Abstract
Alzheimer’s disease is a neurodegenerative disorder of the elderly and advancing age is the major risk factor for Alzheimer’s disease development. Telomere shortening represents one of the molecular causes of ageing that limits the proliferative capacity of cells, including neural stem cells. Studies on telomere lengths in patients with Alzheimer’s disease have revealed contrary results and the functional role of telomere shortening on brain ageing and Alzheimer’s disease is not known. Here, we have investigated the effects of telomere shortening on adult neurogenesis and Alzheimer’s disease progression in mice. The study shows that aged telomerase knockout mice with short telomeres (G3Terc−/−) exhibit reduced dentate gyrus neurogenesis and loss of neurons in hippocampus and frontal cortex, associated with short-term memory deficit in comparison to mice with long telomere reserves (Terc+/+). In contrast, telomere shortening improved the spatial learning ability of ageing APP23 transgenic mice, a mouse model for Alzheimer’s disease. Telomere shortening was also associated with an activation of microglia in ageing amyloid-free brain. However, in APP23 transgenic mice, telomere shortening reduced both amyloid plaque pathology and reactive microgliosis. Together, these results provide the first experimental evidence that telomere shortening, despite impairing adult neurogenesis and maintenance of post-mitotic neurons, can slow down the progression of amyloid plaque pathology in Alzheimer’s disease, possibly involving telomere-dependent effects on microglia activation.
Introduction
Alzheimer’s disease is the most common cause of dementia in the elderly, and is characterized by the extracellular deposits of amyloid plaques and intracellular neurofibrillary tangles (Kidd, 1964; Alzheimer et al., 1995). Advancing age is the major risk factor for the development of Alzheimer’s disease (Tyas et al., 2001) and overall prevalence of Alzheimer’s disease is ∼14 times higher after 85 years of age compared to 65–69 years of age. The increase in average life expectancy and an ageing society has greatly contributed to the growing importance of this disease in recent decades, especially in developed countries (Forstl, 1998). Despite the close intimacy between ageing and Alzheimer’s disease, very limited knowledge is currently available on this aspect.
At the molecular level, ageing is associated with the accumulation of nuclear DNA damage (d’Adda di Fagagna et al., 2003; Sedelnikova et al., 2004) and a variety of studies have provided evidence for accumulation of nuclear DNA damage and activation of DNA damage checkpoints (such as apoptosis and senescence) in ageing tissues (for review see Nalapareddy et al., 2008). Accumulation of DNA damage by oxidative stress and mitochondrial dysfunction during ageing has been attributed to contribute to neuronal toxicity and evolution of neurodegenerative disorders such as Alzheimer’s disease (Brasnjevic et al., 2008; de Souza-Pinto et al., 2008). Parallel to this, an association between defective DNA damage repair and neurodegenerative diseases has been observed previously (Rass et al., 2007). Telomere shortening represents a cell intrinsic mechanism for accumulation of DNA damage that may increase the sensitivity of neurons towards oxidative stress and proteotoxicity. Further it has been reported that oxidative stress can increase telomere shortening (von Zglinicki et al., 1995). A number of studies have documented telomere shortening from different organs during ageing (Satoh et al., 1996; Aikata et al., 2000; Yang et al., 2001). However, in agreement with very limited cell division, telomere shortening was not observed in whole human brain samples (Allsopp et al., 1995). Similar studies on patients with Alzheimer’s disease revealed divergent results, while telomere shortening was accelerated in peripheral blood cells (Panossian et al., 2003; Zhang et al., 2003), opposite results were obtained from neurons (Thomas et al., 2008).
Telomeres shorten as a consequence of cell division and limit the replicative potential of cells (Allsopp et al., 1992). In the adult mammalian brain, glia but not neurons can divide, although active neurogenesis takes place in the subventricular zone of lateral ventricles and the subgranular zone of the dentate gyrus (Abrous et al., 2005). Since the hippocampus plays a key role in learning and memory (Kempermann and Gage, 2002; Jessberger et al., 2009) and neuron loss is the major problem in Alzheimer’s disease, dentate gyrus neurogenesis has been considered as a potential therapeutic target for replenishment of neurons in Alzheimer’s disease. According to this paradigm, age-dependent telomere shortening and decrease in adult neurogenesis could represent a predisposing factor for Alzheimer’s disease progression (Ming and Song, 2005). In agreement with this, a decline in adult neurogenesis is observed during ageing that may contribute to the evolution of cognitive defects in elderly (Ming and Song, 2005). Interestingly, neurogenesis in the hippocampus is enhanced in human patients with Alzheimer’s disease (Jin et al., 2004) and mouse models (Chen et al., 2008), which led to the speculation that this could represent a compensatory mechanism to replace the lost neurons and slow down Alzheimer’s disease progression (Mirochnic et al., 2009).
Apart from adult neurogenesis, telomere shortening could also alter the proliferative capacity of microglia and inflammatory signalling in the ageing brain. Studies on fibroblasts have shown that senescence induces an aberrant secretion of proinflammatory cytokines (Rodier et al., 2009) and similar observation was also made from ageing telomere dysfunctional mice (Ju et al., 2007). Microglia are the immunoregulatory cells of the CNS and play a key role in neurodegenerative diseases (Cunningham et al., 2005). Interestingly, ageing itself is associated with increased proinflammatory status and studies indicate a higher proinflammatory activity of microglia in the aged rodent brain (Sierra et al., 2007; Dilger and Johnson, 2008; Sparkman and Johnson, 2008; Venneti et al., 2009; von Bernhardi et al., 2010).
Despite the intimate association of ageing with telomere shortening and Alzheimer’s disease development, its functional relevance on Alzheimer’s disease progression has not been investigated experimentally. Here, we have analysed the influence of telomere shortening on Alzheimer’s disease progression in a mouse model. The study provides evidence that telomere dysfunction delays the progression of Alzheimer’s disease-associated amyloid pathology in ageing mice, despite reducing adult neurogenesis.
Materials and methods
Animals
Telomerase knockout mice carrying a homozygous germ line deletion for the telomerase RNA component (Terc−/−) gene, that leads to complete loss of Terc expression and telomerase activity (Blasco et al., 1997) were kindly provided by Maria Blasco (National Cancer Research Centre, Madrid) and have been maintained in our laboratory for several years. The APP23 transgenic mice (APP23+) express a mutant form of the human amyloid precursor protein gene carrying the London mutation (V717I) under the neuron specific thy1 promoter (Sturchler-Pierrat et al., 1997). APP23+ mice were provided by Matthias Stauffenbiel from Novartis Pharmaceuticals. These animals were bred and maintained at the central animal facility of Hannover Medical School and University of Ulm. All experiments were approved by the state government. For generating G3Terc−/−APP23+ mice; Terc+/− mice were mated with APP23+ mice to produce Terc+/−APP23+ mice. These mice were then mated with Terc+/− mice to produce first generation (G1) Terc−/−APP23+ mice, which were further mated with Terc−/− mice from the same generation to produce G3Terc−/−APP23+ and G3Terc−/−APP23− mice (Supplementary Fig. 1). All mice were maintained on the C57BL/6 J background. Since aged G3 Terc−/− mice demonstrate severe loss of body weight due to metabolic failure, G3Terc−/−APP23− and G3Terc−/−APP23+ mice were sacrificed before significant loss of body weight. For bromodeoxyuridine labelling of proliferating cells, mice were injected with a single intraperitoneal dose of bromodeoxyuridine at 100 mg/kg body weight for five consecutive days and sacrificed after 1 month.
Tissue processing
For histological analysis, mice were transcardially perfused with 4% formaldehyde and post-fixed for 2 days, followed by cryoprotection in 30% sucrose. Tissues were then frozen on dry ice powder and stored at −80°C until further use. Free floating sections were cut with a cryotome and stored in a cryopreservative at −20°C.
Immunohistology and quantification
For immunohistological analysis, appropriate sections were recovered and washed with 1× Tris-buffered saline followed by blocking with 3% normal donkey (Sigma, D9663) or goat serum (Invitrogen, 016201) and 0.25% Triton-X100, for 1 h at room temperature. Sections were then incubated with primary antibodies diluted appropriately in blocking buffer for 16–24 h at 4°C. Later, after Tris-buffered saline washes, sections were incubated with secondary antibodies for 2 h at room temperature. Unbound antibody was removed by Tris-buffered saline washes. For staining using diaminobenzidine, endogenous peroxidase activity was quenched with 0.6% H2O2 in Tris-buffered saline for 30 min at room temperature followed by Tris-buffered saline washes and incubation with primary antibody. After washing and incubation with biotinylated secondary antibody, Vectastain® Elite ABC kit (Vector labs, PK6100) was used as the tertiary reagent and sections were developed with 1× diaminobenzidine substrate (Roche, 11718096001). For bromodeoxyuridine staining, genomic DNA was denatured with 2 N HCl at 37°C for 30 min followed by neutralization with 0.1 M borate buffer and Tris-buffered saline wash. The following primary antibodies were used in this study: bromodeoxyuridine (Oxford Biotech, OBT0030CX), doublecortin (Santa Cruz, SC-8066), neuronal nuclei (NeuN, Chemicon, MAB377B), γH2AX (Millipore, JBW301), amyloid-β (4G8, Sigma, A1349), glial fibrillary acidic protein (GFAP, Calbiochem, 345860), Iba-1 (Wako, 019-19741), β-amyloid cleavage enhancer-1 (BACE-1, RND, AF931), Presenilin-1 (Biomol, 2158-1), Presenilin-2 (Biomol, 1987-1), ApoE (Santa Cruz, sc-6384), β−actin (Sigma, A5060), Glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Bethyl Labs, A300-641 A). Secondary antibodies used were either donkey or goat polyclonals from Jackson Immunoresearch.
For quantification of doublecortin and bromodeoxyuridine positive cells, 40 µm thick coronal sections were taken from similar regions spanning bregma −1.30 to −3.0 mm. In total, nine serial sections were used with an interval of six and both hemispheres were quantified. For quantification of neurons, 30 µm thick coronal sections spanning bregma 1.00 to 0.2 mm (for cortical layer V neurons) and bregma −1.3 to −3.0 (for CA1 neurons) were used with an interval of six and stained with cresyl violet using standard protocols. Neuronal counting was performed using optical fractionator method with the help of Stereo Investigator 5.05.4 (MicroBrightField Inc). The counting frame size was fixed to 100 × 100 µm with a scan grid size of x:800 µm × y:300 µm (for cortical layer V neurons) and 200 × 200 µm (for CA1 neurons). The desired number of sampling sites was 10. Using these parameters the coefficient of error (Gundersen) was ∼0.1. For counting microglia cells, 20 µm thick coronal sections were stained for Iba-1. For dentate gyrus magnification of ×200 was used and the secondary visual cortex at mediomedial and mediolateral regions from the two hemispheres. For measurement of amyloid plaque load and quantification of microglia and astrocytes density around amyloid plaques, 30 µm thick coronal sections from the frontal cerebral cortex spanning bregma 1.00–0.20 mm were stained for amyloid-β, Iba-1 and glial fibrillary acidic protein. Plaques were pictured at ×400 magnification and the area was measured with the ImageJ program (NIH). Three dimensional morphological analysis of microglia was done with the program Imaris (Bitplane).
Electron microscopy
Synaptic density was measured on electron microscopy images. Briefly, 100 µm vibratome sections from 12 to 13-month-old G3Terc−/−APP23− and Terc+/+APP23− mice were flat-embedded in Epon® resin (Fluka-Sigma), dissected under a microscope and pasted on Epon blocks. Ultra-thin sections were cut at 70 nm, block stained with uranyl acetate and lead citrate and viewed with a Philips EM400T 120KV. Random pictures were taken from the soma-free neuropil areas at ×4600 magnification. Asymmetrical and symmetrical synapses were distinguished (Colonnier et al., 1968; Miranda et al., 2009) and synaptic density, defined as the number of synapse profiles in a given area/length of synaptic profiles (Colonnier et al., 1968; DeFelipe et al., 1999), was determined separately using the ImageJ program.
Measurement of telomere length
Telomere lengths were measured from the CA1 neurons according to the previously described method (Satyanarayana et al., 2003).
Protein and RNA isolation for western blotting and quantitative real-time polymerase chain reaction
For isolation of protein and RNA, brains were sectioned with a cryotome and sections from similar bregma levels were collected and homogenized in either radio-immunoprecipitation assay buffer containing complete protease inhibitor cocktail (Roche, 11836153001) or RNAzol (Wak Chemie, WAK-CS-105) containing glycogen (Invitrogen, 10814). Activated microglia surrounding amyloid plaques were micro-dissected from formalin fixed tissues using laser technology as described previously (Frank et al., 2008). Protein and RNA isolation was carried out according to the standard protocols. Protein concentration was measured with Biorad protein assay reagent (Bio-Rad, 500-0006) and RNA concentration and quality was measured with Agilent Bioanalyser 2100. For complementary DNA synthesis, SuperScript™ III first-stand complementary DNA synthesis kit (Invitrogen, 18580008) was used according to the manufacturer’s protocol. Quantitative gene expression analysis was done using TaqMan® probe for human amyloid precursor protein (Applied Biosystems) and SYBR® green primers for IL1-beta, IL6, MHCII, TGF-beta, TNF-alpha CCl2, CCL5 and CXCL10 (Supplementary Table 1). For single analysis, 1.0 µl of the total complementary DNA was used and gene expression was measured relative to GAPDH as the internal control.
Behavioural studies
After transfer to the behavioural study room, all mice were single caged and allowed to acclimatize to the facility conditions for 1 week. Animals were maintained on a 12 h light/dark cycle and tests were conducted during the light phase. Data were recorded with the help of program Viewer 7 (Biobserve). A modified version of the Morris water maze test was used to assess the spatial navigation learning and reference memory of test mice (Morris, 1984). Briefly, mice had to locate an invisible platform submerged 1 cm below the water level in a circular pool of water (dimensions 120 × 30 cm, temperature 22 ± 1°C), based on the spatial location of strategic cues fixed at distinct positions around the pool. The maximum test duration was 60 s with an intertrial interval of 10 min. Mice that failed to locate the platform were guided to it and allowed to rest for 20 s. The test was conducted over five consecutive days with four trials on each day starting from different directions to the water pool. For the first 3 days mice were trained in the water maze with constant platform position (acquisition phase), on Day 4 the platform position was changed to another quadrant (retention phase). The test was considered positive if the animals were able to locate the platform within 60 s, for a negative test, a penalty of 30 s was given.
Statistics
The programs GraphPad Prism 5 and Microsoft Office Excel 2003 were used for data analysis. P-values were calculated using two-tailed unpaired student’s t-test and a P-value <0.05 was considered statistically significant. All data sets represent mean values ± SEM from a set of replicates.
Results
Telomere shortening ameliorates cognitive impairment in ageing APP23 transgenic mice
In order to examine the effects of replicative ageing and telomere shortening on Alzheimer’s disease progression, we crossed telomerase knockout (Terc−/−) mice with the APP23 transgenic (APP23+) mice. The experimental crosses generated third generation telomerase knockout mice with shortened telomeres carrying or not carrying the APP23 transgene (G3Terc−/−APP23+ or G3Terc−/−APP23−, Supplementary Fig. 1). Mice with long telomeres were used as controls (Terc+/+APP23+ or Terc+/+APP23−).
APP23+ mice start to develop amyloid pathology at ∼6 months of age, which further leads to memory impairments with advancing age (Sturchler-Pierrat et al., 1997). To determine the influence of telomere dysfunction on Alzheimer’s disease-associated memory impairment in APP23+ mice, we compared the learning ability and spatial reference memory of 12–13-month-old mice in the Morris water maze. In agreement with previous studies (Van Dam et al., 2003), the spatial learning ability and reference memory of Terc+/+APP23+ mice was significantly reduced as compared to Terc+/+APP23− mice (Fig. 1A, P = 0.0002 at Day 5). In comparison to Terc+/+APP23− mice, G3Terc−/−APP23− mice showed a significant deficit in learning during the acquisition phase but not at the retention phase (Fig. 1A, P = 0.0092 at Day 2 and 0.0954 at Day 5). However, the expression of APP23 transgene had virtually no effect on the spatial memory of G3Terc−/− mice (Fig. 1A and B) and G3Terc−/−APP23+ mice demonstrated improved spatial memory over Terc+/+APP23+ mice (Fig. 1A, P = 0.0317 at Day 5).
Telomere shortening improves the spatial memory of ageing APP23 transgenic mice. (A) The graph represents the average time taken to locate the hidden platform (escape latency) at different days in the Morris water maze test by 12–13-month-old mice of the indicated genotypes. The test was divided over five consecutive days, for the first 3 days mice were trained in the water maze and their learning ability was measured (acquisition phase), on Day 4 the platform position was changed and their spatial memory was measured (retention phase). G3Terc−/−APP23− mice showed a deficit in spatial learning in the acquisition phase compared to the control Terc+/+APP23− mice (P = 0.0149 at Day 1 and P = 0.0092 at Day 2). Terc+/+APP23+ mice showed significantly impaired spatial learning and memory ability. However, telomere dysfunction improved the spatial memory deficit in G3Terc−/−APP23+ mice (P = 0.0317 at Day 5). (B) Graph represents the differences in escape latency for 12–13-month-old mice at different days of the test. Terc+/+APP23+ mice showed high differences over Terc+/+APP23− mice. In contrast, this difference was very low between G3Terc−/−APP23+ mice and G3Terc−/−APP23− mice (n = 8–11 mice per group, four trials per mouse each day).
Telomere shortening improves the spatial memory of ageing APP23 transgenic mice. (A) The graph represents the average time taken to locate the hidden platform (escape latency) at different days in the Morris water maze test by 12–13-month-old mice of the indicated genotypes. The test was divided over five consecutive days, for the first 3 days mice were trained in the water maze and their learning ability was measured (acquisition phase), on Day 4 the platform position was changed and their spatial memory was measured (retention phase). G3Terc−/−APP23− mice showed a deficit in spatial learning in the acquisition phase compared to the control Terc+/+APP23− mice (P = 0.0149 at Day 1 and P = 0.0092 at Day 2). Terc+/+APP23+ mice showed significantly impaired spatial learning and memory ability. However, telomere dysfunction improved the spatial memory deficit in G3Terc−/−APP23+ mice (P = 0.0317 at Day 5). (B) Graph represents the differences in escape latency for 12–13-month-old mice at different days of the test. Terc+/+APP23+ mice showed high differences over Terc+/+APP23− mice. In contrast, this difference was very low between G3Terc−/−APP23+ mice and G3Terc−/−APP23− mice (n = 8–11 mice per group, four trials per mouse each day).
Telomere shortening reduces the progression of amyloid pathology in ageing APP23 transgenic mice
The improvement in spatial memory of G3Terc−/−APP23+ mice over Terc+/+APP23+ mice indicated that telomere shortening plays a protective role on Alzheimer’s disease progression and memory impairment, which might be associated with amyloid plaque pathology in the brain. In APP23 transgenic mice as well as in human patients with Alzheimer’s disease, the amyloid pathology starts to develop in the frontal cerebral cortex and later gradually spreads to the hippocampus and other regions of the brain (Thal et al., 2002, 2006). Given the limited survival of G3Terc−/− mice to 12–15 months (Choudhury et al., 2007; Schaetzlein et al., 2007) and almost complete absence of plaque pathology in the hippocampus area of APP23 transgenic mice at this age, we analysed amyloid pathology in the frontal cortex of the brain. The analysis revealed a significant reduction in the number of amyloid plaques from 12-month-old G3Terc−/−APP23+ mice (average 23.95 ± 3.741) compared to Terc+/+APP23+ mice (average 42.25 ± 6.580, P = 0.0247, Fig. 2A, C and D). Also, the amyloid plaque load (represents percentage of brain area occupied by plaques) was significantly reduced in G3Terc−/−APP23+ mice (average 5.658 ± 0.8900) compared to Terc+/+APP23+ mice (average 11.02 ± 2.120, P = 0.032, Fig. 2B–D). Western blot analysis on transgenic APP23, BACE-1 (responsible for β-cleavage of amyloid precursor protein), and presenilin proteases (responsible for γ-cleavage of amyloid precursor protein) confirmed unaltered expression of these proteins in G3Terc−/−APP23+ mice compared to Terc+/+APP23+ mice (Supplementary Fig. 2A–E), indicating that the effect of telomere shortening on amyloid pathology was independent of the expression and processing of transgenic APP23.
Telomere shortening reduces the formation of amyloid plaques. (A) Histogram shows the number of amyloid plaques in frontal cerebral cortex of 12–13-month-old G3Terc−/−APP23+ mice (average of 23.95 ± 3.741) compared to Terc+/+APP23+ mice (average of 42.25 ± 6.580, P = 0.0247). (B) Histogram shows the amyloid plaque load (the percentage of brain area occupied by amyloid plaques) in the frontal cerebral cortex of 12–13-month-old G3Terc−/−APP23+ mice (average of 5.658 ± 0.89) compared to Terc+/+APP23+ mice (average of 11.02 ± 2.120, n = 21–24 sections, eight mice per group, P = 0.032). (C) Representative images from the frontal cerebral cortex of G3Terc−/−APP23+ (C) and Terc+/+APP23+ (D) mice. Arrows indicate amyloid plaques. Scale bar = 500 µm.
Telomere shortening reduces the formation of amyloid plaques. (A) Histogram shows the number of amyloid plaques in frontal cerebral cortex of 12–13-month-old G3Terc−/−APP23+ mice (average of 23.95 ± 3.741) compared to Terc+/+APP23+ mice (average of 42.25 ± 6.580, P = 0.0247). (B) Histogram shows the amyloid plaque load (the percentage of brain area occupied by amyloid plaques) in the frontal cerebral cortex of 12–13-month-old G3Terc−/−APP23+ mice (average of 5.658 ± 0.89) compared to Terc+/+APP23+ mice (average of 11.02 ± 2.120, n = 21–24 sections, eight mice per group, P = 0.032). (C) Representative images from the frontal cerebral cortex of G3Terc−/−APP23+ (C) and Terc+/+APP23+ (D) mice. Arrows indicate amyloid plaques. Scale bar = 500 µm.
Telomere shortening is associated with the accumulation of DNA damage foci and loss of neurons in the ageing brain
The protective effect of telomere shortening on Alzheimer’s disease progression was an unexpected finding given its crucial role in cell division and organ maintenance. Previous studies on Terc−/− mice have shown progressive telomere shortening through the successive generations of Terc−/− mice, and late generation mice (G3–G6, depending on the mouse strain) show defects in homeostasis of organ systems with high rates of cellular turnover and a shortened lifespan (Blasco et al., 1997; Lee et al., 1998; Rudolph et al., 1999). However, telomere shortening and presence of dysfunctional telomeres have not been investigated in the brain of ageing Terc deficient mice. Measurement of telomere length from the CA1 neurons by quantitative fluorescence in situ hybridization revealed a significant reduction in 3- and 12-month-old G3Terc−/−APP23− mice compared to age-matched Terc+/+APP23− mice (Fig. 3A, P = 0.0022 and 0.0003, respectively). Also, both cohorts showed an age-dependent reduction in telomere length (Fig. 3A). The reduction in telomere length correlated with the accumulation of γH2AX positive foci [a marker of DNA damage and dysfunctional telomeres (Rogakou et al., 1998)] in the CA1 neurons of 3- and 12-month-old G3Terc−/−APP23− mice compared to age-matched Terc+/+APP23− mice (Fig. 3B, P < 0.0001 for 3- and 12-month-old mice). Moreover, both cohorts showed an age-dependent increase in γH2AX positive foci (Fig. 3B).
Dysfunctional telomeres and loss of neurons in the brain of ageing telomerase deficient mice. (A) Histogram shows the telomere lengths from CA1 neurons in 3- and 12-month-old G3Terc−/−APP23− mice compared to age matched Terc+/+APP23− mice (n = 5 mice per group, P = 0.0022 at 3 months old and P = 0.0003 at 12 months old). (B) Histogram shows the percentage number of γH2AX-positive CA1 neurons in 3- and 12-month-old G3Terc−/−APP23− mice compared to age-matched Terc+/+APP23− mice (n = 5 mice per group). (C) The histogram shows the density of CA1 neurons in 12-month-old G3Terc−/−APP23− mice (average 264.6 ± 9.177) compared to age-matched Terc+/+APP23− mice (average 310.4 ± 13.96, n = 5 mice per group, six serial sections per mouse, P = 0.0255). The difference was not significant in 3-month-old mice (average 291.4 ± 12.77 for G3Terc−/−APP23− mice and 315.5 ± 23.51 for Terc+/+APP23− mice, n = 4–5 mice per group, six serial sections per mouse. P = 0.3708). (D) Histogram shows the density of neurons from layer V of frontal cerebral cortex in 12-month-old G3Terc−/−APP23− (312.2 ± 55.38) mice compared to Terc+/+APP23− mice (443.6 ± 51.12, n = 5 mice per group. P = 0.045). (E) The histogram shows the synaptic density (synapses/mm3) in the frontal cortex of G3Terc−/−APP23− mice (average 1048 ± 111.9) compared to Terc+/+APP23− (average 1480 ± 83.7, n = 4 mice per group. P = 0.0286). (F) Representative electron microscope images from the frontal cortex of 12-month-old G3Terc−/−APP23− and Terc+/+APP23− mice. Arrows indicate synapses.
Dysfunctional telomeres and loss of neurons in the brain of ageing telomerase deficient mice. (A) Histogram shows the telomere lengths from CA1 neurons in 3- and 12-month-old G3Terc−/−APP23− mice compared to age matched Terc+/+APP23− mice (n = 5 mice per group, P = 0.0022 at 3 months old and P = 0.0003 at 12 months old). (B) Histogram shows the percentage number of γH2AX-positive CA1 neurons in 3- and 12-month-old G3Terc−/−APP23− mice compared to age-matched Terc+/+APP23− mice (n = 5 mice per group). (C) The histogram shows the density of CA1 neurons in 12-month-old G3Terc−/−APP23− mice (average 264.6 ± 9.177) compared to age-matched Terc+/+APP23− mice (average 310.4 ± 13.96, n = 5 mice per group, six serial sections per mouse, P = 0.0255). The difference was not significant in 3-month-old mice (average 291.4 ± 12.77 for G3Terc−/−APP23− mice and 315.5 ± 23.51 for Terc+/+APP23− mice, n = 4–5 mice per group, six serial sections per mouse. P = 0.3708). (D) Histogram shows the density of neurons from layer V of frontal cerebral cortex in 12-month-old G3Terc−/−APP23− (312.2 ± 55.38) mice compared to Terc+/+APP23− mice (443.6 ± 51.12, n = 5 mice per group. P = 0.045). (E) The histogram shows the synaptic density (synapses/mm3) in the frontal cortex of G3Terc−/−APP23− mice (average 1048 ± 111.9) compared to Terc+/+APP23− (average 1480 ± 83.7, n = 4 mice per group. P = 0.0286). (F) Representative electron microscope images from the frontal cortex of 12-month-old G3Terc−/−APP23− and Terc+/+APP23− mice. Arrows indicate synapses.
To further understand whether telomere shortening would affect the maintenance of post-mitotic neurons in the adult brain, we quantified the number of CA1 neurons from 3- and 12–13-month-old ageing G3Terc−/−APP23− and Terc+/+APP23− mice (Fig. 3C). The CA1 region of hippocampus lacks active neurogenesis, which makes it suitable to study the effect of telomere shortening on maintenance of post-mitotic neurons in a region without active cell division. While the difference was not significant in 3-month-old mice (Fig. 3C, P = 0.3708), the average density of CA1 neurons was significantly reduced in 12-month-old G3Terc−/−, APP23− mice compared to age-matched Terc+/+, APP23− mice (Fig. 3C, P = 0.0255).
Since in APP23 transgenic mouse model of Alzheimer’s disease, amyloid pathology starts to develop in layer V of the frontal cerebral cortex (Thal et al., 2006), we also quantified the number of layer V neurons from frontal cerebral cortex spanning bregma 1.1 and 0.2. The density of layer V neurons was also reduced in 12-month-old G3Terc−/−APP23− mice (average 443.6 ± 51.12) compared to Terc+/+APP23− mice (average 312.20 ± 21.30, Fig. 3D, P = 0.045). In addition, the average synaptic density in the frontal cortex region was also reduced in 12-month-old G3Terc−/−APP23− mice (average 1048 ± 111.90) compared to age-matched Terc+/+APP23− mice (average 1480 ± 83.78, Fig. 3E and F, P = 0.0286).
Telomere shortening abrogates dentate gyrus neurogenesis in ageing mice
Telomere shortening limits the replicative potential of proliferating cells (Allsopp et al., 1995). In agreement with this, previous work on late generation Terc−/− mice showed a reduction in adult neurogenesis of the subventricular zone (Ferron et al., 2004). Given the key role of hippocampus in learning and memory, we analysed dentate gyrus neurogenesis in 12–13-month-old G3Terc−/−, APP23− and Terc+/+, APP23− mice by bromodeoxyuridine incorporation and quantification of doublecortin positive cells, a marker for active neurogenesis (Couillard-Despres et al., 2005). The number of doublecortin positive cells was reduced more than 4-fold in G3Terc−/−APP23− mice (average 440.0 ± 225.6) compared to Terc+/+APP23− mice (average 1931 ± 138.7, Fig. 4A, P = 0.0005). Similarly, the number of bromodeoxyuridine-positive cells was also strongly reduced in G3Terc−/−APP23− mice (average 28.11 ± 9.585) compared to Terc+/+APP23− mice (average 198.0 ± 29.47, Fig. 4B, P = 0.0006). In agreement with these results, quantification of doublecortin-positive cells also revealed a significant reduction in dentate gyrus neurogenesis in G3Terc−/−APP23+ mice (average 692.6 ± 111.6) compared to Terc+/+APP23+ mice (average 1814 ± 139.1, Fig. 4A, P = 0.0006).
Telomere shortening reduces dentate gyrus neurogenesis in ageing mice. (A) Histogram shows the average number of doublecortin (DCX) positive cells from the dentate gyrus of 12-month-old mice (440 ± 225.6 for G3Terc−/−APP23− mice; 1931 ± 138.7 for Terc+/+APP23− mice; 692.6 ± 111.6 for G3Terc−/−APP23+ mice; 1814.0 ± 139.1 for G3Terc−/−APP23+ mice, n = 5–6 mice per group, nine sections per mouse, P = 0.0005 between G3Terc−/−APP23− and Terc+/+APP23− mice; P = 0.0006 between G3Terc−/−APP23+ mice and Terc+/+APP23+ mice). (B) Histogram shows the average number of bromodeoxyuridine (brdU) positive cells from the dentate gyrus of 12-month-old mice (28.11 ± 9.585 for G3Terc−/−APP23− mice and 198 ± 29.47 for Terc+/+APP23− mice, n = 5 mice per group, nine sections per mouse, P = 0.0006). (C) Histogram shows the average number of bromodeoxyuridine-NeuN double positive cells (19.56 ± 7.584 for G3Terc−/−APP23− mice and 134.4 ± 27.53 for Terc+/+APP23− mice, n = 5 mice per group, nine sections per mouse, P = 0.0038). (D, E) Representative fluorescence images from the dentate gyrus of G3Terc−/−APP23− (D) and Terc+/+APP23− (E) showing doublecortin (red arrow), bromodeoxyuridine (green arrow) and NeuN (blue) positive cells. Scale bar = 20 µm.
Telomere shortening reduces dentate gyrus neurogenesis in ageing mice. (A) Histogram shows the average number of doublecortin (DCX) positive cells from the dentate gyrus of 12-month-old mice (440 ± 225.6 for G3Terc−/−APP23− mice; 1931 ± 138.7 for Terc+/+APP23− mice; 692.6 ± 111.6 for G3Terc−/−APP23+ mice; 1814.0 ± 139.1 for G3Terc−/−APP23+ mice, n = 5–6 mice per group, nine sections per mouse, P = 0.0005 between G3Terc−/−APP23− and Terc+/+APP23− mice; P = 0.0006 between G3Terc−/−APP23+ mice and Terc+/+APP23+ mice). (B) Histogram shows the average number of bromodeoxyuridine (brdU) positive cells from the dentate gyrus of 12-month-old mice (28.11 ± 9.585 for G3Terc−/−APP23− mice and 198 ± 29.47 for Terc+/+APP23− mice, n = 5 mice per group, nine sections per mouse, P = 0.0006). (C) Histogram shows the average number of bromodeoxyuridine-NeuN double positive cells (19.56 ± 7.584 for G3Terc−/−APP23− mice and 134.4 ± 27.53 for Terc+/+APP23− mice, n = 5 mice per group, nine sections per mouse, P = 0.0038). (D, E) Representative fluorescence images from the dentate gyrus of G3Terc−/−APP23− (D) and Terc+/+APP23− (E) showing doublecortin (red arrow), bromodeoxyuridine (green arrow) and NeuN (blue) positive cells. Scale bar = 20 µm.
To analyse the maturation of the newborn neurons, we quantified the number of bromodeoxyuridine and NeuN double positive cells (Mullen et al., 1992; van Praag et al., 2002) in the dentate gyrus of these mice. A significant reduction was also evident in G3Terc−/−APP23− mice (average 19.56 ± 7.584) compared to Terc+/+APP23− control mice (average 134.4 ± 27.53) (Fig. 4C–E, P = 0.0038). Together, these results indicate that telomere shortening has a negative effect on adult neurogenesis as well as on the maintenance of post-mitotic neurons in the ageing mouse brain.
Telomere shortening is associated with the activation of microglia in amyloid-free ageing brain
Microglial activation is believed to play a key role in Alzheimer’s disease pathogenesis since activated microglia may phagocytose amyloid-β and prevent the formation of plaques (Koenigsknecht and Landreth, 2004). To understand the possible reasons for improved spatial memory and amyloid pathology in G3Terc−/−APP23+ mice, we studied the effects of telomere shortening on microglia cells. Studies on human cell lines and ageing mice have revealed experimental evidence that telomere dysfunction induces the expression of proinflammatory cytokines (Ju et al., 2007; Rodier et al., 2009) making it conceivable that such alterations could also influence the activity of microglia in telomerase deficient animals. In order to determine microglia activation, we assessed 3D morphology and density of microglia from different cohorts by Iba-1 immunostaining (Imai et al., 1996). To allow a comparison between animals irrespective of the presence of amyloid plaques, we first analysed microglia in the amyloid-free brain. The average density of microglia cells was slightly higher in 12-month-old G3Terc−/−APP23− mice (average 63.88 ± 1.610) compared to age-matched Terc+/+APP23− mice (average 58.80 ± 0.8627, Fig. 5A, P = 0.027). A detailed 3D morphological analysis of microglia revealed significant microglia activation in the hippocampus and frontal cortex of 12-month-old G3Terc−/−APP23− mice compared to Terc+/+APP23− mice (Fig. 5B–F, Supplementary Fig. 3). In both regions, microglia from G3Terc−/−APP23− mice exhibited reduced dendritic length (Fig. 5C, P = 0.009; Fig. 5E, P = 0.001) and reduced dendritic branching (Fig. 5D, P = 0.017; Fig. 5F, P = 0.012) compared to Terc+/+, APP23− mice. In contrast to these morphological markers of microglia activation, an increase in microglia proliferation (Supplementary Fig. 4A and B, P = 0.5205) or in expression of inflammatory markers, was not observed in G3Terc−/−APP23− mice compared to Terc+/+APP23− mice (Supplementary Fig. 5A). These results indicate that G3Terc−/− mice exhibit a certain degree of microglia activation in absence of amyloid pathology.
Microglia activation in ageing telomerase deficient mice. (A) The histogram shows the average density of Iba-1 positive microglia in G3Terc−/−APP23+ mice (average 65.13 ± 2.038, n = 12 mice); Terc+/+APP23+ mice (average 73.06 ± 2.712, n = 12 mice); Terc+/+APP23− mice (average 58.80 ± 0.8627, n = 9 mice), and G3Terc−/−APP23− mice (average of 63.88 ± 1.61, n = 14 mice). Note the significant increase in Iba-1 positive microglia in G3Terc−/−APP23− mice compared to Terc+/+APP23− mice (P = 0.027). (B) Representative images from the 3D morphological analysis of microglia activation via the measurement of total dendritic length and number of dendritic branch points. (C, E) Histogram represents the total dendritic length of microglia from the cortex (C) and CA1 region (E) of 12-month-old mice of the indicated genotypes. (D, F) Number of dendritic branch points in the cortex (D) and CA1 region (F) in 12-month-old mice of the indicated genotypes. Note, that a decrease in dendritic length and a reduction in branch points indicate a higher microglia activity in G3Terc−/−APP23− mice compared to Terc+/+APP23− mice. Data in C–F are provided as mean ± SEM (n = 9 cells, 3–5 mice).
Microglia activation in ageing telomerase deficient mice. (A) The histogram shows the average density of Iba-1 positive microglia in G3Terc−/−APP23+ mice (average 65.13 ± 2.038, n = 12 mice); Terc+/+APP23+ mice (average 73.06 ± 2.712, n = 12 mice); Terc+/+APP23− mice (average 58.80 ± 0.8627, n = 9 mice), and G3Terc−/−APP23− mice (average of 63.88 ± 1.61, n = 14 mice). Note the significant increase in Iba-1 positive microglia in G3Terc−/−APP23− mice compared to Terc+/+APP23− mice (P = 0.027). (B) Representative images from the 3D morphological analysis of microglia activation via the measurement of total dendritic length and number of dendritic branch points. (C, E) Histogram represents the total dendritic length of microglia from the cortex (C) and CA1 region (E) of 12-month-old mice of the indicated genotypes. (D, F) Number of dendritic branch points in the cortex (D) and CA1 region (F) in 12-month-old mice of the indicated genotypes. Note, that a decrease in dendritic length and a reduction in branch points indicate a higher microglia activity in G3Terc−/−APP23− mice compared to Terc+/+APP23− mice. Data in C–F are provided as mean ± SEM (n = 9 cells, 3–5 mice).
In accordance to the result with amyloid plaques, a significant increase in microglia cell number was observed in Terc+/+APP23+ mice (average 73.06 ± 2.712) compared to Terc+/+APP23− mice, (Fig. 5A, P = 0.0003), but not in G3Terc−/−APP23+ mice (average 65.13 ± 2.038) compared to G3Terc−/−APP23− (Fig. 5A, P = 0.6299). Activation of microglia in the cortex of Terc+/+APP23+ mice in response to plaque formation reduced telomere-related differences in microglia activation (Fig. 5C and D). Since previous studies have shown activation of microglia in response to amyloid plaques, the impaired plaque formation in G3Terc−/−APP23+ mice could provide a plausible explanation for reduced microglia activation in G3Terc−/−APP23+ mice compared to Terc+/+APP23+ mice. However, in the direct vicinity of plaques, microglia displayed an equal phenotype in both cohorts. The density of activated phagocytic microglia surrounding amyloid plaques was similar in G3Terc−/−APP23+ mice (average 4.163 ± 0.1954, P = 0.9241) and Terc+/+APP23+ mice (average 4.223 ± 0.6345) (Supplementary Fig. 5B, P = 0.9241) and quantitative real-time polymerase chain reaction analysis on laser micro-dissected microglia surrounding amyloid plaques revealed similar expression levels of MHC-II in both cohorts (Supplementary Fig. 5C, P = 0.5727), suggesting that a direct contact with amyloid plaques induces microglia activation irrespective of telomerase gene status and telomere function.
Since astrocytes also form a part of the intracranial immune defence against amyloid plaques, we compared the density of glial fibrillary acidic protein positive astrocytes at the vicinity of plaques from the two groups. The average density of astrocytes was also not different from the two groups with an average of 17.20 ± 0.8019 for G3Terc−/−APP23+ mice and 16.34 ± 0.5515 for Terc+/+APP23+ mice (Supplementary Fig. 6, P = 0.7365).
Discussion
The strong association between Alzheimer’s disease and ageing suggests that molecular events associated with ageing may promote development of the disease. Given the close intimacy between ageing and telomere shortening, it was important to determine whether telomere shortening would have any direct influence on Alzheimer’s disease development and/or progression. The current study shows that telomere shortening reduces adult dentate gyrus neurogenesis and impairs the maintenance of post-mitotic neurons in aged late generation telomerase deficient mice. However, the surprising finding of this work is that telomere dysfunction inhibited the progression of Alzheimer’s disease-associated amyloid pathology and the evolution of memory defects in ageing APP23 transgenic mice. While APP23 expression resulted in strong impairment of spatial learning and memory ability in Terc+/+ mice, it virtually had no effects in ageing G3Terc−/− mice.
The decrease in dentate gyrus neurogenesis in aged G3Terc−/− mice stands in agreement with the limiting role of telomeres in cell division (Lee et al., 1998) and impaired subventricular neurogenesis in telomerase-deficient mice (Ferron et al., 2004). Previous studies on mice have shown that dentate gyrus neurogenesis is critical for certain behavioural aspects that involve learning (Tashiro et al., 2007) and even though only 10% of the dentate gyrus neurons undergo turnover (Lagace et al., 2007; Ninkovic et al., 2007; Imayoshi et al., 2009), they seem to have an important role (Aimone et al., 2006 and references within) since the act of learning itself is associated with increased survival of newborn neurons (Gould et al., 1999). The existing study extends previous findings on telomerase-deficient mice by showing that telomere shortening can also influence the maintenance of post-mitotic neurons in the CA1 region of hippocampus as well as neuronal numbers and synaptic density in the frontal cortex. It is tempting to speculate that these factors may contribute to the deficit in learning ability of aged G3Terc−/−APP23− mice compared to wild-type mice.
Despite these inhibitory effects on adult neurons, the current study provides the first experimental evidence that telomere shortening can impair Alzheimer’s disease progression in mice. Specifically, telomere shortening prevented the formation of Alzheimer’s disease-associated amyloid plaques and improved the memory of G3Terc−/−APP23+ mice in Morris water maze compared to Terc+/+APP23+ mice with long telomere reserves. These results indicate that telomere dysfunction prevented the formation of amyloid plaques and memory impairment in the APP23 transgenic mouse model of Alzheimer’s disease.
An active immune response has been shown to exert protective effects in Alzheimer’s disease mouse models (Shaftel et al., 2007). It was recently shown that CX3CR1-deletion leads to microglia activation (Cardona et al., 2006) and reduced amyloid-β load in two different mouse models of Alzheimer’s disease (Lee et al., 2011). Various studies have shown that activated microglia have a higher amyloid-β uptake capacity supporting the concept that microglia can protect from amyloid plaques (Malm et al., 2005; Simard et al., 2006; El Khoury et al., 2007; Shaftel et al., 2007; Koenigsknecht-Talboo et al., 2008). Since our 3D morphological analysis showed an activation of microglia in amyloid-free brain of telomere dysfunctional mice, it is tempting to speculate that similar mechanisms may contribute to impairments in plaque formation in G3Terc−/− mice. However, we did not detect an influence of telomere dysfunction on microglia proliferation, the expression of proinflammatory cytokines or the prevalence of highly activated microglia with a macrophage-like morphology in the amyloid-free brain of ageing mice. These results indicate that telomere dysfunction does not lead to a full-blown proinflammatory activation of microglia cells.
In contrast to the beneficial effects of microglia on Alzheimer’s disease progression, it was recently shown that ablation of microglia in CD11b-HSVTK-amyloid precursor protein mice had no effect on amyloid-β pathology (Grathwohl et al., 2009). In another mouse model it was observed that CX3CR1-deficiency and resulting higher activity of microglia correlated with increased Tau pathology, suggesting that microglia activation can also have adverse effects on Alzheimer’s disease-associated pathology (Bhaskar et al., 2010). The current study revealed a significant and stronger expansion of Iba1-positive microglia cells in APP23 transgenic mice with functional telomeres compared to mice with dysfunctional telomeres. It is possible that reduced expansion of microglia cells in APP23 transgenic mice with dysfunctional telomeres merely reflects the impaired development of amyloid plaques. Alternatively, telomere dysfunction may inhibit reactive proliferation of microglia in response to amyloid-β formation. In aged humans, there is evidence for microglia senescence, which has been implicated to impair microglia function and render the human brain more vulnerable to Alzheimer’s disease development (Conde and Streit, 2007; Streit et al., 2006). Our study indicates that the influence of telomere dysfunction on microglia may not be detrimental but may contribute to the prevention of Alzheimer’s disease-associated amyloid plaque formation and memory loss.
Aside from microglia activation and proliferation, it is possible that telomere dysfunction induces certain systemic effects that may influence the progression of amyloid pathology in APP23 transgenic mice. Another study from our laboratory shows that telomere dysfunction in ageing late generation Terc−/− mice is associated with reduced glucose levels, greater insulin sensitivity, and reduced IGF-1 signalling (Missios P and Rudolph KL, unpublished results). Three recent studies have provided direct experimental evidence that reduced IGF-1 signalling can impair Alzheimer’s disease progression in mice (Cohen et al., 2009; Freude et al., 2009; Killick et al., 2009). Another study has shown that DNA damage can impair somatotrophic signalling including IGF-1 (Niedernhofer et al., 2006). Therefore, it is possible that reduced IGF-1 signalling also contributed to reduced amyloid pathology in ageing telomerase-deficient mice compared to wild-type mice.
In conclusion, the current study has revealed pleiotropic effects of telomere shortening on mouse brain ageing. On one hand, it impaired adult neurogenesis and maintenance of post-mitotic neurons, and on the other, it had protective effects against amyloid pathology in an APP23 transgenic mouse model of Alzheimer’s disease.
Supplementary material
Supplementary material is available at Brain online.
Funding
This work was supported by the Fritz-Thyssen-Stiftung (10.06.1.205) and the DFG (KFO 142: RU745-13) and the EU (Geninca, Telomarker). S.M.H. is supported by the EU (EUMODIC grant # LSHG-2006-037188), the Federal Ministry of Education and Research (NGFN grants # 01GS0850 and # 01GS08133), the Deutsche Forschungsgemeinschaft (SFB 596 TP A12) and the Helmholtz Gemeinschaft (HELMA TP2 Mechanics). The work of KB was supported by Deutsche Forschungsgemeinschaft (DFG) research unit 1336 (For1336).
Acknowledgements
H.R. performed most of the work. H.R. and A.S. did neuronal quantification. A.S. and B.H.L. did DNA damage and qFISH analysis. A.H. did 3D morphological analysis for microglia activation. A.S. and Y.B.N. did RT-PCR analysis on laser micro-dissected microglia. A.S. did microglia proliferation analysis. D.R.T., H.R. and A.S. did synapse measurement. D.R.T., K.B., S.M.H., W.W., D.M.V.W. and D.C.L. supported the study. H.R. and K.L.R. designed the study and wrote the paper.
