A

tmilltdaa©

K

1

A(bbTarAfrcTf

0d

Neurobiology of Aging 32 (2011) 590–603

Sex difference in pathology and memory decline inrTg4510 mouse model of tauopathy

Mei Yue 1, Amanda Hanna 1, Judith Wilson, Hanno Roder, Christopher Janus ∗Department of Neuroscience, Mayo Clinic, 4500 San Pablo Road, Birdsall Bld., R215, Jacksonville, FL 32224, USA

Received 1 December 2008; received in revised form 25 March 2009; accepted 10 April 2009Available online 7 May 2009

bstract

Abnormal phosphorylation of tau protein is a common event in many neurodegenerative disorders, including Alzheimer’s disease and otherauopathies. We investigated the relationship between hyperphosphorylated tau in brain extracts and mnemonic functions in rTg4510 mouse

odel of tauopathy. We report that rTg4510 mice showed rapid deterioration in spatial learning and memory, which paralleled a significantncrease of hyperphosphorylated tau in the brain between 3 and 5.5 months of age. At 5.5 months, rTg4510 females showed significantly higherevels of hyperphosphorylated tau than males, with no evidence of differential tau transgene expression between the sexes. The increasedevels of hyperphosphorylated tau in females were associated with more severe impairment in spatial learning and memory as compared toransgenic males. We also showed that within studied age range, the decrease in memory performance was accompanied by other behavioral

isturbances in the water maze related to search strategy, like thigmotaxic swim and cue response. These findings suggest that the onset ofbnormal tau biochemistry and coincident cognitive deficits in the rTg4510 mouse model is sex-dependent with females being affected earliernd more aggressively than males.

2009 Elsevier Inc. All rights reserved.

differe

fi(ta�p2epAn(q

eywords: Tau; Transgenic mice; Neuronal loss; Memory impairment; Sex

. Introduction

A group of neurodegenerative diseases, includinglzheimer’s disease (AD), collectively termed tauopathies

Iqbal et al., 2005; Lee et al., 2001), is characterisedy age-progressing dementia, profound neuronal loss andrain atrophy (Cotman and Su, 1996; Davies et al., 2005;erry, 2006) with coinciding intra-neuronal accumulation ofggregated hyperphosphorylated tau protein in a form of neu-ofibrillary tangles (NFTs) (Dickson, 2003; Lee et al., 2001).dditionally to genetic mutations implicated in the onset of

amilial cases of AD (FAD) (Selkoe, 1997), there are otherisk factors which can significantly affect the onset and the

ourse of the disease (Erten-Lyons et al., 2009; Rhodin andhomas, 2001; Williams et al., 2006). Among the genetic

actors, the �4 allele of apolipoprotein E (APOE) was identi-

∗ Corresponding author. Tel.: +1 904 953 6414; fax: +1 904 953 7370.E-mail address: janus.christopher@mayo.edu (C. Janus).

1 These authors contributed equally to this work.

Arad

aJ

197-4580/$ – see front matter © 2009 Elsevier Inc. All rights reserved.oi:10.1016/j.neurobiolaging.2009.04.006

nce

ed and documented to be associated with late-onset of ADArendt et al., 1997; Corder et al., 1995a,b). However, the dis-ribution and prevalence of risk factors varies between sexesnd age groups (Azad et al., 2007). The presence of APOE-4 appears to have greater deleterious effect on hippocampusathology and memory in women than men (Fleisher et al.,005), however this risk varies between ethnic groups (Farrert al., 1997). The frequency of APOE-�4 is also higher inrogressive supranuclear palsy (PSP) which has concomitantD pathology, and APOE-�4 increases risk of dementia aftereurotrauma, ischemia, or human immunodeficiency virusHIV) infection (reviewed by Raber, 2004), however the fre-uency distribution related to gender has not been identified.lthough, female sex has been also associated with increased

isk of hypertension, hyperlipidemia, and diabetes (Azad etl., 2007), its linkage to dementias in other neurodegenerative

iseases is yet to be established.

The discovery of over thirty mutations in microtubulessociated protein tau (MAPT) (reviewed by Goedert andakes, 2005) in patients with fronto-temporal dementia with

gy of A

PsHdiilamhd2wt1

cttdel2aswos(taii2ibdrphmg(fl(mgAiwroMom

ef

bomal

2

2

olmttag1Ci(etmhtgtcrgiG1sd2tt(Ntsttao

M. Yue et al. / Neurobiolo

arkinsonism linked to chromosome 17 (FTDP-17), an auto-omal dominantly inherited tauopathy (Clark et al., 1998;utton et al., 1998), and the evidence that intra-neuronaleposition of tau is also a feature of dying neurons dur-ng normal aging (Braak and Braak, 1997) underscored themportance of abnormally modified tau in the neuronal deatheading to cognitive deficits. Hyperphosphorylated tau withcharacteristic abnormal gel mobility is a biochemical hall-ark uniformly present in many tauopathies including AD,

ippocampal tauopathy in cerebral aging, PSP, corticobasalegeneration (CBD), and FTDP-17 (reviewed by Buee et al.,000; Sergeant et al., 2005). It has also been shown that NFTsere better correlated with the degree and duration of demen-

ia in AD than amyloid beta (A�) plaques (Arriagada et al.,992; Berg et al., 1993).

The identification of pathogenic tau mutations enabled thereation of mouse models of tauopathies which confirmedhat the accumulation of aggregated hyperphosphorylatedau protein is directly implicated in neurodegeneration andementia (reviewed by Eriksen and Janus, 2006). The over-xpression of mutated human tau genes in these modelsed to the development of functional motor (Lewis et al.,000), or cognitive (Santa Cruz et al., 2005; Schindowski etl., 2006) deficits, depending on part of the central nervousystem targeted by transgenic tau expression. In our studye used the rTg4510 repressible model of tauopathy, whichver-expresses human P301L mutant tau linked to FTDP-17pecifically in the cortex, limbic system, and basal gangliaSanta Cruz et al., 2005). The model is characterised by spa-ial memory deficits and the formation of a distinct 64 kDabnormally hyperphosphorylated 4R0N isoform of tau andncrease in NFTs, with a rapidly progressing neuronal lossn the hippocampus by 5.5 months of age (Ramsden et al.,005; Santa Cruz et al., 2005; Spires et al., 2006). Whilenvestigating the possibility of circumstantial effects on theehavioral phenotype of this model, we identified a profoundifference between the susceptibility of females and males ofTg4510 mice to tau pathology and corresponding behavioralhenotypes. Sex-dependent differentiation in spatial memoryas been noted before in more complex transgenic mouseodels co-expressing both APP and tau genes. Female trans-

enic APP/tau mice showed worse spatial learning than malesRibé et al., 2005). Separately, in a second model, APP/tauemales showed enhanced neurofibrillary pathology in theimbic system and olfactory cortex as compared to malesLewis et al., 2001). Also, recent studies using a 3 × Tgouse model (co-expressing mutant APP, PS1 and tau trans-

enes) showed that females had significantly exacerbated� pathology over males, but with no gender differences

n tau pathology (Hirata-Fukae et al., 2008). In our study,e characterised spatial reference memory of 5.5-month-old

Tg4510 mice since this memory system is highly dependent

n intact hippocampus (Eichenbaum, 1996; Morris, 1984;orris et al., 1982; Squire, 1992). The significant degree

f neuronal loss in hippocampus already manifest in the 5.5-onth-old rTg4510 (Spires et al., 2006) provides an attractive

sg1e

ging 32 (2011) 590–603 591

xperimental time frame to focus on differences in memoryunction specifically due to tau pathology.

We demonstrate for the first time a direct interactionetween sex and tau pathology. The sex-dependent onsetf the development of tau pathology and compromisednemonic function in the rTg4510 mice may present an

ttractive model to study factors external to the CNS modu-ating the onset and progression of dementia.

. Methods

.1. Mice

The generation of rTg4510 mice was described previ-usly (Santa Cruz et al., 2005). Briefly, a human tau cDNAacking exons 2 + 3 but containing exon 10 with the P301L

utation (4R0N tauP301L) was placed downstream of aetracycline-operon-responder (TRE) construct. To activatehe transgene, the responder has to be co-expressed with anctivator construct, consisted of the tetracycline conditionalene expression system (tet-off, tTA) (Gossen and Bujard,995). The tTA activator system was placed downstream ofa2+-calmodulin kinase II promoter (CaMKII) thus restrict-

ng the expression of TRE mainly to forebrain structuresMayford et al., 1996). The tau transgene responder wasxpressed in the FVB/N (Charles River) mouse strain, whilehe tTA activator system was maintained on 129 S6 (Taconic)

ouse strain. Both congenic parental mouse strains wereemizygous with reference to tau and tTA transgenes, andheir F1 progeny carried both responder and activator trans-enes (tTA/tau, henceforth called rTg4510), necessary forhe expression of the tau transgene, along with littermatesarrying single transgenes (tau responder or tTA activator,espectively) and non-transgenic mice. The FVB/129 F1enetic background of rTg4510 mice and their littermatess suitable for behavioral testing (Banbury Conference onenetic Background in Mice, 1997; Crawley and Paylor,997), being unaffected by recessive traits of parental strainsuch as retinal degeneration present in FVB strain or partialevelopment of corpus callosum in 129 strain (Wahlsten et al.,005). The analysis of the frequency of all the genotypes at theime of weaning (21 + 2 days of age; N = 120) revealed closeo predicted (25%) distribution of pups within each genotype29% tTA/tau, 25% tau, 21% tTA, 25% non-Tg; χ2(3) = 1.7,S). Mice were genotyped by the analysis of tail DNA using

au cDNA-specific primers to exon 1 and exon 5, and primerspecific to tetracycline trans-activator against an internal con-rol (T cell receptor). The rTg4510 mice over-express the tauransgene about 13 times over the endogenous mouse tau,nd this higher expression level of tau likely results in thebserved early onset of tau pathology in the mouse brains,

ince the line with lower, 7-fold over-expression of the trans-ene exhibited accumulation of hyperphosphorylated tau by4 months and NFTs inclusions at 20 months (Santa Cruzt al., 2005). Since our initial characterisation of the model

5 gy of A

iTmuiaA

2

rafcr(hk(fltaSrmwaiMwbwewt(i

2

biiwpwmctopeo

nb

2t

wpmnwd(ptin

2

pptma8tmawamdswctBhSwatf

wpcrt

92 M. Yue et al. / Neurobiolo

ndicated that single transgenic (Tgtau and TgtTA) and non-g littermates showed comparable performance in a wateraze test (Santa Cruz et al., 2005), in the present study we

sed only non-Tg littermates as controls in order to max-mise sample sizes of rTg4510 mice. All experiments werepproved by IACUC and were conducted in accordance withAALAC and IACUC guidelines.

.2. Housing conditions

All experimental mice were bred in a pathogen-free envi-onment. Two cohorts of mice were tested at 5.5 months ofge. First, of 23 mice (Nnon-Tg = 11 (6M, M for males; 5F, F foremales henceforth); NtTA/tau = 12 (8M:4F)), was moved to aonventional colony room outside the pathogen-free envi-onment, while the second cohort of 23 mice (Nnon-Tg = 9,4M:5F); NtTA/tau = 14 (6M:8F)) was kept in a pathogen-freeousing conditions. In both housing conditions mice wereept in a standard mouse 29 × 18 × 13 cm shoe-box cagesup to 4 per same sex groups) filled with corn bedding, andood (Mouse Purina chow #5001) and water available adibitum. Mice in the pathogen-free conditions were main-ained in cages placed on ventilated racks (Allentown) andny manipulation of mice was performed in a Biologicalafety Cabinet. Mice housed outside the pathogen-free envi-onment were kept in shoe-box mouse cages placed on openouse racks. The cages were fitted with micro-isolator topsith minimal restrictions on mice handling and manipulation,

nd contained nestlets, promoting nest building, and mousegloos (BioServ #K3327) which provided additional shelter.

ice kept in these conditions were handled twice per week,hile mice kept in the pathogen-free conditions, apart fromiweekly body weight recording and routine cage cleaning,ere not disturbed otherwise. Mice housed in pathogen-free

nvironment were moved to a conventional colony room 1eek before the onset of testing. A cohort of mice designated

o be tested at 3 months (Nnon-Tg = 15 (8M:7F), NtTA/tau = 129M:3F)) was moved to conventional conditions after wean-ng.

.3. Physical development of mice

The body weight of 5.5-month-old mice did not differetween the genotypes (F(1,38) = 2.2, p = 0.15) and hous-ng conditions (F(1,38) = 0.2, NS). Overall, the body weightncreased with age (F(2,342) = 190.3, p < 0.001), and malesere significantly heavier than females (F(1,38) = 49.2,< 0.001). No interactions between experimental factorsere significant. In the 3-month-old cohort, the rTg4510ice had significantly lower body weight than the non-Tg

ontrols (F(1,22) = 7.9, p = 0.01), with males being heavierhan females (F(1,22) = 63.9, p < 0.001). Overall, the weight

f these mice increased during development (F(2,88) = 99.6,< 0.001), with no significant interactions between anyxperimental factors. At the time of testing, the body weightf 3-month-old mice was 25.5 ± 0.82 and 24.6 ± 0.49 for

Daep

ging 32 (2011) 590–603

on-Tg and rTg4510 mice, respectively (p < 0.05) with maleseing heavier than females (p < 0.001).

.4. Housing conditions did not affect the behavior andau pathology in 5.5-month-old Tg4510 mice

Bi-weekly handling and housing in static cages equippedith igloos and nesting material did not differentiate theerformance of mice in the visible platform and referenceemory water maze tests or in the grip strength test, and

o significant interactions between the experimental factorsere found (data not shown). Similarly, housing conditionsid not have any effect on the accumulation of total tau55 + 64 kDa, Low-S, E1) and of 64 kDa tau (both Low-S and3 fractions) in the brain of the rTg4510 mice. Consequently,he 5.5-month-old cohorts of mice were pooled across hous-ng conditions for final analysis, thus yielding a cohort of 20on-Tg and 26 rTg4510 (14M:12F) mice.

.5. Behavioral tests

The battery of the behavioral tests consisted of a visiblelatform or cued version of the water maze (WM), a hiddenlatform or a reference memory version of the test, and a gripest, administered sequentially in the above order. The water

aze apparatus consisted of a circular pool (1.5 m in diameternd 0.60 m high) made of white plastic, which was elevated6 cm off the floor level (the pool’s rim was 146 cm fromhe floor level). The pool was filled with water (24–25 ◦C),ade opaque by the addition of non-toxic white paint, todepth of 45 cm. An escape platform (16 cm in diameter)as submerged about 0.5 cm under the water level. Duringvisible platform training, the location of a platform wasarked with a centrally mounted post (10 cm high, 1 cm in

iameter) which was painted with black and white horizontaltripes. The search path of a mouse during each swim trialas recorded by a video camera suspended 2.5 m above the

entre of the pool and connected to a video tracking sys-em (HVS Image Advanced Tracker VP200, HVS Image,uckingham, UK). Grip strength apparatus consisted of aorizontally mounted 300 g spring scale (Pesola AG, Baar,witzerland) fitted with a grip bar (a toothpick). The scaleas fitted with a drag pointer which moves down and stops

t the maximum load per measurement. The test measuredhe strength with which a mouse held to a grip bar with itsront paws while an experimenter pulled it gently backwards.

Procedures. During the 3-day-visible platform test, theater maze tank was surrounded by a white curtain, and thelatform locations and the release points of a mouse werehanged semi-randomly each 60-s trial (4 trials per day). Theeference memory WM test was run for 9 days with four 90-sraining trials per day with no curtain surrounding the tank.

ark, contrasting geometrical figures (two to three per wall),cabinet, and curtains screening off the computer and an

xperimenter served as spatial cues in the room. The escapelatform was always positioned in the centre of the same

gy of A

qsomtaaamtpatw

2

eSi

wstpisew(eTtoqoss

Fr3sfec(st

M. Yue et al. / Neurobiolo

uadrant (target quadrant, TQ) throughout training. Four 60-probe trials (as the 1st trial of the day (D)) were administeredn D3, D6, D9 and D10 of training to evaluate the develop-ent of spatial memory for the platform location. During

hese trials the escape platform was removed from the pool,nd a mouse was allowed to search the pool for the entirellocated time. The grip strength test was administered 1 dayfter the completion of the reference memory WM test. Aouse was held in a horizontal position and gently lowered

owards the grip bar until it grasped the bar. It was then gentlyulled backward until it released a grip. The force at the timen animal released the grip was recorded. Three consecutiverials were administered within 1 day and the averaged scoreas used for the analysis.

.6. Data analysis: WM—learning acquisition

Learning acquisition during training in the WM test wasvaluated by the analyses of escape latency and search path.ince both variables are usually correlated, but search path

s less dependent on swim speed differences or floating rate,

sWop

ig. 1. Age-progressing impairment of rTg4510 mice in reference memory version oTg4510 mice were significantly longer than search paths of their non-Tg littermate-month-old mice being less impaired than their older counterparts. (B) Both 3- anearches of the pool in the area equidistant to the platform location from the wall (seor 3-month and p < 0.001 for 5.5-month mice). The 3-month-old rTg4510 mice tenspecially at the later stage of training. (C) Both 3- and 5.5-month-old rTg4510 microssing index (ACI = #crosses of platform site minus an average of crosses in theD) The 5.5-month-old rTg4510 mice searched the target quadrant (TQ) significanearched the TQ comparably to non-Tg littermates. **p < 0.01. Note to panels A anrial revealed no significant differences between the groups in both swim path and c

ging 32 (2011) 590–603 593

e report only the results for a search path. We also analysedwim speed, percent of floating rate and percent of thigmo-axic (wall-hugging or a persistent swim along the wall of theool) swims. In order to understand better the search behav-or of mice during spatial navigation, we analysed chainingearch strategy (a persistent swimming at the distance of anscape platform from the wall of the pool, Fig. 1B, insert)hich represents a non-spatial alternative to spatial strategy

Janus, 2004). Spatial memory for the platform location wasvaluated by the analysis of the percent of search path in theQ, and an annulus crossing index (ACI), henceforth referred

o as memory index. ACI represents the number of crossesver the platform site in TQ, adjusted for search in otheruadrants of the pool (an average of platform site crosses inther three quadrants of the pool is subtracted from platformite crosses in TQ). ACI represents an unbiased evaluation ofpatial memory in the WM test, controlling for non-spatial

earch strategies like chaining (Gass et al., 1998; Janus, 2004;ehner et al., 1990; Wolfer and Lipp, 2000). A strong mem-

ry for the spatial location of the platform is expressed byositive ACI values.

f water maze test. (A) The search paths of both 3-month- and 5.5-month-olds (ps < 0.001), and significantly different from one another (p < 0.01), with

d 5.5-month-old rTg4510 mice used non-spatial chaining strategy (circulare insert)) significantly more frequently than the non-Tg littermates (p = 0.01ded to use chaining strategy less often than 5.5-month-old mice (p = 0.06),e showed a significant impairment in spatial memory evaluated by annulusother three quadrants) as compared to control non-Tg mice (ps < 0.001).

tly less than the control mice (p < 0.001), while the 3-month-old rTg4510d B: t1 denotes the first training trial on day 1. The analysis of data for thathaining.

5 gy of A

2s

mN(trwapdi(m(cfobT1o

2

aawtMsGf(afeBdItfh

2

bb1pH4

sB6m6wipcoScsw07T1tffcsscc2q

2

s1AwsiwiTavsSBm6p(

s

94 M. Yue et al. / Neurobiolo

.7. Non-transgenic mice at 3 and 5.5 months of agehowed comparable performance in the water maze

As expected, the 3-month- and 5.5-month-old non-Tgice showed comparable escape latency (F(1,33) = 0.08,S), search path (F(1,33) = 0.06, NS), thigmotaxic swim

F(1,33) = 0.05, NS) and swim speed (F(1,33) = 4.0, NS) inhe visible platform version of WM at α = 0.01 (Bonfer-oni adjustment for multiple comparisons). Similar resultas obtained in the reference memory version of WM,

nd both 3 and 5.5 months non-Tg mice showed com-arable search paths (F(1,33) = 0.05, NS), which rapidlyecreased during training (F(4,264) = 58.8, p < 0.001), chain-ng search strategy (F(1,33) = 0.7, NS), thigmotaxic swimF(1,33) = 1.8, NS), and swim speed (F(1,33) = 0.3, NS). Theemory index (ACI) significantly increased during training

F(3,99) = 5.0, p < 0.01) with no difference between the ageohorts (F(1,33) = 0.4, NS). To avoid the redundancy of ageactor within non-Tg cohort, the results of 3- and 5.5-month-ld non-Tg mice were pooled together for further analysesetween the genotypes, yielding a control group of 35 non-g mice (15—3 months and 20—5.5 months, respectively,8M:17F) against which 3-month- (N = 12) and 5.5-month-ld (N = 26) groups of rTg4510 mice were compared.

.8. Statistical analysis

The analysis was performed by a factorial analysis of vari-nce (ANOVA) with genotype (non-Tg and rTg4510), age (3nd 5.5 months), and sex as between, and training days asithin subject (repeated measure) factors. Departures from

he assumption of compound sphericity were evaluated byauchly sphericity test with α level set to 0.05. In cases when

phericity was violated, degrees of freedom were adjusted byreenhouse-Geisser �-correction. When appropriate, global

actorial analysis was followed by analyses of simple effectsStevens, 1990) pertaining to each research hypothesis. Innalyses requiring multiple comparisons among means Bon-erroni adjustment of α level to minimise Type I (family-wise)rror rate was used. Post hoc analyses were performed usingonferroni t-test (MODLSD). All statistical analyses wereone using Statistical Package for Social Sciences (SPSSnc., Chicago) version 13 for Macintosh. Due to space limita-ion and considerable number of between- and within-subjectactors only significant results pertaining directly to testedypotheses are reported.

.9. Preparation of brain extracts and tau biochemistry

Mice were sacrificed by cervical dislocation and theirrains were harvested and homogenized in 10× volume ofuffer containing 50 mM Tris, 274 mM NaCl, 5 mM KCl,

mM PMSF, protease inhibitor cocktail 1:100 (Sigma), phos-hatase inhibitors I & II cocktail 1:100 (Sigma), pH 8.0.omogenates were centrifuged at 13,000 × g for 15 min at◦C, and supernatant was saved as the low-speed (Low-S)

ot(n

ging 32 (2011) 590–603

upernatant. Protein concentration was determined using theCA protein Assay Kit (Pierce). We evaluated the levels of4 kDa hyperphosphorylated tau as the total amount of nor-alized 64 kDa tau (expressed as ratio over the total (55 and

4 kDa) of human tau) present in the Low-S supernatant, asell as the 64 kDa hyperphosphorylated tau in the sarkosyl-

nsoluble fraction (p3) following a previously publishedrotocol (Greenberg and Davies, 1990): Briefly, the purifi-ation of sarkosyl-insoluble (p3) fraction comprised a seriesf fractionations. First, an aliquot of about 200 �l of the Low-supernatant, adjusted for equal protein concentration, was

entrifuged at 150,000 × g for 15 min at 4 ◦C. The high-speedupernatant was saved as the soluble fraction. The pelletas re-homogenized in a buffer containing 10 mM Tris–HCl,.8 M NaCl, 10% sucrose, 1 mM EGTA, and 1 mM PMSF, pH.4, then centrifuged again at 150,000 × g for 15 min at 4 ◦C.he supernatant from this centrifugation was then adjusted to% sarkosyl and incubated at 37 ◦C for 1 h, followed by cen-rifugation at 150,000 × g for 30 min at 4 ◦C. The supernatantrom this last centrifugation was saved as sarkosyl solubleraction, and the pellet fraction was suspended in 40 �l bufferontaining 10 mM Tris, 1 mM EDTA (pH 8.0) to represent thearkosyl-insoluble (p3) fraction of tau protein. In a previoustudy using similar methods, we found that the p3 fractionontained invariably less than 5% of the 64 kDa tau that isontained in the original Low-S supernatants (Berger et al.,007), and represents therefore a minor fraction of the overalluantity of hyperphosphorylated tau in brain extracts.

.10. Western blotting

Aliquots containing 4 �g total protein of the Low-Supernatants and 5 �l of the p3 fractions were loaded onto0-well 10% Tris–glycine gels (Invitrogen), and run at 125 V.fter transfer at 200 mA for 90 min, PVDF membranesere blocked with 5% non-fat milk in 10 mM Tris-buffered

aline/0.1% Triton X-100 (TBS-T) for 1 h. Primary antibod-es were added in blocking buffer for overnight at 4 ◦C. Blotsere washed three times for 10 min each with TBS-T and then

ncubated with secondary antibody in 5% non-fat milk withBS-T for 1 h at room temperature. Blots were then washedgain three times for 10 min each, and labeled bands wereisualized using the ECL PLUS kit (PerkinElmer) and expo-ure to film (Kodak Biomax). Bands were quantified usingcionImage (Frederick, MD) by analysing pixel density.and intensities were corrected for background and nor-alized relative to a reference sample of sarkosyl-insoluble

4 kDa tau obtained from a JNPL3 transgenic mouse brainreparation expressing the same P301L tau mutant proteinLewis et al., 2000).

Total transgenic tau was detected with the human-pecific tau polyclonal E1 antibody (1:2000) which rec-

gnizes a phosphorylation-independent N-terminal humanau epitope. Glyceraldehyde-3-phosphate dehydrogenaseGAPDH) monoclonal antibody (1:5000, Biodesign Inter-ational, Kennebunkport, ME) was used to control for

gy of A

eas

3

3i

d(5i(n(sniw(3sfpsoobpttca2mporfc5iag(dgtottoi

bemttsdpaints

3

taFpam(aPrbittl(ct5

3i

rnmtiiicost

M. Yue et al. / Neurobiolo

qual protein loading. Appropriate HRP-labeled secondaryntibodies (1:5000) were utilized from Jackson ImmunoRe-earch, West Grove, PA.

. Results

.1. Age-dependent decline in spatial reference memoryn rTg4510 mice

The groups differed significantly in their search pathsuring training in the reference memory version of WMF(2,67) = 66.6, p < 0.001, Fig. 1A). The paths of both.5-month- and 3-month-old rTg4510 mice were signif-cantly longer than the path of their non-Tg littermatesps < 0.001), with 3-month-old rTg4510 mice showing sig-ificantly shorter paths than the 5.5-month counterpartsp < 0.01). Overall, all mice significantly decreased theirearch paths during training (F(6,408) = 35.1, p < 0.001) andeither genotype nor sex significantly affected their learn-ng. An inspection of Fig. 1A indicated that the groupsere already differentiated during the 1st day of training

F(2,67) = 8.1, p < 0.001). Post hoc analysis revealed that both-month- and 5.5-month-old rTg4510 mice showed longerearch paths than non-Tg mice (ps < 0.01), with no sex dif-erences or interaction present. However, the analysis of theerformance of the 1st trial of day 1 of training revealed noignificant group (F(2,67) = 0.8, NS), sex (F(1,67) = 0.5, NS),r interaction effects, which suggests that both age groupsf rTg4510 and their control littermates showed compara-le initial response to the training conditions in the hiddenlatform WM test (Fig. 1A, t1). In order to elucidate fur-her the nature of impairment in the spatial navigation ofhe rTg4510 mice we analysed their thigmotaxic swim andhaining search strategy which are known to significantlyffect the performance of mice in the water maze test (Janus,004; Lipp and Wolfer, 1998). Since both chaining and thig-otaxis were found to be positively correlated with search

ath (rs = 0.8, p < 0.001), the following analysis providesnly qualitative and supporting explanation of mice behaviorather than an independent analysis of their impairment. Weound a significant difference between the groups in usinghaining strategy (F(2,67) = 27.0, p < 0.001). Both 3- and.5-month-old rTg4510 mice used chaining strategy signif-cantly more often than the non-Tg control mice (p = 0.01nd p < 0.001 for 3- and 5.5-month mice, respectively). Allroups decreased the use of chaining strategy during trainingF(8,16) = 3.5, p = 0.001), however, the rate of this changeiffered between the groups (F(16,536) = 3.7, p < 0.001,roup × day interaction). The 3-month-old rTg4510 miceended to use chaining strategy less often than 5.5-month-ld rTg4510 mice (p = 0.06), especially in the later part of

raining (Fig. 1B). The 5.5-month rTg4510 mice searchedhe pool using chaining strategy at the same rate through-ut training (Fig. 1B). Correspondingly, the groups differedn their thigmotaxic swim (F(2,67) = 81.2, p < 0.001), with

aptf

ging 32 (2011) 590–603 595

oth age cohorts of rTg4510 mice showing comparable lev-ls of thigmotaxis (40% for 3-month, and 50% for 5.5-monthice, p = 0.1), which were significantly higher as compared

o about 30% of thigmotaxic swim observed in non-Tg lit-ermates (ps < 0.001). All groups showed a comparable swimpeed (F(2,67) = 1.9, NS), however, the swim speed changedifferently within groups during training (F(10,367) = 2.1,< 0.02). While the swim speed of non-Tg mice decreaseds training progressed, it remained constant throughout train-ng in both age cohorts of rTg4510 mice. Floating rate didot change during training and was comparable betweenhe groups and sexes, irrespective of the genotype (data nothown).

.2. Impairment in spatial reference memory

All mice improved their spatial memory for the loca-ion of the escape platform during the four probe trialsdministered during training (F(3,201) = 6.4, p < 0.001 and(3,201) = 19.8, p < 0.001 for ACI (averaged across fourrobe trials Annulus Crossing Index, see Section 2)nd %path searched in TQ respectively). However, theemory scores differed significantly between the groups

F(2,67) = 42.2, p < 0.001 for ACI memory index, Fig. 1C,nd F(2,67) = 16.7, p < 0.001 for % path in TQ, Fig. 1D).ost hoc analysis revealed that, both 3- and 5.5-month-oldTg4510 mice showed comparable spatial memory evaluatedy ACI (p = 0.14, Fig. 1C) which in both cases was signif-cantly lower as compared to non-Tg mice (ps < 0.001). Onhe other hand, the analysis of the dwell in the TQ revealedhat only the 5.5-month-old rTg4510 mice spent significantlyess time than the control non-Tg mice searching the TQp < 0.001). The 3-month-old rTg4510 mice searched the TQomparably to non-Tg littermates (p = 0.15), but at the sameime they showed a tendency to search the TQ more than the.5-month-old counterparts (p = 0.09, Fig. 1D).

.3. The impairment in spatial memory of rTg4510 mices more severe in females than in males

The overall analysis including both age cohorts ofTg4510 and control mice revealed that females showed sig-ificantly longer search paths for a hidden platform thanales (F(1,67) = 14.4, p < 0.001) irrespectively of their geno-

ype. However, since the small sample size of females (N = 3)n 3-month-old rTg4510 cohort could obscure the possiblenteraction effects between sex and genotype and bias thenterpretation of the results, we removed the 3-month-oldohort of mice from detailed analyses of the effect of sexn spatial memory. The 5.5-month-old rTg4510 mice wereignificantly impaired in their search path, as compared toheir non-Tg littermates (F(1,42) = 72.4, p < 0.001) and over-

ll females were more impaired than males (F(1,42) = 8.2,< 0.01), with no significant interaction between the geno-

ype and sex present. The inspection of the data indicates thatemales and males of both genotypes performed comparably

596 M. Yue et al. / Neurobiology of Aging 32 (2011) 590–603

Fig. 2. Female rTg4510 mice at the age of 5.5 months were impaired more severely than males in the reference memory water maze. (A) Female mice showedlonger search paths than their male counterparts within each genotype, however the impairment of rTg4510 females was significantly worse than impairmentof rTg4510 males (p < 0.05) and performance of non-Tg littermates (p < 0.001), especially at the end of training period. (B) Similarly, the use of non-spatialchaining strategy was observed more frequently in rTg4510 mice, especially in females which significantly increased the use of this strategy as trainingprogressed. (C) Spatial memory evaluated by annulus crossing index (ACI) was significantly impaired in rTg4510 females as compared to male counterparts( y. rTg4f tly impr

drtttpcrTttPmwffcsittif

ttrowu5seoBocfstaWp

p < 0.05), while female and male non-Tg mice showed comparable memoremale and male controls (p < 0.05). (D) rTg4510 females showed significanTg4510 males and non-Tg littermates (p < 0.01). *p < 0.05; **p < 0.01.

uring the first day of training. The inferior performance ofTg4510 females was clear from day 2 of training and tendedo increase towards the end of the test (Fig. 2A). In con-rast, the non-Tg females showed less pronounced increase inheir search path and by the end of the training they showederformance similar to male littermates. Consequently, weompared the performance of mice at the end of training,unning analysis on averaged across day 8 and 9 scores.he analysis revealed significant differences between geno-

ype and sex (ps < 0.05), but the difference between sexesended to depend on their genotype (F(1,42) = 3.8, p = 0.06).ost hoc analysis revealed that both non-Tg females andales had comparable performance, and their search pathsere significantly shorter from rTg4510 males (p < 0.05), and

rom rTg4510 females (p < 0.001). In contrast, the rTg4510emales continued to differ significantly from their maleounterparts (p < 0.05). Corresponding analysis of chainingtrategy revealed also significantly increased use of chain-ng by rTg4510 mice as compared to non-Tg mice, with a

rend of more frequent use of chaining by rTg4510 femaleshan their male counterparts (F(1,42) = 3.5, p = 0.07). Thenspection of chaining scores clearly shows that rTg4510emales followed chaining strategy more often, especially

t

rm

510 males showed also significant decline in their memory as compared toaired search of the target quadrant (TQ) during probe trials as compared to

owards the end of training (Fig. 2B). Correspondingly, athe end of training (averaged scores of days 8 and 9) theTg4510 females used chaining strategy significantly moreften than their male counterparts and non-Tg mice (p < 0.05)hile both non-Tg females and males and rTg4510 malessed chaining at the comparable level. Lastly, overall the.5-month-old rTg4510 mice spent higher proportion of timewimming along the wall of the pool (ps < 0.001, genotypeffect), but this behavior depended significantly on the sexf mice (F(1,42) = 9.4, p < 0.01, genotype × sex interaction).oth non-Tg females and males spent similar proportionf time (on average 5–7%) on thigmotaxic swim, signifi-antly less than rTg4510 males (18%, p < 0.001) and rTg4510emales, with the latter spending on average 30% of timewimming along the wall (significantly more from male coun-erparts and non-Tg littermates, ps < 0.001). In summary, thenalysis of sex differences during learning acquisition in the

M revealed that rTg4510 females showed longer searchaths, likely due to their increased thigmotaxic swims and

he use of a non-spatial chaining search strategy.

The analysis of memory scores of 5.5-month-old miceevealed that females showed lower memory scores thanales (F(1,42) = 5.2, p < 0.05, for ACI; F(1,42) = 10.1,

M. Yue et al. / Neurobiology of A

Fig. 3. Distribution of memory scores, averaged across four probe trials, in5.5-month-old mice. (A) The rTg4510 mice showed considerable variationin their spatial memory scores as evaluated by the annulus crossing index(ACI), with female scores clustering predominantly at the lower end of thedistribution, often searching the pool in the incorrect area or quadrant (neg-asia

pttibmesttiluoon

mhanNp4ttm(

3a

waimufatm(rrTatlsewmns

tii(iatinmc2w

tive ACI values). (B) All non-Tg littermates showed a positive bias for thepatial location of the platform, albeit also showing considerable variabilityn their memory scores. The distribution of spatial memory of non-Tg femalend male mice was homogenous.

< 0.01, for % of path in TQ) irrespectively of their geno-ype. The memory in males and females showed a tendency ofhe dependence on the genotype in the case of the % of pathn TQ (F(1,42) = 3.0, p = 0.09, sex × genotype interaction)ut not in the case of ACI. The inspection of individual ACIemory scores revealed that rTg4510 mice showed a consid-

rable variability of scores ranging from −1.42 (indicatingearches for the platform in a incorrect quadrant of the pool)o high positive values of 4.25 (Fig. 3A). It became apparenthat, while some 5.5-month-old rTg4510 mice were severelympaired, other mice showed a positive bias for the platformocation, with scores comparable to lower and midrange val-

es of non-Tg littermates (Fig. 3B). Furthermore, the scoresf rTg4510 females clustered predominantly at the lower endf the distribution (Fig. 3A), which was reflected in their sig-ificantly lower average ACI memory scores as compared to

np((

ging 32 (2011) 590–603 597

ale counterparts (t(20) = 2.4, p < 0.05, Fig. 2C). On the otherand, the distribution of ACI memory scores between femalesnd males in non-Tg cohort was homogenous (Fig. 3B), witho differences between their averaged scores (t(18) = 1.05,S, Fig. 3B). Similarly, the % path searching in TQ was com-arable between non-Tg females and males (46.4 ± 2.5 and1.6 ± 2.0, respectively, t(18) = 1.5, NS, Fig. 2D), howeverhe rTg4510 females searched the TQ significantly less thanheir male counterparts (19.3 ± 2.1 for females, 35.6 ± 4.5 for

ales, t(18) = 2.3, p < 0.01), and non-Tg littermates (p < 0.01)Fig. 2D).

.4. Association between the spatial memory declinend hyperphosphorylated 64 kDa tau

To assess the functional context of memory index decline,e investigated the association between the memory indices

nd the levels of hyperphosphorylated 64 kDa tau depositedn the mouse brain. The very low levels of 64 kDa tau in 3-

onth-old rTg4510 mice (ranged from 0.01 to 0.12 (arbitrarynits), and observed only in 5 out of 12 mice), precluded theull analysis of the relationship in this age cohort of mice. Thenalysis of the total levels of hyperphosphorylated 64 kDaau in 5.5-month-old rTg4510 mice revealed that female

ice showed significantly higher levels of 64 kDa tau proteinexpressed as ratio of 64 kDa tau to total (normally phospho-ylated 55 kDa and hyperphosphorylated 64 kDa) tau) thanTg4510 males (t(24) = −3.9, p < 0.001, Fig. 4B left panel).he analysis of sarkosyl-insoluble 64 kDa tau (p3 fraction)lso revealed higher levels of 64 kDa tau in rTg4510 femaleshan in males, albeit the difference did not reach significanceevel at alpha 0.05, likely due to higher variability of thesecores (Fig. 5B). The total levels of tau protein (55 + 64 kDa)xpressed by the transgene in not fractionated Low-S extractere comparable between the females and males of rTg4510ice (Fig. 4B right panel), thus excluding the obvious expla-

ation that sex-dependent tau expression caused differentialeverity of tau pathology.

The analysis of the relationship between levels of 64 kDaau with the averaged across all 4 probe trials spatial memoryndices in 5.5-month-old rTg4510 mice revealed significantnverse correlations between tau levels and both ACIr2 = 0.16, p < 0.05, Fig. 4C) and the percent of path spentn TQ (r2 = 0.23, p < 0.01, Fig. 4D). We used one-tail criticallpha level to evaluate the above associations followinghe hypothesis based on previously published results thatncreased levels of hyperphosphorylated 64 kDa tau affectegatively the development of spatial memory in rTg4510ice (Berger et al., 2007). Our analysis revealed that, despite

onsiderable variability in the memory scores, 16% and3% of variance in ACI and % path in TQ, respectively,ere explained by the variance in the tau levels. Similarly,

egative associations were obtained when sarkosyl-insoluble3 tau was correlated with ACI or % of path in TQ indicesresults not shown). However, the inspection of scatter plotsFig. 4C and D) reveals that the significance of the above

598 M. Yue et al. / Neurobiology of Aging 32 (2011) 590–603

Fig. 4. Total levels of hyperphosphorylated 64 kDa tau are higher in females than in males of rTg4510 mice. (A) Tau species detected in the low speed (Low-S)fraction of brain homogenate using human-specific tau polyclonal E1 antibody. The representative Western blot shows that hyperphosphorylated 64 kDa tau ispresent in all 5.5-month-old rTg4510 mice, however the signal was stronger in female mice. The lower band represents 55 kDa tau. (B) rTg4510 females hadsignificantly higher levels of total 64 kDa tau (normalized to total tau (55 + 64 kDa) present in the homogenate) than male counterparts (left panel). The total taulevels detected in the Low-S fraction of the brain homogenate were comparable between females and males (right panel). (C) A significant negative associationswere found between the total 64 kDa tau (expressed as ratio of 64 kDa tau to total tau detected with the human-specific tau polyclonal E1 antibody) and spatialmemory evaluated by ACI (annulus crossing index), and (D) by the dwell (% of path) in the target quadrant (TQ) in the 5.5-month-old rTg4510 mice. In bothc tau lev( u levelsd ing.

c6maoans

6dwtp

6sbva

3t

ases the significance of the associations was driven by sex differences infor ACI and % path in TQ, respectively) accounted for the differences in taehydrogenase monoclonal antibody) used to control for equal protein load

orrelations was likely driven by the sex differences in4 kDa tau levels and the corresponding differences in theemory scores. Despite a considerable overlap in tau levels

nd memory scores between females and males, the clustersf extreme values of both measures were sex driven. Thenalyses of the relationships within each sex showed onlyon-significant negative trends of the associations betweenpatial memory and levels of 64 kDa tau.

Significant associations were also found between total4 kDa tau and the impaired performance of rTg4510 mice

uring WM training. The high levels of total 64 kDa tauere associated with longer search paths (r2 = 0.16, p < 0.05,

wo-tailed test) and with increased thigmotaxis (r2 = 0.36,< 0.01), but no chaining strategy. Sarkosyl-insoluble p3

pitm

els and in memory scores, with the 16% and 23% of variance in memory. ***p < 0.001. F, female; M, male. GAPDH (glyceraldehyde-3-phosphate

4 kDa tau was significantly associated only with longerearch paths (data not shown). No significant associationsetween any of the behavioral measures recorded during theisible platform version of the water maze test (see below)nd the levels of 64 kDa tau were found (data not shown).

.5. Age-dependent impairment of the rTg4510 mice inhe visible platform version of WM test

Females did not differ significantly from males in their

erformance in the visible platform of the WM in any stud-ed cohort of mice, and no significant interactions betweenhe genotype, age and sex were present. The mice of 3-

onth-, 5.5-month-old and non-Tg cohorts swam readily

M. Yue et al. / Neurobiology of Aging 32 (2011) 590–603 599

Fig. 5. Levels of sarkosyl-insoluble (p3) fraction of hyperphosphorylated64 kDa tau in females and males rTg4510 mice. (A) Sarkosyl-insoluble64 kDa tau, normalized with reference to 64 kDa tau (p3) obtained fromJNPL3 transgenic mice, showed sex-dependent variability, being increasedin female mice. (B) rTg4510 female mice had, on average, higher levels ofsarkosyl-insoluble 64 kDa (p3) tau than males, however the difference did notrrfP

tsmmp3taim(maiawc(scoet

3

n

Fig. 6. rTg4510 mice showed age-progressing impairment in the cuedversion (visible platform) water maze test, without significant differencesbetween sexes. (A) The search paths to a visible platform of 5.5-month-oldrTg4510 mice were significantly longer than their 3-month-old counterparts(p < 0.05) and non-Tg littermates (p < 0.001). The 3-month old rTg4510 micedow

5wgs(tAsto

each significance at α = 0.05. nTg, non-transgenic; Tg, transgenic. JNPL3eference sample of sarkosyl-insoluble 64 kDa tau (p3 fraction) obtainedrom a JNPL3 transgenic mouse brain preparation expressing the same301L tau mutant protein (Lewis et al., 2000) as rTg4510 mice.

o the cued platform (F(2,140) = 79.1, p < 0.001), but theearch paths significantly differed between the groups ofice (F(2,67) = 14.8, p < 0.001). The 5.5-month-old rTg4510ice showed longer paths than their 3-month-old counter-

arts (p < 0.05) and non-Tg littermates (p < 0.001), while the-month-old rTg4510 mice did not differ from non-Tg lit-ermates (Fig. 6A). The analysis of escape latency yielded

similar result (data not shown). The mice also differedn their swim speed (F(2,67) = 14.6, p < 0.01), with non-Tg

ice swimming faster than both age cohorts of rTg4510 miceps < 0.01); in floating rate (F(2,67) = 7.3, p < 0.01), with 3onths rTg4510 mice floating significantly more (p < 0.01)

nd 5.5-month-old mice showing a trend in increased float-ng (p = 0.05) as compared to non-Tg mice. The groupslso differed in thigmotaxic swim (F(2,70) = 28.5, p < 0.01),ith both age cohorts of rTg4510 mice showing signifi-

antly increased thigmotaxic swim than non-Tg littermatesps < 0.001, Fig. 6B). In summary, the rTg4510 mice differedignificantly from non-Tg littermates in several parametersharacterising the performance in the visible platform versionf water maze test. It seems likely that the major factor differ-ntiating the rTg4510 from non-Tg mice was their increasedendency to swim along the wall of the pool.

.6. Grip strength

Since the preliminary analysis revealed that 3-month-oldon-Tg mice showed stronger grip of their front paws than

4

s

id not differ from non-Tg littermates. (B) Both 3-month- and 5.5-month-ld rTg4510 mice showed similar and increased thigmotaxic swim along theall of the pool as compared to the non-Tg mice (ps < 0.001).

.5-month-old counterparts (t(24) = 3.3, p < 0.01), the resultsere analysed with genotype and sex factors within each ageroup. The analysis revealed that overall 3-month-old micehowed stronger grip than the 5.5-month-old counterpartsF(1,54) = 17.1, p < 0.001). No significant effects of geno-ype, sex nor interactions between these factors were found.lso, no significant relationship was found between the grip

trength and body weight, and swim speed, thus excludinghe possibility that subtle differences in physical conditionsf mice could confound the cognitive measures in the study.

. Discussion

We show that female rTg4510 mice experience moreevere deterioration in spatial reference memory than their

6 gy of A

mmarfmMJeIomnptipriricirpaeoeeisbthsonl5damacabittttamts

rtWmptppctpbiWpirpfpsfmitso(sdnAmiorsmmtritaastibst

00 M. Yue et al. / Neurobiolo

ale littermates, and this difference in the memory impair-ent was correlated with significantly increased levels of

bnormally hyperphosphorylated 64 kDa tau in the femaleTg4510 brains. Although there is evidence in rodent speciesor females to generally perform less well in the wateraze than males (rats (Frye, 1995; Frye and Sturgis, 1995;arkowska, 1999; Perrot-Sinal et al., 1996; Warren and

uraska, 1997), laboratory mice (Arters et al., 1998; Benicet al., 2006; Berger-Sweeney et al., 1995; Frick et al., 2000;mwalle et al., 2006), and wild mice (Kavaliers et al., 1996)),ur study demonstrates a more aggressive memory impair-ent in tau transgenic female mice. In contrast, female

on-transgenic mice showed relatively modest decrease oferformance only during learning acquisition phase whileheir spatial memory was comparable to non-Tg males. Sim-lar results were obtained in rats where an initial inferiorerformance of females during training in WM was noteflected in their spatial memory scores at the end of train-ng (Beiko et al., 2004). In our study, the 5.5-month-oldTg4510 females were significantly impaired in both learn-ng acquisition and spatial memory as compared to their maleounterparts and to their non-Tg littermates (Fig. 2). Thiss likely to be related to increased tau pathology as femaleTg4510 mice had significantly higher levels of hyperphos-horylated 64 kDa tau than males at the time of behavioralssessment, a result which could not be accounted by a differ-nt levels of transgene expression between the sexes. Nonef the previously published studies on rTg4510 mice (Bergert al., 2007; Janus, 2008; Ramsden et al., 2005; Santa Cruzt al., 2005; Spires et al., 2005) reported the sex differencen spatial memory and/or pathology, either due to smallerample sizes of the test cohorts and the use of differentehavioral measures, or due to the fact that assessment ofau pathology was not based on the total pool of extractableyperphosphorylated 64 kDa tau, but on less direct mea-ures of tau pathology. Consistent with a strong influencef sex on tau-driven neurodegeneration, the overall stronglyegative association between spatial reference memory andevels of hyperphosphorylated 64 kDa tau in the brain of.5-month-old rTg4510 mice was mainly driven by sex-ependent distribution of extreme memory scores and theirssociated 64 kD tau levels (Fig. 3). It needs to be noted thatultiple behavioral measures in the water maze were associ-

ted with tau pathology and that many of these measures areorrelated and possibly causally interrelated to some degree,lthough they measure different aspects of spatial navigationehavior. The development of cross-correlation between var-ous measures of performance reflects the extend to whichhe behavioral deterioration progresses, and may be useful inhe identification of specific behaviors that may respond toreatment first. Therefore, the most conservative interpreta-ion of our results is that the increased levels of tau pathology

nd associated neurodegeneration in the brain of rTg4510ice had a broader impact on mice behavior, beyond selec-

ive impairment in spatial memory as previous interpretationsuggested (Santa Cruz et al., 2005).

tdso

ging 32 (2011) 590–603

The age range studied here from 3 to 5.5 months, cor-esponded to early and mature stages of NFT formation inhe brain of this mouse model (Santa Cruz et al., 2005).

e demonstrated that at 5.5 months the behavioral impair-ent encompassed both spatial and cue learning (visible

latform) in the WM test. However, the lack of an associa-ion between the impairment in the cue learning task and tauathology suggests that the mice have a significantly com-romised plasticity in the choice of search strategies duringognitively demanding spatial reference memory task, butheir performance in the cognitively less demanding cued-latform task was less affected by neurodegeneration thany other unidentified factors. Hence, it was the significantlyncreased thigmotaxic swim, an initial search strategy in a

M (Janus, 2004; Lipp and Wolfer, 1998) that was the mostrominent characteristic of behavioral impairment observedn rTg4510 mice. Persistent thigmotaxis makes the explo-ation of the pool less efficient and results in longer searchaths (Schenk, 1985). Most importantly, it prevents the micerom finding an escape platform through active search of theool, thus considerably diminishing the reinforcing effect ofearch behavior on learning the location of the escape plat-orm. The increased use of chaining strategy in the referenceemory test, especially by rTg4510 females, indicates the

nability of the rTg4510 mice to develop a spatial map ofhe environment and/or the inability of the efficient use ofpatial strategy to find the platform, an impairment previ-usly documented in some lines of APP transgenic miceBrody and Holtzman, 2006; Janus, 2004). Also, our resultshowed that the interpretation of the memory impairment mayepend on specific memory index employed. In the case ofon-Tg mice, both the percent of search in the TQ and theCI yielded comparable conclusions regarding their spatialemory development and strength. Similarly, both memory

ndices indicated severe memory impairment in 5.5-month-ld rTg4510 mice. However, in the case of 3-month-oldTg4510 mice each memory index led to opposite conclu-ions. While the % of search in TQ indicated that rTg4510ice had a spatial memory comparable to non-transgenicice, the ACI revealed a significant memory impairment in

his age cohort of rTg4510 mice. Such discrepancy likelyeflects different demand on memory recall imposed by eachndex. While the ACI evaluates the focal search for the loca-ion of the platform, and reflects the swims over a smallrea of the pool indicating the precise platform location, alsodjusting for persistent chaining, the dwell in TQ depictsearch of a wider area of the quadrant of the pool containinghe platform, a less demanding task. In most cases, includ-ng the present study, both scores are positively correlated,ut as our results indicate, the ACI index tends to be moreensitive to subtle memory impairments due to either earlyau related pathology, or transgenic background effect. On

he other hand, the dwell in TQ was less affected by mildereficits in spatial memory, and seem to present a better mea-ure of rapid memory decline due to the increasing levelsf hyperphosphorylated toxic tau species. Therefore, eval-

gy of A

ua(pp

spbii(rcesgtcooawcttTiipmeop2etmchmMtStaoytrnrrpda

D

ai

A

RAJhm

R

A

A

A

A

A

B

B

B

B

B

B

B

M. Yue et al. / Neurobiolo

ation of spatial memory based on both measures couldlert to confounding subtle negative transgene dosage effectsTerwel et al., 2005), or be indicative of yet to be identifiedathological changes in neuronal activity preceding overt tauathology.

The unambiguous impact of sex on the phenotype ofingle-tau transgenic mice concomitant with increased tauathology, unencumbered by confounding factors, has noteen documented before. In a double mutant APP/tau modelsnferior performance of females was ascribed to significantlyncreased amyloid burden in female mice. In the TAPP modelLewis et al., 2001), female mice showed enhanced neu-ofibrillary pathology in the limbic system and olfactoryortex as compared to TAPP males, and to transgenic micexpressing mutant tau alone. Tau pathology also developedignificantly faster in females than in males in single trans-enic JNPL3 mice (Lewis et al., 2000). However, in bothhe cases of the JNPL3 and the TAPP model (derived byrossing JNPL3 and APP Tg2576 lines) the interpretationf the effects was obscured by a twofold increase in tauver-expression in JNPL3 females than in males (Lewis etl., 2001). With this interpretation excluded in our study,e demonstrate for the first time that the negative asso-

iation between the levels of hyperphosphorylated 64 kDaau and spatial memory was explained solely by increasedau pathology in females without confounding artefacts.hese results further suggest that the onset of tau pathology

n rTg4510 mouse model of tauopathy may not be lim-ted to a specific postnatal age or over-expression of taurotein, but may depend on yet to be identified develop-ental, gender-specific, and/or environmental factors. The

pidemiologic evidence underscoring that the deregulationf the hypothalamic-pituitary-gonadal (HPG) axis increasesrevalence of Alzheimer disease in women (Atwood et al.,005; Bowen et al., 2002; Casadesus et al., 2006; Webbert al., 2004) focuses on the hormonal imbalance relatedo menopause. The suggested link between APOE-�4, hor-

onal imbalance, and the onset of AD (Raber, 2004) wasonfirmed experimentally when female mice expressinguman APOE-�4 showed significant impairment in the wateraze, at the age of 6 months (Raber et al., 1998, 2000; vaneer et al., 2007). However, treatment of these mice with

estosterone ameliorated these deficits (Raber et al., 2002).ince the rTg4510 mice at 5.5 months were developmentally

oo young to implicate age-dependent hormonal changes asdriving force of increased tau pathology, the exact cause

f increased tau pathology in females of rTg4510 model haset to be elucidated. For this reason it is conceivable thathe rTg4510 model may present an assay to discover natu-al modulators of tau-driven neurodegeneration which couldot be identified in epidemiological studies of AD, where theesponsible factor may no longer be operative due to age-

elated changes, but could be of therapeutic value. This holdsarticularly for hypothetical protective factors in males losturing aging, subject to a hormonal supplementation ther-py.

B

ging 32 (2011) 590–603 601

isclosure statement

The authors, M.Y., A.H., J.W., H.R., and C.J. do not haveny conflict of interest to report regarding the data presentedn this manuscript.

cknowledgements

This work was supported by Alzheimer’s Diseaseesearch Center Pilot Project and American Federation forging Research Grants (CJ). We would like to thank Drs.

ada Lewis, Mike Hutton, and two anonymous reviewers forelpful and insightful comments on an earlier draft of theanuscript.

eferences

rendt, T., Schindler, C., Bruckner, M.K., Eschrich, K., Bigl, V., Zedlick, D.,Marcova, L., 1997. Plastic neuronal remodeling is impaired in patientswith Alzheimer’s disease carrying apolipoprotein epsilon 4 allele. J.Neurosci. 17 (2), 516–529.

rriagada, P.V., Growdon, J.H., Hedley-Whyte, E.T., Hyman, B.T., 1992.Neurofibrillary tangles but not senile plaques parallel duration and sever-ity of Alzheimer’s disease. Neurology 42 (3), 631–639.

rters, J., Hohmann, C.F., Mills, J., Olaghere, O., Berger-Sweeney, J., 1998.Sexually dimorphic responses to neonatal basal forebrain lesions in mice.I. Behavior and neurochemistry. J. Neurobiol. 37 (4), 582–594.

twood, C.S., Meethal, S.V., Liu, T., Wilson, A.C., Gallego, M.,Smith, M.A., Bowen, R.L., 2005. Dysregulation of the hypothalamic-pituitary-gonadal axis with menopause and andropause promotesneurodegenerative senescence. J. Neuropathol. Exp. Neurol. 64 (2),93–103.

zad, N.A., Al Bugami, M., Loy-English, I., 2007. Gender differences indementia risk factors. Gend. Med. 4 (2), 120–129.

anbury Conference on Genetic Background in Mice, 1997. Mutant miceand neuroscience: recommendations. Neuron 19, 755–759.

eiko, J., Lander, R., Hampson, E., Boon, F., Cain, D.P., 2004. Contribu-tion of sex differences in the acute stress response to sex differencesin water maze performance in the rat. Behav. Brain Res. 151 (1–2),239–253.

enice, T.S., Rizk, A., Kohama, S., Pfankuch, T., Raber, J., 2006. Sex-differences in age-related cognitive decline in C57BL/6J mice associatedwith increased brain microtubule-associated protein 2 and synaptophysinimmunoreactivity. Neuroscience 137 (2), 413–423.

erg, L., McKeel, D.W., Miller, J.P., Baty, J., Morris, J.C., 1993. Neuropatho-logical indexes of Alzheimer’s disease in demented and nondementedpersons aged 80 years and older. Arch. Neurol. 50 (4), 349–358.

erger, Z., Roder, H., Hanna, A., Carlson, A., Rangachari, V., Yue, M.,Wszolek, Z., Ashe, K., Knight, J., Dickson, D., Andorfer, C., Rosenberry,T.L., Lewis, J., Hutton, M., Janus, C., 2007. Accumulation of patholog-ical tau species and memory loss in a conditional model of tauopathy. J.Neurosci. 27 (14), 3650–3662.

erger-Sweeney, J., Arnold, A., Gabeau, D., Mills, J., 1995. Sex differ-ences in learning and memory in mice: effects of sequence of testing andcholinergic blockade. Behav. Neurosci. 109 (5), 859–873.

owen, R.L., Smith, M.A., Harris, P.L., Kubat, Z., Martins, R.N., Castel-

lani, R.J., Perry, G., Atwood, C.S., 2002. Elevated luteinizing hormoneexpression colocalizes with neurons vulnerable to Alzheimer’s diseasepathology. J. Neurosci. Res. 70 (3), 514–518.

raak, H., Braak, E., 1997. Frequency of stages of Alzheimer-related lesionsin different age categories. Neurobiol. Aging 18 (4), 351–357.

6 gy of A

B

B

C

C

C

C

C

C

D

D

E

E

E

F

F

F

F

F

G

G

G

G

H

H

I

I

J

J

K

L

L

L

L

M

02 M. Yue et al. / Neurobiolo

rody, D.L., Holtzman, D.M., 2006. Morris water maze search strategyanalysis in PDAPP mice before and after experimental traumatic braininjury. Exp. Neurol. 197 (2), 330–340.

uee, L., Bussiere, T., Buee-Scherrer, V., Delacourte, A., Hof, P.R., 2000.Tau protein isoforms, phosphorylation and role in neurodegenerativedisorders. Brain Res. Brain Res. Rev. 33 (1), 95–130.

asadesus, G., Webber, K.M., Atwood, C.S., Pappolla, M.A., Perry, G.,Bowen, R.L., Smith, M.A., 2006. Luteinizing hormone modulates cog-nition and amyloid-beta deposition in Alzheimer APP transgenic mice.Biochim. Biophys. Acta 1762 (4), 447–452.

lark, L.N., Poorkaj, P., Wszolek, Z., Geschwind, D.H., Nasreddine, Z.S.,Miller, B., Li, D., Payami, H., Awert, F., Markopoulou, K., Andreadis, A.,D’Souza, I., Lee, V.M., Reed, L., Trojanowski, J.Q., Zhukareva, V., Bird,T., Schellenberg, G., Wilhelmsen, K.C., 1998. Pathogenic implications ofmutations in the tau gene in pallido-ponto-nigral degeneration and relatedneurodegenerative disorders linked to chromosome 17. Proc. Natl. Acad.Sci. U.S.A. 95 (22), 13103–13107.

order, E.H., Saunders, A.M., Pericak-Vance, M.A., Roses, A.D., 1995a.There is a pathologic relationship between ApoE-epsilon 4 andAlzheimer’s disease. Arch. Neurol. 52 (7), 650–651.

order, E.H., Saunders, A.M., Strittmatter, W.J., Schmechel, D.E., GaskellJr., P.C., Rimmler, J.B., Locke, P.A., Conneally, P.M., Schmader, K.E.,Tanzi, R.E., et al., 1995b. Apolipoprotein E, survival in Alzheimer’s dis-ease patients, and the competing risks of death and Alzheimer’s disease.Neurology 45 (7), 1323–1328.

otman, C.W., Su, J.H., 1996. Mechanisms of neuronal death in Alzheimer’sdisease. Brain Pathol. 6 (4), 493–506.

rawley, J.N., Paylor, R., 1997. A proposed test battery and constella-tion of specific behaviural paradigms to investigate the behaviouralphenotypes of transgenic and knockout mice. Horm. Behav. 31,197–211.

avies, R.R., Hodges, J.R., Kril, J.J., Patterson, K., Halliday, G.M., Xuereb,J.H., 2005. The pathological basis of semantic dementia. Brain 128 (Pt9), 1984–1995.

ickson, D.W., 2003. Neurodegeneration: The Molecular Pathology ofDementia and Movement Disorders. ISN Neuropath Press, Basel.

ichenbaum, H., 1996. Learning from LTP: a comment on recent attemptsto identify cellular and molecular mechanisms of memory. Learn. Mem.3, 61–73.

riksen, J.L., Janus, C.G., 2006. Plaques, tangles, and memory loss in mousemodels of neurodegeneration. Behav. Genet. 37 (1), 79–100.

rten-Lyons, D., Woltjer, R.L., Dodge, H., Nixon, R., Vorobik, R., Calvert,J.F., Leahy, M., Montine, T., Kaye, J., 2009. Factors associated with resis-tance to dementia despite high Alzheimer disease pathology. Neurology72 (4), 354–360.

arrer, L.A., Cupples, L.A., Haines, J.L., Hyman, B., Kukull, W.A., Mayeux,R., Myers, R.H., Pericak-Vance, M.A., Risch, N., van Duijn, C.M.,1997. Effects of age, sex, and ethnicity on the association betweenapolipoprotein E genotype and Alzheimer disease. A meta-analysis.APOE and Alzheimer Disease Meta Analysis Consortium. JAMA 278(16), 1349–1356.

leisher, A., Grundman, M., Jack Jr., C.R., Petersen, R.C., Taylor, C., Kim,H.T., Schiller, D.H., Bagwell, V., Sencakova, D., Weiner, M.F., DeCarli,C., DeKosky, S.T., van Dyck, C.H., Thal, L.J., 2005. Sex, apolipoproteinE epsilon 4 status, and hippocampal volume in mild cognitive impair-ment. Arch. Neurol. 62 (6), 953–957.

rick, K.M., Burlingame, L.A., Arters, J.A., Berger-Sweeney, J., 2000. Ref-erence memory, anxiety and estrous cyclicity in C57BL/6NIA mice areaffected by age and sex. Neuroscience 95 (1), 293–307.

rye, C.A., 1995. Estrus-associated decrements in a water maze task arelimited to acquisition. Physiol. Behav. 57 (1), 5–14.

rye, C.A., Sturgis, J.D., 1995. Neurosteroids affect spatial/reference, work-

ing, and long-term memory of female rats. Neurobiol. Learn. Mem. 64(1), 83–96.

ass, P., Wolfer, D.P., Balschun, D., Rudolph, D., Frey, U., Lipp, H.P.,Schütz, G., 1998. Deficits in memory tasks of mice with CREB mutationsdepend on gene dosage. Learn. Mem. 5, 274–288.

M

M

ging 32 (2011) 590–603

oedert, M., Jakes, R., 2005. Mutations causing neurodegenerativetauopathies. Biochim. Biophys. Acta 1739 (2–3), 240–250.

ossen, M., Bujard, H., 1995. Efficacy of tetracycline-controlled geneexpression is influenced by cell type: commentary. Biotechniques 19(2), 213–216, discussion 216–217.

reenberg, S.G., Davies, P., 1990. A preparation of Alzheimer pairedhelical filaments that displays distinct tau proteins by polyacry-lamide gel electrophoresis. Proc. Natl. Acad. Sci. U.S.A. 87 (15),5827–5831.

irata-Fukae, C., Li, H.F., Hoe, H.S., Gray, A.J., Minami, S.S., Hamada,K., Niikura, T., Hua, F., Tsukagoshi-Nagai, H., Horikoshi-Sakuraba, Y.,Mughal, M., Rebeck, G.W., LaFerla, F.M., Mattson, M.P., Iwata, N.,Saido, T.C., Klein, W.L., Duff, K.E., Aisen, P.S., Matsuoka, Y., 2008.Females exhibit more extensive amyloid, but not tau, pathology in anAlzheimer transgenic model. Brain Res. 1216, 92–103.

utton, M., Lendon, C.L., Rizzu, P., Baker, M., Froelich, S., Houlden, H.,Pickering-Brown, S., Chakraverty, S., Isaacs, A., Grover, A., Hackett,J., Adamson, J., Lincoln, S., Dickson, D., Davies, P., Petersen, R.C.,Stevens, M., de Graaff, E., Wauters, E., van Baren, J., Hillebrand, M.,Joosse, M., Kwon, J.M., Nowotny, P., Che, L.K., Norton, J., Morris,J.C., Reed, L.A., Trojanowski, J., Basun, H., Lannfelt, L., Neystat, M.,Fahn, S., Dark, F., Tannenberg, T., Dodd, P.R., Hayward, N., Kwok,J.B., Schofield, P.R., Andreadis, A., Snowden, J., Craufurd, D., Neary,D., Owen, F., Oostra, B.A., Hardy, J., Goate, A., van Swieten, J., Mann,D., Lynch, T., Heutink, P., 1998. Association of missense and 5′-splice-site mutations in tau with the inherited dementia FTDP-17. Nature 393(6686), 702–705.

mwalle, D.B., Bateman, H.L., Wills, A., Honda, S., Harada, N., Rissman,E.F., 2006. Impairment of spatial learning by estradiol treatment infemale mice is attenuated by estradiol exposure during development.Horm. Behav. 50 (5), 693–698.

qbal, K., Alonso Adel, C., Chen, S., Chohan, M.O., El-Akkad, E., Gong,C.X., Khatoon, S., Li, B., Liu, F., Rahman, A., Tanimukai, H., Grundke-Iqbal, I., 2005. Tau pathology in Alzheimer disease and other tauopathies.Biochim. Biophys. Acta 1739 (2–3), 198–210.

anus, C., 2004. Search strategies used by APP transgenic mice during spatialnavigation in the Morris water maze. Learn. Mem. 11, 337–346.

anus, C., 2008. Contitionally inducible tau mice—designing a better mousemodel of neurodegenerative diseases. Genes Brain Behav. 7 (Suppl. 1),12–28.

avaliers, M., Ossenkopp, K.P., Prato, F.S., Innes, D.G., Galea, L.A., Kin-sella, D.M., Perrot-Sinal, T.S., 1996. Spatial learning in deer mice: sexdifferences and the effects of endogenous opioids and 60 Hz magneticfields. J. Comp. Physiol. [A] 179 (5), 715–724.

ee, V.M., Goedert, M., Trojanowski, J.Q., 2001. Neurodegenerativetauopathies. Annu. Rev. Neurosci. 24, 1121–1159.

ewis, J., McGowan, E., Rockwood, J., Melrose, H., Nacharaju, P., VanSlegtenhorst, M., Gwinn-Hardy, K., Paul Murphy, M., Baker, M., Yu,X., Duff, K., Hardy, J., Corral, A., Lin, W.L., Yen, S.H., Dickson, D.W.,Davies, P., Hutton, M., 2000. Neurofibrillary tangles, amyotrophy andprogressive motor disturbance in mice expressing mutant (P301L) tauprotein. Nat. Genet. 25 (4), 402–405.

ewis, J., Dickson, D.W., Lin, W.L., Chisholm, L., Corral, A., Jones, G.,Yen, S.H., Sahara, N., Skipper, L., Yager, D., Eckman, C., Hardy, J.,Hutton, M., McGowan, E., 2001. Enhanced neurofibrillary degenerationin transgenic mice expressing mutant tau and APP. Science 293 (5534),1487–1491.

ipp, H.P., Wolfer, D.P., 1998. Genetically modified mice and cognition.Curr. Opin. Neurobiol. 8, 272–280.

arkowska, A.L., 1999. Sex dimorphisms in the rate of age-related decline inspatial memory: relevance to alterations in the estrous cycle. J. Neurosci.19 (18), 8122–8133.

ayford, M., Bach, M.E., Huang, Y.Y., Wang, L., Hawkins, R.D., Kandel,E.R., 1996. Control of memory formation through regulated expressionof a CaMKII transgene. Science 274 (5293), 1678–1683.

orris, R., 1984. Developments of a water-maze procedure for studyingspatal learning in the rat. J. Neurosci. Methods 11, 47–60.

gy of A

M

P

R

R

R

R

R

R

R

S

S

S

S

S

S

S

S

S

T

T

v

W

W

W

W

W

M. Yue et al. / Neurobiolo

orris, R.G.M., Garrund, P., Rawlings, J., O’Keefe, J., 1982. Place naviga-tion impaired in rats with hippocampal lesions. Nature 297, 681–683.

errot-Sinal, T.S., Kostenuik, M.A., Ossenkopp, K.P., Kavaliers, M., 1996.Sex differences in performance in the Morris water maze and the effectsof initial nonstationary hidden platform training. Behav. Neurosci. 110(6), 1309–1320.

aber, J., Wong, D., Buttini, M., Orth, M., Bellosta, S., Pitas, R.E., Mahley,R.W., Mucke, L., 1998. Isoform-specific effects of human apolipoproteinE on brain function revealed in ApoE knockout mice: increased suscep-tibility of females. Proc. Natl. Acad. Sci. U.S.A. 95 (18), 10914–10919.

aber, J., Wong, D., Yu, G.Q., Buttini, M., Mahley, R.W., Pitas, R.E.,Mucke, L., 2000. Apolipoprotein E and cognitive performance. Nature404 (6776), 352–354.

aber, J., Bongers, G., LeFevour, A., Buttini, M., Mucke, L., 2002. Andro-gens protect against apolipoprotein E4-induced cognitive deficits. J.Neurosci. 22 (12), 5204–5209.

aber, J., 2004. Androgens, apoE, and Alzheimer’s disease. Sci. AgingKnowledge Environ. 2004 (11), re2.

amsden, M., Kotilinek, L., Forster, C., Paulson, J., McGowan, E., San-taCruz, K., Guimaraes, A., Yue, M., Lewis, J., Carlson, G., Hutton, M.,Ashe, K.H., 2005. Age-dependent neurofibrillary tangle formation, neu-ron loss, and memory impairment in a mouse model of human tauopathy(P301L). J. Neurosci. 25 (46), 10637–10647.

hodin, J.A., Thomas, T., 2001. A vascular connection to Alzheimer’s dis-ease. Microcirculation 8 (4), 207–220.

ibé, E.M., Perez, M., Puig, B., Gich, I., Lim, F., Cuadrado, M., Sesma,T., Catena, S., Sanchez, B., Nieto, M., Gomez-Ramos, P., Moran, M.A.,Cabodevilla, F., Samaranch, L., Ortiz, L., Perez, A., Ferrer, I., Avila, J.,Gomez-Isla, T., 2005. Accelerated amyloid deposition, neurofibrillarydegeneration and neuronal loss in double mutant APP/tau transgenicmice. Neurobiol. Dis. 20 (3), 814–822.

anta Cruz, K., Lewis, J., Spires, T., Paulson, J., Kotilinek, L., Ingelsson,M., Guimaraes, A., DeTure, M., Ramsden, M., McGowan, E., Forster,C., Yue, M., Orne, J., Janus, C., Mariash, A., Kuskowski, M., Hyman,B., Hutton, M., Ashe, K.H., 2005. Tau suppression in a neurodegen-erative mouse model improves memory function. Science 309 (5733),476–481.

chenk, F., 1985. Development of place navigation in rats from weaning topuberty. Behav. Neural Biol. 43 (1), 69–85.

chindowski, K., Bretteville, A., Leroy, K., Begard, S., Brion, J.P., Hamdane,M., Buee, L., 2006. Alzheimer’s disease-like tau neuropathology leadsto memory deficits and loss of functional synapses in a novel mutatedtau transgenic mouse without any motor deficits. Am. J. Pathol. 169 (2),599–616.

W

ging 32 (2011) 590–603 603

elkoe, D.J., 1997. Alzheimer’s disease: genotypes, phenotypes, and treat-ments. Science 275, 630–631.

ergeant, N., Delacourte, A., Buee, L., 2005. Tau protein as a differentialbiomarker of tauopathies. Biochim. Biophys. Acta 1739 (2–3), 179–197.

pires, T.L., Meyer-Luehmann, M., Stern, E.A., McLean, P.J., Skoch, J.,Nguyen, P.T., Bacskai, B.J., Hyman, B.T., 2005. Dendritic spine abnor-malities in amyloid precursor protein transgenic mice demonstrated bygene transfer and intravital multiphoton microscopy. J. Neurosci. 25 (31),7278–7287.

pires, T.L., Orne, J.D., SantaCruz, K., Pitstick, R., Carlson, G.A., Ashe,K.H., Hyman, B.T., 2006. Region-specific dissociation of neuronal lossand neurofibrillary pathology in a mouse model of tauopathy. Am. J.Pathol. 168 (5), 1598–1607.

quire, L.R., 1992. Memory and the hippocampus: a synthesis from findingswith rats, monkeys, and humans. Psychol. Rev. 99 (2), 195–231.

tevens, J., 1990. Intermediate Statistics: A Modern Approach. LawrenceErlbaum Associates, Inc., Publishers, Hillsdale, NJ.

erry, R.D., 2006. Alzheimer’s disease and the aging brain. J. Geriatr. Psy-chiatry Neurol. 19 (3), 125–128.

erwel, D., Lasrado, R., Snauwaert, J., Vandeweert, E., Van Haesendonck,C., Borghgraef, P., Van Leuven, F., 2005. Changed conformation ofmutant Tau-P301L underlies the moribund tauopathy, absent in progres-sive, nonlethal axonopathy of Tau-4R/2N transgenic mice. J. Biol. Chem.280 (5), 3963–3973.

an Meer, P., Acevedo, S., Raber, J., 2007. Impairments in spatial mem-ory retention of GFAP-apoE4 female mice. Behav. Brain Res. 176 (2),372–375.

ahlsten, D., Cooper, S.F., Crabbe, J.C., 2005. Different rankings of inbredmouse strains on the Morris maze and a refined 4-arm water escape task.Behav. Brain Res. 165 (1), 36–51.

arren, S.G., Juraska, J.M., 1997. Spatial and nonspatial learning across therat estrous cycle. Behav. Neurosci. 111 (2), 259–266.

ebber, K.M., Bowen, R., Casadesus, G., Perry, G., Atwood, C.S., Smith,M.A., 2004. Gonadotropins and Alzheimer’s disease: the link betweenestrogen replacement therapy and neuroprotection. Acta Neurobiol. Exp.(Wars) 64 (1), 113–118.

ehner, J.M., Sleight, S., Upchurch, M., 1990. Hippocampal protein kinaseC activity is reduced in poor spatial learners. Brain Res. 523, 181–187.

illiams, M.M., Xiong, C., Morris, J.C., Galvin, J.E., 2006. Survival and

mortality differences between dementia with Lewy bodies vs Alzheimerdisease. Neurology 67 (11), 1935–1941.

olfer, D.P., Lipp, H.P., 2000. Dissecting the behaviour of transgenic mice:is it the mutation, the genetic background, or the environment. Exp.Physiol. 85 (6), 627–634.