Mapping the exhausted and tolerized myeloid populations in the glioblastoma tumor microenvironment.

2066 Background: Glioblastoma (GBM) is a highly malignant brain tumor characterized by an immunosuppressive microenvironment and rapid development of therapeutic resistance. In GBM, myeloid cells can comprise 30–50% of the tumor and critically regulate its immune microenvironment. Evidence from aging and CNS diseases suggests that microglia can adopt dysfunctional states of exhaustion and tolerance that extend beyond the M1/M2 paradigm, thus prompting us to investigate whether these phenotypes contribute to immune escape and progression of GBM. Methods: Paired RNA-seq of tumor specimens from newly diagnosed and recurrent GBM patients (n=14) was used to examine aging-related gene expression during treatment and progression. Open-access human glioma single-cell RNA-seq data (GSE182109) were analyzed to characterize exhaustion and tolerance across macrophage subpopulations based on molecular/metabolic signatures. Results: Patient tumor RNA-Seq revealed that the aging-related TREM2-APOE axis is present in recurrent GBM, suggesting accelerated brain aging. Single-cell RNA-seq identified nine myeloid clusters, including four microglial states (homeostatic-, activated-, AP-, and a-microglia) and two macrophage populations (anti-inflammatory and immunosuppressive). We found that AP-microglia showed disease-associated activation, with high CD45/CD11C/TREM2 expression and concurrent upregulation of the exhaustion markers PD-L1 and CD38. Moreover, microglial activation-related signaling in AP-microglia includes increased STAT1/2-IRF9 signaling, which can promote immunosuppression by inducing PD-L1. On the other hand, a-microglia is unique in its high expression of activation-restraining and tolerance markers, such as SPRY/P2RY13, rather than the classic microglial marker TMEM119/P2RY12. Both AP-microglia and a-microglia are shown to lose expression of the inflammatory cytokines TNF-α and IL-1β, consistent with the characteristics of exhausted and tolerized microglia. Unlike resident microglia, the immunosuppressive macrophage cluster exhibits a monocyte-tolerized phenotype, with downregulated FABP4 and increased SOD2, CLEC4E, and SLC2A6. It also shows a similar panel of LPS-induced monocyte exhaustion with TLR4/MyD88/SRC and STAT3/IL-10 upregulation. Conclusions: Our analysis indicates that AP-microglia are exhausted and a-microglia are tolerized in GBM, with shared markers of dysfunction across macrophages linking myeloid impairment to immune suppression and treatment-associated brain aging. These findings implicate myeloid exhaustion in GBM immune escape and identify potential therapeutic targets.