Changes occurring in the maternal immune system during pregnancy allow the semiallograft fetoplacental unit to develop and grow. To a degree HLA discordance assists fetal development and reduces the risk of vertical virus transmission such as HIV-1 (1). However, tight immunomodulation is required at the maternal–fetal interface to prevent fetal rejection. The observation that some immune disorders such as rheumatoid arthritis and multiple sclerosis typically improve during pregnancy supports the idea that pregnancy is associated with a state of immune tolerance (2–4). However, not all aspects of the immune system are downregulated, for example, innate immune responses are enhanced and have been associated with preterm labor. Conversely, regulatory T cell (Treg) responses are suppressed, and pro-inflammatory T cell activity is increased in conditions such as pre-eclampsia (5–7).

Various regulatory mechanisms repress maternal immune function and the response against fetal antigens, in particular at the maternal–fetal interface. Programmed death-1 (PD-1) expression on T cells may promote apoptosis of T cells specific for paternal antigens (8, 9). Furthermore, progesterone-regulated HLA-G expression on trophoblast initiates Fas/Fas-ligand (CD95/CD95L)-mediated apoptosis of maternal CD8⁺ T cells specific for paternal antigens (10).

Regulatory T cells have been suggested to modulate the function of the maternal immune system during pregnancy by suppressing effector T cell function (11). For example, Tregs promote effector T cell anergy and inhibit proliferation and cytokine secretion by CD4⁺CD25⁻ T cells under polyclonal stimulus (12), independent of antigen specificity (13). Tregs also mediate immune suppression by reducing the expression of costimulatory molecules CD80 and CD86 on antigen-presenting cells (APCs), and stimulating indoleamine 2,3-dioxygenase (IDO) enzymatic activity in dendritic cells (DCs) and macrophages (13–15). IDO is an enzyme that catalyzes the rate limiting step of the breakdown of the essential amino acid tryptophan (trp) into the kynurenine (kyn) pathway. The combination of trp depletion and accumulation of downstream bioproducts of kyn exert inhibitory activity on T cells in vitro (16), IDO inhibition in vivo causes immune-mediated fetal rejection in semiallogeneic but not syngeneic pregnancies in mice (17). Pregnancy appears to promote immature/tolerant DCs and suppresses their pro-inflammatory cytokine responses (18). Under steady state conditions, with abundant antigenic stimulation, the majority of peripheral DCs display an immature phenotype (19). When matured under the influence of IL-10, immature DCs develop a tolerant DC (DC^IL-10) phenotype. These DCs secrete reduced quantities of pro-inflammatory cytokines and express low levels of MHC class 1 and 2 and costimulatory molecules (19). Tolerogenic DCs classically encourage CD4⁺ T cell differentiation into Treg and T-helper 2 (Th2) subtypes (20).

The majority of DCs are IFN-γ producing plasmacytoid (pDC) and non-lymphoid conventional or myeloid DC (cDC/mDC) (21, 22). Typically, pDC secretes large quantities of IFN-γ in response to virus and prime cytotoxic T cells against viral antigens, whereas mDC maintains self-tolerance and induces specific immune responses to foreign pathogens (23). In pregnancy, proportions of mDC and pDC have been shown to fall in the second trimester but subsequently increase in late pregnancy (24–26). The ratio of mDC and pDC shows a similar increase that is consistent with a predominance of mDCs. In addition, these DC populations become more activated during pregnancy, expressing increasing proportions of costimulatory markers and inflammatory cytokines (24, 26). Neopterin is an established biomarker of immune activation which is a product of guanosine-5′-triphosphate (GTP) catabolism in monocytes (27). Type I interferons are potent inducers of neopterin from activated human monocyte-derived macrophages and DCs (27, 28). Unsurprisingly, neopterin concentrations have been shown to increase with gestation in pregnancy, reflecting increasing DC and monocyte activation (29, 30).

Progesterone is important for the establishment and maintenance of pregnancy and exerts immune-modulatory effects mediated by the lymphocyte-derived protein progesterone-induced blocking factor (PIBF) (31, 32). Effector functions modulated by PIBF include cytokine synthesis, cytotoxic cell activity, and arachidonic acid synthesis (6, 31, 33). PIBF is produced by activated T cell receptor (TCR)-γδ⁺ and CD56⁺ T cells that have interacted with trophoblast and express progesterone receptors (PRs) (34). Serum concentrations of progesterone increase throughout pregnancy and can reach 175–636 nmol/l in maternal circulation mid third trimester (35). However, compared to the maternal–fetal interface, progesterone concentration peripherally is lower in humans and so systemically its immunemodulatory effects may be determined by lymphocyte sensitivity to the hormone (31).

Despite the influence of immune modulation, the expression of activation markers is increased on circulating and decidual T cells in late pregnancy and perhaps prior to the onset of labor (36, 37). This may suggest greater maternal immune awareness of the conceptus and trigger effector responses that could be involved in the process of parturition. In mice, this has been proposed as a cause of immune mediated fetal demise in fetal intervention (38).

In this study, we analyzed the in vitro functional responses to common recall antigens, profiled the changes in markers of peripheral blood T cell activation and maturation associated with pregnancy, and the in vitro expression of PIBF on lymphocytes in response to progesterone. Our results describe for the first time the course of pregnancy as a transition from downregulation to reconstitution of immune responses associated with modulation of T-cell and APC phenotype and function.

Our studies of immune parameters in normal pregnancy describe immune quiescence and inhibition of responses in the second trimester, which reverse in the third trimester in association with greater immune activation and functional immune responses. We suggest that a complex temporal relationship between activation and exhaustion determines the functional status of the maternal immune system during pregnancy.

Previous studies have described gestational increases in plasma IFN-γ and IL-10, and others have shown comparably greater levels of IL-10 and reduced IFN-γ production, measured by multiplex assays, in response to LPS in pregnancy than that observed in unstimulated PBMC suggesting a tendency toward immune repression (43–46). Here, we take a different approach, assessing cytokine responses by T cells following recall antigen stimulation, which are largely mediated by EM and TEMRA subsets (47). Other studies have shown that IL-10 ELISpot PBMC responses are increased in pregnancy (48) and IFN-γ and IL-10 decidual MNC responses are increased in labor (49). The effect of pregnancy may be to negatively affect pleiotropic pro-inflammatory cytokine production and cause immune cells to produce comparably greater IL-10 than in the non-Pr individual (44, 45). Furthermore, in some infective pathology, outside of pregnancy, IL-10 serves to suppress antigen specific (ELISpot) T cell responses ex vivo (50). Our results show that IFN-γ responses are reduced in the second trimester compared to third trimester pregnancies, despite the observed increase in CD4 EM proportion in the second trimester, which have rapid effector functions and homing to inflamed tissues (47). This decline in responses reverses in the third trimester. This observation suggests that effective inhibition of CD4 responses characterizes the second trimester. IL-10 responses showed a pattern similar to IFN-γ, arguing against a role for antigen-induced IL-10 in immune tolerance in mid-gestation. However, IL-10 production may have a key role in the third trimester, holding the ascendant immune activation in check. Of note, although IL-4 and granzyme B responses were largely undetectable, they showed a similar pattern to IFN-γ and IL-10 when measurable. Thus, immune responses in the second trimester are largely muted but are subsequently restored to a non-Pr level in the third trimester. Therefore, we propose that immune tolerance predominates during the second trimester, which is reversed in the third trimester, requiring active immune suppression to prevent exaggerated immune responses. In this study, we sought to understand the potential mechanisms at play.

Programmed death-1 and PD-L1 exhaustion markers are expressed on not only activated lymphocytes and Tregs but also non-hematopoietic and non-lymphoid cells such as placenta (8). The interaction between PD-1 and its ligand can either negatively or positively affect leukocyte activation (8). PD-L1 blockade in mice leads to impaired FoxP3 induction (51), a reduction in Treg proportion and favors a shift toward Th17 T cell phenotype that represses feto-maternal tolerance, resulting in embryo reabsorption (52, 53). Therefore, in the context of pregnancy, PD-1 signaling is required for both the suppression of effector T cells and the maintenance and induction of Tregs. In addition, the PD-1-PD-L1 pathway may induce fetal-specific T cells apoptosis thus limiting their accumulation (8). Our observations suggest that during the second trimester, a selective dampening of primary CD8 T cell immune responses and CD4 EM immune quiescence occurs, which reverts in the third trimester. In addition, we found that Treg proportions were increased in the second trimester compared to controls in keeping with previous reports (54). However, this may not be an accurate reflection of their functional status.

Pregnancy is characterized by significant changes in endocrine hormones and receptors. A close relationship between the endocrine and immune system has previously been reported (55). It is likely that the majority of changes seen in pregnancy are driven by the hormones estrogen and progesterone (56–58). Increasing concentrations of both hormones positively correlate with increasing gestation and successful pregnancy (56, 59, 60). Estrogen has been shown to promote immune tolerance by inhibiting CD4 T cell expansion and promoting apoptosis, increasing Treg proportions, and promoting tolerant DC expressing inhibitory markers (61–63). In addition, T cells express other steroid hormone receptors including glucocorticoid (GR) and androgen (AR) receptors that enable modulation of T cell functions (64, 65). For example, progesterone can bind to GR on T cells to increase Treg proportions, and testosterone binds to AR to regulate Th1 differentiation and promote FoxP3 expressing Tregs (64–66), although the latter is less important in pregnancy.

Progesterone is important for the establishment and maintenance of pregnancy and exerts immunemodulatory effects (67). Some of these are mediated by the lymphocyte-derived protein PIBF (31, 32). PIBF production is mediated through PRs in TCR-γδ⁺ and CD8⁺ T cells. Indeed, in vitro treatment of leukemic cell lines and CD3⁺ T cells with progesterone increases PIBF protein expression (68–71). PIBF binds a GPI-anchored receptor in a heterodimer with the α-domain of the IL-4 receptor, activating STAT6, inducing a Th2 cytokine bias, and suppressing lymphocyte cytotoxicity (33, 72). TCR-γδ T cells comprise >70% of decidual, but only 1–2% of peripheral, lymphocytes. They become progesterone-responsive after exposure to fetoplacental antigens stimulates PR expression (31). Uniquely, they combine adaptive and innate-like responses and are able to interact with non-classical HLA antigens such as those expressed by trophoblast (31, 73). Our results show a fall in proportion of CD4 TCR-γδ⁺ T cells, which may suggest TCR-γδ hypoexpression or apoptosis. However, we have also shown a greater PIBF expression in CD8⁺TCR-γδ⁺ T cells under the influence of progesterone in the second trimester. With greater regulation of responder CD8 T cells in mid trimester, as suggested by the differential expression of exhaustion markers in TEMRA subsets, it seems the second trimester is also characterized by progesterone regulation of MHC unrestricted CD8 TCR-γδ⁺ T cells. Unexpectedly, progesterone did not increase the expression of PIBF in the CD4⁺TCR-γδ subset. This may be due to an absence of antigenic stimulus, or indicative of negative regulation of systemic immune activation at this stage of pregnancy; our samples were taken between 22–27 weeks and 30–39 weeks of gestation (IQR). Indeed, a previous study that demonstrated a concentration-dependent increase in PIBF expression, measured by flow cytometry, and used blood taken at 16–24 weeks of gestation, whereas our study has used a much wider gestational range (71). Urinary PIBF falls in the third trimester despite increasing progesterone levels, perhaps explaining our results (74, 75). In order to further elucidate the impact of pregnancy on adaptive immunity we also analyzed DC phenotype and IDO activity.

Pregnancy-specific cytokines and progesterone have been shown to modify DC maturation and their ability to influence T cell responses by promoting IL-10 production and increasing the number of iTregs (20, 76, 77). Profiling of DC in our experiments showed decreased proportions of mDC but not pDC in peripheral blood, with dampened immune activating phenotypes. It remains unclear whether this loss of DC in the periphery is due to apoptosis or redistribution to other sites. Furthermore, we demonstrated a correlation between mDC proportions, pDC and mDC activation, and neopterin concentration as well as between mDC proportion and IDO activity. These changes are predominately seen in the third trimester and are closely related to T cell activation. Previous studies have shown a falling trp concentration but increasing KTR with advancing gestation suggesting increased IDO activity (78). We found that trp concentration but not KTR showed a gestational reduction, which was not mirrored by increased kyn concentration. An alternative explanation is trp and metabolic steroids can induce trp 2,3-dioxygenase activity in the liver to catabolize trp (79). Thus, the fall in trp concentration maybe due to altered trp bioavailability to favor fetal development rather than a reflection of greater IDO activity (80). However, we cannot exclude the possibility that DC phenotype and IDO activity share the same gestational variation as seen on T cells with a “Yin and Yang” relationship between immune activation and regulation. IDO activity has previously been shown to correlate with neopterin concentrations in pregnancy (17).

The reconstitution of immune responses in the third trimester, suggested by our ELISpot data, is accompanied by greater immune activation. It is already known that pregnancy is associated with increased immune activation, typified by the high proportions of CD38⁺HLA-DR⁺ T cells throughout pregnancy and up to 6 months after delivery (81). Our data show that CD38 expression on CD8⁺ T cells gradually increases with gestation, and HLA-DR expression by CD8 T cells correlated with increasing concentrations of neopterin, which is a surrogate marker for IFN-γ-associated immune activation (82, 83).

Apoptosis has also been proposed as a means of regulating immune responses in pregnancy to favor the semiallograft. Analysis of CD95 expression on T cells in our cohort showed that its expression was increased on both CD4 and CD8 T cells progressively with gestation and in parallel to increased immune activation. The CD95-CD95L pro-apoptotic pathway may potentiate maternal immune tolerance during normal pregnancy. Thus, the expression of CD95L by trophoblast may induce apoptosis of CD95 expressing activated cytotoxic maternal T cells at the maternal–fetal interface. Consistent with this, different CD95 gene polymorphisms have been shown to alter the risk of early pregnancy loss (84, 85). Moreover, the absence of CD95 expression is associated with a reduction in the deletion of fetus-specific T cells, but this does not result in an increased incidence of fetal reabsorption (86). The outcome of CD95 ligation on T cells may depend on the dose of the signal and the cytokine environment suggesting a more complex immunologic scenario than suggested above (87).

Another inhibitory marker that may influence immune responses, TIM-3, is expressed on the surface of terminally differentiated Th1 but not Th2 cells and its signaling downregulates Th1 and Th17 responses by inducing apoptosis (88). These effects are galectin-9 mediated, which is expressed on gastrointestinal epithelium, endothelium, and immune cells. TIM-3 may be hormone and cytokine regulated with increased expression in pregnancy (89, 90). Our data suggest a reduction in CD8⁺TIM-3⁺ T cells in the second trimester of pregnancy that reverts to normal in the third. This reflects similar changes seen in CD95 and CD38 expression.

The reversal of immune tolerance suggested by our data may reflect the processes occurring prior to the onset of parturition. Labor is an inflammatory process and may be driven by fetal antigen exposure and potent innate immune stimulus. Previous reports have shown gestational-specific changes in IFN-γ, TNF-α, IL-2, IL-3, IL-6, IL-10, IL-12, and GM-CSF (43, 44, 91, 92). Inflammation per se can lead to the release of mitochondrial DNA due to tissue necrosis (93). In addition, labor is associated with an increase in corticotrophin-releasing hormone mRNA found in syncytiotrophoblast membrane microparticles (94). Placental-derived mitochondrial DNA and syncytiotrophoblast membrane microparticles-induced monocyte cytokine secretion may contribute to the inflammatory processes occurring before and during parturition (93, 95). Other reports have shown that decidua contains differentiated EM cells poised to respond to antigen and that fetal-specific T cells respond to placental antigens in draining LNs (37, 96–98). Nancy and Erlebacher postulated whether the presentation of antigen at the maternal–fetal interface serves to reinforce the activation of placenta-specific T cells that were first exposed to antigen elsewhere (96). Therefore, parturition appears to involve an array of triggers for cell-mediated and innate responses in a poised and activated immune system.

Collectively, our data indicate that the adaptive immune system from mid to late pregnancy is increasingly activated, with simultaneous negative regulation of responder T cells, suggesting a delicate balance between immune responses and immune tolerance. However, whereas immune suppression dominates in the second trimester, this gradually reverses in the third trimester.

The Research Ethics Committee, Chelsea and Westminster Hospital Trust approved this study; Ref: 11/LO/0971. Human experimentation guidelines of the author’s institution were followed in the conduct of clinical research. Informed written consent was obtained from each donor prior to blood collection. This work was supported by Borne.

NS, NI, and MJ had a substantial contribution to the conception and design of the project and its interpretation. NS, NI, MJ, AeB, and DF were responsible for the acquisition, analysis, and interpretation of the data, and NS, AH, AdB, NI, and MJ drafted the work. All authors contributed to revising of the manuscript and have approved the final version. All authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The handling editor declared a shared affiliation, though no other collaboration, with one of the authors, AB.