Timeless noncoding DNA contains cell-type preferential enhancers important for proper Drosophila circadian regulation

To decrease the expression of tim mRNA, CRISPR-mediated deletions were made in the two regulatory regions known to affect tim reporter gene expression: the region upstream of the transcription start site and a region within the first intron of the tim gene. Both regions include E-box, which almost certainly help recruit the CLK:CYC complex to the chromatin (Fig. 1A). The upstream regulatory region is larger and more complicated than the intronic region as the former contains not only E-boxes but also E-box–like motifs (27, 31).

In the intronic region, a 24 bp deletion resulted (tim_in24) in a deletion of intronic E-box. In the upstream region, we obtained three useful deletions of 10 bp, 122 bp, and 126 bp: tim_up10, tim_up122, and tim_up126, respectively (Fig. 1B). They all removed the E-box, and the two larger deletions also removed the other two E-box–like motifs within this region. The 2nd chromosomes containing these deletions were made homozygous for subsequent biochemical and behavioral analyses. It is worth noting that a canonical E-box was recreated by accident in the tim_up126 deletion; it is ~14 bp upstream of the position of the endogenous E-box (Fig. 1B).

The intronic deletion tim_in24 had little or no effect on tim gene expression (Fig. 1C, yellow vs. teal). In contrast, all deletions in the upstream region decreased tim expression compared to control flies (see Fig. 1C for the smallest 10 bp tim_up10 deletion vs. WT, yellow vs. red). The two larger upstream deletions had even bigger effects (Fig. 1C blue, orange). The data are completely consistent with previous studies using reporter gene assays, which showed that deletions of the upstream E-box are much more important for transcription activation than the intronic E-box (27, 31). Notably, the cycling of tim mRNA expression from the deletion strains appeared largely normal except for its lower amplitude, an effect that is largely if not exclusively due to reduced peak levels with little or no change of trough levels. We also carried out qRT-PCR to assess the expression level of tim in these mutants with similar results (SI Appendix, Fig. S1↗). Not surprisingly, the biggest effect was from the largest E-box containing deletion, tim_up126, which expressed about 15% of WT tim mRNA levels (Fig. 1C). This strain also expressed comparably low levels of TIM protein (Fig. 1 D and E).

These deletion strains appear to constitute an allelic series: tim gene expression progressively decreases, presumably reflecting reduced binding of the positive transcription factor CLK:CYC to this important upstream region (Discussion). As the role of TIM in gene expression regulation is less well-characterized than that of PER, we wondered how the reduced TIM levels would affect the expression of other clock genes that are also direct targets of the CLK:CYC complex. To this end, vrille (vri), per, and PAR domain protein 1 epsilon (pdp1) expression was carefully assessed in the deletion strain series.

Although there is a prominent effect of the deletions on mRNA cycling amplitude, this is largely due to a failure to achieve the low mRNA trough levels characteristic of the WT strain (Fig. 2 A–C). This is best seen for the vri graph (Fig. 2B): the allelic series has progressively increasing trough levels, and the largest deletion tim_up126 has the highest trough levels and the lowest cycling amplitude (blue). The two other genes shown, per and pdp1, have a comparably low cycling amplitude in the largest tim_up126 strain, also due principally to high trough values (Fig. 2 A and C). The data overall indicate an important contribution of TIM to transcriptional repression, i.e., the lower TIM levels compromise repression.

These data (Fig. 2 A–C) have a number of other features that are worth noting. 1) Consistent with the marginal effect of the tim_in24 deletion on tim gene expression peak levels (Fig. 1C), there is almost no effect of this deletion on trough levels of the different mRNAs. 2) There is only a modest effect of most of the deletions on peak mRNA levels. 3) pdp1 mRNA levels are almost unchanged as a function of time in the tim_up126 deletion strain, i.e., they are strongly affected with a cycling amplitude at or near-zero. 4) There is an interesting advanced phase effect on per RNA cycling in the tim_up126 strain. As this effect is matched by the advanced phase of the Clk ChIP profile, it suggests that this is a transcriptional effect due perhaps to the markedly lower levels of the PER–TIM repressor (Fig. 2 A and D). However, other CLK-CYC direct target genes are not so strikingly advanced, further suggesting a somewhat different regulation of per transcription.

Less robust transcriptional repression of CLK-CYC direct target genes might involve less robust release of CLK-CYC from chromatin at times when the transcription rate of these genes is relatively low (32). To address this possibility, we assayed by ChIP-seq (Chromatin-immunoprecipation followed by sequencing) the association of CLK with clock gene regulatory regions in the tim_up126 mutant strain as a function of time of day. We assayed in parallel a WT strain as a positive control, as this result has been previously reported (23, 32). per and pdp1 genes are shown as their CLK binding pattern is approximately characteristic of all well-known direct target genes (Fig. 2D and SI Appendix, Fig. S2↗).

CLK:CYC manifests a canonical association pattern with the per regulatory regions in a wild-type (yw) strain, namely, binding is maximal at ZT10 and ZT14 and very low at the beginning and end of the day, similar to per transcription and RNA profiles (Fig. 2 D, Left). The pattern is very different in the tim_up126 strain (Fig. 2 D, Right), suggesting that the usual temporal association pattern of CLK:CYC with its direct target genes—especially the weak binding normally observed at the beginning and the end of the day when transcription is low—is affected by a lack of sufficient TIM. Although the patterns are not completely identical on all direct target genes this feature is consistent.

It is worth noting in this context that although the tim_up126 deletion accidentally created a canonical E-box (Fig. 1B), the phenotype of this deletion is more severe than that of the smaller tim_up122 deletion, which does not create an E-box (Fig. 2 A–C). This suggests that a simple E-box is insufficient to recruit normal CLK/CYC activity in this in vivo context.

Fly head extracts are a traditional substrate for Drosophila circadian biochemistry and encompass a heterogeneous mix of tissues, including photoreceptors, fat body, neurons, and glia. Indeed, previous studies report that different circadian tissues exhibit highly heterogeneous sets of molecular clock outputs: there are only 14 common cycling transcripts between brain, fat body, gut, and malphigian tubules. Because eight of them are transcribed from the core clock genes tim, vri, per, cry, Clk, cwo, Pdp1, and Cipc (33), the core circadian oscillator is likely much more similar between tissues and cell types than their output transcripts. Nonetheless, core clock components are subject to heterogeneous splicing in an environment- and cell type–specific fashion (34–37), suggesting that there may be other differences in the regulation of core clock component between different head tissues. We were particularly interested whether the detailed regulation of tim transcription within the 150 clock neurons of the fly head is different from that observed in head extracts (Figs. 1 and 2). As mentioned above, this is because clock neurons dictate circadian locomotor activity rhythms and because their biochemical properties are likely invisible in fly head extracts (38).

We first carried out TIM immunostaining around the clock in the tim_up126 strain in constant darkness conditions, i.e., the same conditions under which circadian behavior is normally assayed. Although the TIM expression phase is somewhat shifted and TIM levels somewhat lower in the deletion mutant strain than in the yw strain, (Fig. 3A), the curves are much more similar than the dramatically different TIM western blot results from head extracts (Fig. 1 D and E). PER immunostaining of the tim_up126 strain was similarly shifted compared to yw (Fig. 3B), which is very different from the PER western blot results from head extracts (SI Appendix, Fig. S3↗). Moreover, PDP1 was still cycling in clock neurons of the tim_up126 strain albeit with lower peak levels (Fig. 3C). This is also very different than pdp1 mRNA cycling in heads; levels are high and virtually noncycling (Fig. 2C). The staining results indicate that TIM expression and the molecular clock more generally function quite well in the clock neurons despite the tim_up126 deletion.

The much stronger TIM expression pattern in the clock neurons could be due to altered transcriptional regulation in these cells or to some compensating regulation like enhanced protein stability. To distinguish between these possibilities, we used the single molecule in situ hybridization platform RNAScope to assay the effect of the tim_up126 deletion on tim mRNA levels in the clock neurons and in glia (39). We also considered that the effects of the tim_in24 deletion could be informative and therefore assayed this strain in parallel (40).

The in situ hybridization data parallel what was observed by immunohistochemistry (Fig. 3), namely, a modest effect of the tim_up126 deletion on tim mRNA levels in clock neurons and a strong effect in glia (Fig. 4). The modest effect of this upstream deletion is mirrored by a similarly modest effect of the tim_in24 deletion on clock neuron tim mRNA levels. In contrast, there is little or no effect of the intronic deletion on tim mRNA levels in glia, the opposite of the strong tim_up126 deletion effect in glia. We conclude that tim expression in glia relies predominantly if not exclusively on the 126 upstream enhancer region, whereas enhancer activity responsible for tim expression in clock neurons is shared between this region and the intronic region.

To address this clock neuron vs. glia tim enhancer distinction, we used ATAC-seq to characterize gene regulatory regions from purified cells. We began by examining the nSyb and Repo genes in clock neurons and in glia. Consistent with expectation, the nSyb gene has open chromatin only in clock neurons whereas the Repo gene has open chromatin only in glia (Fig. 5A).

We then focused on the tim regulatory region and how it differs in these same two cell types; we also assayed head chromatin (Fig. 5B). There are two broad accessible regions in all cases, one of which corresponds to the intronic CLK binding site (purple arrow) and the other to the 126 CLK binding site (green arrow). In the case of the head pattern (Fig. 5B), the relative size of these two regions (126 ≫ intron) resembles their relative signal in the head CLK ChIP assay (Fig. 1A). This indicates that the head ATAC-seq pattern simply reflects the much greater CLK binding to the 126 region than to the intronic region in heads and suggests that a similar CLK binding ratio occurs in glia. In fact, Repo-Gal4 cells have even less relative intronic region signal than heads (126 ≫≫ intron; Fig. 5B), suggesting that there is even less relative CLK binding to the intronic region of glia, perhaps because heads are comprised of mixed cell types—some of which may have more intronic CLK binding. The absence of any substantial intronic ATAC-seq signal provides an explanation for why there is little if any glial circadian gene expression without the upstream 126 region.

Clock neurons (Clk856-GAL4 cells) in contrast are very different: They have much more ATAC-seq intronic signal, comparable to the signal found at their 126 region (Fig. 5 B, Top). This provides a mechanistic explanation for why circadian neuron tim gene expression is substantial despite the 126 deletion: The intronic region alone is sufficient to recruit sufficient CLK in circadian neurons but not in glia (Fig. 5; Discussion).

We considered that there might be a developmental factor that opens or reflects a greater opening of the tim intronic E-box region of chromatin in neurons relative to glia. Consistent with this prediction is a striking open chromatin region over the intronic E-box in neurons but not in glia; this open region is present in published ATAC-seq patterns during fly embryonic development (Fig. 5C). As this region appears before the appearance of Clk gene expression (41), these temporal data suggest that another factor opens the chromatin at the tim intronic E-box region during embryonic development, which ultimately enables its subsequent interaction with much more CLK in circadian neurons than in glia.

Alternatively, a quantitative difference in CLK expression between adult clock neurons and glia might exist and contribute to the distinction in ATAC-seq patterns between these cell types. To address this second possibility, we overexpressed CLK in glia with UAS-CLK and Repo-Gal4. Unfortunately, only adult overexpression was possible. This was enabled by the auxin-inducible gene expression system (AGES) and was necessary because the simpler strategy of UAS-CLK expression with Repo-GAL4 throughout development was lethal; this was also the case with nSyb-GAL4; see below. Nonetheless, we considered that adult overexpression might still be effective because mammalian CLK has been shown to act as a pioneer transcription factor by opening closed chromatin (42).

The results indeed indicate that adult CLK overexpression in glia increased the ATAC-seq signal within the intronic E-box region of the tim gene (Fig. 6 A, Left, purple arrow; compare the change of these peaks to the no change in the flanking control peaks). The effect however is modest. Moreover, there is a comparable and perhaps even greater effect on the upstream 126 E-box region, indicating that the dramatic ratio of open chromatin between the two regions (126 ≫ intronic) is not appreciably altered. There is a similar modest positive effect on the ATAC-seq signals corresponding to the CLK-binding peaks within the vri gene regulatory region (Fig. 6B and Discussion), suggesting that the increase in CLK-binding peak signal is general. Although caveats exist (Discussion), this argues against a simple version of this second possibility.

To provide a more general context to the effect of adult CLK overexpression, we expressed UAS-CLK in most if not all adult neurons with nSyb-GAL4 as well as the AGES system (Fig. 6 C and D). The results were qualitatively similar to Repo-gal4 CLK adult overexpression, e.g., an enhanced ATAC-seq signal in the CLK-binding region within the vri gene (Fig. 6B). However, neuron overexpression resulted in a much larger signal increase in this region of the vri gene (Fig. 6C, indicated by purple arrows). CLK overexpression causes an even more dramatic signal increase in the relevant regions of the tim gene (Fig. 6C).

We interpret the dramatic quantitative difference between glia and neurons to the fraction of these cell populations that contain a molecular clock in wild-type fly heads; many, most, and perhaps all glia contain a molecular clock, whereas there are only 150 circadian neurons in the adult fly brain. Although this number is likely to be an underestimate of the number of neurons with a molecular clock, it is likely that most fly brain neurons do not contain a robust molecular clock. Consistent with the notion that the control nSyb ATAC-seq signal reflects the 150 neurons, the ratio of the two tim regions is approximately equal (Fig. 6C), similar to its ratio in clock neurons. We therefore suggest that the very large increase in ATAC-seq signal with nSyb-GAL4 represents the de novo recruitment of naïve adult brain neurons to become circadian neurons. This notion is based on the previously observed effect of ectopic CLK expression in flies (43, 44) as well as its chromatin remodeling properties in mammals (42).

The tim expression and clock gene cycling results (Figs. 3 and 4) as well as the ATAC-seq data (Fig. 5) beg the question, what are the deletion effects on circadian behavior? Not surprisingly perhaps, the key deletion strains have comparable decreases in rhythmicity, most notably the tim_up126 (P = 0.007) and tim_in24 (P = 0.01) strains (Fig. 7A), which is consistent with the comparable ATAC-seq signals on these two regions in clock neurons. This suggests that elimination of either enhancer reduces tim expression to approximately the same extent (Fig. 4), with a comparable effect on the rhythmic index. Although the data do not allow for a more quantitative assessment, it is interesting that a likely modest decrease in tim expression and molecular cycling amplitude within clock neurons affects rhythmic index rather than period length (Discussion).

The deletion effects on the circadian period are different from those on the rhythmic index. The upstream deletion strains have a slightly (0.5 h) longer period than WT with no statistically significant difference between them, whereas the tim_in24 strain is indistinguishable from WT (Fig. 7 B and C). We speculate that this distinction between the two regions may reflect a modest effect of the glial clock on the circadian period.

Flies were housed in standard cornmeal/agar medium with yeast under 12:12 h LD (light:dark) cycles. yw flies were used as the wild-type control. Flies used in this study are in SI Appendix, Table S1↗. CRISPR lines were generated by injecting pCFD3-dU6:3gRNA (Addgene, #49410) or pCFD4-dU6:3tandemgRNAs (Addgene, #49411) containing guide RNAs (gRNAs) corresponding to either the tim promoter or the intronic E-box sequences into embryos using Rainbow Transgenic (Cambridge, MA). tim_in24, tim_up10 were generated by a single gRNA. tim_up122, tim_up126 were generated by tandem gRNAs, the primers used to generate the mutants are listed in SI Appendix, Table S2↗. The CRISPR-generated deletions were validated by sequencing the region. Fly lines homozygous for the CRISPR-generated tim deletions were used for all subsequent biochemical and behavioral analyses.

Male and female flies were entrained in 12:12 LD conditions for 4 d at 25 °C, and then collected at ZT2, ZT6, ZT10, ZT14, ZT18, and ZT22. Total RNA from fly heads was extracted using TRIzol reagent (Invitrogen) following the supplier’s protocol. The resulting RNA was used to make libraries for transcriptome sequencing using the TrueSeq RNA Sample Prep Kit (v2; Illumina). The quality of the libraries was assessed using a Bioanalyzer 2100 (Agilent) and then sequenced on a NextSeq 550 (Illumina). RNA libraries were mapped to Drosophila genome dm6 using Tophat with default setting and expression levels were quantified using ESAT (48, 49). For qRT-PCR, the extracted RNA was quantified by Nanodrop (Thermal Fisher), 500 ng of total RNA was used to synthesize cDNA with a TAKARA kit following the standard protocol. The cDNA was diluted in water and used as a template for quantitative PCR using the qPCRBIO SyGreen Blue Mix (Genesee Scientific). Rpl32 served as an internal control.

Wild-type and tim_up126 mutant flies were entrained in 12:12 LD condition for 4 d at 25 °C, and then collected on dry ice every 4 h throughout the day. Fifty fly heads were homogenized in RIPA buffer with protease inhibitors (Roche) and phosphatase inhibitors (Fisher). Protein extract was heat-denatured in SDS buffer (200 Mm Tris-HCl pH 6.8, 8% Sodium Dodecyl Sulfate (SDS), 0.4% bromophenol blue, 40% glycerol) with 100 °C for 5 min. The samples were separated on a 3 to 8% Tris-acetate gel (Invitrogen) and transferred to nitrocellulose (iBlot; Invitrogen). Blots were blocked in 5% milk in PBST (3.2 mM Na2 HPO4, 0.5 mM KH2PO4, 1.3 mM KCl, 135 mM NaCl, and 0.05% Tween-20, pH 7.4) for 1 h and incubated overnight with either rat anti-TIM at 1:4,000 dilution, rabbit anti-PER at 1:4,000 dilution or mouse anti-Actin (loading control; Santa Cruz) antibodies. Western blot intensity was quantified using Image J.

ChIP-seq was performed on yw;; WT dCLK-V5 (wt) and yw; tim_up126; dCLK-V5 as previously described with the following exceptions (50). Three- to five-day-old flies were entrained in LD condition for at least 3 d and then collected every 4 h throughout the day. Fly heads were homogenized in homogenization buffer (10 mM HEPES-KOH (4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid) at pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 0.8 M sucrose, 0.5 mM EDTA (Ethylenediaminetetraacetic acid), 1 mM DTT (1,4-Dithiothreitol), 1× protease inhibitor, 1× phosphatase inhibitor cocktail) at 4 °C. Homogenates were loaded on equal volumes of sucrose cushion buffer (with 1.0 M sucrose and 10% glycerol in the homogenization buffer) and centrifuged in a HB-6 rotor (Sorvall) at 11,000 rpm for 10 min at 4 °C. Nuclear pellets were suspended in 1 mL 1× PBS with 1% formaldehyde and fixed for 15 min at room temperature and quenched by 0.125 M glycine. Chromatin was extensively washed by 1× PBS and sheared using a biorupter (Diagenode) in 500 µL sonication buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 10% glycerol, 1% SDS). For each IP, 25 µL of sheared chromatin was reserved as input sample. The remaining chromatin sample were diluted by 10 times of IP buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 10% glycerol, 1% NP40 and protease inhibitor tablet) and incubated with 30 µL anti-V5 agarose beads (Sigma) overnight at 4 °C. Beads were washed and eluted and decrosslinked as described previously. The DNA was purified using Minelute columns (Qiagen). Of note, 30 μL of DNA from each IP or 50 ng of input DNA were used to generate ChIP-seq libraries according to the Illumina ChIP-seq protocol. After adaptor ligation, DNA samples were separated by 2% agarose TAE (Tris-acetate-EDTA) gel and gel slices corresponding to 250 to 450 bps were recovered and purified using the QIAGEN gel purification kit. ChIP-seq libraries were sequenced on a Nextseq 550 (Illumina) and the resulting datasets were mapped to the Drosophila genome (dm6) using Bowtie2 and analyzed using MACS2 (51, 52).

First, 3- to 5-d-old male flies were placed into glass tubes containing 2% agarose and 5% sucrose and then entrained for 4 d in 12 h light:12 h dark (LD) conditions, followed by 4 to 5 d in constant darkness (DD) conditions using TriKinetics (Waltham, MA) DAM system (Drosophila Activity Monitors). Analyses were performed with MATLAB as described (53). A rhythmicity threshold of 0.3 was applied.

Immunostaining experiments were performed on 3- to 5-d-old male and female flies as previously described (54). Flies were entrained in LD conditions for 4 d before being subjected to constant darkness for 3 d. Flies were harvested for immunostaining at time points throughout the third day in constant darkness. Fly bodies were fixed in PBS with 4% (vol/vol) paraformaldehyde with 0.5% Triton X-100 for 2 h and 40 min at room temperature. Brains were dissected, washed twice in 0.5% PBST buffer and then blocked overnight in 10% Normalized Goat Serum (NGS; Jackson Immuno Research Lab) at 4 °C. The brains were then incubated in rabbit anti-PER at 1:1,000 dilution, a rat anti-TIM at 1:200 dilution or a guinea pig anti-PDP1 antibody at a 1:1,000 for 2 to 3 d (55–57). The brains were washed 3 times by PBST, then incubated with either Alexa Fluor 488-conjugated anti-rat or anti-mouse (1:1,000 dilution), and Alexa Fluor 633-conjugated anti-rabbit or anti-guinea pig at 1:500 dilutions in 10% NGS. Brains were mounted in Vectashield (Thermal Fisher) and imaged on a Leica SP5 confocal microscope, the z-stack was sequentially imaged in 1 μm sections. Image J was used for signal quantification.

Fluorescence in situ hybridization of Drosophila brains was performed using the RNAScope Multiplex Fluorescent Detection Kit v2 (Advanced Cell Diagnostics, ACD) using the protocol described in ref. 40 with some modifications. Flies were collected at ZT16 directly into 4% formaldehyde in PBST [PBS+ 0.5% Triton X-100] on ice and fixed for 5 h on a rotator at 4 °C in the dark. Brains were then dissected and fixed overnight in 4% formaldehyde in PBST on a rotator at 4 °C. Fixed brains were washed twice with 650 µL PBST and once with 650 µL PBST+ 1% BSA for 10 min on a rotator at room temperature (RT) before being incubated for 5.5 min in a 100 °C heat block in 550 µL prewarmed 1× Target Retrieval Solution (ACD). Brains were washed for 1 min in PBST+ 1% BSA at RT, for 1 min with 650 µL methanol at RT, then for 10 min with PBST+ 1% BSA at RT on a rotator. Brains were then postfixed with 500 µL 4% formaldehyde in PBST for 25 min on a rotator at RT and washed with PBST+ 1% BSA for 10 min on a rotator at RT. Brains were then incubated with 2 drops of Protease Plus solution (ACD) for 10 min at 40 °C. Brains were washed with 650 µL PBST+ 1% BSA for 10 min on a rotator at RT, then with 2 drops of Probe Diluent (ACD) for 1 min. Brains were incubated with 100 µL prewarmed Tim-C1 probe solution overnight at 40 °C.

Signal amplification and labeling were performed according to the manufacturer’s instructions with the following modifications. All incubations at 40 °C were performed with 2 drops of solution in a heat block, and all washing steps were done 2× for 5 min with 650 µL RNAScope Wash Buffer (ACD) on a rotator at RT. Opal 650 dye (Akoya Biosciences) was diluted 1:6,000 in TSA buffer (ACD) for conjugation to Tim-C1 probe. Brains were incubated in 300 µL diluted Opal 650 dye for 30 min at 40 °C.

For colabeling with immunohistochemistry, brains were washed 3× for 10 min on a rotator at RT with PBST after the last RNAScope Wash Buffer washes following incubation with HRP (horseradish peroxidase) Blocker Solution (ACD). Brains were then blocked in 200 µL 10% NGS on a rotator for either 2 h at RT or overnight at 4 °C. To prevent photobleaching of RNAScope signal, all longer incubations were done in the dark. Brains were then incubated with primary antibody [mouse anti-Repo, (DHSB) diluted 1:100 in 10% NGS] for 24 to 48 h at 4 °C. Primary antibody was removed and brains were washed with PBST 3× for 10 min on a rotator at RT. Brains were then incubated with secondary antibody [Alexa Fluor 488 goat-anti-mouse (Invitrogen) diluted 1:200 in 10% NGS] overnight at 4 °C on a rocker. Secondary antibody was removed and brains were washed with PBST 3× for 10 min on a rotator at RT and mounted in Vectashield (Vector Laboratories).

Images were acquired using a Leica Stellaris 8 confocal microscope with a 63× oil immersion objective. Z-stacks were acquired with 0.3 µm steps and 2,500 × 2,500 resolution.

Lightning deconvolution was performed on the tim RNAScope channel following acquisition using default settings. Images were imported into FIJI, where acquired images were exported in three separate tif files: two files with repo and tim RNAScope signal where images were adjusted to aid in segmentation, and one with unadjusted tim RNAScope signal for puncta quantification. Z-slices were carefully selected to avoid overlapping cells.

Segmentation was performed using Cellpose (58). For glia, segmentation was conducted on repo signal using the cyto model, manually fine-tuned using random Z slices across conditions. Masks produced by the fine-tuned model were subject to manual curation. For clock neurons, segmentation was performed manually, defining Clock neurons as the outline formed by high tim signal in regions without repo signal.

Puncta quantification was performed using FISH-quant v2 using default settings and a brightness threshold of 398 (59). Masks generated with Cellpose were used to separate signal into individual cells, with plots created for individual cells. Puncta assignments were manually verified. Statistical analysis and graphs were generated in R.

ATAC-Seq was performed on FACS-sorted clock neurons and glia using Clk856-Gal4 and Repo-Gal4, crossed to UAS-unc84-GFP, respectively. Two-week-old flies were collected at ZT02 and ZT14. Overexpression of CLK in neurons and glia was conducted using a UAS-CLK strain and the AGES-Gal80 system. Adult-specific CLK overexpression was achieved by feeding flies 10 mM auxin for 5 d. For Clk856-Gal4, Repo-Gal4, and nSyb-Gal4, brains were dissected in Schneider’s media and incubated with 0.75 mg/mL Collagenase and 0.4 mg/mL Dispase at room temperature for 30 min. Brains were then triturated 50 to 80 times to obtain a single-cell suspension for FACS.

Transposition on sorted cells was performed as previously described (60). Briefly, sorted cells and nuclei were centrifuged at 1,000 rpm for 10 min at 4 °C and the pellet was resuspended in 50 µL of Transposition mix (25 µL 2× TD buffer, 16.5 µL PBS, 0.05 µL 10% v/v Tween, 0.05 μL 1% v/v Digitonin, and 2.5 µL TDE1 Enzyme (Illumina, San Diego, CA, USA Catalog #20034198) and incubated at 37 °C for 30 min. Tagmented DNA was purified using Zymo DNA clean and concentrator kit. Purified DNA fragments were amplified using single-indexed Nextera primers (Integrated DNA Technologies) for 12 PCR cycles. Amplified libraries were purified using AMPure XP beads (Beckman Coulter, Brea, CA, USA Catalog #A63880), and size distribution was accessed using TapeStation High-Sensitivity D1000 Screentape. Final libraries were sequenced on a Nextseq 550 (Illumina).

ATAC-Seq files were adaptor-trimmed using fastp. Bowtie2 was used to aligned trimmed files using the following parameters: --local --very-sensitive-local --no-unal --no-mixed --no-discordant --phred33 -I 10 -X 700 (51). Samtools was used to remove PCR duplicates and Sambamba was used to remove multimapping reads (61). Tn5 insertion bias was corrected using a custom python script and peaks were called using MACS2 (62).

Peaks called from MACS2 were intersected between replicates using bedtools, and peaks across conditions were merged and converted into a reference annotation in SAF format (63). Featurecounts was used to count reads in each sample within each region in the reference file, and input into DESeq2 in R. DESeq2 was used to perform differential peak quantification, and normalization factors were produced as well. Normalization factors were used to produce normalized bigwig files for visualization using Deeptools bamcoverage (64).

D.M. and M.R. designed research; D.M., P.O., A.D.Y., M.S.A., W.L., and M.M.D. performed research; D.M., P.O., A.D.Y., M.S.A., W.L., E.K., M.M.D., M.W., and W.J.J. analyzed data; M.R. supervised research; and D.M., K.C.A., and M.R. wrote the paper.

The authors declare no competing interest.

The raw and processed sequencing data have been deposited in GEO under accession numbers GSE259243↗ (ATAC-Seq) (65), GSE259245↗ (RNA-Seq) (66), and GSE259247↗ (ChIP-Seq) (67).