The gain-of-function TREM2-T96K mutation increases risk for Alzheimer's disease by impairing microglial function

Abstract

We previously reported that T96K is a gain-of-function mutation in TREM2 based on its ability to increase ligand-dependent activation. Here, we show that TREM2^T96K increases risk for Alzheimer's disease (AD) in a whole-genome sequencing dataset comprised of family-based and case-control samples. Trem2^T96K also reduced clustering of microglia around β-amyloid (Aβ) plaques exclusively in female 5xFAD mice. Furthermore, T96K decreased levels of soluble Trem2 in female 5xFAD mice and human microglial cell cultures. We also observed impaired uptake of Aβ in Trem2^T96K knockin microglial cells. Moreover, Trem2^T96K reduced total area of phagocytic microglia, specifically in female 5xFAD mice. Single-cell RNA sequencing (scRNA-seq) profiling of microglia revealed that Trem2^T96K impairs the transition of homeostatic microglia into disease-associated microglia (DAM) in female 5xFAD mice. Downregulated inflammatory pathways associated with Trem2^T96K included interleukin (IL)-6/JAK/STAT3, complement, and interferon (IFN)-γ response. Collectively, our results indicate that, like the loss-of-function mutation R47H, Trem2^T96K adversely affects microglial function in a sex-dependent manner.

Keywords: Alzheimer’s; T96K; TREM2; amyloid; gain of function; innate immunity; microglia; neuroinflammation; single-cell RNA-seq; uptake.

Figure 1.. The T96K mutation in TREM2 is associated with increased risk for AD

(A) Association statistics of the T96K mutation in TREM2 with AD in the WGS data of three ADSP case-control subcohorts (Non-Hispanic White [NHW], African American [AA], and Hispanic [HISP]) and NIMH families. (B) Association statistics of the R47H mutation in TREM2 with AD in the WGS data of four ADSP case-control subcohorts (NHW, AA, HISP, and Asian [ASIA]) and NIMH families. (C) Number and frequency of individuals with G/G (TREM2^+/+), G/T (TREM2^T96K/+), and T/T (TREM2^T96K/T96K) alleles in indicated subcohorts of ADSP. (D) Graphs show the proportion of AD individuals (affected) without (G/G, non-carriers) and with (G/T and T/T, carriers) TREM2^T96K mutation in the ADSP cohort (ALL) and indicated subcohorts. (E) Boxplots show the median age of onset in AD individuals (affected) with (G/T and T/T, carriers) TREM2^T96K mutation in the ADSP cohort (ALL) and indicated subcohorts. Mean ± SD (D); Fisher test (A–C), Wilcoxon rank sum test (E).

Figure 2.. The T96K mutation does not impact Trem2 expression levels or Aβ plaque morphology in 5xFAD mice

(A) Sequence alignment of Trem2 in indicated species. (B) Targeting strategy for the generation of Trem2^T96K knockin mice. The single-guide RNA (sgRNA), protospacer adjacent motif (PAM, CGG here), codons for T96K, nucleotides for the silent mutation, and restriction sites for HindIII and AflII enzymes are indicated. (C) Representative gel image for genotyping of Trem2^T96K mice. (D) Representative images and their corresponding zoom-ins of cortical fields from female mice of indicated genotypes, labeled with anti-Trem2 (green), anti-Iba1 (purple), and 3D6 antibody (red) for Aβ plaques. Scale bar represents 100 μm, and 25 μm for zoom-ins. (E) Expression levels of Trem2 were summarized as log₂FC of RNA-seq profiling of microglia isolated from 8-month-old female mice. The adjusted p value is shown. Trem2^+/+ (5F), Trem2^T96K/+ (4F), Trem2^T96K/T96K (3F), 5xFAD;Trem2^+/+ (6F), 5xFAD;Trem2^T96K/+ (9F), and 5xFAD;Trem2^T96K/T96K (7F) mice. FC, fold change. (F) Representative images of 3D6 and methoxy-X04 staining in the cortex of 8-month-old female 5xFAD;Trem2^+/+ mice. Scale bar represents 20 μm, and 5 μm for zoom-ins. (G) Percentages of compact, dynamic, and diffuse plaques for each genotype in the cortex of 8-month-old female mice. (F and G) 5xFAD;Trem2^+/+ (13F), 5xFAD; Trem2^T96K/+ (13F), and 5xFAD;Trem2^T96K/T96K (13F) mice. Data are represented as mean ± SEM; one-way ANOVA, Tukey’s test. See also Figures S1 and S2.

Figure 3.. Trem2^T96K leads to reduced total area of Iba1⁺ microglia and limits microglial association with Aβ plaques in 5xFAD mice in a sex-dependent manner

(A and D) Brain sections were labeled with anti-Iba1 (green) and 3D6 antibody (red) for Aβ plaques. Representative images of cortex (A) and hippocampus (D), and their respective zoom-ins, from 8-month-old female mice. Scale bar represents 100 μm, and 25 μm for zoom-ins. (B and E) Quantification of Iba1⁺ microglia area normalized to total area (%) in the cortex (B) and hippocampus (E) of female mice of indicated genotypes. (C and F) Graphs showing the interaction between sex and genotype for the total area of Iba1⁺ microglia in the cortex (C) and hippocampus (F). (G and J) Representative images of the cortex (G) and hippocampus (J) from 8-month-old female mice, processed with Imaris. Iba1 signal is shown in gray, while Aβ plaques are in red. Scale bar represents 100 μm. (H and K) Graphs show frequencies of Iba1⁺ microglia associated with plaques in the cortex (H) and hippocampus (K) of female mice of indicated genotypes. Each dot represents the number of microglia clustering around one plaque. 99–108 and 53–65 plaques in the cortex and hippocampus, respectively, were analyzed per mouse. (I and L) Graphs represent the interaction between sex and genotype for the clustering of Iba1⁺ microglia around plaques in the cortex (I) and hippocampus (L). 5xFAD;Trem2^+/+ (14F/25M), 5xFAD;Trem2^T96K/+ (13F/25M), and 5xFAD;Trem2^T96K/T96K (13F/22M) mice. Mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; one-way ANOVA, Tukey’s test. For sex-specific effects data in (C), (F), (I), and (L), ****p < 0.0001, two-way ANOVA and Šidák’s multiple comparison test. See also Figures S3 and S4.

Figure 4.. The T96K mutation reduces levels of soluble Trem2 in female 5xFAD mice and human microglial cells and impairs Aβ42 uptake in knockin microglial cell lines

(A) Representative immunoblots of sTrem2 in TBS-soluble fractions and full-length Trem2 in cortical tissue lysates from 8-month-old female mice. β-actin served as loading control. (B and C) Densitometry quantifications of sTrem2/full-length Trem2 ratio in 8-month-old male (B) and female (C) mice of indicated genotypes. (D) Graph showing the assessment of interaction between sex and genotype for the sTrem2/full-length Trem2 ratio in the cortex of 8-month-old mice. For (B)–(D), 5xFAD;Trem2^+/+ (25M/11F), 5xFAD;Trem2^T96K/+ (25M/11F), and 5xFAD;Trem2^T96K/T96K (22M/12F) mice. (E) Representative immunoblots of sTREM2 in conditioned media and full-length TREM2 in HMC3 whole-cell lysates. α-Tubulin served as loading control. (F) Densitometry quantifications of sTREM2/full-length TREM2 ratio in HMC3 cells stably expressing human TREM2^WT, TREM2^T96K, or TREM2^R47H. Each dot represents biological replicates (number of wells in a 6-well plate) analyzed from 4 independent experiments. (G) Representative flow cytometry plots of HMC3 cells stably expressing WT or mutant TREM2 constructs, labeled with N-terminal TREM2-PE antibody. (H) Quantification of the mean signal intensity (MSI) of PE⁺ (cell surface TREM2) HMC3 cells. Data represent 5 independent experiments, with samples run in biological replicates (10 data points per cell line). (I) Representative flow cytometry plots of BV2-Trem2^+/+, knockin BV2-Trem2^T96K/+, and BV2-Trem2^T96K/T96K cell lines that take up Aβ42-HiLexa™ Fluor 488. (J) Quantification of the percentage of Trem2^T96K knockin BV2 cells that showed uptake of fluorescently labeled Aβ42 in flow cytometry. (K) ELISA analysis of intracellular Aβ42 in cell lysates of Trem2^T96K knockin BV2 cells that were incubated with unlabeled Aβ42. For (J) and (K), n = 7–10 biological replicates, pooled from 4 independent experiments. Mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; Kruskal-Wallis ANOVA, Dunn’s test (B and C) and one-way ANOVA, Tukey’s test (F, H, J, and K). For sex-specific effects data in (D), two-way ANOVA and Šidák’s multiple comparison test. See also Figure S5.

Figure 5.. Trem2^T96K mutation leads to a sex-specific reduction in total area of phagocytic microglia in 5xFAD mice

(A–F) (A and B) Brain sections were labeled with anti-CD68 (green) and anti-Iba1 (red) antibodies. Representative images of cortex (A) and hippocampus (B) and their respective zoom-ins from 8-month-old female mice. Scale bar represents 100 μm, and 25 μm for zoom-ins. Quantification of area of CD68⁺Iba1⁺ microglia normalized to total area (%) in the cortex (C and E) and hippocampus (D and F) of female (C and D) and male (E and F) mice of indicated genotypes. (G and H) Graphs showing the interaction between sex and genotype for the total area of CD68⁺Iba1⁺ microglia in the cortex (G) and hippocampus (H). 5xFAD; Trem2^+/+ (14F/22M), 5xFAD;Trem2^T96K/+ (13F/22M), and 5xFAD;Trem2^T96K/T96K (13F/20M) mice. Data are represented as mean ± SEM; *p < 0.05; one-way ANOVA, Tukey’s test. For sex-specific effects data in (G) and (H), **p < 0.01, two-way ANOVA and Šidák’s multiple comparison test. See also Figure S6.

Figure 6.. Trem2^T96K results in the suppression of the DAM gene signature and downregulation of inflammatory pathways in female 5xFAD mice

(A) UMAP plot of 90,595 microglial cells, colored by unsupervised clusters including HOM (clusters 1–3), DAM1 (clusters 4–7), DAM2 (clusters 8–11), and IFN-responsive microglia (cluster 12) from 8-month-old female mice: 5xFAD;Trem2^+/+ (4F), 5xFAD;Trem2^T96K/+ (4F), 5xFAD;Trem2^T96K/T96K (5F), Trem2^+/+ (1F), and Trem2^T96K/+ (1F). (B) Relative cell cluster sizes and composition compared among 3 genotypes of female mice. The y axis: percentages of cells per cluster. (C and D) Dot plots representing the scaled average expression and fraction of expressing cells for top differentially expressed HOM (C) and DAM (D) genes in 5xFAD;Trem2^T96K/+ and 5xFAD;Trem2^T96K/T96K microglia versus 5xFAD;Trem2^+/+ (adjusted p value < 0.05). (E and F) GSEA plots showing enrichment score (x axis) and statistical significance (y axis) for top signaling pathways enriched in 5xFAD;Trem2^T96K/+ (E) and 5xFAD;Trem2^T96K/T96K microglia (F) versus 5xFAD;Trem2^+/+. See also Figures S7 and S8 and Data S2.

Figure 7.. Trem2^T96K mutation reduces total area of CD11c⁺ microglia and Itgax expression levels in female 5xFAD mice

(A and B) Brain sections were labeled with anti-Iba1 (green) and anti-CD11c (red) antibodies, and methoxy-X04 for Aβ plaques. Representative images of cortex (A) and hippocampus (B) and their respective zoom-ins from 8-month-old female mice of indicated genotypes. Scale bar represents 100 μm, and 25 μm for zoom-ins. (C and D) Quantification of area of CD11c⁺Iba1⁺ microglia normalized to total area (%) in the cortex (C) and hippocampus (D) of female mice of indicated genotypes. For (A)–(D), 5xFAD;Trem2^+/+ (9F), 5xFAD;Trem2^T96K/+ (8F), and 5xFAD;Trem2^T96K/T96K (8F) mice. (E) Graph showing Itgax mRNA levels assessed by quantitative real-time PCR and normalized to Gapdh in microglia isolated by fluorescence-activated cell sorting (FACS) from 8-month-old female mice. For (E), 5xFAD;Trem2^+/+ (14F), 5xFAD;Trem2^T96K/+ (10F), and 5xFAD;Trem2^T96K/T96K (10F) mice. For (C)–(E), data are represented as mean ± SEM; *p < 0.05; one-way ANOVA, Tukey’s test. See also Figure S8.

Figure 8.. Trem2^T96K leads to defects in the DAM trajectories of microglia in female 5xFAD mice

(A) UMAP plot of 77,666 microglial cells, colored by unsupervised clusters, including HOM (clusters 2, 3, 5, and 9), DAM1 (clusters 0, 7, 8, and 10), DAM2 (clusters 1, 4, and 6), and IFN-responsive microglia (cluster 11) from 8-month-old female mice: 5xFAD;Trem2^+/+ (4F), 5xFAD;Trem2^T96K/+ (4F), and 5xFAD; Trem2^T96K/T96K (5F). (B) Dot plot showing the top 3 enriched genes for each microglia cluster. (C) Average scaled expression levels of selected signature genes per cluster. (D) Log₂ FC of cluster sizes (cell fractions) in 5xFAD;Trem2^T96K/+ and 5xFAD;Trem2^T96K/T96K microglia relative to control 5xFAD;Trem2^+/+. (E) UMAP plot with the five inferred phenotypic cell trajectories: DAM2 trajectory 1 (5, 3, 7, 1, and 6); DAM1 trajectory 2 (5, 2, 8, 0, and 10); DAM2 trajectory 3 (5, 2, 8, and 4); HOM trajectory 4 (5, 2, 8, and 9); and IFN trajectory 5 (5 and 11). (F) Pseudotime distributions for microglia from each genotype along five cell trajectories identified in (E). 5xFAD;Trem2^T96K/+ and 5xFAD;Trem2^T96K/T96K showed impaired differentiation of microglial cells along trajectories 1 and 2, whereas they exhibited two distinct subsets of cells along trajectories 3, 4, and 5. See also Data S3.