This mechanism of natural tolerance in a semiallogeneic context has led to investigate the potential role of HLAG in transplanted patients. To date, clinical studies have demonstrated that HLA-G expression may be induced in some heart, kidney, liver/kidney, lung, pancreas, and kidney/pancreas transplanted patients. Statistical analyses indicate that the presence of HLA-G in plasma and biopsies of transplanted patients correlates with a decreased number of acute rejection episodes and with no chronic rejection, as first described for heart transplants. The direct role of HLA-G in transplantation in vivo was evidenced by skin allotransplantation in HLA-G transgenic mice or in wild-type mice pre-treated with HLA-G tetramer-coated beads. In both experiments the presence of HLA-G significantly delayed skin allograft rejection. For these reasons, and also because it already contributes to the best example of successful tolerance there is: the maternal-fetal tolerance, therapeutic HLA-G molecules for transplantation are actively investigated. Yet, the use of HLA-G molecules as therapeutic agents faces several hurdles, among which the problems of structure and stability. Indeed, HLA-G is a trimolecular complex composed of a heavy chain of 3 globular domains non-covalently associated with the b2-microglobulin and a peptide which is active only as a multimer. Here, we evaluated the tolerogenic function of two types of HLA-G homodimers, whether the alpha-1 domain of HLA-G which is common to all HLA-G isoforms could carry a tolerogenic function by itself as it was originally postulated, and whether the trimolecular complex that constitutes HLA-G could be stabilized by fusing B2M to HLA-G heavy chain while retaining its tolerogenic properties. Our results demonstrate the tolerogenic function of all investigated dimeric forms of HLA-G recombinant proteins in vitro and in vivo, and especially that of the B2M-HLA-G5 dimers in vivo, but do not fully support a tolerogenic function for the alpha-1 domain of HLA-G in human beings, even dimeric. In this work, we investigated the tolerogenic properties of HLAG recombinant proteins. These were B2M-HLA-G heavy chain fusions and HLA-G alpha-1 domains dimerized either through an Fc fragment, or naturally through a C42-C42 disulfide bond. Of note, because of the presence of B2M and the linker in our constructs, the cysteines involved in HLA-G homodimerization were no Paclitaxel longer in position 42, but for the sake of clarity, we kept calling C42 the cysteines of the HLA-G alpha-1 domain that are responsible for homodimerization. Our first aim was to evaluate the tolerogenic function of two types of single-chain B2M-HLA-G homodimers. In this study, we showed that all generated proteins and peptide were multimerized. Furthermore, B2M-HLA-G fusions were properly folded and could bind and activate the ILT2 receptor.