Signaling Domain of Chimeric Antigen Receptors Can Reprogram T Cells

BACKGROUND Chimeric antigen receptors (CARs) redirect T cells to recognize tumor cells, providing a powerful new approach to cancer immunotherapy. However, the attributes of CARs that ensure optimal in vivo tumor destruction and persistence of the CAR remain to be defined. Here, we analyze the influence of the signaling domain on the proliferation and function of CAR T cells. We find that distinct costimulatory domains in the CAR architecture have major effects on cell longevity, memory differentiation, and cell metabolism characteristics. METHODS We developed a novel system of in vitro T cell stimulation to study the consequences of a single round of CAR-specific stimulation in order to analyze signaling through various CAR signaling endodomains. By electroporation of in vitro transcribed mRNA encoding CAR into primary resting human T cells, we achieved >90% CAR- positive T cell population. We expressed anti-mesothelin SS1-CAR constructs with varying intracellular signaling domains (CD3zeta, CD28:z, and 4-1BB:z), all specific for a widely expressed tumor-associated antigen, mesothelin. Upon verifying CAR expression, these T cells were stimulated with recombinant mesothelin immobilized on beads and then cultured. With only a transient expression of CARs, the CAR disappears from the surface after one round of stimulation, allowing the unambiguous analysis of signaling through the CAR alone. This approach thus obviated the requirement for stimulation through the endogenous T cell receptor and permitted for the first time, an analysis of the consequences of signaling through the surrogate antigen of CAR T cells. Experiments were conducted on primary peripheral blood T cells, sorted naïve T cells and cord blood T cells. RESULTS To examine whether intracellular signaling domains influence the in vitro function and antitumor activity of CAR-T cells, we developed a new culture system that permits rapid screening of novel CAR designs. The various CAR constructs demonstrated comparable and specific cytolytic capabilities when cultured with target cells that expressed mesothelin. However, primary T cells stimulated through the 4-1BBz-containing CARs showed superior survival and expansion profiles when compared to CD28-based CARs. Phenotypic analysis of 4-1BB-based CAR T cells revealed that an increased population of cells with central-memory surface markers was generated. On the other hand, cells endowed with CD28z-containing CARs yielded a significantly higher proportion of effector memory cells, with a modest increase in expression of inhibitory PD-1, TIM3 and LAG3 molecules relative to their 4-1BBz counterparts. These results remained consistent whether starting populations of bulk peripheral blood T cells, naïve (CD45RO-CD95- CD62L+CCR7+ sorted) peripheral blood T cells, or cord blood T cells were used. Metabolic profiling of the mesothelin-stimulated cells CAR T cells by the Sea-Horse assay in culture revealed a substantial increase in lipid oxidation in 4-1BBz-CAR stimulated cells compared to their 28z counterpart. Additionally, microarray studies have revealed a unique gene signature in cells that are recovered after stimulation through the different CAR signaling domains. CONCLUSIONS We report the development of a novel system to study signal transduction in primary T cells after stimulation through their surrogate antigen. Following the initial stimulation of CAR T cells by the surrogate antigen mesothelin, both 4-1BB and CD28-based CARs had a burst of more than 5 population doublings. However, the 4-1BB CAR T cells proliferated for longer in vitro. In addition to quantitative differences in proliferation, there were qualitative differences that emerged as a function of the signaling endodomain. 4-1BB-based CAR T cells displayed features of central memory T cells, while CD28-based CARs T cells rapidly differentiated to an effector memory pool of CAR T cells. Other differences between 4-1BB and CD28-based CARs were uncovered, including differential expression of inhibitory checkpoint molecules, metabolic reprogramming, and gene signatures. Together these results may explain the differential survival reported with CD28 and 4-1BB based CAR T cells in clinical trials, and inform the development of CARs with novel signaling domains. These results also provide a new system to rapidly evaluate new CAR designs. Disclosures Scholler: Novartis: Research Funding. Milone:Novartis: Research Funding. June:Novartis: IP licensed by University of Pennsylvania to Novartis. Author entitled to royalties from the University. Patents & Royalties, Research Funding.