Impaired Wnt signaling in dopamine containing neurons is associated with pathogenesis in a rotenone triggered Drosophila Parkinson’s disease model

Parkinson’s disease, which is the one of the most common neurodegenerative movement disorder, is characterized by a progressive loss of dopamine containing neurons. The mechanisms underlying disease initiation and development are not well understood and causative therapies are currently not available. To elucidate the molecular processes during early stages of Parkinson’s disease, we utilized a Drosophila model. To induce Parkinson’s disease-like phenotypes, we treated flies with the pesticide rotenone and isolated dopamine producing neurons of animals that were at an early disease stage. Transcriptomic analyses revealed that gene ontologies associated with regulation of cell death and neuronal functions were significantly enriched. Moreover, the activities of the MAPK/EGFR- and TGF-β signaling pathways were enhanced, while the Wnt pathway was dampened. In order to evaluate the role of Wnt signaling for survival of dopaminergic neurons in the disease model, we rescued the reduced Wnt signaling activity by ectopic overexpression of armadillo/β-catenin. This intervention rescued the rotenone induced movement impairments in the Drosophila model. Taken together, this initial study showed a highly relevant role of Wnt signaling for dopamine producing neurons during pathogenesis in Parkinson’s disease and it implies that interfering with this pathway might by a suitable therapeutic option for the future.


Rotenone induced changes in locomotor behavior in an administration time dependent manner.
This study aims to elucidate the molecular changes occurring in DA neurons during early phases of disease development. PD like symptoms were induced by rotenone application. Previous studies have shown that detectable motor impairments and DA neuron loss start after chronic exposure with rotenone for several days and that this symptomatic phase starts after the majority of DA neurons were already severely damaged 20,23 . This severe motor impairment is known to be not associated with a concurrent demise of dopaminergic cells 24 . In order to monitor the onset and development of the PD-associated pathologies, we used our experimental animals (F1 of TH-GAL4 X 20xUAS-mCD8::GFP) and treated them with rotenone ( Fig. 1). Brains of non-treated animals (Fig. 1A,A′) and of rotenone treated (0.5 mM for 10 d; Fig. 1B,B′) were analyzed. Shown are two regions per treatment type (control, A,A′; rotenone for 10d, B,B′) of representative brains of the above mentioned genotype (TH-GAL4 X 20xUAS-mCD8::GFP) labeled with anti-GFP antibodies. The TH-Gal4 driver line labels about 50% of all dopaminergic cells in the fly's brain 25 . The number and location of neurons remained mostly unchanged, which was in line with previous observations 24 . In non-treated control animals, we observed a mean of 52.13 (±2.1, S.E.M) GFP-positive cells per hemisphere, whereas we identified 49.5 (±5.3, S.E.M.) cells per hemisphere in rotenone-treated animals. This difference was not statistically significant (p = 0.92, N = 8). In contrast, treatment with the sublethal dose of 0.5 mM rotenone had a significant effect on the motor abilities of the treated flies. We quantified their ability to climb a vertical plane (Fig. 1C). There was no obvious locomotor deficit observed in flies exposed to rotenone for up to 3 days. From day 6 onwards, flies showed statistically reduced locomotor ability in comparison to untreated flies, meaning that their mean climbing distance was reduced substantially (Fig. 1C). Thus, we choose flies treated for three days with rotenone for all subsequent transcriptomic analyses. At this time point, we can assume that the first signs of pathological alterations had been induced, but still before the onset of the phenotypical hallmark, the locomotor impairment, of this disease model. Consequently, we performed whole transcriptome analysis with RNA isolated specifically from DA neurons after this time of treatment to ensure that those effects can be identified that occur prior to development of the disease associated phenotype. In parallel, parts of the experimental populations were used for phenotypical analyses of the rotenone induced Parkinsonism over time, which always showed the kinetics as outlined in Fig. 1C. In order to isolate the mCD8::GFP tagged neurons, they were decollated first (Fig. 1D). The figure shows few GFP-positive cells (TH-Gal4 positive) and numerous other, non-labeled neurons. Using an isolation protocol based and anti-mCD8 coupled magnetic beads, the mCD8-GFP-tagged cells were isolated and all non-labeled cells depleted from the preparation. The high degree of enrichment of mCD8-GFP-positive cells is shown in Fig. 1E, where some GFP-positive cells (green arrows), but no non-labeled cells are detectable.
Rotenone induced gene expression changes in DA neurons in an early phase of PD. Although flies treated with 0.5 mM rotenone for three days appear to coordinate their movement normally (Fig. 1B), a number of genes show differential expression in DA neurons obtained from rotenone treated animals compared with those isolated from sham-treated ones. Expression of 765 genes was up-regulated (>1.5 fold) and that of 357 genes was down-regulated (<0.67 fold) if compared with the signals obtained from DA neurons isolated from sham-treated animals. To further filter and identify relevant regulated genes in the two lists, we used different Gene Ontology (GO) analysis programs such as the DAVID 26    cell cycle were also enriched. Applying the DAVID program package to the data revealed additional GO terms as being overrepresented. Those genes with functions related to cellular transport were changed with the highest frequency (83 genes), followed by various others ( Fig. 2A). Other important categories are related to stress response (14 genes), regulation of apoptosis (13 genes), protein folding (12 genes) and ageing (10 genes) ( Fig. 2A). These results might be interpreted as a defense response towards a mild toxic offence given by rotenone. Likewise, a list of down-regulated genes was generated and clustered into 9 relevant functional categories (Fig. 2B).
Collectively, GO terms that were highly enriched in the combined cohorts of up-and down-regulated genes included regulation of apoptosis (Fig. 2C), oxidative phosphorylation (OXPHOS, Fig. 2D), neurotransmitter secretion ( Fig. 2E) or ageing (Fig. 2F). Additional relevant clusters with high enrichment score and p < 0.05 and KEGG pathways filtered by DAVID are listed in Table 1 and Fig. S1, respectively. To validate the differential expression obtained using the microarray data, we performed qRT-PCR experiments with 3 informative genes, namely down-regulation of dat (Dopamine N-acetyltransferase) as well as the enhanced expression of calmodulin and of lethal2NC136 (Fig. S2).

Pathway enrichment analysis.
Since pathway analysis is anticipated to give more instructive data than single gene analysis 28 , further analyses focused on those pathways, whose activity was apparently deregulated. KEGG analysis revealed four pathways to be significantly altered, namely; the mitogen-activated protein kinases (MAPK/EGFR), the transforming growth factor beta (TGF-β), the target of rapamycin (Tor) and the Wnt signaling pathway. While activity of the first three pathways appear to be increased, that of the Wnt pathway appears to be reduced (Fig. 3, Table 1). Genes tightly associated with the MAPK/EGFR signaling are shown in Fig. 3A, while those associated with the Wnt pathway are shown in Fig. 3B. Expression levels of genes overrepresented in TGF-β and Tor signaling are shown in Table S1. In order to evaluate if the activity of the corresponding pathways were indeed induced or impaired, we analyzed either canonical pathway genes of these pathways or essential pathway constituents by qRT-PCR. For the MAPK/EGFR-pathway, expression of the canonical pathway gene spitz 29 was upregulated several fold (*p < 0.05; Fig. 3C). For Wnt signaling we analyzed armadillo, a central regulatory component of the Wnt signaling pathway (the β-catenin homolog). The relative expression level of armadillo in rotenone treated samples was significantly reduced (**p < 0.01; Fig. 3D). For the TGF-β signaling pathway, the expression level of its canonical target gene, Daughter against dpp (Dad) was analysed. Its expression level was significantly increased in rotenone treated samples (*p < 0.05; Fig. 3E), indicative for an increased pathway activity.
In order to evaluate the relevance of modulating the candidate signaling pathways, we employed the Gal4/ UAS-system. Using the TH-Gal4 driver line, this manipulation was restricted to DA containing cells only. We focussed on overexpression of armadillo (using UAS-arm) as the most important signalling pathway impaired by rotenone treatment. While the climbing performance index (relative proportion of animals that were able to cross a 2 cm threshold within 10 s) was identical between controls and those overexpressing armadillo were identical at the beginning of rotenone treatment, the decline in performance from day 10 onwards was dramatic while this was not the case for animals with ectopically increased armadillo expression in dopamine-producing cells only (Fig. 4). These differences were statistically highly significant.

Discussion
The major aim of the current study was to provide new information about the molecular signatures associated with early phases of an induced Parkinson's disease like phenotype that develop specifically in dopamine containing neurons of the brain. In order to reach this ambitious goal, focusing exclusively on these dopamine producing cells being in an early stage was mandatory. Drosophila is ideally suited for this purpose as PD-like phenotypes can be induced easily, their progression can be quantified non-invasively and the dopamine-producing cells can be labeled specifically. Analyzing the early stage is of prime importance, because, at least in humans, clinically noticeable phenotypes occur after an estimated 70% of susceptible DA neurons in the Substantia nigra have already been destroyed 30 . Although Drosophila models of Parkinson's disease recapitulate most aspects of the disease, this appears to be based on massive functional impairments of dopaminergic cells rather than on their death 24 . Thus, we analyzed in the current disease model a very early stage of cell pathology that later develops into the disease-associated functional impairments. We used a toxin-induced model of Parkinson's disease to enhance the freedom of operation for genetic manipulation. Although we decided to utilize rotenone treatment, rotennone and paraquat models are equally well-suited for the induction of Parkinson's disease symptoms in the fly 24,31 . The gene expression analysis carried out in this study is, to our knowledge, the first for rotenone-induced Parkinsonism in Drosophila that was focused on DA neurons. Prior studies of gene expression in Drosophila models of neurodegenerative diseases have been limited to studies of homogenates of brain tissue 32,33 . In contrast, a comparable approach has been performed in mammals, namely in rats. Rotenone treatment followed by laser-dissection and transcriptome analyses of dopaminergic neurons revealed complex sets of regulated genes 34 . A closer comparison of the sets of genes that are regulated in dopamine-containing neurons of rats and flies revealed some surprising commonalities in their response characteristics to rotenone treatment. Regulated Gene Ontologies comprise those associated with cell death and cell cycle as well as those directly associated with general neuronal activities such as neurotransmitter release, which implies that the reaction types of rat and fly dopamine-containing neurons are surprisingly similar. In human brain tissues, the laser capture microdissection (LCM) technique has recently been employed for capturing only DA neurons for transcriptional profiling studies 35,36 . The complex profile of genes with differentially regulated expression in response to rotenone treatment comprises some highly enriched pathways and gene ontologies. Presumably most relevant was the regulation of signaling pathways in these cells that are known to be highly relevant for cell survival and cell death, such as the MAPK/EGFR, TGF-β, Tor and Wnt pathways.
Of special interest in this study was the observed down-regulation of Wnt signaling in response to mild rotenone treatment in young adult flies. Armadillo, the Drosophila β-catenin acts as a transcription factor inducing expression of Wnt target genes. Wnt signaling is known throughout the animal kingdom to be involved in controlling diverse cellular processes including tissue differentiation, neuronal survival, synaptogenesis and plasticity, as well as neurogenesis and neuroprotection 9,37,38 . Moreover, deregulated Wnt signaling is believed to be involved in various neuropathologies including Alzheimer's disease, Schizophrenia and PD 8,10 . Canonical Wnt/β-catenin signaling appears to be highly important for controlling DA neuronal fate decision 39 . Remarkably, both GSK3-β inhibition and β-catenin stabilization increased commitment of neural precursors to develop into DA neurons 40 . Importantly, current studies imply that PD pathophysiology is associated with dysregulation of Wnt signaling 8,9,14,41 . Cantuti-Castelverti and colleagues reported a down-regulation of β-catenin levels in DA neurons of the Substantia nigra in PD patients 36 . In addition, proteins encoded by PARK genes, which were also shown to be involved in hereditary forms of PD, can modify Wnt pathway activity. LRRK2 (leucine-rich repeat kinase 2), which is associated with familial PD 42 was shown to be connected to Wnt signaling 8,13 and Parkin, an E3 ubiquitin ligase, regulates β-catenin protein levels in vivo 14 . Furthermore, gene expression profiling in progressively MPTP-lesioned macaques indicated down-regulation of β-catenin and dysregulation of key components of Wnt signaling 28 . Our results are consistent with these discussed findings, suggesting a general role of Wnt signaling in DA neuron welfare and PD development upon its impairment. Apparently, a certain level of Wnt signaling is necessary to guarantee survival of dopamine-containing neurons, especially in times of stress. Our rescue experiments support the view that downregulation of Wnt signaling is a key event in the neuropathology of PD and that amending this impairment improves the functionality and presumably also the survival of DA neurons at risk. Thus, interfering with Wnt signaling in DA containing neurons may represent a very promising and novel therapeutic strategy.
Very similar as Wnt signaling, MAPK signaling is evolutionarily well conserved. Drosophila melanogaster expresses all three subgroups of MAPKs: Rl (Rolled; ERK homolog), dJNK/Basket (Drosophila homolog of JNK), and dp38a and dp38b (Drosophila homologs of p38) 43,44 . Impairments of MAPK signaling pathways have been associated with many diseases including Alzheimer's, PD and diverse types of cancer 45 . Strikingly, even at this early stage of disease development, expression of four genes associated with the MAPK signaling pathway was approximately 2-fold increased in rotenone treated flies. Moreover, expression of the Egfr, which is the central receptor in this signaling pathway, was also increased. Mice lacking the EGFR develop neurodegenerative diseases and die early 46 . In culture models, EGF stimulated neurite outgrowth, increased dopamine uptake and enhanced long-term survival in cultured dopaminergic neurons. Ectopic activation of ERK1/2 in rotenone rat models of PD, on the other hand, protected dopamine neurons from cell death 47 . In Drosophila, proper functioning of EGFR signaling has been shown to be essential for learning and memory 48 . The increased expression of EGFR and Ras protein coding genes in rotenone treated Drosophila suggests that these neurons launch defensive mechanisms due to stress given by rotenone. These findings support the view that EGFR signaling promotes cell survival also in the vulnerable DA neurons.
The third highly relevant signaling system, the TGF-β pathway controls a surplus of cellular processes in both developing and adult organisms 49,50 . When components of the TGF-β pathway are disrupted, several human diseases, including neurodegeneration and cancer arise [50][51][52] . There is increasing support for a role of TGF-β signaling in neuronal maintenance function and degeneration [52][53][54] .
As most of the other pathways mentioned above, the target of rapamycin (TOR) signaling pathway is evolutionary conserved. It regulates cell proliferation, cell motility, cell survival, protein synthesis and transcription 55,56 . Nucleolar disruption and associated oxidative stress were demonstrated to suppress mTOR activity in DA neurons, thereby providing the basis for neuronal degeneration and the development of parkinsonism 57 . However, major key players of the pathway (TOR, Akt and elF4B) appeared to be up-regulated in flies treated with rotenone in the early phase, which is indicative for induction of survival mechanisms.
Collectively, this study shows that the expression of several genes involved in the MAPK/EGFR, TGF-β and TOR signaling pathways were increased, presumably in order to launch a protective cellular program. The observed down-regulation of Wnt signaling on the other hand may reflect early signs of neurodegeneration. Thus, increasing expression of either of the pathways mentioned above may increase survival of dopamine-containing neurons during disease progression, a hypothesis that is supported by the first experiments performed in this study focusing on enhancing Wnt signaling.

Methods
Fly strains and husbandry. Fly stocks were raised on standard cornmeal-agar medium at 25 °C under 12 h/12 h on/off light cycle. The following flies were used in the experiments described below: TH-Gal4 males (Bloomington Drosophila Stock Center, USA) were crossed to virgin females of either 20xUAS-IVSmCD8::GFP or UAS-arm (Bloomington Drosophila Stock Center, USA) to generate flies overexpressing GFP or armadillo in dopaminergic neurons, respectively. Toxin administration. Rotenone (Sigma Aldrich, Deisenhofen, Germany) was administered orally according to a previously described procedure 23 with minor modifications. Briefly, 3-day-old flies from the crossing mentioned above were exposed to 0.5 mM rotenone in 10% glucose on blotting paper at 25 °C. A volume of 300 µl rotenone/glucose solution was added every 48 h to avoid desiccation. For the control group the same volume of 10% glucose only was added.
In order to evaluate the relevance of selected signaling pathways for survival of dopamine containing neurons in the presence of rotenone, the Gal4/UAS system was employed utilizing the TH-Gal4 driver and effector lines used to ectopically overexpress relevant genes.
Immunohistochemistry. Immunohistochemistry was performed as previously described 58 . Brains were dissected manually in Drosophila Ringer's solution and immediately fixed in 4% paraformaldehyde in PBS for 30 min at room temperature. Subsequently, the samples were washed with PBST (0.3% Triton X-100 in PBS) and blocked in blocking-buffer (10% goat serum in PBST) for 30 min at room temperature, followed by incubation with the primary antibody (1:300 rabbit anti-GFP, Sigma-Aldrich, Taufkirchen, Germany) overnight at 4 °C with subsequent application of the secondary antibody (1:500 anti-rabbit DL488, Jackson ImmunoLabs, Suffolk, UK) for 3 h at room temperature. After washing, the brains were mounted on slides and images were obtained using a fluorescent microscope equipped with an apotome (Zeiss Axio Imager Z1, Göttingen, Germany). To quantify the effects of rotenone on cell numbers, fly brains of treated and non-treated animals (rotenone treatment for 10 days) were dissected and processed as described above. Z-stack images of rotenone treated and control fly brains were acquired using a. The number of GFP-labeled cells in the central brain (without optic lobes) was determined per brain hemisphere with the Fiji Cell Counter plugin, which facilitates cell counting of 3D images 59 .
Behavioural testing-Climbing ability (negative geotaxis) assay. Locomotor ability of adult flies was tested by a negative geotaxis assay as described previously 16 with minor modifications. Twenty male adult flies were placed into a 17 cm long glass tube at a given time point. The flies were tapped to the bottom of the tube and let to climb the tube. After 20 s (or 10 s), a photo was taken in which most of the healthy flies were expected to have crossed the escape line at a height of 6 cm. The height/distance climbed by each fly was analyzed by using Image J.
For the recue experiments, we used the performance index as a quantitative measure 60 . In brief, animals (20 each) were analyzed at the indicated time points by tapping them and counting the number of animals that crossed a 2 cm line within 10 s. The performance index is calculated as follows: PI = 0.5 * ((total number of animals + number of animals above the line − number of animals below the line) divided through the total number of animals). In different types of experiments, we used heights achieved after 10 s or 20 s, depending on preliminary experiments under the respective conditions. Tissue collection and analysis. Seventy heads of F1 generation male adult flies from the crossings between TH-Gal4 and 20xUAS-IVSmCD8::GFP were dissected in pre-chilled HL3 61 buffer. Dopaminergic neurons were sorted based on their mCD8:GFP expression by magnetic Dynabeads MyOne Streptavidin T1 (Invitrogen, Oslo, Norway) according to a previously described protocol 62 . Briefly, the tissue sample was vortexed for 1 sec, the supernatant was discarded and the procedure was repeated 3-4 times until the supernatant became clear. The heads were transferred to a pre-chilled 7 ml Kontes tissue grinder (Fisher Scientific, Leicestershire, UK), which was rinsed with 1% BSA in HL3 buffer in order to avoid the cells from sticking to the glass surface. About 4 ml of HL3 buffer were added to the tissue grinder and gently using a pestle, the tissues were given 30-32 douncing strokes. The solution was then triturated 5 times using the fire-polished glass pipette narrowed to approximately 50% of the standard tip diameter. The level of dissociation was assessed using a fluorescent microscope with a GFP filter (Axiovert S. 100, Zeiss, Jena, Germany). Then the solution was filtered through a 30 µm cell filter (Miltenyi Biotec, Bergisch Gladbach, Germany). The Dynabeads coupled to undiluted biotinylated rat anti-mouse CD8a antibody (eBioscience, Frankfurt, Germany), was then added to the filtrate and incubated for 1 hour on ice. Following this, the tubes were placed on a MagnaRack (Invitrogen, Karlsuhe, Germany) for 2 min to pellet the beads along with GFP positive cells. The supernatant was removed and discarded. Then the cells were washed three times with ice cold HL3 buffer by putting the tubes on the magnet to remove non-specific cell binding. Microarray analysis. Microarray analyses were performed as described earlier 63,64 . Briefly, cDNA was synthesized by using PrimeScript RT (Takara Bio Europe, Saint Germain-en-Laye, France) according to the manufacturer's protocol using a CapFinder approach in order to amplify the entire cDNA population 65,66 . The following primers were employed: CapFinderSp6rG (5′-CAG CGG CCG CAG ATT TAG GTG ACA CTA TAG  A rGrGrG-3′) and OdT T7 I (5′-GAG AGA GGA TCC AAG TAC TAA TAC GAC TCA CTA TAG GGA GAT  TTT TTT TTT TTT TTT TTT T G/A/C-3′). cDNA was amplified with OdT T7 II (5-GAG AGA GGA TCC AAG  TAC TAA TAC GAC TCA CTA TAG G-3′) and Adaptor Sp6rG (5′-GAC GCC TGC AGG CGA TGA ATT TAG G-3′) and LA Taq polymerase. In vitro transcription of cDNA was performed with MEGAscript ® T7 including aminoallyl-UTP and subsequently labeled with Alexa Fluor 647 or 555 (Life Technologies, Darmstadt, Germany) for treated or control sample, respectively. cRNA probes were hybridized to microarray slides (Canadian Drosophila Microarray Centre, Toronto, Canada) and scanned with a Gene Pix 4000B scanner equipped with the Gene Pix Pro 6.0 software (Axon Instruments, Science Products, Hofheim, Germany). Data normalization of the probe signal intensity levels across the arrays was performed with Acuity 4.1 (Axon Instruments, Science Products, Hofheim, Germany). A fold change of >1.5 of the mean signal intensity of a specific gene in at least 2 arrays out of three was considered as up-regulated and fold change <0.67 as down-regulated. The enrichment of Gene Ontology terms, and visualization of genes on KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway maps were done with the help of the online FlyBase database (http://flybase.org/), DAVID (Database for Annotation, Visualization and Integrated Discovery 26 and GOrilla (Gene Ontology enrichment analysis and visualization tool 27 . The microarray raw data have been deposited in the GEO database under the following accession number: GSE74247. Quantitative real-time PCR. Quantitative real-time PCR (qRT-PCR) analysis was performed using cDNA samples prepared as described above utilizing a StepOne TM Real-Time PCR system (Applied Biosystems, Darmstadt, Germany) with the DyNAmo Flash SYBR Green qRT-PCR kit (Fisher Scientific, Schwerte, Germany). Ribosomal protein l 32 (Rpl32) was used as an internal control gene and expression data were analyzed according to Pfaffl 67 . The primer sets used in this analysis are listed in Table S2.
Statistical analysis. Statistical analyses were performed using Unpaired-two-tailed Student's test using Graph Pad Prism software (version 5). The data were presented as mean values +SEM.