Abstract Glioblastoma (GBM) is the most prevalent primary brain malignancy in adults. The current standard of care includes maximal surgical resection followed by radio- and chemotherapy with temozolomide. Yet <5% of GBM patients survive more than five years. This indicates a desperate need for more effective treatments, such as immunotherapy for GBM patients. Unfortunately, most immunotherapy trials, including vaccines, adoptive cellular therapy, CAR-T cells, and checkpoint blockade, showed only modest benefits in GBM patients. A major barrier to immunotherapy efficacy is GBM’s immunosuppressive microenvironment composed of few tumor infiltrating lymphocytes (TILs; <5%) but abundant myeloid cells, making it an immune cold tumor. By contrast, immune hot tumors, characterized by abundant tumoricidal effector T cells necessary to mount a meaningful attack, have consistently responded better to immunotherapy. Hence, a better definition of the heterogeneous cell types in the GBM microenvironment and their function is urgently needed. Fortunately, single cell transcriptomics approaches provide comprehensive and high-resolution cellular and molecular understanding to resolve this heterogeneity. Here we report an integrated, multiregional and -dimensional single cell transcriptomic analysis of 201,986 human glioma and immune cells derived from 44 tissue fragments from 18 human glioma patients. In doing so, we map GBM cellular heterotypia and spatial, molecular, and functional heterogeneity of glioma associated immune cells. We report extensive spatial and molecular heterogeneity of glioma cells, microglia, macrophages, and T cells within the same tumor samples in low grade gliomas, primary GBMs, and recurrent GBMs. Importantly, our analysis of 83,479 glioma infiltrating myeloid cells identifies 9 molecularly distinct myeloid subtypes: 4 microglial, 4 bone marrow derived macrophage and dendritic cells subtypes. Importantly, in multiple independent glioma patient cohorts, 5 of these myeloid cell subtype gene signatures were independent predictors of patient survival. We also provide evidence that cell:cell communication between glioma and immune cells is more robust than glioma:Tcells, indicating that myeloid cells form a communication hub in vivo. Additionally, we identified S100A4 as highly expressed in immunosuppressive macrophages and T cells, and provide in vitro and in vivo evidence that S100a4 plays a critical role in promoting immunosuppressive phenotypes in glioma associated leukocytes. This study not only provides the first comprehensive single cell atlas of GBM to include both glioma and immune cells from same samples but also demonstrates its utility in elucidating cell:cell communication among different cell types in vivo and discovering new therapeutic targets for this poorly immunogenic cancer. Citation Format: Nourhan Abdelfattah, Parveen Kumar, Caiyi Wang, Jia-Shiun Leu, William W. Flynn, Ruli Gao, David S. Baskin, Kumar Pichumani, Omkar B. Ijare, Stephanie Wood, Suzanne Powell, David Haviland, Frederick F. Lang, Sujit Prabhu, Kristin Huntoon, Brittany C. Parker Kerrigan, Wen Jiang Jiang, Betty Y. Kim, Joshy George, Kyuson Yun. A multi-dimensional analysis of human gliomas at the single cell level identifies immune suppressive macrophage molecular signatures and a novel immunotherapy target for GBM [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2540.