T Cells Expressing Engager and Costimulatory Molecules for the Immunotherapy of CD19+ Malignancies

Background: Immunotherapy with anti-CD19/anti-CD3 bispecific engager molecules has shown promise in clinical studies for CD19+ malignancies. However engager molecules have short half-lives and do not accumulate at tumor sites. In addition, co-delivery of other immunostimulatory molecules to enhance antitumor effects is difficult to achieve. We have recently shown that T cells can be genetically modified to secrete bispecific engager molecules (ENG-T cells). ENG-T cells are activated by tumor cells in an antigen-dependent manner, redirect bystander T cells to tumor cells, and have antitumor activity in preclinical models. We now wanted to explore if additional genetic modifications of ENG-T cells can enhance their effector function in vitro and in vivo. Since bispecific engager molecules do not provide co-stimulation, we focused on the provision of co-stimulatory signals by coexpressing CD80 and CD137L on the cell surface of ENG-T cells. Thus, the aim of the study was to compare the effector function of CD19-specific T-cell engagers (CD19-ENG T cells) and CD19-ENG T cells co-expressing CD80 and 41BBL (CD19-ENG/Costim T cells).

Methods: CD19-ENG T cells were generated by transducing T cells with a retroviral vector encoding a CD19-specific T-cell engager and mOrange separated by an IRES (SFG.CD19-ENG-I-mO), and CD19-ENG/Costim T cells were generated by double transducing T cells with SFG.CD19-ENG-I-mO and a 2^nd retroviral vector encoding 41BBL and CD80 separated by an IRES. The effector function of ENG T-cells was evaluated in vitro and in a leukemia xenograft model.

Results: After single or double transduction 60-80% of T cells were positive for mOrange, and ~80% of CD19-ENG/Costim T cells were positive for CD80 and 30-40% positive for 41BBL. In coculture assays CD19-ENG and CD19-ENG/Costim T cells recognized CD19+ lymphoma (Daudi, Raji) and acute leukemia (BV173) cells as judged by IFN-g secretion in contrast to negative controls. While CD19+ target cells that express CD80 and CD86 (Daudi and Raji) induced robust IL2 production of CD19-ENG and CD19-ENG/Costim T cells, CD19-ENG/Costim T cells produced significantly higher levels of IL2 in comparison to CD19-ENG T cells after stimulation with CD19+/CD80-/CD86- negative target cells (BV173). Cytokine production was antigen dependent since ENG and ENG/Costim T cells specific for an irrelevant antigen (EphA2) did not produce cytokines. Specificity was confirmed in cytotoxicity assays. In transwell assays containing inserts preventing T-cell migration, only ENG T cells redirected bystander T cells in the bottom well to CD19+ tumor cells. To assess in vivo anti-tumor activity of CD19-ENG T cells and CD19-ENG/Costim T cells we used the BV173/NSG mouse xenograft model in which BV173 cells are genetically modified with firefly luciferase (ffLuc-BV173) to allow for serial bioluminescence imaging. While therapy with CD19-ENG T cells on day 7 post ffLuc-BV173 injection resulted in the cure of all mice, when therapy was delayed to day 14, only 1/10 mice was alive on day 80. In contrast therapy of mice on day 14 with CD19-ENG/Costim T cells resulted in long-term survival of 7/10 mice. Control T cells (EphA2-ENG T cells or EphA2-ENG/Costim T cells) had no antitumor effects.

Conclusions: We have generated CD19-ENG T cells and CD19-ENG/Costim T cells with the ability to direct bystander T cells to CD19+ malignancies. Both ENG T-cell populations had potent antitumor activity in a preclinical ALL model, and provision of costimulation further enhanced antitumor effects. Genetically modifying T cells to express engager molecules and additional molecules to enhance their effector function may present a promising alternative to current CD19-targeted immunotherapies.