Inducing autophagy by rapamycin before, but not after, the formation of plaques and tangles ameliorates cognitive deficits

Abstract

Previous studies have shown that inducing autophagy ameliorates early cognitive deficits associated with the build-up of soluble amyloid-β (Aβ). However, the effects of inducing autophagy on plaques and tangles are yet to be determined. While soluble Aβ and tau represent toxic species in Alzheimer's disease (AD) pathogenesis, there is well documented evidence that plaques and tangles also are detrimental to normal brain function. Thus, it is critical to assess the effects of inducing autophagy in an animal model with established plaques and tangles. Here we show that rapamycin, when given prophylactically to 2-month-old 3xTg-AD mice throughout their life, induces autophagy and significantly reduces plaques, tangles and cognitive deficits. In contrast, inducing autophagy in 15-month-old 3xTg-AD mice, which have established plaques and tangles, has no effects on AD-like pathology and cognitive deficits. In conclusion, we show that autophagy induction via rapamycin may represent a valid therapeutic strategy in AD when administered early in the disease progression.

Figure 1. Short- and long-term rapamycin treatments do not cause overt side effects.

(A) Schematic of the experimental design. 3xTg-AD and NonTg mice were randomly assigned to one of the following groups: (i) 20 mice/genotype fed rapamycin-containing food starting at 2 months of age for 16 months; (ii) 20 mice/genotype fed control diet for the first 15 months of their life after which they were fed rapamycin-containing food for 3 months; (iii) 20 mice/genotype fed control diet throughout their life. All mice were 18 months of age at the end of the treatment. (B) All mice gained weight throughout the treatment and no statistically significant differences were found among the groups. Data are presented as means ± SEM.

Figure 2. Rapamycin prevents, but does not rescue learning and memory deficits.

(A) Mice were evaluated in the spatial reference version of the MWM. Mice significantly learned the task over the 5 days of training, as indicated by a reduced time to find the escape platform (F = 36.2; p<0.0001 as calculated by a mixed-model repeated-measures ANOVA). There was also a significant genotype/treatment-day interaction (F = 2.68; p = 0.021). Bonferroni post hoc analysis showed that the NonTg^2–18 mice learned the task significantly quicker than the NonTg^CTL mice. In contrast, the NonTg^15–18 mice learned as well as the NonTg^CTL mice. Similarly, 3 months of rapamycin treatment did not improve learning in the 3xTg-AD mice as the 3xTg-AD^15–18 mice performed similarly to the 3xTg-AD^CTL mice. In contrast, we found that the 3xTg-AD^2–18 mice learned the task significantly quicker than 3xTg-AD^CTL mice and as well as NonTg^CTL mice. (B–C) Reference memory, measured 24 hours after the last training trials was significantly improved only in the NonTg^2–18 and 3xTg-AD^2–18 mice compared to the NonTg^CTL and 3xTg-AD^CTL mice, respectively. Three months of rapamycin administration, however, did not have any effect on reference memory. (D–E) Swimming speed and distance traveled during the probe trials were not significantly different among the 6 groups of mice. (F) Mice were also tested using the object recognition task, a cortical-dependent task. One-way ANOVA showed significant changes in the time mice spent exploring the new object across the 6 different groups (Fig. 2F; p = 0.01). Post-hoc analysis showed that short- and long-term rapamycin treatment had no effect on NonTg mice. In contrast, the 3xTg-AD^2–18 mice performed significantly better than the 3xTg-AD^CTL mice. Data are presented as means ± SEM.

Figure 3. Rapamycin does not change APP processing.

(A) Representative Western blots of proteins extracted from 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice (n = 8/group) and probed with the indicated antibodies. (B–C) Quantitative analysis of the blots showed that rapamycin did not change the steady-state levels of full length APP or its two major C-terminal fragments, C99 and C83. Data are presented as means ± SEM and analyzed by one-way ANOVA.

Figure 4. Life-long rapamycin administration reduces Aβ levels and deposition.

(A–I) Representative sections from brains of 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice (n = 8/group) immunostained with an Aβ-specific antibody (A–B, D–E, G–H) and stained with thioflavin S (C, F, I), clearly show that the 3xTg-AD^2–18 mice have less diffuse and fibrillar Aβ deposits compared to 3xTg-AD^CTL and 3xTg-AD^15–18 mice. (J) Semi-quantitative assessment of the number of thioflavin-positive plaques shows no significant change between 3xTg-AD^CTL and 3xTg-AD^15–18 mice. In contrast, the 3xTg-AD^2–18 mice have significantly less plaques compared to the other two groups. One-way ANOVA across the three different groups shows that the changes were highly significant (F = 69.65; p<0.0001). Panels B, E and H represent high magnification views of panels A, D and G. (K–L) Soluble (K) and insoluble (L) Aβ40 and Aβ42 levels were measured by sandwich ELISA. Consistent with the histological results, compared to 3xTg-AD^CTL mice, soluble and insoluble Aβ40 and Aβ42 levels were significantly reduced only in the 3xTg-AD^2–18 mice (F = 40.50; p<0.0001 for the soluble Aβ levels; F = 22.51 and p<0.0001 for the insoluble Aβ levels). Data are presented as means ± SEM.

Figure 5. Tau pathology is significantly reduced in 3xTg-AD^2–18 mice.

(A–F) Representative sections depicting CA1 pyramidal neurons from brains of 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice (n = 8/group) immunostained with the indicated anti-tau antibodies. Note the reduction of AT100- and AT180-positive neurons in the 3xTg-AD^2–18 mice compared to 3xTg-AD^CTL and 3xTg-AD^15–18 mice. (G) Representative Western blots of proteins extracted from 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice (n = 8/group) and probed with the indicated antibodies. (H) Quantitative analysis of the blots shows that rapamycin did not change the steady-state levels of full length tau transgene as measured by the human-specific anti-tau antibody, HT7. In contrast, the levels of tau phosphorylated at the AT100 and AT180 epitopes were significantly reduced in the 3xTg-AD^2–18 mice compared to the 3xTg-AD^CTL and 3xTg-AD^15–18 mice (F = 5.271 and p = 0.018 for AT100; F = 14.25 and p = 0.0003 for AT180). Data are presented as means ± SEM.

Figure 6. Rapamycin reduced the number of activated microglia.

(A–C) Representative sections from CA1/subiculum regions of brains from 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice brains (n = 8/group) immunostained with an anti-CD45 antibody. (D) Semi-quantitative analysis showed that the number of activated microglia was significantly different across the three groups as determined by one-way ANOVA (F = 5.7; p = 0.01). Bonferroni's post hoc analysis showed that the number of activated microglia was significantly lower in the 3xTg-AD^2–18 mice compared to the other two groups. No statistically significant changes were found between 3xTg-AD^CTL and 3xTg-AD^15–18 mice. Data are presented as means ± SEM.

Figure 7. Autophagy is equally induced in 3xTg-AD^2–18 and 3xTg-AD^15–18 mice.

(A) Representative Western blots of proteins extracted from 3xTg-AD^CTL, 3xTg-AD^15–18 and 3xTg-AD^2–18 mice (n = 8/group) and probed with the indicated antibodies. (B–C) Quantitative analysis of the blots showed that rapamycin did not change the steady-state levels of total p70S6K. In contrast, the levels of p70S6K phosphorylated at Thr389 (a site directly phosphorylated by mTOR) were significantly changed by rapamycin administration (F = 23.07; p<0.001). Bonferroni's post hoc analysis showed that the levels of p70S6K phosphorylated at Thr389 were not significantly different between 3xTg-AD^2–18 and 3xTg-AD^15–18 mice. In contrast, phospho-p70S6K levels in both groups were significantly lower compared to the 3xTg-AD^CTL mice (p<0.001). (D–E) Similar results were obtained when we quantified the levels of the autophagy-related proteins, Atg7 and Atg5/Atg12. One-way ANOVA showed that there was a group effect for Atg7 (F = 46.92; p<0.0001) and Atg5/Atg12 (F = 64.37; p<0.0001). Post-hoc analysis confirmed that the levels of Atg7 and Atg5/Atg12 were significantly higher in 3xTg-AD^15–18 and 3xTg-AD^2–18 mice compared to 3xTg-AD^CTL mice, but no significant differences were found between 3xTg-AD^15–18 and 3xTg-AD^2–18 mice. (F–G) Quantitation of the LC3I/II levels showed that while rapamycin did not alter LC3I levels (F = 2.039; p = 0.1552, as calculated by one-way ANOVA), a significant group effect was found for LC3II levels (F = 14.58; p = 0.0001). Consistent with the Atg levels, the groups responsible for this difference were the 3xTg-AD^15–18 and 3xTg-AD^2–18 mice, which showed significantly higher LC3II levels compared to 3xTg-AD^CTL mice, as determined by Bonferroni's post-hoc test (p<0.001). The levels of LC3II were not significantly different between the 3xTg-AD^15–18 and 3xTg-AD^2–18 mice. Data are presented as means ± SEM.

Figure 8. Abnormal autophagosomes in 15-month-old 3xTg-AD mice.

Electron microscope sections obtained from CA1 regions of 15-month-old 3xTg-AD mice. Sections show examples of enlarged autophagosomes containing electron-dense undigested materials. Top-left panel: N = nucleus; the arrowhead points to the nuclear membrane; the arrows point to enlarged autophagosomes. Top-right panel: the arrowhead points to an autophagosome that does not contain undigested material. The arrow points to an autophagosome containing undigested material. Similar structures are also shown in the bottom two panels (arrows).