PROJECT SUMMARY Glioblastoma (GBM) is uniformly lethal and is the most common malignant primary brain tumor. Immunotherapy promises a precise approach, and the hope of durability. One way to deliver precision immunotherapy is with genetically-engineered T cells designed to express a target-specific chimeric antigen receptor, or CAR. In 2017, CAR T cells targeting CD19 were approved by the FDA for B cell malignancies, and several CARs targeting GBM have been described recently. We have developed CARs that target the EGFRvIII tumor mutation, and demonstrated their activity in preclinical studies and in a first-in-human clinical trial. We found that peripheral infusion of CART-EGFRvIII cells was safe and led to elimination of EGFRvIII-expressing glioma cells in patients. However, despite CART-EGFRvIII trafficking to intracranial tumors and targeting of EGFRvIII, the patients ultimately had outgrowth of EGFRvIII-negative disease and tumor progression. Thus, while the CAR T cell platform certainly holds great promise, a critical barrier to clinical impact for brain tumors is targeting a single antigen in an inherently heterogeneous disease. In addition, our study also demonstrated an adaptive increase in immunosuppression within the tumor microenvironment; specifically, the endogenous T cell infiltrate increased in the tumor, but consisted largely of immune-suppressive regulatory T cells (TRegs), rather than effector T cells reflective of antitumor epitope spreading. To simultaneously address antigenic heterogeneity and promote local antitumor activity in GBM, we have now modified CART-EGFRvIII to secrete bispecific antibodies known as bispecific T cell engagers (BiTEs) against wild-type EGFR, which is not expressed in the normal brain but is nearly always expressed in GBM. Delivering BiTES to the brain using T cells as carriers is also attractive because antibodies do not effectively cross the blood-brain barrier. The overall goal of this work is to develop a safe and effective immunotherapy for patients with GBM. We test the hypothesis that anti- EGFRvIII CAR T cells designed to secrete anti-EGFR BiTEs (CAR-BiTE) will lead to potent and durable responses in models of heterogeneous GBM. We will test their mechanism of action by quantifying secretion of BiTEs and systematically testing the role of bystander T cells (Aim 1). Next, we will determine the optimal route of administration of CAR-BiTE products, and the pharmacokinetics and biostribution of these two-in-one active "drugs” (Aim 2). These data are expected to lead to an Investigational New Drug (IND) application and Phase I clinical trial of CART-vIII/BiTE-EGFR in patients with recurrent glioblastoma. Finally, CAR-BiTEs represent a platform that can target multiple combinations of antigens. In Aim 3 we will identify targetable antigens in primary glioma samples and test CAR-BiTEs targeting three or more antigens.