Behavioral and neurophysiological effects of electrical stunning on zebrafish larvae

Abstract

Two methods dominate the way that zebrafish larvae are euthanized after experimental procedures: anesthetic overdose and rapid cooling. Although MS-222 is easy to apply, this anesthetic takes about a minute to act and fish show aversive reactions and interindividual differences, limiting its reliability. Rapid cooling kills larvae after several hours and is not listed as an approved method in the relevant European Union directive. Electrical stunning is a promising alternative euthanasia method for zebrafish but has not yet been fully established. Here we characterize both behavioral and neurophysiological effects of electrical stunning in 4-day-old zebrafish larvae. We identified the electric field characteristics and stimulus duration (50 V/cm alternating current for 32 s) that reliably euthanizes free-swimming larvae and agarose-embedded larvae with an easy-to-implement protocol. Behavioral analysis and calcium neurophysiology show that larvae lose consciousness and stop responding to touch and visual stimuli very quickly (<1 s). Electrically stunned larvae no longer show coordinated brain activity. Their brains instead undergo a series of concerted whole-brain calcium waves over the course of many minutes before the cessation of all brain signals. Consistent with the need to implement the 3R at all stages of animal experimentation, the rapid and reliable euthanasia achieved by electrical stunning has potential for refinement of the welfare of more than 5 million zebrafish used annually in biomedical research worldwide.

Fig. 1. Developing a protocol for humane euthanasia via electrical stunning.

a, An illustration of different euthanasia methods for zebrafish. b, The electrical stunning chamber with cuboid net container allows imaging of behavior. c, The brain activity of immobilized animals can be measured via 2-photon (2p) calcium imaging during visual stimulation and electrical stunning.

Fig. 2. Electric AC fields are well suited for euthanasia of larvae.

a, A schematic of the experimental setup. b, The protocol for assessing mortality. x is the exposure duration that was systematically varied over trials. c, The logical matrix defining under which condition within the experimental procedure a larva was considered dead. d, Mortality rates of different voltage types for different exposure durations at a voltage gradient of 50 V/cm. The asterisks indicate significances by z-test for proportions followed by Bonferroni correction between exposure durations for the same voltage type (colored asterisks) and between voltage types for the same exposure duration (black asterisks); *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The envelopes show 95% Wilson confidence intervals. n = 400 larvae, n = 4 shock exposure trials with 5 larvae each. e, The time to loss of equilibrium for different voltage types (animal numbers are provided next to the hash symbols) For comparison, the time to loss of equilibrium for MS-222 (336 mg/l) is shown in gray. f, Morphological effects of electrical stunning at 50 V/cm, 32 s, directly after exposure. g, Mortality rates of 16 s and 32 s exposure duration for different voltage gradients at AC voltage. The asterisks indicate significance by z-test for proportions followed by Bonferroni correction between voltage gradients and exposure durations; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The envelopes show 95% Wilson confidence intervals. n = 240, n = 4 shock exposures with 5 larvae each. h, The mean power density and current density of different voltage types for different exposure durations at 50 V/cm. Power and current densities were calculated on the basis of measured currents during the experiment. The dashed lines indicate the expected power density without joule heating (light blue, DC; black, AC and PDC). Light blue y axis on the right is for DC and the black y axis on the right for AC and PDC experiments. The envelopes show s.d.

Fig. 3. The AC field drives direct muscle contractions.

a, A schematic of the experimental setup. Agarose surrounding the larva’s tail has been removed. b, Left, a schematic of larva illustrating tail displacement during electrical stunning. The orange arrows indicate the skeleton used for computing the total tail angle. Right, the total tail angle during exposure. The arrows indicate a spontaneous swim bout and the exposure onset. The inset shows the tail angle during exposure, which perfectly matched a 60 Hz sinusoidal wave. c, The length of the larva during (left) and before (right) electrical stunning. d, Top, the lateral tail displacement profile during electrical stunning. Bottom: the lateral tail displacement profile during the spontaneous swim bout. e, Top, the power spectrum during electrical stunning. Bottom, the power spectrum during the spontaneous swim bout. n = 2.

Fig. 4. Absence of sensory-evoked neural activity during and after electrical stunning.

a, A schematic of the experimental setup positioned under the two-photon microscope. The yellow dot indicates the imaged brain area (optic tectum). b, Mean calcium signals for baseline, shock and recovery recording for all fish (one trace per fish). The light-blue bar indicates exposure duration. The LED stimulus trace indicates when a visual stimulus was used or not. The arrows highlight the occurrence of slowly propagating calcium waves occurring after electrical stunning. c, The mean PSD for baseline, shock and recovery recording (green, red and dark blue as used in b). The envelopes show s.d. The dashed line indicates the visual stimulus main frequency of 0.1667 Hz. The asterisks indicate periods with significant differences to baseline recording (two-way repeated-measures ANOVA followed by Tukey’s HSD test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). d, The inset of c for baseline, shock and recovery recordings. e, Example calcium traces of neurons (black) (anti-)correlated with the stimulus (gray), and mean correlation traces (green and dark blue) of 20% best correlated ROIs for baseline and recovery recording. f, ROI cross-correlation matrices for baseline and recovery recording (top and bottom). g, The mean ACF for baseline and recovery recording (top and bottom). The envelopes show s.d. Gray shows the ACF for the stimulus (regressor) itself. h, A histogram of absolute ROI cross-correlation coefficients (Abs(r)) for baseline and recovery recordings. Hash symbols indicate the overall number of neurons (ROIs) used. P value derived from Levene test. n = 7.