Abstract
Free full text
Endogenous Dopamine Suppresses Initiation of Swimming in Prefeeding Zebrafish Larvae
Abstract
Dopamine is a key neuromodulator of locomotory circuits, yet the role that dopamine plays during development of these circuits is less well understood. Here, we describe a suppressive effect of dopamine on swim circuits in larval zebrafish. Zebrafish larvae exhibit marked changes in swimming behavior between 3 days postfertilization (dpf) and 5dpf. We found that swim episodes were fewer and of longer durations at 3 than at 5dpf. At 3dpf, application of dopamine as well as bupropion, a dopamine reuptake blocker, abolished spontaneous fictive swim episodes. Blocking D2 receptors increased frequency of occurrence of episodes and activation of adenylyl cyclase, a downstream target inhibited by D2-receptor signaling, blocked the inhibitory effect of dopamine. Dopamine had no effect on motor neuron firing properties, input impedance, resting membrane potential, or the amplitude of spike afterhyperpolarization. Application of dopamine either to the isolated spinal cord or locally within the cord does not decrease episode frequency, whereas dopamine application to the brain silences episodes, suggesting a supraspinal locus of dopaminergic action. Treating larvae with 10 μM MPTP reduced catecholaminergic innervation in the brain and increased episode frequency. These data indicate that dopamine inhibits the initiation of fictive swimming episodes at 3dpf. We found that at 5dpf, exogenously applied dopamine inhibits swim episodes, yet the dopamine reuptake blocker or the D2-receptor antagonist have no effect on episode frequency. These results led us to propose that endogenous dopamine release transiently suppresses swim circuits in developing zebrafish.
INTRODUCTION
Locomotion is achieved by the rhythmic activity of motor pattern-generating circuits (Grillner 2003; Kiehn 2006). Descending projections to these pattern-generating circuits regulate their activation through the release of fast-acting neurotransmitters and slower-acting neuromodulators (Barrière et al. 2005; El Manira et al. 1997; Li et al. 2006; Marder and Bucher 2001; McLean and Sillar 2003; Nishimaru et al. 2000; Roberts et al. 1998). As an animal develops, its locomotory behavior becomes more flexible and mature (Clarac et al. 2004; Saint-Amant and Drapeau 1998; Sillar et al. 1991) and, in some cases, even undergoes dramatic changes (Combes et al. 2004). Proper maturation of locomotory behavior requires maturational changes in the neural circuits generating motor commands. Neuromodulators have been implicated in triggering the developmental maturation of pattern-generating circuits (Branchereau et al. 2002; Brustein et al. 2003a; Fenelon et al. 2003; Sillar et al. 1995; Straus et al. 2000) and they may achieve this by affecting neurogenesis (Marsh-Armstrong et al. 2004), synaptogenesis (Niitsu et al. 1995), synaptic strength (McDearmid et al. 1997), intrinsic membrane properties of individual neurons within the network (Han et al. 2007; Sillar et al. 1995), or by changing the influence of other neuromodulators on target networks (McLean and Sillar 2004).
Dopamine is a key neuromodulator involved in the control of motor systems in both invertebrates and vertebrates (Crisp and Mesce 2004; Kiehn and Kjaerulff 1996; Marder and Eisen 1984; Schotland et al. 1995). Loss of brain stem dopaminergic neurons leads to movement disorders both in humans and in nonhuman primates, rodents, and fish (Bretaud et al. 2004; Dauer and Przedborski 2003; Lam et al. 2005; McKinley et al. 2005). Furthermore, dopamine receptor-blocking agents prescribed as antipsychotics induce movement disorders (Dauer and Przedborski 2003). The effect of dopamine on the initiation (Kiehn and Kjaerulff 1996; Madriaga et al. 2004; Whelan et al. 2000) and frequency of motor patterns (Schotland et al. 1995; Svensson et al. 2003b) has been well studied. Given the importance of dopamine in the initiation and control of locomotory behavior in established neural circuits, we tested whether dopamine controls the initiation of swimming in a developing vertebrate, i.e., the larval zebrafish.
Locomotion in larval zebrafish evolves from slow tail flips at 18 h postfertilization (hpf), to escape swimming at 28hpf to robust spontaneous swimming at 5 days postfertilization (dpf) (Brustein et al. 2003b; Buss and Drapeau 2001). As late as 3dpf, larvae show very little spontaneous swimming but by 5dpf, larvae swim actively for foraging. In zebrafish, dopaminergic neurons are seen as early as 24hpf (McLean and Fetcho 2004a). By 3dpf, dopaminergic neurons are seen in the ventral diencephalon, the hypothalamus, the preoptic region, and the pretectum (McLean and Fetcho 2004a; Rink and Wullimann 2002). Also, putative dopaminergic fibers densely innervate the mesencephalon, rhombencephalic reticulospinal neurons, and the spinal cord (McLean and Fetcho 2004a,b).
Here, we show that motor patterns generated by larval zebrafish at 3dpf are vastly different from those at 5dpf. The spinal cord in 3dpf zebrafish larvae is capable of initiating a high frequency of spontaneous fictive swimming episodes, but dopamine, acting via D2 receptors, selectively suppresses the initiation of spontaneous fictive swimming episodes. However, at 5dpf, endogenous release of dopamine does not suppress spontaneous swimming episodes, suggesting differential dopamine modulation of circuits involved in the initiation of spontaneous swimming at these two stages.
METHODS
Adult wildtype zebrafish were obtained from a commercial supplier (Scientific Hatcheries, Huntington Beach, CA) and maintained in aquarium tanks at 28°C. Embryos were collected in a trap every morning and maintained in clean fish water in a water bath at 28°C.
Larval swimming behavior
A single larva was placed in a shallow translucent plastic dish filled with fish water. Larvae swam in a 5-cm-wide circular arena. Swimming behavior was recorded for 15 min using a Hamamatsu ORCA ER CCD camera fitted with a Nikon 50-mm zoom lens at 20 frames/s. The position of the larva in each frame was detected by background subtraction and the displacement was calculated from the previous frame. The total displacement for 15 min was calculated by adding the displacements in each frame.
Extracellular suction recordings
Recordings were performed as described in Masino and Fetcho (2005) with minor modifications (see Fig. 2A). Briefly, larvae were anesthetized in 0.02% Tricaine (MS222) and pinned laterally through their notochord onto Sylgard using fine tungsten wire (California Fine Wire, Grover Beach, CA). We then paralyzed the larvae by replacing the MS222 with Danio external saline containing curare (in mM: 134 NaCl; 2.9 KCl; 1.2 MgCl2; 10 HEPES; 10 glucose; 0.01 d-tubocurarine; 2.1 CaCl2; pH 7.8; 290 mmol/kg). Using fine tungsten wire, we peeled the skin to expose the musculature and the brain. Using thin-walled borosilicate capillaries with no filament (Sutter Instrument, Novato, CA), we pulled large-tipped pipettes and filled them with Danio external saline. We positioned these close to the muscles and aspirated the fibers one by one to expose the spinal cord in two or three segments so that bath-applied drugs would permeate easily into the spinal cord. A micropipette filled with Danio saline (0.7–1.5 MΩ) was positioned very close to the myotomal boundary of one of the anterior segments and mild suction was applied. This resulted in the muscles, as well as the axons innervating them, to be drawn up into the micropipette and the action potentials traveling down these axons could be recorded. We recorded mostly from muscle segments in the rostral one third of the animal except during the local dopamine application experiments (see following text). Multiunit spiking activity was recorded using a Multiclamp 700A amplifier and digitized using a Digidata 1320 and pClamp 9.0 suite of software.
For local application of dopamine, patch pipettes were filled with 10 mM dopamine containing sulforhodamineB (Invitrogen, Carlsbad, CA). The third or fourth segment from the caudal end of the spinal cord was exposed by aspirating muscle fibers as explained earlier and the patch pipette containing dopamine was gently inserted into the cord. Mild positive pressure was applied while monitoring the extent of fluorescence. Pressure was applied until the solution traveled at least six segments rostrally. Recordings were made from segments within the extent of dopamine injection.
Whole cell patch-clamp recording
The larva was pinned out and the spinal cord exposed as described earlier. Patch pipettes were pulled from borosilicate glass (Sutter Instrument) and filled with internal solution (in mM: 115 K-gluconate; 15 KCl; 2 MgCl2; 10 HEPES; 10 EGTA; 3.94 Mg-ATP; pH 7.2; 290 mOsm; pipette resistance 10–14 MΩ). The patch pipette was placed in the bath and, in current-clamp mode, the pipette offset and capacitance were calculated. Then the amplifier was switched to voltage-clamp mode and a gigaseal was formed with a ventrally located cell body. After adjusting for pipette capacitance, the seal was broken to achieve whole cell configuration. The amplifier was switched to current-clamp mode and DC current was injected to keep the cell membrane potential near −65 mV. The bridge resistance and pipette capacitance were compensated for. Current pulses of varying amplitudes and about 1-s duration were injected and the resulting membrane potential was recorded. The cell was filled with fluorescent dye included in the internal solution and, at the end of the recording, motor neuronal identity was confirmed from the morphology of the cell.
Drugs
Saline containing drugs at stated concentrations were bath-applied using a switching manifold and drugs tended to have an effect within 10 to 15 min of bath application. Some drugs were dissolved in 0.1% dimethylsulfoxide (DMSO) prior to further dilution in saline. By itself 0.1% DMSO did not have an effect on motor pattern activity (data not shown). Dosages for all drugs used were determined in preliminary experiments. Dopamine, bupropion hydrochloride, N-methyl-d-aspartate (NMDA), and 4-(4-chlorophenyl)-1-(1H-indol-3-ylmethyl)piperidin-4-ol (L741,626) were obtained from Sigma Chemical (St. Louis, MO) and S(−) sulpiride from Tocris Cookson (Ballwin, MO).
Data analysis
Spikes were extracted off-line using Spike2 software (Cambridge Electronic Design) and spike times were sorted into bursts, episodes, and bouts using custom scripts written in Neuroexplorer (Nex Technologies, Littleton, MA) and Matlab (The MathWorks, Natick, MA). Bouts were defined as intervals during which three or more spikes occurred and after which there was an interspike interval (ISI) of ≥10 s. Episodes were defined as time intervals during which three or more spikes occurred and after which there was an ISI of ≥100 ms. Bursts were defined as time intervals during which one or more spikes occurred and after which there was an ISI of ≥10 ms. Burst period and episode period were calculated as the time between successive burst and episode start times, respectively. Episode duration was calculated as the time between the start and end of an episode. Parameters such as episode period, duration, and burst period were calculated using custom scripts written in Matlab. Swim episodes were counted in a 10-min window and plotted.
For the firing rate versus current-injected plots, the instantaneous firing rate was calculated as the inverse of the first ISI evoked by current of a certain amplitude. The slope of the linear part of the firing rate versus current curve was calculated to obtain gain. The input impedance and resting membrane potential were calculated as the slope and the y-intercept, respectively, of the voltage versus current plot. Data were plotted using Microsoft Excel and SigmaPlot. Statistical testing was performed with Statview. In general the Mann–Whitney U test was used to test for significant differences in episode periods and durations because these data were nonnormally distributed. The Student's t-test was used for burst period data because these passed the test for normality. Where present, error bars indicate SE.
MPTP treatment
Embryos were reared in normal fish water for 24 h at 28°C. 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine hydrochloride (MPTP-HCl) was purchased from Sigma in 10 mg rubber-stoppered bottles. A stock solution of MPTP (50 mM) was made by injecting 1 mL fish water containing 0.01% Tween-80 into the bottle. The stock solution was further diluted to 10 μM, in which 1-day-old embryos were placed. Control embryos were reared in 0.01% Tween-80 solution. At 3dpf embryos were rinsed several times in fish water and the spontaneous swim episodes were recorded as described earlier.
Whole-mount immunohistochemistry
Larvae at 3dpf were fixed in 4% paraformaldehyde at 4°C overnight and then rinsed in phosphate-buffered saline (PBS). Larvae were pinned on Sylgard and the skin covering the brain was carefully peeled. The yolk and the eyes were removed. The jaw was removed to expose the ventral surface of the brain because many of the tyrosine hydroxylase (TH)–positive cell groups are located on the ventral side. All incubations were performed at 4°C. After being blocked overnight in 10% goat serum in PBS-Triton-X100 (PBST), larvae were washed several times in PBST and then incubated in 1:400 rabbit anti-mouse tyrosine hydroxylase antibody (MAB318, Chemicon, Temecula, CA) for 2 days. The larvae were washed several times in PBST and then incubated in chicken anti-rabbit IgG coupled to Alexa 546 overnight (Invitrogen). Larvae were washed several times in PBS and the brains were mounted ventral side up on glass slides with Prolong mounting medium (Invitrogen) and coverslipped. Brains were imaged on a Zeiss LSM 510 confocal microscope with 543-nm excitation and a 560-nm long-pass filter. Images were analyzed off-line using LSM Image Browser and Adobe Photoshop software. Control and MPTP-treated larvae were processed for immunohistochemistry in parallel and imaged under the same conditions. Images were analyzed off-line for number of TH-positive cell bodies using ImageJ software.
RESULTS
Newly hatched zebrafish larvae (~2.5dpf) have intact touch-evoked swimming responses but show very little spontaneous swimming. However, by 5dpf, larvae swim robustly in all three dimensions (Brustein et al. 2003b; Buss and Drapeau 2001). To quantify this change in swimming behavior, we recorded movies of larval swimming at 3 and 5dpf and calculated the total displacement of larvae over 15 min (Fig. 1; Supplemental Movies S1 and S2).1 The average total displacement of larvae in 15 min is significantly greater at 5dpf than that at 3dpf (Fig. 1B, P < 0.01, n = 7).
To understand the neural basis for such a dramatic change in swimming behavior, we recorded fictive motor patterns using suction electrodes at the nerve muscle junction in paralyzed preparations (Fig. 2A). In the 3dpf larvae, such recordings revealed motor patterns organized into spontaneous episodes of fictive swimming (subsequently referred to as “episodes”). These episodes were clustered together into bouts of activity and bouts of episodes lasted several seconds with periods on the order of several tens of seconds (Fig. 2B, top trace). Each bout consisted of many episodes of swimming, with the episode periods ranging from hundreds of milliseconds to several seconds (Fig. 2B, second trace). Within each episode of swimming, bursts of spikes occurred with intervals ranging from 20 to 50 ms. Bursts consisted of one to two spikes and the spiking interval within a burst was <10 ms (Fig. 2B, bottom two traces). Figure 2C shows the frequency distribution of the ISIs for the data in Fig. 2B and represents typical ISI distributions seen at 3dpf. ISIs separating spikes in successive bursts (interburst intervals) are centered at about 30 ms and there is a broad distribution of longer ISIs with no detectable peaks. The peak seen at ISI < 10 ms corresponds to spike intervals within bursts (intraburst intervals). These data demonstrate that spontaneous motor patterns are organized into rhythms at multiple timescales.
Fictive swim motor patterns show significant differences between 3 and 5dpf
We then compared fictive swim motor patterns recorded at 3dpf with those recorded at 5dpf. At 3dpf, episodes occurred at relatively low frequency (Fig. 3A, top trace). When these episodes occurred, they were clustered into bouts (Fig. 3A, bottom trace and Fig. 2B). However, at 5dpf, episodes occurred at a much higher frequency (Fig. 3B, top trace) and a bout structure was absent (Fig. 3B, bottom trace). The number of episodes in a 10-min recording window increased from 22.3 ± 4.9 at 3dpf (n = 14) to 222 ± 23 at 5dpf (n = 11; P < 0.001, Fig. 3C). The cumulative probability distribution of episode periods at 3 and at 5dpf (Fig. 3D) shows that at 5dpf, there was no episode period 50 s. The distributions of episode periods at 3 and 5dpf were similar for short episode periods but diverged at the long end of the distribution (P < 0.001, n = 14 at 3dpf and n = 11 at 5dpf). At 3dpf, short episode periods were those occurring inside of a bout of episodes and these were similar to those seen at 5dpf (compare periods for bottom trace in Fig. 3, A and B). However, at 3dpf, bouts were separated by long intervals and these were responsible for the long tail of the period distribution. Episode duration also changed significantly between 3 and 5dpf. At 3dpf, episodes had an average duration of 1.78 ± 0.32 s (n = 14), whereas at 5dpf episode duration was 0.45 ± 0.09 s (n = 11). The distribution of episode durations at 5dpf was shifted toward shorter durations compared with those at 3dpf (Fig. 3E) and the two distributions were significantly different from each other (P < 0.0001). Concomitant with the developmental decrease in episode durations, there was also a decrease in the number of bursts per episode from 42.3 ± 7.5 at 3dpf to 14.2 ± 2 bursts per episode at 5dpf (Fig. 3I, n = 26, P < 0.001). When the motor pattern was observed at faster timescales, we noticed an increase in the number of spikes per burst from 3 to 5dpf (Fig. 3, F and G; P < 0.001, n = 26). However, there was no significant change in the average burst period between 3 and 5dpf (3dpf: 35.1 ± 1 ms; 5dpf: 34.7 ± 1 ms, n = 26, P = 0.934). Thus maturation of the spinal cord swim circuit from 3 to 5dpf was marked by a decrease in the period and duration of episodes, a decrease in the number of bursts per episode, and an increase in the number of spikes per burst.
Dopamine is a modulator of swim-initiating circuits in zebrafish larvae
The output of the swim circuit differed between 3 and 5dpf in zebrafish larvae, suggesting that the swim circuit undergoes developmental modifications during this time. Because dopaminergic innervation of the CNS develops relatively early, it is an ideal candidate for regulating the maturation of the swim motor circuit. Therefore we first tested whether dopamine altered swim-circuit output in larval zebrafish at 3dpf. At 3dpf, bath application of saline containing 10 μM dopamine abolished all episodes in four of four larvae (Fig. 4A); however, episodes could still be evoked by a flash of light in the presence of dopamine in four of four larvae (Fig. 4A, middle trace, arrow). The number of bursts per episode (49.8 ± 24.7), the number of spikes per burst (1.46 ± 0.3), and the burst cycle periods (0.041 ± 0.007 s) in such light-evoked episodes were not significantly different from those seen during spontaneous episodes in control saline (bursts/episode: P = 0.81; spikes/burst: P = 0.346; burst cycle period: P = 0.584). Episodes returned after dopamine was rinsed out of the bath (Fig. 4A).
To test whether dopamine is released endogenously and whether the endogenously released dopamine can suppress fictive swim episodes, we bath-applied 200 μM bupropion hydrochloride, a dopamine reuptake blocker, to 3dpf larvae. Bath application of 200 μM bupropion hydrochloride reversibly silenced episodes in four of four larvae at 3dpf (Fig. 4B). These data indicated that both exogenously applied dopamine and endogenously released dopamine could silence episodes in the 3dpf larval locomotor network.
In contrast to the effect of dopamine and bupropion, application of L741,626, a D2-receptor–specific antagonist, increased the frequency of swim episodes compared with that in control saline (Fig. 4C). In five of five larvae, the number of episodes in 10 min increased after application of L741,626 (Fig. 4D, P < 0.02). When we plotted the cumulative probability distribution of episode periods in control saline and in L741,626 (Fig. 4E), we found that the distributions were significantly different (P < 0.05, n = 175 episodes from five larvae). The distributions were overlapping for episode periods <1 s, although they diverged for longer episode periods. The distribution of episode durations was not significantly different between control and L741,626 (Fig. 4F, P = 0.65, n = 180 episode durations from five larvae). Similarly, application of sulpiride, another D2-specific antagonist, also significantly increased the number of episodes (Fig. 4, G and H, control: 39 ± 23; sulpiride: 194 ± 55, P < 0.01). However, there was no change in the number of spikes per burst, the burst cycle period, or the number of bursts per episode.
Activation of the D2-like family of receptors (consisting of D2, D3, and D4 receptors) suppresses adenylyl cyclase activity (Missale et al. 1998) and leads to a reduction in cellular cyclic adenosine monophosphate levels (Fig. 5A). We asked whether the effect of dopamine on fictive swimming can be occluded by interfering with the downstream signaling of D2 receptors. To this end, we first silenced the spontaneous activity with dopamine and then bath-applied forskolin, a cell-permeable activator of adenylyl cyclase (Fig. 5A). In preparations that were silenced with 100 μM dopamine (Fig. 5B, middle trace), adding 10 μM forskolin to the saline resulted in a high frequency of episodes (Fig. 5B, bottom trace), indicating that forskolin was able to override the suppressive effects of dopamine even at relatively high concentrations of dopamine. In five of five larvae, application of dopamine along with forskolin increased the number of episodes in 10 min (Fig. 5C, control: 39 ± 13; forskolin + dopamine: 397 ± 94; P < 0.02). Again the cumulative distributions of episode periods in control and in forskolin plus dopamine were significantly different (Fig. 5D, 522 episode periods from five larvae, P < 0.001). Forskolin, even in the presence of dopamine, caused a shift of episode periods toward shorter period values and an absence of episode period values >50 s. Importantly, activation of adenylyl cyclase blocked the suppressive effects of dopamine on episodes. Application of forskolin along with dopamine also significantly decreased episode durations: the cumulative distribution of episode durations shifted to the left (Fig. 5E, 539 episode durations from five larvae, P < 0.001), and modestly increased the average burst period (Fig. 5, F and G, control: 0.033 ± 0.003 s, forskolin + dopamine: 0.04 ± 0.0018 s, P < 0.05). Forskolin did not affect the number of spikes per burst or the number of bursts per episode.
Effect of dopamine on motor neuronal firing properties
By 3dpf, the spinal cord is innervated by TH-immunoreactive fibers (McLean and Fetcho 2004a), some of which originate from the dopaminergic neurons of the posterior tuberculum (McLean and Fetcho 2004b). These spinal cord TH fibers are closely apposed to motor neuronal cell bodies and proximal dendrites (McLean and Fetcho 2004b), suggesting a direct effect of dopamine on motor neurons. To investigate whether dopamine affects motor neuronal firing properties, we performed whole cell recordings in current-clamp mode from motor neurons in control saline and in 10 μM dopamine. In control saline, motor neurons showed episodes of spiking (Fig. 6A, left), which correspond to the episodes recorded extracellularly. The same motor neurons showed no spontaneous spiking activity in the presence of 10 μM dopamine, consistent with the results obtained with extracellular recording (Fig. 6A, right). To assay whether dopamine modifies the neuronal input–output relationship, we recorded spikes generated by increasing amounts of injected current in control saline or in 10 μM dopamine (Fig. 6B). The instantaneous firing rate, calculated as the inverse of the first ISI evoked by current of given amplitude, was not affected by dopamine (Fig. 6, B and C). Dopamine did not significantly alter the neuronal gain (the slope of the firing rate vs. current plot; Fig. 6D, P = 0.838), input impedance (Fig. 6E, P = 0.713), resting membrane potential (Fig. 6F, P = 0.969), or afterhyperpolarization amplitude (Fig. 6G, P = 0.232, n = 35 cells in control and 24 cells in dopamine).
Effect of dopamine on spinal circuits
We investigated the effect of dopamine on local spinal cord circuits by using two methods: 1) by assaying the effect of dopamine on the isolated spinal cord and 2) by applying dopamine locally within the cord in the intact animal. We spinalized 3dpf zebrafish by severing the tail at the level of the rostral fourth and fifth segments. After isolation of the tail, no spontaneous activity could be detected (data not shown). NMDA application induced episodic activity as previously reported (McDearmid and Drapeau 2006) (Fig. 7A, top trace). Addition of 100 μM dopamine to the saline containing NMDA did not suppress episodes (Fig. 7A, bottom trace). There was no reduction in the number of NMDA-evoked episodes in 10 min when dopamine was added to the saline (Fig. 7E, P = 0.2). Dopamine had no effect on the number of spikes per burst or bursts per episode but it increased the average burst period (NMDA: 0.047 ± 0.002 s; NMDA + DA: 0.057 ± 0.003 s, P = 0.032, n = 4).
Next, we injected dopamine either into the caudal spinal cord or into the rhombencephalic ventricle while extracellularly recording episodes. When 10 mM dopamine (1,000-fold bath-applied levels) was pressure-injected into the caudal cord, such that four to six segments were exposed to dopamine, episodes were not affected (Fig. 7C, top and middle traces) and the number of episodes in 10 min remained unchanged (Fig. 7D, P = 1, n = 6, paired t-test). The distribution of episode periods was also not significantly altered (316 episode periods from six larvae, P = 0.254). By contrast, in the same animal, when 10 mM dopamine was injected into the rhombencephalic ventricle, episodes were abolished (Fig. 7C, bottom trace) and the number of episodes fell to zero (Fig. 7D). This suggests that the target circuits of dopamine that mediate the suppression of episodes are more likely to lie within the brain and not in the spinal cord.
Effect of dopamine on swimming in 5dpf larvae
We investigated the effect of dopamine in 5dpf larvae, first on swimming behavior and then on fictive swimming in paralyzed preparations. We collected movies of larval swimming behavior in normal fish water and in fish water containing 10 μM dopamine for 15 min each. Larvae at 5dpf showed robust spontaneous swimming (Supplemental Movie S2 and Fig. 8 A, left), whereas the same larvae placed in saline containing 10 μM dopamine showed reduced swimming (Supplemental Movie S3 and Fig. 8A, right). The average total displacement of larvae in 15 min in dopamine was significantly smaller than that in normal saline (Fig. 8B; control: 264.5 ± 46.9 cm; DA: 83.5 ± 24.4 cm, P < 0.05, n = 7).
When we applied dopamine to paralyzed 5dpf larvae, we found a significant decrease in the number of episodes (Fig. 8, C and D; n = 7, P < 0.001). In the presence of dopamine, the cumulative distribution of episode periods was significantly different from that seen in control and had a long tail like that seen at 3dpf (Fig. 8E; 529 episode periods from seven larvae, P < 0.001). The two distributions were similar for short episode periods (<10 s) but diverged for longer episode periods. In particular, at 5dpf, in control saline, period values >50 s were absent, although periods as long as 500 s were seen in the presence of dopamine. Similarly, dopamine shifted the distribution of episode durations toward longer durations (Fig. 8F; 536 episode durations from seven larvae, P < 0.001). There was no effect on the number of spikes per burst or the average burst period in the presence of dopamine.
Endogenously released dopamine does not suppress spontaneous motor episodes at 5dpf
We recorded fictive swim episodes from 5dpf larvae in the presence of bupropion—the dopamine reuptake blocker—and observed that bupropion did not affect episode number (Fig. 9, A and B, P = 0.438, n = 6). Distributions of episode periods in control and in bupropion were similar (Fig. 9C; 1,209 episode periods from six larvae at 5dpf, P = 0.385); however, episodes were of significantly shorter duration in bupropion than in control saline (Fig. 9D; 1,215 episode durations from six larvae at 5dpf, P < 0.001). The number of spikes per burst or the average burst periods were not affected. Thus in contrast to the suppressive effect of bupropion on spontaneous motor episodes at 3dpf, the accumulation of extracellular endogenous dopamine with bupropion treatment was not sufficient to decrease spontaneous activity at 5dpf.
Blocking D2 receptors does not affect episode activity at 5dpf
We asked whether blocking D2 receptors at 5dpf would yield the same effects as it did in the 3dpf larva. We found that bath application of sulpiride—the D2-receptor antagonist—did not affect the number of swim episodes (Fig. 10, A and B, n = 6, P = 0.86). Sulpiride did not affect the distributions of episode periods and durations (Fig. 10, C and D; P = 0.766 for episode periods and P = 0.705 for durations) nor did it change the number of spikes per burst, the burst cycle period, or the number of bursts per episode.
Larvae with compromised dopaminergic circuitry have a higher frequency of occurrence of swim episodes
All of our data thus far suggest that in the 3dpf zebrafish larva, the motor circuitry for generating episodes is functional but suppressed by dopaminergic input and that by 5dpf dopaminergic signaling may not be as effective at suppressing the motor output. If this is the case, specifically lesioning the dopaminergic neurons should relieve this suppression and lead to a higher frequency of fictive swim activity at 3dpf. To test this hypothesis, we used 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine hydrochloride (MPTP), a neurotoxin that has been shown to kill dopaminergic neurons in zebrafish larvae at 50 μM (Bretaud et al. 2004). When we treated larvae with 50 μM MPTP, larvae exhibited a curved body and were completely still. They did not show any escape responses and the musculature appeared rough and irregular. At a lower dose (10 μM), larvae had a normal appearance. To test whether 10 μM MPTP affects dopaminergic neurons, we maintained 1dpf larvae in normal fish water or 10 μM MPTP for 48 h and immunostained them for tyrosine hydroxylase (TH), a marker for catecholaminergic neurons. Dopaminergic cells are found in the olfactory bulb, the subpallium, the pretectum, the preoptic region, the ventral diencephalon, and in the caudal hindbrain. The largest group of TH-positive cells is found in the ventral diencephalon (Fig. 11A, i and ii) (McLean and Fetcho 2004a; Rink and Wullimann 2002). Although MPTP treatment did not appear to affect the number of TH-positive cell bodies in any of the above-cited regions (Fig. 11B), it severely reduced the TH-positive fiber systems in the brain (Fig. 11A, iii–viii). Specifically, TH immunoreactivity (IR) in the postoptic commissure was reduced (Fig. 11A, iii and iv) and TH-IR in the neuropil of the optic tectum was not detectable in the MPTP-treated larvae (Fig. 11A, v and vi). TH-IR fibers in the rhombencephalon that have been shown to innervate reticulospinal circuitry (McLean and Fetcho 2004b) were greatly reduced (Fig. 11A, vii and viii). Overall, 10 μM MPTP treatment reduced innervation of the brain by TH-positive axonal fibers, suggesting that the toxin treatment reduced dopaminergic inputs to target circuits.
When we collected extracellular recordings of spontaneous episodes from control and MPTP-treated larvae, we found that the episodes occurred much more frequently in the latter group (Fig. 11C). Larvae treated with 10 μM MPTP from 1 to 3dpf had significantly more episodes in 10 min at 3dpf compared with control larvae (Fig. 11, C and D, P < 0.001, n = 31 for control, n = 20 for MPTP). The episodes appeared normal and the burst periods in MPTP-treated larvae were similar to those in control larvae (Fig. 11C, bottom trace; compare with Fig. 2B, P = 0.942). The number of spikes per burst and the number of bursts per episode were not significantly different in control and MPTP-treated larvae.
DISCUSSION
Zebrafish larvae show marked changes in swim output between 3 and 5dpf. Episodes of swimming were much more frequent and shorter in duration at 5dpf compared with those at 3dpf. At 3dpf, when we exogenously applied dopamine, or when we increased endogenously released dopamine with a dopamine uptake inhibitor, swim episodes were abolished. Conversely, when we blocked dopaminergic inputs with a D2-receptor antagonist or when we compromised the dopaminergic circuit with the neurotoxin MPTP, larvae exhibited a higher frequency of swim episodes. These data indicate that zebrafish larvae at 3dpf are capable of generating fictive swimming episodes at a high frequency but that the initiation of episodes is suppressed by endogenously released dopamine. Subsequently, at later larval stages, endogenously released dopamine is unable to suppress initiation of swimming and the larva exhibits more frequent episodes of swimming. We propose that dopaminergic neuromodulation helps to keep the functional but immature swimming circuit of a 3dpf larva largely silent.
Transient nature of dopaminergic suppression of swim circuits
We found that exogenous dopamine suppressed swim episodes at 3 and 5dpf. To test whether endogenously released dopamine affected fictive swim episodes at both stages, we used bupropion, which blocks dopamine reuptake. We found that although bupropion silenced episodes at 3dpf, it had no effect on episode frequency at 5dpf, even though dopaminergic neurons and projections to the spinal cord are still present at 5dpf (McLean and Fetcho 2004a; Rink and Wullimann 2002). This suggests that the endogenous release of dopamine is unable to exert a suppressive effect on swim circuits at 5dpf. One caveat is that bupropion also blocks noradrenaline reuptake; thus some of the effects of bupropion that we see might be mediated by noradrenaline instead of dopamine. However, bupropion has been shown to be twice as effective at the dopamine transporter as it is at the noradrenaline transporter (Horst and Preskorn 1998). Further, bupropion has been used effectively in earlier studies in lamprey to establish a role for endogenous dopamine in the modulation of locomotory patterns (Schotland et al. 1995; Svensson et al. 2003b), spinal neuron intrinsic properties (Schotland et al. 1995), and the strength of reticulospinal synaptic inputs (Svensson et al. 2003a). It was also shown by high performance liquid chromatography analysis that when spinal tissue was incubated with bupropion, the concentration of extracellular dopamine significantly increased (Schotland et al. 1995). Last, consistent with the lack of effect of bupropion, we found that application of D2-receptor antagonists did not affect episode frequency at 5dpf. Taken together, these results led us to propose that endogenously released dopamine has a transient suppressive effect on swim circuits in zebrafish larvae. The inability of endogenously released dopamine to suppress the output of the swim circuit at 5dpf may be driven by other ontogenic events such as the maturation of other modulatory inputs (Brustein et al. 2003a) or the increase in excitatory synaptic drive within the cord (Buss and Drapeau 2001).
Locus of dopaminergic action
We found that dopamine had no effect on motor neuronal intrinsic properties and that it did not suppress NMDA-evoked episodes in the isolated cord. In the intact animal, when we injected dopamine locally within the cord so as to expose four to six segments to dopamine, swim episodes were not affected. The most parsimonious explanation for these results is that the locus of dopaminergic suppression of swim episodes is supraspinal, although it is possible that the local injection of dopamine into the spinal cord did not reach a sufficient number of spinal segments to inhibit episode initiation. There are about 30 spinal segments in zebrafish and suppression of 20% of the segmental oscillators may be insufficient to prevent the triggering of episodes. Nevertheless, it is not technically feasible to expose all of the spinal cord without also exposing supraspinal centers to dopamine in this small animal. It should be possible in the future to confirm the locus of dopamine's action through the use of calcium imaging of descending projection neurons.
Dopaminergic terminal loss in response to MPTP treatment
We used a dose of MPTP lower than that used in previous studies in zebrafish (Bretaud et al. 2004; Lam et al. 2005; McKinley et al. 2005). This low dose of MPTP reduced the number of TH-labeled fibers in many areas of the brain. This observation is particularly interesting because in Parkinson's disease, dopaminergic neurons seem to undergo a “dying back” process by which the dopaminergic axonal process progressively dies, culminating in the death of the cell body itself. This view is substantiated by MPTP toxicity studies in the monkey, where loss of striatal dopaminergic terminals precedes loss of substantia nigra cells, and in the rat, where protection of striatal terminals prevents the loss of substantia nigra cells (Dauer and Przedborski 2003). These studies suggest that many of the early clinical symptoms of Parkinson's disease could be the effect of loss in terminals and reduction in dopamine release to postsynaptic targets even before cell death has begun. Consistent with this, cDNA microarray analysis of rat substantia nigra has revealed that MPTP treatment results in reduced expression of messages coding for proteins involved in axonal transport, vesicle docking, and transmitter release (Miller et al. 2004). Taken together with previous studies, our results suggest that a low dose of MPTP is an effective tool to compromise dopaminergic cell function without inducing cell death.
Developmental modulation of locomotory circuits
During development, animals face changing behavioral needs. For example, with increasing body lengths, tadpoles and fish larvae encounter increasing Reynolds numbers. Therefore the force required for propulsion needs to increase. Another example is a change in mode of locomotion, such as in amphibians, where the locomotory networks switch from a premetamorphic tail-driven axial swimming output to a postmetamorphic limb-driven hopping/swimming output (Combes et al. 2004). The maturation of locomotory circuits in line with the animal's behavioral needs can be triggered by neuromodulators. For instance, in zebrafish, serotonin increases swim episode frequency (Brustein et al. 2003a) and it has been suggested that serotonin produces its effects by modulating chloride homeostasis (Brustein and Drapeau 2005). Although serotonergic neurons and fibers are present quite early in development, serotonin fails to affect episode frequency earlier than 4dpf (Brustein et al. 2003a). In fact, maturation of the sertonergic system by 5dpf may partially underlie the relief from dopaminergic suppression of swim episodes that we see.
Neuromodulators can suppress the expression of a motor output until it is needed. The network for generating lung breathing is mature in premetamorphic Rana tadpoles but it is inhibited by γ-aminobutyric acid type B receptor activation (Straus et al. 2000). Similarly, the stomatogastric ganglion in embryonic lobsters is capable of generating adult-like motor patterns but is suppressed from doing so by central modulatory inputs (Fenelon et al. 2003).
Motor patterns in newly hatched Xenopus tadpoles consist of single-spike bursts but in 24 h, this nascent motor pattern evolves to one in which there are several impulses per burst cycle (Sillar et al. 1991). Serotonin acts as a maturation factor for this process (Sillar et al. 1995) and it was shown that the presence of exogenous serotonin increases burst durations in the rostral cord around the time that raphe inputs arrive at the cord. In subsequent stages, more caudal segments of the cord were affected by serotonin (Sillar et al. 1992). Likewise, removal of serotonin or blockade of 5HT1a receptors prevented the occurrence of multispike bursts in posthatchling tadpoles (Sillar et al. 1995). Our results in 3 and 5dpf larvae also show that in larval zebrafish the motor pattern evolves from single-spike bursts to multispike bursts, similar to the ontogeny of the Xenopus motor pattern.
Dopamine is known to activate motor patterns in the rat (Kiehn and Kjaerulff 1996). In mouse, dopamine increases the excitability of motor neurons and the excitatory synaptic transmission impinging on them (Han et al. 2007). In lamprey, dopamine decreased burst frequency when present at high concentrations and had the opposite effect at lower concentrations (Svensson et al. 2003b). Dopamine also reduced the amplitude of the slow afterhyperpolarization in spinal neurons in lamprey (Schotland et al. 1995). Here, we propose a developmental role for dopamine in regulating swim-circuit activity in larval zebrafish. We suggest that endogenously released dopamine acts transiently to regulate swim-episode frequency in larval zebrafish. Such differential neuromodulation during development might be fundamental for the maturation of network function, not only in locomotory circuits but also for neural circuits in general.
GRANTS
This work was supported by the National Institutes of Health Director's Pioneer Award to H. T. Cline.
Acknowledgments
We thank Dr. Mark Masino and Dr. Joseph Fetcho for help with the extracellular recording technique, Dr. James Demas for assistance with the behavior, and members of the Cline lab for helpful discussions.
Notes
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society
Full text links
Read article at publisher's site: https://doi.org/10.1152/jn.90568.2008
Read article for free, from open access legal sources, via Unpaywall: https://europepmc.org/articles/pmc2544474
Free after 12 months at intl-jn.physiology.org
http://intl-jn.physiology.org/cgi/content/full/100/3/1635
Free to read at intl-jn.physiology.org
http://intl-jn.physiology.org/cgi/content/abstract/100/3/1635
Citations & impact
Impact metrics
Citations of article over time
Article citations
Plasticity of Dopaminergic Phenotype and Locomotion in Larval Zebrafish Induced by Brain Excitability Changes during the Embryonic Period.
eNeuro, 10(6):ENEURO.0320-21.2023, 27 Jun 2023
Cited by: 1 article | PMID: 37316295 | PMCID: PMC10306124
mef2ca and mef2cb Double Mutant Zebrafish Show Altered Craniofacial Phenotype and Motor Behaviour.
Biomolecules, 13(5):805, 09 May 2023
Cited by: 0 articles | PMID: 37238675 | PMCID: PMC10216501
A New Bioassay for the Detection of Paralytic and Amnesic Biotoxins Based on Motor Behavior Impairments of Zebrafish Larvae.
Int J Mol Sci, 24(8):7466, 18 Apr 2023
Cited by: 0 articles | PMID: 37108629 | PMCID: PMC10144378
Comprehensive analysis of locomotion dynamics in the protochordate Ciona intestinalis reveals how neuromodulators flexibly shape its behavioral repertoire.
PLoS Biol, 20(8):e3001744, 04 Aug 2022
Cited by: 5 articles | PMID: 35925898 | PMCID: PMC9352054
Zebrafish Embryos and Larvae as Alternative Animal Models for Toxicity Testing.
Int J Mol Sci, 22(24):13417, 14 Dec 2021
Cited by: 19 articles | PMID: 34948215 | PMCID: PMC8707050
Review Free full text in Europe PMC
Go to all (60) article citations
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Mechanisms underlying the endogenous dopaminergic inhibition of spinal locomotor circuit function in Xenopus tadpoles.
Sci Rep, 6:35749, 20 Oct 2016
Cited by: 6 articles | PMID: 27760989 | PMCID: PMC5071771
Dopaminergic modulation of spinal neurons and synaptic potentials in the lamprey spinal cord.
J Neurophysiol, 77(1):289-298, 01 Jan 1997
Cited by: 23 articles | PMID: 9120571
Dopamine D2 receptor activity modulates Akt signaling and alters GABAergic neuron development and motor behavior in zebrafish larvae.
J Neurosci, 31(14):5512-5525, 01 Apr 2011
Cited by: 46 articles | PMID: 21471388 | PMCID: PMC6622698
Changes in electrophysiological properties of lamprey spinal motoneurons during fictive swimming.
J Neurophysiol, 88(5):2463-2476, 01 Nov 2002
Cited by: 13 articles | PMID: 12424286
Prey capture in zebrafish larvae serves as a model to study cognitive functions.
Front Neural Circuits, 7:110, 11 Jun 2013
Cited by: 26 articles | PMID: 23781176 | PMCID: PMC3678101
Review Free full text in Europe PMC