The placenta is essential for sustaining the growth of the fetus during gestation, and defects in its function result in fetal growth restriction or, if more severe, fetal death. Several molecular pathways have been identified that are essential for development of the placenta, and mouse mutants offer new insights into the cell biology of placental development and physiology of nutrient transport.
Survival and growth of the fetus are critically dependent on the placenta. It forms the interface between the maternal and fetal circulation, facilitating metabolic and gas exchange as well as fetal waste disposal. In addition, the placenta produces hormones that alter maternal physiology during pregnancy and forms a barrier against the maternal immune system (14). In humans and rodents, the fully developed placenta is composed of three major layers: the outer maternal layer, which includes decidual cells of the uterus as well as the maternal vasculature that brings blood to/from the implantation site; a middle “junctional” region, which attaches the fetal placenta to the uterus and contains fetoplacental (trophoblast) cells that invade the uterine wall and maternal vessels; and an inner layer, composed of highly branched villi that are designed for efficient nutrient exchange (72). The villi are bathed by maternal blood and are composed of outer epithelial layers that are derived from the trophoblast cell lineage and an inner core of stromal cells and blood vessels.
Many targeted mutations in mice exemplify how single gene mutations can affect placental development or function (Tables 1⇓, 2⇓, and 3⇓). A common feature among these placental mutants is the reduced ability to transport nutrients, which results in fetal growth restriction or, under more serious circumstances, embryonic death. The vast majority of the placental phenotypes that have been described to date result in defects in the establishment or maturation of the placental villi, which in mice comprise the so-called labyrinth layer. Most of the defects are structural in nature, although some of the mutants offer insights into the regulation of nutrient transport.
Placental Development in Mice and Humans
Although the gross architecture of the human and mouse placentas differ somewhat in their details, their overall structures and the molecular mechanisms underlying placental development are thought to be quite similar (72). As a result, the mouse is increasingly used as a model for studying the essential elements of placental development. In mice, placental development begins in the blastocyst at embryonic day (E) 3.5 when the trophectoderm layer is set aside from the inner cell mass (FIGURE 1⇓) (15). At the time of implantation (E4.5), the mural trophectoderm cells, which are those not in contact with the inner cell mass, become trophoblast giant cells that are analogous to human extravillous cytotrophoblast cells (72). These cells stop dividing, yet they continue to replicate DNA (endoreduplication) to become polyploid. In contrast, two diploid cell types emerge from the polar trophectoderm, which are those cells immediately adjacent to the inner cell mass: the extraembryonic ectoderm and the ectoplacental cone (72). Subsequently, the extraembryonic ectoderm will develop into the trophoblast cells of the chorion layer and, later, the labyrinth. While developing, the labyrinth is supported structurally by an ectoplacental cone-derived layer called the spongiotrophoblast. It forms a compact layer of cells sandwiched between the labyrinth and the outer giant cell layer and corresponds to the column cytotrophoblast of the human placenta (72). During later gestation, glycogen trophoblast cells begin to differentiate within the spongiotrophoblast layer, and subsequently they diffusely invade the uterine wall (2).
The vascular portion of the placenta is derived from extraembryonic mesoderm (allantois) that extends from the posterior end of the embryo at E8.0 (14). At E8.5, the allantois and the chorion join together in a process called chorioallantoic attachment. Soon thereafter, the chorion begins to fold to form the villi, creating a space into which the fetal blood vessels grow from the allantois (14). At this time, the chorionic trophoblast cells begin to differentiate into two labyrinth cell types. Multinucleated syncytiotrophoblast cells, formed by the fusion of trophoblast cells, surround the fetal endothelium of the capillaries (see FIGURE 3⇓). A mononuclear trophoblast cell type lines the maternal blood sinuses. Together the trophoblast and fetal vasculature generate extensively branched villi of the labyrinth (comparable with human chorionic villi), which become larger and more extensively branched until birth (E18.5–19.5) (2). Maternal and fetal blood flows in a countercurrent manner within the labyrinth to maximize nutrient transport (2). If the labyrinth is not appropriately vascularized with suitable patterning, branching, and dilation, placental perfusion is impaired, resulting in poor oxygen and nutrient diffusion (63).
The first step in labyrinth development is chorioallantoic attachment, and defects in this process are among the most common causes of midgestation embryonic lethality (72). The allantois and chorion trophoblast cells are derived in parallel from distinct cell populations. Originating from the epiblast, the allantois is composed of extraembryonic mesoderm (16). Many genes are necessary for proper development of the allantois (Table 1⇑). However, the bone morphogenetic protein (BMP) signaling pathway appears to be particularly important. Critical molecules have been knocked out in mice, including Bmp2, -4, -5, and -7 (20, 86, 104) as well as Smad1, a downstream effector of BMP signaling (43). The mutants display mesodermal differentiation defects contributing to abnormal allantoic development. Additionally, the allantois of a Foxf1-deficient mouse embryo is small and shows a loss of BMP4 expression (52), suggesting that this transcription factor is upstream of BMP. The blood vessels in the allantois arise de novo due to vasculogenesis, and this is not dependent on attachment of the allantois to the chorion (16).
The majority of chorionic cells are derived from the extraembryonic ectoderm, although they overlie a thin layer of chorionic mesothelium (72). Both Err2/Errβ, a nuclear hormone receptor (49), and fibroblast growth factor receptor 2 (Fgfr2) (99) are expressed within chorion trophoblast cells and are required for their maintenance. Proper formation of the chorion and allantois are necessary for attachment to occur. In addition, however, many mutants exist in which the allantois and the chorion appear to have formed normally, yet chorioallantoic attachment fails to occur (Table 1⇑). It is known that attachment is dependent on the cell adhesion molecule VCAM1 (25, 42), which is expressed on the allantois, and its ligand α4-integrin (102), which is expressed by the chorionic mesothelium. However, not all Vcam1- or α4-integrin-deficient mice fail in chorioallantoic attachment, suggesting that other redundant adhesion mechanisms are involved. Indeed, other mutants with defects in chorioallantoic attachment also display incomplete penetrance (Table 1⇑). It will be necessary to look more closely at these mutant placentas to determine if this seemingly random collection of genes shares a common molecular pathway, allowing for a better understanding of the attachment process. Importantly, in the event that chorioallantoic attachment does occur in these incompletely penetrant mutants, they will often exhibit later defects in morphogenesis of the labyrinth.
Initiation of Branching Morphogenesis at the Chorioallantoic Interface
At E9.0, immediately after chorioallantoic fusion occurs, primary villi begin to develop, evenly spaced across the chorionic surface (14), and blood vessels soon fill in the villous folds (72). The process is often described as “vascular invasion” of the chorion, but this is misleading because the process requires active participation of chorion trophoblast and allantoic mesoderm. The branchpoints are actively selected by clusters of chorion trophoblast cells that express the Gcm1 gene (4). As each branch elongates, Gcm1 expression remains at the distal tip and continues to be expressed as long as villi are branching. Gcm1 expression also initiates the differentiation of chorionic trophoblast into syncytiotrophoblast (4). Embryos deficient for Gcm1 do not initiate chorioallantoic branching; their chorion layer remains flat, trophoblast cells do not differentiate, and the fetal vasculature remains restricted to the allantois.
Gcm1 mRNA expression is first detected in the chorion before chorioallantoic attachment, and therefore branchpoint selection appears to be independent of allantoic attachment (4). However, the phenotypes of several mouse mutants have suggested that the initiation of morphogenesis after selection has occurred may require the interaction of chorion trophoblast and allantois. For example, the expression of Gcm1 mRNA is not maintained in Mrj mutant mice in which chorioallantoic attachment fails to occur (31) and, in the absence of allantoic mesoderm, chorion trophoblast cells remain undifferentiated (29). In addition, mutations in various genes within the Notch signaling pathway, including Notch1/Notch4 (39), the Notch receptor Delta-like 4 (17), and transcription factors Hey1/Hey2 (19) and Rbpsuh (38), all appear to result in early blocks to chorioallantoic branching. Expression of these genes has only been reported within the allantoic mesoderm/blood vessels, suggesting that the fetal vasculature may be important for initiation of branching of the chorioallantoic interface. There are several caveats with this hypothesis, however. First, it is possible that these mutant mice are simply developmentally delayed or slowed and not arrested at the flat chorion stage, as with Gcm1 mutants. To address this possibility, later-stage placentas should be examined, as has been done with Grb2 (75). Second, Hey1 mRNA has also been detected within the trophoblast cells of the ectoplacental cone at least at E8.5 (K. Dawson and J. C. Cross, unpublished data), and therefore expression of the Notch signaling components is not restricted to allantois. Third, human chorionic villi develop before becoming vascularized (10), implying that vascular interactions are not important for villous development, at least in humans. Given these findings, it is clear that more work needs to be done to address the signaling interactions between chorion trophoblast and allantois during early stages of villous development.
Signaling and Morphogenesis of the Labyrinth
A large number of genes have been identified that are required for labyrinth development (Table 2⇑). However, for most of the genes, the specific cellular phenotype is not clear based on the published studies. Indeed, the most accurate description is that the labyrinth is simply underdeveloped or “small,” meaning that the chorioallantoic interface remains underbranched and as a result there is a relative reduction in the density of fetal blood vessels (FIGURE 2⇓). Some mutants exhibit defects early in labyrinth development such that their chorionic plates remain compact with little branching and little fetal blood vessel growth (8, 9, 23, 30, 64, 101). Embryos in this case will die between E10.5 and E12.5. Many other labyrinth phenotypes manifest slightly later, with some evidence of chorioallantoic branching but with thick trilaminar trophoblast layers and/or reduced vascularization (6, 13, 34, 71, 73, 96). The associated fetuses die either late in gestation or perinatally. The cause of lethality in all cases is a result of insufficient metabolic exchange.
Despite the uncertainty about the specific underlying cellular defects, an important general conclusion to emerge from the study of small-labyrinth mutants is that labyrinth development depends on a number of intercellular signaling pathways. Specific pathways that are critical include Fgf (99), Egf (91), Notch (39), Lif (96), Pdgfb (61), and Wnt (57). Likewise, a number of signaling adaptor proteins downstream of these signaling events are implicated given the similarity of their mutant phenotypes, including Chm (80), CtBP2 (30), Erk2 (27), Erk5 (85, 100), Gab1 (34, 73), Grb2 (75), Mek1 (23), Mekk3 (101), p38α MAPK (1, 60), and Sos1 (69). Based on restricted patterns of expression or chimera experiments, it is apparent that these signaling pathways are largely required in the trophoblast cell compartment of the labyrinth (Refs. 27 and 67; reviewed in Refs. 72, 81, 85, 92, and 100).
In addition to the protein signaling systems, nuclear receptors are also important for morphogenesis of the labyrinth. The retinoid X receptor (RXR) proteins dimerize with a number of different nuclear receptors, including retinoic acid receptors (RARs) and the perioxisome proliferator-activating receptor (PPAR). RXR-α/RXR-β double-mutant mice die at midgestation and show a small-labyrinth phenotype (97). PPAR-γ mutants show a similar phenotype, implying that perhaps PPAR-γ is the critical dimerization partner of the RXRs for labyrinth development (8). In support of this hypothesis, mutations in genes encoding the PPAR-γ-associated proteins PKB-α (103), PRIP/Rap250/AIB3 (5, 41, 108), and PBP (107), as well as the transcriptional target gene Muc1 (79), all have been implicated in labyrinth development.
Direct and Indirect Controls on Vascularization of the Labyrinth
When morphogenesis of the labyrinth is diminished, one of the most obvious differences is that the layer remains cell dense and there are fewer maternal and fetal blood spaces. However, in the vast majority of labyrinth mutants the differences are likely to be secondary effects, and there are only a few examples of mutants with primary vascular defects.
The maternal blood spaces in the labyrinth (termed sinusoids) often appear to be larger than normal in mutants (56, 71), but this can be an indirect effect. The maternal sinusoids within the labyrinth are lined and shaped by trophoblast cells and normally diminish in size as gestation proceeds as a simple consequence of the increasing density of trophoblast villous branching (2). Therefore, whenever chorioallantoic branching is reduced, the maternal blood spaces in the presumptive labyrinth layer will remain larger. The more critical question would be whether the overall maternal blood volume in the presumptive labyrinth is altered in a mutant. This can be difficult to assess accurately in histological sections, however, because the blood will readily leak out during tissue dissection unless the uterine blood vessels are ligated before dissection and tissue fixation (2).
The focus only on fetal blood spaces within the labyrinth can give investigators a false impression about the nature of the primary defect in the labyrinth. For many of the genes whose mutant phenotypes were originally described as “vascular” in nature, they are expressed exclusively within the trophoblast and not the vasculature itself (Table 2⇑). More importantly, since fetal vessels can only grow into the core of villi within the labyrinth, all small-labyrinth phenotypes with fewer villi would also be described as having fewer overall fetal blood spaces. The more accurate way to assess these mutants is to compare vascular density with the density of differentiated villi to determine if the reduction in fetal blood vessel space is simply proportional to reduction in villi (98).
There are perhaps only a few examples of mouse mutants that show a reduction in the vasculature of the labyrinth that is not proportional to the extent of villous development (Table 3⇑). The extracellular matrix protein Cyr61 (56) and the Notch-signaling components Dll4 (17), Notch1/4 (39), Hey1/2 (19), and Rbpsuh (38) are expressed in the vasculature itself, and mutations in their genes result in a poorly vascularized allantois. The Esx1 gene, by contrast, encodes a homeobox transcription factor that is expressed solely in trophoblast cells of the labyrinth (46, 47). Placentas from Esx1 mutants appear to undergo normal chorioallantoic branching morphogenesis but have obvious deficiencies in fetal blood vessel growth into the labyrinth villi (FIGURE 2⇑) (46). This indicates that trophoblast cells are actively involved in the vascularization of the labyrinth and suggests that a possible transcriptional target of Esx1 is a signal from the trophoblast cells that induces or directs vascular morphogenesis. The trophoblast-derived factor(s) that directly influence the development of fetal blood vessels in the labyrinth remain elusive.
An important factor to bear in mind when studying a placental phenotype is that the overall size and extent of vascularization of the placenta can change in an apparent attempt to compensate for primary defects. Esx1 mutant placentas are actually larger than their wild-type counterparts, perhaps suggesting an attempt to compensate for the reduced vascularization and nutrient transport (46). Placentas from mothers who smoke throughout their pregnancy are disproportionately large (66). Impaired oxygen transport caused by an increase in carbon monoxide concentration induces additional angiogenesis of the fetoplacental vasculature (66). Accordingly, one would expect that the majority of mutant placentas with defective labyrinth morphogenesis might also compensate for reduced nutrient and gas exchange. The Rb mutant placenta is a well-documented example of this phenomenon. Rb-deficient labyrinths have abnormal architecture associated with fewer villi due to an inappropriate proliferation of trophoblast cells and block to differentiation (98). The villous surface area of Rb mutant labyrinth trophoblast is reduced to 62% of wild type. However, fetal capillary density is only reduced to 88% and essential fatty acid transport, as a measure of nutrient uptake capacity, is 86% of wild type (98). Therefore, the villi that are able to form in Rb mutants are relatively hypervascularized, and this is apparently able to partially compensate at least for fatty acid transport (FIGURE 2⇑). Proper detailed analyses of other small-labyrinth mutants may reveal similar compensatory measures.
Tissue oxygenation is a critical regulator of vascular development (54). The expression of arylhydrocarbon receptor nuclear translocator (Arnt), also known as hypoxia inducible factor-1β (Hif-1β), in the labyrinth suggests that tissue oxygenation may be a normal regulator of placental growth (37, 72). Arnt heterodimerizes with Hif-1α to mediate transcription of specific genes, including VEGF, in response to oxygen deprivation (3). Particular attention was given to the Arnt knockout mouse when a defect within the labyrinth resulted (37). It was, however, demonstrated by tetraploid chimera experiments that Arnt function is actually required within the trophoblast compartment and not the vascular endothelium (3). As a result, the vascularization defect that was described is secondary to a primary trophoblast defect. Consequently, the Arnt-deficient mutants do not stand apart from the other small-labyrinth mutants (Table 2⇑ and FIGURE 2⇑).
Placental Nutrient Transport
Nutrients are transferred across the placental barrier via several mechanisms, including passive diffusion, facilitated diffusion, and active transport (83). A significant amount of solute flux across the mouse placenta is achieved by passive diffusion (84). Therefore, in addition to the overall surface area and permeability (83), diffusional distance is a major factor influencing overall diffusional capacity of the placenta. In mice, a trilaminar layer of trophoblast cells separates the fetal capillary from the maternal sinusoids: a bilayer of syncytiotrophoblast surrounds the fetal blood vessel endothelium and a layer of mononuclear cells lines the maternal blood sinuses (2). Consequently the nutrients, gases, and waste must diffuse or be transported across four layers to get from one blood compartment to the next (FIGURE 3⇓). Measuring the surface area and thickness of the trophoblast layers in the labyrinth by stereological analysis is a relatively easy way to assess the diffusional ability of mutant placentas (12, 84, 98). Alkaline phosphatase is expressed by trophoblast cells that line maternal blood sinusoids within the labyrinth (2). Accordingly, quantifying this expression is a simple way of assessing diffusional surface area present in the placenta (98).
Placental-specific knockout of the insulin-like growth factor-II gene (Igf2) results in both a thickened diffusional barrier and a smaller overall surface area in the labyrinth (13, 84). Nutrient transport capacity can be assessed either directly by measuring transport of radiolabeled nutrients (13, 36, 84) or by measuring the accumulated content of nutrients such as essential fatty acids in the fetus (98), as has been done with the Igf2 and Rb mutants, respectively.
Few nutrient transporters have been studied in the mouse placenta, and even fewer mutants have been reported that show reduced nutrient transport. Placental-specific mutation of Igf2, in addition to producing anatomic changes, results in altered system A amino acid transporter function (13). Active calcium transport across the placenta is reduced in parathyroid hormone/parathyroid hormone-related peptide receptor mutant mice (36). The gap junction protein connexin 26 (Cx26) has been shown to act in cooperation with the glucose transporter GLUT1 in the facilitated diffusion of glucose across the trophoblast layers of the labyrinth (21) (FIGURE 3⇑). Cx26 mutants die at E11.0 and show a 60% decrease in glucose transport (21). Although this suggests that Cx26 is required for glucose transport, the reduction in glucose transport may also be due to the fact that the surface area of the labyrinth is also probably reduced (21). GLUT1 mutants show haploinsufficiency in that heterozygous embryos die at the blastocyst stage due to increased apoptosis (28). GLUT3 is also expressed in the placenta (44), yet no knockout mouse model has been generated to test its importance in placental glucose transfer.
Work in the past 10 years has resulted in an explosion of information about the regulation of placental development and function, in particular that of the villous placenta that is involved in nutrient uptake. Most of the mouse models that are informative about placental development and function are based on homozygous mutant mice that show embryonic lethality due to the severity of the placental disruption. More work needs to be done to illuminate the cellular basis of most of these mutants as well as to accurately assess whether the placental dysfunction is based on a failure in development of the nutrient transport surface, the function of specific nutrient transporters, or both. The fairly recent development of techniques to address these questions, as described above, should allow rapid progress. Clarifying the cause of the defect will aid in determining if these genes can fit into common or parallel genetic pathways, in addition to creating a better understanding the morphogenesis of chorioallantoic placenta and the causes of fetal growth restriction and fetal death. Using these techniques, it would also be fruitful to examine heterozygotes for known placental mutants. Although the homozygous mutants may have severe defects leading to fetal death, heterozygotes may show the less-severe effect of fetal growth restriction.
Intuitively, placental flux of nutrients throughout gestation is proportional to the size of the fetus, and therefore intrauterine growth restriction (IUGR) in humans may be linked to defects in either placental development or nutrient transport capacity (35, 82). The mouse models based on mutation of Esx1 and the placental-specific isoform of Igf2 have surfaced as the only models that result in IUGR but not fetal death. Both mutations affect nutrient exchange by directly reducing uptake capacity of the villi, although they have distinctive defects in labyrinth trophoblast cells. Loss of Esx1 prevents the normal development of the fetal vasculature into the placenta, and thus nutrients from the mother cannot be passed adequately to the fetal bloodstream. The Igf2 mutants, on the other hand, have reduced diffusional capacity, thereby reducing the amount of nutrient exchange that can occur. Based on this evidence in mice that not all IUGR placentas have the exact same phenotype, it is plausible that IUGR in humans may actually have distinct underlying changes. As a result, it may be important to categorize human IUGR into distinct conditions based on which developmental stage is affected, which cell type is affected, and whether nutrient transporters are properly expressed.
- © 2005 Int. Union Physiol. Sci./Am. Physiol. Soc.