Here, we present a comprehensive behavioral test battery, including the novel tank, Shoaling, and social preference tests, to effectively determine the potential neurotoxic effects of chemicals (e.g., methamphetamine and glyphosate) on adult zebrafish using a single tank. This method is relevant to neurotoxicity and environmental research.
The presence of neuropathological effects proved to be, for many years, the main endpoint for assessing the neurotoxicity of a chemical substance. However, in the last 50 years, the effects of chemicals on the behavior of model species have been actively investigated. Progressively, behavioral endpoints were incorporated into neurotoxicological screening protocols, and these functional outcomes are now routinely used to identify and determine the potential neurotoxicity of chemicals. Behavioral assays in adult zebrafish provide a standardized and reliable means to study a wide range of behaviors, including anxiety, social interaction, learning, memory, and addiction. Behavioral assays in adult zebrafish typically involve placing the fish in an experimental arena and recording and analyzing their behavior using video tracking software. Fish can be exposed to various stimuli, and their behavior can be quantified using a variety of metrics. The novel tank test is one of the most accepted and widely used tests to study anxiety-like behavior in fish. The shoaling and social preference tests are useful in studying the social behavior of zebrafish. This assay is particularly interesting since the behavior of the entire shoal is studied. These assays have proven to be highly reproducible and sensitive to pharmacological and genetic manipulations, making them valuable tools for studying the neural circuits and molecular mechanisms underlying behavior. Additionally, these assays can be used in drug screening to identify compounds that may be potential modulators of behavior.
We will show in this work how to apply behavioral tools in fish neurotoxicology, analyzing the effect of methamphetamine, a recreational drug, and glyphosate, an environmental pollutant. The results demonstrate the significant contribution of behavioral assays in adult zebrafish to the understanding of the neurotoxicological effects of environmental pollutants and drugs, in addition to providing insights into the molecular mechanisms that may alter neuronal function.
The zebrafish (Danio rerio) is a popular model vertebrate species for ecotoxicology, drug discovery, and safety pharmacology studies. Its low cost, well-established molecular genetic tools, and conservation of key physiological processes involved in the morphogenesis and maintenance of the nervous system make zebrafish an ideal animal model for neuroscience research, including neurobehavioral toxicology1,2. The main endpoint for evaluating the neurotoxicity of a chemical was, until recently, the presence of neuropathological effects. Lately, however, behavioral endpoints have been incorporated into neurotoxicological screening protocols, and these functional outcomes are now commonly used to identify and determine the potential neurotoxicity of chemicals3,4. Moreover, behavioral endpoints are highly relevant from an ecological point of view, as even a very mild behavioral change in fish could endanger the survival of the animal in natural conditions5.
One of the most used behavioral assays in adult zebrafish research is the novel tank test (NTT), which measures anxiety-like behavior6,7. In this assay, fish are exposed to novelty (fish are placed in an unfamiliar tank), a mild aversive stimulus and their behavioral responses are observed. NTT is used to assess basal locomotor activity, geotaxis, freezing, and erratic movements of fish, principally. Erratic8 is characterized by abrupt changes of direction (zigzagging) and repeated episodes of accelerations (darting). It is an alarm reaction and is usually observed before or after freezing episodes. Freezing behavior corresponds to a complete cessation of the fish's movements (except for opercular and ocular movements) while on the bottom of the tank, as distinguished from immobility caused by sedation, which causes hypolocomotion, akinesia, and sinking8. Freezing is usually related to a high state of stress and anxiety and is also part of submissive behavior. Complex behaviors are excellent indicators of the state of anxiety of animals. NTT has been shown to be sensitive to pharmacological and genetic manipulation9, making it a valuable tool for studying the neural basis of anxiety and related disorders.
Zebrafish are a highly social species, so we can measure a wide range of social behaviors. The shoaling test (ST) and the social preference test (SPT) are the most used assays to assess social behavior10. The ST measures the tendency of fish to group together11 by quantifying their spatial behavior and movement patterns. ST is useful for studying group dynamics, leadership, social learning, and understanding the social behavior of many fish species12. The SPT in adult zebrafish was adapted from Crawley's preference for social novelty test for mice13 and quickly became a popular behavioral assay for the study of social interaction in this model species14. These two tests have also been adapted for use in drug screening assays and have shown promise for identifying novel compounds that modulate social behavior15,16.
In general, behavioral assays in adult zebrafish are powerful tools that can provide valuable information on the behavior mechanisms or the neurophenotypes of active compounds and abused drugs17. This protocol details how to implement these behavioral tools7 with basic material resources and how to apply them in toxicity assays to characterize the effects of a wide range of neuroactive compounds. In addition, we will see that the same tests can be applied to assess the neurobehavioral effects of acute exposure to a neuroactive compound (methamphetamine) but also to characterize these effects after chronic exposure to environmental concentrations of a pesticide (glyphosate).
Strict compliance with ethical standards guarantees the welfare and proper treatment of the zebrafish used for experimentation. All experimental procedures were carried out under the guidelines established by the Institutional Animal Care and Use Committees (CID-CSIC). The protocols and results presented below were performed under the license granted by the local government (agreement number 11336).
1. Animal housing for behavioral testing
Figure 1: Experimental setups. Three configurations of the square tank to study a wide range of behaviors in adult zebrafish. Please click here to view a larger version of this figure.
Figure 2: Experimental timeline. Two planning proposals for the recording of behavioral assays. Please click here to view a larger version of this figure.
2. Experimental configurations of the tank
3. Video recording for behavioral tests
4. Analysis of recorded videos
5. Statistical analysis
In this section, we will look at some possible applications of these behavioral tools in fish neurotoxicology. The following results correspond to the characterization of the acute or binge effects of methamphetamine (METH), a recreational drug, and the sub-chronic effects of glyphosate, one of the main herbicides found in aquatic ecosystems.
Characterization of a methamphetamine binge neurotoxicity model in adult zebrafish
When evaluating the effect of 40 mg/L METH on NTT (Figure 3), the Kruskal-Wallis test confirmed that the exposed animals presented a positive geotaxis, characterized by a decrease in the exploration time in the upper zone of the experimental tank (H(2) = 35.964, P = 1.55 x 10-8), as well as in the distance traveled in this part (H(2) = 32.272, P = 9.82 x 10-8), and in the number of visits (H(2) = 36.527, P = 1.17 x 10-8). We also observed a significant increase in the latency time preceding the first visit to the upper zone (H(2) = 17.264, P = 0.00018). It is important to remark that the differences observed in the parameters measured in the NTT after METH exposure are consistent over time, as confirmed by the Bonferroni correction (P > 0.8). A significant effect of exposure time was found for freezing behavior (H(2) = 13.120, P = 0.0014).
Figure 3: Anxiety-like behavior assessed in standard 6-min novel tank test (NTT) of adult zebrafish exposed to 40 mg/L methamphetamine (METH) for 3 h and 48 h. Data from each experiment were normalized to the corresponding control values. The combined data is reported as a scatter plot with the median (n = 14-15), **p < 0.01, ***p < 0.001; Kruskal Wallis test with Bonferroni correction for NTT endpoints. Data from 2 independent experiments. This figure has been reproduced with permission from Bedrossiantz et al.15. Please click here to view a larger version of this figure.
Freezing movements can be quantified by assessing the frequency, latency, duration, or location of freezing. The best way to score them is undoubtedly the eye of an experienced observer, which is quite laborious and complex, so we tried an automated alternative using EthoVision software to detect freezing behavior19. We found that the number, latency, and duration of freezing attacks calculated by the software (Table 1A) correlate with good accuracy with the episodes scored manually by the observer (Table 1B). Whereas the two methods are equivalent in terms of results (P = 0.958, Student's test), we used the automated approach to assess the freezing here. After 3 h of exposure to METH, freezing time increased significantly (P = 0.0012), whereas no difference was found with the control after 48 h of exposure (P = 0.16). METH produced no effect on erratic movements at either time.
We used two experimental paradigms to evaluate the effects on social behavior after acute exposure to METH. The ST (Figure 4) revealed that the average distance and farthest distance between individuals were significantly greater for METH-treated fish (H(2) = 53.261, P = 2.72 x 10-12; H(2)=52.504, P = 3.97 x 10-12 for average and farthest interfish distances, respectively), pointing to a behavioral phenotype of social isolation. Again, we remark that no time effect was found using the Bonferroni post hoc test (P > 0.5).
Figure 4: Social behavior of adult zebrafish waterborne exposed to 40 mg/L methamphetamine (METH) for 3 h and 48 h. Shoaling test (ST) results, including the average and the farthest interfish distances. The combined data is reported as a scatter plot with the median (n = 18), *p < 0.05, **p < 0.01, ***p < 0.001; Kruskal Wallis test with Bonferroni correction. Data from 2 independent experiments. This figure has been reproduced with permission from Bedrossiantz et al.15. Please click here to view a larger version of this figure.
In the SPT (Figure 5), treated fish show a significant decrease in time spent and distance traveled in the conspecific zone (F(2,74) = 14.497, P = 4.87 x 10-6; F(2,73) = 13.461, P = 0.00001 for time spent and distance traveled in the conspecific zone, respectively). These results reaffirm the social isolation phenotype suggested by the TS results. Tukey's Honest Significant Difference (HSD) post hoc test ruled out no possible differences between the two analysis times (P > 0.5).
Figure 5: Social behavior of adult zebrafish waterborne exposed to 40 mg/L methamphetamine (METH) for 3 h and 48 h. The social preference test (SPT) results, including time and distance of the fish in each of the three virtual zones of the experimental tank: empty, center, and conspecific. Data from each experiment were normalized to the corresponding control values. The combined data is reported as a scatter plot with the median (n = 17-20), *p < 0.05, **p < 0.01, ***p < 0.001; one-way ANOVA with Dunnett's multiple comparison test. Data from 2 independent experiments. This figure has been reproduced with permission from Bedrossiantz et al.15. Please click here to view a larger version of this figure.
Behavioral effect of sub-chronic exposure to environmental levels of glyphosate
Behavioral analysis of the effects of sub-chronic exposure to 3 µg/L glyphosate on the NTT (Figure 6) reveals a significant decrease in time spent exploring the top (F2,77 = 8.744, P = 0.0004), distance traveled in this part (F2,77 = 9.118, P = 0.0003), and number of visits (F2,77 = 3.441, P = 0.037). These effects are characteristic of positive geotaxis behavior, as is the increased effect observed on the latency time preceding the first visit to the top of the tank (H(2) = 9.628, P = 0.008). The expression of erratic and freezing behaviors of the exposed animals was also analyzed in the NTT. The duration (H(2) = 17.261, P = 0.025) and number of erratic episodes (F2,76 = 10.073, P = 0.0001) were significantly increased by glyphosate. In contrast, no freezing differences were found with the control (Pearson Chi-Square(2) = 2.964, P = 0.253). Applied to an ecological context, the observations made at NTT suggest that glyphosate could significantly decrease the exploratory behavior of fish, jeopardizing their ability to survive in the wild.
Figure 6: Anxiety-like behavior assessed in standard 6-min novel tank test (NTT) of adult zebrafish exposed to 0.3 µg/L and 3 µg/L glyphosate for 2 weeks. Behavioral parameters analyzed, as well as a cartoon of the experimental tank divided into two equal virtual zones, top and bottom. Data reported as a scatter plot with the median (n = 23-29), *p < 0.05, **p < 0.01, ***p < 0.001; one-way ANOVA with Dunnett's multiple comparison test (Total distance, Distance in top, Time in top, Transitions to top, Erratic bouts, High mobility frequency) or Kruskal Wallis test with Bonferroni correction (Latency to the top, Erratic duration). No differences (P > 0.05) were found in the freezing duration and freezing bouts. Data from 2-4 independent experiments. This figure has been reproduced with permission from Faria et al.20. Please click here to view a larger version of this figure.
Schooling, non-polarized groups of conspecifics that are held together by social pressure to protect themselves from predators, is a natural tendency of Danio rerio. The school can "tighten" or "expand" depending on the animals' level of anxiety or fear, a particular visual effect that is very easy to identify experimentally (Figure 7). In the glyphosate experiment, the shoaling test revealed an increase in anxiety in fish exposed to 3 µg/L, reflected by a grouping of the shoal and thus a significant decrease in the average distance and farthest distance between individuals (F2,56 = 5.664, P = 0.006 and F2,56 = 7.413, P = 0.001, for the average and farthest interfish distances, respectively) compared to the control.
Figure 7: Social behavior of adult zebrafish waterborne exposed to 0.3 µg/L and 3 µg/L glyphosate for 2 weeks. Data reported as scatter plot with the median (n = 19-20), *p < 0.05, **p < 0.01, ***p < 0.001; one-way ANOVA with Dunnett's multiple comparison test (Average interfish distance and Farthest distance) Data from 2 to 4 independent experiments. This figure has been reproduced with permission from Faria et al.20. Please click here to view a larger version of this figure.
Table 1: An approximation of freezing behavior using an automated analysis. Data reported in this table come from the same recording (Video 1) analyzed with two different methods. (A) Approximation of freezing behavior by automated calculation with EthoVision V13 software. The variable mobility is calculated from the change of the subject area between two samples, so it depends on the acquisition frequency of this area. We set a very low threshold of immobility (less than 3% mobility) as well as the sample rate to a minimum continuous time of 5 s (more than 150 frames). (B) Analysis of freezing behavior with Behavioral Observation Research Interactive Software (BORIS, free and open-source software). BORIS is an event-logging software for video coding and live observations. With BORIS, the observer can code the freeze episode as a state event, defining the start and end points. Please click here to download this Table.
Video 1: Control fish in the novel tank test. Please click here to download this Video.
Characteristic anxiety behaviors observed in NTT have been positively correlated with serotonin levels analyzed in brains21. For example, after exposure to para-chlorophenylalanine (PCPA), an inhibitor of 5-HT biosynthesis, fish exhibited positive geotaxis as well as decreased brain 5-HT levels22, results very similar to those obtained with METH. Therefore, the decrease in brain serotonin levels and the display of positive geotaxis in METH-exposed zebrafish suggests that the anxiety behavior produced by the drug is mediated by the serotonergic pathway. Interestingly, a similar behavioral phenotype, i.e., an anxiogenic effect on geotaxis, can be seen in adult zebrafish exposed for 2 weeks to 0.3 3 µg/L and 3 µg/L, two environmentally relevant concentrations of glyphosate. An increase in geotaxis was also previously reported for adult zebrafish with the neurotoxicant acrylamide6,23. In all these cases, this behavioral phenotype (an increase of geotaxis in the NTT, characteristic of anxiogenic substance) was associated with the diminution of monoaminergic neurotransmitter levels. Therefore, the NTT paradigm combined with neurochemical analysis of the brain provides ecologically relevant information, exploratory behavior, and foraging efficiency and connects behavior neurophenotypes with neurotransmitter modulations.
On the other hand, an impairment of social behaviors in both assays, the ST and SPT, was also observed in METH-treated fish. The result obtained in this study is consistent with several studies with rats and monkeys, where the acute and chronic exposure of the study animals to METH results in social withdrawal24. Social behavior changes associated with METH abuse have been explained in humans by impairments in social-cognitive function24. An anxiogenic effect on the shoal size was found in zebrafish exposed for 2 weeks to 3 µg/L glyphosate. We observed a phenocopy of this effect in zebrafish exposed to 53 mg/L (0.75 mM) acrylamide for 3 days6,23.
The NTT, ST, and SPT assays allow to effectively determine the potential neurotoxic effects25 of a wide range of chemicals as illustrated by the study of acute methamphetamine and sub-chronic glyphosate toxicity models in adult zebrafish. Behavior is, in toxicology, a relevant apical endpoint, characterizing the effects at organismal levels of a chemical for neurotoxicity and environmental research. Besides being a sublethal endpoint in laboratory conditions, changes in behaviors, such as exploratory or social behavior, can be deleterious in nature. Moreover, the proposed behavioral analysis battery is an easy-to-implement, semi-automated method11 and, therefore, very efficient if the assays are consciously planned (reduction principle)26. Performing these assays as a test battery using a single tank reduces the number of animals and the experimental time and waste generation.
The order of the assays in the battery is an important consideration if we want to study the response profile of an individual in each trial. For this purpose, conducting the individual assays followed (see Figure 2) allows for keeping the animal identified and to relate its exploratory behavior to its social preference. In addition, the animal’s behavioral responses can be related to other biological data, such as its neurotransmitter profile or gene expression, if the fish are kept identified until the end point of sampling (Figure 2A).
Usually, behavioral analysis allows for the observation of differences between groups. First, individual responses are calculated on the basis of animal tracking27 before pooling the data by group. Then, the means and the difference in variance with respect to the control group are compared for each behavioral parameter calculated. With shoaling analysis12, it is critical to be very clear that the unit of variance is the group of test fish, not individual fish because the behavior of each individual fish is influenced by the other fish in the shoal. This is the way used in most papers to process behavioral data28. However, it might be useful to rethink the analysis of behavioral parameters not on a parameter-by-parameter basis but as an overall response per trial. For example, one could calculate the covariance of each measurement made in a trial and report it as a different way of measuring the same thing: anxious, exploratory, or gregarious behavior. There are many ways to calculate and interpret behavioral data28,29. Depending on the number of conditions, type of tests, and image acquisition (2D or 3D)30,31 the analysis can be completely rethought in order to get the best out of the data.
The authors have nothing to disclose.
This work was supported by "Agencia Estatal de Investigación" from the Spanish Ministry of Science and Innovation (project PID2020-113371RB-C21), IDAEA-CSIC, Severo Ochoa Centre of Excellence (CEX2018-000794-S). Juliette Bedrossiantz was supported by a PhD grant (PRE2018-083513) co-financed by the Spanish Government and the European Social Fund (ESF).
Aquarium Cube shape | Blau Aquaristic | 7782025 | Cubic Panoramic 10 (10 L, 20 cm x 20 cm x 25 cm, 5 mm) |
Ethovision software | Noldus | Ethovision XT | Version 12.0 or newer |
GigE camera | Imaging Development Systems | UI-5240CP-NIR-GL | |
GraphPad Prism 9.02 | GraphPad software Inc | GraphPad Prism 9.02 | For Windows |
IDS camera manager | Imaging Development Systems | ||
LED backlight illumination | Quirumed | GP-G2 | |
SPSS Software | IBM | IBM SPSS v26 | |
uEye Cockpit software | Imaging Development Systems | version 4.90 |