Modulation of the histaminergic system and behaviour by α-fluoromethylhistidine in zebrafish

Authors


Address correspondence and reprint requests to Pertti Panula, Institute of Biomedicine/Anatomy, Biomedicum, PB 63, FIN-00014 University of Helsinki, Finland. E-mail: pertti.panula@helsinki.fi

Abstract

The functional role of histamine (HA) in zebrafish brains was studied. Zebrafish did not display a clear circadian variation in brain HA levels. Loading of zebrafish with l-histidine increased HA concentration in the brain. A single injection of the histidine decarboxylase (HDC) inhibitor, α-fluoromethylhistidine (α-FMH), gave rise to a rapid reduction in zebrafish brain HA. Low HDC activity in the brain after injections verified the effect of α-FMH. A reduction in the number of histaminergic fibres but not neurones and an increased expression of HDC mRNA was evident after α-FMH. Automated behavioural analysis after α-FMH injection showed no change in swimming activity, but abnormalities were detected in exploratory behaviour examined in a circular tank. No significant behavioural changes were detected after histidine loading. The time spent for performance in the T-maze was significantly increased in the first trial 4 days after α-FMH injections, suggesting that lack of HA may impair long-term memory. The rostrodorsal telencephalon, considered to correspond to the mammalian amygdala and hippocampus in zebrafish, is densely innervated by histaminergic fibres. These results suggest that low HA decreases anxiety and/or affects learning and memory in zebrafish, possibly through mechanisms that involve the dorsal forebrain.

Abbreviations used
α-FMH

α-fluoromethylhistidine

HA

histamine

HDC

histidine decarboxylase

TH

tyrosine hydroxylase

TM

tuberomammilary nucleus

Mammalian histaminergic neurones are located in the tuberomamillary nucleus (TM) and they innervate the whole brain (Panula et al. 1984; Watanabe et al. 1984; Steinbusch et al. 1986). Histamine (HA) is involved in several regulatory mechanisms in the brain, including alertness and sleep, seizure threshold, hormone secretion and pain (Schwartz et al. 1991; Hough 1999; Brown et al. 2001; Fernandez-Novoa and Cacabelos 2001; Haas and Panula 2003), but its exact functions are poorly established.

HA is produced by decarboxylation of l-histidine. Systemic injections of histidine increase significantly brain HA concentration (Prell et al. 1996b). The only enzyme known to catalyse this reaction is histidine decarboxylase (HDC, EC 4.1.1.22). α-Fluoromethylhistidine (α-FMH) is a potent, irreversible inhibitor of HDC and a useful tool to study the function of the histaminergic system (Kollonitsch et al. 1978).

The available data regarding the effect of modulated central HA content on rodent behaviour, especially on learning and reinforcement, is quite contradictory. After depletion of HA with α-FMH both facilitatory (Tasaka et al. 1985; Kamei et al. 1993) and inhibitory (Cacabelos and Alvarez 1991) effects on conditioned avoidance response have been reported. α-FMH induces an increase in total errors and results in spatial memory impairment in rats (Chen et al. 1999).

An increase in anxiolytic behaviour after lesions, linked to a reduction in HA production, in the rostroventral TM has been reported (Frisch et al. 1998). Studies on mouse suggest that anxiety is provoked through stimulation of H1 receptor, e.g. H1 receptor gene knock-out mice are calmer than wild-type mice (Yanai et al. 1998), whereas H2 receptor stimulation results in an inhibition of experimentally induced anxiety (Yuzurihara et al. 2000). Studies of H3 receptor suggest that this receptor is devoid of an effect on anxiety (Perez-Garcia et al. 1999). HA has been implicated in locomotion, as mice deficient in the H1 receptor or H3 receptor gene show a decrease in locomotion (Yanai et al. 1998; Toyota et al. 2002). Low central HA increases the ethanol-induced motor impairment, possibly through H3 receptors (Lintunen et al. 2002).

The vertebrate histaminergic system is well preserved through the evolution. The zebrafish is an exceptionally suitable model for studies of functions of the neurotransmitter HA in the CNS, because non-neuronal HA is undetectable (Eriksson et al. 1998). Cells expressing HDC in zebrafish are limited to the caudal hypothalamus (Eriksson et al. 1998). A HA receptor activating the Gi/Go class of G proteins binding specifically N-α-[3H]methylhistamine has been described in zebrafish, suggesting existence of an H3-like receptor (Peitsaro et al. 2000). Histaminergic neurones establish extensive connections with other aminergic systems and modulate them in the zebrafish brain. The connections between histaminergic and other aminergic systems are extensive especially in the telencephalon, hypothalamus and raphe area (Kaslin and Panula 2001).

This study addresses the role of endogenous brain HA in zebrafish. To assess the general significance of any significant behavioural effects, it is important to know how brain HA synthesis is regulated in zebrafish compared to mammals.

Materials and methods

Experimental animals

Zebrafish (Danio rerio) larvae and adults, from outbred and AB strains of both sexes were used for the study. The fish were kept at 28.5°C with a light/dark cycle of 14 h/10 h. Breeding and raising was done according to Westerfield (1995). The fish were fed twice daily; 08.30 and 16.00 with regular fish dry food and Artemia nauplia. The permit to carry out these studies was obtained from the Committee for Animal Experiments of Abo Akademi University and the Office of the Regional Government of Western Finland.

Drug administration

Before injection, fish were kept in water containing benzocaine (0.1 g benzocaine/200 mL water) until calm. Fish were injected using Hamilton syringes (Hamilton Bonaduz AG, Bonaduz, Switzerland). Histidine (500 mg/kg,l-histidine hydrochloride monohydrate, Sigma, St. Louis, MO, USA) was injected intramuscularly and fish were kept 1, 2, 3, 4, 5 or 6 h after injection. No food was given during the experiment. α-FMH (100 mg/kg, a generous gift from Dr J. Kollonitch, Merck) and saline (0.9% NaCl solution) injections were performed once, three or five times intramuscularly. Fish injected more than once were injected with a 24-h interval. The time points used in this study were 6 h, 10 h, 24 h or 96 h after the last injection. Fish were fed normally during the experiment. The injected volume was between 4 and 6 µL, depending on the weight of the fish (0.5–0.7 g). Immediately after injection, fish were transferred to fresh highly aerated water to recover. Fish were killed by exposure to ice-cold water.

Embryos were exposed to α-FMH through water. They were collected directly after spawning, cleaned and transferred to small Petri dishes containing aquarium water and α-FMH (10–100 µm). The water/solution was changed once daily. Control and drug-treated groups were raised in parallel.

Quantification of histamine concentration

Freshly dissected zebrafish brains were weighed and homogenized in 10 volumes of 2% perchloric acid and centrifuged at 10 000 × g for 15 min. The HA content in the samples was determined by an automated HPLC-fluorometric method of Yamatodani et al. (1985) as described earlier (Eriksson et al. 1998). HA was measured from brains at different time-points of the day to evaluate the existence of circadian HA in zebrafish. The HA concentration in the brains of histidine-loaded fish was expressed as a percentage after setting the HA concentration in saline-treated brains as 100% ± SEM. Results for α-FMH-exposed fish are presented as percentage HA concentration in α-FMH-treated samples/HA concentration in saline-treated samples. Statistical analyses were performed with Student's t-test.

Immunocytochemistry

The methods have recently been used and described in detail (Kaslin and Panula 2001). Briefly, cryosections of brains fixed with 1-ethyl-3(3-dimethylaminopropyl)carbodiimide (Sigma, St. Louis, MO, USA) were incubated for 12 h at 4°C with an antiserum against HA (1 : 10 000 to 1 : 20 000 HA 19C, Panula et al. 1990) and/or tyrosine hydroxylase (TH) (1/5000, Diasorin, Stillwater, MN). Alexa-conjugated (488, 568) secondary antibodies (1 : 500, Molecular Probes, Eugene, OR, USA) were then applied. The specimens were examined and images were taken with a Leica TCS SP confocal microscope as described (Kaslin and Panula 2001). Images were edited with Adobe Photoshop 6.01 (Adobe systems, San Jose, CA, USA) and compiled with Corel Draw 9.0 software (Corel Corporation, Ottawa, Ontario, Canada).

Determination of HDC activity

HDC activity was determined as described by Niimi et al. (1997) with some modifications. Whole zebrafish brains were homogenized and centrifuged at 10 000 × g for 15 min. Supernatants were transferred to centrifugal filter units (Ultrafree®-MC 30 000 NWL centrifugal filter units, Millipore, MA, USA) and centrifuged three times at 5000 × g for 40 min. The enzymatic reaction was started by addition of histidine to half of the sample; the other half served as a control. After 4 h incubation at 28°C the reaction was stopped with ice-cold perchloric acid (PCA) (final concentration 2%). Produced HA was measured with HPLC and reported as fmol HA/mg tissue produced per h. Statistical analyses were performed with Student's t-test.

Quantification of HDC expression

HDC mRNA levels were analysed in adult fish injected three times with α-FMH or saline and killed 1 or 4 days after the last injection. In situ hybridizations for HDC were carried out with the same oligonucleotide and according to the same procedure as that described by Eriksson et al. (1998). Hybridized sections were exposed to BioMax MR-1 film (Eastman Kodak Company, Rochester, NY, USA) for 7–10 days. The signal intensities were analysed with a computer-based MCID image analysis system (Imaging Research, St. Catharines, Ontario, Canada). The measurements were calibrated according to an autoradiographic 14C-Micro-scale standard exposed to the same film. Values from six brains per group were used. Statistical analysis was performed using Student's t-test.

Zebrafish T-maze

The T-maze used in this study was constructed according to Darland and Dowling (2001) but with gray rather than white walls. Briefly, the system has one longer arm, the initiation arm, and a choice of two arms, of which one leads to a deeper container. Before starting the experiment, fish were pre-selected; fish that never left the initiation place (Darland and Dowling 2001) or otherwise behaved abnormally were not included in the experiment. The α-FMH- and saline-injected fish were carefully transferred to the T-maze system and tested 1 day and 4 days after the last of three repeated injections. When introduced to the maze for the first time, fish were allowed to explore the system for 5 min. The fish were then transferred to the initiation arm and kept there for 30 s before opening the gate. The time taken to find the container was measured (trial 1). The fish were left to swim for 2 min after finding the container, then transferred to the initiation arm and the test was repeated twice (trial 2 and trial 3). The same fish group was tested 1 and 4 days after the last injection. The saline group consisted of six fish and the α-FMH group of 10 fish. Statistical analysis was performed with one-way anova followed by Bonferroni's post test.

Video recording of behaviour

In the test the swimming performance of six fish were recorded simultaneously. The fish were carefully transferred to cylindrical observation tanks (inner diameter 22 cm and water depth 8 cm). In order to minimize effects on social behaviour the fish were individually tracked in separate observation tanks. The experiments were always done in a calm sealed-off area. Fish used for the study were recorded before the start of the experiment to exclude fish with abnormal behaviour, e.g. fish not swimming at all when introduced in the tank were not used in the study. The fish had 1 min of habituation time in the tank before the tracking started. The swimming performance of the individual fish was automatically detected and tracked for 10 min by a digital video camera connected to a standard PC computer system running the EthoVision Pro 2.3 software (Noldus Information Technology, Wageningen, Netherlands). The coordinates of the swimming performance were stored and used for further software analysis of the behaviour. All recordings were done using the sample rate 8.333 samples/s and analyses were performed with a down sample rate 3 giving a real sample rate of 2.77. Statistical analyses were performed with anova followed by Bonferroni's post test.

Results

Circadian variation of histamine

The HA concentration in brains at different time-points of the day was measured to clarify whether brain HA levels in the zebrafish display a circadian variation. The light period started at 08.00 and lasted to 22.00. The time-points measured were 03.00, 07.00, 08.30 (before feeding), 11.00, 15.00, 19.00 and 23.00. The lowest HA was measured at 11.00, 150 ± 12 fmol/mg tissue, and the highest just before the onset of light at 07.00, 227 ± 15 fmol/mg tissue (Fig. 1). The decrease in HA level from 07.00 and 08.30 to 11.00 was significant.

Figure 1.

Circadian brain histamine levels. The brain HA concentration was measured at different time-points during the day (n = 4–6). Data is presented as mean ± SEM. Student's t-test *p < 0,05, **p < 0.01.

L-Histidine injections increase brain histamine concentration

Only adult fish were used for histidine loading. Zebrafish brains were collected 1, 2, 3, 4, 5 and 6 h after injection of histidine (500 mg/kg i.m.) or saline and the HA concentration was measured. A significantly increased HA concentration was measured at all time-points (Table 1). One hour after injection the HA concentration was 316 ± 84 fmol/mg compared to 209 ± 15 fmol/mg in the saline-treated group. This elevation was already of the same amplitude as at the other time-points, e.g. 6 h after injection an increase from 213 ± 44 fmol/mg in the saline group to 318 ± 47 fmol/mg in the histidine-loaded group was measured. The experiment was repeated at different times of the day. No food was given during the experiment.

Table 1.  Histamine concentration in zebrafish brain after l-histidine treatment compared to saline-injected fish. Data is shown as percentage increase compared to saline. The mean histamine concentration following saline is set as 100% ± SEM, h = hours after injection and n = numbers of fish/group
hSalineHistidine
± SEM %(n)± SEM %(n)
  1. Statistical analysis was performed with Student's t-test, *p < 0.05.

1100 ± 7(3)151 ± 27*(4)
2100 ± 5(3)142 ± 20*(4)
3100 ± 8(7)119 ± 8*(5)
4100 ± 3(3)112 ± 6*(5)
5100 ± 10(3)152 ± 28*(3)
6100 ± 21(3)149 ± 15*(4)

α-FMH treatment decreases brain histamine concentration

The brain HA concentration was measured from α-FMH versus saline-treated zebrafish brains. The time points measured were 6 h, 10 h and 24 h after a single injection (100 µg/mg α-FMH), 24 h and 96 h after three injections repeated at 24-h intervals, and 24 h after five injections given with 24-h interval. A single injection of α-FMH gave rise to a diminished HA concentration at all time points measured (Fig. 2a). Six hours after injection, α-FMH-treated fish had 82% of the HA concentration measured in saline-treated fish and a decrease to 57% 24 h after a single injection was observed. Three injections of α-FMH resulted in a larger decrease of brain HA; 1 day after the last injection 44% of the HA measured in the saline-treated group was detected. The lowest level of HA was measured 4 days after the last injection. At this time point, 66% of the brain HA was depleted (59.9 ± 9.0 fmol/mg in α-FMH-treated fish compared to 177.5 ± 14.8 fmol/mg in saline-treated fish). Five injections repeated at 24-h intervals did not cause further reduction; 1 day after the last injection 40% of the normal HA concentration was still detected.

Figure 2.

(a) The HA concentrations in brains after α-FMH injections, given as percentage compared to saline-treated fish. h, sacrificed hours after last injections; n, number of fish in the group. The HA concentration in saline-treated fish, 6, 10 and 24 h after one injection, 24 and 96 h after three injections and 24 h after five injections is set as 100%. (b) The HDC activity in zebrafish brains after three injections of α-FMH and saline (n = 7). No activity is detected in the control samples with no addition of l-histidine. The activity is shown as fmol histamine produced per mg brain tissue in 1 h. (c) The HDC mRNA expression in zebrafish injected three times with α-FMH, analysed 1 and 3 days after last injection (n = 3–6). Data presented as mean ± SEM. Student's t-test p < 0.05, **p < 0.01, ***p < 0.001.

L-Histidine decarboxylase activity is decreased after α-FMH administration

The whole-brain HDC activity was determined at the time point when the lowest HA concentration was measured. Ninety-six hours after three repeated α-FMH injections 5.2 fmol HA/mg tissue/h was produced. When compared to the 21.1 fmol produced in saline controls we concluded that 24.6% of the HDC activity remained (Fig. 2b). In the controls for respective treatments (no histidine added) no HA was synthesized.

L-Histidine decarboxylase expression is increased after α-FMH administration

The HDC mRNA expression determined from in situ hybridization films 1 day after three repeated α-FMH injections was significantly increased by about 15% compared to the saline-injected group (Fig. 2c). Four days after the last injection of α-FMH this difference was not detectable any more.

Altered histamine-immunoreactivity

Adult fish used for immunocytochemistry were injected three times and killed 1 or 4 days after the last injection. Larvae were exposed to α-FMH for 10 days. α-FMH exposure caused a loss in HA-immunoreactivity in the neurones throughout the brain in both larvae and adult fish compared to saline-treated fish. The brains of α-FMH-treated 10-day-old larvae showed reduced HA-immunoreactivity (Fig. 3a). A marked decrease in histaminergic projections in all areas of the brain was evident. The histaminergic neurones in the caudal part of the hypothalamus were still clearly visible and displayed HA immunoreactivity. A closer look at the histaminergic neurones around the posterior recess in the caudal hypothalamus showed that α-FMH-treated fish had fewer but larger HA-containing vesicle aggregates compared to saline-injected ones (Figs 3c and d). One of the main termination areas for histaminergic projections is the telencephalon. The density of HA immunoreactive fibres in telencephalon was reduced both in larval (Figs 3e and f, green) and adult zebrafish. Tissue sections of the telencephalon displayed a clear loss in HA immunoreactivity (Figs 3g and h), whereas no changes in TH immunoreactivity were observed.

Figure 3.

Images of the HA-immunoreactivity in saline- and α-FMH-treated fish. (a) Distribution of HA (green) and TH (red) fibres and neurones in a horizontal whole-mount of a saline-treated larva brain (10 days) scanned from the ventral side as indicated in the insert in the top right corner. The HA-storing neurones and projections to the telencephalon are clearly visible. (b) An α-FMH-treated (10 day) larva brain stained as in (a), a reduction in HA-immunoreactivity can be distinguished but the hypothalamic HA-storing neurones are still detected (arrow). No changes in the TH-immunoreactivity are seen. (c) A larger magnification of the HA-containing neurones around the posterior recess of the diencephalic ventricle in the caudal hypothalamus of adult saline-treated fish and (d) 1 day after three injections of α-FMH. (e) Presentation of a horizontal view of a whole-mount larva showing the bundles of HA fibres (green) in the dorsal telencephalic area in saline-treated fish. TH-immunoreactivity (red) in the telencephalon, olfactory bulbs, dorsal and central hypothalamic nucleus and optic tectum are visualized. (f) A similar picture of an α-FMH-treated larva (10 days exposure). (g) A 20-µm thick cross section of an adult zebrafish telencephalon displaying the histaminergic innervation after saline treatment compared to (h) after α-FMH treatment 1 day after three injections. Dc, central zone of dorsal telencephalic area; Dd, dorsal zone of dorsal telencephalic area; Dl, lateral zone of dorsal telencephalic area; Dm, medial zone of dorsal telencephalic area; Dp, posterior zone of dorsal telencephalic area; V, ventral telencephalic area. Scale bar 100 µm applies for (a) and (b), 10 µm for (c) and (d), 50 µm for (e) and (f), 100 µm for (g) and (h).

The effect of α-FMH on T-maze performance

During the 5-min adaptation period to the T-maze system, the fish usually explored the whole system. After finding the larger container most fish spent the majority of time there. When reintroduced to the system the fish tended to swim to the large container directly, even 3 days after introduction. Zebrafish that received three injections of α-FMH were tested in the T-maze 1 day and 4 days after the last injection. No significant changes were observed in the α-FMH group compared to the saline group 1 day after treatment. The α-FMH group showed a tendency to improve in each trial; they found the container faster for each trial (Fig. 4). Four days after the last injection the same group of fish was tested again. At this time point fish were not allowed to adapt to the system. α-FMH-treated fish were significantly slower in finding the container in the first trial than the saline-treated group (Fig. 4). The α-FMH-injected fish spent significantly less time to find the large container in trial 2 and 3 compared to the time spent in trial 1. Saline-treated fish did not show any significant changes in time spent to find the container.

Figure 4.

Performance in the T-maze system after α-FMH injections compared to saline injections. Values are the mean time in seconds (s) ± SEM taken for the zebrafish to find the larger container in the T-maze system. All fish tested were injected three times with α-FMH or saline. The test was repeated three times (trail 1, 2 and 3) 1 and 4 days after the last injection. n = 5–6, anova*p < 0.01 compared to saline group. ap < 0.05 compared to first trial.

Video tracking of swimming pattern

Zebrafish injected with histidine were recorded 6 h after injection. No significant differences were detected in any of the parameters analysed: velocity, distance moved, meander, turning angle and time spent in the centre of the tank (zone IN). The traces of swimming pattern for histidine-injected compared to saline-injected fish were essentially similar (Fig. 5).

Figure 5.

Zebrafish were analysed with the Ethno vision system 6 h after histidine loading and 24 and 96 h after the last injection of α-FMH. The picture of the swimming pattern of 10 min video recording is shown for one typical fish from each group in the figure, the white square in the pictures represents the fish. No obvious difference between saline and histidine-injected fish can be visualized. A clear difference in the swimming behaviour between the saline and α-FMH-treated group can be visualized. The circle in the saline 1 day panel shows the border between the two zones IN and OUT.

The first recording of fish injected with α-FMH three times was done 1 day after the last injection. The same groups were recorded again 3 days later. The swimming pattern was different in the α-FMH-treated group compared to the saline-treated group on both days tested based on the recordings made during 10 min (Fig. 5). The saline-injected group swam more around the tank compared to the α-FMH-treated group that crossed the centre area of the tank more frequently.

The total distance moved during the 10 min recording period did not show any significant differences between the two groups (Fig. 6a). The minimum, mean (Fig. 6b) and maximum velocity were also similar. To analyse if the α-FMH-treated fish spent more time in the centre of the tank compared to the saline-treated ones, we introduced an inner zone to analyse the time spent in this zone (Fig. 5). α-FMH significantly increased the time spent in the inner zone, 24 h after the last injection (Fig. 6c). At the later time-point, a tendency to increased swimming through the centre of the tank compared to the first time-point was observed in the saline group.

Figure 6.

Software-based analysis of zebrafish swimming pattern based on 10-min recordings in the tanks. The α-FMH data is based on data obtained from 10 fish and the saline group on six fish. The activity was measured as the mean velocity during the recording (a) and the total distance moved (b). The exploratory behaviour was analysed using the total duration in IN zone (see Fig. 4) (c), absolute mean turn angle (d), relative mean angular velocity (e) and relative mean meander (f) as parameters. Data is presented as mean ± SEM. Statistical analysis was performed with anova followed by Bonferroni's post test. *p < 0.05.

The turning behaviour was analysed using the parameters mean turn angle, mean angular velocity and mean meander. The mean of the absolute turn angle was used to study the change in direction. This parameter showed a significant decrease in degrees turned in α-FMH-exposed fish at the first time-point (Fig. 6d). The mean value of the relative angular velocity is seen in Fig. 6(e). This parameter displays the degrees the fish turn per second, indicating the direction clockwise as positive values and counter-clockwise as negative values. The saline-treated fish had a negative mean value indicating more turns counter-clockwise. After α-FMH the mean value was positive, suggesting that these fish turn more clockwise than counter-clockwise. Meander is a parameter that indicates the change in direction related to the distance moved (degrees/cm). The mean of the relative meander displays the same pattern as the mean angular velocity, giving a positive value for the α-FMH-treated fish and a negative value for the control group (Fig. 6f).

Discussion

A clear difference in zebrafish brain HA levels during the light period compared to the dark period was not detected. A significant reduction in HA concentration was, however, identified in the middle of the light period (11.00 compared to 07.00 and 08.30). The concentration at 15.00 was also slightly lower than at the other time-points measured. The levels at the other time-points measured were quite similar, ranging from 202 ± 39 to 227 ± 15 fmol/mg tissue. Circadian variation in the HA concentration of goldfish has been reported (Burns et al. 2003). The goldfish has higher HA levels during the light period than during the dark period, which was not seen in zebrafish. Most zebrafish display a diurnal variation in locomotor activity, being more active during the light period (Hurd et al. 1998; Zhdanova et al. 2001).

Earlier studies on rats show that systemic loading with histidine increases the HA concentration in the brain (Schwartz et al. 1972; Bulfield and Kacser 1975; Prell et al. 1996b). The brain HA concentration in zebrafish was significantly increased at all time-points measured (12–51%). This suggests that in fish, as in mammals, histidine does not saturate HDC under physiological conditions. A 500 mg/kg systemic dose of histidine in rat was reported to give a significant increase in HA levels in the brain, only 1 h after injection. A dose of 1000 mg/kg gave a significantly higher HA concentration at all time-points measured between 1 and 6 h, with a clear peak at 3 h after injection (Prell et al. 1996b). In zebrafish the HA concentration was quite equally elevated at all time-points measured, a slight reduction at time-points 3 and 4 h was, however, seen compared to the other time points. There is a possibility that this is due to a circadian variation in HA concentration due to changes in HDC activity. The brain HA in zebrafish is slightly lower at noon compared to in the morning and late afternoon. However, at all time-points, values were compared to control brains taken at the same time as the treated brains and recalculated as a percentage, which means that circadian variation does not account for the differences. The experiment was also repeated at different times of the day. Our recordings did not show any significant changes in the zebrafish behaviour after histidine injection. Behavioural studies on goldfish after histidine injections have been reported (Medalha et al. 2000; Coelho et al. 2001), but measurements of HA levels were not included in these reports. Results from histidine loading in goldfish suggest that the histaminergic system has an inhibitory modulation on memory and learning processes (Coelho et al. 2001); whether similar effects are seen in zebrafish remains to be clarified.

α-FMH is known to reduce HDC activity and diminish the HA concentration in vivo in several tissues of different mammals (Maeyama et al. 1982; Bouclier et al. 1983; Duggan et al. 1984; Andersson et al. 1992; Andersson et al. 1996; Fujiwara et al. 2001). Despite this, the HA content and HDC activity of the CNS have never been abolished completely by administration of this drug (Garbarg et al. 1980; Maeyama et al. 1982; Duggan et al. 1984; Sakai et al. 1998). The results from this study clearly demonstrate that the α-FMH inhibits HDC in zebrafish, suggesting similar biochemical characteristics for this enzyme in mammals and teleosts. A reduction of the HA concentration was observed already on the injection day and a maximum reduction up to 76.5% was detected. At the time point of the lowest HA concentration the HDC activity was about one-quarter of the normal. The expression of mRNA for HDC was increased in α-FMH exposed zebrafish 1 day after the last injection. No significant change in HDC expression at the time when the HA concentration was as lowest was seen. The HA-immunoreactivity observed in the brains after α-FMH treatment also verified that HA was diminished but not completely depleted. Neuronal HA has a shorter half-life than the mast cell derived HA (Yamatodani et al. 1982; Maeyama et al. 1983). It has been suggested that the remaining HA after α-FMH treatment is derived from mast cells (Schwartz et al. 1991). The finding that mast cell-deficient mice exposed to α-FMH also contain detectable levels of HA lead to the suggestion that the HA present after α-FMH treatment could originate from an α-FMH-resistant pool of histaminergic neurones or a HA pool that is non-neuronal and not mast cells (Sugimoto et al. 1995). Reduced HA detected in zebrafish neurones after α-FMH exposure supports the first hypothesis, since we concluded earlier that the zebrafish does not have mast-cell derived HA (Eriksson et al. 1998).

The histaminergic neurones were still easily detected at all time-points tested after different α-FMH exposures. The histaminergic fibres were weakly fluorescent and hard to visualize in certain areas after α-FMH exposure. A smaller reduction in HA concentration in hypothalamus compared to other brain areas has been reported in rat, possibly due to the synthesis of new HDC producing fresh HA (Sugimoto et al. 1995). Hypothesizing that new HA would be synthesized in TM and transported to nerve terminals, a short delay after detection of HA immunoreactivity in TM before detection in distant areas would be reasonable. Our immunocytochemical data do not show significant intensity differences of the immunoreactivity in the HA-storing neurones in the hypothalamus at any time-point after α-FMH exposure compared to saline. No increase of HA immunoreactivity, indicating distribution of fresh HA to the fibres, was detected at the later time-points. The HA concentration and HDC activity data do not support the hypotheses of production of new HA, as no signs of reversibility were seen 4 days after the last injection of three repeated injections. HA is catabolized through methylation by histamine-methyltransferase (EC 2.1.1.8) to tele-methylhistamine extraneuronally and then further metabolized by monoamine oxidase B (EC 1.4.3.4) (Hough et al. 1982; Barnes and Hough 2002). Possibly, there is a mechanism that prevents the release of all vesicular HA when no new HA is produced. This could also include an aggregation of existing HA vesicles. Another possibility is that there is an effective re-uptake system for HA when the availability of the transmitter is restricted and the HA methyltransferase levels are low (Prell et al. 1996a).

The existing data on the effect of HA on learning and memory in mammals is contradictory. H1 receptor antagonists act as reinforcers (Katz and Goldberg 1986; Kamei and Tasaka 1993; Onodera et al. 1994; Privou et al. 1998), suggesting that HA through H1 receptor has a negative effect on memory. Memory deficit after administration of α-FMH in rats and radial-arm maze test has been reported, as well as a prolonged response latency in active avoidance response after decreased HA content of these brain areas (Kamei et al. 1993; Chen et al. 1999). Intracerebroventricular injections of HA or histidine in old rats cause significant shortening of active avoidance response (Kamei and Tasaka 1993) and a moderate reduction in brain HA levels enhances attention in visuospatial tasks under stressful stimuli (Cacabelos and Alvarez 1991). Studies on goldfish suggest that HA has an inhibitory effect on learning and memory, since the H1 blocker chlorpheniramine facilitates memory (Medalha et al. 2000) and improves appetitive learning (Spieler et al. 1999).

The definitions for short-term and long-term memory in fish have not been established. In this study we refer to the trials done during the same day, repeated at 2-min intervals, as short-term memory. The first trial on day 4 (3 days after the previous trial) is considered to test long-term memory. The performance of the zebrafish in the T-maze suggests that α-FMH does not significantly influence the short-term memory, but a marked increase in time spent to find the large container 3 days after the introduction to the maze was observed. This could be a secondary consequence due to, for example, disruption of motor processing or sedation. Since the performance of the fish improved significantly in the second and third trial, carried out 2 min after finishing the previous trial, and the video recordings did not suggest any disturbances in activity or the control of movement, we consider it more likely a memory-related phenomenon. This suggests that lack of HA impairs long-term memory in zebrafish. At the same time-point a significant shortening in time was seen in trial 2 and 3 after α-FMH exposure compared to controls, from which we might conclude that lack of HA does not affect learning or short-time memory negatively in zebrafish.

The T-maze running time for zebrafish in our experiments was generally shorter than the time reported by Darland and Dowling (2001). This might be due to color differences of the systems. The one previously was light, whereas ours was dark. The zebrafish did not improve their running time in trial 2 and 3 compared to trial 1 in our experiments. This suggests that they know the system after the 5-min adaptation period and remember where the preferred compartment is localized. Thus, the experimental paradigm used in this study does not directly test learning, which would require recording of all arrivals, including the first ones, to the preferred compartment.

No significant changes were found in locomotor activity between α-FMH and control groups, which is in agreement with earlier results from rat (Sakai et al. 1998). In our experiment, both groups of fish swam similar distances at about equal speed. Naive zebrafish tended to swim along the edges of the circular tank when alone. The swimming pattern showed that there were differences between saline- and α-FMH-treated fish: analysis of the time spent in zone IN confirmed that the α-FMH-treated fish spend more time in the centre of the tank. Saline-treated and untreated fish tended to swim along the walls of the tank. On day 4 the saline-treated fish spent more time in the middle of the tank than on day 1. The swimming around the walls can be interpreted as an attempt to find shelter or to be close to an object when the shoal is absent. It is possible that the fish swimming more in the centre feel less anxious, which would suggest that a reduction in HA produces a decrease in anxiety. When fish were reintroduced to the tanks 3 days later, saline-treated fish spent more time in zone IN than on the first day, suggesting that the fish recall the environment and feel secure. These results are in agreement with studies on anxiety and HA on mammals (Imaizumi and Onodera 1993; Frisch et al. 1998) and the less anxious H1 knock-out mice (Yanai et al. 1998).

One of the major termination fields for the histaminergic neurones is in the telencephalon and especially the dorsal telencephalic area. Very dense histaminergic innervation is seen in medial parts and moderate levels in the other parts of the dorsal telencephalon (Eriksson et al. 1998; Kaslin and Panula 2001). The dorsomedial parts of telencephalon in teleosts share similarities with the mammalian amygdala (Bradford 1995; Portavella et al. 2002). The dorsolateral telencephalon is considered the teleost's correspondence to the mammalian hippocampus. In α-FMH-treated fish, HA was depleted quite extensively from these areas and a clear change in swimming pattern and T-maze performance could be detected. The area corresponding to the mammalian amygdala (dorsomedial telencephalon) is more extensively innervated with HA than other parts of telencephalon (Kaslin and Panula 2001), suggesting that HA is more involved in emotional memory (e.g. anxiety) than spatial/temporal memory. Taken together, these findings suggest that HA is involved in memory, learning and anxiety in zebrafish. It still has to be clarified whether the effect of HA is direct or indirect and which of the HA receptors mediate the effect. However, data from GTP-γ-[35S] autoradiography does not indicate any differences in the H3 receptor activity after α-FMH exposure (not shown).

It has been predicted before that histaminergic drugs display few or no effects in peripheral tissues of bony fish without gastrointestinal HA (Reite 1972). This study shows that endogenous HA is regulated in zebrafish in the same manner as in mammals, and changes in it are associated with significant behavioural alterations. The existence of several hundred CNS-affected zebrafish mutants now makes it feasible to study how the brain HA system develops and which genes are essential for its functional regulation.

Acknowledgements

This work was supported by the Magnus Ehrnrooth foundation, Svenska Kulturfonden, Stiftelsen för Åbo Akademi, the Sigrid Juselius Foundation and the Academy of Finland. The authors have contributed to the following parts of the study: Nina Peitsaro: general fish maintenance, treatment of fish with drugs, measurement of HDC activity, expression of HDC mRNA, GTP-γ-S analysis, T-maze analysis, video recordings of behaviour, circadian activity, all statistical analyses and writing of the manuscript; Jan Kaslin: general fish maintenance, treatment of fish with drugs, histamine HPLC, immunohistochemistry, video recordings of behaviour, circadian activity, analysis of fish larvae and comments on manuscript; Oleg Anichtchik: experiments with l-histidine, video recording of behaviour, GTP-γ-S analysis and comments on the manuscript; Pertti Panula: general design of the experimental setups, responsible for analysis and interpretation of results, final comments on all manuscript versions and funding of the project. The authors are grateful to Levente Bascó for help with fish keeping.

Ancillary