SIRT3 mediates hippocampal synaptic adaptations to intermittent fasting and ameliorates deficits in APP mutant mice

Abstract

Intermittent food deprivation (fasting, IF) improves mood and cognition and protects neurons against excitotoxic degeneration in animal models of epilepsy and Alzheimer's disease (AD). The mechanisms by which neuronal networks adapt to IF and how such adaptations impact neuropathological processes are unknown. We show that hippocampal neuronal networks adapt to IF by enhancing GABAergic tone, which is associated with reduced anxiety-like behaviors and improved hippocampus-dependent memory. These neuronal network and behavioral adaptations require the mitochondrial protein deacetylase SIRT3 as they are abolished in SIRT3-deficient mice and wild type mice in which SIRT3 is selectively depleted from hippocampal neurons. In the App^NL-G-F mouse model of AD, IF reduces neuronal network hyperexcitability and ameliorates deficits in hippocampal synaptic plasticity in a SIRT3-dependent manner. These findings demonstrate a role for a mitochondrial protein deacetylase in hippocampal neurons in behavioral and GABAergic synaptic adaptations to IF.

Fig. 1

Behavioral and neuronal network adaptations to IF involve enhanced GABAergic activity. a–d WT mice were maintained on either an ad libitum diet (AL) or a on IF for 1 month. Mice in each diet group were then randomly assigned to behavioral testing in the elevated plus maze and open field after either a 24 h period of food deprivation (FD) or with no food deprivation (NFD). The images at the left show tracings of the walking paths of mice in the AL FD and IF FD groups in the elevated plus maze (a, b) and open field (c, d). The numbers of mice in each group were: AL NFD, 13; AL FD, 15; IF NFD, 12; IF FD, 15. *P < 0.05. e Examples of images of c-Fos immunoreactivity in neurons in the amygdala, hippocampus and CA1 of mice in the indicated groups. The dashed lines in the images of the amygdala demarcate the boundaries of the central nucleus (upper circle) and the basolateral amygdala (lower curved lines). The white frames in the images of the hippocampus demarcate the location of the images of region CA1. f Numbers of c-Fos immunoreactive (cFosIR) neurons in the amygdala and hippocampal CA1 in mice in the four indicated groups (5 mice/group). *P < 0.05, **P < 0.01. g Whole-cell recordings of miniature postsynaptic currents (mIPSCs) in CA1 neurons in hippocampal slices from mice in the four indicated groups. h Amplitudes of mIPSCs (left) and a plot of cumulative event probability as a function of mIPSC amplitude (right) in CA1 neurons of mice in the four different treatment groups (data are from 15 neurons in slices from 5 mice/group). *P < 0.05 compared to the values for each of the other three groups. i mIPSC frequency (left) and a plot of cumulative event probability as a function of mIPSC inter-event interval (right) in CA1 neurons of mice in the four different treatment groups (5 mice/group). j Serum β-hydroxybutyrate (BHB) concentrations in mice in the indicated groups. ***P < 0.001. All error bars are the SEM. All statistical comparisons used ANOVA and Newman–Keuls post hoc tests. Source data are provided in supplemental Source Data file

Fig. 2

SIRT3 is required for behavioral and synaptic adaptations to IF. a Immunoblots show relative levels of SIRT3 protein in hippocampus of mice that had were either fed AL or maintained on alternate day food deprivation (IF) for 1 week or 1 month. The upper blot is samples from mice killed on the second day after diet initiation (at the end of the first 24 h FD period for mice in the IF group). *P < 0.05, ***P < 0.001 compared to the AL NFD and 24 h FD groups (11 mice/group). b, c Time spent (left) and distance traveled (right) in the open arms of the elevated maze (b) and in the center zone of the open field (c) for mice that had either been fed AL or IF for 1 month (12 mice/group). All tests were performed in mice that had been deprived of food for 24 h (FD). *P < 0.05. d Whole-cell recordings of mIPSCs in CA1 neurons in hippocampal slices from mice in the four indicated groups. e mIPSC amplitudes (left) and frequencies (right) of mIPSCs (15 neurons in slices from 5 mice/group). *P < 0.05. f–h fEPSP recordings at CA1 synapses in hippocampal slices from Sirt3^+/+ and Sirt3^–/– mice that had been maintained for 4 months on either AL or IF diets. f shows basal synaptic transmission with the amplitude of the fiber volley plotted against fEPSP slopes (input/output curves). g shows the results of LTP analysis. Panel h shows quantification of LTP magnitude. *P < 0.01 compared to the values for each of other three groups (10 slices from 5 mice/group). i Results of Y-maze testing. j, k Results of water maze testing. Group difference, F (3, 27) = 3.762, P = 0.022. k shows data for escape latencies on testing day 1 trial 4, and testing day 2 trial 1 (left), and latency ratios (escape latency on day 2 trial 1 divided by the escape latency on day 1 trial 4) (right). 8 mice per group, *P < 0.05, **P < 0.01. All error bars are the SEM. All statistical comparisons used ANOVA and Newman–Keuls post hoc tests. Source data are provided in supplemental Source Data file

Fig. 3

IF prevents seizures and ameliorates cognitive and synaptic deficits in App^NL-G-Fmice. a Survival curves for WT and App^NL-G-F mice maintained on either AL or IF (fasting 2 days/week) diets beginning at 52 weeks of age (WT AL, 10 mice; WT IF, 12 mice; AL App^NL-G-F, 19 mice; AD IF, 11 mice. No WT mice or App^NL-G-F mice on the IF diet died, whereas 6 mice App^NL-G-F mice on the AL diet died. b Age of onset of first seizure-like behavior for mice in the four groups. c Immunoblot analysis of SIRT3 protein levels in hippocampus of App^NL-G-F mice in the AL diet group that either did or did not exhibit seizures. Densitometric analysis of SIRT3 band intensity (normalized to actin band; 4 mice with observed seizures, 7 mice without observed seizures). ***P < 0.001. d Images of Aβ immunoreactivity in hippocampus of App^NL-G-F mice that had been on either AL or IF diets for 8 months (scale bar, 100 µm). The graph shows Aβ load values (15 sections from 5 mice/group). e Results of Y-maze testing. Group difference, F (3, 32) = 5.538, P = 0.0035. *P < 0.05, WT-AL compared to values for WT-IF; ^$P < 0.05, App^NL-G-F -AL compared to value for App^NL-G-F -IF; ^#P < 0.05, App^NL-G-F -AL compared to each of the other three groups. f, g Escape latencies on testing day 1 trial 4, and testing day 2 trial 1. *P < 0.05 compared to the WT IF group; ^#P < 0.05, App^NL-G-F-AL compared to the other three groups (8–11 mice/group). h–j Field EPSP (fEPSP) recordings at CA1 synapses. Panel h shows basal synaptic transmission. i shows the results of LTP analysis. j shows quantification of LTP magnitude (10 slices from 5 mice/group). **P < 0.01. k c-Fos immunoreactivity in hippocampal CA1 neurons of an App^NL-G-F IF mouse one month after administration into the dorsal hippocampus of control shRNA lentivirus in the left hippocampus (left panels) and a lentivirus with an shRNA targeting Sirt3 RNA in the right hippocampus (right panels). (Scale bar, upper 250 µm, lower 100 µm). ***P < 0.001. All error bars are the SEM. Student’s t test was used for analyses of data in c, d and k. ANOVA and Newman–Keuls post-hoc tests were used for analyses of data in e, f, g, i and j. Source data for all graphs in this figure are provided in supplemental Source Data file

Fig. 4

SIRT3 is required for GABAergic synaptic scaling. a Immunoblots (left) showing relative levels of SIRT3 protein in cultured WT hippocampal neurons exposed to the indicated treatments for 24 h; 100 μM picrotoxin (PTX) for 12 h; 100 μM PTX for 24 h; 10 μM EUK-134 plus 100 μM PTX for 24 h. (*P < 0.05). b Whole-cell recordings of mIPSCs in cultured hippocampal neurons from Sirt3^+/+ and Sirt3^–/– mice that had been treated with vehicle or 100 μM PTX for 24 h. c Amplitudes of mIPSCs (left) and a plot of cumulative event probability as a function of mIPSC amplitude (right) in Sirt3^+/+ and Sirt3^–/– neurons treated for 24 h with vehicle or PTX (15 neurons/group). d mIPSC frequency (left) and a plot of cumulative event probability as a function of mIPSC inter-event interval (right) in Sirt3^+/+ and Sirt3^–/– neurons treated for 24 h with vehicle or PTX (15 neurons/group). *P < 0.05 compared to the values for each of the other 3 groups. e Whole-cell recordings of miniature postsynaptic currents (mIPSCs) in cultured hippocampal neurons from Sirt3^–/– mice that had been infected with AAV vectors to overexpress either SIRT3 and GFP, or GFP alone; 6 days after infection, cultures were treated with vehicle or 100 μM PTX for 24 h. f Amplitudes (left) and frequencies (right) of mIPSCs in Sirt3^–/– neurons in the four different groups (15 neurons/group). *P < 0.05 compared to the values for each of the other 3 groups. g Whole-cell recordings of miniature postsynaptic currents (mIPSCs) in cultured hippocampal neurons from Sirt3^+/+ mice that had been treated for 24 h with vehicle, 100 μM PTX, 0.3 μM paraquat (PQ) plus PTX, 1.0 μM PQ plus PTX, or 0.3 μM PQ plus vehicle. h mIPSC amplitudes (left) and frequencies (right) in Sirt3^+/+ neurons in the indicated treatment groups (24 h treatment; 12 neurons/group). *P < 0.05. i Results of measurements of mIPSC amplitudes in Sirt3^-/- neurons in the indicated treatment groups (24 h treatment; 12 neurons/group). *P < 0.05. All error bars are the SEM. All statistical comparisons for data in this figure were performed using ANOVA and Newman–Keuls post hoc tests for pairwise comparisons. Source data for all graphs in this figure are provided in supplemental Source Data file