REVIEW ARTICLE: Regulatory T Cells and Their Role in Pregnancy

Authors


Ana Claudia Zenclussen, Department of Experimental Obstetrics & Gynecology, Medical Faculty, Otto-von-Guericke University, Gerhart-Hauptmann-Str. 35, 39108 Magdeburg, Germany.
E-mail: ana.zenclussen@med.ovgu.de

Abstract

Citation Leber A, Teles A, Zenclussen AC. Regulatory T cells and their role in pregnancy. Am J Reprod Immunol 2010

Regulatory T cells emerge in the last years as key players in allowing fetal survival within the maternal uterus. They were shown to be a unique subpopulation of T cells expanding during human and murine pregnancy. The importance of Treg for a normal pregnancy situation was proven by studies showing that their absence impairs murine pregnancy while the adoptive transfer of Treg prevents fetal rejection. In humans, pregnancy pathologies are associated with lower Treg frequencies while therapies that improve pregnancy outcome are able to boost their number. Functional studies have shown that Treg can regulate immune cell responses directly at the fetal–maternal interface either by interacting with other cells or by inducing the expression of immune regulatory molecules. This article revises relevant literature on regulatory T cells in human and murine pregnancy.

Introduction

During pregnancy, semiallogeneic tissue, the fetus, is allowed to grow within the maternal uterus without being rejected by the maternal immune system, which becomes aware of its presence and actively tolerates it. The fetal–maternal interface can be therefore considered as a site of immune privilege, being the conceptus partially responsible for the establishment of a maternal immune response toward a protective, tolerant one. Accumulating body of evidences positions regulatory T cells as important contributors for the establishment and maintenance of active immune tolerance toward the fetus during pregnancy. This review will go through and discuss relevant literature on regulatory T cells in human and murine pregnancy.

Treg: generalities

CD4+ CD25+ regulatory T cells (Treg) were first described as a specialized subpopulation of T cells responsible for the maintenance of immunological self-tolerance by actively suppressing self-reactive lymphocytes.1 They play a major role in preventing autoimmunity and tolerating allogeneic organ grafts.1,2 The forkhead box transcription factor, Foxp3, is exclusively expressed in mouse CD4+ Treg and acts as a master switch in the regulation of Treg development and function.3–5 Several types of cells with regulatory function were described from which the majority is naturally produced in the thymus in a continuous mode expressing both glucocorticoid-induced TNF receptor family-related gene (GITR) and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) and produce soluble TGF-β and Interleukin-10 (IL-10).6 Two other subtypes (Tr1, Th3 cells) are generated extra-thymically and develop from naïve T cells after exposure to antigens in the periphery.7 Other cells, like e.g., the so-called TofB are also claimed to have regulatory functions.8

Treg in human and murine pregnancy

During human pregnancy, a systemic expansion of Treg can be observed already at very early stages.9,10 Convincing evidences indicate that Treg are specific to paternally derived cells,11,12 which points out that their function is to protect paternally derived cells from immune rejection by the mother′s immune system. Infertility has been proposed to be associated with reduced expression of the T-regulatory cell transcription factor Foxp3 in endometrial tissue.13 Sasaki and colleagues14 first reported that spontaneous abortion cases were also associated with lower systemic Treg levels when compared to normally developing pregnancies. Arruvito and colleagues15 further confirmed lower Treg levels in patients with recurrent miscarriage and found that paternal alloimmunization acts to modulate serum levels of factors involved in the IL-6 trans-signaling pathway and increases the frequency of Treg. An expansion of Treg in decidua from normal pregnant women accompanied by a low occurrence of Th17 cells was recently confirmed by Mjösberg and collaborators.16 Furthermore, these authors propose that Treg may be in charge of controlling the Th1 activity that is found locally in normal early pregnancy.16 The association between Treg number or function and pre-eclampsia is not clear as contradictory data have been published.17–21 There is no doubt that Treg expand in the periphery in human pregnancy and are present at important numbers at the fetal–maternal interface, preferentially in the maternal decidua.

During murine pregnancy, CD4+ CD25+ Foxp3+ Treg have been shown to expand systemically as well as locally at the fetal–maternal interface.22,23 Several studies confirmed an increase of Treg already on day 2 of pregnancy in blood, lymph nodes, and thymus followed by a decrease in Treg number from mid-gestation onward until Treg number reaches non-pregnant levels at term or shortly thereafter.22–24 As in humans, murine spontaneous abortion is associated with a decreased frequency of Treg.23,24

Although Treg are well studied in pregnancy models, the mechanisms and factors behind Treg expansion during normal gestation are still under investigation. In this regard, Aluvihare and colleagues22 suggested an alloantigen-independent increase of Treg because of their observation that Treg elevation takes place both in syngeneic and in allogeneic matings. Because of hormonal changes during pregnancy, they proposed that hormones might be involved in Treg expansion. This assumption is supported by a study which showed that estrogen at physiological doses not only expands Treg but also stimulates the conversion of CD4+ CD25 T cells into CD4+ CD25+ T cells in vitro.25 In contrast, Zhao and colleagues26 showed that neither estrogen or progesterone alone nor their combination influenced the frequencies of Treg in ovariectomized mice. Moreover, they detected higher numbers of Treg in allogeneically than in syngeneically pregnant mice suggesting an involvement of paternal antigens in Treg expansion.26 We recently found paternal antigens immediately after insemination in several organs in female mice.27 This emphasizes the possibility that Treg proliferate after encountering paternal antigens derived from semen and presented by antigen-presenting cells (APCs) in lymphoid organs at early pregnancy stages.27 Experiments performed using different pairing combinations including CBA/J females mated with intact Balb/c males, vasectomized Balb/c (lacking sperms), or Balb/c without seminal vesicles (lacking seminal fluid) suggest that both sperm and components of the seminal fluid are essential for Treg expansion (Leber et al. unpublished observations), which strongly suggest an alloantigen-driven Treg expansion. Our data are further supported by studies from Sarah Robertson and colleagues28 who recently showed that seminal fluid can drive Treg expansion. Thus, antigen-independent as well as antigen-dependent mechanisms are likely to be involved in Treg expansion. However, the observation that Treg expand in higher numbers in allogeneic than in syngeneic pregnancies makes it more likely that in naturally occurring matings, namely allogeneic combinations, paternal antigens rather than hormones drive Treg expansion.

Besides the pregnancy-associated Treg increase, it has been shown that Treg vary in their number during the estrus cycle already before a pregnancy is established. They accumulate periodically in the uterus and draining lymph nodes every time a female approaches estrus (Teles et al. unpublished observation).29 This periodic accumulation of Treg is accompanied by matching fluctuations in uterine expression of several chemokines like CCL3, CCL4, and CCL5, which have been shown to play a role in the recruitment and retention of Treg.29 Additionally, pregnancy-related hormones such as estrogen and progesterone might play a role in Treg accumulation as they also fluctuate during the estrus cycle. In this regard, an expansion of CD4+ CD25+ FOXP3+ Treg was detected in fertile non-pregnant women in the late follicular phase of the menstrual cycle. This increase was tightly correlated with serum levels of estradiol and was followed by a dramatic decrease in Treg numbers at the luteal phase.30 Hence, accumulation of Treg during the receptive phase of the menstrual or estrus cycle may facilitate the encounter of paternal antigens by Treg immediately after insemination and is supposed to prepare the uterus for a possible implantation event.

Generation of Treg in pregnancy and antigen specificity

How Treg are generated and expand during pregnancy is still under debate. While an in vitro study supports the influence of hormones on both conversion of CD4+ CD25 T cells into CD4+ CD25+ T cells in vitro and Treg expansion,25 other reports question this being the only factor regulating Treg expansion during pregnancy. In this regard, we have described an increase in systemic regulatory T cell levels in pseudopregnant females after mating with vasectomized males but not after mechanically induced pseudopregnancy.31 Zhao and colleagues26 also demonstrated that neither estrogen or progesterone alone nor both hormones combined had an influence on the frequencies of Treg in ovariectomized mice. Antigenic presentation of paternal structures in female reproductive tissues may be responsible for Treg increase at the beginning of pregnancy and their later expansion.32 This is of importance for the allogeneic fetus to avoid immune attack from the mother. While it was long believed that fetal cells and antigens do not cross the ‘placental barrier’, it is now well documented that microchimerism do occur during murine and human pregnancy.33–37 Khosrotehrani and colleagues38 described the presence of paternal antigens in maternal organs throughout all stages of pregnancy in mouse. This was further confirmed by us after mating wild-type females with GFP+ transgenic males.27 We could detect paternal antigens and cells in maternal lymph nodes, spleen, and blood among other organs, during early, middle, and late pregnancy.27 Tan and colleagues even described the presence of fetal cells in maternal mouse brain post-partum.39 By using sensitive quantitative polymerase chain reaction tests, Dutta and colleagues described the occurrence of maternal microchimerism (MMc) in heart, lungs, liver, and blood of 61% of adults and 100% of neonates.40 MMc was further detected in isolated CD4+ CD11b+ and CD11c+ cells subsets of spleen and in lineage-positive cells in heart.39 Further, we were able to identify paternal antigens in vaginal lumen immediately after fecundation, which provides evidence that paternal antigens are able to be processed early during pregnancy.27 We speculated about the existence of direct and indirect antigenic presentation after observing male APCs in vaginal lumen27 and because of the presence of maternal APCs capable of processing paternal antigens.32 This was also elegantly demonstrated by Moldenhauer and colleagues32 who showed that through female APC-mediated Ag cross-presentation via a TAP-dependent pathway, seminal fluid promotes an activation and expansion of paternal antigen-reactive CD4+ and CD8+ T cell populations. As the presence of paternally derived antigens in maternal organs coincides with the expansion of Treg and some of the paternal cells are APCs, we propose a scenario in which paternal antigens are processed early during pregnancy, which leads to the generation of Treg. Exposure of paternal Ags together with immune-deviating cytokines in seminal fluid at conception may therefore activate Ag-dependent CD4+ Treg and CD8+ cells that then mediate pregnancy tolerance. A continuous release of placental antigens into the maternal circulation40,41 would further allow the maintenance of a Treg population that is specific for paternal antigens and mediates tolerance toward the semiallogeneic fetus until birth time point. Further evidences for antigen specificity of Treg during pregnancy were provided by a study showing an effective in vitro suppression of allo-antigen responses by Treg from peripheral blood of pregnant and non-pregnant women.11 Using a MLC-system with maternal cells and paternal antigens, Mjösberg and colleagues11 observed that cells from normal pregnant women secreted more IL-4 than non-pregnant controls. Treg further suppressed IFN-gamma reactivity against paternal and unrelated alloantigens. Interestingly, Treg suppressed IL-4 secretion against paternal but not unrelated alloantigens during pregnancy. In transplantation studies, it was described that antigen-specific cytotoxic T lymphocytes (Tctl) are directed against fetal inherited paternal alloantigen such as HLA42 and minor antigens.43 By analyzing peripheral blood cells from 17 minor H antigen-disparate mother-offspring pairs, van Halteren and colleagues44 observed that both antigen-specific CD8+ CTLs (cytotoxic T lymphocytes) and CTLA-4-dependent CD8+ Treg are able to emerge during pregnancy and persist over time when mother and offspring differ on minor histocompatibility antigens. Additionally, frequencies of CD4+ CD25+ Treg increase more in allogeneically than in syngeneically pregnant mice thus contributing to a lowered alloreactivity against paternal antigens in allogeneically compared with syngeneically pregnant mice.26 In consensus, Tafuri and colleagues45 reported that mice with a Kb positive conceptus had reduced numbers of Kb-reactive T cells and accepted Kb positive tumor grafts in contrast to syngeneic and third-party allogeneic pregnancies. T cell phenotype and responsiveness were restored after delivery.45 Aluvihare and colleagues22 found that allogeneic fetuses were rejected in the absence of CD25+ cells whereas the same did not occur with syngeneic fetuses. Similar results were obtained by Darrasse-Jèze and colleagues46 who observed that semi-allogeneic pregnancies in mice induce an expansion of Treg, but not of activated CD4+ and CD8+ T cells, in the lymph nodes draining the uterus. The application of a CD25 reactive PC61 monoclonal antibody to female mice before mating resulted in the depletion of Treg and expansion of activated CD4+ and CD8+ T cells in the draining lymph nodes leading to fetal rejection.46 The application of a CD25 reactive PC61 monoclonal antibody on female mice before allogenic mating was also reported to impair implantation.23 Interestingly, it has been recently shown that maternal antigens are present in human fetuses and they are protected from rejection by the generation of Treg that suppress fetal immune response.47 These authors found out that a substantial number of maternal cells do cross the placenta to reside in fetal lymph nodes, inducing the development of CD4+ CD25high FoxP3+ Treg that suppress fetal anti-maternal immunity and persist at least until early adulthood. This breakthrough data clearly contradict an old dogma which postulated that maternal cells do not cross the placenta to the fetus and that the fetal immune system is immature.48 Mold′s findings reveal a form of antigen-specific tolerance in humans, induced in utero and probably active in regulating immune responses after birth. All these evidences obtained in different models and species clearly confirm the existence of an antigen-specific immune response during pregnancy and position Treg in a central role herein.

Kinetics of Treg during pregnancy

The trigger time point of Treg generation during pregnancy is still unknown. In mice, an expansion of Treg population could be detected in lymph nodes draining the uterus already on day 2 of pregnancy.22,24 Previous studies from our group suggest that fetal rejection in DBA/2J-mated CBA/J females could be prevented by adoptively transferring Treg from BALBc-mated CBAJ females. This treatment was only successful when applied at early stages of pregnancy.23 In the same line, it was observed that depletion of Treg using the PC61 monoclonal antibody in the first 2 days of pregnancy caused implantation failure.23 This suggests that Treg are necessary locally before implantation which was further supported by the observation that the expansion of the Treg subpopulation starts early after conception and prior to implantation in several organs as lymph nodes, blood, spleen, and thymus.24 Moreover, studies from Robertson and colleagues28 and Moldenhauer and colleagues32 strongly suggest a role for seminal fluid in the activation of female Treg response prior embryo implantation. Zhao and colleagues26 further describe a gradual decrease in the murine Treg population from mid-gestation to levels similar to non-pregnant mice after fetal delivery. Sasaki and colleagues14 were the first to describe an increase in the CD4+ CD25+ cells in decidual tissue in early human pregnancy. This was supported by the works of Heikkinen and colleagues49 and Somerset and colleagues9 who also observed an increase in the population of CD4+ CD25+ circulating cells in early pregnancy and described a peak of this population during the second trimester and a posterior gradual decline to levels slightly higher than non-pregnant levels during the post-partum period. These cells were lately characterized as being CD4+ CD25high Treg and FOXP3-enriched while they could exert suppressive functions in vitro.14 Additionally, Zhao and colleagues26 also described that late pregnancy and post-partum decrease in the population of human CD4+ CD25high cells is followed by an increase in CD4+ CD25low T cells. There is consensus from human and murine studies that Treg expand early in pregnancy and begin to decline at the end of pregnancy to return to normal non-pregnant levels few days post-partum.

Treg in pregnancy: Thymic versus de novo generation or conversion

First considered to be a homogeneous population of naturally occurring CD4+ CD25+ cells (nTregs) derived from the thymus, Foxp3 CD4+ CD25+ Treg are now known to have two distinct pathways of generation distinguished by having different principal antigen specificities and T-cell receptor signal strength and co-stimulatory requirements: Thymus derived naturally occurring CD4+ cells expressing CD25, the α-chain of the interleukin 2 (IL-2) receptor (nTregs),6,50 and adaptive CD4+ CD25+ cells that are induced from CD25 precursors in the peripheral lymphoid organs (iTreg).51 Aging thymic involution is well known among vertebrates and occurs with several structural and functional alterations linked to the senescence of the immune system.52–55 Despite this, aged mice CD4+ CD25high T cells are still able to suppress the activation of normal young CD4+ CD25 T cells.56 During pregnancy, an ‘involution’ of thymus is also known to occur in a number of species including rat, mouse, and human.57–59 However, contrary to what it happens during aging, there are strong evidences that the so-called ‘involuted’ maternal thymus is active during pregnancy despite suffering several morphological changes.60,61 It is very likely that the so-called involution is just a remodeling process necessary to achieve tolerance. This ‘involution’ occurs though in both allogenic and syngeneic pregnancy and was also reported in pseudopregnant animals62 and can last until after lactation when the regeneration of thymus takes place.63 In mice, this phenomenon was found to be dependent on the progesterone receptors64 and depending on the mating combination, can be prevented by immunizing females to the paternal strain antigens before mating.65 A study from Zoller and colleagues66 shows that this phenomenon occurs in the absence of detectable apoptosis and is followed by a loss of early T cell progenitors in the thymus and reduced proliferation of CD4, CD8, DP, and DN thymocyte subsets. Furthermore, their results show an enrichment of CD4+ CD25+ Treg in the thymus during late pregnancy suggesting the involvement of nTreg in this stage of pregnancy.66 Consistently, we observed a relative augmentation of Treg in thymus from normal pregnant mice compared to virgin ones, which was not detected in animals undergoing spontaneous abortion,23 contrasting with data from Zhao and colleagues26 that show no significant changes in the frequency of CD4+ CD25+ T cells in the thymus in both syngeneic and allogeneic mating combinations. Similar to what is observed in transplantation models where the lymph nodes draining the grafts show an enlargement shortly after transplantation,67 there are reports demonstrating the increased weight of the uterine draining lymph nodes in pregnant animals in comparison with pseudopregnant or non-pregnant animals57 while this could not be observed in mesenteric, axillary or inguinal lymph nodes.62 Work from Chambers and Clarcke shows57 a major increase in the lumbar lymph nodes of syngeneic mated mice. In contrast, studies from Maroni and de Sousa62 showed a higher increase in the draining lymph nodes of allogeneic matings in comparison with syngeneic ones. The contrasting results from both studies could be explained by the different time points studied; since while in the first study, the measurements were done on the 2nd, 4th, and 7th days after mating; in the second one, the measurements were made on virgin animals and on day 6th, 10th, 16th, and 19–20th days of pregnancy. In other study, Beer and Billingham68 observed the enlargement of para-aortic lymph nodes in allogeneic mated rats. Zhao and colleagues26 propose the uterine draining lymph nodes as the main site of Treg conversion as they found an increased frequency of CD4+ CD25+ T cells there in both allogeneic and syngeneic pregnant mice. It can be speculated that during pregnancy iTreg are induced within the peripheral tissues; however, a contribution from nTreg from thymus cannot be discarded.

Adoptive Transfer of Treg in the CBA × DBA abortion-prone model

Using the well-established murine abortion-prone model, we observed that the frequency of Treg in DBA/2J-mated CBA/J females was significantly diminished compared to normal pregnant BALB/c-mated CBA/J females at different time points.23,24,69 By adoptively transferring Treg isolated from thymus and spleen of normal pregnant females into abortion-prone females on days 0–2 of pregnancy, we were able to completely prevent fetal rejection.23,31 As only Treg obtained from pregnant but not from non-pregnant animals were capable to prevent abortion and the males from both combinations share the same major antigen and differ in minor antigens we speculated antigen-specificity after discarding differences in hormonal levels.23 This hypothesis was confirmed after observing that adoptive transfer of Treg obtained from a third party combination was not able to protect the fetus either.31 Antigen specificity and linked immunoprotection from minor antigens were affirmed after observing that vaccination of CBA/J females with BALB/c male antigens before mating with DBA/2J reduced abortion rates and generated Treg, which were in turn able to protect from fetal rejection if isolated and transferred into abortion-prone mice.31,70 The transfer of antigen-specific Treg was able to rescue from fetal rejection by inducing the generation of a transient tolerant milieu at the fetal–maternal interface that was characterized by high levels of TGF-β, HO-1, and Leukemia inhibitory factor (LIF).69 This introduced the concept of Treg generating a local transient tolerance.69 Other groups also employed this model to confirm the protective role of Treg during pregnancy.71–73 The injection of anti-B7-1 and anti-B7-2 mAbs to 4-days pregnant CBA/J females favors the generation of T cells having a regulatory phenotype.71 Anti-B7 treated ‘regulatory’ T cells that were subsequently transferred into abortion-prone females were able to overcome increased fetal rejection.71 The administration of Cyclosporin A (CsA) on day 4.5 of gestation into abortion-prone mice significantly up-regulates the Treg-associated molecule CTLA-4, while down-regulating the levels of costimulatory molecules such as CD80, CD86, and CD28 at the fetal–maternal interface.72 In addition, treatment with CsA induced enhanced growth and reduced cell apoptosis of murine trophoblasts.72 Importantly, CsA injection further reduced the abortion rate of abortion-prone females significantly.72 Miranda and colleagues73 further showed that syngeneic DC administration into DBA/2J-mated CBA/J females diminished the abortion rate that was associated with an increase in the number of CD8 and γδ cells. Additionally, they observed an up-regulation of TGF-β1 and progesterone induced blocking factor expression.73 We have being working with this transfer model in the last years, and our data on Treg using this particular model may be of importance for clinically relevant situations of recurrent abortions.

Migration of regulatory T cells into the fetal–maternal interface

Although the presence of Treg has been proved at the fetal–maternal interface already very early in pregnancy, the mechanisms of Treg migration are still not identified. Knowing that IL-2 that is essential for Treg survival and proliferation in vivo is not available at the fetal–maternal interface in mice and humans,74–76 we hypothesized that Treg need to migrate to the decidua after generation or expansion in draining lymphoid tissue. For migration studies of Treg into the fetal–maternal interface, we first focused on chemokines, known to be potent attractors of lymphocytes from the periphery into tissues. Treg express several chemokine receptors whose ligands are expressed at the fetal–maternal interface. Among these, CCL22 has been detected in cardiac allografts, and migration of Treg in the same graft has been shown to be CCR4 dependent.77 As CCL22 is one of the chemokines produced at the fetal–maternal interface,78 it is tempting to speculate that trafficking of CCR4+ Treg into the uterus is mediated by CCL22. Other chemokine ligands such as CX3CL1,79,80 CCL3, CCL4, and CCL5,81–83 whose receptors are also expressed by Treg84 during pregnancy might further contribute to chemokine-mediated migration to the decidua. Furthermore, other immune cells that become activated at the fetal–maternal interface produce chemokines to attract Treg to regulate them. In this regard, activated B cells and mature DCs produce high amounts of CCL17 and CCL4,85–87 and macrophages and activated T cells produce CCL1.88 This might attract Treg specifically expressing CCR4 and CCR8.89,90 Using in vitro migration assays, we were able to show that Treg are attracted by some of these ligands further supporting the assumption that chemokines might be involved in Treg migration into the fetal–maternal interface (Teles et al., unpublished data). Beside, chemokine-mediated trafficking integrins like CD62L seem to play an important role in Treg migration. Ochando and colleagues91 showed that administration of neutralizing CD62L-specific antibody blocks expansion of Treg in draining lymph nodes and results in allograft rejection which suggests that CD62L is essential for murine Treg to traffic in draining lymph nodes. Moreover, CD62L expression by Treg is associated with high immunosuppressive activity e.g., suppressing autoimmune diabetes92 or after bone marrow transplantation.93 Thus, integrins are also interesting molecules for studying Treg migration during pregnancy. However, the blockage of P- and E-selectin in the CBA/J × DBA/2J model lead to improvement of pregnancy outcome by avoiding the extravasation of Th1 cells.94

The presence of Treg in the periphery and also in the decidua as well as their expansion during human pregnancy has been widely demonstrated. Tilburgs elegantly12 showed that a preferential recruitment of fetus-specific regulatory T cells from maternal peripheral blood to the fetal–maternal interface takes place. The mechanisms involved in the active migration of Treg from the periphery to the decidua were not clear. In this context, we focussed on the ability of the pregnancy hormone human chorionic gonadotropin (hCG) to attract Treg to the fetal–maternal interface. We based our hypothesis on the fact that this hormone is expressed at the fetal–maternal interface at very early pregnancy stages and on evidences that support a chemotaxis function for this hormone. Having confirmed the presence of LH/CG receptors on the surface of Treg, we went on using migration assays to show that both hCG-producing JEG-3 cells (choriocarcinoma cell line) and first trimester trophoblasts efficiently attract Treg.95 In contrast, non-hCG-producing cells like keratinocytes (Hacat cells) were not able to attract Treg. Down-regulation of hCG production after siRNA intervention diminished the ability of trophoblasts to actively attract Treg.95 Our observations on hCG-dependent Treg migration were confirmed after transfecting non-hCG producing colon carcinoma cells with hCG vectors, that leads to effective Treg migration to the cells.95 Our data clearly show that hCG is one main attractor of Treg into the fetal–maternal interface.

Mechanisms of action of regulatory T cells during pregnancy

Treg function during pregnancy is being extensively investigated. The exact mechanisms as to how Treg mediate fetal protection are unclear. They probably act by a variety of mechanisms, which would all work co-ordinately to ensure fetal protection. We have shown that in mice, Treg can protect the allogeneic fetus by creating a ‘tolerant’ privileged microenvironment at the fetal–maternal interface characterized by high expression of HO-1, TGF-β, and LIF.69 Blockade of HO-1 by means of Zn-PP after Treg transfer in abortion-prone mice abrogated the Treg protective effect that further supports a function of HO-1 in Treg suppression (Wafula et al., in revision).

It has been proposed that Treg in general mediate their protective function either directly by cell-cell contact or via secretion of immunosuppressive cytokines. Suppression of activated effector T cells via cell–cell contact has been shown to be mediated through inhibiting molecules like PD-1 or CTLA-4 both expressed by Treg. We observed that blockage of PD-1 led to an abolishment of Treg protection after transfer of Treg in the CBA/J × DBA/2J abortion-prone model.96 This was not the case for the blockage of CTLA-4.96 In contrast to our data, it has been reported that anti-CTLA-4 antibody treatment was able to inhibit Treg function in vivo97 and that decidual CD4+ CD25high Treg from normal human pregnancies express CTLA-4 on their surface. In miscarriages, the number of CTLA-4 expressing Treg is reduced to non-pregnant levels suggesting that decidual CD4+ CD25high Treg expressing CTLA-4 may prevent fetal rejection.14 One mechanism by which CTLA-4 mediates Treg suppression involves the induction of the tryptophan-catabolizing enzyme IDO on DC or macrophages.98 The up-regulated IDO enzyme depletes tryptophan at the fetal–maternal interface and thereby prevents T-cell and NK cell activation.99–101

Immune regulation of Treg via PD-1 is supported by an experiment in which PDL-1 knockout mice showed increased abortion rates in allogeneic but not syngeneic matings.102 In humans, PDL-1 is expressed by trophoblasts throughout pregnancy.103 Therefore, trophoblasts may regulate Treg function through the PD-1/PDL-1 interaction and thereby protect the fetus from maternal T-cell attack.

Recently, Garin and colleagues104 showed that Treg selectively up-regulate the immune regulatory molecule galectin-1. Galectin-1 has been shown to regulate T-cell activation and cytokine production.105,106 In addition, galectin-1 induces cell growth inhibition and promotes apoptosis of activated T cells.104,105,107 Experiments using galectin-1 homozygous null mutant mice showed an reduced regulatory activity in Treg and blockage of galectin-1 binding diminished the inhibitory effects of human and mouse Treg.108 Furthermore, galectin-1 may favor the expansion of Treg.109 These findings suggest that Treg expressing galectin-1 may support fetal acceptance by maternal immune cells. A recent study irrefutably showed that gal-1-exposed DCs promote IL-10-mediated tolerance,110 which is of importance in pregnancy.

Further studies suggest a direct interaction between Treg and DCs that renders the DCs into a tolerogenic phenotype111 that can in turn induce generation of Treg.112 In this regard, we found that HO-1 is important in maintaining maternal DCs in an immature state which then contribute to the expansion of the peripheral Treg population (Wafula et al., in revision).

Treg have been also shown to influence B cells and Mast cells (MCs). Lim and colleagues113 showed a direct suppression of B cell Ig production by Treg while Gari and colleagues114 demonstrated that Treg directly inhibit the FcεRI-dependent degranulation of MCs through OX40-OX40L interactions. In addition, we observed that fetal rescue by means of antigen-specific Treg was associated with more MCs as well as with enhanced expression of MC-related molecules (Tph-1, Mcpt-1 and Mcpt-5) at the fetal–maternal interface. Treg treatment was further associated with an increase in the levels of well-known MC growth factors mSCF and IL-3, while IL-9 remained unaltered. Anti-IL-10 treatment abrogated the protective effect of Treg and down-regulated the levels of Mcpt-1, highlighting a possible function of IL-10 as MC regulator at the fetal–maternal interface. Our results indicate that MCs and their associated molecules might contribute to Treg-induced tolerance at the fetal–maternal interface (unpublished observations).

These data suggest that Treg regulate immune responses at the fetal–maternal interface by altering the function of several immune cell subtypes.

In regard to immunosuppressive cytokines, Treg have been shown to secrete IL-10 and TGF-β and thereby suppress the effector functions of activated leukocytes.115,116 We showed that the blockage of IL-10 but not TGF-β after adoptive transfer of Treg into abortion-prone mice was able to abrogate the Treg protective effect31 positioning IL-10 as one of the molecules by which Treg mediate their protective function during pregnancy. Our assumption is further confirmed by a study which showed that the relative abundance in endometrial tissue of TGFbeta1, TGFbeta2, TGFbeta3 mRNAs was not changed in infertile women, while expression of Foxp3 mRNA was reduced approximately twofold in this tissue.13

IL-10 is up-regulated during human pregnancy117, and its intrauterine production was shown to be reduced in women suffering from recurrent spontaneous abortions.118 However, it has to be further investigated to which extend the different subsets of Treg expressing high amount of either IL-10 (Tr1) or TGF-β (Th3) are involved in fetal protection.

Therapeutic potential

Guerin and collaborators119 have recently systematically reviewed data on Treg in the ovary, testes, uterus, and gestational tissues in pregnancy and put special attention to their function in infertility, miscarriage, and pathologies of pregnancy. As these authors also propose, an attempting approach would be to transfer antigen-specific Treg before implantation as it works in murine models.22,23,31,71,120 Interestingly, the vaccination with paternal cells, a widely applied protocol in cases of recurrent spontaneous abortions,121 is associated with increased Treg levels.30,122 This shows that the augmentation of Treg around implantation may be a good strategy to improve pregnancy outcome. Additionally, vaginal application of TGF-β augments the number of Treg in the periphery and reduces the abortion rate in the CBA/J × DBA/2J model.123 Application of substances that specifically expand Treg, like the CD28 super agonist, seemed to be a very promising therapy.124–129 However, the initial hope to use therapies that expand Treg was hampered by the data provided by Suntharalingam and colleagues130 who reported that the application of the humanized monoclonal antibody to CD28, TGN1412, which should have expanded the Treg population, to young healthy volunteers ended in a medical disaster. The patients first showed a systemic inflammatory response accompanied by headache, myalgias, nausea, diarrhea, erythema, vasodilatation, and hypotension and then became critically ill, with lung injury, renal failure, and disseminated intravascular coagulation. Severe and unexpected depletion of lymphocytes and monocytes occurred within 24 hr after infusion. This study showed that the results obtained in rodents, which encouraged the use of this substance in humans, are not always transferable to the human situation. However, the isolation and ex vivo expansion of Treg have been extensively studied, and defined protocols are well established. It remains to be investigated whether the isolation and expansion of Treg in the presence of paternal antigens represent a useful tool for pregnancy complications, especially recurrent spontaneous abortion.

Conclusions

Regulatory T cells emerge in the last 6 years as key players in allowing fetal survival within the maternal uterus. They were shown to be a unique subpopulation of T cells expanding during human and murine pregnancy. Paternal antigens are likely to contribute to the pregnancy-associated increase of Treg, although a hormonal contribution should not be discarded. The importance of Treg for a normal pregnancy situation was proven by studies showing that their absence impairs murine pregnancy while the adoptive transfer of Treg prevents fetal rejection. In humans, pregnancy pathologies are associated with lower Treg frequencies while therapies that improve pregnancy outcome are able to boost their number. Functional studies have shown that Treg can regulate immune cell responses directly at the fetal–maternal interface. It has been demonstrated that Treg create a tolerant microenvironment both by interacting with other immune cells like DCs and NK cells and by inducing the expression of immune regulatory molecules such as TGF-β, LIF or HO-1 directly at the interface between fetus and mother. Generation, expansion, and migration pathways of Treg during pregnancy are illustrated in Fig. 1. Although substantial effort was put during the last years in developing experimental strategies to expand Treg in vitro and in vivo, there are still much to do before using these cells as a therapeutical tool to combat pregnancy complications. Additionally, much more experimental work needs to be done that will provide more insights in the mechanisms as to how Treg expand and mediate fetal protection. This will significantly contribute to the knowledge of how fetal tolerance is achieved and will eventually further help patients to become pregnant and give birth to healthy, term babies.

Figure 1.

 Possible mechanisms of origin, expansion, migration and function of Treg during pregnancy.

Ancillary