A circadian clock translational control mechanism targets specific mRNAs to cytoplasmic messenger ribonucleoprotein granules

SUMMARY

Phosphorylation of Neurospora crassa eukaryotic initiation factor 2 α (eIF2α), a conserved translation initiation factor, is clock controlled. To determine the impact of rhythmic eIF2α phosphorylation on translation, we performed temporal ribosome profiling and RNA sequencing (RNA-seq) in wild-type (WT), clock mutant Δfrq, eIF2α kinase mutant Δcpc-3, and constitutively active cpc-3^c cells. About 14% of mRNAs are rhythmically translated in WT cells, and translation rhythms for ~30% of these mRNAs, which we named circadian translation-initiation-controlled genes (cTICs), are dependent on the clock and CPC-3. Most cTICs are expressed from arrhythmic mRNAs and contain a P-body (PB) localization motif in their 5′ leader sequence. Deletion of SNR-1, a component of cytoplasmic messenger ribonucleoprotein granules (cmRNPgs) that include PBs and stress granules (SGs), and the PB motif on one of the cTIC mRNAs, zip-1, significantly alters zip-1 rhythmic translation. These results reveal that the clock regulates rhythmic translation of specific mRNAs through rhythmic eIF2α activity and cmRNPg metabolism.

Graphical Abstract

In brief

Castillo et al. show that high daytime levels of phosphorylated translation initiation factor eIF2α establish a circadian post-transcriptional mRNA localization control mechanism that targets specific mRNAs to ribonucleoprotein granules (cmRNPgs) and removes them from the translating cytoplasmic mRNA pool in the daytime.

INTRODUCTION

Organisms are metabolically, physiologically, and behaviorally different at different times of the day as a result of the widespread impact of the circadian clock on rhythmic gene expression.¹ The effect of this regulation is evident in the prevalence of clock-related illnesses in humans, including sleep disorders, psychiatric illness, metabolic syndrome, and cancer.^2–5 Furthermore, over half of the top-selling drugs in the USA target a clock-controlled protein,⁶ underscoring the need to know how, and what, is controlled by the clock. While control of transcription is a major point of regulation by the clock, proteomic analysis of rhythmically accumulating soluble proteins in mouse liver,^7–10 and in protein levels in the clock model organism Neurospora crassa,¹¹ showed that up to 50% of proteins with robust rhythms in abundance had no apparent rhythms in the associated mRNA levels. In addition, ribosome profiling in the mouse liver identified significant numbers of genes with oscillations in ribosome occupancy from constitutively expressed mRNAs.^12,13 These data revealed that rhythms in mRNA levels are not always a predictor of protein rhythms and activity and point to key roles for the clock in post-transcriptional, translational, and/or post-translational mechanisms in generating rhythmic protein levels.^14–17

In N. crassa, the circadian oscillator is comprised of a molecular feedback loop with negative elements FREQUENCY (FRQ), FRQ-INTERACTING RNA HELICASE (FRH), and CASEIN KI NASE I (CKI) inhibiting the activity of the positive elements WHITE COLLAR-1 (WC-1) and WC-2.^1,18 WC-1 and WC-2 heterodimerize to form the White Collar complex (WCC), which activates transcription of the frq gene and downstream target genes important for overt rhythmicity.^19,20 These target genes encode transcription factors, as well as various output pathway components and terminal clock-controlled genes (ccgs).¹⁹ In addition, the N. crassa clock regulates mRNA translation by controlling rhythms in the activity of eukaryotic initiation factor 2 (eIF2) and eukaryotic elongation factor-2 (eEF-2).^21,22 However, not all N. crassa mRNAs are subject to rhythmic translation, and it is unclear how translation of select mRNAs is regulated by rhythmic eIF2 and eEF-2 activities. Because translation is generally limited by initiation and not elongation,^23–25 we focused this study on understanding the mechanism of selection of mRNAs for clock-controlled translation initiation.

A critical and conserved translation initiation control mechanism involves phosphorylation of the α subunit of eIF2 (eIF2α). When eIF2α activity is reduced by phosphorylation (P-eIF2α), translation initiation is inhibited.^26,27 In mammalian cells, eIF2α can be phosphorylated by four different serine/threonine kinases (GCN2, HRI, PERK, and PKR) to induce a program called the integrated stress response (ISR), which coordinates cellular adaptation to extracellular and intracellular stresses.²⁸ Among these kinases, GCN2 is conserved in fungi and mammals.²⁹ In Saccharomyces cerevisiae, GCN2 kinase is activated by conditions that lead to amino acid starvation and other stresses, which result in accumulation of uncharged tRNAs.³⁰ Moreover, activation of GCN2 requires the trans-acting positive effector protein GCN1 that facilitates delivery of uncharged tRNAs to GCN2.³¹ Similarly, N. crassa CPC-3 (the homolog of yeast and mammalian GCN2) activation requires uncharged tRNAs and GCN1.²²

When N. crassa cultures are transferred from constant light (LL) to constant dark (DD) to synchronize the circadian clock in all cells to dusk, a robust circadian rhythm is sustained in the DD cultures. Previous studies showed that at least 30% of N. crassa eIF2α is phosphorylated under the control of the clock when cells are grown in DD, with inhibitory P-eIF2α levels peaking during the subjective day and active eIF2α levels peaking during the subjective night.²² In the clock mutant Δfrq, P-eIF2α levels were arrhythmic, while in the constitutively active kinase mutant cpc-3^c, P-eIF2α levels were arrhythmic and high compared with wild type (WT).²² No P-eIF2α was detected in Δcpc-3 cells grown in DD, demonstrating that CPC-3 is necessary for clock-controlled eIF2α phosphorylation.²²

Clock control of the activity of eIF2α has also been observed in the mouse SCN, the site of the master circadian clock, and this regulation has been implicated in linking ISR and circadian physiology.³² Thus, understanding the mechanisms of circadian clock control of eIF2α activity, and its effect on rhythmic mRNA translation, is critical to understanding the molecular programs that drive rhythmic physiology and behavior in health and disease.

In response to stress and high P-eIF2α levels, translation of some mRNAs is inhibited, and affected mRNAs are sequestered in cytoplasmic mRNA ribonucleoprotein granules (cmRNPgs), including P-bodies (PBs) and stress granules (SGs).^33–36 PBs and SGs are dynamic complexes that temporarily store translationally repressed mRNAs that may later reenter translation.³⁷ PBs are usually associated with mRNA decay machinery, while SGs assemble with translation initiation components.³⁸ Certain conditions promote PB and SG fusion or interactions.³⁸ Thus, it is not surprising that studies in S. cerevisiae and mammals found that many proteins are shared between PBs and SGs.^35,39,40 These features, along with limited studies of PBs and SGs in N. crassa, make it difficult to distinguish between the two structures. Therefore, we refer to PBs and/or SGs as cmRNPgs. In mammalian cells, cmRNPg formation is clock regulated,¹² opening the possibility that cmRNPg formation in N. crassa, and targeting of mRNAs to cmRNPgs, represents a conserved mechanism to control rhythmic translation.

To determine the extent and mechanism of selection of mRNAs for rhythmic translation in N. crassa, ribosome profiling (ribo-seq) was performed in parallel with RNA sequencing (RNA-seq) in WT, Δfrq, Δcpc-3, and cpc-3^c cells. We discovered that ~14% of N. crassa mRNAs are rhythmically translated in WT cells, and translation rhythms for ~30% of these mRNAs were dependent on both the clock and CPC-3. We refer to the genes encoding these mRNAs as circadian translation-initiation-controlled genes (cTICs). Motif enrichment analysis of the 5′ leaders of cTICs identified a CA-rich motif, predicted to be a cytoplasmic PB motif localization sequence. We confirmed that cTIC zip-1 had arrhythmic transcription and rhythmic protein levels that peaked at subjective night in WT cells and that rhythmic ZIP-1 protein levels were dependent on the clock and rhythmic P-eIF2α levels. Moreover, ZIP-1 translation rhythms were abolished in a cmRNPg component knockout, Δsnr-1, and were severely damped when the CA-rich motif in the mRNA 5′ leader was deleted. These data support a model whereby high levels of P-eIF2α during the day, under control of the circadian clock, establish a post-transcriptional localization control of translation mechanism that targets specific mRNAs to cmRNPgs to remove them from the translation pool.

RESULTS

Time-resolved ribosome profiling

Phosphorylation of eIF2α is clock controlled in N. crassa, peaking during the subjective day, and inhibitory P-eIF2α levels led to reduced translation during the day in vitro.²² To determine the extent of clock regulation of translation in vivo, and the role of cycling P-eIF2α levels in this regulation, ribo-seq, in parallel with RNA-seq, was performed on WT cells, clock-defective Δfrq cells, and cells with deleted (Δcpc-3) or constitutively active (cpc-3^c) CPC-3. Duplicate cultures of these strains were synchronized by a light-to-dark transfer (LL to DD), grown in DD, and harvested with a 4 h resolution (Figure S1A). Consistent with previous data, FRQ protein was rhythmic in WT, Δcpc-3, and cpc-3^c cells, confirming that a functional clock was present in these cells²² (Figures S1B, S1D, and S1E). P-eIF2α levels were rhythmic in the same WT protein extracts (Figure S1C) and arrhythmic in the same cpc-3^c protein extracts (Figure S1F). These cells were used to generate ribo-seq and RNA-seq libraries (Figure S1A), and the sequencing data were processed as outlined in Figure S1G. Mapping summaries are reported in Tables S1 and S2 for ribo-seq and RNA-seq data, respectively. The average correlation coefficients (R²) between two biological replicates for each strain at all time points using FPKM values were 0.9382 for ribo-seq and 0.9580 for RNA-seq datasets (Table S3).

To confirm that the sequenced ribosome-protected footprint reads (RPFs) were generated through the authentic protection of mRNAs by ribosomes, the footprints mapped to the coding regions of the genome were examined. As expected, the largest fraction of the average reads (90%) mapped to the protein-coding regions (coding sequence [CDS]), while only 6% and 4% mapped to the 5′ UTRs and 3′ UTRs, respectively, in WT, Δfrq, Δcpc-3, and cpc-3^c cells (Figure 1A). Typical RPF lengths are between 28 and 32 nt, but the most populated read lengths can vary between organisms, cell types, or experimental conditions.⁴¹ In N. crassa, the distribution of the read lengths, as determined using the NGS toolkit Plastid,⁴² was centered between 30 and 31 nt (Figure 1B). The physical process of ribosome movement creates triplet periodicity and allows one to infer the reading frame in which a coding region is decoded when the read alignments are mapped to their P-site offsets.⁴² The most highly phased population of reads, the 31-mers, were examined to determine reading frame preference. Frame analysis revealed a strong preference for the first reading frame (phase 1) for all strains (Figure 1C). Taken together, using a ribo-seq protocol that was optimized for N. crassa resulted in footprints that (1) largely mapped to the CDS, (2) exhibited the expected size of ribosome footprint, and (3) showed a phasing preference for the predicted open reading frame (ORF).

Figure 1. Analysis of N. crassa WT, Δfrq, Δcpc-3, and cpc-3^c ribo-seq data.

(A) Fraction of total reads mapping to the coding sequence (CDS) or 5′ or 3′ untranslated regions (5′ UTRs and-3′ UTRs) for the indicated strains.

(B) Insert size distribution of ribosome-protected footprint (RPF) across the two biological replicates and time points for the indicated strains.

(C) Frame analysis for RPF for the most highly phased population of reads, 31-mers. The RPF are color coded (dark, medium, light) according to the subcodon position alignments (phases 1, 2, and 3) for WT (black), Δfrq (blue), Δcpc-3 (red), and cpc-3^c (green).

(D) Read distribution of normalized RPF (left, black) and RNA abundance (right, gray) along the frq transcript from cells grown in constant dark (DD) and harvested at the indicated times (h).

(E) Plot of the normalized RPF (black, left y axis) and RNA abundance (FPKM) (gray, right y axis) reads of frq mRNA from (D).

Square symbols indicate the values from biological replicate 1 (black) and replicate 2 (gray). The bar at the bottom of the graph represents subjective day (gray) and subjective night (black) in this and all subsequent figures.

Ribosome occupancy of the core clock genes frq, wc-1, and wc-2 and some ccgs was examined. The normalized RPF and mRNA abundance of frq was rhythmic, with frq mRNA levels peaking at DD32 and FRQ ribosome occupancy peaking ~4 h later (DD36) (Figures 1D and 1E), consistent with the reported lag between peak mRNA and protein levels.⁴³ The genes encoding wc-1 and wc-2 were previously found to have rhythmic promoter activity.⁴⁴ In our datasets, wc-1 and wc-2 mRNA and ribosome occupancy levels were rhythmic (Figures S2A and S2B), peaking during early subjective night. Consistent with previous observations, the peak in WC-1 and WC-2 ribosome occupancy occured ~8 h after the peak in FRQ ribosome occupancy.⁴⁵ The 4 h advanced peak of RPF reads compared with the peak in mRNA levels for wc-1 and wc-2 likely reflects normal variation observed with 4 h sampling but could reflect an unknown regulatory mechanism. Several ccgs including ccg-1 (NCU03753), ccg-13 (NCU08907), and ccg-15 (NCU08936) also had rhythmic mRNA levels and ribosome occupancy (Figures S2C–S2E). An additional ccg, os-4, encoding the MAPK kinase (MAPKK) of the OS-4 pathway, was examined.⁴⁶ WCC binds rhythmically to the promoter of os-4 and drives daily rhythms in os-4 mRNA and protein levels. In these datasets, os-4 mRNA accumulation and ribosome occupancy peaked around the same time, DD32 (Figure S2F), in agreement with published data.⁴⁶ Taken together, these results established that the mRNA abundance and ribosome footprint measurements can be applied to determine rhythmic translation across the N. crassa genome.

Clock-controlled eIF2α activity is required for rhythmic translation of specific mRNAs

Rhythms in mRNA abundance and ribosome occupancy were determined using the Extended Circadian Harmonic Oscillator (ECHO) application.⁴⁷ In Δcpc-3 cells deleted for the eIF2α kinase, 640 of the 1,328 clock-controlled translatome became arrhythmic (48%), while 61% (N = 808) were arrhythmic in the constitutive kinase mutant cpc-3^c (Table S4). There were 631 mRNAs and 426 mRNAs that remained rhythmically translated under the control of the clock in Δcpc-3 and cpc-3^c cells, respectively (Figure 2A, left panel). While we expected most of the genes controlled by eIF2α activity to have overall increased translation in Δcpc-3 cells and decreased translation in constitutively active cpc-3^c cells, the data were more complex (Table S5). When RPF reads were averaged across all time points for genes that were no longer rhythmically translated in the mutants and compared with WT, relative read levels were unchanged for most genes; 76% in Δcpc-3 and 78% in cpc-3^c cells compared with WT. In Δcpc-3 cells, 11% had high and 13% had low reads compared with WT, and in cpc-3^c cells, 13% had high and 9% had low reads compared with WT (p < 0.05, fold change > 1.2). In addition, there was little overlap (14%) between genes that had low reads in Δcpc-3 and high reads in cpc-3^c cells (Table S5). Taken together, these data support that other regulatory mechanisms exist to control translation rhythms and levels in N. crassa cells, including, for example, clock control of translation elongation.²¹

Figure 2. A subset of mRNAs requires P-eIF2α for rhythmic translation.

(A) Venn diagrams showing the number of transcripts with rhythmic RPF counts in WT, Δcpc-3, and cpc-3^c cells and arrhythmic RPF counts in Δfrq cells (left) or with rhythmic RPF counts in WT cells and arrhythmic RPF counts in Δfrq, Δcpc-3, and cpc-3^c cells (right).

(B) Heatmaps of the peak phase of genes with rhythmic RPF counts in WT cells and arrhythmic RPF counts in Δfrq, Δcpc-3, and cpc-3^c cells (cTICs, N = 404) grown in DD and harvested at the indicated times (h). Genes are sorted by the peak phase in WT.

(C) Phase distribution based on maximal RPF counts in WT for the 404 cTICs. The concentric circles emanate from zero at the center to increasing frequencies as indicated by the numeric legends. The numbers indicate circadian time (CT), with white shading designating subjective day (CT0–12) and gray shading designating subjective night (CT12–0) in this and all subsequent figures.

(D) GO terms under the biological process category that are significantly enriched (p < 0.05) in rhythmically translated mRNAs that peak during the day (gray bars), and during the night (black bars). For visualization, the p values are plotted as −log₁₀.

See also Tables S4, S5, and S6.

To identify genes whose rhythmic translation is dependent on rhythmic P-eIF2α levels, we focused on the overlapping mRNAs that became arrhythmically translated in both Δcpc-3 and cpc-3^c cells (Figure 2A, right panel). These 404 genes were referred to as cTICs. The relative RPF peak phases of the cTICs over a circadian time course are shown in heatmaps, where the transcripts are ordered by the phase in circadian time (CT) of RPF oscillation in WT cells (Figure 2B). CT represents biological time normalized to 24 circadian hours per cycle in strains or organisms with varying free running periods in constant conditions.

In WT cells, inhibitory P-eIF2α levels peak in the subjective day and are low during the subjective night.²² Thus, cTICs that peak in translation at night are likely sensitive to high P-eIF2α levels and, therefore, have reduced translation during the subjective day. In contrast, cTICs that peak in translation during the day likely use initiation mechanisms that require high P-eIF2α levels for more efficient daytime translation, similar to what has been observed for GCN4.⁴⁸ The phase distribution of the cTICs revealed slightly more day peaking (N = 215) than night peaking (N = 189) cTICs (Figure 2C). These data revealed that a subset of mRNAs are preferentially translated when P-eIF2α levels peak.

Physiological processes impacted by clock regulation of translation by P-eIF2α were examined by performing Gene Ontology (GO) enrichment analyses for biological processes on all daypeaking or all night-peaking cTICs (Figure 2D; Table S6). GO enrichment for mRNAs involved in amino acid biosynthesis and redox stress response peaked during early subjective day, and translation initiation components peaked during late subjective day. During the subjective night, the cTICs are enriched in processes involving mRNA processing and protein glycosylation (Figure 2D; Table S6). Taken together, these data support that clock regulation of P-eIF2α through CPC-3 results in rhythmic translation of specific mRNAs rather than affecting global rhythmic translation.

Genes that depend on clock-controlled eIF2α activity are rhythmically translated from cycling or non-cycling transcripts

To determine the nature of mRNAs that are subjected to rhythmic translation through eIF2α activity, the mRNA profiles of the 404 cTICs were analyzed and divided into three distinct classes: class I, genes whose mRNA and ribosome footprint levels were rhythmic with a phase difference of less than the 4 h sampling time between peak mRNA and ribosome occupancy (24%); class II, genes whose mRNA and ribosome footprint levels were rhythmic with a phase difference of more than 4 h between peak mRNA and ribosome occupancy (18%); and class III, genes rhythmically translated from arrhythmic mRNAs (58%) (Figure 3A; Table S4). For class I cTICs, most of the genes peaked in translation during the night (72%). For class II cTICs, an average phase delay of 10 h was observed for 56 (80%) of class II cTICs, with most (75%) peaking in translation during the day (Figure 3B; Table S4). cTICs in all 3 classes had mRNA levels (averaged across all time points) that either increased, decreased, or stayed the same in both Δcpc-3 and cpc-3^c cells compared with WT cells (Table S5). These changes in mRNA levels are likely due to alterations in the levels or activities of regulatory components that affect mRNA transcription or stability in the mutants.

Figure 3. Genes that depend on clock-controlled eIF2α activity for rhythmic translation arise from cycling and non-cycling transcripts.

(A) Heatmaps of the peak phase of genes with rhythmic RPF counts in WT cells (right panel) and their corresponding mRNA expression (FPKM) profiles (left panel). The 404 cTICs are sorted by the peak phase of WT RPF count for each class. mRNA abundances and RPF levels are standardized within each gene (row) and independently for RNA-seq and ribo-seq columns (Z scores).

(B) Phase distribution based on maximal RPF counts in WT for genes belonging to each class on the left panel: class I, in-phase rhythmic RPF and mRNA, class II, rhythmic RPF and mRNA with phase changes, and class III, rhythmic RPF and arrhythmic mRNAs.

(C) Luciferase activity from HAM-7:LUC translational (black line) and Pham-7::luc transcriptional (gray line) fusions in WT cells grown in DD and recorded every 90 min over 4 days (h DD). The average bioluminescence signal is plotted (mean ± SEM, n = 12 technical replicates). The peak phase values (φ) for each trace are shown.

(D) Luciferase activity from CPC-1:LUC translational (black line) and Pcpc-1::luc transcriptional (gray line) fusions in WT cells grown in DD and recorded every 90 min over 4 days (h DD). The average bioluminescence signal is plotted (mean ± SEM, n = 12 technical replicates), and the phase values are shown as in (C).

See also Table S4.

Rhythmicity of class I gene hyphal anastomosis 7 (ham-7; NCU00881) and class II gene cpc-1 mRNA levels and HAM-7 and CPC-1 protein levels were examined in independent experiments to validate these results. HAM-7 is a GPI-anchored cell wall protein required for development and cell wall stress responses.⁴⁹ In WT cells, a ham-7 promoter luc fusion (Pham-7::luc, a transcriptional reporter) and HAM-7:LUC translational reporter were rhythmic in DD, peaking during the subjective night, with no significant phase difference between the mRNA and protein levels (Figure 3C). In WT cells, the Pcpc-1::luc transcriptional reporter and CPC-1:LUC translational reporter were rhythmic in DD, with CPC-1:LUC peaking during the late subjective day with a phase delay relative to the promoter activity, consistent with the phase difference observed in the genomic datasets (Figure 3D).

More than half of the genes that depend on clock-controlled eIF2 activity for rhythmic translation did not have rhythmic mRNA levels. Of these class III genes, 134 (57%) mRNAs peaked in translation during the day and 100 (43%) peaked during the night (Figure 3B; Table S4). Subsequent analyses focused on class III cTICs, as genes in this category appear to undergo distinct post-transcriptional control mechanisms to generate rhythms in translation from arrhythmic mRNAs.

Functional significance of class III cTICs

A functional enrichment analysis was performed on class III cTICs that peak in translation at different times of the day (Figure S3). Genes that are translated during early to mid-morning (CT0–4) are enriched for carbohydrate metabolism and signaling pathways involving GTP or ATP binding. During late afternoon (CT5–10), genes that function in mRNA splicing and processing were enriched, suggesting that the clock influences other post-transcriptional processes through rhythmic eIF2α activity. Genes involved in translation also peaked near the late afternoon. During the night (CT16–22), genes involved in protein modification, processing, and transport were enriched. Stress response genes were observed throughout the day. These results recapitulate previous findings that showed partitioning of gene products involved in catabolic processes during the day, and protein production and processing during the night⁵⁰ under control of the circadian clock, and support that constitutive expression of their mRNAs allows rapid responses to acute environmental and intracellular stresses through translation regulation.

Class III cTICs are enriched for PB localization signals, and the PB signal is required for robust rhythmic translation of zip-1 mRNA

Although there are several different mechanisms that may contribute to cTIC translation regulation, we focused on identifying possible cis elements in the 5′ leader using de novo motif analysis (MEME Suite v.5.3.0).⁵¹ Two highly represented sequences were identified in the class III cTICs, a CA-rich motif and a CU-rich motif (Figure 4A). These motifs were subjected to GOMo analysis⁵² to identify associated GO terms; however, no significant identities were found for either motif.

Figure 4. Clock control of P-eIF2α levels and cmRNPg sequestration are required for rhythmic translation of zip-1.

(A) Highly represented sequences identified by MEME 5.3.0 from the 5′ leaders of class III cTICs. (B–E) Luciferase activity from ZIP-1:LUC translational (black line) and Pzip-1::luc transcriptional (gray line) fusions in (B) WT cells and ZIP-1:LUC in (C) Δfrq (blue line), (D) Δcpc-3 (red line), and (E) cpc-3^c (green line) cells grown in DD and recorded every 90 min over 5 days (h DD). The data were normalized to the mean, and the average bioluminescence signal is plotted (mean ± SEM, n = 12 technical replicates). ZIP-1:LUC in WT cells was rhythmic as indicated by a better fit to a sine wave (dotted black lines, p < 0.05). ZIP-1:LUC in Δfrq, Δcpc-3, and cpc-3^c cells was arrhythmic as indicated by a better fit of the data to a line (dotted blue, red, and green lines, p > 0.05) by F-test.

See STAR Methods for statistical test for rhythmicity and Table S7.

To refine the motif analyses, we focused on class III cTIC mRNAs with significantly increased translation in Δcpc-3 cells that lack inhibitory P-eIF2α compared with WT and cpc-3^c as these cTICs were predicted to be the most highly impacted by rhythmic P-eIF2α levels. From this analysis, 67 class III cTICs (class III*) met these criteria, with 33 peaking at night and 34 peaking during the day (Table S7). An enriched motif was identified at various locations within the 5′ leader in 62/67 cTICs (Figure 4A; Table S7). Analysis of this motif using GoMo⁵² returned a significant enrichment (p = 2.0 × 10⁻⁵) for cytoplasmic mRNA processing body localization (GO: 0000932) (Figure 4A).

To determine if an association of cTIC mRNAs with cmRNPgs is necessary for rhythmic translational repression, we focused on examining zip-1 (NCU08055, idi-4 homolog in Podospora anserina), a gene with significantly higher ribo-seq RPF coverage in Δcpc-3 cells, and reduced ribo-seq RPF coverage in cpc-3^c cells compared with WT based on IGV read coverage (Figure S4). idi-4 in the filamentous fungus P. anserina encodes a putative basic leucine zipper (bZIP) transcription factor that is induced during heterokaryon incompatibility, a cell death reaction that occurs when hyphae of unlike genotypes fuse.⁵³ The levels of zip-1 mRNA and ZIP-1 protein were examined in WT, Δfrq, Δcpc-3, and cpc-3^c cells. Consistent with the ribo-seq and RNA-seq data, the ZIP-1:LUC translational reporter was rhythmic in WT cells in DD, peaking during subjective night when P-eIF2α levels are low, whereas a zip-1 promoter luc fusion, Pzip-1::luc, was arrhythmic (Figure 4B). ZIP-1:LUC translation was arrhythmic in Δfrq (Figure 4C), Δcpc-3 (Figure 4D), and cpc-3^c cells (Figure 4E). These data confirmed that zip-1 rhythmic translation is dependent on the clock and rhythmic eIF2α activity.

A PB motif resides in the zip-1 5′ leader (Table S7). To determine if rhythmic translation of zip-1 mRNA requires sequestration to cmRNPgs, ZIP-1:LUC translation was assayed in cells deleted for a cmRNPg component. In human and yeast cells, deletion of LSM-1 disrupted PB assembly,^54–56 and a homolog of LSM-1 (NCU01601, called small nuclear ribonucleoprotein 1 [SNR-1]) is present in N. crassa. To confirm that N. crassa SNR-1 is present in cmRNPgs, SNR-1:dsRed was expressed from its endogenous locus in WT N. crassa cells and treated with 3-amino-1,2,4-triazole (3-AT) to induce nutrient stress, phosphorylation of eIF2α, and cmRNPg formation.^22,38 Independent of the time of day, a significant increase in the number and size of SNR-1:dsRed foci was observed in 3-AT-treated WT cells compared with untreated cells, confirming that N. crassa SNR-1 is present in cmRNPgs (Figure S5). To determine if SNR-1 is required for cmRNPg formation in N. crassa, the homolog of yeast cmRNPg component lsm7, SNR-7 (NCU01930), was tagged with dsRed and expressed from its native locus in WT or Δsnr-1 cells. A significant increase in the number and size of SNR-7:dsRed foci was observed in 3-AT-treated WT cells compared with untreated cells (Figures S5A–S5C), whereas SNR-7:dsRed foci were absent in Δsnr-1 cells with or without 3-AT treatment (Figure S5A). In yeast, cmRNPg proteins Lsm7p, Pat1p, and Dhh1p accumulate in the nucleus in Δlsm1 cells.^54,57 In N. crassa Δsnr-1, SNR-7:dsRed signal was observed in structures that were similar in size to nuclei, and staining with Hoechst revealed colocalization of SNR-7:dsRed with the nucleus (Figure S6A). We also examined the homolog of the yeast cmRNPg component dcp2, called DCAP-2 in N. crassa (NCU07889, mRNA-decapping enzyme subunit 2) using an mCherry:DCAP-2 reporter that was used previously to confirm the presence of cmRNPgs in N. crassa meiotic cells.⁵⁸ A significant increase in the number of mCherry:DCAP-2 foci was observed in 3-AT-treated WT cells compared with untreated cells (Figures S5A–S5C) and was absent in Δsnr-1 cells with or without 3-AT treatment. In the absence of cmRNPgs, mCherry:DCAP-2 localized to vesicular structures that colocalized to vacuoles stained with 5(6)-CFDA and Cell Tracker Blue CMAC dye, two membrane-permeant dyes that stain the vacuolar lumen (Figure S6B). These data established that recruitment of SNR-7 and DCAP-2 to cmRNPgs and the assembly of functional cmRNPgs are dependent on SNR-1.

To determine if rhythmic ZIP-1:LUC translation requires sequestration of zip-1::luc mRNA to cmRNPgs, ZIP-1:LUC levels were examined in Δsnr-1 cells. ZIP-1:LUC levels were arrhythmic in Δsnr-1 cells, and consistent with reduced translation in cmRNPgs, the levels of ZIP-1:LUC were generally high throughout the time course (Figure 5A). The clock functioned normally as evidenced by robust rhythms in FRQ and P-eIF2α levels in Δsnr-1 cells (Figure S7). Deletion of the consensus PB motif in the 5′ leader of zip-1 (zip-1^ΔPBM) resulted in rhythms in ZIP-1^ΔPBM:LUC levels with a period similar to WT (Figures 5B and 5C) but with a severely diminished amplitude (Figure 5D). Taken together, these data demonstrated that robust ZIP-1 protein rhythms, which arise from constant mRNA levels, depend on a functional clock and cmRNPg sequestration as a consequence of rhythmic eIF2α activity.

Figure 5. Robust ZIP-1 protein rhythms depend on cmRNPg sequestration.

(A) Luciferase activity from ZIP-1:LUC translational fusion in WT (black line) and Δsnr-1 (yellow line) cells. ZIP-1:LUC in WT cells was rhythmic as indicated by a better fit to a sine wave (dotted black line, p < 0.05), while ZIP-1:LUC in Δsnr-1 cells was arrhythmic as indicated by a better fit of the data to a line (dotted yellow line, p < 0.05). See STAR Methods for statistical test for rhythmicity.

(B) Luciferase activity from ZIP-1:LUC translational fusion in WT (black line) and PB motif deletion ΔPBM (purple line) cells. Cells were grown in DD and recorded every 90 min over 5 days (h DD). The average bioluminescence signal is plotted (mean ± SEM, n = 36 technical replicates).

(C) Mean period in h (mean ± SEM, n = 36 technical replicates, n.s., not significant, p > 0.05) of rhythmic ZIP-1:LUC bioluminescence traces in WT (black bar) and ΔPBM (purple bar) cells.

(D) Mean amplitude (mean ± SEM, n = 36 technical replicates; ****p < 0.0001) of rhythmic ZIP-1:LUC bioluminescence traces in WT (black bar) and ΔPBM (purple bar) cells. p values were calculated by an unpaired t test with Welch correction.

Clock control of cmRNPg formation in N. crassa cells

Previous studies in S. cerevisiae revealed that PBs are slightly larger in size and almost always colocalize with SGs in cells with constitutive high levels of P-eIF2α, suggesting that decreased eIF2 function leads to increased formation of cmRNPgs.^38,59 Similar to N. crassa, P-eIF2α levels cycle in mammalian suprachiasmatic nucleus (SCN) cells, peaking during the day,³² and this peak in P-eIF2a levels corresponds to a peak in PB formation in synchronized U2OS cells.¹² Based on these data, we investigated if components of cmRNPgs have cycling RPF levels in WT cells. RPF rhythms were identified in four N. crassa homologs of S. cerevisiae cmRNPg components snr-1, snr-7, eIF4E transporter (eIF4E-T; NCU02076), and ribonucleoprotein 7 (rnp-7; similar to edc-3 in yeast; NCU00427) (Figure 6A).⁶⁰ Three of the cmRNPg components peaked in translation during the subjective day, coincident with the peak in P-eIF2α levels, whereas rnp-7 translation peaked at subjective late night. These data suggested that in cells grown in DD, high P-eIF2α levels during the subjective day triggers intracellular stress responses that can drive rhythmic expression of cmRNPg components and assembly. To test this idea, cmRNPg levels were examined using SNR-1:dsRed in N. crassa hyphae at DD28 (subjective night) and DD40 (subjective day) corresponding to the trough and peak expressions of P-eIF2α, respectively. Consistent with a rhythm in cmRNPg formation, the average number of SNR-1:dsRed foci per single hypha was significantly higher in WT cells during the subjective day than during the subjective night, and this time-of-day difference was abolished in Δfrq cells (Figures 6B and 6C). However, no time-of-day difference was observed in the average diameters of the foci in both WT and Δfrq cells (Figure 6D). cmRNPgs increase in size in response to various stress conditions.⁶¹ During nutrient stress induced by 3-AT, the average diameter of SNR-1:dsRed foci in WT N. crassa was significantly higher than in untreated conidia (Figure S5D), but no difference in the diameter of SNR-1:dsRed foci was observed in hyphae isolated during the subjective night (DD28) or day (DD40) (Figure 6D). Thus, while increased P-eIF2α levels during the subjective day can trigger cmRNPg formation in a clock-dependent manner, it is not sufficient to drive an increase in cmRNPg size at the level seen in WT cells under acute stress.

Figure 6. The circadian clock regulates cytoplasmic cmRNPg formation.

(A) Averaged normalized RPF reads (black line) traces of rhythmic cmRNPg components eIF4E-T, snr-1, snr-7, and rnp-7 from WT cells grown in DD and harvested at the indicated times (h). Square symbols indicate the values from biological replicate 1 (black) and replicate 2 (gray).

(B) Representative images showing SNR-1:dsRed as a cmRNPg marker and examined in WT or Δfrq cells grown in DD and imaged at the indicated times. Scale bar: 5 μm.

(C) Quantification of the number of SNR-1:dsRed foci in WT or Δfrq hyphae grown in DD and imaged at the indicated times (mean ± SEM indicated by cross in boxplot, n = 60 hyphae, 3 biological replicates with 20 hyphae for each biological replicate representing technical replicates; **** p < 0.0001).

(D) Quantification of the sizes of SNR-1:dsRed foci in WT or Δfrq hyphae grown in DD and imaged at the indicated times (mean ± SEM indicated by cross in boxplot, n = 60 hyphae, 3 biological replicates with 20 hyphae for each biological replicate representing technical replicates; n.s., not significant, p > 0.05). p values were calculated by an unpaired t test with Welch correction.

DISCUSSION

At least half of the cycling proteins in mammalian cells and N. crassa arise from arrhythmic mRNAs, which could be due to rhythmic protein turnover^62,63 and/or clock control of protein synthesis.^7,9,11 One mechanism for circadian clock control of protein synthesis is through the rhythmic phosphorylation and inactivation of translation initiation factor eIF2α.²² Here, we determined the genome-wide impact of rhythmic P-eIF2α levels in N. crassa on translation and identified a mechanism for selection of mRNAs for rhythmic translation through sequestration to cmRNPgs.

We discovered that clock regulation of eIF2α activity through CPC-3 is necessary for the rhythmic translation of some, but not all, mRNAs in DD rather than affecting global rhythmic translation. In mammalian cells, the levels and/or phosphorylation states of several translation initiation factors, including eIF1, eIF4A2, eF4G1, eIF5, and eIF4E, are rhythmic.^10,64,65 The homo logs of some of these translation factors in N. crassa, including eIF4G1 (NCU07868), eIF5 (NCU00366), and eIF4E (NCU02076), have rhythmic ribosome occupancy in WT cells (Table S4). However, the contribution of these factors to rhythmic translation in N. crassa and mammalian cells is not yet known.

Circadian proteomic analysis of N. crassa WT samples reported 1,273 rhythmic proteins that were significantly enriched for genes involved in cellular metabolism.¹¹ Of these, 359 genes had mRNAs with rhythmic ribosome occupancy in our WT data-sets. There are several possible reasons for the low overlap between the two studies. First, 758 proteins from the rhythmic translatome identified here by ribo-seq were not detected in the proteomics study, most likely due to low peptide abundance. Second, 156 genes with rhythmic ribosome occupancy lacked rhythmic protein levels in the proteomic analyses, suggesting that while translation of the mRNA is rhythmic, protein abundance levels are not. This may be due to, for example, posttranslational modifications that affect stability of the protein. Furthermore, comparison of cTICs to the N. crassa rhythmic proteome¹¹ revealed that 252/404 cTIC proteins were detected by mass spectrometry, but only 57/252 were rhythmic at both the translation and protein levels, likely reflecting long protein half-lives. Similar results were observed in comparisons of rhythmic RPFs and rhythmic protein levels in mouse liver.⁶⁶ However, constant protein levels do not necessarily reflect constant activity since newly synthesized proteins are generally more active than older heavily oxidized proteins.⁶⁷ Thus, steady-state protein levels quantified by mass spectrometry may not be a good indicator of rhythmic protein activity from rhythmically translated mRNAs.

To identify potential mechanisms for P-eIF2α-dependent rhythmic translation of the class III cTICs, a de novo motif analysis was performed on the 5′ leaders of the class III cTICs (Figure 4A). Initially, two motifs were identified, but neither of these yielded a predicted function. However, the CU-rich motif in 68 of the cTICs resembled a 5′ terminal oligopyrimidine (TOP) motif typically present immediately downstream of the 7-methylguanosine cap in mRNAs that are sensitive to translation repression by mTOR inhibitors.⁶⁸ Further testing is needed to determine if this CU-rich motif is functional in controlling cTIC translation. The PB motif, identified in a subset of cTICs, provided an attractive mechanism for translational control since cmRNPgs are known to harbor mRNAs that are translationally repressed in response to developmental cues or external and intracellular stress stimuli.^69,70 Moreover, rhythmic ribosome occupancy of the mRNAs specifying the N. crassa orthologs of known cmRNPg components was observed in WT cells, mostly peaking during the subjective day, indicating that the circadian clock could impact the accumulation of cmRNPg components as well as translational regulation through cmRNPgs (Figure 6A).

We observed that ZIP-1 protein rhythms, which arise despite constant mRNA levels, are dependent on a functional clock, rhythmic eIF2α activity and cmRNPg formation in vivo (Figures 4 and 5). The ZIP-1 homolog IDI-4 is induced during cell death by incompatibility, a condition triggered by the fusion of hyphae of unlike genotype or by other stress conditions.⁵³ ZIP-1 function under normal conditions remains unclear, but ZIP-1 in WT cells peaked in translation at night (Figure 4B). P-eIF2α rhythms are necessary for the rhythmic accumulation of ZIP-1 as demonstrated by arrhythmic ZIP-1:LUC protein levels in Δcpc-3 and cpc-3^c cells (Figures 4D and 4E), and loss of an essential cmRNPg component SNR-1 led to arrhythmic ZIP-1:LUC levels (Figure 5A). While deletion of the PB motif in the ZIP-1 5′ leader did not abolish ZIP-1:LUC rhythms, the amplitude of the rhythm was severely diminished (Figures 5B and 5D). The persistence of rhythms in ZIP-1^ΔPBM:LUC cells may be due to other sites in the 5′ leader that share sequence similarities with the deleted PB motif or with specific sequestration in SGs. Alternatively, the low amplitude rhythm may result from cycling inhibitory P-eIF2α levels affecting ZIP-1 mRNA translation initiation at its poor CUG start codon.⁷¹ In support of sequestration to SGs, SG transcriptome analysis performed on S. cerevisiae cells treated with the oxidative-stress-inducing agent sodium azide identified YAP-7, the yeast homolog of ZIP-1, to be among those enriched in SGs.⁷² PBs and SGs are known to physically interact with each other under some conditions, and this interaction may allow mRNA movement between the two.^38,73 Experiments to localize zip-1 and other cTIC mRNAs to cmRNPgs are in progress to examine these possibilities.

There is growing evidence supporting clock control of the formation of cmRNPgs in mammalian cells.^12,74,75 In S. cerevisiae, the brightness and size of cmRNPg foci increased in the constitutive active eIF2α kinase mutant Gcn2c, in which P-eIF2α levels are high compared with WT.³⁸ These data are consistent with our study showing that cmRNPg accumulation in N. crassa peaks during the day when P-eIF2α levels are high under control of the circadian clock. Increasing the pool of translationally inactive mRNA increases cmRNPg formation⁷⁶; thus, the rhythm in cmRNPg accumulation is likely through the coordinate control of P-eIF2α levels and cmRNPg components by the circadian clock.

CmRNPg sequestration is thought to primarily result in RNA degradation; however, RNAs can also be released from cmRNPgs for translation.^38,61,70 We observed that the levels of cmRNPgs are clock controlled. If mRNA sequestration in cmRNPgs were to result in RNA degradation, then cTIC mRNA levels would be expected to be degraded with a circadian rhythm. However, cTIC mRNAs accumulate arrhythmically, so rhythmic cTIC mRNA translation is unlikely due to the rhythmic degradation of cTIC mRNAs in cmRNPgs. Rather, these data support that night-peaking class III cTICs are released from cmRNPgs, presumably at night for translation when P-eIF2α levels go down. Alternatively, the class III* cTICs that peak in translation during the day must be resistant to high levels of P-eIF2α. This would require non-canonical translation initiation mechanisms, such as internal ribosome entry site (IRES)-mediated translation initiation, which is potentially controlled by the day-peaking cTIC eIF3B (NCU02208) (Table S4), a regulatory component of the IRES binding translation factor eIF3.⁷⁷ In addition, these mRNAs may be compartmentalized in cmRNPgs during the day to enable localized translation control, a mechanism observed to control translation of regulatory proteins, including those that form multiprotein complexes.⁷⁰ Consistent with this idea, many of the day-peaking cTICs with PB motifs encode regulatory proteins that are present in multiprotein complexes, such as the 26S proteasome regulatory subunit (Rpn2; NCU09450),⁷⁸ coatomer subunit gamma (NCU01992),⁷⁹ and heat shock protein 70 (NCU02075)⁸⁰ (Table S7). Neither PB nor other motifs were identified in the class I or II cTICs. The class I cTICs peak in translation primarily at night, the same time as the peak in rhythmic mRNA levels, indicating that direct translation regulation occurs in coordination with low levels of inhibitory P-eIF2α at night. Imposing rhythmic translation on top of rhythmic mRNA levels in the class I cTICs may be a mechanism to increase the amplitude of the protein rhythm; however, this possibility, and the mechanisms responsible for the difference in phase observed between the peak in class II cTIC mRNA and translation, still need to be determined.

Clock regulation of CPC-3 activity is necessary for the rhythmic accumulation of P-eIF2α.²² Based on our current findings and previously published data, we propose a model whereby the rhythmic activation of CPC-3 drives rhythmic P-eIF2α levels, which then controls the rhythmic translation of specific mRNAs important for anticipating daytime stress. Most of these translationally controlled mRNAs are constitutively expressed, allowing cells to not only use their clock to anticipate daily stress but also to be able to quickly respond to acute environmental and intracellular signals by elevating translation of existing transcripts at any time of the day. Loss of this regulation would be predicted to prevent the organism from anticipating and responding to both external and internal stresses.

LIMITATIONS OF THE STUDY

We show that clock-controlled cmRNPg formation affects the rhythmic expression of zip-1 mRNA, a class III cTIC with a CA-rich PB motif in its 5′ leader. While our results suggest that rhythms in protein levels from some non-cycling mRNAs are mediated by cmRNPgs, it remains to be determined whether zip-1 or other identified class III cTICs with the PB motif directly interact with cmRNPgs.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Deborah Bell-Pedersen (dpedersen@tamu.edu).

Materials availability

All unique/stable reagents generated in this study are available from the lead contact without restriction.

Data and code availability

• RNA-seq and Ribo-seq raw FastQ and processed BAM files have been deposited at GEO and are publicly available as of the date of publication. The accession number is listed in the key resources table. Original western blot images have been deposited at Mendeley and are publicly available as of the date of publication. The DOI is listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal EIF2S1 (phosphoS51)	Abcam	Cat# 32157; RRID: AB_732117
Rabbit polyclonal EIF2S1	Abcam	Cat# 47508; RRID: AB_869591
Mouse monoclonal anti-FRQ	Dr. Michael Brunner	Clone 3G11–1B10-E2
Anti-rabbit IgG HRP	Bio-Rad	Cat# 170–6515; RRID: AB_11125142
Anti-mouse IgG-HRP	Bio-Rad	Cat# 170–6516; RRID:AB_11125547
Mouse monoclonal V5	Thermo Fisher	Cat# R960–25; RRID:AB_2556564
Chemicals, peptides, and recombinant proteins
Immobilon-P nitrocellulose membrane	Millipore	Cat# IPVH00010
3-Amino-1,2,4-triazole (3-AT)	Sigma-Aldrich	Cat# 8144950100
Basta	Liberty	Cat# 280SL
Glufosinate ammonium	Toronto Research Chem	Cat# G596950
Polyacrylic acid	Sigma-Aldrich	Cat# 181285
SuperSignal™ West Pico PLUS Chemiluminiscent Substrate	Thermo Scientific	Cat# 34077
SuperSignal™ West Femto Maximum Sensitivity Substrate	Thermo Scientific	Cat# 34085
Hygromycin	VWR	Cat# 80055–268
PMSF	Sigma-Aldrich	Cat# P7626
Luciferin	Gold Biotechnology	Cat# LUNCA-300
Sodium ortho-vanadate	Sigma-Aldrich	Cat# S6508
β-glyerophosphate	Sigma-Aldrich	Cat# G6376
Aprotinin	Sigma-Aldrich	Cat# A1153
Leupeptin hemisulfate salt	Sigma-Aldrich	Cat# L2884
Pepstatin A	Sigma-Aldrich	Cat# P5318
Bathocuproinedisulfonic acid (BCS)	Sigma-Aldrich	Cat# B1125
Immobilon-P nitrocellulose membrane	Millipore	Cat# IPVH00010
Cycloheximide	Sigma-Aldrich	Cat# C7698–5G
Turbo DNase	Thermo Fisher	Cat# AM2238
RNase I	Thermo Fisher	Cat# AM2294
CircLigase	Epicentre	Cat# CL4111K
GlycoBlue™	Thermo Fisher	Cat# AM9515
miRNeasy Mini Kit	Qiagen	Cat# 217004
T4 RNA ligase 2 truncated	New England Biolabs	Cat# M0242L
SuperScript™ III Reverse Transcriptase	Thermo Fisher	Cat# 18080044
Phusion Hot Start High-Fidelity DNA Polymerase	Thermo Fisher	Cat# F540L
DNA high-sensitivity chip	Agilent	Cat# 067–4626
Quant-iT™ RiboGreen RNA assay kit	Thermo Fisher	Cat# R11490
Oligo d(T)25 Magnetic Beads	NEB	Cat# S1419S
Turbo DNA-free™ Kit	Ambion	Cat# AM1907
Terminator 5’-Phosphate-Dependent Exonuclease	Lucigen	Cat# TER51020
SENSE Total RNA-seq Library Preparation Kit	Lexogen	Cat# 009
Hoechst 34,580	AAT Bioquest	Cat# 17537
(5-(and-6)-Carboxyfluorescein Diacetate, mixed isomers (5(6)-CFDA)	Invitrogen	Cat# C195
Cell Tracker™ Blue CMAC Dye	Invitrogen	Cat# C2110
HEPES	Sigma	Cat# H4034–500G
Sodium citrate, dihydrate	Millipore Sigma	Cat# 567446
Deposited data
Circadian ribo-seq DD12-DD48 of WT cells	This paper	GEO: GSE181566
Circadian ribo-seq DD12-DD48 of Δfrq cells	This paper	GEO: GSE181566
Circadian ribo-seq DD12-DD48 of Δcpc-3 cells	This paper	GEO: GSE181566
Circadian ribo-seq DD12-DD48 of cpc-3c cells	This paper	GEO: GSE181566
Circadian RNA-seq DD12-DD48 of WT cells	This paper	GEO: GSE181566
Circadian RNA-seq DD12-DD48 of Δfrq cells	This paper	GEO: GSE181566
Circadian RNA-seq DD12-DD48 of Δcpc-3 cells	This paper	GEO: GSE181566
Circadian RNA-seq DD12-DD48 of cpc-3c cells	This paper	GEO: GSE181566
Original western blot scans are deposited at Mendeley Data	This paper	Mendeley Data: https://doi.org/10.17632/vsm3yp3svb.2
Experimental models: Organisms/strains
Neurospora crassa wild type 74-OR23-IV mat a	FGSC	FGSC 4200; DBP 985
Neurospora crassa wild type 74-OR23-IV mat A	FGSC	FGSC 2489; DBP 984
Δfrq::bar, mat a	Bennett et al.81	DBP 1320
Δcpc-3::hyg, mat A	FGSC	DBP 2694
cpc-3c, mat a	Karki et al.22	DBP 3807
Δfrq::bar, mat A	Bennett et al.81	DBP 1228
Δsnr-1::hyg, mat a	FGSC	DBP 3786
Δsnr-7::hyg, mat a	FGSC	DBP 3788
fl; mCherry-dcap-2::hph; mus-51Δ::bar	Xiao et al.58	DBP 4005
WT, mCherry-dcap-2::hph	This paper	DBP 4024
Δsnr-1::hyg, mCherry-dcap-2::hph	This paper	DBP 4053
FRQ::LUC::BAR translational fusion	Larrondo et al.82	DBP 1563
WT, HAM-7::LUC translational fusion	This paper	DBP 2763
WT, Pham-7::luc transcriptional fusion	This paper	DBP 2679
WT, CPC-1::LUC translational fusion	This paper	DBP 3572
WT, Pcpc-1::luc transcriptional fusion	This paper	DBP 3575
WT, ZIP-1::LUC translational fusion	This paper	DBP 3435
WT, Pzip-1::luc transcriptional fusion	This paper	DBP 3790
Δfrq::bar, ZIP-1::LUC translational fusion	This paper	DBP 3436
Δcpc-3::hyg, ZIP-1::LUC translational fusion	This paper	DBP 3763
cpc-3c, ZIP-1::LUC translational fusion	This paper	DBP 3865
Δsnr-1::hyg, ZIP-1::LUC translational fusion	This paper	DBP 3798
ZIp-1 Δpbm::luc translational fusion	This paper	DBP 3876
WT, SNR-1::dsRed translational fusion	This paper	DBP 4065
Δfrq::bar, SNR-1::dsRed translational fusion	This paper	DBP 4072
WT, SNR-7::dsRed translational fusion	This paper	DBP 4067
Δsnr-1::hyg, SNR-7::dsRed translational fusion	This paper	DBP 4069
Oligonucleotides
See Table S8 for a complete list of oligonucleotides	This paper	https://www.idtdna.com/
Software and algorithms
BioDare2	Moore et al.83	https://biodare2.ed.ac.uk/
Cosine Wave Analysis	Lamb et al.46	GraphPad Prism 7.03
GraphPad Prism 7.03	GraphPad	https://www.graphpad.com/
Cufflinks	Trapnell et al.84	N/A
Cutadapt	Martin et al.85	https://github.com/marcelm/cutadapt
DESeq2	Love et al.86	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
ECHO	De Los Santos et al.47	https://github.com/delosh653/ECHO
Fiji	Schindelin et al.87	https://imagej.net/software/fiji/
HTSeq-count	Anders et al.88	https://htseq.readthedocs.io/en/master/
Image J	Schneider et al.89	https://imagej.nih.gov/ij/
RiboCode	Xiao et al.90	N/A
STAR RNA-seq aligner	Dobin et al.91	https://github.com/alexdobin/STAR
Trimmomatic	Bolger et al.92	http://www.usadellab.org/cms/?page=trimmomatic
Other
EnVision Xcite Multilabel Reader	PerkinElmer	Cat# 2105–0010
NanoDrop™ Microvolume Spectrophotometer	Thermo Scientific	Cat# ND-ONE-W
SPEX CertiPrep 6850 Freezer/Mill®	SPEX SamplePrep	Cat# 6850
Varian Cary® 50 UV-Vis Spectrophotometer	American Laboratory Trading	Cat# 20167
2100 Bioanalyzer	Agilent Technologies	Cat# G2939BA
DNA high sensitivity chip	Agilent Technologies	Cat# 5067–4626
Illumina HiSeq 3000	Illumina	Cat# SY-401–3001
Illumina NextSeq 500	Illumina	Cat# SY-415–1001
High Precision Dissecting Micro Scissors	Fisher	Cat# 08–953-1B
Olympus IX70 Inverted Fluorescence Microscope	Olympus	Model IX70
CoolSNAP Monochrome Interline CCD Camera	Princeton Instruments	Model HQ2

EXPERIMENTAL MODEL AND SUBJECT DETAILS

All Neurospora crassa strains used in this study were derived from wild type 74-OR23-IV mating type a (FGSC 4200) or 74-OR23-IV mating type A (FGSC 2489) grown vegetatively from stocks of conidiating cultures stored at −20°C.

METHOD DETAILS

N. crassa strains and growth conditions

Strains, key reagents, and oligonucleotide primers used in this study are listed in the key resources table. Vegetative cells were grown in 1X Vogel’s minimal media supplemented with 2% glucose (V2G), and genetic crosses were performed on synthetic cross medium supplemented with 0.25% biotin.⁹³ All strains containing the hph construct⁹⁴ were grown in V2G supplemented with 200 μg/mL of hygromycin B. Strains containing the bar cassette were maintained on V2G lacking NH₄NO₃ and supplemented with 0.5% proline and 200 μg/mL of Basta.

To assay rhythmic translation, the N. crassa codon-optimized luciferase gene from pRMP57 (GenBank: KC890770.1)⁹⁵ was used to generate translational fusions by 3-way PCR and then co-transformed with hyg^R pBP15⁸¹ into WT cells (FGSC 2489). Primers are listed in Table S8. Hygromycin-resistant transformants were screened for luciferase activity and homologous insertion into the endogenous ham-7 gene (primers ham-7F5 and ham-7R5), cpc-1 gene (primers cpc-1F5 and cpc-1R5), and zip-1 gene (primers zip-1F5 and zip-1R5). To generate transcriptional fusions to luc, promoter regions of ham-7 (1.635 kb amplified with primers ham-7F4 and ham-7R4), cpc-1 (2.254 kb amplified with primers cpc-1F4 and cpc-1R4), zip-1 (0.67 kb amplified with primers zip-1F4 and zip-1R4) were amplified by PCR. PCR products were digested with the appropriate restriction enzymes and cloned into plasmid pRMP57 containing the codon-optimized luciferase gene. The resulting plasmids were linearized by PciI digest, co-transformed with hyg^R pBP15⁹⁶ into WT (FGSC 4200) cells, and hygromycin-resistant transformants were screened for luciferase activity. Transformation was accomplished by electroporation.⁹⁷

ZIP-1:LUC transformants were crossed with Δfrq::bar (DBP 1228), Δcpc-3::hyg (DBP 2694), cpc-3^c (DBP 3291), and Δsnr-1::hyg (DBP 3786) to generate ZIP-1:LUC, WT (DBP 3435), ZIP-1:LUC, Δfrq::bar (DBP 3436), ZIP-1:LUC, Δcpc-3::hyg (DBP 3763), ZIP-1:LUC, cpc-3^c (DBP 3865), and ZIP-1:LUC, and Δsnr-1::hyg (DBP 3798) homokaryons.

To generate the PB motif deletion in ZIP-1:LUC, the 15-bp consensus sequence in the zip-1 5′ leader (Table S7) was deleted and replaced by an EcoRI restriction site. The replacement fragments were generated by 2-way PCR using primers zip-1F1, zip-1^ΔPbmR1, zip-1^ΔPbmF2 and zip-1R1, and co-transformed with the hyg^R pBP15 plasmid into ZIP-1:LUC, WT (DBP 3435). Transformants were screened by PCR for native integration (primers zip-1F5 and zip-1R1), followed by digestion with EcoRI. A ZIP-1^ΔPbm:LUC homo-karyon (DBP 3876) was obtained by microcondia filtration⁹⁸ and the deletion was verified by sequencing.

To detect cmRNPgs and to validate localization of proteins to cmRNPgs, an SNR-1 or SNR-7 translational fusion to the red fluorescent protein tdimer2(12), which contains a 677 bp direct tandem repeat of dsRed (herein referred to as dsRed for simplicity), was generated by 3-way PCR (snr-1F1 and snr1R1 or snr-7F1 and snr7R1, and snr-1F3 and snr1R3 or snr-7F3 and snr7R3 primers) using tdimer2(12) from pMF332 (primers snr-1F2 and snr1R2 or snr-7F2 and snr7R2),⁹⁹ and co-transformed with hyg^R pBP15 into WT (FGSC 4200). Hygromycin-resistant transformants were screened for homologous insertion into the snr-1 or snr-7 gene (primers snr-1F1 and snr-1R1 or snr-7F1 and snr-7R1). An SNR-1:dsRed transformant was crossed with Δfrq::bar (DBP 1228) to generate SNR-1:dsRed, WT (DBP 4065), SNR-1:dsRed, and Δfrq::bar (DBP 4072) homokaryons. An SNR-7:dsRed transformant was crossed with Δsnr-1::hyg (DBP 3786) to generate SNR-7:dsRed, WT (DBP 4067) and SNR-7:dsRed, Δsnr-1::hyg (DBP 4069) homokaryons. mCherry:DCAP-2 in WT (DBP 4024) and Δsnr-1::hyg (DBP 4053) genotypes were obtained by crossing the mCherry:dcap-2:hph; mus-51^Δ:bar strain⁵⁸ to Δsnr-1::hyg (DBP 3786).

Circadian time courses

Circadian time course experiments for western blots and sequencing libraries were performed according to published methods.⁴⁶ Briefly, mycelial mats in Vogel’s minimal media containing 2% glucose (V2G) (pH 6.0) were synchronized to the same time of day by a shift from LL 30°C to DD 25°C. Two biological replicates for each time point, per strain, were grown in LL for a minimum of 4 h and transferred to DD on day 1 (for collection at DD36, 40, 44, 48, 52), day 2 (for collection at DD12, 16, 20, 24, 28, 32), day 3 (for collection at DD8), and harvested either at 9:00 a.m. (DD12, 16, 20, 36, 40, 44) or 5:00 p.m. (DD8, 24, 28, 32, 48, 52) on day 3. Harvested tissue was immediately frozen in liquid N₂.

Protein extraction and western blotting

Protein was extracted using extraction buffer containing 100 mM Tris-HCl pH 7.0, 1% SDS, 10 mM NaF, 1 mM PMSF, 1 mM sodium orthovanadate, 1 mM β-glycerophosphate, 1X aprotinin, 1X leupeptin hemisulfate salt, and 1X pepstatin A.¹⁰⁰ Protein concentration was determined using a NanoDrop^™ Microvolume Spectrophotometer. Protein samples (100 μg) were separated on 10% SDS-PAGE gels and blotted to an Immobilon-P nitrocellulose membrane according to standard methods.

The levels of P-eIF2α were detected using rabbit monoclonal anti-EIF2S1 (phospho S51) antibody) diluted 1:5000 in 5% BSA, 1X TBS, 0.1% Tween, and anti-rabbit IgG HRP secondary antibody diluted 1:10,000. Total eIF2α levels were detected using rabbit polyclonal anti-EIF2S1 antibody diluted 1:5000, and anti-rabbit IgG HRP secondary antibody diluted 1:10,000. FRQ protein was detected using mouse monoclonal anti-FRQ antibody (gift from Dr. Michael Brunner) diluted 1:200 in 7.5% milk, 1X TBS, 0.1% Tween and anti-mouse IgG-HRP secondary antibody diluted at 1:10,000.¹⁰¹ VVD:V5 was detected using mouse monoclonal anti-V5 antibody diluted 1:5000 in 5% milk, 1XTBST, 0.1% Tween, and anti-mouse IgG HRP secondary antibody diluted 1:10,000. All proteins except FRQ were detected using SuperSignal^™ West Pico PLUS Chemiluminiscent Substrate. FRQ was detected using SuperSignal^™ West Femto Maximum Sensitivity Substrate. Densitometry was performed using NIH ImageJ software⁸⁹ and normalized to protein loading using amido black-stained protein.

Ribosome profiling

Frozen mycelia (1 g) was added to pre-chilled grinding vials, along with 2 mL of lysis beads made by dripping lysis buffer into liquid N₂. Samples were pulverized in an SPEX CertiPrep 6850 Freezer/Mill using a 10 min pre-cooling cycle, followed by 3 × 2 min grind cycles, with 1 min re-cooling between each cycle. Pulverized cells were transferred to pre-chilled 50 mL polycarbonate centrifuge tubes and allowed to thaw on ice for approximately 30 min, and then centrifuged at 4°C for 15 min at 16,000 rpm using a JA-20 rotor. Approximately 1 mL of supernatant was carefully removed, avoiding both the pellet and lipid upper layer, and placed in pre-chilled 15 mL conical tubes. Cell extracts were quantified by measuring the A260 using a Varian Cary 50 UV-Vis Spectrophotometer. Fifty A260 units of extract were brought to a final volume of 300 μL with lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂,1 mM DTT, 1% Triton X-100, 25 U/mL Turbo DNase, and 100 μg/mL cycloheximide) and treated with 2 μL of RNase I (100 U/μL) for 45 min at RT with gentle mixing. The extract was immediately transferred to a pre-chilled 13 × 51 mm polycarbonate ultracentrifuge tube, 0.9 mL of a 1 M polysome sucrose cushion underlay was added, and the ribosomes were pelleted by centrifugation at 70,000 rpm at 4°C for 4 h.¹⁰² The ribosomal pellet was resuspended in Qiazol reagent and total RNA, including small RNA, was extracted using the miRNeasy kit. All RNA precipitation steps were done by eluting the RNA with water, GlycoBlue, and 3 M sodium acetate (pH 5.5), followed by ispropanol addition. Ribosome footprint purification was done by resupending the RNA in 10 mM Tris (pH 8) and separation on a 15% (wt/vol) polyacrylamide TBE-urea gel at 200 V for 15 min. Gels were stained for 3 min with 1X SYBR Gold in 1X TBE running buffer with gentle shaking. Dephosphorylation was carried out by the addition of 1X T4 PNK buffer, 20 U SUPERase-In, and 10 U T4 PNK to the RNA samples. Linker ligation was carried out on the dephosphorylated RNA by adding 0.5 μg/mL preadenylylated linker and denaturation for 90 s at 80°C. Next, 1X T4 RnI2 buffer, 15% (w/v) PEG 8000, 20 U SUPERase-In, and T4 RnI2(tr) were added to the RNA and linker mix and the ligation reaction was incubated at room temperature for 2.5 h. An rRNA depletion step was not performed, but instead after heat-inactivation of CircLigase, 2.0 μL of GlycoBlue^™, 6.0 μL 5 M NaCl, 74 μL of DEPC water, and 150 μL of isopropanol were added to each tube and precipitation was carried out over-night at −80°C. The DNA was pelleted by centrifugation for 30 min at 20,000 ×g at 4°C. The pellets were washed with 70% ethanol and allowed to air-dry for 10 min. The pellet was resuspended in 5.0 μL of 10 mM Tris-HCl (pH 8.0), and cDNA synthesis was accomplished with SuperScript III Reverse Transcriptase, followed by barcode addition with PCR amplification using Phusion Hot Start High-Fidelity DNA Polymerase. The PCR mix was 1X Phusion High Fidelity buffer, 0.2 mM dNTPs, 0.5 μM forward library primer, 0.5 μM reverse library primer, 3 μL of the circularized DNA template, 2 U of Phusion polymerase in a 20 μL total volume. PCR conditions were initial denaturation at 98°C for 30 s, and cycling denaturation at 98°C for 10 s, annealing temperature of 65°C for 10 s, and extension temperature of 72°C for 5 s. The number of cycles tested for each sample were 6, 8, and 10, and PCR products were excised and gel-extracted from the cycling condition with purest PCR product. Sequencing libraries were quantified and checked for quality on a 2100 Bioanalyzer using a DNA high sensitivity chip per the manufacturer’s instructions. Sequencing was carried out on an Illumina HiSeq 3000 for WT, Δfrq, and Δcpc-3 libraries, and on an Illumina NextSeq 500 for cpc-3^c libraries.

RNA-seq

Frozen mycelia (0.1–0.2 g) was homogenized using a bead beater for 1 min with 1 g of 180°C autoclaved zirconium beads, 550 μL of extraction buffer (100 mM Tris-HCl pH 7.5, 100 mM LiCl, 20 mM DTT), 800 μL of phenol:chloroform (pH 4.4), and 80 μL of 10% SDS in a 2-mL screw cap tube. The homogenate was mixed by end-to-end rotation at room temperature for 4 min and centrifuged at 16,000 ×g for 1 min at 4°C. The aqueous phase was extracted with 800 μL of phenol:chloroform (pH 4.4) and 800 μL of chloroform. Total RNA was precipitated in 65 μL of 3 M NaOAc and 1 mL of ethanol at −80°C for 30 min, followed by 30-min 16,000 ×g centrifugation at 4°C. The pellet was washed twice with 70% ethanol, air dried briefly, and dissolved in 200 μL of DEPC-treated water. Total RNA was quantified using the Quant-iT RiboGreen RNA assay kit and diluted to 0.8 μg/μL for downstream steps. Total RNA (100 μg) was mixed with 125 μL of 2X binding buffer and 50 μL of preequilibrated Oligo d(T)₂₅ Magnetic Beads by end-to-end rotation at room temperature for 30 min. The beads were collected by sitting the tube on a magnetic stand for 2 min. The supernatant was removed, and the beads were subsequently washed twice with wash solution 1, twice with wash solution 2, and once with low salt wash solution as per the manufacturer’s protocols. The RNA was eluted from the magnetic beads in 50 μL of elution buffer by 2-min incubation at 50°C as per the manufacturer. The eluted RNA was subsequently treated with Turbo DNA-free kit and Terminator^™ 5′-Phosphate-Dependent Exonuclease as instructed by the manufacturer’s protocols. The RNA was ethanol precipitated and dissolved in 20 μL of DEPC-treated water and quantified using the Quant-iT RiboGreen RNA assay kit. Purified mRNA (8 ng) was used as the input to generate RNA-seq libraries using the SENSE Total RNA-seq Library Preparation Kit. Sequencing was carried out on an Illumina HiSeq 3000 for WT, Δfrq, and Δcpc-3 libraries, and on an Illumina NextSeq 500 for cpc-3^c libraries.

Luciferase assays

To examine bioluminescence rhythms arising from strains containing luciferase fusions, 1 × 10⁵ conidia were inoculated into 96 well microtiter plates containing 150 μL of 1X Vogel’s salts, 0.01% glucose, 0.03% arginine, 0.1 M quinic acid, 1.5% agar, and 25 μM firefly luciferin, pH 6. After inoculation of conidia (1 × 10⁵ conidia), the microtiter plate was incubated at 30°C in LL for 24 h and transferred to DD 25°C to obtain bioluminescence recordings using EnVision Xcite Multilabel Reader, with recordings taken every 90 min over 4–5 d. At least 12 technical replicates per sample were assayed. Raw reads were normalized to the mean to graph the data.

Preparation of cells for microscopy

To examine cmRNPgs in the absence or presence of nutrient stress induced by 3-AT, approximately 1 × 10⁶ conidia expressing SNR-1:dsRed, SNR-7:dsRed or mCherry-DCAP-2 were germinated in 25 mL V2G (pH 6.0) at 25°C for 2.5 h before incubation in 10 mM 3-AT for 30 min. Cells were pelleted by centrifugation on a swinging bucket rotor at 4000 rpm for 5 min. Supernatant was decanted, and conidia were resuspended in 5 mL of media. To examine nuclear localization, Hoechst 34,580 was added to the 5 mL conidial suspension, to a final concentration of 0.5 μg/mL and incubated at room temperature for 10 min.¹⁰³ Approximately 20 μL of conidial suspension was inoculated onto a glass slide with a coverslip. At least 20 conidia, which were imaged immediately after mounting, were examined for each strain and treatment. To examine vacuolar localization, 1 × 10⁶ conidia were suspended in 5 mL V2G (pH 6.0) and incubated with mixing at 25°C for 2.5 h. Cells were pelleted by centrifugation on a swinging bucket rotor at 4000 rpm for 5 min. Supernatant was discarded and the conidia were suspended in either 999 μL 50 mM Na-citrate pH 5 with 1 μL of 10 mM 5(6)-CFDA or 990 μL 10 mM HEPES pH 7.4 with 10 μL of 10 mM CellTracker^™ Blue CMAC dye. Cells were incubated in the dark with mixing at 25°C for 30 min. Cells were pelleted at 10,000 rpm for 30 s and washed twice with 1X V2G (pH 6.0). Approximately 20 μL of conidial suspension was inoculated onto a glass slide with a coverslip and imaged.

To examine cmRNPgs at two circadian time points, a mycelial disk was cut using a sterilized 3 mm diameter cork borer from WT or Δfrq strains expressing SNR-1:dsRed. The disks were inoculated into 50 mL V2G (pH 6.0) and 0.2% polyacrylic acid¹⁰⁴ and synchronized by a shift from LL 25°C to DD 25°C. The cultures were grown in LL for a minimum of 4 h and transferred to DD on day 1 (for collection at DD40), and day 2 (for collection at DD28). The mycelia were transferred to a sterile Petri dish. Hyphal pieces were carefully removed from the mycelia using micro scissors and the hyphae were mounted directly onto a glass slide with a coverslip. Three independent slides representing three biological replicates were prepared for each strain and time point, and at least 20 hyphae per slide were examined. Microscope imaging was performed immediately after slide preparation, ensuring that the mounted material was not exposed to light prior to microscopy.

QUANTIFICATION AND STATISTICAL ANALYSIS

Sequencing data analysis

Ribosome profiling-seq reads were trimmed using Cutadapt⁸⁵ to remove the 3′ adapter of the reads (adapter sequence – CTGTAGGCACCATCAAT). Quality controls of ribo-seq and RNA-seq FastQ files were performed by FastQC.¹⁰⁵ Ribo-seq and RNA-seq reads were aligned to the N. crassa genome FungiDB Release 38 with STAR.⁹¹ The mapped RNA-seq reads were quantified and normalized as fragments per kilobase of exon model per million mapped reads (FPKM) values using Cufflinks.⁸⁴ Adapter-trimmed ribo-seq reads were size-filtered to the 28–34 nt range using Trimmomatic.⁹² As ribosome protected footprints (RPF) are enriched in coding regions, the ribo-seq mapped reads originating only from the coding sequences (CDS) were extracted and quantified using HTSeq-count.⁸⁸ The counted reads were normalized using the median of ratios method employed in the DESeq2 analysis workflow.⁸⁶ To ensure reproducibility between the two ribo-seq and two RNA-seq biological replicates for each genotype at different time points, a Spearman correlation score was calculated from the linear regression equation from each correlation plot of ribo-seq or RNA-seq FPKM values. To calculate the ribo-seq read coverage over the 5′ UTR, 3′ UTR, or CDS regions, the geneBody_coverage2.py program from the RSeQC package was used.¹⁰⁶ Ribosome-protected footprint (RPF) features such as read length distribution and sub-codon phasing were determined using NGS toolkit Plastid.⁴² The RPF originating only from the coding sequences (CDS) were also extracted and quantified using HTSeq-count.⁸⁸ The counted reads were normalized using the median of ratios method employed in the DESeq2 analysis workflow.⁸⁶ Ribo-seq and RNA-seq BAM files were visualized using the genome browser Integrative Genome Viewer (IGV) version 2.8.2.¹⁰⁷ To compare sequencing read levels, BAM files were converted to depth normalized bigWig tracks for viewing in a genome browser.¹⁰⁸ To determine differentially expressed genes between WT and Δcpc-3, ribo-seq reads for all time points were used as replicate values for each strain for DESeq2 analysis.⁸⁶ Genes with p < 0.05 are considered differentially expressed with fold change >1.2 being significantly upregulated and <0.8 being significantly downregulated.

Identification of rhythmic genes

The normalized RPF time series was analyzed using the following ECHO settings: single replicate, smoothing function enabled, un-expressed genes excluded from the analysis at the default 70% threshold. The RNA-seq time series was analyzed using the following ECHO settings: two unpaired replicates, smoothing function enabled, unexpressed genes excluded from the analysis at the default 70% threshold, and OE/RE cutoff of 1.25. Genes were considered rhythmic if they had Benjamini-Hochberg p values <0.05, and oscillation types, harmonic, damped or forced.

GO terms, functional categories, and motifs

Determination of the enriched Gene Ontology (GO) terms and functional categories (FunCat) were done using FungiFun version 2.2.8,^109–111 with p value <0.05 as a cutoff for significantly enriched terms. MEME version 5.3.0 analysis⁵¹ was used for de novo identification of enriched motifs in candidate genes.

Statistical test for significance

p-values were calculated by an unpaired t test with Welch correction function in GraphPad Prism version 7.03. p values <0.05 were considered significant.

Statistical test for rhythmicity

Rhythmic data from western blots and luciferase assays were fit to a sine wave or a line.⁴⁶ Nonlinear regression to fit the rhythmic data to a sine wave (fitting period, phase, and amplitude) and a line (fitting slope and intercept), as well as Akaike’s information criteria tests to compare the fit of each dataset to the 2 equations, were carried out using the GraphPad Prism software package version 7.03. The p values reflect the probability that the sine wave fits the data better than a straight line. p values for sine waves of <0.05 were considered significant. Error bars in all graphs represent the SEM from independent experiments.

Image detection and analysis

Mounted conidia or hyphae were visualized using an Olympus IX70 Inverted Fluorescence Microscope at 100X magnification. Images were acquired using a CoolSNAP HQ2 monochrome camera. cmRNPgs were quantified according to previous methods 60,75 with modifications. Digital images of 1392 × 100 pixels and 8 bits were processed with Fiji 108 as follows: 1) Process menu > “Smooth”. 2) Process menu > “Math” > “Subtract”: value = 15. 3) Image menu > “Adjust” > “Threshold”. Step 3 reverses the image, allowing the fluorescent puncta to be “on”. The threshold was adjusted until the sizes of the puncta appeared similar to the original image. 4) Analyze menu > “Analyze particles”: size 3–300 (pixel units checked), circularity 0.5–1, show outlines, display results checked. This generates a table with the number and area of each detected particle in the selected image. CmRNPg diameter can be estimated from the resulting area and the total number of foci for each conidia or hypha can be manually calculated. For imaging the hyphae, an initial step is performed where a single hypha is outlined using the freehand selection tool, prior to the steps outlined above. This step ensures counting particles in non-overlapping hyphae and allows measurement of the individual hyphal area (Analyze menu > “Measure”).

Highlights.

• Clock control of eIF2α activity regulates rhythmic translation of specific mRNAs

• Most circadian translation-initiation-controlled (cTIC) genes have arrhythmic mRNAs

• cTICs with arrhythmic mRNAs contain a P-body localization motif

• cTIC mRNA targeting to cmRNPgs is necessary for rhythmic translation

INCLUSION AND DIVERSITY

We support inclusive, diverse, and equitable conduct of research.