Innate immune cells in the placental bed in healthy pregnancy and preeclampsia

Immune cells in the placental bed are important for adequate development of the placental bed. The most prominent immune cells in the placental bed early in pregnancy are uterine natural killer cells (uNK) cells and macrophages. Also dendritic cells and mast cells can be found in the early placental bed. These cells not only have an immune regulatory function, but are also involved in the regulation of trophoblast invasion, angiogenesis and spiral artery remodeling. In preeclampsia, one of the major complications of pregnancy, decreased trophoblast invasion and spiral artery remodeling has been found. This is associated with decreased numbers of uNK cells, increased numbers of macrophages around the spiral arteries and similar or increased numbers of dendritic cells in the placental bed. In this review, we discuss the current insights in the functions of uNK cells, macrophages, dendritic cells and mast cells in the placental bed in humans during healthy pregnancy and during preeclampsia. As animal models are instrumental in understanding the role of immune cells in the placental bed, we also review studies on the function and phenotype of these innate immune cells in experimental preeclampsia. A better understanding of the dynamics and functional changes of these immune cells in the placental bed may eventually lead to new therapeutic targets for preeclampsia.


Introduction
The human placenta is a hemochorial placenta, which is the most invasive type of placenta. In this placenta there is intimate contact between fetal tissue (trophoblast) and the maternal cells, such as immune cells, at the fetal-maternal interface [1]. Both trophoblast and maternal immune cells are important for proper development of the placental bed, such as spiral artery remodeling, but also for protecting the integrity of the uterus and the mother. The main immune cells present in the placental bed are uterine NK (uNK) cells and macrophages, comprising about 90% of all leukocytes [2]. About 10% of the immune cells are T cells in early human pregnancy [2,3], while dendritic cells, mast cells, granulocytes and B cells can also be found but in lower frequencies [2,4]. A disbalance in immune cell frequencies in the placental bed can lead to disease and pregnancy complications.
Preeclampsia is one of the most important complications in pregnancy, characterized by de novo hypertension accompanied by proteinuria or one or more of the following conditions: renal insufficiency, liver involvement, neurological complications or hematological complications or fetal growth restriction [5]. Preeclampsia can be subdivided into early onset (onset before 34 weeks) and late onset preeclampsia (onset after 34 weeks). Early onset preeclampsia, which is the most severe form of preeclampsia, is associated with impaired spiral artery remodeling and with changes in the number of trophoblasts and immune cells in the placental bed.
In the present review, we will discuss the current insight on the role of the innate immune cells uNK, macrophages, dendritic cells and mast cells in the placental bed in keeping homeostasis in normal pregnancy and in development of preeclampsia. We will mainly focus on the placental bed in humans, but we will also discuss some relevant rodent models that have contributed to the current knowledge on the function of innate immune cells in the placental bed of healthy pregnancy or preeclampsia.

Uterine NK cells in the human placental bed
2.1. Healthy pregnancy NK cells are classified as lymphocytes, with cytotoxic and cytokine producing functions [6] (NK cell marker) and lack of expression of CD3 (T cell marker) [6]. In humans, two populations of NK cells can be distinguished: the cytotoxic NK cells, (CD56+/CD16+) (CD16 is the low affinity Fc receptor), representing about 90% of all NK cells and the cytokine producing NK cells, (CD56++/CD16-), which are also known as regulatory NK cells [6]. Although uNK cells in the human endometrium and placental bed do express high levels of CD56 and lack CD16 expression, in contrast to the circulating cells, they are granulated and express different markers [2]. For instance, they express higher levels of Killer-cell immunoglobulin-like receptors (KIR) and natural killer group (NKG) 2 A/ C/E/D [7] as compared with blood NK cells. Both these receptors can bind to Human Leukocyte Antigen (HLA) C and E expressed by the trophoblast [7]. UNK cells, in contrast to blood NK cells, express CD9, involved in cell adhesion and motility, and CD49a, involved in cell adhesion [7,8].
UNK cell-receptors, KIRs, are important for the function of uNK cells. The expression of these receptors seems to be regulated by oxygen tension in the decidua [9]. KIR receptors bind to Human Leukocyte antigen molecules (HLA) molecules on trophoblast and this binding is important for uNK cell function, for instance for the production of certain growth factors [10]. The binding of uNK cell KIR receptors to trophoblast HLA molecules is also important for trophoblast migration and spiral artery remodeling [10]. The expression of certain KIR receptors (activating KIR receptors), such as KIR2DS1 [10], KIR2DS4 [11] and KIR2DS5 [12] in combination with extravillous trophoblast expression of HLA-C2 is important for normal trophoblast invasion and spiral artery remodeling and therefore for normal placental development.
In the placental bed of early human pregnancy, the numbers of uNK increase [3,13]. After the first trimester, the numbers of uNK cells decrease in the placental bed [14], but they are still present at the end of pregnancy [15,16]. In healthy pregnancy, uNK cells are usually found in the vicinity of invading fetal trophoblast cells [17], and around spiral arteries, that are being remodeled [18]. The potential role of these cells in trophoblast invasion has been shown by in vitro studies. Conditioned medium from uNK cells have been shown to increase chemoattraction of trophoblasts [19,20]. Chemotactic promoting factors may by hepatocyte growth factor [21],IL-8, CXCL10 [22] or IL-6 [23] or protease activity [24]. Interferon γ (IFN γ), on the other hand, may decreased trophoblast migration [25]. Spiral artery remodeling takes place in 2 phases: in the first phase, there is a loss of the musculo-elastic structure and the formation of endothelial cell layer breaks [18]. In the second phase, endovascular trophoblasts are attracted towards the spiral arteries and replace the endothelial lining in the decidua and first part of the myometrium [18]. UNK cells play a role in both phases of spiral artery remodeling [18,26,27]. The mechanism by which uNK cells affect spiral artery remodeling is not exactly clear, however, uNK cells produce many factors, such as cytokines (for instance Tumor necrosis factor-α (TNFα), interleukin-10), growth factors (for instance Macrophage colony-stimulating factor (M-CSF), Placental growth factor (PlGF)), angiogenic factors (such as Vascular endothelial growth factor-C (VEGF-C) Angiopoietin 2 (Ang2)), and other factors, like matrix metalloproteinases (MMPs) [27][28][29][30][31]. These factors play a role in spiral artery remodeling [27,28,32].

Preeclampsia
As already outlined in the introduction, preeclampsia is one of the most important complications of pregnancy, characterized by hypertension in combination with proteinuria, renal insufficiency, liver involvement, neurological complications or hematological complications or fetal growth restriction [5]. Preeclampsia can be divided into early onset (onset before 34 weeks) and late onset preeclampsia (onset after 34 weeks). The early onset form of preeclampsia is associated with decreased trophoblast invasion and spiral artery remodeling as well as with fetal growth restriction [33], resulting in a stressed placenta. The stressed placenta then starts secreting various inflammatory, antiangiogenic and other damaging factors into the maternal circulation resulting in a generalized maternal inflammatory response and endothelial cell dysfunction and finally preeclampsia [33].
Changes in uNK cells numbers in the placental bed have been found in preeclamptic patients. However, some studies found increased numbers of uNK cells in decidua from women with preeclampsia compared with age matched controls [34][35][36], while others have demonstrated decreased numbers of uNK cells in placental bed biopsies of preeclamptic patients [37,38] or observed no difference [15]. Differences in these studies may be due to for instance the use of different experimental methods or the use of tissue from preeclamptic patients, which differed in severity (f.i. preeclampsia with or without growth restriction) or the fact that different placental bed locations were studied (f.i. uNK close to spiral arteries or total numbers of uNK cells in the placental bed). It seems that total numbers of uNK cells are correlated to the presence of fetal growth restriction (FGR), rather than to preeclampsia itself, since the total numbers of uNK cells seems to be decreased in pregnancies with FGR (with or without preeclampsia) [37,38] and correlated with decreased spiral artery remodeling [38].
As indicated above, the expression of certain KIR receptors (i.e. activating KIR receptors such as KIR2DS1, KIR2DS4 and KIR2DS5) is important for normal development of the placental bed and thus normal pregnancy [10][11][12].This suggests that a lack of expression of these receptors or expression of other KIR receptors may be associated with deficient trophoblast invasion and spiral artery remodeling. Indeed, women with mainly inhibitory KIR receptors (KIR AA genotype) (such as KIR2DL1, KIR2DL2, KIR2DL5 and KIR3DL3) in combination with trophoblast expression of HLA-C2 have an increased the risk for developing preeclampsia [39].
Unfortunately, studying uNK cells in biopsies taken from decidua or placental bed after parturition are not informative for the pathophysiology of preeclampsia. Changes found in uNK cells in these biopsies may be the result or the cause of preeclampsia. Therefore, biopsies at the time of trophoblast invasion and spiral artery remodeling are needed. However, collection of placental bed tissue from this time of pregnancy is limited to tissue obtained from termination of pregnancy. The disadvantage of this tissue is that no information is available as to the outcome of the pregnancy. The group of Whitley and Cartwright, however, has developed a strategy to overcome this problem. They have used uterine artery Doppler ultrasound screening to identify women at increased risk for developing preeclampsia and growth restriction, i.e. women with a high uterine artery resistance index i.e. a resistance index of > 95th percentile [40]. Using uNK cells isolated from women with a high resistance index as indicated by uterine artery Doppler screening and from women with a normal resistance index showed that uNK cells from women with a high resistance index were less able to promote invasive behavior of trophoblast in vitro as compared to women with normal resistance index [21]. This may be due to a decreased expression of various uNK cells receptors, such as KIR2DL1/S1 receptors and LILRB1 receptors in women with a high resistance index [41]. This may result in impaired trophoblast-uNK cell interactions [19,41] and differences in secreted factors between uNK cells in women with a high resistance index as compared with women with a normal resistance index [41]. In line with the different functions of uNK cells from women with a high resistance index, uNK cells from these women induced increased endothelial cell activation, destabilized endothelial tube like structures and induced more TNFα production by endothelial cells as compared to uNK cells from women with a normal resistance index [42]. These data suggest that there are changes in uNK cells function in women who later develop preeclampsia during the time of trophoblast invasion and spiral artery remodeling, suggesting that changes in uNK cell function may play a role in the pathophysiology of early onset preeclampsia.

Healthy pregnancy
A lot of knowledge on the role of uNK cells in spiral artery remodeling in healthy pregnancy is derived from mice studies. Similar to humans, uNK cells in the mouse are activated [43] and express cytokines, such as IFNγ and IL-22 [44,45], as well as growth factors, such as and PLGF [45] and angiogenic factors, such as VEGF [46]. Mouse uNK cells also express the mouse equivalent of the KIR receptors, i.e. LY49 [47].
In the mouse, uNK cells play a major role in spiral artery remodeling, since NK cell deficient mice largely lack spiral artery remodeling and have abnormal decidual and myometrial structures at mid gestation [48]. The role of uNK cells is even more apparent for experiments in which NK cells were reconstituted in NK cell deficient mice, which recovered spiral artery remodeling [49]. However, also IFNγ administration to NK cell deficient mice resulted in recovery of spiral artery remodeling [44], suggesting that IFNγ is important for spiral artery remodeling. IFNγ may provide the initiating step in spiral artery modification [44]. Also other factors may affect spiral artery remodeling in the mouse, such as nitric oxide [50] or angiogenic factors, such as VEGF [51].
As the mouse is not an adequate model for deep trophoblast invasion, rat models are used to study deep trophoblast invasion and the interaction between uNK cells and trophoblast invasion [52,53]. Staining of uNK cells and trophoblast in consecutive sections of the mesometrial triangle, which is the equivalent of the human placental bed, on days 15, 17 and 20 of rat pregnancy showed that extravillous trophoblast invasion followed the demise of uNK cells in the direction of the myometrium, suggesting that uNK cells regulate extravillous trophoblast invasion [54]. UNK cells may also regulate endovascular trophoblast invasion, since depletion of NK cells in early pregnancy, resulted in increased endovascular trophoblast invasion at day 13 [55].
Chakraborty et al. [55] have shown that uNK cells promote uterine spiral artery growth towards the ectoplacental cone [55] and that uNK cells cause a partial disruption of the spiral artery tunica media integrity [55]. We showed that uNK cells, but not interstitial or endovascular trophoblasts, were found in the presence of partly remodeled spiral arteries in the mesometrial triangle of day 15 of pregnancy [54].

Preeclampsia
Both mice and rats models are instrumental in understanding the pathophysiology and role of uNK cells in preeclampsia. Mice and rat models for preeclampsia allow studying uNK cells at different time points of pregnancy. One study has shown that in a mouse model for preeclampsia uNK cells are decreased in the mesometrial triangle in mid gestation and that this lower presence of these cells correlated with decreased spiral artery remodeling [56]. A study from our own lab demonstrated that in a rat model for preeclampsia, decreased trophoblast invasion and spiral artery remodeling was associated with decreased numbers of uNK cells [52]. In contrast to depletion of NK cells in healthy pregnant rats, depletion of NK cells in a preeclamptic rat model (female human angiotensinogen × male human renin) reduced trophoblast invasion, with no effect on spiral artery remodeling [57]. In line with studies on the function of uNK cells in humans, these studies therefore also show that the numbers of uNK cells are decreased in preeclampsia and that the function of uNK cells may have changed in preeclampsia; while in normal pregnancy uNK cells inhibit trophoblast invasion, in preeclampsia they may promote trophoblast invasion. Further studies into uNK cell function in preeclampsia and in models for preeclampsia are needed.

Healthy pregnancy
Macrophages are important for the detection, ingestion and processing of foreign material, dead cells and cell debris [58]. Various subsets of macrophages have been described, with M1 (classically activated macrophages) and M2 (alternatively activated macrophages) macrophages at the extreme ends [59,60]. M1 macrophages are microbicidal and proinflammatory, M2 macrophages are immunomodulatory, inducing tolerance and resolution of inflammation [59].
Macrophages are present in the placental bed at all times during pregnancy [61], comprising 20-30% of all decidual leukocytes in early pregnancy [62]. The number of placental bed macrophages decreases with advancing gestational age [37]. Placental bed macrophages are mainly of the M2 subset [16,[63][64][65][66][67] and produce many factors, such as factors associated with immunomodulation, such as the cytokines IL-4, IL-10, TNFα, factors associated with angiogenesis (for instance angiogenin or VEGF) and proteases (MMPs) [68][69][70]. However, they may not be typical M2 macrophages, since they are not typically induced by Th2 cytokines, but by M-CSF and IL-10 [71]. Macrophages can be found near spiral arteries that are in the process of early remodeling, i.e. spiral arteries in the absence of extravillous trophoblast [18]. This may suggest that macrophages prepare spiral arteries for further remodeling by trophoblast cells. Macrophages around the spiral arteries may also engulf apoptotic cells, formed during remodeling, and thereby preventing the release of proinflammatory substances into the decidua [72,73].

Preeclampsia
Studies on macrophages in the placental bed in preeclampsia have mainly been done on placental bed tissue after parturition. Some of these studies reported decreased numbers of macrophages in the decidua of preeclamptic patients [37,74]. However, various other studies have reported increased numbers of macrophages in the placental bed of preeclamptic patients [67,75,76]. Differences between studies may be due to for instance different methods used (flow cytometric study vs immunohistochemical studies), different antibodies used or different locations in the placental bed being studied. This latter is especially important, since differences in macrophage numbers seem to be regional, as increased numbers of macrophages were found around the spiral arteries of preeclamptic patients but not necessarily at other sites [77]. This increased numbers of macrophages around the spiral arteries may be due to an increased presence of chemotactic factors for macrophages [77][78][79] found in the placental bed of preeclamptic patients as compared with the placental bed of healthy control women.
Not only numbers of macrophages were different in preeclamptic patients, macrophages were also differently activated in preeclampsia [77,[79][80][81][82]. Recent in vitro studies showed that macrophages migrate towards invading trophoblast [83], while other groups have shown that activated macrophages in vitro are able to inhibit trophoblast invasion and spiral artery remodeling [84,85]. Therefore, the presence of activated macrophages may play a role in the decreased trophoblast invasion and spiral artery remodeling in preeclampsia.
The different activation status of placental bed macrophages in preeclampsia may be related to a decreased number of M2 macrophages and an increased number of M1 macrophages in the placental bed of preeclamptic women [67]. This is in concordance with increased proinflammatory [86] and decreased anti-inflammatory cytokines in the placenta of preeclamptic women [87,88]. The polarization state of the macrophages may be important for their function in the placental bed, since conditioned medium of M1 macrophages inhibited trophoblast motility and trophoblast tube formation as compared with conditioned medium from M2 macrophages [89].
It is unknown whether the increased numbers and different activational state of macrophages in the placental bed of preeclamptic women is the cause or the consequence of preeclampsia. We have tried to shed light on this question by studying expression of immunological genes in early decidua from women who later developed pregnancyinduced hypertension (PIH; including preeclampsia). We observed that in the early decidua CD68 mRNA expression (a general macrophage marker) was increased and the CD206/CD68 mRNA ratio (CD206 is a marker for M2 macrophages) was decreased in women who later developed pregnancy induced hypertension (including preeclampsia) [90]. These data indicate that the increased numbers of M1 macrophages and decreased numbers of M2 macrophages may already be present before the onset preeclampsia and thus that macrophages may indeed play a role in the pathophysiology of preeclampsia.

Healthy pregnancy
In both mice and rats, macrophages are present in the placental bed, both in the decidua and the myometrium [52,91,92]. They are found scattered throughout the mesometrial triangle and also around the spiral arteries, but with no apparent relation with remodeling [52,93]. At the end of pregnancy, macrophages are the largest population of immune cells in the mesometrial triangle [52,94]. In the rat, we found only a few M2 macrophages in the mesometrial triangle of healthy pregnant rats at the end of pregnancy [52]. It remains to be shown whether macrophages have a role in spiral artery remodeling in the mesometrial triangle in rats and mice.

Preeclampsia
In an animal model for preeclampsia induced by multiple doses of LPS in pregnant rats, decreased trophoblast invasion and spiral artery remodeling after LPS was associated with increased numbers of CD68 positive macrophages [95]. In a different model for preeclampsia (ATP infusion), we found no differences in total macrophages in the mesometrial triangle between control and ATP infused pregnant rats [52], however, numbers of activated macrophages were increased 1 day after ATP infusion, suggesting that activated macrophages may play a role in the pathophysiology of preeclampsia in this model [52]. Further studies into the role of macrophages in the placental bed in animal models may shed more light on the function of macrophages in the placental bed in human pregnancy and preeclampsia.

Healthy pregnancy
Dendritic cells are innate immune cells. They are antigen presenting cells, which are major players in adaptive immune responses. They capture antigen in the periphery, after which they migrate to the lymph nodes, where the present the antigen to T cells. This results in an expansion and polarization of T cells and finally in an antigen specific immune response [96]. Dendritic cells are present in the cycling endometrium and in the placental bed [97]. The number of dendritic cells decrease in early pregnancy as compared with the non-pregnant endometrium [98]. The numbers of dendritic cells are low in the early placental bed, much lower than the number of macrophages [99], and the dendritic cells appear to be scattered throughout the placental bed [100]. Although, both mature (CD83 + ) and immature dendritic cells (DC-SIGN positive) are found in the placental bed, the number of immature dendritic cells is much higher than the number of mature dendritic cells [99]. This may be due to the production of factors by decidual stromal cells inhibiting maturation of dendritic cells in the placental bed [101]. Thus, most dendritic cells appear to be in a resting immature state, which main function is to maintain immune tolerance [96]. This suggestion is in line with the finding in in vitro studies, which suggested that dendritic cells may play a role in skewing immune responses towards Th2 type responses [102]. Moreover, a population of CD14+DC-SIGN positive dendritic cells, which is only present in the pregnant placental bed and not in the non-pregnant endometrium, has been shown to be able to induce Tregs in vitro, suggesting that this population of placental bed dendritic cells may also induce Treg cells in the placental bed [103].
In the human placental bed, DCs seem to closely interact with uNK cells, since DC-SIGN positive dendritic cells were found to have close contact with CD56 + CD16-uNK cells [99]. This suggests close collaboration between uNK cells and DC cells, which is in line with the close collaboration between NK cells and dendritic cells in other tissues [104]. There is indeed evidence of close collaboration of uterine DCs and NK cells from in vitro studies that show enhanced proliferation and activation of NK cells after coculture with decidual DCs [105], while human DCs improve their ability to induce Treg after interaction with uNK cells [106].

Preeclampsia
Only a few studies have evaluated dendritic cells in the placental bed in preeclampsia. Current studies show that the total numbers of dendritic cells in the placental bed may be similar or higher in the placental bed of preeclamptic women [103,107]. The proportion of mature dendritic cells was significantly higher in the decidua of preeclamptic patients as compared with healthy pregnant women [108]. This may suggest activation of dendritic cells in the preeclamptic placental bed, which may result in activation of Th1 cells and inflammation in the placental bed [108]. Although the population of CD14+DC-SIGN positive dendritic cells, as described by Hsu [103], did not change in preeclamptic pregnancies, these cells from the preeclamptic placental bed were less able to induce regulatory T cells [103].

Healthy pregnancy
Mouse models have been instrumental in evaluating the role of DC cells in the placental bed. Similar to the human uterus, also in the mouse uterus, DC are present in the non-pregnant state. Highest DC numbers can be found during the estrus phase of the ovarian cycle [109,110]. Decidual dendritic cells were lower in the decidua as compared with the non-pregnant uterus in mice [111]. DC function is inhibited by the decidua and therefore placental bed DCs do not contribute to an anti-fetal/placental T cell response [111] [112]. However, the dendritic cells do have an immunomodulatory role, which was for instance shown by a study in the abortion prone mouse mating CBA/J x DBA/2J [113]. In the placental bed of these abortion prone mothers, the number of DC is decreased. However, adoptive transfer of syngeneic DCs in these mothers, decreased the number of fetal resorptions [113], by inducing pregnancy protective CD8 and γδ cell populations as well as increased expression of the tolerogenic TGF-beta [113].
Apart from the immunomodulatory role of placental bed DCs, mouse models have shown that DCs are also important in the development of the placental bed. Depletion of DCs in early pregnancy in mice, resulted in decreased size of the implantation sites [114]. These mice also showed signs of fetal resorption [114] and an inhibited development of the decidua. This is associated with impaired expansion of the decidual vascular bed together with decreased vessel permeability and blood flow [115,116]. These data are in line with the suggestion that DCs play an important role in angiogenesis in the placental bed [117].
Increased numbers of DCs at the implantation sites did not affect the implantation size or induced fetal resorptions [114]. However, depletion of uNK cells in the presence of increased numbers of DCs, led to decreased implantation sizes and complete pregnancy failure [114]. Also exogenous administration of IL-10 in mice with increased numbers of DC and depleted NK cells, rescued pregnancy, indicating that IL-10 production by uNK cells, may be important in the NK-DC dialog in early pregnancy [118]. These results suggest that, in accordance with human studies showing in vitro interaction DCs with uNK in the placental bed cells, indeed, in vivo uNK cells and DCs closely collaborate. This dialog between placental bed DC and uNK is not only important for immune regulation in the placental bed, but also for decidual angiogenesis [117]. Moreover, in vitro studies have shown that the dialog between DC and NK cells also affected uterine cell proliferation induced by trophoblast cells [119]. This could have implications for endometrial decidualization in early pregnancy.

Healthy pregnancy
Mast cells are innate immune cells that circulate in the blood in an immature form and migrate to the tissues, such as mucosal tissues, amongst which the placental bed, in which they mature under the influence of various factors, such as factors from the environment. This indicates that mast cells may mature differently in different tissues [120]. Indeed, mast cells in the decidua differ from mast cells in other tissues [121].
Mast cells are present in the endometrium of non-pregnant women, with no differences in numbers of mast cells between the follicular phase and the luteal phase of the ovarian cycle [122,123]. However, premenstrually, mast cells seem to be activated [122]. Mast cells are also present in the uterus during pregnancy [124]. The density of mast cells was higher in the pregnant uterus as compared to the non-pregnant uterus [125]. In the placental bed, mast cells were not only closely located to decidual extra villous trophoblast, they also seem to interact with these cells [126]. In vitro studies have shown that conditioned medium from mast cells increased trophoblast migration [126], suggesting a role for mast cells in trophoblast invasion in the placental bed.

Preeclampsia
A few studies evaluated mast cell numbers in the placenta in preeclamptic patients [127][128][129]. Presence of mast cells was studied in the villous part of the placenta with conflicting results. While two studies found increased numbers of mast cells in the villous part of the placenta [127,129], the other study [128] found decreased numbers of mast cells in the villous part of the placenta in preeclamptic women as compared with healthy pregnant women. However, to the best of our knowledge, no studies have looked at mast cells in the preeclamptic placental bed.

Healthy pregnancy
Most of knowledge on the role of mast cells in development of the placental bed arises from mice studies. In the mouse, mast cells are also already present in the non-pregnant uterus, with numbers peaking at estrus [130] and an increase in number during early pregnancy [130,131]. In abortion prone mice, numbers of mast cells do not increase during early pregnancy [130], suggesting a role for these cells in early healthy pregnancy. Since adoptive transfer of Tregs in these mice increased the numbers of mast cells in the placental bed and rescued pregnancy, Tregs seem to regulate the number of mast cells in the placental bed [130]. The role of mast cells in early pregnancy is further shown by studies in mast cell deficient mice. These studies have shown that mast cells indeed play a role in early pregnancy, mainly in implantation and placentation. Studies in mouse deficient for mast cells, showed a diminished size of the implantation site as compared with control mice with sufficient mast cells [131], while the transfer of bone marrow derived mast cells into mast cell deficient mice restored the size of the implantation sites [131]. Mast cell deficiency in pregnant mice also resulted in decreased spiral artery remodeling, fetal growth restriction [126] and decreased placental size [131]. Mast cells may also play a role in angiogenesis, since degranulation of mast cells induced angiogenesis in the cervix [132,133].
Like dendritic cells, also mast cells seem to interact with uNK cells. Mice deficient in either mast cells or NK cells show impairment in spiral artery remodeling as well as fetal growth retardation [134]. Mice deficient in both mast cells and NK cells show a much more marked impairment in spiral artery remodeling and fetal growth restriction [135]. Together with the observation that mice deficient in mast cells show increased numbers of uNK cells in the placental bed, while mice deficient for uNK cells show increased numbers of mast cells at the implantation site [134], this suggests the both cell types jointly affect placental bed development in early pregnancy and counterbalance each other.

Concluding remarks
Important advances have been made in understanding the role of innate immune cells, such as uNK, macrophages, dendritic cells and mast cells in the placental bed of healthy and preeclamptic pregnancies (Fig. 1). These innate immune cells are important for normal development of the placental bed. Since these cells are often found in the vicinity of invading extravillous trophoblast as well as of spiral arteries, they seem to play a role in trophoblast invasion and spiral artery remodeling. This is also apparent from the fact that decreased numbers of uNK cells are found in association with decreased spiral artery remodeling and increased numbers of macrophages are found around the spiral arteries in preeclampsia. Although there is no consensus on the numbers of dendritic cells in the placental bed in preeclampsia, the number of mature dendritic cells are increased, which may be associated with a different immune regulation, such as increased inflammation and numbers of Th1 cells. Unfortunately, no data are available on mast cells in preeclampsia. In healthy pregnant women, the innate immune cells mainly have a regulatory phenotype (i.e. producing regulatory factors, such as regulatory cytokines, growth factors, angiogenic factors and factors associated with tissue remodeling) (Fig. 1, left panel). In preeclampsia, innate immune cells seem to have a more proinflammatory phenotype. This results in production of different factors, such as increased production of proinflammatory cytokines and decreased production of tissue remodeling factors (Fig. 1, right panel). These changes in preeclampsia may result in decreased spiral artery remodeling and decreased trophoblast invasion. The role of dendritic cells and mast cells needs further investigation.
More studies on the exact role and timing of presence of innate immune cells in the placental bed are needed. Especially the role of dendritic cells and mast cells and the interaction between the different innate immune cells in the placental bed needs further investigation.
As human placental bed tissue samples at the time of trophoblast invasions and spiral artery remodeling are very difficult to collect, studies into these cells should take better advantage of animal models. Relatively low numbers of studies on the role of uNK cells and uterine macrophages in preeclampsia are performed in rat and mouse models, despite the fact that these cells seem to have functions similar to human uNK cells and uterine macrophages. No studies on mast cells and dendritic cells have been performed in animal models for preeclampsia. Studies on innate immune cells during the course of healthy pregnancy and experimental preeclampsia in mice and rats could lead to better insight into the function and dysfunction of these cells during pregnancy and preeclampsia in humans and eventually to novel targets for treatment of preeclampsia.

Conflicts of interest
We have no conflict of interest.