Tuning gated CAR-T cells for enhanced activity against acute myeloid leukemia

Background: Conventional Chimeric Antigen Receptor (CAR) T-cell therapies for Acute Myeloid Leukemia (AML) face persistent challenges, including limited T-cell fitness, resistance emergence, and on-target, off-tumor toxicity. To overcome these limitations, we previously developed logically gated CAR-T cells using Synthetic Notch (SynNotch) circuits targeting two AML antigens, CD33 and CD123. In this system, CD33 engagement triggers expression of a gated CD123 CAR, leading to greater precision, improved metabolic fitness, reduced exhaustion, and longer in vivo persistence. Despite these benefits, gated CAR-T cells exhibit modestly reduced potency compared with conventional CAR-T cells. We hypothesize that incorporating modified intensity signaling elements into the gated architecture will enhance cytotoxicity while preserving the favorable metabolic and immunological profiles of gated CAR-T cells. Methods: To enhance downstream activation, we engineered a panel of CAR signaling constructs, including CD3ζ ITAM-modified and ZAP70-based signaling modules. These designs were assessed using an NFAT reporter assay, allowing quantification of inducible signaling strength across constructs. Results: Gated CAR-T cells expressing these enhanced signaling modules demonstrated accelerated and more robust activation, with a clear dose-dependent increase in NFAT reporter activity. Furthermore, we successfully incorporated the optimized signaling elements into a single open-reading-frame CAR design, enabling efficient expression and positioning these constructs for forthcoming translational testing. Conclusion: Augmenting T-cell receptor-proximal signaling within the gated CAR framework increases activation and improves functional potency. We anticipate that these high-intensity gated CAR-T cells will maintain the metabolic advantages characteristic of SynNotch-gated systems. Ongoing studies are evaluating their efficacy, fitness, and safety in relevant AML models.