Automated GMP-compatible production of universal CAR Tregs for organ-targeted tolerance induction

Abstract Background Adoptive transfer of regulatory T cells (Tregs) has demonstrated safety, feasibility and early signs of efficacy in promoting immunological tolerance in inflammatory conditions such as graft-versus-host disease (GvHD). Chimeric antigen receptor (CAR)-engineered Tregs offer localized activation and suppression compared to polyclonal Tregs, but their clinical translation is limited by high manufacturing costs, lengthy developing times and fixed single-antigen specificity. To address these limitations, we employed the universal adapter Reverse CAR (RevCAR) system, which harbors a peptide epitope lacking intrinsic antigen specificity but provides flexibility in targeting through the use of an antigen-specific RevCAR Target Module (RevTM). As a proof-of-concept, we used a RevTM targeting carcinoembryonic antigen (CEA), which is highly expressed in the gastrointestinal (GI) tract, as a potential strategy to achieve localized immunosuppression in GI acute GvHD. Methods To support clinical translation, we established an automated, GMP-compatible, clinical-scale manufacturing process. Tregs were magnetically enriched from leukapheresis using the CliniMACS ® Plus, followed by high-purity sorting on the MACSQuant ® Tyto ® . The sorted cells were virally transduced and the RevCAR Tregs were expanded on the CliniMACS Prodigy ® to obtain clinically relevant cell numbers. The harvested products were evaluated for phenotype, stability, antigen specificity and suppressive function. Results Across five manufacturing runs, Tregs (CD4 + CD25 high CD127 low FOXP3 + ) with a median initial purity of 94% were expanded to achieve a median therapeutic yield of 602 × 10 6 cells. The final product maintained a high purity (median: 91.9%) and exhibited high RevCAR expression (median: 60% RevCAR + ). Mass cytometry analysis revealed that expanded RevCAR Tregs predominantly exhibited a central memory phenotype with high expression of functional and homing markers. Under experimental pro-inflammatory conditions, the cells maintained stable FOXP3 and Helios expression with minimal pro-inflammatory cytokine production. Importantly, RevCAR Tregs showed antigen-specific activation upon target engagement via the CEA-specific RevTM and robust, dose-dependent suppression. Conclusion The study establishes a scalable, GMP-compatible process for manufacturing pure, stable and functional universal RevCAR Tregs for clinical applications. Furthermore, the RevCAR system offers a promising approach toward an allogenic, off-the-shelf Treg therapy capable of treating diverse immune-mediated diseases with spatial precision.