The emergence of circadian timekeeping in the intestine

During post-embryonic development the intestine undergoes significant growth and remodeling (Fig. 1B). We first determined if the circadian clock affects intestinal development by comparing isogenic wildtype and per¹ mutant flies that exhibit no clock rhythms. Intestinal size, cell density, and the number of cells in AMP / PC islets do not differ significantly in per¹ mutants (Supplementary Fig. 2A, B), indicating that intestinal development is not affected in the absence of circadian clock activity. To determine when Clk/cyc transcriptional activity begins, we used two different reporter constructs (Clock^PER and Clock^TIM) inserted on different chromosomes that use destabilized GFP (dGFP) to visualize Clk/cyc transcriptional activity on per and tim promoters with ~1 h temporal resolution²⁸ (Fig. 1C). The intestine was visualized using the GAL4/UAS system (Myo1A>mCherry) through the marker Myosin 31DF (Myo1A) that is expressed in both larval and adult intestinal cells³⁷. No Clk/cyc transcriptional activity is present in larval (L1, L2, L3) or pupal flies, however, immediately after eclosion (adult day 1), Clk/cyc become active and remain so in later adulthood (Fig. 1D–F, Supplementary Fig. 2C, D) irrespective of sex or reproduction status (Supplementary Fig. 2E–G. To verify our reporter, we probed for per protein using both an antibody and a C-terminal GFP fusion reporter (per-AID-eGFP)³⁸, and cry protein using an N-terminal fusion protein for cry (cry-GFP)³⁹. In all cases, these clock proteins are absent in pupa but present in recently eclosed adults (Fig. 1G and Supplementary Fig. 3A–C). We also verified that the clock is not present in larva and pupa, in a circadian phase opposite to the adults, by checking expression at different timepoints (Supplementary Fig. 3D). These results indicate that circadian clock activity begins only after pupal metamorphosis in the Drosophila intestine.

We next used RT-qPCR to examine the transcriptional expression of the core clock genes (Clk, cyc, per, tim), the secondary stabilizing loop (Pdp1, vri), the blue-light photoreceptor (cry), and doubletime (dbt), a kinase that regulates per to reset Clk/cyc driven transcription^40,41. The intestinal expression of Clk, tim, Pdp1E, and vri increase throughout development to peak in adulthood, whereas dbt, cyc, per, and cry expression increase earlier in development before the stage when Clk/cyc activity exists (Fig. 1H and Supplementary Fig. 3E). Therefore, we hypothesized that the changes in clock activity are likely due to changes in Clk, tim, Pdp1 and/or vri.

When we looked more closely at Clk/cyc activity in the last stages of pupation we noticed few scattered GFP cells (Supplementary Fig. 3F), raising the question of which intestinal cell type expresses Clk/cyc activity first. We measured Clk/cyc activity in pupal and adult ISC/EBs (marked by esg>mCherry) and ECs (marked by Myo1A>mCherry) and found that these early ISC/EBs show similar Clk/cyc activity in the late pupa and adults, unlike ECs that have significantly higher Clk/cyc activity in adults, suggesting that the ISC/EBs establish their circadian clock first (Supplementary Fig. 3G, H). Moreover, when the circadian clock is absent in all cells except ISC/EBs (cyc⁰ clock-dead mutant with esg>cyc rescue) clock activity persists in the ISC/EBs indicating that these cells can develop clocks independently of the other clocks in the body (Supplementary Fig. 3I).

Since the intestinal clock emerges after pupation, we asked whether it is rhythmic immediately, with daily maxima and minima that are characteristic of circadian rhythms, or whether it needs time to synchronize to the environment. We tested wildtype CantonS fly intestines by RT-qPCR to compare adult day 1 (immature) and day 4 (mature) gene expression patterns over a 24-h period. On day 1, the transcripts of Pdp1 and vri are not rhythmic suggesting that the transcriptional cycle is not yet present. These genes increase in rhythmicity three days later, with cry changing the timing of its initial rhythms (Fig. 2A). Of note, the expression of all clock genes, with the exception of cry, are lower on day 1 than day 4, suggesting that establishment of robust transcriptional cycles takes several days.

The gradual emergence of transcriptional rhythmicity suggests Clk/cyc transactivation is not synchronized until the intestine matures. To test this, we imaged the Clock^PER and Clock^TIM reporters and measured fluorescence intensity in the whole intestine following pupation every 3-h over the first four days of adulthood (Fig. 2B, C and Supplementary Fig. 4A, B). When the adult ecloses, Clk/cyc transcriptional activity is present, but its rhythms are not robust 24-h circadian oscillations until day 4 when these consolidate to exhibit a maximum in the morning (ZT21-ZT0) and minimum in the evening (ZT9-ZT12) characteristic of the Drosophila intestine²⁸. In each individual midgut, we did not notice major differences between cells which would indicate that the clocks in individual cells are at drastically different phases. To further test the synchrony between individual cells, we quantified EC fluorescence and noted that, although weaker rhythms are present at day 1, robust EC-specific rhythms are consolidated on day 4 (Supplementary Fig. 4C). This suggests Clk/cyc activity at both per (Fig. 2B) and tim (Fig. 2C) promoters synchronize over the first three days in the adult intestine.

A defining characteristic of circadian rhythms is their ability to continue in the absence of environmental cues such as photoperiod⁴². Accordingly, Clock^TIM intestines were examined when flies were shifted to constant darkness (DD) prior to ZT0 on day 4 and tracked for 48-h (Fig. 2D). The rhythms are similar to those under LD conditions, with a maximum in the morning (CT21-0) and minimum in the evening (CT12), demonstrating that Clk/cyc activity is free-running at day 4.

In a functioning circadian system, clock proteins shuttle between the nucleus and cytoplasm⁴³. We tested a GFP reporter, per-AID-eGFP³⁸ at two timepoints ZT0 and ZT12 that represent nuclear and cytoplasmic localization of per, respectively²⁷. Larva and pupa do not show any expression of per consistent with our qPCR and reporter analysis; the immature adult (day 1) shows signal at ZT0 that remains at ZT12, suggesting that the protein shuttling is not yet established, while the mature adult (day 4) shows strong nuclear signal at ZT0 but not at ZT12 (Fig. 2E). Taken together, our results show that circadian clock development in the intestine has three distinct phases: (1) embryogenesis to late pupa when the circadian clock circuit is absent, (2) between adult day 1 to 3 when Clk/cyc initiate clock rhythmicity but not yet fully synchronized to the environment, and (3) day 4 when the circadian clock is synchronized and rhythmic in the mature intestine.

Previous studies have not been able to examine circadian clock development in individual cells of a tissue, hence it is not clear if clock components mature homogenously, with individual cells gradually expressing all clock genes, or heterogeneously, with individual cells expressing certain clock genes but not others. To further test the development of the circadian pacemaker, we profiled transcriptional changes in CantonS intestinal development over each of the three stages in clock development, from early pupa, immature adults (day 1), and mature adults (day 7–8) using single-cell RNA sequencing (scRNA-seq). Early pupal intestines do not have Clk/cyc activity, immature adult intestines have Clk/cyc activity but no rhythms, and mature adult intestines have rhythmic Clk/cyc activity; we hypothesized that clock development would increase homogenously in all intestinal cell types over these developmental stages. We recovered 5190 cells (2013 pupal, 1197 immature adult, and 1980 adult cells) that show 23 different cell states (Fig. 3A and Supplementary Fig. 5A, B), consistent with previous scRNA-seq studies of the Drosophila intestine^35,44–47, capturing the changing intestinal transcriptome throughout development. Our data provides a resource for cellular development in the insect intestine.

During the transition from pupa to mature adult, the cell population of the Drosophila intestine undergoes cellular changes (Fig. 3B). The ISCs and EEs are present at all stages but ECs are more restricted to later stages. A population of cells we denote as Pupal Cells (PupC) decrease in the immature adult and are completely absent in the mature adult. PupCs express some genes found in the EEs and ECs of the adult (Supplementary Data 4)⁴⁵ and are likely derived from AMPs but represent a separate cell type only present in pupae (Supplementary Fig. 5C), consistent with previous reports that show transient pupal midgut cells disappear during metamorphosis²⁵. Several subtypes of differentiated ECs are not present in the pupae or immature adult and are only present in the mature adult, consistent with the maturation and growth of the Drosophila intestine⁴⁸. To assess circadian gene expression in these cells, we surveyed gene counts in individual cells (Supplementary Fig. 5D and Supplementary Data 1–3). We were able to detect all core clock components (Supplementary Fig. 5D) whose expression at the population level increases as pupa transition to mature adults consistent with the RT-qPCR data (Fig. 1H and Supplementary Fig. 3E). We detected the genes per, Clk, and cyc at much lower levels than tim, Pdp1, vri, dbt, and cwo, a repressor of Clk/cyc. We therefore used the higher-expressed clock genes as a readout of clock development.

Previously, we reported that ISCs, EBs, and ECs, but not EEs have Clk/cyc transcriptional rhythms²⁸. We therefore asked whether individual cells, including EEs, during these developmental stages express certain clock genes, but not the complete clock system. To test this, we focused on clock target genes that are more highly expressed and also feedback to regulate the core clock (cwo, tim, Pdp1 and vri). We graphed all four genes in a multi-dimensional fashion to simultaneously determine changing expression of all four in individual cells (tim is y-axis, cwo is x-axis, vri is size, Pdp1 is color). Individual ISCs show low expression of all four genes in the pupa, increasing co-expression in the immature adults, that further increases in the mature adults (Fig. 3C). This suggests that the arrhythmic Clk/cyc activity on day 1 is, at least in part, due to low expression of the circadian clock program in individual ISCs. Individual ECs show very similar patterns of change, gradually single cells develop that express all four clock genes at higher levels. In contrast, clock genes are not expressed in individual EBs and EEs in the same manner. First, these do not show the robust expression of multiple clock genes at any developmental stage including the mature intestine. For example, only 10% of mature adult EBs coexpress 3–4 of the clock genes compared to 98% of mature adult ISCs. Second, in many cells one or two clock genes are high (Supplementary Fig. 5D), but the remainder are low, suggesting clock components may have non-clock functions in specific subtypes of EEs or EBs but are not co-expressed as part of a circadian expression program. The highest clock gene expression is present in ISCs, that increase expression in the immature and mature adult intestine, whereas the other intestinal cell types showed lower expression levels. Finally, individual PupCs do not express clock genes or express few at very low levels, which is consistent with the absence of Clk/cyc activity in the pupal intestine (Fig. 3D). We conclude that ISCs and ECs increase clock gene expression after pupation, supporting a model where clock gene expression that is absent earlier in development increases in single cells homogenously to reach its maximum in the mature adult intestine.

The sudden emergence of Clk/cyc activity in the immature adult intestine (Fig. 1D–F) suggests that either a regulatory factor initiates the expression of circadian clock genes in the adult, or else suppresses them in the pupa, until the appropriate developmental stage is completed. Hormone nuclear receptors control many aspects of Drosophila physiology and metamorphosis^49,50 and are intimately connected with circadian clock function⁵¹. We therefore hypothesized that clock development is coordinated by hormone nuclear receptor signaling in the developing intestine. We used SCENIC⁵², a bioinformatic approach, that evaluates scRNA-seq transcriptomes to identify active gene networks and infer the underlying transcriptional regulators, to analyze pupal, immature, and mature adult cells. Identification of cell-specific transcription factors revealed enrichment of klu and Sox21a in ISCs, known to regulate ISC differentiation/function^45,53–55, the Notch target E(spl) in ISC/EBs, known to drive ISC differentiation^56,57, nub/Pdm1 in ECs, a well-known marker of this cell type^23,24, tap in EEs⁵⁸, known to promote EE cell fate⁵⁹ (Fig. 4A). Hence SCENIC is able to identify cell-specific transcriptional networks in the Drosophila intestine accurately.

To infer regulators of clock differentiation we focused on the ISC population, since it is present at all three stages in intestinal development (Fig. 3B and Supplementary Fig. 5A). We evaluated differences in regulon activity between the early pupa and immature adults to specifically address the changes occurring during these developmental stages as the clock first emerges (Fig. 4B and Supplementary Data 4). This analysis identified increased activity of 13 transcription factors, and decreased activity of 17 others during pupation. These include known ISC regulators Stat92E^60–64, and GATA⁶⁵, the hormone signaling factors gce/Met (germ cell-expressed bHLH-PAS/Methoprene-tolerant)^66,67, and EcR (ecdysone receptor)^68–70 that are active during metamorphosis^69,71,72, and the clock gene Pdp1. Many other factors were identified that are specific to different cell types in the developing intestine. Our dataset provides a resource for the global transcriptional changes that occur during the growth and development of the Drosophila ISC model system.

Drosophila pupation consists of sequential hormonal pulses with rhythms in adult emergence occurring through circadian control of the EcR⁷³. We find that EcR and the nuclear receptor ftz-f1 (fushi tarazu transcription factor 1), are downregulated in the immature adult ISCs, whereas gce and Met, both of which are involved in Juvenile hormone signaling during metamorphosis^66,67, are upregulated (Fig. 4B, C). Some of these differentially expressed regulons also show clear bimodal distributions indicating that there are distinct cell populations, with and without activity of these transcription factors (i.e. gce, Pdp1, Fig. 4C). This analysis provides a set of potential candidates that regulate the initiation of circadian clock activity.

To test whether these and other hormone nuclear receptors regulate clock development, we performed a screen by depleting genes in a cell-specific manner. Using our scRNA-seq analysis as a guide, we depleted components of hormone signaling in ISC/EBs (esg^TS, 17 genes) using RNAi in Clock^TIM reporters during pupation (Fig. 5A and Supplementary Data 5). We reasoned that the loss of positive regulators of clock differentiation would delay Clk/cyc activity, while negative regulators would hasten it. We found that (as predicted by SCENIC) Clk/cyc activity was reduced by Hr78, Hnf4, and gce knockdown, however, others such as Met, and ftz-f1 did not (Fig. 5A). Several other nuclear receptors were identified that also decrease Clk/cyc activity, but none were found that show an increase. To further test the requirement for hormonal signaling in activating clock gene expression, we overexpressed a subset of nuclear receptors identified in both the SCENIC analysis (Fig. 4B) and tested by RNAi (Fig. 5A). None of these were sufficient to accelerate Clk/cyc activity in pupal stages, suggesting that these pathways are required for circadian maturation but alone are not sufficient to prematurely drive clock emergence (Fig. 5B). However, after eclosion Clk/cyc activity is significantly higher when Hnf4 or Hr78 are overexpressed, suggesting an important role in regulating intestinal clock development (Fig. 5B). Of note, EC reporter expression is also increased when these genes are overexpressed in ISCs, possibly due to the higher clock activity in their precursors. Since common regulatory pathways might affect subsequent clock activity in differentiating ECs as well, we retested EC-specific knockdowns of the same genes tested in ISC/EBs. Consistent with ISC/EBs, components of Hnf4 signaling in ECs decreased Clk/cyc activity on the first day after eclosion, compared to the control (Fig. 5C).

Our SCENIC analysis also suggests that Pdp1 may be a factor, however, Pdp1 loss (Pdp1³¹³⁵ mutant⁷⁴) does not affect initiation of Clk/cyc activity (Supplementary Fig. 6A). This suggests that Pdp1 is not required for clock development, consistent with a role for Pdp1 in regulating clock outputs without being necessary for Clk expression⁷⁵. Together, these data are consistent with the notion that multiple hormone-signaling pathways in the intestine during pupal development regulate the emergence of the clock and highlight Hnf4 signaling of particular importance to circadian clock development.

EBs are progeny of ISCs that differentiate into ECs^23,24, but do not express clock genes robustly (Fig. 3C). We therefore tested clock gene expression in the mature adult ISC to EC lineage. To distinguish between the steps in differentiation, we re-clustered and ordered adult cells from ISCs (cluster 0), EBs (clusters 2, and 3), and three EC populations (clusters 7, 8, 14) to reconstruct these stages in differentiation (Supplementary Fig. 6B). We then followed the longest lineage (ISC-0 to EC-7) to determine whether clock gene expression changes during EC differentiation (Fig. 6A). The expression of a control housekeeping gene, RpL32, is present in all clusters, whereas the gene klu that correlates with the ISC identity^45,76, and the EC marker gene Amy-p⁴⁵ are enriched in their respective populations (Fig. 6B). Clock gene expression was mapped on the same lineages and is reduced during the different stages of EB differentiation: Pdp1, tim, vri, and cwo are initially high in ISCs, then temporarily reduced in the EB populations, re-emerging in ECs when these complete differentiation (Fig. 6B, C). Since nuclear receptors play a role in activating Clk/cyc during development, we tested the expression of hormone signaling components in ISC lineages. Hnf4, Eip75B, and gce also show high expression in ISCs, low in EBs, returning in ECs (Supplementary Fig. 6C). This suggests that nuclear receptors are correlated with clock activity during adult ISC differentiation.

These data indicate that clock genes are repressed in the ISC-to-EC lineage, hence we predicted that individual ISC clones should show Clk/cyc active and inactive cells. We used CD2-flp/out flies, where each CD2-marked clone consists of an initial marked ISC and all of its progeny, to lineage trace individual ISCs with a Clock^TIM reporter. The clones revealed large GFP+ ECs, and a mixture of small GFP+ and small GFP- cells, consistent with the idea that a small GFP + ISC produces small GFP- EBs that go on to differentiate into large polyploid GFP+ ECs (Fig. 6D and Supplementary Fig. 6D, E)²⁸. We further tested this idea by examining Clk/cyc activity specifically in only ISCs (esg^TS, Su(H)GBE-Gal80>mCherry), only EBs (Su(H)GBE>mCherry), or only ECs (Myo1A^TS>mCherry). We predicted that EBs should show lower Clk/cyc activity than their founder cells or their fully differentiated progeny. Indeed, Clk/cyc activity is present in ISCs, absent in EBs, and highest in ECs (Fig. 6E–G). These data are consistent with our scRNA-seq analysis and suggest the transient disappearance of circadian clock activity during lineage commitment of ISC to ECs in the adult intestine (Fig. 6H).

Our results indicate that the intestinal circadian clock gene expression is established during development, however, it is not clear what factors cause the intestine to exhibit synchronous circadian rhythms. The intestine completes its growth and development after adult emergence from pupation^35,48, while it is being synchronized to the environment (Fig. 2B, C). We have previously shown that the Drosophila intestinal clock is synchronized by photoperiod light cycles, with food ingestion also playing a modulating role in timing daily rhythms²⁸. We first asked if photoperiod would be sufficient to synchronize the immature intestinal clock. Flies were completely starved for the first three days of adulthood following pupation and provided food only at ZT0 on adult day 4 when clock synchronization is complete. Even though intestinal size was reduced by lack of nutrients (Supplementary Fig. 7A, B), Clk/cyc activity was robust, and synchronized rhythms emerged one day earlier (day 3 instead of day 4, compare Fig. 7A to Fig. 2C), suggesting that feeding and/or intestine growth delays clock entrainment.

We then performed the opposite experiment, in which photoperiod is absent and feeding is present, by using cry¹ mutants that are unable to detect light to synchronize clock function⁷⁷. We predicted that feeding would synchronize circadian rhythmicity. Without photoperiod, cry¹ intestines exhibit arrhythmic Clk/cyc activity during the first three days but show Clk/cyc activity at day 4 onward similar to wildtype controls (Fig. 7B compare to Fig. 2C). Since light-responsive feeding behavior occurs in cry¹ flies⁷⁸ these data suggest that feeding alone can also complete circadian clock synchronization in the intestine. To further test this, we restricted feeding in cry¹ flies post-pupation to the morning only (ZT0-6), evening only (ZT5-11), or subjected them to alternate days of feeding and starvation to disrupt feeding rhythmicity. We predicted that the timing of food consumption would alter the maxima and minima of intestinal circadian rhythms. Indeed, under these conditions the rhythms of cry¹ intestines were shifted to peak around 6 h after feeding, or rendered non-circadian when disrupted (no clear 24-h period) (Fig. 7C). In contrast, control flies that have the ability to detect photoperiod do not show a shift in Clk/cyc activity on restricted feeding, suggesting that photoperiod is dominant over feeding if both environmental cues are detected (Supplementary Fig. 7C). We further tested if periods of fasting would affect clock rhythms once these are established in adults and noted that these do perturb the clock if photoperiod is detected (Supplementary Fig. 7D, E). We conclude that the combination of daily photoperiod and feeding cycles together synchronize the timing of intestinal circadian transcription cycles during clock maturation.

Drosophila melanogaster strains in this study include: Clock^PER, Clock^TIM described in²⁸. From Jadwiga Giebultowicz: CantonS and per¹(isogenic with CantonS). From Chrysoula Pitsouli esg-Gal4 (II), tubGal80^TS(III). From Yong Zhang PER-AID-GFP. From Steve Jean Su(H)GBE-Gal80, tubGal80ts(III). From Amita Sehgal Pdp1³¹³⁵(III). From Bloomington: #38424 UAS-mCherry (III), UAS-mCherry(II), myo1A-Gal4 (II), #42563 Valium20-cyc(II), cry¹, #76317 CRY-GFP, #4412 act > y + >Gal4,UAS-CD2, RNAi lines: #50712 and #58286 EcR (#1 and #2), #43231 Eip75B, #26718 Eip78C, #61937 and #50868 ERR, #27659 ftz-f1, #61852 gce, #29375 and #64988 Hnf4, #51442 and #27254 Hr3 (#1 and #2), #29376 and #29377 Hr38, #33624 and #27086 Hr39, #54803 and #31868 Hr4, #31990 Hr78, #31603 Luciferase, #26205 met, #36095 tai, #36729 and #27258 Usp (#1 and #2). From FlyORF: F001915 gce, F000656 Eip78C, F000144 and F003392 Hnf4 (#1 and #2), F000423 and F003303 Hr78 (#1 and #2), F004847 and F002207 Eip75B (#1 and #2), F003474 met. Flies were balanced with yw;IF/CyO;MKRS/TM6B. Experimental genotypes are provided in Supplementary Table 1. Flies were raised on cornmeal-glucose media (1.2% w/v yeast, 0.7% w/v soyflour, 5% w/v cornmeal, 0.4% w/v malt, 0.4% v/v agar, 5.3% v/v glucose with propionic acid and tegosept) at 25 °C with 12:12 light/dark photoperiod (lights-on at ZT0). For free-running experiments flies were shifted to constant darkness (DD) during the dark-phase of the final day of LD, CT0 (Circadian Time 0) is the time when lights are expected to turn on. For larval to adult analysis, flies were only allowed to lay eggs for 5 h in the morning (ZT0-5). For adult experiments, only flies that eclosed overnight (ZT12-0) were included, no difference was noted between ZT11-12, ZT23-0, and ZT0-1 eclosion groups. Heatshocks for Flp/out clones were done with adult flies 5–14 days old for 5–10 min at 37 °C. For RNAi and overexpression screening (Fig. 5), flies were raised at 18 °C until pupal stages then vials were shifted to 29 °C and both males and females were collected. Age ranges from first larval instar to adults (day 1–14).

Pupa were transferred to vials containing a piece of water-moistened filter paper using a paintbrush. Flies eclosed onto this filter paper and were kept on water after eclosion either (1) until adult day 4 at ZT0 when flies were transferred to vials containing normal food or (2) for 6 h (from either ZT5-11 or ZT0-5) each day after eclosion. The alternate days of feeding and starvation (Fig. 7C) were done by allowing flies to eclose onto water-moistened filter paper, similar to the restricted feeding, then at ZT0 were flipped onto food (“F”), then the next morning at ZT0 back onto filter paper with water for starvation (“S”), then repeated F (day 3), S (day 4). This alternating cycle was necessary, since unlike the controls the cry¹ flies did not survive the starvation for 2–3 days.

Larva were roughly staged using the following times: 1 day AED (after egg deposition) for L1, 2 days AED is L2, 3–5 days AED is L3 (the day AED may be noted e.g., L3 day 3 refers to three days AED). Early stage pupa were analyzed at the red-eye stage (day 8 AED), and late stage pupae were analyzed at the gray to black wing stage (day 9 or 10 AED)¹⁰⁷. These stages were verified visually to account for individual variation in developmental timings. The first day after eclosion is referred to as adult day 1 (day 9 or 10 AED).

Flies were euthanized in 70% EtOH and then intestines were dissected using fine forceps in 1xPBS. For RNA extractions and single cell analysis, the hindgut (anterior to the Malpighian tubules) and the foregut were removed during dissection to ensure only the intestine was sampled. Dissections were completed in <1 h. Analysis from larva to adult included both males and females but unless otherwise noted, only female flies were analyzed.

Intestines were fixed in 4% paraformaldehyde for 40 min, washed twice in 1xPBS and stained with DAPI (1:5000 in 0.2% Triton-X100 (in 1xPBS)) for 5 min. The intestines were washed twice in 0.2% PBS-T and then mounted on slides using AlexaProLong Gold Antifade Reagent (Invitrogen). For antibody staining, intestines were fixed for 2 h in 4% paraformaldehyde and then washed twice in 1xPBS and blocked for 20 min in 1% bovine serum albumin in 0.2% PBS-T (BSA). The intestines were then incubated in primary antibody (PER (1:1500), histone (1:2000), GFP (1:2000), rat-CD2 (1:2000)) in BSA for 2 h, washed twice in 0.2% PBS-T, incubated in secondary antibody (1:2000) with DAPI (1:5000) in BSA for 1 h and then washed twice in 0.2% PBS-T (at 4 °C) and mounted on slides. Cry-GFP was stained 1-h prior to lights on (ZT23) to avoid light-induced degradation of cry³⁹.

Approximately 15 female flies of CantonS were dissected in 1XPBS before being transferred to tubes containing RNAlater (Qiagen) on ice. RNA was extracted using the RNAMini Kit (Qiagen) with RLT buffer (Qiagen) using a Bullet Blender (Next Advance) for homogenization. cDNA was made using ISCRIPT RT Supermix (Bio-Rad) and RT-qPCR was performed using SYBR green (Bio-Rad) using the Viia7 PCR plate reader, primers: per-F TCATCCAGAACGGTTGCTACG, per-R CCTGAAAGACGCGATGGTGT²⁷, tim-F CCAGCATTCATTCCAAGCAG, tim-R GCGTGGCAAACTGTGTTATG²⁷, Gapdh-F CCAATGTCTCCGTTGTGGA, Gadph-R TCGGTGTAGCCCAGGATT, Clk-F CAAGTTTGGCCTCTGGCTCTC, Clk-R TACAACTAGCTCTGGGCTTCCG, Pdp1-F GAACCCAAGTGTAAAGACAATGCG, Pdp1-R CTGGAAATACTGCGACAATGTGG¹⁰⁸, vri-F TGTTTTTTGCCGCTTCGGTCA, vri-R TTACGACACCAAACGATCGA¹⁰⁸, cyc-F GCGCTGATGGAGTCTCACAAG, cyc-R GTAGCTGTTGTCCTTGCACCG, cry-F CACCGCTGACCTACCAAA, cry-R GGTGGAAGCCCAATAATTTGC¹⁰⁹, dco-F TTGGGAGGAGGGTTAGCAG, dco-R TTACAATGTGGGTGCCTTGC.

Flies from CantonS were euthanized in 70% ethanol in DEPC-treated water at ZT0. Every 10 min during the 1 h dissection time, the intestines were transferred from the 1XPBS well to ice-cold 1% BSA. Once dissections were complete, the intestines were transferred to a drop of 1XPBS on the back of a dissection plate and cut into small pieces using microscissors. The fragments were then transferred to a 1.5 mL tube to a total volume of 350ul (200uL for pupa). Elastase was added to a concentration of 1 mg/mL and incubated at 37 °C for 40 min pipetting up and down 30–40 times every 10 min to dissociate the cells, then 10% BSA was added to a final concentration of 1% BSA. The samples were centrifuged at 300 x g for 15 min at 4 °C, then the pellet was resuspended in 200ul of 0.04% BSA and filtered with a 70um filter centrifuging at 300 x g at 4 °C for 1 min. Orbitrap Density Gradient was used to sort the cells and then the interface was washed with 0.04% BSA. The cell density was determined with a haemocytometer and diluted with 0.04% BSA to a density of 800cells/uL. All samples had a viability >95%. The entire preparation was completed within 2 h. The fresh samples were submitted to the High Content Analysis Core at the University of Alberta by ZT3 for 10X Genomics 3’ Library preparation and to Novogene for sequencing (NovaSeq) and then aligned to the Drosophila reference genome 6.25 using Cell Ranger (10X Genomics, version 3.0.2).

Slides were imaged with either a slide scanner (Zeiss, Axio Scan Z.1) or confocal microscope (Olympus, IX81 FV1000 with 60x Water Immersion Lens or Zeiss Confocal with 20x lens). Whole mount imaging and quantification were done using a fluorescence microscope (Zeiss, VertA.1) or fluorescence stereoscope (Leica M205). Images were analyzed with Zen Blue Software (Zeiss, version 2.3) and processed using Photoshop (Adobe) with adjustments consistent within a panel. Fluorescence intensity measurements are ratios of GFP:DAPI taken for the whole midgut, unless otherwise specified, as previously described¹¹⁰. RT-qPCR data is normalized to Gapdh1. Prism (GraphPad) was used for graphing and statistical analysis. Cosinor analysis was performed using circacompare¹¹¹ and graphed using Prism (GraphPad). Schematics were prepared using Photoshop (Adobe) and Illustrator (Adobe).

Single sequencing data was analysed in R using Seurat (version 4.3.0.1)¹¹² and pseudotime analysis was performed using Slingshot (version 2.8.0)¹¹³. The original dataset of 6835 adult, 2625 pharate, 3816 pupal cells was filtered (genes must be detected in > = 3 cells, %mitochondrial <25, 200<nFeatures<4000, 1000<nCount<100000) and the remaining 1980 adult, 1197 pharate, 2013 pupal cells were used for further analysis. The datasets were integrated (reduction = rpca) and clustered (PCs = 14, resolution = 0.5) and then annotated using differentially expressed genes (method = roc). For lineage tracing 1111 adult cells consisting of clusters 0,2,3,7,8,14 were re-clustered (PCs = 9, resolution = 0.3) or 1933 pupal cells from clusters (0–6,8–14,16,17,18,21) were re-clustered (PCs = 16, resolution = 0.3) prior to lineage tracing to further subdivide these cells. Differential gene analysis was performed using roc. SCENIC analysis (version 1.3.1)⁵² was performed with version 10 Drosophila motif set with minimum number of genes in each regulon = 10, then handed back to Seurat for differential gene expression analysis. In some cases, Prism (GraphPad, version 9) was used for graphing and statistical analysis. Images were processed using Photoshop (Adobe, version 23.4.2) and schematics were made using Photoshop and Illustrator (Adobe, version CS5).

Statistical tests were performed using GraphPad Prism and can be found in Supplementary Information.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary

Nature Communications thanks John Ewer and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

The scRNA-seq data generated in this study have been deposited in the NCBI GEO database under accession code GSE230572↗ [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=gse230572↗]. The alignments were made using Drosophila reference genome 6.25 and are publicly available as of the date of publication. Source data are provided with this paper.