APP transgenic mice for modelling behavioral and psychological symptoms of dementia (BPSD)

Abstract

The discovery of gene mutations responsible for autosomal dominant Alzheimer's disease has enabled researchers to reproduce in transgenic mice several hallmarks of this disorder, notably Aβ accumulation, though in most cases without neurofibrillary tangles. Mice expressing mutated and wild-type APP as well as C-terminal fragments of APP exhibit variations in exploratory activity reminiscent of behavioral and psychological symptoms of Alzeimer dementia (BPSD). In particular, open-field, spontaneous alternation, and elevated plus-maze tasks as well as aggression are modified in several APP transgenic mice relative to non-transgenic controls. However, depending on the precise murine models, changes in open-field and elevated plus-maze exploration occur in either direction, either increased or decreased relative to controls. It remains to be determined which neurotransmitter changes are responsible for this variability, in particular with respect to GABA, 5HT, and dopamine.

1. Pathobiology of Alzheimer's disease

1.1 Cognitive symptoms of Alzheimer's disease

Probable Alzheimer's disease is diagnosed when progressive loss in remembering new items (anterograde amnesia) occurs in conjunction with a deficit in language (aphasia), object use (apraxia), form recognition (agnosia), or step-by-step planning (McKhann et al., 1984). In general, memory and verbal fluency deficits are the initial symptoms (Amieva et al., 2005), progressive amnesia sometimes being the only sign over a period of several years (Weintraub and Mesulam, 1993). A loss of factual information (declarative memory) is found for verbal and spatial items (Amieva et al., 2005; Graham et al., 2004; Helmes and Ostbye 2002; Jacobs et al., 1995; Kessels et al., 2005; Lee et al., 2003), whereas procedural memory is relatively spared, evaluated in pursuit-rotor (Eslinger and Damasio, 1986; Jacobs et al., 1999) and either verbal (Karlsson et al., 2002) or spatial (Kessels et al., 2005) priming tasks. However, some procedural tasks are impaired, such as prepulse inhibition of the acoustic startle response, implying a deficit in sensory gating (Ueki et al., 2006), as well as delay or trace conditioning of the eyeblink response (Woodruff-Pak and Papka, 1996; Woodruff-Pak et al., 1996).

In addition to amnesia, patients with Alzheimer's disease are susceptible to constructional apraxia, characterized by a difficulty in copying geometric figures or reproducing them with building blocks (Graham et al., 2004; Helmes and Ostbye 2002; Henderson et al., 1989; Nielson et al., 1996). Impaired flow perception (Kavcic et al., 2006; Tetewsky and Duffy, 1999), visual search (Parasuraman et al., 2000; Tales et al., 2004), abstract reasoning (Helmes and Ostbye 2002), and executive functions (Collette et al., 1999; Graham et al., 2004; Rainville et al., 2002b) may contribute to this symptom. As a result, they are more likely than controls to be disoriented in a maze (Rainville et al., 2002a), a circular arena (Hort et al., 2007), a hospital ward (Monacelli et al., 2003), and in the streets when driving a motored vehicle (Uc et al., 2004).

1.2 Behavioral and psychological symptoms of Alzheimer's disease

Among emotional symptoms, apathy is one of the most common features of Alzheimer dementia, sometimes accompanied by dysphoria, social withdrawal, and depression (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004). But sometimes agitation is the predominant feature, accompanied by disinhibition and inappropriate euphoria. These opposite signs often overlap in the same patient, as apathy scores have been found to be linearly correlated with disinhibition (Holthoff et al., 2005; Lyketsos et al., 2002). Apathy/anxiety scores, on one hand, and disinhibition/euphoria scores, on the other, were both correlated with caregiver non-compliance (Senanarong et al., 2004), underlying their importance in day-to-day activities. Symptoms may differ from one patient to another with respect to inhibition or disinhibition. Alzheimer patients may either be excessively inhibited due to anxiety, or else display a loss in inhibitory control, often accompanied by irritability and hostility (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004). Excessive aggression may be due to degeneration of brainstem 5-hydroxytryptamine (5HT) neurons, known to influence this symptom (Hardy et al., 1985; Lyness et al., 2003; Whitford 1986).

1.3 Behavioral symptoms linked with metabolic markers

Insight into the neurobiological factors underlying dementia symptoms may be obtained by measuring regional brain metabolism. Alzheimer patients characterized by apathy had lower fluorodeoxyglucose (FDG) uptake in medial prefrontal cortex than a subgroup not displaying this sign (Holthoff et al., 2005; Marshall et al., 2007). These results indicate that apathy may be the result of dysfunction at the level of medial prefrontal cortex. In addition, FDG uptake was lower in temporal neocortex of Alzheimer patients suffering the most from anxiety (Hashimoto et al., 2006). Moreover, Alzheimer patients with misidentification delusions had lower glucose utilization in orbitofrontal, medial temporal, and striate cortex but higher utilization in sensory association cortices (superior temporal and inferior parietal) than those without this symptom (Mentis et al., 1995). Likewise, demented patients with delusions had higher glucose utilization in inferior temporal but lower utilization in medial occipital cortex than those without them (Hirono et al., 1998). In addition to glucose, regional brain metabolism may be estimated by brain perfusion on single photon emission computed tomography (SPECT) scans by measuring the uptake of hexamethylpropelenamine oxime (HMPAO). The HMPAO technique dissociated between patient groups on the basis of emotional reactivity (Hirono et al., 2000). In particular, a mixed group of patients with Alzheimer's or vascular dementia displaying marked aggression had lower neocortical HMPAO uptake than a demented group not displaying this symptom.

2. Pathological characteristics in APP mice

2.1 Introduction

Thanks to the discovery of pathologic mutations, several Alzheimer hallmarks have been attained in generating transgenic mice (Cole and Frautschy, 1997; Dodart et al., 2002; Duff, 1999; Hock and Lamb, 2001; Hsiao, 1998a,b; Janus and Westaway, 2001; Seabrook and Rosahl 1999; van Leuven, 2000). But except for the one expressing APP, PS1, and Mapt (Oddo et al., 2003), APP mutants contain no neurofibrillary tangles, though often with hyperphosphorylated tau, thereby most resembling rare cases of Alzheimer dementia with amyloid deposits and hyperphosphorylated tau without neurofibrillary tangles (Tiraboschi et al., 2004a,b). The genetic characteristics of mice expressing mutated or wild-type APP or its C-terminal fragments are presented in Table 1.

Table 1.

Description of transgenic mice expressing human mutated or wild-type (WT) APP genes with or without mutated PS1 or else expressing Aβ species

Name	Mutation	Promoter	Parenchymal Aβ plaques	References
APP695SWE Tg2576	Swedish	hamster Prp	yes	Hsiao et al., 1996
APP695SWE Tg2123H	Swedish	hamster Prp	no	Hsiao et al., 1995
APP695SWE/Thy1 TgTAS10	Swedish	murine Thy1	yes	Richardson et al., 2003
APP695SWEch	Swedish	murine PrP	no	Borchelt et al., 1996, 1997
APP751SWE TgAPP23	Swedish	murine Thy1	yes	Sturchler-Pierrat et al., 1997
APP770SWE	Swedish	murine Thy1	yes	Moechars et al., 1999b
APP/SWE/Ckm	Swedish	murine Ckm	no	Sugarman et al., 2002
APP/IND TgPDAPP	Indiana	platelet-derived PDGFβ	yes	Games et al., 1995
APP695LD	London	murine Thy1	yes	Moechars et al., 1999b
APP/LD-PDGF	London	platelet-derived PDGFβ	yes	Qin et al., 2000
APP/DU	Dutch	murine Thy1	no	Kumar-Singh et al., 2000
APP/DU/THY1	Dutch	human THY1	no	Andrä et al.,1996
APP/FL	Flemish	murine Thy1	no	Kumar-Singh et al., 2000
APP695SWE/IND TgCRND8 or Tg19959	Swedish+Indiana	hamster Prp	yes	Chishti et al., 2001
APP/SWE/IND J9 and J20 lines	Swedish+Indiana	platelet-derived PDGFβ	yes	Mucke et al., 2000; Palop et al., 2003
APP695SWE/IND/tet	Swedish+Indiana	pTetSplice	yes	Jankowski et al., 2005b
APP751SWE/IND TgAPP-Sw,V717F/B6	Swedish+Indiana	platelet-derived PDGFβ	no	Lee et al., 2004
APP/SWE/ARC	Swedish+Arctic	murine Thy1	yes	Lord et al., 2006
APP751SWE/LD	Swedish+London	murine Thy1	yes	Blanchard et al., 2003
APP/SWE/LD	Swedish+London	murine PRP	yes	Kitamoto et al., 2002
APP/SWE/DI	Swedish+Dutch+Iowa	murine Thy1	yes	Davis et al., 2004
APP/RK	Artificial 684 + 687 positions	murine Thy1	no	Moechars et al., 1996
APP695TRImyc Tg1130H	artificial	hamster Prp	no	Hsiao et al., 1995
APP695WT/Mt2	none	Mt2	no	Yamaguchi et al., 1991
APP695WT/Thy1	none	murine Thy1	no	Moechars et al., 1999b
App695WT Tg1874	none	hamster Prp	no	Hsiao et al., 1995
APP695mycWTTg6209	none	hamster Prp	no	Hsiao et al., 1995
APP751WT	none	rat Nse	no	Quon et al., 1991
APP751WT	none	rat Nse	no	Mucke et al., 1994
APP WT Line I5	none	PDGFβ	no	Mucke et al., 2000
APP/C104	none	human NEFL	yes	Nalbantoglu et al.,1997
APP/C100-Flag	none	Dmd	no	Oster-Granite et al.,1996
APP/C99/IND	Indiana	platelet-derived PDGFβ	no	Lee et al., 2006
APP/C99 Tg13592	none	CMVenhancer/chicken Actb	no	Fukuchi et al., 1996
APP/C100/IND APP/C100/WT	Indiana wild-type	human ACTB	no	Li et al., 2006
pyroglutamate-Aβ	none	murine Thy1	no	Alexandru et al., 2011
Aβ40 and Aβ42 fused with BRI	none	murine Prp	yes for Aβ42	McGowan et al., 2005
Aβ42	none	murine Nefl	no	LaFerla et al., 1995
APP knockout	null	non-applicable	no	Müller et al., 1994
APP695SWE+PS1/M146L APP695SWE+PS1/M146V	bigenic	hamster Prp + PDGFβ2	yes	Duff et al., 1996; Holcomb et al., 1998
APP695SWE+PS1/A246E	bigenic	hamster Prp + human THY1	yes	Dineley et al., 2002 Qian et al., 1998
APP695SWE+PS1/M146L/Thy1 TgTASTPM	bigenic	murine Thy1	yes	Howlett et al., 2004
APP695SWEch+PS1/A246E	bigenic	murine PrP	yes	Borchelt et al., 1996
APP695SWEch+PS1/ΔE9	bigenic	murine PrP	yes	Jankowsky et al., 2005a
APP695SWE/co+PS1/ΔE9	co-injected bigenic	murine PrP	yes	Jankowsky et al., 2004
APP751SW+PS1/G384A	bigenic	murine ThyI for both	yes	Busche et al., 2008
APPswe+PS1/L166P	co-injected bigenic	murine Thy1	yes	Radde et al., 2006
APP695LD+PS1/A246E	bigenic	murine Thy1 for both	yes	Van Dorpe et al., 2000
APP751SWE/LD+PS1/M146L	bigenic	murine Thy1 + human HMGCR	yes	Blanchard et al., 2003
APP751SWE/LD+PS1/M233T+L235P	transgenic + knockin	murine Thy1	yes	Casas et al., 2004
APP/SWE/Ckm+PS1/M146V KI	transgenic + knockin	murine Ckm	no	Sugarman et al., 2002
APP751SWE+PS2/VG TgPS2APP	bigenic	murine Thy1 + murine PrP	yes	Richards et al., 2003
3xTg-AD	trigenic	murine Thy1 + endogenous control + murine Thy1+	yes	Oddo et al., 2003
5xFAD	5-time mutated	murine Thy1	yes	Oakley et al., 2006

2.2 APP mutations

2.2.1 Swedish

The most cited murine Alzheimer model appears to be the Tg2576 mouse, containing an APP transgene with the Swedish mutation of the neuron-specific 695-amino acid isoform driven by the hamster Prp promoter (Hsiao et al., 1996). Detergent-insoluble Congo-red positive Aβ plaques were visible in APP₆₉₅SWE mice as early as 7 months of age and increased with time (Kawarabayashi et al., 2001; Westerman et al., 2002). A second transgene with the same mutation and promoter (Tg2123H) does not harbor Aβ plaques (Hsiao et al., 1995). As seen in Alzheimer brain (Iwatsubo et al., 1994), diffuse plaques in APP₆₉₅SWE mice contain Aβ₄₂ and mature plaques Aβ₄₂ and Aβ₄₀ (Harigaya et al., 2006; Kawarabayashi et al., 2001; Terai et al., 2001). Unlike Alzheimer brain (Thal et al., 2000), cored Aβ plaques appeared earlier, at 12 months, than diffuse ones (Harigaya et al., 2006; Kawarabayashi et al., 2001), a feature appearing in other murine models such as APP₆₉₅SWE/Thy1 (see below) (Howlett and Richardson, 2009), and these are accompanied by hyperphosphorylated tau without neurofibrillary tangles (Tomidokoro et al., 2001). Soluble Aβ dimers appeared in lipid rafts at 6 months (Kawarabayashi et al., 2004). As seen in Alzheimer brain, Aβ₄₂ accumulated intracellularly before plaque formation in multivesicular bodies at pre- and postsynaptic sites (Takahashi et al., 2002) and lysosomes (Shie et al., 2003). A third resemblance with Alzheimer brain (Eikelenboom et al., 2002; McLellan et al., 2003) is the presence of astrocytes and microglia (Irizarry et al., 1997; Wegiel et al., 2001) with reactive oxygen species (McLellan et al., 2003) around fibrillar deposits. Gliosis is observed as soon as Aβ plaques appear (Jacobsen et al., 2006). While astrocytes isolate plaques from healthy tissue, microglia cannot phagocytose them (Wegiel et al., 2001, 2003). Glial-mediated inflammation is mostly seen in the vicinity of fibrillar Aβ plaques (Benzing et al., 1999; Quinn et al., 2003). The Aβ plaques also surround vessel walls, causing microhemorrhages (Domnitz et al., 2005; Frackowiak et al., 2001; Fryer et al., 2003).

Although spine density decreased in the CA1 hippocampal subregion of APP₆₉₅SWE mice (Lanz et al., 2003), neuronal counts were unchanged (Irizarry et al., 1997). The spine density decline in molecular layer of dentate gyrus appeared even before plaque deposition (Jacobsen et al., 2006). White matter tracts appear intact (Harms et al., 2006). Thus, neurodegeneration seems minimal, probably because of enhanced APP-induced gene expression of cell survival pathways (Stein and Johnson, 2003), since chronic infusion of the antibody to the upregulated Aβ-binding protein, transthyretin, caused cell loss in the CA1 pyramidal layer of APP₆₉₅SWE mice (Stein and Johnson, 2002).

A second highly cited model also contains the APP transgene with the Swedish mutation but this time driven by the endogenous PrP promoter of a chimeric (ch) human/murine 695-amino acid isoform (Borchelt et al., 1996, 1997). Despite increased Aβ₄₂ and Aβ₄₀ levels, few or no neuritic Aβ plaques were seen in APP₆₉₅SWEch mouse brain (Savonenko et al., 2003). In contrast, the APP₆₉₅SWE/Thy1 transgene (TgTAS10) driven by the murine neuron-specific Thy1 (thymus cell antigen 1) promoter generated Aβ plaques in cerebral cortex at 12 months of age, acompanied by astrocytosis, microgliosis, and hyperphosphorylated tau (Richardson et al., 2003).

Another highly cited transgene with the Swedish mutation driven by the murine Thy1 promoter was generated on the 751-amino acid isoform (Sturchler-Pierrat et al., 1997). APP₇₅₁SWE mice (APP23) accumulate Aβ in hippocampus and neocortex, with plaque-associated Aβ peptides augmenting 5-fold from 3 to 6 months of age (Van Dam et al., 2003). As seen in Alzheimer brain, neuritic plaques are surrounded by astrocytes and activated microglia and tau is hyperphosphorylated (Bornemann et al., 2001; Stalder et al., 1999; Sturchler-Pierrat et al., 1997). A further resemblance between human (Lüth et al., 2001, 2002) and murine (Lüth et al., 2001) pathology is the overexpression of induced and endothelial forms of nitric oxide synthase (iNOS and eNOS) accompanying astrocytosis. Rodrigo et al. (2004) found increased iNOS but not nNOS concentrations in APP₇₅₁SWE brain. A 10% loss of pyramidal neurons in the CA1 hippocampal subregion was documented (Calhoun et al., 1998), though, unlike advanced Alzheimer's disease, without loss in noradrenergic locus coeruleus neurons (Szot et al., 2009). Neocortical neurons increased by 10% and then declined by the same percentage with aging (Bondolfi et al., 2002). The initial increase is probably caused by the neurotrophic action of secreted APP (Stein and Johnson, 2003; Storey and Cappai, 1999). The neuronal loss was accompanied by increased DNA strand breaks (Bondolfi et al., 2002). Amyloid angiopathy led to weakened vessel walls, aneurysm, vasculitis, and hemorrhage (Beckmann et al., 2011; Calhoun et al., 1999; Winkler et al., 2001), as well as a reduction in their vasodilatory properties (Mueggler et al., 2002, 2003). In neocortex, the water diffusion coefficient was reduced (Mueggler et al., 2004) and size of extracellular space increased (Sykova et al., 2005). The chemical composition of Aβ plaques in APP₇₅₁SWE (Kuo et al., 2001b) as well as the previously APP₆₉₅SWE mutants (Kalback et al., 2002; Kuo et al., 2001a) resemble those found in Alzheimer's disease, except that the plaques examined in the two mouse models were more soluble (Kuo et al., 2001), characteristic of other murine models as well (Kokjohn and Roher, 2009).

Yet another APP model with the Swedish mutation driven by the murine Thy1 promoter includes the 770-amino acid isoform (Moechars et al., 1999b). Aβ plaques were observed in hippocampus and neocortex of these APP₇₇₀SWE mice.

In an attempt to model inclusion body myopathy (IBM), a APP/SWE/Ckm transgene with the Swedish mutation driven by the murine Ckm (creatine kinase, muscle) promoter was generated (Sugarman et al., 2002). As found in patients (Askanas and Engel, 1998, 2001, 2002), Aβ accumulated intracellularly in mouse skeletal muscle (Sugarman et al., 2002). As seen in sporadic IBM but not hereditary IBM (Askanas and Engel 1998), inflammation was evident in the mouse model, with neutrophils instead of T cells as the predominant cell type (Sugarman et al., 2002).

2.2.2 Indiana

APP/IND transgenic mice (PDAPP) carry the V717F Indiana mutation of all three major APP isoforms driven by the PDGFβ promoter (Games et al., 1995). APP/IND mouse brain contains Congo-red positive Aβ plaques, appearing as early as 6 to 9 months of age, surrounded by proliferating microglia and astrocytes (Games et al., 1995; Masliah et al., 1996; Schenk et al., 1997) and reactive oxygen species (McLellan et al., 2003). There is also an age-dependent increase in phosphorylated tau (Masliah et al., 2001). As seen in APP₆₉₅SWE mice (Harigaya et al., 2006; Kawarabayashi et al., 2001), cored Aβ plaques appeared earlier than diffuse ones (Reilly et al., 2003). Cerebral angiopathy is observed with microhemorrhages (Domnitz et al., 2005; Fryer et al., 2003; Kimchi et al., 2001). Although hippocampal volume was reduced (Gonzalez-Lima et al., 2001; Redwine et al., 2003; Weiss et al., 2002), cell counts were unchanged (Irizarry et al., 1997), attributable to dendritic or axonal atrophy, as dendritic length diminished in dentate gyrus granule cells as early as 3 months of age (Wu et al., 2004).

2.2.3 London

Two APP models have been generated with the V717I London mutation, one on the 695-amino acid isoform driven by the murine Thy1 promoter, causing diffuse and neuritic Aβ plaques (Moechars et al., 1999b; Dewachter et al., 2000a,b; Van Dorpe et al., 2000), surrounded by activated astrocytes and microglia, the neurites being linked with hyperphosphorylated tau (Moechars et al., 1999b). The angiopathy in APP₆₉₅LD mice leads to aneurysms but not hemorrhages (Van Dorpe et al., 2000), potentiated after interbreeding with a PS1/A246E line (Van Dorpe et al., 2000). The second model includes an APP₇₅₁LD transgene driven by the PDGFβ promoter, which also causes neuritic Aβ plaques (Qin et al., 2000).

2.2.4 Dutch and Flemish

APP/DU/Thy1 and APP/FL/Thy1 mice with Dutch (E693Q) and Flemish (A692G) mutations, respectively, each driven by the murine Thy1 promoter, have cerebral amyloid angiopathy with elevated C-terminal fragments in parenchymal regions but without Aβ plaques (Kumar-Singh et al., 2000), thus serving as animal models of cerebral amyloid angiopathy caused by such mutations. A second APP/DU/THY1 model with the human THY1 promoter also has cerebral amyloid angiopathy without Aβ plaques (Andrä et al., 1996; Gandy et al., 2010).

2.2.5 Swedish/Indiana, Swedish/Arctic, Swedish/London, and Swedish/Dutch/Iowa

Five models characterized by combined Swedish/Indiana mutations are available. APP₆₉₅SWE/IND mice (TgCRND8 or Tg19959) express the 695-amino acid isoform regulated by the hamster Prp promoter (Chishti et al., 2001; Janus et al., 2000), showing Aβ plaques as early as 3 months of age (Bellucci et al., 2006; Chishti et al., 2001; Janus et al., 2000), together with early onset angiopathy (Domnitz et al., 2005) and activated microglia and astrocytes (Bellucci et al., 2006). As seen in APP₆₉₅SWE (Harigaya et al., 2006; Kawarabayashi et al., 2001), APP/IND (Reilly et al., 2003), and other murine mutants (Howlett and Richardson, 2009), diffuse Aβ plaques appeared after cored Aβ plaques (Chishti et al., 2001). Two APP/SWE/IND constructs driven by the platelet-derived PDGFβ promoter (J9 and J20 lines) also contain Aβ plaques (Mucke et al., 2000; Palop et al., 2003), as well as mice with a human/mouse chimeric APP₆₉₅SWE/IND/tet transgene regulated by the tetracycline-responsive promoter of pTetSplice (Jankowsky et al., 2005b). On the contrary, the APP₇₅₁SWE/IND (APP-Sw,V717F/B6) model driven by the PDGFβ promoter did not harbor plaques up to 14 months of age (Lee et al., 2004). An APP/SWE/ARC line was generated with Swedish and Arctic mutations driven by a murine Thy1 cassette, with Aβ plaques occurring at 10 months of age (Lord et al., 2006, 2009). An APP SWE/LD line with Aβ plaques contains a combined Swedish/London mutation regulated by the murine PRP promoter (Kitamoto et al., 2002).

An APP₇₅₁SWE/LD line with Swedish and London mutations on a transgene driven by the murine Thy1 promoter was also generated, containing Aβ plaques in parenchyma (Blanchard et al., 2003). In addition, an APP/SDI line was generated with Swedish, Dutch, and Iowa mutations causing Aβ plaque accumulation in parenchyma and cerebral vessels (Davis et al., 2004).

2.2.6 Artificial

APP/RK transgenic mice with an artifical double mutation at the α-secretase site were generated under control of the murine Thy1 promoter (Moechars et al., 1996). These mice harbor no Aβ plaques, but neocortex and hippocampus are marked by astrocytosis and DNA fragmentation (Moechars et al., 1998, 1999a). Likewise, no Aβ plaques were observable in APP₆₉₅TRImyc mice (Tg1130H), expressing a triple mutation (V717L, V721A, M722V) with a 3'-myc tag driven by the hamster Prp promoter (Hsiao et al., 1995).

2.3 Murine wild-type App and human wild-type APP

The 695 splice variant was used for generating mice expressing human wild-type (WT) APP₆₉₅under control of the Mt2 (metallothionein IIA) promoter (Yamaguchi et al., 1991). A similar APP₆₉₅WT construct driven by the murine Thy1 promoter is available (Moechars et al., 1999b). Mice expressing human APP₆₉₅mycWT with a 3'-myc tag (Tg6209) and murine App₆₉₅WT (Tg1874) driven by the hamster Prp promoter have also been generated (Hsiao et al., 1995). In addition, two separate mouse lines expressing APP₇₅₁WT regulated by the rat neuron-specific Nse promoter are available (Mucke et al., 1994; Quon et al., 1991), together with APP WT regulated by the PDGFβ promoter (line I5) (Mucke et al., 2000).

2.4 APP carboxy-terminal fragments and Aβ

In view of the neurotoxic properties of C-terminal fragments in cell cultures (Fukuchi et al., 1993; Kim and Suh, 1996) and their accumulation in Alzheimer brain (McPhie et al., 1997), liable to influence disease onset and progression (Turner et al., 2003), it is of interest to examine their neurobehavioral impact. APP/C104 mice with the human 104-amino acid C-terminal fragment transgene driven by the human NEFL promoter have Aβ plaques surrounded by reactive microglia and astrocytes as well as reduced neuronal counts in the CA1 hippocampal subregion (Nalbantoglu et al., 1997). APP/C99/IND transgenic mice express the human 99-amino acid C-terminal fragment with the V717F Indiana mutation driven by the PDGFβ promoter (Lee et al., 2006), while APP/C100/IND mice and APP/C100/WT mice express the human 100-amino acid C-terminal fragment with the V717F Indiana mutation or wild-type (WT), both driven by the Actb (β-actin) promoter (Li et al., 1999). The APP/C100-Flag transgene driven by the Dmd promoter uses the 100 amino acid C-terminal fragment fused with the hydrophilic 8-amino acid Flag epitope (Oster-Granite et al., 1996). APP/C99 mice (Tg13592) expressing the human 99-amino acid C-terminal fragment in brain and skeletal muscle driven by a CMV enhancer/chicken Actb promoter mimick IBM in that Aβ accumulates in muscle but not in brain (Fukuchi et al., 1996, 1998, 2000).

The direct impact of Aβ at different lengths has been examined. For example, transgenic mice overexpressing endogenous Aβ₄₂ driven by the murine Nefl promoter have increased intracellular levels of the peptide (LaFerla et al., 1995, 1996). These mice are marked by DNA fragmentation, increased p53 levels, and morphological markers of apoptosis in neocortex, hippocampus, amygdala, and thalamus. Moreover, transgenic mice overexpressing either Aβ₄₀ or Aβ₄₂ in the form of fusion proteins with the BRI protein driven by the murine Prp promoter have been generated (McGowan et al., 2005). The influence of truncated Aβ species have also been examined, notably in a transgenic mouse model overexpressing pyroglutamate-Aβ driven by the murine Thy1 promoter, causing hippocampal cell loss, astrocytosis, microgliosis, and behavioral signs such as tremor (Alexandru et al., 2011).

2.5 APP null mutation

The only behaviorally characterized APP knockout model contains a deletion mutation of exon 2, resulting in a shorter protein (Müller et al., 1994).

2.6 APP+PS1 mutations

The above-described APP₆₉₅SWE mice (Tg2576) were interbred with mutated PS1 lines 5.1 and 6.2, yielding APP₆₉₅SWE+PS1/M146L and APP₆₉₅SWE+PS1/M146V, respectively, characterized by earlier Aβ plaque onset than the monogenic (Duff et al., 1996; Holcomb et al., 1998, 1999; Kurt et al., 2003; Westerman et al., 2002), with the same features of astrocytosis and microgliosis (Gordon et al., 2002; Holcomb et al., 1998, 1999), the Aβ plaques surrounding vessel walls (Christie et al., 2001; Domnitz et al. 2005) with reduced vasodilation properties (Christie et al., 2001). The same APP₆₉₅SWE line was also bred with a PS1/A246E strain on a Ps1 null background (Qian et al., 1998), once more yielding Aβ plaques with earlier onset than the monogenic (Dineley et al., 2002). A similar APP₆₉₅SWE+PS1/M146L/Thy1 model with Aβ plaques (TgTASTPM) has been described with the mouse Thy1 promoter used for both transgenes, though not on a Ps1 null background (Howlett et al., 2004).

APP₆₉₅SWEch mice (line C3-3) have also been bred with a PS1/A246E line, each transgene driven by the endogenous PrP promoter (Borchelt et al., 1996, 1997). An age-related increase of soluble and insoluble Aβ₄₂ and Aβ₄₀ levels were documented in the APP₆₉₅SWEch+PS1/A246E mice (Marutle et al., 2002, Wang et al., 2003). Neuritic Aβ plaques in hippocampus and neocortex were surrounded by astrocytes and microglia (Borchelt et al., 1997). As mentioned above, APP₆₉₅LD has been interbred with the PS1/A246E line (Van Dorpe et al., 2000). The same APP₆₉₅SWEch model was bred with PS1/ΔE9 (line S-9), each transgene driven by the endogenous PrP promoter, resulting in Aβ plaques at 7 months of age (Jankowsky et al., 2005a; Perez et al., 2005; Savonenko et al., 2005). Due to Aβ deposits in striatum (Perez et al., 2005), neuronal loss has been reported there (Richner et al., 2009), along with decreased 5HT neurons in raphe and decreased noradrenergic neurons in locus coeruleus (Liu et al., 2008).

A co-injected bigenic model (line 85) contains the same 695-amino acid human/murine APP transgene with the Swedish mutation combined with a PS1/ΔE9 transgene driven by mouse Prp (Jankowsky et al., 2004). APP₆₉₅SWE/co+PS1/ΔE9 mice harbor Aβ plaques at 6 months of age (Garcia-Alloza et al., 2006; Jankowsky et al., 2004) and possess fewer capillary segments than controls in neocortical white matter (Lee et al., 2005). Analogous to the previous model, 5HT fibers decreased in frontal cortex of the bigenics (Gimbel et al., 2010). In addition, APP₇₅₁SWE mice were bred with a PS1/G384A line, both under control of the Thy1 promoter (Busche et al., 2008). A second co-injected line contains Swedish-mutated APP combined with a PS1/L166P mutation driven by the mouse Thy1 promoter, containing Aβ plaques and gliosis (Radde et al., 2006).

As mentioned in section 2.2.1, one model with the APP Swedish mutation express APP in skeletal muscle (Sugarman et al., 2002). These mice were crossed with homozygous PS1/M146V knock-in mice, forming a bigenic APP/SWE/Ckm model (Sugarman et al., 2002; Guo et al. 1999; Kitazawa et al. 2006), with higher than normal Aβ42 levels and phosphorylated tau in skeletal muscle with infiltrates of CD8-immunoreactive T cells (Kitazawa et al. 2006).

An APP₇₅₁SWE/LD line with Swedish and London mutations on a transgene driven by the murine Thy1 promoter was crossed with a PS1/M146L line on a transgene driven by the human HMGCR (hydroxy-methyl-glutaryl-Coenzyme A reductase) promoter (Blanchard et al., 2003). The APP₇₅₁SWE/LD +PS1/M146L mice have Aβ plaques as early as 2.5 months of age and age-related microglial activation, first with the appearance of a phagocytic followed by a cytotoxic phenotype (Jimenez et al., 2008), and pyramidal cell loss in hippocampus (Schmitz et al., 2004). The same APP₇₅₁SWE/LD line was crossed with a PS1/M233T+L235P knockin, yielding a mutant also with Aβ plaque onset at 2.5 months of age, glial activation, and pyramidal cell loss in hippocampus (Casas et al., 2004), as well as a loss in dentate gyrus granule cells (Cotel et al., 2008) and CA1–2 thinning (Faure et al., 2011). In addition to hippocampal damage, the APP₇₅₁SWE/LD +PS1/M233T+L235P model is marked by spinal axonopathy and thoracolumbar kyphosis, i.e. spine curvature (Wirths et al., 2008).

As mentioned above, the only model with Aβ plaques and neurofibrillary tangles is the triple APP/SWE+PS1/M146V+Mapt/P301L mutant (Oddo et al., 2003). The 3xTg-AD mice were generated by co-injecting APP with the Swedish mutation and four-repeat Mapt with the P301L mutation into single-cell embryos of PS1/M146V knockin mice. A five-fold transgenic model with Swedish/Florida/London mutations on APP and M146L and L286V mutations on PS1 has been generated with the murine Thy1 promoter (Oakley et al., 2006). The 5xFAD (Tg6799) transgenic mice have Aβ plaques as early as 2 months of age, accompanied by astrogliosis and microgliosis.

2.7 APP+PS2 mutations

To our knowledge, the only model fitting this category is the PS2 transgenic one carrying the N141I Volga-German mutation driven by the endogenous Prp promoter bred with Swedish mutated APP mice driven by the endogenous Thy1 promoter (Richards et al., 2003), yielding APP₇₅₁SWE+PS2/VG (TgPS2APP) mice. The brain of the double mutant contains Aβ plaques surrounded by activated microglial cells and astrocytes.

3. Exploratory activity in Alzheimer mice

Several reviews have appeared on murine Alzheimer models, mainly useful in understanding the impact of amyloid formation on learning or neurobiological processes (Ashe, 2001; Ashe and Zahs, 2010; Duyckaerts et al., 2008; German and Eisch, 2004; Gimenez-Llort et al., 2007; Howlett and Richardson, 2009; Kokjohn and Roher, 2009; McGowan et al., 2006; Mineur et al., 2005; Morgan, 2003; Obulesu and Rao, 2010; Sant'Angelo et al., 2003; Scearce-Levie, 2011; Spires and Hyman, 2005; Van Dam et al., 2005; van Leuven, 2000; Wirths and Bayer, 2010), but none of the authors focused their attention on exploratory activity. In the present review, we analyze the effects of genetic modification of APP on open-field, Y- or T-maze, elevated plus-maze, and emergence tests. In view of the tendency of patients with Alzheimer's disease to be either apathetic or agitated (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004), murine models may be expected to be either hypoactive as a result of apathy or hyperactive as a result of agitation. Moreover, because apathy and disinhibition are linearly correlated in Alzheimer patients (Holthoff et al., 2005; Lyketsos et al., 2002), one may perhaps expect murine models to be hypoactive in one apparatus and hyperactive in another. Another possibility is to uncover opposite changes as a function of aging.

3.1 Open-field, Y-maze, and home-cage

The exploration of novel stimuli has long been assessed in the open-field (Walsh and Cummins, 1976) and Y-maze (King and Arendash, 2002). In contrast, the exploration of familiar stimuli can be examined in the home-cage (Ognibene et al., 2005). In all cases, the main measure used is the distance travelled. Alternative measures include time spent walking, sometimes recorded at different motion speeds. An IntelliCage has been developed, like the open-field consisting of an enclosed arena, but instead of being empty with various objects in it, also used for measuring activity over several months as the mouse's home-cage (Codita et al., 2010). As expected, both hypoactivity and hyperactivity have been reported in APP mutant mice relative to age-matched controls (Table 2).

Table 2.

Motor activity in open-field, Y-maze, or home-cage by transgenic mice expressing human mutated or wild-type (WT) APP genes with or without mutated PS1

Mouse name	Activity	References
APP695SWE (Tg2576)	↑ or normal	Arendash et al., 2004; Chapman et al., 1999; Comery et al., 2005; Deacon et al., 2008; Dineley et al., 2002; Gil-Bea et al., 2007; Gorman and Yellon, 2010; Holcomb et al., 1998; 1999; King and Arendash 2002; Lalonde et al., 2003; Massaad et al., 2009; Middei et al., 2004, 2006; Ognibene et al., 2005 Scholtzova et al., 2009; Tabuchi et al., 2009
APP695SWE (Tg2123H)	normal	Hsiao et al., 1995
APP695SWEch	normal	Savonenko et al., 2003
APP751SWE	↓ and ↑ or normal	Dumont et al., 2004; Heneka et al., 2006; Huang et al., 2006; Lalonde et al., 2002b; Senechal et al., 2008 Van Dam et al., 2003; Vloeberghs et al., 2004
APP695SWE/Thy1	normal	Richardson et al., 2003
APP770SWE	↓	Moechars et al., 1999b
APP751SWE/LD	↑ or ↓	Blanchard et al., 2009
APP/IND	↑ or ↓	Dodart et al., 1999; Gerlai et al., 2002 Huitron-Resendiz et al., 2002
APP695SWE/IND (TgCRND8 and Tg19959)	↑ or normal	Ambrée et al., 2006; Dumont et al., 2010; Görtz et al., 2008 Hyde et al., 2005; Lewejohann et al., 2009; Touma et al., 2004; Walker et al., 2011
APP695SWE/IND/tet	↑	Jankowsky et al., 2005b
APP751SWE/IND	normal	Lee et al., 2004
APP/SWE/IND/J20	normal	Galvan et al., 2006
APP/SWE/ARC	↑	Codita et al., 2010
APP695LD	↓	Moechars et al., 1999b
APP/RK	↓	Moechars et al., 1998
APP695WT/Mt2	↓	Yamaguchi et al., 1991
APP695WT/Thy1	↓	Moechars et al., 1999b
App695WT APP695mycWT, APP695TRImyc	normal	Hsiao et al., 1995
APP751WT	normal	D'Hooge et al., 1996
APP/C104	normal	Nalbantoglu et al., 1997
APP/C100-Flag	normal	Berger-Sweeney et al., 1999
APP/C100/IND	↑	Boon et al., 2010
APP/C100/WT	↑	Boon et al., 2010
APP/C99	↓	Lalonde et al., 2002a
APP695SWE+PS1/M146L	↑ or normal	Arendash et al., 2001; Holcomb et al., 1998, 1999; Sadowski et al., 2004
APP695SWE+PS1/M146L/Thy1	↓ normal	Pugh et al., 2007 Pardon et al., 2009
APP695SWEch+PS1/A246E	↓	Liu et al., 2002
APP751SWE/LD+PS1/M146L	↓	Le Cudennec et al., 2008
APP695SWE+PS1/A246E	normal	Dineley et al., 2002
APP695SWE/co+PS1/ΔE9	↑ or normal	Frye and Walf, 2008; Hartmann et al., 2010; Hoojimans et al., 2009; Lalonde et al., 2004, 2005 Timmer et al., 2010
APP751SWE/LD+PS1/M233+L235P	↑	Faure et al., 2011
APP751SWE+PS2/VG	normal	Richards et al., 2003
3X-Tg-AD	↓ ↑ or normal	Arsenault et al., 2011; Gimenez-Llort et al., 2007; Gulinello et al., 2009; Halagappa et al., 2007; Nelson et al., 2007; Ojha et al., 2011; Pietropaolo et al., 2008; Pietropaolo et al., 2009
APP knockout	↓	Müller et al., 1994; Tremml et al., 1998

↑=increase, ↓=decrease, normal=no difference vs controls; WT=wild-type

3.1.1 Hyperactivity

The most often examined Alzheimer model for motor activity is APP₆₉₅SWE (Tg2576), often found to be more active than non-transgenic controls in the open-field (Deacon et al., 2008, 2009; King and Arendash, 2002; Gil-Bea et al., 2007; Lalonde et al., 2003; Middei et al., 2006; Tabuchi et al., 2009). The same results were reported in the Y-maze (King and Arendash, 2002) and home-cage (Gorman and Yellon, 2010; Ognibene et al., 2005). A positive linear correlation was obtained between open-field and Y-maze values, indicating a common underlying factor (Arendash and King, 2002), one of which may be Aβ-mediated inflammatory responses, since the hyperactivity was reversed after administration of the anti-inflammatory agent, ibuprofen (Lim et al., 2001). However, other experimenters reported no change with this model in the open-field (Arendash et al., 2004; Chapman et al., 1999; Comery et al., 2005; Dineley et al., 2002; Massaad et al., 2009; Scholtzova et al., 2009) or Y-maze (Arendash et al., 2004; Holcomb et al., 1998, 1999; Middei et al., 2004). The age factor is the most likely cause of this discrepancy, since hyperactivity was evident from 14 to 21 months (Deacon et al., 2008, 2009; King and Arendash, 2002; Lalonde et al., 2003; Middei et al., 2006), but generally not at younger ages (Arendash et al., 2004; Chapman et al., 1999; Comery et al., 2005; Holcomb 1998, 1999; Massaad et al., 2009; Middei et al., 2004), except for significant results at 7 (Middei et al., 2006; Tabuchi et al., 2009) and 10 (Tabuchi et al., 2009) months and one negative result at 16 (Scholtzova et al., 2009), implying the additional influence of methodologic factors related to the apparatus, such as lighting conditions, or else animal characteristics, such as different sex distributions or genetic backgrounds, the latter being due to different husbandry practices.

Since the monogenic generally displays age-related hyperactivity, it is no surprise to learn that the APP₆₉₅SWE+PS1/M146L bigenic shows the same tendency, found in the Y-maze as early as 3 months of age, and also at 6 and 9 months (Holcomb et al., 1998, 1999), later confirmed at 5 and 15 months (Arendash et al., 2001). The latter group was also hyperactive in the open-field (Arendash et al., 2001). However, no change in activity was found with the same model evaluated at 8 and 22 months (Sadowski et al., 2004), reflecting task-mediated factors difficult at the present time to be ascertained.

Hyperactivity in the open-field has been reported in four other bigenic models, including APP₇₅₁SWE/LD +PS1/M233T+L235P (Faure et al., 2011), APP₆₉₅SWE/co+PS1/ΔE9 (line 85) (Hoojimans et al., 2009; Lalonde et al., 2005), APP₇₅₁SWE+PS2/VG (Richards et al., 2003), and 3xTg-AD (Gimenez-Llort et al., 2007; Pietropaolo et al., 2008). In 3xTg-AD mice obtained by co-injecting APP with the Swedish mutation and four-repeat Mapt with the P301L mutation into single-cell embryos of PS1/M146V knockins, hyperactivity was found at 6 (Gimenez-Llort et al., 2007) and 12 (Gimenez-Llort et al., 2007; Pietropaolo et al., 2008) months, but, as described below, the opposite pattern of hypoactivity was prominent at different ages. In the APP₆₉₅SWE/co+PS1/ΔE9 model, hyperactivity was reported at 12 (Lalonde et al., 2005) and 15 (Hoojimans et al., 2009) months of age, but the bigenics ran for longer distances only in the center region and not in the periphery of the open-field at the younger age (Lalonde et al., 2005). As with the APP₆₉₅SWE model, variable results are found in younger APP₆₉₅SWE/co+PS1/ΔE9 mice, with open-field hyperactivity detected at 8 months (Hoojimans et al., 2009), but not at 7 (Lalonde et al., 2004), 9 (Frye and Walf, 2008), or 10 (Hartmann et al., 2010). A negative finding was found even from 5 to 12 months, at least with respect to time spent walking (Timmer et al., 2010).

In view of the close neuropathologic resemblance between APP₆₉₅SWE (Tg2576) and APP₇₅₁SWE (APP23) models, it is not surprising to see hyperactivity in both, though once again an age-related pattern is discernable. Indeed, APP₇₅₁SWE mice were more active than controls in the open-field at 3 (Senechal et al., 2008), 10 (Heneka et al., 2006), and 24 months (Dumont et al., 2004). The same result was found in the home-cage from 3 to 12 months (Van Dam et al., 2003; Vloeberghs et al., 2004). But the reverse pattern of less activity in the open-field was reported at 6 (Van Dam et al., 2003) and 16 (Lalonde et al., 2002b) months, with no change in the Y-maze at 3 months (Huang et al., 2006). It is likely that the change from mid-age hypoactivity (Lalonde et al., 2002b) to old-age hyperactivity (Dumont et al., 2004) in the same laboratory is mediated by amyloid-related action on synaptic activity of neurotransmitters, possibly decreased transmission of gamma-amino butyric acid (GABA) as a function of age, since only the older transgenics exhibited home-cage myoclonus and epileptic seizures associated with this neruotransmitter. Likewise, hypoactivity was converted to hyperactivity as a function of aging in APP₇₅₁SWE/LD mutants from 2 to 19 months (Blanchard et al., 2009).

The finding of open-field hyperactivity is not limited to transgenic mice with the Swedish mutation, extending to the APP/IND model (PDAPP) with the Indiana mutation, this time at a very early age, either before (3 months of age) or after (6 and 9 months of age) Aβ plaque formation, the former being attributed to the direct action of mutated APP or to intracellular Aβ (Dodart et al., 1999). Three-month-old APP/IND mice were also hyperactive in their home-cage, although this effect disappeared at 20 months (Huitron-Resendiz et al., 2002). However, the reverse pattern of higher immobility time in this transgenic model was found at 11 months of age, though activity levels were not monitored as such (Gerlai et al., 2002).

Hyperactivity was also reported in two models with combined Swedish and Indiana mutations. As with the previous model, APP₆₉₅SWE/IND mutants (TgCRND8 or Tg19959) were hyperactive in the open-field either before or after Aβ plaque deposition (Dumont et al., 2010; Hyde et al., 2005; Walker et al., 2011) and also in their home-cage (Ambrée et al., 2006). However, no effect was found in the open-field at 3 (Touma et al., 2004) or 4 (Görtz et al., 2008) months of age, a time when plaques are already apparent, perhaps illustrating once again that the phenotype is more apparent in older mice, either because of the long-term action of Aβ in the brain or because of the gradual decline in motor activity as a function of aging in normal mice. Moreover, no change in home-cage activity was discerned in a new procedure with radiofrequency detection (Lewejohann et al., 2009). In any event, hyperactivity was found in a second model with Swedish and Indiana mutations and Aβ plaques, namely APP₆₉₅SWE/IND/tet mutants measured in their home-cage (Jankowsky et al., 2005b).

The same phenotype was discerned in a mutant with Swedish and Arctic mutations containing Aβ plaques, namely APP/SWE/ARC exposed for the first time to the IntelliCage, amounting to the open-field test (Codita et al., 2010). Moreover, since hyperactivity has been reported prior to Aβ plaque formation in models cited above, it is not surprising to find the same phenotype exhibited in two models which never develop Aβ plaques at all, namely APP/C100/IND and APP/C100/WT (Boon et al., 2010).

Since Aβ deposition or C100 overexpression in the limbic system is a prominent feature of most Alzheimer models, it is reasonable to hypothesize that neuronal dysfunction at this level contributes to hyperactivity. Indeed, hyperactivity has often been demonstrated in rats with lesions in the hippocampus (e.g. Roberts et al., 1962) or septum (Köhler and Srebro, 1980) relative to sham-operated controls. A second factor that should be further investigated is the coincident emergence of hyperactivity with myoclonic movements and seizure activity in APP₇₅₁SWE mice (Dumont et al., 2004). These results point to a possible contribution of impaired GABAergic transmission in the hyperactive phenotype, since GABA_A receptor antagonists increase motor activity in rats, for example picrotoxin injected at subconvulsive doses either in the periphery (Chang et al., 2004) or directly into the hippocampus (Bast et al., 2001), amygdala (Chang et al., 2004), or nucleus accumbens (Morgenstern et al., 1984).

It is probable that the hyperactive phenotype is more clearly discriminable in older mice, since the activity of normal mice declines with aging, as demonstrated in C57BL/6 (Dean et al., 1981; Sprott and Eleftheriou, 1974) and NMRI (Lamberty and Gower, 1990, 1992) strains. The phenotype may be obscured in younger mice because of a ceiling effect, i.e. the mutants cannot be demonstrated to be significantly more active because of the high values seen in non-transgenic controls, especially during brief assessments lasting 5 min, most often the case, and especially in particularly active strains such as C57BL/6.

3.1.2 Hypoactivity

In view of the apathy often displayed by patients with Alzheimer's disease (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004), one expects murine models to show hypoactivity, which may be attributed to Aβ-related changes in specific neurotransmitter systems. Moreover, since the App knockout is hypoactive (Müller et al., 1994; Tremml et al., 1998), different results with respect to hyper- or hypo-activity may be due to changes in the functioning of the endogenous Aβ peptide.

As expected, hypoactivity in the open-field has been described in several models, including two mutations with the Swedish mutation, APP₇₅₁SWE (as cited in the previous subsection) (Lalonde et al., 2002b; Van Dam et al., 2003) and APP₇₇₀SWE (Moechars et al., 1999b), one with the London mutation, APP₆₉₅LD (Moechars et al., 1999b), and one with both Swedish and London mutations, APP₇₅₁SWE/LD (Blanchard et al., 2009), together with four multigenic models, namely APP₆₉₅SWEch+PS1/A246E (Liu et al., 2002), APP₇₅₁SWE/LD+PS1/M146L (Le Cudennec et al., 2008), APP₆₉₅SWE+PS1/M146L/Thy1 (Pugh et al., 2007), and 3xTg-AD (Arsenault et al., 2011; Gulinello et al., 2009; Halagappa et al., 2007; Nelson et al., 2007; Pietropaolo et al., 2009). Also as expected, in view of findings with the hyperactive phenotype, this pattern appears age-related, at least with respect to APP₆₉₅SWE+PS1/M146L/Thy1 mutants, found to be hypoactive at 10 months of age (Pugh et al., 2007) but not at 4 (Pardon et al., 2009).

More complicated and difficult to be reconciled results have been obtained with the 3xTg-AD model. As mentioned in the previous subsection, hyperactivity was found at 6 (Gimenez-Llort et al., 2007) and 12 (Gimenez-Llort et al., 2007; Pietropaolo et al., 2008) months of age. Yet the opposite pattern of hypoactivity was noted at 6 (Pietropaolo et al., 2009) and 7 (Nelson et al., 2007) months, as well as 10 and 17 (Halagappa et al., 2007), 12 (Arsenault et al., 2011), or 15 (Gulinello et al., 2009) months, with no effect found at 8 (Ojha et al., 2011). The blatant discrepancies at 6 and 12 months may be due to test duration, as Gimenez-Llort et al. (2007) noted initial hypoactivity followed by a global pattern of hyperactivity. A more interesting neurobiological explanation concerns age-related changes in neurotransmitter systems yet to be uncovered. It is possible that the initial pattern of hyperactivity is converted to hypoactivity as a function of age, since hypoactivity but not hyperactivity has so far been reported in mice above 15 months.

As found with models showing hyperactivity, hypoactivity is also a prominent feature of models without Aβ plaques, notably APP/RK (Moechars et al., 1998), APP₆₉₅WT/Thy1 (Moechars et al., 1999b), APP₆₉₅WT/Mt2 (Yamaguchi et al., 1991), and APP/C99 (Lalonde et al., 2002a). These data point towards direct Aβ-mediated changes on neural activity as being reponsible for the hypoactive phenotype.

It remains to be determined which neurobiologic systems are responsible for amyloid-mediated hypoactivity. In view of the presence of Aβ plaques or APP overexpression in neocortex, it is reasonable to suppose a role for the prefrontal cortex in mediating this phenotype, since anomalies in prefronto-cortical-basal ganglia circuits are known to cause apathy (Levy and Dubois, 2006). Future experimentation should be guided by determining the association between hypoactivity and anomalies in prefrontal cortex-mediated functions. Another point to be considered, as mentioned in the previous subsection, is that mouse activity declines as a function of aging (Dean et al., 1981; Lamberty and Gower, 1990, 1992; Sprott and Eleftheriou, 1974). On one hand, the hypoactive phenotype may be more evident in younger mice, because they are more active at this age. On the other hand, it may take time before Aβ-mediated processes occur in prefrontal cortex or other brain regions, mitigating against the chance of finding an effect at a young age.

3.1.3 Unchanged activity

Since either hypo- or hyper-activity has been found in the open-field, sometimes with the same model (APP₇₅₁, APP₇₅₁SWE/LD+PS1/M146L, 3xTg-AD), it is not surprising to find no change in activity in several models, some with Aβ plaques, namely APP₆₉₅SWE/Thy1 (Richardson et al., 2003), APP₆₉₅SWEch (Savonenko et al., 2003), APP₇₅₁SWE/IND (Lee et al., 2004), APP/SWE/IND/J20 (Galvan et al., 2006), APP₆₉₅SWE+PS1/A246E (Dineley et al., 2002)., and APP/C104 (Nalbantoglu et al., 1997), some without Aβ plaques, namely APP₇₅₁WT (D'Hooge et al., 1996) and APP/C100-Flag (Berger-Sweeney et al., 1999). Likewise, Y-maze activity was normal in APP/SWE/IND/J20 mice with Aβ plaques (Galvan et al., 2006) and APP₆₉₅mycWT, APP₆₉₅TRImyc, App₆₉₅WT, and APP₆₉₅SWE (Tg2123H) mice without Aβ plaques (Hsiao et al., 1995). In most cases, mice were tested at a single age level, thus mitigating against the chance of uncovering a significant effect.

3.1.4 Conclusions

Changes in activity levels irrespective of the direction are concordant with clinical findings of the behavioral and psychological symptoms of dementia (BPSD) (Lanari et al., 2006), apathy and agitation being prominent features of patients with Alzheimer's disease. Although such gross behavioral changes often occur late in dementia progression, these are correlated with caregiver compliance (Senanarong et al., 2004), and for this reason there is an ethical advantage if not a necessity in evaluating exploratory activity in murine models, including during treatment trials, usually limited to learning abilities.

In two murine models, APP₇₅₁SWE (Dumont et al., 2004; Lalonde et al., 2002b) and APP₇₅₁SWE/LD (Blanchard et al., 2009), age-related changes occurred from hypo- to hyper-activity. Similar data seem to apply with 3xTg-AD mice (Arsenault et al., 2011; Gimenez-Llort et al., 2007; Gulinello et al., 2009; Halagappa et al., 2007; Nelson et al., 2007; Pietropaolo et al., 2008, 2009), in need to be re-examined at a wider range within the same study to determine which occurs first. The most likely explanation for these findings is that amyloid exerts different neurochemical changes depending on age. We hypothesize that hyperactivity is the result of decreased transmission of GABAergic pathways, all the more likely in that APP₇₅₁SWE mutants are susceptible to age-related myoclonus and convulsions linked with this neurotransmitter.

Age-related patterns in normal mice are likely to be relevant in determining the sensitivity of the mutation with respect to motor activity. The motor activity of normal mice decrease with age (Dean et al., 1981; Lamberty and Gower, 1990, 1992; Sprott and Eleftheriou, 1974). Thus, if the predominant phenotype in the transgenic is hyperactivity, it should be more apparent in older mice. But if the predominant phenotype is hypoactivity, it should be more apparent in younger mice.

3.2 Spontaneous alternation

After being placed in the stem of a Y- or T-maze and entering one of its arms, spontaneous alternation occurs on the subsequent trial when a mouse enters the opposite arm (Dember and Fowler, 1958; Lalonde, 2002). A cross-maze version may be used with two stems instead of one, with the mouse placed at either end (Jawhar et al., 2011). The spontaneous alternation test estimates the willingness to explore novel stimuli, or to avoid familiar stimuli. In view of the apathy often exhibited by patients with Alzheimer's disease (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004), one may predict reduced alternation rates in murine models. A second reason alternation rates may decline is the result of behavioral dishinbition or a loss in short-term memory, signs also manifest in Alzheimer dementia. In the former case, mice may perseverate as a result of impulsive behavior, choosing the first available arm without considering its previous choice.

As shown in Table 3, APP₆₉₅SWE mutants (Tg2576) have often been described as being deficient in the spontaneous alternation test (Arendash et al., 2004; Deacon et al., 2008; Fukumoto et al., 2010; Holcomb et al., 1998, 1999; Hsiao et al., 1996; King and Arendash, 2002; Lalonde et al., 2003; Middei et al., 2004; Ognibene et al., 2005; Ohno et al., 2004). Logically enough, the same result applies to the APP₆₉₅SWE+PS1/M146L bigenic derived from the monogenic (Holcomb et al., 1998, 1999). However, no such deficit in the bigenic was found in one study (Arendash et al., 2001), possibly due to the use of a different genetic background. Indeed, different normal mouse strains differ quite importantly in terms of alternation rates (Bertholet and Crusio, 1991).

Table 3.

Spontaneous alternation by transgenic mice expressing human mutated or wild-type (WT) APP genes with or without mutated PS1

Mouse name	Alternation rate	References
APP695SWE Tg2576	↓	Arendash et al., 2004; Deacon et al., 2008; Fukumoto et al., 2010; Holcomb et al., 1998, 1999; Hsiao et al., 1996; King and Arendash, 2002; Lalonde et al., 2003; Middei et al., 2004; Ognibene et al., 2005; Ohno et al., 2004
APP695SWE/Thy1	↓ or normal	Richardson et al., 2003
APP751SWE mice APP23	↓ or normal	Dumont et al., 2004 Huang et al., 2006; Lalonde et al., 2002b
APP/SWE/IND/J20	↓	Galvan et al., 2006
APP695SWE/IND	normal	Hyde et al., 2005
APP751WT	↓	Moran et al., 1995
APP/C100/IND	↓	Boon et al., 2010
App695WT	↓	Hsiao et al., 1995
APP695TRImyc	↓	Hsiao et al., 1995
APP695SWEch	normal	Savonenko et al., 2003
APP/C100/WT	normal	Boon et al., 2010
APP/C99	normal	Lalonde et al., 2002a
APP695SWE+PS1/M146L	↓ or normal	Arendash et al., 2001; Holcomb et al., 1998, 1999
APP751SWE+PS1/G384A	↓	Busche et al., 2008
APP695SWEch+PS1/A246E	↓	Filali and Lalonde, 2009; Filali et al., 2011
APP695SWE/co+PS1/ΔE9 Line 85	normal	Bonardi et al., 2011; Frye and Walf, 2008; Harrison et al., 2009a,b; Lalonde et al., 2004, 2005
APP751SWE+PS2/VG	↓ or normal	Richards et al., 2003
3x-Tg-AD	↓	Carroll et al., 2007; Pietropaolo et al., 2009
5xFAD	↓	Jawhar et al. 2011; 2012 Oakley et al., 2006

↓=decrease; normal=no difference vs controls; WT=wild-type

Spontaneous alternation rates declined relative to controls in a second APP model with the Swedish mutation causing an accumulation of Aβ plaques, APP₆₉₅SWE/Thy1 (Richardson et al., 2003). This occurred at 12 and 18 months of age, but not at 24. The negative result at 24 months is probably due the reduced rate of controls relative to that of younger mice. Indeed, a lower alternation rate as a function of aging is a common finding in normal mouse strains (Lalonde, 2002). A reduction in spontaneous alternation rates appeared in a third APP model with the Swedish mutation causing an accumulation of Aβ plaques, APP₇₅₁SWE mice (APP23), found at 3 months of age (Huang et al., 2006), as were APP₇₅₁SWE+PS1/G384A bigenics derived from this monogenic at 6 months of age (Busche et al., 2008). But no deficit was apparent in APP₇₅₁SWE mice at 16 and 24 months of age, presumably at least in part because of the low values obtained in controls (Dumont et al., 2004; Lalonde et al., 2002b). Likewise, spontaneous alternation rates were lower in APP₇₅₁SWE+PS2/VG mutants than controls at 4 months of age, but not at 8, 12, and 16 despite increasing plaque load (Richards et al., 2003). Thus, unlike the hyperactive phenotype discussed in section 3.1.1, this assay appears to lose sensitivity as a function of aging, the low rate in controls being liable to cause a floor effect, i.e. a statistically significant result is unlikely because the values of controls are too low.

Nevertheless, four other Aβ plaque-containing models have been shown to be impaired in this test, namely APP/SWE/IND/J20 (Galvan et al., 2006), APP₆₉₅SWEch+PS1/A246E (Filali and Lalonde, 2009), 3x-Tg-AD (Carroll et al., 2007; Pietropaolo et al., 2009), and 5xFAD (Jawhar et al. 2011; Oakley et al., 2006). However, no such impairment was found in two others carrying Aβ plaques: APP₆₉₅SWE/IND (Hyde et al., 2005) and APP₆₉₅SWE/co+PS1/ΔE9 (Bonardi et al., 2011; Frye and Walf, 2008; Harrison et al., 2009a,b; Lalonde et al., 2004, 2005). These negative results appear despite findings of alternation deficits in four models without Aβ plaques, namely APP₇₅₁WT (Moran et al., 1995), APP/C100/IND (Boon et al., 2010), App₆₉₅WT (Hsiao et al., 1995), and APP₆₉₅TRImyc (Hsiao et al., 1995), though not in three others: APP₆₉₅SWEch (Savonenko et al., 2003), APP/C100/WT (Boon et al., 2010), and APP/C99 (Lalonde et al., 2002a).

The positive findings in mutants with or without Aβ plaques may be explained by amyloid-related dysfunction in brain regions critical for spontaneous alternation, including the hippocampus (Lalonde, 2002). Indeed, hippocampal lesions in rats have often been found to cause spontaneous alternation deficits relative to sham-operated controls (Johnson et al., 1977; Kirkby et al., 1967; Stevens and Cowey, 1973). Identical results have been reported in rats with lesions in the septum (Clody and Carlton, 1969), basal forebrain magnocellular nucleus (Murray and Fibiger, 1986), anterior thalamus (Aggleton et al., 1995), medial prefrontal cortex (Divac et al., 1975), and prelimbic cortex (Delatour and Gisquet-Verrier, 1996). These brain areas either accumulate amyloid in mutants or are liable to be dysfunctional as a result of interrupted brain circuitries secondary to amyloid accumulation, mainly involving the limbic system. The negative findings may be due to insufficient amyloid accumulation in these brain regions, age-related declines in controls, causing a floor effect, or to the use of a genetic background with non-optimal alternation rates.

3.2.2 Conclusions

The spontanenous alternation assay appears sensitive to amyloid loading in mice. These results may be due to an unwillingness to explore novel stimuli or to disinhibitory tendencies, akin to the apathy or disinhibition reported in patients with Alzheimer's disease (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004). Alzheimer patients are liable to perseverate in all sorts of behavior, either because of a lack of initiative in trying out new behaviors, a lack in curiosity, neophobia, or because of a tendency to choose impulsively actions often repeated in the past. Likewise, mice may perseverate as a result of a lack in curiosity, a fear of novel stimuli, or impulsiveness. One way to distinguish between lack in curiosity or impulsiveness as the underlying cause of perseveration is to measure choice latencies, the time taken before mice choose to enter left or right maze arms. One would expect higher choice latencies in apathetic mice and lower latencies in impulsive mice. In one of the rare experiments where this was done, APP₆₉₅SWE mutants deficient in spontaneous alternation did not differ from controls in choice latencies despite being more active in the open-field (Lalonde et al., 2003). On one hand, this negative result may indicate that another factor was involved, such a loss in short-term memory. On the other hand, it is possible that multiple factors are involved, including a lack in motivation to explore and disinhibition, so that choice latencies might even out. In any event, this measure should be used more often as a way of gauging possible underlying causes.

Amyloid-related dysfunction in the limbic system and medial prefrontal cortex may be responsible for the alternation deficits in mutant mice. However, positive findings appear to occur within a narrow time-frame in the mouse's lifetime. Because deficits have been obtained at an early age in mutants with or without Aβ plaques, it seems that early testing is the optimal stragegy to be used for this assay. Aside from age, the absence of a deficit may be caused by differences in regional distribution of Aβ or to the use of a genetic background alternating at low rates.

3.3 Elevated plus-maze

Patients with Alzheimer's disease are subject to anxiety, but sometimes also to the opposite tendency of disinhibition (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004). One may therefore expect either hypoanxious or hyperanxious phenotypes in murine models of this condition. Anxiety levels may be examined in the elevated plus-maze, containing two wall-enclosed arms opposite two arms without walls (Cole and Rodgers, 1993). Open arm entries and duration reflect exploratory activity in an anxiogenic area, whereas enclosed arm entries and duration reflect exploratory activity in a safer one, the ratio between the two providing a specific measure of anxiety. This test has been pharmacologically validated in mice by comparing benzodiazepines versus placebo (Cole and Rodgers, 1993), though dependent on their genetic background (Rodgers et al., 2002). Table 4 illustrates APP mice that have been evaluated in the elevated plus-maze.

Table 4.

Plus-maze exploration by transgenic mice expressing human mutated or wild-type (WT) APP genes with or without mutated PS1

Mouse	Anxiety	References
APP695SWE Tg2576	↓ or normal	Arendash et al., 2004; Gil-Bea et al., 2007; Lalonde et al., 2003; Lassalle et al., 2008; Ognibene et al., 2005; Tabuchi et al., 2009
APP751SWE mice APP23	↓ or normal	Dumont et al., 2004; Lalonde et al., 2002b Vloeberghs et al., 2007
APP695SWEch	↓ or normal	Savonenko et al., 2003
APP695SWE/IND	↓ or normal	Dumont et al., 2010; Görtz et al., 2008 Touma et al., 2004
APP/SWE/IND/J20	↓	Chin et al., 2005
APP/SWE/IND/J9	normal	Chin et al., 2005
APP751SWE/IND	normal	Lee et al., 2004
APP751WT	normal	Moran et al., 1995
APP/C100/IND	normal	Boon et al., 2010
APP/C100/WT	normal	Boon et al., 2010
APP/C99/IND	↑	Lee et al., 2006
APP/C99	normal	Lalonde et al., 2002a
APP695SWE+PS1/M146L	normal	Arendash et al., 2001
APP695SWE+PS1/M146L/Thy1	↓	Pugh et al., 2007
APP751SWE/LD+PS1/M233+L235P	↓	Cotel et al., 2010; Faure et al., 2011
APP751SWE/LD	normal	Blanchard et al. 2009; Le Cudennec et al., 2008
APP695SWEch+PS1/A246E	↑ or normal	Puoliväli et al., 2002; Filali et al., 2011
APP695SWE/co+PS1/ΔE9 Line 85	↓	Lalonde et al., 2004, 2005; Reiserer et al., 2007
3x-Tg-AD	↑ or normal	Pietropaolo et al., 2008, 2009
5xFAD	↓	Jawhar et al, 2011;2012

↑=increase, ↓=decrease, normal=no difference vs controls; WT=wild-type

3.3.1 Hypoanxious APP mice

Relative to non-transgenic controls, APP₆₉₅SWE mutants (Tg2576) had higher open/total arm entries and duration at 17 months of age (Lalonde et al., 2003). The increase in open arm duration occurred in the same mutant at the same age on three different genetic backgrounds (Lassalle et al., 2008). The same increase in open arm exploration occurred at 7 months of age (Gil-Bea et al., 2007; Ognibene et al., 2005; Tabuchi et al., 2009), a time when Aβ plaques begin to form, but not at 5 months (Arendash et al., 2004).

A similar increase in open arm exploration was reported in seven other models with parenchymal Aβ plaques: APP₇₅₁SWE (APP23) (Dumont et al., 2004), APP₆₉₅SWE/IND (Dumont et al., 2010), APP/SWE/IND/J20 (Chin et al., 2005), APP₆₉₅SWE/co+PS1/ΔE9 (Lalonde et al., 2004, 2005; Reiserer et al., 2007), APP₆₉₅SWE+PS1/M146L/Thy1 (Pugh et al., 2007), APP₇₅₁SWE/LD+PS1/M233T+L235P (Cotel et al., 2012; Faure et al., 2011), and 5xFAD (Jawhar et al, 2011, 2012). On two of these, age-dependent effects were found. APP₆₉₅SWE/IND mutants explored the open arms more than controls at 8 months of age (Dumont et al., 2010), but, despite the presence of plaques at this age, not at 3 (Touma et al., 2004) or 4 (Görtz et al., 2008) months. Likewise, the open arm exploration of APP₇₅₁SWE mice increased at 24 months of age (Dumont et al., 2004), but not at 3, 6, 12 (Vloeberghs et al., 2007), or 16 (Lalonde et al., 2002b) months. These results may be due to insufficient Aβ load in the younger animals, a notion supported by the finding that open arm duration increased in the APP/SWE/IND/J20 strain with high but not in the closely associated APP/SWE/IND/J9 strain with lower amyloid load (Chin et al., 2005).

3.3.2 Hyperanxious APP mice

In contrast to the previous models, APP₆₉₅SWEch+PS1/A246E mice containing Aβ plaques had lower open arm durations than controls at 14 months of age (Puoliväli et al., 2002), though not at 6 months (Filali et al., 2011). The same pattern of elevated anxiety levels in this paradigm was reported in 3x-Tg-AD mice with Aβ plaques, though this result was found only in a female (Pietropaolo et al., 2009) and not a mixed (Pietropaolo et al., 2008) subgroup. A hyperanxious phenotype was also revealed in two other APP models, whose brains this time do not contain Aβ plaques, as open arm entries and duration diminished in APP/C99/IND mice (Lee et al., 2006), as did open arm entries in APP₆₉₅SWEch mice, though in the latter only in females on one of two non-congenic lines (Savonenko et al., 2003).

3.3.3 Unchanged anxiety levels in APP mice

Anxiety levels were no different from non-transgenic controls in APP/C99 mutants not harboring Aβ plaques, as only enclosed arm entries decreased (Lalonde et al., 2002a). In a second model without Aβ plaques, APP₇₅₁SWE/IND, both enclosed and open arm entries decreased relative to controls (Lee et al., 2004). Three other models without Aβ plaques did not differ from controls in in this test, namely APP/C100/IND (Boon et al., 2010), and APP/C100/WT (Boon et al., 2010), and APP₇₅₁WT (Moran et al., 1995), as well as two models with Aβ plaques, namely APP₆₉₅SWE+PS1/M146L (Arendash et al., 2001) and APP₇₅₁SWE/LD (Blanchard et al. 2009; Le Cudennec et al., 2008).

3.3.4 Conclusions

Twelve APP models have been shown to be sensitive to anxiety levels in the elevated plus-maze, concordant with the usefulness of this test in modelling the behavioral and psychological symptoms of dementia (BPSD) (Lanari et al., 2006). As with the open-field test, results were found in either direction, in the form of reduced anxiety in eight cases and the opposite pattern in the other four. Unlike the open-field test, no single mutant has yet been reported to display both increases and decreases in exploratory behaviors as a function of age.

Changes in anxiety levels in the elevated plus-maze may be ascribed to alterations in the neurotransmission of GABA. Indeed, benzodiazepines increased open arm exploration in normal mice (Cole and Rodgers, 1993; Griebel et al., 2000; Lalonde and Strazielle, 2010), as did muscimol, the GABAA receptor agonist, injected either peripherally (Dalvi and Rodgers, 1996) or at the level of the central nucleus of the amygdala (Moreira et al., 2007). Moreover, a reduction in anxiety levels was found in rats injected with a benzodiazepine directly into the hippocampus (Menard and Treit, 2001). 5HT-related pathways are also involved on the basis of findings of reduced anxiety levels in mice exposed to the elevated plus-maze following injections of either 5HT_1A (Cao and Rodgers, 1997; Lalonde and Strazielle, 2010; Nunes-de-Souza et al., 2000) or 5HT₂ (Nic Dhonnchadha et al., 2003) receptor agonists. In particular, a reduction in anxiety levels was found in rats injected with a 5HT1A receptor agonist in the hippocampus (Menard and Treit, 1998). Moreover, lesion studies in rats have determined the importance of the septum in this test, as lesions of this brain region increased open arm exploration relative to sham-operated controls (Treit and Menard, 1997). Thus, altered circuitries in the limbic system, comprising hippocampus, septum, and amygdala, liable to Aβ accumulation in most Alzheimer mutants, are likely to be involved in changes of elevated plus-maze behavior. In addition, the influence of the neocortex in these behaviors is underlined by the finding of increased open arm exploration in rats with lesions of the medial prefrontal cortex (Gonzalez et al., 2000).

3.4 Emergence

The emergence test is a second method to screen anxiety in rodents (Holmes, 2001). Mice are placed inside a small enclosure and latencies before emerging into a larger anxiogenic one are measured. As with the elevated plus-maze, this test has been pharmacologically validated in mice injected with a benzodiazepine (Lalonde and Strazielle, 2010). The assay is also sensitive to gene inactivation of corticotropin releasing factor (Smith et al., 1998), 5HT (Tecott et al., 1998), and dopamine (Dulawa et al., 1999) receptors, as well as the GT2 subtype of GTPase activating protein (Schmalzigaug et al., 2009).

To our knowledge, this test has only been used once among APP models. APP₇₅₁SWE mice had equivalent emergence latencies relative to non-transgenic controls despite differing from them in the elevated plus-maze, indicating that the two tests reflect different facets of anxiety (Dumont et al., 2004; Lalonde et al., 2002b).

3.5 Aggression

Increased hostility and aggression are part of the BPSD described in Alzheimer dementia (Lanari et al., 2006). Aggression in male mice may be evaluated by the resident-intruder test, whereby a mouse isolated for several weeks is exposed to an intruder in its home cage (File, 1980). Female mice cannot be used in this test because they do not attack intruders in their territory, defending only their young. Male APP₆₉₅SWE mice (Tg2576) had shorter latencies before the first attack towards an intruder and initiated more attacks than age-matched male controls (Alexander et al., 2011). Likewise, APP/DU/Thy1 and APP/FL/Thy1 residents with amyloid angiopathy but without Aβ plaques in parenchymal regions attacked more often and more quickly intruders than non-transgenic controls (Kumar-Singh et al., 2000). Aggression levels also increased in APP₆₉₅SWE+PS1/M146L/Thy1 mice (TgTASTPM) with Aβ plaques (Pugh et al., 2007) and APP/RK mice without Aβ plaques (Moechars et al., 1996) exposed to the same test. Thus, this test appears useful in the development of anti-aggressive drugs in dementia syndromes, in mice with or without Aβ plaques as well as those with amyloid angiopathy.

4. General conclusions

Willingness to initiate various behaviors, curiosity, anxiety, mood, and aggression are liable to be altered in dementia syndromes including Alzheimer's disease (Chung and Cummings, 2000; Daffner et al., 1992; Frisoni et al., 1999; Hart et al., 2003; Lanari et al., 2006; Lyketsos et al., 2002; Petry et al., 1988; Senanarong et al., 2004). Changes in the behavioral and psychological symptoms of dementia (BPSD) seen in Alzheimer patients (Lanari et al., 2006) can be mimicked in murine models. In particular, open-field activity, Y- or T-maze spontaneous alternation, exploration in the elevated plus-maze, and aggression are susceptible to be modified in APP mice relative to controls. However, changes in the open-field and elevated plus-maze occur in either direction. In some models, sometimes depending on age, open-field activity is increased, in others decreased. Likewise, open arm exploration in the elevated plus-maze may increase or descrease. The underlying reasons have only begun to be evaluated. There is a lack of in-depth analysis of neurotransmitter changes in association with those in behavior as a result of amyloid accumulation, in particular for GABA, 5HT, and dopamine.

Highlights.

/APP mutants display BPSD/

/APP mutants are impaired in spontaneous alternation/

/APP mutants display hyperactivity or hypoactivity/

/APP mutants display hyperanxiety or hypoanxiety/

/APPmutants show increased aggression/