You are on page 1of 8

Behavioural Processes 149 (2018) 35–42

Contents lists available at ScienceDirect

Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc

Extending the analysis of zebrafish behavioral endophenotypes for modeling T


psychiatric disorders: Fear conditioning to conspecific alarm response
⁎⁎
Caio Maximinoa,b, , Daniele L. Meinerzc, Barbara D. Fontanac,d, Nathana J. Mezzomoc,e,
Flavia V. Stefanelloc, Alessandro de S. Prestesd, Cibele B. Batistac, Maribel A. Rubind,

Nilda V. Barbosad, João Batista T. Rochad, Monica G. Limaf, Denis B. Rosembergb,c,d,
a
Laboratório de Neurociências e Comportamento “Frederico Guilherme Graeff”, Instituito de Estudos em Saúde e Biológicas, Universidade Federal do Sul e Sudeste do Pará,
Avenida dos Ipês, S/N, Marabá, PA, Brazil
b
The International Zebrafish Neuroscience Research Consortium (ZNRC), 309 Palmer Court, Slidell, LA 70458, USA
c
Laboratório de Neuropsicobiologia Experimental, Departamento de Bioquímica e Biologia Molecular, Centro de Ciências Naturais e Exatas, Universidade Federal de Santa
Maria, Avenida Roraima, 1000, 97105–900, Santa Maria, RS, Brazil
d
Programa de Pós-graduação em Bioquímica Toxicológica, Universidade Federal de Santa Maria, Avenida Roraima, 1000, 97105–900, Santa Maria, RS, Brazil
e
Programa de Pós-graduação em Farmacologia, Universidade Federal de Santa Maria, Avenida Roraima, 1000, 97105–900, Santa Maria, RS, Brazil
f
Laboratório de Neurociências e Comportamento “Frederico Guilherme Graeff”, Departamento de Morfologia e Ciências Fisiológicas, Universidade do Estado do Pará,
Avenida Hileia, S/N, Agrópolis do INCRA, Marabá, PA, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Anxiety, trauma- and stressor-related disorders are severe psychiatric conditions that affect human population
Fear conditioning worldwide. Given their genetic tractability, evolutionarily conserved neurotransmitter systems, and extensive
Alarm substance behavioral repertoire, zebrafish have become an emergent model organism in translational neuroscience. Here,
Conditioned place aversion we investigate whether a single exposure to conspecific alarm substance (CAS) produces fear conditioning in
Psychiatric disorders
zebrafish using a conditioned place aversion (CPA) paradigm, as well as the persistence of aversive responses at
Zebrafish
different time intervals. While CAS elicited freezing and erratic movements at conditioning phase, zebrafish
showed a robust avoidance for the CAS-paired compartment and increased risk assessment up to 7 days post-
conditioning. Additionally, we observed the existence of two behavioral phenotypes (high- and low-avoider fish)
that present different fear-like responses at conditioning phase and evasion of the conditioning side at post-
conditioning trials. Collectively, we show a prolonged conditioned place aversion in zebrafish after a single CAS
conditioning session, reinforcing the use of fear conditioning protocols as valuable strategies for modeling
psychiatric disorders-related phenotypes in zebrafish.

1. Introduction et al., 2017). Fear conditioning protocols elicit traumatic memories


after exposure to an intense stressor, modeling a psychological stress in
Anxiety, trauma- and stressor-related disorders are severe psychia- the absence of noxious stimuli. These models have been associated with
tric conditions that affect human population worldwide (Shalev, 2009). a vulnerability to phobic fears and other anxiety-related disorders, such
Since the evolution theory has been closely associated with the beha- as panic disorder (PD), agoraphobia, acute stress disorder (ASD), and
vioral neuroscience, animal models of affective disorders attempt to posttraumatic stress disorder (PTSD) (Battle, 2013).
reproduce features of human psychiatric disorders in laboratory ani- Avoidance learning depends on a Pavlovian component which in-
mals, correlating the physiological and behavioral changes associated volves the association of a neutral conditioned stimulus (CS), (e.g., tone
to specific emotional states (Campos et al., 2013). One approach for and visual cues) with a threat signal (unconditioned stimulus, US). Re-
modeling anxiety, trauma- and stressor-related disorders in animals is exposure to the CS will elicit the expectancy of an aversive outcome and
avoidance conditioning, in which subjects learn to minimize and/or thereby activates conditioned defensive responses (e.g., physiological
prevent contact with aversive events (Krypotos et al., 2015; LeDoux arousal, avoidance tendencies) in the absence of US (Bolles et al., 1970;


Corresponding author at: Laboratório de Neuropsicobiologia Experimental. Departamento de Bioquímica e Biologia Molecular, Centro de Ciências Naturais e Exatas, Universidade
Federal de Santa Maria. Avenida Roraima, 1000, 97105–900, Santa Maria, RS, Brazil.
⁎⁎
Corresponding author at: Laboratório de Neurociências e Comportamento “Frederico Guilherme Graeff”, Instituito de Estudos em Saúde e Biológicas, Universidade Federal do Sul e
Sudeste do Pará. Avenida dos Ipês, S/N, Marabá, PA, Brazil.
E-mail addresses: cmaximino@unifesspa.edu.br (C. Maximino), dbrosemberg@gmail.com (D.B. Rosemberg).

https://doi.org/10.1016/j.beproc.2018.01.020
Received 10 October 2017; Received in revised form 29 January 2018; Accepted 30 January 2018
Available online 02 February 2018
0376-6357/ © 2018 Elsevier B.V. All rights reserved.
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

Dinsmoor, 1977, 2001; Fernando et al., 2014; Krypotos et al., 2015; 50 cm × 35 cm × 6 cm (length × width × height), which had equal
LeDoux et al., 2017). Escaping behavior persists depending on instru- divisions for each fish (6 cm × 6 cm × 6 cm – length × width ×
mental learning with multiple sources of reinforcement (Krypotos et al., height) and small perforations (0.5 cm diameter) to allow free water
2015) and impaired safety signal learning has been proposed as an circulation inside the tank. These perforated Plexiglas partitions al-
endophenotype for several brain disorders (Jovanovic et al., 2012; lowed a precise identification of each animal during the experimenta-
Jovanovic et al., 2010; Jovanovic et al., 2009). tion and permitted fish to visualize conspecifics minimizing isolation
Avoidance conditioning has been observed in different vertebrate stress. Water conditions were similar to those of housing tanks, feeding
species (Vargas et al., 2012) to assess various behavioral en- was not interrupted during this period, and animals were kept in the
dophenotypes (de Mooij-van Malsen et al., 2011). In goldfish, for ex- task room to minimize the effects of context variation during the trials.
ample, active avoidance depends on the dorsomedial telencephalon To minimize potential handling stress throughout the experiments, fish
(Portavella et al., 2004a; Portavella et al., 2004b), a basolateral were handled once daily.
amygdala homolog (Maximino et al., 2013). Although inhibitory
avoidance protocols have been proposed using electric shocks as US in 2.2. Conditioned place aversion
zebrafish (Blank et al., 2009; Manuel et al., 2014), a different approach
consists in using aversive chemical cues as fear-inducing stimuli. Con- Conditioned place aversion (CPA) is a protocol derived from con-
specific alarm substance (CAS) is a pheromone-like mixture released in ditioned place preference, in which a drug is administered in a com-
the water as consequence of damage to the epithelial club cells (Doving partment of the apparatus but not in the other. Then, rewarding
and Lastein, 2009; von Frisch, 1938). In zebrafish, CAS elicits fear-like properties of the drug may be assessed by increased occupancy of the
behaviors, such as erratic movements and freezing (Mathuru et al., associated compartment in a posterior drug-free situation (Collier et al.,
2012; Quadros et al., 2016; Speedie and Gerlai, 2008). Moreover, CAS 2014; Prus et al., 2009). In CPA, a potentially aversive stimulus is
serves as US in Pavlovian fear conditioning, because it increases fear paired with one compartment, while its absence is associated with the
responses to CS after paired with a neutral visual stimulus (flashing red other compartment of the apparatus. Thus, explicit relationships be-
light) (Hall and Suboski, 1995a; Hall and Suboski, 1995b). Although tween the presence and absence of an aversive US and context are es-
the re-exposure to the context substantially increases erratic swimming tablished, in which clear safe and danger signals are present during the
24 h after conditioning (Ruhl et al., 2017), the persistence of escaping trials.
responses after distinct time intervals remains unexplored. Since CAS The dimensions of CPA apparatus were 30 cm × 13 cm × 12 cm
can also sensitize anxiety-like behavior after a single exposure, this (length × width × height). Tank was divided into two equally sized
pheromone is a potential stimulus for avoidance conditioning research. compartments (12 cm × 12 cm × 12 cm – length × width × height) by
A brief CAS exposure increases anxiety-like behavior and arousal 24 h a 5-cm gray central partition and two sliding guillotine-type doors
after the stimulus presentation, suggesting a time-dependent sensitiza- (12 cm × 12 cm). To ensure the contrast between experimental subjects
tion for both domains (Lima et al., 2016). Thereby, using CAS in and tank allowing a precise automated behavioral analysis, chambers
avoidance conditioning protocols could expand the breadth of domains were divided by opaque plastic self-adhesive films in white or yellow
to include trauma re-experience, improving the face validity of zebra- colors, covering walls, floor and the corresponding sides of the sliding
fish models of psychiatric disorders. Here, we investigate the aversive guillotine-type doors. Two yellow- and white-covered tanks with
conditioning to a single CAS exposure in zebrafish using a conditioned identical dimensions (12 cm × 12 cm × 12 cm – length × width ×
place aversion (CPA) paradigm and verify the stability of conditioned height) were coupled to the CPA apparatus and used for the con-
responses after different time intervals. ditioning phase. Tanks were illuminated by a 15-W fluorescent lamp
located at 1.80 m above each apparatus, which kept the illumination
2. Material and methods uniform and constant between trials. The CPA tanks and the experi-
mental protocol are shown in Fig. 1.
2.1. Animals and housing
2.2.1. Preconditioning trial
Male and female (∼50:50 ratio) wild type adult zebrafish (Danio The conditioned place preference literature recommends that
rerio) (4–6 months-old, blue short-fin [BSF] phenotype) were obtained ‘biased’ tanks can be used for conditioning because they are highly
from a commercial distributor (Hobby Aquários, RS, Brazil). Since selective, imposing a more “steep” barrier to avoidance conditioning. In
short-fin phenotype fish are genetically heterogeneous, they are ex- biased designs for CPA, the preference of each individual animal for a
pected to better represent the natural populations, decreasing the ef- particular compartment prior to conditioning is evaluated, and the
fects of arbitrary genetic drifts (Speedie and Gerlai, 2008; Lima et al., most-preferred compartment is further associated with the noxious
2016; Parra et al., 2009). In the literature, a large amount of data using stimulus (Prus et al., 2009). We followed this method by introducing a
heterogeneous zebrafish stocks from commercial sources has shown a preconditioning trial, in which fish were placed individually into the
considerable replication among groups when tested in different beha- central partition of the apparatus. After an acclimation period to the
vioral protocols (Egan et al., 2009; Song et al., 2018; Piato et al., 2011; new environment of 30 s, the guillotine-type doors were removed and
Quadros et al., 2016). Fish were previously housed for 14 days in 50-L the animal was free to explore the apparatus for 10 min. The basal
thermostatic aquaria to acclimate to laboratory conditions. Tanks were preference was assessed by measuring the time spent in each com-
filled with non-chlorinated water (temperature set at 27 ± 1 °C, pH partment.
7.2) and kept under constant mechanical and chemical filtration. Fish
were raised in a 14/10 light/dark photoperiod cycle (lights on at 7:00 2.2.2. Conspecific alarm substance extraction and conditioning trial
am and off at 9:00 pm) and lightning was provided by ceiling-mounted CAS was extracted as described previously (Canzian et al., 2017;
fluorescent tubes. Animals were fed thrice daily with commercial flake Lima et al., 2016), where a single donor fish was rapidly euthanized by
fish food (Alcon BASIC®, Alcon, Brazil) and kept in accordance with the decapitation. Afterwards, 10–15 superficial shallow cuts were made on
standard protocols of zebrafish maintenance (Lawrence, 2007). All each side of fish body and care was taken to avoid blood drowning. The
experimental protocols were approved by the Ethics Commission on donor fish was further washed with 10 mL of distilled water to obtain
Animal Use of the Federal University of Santa Maria (process number the skin extract preparation, which was reserved on ice for posterior
106/2014). use.
Two weeks before the experiments, fish were randomly transferred For the conditioning trial, experimental fish were individually
from their housing tanks to an aquarium measuring placed into their most-preferred compartment, and 3.5 mL/L CAS was

36
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

Fig. 1. Schematic representation of the experimental protocol. Zebrafish were kept during 14 days before the experiments in perforated Plexiglas tanks and the basal preference was
assessed at the preconditioning trial. After 24 h, animals were exposed to conspecific alarm substance (CAS) for 5 min and changes on preference were tested at different postconditioning
sessions (DPC). The dimensions of the CPA tanks are shown.

further added to the water (Canzian et al., 2017; Egan et al., 2009; Lima each day (between 10:00 and 16:00) using a webcam located above the
et al., 2016; Quadros et al., 2016). The animal was kept for 5 min in this CPA apparatus. Two trained observers (inter-rater reliability > 0.85)
compartment and immediately further removed from the tank. Then, blind to the experimental condition analyzed the videos using auto-
the tank was washed to remove CAS residuals, and the same animal was mated video-tracking software system (ANY-maze™, Stoelting CO, USA)
later transferred to the non-preferred compartment, where distilled at a rate of 30 frames/s. In order to provide a background noise, a fan
water was added. The animal was kept in this “safe” compartment for was kept on in the experimental room.
5 min. Animals from control group were exposed to distilled water in
both compartments instead of CAS. To allow a better data reproduci- 2.4. Statistics
bility, the order of test was randomized for each fish, and the exposure
to the first compartment occurred between 11:00–12:00, while the Data normality and homoscedasticity were previously analyzed by
exposure to the other chamber occurred between 14:00–15:00. Both Kolmogorov–Smirnov and Bartlett’s tests, respectively, as basic pre-
erratic movements and freezing events were measured at conditioning requisites for running parametric analyses. The basal preference of
phase. These endpoints are characteristic of fear consistently elicited zebrafish was assessed using one-way analysis of variance (ANOVA)
during CAS exposure (Canzian et al., 2017; Quadros et al., 2016; followed by Tukey’s test. To ascertain that initial preference was ran-
Speedie and Gerlai, 2008) and serve the purpose of ‘quality control’. domly distributed between the groups, a Fisher exact test was applied to
Additionally, latent classes were analyzed to identify heterogeneous the proportions of animals preferring each compartment. Effect sizes
trajectories using the pattern of behavior change over time (see below). were reported as Cohen’s d and behavioral endpoints were expressed as
means ± standard error of the mean (S.E.M.). At conditioning phase,
2.2.3. Postconditioning trials CAS effects were evaluated by unpaired Student’s t test and effect sizes
At postconditioning sessions, occupancy was assessed to evaluate were reported as Hedges’ g. Conditioning effects were analyzed by re-
the efficacy of the avoidance conditioning protocol. Animals changing peated measures ANOVA (RM-ANOVAs), with group (CTRL vs. CAS) as
their preference to the “safe” compartment show signs of avoidance fixed effect and trial period (Preconditioning, DPC1, DPC2, DPC3, and
learning. Erratic swimming and freezing were recorded simultaneously DPC7) as the within-subjects factor. Differences between groups were
in each compartment as indexes of CS-US conditioning. Preference was further assessed by Tukey’s test. P-values for trial effects were adjusted
scored at different postconditioning (DPC) periods: 24 h (DPC1), 48 h for sphericity.
(DPC2), 72 h (DPC3), and 7 days (DPC7) after conditioning. All tests Latent Growth Modeling (LGM) methods were used to test for po-
occurred in the same manner: first, each fish was placed into the central pulation heterogeneities in avoidance conditioning over trials. LGM has
partition for 30 s and then the guillotine-type doors were opened. been applied to identify heterogeneous trajectories in longitudinal data
Animals had free access to all compartments for 10 min and the time by identifying latent sub-populations within finite distributions
spent in each chamber was quantified. Aversion was estimated as the (Muthén, 2004; Proust-Lima et al., 2016). LGM provides a statistical
difference (in seconds) of time spent in the CAS-paired compartment framework for determining the number of classes/sub-populations that
during the post and preconditioning phases. We also analyzed the best fit the data (Muthén, 2004). LGM methods have been applied to
number of risk assessment episodes, which were defined as fast (> 1 s) identify distinct patterns of symptoms following trauma exposure
entries in the conditioning compartment followed by re-entry in the (Bonanno et al., 2012; Galatzer-Levy et al., 2013), to avoidance
safe compartment, or as partial entries in the conditioning side (Kalueff learning and threat extinction learning in rats (Galatzer-Levy et al.,
et al., 2013; Mezzomo et al., 2016). After each trial, animals were kept 2013; Galatzer-Levy et al., 2014).
into the perforated Plexiglas tank and manipulated once daily until the A heterogeneous linear mixed model (HLME) was employed using
end of the experimental period. Fish were further anaesthetized in the lcmm package from R (Lima et al., 2015) (https://cran.r-project.
water at 4 °C and euthanized by decapitation. org/web/packages/lcmm/index.html) to identify heterogeneous tra-
jectories by testing for discrete mixture distributions of avoidance
2.3. Behavioral analysis conditioning and retrieval. HLME is an extension of LGM methods that
uses fixed effects, and was employed to reduce the number of para-
All experimental sessions were recorded during the same time frame meters being modeled – a necessity given that small sample sizes

37
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

decrease convergence of parameter fits. A piecewise model, which al- decreased erratic movements (t(df=9.649) = 3.623, p = .0049,
lows modeling separate trajectories for the same subjects, was used g = −1.29) at conditioning session when compared to low avoiders. No
(Flora et al., 2008). Behavior at preconditioning was used as one piece, significant differences were observed in the duration of these behaviors
and the retrieval piece covered the 4 time points of the post- (t(df=10.42) = 0.7998, p = .4417, NS for freezing and
conditioning trials. Those two trajectories were connected by a single t(df=2.611) = 0.7142, p = .5336, NS for erratic movements) (Fig. 5A).
intercept placed at DPC1. Competing models were compared using Finally, RM-ANOVA yielded significant differences between high and
linear slopes only for the preference trials. The best fit was assessed low avoiders regarding the time spent in the conditioning side
based on Bayesian Information Criteria (BIC) and log-likelihood, with (F(3,42) = 6.6, p = .0009, gεs = 0.26 for the interaction term). Both
parsimony and interpretability consistent with previous data on rodents classes did not differ in the number of risk assessment, freezing bouts,
(Galatzer-Levy et al., 2014). The best model was selected based on erratic movements, distance traveled, and crossings across trials. In all
lower values for the criterion indices. cases, except for erratic swimming, a trial effect was observed (Fig. 5B).
After the establishment of the latent classes, individual animal’s Table 1 depicts all experimental data obtained for high and low avoider
probable class assignment was used in further analysis. Behavioral fish.
endpoints measured at conditioning session between classes were
compared by unpaired Student’s t test with Welch’s correction and ef- 4. Discussion
fect sizes were reported as Hedges’ g. Data analyzed across different
time periods were compared using RM-ANOVA with class assignment as Here, we report the use of CPA paradigm to assess the behavioral
a fixed factor and trial as the within-subjects factor. The level of sig- effects of CAS in adult zebrafish. We demonstrate a persistence of CS
nificance was set at p ≤ 0.05 and effect sizes were reported as gen- aversion (up to 7 DPC) following a single CAS exposure and increased
eralized eta squared (gεs). risk assessment and freezing episodes across postconditioning trials.
Interestingly, while freezing and erratic swimming increased at con-
3. Results ditioning phase, erratic swimming did not change after conditioning
session. Additionally, we observed the existence of two behavioral
At preconditioning phase, one-way ANOVA showed that zebrafish phenotypes (high- and low-avoider fish) that present different fear-like
had a preference for the white compartment of the CPA apparatus responses at conditioning phase and avoidance of the conditioning side
(F(2,93) = 9.164, p = .0002, d = 0.7221) (Fig. 2A). Animals were fur- at postconditioning sessions. Considering the persistence of evasion
ther randomly distributed in such a way that CTRL and CAS groups had after a single exposure to a chemical cue that elicits fear (Canzian et al.,
similar baseline place preference (Fisher’s exact test, p = .65, 2017; Nathan et al., 2015; Ogawa et al., 2014) and exerts a time-de-
d = 0.1609) (Fig. 2B). At conditioning phase (Fig. 2C), unpaired Stu- pendent sensitization of anxiety-like behaviors (Lima et al., 2016), our
dent’s t test revealed that CAS increased freezing bouts novel findings expand the analysis of aversive behaviors in zebrafish.
(t(df=30) = 2.425, p = .0215, g = −0.84), freezing duration Because zebrafish display a high genetic and physiological con-
(t(df=30) = 2.693, p = .0115, g = −0.93), number of erratic move- servation (Levin et al., 2015; Stewart et al., 2015), as well as robust
ments (t(df=30) = 3.778, p = .0007, g = −1.30) and duration of erratic aversive behavioral responses (Kalueff et al., 2013), this species is a
movements (t(df=30) = 4.333, p = .0002, g = −1.49) (Fig. 2D). suitable model organism in biological psychiatry and translational
Behavioral data suggest significant changes regarding the color neuroscience research (Caramillo et al., 2015; Lima et al., 2016;
preference in CAS-exposed animals, indicating aversion to the con- Maximino et al., 2015; Stewart et al., 2014). Here, CAS exposure was
ditioning side at different time intervals. RM-ANOVA using group and associated with the most preferred side of the CPA apparatus, eliciting
trial period as factors yielded a significant group effect avoidance tendencies as well as fearful responses (freezing, risk as-
(F(1,90) = 25.501, p < .00001, gεs = 0.222 for the group term). CAS- sessment) which were not observed in the ‘safe’ compartment. Since
exposed group had a stronger avoidance response to the white com- avoidance conditioning carries at least two factors (Pavlovian con-
partment when compared to CTRL at postconditioning trials (DPC1 to ditioning and instrumental conditioning of the avoidance response)
DPC7) (Fig. 3A). Although distance traveled and crossings did not differ (Krypotos et al., 2015), the persistence of freezing are assumed to be
between groups, a trial period effect was observed in distance traveled (classically) a conditioned contextual response. Despite the long-term
(F(3,90) = 8.29, p = .001, gεs = 0.1327 for the trial term) and crossings avoidance and fear responses are useful to assess aversive behaviors,
(F(3,90) = 5.9351, p = .0047, gεs = 0.07 for the trial term), showing they help assess learning and memory in zebrafish. While it is not clear
reduced locomotion across the trials (Fig. 3B). Fig. 4 depicts the the time intervals tested following CAS exposure previously (Hall and
avoidance and fear-like behaviors in the conditioning compartment at Suboski, 1995a; Hall and Suboski, 1995b), classically conditioned CAS
pre and postconditioning phases. In CAS-exposed animals, risk assess- responses are maintained after training (Lima et al., 2016). Although
ment and freezing increased at DPC1–DPC7 (F(1,90) = 26.99, changes in locomotor activity and exploration are not directly involved
p < .0001, gεs = 0.403 for the group term) and DPC1–DPC2 in the preference response, they markedly reduced at postconditioning
(F(3,90) = 5.7, p = .005, gεs = 0.11 for the interaction term), respec- trials, suggesting habituation to the test apparatus in control and CAS-
tively. Regarding the number of erratic movements, RM-ANOVA exposed fish.
yielded no statistical differences after correction for sphericity We have also demonstrated heterogeneity in evasion responses, a
(F(3,90) = 2.2, p = .119, gεs = 0.029 for the interaction term). phenomenon observed in active avoidance in rodents (Galatzer-Levy
Using the HLME, we found that a 2-class model best fits the data et al., 2014) and in inhibitory avoidance in zebrafish (Manuel et al.,
(Akaike’s information criteria: 836.21 for 2-class model, vs. > 841 for 2014). Here, two distinct populations provide better model fit than a
1-class, 3-class, and 4-class models). The largest class (here noted as single population estimate. Specifically, two phenotypes were un-
“high avoiders”) was composed of 81.25% of animals and was char- covered by latent growth modeling: the first (“high avoiders”, 81.25%
acterized by a steady decline in time in conditioning side 24 and 48 h of animals) showed signals of avoidance 24 h after training (DPC1) and
after CAS exposure, reaching an asymptote up to 7 DPC. The second achieved asymptotic levels of avoidance 48 h after training (DPC2). The
class (described herein as “low avoiders”) was composed of 18.75% of second phenotype (“low avoiders”; 18.75% of animals) showed evasion
animals and was characterized by an initial decrease in the time spent at DPC1, but later return to baseline levels. Thus, we demonstrate that
in conditioning side 24 h after CAS exposure, followed by a return to both high and low avoiders display aversive responses following a
control levels. Fig. 5 shows the main behaviors observed for both single CAS exposure, suggesting one-trial learning. Nonetheless, dif-
classes. Unpaired Student’s t test revealed that high avoiders had in- ferently than low avoiders, high avoiders showed pronounced escaping
creased freezing bouts (t(df=6.254) = 2.469, p = 0.0469, g = 1.00) and behavior (DPC1 to DPC7), which reveals a clear persistence of evasion

38
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

Fig. 2. Basal preference of zebrafish (preconditioning session) and CAS-induced fear-like behaviors (conditioning session). (A) Time spent in yellow, central, and white compartments of
the CPA apparatus. Data were expressed as means ± S.E.M and analyzed by one-way ANOVA followed by Tukey’s HSD test (* p ≤ 0.05, *** p ≤ 0.005, n = 32). (B) Initial preference of
zebrafish randomly distributed into control (CTRL) and conspecific alarm substance (CAS) groups. Data were expressed as means ± S.E.M and the proportion of animals preferring each
compartment was assessed using a Fisher exact test, considering p ≤ 0.05 as statistical significance. (C) Schematic cartoons of the conditioning phase. In a random order, CTRL group was
exposed to non-chlorinated water, while CAS group was conditioned to CAS in the most preferred compartment of the CPA apparatus and paired with non-chlorinated in the other
compartment. (D) Fear-related behaviors of zebrafish during conditioning. Data were expressed as means ± S.E.M and analyzed by unpaired Student’s t test (* p ≤ 0.05, *** p ≤ 0.005)
(n = 16 per group).

for at least 7 days postconditioning. While no other differences between alarm reaction – and, therefore, increased fear levels at conditioning
phenotypes were observed after conditioning, high avoiders froze more phase could impair avoidance learning. Studying animals with high and
and displayed less erratic swimming than low avoiders during CAS low avoidance is an interesting strategy to understand the mechanisms
exposure. Conversely, the occurrence of freezing during conditioning is underlying avoidance conditioning (Choi et al., 2010). Acquisition and
incompatible with a successful evasion in active avoidance conditioning retrieval processes following CAS conditioning recruit the dorsomedial
of rodents (Choi et al., 2010). It is unclear whether freezing in zebrafish telencephalon, while only retrieval recruits the dorsolateral tele-
has the same function as freezing in rodents, because it is not con- ncephalon, a hippocampus homolog (Ruhl et al., 2017). Given the
sistently observed during CAS exposure (Canzian et al., 2017; Quadros homology between the dorsomedial telencephalon and the mammalian
et al., 2016; Ruhl et al., 2017; Speedie and Gerlai, 2008). Interestingly, basolateral amygdala, different activities in CNS regions may possibly
CAS had a higher effect on erratic swimming (g = −5.21) than freezing reflect the heterogeneity observed herein. However, more studies are
(g = −3.34), suggesting that erratic swimming is a better index of needed to elucidate how fear-like behaviors influence the persistence of

39
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

Fig. 3. CAS induces conditioned place avoidance without changing locomotion. (A) Changes on preference scores at 1, 2, 3, and 7 DPC, calculated as the difference (in seconds) of time
spent in the CAS-paired compartment at post and preconditioning phases. Data were expressed as means ± S.E.M and analyzed using RM-ANOVA followed by Tukey’s HSD test (* p ≤
0.05, ** p ≤ 0.01). (B) Locomotor activity of zebrafish in the CPA apparatus. Data were expressed as means ± S.E.M and analyzed using RM-ANOVA (n = 16 per group).

Fig. 4. Avoidance and fear-like behaviors at preconditioning and postconditioning phases. The number of risk assessment episodes, freezing bouts, and erratic movements in CAS-paired
compartment were quantified at preconditioning and at different postconditioning intervals. Data were expressed as means ± S.E.M and analyzed using RM-ANOVA followed by Tukey’s
HSD test (n = 16 per group).

Fig. 5. Effects of CAS in zebrafish displaying high avoidance (HA) and low avoidance (LA) to the conditioning compartment. (A) Fear-related behaviors of HA and LA zebrafish at
conditioning phase. Data were expressed as means ± S.E.M and analyzed by unpaired Student’s t test with Welch’s correction (* p ≤ 0.05, ** p ≤ 0.01). (B) Avoidance, fear-like
behaviors, and locomotion of HÁ and LA at preconditioning and postconditioning phases. Data were expressed as means ± S.E.M and analyzed using RM-ANOVA followed by Tukey’s
HSD test.

avoidance responses in zebrafish. aversion. Our novel findings related to different escaping responses
observed across postconditioning trials (as described for high and low
avoiders) raise the possibility to investigate the molecular bases and the
5. Conclusion
neurochemical mechanisms underlying these phenotypes. Additionally,
we reinforce the use of CAS associated with CPA paradigm as a valuable
Collectively, we report for the first time the persistence of a con-
strategy for assessing anxiety, trauma- and stressor-related behavioral
ditioned place aversion across different time intervals in zebrafish after
endophenotypes in order to improve the face validity of zebrafish
a single CAS conditioning session. CAS elicits fear-related behaviors and
models of psychiatric disorders.
avoidance responses for at least 7 DPC, suggesting a persistence of

40
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

Table 1 Doving, K.B., Lastein, S., 2009. The alarm reaction in fishes–odorants, modulations of
Statistics of behavioral endpoints considering class differences. Degrees of freedom (DF), responses, neural pathways. Ann. N. Y. Acad. Sci. 1170, 413–423.
F and p values, and effect sizes, represented as generalized eta squared (gεs) are shown. Egan, R.J., Bergner, C.L., Hart, P.C., Cachat, J.M., Canavello, P.R., Elegante, M.F.,
Significant data are marked in bold font. Elkhayat, S.I., Bartels, B.K., Tien, A.K., Tien, D.H., Mohnot, S., Beeson, E., Glasgow,
E., Amri, H., Zukowska, Z., Kalueff, A.V., 2009. Understanding behavioral and phy-
RM-ANOVA results for class differences siological phenotypes of stress and anxiety in zebrafish. Behav. Brain Res. 205 (1),
38–44.
Fernando, A.B., Urcelay, G.P., Mar, A.C., Dickinson, A., Robbins, T.W., 2014. Safety
Behavioral endpoints Source of DF F p-value gεs
signals as instrumental reinforcers during free-operant avoidance. Learn. Mem. 21,
variation
488–497.
Time spent in the Class 1,14 21.0 0.00043 0.29 Flora, D.B., Curran, P.J., Hussong, A.M., Edwards, M.C., 2008. Incorporating measure-
conditioning Trial 3,42 6.6 0.00093 0.26 ment non-equivalence in a cross-study latent growth curve analysis. Struct. Equ.
compartment Class:Trial 3,42 6.6 0.00092 0.26 Model. 15, 676–704.
Risk assessment episodes Class 1,14 0.84 0.37 0.03 Galatzer-Levy, I.R., Bonanno, G.A., Bush, D.E., Ledoux, J.E., 2013. Heterogeneity in threat
Trial 3,42 13.15 0.0000034 0.28 extinction learning: substantive and methodological considerations for identifying
Class:Trial 3,42 0.29 0.83 0.01 individual difference in response to stress. Front. Behav. Neurosci. 7, 55.
Freezing bouts Class 1,14 0.0033 0.95 0.0 Galatzer-Levy, I.R., Moscarello, J., Blessing, E.M., Klein, J., Cain, C.K., LeDoux, J.E., 2014.
Trial 3,42 7.32 0.00047 0.27 Heterogeneity in signaled active avoidance learning: substantive and methodological
Class:Trial 3,42 2.57 0.07 0.11 relevance of diversity in instrumental defensive responses to threat cues. Front. Syst.
Erratic movements Class 1,14 2.3 0.15 0.054 Neurosci. 8, 179.
Trial 3,42 1.7 0.19 0.072 Hall, D., Suboski, M.D., 1995a. Sensory preconditioning and second-order conditioning of
alarm reactions in zebra danio fish (Brachydanio rerio). J. Comp. Psychol. 109,
Class:Trial 3,42 0.5 0.68 0.023
76–84.
Distance traveled Class 1,14 0.82 0.38 0.017
Hall, D., Suboski, M.D., 1995b. Visual and olfactory stimuli in learned release of alarm
Trial 3,42 7.93 0.00027 0.29 reactions by zebra danio fish (Brachydanio rerio). Neurobiol. Learn. Mem. 63,
Class:Trial 3,42 0.47 0.70 0.023 229–240.
Crossings Class 1,14 0.18 0.89 0.0006 Jovanovic, T., Kazama, A., Bachevalier, J., Davis, M., 2012. Impaired safety signal
Trial 3,42 3.07 0.038 0.11 learning may be a biomarker of PTSD. Neuropharmacology 62, 695–704.
Class:Trial 3,42 0.14 0.93 0.005 Jovanovic, T., Norrholm, S.D., Blanding, N.Q., Davis, M., Duncan, E., Bradley, B., Ressler,
K.J., 2010. Impaired fear inhibition is a biomarker of PTSD but not depression.
Depress. Anxiety 27, 244–251.
Jovanovic, T., Norrholm, S.D., Fennell, J.E., Keyes, M., Fiallos, A.M., Myers, K.M., Davis,
Acknowledgements
M., Duncan, E.J., 2009. Posttraumatic stress disorder may be associated with im-
paired fear inhibition: relation to symptom severity. Psychiatry Res. 167, 151–160.
We recognize financial support and fellowships from Conselho Kalueff, A.V., Gebhardt, M., Stewart, A.M., Cachat, J.M., Brimmer, M., Chawla, J.S.,
Craddock, C., Kyzar, E.J., Roth, A., Landsman, S., Gaikwad, S., Robinson, K., Baatrup,
Nacional de Desenvolvimento Científico e Tecnológico (CNPq),
E., Tierney, K., Shamchuk, A., Norton, W., Miller, N., Nicolson, T., Braubach, O.,
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior Gilman, C.P., Pittman, J., Rosemberg, D.B., Gerlai, R., Echevarria, D., Lamb, E.,
(CAPES), and Federal University of Santa Maria (UFSM). D.L.M. was Neuhauss, S.C., Weng, W., Bally-Cuif, L., Schneider, H., Neuroscience, Zebrafish,
recipient of fellowship from Fundo de Incentivo à Pesquisa (FIPE/ Research, C., 2013. Towards a comprehensive catalog of zebrafish behavior 1.0 and
beyond. Zebrafish 10, 70–86.
UFSM). A.S.P, B.D.F., and N.J.M. were recipients of fellowship from Krypotos, A.M., Effting, M., Kindt, M., Beckers, T., 2015. Avoidance learning: a review of
CAPES. F.V.S. was recipient of fellowship from CNPq. M.A.R., N.V.B., theoretical models and recent developments. Front. Behav. Neurosci. 9, 189.
J.B.T.R., and D.B.R. are recipients of CNPq research productivity grant. Lawrence, C., 2007. The husbandry of zebrafish (Danio rerio): A review. Aquaculture 269,
1–20.
The funders had no influence on the study design, the collection, ana- LeDoux, J.E., Moscarello, J., Sears, R., Campese, V., 2017. The birth, death and resur-
lysis and interpretation of data, as well as writing and submission of this rection of avoidance: a reconceptualization of a troubled paradigm. Mol. Psychiatry
manuscript. 22, 24–36.
Levin, E.D., Kalueff, A.V., Gerlai, R.T., 2015. Perspectives on zebrafish neurobehavioral
pharmacology. Pharmacol. Biochem. Behav. 139 (Pt B), 93.
References Lima, M.G., Silva, R.X., Silva Sde, N., Rodrigues Ldo, S., Oliveira, K.R., Batista Ede, J.,
Maximino, C., Herculano, A.M., 2016. Time-dependent sensitization of stress re-
sponses in zebrafish: A putative model for post-traumatic stress disorder. Behav.
Battle, D.E., 2013. Diagnostic and statistical manual of mental disorders (DSM). Codas 25,
Processes 128, 70–82.
191–192.
Lima, M.G., Silva Sde, N., Silva, R.X., Oliveira, K.R., Batista Ede, J., Maximino, C.,
Blank, M., Guerim, L.D., Cordeiro, R.F., Vianna, M.R., 2009. A one-trial inhibitory
Herculano, A.M., 2015. Putative involvement of the nitrergic system on the con-
avoidance task to zebrafish: rapid acquisition of an NMDA-dependent long-term
solidation, but not initiation, of behavioral sensitization after conspecific alarm
memory. Neurobiol. Learn. Mem. 92, 529–534.
substance in zebrafish. Pharmacol. Biochem. Behav. 139 (Pt B), 127–133.
Bolles, R.C., Grossen, N.E., Hargrave, G.E., Duncan, P.M., 1970. Effects of conditioned
Manuel, R., Gorissen, M., Roca, C.P., Zethof, J., van de Vis, H., Flik, G., van den Bos, R.,
appetitive stimuli on the acquisition and extinction of a runway response. J. Exp.
2014. Inhibitory avoidance learning in zebrafish (Danio rerio): effects of shock in-
Psychol. 85, 138–140.
tensity and unraveling differences in task performance. Zebrafish 11, 341–352.
Bonanno, G.A., Mancini, A.D., Horton, J.L., Powell, T.M., Leardmann, C.A., Boyko, E.J.,
Mathuru, A.S., Kibat, C., Cheong, W.F., Shui, G., Wenk, M.R., Friedrich, R.W., Jesuthasan,
Wells, T.S., Hooper, T.I., Gackstetter, G.D., Smith, T.C., Cohort, Millennium, Study,
S., 2012. Chondroitin fragments are odorants that trigger fear behavior in fish. Curr.
T., 2012. Trajectories of trauma symptoms and resilience in deployed U.S. military
Biol. 22, 538–544.
service members: prospective cohort study. Br. J. Psychiatry 200, 317–323.
Maximino, C., Lima, M.G., Oliveira, K.R., Batista Ede, J., Herculano, A.M., 2013. "Limbic
Campos, A.C., Fogaca, M.V., Aguiar, D.C., Guimaraes, F.S., 2013. Animal models of an-
associative" and "autonomic" amygdala in teleosts: a review of the evidence. J. Chem.
xiety disorders and stress. Rev. Bras. Psiquiatr. 35 (Suppl. 2), S101–111.
Neuroanat. 48–49, 1–13.
Canzian, J., Fontana, B.D., Quadros, V.A., Rosemberg, D.B., 2017. Conspecific alarm
Maximino, C., Silva, R.X., da Silva Sde, N., Rodrigues Ldo, S., Barbosa, H., de Carvalho,
substance differently alters group behavior of zebrafish populations: putative in-
T.S., Leao, L.K., Lima, M.G., Oliveira, K.R., Herculano, A.M., 2015. Non-mammalian
volvement of cholinergic and purinergic signaling in anxiety- and fear-like responses.
models in behavioral neuroscience: consequences for biological psychiatry. Front.
Behav. Brain Res. 320, 255–263.
Behav. Neurosci. 9, 233.
Caramillo, E.M., Khan, K.M., Collier, A.D., Echevarria, D.J., 2015. Modeling PTSD in the
Mezzomo, N.J., Silveira, A., Giuliani, G.S., Quadros, V.A., Rosemberg, D.B., 2016. The
zebrafish: are we there yet? Behav. Brain Res. 276, 151–160.
role of taurine on anxiety-like behaviors in zebrafish: A comparative study using the
Choi, J.S., Cain, C.K., LeDoux, J.E., 2010. The role of amygdala nuclei in the expression of
novel tank and the light-dark tasks. Neurosci. Lett. 613, 19–24.
auditory signaled two-way active avoidance in rats. Learn. Mem. 17, 139–147.
Muthén, B., 2004. SAGE publications ltd. Latent variable Analysis: Growth Mixture
Collier, A.D., Khan, K.M., Caramillo, E.M., Mohn, R.S., Echevarria, D.J., 2014. Zebrafish
Modeling and Related Techniques for Longitudinal Data.
and conditioned place preference: a translational model of drug reward. Prog.
Nathan, F.M., Ogawa, S., Parhar, I.S., 2015. Kisspeptin1 modulates odorant-evoked fear
Neuropsychopharmacol. Biol. Psychiatry 55, 16–25.
response via two serotonin receptor subtypes (5-HT1A and 5-HT2) in zebrafish. J.
de Mooij-van Malsen, A.J., Vinkers, C.H., Peterse, D.P., Olivier, B., Kas, M.J., 2011. Cross-
Neurochem. 133, 870–878.
species behavioural genetics: A starting point for unravelling the neurobiology of
Ogawa, S., Nathan, F.M., Parhar, I.S., 2014. Habenular kisspeptin modulates fear in the
human psychiatric disorders. Prog. Neuropsychopharmacol. Biol. Psychiatry 35,
zebrafish. Proc. Natl. Acad. Sci. U. S. A. 111, 3841–3846.
1383–1390.
Parra, K.V., Adrian, Jr.J.C., Gerlai, R., 2009. The synthetic substance hypoxanthine3-N-
Dinsmoor, J.A., 1977. Escape, avoidance, punishment: where do we stand? J. Exp. Anal.
oxide elicits alarm reactions in zebrafish (Danio rerio). Behav. Brain Res. 205 (2),
Behav. 28, 83–95.
336–341.
Dinsmoor, J.A., 2001. Stimuli inevitably generated by behavior that avoids electric shock
Piato, A.L., Capiotti, K.M., Tamborski, A.R., Oses, J.P., Barcellos, L.J., Bogo, M.R., Lara,
are inherently reinforcing. J. Exp. Anal. Behav. 75, 311–333.
D.R., Vianna, M.R., Bonan, C.D., 2011. Unpredictable chronic stress model in

41
C. Maximino et al. Behavioural Processes 149 (2018) 35–42

zebrafish (Danio rerio): behavioral and physiological responses. Prog. Biochem. Behav. 153, 18–31.
Neuropsychopharmacol. Biol. Psychiatry 35 (2), 561–567. Shalev, A.Y., 2009. Posttraumatic stress disorder and stress-related disorders. Psychiatr.
Portavella, M., Torres, B., Salas, C., 2004a. Avoidance response in goldfish: emotional and Clin. North Am. 32, 687–704.
temporal involvement of medial and lateral telencephalic pallium. J. Neurosci. 24, Song, C., Liu, B.P., Zhang, Y.P., Peng, Z., Wang, J., Collier, A.D., Echevarria, D.J.,
2335–2342. Savelieva, K.V., Lawrence, R.F., Rex, C.S., Meshalkina, D.A., Kalueff, A.V., 2018.
Portavella, M., Torres, B., Salas, C., Papini, M.R., 2004b. Lesions of the medial pallium, Modeling consequences of prolonged strong unpredictable stress in zebrafish: com-
but not of the lateral pallium, disrupt spaced-trial avoidance learning in goldfish plex effects on behavior and physiology. Prog. Neuropsychopharmacol. Biol.
(Carassius auratus). Neurosci. Lett. 362, 75–78. Psychiatry 81, 384–394.
Proust-Lima, C., Dartigues, J.F., Jacqmin-Gadda, H., 2016. Joint modeling of repeated Speedie, N., Gerlai, R., 2008. Alarm substance induced behavioral responses in zebrafish
multivariate cognitive measures and competing risks of dementia and death: a latent (Danio rerio). Behav. Brain Res. 188, 168–177.
process and latent class approach. Stat. Med. 35, 382–398. Stewart, A.M., Ullmann, J.F., Norton, W.H., Parker, M.O., Brennan, C.H., Gerlai, R.,
Prus, A.J., James, J.R., Rosecrans, J.A., 2009. Conditioned place preference. In: Kalueff, A.V., 2015. Molecular psychiatry of zebrafish. Mol. Psychiatry 20, 2–17.
Buccafusco, J.J. (Ed.), Methods of Behavior Analysis in Neuroscience, 2nd ed. Boca Stewart, A.M., Yang, E., Nguyen, M., Kalueff, A.V., 2014. Developing zebrafish models
Raton (FL). relevant to PTSD and other trauma- and stressor-related disorders. Prog.
Quadros, V.A., Silveira, A., Giuliani, G.S., Didonet, F., Silveira, A.S., Nunes, M.E., Silva, Neuropsychopharmacol. Biol. Psychiatry 55, 67–79.
T.O., Loro, V.L., Rosemberg, D.B., 2016. Strain- and context-dependent behavioural Vargas, J.P., López, J.C., Portavella, M., 2012. Amygdala and Emotional Learning in
responses of acute alarm substance exposure in zebrafish. Behav. Processes 122, Vertebrates – A Comparative Perspective. InTech.
1–11. von Frisch, K., 1938. Zur psychologie des Fische-Schwarmes. Naturwissenschaften 26,
Ruhl, T., Zeymer, M., von der Emde, G., 2017. Cannabinoid modulation of zebrafish fear 601–606.
learning and its functional analysis investigated by c-Fos expression. Pharmacol.

42

You might also like