Neuropeptide adrenomedullin remodels stemness and macrophage dynamics in glioblastoma

Abstract

The presence of self-renewing glioblastoma stem cells (GSCs) and infiltrating pro-tumor macrophages constitutes two key hallmarks of glioblastoma (GBM). Here, we identified neuropeptide adrenomedullin (ADM) as a key factor regulating the GSC-macrophage symbiosis. EGFR overexpression upregulates ADM in GSCs to enhance their self-renewal, glycolysis, and tumor growth by activating the STAT3 pathway. GSC-secreted ADM promotes macrophage infiltration and pro-tumor reprograming through activation of ADM receptor (ADMR), thereby engaging both STAT3 and STAT6 pathways. In GBM mouse and PDX models, inhibition of the ADM-ADMR axis, STAT3, or STAT6 suppresses tumor progression, GSC self-renewal, and pro-tumor macrophage abundance, with dual inhibition of STAT3 and STAT6 leading to durable complete tumor regression in a subset of tumor-bearing mice. In human GBM tumors and plasmas, ADM correlates positively with GSC stemness, pro-tumor macrophage abundance, and poor prognosis. These findings highlight the ADM-triggered GSC-macrophage symbiosis as a promising therapeutic target for GBM.

INTRODUCTION

Glioblastoma (GBM) is the most aggressive primary brain tumor in adults, featured by invasive growth, extreme resistance to therapies, and rapid recurrence, resulting in a median survival of only 15–20 months despite standard-of-care^1–3. The poor prognosis, treatment resistance, and tumor recurrence of GBM are in part induced by the presence of glioblastoma stem cells (GSCs), a subset of cancer cells harboring stem cell-like capabilities⁴. Although GSCs only account for ~10% of total cancer cells, they make up the majority of proliferative cancer cells in GBM⁵. Given the critical role of GSCs in regulating tumor progression⁶, angiogenesis⁷, and immune responses^8–11, the development of GSC-targeted therapies holds significant promise^4,12,13. Despite great research efforts have been made, the molecular understanding of GSC stemness and heterogeneity, and their relationship with the tumor microenvironment (TME) are still limited, which impede the development of GSC-targeted therapies^4,14.

Neurons play a crucial role in GBM progression^15,16, and neural stem cells (NSCs) are considered a potential origin of GSCs¹⁷, supporting a hypothesis that secreted neuropeptides might regulate the formation of GSC-TME crosstalk. Tumor-associated macrophages and microglia (TAMs) constitute the most abundant cell types in the GBM TME accounting for up to 50% of total live cells of the whole tumor mass¹⁸. Increasing evidence underscores prominent symbiotic interactions between GSCs and TAMs, where GSCs facilitate the recruitment and activation of microglia and macrophages^11,14,19, and, reciprocally, these TAMs regulate GSC activities, such as stemness maintenance^4,20. Our recent studies have demonstrated that GSCs can express and secrete certain chemokines, such as olfactomedin-like 3 (OLFML3), legumain (LGMN), and tissue factor pathway inhibitor 2 (TFPI2), to trigger microglia infiltration and pro-tumor polarization, which, in turn, promote tumor progression and immunosuppression^8–10,21. These findings uncover the molecular mechanism underlying the interaction between GSCs and microglia; however, the regulatory mechanisms governing GSC-macrophage crosstalk in GBM remain largely unclear. Through unbiased analyses of upregulated neuropeptides in GSCs, followed by functional studies, we have identified adrenomedullin (ADM) as a key effector of GSC-macrophage crosstalk in GBM.

ADM is a peptide hormone and potent vasodilator belonging to the calcitonin gene-related peptide (CGRP) family, which was first discovered from human pheochromocytoma in 1993²². Although the original study demonstrated that ADM was not found in the brain²², increasing evidence suggests that ADM is widely expressed in the nervous system, functioning as a neuropeptide involved in neuroprotection and neurological diseases^23–26. ADM is highly expressed in a variety of tumors and functions as a potent paracrine and autocrine factor that regulates tumor cell survival, angiogenesis, and immunity in the TME^27–30. In GBM, ADM has been shown to promote tumor growth, angiogenesis, and abnormal vascularization^31–33. However, the potential roles of ADM in GSC stemness and its symbiotic relationship with macrophages in the GBM TME remain unexplored.

Here, we identified ADM as a GSC-macrophage symbiosis effector that is highly expressed in GSCs and critical for supporting GSC stemness maintenance via activation of the signal transducer and activator of transcription 3 (STAT3) pathway. Moreover, GSC-secreted ADM triggers macrophage infiltration and pro-tumor polarization via activation of STAT3 a d STAT6 pathways. Inhibition of the ADM–ADM receptor (ADMR) axis and its downstream signal STAT3 or STAT6 extends the survival in GBM mouse and patient-derived xenograft (PDX) models. Notably, dual inhibition of STAT3 and STAT6 further heightens the anti-tumor effect. In combination with results from our clinical-pathological investigations in human GBM tumors and plasmas, these findings point to ADM and its associated downstream signals as promising therapeutic targets for GBM.

RESULTS

ADM promotes GSC self-renewal.

GSC stemness has been shown to be correlated positively with immunosuppression^8–10,14. To identify the specific GSC factors that may mediate the crosstalk between GSCs and immune cells, we performed unbiased analyses using following approaches: 1) identify genes that are enriched in GSCs compared to matched GSC-differentiated non-stem tumor cells (NSTCs)³⁴; 2) focus on 63 genes encoding secreted neuropeptides³⁵; and 3) determine the genes whose expression is higher in GBM tumors than normal brain tissues using The Cancer Genome Atlas (TCGA) GBM dataset^36,37. ADM was identified as the only neuropeptide enriched in GSCs and GBM tumors comparing to NSTCs and normal brain tissues, respectively (Fig. 1A). Next, we analyzed the single-cell RNA sequencing (scRNA-seq) from GBM patient tumors^38,39 and found that ADM was higher in GSCs than GBM cells (Fig. 1B and Fig. S1A). RT-qPCR and Western blotting confirmed that the mRNA and protein levels of ADM were elevated in patient-derived GSC17 (Fig. 1C, D) and GSC272 (Fig. 1E, F) compared to their differentiated NSTCs.

Figure 1. ADM is essential for GSC self-renewal.

(A) Strategy for identification of neuropeptide ADM that is highly expressed in GSCs compared to matched non-stem tumor cells (NSTCs) (GSE20736) and in GBM tumors compared to normal brain in the TCGA GBM RNA-Seq dataset. (B) The expression (left) and quantification (right) of ADM in GBM cell and GSC subclusters of GBM patient tumors as profiled by analyzing scRNA-seq data (GSE182109). (C) RT-qPCR for ADM in GSC17 and their differentiated NSTCs. n = 6 independent samples. (D) Immunoblots for ADM in cell lysates of GSC17 and their differentiated NSTCs. (E) RT-qPCR for ADM in GSC272 and their differentiated NSTCs. n = 6 independent samples. (F) Immunoblots for ADM in cell lysates of GSC272 and their differentiated NSTCs. (G, H) Immunoblots for CD133, SOX2, and ADM in cell lysates of 005 GSC (G) and GSC272 (H) expressing shRNA control (shC) or ADM shRNA (shADM). (I, J) Representative images (left) and quantification (right) of tumorsphere formation in 005 GSC (I) and GSC272 (J) expressing shC or shADM. Scale bar, 200 μm. n = 3 independent samples. (K) Immunoblots for CD133, SOX2, and ADM in cell lysates of 005 GSC expressing control and Adm CRISPR KO. (L) Representative images (left) and quantification (right) of tumorsphere formation of control and Adm KO 005 GSCs. Scale bar, 200 μm. n = 5 independent samples. (M) Immunoblots for CD133, SOX2, and ADM in cell lysates of control and ADM KO GSC272. (N) Representative images (left) and quantification (right) of tumorsphere formation of control and ADM KO GSC272. Scale bar, 200 μm. n = 5 independent samples. (O, P) Immunoblots for CD133, SOX2, and ADM in cell lysates of 005 GSC (O) and GSC272 (P) treated with ADM (100 nM) in the presence or absence of ADM receptor antagonist, AMA (300 nM). (Q, R) Representative images (left) and quantification (right) of tumorsphere formation in 005 GSC (Q) and GSC272 (R) treated with ADM (100 nM) in the presence or absence AMA (300 nM). Scale bar, 200 μm. n = 3 independent samples. Data from multiple replicates are presented as mean ± SEM. **, P < 0.01, ***, P < 0.001, ****, P < 0.0001, Student’s t-test (B, C, E) or one-way ANOVA test (I, J, L, N, Q and R).

See also Figures S1 and S2.

To address the role of ADM in maintaining stemness, we conducted shRNA-mediated ADM depletion in GSCs. Western blotting showed that ADM depletion reduced the expression of stemness-associated factors CD133 and SOX2 in 005 GSC (Fig. 1G), a GSC line isolated from a genetically engineered GBM mouse model^40,41 and CT2A (Fig. S1B), a cell line isolated from mouse GBM tumors harboring GSC features^9,10,41, and human GSC272 (Fig. 1H). Tumorsphere formation assays demonstrated that ADM depletion reduced GSC self-renewal (Fig. 1I, J and Fig. S1C). The reduced CD133 and SOX2 expression and self-renewal were rescued by re-expression of shRNA-resistant ADM cDNA in 005 GSC (Fig. S1D, E) and CT2A cells (Fig. S1F, G), or by the treatment with ADM recombinant protein in GSC272 (Fig. S1H, I). Next, we performed ADM CRISPR knockout (KO) in GSCs and found that ADM KO reduced the expression of CD133 and SOX2 and tumorsphere formation in 005 GSC (Fig. 1K, L), GSC272 (Fig. 1M, N), and CT2A (Fig. S1J, K). Given that ADM is a secreted protein, we investigated whether its receptor ADMR, a complex formed by the calcitonin receptor-like receptor (CALCRL) and a receptor activity-modifying protein (RAMP), is required in this process. Using the brain TIME dataset⁴², we found that CALCRL was highly expressed in CD45⁻ GBM cells/GSCs and macrophages, but not in microglia, CD4 T cells, and CD8 T cells, of IDH1-WT GBM patient tumors (Fig. S1L). Next, we performed flow cytometry in 005 GSC tumors with results confirming that CALCRL was highly expressed in CD45⁻ GSCs, CD45⁺CD11b⁺CD68⁺ macrophages, and CD45⁺CD11b⁺CD68⁺CD206⁺ pro-tumor macrophages (Fig. S1M). Finally, we treated GSCs with ADM recombinant protein in the presence or absence of ADMR antagonist, AMA³⁰. The results showed that ADM protein treatment enhanced the expression of CD133 and SOX2, and GSC self-renewal, and these effects were abolished by the treatment with AMA (Fig. 1O–R and Fig. S1N). In addition, ADM depletion reduced GSC proliferation (Fig. S2A–C), shifted the cell cycle from G0/G1 to G2/M phases (Fig. S2D–F), and increased GSC apoptosis (Fig. S2G–I). Together, these findings demonstrate that ADM is critical for GSC self-renewal.

ADM promotes GSC self-renewal by regulating STAT3 signaling and glycolysis.

To explore the potential ADM-regulated signaling pathways, we performed bioinformatics analysis in TCGA GBM tumors with ADM-high versus ADM-low. Gene set enrichment analysis (GSEA) on hallmark pathways demonstrated that STAT3 was one of the top pathways enriched in ADM-high tumors (Fig. 2A) and has been shown to associate with GSC stemness^9,43. Western blotting results demonstrated that ADM depletion in GSCs reduced P-STAT3 (Fig. 2B, C and Fig. S3A), and the opposite effect was observed when GSCs were treated with ADM recombinant protein (Fig. 2D, E and Fig. S3B). Moreover, ADM-induced P-STAT3 in GSCs was negated by the treatment with AMA (Fig. 2F, G and Fig. S3C). Treatment with STAT3 inhibitor WP1066 negated ADM-induced self-renewal upregulation of 005 GSC (Fig. 2H), CT2A (Fig. S3D), and GSC272 (Fig. 2I). Given the enrichment of glycolysis signature in ADM-high tumors (Fig. 2A) and the important connection between glycolysis and GSC stemness^8,44, we next performed glycolysis assays in GSCs in response to ADM manipulations. Our results showed that ADM depletion or AMA treatment reduced extracellular acidification rate (ECAR) in GSCs (Fig. 2J–M and Fig. S3E, F), while ADM recombinant protein treatment exhibited an opposite effect (Fig. 2L, M). Moreover, ADM-induced ECAR was abolished by the treatment with WP1066 (Fig. 2L, M). Together, these findings suggest that ADM can activate STAT3 and glycolysis to promote GSC self-renewal.

Figure 2. ADM activates STAT3 signaling to promote GSC self-renewal and glycolysis.

(A) Gene Set Enrichment Analysis (GSEA) on hallmark signatures in TCGA GBM tumors showing the enriched pathways in ADM-high group. NES, normalized enrichment score. The pathways related to stemness are highlighted in blue. (B, C) Immunoblots for P-STAT3 and STAT3 in cell lysates of 005 GSC (B) and GSC272 (C) expressing shRNA control (shC) or ADM shRNA (shADM). (D, E) Immunoblots for P-STAT3 and STAT3 in cell lysates of 005 GSC (D) and GSC272 (E) treated with ADM recombinant protein at indicated time points and concentrations. (F, G) Immunoblots for P-STAT3 and STAT3 in cell lysates of 005 GSC (F) and GSC272 (G) treated with ADM protein (100 nM) in the presence or absence of ADM receptor antagonist, AMA (300 nM). (H, I) Representative images (left) and quantification (right) of tumorsphere formation assays in 005 GSC (H) and GSC272 (I) treated with ADM (100 nM) in the presence or absence of STAT3 inhibitor WP1066 (20 nmol/L). Scale bar, 200 μm. n = 3 independent samples. (J, K) Extracellular acidification rate (ECAR) of 005 GSC (J) and GSC272 (K) expressing shC and shADM. ECAR was obtained from the Seahorse experiments and glucose was added at the indicated time point. n = 6 independent samples. (L, M) ECAR of 005 GSC (L) and GSC272 (M) treated by ADM recombinant protein (100 nM) for 24 hrs in the presence or absence of AMA (300 nM) or WP1066 (20 nmol/L). ECAR was obtained from the Seahorse experiments and glucose was added at the indicated time point. n = 6 independent samples. Data from multiple replicates are presented as mean ± SEM. ****, P < 0.0001, one-way ANOVA test.

See also Figure S3.

ADM promotes macrophage migration and pro-tumor polarization.

Our recent studies have demonstrated that GSC stemness is linked to microglia immunosuppression in GBM^8–10,45, leading us to hypothesize a connection between ADM-regulated GSC stemness and tumor immunity. To identify specific type of immune cells linked to ADM expression, we performed GSEA on validated immune cell signatures^8–10,46,47 in TCGA GBM tumors. The results showed that high ADM expression correlated with significant enrichment of macrophages and monocytes, and to a lesser extent, dendritic cells (DCs), CD4 naïve T cells, myeloid-derived suppressor cells (MDSCs), hematopoietic stem cells (HSCs), and granulocytes, while microglia and other immune cells were not affected (Fig. 3A). Since ADM is a secreted protein and CALCRL is highly expressed in macrophages of GBM tumors (Fig. S1L, M), we investigated whether GSC-derived ADM might promote macrophage infiltration. The results showed that the conditioned medium (CM) from 005 GSC and GSC272 with ADM depletion via shRNA (Fig. 3B, C) or CRISPR (Fig. 3D, E) reduced migration of mouse Raw264.7 macrophages and human THP1 macrophages, respectively. The effect induced by shRNA-mediated ADM depletion was rescued by re-expression of shRNA-resistant ADM cDNA in 005 GSC (Fig. S4A). To investigate ADM’s chemoattractant ability in vivo, matrigel plugs with or without ADM supplementation in the presence or absence of AMA were implanted into mice subcutaneously. Immunofluorescence for CD68 demonstrated that macrophage density was significantly increased in the ADM-supplemented matrigel plugs compared with controls, and this enhancement was abolished by AMA (Fig. S4B). Together, these findings suggest that GSC-derived ADM is chemokine for macrophage infiltration in GBM.

Figure 3. ADM triggers macrophage migration and pro-tumor polarization.

(A) Gene set enrichment analysis (GSEA) on indicated immune cell signatures in TCGA GBM tumors with ADM-high versus ADM-low. The significant enriched immune cell signatures (FDR < 0.25) are highlighted in color (green and blue), and macrophage-related signatures highlighted in green. NES, normalized enrichment score. (B, C) Representative images (left) and quantification (right) of relative migration of mouse Raw264.7 macrophages (B) and human THP1 macrophages (C) following stimulation with the conditioned media (CM) from 005 GSC and GSC272, respectively, expressing shRNA control (shC) and ADM shRNA (shADM). Scale bars, 400 μm. n = 3 independent samples. (D, E) Representative images (left) and quantification (right) of relative migration of mouse Raw264.7 macrophages (D) and human THP1 macrophages (E) following stimulation with the CM from control and ADM KO 005 GSC and GSC272, respectively. Scale bars, 400 μm. n = 5 independent samples. (F-I) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD68⁺CD206⁺ (F, H) or ARG1⁺ (G, I) cells in Raw264.7 macrophages treated with the CM from 005 GSC (F, G) or GSC272 (H, I) expressing shC and shADM. n = 3 independent samples. (J-M) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD68⁺CD206⁺ (J, L) or ARG1⁺ (K, M) cells in Raw264.7 macrophages treated with the CM from control and ADM KO 005 GSC (J, K) or GSC272 (L. M). n = 3 independent samples.

Data from multiple replicates are presented as mean ± SEM. ***, P < 0.001, ****, P < 0.0001, Student’s t-test (D, E, J-M) or one-way ANOVA test (B, C, F-I).

See also Figure S4.

Once infiltrating onto the TME, macrophages are usually polarized towards a pro-tumor phenotype promoting tumor progression^48–53. GSEA results showed that pro-tumor macrophage signature was enriched in ADM-high GBM tumors (Fig. 3A), suggesting the potential role of ADM in macrophage polarization. Western blotting demonstrated that ADM treatment upregulated the expression of pro-tumor macrophage markers CD206 and ARG1 in Raw264.7 and THP1 macrophages, and these enhancements were negated by AMA treatment (Fig. S4C). Similar results from flow cytometry showing that recombinant ADM-supplemented medium dramatically increased the percentage of CD68⁺CD206⁺ and ARG1⁺ pro-tumor macrophages, an effect that was abolished by AMA treatment (Fig. S4D, E). To confirm this phenotype in the context of GBM, we performed flow cytometry on macrophages treated with the CM from control and ADM-depleted GSCs. The results showed that the CM from 005 GSC and GSC272 harboring shADM reduced the pro-tumor polarization of Raw264.7 and THP1 macrophages, respectively, by decreasing the percentages of CD68⁺CD206⁺ and ARG1⁺ cells (Fig. 3F–I). The effect was rescued by re-expression of shRNA-resistant Adm cDNA in 005 GSC (Fig. S4F, G). Similarly, the CM from 005 GSC and GSC272 harboring ADM CRISPR KO reduced the pro-tumor polarization of Raw264.7 and THP1 macrophages, respectively (Fig. 3J–M). These findings suggest that GSC-derived ADM is critical for macrophage pro-tumor polarization in GBM.

ADM promotes macrophage migration and pro-tumor polarization by activating STAT3 and STAT6 pathways.

To determine the potential downstream pathways that can mediate ADM-induced macrophage migration and polarization, we performed GSEA on TCGA GBM tumors with ADM-high versus ADM-low. Among the top enriched signatures in ADM-high tumors, the JAK-STAT signature (Fig. 4A) emerged as a key pathway that is important for regulating macrophage biology in the TME⁵⁴. STAT3 and STAT6 are two members of the STAT family involved in regulating macrophage biology, especially for macrophage pro-tumor polarization⁵⁵. scRNA-seq data analyses from GBM patient tumors^38,39 demonstrated that STAT3 was expressed in both GSCs and macrophages, while STAT6 was only highly expressed in macrophages (Fig. S5A–F), supporting our data that STAT3 is required for ADM-induced GSC self-renewal (Fig. 2 and Fig. S3) and our hypothesis that both STAT3 and STAT6 pathways involve in ADM-induced macrophage migration and polarization. Western blotting results confirmed an increased phosphorylation of STAT3 (P-STAT3) and P-STAT6 in macrophages upon the treatment with ADM recombinant protein (Fig. 4B), an effect that was negated by AMA (Fig. 4C). Functionally, transwell migration assays demonstrated that ADM-induced macrophage migration was abolished by pharmacologic inhibition of STAT3 or STAT6 (Fig. 4D). In addition, inhibition of STAT3 or STAT6 impaired ADM-induced macrophage pro-tumor polarization as indicated by reduced expression of CD206 and ARG1 in Raw264.7 and THP1 macrophages (Fig. 4E, F and Fig. S5G, H), as well as downregulated percentages of CD68⁺CD206⁺ and ARG1⁺ cells (Fig. 4G, H). To confirm the relationship between STAT3 and STAT6, control and ADM recombinant protein-conditioned macrophages were treated with or without STAT3 or STAT6 inhibitor. Western blotting assays demonstrated that pharmacologic inhibition of STAT6 upregulated P-STAT3 and was not able to block ADM-induced P-STAT3 in macrophages (Fig. S5I). Similarly, STAT3 inhibition enhanced P-STAT6 and had no effect to inhibit ADM-induced P-STAT6 in macrophages (Fig. S5J). Dual inhibition of STAT3 and STAT6 further negated ADM-induced macrophage pro-tumor polarization (Fig. 4G, H). Collectively, these findings indicate that ADM activates STAT3 and STAT6 signaling to drive macrophage infiltration and pro-tumor polarization in GBM.

Figure 4. STAT3 and STAT6 are responsible for ADM-induced macrophage migration and pro-tumor polarization.

(A) Gene Set Enrichment Analysis (GSEA) on KEGG signatures in TCGA GBM tumors showing the enriched pathways in the ADM-high group. NES, normalized enrichment score. (B) Immunoblots for P-STAT3, STAT3, P-STAT6, and STAT6 in cell lysates of Raw264.7 macrophages treated with ADM protein at indicated time points and concentrations. (C) Immunoblots for P-STAT3, STAT3, P-STAT6, and STAT6 in cell lysates of Raw264.7 macrophages treated with ADM protein (100 nM) for 20 mins in the presence or absence of ADM receptor antagonist, AMA (300 nM). (D) Representative images (left) and quantification (right) of relative migration of Raw264.7 macrophages treated with ADM protein (100 nM) in the presence or absence of STAT3 inhibitor (STAT3i) WP1066 (20 nmol/L) or STAT6 inhibitor (STAT6i) AS1517499 (100 nmol/L). Scale bars, 200 μm. n = 3 independent samples. (E, F) Immunoblots for CD206 and ARG1 in cell lysates of Raw264.7 macrophages treated with ADM protein (100 nM) in the presence or absence of WP1066 (20 nmol/L, E) or AS1517499 (100 nmol/L, F). (G, H) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD68⁺CD206⁺ (G) or ARG1⁺ (H) cells in Raw264.7 macrophages treated with ADM protein (100 nM) in the presence or absence of WP1066 (20 nmol/L) or AS1517499 (100 nmol/L). n = 3 independent samples. Data from multiple replicates are presented as mean ± SEM. *, P < 0.05, **, P < 0.01, ****, P < 0.0001, one-way ANOVA test.

See also Figure S5.

EGFR regulates ADM expression and its function in GSC-macrophage symbiosis.

To investigate the potential mechanism underlying ADM upregulation in GSCs, we performed a differential expression analysis comparing ADM-high tumors to ADM-low tumors, identifying 332 differentially expressed genes (Fig. 5A). Among them, 200 genes were positively correlated with ADM, and three genes (e.g., EGFR, RASL11B, and SEC61G) were found to be amplified in GBM tumors, with EGFR amplification observed in approximately 50% of patients with GBM (Fig. 5B), suggesting a connecting between EGFR and ADM. To confirm it through experimentation, we depleted EGFR in mouse CT2A cells and human GSC272 (both with relatively high EGFR expression), or overexpressed EGFR in GSC23 with relatively low EGFR expression (Fig. 5C–E and Fig. S6A). Western blotting results demonstrated that EGFR depletion reduced ADM expression in CT2A and GSC272 (Fig. 5C, D), whereas EGFR overexpression showed an opposite effect in GSC23 (Fig. 5E). Given that AKT/NF-κB/HIF1A and RAS/RAF/MEK/ERK are key EGFR downstream signaling pathways^56,57, we investigated whether EGFR can activate these signals and whether they are required for EGFR-induced ADM expression in GSCs. We found that EGFR overexpression in GSC23 significantly upregulated P-AKR, P-P65 and HIF1A, but did not affect P-ERK (Fig. S6B). Inhibition of AKT, NF-κB, or HIF1A with their inhibitor LY294002, SC75741, or acriflavine, respectively, negated EGFR overexpression-induced ADM upregulation in GSC23 (Fig. S6C). These findings suggest that ADM expression is regulated by the EGFR/AKT/NF-κB/HIF1A signaling axis in GSCs.

Figure 5. EGFR regulates ADM to affect GSC stemness and macrophage biology.

(A) Heatmap showing the differential gene expression between ADM-high and ADM-low tumors in the TCGA GBM dataset. (B) Genomic alterations of EGFR, SEC61G, and RASL11B in tumors of indicated TCGA GBM datasets. (C, D) Immunoblots for EGFR and ADM in cell lysates of CT2A (C) and GSC272 (D) expressing shRNA control (shC) or EGFR shRNA (shEGFR). (E) Immunoblots for EGFR and ADM in cell lysates of GSC23 expressing control or EGFR overexpression (OE) plasmids. (F) Immunoblots for SOX2 and CD133 in cell lysates of shC and shEGFR GSC272 treated with or without ADM recombinant protein (100 nM). (G) Representative images (left) and quantification (right) of tumorsphere formation assays in shC and shEGFR GSC272 treated with or without ADM recombinant protein (100 nM). Scale bar, 200 μm. n = 3 independent samples. (H) Representative images (left) and quantification (right) of relative migration of Raw264.7 macrophages treated with the conditioned media (CM) from shC and shEgfr CT2A cells in the presence or absence of ADM recombinant protein (200 nM). Scale bar, 400 μm. n = 3 independent samples. (I, J) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD68⁺CD206⁺ (I) or ARG1⁺ (J) cells in Raw264.7 macrophages treated with the CM from shC and shEgfr CT2A cells in the presence or absence of ADM recombinant protein (100 nM). n = 3 independent samples. Data from multiple replicates are presented as mean ± SEM. **, P < 0.01, ***, P < 0.001, ****, P < 0.0001, n.s., not significant, one-way ANOVA test.

See also Figure S6.

Given the role of ADM in GSC self-renewal and macrophage migration and polarization, we further investigated whether these processes are triggered by EGFR. Western blotting for stemness makers SOX2 and CD133 and tumorsphere formation assays demonstrated that EGFR overexpression-induced upregulation of GSC self-renewal was reduced by the treatment with AMA or WP1066 (Fig. S6D, E). In contrast, the reduction in GSC self-renewal induced by EGFR depletion was partially rescued by the addition of recombinant ADM protein (Fig. 5F, G). Similar results were observed for macrophage migration and pro-tumor polarization, as the CM from EGFR-overexpressing GSC23 enhanced THP1 macrophage migration and pro-tumor polarization, effects that were negated by AMA treatment (Fig. S6F, G). Conversely, the impaired macrophage migration and pro-tumor polarization induced by the CM from EGFR-depleted GSC272 was rescued by the addition of recombinant ADM protein (Fig. 5H–J). Together, these findings suggest that EGFR overexpression upregulates ADM to drive GSC self-renewal, and macrophage migration and pro-tumor polarization in GBM.

Inhibition of ADM-regulated GSC-macrophage symbiosis extends the survival of tumor-bearing mice.

To further investigate the role of ADM-mediated biology of GSCs and macrophages in GBM tumor progression in vivo, we utilized shRNA system to deplete ADM in 005 GSC tumors implanted into C57BL/6 mice, revealing that ADM depletion significantly extended survival (Fig. 6A). In a human PDX model of GSC272-bearing nude mice, we found that ADM depletion significantly extended the survival (Fig. 6B). Similarly, pharmacological inhibition of ADMR using AMA extended the survival of nude mice and C57BL/6 mice implanted with GSC272 and CT2A cells, respectively (Fig. 6C, D). Conversely, ADM recombinant protein treatment shortened the survival of CT2A tumor-bearing mice, an effect that was negated by pharmacologic inhibition of STAT3 or STAT6 (Fig. 6D). Given the role of STAT3 in regulating ADM-induced GSC stemness and the effect of STAT3 and STAT6 in ADM-triggered macrophage biology in GBM, we hypothesized that dual inhibition of STAT3 and STAT6 will achieve better survival benefits. Indeed, combined treatment with STAT3 inhibitor WP1066 and STAT6 inhibitor AS1517499 resulted in a complete tumor regression in in a subset of GSC272- and CT2A-bearing mice (Fig. 6C, D).

Figure 6. Inhibition of ADM and its associated pathways reduces GSC stemness, macrophage infiltration and polarization, and tumor growth in vivo.

(A, B) Survival curves of C57BL/6 (A) and nude (B) mice implanted with 005 GSC (2 × 10⁵ cells/mouse) and GSC272 (2 × 10⁵ cells/mouse), respectively, expressing shRNA control (shC) or ADM shRNA (shADM). n = 7–10 mice per group. log-rank test. (C) Survival curves of nude mice implanted with GSC272 (2 × 10⁵ cells/mouse) and treated with STAT3 inhibitor WP1066 (60 mg/kg, oral, three times a week for six doses), STAT6 inhibitor AS1517499 (10 mg/kg, i.p., twice a week for four doses), combined WP1066 and AS1517499, or ADM receptor antagonist, AMA (2 μg/mouse, i.p., daily for two weeks). n = 5–9 mice per group. log-rank test. (D) Survival curves of C57BL/6 mice implanted with CT2A cells (2 × 10⁴ cells/mouse) and treated with STAT3 inhibitor WP1066 (60 mg/kg, oral, three times a week for six doses), STAT6 inhibitor AS1517499 (10 mg/kg, i.p., twice a week for four doses), combined WP1066 and AS1517499, ADM recombinant protein (2 μg/mouse, i.p., daily for two weeks), or ADM receptor antagonist, AMA (2 μg/mouse, i.p., daily for two weeks). n = 5–6 mice per group. log-rank test. (E, F) Representative images (left) and quantification (right) of immunofluorescence staining for CD133 (E) and SOX2 (F) in shC and shAdm 005 GSC tumors. Scale bar, 75 μm. n = 3 independent samples. (G, H) Representative images (left) and quantification (right) of immunofluorescence staining for CD133 (G) and SOX2 (H) in shC and shADM GSC272 tumors. Scale bar, 75 μm. n = 3 independent samples. (I, J) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD45^hiCD11b⁺CD68⁺ macrophages (I) and CD45^hiCD11b⁺CD68⁺CD206⁺ pro-tumor macrophages (J) in size-matched shC and shAdm 005 GSC tumors. n = 3 independent samples. (K, L) Representative images (left) and quantification (right) of flow cytometry for the percentage of CD45^hiCD11b⁺CD68⁺ macrophages (K) and CD45^hiCD11b⁺CD68⁺CD206⁺ pro-tumor macrophages (L) in size-matched CT2A tumors treated with or without WP1066, AS1517499, combined WP1066 and AS1517499, ADM recombinant protein, or AMA. n = 3 independent samples.

Data from multiple replicates are presented as mean ± SEM. **, P < 0.01, ***, P < 0.001, ****, P < 0.0001, one-way ANOVA test.

See also Figure S7.

To confirm whether these observed survival benefits are related to GSC stemness and macrophage biology, we performed histological studies in tumors. Immunofluorescence for CD133, SOX2, Ki67 and cleaved caspase 3 (CC3) demonstrated that GSC self-renewal and GBM cell/GSC proliferation were dramatically reduced, whereas apoptosis was increased upon ADM depletion (Fig. 6E–H and Fig. S7A, B) or AMA treatment (Fig. S7C–F) in both 005 GSC and GSC272 models. Flow cytometry demonstrated that CD45^highCD11b⁺CD68⁺ macrophages and CD45^highCD11b⁺CD68⁺CD206⁺ pro-tumor macrophages were profoundly reduced in ADM-depleted 005 GSC tumors (Fig. 6I, J), upregulated by ADM recombinant protein treatment, an effect that was abolished by inhibition of STAT3 or STAT6 (Fig. 6K, L). Moreover, GBM cell/GSC proliferation and the abundance of macrophages and pro-tumor macrophages were dramatically decreased, whereas apoptosis was significantly increased in combined WP1066 and AS1517499 treatment group compared with the single treatment groups (Fig. 6K, L and Fig. 76G, H). Together, these in vivo findings support the role of GSC ADM and its associated STAT3 and STAT6 signals in GBM by regulating GSC stemness and macrophage biology.

ADM expression correlates with GSC stemness, macrophage biology, and survival in human GBM.

To assess the clinical relevance of ADM in GBM, we first performed bioinformatics analyses in TCGA GBM tumors showing that ADM expression was correlated positively with GSC stemness signature⁵⁸, macrophage signature, and pro-tumor macrophage signature⁴⁷ (Fig. 7A–C). Next, we performed immunohistochemistry for ADM, SOX2 and CD133 in tumors from a cohort of 60 patients with GBM and for ADM and CD206 in a GBM tissue microarray consisting of 35 GBM tumors. We found that ADM expression was positively correlated with SOX2, CD133, and CD206 (Fig. 7D, E). Given that ADM encodes a secreted protein, we performed ELISA assay in the plasma from heathy controls, patients with meningioma, and patients with GBM showing that ADM levels were higher in patients with GBM than healthy controls and patients with meningioma (Fig. 7F). Plasma ADM levels were not related to the status of gender, age, recurrence, and MGMT methylation in patients with GBM (Fig. S8A–D), but correlated negatively with patient survival (Fig. 7G). Together, these findings from GBM patient samples support the role of ADM in regulation of GSC stemness, macrophage infiltration and pro-tumor polarization, and highlight ADM as a potential prognostic biomarker for patients with GBM.

Figure 7. ADM tracks with GSC stemness, macrophages, and survival in patients with GBM.

(A-C) The correlation of ADM with stemness signature (A), macrophage signature (B), and pro-tumor macrophage signature (C) in TCGA GBM dataset (n = 489). R and P values are shown. Pearson’s correlation. (D) Representative images (left) showing the low and high expression levels of ADM, CD133, and SOX2 in GBM patient tumors based on immunohistochemistry staining and correlation analysis (right) between ADM and CD133 or ADM and SOX2 in GBM patient tumors (n = 60). R and P values are shown. Pearson’s correlation. Scale bar, 100 μm. (E) Representative images (left) showing the low and high expression levels of ADM and CD206 in GBM patient tumors based on immunohistochemistry staining and correlation analysis (right) between ADM and CD206 in GBM patient tumors (n = 35). R and P values are shown. Pearson’s correlation. Scale bar, 100 μm. (F) ELISA for ADM in the plasma from healthy controls (n = 10), meningioma (n = 15), and glioblastoma (n = 63) patients. **, P <0.01, ns, not significant, Student t-test. (G) Kaplan-Meier survival curves of patients with GBM relative to high (n = 21) and low (n = 32) serum ADM level. Log-rank test.

See also Figure S8.

DISCUSSION

In this study, we identified neuropeptide ADM as a critical GSC-associated factor that promotes GSC stemness and regulates the migration and pro-tumor reprograming of macrophages, the most enriched myeloid cells in the GBM TME accounting for up to half of cells within the tumor mass^50,59. We established that ADM upregulation in GSCs is triggered by overexpression of EGFR, the most frequently amplified/mutated gene happening in about 50%−60% GBM tumors⁶⁰. The enhanced ADM promotes GSC self-renewal by activating the STAT3 signaling pathway. Through a paracrine effect, ADM triggers macrophage infiltration and pro-tumor polarization via activation of both STAT3 and STAT6 pathways. In GBM mouse and PDX models, blockade of ADM-mediated GSC-macrophage crosstalk significantly impairs tumor progression, and the anti-tumor effect is heightened when tumor-bearing mice receive the treatment with combined STAT3A and STAT6 inhibition. Analyses of tumor and plasma samples from patients with GBM support that ADM is a potential biomarker whose expression correlates with GSC stemness, macrophage abundance, and GBM patient outcome. Together, our work elucidates a mechanism underlying GSC stemness and GSC–macrophage symbiosis and informs a compelling therapeutic strategy targeting ADM-mediated GSC–macrophage symbiosis in GBM.

The lack of effective treatments for GBM is partly due to the presence of GSCs, which play an important role in driving tumor recurrence and therapeutic resistance^14,61. Genetic alterations are key drivers for GBM tumor tumorigenesis and stemness maintenance. Our recent studies have identified CLOCK and TFPI2, which are amplified in about 5% and 4% of human GBM tumors, as important players in GSC stemness maintenance^8,9. Genetic and pharmacologic inhibition of CLOCK and TFPI2 and their associated signaling pathways impairs GSC stemness and tumor progression^8,9. In this study, we provide further evidence supporting that EGFR amplification/overexpression is the key genetic event that upregulates the expression of ADM in GSCs to maintain their self-renewal properties. Comparing to CLOCK and TFPI2 alterations, EGFR amplification/mutation is the dominant genetic event observed in more than half of patients with GBM⁶². However, targeted therapies against the EGFR-receptor tyrosine kinase (RTK) signaling have shown limited anti-tumor efficacy in patients with GBM^62,63, motivating us to identity ADM as a targetable EGFR downstream signaling that connects GSC stemness to the GBM TME.

ADM is a 52-amino acid regulatory peptide playing an important role in tumor growth and metastasis by promoting tumor cell proliferation, survival, and aggressiveness, and tumor angiogenesis⁶⁴. In GBM, limited evidence demonstrates that inhibition of ADM can impair tumor growth and angiogenesis and induces vascular normalization in the GBM TME^31–33. In this study, scRNA-seq analysis of GBM patient tumors and functional studies revealed that ADM is highly expressed in GSCs, where it plays a crucial role in maintaining GSC stemness. These findings not only provide molecular insights into the relationship between EGFR and stemness⁶⁵, but also gain added significance as the GSC model can better recapitulate GBM tumor heterogeneity⁶⁶. In exploring the molecular basis underlying ADM’s role in promoting GSC stemness, we found that ADM activates the STAT3 pathway, which is critical for GSC stemness maintenance^9,43, highlighting the translational potential of targeting STAT3 in ADM-high patients with GBM.

Increasing evidence highlights the importance of cell-cell symbiotic interaction in GBM progression^{48,49,51–53}. ADM is a neuropeptide exhibiting an important function in neuroprotection and neurological diseases^23–26, suggesting a potential role of ADM in tumor-neuron interaction that is critical for GBM progression^15,16. In this study, we revealed that ADM is highly expressed by GSCs, a population of GBM cells originating from NSCs¹⁷ that can establish a GSC-TME symbiosis in a context-dependent manner, influenced by specific genetic alterations^7–9,14,46. We have demonstrated that deficiency of PTEN, a component of the RTK/RAS/PI3K/PTEN pathway, can upregulate the expression of chemokine lysyl oxidase (LOX) to promote macrophage infiltration into the TME⁴⁶. Results from the current study indicate that amplification/overexpression of EGFR, another component of the RTK/RAS/PI3K/PTEN pathway, increases the expression of ADM to promote macrophage infiltration and pro-tumor polarization in the GBM TME. Further studies are needed to elucidate the relationship between LOX and ADM in GSCs and their coordinated effect on macrophage biology in GBM. The function of ADM in macrophage biology is consistent with previous studies in other caner types, showing that ADM promotes macrophage polarization in an autocrine mechanism^30,67. Our findings in this study highlight a paracrine mechanism by which GSC-derived ADM influences macrophage infiltration and pro-tumor polarization. Through an unbiased GSEA followed by functional validations, we identified STAT3 and STAT6 as the key downstream signals responsible for ADM-induced macrophage infiltration and pro-tumor polarization. Further studies are needed to investigate how ADM activates STAT3 and STAT6 in macrophages. However, these findings gain added significance as they provide a connection between ADM and the STAT pathway in macrophages, particularly considering that previous studies have validated the critical roles of STAT3 and STAT6 in macrophage biology^9,47,55. In addition to the scientific significance of identifying STAT3 and STAT6 as downstream signals of ADM in macrophages, our findings highlight these proteins as potential targets for EGFR/ADM-high GBM responder populations, especially considering the potent anti-tumor effects observed in our preclinical trials with STAT3 and/or STAT6 inhibitors.

The function of ADM-mediated GSC-macrophages symbiosis in GBM is supported by our preclinical trials and clinical sample validations. Our animal work demonstrates that inhibition of ADM-AMDR-STAT3/STAT6 signaling pathway extends the survival of tumor-bearing mice. In human GBM, tumor and plasma ADM tracks with GSC stemness, macrophage abundance, and patient survival. Together, our mechanistic insights into the GSC-macrophage symbiosis and anti-tumor responses in preclinical models support the initiation of clinical trials to evaluate the efficacy and safety of targeting this symbiosis in GBM. Immunotherapy is promising for treating several types of solid tumors. However, the benefits of immunotherapy in GBM are only observed in a subset of patients, probably due to the presence of GSCs and immunosuppressive TME, especially the immunosuppressive macrophages^68,69. Given the role of ADM in promoting GSC self-renewal and macrophage immunosuppressive polarization in GBM, it is tempting to speculate that the combination of targeting ADM-mediated GSC-macrophage symbiosis and immunotherapy could benefit patients with GBM, warranting further preclinical and clinical investigations.

Limitations of the study

Although we demonstrated that ADMR is essential for ADM in regulating GSC and macrophage biology, the precise mechanism by which ADM interacts with ADMR on GSCs and macrophages remains unclear. We also showed that STAT3 and STAT6 act as downstream effectors of ADM and are required for ADM-induced macrophage migration and pro-tumor polarization. However, the mechanism by which ADM activates STAT3 and STAT6 in macrophages, as well as their relationship during macrophage migration and pro-tumor polarization in GBM, remains unclear. This manuscript focuses on macrophages, the most abundant immune cells in the GBM TME. Further studies using scRNA-seq and spatial transcriptomics are needed to elucidate the immune landscape of GBM tumors following ADM depletion.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Peiwen Chen (chenp6@ccf.org).

Materials availability

This study did not generate new unique reagents.

Data and code availability

This paper analyses publicly available data. The references for these datasets are listed within the relevant results sections.

This study does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR+METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice and intracranial xenograft tumor models

Female C57BL/6 (#000664) and nude mice (#007850) mice were purchased from the Jackson Laboratory and housed in pathogen-free conditions. All animal experiments were conducted with the approval by the Institutional Animal Care and Use Committee (IACUC). Intracranial CT2A, 005 GSC, and GSC272 tumor models were established as previously described^9,47,70. Mice were euthanized upon reaching humane endpoints and brains of tumor-bearing mice were harvested for following analyses.

GBM patient samples

Peripheral blood plasma from meningioma (n = 15) and glioblastoma (n = 63) patients, and tumor samples (n = 60) from surgically resected IDH-WT glioblastomas were collected at the Northwestern Central Nervous System Tissue Bank (NSTB) under the institutional review board protocol STU00095863. All patients were diagnosed according to the WHO diagnostic criteria. The informed consent for research was obtained from the patients. Patient details are as described in our recent report⁹. The commercially available anonymized and de-identified control plasma (n = 10) were isolated from healthy human blood and purchased from Solomon Park Research Laboratories (#4345). GBM tissue microarray (TMA, n = 35) was purchased from US Biomax (#GL806f).

METHOD DETAILS

Cell culture

THP1 and Raw264.7 cells (from ATCC) were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS, Fisher Scientific, #16140071) and a 1:100 antibiotic-antimycotic solution (Gibco, #15140–122), and differentiated into macrophages using 200 ng/mL of phorbol 12-myristate 13-acetate (Sigma–Aldrich, #P8139) for 24 hrs. CT2A and 293T cells (from ATCC) were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; Gibco, #11995–065) supplemented with 10% FBS and a 1:100 antibiotic-antimycotic solution. 005 GSCs (from Dr. Samuel D. Rabkin at Massachusetts General Hospital), CT2A (for stemness maintenance), and human GSCs, such as GSC272, GSC23, GSC17, GSC20, GSC6–27, and GSC107 (from Dr. Frederick F. Lang at MD Anderson Cancer Center) were cultured in neural stem cell (NSC) proliferation medium (Millipore, #SCM005) containing 20 ng/mL epidermal growth factor (EGF; PeproTech, #AF-100–15) and basic fibroblast growth factor (bFGF; PeproTech, #100–18B). All cell lines were not authenticated; however, they were regularly tested negative for mycoplasma and maintained at 37°C in a 5% CO₂ incubator. CM was collected from number-matched control and ADM shRNA knockdown GSCs after culturing for another 24 hrs in growth factor-free culture medium.

Plasmids, viral transfections, and cloning

shRNAs targeting mouse Adm and Egfr as well as human ADM and EGFR in the pLKO.1 vector (Sigma, #SHC001) were used. Lentiviral particles were produced as we described previously^9,47,70,71. The following human and mouse shRNA sequences (ADM: #1: TRCN0000083200, #2: TRCN0000083202, and #3: TRCN0000271497; Adm: #1: TRCN0000340244 and #2: TRCN0000340245; EGFR: #4: TRCN0000039633 and #5: TRCN0000039635; and Egfr: #3: TRCN0000334048 and #4: TRCN0000055222) were selected after knockdown efficiency validation. The stable EGFRviii overexpression was performed in GSC23 using the EGFRviii open reading frame (ORF) lentiviral vector (from Dr. Ronald DePinho at MD Anderson Cancer Center), and the stable ADM overexpression was performed in 005 GSC and CT2A cells using the Adm ORF lentiviral vector (Origene, #SKU MR201693L4). Sublines were generated by blasticidin selection as we described previously⁸.

CRISPR knockout

The CRISPR knockout experiments were performed as we described previously. Briefly, sgRNA plasmids targeting human ADM (Santa Cruz, #sc-402779) and mouse Adm (Santa Cruz, #sc-419009) were transiently transfected into human (GSC272) and mouse (005 GSC and CT2A) GSCs, respectively, using Lipofectamine 2000. GSCs were harvested 72 hr later and ten GFP-positive cells were sorted into each well of a 96-well plate by flow cytometry, followed by immunoblotting for ADM protein.

Proliferation assay

Cell proliferation was assessed using the CellTrace CFSE Cell Proliferation Kit (Invitrogen, #C34554). Briefly, 10⁶ cells were incubated with the CFSE working solution (1:1000) for 20 mins at 37°C and then cultured in the dark for two days. Cells were then applied to flow cytometry and proliferation was analyzed using FlowJo v10.8.1.

Cell cycle and apoptosis assays

For cell cycle analysis, cells were seeded in 6-well plates and cultured for 24 hrs before being fixed in ice-cold 70% ethanol for 30 mins at 4°C. Then, cells were treated with RNase A solution (Promega, #A797C; 100 μg/ml) for 5 mins at room temperature, followed by incubation with RedX-labeled propidium iodide (PI) (BioLegend, #421301; 50 μg/ml) for 10 mins at 4 °C and analyzed by flow cytometry. For cell apoptosis analysis, cells were incubated with Apotracker (1:10, FITC, BioLegend, #427402) followed by PI (BioLegend, #421301). Apoptotic (FITC+PI⁻) cells were analyzed using a BD FACSymphony flow cytometer.

Tumorsphere formation assay

GSCs at different conditions were seeded at 100 cells/well in a 96-well ultra-low attachment plate (Corning) in NSC media (Millipore, #SCM005). Tumorsphere numbers were imaged and quantified after 14 days.

Migration assay

10⁵ THP1 or Raw264.7 macrophages were suspended in serum-free culture media and seeded into transwell inserts (5 μm pore size, Corning, #3421). GSC conditioned media (CM) or ADM recombinant protein was added to the lower chambers. After 10 hrs, migrated cells were fixed in 4% PFA (Thermo Fisher Scientific, #J61899.AK) for 30 mins, then stained with crystal violet (Sigma-Aldrich, #C-3886) for 30 mins. The migrated cells were captured using an EVOS microscope and quantified using ImageJ software.

Immunofluorescence and immunohistochemistry

Immunofluorescence and immunohistochemistry were performed as we described previously^9,19,47,70. After the staining, images were captured using a Nikon AX/AXR Confocal Microscope System (for immunofluorescence) or EVOS Cell Imaging System (for immunohistochemistry) and analyzed using Image J. The primary antibodies against following proteins were used: Ki67 (Cell Signaling Technology, #9129S), and CC3 (Cell Signaling Technology, #9661S), F4/80 (Cell Signaling Technology, #30325S), CD163 (Invitrogen, #MA5–11458), CD206 R&D Systems, #AF2535), CD68 (Abcam, #ab53444), CD133 (Abcam, #ab5558), and SOX2 (Abcam, #ab97959).

Quantitative real-time PCR (RT-qPCR)

Total RNA was extracted from pelleted cells using the RNeasy Mini Kit (Qiagen, #74106). cDNA was synthesized using the All-In-1 5× RT MasterMix (Applied Biological Materials, #G592) in a T100 Thermal Cycler (Bio-Rad). RT-qPCR was performed using SYBR Green PCR Master Mix (Bio-Rad, #1725275) on a CFX Connect Real-Time PCR Detection System (Bio-Rad). RT-qPCR primers include:

ACTIN-F: CATGTACGTTGCTATCCAGGC; ACTIN-R: CTCCTTAATGTCACGCACGAT ADM-F: ATGAAGCTGGTTTCCGTCG; ADM-R: GACATCCGCAGTTCCCTCTT

Immunoblotting

Protein expression in GSCs and macrophages was analyzed by Western blotting as we detailed previously^9,19,47,70. The following primary antibodies against following proteins were used: Actin (Cell Signaling Technology, #3700S), ADM (Invitrogen, #PA5–24927), EGFR(Cell Signaling Technology, #4267), SOX2 (Abcam, #ab97959), CD133 (Abcam, #ab5558), CD206 (R&D Systems, #AF2535), ARG1 (Cell Signaling Technology, #93668S), STAT3 (Cell Signaling Technology, #9139), P-STAT3 (Cell Signaling Technology, #9145S), STAT6 (Cell Signaling Technology, #9362), P-STAT6 (Abcam, #ab28829), AKT (CST, #4691), P-AKT (CST, #4058), ERK (CST, #4695), P-ERK (CST, #9101), P65 (CST, #6956), P-P65 (CST, #3033), and HIF1A (CST, #36169). Each assay was repeated at least three times and the signaling was imaged under ChemiDoc Touch Imaging System (Bio-Rad).

Flow cytometry

The flow cytometry analyses of macrophages from tumors and cells were performed using an approach as we described previously⁷⁰. For intratumoral macrophage and GSC analysis, following antibodies, PerCP/Cy5.5 anti-mouse CD45 (BioLegend, #103132), PE anti-mouse CD68 (BD Bioscience, #566386), PE/Cy7 anti-mouse/human CD11b (BioLegend, #101216), and AF647 anti-mouse CD206 (BD Bioscience, #565250), were used. For macrophage cells, FITC anti-human/mouse ARG1 (BioLegend, #IC5868F), PE anti-mouse CD68 (BD Bioscience, #566386), AF647 anti-mouse CD206 (BD Bioscience, #565250), and CALCRL (Alomone Labs, #ACR-060-F) were used. Samples were read using either a BD FACSymphony or BD LSRFortessa flow cytometer and analyzed using FlowJo v10.8.1.

Metabolic assay

The extracellular acidification rate (ECAR) of GSCs was measured using the Seahorse XF Cell Mito Stress Test Kit (Agilent Technologies, #103015–100) on a Seahorse XFe96 analyzer, following our recently reported protocol⁴⁷.

Enzyme-linked immunoassay (ELISA)

Plasma levels of ADM from healthy controls, and patients with meningioma and GBM were measured by ELISA using the commercial human ADM kit (ThermoFisher Sientific, #EEL023) following the manufacturer’s instructions.

scRNA-seq data analysis

scRNA-seq data of GSE182109 and GSE135045 were used for analyzing the expression of ADM in GBM cells and GSCs, and analyzing the expression of STAT3 and STAT6 in different populations of cells in GBM patient tumors.

Computational analysis of human GBM datasets

TCGA GBM datasets were downloaded from the GlioVis (http://gliovis.bioinfo.cnio.es/). Using these datasets, analyses for gene expression, correlations between ADM and gene signatures (e.g., stemness, macrophage and pro-tumor macrophages signatures), and GSEA of interesting gene signatures were performed as we recently reported^46,47. The brain TIME dataset⁴² was used to examine the expression level of CALCRL in CD45⁻ GBM cells/GSCs, macrophages, microglia, CD4, and CD8 T cells.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using GraphPad Prism 10. Survival analysis was conducted using the Log-rank (Mantel-Cox) test. Comparisons between two groups were performed using Student’s t-test, while multiple-group comparisons were analyzed using one-way ANOVA test. Data are presented as mean ± SEM. Correlation analyses were performed using Pearson’s correlation test in GraphPad Prism 10. P values were designated as *, P < 0.05; **, P < 0.01, ***, P < 0.001, and ****, P < 0.0001. n.s., not significant (P > 0.05).

KEY RESOURCES TABLE

REAGENT OR RESOURCE	SOURCE	IDENTIFIER
Antibodies
ADM	Invitrogen	Cat#PA5–24927
EGFR	Cell Signaling	Cat#4267
Anti-mouse IgG, HRP-Linked	Cell Signaling	Cat#7076S
Anti-rabbit IgG, HRP-linked	Cell Signaling	Cat#7074S
P-STAT6	Abcam	Cat#ab28829
STAT6	Cell Signaling	Cat#9362
P-STAT3	Cell Signaling	Cat#9145S
STAT3	Cell Signaling	Cat#9139
β-ACTIN	Cell Signaling	Cat# 3700S
RedX-labeled propidium iodide (PI)	BioLegend	Cat#421301
Apotracker Green	Biolegend	Cat#427402
Ki67	Cell Signaling	Cat#9129S
CC3	Cell Signaling	Cat#9661S
F4/80	Cell Signaling	Cat#30325S
CD163	Invitrogen	Cat#MA5–11458
CD206	R&D Systems	Cat#AF2535
CD68	Abcam	Cat#ab53444
CD133	Abcam	Cat#ab5558
SOX2	Abcam	Cat#ab97959
AKT	Cell Signaling	Cat #4691
P-AKT	Cell Signaling	Cat#4058
ERK	Cell Signaling	Cat#4695
P-ERK	Cell Signaling	Cat#9101
P65	Cell Signaling	Cat#6956
P-P65	Cell Signaling	Cat#3033
HIF1A	Cell Signaling	Cat#36169
PerCP/Cy5.5 anti-mouse CD45	BioLegend	Cat#103132
PE anti-mouse CD68	BD Bioscience	Cat#566386
PE/Cy7 anti-mouse/human CD11b	BioLegend	Cat#101216
AF647 anti-mouse CD206	BD Bioscience	Cat#565250
FITC anti-human/mouse ARG1	BioLegend	Cat#IC5868F
FITC anti-mouse/human CALCRL	Alomone Labs	Cat#ACR-060-F
Bacterial and virus strains
N/A	N/A	N/A
Biological samples
Human patient tumor samples from surgically resected IDH-WT GBMs	Northwestern Central Nervous System Tissue Bank	N/A
Chemicals, peptides, and recombinant proteins
Adrenomedullin (ADM) 1–52, human	GenScript	Cat#RP11288
Adrenomedullin (AMA) 22–52, human	GenScript	Cat#RP11293
STAT3 inhibitor WP1066	Selleckchem	Cat# S2796
STAT6 inhibitor AS1517499	Selleckchem	Cat# S8685
AKT inhibitor LY294002	MedChemExpress	Cat# HY-10108
NF-κB inhibitor SC75741	MedChemExpress	Cat# HY-10496
HIF1A inhibitor acriflavine	Sigma	Cat# A8126
Critical commercial assays
BCA protein assay kit	Thermo Fisher Scientific	Cat#23225
ELISA commercial human ADM kit	ThermoFisher Sientific	Cat#EEL023
RNeasy Mini Kit	Qiagen	Cat#74106
CellTrace CFSE Cell Proliferation Kit	Invitrogen	Cat#C34554
Seahorse XF Cell Mito Stress Test Kit	Agilent Technologies	Cat#103015–100
Deposited data
Single-cell sequencing data	Abdelfattah et al. (38)	GEO, GSE182109
Single-cell sequencing data	Xiao et al. (39)	GEO, GSE135045
TCGA expression data	http://gliovis.bioinfo.cnio.es/	N/A
The brain TIME dataset	https://joycelab.shinyapps.io/braintime/	N/A
Experimental models: Cell lines
CT2A	Seyfried et al.	N/A
005 GSC	Dr. Samuel D. Rabkin (Massachusetts General Hospital, Boston, USA)	N/A
GSC272, GSC17, GSC20, GSC23, GSC6–27, GSC107	Dr. Frederick Lang (MD Anderson Cancer Center, Houston, USA)	N/A
THP1	ATCC	Cat#TIB-202; RRID:CVCL_0006
RAW264.7	ATCC	Cat#TIB-71; RRID:CVCL_0493
293T	ATCC	Cat#CRL-11268; RRID:CVCL_1926
Experimental models: Organisms/strains
Mouse C57BL/6	Jackson Laboratory	000664
Nude Mice	Jackson Laboratory	007850
Oligonucleotides
ADM primer #1 (human)-F (RTqPCR)	This paper	ATGAAGCTGGTTTCCGTCG
ADM primer #1 (human)-R (RTqPCR)	This paper	GACATCCGCAGTTCCCTCTT
ACTIN primer #1 (human)-F (RTqPCR)	This paper	CATGTACGTTGCTATCCAGGC
ACTIN primer #2 (mouse)-R (RTqPCR)	This paper	CTCCTTAATGTCACGCACGAT
Recombinant DNA
psPAX2	Addgene	Cat #12260
pMD2.G	Addgene	Cat #12259
pLKO.1	Sigma Aldrich	Cat # SHC001
pLKO.1-human-ADM shRNA #1	Sigma-Aldrich	TRCN0000083200
pLKO.1-human-ADM shRNA #2	Sigma-Aldrich	TRCN0000083202
pLKO.1-human-ADM shRNA #3	Sigma-Aldrich	TRCN0000271497
pLKO.1-mouse-Adm shRNA #1	Sigma-Aldrich	TRCN0000340244
pLKO.1-mouse-Adm shRNA #2	Sigma-Aldrich	TRCN0000340245
pLKO.1-human-EGFR shRNA #4	Sigma-Aldrich	TRCN0000039633
pLKO.1-human-EGFR shRNA #5	Sigma-Aldrich	TRCN0000039635
pLKO.1-mouse-Adm shRNA #3	Sigma-Aldrich	TRCN0000334048
pLKO.1-mouse-Adm shRNA #4	Sigma-Aldrich	TRCN0000055222
ADM CRISPR KO plasmid- human	Santa Cruz	Cat # sc-402779
Adm CRISPR KO plasmid mouse	Santa Cruz	Cat # sc-419009
Adm ORF lentiviral vector	Origene	Cat # SKU MR201693L4
EGFR ORF lentiviral vector	From Dr. Ronald DePinho (MD Anderson Cancer Center, Houston, USA)	N/A
Software and algorithms
Prism GraphPad 10	Prism	https://www.graphpad.com/scientific-software/prism/
ImageJ	Fiji Image	https://imagej.net/Fiji
Image Lab	Bio-Rad	https://www.bio-rad.com/en-us/product/image-lab-software?ID=KRE6P5E8Z
GSEA-4.1.0	Broad Institute	http://software.broadinstitute.org/gsea/index.jsp
FlowJo v10.8.1.	BD Biosciences	https://www.flowjo.com/flowjo/quote-requestv
BioRender	BioRender	https://www.biorender.com/
Other
N/A	N/A	N/A