Moderate early life stress improves adult zebrafish (Danio rerio) working memory but does not affect social and anxiety‐like responses

Early life experiences are often linked to changes in cognitive and behavioral aspects in humans (Pechtel & Pizzagalli, 2011). In animal models, early life stress (ELS) is shown to directly cause long-term changes in several brain functions (Molet, Maras, Avishai-Eliner, & Baram, 2014; Nishi, Horii-Hayashi, Sasagawa, & Matsunaga, 2013; Spinelli et al., 2009). ELS can occur prenatally or Received: 29 September 2019 | Revised: 21 April 2020 | Accepted: 21 April 2020 DOI: 10.1002/dev.21986

In general, because neuropsychiatric studies depict only deleterious outcomes following ELS (Grassi-Oliveira, Ashy, & Stein, 2008;Teicher et al., 2003;Teicher, Tomoda, & Andersen, 2006;Weber & Reynolds, 2004), relatively little attention has been devoted to evaluating positive cognitive outcomes of ELS. Thus, a better understanding of the long-term consequences of ELS could improve prevention strategies for mental disorders, particularly for those that are affected by stress.

| Animals, blinding, and randomization
Experiments were carried out using Tubingen (TU) zebrafish bred in house following breeding from multiple pairs of fish. Fish were sorted into groups of 30-40, and grown to 6-weeks postfertilization in a recirculating nursery system (Tecniplast). At this point, the ELS protocol was carried out (see below for details). All fish were then grown to 4-months postfertilization in groups of ~20, and were housed in groups according to stress protocol (ELS versus handling control; 5 × groups of each condition). The experimental and technical teams were blind as to treatment allocation during rearing.
Adult (4-month postfertilization) animals were tested on one of the three behavioral procedures to ascertain perseveration/working memory (FMP Y-maze test), exploratory and anxiety-related task (novel tank diving test), and sociability (shoaling test). Animals were maintained on a 14/10-hr light/dark cycle (lights on at 9:00 a.m.), pH 7.1, at ∼28.5°C (±1°C) and were fed three times per day with a mixture of live brine shrimp and flake food, except during the weekend, when they were fed once per day. Each animal was tested in only one behavioral protocol to avoid potential effects of multiple testing. Animals were euthanized using 2-phenoxyethanol

| Short-term ELS protocol
The stressors were applied each day for 3 consecutive days. During this time, the fish were exposed to one of three stressors on each day. These included "water change" (water from tank house changed for three times), "shallow water" (2 min in shallow water; 60% of body exposure to the air), and "overcrowding" (10 fish/well in a 6-well plates for 45 min; Figure 1a). The three stressors used here were adapted from previous studies where water change, shallow water, and overcrowding induced behavioral and cortisol changes (Clark, Boczek, & Ekker, 2011;Ramsay et al., 2006). Furthermore, all the three stressors were separately tested to investigate if they were able to modulate cortisol levels in 6-week-old animals. The randomization schedule for exposure to stressors is displayed in Table 1.

| Novel tank diving test
The novel tank diving test is commonly used for analyzing exploratory and locomotor profiles. Moreover, this task is often used to analyze anxiety-like behavior being sensitive to a range of anxiogenic and anxiolytic drugs in the absence or presence of different aversive stimulus (Egan et al., 2009;Maximino et al., 2018;Mezzomo, Silveira, Giuliani, Quadros, & Rosemberg, 2016;Wong et al., 2010).
Zebrafish (n = 23) were placed individually in a novel tank (22-26 cm length × 15 cm height × 9 cm width; Figure 1b). Behavioral activity was recorded in front of the tank using a webcam for 5 min to analyze diving response (Egan et al., 2009;Parker, Millington, Combe, & Brennan, 2012;Rosemberg et al., 2012). The tank was virtually divided into three areas (bottom, middle, and top), and time at the bottom was used to assess anxiety-like phenotypes. The total distance traveled was used to evaluate differences in locomotor profile.
All behaviors were measured with automated video-tracking software (EthoVision, Noldus Information Technology Inc.) at a rate of 60 frames/s.

| Shoaling test
Zebrafish is a social species that presents a natural preference for conspecifics in neutral and mildly stressful situations (De Polavieja & Orger, 2018;Paull et al., 2010;Saverino & Gerlai, 2008). Social cohesion (n = 80, organized into 16 groups of n = 5) was tested in a shoaling assay. Sixteen groups of five fish from each treatment (ELS versus no stress) were placed in a white opaque tank (49 cm length × 15 cm height × 42 cm width) filled with 6 L of F I G U R E 1 Schematic representation of the (A) stressors and (B) behavioral tasks used to assess ELS in adult zebrafish TA B L E 1 Procedure of mild stress protocol in larvae zebrafish aquarium treated water (Figure 1b). Animals were left to habituate for 5 min, then filmed for 10 min from above using a webcam. The unit of replication was shoal (i.e., group of five fish) nested in housing group (n = 8 shoals per treatment). Data analysis was based on previous work (Parker et al., 2014;Parker, Brock, Millington, & Brennan, 2013). Briefly, the arena was split into eight equal sections and the number of fishes in each square (as a function of total number of squares occupied), at any one time, was ascertained every 10 s to calculate a dispersion index. If a fish was between squares, the square containing >50% of the fish was recorded as the occupied square. Shoaling behavior was measured using the automated video-tracking software (EthoVision, Noldus Information Technology Inc.) at a rate of 60 frames/s.

| FMP Y-maze
The FMP Y-maze is distinct from a previously used "Y-maze" proto- was used as it gives a higher chance of identifying random search patterns than using sequences of two or three choices (Frith & Done 1983). Therefore, FMP Y-maze behavior was analyzed through 16 possible four-trial outcomes (tetragrams) for each fish as a proportion of the total number of turns to examine, in detail, their behavior patterns Cleal & Parker, 2018;Gross, Engel, Richter, Garner, & Wurbel, 2011). If behavior was fully random, it would be predicted that each choice would be equally likely (i.e., the predicted frequencies would be equal for each tetragram). However, if behavior was biased toward alternation, or repetition, it would be predicted that the alternation and repetition tetragrams would be significantly different from one another.

| Cortisol levels
For the analysis of cortisol levels, fish (6-week old) were acutely exposed to water change, shallow water, or overcrowding using the conditions previously described. Due to their small size, animals were pooled in 12 per sample. Cortisol levels were assessed using a human salivary cortisol ELISA kit (Salimetrics Salivary Cortisol ELISA, Stratech) as previously described (Cachat et al., 2010;Parker et al., 2012). Fish were killed by immersion in ice-water and the whole bodies were snap-frozen in liquid nitrogen and kept at −80°C until assay. Samples were homogenized in 5-ml ice-cold PBS. Five milliliter of diethyl ether was added, and samples were centrifuged (3,500 g) for 30 min, and the top (organic) layer was removed. This process was repeated three times, and then the diethyl ether was evaporated overnight. The resulting cortisol was reconstituted in 1-mL ice-cold phosphate-buffered saline (PBS) and the ELISA was then performed in 96-well plates as per the manufacturer's instructions. Cortisol concentrations (ng/g) were determined from optical density readings compared against manufacturer's standards. All samples were run in duplicate and the inter-and intra-assay coefficients of variation were determined.

| Statistics
Data were analyzed in IBM SPSS ® Statistics and results were expressed as means ± standard error of the mean (S.E.M). To assess F I G U R E 2 Cortisol levels for non-stressed (control) and stressed animals. Data were represented as mean ± S.E.M and analyzed by one-way ANOVA, followed by Tukey's multiple comparison test. Asterisks indicate statistical differences compared to control group (*P < .05 and **P < .01, n = 3 -4 per group) whether there were any effects of stress on total turns, alterna-

| Cortisol levels are increased in stressed young animals
Initially, we exposed small groups of 6-week-old fish to the three stressors (shallow water, water change, and overcrowding) and examined the cortisol response to the exposures. A one-way ANOVA confirmed that the treatments were all invoking a stress response, with a significant difference between groups for cortisol levels (F (3, 8) = 10.67; P = .0036). Post hoc tests revealed that cortisol levels were significantly increased when acutely exposing 6-weeks animals to each of the three stressors as compared to no treatment controls: control < shallow water (P < .005), control < water change (P < .05), and control < overcrowding (P < .05; Figure 2). All samples were run in duplicate and the inter-and intra-assay coefficients of variation were < 4%.

| Anxiety and social-related phenotypes are not modulated by short-term ELS
We examined the effects of ELS on anxiety using the novel tank test, and social behavior using a social cohesion assay. Student's t-tests (two-tailed) were performed for the analyses of anxiety-like behavior and shoal responses. We observed that ELS did not change any locomotor nor anxiety-related patterns such as distance travelled (t (54) = 0.868; P = .388) and time spent in the bottom section (t (54) = 0.166; P = .86; Figure 3a). There was also no difference in the social

F I G U R E 3 Effects of ELS in stress reactivity and social responses. (A) ELS
do not change distance traveled and time spent in bottom in the novel tank diving task (n = 11-12 per group). (B) Shoal cohesion is not affected by 3 days of ELS in adult zebrafish (n = 8 per group). Data were represented as mean ± SEM and analyzed by Student's t-test cohesion ratio (t (54) = 0.060; P = .952) observed for animals exposed to ELS compared to the control group (Figure 3b).

| Working memory is enhanced in adult zebrafish exposed to ELS
Finally, we analyzed the effects of ELS on working memory by examining turn choices in the FMP Y-maze exploration task. Through analysis of overall turn choices, we found that ELS affected FMP Y-maze performance. Figure 4 depicts the behavioral 16-choice tetragrams for control versus ELS-exposed animals. There was no main effect of

| D ISCUSS I ON
In this study, we evaluated the effects of exposing young zebrafish (16-day old) to unpredictable moderate stress for 3 days on adult behavior (4-month old), through the analysis of three different domains: locomotor and anxiety (novel tank diving test), sociality (shoal cohesion test), and memory (FMP Y-maze test). We showed, for the first time, that 3 days of ELS improves zebrafish working memory by increasing alternations and decreasing repetitions in turn-direction in the FMP Y-maze test. Although learning and memory process were affected by short-term ELS, stress did not modulate social behavior and anxiety-like behavior in adult zebrafish. Collectively, these data support previous evidence that early life exposure to moderate stress can build resilience and have an important adaptive role for different species, including humans. Our study has highlighted F I G U R E 4 Y-maze tetragrams showing the behavioral phenotype of control and ELS groups. Data were represented as mean ± SEM and analyzed by linear mixed effects, followed by Tukey's multiple comparison test. Asterisks indicate statistical differences compared to non-biased group or between biased groups (*P < .05, **P < .01, ***P < .001 and *P < .0001, n = 24 per group)

F I G U R E 5 ELS increases alternations
and decreases repetition in turn-direction without changing the left or right preference of adult zebrafish. Data were represented as mean ± SEM and analyzed by Student's t-test. Asterisks indicate statistical differences compared to control group (*P < .05 and **P < .01, n = 24 per group) that zebrafish is an important animal model for elucidating the developmental processes underlying ELS, and how this impacts adult behavior.
Early life experiences regulate individual differences through both neural plasticity and epigenetic modifications, and these modifications later determine the capacity for flexible and adaptive behaviors in adulthood (McEwen, Bowles, et al., 2015;. Stressful stimuli can exert both positive and negative impacts on neurophysiological aspects being balanced to give the best fitness outcome (Bogdan & Pizzagalli, 2006;Gluckman, Hanson, & Beedle, 2007;Maier, Amat, Baratta, Paul, & Watkins, 2006;Zannas & West, 2014). To confirm if shallow water, water changes, and overcrowding changed zebrafish stress-regulating systems, we analyzed the effects of these three stressors separately in 6-week-old animals. As previously observed in adults (Pavlidis et al., 2013;Song et al., 2018), the acute stressors increased cortisol levels in young animals.
The impact on adult life of ELS follows an inverted U-shaped curve, where exposure to mild/moderate stress during development results in positive/beneficial outcomes such as resilience, and either too little or too much stress exposure (or too high an intensity) leads to negative outcomes (Russo, Murrough, Han, Charney, & Nestler, 2012;Sapolsky, 2015). Concerning this, chronic ELS is related to different behavioral outcomes in different species, including enhanced emotional-and stress reactivity, which are frequently linked to anxiety disorders (Coplan et al., 1996;Nugent, Tyrka, Carpenter, & Price, 2011), aggressiveness (Veenema, 2009;Veenema, Blume, Niederle, Buwalda, & Neumann, 2006), impulsiveness (Lovallo, 2013), and antisocial behavior (Haller, Harold, Sandi, & Neumann, 2014;Kohl et al., 2015). Meanwhile, adolescents exposed to moderately severe ELS events showed blunted depressive symptom responses to changes in proximal stressful events in the previous 9 months, compared to those with fewer ELS events (Shapero et al., 2015). Here, we showed that shortterm ELS did not affect adult zebrafish anxiety-like profile nor social cohesion in both the novel tank diving test and shoal cohesion test. However, ELS increased the number of alternations in turn-direction and decreased the number of repetitions in the FMP Y-maze task.
The FMP Y-maze task comprises three identical arms, where the animal is introduced to the center of the maze and allowed to freely explore. In this task, over the course of multiple arm entries, animals have a natural tendency to enter the recently visited arm less frequently, thus increasing the number of alternations in turn-direction across the test (Drew, Miller, & Baugh, 1973;Hughes, 2004;Kokkinidis & Anisman, 1976;Swonger & Rech, 1972). In general, various configurations of the Y-maze task have been used to study learning and processes in animal models, including rodents (Fu et al., 2017;Ghafouri et al., 2016;Luine, 2015) and fish (Aoki, Tsuboi, & Okamoto, 2015;Cleal & Parker, 2018;Cognato Gde et al., 2012). Both decreased alternations and increased repetitions are affected pharmacologically by muscarinic and NMDA-receptor antagonists, and additionally by β-amyloid peptides in rodents (Cunha et al., 2008;Hiramatsu & Inoue, 2000;Park et al., 2010;Walker & Gold, 1992). Recently, we also observed that zebrafish presented decreased alternations in turn-direction when exposed to moderate alcohol during early nervous system developmental (Cleal & Parker, 2018). Although the FMP Y-maze responses can be influenced by novelty-seeking (Fontana, Cleal, Clay, et al., 2019), we observed here that ELS did not change exploratory responses of animals at the novel tank diving task. Thus, it seems likely, that increased alternations in turn-direction and a coincidental decrease in repetitions are indicative that the ELS animals may have an improvement in learning and memory adaptive flexibility, thus suggesting increased "resilience" as a result of the ELS exposure.
Memory is a highly dynamic process that is built from initial encoding to the new and fragile memory trace that is stabilized during consolidation and reactivated during memory retrieval (Abel & Lattal, 2001;Bisaz, Travaglia, & Alberini, 2014). Memory can also return to an unstable state in which reconsolidation is needed to stabilize it (Dudai, 2006;Lv, 2015). Stress has been shown to affect all phases of memory (encoding, consolidation, memory retrieval, and reconsolidation); however, how stress influences memory depends on when an individual is stressed, and what frequency and intensity of that stress are (Schwabe & Wolf, 2012). When looking at resilience induced by stress, it seems that recovery from stress-inducing changes in neural architecture is not simply a reversal, but instead is a form of neuroplastic adaptation .
Although neural plasticity and epigenetic factors are known to underlie the mechanisms responsible for ELS inducing resilience in adults (Cadet, 2016;Kentner, Cryan, & Brummelte, 2018;, there remains no clear understanding of the mechanisms involved in these processes and how they are functionally associated.

| CON CLUS ION
Here, we show for the first time that adult zebrafish exposed to moderate ELS can build resilience as evidenced by increases in working memory, but without changing anxiety-like behavior or social behavior patterns. We suggest that this protocol could serve as a useful model to understand: (a) translational genetic and physiological aspects are associated with memory adaptive flexibility; (b) the evolutionary and conserved characteristics that are common between zebrafish and humans correspond to stressinduced changes in memory and learning. Overall, future studies should investigate how moderate ELS can modulate neuroplastic adaptation and epigenetics in zebrafish adaptive flexibility and resilience.

ACK N OWLED G M ENTS
This study was financed in part by the Coordenação de is supported by a Science Faculty Studentship from the University