FRQ-CK1 Interaction Underlies Temperature Compensation of the Neurospora Circadian Clock

The circadian period of wild-type Neurospora is temperature compensated between 20°C and 30°C, with a temperature coefficient (Q₁₀) of approximately 1.05 (58) (Fig. 1A). Above 30°C, however, temperature compensation is partially lost, and the period becomes markedly shorter as temperature increases (∼19 h at 34°C), with a Q₁₀ value of about 1.28 (Fig. 1A) (58). To understand the mechanism of temperature compensation, we first examined whether a change in FRQ stability was responsible for the impaired temperature compensation in the wild-type Neurospora strain above 30°C. Comparison of the FRQ degradation rates at 25, 30, and 34°C after the addition of protein biosynthesis inhibitor cycloheximide (CHX) showed that FRQ stability was not significantly different at these temperatures (Fig. 1B). To exclude the effect of CHX at different temperatures, we measured the degradation rate of FRQ following a light-to-dark (LD) transition and found the stability of FRQ was also similar at different temperatures (Fig. 1C). This result indicated that the partial loss of temperature compensation in the wild-type strain above 30°C cannot be explained by a change in FRQ stability.

The long period mutant frq⁷ is a Neurospora clock mutant that has a partial loss of temperature compensation between 20 to 30°C (58, 59). As temperature was increased, the period of frq⁷ shortened more rapidly than the period of the wild-type strain (Fig. 1D). To determine whether the FRQ stability correlated with the temperature compensation defect of frq⁷, we compared FRQ degradation rates at 21°C and 29°C. As expected, FRQ stability was similar at both temperatures for the wild-type strain (Fig. 1E). In frq⁷, however, FRQ was more stable at 29°C than at 21°C (Fig. 1E). Similar degradation profiles were also observed after an LD transition (Fig. 1F). Since increased stability of FRQ is normally correlated with period lengthening, this is opposite to what is predicted if a change of FRQ stability is responsible for its period shortening at high temperatures. Together, these results indicated that maintenance of FRQ stability is not the mechanism underlying temperature compensation in Neurospora.

Because of the critical role of FRQ phosphorylation in regulating circadian period length in Neurospora, we examined whether the inhibition of FRQ phosphorylation affects temperature compensation. The general kinase inhibitor 6-DMAP has been shown to inhibit FRQ phosphorylation and lengthen the circadian period in vivo (17). As expected, 6-DMAP treatment lengthened the period of the conidiation rhythms of the wild-type strain at 25°C in a dose-dependent manner (Fig. 2A). However, as temperature increased from 21°C to 32°C, the effect of 6-DMAP treatment on period length decreased, and 6-DMAP had only a very modest effect on period length at 32°C. Q10 (21 to 30°C) increased from approximately 1.07 in the absence of 6-DMAP to approximately 1.25 at 200 μM 6-DMAP, indicating that the 6-DMAP treatment resulted in a partial loss of temperature compensation in a dose-dependent manner.

Since CK1a (encoded by NCU00685) is the major kinase that phosphorylates FRQ, we examined the involvement of CK1a in temperature compensation. PF-670462 was a CK1-specific inhibitor that was previously shown to inhibit FRQ phosphorylation (45, 60). As expected, PF-670462 treatment resulted in hypophosphorylation of FRQ in the wild-type strain and period lengthening (Fig. 2B; see also Fig. S1A in the supplemental material). Similar to 6-DMAP treatment, PF-670462 treatment also resulted in partial loss of temperature compensation of the circadian conidiation rhythms (Fig. 2B), suggesting that CK1 activity is critical for temperature compensation.

ck1a is an essential gene in Neurospora (47). Although we previously created a ck1a^L (M83I) knock-in mutant that exhibited a long-period circadian phenotype (24), growth and development of this mutant were severely impaired, making it difficult to examine its circadian phenotypes at different temperatures. To obtain genetic evidence of the involvement of CK1a in temperature compensation, we generated another ck1a knock-in mutant in which a conserved histidine residue at position 123 is mutated to tyrosine (Fig. 2C). The same mutation in the budding yeast CK1 homolog YCK2 was previously shown to regulate CK1 activity (61), and a similar mutation in Drosophila doubletime (dbt) resulted in arrhythmicity or long circadian period (62). In Neurospora, the CK1a H123Y mutation led to hypophosphorylation of FRQ and decreased CK1a protein levels (Fig. 2D). The decreased CK1a^H123Y protein levels were not caused by degradation, indicating H123Y mutation affects protein folding of CK1a (Fig. S1B) or posttranscriptional regulation of CK1a (63). As expected, the ck1a^H123Y mutant exhibited a long period at 25°C (Fig. 2E). Comparison of the period lengths at different temperatures showed that the Q₁₀ value of the ck1a^H123Y strain was increased to 1.26, indicating a partial loss of temperature compensation (Fig. 2E). Together, these results indicated that CK1a is involved in regulating the mechanism of temperature compensation.

To investigate how CK1a regulates temperature compensation, we examined whether CK1a protein levels or kinase activity are temperature sensitive above 30°C when temperature compensation of the wild-type strain becomes impaired. The endogenous CK1a (two alternatively spliced CK1a isoforms) levels did not change significantly from 20°C to 35°C (Fig. 3A), indicating that a temperature-dependent change in CK1a protein levels is not the cause of the partial loss of temperature compensation above 30°C. To determine the biochemical activity of CK1a, we expressed the recombinant full-length Neurospora CK1a in Escherichia coli and examined the kinase activity of the purified protein in an assay with two glutathione S-transferase (GST)-tagged recombinant FRQ fragments, FRQ (534-562) and FRQ (425-683), as the substrate at different temperatures. CK1a activity toward FRQ peptides in vitro was relatively temperature insensitive, and there was no significant difference in kinase activity between 30 and 35°C (Fig. 3B). This result suggested that CK1a kinase activity is relatively insensitive to temperature changes; therefore, changes in kinase activity as a function of temperature alone do not explain the partial loss of temperature compensation above 30°C in the wild-type strain. Interestingly, it was previously reported that the mammalian CK1ε/δ activity toward short peptides is also relatively temperature insensitive in vitro (34).

Because the FRQ-CK1a interaction was the major determinant of FRQ phosphorylation and circadian period length at room temperature and the relative level of this interaction was negatively correlated with period length (25), we examined whether the FRQ-CK1a interaction was sensitive to temperature above 30°C in the wild-type strain. A construct for expression of Myc-CK1a was transformed into Neurospora to allow the relative interaction between FRQ and CK1a to be monitored by immunoprecipitation (IP) assays using a c-Myc monoclonal antibody (25). Although the relative level of FRQ-CK1a complex (ratio of CK1a-immunoprecipitated FRQ level to the input FRQ level that is normalized with the CK1a level) was similar between 25°C and 30°C, it was significantly increased at 34°C (Fig. 3C). In contrast, the amount of FRQ-WC2 complex was not affected by temperature (Fig. 3D). To further confirm this result, we performed an in vitro pulldown assay by incubating the recombinant Flag-tagged CK1a with extracts of wild-type Neurospora grown at different temperatures. As expected, relatively more FRQ was precipitated with Flag-CK1a from extracts of Neurospora grown at 34°C than at 25°C and 30°C (Fig. S2). Because an increase of the relative FRQ-CK1a interaction resulted in period shortening (25), these findings suggested that at temperature above 30°C, FRQ-CK1a interaction was specifically increased without affecting the FRQ-WC interaction, resulting in enhanced WC complex phosphorylation and inhibition and leading to a shorter period. Therefore, the increase in FRQ-CK1a interaction should be responsible for the partial loss of temperature compensation of the wild-type strain above 30°C.

If the FRQ-CK1a interaction plays a major role in temperature compensation, it should also explain the temperature compensation defects in clock mutants with impaired temperature compensation. Thus, we examined the FRQ-CK1a interaction in the frq⁷ strain at different temperatures. The CK1a immunoprecipitation assay showed that the relative amount of FRQ precipitated was increased at 29°C compared to 21°C in the frq⁷ strain but not in the wild-type strain (Fig. 3E). This is consistent with the partial loss of temperature compensation in the frq⁷ strain (Fig. 1C). Thus, the altered FRQ-CK1a interaction also explains the partial loss of temperature compensation of the frq⁷ strain.

We hypothesized that a temperature increase above 30°C induces structural changes in FRQ that affect the FRQ-CK1a interaction in the wild-type strain. To test this possibility, we used a limited trypsin digestion assay performed at 30°C to probe the structure of FRQ in Neurospora extracts harvested from the wild-type strain grown at different temperatures in constant light. In extracts from the wild-type culture grown at 34°C, FRQ was significantly more sensitive to trypsin digestion than FRQ from culture grown at 30°C (Fig. 3F), suggesting that a change in the structural conformation of FRQ above 30°C affects FRQ-CK1a interaction.

CK2 was previously shown to be a regulator of temperature compensation in Neurospora (36). In contrast to the wild-type strain, ck2 mutants are temperature overcompensated, meaning that the period lengthens as temperature increases. Although CK2 is also an FRQ kinase, it does not form a stable complex with FRQ (13, 19, 48, 49). Consistent with previous results (36), we observed that the period length of the ck2 mutant ckb^RIP (48) became longer as the temperature increased (Fig. 4A). We compared the relative amounts of FRQ associated with Myc-CK1a at different temperatures in the wild-type and ckb^RIP strains. FRQ was hypophosphorylated in the ckb^RIP strain at 34°C in constant light. In contrast to the significant increase in FRQ precipitated with CK1a at 34°C relative to 25°C observed in the wild-type strain, their relative association was reduced at the higher temperature in the ckb^RIP mutant (Fig. 4B). In addition, the relative level of FRQ-CK1a interaction was lower in the ckb^RIP strain at all temperatures compared to the wild-type strain, consistent with its long-period phenotype. Thus, alterations of the level of the FRQ-CK1a interaction also explain the temperature overcompensation phenotype of the ckb^RIP mutant.

The effect of CK2 on temperature compensation suggested that CK2-dependent FRQ phosphorylation caused a temperature-dependent decrease of FRQ-CK1a interaction and an increased period length at higher temperatures. To confirm this, we screened FRQ phosphorylation site mutants for temperature overcompensation phenotype. The frq M19 mutant has mutations in several phosphorylation sites that match the CK2 consensus sequence (Fig. S3A) (14). Similar to the ckb^RIP strain, the circadian conidiation rhythms of M19 also exhibited a temperature overcompensation phenotype (Q₁₀ of 0.84), and its period was increased from 21.1 h at 21°C to 25.5 h at 32°C (Fig. 4C). As predicted, the relative level of the FRQ-CK1a association in extracts of the M19 mutant was significantly decreased at 34°C compared to lower temperatures (Fig. 4D).

To confirm the sites in FRQ mutated in M19 were phosphorylated by CK2 and not by CK1a, we performed in vitro phosphorylation assays using recombinant Neurospora CK1a or CK2 with purified GST-fusion peptides containing the wild-type region or the M19 region (Fig. S3A). The wild-type GST-fusion peptide was readily phosphorylated by CK2 but not by CK1a; the peptides containing the M19 mutations were not phosphorylated by CK2 (Fig. S3B and C). These experiments indicate that these sites are indeed CK2 sites. Together, these results indicate that some of the CK2-dependent FRQ phosphorylation causes a decrease in the FRQ-CK1a interaction as temperature increases, resulting in the temperature overcompensation phenotype in the ckb^RIP and M19 strains. Therefore, altered FRQ-CK1a interactions can explain both the loss of temperature compensation and the temperature overcompensation phenotypes of the different clock mutants.

Since our data suggested that the FRQ-CK1a interaction is a major factor that determines temperature compensation, we hypothesized that mutation of the FRQ-CK1a interaction domain would change the temperature compensation profile. To test this, we used Neurospora strains with mutations within the FCD1 and FCD2 domains of FRQ that have been previously shown to weaken but not abolish the interaction between FRQ and CK1a, resulting in long-period phenotypes (25). We examined the temperature dependence of circadian conidiation rhythms of mutants with single-amino-acid substitutions in either FRQ FCD1 (L323M and V320I) or FCD2 (L488V, C490G, and Q494N). As expected, all of these FCD mutants displayed partial loss of temperature compensation (Q₁₀ value ranging from 1.22 to 1.32) as their period length shortened more rapidly as temperature increased than the wild-type strain (Fig. 5A to C).

To confirm that the temperature compensation phenotypes of the FCD mutants are indeed due to the alterations in the stability of the FRQ-CK1a complex, we examined the FRQ-CK1a interaction in the Q494N mutant by immunoprecipitating the endogenous CK1a. As expected, the relative amount of the FRQ-CK1a complex in the Q494N mutant was significantly higher at 29°C than at 21°C (Fig. 5D). On the other hand, FRQ became more stable at 29°C than at 21°C in the Q494N mutant following either CHX treatment (Fig. 5E) or LD transition (Fig. S4A), further confirming that temperature-dependent changes in FRQ stability do not mediate temperature compensation. Together, these results strongly suggest that the maintenance of FRQ-CK1a interaction underlies the molecular basis of circadian temperature compensation in Neurospora.

The above-described results suggest that the impaired temperature compensation is due to enhanced CK1a-mediated FRQ phosphorylation at high temperatures (Fig. 3). Thus, we predicted that impaired temperature compensation also can be caused by mutating some of the CK1a target sites on FRQ. FRQ protein is phosphorylated by CK1a and other kinases over 100 sites, and we have previously made different mutants in which multiple putative CK1a phosphorylation sites on FRQ are mutated (13, 14). M9 and M10 are two long-period mutants (29.8 and 32.1 h at 25°C, respectively) in which CK1a phosphorylation sites (FRQ 534-562) downstream of the FCD-2 domain of FRQ are mutated. Race tube assays showed that the increase of temperature from 21°C to 29°C resulted in a marked decrease of the period-lengthening effects in both mutants. The Q₁₀ increased from approximately 1.14 for the wild type (wild-type frq complementation of frq^KO) strain to approximately 1.39 and 1.33 for M9 and M10, respectively (Fig. 6A). These results confirmed that phosphorylation of FRQ mediated by CK1a is involved in the mechanism of temperature compensation.

How does the phosphorylation of the M9 and M10 region affect temperature compensation? In vitro kinase assay showed that the M9 and M10 region of FRQ (534-562) could be readily phosphorylated by CK1a, and the phosphorylation is relatively temperature insensitive (Fig. 3B). The M9 and M10 mutations were previously shown to weaken the FRQ-CK1a interaction (25). Thus, we tested whether the phosphorylation site mutation influences the structural conformation of FRQ, which, in turn, can affect the binding of FRQ with CK1a at different temperatures. We immunoprecipitated FRQ by endogenous CK1a in the M9 or M10 mutant and found that the relative amounts of the FRQ-CK1a complex in the M9 and M10 mutants were significantly higher at 29°C than those at 21°C. This result explains the partial loss of temperature compensation of these mutants (Fig. 6B). In contrast, FRQ became more stable at 29°C than at 21°C in the M9 and M10 mutants following either CHX treatment (Fig. 6C) or LD transition (Fig. S4B), further confirming that the temperature-dependent changes in FRQ stability were not responsible for their temperature compensation phenotypes. Together, these results strongly suggest that the CK1a-mediated FRQ phosphorylation at some sites is important for the maintenance of temperature compensation of circadian clock in Neurospora.

The 87-3 (bd, a) strain or an frq complementation strain (KAJ120) was used as the wild-type strain in this study. M9, M10, M19, frq7, ckb^RIP, V320I, L323M, L488V, C490G, and Q494N strains were created previously (14, 25, 48). Constructs with the qa-2 promoter driving expression of Myc.His.CK1a.hph were introduced into the 87-3 and ckb^RIP strains at the csr locus by homologous recombination. Constructs with the qa-2 promoter driving expression of Myc.His.CK1a.bar were introduced into the M19 strains at the csr locus by homologous recombination. Positive transformants were identified by Western blot analyses, and homokaryon strains were isolated by microconidial purification using 5-μm filters.

Liquid cultures were grown in minimal medium (1× Vogel’s, 2% glucose). When quinic acid (QA) was used to activate the qa-2 promoter, liquid cultures were grown in 1 × 10⁻² M QA (pH 5.8), 1× Vogel’s, 0.1% glucose, and 0.17% arginine. For rhythmic experiments, Neurospora was cultured in petri dishes in liquid medium for 2 days. The Neurospora mats were cut into discs, transferred into medium-containing flasks, and harvested at the indicated time points.

The medium for race tube assays contained 1× Vogel’s salts, 0.1% or 0.03% glucose, 0.17% arginine, 50 ng/ml biotin, and 1.5% agar with or without QA. After entrainment for 24 h in constant light, race tubes were transferred to constant dark and the growth front marked every 24 h. The position of the conidiation band was used to determine the time of banding (circadian period of the Neurospora) relative to the growth front marks. Q₁₀ value was calculated by the equation Q₁₀ = (τ1/τ2)^10/(T2‐T1), where τ1 and τ2 are the periods at temperature T1 and T2, respectively.

To create the ck1a knock-in strain, a wild-type knock-in cassette with a hygromycin resistance gene (hph) was inserted downstream of the ck1a 3′ untranslated region (UTR). The cassette was then transformed into the ku70^RIP strain to select for hph-resistant transformants. The homokaryotic strains were obtained by microconidial purification, crossing with a wild-type (301-6) strain, and confirmed by DNA sequencing (24, 68). Using this method, we created the ck1a^WT and ck1a^H123Y mutant knock-in strains at the ck1a endogenous locus.

Protein extraction, quantification, and Western blot analyses were performed as previously described (69–71). Briefly, tissue was ground in liquid nitrogen with a mortar and pestle and suspended in ice-cold extraction buffer (50 mM HEPES [pH 7.4], 137 mM NaCl, 10% glycerol) with protease inhibitors (1 μg/ml pepstatin A, 1 μg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). After centrifugation, protein concentration was measured using a protein assay kit (Bio-Rad). For Western blot analyses, equal amounts of total protein (40 μg) were loaded in each lane of 7.5% or 10% SDS-PAGE gels containing a ratio of 37.5:1 acrylamide to bisacrylamide. After electrophoresis, proteins were transferred onto polyvinylidene difluoride (PVDF) membranes, and Western blot analyses were performed. Western blot signals were detected by X-ray films and scanned for quantification.

Immunoprecipitation analyses were performed as previously described (69). Briefly, Neurospora proteins were extracted as described above. For each immunoprecipitation reaction, 2 mg protein and 1.5 μl c-Myc antibody, 1 μl WC-2 antibody (72), or 1 μl CK1a antibody were used. After incubation with antibody for 3 h, 40 μl GammaBind G Sepharose beads (GE Healthcare) was added, and samples were incubated for 1 h. For the Flag pulldown assay, 4 mg purified Flag-CK1a fusion protein and 2 μl Flag antibody were incubated with 2 mg Neurospora extracts. After incubation with antibody for 3 h, 40 protein GammaBind G Sepharose beads were added, and samples were incubated for 1 h. Immunoprecipitated proteins were washed three times using extraction buffer before Western blot analysis. IP experiments were performed using cultures harvested in constant light. Quantification of relative FRQ-CK1a or FRQ-WC-2 interaction levels is based on the ratio of IP to input and normalized with CK1a or WC-2 levels, respectively. All FRQ isoforms (hyperphosphorylated and hypophosphorylated FRQ) and all CK1 isoforms (the endogenous CK1a has two spliced isoforms, and quinic acid-induced CK1a has three isoforms) were used for quantifications.

For the in vitro phosphorylation analyses, His-CK1a, His-CK2a, GST-FRQ (534-562), GST-FRQ (425-683), GST-FRQ (972-990), and GST-FRQ (972-990)-M19 fusion proteins were purified from Escherichia coli BL21(DE3) cells (49, 68). His-CK1a or His-CK2a fusion protein (0.5μg) was incubated with GST-FRQ (534-562), GST-FRQ (425-683), GST-FRQ (972-990), or GST-FRQ (972-990)-M19 fusion protein (5 μg) in the reaction buffer (total, 20 μl) containing 25 mM HEPES (pH 7.9), 10 mM MgCl₂, 2 mM MnCl₂, 25 μM ATP, and 2 μCi of [γ-³²P]ATP (49). The reaction mixture was incubated for 15, 30, and 60 min at temperatures ranging from 20°C to 35°C, and the proteins were boiled in 1× SDS loading buffer and separated by SDS-PAGE. After electrophoresis, the gel was dried and the level of phosphorylation was assessed by autoradiography. Each experiment was independently performed at least three times.

To examine FRQ protein stability, discs made from 2-day-old mycelial mats of individual strains were grown in liquid medium for 1 day under constant light conditions. Cultures were grown in constant light for 1 day prior to the addition of CHX to a final concentration of 10 μg/ml or were transferred into constant darkness and harvested at the indicated time.

For the trypsin sensitivity assay, protein extracts were diluted to a protein concentration of 2.5 μg/μl. A 100-μl aliquot was treated with trypsin (final concentration, 1 μg/ml) at 25 °C. A 20-μl sample was taken from the reaction mixture at each time point (0, 5, 15, and 30 min) after addition of trypsin. Protein samples were mixed with protein loading buffer and resolved by SDS-PAGE. To compare trypsin sensitivity of FRQ from different strains, experiments were performed side by side, and the protein samples were transferred to the same membrane for Western blot analysis.

Quantification of Western blot data was performed using Image J software (73). All studies were performed on at least three independent experiments. Error bars are standard deviations for immunoprecipitation assays and standard errors of the means for race tube assays. Statistical significance was determined by Student’s t test.

Materials are available from the corresponding authors upon reasonable request.