Engineering a juxtamembrane-targeting CAR T-cell against mesothelin: a novel binder resilient to shed antigen for enhanced efficacy against ovarian and pancreatic cancer.

Mesothelin is an attractive target for CAR T therapy on a number of cancer types; however, the efficacy of this therapy is diminished because the bulk of the cell surface-expressed mesothelin is shed through naturally occurring proteolysis leaving behind a short juxtamembrane peptide "stump". The two problems this creates are one, the bulk of the target protein is no longer on the tumor cell and two, soluble, shed mesothelin persists in the tumor microenvironment and circulates in blood and other body fluids, where it can bind mesothelin-targeted CAR T cells and act as a decoy that reduces engagement with tumor cell-surface mesothelin. These issues have contributed at least in part to the lack of desired efficacy in human clinical trials utilizing CAR T cells that target membrane distal regions of mesothelin (i.e., the shed domain) such as those utilizing the variable domains of anti-mesothelin monoclonal antibodies SS1 and M5. In addition, there have been safety concerns regarding the targeting of mesothelin on normal tissues. Here we describe CAR T cells that utilize novel phage display-derived antibodies specific for the mesothelin stump domain, thus being unaffected by the natural process of mesothelin shedding. Mesothelin "stump-specific" CAR T cells (CAR 422) had cytotoxicity and in vivo activity that were comparable to previously studied anti-mesothelin CAR T cells. Importantly, CAR 422 T cells were effective against tumor cells that were resistant to conventional anti-mesothelin CAR T cells and showed reduced on-target/off tumor toxicity in a human mesothelin knock-in mouse model. Thus, CAR 422 holds potential as a next-generation therapy for challenging solid tumors.