Almost from the beginning of research in the field, a clear dichotomy was revealed surrounding the role of the conceptus in extending luteal function in primates and domestic ruminants. In primates, the conceptus produces a luteinizing hormone (LH)–like hormone termed chorionic gonadotropin (CG) that acts directly on the corpus luteum (CL) via the blood; an action that was described as luteotropic.1–3 Presence of CG in the blood and urine of primates provides a straightforward mechanism for determining the presence of a viable conceptus in these species and is the basis for many home pregnancy tests.4 In contrast, domestic ruminants (cattle, sheep, goats) produce unique interferons (IFN), closely related to α- and ω-IFN, termed interferon τ (IFN-τ), that do not exhibit luteotropic activity, but rather act locally on the uterus to block luteolysis, an action termed antiluteolytic.1,5,6 Early attempts to identify these substances in the systemic circulation,7–9 urine or cervical mucus10 of ruminants largely failed. There are also species, such as the dog and cat, that do not require a conceptus signal for rescuing CL function.11 In these species, regardless of whether mating establishes a pregnancy, the CL is maintained for a period similar to the length of gestation. Thus, at least during early pregnancy, there is no need for signaling between the uterus and ovary to maintain pregnancy in dogs and cats.
Relative to conceptus effects on luteal lifespan, the antiluteolytic versus luteotrophic hypotheses have weathered years of intense investigation and are routinely taught in the classroom. Evidence supporting this two-pronged view of early conceptus action was so strong that it may have initially diverted focus away from alternative modes of action of these conceptus signals beyond their roles in extending CL function. This is particularly the case related to potential systemic effects of conceptus IFN-τ produced by domestic ruminants, and for potential uterine, and non-luteal effects of primate CG. In this review, we will focus only on those initial conceptus signals (IFN-τ and CG) that are thought responsible for CL rescue and limit our focus to the contribution of ruminant models to understanding the systemic effects of these conceptus signals on circulating immune cell function in primates. Readers desiring information regarding the effects of pregnancy on changes in populations of peripheral or endometrial resident immune cells are directed to recent reviews on this subject in primates,3 ruminants,12 swine13 and horses.14 In addition, there is an excellent recent review on the role of progesterone in altering immune responses during pregnancy.15
Human pregnancy recognition is characterized by production of CG from syncytiotrophoblast cells, beginning approximately 8–10 days after fertilization.3,16 CG is a member of the glycoprotein hormone family that includes LH, follicle stimulating hormone and thyroid stimulating hormone.17 CG arose from a gene duplication event from the LH-β subunit roughly 34–50 million years ago; more than 80 million years after the first appearance of eutherian (i.e., true placental) mammals.18 CG binds to the LH/CG receptor and sustains the CL and progesterone (P4) production until sufficient P4 is produced by the placenta; the highest concentrations of human CG detected in maternal circulation occur during the first trimester of pregnancy. As in other species, humans exhibit significant immunomodulatory adaptations to pregnancy and the changing hormonal milieu is likely a key driving force to these changes in the maternal immune system.19
Forty years after Medawar’s postulates on maternal acceptance of the semiallogeneic conceptus via immunomodulatory mechanisms, Wegmann et al.20 proposed that the immune system shifts to an antibody-based response (Th2) instead of a cell-mediated response (Th1) during pregnancy. The Th1 cytokine profile is associated with greater concentrations of interferon γ (IFN-γ), interleukin-2 (IL-2) and tumor necrosis factor-β (TNF-β). The Th2 cytokine profile is typified by increased levels of IL-4, IL-5, IL-6, IL-10, and IL-13.21,22 There appears to be a delicate balance between Th1 and Th2, with each cytokine profile regulating the other. Disruption of the Th1/Th2 balance has been implicated in miscarriages in a number of species.24 The Th2 cytokine profile can block the activation of Th1 cells, while Th1 cytokines inhibit Th2-cell proliferation. This regulation may be due, in part, to down regulation of nuclear factor kappa-β (NF-κB) in T cells during pregnancy, which biases immune response away from Th1.43 Whether this effect is directly mediated by CG is not clear, as reports show both an activation25 and inhibition26 of NF-κB in monocytes and endometrial stromal cells, respectively. Human CG also exhibits immunomodulatory functions by inducing suppressor T cells27 and has long been known to modulate both B- and T-cell response to mitogen stimulation.28–30 In addition, LH/CG receptors are present on maternal T lymphocytes23 providing for a direct mechanism whereby hCG could alter function of circulating immune cells. During normal pregnancy, there is an elevation of CD25+ CD4+ regulatory T cells (T-reg31), and hCG appears to recruit these cells to the fetal–maternal interface.16,32 Furthermore, CG induced bone marrow–derived, in vitro matured, dendritic cells toward a tolerogenic phenotype characterized by increased IL-10 and indolamine 2,3 dioxygenase production.33
The evidence that P4 shifts the cytokine profile toward Th2 is more compelling.15,34 This action is mediated, in part, through P4-induced production of immunosuppressive molecules including progesterone-induced blocking factor (PIBF115) and glycodelin A,35 among others. Progesterone-induced blocking factor stimulates Th2 cytokine production and can suppress NK cell activity in the uterus and systemically.36 As reviewed by Lea and Sandra,36 P4 induces a number of cytokines in peripheral T cells, including leukemia inhibitory factor, colony stimulating factor-1, IL-4 and IL-5. Together the uterine and systemic effects of P4 paint a fairly consistent picture of a Th2 bias and indirect suppression of uNK cells that promote immunologic recognition of pregnancy and tolerance.
It is increasingly clear, however, that immunomodulation during pregnancy may be more complicated than the Th1–Th2 shift proposed by Wegmann,20,37–39 as evidence by marked activation (as opposed to suppression) of some components of the maternal immune system.40,41 For example, hCG treatment of the baboon uterus upregulates superoxide dismutase 2 and complement component 3, to respond to oxidative stress and enhance phagocytosis, respectively.3,42 In addition, hCG binds to monocytes and increases their trafficking to the endometrium during early pregnancy and increases production of IL-8 via activation of NF-κB.43 From the standpoint of evolution, it would make sense to counter balance the immunosuppressive effects of pregnancy so as not to put the dam at greater risk of infection.44 Clearly, there is evidence that conceptus signals like hCG alter immune cell function in the uterus and peripherally.16,42
Although much work has focused on immunomodulatory mechanisms mediating fetal tolerance and maternal protection, circulating immune cells may play an active role in establishing and maintaining pregnancy.45 Using a luteal cell culture system, Hashii et al.46 showed that peripheral blood mononuclear cells (PBMC) from pregnant women increased P4, IL-4, and IL-10 production. Th2 cytokines, IL-4 and IL-10, stimulated P4 production to concentrations similar to those observed in hCG-treated cultures.46 Nakayama et al.47 provided evidence for a beneficial effect of immune cells on early events of embryo implantation by showing that human peripheral blood leukocytes (PBL) from pregnant mothers enhanced murine embryo spreading and invasion in vitro. In addition, co-culture of PBMC isolated from pregnant women with BeWo-cells (i.e., a trophoblast cell line) enhanced their ability to invade into matrigel compared with PBMC isolated from the secretory phase of the menstrual cycle.48 This effect occurred with the cells isolated on either side of a 0.8 -μm membrane, suggesting the effect was induced by a soluble factor secreted by the PBMC. This factor could, in fact, be CG which was shown to be produced and secreted by PBMC during early pregnancy in women.49 Moreover, Kosaka et al.50 showed PBL promoted attachment of BeWo-cell spheroids to endometrial cells derived from human uteri in the late proliferative and early secretory phases. They suggested that PBL may be able to induce endometrial cells to become ‘receptive’ to embryo implantation. In agreement, Yoshioka et al.51 found that hCG-primed PBMC, in conjunction with freshly prepared PBMC, increased pregnancy rates in women when the PBMC were administered into the uterine lumen 1 day prior to blastocyst transfer. Collectively, these studies highlight how the immune and endocrine system may coordinate events for the establishment and maintenance of pregnancy, and how conceptus signals responsible for rescuing CL function may also play a role in counter-balancing the immunosuppressive effects of progesterone. Surprisingly, there have been relatively few studies examining the effects of hCG on peripheral immune cell gene expression, and no reports of global transcriptional profiling experiments on hCG-stimulated immune cells.
In the early to mid-1990s, several studies revealed potential similarities between CG and the ruminant type I IFN family to which IFN-τ belongs. In those reports, the β-subunit of CG was purported to possess antiviral activity (characteristic of interferons) against HIV.52–54 However, subsequent studies revealed that this activity resulted from contamination of the preparation by lysozyme and RNAses present in the urine of pregnant women from which the CG was purified.55 These results are consistent with those of Gunn et al.56 who were unable to detect antiviral activity in media conditioned by culture with peri-implantation stage human embryos, even if the embryos were producing detectable amounts of CG. However, there is ample evidence that the human placenta produces IFN during gestation that alters local immune cell function.57 For example, interferon-stimulated gene 15 (ISG15) is induced during early pregnancy in the endometrium of the baboon and human.58 ISG15 is an interferon-stimulated ubiquitin homolog that conjugates with cellular and viral proteins to alter their function, half-life and/or distribution in cells.59 ISG15 is expressed in the pregnant endometrium in primates,58 rodents,60 and ruminants.61 However, the conceptus signal(s) inducing these changes in humans has not been defined, but it does not appear to be as a direct effect of hCG production. In addition, ISG15 is increased in PBL in cattle during early pregnancy.62,63 Whether similar increases in ISG15 in peripheral immune cells are induced during human pregnancy is not known.
Nonetheless, hCG clearly alters circulating immune cell function in ways expected to result in immunosuppression. Komorowski et al.64 showed that hCG reduced IL-2 and increased sIL-2 receptor secretion by human PBMC. These two together would result in reduced peripheral T-cell activation in response to paternal alloantigens. In rodents treated with hCG peptides, there was evidence of reduced neutrophil migration to sites of Listeria monocytogenes replication associated with reduced chemokine production.65 These results are consistent with the observations that pregnant humans exhibit increased susceptibility to this pathogen.66 T cells treated with recombinant hCG showed reduced proliferation, decreased IFN-γ secretion and increased IL-10 production.30 Furthermore, hCG reduced the ability of in vitro-matured dendritic cells to stimulate T cells, potentially contributing to peripheral tolerance during pregnancy.67,68 In addition, hCG stimulated PBMC to increase IL-8 production which would support embryo implantation.43 Taken together these studies provide strong support for a systemic, non-luteal action for hCG targeting specific components of the circulating immune system.
With the discovery and characterization of the systemic role for CG from the primate conceptus,1 numerous investigations were launched to determine if similar systemic actions were involved in pregnancy recognition in ruminants.69 These studies identified an early conceptus protein that, when introduced into the uterus in purified form, rescued the CL. This protein, first called ovine trophoblast protein-1 or trophoblastin, and later IFN-τ, blocked development of the uterine luteolytic mechanism through paracrine actions on the uterine endometrium.6 These studies all led to the conclusion that there was little evidence for a systemic effect of the ruminant conceptus on the CL. For example, Godkin et al.7 injected iodinated IFN-τ into the uterus and assayed various tissues, including blood, for radioactivity and found that only very small amounts of label escaped the uterus (<1%). There was no evidence of significant accumulation of labeled IFN-τ in the ovary, nor were they able to demonstrate that IFN-τ could enhance progesterone production by luteal cells in vitro even though luteal membrane preparations specifically bound labeled IFN-τ. Numerous attempts to detect IFN-τ in uterine venous blood by radioimmunoassay were unsuccessful.10 In addition, that report also failed to detect IFN-τ in concentrated urine or cervical mucus collected from pregnant sheep using a western blotting procedure that was sensitive to 1 ng IFN-τ. Furthermore, Lamming et al.9 could not detect IFN-τ in the lymph draining from the pregnant uterus using a sensitive bioassay for antiviral activity. Work by Schalue-Francis et al.8 was the only published report showing low amounts of antiviral activity (characteristic of IFN) in the uterine venous drainage, however, they were not able to detect this activity in jugular blood. It was not clear whether this activity resulted from IFN-τ escaping from the uterus or was the result of an indirect effect of IFN-τ stimulating immune cells trafficking through the uterus to produce a substance(s) with antiviral activity. Taken together, these results were generally interpreted to indicate that the effects of IFN-τ on luteal function were mediated through its paracrine action on the uterine endometrium, which was clearly different than the mechanism of action of hCG.
The antiluteolytic (local) versus luteotrophic (systemic) paradigms for CL rescue have persisted for almost four decades.6 In fact, following the cloning70 and large-scale production of recombinant IFN-τ,71 investigators who previously were unable to consistently improve fertility with systemic rhuIFN-α, undertook studies to determine if exogenous IFN-τ could extend CL function and increase fertility. In light of the previous studies pointing toward a lack of systemic actions of IFN-τ, the hypothesis was not that exogenous IFN-τ could mediate conceptus signaling by actions in the peripheral circulation, but rather if high circulating concentrations of IFN-τ could be achieved to mimic the local antiluteolytic effects in the uterus. These studies were largely unsuccessful in sheep and cattle primarily because of the pronounced pyrogenic effects of exogenous IFN-τ72,73 and further supported the widely held belief that conceptus IFN-τ did not act outside the uterus.
However, more recent evidence has emerged that demonstrates that IFN-τ produced by the ruminant conceptus is also acting systemically.62,63,74–76 This work was the first to show that Type I interferon-stimulated genes that were previously shown to increase in the uterine endometrium in response to IFN-τ were also increased in PBL.63,74 Two of these, the myxovirus resistance or MX proteins (MX1 and MX2), were shown to increase in PBL 48–72 hr after the onset of conceptus elongation and IFN-τ production, with maximal increases occurring between 17 and 19 days after insemination. Abundance of MX proteins remained above concentrations in cyclic ewes out to day 30 after insemination.74 Similarly, studies in cattle62,63,77 showed that these same genes and ISG15 were elevated in PBL collected between days 18 and 21 after insemination. However, while these studies clearly provided evidence of systemic effects of IFN-τ on circulating immune cells in ruminants, it was still not clear whether IFN-τ was directly mediating this effect or whether this was an indirect effect of IFN-τ action on uterine or endometrial immune cells which produced a factor(s) that escaped the uterus.
Definitive evidence that IFN was escaping the uterus was provided by Oliveira et al.75 who demonstrated a 500–1000-fold increase in antiviral activity in the uterine vein compared to the uterine artery or jugular blood of early pregnant ewes. These results provided strong support for the early evidence showing low, but detectable levels of IFN-τ7 and antiviral activity8 in the blood. Work from this same group later demonstrated that the antiviral activity was indeed caused by release of IFN-τ.76 These important studies were the first to definitively demonstrate that IFN-τ had a direct systemic effect, and that this effect could increase CL lifespan.
Interestingly, Tuo et al.78 had previously shown that exogenous IFN-τ had dramatic effects on immune cell recirculation and redistribution in lambs by reducing CD4+, CD5+ and gamma delta + T cells in the peripheral circulation without changing numbers of CD8+ T cells. This effect occurred within 6–12 hr of treatment and peripheral immune cell populations returned to pre-treatment control values by 48 hr. Furthermore, IFN-τ was shown to cause a dose-dependent reduction in lymphocyte proliferation79 and to suppress lymphocyte blastogenesis80in vitro. In contrast, IFN-τ stimulated NK cell activity in sheep PBMC.81 Taken together, these experiments provide evidence that, while ruminants and humans possess different mechanisms for supporting CL function during early pregnancy, there exists the distinct possibility that they may share functions as a result of the fact that they are both present in the peripheral circulation during the very earliest stages of pregnancy recognition signaling and both can apparently bind and alter function of circulating immune cells. Support for this hypothesis is currently limited owing to few of studies examining the effects of either hCG or IFN-τ on circulating immune cell function. However, work carried out in later pregnancy in cattle clearly supports similarities between humans and cattle in alterations in peripheral and endometrial immune cell populations.12 For example, in cattle there was an increase in peripheral cells exhibiting the T regulatory phenotype (CD4+ CD25+) as well as recruitment of these cells to the endometrium. T regulatory cells secrete IL-4 and can induce tolerance to paternal alloantigens and inhibition of T regulatory function is associated with compromised pregnancy.12
We recently conducted a transcriptional profiling experiment to identify genes regulated in PBL by pregnancy and progesterone in cattle82 (Ott and Gifford unpublished). Results from these studies clearly indicated that a large number of known interferon-stimulated genes increase in PBL of early pregnant cows. In addition, some genes not previously thought to be IFN responsive were also increased. For example, chemosensory receptor transporting protein-4 (RTP4) increased in PBL during early pregnancy and, in vitro, in response to IFN-τ.82 We then demonstrated that RTP4 was also expressed in the uterine endometrium, which was surprising because expression of this gene was initially thought to be confined to olfactory neurons. Furthermore, in vitro treatment with IFN-τ increased RTP4 expression by a cell line derived from the uterine glandular epithelium.82 It is not difficult to imagine potential roles for a chemosensory receptor transporting protein in the uterus during early pregnancy because chemokines are proposed to aid in trophoblast attachment and invasion.36 The chemokine CXCL10 was upregulated in the endometrium of pregnant ewes, and the receptor (CXCR3) was localized to the trophectoderm.83 Moreover, chemotaxis assays demonstrated that CXCL10 regulates migration and/or distribution of PBMC in the uterus during early pregnancy. Perhaps RTP4 affects chemokine receptors during early pregnancy to recruit immune cells to the pregnant endometrium.84 Further studies are needed to determine the role(s) of RTP4 in the endometrium during early pregnancy. What this experiment did reveal, however, was that gene expression in PBL during early pregnancy provided a novel and non-invasive mechanism to identify new genes regulated in the uterus during early pregnancy. We hypothesize that by profiling gene expression patterns in PBL, we may be able to identify expression patterns associated with successful and unsuccessfully pregnancy outcome.