A Novel Pharmacologic 'Remote Control' to Steer CAR-T Cell Function In Vivo

Background:

Immunotherapy with chimeric antigen receptor T cells (CAR-T) is emerging as a transformative novel treatment for hematologic malignancies. In contrast to conventional agents, CAR-T are a living drug that is essentially out of the physician's (and patient`s) control after infusion. At present, there is a lack of technologies to maintain control over CAR-T, both for steering efficacy and for preventing toxicity. Here, we present a novel strategy of pharmacologic 'remote control' to precisely steer CAR-T activity and function in real-time.

Methods:

We considered that an effective way for controlling CAR-T function was to interfere with signal transduction through the CAR. We assembled a library of tyrosine kinase inhibitors and screened for their ability to block CAR-T function without affecting CAR-T viability. We performed functional testing with CD8+ and CD4+ CAR-T (n=3 donors) in the presence of titrated doses of the lead compound, and employed CD19- and ROR1-specific CARs comprising 4-1BB or CD28 costimulatory moieties.

Results:

We identified a lead compound, TCI-1, that stood out through its ability to confer a dose-dependent (partial at lower, complete at higher doses) blockade of all CAR-T effector functions, i.e. cytolytic activity, cytokine secretion and proliferation. We confirmed that TCI-1 is effective in both CD8+ and CD4+ CAR-T, and capable of completely blocking CAR-T function independent from CAR specificity and costimulatory moiety included in the signaling module. The onset of CAR-T blockade is immediate after exposure to TCI-1 and is caused by interference with early phosphorylation events in the CAR-CD3zeta module as demonstrated by western blot, and by interference with the induction of transcription factors, as demonstrated with an NFAT-inducible reporter gene. Blockade of CAR-T function is effective for several days if TCI-1 is continuously supplied and instantaneously and fully reversible after removal of the compound. Short- and long-term exposure to TCI-1 does not lead to a reduction in CAR-T viability, and does not hinder the subsequent ability of CAR-T to exert their effector functions. We reasoned that in patients experiencing side effects like cytokine release syndrome (CRS), CAR-T are in a highly activated state and therefore performed comprehensive testing to demonstrate that TCI-1 is able to arrest activated CAR-T that are already in the process of executing their effector functions.

For further proof-of-concept, we employed a xenograft model in immunodeficient mice (NSG/Raji) and demonstrate that TCI-1 is capable of controlling the function of CD19 CAR-T in vivo. In a first scenario, we administered TCI-1 between d3 and d5 after CAR-T infusion and show that TCI-1 is able to arrest activated CAR-T during this time interval and pause the antitumor effect. Importantly, CAR-T immediately resumed their antitumor function once administration of TCI-1 was discontinued (function ON-OFF-ON sequence). In a second scenario, we administered TCI-1 immediately after CAR-T infusion (function OFF-ON sequence). The data show that TCI-1 halts CAR-T activation during the OFF phase and prevents cytokine release, without affecting the subsequent antitumor potency of CAR-T during the ON phase in this lymphoma model.

Conclusions:

Our data introduce TCI-1 as a universally applicable pharmacologic 'remote control' for CAR-T. We show that TCI-1 is capable to exert real-time, ON/OFF control over CAR-T function in vitro and in vivo. TCI-1 has a favorable pharmacokinetic and safety profile in humans, suggesting the potential for rapid clinical implementation. The qualities of TCI-1 with rapid-onset mode of action in addition to complete but fully reversible CAR-T inhibition surpass the qualities of steroids that are toxic to T cells and provide only incomplete control, and complement the existing spectrum of safety technologies with suicide-gene that effectively control CAR-T but also terminate their antitumor effect.