A daily cycle of White Collar Complex dephosphorylation sustains circadian rhythmicity in Neurospora

Significance

At the core of circadian clocks of fungi and animals, a protein heterodimer drives expression of gene(s) whose products inactivate the heterodimer via phosphorylation. To sustain the cycle, the heterodimer activity must be restored, but how this happens has been unclear. In Neurospora, the White Collar Complex (WCC) is the heterodimer and the FRQ–FRH complex (FFC) inactivates it. We determined that WCC activity is restored principally by removal of the inhibitory phosphorylations and that for at least several days no new synthesis of WCC is required. Cells hold a large cellular pool of WCC delicately balanced between FFC-dependent phosphorylation/inactivation and phosphatase-dependent dephosphorylation/reactivation. Each morning, a fraction of the inactive WCC pool is activated, thereby restarting the circadian cycle.

Abstract

The transcription factor complex White Collar Complex (WCC) functions both as a photoreceptor and as the circadian positive element. In response to light, WCC acutely activates ~5% of all genes, whereas in the dark it influences expression of about 40% of the transcriptome. Among WCC targets is frq, which is acutely light-activated through the pLRE (proximal Light-Response Element) and circadian-regulated through the C-box (Clock-box) promoter element that is not responsible for light-driven expression. The FRQ–FRH complex (FFC), which includes CK-1a, represses WCC activity at the C-box by phosphorylating WCC at >95 sites, but FFC has no described role in the light. We validated the expectation that FFC also silences C-box promoters in constant light, thereby confirming two classes of WCC targets: C-box-like genes that are normally repressed in light and pLRE-like genes that remain light-active despite FFC-driven WCC phosphorylation. Derepression of C-box-like promoters in frq-null fungi may explain reported noncircadian phenotypes such as reduced virulence and conidiation. Reanalysis of WCC circadian regulation revealed that, while most WCC is phosphorylated and repressed at dusk, subsequent circadian activation results from transient dephosphorylation of only a small subset of the WCC pool. This small active pool drives frq expression, nucleating the FFC, which rephosphorylates WCC to repress it again, generating a phosphorylation/dephosphorylation cycle that can persist for days without new WCC synthesis. The realization that both FFC and WCC are regulated primarily through phosphorylation rather than protein turnover leaves the circadian oscillator looking much like a “phoscillator,” emphasizing the primacy of posttranslational regulation in timekeeping.

Circadian clocks regulate a wide variety of behavioral, physiological, cellular, and molecular events (1). Dysfunction of biological rhythms has been linked to various human diseases (2, 3). The core oscillators in fungi and animals are intracellular, constructed as autoregulatory transcriptional–translational negative feedback loops wherein a heterodimer of positive elements drives expression of negative elements, and the latter repress transcriptional activity of the former to ultimately impinge on their own expression (4–6). For instance, in mammalian cells, transcription factors CLOCK and BMAL1 heterodimerize via Per–ARNT–Sim (PAS) domains and drive expression of negative elements, PERs and CRYs, as well as elements in ancillary loops that contribute to robustness (5, 7). In Neurospora transcription factors White Collar 1 (WC-1) and WC-2 heterodimerize via PAS domains to form the White Collar Complex (WCC), which drives expression of the core negative regulator FRQ under contrasting light conditions: WCC either binds to the proximal Light-Response Element (pLRE) as a dimer (two copies each of WC-1 and WC-2, with the interaction mediated through the LOV domains of WC-1) to mediate frq-induction in the light (8–10), or binds rhythmically to the Clock-box (C-box) as a monomer (one copy each of WC-1 and WC-2) in the dark, driving the cyclical expression of frq required for rhythmicity (11–14). FRQ interacts with FRH (FRQ-Interacting RNA Helicase) (15–17) and CK-1a (casein kinase 1) (13, 18, 19) to form the FFC (FRQ–FRH complex), which in turn represses circadian (but not light-driven) activity of WCC to terminate its own expression, thereby closing the circadian feedback loop (4, 7, 18).

Multisite phosphorylation has been found to be a central mechanism in tightly controlling and fine-tuning circadian functions of both the positive and negative elements in these loops (7). Regulation of the negative elements in animals and Neurospora has received the most attention as their time-of-day-specific phosphorylation (18, 20–23) leads directly to their loss of activity, although there is debate about whether phosphorylation first leads to inactivation followed by turnover as has been shown in Neurospora (24, 25), or if inactivation is caused by phosphorylation-elicited turnover as is believed for mammalian clocks (5). FRQ, for instance, undergoes myriad temporal phosphorylations by CK-1a, CK2, and additional kinases with over 100 time-of-day-specific phosphorylations, tightly tuning its activity, dynamically determining its binding partners (26, 27), and eventually resulting in its inactivation (18, 22, 23) which happens prior to its turnover (24). Positive element heterodimers in both fungi and mammals are also subject to time-of-day-specific phosphorylation leading to repression by reducing their ability to bind to DNA (28–34) [Displacement Repression (32)]. In Neurospora > 80 phosphosites have been identified on WC-1 and 15 on WC-2 (31). Circadian repression requires phosphorylation of distinct clusters of these residues on each protein, and the time-of-day-specific phosphocode governing rhythmic repression has been elucidated (31). Very recent comparable data from mammalian cells have identified sites on both BMAL1 and CLOCK that must be phosphorylated to achieve repression (35), a result congruent with that seen in Neurospora thus establishing a conserved precedent for repression.

Aside from effects of kinases and phosphorylation, few studies have assessed dosage requirements for circadian positive elements or the role(s) of phosphatases in influencing this. In Neurospora, period length is little affected in strains constitutively or overexpressing WC-1, WC-2, or both together (15, 36), and likewise mammalian cells expressing various constitutive levels of BMAL1 or CLOCK are strongly rhythmic with periods similar to WT (34, 35). Although rhythms in positive element expression are seen both in Neurospora WCC (37–39) and in mammalian BMAL1 (e.g., ref. 30), the rhythms are not viewed as essential but instead as contributing to robustness of the cycle (e.g., ref. 40), in all suggesting that while the circadian heterodimer (WCC or BMAL1/CLOCK) is necessary for rhythmicity, neither its rhythmic expression nor abundance acts as a causal element in determining the pace of the core clock. This, however, poses a question: If the cycle of positive element activity required for the clock depends partly (as in mammals) or wholly (as in Neurospora) on a cycle of phosphorylation-mediated repression, how can the cycle continue unimpeded in the presence of a steady influx of new, unmodified, and nonrepressed heterodimer (e.g., ref. 15)? Such an influx of active positive elements ought to short-circuit the feedback loop leading to constant negative element expression and loss of rhythmicity.

Phosphatases that can reverse the effects of critical clock kinases such as CK1 and CK2 have been suggested as likely clock-relevant proteins and could address this paradox in fungi and mammals, but the phenotypes of mutants are not as unambiguous as could be hoped. Downregulation of PP1 or PP2A lengthens period of the Drosophila clock (41, 42); in fibroblasts PP1 knockdown results in period shortening (43) whereas it is reported to lengthen period in U2OS cells (44), and in fibroblasts reduced PP5 caused modest period lengthening (45). In all cases effects were interpreted as affecting phosphorylation of Per proteins, not their heterodimeric activators. In U2OS cells, PPP4 has been shown to dephosphorylate BMAL1 but paradoxically, elevated PPP4 activity (which results in hypophosphorylation of BMAL1 and enhanced DNA-binding activity) results in reduced transactivation activity with a slightly longer period length, as it does in Drosophila (46). Conversely in Neurospora, the rgb-1 mutant in the essential regulatory subunit of PP2A causes a reduction in PP2A activity leading to WCC hyperphosphorylation, reduced DNA binding causing less active frq expression, and a three-hr period lengthening with no loss of rhythm strength (28, 31), whereas reduction/loss of PPP-1 or PPH-4 yields a 2 to 3 h period shortening (47, 48). Thus in both mammalian and fungal cells the rhythm continues despite loss of salient phosphatase activities (31, 47) suggesting redundancy or a modulatory as distinct from an essential role for dephosphorylation; notably, however, even the phenotypes resulting from loss of phosphatases lack internal consistency. This difficulty in drawing simple conclusions from correlations between phosphatase activities, positive element heterodimer function, and period length is highlighted by the ∆csp-6 strain in Neurospora that lacks the CSP-6 phosphatase and in which, while WCC is always hyperphosphorylated, the core clock still maintains a normal circadian period (49). Taken together, these inconsistencies suggested that there is more to find out about how positive element heterodimers in general, or WCC in particular, renews its activity to sustain circadian rhythms. We find, confirming prior predictions (28), that the answer lies in a more nuanced view of the pool(s) of WCC in the cell: At all times in both light and dark, most of the WCC pool is phosphorylated and inactive at C-box promoters; activation of negative element (here frq) expression relies on dephosphorylation/reactivation of only a small fraction of this cellular WCC pool, which becomes transiently dephosphorylated at times when WCC activity peaks in driving circadian expression of frq. We find that the circadian cycle of WCC repression and reactivation does not require ongoing transcription or translation, and inhibition of protein translation in the cell does not block robust dephosphorylation/reactivation of WCC (31), and even when the induction of wc-1, wc-2, or both is turned off, existing WCC is still able to support sustained circadian oscillations. These data provide an enhanced mechanistic basis for understanding how phosphorylation/dephosphorylation cycles lead to cyclic repression and reactivation of a heterodimeric transcriptional complex driving a circadian oscillation, highlighting significance of the balance between phosphorylation and dephosphorylation of circadian positive elements and thereby illuminating molecular events lying at the core of the circadian oscillator.

Results

An Activity for FRQ in the Light.

As a complex sensing light, WCC undergoes conformational changes to form a dimer (two copies each of WC-1 and WC-2), strongly inducing FRQ expression mainly through the pLRE in the frq promoter (9). In the dark, WCC acts simply as a monomer (one copy each of WC-1 and WC-2), and FRQ complexes with FRH and CK-1a, the FFC, to repress the circadian transcriptional activity of WCC by promoting its phosphorylation (13, 16, 18, 28, 31). FRQ and FFC have described roles only in the circadian system in darkness where, by repressing WCC activity, they impact expression of ~40% of the transcriptome (37); however, we find that FFC also represses WCC transcriptional activity at C-box in the light, an activity that is not surprising in hindsight.

Luciferase (Luc) expression driven by the C-box was tracked in ffc mutants ∆frq and frh^R806H, an frh mutant deficient in WCC repression in the dark (15). WCC activity at the C-box in either ∆frq or frh^R806H was prominently enhanced at the L-to-D transition, as shown by the higher light production at DD0 compared with the WT (Fig. 1A), confirming that FFC also functions in the light to inhibit WCC at the frq promoter. Because WCC is known to bind to and regulate myriad genes during the circadian cycle in darkness that are normally repressed in the light, the finding that FFC also represses this potentially light-driven promoter activity has biological significance: In strains lacking FRQ, C-box-regulated genes could be highly expressed in the light.

Fig. 1.

Enhanced frq C-box promoter activity of frq and frh mutants in the light reveals a function for FRQ in light signaling. (A) C-box promoter-driven luciferase expression was tracked in real-time in ∆frq and frh^R806H. Strains were grown in light overnight at 25 °C, synchronized by a light-to-dark (L-D) transfer, and bioluminescence signals were followed hourly in darkness at the same temperature. Three replicates (in three different colors) are shown with the x- and y-axes representing time (in hours) and signal intensity (in arbitrary units), respectively. (B) ChIP-seq using anti-WC-2 antibody as a proxy for WCC binding at the frq promoter in strains and conditions as marked. The gene map of the frq locus is at the Top; the black left-facing arrow marks the transcription start site, the C-box, pLRE, and AS (antisense) promoter elements are marked, and the ∆frq deletion is marked with the red bar at the Bottom; this deletion is the standard frq deletion from the Neurospora knockout collection (50). Published ChIP data for constant light and DD16 (38) are included for reference. (C) In ∆frq, pLRE retains light-induction that is lost in dark whereas C-box is always active. ChIP-quantitative PCR was carried out using WC-2 antibody with C-box- or pLRE-specific primer sets using strains and conditions as marked. Three technical replicates were performed, and the bars represent average values of ChIP signals plotted as a percentage of the input, with error bars representing the SEM (n = 3). Note that these experiments employed a novel ∆frq strain engineered to remove only the ORF while leaving C-box and pLRE promoter sequences intact. (D) Examples of a C-box-like promoter (WCC-bound in ∆frq but not in WT in LL; frq and WCC-dependent rhythmicity in DD) and pLRE-like promoter (WCC binding not frq-responsive in LL; arrhythmic in DD). (Top) WC-2 ChIP-seq data covering the C-box in ∆frq (Top) and WT (Bottom) strains grown in constant light. NCU04005 (casein kinase-1b [CK-1b]) is an example of a C-box-like promoter while NCU00582 (cryptochrome [cry]) is an example of a pLRE-like promoter. In each panel, transcription start sites are indicated by a black arrow, transcription units by a gray line, and boxes connote translated exons. (Bottom) Promoters of indicated genes were individually fused to the luciferase gene, the constructs were transformed to the cyclosporin resistance-1 (csr-1) locus, and light signals from the transgenic strains were followed in the dark over 5 d. “Δwc-2” means in the genetic background of wc-2-null. Y-axis marks light intensity in arbitrary units; x-axis refers to time in h.

To probe the effect of FRQ on WCC activity in the light, we used a chromatin-immunoprecipitation (IP) assay followed by sequencing (ChIP-seq) to compare binding of WCC to three WCC binding sites in the frq gene, the C-box and pLRE in the sense promoter of frq and a site in the antisense promoter of frq (see also Introduction). In WT following a brief light pulse given in darkness, WCC binds strongly to the pLRE and frq antisense promoters (Fig. 1B; see also ref. 51) whereas binding to the C-box is greatly reduced [Fig. 1B and (51)]; in constant light, binding to the frq antisense promoter is lost but pLRE binding remains. However, in ∆frq grown in constant light, WCC binds strongly to C-box and frq antisense promoters despite the light (Fig. 1B; note that the frq deletion used here removes the pLRE), confirming that FFC removes WCC from the C-box in the light as it does in the dark. In another assay, RT-qPCR was used to track WCC binding to the pLRE and C-box in a different frq deletion strain that just removes coding sequences while leaving both sense promoters intact (Fig. 1C). Loss of FRQ renders binding to C-box always elevated but does not impact light-regulation of pLRE binding: Light still induces pLRE binding (Fig. 1 C, Left) and darkness lessens pLRE binding (Fig. 1 C, Right). In constant light, WCC binds strongly to the pLRE despite high abundance of FRQ in the cell (52, 53), whereas FRQ (presumably with FRH and CK-1a) represses association of WCC with the C-box and the antisense promoter. Taken at face value these data indicate that there are potentially two classes of WCC-responsive directly light-regulatable promoters in the genome, pLRE-like that have been previously described (e.g., refs. 51 and 52) and C-box-like, repressed in the light and cyclically active in the dark.

To further investigate this, we performed ChIP-seq experiments to examine genome-wide WCC binding in WT and in ∆frq strains. We identified both classes, exemplars of which are shown in Fig. 1D, with a more extensive list in SI Appendix, Fig. S1 and compiled in SI Appendix, Table S1. Indeed, one of the C-box-like promoters is the antisense promoter of frq which drives rhythmic transcription of qrf, an antisense transcript (54). We find that roughly half of genomic sites bound by WCC fall into each class.

Unphosphorylated WCC Appears When Its Transcriptional Activity Peaks Over a Circadian Cycle.

A prediction of the data in Fig. 1, especially as regards frq, is that in order to keep C-box-like promoters repressed at the start of night, essentially the entire pool of WCC must be phosphorylated in the light at the Ser/Thr residues found to be essential for circadian repression (31). We tested this hypothesis using Phos-tag gels to distinguish a small fraction of unphosphorylated WC-1 from the much larger expected pool of phosphorylated WCC (Fig. 2A). The phosphosites of WC-1 that are collectively indispensable for feedback loop closure are S971, S988, S990, S992, S994, and S995; in wc-1^{1-970A&996-1167A} all identified WC-1 phosphosites except for S971, S988, S990, S992, S994, and S995 were collectively mutated to alanines; this strain still supports a robust oscillator [(31) and Fig. 2 A, Top]. As predicted, at the time of the L-to-D transition, all the essential residues of WC-1 and most of WC-2 are phosphorylated such that the WCC is inactive with respect to C-box activity (Fig. 2 A, Middle, Phos-tag data). Following the L-D transfer, the amount of WC-1 appears relatively constant, but strikingly most of the cellular pool of WC-1 remains fully phosphorylated throughout the day (Fig. 2 A, Middle). A fraction of this pool becomes dephosphorylated apparently synchronously at all six sites beginning sharply at CT22 (DD8, 8 h into darkness), due either to new synthesis or dephosphorylation, peaking 4 h later and disappearing by CT14 (DD24) (Fig. 2A). These changes are congruent in timing and magnitude with the rhythmic binding of WCC to the C-box that drives frq transcription throughout a day [Fig. 2B; (38, 51)].

Fig. 2.

A small fraction of WC-1 and WC-2 becomes dephosphorylated coincident with FRQ expression. (A) Phosphorylation at key residues of WC-1 and WC-2 cycles in a circadian manner. (Left) WB was performed using protein from C-terminally V5 tagged wc-1^{1-970A&996-1167A} grown at 25 °C and isolated at 4-h intervals across 24 h following a L-D transfer. 15 μg of total protein was loaded per lane in a regular Tris-Acetate SDS-PAGE gel or a Phos-tag gel as indicated. (Right) Quantifications of WB came from three biological replicates and were plotted with x- and y-axes representing time in h and relative band intensities in arbitrary units. “P,” phosphorylated isoforms; “deP,” dephosphorylated isoforms. (B) WC-2 ChIP as a proxy for WCC DNA binding at the frq promoter in the dark (from ref. 38). Samples were cultured for indicated h in the dark, formaldehyde-crosslinked, and harvested; ChIP with Neurospora samples was done with WC-2-specific antibody. (C) Phosphorylation changes at S971, S990, and S433 in wc-1^{1-970A&996-1167A, V5} over 48 h. (Left) WC-1^{1-970A&996-1167A} (C-terminally V5 tagged) was immunoprecipitated by V5 antibody, and WB was performed with indicated antibodies. (Right) Bands from three independent experiments were quantified, averaged, and plotted with error bars representing ±SEM.

To validate the Phos-tag result independently and also to track individual phosphorylation events, we raised custom antibodies against phosphorylated and unphosphorylated S971, a key phosphorylatable residue for the FFC-induced repression of WCC (31), and phosphorylated S990, a representative site for the clustered phosphorylations of S988-S995 (31, 55, 56) on WC-1, as well as phosphorylated S433 of WC-2 (18, 57), another critical phosphosite for WCC repression; all these phosphoevents are highly penetrant in the WCC population (31, 55, 57). To validate the custom antibodies against the phosphorylated- or unphosphorylated-specific residues on WC-1 and WC-2, WC-1 (V5 tagged), along with its associated WC-2, was first immunoprecipitated with V5 antibody and then treated with or without phosphatase. All the affinity-purified/depleted antibodies specifically detected their target residues with or without a phosphorylation modification (SI Appendix, Fig. S2 A and B). To quantify the Western Blot (WB) data from replicate studies, band intensities corresponding to unphosphorylated WC-1[S971] or phosphorylated WC-2[pS433] residues were normalized to total WC-1 or WC-2 from the same times over two circadian cycles in darkness (Fig. 2 C, Right-hand panels). Consistent with the data from the Phos-tag analysis of the grouped phosphoevents (Fig. 2A), the relative level of unphosphorylated S971 increased prominently from ~CT22 (DD8, DD32) with a peak at ~CT2 (DD12, DD36) and declined thereafter in both cycles in darkness (Fig. 2C). Assessing small changes on a large background, levels of phosphorylation of WC-1[S971&S988-S995], bulk WC-2 (SI Appendix, Fig. S3A), or S971 or S990 (SI Appendix, Fig. S3B), did not display robust rhythms, although bulk WC-2 phosphorylation appears to cycle; a careful reading of data from Schafmeier et al. (28) reveals phosphorylation changes in bulk WC-2 consistent with this. However, as strains bearing just one phosphorylatable residue on WC-1 remain weakly rhythmic in a WT wc-2 background (31) (SI Appendix, Fig. S4A), it is possible to follow changes in single residues in appropriately engineered strains; thus, rhythms in the phosphorylation status of just WC-1 S971 or S990 can be followed, suggesting that dephosphorylation events are not strictly dependent on one another (SI Appendix, Fig. S4B). Collectively, the data show that the appearance of unphosphorylated WCC correlated with its ability to bind to the C-box and activate frq expression [as predicted (18, 28, 47, 48, 58)], and also that the transcriptionally active WCC driving the oscillator represents just a small portion of total WCC, a portion that was predicted (28) but biochemically invisible unless it was specifically targeted as a population distinct from the total. This active unphosphorylated WCC must be derived from either new synthesis or dephosphorylation of the existing WCC, but first we wanted to know whether the events yielding phosphorylated inactive WCC were due mostly to FFC or reflected the overall cellular pools of CK-1a.

CK-1a in the FFC Is Involved in Phosphorylation of S971 and S433.

CK-1a has been shown to play a crucial role in FFC-mediated phosphorylation of WCC (13, 18, 19, 59), and the strength of FRQ- CK-1a steady-state interaction has been proposed to influence period length and temperature compensation (59, 60). Because CK-1a is essential for Neurospora (13), to assess the role of CK-1a in mediating phosphorylation at the representative residues of WC-1 and WC-2, we used three documented frq mutants, frq^Δ435-558, frq^{LLCN488-491AAAA}, and frq^{AQLH493-496AAAA} (13, 59), in which the FRQ and CK-1a interaction is severely impaired or even eliminated. We tested whether CK-1a is involved in phosphorylation of WC-1 and WC-2 at the sites of S971 and S433 respectively. Compatible with their performance on race tubes (13, 59), frq^Δ435-558, frq^{LLCN488-491AAAA}, and frq^{AQLH493-496AAAA} were all arrhythmic in the C-box-luc-reporter assay (Fig. 3A). Interestingly, the amount of unphosphorylated WC-1[S971] was remarkably elevated while phosphorylation of WC-2[S433] dropped noticeably in the three frq mutants (Fig. 3B), suggesting that FFC either directly phosphorylates the two sites of WCC or achieves this indirectly through other kinases. Prior investigations have implicated CK-1a in phosphorylating FRQ and governing the overall phosphorylation status of WCC (13, 18, 19, 22, 59). The data here link FRQ-bound CK-1a in the FFC with the crucial phosphorylation events on WCC that lead to circadian feedback repression.

Fig. 3.

Disruption of CK-1a-FRQ interaction promotes the accumulation of unphosphorylated S971 of WC-1 but downregulates the abundance of phosphorylated S433 of WC-2. (A) Luciferase reporter assay in frq^Δ435-558, frq^{LLCN488-491AAAA}, and frq^{AQLH493-496AAAA} confirming loss of rhythmicity in mutant strains compromised in FRQ- CK-1a interactions. Strains grown at 25 °C in the light overnight were transferred to the dark at 25 °C for circadian clock synchronization, and bioluminescence signals were followed every hour with a CCD (charge-coupled device) camera. Three replicates (in three different colors) were plotted with the x- and y-axes representing time (in hours) and signal intensity (in arbitrary units), respectively. The frq mutants were targeted to the frq native locus with a tandem V5 and 6 × histidine (V5H6) tag at their C termini, while WC-1 bore a 3 × FLAG tag at its C-terminus. Strain 661-4a (ras-1^bd, A, his-3::C-box-driven luciferase) served as WT in the assay. (B) (Top) Relative levels of unphosphorylated S971 and phosphorylated S433 in the frq mutants vs. WT. WC-1 (3 × FLAG tagged) was immunoprecipitated by FLAG antibody, and WB was carried out with antibodies indicated on the left of the blots. (Bottom) Quantifications of bands from three independent experiments with error bars representing ±SEM.

Robust Dephosphorylation of WC-1 at S971 Regenerates Active WCC and Proceeds in the Absence of De Novo Translation.

To probe the source of unphosphorylated active WCC appearing at CT22 (DD8) (Fig. 2 and SI Appendix, Fig. S4), the native promoter of wc-1 was replaced by the regulatable qa-2 promoter, a commonly used strategy to control gene expression in Neurospora (14); the strain also bears the C-box-luc reporter (24). In this strain synthesis of WC-1 is dependent on the inducer quinic acid (QA); the background level of expression without QA is so low that it cannot sustain rhythmicity (36) and the induced level is comparable to that seen in WT cultures (15, 36). In the first experiment, cultures were grown in the light and wc-1 was induced by adding 10⁻² M QA (Fig. 4A). The following morning, the Neurospora tissue was washed thoroughly to remove QA and thus stop wc-1 induction, transferred to QA-minus medium, immediately moved to the dark, and sampled every 4 h to probe relative changes in the key phosphorylation events at the residues, S971 and S433. To prevent nonspecific binding and reduce background for better quantification in WB, and to enrich WC-1 and its bound WC-2, WC-1 (V5-tagged) was isolated from cell lysates by V5 IP prior to WB. First, it is plain that when the inducer QA is washed out, WC-1 decays with a half-life of about 8 h indicating that new synthesis of WC-1 is required to maintain the levels seen in WT in Fig. 2A (Fig. 4A); the levels of pS971 and pS990 parallel this decrease, reflecting the total pool of WC-1. However, in contrast to this, after QA removal the relative level of unphosphorylated S971 transiently increased (Fig. 4A), consistent with what was observed from the time-course analysis of the same residue (Fig. 2C). Phosphorylation of WC-2 S433 followed the same trend as WC-1 pS971 and pS990, opposite to that of S971 in the dark (Fig. 4A). These data with those of Fig. 2 are consistent with vigorous dephosphorylation of clock-relevant residues on WCC as predicted (28). It is known, however, that there is a low level of background transcription from the qa-2 promoter even in the absence of the inducer, QA (36), so to further confirm that the unphosphorylated WCC seen after QA washout does not come from new expression, the experiment in Fig. 4A was repeated with the addition of cycloheximide (CHX) at the time of QA withdrawal (as indicated in Fig. 4B). The patterns of WC-1 decay and dephosphorylation were similar to that seen from the CHX-free samples (Fig. 4A), although WC-1 appeared to be slightly stabilized [perhaps reflecting absence of activity-mediated turnover (61)]. Again, levels of S971 rose despite decay of total WC-1, peaking similarly at DD8 and then dropping at DD12 (Fig. 4B). These data demonstrate that clock-relevant sites on WCC such as S971 can be aggressively dephosphorylated in the dark independent of new translation, consistent with prior speculation (28, 47, 48, 58), and that reactivation of WCC arising from this dephosphorylation could itself be sufficient for persistence of rhythmicity.

Fig. 4.

Dephosphorylation of WCC proceeds in the absence of new WC-1 expression. (A) (Top) qa-2-driven wc-1^V5 was grown in LCM containing 0.1% glucose and 10⁻² M QA in the light at 25 °C overnight. Cultures were thoroughly washed, placed in 0.1% glucose LCM without QA and immediately transferred to the dark at 25 °C, and were harvested at 4, 8, or 12 h following the dark transfer. Protein was isolated and V5 IP (for WC-1^V5) was performed to reduce nonspecific binding by phospho- or unphospho-specific antibodies. WB was conducted with indicated antibodies, and representative results are shown. (Bottom) Bands from three replicates were quantified, averaged, ratios calculated as shown, and plotted with error bars representing SEM. (B) The same experiment as in (A) was carried out with the exception of adding CHX at the final concentration of 40 µg/mL to the culture prior to the dark transfer. Quantification and plotting of WB in (A) and (B) were derived from three independent experiments.

A Cycle of Dephosphorylation without New WCC Synthesis Is Sufficient to Operate the Core Oscillator.

Given that termination of induced wc-1 mRNA transcription plus general inhibition of translation did not prevent dephosphorylation of WC-1 at S971 (Fig. 4), we asked whether the core oscillator can persist with only dephosphorylation of existing WCC (Fig. 5A). To this end, the qa-2:wc-1 strain was cultured in QA (10⁻² M)-containing medium overnight, and the following morning the treated tissue was split; half was washed free of QA and cultured in medium without QA while the other was kept with QA (10⁻² M). Both cultures were immediately moved to darkness, and rhythmicity was followed using the frq C-box-luc reporter. Another third parallel culture grown without QA throughout served as a control for background signals generated by the residual WC-1 expressed from the slightly leaky qa-2 promoter. Cultures grown without QA showed low WCC activity (based on luciferase activity) and could be considered arrhythmic although a delayed shoulder of activity appeared in some cultures (Fig. 5 B, Bottom Left). These data suggest that the extremely low level of WC-1 from leaky expression of the promoter was not sufficient to support rhythmicity, consistent with prior observations (15, 36). Interestingly, however, the QA-induced/washed sample remained rhythmic for 3 d and the culture exposed to QA throughout for 4 to 5 d, both with identical ~21 h period lengths (Fig. 5 B, Top Left and Middle Left). The same set of treatments for the qa-2:wc-1 strain were repeated using a new reporter strain, qa-2:wc-1^{1-970A&996-1167A}, in which only the six residues on WC-1 key to the core clock (31) remain phosphorylatable while all other phosphorylatable Ser and Thr in WC-1 were mutated to alanines. The rhythm now persists through three to four cycles of reactivation, the same number as the continually induced culture (Fig. 5 B, Right), and WC-1[1-970A&996-1167A] even appears to be slightly more stable than the WT (Fig. 5C), consistent with the cues driving proteasome-mediated turnover of active transcription factors (30) being among the mutated phosphosites (31). Reflecting this enhanced stability of WC-1[1-970A&996-1167A] it could be noted that even when qa-2:wc-1^{1-970A&996-1167A} was cultured without QA throughout, it still displayed one delayed, severely dampened long period cycle that might be generated from residual WC-1, the long period being consistent with severely reduced WC-1 (36). In any case, given that rhythms can persist for at least 4 cycles in the strain wc-1^{1-970A&996-1167A}, wc-2^15pA that has only the six clock-relevant phosphorylatable residues in the WC-1 (31), these data suggest that new synthesis of WC-1 is not required for a rhythm sustained over several days nor for period-length determination of the core clock, and that derepression of the WCC via dephosphorylation alone may be sufficient for sustained rhythmicity.

Fig. 5.

Circadian rhythms persist for several days in the absence of substantive new WCC synthesis. (A) Schematic illustrating the procedure for overexpressing wc-1, wc-2, or both (as indicated) and subsequently monitoring their activity in sustaining the clock. (B) Luciferase analysis of qa-2-driven wc-1^V5 or wc-1^{1-970A&996-1167A, V5} under different induction conditions. The strain was cultured in LCM (0.1% glucose) with 10⁻² M QA at 30 °C in the light. The following morning the tissue was washed thoroughly, split into two pieces, and each inoculated in LCM (0.1% glucose) in the absence (Upper) or presence (Middle) of 10⁻² M QA in the light at 25 °C overnight, after which the cultures were moved to the dark for tracking bioluminescence signals in the dark at 25 °C. Bottom, another culture was treated in the same way as the other two but never encountered QA. (C) (Top) WB tracking WC-1 and WC-2 levels in strains as indicated at the Top. Leftmost lanes corresponding to Bottom panels in B: Neurospora tissue was cultured in LCM (0.1% glucose) without QA at 30 °C in the light and then transferred to 25 °C in the light and grown overnight. Middle and rightmost lanes: An overnight culture was grown with 10⁻² M QA at 30 °C plus light, washed thoroughly to remove QA and split into two pieces. Half the tissue was cultured in the light at 25 °C overnight in the absence of QA (Middle) while the other with QA at 10⁻² M (Right). Asterisks denoted nonspecific bands showing equal loadings in WB. All cultures were grown in petri dishes without shaking. (Bottom) Bands from three independent experiments were quantified, averaged, and plotted with error bars representing SEM. (D) Luciferase analysis of qa-2-driven wc-2; qa-2-driven wc-1, qa-2-driven wc-2; and qa-2-driven wc-1^{1-970A&996-1167A, V5}, qa-2-driven wc-2 under different induction conditions. Conidia from the indicated strains were inoculated into 2% LCM and incubated overnight at 30 °C. The following day, the medium was replaced with 0.1% glucose, agar-free, race tube medium supplemented with 10⁻² M QA to induce expression of WC-1, WC-2, or both as indicated. Cultures were then incubated overnight at 25 °C under constant light (LL). The next morning, tissue plugs were excised, returned to the same QA-containing medium, and incubated for an additional 6 h at 25 °C under constant light. The plugs were then washed with 0.1% glucose, QA-free, race tube medium to remove QA and transferred to fresh 0.1% glucose, QA-free, agar-free medium for an additional 2-h incubation under constant light. Finally, the samples were moved to darkness at 25 °C for luciferase signal recording using a CCD camera. “No QA throughout” refers to a parallel set of treatments conducted without the addition of QA at any stage.

WC-1 and WC-2 are relatively stable yet low-abundance proteins, whose expression levels correlate positively with their transcriptional activity in sustaining frq expression and the core circadian clock (36, 62). WB analysis was used to follow expression of WC-1 and WC-2 with samples undergoing the same set of QA treatments as for the luciferase assay (Fig. 5B). As expected, WC-1 in the two inducible strains was low when no QA was included in the medium (Fig. 5C), intermediate in the QA- supplemented/washed samples, and highest in samples that encountered QA all the time (Fig. 5C), and appeared slightly higher in the wc-1^{1-970A&996-1167A} background, in all reflecting performance of WCC in maintaining circadian rhythms (Fig. 5B).

To examine whether dephosphorylation of WC-2 is sufficient for recycling WCC’s circadian activity, experiments parallel to those in Fig. 5B for WC-1 were performed for WC-2. It is known that WC-2 levels in wc-2 QA-inducible strains grown without the inducer are markedly lower than those in strains carrying the WT wc-2 under comparable conditions (15) in a manner like WC-1 in Fig. 5C, and in WT cells, WC-2 is always in excess relative to WC-1 so its abundance does not serve as a primary determinant or a limiting component for the feedback loop (31, 63, 64). A qa-2:wc-2 strain was grown in the light in the presence of QA to induce WC-2 expression. QA was then withdrawn to terminate wc-2 transcription, and the activity of a C-box-driven luciferase reporter was monitored in constant darkness. Following QA removal, luciferase rhythms persisted for at least 4 d (Fig. 5 D, Top), indicating that—similar to WC-1—preexisting WC-2 is sufficient to sustain circadian oscillations.

Because WC-1 and WC-2 form a stable WCC, to determine whether the intact WCC could also function independently of new protein synthesis, we performed the same QA induction/withdrawal experiment in a qa-2:wc-1, qa-2:wc-2 strain, in which expression of both WC proteins is driven by the qa-2 promoter (36). This condition also produced multiple robust circadian cycles in constant darkness (Fig. 5 D, Middle), consistent with the idea that preexisting WCC is competent to maintain rhythmic activation/repression activity without continuous synthesis. When this experiment was repeated using the qa-2:wc-1^{1-970A&996-1167A}, qa-2:wc-2 strain, two full circadian cycles were observed (Fig. 5 D, Bottom), suggesting that this mutant version of WC-1 may have an increased dependence on continuous WC-2 expression, but plainly establishing that daily new synthesis of WCC is not needed for rhythmicity. Taken together, our data support that existing positive element complexes in the cell are sufficient to sustain circadian rhythms in the dark through repeated cycles of dephosphorylation/activation and rephosphorylation/repression.

Discussion

The WCC is the prototypic blue light photoreceptor for the Kingdom of Fungi as well as being an exemplar of the heterodimeric transcription factors that drive circadian oscillators in fungi and animals; in Neurospora it mediates both circadian regulation and acute light-regulation of a host of target genes. This biology, driven by the same protein complex, is vast and a long-standing question is why all WCC-bound promoters are not similarly regulated, i.e., why every clock-regulated WCC target is not also transcriptionally light-regulated. Part of the evidence of this confusion, perhaps, lies in the lack of consensus regarding the WCC recognition motif which is variously reported as centered on GATC or ATCG (9, 51, 52) with a great deal of variation in the surrounding context (e.g., ref. 38). As the C-box is weakly light-responsive (9, 51), this ambiguity may now be understood as the confounding of the two promoter types, C-box-like and pLRE-like, that have some similar and some distinct characteristics but are bound by the same transcription factor complex that undergoes conformational changes and multimerization under different conditions (8, 9). Indeed, despite the C-box being quiescent in constant light due to FRQ-based repression (Fig. 1), many C-box-like promoters have been included in lists of acutely light-responsive genes (e.g., refs. 51 and 52). More recent biochemical analyses have identified differences in that the light-activated WCC dimer (Introduction) that binds the pLRE is structurally and functionally distinct from the WCC monomer (Introduction) in the dark that binds the C-box. DNA binding for light-induction via pLRE of frq only needs the Zinc finger (ZnF) DNA binding domain of WC-2 while circadian DNA binding of WCC to C-box requires the ZnF and nearby regions of WC-2 as well as the ZnF and DBD (defective in binding DNA) motif of WC-1 (65). This is consistent with the observation that FFC-promoted phosphorylation of WC-1 does not inhibit the WCC dimer binding to the pLRE. This distinction, now underscored by the division of genome-wide sites into the two classes, makes understanding the structural basis within the WCC for this distinction a worthy goal.

The best studied WCC-responsive promoter is that of frq where WCC binds at both the C-box and pLRE (9, 11). The two sets of unexpected observations reported here stem from trying to understand the importance of phosphorylation in regulating the binding activity of the WCC to these sites. The first observation is that in the light the FFC acts to remove WCC from the C-box but not from the pLRE, giving rise to the two classes of WCC-bound promoters. This finding provides a role in the light for FRQ and the FFC, a protein and complex whose roles in the cell have previously been confined to circadian regulation in the dark. Accepting that WCC may be just one of many TFs impacting expression of its targets, generalizations may still be possible as the two classes of WCC-regulated promoters have distinct behaviors: pLRE-like promoters display high activity at dusk and are relatively inactive at night whereas C-box-like have lower activity at dusk that increases through the night and decreases with the arrival of dawn. It is not difficult to imagine a potential selective advantage for two such temporally distinct and differentially light-responsive promoters.

The second unexpected observation is that during the circadian day essentially the whole pool of WC-1 is fully phosphorylated at clock-essential sites; following from this is realization of the importance of dephosphorylation, and not de novo synthesis, of WC-1 for clock function. WC-1 undergoes FFC-promoted phosphorylation at 80 sites, largely the direct result of FFC-mediated CK-1a activity (Fig. 3), among which only a few have been found to be necessary for circadian repression or regulating transcriptional output ccgs (clock-controlled genes)–circadian amplitude. Although the causes of WCC phosphorylation/inactivation are becoming understood, prior work suggested (e.g., refs. 13 and 28) but could not fully address the source and timing of active hypophosphorylated (or unphosphorylated) WCC, especially the central positive arm protein WC-1. Surprisingly, the pool of WC-1, even when its induction is disallowed, remains able to support several circadian cycles; plainly, de novo synthesis of WCC, especially WC-1, would be needed eventually as WC-1 is consumed when it is transcriptionally active (15, 28, 31, 56) or when rapid cell growth extremely dilutes the existing WCC pool, but daily synthesis of WCC is not needed for circadian function.

Biochemical analysis has revealed that only a small percentage of FFC and WCC interact with each other in cells and do not form abundant stable complexes in the process of feedback repression (28, 66); most WCC resides in the nucleus (28) while most FFC is found in the cytoplasm (28, 62). These observations together imply transient but dynamic interactions between FFC and WCC, probably more akin to the recognition between an enzyme and its substrate (28, 62). This dynamic association enables FFC to efficiently maintain the majority of WCC in a hyperphosphorylated/inactive state, eventually leading to inactivation of the frq promoter; when FFC affinity for WCC is depressed by CK-1a -mediated FRQ phosphorylation, a small share of WCC can evade control by FFC to be dephosphorylated, regaining its DNA binding capability (Fig. 6) and ability to rapidly drive frq expression. As the resulting non- or lightly phosphorylated new and functional FRQ (24), the rate-limiting factor in FFC, increases in abundance, FFC becomes activated again in promoting phosphorylation of WCC to abrogate DNA-binding and terminate frq expression. Interestingly, circadian regulations for FFC and WCC activities share some fundamental mechanistic basis: Although new proteins must be provided to replace those lost to turnover, coupled daily synthesis and degradation are not required for either complex to function in circadian cycles (24, 36), and multisite phosphorylation events determine and modulate their activities (18, 22, 23, 31, 67, 68); this may explain why their activating and repressing phases over circadian cycles last so long relative to proteins working in other signaling pathways, such as WCC acting in light-responses of Neurospora (52).

Fig. 6.

Working model showing how the overall activity of WCC in frq transcription is controlled by FFC and a phosphatase(s). Promoter elements and relevant proteins acting on the frq locus are marked. In the light, FRQ is continually expressed due to dimeric WCC action on the pLRE, and sites on WCC essential for binding to the C-box are phosphorylated maximally due to high FRQ abundance/activity, abrogating WCC binding to the C-box. When the light disappears (Dark-Inactive phase), new FRQ synthesis stops but existing light-induced FRQ ensures FFC-mediated WCC phosphorylation; WCC cannot bind the C-box, and the dimer of WCCs dissociates and is unable to bind the pLRE. FRQ becomes phosphorylated so FFC gradually loses its capacity to promote phosphorylation of WCC while the unknown phosphatases in the cell, including but not limited to PP2A (28, 31), gain the upper hand over FFC, leading to dephosphorylation/activation of a small fraction of WCC (Dark-Active phase); WCC binds the C-box driving frq transcription. New FRQ rapidly nucleates active FFC, again promoting phosphorylation of WCC in excess of the ability of the phosphatases to remove it, leading to the removal of WCC from the frq promoter, termination of FRQ expression, and reestablishment of the Dark-Inactive phase.

We do not know yet what phosphatases are responsible for dephosphorylating WCC to reactivate its circadian DNA binding activity, although several findings suggest that the actions of multiple phosphatases are needed. In particular, only period modifications rather than complete loss of rhythmicity have been noted in the nonessential phosphatase-deficient strains examined, and even then the effects are not internally consistent (31, 47, 48, 69). For instance, although hyperphosphorylated WCC was noted in the mutants of both RGB-1 (affecting PP2A activity), PPH4, and CSP-6 (28, 47, 49), rgb-1^RIP results in a long period clock (31), whereas loss of PPH4 and PPP-1 shortens period length (47, 48) and loss of CSP-6 has negligible effects on period, although Δcsp-6 fails to show an overt rhythm (49).

Taken together, in darkness, WCC monomer activates FRQ expression, leading to the assembly of the FFC, which subsequently phosphorylates and represses the circadian activity of WCC at the C-box-like promoters. As the FFC gradually loses its repressive function by CK-1a-mediated phosphorylation, dephosphorylation of WCC by unknown phosphatase(s) restores its activity at these promoters. This cyclic activation and repression establishes circadian oscillations. WCC monomer does not associate with pLRE-like promoters. Upon light exposure, WCC monomer is converted to WCC dimer (Introduction), which can bind and activate pLRE-like promoters. The elevated level of FFC assembled from continuously expressed FRQ enhances the phosphorylation of WCC, displacing it from C-box-like promoters if it was bound there in the dark prior to illumination. In contrast, WCC dimer associated with pLRE-like promoters is resistant to FFC-mediated repression. Continuous FRQ production in the light maintains high FFC activity, which keeps WCC in a constitutively phosphorylated state with no opportunity for dephosphorylation, thereby persistently preventing it from rebinding to C-box-like promoters.

Our findings in fungi may shed some light on mammalian clocks that share a nearly identical regulatory architecture: A heterodimer of proteins interacting via PAS domains constitute a transcription factor complex that drives expression of genes whose protein products enter a complex that recruits CK1 which, in turn, phosphorylates and thereby represses the activity of the heterodimer by reducing its ability to bind to DNA. Although expression rhythms are seen in both BMAL1 and WC-1 (1, 5, 70), constitutive expression of either protein results in an essentially WT period length (34, 36, 71, 72). In mammals, both inhibition of CK1δ and activity of PPP4 leads to hypophosphorylated BMAL1 which has greater ability to bind DNA (32, 33, 46); inhibition of CK1 lengthens period in fungi and animals and inhibition of PPP4 in U2OS cells (PPH-4 in Neurospora) shortens period in both systems (46, 47). A unifying mechanistic model would include a dynamic balance of phosphorylation/dephosphorylation in controlling the positive arm activity, although clock proteins may act slightly differently to perform the same tasks among circadian systems. The negative element complexes (either FRQ/FRH in Neurospora or Pers/Crys in mammals) recruit CK1 to repress the circadian activity of their transcription factor complexes (either WCC in Neurospora or BMAL1/CLOCK in mammals) via phosphorylation promotion; the positive element complexes are restored to their circadian functions through timely dephosphorylation.

We have shown here that the circadian oscillator can run for several days by creating/recycling active WCC from the larger phosphorylated/inactive pool. Similarly, we have previously described a daily cycle in the ability of FFC to interact with WCC, in which newly made unphosphorylated FRQ interacts well but old heavily phosphorylated FRQ does not (18). Phosphorylated FRQ is unstable, but we now know that FRQ stability can be uncoupled from period determination (24); it is instead the phosphorylation status of FRQ that drives the cycle of FFC/WCC interaction, and thereby drives the clock. Indeed, if unphosphorylated FRQ could be generated by rapid dephosphorylation of old FRQ instead of through de novo synthesis, the clock still ought, in theory, to cycle. Stated in this way, the classic Transcription/Translation Feedback Loop (TTFL) oscillator cycle begins to take on distinctive aspects of a “phoscillator” wherein all the clock-relevant circadian parameters, those determining periodicity, persistence, and compensation, might be explained by cycles in phosphorylation of the key players, without a requirement for daily transcription/translation at all.

Materials and Methods

Neurospora Strains.

Neurospora strain 661-4a (ras-1^bd, A, his-3::C-box-driven luciferase) was used as the WT strain for luciferase assays. wc-1, wc-2, and frq mutants were generated using a previously described yeast-based homologous recombination method (73). In ∆frq in Fig. 1C, the ORF (from the starting codon [the 1st ATG] to the stop codon TAG) of frq was replaced with the bar gene (conferring ignite resistance) while all its promoters (C-box, pLRE, and AS [antisense]) remain intact. Mutations were confirmed by Sanger sequencing using frq-specific primers at the Dartmouth Genomics Core facility. Primer sets used to generate promoter–luciferase fusion constructs are listed in SI Appendix, Table S2. Neurospora transformation was performed by electroporation (BTX, Model # ECM630; 1,500 V, 600 Ω, 25 μF) (23, 65).

Growth Media.

Vegetative cultures were grown on complete-medium slants containing 1 × Vogel’s salts, 1.6% glycerol, 0.025% casein hydrolysate, 0.5% yeast extract, 0.5% malt extract, and 1.5% agar (74). Neurospora sexual crosses were carried out at 25 °C with Westergaard’s agar plates bearing 1 × Westergaard’s salts, 2% sucrose, 50 ng/mL biotin, and 1.5% agar (66). Liquid culture medium (LCM) used for culturing Neurospora for IP and WB contains 1 × Vogel’s, 0.5% arginine, 50 ng/mL biotin, and 2% glucose (75); for QA induction experiments, glucose was reduced to 0.1%.

ChIP.

Freshly harvested Neurospora tissue was treated with 3% formaldehyde for 15 min and then with 0.25 M glycine for 5 min. Samples were washed three times with PBS, vacuum-dried, and 0.4 g of tissue was cut into nine pieces and each soaked in 0.5 mL SDS lysis buffer (50 mM Tris-HCl [pH 8.0], 1% SDS, 5 mM EDTA) supplemented with protease inhibitors (Roche, Sigma-Aldrich, Cat. #11836170001).

The tissue was first sonicated three times for 8 s at 30% amplitude using a Branson sonicator equipped with a microtip, followed by five cycles of 5 min sonication in a water-bath sonicator (30 s on/30 s off). The lysate was clarified by centrifugation at 8,000 rpm for 5 min twice. A 200 μL aliquot of the supernatant was mixed with 1.8 mL RIPA buffer, and IP was performed by adding 15 μL of protein A magnetic beads and 5 μL of WC-2 antibody, followed by overnight rotation at 4 °C.

The next day, immunoprecipitates were sequentially washed once each with buffers A–D (buffer A: 20 mM Tris-HCl [pH 8.0], 0.1% SDS, 2 mM EDTA, 1% Triton X-100; buffer B: same as A plus 500 mM NaCl; buffer C: 10 mM Tris-HCl [pH 8.0], 0.25 M LiCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate; buffer D: 10 mM Tris-HCl [pH 8.0], 1 mM EDTA). DNA–protein complexes were eluted with elution buffer (0.1 M NaHCO₃, 1% SDS), reverse cross-linked with NaCl at a final concentration of 0.2 M, and purified using a Qiagen PCR purification kit. The purified ChIP DNA was analyzed either by sequencing or by PCR with gene-specific primers (31).

Phos-Tag Gel.

To better resolve WC-1 and WC-2 phosphorylation events, Phos-tag reagent (ApexBio) was incorporated into a 6.5% SDS-PAGE Tris-Glycine gel at a final concentration of 20 μM, with an acrylamide/bisacrylamide ratio of 149:1 (23, 31).

Custom Antibodies against the Phosphorylatable Residues of WC-1 and WC-2.

Custom antibodies against phosphorylated S971, phosphorylated S990, or unphosphorylated S971 of WC-1 and phosphorylated S433 of WC-2 were ordered from GenScript Biotech Corporation and raised using New Zealand rabbits. Immunogenic peptides used in rabbit injections included (KKSN[p{phosphorylated}Ser]PSHSSPLHRC)-KLH (keyhole limpet hemocyanin) conjugate for phosphorylated S990, (TGNA[pSer]PTLIKGDAGC)-KLH conjugate for phosphorylated S433, (GRV[pSer]PRTSSRGGNGC)-KLH conjugate for phosphorylated S971, and (GRVSPRTSSRGGNGC)-KLH conjugate for unphosphorylated S971. Antibodies from rabbit sera were first affinity-purified with the immunogenic peptides and then immunodepleted against the immunogenic peptides with or without a phosphorylation modification at the indicated site for unphospho- or phospho-specific antibodies respectively. More detailed information on the purification and depletion of these antibodies is proprietary and was not disclosed by the manufacturer. All the purified antibodies as the primary antibodies in WB were used by a 1: 750 dilution (20 μL antibodies to 15 mL PBST [0.3% Tween-20]), being incubated with wet-transferred protein-coated PVDF (polyvinylidene fluoride) blots in a cold room (4 °C) overnight.

Protein Extraction, IP, and WB.

Protein extraction, IP, and WB were performed as previously described (66). Protein-extraction buffer (10 mL; 50 mM HEPES [pH 7.4], 137 mM NaCl, 10% glycerol, 0.4% NP-40) supplemented with one tablet of complete, Mini, EDTA-free Protease Inhibitor Cocktail (Roche, Catalog # 04693159001) was added to the ground powder and mixed by vortexing. The mixture was alternately vortexed (10 s) and chilled on ice (10 s) for 2 min, incubated on ice for an additional 10 min, and clarified by centrifugation at 12,000 rpm at 4 °C for 10 min.

For IP using Neurospora lysate, cleared extracts containing 2 mg of protein, as determined by absorbance at 280 nm, were incubated with 25 μL of FLAG M2 agarose (Sigma-Aldrich, Catalog # A2220) or 25 μL of V5 agarose (Sigma-Aldrich, Catalog # A7345) followed by rotating at 4 °C for 2 h (76). Protein-bound agarose beads were washed twice with protein extraction buffer containing protease inhibitors, eluted in 100 μL of 5 × SDS sample buffer at 99 °C for 5 min, and 10 μL of the eluate was loaded per lane.

For WB, 15 μg of cleared whole cell protein lysate was loaded per lane in a commercial 3 to 8% Tris-Acetate SDS gel (1.5-mm thickness, 15-well format; Thermo Fisher Scientific, Catalog # EA03785BOX) and electrophoresed using 1 × NuPAGE Tris-Acetate SDS Running Buffer (Thermo Fisher Scientific, Catalog # LA0041). Custom rabbit FRQ, WC-1, and WC-2 antibodies used for WB have been described previously (15, 53, 64, 77); the application of custom phospho- or unphospho-specific antibodies is detailed in the section titled “Custom antibodies against the phosphorylatable residues of WC-1 and WC-2.” Phos-tag gels (6.5% SDS-PAGE containing 20 mM Phos-tag with an acrylamide:bisacrylamide ratio of 149:1) were prepared following a previous publication (31).

For Fig. 4A, the strain was grown in 0.1% LCM (1× Vogel’s, 0.5% arginine, 50 ng/mL biotin, and 0.1% glucose) supplemented with 10⁻² M QA under constant light at 25 °C overnight. The next day, cultures were thoroughly washed, transferred to fresh 0.1% glucose LCM without QA, and immediately shifted to darkness at 25 °C. Samples were collected 4, 8, and 12 h after the dark transfer. Proteins were extracted, and V5 IP (for WC-1^V5) was performed to reduce nonspecific binding by antibodies. For Fig. 4B, the same procedure was followed as in Fig. 4A, except that cycloheximide (CHX) was added to the culture at a final concentration of 40 µg/mL immediately before the dark transfer.

Luciferase-Reporter Assays.

Luciferase assays were performed as previously described (78). Briefly, opaque 96-well plates containing race-tube medium were inoculated with a conidial suspension and grown at 25 °C in constant light for 16 to 24 h, then transferred to darkness. Bioluminescence was recorded hourly using a CCD (charge-coupled device) camera. Data were acquired with ImageJ and a custom macro, and period lengths were determined manually.

Assays Examining the Capacity of Preexisting WCC to Sustain the Clock.

In Fig. 5B, strains were initially grown overnight in 0.1% LCM (1× Vogel’s, 0.5% arginine, 50 ng/mL biotin, and 0.1% glucose) containing 10⁻² M QA at 30 °C in the light. The following morning, the tissue was thoroughly washed, divided into two portions, and each was transferred to 0.1% LCM with or without 10⁻² M QA and incubated overnight in the light at 25 °C. The cultures were then shifted to darkness at 25 °C to record bioluminescence rhythms. As a control, an additional culture (“no QA throughout”) was processed identically but never exposed to QA. In Fig. 5D, conidia from the indicated strains were inoculated into 2% LCM and incubated overnight at 30 °C. The next day, the medium was replaced with 0.1% glucose, agar-free race tube medium containing 10⁻² M QA to induce expression of WC-1, WC-2, or both, as indicated. Cultures were then incubated overnight at 25 °C under constant light (LL). The following morning, tissue plugs were excised, transferred to fresh QA-containing medium, and incubated for another 6 h at 25 °C in the light. After induction, the plugs were washed with 0.1% glucose, QA-free, agar-free race tube medium and transferred to fresh 0.1% glucose, QA-free, agar-free race tube medium for an additional 2 h under constant light. Finally, the samples were shifted to darkness at 25 °C, and luciferase signals were recorded using a CCD camera. The “no QA throughout” control underwent the same treatment but without QA at any step.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Previously published data were used for this work (38). All other data are included in the manuscript and/or SI Appendix.