A Novel Humanized Murine Model to Identify Neoantigen-Specific T Cells in CBFA2T3-GLIS2 Positive Acute Megakaryoblastic Leukemia

CBFA2T3-GLIS2 is an oncogenic fusion gene that acts as the driver mutation in ~18% of pediatric patients with acute megakaryoblastic leukemia (AMKL) that do not have Down syndrome. CBFA2T3-GLIS2 driven leukemia carries the worst prognosis among all pediatric acute myeloid (AML) subtypes, with event free and overall survival probabilities of 8% and 14% respectively. Chemotherapy and stem cell transplantation fail to cure this disease, thus innovative treatment approaches are needed. While monoclonal antibody and chimeric antigen receptor T-cell (CAR-T) therapies have proven successful in acute lymphoblastic leukemia (ALL), similar successes haven't been realized in AML due to the lack of targetable antigens that eradicate leukemia cells while minimizing off target toxicities such as depletion of normal myeloid cells which can lead to prolonged neutropenia. An alternative is T cell receptor engineered T cell (TCR-T) immunotherapy which use heterodimers consisting of alpha and beta peptide chains to recognize polypeptide fragments presented by MHC molecules on the tumor cells. An advantage of this approach is the ability to recognize intracellular tumor specific and tumor associated antigen fragments in addition to extracellular proteins which results in a wider range of targets. To investigate the immunogenicity of CBFA2T3-GLIS2 positive AMKL cells, we established a humanized patient derived xenograft (PDX) murine model by conditioning 15 immunodeficient NSG-SGM3 mice with 200 rads 24 hours prior to transplantation of 3.5 million CBFA2T3-GLIS2 positive PDX cells derived from an individual female patient. After 20 days of engraftment, 2.5 million HLA class I matched PBMCs from an unrelated female donor were introduced. Experimental mice (engrafted with both PDX cells and PBMCs), leukemia only control mice, and PBMC only control mice were sacrificed at days 30, 35, and 40. Flow cytometry analysis at day 40 showed dramatic reduction of leukemic blasts (huCD45lo, huCD34+) when exposed to PBMCs compared to the leukemia only control (1.48% and 93.7%, respectively). Circulating human Granzyme A was elevated at day 40 in the PDX mice with PBMCs compared to the PBMC only control (9665.4 pg/ml vs. 789.7 pg/ml, p<0.001), indicating a potential leukemia-directed cytokine lymphocyte response. Single cell TCR repertoire analysis of CD8+ T cells from day 40 revealed a decrease in alpha beta TCR diversity when exposed to PDX cells compared to the PBMC only control, consistent with a clonal expansion of neoantigen-directed TCRs. Single TCR expressing stable cell lines have been generated for the top expanded clones. In ongoing experiments, we are utilizing these cell lines to assess leukemia clearance in our model and measuring specificity and functionality for the putative neoantigens through tetramer staining and peptide stimulation assays, respectively. For the latter studies, we have taken two approaches for the identification of potential neoantigens that may be directing this observed clonal expansion. First, to interrogate potential neoantigens spanning the fusion junction we used the algorithm from NetMHCcons which identified two peptides with intermediate binding affinity (defined by an IC50 of 150 - 500 nM). Second, we utilized immunopeptidomics to identify HLA class I presented peptides on the surface of the PDX cells. One of our top significant hits was a peptide that corresponds to the 5' untranslated region (UTR) of bone morphogenetic protein 2 (BMP2). This peptide fragment was not present in the human Uniprot database, suggesting the 5' UTR is not expressed in normal tissues. Although its role in leukemia is not well described, BMP2 is aberrantly upregulated in CBFA2T3-GLIS2 positive AMKL and required for serial replating in colony formation unit assays; therefore, this peptide is of great interest as a candidate leukemia-associated antigen. In summary, our novel humanized PDX model sheds light on the immunogenicity of CBFA2T3-GLIS2 mediated AMKL and the potential for TCR-T immunotherapy in the setting of minimal residual disease for this chemotherapy resistant malignancy. Studies are ongoing to identify the TCR clone(s) responsible for the anti-leukemic response seen in our model and the leukemia-associated antigens that are mediating this response.