Transcriptional repression of frequency by the IEC-1-INO80 complex is required for normal Neurospora circadian clock function

To characterize the transcription factors that are involved in the regulation of frq expression, we examined the conidiation rhythms of available knockout transcription factor mutants with race tube assays. As shown in Fig 1A, the strain with deletion of the iec-1 (ras-1⁺) (NCU03206) gene showed arrhythmic conidiation in a race tube assay compared to the wild-type strain, suggesting that IEC-1 plays a critical role in Neurospora circadian conidiation rhythm. The iec-1 gene codes a C2H2 finger domain-containing protein IEC-1, a homolog of mammalian Yin Yang 1 (YY1), Drosophila PcG PHO, and yeast Iec1 [39]. When the IEC-1 protein sequence was used in a BLAST search against protein databases, its homologs were found to be highly conserved in ascomycetes (Fig 1B). To confirm the phenotype of the iec-1 mutant, the iec-1 gene, which is driven by the quinic acid (QA)-inducible qa-2 promoter, was reintroduced to the iec-1^KO strain. In addition, five copies of the c-Myc epitope with six histidine residues [42] were inserted at the N-terminus of the IEC-1 ORF to facilitate the detection of the expression of Myc-IEC-1 using a c-Myc monoclonal antibody (9E10). In QA-containing race tubes, the circadian conidiation rhythms of the iec-1^KO, qa-Myc-IEC-1 strains were similar to those of the wild-type strains (Fig 1C). These data indicate that Myc-IEC-1 can partially complement the function of the endogenous IEC-1 protein in the iec-1^KO strain. To test whether light exposure could entrain the conidiation rhythms, the wild-type strain and iec-1^KO strains were assayed under light/dark cycles (LD) at 25°C. The race tube assays showed that the conidiation process of the iec-1^KO strains was still entrained by LD cycles (S1A Fig). Meanwhile, a reduced light induction of FRQ expression by light was observed in the iec-1 mutant (S1B Fig). Notably, the basal level of FRQ is higher in the mutant. Nonetheless, this result suggests that IEC-1 also affect the light induction of FRQ expression. These data suggested that the clock phenotype of the mutant strain is not caused by a defect in the input pathway of clock. To further confirm the arrhythmic phenotype of the iec-1^KO strain, we introduced bioluminescence reporter constructs (frq-luc) into the his-3 locus of the iec-1^KO strains and the wild-type strains, in which the luciferase expression of both strains are driven by the frq promoter [43]. The sequence analysis of genomic DNA showed that the frq promoters in the luciferase-expressing strains were intact (S2 Fig). Consistent with the phenotype in the race tube assay, the robust bioluminescence rhythm of the wt, frq-luc strain was abolished in the iec-1^KO, frq-luc strains (Fig 1D and S3A Fig), indicating that IEC-1 is critical for circadian clock function at the molecular level.

To further determine the role of IEC-1 in the circadian clock, we examined the FRQ expression profile at different time points in constant darkness (DD). Both FRQ rhythms and FRQ phosphorylation profiles were disrupted in the iec-1^KO strain (Fig 2A). Northern blot analyses showed that the abolished FRQ rhythms were due to constant transcription of the frq gene in the iec-1^KO strain (Fig 2B). To test whether IEC-1 directly regulates frq transcription, a chromatin immunoprecipitation (ChIP) assay was performed by using an IEC-1-specific polyclonal antibody (Fig 2C). Because the IEC-1 antibody recognizes nonspecific bands, the iec-1^KO strain was used as the negative control in the ChIP assays. The ChIP assay revealed enrichment of IEC-1 at the frq promoter, ORF 3’, and the 3’UTR but not at ORF 5’ and the middle regions at DD18 (Fig 2D). Furthermore, the association of IEC-1 with C-box fluctuated from DD10 to DD42 (Fig 2E). Altogether, these results indicate that IEC-1 functions at the frq locus to regulate frq transcription.

Previous studies showed that yeast Iec1or mammalian Yin Yang 1 (YY1) co-purified with the INO80 complex in yeast or mammalian cells [39,44]. These transcription factors are required for the recruitment of the INO80 complex to target genes. Since the INO80 complex acts as a conserved co-factor of Iec1/YY1 in different organisms, we performed an IP assay to test the association of the INO80 complex with IEC-1 in Neurospora. The results showed that FLAG-IEC-1 was immunoprecipitated (IP) with INO80 (NCU08919) by the INO80 antiserum but not immunoprecipitated with the preimmune (PI) serum (S4A Fig). In the INO80 complex, INO80 is the catalytic subunit and IES-1 is a structural subunit [33,37,45]. To further identify whether the INO80 complex functions in the same pathway as IEC-1 in the regulation of the Neurospora circadian clock, we generated ino80 (NCU08919) and ies-1 (NCU01362) deletion mutants. As expected, both ino80^KO and ies-1^KO strains exhibited arrhythmic conidiation phenotypes in the race tube assays (S4B Fig). We also tried to generate mutants with a band background (ras-1^bd), but we only obtained heterokaryotic strains. These strains also showed arrhythmic conidiation phenotypes in the race tube assays (S4 Fig). To further confirm that deletion of ino80 or ies-1 is the only cause of the clock defect in these mutants, rescue strains of the mutants were generated. The conidiation rhythms of the ino80^KO, qa-Myc-INO80 and ies-1^KO, qa-Myc-IES-1 transformants were not restored in the race tubes in the absence of QA (S4C and S4D Fig). Meanwhile, obvious conidiation rhythms were observed in the rescue strains but not in the knock-out strains in the race tube assays with 10⁻³ M QA (S4C and S4D Fig). However, we also observed a phase difference between the wild-type and the rescue strains. The results suggest that qa-2 promoter-driven expression of Myc-INO80 could not fully complement the function of endogenous INO80 due to an abnormal expression level or the presence of the Myc tag. The race tube assays showed that the conidiation processes of the ino80^KO and ies-1^KO strains are entrained by LD cycles, excluding the defect in the input pathway of clock in these mutants (S4E Fig). Notably, the growth rates of ino80^KO strains were strongly dependent on glucose, indicating that ino80^KO strain might be sensitive to glucose or carbon starvation. Bioluminescence rhythms in ino80^KO, frq-luc and ies-1^KO, frq-luc strains were also abolished (S4F, S3B and S3C Figs). Western blots showed disruption of both the FRQ rhythms and FRQ phosphorylation profiles in the ino80^KO and ies-1^KO strains (S4G Fig). Northern blot analyses showed that the abolished FRQ rhythms were due to constant transcription of the frq gene in the ino80^KO and ies-1^KO strains (S4H Fig). Altogether, these results indicate that the INO80 complex is critical for frq transcription repression.

Previous studies have shown that the INO80 complex can be recruited to target genes by Iec1 or YY1 in yeast or mammalian cells [39,40]. To investigate the occupancy of the INO80 complex at the C-box of the frq gene, we performed a ChIP assay using INO80-specific antiserum at DD18 (Fig 3A). The ChIP results revealed that binding of INO80 to the frq gene peaks at the C-box and 3’-UTR (Fig 3B), similar to the binding pattern of IEC-1 (Fig 2D). Next, we assessed whether IEC-1 is required for recruitment of the INO80 complex at the C-box of the frq promoter. The ChIP assays conducted with an INO80 antibody showed a dramatic decrease in enrichment of INO80 at the C-box in iec-1^KO strains (Fig 3C), indicating that efficient binding of the INO80 complex at the frq C-box is dependent on the expression of IEC-1.

To determine whether the association of INO80 with the C-box is rhythmic during circadian cycles, the recruitment of INO80 at four different time points in constant darkness were examined using a ChIP assay. The results demonstrated that the association of INO80 with the C-box of the frq promoter was rhythmic, peaking at DD18 (Fig 3C). In yeast or mammalian systems, recruitment of the INO80 complex by Iec1 or YY1 is a key event in gene expression [39,40]. Thus, these results suggest that IEC-1-dependent INO80 recruitment at the frq promoter is required for suppression of frq transcription after WCC-inactivation by FRQ. In the Neurospora circadian system, the transcriptional activators WC-1 and WC-2 are responsible for the transcriptional activation of the frq gene during circadian cycles [7,9–11,13,46,47]. In the frq⁹ mutant, constant frq transcription [48] causes a low H3 density at the frq promoter, whereas, in the wc-2^KO strain, a lack of frq transcription [10] is associated with a high H3 density at the frq promoter (Fig 3D). These findings indicate that WC-mediated transcriptional activation leads to open chromatin states by nucleosomal removal, which increases the accessibility of the frq promoter. It is also possible that WC-mediated transcriptional activation of frq is required for rhythmic recruitment of the INO80 complex during the period of high expression of frq to prepare for repression of frq transcription. To assess this possibility, we examined enrichment of the INO80 complex at the frq C-box in the wc-2^KO and frq⁹ strains. The ChIP results showed that the levels of INO80 enrichment in the wc-2 mutant were lower than that in the wild-type strain at the DD18 time point (Fig 3E). These data suggest that the rhythmic recruitment of the INO80 complex is dependent on WC-mediated transcriptional activation of frq. In contrast, in frq⁹ strains, the constant elevation of the levels of INO80 enrichment are due to the constant high elevation of frq expression, which is caused by malfunction of the negative feedback loop (Fig 3E). To verify that the high levels of recruitment of the INO80 complex in the frq⁹ mutant are due to the activity of the WC complex, we generated the wc-2^KOfrq⁹ double-mutant strain, and examined the recruitment of INO80 at the C-box. Loss of WC-2 restored the low levels of recruitment of INO80 at the C-box in the frq⁹ mutant (Fig 3E). Meanwhile, the INO80 expression of these mutants was similar to that of the wild-type strain (Fig 3F). These data indicate that the highly activated transcription of frq by the WC complex also contributes to the rhythmic recruitment of the INO80 complex to the C-box of frq for subsequent suppression of frq transcription during circadian cycles.

In the Neurospora circadian system, the WC complex, the positive transcription factor that binds to the C-box of frq, is responsible for rhythmic frq expression [7–9,13]. To test whether frq transcription is driven by the WC complex in ino80 and iec-1 mutants, we first examined the transcriptional activity of WC-2 in the ino80^KO and iec-1^KO strains. The ChIP results revealed that the rhythmic enrichment of WC-2 at the frq C-box corresponded to the rhythmic expression of frq in the wild type strain (Fig 4A). In contrast, the recruitment of WC-2 was dramatically decreased in the ino80^KO and iec-1^KO strains (Fig 4A), which was inconsistent with the constant expression of frq in these mutants. This finding suggests that the transcriptional activities of WC-1 and WC-2 are decreased in the ino80 and iec-1 mutants. Previous studies showed that hypophosphorylated WC-1 and WC-2 efficiently bind to the C-box for frq transcriptional activation [12,21] while hyperphosphorylated WC-1 and WC-2 exhibit lower binding activity at the C-box of frq [14,49]. Western blot analysis showed that the phosphorylation levels of WC-1 and WC-2 were increased in the ino80^KO, ies-1^KO and iec-1^KO strains compared to those in the wild-type strains at different time points in DD (Fig 4B and 4D). In addition, we found no significant changes in the protein levels of WC-1 and a slight decrease in the levels of WC-2 in the ino80^KO, ies-1^KO and iec-1^KO strains compared to those in the wild-type strains (Fig 4C and 4E). The levels of WC-1 protein did not exhibit a robust circadian rhythm in our hands, which is consistent with previous studies showing that the amplitudes of WC rhythms are variable. Since WC activity is mostly known to be regulated by phosphorylation, the variability of the WC-1 rhythms might be due to the use of different WC-1 antibodies in different laboratories which may have different sensitivity to different isoforms of WC-1. These results indicate the increased frq transcripts in these mutants are not caused by the increase of transcriptional activity of WCC. These data also suggest that the high levels of frq transcription could be partially independent of WC expression in these mutants. To further confirm this possibility, we generated an ies-1^KOwc-1^RIP double mutant and compared its FRQ levels to that of a wc-1 single mutant. Consistent with previous results, the FRQ levels were extremely low in the wc-1 single mutant (Fig 4F) [10]. However, the levels of FRQ protein and frq mRNA in the ies-1^KOwc-1^RIP double mutant were well detected with Western blot or Northern blot analyses (Fig 4F and 4G), indicating that WC-independent frq transcription exists in the ies-1 mutants. Altogether, these results demonstrate that recruitment of the INO80 complex at the C-box through high transcriptional activation prepares for suppression of frq transcription after WCC inactivation.

In S. cerevisiae, the first four nucleosomes at the transcription start site of genes have critical roles in the process of transcription initiation [50]. In Drosophila, the +1 nucleosome strongly inhibits the normal function of RNA polymerase II [51]. Although the INO80 complex occupied and peaked at the boundary of the genes, genome-wide ultra-high-resolution ChIP-exo data showed that the Arp5 subunit of the INO80 complex was particularly enriched at the +1 position in yeast [36,52]. Antisense transcripts qrf in the Neurospora circadian system revealed that induction of qrf promotes frq gene expression by creating a more accessible local chromatin environment, even in the absence of the WC complex [28]. Given that qrf-induced frq transcripts share the same properties with the WC-independent frq transcripts, the chromatin states in these two pathways should be similar. The ChIP assay showed that the H3 levels were dramatically reduced at both the C-box and TSS in the ino80^KO and ies-1^KO strains compared to that of the wild-type strains (Fig 5A). However, the H3 levels at the frq ORF in mutants were similar to that of the wild-type strains (Fig 5A). These results indicate that nucleosome density is decreased at the C-box and TSS of the frq gene in mutants.

SET-2-mediated H3K36me3 is an important landmark on chromatin during transcription elongation [53]. To confirm the enhanced transcription of frq induced by the decreased nucleosomal barrier in the ino80^KO strains, a ChIP assay was performed to examine the recruitment of SET-2 and enrichment of H3K36me3 at the 3’ORF of the frq gene. As expected, the recruitment of SET-2 and the enrichment of H3K36me3 were significantly increased at the 3’ ORF of the frq gene in the ino80^KO strains compared to those of the wild-type strains (Fig 5B). Together, these results demonstrate that the INO80 complex contributes to establishment of the dense chromatin environment at the frq promoter, which is essential for suppressing WC-independent frq transcription.

The wild-type strain (4200) was used as a control. The iec-1 or ies-1 genes were deleted by the replacement of their ORFs with a hygromycin resistance gene (hph) on the ku70^RIP (bd, a) background strain. The ku70^RIPiec-1^KO and ku70^RIPies-1^KO strains were crossed with the 774–10 (A, his-3^–) strain to obtain homokaryotic iec-1^KO and ies-1^KO strains, respectively. The ku70::bar ino80^KO strain was generated by the replacement of its ORF with the hygromycin resistance gene (hph) on the ku70::bar background strain and microconidia purification. The ku70::bar ras-1^bdino80^KO and ras-1^bdino80^KO strains were generated in the same manner as the ku70::bar ino80^KO strain. For rescue strains, the plasmid qa-5Myc-6his-IEC-1 was transformed into the iec-1^KO strain. The ino80^KO, qa-5Myc-6his-INO80 and ies-1^KO, qa-5Myc-6his-IES-1 transformants were obtained in the same manner as the iec-1^KO, qa-5Myc-6his-IEC-1 transformants. A plasmid containing the full-length frq promoter fused to luciferase was transformed into the iec-1^KO, ino80^KO and ies-1^KO strains to generate the iec-1^KO, frq-luc, ino80^KO, frq-luc and ies-1^KO, frq-luc strains. Liquid culture conditions were the same as a previously published method [56]. The wc-2^KO, frq⁹, wc-2^KOfrq⁹, wc-1^RIP and ies-1^KOwc-1^RIP all contain the band mutation.

The race tube medium contained 1× Vogel’s salts, 0.1% glucose, 0.17% arginine, 50 ng/mL biotin and 1.5% agar supplemented with or without 10 mM H₂O₂. Conidia of different strains were inoculated at one end of each race tube and were grown under constant light (LL) for 1 day to synchronize the clock. The race tubes were then transferred to constant darkness (DD), and the position of the advancing mycelia front was marked at 24 h intervals on the tube. When growth was completed, tubes were scanned, and the growth period of each strain was calculated.

GST-INO80 (containing INO80 amino acids 1–341) and GST-IEC-1 (containing IEC-1 amino acids 9–175) fusion proteins were expressed in BL21 cells by induction of IPTG. After purification, the recombinant proteins were used as antigens to immunize rabbits, which yielded rabbit polyclonal antiserums [57].

ChIP assay was performed as previously described [55]. Neurospora tissues were fixed by shaking in 1% formaldehyde for 15 min at 25°C, and cross-linking reactions were stopped by adding glycine at a final concentration of 125 mM. The cross-linked chromatin was sheared by sonication to approximately 200–500 bp fragments. A 1 mL aliquot of protein (2 mg/mL) was used per immunoprecipitation, and 10 μL was maintained as the input DNA. The chromatin immunoprecipitation reaction was carried out with 2 μL antibody to WC-2, 2.5 μL antibody to H3 (2650; Cell Signaling Technology), 2.5 μL antibody to IEC-1, 5 μL antibody to INO80, 5 μL antibody to SET-2, and 2 μL antibody to H3K36me3 (4909; Cell Signaling Technology). Immunoprecipitated DNA was quantified using real-time PCR. The primers for real-time PCR were designed according to a previously published protocol [27]. The ChIP-qPCR data were normalized by the input DNA, and the results were presented as the percentage of input DNA. Each experiment was independently performed at least three times.

Protein extraction, quantification and western blot analysis were performed as previously described [58,59]. Western blot analyses were performed by using antibodies against the proteins of interest. Equal amounts of total protein (40 μg) were loaded in each lane. After electrophoresis, proteins were transferred onto PVDF membranes, and western blot analysis was performed.

RNA was extracted by TRIzol [60] and analyzed by northern blotting as previously reported [56]. Shortly, equal amounts of total RNA (20 μg) were loaded onto agarose gels for electrophoresis, and the gels were blotted and probed with an RNA probe specific for frq mRNA.

The luciferase reporter assay was performed as previously reported [54]. The bioluminescence reporter construct (frq-luc), in which luciferase expression is driven by the frq promoter, was introduced into the his-3 locus of the iec-1^KO, iec-1^KO, ino80^KO and wild-type strains. One drop of conidia suspensions in water was placed on AFV medium and grown in constant light (LL) overnight at 25°C. The cultures were then transferred to constant darkness, and luminescence was recorded in real time using LumiCycle after one day in DD at 25°C. The data were then normalized with LumiCycle analysis software by subtracting the baseline luciferase signal, which increases as the cells grow. Under our experimental condition, luciferase signals are highly variable during the first day in the LumiCycle but become stabilized afterwards, which is likely due to the light-dark transfer of the cultures. Thus, the results were recorded after one day in DD at 25°C.