Zebrafish of the Zebrafish International Resource Center WIC (ZIRC WIC) lineage were bred and housed at Philadelphia College of Osteopathic Medicine (PCOM) laboratory animal research facilities (Philadelphia, PA, United States). Animals were housed in 3 L tanks on a 14:10 h light dark cycle. Water was recirculated and conditioned using a Pentair G-Hab (Cary, NC, United States) or Tecniplast Blue Active (Milan, Italy) aquatic habitat. Conductivity was maintained between 500 and 1,000 mΩ–1, pH between 7.0 and 7.5, and temperature between 27 and 28°C. Zebrafish were fed GEMMA Micro (Westbrook, ME, United States). Animals used for behavioral experiments were kept as breeders or euthanized humanely. All protocols were carried out in accordance with the regulations set by the PCOM and St. Joseph’s University Institutional Animal Care and Use Committees (IACUC).

Our pilot studies using the single-day PTZ-induced seizure model first described by Baraban et al. (2005) did not show significant effects on y-maze or social behavior (Supplementary Figure 1). Therefore, we modified the previously published seizure model by increasing seizure induction sessions from one to three (once daily for 3 days), and decreasing the concentration of PTZ (MilliporeSigma, Burlington, MA, United States) to compensate for the increased amount of time the fish were exposed to PTZ. Specifically, at 5 days post-fertilization (dpf), zebrafish were randomly divided into three groups: untouched controls (UC), handling controls (HC), and seizure (PTZ group) and kept in 250 mL beakers with aquarium system water. HC and PTZ fish were pipetted individually into 24-well plates with 500 uL tank water, where they acclimated for 20 min while UC group remained in the incubator undisturbed. Temperature within the experimental arena was kept at 26–28^oC using a heating pad under the arena. Next, 500 uL of clean tank water was added to the HC group and 500 uL of 10 mM PTZ mix with aquarium water was added to the PTZ group (final concentration of 5 mM PTZ). Behavior of HC and PTZ fish was recorded using EthoVisionXT software with a GigE camera (Noldus, Leesburg, VA, United States) for 40 min. Fish were returned to their home beakers after another 40 min and kept in a 28^oC incubator overnight. This procedure was repeated on 6 and 7 dpf. During PTZ exposure, zebrafish showed Stage 2 seizure behavior typical of what has been reported in single day seizure models. This included circular “whirlpool” swimming (Baraban et al., 2005; Bandara et al., 2020), increased distance moved, increased overall swim velocity, and increased time spent swimming at high speed (≥2 cm/s) (Winter et al., 2017), compared to HC fish not exposed to PTZ (Supplementary Figure 2). Fish that did not show Stage 2 seizure behavior at 5 dpf (approximately 1–2%) were removed from the study. All three groups of zebrafish were transferred to the larger rack system tanks on 8 dpf by gently pouring the entire contents of the beaker (water and fish) into the tank to avoid additional handling. Groups remained separated for identification purposes.

To assess learning and memory, we used a y-maze task, as described by Cognato Gde et al. (2012), in fish 8–14 weeks post-fertilization. The y-maze consisted of an acrylic apparatus with three arms of equal dimensions, 25 cm long, 8 cm wide, and 15 cm high. The external surface of each arm was covered with a black adhesive film to block outside stimuli. Each arm was lined with white cut-outs of squares, triangles, or circles (one shape per arm). These specific shapes were selected because zebrafish exhibit approximately the same amount of preference for each and did not show signs of aversion in previous studies (Cognato Gde et al., 2012). The y-maze apparatus was filled with 3 L of tank water, with all shapes below the waterline. An opaque black partition was used to block off one of the arms at random. One of the remaining two open arms was randomly chosen as the starting arm. The experiment consisted of three trials; acclimation (Trial 1), observation (Trial 2), and recall (Trial 3). In this task, the zebrafish acclimate to a test apparatus with two of the three arms open for exploration (Trial 1; Figure 1A). Next (Trial 2), the blockade to the third arm is removed, and the fish continued exploring for 5 min. During Trial 1, a zebrafish was placed in the starting arm and allowed to freely explore the two open arms for 20 min while acclimating to the test apparatus (Figure 1A). During trial 2, the partition was lifted, supplying novel stimuli (space and visual shape), and the fish was allowed to explore for additional 5 min (Figure 1A). At the end of trial 2, the fish was removed from the maze and transferred to a 2 L tank in a 27.5°C incubator for a 90-min inter-trial interval (ITI). During trial 3, the fish was returned to the original starting arm, with all three arms open, and given an additional 5 min to explore freely. During each trial, the fish was recorded and movement was tracked using a GigE camera and Noldus EthoVision XT software. Time spent in each arm was measured as a percent of the total time in the trial. Any fish that spent less than 15% of the total time in the novel arm in T2 were excluded from T3 final analysis (Supplementary Figure 3; total excluded = 1 UC, 3 HC, 4 PTZ).

At 3 weeks of age, HC, UC, and PTZ zebrafish were transported from PCOM to St. Joseph’s University. Fish were allowed to acclimate to their environment for 1 week, then tested for social behavior at 28–32 dpf using a shoaling assay (Way et al., 2016). The shoaling assay apparatus consisted of a plastic tank (30 mm-wide × 180 mm-long × 40 mm-deep) divided into three equal compartments with clear plastic panels. Three preference zones within the central compartment were delineated by vertical lines marked the outside surface of the tank, 20 mm from each end compartment. Five target zebrafish of the same age and lineage were added to either the left or right compartment at random. The preference zone closest to the target fish was denoted as +1. The center was preference zone 0 and the preference zone furthest from the target fish was –1. For each assay, a single 30–40 dpf zebrafish was added to preference zone 0 and allowed to acclimate to the tank for 1 min. After 1 min, fish were recorded for the following 10 min using an Apple iPad. Fish then acclimated for an additional 49 min in the testing apparatus, for a total time in-tank time of 1 h, then recorded for another 10 min. Videos were transferred to Noldus EthoVision XT software for tracking the time spent in each sub-zone by the test fish. Time spent in the +1 zone was considered a positive shoaling response or normal social behavior, while time spent in zones 0 or –1 was considered a negative shoaling response or abnormal social behavior. Shoaling baseline varied by clutch, therefore, data were normalized to the average shoaling of the UC group for each clutch at 1 min.

Freezing behavior was used as a measure of anxiety (Egan et al., 2009). Freezing was assessed using Noldus Ethovision XT behavioral tracking software to quantify cumulative time immobile, and defined as a less than 0.5% change in body pixels between frames. This was then normalized to the total time the fish was tracked, resulting in percent time immobile. Measurements were taken from 28–32 dpf zebrafish during the shoaling task and from zebrafish 12–14 week post-fertilization during the y-maze task. Data were analyzed by ANOVA on GraphPad Prism software.

To assess escape response, we used the acoustic startle response (ASR) assay described by Best et al. (2008). Larval zebrafish (12–13 dpf) acclimated in 12-well plates within a temperature and sound controlled DanioVision, Noldus, Leesburg, VA, United States, chamber for 25 min prior to exposure to the acoustic stimulus. The stimulus consisted of an approximately 83 dB tap, a stimulus that was loud enough to cause a startle response in some but not all (91.7%) of UC fish tested. The onset of the tap was controlled using EthoVision software. For the duration of the experiment, fish were recorded using a Basler acA1300-60 gm GigE camera. Pre-tone baseline movement was quantified using the first 20 s prior to the first tap for each fish during the acclimation period. The acoustic startle response for each larval zebrafish was quantified by measuring the distance moved (cm) for 1 second from the beginning of each tapping stimulus. HC and PTZ larvae responses were compared to that of UC larvae, before and after the tap, by 2-way repeated measures ANOVA.

Statistical tests were performed using GraphPad Prism software, San Diego, CA, United States, version 9.2. All data were tested for normality. Data and graphs are represented as mean ± standard error unless stated otherwise. Significant differences were defined as p < 0.05.

Zebrafish tend to explore novelty and will investigate novel 2-D shapes and colors on the walls of their tank (Cognato Gde et al., 2012). The Y-maze task takes advantage of this behavior, allowing for quantification of zebrafish learning and memory (Cognato Gde et al., 2012). Most zebrafish preferred the novel arm in Trial 2 regardless of the experimental group (Supplementary Figure 1). For trial 3, the expected response in controls is recall and preference for the novel arm (Cognato Gde et al., 2012). In 8–9 weeks old fish, the control fish did not show consistent evidence of robust recall, even at a shorter 60-min inter-trial interval (ITI) (p = 0.975; F_(2,33) = 0.025 by ANOVA; n = 13 UC, 12 HC, 11 PTZ; Figure 1B). Therefore, we next tested 12–14 week old zebrafish. We found that the UC fish demonstrated the expected strong preference for the novel arm in Trial 3, spending an average of 44 ± 3.2% of the total time of Trial 3 in the novel arm. However, the PTZ fish showed a significant deficit in recall, spending less than a third of the total time of Trial 3 in the novel arm (29.7 ± 2.1%; p = 0.004 by ANOVA, F_(2,40) = 6.223, with p = 0.002 by Dunnett’s multiple comparisons post-hoc test of UC vs. PTZ; Figure 1C). The HC group showed a slight decrease in recall that was not significantly different from the UC recall and still spent greater than one-third of trial 3 in the novel arm (36.6 ± 3.0%; with p = 0.15 by Dunnett’s multiple comparisons post hoc test of UC vs. HC; n = 18 UC, 12 HC, 13 PTZ fish). The decreased y-maze recall of the PTZ group demonstrates that seizures in larval zebrafish can have long-term effects on learning.

In addition to learning deficits, social behavior deficits are another outcome commonly linked to ELS (Lippman-Bell et al., 2013; Dal Canto et al., 2018). To determine whether larval zebrafish seizures delay or alter social behavior development, we assessed shoaling, the likelihood of fish to cluster in social groups (Ledesma and McRobert, 2008; Miller and Gerlai, 2011). If larval seizures caused a developmental delay in social behavior, any deviation from controls would be most stark at the earliest age at which zebrafish begin to show shoaling behavior. Prior studies show a significant increase in shoaling between 26 and 48 dpf (Buske and Gerlai, 2012). Thus, we tested shoaling in 28–32 dpf. After 1 min of acclimation, test fish shoaled for similar amounts of time regardless of experimental group (UC, HC, and PTZ). After 1 h, the PTZ test fish still shoaled as much as the UC fish, but the HC fish shoaled significantly less than the PTZ group. [Mixed-effects analysis showed a significant main effect of experimental group p = 0.046; F_(2,85) = 3.197; post-hoc analysis shows a significant decrease of HC at 1 h compared to PTZ at 1 h (p = 0.01) with no significant changes between any other groups or time points; Figures 2A,B]. Therefore, we concluded early-life handling, but not ELS, influenced later-life shoaling behavior.

We next assessed anxiety behaviors by measuring the freezing behavior during the shoaling and y-maze tasks, as freezing in the absence of a predator is thought to be analogous to anxiety (Stewart et al., 2012). All zebrafish showed similar levels of freezing behavior after the first minute of acclimation in the shoaling task. However, once the fish had acclimated to the shoaling tank for 1 h, only zebrafish in the HC group displayed significantly more freezing than the UC group; freezing behavior in the PTZ group did not differ significantly from naïve controls (Figure 3A). [Two-way repeated measures ANOVA showed a main effect of experimental group (F_(2,139) = 3.373, p = 0.037), with a post-hoc test showing a significant difference between HC and UC groups at the 1 h acclimation time point; p = 0.027].

To confirm that the increased freezing behavior in the HC zebrafish was indicative of a heightened stress response, we next measured cortisol levels in each group after gently netting the fish for 2–3 s. HC fish showed elevated levels of cortisol compared to both the UC and PTZ groups, with no change between the PTZ and UC groups [p = 0.008, F_(2,12) = 7.327 by ANOVA, with post hoc analysis showing a significant difference of HC vs. UC (p = 0.02) and PTZ (p = 0.004); UC vs. PTZ p = 0.3; mean = 100.0 ± 6.5 for UC, 144.1 ± 13.53 for HC, and 79.99 ± 14.94 for PTZ; n = 5 UC, 6 HC, and 4 PTZ samples with two zebrafish pooled per sample; Figure 3B]. Taken together with the y-maze data, these results indicate that early-life seizures attenuated the development of the heightened stress response after early-life stress seen in handled controls.

Elevated cortisol levels occur with both acute and chronic stress (Manuel et al., 2014). However, chronic stress can be marked by attenuated expression of one of the primary cortisol receptors, the glucocorticoid receptor (Kitraki et al., 1999; Piato et al., 2011). Therefore, to distinguish between an acute stress response and chronically increased stress, we next measured levels of glucocorticoid receptor mRNA. Unlike cortisol levels, glucocorticoid receptor expression remained similar across experimental groups at 30 dpf (Figure 3C). Thus, the elevated cortisol levels in the HC fish likely represent an acute heightened response to the mild stressor and not chronic anxiety.

By 60 dpf, the freezing behavior and the elevated cortisol response in the HC fish returned closer to UC levels, and glucocorticoid receptor mRNA expression remained similar between groups (Figures 3D–F). This return to control levels supports the idea that HC fish did not have chronic anxiety, but a heightened stress response that dissipated over time as the fish lived on the tank system, undisturbed by outside stressors.

A heightened stress response after an early-life stressor, such as handling, can represent a learned, adaptive response (Belda et al., 2008; Berry and López-Martínez, 2020); multiple animal models demonstrate that mild stressors change later behavior to protect the animal from further harm (Finsterwald and Alberini, 2014; Berry and López-Martínez, 2020). If this were the case in the handled larval zebrafish, we would expect the HC group to develop behaviors advantageous to their survival. Therefore, we compared the HC acoustic startle response (ASR) to that of the PTZ and UC groups to measure a potential advantage for escape.

Zebrafish will show a startle response to a range of stimuli (Colwill and Creton, 2011). To evoke an ASR, zebrafish were acclimated to a 24-well plate in the DanioVision chamber for 25 min, then exposed to a loud tap. The response of each fish was quantified by distance traveled in the first second after the tap. Two-way repeated-measures ANOVA revealed a significant effect of tap x experimental group [F_(2,68) = 4.826, p = 0.011], with post-hoc Sidak’s multiple comparison test showing a significant increase between baseline and first tap (p < 0.0001) for all three experimental groups. These data indicate that all three experimental groups responded to the acoustic tapping stimulus and displayed an ASR, moving a greater distance than during the pre-tap baseline (Figure 4A). In addition, the 2-way repeated measures ANOVA revealed a significant effect of tapping [F_(1,68) = 139.5, p < 0.0001]. A post hoc Sidak’s multiple comparison test showed no differences between groups at baseline [mean distance moved (cm) = 0.81 ± 0.04 for UC, 0.77 ± 0.06 for HC, and 0.77 ± 0.04 for PTZ group], but revealed a significantly increased ASR in the HC group compared to both UC (p = 0.007) and PTZ (p = 0.014), showing that the HC fish moved significantly farther in the one second from the start of the tap than either of the other two groups [mean distance moved at tap = 1.29 ± 0.09 for UC, 1.64 ± 0.13 for HC, and 1.32 ± 0.09 for PTZ group; n = 24 UC, 23 HC, and 24 PTZ]. Notably, the ASR of the PTZ group did not differ from that of the UC group (p = 0.995). HC fish also reached a higher max acceleration in the first second from the start of the tap, compared to both UC [p = 0.001 by Kruskal-Wallis with a post hoc Dunn’s multiple comparisons test] and PTZ groups (p = 0.029; Figure 4B). These data demonstrate that the HC group had an enhanced response to a startling stimulus, indicating that they may have an escape advantage over the other groups.