MS cholinergic neurons are active during REMS and wakefulness and predominantly project to hippocampal and cortical regions

(A) AAV-DIO-GCaMP6s was injected into the MS of ChAT-Cre mice to target cholinergic neurons (scale bar: 100 μm). An optic fiber was implanted above the MS, and EEG/EMG electrodes enabled simultaneous fiber photometry and polysomnographic recordings (n = 6 mice). (B) GCaMP fluorescence signals increased during transitions from NREMS to REMS and from NREMS to wakefulness, but showed minimal changes during wake-to-NREMS transitions, indicating selective activation of MS cholinergic neurons during REMS and wakefulness. During NREMS-to-REMS transitions, 97% of events were associated with increased activity, whereas 76% of NREMS-to-wakefulness transitions showed elevated firing. Transitions from wakefulness to NREMS displayed a balanced distribution of increases and decreases. (C) AAV-DIO-ChR2-EYFP was injected into the MS of ChAT-Cre mice (n = 2) to visualize cholinergic axons (green; scale bar: 200 μm). Serial coronal sections were used to map cholinergic projections. Gradient color-coding indicates the rostrocaudal position of each brain slice in whole-brain quantified immunohistochemical data. (D) Coronal brain sections show GFP-positive cholinergic projections (green) and Hoechst-stained nuclei (blue) including cortical, hippocampal, thalamic, and hypothalamic regions (scale bar: 100 μm). Crop identifiers (C1.1-C7.2) correspond to red squares in the schematic overview shown in panel C. Statistical analysis: two-sided Student’s paired t-test; mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001. Parts of this figure were created with the help of BioRender.com.

MSChAT-AppNL-G-F mice exhibit local Aβ deposits, while AppNL-G-F/NL-G-F knock-in mice show marked increase in Aβ deposition

(A) MSChAT-AppNL-G-F mice were generated by injecting AAV-DIO-AppNL-G-F-2A-EGFP into the MS of ChAT-Cre mice. EEG/EMG recordings were performed every four months over the year post-injection to monitor sleep-wake cycles. From 12 months onward, behavioral assessments, locomotor activity monitoring, and baseline stress level measurements were conducted prior to brain collection for histological analyses (scale bar: 100 μm). (B) Serial coronal brain sections spanning the rostrocaudal extent of the MS were analyzed. In plots of quantified immunohistochemical BF data, gradient color-coding indicates the rostrocaudal position of each brain slice from anterior to posterior. (C) Coronal brain sections from MSChAT-GFP (left) and MSChAT-AppNL-G-F (right) mice show Aβ immunostaining (D54D2, red) and cell nuclei (Hoechst, blue) (scale bar: 200 μm). (D) AppNL-G-F/NL-G-F knock-in mice, carrying familial Alzheimer’s disease-linked APP mutations, were monitored using the same longitudinal design as MSChAT-AppNL-G-F mice, with EEG/EMG recordings every four months from birth and behavioral, locomotor, and stress assessments from 12 months onward. Coronal brain sections from AppWT (control; left) and AppNL-G-F/NL-G-F (right) mice show Aβ immunostaining (D54D2, red) with Hoechst counterstaining (blue) (scale bar: 200 μm). (E) MSChAT-AppNL-G-F mice exhibited increased Aβ accumulation in the MS, with no significant changes in the VDB or HDB compared with MSChAT-GFP controls. In contrast, AppNL-G-F/NL-G-F mice displayed robust increases in Aβ deposition across the MS, VDB and HDB relative to AppWT controls. Aβ levels were also higher in AppNL-G-F/NL-G-F mice than in MSChAT-AppNL-G-F animals across all BF regions. See panel B for gradient color-coding. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 4 males/group, n = 4 females/group); mean ± SEM. Asterisks on MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F groups denote significance relative to their corresponding controls. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001. Parts of this figure were created with the help of BioRender.com.

MSChAT-AppNL-G-F mice exhibit broadcasted Aβ deposits

(A) Coronal whole-brain sections (bregma -1.70mm) showing Aβ immunostaining (D54D2, red) and nuclei (Hoechst, blue) in MSChAT-GFP (left) and MSChAT-AppNL-G-F (right) mice (scale bar: 500 μm). (B) Aβ spread and accumulation in the hippocampus were examined in MSChAT-AppNL-G-F mice. Coronal hippocampal sections show Aβ immunostaining (D54D2, red) and nuclei (Hoechst, blue) in MSChAT-GFP (top) and MSChAT-AppNL-G-F (bottom) mice (scale bar: 200 μm). MSChAT-AppNL-G-F mice showed significantly higher Aβ accumulation in the hippocampal regions (CA areas, dentate gyrus, and subiculum) compared with MSChAT-GFP mice. See Figure 1C for gradient color-coding. (C) Quantification of Aβ deposition in other projection fields of MSChAT neurons revealed significantly higher Aβ accumulation in the thalamus, amygdala, medial prefrontal cortex (prelimbic and cingulate cortices), primary sensory cortices (somatosensory, motor, and visual), and hypothalamus in MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls. No significant differences were observed in the medial habenula, olfactory bulb, or entorhinal cortex. See Figure 1C for gradient color-coding. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 3-4 males/group, n = 4 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Loss of BF cholinergic neurons in aged MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice, and in MSΔChAT mice

(A-B) Coronal sections showing cholinergic neurons (ChAT, orange) and nuclei (Hoechst, blue) in the MS, VDB, and HDB of (A) MSChAT-GFP (control; left) and MSChAT-AppNL-G-F(right) mice and (B) AppWT (control; left) and AppNL-G-F/NL-G-F (right) mice (scale bar: 200 μm). (C) Coronal MS sections showing cholinergic neurons (ChAT; orange) and nuclei (Hoechst, blue) in ChAT-Cre mice after MS-targeted injection of AAV-DIO-EGFP (MSChAT-GFP, control; top) or AAV-DIO-CASP (MSΔChAT; bottom) (scale bar: 100 μm). (D) MSChAT-AppNL-G-F mice exhibited a significant reduction in cholinergic neuron number in the MS compared with MSChAT-GFP mice, while the VDB and HDB were unaffected. AppNL-G-F/NL-G-F mice showed reduced cholinergic neurons in the MS and HDB relative to AppWT controls, with no change in the VDB. MSΔChAT mice also displayed a marked MS-restricted loss of cholinergic neurons compared to MSChAT-GFP and AppWT mice. See Figure 2B for gradient color-coding. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 4 males/group, n = 4 females/group); mean ± SEM. Asterisks on MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F groups denote significance relative to their corresponding controls. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Partially overlapping cognitive and emotional alterations in MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F and MSΔChAT mice

(A) Home-cage locomotor activity monitoring showed that AppNL-G-F/NL-G-F mice displayed increased locomotor activity relative to AppWT controls during the dark phase. Females were more active than males (see Supplemental information for detailed statistics). (B) Basal stress levels were assessed by quantifying fecal corticosterone metabolites (FCM) across five time windows spanning a 28 h period (gray bar). FCM levels in MSChAT-AppNL-G-F mice did not differ from MSChAT-GFP controls, whereas AppNL-G-F/NL-G-F mice exhibited significantly elevated levels compared with AppWT counterparts. In both models, female mice showed higher FCM levels than males (see Supplemental information for detailed statistics). (C) Spatial working memory was assessed using the Y-maze test. MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F, and MSΔChAT mice, exhibited significantly fewer spontaneous alternations compared to their respective controls, indicating impaired working memory. (D) Non-spatial working memory was evaluated using the novel object recognition task (NORT). During the test phase (maximum 20 s exploration), MSChAT-AppNL-G-F and MSΔChAT mice did not display a significant preference for the novel object, indicating a working memory deficit, whereas MSChAT-GFP controls showed a clear preference. In contrast, both AppNL-G-F/NL-G-F and AppWT mice exhibited significant preference for the novel object. (E) Anxiety-like behavior was assessed using the open field test. MSChAT-AppNL-G-F mice spent significantly more time in the center compared with MSChAT-GFP controls, indicating reduced anxiety-like behavior. AppNL-G-F/NL-G-F mice showed an even greater increase in center time relative to both AppWT controls and MSChAT-AppNL-G-F mice, whereas MSΔChAT mice did not differ from controls. (F) Anxiety-like behavior was further assessed using the light-dark box test. MSChAT-AppNL-G-F mice spent significantly more time in the illuminated compartment compared with MSChAT-GFP controls, consistent with reduced anxiety-like behavior. In contrast, AppNL-G-F/NL-G-F and MSΔChAT mice did not differ from their respective controls. Statistical analysis: linear mixed-model and standard ANOVAs with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate, and one-sample Student’s t-test for NORT (n = 5-8 males/group, n = 5-7 females/group); mean ± SEM. Asterisks on MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F groups denote significance relative to their corresponding controls, except for NORT, where they all denote deviation from chance-level exploration (10 s for each object) Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Comparable REMS phenotypes in aging MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice, and in MSΔChAT mice

(A) During the light phase, REMS duration in MSChAT-AppNL-G-F mice diverged from MSChAT-GFP controls at 8 and 12 months following AAV injection, the former group showing overall less REMS. A similar pattern was observed in AppNL-G-F/NL-G-F mice relative to AppWT controls at matched stages of amyloid pathology progression. At both stages, AppNL-G-F/NL-G-F mice exhibited less REMS than MSChAT-AppNL-G-F mice. Notably, MSChAT-GFP and AppWT controls also differed at 12 months. (B) REMS bout length was overall increased during the light phase in both MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice compared with their respective controls, irrespective of time scale. Time-specific increases were observed at 4 months post-injection in MSChAT-AppNL-G-F mice and at later stages of pathology in AppNL-G-F/NL-G-F mice. (C) In MSChAT-GFP mice, the number of REMS bouts increased with time following AAV injection during the light phase, whereas it remained stable in MSChAT-AppNL-G-F mice, resulting in significant differences at 8 and 12 months. During the dark period, differences between these groups emerged only at 12 months post-injection. In AppNL-G-F/NL-G-F mice, REMS bout number was reduced compared with AppWT controls at comparable stages of pathology in the light phase, with an additional reduction observed at earlier stages in the dark phase. In the light phase, REMS bout number was also consistently lower in AppNL-G-F/NL-G-F mice than in MSChAT-AppNL-G-F mice at matched stages of amyloid progression. MSChAT-GFP and AppWT controls also differed at 12 months in both light and dark periods. (D) Six months after AAV injection, MSΔChAT mice displayed reduced REMS duration during the light phase compared with MSChAT-GFP mice at 12 months of amyloid pathology. (E) REMS bout length did not differ between MSΔChAT mice and other groups. (F) During the light phase, the number of REMS bouts was reduced in MSΔChAT mice compared with MSChAT-GFP controls. (G) EEG spectral analysis (FOOOF) revealed decreased integrated periodic delta (1-4.4 Hz) and theta (6-9 Hz) power in both light and dark phases in MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice relative to their respective controls, as well as in MSΔChAT mice compared with MSChAT-GFP controls. Theta power also differed between MSChAT-GFP and AppWT controls in both phases. Statistical analysis: linear mixed-model and standard ANOVAs with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 5-8 males/group, n = 5-8 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Distinct contributions of amyloid burden and cholinergic loss to phenotypic outcomes

(A) Partial Kendall’s tau correlations across models exhibiting cholinergic loss (MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F, and MSΔChAT) revealed positive associations between cholinergic neuron number and both REMS duration and theta power during the light phase. (B) The number of interictal epileptiform spikes was increased in MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice compared with their respective controls during both light and dark phases. AppNL-G-F/NL-G-F mice exhibited a higher incidence of spikes than MSChAT-AppNL-G-F mice, whereas such events were not observed in MSΔChAT mice, suggesting a contribution of amyloid pathology to their generation. (C) Coronal sections of the MS showing GFAP immunostaining (astrocytes, orange) and nuclei (Hoechst, blue) (scale bar: 100 μm). MSChAT-AppNL-G-F mice exhibited increased GFAP expression selectively in the MS compared with MSChAT-GFP and MSΔChAT mice, consistent with the local accumulation of amyloid, whereas no increase was observed in the VDB or HDB. In the hippocampus, where amyloid is also distributed in MSChAT-AppNL-G-F mice, GFAP staining was similarly elevated compared with MSChAT-GFP controls. Statistical analysis: linear mixed-model and standard ANOVAs with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate, and partial Kendall’s tau correlations controlling for group and sex (n = 4-8 males/group, n = 4-8 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Enhanced colocalization between the 6E10 amyloid marker and cerebral vasculature outside the BF in MSChAT-AppNL-G-F mice

(A) Coronal MS sections showing amyloid deposits (6E10, orange) and cell nuclei (Hoechst, blue) in MSChAT-GFP (control; left) and MSChAT-AppNL-G-F (right) mice (scale bar: 100 μm). (B) Quantification revealed significantly increased Aβ accumulation in the MS, VDB, and HDB of MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls. See Figure 2B for gradient color-coding. (C) Representative hippocampal section from a MSChAT-AppNL-G-F mouse show colocalization of amyloid deposits (6E10; orange) with blood vessels (laminin, red), highlighted by white arrows in the dentate gyrus (nuclei stained with Hoechst, blue) (scale bar: 50 μm). (D) Quantitative analysis using Mander’s coefficient (M1) showed that the proportion of 6E10 signal overlapping with laminin-positive vasculature was higher in the hippocampus than in the MS, VDB, and HDB in both MSChAT-GFP and MSChAT-AppNL-G-F mice. See Figure 1C for gradient color-coding. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 3 males/group, n = 3 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Broadcasted Aβ deposits into the hippocampus of MSChAT-AppNL-G-F mice do not induce neuronal loss

Coronal hippocampal section showing the neuronal marker NeuN (orange) and cell nuclei (Hoechst, blue) (scale bar: 200 μm). Quantification of NeuN-positive cells revealed no difference between MSChAT-GFP and MSChAT-AppNL-G-F mice. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 4 males/group, n = 4 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

GABAergic and glutamatergic neurotransmission is preserved in MSChAT-AppNL-G-F mice but impaired in AppNL-G-F/NL-G-F mice

(A) Coronal MS sections showing immunostaining for glutamic acid decarboxylase 67 (GAD67, red), and nuclei (Hoechst, blue) across genotypes from left to right: MSChAT-GFP (control), MSChAT-AppNL-G-F, AppWT (control), and AppNL-G-F/NL-G-F mice (scale bar: 100 μm). (B) GAD67 expression in the MS, VDB, and HDB did not differ between MSChAT-GFP and MSChAT-AppNL-G-F mice. In contrast, AppNL-G-F/NL-G-F mice showed significantly reduced GAD67 expression across all three BF regions compared with AppWT controls. See Figure 2B for gradient color-coding. (C) Coronal MS sections showing immunostaining for vesicular glutamate transporter 2 (VGluT2, orange) and nuclei (Hoechst, blue) across genotypes arranged from left to right: MSChAT-GFP (control), MSChAT-AppNL-G-F, AppWT (control), and AppNL-G-F/NL-G-F mice (scale bar: 100 μm). (D) VGluT2 expression in the MS, VDB, and HDB did not differ between MSChAT-GFP and MSChAT-AppNL-G-F mice. In contrast, AppNL-G-F/NL-G-F mice exhibited significantly reduced VGluT2 expression in all BF regions compared with AppWT controls. VGluT2 levels were also consistently lower in AppNL-G-F/NL-G-F mice than in MSChAT-AppNL-G-F mice. See Figure 2B for gradient color-coding. Statistical analysis: linear mixed-model ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 4 males/group, n = 4 females/group); mean ± SEM. Asterisks on MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F groups denote significance relative to their corresponding controls. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Shared alterations in NREMS phenotypes in MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F and MSΔChAT mice

(A) Across the one-year monitoring period, total NREMS duration remained comparable between groups during both light and dark phases. (B) During the light period, NREMS bout length was significantly increased in MSChAT-AppNL-G-F mice at 8 and 12 months post-injection compared with MSChAT-GFP controls. A similar increase was observed in AppNL-G-F/NL-G-F mice relative to AppWT controls at 12 months, mirroring the phenotype seen in MSChAT-AppNL-G-F animals. (C) MSChAT-GFP mice showed an age-dependent increase in NREMS bout number during the light phase, whereas this effect was absent in MSChAT-AppNL-G-F mice, resulting in significant differences at 8 and 12 months post-injection. In AppNL-G-F/NL-G-F mice, NREMS bout number was reduced compared with AppWT controls at 12 months in the light period, with an additional reduction observed at 4 months during the dark phase. At 12 months of amyloid pathology, group differences were also evident in the light phase, with MSChAT-AppNL-G-F and MSChAT-GFP mice respectively differing from AppNL-G-F/NL-G-F and AppWT controls. (D) Six months after AAV injection, MSΔChAT mice did not differ in NREMS duration compared with other groups at 12 months of amyloid pathology. (E) NREMS bout length was comparable between MSΔChAT mice and all other groups. (F) During the light phase, MSΔChAT mice showed a reduced number of NREMS bouts compared with MSChAT-GFP controls. In addition, during the dark phase, MSΔChAT mice exhibited a higher number of NREMS bouts relative to AppWT controls. (G) EEG spectral analysis (FOOOF) revealed decreased integrated periodic delta (1-4.4 Hz) and theta (6-9 Hz) power during both light and dark phases in MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls. In AppNL-G-F/NL-G-F mice, only theta power was significantly reduced across both phases compared with AppWT controls. Similar to the MSChAT-AppNL-G-F model, MSΔChAT mice also showed reduced delta and theta power in both light and dark phases relative to MSChAT-GFP controls. Statistical analysis: linear mixed-model and standard ANOVAs with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 5-8 males/group, n = 5-8 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Differential wakefulness phenotypes in MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F and MSΔChAT mice

(A) Total wakefulness duration remained comparable between MSChAT-AppNL-G-F and MSChAT-GFP mice across the one-year recording period in both light and dark phases. In contrast, AppNL-G-F/NL-G-F mice exhibited increased wakefulness compared with AppWT controls at 8 and 12 months during the light phase only. At 12 months of amyloid pathology, AppNL-G-F/NL-G-F mice also differed from MSChAT-AppNL-G-F mice. (B) Wakefulness bout length remained stable and did not differ between MSChAT-AppNL-G-F and MSChAT-GFP mice, nor between AppNL-G-F/NL-G-F and AppWT mice, across all time points and circadian phases. (C) The number of wakefulness bouts was also stable across time and did not differ between MSChAT-AppNL-G-F and MSChAT-GFP mice or between AppNL-G-F/NL-G-F and AppWT controls, in both light and dark phases. (D) Six months after AAV injection, MSΔChAT mice did not differ in total wakefulness duration compared with other groups at 12 months of amyloid pathology. (E) Wakefulness bout length was comparable between MSΔChAT mice and all other groups. (F) MSΔChAT mice exhibited an increased number of wakefulness bouts during both light and dark phases compared with MSChAT-GFP controls. During the dark phase, this increase was also significant relative to AppWT controls. (G) EEG spectral analysis (FOOOF) revealed decreased integrated periodic delta (1-4.4 Hz) and theta (6-9 Hz) power during both light and dark phases in MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls. In AppNL-G-F/NL-G-F mice, delta power was reduced in the dark phase relative to AppWT controls, while theta power was lower in AppWT mice compared with MSChAT-GFP in both phases. Similar to MSChAT-AppNL-G-F mice, MSΔChAT mice showed reduced theta power in both light and dark phases relative to MSChAT-GFP controls, with a reduction in delta power restricted to the light phase. Statistical analysis: linear mixed-model and standard ANOVAs with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 5-8 males/group, n = 5-8 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

Aperiodic EEG features across vigilance states in MSChAT-AppNL-G-F, AppNL-G-F/NL-G-F, and MSΔChAT mice

(A) In REMS, the aperiodic offset was reduced in MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice compared with their respective controls during both light and dark phases. A similar reduction was observed in MSΔChAT mice relative to MSChAT-GFP controls. (B) The aperiodic exponent in REMS did not differ between groups. (C) In NREMS, the aperiodic offset was selectively reduced in MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls in both light and dark phases. (D) The aperiodic exponent in NREMS remained unchanged across groups. (E) During wakefulness, the aperiodic offset was reduced in MSChAT-AppNL-G-F and AppNL-G-F/NL-G-F mice relative to their respective controls in both phases, whereas MSΔChAT mice did not differ from other groups. (F) The aperiodic exponent in wakefulness showed no group differences. Statistical analysis: standard ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 5-7 males/group, n = 5-7 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.

EEG spectral changes expressed as percentage of total power

(A) During REMS in the light phase, conventional EEG spectral analysis (expressed as percentage of total power) revealed a selective decrease in theta power in MSChAT-AppNL-G-F mice compared with MSChAT-GFP controls. In contrast, MSΔChAT mice exhibited an increase in delta power relative to MSChAT-GFP controls. (B) During NREMS in the light phase, theta power was significantly reduced in AppNL-G-F/NL-G-F mice compared with MSChAT-AppNL-G-F animals. (C) During wakefulness, no significant spectral differences were detected between groups. Statistical analysis: ANOVA with Holm-Bonferroni correction for multiple pairwise comparisons where appropriate (n = 5-7 males/group, n = 5-7 females/group); mean ± SEM. Symbols: ▪ males, ● females, ♦ males and females, * p < 0.05, ** p < 0.01, *** p < 0.001.