Research ArticleGENETICS

Extensive tissue-specific expression variation and novel regulators underlying circadian behavior

See allHide authors and affiliations

Science Advances  29 Jan 2021:
Vol. 7, no. 5, eabc3781
DOI: 10.1126/sciadv.abc3781
  • Fig. 1 Tissue-specific gene expression cycling and detection of previously uncharacterized circadian regulators based on the reference w1118 time series experiments.

    (A) Schematic representation of the tissue-specific samples obtained from w1118. (B) Heatmap of the expression of TSC genes averaged across 2 days of observation. The presence of the characteristic circadian patterns in the tissue is indicated with black rectangles around the heatmaps. The number at the left designates the number of detected TSC genes. (C and D) Heatmap of cycling in two (C) and three tissues (D) gene expression averaged across 2 days of observation. Black rectangles indicate tissues in which genes were found to be cycling. For (B) to (D), lighter areas or the heatmaps correspond to lower expression levels. The time is indicated at the bottom of the heatmap. (E) Examples of expression across tissues of the TSC genes during two consecutive days of observation (12-hour/12-hour LD + 24-hour DD). The rectangle at the bottom shows the presence and absence of light. (F) The majority of TSC genes are not tissue-specifically expressed. Each panel is dedicated to the genes found cycling only in the corresponding tissue. The X axis denotes the expression pattern of genes across tissues, and the Y axis shows the number of TSC genes with the considered expression pattern. The color of the bar indicates a differential expression status of the gene compared to specific tissues as listed on the X axis. Black (“Multi”) indicates genes with a varying direction of differential expression. In (B) to (E), FB stands for fat body and MT for Malpighian tubules, respectively. DE, differentially expressed.

  • Fig. 2 Genes cycling in all examined tissues.

    (A) Rhythmic expression of the canonical circadian genes and previously uncharacterized genes (bold) cycling in all four tissues during two consecutive days (12-hour/12-hour LD + 24-hour DD). Rectangles show the presence and absence of light. (B and C) Knockdown effect of four putative core circadian genes in tim-expressing (B) and Pdf-expressing (C) clock neurons (LNvs) on locomotor behavior: % rhythmic flies (left), period (middle), and rhythmicity index (right). The Y axis indicates gene names, and the subscript denotes an RNAi line stock identifier. The first letter of the subscript designates the RNAi line stock center: (B)loomington and (V)DRC. Column N specifies the number of flies used in the experiment. The red dashed line depicts the mean of measurements in control flies. For the period plot, only measurements on rhythmic flies were used. A test of equal proportions was used to assess the statistical significance of the % rhythmic flies deviating from the control, and analysis of variance (ANOVA) was used for period and rhythmicity strength. Several RNAi lines were tested per gene. Amph, CG2277, CG5793, and Usp1 showed robust and consistent effects in both RNAi constructs and both gene drivers. Figure S1 displays an overview of all tested lines. The stars indicate the FDR-adjusted P values: ***P < 0.001, **P < 0.01, *P < 0.05, and ✕P < 0.1.

  • Fig. 3 Mammalian orthologs of genes cycling in all tissues and phase shift in the expression across tissues.

    (A) Putative core clock regulators orthologs’ expression in baboon and mouse. The data were derived from (27) and (26), respectively. (B) Examples of genes cycling in two tissues with the most significant phase shift. ⍵ in plot headers denotes the phase shift value. Vertical lines correspond to the tissue-specific expression’s molecular peak times. Tissue-specific molecular peak times on the second day were calculated as the sum of tissue-specific molecular peak time on the first day and tissue-specific gene period. (C) Examples of genes cycling in three tissues with the most significant phase shifts per tissue combination. Spin and CG5789 have a phase shift in all pairs of tissues they cycle in. ⍵ in the plot header for Spin denotes the phase shift between the brain and gut. Phase shifts for the brain–fat body and fat body–gut were 3.72 and 3.20 hours, respectively. ⍵ in the plot header for CG5789 denotes the phase shift between fat body and gut. Phase shifts for the fat body–Malpighian tubules and gut–Malpighian tubules were 5.50 and −5.45 hours, respectively. (D) Phase shift in the expression patterns of glutathione metabolism–associated genes.

  • Fig. 4 Regulatory network and motif analyses of TSC gene expression.

    (A) GRN based on genes cycling in at least one tissue. The color of the node denotes the tissue(s) a gene cycles in. The size of the node designates the node out-degree. The color of the edges represents the tissue in which the edge was derived. Shaded areas indicate the TSC gene regulatory modules with the names of the TFs arranged in the same order as the nodes. TFs are denoted as diamonds. Nodes with an orange rim indicate direct targets of Clk/Cyc. For network clarity, only the top 200 links ranked by weight from every tissue are visualized. (B) Examples of Clk/Cyc binding profiles at the vicinity of TSC TF–coding genes as derived from (2). The name of the TSC TF–coding gene is indicated at the top left corner of each column. Rows correspond to ChIP-seq samples, and columns denote tissues for which a cycling TF-coding gene was found to be specific. The black bar at the bottom of each column designates the ChIP-seq peak location. (C) Motif enrichment in TSC gene promoters. The motif’s name is indicated at the bottom, and its graphical representation is depicted at the top. The number inside cells refers to the fold enrichment of a motif in the target regions over the random background. The color bar indicates the q-value for motif detection.

  • Fig. 5 Transcriptomic-based physiological circadian time assessment for tissue-specific DGRP profiles.

    (A) Experimental strategy to reconstruct dynamic circadian expression profiles based on static transcriptomes from DGRP lines. Dots represent the expression value of the same circadian gene derived from the transcriptome of a random DGRP line sampled at the time indicated on the X axis. (B) The dots recapitulate the dynamic expression pattern of a cycling gene as it would be derived from a regular time series analysis (shown for timeless). (C) MTT-inferred physiological time versus sampling time. Colored areas feature a difference of >3.4 hours between the physiological and sampling times. (D) Time shift of outlier lines across tissues. Gray circles indicate a nonsignificant (n.s.) shift <3.4 hours, and the absence of a circle denotes missing data. The period value was derived from (16). The first four columns show samples for line DGRP-774 used as a proof of concept (see Materials and Methods), which has an 18.9-hour period in males (16). N/A, not applicable. (E) Brain and gut expression profiles of timeless during a 12-hour/12-hour LD followed by 24-hour DD in w1118 and DGRP-796. The rectangle indicates light presence versus absence. (F) Venn diagram showing the overlap between cycling genes in w1118 and DGRP-796 time series in brain and gut. The color code is as in (E). (G) Amplitude distribution of the top 50 cycling genes ranked by the amplitude detected in the w1118 and DGRP-796 time series. ***P < 0.001; t test (N = 50).

  • Fig. 6 Behavioral characterization of DGRP-796.

    (A) Double-plotted activity measurements for w1118 (right), Canton-S (middle), and DGRP-796 flies (left) in DD after entrainment in standard LD conditions. The number of flies used in each experiment is indicated in the parenthesis next to the genotype. Each row of the diagram represents a histogram of flies’ activity during the day concatenated with the data from the previous day. The first row displays the data for observation days 1 and 2, the second row shows day 2 and 3, etc. The dark gray background indicates the absence of light, and white background designates the presence of light. The activity of w1118 focused in two peaks at dusk and dawn, and that of DGRP-796 was concentrated mostly around dusk, whereas Canton-S was active uniformly across the day. (B) Double-plotted activity measurement of w1118 (right), Canton-S (middle), and DGRP-796 flies (left). The yellow rectangle indicates the light pulse of 20 min (CT15). Upon subjection to the light pulse, w1118 and Canton-S showed a phase delay of approximately 3.4 hours, which could be seen as a shift to the bottom right of the activity peak between the diagram rows. DGRP-796 did not respond to the light pulse.

  • Fig. 7 Histological characterization of DGRP-796.

    (A) In the absence of light, Tim (first column) is expressed in LNds and l-LNv neuronal groups in both Canton-S and DGRP-796. L-LNvs are also marked by expression of Pdf (second column). No significant difference is observed between the fly lines. (B) Influence of a light pulse on neuronal groups in Canton-S and DGRP-796. Upon light pulse exposure, Tim is effectively degraded in the LNds in Canton-S (first row) but to a much lesser extent in DGRP-796 [third row versus no light pulse (B)]. Large LNvs expressing Pdf (second column) are not affected by the light pulse in either of the lines (second and fourth rows) (47), constituting an internal technical control. (C) Quantification of Tim staining in LNds in Canton-S and DGRP-796. (D) Quantification of Tim staining in l-LNvs in Canton-S and DGRP-796. The number of Tim-expressing LNds (C) and l-LNvs (D) before and after the light pulse was counted in both fly lines, and a two-sided t test was performed. The number of used brains is indicated at the bottom.

  • Fig. 8 Effect of the identified novel cry allele on protein structure.

    (A) Chromosome mapping strategy involving phase shifts of w1118, DGRP-796, and their crosses after the light pulse at CT15 (yellow diamond). (B) Alignment of Cry protein sequences for the wild-type (WT) dCry, DGRP-796 (dCry DGRP-796), and its mouse (mCry1 and mCry2) and human (hCry1 and hCry2) homologs. (C) WT (yellow) and DGRP-796 (light blue) Cry protein structure models. (D) Enlarged w1118 (yellow) and DGRP-796 (light blue) Cry protein structure models in the Asp410-Arg430 region. The DGRP-796 cry mutation disrupts the secondary structure of the ⍺ helix and causes a reorientation of Trp420. (E) The percentage of rhythmic and arrhythmic flies in w1118 mutant flies under LL conditions. The numbers at the right indicate population size. FDR-adjusted P values: ***P < 0.001, **P < 0.01, *P < 0.05 and ×P < 0.1.

Supplementary Materials

  • Supplementary Materials

    Extensive tissue-specific expression variation and novel regulators underlying circadian behavior

    Maria Litovchenko, Antonio C. A. Meireles-Filho, Michael V. Frochaux, Roel P. J. Bevers, Alessio Prunotto, Ane Martin Anduaga, Brian Hollis, Vincent Gardeux, Virginie S. Braman, Julie M. C. Russeil, Sebastian Kadener, Matteo dal Peraro, Bart Deplancke

    Download Supplement

    The PDF file includes:

    • Supplementary methods
    • Figs. S1 to S3
    • Legends for tables S1 to S7
    • References

    Other Supplementary Material for this manuscript includes the following:

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article