Engineering advanced logic and distributed computing in human CAR immune cells

The human immune system exhibits remarkable capabilities in sensing and responding to various antigens and signals. T cells, for example, possess intricate biocomputation circuitries that integrate signals from different receptors to respond to pathogens or tumors. The immune system also utilizes consortia of specialized immune cells to perform distributed computing, where cells collaboratively address challenges, each type contributing specific inputs and outputs. Furthermore, intercellular communication facilitates temporally coordinated responses. The complex interplay between innate and adaptive immune cells during infection exemplifies this sophistication in distributed processing and communication. This complexity is crucial for immune homeostasis and disease prevention. While therapies like antibody therapy modulate immune cell sensing and interactions, their applicability is limited by their inability to discriminate targets based on multiple antigens or engage various cell types. Engineering complex logic in human immune cells can significantly improve immunotherapy specificity and potential. Chimeric antigen receptors (CARs), typically composed of a single-chain variable fragment fused to signaling domains, have shown high efficacy in redirecting T-cell specificity against cancer cells. However, conventional CAR designs are limited to single-antigen detection, lacking the specificity needed for many applications. Efforts to incorporate logic and control functions into CARs have focused on creating systems active only when two antigens are present (AND gate). These systems typically involve transducing T cells with a CD3ζ CAR targeting one antigen and a chimeric co-stimulatory receptor targeting another. However, these systems lack tunability after engineering and require careful balancing of CAR activities to function effectively. Engineering distributed processing and cell-cell communication could lead to synthetic immune cell consortia, offering improved safety and efficacy in immunotherapy. The split, universal, and programmable (SUPRA) CAR system addresses limitations of conventional CARs by separating the antigen-binding portion (zipFv) from the signal transduction receptor (zipCAR) expressed on T cells. The leucine zipper-mediated interaction between zipFv and zipCAR enables ON/OFF switching, fine-tuned activation, and AND logic computation. Orthogonal SUPRA CARs can independently control different T-cell subsets, highlighting the potential for complex biocomputation at both single-cell and consortium levels.

Several studies have explored the use of chimeric antigen receptors (CARs) to improve the specificity and efficacy of cancer immunotherapy. Early CAR designs primarily focused on redirecting T-cell specificity towards cancer cells expressing a single target antigen. While effective in some cases, the limited specificity of these single-antigen CARs led to off-target effects and limited therapeutic potential. Subsequent research sought to enhance the specificity of CAR T-cell therapy by incorporating logic gates, allowing for more precise control of T-cell activation based on the presence or absence of multiple target antigens. For instance, AND gate CARs have been developed that require the simultaneous presence of two target antigens for T-cell activation. However, these systems often suffer from limited tunability and require precise balancing of signaling domains. Furthermore, relatively few studies have integrated multiple cell types to create a distributed computation system, mimicking the complexity and interconnectivity of the native immune system.

The study utilized the split, universal, and programmable (SUPRA) CAR system, which consists of a soluble antigen-binding portion (zipFv) and a universal signal transduction receptor (zipCAR). Seven distinct innate and adaptive immune cell types (CD8+ T cells, Th1 cells, Th2 cells, Treg cells, γδ T cells, NK cells, and macrophages) were transduced with zipCARs. The functionality of the SUPRA CAR system in these cell types was evaluated by measuring cell killing (for cytotoxic cells) and cytokine production (for other immune cells). To achieve tunable AND logic, orthogonal zipCARs containing CD3ζ or co-stimulatory (CD28 or 4-1BB) domains were introduced. The study also explored the development of a NOT gate using an inhibitory domain (BTLA), creating a three-input (A AND B) AND NOT C logic system. The versatility of the SUPRA CAR system was demonstrated by creating a synthetic immune cell consortium that inducibly controlled macrophage polarization, enabling multicellular distributed computing via Treg-mediated suppression of conventional CD4+ T cells. A direct cell-cell communication channel was created using a zipFv secretion system, implementing an intercellular AND logic circuit. In vivo experiments using a human xenograft tumor model with NK cells expressing a CD3ζ/BTLA NOT gate were conducted to assess the in vivo functionality of the NOT gate. Finally, three-input multilogic in a single T cell was achieved using SUPRA CAR, combining AND and NOT logic.

The SUPRA CAR system successfully redirected antigen specificity in seven distinct innate and adaptive immune cell types. Tunable AND logic was achieved in CD8+ T cells and Tregs, demonstrating synergistic upregulation of target cell killing and activation markers (CD69 and CTLA-4) upon simultaneous activation of CD3ζ and co-stimulatory domains. A BTLA-derived co-inhibitory domain enabled a functional NOT gate in CD4+ and CD8+ T cells and NK cells, significantly inhibiting IFN-γ secretion or target cell killing. In vivo experiments confirmed the NOT gate's function, showing a significant increase in tumor burden when the BTLA domain was activated. A three-input logic gate (A AND B) AND NOT C was successfully implemented in single CD8+ and CD4+ T cells. The SUPRA CAR system enabled the orthogonal and local control of macrophage polarization via Th1/Th2 cell activation. An inducible, antigen-dependent multicellular logic circuit was created, regulating immune activation and suppression through Treg-mediated suppression of conventional CD4+ T cells. Finally, a kill switch system, utilizing a V5 epitope tag and CAR-NK cells, was created for eliminating engineered immune cells, acting as a safety mechanism against potential toxicities.

This study demonstrates the successful engineering of advanced logic and distributed computing capabilities in human immune cells using the SUPRA CAR system. The ability to create tunable AND and NOT gates, culminating in a three-input logic circuit, represents a significant advance in the field of synthetic immunology. This approach offers a powerful tool to address the limitations of traditional CAR T-cell therapies, such as off-target effects and lack of specificity. The implementation of these logic gates in multiple immune cell types, along with the creation of a synthetic intercellular communication channel, allows for the engineering of complex multicellular circuits capable of performing distributed computations. The findings highlight the importance of signal strength tunability in achieving optimal logic performance, with the split design of the SUPRA CAR system offering superior control over CAR activity compared to other methods. The creation of a kill switch provides a valuable safety mechanism to mitigate potential toxicities associated with CAR T-cell therapy. Further development of this technology holds the potential for creating highly specific and safe immunotherapies for a range of diseases, beyond cancer.

This research demonstrates the successful application of the SUPRA CAR system to engineer advanced logic and distributed computing functions in human immune cells. The creation of tunable AND and NOT gates, culminating in a three-input logic gate, along with the development of a kill switch and intercellular communication channels, provides a robust platform for the design of sophisticated synthetic immune consortia. Future studies could explore further applications of this system, such as investigating more complex logic gates, expanding the range of controllable immune cells, and optimizing the design for specific therapeutic applications. This technology holds significant promise for revolutionizing immunotherapy.

The study primarily focused on population-level responses, with the single-cell-level immune response remaining largely unexplored. While the in vivo experiments demonstrated the functionality of the NOT gate, further investigation with larger sample sizes and more diverse tumor models is warranted. The long-term effects of the engineered immune cells and potential for off-target effects require further study. The translation of this technology to clinical settings necessitates careful consideration of manufacturing scalability, cost-effectiveness, and potential toxicity.