Reinforcement learning-guided control strategies for CAR T-cell activation and expansion.

Reinforcement learning (RL), a subset of machine learning (ML), could optimize and control biomanufacturing processes, such as improved production of therapeutic cells. Here, the process of CAR T-cell activation by antigen-presenting beads and their subsequent expansion is formulated in silico. The simulation is used as an environment to train RL-agents to dynamically control the number of beads in culture to maximize the population of robust effector cells at the end of the culture. We make periodic decisions of incremental bead addition or complete removal. The simulation is designed to operate in OpenAI Gym, enabling testing of different environments, cell types, RL-agent algorithms, and state inputs to the RL-agent. RL-agent training is demonstrated with three different algorithms (PPO, A2C, and DQN), each sampling three different state input types (tabular, image, mixed); PPO-tabular performs best for this simulation environment. Using this approach, training of the RL-agent on different cell types is demonstrated, resulting in unique control strategies for each type. Sensitivity to input-noise (sensor performance), number of control step interventions, and advantages of pre-trained RL-agents are also evaluated. Therefore, we present an RL framework to maximize the population of robust effector cells in CAR T-cell therapy production.

Inhibiting the cGAS-STING Pathway in Ulcerative Colitis with Programmable Micelles.

Ulcerative colitis is a chronic condition in which a dysregulated immune response contributes to the acute intestinal inflammation of the colon. Current clinical therapies often exhibit limited efficacy and undesirable side effects. Here, programmable nanomicelles were designed for colitis treatment and loaded with RU.521, an inhibitor of the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway. STING-inhibiting micelles (SIMs) comprise hyaluronic acid-stearic acid conjugates and include a reactive oxygen species (ROS)-responsive thioketal linker. SIMs were designed to selectively accumulate at the site of inflammation and trigger drug release in the presence of ROS. Our in vitro studies in macrophages and in vivo studies in a murine model of colitis demonstrated that SIMs leverage HA-CD44 binding to target sites of inflammation. Oral delivery of SIMs to mice in both preventive and delayed therapeutic models ameliorated colitis's severity by reducing STING expression, suppressing the secretion of proinflammatory cytokines, enabling bodyweight recovery, protecting mice from colon shortening, and restoring colonic epithelium. In vivo end points combined with metabolomics identified key metabolites with a therapeutic role in reducing intestinal and mucosal inflammation. Our findings highlight the significance of programmable delivery platforms that downregulate inflammatory pathways at the intestinal mucosa for managing inflammatory bowel diseases.