Ly6G+ inflammatory cells enable the conversion of cancer cells to cancer stem cells in an irradiated glioblastoma model

Abstract

Most glioblastomas frequently recur at sites of radiotherapy, but it is unclear if changes in the tumor microenvironment due to radiotherapy influence glioblastoma recurrence. Here, we demonstrate that radiation-induced senescent glioblastoma cells exhibit a senescence-associated secretory phenotype that functions through NFκB signaling to influence changes in the tumor microenvironment, such as recruitment of Ly6G⁺ inflammatory cells and vessel formation. In particular, Ly6G⁺ cells promote conversion of glioblastoma cells to glioblastoma stem cells (GSCs) through the NOS2-NO-ID4 regulatory axis. Specific inhibition of NFκB signaling in irradiated glioma cells using the IκBα super repressor prevents changes in the tumor microenvironment and dedifferentiation of glioblastoma cells. Treatment with Ly6G-neutralizing antibodies also reduces the number of GSCs and prolongs survival in tumor-bearing mice after radiotherapy. Clinically, a positive correlation exists between Ly6G⁺ cells and the NOS2-NO-ID4 regulatory axis in patients diagnosed with recurrent glioblastoma. Together, our results illustrate important roles for Ly6G⁺ inflammatory cells recruited by radiation-induced SASP in cancer cell dedifferentiation and tumor recurrence.

Fig. 1

Glioblastoma grown after irradiation exhibits increased stem cell markers and microenvironmental changes. a Experimental scheme for the mouse model of GBM tumors treated with or without irradiation. For the irradiation of GBM mouse model, whole mouse brains were treated 5 times with 2 Gy ionizing radiation (2 Gy × 5) daily from day 21 to 25 following orthotopic injection (n = 4). b Hematoxylin–eosin (HE) stained images showing U87MG and LN229 grown tumors after fractionated irradiation. Average survival time is indicated. Scale bar represents 500μm. c Representative images showing the primary and grown tumors stained with several stem cell markers (OCT4, Nestin, and CD133). Scale bar represents 50 μm. d Representative images indicating CD31⁺ vessels, IBA1⁺ microglia, and Ly6G⁺ cells in the primary and grown tumors. Scale bar represents 50μm. e Quantifications of OCT4-, Nestin-, and CD133-positive cells shown in Fig. 1c and CD31-positive vessel numbers, and IBA1- and Ly6G-positive cells (*p < 0.05, **p < 0.01). f Box plots showing enrichment score (ES) for a gene set of neutrophil and macrophage markers of the patients with primary and recurrent GBM (n = 30). g Correlation plots of POU5F1 mRNA levels and neutrophil markers (MPO, CD66B) or macrophage markers (AIF, CD68) mRNA level (n = 30). A Pearson product–moment correlation coefficient (r) was used to determine the linear correlation between two variables. Data in this figure are expressed as means ± standard error of the mean (SEM)

Fig. 2

Irradiated glioblastoma cells trigger glioblastoma cell dedifferentiation and Ly6G⁺ inflammatory cell recruitment. a Experimental scheme for orthotopic co-injection models (n = 4). Mice were injected with non-irradiated glioblastoma cells (GFP-N; 1 × 10⁵) alone or with a combination of GFP-N (1 × 10⁵) and irradiated glioblastoma cells (I-Red; 2 × 10⁵). b HE stained images of orthotopic co-injection models. Scale bar represents 500 μm. c Representative immunofluorescence images of the indicated tumors stained with several stem cell markers. OCT4-, Nestin-, and CD133-positive cells were quantified (*p < 0.05, **p < 0.01). Scale bar represents 50μm. d Representative immunohistochemistry images of the indicated tumors stained with antibodies against CD31, Ly6G, F4/80, and IBA1. Intensity of CD31-, Ly6G-, F4/80- and IBA1-positive regions was quantified (*p < 0.05, **p < 0.01). Scale bar represents 50μm. Data in this figure are expressed as means ± SEM

Fig. 3

Tumor-derived GFP-P cells exhibit stem cell-like properties and ID4 expression. a, d In vitro limiting dilution assay showing stem cell sphere-forming frequency of the T-OCT4-p-GFP⁺ (T-GFP-P) and T-OCT4-p-GFP^- (T-GFP-N) cells derived from a subcutaneous co-injection model (Supplementary Fig. 2a). b, e FACS analysis comparing the expression of several stem cell markers (Nestin, CD133, OCT4, SOX2, and NANOG) in T-GFP-P and T-GFP-N cells (*p < 0.05, **p < 0.01). c, f Colony-formation assay demonstrating the radioresistance of T-GFP-P and T-GFP-N cells on day 14 after 0, 2, or 3 Gy irradiation (*p < 0.05, **p < 0.01). g Tumor-initiating ability of T-GFP-P and T-GFP-N cells. h Western blot analysis showing ID family expression in T-GFP-P and T-GFP-N cells. i Representative images indicating ID4-positive cells (red) within the tumors. ID4-positive cells were quantified (*p < 0.05, **p < 0.01). Scale bar represents 50 μm. j Western blot analysis showing ID4 expression (upper) and sphere-forming assay (bottom) of U87MG-GFP-N-puro and -ID4 cells. k MTS assay for detecting radioresistance of U87MG-GFP-N-puro and -ID4 cells. Cell survival was detected 3 days post irradiation. l Western blot analysis showing ID4 expression (upper) and sphere-forming assay results (bottom) of LN229-GFP-N-puro and -ID4 cells. m MTS assay for detecting radioresistance of LN229-GFP-N-puro and -ID4 cells. Cell survival was detected 3 days post irradiation. Data in this figure are expressed as means ± SEM

Fig. 4

Irradiated glioblastoma cells promote tumor microenvironment changes via NFκB signaling in vitro and in vivo. a Representative images showing SA-β-gal- and c-CASP3-positive cells in U87MG and LN229 xenograft models after fractionated irradiation (2 Gy × 5). Scale bar represents 50 μm. b The SA-β-gal staining assay of U87MG and LN229 cells overexpressing IκBα* on day 3 after irradiation with 20Gy. Representative images (left) and quantification of the SA-β-gal-positive glioblastoma cells (right, n.s.: not significant). Scale bar represents 50μm. c qRT-PCR assay showing mRNA levels of SASP genes in U87MG and LN229 cells overexpressing IκBα* on day 3 after irradiation with 20 Gy (*p < 0.05, **p < 0.01 between puro and I-puro; ^#p < 0.05, ^##p < 0.01 between I-puro and I-IkBα*). d In vitro tubule formation assay of HRECs incubated with CM obtained from I-puro cells or I-IκBα* cells. Quantification of the tube numbers (*p < 0.05, **p < 0.01 between control-puro and IR-puro; ^#p < 0.01 between IR-puro and IR-IkBα*). e Transwell invasion assay of BV2 microglial cells in the upper chamber with CM from I-puro cells or I-IκBα* cells in the bottom chamber. Quantification of the total number of invaded cells (*p < 0.05, **p < 0.01 between control-puro and IR-puro; ^##p < 0.01 between IR-puro and IR-IkBα*). f Transwell migration assay of dHL-60 cells in the upper chamber with CM from I-puro cells or I- IκBα* cells in the bottom chamber. (*p < 0.05, **p < 0.01 between control-puro and IR-puro; ^##p < 0.01 between IR-puro and IR-IkBα*). g Experimental scheme for the orthotopic co-injection models (n = 4). Mice were injected with GFP-N alone (1 × 10⁵), GFP-N (1 × 10⁵) and I-puro (2 × 10⁵), or I-IκBα* alone (2 × 10⁵). h Representative immunofluorescence images of the indicated tumors stained with Ly6G, IBA1, CD31, and GFP. Ly6G-, IBA1-, CD31-, and GFP-positive cells were quantified (*p < 0.05, **p < 0.01 between GFP-N and GFP-N + I-puro; ^#p < 0.05, ^##p < 0.01 between GFP-N + I-puro and GFP-N + I-IkBα*). Scale bar represents 50μm. Data in this figure are expressed as means ± SEM

Fig. 5

Infiltrated Ly6G⁺ inflammatory cells promote dedifferentiation of glioblastoma cells to GSCs via the NO-ID4 axis. a qRT-PCR assay showing mRNA levels of pro-inflammatory cytokines and chemokines (Ccl2, Ccl3, Tnf-a, Vegfa, Il-1a, Il-1b, Il-6, Cxcl1, Nos1, Nos2, and Nos3) in Ly6G⁺ cells isolated from normal spleen and tumor tissues (*p < 0.05, **p < 0.01; n.d. not detectable). b Quantification of nitrite levels secreted by I-Red cells derived from U87MG and LN229 cells. c Representative immunofluorescence images showing Ly6G (green)/NOS2 (red) double-positive cells. Ly6G⁺NOS2^- and Ly6G⁺NOS2⁺ cells were quantified (**p < 0.01). d Representative images showing GFP-, ID4-, Nestin-, and CD133-positive GSCs (red) located in close proximity to the Ly6G⁺ cells (green). The stem cell marker-positive cells located in close proximity to the Ly6G-positive cells (≤100μm diameter regions) were quantified (*p < 0.05, **p < 0.01). Data in this figure are expressed as means ± SEM

Fig. 6

Ly6G⁺ inflammatory cell inhibition delays tumor growth after irradiation. a Experimental scheme for the recurrent GBM model treated with either control IgG or anti-Ly6G. The mice were treated 5 times with 2 Gy ionizing radiation (2 Gy × 5) from day 21 to 25 after orthotopic injection (n = 5). In one group of mice (pre-anti-Ly6G treatment), IgG or anti-Ly6G antibody was injected at a dose of 300 μg on day 19 prior to irradiation and then at 100 μg after irradiation. In a second group (post-anti-Ly6G treatment), 100 μg of antibodies were injected every 72 h from day 26 until the final treatment. b Survival rate of U87MG xenograft mice injected with either IgG or anti-Ly6G before and after exposure to irradiation (0 Gy or 2 Gy × 5; *p < 0.05, **p < 0.01; n.s. not significant). c Representative immunofluorescence images and quantification of Ly6G⁺ and IBA1⁺ cells. (*p < 0.05, **p < 0.01). d Representative images showing ID4⁺, Nestin⁺, and CD133⁺ GSCs (red) located in close proximity to the Ly6G⁺ cells (green). The stem cell marker-positive cells located in close proximity to the Ly6G⁺ cells (≤100 μm diameter regions) were quantified (*p < 0.05, **p < 0.01). Data in this figure are expressed as means ± SEM

Fig. 7

Neutrophils correlate with TAN and dedifferentiation gene sets in recurrent GBM patient samples. a, b Heatmap of TAN (a) and cytokine/chemokine (b) genes observed in recurrent GBM patient samples with MPO high vs. primary GBM patient samples with MPO low. c, d GSEA data showing the enrichment of OCT4, SOX2, and NANOG gene sets (c) and ID4, NOS, NFκB, and STAT3 gene sets (d) in recurrent GBM patient samples with MPO high. e A schematic diagram indicating the mechanism of GBM recurrence after radiotherapy. Ly6G⁺ inflammatory cells recruited by radiation-induced SASP promote GBM recurrence by inducing the dedifferentiation of glioma cells to GSCs via the NOS-ID4 signaling axis