Humanized mouse models are misleading to evaluate CAR-Treg function
Michel Tenspolde1, Fatih Noyan1, Tobias Riet1, Michael Hust1, Katharina Zimmermann1, Matthias Hardtke-Wolenski1, Elmar Jaeckel1,2.
1Dept. Gastroenterology, Hepatology & Endocrinology, Hannover Medical School, Hannover, Germany; 2Toronto Liver Transplant Program, Ajmera Transplant Center, University of Toronto, Toronto, ON, Canada
Introduction: Chimeric antigen receptors CARs can be used to change the specificity of FOXP3+ regulatory T cells (Tregs) for the induction of tissue specific tolerance. We and others have shown that CAR-Tregs specific for transplant mismatched HLA-A*0201 (A2-CAR) can be used for tolerance induction in humanized mouse models.
Methods: We now wanted to study the therapeutic potential of MHC-specific CAR Tregs in an immunogenic, non-lymphopenic mouse model. To this end, we generated H-2Kb-specific scFv from phage display libraries to create second generation CARs with a CD28/CD3z signaling domain.
Results: The CARs showed good surface expression and strong activation of Tregs with an exclusive Kb-specificity. They were much more effective in suppression of CD4+ and CD8+ effector T cell response in allogeneic mixed lymphocyte reaction as compared to nTregs. Kb-CAR-Tregs prevented skin transplant rejection in BALB/c RAG1-/-after cotransfer with equal numbers of effector T cells. To our major surprise adoptive transfer of CAR-Tregs in an immunogenic C57BL/6 to BALB/c skin transplant did not result in any meaningful prolongation of transplant survival in a non-lymphopenic setting. We therefore cloned our anti-MHC*0201 scFv from the initial experiments in our humanized mouse models into a murine CAR vector. However, the rejection of A2-transgenic C57BL/6 skin grafts into C57BL/6 recipients (single MHC mismatch) was also not prolonged by adoptive transfer of A2-CAR Tregs. Although lymphodepletion and rapamycin prolonged graft survival, the addition of Kb-CAR Tregs had no added benefit, although we observed long-term persistence of CAR-Tregs and a strong accumulation of the transferred Tregs within the graft. In fact, the transferred CAR-Tregs made up to 40% of all intragraft Tregs ruling out too small local Treg numbers as the reason for the low in vivo efficiency.
Conclusion: Weakly immunogenic humanized mouse models can be misleading to evaluate the effectivity of CAR-Tregs for clinical translation. Lymphodepleting strategies strongly increase the accumulation of graft-specific Tregs within the grafts. In order to develop clinically meaningful protocols for CAR-Treg therapies more immunogenic, non-lymphopenic models are required.
