Abstract
Parkinson's disease (PD) is the second most prevalent neurodegenerative disorder in the Western world. PTEN (phosphatase/tensin homolog on chromosome 10)-induced putative kinase 1 (PINK1), a putative kinase that is mutated in autosomal recessive forms of PD, is also implicated in sporadic cases of the disease. Although the mutations appear to result in a loss of function, the roles of this protein and the pathways involved in PINK1 PD are poorly understood. Here, we generated a vertebrate model of PINK1 insufficiency using morpholino oligonucleotide knockdown in zebrafish (Danio rerio). PINK1 knockdown results in a severe developmental phenotype that is rescued by wild-type human PINK1 mRNA. Morphants display a moderate decrease in the numbers of central dopaminergic neurons and alterations of mitochondrial function, including increases in caspase-3 activity and reactive oxygen species (ROS) levels. When the morphants were exposed to several drugs with antioxidant properties, ROS levels were normalized and the associated phenotype improved. In addition, GSK3β-related mechanisms can account for some of the effects of PINK1 knockdown, as morphant fish show elevated GSK3β activity and their phenotype is partially abrogated by GSK3β inhibitors, such as LiCl and SB216763 [3-(2,4-dichlorophenyl)-4-(1-methyl-1H-indol-3-yl)1H-pyrrole-2,5-dione]. This provides new insights into the biology of PINK1 and a possible therapeutic avenue for further investigation.
Introduction
Parkinson's disease (PD) is the second most prevalent neurodegenerative disorder in the Western world. To date six genes with pathogenic mutations have been identified as contributing to Mendelian forms of PD. These include two autosomal dominant genes, α-synuclein (SNCA) and leucine-rich repeat kinase 2 (LRRK2), and three autosomal recessive genes, parkin, DJ-1, and phosphatase/tensin homolog on chromosome 10 (PTEN)-induced putative kinase 1 (PINK1) (Valente et al., 2004; Klein and Schlossmacher, 2006). An apparently dominant mutation in a sixth gene, ubiquitin C-terminal hydrolase L1 (UCHL1), has only been found in one small family and its relevance to familial PD is still debated (Healy et al., 2006). PINK1 mutations are the second most common cause of autosomal recessive PD after parkin (Li et al., 2005), and PINK1 is also implicated in sporadic cases of the disease (Valente et al., 2004). PINK1 is a highly conserved 581 aa protein that is widely expressed in the human brain and localizes to mitochondrial membranes via an N-terminal targeting motif (Gandhi et al., 2006). It shares homology with calmodulin-dependent protein kinase 1, contains a catalytic serine–threonine kinase domain, and has autophosphorylation activity in vitro (Sim et al., 2006). PD-causing mutations appear to result in loss of function (Sim et al., 2006).
Recent studies of PINK1 in Drosophila melanogaster suggest its importance for mitochondrial function. Loss-of-function mutants of PINK1 exhibit male sterility, muscle and dopaminergic neuron degeneration, and increased sensitivity to stressors (Clark et al., 2006; Park et al., 2006; Yang et al., 2006). PINK1 mutant flies have decreased expression of parkin and have phenotypes similar to flies with loss of parkin function. Because overexpression of parkin rescues phenotypes caused by PINK1 deficiency, but not vice versa, it has been proposed that parkin acts downstream of PINK1, also suggesting that understanding PINK1 function may be relevant for PD caused by either parkin or PINK1 mutations. The molecular targets of PINK1 kinase activity are poorly studied. Recently, the phosphorylation of mitochondrial molecular chaperone TRAP1 was attributed to PINK1 (Pridgeon et al., 2007), further strengthening the mitochondrial relevance of PINK1.
Materials and Methods
Zebrafish strain husbandry and in situ hybridization.
Wild-type zebrafish of the AB strain were maintained at 28.5°C under standard conditions (Westerfield, 2000) in compliance with Animal Welfare legislation. Embryos were collected after natural spawning, staged as described previously (Kimmel et al., 1995), and raised in embryo medium (Westerfield, 2000). Whole-mount in situ hybridization was done as described previously (Jowett and Lettice, 1994). Antisense digoxigenin-labeled RNA probes were generated using cDNAs of pink1, parkin, reelin (Costagli et al., 2002), fezl (Levkowitz et al., 2003), and neurogenin-1 (ngn-1) (Blader et al., 1997). All husbandry and experimental procedures were performed in accordance with the Animals (Scientific Procedures) Act.
In silico identification of zebrafish PINK1 orthologue.
To identify the zebrafish PINK1 orthologue, a basic local alignment search tool (BLAST) search of the Ensembl zebrafish cDNA database (http://www.ensembl.org/Danio_rerio/index.html) was performed using the human PINK1 amino acid sequence (ENSG00000158828). Based on the obtained sequences, primers were designed to amplify the complete sequence of the zebrafish gene. mRNA, isolated from different age embryos, was reverse transcribed with the First-Strand Synthesis Kit (Invitrogen). PCR products were analyzed by agarose gel electrophoresis and sequenced.
Morpholino and mRNA injections and drug treatment.
Morpholino oligonucleotides (MOs) (ATG targeting, 5′-GCT GAG AAC ATG CTT TAC TGA CAT T-3′; 5′-untranslated region (UTR) targeting, 5′-ATA TTG ACT ATG AGA GGA AAT CTG A-3′; ATG-targeting five-mispair control, 5′-GCT CAC AAC ATC CTT TAG TGA GAT T-3′; 5′-UTR targeting five-mispair control, 5′-ATA TTC AGT ATC AGA GCA AAT GTG A-3′) targeting either the ATG or 5′-UTR region of the zebrafish PINK1 gene were obtained from Gene Tools and dissolved in 1× Danieau medium [containing (in mm) 58 NaCl, 0.7 KCl, 0.4 MgSO4, 0.6 Ca(NO3), 2.5 HEPES, pH 7.6]. One-cell stage embryos were injected with a combination of ATG and 5′-UTR morpholinos at 7, 9, and 13 ng each. Five-mispair control morpholinos were used to determine the concentrations of injected morpholino that produced phenotypes that were disctinct from morpholino toxicity. Additional controls to monitor the quality of the injection technique included injections of a standard control morpholino (5′-CCT CTT ACC TCA GTT ACA ATT TAT A-3′), which targets a splice site in the human β-globin gene (negative control), and a morpholino targeting the zebrafish chordin gene (5′-ATC CAC AGC AGC CCC TCC ATC ATC C-3′) (positive control) (all from Gene Tools).
Capped mRNAs were synthesized from cDNAs cloned into pcDNA/V5-DEST, encoding human full-length wild-type PINK1 gene using the T7 mMessage mMachine kit (Ambion). Sites corresponding to the clinically relevant mutations A168P and W347X were altered using the QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene). Sixty-five picograms of mRNA were coinjected together with PINK1 MOs into embryos at the one-cell stage.
For the pharmacological rescue experiments, LiCl (0.5–50 mm; Sigma-Aldrich), 3-(2,4-dichlorophenyl)-4-(1-methyl-1H-indol-3-yl)1H-pyrrole-2,5-dione (SB216763; 1–10 μm; Tocris Bioscience), reduced glutathione (GSH; 100 μm) and N-acetyl-cysteine (NAC; 100 μm) were dissolved in DMSO or water and were introduced into the embryonic media 6 h after morpholino injection The final concentration of DMSO was 0.2% in the water. Fish were analyzed 24 and 48 h after injection.
For quantification of the rescue results, data were gathered from 15–20 embryos at each stage and age. The normal phenotype was expressed as 100%, and the changes in the morphology were scored and calculated accordingly. Data from at least three experiments were pooled for statistical analyses.
We used five-mispair control MOs as the “control state” in the present study unless indicated otherwise. Generally, five-mispair control MO at the concentration used did not produce any phenotype different from naive fish.
Immunohistochemistry and PAGE.
Immunohistochemistry for acetylated tubulin (Sigma-Aldrich), tyrosine-hydroxylase, serotonin (5-HT) (Immunostar), and PINK1 (Cayman Chemicals) was performed as described previously (Kaslin and Panula, 2001). Images were taken using an Olympus SZX12 or BX51 fluorescence microscope equipped with digital camera, and processed using AnalySISD 5.0 software (Soft Imaging System). No adjustment of brightness or contrast was made after image acquisition. For Western blotting, embryos of 24–48 h postfertilization (hpf) were dechorionated, deyolked (Link et al., 2006), homogenated in radioimmunoprecipitation assay (RIPA) buffer with protease and phosphatase inhibitors, and processed for SDS-PAGE. Immunoreactive bands were detected with ECL reagent (GE Healthcare Life Sciences) and chemiluminescent signal was visualized by the exposing membrane to ECL Hyperfilm (GE Healthcare Life Sciences). Films were scanned using a desktop scanner HP ScanJet 3800, and densitometric analysis of blots was performed using ImageJ software (National Institutes of Health). The background intensity of the film was subtracted from the band intensity. At least three separate experiments were analyzed, and band intensities were normalized to the loading control band intensity. Quantification of tyrosine hydroxylase (TH)-positive neurons was done in 24 and 48 hpf whole-mount, deyolked, flat-mounted embryos by counting all fluorescent cell bodies in the brain using a 40× objective. The number of neurons in control five-mispair MO-injected embryos was presented as 100%, and the changes in the PINK1 MO group was calculated accordingly. Each experiment analyzed at least seven embryos in each group, and data from at least four experiments were pooled for the statistical analyses.
Detection of cell death.
Apoptotic/necrotic cells were detected by incubation of live 24- to 48-hpf-old embryos in acridine orange solution (5 μg/ml) (Furutani-Seiki et al., 1996). Cells loaded with the dye were visualized using a tetramethylrhodamine isothiocyanate filter on the Olympus SZX12 fluorescent microscope. To detect the activity of caspase-3 in zebrafish embryos, fish were homogenated in ice-cold RIPA buffer and the homogenate was mixed with an equal volume of 0.39 mm acetyl Asp-Glu-Val-Asp 7-amido-4-methylcoumarin (Sigma-Aldrich), a selective caspase-3 substrate, resulting in the release of the fluorescent 7-amino-4-methylcoumarin (AMC). The excitation and emission wavelengths of AMC (360 and 460 nm, respectively) were detected using a Victor3 plate reader (PerkinElmer Life And Analytical Sciences). The reactions were normalized with total protein concentrations. Embryos treated with UV light (400 μJ/cm2, 0.1 s) for 6 h before the assay served as a positive control of apoptosis-induced caspase-3 activation. Incubation with 100 μm pan-caspase-3 inhibitor benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone (zVAD; R&D Systems) was used as a negative control.
Reactive oxygen species (ROS) accumulation, assessed by the levels of the oxidized form of the cell-permeant ROS indicator acetyl ester of 5-(and 6-)chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA) (Invitrogen), was detected in live fish as follows. Injected embryos were dechorionated, several groups of three embryos were put in the wells of a 96-well plate, and excess medium was removed using a fire-polished glass Pasteur pipette. Embryos were incubated in 1 μg/ml CM-H2DCFDA for up to 2 h, and the fluorescence was measured with a Victor3 plate reader (PerkinElmer Life And Analytical Sciences) (excitation 485 nm, emission 560 nm) after every 10 min. The spontaneous oxidation of CM-H2DCFDA was monitored in the wells without fish, and the levels were low and did not change with time. H2O2 solution served as a positive control.
Isolation of mitochondria.
Crude mitochondrial and cytoplasmic fractions were isolated using a Mitoiso 1 isolation kit (Sigma-Aldrich). Briefly, embryos 80–100 24 hpf were collected, dechorionated, and homogenated in the supplied extraction buffer (10 mm HEPES buffer, pH 7.5, containing 200 mm mannitol, 70 mm sucrose, 1 mm EGTA, and 2 mg/ml BSA). Homogenates were spun at 600 × g in 4°C, and supernatants were then spun at 11,000 × g for 10 min in 4°C. Supernatants, representing cytoplasmic fractions, were removed and stored in ice. Pellets were resuspended in extraction buffer, and the centrifugations were repeated. The resulting pellets were resuspended in storage buffer [10 mm HEPES buffer, pH 7.5, containing (in mm) 250 sucrose, 1 ATP, 0.08 ADP, 5 sodium succinate, 2 K2HPO4, and 1 DTT], and the protein concentration was assayed using the Bradford method. 5,5′,6,6′-Tetracloro-1,1′,3,3′-tetraethylbenzimidazolcarbocyanine iodide (JC-1) uptake, an indicator of mitochondrial Δψ membrane potential, was then assessed by incubation adjusted by protein levels. The preparations were incubated for 10 min in the dark, and fluorescence was detected using a Victor3 plate reader (PerkinElmer Life and Analytical Sciences) (excitation, 490 nm; emission, 525 and 590 nm). The ratio of 590/530 nm (orange/green) was considered as an indicator of mitochondrial Δψ membrane potential. Incubation with 1 μg/ml valinomycin resulted in decrease of JC-1 orange/green fluorescence, and was used as a positive control. The purity of the resulting preparation was assessed, using actin as a control for cytoplasm, and porin as a mitochondrial loading control (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
Statistics.
Pooled estimates for the changes in tail phenotype, resulting from perturbations assessed in multiple experiments, were calculated as odds ratios with 95% confidence intervals. We used this method frequently in the past to allow analysis of data from multiple independent experiments (Wyttenbach et al., 2001, 2002; Carmichael et al., 2002). Odds ratios and p values were determined by unconditional logistical regression analysis, using the general log-linear analysis option of SPSS 9 software. In other experiments, two-way ANOVA and t test (GraphPad Prism 4) were used when applicable.
Results
Identification of zebrafish PINK1
Here, we inhibited translation of PINK1 in zebrafish with MOs to examine its roles in vertebrate development. To identify the zebrafish gene, we performed BLAST searches with the human PINK1 protein sequence (NP_115785) in the Ensembl zebrafish database and identified putative zebrafish PINK1 transcripts, which were confirmed by PCR cloning. Zebrafish PINK1 (NM_001008628) encodes a 574 aa protein with a predicted molecular mass of 64 kDa and is 54% identical to the human PINK1 protein (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). Analysis of the zebrafish primary sequence revealed a putative mitochondrial targeting domain at the N terminus of the protein with a possible cleavage site at amino acid 88 (MitoProt) (Claros and Vincens, 1996) and a putative Ser-Thr protein kinase active site at amino acids 346–358, which is highly similar to human PINK1 (supplemental Fig. 2, available at www.jneurosci.org as supplemental material).
Using RT-PCR, zebrafish PINK1 mRNA was detected as early as 2.5 hpf and throughout development up to 144 hpf (Fig. 1A). In situ hybridization detected PINK1 mRNA that was distributed fairly uniformly in zebrafish larvae at 24 and 48 hpf, with increased intensity in the nervous system and muscle (Fig. 1B). Polyclonal antibodies raised against human PINK1 peptide detected signals within the paraventricular regions of the adult zebrafish brain, including the hypothalamus and the deep layers of the optic tectum (Fig. 1C). In some areas (e.g., the hypothalamus), we observed overlapping signals for PINK1 and TH, suggesting the presence of PINK1 in some dopaminergic neurons. We detected two PINK1-positive bands in the whole zebrafish lysates: a smaller one, ∼33 kDa, and a larger one, ∼66 kDa. Injection of PINK1 MOs led to a dose-dependent decrease of the intensity of the PINK1 bands from 24 hpf fish (Fig. 1D). We observed these bands in both cytoplasmic and mitochondrial fractions of whole 24 hpf embryos homogenates (supplemental Fig. 3, available at www.jneurosci.org as supplemental material). PINK1 bands approximating the size of the smaller fragment we observed in zebrafish have been reported recently in human cell lines (Lin and Kan, 2008).
Characterization of zebrafish PINK1. A, RT-PCR analysis of PINK1 mRNA expression at different hours after fertilization. B, In situ hybridization showing expression of zebrafish PINK1 mRNA in 24 and 48 hpf embryos. Right, Negative control staining with sense probe. C, Immunohistochemical analysis of PINK1 protein (green) and TH (red) in adult zebrafish brain (i–iii). PINK1 protein is localized ubiquitously with the tendency to be enriched along ventricular surfaces. In some areas (hypothalamus), PINK1 protein is colocalized with tyrosine hydroxylase. i, Posterior hypothalamus; ii, optic tectum; iii, areas around diencephalic ventricle; iv, posterior tuberculum (PINK1 protein, red; TH, green). D, The PINK1 antibody used (Cayman Chemicals, catalog #10006283) detects bands of ∼66 and ∼33 kDa in mouse liver mitochondria preparations. Similarly sized bands are seen in the wild-type zebrafish, and these disappear in PINK1 MO-treated fish. The trend in this gel is representative of at least five replicate experiments. E, PINK1 morphants at 24 hpf after injection of 7 ng of MOs show long-tail, short-tail, and no-tail phenotypes. Five-mismatch MO control zebrafish look the same as uninjected siblings. F, Phenotype of PINK1 morphants at 48 hpf after injection of 7 ng of MO. Enlarged brain ventricle, slim tail-yolk extension, and curved spines are indicated by arrows in mildly (middle) and severely (right) affected morphants. Scale bars: B, E, F, 100 μm; C, 50 μm.
Phenotype of PINK1 knockdown
To examine morphological changes caused by PINK1 deficiency, embryos were injected with MOs directed against the 5′-UTR and the ATG start codon of zebrafish PINK1 (7 ng of each ATG and UTR MO in combination). Using otherwise identical MOs with 5 bp substitutions (control MO), no morphological abnormalities were observed up to 6 d postfertilization (dpf). MOs that block zebrafish PINK1 translation did not affect gastrulation but resulted in three distinct phenotypic groups at 24 hpf (Fig. 1E). The first group had dark, nontransparent heads but properly developed trunks/tails [“long-tail” group, 13.8 ± 2.71% (SD) of embryos]; the second group had dark, nontransparent heads, small eyes, and short, curved tails with misshapen somites (“short-tail” group, 79 ± 3.5% of embryos); and the third group appeared to have halted head/tail development (“no-tail” group, 6.6 ± 2.8% of embryos). Injection of higher doses of MO (>10 ng) resulted in an increase in the size of no-tail group and disappearance of the long-tail group. At 48 hpf, the predominant phenotype (with 7 ng of MO) was in the trunk/tail: MO-injected fish had various degrees of ventral curvature of the spine with misshapen tail ends (Fig. 1F). In addition, they had enlarged brain ventricles, small heads, and slim yolk extensions and occasionally manifested cardiac edema. These embryos failed to swim away when touched on the tail (impaired escape response). By 72 hpf, the most severely affected larvae died, whereas surviving larvae showed some recovery. Note that the effects of MOs decrease with time after injection, with most activity being lost by ∼5 d. In general, MO-induced translational inhibition is most efficient during the first 2–3 d of development (Sumanas and Larson, 2002).
Injection of PINK1 MO resulted in structural alterations in the axonal scaffold of the larval nervous system: commissures were less prominent, sometimes the tegmental commissure failed to develop and the paramedial descending axonal tract in the hindbrain and spinal cord were disorganized (Fig. 2A,B). The number of TH-positive neurons in the diencephalon of MO-injected zebrafish at 48 hpf was reduced to 62.7 ± 6.05% SEM of control levels 100% (p = 0.012; n = 3), whereas serotoninergic (5-HT) neurons were not affected (Fig. 2C,D, supplemental Fig. 4, available at www.jneurosci.org as supplemental material). Several in situ hybridization probes were used to examine neuron development (Fig. 2E). Expression of reelin, a general neuronal marker (Costagli et al., 2002), was decreased in PINK1 morphants. In accordance with the Drosophila PINK1 knock-out data (Yang et al., 2006), parkin expression was also reduced in the brain of PINK1 morphants. However, fezl mRNA, one of the factors required for the development of the dopaminergic system in zebrafish, was selectively increased in the posterior diencephalon of PINK1 morphants. Neurogenin-1, a downstream target of fezl (Jeong et al., 2006), was similarly upregulated in PINK1 morphants (Fig. 2E). In the spinal cord, PINK1 morphants showed decreased islet-1 and acetylated tubulin staining, suggesting loss of peripheral neurons (Abdelilah et al., 1996; Devine and Key, 2003) (Fig. 2F–I).
Morphological characterization of PINK1-deficient fish. A, B, Acetylated tubulin immunostaining in MO-injected (A) and five-mispair MO control (B) embryos at 48 hpf. The decrease in staining in the tectal area is indicated by an arrow. C, D, TH and 5-HT staining in representative MO-injected (C) and five-mispair MO control (D) embryos at 48 dpf. The diencephalic cluster of dopaminergic cells is outlined. E, In situ hybridization of PINK1 MO-injected and control embryos at 48 hpf. The increase in fezl and ngn-1 expression in morphants is indicated by arrows. F, G, Immunohistochemistry with antibody 40.2D6 (islet-1) in the trunk of control (F) and MO-injected (G) fish, showing decreased staining in MO fish. H, I, Acetylated tubulin staining in the trunk of control (H) and MO-injected (I) fish. Changes in the morphology (e.g., decrease in neuron numbers) of spinal neurons are indicated by arrows. RB, Sensory Rohon-Beard neuron; MN, primary motoneuron. Scale bars, 100 μm.
To test whether the zebrafish PINK1 knockdown phenotype was relevant to human PINK1 function, experiments were performed to ascertain whether the MO phenotype could be rescued by wild-type human PINK1 mRNA. mRNA (65 pg, which by itself did not cause an obvious external phenotype) was injected together with the MOs and the phenotype scored at 48 hpf (Fig. 3A). The mRNA mostly rescued the PINK1 MO phenotype as changes of the shape of tail yolk, curvature of the spine/tail and loss of escape response were reversed. Although 71.7% (±6.6%, SEM) of the MO treated fish had the slim tail yolk/no tail yolk phenotype, this was reduced to 35.8% (±7.3%, SEM; p < 0.001; n = 6) with coinjection of the human PINK1 mRNA with the MO. Likewise, the curved spine phenotype in the MO fish (82.4 ± 6.3%) was improved to 41.8% (±14.2%, SEM; p < 0.001; n = 4) by coinjection of the human PINK1 mRNA, and the escape response in the MO fish (30.9 ± 8.5%, SEM) was improved to 55.3% (± 7.2%, SEM; p < 0.001; n = 7) by coinjection of the human PINK1 mRNA. In addition, overexpression of human PINK1 mRNA partially restored decreased levels of PINK1 protein, confirming the specificity of the used antibodies (Fig. 3B, supplemental Fig. 5A, available at www.jneurosci.org as supplemental material). The injection of mRNA into five-mispair control MO-injected embryos at this concentration did not produce obvious changes in the normal phenotype.
Rescue of PINK1 MO phenotype with full-length human PINK1 mRNA. A, Rescue of the overall MO phenotype at 48 hpf by coinjected full-length human PINK1 mRNA. The phenotypes assessed (curved back, small tail-yolk extension) are indicated with arrows. Scale bar, 100 μm. B, Coinjection of MOs with human full-length mRNA partially restores PINK1 levels in 24 hpf embryos. C, Coinjection of mutant human mRNA did not produce a rescue of PINK1-deficient phenotype. The bars represent changes in the assessed phenotype relative to the MO-induced phenotype in three independent experiments. Error bars indicate SEM.
Several mutations of PINK1 that alter its activity have been described in humans (Valente et al., 2004; Silvestri et al., 2005). We introduced two such mutations, A168P and W437X, into the human full-length PINK1 mRNA. These mutated mRNAs did not rescue the phenotypes observed with PINK1 MO in the same way that we observed with the wild-type PINK1 mRNA (Fig. 3C).
PINK1 knockdown influences AKT-GSK3β axis
The tail phenotype of PINK1 MO-injected zebrafish at 24 hpf (short tail) looked remarkably similar to wnt pathway mutants (e.g., pipe-tail) (Rauch et al., 1997). Furthermore, fezl, the expression of which is upregulated in our PINK1 morphants, is also induced in zebrafish embryos expressing the wnt pathway inhibitor DKK1 (Hashimoto et al., 2000a). Accordingly, we hypothesized that PINK1 knockdown enhanced GSK3β activity (a consequence of wnt inhibition). Indeed, PINK1 morphants showed a decrease in the amount of inactive Serine 9 (Ser9)-phosphorylated forms of GSK3β (Fig. 4A). Active GSK3β phosphorylates β-catenin, enhancing its degradation. Consistent with predictions, active β-catenin levels were decreased in PINK1 morphants (Fig. 4B).
Rescue of PINK1 MO phenotype via modulation of AKT/GSK3β pathway. A, Decrease of Ser9-phosphorylated GSK3β in lysates of 24 hpf PINK1 MO embryos. Because this antibody detects both Ser9 GSK3β (47 kDa) and GSK3α (51 kDa), the lower band (GSK3β) was used for quantification. Blots obtained from three independent experiments were used for quantification and the levels of Ser9 GSK3β were normalized to the corresponding levels of the total GSK3β. Ser9 GSK3β phosphorylation was reduced by the morpholino to 27.6% (±16.56%, SEM) of the control level (set to 100%). p = 0.01 as analyzed by t test. B, Decrease of the active form of β-catenin in 24 hpf embryos injected with 7 ng of PINK1 MO. Active β-catenin bands were quantified as a function of the actin loading control in three independent experiments. Active β-catenin levels were reduced by the morpholino to 40.2% (±11.1%, SEM) of the control levels (set to 100%). p < 0.01 as analyzed by t test. C, Rescue of the short tail phenotype in 24 hpf PINK1 MO-injected embryos by nonspecific (50 mm of LiCl) and specific (10 μm of SB216763) GSK3β inhibitors. Bars represent three individual experiments, with 15–20 fish in each group per experiment. *p < 0.05, **p < 0.01, as analyzed by unconditional logistical regression analysis. D, Recovery of active β-catenin levels in PINK1 MO-injected embryos after treatment with 50 mm LiCl. The numbers under the blots are the ratios between the active β-catenin band intensities and actin loading control band intensities in relation to the control values. Independent replicate of the experiment could be seen in Figure 6C. E, The activity of AKT is decreased in PINK1 MO-injected embryos at 24 hpf. Ser473 AKT bands were quantified as a function of total AKT in three independent experiments. ***p < 0.01, as analyzed by t test. Error bars indicate SEM.
To test whether enhanced GSK3β activity contributed to the phenotypes we observed in PINK1 morphants, we treated them with nonspecific (LiCl) and specific (SB216763) GSK-3β inhibitors. After performing preliminary concentration–response curves with LiCl (supplemental Fig. 5B, available at www.jneurosci.org as supplemental material), we used 50 mm for a series of independent replicate experiments. Both compounds were able to rescue ∼20% of short tail morphants at 24 hpf (Fig. 4C). Moreover, treatment with LiCl partially restored decreased levels of active β-catenin in PINK1 MO embryos (Fig. 4D). However, both treatments failed to reverse the dark-head phenotype.
AKT is an important regulator of GSK3β activity as it phosphorylates GSK3β on serine 9, resulting in inhibition of the enzyme. AKT activity, assessed by the amounts of its phosphorylated serine 473 form, was decreased in whole lysates of 24 hpf PINK1 morphants (Fig. 4E). As expected, LiCl, which rescued tail phenotype of PINK1 morphants and restored Ser9 GSK3β levels, did not affect decreased levels of Ser473 AKT.
Loss of PINK1 leads to cell death and accumulation of ROS
PINK1 morphants exhibited dark, nontransparent heads as early as 24 hpf. Later, at 48 and 72 hpf, the heads of MO-injected fish were notably smaller, nontransparent, and occasionally lacking the hindbrain–midbrain boundary. These phenotypes have been attributed to ongoing neurodegeneration (Furutani-Seiki et al., 1996). Here, the lack of PINK1 protein led to an increase of acridine orange accumulation in the brain and throughout the body at 24 hpf (Fig. 5A). Because acridine orange accumulation in vivo is indicative of the cell death, we examined the activity of caspase-3, which was increased in the whole fish lysates already at 24 hpf (Fig. 5B) above the levels seen in the positive control group (UV-exposed embryos). Exposure to 50 mm LiCl for 18 h led to the partial decrease of caspase-3 activity in 1 dpf PINK1 MO-injected embryos (Fig. 5C). Administration of the pan-caspase inhibitor zVAD did not rescue the tail phenotype like LiCl did (data not shown), suggesting that although GSK3β overactivity does contribute to the caspase-3 activation, the effect of GSK3β inhibitors on the tail phenotype in PINK1 MO is not related to the caspase-3 activity.
Increase of apoptosis in PINK1 MO-injected fish. A, Acridine orange staining of MO-injected embryos. Stained cells (fluorescent) represent dead cells. The arrows show increased acridine orange accumulation in brain and spine of short-tail PINK1 MO-injected embroys. B, Caspase-3 activity is increased in MO-injected fish at 24 hpf. Values represent means of three samples, each of which is the total lysate of 15 fish. C, LiCl (50 mm) rescues the increased levels of caspase-3 in 24 hpf PINK1 MO-injected embryos. Values represent means of three samples, each of which is the total lysate of 15 fish. *p < 0.05; **p < 0.01; ***p < 0.01, as analyzed by one-way ANOVA with Bonferroni post hoc test. D, The mitochondrial Δψ membrane potential, represented as the ratio 590/530 nm (orange/green) of JC-1, is decreased in PINK1 morphants. LiCl (50 mm) does not change decreased levels of JC-1 in MO embryos. Bars represent three independent experiments, each of which uses the total lysate of 80–100 24 hpf embryos. Scale bars, 100 μm. Error bars indicate SEM.
To assess the integrity of the mitochondrial outer membrane, mitochondria were isolated from ∼100 embryos at 24 hpf and incubated with JC-1. MO-injected embryos displayed an ∼30% decreased JC-1 orange/green ratio in the mitochondrial fraction, indicating a lower mitochondrial potential (Δψ) (Fig. 5D). Inhibition of GSK3β with LiCl did not improve mitochondrial potential in PINK1 MO embryos. PINK1 morphants also had increased levels of ROS: 24 hpf MO-injected embryos had increased levels of ROS, as detected by accumulation of oxidized CM-H2DCFDA probe in vivo (Fig. 6A). Intracellular oxidation of CM-H2DCFDA can be caused by various reactive species, including H2O2, peroxy products, peroxyl radicals, and singlet oxygen. The increased ROS levels caused by the PINK1 MO were mostly normalized by 50 mm LiCl or antioxidant drugs (100 μm of either GSH or NAC) (Fig. 6A). In addition, these antioxidant drugs resulted in a small improvement in the tail phenotype (Fig. 6B), and the treated fish looked generally less severely affected by the loss of PINK1. As expected, LiCl rescued the decrease in active catenin levels in the PINK1 MO-treated fish, whereas GSH and NAC did not (Fig. 6C).
ROS levels are increased in PINK1 MO-injected fish. A, Accumulation of the CM-H2DCFDA dye, as the indicator of ROS levels, in live 24 hpf PINK1 MO-injected embryos. The increased levels of ROS are decreased using 100 μm GSH or 100 μm NAC, compounds with antioxidant qualities. Values represent means of three samples, each of which is the total lysate of 15 fish. B, Quantification of the short-tail phenotype at 24 hpf in embryos from A representing three independent experiments, each of which is the quantification of 15–20 24 hpf embryos. *p < 0.05, one-tailed t test. Error bars indicate SEM. C, Detection of active catenin in PINK1 MO-injected fish, treated with 50 mm LiCl, 100 μm GSH, and 100 μm NAC. Gel is representative of data from two independent experiments.
Discussion
Here, we describe a vertebrate model of PINK1 insufficiency. Decreased PINK1 activity in embryonic zebrafish resulted in a general developmental delay, aggravated by a severe mispatterning of axonal scaffolds and reduction of neuronal populations, in the dopaminergic system but also in other areas (e.g., spinal motor neurons). We observed a relatively modest decrease of TH-positive cells in the brains of 2 dpf PINK1 MO-injected embryos, ∼30% from the control. It is difficult to assess whether this decrease is attributable to the developmental retardation of dopaminergic neurons, or to the degeneration of already developed neurons. However, neurodegeneration and increased apoptosis (as judged by acridine orange staining and elevated caspase-3 activity) does occur as a result of PINK1 knockdown, and it is thus likely that this contributes significantly, at least in part, to the neuronal loss observed.
We found that PINK1 knockdown was associated with elevation of GSK3β activity, which contributes to at least some aspects of the knockdown phenotype, because a degree of rescue was seen with GSK3β inhibitors. Our data are compatible with the elevation of GSK3β activity attributable to the inhibition of AKT via dephosphorylation at Ser9. AKT is one of main contributors to the phosphorylation of GSK on Ser9, producing the inhibition of the enzyme.
The question remains how the loss of function of mitochondrially located PINK1 would affect the functional activity of cytoplasmic proteins, such as GSK3β and AKT, and further studies will be required to elucidate the molecular events linking PINK1 to AKT/GSK3β. However, GSK3β has been identified within mitochondria (Hoshi et al., 1996), as well as in the cytoplasm. Because the LiCl rescue is restricted to the tail/trunk organization and development, it is possible that the apparently pleiotropic phenotypes seen in the PINK1 knock-out fish are related to mitochondrial GSK3β and general mitochondrial dysfunction. For instance, zebrafish morphants with specific inner mitochondria membrane pathologies display severe tail/trunk anomalies that are similar to those we saw with PINK1 knock-out (Guo et al., 2004), and PINK1 RNA is enriched in mitochondria rich tissues such as muscles and brain (Unoki and Nakamura, 2001). The roles of PINK1 in mitochondria that have been elucidated to date are compatible with the pleiotropic phenotype model we proposed. Recently, it was shown that the activation of HtrA2/Omi and consequent protection of mitochondria from stress is PINK1-dependent (Plun-Favreau et al., 2007). In addition, TRAP1 and parkin have been shown to operate downstream of PINK1 (Clark et al., 2006; Park et al., 2006; Pridgeon et al., 2007). All of these proteins are involved in the protection of the cell from various insults, further strengthening the idea of PINK1 being the important neuroprotective mitochondrial protein.
We detected two PINK1-immunopositive bands in the lysates of whole zebrafish embryos. The intensity of both bands was decreased after MO injection and increased after overexpression of wild-type human PINK1 mRNA, supporting the specificity of the observed signals. Whereas the full-length zebrafish PINK1 has predicted molecular mass of 64 kDa and is most likely represented by the bigger band in our experiments, the nature of the smaller band (∼30 kDa) is not clear. The truncation of N terminus of the protein could contribute to the generation of such a fragment (as our antibodies detect C terminus of the protein). Also, in experiments in which films had long exposures, we observed another band (e.g., 50 kDa) (supplemental Fig. 5A, available at www.jneurosci.org as supplemental material), which could represent another splice variant. This sized PINK1 band has been described in the human brain (Gandhi et al., 2006)
Consistent with loss of PINK1 causing increased GSK3β activity, we observed an increase in the expression of fezl (a gene regulated by the wnt/GSK3β pathway) (Hashimoto et al., 2000b) and its downstream target neurogenin-1 in our PINK1 morphants. Our PINK1 morphants had decreased numbers of dopaminergic neurons, whereas neurogenin-1 overexpression in otherwise normal zebrafish leads to an increase in this neuronal population (Jeong et al., 2006). Thus, it is likely that the phenotype mediated by GSK3β activity in the brain is independent of neurogenin-1 and is mediated by other pathways regulated by this broad-acting enzyme. We observed no obvious benefit of LiCl treatment on the neurodegeneration in the brain of morphants, in contrast to the rescue seen with the tail/trunk phenotype. These findings raise the possibility that PINK1 deficiency may selectively affect mitochondrial function and consequently, the development of mitochondria-rich tissues.
We observed that zebrafish PINK1 morphants had increased levels of ROS as early as at 24 hpf. This could be caused by mitochondrial malfunction, as the mitochondrial membrane potential, assessed by accumulation of JC-1, was decreased in PINK1 morphants. However, we cannot exclude the possibility that increased ROS levels result from activation of cell death pathways in these fish. We have attempted to test whether the GSK3β and ROS pathology seen in the PINK1 morphants were related. The antioxidants used did not affect GSK3β activity. However, LiCl did reduce ROS and caspase-3 levels while not affecting the decrease of mitochondrial membrane potential, suggesting that at least some of the ROS accumulation may be downstream of GSK3β in this model. Furthermore, LiCl failed to rescue the TH+ neuronal deficit in our morphants (data not shown). This suggests that GSK3β overactivity is related to PINK1 function in peripheral tissues, and the decrease in TH+ neurons is primarily attributable to factors distinct from GSK3β.
In our study, we obtained similar rescue of the PINK1 knockdown phenotype with a “specific” GSK3β inhibitor or with lithium, which inhibits this enzyme but also has other effects. Although it is very likely that the effects we observed are caused by GSK3β inhibition, they are probably not caused by a complete inhibition of enzyme activity. The inhibition of GSK3α and β activity results in severe apoptosis in the CNS, accompanied with abnormalities in body axis formation and heart development (Lee et al., 2007). These data indicate that the complete inhibition of this enzyme results in very severe consequences. [It would not really be informative to try to test whether Akt overexpression had similar effects to lithium, as Akt has protective effects independent of GSK3β signaling and has diverse signaling consequences (in addition to GSK3).]
The reduction of PINK1 levels in zebrafish also led to a decreased expression of parkin (Fig. 2F). In future studies it will be interesting to test whether overexpression of parkin can rescue aspects of the PINK1 knockdown phenotype in a similar way to what was described previously in Drosophila (Clark et al., 2006; Park et al., 2006; Yang et al., 2006). However, such studies will be a major undertaking, as the expression patterns of zebrafish parkin are uncharacterized, and as it is not clear whether zebrafish parkin is functionally ortholgous to human parkin, knockdown and overexpression experiments will be required for robust interpretations.
In conclusion, decreasing levels of PINK1 protein produces significant developmental and neurochemical alterations in embryonic zebrafish. These alterations are clearly associated with mitochondrial alteration, leading to the activation of caspases and accumulation of ROS and with activation of GSK3β. These data lead to insights into normal PINK1 function, which is believed to be lost in PD cases with mutations in this gene (Sim et al., 2006). Our data raise the speculative possibility that drugs like lithium chloride and more specific GSK3β inhibitors, as well as antioxidants, may be of value in some forms of PD, particularly in familial cases caused by PINK1 mutations.
Footnotes
-
H.D., A.F., and A.R. are employees of Summit plc and have share options in this company. A.F., P.G., and A.R. are shareholders in Summit plc.
-
This work was supported by the Medical Research Council and the Department of Trade and Industry. D.C.R. is a Wellcome Trust Senior Research Fellow in Clinical Science. We acknowledge the Summit (Cambridge) aquarium staff for support and maintenance of the fish facility.
- Correspondence should be addressed to either of the following: Oleg Anichtchik at his present address, Cambridge Centre for Brain Repair/Department of Clinical Neuroscience, E. D. Adrian Building, Forvie Site, Robinson Way, Cambridge CB2 0PY, UK, oa220{at}cam.ac.uk; or David C. Rubinsztein, Department of Medical Genetics, Cambridge Institute for Medical Research, Wellcome/MRC Building, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2XY UK, dcr1000{at}hermes.cam.ac.uk
