Background Chimeric antigen receptor (CAR)-NK and -T cells have shown promise in treating hematological malignancies but struggle to address solid tumors. Issues include poor performance in the immunosuppressive tumor microenvironment (TME) and on-target/off-tumor toxicity due to the expression of tumor-associated antigens (TAA) on tissues of vital organs. Various strategies have been proposed to address these challenges, ranging from ‘arming’ CAR immune cells with constitutive expression of cytokines to combat TME-mediated suppression to multi-receptor logic gated gene circuits (such as NOT, AND, and OR gates) to precisely distinguish normal cells from cancerous ones. These strategies and many others involve constitutive expression of multiple genes (receptors, cytokines, etc.), which are often strung together in a multicistronic payload under control of a single promoter. As a result, typical designs are highly dependent on promoter strength to achieve high expression—and therefore function—of all components.

Methods We created a technology platform for engineering promoters that drive multi-component cell therapy payloads in immune cells. Our multistage approach nominates promoter sequences, then further engineers them into next-generation promoters through multiple design-build-test-learn cycles. At each stage we leveraged multiple high-throughput screening technologies, including pooled screens of large libraries of over 10,000 putative promoters, and arrayed screening of sublibraries of 100s of promoters. The arrayed process utilized a purpose-built automatic liquid handling system to perform transfection, viral production, primary cell transduction and culture, and expression quantification in a continuous, end-to-end process. Library diversity was generated by complementing unbiased combinatorial rearrangements with computer-aided design driven by statistical and machine learning models trained on large promoter datasets.

Results Using this platform, we discovered promoters capable of driving expression of 3- or 4-component multicistronic payloads, each consisting of multiple CARs and cytokines, in primary NK cells at levels significantly higher than commonly used promoters in cell therapy, such as SV40, SFFV, or EF-1a, while also being more compact. These promoters remained active across multiple cell states and did not cause toxicity or phenotypic defects. Significantly enhanced expression resulted in higher performance in terms of all aspects of the multi-component payload design, including CAR-mediated cytotoxicity, NOT gate-mediated healthy cell protection, and secretion of armoring cytokines.

Conclusions Our discovery platform is capable of creating promoters that enhance the expression and function of gene circuits for next generation cell therapies, including CAR-NK and CAR-T, for new treatments of solid tumors.