Development of a humanized mouse model of graft-versus-host disease to assess human regulatory T cell function

Regulatory T cells (Treg)-based therapies are increasingly used for treating autoimmune or graft-versus-host (GVHD) disease. Given their low frequency, several approaches aimed at amplifying them or at increasing their immunosuppressive activities are currently investigated. Unfortunately, the impact of these strategies on human Treg function has remained difficult to assess in vivo. Here, we report the development of a novel humanized mouse model of allogeneic and xenogeneic GVHD intended to characterize the immunosuppressive activity of human Treg in vivo.

In this model, GVHD is induced by injecting CD25-depleted HLA-A2^- peripheral blood mononuclear cells (PBMC) into irradiated NSG-HLA-A2-HHD mice. The CD25⁺ (Treg) fraction is maintained in vitro for 48h during which Treg-promoting treatments can be tested before infusion into mice. We took advantage of this model to investigate whether tumor necrosis factor alpha (TNF-α) priming of Treg would increase their suppressive function and improve their ability at preventing xenogeneic GVHD, after assessing the effect of TNF-α on Treg in vitro.

In vitro, single cell RNAseq analyses showed that eleven Hallmark pathways, including Interferon response, IL-6-JAK-STAT3 signaling, mTORC1 signaling, and TNF-α signaling via NF-κB, were significantly upregulated in Treg following TNF-α priming. In vivo, Treg infusion resulted in higher Treg levels, lower counts of human cells, lower conventional CD4⁺ (Tconv) and CD8⁺ T-cell counts, lower KI67 and HLA-DR expression by Tconv and lower Granzyme B expression by CD8⁺ T-cells in peripheral blood. No significant impact of TNF-α priming of Treg on survival was observed.

These results emphasize the importance of reliable techniques to assess Treg in vivo as efficient methods to activate them in vitro do not always result in an enhanced function in the in vivo setting. In summary, we present here the development of a novel humanized model of GVHD designed to evaluate the in vivo functionality of human Treg. Taking advantage of that model, we observed that TNF-α priming of human Treg did not increase their suppressive activity in vivo.