Impaired hippocampal circuit function underlying memory encoding and consolidation precede robust Aβ deposition in a mouse model of alzheimer's disease

Abstract

Current therapeutic strategies for Alzheimer's disease (AD) target amyloid-beta (Aβ) fibrils and high molecular weight protofibrils associated with plaques, but molecular cascades associated with AD may drive neural systems failure before Aβ plaque deposition in AD. Employing hippocampal electrophysiological recordings and dynamic calcium imaging across the sleep-wake cycle in a freely behaving mouse model of AD before Aβ plaques accumulated, we detected marked impairments of hippocampal systems function: In a spatial behavioral task, phase-amplitude coupling (PAC) of the hippocampal theta and gamma oscillations was impaired and place cell calcium fluctuations were hyper-synchronized with the theta oscillation. These changes were not observed in REM sleep. In subsequent slow wave sleep (SWS), place cell reactivation was reduced. These degraded neural functions underlying memory encoding and consolidation support targeting pathological processes of the pre-plaque phase of AD to treat and prevent hippocampal impairments.

Keywords: Alzheimer’s disease; Amyloid-beta; Electrophysiology; Hippocampus; Place cell; Sleep.

Fig. 1

Young APP/PS1 mice expressing minimal Aβ plaque burden and abundant soluble AβO showed preserved place cell function and no calcium hyperactivity across the sleep-wake cycle. (a) Schematic of hippocampal calcium imaging and simultaneous ipsilateral LFP recording across the sleep-wake cycle. Insert: Examples of LFP spectrogram (color bar indicates power) and one CA1 neuron’s calcium trace across behavioral states; and example dataset of place cells sorted by their place fields (color bar indicates probability) in inbound (LR) and outbound (RL) directions. Yellow shows the place field of each cell. Exploratory behavior (RUN); quiet wakefulness (QW); slow wave sleep (SWS); rapid eye movement (REM) sleep. (b) Immunohistochemistry against Aβ in hippocampus of a wild-type control mouse (6 months) and APP/PS1 mice at 6 and 12 months. Green: Aβ plaques (D54D2). (c) Aβ plaque counts in hippocampus and cortex were low in 4–6 month-old APP/PS1 mice and increased markedly at 12 months. (d) Levels of soluble AβO in hippocampus and cortex increased with age in APP/PS1 mice. (e) Spatial information (SI) and (f) place field width were comparable across genotypes. (g) Calcium event rates of place cells and non place cells in each behavioral state were similar to those of littermate control mice, and calcium event rates of place cells were higher in RUN than SWS in both genotypes. (h) In RUN, APP/PS1 had fewer low rate and high rate cells; In QW and SWS, APP/PS1 had more low rate cells and fewer high rate cells. In REM, APP/PS1 had fewer high rate cells. * p < 0.05, ** p < 0.01, ***p < 0.001; c, one-way ANOVA with post-hoc comparisons; d, Pearson correlation; e, f, t-test; g, two-way ANOVA with post-hoc comparisons; h, mixed effects model. b-d: each group n = 4 mice; e-h: Control n = 5 mice, 395 place cells, 2169 total cells; APP/PS1 n = 6 mice, 237 place cells, 2133 total cells. Data are represented as mean ± S.E.M.

Fig. 2

Young APP/PS1 mice showed reduced place cell pairwise reactivation but intact oscillations of slow wave sleep. (a) Calcium transient time cross-correlations of 30 randomly selected place cells from one control and one APP/PS1 mouse, sorted by place field location on the track (colorbar). Grey line: coactivation. (b) The percentage of place cell pairs that were temporally reactivated in slow wave sleep (SWS) was reduced in APP/PS1 compared to control mice, but not in quiet wakefulness (QW) or REM sleep (REMs). (c) In SWS, power in the cortical slow oscillation, delta, spindle (left panel) and hippocampal ripple band (right panel) was comparable between genotypes. Inset: cortical slow oscillation and delta power. (d) Cortical spindle-triggered ripple power was comparable between genotypes. **p < 0.01, two-way ANOVA with post-hoc comparison. Control n = 5 mice, 395 place cells, 2169 total cells; APP/PS1 n = 6 mice, 237 place cells, 2133 total cells. Data are represented as mean ± S.E.M.

Fig. 3

In pre-plaque APP/PS1 mice, theta-gamma phase-amplitude coupling in exploratory behavior was impaired and dynamic calcium fluctuations were excessively synchronized with hippocampal theta. (a) In RUN, hippocampal theta and gamma power in APP/PS1 and control mice were comparable, and hippocampal theta power was not significantly correlated with age in either group. (b) Intra-hippocampal theta (6–8 Hz) - gamma (40–80 Hz) phase - amplitude coupling (PAC) was reduced in APP/PS1 compared to control mice. (c) Intrahippocampal theta-gamma PAC was negatively correlated with the age of APP/PS1 mice but not control mice. (d) Cortical theta power was reduced in APP/PS1 mice while cortical gamma power was unchanged. Cortical theta power was not significantly correlated with age in either group. (e) PAC of hippocampal theta (6–8 Hz) - cortical gamma (40–80 Hz) was reduced in APP/PS1 compared to control mice. (f) The reduction of hippocampal theta- cortical gamma PAC was negatively correlated with age in APP/PS1 but not control mice. (g) In APP/PS1 but not controls, the extent of place cell pairwise reactivation in SWS correlated with the magnitude of intra-hippocampal theta-gamma PAC in RUN. (h) The proportion of reactivated place cell pairs in SWS was not correlated with age. (i) Average theta-range LFP-cellular coherence of place cells was higher in APP/PS1 than control mice. *p < 0.05; a, c, d, f, g, h Pearson correlation; a, b, d, e, 2 sided t-test; i, two-way ANOVA with post-hoc comparisons. a – f: Control n = 7 mice, APP/PS1 n = 8 mice; g – i: Control n = 5 mice, 395 place cells, 1774 non-place cells, 2169 total cells; APP/PS1 n = 6 mice, 237 place cells, 1896 non-place cells, 2133 total cells. Data are represented as mean ± S.E.M. pc, place cells. Non-pc, nonplace cells.

Fig. 4

In young APP/PS1 mice, theta-gamma phase amplitude coupling in REM sleep was intact and correlated with theta-gamma phase-amplitude coupling during exploratory behavior. (a) In REM sleep, hippocampal theta and gamma power were comparable between groups. Hippocampal theta power of APP/PS1 but not control mice was negatively correlated with age. (b) Intra-hippocampal theta (6–8 Hz) - gamma (40–80 Hz) phase-amplitude coupling (PAC) was comparable in APP/PS1 and control mice. (c) Intrahippocampal theta-gamma PAC showed only a trending negative correlation with the age of APP/PS1 mice but not control mice. (d) Cortical theta and gamma power were comparable between groups. Cortical theta power of APP/PS1 mice but not control mice was negatively correlated with age. (e) PAC of hippocampal theta (6–8 Hz) - cortical gamma (40–80 Hz) was comparable across genotypes. (f) Hippocampal theta- cortical gamma PAC was negatively correlated with age in APP/PS1 mice but not control mice. (g) Intra-hippocampal theta-gamma PAC between RUN and REM sleep showed a trending correlation in both genotypes. (h) Hippocampal theta - cortical gamma PAC in RUN and in REM sleep were significantly correlated in both APP/PS1 and control mice. (i) Average theta-range LFP-cellular coherence in REM sleep was comparable across APP/PS1 and control mice. a, c, d, f, g, h, Pearson correlation; a, b, d, e, 2 sided t-test; i, two-way ANOVA with post-hoc comparisons. a – g: Control n = 7 mice, APP/PS1 n = 8 mice; i: Control n = 5 mice, 395 place cells, 2169 total cells; APP/PS1 n = 6 mice, 237 place cells, 2133 total cells. Data are represented as mean ± S.E.M. pc, place cells. Non-pc, nonplace cells.