6 months old APP-GAD-Cre mice exhibited impaired NREM sleep, decreased delta power, and increased sleep fragmentation

To investigate the role of the GABAergic interneurons in the context of AD, we generated a mouse model where the general GABAergic (glutamic acid decarboxylase, or GAD-expressing) neuronal population could be targeted in the presence of amyloidosis. We crossed heterozygous APP/NTG mice [13] with homozygous GAD-Cre animals [30], thus generating APP-GAD-Cre mice, and NTG-GAD-Cre mice that served as controls (Fig. 1A). Previous literature suggested that APP mice exhibited sleep disruptions early in the disease progression preceding the cognitive deficits and AD neuropathology [16]. Thus, we first examined whether sleep disruptions were present in 6-month-old APP-GAD-Cre mice when compared to NTG-GAD-Cre controls, using an EEG/EMG telemetry system (Fig. 1B). Mice exhibited NREM and REM sleep as well as awake states while maintained individually in home cages (Fig. 1C).

Fig. 1

Impaired NREM sleep, decreased delta power, and increased sleep fragmentation in APP mice at 6 months of age. (A) Left, diagram showing viral injection strategy to target GABAergic neurons with mCherry or ChR2-mCherry. Right, representative photomicrograph showing GABAergic neurons expressing ChR2-mCherry with DAPI (Gray). The dashed line shows the approximate location of the cannula track. Scale bar, 1 mm. (B) Left, diagram showing placement of EEG/EMG implant. Right, placement of EEG, EMG electrodes, and fiber-optic cannula on the skull. EMG electrodes were placed within the nuchal musculature. (C) Representative EEG and EMG traces during NREM, REM and wake states. (D and E) Overall 24-hour sleep pattern and sleep architecture of the NTG (D) and APP (E) mice. (F) Averaged time spent in each sleep-wake cycle stage (NREM, REM and wake) during 24-hour, 12-hour dark phase and 12-hour light phase of NTG and APP mice. (G) Time course of the changes in NREM sleep in NTG and APP mice. (H-J) Relative power spectral density of NREM sleep during 24-hour, 12-hour llight phase and 12-hour dark phase of NTG and APP mice. (K) The average EEG power density in the delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), sigma (12–16 Hz) and beta (16–24 Hz) bands during NREM sleep during 24-hour, 12-hour light phase and 12-hour dark phase of NTG and APP mice. (L and M) Average length (L) and bouts count (M) in each sleep-wake cycle stage (NREM, REM and wake) during 24-hour, 12-hour dark phase and 12-hour light phase of NTG and APP mice. All data are expressed as means ± standard error. The number of mice examined: NTG = 11 mice; APP = 12 mice. *P < 0.05, **P < 0.01, and ***P < 0.001. n.s. not significant

Analysis of sleep architectures revealed that both NTG-GAD-Cre (NTG) and APP-GAD-Cre (APP) mice were nocturnal and had a sleep pattern with the main sleep period identified during the light phase, while also spending a considerable amount of time sleeping during the dark phase. In general, APP mice exhibited disrupted sleep patterns, with the hypnogram showing interruptions in NREM sleep (Fig. 1D, E). APP mice spent more time awake and less time in NREM sleep during the 24-hour period (Fig. 1F). Furthermore, APP mice spent significantly more time awake and less time in NREM sleep during both light and dark periods, indicating poor sleep quality compared to that of NTG controls (Fig. 1F, G). We did not detect significant differences in time spent in REM sleep (Fig. 1F). We then performed Fourier transform analysis to determine the power-frequency relationships. APP mice exhibited lower delta power (0.5-4 Hz) during NREM sleep when assessed across the 24-hour period, during the light period or during the dark period. A decrease in delta power was accompanied by an increase in sigma and beta power in APP mice (Fig. 1H-K), suggesting hyperactivity within the circuit. Analyses of powers during REM sleep (Supplemental Fig. 11A-C, G) and wakefulness (Supplemental Fig. 1D-F, H), further confirmed a shift to higher frequencies, which is consistent with previous sleep research using AD mouse models [34]. Moreover, we discovered increased sleep fragmentation in APP mice with a significant reduction in NREM bout length and bout number comparable to that in NTG mice (Fig. 1L, M).

Overall, these results demonstrate that APP mice exhibit deficits in NREM sleep, erosion of delta power and decreases in individual bout length at 6 months of age. These data are consistent with increases in circuit hyperactivity evident due to increases in powers of higher frequency bands.

Optogenetic stimulation of GABAergic interneurons ameliorated sleep deficits in APP-GAD-Cre mice

Previous studies reported that GABAergic interneurons are important for sleep generation and regulation, particularly for SWA during NREM sleep [35, 36]. Excessive Aβ induces neuronal dysfunction and downregulates the activity of GABAergic interneurons, which eventually leads to circuit failure in AD [37]. We previously reported deficits in inhibitory elements of the circuit, specifically decreases in GABA, GABAA and GABAB receptors in young APP mice. Furthermore, GABA administration rescued slow wave power in APP mice [14]. Thus, in this study, we determined whether optogenetic stimulation of cortical GABAergic interneurons at the slow wave frequency could rescue sleep deficits and restore NREM delta power in APP mice.

We used Cre/Lox recombination technology to express ChannelRhodopsin-2 (ChR2) exclusively in GABAergic interneurons in the anterior cortex of APP-GAD-Cre (APP) and NTG-GAD-Cre (NTG) mice (Fig. 1A). We optogenetically targeted interneurons in the anterior cortex because the anterior cortex is known to be the origin of endogenous slow oscillations [38]. APP mice exhibited sleep rescue during optogenetic stimulation of ChR2 (APP-ChR2-opto), compared to those prior to stimulation (APP-ChR2) (Fig. 2A, B). Optogenetic stimulation of GABAergic interneurons increased the amount of time APP mice spent in NREM sleep. This phenomenon was observed in both light and dark phases (Fig. 2C, D). APP mice spent less time awake during stimulation. REM sleep was not affected significantly by stimulation (Fig. 2C). Fourier transform analysis revealed an increase in delta power (0.5-4 Hz) during stimulation compared to baseline (Fig. 2E-H). SWA (0.5-1 Hz), the major restorative feature of NREM sleep, was significantly improved during stimulation across the 24-hour period, during light and dark periods (Fig. 2I). Additionally, optogenetic stimulation ameliorated sleep fragmentation as indicated by prolonged NREM bout lengths particularly in the light phase (Fig. 2J), while maintaining the bout numbers (Fig. 2K).

Fig. 2

Optogenetic stimulation of GABAergic neurons increased SWA and delta power, rescued NREM sleep and promoted sleep integrity in APP mice. (A and B) Overall 24-hour sleep pattern and sleep architecture of the APP mice before (A) and during (B) stimulation on ChR2. (C) Averaged time spent in each sleep-wake cycle stage (NREM, REM and wake) during 24-hour, 12-hour dark phase and 12-hour light phase of APP mice before and during stimulation on ChR2. (D) Time course of the changes in NREM sleep in APP mice before (APP-ChR2) and during stimulation (APP-ChR2-opto) on ChR2. (E-G) Relative power spectral density of NREM sleep during 24-hour, 12-hour dark phase and 12-hour light phase of APP mice before and during stimulation on ChR2. (H) The average EEG power density in the delta (0.5–4 Hz) and theta (4–8 Hz) bands during NREM sleep during 24-hour, 12-hour dark phase and 12-hour light phase across conditions. (I) The average SWA (0.5–1 Hz) power during NREM sleep during 24-hour, 12-hour dark phase and 12-hour light phase across conditions. (J and K) Average length (J) and bouts count (K) in each sleep-wake cycle stage (NREM, REM and wake) during 24-hour, 12-hour dark phase and 12-hour light phase across conditions. All data are expressed as means ± standard error. The number of mice examined: APP = 6 mice/group. *P < 0.05, **P < 0.01. n.s. not significant

In contrast, light stimulation of mCherry in absence of ChR2 failed to significantly affect sleep architectures in both NTG (Supplemental Fig. 2) and APP mice (Supplemental Fig. 3). No significant differences in NREM, REM sleep and wake duration were found in mCherry groups during light and dark phases (Supplemental Fig. 2A, B; Supplemental Fig. 3A, B). In addition, Fourier transform analysis revealed that the power spectra density plots were comparable (Supplemental Fig. 2C-G; Supplemental Fig. 3C-G). Similarly, there were no significant differences in sleep/wake durations and power density between APP-mCherry or APP-ChR2 groups in absence of light stimulation (Supplemental Fig. 4A-C). Thus, light stimulation of ChR2 resulting in optogenetic activation of GABAergic interneurons was necessary for NREM sleep rescue.

Altogether our data suggest that optogenetic stimulation of GABAergic interneurons rescued sleep deficits, and improved delta power as well as SWA in APP mice.

Chronic optogenetic stimulation of GABAergic interneurons reduced amyloid plaque deposition in APP mice

Since sleep disruptions were shown to facilitate Aβ accumulations [39, 40], we next determined whether restoring sleep deficits could slow Aβ deposition. To achieve this, 6-month-old APP mice were treated with chronic optogenetic stimulation of ChR2 (ChR2-opto) targeted to cortical GABAergic neurons. Since APP mice showed deficits in NREM sleep during the day and night, Optogenetic treatment was performed continuously for 4 weeks. After treatment, a cranial window was installed over the right posterior cortex, contralateral to the stimulation site (Fig. 3A, B). Since slow oscillations activated in the anterior cortex propagate to the contralateral side, we chose to stimulate ChR2 in the anterior left cortex and image amyloid deposition in the posterior right. That way, we were able to avoid the confound of imaging the neurons that were directly activated by optogenetics. Amyloid plaques were labeled with Methoxy-X04 and monitored using multiphoton microscopy in anesthetized mice (Fig. 3C).

Fig. 3

Effect of chronic optogenetic stimulation of GABAergic neurons on amyloid plaque deposition in APP mice. (A and B) Experimental design. After AAV infusion and cannula installation, mice received 1-month of continuous optogenetic stimulation. Multiphoton imaging (B) was performed after treatment. (C) Representative multiphoton images of Methoxy-X04 positive amyloid plaques (Cyan) and blood vessels (Red) in APP mice in absence of stimulation (APP-mCherry-no opto), in presence of stimulation of mCherry (APP-mCherry-opto) and during stimulation of ChR2 (APP-ChR2-opto). (D) Amyloid plaque number across conditions. (E) Amyloid plaque size across conditions. (F) Amyloid plaque burden across conditions. (G) Representative images of 6E10, 82E1 and Methoxy-X04 positive amyloid plaques in cortex and hippocampus within postmortem sections. (H-J) 6E10 (H), 82E1 (I) and Methoxy-X04 (J) positive amyloid plaque burden across conditions. All data are expressed as means ± standard error. The number of mice examined: n = 6–7 mice/group. *P < 0.05, **P < 0.01. n.s. not significant. Scale bars: 50 μm

Amyloid plaque number was significantly lower after optogenetic stimulation in the ChR2-opto group when compared to other conditions in APP mice (Fig. 3D). Plaque size was comparable in all three groups (Fig. 3E). Amyloid plaque burden, which considers the number and size of plaques, was also significantly lower in optogenetically treated APP mice (Fig. 3F). These data demonstrated that chronic optogenetic stimulation of GABAergic neurons resulted in lower amyloid deposition in APP mice compared to APP-mCherry-opto control mice. Moreover, amyloid burden did not differ significantly between APP-mCherry-opto and APP-mCherry-no opto groups, demonstrating that the blue light used here did not result in detectable toxicity (Fig. 3D-F).

Since multiphoton microscopy allowed monitoring amyloid under the cranial window in a small cortical region, we verified amyloid plaque data in a greater cortical region as well as hippocampus, a deeper brain region inaccessible by multiphoton microscopy. We performed immunostaining on post-mortem brain tissue of the same APP mice chronically treated with optogenetic activation of ChR2 or mCherry as described above. Brain sections were immunostained with anti-amyloid β antibody 6E10 reactive to amino acid residue 1–16 of amyloid β, and 82E1 which recognized the N-terminus of Aβ but not full-length APP. Immunostaining was compared with Methoxy-X04. While Methoxy-X04 labeled dense cores of amyloid plaques, 6E10 and 82E1 decorated the periphery as well as the cores (Fig. 3G). Chronic optogenetic stimulation resulted in lower amyloid plaque burden both in the cortex and the hippocampus. 6E10 and 82E1 immunoreactivity revealed that amyloid plaque burden was significantly lower as a result of chronic optogenetic stimulation in the ChR2 group compared to the control mCherry group (Fig. 3G-I). Similarly, the Methoxy-X04 data revealed lower plaque burden after chronic optogenetic stimulation (Fig. 3G, J). These post-mortem data were all consistent with our in vivo findings using multiphoton microscopy. Finally, APP mice expressing ChR2 in absence of optogenetic stimulation (ChR2-no opto) failed to show lower plaque deposition (Supplemental Fig. 5A-C).

Taken together, chronic optogenetic stimulation of GABAergic neurons ameliorated amyloid plaque deposition in APP mice.

Chronic optogenetic stimulation of GABAergic interneurons improved neuronal calcium homeostasis in APP mice

In addition to depositing amyloid plaques, APP mice contain a small cortical neuronal population that is vulnerable to amyloid β-dependent calcium dysregulation resulting in calcium elevations [31]. This results in calcium overload within neuronal processes, neurites [41]. To determine whether chronic optogenetic stimulation of GABAergic neurons improved neuronal calcium homeostasis, we expressed the radiometric calcium sensor, Yellow Cameleon 3.6 (YC3.6), in the cortex of APP mice and NTG controls (Fig. 3A A, B). YC3.6 is a genetically encoded calcium sensor containing YFP and CFP [42]. A YFP/CFP ratio greater than two standard deviations above the NTG mean, 1.79, constituted calcium overload (Fig. 4C, red box, D). This value translated into an intracellular calcium concentration greater than 235nM [31]. Neurites exhibiting calcium overload are shown (Fig. 4B, red neurites, white arrowheads). Therefore, restoration of neuronal calcium levels would serve as a functional indicator of treatment efficacy [43].

Fig. 4

Effect of chronic optogenetic stimulation of GABAergic neurons on Calcium overload in NTG and APP mice. (A) Diagram of AAV-CBA-YC3.6 construct. (B) Representative multiphoton images pseudocolored according to the intraneuronal calcium concentration in NTG and APP mice in absence of optogenetic stimulation (mCherry-no opto), in presence of optogenetic stimulation of mCherry (mCherry-opto) and optogenetic stimulation of ChR2 (ChR2-opto). Neuronal processes exhibiting calcium overload are shown in red (see arrowheads). (C) Histogram showing the distribution of YFP/CFP ratios in neurites expressing YC 3.6 of APP mice. (D) The percentage of neurites exhibiting calcium overload across conditions in APP mice. (E) Histogram showing the distribution of YFP/CFP ratios in neurites expressing YC 3.6 of NTG mice. (F) The percentage of neurites exhibiting calcium overload across conditions in NTG mice. All data are expressed as means ± standard error. Neuronal calcium overload was defined as a YFP/CFP ratio larger than 2 standard deviations above the average YFP/CFP ratio in the neurons of NTG mice. The ratio of YFP/CFP > 1.73 was considered calcium overload. The number of mice examined: n = 5–7 mice/group. ***P < 0.001. n.s. not significant. Scale bar: 50 μm

As anticipated, APP mice expressing mCherry contained more neurites with calcium overload compared to healthy NTG controls (Fig. 4B, top left and middle panels, red neurites, white arrowheads, Fig. 4B bottom left and middle panels). Optogenetic treatment of ChR2 led to a lower percentage of neurites with calcium overload in APP mice (Fig. 4B, upper right panel). Histogram analysis revealed the presence of a small yet vulnerable neuronal population with calcium elevations, exhibiting YFP/CFP ratio of > 1.79 in APP mice with mCherry (Fig. 4C, red box). Compared to mCherry group, optogenetic stimulation of ChR2-expressing GABAergic interneurons resulted in a lower percentage of neurites with calcium overload (Fig. 4C, D). To control blue light toxicity, we examined APP mice expressing mCherry with and without light treatment. APP-mCherry-opto mice exhibited neuronal calcium overload comparable to APP-mCherry-no opto mice, demonstrating that blue light had no major toxic impact on neuronal calcium homeostasis in APP mice (Fig. 4B-D).

In addition, optogenetic stimulation of ChR2-expressing GABAergic interneurons in healthy NTG mice did not significantly alter the percentage of neurites with calcium overload compared to the NTG-mCherry-opto group (Fig. 4B, E, F). Thus, optogenetic stimulation of GABAergic interneurons had no significant impact on calcium homeostasis in healthy NTG mice. Similarly, NTG-mCherry-opto and NTG-mCherry-no opto groups had a comparable percentage of neurites with calcium overload (Fig. 4B, E, F).

Thus, chronic optogenetic stimulation of GABAergic neurons decreased neuronal calcium overload and restored neuronal calcium homeostasis in APP mice.

Chronic optogenetic stimulation of GABAergic interneurons improved memory consolidation in APP mice

We next assessed whether optogenetic stimulation of GABAergic neurons affected memory. We subjected a different cohort of NTG and APP mice to a battery of behavioral tests after 2 weeks of optogenetic treatment of ChR2 or light treatment of mCherry. The mice were first tested in an open field to determine whether optogenetic stimulation altered the locomotor activity or induced any anxiety-like behaviors (Fig. 5A). The total distance traveled was comparable across conditions (Fig. 5B). Time spent in the center was similar across conditions, as was time spent in the border of open field (Fig. 5C). Thus, APP mice did not exhibit any locomotor impairments, nor anxiety-like behaviors.

Fig. 5

Chronic optogenetic stimulation of GABAergic neurons improved memory performance in APP mice. (A) Schematic diagram of the open field test. (B) The total distance traveled during open field test across conditions. (C) Time spent exploring the center or border zones. (D) Schematic diagram of fear conditioning test. (E) Percentage of time spent freezing during fear recall. (F) Schematic diagram of the Y-maze spontaneous alternation is shown. (G) Percentage of spontaneous alternation (% alternation) activities across conditions. (H) The number of total entries. All data are expressed as means ± standard error. The number of mice examined: n = 10 mice/group. *P < 0.05. n.s. not significant

We performed contextual fear conditioning to test the hippocampus-dependent associative emotional memory in the animals [44]. The APP mice were reported to exhibit cognitive deficits in contextual memory as early as 4–6 months of age [45]. On the first day, animals were subjected to electric foot shocks. After the fear acquisition, animals were allowed to sleep. The next day, their sleep-dependent memory consolidation was tested in the same chamber yet without foot shocks (Fig. 5D). The percentage of time spent freezing was used as an index of fear memory. As expected, the APP-mCherry-opto mice showed impairments in memory consolidation with less time spent freezing compared to the NTG-mCherry-opto group (Fig. 5E). Interestingly, the freezing levels were significantly higher in APP-ChR2-opto group, indicating that optogenetic stimulation of GABAergic neurons restored memory consolidation in APP mice (Fig. 5E).

The amygdala is required for the acquisition and expression of learned fear responses in mice, and the hippocampus-amygdala circuit is engaged during contextual fear conditioning tests [44, 46]. Therefore, we examined plaque deposition in the amygdala after treatment. Optogenetic stimulation of GABAergic neurons failed to reduce plaque deposition in the amygdala of APP mice (Supplemental Fig. 6A-D). This might be explained by the fact that the amygdala had lower plaque load compared to the cortex and hippocampus during the ages tested. In addition, we tested the animals in the Y-maze test (Fig. 5F) to measure spatial working memory, as a spontaneous alternation score [47]. We observed no impairments in APP mice at this age. Chronic optogenetic treatment did not alter the score significantly (Fig. 5G, H).

Overall, these data demonstrate that optogenetic stimulation of GABAergic interneurons rescued sleep-dependent memory consolidation in APP mice without altering sleep-independent working memory. Furthermore, optogenetic treatment did not significantly affect the locomotion or anxiety levels of the APP mice.

Chronic optogenetic stimulation of GABAergic interneurons induced alterations of microglial morphological features and phagocytic ability

Accumulating evidence suggests that microglia are regulated by sleep and play a role in AD pathology [25, 27, 29]. Our data demonstrated that sleep deficits were rescued by optogenetic stimulation of GABAergic neurons in APP mice. Thus, we investigated the effect of 2-week long optogenetic treatment on microglia. First, we immunostained the post-mortem brain sections of chronically treated animals with antibodies against the microglial marker Iba1. Amyloid plaques were labeled with Methoxy-XO4. Images were acquired using a confocal microscope with a 40X objective at identical settings (Fig. 6A). The number of reactive microglia was significantly elevated in mice in APP-ChR2-opto condition compared to those in APP-mCherry-opto condition (Fig. 6B). Increases in cell body size and decreases in process length were observed indicating a shift towards a phagocytic state as a result of optogenetic treatment (Fig. 6C, D). Additionally, microglia-Aβ clustering analyses were performed to evaluate the microglial clustering patterns around amyloid plaques (Fig. 6A). Using three-dimensional images of each amyloid deposit, the number of microglia located within a 25 μm radius of the deposit was quantified using Imaris. Optogenetic treatment increased the number of reactive microglia surrounding amyloid plaques (Fig. 6A, E). Overall, our data showed that optogenetic stimulation resulted in unique morphological changes within microglia.

Fig. 6

Chronic optogenetic stimulation of GABAergic neurons altered microglia number and morphology, upregulated expression of phagocytic markers and enhanced phagocytic activity in APP mice. (A) Representative confocal Z-projections depicting Iba-1 positive microglia (Green) and Methoxy-XO4 positive plaques (Blue). (B-E) Microglia number (B), microglia cell body volume (C), microglia process length (D) and the number of plaque associated microglia (E). (F) Gating strategy used to identify CD11b + CD45lo microglia. The proportion of Aβ-phagocytic microglia (MeX04 + microglia). (G) Quantitation of the microglial cell population. (H) Quantification of the Methoxy-X04+CD11b+CD45low microglia. (I-K) Quantitation of microglial CD36 (I), CD68 (J) and CSF-1R (K) expression. All data are expressed as means ± standard error. The number of mice examined: n = 6 mice/group. *P < 0.05 and **P < 0.01. n.s. not significant. Scale bars: 50 μm

To further assess microglial function, we performed flow cytometry on cells isolated from fresh tissue and identified microglia with CD11b and CD45 markers (Fig. 6F). We saw an increase in microglia number as a result of optogenetic treatment consistent with immunohistochemical data (Fig. 6G, B). To investigate the phagocytic ability of microglia, the percentage of viable Methoxy-X04+CD11b+CD45low microglia as well as phagocytic markers CD36 and CD68 expression was measured. Optogenetic treatment resulted in an increased percentage of Methoxy-X04+CD11b+CD45low microglia (Fig. 6H) as well as in elevated expression of CD68 (Fig. 6J) compared to the mCherry group. This indicated increased Aβ clearance ability by microglia in the APP-ChR2-opto group. In addition, optogenetic treatment increased the expression of the microglia proliferation marker CSF-1R (Fig. 6K), consistent with the increased number of microglia observed. Moreover, APP mice expressing ChR2 without optogenetic stimulation (ChR2-no opto) showed no significant difference in microglia number or phagocytic ability (Supplemental Fig. 5D, E).

Taken together, these results demonstrate that optogenetic stimulation of ChR2 targeted to GABAergic neurons increased reactive microglial proliferation, induced a clustering phenotype around plaques, and promoted Aβ phagocytosis by microglia that resulted in clearance of Aβ.

Chronic optogenetic stimulation of GABAergic interneurons combined with sleep deprivation as well as 40 Hz stimulation failed to result in microglia-dependent amyloid plaque clearance in APP mice.

Increasing evidence suggests that sleep disruption may also play a role in AD pathogenesis [1, 2, 15, 48]. Since optogenetic stimulation of GABAergic neurons improved sleep and slowed AD progression, it was important to determine whether sleep restoration was necessary. Thus, we tested whether sleep restoration resulting from optogenetic activation of interneurons led to slowing of AD progression in APP mice. Sleep deprivation is known to reduce total sleep time in mice [33]. APP mice were treated with optogenetic activation of ChR2 while being sleep deprived for 6 h each day (APP-ChR2-opto-SD). Their sleep patterns were compared to their baseline levels prior to optogenetic stimulation and sleep deprivation in the same mice (APP-ChR2). Interestingly, sleep deprivation ablated previously observed improvements in NREM sleep associated with optogenetic treatment (Supplemental Fig. 7A, Fig. 2C). We further examined the effect of sleep deprivation in the presence of stimulation on plaque deposition and microglia numbers. Amyloid plaque burden remained high in the cortex and the hippocampus of APP-ChR2-opto-SD mice (Supplemental Fig. 7B-D). Moreover, there was no significant difference in microglia number or its Aβ phagocytic ability in the APP-ChR2-opto-SD group compared to the control (Supplemental Fig. 7E, F).

Moreover, optogenetic stimulation of cortical interneurons at 40 Hz failed to significantly alter amyloid plaque deposition and microglia numbers in APP mice compared to non-treated APP mice (Supplemental Fig. 8).

Therefore, the frequency of optogenetic stimulation at 0.6 Hz was critical to elicit the protective effects of optogenetic stimulation of GABAergic neurons on plaque deposition and microglia through sleep restoration in APP mice.