Alzheimer risk-increasing TREM2 variant causes aberrant cortical synapse density and promotes network hyperexcitability in mouse models

Abstract

The R47H variant of triggering receptor expressed on myeloid cells 2 (TREM2) increases the risk of Alzheimer's disease (AD). To investigate potential mechanisms, we analyzed knockin mice expressing human TREM2-R47H from one mutant mouse Trem2 allele. TREM2-R47H mice showed increased seizure activity in response to an acute excitotoxin challenge, compared to wildtype controls or knockin mice expressing the common variant of human TREM2. TREM2-R47H also increased spontaneous thalamocortical epileptiform activity in App knockin mice expressing amyloid precursor proteins bearing autosomal dominant AD mutations and a humanized amyloid-β sequence. In mice with or without such App modifications, TREM2-R47H increased the density of putative synapses in cortical regions without amyloid plaques. TREM2-R47H did not affect synaptic density in hippocampal regions with or without plaques. We conclude that TREM2-R47H increases AD-related network hyperexcitability and that it may do so, at least in part, by causing an imbalance in synaptic densities across brain regions.

Keywords: Alzheimer's disease; Amyloid precursor protein; EEG; Epilepsy; Glia; Hyperexcitability; Microglia; Network dysfunction; Synapses; TREM2.

Figure 1.. Increased epileptiform activity after excitotoxin challenge in TREM2^H/+ mice.

(A to H) Video-EEG recordings were obtained from 5–6-month-old female mice before and after four low-dose KA injections. Knockin mice were compared with WT controls from the respective line as indicated. (A) Representative EEG traces before (baseline) and after (seizures and interictal spikes) KA injections from a 5-month-old TREM2^H/+ mouse and an age-matched WT control. (B) Spike numbers per 30-min bins recorded during 7 h after the first KA injection, independent of whether spikes occurred during or between seizures. (C) Total number of spikes during 7 h after first KA injection. (D) Number of ictal spikes during 7 h after first KA injection. (E) Number of interictal spikes during 7 h after first KA injection. (F) Number of seizures during 7 h after first KA injection; no seizure lasted >5 min. (G) Cumulative amount of time for which mice had seizures during 7 h after first KA injection. (H) Maximal spike frequency reached within any 30-min bin during 7 h after first KA injection. n = 9–12 mice per genotype. Data were analyzed by Welch’s one-way ANOVA (B), Kruskal-Wallis test (G), or generalized linear model (C to F, and H). These analyses revealed significant group differences in (B to E, and H; P < 0.01) but not (G). The number of seizures (F) showed a trend-level difference between TREM2^H/+ mice and WT mice from the TREM2^R/+ line. Pairwise comparisons of areas under the curve (AUC) in (B) revealed a significant difference between TREM2^H/+ mice and WT controls from the TREM2^H/+ line (P < 0.05) and a strong trend toward a difference between TREM2^H/+ and TREM2^R/+ mice (P = 0.06). ^#P = 0.09, *P < 0.05, **P < 0.01, ***P < 0.001 by Mann-Whitney test followed by Holm-Sidak correction (G) or pairwise comparisons of estimated marginal means followed by Holm-Sidak correction (C to F, and H). In this and the other figures, only statistically significant differences were indicated by brackets and asterisks. Dots in (C to H) represent individual mice. Dots with error bars in (B) and bars in (C to H) are means ± s.e.m.

Figure 2.. TREM2^H increases the frequency of thalamocortical SWDs in App^FAD mice.

(A to F) Electrophysiological recordings were obtained from freely behaving female and male mice at 10–12, 13–15, and 16–18 months of age. (A to C) Representative results from a single female TREM2^H/+/App^FAD mouse. (A) Electrode locations (left) and corresponding traces (right). The EEG was recorded from the primary somatosensory (S1) cortex, the differential local field potentials (dLFPs) from the hippocampus and thalamus, and the EMG from neck muscles. The spectrogram shown above the EMG trace was generated from the EEG recording. Note that the SWDs (pink shadings) in the cortex and thalamus were associated with changes in the spectrogram (see panel (C) for details) and with reduced EMG activity (indicating behavioral arrest). (B) Illustration of SWD detection. EEG signals were band-pass (BP) filtered at 6–10 Hz and an SWD detection threshold (Thr.) was applied as described in Methods. dB, decibels. Freq., frequency. (C) Power spectra calculated from SWDs (white) and baseline (grey) EEG activity. Note the dominant peak frequencies (8–9 and 16–18 Hz) during SWDs. (D) SWD frequencies detected in female and male mice of the indicated genotypes and ages. (E) Quantification of SWD frequency in female (empty dots) and male (black dots) mice at 13–15 months of age. No SWDs were detected in WT and TREM2^H/+ mice at this age. n = 5–8 mice per genotype and sex. (F) Percentage of female (empty bars) and male (grey bars) Trem2^+/+/App^FAD and TREM2^H/+/App^FAD mice with SWDs from (D). n = 11–16 mice per genotype. For (D), we calculated the area under the curve (AUC) and analyzed the data by Kruskal-Wallis test and post hoc Mann-Whitney tests. All pairwise comparisons revealed significant (P < 0.05) differences except for Trem2^+/+/App^+/+ vs. TREM2^H/+/App^+/+ (P = 0.13). For (E), we used a generalized linear mixed-effects model that treated age, Trem2 genotype, App genotype, and the interaction among them as fixed effects and individual mice as a random effect. **P < 0.01 based on a pairwise comparison of estimated marginal means followed by Holm-Sidak correction. For (F), Fisher’s exact test revealed a significant difference among groups (P < 0.01). *P < 0.05 by pairwise Fisher’s exact test followed by Holm-Sidak correction. Dots in (E) represent individual mice. Dots with error bars (D) and bars in (E) are means ± s.e.m. Bars in (F) represent percentage.

Figure 3.. TREM2^H does not alter plaque levels in the cortex or hippocampus of App^FAD mice.

(A to G) Coronal brain sections from 21–22-month-old female and male mice of the indicated genotypes were stained with ThioS (magenta) and DAPI (cyan) and the parietal cortex was imaged by wide-field microscopy. (A) Representative images of ThioS-positive plaques in the cortex of female TREM2^+/+/App^FAD (left) and TREM2^H/+/App^FAD (right) mice. Scale bar: 50 μm. (B) Number (No.) of plaques per mm² of cortical area. (C) Plaque size (area covered per plaque) in cortex. (D) Fluorescence intensity of plaques in cortex. (E) Number of plaques per mm² of hippocampal area. (F) Plaque size (area covered per plaque) in hippocampus. (G) Fluorescence intensity of plaques in hippocampus. n = 7–8 mice per group. Data were analyzed with Student’s t-tests. No significant genotype effects were identified. In (B to G), dots represent individual female (empty) and male (black) mice, and bars are means ± s.e.m.

Figure 4.. TREM2^H increases cortical synapse density.

(A to G) Coronal brain sections from 21–22-month-old female and male mice of the indicated genotypes were immunolabeled for the presynaptic marker bassoon (green) and the postsynaptic marker homer 1 (magenta), and stained for amyloid plaques with ThioS. The parietal cortex was imaged by confocal microscopy. Synaptic puncta were operationally defined as pairs of bassoon puncta and homer 1 puncta that were <0.5 μm apart. In Trem2^+/+/App^FAD and TREM2^H/+/App^FAD mice, we compared cortical regions that were distal (>20 μm away from plaque perimeter) or adjacent (<10 μm from plaque perimeter) to plaques. (A) Representative images from the cortex of a female WT mouse illustrate our approach to identifying bassoon-positive presynaptic puncta (green), homer 1-positive postsynaptic puncta (magenta), and synaptic puncta (merge and inset). Scale bars: 5 μm, 1 μm (inset). White asterisks in inset indicate synaptic puncta. (B to D) Relative densities of bassoon puncta (B), homer 1 puncta (C), and synaptic puncta (D) in plaque-free regions of the cortex in female (empty dots) and male (black dots) mice. For each measure, the mean density in WT mice was defined as 100%. (E to G) Relative densities of bassoon puncta (E), homer 1 puncta (F), and synaptic puncta (G) in cortical regions that were distal (Dist., data from (B to D)) or adjacent (Adj.) to plaques in female (empty dots) and male (black dots) mice. For each measure, the mean density distal to plaques in Trem2^+/+/App^FAD mice was defined as 100%. n = 5–12 mice per group. For (B to D), two-way ANOVA revealed a significant effect of the Trem2 genotype (P < 0.001) but not the App genotype (P > 0.45) and no interaction between them (P > 0.34). For (e–g), two-way ANOVA revealed significant effects of the Trem2 genotype (P < 0.01) and plaque adjacency (P < 0.01) and a significant interaction between them for (F and G; P < 0.05) but not (E; P = 0.69). *P < 0.05, **P < 0.01, ***P < 0.001 by Holm-Sidak test. In (B to G), dots represent individual female (empty) and male (black) mice. Bars are means ± s.e.m.

Figure 5.. TREM2^H does not alter hippocampal synapse density.

(A to F) Coronal brain sections from 21–22-month-old female and male mice of the indicated genotypes were immunolabeled for bassoon and homer 1, and stained for amyloid plaques with ThioS. The hippocampus was imaged by confocal microscopy. Conventions were otherwise as in Fig. 4. (A to C) Relative densities of bassoon puncta (A), homer 1 puncta (B), and synaptic puncta (C) in plaque-free areas of the hippocampal CA1 region in female (empty dots) and male (black dots) mice. For each measure, the mean density in WT mice was defined as 100%. (D to F) Relative densities of bassoon puncta (D), homer 1 puncta (E), and synaptic puncta (F) in CA1 areas that were distal (Dist., data from (A to C)) or adjacent (Adj.) to plaques in female (empty dots) and male (black dots) mice. For each measure, the mean density distal to plaques in Trem2^+/+/App^FAD mice was defined as 100%. n = 5–12 mice per group. For (A to C), two-way ANOVA revealed no significant effects of the Trem2 (P > 0.46) or App (A: P = 0.0852; B, C: P > 0.13) genotypes and no interaction between them (P > 0.13). For (D to F), two-way ANOVA revealed no effect of the Trem2 genotype (P > 0.45) but a significant effect of plaque adjacency (D to F; P <0.01). *P < 0.05, **P < 0.01 by Tukey test. In (A to F), dots represent individual female (empty) and male (black) mice. Bars are means ± s.e.m.