The small G protein RAS2 is involved in the metabolic compensation of the circadian clock in the circadian model Neurospora crassa

Accumulating evidence from both experimental and clinical investigations indicates a tight interaction between metabolism and circadian timekeeping; however, knowledge of the underlying mechanism is still incomplete. Metabolic compensation allows circadian oscillators to run with a constant speed at different substrate levels and, therefore, is a substantial criterion of a robust rhythm in a changing environment. Because previous data have suggested a central role of RAS2-mediated signaling in the adaptation of yeast to different nutritional environments, we examined the involvement of RAS2 in the metabolic regulation of the clock in the circadian model organism Neurospora crassa. We show that, in a ras2-deficient strain, the period is longer than in the control. Moreover, unlike in the WT, in Δras2, operation of the circadian clock was affected by glucose; compared with starvation conditions, the period was longer and the oscillation of expression of the frequency (frq) gene was dampened. In constant darkness, the delayed phosphorylation of the FRQ protein and the long-lasting accumulation of FRQ in the nucleus were in accordance with the longer period and the less robust rhythm in the mutant. Although glucose did not affect the subcellular distribution of FRQ in the WT, highly elevated FRQ levels were detected in the nucleus in Δras2. RAS2 interacted with the RAS-binding domain of the adenylate cyclase in vitro, and the cAMP analogue 8-bromo-cyclic AMP partially rescued the circadian phenotype in vivo. We therefore propose that RAS2 acts via a cAMP-dependent pathway and exerts significant metabolic control on the Neurospora circadian clock.

Accumulating evidence from both experimental and clinical investigations indicates a tight interaction between metabolism and circadian timekeeping; however, knowledge of the underlying mechanism is still incomplete. Metabolic compensation allows circadian oscillators to run with a constant speed at different substrate levels and, therefore, is a substantial criterion of a robust rhythm in a changing environment. Because previous data have suggested a central role of RAS2-mediated signaling in the adaptation of yeast to different nutritional environments, we examined the involvement of RAS2 in the metabolic regulation of the clock in the circadian model organism Neurospora crassa. We show that, in a ras2-deficient strain, the period is longer than in the control. Moreover, unlike in the WT, in ⌬ras2, operation of the circadian clock was affected by glucose; compared with starvation conditions, the period was longer and the oscillation of expression of the frequency (frq) gene was dampened. In constant darkness, the delayed phosphorylation of the FRQ protein and the long-lasting accumulation of FRQ in the nucleus were in accordance with the longer period and the less robust rhythm in the mutant. Although glucose did not affect the subcellular distribution of FRQ in the WT, highly elevated FRQ levels were detected in the nucleus in ⌬ras2. RAS2 interacted with the RAS-binding domain of the adenylate cyclase in vitro, and the cAMP analogue 8-bromo-cyclic AMP partially rescued the circadian phenotype in vivo. We therefore propose that RAS2 acts via a cAMP-dependent pathway and exerts significant metabolic control on the Neurospora circadian clock.
Circadian rhythms are endogenously generated at the cellular level. Core clock mechanisms control rhythmic expression of a large set of genes, which, in turn, regulate various biological processes, including cell growth, proliferation, and metabolism. An important feature of the circadian clock is the ability to display an endogenous rhythm with a constant period length under different environmental conditions. Among these adap-tation mechanisms, the most intensively investigated process is temperature compensation. Although a tight interaction between metabolism and the circadian clock has been shown at almost all levels of organisms (1)(2)(3), it is still poorly understood how molecular timekeeping is compensated against changes in nutrient availability.
Neurospora crassa belongs to the most extensively examined model systems in the field of circadian research and has served as a useful tool for the investigation of different aspects of circadian regulation, including metabolic compensation of the circadian clock (4). Monitoring the conidiation rhythm, the time-dependent formation of asexual spores, allows easy study of the effect of genetic changes or pharmacologic manipulations on the output of the circadian oscillator (5). The molecular clockwork of Neurospora is well characterized (recently reviewed in Refs. 6,7). The transcription factors White Collar 1 (WC-1) 3 and WC-2 and the negative factor Frequency (FRQ) represent the core clock components. Other important regulators of the molecular oscillator include kinases, phosphatases, exosome components, and factors controlling the chromatin status (e.g. Refs. 8 -13). WC-1 and WC-2 form the WC complex (WCC), which supports expression of frq. The FRQ protein interacts with FRQ-interacting RNA helicase and casein kinase 1a (9,14) and inhibits the activity of the WCC by facilitating its phosphorylation. FRQ undergoes several steps of phosphorylation that affect distinct functions of the protein (15)(16)(17). Shortly after its synthesis, low levels of hypophosphorylated FRQ forms are accumulated in the nucleus and are efficient in negative feedback. In later phases of the circadian cycle, hyperphosphorylation of FRQ interferes with nuclear import of the protein, and the cytoplasmic FRQ supports accumulation of the WCC, thereby forming the positive feedback loop. Hyperphosphorylated FRQ interacts with F-box/WD-40 repeat-containing protein (FWD-1), which facilitates ubiquitination and, thus, degradation of the protein (18). When FRQ is cleared from the nucleus, the WCC is gradually released from inhibition, and a new cycle starts. The WCC is also the main photoreceptor of Neurospora, controlling transcription of light-inducible genes and transducing light inputs to the circadian oscillator (19 -22). In light/dark cycles, the light-inducible blue light receptor VIVID inhibits the light-activated WCC by forming an additional negative feedback loop (23)(24)(25).
Regarding the interconnections between metabolism and circadian timekeeping, the period of the Neurospora clock was shown to be independent of the glucose concentration of the medium, and conidial separation 1 (SCP-1), an important regulator of energy metabolism, was characterized as a key component of the adaptation of clock function to different sugar availabilities (4). According to the model of metabolic compensation proposed by Sancar et al. (4), glucose generally increases the rate of protein synthesis and, thus, translation of WC-1, the limiting subunit of the WCC. Glucose-dependent suppression of wc-1 transcription by CSP1, however, compensates for the translation elevation, resulting in glucose-independent expression of the WCC and, thus, stabilization of the period. Recently, PERIOD 1 (PRD-1), an RNA helicase, was also found to control cycle length in the presence of glucose (26,27).
In yeast, the RAS-mediated pathway is central in the signal transduction of glucose sensing. This pathway controls the transcription of at least one hundred genes and thus contributes to the adaptation of cellular functions to changes in the metabolic environment (28,29). Because yeast cells and N. crassa express similar RAS proteins (30), and the Neurospora clock is thoroughly characterized at the molecular level, Neurospora may represent a good tool for the investigation of the interactions between RAS signaling and metabolic regulation of the circadian clock.
RAS small G proteins are universal eukaryotic signal transducers that belong to the superfamily of small monomeric GTPases. In most organisms, several RAS isoforms are expressed. Although they share common sets of upstream regulators and downstream effectors, various experimental data support the functional specificity of the different RAS isoforms (31). RAS proteins play a critical role in the regulation of metabolism, cell growth, proliferation, and oncogenic transformation (32,33), all representing processes that are interlinked with the circadian rhythm. Indeed, an increasing body of evidence suggest interconnections between RAS-mediated pathways and the circadian clock in different organisms. Expression of RAS family members was found to be regulated by the circadian clock in some mammalian tissues (34 -36), and functional analyses also revealed that the RAS/MAPK pathway may serve as output and/or is able to modulate outputs of the circadian clock (37)(38)(39)(40)(41)(42)(43)(44)(45). In Neurospora, activating mutation of RAS-1 (present in the band (bd) strain) enhances the conidiation rhythm and light-induced transcription of the positive factor WC-1 (46). Moreover, we recently showed that the phase of conidiation is more sensitive to temperature changes in the bd strain than in the WT (47). Although, in most cases, the molecular basis is not well understood, RAS GTPases seem to be involved in clock control mechanisms in higher organisms as well. In Drosophila, RAS regulates the activity of the positive clock factor CLOCK/ CYCLE (48). In mice, the dexamethasone-induced RAS protein 1 (DEXRAS1) is suggested to be an input factor of the circadian clock (49 -52). Both experimental data and bioinformatic analysis suggest that oncogenic RAS proteins are involved in deregulation of the circadian clock in cancer cells (53). Finally, it was recently shown that RAS signaling affects the circadian clock in the suprachiasmatic nucleus of mice, acting, at least partially, via ERK and glycogen synthase kinase, thereby fine-tuning the circadian period (54).
Our goal in this study was to investigate whether RAS2 is involved in metabolic compensation of the circadian clock of N. crassa. Our results indicate that stability of the molecular clock function against changes in glucose availability is dependent on RAS2. We show that both phosphorylation and subcellular localization of FRQ are dependent on RAS2. We propose that RAS2, acting via a cAMP-dependent pathway, links the molecular oscillator to the glucose-sensing pathway and therefore is an important component of the nutritional compensation of the circadian period.

The ras2 mutation affects conidiation rhythm
To study the possible interplay between the RAS2-mediated signal transduction pathway and the circadian clock, we aimed to examine the circadian behavior of the ras2 deletion strain. ⌬ras2 (FGSC 12467) was generated during the Neurospora Genome Project (55) and had a phenotype similar to that of the smco7 mutant described earlier (56). On agar slants, ⌬ras2 formed less aerial hyphae than the WT and, accordingly, produced relatively few conidia, located primarily in a crescent form at the upper top of the medium (Fig. 1A). When inoculated in liquid medium, ⌬ras2 grew slowly compared with the WT and produced mycelia dominantly in colonial form (small balls) instead of mycelial mats (supplemental Fig. S1A).
To confirm that deletion of ras2 is responsible for the morphological defects of the mutant, we generated a strain expressing a tagged form of RAS2 under the control of the cpc-1 promoter in the ⌬ras2 background. Comparable ras2 transcript levels were detected in ⌬ras2, cpc-1-ras2 and the WT, and an anti-FLAG antibody recognized a protein band with the expected molecular mass (29.5 kDa) in the total cell lysates of ⌬ras2, cpc-1-ras2 but detected no signal in the lysates prepared from either ⌬ras2 or WT cells (Fig. 1B). Expression of RAS2 FLAG rescued the morphological defects observed in the mutant, indicating that the fusion protein is functionally active and that the altered morphology was a consequence of ras2 deficiency ( Fig. 1A and supplemental Fig. S1A).
To analyze the conidiation rhythm of ⌬ras2, race tube assays were performed under constant conditions (Fig. 1C). Similar to the smco7 mutant (56), ⌬ras2 had a reduced growth rate compared with the WT. In accordance with literature data (7,46), WT Neurospora did not display a conidiation rhythm on minimal medium but showed sustained banding when the reactive oxygen species generator menadione was present. In ⌬ras2, however, no conidiation rhythm was detected, even when high concentrations (100 M) of menadione were applied, suggesting that deletion of ras2 affects the rhythmic output.
Because entrained conditions generally support conidiation rhythm (47), we incubated race tube cultures in 12/12-h light/ dark cycles. Under these conditions, ⌬ras2 also displayed sustained banding for several days (supplemental Fig. S1B). However, when the phase of conidiation was thoroughly analyzed and compared with that of the WT, a significant delay was

Metabolic compensation of the clock by RAS2
observed in ⌬ras2, suggesting that RAS2 activity also affects clock function under entrained conditions (Fig. 1D). In ⌬ras2, cpc-1-ras2, the phase of banding was similar to the phase detected in the WT, indicating that expression of RAS2 FLAG in the mutant background results in rescue of the clock function.

Involvement of RAS2 in glucose compensation of the circadian clock
To further analyze the effect of ras2 deficiency on clock function, we generated a ⌬ras2 strain that expressed luciferase under the control of the frq promoter. In the first experiments, we used a medium containing no glucose and followed the rhythm of frq promoter activity in constant darkness ( Fig. 2A, top panel). Although a robust rhythm was detected in both the ⌬ras2 and the WT background, the circadian period was significantly longer in the mutant than in the control strain, suggesting that RAS2-mediated signaling interacts with the molecular clock (Fig. 2B). As expected from earlier data (4), addition of glucose to the medium did not affect either the robustness or period of the rhythm in the WT (Fig. 2

, A, bottom panel, and B).
In the ras2-deficient strain, however, a dampening in the amplitude was observed, and the period was more than 2 h longer compared with the WT. These data indicate that compensation of the oscillator function against glucose requires the action of RAS2.
Next, we examined whether expression of ras2 is controlled by the circadian clock in the WT. As shown in Fig. 3, although frq levels displayed rhythmic changes in our samples, ras2 mRNA did not oscillate, indicating that ras2 is not a clockcontrolled gene in Neurospora.

Expression and phosphorylation of the molecular clock components is affected by the ras2 mutation
In the next experiments, we compared the expression and phosphorylation of the main clock components in the WT and ⌬ras2. The electrophoretic mobility of the core clock proteins FRQ, WC-1, and WC-2 is dependent on their phosphorylation status, i.e. phosphorylation is reflected by the presence of protein forms with slower electrophoretic mobility (57,58). On the other hand, the activity of the clock components correlates with their phosphorylation status, i.e. hypophosphorylated nuclear FRQ is active in negative feedback by supporting phosphorylation and, thus, inactivation of the WCC (16).
As the results of the in vivo luciferase measurements suggested that the impact of the ras2 mutation on the clock function depends on glucose, Neurospora was incubated in glu- . Right panel, the indicated strains were grown in liquid cultures in LL. Whole cell extracts were analyzed by Western blotting using an anti-FLAG antibody. Asterisk, unspecific band. a.u., arbitrary unit. C, menadione (men) does not induce conidiation rhythm in ⌬ras2. Race tubes were inoculated with WT and ⌬ras2 and contained menadione as indicated. Following incubation in constant light for 2 days, race tubes were transferred to constant darkness. Representative race tubes are shown. The first black line on each race tube marks the growth front at the time point of the LD transition. D, the phase of banding is delayed in ⌬ras2 under entrained conditions. Race tubes inoculated with conidia of ⌬ras2, WT or ⌬ras2, cpc-1-ras2 were incubated in 12/12-h light/dark cycles. For better comparison of the position of conidial bands, images were fitted so that the daily growth distances were similar in the parallel samples. Representative race tubes are shown. Average phases of the peak of conidiation were calculated and statistically analyzed (n ϭ 7 (WT), 18 (⌬ras2), 14 (⌬ras2, cpc-1-ras2); S.E.; ***, p Ͻ 0.005, two-sample t test).

Metabolic compensation of the clock by RAS2
cose-containing standard medium. In constant light (LL), only a slight difference in FRQ protein level was detected, and this was represented by a higher accumulation of hypophosphorylated FRQ forms in ⌬ras2 (Fig. 4A, left panel). WC-1 levels were higher and the hyperphosphorylated fraction of the protein dominated in ⌬ras2 compared with the WT. Although WC-2 expression levels were similar in both strains, a slightly higher fraction of the hyperphosphorylated forms was detected in ⌬ras2. To further examine the effect of the ras2 mutation on the molecular clock, we performed subcellular fractionation and analyzed the expression of clock proteins in the cytosol and nucleus (Fig. 4A, right panel). A higher fraction of the hypophosphorylated FRQ forms was detected in the nucleus of ⌬ras2 compared with the WT. In accordance with the higher activity of hypophosphorylated FRQ in promoting phosphorylation of the WCC, hyperphosphorylated forms of both WC proteins dominated in ⌬ras2 nuclei. Moreover, a substantial fraction of WC-1 was detected in the cytosol of the mutant, indicating that the excess of WC-1 in ⌬ras2 is mainly present in an inactive cytosolic form.
To further investigate the molecular oscillator, we analyzed the expression of FRQ in cultures grown in constant darkness (DD). Although both strains displayed time-dependent changes in FRQ levels on the first day ( Fig. 4B and supplemental Fig. S2A), FRQ signals peaked later in ⌬ras2 than in the WT. This difference was also well represented by the reduced frq RNA expression and the very low level of newly synthesized hypophosphorylated FRQ forms in ⌬ras2 at DD12 (Fig. 4C). During the second circadian cycle, both the levels and the phosphorylation of FRQ displayed oscillation in the WT, whereas less pronounced changes in the phosphorylation and expression of the protein were detected in ⌬ras2 ( Fig. 4B and supplemental Fig.  S2A). In the WT, newly synthesized FRQ becomes progressively phosphorylated from DD16; therefore, we investigated both expression levels and phosphorylation of FRQ between DD16 and DD26 with a more detailed resolution on the same gel (Fig. 4D). Overall FRQ levels were lower in ⌬ras2 than in the WT. To assess FRQ levels in ⌬ras2 relative to the WT, we constructed a calibration line by loading the gel with increasing quantities (25-125%) of the WT protein extracts and determined relative FRQ levels at DD16, when newly synthesized FRQ is already well detectable in both strains (supplemental Fig. S2B and Fig. 4E, left panel). We found significantly lower FRQ expression in ⌬ras2 lysates compared with WT samples. In addition, although, in the WT, FRQ was gradually and continuously shifted toward the hyperphosphorylated forms, in ⌬ras2, this shift was slow, and the highly phosphorylated forms of the protein were absent, even at DD26 (Fig. 4D). When the ratio of hypo-and hyperphosphorylated FRQ forms was determined, the results indicated that distribution of these FRQ forms displayed a marked difference over time in the two strains (significant time ϫ strain interaction of repeated measures ANOVA) (Fig. 4E, right panel). Thus, the absence of RAS2 also affects phosphorylation of FRQ under constant conditions.

Metabolic compensation of the clock by RAS2
According to data in the literature, the rhythm of WC-1 expression is, in most cases, not as robust as that of FRQ (59,60). Although WC-1 showed time-dependent changes in the WT in our hands, with the first trough and peak phases similar to those reported by others (59,61), no rhythm of WC-1 expression could be detected in ⌬ras2 (supplemental Fig. S2, C and D). This finding also suggests that the molecular clock function is less robust in the mutant than in the WT. We quantified and compared WC-1-specific signals in the WT and ⌬ras2 at DD24 and found no significant difference (supplemental Fig. S2E).
Because phosphorylation of FRQ may influence nuclear accumulation of the protein, we analyzed cytosolic and nuclear FRQ fractions during a circadian cycle in the WT and the ras2deficient strain (Fig. 5). Following the LD transition, in the first 8 h, a similar decrease of FRQ levels was observed in both strains in both cellular compartments. In the WT, FRQ levels peaked at DD16 and then became rapidly reduced, with kinetics similar to those described earlier (62). In accordance with the delayed synthesis of FRQ, cytosolic protein levels increased more slowly in ⌬ras2 than in the WT. The increase in the nuclear amount of the protein was, however, similar to that in the WT, and instead of a rapid reduction in nuclear FRQ levels seen in the WT, a substantial fraction of FRQ was detected in the nucleus, even at DD24. To analyze this difference, we determined nuclear/cytosolic FRQ-specific signal ratios. At both DD16 and DD20, significantly higher ratios were obtained for ⌬ras2 than for the WT (Fig. 5). In addition, repeated measures ANOVA with two factors (time and strain) showed a significant strain effect for the whole period analyzed (supplemental Fig.  S3). In summary, the above data suggest that RAS2 is involved in the control of subcellular localization of FRQ under constant conditions.

Response of the molecular clock to glucose is dependent on RAS2
Based on our results showing that the circadian phenotype of ⌬ras2 is dependent on glucose, we aimed to examine how FRQ expression and subcellular distribution respond to glucose in ⌬ras2 and the WT. We cultured mycelia in starvation medium overnight in LL and then treated them with different levels of glucose for 4 h (Fig. 6). When a relatively low concentration (0.5%) of glucose was applied, high levels of FRQ accumulated in the nuclei of ⌬ras2, whereas nuclear FRQ levels remained low in the WT. Addition of glucose at a higher concentration (4%) effectively increased the accumulation of hypophosphorylated and, thus, total levels of FRQ in the WT as well. However, when the protein signals were quantified, a significantly higher  Fig. S2B) was constructed by loading the gel with different quantities (25-125%) of WT protein extract, and FRQ levels in ⌬ras2 relative to the WT were calculated. 100% corresponds to mean FRQ levels in the WT samples (n ϭ 4; S.E.; ***, p Ͻ 0.005; paired t test). Right panel, hyper-and hypophosphorylated FRQ forms (see D) were quantified by densitometry, and the ratio of hyper-and hypophosphorylated protein forms in the WT and ⌬ras2 is shown at the indicated time points (n ϭ 3; S.E.; ***, p Ͻ 0.005, significant time ϫ strain interaction (ANOVA with two factors: time, repeated; strain, non-repeated); Tukey honest significant difference test).

Metabolic compensation of the clock by RAS2
fraction of FRQ was detected in the nuclei of ⌬ras2 than in the nuclei of the WT. To exclude the possibility that osmotic changes contributed to the observed effect of high glucose levels in ⌬ras2, we repeated the experiment by adding NaCl instead of glucose at the same osmotic concentration. NaCl did not affect the nuclear levels of FRQ in ⌬ras2 (supplemental Fig.  S4A). Moreover, FRQ was also shifted toward the nucleus in ⌬ras2 when cultures were incubated overnight in a glucosecontaining medium (supplemental Fig. S4B). Our observations indicate that RAS2 plays an important role in the control of nuclear levels of the oscillator protein in response to glucose.

Analysis of the signaling pathway between RAS2 and the circadian clock
RAS2 was found to control both a MAP kinase cascade and the adenylate cyclase-dependent pathway in Saccharomyces cerevisiae (63)(64)(65). The MAPK component of the cascade is represented by Kss1. Therefore, we examined the levels of the active forms of MAK2, the Neurospora ortholog of Kss1 (66,67). Based on the sequence homology (67), we used a phosphospecific commercial antibody that recognizes the phosphorylated and thus active form of the mammalian MAPK (ERK). According to data in the literature (67,68), a protein band with the approximate molecular mass of 43 kDa could be detected in both strains (Fig. S5). Although phospho-MAK2 levels showed relative large variations when independent cultures were compared, we found no consistent strain-specific differences.  WT and ⌬ras2 mycelia grown in standard liquid medium were transferred into a medium containing no glucose. Following incubation for 16 h, cultures were treated with either 0.5% or 4% glucose for 4 h, as indicated. The experiment was performed in LL. FRQ expression in whole-cell extracts (T) or cytosolic (C) and nuclear (N) fractions was analyzed by Western blotting. Left panel, representative Western blots are shown. RGB-1 was detected as a cytosolic marker. Right panel, nuclear/total signal ratios were calculated for samples incubated without glucose or with 4% glucose, as indicated. The values of ⌬ras2 samples were normalized to the control ratio determined for the corresponding WT sample on the same Western blot (n ϭ 4; S.E.; *, p Ͻ 0.05; one-sample t test; n.s., not significant).

Metabolic compensation of the clock by RAS2
Based on the data showing that RAS2 activates adenylate cyclase in yeast (28), in the next experiments we addressed whether Neurospora RAS2 similarly interacts with adenylate cyclase. We generated an Escherichia coli strain expressing a fusion protein containing the RAS-binding domain of adenylate cyclase and an N-terminal GST tag. The fusion protein was coupled to glutathione-agarose, and the resin was incubated with the lysate of ⌬ras2, cpc-1-ras2 containing RAS2 FLAG . RAS2 FLAG was specifically bound to the resin, suggesting that RAS2 may be an interaction partner of adenylate cyclase (supplemental Fig. S6A). To further examine the possible involvement of the RAS/adenylate cyclase-mediated pathway in the glucose response of the clock, we tested the effect of the cellpermeable and relatively stable cAMP analogue 8-Br-cAMP in the presence of glucose. 8-Br-cAMP had no effect on the rhythm in WT frq-luc (Fig. 7A), whereas it slightly but significantly shortened the period in ⌬ras2 frq-luc (Fig. 7B). Thus, the presence of the cAMP analogue partially rescued the impaired function of the circadian clock of ⌬ras2. Moreover, addition of 8-Br-cAMP reduced the glucose-induced nuclear accumulation of FRQ in the mutant (supplemental Fig. S6B), also suggesting that a cAMP-mediated pathway may link RAS2 to the molecular clock.

Discussion
From fungi to mammals, RAS family GTPases play important roles in signaling systems controlling cellular metabolism, cell growth, and differentiation (32,33). In mammals, the importance of RAS proteins as regulators of cell growth is emphasized by the prevalence of up-regulation of RAS-mediated signaling in many tumor types (69). Expression of the mammalian RAS protein can rescue the loss of endogenous RAS in yeast, suggesting functional conservation of RAS proteins among eukaryotic organisms (70). We found that, in Neurospora, the circadian period is significantly longer in the ⌬ras2 mutant than in the WT, suggesting that RAS2 activity is crucial for normal function of the circadian clock. Moreover, in contrast to the WT, both the period and robustness of the rhythm were sensitive to glucose in ⌬ras2; the period was significantly increased, and the oscillation became dampened in glucosecontaining medium. These findings indicate that activation of RAS2-mediated signaling is part of the compensatory mechanisms needed for robust clock function under different metabolic conditions. Both the expression and phosphorylation of molecular clock components are characteristically affected by deletion of ras2. Increased abundance in ⌬ras2 of the hypophosphorylated forms of FRQ, which are therefore more active in

Metabolic compensation of the clock by RAS2
negative feedback, is in accordance with enhanced accumulation of the hyperphosphorylated and thus inactive forms of the WCC. In a recent work, the speed of the phosphorylation cascade affecting FRQ activity in negative feedback was proposed as the key factor determining the circadian period (15). Our data showing delayed phosphorylation of FRQ in DD in ⌬ras2 together with a longer period are in good accordance with this model. Based on the data presented in this study, we propose that RAS2 may fine-tune the circadian clock by controlling a cAMP-dependent pathway. In our model, elevation of glucose levels could lead to an increased translation rate and, thus, accumulation of newly synthesized hypophosphorylated FRQ forms and enhanced activity of the RAS2-dependent pathway. As a distal component of this pathway, PKA could directly or indirectly support phosphorylation of FRQ and thereby counteract the nuclear accumulation of FRQ forms active in negative feedback. Conversely, in the ras2 deletion strain in the presence of glucose, the tardy clearance of FRQ from the nucleus may result in prolonged inhibition of the WCC and, thus, an increase in the period and a dampening of the oscillation.
FRQ can be phosphorylated by PKA in vitro (71), suggesting that FRQ may be a direct substrate of the kinase. Moreover, the pkac-1 ko strain lacking the major catalytic subunit of PKA partially resembles the phenotype of the ⌬ras2 strain. Reduced growth rate and dampened rhythm of FRQ oscillation, although to different extents, were detected in both strains. However, expression of the WCC was differentially affected by the lack of PKAC-1 and RAS2. Although expression of the WCs was low in pkac-1 ko , WC-1 levels in ⌬ras2 were elevated in LL and similar to the WT levels in DD, suggesting that RAS2-independent PKA activity is sufficient to support expression of the WC proteins in the mutant. Expression of WC-1 is controlled by both light-dependent and light-independent mechanisms and is posttranslationally supported by FRQ (16,19,61,72,73). How the absence of RAS2 influences this complex regulation is still unclear.
Beside cAMP-mediated signaling, a MAPK module homologous to the mammalian Raf/Mek/Erk pathway can also be activated by the RAS system in different fungi (66,74). However, we did not find a clear difference in MAK2 activity between ⌬ras2 and the WT in our samples. As MAP kinases other than Erk (JNK and p38) were also described as downstream effectors of RAS GTPases in higher eukaryotes, and their homologous are also present in Neurospora (5, 75), a role of these pathways in the metabolic regulation of the clock of Neurospora cannot be excluded. Moreover, as MAPKs are usually sensitive to the metabolic state of the cell (e.g. Ref. 76), their contribution to the control of the circadian clock may depend on culture conditions. Although data from the literature already implied that RASmediated pathways affect circadian timekeeping in different organisms, our study indicates for the first time that RAS GTPases may constitute a link between metabolism and the molecular clock. Our data suggest that, in Neurospora, RAS2 is required to maintain a robust rhythm with a constant circadian period independent of changes in glucose availability in the environment. We propose that, in addition to CSP1 and PRD-1 (4, 26, 27), RAS2 is a mediator of metabolic compensation of the Neurospora clock. Although loss of CSP1 shortens the period in response to glucose, prd-1 and ras2 mutants show period lengthening under these conditions. This suggests that the operation of counteracting systems may be required for stabilization of the circadian period in a changing metabolic environment. The molecular mechanisms underlying this metabolic control are only partially understood, and a plausible model has only been provided for the action of CSP1. Although CSP1 controls WC-1 levels, the RAS2-mediated pathway seems to primarily influence the action of FRQ in a glucose-dependent manner.

Plasmid construction
cpc-1-FLAG-Strep-ras2 was constructed in two steps. By using Neurospora genomic DNA as a template, a PCR product including the coding region of ras2 and nucleotides coding for an N-terminal Strep tag was amplified and inserted between the XbaI and HindIII sites of a p3XFLAG-CMV TM -7.1 vector (Sigma-Aldrich). This construct served as a template for the PCR product containing the coding region of ras2 and nucleotides coding for two N-terminal FLAG tags and a Strep tag. This product was cloned as an AscI-SpeI fragment into cpc-1-vvd-Strep (25), resulting in replacement of the vvd-Strep region with the FLAG(2x)-Strep-ras2 fragment.
Liquid cultures were incubated at 90 rpm and 30°C unless indicated otherwise. Vogel's media were supplemented with 0.5% L-arginine, 10 ng/ml biotin, and the indicated amount of glucose. The standard liquid medium contained 2% glucose. Race tubes were prepared as described previously (72). When indicated, the race tube medium was supplemented with menadione. Race tubes were incubated at 30°C. Analysis of race tubes was performed by densitometry using the ChronOSX 1.0.7 ␤ software (T. Roenneberg, Ludwig-Maximilians-Universität München) and according to a protocol described previously (47). The first day of the entrainment was excluded from the analysis of race tubes. In each series of experiments, data derived from the same days were analyzed for all race tubes. 8-Br-cAMP was dissolved in NH 4 OH, and the pH level of the medium was restored by addition of HCl.

Protein analysis
Extraction of Neurospora protein and Western blots were performed as described previously (13,47). For subcellular frac-

Metabolic compensation of the clock by RAS2
tionation, the method described by Luo et al. (62) was used with slight modification. Briefly, buffer volumes were reduced to 1/10, and the centrifugation step to separate the cytosol and nuclei was modified to 8800 ϫ g for 2 min at 4°C. 450 and 300 g of protein from the cytosolic and nuclear fractions were analyzed by SDS-PAGE, respectively. Detection of total protein (TP) levels by Ponceau staining was used as a loading control. Detection of RGB-1 as a cytosolic marker was used to control subcellular fractionation.
For the analysis of protein levels, the optical density of the protein bands was quantified using ImageJ software. Density values were corrected for loading differences using a well-defined region of the total protein stain as described previously (80). When all replicates could not be loaded onto the same gel, values obtained in ⌬ras2 samples were normalized to the control values determined in the corresponding WT or untreated samples on the same Western blot.

RNA analysis
Total RNA was extracted with TriPure isolation reagent (Roche), and transcript levels were quantified as described earlier (47) and in the supplemental Experimental Procedures.

In vivo luciferase assay
96-well white plates with solid medium containing Vogel's salt solution, 0.1% L-sorbose, 1% agarose, 10 ng/ml biotin, and 150 M beetle luciferin (Promega) were inoculated with the luciferase-expressing strains. When indicated, 0.05% glucose and 1 mM 8-bromo-cAMP were added to the medium. Cultures were incubated for 1 day in constant darkness and subsequently for 1 day in constant light at 30°C. Following a light-dark transfer, luminescence was detected for 4 -5 days, and data were analyzed as described previously (47).

Statistical analysis
For statistical analysis, Statistica 7.0 (Statsoft Inc., Tulsa, OK) software was used. The data in all figures are presented as mean Ϯ S.E. Results were considered to be statistically significant when p Ͻ 0.05.
Author contributions-N. G. and K. K. designed the experiments and wrote the manuscript. N. G., A. S., and K. E. performed the experiments.