Lack of ST2 aggravates glioma invasiveness, vascular abnormality, and immune suppression

Abstract

Background: Glioblastoma (GBM) is the most common primary malignant brain tumor in adults, characterized by aggressive growth and a dismal prognosis. Interleukin-33 (IL-33) and its receptor ST2 have emerged as regulators of glioma growth, but their exact function in tumorigenesis has not been deciphered. Indeed, previous studies on IL-33 in cancer have yielded somewhat opposing results as to whether it is pro- or anti-tumorigenic.

Methods: IL-33 expression was assessed in a GBM tissue microarray and public databases. As in vivo models we used orthotopic xenografts of patient-derived GBM cells, and syngenic models with grafted mouse glioma cells.

Results: We analyzed the role of IL-33 and its receptor ST2 in nonmalignant cells of the glioma microenvironment and found that IL-33 levels are increased in cells surrounding the tumor. Protein complexes of IL-33 and ST2 are mainly found outside of the tumor core. The IL-33-producing cells consist primarily of oligodendrocytes. To determine the function of IL-33 in the tumor microenvironment, we used mice lacking the ST2 receptor. When glioma cells were grafted to ST2-deficient mouse brains, the resulting tumors exhibited a more invasive growth pattern, and are associated with poorer survival, compared to wild-type mice. Tumors in ST2-deficient hosts are more invasive, with increased expression of extracellular matrix remodeling enzymes and enhanced tumor angiogenesis. Furthermore, the absence of ST2 leads to a more immunosuppressive environment.

Conclusions: Our findings reveal that glia-derived IL-33 and its receptor ST2 participate in modulating tumor invasiveness, tumor vasculature, and immunosuppression in glioma.

Figure 1.

Expression of IL-33 in GBM patients and in xenograft models of glioma. (A) Analysis of publicly available IL-33 mRNA expression data from 3 GBM cohorts compared to control brains. Expression levels were normalized and presented as fold change. (B) Heat map showing quantification of IL-33 immunolabeling in 22 tumor cores from GBM patients in a tissue microarray (TMA). Automated quantification is presented as staining intensity (weak, intermediate, or strong). (C,D) Violin plots showing quantification of the immunolabelling intensity of IL-33 staining control brain and GBM. Two-way ANOVA of control brain cores versus all GBM cores, P = .075. (E, G, H) Expression of murine IL-33 in xenograft mouse models derived by injecting human GBM cells U3013, and (F, I, J) U3024 cells in immunodeficient mouse striatum. The distribution of murine IL-33-expressing cells was examined in cells directly surrounding the tumor mass (E, F, demarcated by dotted line), gray matter of striatum (G, I), and white matter of corpus callosum (H, J). Human cells, stained with STEM121 antibodies. Blue DAPI nucelar stain. Representative data from three mice for each GBM cell line injected. Data are presented as mean ± SD, ****P < .0001. Scale bar 50 µm.

Figure 2:

IL-33 expression in the tumor microenvironment of mouse glioma models. (A) Immunohistochemical analysis showing IL-33 expression in the mouse brain with CT-2A tumor cells. IL-33 expression in the contralateral hemisphere (A, insert). (B-C) Co-expression of IL-33 and CNPase in the white matter of the corpus callosum (WM, B), and in the gray matter of the cortex (GM, C). (D) Co-localization of IL-33 with Olig2 or (E) GFAP. (F) Quantification of co-localization of IL-33 with Olig2, GFAP or CNPase in the tumor brain. (G-J) Proximity ligation assay detects protein complexes formed by IL-33 and ST2 in CT-2A glioma. (G) control brain, (H) tumor core, (I) area surrounding the tumor, and (J) contralateral hemisphere. Results are representative of three biological replicates with 6 mice per group in each repeat. Data are presented as mean ± SD, ****P < .0001. Scale bar 50 µm.

Figure 3.

Increased glioma aggressiveness in ST2-deficient mice. (A) Kaplan–Meier analysis showing shorter survival times in ST2-deficient mice (n = 30) with CT-2A glioma, compared to wt mice (n = 30). (B-C) Image processing to measure the length of the tumor border in ST2-deficient and wt mice. (D) Quantification of the tumor border length, with longer borders observed in ST2-deficient mice compared to wt. (E) Increased matrix metalloproteinases (MMP2, MMP3, and MMP9) in the serum of ST2-deficient mice, further enhanced in ST2-deficient mice with tumors. Scale bars: 1 mm (B) and 50 µm (C). (B-D) Results are representative of three biological replicates with 10 mice per group in each repeat. Data are presented as mean ± SD. Scale bar 50 µm.

Figure 4.

Aggravated tumor vasculature in the absence of ST2 in mouse glioma. (A-B) Immunofluorescence staining showing increased CD31 expression and vascular density in gliomas from ST2-deficient mice. In the absence of st2, CD31 expression increases, as evidenced by (C) area marked by CD31 antibodies and (D) intensity of the staining close to the tumor border. There is a higher number of vessels (E), extended vessel length (F), and larger vessel volume (G). (H) Vessels in st2^-/- mice are characterized by a greater cross-sectional area and (I) an increased diameter. (J-L) IgG labeling reveals that ST2-deficient glioma-bearing mice exhibit more leakage than wt mice in close proximity of the tumor border (M) Elevated levels of proteins associated with angiogenesis in serum of ST2-deficient mice with brain tumors, compared to control mice, and further enhanced in st2^-/- mice with glioma. Results are representative of three biological replicates with 10 mice per group in each repeat. Scale Bars: 50 µm (A, B) and 500 µm (J, K).

Figure 5.

Absence of ST2 results in suppressed immune response in CT-2A gliomas. Flow cytometry data of immune cell profiles in CT-2A brain tumors from wt and st2^-/- mice. (A) Gating strategy used to identify innate myeloid populations in the brain of CT2A-bearing wt and st2^-/- mice by flow cytometry, namely CD11b⁺CD11c^- macrophages (MΦ), CD11b^-CD11c⁺ cDC type 1 (cDC1) and CD11b⁺CD11c⁺ cDC type 2 (cDC2). The overall macrophage, cDC1, and cDC2 populations were identified by gating on CD45^high cells, followed by gating based on CD11b and CD11c expression. Alternatively, non-tissue resident cDC1s were identified as CD45^highCX3CR1^-CD11b^-CD11c⁺. Macrophages (CD11b + CD11c-) from st2^-/- mice have reduced CD86 expression (B,C) and increased Arg1 levels (D, E), indicating suppressed activity. The gray box in (D) highlights a population of CD86^hi cells that were underrepresented in st2-/- mice. (D,E) The dotted line in (E) indicates increased expression of Arg1 in st2-/- mice. Type 1 conventional dendritic cells (CD11b-CD11c + cDC1s) in st2^-/- mice show lower CD86 (F, G), and nonresident CX3CR1- cDC1s display reduced MHC-II expression (H, I), crucial for T cell activation. The gray box in (G) highlights a population of CD86^hi cells that were underrepresented in st2-/- mice while the gray box in (I) highlights a population of MHC-II^- cells that were overrepresented in st2-/- mice. Similarly, type 2 cDCs (CD11b + CD11c + cDC2s) display decreased CD86 (J, K) with heightened Arg1 (L, M) and IL-10 production (N, O), suggesting immune inhibition. The gray box in (K) and (O) highlights a population of CD86^hi and IL-10^hi cells that were underrepresented in st2-/- mice. The dotted line in (M) indicates increased expression of Arg1 in st2-/- mice. (P) Gating strategy used to produce data shown in panels Q-S. The two major T cell subpopulations were separated based on their CD8 and CD4 expression. CD8⁺PD1⁺, as well as CD4⁺CD69⁺ and CD4⁺CD107a⁺ T cells, were identified using FMO-based gates. An increase in exhausted CD8 + PD1 + T cells (Q) and a decrease in activated (CD69+) (R) and cytotoxic (CD107a+) (S) CD4 + T cells in st2^-/- mice indicate T cell suppression. Results are representative of three biological replicates with 10 mice per group in each repeat.