The Human Placenta in Diabetes and Obesity: Friend or Foe? The 2017 Norbert Freinkel Award Lecture

Owing to its position between the maternal and fetal circulation, the placenta plays a central role in fetal nutrition to sustain fetal development and growth. Changes in either maternal or fetal circulation may alter placental structure and function with potential consequences for fetal growth and development. The question, which has puzzled me throughout my career, has been: does the placenta contribute to the fetal phenotype of diabetes, acting as a foe, or does it protect the fetus from adverse effects of the diabetic environment and is it, thus, a friend of the growing fetus?

In my quest to address this question, I have been considerably inspired by Norbert Freinkel. In particular, his concept of critical time periods in pregnancy during which different organs and structures of the fetus are susceptible to perturbations in the intrauterine environment, as seen with, for instance, maternal diabetes (and for sure also obesity), has influenced my thoughts (1,2). We have adapted this concept to the placenta. The placenta may be exposed to metabolic changes in the mother during critical periods of its development (Fig. 1), and, depending on the timing, different placental structures and processes could be affected. Therefore, we postulated that pregestational and gestational diabetes mellitus (GDM) affect the placenta differently, since their associated metabolic changes are present in different developmental periods (3). Recent emerging evidence demonstrates that metabolic changes resulting in a diagnosis of GDM later in pregnancy are present early in gestation. However, in pregestational diabetes and maternal obesity, these changes are present as early as at conception. Hence, the placenta may be exposed to metabolic changes in the mother during different critical periods of its development (Fig. 1). Although at week 12 of pregnancy the placenta weighs only about 5% of its final weight at term of pregnancy, placental growth kinetics, i.e., placental weight change per week, is at its maximum between pregnancy weeks 10 and 14 (Fig. 2). Rapidly growing tissues are very susceptible to environmental perturbation (4), and metabolic or proinflammatory disturbances in early pregnancy may, therefore, have profound effects on placental growth. We have recently proposed that a compromised early placental development may track throughout pregnancy and contribute to the fetal phenotype through, among others, the glucose steal phenomenon (5,6). In the following sections, I have separated the discussion into two periods, early versus late pregnancy, mostly because there is no information available about the time in between. In the end, I shall try to synthesize the discussion into a comprehensive framework and provide the answer(s) to the main question of this review.

Early pregnancy is characterized by a wide range of processes during which the blastocyst implants and differentiates to establish the two main component tissue types: inner cell mass, from which the fetus proper originates, and trophectoderm, which develops into trophoblast and placental tissue. In the first weeks of pregnancy, before remodeling of the spiral arteries in the decidua, the placenta develops in a low-oxygen environment. Subsequent spiral artery remodeling is accompanied by an increase of oxygen tension in the intervillous space (7), which results in a transient physiologic oxidative stress in the trophoblast (8), the placental cell type exposed to maternal blood. If this physiologic oxidative stress is superimposed by oxidative and/or proinflammatory stress, due to maternal diabetes or obesity, then a reduction in trophoblast growth results (9). The specific consequences of the changes in the first trimester in oxygen tension supplied to the placenta are unknown. The detailed mechanisms of placental responses also remain to be determined. This is a complex endeavor, because trophoblast invasion and, hence, spiral artery remodeling may be modified by maternal obesity (10) and diabetes (11).

According to this concept, the increasing oxygen tension in early pregnancy, in combination with hyperglycemia and/or proinflammatory stress, may have an inhibitory effect on trophoblast and placental growth by inhibiting invasion and thereby delaying spiral artery remodeling, by directly inhibiting trophoblast proliferation, or by a combination of the two.

However, other components of the maternal environment may have opposing effects, i.e., stimulate placental growth. Insulin may be one candidate with a sustained effect that tracks throughout pregnancy. The recent demonstration of associations between the maternal insulin secretory response in early pregnancy (weeks 12–14) with placental volume at week 20 and placental weight at the end of gestation suggests such a sustained insulin effect (12).

Mechanistically, insulin cannot directly promote placental growth because of spatial reasons: the insulin receptor is located predominantly on the plasma membrane of the syncytiotrophoblast, whereas the cytotrophoblast is the proliferating trophoblast stem cell population (13) positioned subjacent to the syncytiotrophoblast. Placental growth at this stage of pregnancy requires an expansion of cytotrophoblast number and their fusion with the syncytiotrophoblast. This is a complex and tightly regulated process, in which a syncytiotrophoblast matrix metalloproteinase (MMP14) is directly involved (14). It can be upregulated by insulin and tumor necrosis factor-α and is present at higher levels in the first trimester placenta of women with type 1 diabetes (T1D) (15). In this situation, placental MMP14 levels directly associate with the total daily insulin dose these women received for managing their glycemia. These early placental responses to insulin might be modified by maternal obesity due to reduced insulin sensitivity (16).

The temporal sequence of these events is unknown, but we envisage a scenario in pregestational diabetes or obesity in which early placental growth is initially reduced or delayed due to increased oxidative stress. This is paralleled by an early fetal growth delay as found in some T1D pregnancies (17–19). Subsequently, a placental growth spurt occurs in late first trimester (17–19). The larger placenta as a result of this growth spurt leads to enhanced glucose transfer, which relaxes the restraint on fetal growth. Thus, once the placenta has grown, fetal catch-up growth will follow. This scenario may explain biphasic fetal growth in T1D (20,21).

Why is early placental growth so important for fetal growth and development? It is unknown to date how glucose transfer at this early period is governed, but, as mentioned before, placental size may be one factor: more specifically, placental volume and, hence, surface area may define the total number of glucose transporters. Subtle increases in transporter number at this stage may contribute to an early onset of enhanced transplacental glucose flux. As a consequence, the maturation of the stimulus-secretion coupling mechanisms of the fetal pancreas may be accelerated. This leads to an earlier onset of fetal β-cell response to elevated glucose levels and results in elevated fetal insulin levels. These in turn stimulate fetal glucose uptake into peripheral tissues. As a consequence, fetal glucose levels will be transiently lowered, thereby steepening the maternal-to-fetal glucose concentration gradient. This steepened concentration gradient will drive maternal glucose flux to the fetus even under seemingly normal glucose control of the mother (glucose steal phenomenon) (6). The onset of this phenomenon is unknown, is likely multifactorial, and varies from pregnancy to pregnancy.

Collectively, it is currently unclear whether the placenta early in pregnancy acts as a friend or foe of the fetus.

Almost all studies on the effects of maternal diabetes and obesity on the placenta have been carried out at the end of pregnancy. Based on the concept that effects of changes in maternal environment on placental growth and development differ per period of pregnancy (Fig. 1), one can predict that in the second and third trimester, placental functional differentiation and mass expansion are mostly affected. The results seen at the end of gestation are then the manifestation of these changes that have occurred earlier on.

Over the years we have identified many changes associated with maternal diabetes and/or obesity. In our view, these can be regarded as adaptive responses to protect the fetus rather than contributors to excessive growth and fat accretion. The two examples that follow support this notion.

It has been known for a long time that pregnancies of women with diabetes carry the risk for the fetus to be low in oxygen, a condition that can be transient or persist for a longer period of time. This is the result of augmented fetal oxygen demand because of enhanced fetal metabolism. Fetal metabolism is enhanced due to the elevated insulin levels paralleling hyperglycemia. At the same time, more maternal hemoglobin is glycosylated in diabetes. HbA_1c has a higher oxygen affinity than HbA₁. Hence, oxygen delivery to the intervillous space for fetal supply is reduced. This imbalance in maternal supply and fetal demand results in low fetal oxygen. The fetus responds by increasing its erythropoietin levels (22,23), thereby enhancing erythropoiesis, which eventually results in more nucleated red blood cells in the neonatal circulation (24). Although the placenta synthesizes erythropoietin in an oxygen-sensitive manner, its contribution to elevated fetal levels is unlikely because placental synthesis is restricted to the trophoblast populations facing the maternal circulation (25). Fetal insulin also stimulates growth of erythroid progenitors in cord blood and contributes to a higher number of erythroid cells (26).

The placenta also responds to the enhanced fetal demand for oxygen by expanding its surface area of exchange (27,28). This has been described mostly in T1D. Surface enlargement is achieved mostly by enhancing fetoplacental vascular growth through angiogenesis. The new, usually small-caliber vessels are homogeneously distributed across the core of the placental villi; in contrast, in pregnancies of women without diabetes, the small vessels are preferentially arranged in a subtrophoblastic location (29).

One of the key regulators for placental angiogenesis under these conditions is fetal insulin (30,31). We have delineated several of the mechanisms and pathways through which fetal insulin contributes to vascular growth in the placenta. These include 1) degradation of extracellular matrix by a matrix metalloproteinase (MMP14), 2) activation of endothelial nitric oxide synthesis and enhanced formation of nitric oxide, which stimulates endothelial tube formation, and 3) stabilization of the F-actin cytoskeleton, which is necessary for lamellipodia formation and cell migration. Insulin activates these individual processes through the protein kinase-3 pathway. Surprisingly, insulin does not activate the mitogen-activated protein kinase pathway to stimulate endothelial cell proliferation (31,32). The fetal signals for enlargement of surface area are likely generated before week 32 in pregnancy, as the placental surface area correlates with the day-to-day variation in maternal blood glucose levels only between week 12 and 32 of gestation and not thereafter (33).

The placental response to increased fetal oxygen demand is not limited to facilitating supply of maternal oxygen. As a further measure, it also facilitates the supply of iron needed for enhanced erythropoiesis. This is accomplished by upregulating the placental transferrin receptor, although more pronounced receptor glycosylation associated with hyperglycemia may reduce its affinity for transferrin binding (34–36). A direct effect of fetal oxygen on placental vascular function cannot be ruled out, but remains to be demonstrated.

Collectively, the fetoplacental unit adapts structurally and functionally to facilitate oxygen delivery to the fetus in normal pregnancy and beyond in situations of metabolically induced enhanced oxygen demand. These changes result in a shorter overall oxygen diffusive conductance (37,38).

Another example of placental responses aiming to protect fetal development can be found in the mechanisms preventing preatherosclerotic lesions in the fetoplacental circulation. On the maternal side, atheromas in spiral arteries have been found and are more frequent in T1D (39). Foam cells can form in the intervillous space in situations of maternal hypercholesterolemia (40), and in the fetal aorta, fatty streaks can be formed (41). However, no similar lesions have been identified yet in the fetoplacental circulation. This suggests that highly efficient mechanisms for cholesterol efflux from placental vessels must be in place. Indeed, we have identified cholesterol efflux transporters ABCA1 and ABCG1 on fetoplacental endothelial cells and have delineated a two-step mechanism for the endothelial cells to efflux cholesterol (42). The primary cholesterol acceptor on the fetal side is nascent HDL followed by HDL₃. The small, dense HDL₃ is modified by endothelial cell phospholipid transfer protein (PLTP) to cholesterol-rich HDL₂ and nascent HDL, which in turn will pick up cholesterol released from the endothelial cells. Fetal HDL₂ will then be taken up by the liver. This represents an efficient mechanism for fetoplacental endothelial cells to shuttle their cholesterol to the fetal liver, thus protecting against formation of preatherosclerotic lesions in the vasculature of the placenta proper. In GDM, placental endothelial cells synthesize more cholesterol because of upregulation of HMG-CoA reductase, the rate-limiting enzyme in the cholesterol biosynthetic pathway. Yet total cholesterol content in the endothelium is unaffected by GDM, because cholesterol efflux mechanisms are upregulated by oxysterols in the fetal circulation, which form de novo from cholesterol through reactive oxygen species–induced oxidation (43). Enhanced cholesterol efflux in GDM leads to more HDL₃ in the fetal circulation, but, in parallel, PLTP is also upregulated. Fetal insulin is one of the regulators of PLTP activity (44).

Thus, in GDM, all mechanisms that ensure cholesterol homeostasis at the fetoplacental interface are upregulated to avoid vascular damage and compromised blood flow due to the formation of preatherosclerotic lesions. The mechanisms related to cholesterol homeostasis are further supported by a reduction of both forms of intercellular adhesion molecule 1 (ICAM-1)—the surface bound and the soluble form—on placental endothelial cells. Their levels are in an inverse relationship with maternal BMI. These proteins are markers of vascular inflammation (45). The low ICAM-1 protein levels in GDM can be regarded as an additional protective mechanism to avoid transmigration of fetal leukocytes into the subendothelial space of the fetoplacental circulation, thereby also contributing to reducing the risk of preatherosclerotic lesions.

In both examples above, the fetus provides the signals inducing the cellular and molecular mechanisms that facilitate placental adaptation (Fig. 3). This represents a change in paradigm: previously, maternal control of placental development and function was presumed; the findings discussed above establish the importance of fetoplacental signaling. On the basis of the developmental shift of the insulin receptor location from the surface facing the maternal circulation to that facing the fetal circulation, we have previously proposed that maternal factors govern placental development and function mostly early in pregnancy, whereas the fetus gradually takes over control of its own organ in later stages of pregnancy (46).

The processes described above are just examples of how the placenta protects the fetus at the end of pregnancy and, in doing so, acts as a friend for the fetus. However, these examples of placental changes to maintain a stable fetoplacental interaction through homeostatic mechanisms have been established under certain experimental conditions, and generalization is difficult at this point. There are some aspects that may modify the protective function or at least shift the level of tolerance toward metabolic and inflammatory disturbance. If maternal disturbances affect fetal growth and development, this depends not only on the placental homeostatic capacity, but also on the metabolic load to which the fetoplacental unit is exposed (Fig. 4). This model resembles the metabolic capacity/metabolic load model for fetal and neonatal programming (47).

The placenta is well known to show sex-specific differences in expression of genes and function, and the nature of these differences depends on placental cell type (48–50). Also, the response to environmental stimuli, such as maternal dietary factors, differs between sexes (51). This seems to favor flexibility and adaptive responses in placentas of female fetuses. Insulin resistance found in female neonates born to women with diabetes serves as a cellular antioxidant defense mechanism (52) and is a further protective strategy of the fetoplacental unit designed by evolution to favor reproductive outcome in females. The association of cord blood insulin with subcutaneous fat in the newborn is sex specific (53). This adds evidence to the notion of differential susceptibility of fetal tissues to diabetes- and/or obesity-associated neonatal hyperinsulinemia depending on fetal sex.

Therefore, it can be expected that some sex-specific differences exist in the placental capacity to mount adaptive responses to the maternal diabetic and/or obese environment. The presence of transcriptome sex differences in the late first trimester placenta suggests different susceptibility and stress response mechanisms early on (54).

The protective mechanisms described above were all established with endothelial cells from female fetoplacental units, and it remains to be seen whether they are also operative to a similar extent in male endothelial cells.

While it is obvious that distinct placental cell types respond differently to environmental cues, even within distinct placental cell types, regional differences in the protein repertoire may lead to different responses even within distinct cell types. A clear example of such regional differences is the presence and density of endothelial insulin receptors along the vascular tree (55), with high density in areas destined to drive vascular growth. Thus, the response to fetal hyperinsulinemia likely differs by region. Currently, it is unclear whether regions with low receptor density show similar responses as described previously. It can be hypothesized that regional differences are also present for all other cell-surface or intracellular molecules. Hence, functional responses of the placenta to changes in utero and the fetoplacental circulation may vary by region.

Most placenta studies in diabetes and/or obesity have been carried out in mild forms of the conditions, and only modest changes in the placenta were found. For example, proteomic analyses of placentas in well-controlled GDM at the end of pregnancy demonstrated only minor changes, although some proteins were hyperglycosylated (56,57).

Placental lipid metabolism is strikingly and nonlinearly affected by maternal conditions. The enzyme endothelial lipase (58) is the predominant lipase on the microvillous membrane of the syncytiotrophoblast. It is involved in hydrolyzing phospholipids and triglycerides, and the fatty acids released by lipolysis can be taken up by the syncytiotrophoblast for further metabolism. The enzyme expression is unaffected by maternal obesity or GDM, but it is profoundly upregulated by 100% when GDM is combined with obesity. This suggests the presence of a threshold of tolerance to inflammatory mediator levels such as leptin and tumor necrosis factor-α levels associated with maternal obesity and low-grade inflammation, which regulate endothelial lipase. The enzyme responds only when threshold levels are exceeded, again demonstrating a protective effect for the fetoplacental unit.

The free fatty acids taken up by the placenta enter metabolic pathways, one of which is their re-esterification and storage as triglycerides in syncytiotrophoblast lipid droplets (59,60). The removal of free fatty acids from the cytoplasm protects the syncytiotrophoblast from lipotoxicity (61) and, hence, can be regarded as a further protective mechanism. In mothers with modest obesity (obesity class I), the placental triglyceride content is higher than in lean women. However, more pronounced maternal obesity (obesity classes II and III) is not associated with a further increase in placental triglyceride content, which levels off, demonstrating the capacity limit of protection. It is unknown whether the excess fatty acids in the syncytiotrophoblast not stored in triglycerides affect trophoblast function or spill over to the fetal circulation and contribute to fetal fat accretion.

In this article, I have attempted to bring together the available evidence and embed it in a conceptual framework encompassing early and late pregnancy periods.

In the past decades, during which I have attempted to understand the role of the placenta for the growing fetus especially in conditions characterized by maternal overnutrition, I have gradually come to the conclusion that the placenta at the end of pregnancy is regulated/controlled by the fetus and acts more as a friend of the fetus. Despite manifold studies, there is no evidence for a placental role late in pregnancy contributing to adverse outcomes. However, these studies including our own have the major limitation that 1) only mild forms of maternal metabolic, endocrine, and inflammatory disturbance have been included and 2) they have focused exclusively on the end of pregnancy. In other situations, placental response may be different.

Hence, the answer to my question raised at the outset of this article, of the placenta being a friend or foe for the fetus, is not straightforward but rather more complex. The major determinant for the answer to the friend/foe question certainly is the period in pregnancy in which the maternal metabolic and/or inflammatory disturbances begin to act on the placenta.

The scarce information about effects of the diabetic or obese environment on the placenta early in pregnancy does not yet allow a conceptual conclusion. This pregnancy period is an underresearched area because of ethical constraints. Animal studies do not help because animals follow different growth strategies than humans and their placentas are structurally and functionally different. In vitro studies need to take into account the changes in ambient oxygen tension during the first trimester as well as the week of pregnancy when the placental tissue was collected for research.

Clinical research can strongly contribute by measuring crown-rump length as well as placental volume from as early as possible in the first trimester and by correlating both measures with maternal metabolic and endocrine factors. This would allow determination of the effects of early metabolic changes on placental and fetal growth throughout pregnancy as well as on neonatal outcomes, including body composition.

Much more research is needed in more severe metabolic, endocrine, and inflammatory conditions of the mother at the end of gestation. The continuous improvement in maternal metabolic control has made it difficult to prospectively carry out these studies. Analyzing archival material with state-of-the-art methods developed especially for these types of samples may be an option. This will lead to a better understanding of which cellular and molecular processes are most susceptible to profound environmental perturbations. Establishing exposure-response relationships in the placenta will also allow assessment of whether and how these perturbations affect fetal growth and development and how this is manifested in the neonate. Explanations for the most severe adverse pregnancy outcome in maternal diabetes and obesity, i.e., stillbirth, may then be found. Placental failure to mount adaptive hypervascularization and to upregulate maternal-to-fetal iron transfer efficiency may turn out to be contributors.

Finally, in an ideal world, rather than associating placental changes with maternal or fetal/neonatal parameters, placental tissue collected from randomized controlled trials could be used to demonstrate cause-and-effect relationships. Statistical path analyses could then help to identify placental mediators of maternal changes early in pregnancy, leading to altered neonatal outcomes.

Emphasis on the early pregnancy period was suggested almost 35 years ago by Claude Michèle Poissonet, who raised the question, “…should diabetes in pregnant women be carefully managed in early gestation in the hope of preventing excessive body fat deposition in the conceptus?” (62). The time is now ripe to focus our research on this period, as has also been recommended by some international organizations (63,64).

Acknowledgments. I am especially grateful to my mentors Eleazar Shafrir, Hadassah University Hospital, Jerusalem, Israel, and Peter Kaufmann, Rheinisch-Westfälische Technische Hochschule Aachen, Aachen, Germany, who both have patiently introduced me to the complexities and intricacies of diabetes and the placenta. I am also grateful to all members of the team in the laboratory, past and present, who have enthusiastically worked and generated most of the data on which the framework presented in this article is based. I am greatly indebted to a large number of people around the globe for many exciting and inspiring discussions and exchange of thoughts as well as to reviewers of manuscripts and discussants of talks, who, with often seemingly simple questions, have made me think deeply to reframe and refine the concepts and to develop a more global view. I also wish to thank Mireille van Poppel, University of Graz, Graz, Austria, for critically reading the manuscript and for her important input. The artwork of Figs. 1–4 was reproduced with permission of biolution GmbH.

Funding. The work of the author has been mostly funded by the Austrian Science Fund (FWF), the Jubilee Fund of the Austrian National Bank, and the European Commission.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.