Triggering receptor expressed on myeloid cells 2 (TREM2) regulates phagocytosis in glioblastoma

Abstract

Glioblastomas (GBMs) are tumors of the central nervous system that remain recalcitrant to both standard of care chemo-radiation and immunotherapies. Emerging approaches to treat GBMs include depletion or re-education of innate immune cells including microglia (MG) and macrophages (MACs). Here we show myeloid cell restricted expression of triggering receptor expressed on myeloid cells 2 (TREM2) across low- and high-grade human gliomas. TREM2 expression did not correlate with immunosuppressive pathways, but rather showed strong positive association with phagocytosis markers such as lysozyme (LYZ) and CD163 in gliomas. In line with these observations in patient tumors, Trem2^-/- mice did not exhibit improved survival compared to wildtype (WT) mice when implanted with mouse glioma cell lines, unlike observations previously seen in peripheral tumor models. Gene expression profiling revealed pathways related to inflammation, adaptive immunity, and autophagy that were significantly downregulated in tumors from Trem2^-/- mice compared to WT tumors. Using ZsGreen-expressing CT-2A orthotopic implants, we found higher tumor antigen engulfment in Trem2⁺ MACs, MG, and dendritic cells. Our data uncover TREM2 as an important immunomodulator in gliomas and inducing TREM2 mediated phagocytosis can be a potential immunotherapeutic strategy for brain tumors.

Keywords: TREM2; glioblastoma; microglia; phagocytosis; tumor associated macrophages.

Figure 1.. TREM2 is associated with myeloid cells and highly expressed in IDH-wt GBMs.

(A) Bar graphs showing TREM2 mRNA expression levels on lymphoid vs. myeloid cells across glioma types. NGB=non glioma brain, IMP=IDH-mut primary, IMR=IDH-mut recurrent, IWP=IDH-wt primary, IWR=IDH-wt recurrent. Wilcoxon signed-rank test was used to derive p values. (B) Violin plot of TREM2 expression in myeloid cell subsets as defined by Gupta, et al. (C) Representative staining of TREM2 IHC in primary and recurrent IDH-wt and IDH-mut gliomas. Scale bar: 20 µm. Table shows TREM2⁺ cells were classified as ramified or ameboid based on extent of branching. p value was calculated with Chi-square goodness of fit. (D) Representative IHC images of TREM2 in primary and recurrent IDH-wt and IDH-mut tumors. Scale bar: 20 µm. The contingency table shows the number of gliomas with <10%, 10–30%, or >30% TREM2⁺ cells in the tumor core. p value was calculated using Chi-square goodness of fit. (E) Kaplan-Meier curves from TCGA GBM datasets (analyzed via RNA-Seq, HG-U133A, and Agilent-4502A, respectively) showing no significant survival differences in TREM2 high vs. TREM2 low GBMs. For RNA-Seq, log-rank p value=0.6267 and Wilcoxon p value=0.83553. For HG-U133A, log-rank p value=0.0951 and Wilcoxon p value=0.1032. For Agilent-4502A, log-rank p value=0.2606 and Wilcoxon p value=0.8442.

Figure 2.. TREM2 is associated with phagocytosis in gliomas.

(A) Linear regression plots showing lack of correlation between TREM2 and immunosuppressive markers from TCGA GBM HG-U133A. r value was computed using Pearson’s product moment correlation. For all plots, p<0.001. (B) 2D density plot showing distribution of TREM2- and TREM2⁺ MG, MDMs, and MACs based on their phagocytosis and TREM2 pathway score. (C) Gating strategy showing TREM2 expression on both MG and non-MG myeloid subsets. MG are defined as CD11b⁺ P2RY12⁺, and non-MG myeloid cells are CD11b⁺ P2RY12-. In both MG and non-MG subsets, TREM2⁻ and TREM2⁺ populations are observed. (D) Heat map of median fluorescence intensity (MFI) values of LYZ and CD163 in TREM2⁺ and TREM2⁻ sorted myeloid populations across gliomas. Each row in the heatmap is data from a single patient. (E) Linear regression plots showing correlation of TREM2 expression with CD74, LYZ, and CD163 in TCGA GBM dataset HG-U133A. The r value was computed using Pearson’s product moment correlation. For all correlations, p<0.001.

Figure 3.. Deletion of Trem2 does not slow tumor growth in mouse models of glioma.

(A) Kaplan-Meier curve showing survival of wildtype and Trem2^−/− mice challenged with CT-2A. p=0.3072, Log-Rank (Mantel-Cox) test. (B) Kaplan-Meier curve of wildtype and Trem2^−/− mice bearing CT-2A-luc; p = 0.2837, Log-Rank (Mantel-Cox) test. (C) Bioluminescence imaging of WT and Trem2^−/− mice bearing CT-2A-luc. (D) Quantification of bioluminescence in WT and Trem2^−/− mice throughout tumor progression. p=0.0135 (mixed-effects analysis). (E) Pathway signature scoring between WT and TREM2 KO mice. Trend plots using NanoString ncounter data were used to define pathway scores based on the expression of key genes across WT and Trem2^−/− mice. Lines in the trend plot represent each pathway’s average score across WT and Trem2^−/− groups. Trem2^−/− mice display a drastic decrease in the pathway scores associated with adaptive immune response, apoptosis, autophagy, astrocyte function, neurons and neurotransmission, MG function, cytokine signaling, innate immune response, growth factor signaling and inflammatory signaling. (F) Boxplots showing the distribution of normalized log2 transformed counts from nanoString ncounter data of TREM2, C3, Tgm2, Mertk, Dock2 and Syk genes between WT and Trem2^−/− mice. Median expression values of the selected genes in WT mice are higher as compared to Trem2^−/− mice, and the difference is statistically significant (p-value < 0.05).

Figure 4.. Trem2 is associated with increased ZsGreen uptake in an orthotopic murine glioma model.

(A) Schematic outlining implantation of CT-2A-ZsGreen, phagocytosis and uptake of tumor antigen by myeloid cells, and flow cytometry analysis. Schematic illustration made in BioRender. (B) Uniform manifold approximation and projection (UMAP) showing MG (green), MAC (red), and DC (blue) in CT-2A-ZsGreen-bearing mouse brains. MG were defined as P2RY12⁺ CD11b⁺, MACs were defined as P2RY12⁻ CD11b⁺ F4/80⁺ MHC II⁺, and dendritic cells were defined as P2RY12⁻ CD11b⁺ CD11c⁺ MHC II⁺. (C) Histogram showing intensity of Trem2 expression on these distinct myeloid populations. (D) FSC vs. ZsGreen as gated on the Trem2⁻ population shows low proportions of Trem2⁻ZsGreen⁺ cells. (E) FSC vs. ZsGreen gated on Trem2⁺ populations show a large frequency of Trem2⁺ZsGreen⁺ cells. (F) Quantification of ZsGreen⁺ cells in Trem2⁻ vs. Trem2⁺ DCs. All mice showed higher proportions of ZsGreen in the Trem2⁺ DCs. A paired samples t test indicates that Trem2⁺/ZsGreen⁺ DCs are significantly higher in number than Trem2⁻ZsGreen⁺ DCs (p=0.0448). (G) Quantification of ZsGreen⁺ cells in Trem2⁻ vs. Trem2⁺ MG. A paired samples t test indicates no significant difference in Trem2⁺ZsGreen⁺ and Trem2⁻ZsGreen⁺ MG populations (p=0.2349).