Tolerance mechanisms in pregnancy: a reappraisal of the role of class I paternal MHC antigens.

Problem: Allogeneic pregnancies have a survival advantage over syngeneic pregnancies, and paternal Class I MHC antigens have been implicated. In humans, HLA-C and HLA-G and E are expressed by subpopulations of fetal trophoblast. In mice, Qa-2, a Class Ib antigen, and classical H-2K antigens have been described. However, the mechanism of prevention of embryo demise in utero has not been critically assessed, and a number of conflicting ideas have not been addressed. The alphabeta T-cell receptor recognizes peptide bound to the groove in Class I MHC, and peptides have profound effects on the interaction of KIR receptors on T and NK cells with Class I MHC.

Methods: Data on prevention of pregnancy loss (abortion) in poly IC-treated mice were reviewed along with information about prevention of losses in the abortion-prone CBA x DBA/2 model. This information was combined with data on paternal antigen expression at different times in pregnancy when key events determining outcome are thought to transpire, and role of tolerance signaling molecules such as CD200. Current data on models supporting a role for 'true' uterine NK cells (TuNKs) versus blood NK cells in the uterus (BuNKs) and role of MHC-KIR interaction were reviewed along with incompatible data in the literature.

Results: Whilst paternal Class I MHC appears important, there is an important role for paternal non-MHC minor antigens (small peptides) that bind to the antigen-presenting groove of Class I MHC. BuNKs along with CD8(+) T cells and Treg cells appear more important than TuNKs where the role of the latter appears primarily to promote angiogenesis. When during pregnancy the maternal immune system cells are first exposed to paternal Class I + peptide is uncertain, but at the time of implantation, if not earlier, seems likely.

Conclusion: Suppression of pregnancy loss by paternal/embryo Class I MHC depends on the presence of paternal peptides. This greatly complicates existing models of Class I-KIR interactions in feto-maternal tolerance or rejection. It is important to consider all the data when devising explanatory models.