|Home | About | Journals | Submit | Contact Us | Français|
Transgenic APPSwe/PS1dE9 (APP/PS1) mice that overproduce amyloid beta (Aβ) are extensively used in the studies of pathogenesis and experimental therapeutics and new drug screening for Alzheimer’s disease (AD). However, most of the current literature uses young or adult APP/PS1 mice. In order to provide a broader view of AD-like phenotype of this animal model, in this study, we systematically analyzed behavioral and pathological profiles of 24-month-old male APP/PS1 mice. Aged APP/PS1 mice had reference memory deficits as well as anxiety, hyperactivity, and social interaction impairment. Consistently, there was obvious deposition of amyloid plaques in the dorsal hippocampus with decreased expression of insulin-degrading enzyme, a proteolytic enzyme responsible for degradation of intracellular Aβ. Furthermore, decreases in hippocampal volume, neuronal number and synaptophysin expression, and astrocyte atrophy were also observed in aged APP/PS1 mice. This finding suggests that aged APP/PS1 mice can well replicate cognitive and noncognitive behavioral abnormalities, hippocampal atrophy, and neuronal and astrocyte degeneration in AD patients, to enable more objective and refined preclinical evaluation of therapeutic drugs and strategies for AD treatment.
Alzheimer’s diseases (AD) is the most common neurodegenerative disease among the elderly, characterized by progressive memory loss and declining cognitive function, in addition to a variety of psychiatric and behavioral abnormalities (Holtzman et al. 2011). The pathogenesis of AD remains unclear and may likely be attributed to the combined effects of numerous factors, such as hereditary, environment, lifestyle, metabolic disturbance, and brain trauma (Querfurth and Laferla 2010; Jiang et al. 2013). However, aging is a significant risk factor for the onset of AD (Ballard et al. 2011; Bernick et al. 2012; Langbaum et al. 2013). Because of improvements in quality of life and health care, there has been a dramatic rise in human life expectancy. The total number of individuals and the proportion of aged population with AD are increasing year by year, causing tremendous social, financial, and public health burdens. Therefore, it is crucial to explore the interacting mechanisms between advancing age and AD.
Early-onset AD is caused primarily by mutations in the presenilin (PS)1, PS2, and/or Aβ precursor protein (APP) genes, while late onset AD may be associated with apolipoprotein E alleles (Bertram and Tanzi 2008; Rademakers and Rovelet-Lecrux 2009; Ferencz and Gerritsen 2015). Nevertheless, both types of AD show similar Aβ-related neuropathological changes (Huang and Mucke 2012). Correspondingly, various transgenic rodent models, such as those overexpressing mutant human APP or with both PS1 and tau, have been developed, providing valuable insights into specific aspects of AD pathogenesis (Bales 2012; Braidy et al. 2012; Bilkei-Gorzo 2014; Webster et al. 2014). Among these, APP/PS1 double transgenic mice are an extensively used model, showing progressive age-related development of Aβ accumulation and cognitive deficits (Jankowsky et al. 2001; Trinchese et al. 2004; van Groen et al. 2006; Meyer-Luehmann et al. 2009; Filali et al. 2011; Janus et al. 2015). Furthermore, previous studies have revealed positive results, such as demonstrating various antioxidants (Garcia-Alloza et al. 2010; Varamini et al. 2014; Yanagisawa et al. 2015), neuroprotective agents (Kilgore et al. 2010; Peng et al. 2012; Li et al. 2014), anti-neuroinflammation (Cherry et al. 2015; Guo et al. 2015), immune therapy (Sudduth et al. 2013; Carrera et al. 2013; Zhang et al. 2015), and nonpharmacological interventions (Liu et al. 2011; Jankowsky et al. 2005), which can attenuate or even reserve AD-like pathology in APP/PS1 mice. Unfortunately, none of these potentially useful results have been able to be reproduced in clinical therapies for AD patients. Thus, transgenic mice including APP/PS1 mice as a common tool for the study of AD, demonstrate an extremely low translational value from basic research to the clinical utilization (Zahs and Ashe 2010; Webster et al. 2014).
As an irreversible neurodegenerative disease, AD patients often pass away within 10–20 years of disease onset (Tanzi and Bertram 2008). Contrary to humans, the overall survival rate of transgenic AD mice is similar to that of the wild-type (WT) littermate controls, but the underlying mechanism for this discrepancy remains unclear (Zahs and Ashe 2010; Webster et al. 2014). Furthermore, to save experimental time and costs, most studies select young or adult transgenic mice, leaving out elderly subjects, as research participants (Bales 2012; Braidy et al. 2012; Bilkei-Gorzo 2014). This is inconsistent with the fact that AD is essentially an age-related disease. The largely overwhelmingly positive results obtained when testing novel therapies in young or adult transgenic AD mice have failed to translate into similar positive clinical outcomes. Therefore, to provide theoretical basis for the rational use of transgenic AD models, it is necessary to systematically analyze their behavior and pathological phenotypes in old age. In this study, 24-month-old APP/P1 mice were used to assess AD-like pathology. The results suggested that aged APP/PS1 mice can better model pathophysiological features seen in human AD, so that may increase the success rate for therapeutic drugs and strategies advancing into clinical treatment of AD.
APPSwe/PS1dE9 (APP/PS1) mice (Jankowsky et al. 2001) overexpressing chimeric mouse and/or human-mutated APP with Swedish mutation (APP695Swe), and mutant human presenilin 1 with exon 9 deletion (PSEN1dE9) genes, were used in the study. Postnatal day 28 male mice were weaned and separated from their female littermates and housed (two to four mice per cage) under standard laboratory conditions (temperature- and humidity-controlled conditions with a 12-h light-dark cycle with food and water ad libitum) until 24 months of age. There were no observable differences in mortality rates between APP/PS1 mice and wild-type (WT) mice in this study. A general health check was performed with all mice before each testing to ensure that behavioral findings were not the result of deteriorated physical conditions of the animals.
All experiments were conducted in accordance with international standards on animal welfare and the guidelines of the Institute for Laboratory Animal Research of Nanjing Medical University. All efforts were made to minimize animal suffering and to reduce the number of animals used.
The Y-maze test was performed to measure working memory, as previously described (Wang et al. 2013). In brief, the Y-maze test contains two 5-min stages with an interval of 2 h. During the first stage, one arm, the novel arm, was blocked by a black baffle, and mice were allowed to move freely within the other two arms. During the second stage, the novel arm was opened and mice could freely move throughout three arms. The percentage of time traveled in the novel arm, number of entries into the novel arm, and total distance traveled, as well as traveling speed, were counted.
The open field test was carried out to measure exploration and anxiety-related behaviors as described (Chen et al. 2006). The open field box consisted of four identical chambers (60 × 60 × 25 cm), with an outlined center area (30 × 30 cm). Each animal was placed in the middle of chamber, which served as a starting point, then allowed to move freely for 10 min within the box. The time spent in the center area and number of entries into the center area were measured. The total distance traveled and mean speed of movement during the test were also counted.
The elevated plus maze test was utilized to assess anxiety-related behavior in response to a potentially dangerous environment (Webster et al. 2013). The maze consists of two enclosed arms (50 × 10 × 40 cm), two open arms (50 × 10 cm), and center area (10 × 10 cm) elevated 100 cm above the floor. Each mouse was placed in the center of the maze, and the time spent in the open arms and number of entries into the open arms were analyzed.
Social interaction behavior was evaluated in an open field apparatus made of clear Plexiglas (40 × 40 × 30 cm). The procedure for the social interaction test included three phases: habituation, sociability test, and social memory test, as previously described (Lo et al. 2016). During the habituation phase, each of the two side chambers contained an inverted empty wire cup. Each mouse was given 10 min to explore the environment freely. The mice were habituated for 2 days. During the sociability phase, an unfamiliar mouse (stranger 1) was enclosed in one of the wire cup in a side chamber. The location of the unfamiliar mouse alternated between the two side chambers across test. During the social memory phase, a new unfamiliar mouse (stranger 2) was enclosed in the wire cup that had been empty during the sociability phase. Exploring novelty was defined as when a test mouse climbed on the cup or oriented toward the cup with the distance between the nose and the cup less than 1 cm. The time spent in each chamber, and exploration of novel mice or empty cups, were recorded for the first 4min of each session. All stranger mice were familiarized to being enclosed in inverted wire cups for 15min daily on two consecutive days prior to the experiment.
Mouse activity in the above behavioral apparatuses was collected by a digital video camera connected with a computer-controlled system (Beijing Sunny Instruments Co. Ltd., China). All tests were performed by two independent experimenters who were blind to the treatment schedule.
Following deep anesthesia, mice were transcardially perfused with PBS followed by 4 % buffered formalin. Brains were dissected in the mid-sagittal plane, postfixed overnight at 4 °C, and embedded in paraffin following standard procedures. Both paraffin embedded half-brains were serially cut into sagittal sections at 5 μm and divided into five sets.
Immunohistochemical staining was performed as previously described (Xu et al. 2013). After deparaffinization and rehydration, sections were incubated overnight at 4 °C with one of primary antibodies listed in Table Table11 and incubated with secondary antibody for 1.5 h at room temperature. The reaction was visualized with DAB (Sigma-Aldrich).
The sets (every 25th sections, about 12–16 sections per mouse) of dorsal hippocampus stained with Cresyl violet, Thioflavin-S, or antibodies listed in Table Table11 were photographed at ×100 magnification using a digital microscope (Leica Microsystems). Individual images were exported to Image-Pro Plus 6.0 Analysis System (Media Cybernetics Inc., San Francisco, CA, USA). The estimates of hippocampal volume were performed on serial Nissl-stained sections. The boundary of hippocampus and gray region that consists of pyramidal cell layer and granular cell layer was manually delineated. The total hippocampal area and gray matter area per section was per section were measured. Total hippocampal or gray matter volume per mouse was also calculated according to the principles of Cavalieri basic (Pereda-Pérez et al. 2013). The hippocampal white matter volume was obtained by the total hippocampal volume subtracting the hippocampal gray matter volume. The total neuron number (T) in the pyramidal layer of CA1 and granular cell layer of the dentate gyrus was quantified on NeuN-stained sections using the formula: T = (N × V)/t. Where N, V, and t are the cell density, volume of the structure, and thickness of the section, respectively (Beauquis et al. 2014).
For quantitative analysis of Aβ plaque load, thioflavin-S, or 6E10-positive signals were determined by a standardized region of interest grayscale threshold analysis (Kim et al. 2009). The percentage of area occupied by thioflavin-S or 6E10 in the whole hippocampus, gray matter area, and white matter area was measured. In addition, expression levels of disintegrin and metalloproteinase 10 (ADAM10), β-site amyloid precursor protein-cleaving enzyme 1 (BACE1), PS1, neprilysin (NEP), insulin degrading enzyme (IDE), and synaptophysin (SYP) in the hippocampus or gray/white subareas was assessed by measurement of mean integrated optical density (MIOD = IOD/total area), respectively (Xu et al. 2013).
Glial fibrillary acidic protein (GFAP) or ionized calcium-binding adaptor molecule 1 (Iba-1) immunostained in the stratum radiatum of CA1 was semi-quantified as previously described (Calvo-Ochoa et al. 2014). Briefly, only cells with clear cell bodies and complete processes were counted and expressed as the number of cells per square millimeter. The mean area of each GFAP-positive astrocytes and Iba-positive microglia was also measured. All quantification was done blind to animal genotype.
Hippocampal tissues were homogenized and centrifuged at 4 °C and 12,000 rpm for 15 min. The samples were resolved on SDS-PAGE, transferred onto PVDF membranes using a Bio-Rad miniprotein-III wet transfer unit, then blocked with 5 % skim milk dissolved in TBST (pH 7.5, 10 mM Tris–HCl, 150 mM NaCl, and 0.1 % Tween 20) at room temperature for 1 h. Membranes were probed at 4 °C overnight with one of the primary antibodies listed in Table Table1.1. Horseradish peroxidase-conjugated secondary antibodies (Vector Laboratories, Burlingame, CA, USA) were used, and bands were visualized using ECL plus detection system. β-tubulin was used as an internal control for protein loading and transfer efficiency.
All results were expressed as means ± SEM. The data were analyzed by two-tailed Student’s t test. A value of P < 0.05 was considered statistically significant.
APP/PS1 mice showed working memory impairment in the Y-maze test, as revealed by decreased time spent in the novel arm (P = 0.042; Fig. Fig.1a)1a) and the number of entries into the novel arm (P = 0.005; Fig. Fig.1b).1b). In addition, neither movement distance nor speed was significantly different between APP/PS1 mice and WT mice (both P > 0.05; Fig. Fig.1c,1c, d).
APP/PS1 mice displayed increased levels of general anxiety in the open field test, as shown by decreased time spent (P = 0.005; Fig. Fig.2a)2a) in the center area, as well as the number of crosses into the center area (P = 0.028; Fig. Fig.2b).2b). APP/PS1 mice also exhibited mild hyperactivity with a tendency to increase movement distance and speed, compared with WT controls (both P > 0.05; Fig. Fig.2c,2c, d).
The elevated plus maze test confirmed anxiety-like behaviors in APP/PS1 mice. They exhibited decreases in total movement time (P = 0.017; Fig. Fig.3a)3a) and distance (P = 0.005; Fig. Fig.3b)3b) in the open arm and number of entries into the open arm (P = 0.028; Fig. Fig.3c).3c). APP/PS1 mice showed a slightly longer movement distance and high movement speed compared with controls (both P > 0.05) but with high individual variability in the both genotypes (Fig. (Fig.33d).
Social preference tests results show that WT mice preferred to enter into the compartment housing a stranger mouse (P = 0.047, Fig. Fig.4a),4a), with a preference for exploring the novel mouse over the empty cup (P = 0.003; Fig. Fig.4b).4b). Alternatively, APP/PS1 mice were remained in the empty compartment (P = 0.001; Fig. Fig.4a)4a) and displayed no exploring preference between the novel mouse (stranger 1) and empty cup (P = 0.501; Fig. Fig.4b).4b). Social memory test results show that WT mice preferred to enter the compartment housing a new stranger mouse (stranger 2) compared with that housing stranger 1 (P = 0.034; Fig. Fig.4c).4c). By contrast, APP/PS1 mice did not show clear preference for staying in the chamber containing stranger 2 (P = 0.398; Fig. Fig.4c).4c). In addition, both groups showed preference for exploring stranger 2 over stranger 1 (P = 0.004 for WT mice; P = 0.034 for APP/PS1 mice), but contact time with stranger 2 was much longer in WT mice than APP/PS1 mice (P = 0.013; Fig. Fig.44d).
Hippocampal atrophy and marked neuronal loss are the pathological basis for cognitive dysfunction and psychological symptoms of AD patients (Caselli and Reiman 2013; Schröder and Pantel 2016) but do not occur in young/adult APP/PS1 mice (Bilkei-Gorzo 2014; Webster et al. 2014). Therefore, we investigated whether these pathological hallmarks were able to be observed in aged APP/PS1 mice. Quantitative analysis of HE-stained hippocampal serial sections (Fig. (Fig.5a)5a) revealed that aged APP/PS1 mice had a smaller dorsal hippocampus than their littermate controls, with significant reduction in section numbers (P = 0.032; Fig. Fig.5b)5b) and hippocampal volume (P = 0.004; Fig. Fig.5c).5c). Notably, white matter volume decreased more than that of the gray matter, the location of cellular bodies of pyramidal and granular neurons (P = 0.005; Fig. Fig.55d).
Immunohistochemistry for neuronal marker NeuN followed by unbiased stereology analysis was performed to estimate the number of hippocampal neurons. The number of NeuN-positive hippocampal neurons in the CA1 and dentate gyrus of APP/PS1 mice was reduced to 86.6 and 79.6 %, respectively, compared with WT mice (P = 0.041, P = 0.033, respectively; Fig. Fig.5e,5e, f). Furthermore, both semi-quantitative immunohistochemistry (Fig. (Fig.5g,5g, h) and Western blot analysis (Fig. (Fig.5i,5i, j) demonstrated that hippocampal expression levels of synaptic marker SYP were reduced to 35 and 50 %, respectively (both P < 0.001). This indicates that synaptic loss was more severe than neuronal loss in the hippocampus of this AD model.
Both 6E10 immunostaining (Fig. (Fig.6a)6a) and thioflavin-S fluorescent staining (Fig. (Fig.6b)6b) displayed a large number of Aβ plaques, predominately large-sized plaques, were present in the hippocampus and cerebral cortex of aged APP/PS1 mice. Furthermore, consistent with more severe atrophy of the hippocampal white matter, quantitative data revealed that the percentage area occupied by thioflavin-S or 6E10-labeled plaques was higher in the white matter than the gray matter (both P < 0.001; Fig. Fig.66c).
It is known that brain Aβ accumulation is a consequence of the imbalance between Aβ generation and Aβ clearance (Tanzi and Bertram 2008). Aβ is cleaved from its precursor protein APP by β-secretase and γ-secretase, but it is most precluded by α-secretase to generate neuroprotective and soluble APPα fragments (sAPPα) (Morishima-Kawashima 2014). Neprilysin (NEP) and insulin-degrading enzyme (IDE) are responsible for proteolytic degradation of Aβ peptides from brain parenchyma (Tanzi et al. 2004; Tarasoff-Conway et al. 2015). Therefore, we determined whether different Aβ burden between the gray matter and white matter was associated with different expression patterns of these Aβ metabolism-related enzymes. Within the cell bodies of both pyramidal and granular neurons of aged APP/PS1 mice, there was high expression of α-secretase, IDE and NEP, but low expression of β-secretase and γ-secretase. In contrast, there was low expression of α-secretase, IDE and NEP, but high expression of β-secretase and γ-secretase in the white matter (Fig. (Fig.6d).6d). The semi-quantitative analysis confirmed that the gray matter and white matter were differences in expression levels of α-secretase (P < 0.001), β-secretase (P = 0.003), γ-secretase (P < 0.001), IDE (P < 0.001), and NEP (P < 0.001) (Fig. (Fig.6e).6e). Western blot revealed that the expression levels of α-, β-, and γ-secretases and NEP were similar between aged APP/PS1 mice and WT mice (all P > 0.05), but IDE levels significantly decreased (P = 0.036) in the hippocampal neurons of aged APP/PS1 mice (Fig. (Fig.6f,6f, g).
Expression and distribution of astrocytes in the hippocampus was observed by immunohistochemical staining for GFAP. In the hippocampus of aged APP/PS1 mice, there were obvious GFAP-positive signals around Aβ plaques, but astrocytes lacked normal morphology. Astrocytes also underwent apparent atrophy in the brain parenchyma distant to the Aβ plaques. By contrast, a large number of GFAP-positive astrocytes were present throughout the hippocampus of WT controls, with most being of the activated phenotype characterized by hypertrophic cellular bodies and processes (Fig. (Fig.7a).7a). Astrocyte loss and atrophy in the hippocampus of APP/PS1 mice was confirmed by quantitative analyses revealing marked decreases in GFAP-positive cell number (P = 0.037; Fig. Fig.7b)7b) and cell area (P = 0.019; Fig. Fig.7c).7c). Therefore, similar to neuronal elements, astrocytes also underwent degeneration during the later stages of Aβ pathology in APP/PS1 mouse model. In addition, activated Iba-1-positive microglia were observed in the hippocampus of both genotypes, but many amoeba-like microglia were present around Aβ plaques (Fig. (Fig.7a).7a). Quantitative analysis revealed a high number of microglia in the hippocampus of APP/PS1 mice, suggesting the occurrence of long-term chronic inflammation (P = 0.042; Fig. Fig.77c).
Brain-derived neurotrophic factor (BDNF) is a central regulator of hippocampal neuron survival and synaptic plasticity (Lipsky and Marini 2007). Levels of BDNF have been reported to be decreased in the preclinical stages of AD (Peng et al. 2005). Recent studies revealed that BDNF is associated with age-related decline in hippocampal volume (Erickson et al. 2010) and prevents hippocampal atrophy in APP/PS1 mice (Rantamäki et al. 2013). Therefore, we determined whether BDNF was downregulated in the hippocampus of aged APP/PS1 mice. As expected, both semi-quantitative immunohistochemistry and Western blot demonstrated that aged APP/PS1 mice had lower levels of BDNF in the hippocampus than WT mice (both P < 0.001, Fig. Fig.88a–d).
Aging is the greatest risk factor for AD, occurring primarily in the 5th quintile of life. APP/PS1 transgenic mice expressing mutant human APP and PS1 genes are a useful research tool to study Aβ-related pathogenesis. There are several studies characterizing cognitive as well as noncognitive abnormality in APP/PS1 mice at 15–24 months of age (Morgan et al. 2000; Arendash et al. 2001; Sadowski et al. 2004; Wilcock et al. 2004; Hooijmans et al. 2007; Minkeviciene et al. 2008; Pistell et al. 2008; Mei et al. 2010; Ke et al. 2011; Gengler et al. 2012; Manczak and Reddy 2012; Wang et al. 2012; Duffy and Hölscher 2013; Do et al. 2014; Huang et al. 2015; Janus et al. 2015; Sahlholm et al. 2015; Yousefi et al. 2015) (A detailed summary is available in Table Table2).2). However, the majority of experiments utilize young or adult APP/PS1 mice, which does not replicate typical hallmarks of AD, such as severe hippocampal atrophy and extensive neuronal loss (Bales 2012; Bilkei-Gorzo 2014). This shortcoming is then seen when positive therapeutic outcomes from APP/PS1 mice fail in clinical treatment on patients with AD (Bales 2012; Bilkei-Gorzo 2014). In order to provide a broader view of AD-like phenotype of this animal model, in this study, we systematically assessed behavioral and pathological changes of 24-month-old APP/PS1 mice. The results indicate that aged APP/PS1 mice could enable more objective and refined preclinical evaluation of therapeutic drugs and strategies for AD treatment.
The progressive impairment of cognitive function is one typical clinical manifestation of AD. Previous studies report that APP/PS1 mice show reference memory impairments by 3 months of age and persist throughout life (Reiserer et al. 2007; Bernardo et al. 2009; O’Leary and Brown 2009). By comparing these results with our previous Y-maze test for behavioral analysis of 11–12-month-old APP/PS1 mice (Xu et al. 2013, 2015), we found that 24-month-old APP/PS1 cohorts show more severe working and reference memory impairment. This suggests that the 24-month-old APP/PS1 mouse model can imitate age and disease course-related cognitive decline of AD. In addition, we found that there is considerable individual variability in cognitive performance among aged APP/PS1 mice as well as normal aged mice. Further studies are necessary to reveal the underlying mechanism for individual heterogeneity of age-associated cognitive decline, which might be beneficial to find key targets for prevention and treatment of brain aging and AD.
Apart from progressive cognitive dysfunction, AD patients often display anxiety, apathy, emotional changes, and varying degrees of social interaction disability, but there is no clear correlation between noncognitive abnormalities and cognitive impairment (Deutsch and Rovner 1991; Apostolova et al. 2014). It is known that anxiety-related behavior often manifests prior to cognitive impairment in the longitudinal course of AD (Ramakers et al. 2013), a characteristic that also occurs in APP/PS1 mice. Previous studies reported that 3-month-old APP/PS1 mice exhibit anxiety-like phenotype, although without spatial cognitive dysfunction, when compared with age-matched WT controls (Trinchese et al. 2004; Webster et al. 2014). The open field and elevated plus maze test confirmed that anxiety-like behaviors continue to exist in the late stages of this mouse AD model, similar to the Tg2576 (Gil-Bea et al. 2007; Lassalle et al. 2008), J20 (Harris et al. 2010; Cissé et al. 2011), and 5xFAD mice (Wirths et al. 2010; Shukla et al. 2013), all of which also exhibit anxiety-related behavior deficits in the elevated plus maze task. However, there is no anxiety-related behavior abnormality in the APP/PS1 double knock-in mice from 7 to 24 months old (Webster et al. 2013); while 3xFAD mice show decreased anxiety-related behavior in the open field and elevated plus maze test (Nelson et al. 2007; Filali et al. 2012). The variability of anxiety-related behavior in these commonly used AD mouse models supports a heterogeneous range of noncognitive symptoms caused by Aβ-related pathogenesis.
Furthermore, compared with younger cohorts, aged APP/PS1 mice display more advanced social behavioral abnormalities. Previous studies reported that 6-month-old APP/PS1 mice experience mild decline in social interaction function, exhibiting no preference to stay in the empty chamber versus the one having unfamiliar mouse (Zhang et al. 2013; Hsiao et al. 2014). The present study revealed that 24-month-old APP/PS1 mice spent more time in the empty chamber than with the unfamiliar partner, indicating the occurrence of social interaction impairment. Aged APP/PS1 mice also show impairment of social interaction memory, revealed by no preference to stay with the novel mouse versus the familiar one. In addition, similar to later stages of AD patients, aged APP/PS1 show mild hyperactivity, reflected by increased movement speed in the elevated plus maze test and open field. This is not observed in 6–18-month-old APP/PS1 mice (Lalonde et al. 2004; Cao et al. 2007; Sood et al. 2007). In addition to evident cognitive impairment, behavioral results show that 24-month-old APP/PS1 mice emulate other psychiatric and behavioral symptoms including anxiety, social behavior withdrawal, and hyperactivity.
During AD the hippocampus is one of the first regions of the brain to suffer damage. The mechanism of hippocampal atrophy in AD patients is not entirely clear and may be due to neuronal loss, as well as neurite and synaptic degeneration caused by Aβ, highly phosphorylated tau protein, and neurofibrillary tangles, together with other damage factors such as hypotension, diabetes, hyperlipidemia, epilepsy, stress, and affective disorders (Dhikav and Anand 2012; Roberts et al. 2014; Elcombe et al. 2015). Accumulated evidence shows that there is a close correlation between hippocampal atrophy and cognitive impairment in patients with AD (Yavuz et al. 2007; Dhikav and Anand 2011). Furthermore, hippocampal atrophy and dysfunction have also occurred in mild cognitive impairment (MCI) patients (van de Pol et al. 2007). Hippocampal volume reduction rate is an effective tool to help predict whether the MCI will develop into AD (Perrin et al. 2009; Nasrallah and Wolk 2014).
Conversely, various 6–12-month-old AD mouse models, including PDAPP, TG2576, APP23, APP/PS1, 5 × FAD, and 3 × Tg-AD do not show obvious hippocampal atrophy and neuronal loss (Bales 2012; Braidy et al. 2012; Bilkei-Gorzo 2014; Webster et al. 2014). Based on stereological analysis, we found that dorsal hippocampal volume and NeuN-positive neurons in 24-month-old APP/PS1 mice are reduced by 24.5 and 19.4 %, respectively (Fig. (Fig.5).5). These pathological changes suggest that aged APP/PS1 mice display hippocampal atrophy and neuronal loss, although this characteristic pathological change occurs in the early stages of AD patients (Schröder and Pantel 2016). Further research is needed to identify a precise beginning to the hippocampal atrophy and massive loss of neurons in APP/PS1 model, which will be helpful for objective preclinical evaluation of neuroprotective drugs for AD.
Excessive accumulation of neurotoxic Aβ peptides and subsequent formation of senile plaques in the brain parenchyma is the fundamental basis for the gradual, progressive development of AD pathology (Mawuenyega et al. 2010; Holtzman et al. 2011). An earlier study reported that diffuse amyloid plaques begin to appear in the hippocampus and cerebral cortex of APP/PS1 mice at 3 months old (Trinchese et al. 2004). Subsequent studies have revealed an age-dependent pattern of Aβ burden in this AD model (Samaroo et al. 2012; Janus et al. 2015). By use of thioflavin-S and 6E10 immunostaining, we found that 24-month-old APP/PS1 mice versus their 20-month counterparts have a higher percentage of area occupied by thioflavin-S-labeled core plaques (7.24 vs. 3.15 %) or 6E10-labeled diffusing plaques (3.02 vs. 3.65 %) in the hippocampus (Huang et al. 2015). These results suggest that even during 20 to 24 months of age, senile plaques in APP/PS1 mouse brain are still in a dynamic change state.
Furthermore, as mentioned above, APP cleavage is controlled by both α-secretase-mediated nonamyloidogenic pathway and β- and γ-secretase-mediated amyloidogenic pathway (Morishima-Kawashima 2014). Previous studies revealed that young APP/PS1 mice have high levels of amyloidogenic enzyme PS1 (Morishima-Kawashima 2014). In contrast, we demonstrated that aged APP/PS1 mice have no changes in Aβ enzymes including ADAM10, BACE1, and PS1 in the hippocampus, compared with their littermates.
We hypothesized that the relative reduction of γ-secretase in the hippocampus of aged APP/PS1 mice could reduce the production of Aβ. This presumption is supported by the presence of only mature fibrillary plaques, without new inflammatory plaques, in the hippocampus of 24-month-old APP/PS1 mice. Similar to adult APP/PS1 mice, the deposition of Aβ plaques was mainly present in the molecular layer and stratum lacunosum moleculare, while being virtually absent in the pyramidal cell layer and the granule cell layer (Bas Orth et al. 2005; Oksman et al. 2006; Donkin et al. 2010; Scott et al. 2012; Xu et al. 2012, 2016). Furthermore, our previous and present results suggest that laminar heterogeneity and Aβ load is correlated with laminar variability in expression of APP secretases (Xu et al. 2016). As shown in Fig. Fig.6,6, cellular bodies of pyramidal neurons express high levels of α-secretase, but low levels of β-secretase and γ-secretase. In contrast, dendrites of pyramidal cells located in the oriens, radiatum, and lacunosum moleculare layers display high expression of β-secretase and low expression of α-secretase. In addition, it has been shown that over-expression of human sAPPα ameliorates Aβ pathogenesis in APP/PS1 mice. Correspondingly, immunoneutralization of sAPPα by antibody causes hippocampal neuronal apoptosis in this model (Obregon et al. 2012). These evidences, taken together, highly suggest that elevated α-secretase levels in neuronal cell bodies, give them the ability to produces neuroprotective sAPPα and maximally reduce neurotoxic Aβ production, in turn protects hippocampal neurons in the relatively long-term survival in AD-like pathology.
We also determined that human hippocampal neuronal cell bodies expressed high levels of β-secretase and γ-secretase but practically did not express α-secretase. This could potentially increase Aβ production, causing progressive self-death during the AD process (Xu et al. 2016). Different expression patterns of APP secretase in human and mouse neurons may contribute to a significant prolonging of the relative survival time of APP/PS1 mice compared with patients with AD. However, the intrinsic mechanism for species differences in neuronal susceptibility to Aβ pathogenesis needs further investigation.
It is known that brain Aβ homeostasis depends on the balance between the generation and clearance of Aβ (Tanzi and Bertram 2005). Aged APP/PS1 mice show decreases in IDE levels, without changes in Aβ-generating enzymes including ADAM10, BACE1, and PS1 in the hippocampus, compared with their WT littermate controls. This result indicates that there is decreased Aß clearance, without increased Aß production, in the late stage of this AD model. This phenotype is consistent with previous human studies that Aβ clearance is impaired in both early-onset and late-onset forms of AD (Mawuenyega et al. 2010; Potter et al. 2013). Based on this, it seems that aged APP/PS1 mice may be available for screening potential anti-AD agents targeting at Aβ clearance.
NEP and IDE are two key metabolic enzymes used in brain Aβ degradation, and each plays an important role in maintaining the homeostasis of brain Aβ. The primary effect of IDE is degradation of Aβ polymers and fibers (Eckman and Eckman 2005; Qiu and Folstein 2006), while NEP is used predominantly in degradation of Aβ monomers and oligomers (El-Amouri et al. 2008). Our results show that IDE content in the hippocampus of APP/PS1 mice clearly decreases, whereas NEP levels do not differ from those in WT controls. This result is also consistent with the feature of hippocampal Aβ load, wherein the percentage of area occupied by thioflavin-S-positive fibrillary plaques is higher than 6E10-immunopositive diffuse plaques. Unlike aged APP/PS mice, previous studies have reported that the protein levels and activities of NEP are significantly decreased in the brain of patients with AD, while IDE is not significantly altered (Yasojima et al. 2001; Caccamo et al. 2012; Wang et al. 2005). Therefore, for AD patients, impaired Aß clearance could be mainly due to decreased degradation of Aβ monomers and oligomers, whereas, for aged APP/PS1 mice, it is primarily attributed to reduced degradation of Aβ polymers and fibers. It is known that Aβ monomers and oligomers are the main components of neurotoxic substances in the brain, and the toxicity of amyloid plaques is relatively weak (Schnabel 2011). Therefore, we speculate that decreased IDE, but not NEP impairment, may be another key factor for the increased survival time of APP/PS1 mice when compared with patients with AD.
Astrocytes are responsible for maintaining the homeostasis of brain and neural network, by playing a variety of essential functions, including maintaining cerebral water, ions, pH, and neurotransmitter homeostasis, modulating cerebral microcirculation and synaptic transmission and producing neurotrophic factors and antioxidants (Allen and Barres 2009; Baxter 2012). Furthermore, recent studies, including from our laboratory, have demonstrated that astrocytes, via specific expression of the water channel aquaporin 4, are responsible for perivascular clearance of Aβ and other pathological proteins from the brain (Iliff et al. 2012; Xie et al. 2013; Arbel-Ornath et al. 2013; Xu et al. 2015). Astrocytes make rapid responses to numerous varieties of damage and pathological changes through a process referred to as reactive astrogliosis. Evidence of astrogliosis has been documented in early stages of AD (Schipper et al. 2006; Owen et al. 2009; Carter et al. 2012), whereas astrocyte death or atrophy in AD brains from patients and animal models has also been reported (Smale et al. 1995; Kobayashi et al. 2002, 2004). Extensive gliosis is present in the hippocampus of adult APP/PS1 mice (Xu et al. 2013, 2015). Whereas in the 24-month-old APP/PS1 mice, a large number of astrocytes undergo atrophy throughout the hippocampus and even death surrounding Aβ plaques. There is no doubt that degeneration of astrocytes affects their normal functions, including BDNF production, which subsequently exacerbates AD-like pathology in aged APP/PS1 mice. In addition, a large number of microglia were excessively activated, exhibiting amoeba-like shape in aged APP/PS1 mice. Although phagocytic microglia can facilitate Aβ degradation and clearance, protecting adjacent neuronal elements, they may also produce a variety of inflammatory factors, causing neurotoxic effects (Derecki et al. 2014; Mhatre et al. 2015). Therefore, functional impairment of glial cells and microglia may act as a key player in the late stages of AD and protect their functions could serve as an effective strategy against moderate-to-severe AD (Osborn et al. 2016).
BDNF is important in neuronal growth and survival and synaptic processes of memory (Lipsky and Marini 2007). Evidence has suggested that dysfunction of BDNF is a possible contributor to the pathology and symptoms of AD (Tapia-Arancibia et al. 2008). BDNF is found to be decreased in APP/PS1 mice even in the early stages of AD-like pathology (Peng et al. 2009; Hsiao et al. 2014). Apart from BDNF, other neurotrophic factors, such as nerve growth factor (NGF) and glial cell-derived neurotrophic factor (GDNF), are also involved in normal aging and AD (Allen et al. 2013; Budni et al. 2015). It is a limitation of this study that we have only investigated BDNF levels in aged APP/PS1 mice.
In conclusion, via systematic behavioral and pathological analysis, our results have revealed that 24-month-old APP/PS1 mice can emulate various clinical features of AD patients, such as declines in the spatial memory and exploratory ability, and neuropsychiatric symptoms including anxiety and social behavior dysfunction. The mice also show Aβ plaque deposition and other similar pathological characteristics seen in AD patients, such as hippocampal atrophy, neuronal loss and severe glial cell atrophy, none of which occurs in adult APP/PS1 mice. In addition, when compared with AD patients, aged APP/PS1 mice have unique characteristics of AD-like pathology, including a considerable number of survival hippocampal neurons with high expressions of α-secretase and NEP. Nonetheless, it should be pointed out that we did not assess the long-term memory impairment of 24-month-old APP/PS1 mice, due to the fact that aged mice may be unable to tolerate high physical activity during the water maze training. To reduce the number of animals used, we also did not perform systematic evaluation of AD-like progress in this model by using APP/PS1 mice at different months of age. In addition, it will be interesting to determine whether APP/PS1 mice have an age-dependent relationship between hippocampal atrophy and Aβ accumulation, and the underlying mechanisms for the species difference between humans and mice in the sensitivity of the central nervous system to Aβ neurotoxicity. These studies will contribute to a more comprehensive understanding of the AD-like phenotype of APP/PS1 mice, which will be beneficial for more a refined preclinical therapeutic evaluation of AD.
This work was supported by the grants from the National Natural Science Foundation of China (30971020 and 30973517) and the Natural Science Foundation of Jiangsu Educational Department (09KJA310003).