Forty-hertz light stimulation does not entrain native gamma oscillations in Alzheimer's disease model mice

Abstract

There is a demand for noninvasive methods to ameliorate disease. We investigated whether 40-Hz flickering light entrains gamma oscillations and suppresses amyloid-β in the brains of APP/PS1 and 5xFAD mouse models of Alzheimer's disease. We used multisite silicon probe recording in the visual cortex, entorhinal cortex or the hippocampus and found that 40-Hz flickering simulation did not engage native gamma oscillations in these regions. Additionally, spike responses in the hippocampus were weak, suggesting 40-Hz light does not effectively entrain deep structures. Mice avoided 40-Hz flickering light, associated with elevated cholinergic activity in the hippocampus. We found no reliable changes in plaque count or microglia morphology by either immunohistochemistry or in vivo two-photon imaging following 40-Hz stimulation, nor reduced levels of amyloid-β 40/42. Thus, visual flicker stimulation may not be a viable mechanism for modulating activity in deep structures.

Extended Data Fig. 1 |. Histological analysis.

a, b) Heatmaps of Aβ plaque load in various coronal sections in 5xFAD mice after sham (a) and 40-Hz flicker (b) stimulation. c) Representative coronal sections of Aβ expression (LifeCanvas Technologies, Cambridge, MA 02141). Was replicated in n = 2 mice. d) Posterior visual cortex section (marked by red dash lines), used for quantification of V1. e) Group difference (medians, interquartile ranges, maxima and minima) of the percent area occupied by plaques after sham (n = 11 mice) and 40-Hz flicker (n = 13 mice) stimulation (ns, not significant by Wilcoxon test: p = 0.12). f) Cortex plaque load variability among six cohorts (mean + /− s.d; two-sided Wilcoxon test; p = 0.915, p = 0.0368, p = 0.879, p = 0.988, p = 0.0878, p = 0.379; ns, not significant).

Extended Data Fig. 2 |. ELISA analysis by groups.

a, b) Group difference (medians, interquartile ranges, maxima and minima) Aβ40 peptide levels in the (a) hippocampus and (b) V1 separated by sex (ns, not significant; two-sided Wilcoxon test). (n = 7 male and n = 5 female 40-Hz treated; n = 7 male and n = 4 female non-treated controls) c, d) Group difference (medians, interquartile ranges, maxima and minima) Aβ42 peptide levels in the (c) hippocampus (p = 0.38; p = 1;p = 0.19; p = 0.19 two-sided Wilcoxon Test) and (d) V1 (p = 0.02; p = 0.46;p = 0.68; p = 0.73; two-sided Wilcoxon Test) separated by sex and ELISA test. To test the reliability and consistency of the ELISA method, we performed two separated analyses for Aβ42 from the same aliquot samples on different days separated by one week (sample 1, sample2). The corresponding aliquots in the two tests are connected by lines. (n = 7 male and n = 5 female 40-Hz treated; n = 7 male and n = 4 female non-treated controls) e) The correlation between the two ELISA tests in c and d. Two conclusions may be drawn from this repeated analysis. First, the two tests were correlated strongly and significantly (R = 0.78). Second, the values of the second ELISA test were, on average, lower. This latter observation suggests that some epitope degradation of the sample may take place with time. These findings points to further potential sources of variability across studies, which may have different delays between tissue processing and ELISA Analysis.

Extended Data Fig. 3 |. Allen Brain Institute 4-Hz light response.

a) Peristimulus time histograms showing the responses of individual neurons (bottom panels) and the average (mean ± s.e.m.) response per area, from primary visual cortex (VISp), the 3 major hippocampal areas (DG, dentate gyrus), the thalamus (TH) and midbrain (MB). b) Anatomical locations of 35,910 units included in the dataset, color coded according to their modulation index. c) Box and whisker plot (medians, interquartile ranges, maxima and minima) showing the distributions of modulation index in the different areas p = 0, p = 0, p = 0, p = 0, p = 0, p = 4.9e-14, p = 0.013,p = 0.999,p = 0.970, p = 8.97e-19, p = 0.0006, p = 0.0434, p = 0.940, p = 0.359, p = 0.962 for VISp-CA1, VISp-CA3, VISp-DG, VISp-TH, VISp-MB, CA1-CA3, CA1-DG, CA1-TH, CA1-MB, CA3-DG, CA3-TH, CA3MB, DG-TH,DG-MB,TH-MB, respectively; KW; *p < 0.05, **p < 0.01, ***p < 0.001. d) Bar graph shows the fractions of significantly modulated cells per area (VISp: 1597/3439; CA1: 1225/5833; CA3:145/835; TH:2329/6249; MB: 638/1908) (n = 49 mice).

Extended Data Fig. 4 |. Electrophysiological methods.

a) Relationship between electrode impedance and 40-Hz power in the hippocampus. Note that artifactual 40-Hz power can occur at high impedance sites. b) Power spectra in a high impedance channel in the hippocampus during 40-Hz stimulation and no-stimulation epochs (10 s on, 10 s off). Note large peak at 40-Hz during stimulation. c) Example artifacts at a high-impedance channels (red traces) in the hippocampus show 40-Hz when the light is turned on. d) Comparison (medians, interquartile ranges, maxima and minima) of the two methods we used to quantify phase modulation of spikes: bootstrap and Rayleigh’s methods (Example from the hippocampus; ns, nonsignificant p = 0.193; two-sided Wilcoxon test). Vector length distribution. Vertical red lines separate nonsignificant and significant events. (59 sessions in 15 mice). e) Significant difference (medians, interquartile ranges, maxima and minima) in the vector lengths of neurons statistically modulated by 40-Hz in different brain regions. Note that despite very few CA1 and EC neurons show significant phase-locking to 40-Hz stimuli (Fig. 2), the few that do show comparable vector lengths in all three structures (***p < 0.001; V1:p = 3.12e-41;CA1: p = 1.6e-30; EC: p = 0.239; two-sided Wilcoxon test) (V1 = 14 sessions in 5 mice, EC = 7 sessions in 3 mice, CA1 = 59 sessions in 15 mice). f) Example raster plots of significantly modulated putative interneurons in CA1 (red) and EC (orange) regions. g) Separation of putative pyramidal cells from interneurons using spike duration and burst index. Bar graphs show the fraction of significantly modulated putative pyramidal cells and interneurons.

Extended Data Fig. 5 |. Firing rates changes during stead state driving at 40-Hz stimulation.

Firing rates of all neurons during onset and offset of 40-Hz trains every 10 second for the visual cortex (a; two-sided paired t-test; p = 0.0066), hippocampus (b; two-sided paired t-test; p = 0.2581), and entorhinal cortex (c; two-sided paired t-test; p = 0.2629), respectively. Bin size= 0.005 s. 40-Hz modulated cells are marked by color circles.

Extended Data Fig. 6 |. Aversive response to 40-Hz flicker and control Ach.

a) Time spent in the 40-Hz compartment versus in the compartment with continuous light (n = 14 mice). b) Ach response to continuous white light (mean + /− s.e.m; p = 0.59, two-tailed paired t-test; n = 5 mice). Note lack of sustained activity, in contrast to the sustained Ach activation with 40-Hz flickering light (Fig. 4).

Fig. 1 |. Effect of acute 40-Hz visual simulation on Aβ load and microglia.

a, Schematic of experimental design. Baseline of 10 min followed by 1 h of control light or 40-Hz stimulation, after which the animal was killed for histological analysis. Parts created with BioRender.com. b, Representative IHC with superimposed anti-Aβ (D54F2, green), anti-Iba1 (microglia, red) and DAPI (blue) in two 4-month-old 5xFAD mice exposed to continuous white light (control) or 40-Hz flicker light (replicated in n = 34 mice; Supplementary Table 1). c, Schematic of brain areas used to quantify Aβ load in neocortex (dark blue), V1 (light blue) and hippocampus (red). d–f, Group difference (medians, interquartile ranges, maxima and minima) of the percentage area occupied by plaques in neocortex (d) (NS, P = 0.12; two-sided Wilcoxon test), hippocampus (e) (NS, P = 0.296; two-sided Wilcoxon test) and V1 (f) (NS, P = 0.15; two-sided Wilcoxon test) (control, 20 mice; 40-Hz flicker, 22 mice). Squares are females and circles are males. The different colored dots correspond to different littermate cohorts (Extended Data Fig. 1e). g, High-magnification IHC pictures stained with anti-Aβ (D54D2, green) and anit-Iba1 (microglia, red) antibodies in V1 of the same two 5xFAD mice as in b (replicated in n = 34 mice for 40-Hz stimulation and n = 31 control mice; Supplementary Table 1). h, PCC between microglia and Aβ in control and 40-Hz mice (medians, interquartile ranges, maxima and minima; NS by two-sided Wilcoxon test: P = 0.538) (control, 15 mice; 40-Hz flicker, 15 mice). F, female; M, male; NS, not significant.

Fig. 2 |. Effect of chronic 7-d 40-Hz visual simulation.

a, Group differences (medians, interquartile ranges, maxima and minima) of the percentage area occupied by Aβ in the neocortex (NS, P = 0.13; two-sided Wilcoxon test), hippocampus (NS, P = 0.479; two-sided Wilcoxon test) and V1 (NS, P = 0.309; two-sided Wilcoxon test). b, PCC between Aβ (anti-Aβ, D54F2) and microglia (anti-Iba1, microglia) (NS, P = 0.406; two-sided Wilcoxon test; medians, interquartile ranges, maxima and minima). c, Relative human Aβ42 (hippocampus: NS, P = 0.096; two-sided Wilcoxon test; V1: NS, P = 0.242; two-sided Wilcoxon test) and Aβ40 (hippocampus: NS, P = 0.089; two-sided Wilcoxon test; V1: NS, P = 0.096; two-sided Wilcoxon test) peptide levels in hippocampus and V1 measured by ELISA and normalized to overall protein levels detected with Bradford test (n = 11 control; n = 12 40-Hz stimulation mice; medians, interquartile ranges, maxima and minima).

Fig. 3 |. In vivo monitoring of plaques and microglia.

a, Schematic of two-photon experiment. Mice were first imaged before and after 1 h of continuous white light stimulation, after which they were allowed to rest for 1 h in the homecage. Following rest, they were placed back in the imaging set-up and images were taken before and 1 h after 40-Hz stimulation. Parts created with BioRender.com. b, Differences (medians, interquartile ranges, maxima and minima) between the fractional area occupied by plaques (PA difference) in the imaged V1 before and after continuous light stimulation (sham) and before and after 40-Hz light flicker (sham, 7 mice; no stimulation, 2 mice; 40-Hz flicker, 9 mice; NS, two-sided Wilcoxon test: P = 0.758). c, Percentage plaque area (medians, interquartile ranges, maxima and minima) for the n = 9 mice before and after 40-Hz stimulation (NS, P = 0.934; two-sided Wilcoxon test). d, Aβ plaque area (medians, interquartile ranges, maxima and minima) did not change following 40-Hz stimulation (NS; sham–pre P = 0.382; pre–post P = 0.383; sham–post P = 0.915; two-sided Wilcoxon test) in n = 7 mice. In this comparison, individual plaques were compared before and after flicker stimulation. e, Example Aβ plaque image and its mask after sham stimulation, before and after 40-Hz flicker exposure (replicated in n = 8 mice; Supplementary Table 2). f, Immunohistochemical quantification of percentage Aβ area (medians, interquartile ranges, maxima and minima) in the entire V1 in a subset of 7 mice used for two-photon experiments compared with 3 age-matched, nonstimulated control mice (NS, P = 0.287; two-sided Wilcoxon test). g, Example two-photon images of plaques. Red, methoxy-X04); green, microglia in Cx3cr1^GFP/+:5xFAD^+/− mice and their merged version before and after 40-Hz flicker stimulation (replicated in n = 6 mice; Supplementary Table 2). h, Percentage occupancy of microglia (medians, interquartile ranges, maxima and minima) in the imaged area before and after 40-Hz flicker stimulation (bottom; n = 6 mice; NS, P = 0.932; two-sided Wilcoxon test). i–m, Further morphological analyses for microglial parameters (ref. ) (medians, interquartile ranges, maxima and minima). i, Glial area, quantified as the total number of pixels present in the filled shape (µm²; NS, P = 0.484; two-sided Wilcoxon test). j, Cell perimeter (the number of pixels in µm that outline the cell; NS, P = 0.589; Wilcoxon test). k, Density = area/convex hull area (NS, P = 0.484; two-sided Wilcoxon test). l, Number of branches in each microglia (NS, P = 0.766; two-sided Wilcoxon test). m, The soma size (the pixel area µm² occupied by the microglial soma mask; NS, P = 0.793; two-sided Wilcoxon test).

Fig. 4 |. A 40-Hz visual simulation does not engage native gamma oscillations.

a, Schematic of the experimental design. Headgear is shielded by a 3D-printed copper mesh cap. Baseline of 10 min followed by 30 min of 40-Hz stimulation (10 s on, 10 s off) and 10 min of control recording. Parts created with BioRender.com. b, Mice were implanted with silicon probes in V1, CA1 or EC. Example LFP depth profiles for each brain region. c, Mean power spectra ± s.e.m. from V1 (blue) and hippocampal CA1 (red) recording sites during interleaved 40-Hz stimulation and no-stimulation epochs, overlayed. Narrow-band 40-Hz response is only seen in V1 during light stimulation (inset shows enlarged part of 40-Hz response). d, Example time-resolved spectrogram of V1, CA1 and EC recordings. White and black vertical lines mark the onset and offset of 40-Hz stimulation, respectively. Curved arrow, narrow-band, 40-Hz (γ) in V1. Note intermittent, brain-state-dependent responses despite continuous 40-Hz stimulation. e, Example average LFP traces (black) and current source density map in V1, triggered by the onset of 40-Hz trains every 10 s. The color scale indicates current density (a.u.). f, z-scored single neuron responses in V1 (n = 458 neurons in 5 mice) to the onset of 40-Hz trains every 10 s. Overlayed in black is the average response signal. g, Example peristimulus time histograms of example neurons in V1 (blue), CA1 (red) and EC (orange). Black, average firing rate. h, Distribution of z-scored light responses. Note low fraction of significantly responding neurons (red line, z-score > 1.96) in CA1 (red) and EC (orange). i, Phase distribution of significantly phase-modulated V1 neurons. j, Example of same V1 cell (g) peristimulus time histogram around 40-Hz stimulus. k, Z-scored single neuron responses in V1 (n = 458 neurons in 5 mice), CA1 (n = 1,874 neurons in 15 mice) and EC (n = 211 neurons in 3 mice) to 40-Hz stimulation. The horizontal line separates nonmodulated cells (above) from modulated neurons (below). l, Percentage of 40-Hz-modulated cells in V1 (blue), CA1 (red) and EC (orange). FFT, Fast Fourier transforms.

Fig. 5 |. A 40-Hz flickering stimulation is aversive.

a, Example of a mouse’s movement trajectory in a two-compartment (40-Hz flickering light and continuous light, respectively) apparatus. b, Group average time spent in the respective compartments (n = 14 mice; ***P = 3.59 × 10⁻⁴, two-sided Wilcoxon test). c, Ach3.0 fluorescence signal, measured by GRAB fiber photometry (**P = 0.0014, two-tailed paired t-test) and movement (NS, P = 0.67, two-tailed paired t-test) recorded during alternating 10 s on, 10 s off 40-Hz light stimulation (40 min).