Hypomethylating Agent Azacitidine Induces FoxP3 Negative HLA-G Expressing Immunoregulatory T Cells

The major obstacles in using FOXP3+ T regulatory cells as T-cell based immunotherapy against GVHD after allogeneic hematopoietic cell transplantation are their low numbers in the circulation and the lack of specific cell surface markers for efficient purification. DNA methylation has been considered to play a role in the regulation of T-cell effector function and cytokine gene expression, indicating a promising role of hypomethylating agents for immunomodulation. Recently it was shown that in-vitro treatment of conventional T-cells with hypomethylating agent azacitidine (aza) induced FoxP3 expression and converted CD4⁺CD25^- cells into immunosuppressive T-cells the suppressor function of which is independent of FOXP3 expression (Choi et al Blood 116:129;2010), suggesting that aza induced suppressor function depends on the modification of other hypomethylated genes. Human leukocyte antigen-G (HLA-G) is a non-classical HLA class I molecule, shown to exert immunoregulatory functions, the expression of which is epigenetically regulated. In this study we investigated whether hypomethylating agent aza can induce HLA-G+ immunoregulatory T cells. We used CD3⁺ T cells from peripheral blood of healthy individuals isolated by MACS negative selection. T cells were stimulated with anti-CD3⁺ plus anti-CD28⁺ coated magnetic beads and then were treated for 72 hours with aza (0.5-15 mM) in the presence of 50U/ml inteleukin-2 (IL-2). Detailed phenotypical characterization of in-vitro aza treated cells was performed with flow cytomentry. Aza-induced FACS sorted HLA-G⁺ T cells, were irradiated and then used as third party cells in allogeneic mixed lymphocyte cultures (MLC). For the analysis of the in-vivo effect of aza on HLA-G expression, peripheral blood of patients with myelodysplastic syndrome (MDS) was obtained at baseline and after Vidaza treatment. In-vitro treatment of CD3⁺ T cells with aza increases the percentage of HLA-G+ cells, with maximum induction at a concentration of aza 7.5 mM in comparison to control (aza-0mM) (n=4, 9,11±0,6% vs 0,76±0,74%, p<0,0001). However optimum aza concentration for the maximum HLA-G induction with the lowest toxicity in CD3 T cells was determined at 5 mM (HLA-G⁺:6,88±3,9%, p=0,0022). Aza treatment increases the percentage of CD4⁺CD25^highFoxP3⁺ cells in culture. Maximum HLA-G induction was observed on CD4⁺CD25⁺ population (n=2, 9,21±5,5%). Aza treatment of FACS-sorted CD4⁺CD25^negHLA-G^neg cells induced CD4^lowCD25⁺HLA-G positive cells, revealing that aza induced HLA-G positive cells are not the result of a selectively expanded proexisting HLA-G⁺ population. Strickingly aza induced CD4^lowCD25⁺HLA-G⁺ are FoxP3 negative. Aza induced HLA-G+ cells showed suppressive function (20% inhibition of proliferation) in allo MLCs, whereas HLA-G- selected cells had no effect.The percentage of peripheral blood HLA-G⁺ lymphocytes of healthy donors was 1,33±0,36% (n=4) in comparison to MDS patients which was 2,08±0,99% (n=2). In accordance to the in- vitro studies, preliminary data from MDS patients under Vidaza show up to 10 fold increase in CD4⁺CD25^highHLA-G⁺ cells on day 28 post treatment. In conclusion, aza induces de-novo HLA-G expression on CD4⁺ T cells in-vitro and in-vivo. By using hypomethylating agent aza in-vitro, we generated a CD4^lowHLA-G⁺ FoxP3^neg immunoregulatory population, that is highly possible to be the aza induced FoxP3 independent T regulatory population.This strategy may provide an easy methodology to generate ex-vivo immunoregulatory T cells for adoptive immunotherapy.