PRD-2 mediates clock-regulated perinuclear localization of clock gene RNAs within the circadian cycle of Neurospora

Abstract

The transcription-translation negative feedback loops underlying animal and fungal circadian clocks are remarkably similar in their molecular regulatory architecture and, although much is understood about their central mechanism, little is known about the spatiotemporal dynamics of the gene products involved. A common feature of these circadian oscillators is a significant temporal delay between rhythmic accumulation of clock messenger RNAs (mRNAs) encoding negative arm proteins, for example, frq in Neurospora and Per1-3 in mammals, and the appearance of the clock protein complexes assembled from the proteins they encode. Here, we report use of single-molecule RNA fluorescence in situ hybridization (smFISH) to show that the fraction of nuclei actively transcribing the clock gene frq changes in a circadian manner, and that these mRNAs cycle in abundance with fewer than five transcripts per nucleus at any time. Spatial point patterning statistics reveal that frq is spatially clustered near nuclei in a time of day-dependent manner and that clustering requires an RNA-binding protein, PRD-2 (PERIOD-2), recently shown also to bind to mRNA encoding another core clock component, casein kinase 1. An intrinsically disordered protein, PRD-2 displays behavior in vivo and in vitro consistent with participation in biomolecular condensates. These data are consistent with a role for phase-separating RNA-binding proteins in spatiotemporally organizing clock mRNAs to facilitate local translation and assembly of clock protein complexes.

Keywords: Neurospora; cell biology; circadian rhythms; liquid–liquid phase separation; smFISH.

Fig. 1.

frequency transcripts can be detected and quantified using smFISH. (A) Wild-type (WT) N. crassa hybridized with frq mRNA–specific Quasar 647–conjugated oligonucleotides (magenta) and Hoechst 33342–stained nuclei (green) shown as individual channels and a composite of both channels. (B) Composite smFISH images of frq-specific and cln-1–specific Quasar 647–conjugated oligonucleotides (magenta) and Hoechst 33342–stained nuclei (green) in a Δfrq mutant strain of N. crassa. (C) Histograms of integrated density values for detected frq mRNA foci in 50 images from four biological replicates for both constant light and the first circadian mRNA peak in free running conditions (DD16/CT5), both used to quantify individual transcripts. (D) Boxplots representing the ratio of frq mRNA to nuclei and volume in LL and DD16. Hyphal boundaries are indicated by gray outlines. (Scale bars, 10 µm.)

Fig. 2.

frequency mRNA cycles in abundance throughout the day. (A) Composite images of frq mRNA (magenta) and nuclei (green) in wild-type free running (constant darkness) N. crassa sampled every 4 h for 36 h. (Scale bar, 5 μm.) (B) Composite images of frq mRNA (magenta) and nuclei (green) shown as maximum-intensity Z projections. Arrows indicate nuclei containing colocalized frq transcripts. Hyphal boundaries are indicated by gray outlines. (Scale bars, 5 μm.) (C) Boxplot approximating the percentage of nuclei actively transcribing frq as determined by the number of nuclei containing colocalized frq transcripts. (D) Boxplot comparing the ratio of frq mRNA to nuclei in multiple images (n = 20) at each time point in a free running circadian experiment.

Fig. 3.

frequency mRNA clusters with nuclei. (A) Spatial point patterning analysis using Ripley’s K. Three-dimensional rendering of objects from real data detected by image analysis software. Transcripts (magenta) appear clustered near a nucleus (green) (i) or having no particular association (random) with a nucleus (ii). Black dots indicate the center of mass for the nuclei. The number of transcripts within spheres at increasing distances extending outward from the nuclear center of mass is quantified using Ripley’s K function (iii and iv) and plotted (v). Nuclei with a high number of transcripts clustered within a sphere with radius x from the nuclear center of mass will show enrichment on the corresponding magenta line on the plot (v), whereas nuclei with transcripts randomly distributed from the nuclear center of mass will simply increase in a near-linear manner as shown with the black line on the plot (v). A 95% CI of random distribution is determined by a 100× simulation and represented by black dotted lines (vi). Data are normalized to the upper bounds of the 95% CI and a median clustering index is plotted in magenta (vii). Values where the curve is above the 95% CI random range indicate clustering with P < 0.05. (B) Clustering index of frq and ad-18 in constant light. (C) Clustering index of frq throughout the day in free running (constant darkness) conditions.

Fig. 4.

RNA-binding protein PERIOD-2 influences frq localization. (A) Clustering index of frq in constant light (purple) and at CT5 (yellow) in a Δprd-2 mutant strain, compared with wild type (thin dotted lines). (B) Boxplot of the ratio of frq mRNA to nuclei in Δprd-2 hyphae in constant light and at CT5. Mean values are not statistically different from Fig. 1D wild type. (C) qRT-PCR performed on uvCLIP samples of PRD-2^FLAG and Puf4^FLAG using anti-FLAG–coated magnetic beads shows enrichment of frq and ck-1a in PRD-2 but not Puf4 samples (N = 3 biological replicates, error bars +/- SD). The mRNA produced by the actin gene was used for normalization.

Fig. 5.

PERIOD-2 is an intrinsically disordered cytoplasmic protein. (A) PRD-2 is heterogeneously localized in the cytoplasm as shown by PRD-2^mNeonGreen and histone H1^mApple in live cells. (Scale bar, 5 µm.) (B) Multicolored lookup table showing heterogeneous patterning of PRD-2^mNeonGreen. Approximately circular black regions within the cell are nuclei, from which PRD-2 is excluded. Arrows indicate a patch of enriched PRD-2 associating with a nucleus and being pulled apart into two distinct patches. (Scale bar, 4 µm.) (C) PRD-2 is predicted to be highly disordered by IUPred2 and ANCHOR2. Regions of the plot where a line exceeds a score of 0.5 are predicted to be part of a disordered region. (D) Intrinsically disordered proteins FRQ and PRD-2 remain in solution after incubating a lysate at 100 °C for 10 min, whereas the highly ordered protein FRH precipitates out.