Introduction

The success of biomedical research in the past few decades has provided the medical community with valuable information on numerous human diseases, including cancers and heart disease. Whereas human neurodegenerative diseases are among those most prominent, the application of genetics in simple model systems to address mechanisms of these diseases has only recently received attention. A number of major breakthroughs have been made in the last decade on these human diseases including the identification of a number of disease-related genes involved in Alzheimer's, Parkinson's and Huntington's diseases. The identification of the Huntington's Disease gene is the classical example of positional cloning of a human disease gene.1,2 With the cloning of such genes, pathogenic mechanisms of the diseases can then be addressed.3

Studies in model organisms have been found to be invaluable in elucidating the cellular and molecular basis of normal cellular processes, and what can go awry in disease pathogenesis. For example, transgenic and knock-out mouse models are a powerful approach to elucidate the molecular genetic basis of disease progression.4 However, genetic manipulations in the mouse are costly and time consuming, providing a need for even simpler systems with a greater array of genetic approaches that provide a faster time course. Such readily available systems that have been applied to elucidate molecular mechanisms of human neurodegenerative diseases include yeast, C. elegans and Drosophila (Figure 1). Genes between flies and humans are highly conserved,5 and Drosophila has a complex nervous system and displays complex behaviours such as learning and memory, making it a particularly attractive system for study of neuronal dysfunction and loss that proceeds from neurodegenerative disease. Here, we emphasize advances toward the understanding of select human neurodegenerative disorders through the use of Drosophila genetics. Other recent reviews emphasize additional aspects of cellular loss and degeneration in fly mutants and technical approaches,6,7,8 including other reviews of this issue.

Figure 1
figure 1

Approaches to addressing gene function in Drosophila. Function of fly genes homologous to human disease genes can be studied through classical mutagenesis, as well as transgenic and reverse genetic approaches (‘knock-in’/‘knock-out’ technology). Transgenic models of disease allow mechanisms of disease progression to be addressed using the wealth of techniques available in Drosophila. Genetic screens allow the identification of genes that influence disease pathogenesis through modifier screens for enhancer and suppressor mutations

Full size image

Alzheimer's disease

Alzheimer's disease (AD) is one of the most common degenerative brain diseases, affecting 11% of the population over 65 years of age and 50% over the age of 85 (for reviews, please refer to9,10). AD disease brain tissue displays several unique pathological hallmarks including senile neuritic plaques and neurofibrillary tangles. Senile neuritic plaques are extracellular deposits consisting of β-amyloid peptides, while neurofibrillary tangles are cytoplasmic aggregates composed of the paired helical filaments of hyperphosphorylated tau (a brain-specific microtubule-associated protein). These abnormal deposits form primarily in brain regions that are essential for cognition and memory, with AD patients typified by dementia.

A genetic basis for AD was elucidated almost a decade ago after linkage between familial AD patients and mutations in the β-amyloid precursor protein (APP) gene was found. The APP gene encodes a type I integral membrane protein (770 amino acids in length) that is susceptible to endoproteolytic cleavage through proteolytic pathways of α-secretase and β-secretase. Cleavage of APP results in a 100–120 kDa N-terminal extracellular fragment and a 10–12 kDa membrane bound fragment. The membrane bound fragment is further cleaved by γ-secretase and leads to the secretion of the p3 (from the α-secretase pathway) or the Aβ40 (from the β-secretase pathway) fragment (for review, see11). Missense mutations in the APP gene identified in familial AD (FAD) patients appear to favour the production of an amyloidogenic Aβ42 peptide instead of Aβ4012; Aβ42 is prone to aggregation and is the main constituent of the senile neuritic plaques.

Another group of mutations associated with FAD patients falls into the Presenilin genes (PSN1 and (PSN213). Presenilin proteins possess eight proposed transmembrane domains and a large hydrophilic loop between transmembrane domains (TM) 6 and 7. Most Presenilin mutations are located around the TM domain encoding regions.14 Presenilin is suggested to be the γ-secretase15 or in close association with the γ-secretase, suggesting a mechanism by which altered Presenilin activity can lead to an increase in production of Aβ42.

Functions of APP in flies

Although the processing mechanism of APP has been studied extensively, the function of the APP protein and role of processing still remain elusive. Drosophila has an APP-like (APPL) orthologue,16 allowing functional analysis of the protein in the context of a developing nervous system (Table 1). Expression of the Appl gene in flies is observed mainly in the nervous system, starting from mid-embryogenesis through adulthood.17 At the protein level, APPL is first synthesized as a 145 kDa membrane-associated protein, which is then rapidly cleaved into a 130 kDa soluble fragment that lacks the C-terminal domain.18

Table 1 Neurodegenerative genes in Drosophila
Full size table

Flies that are homozygous for Appl null mutations are viable and display no gross morphological defects, suggesting that Appl is not essential for viability. However, behavioural defects in phototaxis are detected.19 These defects can be rescued by directed expression of the Appl transgene, but not by a mutant form of the protein that is unable to produce the cleaved 130 kDa soluble APPL domain. This suggests that the cleavage event is at least in part essential for the normal cellular function of APPL. Importantly, behavioural defects of Appl mutants can also be rescued by a human APP transgene, suggesting that human APP and Drosophila APPL are at least partly functionally interchangeable. Moreover, upon expression in flies of a human APP transgene that contains only the integral membrane and C-terminal domains (including the Aβ moiety) in flies, Aβ40 is detected by Western blot, suggesting that Drosophila possesses functional APP processing machinery.20 This finding provides the foundation for genetic approaches in Drosophila to elucidate additional details of APP processing.

In addition to expression in the central nervous system, APPL protein is detected at the neuromuscular junction (NMJ) and its accumulation is dependent on synaptic activity.21 When synaptic activity is enhanced, APPL is found to be densely localized at the NMJ. However, in the absence of APPL, the number of synaptic boutons (the interface between motor neurons and muscles) is reduced. In contrast, when APPL is over-expressed at the NMJ, extra numbers of boutons with abnormal appearance are observed.21 These results suggest that localization of APPL may be an activity-dependent event, and that APPL is also involved in the regulation of synaptic morphology. Another role of APPL in vesicular trafficking is revealed by monitoring axonal transport of a synaptic protein, synaptotagmin.22 Normally, synaptotagmin is transported through motor axons and localized at the NMJ. Upon over-expression of APPL, retention of synaptotagmin along motor axons is observed, suggesting a disruption of normal axonal transport.

The role of Presenilin in Drosophila

A critical aspect of AD research is to elucidate mechanisms of APP processing and identify genes that are involved in the production of the amyloidogenic Aβ42 peptide. Analogous to APP processing, Presenilins have recently been shown to be involved in proteolytic processing of Notch–a role revealed by study of Presenilin function in Drosophila.23,24,25,26 A role of Presenilin genes in Notch pathways was first revealed by studies in C. elegans of the analogous lin-12 pathway.27 In Drosophila, Presenilin mutants display a partial Notch-like phenotype23 and mutations in Presenilin enhance dominant Notch phenotypes,25 suggesting that Presenilin and Notch are involved in the same pathway in flies. Directed expression of Presenilin to the fly eye causes disruption of the normal eye structure, due to death of cells through apoptotic pathways. The phenotype can be suppressed by either up-regulation of the Notch pathway or by co-expression of the viral anti-apoptotic gene P35. Compared to wild-type Presenilin, mutations of Presenilin analogous to those observed in human FAD show less activity than normal, suggesting that FAD mutations in Presenilin are loss-of-function mutations.28,29

The Drosophila Presenilin protein shows ∼50% sequence identity to its human counterpart, and like human Presenilins, undergoes proteolytic cleavage to give rise to a 30–35 kDa N-terminal and 25 kDa C-terminal fragments.25,30 Two aspartate residues (one from each cleaved fragment), that are thought to be critically involved in γ-secretase function of human Presenilins, are also conserved in the fly protein.

Parkinson's disease

Parkinson's disease (PD) is the most common movement disorder, affecting approximately 1 million people in the United States.31 This disease is mostly sporadic and typically affects people that are between 50 and 60 years of age. A small fraction of PD cases have been linked to mutations in specific genes, including the α-synuclein, parkin and ubiquitin C-terminal hydrolase L1 (UCHL1) genes.32,33,34,35 Cytoplasmic aggregates called Lewy bodies are usually found in the substantia nigra of brain tissue, except for a few variant forms of PD. Dopaminergic neurons are the type of nerve cells that are most susceptible to degeneration in PD. Not only have mutations in the α-synuclein gene been linked to PD, α-synuclein protein is the major building block of Lewy bodies36 in both familial and sporadic PD, suggesting similar mechanisms of disease pathology.

α-Synuclein is a small soluble protein (140 amino acids in length) that contains six N-terminal degenerate KTKEGV repeats and an acidic C-terminal domain. The protein is transported via two axonal transport mechanisms in neurons (fast and slow components37). The first four N-terminal imperfect repeats have been demonstrated to be essential for axonal transport of α-synuclein.37 Under defined conditions in vitro, α-synuclein is able to form insoluble filaments with β-sheet structure.38,39 Mutations in α-synuclein identified in familial PD patients are A53T and A30P.32,33 In vitro, the A53T mutant protein displays a higher rate of insoluble filament assembly compared to wild-type α-synuclein,38 whereas A30P impairs the ability of α-synuclein to bind to brain vesicles in rat brain.37 Lewy body formation may be achieved by enhancing the rate of β-sheet insoluble filament assembly through α-synuclein mutation, sporadic effects, or defective axonal transport of the protein causing local accumulation. Since wild-type α-synuclein protein forms insoluble filaments (Lewy bodies) in sporadic PD in the absence of α-synuclein mutation, this suggests that other yet-to-be-defined genetic factors40 as well as environmental factors, for example cellular redox conditions,41 are likely to contribute to disease pathogenesis.

Modelling PD in flies

Expression of human α-synuclein in flies does not cause any gross morphological changes, but leads to selective loss of dopaminergic neurons in the adult brain over time.42 The cellular loss displays specificity in that serotonic neurons, in contrast to dopaminergic neurons, are unaffected. Loss of dopaminergic neurons in aged flies is observed upon expression of normal as well as mutant forms of α-synuclein (A30P and A53T), thus displaying no selective characteristics of the mutant forms of the protein in flies in vivo.

Transgenic flies expressing α-synuclein also display accumulation of protein in cytoplasmic inclusions. The aggregates are observed in a sequential manner, with diffuse and cytoplasmic staining present in young brain tissue and protein accumulations noted at later stages, prior to neuronal loss. These cytoplasmic inclusions appear to be present in a more widespread manner than only dopaminergic neurons. A lack of spatial correlation between Lewy body formation and loss of dopaminergic neurons is also observed in patients with diffuse Lewy body disease. This possible selective vulnerability of dopaminergic neurons to human α-synuclein in flies may reflect a special property of α-synuclein in dopaminergic neurons or merely sensitivity of dopaminergic neurons to ectopic insults. Pan-neural expression of α-synuclein (wild-type, A53T and A30P mutants) results in premature loss of climbing ability, suggesting that transgenic expression of synuclein impairs locomotor ability in flies. A link between these behavioural defects and loss of dopaminergic neurons remains to be established.

Human polyglutamine disease

Numerous cellular proteins contain stretches of repeats of the amino acid glutamine. A class of human disease has been found to be associated with expansion of such polyglutamine domains. To date, at least eight human neurological diseases are caused by polyglutamine expansion (typically from a normal range of 4–36 residues to a pathogenic range of 36–306 residues, with the exception of SCA6) in the respective cellular proteins.3 These diseases are dominantly inherited (except for spinobulbar muscular atrophy which is X-linked), suggesting that the expanded polyglutamine domain confers a novel toxic property on the cellular proteins. Moreover, expanded polyglutamine proteins form aggregates, typically in the form of nuclear inclusions (NIs). Except for the polyglutamine, no other sequence homology is found between these different human disease proteins; most of them are novel proteins with currently unknown cellular functions.

Nuclear inclusions have been proposed to be the cause of neurotoxicity of polyglutamine diseases, however a direct relationship between aggregate formation and neurotoxicity is still lacking. Other models have also been suggested to explain toxicity of the polyglutamine proteins.43 Aggregation of mutant proteins is clearly a characteristic of the disease proteins, indicating that these abnormal aggregates, whereas perhaps not causal, reflect a novel property gained by the mutant protein.

Drosophila models for human polyglutamine disease

Polyglutamine-induced neurodegeneration has been re-capitulated in model organisms using transgenic techniques.44 In Drosophila, two disease models have been generated45,46 (Figure 2), as well as models using pure polyglutamine domains or an expanded polyglutamine domain inserted into a normal cellular protein.47,48 Polyglutamine disease models in flies include an Machado-Joseph Disease (MJD) disease model and a Huntington's Disease model. By expressing a truncated version of the pathogenic human MJD gene (also called SCA3, spinocerebellar ataxia type 3) in the fly, Warrick et al.46 observed late-onset progressive degeneration. Using a truncated form of the human Huntington's disease gene, Jackson et al.45 directed expression to the fly eye, showing progressive loss of photoreceptor neurons. In humans, severity of disease is proportional to the length of the expanded polyglutamine run, with longer expansions causing an earlier onset, more severe disease. In Drosophila, degeneration is also observed to be earlier onset and more severe with a run of 102 glutamines compared to 75 glutamines, although a quantitative comparison of the expression levels between the two transgenes was not reported.45 Co-expression of mutant disease proteins with the viral anti-apoptotic gene P35, has little or no effect on degeneration, suggesting that polyglutamine-induced cell loss appears not to be greatly influenced by P35-inhibitable cell death pathways.

Figure 2
figure 2

A model for human neurodegenerative disease in Drosophila. (A) Left is a fly expressing a human mutant disease gene for the neurodegenerative disease Machado-Joseph disease in the fly eye. The mutant protein induces severe degeneration, illustrated by the lack of pigmentation and collapse of the eye. Right is a fly expressing the same human mutant disease gene, but now also expressing a human gene encoding the molecular chaperone Hsp70. Although the disease protein is still present, neurodegeneration does not occur, demonstrating the power of Drosophila to elucidate potential suppressor mechanisms to prevent neurodegenerative disease. (B) Abnormal aggregates formed upon expression of the human mutant disease protein in developing eye tissue. Many different human neurodegenerative diseases are associated with abnormal protein aggregates, including AD, PD and polyglutamine disease. As illustrated here, not only the neurodegenerative phenotype, but also abnormal protein aggregation, is modelled in Drosophila. Arrow marks morphogenetic furrow

Full size image

Both Kazemi-Esfarjani and Benzer,47 and Marsh et al.48 have shown that merely a run of polyglutamine peptide is sufficient to induce degeneration in the fly nervous system. Marsh et al.48 showed that addition of non-polyglutamine sequence at the ends of a polyglutamine run reduces the toxicity of the peptides, confirming that the toxicity of polyglutamine-induced neurodegeneration is, at least in part, dependent upon protein context. The authors also expanded a polyglutamine domain from 28 to 108 residues within an endogenous fly protein, Dishevelled (Dsh), and expressed it via endogenous dsh promoter elements. When compared to the activity of normal Dsh protein, the ‘expanded Dsh’ partially rescued the dsh mutant phenotype, suggesting that polyglutamine expansion does not completely abolish normal protein activity, but that some degree of normal cellular function is retained despite polyglutamine expansion.

Using fly genetics to elucidate mechanisms of toxicity and prevent neurodegeneration

The power of modelling human neurodegenerative disease in Drosophila is to pioneer ways to understand degenerative mechanisms and effect suppression. Toward this end, molecular chaperones have been revealed as powerful suppressors of neurodegeneration in Drosophila. Heat shock proteins (Hsps) comprise a large group of proteins that are rapidly induced during stress conditions. Induced Hsps serve different functions, including assistance in the folding of misfolded proteins and degradation of irreparably damaged proteins.49,50 Hsp70 is the main stress-induced molecular chaperone in Drosophila. By expressing a human Hsp70 transgene in flies, Warrick et al51 were able to suppress neurodegeneration in the Drosophila MJD disease model (Figure 2). Moreover, expression of a dominant-negative form of constitutive Hsp70 enhanced degeneration, suggesting a central role of Hsp70 in disease progression. Interestingly, in the situation when polyglutamine toxicity is ameliorated, the protein remains aggregated as visualized by immunohistochemistry, suggesting that reduced toxicity is not correlated with gross loss of NIs.51

Kazemi-Esfarjani and Benzer47 found additional suppressors through a mis-expression screen, looking for modifiers of the eye degenerative phenotype caused by ectopic expression of a pathogenic polyglutamine peptide. Two suppressors were defined that belong to a second class of chaperone family. One of them is a fly Hsp40 orthologue dHdj1. Hsp40 proteins assist Hsp70 in protein folding by stimulating the ATPase activity and substrate binding of Hsp70.52 Up-regulation of dHdj1 or other J-domain containing proteins may therefore increase the efficiency of the protein folding function of Hsp70 and hence modulate folding of polyglutamine proteins. These studies illustrate the power of Drosophila genetics to elucidate potential mechanisms of suppression of neurodegeneration.

The post-genomic era of biomedical Drosophila research

With the recent completion of the Drosophila genomic sequence, comparative genomic analysis has led to the discovery of several previously unidentified disease-related genes including orthologues of human SCA2 (the gene mutated in spinocerebellar ataxia type 2) and parkin (juvenile onset parkinsonism).5 Both classical and reverse Drosophila genetics have proven to be invaluable tools in elucidating underlying mechanisms of many cellular processes, as illustrated here. Moreover, with the recent establishment of a gene targeting technique in flies,53 it should be possible to target endogenous wild-type copies of ‘disease genes’ in the fly genome for inactivation (knock-out); defined mutations can also be ‘engineered’ (knock-in) into respective endogenous genes, e.g. extra glutamine residues in Sca2, huntingtin, to create gain-of-function models. Potentially, human disease genes can also be used to replace endogenous copies of their Drosophila orthologues, allowing expression of the human disease gene under the control of endogenous enhancer/promoter elements of the fly.

Gene expression profiles in disease conditions are always different from normal conditions.54 Monitoring changes of gene expression profiles in disease and non-disease conditions may lead us further toward the understanding of molecular mechanisms of disease pathogenesis. With the whole ‘dictionary of the fruit fly’ in hand, genes that have altered expression patterns can be determined in vivo through the use of DNA microarray analysis.55 As a step forward, in vivo analysis of gene expression profiles from a suppressed (e.g. Hsp70 suppression) or enhanced condition, or other more complicated genetic scenarios, will provide us with additional handles into the molecular mechanisms of disease pathogenesis in a whole organism.

For almost a century, fruit flies have been providing a useful tool to study various different subjects: from the chemical basis of mutagenesis, to the definition of genes; from developmental biology, to animal behaviour. The ability to use Drosophila as a powerful tool to approach pathogenic disease mechanisms of human diseases speaks to a tremendous application in biomedical research.