REVIEW ARTICLE: Immunology of Pre-Eclampsia

Authors


CWG Redman, Nuffield Department of Obstetrics and Gynaecology, University of Oxford, John Radcliffe Hospital, Oxford OX3 9DU, UK.
E-mail: christopher.redman@obs-gyn.ox.ac.uk

Abstract

Citation Redman CWG, Sargent IL. Immunology of Pre-eclampsia. Am J Reprod Immunol 2010

Pre-eclampsia develops in stages, only the last being the clinical illness. This is generated by a non-specific, systemic (vascular), inflammatory response, secondary to placental oxidative stress and not by reactivity to fetal alloantigens. However, maternal adaptation to fetal (paternal alloantigens) is crucial in the earlier stages. A pre-conceptual phase involves maternal tolerization to paternal antigens by seminal plasma. After conception, regulatory T cells, interacting with indoleamine 2,3-dioxygenase, together with decidual NK cell recognition of fetal HLA-C on extravillous trophoblast may facilitate placental growth by immunoregulation. Complete failure of this mechanism would cause miscarriage, while partial failure would cause poor placentation and dysfunctional uteroplacental perfusion. The first pregnancy preponderance and partner specificity of pre-eclampsia can be explained by this model. For the first time, the pathogenesis of pre-eclampsia can be related to defined immune mechanisms that are appropriate to the fetomaternal frontier. Now, the challenge is to prove the detail.

Introduction

Pre-eclampsia is relatively common, affecting about 3% of pregnancies. It originates in the placenta and causes variable maternal and fetal problems. At its worst, it may threaten maternal and perinatal survival. It is defined as a syndrome (a pattern of clinical features) and is probably heterogeneous in origin as it is in presentation.

Placentation, the Uteroplacental Circulation and Staging of Pre-Eclampsia

Pre-eclampsia results from imbalance between factors produced by the placenta and maternal adaptation to them. There are two broad classes of pre-eclampsia: maternal and placental, although many cases are a mix of the two.1 Placental pre-eclampsia is the outcome of poor placentation in early pregnancy (weeks 8–18). It comprises failure to remodel spiral arteries supplying the uteroplacental circulation. Normal placentation requires that cytotrophoblast invade the placental bed where they come into contact with maternal tissues. The distal ends of the spiral arteries are then transformed into widely dilated, structureless conduits. Spontaneous miscarriage, especially if recurrent, is also associated with poor placentation2 and is the extreme end of a spectrum of compromise (Fig. 1).

Figure 1.

 Establishing the intervillous circulation. Before 8 weeks, the spiral arteries are plugged by cytotrophoblast and there is no intervillous perfusion. During the next 4 weeks, the arteries progressively unplug. With inadequate trophoblast invasion of the placenta bed, this happens prematurely. Then either miscarriage ensues or pregnancy continues with dysfunctional placental perfusion which leads on to pre-eclampsia. Adapted from Burton and Jauniaux (2004).2

Pre-eclampsia has pre-clinical (symptomless) and clinical stages;1 until recently, only the final phase could be detected by clinical screening. The latter is the outcome of placental dysfunction secondary to oxidative stress and a maternal inflammatory reaction to its presence.3 Poor placentation is probably not the primary cause of pre-eclampsia. Changes in circulating trophoblast–derived factors associated with an increased risk of pre-eclampsia can be detected before placentation is completed.4 It is probable that pre-eclampsia results from abnormal trophoblast growth and differentiation, at any time after the earliest stages of implantation.4 The primary dysfunction may be immunologic leading to the concept of a three-stage disease.5

This review is mainly focused on placental pre-eclampsia. The central issue is whether maternal immune responses to trophoblast generate normal or abnormal placentation?

Throughout, we emphasize that, in overt pre-eclampsia, non-specific inflammatory responses contribute to pathogenesis. Placentation in early pregnancy appears to involve specific maternal immune responses to fetal alloantigens, which generate partner specificity.

Clinical and Epidemiological considerations

Pre-eclampsia occurs mainly in first pregnancies. This has been explained by invoking immune mechanisms linked to the belief that a genetically foreign fetus challenges the maternal immune system.6 The hypothesis is that the maternal immune system ‘learns’ to accommodate the fetus. Such adaptation may be relatively defective in a first pregnancy but less so in subsequent pregnancies. Furthermore, there may be partner specificity,7 which strengthens the argument that pre-eclampsia results from a relative failure to induce maternal tolerance of paternal alloantigens.8

Partners, Coitus, Sperm, and Semen

A change of partner seems to restore the risk of pre-eclampsia to that of primiparity in multigravid women, for example Zhang and Patel 2007.9 But, a change of partner is also associated with a long inter-pregnancy interval.9,10 After correction for the latter, the association with the former disappears.11 However, a short interval between first coitus and conception (coital interval) with the same partner also increases the risk of pre-eclampsia.12,13 In this context, there is clear evidence for partner specificity. These features suggest that exposure to paternal sperm or seminal plasma or both tolerizes the mother to fetopaternal alloantigens and failure of this immunoregulation increases the risk of pre-eclampsia (Stage 0 of pre-eclampsia).

Immune priming, by coitus, has been demonstrated in mice.14 Seminal plasma appears to be more important than sperm. It contains paternal type I and type II MHC antigens and high concentrations of transforming growth factor-β (TGFβ). TGFβ induces regulatory T cells (T(reg)) or, in a more pro-inflammatory environment, Th17 cells.15 In mice, exposure to seminal fluid at mating induces tolerance to paternal alloantigens and an accumulation of T(reg) in the uterine draining lymph nodes, which may facilitate implantation.16

Pregnancy itself confers further protection to pre-eclampsia in later pregnancies by the same partner. This is probably true even after an abortion, although the evidence is inconsistent.17 Both forms of protection (before or after conception) appear to be relatively short lived, explaining the increased risk of pre-eclampsia after a long inter-pregnancy interval.

The importance of pre-conceptual exposure to semen can explain why artificial insemination with donor sperm, intracytoplasmic sperm injection, or barrier methods of contraception, summarized by Dekker (2002),18 are all associated with increased risks of pre-eclampsia (Fig. 2).

Figure 2.

 Maternal tolerance to fetal alloantigens. A longer pre-conception duration of coitus reduces the risk of pre-eclampsia by promoting maternal tolerance to paternal antigens. The tolerizing mechanism is not activated with some forms of assisted conception. Pregnancy itself enhances this tolerance which is slowly lost after delivery. Whether a change of partner increases the risk of pre-eclampsia depends on the coital interval with the new partner, not the duration of time since the last pregnancy. *IVF: in-vitro fertilisation; ICSI: intracytoplasmic sperm injection.

Nature’s Transplant and Maternal–Placental Interfaces

Reproductive immunology has focused on the concept of the semi-allogeneic fetus and its possible rejection by classical T-cell responses to alloantigens. But, the placenta not the fetus comprises the ‘transplant’. The placental cell of interest is trophoblast. Maternal immune cells are confronted by different trophoblast subtypes at multiple maternal–placental interfaces depending on gestational age and anatomic location (Table I). The interfaces may be in maternal tissues (decidua) or maternal blood (intervillous space). The earliest comprises syncytiotrophoblast of the implanting embryo (Interface I), the only time that syncytiotrophoblast lies at a tissue interface. The intervillous circulation is established after 8 weeks. Before then, chorionic villi are bathed in uterine gland secretions – histiotrophic nutrition to the first-trimester fetus.19 By the end of the first trimester, Interface I is replaced by three others: formed by invasive cytotrophoblast in the placental bed (Interface II), with the chorion leave (Interface III), and by syncytiotrophoblast bathed by maternal blood in the intervillous space (Interface IV). These concepts extend our earlier proposals.20 In the first half of pregnancy, Interfaces I and II dominate. Interface II diminishes after 16 weeks,21 while Interface IV is activated with onset of the uteroplacental circulation at 8–9 weeks5 and enlarges with placental growth to become the dominant interface after 20 weeks. The early interfaces are concerned with implantation, placentation, and stages 1 and 2 of placental pre-eclampsia. Interface IV is largest and, at the end of pregnancy, generates the final stage 3 of the disorder.

Table I.   The four maternal-fetal immune interfaces and preeclampsia Thumbnail image of

Immune Mechanisms and Causes of Pre-eclampsia: Where, When, and How?

If immune mechanisms are relevant, there must be maternal immune recognition of trophoblast, which regulates implantation, placental growth, and placentation. It needs to be partner specific and to generate immune memory. It should be consistent with the structure of the immune maternal–fetal interfaces, with the types of maternal immune cells positioned at the interfaces and the antigens expressed by trophoblast. Partner specificity requires trophoblast expression of paternal alloantigens and their recognition by maternal immune cells. Immunoregulation implies a mechanism, which may partially fail to cause deficient placentation or, by extension, fail completely to cause spontaneous abortion. Memory points to T-cell involvement.

HLA (Human Leukocyte Antigen) Expression by Human Trophoblast

Human trophoblast has limited expression of strong transplantation antigens (Fig. 3). These include non-polymorphic HLA-E, F, and G which cannot signal paternal specificity whereas HLA-C, on extravillous (invasive) cytotrophoblast in Interface II, can. But this interface regresses in the second half of pregnancy.22 Stage 3, the clinical stage of pre-eclampsia, occurs when interface IV, the villous syncytium, is dominant. This is devoid of HLA expression. Immune mechanisms are of interest in the first and second stages of pre-eclampsia. The clinical third phase is generated by a maternal systemic inflammatory response, which is unlikely to be alloantigen driven.

Figure 3.

 HLA expression by subsets of human trophoblast. Adapted from van Maurik et al. (2009).21

Stage 1 Pre-Eclampsia: Peri-implantation and the Histiotrophic Placenta

At this stage, specific immune mechanisms in the placental bed are almost inaccessible to study. If immunoregulatory mechanisms fail completely, spontaneous abortion would ensue whereas partial failure might lead to continuing pregnancy with a small placenta. This creates a continuum between abortion and pre-eclampsia.2,23

Indoleamine dioxygenase 2,3-dioxygenase (IDO), an enzyme involved in the catabolism of tryptophan, is an important immune regulator. It is activated in antigen-presenting cells by various inflammatory stimuli including interferon-γ. Tryptophan is an essential amino acid. Its depletion by catabolism ‘starves’ T cells in the tissue micro-environment. This promotes their differentiation into regulatory T cells [T(reg)].24 IDO acts as a switch: in its presence, T(reg) are promoted; in its absence, T(reg) are reprogrammed to acquire a Th17 phenotype, with more pro-inflammatory activity.25 T(reg) modulate immune responses in an antigen-specific way, hence their importance for allogeneic pregnancy, for which they are essential (see elsewhere in this issue). Pregnant mice treated with IDO inhibitor (4.5 days post conception) reject allogeneic but tolerate syngeneic fetuses.26 If the inhibitor is given later (6.5 days post conception), allogeneic pregnancies continue but significant systolic hypertension develops compared to mice with syngeneic pregnancies.27 Proteinuria is inconsistently present but, in this report, the mice were killed relatively early (16.5 days) and the full evolution of the pregnancies was not defined. Hence in a mouse model, antigen- and stage-specific immune mechanisms are highly important. Early dysfunction leads to pregnancy loss; but dysfunction at a slightly later stage is associated with one feature of pre-eclampsia (hypertension) toward the end of a continuing pregnancy.

Expression of IDO in human endometrium begins in the luteal phase in the glandular epithelium and stromal leukocytes. This pattern regresses during the second trimester when expression begins in trophoblast,28 particularly strongly in invasive cytotrophoblast,29 where it might be expected to modulate interactions with decidual immune cells. In short, its position at Interfaces I and II is highly relevant to potential immune mechanisms in stages 1 and 2 of pre-eclampsia.

In summary, together with TGF-β, the components of an alloantigen-specific immunoregulatory mechanism [IDO and T(reg)] are located at interface I, are primed by pre-conceptual exposure to partner’s semen and, if dysregulated, could explain why primiparity and primipaternity are risk features for pre-eclampsia.

However, there is little evidence available at the moment for two reasons. Interface I can only be studied before the outcome of pregnancy can be known. If the pregnancy is continuing, then the interface is accessible only to indirect measurements. Secondly, there is doubt about how best to identify and measure T(reg) during human pregnancy.30 No measurements have been reported in stages 1 and 2 of pre-eclampsia where they would be most likely to influence allogeneic responses at interface II. Whether measures in peripheral blood are a good indicator of decidual activity is not known. Many authors look for concurrent expression of CD4 and CD25 or CD25bright to identify T(reg) and most find that their proportions are increased in normal pregnancy, for example Somerset et al. (2004).31 However, the data are inconsistent, as are the methods used. T(reg) themselves are phenotypically and functionally heterogeneous32 so the relevant subset to study is not known.

Recognition of HLA-C in the Early Human Decidua

HLA-C is a key trophoblast molecule in relation to pre-eclampsia in stages 1 and 2. Both decidual T cells and the more abundant decidual NK cells can recognize paternal HLA-C. The importance of decidual NK cells is described elsewhere in this issue. They express killer immunoglobulin-like receptors (KIR) for which HLA-C is the dominant ligand. HLA-C has more than 100 alleles. Its KIR receptors are themselves extremely polymorphic. There are up to 17 different human KIR genes each with its own polymorphisms, some that activate, others that inhibit. The number of KIR genes in different genotypes varies. One genotype may itself be variably expressed by differential expression of KIR genes, fixed by methylation with the phenotype passed to daughter cells.33 In other words, maternal–fetal immune recognition at the site of placentation appears to be highly individualized by two polymorphic gene systems, maternal KIRs and fetal HLA-C molecules.

KIR haplotypes fall into two groups, A and B, the latter distinguished by additional activating receptors. Uterine NK cells make chemokines and angiogenic cytokines, which promote trophoblast invasion. Their secretion is enhanced by ligation of a stimulatory KIR receptor (haplotype B) and reduced by ligation of haplotype A receptor as is summarized by Moffett and Hiby (2007).34 The former combination, in vivo, would be predicted to protect from pre-eclampsia. Possible maternal KIR genotypes could be AA (no activating KIR) or AB/BB (presence of one or more activating KIRs).

HLA-C haplotypes can also be grouped as C1 and C2, depending on a single amino acid dimorphism in the alpha-1 domain. HLA-C2 interacts with KIRs more strongly than HLA-C135 so that maternal HLA-C2 with fetal KIR B/B could be the best combination for promoting adequate placentation and avoiding pre-eclampsia. This is what is observed. In a converse manner, Kir AA mothers confronted with HLA-C2 fetuses are the most susceptible to pre-eclampsia.36 Similar patterns occur in recurrent miscarriages which, as already mentioned, may share the same causation with pre-eclampsia, at the most extreme end of the spectrum.37

Although this form of maternal–fetal immune recognition can explain the partner specificity for pre-eclampsia, it does not readily explain protection conferred by a previous pregnancy by the same partner.

Pre-eclampsia and Immunologic ‘Memory’

Protection from pre-eclampsia has relatively short-term partner specificity. This could imply involvement of T cells and T-cell memory. The evidence is that paternal HLA-C is recognized by decidual T(reg), which can down regulate anti-paternal responses.38,39 The stability of T(reg) memory is still being determined. Natural T(reg) seem to be stable whereas those that are inducible probably are not.40 It is therefore likely that decidual T(reg) bestow, at least, short-term memory which could protect from pre-eclampsia in a second pregnancy with a short inter-pregnancy interval but this needs further investigation. The specific involvement of T(reg) in the genesis of pre-eclampsia is difficult to investigate; there are no data at the moment.

Recently, it has become evident that NK cells themselves can sustain memory and mount the equivalent of a secondary immune response.41 Indeed, macrophages can generate a form of memory by epigenetic modification of specific TLR genes.42 Thus, an immune explanation for the primiparity and primipaternity effects of pre-eclampsia has become more plausible but awaits further investigation.

The Third Stage of Pre-eclampsia and Maternal Systemic Inflammation

Poor placentation leads to small muscular spiral arteries which supply high pressure pulsatile flow to the intervillous space. Damage from rapid changes in oxygenation and hydrostatic stress ensues.43 The arteries retain responsiveness to vasoconstrictors, which can exacerbate oxidative stress further. Oxidative and ER stress are both features of the pre-eclampsia placenta.44 They are also powerfully pro-inflammatory. We propose that the inflammatory drive of pre-eclampsia results from the three-way interactions between these two stresses and inflammatory responses3,45,46 in the placenta (Fig. 4). We further suggest that the third stage of pre-eclampsia is not caused by ‘rejection’ of an allogeneic placenta but by a global maternal inflammatory response to a damaged placenta.

Figure 4.

 Three stages of pre-eclampsia. Stages 1 and 2 lead to dysfunctional uteroplacental perfusion and placental oxidative stress. This and the associated inflammatory and endoplasmic reticulum stresses lead to abnormal placental–maternal signalling and overt pre-eclampsia.

This is superimposed on a background of systemic inflammatory response common to all normal pregnancies, which starts in the luteal phase of the menstrual cycle,47 when an allogeneic fetus cannot be implicated. The normal response intensifies with advancing gestational maturity.48 We have emphasized the diverse and widespread nature of this sort of response (Fig. 5), which can explain the complex clinical presentations of severe pre-eclampsia, including endothelial, metabolic, coagulation, and complement abnormalities.49 We previously highlighted that pre-eclampsia is not an intrinsically different state from normal pregnancy but the extreme end of a continuous spectrum of responses that are a feature of pregnancy itself. Many ‘physiological’ changes of normal pregnancy itself are simply less severe manifestations of the same changes in pre-eclampsia.

Figure 5.

 The systemic inflammatory network has metabolic components as well as involving the intravascular coagulation and complement systems. The main players are inflammatory leukocytes and endothelium. All are activated in pregnancy relative to non-pregnancy and further activated in pre-eclampsia relative to normal pregnancy.3

Trophoblast Derived Factors and Pre-eclampsia

Elsewhere we summarize the several possible trophoblast-derived factors that potentially contribute to maternal systemic inflammation.3 These include growth factors, activin-A, corticotrophin-releasing hormone, and leptin, all of which have documented pro-inflammatory actions. Another pro-inflammatory placental factor has been recently described, namely free heme, which is strongly pro-oxidant and pro-inflammatory.50 It is ectopically synthesised in the placenta and placental bed in pre-eclampsia. A final candidate factor comprises placental microvesicles.

Activated cells release microvesicles (200–500 nm) by blebbing of the cell membrane. Apoptotic cells release larger microvesicles or apoptotic bodies. Cells can also secrete nanovesicles or exosomes (40–100 nm). Microvesicles are shed from cell surfaces after activation, while nanovesicles are stored in multivesicular bodies and secreted constitutively. Microvesicles are normally detectable in blood from healthy individuals and are increased where there is vascular or inflammatory stress. They are derived from various intravascular sources (platelets, leukocytes, endothelial cells, and so on) and can be identified by the specific markers they express. There is increasing awareness of their potential as markers for arterial, inflammatory, and malignant diseases.51

Micro- and nanovesicles are of interest in pre-eclampsia because they can be markers of inflammatory and endothelial dysfunction, are immunoregulatory, and because there is a pregnancy-specific component derived from trophoblast. Syncytiotrophoblast microvesicles and nanovesicles are shed in normal pregnancy and in significantly increased amounts in pre-eclampsia.52–54 They have an anti-endothelial effect52 and are also pro-inflammatory in vitro.53

Microvesicular shedding from the syncytial surface would be expected to increase in two situations. The first is with increased placental size. Pre-eclampsia is predominantly a disorder of the third trimester when the placenta reaches its greatest size. Multiple pregnancies are associated with a higher risk of pre-eclampsia and also larger placentas. The second situation would be associated with placental oxidative stress in the most severe pre-eclampsia, typically of early onset. The placentas are usually abnormally small but may generate microvesicles with a more intense inflammatory stimulus, for example, by an increased content of peroxidized lipids.55 Whether these microvesicles are a cause or consequence of systemic vascular stress is not known.

Immunoregulation in Stage 3 Pre-eclampsia

Only two aspects will be covered in detail: T(reg) and the role of IDO.

Elsewhere in this issue, the concept of the Th2 bias of normal pregnancy and the need to update the Th1/Th2 paradigm in terms of Th17 and regulatory T cells [T(reg)] is discussed. Pre-eclampsia is associated with a Th1 bias; but this extends beyond Th cells to a type I bias of NK cells.56 However, in normal pregnancy, there may also be a bias away from IL-17 producing cells, which is not found in pre-eclampsia.57 Th17 stimulates autoimmunity and coordinates the inflammatory response to infection. Its potential to harm pregnancy is not yet defined.

The increase in circulating T(reg) found by some authors in normal pregnancy appears to be lost in pre-eclampsia (several authors, for example Sasaki et al. (2007).58 However, the available data are inconsistent. T(reg) are heterogeneous without specific markers59 so that their identification, for example by flow cytometry, is problematic. In addition, they function in tissues so that measures in peripheral blood are not easy to interpret. That Stage 3 pre-eclampsia is caused by loss of immune regulation to an allogeneic fetus owing to T(reg) dysfunction is inconsistent with the evidence that the end-stage of the disease is a non-specific inflammatory response to a damaged placenta. A simpler interpretation is that a Th1 environment tends to suppress the generation of T(reg).60 The interactions between IDO and T(reg) have already been mentioned, T(reg) differentiation being stimulated by IDO. Indirect measures suggest that IDO is less active in stage 3 pre-eclampsia.61 Given that 3rd stage pre-eclampsia involves maternal systemic inflammation, it would be predicted that IDO would be activated. So this observation is unexplained. However, there is two way interaction between T(reg)62 and IDO such that underactive T(reg) responses could cause less IDO activity. The full detail awaits further analysis.

Pre-eclampsia and Autoantibodies

Certain function perturbing autoantibodies can increase the risk of pre-eclampsia most notably anti-phospholipid and anti-angiotensin II type I receptor antibodies.63,64 It is difficult to argue that these are the specific cause of the syndrome even though the latter autoantibodies can generate a pre-eclampsia-like disorder after administration to pregnant mice.65 In a case–control study, we found that such antibodies were relatively common in normal pregnancy controls and not present in a substantial minority of cases.66 Cross-reactivity with parvovirus B19 VP2 protein was demonstrated although serological evidence of previous parvovirus infection was not more common in pre-eclampsia cases. Th17 activity predisposes to autoimmunity.67 If the possible Th17 bias of pre-eclampsia is confirmed, this might amplify the induction of these antibodies in at least some cases, which might contribute to the pathology.

Conclusion

Different immune mechanisms operate at different interfaces during the three stages of pre-eclampsia. The diversity of the changes in the final stage is well explained by a maternal systemic inflammatory response secondary to oxidative placental damage. Now, for the first time, there is the potential to explain the early phases of pre-eclampsia in terms of immune mechanisms that are appropriate and located at Interfaces I and II and account for the first pregnancy preponderance of pre-eclampsia. The challenge is to turn potential into reality. A provisional scheme of the immune mechanisms of pre-eclampsia is in Fig. 6.

Figure 6.

 Summary of immune events in pathogenesis of pre-eclampsia. The mechanisms of Stage 0 and Stage 1 have yet to be proved; the involvement of T(reg), dendritic cells, and Indoleamine dioxygenase 2,3-dioxygenase (IDO) is speculative but supported by experiments in mice. Stage 0–2 mechanisms largely concern uterine mechanisms. Stage 3 disease is a systemic inflammatory disorder.

Acknowledgments

We acknowledge support from the Wellcome Trust Grant Ref GR079862MA. Our work is also supported by the Oxford Partnership Comprehensive Biomedical Research Centre with funding from the Department of Health’s NIHR Biomedical Research Centres funding scheme. The views expressed in this publication are those of the authors and not necessarily those of the Department of Health.

Ancillary