T-cell Tolerance to the Developing Equine Conceptus

Authors


Author’s address (for correspondence): DF Antczak, Baker Institute for Animal Health, College of Veterinary Medicine, Cornell University, Ithaca, NY 14853, USA. E-mail: dfa1@cornell.edu

Contents

One of the most intriguing and dramatic examples of immunological tolerance is displayed by the mammalian foetal–placental unit, which thrives as a semi-allograft in the mother’s uterus during pregnancy. The success of the so-called foetal allograft stands in stark contrast to the failure of most tissue and organ grafts to survive without genetic matching of donor and recipient or drastic immunosuppression of the recipient’s immune system. Experiments conducted over the past 60 years have revealed multiple mechanisms that enable the conceptus to avoid immunological detection or destruction. Many of these mechanisms are directed towards evading immune-mediated damage by maternal T lymphocytes, and they can be grouped into three classes: (i) downregulation of major histocompatibility complex (MHC) gene expression in placental trophoblast cells; (ii) local and systemic alterations in maternal immune reactivity; and (iii) innate defence mechanisms of the trophoblast cells that comprise the barrier between foetal and maternal tissues. The redundancy in these protective mechanisms helps ensure the transmission of life from generation to generation and provides a rich field of study of ways in which functional immunological tolerance can be manifest. The variation in placental forms and function among mammalian species present opportunities to discover and understand novel tolerogenic mechanisms that may have broad application in biology, medicine and animal husbandry. This review focuses on the evidence obtained from studies of pregnancy in the mare that support the case for selective T-cell tolerance to the mammalian conceptus.

The Scope of the Immunology of Pregnancy

The field of pregnancy immunology includes immunological control of fertility, autoimmune causes of infertility, infectious diseases of the reproductive tract, mucosal immunity of genital tissues, the development of the foetal immune system and passive transfer of immunity from mother to foetus. Most relevant to this review is the additional issue of tissue incompatibility between mother and foetus. In this case, ‘incompatibility’ has a strict immunological meaning, referring to the differences between cells, tissues or organs determined by histocompatibility genes. The histocompatibility genes encode polymorphic molecules called alloantigens, which stimulate the immunological responses known as transplantation reactions that result in graft rejection (Klein 1986). Only a small subset of the polymorphic genes of mammals encodes proteins that stimulate transplantation reactions. These fall into three main categories.

  • 1 Most important are the major histocompatibility complex (MHC) class I and class II genes, which code for genetically linked, highly polymorphic, cell surface molecules that play a major role in transplantation and in governing immune responses to conventional antigens (Klein 1986; Murphy 2012). The MHC class I molecules are expressed on virtually all nucleated cells, while the MHC class II molecules have a limited expression pattern limited largely to antigen-presenting cells such as dendritic cells and B lymphocytes.
  • 2 Second, the so-called minor histocompatibility antigens are encoded by polymorphic genes located throughout the genome. The protein products of these genes are recognized by allogeneic T cells as peptide fragments held in the binding groove of MHC class I molecules (Simpson et al. 2002). Antibodies are not produced against minor histocompatibility antigens, and these antigens can be tissue-specific. Individually, minor histocompatibility antigens have only a small transplantation effect, but in aggregate multiple minor histocompatibility, differences can evoke strong graft rejection (Simpson et al. 1989). Maternal sensitization to foetal minor histocompatibility antigens has been demonstrated in humans (Verdijk et al. 2004) and in mice (James et al. 2003).
  • 3 Finally, the well-known genetic differences in blood groups can result in antibody production and disease after transfusion between donors and recipients that are not matched, and in haemolytic disease of the newborn in humans (Moise 2002) and horses (Boyle et al. 2005).

The remaining type of molecule that could stimulate a maternal immune rejection reaction is one that is expressed in a tissue-specific manner in the placenta. This issue raises a paradox of mammalian pregnancy. The mammalian placenta arose late in evolution (Mess and Carter 2007), when the gene set of vertebrates was largely complete. Thus, to carry out many of its functions, such as removal of foetal waste and delivery of nutrients and oxygen, the placenta uses genes and proteins also in use in other tissues and organs. Indeed, in early pregnancy, there is evidence for a disproportionate use of ancient conserved genes in embryo and placental development (Knox and Baker 2008). However, the placentae of different species have also been shown to express a variety of placenta-specific and species-specific genes (Knox and Baker 2008; Wildman 2011). In humans and rodents, these genes are primarily recently duplicated genes, such as the cathepsins, prolactin-like genes and pregnancy-specific glycoproteins that are expressed in late pregnancy (Knox and Baker 2008). Despite this evidence for the expression of placenta-specific genes to which the mother should not be tolerant, very few antigenic molecules specific to trophoblast have been identified (Anderson et al. 1987). The reason for the lack of antigenicity of these placenta-specific molecules remains an enigma.

The principle of self-tolerance dictates that individuals develop tolerance to their so-called self-antigens (Zouali 2001; Murphy 2012), but mammalian pregnancy presents a unique challenge with the close approximation of placental tissues to maternal immune cells. This review addresses the question of how the foetus escapes destruction by the maternal immune system during pregnancy, with a focus on mechanisms that result in functional T-cell tolerance to the conceptus. Because T cells recognize foreign molecules only in the context of MHC class I and class II molecules expressed on cell surfaces (Murphy 2012), attention has focused on how the conceptus manipulates the expression of MHC gene products on the placental cells that come into contact with maternal immune cells.

Immunological Tolerance in Pregnancy

The concept of immunological tolerance is understood operationally by virtually all biologists, but tolerance is a complicated state that can be achieved through many pathways and mechanisms, including low-dose and high-dose tolerance, clonal deletion and active suppression (Zouali 2001). Immunological tolerance is easy to recognize, but hard to define precisely, and thus it is important to be as specific as possible about the type of tolerance under consideration. In the context of mammalian pregnancy, maternal tolerance of the conceptus means simply that the mother ‘tolerates’ the presence of her antigenically foreign foetus. Determining how this is achieved has been a continuing saga that has extended since the problem was first defined in 1953 by the late Sir Peter Medawar. For example, in a number of species, there is clear evidence for antigenicity of the foetal–placental unit and for maternal immunological recognition or ‘awareness’ during pregnancy (Antczak 1989). How this immunological recognition relates to maternal tolerance of the foetus remains a key unanswered question.

Downregulation of MHC Gene Expression

Perhaps the single most important mechanism preventing maternal immune recognition of the foetal–placental unit is the control of expression of the membrane-bound MHC molecules on trophoblast cells. In all species examined thus far, two observations are most important. First, most trophoblast populations downregulate the expression of both MHC class I and class II molecules. Second, in many species, a population of trophoblast cells expresses some form of MHC class I molecule during gestation (Table 1). Both strategies, the expression of MHC molecules and their repression, have important functional consequences.

Table 1.   Expression of major histocompatibility complex (MHC) molecules on trophoblast cells of various species
SpeciesMHC molecule/locusTrophoblast populationReferences
AllMHC class II, all lociAll – no expression detectedHoltz et al. (2003)
HumanMHC class I HLA-C, HLA-E, HLA-G lociExtravillous trophoblastHunt et al. (2000), Ishitani et al. (2003), Moffett and Loke (2006), Apps et al. (2009)
MouseMHC class I H2-KTrophoblast giant cellsMadeja et al. (2011)
CowMHC class I, classical and non-classical lociSeveral trophoblast populations, expression varies with stage of gestationLow et al. (1990), Davies et al. (2000), Bainbridge et al. (2001), Davies et al. (2006)
HorseMHC class I multiple classical and non-classical lociChorionic girdle and early endometrial cupsDonaldson et al. (1988, 1990), Bacon et al. (2002)

Trophoblast cells completely suppress the expression of MHC class II molecules and genes (Holtz et al. 2003). MHC class II molecules are critically important in the activation of CD4+ helper T lymphocytes (Th cells), and therefore trophoblast cells should be deficient in this aspect of direct alloantigen presentation. MHC class II genes are downregulated via transcriptional control mechanisms (Holtz et al. 2003); thus, no mRNA for MHC class II genes can be detected in trophoblast cells.

For the MHC class I molecules, the situation is more complex. In all species that have been studied (human, mouse, rat, cow, horse, pig), the majority of the trophoblast cells fail to express MHC class I molecules (Billington 2003), which are the targets for allo-antibodies and CD8+ cytotoxic T lymphocytes (CTL) (Klein 1986; Murphy 2012). The MHC class I-negative trophoblasts include the human villous cytotrophoblast and syncytiotrophoblast, and the allantochorion trophoblast of the horse (Noronha and Antczak 2010). However, a minor subpopulation of trophoblast cells does express MHC class I proteins (Table 1).

The lack of MHC expression represents a strategy of evasion of recognition by the maternal immune system. Observations in several species support this strategy – few pregnant females make alloantibodies to paternal MHC antigens (Doughty and Gelsthorpe 1976; Newman and Hines 1980; Bell and Billington 1981, 1983; Davies et al. 2000). Direct evidence for maternal T-cell recognition of paternally inherited foetal MHC antigens has been even more difficult to detect (Moffett-King 2002). Thus, the principal result of repression of expression of MHC antigens in trophoblast is that the maternal immune system by and large ignores the potential antigenic challenge of pregnancy. Lack of expression of MHC class I and class II molecules makes these trophoblast cells poorly antigenic as a transplant. In particular, they would be unable to be recognized by effector T cells.

In what seems to be a counter intuitive evolutionary strategy, in several species, a minor population of trophoblast cells do express MHC class I molecules, and in some species, these are the cells that are most invasive (Table 1). MHC class I molecules are expressed on human extravillous trophoblasts (Apps et al. 2009), mouse trophoblast giant cells (Madeja et al. 2011) and equine chorionic girdle and early endometrial cup trophoblast cells (Donaldson et al. 1990, 1992). These cells come into closest contact with maternal immune cells, and thus are most likely to be recognized as foreign and destroyed. The MHC class I loci expressed in the various species are different and not homologous. In humans, the invasive extravillous trophoblast cells express the relatively non-polymorphic MHC class Ib antigens HLA-E, and HLA-G, but also HLA-C antigens (Ishitani et al. 2003; Moffett and Loke 2006; Apps et al. 2009). The HLA-G gene is expressed as a number of different protein isoforms, including cell-bound and soluble variants (Hunt et al. 2000). The mouse expresses classical polymorphic MHC class I molecules, but primarily those from the H2-K locus (Madeja et al. 2011). Both classical polymorphic and non-classical, pauci-polymorphic MHC class I molecules are expressed on equine (Bacon et al. 2002) and bovine (Low et al. 1990; Davies et al. 2000, 2006) trophoblast cells.

The possible advantage of selective expression of MHC class I molecules by trophoblast cell subpopulations is difficult to understand. One negative result is the potential danger of maternal immunological recognition of foreign foetal (paternal) MHC antigens. It is not clear why the foetus would choose to advertise its presence to the maternal immune system so blatantly by expressing the highly antigenic MHC class I antigens, particularly in the light of the evidence that most trophoblast cells do not express MHC antigens. Nature has solved the problem of how to repress the expression of MHC antigens on trophoblast. Why make the effort to re-express those molecules on invasive trophoblast cells? In humans, it is thought that the non-polymorphic HLA-C, HLA-E and HLA-G molecules might provide protection for trophoblast cells from destruction by natural killer (NK) cells that are abundant in the human uterus during pregnancy (Moffett and Loke 2006). Because human trophoblast does not express the highly antigenic HLA-A and HLA-B antigens, it is thought that maternal sensitization in women is owing to migration of foetal leucocytes to the mother (Iverson et al. 1981). This would account for the production of maternal antibodies to both MHC class I and class II antigens in human pregnancy.

In summary, this first line of defence against maternal immune responses is based upon the strategy of evasion of detection. If complete, this strategy should allow the maternal immune system to remain intact during pregnancy so that it can respond to threats from infectious diseases. The selective expression of MHC class I molecules by trophoblast subpopulations may have been the driving force for the evolution of other immuno-protective mechanisms described later. When maternal immune recognition of trophoblast occurs, the target antigens are almost always MHC class I molecules.

Local and Systemic Alterations in Maternal Immune Reactivity

Pregnancy-associated changes in maternal T-cell responsiveness have been described both systemically (i.e. in circulating lymphocytes) and locally at the placental–uterine interface. A number of experimental approaches have identified both antigen-specific and non-specific tolerogenic mechanisms that decrease systemic T-cell reactivity towards the foetus (Table 2). Antigen-specific mechanisms have the great advantage of leaving the remainder of the mother’s immune system intact, allowing her to defend herself against infection during pregnancy. The antigen non-specific mechanisms do not require information about the specific histocompatibility challenge of the foetus, and therefore they complement evasion mechanisms like the downregulation of MHC genes. It is not clear whether and how the antigen-specific and non-specific effects are related.

Table 2.   Evidence for systemic decreases in maternal T-cell reactivity during pregnancy
SpeciesSystem/evidenceReferences
MouseTransgenic CD8+ T cells/Induction of tolerance in alloantigen-specific T cellsTafuri et al. (1995)
MouseH-Y transgenic T cells/Clonal deletion and induction of tolerance in foetal antigen-specific T cellsJiang and Vacchio (1998), Vacchio and Jiang (1999)
MouseLeishmania infection/Shift from Th1 to Th2 cytokine response profile/increase in susceptibilityKrishnan et al. (1996a,b)
MouseListeria infection/Reduction in genetic resistance to infection by intragastric inoculationPoulsen et al. (2011)
HumanAutoimmune disease/Shift from Th1 to Th2 cytokine profile and remission of disease symptoms during pregnancyMattsson et al. (1991), Beagley and Gockel (2003)
Mouse, HumanNormal pregnancy/Systemic expansion of regulatory T cells, suppression of allogeneic and autoimmune responsesAluvihare et al. (2004), Zenclussen et al. (2006), Tilburgs et al. (2008)
HorseNormal pregnancy/Decrease in CTL reactivityBaker et al. (1999), Noronha and Antczak, 2012

Antigen-specific tolerogenic effects have been demonstrated in transgenic mice engineered to express a single type of T-cell receptor on all T cells (Tafuri et al. 1995; Vacchio and Jiang 1999). In these studies, T cell numbers decreased during pregnancy as a result of expression of the paternal antigens recognized by the transgenic T cells. These states of T-cell tolerance were reversible; they could be measured only during pregnancy. In contrast, there is also evidence for specific T-cell sensitization, and not tolerance, to paternal histocompatibility antigens in humans (Verdijk et al. 2004) and mice (James et al. 2003). These studies were associated with multiparity and described long-lasting effects that in women were still detectable years after the last pregnancy. It is not yet possible to reconcile this evidence for transient tolerance and long-term sensitization.

The evidence for non-specific T-cell tolerance in pregnancy comes from observations and experimental studies. For many years, physicians have recognized that during pregnancy many women experience a temporary remission of symptoms of certain autoimmune diseases where pro-inflammatory cytokines are thought to be responsible for pathology (Mattsson et al. 1991; Beagley and Gockel 2003). These observations fit well with the hypothesis that the state of pregnancy induces a shift in immune system cytokines away from proinflammatory molecules such as TNF alpha and Interferon gamma and towards cytokines that promote antibody responses, such as Interleukin 4 (IL4) and IL10 (Krishnan et al. 1996a,b).

Superimposed on these cytokine profile changes is the evidence for increases in so-called regulatory T cells (Tregs) during pregnancy in mice (Aluvihare et al. 2004) and humans (Saito et al. 2005). Tregs are a CD4+ CD25+ T-cell subpopulation that also expresses the FOXP3 transcription factor (Ramsdell 2003; Wood and Sakaguchi 2003; Nagler-Anderson et al. 2004; Waldmann et al. 2008). Increases in regulatory T cells have been detected both in the circulation (Tilburgs et al. 2008) and locally at the foetal–maternal interface (Zenclussen et al. 2006). Elimination of Tregs resulted in pregnancy loss that was specific for allogeneic conceptuses, but the activation of the Tregs was also observed in syngeneic pregnancy (Aluvihare et al. 2004).

Another potent modulator of maternal immune reactivity in the pregnant uterus is progesterone. The immunosuppressive actions of this hormone have been demonstrated very dramatically in allograft (Padua et al. 2005) and xenograft experiments in sheep (Majewski and Hansen 2002), where prolonged graft survival is associated with exogenous administration of progesterone. This immunosuppressive effect is clearly not induced or expressed in an antigen-specific fashion.

The evidence for non-specific reductions in maternal T-cell-mediated immunity during pregnancy raises the important question of the effect of such immunosupression on the ability of the mother to fight off infectious diseases during pregnancy. Indeed, there is evidence that pregnancy leads to increases in susceptibility to infections where T-cell responses are most important for protection (Poulsen et al. 2011). Mammalian pregnancy balances the need to protect the foetus from immune destruction while maintaining immune defences against the wide variety of pathogens. This is no doubt a compromised situation.

Innate Defence Mechanisms of Trophoblast

The third layer of protection of the foetus from maternal T cells consists of local mechanisms innate to the trophoblast cells themselves (Table 3). These include cell surface molecules that can induce apoptosis in T cells (Fas ligand and Galectin-1), and secreted products that have negative effects on T cells (Indolemine 2,3 deoxygenase or IDO). The local mechanisms would act on the efferent arm of the immune system after immune responses have been generated. The immuno-protective mechanisms described earlier in this review usually result in the mother ignoring the foetus or being suppressed in her ability to generate anti-foetal immune responses, leading to a lack of induction of immunity. The innate mechanisms of trophoblast would only take effect if the conceptus is antigenic, and if the mother mounts a deleterious immune response against it.

Table 3.   Innate defence mechanisms that can protect trophoblast from destruction by maternal T cells
SpeciesImmune effectorProtective mechanismReference
Human, MouseCytotoxic T cells that express FasFas Ligand (induces apoptosis in the immune effector cell)Hunt et al. (1997)
Human, MouseActivated T cellsIDO (depletes tryptophan required by T cells)Mellor and Munn (2000)
Human, MouseActivated T cellsGalectin-1 (induces apoptosis of T cells)Blois et al. (2007), Than et al. (2009)
MouseActivated T cellsProgrammed Death Ligand 1 (PDL1) (maintenance of tolerance by unexplained mechanisms)Guleria et al. (2005), Habicht et al. (2007), D’Addio et al. (2011)

Experiments in mice in which local defence mechanisms have been knocked-out by gene inactivation or neutralized by antibody or drug treatment often result in the loss of allogeneic, but not syngeneic pregnancies (e.g. Mellor and Munn 2000). This supports the case for a protective effect of these molecules. These results strongly suggest that the conceptus can be antigenic and that the mother’s immune system is not completely tolerant of the foetus. However, the situation is complicated. In most of the studies in mice in which pregnancy loss has been documented, there is no evidence for specific immune responses against the conceptus – usually the only experimental read-out is pregnancy loss, and in mice, this sometimes means reduced litter size, an outcome that is difficult to interpret. Furthermore, there is no evidence that activated T cells are responsible for the pregnancy loss. In some cases, the loss has been attributed to the activation of maternal complement through the alternate pathway of the innate immune system (Caucheteux et al. 2003), or NK cells (Erlebacher et al. 2002), but seldom (ever?) to classical acquired immune responses with the demonstration of the cardinal features of immunological specificity and/or memory (Dosiou and Giudice 2005; Kwak-Kim et al. 2009). Thus, the principal tenets of the acquired immune system have not been demonstrated.

Is There Evidence for Maternal Immune Destruction of the Conceptus?

One of the most intractable issues in pregnancy immunology is the question of susceptibility of trophoblast cells to destruction by the maternal immune system (Smith 1983; Billington 1993; Erlebacher et al. 2002). This question is critical in the context of unexplained pregnancy loss in women, where immune mechanisms have often been postulated, but seldom proven (Clark 2003; Salmon 2004; Niederkorn 2006). This has lead to widespread immunological testing in cases of infertility that may have little value (Kallen and Arici 2003); and the use of cell (Porter et al. 2006) or serum-based immunotherapies (Stephenson et al. 2010) that have not been shown to have therapeutic efficacy. Unfortunately, as described earlier, experimental model systems have not yet provided clear evidence for antigen-specific T-cell-mediated pregnancy loss.

The Place of the Horse in Pregnancy Immunology

Equine pregnancy has several attributes that have made the horse a particularly useful species for studies of pregnancy immunology. These have been reviewed recently (Noronha and Antczak 2010) and will be summarized briefly here. First, there is overwhelming evidence for specific maternal immunological recognition of the developing equine conceptus, in the form of antibodies to paternal MHC class I antigens inherited by the foetus (Antczak et al. 1982, 1984). Maternal antibody responses are routinely detected by day 60 of gestation in primiparous mares, much earlier than in any other studied species with comparable gestation lengths (Antczak et al. 1984), and the frequency of sensitization in mares carrying MHC incompatible foetuses approaches 100% (Antczak et al. 1982, 1984). The source of the stimulating MHC class I antigens is the invasive trophoblast tissue of the chorionic girdle, which migrates into the endometrium at days 36–38 of gestation (Donaldson et al. 1990). The girdle cells express high levels of MHC class I antigens that can be easily detected using immunohistochemistry (Donaldson et al. 1992) and which are also reflected in abundant levels of mRNA (Bacon et al. 2002). The alloantibody responses are evidence that the humoral arm of the mare’s immune system is intact during pregnancy.

Second, studies of equine pregnancy have also provided support for T-cell recognition of the conceptus. Shortly, after the chorionic girdle cells invade the endometrium to form the equine chorionic gonadotrophin (eCG)-secreting endometrial cup trophoblasts, the maturing cup cells are surrounded by striking accumulations of maternal lymphocytes (Grunig et al. 1995). Most of these lymphocytes carry cell surface markers of equine T cells (e.g. the CD3 pan T-cell molecule) and include separate CD4 (T helper cell marker) and CD8 (cytotoxic T-cell marker) subpopulations (Grunig et al. 1995; de Mestre et al. 2010). The lymphocyte accumulations occur only at the site of invasion of the MHC class I antigen-positive chorionic girdle cells, and not along the endometrium–allantochorion border, where only scattered T cells can be detected (Grunig et al. 1995; de Mestre et al. 2010).

Similar accumulations of lymphocytes were observed in the endometrial cups from experimental MHC compatible equine pregnancies, where no cytotoxic antibody was detectable in maternal serum (Allen et al. 1984). The maternal lymphocytes in MHC compatible pregnancies could be specific for minor histocompatibility antigens expressed by the invading MHC class I chorionic girdle cells (Simpson et al. 1989). Minor histocompatibility antigens have been shown to be antigenic in murine and human pregnancy (James et al. 2003; Verdijk et al. 2004). Alternatively, they might be T cells with regulatory function (Zenclussen et al. 2006; de Mestre et al. 2010). The lifespans of the cup trophoblasts in MHC compatible pregnancies were not extended (Allen et al. 1984), raising the possibility that the lifespan of endometrial cups may be determined by the trophoblast cells themselves, and not limited by a maternal immune response against them (de Mestre et al. 2011).

The maternal T cells appear to be associated with the eventual death of the endometrial cups at day 90–120 of pregnancy, but by that time, the endometrial cup trophoblasts have lost expression of their MHC class I antigens (Donaldson et al. 1992; Maher et al. 1996), and thus the T cells should be unable to recognize the trophoblast cells. NK cells should be able to kill these MHC antigen-negative trophoblasts, and there is recent molecular evidence for NK cells in the endometrial cups (Noronha et al. 2012a). Good antibody markers for equine NK cells have not been available, but the recent production of monoclonal antibodies to equine CD16 (Noronha et al. 2012b) and NKp46 (Noronha et al. 2012c) should facilitate study of equine uterine NK cells. Recent work has also demonstrated an enrichment for markers for regulatory T cells among the endometrial cup lymphocyte populations compared to peripheral blood samples obtained from the same mares on the same day of pregnancy (de Mestre et al. 2010). Thus, although the role of maternal T lymphocytes in the destruction of the endometrial cup trophoblasts is still unclear, the possibility of local regulation of T cells seems likely.

Third, there is evidence for a systemic decrease in maternal T-cell reactivity in the mare. During pregnancy, there is a reduced capacity for maternal T cells to develop into cytotoxic T cells (CTL) after in vitro stimulation with irradiated paternal lymphocytes (Baker et al. 1999). This reduction was observed in both horse and donkey pregnancy, and it was detected at every stage of gestation tested (Baker et al. 1999). Recent studies have shown that the decreased T-cell reactivity is not antigen-specific and that it occurs in MHC compatible pregnancy, where there is no MHC difference between mother and foetus (Noronha and Antczak 2012). Thus, it is likely that this type of T-cell tolerance is induced by the state of pregnancy itself, and not by antigenic differences between mother and foetus.

Fourth, and finally, there is very strong evidence that equine trophoblast cells have innate mechanisms that protect them from immune destruction. Our laboratory has developed a novel method for trophoblast transplantation to ectopic sites in non-pregnant recipients (Adams and Antczak 2001). This allows detailed scrutiny of in vivo interactions between host immune cells and trophoblast outside of the uterus, where the local uterine environment and the hormonal milieu of pregnancy are eliminated as contributing factors to trophoblast survival. In this system invasive, MHC class I-positive trophoblast cells are transplanted to the vulvar mucosa, where the cells evoke antibody and T-cells responses in the recipients in very similar fashion to the situation in normal pregnancy (Adams and Antczak 2001). The transplanted trophoblasts survive for up to 105 days and secrete sufficient eCG into the circulation to cause ovulations and corpora lutea formation that prevent the graft recipients from coming into oestrus for 3 months (de Mestre et al. 2011). The trophoblast cells can be cultured for short periods before transplantation (de Mestre et al. 2008), and thus may enable gene knockdown studies in the trophoblast to be performed. The survival of transplanted equine trophoblast cells in hostile environments marked by antibody production and T-cell accumulation points to the existence of strong defensive mechanisms that are innate to the trophoblasts themselves. Equine lymphocytes co-cultured with chorionic girdle trophoblast cells, but not foetal fibroblasts, have a reduced capacity for proliferation in response to mitogens (Flaminio and Antczak 2005). This effect could be mediated by a soluble factor released from the trophoblast cells and did not require lymphocyte–trophoblast contact (Flaminio and Antczak 2005). Such an effect might aid in the survival of transplanted horse trophoblasts.

In another approach, our laboratory has used whole genome expression arrays to identify genes with possible immunomodulatory function that are expressed in equine invasive chorionic girdle trophoblast. One of the genes discovered is the cytokine Interleukin 22 (IL-22), which functions in antimicrobial immunity, inflammation and tissue repair at mucosal surfaces (Sonnenberg et al. 2011). Until our study IL-22 had been identified only in lymphoid cells. However, during normal equine pregnancy, IL-22 mRNA was detected at very high levels in the developing chorionic girdle and early endometrial cup trophoblasts, but not in any other foetal or placental tissues studied (Brosnahan et al. 2012). IL-22 may be expressed in the placentas of other mammals, but if the expression is as limited in time as it is in the horse, it may be difficult to demonstrate unless the correct samples can be obtained. The case of IL-22 in equine trophoblast appears to be an example of an immune system gene co-opted by the placenta. Although the function of IL-22 in equine pregnancy is not known, it may involve immunological survival of the conceptus. It is likely that other surprising examples of gene expression in trophoblast will be discovered through the application of whole genome technologies in domesticated species.

Conclusion

Studies of the immunology of pregnancy are challenging because of the large number of independent mechanisms that provide protection to the foetus and placenta. Several critical questions remain to be answered. Among the most important are why some trophoblast cells express MHC class I molecules, and how maternal T cells are regulated during pregnancy. It is tempting to speculate that the limited expression of MHC class I molecules by trophoblast subpopulations and their subsequent recognition by the maternal immune system is beneficial to pregnancy. Otherwise, why would this strategy have evolved independently in so many species? For many years, the expression of polymorphic MHC class I antigens in the horse placenta seemed to be unusual, but recent research suggests otherwise. It now appears that any species with an invasive trophoblast cell type might express MHC class I molecules from one or more classical loci (Table 1).

With respect to maternal T cells, the majority of studies suggest non-specific reductions in activity that may reflect functional tolerance. Because antibody production requires T cell help, it is unlikely that all T-cell functions are diminished during pregnancy. It appears that the cytotoxic lymphocyte pathways are most affected. Further studies of regulatory T cells in pregnancy should be fertile ground for new discoveries.

A convenient way to frame the combined evidence for immune recognition and regulation during pregnancy is through the hypothesis of ‘split immunological tolerance’ (de Mestre et al. 2010). In such a state, the B-cell component of the maternal immune system would be left largely intact, while the most damaging types of T-cell responses would be controlled through a variety of mechanisms, including a final set of defences innate to the trophoblast cells themselves. Studies of the horse may be illuminating in testing this hypothesis of split immunological tolerance to trophoblast, because of natural features of equine reproduction and pregnancy immunology.

Acknowledgements

The author thanks Mr Donald Miller and Ms Rebecca Harman for assistance with manuscript preparation. This work was supported by grants from the US NIH (R01 HD049545) and the Harry M. Zweig Memorial Fund for Equine Research. DFA is an investigator of the Dorothy Russell Havemeyer Foundation, Inc.

Conflicts of Interest

None of the authors have any conflicts of interest to declare.

Ancillary