Hypokinesia and Reduced Dopamine Levels in Zebrafish Lacking β- and γ1-Synucleins^*

Zebrafish

Experiments were carried out in accordance with National Institutes of Health guidelines and Institutional Animal Care and Use Committee approval. Adult strain AB zebrafish were maintained at 28.5 °C and euthanized by deep tricaine anesthesia followed by exposure to ice-cold water. Embryos were raised in E3 buffer (5 mm NaCl, 0.17 mm KCl, 0.33 mm CaCl₂, 0.33 mm MgSO₄), supplemented where necessary with 0.003% 1-phenyl-2-thiourea to inhibit pigmentation.

Synteny Analysis

Online tools (NCBI genome viewer and Cinteny (35), both available on the World Wide Web) were used to evaluate syntenic relationships. Proposed orthologues were verified by multiple sequence alignments, and synteny maps were drawn manually.

RT-PCR and 5′-RACE

First strand cDNA was generated by oligo(dT)-primed reverse transcription of total RNA from whole adult zebrafish brains or pooled embryo lysate (SuperScript III, Invitrogen). Nested 5′-RACE was carried out (RLM-RACE, Ambion, Austin, TX) as described previously (24, 36, 37). Primer sequences are listed in supplemental Table 1.

RNA Hybridization and Immunofluorescence

Northen blots (36), whole mount RNA in situ hybridization (24), RNA in situ hybridization to brain cryosections (24), double label ISH/IHC (38), and indirect immunofluorescence and confocal microscopy (24, 39) were carried out as reported in our previous work. Rabbit anti-α-synuclein (1:300; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) was used to detect human α-synuclein in tissue sections.

Morpholino Oligonucleotides

Antisense morpholino oligonucleotides (MOs; Gene Tools LLC, Philomath, OR) were designed to inhibit zebrafish synuclein gene expression by targeting splice signals in the primary transcripts. Expression of sncb was disrupted by MOs targeted against splice donor sites of exons 2 and 3 (sncb E2/I2, aatgttgcatgacttacCCACAAAC; sncb E3/I3, ttgcagaatcaccttacCTGTTGCA; sequence complementary to the 5′-end of each exon is shown in capital letters). Expression of sncg1 was disrupted by MOs targeted against the splice donor sites of exons 1 and 3 (sncg1 E1/I1, tatacatcagcacttacCCACATAC; sncg1 E3/I3, ctgattgctattcactcacCGGGT). Control MOs (random MO (N₂₅) and non-targeting MO CCTCTTACCTCAGTTACAATTTATA) were used to exclude nonspecific effects arising from morpholino exposure. 3 nl of MO dissolved in 1× Danieau Buffer (58 mm NaCl, 0.7 mm KCl, 0.4 mm MgSO₄, 0.6 mm Ca(NO₃)₂, 5.0 mm HEPES, pH 7.6) was injected into the yolk sac at the 1–2-cell stage. For sncb, 3 ng of sncb E3/I3 MO and 5 ng of sncb E2/I2 MO were injected (β MO). For sncg1, 4 ng of each sncg1 E1/I1 and sncg1 E3/I3 were injected (γ1 MO). To target both sncb and sncg1, all four MOs were injected (β + γ1 MO), total dose 14 ng. Control MO was injected at 14 ng/embryo.

mRNA Rescue

A NotI/XhoI fragment of pIRES2-EGFP (Clontech, Mountain View, CA) was inserted into the XbaI/XhoI sites of pCS2⁺ (40) to generate plasmid pCS-IG. Human α-synuclein cDNA was amplified from NT2 cells by RT-PCR, sequenced, and inserted into the XhoI/XmaI sites of pCS-IG to generate pCS-syn-IG. pCS-IG and pCS-syn-IG were linearized at the NotI site 3′ to the poly(A) signal, and mRNA was transcribed from the Sp6 promoter (mMessage Machine, Ambion, Austin, TX). Intracellular injection of 1–1.25 nl of mRNA solution (250 pg/nl mRNA, 0.25% phenol red, 120 mm KCl, 20 mm HEPES, pH 7.4) was carried out at the single cell stage, followed by MO injection into the yolk sac as above.

Western Blot

Western blots were carried out as described previously (36, 39). Proteins were transferred to PVDF membrane (Immobilon, Millipore, Billerica, MA), and blots were blocked in Odyssey Blocking Buffer (LI-COR, Lincoln, NE). Primary antibodies were as follows: α-synuclein (mouse; 1:1000; BD Biosciences) and β-actin (rabbit; 1:2000; Sigma). Secondary antibodies conjugated to different infrared fluorophores (anti-mouse IRdye®800 and anti-rabbit IRdye®680; 1:10,000 each; LI-COR) allowed simultaneous detection of both proteins using a two-channel infrared imaging system (LI-COR).

Motor Analysis

Larval movement was analyzed using a proprietary system (Zebralab; ViewPoint Life Sciences, Montreal, Canada) as reported previously (34). Zebralab software was used in “quantization mode” with detection sensitivity set to 16 and with burst and freezing activity thresholds set to 10 and 2, respectively. Groups of 15 embryos were analyzed in 12-well plates containing 3 ml of E3 buffer/well. Spontaneous activity was measured for an 8-h period daily, from 24 h postfertilization (hpf) to 7 days postfertilization (dpf).

Neurochemical Measurements

Neurochemical measurements were performed by adapting a previously reported method (41). 30 larvae were sonicated in 150 μl of 4 m perchloric acid, 0.1% Na₂S₂O₅, 0.1% EDTA at 4 °C. The sample was centrifuged at 17,000 × g at 4 °C, and the supernatant was filtered through a 0.22-μm nylon membrane (Spin-X, Corning Glass). 25 μl was loaded into a Waters 2695 HPLC separation module (Waters, Milford, MA). The HPLC mobile phase consisted of 0.06 m sodium phosphate monobasic, 0.03 m citric acid, 8% methanol, 1.1 mm 1-octanesulfonic acid, 0.1 mm EDTA, 2 mm sodium chloride, pH 3.5. The flow rate was 1.0 ml/min. Neurotransmitters were separated on a Waters XBridge C18 4.6 × 150-mm column, particle size 3.5 μm, at 34 °C, and detected using a Waters 2465 electrochemical detector with a glassy carbon electrode set at 750 mV, referenced to an ISAAC electrode. Neurotransmitters were quantified as pg/larva, using a standard curve generated from injection of high purity standards. Each data point was then normalized to the value for wild type animals in the same experiment.

Statistical Analysis

Quantitative data were shown to be normally distributed by Pearson's and D'Agostino's tests and were analyzed using parametric statistical tests. All experiments were repeated at least in triplicate. Graphs show means ± S.E. One-way ANOVA was carried out after confirming the assumption of equal variance using Bartlett's test and was followed by Dunnett's post hoc test to evaluate statistically significant differences between test groups.

Zebrafish Synucleins

Identification of zebrafish synucleins was recently reported by two other groups (42, 43). We independently cloned the same genes at the start of this study (not shown). sncb encodes the zebrafish orthologue of human β-synuclein, and sncg1 and sncg2 encode dual paralogues of human γ-synuclein. Zebrafish synucleins show extensive sequence similarity to human synucleins, except that the C termini of the two zebrafish γ-synucleins are relatively divergent (Fig. 1). We did not identify an orthologue of human α-synuclein in zebrafish genomic, mRNA, or expressed sequence tag databases, which was unexpected because α-synuclein orthologues have been identified in mammals, birds, reptiles, amphibians, and other fish species. Despite employing a variety of RT-PCR and RACE strategies to amplify a putative snca transcript (supplemental Table 4), we did not find evidence of a zebrafish α-synuclein orthologue expressed in the CNS.

In order to further understand how human and zebrafish synuclein genes are related, we constructed synteny maps (Fig. 2). Two of the genes flanking the sncb gene on zebrafish chromosome 13 are orthologues of genes flanking the human SNCB gene on chromosome 5 (Fig. 2A), suggesting that the genes are derived from a common ancestral locus. A large genomic region surrounding zebrafish sncg1 and sncg2 is duplicated and is syntenic with the human genomic region containing SNCG (Fig. 2B). These observations suggest that sncg1 and sncg2 are derived from duplication of an ancestral locus shared with human SNCG. Zebrafish orthologues of genes immediately flanking human SNCA were not found in the available zebrafish genome sequence, whereas the wider genomic region in the human showed two blocks of genes that were syntenic to different regions of the zebrafish genome (Fig. 2C). It is possible that the region between these syntenic blocks was deleted from an ancestral chromosome during evolution, accounting for the lack of zebrafish genomic or mRNA sequences with homology to human SNCA.

mRNA Transcripts and Genomic Organization of Zebrafish Synucleins

Zebrafish sncb and sncg1 transcripts were detected abundantly in adult brain by Northern blot (Fig. 3A). Each gene gave rise to a single mRNA species of 1.45 kb (sncb) or 2.7 kb (sncg1). sncg2 was alternatively spliced to yield two transcripts encoding isoforms with divergent C termini; although both transcript variants were amplified from CNS by PCR (Fig. 3B), we were unable to identify sncg2 transcripts in the CNS by Northern blot hybridization or RNA in situ hybridization and conclude that they are present at low abundance. The remainder of this study focused on sncb and sncg1, encoding the predominant CNS synucleins in zebrafish.

5′-RACE (Fig. 3C), using the tobacco acid pyrophosphatase (TAP) method (as described previously (24, 36, 37)), was employed to map the 5′-ends of the sncb and sncg1 transcripts. TAP⁻ and RT-PCR controls verified that the boundary between the RACE adapter and the cDNA sequence in the RACE products occurred at the mRNA cap site. Both sncb and sncg1 promoter regions lack canonical TATA motifs and show multiple transcriptional start sites from motifs with homology to the initiator consensus 5′-YYA⁺¹NWYY-3′ (44) (Fig. 3, D and E). We next examined the genomic organization of sncb, sncg1, and sncg2 by comparing cDNA sequences with the zebrafish genomic sequence (Fig. 3, F–H, and supplemental Tables 2 and 3). Like its human orthologue, sncb contains six exons, the first of which is non-coding. Similar to their human paralogue, both sncg1 and sncg2 contain five exons. The positions and phases of the splice boundaries in sncb, sncg1, and sncg2 are conserved with respect to their human paralogues, providing further evidence of a common origin (supplemental material).

Expression Patterns of Zebrafish Synucleins

The expression patterns of zebrafish synucleins were determined by RNA in situ hybridization. As reported previously during early development (42, 45), sncb and sncg1 were expressed in the CNS, whereas sncg2 was found in the notochord (Fig. 4, A–C). In the adult zebrafish CNS, both sncb and sncg1 were expressed in cells with neuronal morphology, widely distributed throughout gray matter of the brain and spinal cord. However, their topographical expression patterns differed significantly; sncb was more robustly expressed rostrally, whereas sncg1 expression was more prominent caudally (Fig. 4, D and E). Differences in the expression patterns of sncb and sncg1 in specific brain regions are illustrated in Fig. 4F. Whereas prominent sncb expression was detected in the olfactory bulb and dorsal telencephalon, sncg1 was not expressed at high levels in these regions. Conversely, robust sncg1 expression in the medulla oblongata and habenula was accompanied by weak expression of sncb in these areas. Even within some brain regions where there was overlap between sncb and sncg1 expression, the two genes were expressed in different cellular populations. For example, within the cerebellum, sncb was expressed within granule cells, whereas sncg1 was expressed most prominently in the Purkinje cell layer (Fig. 4G).

We next examined expression of sncb and sncg1 in catecholaminergic neurons in the zebrafish CNS. In the absence of available antibodies to zebrafish β- or γ1-synucleins, we used a double label ISH/IHC technique, allowing us to detect sncb or sncg1 and tyrosine hydroxylase simultaneously in the same tissue sections. Examples are shown in Fig. 4H; dopaminergic neurons of the periventricular hypothalamus were identified by their morphology and by robust tyrosine hydroxylase immunoreactivity in their processes. Strong expression of sncb (Fig. 4H, left) and sncg1 (right) was also evident in the perinuclear cytoplasm. Using this method, we examined each of the major catecholaminergic cell groups within the olfactory bulb, telencephalon, diencephalon, and hindbrain. With the exception that sncg1 expression was relatively weak in the olfactory bulb, all tyrosine hydroxylase-immunoreactive groups also expressed both sncb and sncg1. These data suggest that sncb and sncg1 may be co-expressed in catecholaminergic neurons.

β- and γ1-Synucleins Are Not Required for Early Embryogenesis or CNS Morphogenesis

We next asked whether perturbing expression of sncb or sncg1 using antisense MOs would result in a neurodevelopmental phenotype that might be informative regarding the normal functions of β- and γ1-synucleins. Antibodies to zebrafish synucleins are not available, so verification of translation-inhibiting MOs was not possible; consequently, we inhibited gene expression by targeting pre-mRNA splice sites because splicing could be verified by RT-PCR. Pairs of MOs targeting two splice boundaries in each transcript greatly reduced expression, such that sncb or sncg1 mRNA could not be detected by RT-PCR up to 72 hpf (Fig. 5, A and B). We refer to the pair of MOs targeting sncb as β MO and the pair of MOs targeting sncg1 as γ1 MO. As a control for nonspecific effects of MOs, we used both non-targeting and random MOs (termed Ctrl MO in subsequent figures) at the same concentrations. These did not affect expression of either sncb or sncg1 (Fig. 5, A and B).

Zebrafish lacking β-, γ1-, or both synucleins were viable and showed normal development of the basic body plan, internal organs, and circulation (Fig. 5C). In order to examine the morphology of the nervous system in more detail, we employed Tg(eno2:egfp) zebrafish, which express GFP in a pan-neuronal pattern (24), allowing delineation of CNS structure and examination of spinal morphology and nerve roots. Compared with controls, we did not detect any abnormalities of CNS or PNS structure in Tg(eno2:egfp) larvae lacking β-, γ1-, or both synucleins. In particular, the CNS appeared structurally unremarkable at 4 dpf, the time point at which the most robust neurobehavioral abnormalities were found (see below). These data show that β- and γ1-synucleins are not necessary for development of normal CNS morphology.

β- and γ1-Synucleins Are Required for Normal Development of Motor Function

To evaluate the role of β- and γ1-synycleins in neuronal function, we employed an assay of spontaneous larval movement. A live video stream showing zebrafish larvae housed in multiwell plates was analyzed for movement by quantifying the changes in pixel gray scale values that accompany larval displacement from frame to frame. In order to detect changes in the frequency with which spontaneous movements were executed, we measured the proportion of video frames in which pixel gray scale changes exceeded the threshold for detection of movement above stochastic pixel noise. To confirm that this method could quantify spontaneous movement under the conditions used for our study, we housed 0, 2, 4, 8, or 16 zebrafish in each well of a 12-well plate and quantified motor activity at 2, 3, and 7 dpf (Fig. 6A). No movement was detected in empty wells, and the relationship between the number of larvae in a well and measured motor activity was linear.

We next analyzed movement of groups of 15 larvae from each experimental group (WT, buffer-injected, Ctrl MO, β MO, γ1 MO, or β + γ1 MO). Six independent experimental replicates were carried out at each time point between 2 and 7 dpf (Fig. 6B). In uninjected larvae, little movement was detected at early developmental points; however, motor activity increased sharply at 4 dpf and remained at approximately the same level until 7 dpf. At each time point, there were no statistically significant differences between uninjected larvae, larvae injected with buffer only, and larvae injected with control MOs at the highest doses used in gene targeting experiments. Thus, neither the mechanical consequences of microinjection nor nonspecific toxic effects of exposure to MOs induced changes in spontaneous movement in this assay. At 2 dpf, small statistically significant reductions in activity were apparent between controls and larvae lacking β-, γ1-, or both synucleins, but these changes resolved by 3 dpf. Between 3 and 4 dpf, larvae lacking β- or γ1-synucleins showed only a modest increase in activity, compared with the large increase seen in controls. Consequently, at 4 dpf, larvae lacking β- or γ1- synucleins showed a robust reduction in spontaneous movement compared with controls. Loss of motor activity was more severe in larvae lacking both β- and γ1-synucleins than in larvae lacking one or other synuclein. As gene expression recovered later in development, the hypokinetic phenotype started to resolve. By 5 dpf, spontaneous movement in synuclein knockdown animals had almost returned to control levels, and only animals injected with β + γ1 MO showed statistically significant differences from controls. By 7 dpf, the groups were statistically indistinguishable, although measured motor activity remained lower in larvae exposed to β + γ1 MO than in the other experimental groups.

In order to firmly establish whether these motor abnormalities were attributable to loss of synucleins, we next asked whether this hypokinetic phenotype could be rescued by an exogenous synuclein. Human α-synuclein was expressed in zebrafish embryos by microinjecting a bicistronic mRNA encoding both α-synuclein and GFP. mRNA encoding GFP alone was used as a control. Because the IRES element functions weakly in zebrafish (not shown), human α-synuclein expression was verified by other methods. Western blot showed a single 16-kDa α-synuclein immunoreactive band specific to microinjected embryos; immunofluorescence demonstrated robust α-synuclein expression throughout microinjected embryos, including the neural tube and developing retina (Fig. 6C). Expression of either GFP or human α-synuclein did not alter WT larval motor activity (Fig. 6D). Reduced motor activity at 4 dpf was seen in GFP-expressing larvae lacking β-, γ1-, or β- and γ1-synucleins, similar to results in larvae expressing no exogenous mRNA. In contrast, animals expressing human α-synuclein showed normal levels of motor activity in the absence of β- or γ1-synuclein. In the absence of both β- and γ1-synucleins, it was necessary to express a higher level of human α-synuclein in order to prevent loss of spontaneous movement. These data show that the hypokinetic abnormality observed after loss of β-, γ1-, or β- and γ1-synucleins is attributable to loss of synucleins rather than an artifact of the experimental system. Together, these findings demonstrate that β- and γ1-synuclein expression in neurons is essential for physiological functions that underlie normal motor behavior during development.

β- and γ1-Synucleins Are Required for Development of the Dopamine System

We next addressed the hypothesis that the motor phenotype resulting from loss of β- or γ1-synuclein was attributable to abnormalities of the dopamine system. We first asked whether development of dopaminergic neurons differed between controls and zebrafish lacking β-, γ1-, or both synucleins. RNA in situ hybridization was carried out at 48, 72, and 96 hpf, using a cRNA probe to slc6a3 (dopamine transporter, dat), which is specifically expressed in dopamine neurons (38, 46) (Fig. 7A). No qualitative differences were observed in the distribution or staining intensity of slc6a3-expressing cells between controls and larvae lacking β-, γ1-, or both synucleins. However, quantification by an observer blinded to MO exposure showed a modest reduction in the number of slc6a3⁺ cells in larvae lacking both β- and γ1-synucleins at 48 hpf (Fig. 7B). This abnormality had recovered by 72 hpf, such that there was no difference in the number of slc6a3⁺ cells between WT larvae and those exposed to β + γ1 MO (Fig. 7C). Expression of GFP or human α-synuclein did not alter the number of slc6a3⁺ neurons in control larvae (Fig. 7D). However, expression of human α-synuclein was sufficient to prevent a reduction in the number of slc6a3⁺ cells in the absence of β- and γ1-synucleins, whereas GFP expression did not rescue this phenotype. These data show that both β- and γ1-synuclein are necessary for early development or differentiation of dopamine neurons. However, there was a normal complement of dopamine neurons by 4 dpf, when the hypokinetic phenotype in zebrafish lacking synucleins was most prominent.

Finally, we asked whether abnormalities of monoaminergic neurochemistry might explain the hypokinesia resulting from loss of β- or γ1-synucleins. We examined steady-state whole larval dopamine levels at 7 dpf, which was the earliest time point at which we could measure neurotransmitters with sufficiently low variability to enable detection of differences between multiple experimental groups. Although gene expression had recovered by this time point, zebrafish lacking both β- and γ1-synucleins earlier in development showed significantly reduced dopamine levels at 7 dpf compared with Ctrl MO-injected animals (Fig. 7E). None of the other experimental groups showed significant differences from controls. Transient expression of GFP or human α-synuclein early in development did not change steady state levels of dopamine in WT larvae at 7 dpf. However, α-synuclein but not GFP prevented the loss of dopamine that occurred in the absence of zebrafish β- and γ1-synucleins (Fig. 7F). These data show that early developmental expression of both β- and γ1-synucleins is necessary for establishment of normal dopamine levels. Interestingly, 5-hydroxytryptamine levels were not reduced in larvae lacking synucleins (Fig. 7G), suggesting that the relevant function of synucleins in this context may be specific to dopaminergic neurons.