Deconvoluting CAR T cell heterogeneity to engineer durable antitumor protection

ABSTRACT Chimeric antigen receptor (CAR) T cells are remarkably effective against advanced hematologic tumors, mediating complete remission in >50% of patients. Yet they have limited success against solid tumors due to the immunosuppressive tumor microenvironment (TME). To improve efficacy, CAR T cells can be “armored” with growth factors, such as IL-2 and IL-15, to promote expansion and survival within the TME. Alternatively, “arming” CAR T cells with inflammatory cytokines amplifies antitumor responses by activating host T cells. We recently armored and armed CAR T cells with an IL-2 superkine (Super2) and tissue alarmin IL-33 (hereafter, Super2+33). These cytokines act synergistically and, with multiple CAR constructs, to alter the TME to promote regression of primary and metastatic solid tumors. However, the ability of CAR T cells to provide durable antitumor immunity remains an open question. Analysis of two long-term survivors who received CD19-targeting CAR T cells shows decade-long persistence of CAR T cells, leukemia regression, and loss of normal B cells. Clinical responses are also positively associated with transcriptional features of stem cell memory T (TSCM) cells in the manufactured CAR T-cell product. These data have spurred the field to optimize in-vitro manufacturing conditions to increase memory T-cell potential within the CAR T-cell product. However, the in-vivo fate of individual CAR T cells within a heterogenous CAR T cell population, such as each cell's ability to proliferate, infiltrate tumors, kill, and differentiate into memory T cells, remains unknown. Moreover, it is unclear how CAR engineering and host environmental factors impact the heterogeneity of CAR T cell effector and memory responses. We will test the overall hypothesis that CAR T-cell heterogeneity can be deconvolved to understand how individual T cell subsets behave in vivo and that principles gained through these analyses can help us rationally engineer CAR T-cell populations that balance antitumor effector responses with generation of durably protective antitumor CAR and host circulating (TCIRC) and widely distributed tissue resident (TRM) memory T cells. This hypothesis will be tested in three aims. Aim 1 will use using dynamic lineage recording to deconvolve CAR T cell heterogeneity to reveal individual differentiation trajectories and identify engineering strategies to program CAR stem cell-like (TSCM) and TRM cell fates. Aim 2 will determine the respective requirements for host and CAR T-cell memory. Aim 3 will identify the effect of tumor burden and neoadjuvant immune checkpoint blockade on T-cell precursor subsets and states on CAR T cell differentiation. Successful completion of this study will advance the development of durable adoptive CAR T-cell therapy for melanoma that results in globally distributed memory T cells with optimal fitness and function.