Minireview: The Impact of Antenatal Therapeutic Synthetic Glucocorticoids on the Developing Fetal Brain

Abstract

The life-threatening, emotional, and economic burdens of premature birth have been greatly alleviated by antenatal glucocorticoid (GC) treatment. Antenatal GCs accelerate tissue development reducing respiratory distress syndrome and intraventricular hemorrhage in premature infants. However, they can also alter developmental processes in the brain and trigger adverse behavioral and metabolic outcomes later in life. This review summarizes animal model and clinical studies that examined the impact of antenatal GCs on the developing brain. In addition, we describe studies that assess glucocorticoid receptor (GR) action in neural stem/progenitor cells (NSPCs) in vivo and in vitro. We highlight recent work from our group on two GR pathways that impact NSPC proliferation, ie, a nongenomic GR pathway that regulates gap junction intercellular communication between coupled NSPCs through site-specific phosphorylation of connexin 43 and a genomic pathway driven by differential promoter recruitment of a specific GR phosphoisoform.

Glucocorticoids (GCs) are essential for mobilizing biological processes not required for fetal viability during the transition of the mammalian fetus from intrauterine to extrauterine life. The overall action of endogenous fetal GCs is to trigger organ maturation, enabling the lungs, liver, gastrointestinal tract, thyroid gland, adrenals, and kidneys to function and sustain life outside the uterine environment (1). Various tissues in the developing fetus express the glucocorticoid receptor (GR), and it is these primary organs that undergo a maturational shift to prepare the infant for parturition and ex utero survival. For example, in the lungs GCs trigger thinning of the alveolar septae and rapid maturation of alveoli, production of collagen and elastin, and production and release of surfactant proteins and phospholipids (2–5). GCs also improve the ability of the lungs to resorb fluids by increasing ion channels in the pulmonary epithelium and up-regulating β-adrenergic receptors (4, 6, 7). In the liver, GCs increase protein and glycogen synthesis as well as alter the expression of gluconeogenic enzymes, fatty acid synthase, aminotransferases, and thyroid hormone metabolism (1, 8–12). GC responses in the gut lead to increases in the number and height of villi and migration of enterocytes. As a result, digestive activity and hormone release are augmented (1, 13–18). The increase in the fetal kidney's resorptive ability and decrease in the fraction of excreted sodium are due in part to GC up-regulation of the Na⁺/H⁺ exchanger and Na⁺/K⁺ ATPase (19–23). Erythropoietin production decreases as do renin levels and angiotensin II receptor expression, but the renin-angiotensin system becomes more responsive to hypovolemia (24–28). GCs also hasten thyroid maturation, increasing thyroid hormones that are critical to neurodevelopment (1). Finally, in the fetal adrenals GCs impact cytoarchitecture of the zona fasciculata and induce cytochrome P450s, phenylethanolamine N-methyltransferases, and ACTH receptors (1). These specific developmental requirements for GCs are reflected in the ontogeny of circulating GC levels in the fetus. Specifically, human fetal serum cortisol levels as measured in the umbilical cord demonstrate a fall in midgestation and a rapid rise in late gestation (Table 1) (29).

Table 1.

Serum Cortisol Levels in the Human Fetus (29)

Weeks of Gestation	Average Fetal Serum Cortisol, ng/mL
15–17	8.4
17.5–20	4
35–36	20
37.5–40	45.1

Therapeutic use of synthetic GCs in pregnant women to reduce complications of premature birth

Synthetic GCs are prescribed to women at risk for preterm labor to decrease the morbidity and mortality associated with prematurity. This therapeutic modality was originally tested in fetal sheep in the late 1960s (30) and became a part of clinical practice after a 1972 landmark study. In that report, Liggins and Howie (30, 31) found that a 2-day course of synthetic antenatal GCs decreased respiratory distress syndrome (RDS) and intraventricular hemorrhage (IVH), specifically in infants under 32 weeks' gestation whose mother had at least 24 hours between steroid administration and delivery. Subsequent large cohort studies determined that antenatal GC therapy also reduced necrotizing enterocolitis and infant mortality (32, 33). Two types of synthetic GCs have historically been used for antenatal administration, dexamethasone (Dex) and betamethasone (Beta). Although Dex and Beta are very similar structurally, varying only in the position of one methyl group, they differ in their recommended dosing schedules, placental metabolism, affinity for GR in some fetal tissue, and potency of nongenomic effects (34–36). Recent studies suggest that the preparation of Dex or Beta may also contribute to their differing clinical effects as well (37, 38).

Currently used regimens for antenatal GCs were determined empirically and later established by a 1994 National Institutes of Health Consensus Conference and a meta-analysis of 18 trials by Crowley (32) and have remained unchanged since the study by Liggins and Howie in 1972 (31). The plasma unbound GC level achieved by this regimen is physiological; Ballard et al (39) found that it is comparable with stress-level plasma GCs as a result of endogenous cortisol production in premature infants with RDS. However, decreasing GCs to one single dose of Beta has long been considered, and a study using sheep has found that this permits lung maturation and improved cardiac and renal function (40, 41). Historically, multiple courses of GCs have been administered if the delivery does not occur within 7 days, although the recently published Multiple Courses of Antenatal Corticosteroids for Preterm Birth study suggests that multiple courses of antenatal GCs lead to a decreased length, weight, and head circumference at birth, compared with a single course (42). However, in multiple placebo controlled trials using weekly administration of GCs, the incidence of RDS and mortality was similar to one single course (43).

Complicating the issue is a recent Cochrane review suggesting there is benefit with regard to RDS in multiple courses of antenatal GCs when given to women who remain at risk of preterm delivery after 7 days since the initial course (44). Currently the same dosing regimen is used for singletons and multiple gestations (45), including whether multiple gestations are composed of one or more placentas. A recent study suggests that, at least for Beta, umbilical cord concentrations of Beta are similar for singleton or multigestational pregnancies and also suggests that there is no significant difference in cord concentrations if the mother is obese or of optimal habitus (46). The route of administration has traditionally been intramuscular because oral antenatal GCs have been linked to an increased risk of early-onset neonatal sepsis and IVH (47). Finally, neither gender nor race is taken into account in dosing regimens, despite a clear difference in some outcomes of Caucasian males and females compared with those of African descent (48, 49).

Recent studies have examined whether one particular synthetic GC preparation is substantially more beneficial than another. Although Dex may generate a more pronounced decrease in IVH and possibly time under neonatal intensive care (50), Beta may be more effective at preventing respiratory complications in very low-birth weight infants (51) and in mice (37) and may decrease incidence of periventricular leukomalacia in human studies (52). However, conflicting results have precluded the designation of an optimal GC dosing regimen (50, 51). In the United States, Beta is preferred over Dex for antenatal use in women at risk for preterm delivery.

A more controversial historical use of antenatal GCs is for the treatment of congenital adrenal hyperplasia (CAH). CAH is an autosomal recessive disorder primarily affecting the adrenal cortex. Ninety-five percent of CAH cases are caused by deficiency of 21-hydroxylase, the enzyme that converts 17-hydroxyprogesterone to 11-deoxycortisol. This results in deficiencies of cortisol and aldosterone and excess of androgens (53). Although classical CAH may result in childhood virilization in males, in utero virilization and the development of ambiguous genitalia uniquely affect females, and for this indication, GCs have been given early in gestation and continued throughout pregnancy for any fetus at risk for CAH based on parental genetics (53, 54). However, seven eighths of infants treated, all males and unaffected females, receive no benefit due to the need to treat before the earliest possible time for prenatal genetic testing (54). Additionally, in both rodent and primate models, GC treatment for CAH has been associated with hypertension, hyperinsulinemia, hyperglycemia, fatty liver on a high-fat diet, hypothalamus-pituitary-adrenal (HPA) hyperreactivity, and poorer learning and memory (55–57). Children who received treatment have experienced low birth weight (LBW), failure to thrive, developmental delay, mood disturbance, poor school performance, and social anxiety (58–61). Given this, antenatal GCs for prevention of female virilization in CAH are currently reserved for clinical trials (62).

Unlike endogenous GCs, Dex and Beta are not inactivated by enzymes found in the placenta (63). Furthermore, GCs easily cross the blood-brain barrier via simple diffusion (64) and therefore have access in the fetal brain to GR. In contrast to their beneficial effects on various fetal organ systems, synthetic GCs may negatively impact development of the fetal brain. In nonhuman primates treated with antenatal GCs and either delivered prematurely or at term, treatment was associated with lower brain and cerebellum weight and reductions in the dentate gyrus and cornu Ammon (CA) region of the hippocampus. These cellular abnormalities are associated with lower levels of the presynaptic protein synaptophysin and microtubule associated proteins in the frontal region (65). At 20 months of age, young monkeys who received antenatal GCs still had a reduction in hippocampal volume, which may be due to a chronically elevated cortisol level causing neurotoxicity (66). In nonhuman primates, antenatal GCs were associated with poorer concentration, learning and memory, and hyperactivity (67, 68). Similarly in rats and mice, antenatal GCs were associated with decreased learning and memory along with increased anxiety (69, 70). Antenatal GC administration was associated with changes in basal and stress-induced HPA axis function in a variety of mammals, but these effects are not generalizable across age and species. In species with young born at an advanced stage of development, antenatal synthetic GCs suppress HPA axis function early in life and in adulthood and briefly increase it in the juvenile period. In species with young born in an underdeveloped state, the HPA axis function is generally elevated by antenatal GCs throughout the life span (71).

In humans, antenatal GCs have been associated with structural changes in the brain. For example, antenatal GCs trigger cortical thinning specifically in the rostral anterior cingulate cortex in children 6–10 years of age who were born at term (63) as well as decreased cortical surface area and complexity of cortical folding in infants born at term (72) and periventricular leukomalacia in premature infants assessed at 2 years adjusted age (73). In clinical studies, antenatal GCs have been linked to neuropsychiatric changes in children, including attention deficits (74), decreased scores in cognitive tests (75), distractibility, and aggressive behavior (76). Thinning of the left rostral anterior cingulate cortex, as reported above, is associated with increased incidence of affective disorders (63). Although the mechanism(s) responsible for these alterations are not fully established, they may involve changes in neural stem cell proliferation and/or differentiation, thereby disrupting the development of neuronal circuits essential for higher order cognitive or behavioral function. GC effects in isolated neurons and glia are widespread, triggering a decrease in glucose uptake, inhibition of proliferation, decreased neuronal excitability, and increased dendritic atrophy (64). In a rodent model of antenatal GC administration, a single course of Beta produced significant anatomical differences in interneurons of the hippocampus (77). Antenatal GCs leads to potentially long-lasting alterations of the HPA axis in humans as well (78). For example, antenatal GCs are associated with increased stress reactivity (greater elevation in cortisol level when faced with stress) in female children, even when these children were born at term and maternal stress during pregnancy was controlled for (79).

Maternal stress and programming of the fetal brain

In addition to synthetic GCs, maternal stress has been shown to detrimentally affect fetal programming and brain development through elevated levels of endogenous GCs. In several animal studies, prenatal stress has been associated with reduced volume in several brain regions, including the amygdala, cerebral cortex, hippocampus, and the corpus callosum (80). Human studies have also shown that the timing of maternal stress has differential effects on fetal development. Maternal stress during late gestation leads to a decreased stress response in childhood and adolescence as measured by cortisol levels. By contrast, maternal stress throughout the second half of gestation increased stress responses. Girls born to mothers who experienced anxiety during the first trimester were found to have an increased susceptibility to affective disorders, whereas children whose mothers experienced stress during late gestation were more prone to developing attention deficit hyperactivity disorder (71).

A possible mechanism by which maternal stress alters fetal development is by changing gene expression patterns in the placenta. Male placentas of mice exposed to prenatal stress early in gestation (early prenatal stress) were found to have significantly increased expression of genes regulating growth and development. These changes were associated with maladaptive stress behaviors in adult early prenatal stress male mice (81). In humans, endogenous GCs have also been found to affect placental function by stimulating the placenta to synthesize and release CRH, which could subsequently stimulate the fetal HPA axis (82). Importantly, maternal stress affects pathways distinct from synthetic GCs. For instance, endogenous GCs bind to both GR and mineralocorticoid receptor with equal affinity, whereas synthetic GCs bind only to GR. Synthetic GCs can also bind other orphan nuclear receptors in the brain. Furthermore, placental 11β-hydroxysteroid dehydrogenase (HSD) preferentially inactivates endogenous GCs to cortisone during early gestation; however, levels of 11β-HSD decrease during late gestation, a period in which maternal stress may potentiate the actions of synthetic GCs (82). Finally, endogenous GCs are secreted in a pulse-like fashion, resulting in transient, acute changes in transcription unlike synthetic GCs, which continue to affect transcription after they have been withdrawn. This biological rhythmic secretion of GCs may serve a physiological role by programming cells to respond rapidly to stress, a response that may be impaired due to prolonged exposure to synthetic GCs (83).

Studying the impact of GCs on neural development using primary stem cell cultures

To provide mechanistic details of the neurodevelopmental consequences of antenatal GC exposure, our group (84, 85) and others have studied the effects of GCs on primary murine fetal neural stem/progenitor cell (NSPC) cultures. GCs exert an antiproliferative effect on NSPCs via multiple mechanisms. In rat NSPCs cultured as neurospheres, Dex treatment triggers cyclin D1 degradation via the ubiquitin-proteasome system, thereby inhibiting cell-cycle progression and proliferation (86). In a follow-up study, Dex mediated a decrease in NSPC proliferation through an independent mechanism, the down-regulation of BRUCE/Apollon, a member of the inhibitors of apoptosis protein family found in neuroblasts (87). Dex regulates BIR repeat-containing ubiquitin-conjugating enzyme (BRUCE) at the mRNA and protein level, partially by up-regulating the deubiquitinating enzyme Usp8/Ubpy, which then stabilizes the ubiquitin ligase Nrdp1 and enables it to target BRUCE for degradation (87). Similar studies using gene expression profiling in NSPCs derived from whole brains, found that Dex up-regulated ferritin heavy chain 1 and IGF binding protein 3, which exerted inhibitory effects on cyclin D1 and nestin, a marker of NSPCs (88). These studies also demonstrated that Dex down-regulated the endothelin receptor type B, a change that has been associated with apoptosis in the dentate gyrus and cerebellum and decreased proliferation in the cerebellum (88).

The antiproliferative effects of GCs on NSPCs have also been demonstrated in vivo. Embryonic rats treated with Dex at embryonic day (E) 14 or E15.5 for 3 days exhibited decreased levels of proliferating NSPCs in the striatum and hippocampus (86) as well as dentate gyrus (70). Administering Dex to neonatal rats from postnatal days 1–7 led to apoptosis and depletion of the NSPC pool in the subgranular zone of the dentate gyrus (89, 90). Endogenous GCs may in fact underlie the reduction in the NSPC pool with increasing age (89). Although not within the scope of this review, GCs have also been found to have similar antiproliferative effect on adult NSPCs in vitro and in vivo (91–93).

Early prenatal exposure to GCs is associated with behavioral changes later in life. Evidence that early exposure to GCs leads to long-lasting changes in the molecular profile of NSPCs comes from several sources. In NSPCs isolated from E15 rat cerebral cortices, the antiproliferative effects of Dex are associated with acute and chronic up-regulation of the cell cycle inhibitors p16 and p21. These alterations are accompanied by changes in target genes involved in senescence, such as Bmi1 and Hmga1 (94). Because senescence is related to mitochondrial dysfunction and susceptibility to oxidative stress, Dex down-regulated the mitochondrial proteins nicotinamide adenine dinucleotide hydroxide dehydrogenase 3 and cytochrome b and increased the production of reactive oxygen species and apoptosis when challenged with an oxidative stress inducer (94). The Dex-induced changes in mitochondrial and senescence genes and Dex-induced changes in DNA methylation after several passages suggest an epigenetic reprogramming of NSPCs (94).

Evidence for cell type-specific effects of Dex has come from several studies. Whereas prior studies used 10⁻⁶ M Dex in neural stem cells (94), Yu et al (89) found that 10⁻⁵ M Dex induced apoptosis in a rat hippocampal culture affecting mitotic and resting cell populations and both neurons and NSPCs but not astrocytes. In human NSPC cultures derived from gestation weeks 16–19, Dex similarly decreased proliferation but also decreased the percentage of neurons in differentiating cultures while increasing the proportion of glia (95). These Dex effects were mediated by GR binding to the promoter of Dickkopf1, leading to up-regulation of its expression and subsequent repression of canonical Wnt signaling (95). Several lines of evidence also support a role for GC-induced changes in oligodendrocytes (96, 97). In utero exposure to GC is associated with hypomyelination, although these effects are not due to direct changes in oligodendrocyte progenitor cell proliferation, maturation, or survival and are thought to be secondary to deficits in the primary cellular targets, microglia, and astrocytes (98). Dose- and drug-specific responses determine cellular outcome with high doses of Dex, reducing myelination by altering genomic responses (such as Olig 1 in microglia), whereas astrocyte production relies on the GR nongenomic or rapid pathway (99).

Although many studies demonstrate antiproliferative and proapoptotic effects of GCs, one study reported a dose-dependent proproliferative effect of GCs, including Dex, Beta, and hydrocortisone, on human induced pluripotent stem cell-derived NSPCs, leading to an increase in the number of microtubule-associated protein 2-positive neurons (100). In particular, hydrocortisone stimulated NSPC and neuronal proliferation, even under conditions of oxidative stress, which was hypothesized to be due to the effects of mineralocorticoid receptor activation by hydrocortisone or selective 11β-HSD2 inactivation of hydrocortisone (100).

Similar to Li et al, our studies use primary NSPCs isolated from E14.5 mouse cerebral cortex and cultured as three-dimensional neurospheres (84). To maintain their undifferentiated state, neurospheres are cultured on ultralow adherence plates in the presence of epidermal growth factor (EGF) and/or fibroblast growth factor-1 (FGF-1) (Figure 1). EGF and FGF are crucial for promoting renewal and expansion of the NSPCs. NSPCs derived from E14.5 brain are advantageous for analysis of antenatal GC effects because GR is expressed in NSPC domains in the developing cerebral cortex in vivo at this age (101). Notably, this is a time when embryonic exposure to endogenous GCs is minimal. The major advantage of this model lies in the ability of neurospheres to properly recapitulate the in vivo differentiation order of neurons and glia under in vitro conditions (102, 103) as well as in the maintenance of cell-cell coupling even in the in vitro setting (85), allowing the study of gap junction intracellular communication. Finally, alterations in the development of the cerebral cortex, such as those caused by a fetal exposure to GCs can contribute to neurodevelopmental diseases such as schizophrenia, anxiety, depression, autism spectrum disorders, and attention deficit hyperactivity disorder (104).

Figure 1.

Outline of NSPC culture system. Cerebral cortices from E14.5 mice are dissociated into single cells and cultured in the presence of EGF and FGF-1 to form neurospheres (85). After the dissociation of neurospheres, continued growth in EGF and FGF-1 will produce secondary neurospheres, whereas plating in the absence of EGF and FGF-1 can lead to differentiation.

Using NSPCs isolated from the cerebral cortex, recent work from our laboratory has identified a novel rapid signaling pathway that impacts synchronized calcium waves in coupled cells and proliferation (85). Specifically, plasma membrane-associated GR regulates gap junction intercellular communication in coupled NSPCs through a rapid c-Src- and MAPK-dependent phosphorylation of the gap junction protein connexin 43 (Cx43) (85). Furthermore, the mobilization of this rapid GR signaling pathway occurs in lipid rafts and requires caveolin-1 (Cav-1) (85). The impact of Cx43-dependent propagation of synchronized calcium waves on NSPC proliferation observed in our primary cultures (85) has also been observed in the developing cerebral cortex in vivo and may therefore be a general feature of coordinated NSPC proliferation responsible for the generation of appropriately connected neural circuits (105).

As observed with other steroid receptors (106), the rapid signaling pathway we identified in NPSCs is linked to classical genomic GR signaling leading to gene-specific effects on GR targets (84). Specifically, analysis of Dex treated (ie, 4 h) NSPCs derived from mice null for Cav-1 by microarray revealed approximately 100 genes with altered GR transcriptional regulation (84). The mechanistic basis for Cav-1 effects on GR transcription targets is due in part to its impact on site-specific GR phosphorylation. For example, in the absence of Cav-1, GC-induced GR phosphorylation at serine 224 (mouse equivalent of human GR serine 211) is dramatically reduced, whereas phosphorylation at serine 234 (mouse equivalent of human GR serine 226) is unaffected (84). Furthermore, some GR target genes with reduced transcriptional responses to GC in Cav-1 null NSPCs (ie, Fkbp-5 and Sgk-1) exhibit diminished promoter recruitment of GR phosphorylated at serine 224 as revealed by chromatin immunoprecipitation assays (84). Analogous to the unique cistromes of individual GR phosphoisoforms (107), selective effects of the Cav-1 on GR phosphorylation could impact GR target gene selection in NSPCs and thereby influence various responses of these cells to GCs (Figure 2). For example, the lack of an antiproliferative response to GCs in Cav-1 null NSPCs could be due to the loss of hormone induction of Sgk-1, a gene previously established to mediate antiproliferative responses of GCs in cultured human hippocampal progenitor cells (108).

Figure 2.

Model of integration of genomic and nongenomic GR signaling in NSPCs. Ligand binding (Dex shown in figure) to plasma membrane-associated GR can trigger nongenomic activation of specific protein kinases that can either directly regulate gap junction intercellular communication (GJIC) through phosphorylation of Cx43 or genomic actions through direct phosphorylation of GR or other transcription factors.

Future perspectives: reaching an improved therapy: how can the study of NSPCs inform antenatal synthetic GC treatment?

As stated above, both Dex and Beta are administered antenatally to reduce the complications of prematurity. Although upon initial glance the two GCs appear functionally equivalent, the action of these two hormones varies significantly and current studies reveal no consistently better GC for antenatal use (37). Strikingly, although Beta and Dex have similar affinities for the GR and similar genomic potencies, Dex has a 5-fold greater potency for the rapid action of GR than Beta in rat thymocytes (36). Different GR signaling modalities (ie, genomic vs rapid) may exert discrete influences on the developing brain and peripheral organ systems. Thus, the analysis of GR signaling pathways and functional responses of NSPC to novel therapeutic alternatives to synthetic GC administration could provide insights into treatment modalities that maintain beneficial effects in vulnerable organ systems in the premature infant (eg, lung) but limit their adverse neurodevelopmental effects. In addition, to improve antenatal GC therapies, we need a better understanding of the genetic or uterine environmental factors that could modulate GC responses. In term-born children treated during the fetal period with GCs, there is no difference in cortisol reactivity if the child was treated with Dex or Beta (79). However, despite reporting sexually dimorphic changes in cortisol secretion, this study did not separate the Dex/Beta-treated groups by sex prior to analysis. Gender-specific responses of NSPCs to GCs may exist because there are significant gender differences in outcome and infant mortality in response to antenatal GCs in both humans and animal models (79, 109, 110) (reviewed in reference 111). For example, LBW males have a higher risk of IVH and greater mortality than LBW girls (109). Fetal gender is not taken into account in current treatment guidelines, so there is a clear need to develop antenatal GC regimens that would be selectively robust in preterm male vs preterm female infants.

In summary, as additional molecular details of GR action in NSPCs are uncovered, antenatal synthetic GC regimens established more than 40 years ago may be updated to continue to provide life-saving benefits of GCs to premature infants while limiting adverse neurological outcomes that may take many years to be revealed.