Disrupted neural correlates of anesthesia and sleep reveal early circuit dysfunctions in Alzheimer models

Abstract

Dysregulated homeostasis of neural activity has been hypothesized to drive Alzheimer's disease (AD) pathogenesis. AD begins with a decades-long presymptomatic phase, but whether homeostatic mechanisms already begin failing during this silent phase is unknown. We show that before the onset of memory decline and sleep disturbances, familial AD (fAD) model mice display no deficits in CA1 mean firing rate (MFR) during active wakefulness. However, homeostatic down-regulation of CA1 MFR is disrupted during non-rapid eye movement (NREM) sleep and general anesthesia in fAD mouse models. The resultant hyperexcitability is attenuated by the mitochondrial dihydroorotate dehydrogenase (DHODH) enzyme inhibitor, which tunes MFR toward lower set-point values. Ex vivo fAD mutations impair downward MFR homeostasis, resulting in pathological MFR set points in response to anesthetic drug and inhibition blockade. Thus, firing rate dyshomeostasis of hippocampal circuits is masked during active wakefulness but surfaces during low-arousal brain states, representing an early failure of the silent disease stage.

Keywords: Alzheimer's disease; DHODH; NREM; calcium imaging; firing rate homeostasis; general anesthesia; hippocampus; hyperexcitability; single-unit recordings; sleep.

Graphical abstract

Figure 1

No deficits in the CA1 network activity and CA3-CA1 synaptic transmission during active wakefulness in APP/PS1 mice (A) Large-scale Ca²⁺ imaging of excitatory cells expressing GCaMP6f using wide-field, head-mounted miniaturized fluorescence microscope in freely behaving mice during exploration of familiar environment. (B) Relative fluorescence change traces of 10 cells randomly selected from the spatial locations depicted in Figure S3A, representing Ca²⁺ transients of cells detected in WT (left, blue traces) and APP/PS1 (right, red traces). Scale bars: 5 min; 10 Z scores. (C) Average mCaR distributions of CA1 neuronal populations in WT (6 mice, 3,846 cells) and APP/PS1 (6 mice, 3973 cells) during active wakefulness. (D–F) The median mCaRs of neurons (p = 0.94, D), number of N_as (p = 0.99, E), and total activity (mCaR^∗N_a, p = 0.82, F) were not different between WT and APP/PS1 mice (same data as in C, averaged per mouse) during active wakefulness. (G) Representative raster plots demonstrating CA1 single-unit spiking activity during active wakefulness in WT (blue) and APP/PS1 (red) mice. (H) MFRs of regularly spiking neurons during active wakefulness were not different (p = 0.82) between WT (2.01 ± 0.195; 4 mice, 103 single units) and APP/PS1 (1.88 ± 0.15; 4 mice, 98 single units) mice. (I) Dynamics of MFR change of regular spiking neurons across all active wakefulness episodes, concatenated (bold line represents the average MFR) in WT and APP/PS1 mice (same data as in H). Shaded area represents SEM. (J) No changes (p = 0.19) in CA3-CA1 basal synaptic transmission between awake WT (8 mice) and APP/PS1 (6 mice). Right: representative traces of fEPSPs of WT versus APP/PS1 evoked by 40, 50, and 60 μA stimulation. (K) No changes (p = 0.42) in short-term synaptic facilitation in response to 50-Hz burst between awake WT and APP/PS1 mice. Right: representative traces of fEPSPs evoked by five stimuli at 50 Hz. Unpaired Mann-Whitney nonparametric test (D–F and H). Two-way ANOVA with Sidak's multiple comparisons test (J and K) were used for the analysis. Error bars represent SEM. ns, non-significant. See also Figures S1–S3 and Videos S1 and S2.

Figure 2

Reduction of CA1 population activity during NREM sleep is impaired in APP/PS1 mice (A and B) An example of wake-dense (A) and sleep-dense (B) recordings from a WT mouse. Top: hypnograms, generated by manual brain state segregation. Brain states are color coded: yellow, active wake; gray, quiet wake; blue, NREM sleep; purple, REM sleep. Middle: Fourier transform-based LFP power spectrograms. Bottom: EMG traces (scale bars: 0.5 mV, 1 min). (C and D) Representative raster plots of CA1 Ca²⁺ event rates in active wake-dense episode (C) and NREM-dense episode (D) in a WT mouse (time interval corresponds to dashed rectangles in A and B). (E–H) Same as (A)–(D) for APP/PS1 mice. (I) Average mCaR distributions in CA1 of WT mice (5 mice) during active wake (3,319 cells) and NREM sleep (1,741 cells) states. (J) Average mCaR distributions in CA1 of APP/PS1 mice (4 mice) during active wake (2,595 cells) and NREM sleep (2,224 cells) states. (K) Relative change in total activity in WT versus APP/PS1 mice by NREM sleep in comparison to active wake state (the same data as in I and J). Two-way ANOVA with Sidak's multiple comparisons test (K, inter-group analysis); two-way ANOVA (K, intra-group analysis). ^∗∗p < 0.01. Error bars represent SEM. See also Figures S4A–S4C.

Figure 3

Local dysregulation of CA1 firing rates by NREM sleep precedes global SWA deterioration in APP/PS1 mice (A and B) Raw representative traces from a tetrode in WT (A) and APP/PS1 (B) mice during NREM sleep. Scale bars: 0.5 mV, 100 ms. (C) NREM sleep causes a reduction (p < 0.0001) in CA1 MFR of RS neurons from 2.01 ± 0.195 Hz in active wakefulness (AW) to 1.51 ± 0.12 Hz in WT mice (4 mice, 103 single units). (D) WT neurons whose MFRs were below the median firing rate during AW (1.4 Hz) showed no change in the gain factor in NREM sleep (+1.7%, confidence interval [CI] = [−0.8%, 3.1%]; p = 0.055), while cells whose MFRs were above the median showed a decrease in gain factor of −21% in NREM sleep (CI = [−27%, −11%], p < 0.0001) during sleep (same data as in C). (E) MFR of CA1 RS neurons is not different (p = 0.53) between AW (1.88 ± 0.15 Hz) and NREM sleep (1.91 ± 0.13 Hz) in APP/PS1 mice (4 mice, 98 units). (F) In APP/PS1 mice, cells whose MFRs were below the median firing rate during AW showed an increase of MFR in NREM sleep with median gain factor +12% (CI = [2.7%, 22%]; p < 0.0001), while cells whose MFRs were above the median showed no change in NREM sleep (−4%, CI = [−5.8%, 1.4%]; p = 0.22, same data as in E). (G) Frontal EEG spectra during NREM sleep in WT (n = 5) and APP/PS1 (n = 5) mice. Post hoc comparisons did not reveal significant genotype differences for any of the SWA frequency bins in NREM state (p = 0.68). (H) CA1 LFP spectra recorded by the same electrodes as single units in WT (n = 4) and APP/PS1 (n = 4) mice during NREM sleep. Slow-wave power during NREM is not significantly different between WT and APP/PS1 mice (p = 0.14). Non-parametric, paired, two-tailed Wilcoxon test (C and E), two-way ANOVA with Sidak's multiple comparisons test (G and H), one-sample Wilcoxon test comparing medians of the samples to 100% (D and F). ^∗∗∗∗p < 0.0001. Error bars represent SEM. See also Figures S4D–S4G, S5, and S6.

Figure 4

Loss of neuronal inhibition across anesthetic depth in APP/PS1 mice (A and B) Representative traces demonstrating raster plots of Ca²⁺ transients (top), raw LFP (middle), and corresponding scalograms based on wavelet transformation (bottom, log scale) for both WT (A) and APP/PS1 (B) under exploration in familiar environment (active wake), moderate anesthesia (moderate), and under deep anesthesia (deep). Imaging was performed in parallel to CA1 LFP recordings. (C and D) Average mCaR distribution of CA1 neuronal populations in WT (C, 5 mice per condition) and APP/PS1 (D, awake and moderate: 7 mice; deep: 4 mice) under all three conditions described in (A) and (B). Note higher mCaR and number of active neurons in anesthetized APP/PS1 versus WT mice. (E) Total activity is higher in APP/PS1 versus WT mice under both moderate and deep anesthesia (same data as in C and D). (F and G) Differences in temporal patterns of Ca²⁺ transients in CA1 network of WT versus APP/PS1 anesthetized mice. APP/PS1 mice display a larger number of cells that participate in network burst (F, p = 0.0001 for moderate [top] and deep [bottom]) and a larger number of spikes that constitute a network burst (G, p = 0.0001 for moderate [top] and deep [bottom]). Two-way-ANOVA with Sidak's multiple comparisons test (E, inter-group analysis), two-way ANOVA with Tukey's multiple comparison test (F and G), and one-way ANOVA with Dunnett's multiple comparisons test (E, intra-group analysis) were used for the analysis. ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001; ^nsp > 0.05. Error bars represent SEM. See also Figures S7–S9 and Videos S3 and S4.

Figure 5

CA1 is hyperexcitable under anesthesia in different fAD models (A) Representative traces of raw CA1 LFP recordings during awake behavior (left) and moderate isoflurane anesthesia (right). Notice the appearance of abnormal spikes under anesthesia. LFP recordings were performed across different mouse fAD models: APP/PS1, 5XFAD, and APP-KI. Scale bars: 10 s, 1 mV. (B) Representative epileptiform spikes detected during moderate anesthesia across distinct fAD models. Scale bars: 50 ms, 1 mV. (C) On average, all three fAD models displayed higher frequency of epileptiform spikes compared to WT across isoflurane-induced moderate (WT: 17 mice, APP/PS1: 18 mice, FADX5: 5 mice, APP-KI: 6 mice) and deep (WT: 22 mice, APP/PS1: 22 mice, FADX5: 8 mice, APP-KI: 9 mice) anesthesia. Two-way-ANOVA with Dunnett's multiple comparisons test (C) was used for the analysis. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗∗p < 0.0001. Error bars represent SEM. See also Figures S7E and S7F and Videos S5 and S6.

Figure 6

Dysregulation of downward MFR homeostasis by fAD mutations in hippocampal networks ex vivo (A and B) Raster plots from MEA recordings showing activity of the same channels in baseline, after 2 and 24 h of isoflurane (1%, 40mL/min) application in WT (A, blue, 97 channels) and APP/PS1 (B, red, 99 channels) cultures. (C) The MFR of the network was stably reduced after isoflurane (ISO) application to 42.0% ± 0.9% of baseline in WT cultures (n = 8 experiments, 633 channels). (D) The MFR of the network was decreased and rapidly compensated to 112.5% ± 8.7% (p > 0.9) of baseline after ISO application in APP/PS1 cultures (n = 5 experiments, 384 channels). (E) Summary of isoflurane (24 h) effect on MFRs in WT and APP/PS1 cultures (same data as C and D). (F and G) A typical MFR homeostatic response to chronic inactivity induced by a GABA_B receptor agonist baclofen (10 μM) in WT (F) and APP/PS1 (G) networks. MFR was homeostatically compensated during the first day of the perturbation to the MFR set-point level in WT (7 experiments, 394 channels, p = 0.76) and APP/PS1 (n = 6, p = 0.64, 377 channels) neurons. (H) Summary of baclofen (48 h) effect on MFRs in WT and APP/PS1 cultures (same data as F and G). (I and J) MFR was renormalized to the baseline level following 2 days of gabazine (GBZ, 30 μM) application in WT neurons (p = 0.31, n = 6 experiments, 398 channels), but MFR renormalization to GBZ was disrupted in APP/PS1 neurons (n = 4 experiments, 270 channels). (K) Summary of gabazine (48 h) effect on MFRs in WT and APP/PS1 cultures (same data as I and J). Paired two-tailed t test between the baseline and the perturbation (C, D, F, G, I, and J) and Mann-Whitney nonparametric two-tailed test (E, H, and K) were used for the analysis. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001. The error bars represent SEM. See also Figures S10A and S10B.

Figure 7

Teriflunomide reduces CA1 hyperexcitability in anesthetized APP/PS1 mice (A) Dose-response effect of TERI on the fEPSP amplitude in CA3-CA1 synaptic connections in hippocampal slices of WT (8 mice, 13 slices) and APP/PS1 (8 mice, 13 slices). IC₅₀: 17.6 ± 1 μM in WT, 19.2 ± 1 μM in APP/PS1. Right: representative fEPSP traces before (dark blue/red) and after 25 μM TERI (light blue/red). Scale bars: 0.5 mV, 20 ms. (B) Orotate concentration is not different between hippocampi of anesthetized WT and APP/PS1 mice (6 WT and 6 APP/PS1 hippocampi, p = 0.92). (C) Representative traces of CA1 LFP recordings depict epileptiform spikes in the baseline (Bsl) and after TERI i.c.v. injection in anesthetized (1.5% isoflurane) APP/PS1 mice. Scale bars: 1 mV, 1 min. (D) TERI i.c.v. injection caused a significant reduction in the frequency of epileptiform spikes from 7.6 ± 1.4 to 3.8 ± 0.95 per min (n = 9 mice, p = 0.004). (E) VEH i.c.v. injections did not affect the frequency of epileptiform spikes (n = 7 mice, p = 0.58). (F) On average, the TERI-treated group displayed 50.7% ± 7.2% reduction (p = 0.0003) in the frequency of epileptiform spikes (the same data as in D and E). Wilcoxon matched pairs (D and E) and Mann-Whitney nonparametric two-tailed test (B and F) were used for the analysis. ^∗∗p < 0.01; ^∗∗∗p < 0.001. The error bars represent SEM. See also Figures S10C and S10D.