Syntaxin1A-mediated Resistance and Hypersensitivity to Isoflurane in Drosophila melanogaster

ELUCIDATING how general anesthetics work has been an ongoing debate in medical research.¹  Why has uncovering the mechanism of general anesthesia remained problematic even though some molecular targets have now been identified² ? One reason perhaps relates to teasing apart the plethora of effects that general anesthetics have on various targets in the brain, both at the molecular and circuit levels.²  Recent evidence suggests that there is a relation between endogenous sleep pathways and general anesthesia. General anesthetics may take effect through arousal pathways via disinhibition (or activation) of sleep-promoting circuits.^2–8  Consistent with this idea is evidence that general anesthetics, particularly intravenous anesthetics such as propofol, can potentiate γ-aminobutyric acid (GABA) type A receptors,^9–12  and the arousal centers in the brain are inhibited by GABAergic input from sleep-promoting centers.⁶  Yet, not all animals sleep¹³  but all animals can be rendered unresponsive by general anesthetics.¹⁴  This suggests that other anesthetic mechanisms might exist in addition to those targeting GABAergic sleep–arousal pathways. Also, general anesthetics promote a more profound loss of responsiveness than can ever be achieved by sleep, so other mechanisms clearly must be involved.

Another mechanism through which general anesthetics could act is by disrupting neuronal communication, by directly targeting synaptic transmission. In the mouse hippocampus, the volatile anesthetic halothane has been shown to impair transmitter release from glutamatergic synapses.¹⁵  This decrease in glutamate release is independent of any increase in GABAergic transmission,¹⁶  indicating that anesthetics are affecting glutamate release directly. Such decreased release could reflect anesthetic action on the transmitter release machinery. Indeed, a genetic screen for anesthetic sensitivity in the nematode model Caenorhabditis elegans identified specific mutations in the synaptic protein syntaxin1A that produced animals that were either resistant or hypersensitive to volatile anesthetics,¹⁷  suggesting that volatile general anesthetics may directly target the neurotransmitter release machinery. In particular, a syntaxin1A isoform containing a deletion in the H3 domain of the protein was found to confer a high level of resistance to isoflurane and halothane.¹⁷  An equivalent mutation engineered in mammalian cell lines was found to reduce the effects of isoflurane¹⁸  and propofol¹⁹  on transmitter release, suggesting that syntaxin1A is central to general anesthetic mechanisms. Because the H3 domain of syntaxin1A is extremely conserved across all animals,²⁰  we hypothesized that these anesthetic resistance effects on the transmitter release machinery are likely to be preserved across species, representing a conserved target mechanism for general anesthesia, in addition to sleep-related mechanisms.

General anesthetics first produce unconsciousness by activating endogenous sleep pathways. Accordingly, in previous work, we have identified a sleep–wake pathway that controls sensitivity to isoflurane in Drosophila melanogaster.³  In this study, we investigate synaptic release as an alternate anesthetic mechanism, by assaying the effects of two mutations in the H3 domain of syntaxin1A, across different behavioral and electrophysiological endpoints. We found that these syntaxin1A mutations produced both resistance and hypersensitivity, mirroring the nematode results. Remarkably, syntaxin1A-induced resistance to isoflurane was observed across all endpoints, ranging from behaviors in adults and larvae to effects at the larval neuromuscular junction (NMJ). Our results suggest that syntaxin1A-mediated neurotransmitter release represents a parallel target process of general anesthetics that is independent of species-specific sleep circuitry.

D. melanogaster were cultured on a yeast–sugar–agar medium in vials at 25°C on a 12-h light–dark cycle. Female flies (3 to 5 days old) were selected at random for behavioral experiments by brief carbon dioxide exposure and kept in food vials overnight before experiments. The control strains used in this study were wild-type Canton-S strain (CS) (Bloomington Stock Center, USA) and isoCJ1.^21,22  The syntaxin1A mutants used in this study have been described previously: the deletion mutant, syxH3-C²³  and syxKARRAA.²⁴  These strains express a syntaxin1A-mutant protein in a heterozygous null syx²²⁹ background: syxH3-C, genotype: yw; P(syx[H3-C]); syx1A^Δ229/TM6 (gift from Hugo Bellen, Ph.D., Department of Molecular and Human Genetics and Neuroscience, Howard Hughes Medical Institute, Houston, Texas); syxKARRAA, genotype: yw;P[CaryP]attP40 (HA-syx1A^KARRAA); syx1A^Δ229/TM6b. The genetic control strain for syxKARRAA is syxWT: yw;P[CaryP]attP40 (HA-syx1A^WT); syx1A^Δ229/TM6b (gifts from Patrik Verstreken, Ph.D., V.I.B. Center for the Biology of Disease, Leuven, Belgium). Syntaxin1A mutations were also placed on a common wild-type background (isolated from the null background) by outcrossing for five generations to isoCJ1.

Flies were frozen in dry ice, and their heads collected with a sieve (no. 25 and no. 40; Thermo Fisher, Australia). Heads were homogenized in sample buffer (50 mM Tris-HCl, pH 6.8, 2% sodium dodecyl sulfate, 10% glycerol, 12.5 mM EDTA, 1 mM dithiothreitol, plus protease inhibitor cocktail [Roche, Australia]). Proteins were added to commercial sample buffer before sodium dodecyl sulfate–polyacrylamide gel electrophoresis analysis (NuPage LDS Sample Buffer; Life Technologies, Australia). Proteins were separated using a 12% NuPage polyacrylamide gel (Life Technologies) with buffers prepared and protocol followed as per the manufacturer’s recommendations. After separation, the proteins were transferred to a polyvinylidine fluoride membrane (Merck Millipore, Australia) in transfer buffer (25 mM Tris, 0.2 M glycine, and 20% v/v methanol). The transfer apparatus was assembled to allow the proteins to transfer to the polyvinylidene fluoride membrane using settings recommended by the manufacturer. Proteins were detected using a syntaxin antibody (mouse anti-8c3, 1:1,000; Developmental Studies Hybridoma Bank, USA).

The fly coordination assay was developed from previous Drosophila general anesthesia assays.^25,26  Approximately 10 flies are loaded into 100 ml cylindrical glass tubes (length 20 cm and diameter 2.5 cm) with a side arm (Pyrex, USA), custom fit with a rubber stopper. Into the side arm of the tube, 3 μl isoflurane (Attane; Baxter Healthcare, Australia) was added using a Hamilton syringe (Hamilton Company, USA), which corresponds to 0.15 vol% isoflurane as quantified by gas chromatography (table 1). Experiments with halothane (Sigma-Aldrich, Australia), which is more potent than isoflurane, were performed with 2 μl injected as a liquid. Immediately after anesthetic injection, the number of flies in the bottom 2 cm of the tube (and therefore unable to hold on to or climb onto the sides) was counted every 10 s. This was repeated until the last fly had dropped to the bottom of the tube.

The startle-induced locomotion assay assesses flies’ movement response to a startle-inducing vibration stimulus and has been described previously.³  Flies are loaded into individual glass tubes (Trikinetics, USA; length 65 mm and diameter 3 mm), with paper and cotton rolled together on either end of the tubes to allow anesthetic gas to reach the fly. Tubes are placed on two custom-made scaffolds set into a closed chamber at least 15 min before the beginning of an experiment. Each scaffold apparatus holds 20 tubes, enabling comparison of a total of 40 flies per experiment. Startle stimuli were delivered using four shaft-less vibrating motors (model 312-101, Precision Microdrives, United Kingdom). Stimulus intensities were controlled using a custom MATLAB program (Mathworks, USA) (Drosophila Arousal Tracking²⁷ ) interfacing with the analog output channels of a USB data acquisition device (Measurement Computing, USA). The startling stimulus used in this study was a 5 × 200 ms vibration set at 1.3 g, delivered every 1 s. The startle stimulus amplitude and sequence chosen was optimized for the common isoCJ1 genetic background strain (see Supplemental Digital Content 1, fig. 1, https://links.lww.com/ALN/B139, for responses of flies to startle stimulus presentations without isoflurane gas).

Fly activity is filmed continuously using a webcam (5 frames per second; Logitech, Australia) all throughout the duration of an experiment (approximately 70 min). The experimental protocol consists of baseline and startle behavioral metrics, which are both measured as the velocity of the flies (mm/s) during a 1-min period immediately before and after the stimulus respectively. Baseline reflects the general locomotion capabilities of the fly (in air and at different drug concentrations), and the startle is the movement of the flies after the vibration stimulus. During each anesthesia trial, baseline is defined as the 1 min before the vibration stimulus, and the startle response is defined as the 1 min of fly activity immediately after the vibration stimulus. After this 1 min of locomotion activity, the concentration of isoflurane gas is increased until the concentration reaches 1 vol% isoflurane. For each experiment, the concentrations used were 0, 0.12, 0.25, 0.37, 0.5, 0.75, and 1 vol% isoflurane. The startle stimulus is delivered automatically every 10 min until the end of the experiment. Data were extracted and analyzed using custom MATLAB software.²⁷ 

Nine third-instar larvae were brushed from food vials and pipetted into the bottom of 100 ml cylindrical glass tubes (length 20 cm and diameter 2.5 cm; Pyrex). Larvae were allowed to recover for 1 min after this manipulation, during which time they typically began crawling up the sides of the tube. After this 1-min recovery period, a set volume of volatile anesthetic (isoflurane or halothane) was added into the side arm of the tube using a Hamilton syringe (Hamilton Company). After 4 min of anesthesia equilibration time, the larvae’s position was circled with a marker on the outside of the glass. After a further 1 min, the number of larvae that had moved at least one-body-length outside the marked circle was noted, and data converted to express the number of larvae that moved as a proportion of the total.

Locomotor activity during several days was monitored using similar materials as per the startle-induced locomotion assay (see Materials and Methods, Startle Assay). Flies were loaded into individual tubes and placed into behavioral scaffolds holding 17 tubes per apparatus. Tubes were sealed with food capped with wax and rolled cotton. The behavioral apparatus was placed into a temperature-controlled incubator set to 24°C with 12-h light–dark cycle to study fly activity during 3 consecutive days. Fly activity and responsiveness to startle vibration stimuli is monitored as described for the startle-induced locomotion assay. In brief, fly activity is filmed continuously with a webcam (Logitech). Custom MATLAB software was used to deliver a 5 × 200 ms vibration pulse to the flies every hour across the experiment duration. The same software package²⁷  was used to analyze sleep duration metrics and fly responsiveness to the periodic vibration stimuli. Sleep was determined as 5 min or more without activity,²⁸  allowing for cumulative sleep bouts to be tallied across multiple days and nights. The responsiveness to vibration stimuli was characterized as an average velocity curve, normalized to baseline locomotion.²⁷ 

For the startle assay, humidified isoflurane gas was delivered to the sealed chamber by an isoflurane evaporator (Mediquip, Australia) under a constant flow of 2.5 l/min, and gas was vacuumed out of the chamber to ensure a constant gas flow and pressure. Isoflurane should equilibrate within fly tissue in less than 1 min.²⁹  The concentration delivered into the behavioral chamber from the evaporator was verified using gas chromatography as described previously.³ 

Isoflurane concentrations in saline at the NMJ was determined by gas chromatographic headspace analysis (PerkinElmer Clarus 680 GC-FID; Perkin Elmer, USA), performed as described previously.³⁰  In brief, 1 ml of perfusate was placed into 10 ml headspace vials and sealed immediately with lids containing a polytetrafluoroethylene septum. Samples were heated to 60°C, and 1 ml of headspace gas was injected into the gas chromatograph via an autosampler. All samples were analyzed in duplicate. The concentration of isoflurane was determined by comparing to a saturated isoflurane solution.

Sharp intracellular recordings were made from the larval NMJ as described previously.³¹  Wandering third-instar larvae were dissected in ice-cold Schneider insect medium (Sigma-Aldrich) and pinned onto glass dissection plates to expose the body wall muscles. Intracellular electrodes (50 to 80 MΩ) were filled with a 2:1 mixture of 3 M potassium chloride and 3 M potassium acetate. Recordings were conducted at room temperature in HL3 hemolymph-like solution^32,33  with [Ca²⁺] = 0.7 mM and [Mg²⁺] = 20 mM, from muscle 6, abdominal segment A3. Analysis was performed on recordings with membrane potentials lower than −65 mV. Data for calcium dependence of transmitter release included experiments performed in [Ca²⁺] of 0.5 and 0.6 mM.

Signals from intracellular recordings were amplified using an Axoclamp2B amplifier (Molecular Devices, USA) in bridge mode. Signals were captured and stored using the Chart software (v.5.5.4; 2-kHz sampling rate) and hardware incorporated with the PowerLab/4s data acquisition system (ADInstruments, Australia).

Isoflurane solutions were prepared as described previously.³⁰  HL3 saline was placed into a 20 ml vial, and a set volume of isoflurane was added to the saline using a Hamilton syringe (Hamilton Company) and immediately vortexed for 1 min. Saline was placed into a syringe and perfused onto the dissected larvae using a syringe pump (KD Scientific, USA) at a rate of approximately 1 ml/min with Teflon tubing (2 mm inner diameter; Gecko Optical, Australia). Recordings begin with 3 min of baseline excitatory junctional potentials (EJPs), stimulated at a frequency of 1 Hz. Isoflurane perfusion is then initiated and continues until the recording has lasted 10 min or the muscles start contracting³⁰  and the impalement is lost, whichever occurs first.

All statistical comparisons were performed using Prism (GraphPad, USA). It was not feasible to blind experimenters to fly genotypes. Sample sizes for all assays were selected based on previous experience.³  The number of reported animals was the same as the number tested for all experiments.

Data were converted to express the number of flies in the bottom of the tube as a proportion and tested for normality using the Lilliefors test.³⁴  The mean time for 50% of the flies to fall to the bottom was compared across genotypes. Normally distributed data was tested for significance (P < 0.05) by two-tailed t test comparing means. Otherwise, for the nonparametric data, a Mann–Whitney test for unpaired comparisons was used.

Velocity data for the flies were normalized to the behavior at 0 vol% isoflurane to enable comparisons across experiments. Data were fit by nonlinear regression (Prism 6; GraphPad) to estimate an EC₅₀ and standard error of the estimate. The following logistic equation was used for curve fitting: Y = Min + (Max − Min) / (1 + 10^{((LogEC50 − X) × HillSlope)}), with Y representing the behavioral response and X the gas concentration. The EC₅₀ represents the isoflurane concentration at which the behavior is half-maximal and was calculated using the best-fit parameters for each genotype to give the lowest standard error of the estimate. This was obtained by constraining the maximum value to 1 and the minimum value to 0. Separate curves were compared for significant differences by simultaneous curve fitting, where all data are fit together while constraining the EC₅₀ to be shared,³⁵  with significance indicating rejection of the null hypothesis that both datasets share a common EC₅₀ (P < 0.05). EC₅₀ data represent isoflurane volume % atmospheres mean ± standard error of the estimate with 95% CIs reported.

Data were tested for normality using the Lilliefors test.³⁴  A two-tailed, unpaired t test was used to compare experimental and control strains with normal distributions. Otherwise, for nonparametric comparisons, the Mann–Whitney test was used for unpaired comparisons, with significance thresholds set at P value less than 0.05.

Recordings were processed in Axograph X (version 1.5.4; Axon Instruments, Inc., USA) to obtain the amplitude and baseline offset of EJPs and spontaneous miniature EJPs (mEJP). To process recordings, a template function was created following instructions in the Axograph manual. This template was used to process all recordings. The signal-to-noise ratio was set to 3.5, and this value was determined by comparing Axograph measures to those obtained with manual measures obtained in Chart (ADInstruments, cursor comments function). Values obtained from the two different analyses were not significantly different. Quantal content was calculated by dividing the mean EJP amplitude by the mean mEJP amplitude. Evoked responses were corrected for nonlinear summation³⁶  before calculations. Tests for significant differences between control and isoflurane perfusion were conducted using one-way ANOVA with Dunnett multiple comparisons test. To test for significant differences between genotypes under isoflurane perfusion, two-way ANOVA with Sidak multiple comparison test was used. Differences were considered significant at P value less than 0.05. To analyze the calcium dependence of transmitter release, quantal content and calcium concentrations were plotted on log–log scales, with slopes of regression fitted to these points.

Evidence from both nematodes¹⁷  and mammalian cell lines¹⁸  suggests that a target site for volatile general anesthetics might involve the transmitter release machinery, in particular, the H3 domain of the protein syntaxin1A (fig. 1A). We investigated the relevance of syntaxin1A to anesthetic sensitivity in the fruit fly, D. melanogaster. Two mutant strains were acquired in which regions of the H3 domain of syntaxin1A were modified: syxH3-C,²³  which contains a deletion of 14 amino acids in the C-terminal region of the H3 domain, and syxKARRAA,²⁴  a strain with two amino acid substitutions in the H3 domain (fig. 1, A and B). These strains therefore express both mutant syntaxin1A protein and wild-type syntaxin1A protein. To verify the expression of the syntaxin1A mutations in these fly strains, we performed Western blot in adult animals (fig. 1A). syxH3-C mutants expressed wild-type syntaxin1A protein (33 kDa) and a syntaxin1A protein of smaller size, consistent with predicted size of the deletion protein. In previous C. elegans and cell culture studies, coexpression of mutant syntaxin1A containing H3 deletions resulted in anesthesia resistance,^17,18  whereas amino acid substitutions in the H3 domain produced anesthesia hypersensitivity in C. elegans.¹⁷ 

To investigate whether syntaxin1A mutations might modulate general anesthesia in Drosophila, we tested flies in the startle-induced locomotion assay³  (fig. 1C). Two anesthesia endpoints were derived from this assay: baseline locomotion and startle-induced locomotion (fig. 1D). In air and under low concentrations of isoflurane (>0.25 vol %), wild-type (CS) flies walked significantly faster after the vibration stimulus compared with before the stimulus (P < 0.0001, t test, see Supplemental Digital Content 1, fig. 2, https://links.lww.com/ALN/B139, for raw data baseline and startle behavioral responses in wild-type flies). This significant increase in locomotion after the vibration stimulus is lost at 0.37 vol% isoflurane (see Supplemental Digital Content 1, fig. 2, https://links.lww.com/ALN/B139). The startle-induced locomotion endpoint EC₅₀ is 0.30 ± 0.005 (95% CI, 0.29 to 0.31; n = 60 flies) and is more sensitive to the effects of isoflurane than baseline locomotion (EC₅₀ = 0.35 ± 0.01; 95% CI, 0.32 to 0.38; n = 60 flies). The EC₅₀ for startle-induced locomotion is significantly lower (P < 0.01; fig. 1D), indicating that this form of behavioral responsiveness is a more sensitive anesthesia endpoint than baseline locomotion.

We used the startle-induced locomotion endpoint to characterize the isoflurane sensitivity of the syntaxin1A-mutant strains syxH3-C and syxKARRAA. We found that syxH3-C was resistant to isoflurane compared with the common genetic background (syx²²⁹/+, a syntaxin1A null mutation³⁷ ) (EC₅₀ = 0.33 ± 0.01; 95% CI, 0.30 to 0.37; n = 40; P < 0.001; fig. 1E). In contrast, syxKARRAA was found to be hypersensitive compared in the same genetic background (EC₅₀ = 0.24 ± 0.005; 95% CI, 0.22 to 0.26; n = 40; P < 0.001; fig. 1E).

To exclude any effects of the syx²²⁹/+ genetic background, we outcrossed the two different syntaxin1A-mutant strains to a common wild-type genetic background (isoCJ1) and tested these flies for isoflurane sensitivity in the startle-induced locomotion assay. We found a consistency of general anesthetic effects in the isoCJ1 background for the startle endpoint, with syxH3-C still showing resistance to isoflurane (EC₅₀ = 0.28 ± 0.004; 95% CI, 0.27 to 0.30; n = 40; P < 0.0001; fig. 1F) and syxKARRAA still being hypersensitive compared with the isoCJ1 control strain (EC₅₀ = 0.24 ± 0.008; 95% CI, 0.20 to 0.27; n = 40; P < 0.01; fig. 1F). This confirms that the isoflurane resistance and hypersensitivity phenotypes observed in these strains are respectively attributable to the coexpressed syntaxin1A mutations, syxH3-C and syxKARRAA. When testing the overall responsiveness of either strain to the startle stimulus, we find both mutants are strongly responsive to the vibration stimulus (fig. 1, G and H). The isoflurane hypersensitive strain, syxKARRAA, is actually even more responsive than the resistant strain syxH3-C (P < 0.01, t test; fig. 1H). This excludes the possibility that the resistance or hypersensitivity effects stem from different locomotion or responsiveness levels in the absence of the drug. Rather, these are anesthesia-specific effects.

We next addressed whether the syntaxin1A-mediated resistance and hypersensitivity effects were specific to the startle endpoint (thereby indicating behavioral responsiveness circuits³⁸  as possible targets), or if these same anesthesia phenotypes were present for baseline locomotion. We found syxH3-C was also resistant to isoflurane for baseline locomotion compared with isoCJ1 (EC₅₀ = 0.35 ± 0.01; 95% CI, 0.32 to 0.39; n = 40; P < 0.001; fig. 2A) and syxKARRAA was also hypersensitive for baseline locomotion compared in the isoCJ1 background (EC₅₀ = 0.30 ± 0.01; 95% CI, 0.27 to 0.33; n = 40; P < 0.01; fig. 2A). This suggests that the syntaxin1A effects on isoflurane sensitivity are not restricted to a circuit relating only to behavioral responsiveness.

To further explore baseline behavioral capabilities of flies under general anesthesia, we devised a fly coordination assay (fig. 2B). This simple general anesthesia assay assessed a fly’s capability to display negative geotaxis and move upwards against gravity, displaying climbing behavior.²⁵  This assay again confirmed our anesthesia phenotypes found for the other endpoints: in the isoCJ1 background, flies expressing the syxH3-C protein were resistant to isoflurane (mean time for 50% to fall to bottom: 113 ± 71 s SD) and halothane (mean time for 50% to fall to bottom: 116 ± 85 s SD), with these flies taking significantly longer to fall to the bottom of the tube (P < 0.05, t test; fig. 2, C and D). In contrast, syxKARRAA are hypersensitive to isoflurane compared with the isoCJ1 control (mean time for 50% to fall to bottom: 26 ± 13 s SD, P < 0.05, t test; fig. 2C). syxKARRAA shows no change in sensitivity to the more potent volatile anesthetic, halothane, compared with controls (P = 0.78, t test; fig. 2D). Taken together, our general anesthesia results suggest that the syntaxin1A mutations modulate different behavioral capabilities under general anesthesia rather than any specific behavior.

We wondered whether the opposing anesthesia phenotypes in syxH3-C and syxKARRAA reflected indirect effects on the sleep circuitry. In previous work, we found a negative correlation between sleep and general anesthesia: flies that were resistant to isoflurane slept less during several days and nights, and flies that were hypersensitive to isoflurane slept more.³  This strong correlation was found in strains that carried specific genetic manipulations directly targeting particular circuits within the fly brain, including the sleep–wake circuitry. We questioned whether the syntaxin1A mutants would also show a negative correlation between sleep and general anesthesia because these strains expressed syntaxin1A mutations not only in the sleep–wake circuitry but also in all neurons. Consistent with the correlation,³  we found that syxH3-C which is resistant to isoflurane slept less than the control strain (isoCJ1) (see Supplemental Digital Content 1, fig. 3, https://links.lww.com/ALN/B139, for sleep profiles of isoCJ1 and syxH3-C and quantification of time spent asleep). Unexpectedly, sleep duration was not increased in the hypersensitive syxKARRAA strain compared with controls (see Supplemental Digital Content 1, fig. 3, https://links.lww.com/ALN/B139, for sleep profile of syxKARRAA and quantification of time spent asleep). This suggests that syntaxin1A-mediated effects on isoflurane sensitivity do not fit the model predicted by sleep–wake circuits on general anesthesia although the resistance effects still match the prediction.

Both syxH3-C and syxKARRAA modulate isoflurane sensitivity in a wild-type background (isoCJ1), suggesting a gain of function (or neomorphic) effect. We determined whether one copy of the mutant syntaxin1A protein was therefore sufficient to produce resistance or hypersensitivity to isoflurane in adult flies. After verifying that these flies still expressed the syntaxin1A deletion protein through Western blot (fig. 3A), we found that flies expressing only one copy of syxH3-C were as resistant to isoflurane as the homozygous strain (EC₅₀ = 0.28 ± 0.007; 95% CI, 0.26 to 0.30; n = 40; P < 0.001; fig. 3B). Interestingly, syxKARRAA is also dominant, with heterozygotes remaining as hypersensitive compared with genetic controls (EC₅₀ = 0.25 ± 0.01; 95% CI, 0.24 to 0.26; n = 40; P < 0.01; fig. 3B). These results show that both syntaxin1A lesions are dominant and that half the amount of either mutant protein is sufficient to confer either opposing phenotype.

Given both syntaxin1A lesions act dominantly, this raises the question: which general anesthesia phenotype will predominate if both resistant and hypersensitive syntaxin1A alleles are combined? We tested flies carrying both syntaxin1A mutations as a transheterozygote (on an isoCJ1 wild-type background). We found that flies carrying both syxH3-C and syxKARRAA are resistant, showing no significant difference compared with syxH3-C/+ (EC₅₀ = 0.28 ± 0.008; 95% CI, 0.25 to 0.30; n = 40; P = 0.719; fig. 3B).

Larvae are an immature stage of the flies’ lifecycle, with a different central nervous system compared with the adult fly^39,40  and probably lacking the sleep-promoting circuits found in adults.⁴¹  We questioned whether the syntaxin1A mutations would also confer general anesthetic resistance and hypersensitivity as larvae. The syntaxin1A deletion protein is expressed in larvae (fig. 4A), so effects at this early stage might indicate a common mechanism of action for general anesthetics that does not encompass a developed sleep circuitry.^3,21,42 

We devised a coordination assay to test larval behavior under general anesthesia (fig. 4A, and see Materials and Methods). Under increasing concentrations of isoflurane, the proportion of wild-type (isoCJ1) larvae that can display coordinated movement decreased significantly (P < 0.01, t test; fig. 4B). We found that syxH3-C were also resistant as larvae: a greater proportion of syxH3-C larvae (on the isoCJ1 background) were capable of coordinated movement compared with the isoCJ1 controls (P < 0.01, t test; fig. 4C). Also, consistent with our observations in adult flies, syxKARRAA larvae remain hypersensitive to isoflurane compared with isoCJ1 controls (P < 0.01, t test; fig. 4C). These experiments show that the syntaxin1A-mediated effects are independent of the life stage, and therefore, do not require adult-specific circuitry, such as sleep–wake promoting pathways.^3,21,42  Importantly, resistance and hypersensitivity also were evident when these syntaxin1A mutants were exposed to another volatile anesthetic, halothane (P < 0.05, t test; fig. 4D). Thus, our behavioral analysis shows that syntaxin1A mutations produce consistent anesthesia phenotypes across the Drosophila life cycle in animals with vastly different brains and sleep requirements. This suggests that the syntaxin1A-mediated effects on isoflurane sensitivity are unlikely to involve interactions with the sleep–wake pathway that has been identified in adult flies.^3,21,42  Instead, a process common to adult and larval nervous systems must be involved.

To investigate whether the decrease in larval coordination under isoflurane reflected a change in transmitter release, we examined synaptic transmission at the larval NMJ. The larval NMJ is a model glutamatergic synapse, sharing many similarities with mammalian glutamatergic synapses.⁴³  To characterize the general anesthetic effects on synaptic transmission, we recorded mEJPs and EJPs in wild-type NMJs (fig. 5, A and B) before and during exposure to isoflurane (fig. 5, C and D). A dose–response characterization of the effects of isoflurane on synaptic transmission in Drosophila larvae has been previously reported, using a slightly different recording technique.³⁰  We focused our investigations on isoflurane concentrations around the reported EC₅₀ for larval locomotion: 0.4 vol% isoflurane for wild-type (CS) larvae, which closely corresponded with the EC₅₀ of 0.17 mM isoflurane that decreased EJP amplitudes.³⁰ 

At wild-type NMJs, we found isoflurane significantly decreases the amplitude of evoked responses (P < 0.05, one-way ANOVA; fig. 5, C and E; see Supplemental Digital Content 1, fig. 4, https://links.lww.com/ALN/B139, for electrophysiology response measures over time). In contrast, isoflurane does not affect miniature endplate potential amplitudes (P = 0.06, one-way ANOVA; fig. 5, D and F). Quantal content (average number of vesicles released per action potential) was significantly decreased by isoflurane (P < 0.05, one-way ANOVA; fig. 5G). This suggests that a decrease in transmitter release could underlie the coordination defects observed in wild-type larvae under isoflurane exposure (fig. 4, A and B).

We showed earlier that syxH3-C larvae were resistant to the effects of general anesthetics, being capable of coordinated movement at isoflurane concentrations where control larvae had stopped moving (fig. 4, C and D). To investigate whether these resistance effects in syxH3-C reflected altered transmitter release, we analyzed quantal content in syxH3-C-mutant and control synapses. As in wild-type CS larvae, quantal content was significantly decreased in isoCJ1 after isoflurane perfusion compared with before perfusion (P < 0.05, one-way ANOVA; fig. 5G). In contrast, quantal content in syxH3-C was not significantly decreased at the same perfusion time point (P = 0.16, one-way ANOVA; fig. 5G) and is significantly higher than in isoCJ1 (P < 0.05, two-way ANOVA; fig. 5G). In addition, in the presence of isoflurane, the percentage of failures in nerve-evoked transmitter release was lower at syxH3-C NMJs compared with control NMJs (P < 0.05, t test; fig. 6, A and B). Accordingly, we found no change in quantal content in syxH3-C with isoflurane, whereas transmitter release has decreased in isoCJ1 (fig. 5G). These observations suggest that at syxH3-C NMJs, transmitter release is partially protected from the inhibitory effects of isoflurane.

Evoked transmitter release is calcium dependent. We explored whether isoflurane impacted transmitter release by changing the calcium dependence of transmitter release (calcium sensitivity and/or cooperativity) at wild-type and syxH3-C synapses. By constructing log–log plots of quantal content versus the extracellular calcium concentration (fig. 6, C–F), we investigated whether changes in transmitter release were produced through changes in calcium dependence.^44,45  Changes in calcium cooperativity (linear slope of the log–log plot) indicate changes in the number of calcium ions binding with the calcium sensor(s) to evoke transmitter release, whereas changes in the calcium sensitivity (left or right shift of the line) indicate changes in the amount of calcium required to evoke neurotransmitter release.^44,45  Before isoflurane perfusion, we found that calcium dependence was not different at isoCJ1 synapses and syxH3-C synapses (P = 0.294; fig. 6C), indicating similar transmitter release kinematics in both strains in the absence of isoflurane. After isoflurane perfusion, the calcium cooperativity increased in both strains (P < 0.05; fig. 6, D and E), as evident from changes in the slope of the relation, although to a greater extent in isoCJ1 (7.6; fig. 6D) compared with syxH3-C (4.4; fig. 6E). The right shift of the lines under isoflurane (fig. 6, D–F) indicated changes in calcium sensitivity, where more calcium was required to evoke similar levels of transmitter release than before isoflurane perfusion. These results confirm a presynaptic mechanism for isoflurane, indicating that with isoflurane, the dependence on calcium is greater to achieve a similar level of transmitter release (P < 0.05; fig. 6F). However, it is also clear that the calcium dependence of transmitter release is less in syxH3-C under isoflurane perfusion (fig. 6F), even though calcium dependence is similar without isoflurane (fig. 6C). This suggests that the mutant syntaxin1A protein is specifically interfering with the action of isoflurane rather than merely increasing synaptic efficacy in general.

Our understanding of general anesthesia has mirrored our evolving appreciation of how the brain works, so it is not entirely surprising that the mechanism of general anesthesia remains somewhat of a mystery. Early work on the structure and excitability of neurons pointed to a nonspecific role for these drugs disrupting cellular excitability by interfering with the lipid membrane of neurons.⁴⁶  A subsequent elucidation of the roles of proteins embedded in the membrane, notably ion channels, led to the realization that inhibitory GABA receptors are the most likely target for these drugs.^2,11  More recent studies have highlighted the involvement of the sleep–wake circuitry in the mammalian and fly brain, showing that endogenous sleep pathways are disinhibited by some general anesthetics.^2–6,8  Moving beyond dedicated circuits, a new systems-level view of whole-brain dynamics now proposes that these drugs disrupt long-range communication, coherence, and integration across the brain.^47–49 

We have proposed recently that general anesthesia may primarily be a two-step process, whereby sleep-promoting pathways are activated first at low drug concentrations (thereby producing a “gentle” loss of consciousness for GABAergic drugs such as isoflurane), and synaptic mechanisms are attenuated at the higher drug concentrations required for surgery.⁷  In a previous report, we have uncovered a sleep–wake circuit supporting the first step in this process.³  In this study, we examine syntaxin1A mutants in Drosophila to explore the second step, synaptic release mechanisms.

We characterized the general anesthesia phenotypes of two Drosophila strains containing modifications to the H3 domain of syntaxin1A: syxH3-C, which expresses a syntaxin1A protein with 14 amino acids deleted within the H3 domain,²³  and syxKARRAA, a strain with two amino acid substitutions (lysine to alanine) in the H3 domain adjoining the transmembrane portion of the protein.²⁴  We found that syxH3-C is resistant to isoflurane and syxKARRAA is hypersensitive to isoflurane across a variety of behavioral endpoints in both adults and larvae (figs. 1–4). These findings provide two important conclusions. First, neomorphic modifications in one synaptic protein can produce both hypersensitivity and resistance to isoflurane. Interestingly, anesthesia effects were not consistent with another volatile anesthetic, halothane, with syxH3-C being resistant but syxKARRAA showing hypersensitivity only as an immature larva. This suggests that the syxH3-C mutation more strongly affects anesthetic sensitivity than syxKARRAA. Indeed, when both mutations are placed together as a transheterozyote, the prevailing anesthesia phenotype is resistance (fig. 3B). This suggests a central role for syntaxin1A function in mediating general anesthesia. Second, the conservation of resistance and hypersensitivity effects for isoflurane across behavioral endpoints and life stages suggest that these are general mechanisms that are not necessarily linked to any specific brain circuitry, such as arousal pathways in the adult brain.

What could account for the general anesthesia phenotypes we have uncovered in the syntaxin1A mutants? For either strain, it is unlikely that the differences in anesthetic sensitivities we have found stem from altered anesthetic uptake, as behaviors were assayed after the anesthetics had equilibrated for several minutes. More specifically, syxKARRAA was created to study the effects of electrostatic interactions in syntaxin1A clustering. Changing two positive lysine residues to neutral alanine within the juxtamembrane region of syntaxin1A abolishes the ability of the phosphoinositide PI(3,4,5)P₃ to cluster syntaxin1A.²⁴  This lack of syntaxin1A clustering compromises transmitter release, with evoked release significantly decreased in syxKARRAA compared with control.²⁴  A decrease in transmitter release is therefore a likely explanation for the general anesthetic hypersensitivity in syxKARRAA. Interestingly, the anesthesia phenotype of syxKARRAA is dominant, suggesting endogenous syntaxin1A cannot compensate for the compromised transmitter release, which is consistent with the electrophysiology data.²⁴ 

In contrast, the general anesthesia resistance effects of syxH3-C are likely to stem from altered interactions with soluble NSF attachment protein receptor (SNARE)–binding partners. Indeed, syxH3-C was originally created to identify putative syntaxin1A binding partners. Wu et al.²³  showed that this syntaxin1A deletion spans the calcium effector domain and synprint binding site and that syxH3-C has defects in binding other SNARE proteins, yet the core complex can still form in these mutants. The fidelity of calcium-triggered transmitter release in this mutant was severely affected when assayed as embryos in a null homozygous background (these animals do not survive beyond the embryonic stage).²³  In contrast, we see that when the mutant is coexpressed together with wild-type syntaxin1A, calcium-triggered release is normal (fig. 6C), and instead a phenotype only emerges under isoflurane perfusion (fig. 6F). This suggests that the syntaxin1A deletion protein confers an isoflurane resistance–promoting effect on the drug’s target mechanism, specifically. Exactly how this resistance effect is mediated requires further elucidation.

Our work suggests that syntaxin1A represents a conserved target of general anesthetics across animals. Interestingly, the resistance effects from syxH3-C in Drosophila are modest compared with the original C. elegans md130 truncation.¹⁷  One reason for this may relate to the lack of conserved sleep-promoting circuitry in adult nematodes.⁷  If the entry into general anesthesia involves activation of sleep pathways followed by global attenuation of synaptic release,⁷  animals that lack sleep circuitry will unmask other relevant targets of general anesthetics. Consequently, mutations in synaptic release proteins will produce greater effects on general anesthetic sensitivity in animals that do not sleep.

There are only three previous reports on the effects of general anesthetics on transmitter release properties at the larval NMJ, one with halothane,⁵⁰  and two isoflurane studies.^30,51  For wild-type fly larvae, consistent with the previous reports, we found that evoked responses decrease after isoflurane exposure (fig. 5E), whereas the amplitude of miniature endplate potentials is unaffected (fig. 5F). We report here for the first time in Drosophila that isoflurane decreases quantal content (fig. 5G). Interestingly, we found preserved quantal content in syxH3-C after isoflurane exposure (fig. 5G). The lack of isoflurane inhibition of transmitter release may account for why syxH3-C larvae are capable of coordinated movement under general anesthesia compared with genetic controls (fig. 4, C and D), but also why syxH3-C adult flies are resistant to isoflurane across a variety of behavioral endpoints (fig. 1–3).

Previous work by others indicates that isoflurane decreases transmitter release by decreasing release probability (the chance of a vesicle undergoing exocytosis after an action potential).^30,52  Release probability is calcium dependent.^53,54  Therefore, to further understand the changes in quantal content with isoflurane, we investigated the calcium dependence of transmitter release (fig. 6). At control synapses, our results indicate that transmitter release is more calcium dependent after isoflurane exposure than before exposure (fig. 6, C–F). syxH3-C synapses also showed this calcium dependency, however, to a lesser extent under isoflurane perfusion than control synapses. Because syxH3-C fails to bind the calcium sensor synaptotagmin⁵⁵  while still forming part of the core SNARE complex,²³  it is possible the mutant syntaxin1A protein might be less dependent on calcium for transmitter release under isoflurane. Alternatively, the deletion protein might be accomplishing another SNARE-related function such as preserving SNARE clustering^56,57  on the plasma membrane under isoflurane anesthesia.

One reason why syxH3-C is less dependent on calcium may lie with how isoflurane interacts with the transmitter release machinery. A recent study suggests that synaptotagmin1 and syntaxin1A together form binding pockets for isoflurane.⁵⁸  Synapses where the synaptotagmin and syntaxin interaction has been modified display altered calcium cooperativity^59,60  even in the absence of general anesthetics. Our results therefore suggest that volatile anesthetics such as isoflurane may partially disable interactions between syntaxin1A and the calcium sensor(s) although we cannot exclude the possibility that other syntaxin1A-interacting proteins, such as UNC13,⁶¹  may also be involved.

Our combined results point to a mechanism of general anesthesia that is distinct from the better-understood GABAergic sleep pathway.^2–4,6,8 We propose that general anesthetics such as isoflurane and halothane indeed target the sleep pathway as proposed,^3,5  thereby producing unconsciousness in all animals that sleep, but that these drugs also target synaptic release mechanisms in general, as is now evident from the Drosophila model as well as a number of other studies in simpler systems.^16–19  Anesthetic effects on synaptic release are likely to impair information processing across the brain, even if the effect at each individual synapse may be small.

This work was supported by an Australian Research Council (Canberra, Australia) fellowship (grant nos. FT100100725 and DP1093968 to Dr. van Swinderen); by the Australian Research Council (grant no. LEI130100078); and grants from the Griffith University, Gold Coast, Queensland, Australia (to Dr. Karunanithi).

The authors declare no competing interests.