Clinical and Experimental Pharmacology and Physiology (2014) 41, 798–806

doi: 10.1111/1440-1681.12277

ORIGINAL ARTICLE

Comparative studies using the Morris water maze to assess spatial memory deficits in two transgenic mouse models of Alzheimer’s disease Stephen R Edwards,* Adam S Hamlin,†‡ Nicola Marks,† Elizabeth J Coulson† and Maree T Smith* *Centre for Integrated Preclinical Drug Development and †Queensland Brain Institute, The University of Queensland, Brisbane, Qld, Australia

SUMMARY Evaluation of the efficacy of novel therapeutics for potential treatment of Alzheimer’s disease (AD) requires an animal model that develops age-related cognitive deficits reproducibly between independent groups of investigators. Herein we assessed comparative temporal changes in spatial memory function in two commercially available transgenic mouse models of AD using the Morris water maze (MWM), incorporating both visible and hidden platform training. Individual cohorts of cDNA-based ‘line 85’-derived double-transgenic mice coexpressing the ‘Swedish’ mutation of amyloid precursor protein (APPSwe) and the presenillin 1 (PS1) ‘dE9’ mutation were assessed in the MWM at mean ages of 3.6, 9.3 and 14.8 months. We found significant deficits in spatial memory retention in APPSwe/PS1dE9 mice aged 3.6 months and robust deficits in spatial memory acquisition and retention in APPSwe/PS1dE9 mice aged 9.3 months, with a further significant decline by age 14.8 months. b-Amyloid deposits were present in brain sections by 7.25 months of age. In contrast, MWM studies with individual cohorts (aged 4–21 months) of single-transgenic genomic-based APPSwe mice expressing APPSwe on a yeast artificial chromosomal (YAC) construct showed no significant deficits in spatial memory acquisition until 21 months of age. There were no significant deficits in spatial memory retention up to 21 months of age and b-amyloid deposits were not present in brain sections up to 24 months of age. These data, generated using comprehensive study designs, show that APPSwe/PS1dE9 but not APPSwe YAC mice appear to provide a suitably robust model of AD for efficacy assessment of novel AD treatments in development. Key words: Alzheimer’s disease, b-amyloid, APPSwe YAC mice, APPSwe/PS1dE9 mice, cognitive deficits, Morris water maze, spatial memory, transgenic mouse models.

Correspondence: Maree T Smith, Centre for Integrated Preclinical Drug Development, The University of Queensland, St Lucia, Qld 4072, Australia. Email: [email protected] ‡ Present address: School of Biomedical Sciences, Charles Sturt University, Wagga Wagga, NSW, Australia Received 15 April 2014; revision 5 June 2014; accepted 8 June 2014. © 2014 Wiley Publishing Asia Pty Ltd

INTRODUCTION Alzheimer’s disease (AD) is the most common form of dementia, affecting 1–5% of the population aged > 65 years and 20–40% of the population aged > 85 years.1 Patients with AD exhibit progressive memory loss, cognitive decline and behavioural changes. The neuropathology of AD is characterized by the presence of neuritic plaques containing deposits of b-amyloid (Ab) that are associated with reactive microglia and astrocytes, intraneuronal neurofibrillary tangles and neuronal loss.2 Numerous transgenic mouse models of AD have been developed. Although they do not recapitulate all aspects of the human disease, they are widely used to investigate the pathophysiology of AD and for assessing potential novel treatments and biomarkers of disease progression.3 Transgenic mice expressing mutational variants of human amyloid precursor protein (APP) found in familial forms of AD develop key features of AD, including Ab deposits surrounded by reactive astrocytes and microglia, as well as hyperphosphorylated tau and local neuronal loss in proximity to Ab deposits.4 However, assessment of memory impairments in AD transgenic mouse models has produced less consistent results. For instance, in Tg2576 mice, Ab deposits develop in the cortex from 12 months of age.5 Although some studies have reported robust impairment of spatial memory acquisition and retention in the Morris water maze (MWM) by 12–18 months of age in Tg2576 mice,6 deficits in acquisition and retention in the MWM were not apparent by 19 months of age in work reported by others.7 For an animal model of AD to have validity, it should exhibit progressive AD-like neuropathology and cognitive deficits that have been verified by several independent investigators.8 Hence, investigations characterizing age-related impairments in behavioural memory performance in transgenic mouse models in association with AD-like neuropathology are essential. Thus, the aim of the present study was to undertake a comparative evaluation of two commercially available transgenic mouse models of AD supplied by Jackson Laboratories (Bar Harbor, ME, USA; http://jaxmice.jax.org, accessed 22 July 2014), to assess their comparative utility for future use in studies assessing the efficacy of novel therapeutic agents and/or utility of proposed biomarkers of AD progression. Single-transgenic genomic-based ‘B6-R1.40’ mice developed by Lehman et al.9 expressing the ‘Swedish’ mutation of amyloid precursor protein (APPSwe) contain a genomic copy of APP with

APPSwe/PS1dE9 mice for efficacy testing the ‘Swedish’ mutation carried on a yeast artificial chromosome (YAC) and reportedly develop Ab deposits by 13.5 months of age.9 Double-transgenic cDNA-based ‘line 85’-APPSwe/presenillin 1 (PS1) ‘dE9’ mutation (PS1dE9) mice developed by Jankowsky et al.10 coexpress a chimeric mouse–human APP containing the ‘Swedish’ mutation with a mutational variant of human PS1 (PS1dE9) produced by coinjection of both vectors and reportedly develop Ab deposits by 6 months of age.10 Herein, we report data from comparative studies using the MWM to assess spatial memory function in individual cohorts of APPSwe/PS1dE9 and APPSwe YAC mice out to 14.8 and 21 months of age, respectively. Histological assessment of brain sections for the presence of Ab deposits was performed on completion of behavioural testing.

RESULTS Double-transgenic APPSwe/PS1dE9 mouse MWM studies For visible platform (VP) training (Fig. 1a,c,e), there was a significant effect of day on the escape latencies for each age group tested (3.6 months: F3,48 = 31.93, P < 0.0001; 9.3 months: F3,48 = 43.51, P < 0.0001; 14.8 months: F3,48 = 34.73, P < 0.0001). Mean (
SEM) VP escape latencies were significantly lower (P < 0.0001) on Day 4 compared with Day 1 for both transgenic and wild-type mice in each age group, indicating learning of this simple task and visual acuity of mice of both genotypes was intact. In all, one transgenic and two wild-type mice were deemed to be ‘poor performers’ and were excluded from the study. (a)

(b)

(c)

(d)

(e)

(f)

Fig. 1 Mean (
SEM; n = 8–10) escape latencies during (a, c, e) visible platform and (b, d, f) hidden platform training for separate cohorts of APPSwe/PS1dE9 (○) and wild-type (□) mice aged 3.6 (a, b), 9.3 (c, d) and 14.8 months (e, f). *P < 0.05, **P < 0.005 compared with wild-type mice on the same day (repeated-measures two-way ANOVA followed by post hoc pairwise comparisons using Fisher’s LSD test).

799

There was a significant effect of transgene on VP escape latencies for mice aged 14.8 months only (3.6 months: F1,16 = 0.36, P > 0.05; 9.3 months: F1,16 = 3.93, P > 0.05; 14.8 months: F1,16 = 5.55, P < 0.05). For transgenic mice aged 14.8 months, mean values of VP escape latency were significantly higher (P < 0.05) than those from wild-type mice on Days 1 and 2 of training, but importantly they did not differ significantly (P > 0.05) on Days 3 and 4 of training. Hence, this difference was most likely due to a cognitive deficit rather than a sensorimotor deficit. For hidden platform (HP) training (Fig. 1b,d,f), there was a significant effect of day on the escape latencies for mice in each age group tested (3.6 months: F9,135 = 4.54, P < 0.0001; 9.3 months: F9,144 = 5.21, P < 0.0001; 14.8 months: F9,144 = 3.13, P < 0.005). Mean HP escape latency values were significantly lower (P < 0.005) on Day 10 compared with Day 1 for wild-type mice from each age group tested. However, although mean HP escape latencies were significantly lower (P < 0.05) on Day 10 compared with Day 1 for transgenic mice aged 3.6 and 9.3 months, they were not significantly different (P > 0.05) for transgenic mice aged 14.8 months. There was a significant effect of transgene on HP escape latencies for mice aged 9.3 months (F1,16 = 11.41, P < 0.005) and 14.8 months (F1,16 = 33.0, P < 0.0001), but not 3.6 months (F1,15 = 3.27, P > 0.05). For mice aged 9.3 months, mean HP escape latencies were significantly higher (P < 0.05) for transgenic compared with wild-type mice on Days 1, 2, 7, 9 and 10. For mice aged 14.8 months, mean HP escape latencies were significantly higher (P < 0.005) for transgenic compared with wildtype mice on Days 2, 3 and 5–10. For probe trial values of the percentage of time spent in the training quadrant (%Time; Fig. 2a,c,e), there was a significant effect of day for mice aged 9.3 months (F2,32 = 7.0, P < 0.005), but not 3.6 months (F2,30 = 1.21, P > 0.05) or 14.8 months (F2,32 = 2.57, P > 0.05). Mean values for %Time were significantly higher (P < 0.05) for Day 10 compared with Day 4 probe trials for both transgenic and wild-type mice aged 9.3 months. There was a significant effect of transgene on %Time for mice aged 9.3 months (F1,16 = 14.9, P < 0.005) and 14.8 months (F1,16 = 10.8, P < 0.005), but not 3.6 months (F1,15 = 1.90, P > 0.05). Mean values of %Time were significantly lower (P < 0.05) for transgenic compared with wild-type mice aged 9.3 months on Days 4, 7 and 10, as well as for transgenic compared with wild-type mice aged 14.8 months on Days 4 and 7. For probe trial values of the average proximity to the platform location (Fig. 2b,d,f), there was a significant effect of day for mice aged 9.3 months (F2,32 = 10.6, P < 0.0005) and 14.8 months (F2,32 = 3.32, P < 0.05), but not 3.6 months (F2,30 = 1.02, P > 0.05). Mean values of average proximity were significantly lower (P < 0.05) for Day 10 compared with Day 4 probe trials for transgenic and wild-type mice aged 9.3 months and for transgenic, but not wild-type, mice aged 14.8 months. There was a significant effect of transgene on average proximity for mice aged 9.3 months (F1,16 = 8.67, P < 0.01) and 14.8 months (F1,16 = 13.9, P < 0.005), but not 3.6 months (F1,15 = 2.62, P > 0.05). However, for mice aged 3.6 months, there was a significant interaction between day and transgene (F2,30 = 5.35, P < 0.05). Pairwise comparison using Student’s unpaired t-test found that mean Day 10 values of average

© 2014 Wiley Publishing Asia Pty Ltd

SR Edwards et al.

800 (a)

(b)

compared with wild-type mice aged 9.3 months on Days 7 and 10 and for transgenic compared with wild-type mice aged 14.8 months on Days 4, 7 and 10. Data for cumulative distance to the escape platform and mean velocity during VP and HP training are shown in Figs S1 and S2, available as Supplementary Material to this paper.

(c)

(d)

Single-transgenic APPSwe YAC mouse MWM studies

(e)

(f)

Fig. 2 Mean (
SEM; n = 8–9) values of (a, c, e) percentage time spent in the in the training quadrant (%Time) and (b, d, f) average proximity to platform location during Day 4, 7 and 10 probe trials for separate cohorts of APPSwe/PS1dE9 (□) and wild-type ( ) mice aged 3.6 (a, b), 9.3 (c, d) and 14.8 months (e, f). *P < 0.05, **P < 0.005 compared with wild-type mice on the same day (repeated-measures two-way ANOVA followed by post hoc pairwise comparisons using Fisher’s LSD test).

proximity for mice aged 3.6 months were significantly higher (P < 0.05) for transgenic compared with wild-type mice. Mean values of average proximity were significantly higher (P < 0.05) for transgenic compared with wild-type mice aged 9.3 months on Days 7 and 10 and for transgenic compared with wild-type mice aged 14.8 months on Days 4, 7 and 10. In summary, VP escape latencies for transgenic and wild-type mice aged 3.6 and 9.3 months were not significantly different (P > 0.05). For mice aged 14.8 months, VP escape latencies were significantly higher (P < 0.05) for transgenic compared with wild-type mice on Days 1 and 2 of training, but they were not significantly different (P > 0.05) on Days 3 and 4. The HP escape latencies for transgenic and wild-type mice aged 3.6 months were not significantly different (P > 0.05). For mice aged 9.3 months, HP escape latencies were significantly higher (P < 0.05) for transgenic compared with wild-type mice on Days 1, 2, 7, 9 and 10 of training. For mice aged 14.8 months, HP escape latencies were significantly higher (P < 0.005) for transgenic compared with wild-type mice on Days 2 and 5–10 of training. Probe trial %Time values for transgenic and wild-type mice aged 3.6 months were not significantly different (P > 0.05). The %Time values were significantly lower (P < 0.05) for transgenic compared with wild-type mice aged 9.3 months on Days 4, 7 and 10 and for transgenic compared with wild-type mice aged 14.8 months on Days 4 and 7. Probe trial average proximity values were significantly higher (P < 0.05) for transgenic compared with wild-type mice aged 3.6 months on Day 10. Average proximity values were significantly higher (P < 0.05) for transgenic

For VP training (Fig. S3), there was a significant effect of day on escape latencies in each age group tested (4 months: F3,69 = 55.9, P < 0.0001; 7 months: F3,108 = 118.5, P < 0.0001; 10 months: F3,96 = 84.5, P < 0.0001; 16 months: F3,111 = 126.6, P < 0.0001; 18 months: F3,99 = 84.4, P < 0.0001; 21 months F3,96 = 68.2, P < 0.0001). Mean values of VP escape latency were significantly lower (P < 0.0001) on Day 4 compared with Day 1 for both transgenic and wild-type mice in each age group, indicating that learning of this simple task and visual acuity of both genotypes was intact. No mice were excluded from the study as ‘poor performers’. However, there was no significant effect of transgene on VP escape latencies for each age group tested (4 months: F1,23 = 0.07, P > 0.05; 7 months: F1,36 = 0.48, P > 0.05; 10 months F1,32 = 0.46, P > 0.05; 16 months: F1,37 = 0.002, P > 0.05; 18 months F1,33 = 0.32, P > 0.05; 21 months: F1,32 = 0.37, P > 0.05). Hence, the performance of transgenic mice was equivalent to that of their wild-type counterparts during VP training. For HP training (Fig. 3), there was a significant effect of day on escape latencies in each age group tested (4 months: F9,207 = 7.0, P < 0.0001; 7 months: F9,324 = 19.3, P < 0.0001; 10 months: F9,288 = 11.5; P < 0.0001; 16 months: F9,333 = 16.6, P < 0.0001; 18 months: F9,297 = 12.4, P < 0.0001; 21 months: F9,288 = 15.5; P < 0.0001). Mean values of HP escape latency were significantly lower (P < 0.0001) on Day 10 compared with Day 1 for both transgenic and wild-type mice aged 7, 10, 16, 18 and 21 months. For mice aged 4 months, mean values of HP escape latencies were significantly lower (P < 0.0001) on Day 10 compared with Day 1 for wild-type mice, but were not significantly different (P > 0.05) for transgenic mice. There was no significant effect of transgene on HP escape latencies for mice aged 4–18 months (4 months: F1,23 = 0.07, P > 0.05; 7 months: F1,36 = 2.90, P > 0.05; 10 months: F1,32 = 2.90, P > 0.05; 16 months: F1,37 = 0.26, P > 0.05; 18 months: F1,33 = 0.26, P > 0.05). However, there was a significant effect of transgene for mice aged 21 months (F1,32 = 4.65, P < 0.05). Mean values of HP escape latency were significantly higher (P < 0.05) for transgenic compared with wild-type mice on Days 8 and 10. For probe trial %Time values (Fig. 4), there was a significant effect of day for mice in each age group tested (4 months: F2,46 = 3.75, P < 0.05; 7 months: F2,72 = 21.7, P < 0.0001; 10 months: F2,64 = 8.54, P < 0.001; 16 months: F2,74 = 9.63, P < 0.0005; 18 months: F2,66 = 6.92, P < 0.005; 21 months: F2,64 = 13.2; P < 0.0001). For mice aged 4 months, mean values of %Time were significantly higher (P < 0.05) for Day 10 compared with Day 4 probe trials for wild-type, but not transgenic, mice. At 7, 10, 16, 18 and 21 months of age, mean values of %Time were significantly higher (P < 0.05) for Day 10 compared with Day 4 probe trials for both transgenic and wildtype mice.

© 2014 Wiley Publishing Asia Pty Ltd

APPSwe/PS1dE9 mice for efficacy testing (a)

(b)

(c)

(d)

(e)

(f)

Fig. 3 Mean (
SEM; n = 10–21) escape latencies during hidden platform training for separate cohorts of APPSwe YAC (○) and wild-type (□) mice aged (a) 4, (b) 7, (c) 10, (d) 16, (e) 18 and (f) 21 months. *P < 0.05 (21 months) compared with wild-type mice on the same day (repeated-measures two-way ANOVA followed by post hoc pairwise comparisons using Fisher’s LSD test).

(a)

(b)

(c)

(d)

(e)

801

However, there was no significant effect of transgene on % Time for mice in each age group tested (4 months: F1,23 = 0.96, P > 0.05; 7 months: F1,36 = 0.03, P > 0.05; 10 months: F1,32 = 1.41, P > 0.05; 16 months: F1,37 = 0.03, P > 0.05; 18 months: F1,33 = 1.54, P > 0.05; 21 months: F1,32 = 1.84, P > 0.05). For probe trial average proximity values (Fig. S4), there was a significant effect of day for mice in each age group tested (4 months: F2,46 = 5.82, P < 0.01; 7 months: F2,72 = 24.4, P < 0.0001; 10 months: F2,64 = 6.97, P < 0.005; 16 months: F2,74 = 10.2, P < 0.0005; 18 months: F2,66 = 10.3, P < 0.0005; 21 months: F2,64 = 12.9, P < 0.0001). For mice aged 4 months, mean values of average proximity were significantly lower (P < 0.05) for Day 10 compared with Day 4 probe trials for wild-type, but not transgenic, mice. At 7, 10, 16, 18 and 21 months of age, mean values of average proximity were significantly lower (P < 0.05) for Day 10 compared with Day 4 probe trials for both transgenic and wild-type mice. However, there was no significant effect of transgene on average proximity for mice in each age group tested (4 months: F1,23 = 0.74, P > 0.05; 7 months: F1,36 = 0.46, P > 0.05; 10 months: F1,32 = 1.85, P > 0.05; 16 months: F1,37 = 0.55, P > 0.05; 18 months: F1,33 = 2.68, P > 0.05; 21 months: F1,32 = 1.50, P > 0.05). In summary, VP escape latencies for transgenic and wild-type mice aged 4, 7, 10, 16, 18 and 21 months were not significantly different (P > 0.05). The HP escape latencies for transgenic and wild-type mice aged 4, 7, 10, 16 and 18 months were not significantly different (P > 0.05). However, for mice aged 21 months, HP escape latencies were significantly higher (P < 0.05) for transgenic compared with wild-type mice on Days 8 and 10 of training. Probe trial %Time and average proximity values for transgenic and wild-type mice were not significantly different (P > 0.05) for each age group tested. Data for cumulative distance to the escape platform and mean velocity during VP and HP training are shown in Figs S5–S8. b-Amyloid deposition in the cortex and dorsal hippocampus of APPSwe/PS1dE9 and APPSwe YAC mice Manual counts of congophilic Ab deposits (Fig. 5) were significantly higher (P < 0.0001) in the cortex (3.4-fold) and dorsal hippocampus (2.2-fold) of APPSwe/PS1dE9 mice aged 12 months than mice aged 10.5 months. Representative images of Ab deposits in cortex and dorsal hippocampus from an APPSwe/PS1dE9 mouse aged 12 months are shown in Fig. 6. Congophilic Ab deposits were present in the cortex and hippocampus of brain sections from APPSwe/PS1dE9 mice aged 7.25 months, but were not readily detected in cortical and hippocampal sections from APPSwe/PS1dE9 mice aged 3.6 months. Cortical and hippocampal congophilic Ab deposits were not detected in brain sections from APPSwe YAC mice at any age from 4 to 24 months.

(f)

Fig. 4 Mean (
SEM; n = 10–21) percentage time spent in the training quadrant (%Time) during Day 4, 7 and 10 probe trials for separate cohorts of APPSwe YAC (□) and wild-type ( ) mice aged (a) 4, (b) 7, (c) 10, (d) 16, (e) 18 and (f) 21 months. There was no significant effect (P > 0.05) for transgene status at any age group (repeated-measures two-way ANOVA).

DISCUSSION Efficacy assessment of novel agents for the potential treatment of AD requires an animal model that develops progressive AD-like neuropathological changes and cognitive deficits that are

© 2014 Wiley Publishing Asia Pty Ltd

SR Edwards et al.

802 (a)

(b)

Fig. 5 Mean (
SEM) manual counts of congophilic b-amyloid (Ab) deposits in the (a) cortex and (b) dorsal hippocampus of APPSwe/PS1dE9 transgenic mice aged 10.5 and 12 months (n = 3 mice per group). Bilateral counts of Ab deposits were performed in five to six sections from each mouse. *P < 0.0001 compared with 10.5 months (Student’s unpaired t-test).

(a)

(b)

Fig. 6 Representative images of b-amyloid (Ab) deposits in the (a) cortex and (b) dorsal hippocampus from an APPSwePS1/dE9 mouse aged 12 months. Bars, 1 mm.

reproducible.8 In addition, recent articles published in Nature highlight the importance of replication studies by independent groups of investigators to verify the validity of preclinical

research findings.11,12 To this end, our data reported herein, generated using comprehensive study designs using the MWM with both VP and HP training, show marked differences in the temporal profiles for performance in the MWM task between two commercially available transgenic mouse models of AD, namely APPSwe/PS1dE9 and APPSwe YAC mice. Specifically, ‘line 85’-derived hemizygous APPSwe/PS1dE9 mice exhibited a significant deficit in spatial memory retention in Day 10 MWM probe trials by 3.6 months of age. In addition, APPSwe/PS1dE9 mice aged 9.3 months exhibited robust, significant deficits in spatial memory acquisition during MWM HP training and significant performance deficits were present on Days 1, 2, 7, 9 and 10 of training compared with their wild-type counterparts. There was a further significant decline in APPSwe/ PS1dE9 mice aged 14.8 months and significant performance deficits were present on Days 2, 3 and 5–10 of HP training compared with their wild-type counterparts. Significant deficits in spatial memory retention were also apparent in Day 4, 7 and 10 MWM probe trials for APPSwe/PS1dE9 mice aged 9.3 and 14.8 months. Consistent with these spatial memory deficits in APPSwe/ PS1dE9 mice, histological assessment confirmed the presence of congophilic Ab deposits in the cortex and hippocampus of brain sections from APPSwe/PS1dE9 mice aged 7.25 months. In addition, congophilic Ab deposits increased significantly in the cortex (3.4-fold) and hippocampus (2.2-fold) of APPSwe/PS1dE9 mice aged 12 months compared with mice aged 10.5 months. Our findings are aligned with work by others, whereby Ab deposits in brain sections from APPSwe/PS1dE9 mice were detected using thioflavin S by 4 months of age and Ab deposition increased progressively between 4 and 12 months of age.13 Importantly, in the present study there were no significant differences between APPSwe/PS1dE9 mice and their wild-type counterparts during VP training in the MWM task at 3.6 and 9.3 months of age, indicating that sensorimotor deficits were not present in APPSwe/PS1dE9 mice at these ages. Although performance of 14.8-month-old APPSwe/PS1dE9 mice was significantly poorer on Days 1 and 2 of VP training compared with their wild-type counterparts, Day 3 and 4 VP escape latencies for both APPSwe/PS1dE9 and wild-type mice were equivalent, indicating that the impaired performance of 14.8-month-old APPSwe/ PS1dE9 mice in the MWM task was most likely due to cognitive deficits and not related to sensorimotor deficits. The HP escape latencies were significantly lower on Day 10 compared with Day 1 for APPSwe/PS1dE9 and wild-type mice aged 3.6 and 9.3 months, indicating that spatial memory acquisition improved over the training period for both genotypes at these ages. Hence, although APPSwe/PS1dE9 mice aged 9.3 months demonstrated deficits in spatial memory acquisition compared with their wild-type counterparts, their acquisition still improved over the 10 day training period. However, for mice aged 14.8 months, HP escape latencies were significantly lower on Day 10 compared with Day 1 for wild-type, but not APPSwe/ PS1dE9, mice, indicating that spatial memory acquisition did not improve during HP training for APPSwe/PS1 mice aged 14.8 months. Our findings are consistent with recent findings reported by others using less comprehensive study designs than those used herein, namely that ‘line 85’-derived APPSwe/PS1dE9 mice aged

© 2014 Wiley Publishing Asia Pty Ltd

APPSwe/PS1dE9 mice for efficacy testing 7 and 8 months had significant deficits in spatial memory acquisition and retention in the MWM task.14,15 A particular strength of the MWM paradigm used herein, in contrast with the foregoing studies,14,15 is the use of VP training to exclude the potential of sensorimotor deficits to confound performance of APPSwe/ PS1dE9 mice in the MWM task. However, in other studies in which APPSwe/PS1dE9 mice underwent VP and HP training for 2 and 6 days, respectively, there were no significant differences in their MWM performance at 12 months of age compared with their wild-type counterparts.16 More recently, significant deficits in reversal learning in the MWM task were found for female but not male APPSwe/ PS1dE9 mice aged 9 months compared with their wild-type counterparts.17 Factors potentially contributing to the aforementioned betweenstudy differences include gender, background strain of the mice being tested, as well as procedural variables used in the MWM task (see Tables 1 and 2). For example, genetic background is an important variable influencing learning in the MWM task.18 In one study,15 deficits in acquisition and retention in the MWM task were reported for male APPSwe/PS1dE9 mice on a C57BL/ 6J 9 C3H hybrid background by 7 months of age, whereas in another study16 performance deficits in the MWM task were not apparent in female APPSwe/PS1dE9 mice on a B6C3F1/J background by 12 months of age (Table 1). Similarly, there have been replication difficulties in MWM studies that used Tg2576 mice where differences in gender and

803

background strain may have been contributing factors (Table 1). For example, the significant deficits in spatial memory retention by 6–11 months of age and significant deficits in both spatial memory acquisition and retention by 12–18 months of age for Tg2576 mice on a B6SJLF1 background6 were not reproduced in Tg2576 mice on a C57BL/6 background.7 In the latter study,7 there was no overall impairment in spatial memory acquisition and retention up to 19 months of age, although acquisition was significantly impaired for female (but not male) Tg2576 mice aged 19 months.7 However, more recently, for male Tg2576 mice on a C57BL6 SJL F1 background, deficits in spatial memory retention were evident at 10 months of age.19 Herein, we report no significant difference between hemizygous single-transgenic ‘B6-R1.40’-derrived APPSwe YAC mice and their wild-type counterparts during VP training in the MWM task for individual cohorts of mice aged from 4 to 21 months. In addition, VP escape latencies were significantly lower on Day 4 compared with Day 1 for mice of both genotypes in each age group tested. Hence, sensorimotor deficits were not present in APPSwe YAC mice up to 21 months of age. Spatial memory acquisition was significantly impaired during MWM HP training for APPSwe YAC mice aged 21 months, with significant performance deficits on Days 8 and 10 of HP training compared with their wild-type counterparts. However, spatial memory retention during the MWM probe trials on Days 4, 7 and 10 was not significantly impaired in individual cohorts of APPSwe YAC mice up to 21 months of age. Together, our

Table 1 Between-study comparisons of variables potentially affecting behavioural outcomes for transgenic mice in the Morris water maze task Reference

AD mouse

Background strain

6

Tg2576

B6SJLF1

7

Tg2576

14 15 16 17 19 20

Gender

Ages tested (months)

Outcomes

Not specified

4–5, 6–11, 12–18, 20–25

C57BL/6

Male and female

3, 9, 14, 19

APPSwe/PS1dE9

Not specified

Not specified

3, 5, 8

APPSwe/PS1dE9 APPSwe/PS1dE9 APPSwe/PS1dE9 Tg2576 APPSwe YAC

C57BL6/JxC3H B6C3F1/J Not specified C57BL6 SJL F1 C57BL/6Jx129/SvJ x129Sv- +P +Tyr–c +Kitl-SlJ

Male Female Male and female Male Not specified

7 12 9 3, 5, 6, 10 19

Deficits in acquisition (12–18 months) and retention (6–11 months) Combined no deficits; 19 month female acquisition deficits Deficits in acquisition and retention at 8 months Deficits in acquisition and retention No deficits Deficits in reversal training for females Deficits in retention at 10 months Deficits in long-term retention

Table 2 Between-study comparisons of methodological variables for the Morris water maze task potentially affecting behavioural outcomes Reference

Pool/platform diameter

Visible platform

Trials/day

Time on platform

6 7 14 15 16 17 19 20

1 m/12 cm 1 m/9 cm 1.8 m/8 cm 1 m/n.s. 1.2 m/12 cm 1.2 m/15 cm 1.2 m/10 cm 1.2 m/11 cm

3 days 4 days Not performed Pretraining only 2 days Not performed 4 days Pretraining only

8 4

30 s 30 s 5s 20 s 30 s 20 s 30 s 30 s

6 2

Water temperature

Duration of acquisition

25–27°C n.s. 20–24°C 24–25°C 24–27°C 22–23°C 22–26°C n.s.

9 days 10 days 6 days 4 days 6 days 5 days 4 days 8 days

n.s., not specified.

© 2014 Wiley Publishing Asia Pty Ltd

Trials/day Probe trials

4 4 4 4 4 4 2 4

Days 4, 7, 10 Day 11 Day 7 Day 5 Days 2 & 6 Not performed Day 4 Day 9, 18, 26

Maximum trial length/ inter-trial interval 60 60 60 60 60 60 80 60

s/20 min s/n.s. s/15 min s/20–30 min s/10 min s/5 min s/1 h s/12–15 min

SR Edwards et al.

804

findings indicate that there is minimal effect of the APP transgene on spatial memory acquisition and retention in hemizygous APPSwe YAC mice up to 21 months of age. Consistent with these behavioural findings, congophilic Ab deposits were not present in brain sections from APPSwe YAC mice up to 24 months of age. In comparison, studies by Hock et al.20 using homozygous ‘R1.40’ APPSwe YAC mice found no deficits in spatial memory acquisition in the MWM task at 19 months of age, although Ab deposits were detected by immunoperoxidase staining in brain sections at this age. However, there were impairments in longterm retention in probe trials performed 19 days after completion of HP training.20 In the present study, the APPSwe YAC mice were hemizygous and on a C57BL6/J background, whereas those used by Hock et al.20 were homozygous and on a C57BL/ 6J 9 129/SvJ 9 129Sv- +P +Tyr–c +Kitl–S1J background. Genetic background has been shown to regulate APP processing and Ab deposition in ‘R1.40’ APPSwe YAC mice.9 Hence, there is a need for studies by independent groups of investigators to evaluate the reproducibility of behavioural changes in this transgenic mouse model of AD. Between-laboratory differences in the specifics of how the MWM task is implemented, together with interactions with factors such as age, gender and background strain of the transgenic mice, likely contribute to between-study differences in MWM testing outcomes, and these are summarized in Tables 1 and 2. Arguably, the most important methodological variable impacting on study outcomes in the MWM task is water temperature.21 In preliminary studies, a water temperature of 24°C and an inter-trial interval of 13 min were required to prevent hypothermia and reduced swimming speeds in transgenic mice carrying APPSwe and PS1 mutations.21 The present study used an in-built thermostatically controlled heating system to ensure that water temperature remained constant at 25–27°C for the duration of the experimental procedures and an inter-trial interval of 20 min. In conclusion, the findings of the present study show a significant deficit in spatial memory retention in the MWM task for double-transgenic hemizygous ‘line-85’-derived APPSwe/PS1dE9 mice bred on a C57BL6/J background by 3.6 months of age, as well as significant deficits in both spatial memory acquisition and retention at 9.3 and 14.8 months of age. Our findings replicate and extend previous work by others and provide further support that ‘line 850 -derived APPSwe/PS1dE9 transgenic mice develop reproducible age-related cognitive deficits. In addition, it is clear that rigorously designed studies using transgenic mouse models of AD are needed to evaluate the extent to which temporal changes in memory performance can be reproduced by independent groups of investigators. Such replication studies are essential to assess the robustness of AD animal models and hence identify the most suitable model for use in efficacy profiling of novel AD treatments from discovery programs.

METHODS Animals Animal ethics approval was obtained from The University of Queensland’s Animal Ethics Committee. Congenic single-transgenic hemizygous APPSwe YAC ‘B6.129-Tg(APPSw)40Btla/J’

mice developed by Lehman et al.9 and double-transgenic hemizygous APPSwe/PS1dE9 ‘B6.Cg-Tg(APPswe,PSEN1de9)85Dbo/J’ mice developed by Jankowsky et al.,10 each on a C57BL6/J background, were sourced originally from Jackson Laboratories. Mice used in the present study were bred and aged by The University of Queensland’s Biological Resources (UQBR) and the colonies were maintained by mating transgenic-positive male mice with wild-type C57BL6/J female mice. Offspring were genotyped by polymerase chain reaction analysis of genomic DNA taken from blood samples. Mice were housed in groups of two to five depending on litter size in Optimice (Animal Care Systems, Centennial, CO, USA) individually filtered plastic cages and maintained on a 12 h light–dark cycle (lights on at 0700 hours), with food and water provided ad libitum. Study design The MWM testing was conducted on separate cohorts of mice at the following ages: single-transgenic APPSwe YAC mice (n = 205 in total) were tested at 4, 7, 10, 13, 16, 18 and 21 months of age (within
1 week); double-transgenic APPSwe/PS1dE9 mice (n = 54 in total) were tested at mean ages of 3.6 months (range 3.25–4 months), 9.3 months (range 7.25– 11 months) and 14.8 months (range 14–15.75 months). The individual numbers of transgenic and wild-type mice in each age group that underwent testing in the MWM are given in Table S1. Morris water maze procedures The MWM comprised a tank 2 m in diameter divided into four quadrants and filled to a depth of 40 cm with water made opaque by the addition of Akaton Opacifier 621 (Wardel Chemicals, Surrey, UK). The temperature was maintained at 25–27°C using an in-built thermostatically controlled heating system. A Plexiglas escape platform with a diameter of 15 cm was placed in the centre of the training quadrant approximately 1 cm beneath the water surface. Posters were placed on the walls around the tank to act as external navigation cues; the position of these posters remained unchanged for the duration of the investigations. Trials were recorded by a tracking camera positioned above the tank and acquired using videotracking software (EthoVision XT v.4; Noldus Information Technology, Leesburg, VA, USA). Images were captured every 200 ms. The MWM testing procedure was based on the method described by Westerman et al.6 Cohorts of up to 12 mice at a time were tested by the same experimenter for the duration of the testing procedure. The VP training was initially performed over a 4 day period. A 10 cm high black tube was fixed to the centre of the escape platform to mark its position. On each day of VP testing, the platform was placed in a different quadrant, which varied pseudorandomly, to ensure that the testing procedure was independent of external navigation cues. The same sequence was used for each cohort of mice. Mice were released into the pool from the diagonally opposite side, with their head located toward the side of the tank. Mice underwent eight trials per day, separated by 20 min intervals, and were left on the platform for 30 s before being dried with tissues and returned to their home cages. The maximum duration of each trial was 1 min. Mice failing to swim

© 2014 Wiley Publishing Asia Pty Ltd

APPSwe/PS1dE9 mice for efficacy testing to the platform were guided to the platform using an orange plastic scoop handle. Any mice failing to swim to the platform in more than one trial on Day 4 of VP training were deemed to be ‘performance incompetent’ and excluded from the study. Hidden platform training was subsequently performed over the next 10 days. The platform remained in a fixed position in the training quadrant and mice were released into the pool from one of four points located at the boundaries of each quadrant. Although the release point varied pseudorandomly on each successive day, the same release sequence was used for each cohort of mice that underwent testing. Mice underwent four trials per day, separated by 20 min intervals, with a maximum trial duration of 1 min and were left to remain on the platform for 30 s before return to their home cages. Mice failing to locate the escape platform were guided to the platform using the orange plastic scoop handle. Probe trials were conducted on Days 4, 7 and 10 of HP training and were performed prior to recommencement of HP training. The platform was removed and mice were released into the tank from the side diagonally opposite to where the platform was located and were allowed to swim freely for 1 min. EthoVision XT v.4 software (Noldus Information Technology) was used to determine the escape latency, cumulative distance to the escape platform and mean velocity for each VP and HP training trial. For probe trials, EthoVision XT v.4 software (Noldus Information Technology) was used to determine the time in each quadrant and average proximity to the platform location.

805

Data analysis Statistical comparisons were performed using GraphPad Prism (v6.00) software (GraphPad Software, La Jolla, CA, USA). For each mouse, the mean values of escape latencies for visible and hidden platform trials on each day of training in the MWM were determined and subsequently analysed by two-way (day and genotype) repeated-measures ANOVA, followed by post hoc pairwise comparisons using Fisher’s least significant difference (LSD) test. Data for the percentage of time spent in the training quadrant (% Time) and average proximity to the platform location during probe trials on Days 4, 7 and 10 were analysed using the same procedure. Comparison of pooled values for manual counts of Ab deposits in the cortex and hippocampus from APPSwe/PS1dE9 mice aged 10.5 and 12 months was performed using Student’s unpaired t-test. The criterion for statistical significance was P < 0.05 (two-sided).

ACKNOWLEDGEMENTS This work was funded by a National–International Research Alliance Program grant from the Queensland State Government and used infrastructure purchased using investment funds from the Queensland Government Smart State Research Facilities Fund (SSRFF). The authors thank Yuen Dieu, Helen Gooch and Mei-Fong Ho for expert technical assistance.

DISCLOSURE The authors report no conflicts of interest.

Tissue preparation After completion of MWM studies, mice were administered a lethal dose of sodium pentobarbitone (600 mg/kg, i.p.) and were perfused transcardially with approximately 20 mL of 0.9% saline, followed by 100 mL of 4% paraformaldehyde in phosphate-buffered saline (PBS, pH 7.4). The brain was then removed, postfixed for approximately 24 h in the same fixative solution and placed into PBS containing 0.1% sodium azide until sectioning. Prior to sectioning, brains were placed in 30% sucrose solution for 24 h. Coronal sections were cut using a freezing microtome at 40 lm into sets of three or six serially adjacent sections and collected into 24-well plates filled with PBS containing 0.1% sodium azide. Histology Congo red was used to detect Ab deposits. Sections were mounted onto slides coated with gelatine/chrome-alum and allowed to dry overnight, before staining with Congo red using standard techniques. Briefly, sections were passed through ethanol and xylene, then back through ethanol again, before staining with Congo red and counterstaining with cresyl violet. Sections were subsequently passed through ethanol, cleared in xylene and cover-slipped. Manual bilateral counts of cortical and hippocampal Ab deposits stained with Congo red were conducted on brain sections containing the dorsal hippocampus taken from APPSwe/PS1dE9 mice aged 10.5 and 12 months (n = 3 per group). Counts were performed on five to six brain sections at least 240 lm apart across the dorsal hippocampus in each mouse.

REFERENCES 1. Gold TE. Alzheimer disease therapy: Can the amyloid cascade be halted. J. Clin. Invest. 2003; 111: 11–18. 2. Selkoe DJ. Alzheimer’s disease. Genes, proteins, and therapy. Physiol. Rev. 2001; 81: 741–66. 3. Higgins GA, Jacobsen J. Transgenic mouse models of Alzheimer’s disease: Phenotype and application. Behav. Pharmacol. 2003; 14: 419–38. 4. Howlett DR, Richardson JC. The pathology of APP transgenic mice: A model of Alzheimer’s disease. Histol. Histopathol. 2009; 24: 83–100. 5. Kawarabayashi T, Younkin LH, Saido TC, Shoji M, Ashe KH, Younkin SG. Age-dependent changes in brain, CSF, and plasma amyloid b protein in the Tg2576 transgenic mouse model of Alzheimer’s disease. J. Neurosci. 2001; 15: 372–81. 6. Westerman MA, Cooper-Blacketer D, Mariash A et al. The relationship between Ab and memory in the Tg2576 mouse model of Alzheimer’s disease. J. Neurosci. 2002; 22: 1858–67. 7. King DL, Arendash GW. Behavioural characterization of the Tg2576 transgenic model of Alzheimer’s disease through 19 months. Physiol. Behav. 2002; 75: 627–42. 8. Janus C, Westaway D. Transgenic mouse models of Alzheimer’s disease. Physiol. Behav. 2001; 73: 873–86. 9. Lehman EJH, Kulnane LS, Gao Y et al. Genetic background regulates b-amyloid precursor protein processing and b-amyloid deposition in the mouse. Hum. Mol. Genet. 2003; 12: 2949–59. 10. Jankowsky JL, Fadale DJ, Anderson J et al. Mutant presenilins specifically elevate the levels of the 42 residue b-amyloid peptide in vivo: Evidence for augmentation of a 42-specific b secretase. Hum. Mol. Genet. 2004; 13: 159–70. 11. Prinz F, Schlange T, Asadullah K. Believe it or not: How much can we rely on published data on potential drug targets? Nat. Rev. Drug Discov. 2011; 10: 712–13.

© 2014 Wiley Publishing Asia Pty Ltd

806

SR Edwards et al.

12. Begley G, Ellis LM. Raise standards for preclinical cancer research. Nature 2012; 438: 531–3. 13. Garcia-Alloza M, Robbins EM, Zhang-Nunes SX et al. Characterization of amyloid deposition in the APPSwe/PS1dE9 mouse model of Alzheimer disease. Neurobiol. Dis. 2006; 24: 516–24. 14. Chen S, Cai Q, Shen Y et al. Age-related changes in brain metabolites and cognitive function in APP/PS1 transgenic mice. Behav. Brain Res. 2012; 235: 1–6. 15. Huang HJ, Liang KC, Ke HC, Chang YY, Hsieh-Li HM. Longterm social isolation exacerbates the impairment of spatial working memory in APP/PS1 transgenic mice. Brain Res. 2011; 1371: 150–60. 16. Montgomery KS, Simmons RK, Edwards G III et al. Novel agedependent learning deficits in a mouse model of Alzheimer’s disease: Implications for translational research. Neurobiol. Aging 2011; 32: 1273–85. 17. Gallagher JJ, Minogue AM, Lynch MA. Impaired performance of female APP/PS1 mice in the Morris water maze is coupled with

18.

19.

20.

21.

increased Ab accumulation and microglial activation. Neurodegener. Dis. 2013; 11: 33–41. Owen EH, Logue SF, Rasmussen DL, Wehner JM. Assessment of learning by the Morris water task and fear conditioning in inbred mouse strains and F1 hybrids: Implications of genetic background for single gene mutations and quantitative trait loci analyses. Neuroscience 1997; 80: 1087–99. Kishimoto Y, Higashihara E, Fukuta A, Nago A, Karino Y. Early impairment in a water finding test in a longitudinal study of the Tg2576 mouse model of Alzheimer’s disease. Brain Res. 2013; 1491: 117–26. Hock BJ, Lattal KM, Kulnane LS, Abel T, Lamb BT. Pathology associated memory deficits in Swedish mutant genome-based amyloid precursor protein transgenic mice. Curr. Aging Sci. 2009; 2: 205–13. Iivonen H, Nurminen L, Harri M, Tanila H, Puolivali J. Hypothermia in mice tested in Morris water maze. Behav. Brain Res. 2003; 141: 207–13.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Figure S1. Cumulative distance to the escape platform during visible and hidden platform training for APPSwe/PS1dE9 and wild-type mice. Figure S2. Mean velocities during visible and hidden platform training for APPSwe/PS1dE9 and wild-type mice. Figure S3. Escape latencies during visible platform training for APPSwe YAC and wild-type mice. Figure S4. Average proximity to the platform location during probe trials for APPSwe YAC and wild-type mice. Figure S5. Cumulative distance to the escape platform during visible platform training for APPSwe YAC and wild-type mice. Figure S6. Cumulative distance to the escape platform during hidden platform training for APPSwe YAC and wild-type mice. Figure S7. Mean velocities during visible platform training for APPSwe YAC and wild-type mice. Figure S8. Mean velocities during hidden platform training for APPSwe YAC and wild-type mice. Table S1. Numbers of transgenic and wild-type APPSwe YAC and APPSwe/PS1dE9 mice for each age group tested in the MWM.

© 2014 Wiley Publishing Asia Pty Ltd