Introduction

This chapter focuses on the unique process by which the human placenta normally forms, and how changes can lead to serious pregnancy complications such as preeclampsia. Special emphasis will be placed on work from the principal author’s laboratory that led to the discovery that the subset of placental cells, termed cytotrophoblasts, that invade the uterus and form vascular connections with the resident maternal vessels undergo a novel transformation from an epithelial cell of ectodermal origin to a vascular-like cell with a myriad of endothelial-like properties.

The Microanatomy of Normal Human Placentation

The human placenta’s unique anatomy ( Figs. 5.1 and 5.2 ) is due in large part to differentiation of its ectodermally derived progenitors, termed cytotrophoblasts. How these cells differentiate determines whether chorionic villi, the placenta’s functional units, float in maternal blood or anchor the conceptus to the uterine wall. In floating villi, cytotrophoblasts differentiate by fusing to form multinucleate syncytiotrophoblasts whose primary function – transport – is ideally suited to their location at the villus surface. In anchoring villi, cytotrophoblasts also fuse, but many remain as single cells that detach from their basement membrane and aggregate to form cell columns. Cytotrophoblasts at the distal ends of these columns attach to, then deeply invade the uterus (interstitial invasion) and its arterioles (endovascular invasion). As a result of endovascular invasion, the cells replace the endothelial and muscular linings of uterine arterioles, a process that initiates maternal blood flow to the placenta and greatly enlarges the vessel diameter. Paradoxically, the cells invade only the superficial portions of uterine venules.

(A) Diagram of a longitudinal section of an anchoring chorionic villus (AV) at the fetal–maternal interface at about 10 weeks of gestational age. The anchoring villus (AV) functions as a bridge between the fetal and maternal compartments, whereas floating villi (FV) are suspended in the intervillous space and are bathed by maternal blood. Cytotrophoblasts (CTB) in AV (Zone I) form cell columns (Zones II and III). CTB then invade the uterine interstitium (decidua and first third of the myometrium: Zone IV) and maternal vasculature (Zone V), thereby anchoring the fetus to the mother and accessing the maternal circulation. Zone designations mark areas in which CTB have distinct patterns of stage-specific antigen expression. (B) Diagram of a uterine (spiral) artery in which endovascular invasion is in progress (10–18 weeks of gestation). Endometrial and then myometrial segments of spiral arteries are modified progressively. (a) In fully modified regions the vessel diameter is large. CTB are present in the lumen and occupy the entire surface of the vessel wall. A discrete muscular layer (tunica media) is not evident. (b) Partially modified vessel segments. CTB and maternal endothelium occupy discrete regions of the vessel wall. In areas of intersection, CTB appear to lie deep to the endothelium and in contact with the vessel wall. (c) Unmodified vessel segments in the myometrium. Vessel segments in the superficial third of the myometrium will become modified when endovascular invasion reaches its fullest extent (by 22 weeks), while deeper segments of the same artery will retain their normal structure.

In normal pregnancy (left) fetal cytotrophoblasts (stained with anti-cytokeratin) from the anchoring villi (AV) of the placenta invade the maternal uterine blood vessels (BV). In preeclampsia (right) fetal cells fail to penetrate the uterine vasculature (arrows and arrowheads).

Cells that participate in endovascular invasion have two types of interactions with maternal arterioles. In the first, large aggregates of these fetal cells are found primarily inside the vessel lumen. These aggregates can either lie adjacent to the apical surface of the resident endothelium or replace it such that they appear directly attached to the vessel wall. In the second type of interaction, cytotrophoblasts are found within the vessel wall rather than in the lumen. In this position, they colonize the smooth-muscle layer of the vessel and lie subjacent to the endothelium. These different types of interactions may be progressive stages in a single process, or indicative of different strategies by which cytotrophoblasts accomplish endovascular invasion. In either case, the stage in which fetal cytotrophoblasts cohabit with maternal endothelium in the spiral arterioles is transient. By late second trimester these vessels are lined exclusively by cytotrophoblasts, and endothelial cells are no longer visible in either the endometrial or the superficial portions of their myometrial segments.

The Microanatomy of Abnormal Human Placentation in Preeclampsia

Preeclampsia is a disease that adversely affects 7–10% of first pregnancies in the United States. The mother shows signs and symptoms that suggest widespread alterations in endothelial function (e.g., high blood pressure, proteinuria, and edema ). In some cases fetal growth slows, which leads to intrauterine growth retardation. The severity of the disease varies greatly (see Chapter 2 ). In its mildest form the signs/symptoms appear near term and resolve after birth, with no lasting effects on either the mother or the child. In its severest form the signs/symptoms often occur in the second or early third trimesters. If they cannot be controlled, the only option is delivery, with consequent iatrogenic fetal prematurity. Owing to the latter form of the disease, preeclampsia and hypertensive diseases of pregnancy are leading causes of maternal death and contribute significantly to premature deliveries in the United States.

Although the cause of preeclampsia is unknown, the accumulated evidence strongly implicates the placenta. Anatomic examination shows that the area of the placenta most affected by this syndrome is the fetal–maternal interface ( Fig. 5.2 ). Cytotrophoblast invasion of the uterus is shallow, and endovascular invasion does not proceed beyond the terminal portions of the spiral arterioles. The effect of preeclampsia on endovascular invasion is particularly evident when interactions between fetal cytotrophoblasts and maternal endothelial cells are studied in detail. Serial sections through placental bed biopsies show that few of the spiral arterioles contain cytotrophoblasts. Instead, most cytotrophoblasts remain at some distance from these vessels. Where endovascular cytotrophoblasts are detected, their invasion is limited to the portion of the vessel that spans the superficial decidua. Even when the cytotrophoblasts gain access to the lumina, they usually fail to form tight aggregates among themselves, or to spread out on the vessel wall, as is observed for cytotrophoblasts in control samples matched for gestational age. Instead they tend to remain as individual rounded cells, suggesting that they are poorly anchored to the vessel wall. Thus, cytotrophoblasts in preeclampsia not only have a limited capacity for endovascular invasion but also display an altered morphology in their interactions with maternal arterioles.

Because of these alterations in endovascular invasion, the maternal vessels of preeclamptic patients do not undergo the complete spectrum of physiologic changes that normally occur (e.g., loss of their endothelial lining and musculoelastic tissue); the mean external diameter of the myometrial vessels is less than half that of similar vessels from uncomplicated pregnancies. In addition, not as many vessels show evidence of cytotrophoblast invasion. Thus, the architecture of these vessels precludes an adequate response to gestation-related fetal demands for increased blood flow.

The Road to Preeclampsia

Several reviews summarize the current state of knowledge regarding preeclampsia. This interest reflects the clinical importance of this condition, which continues to be a primary driver of obstetric care in the developed world. This is due, in large part, to the sometimes rapid onset of preeclampsia – hence the name “preeclampsia” (derived from the Greek εχλαμψιεζ, sudden flash or development). A somewhat clearer picture of the pathogenesis of preeclampsia has begun to emerge. A two-stage model has been proposed in which the initiating event, poor placentation, is thought to occur early on. This concept is supported by several studies that document the association between reduced blood flow to the placenta before 20 weeks of gestation, as determined by color Doppler ultrasound evaluation of spiral arterial blood flow, and a greatly increased risk of developing preeclampsia ( Fig. 5.1 ).

The second stage of preeclampsia is the maternal response to abnormal placentation. Systemic endothelial dysfunction is thought to be an important common denominator. For example, serum from pregnancies complicated by preeclampsia induces activation of umbilical endothelial cells in vitro , experimental evidence in support of this theory. Interestingly, increased pressor sensitivity and abnormal-flow-induced vasodilatation can be demonstrated before the clinical signs appear, as can sensitivity to angiotensin II. Much of the current data point to an imbalance in circulating factors with relevant biological activities. For example, molecules that are associated with endothelial dysfunction, such as fibronectin, factor VIII antigen, and thrombomodulin, are found at higher levels in the blood of women with preeclampsia. The concentration of S -nitrosoalbumin, a major nitric oxide reservoir, also increases as does endothelin. Concomitantly, the levels of endothelial-derived vasodilators, such as prostacyclin, decrease. Activators of peroxisome proliferator-activated receptor gamma, which functions in metabolic and immune responses, show the same pattern, with abnormally low levels evident weeks and in some cases months before the clinical diagnosis of preeclampsia is made. Maternal blood levels of the soluble form of VEGFR-1 (sFlt-1) rise as do those of the TGFβ receptor, endoglin, with a concomitant failure of placental growth factor levels to reach normal values (see Chapter 6 ). These changes have been attributed by one group of investigators to a deficiency in catechol- O -methyltransferase and 2-methoxyestradiol. Recent work in animal models has also implicated agonistic autoantibodies to the angiotensin receptor in the cascade of events that lead to preeclampsia.

Oxygen Tension Regulates Human Cytotrophoblast Proliferation and Differentiation In Vitro

Information about the morphological and molecular aspects of cytotrophoblast invasion in normal pregnancy and in preeclampsia has been used to formulate hypotheses about the regulatory factors involved. Specifically, in normal pregnancy cytotrophoblasts invade large-bore arterioles, where they are in contact with well-oxygenated maternal blood. But in preeclampsia, invasive cytotrophoblasts are relatively hypoxic. Another important consideration is that placental blood flow changes dramatically during early pregnancy. During much of the first trimester there is little endovascular invasion, so maternal blood flow to the placenta is at a minimum. The oxygen pressures of the intervillous space (that is, at the uterine surface) and within the endometrium are estimated to be approximately 18 mm Hg and 40 mm Hg, respectively, at 8–10 weeks of gestation. Afterwards, endovascular invasion proceeds rapidly; cytotrophoblasts are in direct contact with blood from maternal spiral arterioles, which could have a mean oxygen pressure as high as 90–100 mm Hg. Thus, as cytotrophoblasts invade the uterus during the first half of pregnancy, they encounter a steep, positive oxygen tension gradient. These observations, together with the results of our own experiments conducted on isolated cytotrophoblasts, suggested that oxygen tension might regulate cytotrophoblast proliferation and differentiation along the invasive pathway.

First, we used immunolocalization techniques to study the relationship between cytotrophoblast proliferation and differentiation in situ . Cytotrophoblasts in columns (i.e., cells in the initial stages of differentiation) reacted with an antibody against the Ki67 antigen, which is indicative of DNA synthesis. Distal to this region, anti-Ki67 staining abruptly stopped, and the cytotrophoblasts intricately modulated their expression of stage-specific antigens, including integrin cell adhesion molecules, matrix metalloproteinase-9, HLA-G (a cytotrophoblast class Ib major histocompatibility complex molecule ), and human placental lactogen. These results suggested that during differentiation along the invasive pathway, cytotrophoblasts first undergo mitosis, then exit the cell cycle and modulate their expression of stage-specific antigens.

As an in vitro model system for testing this hypothesis, we used organ cultures of anchoring villi explanted from early gestation (6–8-week) placentas onto an extracellular matrix substrate. Some of the anchoring villi were cultured for 72 hours in a standard tissue culture incubator (20% O ₂ or 98 mm Hg). Figure 5.3A shows a section of one such control villus that was stained with an antibody that recognizes cytokeratin to demonstrate syncytiotrophoblasts and cytotrophoblasts. The attached cell columns were clearly visible. To assess the cells’ ability to synthesize DNA, the villi were incubated with bromodeoxyuridine (BrdU). Incorporation was detected in the cytoplasm, but not the nuclei, of syncytiotrophoblasts. Few or none of the cells in columns incorporated BrdU ( Fig. 5.3B ). Other anchoring villi were maintained in a hypoxic atmosphere (2% O ₂ or 14 mm Hg). After 72 hours, cytokeratin staining showed prominent cell columns ( Fig. 5.3C ), and the nuclei of many of the cytotrophoblasts in these columns incorporated BrdU ( Fig. 5.3D ). Because cytotrophoblasts were the only cells that entered S phase, we also compared the ability of anchoring villus explants cultured under standard and hypoxic conditions to incorporate [ ³ H]thymidine. Villus explants cultured under hypoxic conditions (2% O ₂ ) incorporated 3.3±1.2 times more [ ³ H]thymidine than villi cultured under standard conditions (20% O ₂ ). In contrast, [ ³ H]thymidine incorporation by explants cultured in a 6% O ₂ atmosphere (40 mm Hg) was no different than in control villi. Taken together, these results suggest that a hypoxic environment, comparable to that encountered by early gestation cytotrophoblasts in the intervillous space, stimulates the cells to enter S phase.

Low oxygen (2% O ₂ ) stimulates cytotrophoblast BrdU incorporation in vitro . Anchoring villi (AV) from 6–8-week placentas were cultured on Matrigel (m) for 72 hours in either 20% O ₂ (A, B) or 2% O ₂ (C, D). By the end of the culture period, fetal cytotrophoblasts migrated into the Matrigel (F→m). To assess cell proliferation, BrdU was added to the medium. Tissue sections of the villi were stained with anti-cytokeratin (ck; A, C), which recognizes syncytiotrophoblasts (ST) and cytotrophoblasts (CTB) but not cells in the villus core (vc); and with anti-BrdU (B, D), which detects cells in S phase. Villus explants maintained in 2% O ₂ (C) formed much more prominent columns (COL) with a larger proportion of CTB nuclei that incorporated BrdU (D) than explants cultured in 20% O ₂ (A, B).

(Reprinted with permission. Copyright 1997 American Association for the Advancement of Science.)

Cytokeratin staining also showed that the cell columns associated with anchoring villi cultured under hypoxic conditions were larger than cell columns of control villi cultured under standard conditions (compare Figs. 5.3A and 5.3C ). To quantify this, we made serial sections of villus explants maintained in either 20% or 2% O ₂ , and then counted the number of cells in columns. Under hypoxic culture conditions, the columns contained triple the number of cells present in columns maintained in 20% oxygen. These results indicate that hypoxia stimulates cytotrophoblasts in cell columns to proliferate.

The next series of studies were based on the hypothesis that the hypoxia-induced changes in the cells’ proliferative capacity would be reflected by changes in their expression of proteins that regulate passage through the cell cycle ( Fig. 5.4 ). With regard to the G2 to M transition, we were particularly interested in their cyclin B expression, since threshold levels of this protein are required for cells to enter mitosis. Immunoblotting of cell extracts showed that after 3 days in culture, anchoring villi maintained in 2% O ₂ contained 3.1 times more cyclin B than did villi maintained in 20% O ₂ ( Fig. 5.4A ). Immunolocalization experiments confirmed that cyclin B was primarily expressed by cytotrophoblasts (not shown). Since p21 ^WF1/CIP1 abundance has been correlated with cell cycle arrest, we also examined the effects of oxygen tension on cytotrophoblast expression of this protein. Very little p21 ^WF1/CIP1 expression was detected in cell extracts of anchoring villi maintained for 72 hours in 2% O ₂ , but expression increased 3.8-fold in anchoring villi maintained for the same time period in 20% O ₂ ( Fig. 5.4B ). Immunolocalization experiments confirmed that p21 ^WF1/CIP1 was primarily expressed by cytotrophoblasts. These results, replicated in five separate experiments, confirm that culturing anchoring villi in 20% O ₂ induces cytotrophoblasts in the attached cell columns to undergo cell cycle arrest, whereas culturing them in 2% O ₂ induces them to enter mitosis.

Hypoxia induces changes in cytotrophoblast expression of proteins that regulate progression through the cell cycle. (A) Villus explants cultured for 72 hours in 2% O ₂ contained 3.1 times more cyclin B than did villi maintained under standard culture conditions (20% O ₂ ) for the same length of time. (B) Expression of p21 increased 3.8-fold in anchoring villi cultured for 72 hours in 20% O ₂ as compared to 2% O ₂ .

(Reprinted with permission. Copyright 1997 American Association for the Advancement of Science.)

Changes in proliferative capacity are often accompanied by concomitant changes in differentiation. Accordingly, we investigated the effects of hypoxia on the ability of cytotrophoblasts to differentiate along the invasive pathway ( Fig. 5.5 ). Under standard tissue culture conditions, cytotrophoblasts migrated from the cell columns and modulated their expression of stage-specific antigens, as they do during uterine invasion in vivo . For example, they began to express integrin α1, a laminin-collagen receptor that is required for invasiveness in vitro . Both differentiated cytotrophoblasts and villus stromal cells expressed this antigen ( Fig. 5.5B ). When cultured under hypoxic conditions, cytotrophoblasts failed to stain for integrin α1, but stromal cells continued to express this molecule, suggesting the observed effects were cell-type specific ( Fig. 5.5E ). Hypoxia also reduced cytotrophoblast staining for human placental lactogen, another antigen that is expressed once the cells differentiate. However, lowering the O ₂ tension did not change cytotrophoblast expression of other stage-specific antigens, such as HLA-G ( Figs. 5.5C and 5.5F ) and integrins α5β1 and αVβ3 (not shown). These results suggest that hypoxia produces selective deficits in the ability of cytotrophoblasts to differentiate along the invasive pathway.

Some aspects of cytotrophoblast differentiation/invasion are arrested in hypoxia. Anchoring villi (AV) from 6–8-week placentas were cultured on Matrigel (m) for 72 hours in either 20% O ₂ (A–C) or 2% O ₂ (D–F). Tissue sections of the villi were stained with anti-cytokeratin (ck; A, D), anti-integrin α1 (B, E) or anti-HLA-G (C, F). Cytotrophoblasts (CTB) that composed the cell columns (COL) of villus explants that were cultured in 20% O ₂ upregulated both integrin α1 (B) and HLA-G expression (C). In contrast, cytotrophoblasts in anchoring villus columns maintained in 2% O ₂ failed to express integrin α1, although constituents of the villus core continued to express this adhesion molecule (E). But not all aspects of differentiation were impaired; the cells upregulated HLA-G expression normally (F). ST, syncytiotrophoblast; vc, villus core.

(Reprinted with permission. Copyright 1997 American Association for the Advancement of Science.)

The effects of oxygen tension on the proliferative capacity of cytotrophoblasts could help explain some of the interesting features of normal placental development. Before cytotrophoblast invasion of maternal vessels establishes the uteroplacental circulation (≤10 weeks), the conceptus is in a relatively hypoxic environment. During this period, placental mass increases much more rapidly than that of the embryo proper. Histological sections of early-stage pregnant human uteri show bi-laminar embryos surrounded by thousands of trophoblast cells. The fact that hypoxia stimulates cytotrophoblasts, but not most other cells, to undergo mitosis could help account for the discrepancy in size between the embryo and the placenta, which continues well into the second trimester of pregnancy. Although this phenomenon is poorly understood at a mechanistic level, we have recently shown (see below) that cytotrophoblasts within the uterine wall mimic a vascular adhesion molecule phenotype. In other tissues hypoxia induces vascular endothelial growth factor production, which stimulates endothelial cell proliferation. This raises the possibility that similar regulatory pathways operate during placental development.

The effects of oxygen tension on cytotrophoblast differentiation/invasion could also have important implications. Relatively high oxygen tension promotes cytotrophoblast differentiation and could help explain why these cells extensively invade the arterial rather than the venous side of the uterine circulation. Conversely, if cytotrophoblasts do not gain access to an adequate supply of maternal arterial blood, their ability to differentiate into fully invasive cells may be impaired. We suggest that the latter scenario could be a contributing factor to pregnancy-associated diseases, such as preeclampsia, that are associated with abnormally shallow cytotrophoblast invasion and faulty differentiation, as evidenced by their inability to upregulate integrin α1 expression. These results also prompted us to consider the possibility that the profound effects of oxygen on invasive cytotrophoblasts might be indicative of their ability to assume a vascular-like phenotype.

During Normal Pregnancy, Invasive Cytotrophoblasts Modulate their Adhesion Molecule Repertoire to Mimic That of Vascular Cells

We tested the hypotheses that invasive cytotrophoblasts mimic broadly the adhesion phenotype of the endothelial cells they replace, and that these changes in adhesion phenotype have the net effect of enhancing cytotrophoblast motility and invasiveness. To test these hypotheses, we first stained tissue sections of the fetal–maternal interface for specific integrins, cadherins, and immunoglobulin family adhesion receptors that are characteristic of endothelial cells and leukocytes. Subsequent experiments tested the functional consequences for cytotrophoblast adhesion and invasion of expressing the particular adhesion receptors that were upregulated during cytotrophoblast differentiation.

First, we examined the distribution patterns of αV integrin family members. These molecules are of particular interest because of their regulated expression on endothelial cells during angiogenesis and their upregulation on some types of metastatic tumor cells. αV family members displayed unique and highly specific spatial staining patterns on cytotrophoblasts in anchoring villi and the placental bed. An antibody specific for the αVβ5 complex stained the cytotrophoblast monolayer in chorionic villi. Staining was uniform over the entire cell surface. The syncytiotrophoblast layer, and cytotrophoblasts in cell columns and the placental bed, did not stain for αVβ5. In contrast, anti-αVβ6 stained only those chorionic villus cytotrophoblasts that were at sites of column formation. The cytotrophoblast layer still in contact with basement membrane stained brightly, while the first layer of the cell column showed reduced staining. The rest of the cytotrophoblasts in chorionic villi, cytotrophoblasts in more distal regions of cell columns, and cytotrophoblasts within placental bed and vasculature did not stain for αVβ6, documenting a specific association of this integrin with initiation of column formation. In yet a different pattern, staining for anti-αVβ3 was weak or not detected on villus cytotrophoblasts or on cytotrophoblasts in the initial layers of cell columns. However, strong staining was detected on cytotrophoblasts within the uterine wall and vasculature. Thus, individual members of the αV family, like those of the β1 family, are spatially regulated during cytotrophoblast differentiation. Of particular relevance is the observation that αVβ3 integrin, whose expression on endothelial cells is stimulated by angiogenic factors, is prominent on cytotrophoblasts that have invaded the uterine wall and maternal vasculature.

Since blocking αVβ3 function suppresses endothelial migration during angiogenesis, we determined whether perturbing its interactions also affects cytotrophoblast invasion in vitro . Freshly isolated first-trimester cytotrophoblasts were plated for 48 hours on Matrigel-coated Transwell filters in the presence of control mouse IgG or the complex-specific anti-αVβ3 IgG, LM609. Cytotrophoblast invasion was evaluated by counting cells and cellular processes that had invaded the Matrigel barrier and extended through the holes in the Transwell filters. LM609 reduced cytotrophoblast invasion by more than 75% in this assay, indicating that this receptor, like the α1β1 integrin, contributes significantly to the invasive phenotype of cytotrophoblasts.

Next, we examined cadherin switching during cytotrophoblast differentiation in vivo ( Fig. 5.6 ). The cytotrophoblast epithelial monolayer stained strongly for the ubiquitous epithelial cadherin, E-cadherin, in a polarized pattern ( Fig. 5.6A ). Staining was strong on the surfaces of cytotrophoblasts in contact with one another and with the overlying syncytiotrophoblast layer, and was absent at the basal surface of cytotrophoblasts in contact with basement membrane. In cell columns, E-cadherin staining intensity was reduced on cytotrophoblasts near the uterine wall and on cytotrophoblasts within the decidua. This reduction in staining was particularly pronounced in second-trimester tissue. At this stage, E-cadherin staining was also very weak or undetectable on cytotrophoblasts that had colonized maternal blood vessels and on cytotrophoblasts in the surrounding myometrium. All locations of reduced E-cadherin staining were areas in which invasion is active during the first half of gestation. Interestingly, the staining intensity of E-cadherin was strong on cytotrophoblasts in all locations in term placentas, at which time cytotrophoblast invasive activity is poor. Taken together, these data are consistent with the idea that cytotrophoblasts transiently reduce E-cadherin function at times and places of their greatest invasive activity.

E-cadherin staining is reduced and VE-cadherin staining is upregulated in normal, differentiating second-trimester CTB. Sections of second-trimester placental bed tissue were stained with antibody against E-cadherin (A), VE-cadherin (B and D) or cytokeratin (CK, 7D3: C). (A) E-cadherin staining was strong on anchoring villus (AV) CTB and on CTB in the proximal portion of cell columns (Zone II). Staining was sharply reduced on CTB in the distal column (Zone III) and in the uterine interstitium (Zone IV). Staining was not detected on CTB within maternal vessels (Zone V). (B) VE-cadherin was not detected on CTB in AV (although fetal blood vessels in villus stromal core are stained). VE-cadherin was detected on column CTB (B) and on interstitial and endovascular CTB (D). VE-cadherin is also detected on maternal endothelium in vessels that have not been modified by CTB (For description of zones, see Fig. 5.1 ).

Cadherin switching occurs frequently during embryonic development when significant morphogenetic events take place. We therefore stained sections of first- and second-trimester placental tissue with antibodies to other classical cadherins. These tissues did not react with antibodies against P-cadherin, but did stain with three different monoclonal antibodies that recognize the endothelial cadherin, VE-cadherin ( Fig. 5.6B ). In chorionic villi, antibody to VE cadherin did not stain villus cytotrophoblasts, although it stained the endothelium of fetal blood vessels within the villus stroma. In contrast, anti-VE-cadherin stained cytotrophoblasts in cell columns and in the decidua, the very areas in which E-cadherin staining was reduced. VE-cadherin staining was stronger in these areas in second-trimester tissues. In maternal vessels that had not yet been modified by cytotrophoblasts, anti-VE-cadherin stained the endothelial layer strongly. Following endovascular invasion, cytotrophoblasts lining maternal blood vessels also stained strongly for VE-cadherin ( Fig. 5.6D ). Thus, cytotrophoblasts that invade the uterine wall and vasculature express a cadherin characteristic of endothelial cells.

Next, we used function-perturbing anti-cadherin antibodies, in conjunction with the Matrigel invasion assay, to assess the functional consequences of cadherin modulation for cytotrophoblast invasiveness. We plated isolated second-trimester cytotrophoblasts for 48 hours on Matrigel-coated filters in the presence of control IgG or function-perturbing antibodies against VE-cadherin or E-cadherin. By 48 hours, significant invasion was evident in control cytotrophoblasts. In cultures treated with anti-E-cadherin, cytotrophoblast invasiveness increased more than 3-fold, suggesting that E-cadherin normally has a restraining effect on invasiveness. In contrast, antibody against VE-cadherin reduced the invasion of cytotrophoblasts to about 60% of control. This suggests that the presence of VE cadherin normally facilitates cytotrophoblast invasion. Taken together, these functional data suggest that as they differentiate, the cells modulate their cadherin repertoire to one that contributes to their increased invasiveness.

Our data presented thus far indicate that, as they differentiate, cytotrophoblasts downregulate adhesion receptors highly characteristic of epithelial cells (integrin α6β4 and E-cadherin) and upregulate analogous receptors that are expressed on endothelial cells (integrins α1β1 and αVβ3, and VE-cadherin). These observations support our hypothesis that normal cytotrophoblasts undergo a comprehensive switch in phenotype so as to resemble the endothelial cells they replace during endovascular invasion.

We hypothesize that this unusual phenomenon plays an important role in the process whereby these cells form vascular connections with the uterine vessels. Ultimately these connections are so extensive that the spiral arterioles become hybrid structures in which fetal cytotrophoblasts replace the maternal endothelium and much of the highly muscular tunica media. As a result, the diameter of the spiral arterioles increases dramatically, allowing blood flow to the placenta to keep pace with fetal growth. Circumstantial evidence suggests that several of the adhesion molecules whose expression we studied could play an important role in forming these novel vascular connections. In the mouse, for example, targeted disruption of either vascular cell adhesion molecule (VCAM)-1 or α4 expression results in failure of chorioallantoic fusion. It is very interesting to find that cytotrophoblasts are the only cells, other than the endothelium, that express VE-cadherin. In addition, VE-cadherin and platelet-endothelial cell adhesion molecule (PECAM)-1 are the first adhesion receptors expressed by differentiating endothelial cells during early development. αVβ3 expression is upregulated on endothelial cells during angiogenesis by soluble factors that regulate this process. Thus, adhesion receptors that are upregulated as normal cytotrophoblasts differentiate/invade play vital roles in differentiation and expansion of the vasculature.

In Preeclampsia, Invasive Cytotrophoblasts Fail to Switch their Adhesion Molecule Repertoire to Mimic That of Vascular Cells

The next hypothesis tested was that preeclampsia impairs the ability of cytotrophoblasts to express the adhesion molecules that are normally modulated during the unique epithelial-to-vascular transformation that occurs in normal pregnancy. First, we compared cytotrophoblast expression of three members of the αV family (αVβ5, αVβ6, and αVβ3) in placental bed biopsies obtained from control and preeclamptic patients that were matched for gestational age. Preeclampsia changed cytotrophoblast expression of all three αV-family members. When samples were matched for gestational age, fewer preeclamptic cytotrophoblast stem cells stained with an antibody that recognized integrin β5. In contrast, staining for β6 was much brighter in preeclamptic tissue and extended beyond the column to include cytotrophoblasts within the superficial decidua. Of greatest interest, staining for β3 was weak on cytotrophoblasts in all locations; cytotrophoblasts in the uterine wall of preeclamptic patients failed to show strong staining for β3, as did cytotrophoblasts that penetrated the spiral arterioles. Thus, in preeclampsia, differentiating/invading cytotrophoblasts retain expression of αVβ6, which is transiently expressed in remodeling epithelium, and fail to upregulate αVβ3, which is characteristic of angiogenic endothelium. Therefore, as was the case for integrin α1, our analyses of the expression of αV-family members suggest that in preeclampsia, cytotrophoblasts start to differentiate along the invasive pathway but cannot complete this process.

Preeclampsia also had a striking effect on cytotrophoblast cadherin expression ( Fig. 5.7 ). In contrast to control samples, cytotrophoblasts in both the villi and decidua showed strong reactivity with anti-E-cadherin ( Fig. 5.7A ), and staining remained strong even on cytotrophoblasts that had penetrated the superficial portions of uterine arterioles (data not shown). Interestingly, in preeclampsia cytotrophoblasts within the uterine wall tended to exist as large aggregates, rather than as smaller clusters and single cells, as is the case in normal pregnancy. This observation is in accord with the likelihood that E-cadherin mediates strong intercellular adhesion between cytotrophoblasts, as it does in all other normal epithelia examined.

In preeclampsia, E-cadherin staining is retained on placental bed CTB and VE-cadherin staining is not detected. Sections of 27-week severe preeclamptic (SPE: A) and 26-week HELLP tissue (B, C, D) were stained with antibody against E-cadherin (A), VE-cadherin (B and D) or cytokeratin (CK, 7D3: C). (A) E-cadherin staining was strong on CTB in almost all locations. CTB also appeared to be in large aggregates. VE-cadherin was not detected on CTB in cell columns (B) or near blood vessels (D). But the endothelial cells (EC) that line the vessel did stain (arrows).

Strikingly, no VE-cadherin staining was detected on cytotrophoblasts in any location in placental bed specimens obtained from preeclamptic patients; neither cytotrophoblasts in the cell columns ( Fig. 5.7B ) nor the few cells that were found in association with vessels in the superficial decidua expressed VE-cadherin ( Fig. 5.7D ). However, staining for this adhesion molecule was detected on maternal endothelium in the unmodified uterine vessels in preeclamptic placental bed biopsy specimens. Thus, cadherin modulation by cytotrophoblasts in preeclampsia was defective, as shown by the persistence of strong E-cadherin staining and the absence of VE-cadherin staining on cytotrophoblasts in columns and in the superficial decidua.

The results summarized above raise the interesting possibility that the failure of preeclamptic cytotrophoblasts to express vascular-type adhesion molecules, as normal cytotrophoblasts do, impairs their ability to form connections with the uterine vessels. This failure ultimately limits the supply of maternal blood to the placenta and fetus, an effect thought to be closely linked to the pathophysiology of the disease. We also hypothesize that the failure of preeclamptic cytotrophoblasts to make a transition to a vascular cell adhesion phenotype might be part of a broader-spectrum defect in which the cells fail to function properly as endothelium. Such a failure would no doubt have important effects on the maintenance of vascular integrity at the maternal–fetal interface. Clearly, in preeclampsia undifferentiated cytotrophoblasts that fail to mimic the adhesion phenotype of endothelial cells are present in the termini of maternal spiral arterioles. Whether or not the observed defects are related to their propensity to undergo apoptosis is not yet known (reviewed in ). Whether their presence also affects the phenotype of maternal endothelium in deeper segments of the same vessels and/or is linked to the maternal endothelial pathology that is a hallmark of this disease remains to be investigated.

The Pathological Consequences of Abnormal Cytotrophoblast Invasion and Failed Spiral Artery Remodeling

The major effect of abnormal cytotrophoblast invasion and failed remodeling of the maternal spiral arteries is abnormal perfusion of the intervillus space. Before conception the spiral arteries, the terminal branches of the uterine arteries, are typical small muscular vessels, and are richly innervated. During pregnancy there is modification of much of the uterine artery, the most prominent being that of the spiral branches that perfuse the placenta. with a 4 fold dilatation of the terminal portion of the arteries. However, there is also remodeling of the vascular wall throughout the spiral artery’s length, extending as far as the inner third of the myometrium. There is a striking loss of smooth muscle and elastic tissue, rendering the artery unresponsive to neural or humoral signals. The depth of the remodeling is also important. In normal pregnancy the loss of muscle extends beyond the decidua into the inner third of the myometrium. In the non-pregnant state the spiral artery at the junction of the uterine mucosa (“endometrium” before conception, decidua in pregnancy) acts as a “functional sphincter” during menses. Constriction of this portion of the vessel serves to terminate bleeding after endometrial shedding. During normal pregnancy the depth of remodeling eliminates this functional sphincter and its constrictor responsiveness.

In preeclampsia the net result of failed remodeling is failure of terminal spiral artery dilatation to occur. The depth of remodeling is also compromised and does not extend beyond the junction of decidua and myometrium, leaving the “functional sphincter” intact, resulting in major consequences. This failure of terminal dilatation is the pathology that has received most attention, proposed, for example, to dramatically reduce intervillus perfusion (according to Poiseuille’s law, flow increases with the fourth power of the radius). However, as has been elegantly pointed out by Graham Burton and colleagues this vascular dilatation is essentially confined to the terminal portion of the vessel and thus has minimal effect on perfusion. The major impact actually is upon the velocity of blood flow as it leaves the spiral artery. The increase in cardiac output and redistribution of blood that characterizes normal pregnancy would result in an enormous increase in the velocity of blood exiting the spiral artery (2–3 meters/second). The terminal dilatation of the spiral artery can be predicted to reduce blood flow velocity to 10 centimeters/second. The dramatically increased blood velocity with the failed modeling and consequent non-dilated terminal arteries would be predicted to lead to damage to the chorionic villae that with a hemochorial placenta are in direct contact with maternal blood. The accelerated velocity of blood also reduces time for extraction of blood and nutrients from the intervillus blood.

Another aspect of the failed remodeling should also have important consequences. The maintenance of smooth muscle in the spiral arteries results in vessels, which, unlike the situation in normal pregnancy, remains responsive to external signals. The reduced depth of remodeling of the vessels in preeclampsia that does not extend beyond the decidua could be particularly relevant since this results in the maintenance of the “functional sphincter” at the junction of the decidua and myometrium in the spiral artery.

The consequence of the maintenance of responsiveness in the unremodeled spiral arteries is an increased risk of intermittent reduction of blood flow to the intervillus space. This intermittent reduction would set the stage for a hypoxia reperfusion scenario with subsequent oxidative stress. Oxidative stress occurs when the production of reactive oxygen species exceeds the local capacity for buffering by labile antioxidants or antioxidant enzymes. Once the balance is tipped in favor of oxidative stress there is an explosive feed-forward generation of free radicals that damage proteins, lipids, and DNA. In a setting of high oxygen demand (as is present in the placenta) and intermittent flow reduction there is inadequate oxygen available for ATP function and ATP is degraded eventually to uric acid and either NADPH or superoxide. The enzyme that forms uric acid as the terminal step in nucleotide metabolism, xanthine oxidase/dehydrogenase, is a bifunctional enzyme forming either NADPH or the reactive oxygen species, superoxide and H ₂ O ₂ . With hypoxia, the preferential end products are superoxide and H ₂ O ₂ . Further free radicals are generated by NADPH oxidase that is activated by inflammatory activators either from injury secondary to superoxide and H ₂ O ₂ or as a response to the augmented inflammation characteristic of preeclampsia. The superoxide generated also affects the function of nitric oxide synthase, the enzyme responsible for the formation of nitric oxide. The enzyme becomes uncoupled and no longer makes NO (which is an antioxidant) but rather more superoxide. To compound the problem reactive oxygen species downregulate one of the major antioxidant enzymes, superoxide dismutase. These concepts also are reviewed in Chapter 9 .

Closely linked with oxidative stress is endoplasmic reticulum stress. Reduced oxygen delivery can result in both oxidative and endoplasmic reticulum stress. Inflammatory mediators can induce both and both can produce these signaling molecules. Endoplasmic reticulum stress is a cellular mechanism to reduce protein synthesis in settings in which nutrient and oxygen delivery is not sufficient to fully process proteins. This results in a characteristic response, the “unfolded protein response” (UPR). The UPR turns off protein synthesis and could account for the small placenta associated with growth restriction. Apoptotic cell death results with profound stress. UPR also induces the formation of reactive oxygen species. Oxidative stress and endoplasmic reticulum stress not only affect local placental function but also participate in signal generation leading to the systemic features of preeclampsia. Free radicals modify lipid structure and the altered structure and apoptosis increase the shedding of trophoblast fragments. The shedding is augmented by the increased velocity of intervillus blood flow through the unremodeled spiral arteries. Redman and Sargent have championed the idea that such fragments have the capacity to activate immune cells and perhaps directly injure endothelial cells ( Chapter 8 ). Because of the oxidized lipids in these particles they have the potential to transfer oxidative stress systemically. In addition, inflammatory cells passing through the intervillus space, which is replete with free radicals, could also be activated. Cytokines produced in the placenta also enter the systemic circulation. Furthermore, with tissue damage xanthine oxidase/dehydrogenase can be released into the circulation where it can target endothelial cells.

Novel Unbiased Approaches for Addressing the Complexities of the Preeclampsia Syndrome

An important outstanding question has been the extent to which the defects in cytotrophoblast differentiation/invasion that are observed in preeclampsia, particularly the severe forms that occur early in gestation, are a unique feature of this pregnancy complication rather than a default pathway that is activated in response to other pathological conditions. This point has been particularly difficult to address because tissue samples of the maternal–fetal interface from uncomplicated pregnancies are not available after 24 weeks of gestation, when elective terminations are no longer permitted in many countries. Thus, it is impossible to obtain true control samples for studies of the severe forms of PE, which usually occur in the 24- to 32-week interval. The importance of this problem was recently emphasized by the results of a microarray study in which we compared changes in gene expression at the maternal–fetal interface over five time intervals between 14 to 24 and 37 to 40 weeks of gestation. The results showed surprisingly few differences before 24 weeks and hundreds of changes by term, evidence that gestational age is an important variable during this time period.

Accordingly, we tested the hypothesis that the constellation of morphological and molecular defects that are associated with preeclampsia are unique to this condition. Specifically, we compared the histology of the maternal–fetal interface and cytotrophoblast expression of stage-specific antigens in preeclampsia and in preterm labor, with or without inflammation. In the absence of inflammation, biopsies obtained after preterm labor were near normal at histological and molecular levels. In accord with previously published data, preeclampsia had severe negative effects on the endpoints analyzed; biopsies obtained after preterm labor with inflammation had an intermediate phenotype. Thus, our results suggest that the maternal–fetal interface from cases of preterm labor without inflammation can be used for comparative purposes, e.g., as age-matched controls, in studies of the effects of preeclampsia on cells in this region.

As a further proof of principle, we conducted a global analysis of gene expression at the maternal–fetal interface in preeclampsia ( n =12; 24–36 weeks) vs. samples from women who delivered due to preterm labor with no evidence of infection (PTL; n =11; 24–36 weeks). Using the HG-U133A&B Affymetrix GeneChip platform and statistical significance set at log odds-ratio of B >0, 55 genes were differentially expressed in preeclampsia. They encoded proteins previously associated with preeclampsia (such as VEGFR-1, leptin, CRH, and inhibin) and novel molecules, e.g., sialic acid binding immunoglobulin-like lectin 6 (Siglec-6), a potential leptin receptor, and pappalysin 2 (PAPP-A2), a proteinase that cleaves insulin-like growth factor binding proteins. We used quantitative-PCR to validate the expression patterns of a subset of the up- or downregulated genes. At the protein level, we confirmed preeclampsia-related changes in the expression of Siglec-6 and PAPP-A2, which localized to invasive cytotrophoblasts and syncytiotrophoblasts. Notably, Siglec-6 placental expression is uniquely human, as is spontaneous preeclampsia. The functional significance of these observations may provide new insights into the pathogenesis of preeclampsia, and assaying the circulating levels of these proteins could have clinical utility for predicting and/or diagnosing this syndrome. These new molecular concepts are developed further in the Appendix on trophoblast gene expression that follows this chapter.

Summary and Future Directions

We now understand a great deal about cytotrophoblast defects in the placentas of patients whose pregnancies are complicated by preeclampsia. In a landmark study published over 35 years ago, Brosens, Robertson, and Dixon first described the abnormally shallow cytotrophoblast invasion that is observed in preeclampsia and a substantial proportion of pregnancies complicated by intrauterine growth retardation. These investigators considered the lack of invasion of the spiral arterioles to be particularly significant. Building on this foundation, our recent studies have shown that cytotrophoblast invasion of the uterus is actually a unique differentiation pathway in which the fetal cells adopt certain attributes of the maternal endothelium they normally replace. In preeclampsia, this differentiation process goes awry.

Currently, we are very interested in using these findings as a point of departure for studies of the disease process from its inception to the appearance of the maternal signs. With regard to its inception, understanding the nature of the phenotypic alterations that are characteristic of cytotrophoblasts in preeclampsia offers us the exciting opportunity to test hypotheses about the causes. From a reductionist viewpoint, preeclampsia can be considered as a two-component system in which the two parts – the placenta and the mother – fail to connect properly. In theory, this failure could be due to either component. For example, it is inevitable that cytotrophoblast differentiation must sometimes go awry. The high frequency of spontaneous abortions that are the results of chromosomal abnormalities is a graphic illustration of the consequences of catastrophic failure of cytotrophoblast differentiation. But the observation that confined placental mosaicism can be associated with IUGR is especially relevant to the studies described in this chapter. Conversely, there is interesting evidence that in certain cases the maternal environment may not permit normal trophoblast invasion. For example, patients with preexisting medical conditions, such as lupus erythematosus and diabetes mellitus, or with increased maternal weight, are prone to developing pregnancy complications, including preeclampsia. Finally, the mother’s genotype may also play a role; expression of an angiotensinogen genetic variant has been associated with the predisposition to develop preeclampsia.

An equally interesting area of study is how the faulty link between the placenta and the uterus leads to the fetal and maternal signs of the disease. It is logical that a reduction in maternal blood flow to the placenta could result in fetal IUGR, a fact that has been confirmed in several animal models. But how this scenario also leads to the maternal signs is much less clear. Since the latter signs rapidly resolve once the placenta is removed, most investigators believe that this organ is the source of factors that drive the maternal disease process. Another important consideration is that the local placental abnormalities eventually translate into maternal systemic defects. Thus, it is likely that the causative agents are probably widely distributed in the maternal circulation. But their identities are not yet known. Candidates include macromolecular entities such as fragments of the syncytiotrophoblast microvillous membrane that are shed from the surface of floating villi and can damage endothelial cells. Molecular candidates include the products of hypoxic trophoblasts, whose in vitro vascular effects mimic in vivo blood vessel alteration in patients with preeclampsia and vasculogenic molecules of placental origin (see Chapter 6 , Chapter 8 ).

In the end, the utility of the observations discussed in this chapter will rest in our ability to use this newfound knowledge to make improvements in the clinical care offered to pregnant women. In this regard, the time is right to mount a systematic attack for discovering potential biomarkers of preeclampsia that circulate in maternal blood, before the signs appear and/or at the time of diagnosis. We think that all the requisite elements for these studies are in place. On the biology side, we now have a good understanding of the underlying defects in placentation that are thought to eventually lead to the full-blown manifestations of this condition. On the technology side, many new platforms for biomarker discovery are being developed, including those that employ powerful mass spectrometry-based approaches. Although current efforts are focusing on prediction and/or diagnosis, these studies might also reveal therapeutic targets. Thus, we are entering an exciting period when years of basic science will move from a research setting to clinical laboratories, a prospect that has the potential to transform obstetrical care from a 20th-century enterprise to a 21st-century venture.