Graft-versus-host disease (GVHD) is a major complication of hematopoietic stem cell transplantation. Mesenchymal stem cells (MSCs) can modulate immune response and have been used as a treatment for aGVHD. The immune-modulating factors of MSCs are secreted and reside in supernatant fractions that are enriched for extracellular vesicles (EVs). MSC-derived EVs (MSC-EVs) also exhibit immunosuppressive activity, providing many advantages compared to MSCs and have been proven therapeutic in aGVHD. Arsenic trioxide (ATO) exhibits potent antitumor effects and increasing studies indicate its immunosuppressive effects. However, ATO at high concentrations can cause severe adverse effects. If encapsulated in some kind of drug vehicles, ATO can be made less toxic. Therefore, we believed that the combination of MSC-EVs with a low dose of ATO would be an effective therapy for aGVHD.


We used a classical GVHD model (BALB/c→B6) and developed 4 groups: the control group (TCD-BM), the GVHD control group (TCD-BM + spleen T cells), the MSC-EVs treatment group and the MSC-ATO-EVs (MSC-derived ATO-encapsulating EVs) treatment group. OS, GVHD clinical and histological scores were evaluated. A20-luc lymphoma cells were injected to generate the GVL model. Using flow cytometry analysis, we analyzed Th cell subsets, cytokines and transcription factors (Th1*IFN-γ/TNF-α*T-bet, Th2*IL-4*GATA3, Th17*IL-17*RORγt, Treg*IL-10*Foxp3) and sorted CD8+ SLECs, and CD8+ MPECs in BM and spleen of recipients. Dll4 expression was analyzed on DCs. B6 cells were incubated with or without BALB/c spleen cells and complete medium alone, with 10 mM ATO alone. T cell apoptosis was determined with Yopro-1 staining. We used MLR assays to examine Th subsets, cytokines and notch targeted genes with or without ATO or neutralizing Ab specific to Dll4 (anti-Dll4).


BALB/c mice receiving B6 TCD-BM alone developed no sign of GVHD, whereas all BALB/c mice receiving B6 donor TCD-BM + spleen T cells died of GVHD. In contrast, injection of MSC-EVs and MSC-ATO-EVs inhibited GVHD in T cell recipients, with 20% and 29% of them surviving without severe GVHD, respectively. These survival rates were accompanied by significantly lower clinical and histological scores. GVL effects mediated by MSC-EVs and MSC-ATO-EVs were comparable to those obtained in the GVHD control group.

Compared to the control group, CD4+T and CD8+T cells increased substantially in T cell recipients, resulting in severe GVHD. In contrast, treatment with MSC-ATO-EVs significantly reduced the number of CD4+T and CD8+T cells, while MSC-EVs recipients retained approximately the same number of T cells as the GVHD group.

Compared to the GVHD control group, Th2 and Treg cells derived from the spleen increased, while Th1 and Th17 cells were reduced significantly in both the MSC-EVs and MSC-ATO-EVs groups. We also detected lower serum levels of TNF-α and IFNγ as well as lower expression of RORγt and T-bet in blood and BM CD4+ T cells in these two groups, while the expression of GATA3 and Foxp3 increased significantly. Treatment with MSC-ATO-EVs markedly raised the MPEC/SLEC ratio compared to the MSC-EVs and GVHD control groups. We also examined Dll4high DCs in different organs and different groups and found that only MSC-ATO-EVs significantly reduced the Dll4high DCs, especially in the spleen and intestine.

Treatment of stimulated B6 CD4+ T and CD8+ T cells with ATO increased production of H2O2. Yopro-1 staining of activated B6 CD4+ T and CD8+ T cells indicated that ATO dramatically triggered apoptosis in those cells.

DCs were isolated and cultured with B6 mouse-derived CD4+ T or CD8+ T cells, with or without addition of ATO or anti-Dll4. ATO and anti-Dll4 both led to significant reduction of IFN-γ and TNF-α, while IL-4 and IL-10 increased slightly. We next assessed the notch pathway targeted genes in T cells and found there were significantly increased GATA3 and reduced Dtx expression levels.


Altogether, our findings demonstrate that MSC-ATO-EVs might be a highly promising therapy for aGVHD through reducing T cell amounts and modulating Th subsets and CD8+ T cell differentiation. These effects can be explained with the inhibition of the Dll4-notch pathway by ATO. Therefore, further exploitation of the potential application of ATO in aGVHD and the mechanisms of action of ATO may improve outcomes after allo-HSCT.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.