Genetic models of cleavage-reduced and soluble TREM2 reveal distinct effects on myelination and microglia function in the cuprizone model

Abstract

Triggering receptor expressed on myeloid cells 2 (TREM2) is a cell-surface immunoreceptor expressed on microglia, osteoclasts, dendritic cells and macrophages. Heterozygous loss-of-function mutations in TREM2, including mutations enhancing shedding form the cell surface, have been associated with myelin/neuronal loss and neuroinflammation in neurodegenerative diseases, such as Alzheimer`s disease and Frontotemporal Dementia. Using the cuprizone model, we investigated the involvement of soluble and cleavage-reduced TREM2 on central myelination processes in cleavage-reduced (TREM2-IPD), soluble-only (TREM2-sol), knockout (TREM2-KO) and wild-type (WT) mice. The TREM2-sol mouse is a new model with selective elimination of plasma membrane TREM2 and a reduced expression of soluble TREM2. In the acute cuprizone model demyelination and remyelination events were reflected by a T2-weighted signal intensity change in magnetic resonance imaging (MRI), most prominently in the external capsule (EC). In contrast to WT and TREM2-IPD, TREM2-sol and TREM2-KO showed an additional increase in MRI signal during the recovery phase. Histological analyses of TREM2-IPD animals revealed no recovery of neuroinflammation as well as of the lysosomal marker LAMP-1 and displayed enhanced cytokine/chemokine levels in the brain. TREM2-sol and, to a much lesser extent, TREM2-KO, however, despite presenting reduced levels of some cytokines/chemokines, showed persistent microgliosis and astrocytosis during recovery, with both homeostatic (TMEM119) as well as activated (LAMP-1) microglia markers increased. This was accompanied, specifically in the EC, by no myelin recovery, with appearance of myelin debris and axonal pathology, while oligodendrocytes recovered. In the chronic model consisting of 12-week cuprizone administration followed by 3-week recovery TREM2-IPD displayed sustained microgliosis and enhanced remyelination in the recovery phase. Taken together, our data suggest that sustained microglia activation led to increased remyelination, whereas microglia without plasma membrane TREM2 and only soluble TREM2 had reduced phagocytic activity despite efficient lysosomal function, as observed in bone marrow-derived macrophages, leading to a dysfunctional phenotype with improper myelin debris removal, lack of remyelination and axonal pathology following cuprizone intoxication.

Keywords: Cuprizone; MRI; Microglia; Neuroinflammation; Phagocytosis; TREM2.

Fig. 1

Characterization of BMDM from wild-type, TREM2-IPD, TREM2-sol and TREM2-KO mice. a Schematic representation of the TREM2 receptor. Black, red and yellow asterisks indicate the cleavage site in WT, the mutated cleavage site in TREM2-IPD and in TREM2-sol, respectively. b Flow cytometry analysis of murine cell surface TREM2 on BMDM. MFI: median fluorescence intensity. Sheddase inhibitor: DCP333 (DPC), sheddase activator PMA. c Analysis of supernatants from b of murine soluble TREM2 from BMDM. d ATP-based cell survival assay of BMDM upon M-CSF deprivation for 2 and 3.5 days. ATP levels of cells cultured with M-CSF (n = 7) were set as 100% survival and compared to the ATP concentration after 2 (n = 4) and 3.5 days (n = 3) without M-CSF for each genotype. Statistics: Holm–Sidak’s two-way ANOVA multiple comparisons (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). e In vitro phagocytosis capacity of BMDM over 12 h (area-under-the curve) with 5 µg pHrodo-myelin per well (n = 3). Fluorescence measurements in wells without prey were used as controls (data not shown). f Representative images of the Cathepsin B activity assay taken by the In-Cell Analyzer. The nuclei are stained with DAPI (blue), and the red fluorescence signals are derived from cleaved Magic red. g Quantification of the Cathepsin B assay images. The fluorescence integrated density of the Magic red signal was measured and normalized to the nuclei count. A significant (p < 0.05) increase in normalized fluorescence between the DMSO control and K-18 within one genotype is marked by #. Statistics for d and g: Holm–Sidak’s two-way ANOVA with multiple comparisons (*p ≤ 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001). Statistics for e: One-way ANOVA test with Holm–Sidak’s multiple comparisons test (***p < 0.001; ****p < 0.0001). All data are presented as means ± SEM

Fig. 2

MRI indicated myelination deficits in TREM2-sol and TREM2-KO in the acute cuprizone model. a Schematic diagram of the experimental setup for the cuprizone treatment and recovery. Groups consisted of mice treated for 5 weeks with control food or 0.2% cuprizone in food and then switched back to control food (normal food) for the 4-week recovery. MRI measurements were performed at week 0 (baseline), week 3 (except for TREM2-KO and wt2) and week 5 of cuprizone intoxication, at week 7 (2 weeks of recovery on control food, except for TREM2-KO and wt2) and at week 9 (4 weeks of recovery on control food). Mice were culled at week 9 immediately after the last MRI measurement. b Representative MRI images acquired from three mice at baseline, at maximal pathology (5 weeks of receiving 0.2% cuprizone) and at recovery (4 weeks after switching to control food) for the different genotypes. c Corresponding T2-weighted MRI signal intensity (relative to the signal intensity at baseline) in external capsule (EC). Group sizes: n = 7–9 for all genotypes and timepoints. Data are shown as means ± SEM. Statistics: ANOVA with random effects comparisons indicated significant differences with respect to WT mice: *0.01 < p < 0.05, ***0.0001 < p < 0.001, ****p < 0.0001. For each group examined, T2-weighted signals were significantly increased with respect to baseline values (significances not shown). d Analysis of TREM2 levels in the brain for mice receiving control food (ctrl), at peak of cuprizone intoxication (week 5, cpz) and after 4-week recovery (rec). n.d. not detected. Statistics: Holm–Sidak’s multiple comparison test one-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001, ⁺⁺⁺⁺p < 0.0001 to the respective wt group). Wt1, as well as wt ctrl1, wt cpz1 and wt rec1 are the respective wild-type groups for the study with TREM2-IPD and TREM2-sol, wt2 is the wild-type group for the TREM2-KO study

Fig. 3

TREM2-sol and TREM2-KO display myelin debris, lack of remyelination and axonal pathology in the EC. Representative pictures for the different genotypes and timepoints from histological stainings detecting a myelin with Luxol Fast Blue (LFB) and corresponding quantitative optical density (OD) analysis of LFB in the EC (normalized to WT at control food), b mature oligodendrocytes (GST-π) and corresponding image analysis in EC (GST-π soma area in %), c myelin basic protein debris (dMBP) and corresponding image analysis in EC (dMBP-stained area in %), d neurofilament (SMI312) and corresponding image analysis in EC (SMI312-stained area in %). Group sizes: n = 3-7 for all genotypes and timepoints. Data shown as means ± SEM. wt: wild-type, TREM2-IPD: TREM2 cleavage-reduced, TREM2-sol: TREM2 soluble-only, TREM2-KO: TREM2 knockout. Ctrl: control food, cpz: cuprizone food for 5 weeks, rec: recovery on control food for 4 weeks. EC: external capsule, CC: corpus callosum. Scale bars: 300 µm (overview), 50 µm (close-up). Statistics: Holm–Sidak`s multiple comparison test one-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Comparisons not indicated are non-significant. Wt ctrl1, wt cpz1 and wt rec1 are the respective wild-type groups for the study with TREM2-IPD and TREM2-sol, Wt ctrl2, wt cpz2 and wt rec2 are the wild-type groups for the TREM2-KO study. Only statistical analysis within a study was performed. For c the analysis of the respective wt group for TREM2-KO was omitted as no dMBP signal was observed

Fig. 4

Increase of NF-L in plasma from TREM2-sol and TREM2-KO in the acute cuprizone model. NF-L measurements in plasma of WT, TREM2-IPD, TREM2-sol and TREM2-KO mice receiving control food (ctrl), at 5 weeks of cuprizone intoxication (cpz) and at 4-week recovery on normal food (rec). Group sizes: wt ctrl (n = 4), wt cpz (n = 4), wt rec (n = 4), TREM2-IPD ctrl (n = 3), TREM2-IPD cpz (n = 7), TREM2-IPD rec (n = 7), TREM2-sol ctrl (n = 4), TREM2-sol cpz (n = 7), TREM2-sol rec (n = 4), TREM2-KO cpz (n = 4). One-way ANOVA Holm–Šídák's multiple comparisons test, *p < 0.05, ***p < 0.001, ****p < 0.0001. Comparisons not indicated are non-significant. Wt ctrl1, wt cpz1 and wt rec1 are the respective wild-type groups for the study with TREM2-IPD and TREM2-sol, wt ctrl2, wt cpz2 and wt rec2 are the wild-type groups for the TREM2-KO study. Only statistical analysis within a study was performed

Fig. 5

TREM2-IPD and TREM2-sol mice show both sustained microglia/astrocyte activation and enhanced LAMP-1 in the EC. Representative images for the different genotypes and timepoints from histological stainings detecting a Iba1 and corresponding image analysis of Iba1-positive soma numbers (normalized to WT at week 5 cuprizone), b astrocytes (GFAP) and corresponding image analysis (GFAP-stained area in %), c LAMP-1 (lysosomal-associated membrane protein 1) and corresponding image analysis (LAMP1-stained area in %), as well as d TMEM119 (homeostatic marker) and corresponding image analysis (TMEM119-stained area in %). Group sizes: n = 2-7 for all genotypes and timepoints. Data are shown as means ± SEM. WT: wild-type, TREM2-IPD: TREM2 cleavage-reduced, TREM2-sol: TREM2 soluble-only, TREM2-KO: TREM2 knockout. Ctrl: control food, cpz: cuprizone food for 5 weeks, rec: recovery on control food for 4 weeks. EC: external capsule. CC: corpus callosum. Scale bars: 300 µm (overview), 50 µm (close-up). Statistics: Holm–Sidak’s multiple comparison test one-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Comparisons not indicated are non-significant. wt ctrl1, wt cpz1 and wt rec1 are the respective wild-type groups for the study with TREM2-IPD and TREM2-sol, wt ctrl2, wt cpz2 and wt rec2 are the wild-type groups for the TREM2-KO study. Only statistical analysis within a study was performed. For d the analysis of the respective wt group for TREM2-KO was omitted as no relevant TMEM119 signal was observed

Fig. 6

Brain cytokine/chemokine response is reduced in TREM2-sol and TREM2-KO, but enhanced in TREM2-IPD. Cytokine/chemokine measurements in brain detecting MIP-1a, MIP-1b, IP-10 and MCP-1 in WT, TREM2-IPD, TREM2-sol and TREM2-KO with control food (ctrl), at 5 weeks of cuprizone intoxication (cpz) and at 4-week recovery on normal food (rec). Measurements are normalized to wt cpz. Group sizes: wt ctrl (n = 4), wt cpz (n = 4), wt rec (n = 4), TREM2-IPD ctrl (n = 4), TREM2-IPD cpz (n = 7), TREM2-IPD rec (n = 7), TREM2-sol ctrl (n = 4), TREM2-sol cpz (n = 7), TREM2-sol rec (n = 4), TREM2-KO ctrl (n = 7), TREM2-KO cpz (n = 4), TREM2-KO rec (n = 7). Statistics: ordinary one-way ANOVA Holm–Šídák's multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Comparisons not indicated are non-significant. wt ctrl1, wt cpz1 and wt rec1 are the respective wild-type groups for the study with TREM2-IPD and TREM2-sol, wt ctrl2, wt cpz2 and wt rec2 are the wild-type groups for the TREM2-KO study. Only statistical analysis within a study was performed

Fig. 7

TREM2-IPD showed enhanced myelination in the chronic cuprizone model. a Schematic diagram of the experimental setup for the cuprizone treatment and recovery. Groups consisted of mice treated for 12 weeks with control food (normal food) or 0.2% cuprizone in food and then switched back to control food for the 3-week recovery. MRI measurements were performed at timepoints indicated. Mice were culled at week 15 immediately after the last MRI measurement. b T2-weighted signals in the CC and EC during the 12-week intoxication period and the recovery phase were significantly increased with respect to baseline values and compared to analyses for mice receiving control diet throughout the experiment. The significance levels ^#0.01 < p < 0.05 and ^###p < 0.001 correspond to ANOVA with random effects comparisons between WT and TREM2-IPD animals. Representative images for the different genotypes and at week 15 from histological stainings detecting c myelin with Luxol Fast Blue (LFB) and corresponding quantitative optical density analysis of LFB in the EC and CC, and d mature oligodendrocytes (GST-π) and corresponding image analysis in EC and CC (GST-π soma area in %). Group sizes: n = 5–7 for all genotypes and timepoints. Male mice were used for the cuprizone groups. Data are shown as means ± SEM. WT: wild-type, TREM2-IPD: TREM2 cleavage-reduced. Ctrl: control food, rec: recovery on control food for 3 weeks. Control refers to TREM2-IPD mice receiving normal food throughout the study. EC: external capsule, CC: corpus callosum. Scale bars: 500 µm. Statistics: ordinary one-way ANOVA Holm–Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant