MPTP and MPP+ target specific aminergic cell populations in larval zebrafish


  • V Sallinen,

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • V Torkko,

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • M Sundvik,

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • I Reenilä,

    1. Institute of Biomedicine/Pharmacology, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
    • 1

      The present address of I. Reenilä is the Division of Pharmacology and Toxicology, Faculty of Pharmacy, Viikinkaari 5E, 00014, University of Helsinki, Helsinki, Finland.

  • D Khrustalyov,

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • J Kaslin,

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
    • 2

      The present address of J. Kaslin is the Max Planck Institute for Molecular Cell Biology and Genetics, Pfotenhauerstrasse 108, 01307 Dresden, Germany.

  • P Panula

    1. Neuroscience center and Institute of Biomedicine/Anatomy, Haartmaninkatu, University of Helsinki, Helsinki, Finland
    Search for more papers by this author

Address correspondence and reprint requests to Pertti Panula, Neuroscience Center, Institute of Biomedicine/Anatomy, POB 63, 00014, University of Helsinki, Helsinky, Finland. E-mail:


Larval zebrafish offers a good model to approach brain disease mechanisms, as structural abnormalities of their small brains can be correlated to quantifiable behavior. In this study, the structural alterations in one diencephalic dopaminergic nucleus induced by 1-methyl-4-phenylpyridinium (MPP+), a toxin inducing Parkinson’s disease in humans, and those found in several neuronal groups after 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), the pretoxin, were associated with decreased swimming speed. Detailed cell counts of dopaminergic groups indicated a transient decline of tyrosine hydroxylase expressing neurons up to about 50% after MPTP. The MPTP effect was partly sensitive to monoamine oxidase inhibitor deprenyl. Detailed analysis of the developing catecholaminergic cell groups suggests that the cell groups emerged at their final positions and no obvious significant migration from the original positions was seen. One 5-HT neuron group was also affected by MPTP treatment, whereas other groups remained intact, suggesting that the effect is selective. New nomenclature for developing catecholaminergic cell groups corresponding to adult groups is introduced. The diencephalic cell population consisting of groups 5,6 and 11 was sensitive to both MPTP and MPP+ and in this respect resembles mammalian substantia nigra. The results suggest that MPTP and MPP+ induce a transient functional deficit and motility disorder in larval zebrafish.

Abbreviations used

blood–brain barrier


dopamine transporter


monoamine oxidase A


monoamine oxidase B






phosphate-buffered saline


Parkinson’s disease


paraventricular organ anterior


paraventricular organ intermediate


paraventricular organ posterior


tyrosine hydroxylase

Parkinson’s disease (PD) is a progressive and common neurodegenerative disease, affecting over 1% of the population aged over 60 years (Nussbaum and Ellis 2003). It is characterized by resting tremor, hypokinesia and rigidity caused by the loss of dopaminergic cells in the substantia nigra pars compacta and subsequent loss of dopaminergic input into the striatum (Hornykiewicz and Kish 1987). The etiology of PD is unclear, and no therapy based on specific disease mechanisms exists. More information of the disease pathogenesis is needed for new therapies.

The MPTP-induced model of PD, which has been widely used in several species, appears to be the best one available (Beal 2001). MPTP causes cell death of dopaminergic cells of substantia nigra by inhibiting the complex 1 of mitochondrial respiratory chain via its toxic metabolite MPP+ produced by monoamine oxidase B (MAO B, EC (Nicklas et al. 1985).

Zebrafish models are being increasingly used to study human diseases (Hsu et al. 2007), much due to its low housing cost, large progeny size, rapid and transparent development and ease of manipulation with genetic and pharmacological methods (Streisinger et al. 1981). Since zebrafish is a vertebrate, studies on it may be more directly applicable to humans than those performed on non-vertebrate models such as Drosophila melanogaster and Caenorhabditis elegans. Zebrafish larvae are an attractive model to study neurological diseases, since whole neurotransmitter systems can be imaged three-dimensionally (Panula et al. 2006).

Adult zebrafish are sensitive to MPTP and exhibit a clear behavioral pattern accompanied with changes in brain catecholamines (Anichtchik et al. 2004). The aim of this study was to characterize the MPTP-model of PD in larval zebrafish. This was achieved by first studying the development of dopaminergic and noradrenergic neurotransmitter systems and then examining the effects of MPTP and MPP+ on these neuron populations and behavior of larval zebrafish. Since previous studies have provided somewhat conflicting results (Bretaud et al. 2004; Lam et al. 2005; McKinley et al. 2005; Wen et al. 2008), we chose to analyze detailed cell counts and quantify the behavioral alterations.

Materials and methods

Fish handling and preparation

Zebrafish of an outbred Turku strain maintained in the laboratory were used. The Turku strain has been maintained in the laboratory for more than 10 years. It has been used in several earlier studies in our laboratory (Kaslin and Panula 2001; Anichtchik et al. 2004, 2006; Panula et al. 2006). The Turku strain fish were used because of better yield of viable embryos than some other tested breeds (e.g. AB strain). Fish feeding, breeding and maintenance were performed according to Westerfield (1995). Fish were killed by immersion in ice-cold water followed by decapitation. Brains of 4-day-old or older larvae were dissected prior to whole-mount staining procedure. The permits for the experiments were obtained from the Office of the Regional Government of Southern Finland in agreement with the ethical guidelines of the European convention.


All incubation steps were carried out on a shaker. Specimens fixed with 4% paraformaldehyde or 4% 1-ethyl-3,3(dimethyl-aminopropyl) carbodiimide (EDAC, CMS Chemicals, Abingdon, UK) and 2% paraformaldehyde were washed in phosphate-buffered saline containing 0.3% Triton X-100 (PBS-T) for 3 × 30 min at room temperature (RT), then pre-incubated with PBS-T containing 1% dimethylsulfoxide and 4% normal goat serum overnight (o/n) at 4°C. Incubation with monoclonal mouse anti-tyrosine hydroxylase (TH) antibody (Diasorin, Stillwater, MN, USA) diluted 1 : 1000, rabbit anti-serotonin (Sigma, St. Louis, MO, USA) diluted 1 : 1000 or rabbit anti-histamine (Panula et al. 1990) diluted 1 : 5000 was performed in PBS-T containing 2% normal goat serum o/n at 4°C. Specimens were then washed in PBS-T 10 min + 3 × 30 min at RT. Incubation with goat anti-mouse or anti-rabbit Alexa-conjugated secondary antibodies (488 or 568, Molecular Probes, Carslbad, CA, USA) was performed in PBS-T o/n at 4°C and then washed in PBS-T 10 min + 3 × 30 min at RT. Specimens were infiltrated with 80% glycerol in PBS o/n and then mounted on microscope slides with glass spacers to prevent compression and subsequent distorsion artifacts. The TH and serotonin antibodies have been characterized earlier (Kaslin and Panula 2001).

Monoamine oxidase enzyme histochemistry

Larvae fixed with 4% EDAC o/n were washed 2 × 15 min in 0.05 M Tris–HCl buffer pH 7.6, followed by 2 h incubation in Tris–HCl containing 6 mg/mL NiSO4, 1 mg/mL tyramine, 1 mg/mL horseradish peroxidase and 0.8 mg/mL diaminobenzidine to demonstrate MAO enzyme activity as described in detail (Anichtchik et al. 2006). Samples were then washed 2 × 15 min in PBS and then infiltrated with 80% glycerol and mounted on microscope slides as described in the previous chapter for immunohistochemistry. All reagents were purchased from Sigma.

HPLC measurements

Tissue samples, each consisting of 20 pooled 5-day-old zebrafish larvae, were weighed and sonicated on ice in 150 μL of solution containing 0.4 M perchloric acid, 0.1% Na2S2O5 (w/v) and 0.1% EDTA (w/v) followed by centrifugation for 20 min at 27 000 g at + 4°C. Samples were then filtered through Acrodisk syringe filters (13 mm) with 0.45 μm polyvinylidene difluoride membrane and stored at −80°C until determination. Details of the HPLC method are described in Supporting information.

Drug treatment protocol and statistics

Fish were raised in six-well plates (30 fish per well) containing E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4) with or without the drugs in a total volume of 3 mL. They were briefly transferred to a petri dish at 24 h post-fertilization (hpf) for manual dechorionation. The solution of each group was disposed of completely daily and replaced with a fresh pre-warmed solution. Fish were exposed to a solution containing MPTP (Sigma) 100 μM or 1000 μM on days 1–4 post-fertilization and/or 100 μM deprenyl (Sigma) on days 0–4 (for behavior and HPLC) or on days 0–5 (for TH-immunohistochemistry) dpf. MPP+ (Sigma) was used similarly at concentrations 100, 500 and 1000 μM on days 1–4. Control groups received no drugs. Drug treatments were carried out as three independent experiments and results were combined. MPTP and MPP+ exposures were carried out and toxins were disposed of according to current safety protocols (Przedborski et al. 2001).

Statistical analyses were performed using GraphPad Prism 4.01 software (GraphPad Software Inc., San Diego, CA, USA). For multigroup analysis, anova was used followed by Dunnett, Newman–Keuls or Bonferroni post-hoc test. Statistical analysis and number of samples used in each experiment is stated in the figure legend. All bars indicate mean ± SEM.

Behavioral analysis

Larval behavior was analyzed in a calm sealed-off area using a digital video tracking system consisting of a CCD camera connected to a computer, and the acquired images were processed using EthoVision 3.1 video tracking software (Noldus Information Technologies, Wageningen, The Netherlands) as described (Panula et al. 2006). Only fish exhibiting normal gross phenotype were used for the analysis. A few abnormal larvae were found in all groups during the days of experiments (Supporting information Table S1). In order to minimize effects on social behavior the fish were individually tracked in separate observation chambers. The larvae swam during the tracking in a hemisphere-shaped dish (diameter 4 cm, volume 12 mL) or in the wells of a 48-well plate. In the test, the swimming performance was recorded simultaneously for up to 48 fish. Details of the behavioral analysis are described in Supporting information methods.

Microscopy and image analysis

Intact fish were examined using a Leica DM IRB inverted microscope (Leica Microsystems, Mannheim, Germany) with a Leica DFC 490 Color Digital Camera connected to a computer and imaged using Leica Application Suite 2.7.0 R1 software. MAO enzyme histochemistry and in situ hydridization samples were examined using Olympus AX70 light microscope (Olympus Corporation, Tokyo, Japan) with Olympus DP 71 camera and imaged using CellA software (Olympus Soft Imaging Solutions GmbH, Münster, Germany). Immunofluorescence samples were examined using a Leica TCS SP2 confocal microscopy system, using an Argon laser (488 nm) and green diode laser (561 nm). Sequential scanning and frame averaging were used to reduce the noise and cross-talk between the channels. Stacks of images taken at 0.2–1.2 μm intervals were compiled to produce maximum intensity projection images with Leica Confocal Software. Numbers of cells in different nuclei were counted from whole stacks, which were scanned through the entire cell populations to obtain the absolute number of cells in a population. Within each individual experiment all the samples were scanned using the same parameters. Images were further processed using ImageJ 1.37a software (National Institutes of Health, Bethesda, MD, USA). Final images were compiled using Corel Draw 13.0 software (Corel Corporation, Ottawa, Canada). The animation was made using Imaris 6.0 software (Bitplane AG, Zurich, Switzerland) and compiled using Quicktime 7.4.1 Pro software (Apple Inc., Cupertino, CA, USA). Anatomical structures were named and/or numbered using the neuroanatomical atlas of adult zebrafish brain (Wullimann et al. 1996) and an atlas based on location of aminergic neurons in adult zebrafish (Kaslin and Panula 2001).


MPTP affects the locomotion of zebrafish larvae

The two MPTP concentrations used (100 and 1000 μM) were chosen after pilot studies. Only the 1000 μM concentration produced a consistent significant locomotor deficit (Fig. 1a and b). Swimming speed and thus the total distance moved decreased in the groups treated with 1000 μM MPTP. The alteration was delayed, emerging 1–2 days after the exposure to MPTP, i.e., at 5–6 dpf (Fig. 1a), being maximal at 6 dpf. The effects were transient and the swimming was recovered at 7–8 dpf. A typical behavioral effect on the swimming pattern from one experiment is shown (Fig. 1b). In mammals, MAO B activity is required for MPTP toxicity. Deprenyl has been shown to alleviate the neurotoxicity of MPTP by inhibiting MAO and thus the production of the toxic compound MPP+ (Heikkila et al. 1984). Deprenyl attenuated the behavioral deficit induced by 1000 μM MPTP by increasing the total distance moved 135% at 5 dpf (Fig. 1a). Deprenyl could not alleviate the deficit significantly at 6 and 7 dpf (Fig. 1a and b), although a tendency was seen.

Figure 1.

 (a–b) The behavioral effect of MPTP (days 1–4) and/or deprenyl (days 0–4) on 5–7-day-old larvae. (a) Total distance moved at different time points. n = 9–12 per group, *p < 0.05, **p < 0.01, ***p < 0.001, anova followed by Bonferroni post-hoc test for selected pairs: control versus deprenyl 100 μM, control versus MPTP 100 μM, control versus MPTP 1000 μM, MPTP 100 μM versus MPTP 100 μM + deprenyl 100 μM, MPTP 1000 μM versus MPTP 1000 μM + deprenyl 100 μM. Comparisons were made within each age-group. (b) Characteristic swimming patterns of 5 dpf larvae. (c) Concentrations of DA, NA and 5-HT in MPTP treated 5-day-old zebrafish larvae. 1000 μM MPTP decreased DA, NA and 5-HT, while 100 μM MPTP did not affect the levels of these amines. 100 μM deprenyl prevented the decrease of these amines significantly. *p < 0.05, **p < 0.01, ***p < 0.001. n = 3. anova followed by Newman–Keuls post-hoc test. (d) Effect of MPTP and/or deprenyl treatment on the number of TH-ir cells at 5 dpf. Both concentrations of MPTP decrease the number of TH-ir cells in populations 3–4, 5,6,11, 7 and 13. Exposure to MPTP 1000 μM decreased the number of the cells more than exposure to MPTP 100 μM. Deprenyl treatment significantly prevented the loss of TH-ir cells after exposure to MPTP 1000 μM in these populations. Samples are merged from three individual experiments. * or #p < 0.05, **p < 0.01, *** or ###p < 0.001, anova followed by Bonferroni post-hoc test for selected pairs: control versus MPTP 100 μM, Control versus MPTP 1000 μM, MPTP 100 μM versus MPTP 100 μM + deprenyl 100 μM, MPTP 1000 μM versus MPTP 1000 μM + deprenyl 100 μM. n = 12–16.

Effects of MPTP on catecholamine and 5-HT levels

To assess the neurochemical alterations caused by MPTP, DA, NA and 5-HT levels were measured in whole larvae using HPLC at 5 dpf. 100 μM MPTP did not significantly decrease DA, NA or 5-HT levels, whereas 1000 μM MPTP decreased DA levels 61%, NA levels 40% and 5-HT levels 51% (Fig. 1c). When fish treated with 1000 μM MPTP received also 100 μM deprenyl, DA, NA and 5-HT levels were 69%, 32% and 71% higher than in fish receiving only 1000 μM MPTP, respectively (Fig. 1c). This means that deprenyl significantly prevented the decrease of these amines induced by 1000 μM MPTP.

The development of dopaminergic and noradrenergic systems

In order to precisely locate dopaminergic cell populations and accurately quantify the cell numbers, a three-dimensional topological map of the dopaminergic and noradrenergic cell populations was created (See Supporting information Video Clip S1 and Fig. S1). Cell groups were assigned numbers from rostral to caudal groups, which were distinguishable after 3D examination. This nomenclature is different from the one proposed by Rink and Wullimann (2002), which focused primarily on the diencephalic populations. The development is briefly described here with focus on the timed appearance and the relationship of the cell groups to the adult populations (summarized in Fig. 2a and b and presented as an atlas in Supporting information Fig. S1).

Figure 2.

 (a) Sagittal view of the distribution of catecholaminergic (green) cell clusters in the adult zebrafish brain numbered in rostrocaudal direction (correlated to the maps of Kaslin and Panula 2001). (b) Horizontal schematic overviews outline the catecholamine containing cell clusters in the 72 hpf zebrafish brain. The cell populations are stacked on top of each other. The color code shows the dorsoventral distribution of the catecholaminergic cell clusters. The catecholaminergic populations with a more dorsal position are darker green and the ones with a ventral position are light green. Correlation to adult populations: 1: olfactory bulb (Ob); 2: ventral telencephalic nuclei (Vv, Vd, Vs); 3–4: parvocellular preoptic nucleus, anterior part (PPa); 5: parvocellular preoptic nucleus, posterior part (PPp); 6: anterior, intermediate, ventrolateral and ventromedial thalamis nuclei (A, I, VM, VL); 7: periventricular pretectal nucleus, ventral part (PPv); 8: paraventricular organ, anterior part (PVOa); 9: paraventricular organ, intermediate part; 10: paraventricular organ, posterior part (PVOp); 11: periventricular nucleus of posterior tuberculum (TPp); 12: paraventricular organ (PVO), posterior tuberculum (PTN); 13: periventricular hypothalamus (Hc, Hd), posterior tuberculum (PTN); 14: locus coeruleus (LC); 15–16: vagal lobe (LX), internal retinal formation (IRF); 17: vagal lobe (LX), area postrema (AP). (c) First TH-ir neurons appear early in the development and most of the adult aminergic cell clusters are readily detectable at 3 dpf. Number of catecholaminergic neurons increase rapidly during 2–3 dpf. A sagittal view of a 50-hpf-old zebrafish embryo showing an overview of the distribution of TH-ir cell clusters. The catecholaminergic domains in the telencephalon, hypothalamus, locus coeruleus and hindbrain are all readily detectable. (d) A horizontal overview from the dorsal side of a 60-hpf-old zebrafish embryo. Note the prominent descending TH-ir fiber projections and the sparse innervation of TH-ir to the telencephalon. (e) A horizontal overview from the ventral side showing the distribution TH-ir neurons and fibers in the forebrain of 3-dpf-old zebrafish with depth mapping of the TH-ir clusters and numbering of clusters in respect to their counterparts in the adult. The depth is indicated by colors (blue, yellow, red). The most ventral structures are shown in blue, intermediate structures are light blue to green and the most dorsally lying structures are yellow and red. (f) An overview of the hypothalamic TH-ir (yellow) and HA-ir neurons (blue) shown with higher magnification. The HA-ir neurons were distributed around the TH-ir neurons that were lining the posterior recessus. (g) A single optical plane located close to the ventral surface. Ascending TH-ir (yellow) fiber tracts originating from the neurons located around the posterior recessus were detected. The pituitary is innervated by TH-ir fibers from the preoptic area (arrows). Scale bar 100 μm in (c) and 50 μm in (d–g). (h–k) TH-ir populations in 5 dpf zebrafish brain. Populations are numbered in respect to their counterparts in the adult (see Fig. 3a and b). Images are thick (∼20–40 μm) maximum intensity projections from a larger stack montage made of four stacks. They are imaged from the ventral side and thus because of the thickness of the sample dorsal structures appear dimmer. Level of projection is shown in the illustration on the lower left corner of the images. (h) TH-ir neurons were detected in olfactory bulbs (1) and medial telencephalon (2). The telencephalic population (2) sent lateral projections into the telencephalon. The most dorsally detected TH-ir cluster was found in the pretectum (7). Lateral projections from this population into the optic tectum were seen (arrow). Scale bar is 50 μm (applies to h–k). (i) TH-ir fibers from the population 10 can be followed into the telencephalon and the Cpop (arrows). (j) An ovoid population (5, 6, 11) of TH-ir small neurons is detected rostrally to the thin column of large TH-ir neurons (12). The TH-ir neurons in the LC (14) are readily detected at this level. Note the prominent connections from LC to raphe area (R). Caudal populations (15–17) are also readily detected at this level. (k) The ventral preoptic TH-ir cells (3–4) are situated in a domain that stretches from the Cpop to the Cant. The medially situated fountain- shaped complex (13) of TH-ir neurons is easily distinguished at this level. A crescent- shaped population (8) of periventricular neurons was found anterior to this complex. Note the prominent tract of ascending TH-ir fibers (arrows) that emanates from cell cluster 13. A population of TH-ir neurons (9) is located in the proximity to the lateral recessus.

The first tyrosine hydroxylase-immunoreactive (TH-ir) neurons were detected in the rostral diencephalon just before 24 hpf in close proximity to the nucleus of the tract of the post-optic commissure (nTPOC) also known as the ventrorostral cluster. The onset of TH and dopamine transporter (DAT) mRNA expression has been reported in a similar position from 18 hpf and onwards (Holzschuh et al. 2001). As soon as the TH-ir neurons were detected they emanated prominent descending axons that projected through the diencephalon into the spinal cord. The morphology of these first cells resembled the ones later found in the adult population 12 (Fig. 2a–e and j). Between 24–48 hpf the number of TH-ir neurons increased significantly in the hypothalamus (Fig. 2c). At 48 hpf all the major TH-ir cell types that are seen in the adult brain were present in this diencephalic complex (small, large and bipolar periventricular cells). No evident migration of TH-ir neurons was seen, and as soon as the TH-ir neurons were detected they were localized in similar position and displayed similar morphology with their adult counterparts. TH-ir neurons of the diencephalic complex (populations 12–13) were detected from 24–48 hpf. A domain of TH-ir neurons (populations 5, 6, 11) was distinguishable in the thalamus and posterior tuberculum from 48 hpf (Fig. 2a–e and j). The periventricular hypothalamic populations of TH-ir neurons were noticed in a sequential order. The paraventricular organ anterior (APVOa, population 8) was detected at 48 hpf, the paraventricular organ posterior (APVOp, population 10) at 60 hpf and the paraventricular organ intermediate (APVOi, population 9) at 72 hpf (Fig. 2a–e, i and k). Histaminergic neurons were detected at 60 hpf in the PVOp among the TH-ir neurons (Fig. 2f and g). A group of TH-ir neurons was detected in the olfactory bulb and telencephalon (populations 1–2) on 48 hpf (Fig. 2a–e and h). The pretectal population (population 7) of TH-ir neurons was detected from 60 hpf (Fig. 2a, b, e and h). A population of TH-ir neurons was detected in the ventral preoptic area (populations 3–4) at about 60 hpf (Fig. 2a, b, e and k). The pituitary was innervated by fibers emanating from this population (Fig. 2g). The preoptic population originated near the ventral floor and in the proximity to the posterior commissure (Cpop). A small population of TH-ir neurons was detected in the locus coeruleus (population 14) from 24 hpf. Two bilateral columns (populations 15–17) of TH-ir neurons were detected caudally in the rhombencephalon from 36 hpf (Fig. 2a, b and j). Neuron numbers in diencephalic TH-ir groups are given in Fig. 1d.

All adult TH-ir populations could be detected already at 72 hpf. However, all clusters were not easily separable before 5 dpf (populations 1–2, 3–4, 5,6,11, 9–10 and 15–16). Populations 5, 6, 11, forming the adult of posterior tuberculum (11) and post-optic (5) and thalamic parts (6), were not separable even then. Comparison with other publications concerning the development of the catecholaminergic cell populations can be found in Supporting information Table S2.

MPTP affects several dopaminergic neuron populations

MPTP exerted effects on several TH-ir neuron populations. Both concentrations used – 100 and 1000 μM – produced a dose-dependent loss of TH-ir neurons in the same cell populations at 5 dpf. The pretectal (7) and diencephalic (5,6,11) populations were most severely affected, followed by preoptic (3–4) and hypothalamic (13) cell groups (Figs 1d and 3). The cell number in the anterior and posterior parts of the paraventricular organ (8, 10), locus coeruleus (14) and posterior tuberculum (12) were not affected (data not shown).

Figure 3.

 Effect of MPTP and/or deprenyl treatment on TH-immunoreactive cell populations on 5-days-old zebrafish larvae. Both MPTP concentrations decrease cell number in populations 13, 7, 3–4 and 5,6,11. (a–e) Population 13 (f–j) Population 7 (pretectum) (k–o) Population 5,6,11 (p–t) Population 3–4 (preoptic). Abbreviations: M – MPTP, D – deprenyl. Scale bars in A = 20 μm (applies a–e), F = 20 μm (applies f–j), K = 20 μm (applies k–o), p = 20 μm (applies p–t).

Deprenyl, a MAO B inhibitor, alleviated significantly the TH-ir cell loss induced by 1000 μM MPTP in all cell populations studied, but not using MPTP in concentration at 100 μM (Figs 1d and 3). This may be explained by smaller cell loss at the lower concentration. Thus, the recovery could not reach statistical significance although a tendency was seen at lower MPTP dose.

As the behavioral effect was transient, TH-ir was also studied on 7 dpf, i.e. after a two-day recovery period following the MPTP exposure. Remarkably, all cell populations affected at 5 dpf recovered during the two-day-long recovery period and no differences in TH-ir could be observed between control and MPTP exposed fish after this (Supporting information Figs. S2 and S3). Additionally, we failed to show cell death or apoptosis at several time points (3, 5 and 7 dpf) using TUNEL staining, lectin and acridine orange staining (data not shown). Positive control experiments (Dnase treatment for TUNEL) indicated that the method worked.

Co-localization of MAO activity with catecholaminergic and histaminergic markers

As MAO is the key enzyme in the toxicity of MPTP, the co-localization of MAO activity with histamine-immunoreactivity and TH-ir was studied at 4 dpf, when all adult catecholaminergic cell populations had emerged. MAO activity was not detected in catecholaminergic cell populations 1–4 (Supporting information Fig. S4a and S4b). However, MAO activity was detected in cells rostral to TH-ir cells in the preoptic area (Supporting information Fig S4b and d). MAO activity was detected in the TH-ir cells of the rostral part of the diencephalic complex (12–13), but not in the TH-ir cells in the caudal part (Supporting information Fig. S4b, e and f). Also the TH-ir fiber tract to the pituitary contained MAO activity (Supporting information Fig. S4a). TH-ir cells in the PVOp (10) did not contain MAO activity (Supporting information Fig. S4c). Histaminergic cells in the PVOp area contained MAO activity (Supporting information Fig. S4g and h).

MPP+ affects behavior and one dopaminergic cell population

A possible reason for recovery and lack of cell death might be that MPTP is a relatively weak substrate for zebrafish MAO (Anichtchik et al. 2006), and thus might not produce the toxic metabolite MPP+ in sufficient quantities. Consequently, we tested whether MPP+ could induce loss of TH-ir cells and behavioral alterations in larval zebrafish. After treatment of larval zebrafish from 1 to 4 dpf with MPP+, the same kind of delayed transient decrease in total distance moved was observed (Fig. 4f–h). A dose-dependent effect of MPP+ was visible at 6 dpf (Fig. 4g) and returned to control level at 7 dpf (Fig. 4h). These results show that the effect of MPP+ on locomotor activity is similar with the effect induced by MPTP.

Figure 4.

 (a–e) Effect of different concentrations of MPP+ on dopaminergic population 5,6,11 in 5-day-old zebrafish larvae. (a–d) Maximum intensity projection image of the whole population 5,6,11 showing that MPP+ dose-dependently decreases the number TH-ir positive cells. Scale bar in A is 20 μm (applies to a–d). (e) Quantitative analysis of cell number of population 5,6,11 at 5 dpf after MPP+ treatment. n = 9 per group. Samples are merged from 3 individual experiments. (f–h) Effect of different concentrations of MPP+ on larval zebrafish locomotion after treatment on days 1–4. Total distance moved and swimming patterns of (f) 5 dpf larvae (g) 6 dpf larvae (h) 7 dpf larvae. Number of samples (n) is stated on the bars. Samples are merged from 3 individual experiments. (i–k) Concentrations of DA (i), NA (j) and 5-HT (k) in MPP+ treated 5-day-old zebrafish larvae. 500 μM and 1000 μM MPP+ decreased the levels of DA. 1000 μM, but not 500 μM MPP+, decreased the levels of NA. MPP+ did not significantly alter the levels of 5-HT. n = 3. *p < 0.05, **p < 0.01. anova followed by Dunnett post-hoc test (applies e–k).

DA and NA levels in MPP+ treated larvae were measured at 5 dpf using HPLC. 100 μM MPP+ did not have significant effect on the levels of DA or NA (Fig. 4i and j). 500 μM MPP+ decreased DA levels 41% (Fig. 4i), but did not decrease NA levels significantly (Fig. 4j). 1000 μM MPP+ decreased DA levels 48% (Fig. 4i) and NA levels 38% (Fig. 4j). 5-HT levels were not significantly altered by any of the MPP+ concentrations used (Fig. 4k).

It was initially hypothesized that MPP+ would induce higher loss of TH-ir cells than MPTP. Surprisingly, MPP+ caused a dose-dependent loss of TH-ir cells in the dopaminergic population 5,6,11 while leaving other studied catecholaminergic populations unaffected (Fig. 4a–e, Supporting information Fig. S5).

Non-catecholaminergic effects of MPTP and MPP+

Mortality of the MPTP or MPP+ treated fish did not differ significantly from the controls (Supporting information Table S1). A few fish treated with MPTP or MPP+ exhibited pericardial edema, while most of the fish were indistinguishable from the control fish (Supporting information Fig. S6). Serotonergic cell populations (groups 1–8) were studied and 5-HT-ir cells were counted. Numbering of the populations is based on the classification described in Sallinen et al. (in preparation). 1000 μM MPTP caused loss of 40% of 5-HT-ir cells in the intermediate part of the paraventricular organ (PVOi, 5-HT population 3, Fig. 5a–c). No significant alterations were detected in other populations, and 100 μM MPTP did not affect any 5-HT cell population (Fig. 5a, Supporting information Fig. S7). MPP+ did not cause significant alterations in any 5-HT populations studied (Fig. 5a, Supporting information Fig. S7).

Figure 5.

 Effects of MPTP and MPP+ on serotonergic intermediate part of paraventricular organ (PVOi) (a) 1000 μM MPTP (1–4 dpf) caused 40% loss of 5-HT-ir cells in PVOi. 100 μM MPTP or MPP+ did not cause significant loss of cells. **p < 0.01, anova followed by Dunnett post-hoc test. n = 5–6. (b) Maximum projection image of PVOi of control. (c) Maximum projection image of PVOi of larvae treated with 1000 μM MPTP. Scale bar in (b) is 25 μm (applies to b and c).


Zebrafish is increasingly being used to model human diseases (Hsu et al. 2007). Despite the advantages of zebrafish, such as cost efficiency, screening potentials and small size, there are caveats as well. An important factor to consider is the relative similarity of the zebrafish systems compared with human systems. Thus, the basic properties of catecholaminergic systems and parkinsonian models must be understood before more complex experimental designs or proceeding to large-scale screening. In order to evaluate the MPTP-model of Parkinson’s disease in larval zebrafish, it was necessary to first study the developing catecholaminergic systems and their relationship to the same cell populations in adult zebrafish and other species.

The development of catecholaminergic systems

Reports of the development of catecholaminergic cell populations have been published earlier. While some of these nicely report the timed appearance of catecholaminergic cell groups, they fail to deal with the diencephalic complex as separate populations (populations 5–6 and 8–13) (Guo et al. 1999; Holzschuh et al. 2001). Another report finely dissects these populations, but lacks detailed description of the time scale and hindbrain populations (Rink and Wullimann 2002). Comparison to these publications can be found in the Supporting information Table S2. Our analysis essentially fills these gaps and provides a method for counting the absolute number of cells. In addition, fibers can be readily detected and followed. Contrary to earlier reports, we found that all adult catecholaminergic cell populations are detectable already at 72 hpf, while they are more readily separable at 5 dpf. The difference in the appearance of the populations might be related to the fact that the method used is more sensitive and cells with lower concentration of TH can be detected.

The development of the catecholaminergic system in zebrafish followed a very similar spatiotemporal sequence of appearance to the other reported teleosts (Manso et al. 1993). In all these teleosts, the catecholaminergic cells are first detected in the diencephalon and the locus coeruleus, followed by cells in the suprachiasmatic/tuberal area and the rhombencephalon. Later, just before hatching catecholaminergic cells are found in the pretectum, rostral preoptic area and the periventricular hypothalamus. The telencephalic catecholaminergic cells are detected earlier in zebrafish than in the other teleosts (Manso et al. 1993). In mammals, appearance of the midbrain dopaminergic neurons preceed the appearance of diencephalic and telencephalic neurons (Di Porzio et al. 1990; Kalsbeek et al. 1992). The homologue of midbrain dopaminergic neurons in teleost is currently unknown. Generally, the zebrafish catecholaminergic systems share similarities with other vertebrates, but also differ from them significantly. These differences must be taken into account when using zebrafish to study the dopaminergic system.

MPTP induced behavioral deficit

The locomotor network of zebrafish develops through different steps starting with spontaneous tail coilings at 17 hpf and reaching a mature beat-and-glide swimming pattern at 5 dpf (Brustein et al. 2003). In our study, a large-scale behavioral motility analysis was applied to 5–7-day-old zebrafish. MPTP produced a delayed and transient locomotor activity deficit in larval zebrafish, which was attenuated by deprenyl treatment. MPTP decreased the levels of DA, NA and 5-HT, which all may contribute to the decrease in locomotion. Deprenyl restored the locomotion of MPTP treated fish at 5 dpf, and prevented the decrease in the amine levels significantly. We have noticed that deprenyl alone does not affect DA or NA levels but it increases the levels of 5-HT and the increased 5-HT levels decrease spontaneous locomotion (Sallinen et al., in preparation), indicating that deprenyl does not increase motility per se. Attenuation of the MPTP-induced motility deficit and restoration of amine levels by deprenyl could be due to inhibition of the MPTP conversion to MPP+. This is supported by the fact that MPP+ produced similar behavioral deficits as MPTP and also decreased the levels of catecholamines. Similar transient decline in the locomotor activity is seen in adult zebrafish after MPTP or 6-hydroxydopamine treatment (Anichtchik et al. 2004). Also, in adult zebrafish NA levels were decreased along with DA levels (Anichtchik et al. 2004), a finding similar to our findings in larval fish. An earlier report showed non-quantified slowness in tail touch response in 3-day-old zebrafish after exposure to MPTP (Lam et al. 2005). However, at 3 dpf the locomotor system is not mature yet, and 3-day-old zebrafish do not swim spontaneously. Thus, comparison of these results to our study is difficult. McKinley et al. (2005) reported that 4-day-old larvae were immobile and unresponsive to touch at 4 dpf. We have observed similar behavior, which could be caused by acute non-specific toxicity of MPTP. Hence, we analyzed the behavior earliest at 1 day after the treatment, i.e., at 5 dpf. Another report analyzed ten 7-day-old zebrafish larvae swimming in the same tank, and reported quantified decrease in the locomotion after 1–5 dpf treatment with 9 mg/L (42.9 μM) or 45 mg/L (214.6 μM) MPTP (Bretaud et al. 2004). We did not notice any decrease in the locomotion after treatment with 100 μM MPTP. It is also noteworthy that this dose did not affect catecholamine levels either. These differences in the dose–response to MPTP might be due to the differences in the MPTP lot and/or in the zebrafish strain used as it is the case with mice. None of the earlier studies report effects of deprenyl on the behavior of zebrafish larvae. The behavioral response may arise from the diencephalic dopaminergic population 5,6,11, since it is affected by both MPTP and MPP+.

Mechanism and targets of MPTP action

The zebrafish cell population homologous to mammalian substantia nigra is unclear at the moment. Immunohistochemical and tracer studies in adult zebrafish suggest that the strongest candidade for zebrafish substantia nigra is the dopaminergic cell group of the posterior tuberculum (larval population 5,6,11) (Kaslin and Panula 2001; Rink and Wullimann 2001).

In larval zebrafish, MPTP has been reported to cause cell death of the pretectal population at 5 dpf (McKinley et al. 2005), putative posterior tuberal nucleus at 2 dpf (Lam et al. 2005) or diencephalic dopaminergic neurons at 5 dpf (Bretaud et al. 2004). All these studies lack the description of the exact populations affected and diencephalic populations are considered as a whole, which renders it difficult to compare them with our results. Loss of cells is quantified by counting only in Bretaud et al. 2004;. Lam et al. (2005) describe cell loss at 2–3 dpf. Analysis at this early age presents issues that need consideration. Firstly, several dopaminergic cell populations have not developed yet, among them the putative group corresponding to mammalian mesencephalic neurons, population 5,6,11. Secondly, MAO activity is very weak before 2 dpf (Sallinen et al., in preparation), suggesting that conversion of MPTP to toxic MPP+ is modest. A recent study, which was published after the completion of this study, reported moderate, yet non-quantified, loss of cells in the VMAT2-positive neural clusters (Wen et al. 2008). A detailed comparison of the findings of previous reports is shown in Supporting information Table S3.

In line with previous reports, we detected TH-ir cell loss in the pretectal population (7) and hypothalamic population (13) following MPTP. Additionally, we also detected TH-ir cell loss in the diencephalic (5,6,11) and preoptic (3–4) populations, which were not described in earlier reports. Pretectal (7) and diencephalic (5,6,11) populations were the most severely affected populations. Concerning 5-HT populations we found [contrary to the findings of Lam et al. (2005) and Wen et al. (2008)] that MPTP affected only PVOi population, while MPP+ did not affect the 5-HT populations significantly. We also showed decrease in the DA, NA and 5-HT levels in the larval zebrafish after MPTP exposure. In line with the immunohistochemical results, MPP+ did not affect 5-HT levels significantly, while it decreased DA and NA levels. Despite several approaches we failed to show actual cell death in the brain. Furthermore, the loss of TH-ir recovered rapidly and completely. The quick recovery might be the reason why no cell loss has been detected in adult zebrafish after MPTP exposure, although a decline in the DA and NA content has been reported (Anichtchik et al. 2004; Bretaud et al. 2004). These results suggest that MPTP is incapable of producing dopaminergic cell death in zebrafish, but causes a transient loss-of-function of the cells.

As MPP+ in mammals gains intracellular access via DAT, the expression of DAT might affect the susceptibility of different TH-ir populations to MPTP or MPP+. TH-ir cells in the preoptic do not express putative DAT, whereas pretectal (7) and hypothalamic (13) TH-ir populations express it (Holzschuh et al. 2001). The data regarding DAT expression in the region of population 5,6,11 is vague. DAT positive, but TH negative cells are reported at this site along with TH positive but DAT negative cells (the region is termed as PT in Holzschuh et al. 2001). It could be that the TH signal is below detection level in DAT-positive TH-negative cells and these cells corresponds to TH-ir cells in population 5,6,11. It is noteworthy that even though putative zebrafish DAT protein has high similarity to human DAT, the functional properties, such as substrate specificity, are unknown. Further studies are needed to confirm functional homology.

In our study, deprenyl prevented the MPTP toxicity partly, raising a question whether MPTP-toxicity is MAO-dependent in larval zebrafish. MPTP is a relatively weak substrate for zebrafish MAO in contrast to mammalian MAO B (Anichtchik et al. 2006), suggesting that MPTP oxidation to MPP+ may not be very efficient. On the other hand, deprenyl inhibits developing zebrafish MAO only partially (Sallinen et al., in preparation), which might result in incomplete rescue. So far, one form of MAO has been found in zebrafish genome (Setini et al. 2005; Anichtchik et al. 2006), but there may be other enzymes present in zebrafish capable of converting MPTP into MPP+. Interestingly, MAO activity and mRNA expression were detected in close proximity to population 5,6,11, but not pretectal population 7. Also MAO activity was detected in the rostral cells of the diencephalic population 13, but not in the preoptic population 3–4 (this study and Sallinen et al., in preparation). This suggests that loss of TH-ir cells in some of the dopaminergic populations might be MAO-independent. The close proximity of MAO activity to the serotonergic PVOi might be the reason for PVOi being the only serotonergic population affected by MPTP. One of the sites for conversion of MPTP to MPP+ in mammals is the histaminergic neurons in the hypothalamus (Nakamura and Vincent 1986). MAO was detected in the histaminergic neurons in the PVOp, and thus these cells might contribute to the conversion of MPTP to MPP+ also in larval zebrafish. To address the question whether the MPTP-toxicity is due to direct toxicity of MPTP itself or a MAO-dependent toxicity in which MPTP is converted to MPP+, we exposed the larvae to MPP+. In mammals MPP+ cannot cross the blood–brain barrier (BBB). Thus it is not neurotoxic if administered systemically (Perry et al. 1985). Available data regarding the properties and development of zebrafish BBB is scarce. Intraperitoneal injections of 75 mg/kg MPP+ do not induce loss of TH-ir or behavioral deficit in adult zebrafish (Bretaud et al. 2004), suggesting that BBB develops later to prevent the entry of MPP+ into CNS. In larval zebrafish, MPP+ caused similar locomotor deficit as MPTP, but loss of TH-ir was detected only in the population 5, 6, 11. These results suggest that the MPTP toxicity towards population 5,6,11 is MAO-dependent and due to the conversion to MPP+, whereas the MPTP-neurotoxicity affecting the pretectal population 7 may be more MAO-independent. On the other hand, permeability of the zebrafish BBB might have regional differences and thus MPP+ might only penetrate the BBB in certain areas. These findings also support the concept that the population 5,6,11 is most sensitive to MPTP/MPP+ and also in this respect corresponds to mammalian substantia nigra.

General conclusions

The development of catecholaminergic system in zebrafish is similar to that of other vertebrates and thus zebrafish might be a good model to study the disorders of the dopaminergic and noradrenergic systems. Identification and detailed cell counts and atlas of the catecholaminergic cell populations provided here enable the comparison of the larval populations to the adult counterparts. MPP+ may be a more useful tool than MPTP in larval zebrafish to model PD, since it seems more specific than MPTP. The results also suggest that the diencephalic population 5,6,11 is involved in the control of locomotion in larval zebrafish and is the target of both MPTP and MPP. Additionally, one 5-HT cell population was also affected by MPTP and MPTP decreased 5-HT levels, a finding similar to that of human PD patients (Kish et al. 2008).


We thank Henri Koivula, Jenny Bergqvist and Anna Harju for expert technical assistance. This study was funded by Technology Development Fund (TEKES), Finnish Parkinson Foundation, Academy of Finland and Sigrid Juselius Foundation.