A Novel Switchable CAR-T Platform for Acute Lymphoblastic Leukemia Treatment

Treatment of Acute Lymphoblastic Leukemia (ALL) with Chimeric Antigen Receptor (CAR) -T cells has shown tremendous results in clinical trials achieving complete responses in 90% of heavily pretreated patients. However serious side effects, most commonly Cytokine Release Syndrome (CRS), have also been encountered. Mechanisms to control CAR-T cell activity and adverse events by steroidal immune-suppression or specific CAR-T cell depletion may mitigate safety concerns, however antigen loss relapse remains a major source of resistance. Current CAR-T constructs in the clinic are fixed in their ability to target a single antigen, limiting the anti-tumor efficacy once loss or mutation of the targeted antigen occurs.

We have previously shown that T cells can be redirected towards specific antigens via transduction with a "switchable" CAR (sCAR) construct that recognizes a short peptide engrafted into a Fab antibody fragment which recognizes the tumor (the switch) when dosed separately. In preclinical models this platform has demonstrated potent in vivo activity and tumor regression targeting multiple antigens and exhibits exquisite control of the CAR-T cell activity by switch titration. This may control the severity of side effects in the clinic.

Herein, we have taken advantage of synthetic E/K coiled-coil peptidic structures to create a new species of switchable CAR that do not rely on the conventional scFv structure. The coiled-coil system is based on the specific interaction of two peptidic alpha helices, the sequence which is characterized by a heptad repeat and a continuous hydrophobic core that stabilizes the helix and generates a left handed supercoil. By linking one peptide-coil to the end of the extracellular part of the CAR and the other one to a Fab antibody fragment we have created a unique and versatile platform for switchable CAR design. In silico analysis of each coil-peptide linked to the Fab antibody fragment or to the N-terminus of the CAR showed a low probability of causing an immunogenic response in humans. This novel coiled-coil switchable CAR-T cell afforded highly specific killing of CD19 positive leukemia cells in vitro in a dose titratable manner with no non-specific killing of CD19 negative cell lines. Optimization of the overall distance and orientation of the immunological synapse and the switch valency resulted in pico Molar EC₅₀ range with robust multi-cytokine release in in vitro cytotoxicityassays. Moreover, in an aggressive xenograft model of human B-cell leukemia the switchable CAR-T efficiently controlled tumor growth in the presence of switch.

The coiled-coil switchable CAR (ccsCAR) platform shows great potential for tuning switch-CAR affinity and specificity. The design of coiled-coil pairs with variable iterations of helices and/or peptide sequences determine their specificity and affinity for each other. This may increase the effector functions of the ccsCAR while generating new specific coiled-coil pairs. A shorter coiled-coil would decrease the genetic content of the CAR construct and increase CAR expression on the T cell surface while decreasing the potential for immunogenicity.

In summary, we show here that the coiled-coil platform can be used to design fully controllable CAR-T cells without the use of a scFv. This is expected to allow the targeting of multiple antigens (ie. CD19, CD20, and CD22 can be simultaneously or sequentially targeted) with the same CAR-T cell construct to potentially decrease tumor immune escape and prevent relapse prevalence while increasing the safety of CAR-T cell based therapies. Finally, this platform could be extended to other malignancies and applied under a personalized medicine criteria by characterizing the tumor phenotype prior to treatment initiation.