Differential regulation of phosphorylation, structure, and stability of circadian clock protein FRQ isoforms

Neurospora crassa is an important model organism for circadian clock research. The Neurospora core circadian component FRQ protein has two isoforms, large FRQ (l-FRQ) and small FRQ (s-FRQ), of which l-FRQ bears an additional N-terminal 99-amino acid fragment. However, how the FRQ isoforms operate differentially in regulating the circadian clock remains elusive. Here, we show l-FRQ and s-FRQ play different roles in regulating the circadian negative feedback loop. Compared to s-FRQ, l-FRQ is less stable and undergoes hypophosphorylation and faster degradation. The phosphorylation of the C-terminal l-FRQ 794-aa fragment was markedly higher than that of s-FRQ, suggesting the l-FRQ N-terminal 99-aa region may regulate the phosphorylation of the entire FRQ protein. Quantitative label-free LC/MS analysis identified several peptides that were differentially phosphorylated between l-FRQ and s-FRQ, which were distributed in FRQ in an interlaced fashion. Furthermore, we identified two novel phosphorylation sites, S765 and T781; mutations S765A and T781A showed no significant effects on conidiation rhythmicity, although T781 conferred FRQ stability. These findings demonstrate that FRQ isoforms play differential roles in the circadian negative feedback loop and undergo different regulations of phosphorylation, structure, and stability. The l-FRQ N-terminal 99-aa region plays an important role in regulating the phosphorylation, stability, conformation, and function of the FRQ protein. As the FRQ circadian clock counterparts in other species also have isoforms or paralogues, these findings will also further our understanding of the underlying regulatory mechanisms of the circadian clock in other organisms based on the high conservation of circadian clocks in eukaryotes.

Neurospora crassa is an important model organism for circadian clock research. The Neurospora core circadian component FRQ protein has two isoforms, large FRQ (l-FRQ) and small FRQ (s-FRQ), of which l-FRQ bears an additional Nterminal 99-amino acid fragment. However, how the FRQ isoforms operate differentially in regulating the circadian clock remains elusive. Here, we show l-FRQ and s-FRQ play different roles in regulating the circadian negative feedback loop. Compared to s-FRQ, l-FRQ is less stable and undergoes hypophosphorylation and faster degradation. The phosphorylation of the C-terminal l-FRQ 794-aa fragment was markedly higher than that of s-FRQ, suggesting the l-FRQ N-terminal 99-aa region may regulate the phosphorylation of the entire FRQ protein. Quantitative label-free LC/MS analysis identified several peptides that were differentially phosphorylated between l-FRQ and s-FRQ, which were distributed in FRQ in an interlaced fashion. Furthermore, we identified two novel phosphorylation sites, S765 and T781; mutations S765A and T781A showed no significant effects on conidiation rhythmicity, although T781 conferred FRQ stability. These findings demonstrate that FRQ isoforms play differential roles in the circadian negative feedback loop and undergo different regulations of phosphorylation, structure, and stability. The l-FRQ N-terminal 99-aa region plays an important role in regulating the phosphorylation, stability, conformation, and function of the FRQ protein. As the FRQ circadian clock counterparts in other species also have isoforms or paralogues, these findings will also further our understanding of the underlying regulatory mechanisms of the circadian clock in other organisms based on the high conservation of circadian clocks in eukaryotes.
Circadian clocks enable organisms to predict and adjust their physiology and behavior to the daily changing environment (1). It has been demonstrated that misalignment in circadian rhythms leads to maladaptation to the daily alternating environment in many species (2). In humans, the circadian clock is considered one of the most important hallmarks of health, and circadian misalignment causes a variety of diseases (3).
The circadian clock is finely regulated at the molecular level. The filamentous fungus Neurospora crassa is a classical model for circadian research, which has substantially contributed to the understanding of circadian systems in other organisms (4). In Neurospora, White Collar 1 (WC-1) and WC-2 are the two positive elements that form the White Collar Complex (WCC) to activate the transcription of the frequency (frq) gene; and the latter encodes the FRQ protein which is the negative element (5,6). FRQ is the negative component in the Neurospora circadian clock, which represses the function of WCC as transcription factors in the negative feedback loop and supports WCC levels in an interlocked positive limb (7)(8)(9). FRQ proteins form dimers via the coiled-coil domain that is critical for its circadian clock function (10). FRQ also recruits casein kinases to regulate the phosphorylation of WCC proteins (11). Hyperphosphorylated WCC proteins are stable and inactive in transcription (12)(13)(14). Phosphatase 2A dephosphorylates and reactivates WCC by countering the function of casein kinases (11)(12)(13)(14)(15).
Dynamic phosphorylation of circadian clock proteins is highly conserved in eukaryotes (16). In Neurospora, 101 in vivo/ in vitro phosphorylation sites have been identified in FRQ, and it is very likely that FRQ may harbor even more phosphorylation sites (17,18). The phosphorylation of FRQ is dynamically controlled by a number of protein kinases including casein kinas 1 (CK1), CK2, checkpoint kinase 2, calcium/calmodulindependent kinase-1, Protein kinase A (PKA), in which FRQbound CK1a plays multiple roles in the circadian negative feedback loop (17)(18)(19). The degradation rate of FRQ is primarily mediated by the affinity of FRQ-CK1 interaction, which roughly correlates with the circadian period (20).
The open reading frame (ORF) of frq contains three initiation codons, and the second codon is rarely used (21). Translation from the first initiation codon results in the ‡ These authors contributed equally to this work. synthesis of a protein product with 989 amino acid residues (aa), which is called large FRQ (l-FRQ). Frq also possesses up to six introns that undergo complex control of alternative splicing, producing eight splicing variants in total (21)(22)(23)(24). Removal of the sixth intron (I-6) in frq pre-mRNA by alternative splicing eliminates the translation initiation from the first codon, which results in the synthesis of a protein isoform lacking the N-terminal 99 aa (N-99), which is called small FRQ (s-FRQ) (22,23). Several splicing factors or regulators, for example, PRP5, PRMT5, and the snRNA U4-2, U5, are implicated in the regulation of frq splicing and/or circadian rhythms. Complexes for mRNA processing and surveillance, including the exosome and nonsense-mediated RNA decay machinery, have been demonstrated to regulate the splicing and decay of transcript variants (25)(26)(27). The frq gene has several untranslated ORFs in its 5 0 untranslated region, which regulates the temperature sensitivity during FRQ translation (21)(22)(23).
The synthesis ratio of l-FRQ/s-FRQ changes according to the ambient temperature. Each of the FRQ isoforms can independently sustain rhythmicity at certain temperatures although synergetic expression of both is critical for fine tuning of clock (21,23,28). The strain exclusively expressing l-FRQ (l-frq strain) displays a shorter period while the strain expressing only s-FRQ (s-frq strain) displays a longer period, in comparison to the wild type, respectively. S-FRQ loses its function at high temperatures, while l-FRQ becomes dysfunctional at low temperatures (21), suggesting that both s-FRQ and l-FRQ are required for clock and carry out differential circadian functions. A significantly higher proportion of s-FRQ is localized in the nucleus than l-FRQ (29), also suggesting the differences in phosphorylation and function between s-FRQ and l-FRQ. In addition, l-FRQ is more effective in promoting the day/night growth ratio (26). However, little is known about how FRQ isoforms operate differentially in regulating the circadian clock.
In this work, we demonstrate the different roles of l-FRQ and s-FRQ and the underlying mechanisms in the regulation of the circadian clock. The N-99 aa fragment of l-FRQ plays an important role in controlling the phosphorylation, stability, conformation, and function of FRQ. In addition, we identified two novel FRQ phosphorylation sites and characterized their effects on circadian rhythms.

l-FRQ and s-FRQ act differently in circadian feedback loops
Both splicing and ambient temperature contribute to the production of s-FRQ and l-FRQ (22,23,27). To validate whether alternative splicing directly regulates the ratio of FRQ isoforms, we treated the mdr3 KO strain with spliceostatin A (SSA), an inhibitor of spliceosome assembly (30). Multiple drug resistance 3 (MDR3) is a protein responsible for drug metabolism (31); therefore, the lack of MDR3 in the mdr3 KO strain could prevent its potential resistance against SSA. The results from dephosphorylation experiments showed that the samples in the presence of SSA showed a lower proportion of s-FRQ than the untreated samples in the mdr3 KO strain (Fig. S1A), indicating that suppression of the splicing of frq I-6 directly affects the production of the synthesis ratio of FRQ isoforms.
WC-2 binds to WC-1 to form the WCC complex that binds to the promoters of frq and other clock-controlled genes in a circadian fashion, and FRQ represses the function of WCC in a negative feedback loop (5,7). To compare the functional difference between FRQ isoforms in the circadian negative feedback loop, we conducted chromatin immunoprecipitation (ChIP) assays with WC-2 antibody to assess the binding efficiency of WCC to the frq promoter in the l-frq and s-frq strains at 20 C and 27 C. In the results at 20 C, WC-2 showed the highest binding to frq promoter at DD14 in the s-frq strain and DD22 in the l-frq strain; WC-2 showed the highest binding to frq promoter at DD14 in both s-frq and l-frq strains but lower binding at LL, DD18, and DD22 in l-frq strain at 27 C (Fig. 1A). These data support that l-FRQ and s-FRQ may function differentially to sustain circadian rhythmicity in the negative feedback loop in a temperature-sensitive manner (21).
As the ratio of FRQ isoforms differs at high and low temperatures, changed s-FRQ/l-FRQ ratios may affect the levels of WCC proteins (21,23,28). We next performed Western blot analysis to examine the levels of the clock proteins FRQ, WC-1 and WC-2 in l-frq, s-frq, and WT strains at three temperatures, which were chosen according to the previous reports (21,23,28). The results showed that with increasing temperature, the levels of both WC-2 and FRQ were significantly induced in the l-frq strain. By contrast, the change in FRQ and WC-2 was marginal in the s-frq strain. In kaj120, a rescued strain expressing both s-FRQ and l-FRQ in the background of the frq knockout strain (frq 10 ), the levels of WC-2 and FRQ were between those of s-frq and l-frq, reflecting a mixed effect. By contrast, the change in WC-1 levels was comparable in all three strains (Fig. 1, B and C), which is in agreement with previous reports (21,23). However, when considering the levels of FRQ in different strains, it is noteworthy that the amount of WC-1 but not WC-2 in the s-frq strain was much higher than that in the other two strains at 28 C (Fig. 1D).
Besides its role in the circadian negative feedback loop, FRQ also supports the transcription of wc-2 and the accumulation of WC-1 protein through different positive limbs (8,9,32). To address the different expressions of WC-1 and WC-2, we checked the mRNA levels of wc-1 and wc-2 which showed comparable expression at both 20 C and 28 C (Fig. S1B). WC-2 protein showed a faster degradation rate at 28 C than that at 20 C, while the degradation rates of WC-1 or WC-2 were comparable in s-frq or l-frq strains at both 20 C and 28 C (Fig. S1, C-F). Therefore, together, these facts suggest that the increased degradation of WC-2 at 28 C in s-frq may account for the difference in WC-2 expression from that in lfrq.
In addition to the repression of WCC function in promoting transcription, FRQ also facilitates the phosphorylation and accumulation of WCC which constitutes a positive loop in the circadian clock negative feedback circuits, which leads to reduced WCC turnover. Therefore, we measured the levels of WCC proteins as this way the function of FRQ could be determined (8,14). And the results suggest that s-FRQ is more efficient in promoting the level of WC-1 at higher temperatures.
FRQ binds to the WCC complex to repress its own transcription, which closes the negative feedback loop. Next, we conducted co-immunoprecipitation (CO-IP) with WC-2 antiserum in the l-frq and s-frq strains to assess the association between WCC and s-FRQ and l-FRQ at 20 C and 30 C, respectively. The CO-IP results showed that s-FRQ bound to more WCC than l-FRQ at both temperatures ( Fig. 1, E and F), which suggests that s-FRQ exhibits higher binding to WCC in the negative feedback loop.
The FRQ-CK1a association determines the prevalent phosphorylation of FRQ and the circadian period (20,33). Compared to l-FRQ, the ratio of s-FRQ bound to CK1a was significantly lower than l-FRQ and wild type (Fig. 1G), suggesting that the phosphorylation of FRQ isoforms is differentially regulated by CK1a. WCC proteins undergo phosphorylation and dephosphorylation by related kinases and phosphatases, and the WCC phosphorylation levels determine their turnover (19). We conducted λ phosphatase treatment and Western blot analysis to investigate the expression, and surprisingly, phosphorylation of WCC proteins in the l-frq and s-frq strains and found that both WC-1 and WC-2 were hyperphosphorylated in s-frq compared to that in l-frq (Fig. 1H).
l-FRQ shows faster phosphorylation, turnover, and it possesses a looser structure compared to s-FRQ Phosphorylation and stability are two crucial aspects of FRQ protein, which are closely related to its function (17,34). However, the difference in phosphorylation, structure, and degradation between s-FRQ and l-FRQ remains elusive. To this end, we assessed the difference in the degradation rates between l-FRQ and s-FRQ by cycloheximide (CHX) treatment. CHX is a translation inhibitor that allows us to measure the turnover rate of proteins of interest. The Western blotting results showed that the degradation of l-FRQ was faster than s-FRQ at all of the three tested temperatures (20 C, 25 C, and 30 C) (Fig. 2, A-C). To exclude the possibility of artificial effects caused by CHX, we also examined the degradation rates of l-FRQ and s-FRQ by LD transition experiments. FRQ protein maintains a relatively high level in constant light and undergoes gradual degradation in the first several hours after Characterization of differences in Neurospora FRQ isoforms transition into darkness, and the decrease in l-FRQ level was faster at 20 C and 25 C (Fig. 2, D-F).
We also compared the changes in phosphorylation levels of l-FRQ and s-FRQ during their degradation. The CHX-treated protein samples of l-frq were loaded for electrophoresis for approximately 10 min, and then the CHX-treated protein samples of s-frq were loaded on the same gel, to allow the samples of s-frq and l-frq to run at comparable levels for better comparison of their phosphorylation. The phosphorylation of l-FRQ reached its highest level after 3 h while the phosphorylation of s-FRQ was significantly delayed (Fig. S2, A-C). These results demonstrate the differential phosphorylation and turnover between the FRQ isoforms.
As the entire sequence of s-FRQ is identical to the C-terminus of l-FRQ (100 aa-989 aa), the higher phosphorylation level of l-FRQ might be attributed to higher phosphorylation of the 890 aa C-terminus. Another possibility is that the l-FRQ N-99 fragment itself but not the rest part is more phosphorylated in l-FRQ. A third possibility is the combination of both.
Sequence analysis indicates that both l-FRQ and s-FRQ harbor one common thrombin cleavage site, which is located between R195 and R196 in l-FRQ and between R96 and R97 in s-FRQ (Fig. 2G). The constructs expressing l-FRQ-c-Myc-His and s-FRQ-c-Myc-His bearing were transformed into the frq 10 strain, respectively, and the resultant transformants frq10, sfrq-c-myc-his, and frq10, l-frq-c-myc-his were obtained. The l-FRQ-c-Myc-His and s-FRQ-c-Myc-His proteins were purified and subjected to thrombin cleavage and λ-phosphatase for dephosphorylation. The migration of cleaved products was obviously slower in l-FRQ-c-Myc-His than that in s-FRQ-c-Myc-His (Fig. 2H), suggesting that at least the C-terminus of FRQ is more hyperphosphorylated in l-FRQ than that in s-FRQ and the N-99 aa fragment of l-FRQ regulates the phosphorylation of FRQ C-terminus. Phosphorylation and degradation of FRQ are tightly linked to its structure (17,34). FRQ is very likely an unstructured protein, and its phosphorylation is very extensive and dynamic (35,36), which complicates the analysis of its structure. To assess the difference in the structure of s-FRQ and l-FRQ, we performed freeze-thaw and limited trypsin digestion experiments, respectively. After treatment for up to 12 freeze-thaw cycles, l-FRQ levels decreased more rapidly than s-FRQ (Fig. 3A). Similarly, the results from partial trypsin digestion showed that l-FRQ decreased more rapidly than s-FRQ (Fig. 3B). Together, these results indicate that l-FRQ might possess a looser structure that is more subject to phosphorylation and degradation, and the differential conformation of FRQ isoforms might result from the N-99 aa region itself and/ or the effects of N-99 aa region on the rest parts of FRQ.

Mapping and analysis of the critical region in the l-FRQ Nterminus
To map the minimal region involved in the determination of the function of the N-99 region, we generated a series of constructs with deletions in the N-99 region, which were transformed into frq 10 (Fig. 4, A and B). The results of the race tube assay showed that l-FRQΔ(71-77) and l-FRQΔ(78-99) displayed arrhythmic phenotypes while other strains displayed rhythms of conidial zonation. The period of l-FRQΔ(2-33) was shorter while l-FRQΔ(34-55) was longer than l-frq (Fig. 4C).
We next introduced a luciferase reporter construct that is under the control of the frq promoter into the l-FRQΔ(71-77) and l-FRQΔ(78-99) strains and found that, consistently, robust rhythmic luciferase activity was abolished in these two strains (Fig. 4D). These data suggest that the aa 71 to 99 region is required for the function of the N-99 region, and deletion of it may compromise the function of the entire FRQ protein. In contrast, the strains harboring mutations in this region, including the strain with mutation of S73A and the M2 strain with mutations of S72A, S73A and S76A, displayed conidiation rhythmicity (17,18). In these strains, functional s-FRQ might restore circadian rhythmicity as the third initiation codon remains intact.
We measured the expression and phosphorylation of clock proteins in the deletion strains, and the Western blotting results showed that deletion of l-FRQ(71-77) led to a dramatic decrease in the level of hyperphosphorylated FRQ proteins compared to other deletions (Fig. 5A). Deletion of these FRQ regions at its N terminus led to different influences on the phosphorylation of FRQ and WC-1. FRQ displayed overt hypophosphorylation in the strain l-FRQΔ(71-77) compared to other deletion strains; the WC-1 phosphorylation levels in l-FRQΔ(34-55) and l-FRQΔ(71-77) strains were higher than those in the strains of l-FRQΔ(2-33), l-FRQΔ(55-70) and l-FRQΔ(78-99). These data suggest that these regions differentially affect FRQ phosphorylation and contribute to the positive limb of the circadian clock (Fig. 5A). The turnover of l-FRQΔ(71-77) and l-FRQΔ(78-99) was significantly slower than that of l-FRQ (Fig. 5, B and C), suggesting that although these deletion strains cannot precisely mimic the phenotype of the s-frq strain, the 71 to 99 aa region plays a critical role in regulating FRQ function and circadian rhythms.

Comparison of phosphorylated peptides between l-FRQ and s-FRQ
As the phosphorylation level is higher in l-FRQ than that in s-FRQ, we conducted label-free LC-MS to quantitatively detect the phosphorylation sites in l/s-FRQ proteins. L-FRQ-c-Myc-His and s-FRQ-c-Myc-His proteins were purified from frq 10 , l-frq-c-myc-his, and frq10, s-frq-c-myc-his strains which were subsequently subjected to quantitative label-free LC-MS analysis (Fig. 6, A-C).
The results from label-free quantitative LC/MS analysis revealed 19 peptides with differential phosphorylation patterns between s-FRQ and l-FRQ. The preferential phosphorylation at the sites in the N-99 aa region supports the validity of the present analysis. In addition, phosphorylation occurred in s-FRQ and l-FRQ in an interlaced pattern, in which even some neighboring sites showed opposite preferential phosphorylation ( Fig. 6D and Table S1). In addition, eight phosphorylation sites were identified, in which S28 and S32 are located in l-FRQ but not s-FRQ and S286 was prevalently phosphorylated in l-FRQ. Two novel sites (S765 and T781) were also identified and the phosphorylation of which was present in both s-FRQ and l-FRQ (Fig. 6, D and E).

Identification and characterization of two novel phosphorylation sites on FRQ
Previous studies have revealed that FRQ harbors up to 101 phosphorylation sites, which change in a circadian fashion (17,18). In this work, the label-free quantitative LC/MS analysis identified two novel phosphorylation sites in FRQ, S765, and T781 (Figs. 6, D and E and 7A) which showed no significant difference in phosphorylation between s-FRQ and l-FRQ. Both of these sites are located in the FRQ6 region, and especially, T781 is located in the FRQ6B region which is critical for FRQ-FRH interaction and circadian rhythmicity (37)(38)(39).
We generated mutants harboring S765A (l-FRQ S765A ) and T781A (l-FRQ T781A ) in the l-frq strain, respectively. In addition, we also generated mutants of T782A (l-FRQ T782A ) and T781A,T782A (l-FRQ T781A,T782A ) although phosphorylation of T782 was not found in this work or in previous studies (17,18). In race tube assay, the strains of l-FRQ S765A and l-FRQ T781A showed comparable conidiation periods to that of lfrq strain, a comparable conidiation period to l-frq (Fig. 7B). L-FRQ T781A but not l-FRQ S765A showed lower phosphorylation levels under constant light compared to that of l-FRQ (Figs. 7C  and S3, A and B). Co-IP was conducted to measure the association between mutated FRQ and FRH, and the results showed that mutations of the three sites, S765A, T781A, and T782A, did not abolish the FRQ-FRH interaction (Fig. 7D). In most cases, a higher phosphorylation level is correlated with a decreased stability of FRQ (18), and consistently, the Western blotting results revealed that the T781A mutation stabilized FRQ while S765A showed no significant effects on FRQ turnover (Fig. 7E). The two mutations on s-FRQ, s-FRQ S666A and s-FRQ T682A which are counterparts to l-FRQ S765A and l-FRQ T781A , respectively, showed no overt effects on conidiation rhythmicity, and mutation of s-FRQ T682A but not s-FRQ S666A caused a slower degradation rate compared to the s-frq strain (Fig. S4, A-C). The functional significances of phosphorylation on l-FRQ T781 and s-FRQ T682 remain to be further investigated.

Discussion
The alternative splicing of frq I-6 results in the production of two FRQ isoforms, l-FRQ and s-FRQ (21)(22)(23). In this work, we show that s-FRQ more effectively promotes WC-1 expression at high temperatures (Fig. 1, B-D) and demonstrate that l-FRQ is more effective in recruiting CK1 which leads to its higher phosphorylation (Figs. 1H and 2H). L-FRQ shows faster phosphorylation and degradation than s-FRQ (Figs. 2, A-H and S2, A-C). All of these findings support that FRQ isoforms play differential roles in regulating Neurospora circadian rhythms.
The phosphorylation and degradation of FRQ are crucial for clock functions (40)(41)(42)(43). Usually, accelerated phosphorylation of FRQ which is determined by the FRQ-CK1 interaction tends to decrease its stability and leads to the shortening of circadian periods (5,20,40). The N-99 region of l-FRQ contains several conserved and dynamically phosphorylated sites, which are phosphorylated in the late time of a circadian day (17,18). In l-FRQ, the N-terminal specific region has been proposed to promote the degradation of the entire protein, and extensive mutations of phosphorylation sites in the FRQ N-terminus led to decreased phosphorylation and increased stability in clock proteins, suggesting that phosphorylation of the N-terminus might be necessary for regulating FRQ turnover (17,36). In this work, the thrombin cleavage results directly demonstrated that both of the N-99 region and the rest part of l-FRQ contribute to its different phosphorylation pattern from that of s-FRQ (Figs. 2H and 6D), which might contribute to the differential turnover, conformation and function between FRQ isoforms. Bake et al. reported that FRQ mutation S73 led to an increased period length (1.8 h-2.3 h) in the WT strain (18), which also suggests that in the N-99 region, this site may importantly contribute to the period difference between s-frq and l-frq strains.
Mutations of most phosphorylation sites in the middle part of FRQ lead to longer periods (17,18). We identified that the phosphorylation pattern in some regions that contain multiple potential phosphorylation sites was different between s-FRQ and l-FRQ (Fig. 6D), which might be attributed to differential phosphorylation of the same sites or phosphorylation of different sites in the same regions. This interlaced phosphorylation pattern suggests that some of these phosphorylated regions may have distinct functions in controlling the circadian clock, and the N-terminal region of l-FRQ may play an important role in modulating phosphorylation, stability, conformation, and accessibility to kinases of the FRQ protein. Moreover, we identified two novel phosphorylation sites in the FRQ protein, S765 and T781 (Fig. 7A). Based on this report and previous reports (17,18), 103 in vivo and in vitro phosphorylation sites have been found in FRQ in total (Fig. 6E). The very high abundance of phosphorylation sites indicates that FRQ has a very intricate structure that undergoes complex regulation. The degradation rate of FRQ is regulated by CK1, which is negatively associated with Neurospora circadian period. Nonetheless, there are some exceptions, for example, in the strain lacking the F-box protein FWD-1, an E3 ubiquitin ligase, FRQ oscillates with a period of 24 h despite its severely compromised turnover (44). Mutations of S781A on l-FRQ and S682A on s-FRQ showed no significant effects on the conidial period despite their confers decreased turnover rates of FRQ protein (Figs. 7, B and E and S4, A-C), which supports that the FRQ stability does not always correlate with the circadian period (44). FRQ-bound CK1a modulates the phosphorylation of FRQ and WCC; similarly, mammalian PER acts as a scaffold of CK1 to phosphorylate itself and CLOCK (19,45,46). The frq7 and frq1 strains display a longer period and a short period, respectively (47). Liu et al. (20) found that the binding of CK1a to FRQ7 was weaker than FRQ1, suggesting that the association affinity between FRQ and CK1a determines the circadian period length. Consistently, in this work, the l-frq strain with a shorter period showed a stronger CK1a-FRQ association compared to s-frq strain (Fig. 1G), and consistently, l-FRQ displayed higher phosphorylation and faster turnover rate (Figs. 2, A-F and H and S2, A-C). These data provide more evidence that the recruitment of CK1 by the circadian negative components acts in the positive and negative feedback loops.
The association between FRQ and FRH plays multiple important roles in regulating FRQ phosphorylation, stability, interaction with other partners, and clock function (5,(48)(49)(50). FRQ6B2 and FRQ6B5 are two subdomains located in FRQ6B, the deletion of which abolishes the FRQ-FRH interaction (37,49). T781 and T782 are the only two threonine sites in FRQ6B2 and FRQ6B5 and mutations of these two sites showed no overt effects on the FRQ-FRH interaction (Fig. 7D), suggesting that the formation of the FRQ-FRH complex is independent of the phosphorylation status of FRQ.
FRQ is an intrinsically disordered protein (35,36), and it has been proposed that FRQ might exist as a non-globule-shaped molecule (18). In Drosophila, phosphorylation of PER leads to a more open conformation than the hypophosphorylated isoforms, and phosphorylation decreases the stability of PER (51). Hyperphosphorylated FRQ possesses a looser conformation than hypophosphorylated species (36). Consistently, in this work, we revealed that l-FRQ might have a structure looser than s-FRQ (Fig. 3, A and B). We also observed lower phosphorylation of WCC proteins in the l-frq strain at 22 C, 25 C, and 28 C (Fig. 1H), which cannot be explained by the higher l-FRQ-CK1 affinity (Fig. 1G). It is possible that although l-FRQ shows higher binding affinity with CK1, it may be less efficient to promote the phosphorylation of WCC due to its looser conformation. Alternatively, the phosphorylation difference in WCC proteins between s-FRQ and l-FRQ may be controlled by other kinase(s) rather than CK1. These issues remain to be investigated.
The extensive phosphorylation of FRQ occurs in a dynamic and circadian fashion, in which phosphorylation of the Nterminus of l-FRQ occurs late in the circadian cycle (17). The N-terminal portion of FRQ is positively charged while the Cterminal portion is negatively charged, and it has been proposed that the N and C parts of FRQ bind together to form an intramolecular hinge-like structure (36). However, this hingelike structure fails to explain the late phosphorylation phase of the l-FRQ N terminus (51). Since FRQ proteins form dimers via their coiled-coil (CC) domains, which are critical for maintaining WCC expression and circadian rhythmicity (10), an FRQ dimer might be composed of two hinge-like FRQ proteins bound together via their CC domains (Fig. 7F). A looser conformation in the middle part of l-FRQ would facilitate the accessibility of CK1a with FRQ-CK1a interaction domains (FCDs) which might explain the late phase of phosphorylation of the N terminus (17). In contrast, the intermolecular binding between the FRQ dimer via the adjacent CC domains might be so compact that it probably hinders the accessibility of kinases and phosphorylation as consequence, which may explain the late phosphorylation phase of the N terminus of l-FRQ (Fig. 7F).
Synchronization is critical for the circadian clock. In the WT strain, FRQ proteins might exist as three forms of dimers: l-FRQ/l-FRQ, s-FRQ/s-FRQ, and s-FRQ/l-FRQ, which implies that even in one single cell, the regulation of each set of clock machinery may not be homologous. In Drosophila and mammals, PER and CRY/TIM proteins are the counterparts of FRQ in the circadian negative feedback loop, and PER and CRY contain paralogues in mammals (4,52,53). It will be interesting to investigate how the pool of different clock protein isoform dimers acts in concert to generate molecular rhythms.

Conclusion
In this study, we provided further evidence supporting the differential regulation of phosphorylation, structure, and stability of FRQ isoforms which control the Neurospora circadian clock; the specific N-99 region plays a critical role in regulating the post-translational modification and function of l-FRQ. In addition, we identified the differentially phosphorylated sites on FRQ isoforms and two novel phosphorylation sites. The regulatory mechanisms of circadian clocks are highly conserved in eukaryotes. These findings will also further our understanding of the underlying regulatory mechanisms of the circadian clock in other organisms. In the future, resolving the structure of FRQ and its dimers probably through Cryo-EM will shed light on the in-depth understanding of the differences in post-translational modification, conformation, and function between FRQ isoforms.

Experimental procedures
Neurospora strains and culture 303-3 (bd, his3, frq 10 ), the frq null strain, is the host strain for all FRQ mutant constructs. Pkaj120 plasmids bearing different deletions were transformed into the frq 10 host strain at the his-3 locus. L-frq and s-frq strains were generated by transforming plasmids harboring mutations at the first and the third initiation codons into frq 10 , respectively (21). Pkaj120 is an frq null strain transformed wild-type kaj120 plasmid as a control strain (18). The 301-5 (bd) strain was also used as wildtype control in indicated experiments (26). The mdr3 KO strain was purchased from the Fungal Genetics Stock Center (FGSC13545). To construct the pkaj120 plasmids comprising different mutations or deletions, the mutations were first generated in partial frq sequence cloned in puc19 vector by site-directed mutagenesis strategy through overlap extension using PCR with primers harboring respective mutations. Next, the fragments were subcloned into pkaj120 through specific restriction enzyme sites.

Race tube assay
A race tube is a long glass tube containing a layer of solid media inside, used for the determination of the circadian period of Neurospora in constant darkness. In race tube assay, Neurospora is inoculated inside the race tube which allows it to grow in one-way. During growth, Neurospora releases asexual conidia in a circadian fashion, and the circadian period can be calculated according to the interval time between the conidiation bands (54). The medium for race tube assays contained 1× Vogel's medium, 2% glucose, 50 ng of biotin/ml, and 1.5% agar. After transfer into a dark room, the growth fronts were marked on the race tubes at certain time points every 24 h.

Protein analysis and ChIP assay
Protein extraction, quantification, Western blot analysis, and immunoprecipitation assays were performed as described previously (49). For analysis of the phosphorylation of WCC proteins, specific SDS-PAGE gel (acrylamide-bisacrylamide 149:1) was prepared (13). Transfer of protein from gel to the membrane in Western blot was conducted under 120 mA for 2.5 h using the Mini-PROTEAN Tetra (Bio-Rad) apparatus. ChIP assays were performed as described previously (55). Immunoprecipitation was performed with a WC-2 antibody. The following primers were used in the ChIP assay: forward, 5 0 -tgtccaagcgggaagctggagt-3 0 ; reverse, 5 0 -ccacgcttagggtaagtaactg-3 0 .

Freeze-thaw experiments and incomplete trypsin digestion analysis
To perform protein freeze-thaw assays, the protein extracts were diluted to a concentration of 5 μg/μl, then frozen in liquid nitrogen, and thawed in a water bath at room temperature, for a specific number of cycles. The trypsin digestion assay was performed as described previously (36,56). Briefly, 100 μl of the protein extracts (2.5 μg/μl) was treated with 50 ng trypsin (final concentration 0.5 μg/ml) at 25 C, and the aliquots of 20 μl were taken out after trypsin addition at 0, 5, 15, 30 min. The samples were subjected to Western blot analysis.

Protein purification and thrombin cleavage assay
The thrombin cleavage site of FRQ was predicted by the Peptidecutter online tool (https://web.expasy.org/peptide_ cutter/) (57). L-FRQ-c-Myc-His and s-FRQ-c-Myc-His were expressed and enriched with anti-c-Myc beads (Sigma), which were further cleaved by thrombin (final concentration: 10 ng/ μl) at 4 C for 5 h. Subsequently, the samples were treated with λ-PPase (final concentration: 20 U/μl) by incubating at 30 C for 30 min. The cleaved protein was subject to Western blot analysis using FRQ antibody.
Protein purification and label-free quantitative LC/MS analysis of s/l-FRQ Label-free quantitative LC/MS analysis was used to identify FRQ peptides with differential phosphorylation patterns (58). The label-free LC/MS experiments were conducted by Novogene Co, LTD.

Statistical analysis
Data are the mean values ± SD or mean values ± SE as indicated. Experimental data were analyzed using Student t test in Microsoft Excel software. A p-value of <0.05 was considered significant. *p < 0.05, **p < 0.01, # p < 0.001.

Data availability
All data in this study are available within the article, supporting information, and/or from the corresponding author on reasonable request.
Supporting information-This article contains supporting information.