miR-124 acts during Drosophila development to determine the phase of adult circadian behavior

Abstract

The circadian clock enables organisms to optimize their metabolism, physiology, and behavior with the time-of-day. However, circadian rhythms benefit organisms only if they are properly synchronized with the day/night cycle; circadian misalignment can have detrimental effects on animals’ wellbeing and survival. We previously showed that in Drosophila, loss of the microRNA miR-124 advances the phase of circadian evening locomotor activity by several hours under constant darkness conditions. Interestingly, we now report that loss of miR-124 also delays morning activity under a light/dark cycle with a short photoperiod. We recapitulated these opposite phase phenotypes by eliminating miR-124 during larval development, but not when this microRNA is lost during pupation to adulthood. The loss of miR-124 results in significant miswiring within the circadian neural network and severely alters neural activity rhythms in the ventral Lateral Neurons (s-LNvs) and the posterior Dorsal Neurons 1 (DN1ps), which control the timing of morning and evening activity. Silencing the s-LNvs in miR-124 mutant flies restores the phase of evening activity, while activating the DN1ps rescues the phases of both morning and evening activities. Our findings thus reveal the pivotal role of miR-124 in sculpting the Drosophila circadian neural network during development and its long-lasting impact on circuit activity and adult circadian behavior.

Introduction

Most living organisms, including bacteria, fungi, plants, insects, mammals, and humans, experience daily changes in their surrounding environment, such as variations in light and temperature (Panda et al., 2002; Bhadra et al., 2017). To adapt to and anticipate these environmental cycles, they have evolved an internal timekeeping system known as the circadian clock. This system helps organisms orchestrate their temporal relationship with the ever-changing environment, enhancing their efficiency in resource utilization and activity scheduling. However, circadian clocks remain rhythmic with a period of ca. 24-hour even under constant environmental conditions, such as constant darkness (DD). Thus, this intrinsic rhythm allows organisms to sustain biological processes that rely on timing, ensuring continuity in their behavioral patterns and physiological responses, even in the absence of external Zeitgebers (Pittendrigh and Minis, 1972; Sehgal, 2017; Patke et al., 2020). These circadian clocks consist of self-sustained transcriptional feedback loops that function as endogenous oscillators, driving the rhythmic expression of a wide range of genes involved in diverse biological processes, from metabolic pathways to physiology and complex behaviors (Young and Kay, 2001; Zhang et al., 2014). In Drosophila melanogaster, the molecular clock comprises three interconnected feedback loops (Somers et al., 2018). In the core loop, the transcriptional activators CLOCK (CLK) and CYCLE (CYC) dimerize via PAS domains to drive the transcription of two genes encoding transcriptional repressors, PERIOD (PER) and TIMELESS (TIM). These repressors form a complex that translocates into the nucleus and represses CLK/CYC activity (Benito et al., 2007; Allada and Chung, 2010). This molecular clockwork is found in approximately 240 neurons in the adult Drosophila brain, which can be classified into several subsets based on their anatomical localization and size (Nitabach and Taghert, 2008; Reinhard et al., 2024). Different groups of clock neurons display distinct gene expression patterns (Ma et al., 2021) and phases of neural activity, which regulate specific aspects of daily locomotor rhythms, such as the timing of morning and evening activities (Grima et al., 2004; Stoleru et al., 2004; Liang et al., 2016). Through communication via neuropeptides and neurotransmitters, the ca. 240 clock neurons form a plastic network that adapts to different environmental conditions (Muraro et al., 2013; Chatterjee et al., 2018).

The synchrony between intrinsic circadian clock and rhythmic environmental cues provides an adaptive advantage, while their misalignment has detrimental effects on survival and health, including in humans (Roenneberg and Merrow, 2016; Walker et al., 2020; Fishbein et al., 2021). Evidence from various species also shows that mutants with altered circadian periodicity, when placed in non-matching environmental cycles, experience a reduction in lifespan. In contrast, synchrony between the circadian clock and environmental cycles extends lifespan and improves fitness (Pittendrigh and Minis, 1972; Ouyang et al., 1998; Dodd et al., 2005; Wyse et al., 2010). Both in nature and in laboratory conditions, individuals with circadian periods closely matching environmental cycles are favored under competition (Kyriacou et al., 2008; Kannan et al., 2012). For human beings, besides circadian gene mutations, modern lifestyles significantly contribute to the disharmony between our body clocks and external time. Factors such as untimely eating, shift work, travel across different time zones, social jetlag, and extended artificial lighting disrupt the proper entrainment of our circadian systems to local environmental cycles (Roenneberg et al., 2012; Chang et al., 2015; Fishbein et al., 2021). This irregularity prevents our circadian clocks from functioning optimally, leading to various health issues. It is therefore essential to understand how the phase of various circadian rhythms is matched with the day/night cycle.

We now have a good understanding of the mechanisms underlying circadian molecular clocks and their entrainment to environmental cycles. However, the mechanisms by which these clocks generate properly phased circadian outputs, such as behavioral rhythms, is not well understood. Genetic screens conducted for example in Drosophila have been remarkably efficient at identifying circadian pacemaker genes. A great number of period altering mutants have been isolated. As expected, such mutants usually do not just affect the period of circadian rhythms under constant conditions, but also their phase when mutant animals are exposed to environmental cycles. Most famously perhaps, flies with long or short period mutations in the per gene display delayed and advanced circadian phases under Light:Dark (LD) cycles, respectively (Konopka et al., 1994; Horn et al., 2019). Gene mutants that exclusively affect the phase of a circadian rhythm, without impacting the period of the underlying circadian pacemaker, are rather rare. This explains why our knowledge of the mechanisms by which the circadian pacemaker determines the phase of overt circadian rhythms is quite limited.

Circadian behavior in Drosophila is characterized by two peaks of activity at dawn and dusk. These peaks are observed under both LD and DD conditions, although the morning peak of activity tends to progressively fade under constant darkness (Dubowy and Sehgal, 2017). We and others previously reported that the loss of miR-124 in Drosophila leads to an advanced evening phase under DD conditions, but importantly does not impact the period and phase of the molecular clock in the circadian neurons that controls rhythmic behavior (Garaulet et al., 2016; Zhang et al., 2016). miR-124 thus provides an entry point to elucidate the mechanism of circadian phase determination.

miR-124 is an evolutionarily conserved neuronal microRNA (Lagos-Quintana et al., 2002; Conaco et al., 2006). Its abundant expression in the CNS suggests an indispensable role in promoting and maintaining normal CNS function. In vertebrates, miR-124 is critical for various physiological processes, including neural stem cell proliferation, neuronal differentiation, neurite growth and neuron migration (Gu et al., 2023; Zhang et al., 2023). Abnormal expression of miR-124 or its target genes has been linked to numerous neurodevelopmental and neurodegenerative disorders (Sun et al., 2015; Xu et al., 2022). Interestingly, miR-124 appears to be less critical for the central nervous system of invertebrates. In C. elegans, miR-124 knockout mutants are viable and show no gross abnormalities, nor are there evident defects in the number or differentiation of sensory neurons, where miR-124 is predominantly expressed (Clark et al., 2010). In Drosophila, previous work indicated that loss of miR-124 leads only to mild neurodevelopmental defects. Specifically, the absence of miR-124 leads to a reduction in the branch length and bouton numbers of neuromuscular junctions (NMJ) 6/7, as well as a decrease in the branch numbers of dendritic arborization (DA) neuron ddaE in the third larvae (Wang et al., 2014). Additionally, it increases the variability in dendrite numbers of sensory neurons, particularly ddaD neurons, in larvae (Sun et al., 2012). Despite these subtle developmental abnormalities, mutant flies display a range of physiological and behavioral impairments, including deficits in locomotion, flight ability, reduced lifespan, and decreased fertility (Sun et al., 2012; Weng et al., 2013; Wang et al., 2014; Kong et al., 2015).

In this study, we utilized CRISPR/Cas9 to disrupt miR-124 in neurons. This approach revealed that miR-124 functions during development to regulate circadian phase. Additionally, the loss of miR-124 resulted in important miswiring of the circadian circuits, and abnormal activity of neurons controlling morning and evening activities. These findings suggest a critical role for miR-124 in the development and proper functioning of the circadian neural network.

Material and Methods

Drosophila strains and husbandry

All flies were reared on standard low yeast brown food at 25°C under 12:12 light/dark cycles unless otherwise specified. The following fly lines were used for this study: w1118; miR-124^KO and rescued miR-124^KO (39N16) were generated by Sun et al. (Sun et al., 2012), y w; Pdf-GAL4/CyO was described by Renn et al. (Renn et al., 1999), clk856-GAL4 and MB122B-GAL4 have been described previously (Gummadova et al., 2009; Guo et al., 2017), UAS-Kir2.1 was a gift from Dr. Michael Rosbash, UAS-dTrpA1 was from Dr. Paul Garrity. Clk4.1M-GAL4 was generated by Zhang et al (Zhang et al., 2010b). UAS-uMCas9 (VDRC 340009) and VT027231-GAL4 (VDRC 205530) were obtained from the Vienna Drosophila Resource Center (VDRC). The following strains were ordered from the Bloomington Stock Center: Clk856-GAL4 (BDSC 93198), R18H11-GAL4 (BDSC 48832), UAS-CaMPARI2 (BDSC 78316), elav-GAL4;tubP-GAL80^ts (BDSC 67058), UAS-cas9 (BDSC 58985), UAS-myrGFP.QUAS-mtdTomato-3xHA;trans-Tango (BDSC 77124), w[*];trans-Tango (BDSC 77123), UAS-myrGFP P,QUAS-mtomato-3xHA (BDSC 77479), elav-GAL4;UAS-cas9.P2 (BDSC 67073), GMR56H10-GAL4 (BDSC 61644), GMR92G05-GAL4 (BDSC 48416), GMR57F07-GAL4 (BDSC 46389), Hug-GAL4.S3 (BDSC 58769), Dh44-GAL4.TH (BDSC 51987), GMR21A04-GAL4 (BDSC 49854), ple-2A-GAL4 (BDSC 86289).

Fly line generation

We constructed the UAS-miR-124-sgRNA plasmid by following the protocol published by F. Port and S. L. Bullock (Port and Bullock, 2016). Briefly, three gRNAs targeting pre-miR-124 were selected according to flyCRISPR (https://flycrispr.org/target-finder/). GAL4/UAS controlled gRNA-expressing pCFD6 vector (Addgene #73915) was digested with BbsI-HF (NEB #R3539S) and gel purified. Two inserts carrying the three gRNAs were generated by PCR reactions using pCFD6 as the template. The resulting two PCR fragments and linearized pCFD6 were then assembled in a single reaction by NEBuilder HiFi DNA Assembly (NEB #E2621S). The construct was inserted into the third chromosome using phiC31 integrase (attP2, BL8622) by the BestGene Inc (Chino Hills, CA).

The following gRNA and primers sequences were used:

gRNA1: TTTCTCCTGGTATCCACTGTAGG

gRNA2: TATTTCCACCATAAGGCACGCGG

gRNA3: TGTAGAACTGCGTTCGCTCTTGG

gRNA1F: CGGCCCGGGTTCGATTCCCGGCCGATGCATTTCTCCTGGTATCCACTGTGTTTCAGAGCTATGCTGGAAAC

gRNA1R: CGTGCCTTATGGTGGAAATA TGCACCAGCCGGGAATCGAACC

gRNA2F: TATTTCCACCATAAGGCACGGTTTCAGAGCTATGCTGGAAAC

gRNA2R:ATTTTAACTTGCTATTTCTAGCTCTAAAACAGAGCGAACGCAGTTCTACATGCACCAGCCGGGAATCGAACC

RNA extraction and qRT-PCR

miR-124 expression was measured with the miScript PCR System (Qiagen). Small RNA including miRNA from fly heads was extracted using miRNeasy Micro Kit (QIAGEN, Cat No. 217084) following the manufacturer’s instructions. The cDNA was synthesized using miScript II RT kit (QIAGEN, Cat No. 218161) according to the supplier’s protocol. To quantify mature miR-124, qRT-PCR was performed by using miScript SYBR Green PCR Kit (QIAGEN, Cat No. 218073) and Bio-Rad thermocycler CFX96. Reaction mix contained a universal reverse primer and a miR-124-specific forward primer provided by QIAGEN.

Behavior recording and data analysis

Individual adult male flies (aged 2–7 days) were used to measure locomotor activity rhythms. In most experiments, flies were entrained to a 12:12 LD cycle for 3 days followed by at least 6 days of DD at 25°C, in Percival I36-LL incubators. To test flies under long or short photoperiod, 6h:18h LD or 16h:8h LD cycles at 25°C were used. The circadian locomotor activity of flies expressing UAS-TrpA1 was monitored at 27°C to induce moderate neuronal activity in DN1p neurons. Locomotor activity was recorded in Trikinetics Drosophila Activity Monitors (Waltham, MA). Group activity for each genotype was analyzed using the FaasX software (Grima et al., 2002). Under LD conditions, data from three days were analyzed. For DD conditions data from 4–5 days of activity were used to determine the evening activity phase, unless otherwise indicated. Morning anticipatory index under 12:12 LD cycles was determined with the dedicated FaasX routine, which divides the average activity 3h prior to lights-on with the activity 6h before lights-on.

Under 6:18 LD cycles, the phase of the M-peak was quantified as the highest average activity time point before ZT0. Similarly, the E-peak phase was calculated as the highest average activity time point before CT12. All phase quantifications were manually performed by an observer who was blinded to the genotypes, using FAAS graphic outputs.

The Morning/Evening (M/E) activity ratio (figure 3F”) was calculated by dividing the average morning activity (CT0–2) with the average activity over 2.5 hours centered around the evening peak of activity.

Fig. 3. Loss of miR-124 represses neuronal activity in DN1p^Es and DN1p^Ms.

A-B, Confocal imaging of DN1p^E (A) and DN1p^M (B) calcium dynamics in WT and mutant brains dissected at different time points (circadian time, CT) during the first day of DD. C-D, Quantification of CaMPARI photoconversion efficiency across the indicated circadian time points in DN1p^Es (C) and DN1p^Ms (D) (n = 6–8 hemispheres per time point). Error bars indicate SEM in C and D. Two-way ANOVA with Tukey’s post hoc test. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. These comparisons are between genotypes for each time point. ANOVA also revealed time-dependent differences for both DN1p^Es and DN1p^Ms in wild-type flies, but not in mutant animals (# signs: comparison with peak ratio). E-G, Representative locomotor activity profiles of miR-124^KO flies with activation of DN1ps (and their control) in 12:12 LD (E), first day of DD (F) and short photoperiods (G) (E-F, n ⩾ 16 per genotype; G, n ⩾ 9 per genotype). E’-G’, Quantification of morning anticipation index in 12:12 LD (E’), evening peak phase in DD (relative to the subjective end of the day) (F’) and morning peak phase under 6:18 LD (F’). Because behavioral rhythms degrade rapidly in DD upon DN1p activation in miR-124^KO mutants, we used only the first day of DD to assess changes in evening phase. F”, Quantification of the ratio between morning and evening peak activity under DD1. G”, Since morning activity was quite flat in miR-124 mutant flies with re-activated DN1ps, we also measured the time at which 90% of morning peak activity was reached in 6:18 LD. This confirmed that activity phase was indeed advanced in these flies compared to control. Legends for E’-G’ and F”-G” are above the panels, next to panel D. Mean ± SEM is shown in E’-G’. One-way ANOVA with Tukey’s post hoc test. *, P<0.05; **, P<0.01; ***, P<0.001; ns, not significant. (N ⩾ 3 independent behavioral experiments).

Conditional miR-124 knock-out using the TARGET system

To induce miR-124 knock-outs in adult flies only with the TAGRET system, flies of experimental genotypes were maintained at 18°C (restrictive temperature, GAL80ts active, no gRNA expression) during their entire development, and moved to 29°C (permissive temperature, GAL80ts inactive, gRNA expression) three days prior to the beginning of the locomotor activity assays. Control flies were exposed to the same temperature conditions. To induce the knock-out early during development, flies were exposed to 29°C throughout their development and adulthood.

To induce miR-124 deletion at specific developmental stages, parental flies were kept at 25°C for 24 hours to allow for mating and egg laying. Eggs were then moved to 18°C to cause GAL80^ts repression and thus block expression of the miR-124 gRNAs. At specific developmental stages, animals were transferred to 29°C until adulthood, to inactive GAL80ts and thus induce gRNA expression. Developmental stages were visually assessed, and genetic controls were raised alongside experimental flies to account for any temperature effects on development.

Immunohistochemistry and Image Analysis

Third instar larvae or adult fly heads were dissected in dissection buffer (KCl 13.6g, NaCl 2.7g, CaCl₂ 0.33g, Tris 1.21g in 1L water, pH 7.2) and immediately fixed in 4% paraformaldehyde diluted in PBS for 30 minutes at room temperature. Following two quick rinses, brains were washed three times with PBST (0.1% Triton X-100) for 10 minutes each. The brains were then blocked in 10% normal donkey serum in PBST for one hour at room temperature, followed by incubation with the primary antibodies at 4°C overnight. Afterward, the brains were washed six times in PBST (20 minutes each) at room temperature and incubated with the secondary antibodies at 4°C overnight. Finally, the brains were washed six times in PBST (20 minutes each) at room temperature and mounted on glass slides using Vectashield mounting medium (Vector Laboratories). Primary antibodies were: mouse anti-PDF (Developmental Studies Hybridoma Bank, PDF; 1:400), guinea pig anti-TIM (Rakshit et al., 2012) (1:400), rabbit anti-GFP (Thermo Fisher Scientific, A11122; 1:1000), mouse anti-HA (Covance, MMS-101P; 1:250). Secondary antibodies were diluted at 1:200–400 and were as follows: donkey anti-rabbit Alexa Fluor^® 488, Goat anti-mouse Rhodamine Red, donkey anti-mouse Alexa Fluor^® 546, donkey anti-guinea pig Alexa Fluor^® 647. For TIM staining, flies were dissected at ZT 21, when TIM levels are high, after 3 days of LD entrainment to enhance signal detection.

All brains were imaged using a Zeiss LSM700 confocal microscope with ZEN software, or Zeiss LSM800 confocal microscope with ZEN blue software, keeping the laser settings constant within each experiment. Images were processed using Fiji software (http://fiji.sc). To count the number of DN1ps postsynaptically connected with the s-LNvs, an observer blind to genotypes counted the number of DN1s that showed postsynaptic trans-tango signal surrounding nuclear TIM signal.

CaMPARI imaging:

Flies expressing UAS-CaMPARI crossed with clk856-GAL4 (Gummadova et al., 2009) were used to monitor calcium dynamics in clock neurons. Male progenies were entrained to 12h:12h light-dark cycles at 25°C for 4–5 days before transitioning to constant darkness (DD) at the same temperature. Flies were collected at the indicated time points during the first day of DD.

Flies were mounted on Petri dishes and exposed to UV light (395 nm, 10mW/cm²) for 5 minutes to induce CaMPARI’s calcium-dependent photoconversion. Their brains were then immediately dissected in cold extracellular fly buffer (Chen et al., 2021) under dim red light. Brain tissues were then imaged using a Zeiss 800 confocal microscope.

Analysis involved processing images in ImageJ, with background subtraction to measure the green (F_green) and red (F_red) fluorescence intensities in target neurons. The CaMPARI photoconversion efficiency was calculated as the F_red/F_green ratio.

Statistical analysis;

Two-sample comparisons were performed using a two-tailed Student’s t-test. For multiple group comparisons, one-way or two-way ANOVA followed by Tukey’s post hoc test was used. Statistical significance was defined as P < 0.05. Error bars represent the standard error of the mean (SEM).

Results

Loss of miR-124 causes opposite shifts in morning and evening phases

Previous studies showed that loss of miR-124 results in an advanced phase of evening activity under DD, without affecting the free-running period of activity rhythms (Garaulet et al., 2016; Zhang et al., 2016). However, under a 12 h/12h LD cycle, miR-124^KO flies display a completely normal evening peak phase, indicating that light input can correct the phase defect. This likely occurs through plastic recruitment of different evening neurons under LD and DD (Chatterjee et al., 2018).

To further investigate the effects of light on circadian phase, we examined the circadian locomotor activity of miR-124^KO flies under different LD cycles. Consistent with previous studies (Garaulet et al., 2016; Zhang et al., 2016), under a regular LD cycle (LD12:12), the phase of evening activity was not affected in miR-124^KO flies, although its amplitude was reduced (Fig. 1A). Also, both morning activity and the startle response resulting from the light-on transition were absent in the mutants (Fig. 1A and 1A’). Nocturnal activity was also elevated in these mutants. Following release into DD, miR-124 mutants displayed as previously observed (Garaulet et al., 2016; Zhang et al., 2016) a unimodal activity pattern with a markedly broad and ca. 5h advanced evening peak (Fig. 1B and 1B’). These defects were corrected with a transgene containing miR-124. Similar phenotypes were observed under long photoperiods (LD16:8; Fig. S1A). Both the morning anticipation and the light-on startle response were lost, but the evening activity phase was comparable to that of wild-type flies. In contrast, miR-124^KO flies showed robust morning activity under short photoperiods (LD 6:18; Fig. 1C). However, its phase was severely delayed compared to wild-type controls and rescued miR-124 mutant flies (Fig. 1C–1C’). Thus, while loss of miR-124 advances evening activity in DD, it delays the morning peak of activity under LD cycles. We note that a truncated, delayed morning peak of activity was actually also detectable in per^S;miR-124^KO double mutant flies in our previous study (Zhang et al., 2016). We thus conclude that miR-124 mutants can display a delayed morning peak of activity, as long as its onset falls within the dark period (Fig. 1E and S1B). Under LD12:12, it falls during the light period and is thus suppressed.

Fig. 1. Loss of miR-124 shifts morning and evening peak phases in opposite directions.

A-C, Panels depict behavior of w¹¹¹⁸, miR-124^KO, and rescued miR-124^KO male flies. The rescued flies contain a 19 kb genomic transgene [39N16]. Open arrows indicate the morning peaks of activity, and regular arrows the evening peak of activity. This will be the case throughout this report.

A, Averaged locomotor activity profiles under 12:12 LD cycle. White bars indicate activity during day, black bars at night (n ⩾ 10 per genotype). A’, Quantification of morning anticipation index under 12:12 LD. B, Averaged locomotor activity profiles under DD (n ⩾ 8 per genotype). Gray shades represent the subjective night. B, Quantification of evening peak phase under DD, relative to the subjective light-off transition. C, Averaged locomotor activity profiles under short photoperiods (6:18 LD) (n ⩾ 10 per genotype). C’, Quantification of morning peak phase under short photoperiods is plotted, relative to the light-on transition. D, Quantification of free-running periods over five DD cycles. E, Schematic diagram summarizing the distinct locomotor activity patterns of miR-124^KO mutants compared to WT flies under different environmental conditions. White bars indicate light phase (day), black bars indicate dark phase (night). Quantifications are mean ± SEM from at least 3 independent behavioral experiments, and behavioral panels are from a single representative experiment. This will be the case throughout this report. One-way ANOVA with Tukey’s post hoc test. ***, P<0.001; ****, P<0.0001; ns, not significant.

In summary, miR-124^KO flies display specific behavioral phenotypes under various environmental conditions, due to changes in activity phase. In both normal LD cycles and long-day conditions, miR-124^KO flies exhibit a unimodal activity pattern due to the absence of the morning peak and the lights-on startle response, while the evening phase remains normal. Strikingly, the loss of miR-124 advances the evening activity in DD, but delays morning activity under short-day conditions (Fig. 1E). It is possible that morning activity is also delayed under DD and merges with the higher amplitude advanced evening activity, thus contributing to the broad peak observed under these conditions in miR-124^KO flies. Because neither the free-running period nor the phase of the neural molecular clocks are affected in miR-124^KO mutants (Garaulet et al., 2016; Zhang et al., 2016) (Fig. 1D), this indicates that miR-124 specifically regulates the circadian phase of rhythmic behavior.

Loss of miR-124 disrupts the activity of circadian pacemaker neurons

Although the molecular clocks of various circadian pacemaker neurons are synchronized, their Ca²⁺ oscillations and thus their neural activity are asynchronous. These Ca²⁺ rhythms usually coincide with locomotor activity bouts mediated by the corresponding neural pacemakers (Liang et al., 2016). Since molecular clocks in all pacemaker groups function normally in miR-124^KO flies (Garaulet et al., 2016; Zhang et al., 2016), we investigated whether neuronal activities in circadian clusters were altered in miR-124^KO mutants. We expressed the photoconvertible protein CaMPARI (Fosque et al., 2015) (Calcium Modulated Photoactivatable Ratiometric Integrator) to monitor the integrated calcium activity of different pacemaker groups over distinct time points. These experiments were conducted under DD, to avoid acute effects of light on neuronal activity. We first detected calcium activity in s-LNvs, as they are responsible for rhythmicity in DD, and dominate under short photoperiods (Stoleru et al., 2007), conditions where miR-124^KO flies show behavioral phase defects (Fig. 1B–1C and 1B’–1C’). The s-LNvs also control the morning peak of activity, which is phase-shifted in miR-124 mutant flies (Fig. 1C and 1C’). In wild-type flies, Ca²⁺ levels in the s-LNvs peaked near the end of the night (Fig. 2A–2B), as previously reported, consistent with their role as morning anticipation-driving neurons (Grima et al., 2004; Stoleru et al., 2004; Liang et al., 2016). Intriguingly, in miR-124^KO mutants, the Ca²⁺ peak of s-LNvs occurred during the subjective day under DD (Fig. 2A and 2B), coinciding with the broad peak of locomotor activity observed under these conditions (Fig. 1B and 1B’). To determine whether the shifted neuronal activity in s-LNvs is responsible for the evening phase advance of behavioral activity in miR-124^KO flies under DD, we inhibited the s-LNvs by expressing the inwardly rectifying K⁺ channel Kir2.1 (Ho et al., 1993) in miR-124^KO mutants. Indeed, repression of s-LNvs in the miR-124^KO was able to correct the evening phase (Fig. 2C–2D). Note that behavior was observed during the first day of DD only, and we could not assess morning activity phase under LD since inhibition of s-LNvs causes behavioral arrhythmicity under prolonged DD exposure and blunts morning activity (Nitabach et al., 2002; Depetris-Chauvin et al., 2011).

Fig. 2. Loss of miR-124 shifts the neuronal activity of s-LNvs.

A, Confocal imaging of s-LNv calcium dynamics in WT and mutant brains dissected at different time points (circadian time, CT) during the first day of DD. B, Quantification of CaMPARI photoconversion efficiency across the indicated circadian time points (n = 6–8 hemispheres per time point). Error bars indicate SEM. Two-way ANOVA with Tukey’s post hoc test. **p<0.01, #p<0.05. Comparisons are between genotypes for each time point. ANOVA also revealed statistically significant time-dependent differences in miR-124 mutants (# signs, comparison with trough ratio). Wild-type animals showed as expected (Liang et al., 2016; Liang et al., 2019) the highest CAMPARU ratio (i.e. highest neural activity) at ZT0, although statistical significance was not reached as variability was unusually high for this time point (P=0.09). C, Averaged locomotor behavior under three days of LD cycles (upper panel) and during the first day of DD (lower panel) for flies expressing Kir2.1 in M-cells, and their controls (n ⩾ 10 per genotype). D, Quantification of evening peak phase on the first day of DD (DD1; N ⩾ 3 independent experiments). Only the first day of DD was used, because behavioral rhythms of flies expressing Kir2.1 in M-cells degrade rapidly in DD. Error bars indicate SEM. One-way ANOVA with Tukey’s post hoc test. ***, P<0.001; ns, not significant.

A subset of the posterior dorsal neurons 1 (DN1ps), which includes both morning- and evening-activity promoting cells, is directly targeted by the s-LNvs (Zhang et al., 2010a; Zhang et al., 2010b; Chatterjee et al., 2018). We thus also investigated the role of miR-124 in regulating neural activity in these neurons. In wild-type flies, DN1p-evening neurons (DN1p^E) reach Ca²⁺ peak levels in the middle of the day (Fig. 3A and 3C), while DN1p-morning neurons (DN1p^M) peak during the late night (Fig. 3B and 3D). This is consistent with their respective roles in morning and evening anticipatory activities. Surprisingly, in the miR-124^KO mutants, neural activity was severely depressed at all time points in both DN1p subtypes (Fig. 3A–3D). When we simultaneously activated both DN1p^Ms and DN1p^Es using the Clk4.1M-GAL4 driver (Zhang et al., 2010b), the behavioral phenotypes of miR-124 mutants were partially corrected (Fig. 3E–3G and 3E’–3G’). In LD and DD conditions, miR-124^KO flies now displayed bimodal activity patterns due to the re-emergence of morning anticipatory activity combined with reduction in nighttime activity (Fig. 3E–3F and 3E’–3G’). Evening and morning activity were of approximately equal amplitude on DD1 (Fig. 3F), and morning activity and evening phase were significantly improved (Fig. 3E’–F’ and 3F”). Under short photoperiod conditions, miR-124^KO mutants with activated DN1p^Ms and DN1p^Es showed advanced morning activity phase compared to their controls (Fig. 3G, 3G’ and 3G”). However, the amplitude was not fully restored, presumably because of the constant activation of DN1ps. Notably also, miR-124^KO flies, which are as mentioned above more nocturnal than control animals, switched to a wild-type-like diurnal activity pattern upon activation of DN1p neurons. As expected, the lights-on startle responses of miR-124^KO mutants could not be restored by DN1p activation or s-LNv inhibition (Fig. 2C, 3E and 3G). Indeed, these responses are not controlled by the circadian clock (Allada et al., 1998).

Taken together, these results demonstrate that the loss of miR-124 either shifts or suppresses the neuronal activities of circadian clusters. Silencing s-LNvs or activating DN1ps in miR-124^KO mutants can correct phase defects in morning and evening activities, indicating that miR-124 regulates circadian phases through these neurons and their activity.

Pan-neuronal CRISPR/Cas9-mediated disruption of miR-124 mimics miR-124^KO

To study further the mechanisms by which miR-124 regulates circadian behavior phase, we turned to conditional CRISPR/Cas9-mediated gene knock-out. Indeed, this approach works efficiently in the Drosophila circadian neural network (Delventhal et al., 2019; Schlichting et al., 2019a). We followed the protocol provided by Port and Bullock (Port and Bullock, 2016), and designed three sgRNAs targeting three different regions of the miR-124 precursor (pre-miR-124) to ensure efficient gene disruption (Fig. 4A). One of the three gRNAs targeted the coding sequence for the mature miR-124. Given that miR-124 is a neuron-specific miRNA (Weng and Cohen, 2012), we expressed miR-124-sgRNAs and Cas9 pan-neuronally with the elav-GAL4 driver. This led to a substantial reduction of miR-124 level, which was barely detectable in head extracts (Fig. 4B). As expected, miR-124^CRISPR flies showed very similar behavioral phenotypes to those of miR-124^KO mutants under LD cycles with regular (Fig. 4C and 4C’) or short photoperiod (Fig. 4E and 4E’), and under DD (Fig. 4D and 4D’). Given that pan-neuronal CRISPR-mediated miR-124 deletion fully recapitulates the phenotypes observed in the constitutive KO mutants, we infer that the molecular clock is also unaffected in circadian neurons of the conditional KO flies, as observed in the constitutive mutants (Garaulet et al., 2016; Zhang et al., 2016).

Fig. 4. Pan–neuronal CRISPR-mediated disruption of miR-124 phenocopies miR-124KO.

A, Diagram showing three sgRNA target sites for pre-miR-124. B, Quantitative real-time PCR analysis of miR-124 abundance in deletion mutants and controls. elav indicates the elav-GAL4 driver, miR-124^CRISPR refers to the genotype UAS-Cas9/+;UAS-tgRNA124/+. C-E, Averaged locomotor behavior of flies with elav-Gal4-driven disruption of miR-124 in 12:12 LD (C), DD (D) and 6:18 LD (E) (C-E, n ⩾ 10 per genotype). See also Figure S2 for additional data. C’-E’, Quantifications of morning anticipation index in 12:12 LD (C’), evening peak phase under DD (D’) and morning peak phase in short photoperiods (E’) (N = 3 independent behavioral experiments). The legends for C’-E’ are above the panels.

Error bars represent SEM. One-way ANOVA followed by a Tukey post-hoc test was performed. **, P<0.01; ***, P<0.001; ****, P<0.0001; ns, not significant.

Taken together, these data demonstrate that miR-124 acts in neurons to regulate circadian phase, consistent with its role as a conserved neuron-specific microRNA. Additionally, CRISPR/Cas9-mediated miR-124 deletion in neurons is highly efficient and can thus be used to study miR-124 temporal and spatial requirements.

Developmental miR-124 expression is essential for properly phased adult circadian behavior

Next, we attempted to narrow down the specific neurons required for miR-124-mediated circadian phase regulation by conducting a neuronal GAL4 screen. We crossed flies carrying both UAS-Cas9 and UAS-miR-124-sgRNA transgenes with various GAL4 lines, targeting clock neurons (Tataroglu and Emery, 2014) and non-clock neurons related to circadian output pathways (Cavanaugh et al., 2014; Cavey et al., 2016; King et al., 2017; Liang et al., 2019), to disrupt miR-124 in these neurons. However, none of these GAL4 lines recapitulated the behavioral defects observed in pan-neuronal miR-124^CRISPR flies (Fig. S2). The elav-GAL4 line we used is an enhancer trap line that fully mirrors the expression pattern of endogenous elav (Berger et al., 2007), with expression first detected in the embryonic nervous system at stage 12. The absence of phenotype with the other GAL4 lines we tested might thus be because these lines are expressed too late during the development of the relevant neurons. To test whether miR-124 is indeed required early during development, we employed the temporal and regional gene expression targeting (TARGET) system (McGuire et al., 2004) to induce miR-124 deletion at different developmental stages. Pan-neuronal disruption during development advanced evening phase in DD (Fig. 5B and 5B’), delayed morning phase under short photoperiods (Fig. 5C and 5C’), and disrupted morning anticipation and light-on startle response under 12:12h LD cycles (Fig. 5A and 5A’), recapitulating the behaviors seen with constitutive miR-124 deletion (Fig. 1A–1C). In contrast, disruption confined to the adult stage had no effect on circadian phase, with behaviors comparable to genetic controls (Fig. S3). miR-124 is therefore required during development to control the phase of adult circadian behavior.

Fig. 5. miR-124 acts during development to regulate circadian phase.

A-C, Averaged locomotor behavior of flies with embryonic miR-124 deletion in 12:12 LD (A), DD (B) and 6:18 LD (C) (A-B, n ⩾ 8 per genotype; C, n ⩾ 9 per genotype). A’-C’, The quantification of morning anticipation index in 12:12 LD (A’), evening peak phase (hours) under constant darkness (B’) and morning peak phase (hours) in 6:18 LD (C’). D, Quantification of morning anticipation index (left), evening phase (middle) and morning phase (right) under 12:12 LD, DD and 6:18 LD for control flies and flies with 1^st instar larval and 2^nd instar larval deletion of miR-124. E, Quantification of morning anticipation index (left), evening phase (middle) and morning phase (right) under 12:12 LD, DD and 6:18 LD for control flies and flies with 3^rd instar larval and pupae stage deletion off miR-124. The legends for A’-C’ and D-E are on the right side of panel D. See also Figure S3–S5 for additional data. Quantifications are means ± SEM in A’-C’ and D-E (N = 3 independent behavioral experiments). One-way ANOVA with Tukey’s post hoc test. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001; ns, not significant.

To determine more precisely when miR-124 is required, we induced miR-124 disruption at different developmental stages. Pan-neuronal miR-124 deletion induced during the first or second instar larvae resulted in the same circadian phase defects as those observed with constitutive loss of miR-124 (Fig. 5D and Fig. S4A–4B). In contrast, miR-124 disruption induced during the third instar larvae or at the pupal stages had no effects on circadian behavior (Fig. 5E and Fig. S5A–5B). Thus miR-124 is required around the second instar larval stage to determine adult behavior phase.

Loss of miR-124 leads to defective circadian network wiring

Given that miR-124 functions during larval development and alters the activity of circadian neurons in miR-124 mutants, we hypothesized that the loss of miR-124 affects the wiring of the circadian neural network. To test this, we employed the unbiased anterograde trans-synaptic tracing system, trans-Tango, to map synaptically connected neurons within circadian circuits (Talay et al., 2017).

The trans-Tango signal is known to be both temperature- and age-dependent. It increases with age and decreases with higher temperature (Talay et al., 2017). Therefore, to assess robust synaptic connectivity within the circadian neural network of miR-124^KO mutants, we raised and entrained 5–7 days old flies at the stringent 25°C temperature. We first traced and defined circuits downstream of the s-LNvs by co-expressing the trans-Tango ligand and myrGFP using pdf-GAL4. In both WT and miR-124^KO brains, we observed the typical innervation of the posterior dorsal protocerebrum by GFP-expressing s-LNvs (Fig. 6A and Fig. S6), along with optic lobe and contralateral projections from l-LNvs. However, in miR-124^KO brains, the dorsal projections of the s-LNvs appeared to be less fasciculated (Fig. 6A and S6B), with projection branching out of the main bundle more frequently ventrally. Similar observations had been made previously (Zhang et al., 2016), although projection defects seemed more pronounced in the present study, perhaps because of improved imaging. These projection abnormalities were more clearly observed in larvae (Fig. S7). In WT larval brains, all s-LNv axons projected medially and terminated in the anterior region of the central brain (Fig. S7A). However, in miR-124^KO brains, some s-LNv axons displayed aberrant wandering near the somas in the posterior brain region and even formed loop-like structures (Fig. S7B). Interestingly, these axonal projection defects were usually more pronounced in one hemisphere compared to the other (Fig. S7B–7C). Overall, these observations indicate that miR-124 plays an important role in s-LNv projection development, possibly impacting axon guidance, pruning or retraction.

Fig. 6. Loss of miR-124 causes excessive synaptic connections in the circadian network.

A, Postsynaptic partners of s-LNvs in WT and miR-124^KO mutants revealed with trans-Tango. From left to right: Presynaptic s-LNvs (anti-GFP, green), postsynaptic trans-Tango signal (anti-HA, red), and merged images. Genotypes are indicated on the left side of the panels. The scale bars represent 40 μm. See also Figure S6–S7 for additional data. B, Direct synaptic contacts between s-LNvs and DN1ps in A revealed by co-staining with anti-TIM (blue). Brains were dissected at ZT21, when TIM expression peaks, to facilitate optimal identification of clock neurons. White open arrowheads indicate the DN1ps directly targeted by s-LNvs. See also Figure S8. C, Quantification of the numbers of DN1ps contacted by s-LNvs per hemisphere. Error bars indicate SEM. An unpaired two-tailed Student’s t-test was used. ****, P<0.0001. D-E, Postsynaptic partners of DN1p^Es (D, scale bar, 50um) and DN1p^Ms (E) in WT and miR-124^KO mutants revealed with trans-Tango. White arrows represent DN1p^M, white diamond arrows indicate non-DN1p^Ms in E. See also Figure S9–S11 for additional data.

In addition to projection defects, adult miR-124^KO brains displayed significantly increased Tomato signal in neurons postsynaptic to the s-LNvs relative to WT controls (Fig. 6A and Fig. S6), suggesting that s-LNvs target a greater number of neurons in miR-124^KO flies. This increased signal was observed in both the dorsal part of the brain, and near the midline. Since the DN1ps are downstream of the s-LNvs and project fibers into the dorsal protocerebrum, we investigated whether the loss of miR-124 affected their connections. By confirming their identity as clock neuron through TIMELESS (TIM) protein staining (Fig. 6B and S8), we observed that in WT, 1.9±0.1 (mean±SEM) DN1ps per hemisphere were postsynaptic to the s-LNvs. However, in miR-124^KO brains, 4.6± 0.2 DN1ps per hemisphere were targeted by the s-LNvs (Fig. 6C). This indicates that the loss of miR-124 leads to increased synaptic connectivity between the s-LNvs and DN1ps.

Next, we sought to trace postsynaptic partners of both DN1p subtypes in wild-type and mutant flies. First, we co-expressed the trans-Tango ligand and myrGFP with the R18H11-GAL4, which specifically labels DN1p^Es. In WT brains, most projections from R18H11-positive DN1ps are located in the dorsal brain, innervating the anterior optic tubercle (AOTU) and the pars intercerebralis (PI) (Guo et al., 2018; Lamaze et al., 2018) (Fig. 6D and Fig. S9A). However, in miR-124^KO brains, in addition to these dorsal projections, numerous projections terminated more ventrally, near the antennal lobe (AL), and there were even sparse projections in the suboesophageal ganglion (SOG) (Fig. 6D and Fig. S9B). This extensive branching of DN1p^Es in miR-124^KO brains correlated with much broader and denser postsynaptic mtdTomato labeling. In WT brains, mtdTomato-positive projections were mainly restricted to the lateral and dorsal protocerebrum (Fig. 6D and Fig. S9A), whereas in miR-124^KO brains, robust mtdTomato-positive projections were found throughout a large area of the central brain (Fig. 6D and Fig. S9B). As a result, it was challenging to accurately identify specific synaptic connections. However, using TIM staining, we observed that some DN1p^Es form synapses with each other (Fig. S10A). The loss of miR-124 did not appear to affect this connectivity, as 4 out of 7–8 R18H11-GAL4-labelled DN1p neurons (Kunst et al., 2014) made contact with each other in both WT and miR-124^KO flies (Fig. S10B).

We then turned our attention to the DN1p^Ms. VT027231-Gal4 labels the DN1ps which are sufficient to drive morning anticipation (Chatterjee et al., 2018), so we used this driver to express the trans-Tango ligand and myrGFP in the DN1p^Ms. Note that this driver is also expressed in a few non-DN1ps neurons. In WT brains, GFP-expressing fibers from the DN1ps are primarily concentrated in the lateral and dorsal brain (Fig. 6E and Fig. S11A). However, in miR-124^KO brains, these fibers spread extensively throughout the entire central brain (Fig. 6E and Fig. S11B). This widespread and dense arborization was associated with highly dense trans-Tango-dependent signal. (Fig. 6E and Fig. S11). This robust mtdTomato labeling surrounding the DN1pMs strongly indicates that these circadian neurons form additional synaptic connections in the absence of miR-124, although we cannot entirely exclude that this dorsal signal is at least in part due to non-DN1p neurons labeled by VT027231-Gal4. However, this possibility seems unlikely, because these extra neurons are located much more ventrally than the DN1pMs. They are more likely to contribute to the trans-Tango signal observed in the central ventral brain. Taken together, these observations indicate that both DN1p^Es and DN1p^Ms are similarly affected by the loss of miR-124, exhibiting broader and denser projections in miR-124^KO brains, as well as increased connectivity.

Collectively, these results suggest that the absence of miR-124 leads to developmental miswiring within circadian circuits, likely disrupting neuronal activities in the circadian network and causing shifts in circadian phase.

Discussion

We demonstrate here that miR-124 plays a crucial role in the proper development and neuronal wiring of the Drosophila circadian neural network. Its absence leads to excessive synaptic connections within the circadian neural circuit, and between circadian and non-clock neurons. Since previously reported neural defects, observed in the larvae, were quite subtle, we were surprised by the extent of the wiring defects in the adult circadian neural network. Synaptic defects related to miR-124 loss have been observed in other species. In Aplysia californica, miR-124 modulates synaptic plasticity through CREB (Rajasethupathy et al., 2009), while in rat hippocampal neurons, miR-124 controls input-specific synaptic plasticity via synaptopodin (SP) (Dubes et al., 2022). In mice, miR-124 regulates dendritic morphogenesis and increases spine density in the olfactory bulb, highlighting its role in synaptic formation and plasticity (Li and Ling, 2017). Collectively, these data suggest that regulating synapse function is an ancient and conserved function of miR-124 across the animal kingdom.

The precise formation of synaptic connections between neurons is critical for proper information perception, processing, and transmission, which underpins all central nervous system functions (Luo, 2021). Despite their distinct roles, the various clock neuron groups in the Drosophila brain are highly interconnected, allowing the flies to adapt to ever-changing environmental conditions (Muraro et al., 2013; Chatterjee et al., 2018; Ahmad et al., 2021). The s-LNvs, the master pacemakers in Drosophila, maintain free-running rhythms under DD and are essential for anticipating dawn under LD conditions (Renn et al., 1999; Grima et al., 2004; Stoleru et al., 2004). They also modulate the timing of the evening peak of locomotor activity (Schlichting et al., 2019b). Meanwhile, the DN1ps consist of subgroups that promote either morning or evening activity (Chatterjee et al., 2018). We focused on the neuronal connectivity of these two clock neuron subgroups, as the loss of miR-124 affected both morning and evening phases in miR-124^KO flies (Fig. 1). Our trans-Tango analysis revealed that in WT flies, only two DN1ps are postsynaptic to the s-LNvs (Fig. 6B–6C), which aligns with findings from the electron microscopy (EM) hemibrain connectome (Scheffer et al., 2020), where s-LNvs are seen to form very few synaptic connections with other clock neuron groups (Shafer et al., 2022). This is consistent with previous studies showing that s-LNvs primarily communicate with DN1ps and other circadian groups through volumetric neuropeptide PDF secretion to set their pace (Liang et al., 2017; Fernandez et al., 2020; Ahmad et al., 2021). In WT flies, the s-LNvs differentially activate the DN1p^Ms and inhibit the DN1p^Es through PDF signaling (Chatterjee et al., 2018). Additionally, sNPF from both s-LNvs and LNds also contributes to shaping Ca²⁺ activity pattern in DN1ps (Liang et al., 2017). Interestingly, the s-LNvs apparently use glycine, which usually inhibits neuronal activity, as its sole fast synaptic neurotransmitter (Frenkel et al., 2017). In miR-124^KO brains, we observed that the s-LNvs connect to 2–3 additional DN1ps. These additional glycinergic synaptic connections may thus constantly inhibit both DN1p^Ms and DN1p^Es, since these neurons are connected to each other through electric synapses (Tabuchi et al., 2018). Increased sNPF signaling, and for the DN1p^Es, increased PDF signaling, could also contribute to DN1 inhibition (Shang et al., 2013; Vecsey et al., 2014; Chatterjee et al., 2018).

Interestingly, the phase of s-LNv activity was altered in miR-124 knockout animals, with activity shifted toward the middle of the day, rather than at the end of the night. This could be the result of altered neural input onto the s-LNvs, or miR-124 could modulate the expression of circadian output genes that determine neural activity rhythms. It is noteworthy that although the loss of miR-124 results in an increase in synaptic connections made by the s-LNvs and the DN1ps and drastically alters their activity pattern, only the phase of circadian rhythms is affected, not its period, nor the oscillations of neuronal molecular pacemakers (Garaulet et al., 2016; Zhang et al., 2016). For the s-LNvs, this fits with the observation that their neural activity can be eliminated without major consequences on the oscillation and phase of their circadian pacemaker (Depetris-Chauvin et al., 2011). However, in the case of miR-124 mutants, the s-LNvs are rhythmically active, with an abnormal phase. Thus, the circadian molecular pacemaker of the s-LNvs is quite resistant to (or protected from) changes in neural activity, both in terms of period and phase. Since the s-LNvs function as the main circadian pacemaker neurons, this insulation might be needed to avoid overreaction to environmental inputs such as temperature changes (Busza et al., 2007).

The DN1ps are controlled by the s-LNvs through PDF signaling, which maintain their rhythmicity and clock speed (Zhang et al., 2010a; Yao et al., 2016; Chatterjee et al., 2018). Altering the molecular clock in either s-LNvs or DN1ps readily shifts behavioral phase under DD, whereas similar molecular manipulations in LNds do not (Chatterjee et al., 2018), emphasizing the pivotal role of the s-LNv-DN1p axis in circadian phase regulation. The simultaneous re-activation of both morning and evening DN1ps significantly rescued circadian behavior in miR-124^KO mutants (Fig. 3E–3G), indicating that neuronal activity and likely communication within DN1ps are essential for maintaining normal behavioral phases. Indeed, we (Fig. S10) and others observed interconnections between DN1ps (Tabuchi et al., 2018). Interestingly, in miR-124 mutants, DN1p activity is very low in both the E and M subtypes, and behavior phase correlated quite closely with s-LNv activity under DD: both showed a broad peak in the middle of the subjective day. This suggests that the s-LNvs can bypass the DN1ps to control behavior. The DN1ps respond to both light and temperature inputs (Zhang et al., 2010b), and might thus be at time acutely inhibited. This could allow the s-LNvs to control behavior directly under specific environmental conditions.

Synapse formation requires a complex set of genes, including genes encoding cell-surface proteins, which differ among neuron types to specify their unique connectivity (Xie et al., 2022). miR-124 is predicted to directly regulates hundreds of genes, and the loss of miR-124 affects the expression of thousands, particularly those involved in central nervous system (CNS) functions (Sun et al., 2012). It is therefore likely that the wiring defects we observed is the result of the combined misregulation of multiple genes. Supporting this idea, reduction in BMP signaling - which impact multiple aspects of Drosophila development including synapse formation and function (Berke et al., 2013; Sulkowski et al., 2014; Heo et al., 2017; Furusawa et al., 2023; Kaneko et al., 2024) - only partially rescued the evening miR-124^KO phase phenotype, and this only under DD (Garaulet et al., 2016). Clearly, miR-124 is required quite early during development. Conditional knock-out indicates that miR-124 is required during mid or early larval development, and larval s-LNvs show frequent axonal projection defects when miR-124 is missing. It is however curious that in adults, these striking defects have been largely corrected, with at best a minor increase in aberrant axonal projections (Fig. 6A and S6) (Zhang et al., 2016). It is therefore likely that the aberrant projections observed in larvae are pruned during pupation, or perhaps even earlier since we usually observed axonal projection defects in only one of two brain hemispheres. The main defect of adult s-LNvs thus appear to be the excess connectivity with DN1ps and other unidentified neurons. The origin of this hyper-connectivity, also present in DN1ps, is presently unclear. It could be the result of more subtle axon projection defects than those observed at the larval stages, or insufficient pruning. Neuronal projection and their development is controlled by a complex interplay of biological processes (de la Torre-Ubieta and Bonni, 2011), including neuronal polarization, axon outgrowth and guidance, cell adhesion, transcriptional control, and activity-dependent plasticity. These processes are tightly regulated and coordinated during development to ensure the establishment of accurate and functional neural circuits. The loss of miR-124 may disrupt one or more of these mechanisms, or coordination among them, ultimately leading to the projection defects we observed.

Our results suggest that miR-124 functions non-cell-autonomously. Indeed, pan-neuronal conditional knock-out during the 2^nd instar larvae recapitulates all the phenotypes observed in miR-124^KO mutants, but circadian GAL4 drivers (clk856-GAL4, pdf-GAL4, tim-GAL4) which are already active in the 1^st instar larvae, do not. Moreover, unlike the s-LNvs and four dorsal neurons which differentiate during the first instar larval stage, the DN1ps emerge and develop much later, during the third instar larval and pupal stages (Liu et al., 2015; Poe et al., 2022). The development, morphology and connectivity defects observed in DN1ps of miR-124^KO mutants thus likely arise through non-autonomous mechanisms, perhaps induced by defects in early-developing neurons such as the s-LNvs. However, we cannot exclude that miR-124 regulates DN1p neuronal projections in a cell-autonomous manner at later developmental stages without leading to detectable behavioral alterations.

Our findings demonstrate that miR-124 regulates circadian phase of rhythmic behavior by coordinating the development of circadian neural circuits. They reveal the importance of proper circuit wiring in the control of behavior phase. Defective neural connectivity may also help explain other physiological and behavioral abnormalities caused by the loss of miR-124, not only in Drosophila but potentially also in other animals, given that miR-124’s mature sequence and brain-enriched expression are conserved from worms to humans.

Data and material availability

Any additional information and materials reported in this paper are available from the corresponding authors upon request.