Anxiolytic effects of fluoxetine and nicotine exposure on exploratory behavior in zebrafish

Subjects

Experimental fish were bred from adult ScientificHatcheries strain (Huntingdon, CA) that has been maintained in our facility.Water in our Aquaneering Inc. (San Diego, CA) system was constantly circulating and kept at a temperature of 28.5°C on a 14 hour light:10 hour dark cycle. The fish were fed a diet of brine shrimp twice and flake food (Tetramin) once for a total of three daily feedings. At the time of data collection, the fish were four months old and housed in three-liter tanks in groups of five to achieve maximal growth rates. Though zebrafish stocked at this density are known to develop social hierarchies that can influence stress and behavior (Pavlidis et al., 2013), we randomly assigned individuals to a drug treatment group such that these effects should be equally distributed across treatments. All aspects of this study were approved by the University of Idaho’s Animal Care and Use Committee under protocol 2014-14.

Dosing

Fluoxetine (generic (Teva Pharmaceuticals) from Wal Mart) and nicotine (Sigma Aldrich) treatments were administered at concentrations of 10 mg/L for the fluoxetine and 1 mg/L for the nicotine. These doses vary from standard doses in the zebrafish literature. Fluoxetine is often given at concentrations up to 100 μg/L, but administered chronically over a two-week period (Egan et al., 2009). We used a higher dose than the chronic concentrations reported in the literature; however, it is important to note that this choice could yield non-target effects due to higher concentrations. Nicotine is often administered as a ditartrate salt at concentrations up to 100 mg/L (Levin, Bencan & Cerutti, 2007). We used pure nicotine and were unsure at the time of the experiment how the two forms compared with each other. We chose our dose based on the LD50 concentration (4 mg/L) to avoid lethal effects on our subjects. Each drug was dissolved in system water to make a working solution each morning of administration. A third treatment of only system water served as a control. Fish were netted from their home tank and immediately placed into a beaker containing 100 mL of one of the three treatments. After three minutes of exposure to the drug dose, the fish were transferred to an open field test tank filled with untreated system water for behavioral recording. Dosing and behavioral observations were made on one fish at a time and the treatment type and order were randomized across individuals.

Behavior assay & video tracking

The fish were placed in a rectangular tank with interior dimensions measuring 25 cm wide, 12 cm high (from water level to bottom), and 6 cm thick (front to back). The volume of water in the tank was approximately 2 L. Each fish was filmed for four minutes (240 s) at 25 frames/second beginning from the time that the subject entered the water. The camera and operator were hidden behind a blind during the recorded observation time. The tank was backlit with an opaque diffuser for the purposes of creating a silhouetted object for motion tracking. After the four-minute period, the fish was netted out of the test tank, placed into its own individual 1.5 L housing and returned to the main system. Observations were recorded over three days between the hours of 10:00 am and 2:00 pm. After all subjects had been recorded, weight and standard length measurements were obtained by first anesthetizing the individual in MS-222 solution and blotting excess water with a paper towel. At this time, we also recorded the sex of the individual using visual cues: larger, rounded abdomen and dull fins for females, smaller and leaner abdomen and bright yellow fins for males.

Videos were digitally tracked using ProAnalyst^® (Xcitex, Cambridge, MA). Tracking began with the frame in which the fish hit the surface of the water, and proceeded to the end of the video. The tracking data were imported into R for cleaning and processing. Each track was truncated to exclude the first five seconds during which the fish would sink, but remain otherwise motionless, as it recovered from the initial shock of being released from the net. Tracks were then standardized to 4 min, or 6,000 frames. We computed velocity from the x–y data points. Since the tracking software did not always track the exact same position on the fish, velocity was estimated using the change in coordinates between two frames before and two frames after the focal frame. This algorithm sufficiently smoothed the speed data while retaining detail at small time intervals.

Analysis

Multivariate

The full model Type-II MANOVA included the effects of weight, sex, drug treatment, and the sex by drug interaction on average depth, variance of depth, average speed, variance of speed, percent of time spent in the top half, number of crosses into the top half, latency to enter the top half, and proportion of time spent in the middle third horizontally (i.e., away from the edges). There was a non-significant effect of sex (Λ = 0.17896, F_8,73 = 1.9889, p = 0.05974) and a significant effect of drug treatment (Λ = 0.56646, F_16,148 = 3.6551, p = 0.00001305) on behavior, but no significant interaction. There was no significant effect of weight as a co-variate, and including weight in the model showed no improvement over removing it (Λ = 0.95793, F_5,76 = 0.66755, p = 0.6492). With the reduced model, we observed a significant effect of sex (Λ = 0.22404, F_8,74=2.6707, p = 0.01237) and drug treatment (Λ = 0.56659, F_16,150 = 3.7057, p = 0.00001014). Therefore, for all subsequent analyses we considered only the effects of sex, drug treatment, and the interaction term.

Individual components of behavior

We observed no significant interactions between sex and drug treatment in any of the individual behavior components (see Table 1), consistent with the results of the MANOVA above. All components indicated a significant effect of drug treatment (p < 0.05) except for freezing occurrence and freezing duration. The subsequent descriptions describe the results of the post-hoc pairwise comparisons of the drug treatments using the least-squared means and Tukey adjusted p-values based on 3 tests. We also observed a significant effect of sex with regard to average swimming speed (F_1,81 = 10.7178, p = 0.001562) and consistency (variance) of swimming speed (F_1,81 = 13.9196, p = 0.0003528). Males were on average faster than females, but also exhibited less consistency in their swimming speeds. These were the only instances in which the sexes differed in their behavior.

Freezing behavior

Freezing behavior is a commonly observed anxiety related behavior in zebrafish (Egan et al., 2009). Of the 87 individuals observed, 52 exhibited freezing behavior. Though males tend to be more likely to freeze than females on average, this difference was not statistically significant (χ² = 3.7866, p = 0.05167). We also failed to observe a significant effect of drug treatment on freezing occurrence (χ² = 3.7964, p = 0.14983) as well as a sex by drug interaction (χ² = 0.3949, p = 0.82083). For freezing duration, or latency to explore, we only included the 52 individuals that exhibited freezing behavior (control: F = 11, M = 10; fluoxetine: F = 7, M = 8; nicotine: F = 6, M = 10). This improved the assumptions of normality required for the ANOVA. Results of the type II ANOVA suggest that neither sex nor drug treatment have any significant effect on freezing duration (Sex: F_1,46 = 1.9604, p = 0.1682; Drug: F_2,46 = 1.3707, p = 0.2641). Figure 1 shows the results of freezing behaviors.

Speed

When analyzing only the active swimming data from the trials, fish from both drug treatments appear to reduce their average swimming speed compared with the control, however this pattern is only significant in the nicotine treatment (t = 3.373, p = 0.0032, see Fig. 2). Drugged fish also swam at a more consistent speed than the undrugged control fish (F_2,81 = 4.0654, p = 0.0207731), but again this trend was only significant in the nicotine treatment (t = 2.818, p = 0.0166).

Depth

Both the subjects dosed with nicotine and fluoxetine positioned themselves higher in the water column than the control fish (nicotine: t = − 2.462, p = 0.0417; fluoxetine: t = − 4.711, p < .0001). Fish dosed with fluoxetine explored more of the water column than control subjects (t = − 3.172, p = 0.0060). Subjects dosed with nicotine also exhibited more variation in depth use on average than the control subjects, but this difference was not significant (see Fig. 3).

We also divided the tank into two discrete and equal vertical zones and compared the proportion of time spent in the upper half (Fig. 4). Subjects dosed with fluoxetine tended to spend more than twice as much time in the upper half as control subjects and this difference is significant (t = − 3.883, p = 0.0006). Subjects in both the nicotine and fluoxetine treatments exhibited a reduced latency time to first enter the top half than control subjects (nicotine: t = 3.333, p = 0.0037; fluoxetine: t = 2.652, p = 0.0258). When comparing the total number of visits to the top half, only the fluoxetine group showed a significant increase over the control (t = − 3.801, p = 0.0008).

Differences in fluoxetine and nicotine behavioral responses

Small prey fish such as zebrafish tend to behave in such a way as to reduce risk of predation. When placed in a novel open field, such behavioral strategies include diving to the bottom and remaining motionless (Egan et al., 2009), and avoiding potentially risky locations such as the surface of the water (Wilson & Godin, 2009; Oswald, Singer & Robison, 2013). Exposure to anxiolytic drugs alters these behaviors in ways that may indicate an association between anxiety related behaviors and risk management. We observed a decrease in bottom dwelling and an increase in time spent in the top half of the tank in fish exposed to fluoxetine (Figs. 3 and 4). This is consistent with patterns observed by Egan et al. (2009) who also report an increased use of the top of the water column by zebrafish exposed to fluoxetine. However, the study by Egan et al. (2009) also reports a reduction in freezing bouts and freezing time, a pattern we failed to observe. One explanation for this discrepancy could be differing effects of chronic and acute dosing. Fluoxetine is metabolized into norfluoxetine, its active metabolite, in the liver by cytochrome P450 enzymes (Rasmussen et al., 1995). It then travels through the bloodstream to the brain where it blocks the reuptake of serotonin (Beasley, Masica & Potvin, 1992). Metabolism of the drug could delay its effect until after the animal had already recovered from freezing behavior.While most fluoxetine studies utilize chronic exposure, we have shown that similar behavioral changes can occur with just a single acute dose. Acute exposure to fluoxetine has also been shown to reduce cortisol levels of zebrafish exposed to a stressful environment (De Abreu, Koakoski & Ferreira, 2014). We speculate that the behaviors we observed may be due to a reduction in physiological stress response resulting from exposure to the drugs, though more experiments are needed to confirm this.

We observed changes in swimming speed, average depth, and latency to enter the top in fish exposed to nicotine. Fish exposed to nicotine were quicker to enter the top and swam higher in the water column on average compared to control fish. This is consistent with a reduction in anxiety related behaviors as seen in the fluoxetine treatment group. Exposure to nicotine and fluoxetine appeared to decrease swimming speed while increasing the consistency at which the fish swam. The increased consistency (reduction of individual variance) might be explained by a reduction in anxiety, where individuals that are calm should move at a fairly normal and constant pace, while anxious individuals may constantly alter their swimming speeds in an erratic fashion Gerlai, Fernandes & Pereira (2009). Egan et al. (2009) reported an increase in average swim speed with exposure to fluoxetine, which contrasts with our observations of slower average swim speeds with exposure to either fluoxetine or nicotine. Sackerman et al. (2010) suggests that nicotine may have sedating effects which could account for the slower swim speeds. However, we also observe slower average swim speeds in the fluoxetine treatment, and though the difference is not statistically different from the control, it is also not different from the nicotine effect. We observed a similar pattern in the nicotine treatment with respect to the time spent at the top and the variation in depth use, where the nicotine treatment was statistically indistinguishable from both the control and the fluoxetine treatments. In these two instances, it is likely that the nicotine is having an anxiolytic effect, but that we used too low a dose to observe an effect that is different from the control. Sackerman et al. (2010) also failed to observe an effect of nicotine on swim depth using a low dose of 25 mg/L, but noted that higher doses such as 50 mg/L and 100 mg/L do produce a significant effect (Levin, Bencan & Cerutti, 2007). Our dose of 1 mg/L is noticeably lower than other studies of nicotine in adult Zebrafish, accounting for the our use of pure nicotine liquid while the other studies used a nicotine tartrate salt (Levin, Bencan & Cerutti, 2007; Sackerman et al., 2010). It should be noted that the relationship between the tartrate salt and pure form is about 0.325, such that a concentration of 100 mg/l of the tartrate equates to a concentration of 32.5 mg/l of pure nicotine (Matta et al., 2007).

Both nicotine and fluoxetine affected behavior in ways indicative of a reduction of anxiety. However, the two drugs also appear to affect different components of behavior. Nicotine had its highest effect on swimming speed, while fluoxetine mostly affected behaviors related to vertical positioning. This suggests that anxiety is not a simple condition, but rather a complex idea encompassing a number of components that are sometimes correlated, but not always connected. These behavioral components may be separated by different physiological pathways which could explain why different classes of drugs affect specific behaviors.

The effect of sex on behavior and drug efficacy

Sex differences in anxiety behaviors have been described in a number of species including rats (Mehta, Wang & Redei, 2013), stickleback (King et al., 2013), and guppies (Harris et al., 2010). While most of these studies find that males are typically more bold (less anxious) than females, our lab has previously observed the opposite trend in the Scientific Hatcheries strain of zebrafish with regard to association with humans, vertical position, and feeding latency in individual home-tank observations (Oswald et al., 2013; Oswald, Singer & Robison, 2013; Benner et al., 2010). These differences are the basis for our inquiry as to whether substances known to alter these behaviors might work at different efficacy in males and females. In the present study, we only observe significant behavioral differences between the sexes with respect to swimming speed. While males swim slightly faster than females, it’s the females that swim at a more constant rate. In addition, males seem to show a higher probability to exhibit freezing behavior across all three treatments, and even though this trend isn’t statistically significant, it still leads us to suggest that males could be behaving with higher anxiety levels than females.

With the active swimming behaviors, we fail to observe differences between the sexes, and across all of the behaviors, the data do not suggest any indications of sex-specific effects of either drug. There is plenty of literature in mammalian models that contradict these findings (Mitic et al., 2013; Leuner, Mendolia-Loffredo & Shors, 2004; Lifschytz et al., 2006; Monleón et al., 2002; Hodes et al., 2010). One possible explanation for our lack of sex-specific effects stems from our general lack of sex differences in the behaviors analyzed, and perhaps a baseline difference in behavior is necessary to elicit a sex-specific effect. The results of Mitic et al. (2013); Leuner, Mendolia-Loffredo & Shors (2004) and Lifschytz et al. (2006) in rats all observe sex-specific responses to fluoxetine only when the sexes differed in behaviors without the drug. We do not have adequate data to confirm this explanation and more experimentation along with physiological data would be necessary.

Another possible explanation for our lack of sex-specific drug effects could be our choice of dose. Our choice of 1mg/L of nicotine is quite low compared with other studies in zebrafish (Levin, Bencan & Cerutti, 2007; Sackerman et al., 2010), and while our dosage of fluoxetine was much higher than is typically reported (Egan et al., 2009; Wong, Oxendine & Godwin, 2013), it is typically administered chronically. We would also like to note that the sex-specific results of Faraday, O’Donoghue & Grunberg (1999) utilizing nicotine in rats was only observed in one of the two strains used. Zebrafish are highly genetically diverse (Parichy, 2015) and strain differences in behavior (Benner et al., 2010; Egan et al., 2009) and drug efficacy (Sackerman et al., 2010) have been reported. Therefore the possibility exists for sex-dependent drug effects to be observed in another strain.

Finally, we cannot dismiss the possibility that zebrafish simply don’t exhibit sex-specific effects with fluoxetine or nicotine. While there is no literature in this species to compare our results with, a recently published study utilizing medaka (Oryzias latipes), another small teleost fish from southeast Asia, fails to find sex-specific effects of chronic fluoxetine on many of the same behaviors described in the present study (Ansai et al., 2016). More research is necessary to confirm any of the explanations given for our lack of observed sex-drug interactions. The absence of studies considering sex-specific effects of drugs is problematic if zebrafish are to remain a relevant model of pharmacology research. The topic has become a concern in all animal models that NIH is going to start requiring all animal research to include sex as part of the study unless deemed unnecessary (Clayton & Collins, 2014). If it turns out that strain is a major factor influencing our results, then the abundance of genetically diverse populations could make zebrafish an exciting tool to aid in the growing field of pharmakogenetics and personalized medicine in which genetic background, among other traits, will be important for determining what drugs will be most effective for treating disorders.