All the flies used have been validated and published previously. For the reporter experiments, the G0338-or G0431-Gal4 driver lines (which are both expressed in DAL neurons) are combined with the two other transgenes (period-promoter-FRT-luciferase and 20xUAS-FLP) that comprise the LABL system (Chen et al., 2012; Lin et al., 2021; Johnstone et al., 2022). For olfactory behavior, circadian and sleep experiments, progeny from a cross between one parent homozygous for two different transgenes and a second parent homozygous for a third transgene are used. The DAL specific driver line (G0431) is combined with the tubulinP-Gal80^ts transgene in the doubly transgenic parent (w¹¹¹⁸; G0431-Gal4; tubulinP-GAL80^ts). For the rest of this report, this genotype will be shortened to G0431; tubulinP-Gal80^ts. The other parent is one of: [w¹¹¹⁸; +/+; UAS-clk^Δ (Tanoue et al., 2004)], [w¹¹¹⁸; UAS-cyc^Δ; +/+ (Tanoue et al., 2004)], [w¹¹¹⁸; UAS-clk^RNAi; +/+ (VDRC 107576)] or [w¹¹¹⁸; UAS-cyc^RNAi; +/+ (VDRC 110455)]. These four genotypes will be shortened to: UAS-clk^Δ, UAS-cyc^Δ, UAS-clk^RNAi and UAS-cyc^RNAi. The tubulinP-Gal80^ts transgene makes the Gal80^ts protein ubiquitously. In the progeny flies that contain all three transgenes, at low temperature (20°C) the Gal80^ts protein is active and represses Gal4 from acting. At the restrictive temperature (29°C), the Gal80^ts protein is inactive and Gal4 becomes active. For assays of circadian locomotor activity and sleep, the parents and progeny from two of the crosses (those involving UAS-clk^Δ _or UAS-cyc^Δ) are used. The UAS-cyc^Δ _and UAS-clk^Δ _flies are from P. Hardin, although the UAS-clk^Δ stock is available through BDSC (#36319). The cyc^o, tim^o stocks along with their isogenic wild-type control flies are from A. Sehgal. The G0338 and G0431-Gal4 driver lines and the doubly transgenic parent (G0431; tubulinP-GAL80^ts) are from T. Tully.

Flies of the correct genotype (with the DAL-specific Gal4 driver lines [either G0338 or G0431], the LABL transgenic reporter and the UAS-FLP transgene) are assayed and analyzed as described previously (Johnstone et al., 2022).

Behavioral training is done as described previously with a few modifications to the induction conditions (Yin et al., 2023). Young flies (roughly 1 week of age) are used for behavior. Flies are collected and kept in groups of 100–150 individuals/vial and entrained to a 12-h lights on: 12-h lights off schedule at 20°C for 3d prior to training. Training starts between ZT = 14 and ZT = 16, and flies are returned to their dark period and kept at 20°C until testing. Testing always occurs around ZT = 16 to ZT = 18. For flies that undergo heat-shock induction, vials are shifted to incubators at 29°C under the same light:dark regimen.

Flies are trained in the olfactory avoidance-training paradigm developed by Tully and Quinn and modified to allow for automated training sessions (Tully and Quinn, 1985; Tully et al., 1994). A single-cycle of training consists of 90 s exposure to ambient air; 60 s of electric shock [the unconditioned stimulus which consists of 70 V pulses lasting 1.5 s and administered every 5 s (12 total)] accompanied by simultaneous exposure to 1 odor (the conditioned stimulus, CS+); 45 s of ambient air exposure to clear the first odor; 60 s of exposure to the second odor with no shock (the CS-condition), 45 s of ambient air to clear the second odor. This single training trial takes about 2.6 min. Spaced training consists of 10 single cycles separated by 15-min rest intervals. This training requires about 2 h 50 min of time. The odors that are used are 3-octanol and 4-methylcyclohexanol carried on an airstream.

Testing is done by placing flies in a choice point and giving them 2 min to decide between the CS + and CS-stimuli. The performance index = [the number of flies making the correct choice]−[the number of flies making the incorrect choice]/total number of flies, multiplied by 100. To avoid odor-avoidance biases, we calculate the performance index of every single N by taking an average performance of two groups of flies, one trained with 3-octanol as CS+, and the other with 4-methylcyclohexanol as the CS+. Flies were trained in a balanced manner, such that the sequence of shock-paired odors alternates, as well as the assignment of left vs. right arm at the choice point during testing. Data is presented as the standard error of the mean, and the Student’s T-test was used to evaluate statistical significance in pairwise comparisons.

Locomotor activity is measured as described previously using both parents and progeny from the behavioral crosses involving G0431; tubulinP-Gal80^ts and UAS-clk^Δ or UAS-cyc^Δ flies. Young flies (about 1-2-days post-eclosion) are entrained on a 12:12 light–dark cycle for 2 days before being loaded into the Drosophila Activity Monitor (DAM) System. The flies are allowed to recover for 2 days under the same light–dark cycle, at 20°C, shifted to constant darkness for 2 days, and then shifted to 29°C heat-shock until the end of the experiment. The SleepMat program is used to analyze periodicity and to generate the periodogram, and a custom program described previously is used to analyze and plot circadian locomotor activity (Sisobhan et al., 2022; Yin et al., 2023).

Sleep is measured as described previously using both parents and progeny from the behavioral crosses involving G0431; tubulinP-Gal80^ts and UAS-clk^Δ or UAS-cyc^Δ flies. Young flies (about 1-2-days post-eclosion) are entrained on a 12:12 light–dark cycle for 2 days at 20°C before being loaded into the DAM System. The flies are allowed to recover under the same light–dark cycle and temperature for 2 days, followed by a shift to 29°C starting at ZT 23 on day 3 until the end of the experiment. A custom program described previous is used to analyze and plot sleep figure (Yin et al., 2023).

Previously we showed that ubiquitous, post-training, inducible disruption of the circadian system interferes with 3-day (3d) memory formation (Yin et al., 2023). Chen et al. inferred that the circadian genes are expressed in the dorsal-anterior lateral (DAL) neurons, a pair of neurons located outside of the canonical central clock network, and showed that DAL neurons are important for 1d memory (Chen et al., 2012). Subsequent single-cell transcriptomic work shows that the circadian genes cycle, pdp1 and cry are likely expressed in the DAL neurons (Crocker et al., 2016; Supplemental Table 6, #287, #676 and #712). In this report we ask whether the circadian genes in those neurons are important for 3d memory.

Initially we use the LABL reporter system to ask if there is detectable oscillatory transcriptional activity in DAL neurons (Johnstone et al., 2022). Reporter activity is an indirect measure for CLK and CYC activity since the LABL reporter contains a minimal period promoter that results in oscillatory activity if CLK and CYC bind to a consensus E-box sequence (Bargiello et al., 1987). Figure 1A cartoons the necessary components in this system, and we test it in DAL using the G0338 or G0431 Gal4 driver lines (Chen et al., 2012).

Figure 1B shows reporter activity when FLP is expressed in DAL neurons. Luminescence (in arbitrary units) is plotted as a function of time for the G0338- (left side plot in orange) and G0431-driven (right side plot in purple) reporters, with the subjective day (light gray) and subjective night (dark gray) periods indicated. The closed circles represent the best fits for maxima and minima of the oscillating activity each day. Although the amplitudes are low, both G0338-and G0431-driven reporter activities oscillate. Morlet wavelet analysis (Figure 1C) indicates that the periodicity of the oscillations cluster around 24 h for all confidence intervals, consistent with circadian-regulated transcription. The violin plots to the right of each Morlet wavelet analysis show that the periodicities across the entire 9d interval cluster around 24 h for reporters activated using both drivers. Since CLK/CYC heterodimers are major contributors to oscillating transcription from the period promoter, they are likely to be expressed and active in DAL neurons, confirming findings from prior work (Bargiello et al., 1987; Chen et al., 2012; Crocker et al., 2016). Crocker et al. reported detecting the expression of the circadian genes cyc, pdp1 and cry in DAL neurons using a scRNA sequencing approach (Crocker et al., 2016).

Since CLK and CYC are likely to be expressed in DAL neurons and to function together to contribute to rhythmic LABL activity (Figures 1B,C), we inducibly disrupt them in DAL neurons to ask about the possible requirement for circadian genes in memory formation. The G0431-Gal4 driver is combined with the ubiquitously expressed tubulinP-Gal80^ts transgene to test four different target UAS-transgenes in triply transgenic flies. At low temperature (20°C), the Gal80^ts protein binds Gal4 and prevents expression from the UAS transgene. At the non-permissive temperature (29°C), the Gal80^ts protein becomes inactive and Gal4 activates transcription from the UAS target transgene. The four different UAS-transgenes make truncated, “dominant negative” versions of CYC or CLK (CYC^Δ or CLK^Δ), or RNAi directed against each respective gene. All parental stocks or triply transgenic (progeny) flies are trained with 10 cycles of spaced training, induced or not beginning 6–24 h after the end of training, and tested for 3d memory. Figure 2 shows the results when we test the effects of post-training, inducible disruption on 3d memory.

Although baseline performance varies among the different stocks, induction in any of the parental lines has no effect on 3d memory (compare the 4 leftmost pairs of blue and blue-striped bars). However, triply transgenic flies (red and red-striped bars) exhibit deficits in 3d memory upon induction. The magnitude of the impairment is nearly identical whether the induced product is a dominant negative CYC^Δ or CLK^Δ protein, or an RNAi molecule targeted against the cyc or clk gene. Induction of the CYC^Δ encoding transgene 3 h prior to behavioral testing (rightmost pair of bars) does not affect performance, showing that the CYC^Δ protein is not likely to affect retrieval. Taken collectively it is very likely that post-training disruption of cyc or clk in DAL neurons disrupts 3d memory.

Chen et al. previously showed that “classical” circadian mutants in Drosophila (other than period) do not affect 1-day (1d) memory of olfactory avoidance behavior (Chen et al., 2012). Sakai et al. reported identical results using the suppression of male courtship behavior testing at a later timepoint (Inami et al., 2021; Inami et al., 2022). Since we show inducible disruption of cyc or clk affects 3d memory (Figure 2), we test two widely studied mutants (tim^o and cyc^o) for their possible effects on 3d memory (see Figure 1D). The mutants and their isogenic wild-type control stocks were entrained to a 12:12 light:dark schedule for 3 days and then trained with 10 cycles of spaced training beginning about ZT16 under dim red lighting. Following training flies are returned to their pre-training light:dark cycle and tested for 3d memory. The solid bars represent the performance of the wild type controls, while the hatched bars show the results for the mutants. Under these conditions we do not detect any significant differences in 3d memory between mutants and wild-type flies.

Mechanistically, how is our inducible intervention affecting memory consolidation? To assess the possible effects of induction on circadian rhythmicity (Figure 3), we use the DAM to measure locomotor activity from the two different crosses ([G0431; tubulinP-GAL80^ts x UAS-cyc^Δ] and [G0431; tubulinP-GAL80^ts x UAS-clk^Δ]) that are used for behavior. Flies are entrained to a 12:12 light:dark cycle for two days, shifted to constant darkness for two days, and then maintained in darkness but shifted to 29°C to induce the heat-shock transgenes. Locomotor activity is recorded and analyzed over the following three full days. Rhythmicity is tested using a previously published method, and the resulting periodogram is presented in Figure 3A (Sisobhan et al., 2022). The quantitative analysis is shown in Tabular form at the bottom of Figure 3A. Five different groups are evaluated (the G0431; tubulin-Gal80^ts parent, the UAS-cyc^Δ parent, the UAS-clk^Δ parent, and the two progeny from the crosses). Because the two UAS parents behave identically, they are grouped together in the periodogram at the top of Figure 3A and represented using a blue trace. Similarly, the two progeny are nearly identical and grouped together and represented using a green trace. The G0431-Gal4; tubulin-Gal80^ts parent is represented using a red trace. The frequency of different periods is plotted as a function of period length. Induction does not substantially alter the periodicity of any of the groups including the progeny which are the only groups that disrupt memory upon induction.

Figure 3B shows the locomotor activity profiles for the same three groups (UAS parents, G0431; tubulin-Gal80^ts parent, and the progeny) plotted as a function of time across the experiment. The origin of the x-axis indicates when heat-shock induction (20°C to 29°C) begins and the negative and positive numbers denote the hours prior to, and after, induction. Locomotor activity of the progeny flies (green trace) appears as an arithmetic average of that of their parents (red and blue traces) and all flies maintain a similar general pattern of rhythmicity. In comparing the activity pre-and post-induction, both the parents (no memory deficits) and progeny (memory deficit) exhibit a subtle shift after induction (Figure 3B). They display a broadly distributed single peak during the subjective daytime period of constant darkness (−36 to −24 and −12 to 0 h prior to induction) and a much sharper and centered doublet activity peak post-induction (+24, +48). Since this change is not exclusive to the progeny of the crosses, this change is unlikely to contribute to the memory problems in those groups. ANOVA tests of the average periodicity for all genotypes also shows no significant statistical differences between groups, reinforcing the view that the subtle changes to circadian locomotor activity are unlikely to contribute to the memory deficits.

It is generally accepted that sleep is important for memory consolidation. Are our behavioral deficits in 3d memory attributable to disruptions to sleep? To investigate this, we use the DAM system to assess both steady-state and post-induction sleep profiles of the parents and progeny from the same crosses that are used for behavior. In Figure 4, we show the average sleep profiles across two days both prior to and after induction for the parents and progeny of one of the crosses. Panels A and B show the average sleep per hour for the G0431; tubulin-Gal80^ts and UAS-cyc^Δ parents. Panel C shows the sleep profile for the progeny from this cross. Analysis of the total sleep during the day and nighttime suggests that the values for the progeny are the average of the values for the two parents regardless of whether the flies are induced or not. This averaging is most apparent when the “theoretical” (arithmetic average, Panel D) of the parents is compared to the actual data of the biological cross (the progeny, Panel C). The same result is true for the other cross as well (see Supplementary Figure 1). The detailed analyses of sleep for all five groups are presented in Supplementary Figure 2. We also show the kinetics of the effects of induction on sleep levels and on the architecture of sleep in the progeny from both crosses (grouped together, see Supplementary Figure 3). In general, heat-shock induction increases the amount of daytime sleep and the average length of daytime bouts, which are consistent with a more consolidated sleep state. Part of this effect is attributable to heat-shock itself increasing sleep (Lenz et al., 2015). However, the average number of daytime bouts also increases, and this pattern is usually associated with poorer quality of sleep. The effects of heat-shock on nighttime sleep all trend in the direction of less sleep consolidation—decreases in total amounts, with increases in bout number and decreases in bout lengths. However, like the effects of induction on circadian rhythms, the changes in sleep in the behaviorally relevant (progeny) flies are very close to an arithmetic average of the effects of induction on the parents. The presumed induced protein(s) do not appear to appreciably alter either circadian locomotor activity or sleep patterns more than what heat-shock induction itself does in the parents. Thus, it seems quite unlikely that any of the circadian locomotor-activity or sleep-related changes are contributing to the behavioral deficits.