(a) The upper graph shows the schematic of nr1d1:VNP construct design. The lower graph shows a magnified view of the nr1d1 promoter sequence used for driving the circadian expression of VNP. The putative RRE (Nr1d1/2 binding site), E-box (Bmal1/Clock binding site), Crx, Otx5, and Crx/Otx5 binding sites were indicated by red oval, blue rectangle, light green oval, yellow rectangle, and dark green oval, respectively. (b) Plot of real-time PCR results of nr1d1 and Venus-pest expression. Each time point has 2 replicates, and each replicate is the pool of 7–10 fish. The dots show the original value, while the solid lines and the bars represent mean ± SEM. (c) Whole-brain two-photon image shows the spatial distribution of nr1d1:VNP-positive cells in different brain regions at 7.5 dpf. (d) Experimental design to examine the developmental dynamics of nr1d1:VNP expression in the whole zebrafish brain. (e) Single-cell tracing result of nr1d1:VNP-positive cells in the pineal gland (red), the optic tectum (green), and the cerebellum (blue) for one fish. The cells were tracked from 3.5 dpf in the pineal gland, 5.5 dpf in optic tectum, and 6.5 dpf in cerebellum, respectively. Each thin line represents one cell, while the thick line represents the mean value of cells in each brain region. (f) Illustration of the setup for time-lapse imaging. The fish were fixed in a chamber with circulating egg-water without anesthesia at temperature of 28 ± 0.5°C. (g) Single-cell tracing results of all 48 nr1d1:VNP-positive cells imaged at 1-hour resolution. The blue dots represent the original fluorescence signals, while the solid red line represents the smoothed curve fitted by the cosine functions. The numerical values for panels b, e, and g are in S1 Data. dpf, days postfertilization; PA, poly(A) site; Tol2, transposable element of Oryzias latipes 2; VNP, Venus-NLS-PEST.

(a) t-SNE visualization of brain cell clusters. The clusters were annotated by comparing to the adult zebrafish scRNA-seq data. (b) Violin plot demonstrated the clusters with enriched expression of nr1d1:VNP. Y axis denotes the normalized expression value. Note the highest levels in the pineal gland. (c) Fluorescence images show the co-expression of nr1d1:VNP (nuclear signal) and aanat2:mRFP (cytoplasmic signal) in the zebrafish pineal gland. The left graph illustrates the 3D location of the pineal gland in zebrafish larva. (d) Three-dimensional reconstruction of a common zebrafish pineal gland by aligning and averaging of six fish. The gray sphere represents the boundary of the pineal gland and the green color represents the density distribution of nr1d1:VNP-positive cells in the pineal gland. (e) Violin plot showed the expression of rod cell markers (rho, gnat1), cone markers (gnat2, gngt2a), and nr1d1:VNP in photoreceptor clusters. Y axis denotes the normalized expression value. (e) Fluorescence images showed the co-expression of nr1d1:VNP (nuclear signal) with the Tg(xops:nfsB-mCherry) fish line (cytoplasmic signal) and Tg(lws2:nfsB-mCherry) fish line (cytoplasmic signal) in the zebrafish pineal gland. The numerical values for panel b and e are in S1 Data. scRNA-seq, single-cell RNA-seq; t-SNE, t-distributed stochastic neighbor embedding; VNP, Venus-NLS-PEST.

(a) Experimental design to examine the developmental dynamics of nr1d1:VNP expression. (b) Fluorescence images illustrate the tracing of one nr1d1:VNP cell during development from 3.5 dpf to 6.5 dpf. This cell was highlighted in green in Fig 3C. (c) Raw fluorescence intensity of the traced cells (six fish). (d) Regression model used to fit the fluorescence intensity (the cell is illustrated in Fig 3B and highlighted in Fig 3C). The dots represent the fluorescence intensity, while the line represents the fitting curve. (e) Scatterplot demonstrated the relationship between coefficients of developmental effect (A) and circadian oscillation (B). The histograms in orange and blue showed the distribution of coefficients of developmental effect (A) and the distribution of coefficients of circadian oscillations (B), respectively. ***Pearson’s correlation P < 0.001. (f) Raw fluorescence intensity of rod-like cells and non-rod-like cells (six fish). Each thin line represents one cell and each dot represents raw fluorescence intensity. The thick lines represent the loess-smoothed curves for all rod-like cells in red and non-rod-like cells in blue, respectively. The shaded areas show the 95% confidence level of the smoothed curve. (g) A comparison of baseline expression (C) between rod-like and non-rod-like cells. Two-tailed Student t test was applied to calculate the levels of significance between the two types of cells. **P < 0.01. The numerical values for panels c–g were in S1 Data. dpf, days postfertilization.

(a) Experimental design to examine the developmental dynamics of nr1d1:VNP expression at higher temporal resolution. (b) Fluorescence images illustrate the tracing of one nr1d1:VNP example cell every 2 hours. (c) Single-cell tracing results of all 117 nr1d1:VNP-positive cells in two zebrafish pineal glands (two fish). The blue dots represent the original fluorescence signals, while the solid red line represents the smoothed curve fitted by the cosine functions. The example cell in (b) is highlighted by a green square. (d) Circadian phase distributions of the 117 nr1d1:VNP-positive cells. (e) t-SNE visualization of the clustering result of the 117 nr1d1:VNP-positive cells. (f) Raw fluorescence intensity traces of the two types of nr1d1:VNP-positive cells in (e). (g and h) The comparison of the baseline expression (g) and relative circadian amplitude (h) between the two types of cells. The colors of the boxes correspond to (e). Two-tailed Student t test was applied to calculate the levels of significance between the two types of cells. ***P < 0.001. (i and j) Scatterplot demonstrates the relationship between baseline intensity and relative oscillation amplitude for fish1 (i) and fish2 (j), respectively. ***Pearson’s correlation P < 0.001. The numerical values for panels c–j are in S1 Data. dpf, days postfertilization; t-SNE, t-distributed stochastic neighbor embedding; ZT, Zeitgeber time.

(a) Experimental design to investigate the effect of light on the onset of circadian clock development. (b) Expression patterns of all cells under DD condition during development (four fish). Each thin line represents one cell. Each thick line represents the loess-smoothed curve of all cells in each individual fish. The shaded area shows the 95% confidence level of the smoothed curve. (c) Comparison of oscillating coefficient (B) between LD (six fish) and DD (four fish) cells. Two-tailed Student t test was applied to calculate the levels of significance between the two types of cells. ***P < 0.001. (d) Comparison of developmental coefficient (A) between LD (six fish) and DD (four fish) cells. Two-tailed Student t test was applied to calculate the levels of significance between the two types of cells. ***P < 0.001. (e) Experimental design to examine the single-cell circadian clocks under DD after transferred from LD (LD_DD). (f) Expression patterns of LD_LD (three fish) and LD_DD cells (three fish) from 6.0 to 7.5 dpf. Each thin line represents one cell and each dot represents raw fluorescence intensity. Thick lines represent the loess-smoothed curves for all the LD_DD cells in red and LD_LD cells in blue, respectively. The shaded areas show the 95% confidence level of the smooth curve. (g) Log2-transformed dusk–dawn ratios of VNP fluorescence intensities of all the cells in (f). Each data point represents one cell and the colors correspond to different fish. The red dashed line represents y = 0. n.s. represents no significant difference in dusk–dawn ratios between different time points (two-tailed ANOVA test). (h) Experimental design to examine the expression pattern of DD cells at higher temporal resolution. (i) Expression patterns of LD (two fish) and DD cells (three fish) across one day at 2-hour resolution. Each thin line represents one cell and each dot represents raw fluorescence intensity. Thick lines represent the loess-smoothed curves for all the LD in blue and DD cells in red, respectively. The shaded areas show the 95% confidence level of the smooth curve. (j) Comparison of the absolute amplitudes of LD and DD cells in (i) from JTKcycle. The red dashed line represents y = 100. (k) Percentages of oscillating cells (JTKcycle adjusted P < 0.05 and absolute amplitude >100) in each LD and DD fish. The orange bars represent the percentage of oscillating cells, while the blue bars represent the percentage of non-oscillating cells. The numerical values for panels b, c, d, f, g, i, j, and k were in S1 Data. DD, constant dark; dpf, days postfertilization; LD, light–dark; VNP, Venus-NLS-PEST; ZT, Zeitgeber time.