form of FRQ. Accumulating data suggest that FRQ modulates the stability of total WC-1 protein, promoting its own accumulation and turnover [68]. Physical interaction between FRQ and WC-1 is essential for this process, and it has been suggested that FRQ-dependent clearance of WC-1 from the nucleus is a closure mechanism of the negative feedback loop mediated by oscillation in the FRQ: WCC ratio [134]. Translocated WC-1 can be reactivated in the cytoplasm in an FRQ-mediated manner [23,66]. Though it is not clear how FRQ can reactivate WC-1, it probably mediates the formation of the WCC.

Specific interactions between light-responsive regions (LRR) of the light-dependent genes and WCC have been demonstrated in vivo [27]. In vitro DNA binding assays using recombinant zinc-finger domains of WC-1, NIT-2, and CYS-4 did not preferentially bind to their specific GATA repeats on naked DNA segments consisting of target promoters [149]. This suggested the existence of other not yet described factors necessary in this interaction might be important in conferring specificity to the response. The correlation between chromatin remodeling and light responses clarified this point. The first evidence of this correlation was observed at histone H3 transient acetylation in the promoter region of al-3 [130]. An increase in acetylation was observed 20 min after multiple light pulses delivered to wild-type Neurospora strains in different light conditions, with a transient kinetic comparable to that of al-3 mRNA [150]. Lysine 14 (K14) has been shown to be involved in rhythmic acetylation after light pulse. Correlations between acetylation of acH3 K14 and transcriptional activation was confirmed in several other systems [151,152]. This was the first evidence that Neurospora light signal transduction involves chromatin modifications. Conversantly analysis of the acetylation state of core histones including AcH3 K9/K14 at the LRE region of the frq promoter did not show changes over the circadian cycle. Though this apparently contrasts with descriptions of the acetylation state of core histones at LRE regions, the discrepancy between these results may be due to differences in the loci examined and the regions therein[130,153].

A correlation between the transient acetylation of AcH3K14 and WCC activity was also demonstrated by the loss in light-dependent chromatin acetylation in a wc-1 null strain. Moreover, a Neurospora h3^k14q mutant, who carried an ectopic mutant copy of histone H3 (H3K14Q), engineered to substitute a lysine to glutamine at residue 14, showed the same light-dependent aberrant phenotype of the wc-1 null mutant [130]. Light-induced carotenogenesis was strongly impaired in the hH3K14Q strain as compared with the isogenic wild type and this was coupled with a loss of light-inducible al-3 mRNA oscillation. Neurospora Gcn Five-1 (NGF-1), a histone acetyl transferase homolog of yeast Gcn5p has also been found directly responsible for K14 H3 acetylation. This correlation was observed in the ngf-1^RIP1 mutant strain in which the absence of NGF-1 HAT activity was coupled with a loss of histone acetylation at K14 of histone H3. Finally, ngf-1^RIP1 and wc-1 null mutants exhibited identical phenotypes, suggesting that they are affected in the same signal transduction pathway. NGF-1 needs to physically interact with WC-1 in order to exert its own HAT activity [129]. Although this complex was preassembled in the dark, NGF-1 acetylated histones only after multiple light pulses (Figure 4). The more reasonable explanation is that, after multiple light pulses, a conformation shift in the region of the LOV domain of WC-1 changes the NGF-1 position on the nucleosome, thus activating its HAT function [154]. NGF-1 is unable to acetylate histones without correct binding with the C-term region of WC-1. It has been found that the LXXLL consensus sequence of WC-1 is involved in the interaction with NGF-1. This is a conserved consensus used for the binding between nuclear receptors and coactivators [155,156]. This means that WC-1 acts as a nuclear receptor. This was the first time that this mechanism was observed in fungi and the first demonstration of WCC-mediated chromatin remodeling in light-dependent events in N. crassa [157].

Multiple evidences suggest that dynamic histone accumulation and modification are under the control of the circadian cycle. Mouse Period 1 and Period 2 (mPer1 and mPer2) genes, which encode central clock components functionally analogous to FRQ, have rhythmic AcH3 (K9, K14) modifications, and RNA polymerase II (Pol II) binds rhythmically to these promoters [144,151]. Since the mammalian clock gene mPer2 has promoter methylation, chromatin remodeling and histone occupancy at the frq promoter region were analyzed. Transcriptional regulation at the frq promoter occurs through binding of the WC proteins to Clock box (C box) required for rhythmic oscillation in constant darkness. It occurs by binding with proximal light-regulated elements (PLREs) as well, necessary for frq light-dependent activation [24,51]. Occupancy of the frq promoter by WCC and DNase hypersensitivity of this region show circadian oscillation, contrasting with the common idea of a stable, preassembled and obligated WCC in all daily conditions as described above [153,158]. ChIP assays showed rhythmic association of WC-2 but not of WC-1 at the C-box, indicating a dynamic regulation that includes complex formation or disassembly. Moreover, at the PLRE region it was observed that WC-1 but not WC-2 is present in all circadian times. This means that accessibility to DNA by transcription factors is cyclically regulated during the day.

The ATP-dependent chromatin-remodeling enzyme CSW-1 is thought to be responsible for the closing catalytic activity [153]. Molecular and genotypic data showed that CSW-1 negatively regulates WCC activity by changing chromatin accessibility at the frq promoter. Indeed, strains lacking CSW-1 show a relatively open chromatin structure at the frq promoter. CSW-1 facilitates chromatin compaction and decrease of WC-2 levels at the frq promoter, thus promoting its repression. Translated FRQ cooperates to increase nucleosome occupancy and inhibit WCC binding at the C-box. Also, other enzymes that modify DNA methylation cooperate in chromatin compaction. The DNA methyltransferase (DNMT) DIM-2/HP1 (defective in methylation; heterochromatin protein-1) and DIM-5 are required for DNA methylation at frq promoter in circadian rhythmicity [159]. Deletion of dim-2 results in a small phase defect (approximately 2 h) and changes in DNA methylation at frq. This indicates a link between DNA methylation and phasing of clock gene expression. Interestingly, DNA methylation at the frq promoter decreased with time in constant darkness, although not in an overtly circadian manner. Moreover, promoter methylation appeared greatly reduced in cultures grown in constant light for extended periods (48 h in light), providing evidence for another strong correlation between epigenetic modification and light signal transduction. Dim-5-dependent H3K9me3 is required for frq promoter methylation. Loss of DIM-5 can also suppress the hypermethylation phenotype of chd1 and the mutant show a shorter phase advantage compared with dim-2 mutant period [159]. DIM-5 DNA activity is also established in response to the light and loss of H3K9me3 results in elevated WCC binding to the frq promoter and an increase in light activated frq expression. All together these data suggest that H3K9me3 appears to be involved light-mediated expression and is likely more important in the diurnal expression confirming that DNA methylation is required in the circadian entrainment. Antisense qfr RNA is also essential for normal DNA methylation in this region. This DNA methylation at frq is surprising. Indeed, in various Neurospora strains analyzed in many different conditions in which about 2% of cytosines are methylated, methylation exists exclusively as a relic of repeat-induced point (RIP) mutation and rDNA [160,161]. Since all these factors are indispensable for chromatin compacting, others are involved in chromatin opening for the next round of activation. Clock ATPase (CATP) counteracts CSW-1 activity, opening the chromatin structure and positively regulating nucleosome occupancy at the frq locus [162]. It has been demonstrated that CATP promotes the removal of nucleosomes at the frq locus and enhances WCC binding. CATP activity, but not CATP expression, shows circadian oscillation, counteracting the increase in nucleosome occupancy at the C-box. CATP specifically associates with the frq locus, suggesting that CATP regulates nucleosome occupancy directly on chromatin. The catp^Ko strain shows a lower expression of FRQ, higher occupancy at the frq C-box, and minor WCC occupancy during the day as compared with the wild-typestrain. Though it is not clear how CATP works, both the ATPase domain and the bromodomain of CATP are essential for its function in the clock.

Chromo-domain helicase DNA-binding protein 1 (CHD1), another chromatin remodeler, has been found to be involved in frq chromatin remodeling through which it promotes the open state of chromatin [158]. Loss of CHD1 generates hypermethylation on the frq locus. It was demonstrated that CHD1 is required for maintaining regular high amplitude rhythms of WC-2 binding, which contributes to the apparent changes in chromatin structure. DNA methylation has been also found in the promoter of wc-1 gene in the dchd1 mutant strain but absent in the wild-type strain. These data suggest that CHD1 acts as a demethyltransferase that counteracts DIM2 and qfr activity, though the molecular mechanism is not clear. Finally, these data highlight that DNA and histone methylation modulates light responses and circadian rhythms. Figure 4 illustrates the chromatin rearrangements on light-activated genes. Table 3 presents the chromatin remodelers and their activity.

Other factors cooperate with CSW-1 for chromatin compaction. The KMT2 enzyme, called SET1 (FGSC15827), has been isolated as a candidate for circadian and light-dependent histone methylation [163]. SET1 is part of a large holoenzymatic complex (complex associated with SET1 [COMPASS]) that contains several additional subunits involved in H3K4 methylation [164]. Strains lacking set1 gene or a subset of other SET1/COMPASS subunits have an apparent arrhythmic clock phenotype on race tubes. Moreover, in analysis at transcriptional and translational levels confirmed the defect of frq in the set1 mutant strain indicating that H3K4 methylation is required for efficient feedback inhibition of the central clock gene.

ChIP data indicate that H3K4me3 is present at the frq locus throughout the entire circadian cycle, with peaks occurring during the transcriptional repressive phase (CT9 to CT13) corresponding to the repressive phase of frq transcription. H3K4me3 is predominantly present in the frq-coding region, which is consistent with reports that H3K4 methylation is normally localized to coding regions and the transcriptional start site of genes [165]. This methylation is strongly regulated by SET1. Indeed, there is a delay in the temporal H3K4 methylation of the frq promoter in Δset1 mutant strains compared with wild-type strains. This delay increases the amplitude of frq expression under circadian and light-dependent conditions. This further suggests that H3K4me3 may play a role in the down-regulation of frq. Moreover, lack of functional SET1 causes a temporal alteration of rhythms in conidia formation.A defect in WC-2 binding to the C-box element in Δset1 mutant has been observed as consequence of the altered histone methylation state.

In wild-type Neurospora strains, no changes in H3K4 methylation were observed after exposure to light for 15 min at the PLRE. After 30 min, a time corresponding to the start of VIVID (VVD)-dependent light repression, the H3K4me3 levels rose and remained elevated, followed by a slow and steady decrease over the course of the next 2 h. This suggests that SET-1 can play a role also in light-dependent processes. These data confirm that histone methylation is necessary for modulating circadian and light responses by chromatin compaction. Figure 4 summarizes the factors involved in Chromatin methylation.