Whole brain fluorescence imaging in Drosophila reveals spreading depolarization and its initiation, propagation, and resilience dynamics Authors: Allison L. Harris, Grace Crowe, Ella Nelson, Mohammad Al Mokbil, Wolfgang Stein Published: 2026 ABSTRACT SummarySpreading depolarization (SD) is a propagating wave of neural depression associated with migraine, stroke, and brain injury. We develop high-throughput, whole-brain fluorescence imaging assays in Drosophila melanogaster ideal for observing and mechanistically understanding SD. We reveal key characteristics of SD, including that it originates at multiple, variable sites across and within individuals, and that it spreads throughout the nervous systems of adults and larvae. This reveals that SD is a fundamental phenomenon not restricted to specific brain regions, neuronal populations, or developmental stages. Furthermore, we demonstrate that single SD events cause a depolarization-independent transient refractory period with elevated SD threshold, suggesting an alteration of homeostatic processes following SD. This work and others, combined with Drosophila’s extensive genetic toolkit, establishes the fly as a high-throughput model for understanding conserved cellular and circuit-level mechanisms relevant to human neurological disorders involving SD. Introduction Spreading depolarization (SD) is a dramatic neurophysiological phenomenon that occurs during disorders such as migraine, stroke, cerebral ischemia, and brain injury.1 During SD, a wave of neuronal hyperexcitability and subsequent inactivation propagates through large areas of the brain (often cortex) at a speed of a few millimeters per minute.2 At a cellular level, SD is characterized by rapid neuronal firing followed by a loss of spiking due to depolarization block. It typically manifests with negative symptoms but has recently been hypothesized to have a protective effect against more severe disorders.3 Studies in mammalian brains, alongside computational models, have contributed to characterizing the dynamics and physiological mechanisms of SD spread.1,2,4,5,6,7,8 These include dramatic changes in intra- and extracellular ion concentrations, such as elevated extracellular potassium and intracellular sodium. The elevated potassium concentration, in particular, propagates at a characteristic slow velocity through the extracellular medium via diffusion, helping to mediate the spread of the hyperactivity and subsequent depolarization block.9 The consequences of this disrupted ion homeostasis include cell swelling, altered blood flow, inflammation, and in the case of migraine, headaches (for review, see Ayata and Lauritzen1). Despite decades of study on cellular-level dynamics and SD spread,1,2,4,5,6,7,8 the mechanisms that initiate, modulate, and affect SD susceptibility remain elusive. Current mammalian systems for studying SD struggle to provide large-scale and high-throughput recordings from intact brains that can identify the underlying mechanisms that alter SD susceptibility. Most commonly, studies of SD have imaged small brain regions with in-vivo experiments (e.g., Leo et al.10 and Chung et al.11 but see Joshi and Andrew12) or elicited SD in in-vitro brain slices using artificial stimuli (e.g., Harriott et al.,13 Berger et al.,14 and Anderson and Andrew15), such as high focal potassium application or electrical or mechanical brain stimulation. In contrast, insect nervous systems offer a more tractable system in which to study SD in fully intact animals. Recently, chill coma, a reversible state of paralysis that occurs in many insects when exposed to low temperatures, has been proposed to be associated with SD.16,17,18,19,20,21 This environmentally occurring, experimentally controllable stimulus enables the study of SD susceptibility through a process that is part of many animals’ natural experience. Initial work linking SD with chill coma has shown a dramatic extracellular potassium accumulation as insects are cooled to a few degrees Celsius,19,20,21 accompanied by neuronal hyperactivity. Additionally, measurements of the DC field potential at two separate sites show a sudden drop16,19,20 that occurs with some time delay and has been interpreted as a spreading wave. The dynamics of these potential changes have been shown to be temperature and acclimation dependent.22 However, no direct evidence of large-scale spread has yet been observed or measured for cold-induced SD, and it remains unclear whether the changes in field potential are correlated or occur independently. These studies suggest a link between SD and chill coma, and show that locally, the ion concentration and electrophysiological changes observed in insects are similar to those in mammals (for comparison of insect and mammalian SD, see Spong et al.23). However, electrophysiological and ion concentration measurements are technically challenging, low-throughput, and provide only a limited picture of the physiological dynamics of SD initiation, spread, and susceptibility. We introduce a method for observing and quantifying SD in Drosophila melanogaster using whole brain fluorescence imaging in which both the spatial and temporal dynamics of SD can be measured. We present evidence that SD can be elicited using rapid cooling and extracellular potassium application, and show that SD spreads across large areas of the nervous system. Our technique allows us to capitalize on the many advantages of the Drosophila study system, including targeted gene expression and high-throughput imaging techniques.24,25,26,27,28,29 By leveraging genetically encoded pan-neuronal calcium and voltage sensors, we show that SD is not restricted to particular brain areas but is initiated at different and multiple sites that vary across and within animals. Furthermore, we reveal that SD exists not only in adult Drosophila but also in all three larval stages, underscoring its ubiquity across widely varying neuronal connectivities. Our assays for SD measurements in larvae provide a simple and high-throughput technique for observing SD in up to 16 animals simultaneously, enabling the rapid testing of SD initiation stimuli. Through repeated exposure to cold in our high-throughput assay, we demonstrate that a singular SD event confers resilience for up to 45 min against future SD events, during which time the threshold to elicit SD is elevated. This refractory period is reminiscent of a similar feature observed in mammals,30,31 but is shown here in fully intact animals under naturally experienced conditions. Overall, our findings establish that SD is a pervasive and large-scale phenomenon in both adult and larval Drosophila, indicating that its initiation and propagation are not restricted to specific regions, neuronal populations, or developmental stages. We further show that a single SD event lowers the susceptibility to future SD events, conferring resilience for up to 45 min. Together, these results and techniques position the fly as a powerful model for the study of conserved cellular and circuit-level mechanisms governing SD initiation, propagation, and adaptive regulation. Results Cooling elicits SD in adult fly brains We tested the hypothesis that SD propagates across adult fly brains by measuring pan-neuronal calcium fluorescence (GCaMP6m) during rapid cooling. Adult flies were placed on a custom-built cooling platform (Figure 1A) with the dorso-posterior head cuticle and trachea removed to expose the brain (Figure 1B). Large-scale neuronal activity was measured using genetically encoded fluorescent reporters (e.g., GCaMP6m; Figures 1C and 1D). At room temperature, spontaneous activity was observed and was particularly prevalent in the mushroom body. For example, Figure 1D shows the spontaneous fluorescence changes at room temperature at six different locations across the mushroom body. The fluorescence traces of the left mushroom body showed correlated but propagating activity that was independent of that observed in the right mushroom body. Overall, the changes in detected spontaneous GCaMP6m fluorescence values ranged between 3% and 50% of the background fluorescence of the location where they were measured (Video S1). Brains in which no spontaneous activity was detected were considered damaged during dissection and discarded.Figure 1Whole brain fluorescence imaging of adult flies(A) Schematic of experimental setup to rapidly cool from room temperature to near 0°C while measuring GCaMP6m or Arclight fluorescence in adult flies.(B) Brightfield image of an adult fly head with the posterior head cuticle removed to expose the brain.(C) Same fly as in (B), expressing pan-neuronal GCaMP6m under fluorescent light (470 nm illumination, 525 nm detection). The mushroom body (MB), ellipsoid body (EB), and fan-shaped body (FB) are clearly visible.(D) Left: regions of interest (ROIs) used to measure spontaneous activity in the brain at room temperature. Right: fluorescence (arbitrary units) measured in each ROI at room temperature. Traces are separated vertically for clarity. See also Video S1. Whole brain fluorescence imaging of adult flies (A) Schematic of experimental setup to rapidly cool from room temperature to near 0°C while measuring GCaMP6m or Arclight fluorescence in adult flies. (B) Brightfield image of an adult fly head with the posterior head cuticle removed to expose the brain. (C) Same fly as in (B), expressing pan-neuronal GCaMP6m under fluorescent light (470 nm illumination, 525 nm detection). The mushroom body (MB), ellipsoid body (EB), and fan-shaped body (FB) are clearly visible. (D) Left: regions of interest (ROIs) used to measure spontaneous activity in the brain at room temperature. Right: fluorescence (arbitrary units) measured in each ROI at room temperature. Traces are separated vertically for clarity. See also Video S1. Video S1. False-color video of spontaneous activity in the adult fly brain, related to Figure 1GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top. Video S1. False-color video of spontaneous activity in the adult fly brain, related to Figure 1 GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top. Figure 2A shows a representative example of calcium fluorescence during cooling and subsequent heating. Fluorescence changes could be classified into 4 distinct regimes that correlate with expected physiological changes during SD (Figure 2A and 2B, and Video S2). During the initial temperature drop (18°C–13°C; regime 1), there was a slow and monotonic rise in fluorescence, indicating increased calcium accumulation across the brain. In regime 2, fluorescence rapidly and dramatically increased, beginning on average at 6.7 ± 0.6°C, peaking at 5.6 ± 0.6°C, and taking 44.2 ± 5.6 s (N = 15). This increase indicated a dramatic change in the state of the neurons, as would be expected for neuronal hyperactivity and depolarization during SD. During regime 3, fluorescence gradually decreased even when cooling continued to an average coldest value of 3.7 ± 0.6°C (N = 15). Finally, during heating back to 18°C (regime 4), fluorescence rapidly decreased to below its initial level, suggesting that neuronal activity recovered from its depolarized state. Figure 2C shows that the mean normalized fluorescence at each time point (i–v) was significantly different from the others, demonstrating that the shape of the fluorescence curve in Figure 2B was common across animals. Notably, point iii showed the largest change, averaging 3.0 ± 0.4 times the fluorescence at 18°C. The most dramatic and rapid changes were observed during regime 2, with an average 1.3 ± 0.2-fold increase from beginning to end of the regime (Figure 2D) and a slope more than 50 times greater than regime 1 (Figure 2E).Figure 2Fluorescence changes during rapid cooling and rewarming can be classified into 4 regimes(A) Representative recording of GCaMP6m fluorescence during cooling from room temperature to near 0°C and then rewarming back to room temperature. (i) End of regime 1, 13°C. (ii) Start of regime 2, 5.9°C. (iii) End of regime 2 = start of regime 3, 2°C. (iv) End of regime 3 = start of regime 4, 4.6°C. (v) End of regime 4, 18°C.(B) Representative whole brain GCaMP6m fluorescence over time showing the 4 regimes (R1–R4) and their starting and ending points (i–v). The temperature stimulus used is shown in yellow.(C) Quantification of normalized GCaMP6m fluorescence at points (i)–(v). For each animal, the mean fluorescence was normalized to the mean fluorescence at 18°C at the start of the cooling ramp. The normalized fluorescence at each point was significantly different from the other points (Friedman chi-square test, N = 15, χ2 = 52.000, dF = 4, p < 0.001, Student-Newman-Keuls post hoc test at p < 0.05 [∗]). Data are represented as mean ± SEM.(D) Fold change in GCaMP6m fluorescence during each regime is significantly different from the other regimes (Friedman chi-square test, N = 15, χ2 = 39.000, dF = 3, p < 0.001, Student-Newman-Keuls posthoc test at p < 0.05 [∗]). Data are represented as mean ± SEM.(E) Slope of GCaMP6m fluorescence over time during each regime was significantly different from the others (Friedman chi-square test, N = 15, χ2 = 39.000, dF = 3, p < 0.001, Student-Newman-Keuls post hoc test at p < 0.05 [∗]). Data are represented as mean ± SEM.(F) Same as (A) but for Arclight fluorescence. Temperatures are indicated in the figure.(G) Same as (B) but for Arclight fluorescence.(H) Original recordings of GCaMP6m fluorescence during regime 2 show that the fluorescence wave spreads through the mushroom body. Times are shown from the start of the spreading wave (t = 0). The right image shows 4 regions of interest (ROIs) used to measure the fluorescence in (I); see also Video S2.(I) Normalized fluorescence for the ROIs shown in (H). Arrows indicate the start of regime 2 and the start of SD for each ROI. They are clearly separated in time, indicating that the fluorescence wave shows a spatial spread. Fluorescence changes during rapid cooling and rewarming can be classified into 4 regimes (A) Representative recording of GCaMP6m fluorescence during cooling from room temperature to near 0°C and then rewarming back to room temperature. (i) End of regime 1, 13°C. (ii) Start of regime 2, 5.9°C. (iii) End of regime 2 = start of regime 3, 2°C. (iv) End of regime 3 = start of regime 4, 4.6°C. (v) End of regime 4, 18°C. (B) Representative whole brain GCaMP6m fluorescence over time showing the 4 regimes (R1–R4) and their starting and ending points (i–v). The temperature stimulus used is shown in yellow. (C) Quantification of normalized GCaMP6m fluorescence at points (i)–(v). For each animal, the mean fluorescence was normalized to the mean fluorescence at 18°C at the start of the cooling ramp. The normalized fluorescence at each point was significantly different from the other points (Friedman chi-square test, N = 15, χ2 = 52.000, dF = 4, p < 0.001, Student-Newman-Keuls post hoc test at p < 0.05 [∗]). Data are represented as mean ± SEM. (D) Fold change in GCaMP6m fluorescence during each regime is significantly different from the other regimes (Friedman chi-square test, N = 15, χ2 = 39.000, dF = 3, p < 0.001, Student-Newman-Keuls posthoc test at p < 0.05 [∗]). Data are represented as mean ± SEM. (E) Slope of GCaMP6m fluorescence over time during each regime was significantly different from the others (Friedman chi-square test, N = 15, χ2 = 39.000, dF = 3, p < 0.001, Student-Newman-Keuls post hoc test at p < 0.05 [∗]). Data are represented as mean ± SEM. (F) Same as (A) but for Arclight fluorescence. Temperatures are indicated in the figure. (G) Same as (B) but for Arclight fluorescence. (H) Original recordings of GCaMP6m fluorescence during regime 2 show that the fluorescence wave spreads through the mushroom body. Times are shown from the start of the spreading wave (t = 0). The right image shows 4 regions of interest (ROIs) used to measure the fluorescence in (I); see also Video S2. (I) Normalized fluorescence for the ROIs shown in (H). Arrows indicate the start of regime 2 and the start of SD for each ROI. They are clearly separated in time, indicating that the fluorescence wave shows a spatial spread. Video S2. False-color video of cooling-induced spreading wave of fluorescence in the adult fly brain, related to Figure 2GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top. Video S2. False-color video of cooling-induced spreading wave of fluorescence in the adult fly brain, related to Figure 2 GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top. The fluorescence signal thus strikingly resembled the expected shape of membrane polarization that neurons experience during SD, i.e., a slow depolarization that leads to increased activity and is followed by a sustained depolarization with depressed activity, and ultimately a sudden recovery.2 To further support that the observed fluorescence changes were due to neuronal depolarization, we first confirmed that they were not due to a temperature response of the GCaMP6m (Figure S1). We then repeated the cooling experiments using a pan-neuronal expression of the voltage sensor Arclight (see STAR Methods). Arclight fluorescence decreases with membrane depolarization.32 Accordingly, the Arclight fluorescence response was inverted from that of GCaMP6m and exhibited the same four regimes. Figures 2F and 2G show a representative example of the observed Arclight fluorescence during the cooling and rewarming protocol at the same milestones as Figure 2A. During regime 1, there was a slow increase in fluorescence, suggesting an initial slight hyperpolarization that was followed by a subsequent depolarization after regime 1 ended. During regime 2, the fluorescence sharply dropped to its smallest value, indicative of large-scale neuronal depolarization. This drop occurred on average at 4.5 ± 0.4°C (N = 7) and reached its minimum at 2.5 ± 0.3°C (N = 7). Fluorescence during regime 2 dropped by 35.0 ± 5.3%. Fluorescence slowly increased during regime 3 as the temperature began to warm. Finally, during heating back to 18°C (regime 4), fluorescence rapidly increased, suggesting an end to neuronal depolarization. Together, our Arclight and GCaMP6m data suggest that fly neurons undergo physiological changes resembling those observed during SD, with a sudden, long-lasting, and reversible depolarization that is triggered by cold temperatures. Lastly, if the observed fluorescence changes are indicative of SD, they should propagate slowly across the brain. Further inspection of regime 2 revealed that the rapid rise in GCaMP6m fluorescence did not occur simultaneously in all areas of the brain (Figures 2H and 2I). There was a clear initiation point from where fluorescence started before it spread slowly to neighboring areas, leading to the increased mean brightness observed in regime 2. Figure 2H shows an example of a wave initiated in the right α lobe of the mushroom body that spread medially toward the β lobe, then jumped to the β lobe of the left mushroom body before ultimately reaching the left α lobe. Our measurements also revealed that there was no single, fixed initiation point, but instead waves could be initiated in multiple locations. In Figure 3A, two independent and non-contiguous waves were initiated with the first wave in the right mushroom body, and the second shortly after in the left. Plotting the normalized fluorescence for a set of regions of interest (ROIs; Figure 3B) placed along the brain areas affected by the spreading fluorescence wave confirmed the two initiation sites. On average, we identified 3.0 ± 0.7 initiation points per brain, with some initiation points leading to multiple spreads in different directions. Spreading waves initiated by rapid cooling traveled an average distance of 63.7 ± 9.1 μm at a velocity of 0.77 ± 0.1 mm/min (N = 40 spreads, 8 animals), which is slower than that of previous two-point measurements.19,20,33 The difference may be partly due to the limitations of a two-point measurement, which cannot know that both recorded events came from the same SD event or that the SD traveled directly from one measurement site to the other. For example, if SD was initiated between the two measurement sites, the recorded time delay would be artificially low, thereby increasing the presumed velocity. Additionally, because the propagation of SD relies in part on diffusion, colder temperatures are likely to lead to a slower velocity. In fact, our data validate extrapolated estimates of previous studies.22 Our measurements are better able to measure spread velocity but likely underrepresent the number of initiation points because not all brain areas were monitored (e.g., the optic neuropils were outside of our field of view) and only a single focal plane was measured.Figure 3Multiple waves of spreading fluorescence were observed that started in different locations when the brain was cooled(A) Representative recordings of GCaMP6m fluorescence showing spreading waves were initiated in two different locations. (i) Before spreading waves were initiated. 21 ROIs (blue squares) were drawn along the path that the spreading waves traveled. (ii) Begin of spread in the right mushroom body. (iii) High fluorescence spreads medially in the right mushroom body. (iv) A second wave of high fluorescence begins in the left mushroom body while the first wave continues spreading medially in the right mushroom body. (v) High fluorescence in the left mushroom body spreads medially and laterally.(B) Normalized fluorescence of each ROI defined in (Ai) over time. The white line traces the maximum fluorescence of each ROI. The two wave initiation points are indicated by the arrows.(C) Original recording of GCaMP6m fluorescence during SD in an adult fly. ROIs 1–7 are shown along the path of SD spread. A sharp microelectrode was impaled at ROI 6 and the transperineurial potential (TPP) was measured. Left: normalized GCaMP fluorescence (gray lines) over time for ROIs 1–7. The trace for ROI 6 is highlighted in orange. The TPP measured from the sharp microelectrode located at ROI 6 coincided with the occurrence of the SD wavefront as indicated by the change in fluorescence at ROI 6. For display purposes, the TPP was inverted. Ant, anterior; post, posterior. Multiple waves of spreading fluorescence were observed that started in different locations when the brain was cooled (A) Representative recordings of GCaMP6m fluorescence showing spreading waves were initiated in two different locations. (i) Before spreading waves were initiated. 21 ROIs (blue squares) were drawn along the path that the spreading waves traveled. (ii) Begin of spread in the right mushroom body. (iii) High fluorescence spreads medially in the right mushroom body. (iv) A second wave of high fluorescence begins in the left mushroom body while the first wave continues spreading medially in the right mushroom body. (v) High fluorescence in the left mushroom body spreads medially and laterally. (B) Normalized fluorescence of each ROI defined in (Ai) over time. The white line traces the maximum fluorescence of each ROI. The two wave initiation points are indicated by the arrows. (C) Original recording of GCaMP6m fluorescence during SD in an adult fly. ROIs 1–7 are shown along the path of SD spread. A sharp microelectrode was impaled at ROI 6 and the transperineurial potential (TPP) was measured. Left: normalized GCaMP fluorescence (gray lines) over time for ROIs 1–7. The trace for ROI 6 is highlighted in orange. The TPP measured from the sharp microelectrode located at ROI 6 coincided with the occurrence of the SD wavefront as indicated by the change in fluorescence at ROI 6. For display purposes, the TPP was inverted. Ant, anterior; post, posterior. Another hallmark of SD is its association with a large increase in extracellular potassium that is coincident with neuronal hyperactivity and by itself causes further neuronal depolarization. To determine whether the observed fluorescence wave was associated with an extracellular potassium increase, we measured the transperineurial potential (TPP), which was previously shown to reflect such changes in extracellular potassium.16,20 Figure 3C shows that during SD elicited by rapid cooling, dramatic changes in the TPP coincided with the arrival of the fluorescence wavefront at the recording site. Thus, rapid changes in extracellular ion concentrations occurred at times when the neurons underwent strong cooling-induced depolarization, further supporting the hypothesis that cold temperatures elicit SD. It is well established that increasing the extracellular potassium concentration through application of potassium chloride (KCl) can elicit SD in mammals.34 We thus performed calcium imaging of adult fly brains while we spiked the saline bath that surrounded the fly brain with a drop (4 μL) of 1M KCl solution (increasing the potassium concentration from 5 mM to 12.9 mM). In all cases (N = 9), spreading fluorescence waves were observed. Figure 4A shows a wave initiated in the right mushroom body and spreading laterally, followed by a second wave in the left mushroom body, also spreading laterally. Finally, a delayed third wave traveled toward the calyx after the fluorescence in the right hemisphere began to diminish (Video S3). The mean fluorescence over time plot of the whole brain (Figure 4B) resembled those of the cooling experiments but showed additional features indicative of the three waves. When the wave began in the right mushroom body, a rapid rise of fluorescence occurred (peak at ii, orange). This was followed by the wave in the left mushroom body, which caused a subsequent steep rise in fluorescence (peak at iii, green). Lastly, the delayed wave near the calyx caused a smaller peak after the mean fluorescence had already reached its maximum (iv, red). Figure 4C shows the timing of the wave spreading across the brain, starting with the right mushroom body.Figure 4Increased potassium concentration facilitates the initiation of a spreading wave of calcium fluorescence(A) Representative recordings of GCaMP6m fluorescence at room temperature while the brain was bathed in saline and then spiked with increased potassium. (i) Before KCl application; (ii) shortly after KCl application; (iii) at peak fluorescence; (iv) after fluorescence started to diminish (Video S3).(B) Whole brain mean fluorescence over time for the images in (A).(C) Normalized mean fluorescence at the ROIs shown in (Aiii). The minimum to maximum fluorescence was normalized to 0 to 1.(D) Whole brain mean fluorescence changes over time in an animal that exhibited multiple waves of spreading fluorescence.(E) Normalized mean fluorescence at the start (i) and end (iii) of the spreading wave when maximum fluorescence occurred (iii). The change in fluorescence was highly significant (N = 9, p < 0.01 [∗], paired t test). For each animal, mean fluorescence was normalized to the mean fluorescence value at the start of the experiment. Data are represented as mean ± SEM.(F) Quantification of the temperature at which the spreading fluorescence wave was initiated in experiments where brains were cooled in standard physiological saline (sal) and in saline with enhanced (1.25×) KCl concentration. The initiation temperature in the KCl-enhanced saline was significantly higher than the temperature in standard saline (N = 7, p = 0.03 [∗], paired t test). Data are represented as mean ± SEM. Increased potassium concentration facilitates the initiation of a spreading wave of calcium fluorescence (A) Representative recordings of GCaMP6m fluorescence at room temperature while the brain was bathed in saline and then spiked with increased potassium. (i) Before KCl application; (ii) shortly after KCl application; (iii) at peak fluorescence; (iv) after fluorescence started to diminish (Video S3). (B) Whole brain mean fluorescence over time for the images in (A). (C) Normalized mean fluorescence at the ROIs shown in (Aiii). The minimum to maximum fluorescence was normalized to 0 to 1. (D) Whole brain mean fluorescence changes over time in an animal that exhibited multiple waves of spreading fluorescence. (E) Normalized mean fluorescence at the start (i) and end (iii) of the spreading wave when maximum fluorescence occurred (iii). The change in fluorescence was highly significant (N = 9, p < 0.01 [∗], paired t test). For each animal, mean fluorescence was normalized to the mean fluorescence value at the start of the experiment. Data are represented as mean ± SEM. (F) Quantification of the temperature at which the spreading fluorescence wave was initiated in experiments where brains were cooled in standard physiological saline (sal) and in saline with enhanced (1.25×) KCl concentration. The initiation temperature in the KCl-enhanced saline was significantly higher than the temperature in standard saline (N = 7, p = 0.03 [∗], paired t test). Data are represented as mean ± SEM. Video S3. Video of KCl-induced spreading wave of fluorescence in the adult fly brain, related to Figure 4GCaMP6m was expressed pan-neuronally. Lighter regions represent higher fluorescence. The video is shown at the original speed. Anterior is to the top. Video S3. Video of KCl-induced spreading wave of fluorescence in the adult fly brain, related to Figure 4 GCaMP6m was expressed pan-neuronally. Lighter regions represent higher fluorescence. The video is shown at the original speed. Anterior is to the top. In some experiments, multiple waves across the same brain areas were observed, even though potassium was only applied once (e.g., three independent waves in Figure 4D). This is consistent with prior measurements of K+-induced SD35 and provides additional confirmation that the observed phenomenon is SD. On average, the first wave was observed within 14.0 ± 4.4 s of the application with a maximum fluorescence significantly larger than before the wave started (before application: 1.0 ± 0.04, after: 2.3 ± 0.3, Figure 4E). We also tested the hypothesis that elevated extracellular potassium facilitates SD initiation, even at concentrations that are insufficient to elicit SD itself. We first cooled the brain in regular saline to elicit SD. After recovery to room temperature, we exchanged the saline with elevated potassium saline (25% higher, 6.25 mM) and after waiting 25 min, we repeated the cooling. We predicted that if potassium changes in the extracellular space are causally involved in eliciting the cooling-induced SD, then elevated potassium levels should cause SD to occur more easily, i.e., at a higher temperature. This was indeed the case (control saline 4.4 ± 0.2°C; elevated potassium: 5.5 ± 0.3°C, Figure 4F). Taken together, we have identified a slowly propagating wave of sustained neuronal depolarization that is induced by cooling or high potassium application, is modulated by extracellular potassium concentration, and is accompanied by changes in the TPP. These observations are reminiscent of the hallmarks of SD, and thus consistent with the hypothesis that adult fly brains can exhibit SD. Cooling elicits SD in larval fly brains Existing literature suggests that SD results from underlying fundamental mechanisms and properties shared between different circuits and brain structures (for reviews see Somjen2 and Robertson et al.36). We thus hypothesized that SD can also be induced in larval fruit flies and is based on the same mechanisms as those in the adult fly brain, despite differences in nervous system structure, connectivity, and neuron number. To test our hypothesis, up to 16 larvae were secured on a custom-built cooling platform (Figure 5A) either as intact animals (Figures 5B and 5C) or with the fillet dissection37 (Figures 5D and 5E) and imaged simultaneously. Large-scale neuronal activity was measured using genetically encoded fluorescent reporters (e.g., GCaMP6m; Figures 5C and 5F). We focused on the ventral nerve cord (VNC) to identify neuronal activity. As expected, and reported previously38,39 at room temperature, clearly visible slow waves of calcium fluorescence traveled posteriorly along the nerve cord (Figure 5F). These waves were coordinated and had high signal-to-noise ratio (Video S4), suggesting that they resulted from neuronal activity that drives larval crawling behaviors.Figure 5Whole brain fluorescence imaging of larval flies(A) Whole nervous system fluorescence imaging of larval flies. Schematic of high-throughput experimental setup to rapidly cool larvae from room temperature to near 0°C while measuring GCaMP6m or Arclight fluorescence in the nervous system. Up to 16 second instar larvae can be imaged simultaneously.(B) Brightfield image of 6 second instar larvae. The anterior ends of the larvae with visible mouthhooks are pinned at the center and the posterior ends are pinned at the outer edge of the image.(C) Same larvae as in (B), expressing pan-neuronal GCaMP6m under fluorescent light (470 nm illumination, 525 nm detection). The brain lobes and VNCs are clearly visible through the cuticle.(D) Schematic of larval fillet dissection used for high-resolution imaging or for pharmaceutical applications. The anterior end of the larva was pinned at the top of the schematic and the posterior end was pinned at the bottom of the schematic.(E) Fluorescence image of fillet-dissected larva, dorsal view. The fluorescent brain and VNC are clearly visible after removal of the dorsal cuticle and gut.(F) Spontaneous activity in the larval nervous system. Left: regions of interest (ROIs) used to measure spontaneous activity in theVNC. Right: fluorescence (arbitrary units) measured in each ROI at room temperature. Traces are separated vertically for clarity. See also Video S4. Whole brain fluorescence imaging of larval flies (A) Whole nervous system fluorescence imaging of larval flies. Schematic of high-throughput experimental setup to rapidly cool larvae from room temperature to near 0°C while measuring GCaMP6m or Arclight fluorescence in the nervous system. Up to 16 second instar larvae can be imaged simultaneously. (B) Brightfield image of 6 second instar larvae. The anterior ends of the larvae with visible mouthhooks are pinned at the center and the posterior ends are pinned at the outer edge of the image. (C) Same larvae as in (B), expressing pan-neuronal GCaMP6m under fluorescent light (470 nm illumination, 525 nm detection). The brain lobes and VNCs are clearly visible through the cuticle. (D) Schematic of larval fillet dissection used for high-resolution imaging or for pharmaceutical applications. The anterior end of the larva was pinned at the top of the schematic and the posterior end was pinned at the bottom of the schematic. (E) Fluorescence image of fillet-dissected larva, dorsal view. The fluorescent brain and VNC are clearly visible after removal of the dorsal cuticle and gut. (F) Spontaneous activity in the larval nervous system. Left: regions of interest (ROIs) used to measure spontaneous activity in theVNC. Right: fluorescence (arbitrary units) measured in each ROI at room temperature. Traces are separated vertically for clarity. See also Video S4. Video S4. False-color video of spontaneous activity in the third instar larval ventral nerve cord, related to Figure 5GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the bottom left. Video S4. False-color video of spontaneous activity in the third instar larval ventral nerve cord, related to Figure 5 GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the bottom left. We then rapidly cooled larval flies (first, second, and third instars) with pan-neuronal GCaMP6m expression. Figures 6A and 6B show an example of a third instar larva. As with adult brains, the changes in GCaMP6m fluorescence could be characterized into 4 regimes: (1) a slow global increase in fluorescence during initial cooling, (2) a rapid rise in fluorescence starting on average at 6.0 ± 0.3°C (N = 30) across all larval stages, (3) a slow global decrease in fluorescence, and (4) a rapid decrease in fluorescence during warming. The observed changes in fluorescence occurred in all tested animals, regardless of larval stage, suggesting that SD was present independently of the developmental state of nervous system. To characterize SD in the different larval stages, we first compared results from experiments in which the same cooling rate (3°C/min) was used. This avoided a potential influence of stimulus dynamics on the observed results. Figure 6C shows that the fluorescence change in all three larval stages was significant. However, there was an overall significant effect of larval stage on SD initiation temperature with third-instar larvae requiring the coldest stimulus (Figure 6D). Specifically, in these experiments, we paired all three larval stages on the same cooling plate to directly compare the effects of cooling between developmental stages. Finally, we tested the effects of different cooling rates on SD temperature. Second instar larvae were exposed to either 0.8 °C/min, 1.5 °C/min, or 3 °C/min cooling ramps. The slowest cooling rate required significantly colder temperatures to elicit SD (Figure 6E), suggesting that the neuronal processes involved in the cooling response possess temporal dynamics that modify SD susceptibility. To better determine the effects of increasing cold stress on the dynamics of the neuronal calcium signal during prolonged cooling, we extended the cold period by holding the temperature at a constant, cold temperature after SD had occurred (L2 larvae, >10 min at 0.6°C; Figure S2A). Like in previous experiments, all larvae showed a clear SD, with a rapid rise of the calcium signal that peaked while the temperature was dropping. Figures S2B and S2C show the time course of the aligned fluorescence signals of all animals and their average, which exhibited the four characteristic regimes described above. After the peak, there was an extended plateau, during which the calcium signal slowly diminished (with a time constant of 221.4 ± 26 s; Figure S2D). Shortly after the beginning of warming, the calcium signal of all larvae dropped rapidly until it reached approximately pre-SD levels. After this, the calcium signal continued to diminish but with a much slower time course, until room temperature was reached. At this point, the fluorescence was no longer distinguishable from the fluorescence at room temperature at the beginning of the trial (Figure S2C).Figure 6Changes in GCaMP6m fluorescence during cooling and rewarming of the larval nervous system can be classified into the same 4 regimes as the adult brain(A) Representative recording of GCaMP6m fluorescence in a third instar larva during cooling to near 0°C and then rewarming back to room temperature. (i) End of regime 1, 13°C. (ii) Start of regime 2, 8.5°C. (iii) End of regime 2 = start of regime 3, 6.1°C. (iv) End of regime 3 = start of regime 4, 8.1°C. (v) End of regime 4, 18°C.(B) Representative mean fluorescence of the whole nervous system over time showing the 4 regimes (R1 – R4) and their starting and ending points (i–v). The temperature stimulus used is shown in yellow.(C) Quantification of normalized mean fluorescence at points (ii) (start of the quick fluorescence rise) and (iii) (peak fluorescence). There was a significant increase in fluorescence in all larval stages (instars 1, 2, and 3; separate paired t tests, p < 0.001 [∗∗∗] each, N = 10). For each animal, mean fluorescence was normalized to the mean fluorescence at 18°C at the start of the cooling ramp. Data are represented as mean ± SEM.(D) Comparison of SD initiation temperature in the three larval stages. Third-instar larvae required significantly colder temperatures to elicit SD (RM ANOVA, N = 10, F(2,18) = 16.321, Student-Newman-Keuls post hoc test with p < 0.05 [∗]). Data are represented as mean ± SEM.(E) Comparison of different cooling speeds on SD initiation temperature (slow: 0.8 °C/min, medium: 1.5 °C/min, fast: 3 °C/min). The SD temperature of the slowest cooling rate was significantly lower (overall effect of cooling p = 0.002, one-way ANOVA, N ≥ 10, F(2,28) = 7.764), with Student-Newman-Keuls post hoc test with p < 0.05 [∗]). Data are represented as mean ± SEM.(F) Original recordings of GCaMP6m fluorescence during regime 2 showing that the fluorescence wave spreads from the VNC to the brain hemispheres. Times of the frames are shown from the start of the spreading wave (t = 0). The right image shows 4 ROIs used to measure the fluorescence in (G); see also Video S5.(G) Normalized fluorescence for the ROIs shown in (F). Arrows indicate the start of regime 2 and the start of the SD for each ROI. They are clearly separated in time, indicating that the fluorescence wave shows a spatial spread.(H) Same as (A) but for Arclight fluorescence. Lateral view of the nervous system. Temperatures are indicated in the figure.(I) Same as (B) but for Arclight fluorescence in second instar larvae.(J) Quantification of normalized mean fluorescence at points (ii) (start of the quick fluorescence drop) and (iii) (trough of fluorescence) showing a significant decrease in fluorescence (paired t test, p < 0.001 [∗∗∗]). Data are represented as mean ± SEM. Changes in GCaMP6m fluorescence during cooling and rewarming of the larval nervous system can be classified into the same 4 regimes as the adult brain (A) Representative recording of GCaMP6m fluorescence in a third instar larva during cooling to near 0°C and then rewarming back to room temperature. (i) End of regime 1, 13°C. (ii) Start of regime 2, 8.5°C. (iii) End of regime 2 = start of regime 3, 6.1°C. (iv) End of regime 3 = start of regime 4, 8.1°C. (v) End of regime 4, 18°C. (B) Representative mean fluorescence of the whole nervous system over time showing the 4 regimes (R1 – R4) and their starting and ending points (i–v). The temperature stimulus used is shown in yellow. (C) Quantification of normalized mean fluorescence at points (ii) (start of the quick fluorescence rise) and (iii) (peak fluorescence). There was a significant increase in fluorescence in all larval stages (instars 1, 2, and 3; separate paired t tests, p < 0.001 [∗∗∗] each, N = 10). For each animal, mean fluorescence was normalized to the mean fluorescence at 18°C at the start of the cooling ramp. Data are represented as mean ± SEM. (D) Comparison of SD initiation temperature in the three larval stages. Third-instar larvae required significantly colder temperatures to elicit SD (RM ANOVA, N = 10, F(2,18) = 16.321, Student-Newman-Keuls post hoc test with p < 0.05 [∗]). Data are represented as mean ± SEM. (E) Comparison of different cooling speeds on SD initiation temperature (slow: 0.8 °C/min, medium: 1.5 °C/min, fast: 3 °C/min). The SD temperature of the slowest cooling rate was significantly lower (overall effect of cooling p = 0.002, one-way ANOVA, N ≥ 10, F(2,28) = 7.764), with Student-Newman-Keuls post hoc test with p < 0.05 [∗]). Data are represented as mean ± SEM. (F) Original recordings of GCaMP6m fluorescence during regime 2 showing that the fluorescence wave spreads from the VNC to the brain hemispheres. Times of the frames are shown from the start of the spreading wave (t = 0). The right image shows 4 ROIs used to measure the fluorescence in (G); see also Video S5. (G) Normalized fluorescence for the ROIs shown in (F). Arrows indicate the start of regime 2 and the start of the SD for each ROI. They are clearly separated in time, indicating that the fluorescence wave shows a spatial spread. (H) Same as (A) but for Arclight fluorescence. Lateral view of the nervous system. Temperatures are indicated in the figure. (I) Same as (B) but for Arclight fluorescence in second instar larvae. (J) Quantification of normalized mean fluorescence at points (ii) (start of the quick fluorescence drop) and (iii) (trough of fluorescence) showing a significant decrease in fluorescence (paired t test, p < 0.001 [∗∗∗]). Data are represented as mean ± SEM. Quantification of the fluorescence change at several locations throughout the nervous system confirmed the presence of a spreading wave. Figures 6F and 6G show an example where SD started in the posterior section of the VNC and slowly spread toward the anterior (Video S5). Between animals, waves differed in starting location, but they were typically either initiated in the anterior brain lobes or in the posterior third of the VNC. They always spread throughout the entirety of the larval nervous system, however with velocities slower than those in the adult brain (0.37 ± 0.04 mm/min, N = 12, p = 0.049, t test). Video S5. False-color video of cooling-induced spreading wave of fluorescence in the third instar larval ventral nerve cord, related to Figure 6GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top left. Video S5. False-color video of cooling-induced spreading wave of fluorescence in the third instar larval ventral nerve cord, related to Figure 6 GCaMP6m was expressed pan-neuronally. Warmer colors represent higher fluorescence. The video is shown at five times the original speed. Anterior is to the top left. We confirmed that the calcium fluorescence wave was due to increased neuronal membrane depolarization using pan-neuronal Arclight expression. Similar to adult brains, the overall Arclight fluorescence response was inverted from that of GCaMP6m (Figure 6H and 6I), with an initial slow change in fluorescence in regime 1 and a rapid drop in regime 2 (Figure 6J) indicative of a rapid depolarization. This occurred at 5.3 ± 0.9°C (N = 8). Minimum fluorescence was reached at 3.0 ± 0.7°C (N = 8). Upon warming, the fluorescence continued to slowly decrease (Figure 6H iv; 8.1°C) but then rapidly increased as the temperature returned to 18°C (v). Our Arclight and GCaMP6m data thus suggest that larval fly brains resemble those of adults during cooling and that they undergo the physiological changes that have been described for SD.2 To test whether the spreading fluorescence wave observed in larvae can also be initiated by increasing the extracellular potassium concentration, we performed calcium imaging on fillet-dissected37 larvae while we rapidly spiked the saline bath with 1M KCl. Several animals were mounted in each experiment (range: 2–4), and all animals (N = 12) exhibited a spreading wave of rapidly increasing fluorescence that was initiated within 3.7 ± 1.3 s of the application. Figure 7A shows an example of a wave initiated in the left brain lobe that spread posteriorly to the VNC and laterally to the right brain lobe. The mean fluorescence of the whole nervous system (Figure 7B) closely resembled the cooling experiments, with a rapid rise in fluorescence shortly after KCl application. Peak fluorescence was 5.1 ± 0.3 times higher than before KCl application (Figure 7C) and a spreading wave was observed in all animals (N = 12). In the example in Figures 7A and 7D, the wave started in the left brain lobe (blue), reached the medial VNC (orange), then spread to the right brain lobe (purple) and the posterior VNC (green).Figure 7Increased potassium concentration facilitates the initiation of SD in the larval nervous system(A) Representative recordings of GCaMP6m fluorescence at room temperature while the nervous system was bathed in saline and then spiked with increased potassium. (i) before KCl application; (ii) shortly after KCl application; (iii) at peak fluorescence; (iv) after fluorescence started to diminish.(B) Whole nervous system mean fluorescence over time for the images in (A).(C) Normalized mean fluorescence at the start (i) and end (iii) of the spreading wave when maximum fluorescence occurs. The change in fluorescence was highly significant (N = 12, p < 0.001 [∗∗∗], paired t test). For each animal, mean fluorescence was normalized to the mean fluorescence value at the start of the experiment.(D) Normalized mean fluorescence at the ROIs shown in (Aii). The minimum to maximum fluorescence was normalized to 0 to 1.(E) Peak mean fluorescence time at the closest and farthest points from the location of the KCl application were not significantly different (N = 10, paired t test, p = 0.70).(F) Quantification of the temperature at which the spreading fluorescence wave was initiated in experiments where nervous systems were cooled in standard physiological saline (sal) and in saline with elevated (1.25×) KCl concentration. The initiation temperature in the elevated KCl saline was significantly higher than the temperature in standard saline (N = 11, p = 0.0012 [∗], paired t test).(G) SD confers resilience to subsequent SDs (see text for statistical details). Quantification of SD temperatures for SDs elicited by repeated cooling of second instar nervous systems. Subsequent SDs required colder temperatures to elicit SD.(H) With sufficient time between cooling cycles, there was no significant effect on SD temperature. Only cooling cycles 1 and 5 were applied. (N = 20, paired t test, p = 0.65).(I) When SD was elicited with KCl application to the nervous system, the effect was re-established, and the subsequent cooling-induced SD required a colder temperature to be elicited (N = 9, paired t test, p = 0.0093 [∗∗]). Only cooling cycles 1 and 5 were applied. KCl was applied instead of cooling cycle 4.(J) Quantification of normalized Arclight fluorescence at room temperature before the first cooling cycle, the end of the first cooling cycle, and 5, 10, and 20 min later. Only the fluorescence at the end of the cooling cycle was significantly lower (RM ANOVA, N = 8, F(4,28) = 12.506, Student-Newman-Keuls post hoc test, p < 0.05 [∗], unnormalized data.(K) Quantification of normalized GCaMP6m fluorescence at room temperature before the first cooling cycle, the end of the first cooling cycle, and 5, 10, and 20 min later (RM ANOVA, N = 12, F(4,44) = 12.717, Student-Newman-Keuls post hoc test, p < 0.05 [∗], unnormalized data).(L) Whole nervous system mean fluorescence measured at the start of each cooling cycle across 5 cooling-induced SD events showed a significant drop in fluorescence with each SD (p < 0.001, RM ANOVA, N = 8, F(4,28) = 121.18, Student-Newman-Keuls post hoc test, p < 0.05 [∗]).(M) Whole nervous system fluorescence when only the first cooling-induced SD was elicited. Cooling cycles 2–4 were not applied, and no SD was elicited, but the fluorescence light was turned on for the corresponding time. Fluorescence was significantly lower before cooling cycle 5 (paired t test, N = 11, p = 0.0015 [∗∗]).(N) Whole nervous system mean fluorescence when only the first cooling-induced SD was elicited and the fluorescence light was kept off after the first cooling. Fluorescence was still significantly lower before cooling 5 (paired t test, N = 7, p = 0.003 [∗∗]).(O) Comparison of whole nervous system fluorescence before cooling cycle 5 from experiments shown in (L)–(N), normalized to the fluorescence before the respective first cooling trials. Fluorescence was significantly lower when repeated SDs were elicited (One-way ANOVA, 7 ≤ N ≤ 11, F(2,23) = 16.259, Student-Newman-Keuls post hoc test, p < 0.05 [∗]). Having the fluorescence light on did not significantly affect the measured fluorescence. In all graphs, pooled data are represented as mean ± SEM. Increased potassium concentration facilitates the initiation of SD in the larval nervous system (A) Representative recordings of GCaMP6m fluorescence at room temperature while the nervous system was bathed in saline and then spiked with increased potassium. (i) before KCl application; (ii) shortly after KCl application; (iii) at peak fluorescence; (iv) after fluorescence started to diminish. (B) Whole nervous system mean fluorescence over time for the images in (A). (C) Normalized mean fluorescence at the start (i) and end (iii) of the spreading wave when maximum fluorescence occurs. The change in fluorescence was highly significant (N = 12, p < 0.001 [∗∗∗], paired t test). For each animal, mean fluorescence was normalized to the mean fluorescence value at the start of the experiment. (D) Normalized mean fluorescence at the ROIs shown in (Aii). The minimum to maximum fluorescence was normalized to 0 to 1. (E) Peak mean fluorescence time at the closest and farthest points from the location of the KCl application were not significantly different (N = 10, paired t test, p = 0.70). (F) Quantification of the temperature at which the spreading fluorescence wave was initiated in experiments where nervous systems were cooled in standard physiological saline (sal) and in saline with elevated (1.25×) KCl concentration. The initiation temperature in the elevated KCl saline was significantly higher than the temperature in standard saline (N = 11, p = 0.0012 [∗], paired t test). (G) SD confers resilience to subsequent SDs (see text for statistical details). Quantification of SD temperatures for SDs elicited by repeated cooling of second instar nervous systems. Subsequent SDs required colder temperatures to elicit SD. (H) With sufficient time between cooling cycles, there was no significant effect on SD temperature. Only cooling cycles 1 and 5 were applied. (N = 20, paired t test, p = 0.65). (I) When SD was elicited with KCl application to the nervous system, the effect was re-established, and the subsequent cooling-induced SD required a colder temperature to be elicited (N = 9, paired t test, p = 0.0093 [∗∗]). Only cooling cycles 1 and 5 were applied. KCl was applied instead of cooling cycle 4. (J) Quantification of normalized Arclight fluorescence at room temperature before the first cooling cycle, the end of the first cooling cycle, and 5, 10, and 20 min later. Only the fluorescence at the end of the cooling cycle was significantly lower (RM ANOVA, N = 8, F(4,28) = 12.506, Student-Newman-Keuls post hoc test, p < 0.05 [∗], unnormalized data. (K) Quantification of normalized GCaMP6m fluorescence at room temperature before the first cooling cycle, the end of the first cooling cycle, and 5, 10, and 20 min later (RM ANOVA, N = 12, F(4,44) = 12.717, Student-Newman-Keuls post hoc test, p < 0.05 [∗], unnormalized data). (L) Whole nervous system mean fluorescence measured at the start of each cooling cycle across 5 cooling-induced SD events showed a significant drop in fluorescence with each SD (p < 0.001, RM ANOVA, N = 8, F(4,28) = 121.18, Student-Newman-Keuls post hoc test, p < 0.05 [∗]). (M) Whole nervous system fluorescence when only the first cooling-induced SD was elicited. Cooling cycles 2–4 were not applied, and no SD was elicited, but the fluorescence light was turned on for the corresponding time. Fluorescence was significantly lower before cooling cycle 5 (paired t test, N = 11, p = 0.0015 [∗∗]). (N) Whole nervous system mean fluorescence when only the first cooling-induced SD was elicited and the fluorescence light was kept off after the first cooling. Fluorescence was still significantly lower before cooling 5 (paired t test, N = 7, p = 0.003 [∗∗]). (O) Comparison of whole nervous system fluorescence before cooling cycle 5 from experiments shown in (L)–(N), normalized to the fluorescence before the respective first cooling trials. Fluorescence was significantly lower when repeated SDs were elicited (One-way ANOVA, 7 ≤ N ≤ 11, F(2,23) = 16.259, Student-Newman-Keuls post hoc test, p < 0.05 [∗]). Having the fluorescence light on did not significantly affect the measured fluorescence. In all graphs, pooled data are represented as mean ± SEM. To test whether the location of the KCl application biases the SD initiation site, we simultaneously monitored several larvae that were pinned in different orientations (see STAR Methods). We determined the time of peak fluorescence from application at the closest and farthest points of the nervous system from where KCl was applied. There was no significant difference between peak times, indicating that the location of the drop relative to the nervous system did not bias spread initiation (Figure 7E). Additionally, there was no correlation between the drop location and at which time the individual animals exhibited SD. Finally, we tested the hypothesis that elevated extracellular potassium facilitates SD susceptibility in larval fly brains by bathing second instar nervous systems in elevated potassium saline solution (25% higher, 6.25 mM). This concentration was insufficient to elicit SD at room temperature. We first cooled the nervous system in regular saline to elicit SD. After recovery to room temperature, we exchanged the saline with elevated potassium saline and after 25 min, we repeated the cooling. As predicted by our results in adult brains (Figure 4F), we found that SD temperature was significantly higher in potassium elevated saline than in control (saline: 4.0 ± 0.4; potassium elevated saline: 6.2 ± 0.4, Figure 7F). Together, our data indicate that the larval fruit fly nervous system reliably experiences SD during cooling and that extracellular potassium changes are causally involved in this process. SD reduces susceptibility to subsequent SD To address whether there are lasting effects of SD, we performed calcium imaging of second instar brains with five repeated cooling cycles (3 °C/min, 5-min breaks, ∼15 min total per cycle). We found an overall significant effect of repeated cooling cycles on SD temperature (Figure 7G, p < 0.001, RM ANOVA, N = 8, F(4,28) = 21.341), with cycles 1, 2, and 5 being significantly different from all others (Student-Newman-Keuls posthoc test, p < 0.05, [∗]). SD was the easiest to elicit in cycle 1 (SD temperature: 7.5 ± 0.3°C, N = 8) and the hardest in cycle 5 (3.6 ± 0.7°C, N = 8). SD temperatures in cycles 3 and 4 were not different from one another, but significantly different from all others (Student-Newman-Keuls posthoc test, p < 0.05 [#]). The same effect was observed in first and third-instar larvae and adults (Figure S3). To exclude potential effects of experiment duration, we used second instar larvae and only carried out the first cooling cycle, skipped cycles 2 through 4 by leaving the animal at room temperature, and finally carried out cooling cycle 5. We kept overall times the same so that cycle 5 started at its usual time (60 min after the start of cycle 1). In this case, SD temperatures between cycles 1 and 5 were no longer significantly different (Figure 7H; N = 20, paired t test, p = 0.65). To confirm that the change in SD susceptibility with repeated cooling cycles was due to SD itself and not the exposure to cold temperature, we skipped cooling cycles 2 to 4 but elicited SD in cycle 4 with 1M KCl application. KCl was then washed out with regular saline, and the fifth cooling cycle was started at its usual time (60 min after the start of cycle 1). The SD temperature of cycle 5 was again significantly lower than cycle 1 (Figure 7I), suggesting that the initial SD event provided lasting effects that reduced SD susceptibility and that a single event was sufficient to cause that effect. To identify potential lasting effects of repeated SDs, we compared Arclight and GCaMP6m fluorescence over extended periods of time at room temperature after the first cooling-induced SD. At the end of the first cooling cycle, Arclight fluorescence was significantly lower than at the beginning (Figure 7J). However, from 5 min after the cycle and beyond, fluorescence was no longer different from baseline, suggesting that SD had no lasting effect on neuronal activity. For GCaMP6m, however, the fluorescence at the end of the first cycle was significantly higher than at the beginning (Figure 7K). It returned to baseline 5 min after the cooling cycle but then was significantly lower after 10 and 20 min. This effect was enhanced by repeated cooling cycles (and thus SDs), which resulted in significantly reduced fluorescence levels at the start of each cycle (Figure 7L). To test whether this lasting change in calcium fluorescence was due to the repeated SD events or was a consequence of the long experimental time and continuous exposure to the blue excitation light used for imaging, we carried out two control experiments using calcium imaging. First, instead of 5 repeated cooling cycles, we only carried out the first one, then skipped cycles 2 through 4 and measured fluorescence at the start of cycle 5. During skipped cooling cycles, the animals were left at room temperature and exposed to the excitation light as usual for trials 2–4. We found a significant decrease in fluorescence (Figure 7M), indicating a potential effect of the experiment duration or the continued light stimulus on either the GCaMP6m proteins or the calcium levels in the neurons. In a second control experiment, we thus did not expose animals to excitation light or cooling for cycles 2–4. This still resulted in a significant reduction in fluorescence (Figure 7N), suggesting that the long duration of the experiment diminished GCaMP6m fluorescence. To test, whether this effect could explain the reduction in fluorescence observed with repeated SD events (Figure 7L), we measured the relative change in fluorescence between the starts of cycles 1 and 5 for three groups: those with all cooling cycles, those with only the first cooling cycle plus light exposure, and those with only the first cooling cycle but without light exposure (Figure 7O). Experiments with repeated cooling cycles showed significantly larger reduction in fluorescence than the two experiments without cooling. The latter were not different from one another. Collectively, these experiments demonstrate that while there was a small effect of time on our fluorescence signal, repeated SDs significantly altered the baseline calcium fluorescence in larval neurons. Together, our data show that cooling-induced SD events provide a lasting effect that reduces susceptibility to subsequent SDs, independent of the brain’s developmental stage, and that the mechanisms providing this resilience are temperature-independent and directly activated by the occurrence of SD. Discussion We demonstrate that SD can be observed in whole brain calcium and voltage imaging in adult and larval Drosophila. SD could either be elicited through rapid cooling, a naturally occurring environmental condition, or through high potassium chloride application. Four distinct regimes corresponding to characteristic physiological features of SD were identified: an initial increase in intracellular calcium followed by sudden depolarization and accompanying increase in calcium, sustained depolarization, and rapid recovery. These regimes are indicative of the increased firing frequency and depolarization block2,8,40 associated with SD. Prior studies had provided evidence of SD in insects, including a sharp decrease in TPP at two spatially and temporally separated points during rapid cooling of the adult Drosophila nervous system16,19,20 and correlated measurements for triggering SD with high K+ while measuring the response at a separate location.22 However, clear confirmation of the dynamic connection between the events was not provided. Our results definitively confirm that a slowly traveling wave of depolarization spreads across the brain and that it coincides with the TPP decrease. Our measurements of SD threshold in elevated potassium saline applied to the bath show that the process is influenced by potassium with increased extracellular potassium facilitating SD initiation. In adult flies, SD initiation occurred at multiple sites with a singular stimulus and could be initiated in different locations with repeated stimuli, indicating that SD does not result from a singular defect and is not specific to neural architecture, connectivity, or cell type population. SD spread through the mushroom body was prevalently observed, perhaps indicating that this region is more prone to SD or that it is more readily observable here. We observed that SD typically began near the calyx (densely packed cell bodies) of one hemisphere before spreading to the pedunculus and alpha lobe, followed by the beta and gamma lobes (axonal projections) of that hemisphere. It often then jumped to the gamma and beta lobes of the other hemisphere and spread in a reverse pattern into the calyx. This suggests SD propagation through sites with high cell body density and many axons is mediated through local changes. If instead SD spread was mediated through electrical signaling in the axons (i.e., action potential propagation), a rapid jump from the initiation site (e.g., calyx) to where axons terminate (e.g., mushroom body lobes) would be expected. This was not observed and is consistent with mammalian studies showing that SD does not cross white matter, preventing spread into different brain regions.2 Likewise, in adult insects, there is evidence that SD does not cross between ganglia.33 Our larval data show that SD spreads between the brain lobes and VNC, which unlike in adult insects, is a single consolidated ganglion. This further supports the idea that local diffusion, rather than neuronal activity propagates SD. We also observed spread in other neuropils (superior lateral protocerebrum, lobula, antennal lobes), in which neuronal densities and synaptic connectivity differ, thus indicating the universality of SD. In addition to demonstrating SD existence in adult flies, we show that SD exists in all larval instars with properties similar to SD in adults. Structure, neuronal connectivity, and neuronal numbers are distinct between the larval stages and adults, thus pointing to the fundamentality of SD and its independence of these features. This idea is also supported by the occurrence of SD in animals from different clades (mammals, birds, and insects).2,23,41 The evolutionary origins of SD have been hypothesized to arise from the development of high neuronal densities, tightly regulated ion homeostasis, and the ability to rapidly fire action potentials in response to stimuli.36 Larvae, like adults, regulate ion homeostasis through glial cells, Na+/K+ pumps, and ion channels but have reduced sensory functions that do not require the same rapid responses to sensory stimuli as adults. Our results also indicate differences between larval and adult SD, which may help identify relevant characteristics. In larvae, the fold change in calcium fluorescence was less and the spread velocity was slower than in adults, suggesting that neuronal excitation may be lower in larvae than adults. One contributing factor may be the different proportions of spiking to nonspiking neurons across development. Drosophila have only one type of sodium channel (para42) but with many isoforms. In third-instar larvae, para is only expressed in 23% of the neurons,42 suggesting that only this subset is able to fire Nav-dependent action potentials. In contrast, the adult central nervous system widely expresses para and therefore many more neurons may be able to fire action potentials.42 If the accepted working hypothesis of SD is correct, and neurons undergo a rapid increase in firing frequency followed by depolarization block,8,40 then a nervous system with fewer spiking neurons (larvae) would be expected to behave differently than one with many (adults). Thus, in larvae there are fewer neurons to be trapped in depolarization block, resulting in a lower overall depolarization and delayed or reduced ion homeostasis disruption. This manifests in a reduced average calcium fluorescence in larvae. Additionally, because extracellular potassium diffusion is hypothesized to help mediate SD spread, a larval system with fewer neurons in depolarization block would experience slower potassium accumulation and a reduced spread velocity. Despite the differences between larvae and adults, SD was consistently elicited in both with no difference in the initiation threshold. Additionally, we showed that elevated extracellular potassium levels are causally related to SD initiation in both larvae and adults. Thus, the reduced number of spiking neurons in larvae does not affect SD susceptibility or its dependence on extracellular potassium. We also show that a single SD reduces susceptibility to future SDs, independent of its initiation mechanism. This effect was present within minutes, pointing to a cellular- or circuit-level memory trace from a lingering adaptation of the nervous system. Such lasting effects were suggested in the initial reports of SD43 and subsequently reported as a refractory period lasting several minutes in mammalian and avian models.30,44 In insects, prior exposure to temperature changes through either acclimation or rapid cold hardening has been shown to alter SD susceptibility and dynamics.22,45 Lasting effects have also been observed in SD-related pathologies, including migraine disorders, although changes to susceptibility remain unclear.46 Possible reasons include disrupted ion homeostasis, neuronal excitability changes, alterations to ion pump activities, or persistent changes in neurotransmitter levels or synaptic activity. All SD events are associated with a dramatic increase in extracellular potassium levels. It is unlikely, however, that residually elevated extracellular potassium directly contributes to the refractory state after SD as it should increase neuronal excitability. Neuronal excitability is a modulator of SD susceptibility, and an increase should lead to a lower, not a higher threshold for SD. However, residually elevated extracellular potassium may play an indirect role by activating sustained homeostatic processes that influence neuronal activity. For example, in rats, sustained exposure to elevated extracellular potassium leads to persistent changes in neuronal excitability mediated by calcium-dependent processes that alter membrane potential and input resistance47 and causes long-lasting alterations in calcium channel function.48 These effects are reminiscent of slow activity-dependent homeostatic plasticity acting through changes in gene expression to restore cell excitability over many hours and days (e.g., Turrigiano49 and Golowasch50). They are thus unlikely to have contributed to our observed memory trace after SD. The short-term effects of SD on mammalian cortical networks are less apparent: only minor changes to the pyramidal cell excitability were observed (slightly increased rheobase) that neither affected input resistance nor frequency-current (F/I) curves.30 In contrast, there is evidence for rapid and short-term (seconds to minutes) excitability changes after exposure to elevated extracellular potassium in crustacean stomatogastric neurons.51 While the specific mechanisms remain unclear, phosphorylation of voltage-gated ion channels has been speculated to rapidly change cell excitability. Additionally, SD-induced changes in potassium and sodium concentrations are likely to increase Na+/K+ pump activity.52 The consequences for cell excitability have been seen after bursts of high neuronal firing where pump activity is a contributor to spike frequency adaptation and increased after-hyperpolarization that outlasted burst activity for several seconds.53,54,55 Indeed, our Arclight experiments suggest sustained hyperpolarization after SD but also that it disappears within 5 min. In contrast, our calcium data imply a longer-lasting reduction of neuronal excitability (>30 min). These results are consistent with modeling studies of Drosophila larval neurons that suggest that Na+/K+ pump mediated reductions of cell excitability can outlast the rather short effects of the pump on membrane potential and persist in the absence of membrane potential changes.55 These effects are predicted to be mediated by the interdependent actions of ion concentration changes, diffusion, and pump activation. Neuronal hyperexcitation during SD can lead to excessive neurotransmitter release (e.g., glutamate and GABA2) with lasting changes to release properties and presynaptic neuron firing causing a sustained shift in the ratio of excitatory to inhibitory synaptic inputs.30 Similar effects could lead to the protective effect we observed after SD. Such lasting effects could be mediated by synaptic plasticity (e.g., depression) in response to the excessive synaptic release during SD, or by enhanced and prolonged enzyme degradation of the released transmitters and facilitated transport into glial cells such as astrocyte-like glia or sub-perineurial glia. Astrocyte-like glia are in intricate proximity to the neuropil and responsible for neurotransmitter clearance, e.g., glutamate uptake through the excitatory amino acid transporter (EAAT56). Sub-perineurial glia regulate transport from the hemolymph into the ganglion, acting as a blood-brain barrier.57 They hypertrophy during the third instar, which may explain why in the comparison between the different larval stages, the third-instar larvae were the most difficult in which to elicit SD. Combined, our results demonstrate that SD is a phenomenon that persists across developmental stages and in networks with vastly different sizes, connectivity, and neuronal populations. Its occurrence in Drosophila shows similar characteristics to that of mammals, with slow spread across large brain areas and dramatic changes in neuronal behavior. We have shown that SD is facilitated by increased extracellular potassium, can originate from multiple independent sites, and that these sites vary with repeated events. A singular SD causes lasting effects observable in the depressed baseline calcium signal of the neurons and offers a limited-term protection from future SDs. Limitations of the study This paper demonstrates that Drosophila is a powerful model for investigating SD, but we recognize several limitations that should be considered. First, the wide-field imaging technique used is restricted to a single focal plane and limits the analysis of the spatial spread of SD, particularly in adult brains. As a result, the three-dimensional aspects of SD dynamics remain to be further explored. Our approach also does not have the temporal and spatial resolution to provide direct measures of individual neuronal activities and membrane depolarization, leaving the cellular underpinnings of SD initiation and spread to future investigations. Second, while cooling represents a naturally occurring environmental stress for insects and enables reproducible SD induction, the temperature-dependent processes involved in the neuronal depolarization that lead to SD initiation at cold temperatures remain unresolved. Third, our demonstration of SD across larval and adult stages points to a fundamentality of the phenomenon, although developmental differences in SD initiation and propagation dynamics were observed. The mechanistic underpinnings for these differences remain speculative, as do the minimal circuit and developmental requirements for the presence of SD, with differences in neuronal excitability, ion channel expression, and glial function being potentially important contributing factors. Finally, while we demonstrate a robust and prolonged reduction in SD susceptibility following a single SD event, the cellular mechanisms underlying this refractory period remain unresolved. Dissecting these mechanisms will be an important direction for future studies. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Wolfgang Stein (wstein@ilstu.edu). Materials availability This study did not generate new unique reagents. Data and code availability The data have been deposited at Mendeley Data and are publicly available as of the date of publication at Mendeley Data: https://doi.org/10.17632/32vnnb77kh. All data reported in this paper will be shared by the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. All original code has been deposited and is publicly available at Mendeley Data: https://doi.org/10.17632/32vnnb77kh and Mendeley Data: https://data.mendeley.com/datasets/32vnnb77kh. Acknowledgments We would like to thank the College of Arts and Sciences and the Physics Department at Illinois State University for providing financial support. We would also like to thank the 10.13039/100009500School of Biological Sciences for providing laboratory space and commodities. Additional financial support through Firebird and Birdfeeder grants was provided by the Office of Student Research at Illinois State University. Further thanks go to Pedro Galvan and Abigail Spena for carrying out preliminary experiments. Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) were used in this study. Author contributions Conceptualization, A.L.H. and W.S.; data curation, A.L.H., W.S., G.C., E.N., and M.A.M.; formal analysis, A.L.H., W.S., G.C., E.N., and M.A.M.; funding acquisition, A.L.H., W.S., G.C., E.N., and M.A.M.; investigation, A.L.H., W.S., G.C., E.N., and M.A.M.; methodology, A.L.H. and W.S.; project administration, A.L.H. and W.S.; resources, A.L.H. and W.S.; software, A.L.H. and W.S.; supervision, A.L.H. and W.S.; validation, A.L.H., W.S., G.C., E.N., and M.A.M.; visualization, A.L.H. and W.S.; writing – original draft preparation, A.L.H. and W.S.; Writing – review and editing, A.L.H. and W.S. Declaration of interests The authors declare no competing interests. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERChemicals, peptides, and recombinant proteinsFormula 4–24 Instant Drosophila Medium, PlainCarolina Biological Supply CompanyItem #: 173200Sylgard 184SigmaN/ADeposited dataDataMendeley Datahttps://doi.org/10.17632/32vnnb77khhttps://data.mendeley.com/datasets/32vnnb77khExperimental models: Organisms/strainsnSyb-GAL4: w[1118]; P{y[+t7.7] w[+mC] = GMR57C10-GAL4}attP2Bloomington Drosophila Stock CenterBDSC # 39171UAS-GCaMP6m: w[1118]; P{y[+t7.7] w[+mC] = 20XUAS-IVS-GCaMP6m}attP40Bloomington Drosophila Stock CenterBDSC # 42748UAS-Arclight: w[∗]; P{y[+t7.7] w[+mC] = UAS-ArcLight}attP40/CyOBloomington Drosophila Stock CenterBDSC # 51057Software and algorithmsFijihttps://fiji.sc/RRID:SCR_002285CorelDraw X7Corel CorporationRRID:SCR_014235Excel 365MicrosoftRRID:SCR_016137Adobe IllustratorAdobeRRID:SCR_010279VSDC video editorFlash-Integro LLCN/ASpike2Cambridge Electronic Design LimitedRRID:SCR_000903SigmaPlotSystat SoftwareRRID:SCR_003210Custom Analysis CodeMendeley Datahttps://doi.org/10.17632/32vnnb77khhttps://data.mendeley.com/datasets/32vnnb77kh REAGENT or RESOURCESOURCEIDENTIFIERChemicals, peptides, and recombinant proteinsFormula 4–24 Instant Drosophila Medium, PlainCarolina Biological Supply CompanyItem #: 173200Sylgard 184SigmaN/ADeposited dataDataMendeley Datahttps://doi.org/10.17632/32vnnb77khhttps://data.mendeley.com/datasets/32vnnb77khExperimental models: Organisms/strainsnSyb-GAL4: w[1118]; P{y[+t7.7] w[+mC] = GMR57C10-GAL4}attP2Bloomington Drosophila Stock CenterBDSC # 39171UAS-GCaMP6m: w[1118]; P{y[+t7.7] w[+mC] = 20XUAS-IVS-GCaMP6m}attP40Bloomington Drosophila Stock CenterBDSC # 42748UAS-Arclight: w[∗]; P{y[+t7.7] w[+mC] = UAS-ArcLight}attP40/CyOBloomington Drosophila Stock CenterBDSC # 51057Software and algorithmsFijihttps://fiji.sc/RRID:SCR_002285CorelDraw X7Corel CorporationRRID:SCR_014235Excel 365MicrosoftRRID:SCR_016137Adobe IllustratorAdobeRRID:SCR_010279VSDC video editorFlash-Integro LLCN/ASpike2Cambridge Electronic Design LimitedRRID:SCR_000903SigmaPlotSystat SoftwareRRID:SCR_003210Custom Analysis CodeMendeley Datahttps://doi.org/10.17632/32vnnb77khhttps://data.mendeley.com/datasets/32vnnb77kh Experimental model and study participant details Fly stocks and maintenance Fly stocks were obtained from the Bloomington Drosophila Stock Center (Bloomington, IN). Flies were reared and maintained at room temperature (22°C) with the natural daily light cycle in vials with food (Formula 4-24 Instant Drosophila Medium, Plain) from Carolina Biological Supply Company (Burlington, NC). The first filial 1 (F1) generation was obtained by crossing virgin females of an nSyb-GAL4 line, w[1118]; P{y[+t7.7] w[+mC] = GMR57C10-GAL4}attP2 (BDSC # 39171) with males of either a GCaMP6m line w[1118]; P{y[+t7.7] w[+mC] = 20XUAS-IVS-GCaMP6m}attP40 (BDSC # 42748) or an Arclight line w[∗]; P{y[+t7.7] w[+mC] = UAS-ArcLight}attP40/CyO (BDSC # 51057). Adult F0 flies were allowed to lay eggs for 3–4 days and then larvae of the first, second, or third instar stage were separated from the food 1 to 4 days later. A 20% sucrose solution in water was added to the fly vial and floating larvae were collected after 10 min. Larval stage was identified based on size. For adult F1 flies, the larvae were allowed to mature to adulthood and adults were collected for a maximum of 7 days post-eclosion of the first F1 fly. Adults were then transferred weekly to new vials. Male and mated female adult flies of age 17 to 67 days (mean: 27 ± 3.6) were used in approximately equal proportions in the experiments. Method details Reagents A 20% sucrose solution in water was used to extract larvae from the food. Dissections and recordings were performed using physiological saline58 consisting of (in mM): 108 NaCl, 5 KCl, 4 MgCl2∗6H2O, 3 CaCl2∗2H2O, 1 NaH2PO4, 4 NaHCO3, 5 trehalose, 10 sucrose, 5 HEPES. pH was adjusted to 7.5 with NaOH. For some experiments, a 1.25 x KCl saline (6.25 mM) was used. All reagents were obtained from Sigma Aldrich (St. Louis, MO). For determining a possible intrinsic temperature dependence of the GCaMP6m protein, brains from 16 adult flies less than 5 days old were dissected out in a calcium-free buffer solution consisting of (in mM): 10 HEPES, 2 EGTA, 135 NaCl. Brains were placed in 50 μL of the calcium-free buffer and homogenized using a homogenizer (Bel-Art ProCulture; Thermo Fisher Scientific, Waltham, MA). 1 μL of the homogenized brain solution was mixed with 1 μL of a 100 μM calcium buffer solution containing (in mM): 10 HEPES, 2 EGTA, 135 NaCl, 100 CaCl2∗2H2O. Fluorescence imaging under 470 nm illumination light was performed on the final 50 μM brain solution as the temperature was cooled from 20°C down to 0°C. As a control to detect potential autofluorescence, the brain solution was also imaged under 525 nm light at room temperature. No autofluorescence was observed. Finally, 2 μL of 50 μM calcium solution without brain tissue was imaged under 470 nm light. No fluorescence was observed. Dissection and experiment holders Custom-made dissection dishes and animal holders were used during the experiments. For adults, a 3.75 × 2.5 × 0.025 cm (length x width x thickness) piece of aluminum was used as a base for securing the flies with dental cement (Figure 1A). For larvae, 25 μm thick aluminum was used to create a 6.25 × 6.25 × 0.95 cm (length x width x height) square tray. A layer of 0.25 cm thick Sylgard 184 (Sigma, St. Louis, MO) was added to the holder. Dissections and mounting of animals were performed using a stereo microscope (Leica, Wetzlar Germany) with fluorescence adapter (Kramer Scientific, Amesbury, MA). Fluorescence excitation (470 nm) was provided by a Mightex BLS-Series High-Power LED Collimator Source (Mightex, Toronto, Ontario, Canada). Adult cooling experiments Adult flies (aged 17 to 67 days) were secured to the adult fly holder using dental cement (Protemp, ESPE, St. Paul, MN). The wings and legs were placed in the cement to eliminate movement. A petroleum jelly well was created around the head and 0.5 mL of physiological saline was added. The abdomen and thorax of the fly remained outside the saline well. Care was taken to not obstruct the spiracles. The dorso-posterior head cuticle and trachea were removed to expose the brain (Figures 1A and 1B). The holder with flies was then placed on the top of a 40 × 40 mm Peltier chip (Peltier Module TEC1-12706) that was controlled by a regulated power supply (Mastech HY3010Ex, Mastach. San Jose, CA). A water-cooling block below the Peltier chip ensured efficient heat transfer (Figure 1A). A temperature logger (TC0520, PerfectPrime, New York, NY) recording at 0.2 Hz was placed into the physiological saline well. This setup allowed us to record large-scale neuronal activity by measuring GCaMP fluorescence (Figure 1C), which correlates with neuronal depolarization and is particularly reliable in indicating the rise of neuronal activity.59 Brains exhibiting spontaneous activity were cooled from room temperature (20°C–22°C) to near 0°C with recordings taken between 18°C and 0°C. We tested cooling rates between 0.8 °C/min and 3 °C/min, but for each experiment, the rate of cooling was kept constant. In all cases, cooling was followed by a warming back to room temperature at a rate of 3 °C/min. Larval cooling experiments For high-throughput experiments, up to 16 larvae were secured to the larval fly holder that was mounted on a cooling platform (Figures 5A and 5B). Calcium fluorescence of all animals was imaged simultaneously (Figure 5C). Recordings were aided by the fact that the Drosophila larval stages are translucent, which allowed us to determine whether SD occurred in fully intact animals without the need to dissect out the nervous system. Each larva was secured using 2 pins, one near the mouth hooks and another near the spiracles at the posterior end. A 0.2 mL drop of physiological saline was added to cover all larvae on the holder. In a second assay, we used the “fillet” dissection37 to open the dorsal cuticle and expose the nervous system (Figures 5D and 5E). Larvae were individually pinned with their dorsal side up using 2 pins, one near the mouth hooks and another near the spiracles at the posterior end. A 0.2 mL drop of physiological saline was added to cover all larvae on the holder. Two small transversal cuts were made at the anterior and posterior ends of each larva, and a longitudinal incision was made along the dorsal body wall of the animal to fillet it. Additional pins were used to open and secure the body wall. The gut and trachea were removed to expose the brain. As in the case with the adults, the holder with the larvae was placed on a water-cooled Peltier chip. A temperature probe recording at 0.2 Hz was placed into the physiological saline surrounding the larvae and the temperature was adjusted by changing the voltage supplied to the Peltier chip. This setup allowed us to track fluorescence changes in more detail and in higher resolution due to the exposed nervous system and to manipulate the saline solution surrounding the nervous system. Figure 5F shows a GCaMP6m fluorescence image of a single third instar nervous system at room temperature. Like in adults, we cooled nervous systems at rates between 0.8 °C/min and 3 °C/min from room temperature (20°C–22°C) to near 0°C and then warmed back to room temperature at a rate of 3 °C/min. Cooling ramps were kept consistent between trials for the same experiment, but different experiments tested different ramps. Potassium chloride application experiments To test whether extracellular KCl application could elicit SD, we increased the potassium concentration in the saline bath that surrounded the fly brain (in adults) or the brain hemispheres and VNC (in larvae). A drop (4 μL) of 1M KCl solution was added to the bath, and the location and time of application were noted. Fluorescence imaging continued throughout the application. The application rapidly increased the potassium concentration from the normal 5 mM–12.9 mM, with potential higher local concentrations during the application. Electrophysiology Transperineurial potentials were obtained by impaling the adult brain using 20–30 MΩ glass microelectrodes filled with 0.6 M K2SO4 + 20 mM KCl electrolyte solution. Electrodes were pulled using a Sutter P97 puller (Sutter Instruments, Novato, CA). Signals were filtered and amplified through an Axoclamp 900A amplifier (Molecular Devices, San Jose, CA) in bridge mode. Files were recorded, saved, and analyzed using Spike2 Software at 10 kHz (version 7.18; CED, Cambridge, UK) and a Power 1401 (CED). Fluorescence imaging and analysis Fluorescence data were recorded at 10 Hz using an Olympus-BX51 epifluorescence microscope with either a UMPLFLN 10XW (high-resolution experiments; 0.3NA) or MPLFLN 4× objective (high-throughput experiments; 0.15NA), CoolLed PE-4000 fluorescence illuminator (470 nm; 505 nm excitation cut-off filter, 525/50 nm emission filter), and Basler ace acA4024-29μm camera (Basler, Ahrensburg, Germany). Images were imported into FIJI60 and regions of interest (ROIs) were drawn around either the entire brain (to determine start and end of SD) or small regions of the brain (for propagation analysis). The mean fluorescence value for each ROI was plotted as a function of time, which was then correlated with temperature. A custom-made script was used to identify the starting and ending times of SD. The ending time was defined to be the time of maximum fluorescence. The starting time was defined to be the time where fluorescence began to increase rapidly. This point was found by shifting and rotating the coordinate axes of the fluorescence vs. time curve (Figure S4). A new origin was defined at the initial fluorescence value (t = 0) and the x- and y axes were rotated such that the new y axis was along a line from the new origin to the maximum of the original fluorescence vs. time curve. The sharp change in slope identifying the start of SD could then be found from the point in the fluorescence curve that had the largest x value in the new coordinate system. To describe the temporal dynamics of the fluorescence signal, we used 1st order exponential curve fits to determine the time constant of change. Time constants were determined in each animal and then averaged. Curve fits were created using the Spike2-internal fit function (F(t)=a0∗e−ta1+a2). Quantification and statistical analysis For normally distributed and paired data, either paired t-tests (for comparison of two conditions) or RM ANOVA with Student-Neuman-Keuls (SNK) posthoc tests at p < 0.05 were used. For other paired data, nonparametric Friedman RM ANOVA on Ranks tests were used. For unpaired normally distributed data, either t-tests (for two group comparisons) or One-Way ANOVAs with SNK posthoc tests were used. For other unpaired data, One-Way ANOVAs on Ranks were used. Statistical tests were calculated with Sigmaplot (v15, Grafiti LLC, Palo Alto, CA). Unless stated otherwise, data are presented as mean ± SEM. Alternatively, individual data points for each animal are given. Significant differences are stated as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Figures were prepared with CorelDraw X7 (Corel Corporation, Ottawa, Canada), Excel 365 (Microsoft, Redmond, WA), Adobe (San Jose, CA), SigmaPlot (version 11, Systat Software, San Jose, CA), and VSDC video editor (Flash-Integro LLC, Tashkent, Uzbekistan).