SIRT1 suppresses beta-amyloid production by activating the alpha-secretase gene ADAM10

Abstract

A hallmark of Alzheimer's disease (AD) is the accumulation of plaques of Abeta 1-40 and 1-42 peptides, which result from the sequential cleavage of APP by the beta and gamma-secretases. The production of Abeta peptides is avoided by alternate cleavage of APP by the alpha and gamma-secretases. Here we show that production of beta-amyloid and plaques in a mouse model of AD are reduced by overexpressing the NAD-dependent deacetylase SIRT1 in brain, and are increased by knocking out SIRT1 in brain. SIRT1 directly activates the transcription of the gene encoding the alpha-secretase, ADAM10. SIRT1 deacetylates and coactivates the retinoic acid receptor beta, a known regulator of ADAM10 transcription. ADAM10 activation by SIRT1 also induces the Notch pathway, which is known to repair neuronal damage in the brain. Our findings indicate SIRT1 activation is a viable strategy to combat AD and perhaps other neurodegenerative diseases.

Figure 1. SIRT1 levels regulate the pathology in a mouse Alzheimer's Disease model

(A) Cortical frozen sections (6 per mouse) of fixed brains from indicated mice (ages in months (m)) stained with Aβ specific 6E10 antibody. The graph on the right shows the quantification of the amyloid plaques according to their sizes in APPswe,PSEN1dE9 mice with endogenous SIRT1 (AD) and with a SIRT1 transgene (AD-Tg), n=5. Left shows examples of stained samples from these and wild type (wt) mice. Scale bar, 50 μm. * indicates p< 0.01 throughout in students t test. (B) Western blot analysis of SIRT1 expression in wild type (wt) and SIRT1 brain specific knockout mouse (BSKO) whole brains. Deletion of exon 4 resulted in the truncated inactive form of SIRT1 in BSKO mice. β-actin served as a loading control. (C) Survival curve of the percent viability of the AD, BSKO and AD-BSKO mice. n= 47 for AD-BSKO and BSKO and n= 45 for AD mice. Both BSKO and AD mice showed 100 % viability. (D) Cortical frozen sections of fixed brains from indicated mice stained with Aβ specific 6E10 antibody. The graph on the right shows the quantification of the amyloid plaques. Scale bar, 50 μm. n=5 (E) Frozen sections of fixed brains from indicated mice stained with GFAP (glial fibrillary acidic protein) antibody for gliosis. The graph on the right shows the quantification of the GFAP staining. Scale bar, 50 μm. (see also Figure S1)

Figure 2. SIRT1 improves the behavioral phenotype of Alzheimer's Disease mouse model

(A) Schema of the fear conditioning experiment. (B) The graph shows the total percentage of freezing (on day 2) of the wt, Tg, AD and AD-Tg mice of the indicated ages across the entire 5-min tone retrieval test. n=10-15 for each bar. (C) The graph shows the total percentage of freezing of the wt, BSKO, AD and AD-BSKO mice at 4 months of age across the entire 5-min tone retrieval test. n=10-15 (D) Morris water maze experiment (D-F). Escape latency (time to find the platform) was plotted for 7 days for 11, 8 and 4 months old of AD and AD-Tg mice. n= 6-12 (E) Escape latency plotted over 7 days for 11, 8 and 4 months old of Tg and wt mice. n=6-12 (F) Escape latency plotted over 7 days for 4 months old of wt, BSKO, AD and AD-BSKO mice. n=6-12 (see also Figure S1)

Figure 3. SIRT1 reduces Aβ levels and increases α-secretase activity

(A) Aβ 1-42 levels measured using ELISA from the whole brains of the indicated mice. n=6 for each bar. (B) α-secretase activity from the whole brains of indicated mice using an activity assay. n=6 AFU, arbitrary fluorescence units (C) β-secretase activity in the whole brains of indicated mice. n=6 (D) The scheme of the proteolytic processing of APP. Aβ 40/42, amyloid β peptide with 40 (Aβ 1-40) or 42 amino acid residues (Aβ 1-42); APP, amyloid precursor protein; AICD, APP intracellular domain; sAPPα, soluble APP after α-secretase cleavage (alpha-fragment); sAPPβ, soluble APP after β-secretase cleavage; C99, C-terminal fragment of APP of 99 amino acids after β-secretase cleavage; C83, C-terminal fragment of APP of 83 amino acids after α-secretase cleavage (α-CTF); P3, N-terminal fragment of C83 after α-secretase cleavage. (E) Western blotting of the whole brains of indicated mice of 6 months of age using 6E10 antibody. β-actin serves as the loading control. Quantification of the APPα/APP ratio is shown on the right as percent of control. (F) Western blotting of the whole brains of indicated mice of 6 months of age using APP C-terminal antibody. β-actin serves as the loading control. Quantification of the α-CTF is shown on the right as percent of control. (G) Western blotting of the whole brains of indicated mice of 6 months of age using APPβ antibody. β-actin serves as the loading control. Quantification of the APPβ fragment is shown on the right as percent of control. (H) Western blotting of the whole brains of indicated mice of 6 months of age using a different APP C-terminal antibody (see Experimental Procedures) from Figure 3F which can detect both α-CTF and β-CTF. β-actin serves as the loading control. Quantification of the α-CTF and β-CTF are shown on the right as percent of control. (see also Figure S2)

Figure 4. SIRT1 increases α-secretase levels by activating ADAM10

(A) ADAM10 RNA levels from whole brains of indicated mice quantified by q-PCR. The numbers on x-axis show the age of mice in months. n=5 for each bar. (B) Western blotting of whole brains from mice of indicated ages in months (m) and years (y) using ADAM10 antibody. β-actin serves as the loading control. ADAM10-P shows the unprocessed ADAM10 precursor protein. Quantification of ADAM10-P (top) and ADAM10 (bottom) are shown below each gel as percent of control. (C) Western blotting of whole brains from indicated mice using NICD (Notch intracellular domain) antibody. β-actin serves as the loading control. Quantification of NICD is shown below as percent of control. (D) HES1 and HES5 RNA levels quantified from whole brains of mice by q-PCR. The numbers on x-axis show the age of mice in months. n=5 (see also Figure S3)

Figure 5. SIRT1 suppression of Aβ production is mediated by ADAM10

(A) Chromatin-immunoprecipitation on whole brains of 6 months old SIRT1 Tg, AD-Tg and SIRT1^-/- mice with anti-SIRT1 antibody or IgG. The scheme illustrates the highest activity region of the ADAM10 promoter and the primers used for q-PCR in assay. q-PCR was performed using primers as indicated (see Experimental Procedures). SIRT1^-/-mice were in a mixed genetic background to allow recovery of viable adults (McBurney et al. 2003). (B) ADAM10 RNA levels quantified by q-PCR from the SIRT1^+/+ or SIRT1^-/- MEFs transfected with mSIRT1 or mSIRT1-HY and/or treated with retinoic acid (RA). (C) NICD levels determined by western blotting from the SIRT1^+/+ or SIRT1^-/- MEFs transfected with mSIRT1 or mSIRT1-HY and/or treated with retinoic acid (RA). β-actin serves as the loading control. Quantification of NICD levels is shown on the right as percent of control. (see also Figure S4)

Figure 6. SIRT1 deacetylates RARβ

(A) Luciferase reporter assays of MEFs overexpressing SIRT1 (mSIRT1), lacking SIRT1 (SIRT1^-/-) or controls (wt). Cells were transfected with luciferase reporter driven by the RARE (Retinoic Acid Receptor Element). 24 hr later, cells were incubated with RA (Retinoic Acid), LE-135 or DMSO for an additional 24 hr and assayed as described in Experimental Procedures. (B) Cell lysates from wild type (wt) and SIRT1^-/- (SIRT1^-/-) MEFs immunoprecipitated with normal rabbit serum (NRS) or anti-SIRT1 antibody and blotted with anti-SIRT1 and anti-RARβ antibodies. The two proteins are shown to interact at endogenous levels. (C) Cell lysates from wild type (wt), SIRT1-overxpressing (mSIRT1) MEFs, SIRT1^-/- MEFs and SIRT1^-/- MEFs overexpressing SIRT1 catalytic mutant HY were subjected to immunoprecipitation with anti-RARβ antibody or NRS. Immunoprecipitates were analyzed by western blotting with anti-RARβ, anti-RXR or anti-pan acetylated lysine (Ac-K) antibodies. Quantification of the acetylation levels were shown below. (see also Figure S5) (D) A model for the role of SIRT1 in suppressing Aβ production and activating notch pathway. SIRT1 deacetylates RARβ, causes RA to bind to RARβ-RXR heterodimer which binds to ADAM10 promoter. ADAM10 transcription and protein level is increased and APP processing is directed towards α-secretase pathway. This will lead to less β-secretase cleavage of APP resulting in less Aβ production, fewer β-amyloid plaques and less gliosis. ADAM10 activation by SIRT1 also induces the Notch pathway, which is known to repair neuronal damage in brain, providing neuroprotection.

Figure 7. SIRT1 suppresses Aβ production by activating ADAM10

(A) N2A cells stably overexpressing APPswe/PSEN1dE9 transgenes transfected with SIRT1 (mSIRT1). 24 hr after SIRT1 transfection, cells were transfected with 3 different ADAM10-shRNA vectors (1, 2 or 3). N2A cells not overexpressing SIRT1 were also transfected with 3 different ADAM10-shRNA vectors (1, 2 or 3) to silence ADAM10. Separately, SIRT1 was silenced by 3 different shRNAs (1, 2 or 3). 48 hr later, Aβ 1-42 concentration in the conditioned medium was assessed by ELISA (Experimental Procedures). N2A cells without APPswe,PSEN1dE9 transgenes showed no detectable Aβ 1-42 in this assay (not shown). (B) Western blotting of the extracts of these cells using anti-ADAM10 antibody. Quantification of the ADAM10-P (top) and ADAM10 (bottom) were shown on the right as percent of control. (C) Western blotting of the extracts of the cells using anti-NICD antibody. Quantification of the NICD are shown on the right as percent of control. (D) HES1 RNA levels were determined by qPCR from RNA extracted from cells. (E) N2A cells stably overexpressing APPswe/PSEN1dE9 were transfected with SIRT1 (mSIRT1). 24 hr later, cells were incubated with RA (Retinoic Acid), LE-135 or DMSO for an additional 24 hr and assayed by ELISA. (see also Figure S6 and S7)