Immunotherapy has demonstrated increasing success in the treatment of cancer, and emerging clinical studies demonstrate that this new approach will provide a fourth pillar for treatment of childhood cancer. Several new classes and types of immunotherapeutics are emerging that open up the possibility of applying immunotherapy to an ever-increasing array of targets and cancers. This project leverages the explosion of technologies and insights emerging in the immunotherapy domain to bring new immunotherapy agents to bear on childhood cancer. We seek to answer fundamental questions about similarities and differences between childhood vs adult cancers as it relates to immune based therapies and to extend our understanding of specific factors that limit the effectiveness of immune based therapies for solid tumors compared to liquid tumors. One major accomplishment in FY14 was creation of an entirely novel chimeric antigen receptor targeting the CD22 antigen, which is expressed on B cell leukemia, the single most common cancers of childhood. State-of-the-art technology previously employed to generate chimeric antigen receptors targeting other antigens (e.g. CD19 for leukemia) was used to generate an entire series of new receptors targeting CD22. The CD22 targeted chimeric antigen receptor that showed similar potency to a CD19 targeted receptor, and thus is predicted to have activity in clinical trials. This work was published in Blood in 2013 (Haso et al), a patent was filed, and the new receptor has been licensed by a commercial entity and will be tested in children and young adults in an early phase trial in 2014. A second major accomplishment in FY14 was publication of a report that defined myeloid derived suppressor cells as the basis for immune escape in murine rhabdomyosarcoma following treatment with the immune checkpoint inhibitor anti-PD1. Anti-PD1 is part of a new class of immunotherapeutics that circumvent immunosuppressive mechanisms used by cancer and allow natural, endogenous antitumor immune responses to emerge. Response rate of approximately 10-30% have been reported in malignant melanoma, lung cancer and renal cell carcinoma. This work used anti-PD1 in a model of a pediatric solid tumor, namely rhabdomyosarcoma. Briefly, treatment of mice with anti-PD1 prior to tumor inoculation completely prevented development of cancer, but delaying anti-PD1 treatment for even 7 days after tumor inoculation resulted in minimal benefit. We identified that rhabdomyosarcoma-induced expansion of myeloid derived suppressor cells that traffic to tumors via a CXCL1/CXCR1 axis. We therefore hypothesized that interruption of trafficking of myeloid derived suppressor cells to rhabdomyosarcomas could enhance the efficacy of anti-PD1 therapy. To do this, we created chimeras lacking CXCR2 on hematopoietic cells and we also treated animals with anti-CXCR2 blocking antibodies. In both cases, we observed regression of established rhabdomyosarcoma with anti-PD1 therapy. This work establishes MDSC as important mediators of immune escape in pediatric solid tumors and identified distinctions between the cells induced in adult tumors vs those induced in tumors of childhood. We also sought to determine whether the CXCL/CXCR axis was likely to play a role in human pediatric sarcomas. We first demonstrated that human sarcoma cell lines produce several CXCR ligands and furthermore, discovered that serum from patients treated on trials of immunotherapy had elevated levels of CXCL1 and CXCL8. Moreover, among patients with metastatic sarcoma, elevated CXCL8 levels were associated with poor prognosis. We therefore believe that the CXCL/CXCR chemokine axis represents a potential basis for developing therapies that could be used to augment the efficacy of immunotherapy for childhood cancer. This work was published as the cover article in Science Translational Medicine in 2014 and was also highlighted in Nature Reviews Cancer. A third accomplishment inFY13 is near completion of a study demonstrating that T cells genetically engineered to recognize a self antigen expressed in the thymus develop T cell leukemia through a process whereby the T cell receptor itself serves as an oncogene. We developed a T cell transgenic mouse in which the vast majority of T cells have specificity for "survivin" an anti-apoptotic molecule considered to be a universal tumor antigen. We hypothesized that survivin TCR transgenic mice would be immune to tumor growth as most tumors express high levels of survivin. Unfortunately and remarkably, these mice did not resist growth of implanted tumors, but rather T cell lymphoblastic leukemia developed in essentially all mice. This was not due to insertional mutagenesis. Rather, self-reactivity with survivin antigens expressed in the thymus led to expansion of early thymic progenitors via signaling of the transgenic TCR in the thymus in response to thymic expression of the survivin gene. TCR signaling at this early stage of T cell development leads to expansion of progenitors, followed by acquisition of NOTCH mutations in multiple clones. The role for TCR signaling in this process is confirmed by the fact that mice which cannot present peptides derived from the survivin antigen within the thymus (e.g. b2m-/- mice) have a substantially diminished incidence of leukemia. These insights provide novel clues to potential dangers of genetic engineering that involves targeting developing thymocytes toward self-antigens. This work is being prepared for publication. A fourth accomplishment was substantial progress in developing chimeric antigen receptors targeting solid tumors of childhood. We generated chimeric antigen receptors targeting a variety of antigens expressed on pediatric solid tumors, including GD2, B7H3, ALK and CSPG4. These chimeric antigen receptors (also known as CARs) are genetically engineered into a retroviral vector and expressed in human T cells. Through this work, we have conducted extensive comparison studies of a very potent CAR targeting CD19 vs the less potent CAR targeting GD2 to identify those factors limiting the efficacy of the GD2 CAR. Our early hypothesis was that the less effective GD2-CAR CAR was limited by lesser levels of antigen engagement and lesser T cell activation, but we have now made the remarkable observation that GD2-CAR expressing T cells actually are actually "superactivated" compared to their CD19-specific counterparts. These GD2-CAR expressing T cells show hallmark features of exhaustion early during their development and postulate that this exhaustion limits their effectiveness in vivo. Ongoing work is underway to specifically delineate the signaling pathways required for exhaustion in this model system, to identify the role that altered metabolism plays in the development of exhaustion and to combine GD2-targeted CAR cells with other agents that may optimize their efficacy in vivo. We are currently attempting to optimize approaches for solid tumor immunotherapy using chimeric antigen receptors by incorporating regimen that modulate myeloid derived suppressor cells and optimize CAR signaling capacity.