Flies were fed standard Drosophila food containing molasses, cornmeal, and yeast. They were housed in a 25°C incubator (DR-36VL, Percival Scientific, Perry, IA, United States) under 12 h:12 h light-dark cycles and 65% humidity. To target DN1p neurons, the wake-Gal4 line was used (Liu et al., 2014). To visualize localization of SLOB and NaKβ, UAS-SLOB-HA and UAS-NaKβ-myc lines were used (Tabuchi et al., 2018). For electrophysiological and immunocytochemical experiments, newly eclosed adult wake-Gal4 > UAS-CD8::GFP, > UAS-SLOB-HA, or > UAS-NaKβ-myc female flies were collected and transferred to food vials at a density of ∼10 flies/vial and maintained at 25 °C under a 12 h:12 h light-dark cycle with 65% humidity. Flies were transferred into fresh food vials every 2 days. Flies that were 2–4 days old were considered “young,” and flies that were approximately 2 months old (60–67 days old) were considered “aged.”

We conducted electrophysiological recordings from DN1p neurons with ex vivo configuration (i.e., isolated brain preparation). Flies were anesthetized by chilling on ice (up to 10 min). Their heads were then isolated and placed in a dissecting chamber. Brains were removed and dissected in a Drosophila physiological saline solution (101 mM NaCl, 3 mM KCl, 1 mM CaCl₂, 4 mM MgCl₂, 1.25 mM NaH₂PO₄, 20.7 mM NaHCO₃, and 5 mM glucose; pH 7.2) pre-bubbled with 95% O₂ and 5% CO₂. To increase the likelihood of successful recordings, brains were treated with an enzymatic cocktail of collagenase (0.1 mg/mL), protease XIV (0.2 mg/mL), and dispase (0.3 mg/mL) at 22°C for 1 min. Then the glial sheath surrounding the brain was focally and carefully removed using sharp forceps. The surface of the cell body was cleaned with a small stream of saline pressure-ejected from a large-diameter pipette under visualization of a dissecting microscope. DN1p neurons were visualized with GFP fluorescence with a PE300 CoolLED illumination system (CoolLED Ltd., Andover, United Kingdom) on a fixed-stage upright microscope (BX51WI; Olympus, Japan). One neuron per brain was recorded.

Perforated patch-clamp recordings of DN1p neurons were performed as described (Tabuchi et al., 2018). Patch pipettes (9–12 MΩ) for perforated patch-clamp were fashioned from borosilicate glass capillary (without filament) using a Flaming-Brown puller (P-97, Sutter Instrument) and further polished with a microforge (MF200, WPI) before filling the internal pipette solution with 102 mM potassium gluconate, 0.085 mM CaCl₂, 0.94 mM EGTA, 8.5 mM HEPES, 4 mM Mg-ATP, 0.5 mM Na-GTP, 17 mM NaCl, pH7.2. Escin (Santa Cruz Biotechnology) was prepared as a 50 mM stock solution in water (stored up to 2 weeks at −20°C) and added fresh into the internal pipette solution to a final concentration of 50 μM. Due to the lightsensitivity of escin, filling syringes were wrapped with aluminum foil. Pipette tips were dipped into a small container with escin-free internal pipette solution for approximately 1 s, and then backfilled with the escin-containing solution from the filling syringe. Air bubbles were removed by gentle tapping. Escin pipette solutions remained stable for several hours after mixing in the filling syringe, with no evidence of precipitation. Junction potentials were nullified, a high-resistance seal was formed, and perforated patches were allowed to develop spontaneously over time. After breakthrough was evident (based on the gradual development of a large capacitance transient in the seal test window), access resistance was first monitored with the membrane test function and then continuously during the final steps of the perforation process until it became stable (access resistance stably < 40 MΩ). Cells that showed signs of “mechanical” break-in (i.e., a significant increase of time constant of transient capacitive current) were excluded from further data acquisition. During the recording, the bath solution was continuously perfused with saline with a gravity-driven system. Recordings were acquired with an Axopatch 1D, 200A, or 200B amplifier (Molecular Devices) and sampled with PCIe-6341 interface (National Instrument) controlled by Wavesurfer¹ software (for Axopatch 1D) or Digidata 1550B (Molecular Devices) controlled by pCLAMP 11 (Molecular Devices). The voltage signals were sampled at 10 kHz and low-pass filtered at 1 kHz.

To clearly dissociate action potential spikes from postsynaptic potentials (PSPs), sharp electrode intracellular recordings of DN1p neurons were performed as described (Liu et al., 2017). Sharp electrodes from quartz glass with a filament (OD/ID: 1.2/0.6 mm) were fabricated with a laser-based micropipette puller (P-2000, Sutter instrument) and backfilled with 1 M KCl, with resistances of 120–190 MΩ. Solutions were filtered using a 0.02-μm syringe filter (Anotop 10, Whatman). We inserted an electrode into the cell body of DN1p neurons expressing GFP. Impalements of the intracellular electrode were induced using the shortest “buzz” pulses and only buzzing when the electrode was not moving. Because stabilizing the cell membrane potential takes at least 1 min, recordings of membrane potentials began after the cell membrane potential was stable. To separately measure PSPs, the tonic hyperpolarizing current was injected into the targeted cell. We aimed to keep the somatic membrane potential between –105 and −110 mV. The measured PSPs were likely a mix of excitatory and inhibitory synaptic events because targeted cells were hyperpolarized to values near or beyond the reversal potential for inhibitory synaptic currents (up to −110 mV at somatic observation based on the lack of hyperpolarizing PSPs). Recordings were acquired with an Axoclamp 2B with HS-2A × 1 LU headstage (Molecular Devices) and sampled with Digidata 1550B interface, both of which were controlled on a computer using pCLAMP 11 software. The signals were sampled at 10 kHz and low-pass filtered at 1 kHz.

Electrophysiological analysis was performed in MATLAB (MathWorks). To quantify spontaneous firing activity, we used the coefficient of variation (CV) of interspike intervals (ISI), a global measure of irregularity defined as the dispersion of the ISIs (Holt et al., 1996). We also calculated the local variation (LV) as alternative measures of local irregularity by computing the dispersion of the two adjacent ISIs. LV is defined as

where ISI_i is the ith ISI and n is the number of ISIs (Shinomoto et al., 2003). To assess the shape of the ISI distribution, we calculated skewness and kurtosis based on the histograms. Skewness indicated the symmetry of the ISI distribution around the mean, and kurtosis mirrored the peak of the ISI histogram. To quantify evoked spiking activity, spikes were elicited in response to current injections with 300-ms stepping pulses at 5-pA increments up to 35 pA. The f-I curve was computed by sorting the level of injected current. The current threshold (minimal current to evoke spiking) and the slope of the f-I curve were determined by linear regression of the curve from the point of initial spiking. Input resistance was calculated from the voltage change obtained by injecting a hyperpolarizing current of 10 pA. Degrees of spike frequency adaptation were assessed by plotting the distribution of instantaneous spike frequency during the depolarizing current injection. Half-life and plateau variables of the process curve of spike frequency adaptation were determined by fitting with a monoexponential function. To detect and quantify PSPs, a median filter with a time constant of 3 ms was applied to the unfiltered membrane potential. Background noise power was computed based on root mean square values from the all-points amplitude in each data set. These computations were used to define both event-finding and noise-rejection criteria, which consists of a minimum-allowed amplitude. To manually check computed results, we visualized all detected PSP events and their peak and quantified amplitudes. We sorted individual PSP epochs and defined their amplitude as the difference between the maximum/minimum potential of each PSP and the mean membrane potential of the entire trace. The relative rising slope of the PSP was calculated as the change in voltage amplitude per millisecond. To estimate the temporal structure of a state transition between miniature PSP (mPSP) and spike-induced PSP (PSP), discrete-time Markov chain was used. We first classified the estimated mPSP and PSP with the k-means clustering algorithm (Dorgans et al., 2019), which was based on average PSP amplitude of each dataset. We confirmed that the separation is a binary distribution by estimating that mPSP has a smaller amplitude and PSP has a larger amplitude (Wierenga and Wadman, 1999). We then created a transition matrix to calculate transition probability.

Brains were dissected in a Drosophila physiological saline solution, fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30 min at room temperature, and then washed in PBS. To improve the penetration of antibodies when staining brains, the glial sheath enveloping the brain was carefully removed. Samples were incubated with rat anti-HA at 1:100 (3F10, Roche) or mouse anti-MYC (9E10, Sigma-Aldrich) at 1:50 on a shaker at 4°C for 48 h. After washing 3 times with PBS + 0.1% Tween 20 for 15 min each, samples were then incubated with Alexa Fluor 488-conjugated anti-rat (Invitrogen, 1:1,000) for SLOB-HA, or Alexa Fluor 488-conjugated anti-mouse (Invitrogen, 1:1,000) for NaKβ-myc on a shaker at 4°C for 48 h. After washing 3 times with PBS + 0.1% Tween 20 for 15 min each, samples were cleared in 70, 80, 90, and 100% glycerol in PBS for 5 min at room temperature. Samples were then mounted with a coverslip and Vectashield mounting medium (H−1000, Vector Laboratories). Images were taken under a 100 × magnification objective lens using a Leica TCS SP8 gated stimulated emission depletion 3X system (Leica Microsystems) and acquired as 1,024 × 1,024 pixels (16 bit). A slice having maximum nuclear diameter was used to quantify the appropriate region of interest from each cell. After the acquisition, images were preprocessed by the iterative deconvolution algorithm of the Huygens Professional deconvolution package (Scientific Volume Imaging, Netherlands). Fiji (ImageJ) was used to quantify the intensity of the total, plasma membrane, and perinuclear signals. Background intensity adjacent to the region of interest was measured and subtracted.

Statistical analyses were performed using Prism software (GraphPad, version 9.3.1.). To compare two groups of data, normally distributed data were compared with t-tests, and non-normally distributed data were compared with Mann–Whitney U-tests. To compare more than two multiple-group comparisons, two-way ANOVA with multiple comparisons was used. A p-value < 0.05 was considered statistically significant, with asterisks indicating p-values as follows: *p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001. ns indicates non-significance. All error bars represent means ± SEM averaged across experiments.

The molecular clock dynamically modulates the neural activities of clock networks to regulate circadian physiology and behavior (Allen et al., 2017). Age-related electrophysiological changes in clock neural networks occur in both Drosophila and mammals (Watanabe et al., 1995; Curran et al., 2019). However, most research has focused on the clock neuron firing rate, and we do not know if aging influences the temporal structural pattern of clock neuron firing. We focused on posterior DN1 (DN1p) neurons because of their contributions to sleep regulation. Moreover, specific temporal patterns of spontaneous firing activity in DN1p neurons are different at midday (ZT6–8, Zeitgeber Time 6–8) and mid-night (ZT18–20). Thus, they differentially affect sleep architecture, even if the mean firing rate is unchanged. Thus, we compared spontaneous activity in DN1p neurons between young and aged flies. In young flies (2–4 days old), DN1p neurons exhibited irregular firing at ZT6–8 but regular firing at ZT18–20 (Figure 1A). These observations are consistent with another report that used a slightly different age range (4–8 days old). In contrast to young flies, aged flies (60–67 days old) appeared to exhibit similar firing at ZT6–8 and ZT18–20 (Figure 1A).

To identify biophysical parameters underlying age-associated electrophysiological changes, we quantified intrinsic membrane properties during spontaneous activities. We found that the resting membrane potential (Figure 1B), mean firing rate (Figure 1C), spike waveform kinetics (e.g., afterhyperpolarization) (Figure 1D), and spike onset depolarization duration (time from threshold to peak depolarization) (Figure 1E) significantly differed between young and aged DN1p neurons at ZT18–20, but not at ZT6-8. These results suggest that age-related differences in DN1p activity patterns may be due, in part, to changes in intrinsic membrane properties at ZT18–20 rather than ZT6–8.

The DN1p circadian clock adjusts ionic flux in a time-dependent manner to alter its own neural activity. Also, spontaneous activity in DN1p neurons shows the clock-generated formation of unique temporal spiking patterns, defined by the second-order temporal structure of ISIs. Because we found remarkable age-associated changes in the pattern of DN1p neural activity, we analyzed how aging alters the temporal structure of ISIs. First, we quantified the autocorrelation function based on spike timing. The structured autocorrelation function only occurred in DN1p neurons of young flies at ZT18–20 (Figure 2A). Next, we examined the relationship between spike irregularity and aging. The CV of ISIs was used as a global metric of spike irregularity statistics (Figure 2B), the LV of adjacent ISIs was used as a local metric of spike irregularity statistics (Figure 2C). In both cases, the temporal structural irregularity was significantly greater in aged vs. young DN1p neurons at both ZT6–8 and ZT18–20.

To define the shape of the ISI histogram (Supplementary Figure 1), we quantified the degrees of skewness and kurtosis. The spiking patterns in aged DN1p neurons showed significantly different skewness (Figure 2D) and kurtosis (Figure 2E) in young vs. aged DN1p neurons at ZT18-20, but not ZT6-8. We assessed how activity fluctuations differ in the frequency domain with continuous wavelet transform. We found that aged DN1p neurons showed an altered spectral power structure (Supplementary Figure 2). These results indicate that aging dramatically alters the temporal structure of spontaneous firing patterns of DN1p neurons, especially during nighttime.

To further delineate biophysical parameters, we assessed age-associated changes in intrinsic membrane properties by measuring the membrane potential dynamics of DN1p neurons in response to current injections (Figure 3A). The frequency-current relationships (f-I curve) showed greater excitability in aged DN1p neurons vs. young DN1p neurons (Figure 3B). However, based on the linear regression slope, the f-I curve did not significantly differ between young and aged flies, regardless of the circadian timing (Figure 3C).

Next, we quantified the degree of spike frequency adaptation by fitting instantaneous spike frequency with a monoexponential relation (Figure 3D). When sustaining a depolarizing current injection, all experimental conditions showed a spike frequency adaptation. We also compared the relation, half-life (Figure 3E), and plateau (Figure 3F) variables of the fitting. Although the half-life did not significantly differ between young and aged DN1p at ZT6-8, it was significantly less in young vs. aged DN1p neurons at ZT18-20 (young at ZT6-8: 6.9 ± 1.7 ms, aged at ZT6-8: 8.7 ± 2.5 ms, young at ZT18-20: 2.8 ± 0.67 ms, aged at ZT18-20: 12.8 ± 3.79 ms) (Figure 3E). These findings indicate that aging weakened the adaptation strength. The plateau significantly differed between young and aged DN1p neurons at both ZT6-8 and ZT18-20 (young at ZT6-8: 21.03 ± 1.92 ms, aged at ZT6-8: 26.5 ± 1.56 ms, young at ZT18-20: 20.7 ± 1.76 ms, aged at ZT18-20: 27.1 ± 2.1 ms) (Figure 3F), suggesting that aged DN1p neurons have a lower baseline of depolarization sensitivity.

We also measured input resistance as a passive membrane property. The input resistance did not significantly differ between young and aged DN1p neurons at both ZT6-8 and ZT18-20 (Figure 3G).

To test whether synaptic input contributes to the age-related changes of spike patterns in DN1p neurons, we used quartz glass with a sharp electrode to measure spontaneous synaptic input. While constantly injecting a hyperpolarizing current from a sharp electrode, spontaneous postsynaptic potential (PSP) readily occurred (Figure 4A). Individual PSP epochs were sorted to show the cumulative probability of individual amplitude (Figure 4B) and inter-event interval (Figure 4C). These PSP events were quantified in functional biophysical parameters. Whereas the relative rising slope of the PSPs did not significantly differ between young and aged DN1p neurons at both ZT6-8 and ZT18-20 (Figure 4D), the PSP amplitude was significantly greater in aged DN1p neurons at ZT6-8 but not at ZT18-20 (Figure 4E). In contrast, the PSP frequency was significantly lower in aged samples at ZT18-20 but not at ZT6-8 (Figure 4F).

We also assessed if aging alters state transition probability in the temporal structure of synaptic input sequences. To do this, we created a discrete-time Markov chain showing the state transition between estimated miniature PSP (mPSP) and PSP (Figure 4G). The state transition probability increased from PSP to mPSP in young and aged DN1p neurons at ZT6-8 but not at ZT18–20. These results indicate that aging leads to directional changes in synaptic inputs of DN1p neurons between daytime and nighttime. These changes should contribute to age-related changes in DN1p spike patterns by interacting with intrinsic membrane properties.

Slowpoke binding protein (SLOB) and Na⁺/K⁺ ATPase β subunit (NaKβ) contribute to membrane potential dynamics in DN1p neurons (Tabuchi et al., 2018). Moreover, these molecules showed circadian-dependent changes in their subcellular localization when regulated by clock and wake signaling in circadian neuronal networks. We hypothesized that aging disrupts the localization of SLOB and NaKβ, leading to changes in electrophysiological properties. To determine if the subcellular localization of these factors changes with age, we used transgenic flies expressing HA-tagged SLOB (SLOB-HA) and myc-tagged NaKβ (NaKβ-myc) in their DN1p neurons. Because DN1p neurons are small and have limited accessibility for quantification, we used stimulated emission depletion microscopy to quantify the subcellular patterns in these neurons. With this approach, we readily observed the subcellular patterns of SLOB-HA and NaKβ-myc (Figures 5A, 6A). We found that total SLOB expression was significantly greater in young vs. aged DN1p neurons at ZT18-20 but not at ZT6-8 (Figure 5B). This age-related effect was remarkably greater at the plasma membrane (Figure 5C). Additionally, we found that SLOB showed a significant increase in the perinuclear region in aged flies compared to young flies (Figure 5D), suggesting a possible mistargeting to membrane trafficking pathways. We also quantified the heterogeneity of SLOB expression within DN1p neurons based on signal variability (Figure 5E). We found that the signal variability was remarkably greater in young vs. aged DN1p neurons at ZT6-8, but not significantly different at ZT18-20. Subcellular variability also did not differ between the plasma membrane (Figure 5F) and the perinuclear region (Figure 5G).

We also assessed the expression patterns of myc-NaKβ in DN1p neurons (Figure 6A). NaKβ expression was significantly greater in young vs. aged DN1p neurons at ZT18-20 but not at ZT6-8 (Figure 6B). Similar to SLOB, NaKβ expression was remarkably greater at the plasma membrane (Figure 6C) but not at the perinuclear region (Figure 6D) in aged DN1p neurons. We also analyzed changes in the signal variability of NaKβ. Similar to SLOB, the signal variability of NaKβ was most heterogeneous in young DN1p neurons at ZT6-8 and significantly less heterogeneous in aged DN1p neurons (Figure 6E). The signal variability did not significantly differ in the perinuclear compartments under any experimental condition (Figure 6G), suggesting that such heterogeneity was largely driven by differences in the plasma membrane compartments (Figure 6F).