Tumor progression is independent of tumor-associated macrophages in cell lineage-based mouse models of glioblastoma

Abstract

Macrophage targeting therapies have had limited clinical success in glioblastoma (GBM). Further understanding the GBM immune microenvironment is critical for refining immunotherapeutic approaches. Here, we use genetically engineered mouse models and orthotopic transplantation-based GBM models with identical driver mutations and unique cells of origin to examine the role of tumor cell lineage in shaping the immune microenvironment and response to tumor-associated macrophage (TAM) depletion therapy. We show that oligodendrocyte progenitor cell lineage-associated GBMs (Type 2) recruit more immune infiltrates and specifically monocyte-derived macrophages than subventricular zone neural stem cell-associated GBMs (Type 1). We then devise a TAM depletion system that offers a uniquely robust and sustained TAM depletion. We find that extensive TAM depletion in these cell lineage-based GBM models affords no survival benefit. Despite the lack of survival benefit of TAM depletion, we show that Type 1 and Type 2 GBMs have unique molecular responses to TAM depletion. In sum, we demonstrate that GBM cell lineage influences TAM ontogeny and abundance and molecular response to TAM depletion.

Keywords: CSF1R inhibition; cell of origin; glioblastoma; microglia; tumor-associated macrophages.

Fig. 1.

Cell of origin influences TAM ontogeny in spontaneous GEMMs of GBM. (A) Schematic illustrating the spontaneous GEMMs used to examine the impact of GBM cell of origin on the tumor immune microenvironment. (B and C) IF images and quantification of the total TAMs in Type 1 and Type 2 spontaneous GBMs (N = 6 Type 1, 4 Type 2 tumors, data represented as mean with SEM, P = 0.0379, unpaired t test, each dot represents one FOV) and the average number of processes per TAM cell per tumor (N = 3 Type 1, 5 Type 2, TAMs quantified from five 20× images per tumor, data represented as mean with SEM, P = 0.0007, unpaired t test). (D) Multicolor flow cytometry measuring the levels of all immune infiltrates (P = 0.0006), MDMs (P = 0.0017 CD45^hi, CD11B⁺, P = 0.0082 CD45⁺, CD11B⁺, CD49D⁺), microglia (P = 0.0086), neutrophils, and monocytes (P = 0.0489), data represented as mean with SEM, unpaired t test, each dot represents one mouse. (E) mRNA expression levels of pan-TAM (AIF1), MDM-associated (ITGA4, ITGAL), pan-immune (PTPRC), and microglia-associated (SALL1) genes in Type 1 and Type 2 subtyped human TCGA GBM samples (N = 52 core Type I, 38 core Type II).

Fig. 2.

Combination of a genetic and pharmacologic system robustly depletes TAMs. (A) Representative IF images of microglia-depleted vs. control nontumor bearing brains. (B) Box and whisker plot comparing microglia depletion strategies. N = 2 mice per group, three images quantified per mouse per brain region, middle line displays median value, whiskers display min and max values. Comparing the median values for the cortex region gives the following TAM depletion percentages by treatment; 66% (CSF1R only 1 wk), 81% (CSF1R only 2 wk), 87% (DT+CSF1R 1 wk), 86% (DT+CSF1R 2 wk). All P values with four stars indicate P < 0.0001. Other P values for comparisons are cortex 1 wk DT+PLX5622 vs. PLX5622 only (P = 0.041), cortex 2 wk DT+PLX5622 vs. PLX5622 only (P = 0.016), thalamus 1 wk DT+PLX5622 vs. PLX5622 only (P = 0.023), thalamus 2 wk DT+PLX5622 vs. PLX5622 only (P = 0.019), cerebellum 1 wk DT+PLX5622 vs. PLX5622 only (P = 0.05), cerebellum 2 wk DT+PLX5622 vs. PLX5622 only (P = 0.0015), cortex 1 wk control vs. PLX5622 only (P = 0.0037), medulla 1 wk control vs. PLX5622 only (P = 0.0014), and cerebellum 1 wk control vs. DT+PLX5622 (P = 0.0002). (C) Multicolor flow cytometry levels of total TAMs (Type 1 P = 0.0056, Type 2 P = 0.02), monocytes, and neutrophils in control vs. TAM-depleted Type 1 and Type 2 GBMs, data represented as mean with SEM, unpaired t test, each dot represents one mouse. TAM depletion percentage was estimated by comparing the percentages of total TAMs between TAM-depleted and control samples (84% for Type 1, 97% for Type 2).

Fig. 3.

TAM depletion does not extend survival in Type 1 and Type 2 GBM. Kaplan–Meier survival curves of TAM-depleted vs. control (treatments initiated 3 to 4 wk after tumor implantation) Type 1 (DT+PLX5622 N = 11, Control N = 14, median survival of 7.2 wk for control, 6.3 wk for DT+PLX5622) (A) and Type 2 (DT+PLX5622 N = 10, Control N = 14, median survival of 9.7 wk for control, 9.7 wk for DT+PLX622) (B) GBM mice, significance calculated with Mantel–Cox test. (C) Phospho-histone H3 IF staining images and quantification of TAM-depleted and control GBMs, data represented as mean with SEM, N= 5 control mice, 5 TAM-depleted mice, unpaired t test, symbols represent %PH3+ events per dapi+ event for each FOV (5 FOV per mouse). (D) Kaplan–Meier survival curve of TAM-depleted vs. control (treatments initiated 1 wk after tumor implantation) Type 2 (DT+PLX5622 N = 8, control N = 8, median survival of 8.5 wk for control, 9.3 wk for DT+PLX622) GBM mice, significance calculated with the Mantel–Cox test. (E) Kaplan–Meier survival curve of TAM-depleted vs. control (initiated 3 to 4 wk after tumor implantation) Nf1-sufficient (DT+PLX5622 N = 6, Control N = 6, median survival of 7.8 wk for control, 7.9 wk for DT+PLX5622) GBM mice, significance calculated with the Mantel–Cox test.

Fig. 4.

Type 1 and Type 2 GBMs have distinct molecular responses to TAM depletion. (A) PCA plot showing Type 1 TAM-depleted vs. control samples. (B) PCA plot showing Type 2 TAM-depleted vs. control samples. (C) Volcano plot showing the DEGs when comparing Type 1 TAM-depleted vs. all control tumors (padj < 0.05, LFC > 1.5 or <−1.5). (D) Volcano plot showing the DEGs when comparing Type 2 TAM-depleted vs. all control tumors (padj < 0.05, LFC > 1.5 or <−1.5). (E) Hallmark pathway GSEA of Type 1 TAM-depleted vs. control tumors. Select enrichment plots for mitotic spindle and inflammatory response shown. (F) Hallmark pathway GSEA of Type 2 TAM-depleted vs. control tumors. Select enrichment plots for oxidative phosphorylation and interferon gamma response shown.