AD (Alzheimer's disease) is a neurodegenerative disorder characterized by the abnormal hyperphosphorylation and aggregation of the microtubule-associated protein tau and the misfolding and deposition of Aβ peptide. The mechanisms by which tau and Aβ become abnormal is not clearly understood, neither is it known what role either protein plays in the neurodegenerative process underlying AD. We have modelled aspects of AD in Drosophila melanogaster to shed light on these processes and to further our understanding of the relationship between tau and amyloid in this disease.
- Alzheimer's disease
- wnt/wingless signalling
Oligomeric and aggregated forms of the Aβ (amyloid β-peptide) and hyperphosphorylated and filamentous forms of the microtubule-associated protein tau are the precursors of neuritic plaques and neurofibrillary tangles, the pathological hallmarks of AD (Alzheimer's disease). Since both Aβ and tau are normal proteins found in healthy neurons, there is a great deal of interest in understanding what triggers their conversion into the abnormal and aggregated states that are found in AD. Similarly, research efforts are directed towards discerning how the pathological alteration of both proteins affects their physiological function (which in the case of Aβ is still not known). The relationship between tau and Aβ is also a matter of debate, with the general consensus view supporting the hypothesis that Aβ pathology lies upstream of tau pathology in AD, but that the latter pathology is responsible for the cognitive impairments that characterize this disease .
Numerous cellular, ex vivo slice and in vivo animal model systems have been established to answer these questions. The earlier models emulated the amyloid pathology by overexpressing mutant forms of APP (amyloid precursor protein) and/or PS (presenilin) mutations in transgenic rodent and invertebrate models (reviewed in [2,3]). After the discovery of familial forms of tauopathies which are linked to mutations in the tau gene, transgenic models were generated which modelled aspects of the tau pathology in rodents (reviewed in ) and invertebrates [5–9]. More recently, transgenic rodent models have been created that model both the amyloid and the tau pathologies by co-expression of mutant APP/PS and FTDP-17 (frontotemporal dementia with parkinsonism linked to chromosome 17) tau genes [10,11].
We have established various Drosophila models of tauopathy in which we express either wild-type or mutant forms of human 0N3R or 0N4R genes alone [5,7,8] or in combination with Aβ40 or Aβ42 . We have used these models both to further our understanding of the mechanisms by which hyperphosphorylated tau mediates neuronal dysfunction and to probe the relationship between the tau and amyloid pathologies.
Using Drosophila to model human disease
Drosophila has been used as a model organism for almost a century and has been pivotal for the elucidation of a plethora of biological processes in this time, ranging from the rules of genetic inheritance  to an understanding of complex processes such as circadian rhythms , learning and memory , development  and aging , to mention a few. Furthermore, the fruitfly genome has been sequenced completely. As a result of all of these studies, there is a wealth of genetic tools that can be employed to study and manipulate almost any cellular process and cell within this organism. The UAS (upstream activating sequence)/GAL4 gene expression system is one example of such a genetic tool that allows for experimental control over both temporal and spatial transgene expression and has thus been described as the fruitfly geneticist's Swiss Army knife . Furthermore, Drosophila are small in size and have a relatively simple well-studied anatomy which allows one to carry out experiments at the level of single easily identifiable cells (such as the well-characterized motor neurons and their neuromuscular junctions ). Moreover, its short life cycle makes it highly amenable to enhancer/suppressor studies  that can shed light on novel modifiers of disease processes. This is invaluable for the study of human diseases such as tauopathies, where such knowledge will pave the way for disease-modifying therapies. Another very important attribute of Drosophila that make it particularly useful for studying chronic and progressive human neurodegenerative diseases is the transparent cuticle of the larvae which allows the study of the progression of the disease process, and its impact on neuronal function, non-invasively in living intact animals , . All of these attributes of Drosophila make it one of the most experimentally tractable model organisms. The fact that many insights into human diseases such as Huntington's disease , Parkinson's disease  and AD have come from Drosophila models of these diseases testifies to the usefulness of this experimental system for elucidating the mechanisms that underpin neurodegeneration (reviewed in [24,25]).
We have employed the UAS/GAL4 system to induce expression of human tau in either motor neurons alone or in all neurons of Drosophila. We have then carried out studies to investigate the effects of highly phosphorylated human tau on aspects of neuronal function such as axonal transport , microtubular cytoskeletal integrity  and synaptic structure and function . In some instances, these studies were carried out in living intact animals, thus they provided insights into the disease processes as they unfolded in vivo.
More recently we co-expressed the human tau with Aβ40 and Aβ42 to ascertain whether or not the presence of the amyloid pathology in the same neurons as the highly phosphorylated tau influences the tau-mediated neuronal dysfunction .
Soluble highly phosphorylated tau disrupts neuronal and synaptic function
Since the realization that hyperphosphorylated tau is the main constituent of neurofibrillary tangle pathology in AD  the pathological significance of tau hyperphosphorylation and aggregation has been vigorously investigated. It has been hypothesized that tau phosphorylation leads to a loss of tau function (tau–microtubule hypothesis, reviewed in ), whereas tau aggregation may lead to a toxic gain-of-function . Since tau hyperphosphorylation is evident before filament formation in post-mortem AD-affected brains , it has been hypothesized that hyperphosphorylation primes tau for aggregation . Thus it is conceivable that a pathological cascade unfolds in tauopathies in which hyperphosphorylation of tau is stimulated in the early stages and this in turn leads to tau aggregation and tangle formation in late stages. Neuronal dysfunction, mediated by highly phosphorylated tau, may characterize the early stages, whereas neuronal death mediated by toxic tau filaments may be a feature of the late stages of disease.
One aim of our research is to test this hypothesis in Drosophila. We employ Drosophila models of tauopathy (i) to study the neuronal consequences of inducing tau hyperphosphorylation and filament formation, and (ii) to investigate the causal relationship between the development of both tau aberrations. By investigating the neuronal consequences of inducing tau hyperphosphorylation, we have effectively been testing the tau–microtubule hypothesis in vivo. This is a long-standing hypothesis that is based primarily on evidence obtained from in vitro observations, and our studies in Drosophila are the first to test all aspects of this hypothesis in one in vivo experimental paradigm (reviewed in ). We have shown that, following expression of highly phosphorylated human 0N3R tau in Drosophila motor neurons, there is a profound disruption of axonal transport, leading to significant synaptic dysfunction and behavioural impairments ([7,8] and reviewed in ). More recently, we have shown that the molecular mechanism by which highly phosphorylated human tau mediates this neuronal dysfunction is 2-fold: first, as predicted by the tau–microtubule hypothesis, highly phosphorylated tau is unable to bind to microtubules effectively and thus, in the presence of human hyperphosphorylated tau, the microtubule cytoskeleton becomes destabilized and disorganized . Secondly, this highly phosphorylated human tau exhibits another pathological property, one not predicted by the tau–microtubule hypothesis: it binds to and sequesters endogenous Drosophila tau, compromising its microtubule-binding capacity . Both the destabilization of the cytoskeleton and the sequestration of normal endogenous Drosophila tau could be reversed by treatment of the human-tau-expressing larvae with LiCl, a drug that has been shown to be capable of suppressing tau phosphorylation . These results demonstrate that highly phosphorylated tau is pathogenic and exhibits both a loss of normal function and also a gain of toxic function. Furthermore, the highly phosphorylated tau does not need to aggregate to mediate these effects because tau filament formation was not evident when disruptions to microtubular integrity or binding to Drosophila tau were seen. Moreover, although the human-tau-expressing neurons exhibited all of these structural and functional aberrations, there was no evidence of neuronal loss or overt neurodegeneration. Thus soluble hyperphosphorylated tau proteins evident in neurons before tangle formation may cause significant neuronal dysfunction and this may be responsible for the early symptoms of dementia in tauopathies.
Aβ42 exacerbates tau-mediated neuronal dysfunction
The relationship between the tau and amyloid pathologies in AD has been researched intensively and is highly debated. The finding that mutations in the tau gene cause familial forms of tauopathies (such as familial forms of FTDP-17) wherein tangle pathology is evident in the absence of amyloid pathology suggests that tangle pathology lies downstream of amyloid pathology in AD. This is supported by studies in rodent models that exhibit both the tau and amyloid pathologies in which the co-existence of the amyloid pathology accelerates and exacerbates the emergence of the tangle pathology . Moreover, in these models, it has been noted by many that the accumulation of oligomeric Aβ precedes the induction of the hyperphosphorylated tau [10,33]. These findings imply that oligomeric Aβ42 can stimulate hyperphosphorylation and accumulation of tau. Indeed, results from various in vitro and in vivo studies lend support to this hypothesis: exposure of primary cells in culture to oligomeric Aβ42 can induce phosphorylation of tau [34–36], and intracerebroventricular infusion of Aβ oligomers into rodent brains also stimulates tau phosphorylation .
We were interested in investigating whether co-expression of amyloid with human tau in our Drosophila model of tauopathy induced tau hyperphosphorylation and exacerbated the tau phenotypes. We demonstrated that expression of Aβ42 with human 0N4R tau in this model resulted in greater phosphorylation of the human tau at some (AT8, AT180) but not all (PHF-1) sites . Furthermore, in the presence of Aβ42, tau-mediated phenotypes in both larval and adult fruitflies were significantly exacerbated (Figure 1) . In larvae expressing either Aβ42 alone or human tau alone within their motor neurons, there were conspicuous morphological abnormalities in the neuromuscular junctions . This is to be expected since both proteins have been shown to cause structural and functional dysfunction within synapses in various models of AD [8,38,39]. It was therefore not surprising that when both proteins were co-expressed in larval motor neurons, there was an additive exacerbation of the NMJ (neuromuscular junction) phenotype induced by expression of either protein alone . In contrast, expression of Aβ42 in larval motor neurons did not result in significant axonal transport disruption, and yet co-expression of Aβ with human tau resulted in a 3-fold exacerbation of the disruption caused by expression of tau alone (Figure 1A) . Similarly, pan-neuronal expression of Aβ42 in adults did not induce significant locomotor impairment in an adult climbing assay, whereas expression of human 0N4R tau led to a progressive age-dependent impairment which was potentiated severalfold by co-expression of Aβ42 (Figure 1B) . These results demonstrate that, even when Aβ42 does not by itself cause detectable neuronal dysfunction, it significantly exacerbates tau-mediated neuronal dysfunction.
In attempting to unravel the mechanism by which Aβ interacted with tau in this model, we assessed the activity of the tau kinase, GSK3β (glycogen synthase kinase 3β), in Drosophila expressing either protein alone compared with those expressing both. We hypothesized that Aβ exacerbated the tau phenotypes by activating GSK3β because several reports have demonstrated that exposure to cells in culture to oligomeric Aβ results in increased activity of GSK3β as well as tau hyperphosphorylation [35,36,40]. In line with this, we found that treatment of tau/Aβ42 co-expressing larvae to LiCl, a well-known GSK3 inhibitor, ameliorated the Aβ42 exacerbation of the tau phenotype (Figure 1C) . However, when we examined the actual activity of GSK3β in these tau/Aβ42 bigenic fruitflies using antibodies against active and inactive phospho-epitopes of GSK3β, we found no difference between any of the lines . These results suggest that although GSK3β may be the kinase that mediates the interaction between tau and amyloid in this model, this does not occur via an Aβ42-induced increased enzymatic activity of GSK3β.
Apart from phosphorylation to activate or inhibit GSK3β, compartmentalization with its substrates is another cellular mechanism by which its activity can be regulated. Such a mode of regulation of GSK3β activity is utilized in the wnt or wingless signalling pathway which plays a pivotal role in determining cell fate during development. In the absence of a wnt signal, GSK3β forms a complex with its substrate β-catenin and phosphorylates it, targeting it for degradation. Upon activation of frizzled receptors by wnt ligands, an intracellular effector protein dishevelled (Dsh) is recruited to destabilize the GSK3β–β-catenin complex, resulting in reduced phoshorylation and degradation of β-catenin. Apart from β-catenin, GSK3β has also been shown to form a complex with tau  and it has been demonstrated that components of the wnt signalling pathway can suppress its phosphorylation of tau in vitro [42,43], presumably by disrupting the complex it forms with tau as it does with β-catenin.
It has been shown by many that Aβ42 can interact with components of the wnt signalling pathway and antagonize its effects [44–46]. We therefore hypothesized that Aβ42 could enhance GSK3β-mediated phosphorylation of tau by suppressing endogenous wingless signalling in the tau/Aβ42 bigenic fruitflies. We thus predicted that up-regulating wingless signalling by overexpressing Dsh in tau-expressing Drosophila should have the opposite effect to Aβ42 co-expression, and lead to a suppression of the tau phenotypes. Indeed, in transgenic animals that co-express both tau and Dsh, there is a significantly improved axonal transport and locomotor behaviour and a concomitant reduction of tau phosphorylation . Thus it is possible to speculate that GSK3β activity towards its substrates such as tau are regulated by a low level of wingless signalling in the adult brain and that if this becomes deregulated (such as may happen following Aβ accumulation in the aged brain), tau hyperphosphorylation ensues, leading to neuronal dysfunction and cognitive decline (Figure 2). Thus wingless signalling may be the link between the tau and amyloid pathologies although components of this developmental signalling pathway need to be demonstrated in the adult brain to validate this idea.
We have modelled aspects of AD in Drosophila to further our understanding of the mechanisms by which hyperphosphorylated tau and oligomeric Aβ42 disrupt neuronal function in vivo. Using these Drosophila models, we have been able to gain an insight into these pathological mechanisms as they unfolded in living intact animals. The genetic tractability of Drosophila allowed for tight spatial and temporal control over the human tau and amyloid expression and thus enabled us to conduct these studies at the level of individual easily identifiable neurons as well as facilitating the functional consequences at the organismal level.
From these studies, we can conclude that hyperphoshorylated human tau exhibits a profound loss of function (leading to destabilization of the microtubule cytoskeleton and a disruption of axonal transport) as well as an unexpected gain of toxic function (by sequestering normal tau and functionally compromising it). Furthermore, aggregation of tau is not required for it to mediate these effects. Our studies also indicate that Aβ42 acts synergistically with human tau to exacerbate its pathological effects. Although the cellular process that facilitates the interaction between tau and amyloid is unclear, our results point to the wingless signalling cascade as a potential pathway that may link the two proteins in AD.
Our studies highlight the utility of Drosophila models for enhancing our understanding of the pathological mechanisms that underpin human neurodegenerative diseases.
This work was supported by the Alzheimer's Society UK and the Wessex Medical Trust.
The Biology of Tau and its Role in Tauopathies: A Biochemical Society Focused Meeting held at Robinson College, Cambridge, U.K., 7–8 January 2010. Organized and Edited by Amrit Mudher (Southampton, U.K.) and Makis Skoulakis (Alexander Fleming Research Center, Greece).
Abbreviations: Aβ, amyloid β-peptide; AD, Alzheimer's disease; APP, amyloid precursor protein; FTDP-17, frontotemporal dementia with parkinsonism linked to chromosome 17; GSK3β, glycogen synthase kinase 3β; PS, presenilin; UAS, upstream activating sequence
- © The Authors Journal compilation © 2010 Biochemical Society