In an aging society, research involving neurodegenerative disorders is of paramount importance. Over the past few years, research on Alzheimer's and Parkinson's diseases has made tremendous progress. Experimental studies, however, rely mostly on transgenic animal models, preferentially using mice. Although experiments on mice have enormous advantages, they also have some inherent limitations, some of which can be overcome by the use of Drosophila melanogaster as an experimental animal. Among the major advantages of using the fly is its small genome, which can also be modified very easily. The fact that its genome lends itself to diverse alterations (e. g. mutagenesis, transposons) has made the fly a useful organism to perform large-scale and genome-wide screening approaches. This has opened up an entirely new field of experimental research aiming to elucidate genetic interactions and screen for modifiers of disease processes in vivo. Here, we provide a brief overview of how flies can be used to analyze molecular mechanisms underlying human neurodegenerative diseases.
Neurodegenerative diseases like Alzheimer's disease, Parkinson's disease (PD) or frontotemporal lobar degeneration are devastating age-related disorders. The mechanisms leading to the development and onset of such diseases are still elusive. To date, we have found neither a cure for these diseases nor any means to arrest their progression. Thus, patients suffering from a neurodegenerative disease only receive symptomatic treatment. An in-depth understanding of disease etiology and the mechanisms involved in disease onset and progression, therefore, are a prerequisite for a rational design of potential therapies.
In order to shed light on the pathogenesis of neurodegenerative disorders, researchers utilize animal model organisms. Invertebrate model systems like Drosophila melanogaster are particularly suited to address certain questions pertaining to neurodegenerative disorders. Although evolutionarily separated, flies and humans share basic molecular mechanisms. In the context of neurodegenerative diseases, the fact that around 70% of disease-associated human genes have a fly homolog (Bier 2005) makes research with this organism feasible. Moreover, there is a reasonable similarity between the central nervous systems of flies and humans, with both consisting of neurons and glia and utilizing the same neurotransmitters. In combination with the great variety of established genetic tools, these similarities render Drosophila a useful model organism to study the etiology of human neurodegenerative diseases (Shulman et al. 2003). In this review, we would like to provide a brief overview of how research using Drosophila has led to novel insights into human neurodegenerative diseases.
Screens: unbiased high throughput analysis in Drosophila
Among the advantages of using flies for research are low costs and efficient handling in terms of required room and time (Greenspan 2004), which allow maintenance of large collections of fly lines in stock centers with public access. Owing to the small and by now fully sequenced genome of flies (Adams et al. 2000; Rubin et al. 2000), large-scale screening approaches are easy to perform. Consequently, a plethora of classical forward and backward genetic screens have been used to identify specific genes in flies and elucidate their functions.
In forward screens, randomly mutagenized flies, e. g. by chemicals or X-ray radiation, are screened for disturbances of a pre-defined phenotype/process. Such screens have proved very helpful as highlighted by the Nobel prize to C. Nüsslein-Volhard, E. Wieschaus and E.B. Lewis in 1995. Here, forward genetics were used to screen for genes involved in early development, namely the segmentation of the embryo (Nusslein-Volhard and Wieschaus 1980). A drawback of such screens was the difficult and laborious fine-mapping of random mutations (usually loss-of-function mutations) to specific genes. In former times, mapping of mutations was only possible if the mutant allele caused an obvious, easy-to-score phenotype or lethality. This problem has been partially overcome by the development of single nucleotide polymorphism maps (Berger et al. 2001) or whole-genome sequencing (Blumenstiel et al. 2009). Another way of generating loss-of-function mutations in flies is transposon-mediated mutagenesis (e. g. by P-elements see Bellen et al. 2004; Cooley et al. 1988). Here, transposable elements are mobilized with the aim of generating new insertion sites and thereby disrupting genes (Fig. 1). The benefit of using transposable elements is the straight-forward mapping of the insertion site, which allows identification of the disrupted gene. However, in this way only a small percentage of the Drosophila chromosomes can be mutagenized because transposable elements do not randomly integrate into the genome but do so preferentially in specific hot-spots (Spradling et al. 1995; Bartolome et al. 2002; Peter et al. 2002). Despite that, non-random integrating transposons have been generated, for example minos (Metaxakis et al. 2005) or piggyBac (Thibault et al. 2004). Nevertheless, transposable elements are still useful tools, as they can be modified in a fashion to suit specific requirements and applications (e. g. gene disruption, enhancer trapping, introduction of recombination sites, over-/misexpression etc. see for example Ryder and Russell 2003; Akimoto et al. 2005; Hoehne et al. 2005; Hummel and Klambt 2008; Venken and Bellen 2012). This is why nowadays large collections of transposon-inserted fly lines are freely available (O'Kane and Gehring 1987; Bier et al. 1989; Thibault et al. 2004; Bellen et al. 2011), e. g. at the Bloomington Drosophila Stock Center (http://flystocks.bio.indiana.edu/), the Exelixis Collection at the Harvard Medical School (https://drosophila.med.harvard.edu/) and the Drosophila Genetic Resource Center in Kyoto (http://www.dgrc.kit.ac.jp/en/index.html).
Reverse genetic screens explore the functions of predefined genes (St Johnston 2002). After the disruption of a targeted gene, the resulting phenotype is analyzed. The fly provides a large variety of methods to disrupt target genes. Mobilization of transposable elements allows random generation of mutant alleles (Fig. 1) (Ryder and Russell 2003; Kim et al. 2012; Kao and Lee 2013). In addition, classical homologous recombination can be employed for knockout or replacement of genes/alleles (Rong et al. 2002; Choi et al. 2009). Gene silencing by RNA interference (RNAi) has been established (for a review see Kennerdell and Carthew 2000; Rao et al. 2009), representing means of mimicking classical knockout strategies. Using the binary UAS/Gal4 expression system (Fig. 2), short hairpin RNAs (shRNAs) can be expressed to efficiently induce RNAi-mediated silencing of endogenous genes in a spatiotemporal manner (Brand and Perrimon 1993). This permits analysis of the detrimental effects of gene inactivation in postmitotic cells of the nervous system. The Vienna Vienna Drosophila RNAi Center (VDRC), http://stockcenter.vdrc.at/control/main) implemented a collection of more than 22 000 transgenic Drosophila strains, each containing an inducible UAS-shRNA construct targeting a single protein-coding gene. More than 12 000 genes, or 88.2% of the fly genome, are represented in this collection (Dietzl et al. 2007).
One intriguing example of how reverse genetics can provide novel insights into human neurodegenerative diseases is the PTEN-induced kinase 1(PINK1)/Parkin pathway. To date, both PD-linked gene products are well accepted in mitochondrial quality control and mitophagy (removing dysfunctional mitochondria by authophagy). However, first in vivo evidence that PINK1 and Parkin might act in the same cellular pathway was derived from fly research. Analyzing loss-of-function mutations in Drosophila Pink1 (generated by imprecise excision of a P-element, see Fig. 1), it became evident that Pink1-deficient flies display almost identical phenotypes as described for parkin-deficient flies. This already suggested that both gene products might participate in the same pathway. Two groups independently confirmed this assumption by showing that overexpression of Parkin (using the UAS/Gal4-system) was able to rescue phenotypes observed in Pink1-deficient flies (Clark et al. 2006; Park et al. 2006).
Modifier screens combine the advantages of forward and reverse genetic screens (St Johnston 2002). Modifier screens require predefined phenotypes that are easily accessible and sensitive to genetic modifications. In the context of neurodegenerative diseases, for example, the expression of a toxic, disease-linked gene product is targeted to the fly eye. This might result in a so-called rough eye phenotype (REP, for a review see Kumar 2012) caused by degeneration of eye-specific cells, e. g. photoreceptors. Usually, these REPs are quite robust and display only small variability. Additionally, the severity of the REP correlates with the degree of cell loss. As the fly eye is easily accessible from the outside, and the enhancement or suppression of photoreceptor loss is reflected by changes in REP appearances, REPs provide an ideal readout for screens. An example of such a screening approach is provided in Fig. 3.
In modifier screens, flies displaying a REP are crossed with flies with either loss-of-function mutations or misexpression of endogenous genes under UAS control. The F1 generation is then screened for obvious changes in REP appearance. Key benefits of such screens are: first, they can be easily and quickly conducted, allowing the screening of large collections of potential modifiers in vivo. Second, by external investigation of the REP, the degree of photoreceptor loss as an indicator for neurotoxicity can be evaluated. Third, epistatic interactions can be revealed and therefore even genes that would not be normally detected in a traditional forward screen may be identified. Modifier screens are the current standard and variations mainly exist in the choice of effector lines. A rough summary of performed screens on neurodegenerative diseases in flies is presented in Table 1. In the following paragraphs, we present some screens in more detail to illustrate the general screening approaches, providing examples of how results from such screens have enriched our knowledge of neurodegenerative diseases.
Table 1. Overview and summary of screens identifying modifiers of neurodegenerative disease pathology in Drosophila
List of modifier screens in neurodegeneration
(a) Protein (b) Screened system
GOF, gain-of-function; LOF, loss-of-function; REP, rough eye phenotype; GMR, glass multimer reporter; Elements, EP, EY and Mae-UAS;6.11 are transposable elements, randomly integrated in the fly genome in which UAS sites facilitate either overexpression or RNAi-mediated silencing of neighboring genes; EMS, ethyl methanesulfonate used for mutagenesis; SCA, Spinocerebellar ataxia; S2 cells, Schneider 2 cells, derived from a primary culture of late stage (20–24 h old) Drosophila melanogaster embryos, likely from a macrophage-like lineage; BG2-c2 cells, cell line derived from central nervous system of 3rd instar larvae; VDRC, Vienna Drosophila RNAi Center; UTR, untranslated region; NIG-Fly, fly stock collection at the National Institute of Genetics, Japan.
Alzheimer's disease (including tauopathies)
(a) Tau [V337M](b) GOF
Readout: REP induced by eye-specific (GMR) expression of Tau [V337M]
Several examples of modifier screens in Drosophila are summarized below. In one of the first modifier screens, a fly line with eye-specific expression of a polyglutamine (polyQ) tract derived from Huntingtin harboring 127 glutamine repeats was generated. These flies were crossed with 7000 de novo-generated autosomal P-element insertion strains. The F1 generation was analyzed for changes in the polyQ-dependent REP, which served as readout for genetic interactions. In this screen, dHDJ1 (Drosophila NH2-terminal J domain with homology to human HSP40/HDJ1) and dTPR2 (Drosophila tetratricopeptide repeat protein 2) were identified as suppressors of toxicity (Kazemi-Esfarjani and Benzer 2000). Shulman and Feany used a very similar approach to identify modifiers of REP generated by Tau-induced toxicity. In their screen, alterations of a Tau-induced REP served as a readout (Shulman and Feany 2003). In contrast to the screen by Kazemi-Esfarjani and colleagues, Shulman and Feany used a collection of roughly 2000 lines containing enhancer/promoter (EP) elements. Inserted in close vicinity to a given gene, UAS sites within the EP element allow gene overexpression or silencing (depending on the orientation of the gene with respect to the EP element) under GAL4 control (Fig. 1). Accordingly, the screen by Shulman and Feany allowed the identification of gain-of-function and loss-of-function modifiers of Tau toxicity. Interestingly, one-third of the modifiers identified in the screen encoded for protein kinases and phosphatases, of which some were shown to phosphorylate Tau in vitro. One of these modifiers was the kinase Par-1. In subsequent studies, Par-1 has been shown to play an initiator role in Tau phosphorylation, triggering additional, temporally ordered phosphorylations of Tau by downstream kinases like Cdk5 and glycogen synthase kinase 3 (GSK-3β). In summary, the sequential phosphorylations result in the generation of toxic Tau species (Nishimura et al. 2004; Chatterjee et al. 2009; Ambegaokar and Jackson 2011). To date, GSK-3β is accepted to be one of the main kinases in Tau phosphorylation. Of note in this context, Jackson and co-workers established a direct link of altered GSK-3β levels and Tau toxicity in vivo. The authors were able to show that GSK-3β overexpression enhanced Tau phosphorylation and toxicity, while reducing GSK-3β levels suppressed Tau toxicity (Jackson et al. 2002).
An elegant variation of the classical modifier screens using EP elements was used by Bilen and Bonini. They performed a genome-wide screen for new modifiers of the REP generated by polyQ-induced toxicity. Initially, they performed a screen with a subset of 2300 available EP-element insertion lines. Furthermore, de novo EP-element insertions were generated and only those novel insertions that modified polyQ toxicity were selected for further analysis (Bilen and Bonini 2007). In continuation of this work, Lessing and Bonini showed that toxicity induced by a truncated form of Ataxin-3 with a polyQ expansion is dependent on normal activity of Ataxin-2. This interaction depends on a conserved protein-interaction motif of Ataxin-2 and binding of cytoplasmic poly(A)-binding protein to this motif. These findings suggest that the normal roles of Ataxin-2 and poly(A)-binding protein regulate translation of specific mRNAs, which are critical to SCA3 disease (Lessing and Bonini 2008). Finally, Voßfeldt and co-workers were the first to perform a large-scale screen using a collection of UAS-shRNA lines (Fig. 3). The question that was addressed by their study was whether or not eye-specific silencing of almost 7000 genes representing human orthologues had an impact on the REP induced by the expression of a toxic Ataxin-3 species. The candidate interactor genes (roughly 500, involved in various cellular processes) obtained in this study constitute a valuable pool for future research on modifiers of genes involved in neurodegenerative disorders (Voßfeldt et al. 2012).
In summary, modifier screens using alterations in REPs induced by eye-specific expression of disease-linked, toxic gene products are valuable tools. In addition to the REP, there are certainly multiple other readout systems to address neurodegeneration and neuronal dysfunction in Drosophila. Some of these readout systems address parameters of fly behavior like locomotion, flight, vision and longevity. Electrophysiological recording of neurons allows direct assessment of neuronal dysfunction (e. g. recording of synaptic transmission in the giant fiber pathway or retinogram). Moreover, histological analysis is frequently used to address neurodegeneration and cell death (e. g. vacuolization of fly brains upon neuron loss). In the context of neurodegenerative diseases, these assays are more significant than the REP, which explains their frequent use in the verification of genetic interactions found in REP-based screens. However, in conjunction with large-scale screen, the REP is the easiest to score, facilitating the fastest readout.
Depending on the disease model analyzed, other readouts might allow for large-scale screens as well. One such example is described below. Eye-specific expression of the well-known PD-linked protein α-Synuclein does not cause a REP. Nevertheless, expression of α-Synuclein in aminergic (serotonergic and dopaminergic) neurons of the fly is detrimental. Depending on the strength of α-Synuclein expression, flies show an age-dependent decline in locomotion and earlier mortality (Butler et al. 2012). Interestingly, dopamine (DA) levels in heads derived from flies with aminergic α-Synuclein expression are reduced, indicating that α-Synuclein causes dysfunction of DA neurons. DA levels in fly heads are easy to determine by HPLC. Butler and coworkers used the α-Synuclein-induced decline in DA as readout to conduct a screen (Butler et al. 2012). One candidate derived from this screening approach was the HSP90-like mitochondrial chaperone TNF receptor-associated protein 1 (TRAP1). The authors showed that reduction of TRAP1 significantly enhanced DA decline and other detrimental effects of α-Synuclein in flies (reduced climbing ability, decline in longevity, loss of DA neurons), while expression of human TRAP1 provided rescue. As an extension of this work, two groups independently showed that TRAP1 (fly and human) is also able to rescue PINK1 but not Parkin loss-of-function phenotypes (Costa et al. 2013; Zhang et al. 2013). These results were confirmed in cultured human cells, emphasizing the conserved role of TRAP1 in detrimental effects of α-Synuclein and its function within the PINK1/Parkin pathway (Butler et al. 2012; Zhang et al. 2013). In summary, the data suggest that mitochondrial TRAP1 is an important factor in α-Synuclein and PINK1-induced PD and that enhancing TRAP1 activity might represent a strategy for therapeutic approaches. The data also tightly link α-Synuclein toxicity to mitochondrial dysfunction, implicating that dominantly inherited forms of PD may involve mitochondrial pathology.
All in all, Drosophila melanogaster is now an established model organism to study human neurodegenerative diseases. The high degree of conservation in molecular pathways between flies and humans has led to the discovery of novel pathomechanisms in disease, which we have sought to emphasize here. In conclusion, future biased and unbiased research using Drosophila will help shed light on disease mechanisms in human neurodegenerative diseases.
This work was funded by the ‘Bundesministerium für Bildung und Forschung’ (‘Nationales Genomforschungsnetz (NGFN+): 01GS08137-5/01GS08137-6a to AV and JBS, and ‘Kompetenznetz Degenerative Demenzen’ (KNDD): 01GI0703/01GI1005C to JBS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. SL and AV are supported by the Alzheimer Forschung Initiative e.V. (project number 12812).