Animal Models of Neurodegenerative Diseases 55 Table 1 Amyloid Transgenic Rodent Models of AD Pathology Name Gene Mutation Promoter Amyloid Deposits (DP, AP) P-Tau NFT Neuronal loss Memory deficits Inflammatory reaction NSEAPP mice APP 751 NSE DP (8 M) No No No Yes Rare (5%) PDAPP mice APP V717F PDGFβ AP (6–8 M) Yes (18 M) No No Yes Yes Tg2576 mice APP 695(K670N,M671L) PrP AP (9–11 M) Yes No No Yes Yes APP23 mice APP 751(K670N,M671L) Thy-1 AP (6 M) CAA (6 M) Yes No 14% (14–18 M) Yes Yes TgCRND8 mice APP 695(K670N,M671L, V717F) PrP AP (3 M) NR No NR Yes Yes J20 mice APP K670N,M671L, V717F PDGFβ AP NR No NR Yes No PSAPP mice APP 695(K670N,M671L) PS-1 M146L PrP PDGFβ AP (6 M) Yes NR No Yes Yes TgCRND8/PS-1 mice APP 695(K670N,M671L, V717F) PS-1 PrP AP (1 M) Yes NR NR Yes Yes APP SWE /PS-1 dE9 mice APP 695(K595N,M596L) PS-1 dE9 PrP PrP AP (4–6 M) CAA (6 M) NR No Monoaminergic (50%) Yes Yes APP SL /PS-1 mice APP 751(K670N,M671L and V717I) PS-1 M146L Thy-1 HMG AP (6 M) Yes (4 M) No 35% (17 M) Yes Yes APP/PS-1 KI mice APP 751(K670N,M671L and V717I) PS-1 M233T,L235P Knock-in Thy-1 PS-1 mouse AP (6 M) Yes (11 M) NR 50% (CA1, 6 M) 44% (DG, 12 M) Yes Yes APP rat APP 695(K670N,M671L) PrP No No No No Yes No PSAPP rat APP 695(K670N,M671L) PS-1 M146V PDGFβ Rat synapsin 1 and human PS-1 AP (9–11 M) CAA () Yes (19–22 M) No No Yes Yes AP: amyloid neuritic plaques; CAA: cerebral amyloid angiopathy; DG: Dentate gyrus; DP: diffuse (preamyloid) plaques; NFT: neurofibrillar tangles; NR: not reported; P-Tau: hyperphosphorylated MAPT tau protein. 56 I. Ghorayeb et al. evident except in the model APP SL /PS-1 M146L mice developed by Blanchard et al. (2003) where a loss (35%) of neurons in the pyramidal cell layer of the hippocampus was seen at 17 months of age (Schmitz et al., 2004). Recently, an intense sub- cortical monoaminergic neurodegeneration (50% neuronal loss) was observed in APP SWE /PS-1 dE9 (Liu et al., 2008b). It is to be noted that none of these mouse models showed any NFTs. However, many studies described a behavioural pheno- type in various APP/PS-1 transgenic mice (Higgins and Jacobsen 2003; Janus and Westaway 2001; Reiserer et al., 2007; Savonenko et al., 2005). In particular, the performance on the Y-maze, that measures spatial working memory, was impaired before amyloid deposits in PSAPP mice (Holcomb et al., 1999). Taken together, these findings show that it is difficult to obtain a mouse model reproducing AD perfectly, especially with both the neuronal loss and NFTs. However, Casas et al. (2004) produced a new model with many features of AD, the APP SL /PS-1 knock-in mice. These transgenic mice have two mutations in the human APP gene at K670N/M671L and V717I sites corresponding to β- and γ-secretase sites, respectively. In addition, their endogenous ps-1 gene carries the M233T and L235P mutations known to be linked to very early onset FAD at 29 and 35 years of age, respectively. These mice displayed a massive neuronal loss (49% in the ten-month-old APP SL /PS-1 KI mice) in the CA1 region of the hippocam- pus with an intense neuronal apoptosis (Casas et al., 2004; Page et al., 2006). This neuronal loss distribution closely parallels the strong intraneuronal Aβ immunos- taining and intracellular thioflavine-S-positive material but does not correlate with extracellular deposits (Christensen et al., 2008b). Furthermore, the authors also described a loss of neurons (44%) in the dendate gyrus granule layer (Cotel et al., 2008). The APP SL /PS-1 KI mouse model exhibits early robust brain and spinal cord axonal degeneration (Wirths et al., 2007, 2006). At the same time-point, a dra- matic age-dependent reduced ability to perform working memory and motor tasks is observed. These mice are smaller and show development of a thoracolumbar kypho- sis, together with an incremental loss of body weight (Wirths et al., 2008b). Onset of the observed behavioural alterations correlates well with robust axonal degeneration in brain and spinal cord and with abundant hippocampal CA1 neuron loss (Bayer and Wirths 2008). Although our group detected hyperphosphorylated tau protein in cell bodies of neurons in 11-month-old APP SL /PS-1 KI mice (unpublished data obtained by G Page), NFTs were not reported yet. Contrary to studies showing a minor loss of cholinergic interneurones in the motor cortex of APP SWE /PS-1 (deltaE9) mice (Perez et al., 2007), the APP SL /PS-1 KI mouse model shows a loss of choline acetyl transferase-positive neurons only in the motor nuclei Mo5 (motor trigeminal nucleus) and 7 N (facial motor nucleus) accumulating various intracellular Aβ species (Christensen et al., 2008a). The cholinergic forebrain complex consisting of Ch1-4 showed no Aβ pathology, with neither extracellular Aβ plaque deposition, nor intracellular accumulation of Aβ peptides. These fibres from this region displayed swollen ChAT-positive dystrophic neurites surrounding Aβ plaques in the cortex and hippocampal formation. Another neuropathological alteration is the inflammatory processes, such as microglial activation and astrocyte reactivity, that occurs early during the course of Animal Models of Neurodegenerative Diseases 57 the disease (Eikelenboom et al., 2006). At the age of six months, the APP SL /PS-1 KI mouse model upregulates different astro- and microglia markers in both brain and spinal cord including GFAP, cathepsin D, members of the Toll-like receptors family, TGFβ-1, and osteopontin (Casas et al., 2004; Damjanac et al., 2007; Wirths et al., 2008a). Another interesting feature is the occurrence of ganglioside alterations and an accumulation of ceramide species in the cerebral cortex of APP SL /PS1 KI mice as it was shown in human AD brains (Barrier et al., 2007, 2008). As early as three months, these lipid alterations were increased and could be linked to the massive neuronal death observed at sixmonths (Table 1). APP/BACE Mice The type I transmembrane aspartyl proteinase β-site APP cleaving enzyme (BACE1) was identified as the major β-secretase for generation of Aβ peptides by neurons (Luo et al., 2003). BACE cleaves APP at Asp1 and Glu11, whereas subsequent cleavage by γ-secretase gives rise to the Aβ (1–40/42) and Aβ (11–40/42) amyloid peptides. Deficiency of BACE1 in a double transgenic combination with human mutant APP rescued the early hippocampal memory deficits and correlated with a dramatic reduction in Aβ levels (Ohno et al., 2004). On the other hand, mice overexpressing BACE1 in addition to human wild-type (WT), APP, or mutant APP increased the amyloidogenic processing of APP as revealed by increased levels of the APP metabolites sAPPβ, β-CTF, and Aβ peptides (Willem et al., 2004). No CAA was observed probably due to the higher rate of self-association and fibrillogenic capacity of the shorter and less soluble N-truncated Aβ11–42 peptides that form amyloid deposits in the parenchyma, indicating that BACE1 is in tight control of the balance in amyloid pathology in brain, promoting either parenchyma or vasculature. APP/ApoE Mice In epidemiological investigations, it has been found that the ApoE4 allele is genet- ically associated with sporadic AD with a frequency of 45% compared with 15% in the general population (Corder et al., 1993). The pathological contributions of ApoE to amyloid and tau pathology in AD have been studied in different types of transgenic mice deficient in endogenous murine ApoE and/or overexpressed differ- ent ApoE isoforms, including various combinations with mutant human APP and PS-1 (Holtzman 2004). ApoE knock-out mice have significantly decreased synap- tophysin and MAP 2 staining, supporting the role of ApoE in the maintenance of synapses and dendrites during aging (Masliah et al., 1995). The finding that ApoE deficiency delayed amyloid plaque deposition in mice, whereas overexpression of human ApoE4 and not ApoE3 by transferring gene promoter accelerated plaque formation in transgenic mice, suggested a gain of function of ApoE4 (Bales et al., 1999; Carter et al., 2001). Authors showed that ApoE4 did not change the balance of amyloidogenic to nonamyloidogenic pathways. Nevertheless, the levels of Aβ42 and Aβ40 increased by ApoE4 overexpression, indicating that ApoE4 acted down- stream of the production of amyloid peptides, that is, slowed down the degradation 58 I. Ghorayeb et al. and clearance of Aβ peptides (Van Dooren et al., 2006). Furthermore, the neuron- specific proteolysis of ApoE4 was linked to increased phosphorylation of tau in the brain of ApoE mice (Brecht et al., 2004). Another most interesting finding is the development of CAA in the cortex, hippocampus, and thalamus of APP/ApoE4 and APP/ApoE4/PS-1 mice ( Fryer et al., 2003). APP/ADAM Mice The endoproteolysis of APP within the Aβ sequence by the α-secretase can pre- clude the formation of any Aβ peptides. In addition, cleavage by α-secretase releases the N-terminal soluble ectodomain of APP, known as APPα, which has been claimed to exert neurotrophic and neuroprotective properties (Mattson, 1997). Proteinases belonging to the ADAM family (A Desintegrin and Metalloproteinases) were the main candidates as physiologically relevant α-secretases. ADAM10 and ADAM17 single knock-outs have been shown to be lethal embryonically, whereas ADAM9 knock-outs are viable. Transgenic mice are developed with overexpres- sion of ADAM10 or a dominant negative catalytically inactive ADAM10 mutant with human mutant APP in double transgenic mice. Moderate neuronal levels of ADAM10 increased the secretion of APPα concomitantly with a reduction in the production of Aβ peptides, preventing any deposition of amyloid plaques. long-term potentiation (LTP) and cognitive deficits were also improved, suggesting a funda- mental rescue of synaptic function via the increased activity of α-secretases (Postina et al., 2004). Tau and Tau/APP Mice Although mutations in tau do not lead to AD, they produce dementia such as frontotemporal dementia with Parkinsonism associated with chromosome 17 (FTDP-17). In AD and other tauopathies, the MAP tau protein is abnormally hyperphosphorylated and is accumulated as intraneuronal tangles of PHF in cell bodies of neurons. Furthermore, the number of tangles correlates significantly with the degree of dementia, more so than the amyloid plaque numbers. Tau exists in six isoforms (352–441 amino acids) by alternative splicing of exons 2, 3, and 10, with isoforms containing either three or four C-terminal tandem microtubule- binding domain repeats and either no, one, or two shorter N-terminal domains (Fig. 1). Preparations of PHFs from AD brains reveal only three isoforms of tau corresponding to abnormally phosphorylated tau (Goedert et al., 1992). Pathogenic mutations in the tau gene that cause FTD and FTDP-17 either reduce the ability of tau to bind to microtubules or alter the splicing of exon 10 result- ing in increased 4 repeat tau isoforms. The first transgenic tau models (ALZ7 line) expressing wild-type human tau were generated in 1995 before pathogenic tau mutations had been identified. Overexpressing the longest isoform of human tau (4 repeats) under the human Thy-1 promoter resulted in hyperphosphorylation of tau and somatodendritic localization (Gotz et al., 1995). There are no NFTs, but these mice suffered from a severe axonopathy instead, with progressive paralysis of the hindlimbs, extending to the forelimbs, and age-related increased impairment Animal Models of Neurodegenerative Diseases 59 Fig. 1 Schematic representation of six human brain tau isoforms, White boxes: three or four tubulin-binding domains, Grey boxes: inserts from exon 2 near the N-terminus, Vertical lines in boxes: inserts from exon 3 near the N-terminus, and Black boxes: inserts from exon 10 near the N-terminus in the performance of tasks such as beam walking and rotarod (Spittaels et al., 1999). Overexpressing wild-type human 3 repeat tau under the mouse PrP promoter also resulted in hyperphosphorylation of tau and axonopathy in the spinal cord with NFTs in the hippocampus, amygdala, and entorhinal cortex, albeit at very old ages (18–20 months) (Ishihara et al., 2001). Because overexpression of wild-type human tau in mice replicated very limited aspects of tau pathology in AD, many groups turned to the discovered pathogenic tau mutations for use in animal models. In 2000, Lewis et al. (2000) published their J NPL3 mouse model where the transgene con- tains the most common tau mutation (P301L) associated with FTDT-17 under the mouse PrP promoter. These mice with no amyloid pathology developed NFTs asso- ciated with astrogliosis, apoptosis in the spinal cord, and motor and behavioural disturbances. Before producing a bigenic APP/tau mouse model, some authors injected a synthetic Aβ42 into the somatosensory cortex and contralateral hippocampus of P301L mice resulting in a fivefold increase in NFT numbers in the amygdala, which receives projections from both cortex and hippocampus (Gotz et al., 2001). However, Aβ was not capable of inducing NFT formation in non-NFT-forming WT tau transgenic mice (Gotz et al., 2001). Crossing Tg2576 mice with JNPL3 tau mice resulted in a double transgenic mouse line showing a more than sevenfold increase in NFT numbers at 9–11 months of age compared to JNPL3 mice. However, the presence of tau did not affect amyloid pathology (Lewis et al., 2001). To address the relationship of plaques and NFTs, Oddo et al. (2003) developed a triple trans- genic mouse model named 3xTg-AD. These mice harbor mutations of APP SWE , PS-1 M146V KI, and tau P301L and develop senile plaques first in the cortex (around 3 months of age) that spread to the hippocampus by 6 months. Tangles develop after amyloid pathology with hippocampal origin at 12 months of age and extend to the cortex. This regional and temporal development of pathology closely mimics the development of pathology in AD. These mice also exhibit synaptic dysfunc- tion, including LTP deficits that precede senile plaques and tangles formation (Oddo et al., 2003). In this triple transgenic model, cognitive deficits are observed whereas no cell loss is depicted. In vitro, many kinases can phosphorylate tau, but it is very difficult to establish the equivalent in the brain in vivo and to define exactly which kinases are responsible 60 I. Ghorayeb et al. for the phosphorylation of tau at precise amino acid residues. Two kinases that are the most likely candidates in vivo are glycogen synthase kinase-3β (GSK-3β) and cyclin-dependent kinase 5 (cdk-5). Neuronal overexpression of GSK-3β by itself reduced the brain size without any phenotypic repercussion or development of tauopathy despite increased phosphorylation of tau (Spittaels et al., 2000, 2002). Surprisingly, in the tau-4R × GSK-3β double combination, the axonopathy was practically completely rescued with elimination of axonal dilations in brain and spinal cord, reduction in axonal degeneration and muscular atrophy, and the alle- viation of all motor problems. The amount of tau associated with microtubules was reduced by 50% compared to single htau-4R transgenic animals and unbound tau was phosphorylated, leading to the conclusion that hyperphosphorylation of protein tau does not cause tauopathy per se. Recently, two novel bigenic mouse models, APP L /Tau P301L with amyloid and tau pathology and GSK-3β/Tau P301L with tauopathy only showed remarkable parallels: aggravation of tauopathy, severe cognitive and behavioural defects in young adults before the onset of amyloid deposits or tauopathy and activated GSK-3β with pathological phospho-epitopes of tau (S396/S404, characteristic GSK-3β motif). These findings indicate that Aβ induces tauopathy through activation of GSK-3β (Terwel et al., 2008). In addition, cdk-5 and its activating subunit p35 or its N-truncated p25 product have been inactivated or expressed in transgenic mouse brain with various degrees of success. The expression of cdk-5 with p35 and tau-4R in triple transgenic mice has yielded no additional new insights to the problem. Then, inducible p25 mice controlled by tetracycline displayed a dramatic neurodegeneration and neuroinflam- mation. A 30% decrease in brain weight was evident in a three-month observation period after the induction of p25 at the age of 6 weeks (Muyllaert et al., 2008; Table 2). 2.2.3 Transgenic Rat Models of Alzheimer’s Disease Parallel to the generation of transgenic mice, several transgenic rat models have also been produced as rats are a better rodent model for studies involving neurobe- havioural testing, cannulation, sampling of cerebrospinal fluid, electrophysiology, neuroimaging, and cell-based transplant manipulations (Abbott 2004). The first transgenic rat line was generated by Flood et al. (2007). Rats have human APP SL and human PS1 M146V gene mutations and developed amyloid deposits at 9 months of age. APP SWE rat model was reported but no amyloid pathology was observed except for a low intracellular accumulation of Aβ (Echeverria et al., 2004). Another APP SWE rat model has been generated and produced mild, extracellu- lar Aβ immunostaining and failed to develop compact, mature amyloid plaques by the age of 22 months (Folkesson et al., 2007). Recently, the model of Flood et al. (2007) has been more characterized: from the age of 9 months on, this rat model of AD had amyloid deposits in both diffuse and compact forms associated with activated microglia and reactive astrocytes; two months before the appearance of amyloid plaques, impaired LTP was revealed on hippocampal slices, accompa- nied by impaired spatial learning and memory in the Morris water maze; a mild Animal Models of Neurodegenerative Diseases 61 Table 2 Tau Transgenic Mouse Models Name Gene Mutation Promoter Amyloid Deposits P-Tau NFT Axonopathy Neuronal Loss Memory Deficits Inflammatory reaction ALZ7, ALZ17 4R tau hThy-1, mThy-1 No Yes No Yes No No NR 7TauTg 3R tau PrP No Yes Yes Yes NR NR NR JNPL3 4R tau P301L PrP No Yes Yes Yes Yes Yes Yes pR5 4R tau P301L Thy-1.2 No Yes Yes Yes Yes NR NR TAPP APP 695(K670N,M671L) 4R tau P301L PrP PrP AP (9–11 M) Yes Yes Yes Yes Yes Yes 3xTg-AD APP SWE PS-1 M146V KI Tau P301L Thy-1 Thy-1 Thy-1 AP Yes Yes Yes NR Yes Yes AP: amyloid neuritic plaques; h: human; m: mouse; NFT: neurofibrillar tangles; NR: not reported; P-Tau: hyperphosphorylated MAPT tau protein. 62 I. Ghorayeb et al. amyloid angiopathy was also described on the leptomeningeal blood vessels (Liu et al., 2008a). 2.3 Invertebrate Models Species as diverse as the fly Drosophila Melanogaster, the nematode Caenorhabditis elegans, and the sea lamprey Petromyzon marinus have been employed to provide new insight into the pathogenesis of AD. As we also show in other animal models of neurodegenerative disorders, these lower species offer sev- eral advantages compared to rodent models. The sea lamprey was used to study the degenerative changes linked to tau overexpression as it presents s ix giant neurons in the hindbrain which resemble most large vertebrate neurons and are readily acces- sible for manipulation (Hall et al., 1997). Flies and worms have other advantages: easy and fast to breed, cheap, no ethical limitations, powerful genetics, and modifier (suppressor and enhancer) screens and drug screenings are possible. These mod- els are useful to understand the normal functions and regulation of APP, PS, and tau genes. Genetic approaches could identify cellular processes that can suppress Aβ- or tau-dependent pathology. The fly APP homologue, APPL, does not contain the segment of APP cleaved to generate pathogenic amyloid peptides. Therefore, some authors studied the physi- ological functions of APP and APPL in Drosophila. Both proteins were shown to function as vesicular receptors for kinesin 1, a motor-mediated anterograde vesi- cle trafficking. Flies lacking APPL or overexpressed of WT and mutant APP have axonal transport defects and only APP overexpression increased cell death in the larval brain (Cauchi and van den Heuvel 2006). Other authors introduced FAD- linked mutations at conserved residues in the Drosophila PS gene or overexpressed APP/BACE and showed an increased neurotoxicity in the fly with production of amyloid peptides (Sang and Jackson 2005). Modelling AD in the fly was also attempted by delivering transgenes encoding Aβ40 and Aβ42 peptides. Results with Aβ42 peptides specifically expressed in brain tissue showed a reduced longevity, locomotor deficits, impaired olfactory memory, and neurodegeneration whereas Aβ40 flies were not affected. As for the fly, C. elegans has an APP homologue, APL-1. The RNAi knock- down results in a more severe uncoordinated phenotype and genetic deletion results in embryonic or larval lethality (Link, 2005; Segalat and Neri, 2003). Furthermore, a transgenic C. elegans expressing human Aβ has been developed and shown neurodegeneration, amyloid deposits, oxidative stress, and upregulation of many stress-related genes (Wu and Luo, 2005). In contrast to APP, the deletion of the worm tau homologue or the fly tau homologue does not result in any detectable phenotype, probably due to compensation by other MAPs. Authors produced trans- genic Drosophila and C. elegans by introducing either the normal human tau gene or various mutant forms of the human tau gene. Invertebrate animals displayed neurodegeneration, a shortened life span, axonal transport defects, vacuolization in the cortex of the fly, and positive immunostaining for a series of NFT-specific Animal Models of Neurodegenerative Diseases 63 epitopes without insoluble tau fibrils in Drosophila contrary to C. elegans (Gotz et al., 2004). 2.4 Primate Models 2.4.1 Spontaneous Approaches Although nonhuman primates do not spontaneously develop AD, age-related behavioural and neurodegenerative changes occur in monkeys. Indeed, it has been shown that nonhuman primates of several species exhibit cerebrovascular and parenchymal Aβ amyloidosis but without or with paucity of tau lesions (Gearing et al., 1994, 1997; Martin et al., 1991; Struble et al., 1985; Walker et al., 1990; Wisniewski et al., 1973). Significant intraneuronal tau pathology was only recently documented in an aged chimpanzee (Rosen et al., 2008). Although the lesion pro- file in this chimpanzee differed somewhat from that in AD, the occurrence of both tau-immunoreactive paired helical filaments and Aβ-amyloidosis indicates that the molecular mechanisms for the pathogenesis of the two key hallmarks of AD, namely NFTs and senile plaques, are present in aged chimpanzees. In this monkey, although age probably played a role in the pathogeny of tauopathy, additional factors are sug- gested to be involved because it is unusual to encounter tau-immunoreactive neurons and processes in older animals (Gearing et al., 1994). Similarly to brain pathology, it was also found that the monkeys undergo an age- related decline in several domains of cognitive function (Bartus et al., 1979, 1978; Lai et al., 1995;Mooreetal.,2006; Rapp 1993, 1990;Voytko1999). However, these changes were not correlated with neuronal loss in memory-related brain regions such as the hippocampus and entorhinal cortex (Peters et al., 1996) but with exten- sive loss of neurons in subcortical cholinergic basal forebrain regions similar to AD (Smith et al., 1999). The validity of these spontaneous models remains, how- ever, questionable because by contrast to patients with AD in whom severe neuronal cell loss in the hippocampus can be found, the brain of normal aging subjects dis- played almost no neuron loss in this region (West et al., 1994). It can be concluded that the neurodegenerative processes associated with normal aging and with AD are qualitatively different and that human AD is not accelerated by aging but is a distinct pathological process. The validity of such models is also weakened by the lack of correlation between the degree of amyloid plaque accumulation and cogni- tive decline in aged monkeys (Sloane et al., 1997). Therefore, the pathological and cognitive changes observed in the aged nonhuman primates emphasize their value as animal models for studies of human aging but question their relevance to the human AD. 2.4.2 Lesioning Approaches Lesioned animal models are based upon the assumption that the destruction of basal forebrain cholinergic neurons by injection of a neurotoxin, such as ibotenic acid, is sufficient to reproduce some of the cognitive impairments associated with AD, mainly memory and learning deficits. Indeed, pathology in the basal forebrain 64 I. Ghorayeb et al. cholinergic neurons is a prominent feature of AD and it may be responsible for the severe memory deficits observed in these patients. Impaired learning abilities, visual discrimination, and memory deficit were thus elicited following lesions of the nucleus basalis of Meynert (NBM) (Irle and Markowitsch, 1987; Ridley et al., 1985; Roberts et al., 1990). In other studies, however, large lesions of the NBM did not impair memory or produced only transient mild deficit in visual recognition mem- ory (Aigner et al., 1987, 1991; Voytko et al., 1994). Surprisingly, a cholinesterase inhibitor (physostigmine) produced modest improvement in performance in the control group but not in the experimental animals. Thus, no consensual conclusion is available about the behavioural and cogni- tive effects of basal forebrain cholinergic neuron lesions in nonhuman primates. However, one has to keep in mind that AD is not solely a disease of the cholinergic system. 2.4.3 Pharmacological Approaches Based on evidence of modest improvements in cognitive function in patients with AD and in normal human volunteers with the augmentation of central cholin- ergic neurotransmission by cholinesterase inhibitors such as physostigmine and tacrine, many animal studies have investigated the effects of systemic adminis- trations of direct or indirect cholinergic modulators. The ability of cholinesterase inhibitors to reverse cognitive impairments induced by the muscarinic antagonist scopolamine has been demonstrated in nonhuman primates and has been the most widely exploited approach used in preclinical animal assays to identify potential therapies for AD (Aigner and Mishkin 1986; Bartus and Johnson 1976; Fitten et al., 1988; Rupniak et al., 1991, 1989; Rupniak et al., 1997; Tang et al., 1997). Although these models have provided a framework to understand AD and to test the preclinical development of drugs to treat the cognitive symptoms, fundamen- tal questions persist regarding the validity of measures of behavioural function in animals in terms of reflecting clinically relevant measurements of cognition. 2.5 Perspectives Several animal models of AD have been developed in species ranging from worms to primates. Although none completely recapitulate the disease process, they have proven to be useful models for neuropathological changes. As discussed below for all animal models of neurodegenerative diseases, there is no perfect animal model for AD. It all comes to the question that is to be answered. For instance, for screening purposes of molecules against Aβ aggregation, one can use an Aβ-injected rodent; t o explore mechanisms involved in neuronal death, an animal model with amyloid and tau pathology and neuronal loss such as APP SL PS-1 KI or Tau/APP mice are per- haps better suited. However, it is important to note that amyloid plaques are probably not at the origin of neuronal death inasmuch as the active vaccination in AD patients did not rescue the cognitive impairment whereas amyloid plaques were suppressed . human brain tau isoforms, White boxes: three or four tubulin-binding domains, Grey boxes: inserts from exon 2 near the N-terminus, Vertical lines in boxes: inserts from exon 3 near the N-terminus,. deposits in the parenchyma, indicating that BACE1 is in tight control of the balance in amyloid pathology in brain, promoting either parenchyma or vasculature. APP/ApoE Mice In epidemiological investigations,. Tau exists in six isoforms (352–441 amino acids) by alternative splicing of exons 2, 3, and 10, with isoforms containing either three or four C-terminal tandem microtubule- binding domain repeats