Abstract The COVID vaccination campaigns have highlighted the promise of mRNA-mediated delivery as a novel therapeutic modality. One particular application is adoptive cell therapy, where immune cells are equipped with novel functions to treat diseases, e.g., expressing chimeric antigen receptors (CARs) in T cells to ablate cancer and other undesirable cells. Compared to DNA-based engineering of immune cells, mRNA has several advantages as a delivery vector, especially its superior safety profile, because it eliminates the risk of randomly inserting into the host genome and causing mutations, and its short half-life mitigates long-term adverse effects due to the persistence of the engineered cellular function. Instead of extracting cells from the patient, engineering them ex vivo, and then reinducing them, researchers have even directly delivered CAR-encoding mRNAs and created functional CAR T cells in vivo. This is appealing because it has the potential to make adoptive cell therapy accessible to the general public, its logistics almost as straightforward as manufacturing, distributing, and administering vaccine shots, in contrast to the costly ex vivo engineering process that will be limited to the privileged few. Despite the great potential of mRNA-mediated adoptive cell therapy, there remains a critical need for tools that enhance targeting precision. It is very rare for a single surface marker, targeted by CAR, to unambiguously identify one target cell population. Therefore, “on-target/off-caner” killing is a major concern for CAR T cell therapies against cancer, and the same concern also applies to scenarios of ablating other cells, such as active fibroblasts in heart infarction or senescent cells. One elegant solution is synthetic receptors, most notably synthetic Notch, that detect a second marker and express CAR in response, effectively forming AND logic, where target cells are only killed when both inputs to the synthetic receptor and CAR are present. However, to our knowledge, all existing synthetic modular, programmable receptors operate at the DNA level, and are therefore incompatible with mRNA-mediated delivery. Here we design and demonstrate the feasibility of a first-in-class synthetic modular receptor that operate at the RNA level. It converts ligand-induced dimerization events into the expression of arbitrary output proteins. Through extensive computational simulation and experimental optimization, we will expand the input/output repertoire of this receptor, establish its design principle, enable its encoding on single transcripts and delivery by mRNA, and take first steps towards improving the precision of ablating active fibroblasts to treat heart infarction. The impact of such novel receptors is beyond adoptive cell therapy. For example, they can facilitate basic research by recording cells’ (e.g., neurons) exposure to specific signals (e.g., dopamine). They will benefit a variety of other biomedical applications too, from expressing antigens or cytokines in response to extracellular cues to enhance vaccine efficacy, to generating novel sense-and-respond capabilities for tissue engineering.

Relevance to Public Health This project aims to create receptors that can detect a wide variety of inputs that represent cell states and external cues, and produce arbitrary proteins in response. Such receptors would enhance the precision of therapies where engineered cells are used to seek and destroy disease- causing target cells. In contrast to prior art, our receptors operate at the RNA level, and is therefore uniquely suitable for leveraging all the efficiency and safety benefits of RNA-mediated delivery.