Expression of Molecules Involved in Fetal-Maternal Tolerance Allows Human Mesenchymal Stem Cells to Engraft at High Levels across Immunologic Barriers

The ability of HLA-G to inhibit NK cells, cytotoxic T cells, and the maturation of dendritic cells is thought to play a crucial role in the maintenance of fetomaternal tolerance during pregnancy. Recent studies have provided evidence that MSC are not as immuno-inert as once thought, since they are capable of expressing HLA-II if stimulated by IFN-γ and of activating and serve as targets of NK-mediated lysis. Here we investigated whether transducing human MSC with a retroviral vector encoding HLA-G1 (MSC-G1) or G5 (MSC-G5) would render these cells and their differentiated progeny undetectable by the immune system, and thereby allow efficient engraftment into immuno-competent xenogeneic recipients. Mixed lymphocyte reactions showed that HLA-G1 expression reduced MSC’s ability to stimulate human and sheep lymphocytes by at least 80%, while HLA-G5 expression reduced activation of human lymphocytes by 35–63% and sheep lymphocytes by 48–64%. Furthermore, MSC-G1 exhibited a 35–60% reduction in NK cell lysis depending on the effector:target ratio, and MSC-G5 exhibited NK cell inhibition of 12–44%. These results indicated that cell contact-mediated NK cell inhibition by the transmembrane form of HLA-G (HLA-G1) was more effective than inhibition mediated by the soluble HLA-G5. To assess whether these in vitro results were indicative of true immune-evasion ability, we next transplanted (Tx) 10⁵ unMSC, MSC-G1, or MSC-G5 into fetal sheep recipients during the pre-immune period (55 days; n=9), or after immunocompetence was achieved at 82 days (n=10) or 104 days (n=8) of gestation. Evaluation of the recipients’ hematopoietic system at 42 days post-transplant for the presence of human cell engraftment revealed that Tx at 55 days resulted in similar levels of engraftment for all cell types (MSC:4±0.9; MSC-G1:6±0.3; MSC-G5:5±0.3%). In contrast, while unMSC engrafted at very low levels at 82d and 104d, Tx of MSC-G1 and-G5 at these later time points resulted in levels of engraftment that were considerably higher than those achieved during the pre-immune period (82d:7±1;18±2.% and 104d:12±2;16±2%). A similar outcome was seen with liver engraftment and hepatic differentiation, with MSC-G1 and MSC–G5 giving rise to 3–5 times more hepatocytes than unMSC at later time points. However, even at the early Tx time point of 55days, MSC-G1 and MSC–G5 gave rise to, respectively, 5 times and 2 times more donor-derived hepatocytes than their unMSC counterpart, suggesting that, even at the earlier stages of development, the fetus may not be completely immuno-naive. Moreover, while no donor cells were found in the lungs of animals that received the Tx at 56 days of gestation, donor cells were consistently observed primarily within the epithelium and the fibro-elastic layer in the lungs of animals that received either MSC-G1 or MSC-G5 at 82 and 104 days of gestation, and engraftment efficiency appeared to increase as a function of gestational age at the time of Tx. Serial sectioning of these tissues demonstrated that transduced cells engrafting within the lung produced HLA-G and Prosurfactant B, suggesting differentiation to alveolar Type II cells. Our studies demonstrate that the forced expression of HLA-G1 or G5 enables MSC to evade a competent recipient immune system and engraft at significant levels at times in gestation when donor cells are normally rejected. These findings suggest this approach could enable the treatment of diseases affecting tissues that develop after fetal immune development such as cystic fibrosis and may allow broadening of the use of MSC to diseases in which an underlying MSC defect precludes the use of the patient’s own MSC