TET2-mutant myeloid cells mitigate Alzheimer's disease progression via CNS infiltration and enhanced phagocytosis in mice

Abstract

Clonal hematopoiesis (CH) is associated with many age-related diseases, but its interaction with Alzheimer's disease (AD) remains unclear. Here, we show that TET2-mutant CH is associated with a 47% reduced risk of late-onset AD (LOAD) in the UK Biobank, whereas other drivers of CH do not confer protection. In a mouse model of AD, transplantation of Tet2-mutant bone marrow reduced cognitive decline and β-amyloid plaque formation, effects not observed with Dnmt3a-mutant marrow. Bone-marrow-derived microglia-like cells were detected at an increased rate in Tet2-mutant marrow recipients, and TET2-mutant human induced pluripotent stem cell (iPSC)-derived microglia were more phagocytic and hyperinflammatory than DNMT3A-mutant or wild-type microglia. Strikingly, single-cell RNA sequencing (scRNA-seq) revealed that macrophages and patrolling monocytes were increased in brains of mice transplanted with Tet2-mutant marrow in response to chemokine signaling. These studies reveal a TET2-specific protective effect of CH on AD pathogenesis mediated by peripheral myeloid cell infiltration.

Keywords: Alzheimer’s disease; DNMT3A; TET2; clonal hematopoiesis; inflammation; microglia, peripheral immune cells; myeloid cell activation; phagocytosis.

Figure 1.. Loss of Tet2, but not Dnmt3a improves AD pathogenesis

(A) Odds ratios (OR) for overall CH and driver subtypes relative to incident LOAD in the UKB cohort, adjusted for age, sex, smoking and alcohol related behaviors, APOE genotype, and Townsend Deprivation Index (See also Figure S1A) (B-C) Frequency of the top 10 CH driver mutations in patients from the UKB. (B) All patients (C) patients with incident LOAD. (D+H) Mouse transplant model. Six-eight week old 5xFAD recipient mice were irradiated and transplanted with indicated WBM. Eight weeks after transplantation, mice were treated with weekly doses of LPS for an additional eight weeks. (D) Following transplant, recipient mice are referred to as WT^AD (5xFAD with Tet2^+/+ WBM) and Tet2^AD (5xFAD with Tet2^−/− WBM) or (H) WT^AD (5xFAD with Vav-iCre⁻/Dnmt3a^fl/fl WBM) and Dnmt3a^AD (5xFAD with Vav-iCre⁺/Dnmt3a^−/− WBM). (E-K) Behavior assays on transplanted mice following LPS treatment. (E-G) WT^AD and Tet2^AD mice, n=14–16. (I-K) WT^AD and Dnmt3a^AD mice, n=14. (E+I) Novel object recognition, shown as the ratio of time spent at novel object vs time spent at familiar object (% index). Performed following 2–3 doses of LPS at age 16–18 weeks. (F+J) Open field behavior, shown as the ratio of center to total distance. Performed following 2–3 doses of LPS at age 16–18 weeks. (G+K) Conditioned fear, shown as % cued freezing behavior. Performed following 7–8 weekly doses of LPS at age 22–25 weeks. (G) Two Tet2^AD mice died before completing CF testing (K) One outlier was removed from Dnmt3a^AD group using ROUT test, Q=1 (See also Figure S1J–K) (L-M) β-amyloid plaque staining in cortex region of the brain of (L) WT^AD and Tet2^AD mice, n=8–10 (M) WT^AD and Dnmt3a^AD mice, n=7–8, following 8 weekly doses of LPS (Left) Representative images. Scale bars are 100μm (Right) Quantification of the number of plaques per area of cortex in mm². β-amyloid was detected using Thermo Fisher Cat# 71–5800. Representative of two independent experiments. Bar graphs represent mean ± SD. Unpaired t test. **** p < 0.0001; ** 0.001 ≤ p < 0.01; * 0.01 ≤ p < 0.05; ns, not significant

Figure 2.. TET2 mutant microglia are hyperinflammatory and have enhanced phagocytosis

(A-D) Flow cytometry analysis of cells isolated from whole brain homogenates from (A-B) WT^AD and Tet2^AD mice, n=12 (C-D) WT^AD and Dnmt3a^AD mice, n=7–8, following 8 weekly doses of LPS. Microglia are defined as CD45⁺ CD11a⁻ CD11b⁺ Tmem119⁺ (A+C) % microglia, shown as % of total CD45⁺ cells (B+D) Mean fluorescence intensity (MFI) of microglia activation markers CD68, CD11c and CD45. Representative of two independent experiments. (See also Figure S2A–C) (E-F) Donor-derived microglia-like cells, defined as CD45.1/2⁺ CD11a⁺ CD11b⁺ Tmem119⁺. (E) Representative staining scheme (F) % of total microglia. n=10–11 (See also Figure S2D–E) (G-L) Human iPSC-derived microglia (iMGLs). (G) Schematic of generation of isogenic iPSC lines through CRISPR/Cas9-mediated gene editing of a common normal parental iPSC line. Resulting iMGLs are referred to as DNMT3A^R882H/WT and TET2^DelE3-E11/WT, respectively. (See also Figure S2F–G) (H-I) In vitro phagocytosis assays. MFI of the indicated genotype relative to isogenic WT iMGLs. Each data point represents an independent iMGL differentiation experiment from two iPSC lines per genotype. (H) iMGLs incubated with pHrodo-Red labeled myelin, n=5–6 (I) iMGLs incubated with Fluor 488-labeled amyloid-β peptide following 24hr treatment with 100ng/mL LPS, n=3. (See also Figure S2H). (J) Volcano plot of differentially expressed genes between TET2^DelE3-E11/WT vs DNMT3A^R882H/WT iMGLs following 24hr treatment with 100ng/mL LPS. Phagocytosis genes PTK2B, APBB3, SPP1, CCL2, RHOH, ABI3, GRN, PTK2 and P2RY12 are shown. (See also Figure SI-K) (K) Heatmap showing normalized Z-scores of cytokines measured by Luminex multiplex assay in the supernatant of iMGLs untreated or treated with 100ng/mL LPS for 24hrs (triplicate wells). (L) Expression levels of key chemokines CCL2, CCL7, CCL5 and CX3CL1 in iMGLs following 24hr treatment with 100ng/mL LPS. Bar graphs represent mean ± SD. (A-D + F) Unpaired t test. (H-I + L) One-way ANOVA with Tukey multiple comparison test. **** p < 0.0001; *** 0.0001 ≤ p < 0.001; ** 0.001 ≤ p < 0.01; * 0.01 ≤ p < 0.05; ns, not significant

Figure 3.. Loss of Tet2 leads to increased peripheral myeloid cell infiltration to the brain

(A-G) Flow cytometry analysis of bone-marrow derived infiltrating immune cells from whole brain homogenates from (B+D+F) WT^AD and Tet2^AD mice, n=12 (C+E+G) WT^AD and Dnmt3a^AD mice, n=7–8, following 8 weekly doses of LPS. (A) Representative flow plot showing range of % CD45⁺ CD11a⁺ of total cells (B-C) Total infiltrating cells, shown as % CD11a⁺ of total CD45⁺. (D-E) Infiltrating CD4⁺ and CD8⁺ T cells, shown as % CD4⁺/CD8⁺ CD3⁺CD11b⁻ of total CD45⁺ CD11a⁺ cells. (E) Outlier removed from CD4⁺ Dnmt3a^AD group, ROUT test Q1 (F-G) Infiltrating CD11b+ myeloid cells, shown as % CD11b⁺Tmem119⁻ of total CD45⁺ CD11a⁺ cells. (G) Outlier removed from WT^AD group, ROUT text Q=1. Representative of two independent experiments (See also Figure S3A) (H) In vitro phagocytosis assay on CD11b-enriched myeloid cells from WBM of WT, Dnmt3a^−/− and Tet2^−/− mice. Treated for 24hrs with 100ng/mL LPS and incubated with Fluor 488-labeled amyloid-β peptide. Shown as fold change over WT, n=5. Representative of three independent experiments. (I-L) Single cell RNA sequencing (scRNA seq) on CD11a-enriched cells isolated from whole brain homogenates from WT^AD, Tet2^AD and Dnmt3a^AD mice following 8 weekly doses of LPS. (I) Uniform manifold approximation and projection (UMAP) embeddings of all cells with major cell types annotated in the main cohort. (J) UMAP embeddings of major cell types with Milo differential abundance testing results in the comparison of Tet2^AD versus WT^AD (J) and Dnmt3a^AD versus WT^AD (K). The notes represent neighborhoods constructed based on transcriptional similarities between cells, colored by the abundance log2 fold changes between genotypes. (L) Proportion of cells in each genotype in the main cohort. (See also Figure S3B) (M) Proportion of cells in each population from scRNAseq on CD45-enriched cells from the brains of an independent validation cohort of 5xFAD mice that were transplanted with bone marrow from Vav-Cre⁺ wildtype (WT^AD−2), Vav-Cre⁺ Tet2^−/− (Tet2^AD−2) or Vav-Cre⁺ Dnmt3a^−/− (Dnmt3a^AD−2) mice without LPS treatment. (See also Figure S3C–E) Bar graphs represent mean ± SD. Unpaired t test **** p < 0.0001; *** 0.0001 ≤ p < 0.001; ** 0.001 ≤ p < 0.01; * 0.01 ≤ p < 0.05; ns, not significant.

Figure 4.. Tet2 loss leads to an enhanced migratory phenotype in monocytes and macrophages

(A, C-F) scRNA seq on CD11a-enriched cells isolated from whole brain homogenates from WT^AD, Tet2^AD and Dnmt3a^AD mice following 8 weekly doses of LPS. (A) Raincloud plots showing the pseudo-bulk expression of chemokine-related genes Ccr2, Ccr1, Ccl4, Ccl6, Cxcl2, Cxcl10 and Vegfa in macrophages. (B) Flow cytometry analysis of infiltrating Ccr2+ myeloid cells from whole brain homogenates of WT^AD, Tet2^AD and Dnmt3a^AD mice following 8 weekly doses of LPS. Shown as % Ccr2+ CD11b+ of CD11a+ cells. (C) UMAP embeddings of subclasses of macrophages, monocytes and microglia. (D) Violin plot showing log2 fold changes of cell types grouped by subclass in Tet2^AD vs WT^AD (upper panel) and Dnmt3a^AD vs WT^AD (lower panel). (E-F) GO Terms differentially regulated in non-classical monocytes. (E) Tet2^AD vs WT^AD (F) Dnmt3a^AD vs WT^AD. Bar graphs represent mean ± SD. Unpaired t test **** p < 0.0001; *** 0.0001 ≤ p < 0.001; ** 0.001 ≤ p < 0.01; * 0.01 ≤ p < 0.05; ns, not significant.