Introduction

Nicotine has been shown to significantly improve cognitive function in mice, rats, monkeys, and humans (Buccafusco and Jackson 1991; Levin and Simon 1998). Attention, memory, and learning have been shown to be improved by nicotine. Other nicotinic agonists have shown similar effects (Buccafusco et al. 1995). However, some studies have not found nicotine-induced improvements or have documented negative actions (Dunnett and Martel 1990; Mundy and Iwamoto 1988; Welzl et al. 1988). This may be due to the inverted U-shaped dose–effect function seen with drugs that improve cognitive function when other less specific actions become expressed at higher doses and impair performance. In addition, task requirements in some procedures penalize retention of sensory stimuli and response patterns past a certain point. Such is the case with proactive interference, where memory of the cues of previous trials can interfere with performance on the current trial. Nicotine can impair accuracy in tasks with proactive interference (Dunnett and Martel 1990). Substantial effort is being exerted to develop novel nicotinic agonists that would effectively treat cognitive impairments associated with Alzheimer's disease, attention-deficit/hyperactivity disorder, and schizophrenia (Levin and Rezvani 2002; Newhouse and Kelton 2000; Newhouse et al. 1997). Receptor assays can be used for high-throughput screening of novel compounds, but the next step of identifying which compounds are worth pursuing in the development of new cognitive-enhancing drugs is often quite slow. The development of a small nonmammalian model for quick assessment of cognitive function could be an asset for the drug development process. Zebrafish can be useful in this effort.

There is a long history of cognitive function studies in fish. For example, decades ago, goldfish were shown to demonstrate temporal discrimination as well as avoidance discrimination (Behrend and Bitterman 1964; Eskin and Bitterman 1960; Rozin 1965). Recently, we have been developing methods for assessing cognitive function in zebrafish. Zebrafish will learn escape response, position discrimination, color discrimination, and their reversal (Arthur and Levin 2001), as well as delayed spatial alternation (Levin and Chen 2004).

Zebrafish have been shown to be useful in neurobehavioral toxicology. They are sensitive to persisting adverse cognitive effects caused by developmental exposure to the pesticide chlorpyrifos (Levin et al. 2003), much in the same way as rats. Likewise, early developmental exposure to ethanol (Bilotta et al. 2004; Loucks and Carvan 2004; Reimers et al. 2004) has been found to produce neurobehavioral impairments, which are analogous to teratological effects in mammals.

Zebrafish can also be used in the study of potential therapeutic effects of drug treatments. Recently, we have found that nicotine causes significant improvements in memory function as measured by delayed spatial alternation (Levin and Chen 2004). Zebrafish have an inverted U-shaped dose–response curve, with moderate doses improving cognitive function and higher doses having less of a positive effect. This is quite similar to the effects of nicotine and other cognitive-enhancing drugs in rodents, monkeys, and humans (Levin and Rezvani 2002; Levin and Simon 1998). However, the methods used to date have been quite labor-intensive and time-intensive.

In the current study, we used a rapid, single-session position discrimination task to determine the effects of nicotine on learning in zebrafish. Time-effect and dose-effect functions were assessed to determine the parameters for the peak effect of nicotine. Then, we investigated the role of nicotinic receptor activation in the induction and expression of nicotine effects with the use of the nicotinic receptor antagonist mecamylamine. If reliable effects of nicotine can be demonstrated in this rapid and inexpensive assay, it may be useful in the identification of promising new nicotinic and other promnestic drugs for the treatment of cognitive dysfunction.

Methods

Subjects

The experimental protocol was approved by the Duke University Institutional Review Committee for the use of animal subjects. Zebrafish (Danio rerio) were kept at approximately 28.5°C on a 12-h light/dark cycle, and behavioral testing took place during the light phase between 8:00 a.m. and 5:00 p.m. Tank water consisted of deionized H2O and sea salts (1.2 g/20 l of water, Instant Ocean). The tanks with groups of adult fish were maintained with constant filtration and aeration. Fish were fed daily with brine shrimp and flake fish food.

Drug administration

Zebrafish were immersed for 3 min in a beaker with 0, 50, 100, and 200 mg/l nicotine ditartrate. The dose of nicotine was calculated based on the weight of ditartrate salt. Delays of 0–160 min were imposed between the end of dosing and the beginning of the test session. The fish were exposed to nicotine in a separate beaker and then were placed singly into a holding tank without nicotine for the interval between exposure and testing. In the nicotine + mecamylamine experiment, nicotine (0 or 100 mg/l) was administered for 3 min, 40 min before testing. The nicotinic antagonist mecamylamine (0 or 200 mg/l) was administered either concurrently with nicotine (early mecamylamine) or later, 5 min before testing (late mecamylamine). This was done to determine the impact of high-affinity nicotinic receptor blockade either for the induction or for the expression of nicotine effect on choice accuracy. There was no nicotine exposure in either the home tank or the test chamber. All the fish were drug-naïve, and each fish was used only once. There were 10–30 fish per condition.

Behavioral testing

The fish were tested in a three-chambered test tank. A Plexiglas maze was fitted into a 29-l fish tank (22 cm wide ×44 cm long ×30 cm high we have used in previous studies (Levin et al. 2003; Levin and Chen 2004). The behavioral testing tank was divided into three compartments: one central compartment (18×22 cm) and two side compartments (13×22 cm). Vertically sliding doors (12 cm high and 10 cm wide on either side of this central start area) led to the two choice areas, one on each side. The partitions were mounted on rails so that they could be moved to 1 cm from the end wall of the tank to restrict the movement of the fish as the aversive consequence of choosing the incorrect side. A dark panel was placed along the rear side of the tank to provide an axis of orientation for position discrimination.

Following each drug treatment, the fish was placed in the middle chamber of a three-chambered tank for 60 s. After 60 s, the partitions to both sides of the chamber were lifted such that the fish could choose to swim to a particular side. As soon as the fish chose a side, the partitions were closed. After the initial choice, the partition located on the side to which the fish swam was moved such that only 1 cm remained between the partition and the end of the tank. The fish would then remain in this confined area for 10 s. After this punishment, the fish was allowed to return to the central chamber by opening the door to the central compartment, and the partition was moved back to its original position. There was then another 60-s interval until the next trial. On subsequent trials, if the fish chose the side it had chosen initially (the “incorrect” side), it would again be confined. Upon choosing the “correct” side, the fish was allowed to stay in that compartment for 30 s before being allowed to return to the central chamber. If, after 20 s, it had not made a decision, a fish net was placed into the center of the tank and moved parallel to the partitions for a maximum of 120 s in which the fish had the opportunity to make a choice. After the initial punishment run, the fish performed six subsequent trials, making the total number of trials seven per fish. Choice accuracy and response latency were recorded. Performance was assessed with three-trial blocks averaged for analysis during acquisition. Trial and response times were monitored by an electronic stopwatch.

Statistical analysis

Behavioral data were assessed by analysis of variance (ANOVA). The between-subjects factors were dosing to testing interval and nicotine dose. The repeated-measures factor was trial block. Dependent measures were percent correct performance and seconds per trial. Planned comparisons to control were made of the dose conditions.

Results

The 100-mg/l nicotine ditartrate dose caused a significant improvement in percent correct performance relative to vehicle control treatment (Fig. 1). Nicotine-induced improvement was delayed. Across the range of intervals between the end of nicotine dosing and the beginning of testing (between 0 and 160 min compared to untreated controls), nicotine caused a significant (p<0.025) linear interval-related improvement. In addition, there was a significant quadratic trend (p<0.05) reflecting a rise in accuracy with increasing delay, followed by a decline back toward control performance with even longer delays. There were significant improvements at the 20-min (p<0.05), 30-min (p<0.05), and 40-min (p<0.01) intervals compared with controls (Fig. 1). No significant nicotine effects were seen at 0-, 5-, 10-, 80-, or 160-min postdosing interval.

Fig. 1
figure 1

The effect of varying the interval between the end of 3-min dosing by immersion in 100 mg/l nicotine ditartrate and the behavioral testing for position discrimination, percent correct (mean±SEM). Control (Con) is vehicle treatment and 0–160 are minutes elapsed between the end of nicotine treatment and the beginning of behavioral testing. The peak effect is 20–40 min after dosing

Full size image

Response latency was also significantly affected by nicotine. The timing of the effect differed from nicotine effect on choice accuracy. Nicotine significantly increased response latency with the 5-min (p<0.001) and the 20-min (p<0.025) intervals. No significant effects were seen with longer intervals (Fig. 2). When the response latency data were broken down by latency for correct vs. incorrect response, no differential treatment was seen on latencies in the two different types of response. Rarely, the fish did not make a choice in the 120-s period permitted. The rate of nonresponse ranged from 0 to 0.5 trials per six-trial choice session between the treatment groups, with no significant treatment effect seen.

Fig. 2
figure 2

The effect of varying the interval between the end of 3-min dosing by immersion to 100 mg/l nicotine ditartrate and the behavioral testing for position discrimination, seconds per trial (mean±SEM). Control (Con) is vehicle treatment, and 0–160 refer to minutes elapsed between the end of nicotine treatment and the beginning of behavioral testing

Full size image

The dose effect of nicotine at the 40-min interval was significant with the linear trend (p<0.005), and the 100-mg/l (p<0.01) and 200-mg/l (p<0.05) doses resulted in significant improvements relative to untreated controls (Fig. 3). No significant effects on latency were seen, either as an overall effect or when broken down by latency for correct and incorrect choices. As in experiment 1, there were rare incidences of nonresponse ranging from 0 to 0.2 trials per six-trial choice session between the treatment groups with no significant treatment effect seen.

Fig. 3
figure 3

The effect of varying the dose at 40 min between 3-min dosings by immersion in 0, 50, 100, and 200 mg/l nicotine ditartrate, percent correct (mean±SEM). This shows the threshold of accuracy improvement between 50 and 100 mg/l

Full size image

The high-affinity nicotinic receptor antagonist mecamylamine effectively reversed (p<0.05) the choice accuracy improvement caused by the benchmark nicotine dose (100 mg/l) that was administered 40 min before testing (Fig. 4). Interestingly, this effect was seen when mecamylamine was administered shortly (5 min) before testing, (p<0.05, late mecamylamine + nicotine vs nicotine alone), but not when mecamylamine was administered concurrently with nicotine 40 min before testing. None of the nicotine or mecamylamine treatments or combinations of treatments significantly affected overall response latency. As in earlier experiments, there was no differential effect of treatment on latency to make correct or incorrect responses. There were also no significant drug treatment effects on the rare incidences of response omissions, which ranged from 0 to 0.4 trials per session in the different treatment groups.

Fig. 4
figure 4

The effect of the high-affinity nicotinic receptor antagonist mecamylamine (200 mg/l) administered concurrently with nicotine (100 mg/l) 40 min before testing (Early) or after nicotine 5 min before testing (Late), percent correct (mean±SEM). This shows reversal of the nicotine-induced accuracy improvement with late, but not early, mecamylamine exposure

Full size image

Discussion

Nicotine exposure in zebrafish caused a significant improvement in the rapid spatial discrimination task. The single-session determination of the promnestic effect of nicotine in zebrafish may be useful for rapid throughput screening of nicotinic and other potential drugs being developed for treating cognitive impairment.

One of the more interesting findings was that the appearance of the promnestic effect of nicotine was delayed. The nicotine peak accuracy improving effect was seen 40 min after the end of nicotine administration. No sign of improvement was seen when the fish were tested just after dosing. Delay intervals of 5 or 10 min also did not result in improvements, although there was a suggestion of an improvement after 10 min of delay. Starting at 20 min, there was a significant nicotine-induced improvement, an effect that was also seen after 30 and 40 min of delay, with the peak effect seen at 40 min. Longer delay intervals of 80 and 160 min diminished choice accuracy improvement, which was not significantly different from untreated controls.

Persistence of nicotine effects that improve cognitive function, such as those indicated in the current zebrafish study, has been seen previously. Buccafusco and Terry (2001) and Levin and Rezvani (2001) have seen this effect in monkeys, and we have seen it in rats. This is also seen with nicotinic agonists. We found that the nicotinic agonist metanicotine caused significant improvements in the memory of rats for 6 h after oral administration, far outlasting the presence of the drug in the body (Levin and Christopher 2002). There may be some persisting positive effects of desensitizing the nicotinic receptor, which could account for the longer-term positive nicotine effects on cognition.

The delayed effect may have also been due to an attenuation of the dose to a more optimal level over time. The increased response latency in the early time intervals after dosing may have been an indication of adverse effects of too high a dose of nicotine. However, an increase in latency per se did not interfere with improved choice accuracy inasmuch as both significantly increased latency and significantly improved choice accuracy as seen at the 20-min interval.

Interestingly, the effect of mecamylamine blocking nicotine-induced choice accuracy improvement was seen when mecamylamine was administered shortly (5 min) before testing, but not when it was administered concurrently with nicotine 40 min before testing. This suggests that activation of high-affinity nicotinic receptors blocked by mecamylamine was necessary for the expression of nicotine effect on choice accuracy, but not on the induction of the effect. Nicotine may have been acting on low-affinity nicotinic receptors to induce its accuracy-improving effects. Alternatively, nicotine-induced desensitization of nicotinic receptors may have been important for the induction of accuracy improvement 20–40 min later. Then as the receptors became resensitized, they may have been better able to be stimulated by their endogenous ligand acetylcholine in the neural bases needed for the cognitive task. This fits well with the observed effects of mecamylamine. Mecamylamine attenuates the stimulation of nicotinic receptors by nicotine (Yin and French 2000), but as with other noncompetitive antagonists (Ochoa et al. 1989; Quick and Lester 2002), mecamylamine may not block the desensitization of nicotinic receptors by nicotine, although the action of mecamylamine in this regard appears to vary with the nicotinic receptor subtype involved (Gentry and Lukas 2002). That is, mecamylamine may have been ineffective in blocking nicotine actions when given concurrently 40 min prior to testing because the critical inciting action of nicotine was to desensitize nicotinic receptors. Then mecamylamine may have effectively blocked the expression of nicotine effects when given shortly before testing because mecamylamine would have blocked the stimulation of the resensitized nicotine receptors by either residual nicotine given 40 min earlier or the endogenous nicotinic ligand acetylcholine.

Nicotinic receptors in zebrafish show a high degree of similarity to those in mammals (Zirger et al. 2003), although the exact degree of similarity remains to be determined. Relevant brain regions for learning and memory in zebrafish certainly differ from those in mammals given the different neural architecture. Functional similarity in pharmacological response will be determined in studies of behavioral responses to drug exposure. The current study is among those studies at the beginning of the series that aim to determine the similarities and differences of zebrafish neurocognitive pharmacology relative to rodents and primates, including humans.

It will be important to determine both pharmacokinetics and differential pharmacokinetics in zebrafish vs mammals. Little is known about the uptake, distribution, and metabolism of nicotine in zebrafish. In the current study, the zebrafish are dosed by immersion, with uptake of nicotine presumably taking place via absorption across the gills, through the skin, or perorally. Nicotinic acetylcholine receptors, the primary sites of action of nicotine, are found in a variety of areas of the rat brain, such as the hippocampus and frontal cortex, which are important for cognitive function (Levin et al. 1997, 2005). Given that the telencephalon of zebrafish is quite limited, it is likely that the midbrain is more important for cognitive function. In mammals, nicotinic receptor density in the optic tectum (superior colliculus) is very high (Clarke et al. 1985). The optic tectum is the major processing area for visual processing in zebrafish (Kaethner and Stuermer 1997) and constitutes a considerable portion of the zebrafish brain, positioned much like the neocortex in mammals (Wullimann et al. 1996). Further research is necessary to determine the possible role of nicotinic receptors in the zebrafish optic tectum for cognitive function.

The rapid single-session assessment of nicotinic drugs with zebrafish could serve as a useful preliminary screening for further development of new promnestic drugs. This zebrafish model could also have a value in the broader context of nonnicotinic drugs as well.