Dual PDF Signaling Pathways Reset Clocks Via TIMELESS and Acutely Excite Target Neurons to Control Circadian Behavior

PDFR is expressed in most of the pacemaker neurons and most of these neurons respond to PDF application in ex vivo preparations with increases in cAMP levels [46],[47]. We tested the function of the cAMP-dependent protein kinase A (PKA) in clock neuron-driven behavior. cAMP is the canonical activator of PKA activity. cAMP binds the PKA regulatory (R) subunit releasing the catalytic (C) subunit to phosphorylate substrates (reviewed in [48]). While PKA signaling has been implicated in circadian clock function [49],[50], its precise role in mediating PDFR signaling has not been defined.

Using the Gal4-UAS system, we expressed a type I regulatory subunit of PKA that is defective for cAMP binding (U-PKA-R1dn), thereby rendering endogenous catalytic subunits insensitive to cAMP activation [51],[52]. Initially we drove expression using circadian drivers, such as cwo-G4 (also known as c632a; [5]) and examined behavior under standard 12:12LD conditions. cwo-G4 is a GAL4 enhancer trap insertion just upstream of the transcription start site of core clock gene cwo and drives expression in clock neurons [5]. In addition to expression in the PDF clock neurons [5], we examined cwo-G4 driven nuclear green fluorescent protein (GFP) expression in pacemaker neurons using PER co-labeling. We observed GFP expression mainly in the circadian pacemaker neurons, with limited expression in other brain regions (Figure S1; unpublished data). Among pacemaker neurons, we observed GFP in non-PDF clock neurons, including all of the LNds and the PDF(−) s-LNv, as well as a number of DN1s and DN3s (Figure S1). Broad expression of PKA-R1dn with cwo-G4 (Figure 1C) was able to phenocopy many features seen in pdfr mutants (Figure 1D). These flies exhibited reduced morning anticipation and phase advanced evening activity under LD cycles, and were nearly arrhythmic in constant darkness, thus mimicking the three canonical behavioral phenotypes of pdf and pdfr mutants (Figure 1A–1E; Tables 1 and 2).

Given that the majority of PDFR functions mapped to non-PDF clock neurons, we then asked whether these PKA-R1dn effects were due to its expression in non-PDF neurons by blocking PDF neuron expression with pdf-G80. Importantly, we did not detect cwo-G4 driven GFP expression in the PDF(+) neurons, verifying effective inhibition of cwo-G4 activity in those neurons by pdf-G80 (Figure S1). We observed once again that these flies behaved similarly to pdf and pdfr mutants, exhibiting the three hallmark phenotypes of reduced morning anticipation, phase-advanced evening activity, and reduced rhythmicity in constant darkness (Figure 1F–1J; Table 1), providing strong evidence for PKA function in non-PDF neurons in mediating PDF effects in circadian neurons. The addition of pdf-G80 did improve the rhythmicity (power-significance [P-S]) in constant darkness, suggesting that PKA activity in the PDF neurons contributes to a portion of this phenotype (Table 2, p<0.005). Nonetheless, the morning and evening phenotypes continue to map to non-PDF neurons.

Given that cwo-G4 also drives limited expression in some non-circadian areas we decided to examine PKA function using an independent circadian driver. Previous work from our laboratory demonstrated that morning and evening anticipation phenotypes of pdfr mutant behavior can be rescued by expressing U-pdfr in non-PDF neurons using cry13-G4 and pdf-G80[35]. Here we expressed U-PKA-R1dn in PDF(−) circadian cells using this Gal4/Gal80 combination and found that inhibition of PKA activity also results in morning and evening anticipation similar to pdfr- (Figure 2A, 2B, 2D, 2E; Table 1). If PDFR and PKA are operating in the same pathway, we would expect that U-PKA-R1dn expression in a pdfr^han5304 (pdfr-) mutant background would exhibit phenotypes comparable to either PKA-R1dn or pdfr mutants alone. Indeed, we found no differences in morning or evening behaviors among pdfr-, PKA-R1dn expression, or PKA-R1dn expression in the pdfr- background using cry13-G4/pdf-G80 (Figure 2B, 2C; Table 1), suggesting that PDFR and PKA operate in a common pathway within these cells.

Rescue or suppression of mutant receptor function by expression of activated downstream signaling components is a powerful in vivo method to elucidate signaling pathways [53]. We also attempted to suppress pdfr mutant phenotypes by expressing a PKA catalytic subunit (U-PKA-mC*), which is defective for regulatory subunit binding and is therefore constitutively active [52]. We expressed PKA-mC* in PDF(−) circadian neurons in pdfr mutants and observed that it rescued the reduced morning anticipation and phase-advanced evening activity onset characteristic of pdfr mutants (Figure 2; Table 1). There were no notable effects of PKA-mC* expression in a wild-type background in LD (Figure 2F–2J; Table 1). While PKA-mC* rescued LD phenotypes, we did not observe rescue of DD rhythms, suggesting that PKA-mC* is not expressed at the appropriate levels, cAMP inducibility may be required, and/or that non-PKA signaling pathways may contribute to PDFR function in DD (Table 2). We obtained similar results in the absence of pdf-G80, thus allowing PKA-mC* expression in the PDF neurons, suggesting that lack of PKA activity in PDF neurons is not responsible for the lack of DD rhythm rescue by PKA-mC* (Table 2). Taken together, our genetic evidence, especially our ability to bypass PDFR function with an activated form of PKA, indicate that PKA is a major mediator of PDFR signaling in non-PDF clock neurons. PKA activity is both necessary and sufficient for the execution of most PDFR-mediated behaviors.

We have previously shown that we can rescue pdfr mutant morning anticipation and DD rhythmicity, but not evening anticipation, by expressing wild-type pdfr only in DN1p circadian neurons using the Clk4.1-G4 driver [41]. To test whether PKA also functions in the DN1p we expressed PKA-R1dn using Clk4.1-G4 and found modestly reduced morning anticipation and DD rhythmicity, but no change in evening anticipation, complementing our observations for pdfr rescue (Figure 3A–3D; Tables 1 and 2) [35],[41]. Given that cwo-G4/pdf-G80 driven PKA-RIdn exhibited more robust morning and DD phenotypes (Tables 1 and 2), we hypothesize that non-PDF, non-DN1p cells may also contribute to these phenotypes and/or this driver combination more strongly inhibits PKA within the DN1p than Clk4.1-G4. To address the neural substrates of PKA function in evening anticipation, we used the mai179-G4 driver in combination with pdf-G80, which drives expression in the single PDF(−) sLNv, and the CRY(+) subset of LNd with variable expression in a 1–2 DN1s [30],[54],[55]. Here we found that PKA inhibition phase advances evening anticipation similarly to pdfr mutants (Figure S2). In keeping with a model in which the CRY+ LNd and fifth PDF(−) sLNv selectively regulate evening anticipation, PKA-R1dn driven by mai179-G4/pdf-G80 had no effect on morning anticipation (Figure S2; Table 1) or DD rhythms (Table 2).

Rhythms in PDF levels are apparent in the terminals of LNv neurons and rhythmic PDF release is thought to contribute to the temporal encoding of the PDF signal; however, rhythmic PDF may not be necessary for rhythmic behavior or clock function [28],[40],[56]. We hypothesized that rhythmic control of signal processing within cells receiving the PDF signal may contribute to the robustness of this pathway. To determine whether PKA subunit transcripts are under circadian clock control in DN1p clock neurons, we expressed UAS-membrane GFP (U-mGFP) using Clk4.1-G4 and isolated these neurons by fluorescence-activated cell sorting (FACS). After RNA isolation and linear amplification (see Materials and Methods), we examined the transcript levels of the three catalytic PKA subunits (C1, C2, and C3) and two regulatory subunits (R1 and R2) by quantitative real-time (RT)-PCR. Transcript levels of three PKA subunits (PKA-C1, PKA-R1, and PKA-R2) oscillate in phase, with coincident peaks in the mid-day (ZT4) (Figure 3E–3G). Peak transcript levels were reduced and rhythms were not detected in the per¹ mutants consistent with circadian clock control. The two other PKA transcripts (C2, C3) were near the limits of quantitative detection. Thus, not only is PKA activity likely controlled via rhythmic inputs of PDF-driven cAMP production but PKA is also rhythmically controlled at the level of gene expression. We hypothesize that these dual mechanisms collaborate to provide a robust time-of-day PKA signal to synchronize non-PDF to PDF oscillators.

The ability of PDF neurons to reset molecular clocks in non-PDF neurons has been powerfully demonstrated by selective manipulation of circadian period in PDF neurons and examination of molecular oscillations in non-clock cells [39],[41],[55]. To determine the direct molecular targets of PDF neuron signaling, we rescued clock function selectively in PDF neurons of arrhythmic per¹ mutants and assayed molecular oscillations in per-less non-PDF neurons. By examining molecular changes in per¹ non-PDF neurons, we removed the possibility that identified changes would be indirect through a functioning circadian clock and/or per. In addition, we examined molecular changes on the first day of constant darkness, removing the possibility that light signaling via CRY is responsible for any changes. We rescued per in PDF cells (per¹;pdf-G4/+;U-per16/+; “pdfPER” flies) and examined non-PDF-expressing circadian cells (LNd and DN1) in brains stained for the clock components TIM (Figure 4) and PDP1ε (Figure 5). First we demonstrated that transgenically supplied PER cycles in the sLNvs and rescues oscillations in TIM levels and nuclear localization in those cells (CT24), indicating at least a partially rescued sLNv clock consistent with prior studies (Figure 4A) [55].

We then examined the consequences of the rescued sLNv clock on non-PDF per¹ DN1 and LNd neurons. Consistent with the fact that these cells lack per or a fully functioning circadian clock, TIM is predominantly expressed in the cytoplasm at all times of day (Figure 4) [13]. However we found stark changes in the levels and phase of TIM oscillation in DN1 neurons despite the lack of PER. In per¹ controls, we observed a low amplitude TIM oscillation with an inappropriate day-time peak (Figure 4). In pdfPER rescue flies, TIM cycling amplitude increases with elevated peak levels and its oscillation phase is synchonized with that of TIM in the PDF neurons (Figure 4). TIM levels and phase in LNds are also modified by the PDF-neuron clock, but these effects are much smaller than those in the DN1s (Figure 4), consistent with our prior finding that the DN1 are more strongly reset by PDF neurons than the LNd [41]. To determine if these effects on TIM are specific, we also assayed a second clock component PDP1ε, a core circadian transcription factor directly activated by CLK/CYC [3]. We found that PDP1ε exhibits comparable levels between per¹ and pdfPER rescue flies with only a modest change at ZT18 in the LNd (Figure 5). The absence of strong effects on PDP1ε suggests that PDF-mediated inputs to the molecular clock are unique to TIM. This result also argues strongly against large PDF effects on CLK/CYC driven transcription, a major determinant of PDP1ε levels [3]. The strength of the effects on TIM evident in the absence of a functioning clock, per and light, suggest that TIM is a direct target of PDF signaling.

Our results suggest PKA mediates PDF effects and that PDF targets TIM. To test whether PKA can influence TIM levels, we expressed PKA-R1dn in non-PDF circadian cells using cwo-G4 in combination with pdf-G80 in a per¹ mutant background and examined TIM levels (Figure 6). Flies were entrained and dissected at four time points across the LD cycle, and stained for TIM. We observe reduced peak levels of TIM in both the LNd and DN1 at ZT24 with a non-significant trend developing by ZT12 during the light period. In these per¹ flies, TIM remains in the cytoplasm throughout the 24-h cycle. These results indicate that PKA activity positively regulates TIM accumulation in the LNds and DN1s in the absence of a functioning clock suggesting a direct effect on TIM.

We have shown that the PDF-cell clock specifically regulates TIM in non-PDF circadian cells (Figures 4 and 5) and that reducing PKA activity in non-PDF cells leads to reduced TIM levels (Figure 6). Our behavior data show that PKA acts in the PDF signaling pathway downstream of PDFR (Figures 1 and 2). We therefore determined whether loss of PDF would mimic reduced PKA activity and result in reduced TIM. Here we examined loss of PDF (pdf¹) in an arrhythmic per¹ mutant background and examined the effects of the PDF peptide on TIM at the end of DD1 (CT24). We compared per¹ and per¹;;pdf¹ flies and observed that TIM is strongly reduced in the absence of PDF in both LNd and DN1 neurons (Figure 7). We observe a nearly 50% reduction in TIM staining intensity in the absence of PDF, which is comparable to or even larger than the effect than we observed with PKA inhibition. One possibility is that expression of PKA-R1dn may not completely interrupt the signaling cascade, whereas pdf¹ is a confirmed null mutation [37]. These results provide further independent support for the hypothesis that the PDF>PDFR>PKA signaling pathway influences the core molecular clock by promoting the accumulation or stability of TIM.

Reduced morning anticipation in pdfr mutants coupled to an absence of significant core clock effects in LD suggested that PDF morning function is via effects on neuronal output [35]. We hypothesized that these effects may be mediated by direct effects on neuronal activity. Our previous work had identified the DN1p as functional targets of PDF on morning anticipation using rescue of pdfr mutants [41]. To determine if this reflects a direct interaction, we selectively labeled the DN1p using U-GFP in combination with Clk4.1-G4 and the LNv using anti-PDF and found that the arbors of each extensively co-mingle (Figure 8A). To test if DN1p neurons are direct targets of PDF neurons, we employed GFP reconstitution across synaptic partners (GRASP) [57]. Here we expressed one fragment of GFP (GFP11) on the extracellular surface of the LNv neurons using pdf-LexA and its complementary fragment (GFP1-10) on the extracellular surface of the DN1p neurons using Clk4.1-G4. We observe robust fluorescence in the dorsal terminals consistent with extensive physical contacts (Figures 8B, 8C, and S4). Co-labeling of sLNv-DN1p GRASP with PDF finds that PDF signal is in close proximity to GRASP signals suggesting that sites of physical contact are potential release sites for PDF (Figure 8B and 8C).

To examine PDF signaling mechanisms that control neuronal output, we first performed live imaging on the DN1p neurons on explanted brains. Using the Clk4.1-G4 driver in combination with the FRET sensor U-Epac1(50A) [58], we measured the variation of [cAMP] following focal PDF application specifically in the DN1p neurons. Prior studies had used bath application of PDF to examine changes in cAMP and thus, the observed effects could be due to indirect activation [47]. Following focal PDF application to the DN1p neurons (Figure S3A), we observed a decrease in the ratio YFP/CFP indicating an increase of [cAMP] (Figure S3B–S3D). Thus, as suggested by prior studies [47], we confirm that PDF increases cAMP levels in the DN1p.

To resolve whether PDF acutely controls DN1p neuronal activity, we developed patch clamp electrophysiology of the DN1p subset of pacemaker neurons and assayed the response of these cells to focal PDF application. We performed cell-attached recordings in combination with live calcium imaging on the DN1ps by simultaneously recording firing frequency and [Ca²⁺]_i using Clk4.1-G4 driving expression of the GCaMP6f calcium indicator [59]. Focal PDF application acutely stimulates the DN1ps by increasing the instant firing frequency of the neurons (Figure 9A and B). This increase in neuronal activity is directly correlated with an increase in [Ca²⁺]_i as measured by the GCaMP6f calcium indicator (Figure 9C).

Using whole-cell current clamp recordings, we found that PDF both acutely depolarizes and increases action potential firing rates (Figure 10A). This excitatory effect of PDF is dependent on its receptor as we could not detect any effect of PDF on the membrane potential or firing frequency in cells lacking PDFR (Figure 10A). The PDF-evoked depolarization was present after blocking action potential firing, and thus most synaptic transmission, with the voltage gated sodium channel blocker TTX indicating that PDF acts on the DN1ps directly (Figure 10B). Surprisingly the PKA inhibitor H89 did not block these effects indicating that PDF activates a PKA-independent pathway to acutely activate neurons (Figure 10C). To independently confirm the dispensability of PKA signaling we recorded from DN1p neurons expressing the dominant-negative PKA-R1 (Figure 10D and 10E). PDF application still depolarizes and increases in firing frequency further supporting the hypothesis that PDF acts independently of PKA activation to regulate membrane excitability.

We next examined the potential role of cAMP as the intracellular component mediating the acute PDF effect on membrane activity. First, we demonstrated that the adenylate cyclase inhibitor MANT-GTPγS blocks the PDF induced excitation (Figure 11A–11C). Conversely, forskolin (an adenylate cyclase activator) and direct dialysis of cAMP into the cell induces a depolarization similar to the PDF evoked response (Figure 11D and 11E). Finally, the cAMP induced depolarization and activation was present in the neurons expressing PKA-R1dn (Figure 11E). Taken together these data indicate that cAMP, rather than other upstream signaling components is responsible for the PDF effects on excitability.

In voltage clamp mode, PDF application acutely induces an inward current at negative potentials and a positive shift in the reversal potential (Figure 12A–12C). This inward current is TTX-insensitive (Figure 12C) and is attenuated after reduction of extracellular sodium (Figure 12F). Furthermore, focal application of the adenylate cyclase activator forskolin or direct intracellular dialysis of cAMP induce an inward current like PDF (Figure 12D and 12E). The properties of the observed current—neuropeptide induction, TTX resistance as well as PKA independence—are consistent with a cyclic-nucleotide-gated channel (CNG) [60]. Thus, using this novel patch clamp analysis and focal application of PDF, we demonstrate that PDF acts as an excitatory neurotransmitter that acutely increases firing rate and calcium, likely in a PKA-independent manner. The rapidity of PDF effects on excitablility argues strongly that they are direct and not via clock resetting.

Fly lines carrying U-PKA-mC* and U-PKA-R1dn (also known as BDK33) were a generous gift from Daniel Kalderon [52]. UAS-CD4::spGFP1-10 and LexAop-CD4:spGFP11 flies were the gift of Kristen Scott [71]. The latter transgene was recombined with pdf-LexA (a gift of Michael Rosbash [72]). Lines carrying combinations of these and other transgenes or mutants were constructed using standard genetic crosses. Tim was genotyped for the s/ls alternative start site polymorphism using previously described primers (Peschel 2004). TIM staining in the per¹;pdf-G4;U-per16 rescue context was completed three times: in two trials the tim genotypes were s/ls, and in one trial the control was ls/ls, while the pdfPER flies were s/ls. The staining results were comparable between these conditions and were combined. Strains for TIM staining with PKA-R1dn expression (Figure 6) were s/ls.

Fly behavior was recorded using the Drosophila Activity Monitoring system (Trikinetics) and analyzed using ClockLab and the Counting Macro as described [73]. Briefly, male flies were fed on 5% sucrose-agar medium in 5LD7DD conditions at 25C. LD eductions were obtained using averaged data in 30-minute bins across days 2–5 of the behavior run. DD period and rhythmicity data were calculated in ClockLab with period measurements taken only from flies in which the Power-Significance (P-S) ≥10.

Morning anticipation was calculated using a variant of the method described in [32]. Activity from each of four days of LD behavior recorded for each individual fly were analyzed such that the morning index (MI) = ((total activity 3 h prior to lights-on)/(total activity 6 h prior to lights-on)) − (0.5). 0.5 was subtracted so flat activity over the six hours analyzed is equal to 0. In cases where no activity counts occurred in the 6 hours before lights-on, resulting in an undefined 0/0, the ratio was set to 0.5, indicating no change in activity over that time period.

The timing of evening activity onset was calculated as previously described [35] with the onset time defined as the first time point in the four 30-min bin sliding window with the largest increase in activity prior to lights-off.

Genotypes were compared by Student's two-tailed t-test.

Flies to be stained were entrained for five to seven 12-h light, 12-h dark (LD) cycles at 25°C and either dissected and fixed at the indicated timepoint for LD staining (Figure 6) or transferred to constant darkness and dissected and fixed for DD1 staining (Figures 4 and 5). Brains were dissected in PBS (pH 7.5) and fixed in 3.7% formaldehyde in PBS for 1 h shaking at room temperature. Brains were then washed 3× in PBS and primary antibody solution was added. Guinea pig anti-TIM (1∶2,000), rabbit anti-PER (1∶16,000), and mouse anti-PDF (1∶500) (Developmental Studies Hybridoma Bank) were incubated overnight shaking at 4°C in a solution of PBS, 10% goat normal serum (GNS), and 0.3% Triton X-100. For stains involving rabbit anti-PDP (1∶200) brains were dissected and fixed as above, except after fixation and 3× washes with PBS, brains were subject to a 1-h permeablization in PBS +1% Triton X-100 and primary antibody solution was incubated for 3 days in PBS with 0.3% Triton X-100 and 10% GNS. After the primary incubation, for all stains, brains were washed 3× in PBS +0.3% Triton X-100 and secondary antibodies (for PDF, PER, TIM staining: anti-mouse Alexa647, anti-guinea-pig Alexa 488, anti-rabbit Alexa 594; for PDF, PDP1ε staining: anti-mouse Alexa 594, anti-rabbit Alexa 488) (all dyes from Molecular Probes - Invitrogen) were each added at 1∶500.

Brain images were taken on a Nikon E800 laser-scanning confocal microscope using a 60× A 1.40 N.A. objective with laser, filter, and gain settings remaining constant within each experiment. Channels were scanned sequentially. Confocal Z-stacks were analyzed in NIH ImageJ software. Intensity measurements were taken from single confocal sections at approximately the middle of each cell. Nearby areas of similar area to the cells being measured were selected for each cell group in each hemisphere as a measurement of background staining. The background measurement for each cell group in each hemisphere was subtracted from the intensity measurement for each cell in that group. Background-subtracted values were then averaged across all brains in that experiment. Image measurements were normalized prior to combining data from independent experiments. For Figures 4 and 5, each experiment was individually normalized such that pdfPER Rescue at CT6 = 1. For Figure 6, data were normalized to Control at ZT6 = 1. For Figure 7, data were normalized to per¹ at CT24 = 1. Data from independent experiments were combined post-normalization to obtain the final graphs. Images from Figures 4, 5, and 6 are displayed using the inverted 5 Ramps lookup table within ImageJ for ease of viewing images with low signal. Staining data were statistically analyzed by one-way ANOVA and Tukey's pairwise comparisons.

GRASP signal and mouse anti-PDF stained brains were fixed, mounted, and imaged as above, except using anti-mouse Alexa 594 to label PDF. Both 40× and 60× objective images were collected in 1 micron steps through the region containing staining, or through the entire dorsal brain for non-labeled parental control lines.

Cells were processed as described previously [74]. Before FACS cell sorting cells were filtered using 100 micron filter. Propidium iodide (Sigma, 130 ng/ul) was added to distinguish between dead and alive cells. Cells were sorted on Aria II FACS Cell Sorter (BD Biosciences) into an extraction buffer from the PicoPure RNA extraction kit (Arcturus). Transcripts were obtained from 40 to 45 brains (yielding 300–500 DN1p neurons) per time-point. Subsequently, the cells were lysed and stored at −80°C until RNA extraction as described previously [74].

Cells were processed as described previously [75]. cDNA from two independent replicates per genotype were analyzed per time-point on a BioRad CFX384 real-time PCR system. mRNA was quantified as described previously [74]. One-way ANOVA was used to determine statistically significant differences between time-points within each genotype (p<0.05). The following primers were used to examine pka expression: pka-R1, F primer, 5′-ACTTTGGCGAGATTGCTCTG-3′; R primer, 5′-CGGACAACGATACGAAACTG-3′; pka-R2, F primer, 5′-CTACGAACGCATGAATCTGG-3′; R primer, 5′-GCCGAAGTACTGTCCCTTGC-3′; pka-C1, F primer, 5′-ATCGCTGGCATCGTAGTCG-3′; R primer, 5′-AAGGCGCTTGGTTAAGACG-3′.

Brains from male adults Drosophila (7–14 days old) were removed from their heads in ice-cold recording solution. After removing the connective tissue, air sacs, and trachea with fine forceps, the brains were transferred to a recording chamber and were held ventral side down by a harp slice grid (ALA scientific). No enzymatic treatment was used to avoid altering ion channels function on the cell surface. Brains were allowed to rest in continuously flowing oxygenated saline (95% oxygen and 5% carbon dioxide) for at least 10 min and no more than 2 h before recording. Perfusion with oxygenated saline was continued throughout the recording period. Whole brain electrophysiology and imaging experiments were performed on an Ultima two-photon laser scanning microscope (Prairie Technologies) equipped with galvanometers driving a Coherent Chameleon laser. Fluorescence was detected with photomultiplier tube. Images were acquired with an upright Zeiss Axiovert microscope with a 40×0.9 numerical aperture water immersion objective at 512×512 pixel resolution and 1-µm steps. Current-clamp recordings were performed with pipettes (10–14 MΩ) filled with internal solution. To visualize the recorded cell, Alexa Fluor 594 biocytin (10 µM) was added into the intracellular solution. Recordings were made using Axopatch 200B patch-clamp amplifier, Digidata 1320 A, and pCLAMP software (Axon Instruments). The extracellular recording solution contains in mM: 101 NaCl, 1 CaCl₂, 4 MgCl₂, 3 KCl, 5 glucose, 1.25 NaH₂PO₄, and 20.7 NaHCO₃ (pH 7.2, 250 mOsm). The internal solution contains in mM: 102 K-gluconate, 0.085 CaCl₂ 1.7, MgCl₂, 17 NaCl, 0.94 EGTA, 8.5 HEPES, 4 Mg-ATP, 0.3 Tris-GTP, and 14 phosphocreatine (di-tris salt) (pH 7.2, 235 Osm). For simultaneous cell attached and live calcium-imaging recordings, the Drosophila DN1ps neurons were visualized with GCaMP6f indicator [59]. The x-y images of GCaMP6f fluorescence were acquired at 10–20 Hz. GCaMP6f fluorescence was excited at 840 nm and was captured at wavelengths between 490 and 540 nM using a bandpass filter.

Changes in intracellular cAMP concentration were imaged using the Epac1-cAMPs indicator. Whole brain imaging experiments were performed using hemolymph-like HL3 saline [76] (in mM: NaCl 70, KCl 5, CaCl₂ 1.5, MgCl₂ 20, NaHCO₃ 10, D-trehalose dihydrate 5, sucrose 115, Hepes 5, pH adjusted at 7.1 with NaOH 1 M). After dissection, whole brains were placed with HL3 solution in the experimental chamber (POC-R perfusion chamber, Zeiss) and placed on the stage of an Axiovert 200 M inverted microscope attached to a Zeiss 510 Meta/ConfoCor3 Laser Scanning unit (Zeiss) available through the Northwestern University Biological Imaging Facility. The x-y confocal images of Epac1-camps fluorescence were acquired at 2–4 Hz using a Zeiss planApochromat 20×0.8 N.A. objective. Epac1-camps fluorescence were excited at 454 nm by a 200 mW argon ion laser and were captured at wavelengths between 470 and 500 nM for CFP and between 510 and 550 nM for YFP using a bandpass filter. The pinhole was set to provide a confocal optical slice of 10 µm. Epac1-camps fluorescence intensity was normalized to the average fluorescence intensity in the images captured before neurotransmitter application and the ratio YFP/CFP was calculated.

PDF (50 µM, dissolved in recording solution, GenScript) or forskolin (Sigma) was applied focally for 10 s to the recorded cells via pressure ejection (0.5–1 psi) from a glass pipette (5–10 µM) placed in the vicinity of the cell. TTX (Tocris) and/or H89 (Sigma) were bath applied by exchanging the recording solution. cAMP (10 µM, Sigma) and MANT-GTPγS (500 nM, Sigma) were diluted into the intracellular solution.