Circadian clock disruption promotes the degeneration of dopaminergic neurons in male Drosophila

To test whether the vulnerability of PAM neurons to oxidative stress is rhythmic, we next performed the short H₂O₂ treatment in a circadian fashion. Seven-day-old w¹¹¹⁸ flies entrained to LD (12 h:12 h light-dark cycles) were exposed to 10% H₂O₂ for 4 h, following a 5-h food and water deprivation. H₂O₂ exposure was initiated at a different Zeitgeber Time (ZT) across the day, and DA neurons were analyzed 7 days post-treatment (Fig. 1c). Whereas the H₂O₂ treatment at any time caused significant PAM neuron loss compared to the water-only control, the treatment at ZT20 (8 h after lights-off) resulted in a more significant loss of PAM neurons than the other time points (Fig. 1d). The highest sensitivity at ZT20 was replicated in the assay with the 4-h 5% H₂O₂ treatment, validating its specificity (Supplementary Fig. S1c). We performed the same experiment in constant darkness (DD) to determine the role of endogenous clocks and LD cycles. Newly hatched flies were entrained for 3 days in LD and then placed in DD for the remainder of the experiment and treated with H₂O₂ on the fourth day in DD (DD4) (Fig. 1c). The treatment at CT4 (4 h after subjective lights-on) with 10% (Supplementary Fig. 1c) or 5% H₂O₂ (Supplementary Fig. S1d) caused a more significant PAM neuron loss than the other time points. These results suggest intrinsic rhythmicity in the vulnerability of PAM neurons to oxidative stress modulated by light.

Because a previous study reported that H₂O₂ exposure elicits a higher rate of mortality in per⁰ than in wild-type flies³⁴, we also examined the effect of various concentrations of H₂O₂ on PAM neuron survival in per⁰ flies at a single timepoint; 2.5% H₂O₂, which did not affect PAM neurons in w¹¹¹⁸ and Canton-S (CS), robustly induced PAM neurodegeneration in per⁰ mutants (Fig. 2d). This finding suggests that loss of per null mutation enhances the vulnerability of PAM neurons to oxidative stress-induced degeneration. We next performed the circadian oxidative treatment with 5% H₂O₂ in LD and DD on per⁰ flies. The per⁰ flies treated with H₂O₂ at any timepoint showed a significant loss of PAM neurons as compared to the control group treated with water only. In LD, the treatment at ZT12 resulted in a more significant loss of PAM neurons than the other time points (Fig. 2e). In contrast, the difference among time points was abolished in DD (Fig. 2f). These findings suggest that the circadian clock gene per controls the magnitude and temporal variations of PAM neuron sensitivity to oxidative stress. In the absence of functional clocks, light can impose diurnal rhythms of PAM neuron vulnerability.

Because PAM neurons are not clock-containing cells, the rhythmic vulnerability of PAM neurons is non-cell-autonomously controlled by clocks located elsewhere. Several mechanisms might drive rhythms in non-clock-containing neurons, including the rhythmic H₂O₂ uptake by feeding/drinking rhythms, systemic rhythmic changes in the brain environment, and the rhythmic modulation of the physiology of PAM neurons by circadian pacemaker neurons.

The circadian H₂O₂ feeding assay requires flies to be starved before and during H₂O₂ exposure. We next tested whether the starvation and dehydration associated with the assay led to the rhythmic PAM neurodegeneration. To this end, flies were removed from food and water for 9 h, 5 h before and 4 h after ZT20. Seven days later, PAM neurons were examined by anti-TH immune staining. The control group was given access to water for 4 h after a 5-h starvation/water deprivation period. The results showed no changes in PAM neuron count between the groups (Fig. 3b). This finding suggests that the loss of PAM neurons and their rhythmicity are not caused by starvation or dehydration before or during the H₂O₂ treatment.

The results obtained so far suggest that the loss of PAM neurons can be directly attributed to the increased levels of reactive oxygen species (ROS) in PAM neurons resulting from H₂O₂ treatment rather than the systemic effects triggered by H₂O₂ ingestion. To test this further, we expressed catalase, an enzyme that breaks down H2O2 into water and oxygen, specifically in the PAM neurons using the R58E02 driver. The expression of catalase effectively prevented the H₂O₂-induced loss of PAM neurons in both LD (treated at ZT20) and DD (at CT4 and CT16) (Fig. 3c, d). These results indicate that the elevation of ROS within the PAM neurons directly leads to the neurodegeneration of PAM neurons following exposure to H₂O₂.

Endogenous oxidative stress levels vary daily in flies and mammals^31,37. We wondered whether the rhythms in endogenous ROS levels in the fly brain and H₂O₂ administered at specific times would accumulate to trigger a rhythmic neuronal loss in the PAM cluster. To test this possibility, we used MitoSOX (a ROS-sensitive fluorescent dye targeted to mitochondria) to monitor mitochondrial ROS levels across the day in PAM neurons labeled with R58E02>UAS-EGFP in w¹¹¹⁸ flies. In 7-day-old flies, MitoSOX fluorescence intensity in PAM neurons in LD showed daily variations and peaked at ZT20 (Fig. 3e), the time when PAM neurons are the most vulnerable to H₂O₂ (Fig. 1d). This finding suggests that oscillating levels of endogenous ROS combined with the H₂O₂ treatment might account for the differential loss of PAM neurons in LD. In contrast, no significant differences in MitoSOX levels were observed between time points in DD (Fig. 3e, Supplementary Fig. S3). Therefore, another mechanism under the control of circadian clocks gates the vulnerability of PAM neurons in DD.

To test if any subclass of clock neurons is essential for the role of PER in regulating PAM neuron survival, we next performed per⁰ genetic rescue using Pdf-GAL4, expressed in the s- and l-LNvs except the fifth s-LNv^39,40, and DvPdf-GAL4, which targets all LNvs, three CRY-negative LNds, and one CRY/ITP double-positive LNd^41,42. Both rescue genotypes restored the number of PAM neurons to the level of w¹¹¹⁸ under basal conditions (Fig. 4a). The number of surviving PAM neurons after H₂O₂ treatment was also significantly greater in both rescue genotypes than in control expressing only the driver (Fig. 4b). PER rescue with Pdf-GAL4 or Clk1982-GAL4 showed similar levels of protection against H₂O₂, whereas, with DvPdf-GAL4, the protection was partial. This might be due to differential expression levels within the LNvs between Pdf-GAL4 and DvPdf-GAL4.

Additionally, we examined whether the genetic rescue of per⁰ in clock neurons could prevent PAM neuron loss in DD. Clk1982-GAL4 is highly expressed in most clock neurons, except in the LPN, but has an ectopic expression in non-clock cells, such as the Kenyon cells. Therefore, we included a second pan-clock neuron driver, Clk856-GAL4, which covers LPN and has a reduced ectopic expression³⁸ in this experiment. H₂O₂ treatment was performed at CT4 and CT16. The number of PAM neurons was significantly higher in all the rescue genotypes compared to per⁰ at both time points (****p < 0.0001 for each pairwise comparison between per⁰ and the rescue genotypes). Flies rescued with either Clk1982-GAL4 or DvPdf-GAL4 exhibited partial protection against H₂O₂-induced PAM neuron loss, whereas flies rescued with Clk856-GAL4 or Pdf-GAL4 were completely resistant to H₂O₂ at both time points (Fig. 4c). These findings suggest that Clk856-GAL4 and Pdf-GAL4 may drive the expression of PER above endogenous levels, thereby bolstering the resistance of PAM neurons beyond that in wild-type flies. Taken together, these findings suggest that PER expression within the PDF-positive LNvs plays a significant role in the regulation of survival of PAM neurons following oxidative insults.

The finding that PER expressed in circadian pacemaker neurons controls rhythmic PAM neuron vulnerability suggests that PAM neurons are directly or indirectly downstream of circadian neural circuits. To test this possibility, we examined postsynaptic targets of DvPdf-GAL4-expressing neurons using trans-Tango⁴³, an anterograde transsynaptic tracing tool. Several neurons in the most anterior region of the PAM cluster were marked with the postsynaptic signal, indicating the direct projection from the DvPdf-GAL4-labeled neurons to 6 to 7 PAM neurons (Fig. 4d). In contrast, the trans-Tango assay with Pdf-GAL4 did not result in any positive signal within the PAM cluster (Fig. 4d), suggesting that the s- and l-LNv neurons are not presynaptic to PAM neurons. These findings taken together suggest that all or some DvPdf-positive but Pdf-negative cells (i.e., the three CRY-negative LNds, one CRY/ITP double-positive LNd, and the fifth LNv) are presynaptic to PAM neurons (Fig. 4e).

To identify the vulnerable PAM subgroup within the R58E02-positive cells, we tested split-GAL4s. From the available split-GAL4 drivers created at the Janelia Research Campus⁴⁵, we selected all lines with activation domains driven with the R58E02 promoter, i.e., MB312B, MB043B, MB299B, and MB315C. These drivers label the neurons innervating different but partly overlapping areas of the MBs: MB312B-neurons send projections to the γ-lobe (γ4 subdomain), MB043B-neurons to the α- β-, and β’-lobes (α1, β1, β’1ap, and β’1m subdomains), MB299B neurons to the α-lobe (α1 subdomain) and MB315C-neurons to the γ-lobe (γ5 subdomain)⁴⁴. UAS-GFP was expressed with each of these drivers, and the number of remaining GFP-positive cells after short-term H₂O₂ treatment was scored. Neuronal loss was observed within the clusters labeled by the MB315C and MB299B drivers (Fig. 5b, c). We tested one more split-GAL4 driver (MB441B) that does not contain the R58E02-driven component but is expressed in the PAM neuron subset projecting to the MB γ-lobe (γ3 subdomain). H₂O₂ treatment caused a loss of neurons within the subpopulation labeled by MB441B (Fig. 5b, c). These results suggest that PAM neurons innervating the MB γ3, γ5, and α1 subdomains (PAM γ3, γ5, and α1 subgroups, respectively) are subpopulations vulnerable to neurodegeneration after oxidative insults (Fig. 5d). Of these, the PAM α1 subgroup labeled by MB299B was the most vulnerable, as more than 50% degenerated after the H₂O₂ treatment (Fig. 5b–d).

We next determined whether these PAM subpopulations were rhythmically vulnerable to oxidative stress, focusing on the most vulnerable MB299B-labeled neurons (PAM-α1). We treated 7-day-old flies expressing GFP with the MB299B split-GAL4 at ZT8 and ZT20 with 10% H₂O₂, and the GFP-positive cell count was scored 7 days later. The H₂O₂ treatment at ZT8 caused the loss of around three to four neurons, whereas, on average, 7 neurons were lost from the treatment at ZT20. Thus, PAM-α1 indeed shows rhythmicity in vulnerability to oxidative stress, at least in LD (Fig. 5e, f). In DD, exposure to H₂O₂ at CT4 and CT16 both induced a significant loss of the MB299B-labeled neurons, but no difference between time points was observed (Fig. 5g). This finding suggests the presence of another PAM subpopulation that displays rhythmic vulnerability in DD.

The finding that MB299B-positive, PAM-α1 neurons are rhythmically vulnerable to oxidative insults in LD and postsynaptic to LNds raises the possibility that they exhibit physiological rhythms. Therefore, we measured levels of intracellular Ca²⁺, a second messenger that controls several cellular functions, in PAM-α1 neurons across 24 h in live flies (Fig. 6b). The calcium sensor GCaMP7s⁴⁸ was expressed with the MB299B driver, and the 3-day-old flies were entrained to LD cycles for four days before imaging in DD. Ca²⁺ levels in these neurons showed temporal variations, with peaks twice a day, around CT5 and CT17 (Fig. 6c). In per⁰ flies, GCaMP fluorescence levels were uneven across time points, and the bimodal pattern in w¹¹¹⁸ was lost, with a peak around CT17 (Fig. 6d). These results suggest that circadian inputs to the PAM-α1 neurons rhythmically modulate their activity that peaks just before midday and before midnight.

We next wanted to know whether decreased excitability could be neuroprotective. Therefore, we silenced PAM neurons by overexpressing the inward rectifier K+ channel, Kir2.1⁴⁹, with the R58E02 driver and tested whether this silencing prevents the degeneration of PAM neurons after the oxidative stress challenge. However, significantly fewer PAM neurons were present in the 14-day-old flies expressing Kir2.1 than in the control without H₂O₂ treatment, probably due to developmental impairments. Moreover, Kir2.1 overexpression did not prevent H₂O₂-induced PAM neuron loss (Supplementary Fig. S5). This finding suggests that neuronal activity levels are unlikely to be the cause of the selective and rhythmic vulnerability of PAM-α1 neurons.

Four-hour treatment with 10% H₂O₂ at ZT20 leads to the loss of approximately 13 DA neurons in the PAM cluster, most of which belong to the PAM-α1 subclass (Fig. 5a–d). We wondered whether this loss of approximately 10% of PAM neurons would have immediate or long-term effects on motor or non-motor functions. It was reported that synaptic dysfunction of a small subset of PAM neurons (expressing NP6510-GAL4) could impair the fly’s ability to climb¹⁴. To test whether the PAM-α1 subgroup has a similar role in climbing ability, we performed a startle-induced negative geotaxis assay²¹ after the short-term H₂O₂ treatment with increasing age. Climbing abilities of control and treated flies declined with age, as expected and as previously reported¹⁴. However, there was no difference between the control and experimental groups up to day 56, i.e., 49 days post-H₂O₂ treatment (Supplementary Fig. S6), except for day 49. These findings suggest that PAM-α1 is not involved in the control of climbing behavior.

H₂O₂ administration may affect sleep independently of its effect on PAM- α1 neurons. To distinguish these possibilities, we measured sleep while blocking the output of PAM-α1 neurons using a temperature-sensitive mutant of dynamin, Shibire^ts (Shi^ts). UAS-Shi^ts was driven with MB299B, and sleep and activity were monitored under 30 °C between day 11 and 13 after eclosion in flies that were raised at 19 °C. In comparison to control flies carrying only UAS-Shi^ts, MB299B>Shi^ts flies showed an increase and decrease in total sleep duration during the day and the night, respectively (Supplementary Fig. S7a). Sleep was more consolidated during the day (Supplementary Fig. S7b); however, the mean activity count during the waking period was reduced across 24 h, indicating hypoactivity (Supplementary Fig. S7c). Thus, inhibition of PAM- α1 synaptic output not only reduces activity levels but also modulates sleep. Although the effect on sleep by H₂O₂ exposure and synaptic block are not identical, possibly due to the temperature difference and the incomplete blockage of neuropeptide release by Shi^ts51,52, these results collectively suggest the involvement of PAM-α1 neurons in the regulation of sleep and activity. However, it remains to be determined if the effects on sleep by the H₂O₂ treatment result from the loss of PAM- α1 neurons.

These findings support the multiple-hit hypothesis and suggest that an insult in addition to H₂O₂ might accelerate PAM neurodegeneration. We speculated that per⁰ mutation might constitute the second “hit,” following our finding that per⁰ mutants display age-dependent PAM neurodegeneration (Fig. 2a–c). Therefore, we counted the number of PAM neurons over the course of aging in per⁰ flies after the H₂O₂ treatment. On day 14, per⁰ showed a substantial loss of PAM neurons as expected; however, after that, the rate of PAM neurodegeneration in the H₂O₂-treated flies was not higher than in non-treated flies, reaching the same number of PAM neurons on day 28 (Fig. 8b).

Increased mortality is a characteristic of PD patients⁵⁴. We monitored the life spans of w¹¹¹⁸ and per⁰ flies with and without H₂O₂ treatment. Whereas the H₂O₂ treatment did not affect the life spans of w¹¹¹⁸ flies, H₂O₂-treated per0 flies exhibited a significantly reduced lifespan compared to the control group (Fig. 8c, d). These results demonstrate that the combination of an oxidative insult and the per null mutation causes premature death of the animal.

Flies were reared in 12-h LD cycles in a humidified chamber at 25 °C on a standard corn meal medium. Clk856-GAL4³⁸ was a gift from Ralf Stanewsky. The following lines were previously described: HL9-GAL4⁶⁷, R58E02-GAL¹⁶, per¹,w⁶⁸, w;tim¹⁶⁹Pdf-GAL4⁷⁰, Clk1982-GAL4³⁸, DvPdf-GAL4⁴¹UAS-per16⁷¹, and UAS-Shibire^ts72. NP6510-GAL4 (113956) was obtained from the Kyoto Stock Center. The following lines were obtained from the Bloomington Stock Center: TH-GAL4 (Stock number 8848), R48B04-GAL4 (50347), 0104-GAL4 (62639), MB312B (68314), MB043B (68304), MB299B (68310), MB441B (68251), MB315C (68316), UAS- GCaMP7s (79032), UAS-Kir2.1 (6596), UAS-Catalase (24621), w¹¹¹⁸ (5905), Canton-S (64349), UAS-myrGFP, QUAS-mtdTomato(3xHA), and trans-Tango (77124).

Male flies were collected after hatching and entrained to 12 h/12 h-LD cycles in an incubator at 25 °C. Seven-day-old flies were transferred to an empty vial for 5 h, and then a filter paper soaked with 100 µl H₂O₂ (Sigma Aldrich, 216763) of a given concentration or H₂O was inserted in the vial for a defined duration. Flies were then placed in a vial containing standard corn meal agar food for 7 days before the analysis of DA neuron integrity. For circadian H₂O₂ treatment in LD, newly hatched flies were entrained to LD cycles, and on day 7, flies were treated with 5% or 10% H₂O₂ for 4 h at one of the 6 ZT time points (ZT0, 4, 8, 12, 16 and 20) following a 5-h food and water deprivation. After the H₂O₂ exposure, flies were placed in the vial with the standard food and maintained under LD cycles at 25 °C for 7 days. For circadian treatment in DD, newly hatched flies were entrained to LD cycles for 3 days and then released to DD. After three DD cycles, flies were treated with H₂O₂ at one of the CT time points (CT0, 4, 8, 12, 16 and 20)_, as described above. Treatment and subsequent incubation were performed in DD.

Immunostaining was performed on whole fly brains. Flies were decapitated, and the whole heads were fixed in 4% paraformaldehyde + 0.3% Triton X-100 on ice for 1 h. The heads were washed with 0.3% Triton X-100 in phosphate-buffered saline (PBT) three times at room temperature. The head cuticles were partly opened, and the heads were incubated in a blocking solution (5% normal goat serum in PBT) for 1 h at room temperature. The incubation with primary antibodies was performed over two nights at 4 °C. Then, the heads were washed three times and incubated overnight with secondary antibodies at 4 °C. After three washes, the brains were dissected entirely by removing the remaining cuticle and trachea and mounted in Vectashield mounting medium (Vector laboratories, H-1000-10). The slides were stored at 4 °C and protected from light until imaging. The primary antibodies and the concentrations were as follows: rabbit polyclonal anti-TH (Millipore, ab152) 1:100; mouse monoclonal antibody nc82 (Developmental Studies Hybridoma Bank) 1:100; polyclonal rabbit anti-GFP (Invitrogen, A6455) 1:500; and monoclonal mouse anti-HA (Covance, MMS-101P) 1:200. The secondary antibodies and their concentrations were as follows: Alexa Fluor anti-rabbit 488 (ThermoFisher, A21052) 1:250 and Alexa Fluor anti-mouse 633 (ThermoFisher, A11008) 1:250.

The feeding assay was performed as previously described with minor modifications³⁵. Male flies were starved for 5 h before the assay (beginning of feeding). A filter paper soaked with a 0.125 mg/ml solution of Brilliant Blue FCF was then inserted in the vial for a control group. In the viral containing the flies of the experimental group, a filter paper soaked with a 0.125 mg/ml solution of Brilliant Blue FCF with 10% H₂O₂ was inserted. The filter paper was kept in the vials for 4 h, after which the flies were flash-frozen in liquid nitrogen. Then, the flies were vortexed to detach their heads from bodies. A standard sieve (No. 25) was used to remove heads, and the bodies were transferred to 1.5-ml Eppendorf tubes. The bodies were homogenized with a motorized pestle in 1 ml of PBT and centrifuged at 13,000 rpm for 15 min to clear the debris. The absorbance of the supernatant was measured at 625 nm on a spectrophotometer. The measured value was divided by the number of flies homogenized in one tube.

MitoSOX Red (Invitrogen, M36008↗), a cell-permeable mitochondrial superoxide indicator, was used to measure endogenous ROS levels. Male flies were decapitated, and brains were dissected in Hanks’ Balanced Salt Solution (HBSS), followed by incubation in 5 μM MitoSOX for 15 min at 37 °C and three washes with warm HBSS. The brains were mounted in HBSS and imaged immediately with a Leica TCS SP5 confocal microscope or a Nikon Ax confocal microscope.

The trans-Tango assay was performed as described previously⁴³. Briefly, male flies were aged for 10 days at 18 °C to maximize trans-Tango expression with the optimal signal-to-noise ratio. The immunohistochemistry protocol described above was used to visualize the neurons and their projections.

The calcium sensor GCaMP7s⁴⁸ was expressed with MB299B split-GAL4, and the 3-day-old flies were entrained to LD cycles for 4 days. On the first day in DD following the LD entrainment, flies were placed under the fluorescence microscope in the perfusion chamber after the brief operation to remove a small part of the cuticle of the head capsule. Each experiment consisted of a session of 6-h to 8-h live imaging at 30-min intervals, starting at four different circadian times (CT4, CT10, CT16, and CT22). During each session, 4 flies were recorded, and on average, 13 (ranging from 5 to 19) GFP-positive neurons were imaged per fly. Images were acquired using the Leica DM550 fluorescent microscope. The exposed brains were constantly perfused with oxygenated saline to prevent desiccation or deterioration of the tissue during imaging. Throughout the experiment, the flies retained the ability to move their abdomen and legs. The death of a fly was determined by the absence of movement. Additionally, a sudden drop in fluorescence intensity and a rapid increase in background autofluorescence were considered indicators of tissue deterioration. Data from the flies that died or displayed signs of deterioration during the experiment were excluded from the analysis. The fluorescence intensity of individual cells was measured using the freehand selection tool of Fiji/ImageJ. The corrected total cell fluorescence (CTCF) value was obtained by subtracting the measured fluorescence value outside of MB299B neurons from the measured fluorescence value inside MB299B neurons. The following formula was applied to calculate CTCF: CTCF = integrated density − (area of the region of interest x mean fluorescence of background). The mean of the CTCF of individual cells per timepoint was calculated, and the normalized CTCF time series was obtained by dividing the mean CTCF of each timepoint by the lowest mean CTCF over 24 h.

The fly sleep assay was performed as described previously with minor modifications²⁷. Seven-day-old male flies were deprived of food and water for 5 h and treated with 10% H₂O₂ for 4 h starting at ZT20. The control group was treated with water only. After the treatment, flies were placed in individual glass tubes containing 5% agarose with 2% sucrose in the Drosophila Activity Monitoring (DAM) System 3 (Trikinetics) and entrained in 12 h:12 h-LD cycles at 25 °C for 3 days. Locomotor activity data were recorded during the subsequent 12 days in LD. We extracted sleep quantity from the activity data collected in 1-min bins (defined as bouts of inactivity lasting for 5 min or longer)⁷³. Sleep analysis was performed in MATLAB (MathWorks, version R2017a) using SCAMP v2 (Sleep and Circadian Analysis MATLAB Program) according to the instruction manual (Vecsey laboratory)⁷⁴. MB299B>Shi^ts and UAS-Shi^ts flies were raised at 19 °C in 12 h:12 h-LD cycles until 7 days old and placed in the activity monitor at 19 °C, 12 h:12 h-LD cycles. When they were 11 days old, the temperature was shifted to 30 °C until the end of the experiment. Locomotor activity and sleep data between day 11 and 13 were analyzed as described above.

The startle-induced negative geotaxis assay was performed as previously described⁷⁵ with minor modifications²¹. Twenty male flies were anesthetized with CO₂ and placed in a 100-ml graduated glass cylinder. The cylinder was divided into five equal zones and graded from 1 to 5 from the bottom to the top. The bottom of the cylinder was marked as zone 0. After 1 h of recovery from the CO₂ exposure, the climbing assay was performed at ZT2. Flies were gently tapped to the bottom of the column to startle them, from which they recovered quickly and started climbing the cylinder wall. The whole experiment was video-recorded for subsequent analyses. The number of flies that climbed up to each zone within the 20 s after startling was manually calculated, from which climbing index (CI) was calculated using the following formula: CI = (0 × n₀ + 1 × n₁ + 2 × n₂ + 3 × n₃ + 4 × n₄ + 5 × n₅) / n_total, where n_total is the number of total flies used in the experiment and n_x is the number of flies that reached the zone X. Three trials for each experimental group were performed at 30-s intervals, from which the median values were chosen as final data of an experiment. Three to four independent experiments were performed per group per age.

Lifespans of flies were assayed as previously described with minor modifications⁷⁶. Briefly, 1-day-old male flies were placed in the vials containing cornmeal agar food, 10 flies per vial, and maintained under 12 h:12 h-LD cycles at 25 °C. At 7 days old, treatment groups were exposed to 10% H₂O₂ for 4 h starting at ZT20, as described above. Control groups were treated with water only. Following the treatment, flies were returned to the vials with food. Flies were transferred to new vials twice a week, and the number of dead flies was recorded daily throughout the experiment.

Fly brains were imaged using a Leica TCS SP5 confocal microscope and a Nikon Ax confocal microscope. All images were analyzed using Fiji/ImageJ software (NIH). To count DA neurons, neurons positive for TH immunostaining or neurons labeled with fluorescent protein were counted manually using the cell-counter plugin of ImageJ through individual Z-stacks of the confocal images. To quantify the fluorescence intensity of the MitoSOX Red in the region of PAM neurons expressing GFP, the measurement feature of ImageJ was used after adjusting a proper threshold. To calculate the resulting fluorescence intensity of the MitoSOX Red in the region of PAM neurons, the CTCF value was obtained by subtracting the measured fluorescence value outside of the PAM region from the measured fluorescence value inside PAM neurons.

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

Supplementary Information Peer Review File Reporting Summary

Source Data