Regulation of the Activity and Cellular Localization of the Circadian Clock Protein FRQ

Abstract

Eukaryotic circadian clocks employ autoregulatory negative feedback loops to control daily rhythms. In the filamentous fungus Neurospora, FRQ, FRH, WC-1, and WC-2 are the core components of the circadian negative feedback loop. To close the transcription-based negative feedback loop, the FRQ-FRH complex inhibits the activity of the WC complex in the nucleus by promoting the casein kinases-mediated WC phosphorylation. Despite its essential role in the nucleus, most FRQ is found in the cytoplasm. In this study, we mapped the FRQ regions that are important for its cellular localization. We show that the C-terminal part of FRQ, particularly the FRQ-FRH interaction domain, plays a major role in controlling FRQ localization. Both the mutation of the FRQ-FRH interaction domain and the down-regulation of FRH result in the nuclear enrichment of FRQ, suggesting that FRH regulates FRQ localization via a physical interaction. To study the role of FRQ phosphorylation, we examined the FRQ localization in wild-type as well as an array of FRQ kinase, FRQ phosphatase, and FRQ phosphorylation site mutants. Collectively, our results suggest that FRQ phosphorylation does not play a significant role in regulating its cellular localization. Instead, we find that phosphorylation of FRQ inhibits its transcriptional repressor activity in the circadian negative feedback loop. Such an effect is achieved by inhibiting the ability of FRQ to interact with WCC and casein kinase 1a. Our results indicate that the rhythmic FRQ phosphorylation profile observed is an important part of the negative feedback mechanism that drives robust circadian gene expression.

Introduction

The eukaryotic circadian oscillators are comprised of autoregulatory negative feedback loops (1–5). Despite the evolutionary distance between the filamentous fungus Neurospora crassa and higher eukaryotes, their circadian oscillator mechanisms share remarkable similarities (6–8). In N. crassa, two PAS (PER-ARNT-SIM) domain-containing transcription factors, WC-1 and WC-2, form a complex (WCC)² that activates the transcription of the frq gene by binding to its promoter (9–13). FRQ protein binds FRH to form the FRQ-FRH complex (FFC), which acts as the negative element in the circadian negative feedback loop (14–17). To close the circadian negative feedback loop, FFC decreases frq mRNA-levels by promoting frq mRNA degradation and by inhibiting frq transcription. To repress frq transcription, FFC inhibits WCC activity by recruiting casein kinase 1a (CK-1a) and CKII to phosphorylate the WC proteins, resulting in a decrease of WCC DNA binding activity and an increase of WCC nuclear export (9, 10, 14–24).

Like the animal PER proteins, FRQ is progressively phosphorylated upon its synthesis and becomes extensively phosphorylated before its disappearance, resulting in robust oscillations of its level and phosphorylation profile (25). Under normal physiological conditions, FRQ is phosphorylated by CK-1a, CKII, and PKA (21, 22, 25–28). On the other hand, FRQ is also dephosphorylated and stabilized by protein phosphatases PP1 and PP4 (20, 29). In the ck-1a, cka, and ckb-1 mutants, FRQ is hypophosphorylated and more stable than in the wild-type strain, leading to arrhythmic or long period conidiation rhythms in constant darkness (22, 26, 27). These results suggest that both casein kinases phosphorylate FRQ to promote its degradation. Conversely, PKA counters the role of casein kinases by stabilizing FRQ (21). Thus, FRQ phosphorylation plays opposing roles in regulating the stability of FRQ and setting the period length of the clock. Recently, we and others have identified >80 phosphorylation residues on FRQ by mass spectrometry analyses (30, 31) In addition, we have shown that CK-1a and CKII can phosphorylate most of the identified phosphorylation sites, indicating that these two casein kinases are the two major FRQ kinases (30). Systematic mutagenesis of identified phosphorylation sites showed that although most FRQ phosphorylation events promote its degradation, phosphorylation of the C-terminal part of FRQ results in its stabilization (30, 31). The phosphorylation-triggered FRQ degradation is mediated by the E3 ubiquitin ligase SCF^FWD-1, in which FWD-1 is the substrate recruiting subunit that interacts with phosphorylated FRQ (32). It is unclear whether FRQ phosphorylation has additional roles in the function of the clock.

The partner of FRQ, FRH, is a homolog of Mtr4p in Saccharomyces cerevisiae and sequence homologs are found from fungi to mammals (17). The entire pool of FRQ is in complex with FRH in Neurospora, and down-regulation of frh expression abolishes circadian rhythmicities and results in high frq RNA levels. In addition to its role in repressing WCC activity, we recently showed that the FFC also regulates frq mRNA levels posttranscriptionally by promoting frq RNA degradation through the exosome complex (16). Furthermore, the FRQ-FRH interaction is important for maintaining the steady state levels of FRQ (17, 33). Mutants in which the FRQ-FRH interaction has been disrupted, have low levels of FRQ resulting from its degradation through an FWD-1-independent pathway (33).

Although FRQ nuclear localization is essential for its function in the circadian clock, most FRQ is found in the cytoplasm (34). An SV-40-like nuclear localization signal located downstream of the coiled-coil domain is required and sufficient for the nuclear localization of FRQ, suggesting that the FRQ cellular distribution profile is also regulated by an active nuclear export process. Indeed, a recent study using a FRQ-FRB fusion protein showed that there is rapid nuclear-cytoplasmic shuttling of FRQ (35), but how FRQ cellular localization is regulated is not known. Previously, we and others (19, 20) have shown that phosphorylation of the WCs promotes their cytoplasmic localization, a process that is regulated by PP4 and probably PP2A. By monitoring the nuclear-cytoplasmic shuttling of hyperphosphorylated FRQ-FRB fusion protein, it was proposed that the phosphorylation of FRQ inhibits its nuclear import process (35). However, because hyperphosphorylation of FRQ-FRB in these experiments was achieved using cycloheximide, which blocks the synthesis of all proteins, it is not clear whether this observation is physiologically relevant.

In this study, by creating a series of FRQ deletion mutants, we mapped the FRQ domains that are important for FRQ cytoplasmic localization. We show that the C-terminal part of FRQ, particularly the FRQ-FRH interaction domain, plays a major role in determining the proper FRQ cellular localization profile. Mutation of the FRQ-FRH interaction domain and down-regulation of FRH both result in significant nuclear enrichment of FRQ, suggesting that the formation of the FFC is important for FRQ to adopt a proper structural conformation required for its nuclear export. We examined FRQ localization at different time points during a circadian cycle in a wild-type strain, in FRQ kinase mutants, in a series of mutants with FRQ phosphorylation sites mutated, and in strains that have hyperphosphorylated FRQ. Collectively, our results strongly suggest that FRQ phosphorylation and the major FRQ kinases do not play a significant role in regulating FRQ cellular localization. On the other hand, we find that FRQ phosphorylation inhibits its role as a transcriptional repressor by repressing FRQ-WC and FRQ-CK-1a interactions. Therefore, the robust oscillation of FRQ phosphorylation profiles during a circadian cycle is an important process that drives rhythmic transcriptional events in Neurospora.

EXPERIMENTAL PROCEDURES

Strains and Culture Conditions

The 87-3 (ras-1^bd, a) strain was used as a wild-type strain in this study. The FRQ mutant strains used in this study are described below (see Fig. 1A and Table 1). To generate FRQ deletions or phosphorylation site mutations, QuikChange site-directed mutagenesis (Stratagene) was carried out using pUC19Mfrq as the template. The mutated frq constructs were subcloned into pKAJ120 (containing the entire wild-type frq gene), and the resulting constructs were transformed into a bd frq¹⁰ his-3 strain at the his-3 locus by electroporation. All mutation constructs were confirmed by DNA sequencing. Liquid cultures were grown in minimal media (1× Vogel's, 2% glucose). When quinic acid was used to activate the qa-2 promoter, liquid cultures were grown in 0.01 m quinic acid (pH 5.8), 1× Vogel's, 0.5% glucose, and 0.17% arginine.

FIGURE 1.

Mapping of the domains important for FRQ localization. A, diagrams showing the domain structure of FRQ and deletion mutants used. CC, coiled coil domain; NLS, nuclear localization signal; FCD, FRQ-CK-1a interaction domain; FFD, FRQ-FRH interaction domain. FRQ amino acids deleted in each mutant are indicated by numbers. Alanine substitution mutations used in the following experiments are designated as FRQ4A11, FRQ4A12, and FRQ6B2. B and C, Western blot analyses showing the level of FRQ in the total extracts (T), nuclear (N), or cytosolic (C) fractions of various deletion mutants. WC-2 signals were used to verify nuclear fractions as this protein is known to be highly enriched in the nucleus. The use of α-tubulin antibody indicated that the nuclear fractions were not contaminated with cytosolic protein. The asterisks indicate nonspecific bands detected by our WC-1 antibody, which can serve as a loading control. Note that CD1∼CD3 can express both lFRQ and sFRQ, whereas sFRQ6∼8 can only express sFRQ. D, Western blot analyses comparing the nuclear/cytosolic distribution of sFRQ and lFRQ. Paired Student's t test showing the statistical significance of the results is indicated. Error bars indicate S.D.

TABLE 1.

FRQ mutant strains used

Strain	Description	Ref.
FRQ6B2	773–775 (DHF) replaced by alanines	Ref. 33
sFRQ and lFRQ	Two different isoforms generated by alternative splicing	Ref. 40
FRQ4A11	488–491 (LLCN) replaced by alanines	Ref. 22
FRQ4A12	494–496 (QLH) replaced by alanines	Ref. 22
M5	Ser211, Ser215, Ser216, Ser218, Thr219, and Ser220 replaced by alanines	Ref. 30
M7	244–282 deleted	Ref. 30
M9	Thr534, Ser537, Ser538, Ser540, and Ser541 replaced by alanines	Ref. 30
M10	Ser545, Ser548, Thr551, and Thr554 replaced by alanines	Ref. 30
M18	Ser967, Ser968, and Thr971 replaced by alanines	Ref. 30
C23A	Ser797, Ser800, Thr803, Ser818, Ser824, Ser900, Thr904, Thr915, Thr917, Ser923, Ser929, Ser950, Thr956, Ser967, Ser968, Thr971, Ser977, Ser980, Ser981, Ser982, Ser987, Ser988, and Ser989 replaced by alanines	This study

Protein Analyses

Nuclear and cytosolic protein extracts were prepared as described previously (34), and each fraction (20 μg) was loaded onto SDS-PAGE for Western blot analysis. The monoclonal α-tubulin antibody (Sigma) was used as a cytosolic marker. Our nuclear fractions did not show a tubulin signal, indicating the absence of cytosolic protein contamination. The bands corresponding to FRQ were scanned and quantified by ImageJ. Student's t tests were used to examine the statistical significance of the data.

To analyze the phosphorylation profiles of WC-2, 10% SDS-PAGE gels containing a ratio of 149:1 acrylamide/bisacrylamide were used (22). Otherwise, 7.5% SDS-PAGE gels containing a ratio of 37.5:1 acrylamide/bisacrylamide were used.

Immunoprecipitation

2 mg of soluble protein extracts prepared in extraction buffer (25) were incubated at 4 °C for 3 h with WC-2 antiserum (1:400 dilution). GammaBind G-Sepharose beads (GE Healthcare) were added and incubated at 4 °C for 1 h and then washed three times with ice-cold extraction buffer prior to the resuspension of immunoprecipitation products in protein loading buffer.

RNA Analyses

Total RNA was acquired and quantitative real time PCR experiments were performed as described previously (36, 37). For these experiments, the β-tubulin gene was used as a control. All of the primer sequences are available upon request.

RESULTS

Mapping of FRQ Domains Important for FRQ Cellular Localization

Fig. 1A shows the domain structure of FRQ, in which the coiled-coil domain, FRQ-CK-1a interaction domain, and FRQ-FRH interaction domain (FFD) were identified previously to mediate the FRQ-FRQ, FRQ-CK-1a, and FRQ-FRH interactions, respectively (17, 22, 33, 38). The nuclear localization signal of FRQ resides downstream of the coiled-coil domain and is essential for its nuclear localization (34). Two PEST domains also have been reported (28, 39). We created a series of FRQ deletion constructs that can express either large FRQ (lFRQ) or small FRQ (sFRQ) carrying various C-terminal or internal in-frame deletions (Fig. 1A, bottom). These FRQ expression constructs were individually transformed into a frq null strain (frq¹⁰), and the cellular localization of FRQ in the resulting transformants was examined in cultures grown in constant light. Similar to earlier results (17, 34), although a low level of FRQ was found in the nuclear fraction of the wild-type strain, most FRQ was found in the cytosolic fraction (Fig. 1B). The lack of tubulin signal in the nuclear fractions indicated that our nuclei preparations were free of cytoplasmic contamination. In contrast, the deletion of a C-terminal region (aa 698–989) in the CD1 strain resulted in a dramatic nuclear enrichment of FRQ protein (Fig. 1B), to a level that is comparable with those of the WC proteins (17, 20, 27). Deletion of aa 793–989 (CD2) also modestly increased the nuclear FRQ level, whereas the deletion of the C-terminal tail of FRQ in CD3 (aa 905–989) did not result in significant changes in FRQ cellular distribution. We then used a series of previously created sFRQ internal deletion mutants to further map the FRQ domain important for its cytoplasmic localization (17, 22). As shown in Fig. 1C, the internal deletion of aa 696–781 (sFRQ6) significantly increased the nuclear sFRQ levels. Taken together, these results suggest that the region between aa 698–781 plays a major role in promoting the cytosolic localization of FRQ.

Interestingly, even though the C-terminal deleted regions in the CD3 and sFRQ8 strains were very similar, we found that there was significantly more nuclear FRQ in the sFRQ8 mutant than in the wild-type strain, suggesting that the N-terminal part of FRQ that is also missing in sFRQ8 may promote FRQ cytoplasmic localization. We thus compared FRQ localization between strains that can only express lFRQ or sFRQ (40). As shown in Fig. 1D, the relative nuclear FRQ level was significantly higher in the sFRQ strain than in the lFRQ strain, indicating that the N-terminal 99 aa of FRQ also contribute to its cytoplasmic localization.

FRH Regulates Nuclear/Cytosolic Distribution of FRQ

We have previously shown that aa 728–797 of FRQ contain the FFD (17, 33). To understand the role of the FFD and FRH in the localization of FRQ, we examined FRQ localization in the FRQ6A and FRQ6B mutants of which only the former can interact with FRH (33). Although the nuclear enrichment of FRQ was not affected in the FRQ6A strain, it was significantly increased in the FRQ6B strain (Fig. 2A). Furthermore, nuclear enrichment of FRQ was also observed in the FRQ6B2 strain, which carries point mutations within the FFD that severely impair the FRQ-FRH interaction and make the FRQ protein unstable (Fig. 2B). These results suggest that the FRQ-FRH interaction regulates the nuclear/cytosolic distribution of FRQ.

FIGURE 2.

FRH and the FRQ-FRH interaction regulate FRQ localization. A and B, Western blot analyses showing FRQ localization in two internal deletion mutants (A) and an alanine-substitution mutant (B) of FRQ. Because the level of FRQ protein in FRQ6B2 is low, a longer exposure (long) of the same blot is also presented. C, down-regulation of frh results in an increase of relative nuclear FRQ levels. Quinic acid (QA) was used to induce silencing of frh expression in the dsfrh strain. FRQ localization was examined by Western blot analysis. T, total extracts; N, nuclear fractions; C, cytosolic fractions. Error bars indicate S.D. The asterisks indicate nonspecific bands.

To elucidate the role of FRH in regulating FRQ localization, we examined the FRQ nuclear/cytoplasmic distribution in a strain (dsfrh) in which the expression of frh can be inducibly silenced by the expression of an frh-specific hairpin RNA (17). As shown in Fig. 2C, the silencing of frh results in a significant increase in the relative levels of nuclear FRQ despite the large decrease in the steady state levels of FRQ. This result indicates that FRH, through its interaction with FRQ, promotes the cytoplasmic localization of FRQ.

Relationship between FRQ Phosphorylation Profile and Its Cellular Localization

The progressive phosphorylation of FRQ by several kinases function primarily to regulate the circadian period length by controlling FRQ degradation through the ubiquitin/proteasome pathway (30–32). Recently, it was proposed that CKI-mediated phosphorylation of FRQ may inhibit its nuclear localization as the hyperphosphorylation of a FRQ-FRB fusion protein induced by cycloheximide treatment resulted in a decrease of nuclear import (35). To test this hypothesis, we examined FRQ localization at different times in constant darkness (DD) and reasoned that the FRQ cellular localization profile should be rhythmic during a circadian cycle due to the rhythmic FRQ phosphorylation profiles. As shown in Fig. 3A, although both FRQ level and its phosphorylation profile were robustly rhythmic from DD16 (FRQ hypophosphorylated) to DD28 (FRQ hyperphosphorylated), and there was a rhythm of nuclear FRQ levels, the relative nuclear enrichment of FRQ was maintained at a similar level across all time points. This result suggests that the dynamics of FRQ nuclear/cytoplasmic shuttling does not change significantly in a circadian cycle.

FIGURE 3.

Major FRQ kinases do not play a significant role in regulating FRQ localization. A, Western blot analyses showing the circadian rhythms of FRQ levels and phosphorylation profile in nuclear and cytosolic fractions. Cultures of the wild-type strain were harvested at the indicated time points in DD. Note that the relative nuclear FRQ levels (Nuclear/Total ratio) were not significantly changed at different time points. B, Western blot analyses showing FRQ localization in ck-1a^L and cka^RIP mutants. C, Western blot analyses showing the localization of FRQ in FRQ mutants that cannot interact with CK-1a. T, total extracts; N, nuclear fractions; C, cytosolic fractions.

CK-1a and CKII are the two major kinases that phosphorylate and regulate FRQ (22, 27, 30, 41). We have previously shown that in both the ck-1a^L strain, in which CK-1a is only partially functional due to a point mutation, and in the cka^RIP strain, in which the sole catalytic subunit of CKII is disrupted by repeat induced point mutations, FRQ is hypophosphorylated, and the normal circadian clock functions are abolished (22, 26, 27). Comparison of these mutants with a wild-type strain showed that in constant light, the relative nuclear FRQ level was not significantly changed by mutation of the two major FRQ kinases (Fig. 3B). We further examined the role of CK-1a in FRQ cellular localization using two additional mutants, FRQ4A11 and FRQ4A12, grown in constant light. In these two mutants, point mutations within the FRQ-CK-1a interaction domain severely impair the interaction between FRQ and CK-1a and result in hypophosphorylation of FRQ. As shown in Fig. 3C, the relative nuclear levels of FRQ in these two mutants were not significantly different from that of the wild-type strain. To rule out the possibility that differences in FRQ localization between these mutants and the wild-type strain are masked by potential active nuclear export during nuclei preparation, we also isolated nuclei from cultures that were fixed with formaldehyde before harvesting. Again, we found that there was no significant difference in FRQ localization between the mutants and the wild-type strains in these cultures (data not shown). These results suggest that the major FRQ kinases do not play a major role in regulating FRQ localization under normal growth conditions.

We and others (30, 31) previously identified >80 FRQ phosphorylation sites and characterized their roles in the circadian clock through systematic mutagenesis. Mutation of many of these phosphorylation sites lengthened or shortened the period of the clock, which can be attributed to increased or decreased FRQ stability in these mutants, respectively. To further assess the potential role of FRQ phosphorylation in controlling its cellular localization, we examined whether FRQ localization was altered in these mutants. As shown in Fig. 4A, the relative nuclear FRQ levels in the long period mutants (M5, M7, M9, and M10) and in a short period mutant (M18) were comparable to that of the wild-type strain even though these mutants exhibit significantly altered periods (30).

FIGURE 4.

The known phosphorylation events of FRQ do not play a significant role in regulating its localization. A, FRQ localization was examined in some of the previously reported FRQ phosphorylation site mutants. Quantification of Western blot analyses results (left) are shown on the right. B, in the C23A strain, 23 phosphorylation sites in the C-terminal region of FRQ were mutated to alanines (Table 1). Race tube assays show that the C23A strain has short period conidiation rhythms. Black marks indicate the daily growth front. C, Western blot analyses showing the localization of FRQ in the C23A mutant. The quantification of three independent Western blot results is shown on the right. Error bars indicate S.D. T, total extracts; N, nuclear fractions; C, cytosolic fractions. The asterisks indicate nonspecific bands.

As FRQ has many phosphorylation sites, we wondered whether its localization is regulated by multiple, redundant phosphorylation events. Because the C-terminal part of FRQ plays a major role in promoting its cytoplasmic localization, we created a strain (C23A), in which 23 known Ser/Thr phosphorylation sites near the C-terminal end of the protein were mutated to alanine (Table 1). Consistent with our previous results that FRQ phosphorylation at the C-terminal end stabilizes FRQ (30), the C23A mutant exhibited a short period circadian conidiation rhythm (∼18.9 h), and its FRQ was hypophosphorylated (Fig. 4, B and C). However, despite the elimination of a large number of FRQ phosphorylation sites in this mutant, its relative nuclear FRQ level was comparable with that of the wild-type strain. These results suggest that the known phosphorylation events of FRQ do not play a significant role in regulating FRQ localization.

We next examined FRQ localization in Neurospora mutants that accumulate hyperphosphorylated FRQ. PP1 and PP4 are two major phosphatases that dephosphorylate FRQ (20, 29). In the ppp-1^RIP pp4^KO double mutant, which has a partially functional PP1 catalytic subunit resulting from point mutations and is null for pp4, FRQ is hyperphosphorylated, resulting in its rapid degradation and a short period phenotype (20). On the other hand, disruption of fwd-1, which encodes for the substrate-recruiting subunit of the ubiquitin E3 ligase that mediates FRQ degradation, results in the accumulation of hyperphosphorylated FRQ due to deficient phosphorylation-triggered FRQ degradation (32, 42). As shown in Fig. 5, A and B, despite their hyperphosphorylated FRQ profiles, these two mutants did not show a significant difference in the nuclear/cytosolic distribution of FRQ when compared with the wild-type strain. Taken together, we have presented several lines of evidence strongly suggesting that FRQ phosphorylation does not play a major role in regulating FRQ cellular localization. Recently, examination of a fluorescent FRQ-mCherry fusion protein suggested that there are two peaks of FRQ enrichment in the nucleus within a circadian cycle (43). Although it is not clear how the cellular distribution of FRQ-mCherry correlates with its phosphorylation states, this result suggests that regulation of FRQ cellular localization may be complex.

FIGURE 5.

FRQ nuclear/cytosolic distribution was not significantly altered in ppp-1^RIPpp-4^KO and fwd-1^RIP strains. Western blot analyses showing FRQ localization in the ppp-1^RIPpp-4^KO (A) and fwd-1^RIP (B) strains. Note that FRQ in the ppp-1^RIPpp-4^KO and fwd-1^RIP strains is hyperphosphorylated. The quantification of three independent Western blot results is shown on the right. Error bars indicate S.D. T, total extracts; N, nuclear fractions; C, cytosolic fractions. The asterisks indicate nonspecific bands.

Phosphorylation of FRQ Inhibits Its Transcriptional Repressor Activity and FRQ-WCC and FRQ-CK-1a Interactions

In constant darkness, although FRQ levels oscillate, the typical amplitude of FRQ oscillation is not high: a peak to trough ratio of 2–3. Such a low amplitude rhythm raises the possibility that additional regulations are responsible for the high amplitude circadian rhythms observed under normal conditions. The robust circadian oscillation of FRQ phosphorylation profile in a circadian cycle but the apparently insignificant role of FRQ phosphorylation in regulating cellular localization prompted us to examine other potential roles for FRQ phosphorylation. We decided to examine how FRQ phosphorylation affects its role in repressing frq transcription, which is its major role in the circadian negative feedback loop. Unlike the FRQ kinase or phosphatase mutants, which display altered phosphorylation of the WCs, the fwd-1^RIP mutant accumulates hyperphosphorylated FRQ without a direct effect on WCC activity and phosphorylation (32). As shown in Fig. 6A, in contrast to the robust circadian oscillations of FRQ level and phosphorylation profile in the wild-type strain, FRQ was hyperphosphorylated in the fwd-1^RIP strain and accumulated to high levels at all time points in DD. Despite the high levels of FRQ in the mutant, which was expected to result in constant inhibition of WCC activity, frq mRNA levels in the mutant were slightly elevated when compared with the wild-type strain at equivalent time points (Fig. 6B). In addition, the mRNA levels of ccg-1 (clock controlled gene-1), which are negatively regulated by the functional FRQ-based circadian oscillator (44), are significantly higher in the fwd-1^RIP mutant than the wild-type strains (Fig. 6C). These results suggest that despite its high levels, the hyperphosphorylated FRQ in the fwd-1^RIP mutant is severely impaired in its function as a transcriptional repressor in the circadian negative feedback loop.

FIGURE 6.

Phosphorylation of FRQ inhibits its functions in the circadian negative feedback loop. A, Western blot analyses showing the levels of FRQ protein in the wild-type and fwd-1^RIP strains. Densitometry of the Western blot results is shown at the bottom. B and C, quantitative real-time PCR results showing the levels of frq and ccg-1 mRNA in the wild-type and fwd-1^RIP strains at different time points in DD (h. DD). D, immunoprecipitation assay using an WC-2 antibody showing the association of FRQ with the WCC in the wild-type and fwd-1^RIP strains. Preimmune serum (PI) was used as a negative control. The ratio of FRQ signals in immunoprecipitation to input lanes was calculated for each strain. The ratio of FRQ associated with WCC in the wild-type strain was set as 1 to facilitate comparisons with the mutant. The error bar indicates the S.D. from three independent experiments. The result from the Student's t test is also indicated.

Because FRQ mediates its transcriptional suppressor role by interacting with the WCC (38, 45), we wanted to determine whether phosphorylation of FRQ affected its interaction with the WCs. To this end, we performed immunoprecipitation assays using a WC-2 antibody to compare the FRQ-WCC interaction between the wild-type and fwd-1^RIP strains. As shown in Fig. 6D, although FRQ levels were much higher in the fwd-1^RIP mutant, a comparable amount of FRQ was pulled down in the wild-type and fwd-1^RIP strains. Compared with the wild-type strain, the relative amount of FRQ associated with WC-2 was reduced by half in the mutant, suggesting that the hyperphosphorylated FRQ has a low affinity to the WCC. Consistent with this notion, the FRQ associated with WC-2 in the fwd-1^RIP strain was not enriched for the hyperphosphorylated forms despite the high levels of hyperphosphorylated FRQ found in total extracts of the mutant. These results suggest that in addition to regulating FRQ stability, phosphorylation also inhibits the ability of FRQ to repress WCC activity by affecting the FRQ-WCC interaction. Consistent with our results, it was previously shown that relative level of the FRQ-WC-2 interaction is highest when FRQ is newly synthesized and hypophosphorylated (31).

The significant but modest decrease of the FRQ-WC interaction does not fully explain the severe impairment of FRQ function in the fwd-1^RIP strain, indicating that FRQ phosphorylation has an additional role in regulating its activity. FRQ inhibits WCC activity by recruiting CK-1a and CKII to phosphorylate the WC proteins and the FRQ-CK-1a interaction is essential for the negative feedback process (22). Thus, we compared the FRQ-CK-1a interaction in the wild-type and fwd-1^RIP strains. Both strains were transformed with a construct that expresses the Myc-tagged CK-1a. As shown in Fig. 7A, in contrast to the wild-type strain, very little FRQ protein was pulled down with Myc-CK-1a despite the high levels of the hyperphosphorylated FRQ in the fwd-1^RIP strain. This result suggests that phosphorylation of FRQ strongly inhibits the FRQ-CK-1a interaction, resulting in low FRQ activity in repressing WCC activity. Furthermore, we found that WC-2 protein was hypophosphorylated in the fwd-1^RIP strain, consistent with the impairment of FRQ to promote WCC phosphorylation (Fig. 7B). Together, our results suggest that the rhythmic FRQ phosphorylation profile is an important mechanism that leads to rhythmic changes in FRQ activity. Thus, the combined rhythms of FRQ level and its phosphorylation profile result in robust oscillation of FRQ activity and frq transcription.

FIGURE 7.

Phosphorylation of FRQ inhibits the FRQ-CK-1a interaction, resulting in reduced phosphorylations of WC-2. A, immunoprecipitation assay using a Myc antibody showing the association of FRQ with Myc-CK-1a in the wild-type and fwd-1^RIP strains. The wild-type strain without the Myc-CK-1a construct was used as a negative control. The asterisk indicates the IgG signal. The ratio of FRQ signals in immunoprecipitation (IP) to input lanes was calculated for each strain, and the ratio in the wild-type strain with Myc-CK-1a was set as 1. The error bar indicates the S.D. from three independent experiments. The result from the Student's t test is also indicated. B, Western blot analysis showing the phosphorylation profiles of WC-2 in the wild-type and fwd-1^RIP strains. The arrows indicate the hyperphosphorylated WC-2 species absent in the fwd-1^RIP strain.

DISCUSSION

Regulation of the cellular localization of clock proteins plays important roles in the control of eukaryotic circadian clocks. Although the functions of the Neurospora clock protein FRQ require its nuclear localization, most FRQ resides in the cytoplasm due to nuclear export (34, 35). In this study, we examined the regulation of the nuclear/cytosolic distribution of FRQ protein. By creating a series of FRQ deletion mutants, we showed that the C-terminal part of FRQ plays a major role in its cytosolic localization. Importantly, deletions and point mutations of the previously identified FRQ-FRH interaction domain result in the significant enrichment of nuclear FRQ, suggesting that the FRQ-FRH interaction is important for FRQ cellular localization. Further supporting this conclusion, we showed that the down-regulation of FRH led to an increase in relative FRQ levels in the nucleus. This is particularly interesting because the FFC acts as the negative element of the circadian negative feedback loop, and all FRQ proteins are found in complex with FRH (16, 17). The interaction between FRQ and FRH is also important for FRQ stability as the down-regulation of FRH results in lower cellular FRQ levels due to degradation through an FWD-1-independent pathway (33). Taken together, these data suggest that the formation of the FFC allows FRQ to adopt a proper structural conformation that is essential for its stability, function, and cellular localization.

In addition to the FRQ-FRH interaction domain, deletion of the N-terminal 99 aa of FRQ or the region downstream of the FRQ-FRH interaction domain also modestly affect FRQ localization. Several putative leucine-rich nuclear export signals (46) (such as aa 488–501 and aa 952–963) exist on FRQ, but mutations of these putative nuclear export signals had no effect on FRQ nuclear/cytosolic distribution (data not shown). Together, our results suggest that FRQ cytoplasmic localization does not employ a conventional cytoplasmic export signal but rather is predominantly controlled by nuclear import and export processes that are regulated by the three-dimensional structure of FRQ. Future studies on the mechanisms of nuclear import and export of FRQ will help us understand how FRQ cellular localization is determined.

In Drosophila, the cellular localization of the PER protein is also regulated by its protein partner TIM, although TIM is not required for PER nuclear localization (47–49). In addition, PER levels are low in tim⁰¹ mutants (50), suggesting that the PER-TIM complex may also allow PER to adopt a proper structural conformation that is important for its levels and cellular localization.

Phosphorylation of clock proteins by casein kinases is a common theme in the eukaryotic circadian oscillators from Neurospora to mammals (7, 8, 27, 51–54). In Neurospora, we and others (9, 18–20, 22, 24) have previously shown that the phosphorylation of the WC proteins inhibits their DNA binding activities and promotes their cytoplasmic localization. In addition, we showed that PKA and PP4 play opposing roles in regulating WCC localization; PKA inhibits WCC nuclear localization, whereas PP4 promotes WCC nuclear entry (20).

Recently, Diernfellner et al. (35) showed that the hyperphosphorylation of a FRQ-FRB fusion protein induced by cycloheximide treatment impaired its nuclear import. In addition, a CKI-specific inhibitor increased the relative nuclear levels of the FRQ-FRB fusion protein after cycloheximide treatment. These results led to the hypothesis that CK-1a-mediated FRQ phosphorylation regulates FRQ cellular localization. In the current study, we used various mutants and performed a series of experiments under normal growth conditions to examine the role of FRQ phosphorylation in regulating its localization. Collectively, our results strongly suggest that phosphorylation of FRQ does not play a major regulatory role in its cellular localization. First, although there was a robust circadian oscillation of FRQ phosphorylation profiles in wild-type cultures grown in constant darkness, the relative nuclear enrichment of FRQ was not rhythmic, suggesting that phosphorylation profiles of FRQ does not affect its cellular localization significantly under normal circadian conditions. Second, the relative nuclear enrichment of FRQ levels in the ck-1a^L and cka^RIP strains was not significantly different from that of the wild-type strain. Furthermore, we found that in strains that have a mutated FRQ-CK-1a interaction domain, the relative nuclear FRQ levels were normal, despite their hypophosphorylated profiles resulting from impaired CK-1a-mediated phosphorylation. These results strongly suggest that CK-1a and CKII, the two major FRQ kinases, do not have a major role in regulating FRQ localization. Third, we examined a series of FRQ phosphorylation site mutants that exhibit significant clock phenotypes (30, 31), and we found no defect in FRQ cellular localization in these strains. In addition, a mutant in which all known FRQ phosphorylation sites near the C-terminal end were mutated also exhibited a normal FRQ nuclear/cytoplasmic distribution profile. These results suggest that the known FRQ phosphorylation events do not play a significant role in regulating its cellular localization. Finally, we examined FRQ localization in ppp-1^RIP, pp-4^KO, and fwd-1^RIP, two mutant strains in which FRQ is constantly hyperphosphorylated due to the impaired FRQ dephosphorylation process or phosphorylation-mediated degradation, respectively (20, 29, 32), and found that both strains display FRQ cellular localization profiles similar to that of the wild-type strains. Taken together, our results argue strongly that FRQ phosphorylation is not a major factor that influences its cellular localization profile. Although we do not know why the cycloheximide-induced hyperphosphorylation of the FRQ-FRB fusion protein inhibits its nuclear import, it is possible that blocking protein synthesis may also inhibit some proteins that are important for FRQ nuclear import. It is also worth noting that the degree of FRQ hyperphosphorylation induced by cycloheximide treatment is not observed under normal growth conditions (23, 40).

Although our results do not support a role for FRQ phosphorylation in regulating its localization, they do suggest that phosphorylation of FRQ inhibits its ability to repress WCC activity. In the fwd-1^RIP strain, FRQ is hyperphosphorylated and remains at a high level due to the inhibition of phosphorylation-mediated FRQ degradation through the ubiquitin/proteasome pathway (32, 42). Despite the high levels of FRQ in this mutant, frq mRNA levels are slightly higher than those of the wild-type strain in constant darkness, suggesting that the hyperphosphorylated FRQ is severely compromised in its role as a repressor of WCC activity. Furthermore, we showed that this deficiency is due to reduced FRQ-CK-1a and FRQ-WCC interactions in the fwd-1^RIP strain. Consistent with a role for FRQ phosphorylation in regulating the FRQ-WCC interaction, it was previously shown that in constant darkness, relative levels of the FRQ-WC-2 interaction are high when FRQ is newly synthesized and hypophosphorylated and low when FRQ is extensively phosphorylated (31). Taken together, these results suggest that in addition to regulating its stability, FRQ phosphorylation also inhibits its ability to repress WCC. Thus, the rhythmic phosphorylation profiles of FRQ will result in rhythmic FRQ activity. In combination with the rhythm of FRQ levels, the rhythm of FRQ phosphorylation profile results in a high amplitude rhythm in FRQ activity in a circadian cycle, leading to robust rhythms of frq transcription and the expression of clock-controlled genes. Due to the similarities of clock mechanisms between Neurospora and animals, our results raise the possibility that the phosphorylation of the PER proteins may regulate their activities in the circadian negative feedback loops.