Summary The diversity of molecular states and cellular plasticity of immune cells in the glioblastoma environment is still poorly understood. Here, we performed scRNA sequencing of the immune compartment and mapped potential cellular interactions leading to an immunosuppressive microenvironment and dysfunction of T cells. Through inferring the dynamic adaptation during T cell activation, we identified three different terminal states with unique transcriptional programs. Modeling of driver genes for terminal T cell fate identified IL-10 signaling alterations in a subpopulation of HAVCR2(+) T cells. To explore in depth cellular interactions, we established an in-silico model by the integration of spatial transcriptomic and scRNA-sequencing, and identified a subset of HMOX1 + myeloid cells defined by IL10 release leading to T cell exhaustion. We found a spatial overlap between HMOX(+) myeloid and HAVCR2(+) T cells, suggesting that myeloid-lymphoid interaction causes immunosuppression present in tumor regions with enriched mesenchymal gene expression. Using human neocortical GBM model, coupled with patient-derived T cells, we confirmed that the functional interaction between myeloid and lymphoid cells, leads to a dysfunctional state of T cells. This IL-10 driven T cell exhaustion was found to be rescued by JAK/STAT inhibition. A comprehensive understanding of the cellular states and plasticity of lymphoid cells in GBM will aid towards successful immunotherapeutic approaches.

Lineage and Spatial Mapping of Glioblastoma-associated Immunity