The γδT cells are present in endometrium of all mammals throughout pregnancy 9. It has been shown that these cells specifically colonize the non-pregnant murine and sheep endometrium and show a dramatic increase during pregnancy suggesting a special role at the feto-maternal interface 34. The number of γδT cells in the uterus is higher in allogeneic than syngeneic pregnancy, and the expression of the TCRγδ in the pregnant uterus is shown to be hormonally controlled 33. The decidual γδT cells are large granular lymphocytes rich in intracytoplasmic granules and express neither CD4 nor CD8 (double negative) 101113. They are CD2⁺, express the activation marker CD69, the memory/activation marker CD45RO, the NK receptors CD94 1013 and NKG2D (manuscript in preparation), and CTLA 4 (fig. 2), a marker associated with regulatory T cells. The human γδT cells comprise a heterogeneous population: double positive TCRγδ⁺/CD56 ^{+ dim} cells and TCRγδ single positive cells 13. The counterpart of these cells in the murine system seems to be the TCRγδ⁺/ asialoGM⁺ cells and the single positive murine γδT cells as described by Arck et al 2. The surface density of the TCR/CD3 is low in freshly isolated decidual γδT cells (10, 35) but can be up-regulated in vitro 10.

The vast majority of the human decidual γδT cells are Vδ1⁺ 1336. Itohara et al. 37 and Heyborne et al. 42 have shown that the Vδ1 chain is also preferentially used by γδT cells in the uterus of normal and pregnant mice. Thus, the γδT cells in pregnant uterine mucosa, like other mucosa-associated γδT cells, are resident Vδ1⁺ cells.

Similar to resident γδT cells at other mucosal sites 20, the decidual γδT cells are activated but resting. We have shown that they posses a cytotoxic potency and express five major cytolytic molecules: perforin (Pf), granzyme A, granzyme B, granulysin and Fas ligand (FasL), and store them in microvesicles in intracytoplasmic cytolytic granules 26. Like other cytotoxic lymphocytes 39 the decidual γδT cells do not express FasL on their surface but store preformed FasL in the granules, and can rapidly mobilize it to the cell surface upon stimulation. Thus, the two major cytotoxic mechanisms – Pf- and FasL-mediated – are performed by one common secretory pathway based on cytolytic granule exocytosis 26. Cytotoxic mechanisms play a crucial role in the clearance of viral and bacterial infections, tumor surveillance, transplant rejection, homeostatic regulation of immune responses and peripheral tolerance 40. Logically these mechanisms should have an important function at the feto-maternal interface by protecting the maternal-fetal unit against pathogens, controlling invasion of placental trophoblast, and creating a local transient immunotolerance toward the semiallogeneic conceptus through deletion of fetus-reactive lymphocyte clones. Indeed, recent studies of Pf- and FasL-deficient mice have shown that although functional deletion of Pf or FasL alone does not appear to affect fertility, the combined absence of these two effector molecules induces infertility 40.

Interestingly, we were able to stain γδT cells in mitosis 13 proving that the γδT cells divide in human decidua. As a rule, the plasma membrane of the mitotic cells was strongly stained with the reaction product indicating a high level of γδT cell receptor expression 13. Our finding of γδT cells dividing in situ is in line with previous suggestion that γδT cells might expand in epithelial sites exposed to external environmental antigens, and, in some cases, recognize self-antigens, specific to a particular local environment 72027. By analogy, decidual Vδ1⁺ T cells may recognize trophoblast-related antigens and be involved in controlling trophoblast invasion during placenta formation 41.

In previous reports, Hayakawa et al. 17 in the human system and Kimura et al. 18 in the murine system have shown expression of mRNA for RAG-1 and RAG-2 proteins, which are required for TCR rearrangement, in human CD56^bright/CD16^- cells and in murine decidual mononuclear cells respectively. We have confirmed and extended these results showing that transcripts of RAG can be easily detected in purified CD56⁺, CD2⁺, c-kit⁺ or IL-7R⁺ decidual cells implying an ongoing process of TCR gene rearrangement 13. There is no doubt that ongoing rearrangement of TCRγδ takes place in decidua, probably for two purposes: 1) local extrathymic differentiation of γδT cells by TCR receptor rearrangements and 2) secondary TCRγδ rearrangement, permitting editing of antigen receptors on mature cells, thus adjusting the decidual γδT-cell repertoire to the ongoing pregnancy. Re-induced RAG expression, involved in receptor editing is a phenomenon observed in immature T cells in thymus and seems to be required for the generation of normal T-cell repertoire 42. Although not proven yet it is reasonable to assume that both local TCRγδ receptor rearrangement and editing are equally used in decidua.

Is there a need for T-cell differentiation in decidua? What purpose and biological significance there might be for extrathymic T-cell differentiation during pregnancy?

We can argue for at least two different reasons for extrathymic maturation in pregnancy. The first reason is priming the maternal immune system to the fetus. The meeting between the mother and the fetus is dual: 1) between the maternal blood and syncytiotrophoblast cells of the chorion villi of the placenta and 2) between the extravillous trophoblast and the maternal epithelial, stromal, endothelial and immune cells in decidua when placenta is formed. It is reasonable to assume that the first encounter and antigen presentation of fetal antigens to the immune system takes place in decidua. Decidua/endometrium might enrich CD56⁺ progenitor cells of bone marrow origin which will further differentiate/rearrange locally (or naive thymus-derived T cells will edit their TCR) upon the encounter of fetal antigens. The extrathymic maturation in decidua might be one of the mechanisms adjusting the immune system and the T-cell repertoire towards acceptance of the ongoing pregnancy. Heyborne et al have shown that murine decidual γδT cells recognize trophoblast-derived antigens. Immune cells, locally primed in decidua might then repopulate the peripheral blood of the pregnant woman as suggested by published reports [reviewed in 32].

The second reason for extrathymic maturation in decidua might be the temporary thymic involution taking place during pregnancy. A great loss in thymic weight during pregnancy occurs due to increased cell death of small lymphocytes from the cortex. It appears that primarily CD4⁺/CD8⁺ cortical thymocytes are lost whereas most other subsets are retained; B-lymphopoiesis is depressed 43. The cortical involution is at its greatest by the end of pregnancy and is maintained until lactation ceases. The implications of the observed thymic changes can be anticipated if they are correlated to the known thymic function [44, Fig. 4]. It is reasonable to assume that such radical rearrangements are likely to have effects on the maternal immune system and to influence the mother's ability to protect the fetus from harmful maternal responses to paternally inherited fetal antigens. The involution of cortex might mean deleting clones or unresponsiveness to paternally derived antigens 45. It is tempting to interpret the enlargement of the medulla as a potential increase of regulatory T cells needed to modulate the immune responses. Is there a role for decidua in the context of the thymic involution during pregnancy? The decidua as an extrathymic maturation site can be complementary to the thymic changes in at least two ways: 1) The need for positive selection ablated by cortex involution might be compensated for by extrathymic differentiation of T cells which will be primed on pregnancy-derived antigens in the decidual microenvironment and will allow to eliminate/silence fetus-reactive T cell clones. 2) Naive T cells generated in the medulla (e.g. regulatory cells) might be re-edited in the decidua thus adjusting their T-cell receptor repertoire to the ongoing pregnancy. The γδT cells, differentiated locally in decidua 131718, will thus be specifically primed on the ongoing pregnancy.

Cytokines at the fetomaternal interface play a pivotal role for the establishment and maintenance of normal pregnancy. Several well-performed studies in humans and mice have shown beyond doubt that there is a T-helper (Th) 2 bias in the cytokine response [reviewed in 46]. But the role of cytokines in pregnancy cannot solely be explained by the Th2/Th1 paradigm. Although very attractive, there is a serious risk of oversimplifying this concept. First, a critical feature of the Th1/Th2 model is that the two cell types counter regulate one another via cytokine production. But the polarization of the Th1 versus Th 2 effector cells is rarely complete and simultaneous Th1 and Th2 responses are possible. Second, this concept is derived from results of in vitro experiments and experimental models with immunologically inactive, inbred laboratory mice. In reality, when faced with established responses, the Th1 effectors have little ability to down-regulate Th2 responses 47. Similarly, Th2 effector cells, carefully separated from the Th2-like regulatory cells, have been shown to aggravate, rather that inhibit Th1-mediated inflammatory responses 47. There is a compiling body of evidence that the T-cell function at the feto-maternal interface in successful pregnancy is modulated by a cytokine environment of IL-10 and TGF-β, cytokines that are not always viewed as Th2-type only 48. Abandoning the Th1/Th2 bias one can ask the question if other, non-Th1/Th2 cells and responses operate at the fetomaternal interface. Careful studies of decidual γδT cells in the murine system have shown that, TCRγδ⁺/asialoGM1⁺ cells and TCRγδ single positive cells 2 play a decisive role in pregnancy outcome depending on their cytokine response. At early preimplantation stage, murine γδT cells produce TNF-α, IFN-γ and probably IL-2 and promote abortions by activation of decidual NK cells and macrophages. At a later stage, during the time of implantation and placenta formation, the γδT cells in murine decidua produce TGF-β and IL-10 and exert anti-abortogenic effect 249. Using quantitative RT-PCR we have analyzed the cytokine profile of the two main subpopulations of γδT cells in human decidua: TCRγδ⁺/CD56⁺ and TCRγδ single positive cells 1013. Our results 50 show that the TCRγδ⁺/CD56⁺ cells almost exclusively express mRNA for TGF-β1 and IL-10 cells suggesting orientation towards an immunosuppressive profile 48. Then as they further develop into primed TCRγδ single positive cells their IL-10 and TGF-β1 expression is strongly enhanced. Additionally, the TCRγδ single positive cells transcribe two more cytokines-IL-6, suggesting an orientation toward the pregnancy-promoting Th 2 response and IL-1β, a cytokine considered in general to have a function promoting maturation and clonal expansion of other lymphocyte subpopulations. In pregnancy in particular, IL-1β is considered to be an important factor for the implantation of the blastocyst in the uterine cavity acting through up-regulation of adhesion molecule expression 51. Our results 50 based on quantitative cytokine mRNA measurement in these two subpopulations of human decidual γδT cells indicates that these cells, by virtue of the strong dominance /exclusivity of IL-10/TGF-β mRNA expression can be ascribed to the newly "reborn" suppressor/regulatory T (Treg) cells. Furthermore, these cells express the regulatory T cell marker CTLA4 (Fig. 2). A brief summary of some of the characteristics of the Treg cells is given in Table 2.

Recent data provide convincing evidence that a specialized population of Treg cells both do exist and play a critical role in the generation and maintenance of immunological tolerance in humans [reviewed in 4852]. The Treg cells exert their suppressive activities on effector cells in remarkably low numbers 53. Several subsets of Treg cells have been described in a variety of experimental models. In the context of this discussion it is interesting to note that two human CD4⁺ Treg-cell subsets exert their regulatory effect via secretion of the immunosuppressive cytokines IL-10 and TGF-β. One subset includes the so-called T helper 3 (Th3) cells, distinguished from other T helper cells by their ability to produce high levels of TGF-β and varying levels of IL-10 and IL-4 52. The other subset, termed T regulatory type 1 (Tr1) cells produces a significant level of IL-10, various levels of TGF-β and no IL-4 54. The decidual γδT cells produce high amount of IL-10 followed by TGF-β and no IL-4 50 and thus belong to the Tr1 regulatory type of cells. The selective generation of different Treg cell subsets is determined by the cytokine microenvironment in which naive CD4⁺ T cells encounter antigen. This means that if priming of naive Th0 cells occurs in the presence of IL-10 or TGF-β they differentiate to Tr1 or Th3 cells respectively and become polarized to synthesis of IL-10 or TGF-β [reviewed in 55]. The cell type(s) responsible for the creation of such unique cytokine microenvironments in vivo is a subject of discussions. γδT cells are excellent candidates for this role, because these cells can respond to broadly distributed self-antigens in stressed, damaged and transformed tissues and do not require classical antigen processing and MHC-restricted presentation 27. γδT cells have been implicated in the down regulation of immune responses in various inflammatory disorders and may acquire immunoregulatory properties at mucosal sites [reviewed in 56]. A population of γδT cells producing the Tr1/Th3 – type cytokines IL-10 and TGF-β has been isolated from tumor-infiltrating lymphocytes. These γδT cells could play a role in the inhibition of immune responses to tumors 56. It was shown that aerosol or nasal inoculation of intact insulin resulted in expansion of γδT cells with an immunosuppressive anti-diabetogenic effect, mediated by IL-10 56. Remarkably, only a small fraction of γδT cells was enough to prevent adaptive transfer of diabetes 56. Furthermore, γδT cells producing IL-10/TGF-β were reported to be a critical cell population for the induction of allograft- and testicular tolerance 5758.

Summing up the accounted data above an attractive hypothesis is that γδT cells act as cytokine-producing cells to create a decidual environment that actively tolerates the fetus 50. We suggest that pregnancy-related antigen(s) can activate decidual γδT cells causing them to release the immunosuppressive Tr1- and Th3-type cytokines IL-10 and TGF-β. Figure 3 illustrates two possible mechanisms by which these cells could induce local uterine tolerance towards the fetus. In the direct pathway the effector cells (cytotoxic T lymphocytes, NK cells, macrophages, dendritic and B cells) at the feto-maternal interface could be directly inhibited by IL-10 and TGF-β50. In this pathway γδT cells function as Treg cells 56. In the indirect pathway γδT cells could mediate the tolerogenic effect through generation of primed Th0, mainly TCRαβ⁺ CD4⁺ (and probably CD8⁺) cells. Under the influence of IL-10 and TGF-β, these cells differentiate into IL-10 producing Tr1-type of cells and TGF-β producing Th3 type of cells which in their turn act suppressively on the effector cells. In this pathway the γδT cells are needed for generation of efferent suppressor cells, but are not suppressors themselves 56. These two pathways might function in parallel and exert immunosuppression in concert with each other. It cannot be excluded that other types of decidual cells such as dendritic cells could also participate in the immunoregulation 59. The model presented [50, Fig. 3] is simplified but comprises one important mechanism for immunomodulation at the feto-maternal interface.