Soluble ICAM-1 a Pivotal Communicator between Tumors and Macrophages, Promotes Mesenchymal Shift of Glioblastoma

Abstract

Despite aggressive clinical treatment, recurrence of glioblastoma multiforme (GBM) is unavoidable, and the clinical outcome is still poor. A convincing explanation is the phenotypic transition of GBM cells upon aggressive treatment such as radiotherapy. However, the microenvironmental factors contributing to GBM recurrence after treatment remain unexplored. Here, it is shown that radiation-treated GBM cells produce soluble intercellular adhesion molecule-1 (sICAM-1) which stimulates the infiltration of macrophages, consequently enriching the tumor microenvironment with inflammatory macrophages. Acting as a paracrine factor, tumor-derived sICAM-1 induces macrophages to secrete wingless-type MMTV integration site family, member 3A (WNT3A), which promotes a mesenchymal shift of GBM cells. In addition, blockade of either sICAM-1 or WNT3A diminishes the harmful effect of radiation on tumor progression. Collectively, the findings indicate that cellular crosstalk between GBM and macrophage through sICAM-1-WNT3A oncogenic route is involved in the mesenchymal shift of GBM cells after radiation, and suggest that radiotherapy combined with sICAM-1 targeted inhibition would improve the clinical outcome of GBM patients.

Keywords: glioblastoma (GBM); macrophages; mesenchymal shift; soluble intercellular adhesion molecule-1 (sICAM-1); tumor microenvironment.

Figure 1

Radiation‐induced soluble ICAM‐1 promotes mesenchymal shift of GBM. A) Gene set enrichment analysis of GBM patients according to recurrence after radiation therapy (GSE7696). B,C) GSEA plots of normalized enrichment score (NES) of radiation exposure‐responsive gene set in human U87MG compared to non‐irradiated cells (GSE56937). D) Table summarizing the quantification of cytokines in the conditioned media of U87MG cells exposed to radiation with either a fractionated dose, 2Gy/d × 3 (2Gy per day for 3 days), or single dose (6, 10, or 20Gy). E) Overall survival curves for ICAM‐1, IL‐6, IL‐18, and TNF‐α in patients with GBM based on low and high expression (GSE4271, n = 77). F) Cytokine expression levels in normal human brain and GBM (GSE66354). G) Schematic of the animal experimental design. U87MG cells were injected intracranially into BALB/c nude mice (n = 6 per group). H) IHC analysis of ICAM‐1 and I) ELISA of sICAM‐1 level in U87MG orthotopic xenograft tumor in BALB/c nude mice after cranial irradiation. Scale bar, 200 µm. J,K) FACS analysis was performed using U87MG orthotopic xenograft tumors to assess ICAM‐1 expression in tumors, endothelial cells, and macrophages. L) IHC analysis and graphical depiction of the ratio of post‐CCRT to pre‐CCRT IHC intensity in specimens from patients with GBM. Scale bar, 100 µm. M) Schematic of the animal experimental design. U87MG^sh‐Ctrl or U87MG^{sh‐ICAM‐1} cells were injected orthotopically into BALB/c nude mice (n = 6 per group). N) H&E and IHC analysis of FN, VIM, and YKL‐40 in U87MG^sh‐Ctrl or U87MG^{sh‐ICAM‐1} orthotopic xenograft tumors after irradiation. Scale bar, 200 µm. O) Schematic diagram of expression constructs encoding ICAM‐1 wild type and soluble ICAM‐1. P) H&E and IHC analysis of FN, VIM, and YKL‐40 in U87MG^Mock and U87MG^sICAM‐1 orthotopic tumors (n = 6 per group). Scale bar, 200 µm. Q) RT‐qPCR of FN, VIM, and YKL‐40 in U87MG^Mock and U87MG^sICAM‐1 orthotopic tumors. Data are presented as mean ± SD. β‐actin was used as control for normalization of expression. n.s, non‐significant; *, p < 0.05 versus control; **, p < 0.01 versus control; ***, p < 0.001 versus control. A two‐tailed Student's t‐test was used to compare data between two groups.

Figure 2

Radiation‐induced sICAM‐1 recruits bone marrow‐derived macrophages to the tumor, resulting in mesenchymal shift of GBM. A) GSEA for 17 types of immune cells in ICAM‐1^High and ICAM‐1^Low GBM patients from GEO dataset. B) GSEA plots for a significant correlation between high expression of ICAM‐1 and gene sets of macrophage migration and chemotaxis. C) IHC analysis for IBA‐1 and ICAM‐1 in GL261 orthotopic syngeneic tumors irradiated (2.5 Gy per day for 3 days) or not irradiated (n = 7 per group). Scale bar, 100 µm. D) Quantification of mobility of THP‐1 macrophages and PBMCs in CM of U87MG cells irradiated or non‐irradiated, and following treatment with ICAM‐1 antibody. E) Quantification of mobility of THP‐1 macrophages and PBMCs in CM of U87MG cells transfected with sICAM‐1 or mock construct. F) Effect of rh‐sICAM‐1 on mobility of THP‐1 macrophages and PBMCs. G) Schematic of the animal experiment design for preferential recruitment of THP‐1 macrophages to U87MG^sh‐Ctrl or U87MG^{sh‐ICAM‐1} xenograft tumors in BALB/c nude mice after irradiation (n = 3 per group). H) IHC analysis of mCherry‐labeled THP‐1 macrophages in U87MG^sh‐Ctrl or U87MG^{sh‐ICAM‐1} xenograft tumors after irradiation. Scale bar, 200 µm. I) Schematic of the animal experiment design for preferential recruitment of THP‐1 macrophages to U87MG^sh‐Ctrl or U87MG^{sh‐ICAM‐1} orthotopic xenograft tumors in BALB/c nude mice after irradiation (n = 3 per group). J) IHC analysis of mCherry‐labeled THP‐1 macrophages in orthotopic xenograft mice. Scale bar, 200 µm. Data are presented as mean ± SD. β‐actin was used as control for normalization of expression. n.s, non‐significant; *, p < 0.05 versus control; **, p < 0.01 versus control; ***, p < 0.001 versus control. A two‐tailed Student's t‐test was used to compare data between two groups.

Figure 3

Radiation‐induced sICAM‐1 promotes WNT3A secretion by bone marrow‐derived macrophages after recruitment to GBM site. A) Top 50 list of cytokine profiling antibody array for THP‐1 macrophages after treatment with control medium or CM of U87MG cells that are irradiated or non‐irradiated (detailed in Table S1, Supporting Information). B) Analysis of cytokines induced by CM of irradiated U87MG cells. The network was built based on the KEGG pathway. C) RT‐qPCR analysis of soluble factor expression induced in THP‐1 macrophages after treatment with rh‐sICAM‐1. D) RT‐qPCR analysis of WNT3A expression in THP‐1 macrophages cocultured with irradiated or non‐irradiated GBM cells. E) RT‐qPCR analysis of WNT3A expression in THP‐1 macrophages cocultured with U87MG^Mock or U87MG^sICAM‐1 cells. RT‐qPCR analysis of WNT3A expression in THP‐1 macrophages F) when cocultured with U87MG GBM cells transfected with ICAM‐1 siRNA or G) treated with MMP‐9 inhibitors prior to irradiation. H) Western blot analysis of β‐catenin in U87MG or U251MG cells alone or cocultured with THP‐1 macrophages. I) Effect of rh‐WNT3A on migration of GBM cells in Transwell plates. J) IHC analysis of WNT3A in GL261^Mock or GL261^sICAM‐1 orthotopic syngeneic tumor (n = 7 per group). Scale bar, 200 µm. K) RT‐qPCR analysis of WNT3A expression in GL261^Mock or GL261^sICAM‐1 orthotopic syngeneic tumor. L) The amount of WNT3A in blood extracted from Heart in GL261‐bearing syngeneic mice. M) IHC analysis of CCR2 and WNT3A in GL261^Mock or GL261^sICAM‐1 orthotopic syngeneic tumor. Scale bar, 100 µm. N) IHC analysis of WNT3A levels in a pair of specimens of GBM patient before CCRT (pre‐CCRT) and post‐CCRT. Scale bar, 200 µm. O) GSEA plots (GSE4412, 42669, and 22866) for a significant correlation between high expression of ICAM‐1 and WNT signaling. Data are presented as mean ± SD. β‐actin was used as control for normalization of expression. n.s, non‐significant; *, p < 0.05 versus control; **, p < 0.01 versus control; ***, p < 0.001 versus control. A two‐tailed Student's t‐test was used to compare data between two groups.

Figure 4

Blocking sICAM‐1 or WNT3A inhibits the infiltration and mesenchymal shift of GBM by suppressing macrophages recruitment. A) Invasion assay of U87MG cells cocultured with THP‐1 in combination treatments with radiation and/or a neutralizing antibody against ICAM‐1 or WNT3A as indicated. B) RT‐qPCR analysis of mesenchymal markers in U87MG cells cocultured with THP‐1 in combination treatments with radiation and/or a neutralizing antibody against ICAM‐1 or WNT3A. C) Invasion assay of U87MG cells cocultured with THP‐1 in combination treatments with sICAM‐1 transfection and/or a neutralizing antibody against ICAM‐1 or WNT3A as indicated. D) RT‐qPCR analysis of mesenchymal markers in U87MG cells cocultured with THP‐1 in combination treatments with sICAM‐1 transfection and/or a neutralizing antibody against ICAM‐1 or WNT3A as indicated. E) Schematic of the animal experiment design for evaluating therapeutic effect of neutralizing antibody against ICAM‐1 or WNT3A in U87MG xenograft BALB/c nude mice model (n = 5 per group). F) H&E and IHC analysis of FN, VIM, and YKL‐40 in U87MG orthotopic xenograft tumors after combination treatment with irradiation and ICAM‐1 or WNT3A neutralizing antibody. Scale bar, 200 µm. Data are presented as mean ± SD. β‐actin was used as control for normalization of expression. n.s, non‐significant; *, p < 0.05 versus control; **, p < 0.01 versus control; ***, p < 0.001 versus control. A two‐tailed Student's t‐test was used to compare data between two groups.

Figure 5

Increase in sICAM‐1 in patients with GBM after radiotherapy correlates with bone marrow‐derived macrophage recruitment, mesenchymal shift of tumors, and high levels of WNT3A. IHC analysis of A) ICAM‐1 and IBA‐1‐expressing cells or B) CCR2 expressing cells of GBM specimens from patients #13 and #16 pre‐CCRT and post‐CCRT. Scale bar, 200 µm. C) GSEA plots for correlation between gene sets of recurrent GBM after radiation and gene sets of macrophage chemotaxis or WNT signaling (GSE7696). D) ICAM‐1 expression levels in four different human GBM subtypes and normal human brain queried from the TCGA database. E) IHC analysis of ICAM‐1, IBA‐1, WNT3A, and VIM levels in GBM proneural and mesenchymal subtypes from patients. Scale bar, 200 µm. F) Quantification of IHC analysis in (E). G) Kaplan‐Meier survival curves of patients with GBM using Rembrandt database; ICAM‐1^High versus ICAM‐1^Low. H) Schematic summarizing a model of mesenchymal shift of GBM cells caused by WNT3A signaling in tumor microenvironment triggered by a crosstalk between bone marrow‐derived macrophages and GBM cells in response to radiation‐induced sICAM‐1. Data are presented as mean ± SD. β‐actin was used as control for normalization of expression. n.s, non‐significant; *, p < 0.05 versus control; **, p < 0.01 versus control; ***, p < 0.001 versus control. A two‐tailed Student's t‐test was used to compare data between two groups.