Cooperativity and complementarity

Abstract

Glucocorticoids (GCs) are an ubiquitous class of steroid hormones that exert a wide array of physiological effects. Traditionally, GC action has been considered to primarily involve transcriptional effects following the binding of hormone to the glucocorticoid receptor (GR) and subsequent activation or repression of target genes. However, a number of findings suggest that cellular responses following GC exposure may be mediated by transcription-independent, or “non-classical,” mechanisms. We have added to this growing body of work by recently uncovering a novel GC signaling pathway that operates through plasma membrane GRs to limit gap junction intercellular signaling and limit the proliferation of neural progenitor cells (NPCs). In this review, we highlight our current state of knowledge of non-classical GR signaling, in particular as it applies to neuronal function. Using NPCs as a cellular model, we speculate on the components of this non-classical pathway and the mechanisms whereby a number of cytoplasmic and nuclear signaling events may be integrated.

Glucocorticoid hormones (GCs) mediate a wide array of physiological actions following their binding to the glucocorticoid receptor (GR). According to the traditional view of GR action, the effects of GCs are thought to be largely mediated by transcriptional responses (i.e., activation or repression) that follow either the direct binding of a GR-ligand complex to glucocorticoid response elements (GREs) contained within target genes or the indirect association of the receptor with other DNA elements or DNA-bound transcription factors.^1,2 However, more recent evidence suggests that GR may also act via transcription-independent (heretofore termed “non-classical”) mechanisms to mediate rapid cellular responses to GCs in the absence of measurable alterations in gene expression.^2-4 Non-classical steroid actions are characterized by rapid signaling (typically within a few seconds to minutes), insensitivity to transcriptional and/or translational inhibitors, and continued hormone action despite the use of cell-impermeable hormone conjugates.⁵

The idea that steroid hormone receptors and their ligands may have transcription-independent functions has been in existence for at least 60 y. The first report of non-classical GC actions was by Hans Selye, who postulated that GCs may regulate “rapid adaptations to stress.”⁶ In the 1960s, evidence of rapid increases in cyclic adenosine monophosphate (cAMP) following a 15 sec pulse of 17-β-estradiol (E2) as well as evidence of E2 binding sites on the surface of endometrial cells provided some of the first evidence of non-classical signaling by steroid hormones.^7-9 Despite these early discoveries, the field of non-classical steroid hormone signaling largely remained unexplored until the last decade, when novel discoveries on non-classical signaling by estrogen receptor (ER) and GR led to a rapid expansion of our understanding of the role of transcription-independent steroid hormone signaling.⁷

Following a brief overview of classical GR signaling, this review will focus on the implications of more recent discoveries in our laboratory and others on non-classical GR signaling, with a particular focus on our work examining GR effects in neural progenitor cells (NPCs). Interestingly, rapid non-classical GR signaling appears to follow many of the same principles of action previously documented for non-classical ER signaling, suggesting a conserved set of processes utilized by steroid hormone receptors. In addition, the presence of non-classical signaling suggests a greater complexity and diversity to GC and other steroid hormone signaling. Importantly, the strong synergies between rapid, non-classical and slower acting classical signaling appear to have important implications for physiological responses of GC signaling. Given the critical role of GC hormones in normal human physiology, human disease and as pharmacological agents, these findings may have important clinical and therapeutic implications.^10,11

Classical GR Signaling

According to the “classical,” or “genomic,” view of nuclear hormone receptor action, unliganded GR dynamically associates with a number of chaperone proteins. These include the heat shock proteins Hsp90 and Hsp70 and the immunophilin FKB56 among others.^12,13 Their association with chaperone proteins generally restricts GRs to the cytoplasmic compartment.¹² Hormone binding leads to a disassociation of chaperone proteins and nuclear translocation of the GR-ligand complex.⁵ GRs can then bind, usually as homodimers, directly to GREs linked to GC-responsive promoters in various orientations and positions or tether to other DNA-bound transcription factors through specific protein-protein interactions.^5,14 Like other nuclear receptors, GR serves as a platform for the recruitment of a wide variety of coregulator and chromatin remodeling proteins that impact the transcriptional output from the linked promoter.¹⁴

Non-Classical Steroid Hormone Signaling

Pathways and principles.

Non-classical estrogen signaling mediated by its cognate nuclear estrogen receptors (ERα and ERβ) has been the most well characterized of all rapid nuclear receptor signaling pathways. Therefore, our understanding of non-classical ER signaling can serve as a useful model for analogous non-classical signaling by other steroid hormone receptors, such as GR.⁵ Although an area of ongoing debate, data from mouse endothelial cells (ECs) and the human breast cancer cell line MCF-7 suggest that non-classical ER signaling is mediated by estradiol-activated ERα and ERβ localized within the plasma membrane.^7,15 In particular, rapid ER signaling is absent in ECs from ERα/ERβ-knockout mice, and siRNA directed against ERα/ERβ in MCF-7 cells also abrogates rapid ER signaling.⁷ Immunoprecipitation and sucrose gradient fractionation experiments have been used to establish the presence of ER within the plasma membrane and its association with caveolae-containing lipid rafts.⁷ It appears that caveolae rafts provide a physical space, where a number of signaling proteins, including the steroid hormone receptor, mitogen-activated protein kinases (MAPKs), G-proteins and other molecules, can interact.⁷

The membrane localization of ER has been shown to be dependent on palmitoylation of cysteine 447. Mutation of this site prevents plasma membrane localization.¹⁶ Interestingly, a highly related palmitoylation motif including a potentially modified cysteine and the surrounding nine amino acids is present in GR as well, suggesting that GR membrane localization may be dependent on a similar post-translational lipidation¹⁷ (Fig. 1). Curiously, the mineralocorticoid receptor (MR), which shares a high degree of relatedness to GR in this putative palmitoylation motif, does not contain a cysteine residue that would be the site of palmitoylation. Since MR has also been found to be localized to the plasma membrane and to be involved in rapid non-classical signaling, alternative lipid modification sites or distinct mechanisms may exist to target this receptor to the plasma membrane.¹⁷

Figure 1. Palmitoylation Sequences in E-domain of nuclear receptors. The consensus palmitoylation motif for mouse/human ER, AR, PR and GR are remarkably similar sequences with a cysteine “C” at the third position surrounded by 9–11 highly related amino acids. Ω, aromatic; ϕ, hydrophobic; ζ, hydrophilic. Number represents amino acid number from the beginning of the E-domain.⁶³

Similar principles of non-classical action have also been demonstrated in GC/GR signaling, although the precise molecular mechanisms remain less well-defined.^4,5 In addition, while a significant body of the ER literature has focused on the role of non-classical ER signaling on cell proliferation and survival, non-classical GR studies have mainly examined rapid effects of GCs that modulate stress responses emanating from the central nervous system (CNS).¹⁷ There is an ultradian pattern of GC release by the adrenal glands following activation of the hypothalamic-pituitary-adrenal (HPA) axis.¹⁷ The period of this pattern of troughs and peaks is approximately 1 h and lends itself to rapid classical effects of GR on target genes, as the circulating levels of GCs drop below the concentration needed to maintain GR activity during troughs.^17,18 The impact of these ultradian cycles of GC secretion on rapid non-classical signaling has not been explored.

Based on these findings, there appear to be general patterns and principles of rapid, non-classical GR effects that can be instructive. One general principle from these studies is that rapid, non-classical GC effects often seem to be a precursor for more slow acting, but longer lasting, genomic GC effects. In the basolateral amygdala, corticosterone increased the frequency of miniature excitatory postsynaptic currents (mEPSCs) within 15 min.¹⁹ While this rapid effect was found to be non-classical, prolonged GC exposure also increased the frequency of mEPSCs in a manner that was sensitive to protein synthesis inhibitors and was, therefore, a genomic/classical effect.¹⁹ Similarly, in the anterior pituitary, GR activation led to a decrease in ACTH release within 1 min that was sustained 2 hours after the GC administration.¹⁷ The former effect was shown to be non-classical and dependent on the rapid phosphorylation and membrane translocation of annexin-1, whereas the latter was found to be a genomic effect dependent on protein synthesis.^17,20 A second general principle is that the presence of GR (or MR) on the cell surface provides a strong indication for a role for non-classical hormone signaling. A final general principle of rapid non-classical GR signaling is that GC stimulation often acts in a “permissive” manner.¹⁷ That is, the rapid effects of GCs tend to alter pre-existing activity or the threshold of activity of neurotransmitters and/or ion channels rather than inducing or inhibiting function in an all or none fashion. These insights, as well as knowledge of non-classical signaling from other hormone receptors (such as ER) provide a strong foundation upon which to explore GR action in different biological systems, particularly during neurodevelopment, where rapid GC action has been largely unexplored.

GC Signaling

Clinical applications and neurodevelopmental consequences.

The true importance of fully understanding the breadth of GC signaling, including non-classical actions, stems partly from the multiple biological functions and widespread therapeutic use of glucocorticoid hormones. GCs play a major role in many vital processes, including growth, reproduction, intermediary metabolism, water and electrolyte homeostasis, cardiovascular function and immune and inflammatory functions.

Since the landmark study by Liggins and Howie in the 1970s, dexamethasone (DEX) or betamethasone has been used to promote lung maturation in premature infants and decrease neonatal mortality.²¹ The mechanism(s) of this beneficial effect includes induction of the type II alveolar cells to increase surfactant production and stimulate structural development of the fetal lung, resulting in improved neonatal respiratory function.²² Cohort and animal studies using repeated courses of antenatal glucocorticoids have demonstrated no significant additional benefits but raised the possibility of deleterious effects on birth weight, head size, brain growth and blood pressure.²³ Indeed recent animal data suggest that antenatal glucocorticoid exposure may have long-term programming consequences, increasing risks for diabetes mellitus and hypertension.²⁴

DEX has also been administered postnatally for the prevention of respiratory distress syndrome.²⁵ Yeh and colleagues conducted a double-blind placebo controlled study of early post-natal DEX treatment; they found that the treated children were shorter, had smaller head circumferences and had deficits in cognition and neuromotor skills.¹¹

In addition, DEX has been administered to some women at risk of having a child with classical forms of congenital adrenal hyperplasia (CAH), a group of autosomal recessive disorders. The most common type of CAH is a 21-hydroxylase deficiency due to mutations in the CYP21A2 gene. The loss of negative feedback inhibition leads to excessive androgen secretion during gestation, which induces virilization of the external genitalia with clitoral enlargement, fusion of labioscrotal folds and displacement of the urethral meatus in affected female fetuses.²⁶ The distress and embarrassment regarding the external genital appearance can negatively impact quality of life for the affected girl and her family. To diminish prenatal adrenal androgen production and decrease the extent of virilization of the external genitalia, prenatal DEX has been administered to mothers at risk for another child with CAH since the 1980s.^27,28

Since the process of external genital differentiation begins by 6–7 weeks after conception, mothers need to start DEX early in pregnancy to prevent the genital anomalies. Genetic diagnosis cannot be safely performed until 10–12 weeks of gestation. Thus, all at-risk pregnancies are treated. Available data show decreased birth weight in treated infants compared with untreated infants.²⁹ Although one study showed no significant differences regarding nine social/developmental scales, unaffected children exposed to DEX during the first trimester had an impaired verbal working memory, low self-perceived scholastic competence and increased self-rated social anxiety.^30-32 In light of the available outcome data, the Endocrine Society has reviewed the role of prenatal DEX treatment for infants at risk for classical CAH and concluded that “prenatal therapy continue to be regarded as experimental.”³³

In adults, exposure to chronic levels of endogenous stress hormones such as the GC cortisol is associated with mood disorders and cognitive deficits that may be linked to hormone effects on cell proliferation.¹⁰ Specifically, these illnesses have been partly attributed to GC inhibition of adult NPC proliferation.³⁴ One of two major sites of adult neurogenesis is in the dentate gyrus (DG) of the hippocampus. Chronic stress has been shown to lead to hippocampal atrophy, which is also correlated with increased incidence of depression as well as deficits in learning and memory.³⁵ Stress-induced suppression of cell proliferation in the DG of the hippocampus has been observed in a number of different mammals, including rats and mice, in response to a variety of stressors including footshock, restraint stress and predator odor.³⁴

GCs have specifically been cited as important in the reduction of cell proliferation from these stress-inducing manipulations, because adrenalectomy or blockade of HPA axis receptors increase cell proliferation.^34,36 The inhibitory effect of GCs on DG cell proliferation is likely mediated through GR but not MR. Even though both of these GC receptors exist in the adult hippocampus, the high affinity MR is occupied under baseline (unstressed) levels of GC circulation, unlike GR.³⁴ Furthermore, pharmacological blockade of GR in rats prevented the loss of cell proliferation produced by elevated corticosterone levels.³⁷

Based on these clinical findings, gaining a better understanding of the mechanisms that underlie GC effects on NPC proliferation has the potential to have far-reaching clinical implications for a number of diseases that affect both adults and children. Recent findings in our laboratory and others has led to a role for a seemingly unlikely candidate, gap junctions, in regulating neurodevelopment and mediating GC effects on NPC proliferation.^38-41

Gap Junctions

A mediator of GC effects on NPC proliferation.

Gap junctions form intercellular channels between adjacent cells that allow the passage of ions and molecules less than 1 kD in size.⁴² Mammalian gap junctions are composed of connexin proteins, six of which make up a single connexon, or hemichannel.⁴² Two opposing hemichannels on adjacent cells, in turn, constitute a gap junction⁴³ (Fig. 2).

Figure 2. Gap junctions are composed of connexin proteins. Six individual connexin proteins combine to form a gap junction hemi-channel. Two hemi-channels from adjacent cells form a gap junction through which metabolites and small molecules less than 1 kD in size can pass. Connexin 43 gap junctions can be phosphorylated by ERK-1/2, and this phosphorylation leads to inhibition of gap junction intercellular communication.⁴⁴

Over 20 connexin genes have been identified and classified according to predicted molecular weight. Furthermore, expression of the various connexins is highly tissue-specific.⁴² The most ubiquitous and most extensively studied connexin is connexin 43 (Cx43).⁴⁴

Of the 20 connexin subtypes identified, at least five connexins are known to be expressed in the rodent cerebral cortex.^42,45 While there are clear temporal and spatial variations in the expression of connexin isoforms within the CNS, both Cx43 and Cx26 are localized to proliferating and undifferentiated neuronal progenitors.^42,46

Gap junctions and their constituent connexin proteins may play a number of important roles in the development of the embryonic brain.^42,45 Gap junction coupling has been demonstrated during most stages of embryonic cortical development and remains prominent during the early postnatal period.⁴⁷ Pharmacologic disruption of GJIC in NPCs decreases the rate of proliferation and prevents cells from entering the cell cycle.⁴⁷ Inhibition of Cx43 gap junctions in in vitro NPC culture led to decreased proliferation and increased differentiation of these cells. In addition, NPCs remained in a proliferative state following basic fibroblast growth factor (bFGF) withdrawal if Cx43 was simultaneously overexpressed, indicating that Cx43 mediated GJIC may be necessary for NPC proliferation.³⁸

The specific mechanism whereby gap junctions facilitate NPC proliferation remains an active area of investigation. One area of focus has been the role of gap junctions and hemichannels in the propagation of Ca²⁺ waves. Intercellular Ca²⁺ signaling was significantly decreased following pharmacological inhibition of GJIC in coupled HEK293 cells.⁴⁸ In addition, the propagation of spontaneous Ca²⁺waves through Cx43 hemichannels was shown to be necessary for the proliferation of radial glial cells in the embryonic ventricular zone (VZ).⁴¹ Furthermore, gap junction-mediated passage of small molecules such as cAMP or cell cycle proteins have also been posited to influence cell proliferation.^49,50 More recent work has also suggested that Cx26 and Cx43 may act as adhesive proteins facilitating radial glial cell migration during embryonic cerebral cortical development.³⁹ According to these studies, decreased cell numbers that are observed following gap junction inhibition or knockout during cortical development may be partly explained by connexin-dependent deficiencies in progenitor cell migration.

GCs: Non-Classical Effects on NPC Proliferation

Recent work from our laboratory provides further evidence of a role for connexins in NPC proliferation, and suggests that a novel non-classical GC signaling dependent pathway may be involved in modulating neurodevelopment. Using embryonic murine NPCs, we identified a non-classical GR signaling pathway that impacted proliferation in vitro through inhibitory effects on GJIC.⁴⁰ Phosphorylation of specific connexin proteins has been shown to regulate GJIC in other systems.⁵¹ In agreement, we provided evidence for rapid activation of extracellular signal regulated kinase-1/2 (ERK-1/2) by GCs that triggers site specific phosphorylation of Cx43. This phosphorylation event, in turn, leads to reduced GJIC. Interestingly, GCs do not appear to influence Cx43 (or Cx26) protein expression or subcellular trafficking in murine NPCs. Rapid GR-dependent activation of ERK-1/2 requires a c-src family member and may be initiated by a signaling complex assembled at the plasma membrane through GR interactions in lipid rafts containing caveolin-1. Results from experiments using caveolin-1-knockout NPCs corroborated the role of caveolin-1 in mediating the anti-proliferative effects of GCs, which was established previously in mouse embryonic fibroblasts.⁵² These rapid GR responses contribute to both a reduction in S phase progression and enhanced cell cycle exit of NPCs. Interestingly, inhibiting GJIC alone with 1-heptanol reduced S phase entry but, unlike GC exposure, did not enhance cell cycle exit. Lastly, and in agreement with previous results suggesting a critical role for Ca²⁺ signaling in neurodevelopment, transient inhibition of GJIC led to a decrease in synchrony of spontaneous Ca²⁺ transients in NPCs.^40,41

The molecular framework for non-classical GR signaling that we characterized is remarkably similar to what has been previously demonstrated during non-classical ER signaling, and these discoveries suggest a number of interesting possibilities.^7,15 For example, GC exposure may modulate NPC proliferation through dual mechanisms, first through rapidly acting effects on GJIC that reduce S-phase entry, and second through a complementary and transcriptional mechanism that induces cell cycle exit (Fig. 3). Delineating these dual mechanisms presents an exciting challenge for future studies, and many signs seem to point toward a central role for Ca²⁺signaling in this process.

Figure 3. Activation of non-classical and classical pathways by GCs alters NPC proliferation. Hormone (GC) treatment leads to rapid signaling by membrane GR associated with caveolin-1 (Cav-1). Rapid activation of c-src leads to ERK-1/2 activation, phosphorylation of Cx43, and reduction of GJIC. This non-classical signaling reduces S-phase entry. In addition, GRs participate in classical transcription-dependent processes that also reduces NPC proliferation.

Ca²⁺, a GR-Activated Modulator of NPC Proliferation?

In light of our observation of GJIC inhibition and subsequent loss in synchrony of spontaneous Ca²⁺ waves, it is interesting to speculate whether the decreased rate of S-phase progression that we have observed in NPCs results from a loss of Ca²⁺ wave propagation.⁴⁰ The propagation of spontaneous Ca²⁺ waves through gap junction hemichannels has been proposed to be an essential component of neuronal proliferation in the developing cerebral cortex.⁴¹ In particular, Weissman et al. demonstrated that spontaneous Ca²⁺ waves in radial glial cells in the rat embryonic ventricular zone are mediated by gap junction hemi-channels. Inhibition of gap junction communication with the inhibitor carbenoxolone diminished these waves, which, in turn, reduced VZ cell proliferation.⁴¹

The importance of Ca²⁺ release in NPC proliferation was also shown by Lin et al., who demonstrated that Ca²⁺ release was dependent on ATP activation of P2Y1 surface receptors on NPCs.⁵³ Interestingly, ATP was emitted in spontaneous bursts from the proliferating NPCs, and this release was decreased in serum-exposed NPCs that were beginning to differentiate. In addition, loss of the calcium wave, inhibition of the upstream P2Y1 receptor or inhibition of ATP release were each found to diminish cell proliferation, demonstrating an intimate link between these pathways and cell cycle progression (Fig. 4).^41,53

Figure 4. Gap junctions facilitate the movement of spontaneous ATP and Ca⁺ waves needed for proliferation. Open gap junctions may allow passage of ATP, which binds to and activates P2Y1 receptors leading to Ins(1,4,5)P₃-mediated Ca²⁺ release from the endoplasmic reticulum (ER). The Ca²⁺ itself and the Ins(1,4,5)P₃ could travel between cells in a gap junction-dependent manner and thereby regulate cell proliferation in coupled cells.

Cooperation Between Classical and Non-Classical GR Signaling

Importantly, previous discoveries on non-classical signaling as well as our own findings suggest a pattern of cooperation between non-classical and classical signaling in achieving a given biological endpoint. In our studies on NPCs, we found that a reduction in GJIC by heptanol exposure for 1 h leads to a reduction in the number of NPCs in S-phase of the cell cycle. In contrast, a 1 h GC exposure, limited by subsequent RU-486 treatment, forces NPCs out of the cell cycle entirely. This important difference suggests that GC exposure modulates NPC proliferation through a reduction of GJIC as well as through other means. Therefore, while GC-mediated loss of GJIC may affect NPC proliferation rapidly via Ca²⁺ effects, as hypothesized above; GRs may also utilize direct transcriptional targets to impact NPC proliferation. One potential set of candidate genes for GR effects on proliferation are the established regulators of the cell cycle. GR may upregulate the expression of factors that cause cell cycle arrest or repress certain factors that promote cell cycling. In particular, GR has been shown to activate the cyclin-dependent kinase inhibitors (CDIs) p27 and p21, and GR has also been shown to repress the expression of the cyclin-dependent kinases (CDKs) CDK4 and CDK6.⁵⁴

CDKs associate with cyclins at particular points in the cell cycle, and CDK-cyclin complexes phosphorylate and activate proteins that promote cell cycle progression. CDIs bind to the CDK-cyclin complexes and inhibit their kinase activity, thus inhibiting cell cycle progression.⁵⁴ DEX-activated GR in the SAOS2 human osteosarcoma cell line upregulates the expression of both p27 and p21 and causes a decrease in proliferation.⁵⁴ The p21 promoter lacks a consensus GRE but is activated rapidly by GCs even in the presence of translational inhibition with cycloheximide treatment. The promoter does, however, contain a number of half-GRE sequences. In addition, a GR dimerization mutant (i.e., a GR that is unable to form the homodimers that are typically needed for GR mediated transcription) was able to activate p21 induction.⁵⁵ Taken together, these results indicate that ligand-activated GR directly upregulates p21 expression in an atypical manner in SAOS2 cells. Specifically, the rapid induction of p21 mRNA despite the presence of only a half-GRE sequence suggests that GR monomers interact with co-activator proteins to directly upregulate p21.

GC induction of p27 mRNA expression requires 24 h of hormone exposure and does not occur in cells expressing the GR dimerization mutant. Furthermore, this increased mRNA expression was sensitive to cycloheximide treatment. These results suggest that GR may activate p27 transcription indirectly by utilizing direct targets of GR transcriptional responses that require receptor dimerization.⁵⁵

In contrast, DEX-activated GR had no effect on p27 or p21 in the U2OS human osteosarcoma cell line but did repress CDK4, CDK8 and cyclin D3 activity, which also led to a reduction in cell proliferation.⁵⁴ The precise mechanism whereby GR repressed CDK and cyclin activity has not been established. However, the deletion of the N-terminal transcriptional activation domain had no effect on ligand-induced GR repression, while deletion of the GR zinc finger domain that is critical for certain GR interactions with other proteins abrogated the repression activity. Thus, an interaction between ligand-bound GR and an unidentified transcriptional repressor(s) was likely responsible for the effects observed in U2OS cells.⁵⁴ These examples suggest that GR reduces cell proliferation by multiple mechanisms and in a cell type-specific manner (Fig. 5). GC-mediated inhibition of NPC proliferation may also partly be a result of activation of CDIs or repression of CDKs. Importantly, our data suggest that even a 1 hr exposure to DEX is sufficient to induce the transcription of GR target genes, making GR transcriptional effects on proliferation a plausible outcome, even after a GC exposure limited to 1 hr.

Figure 5. GR May Alter NPC Proliferation by non-classical and classical mechanisms. Rapid, non-classical GR signaling inhibits GJIC and may disrupt Ca²⁺ waves. This may alter the activity of Ca²⁺-dependent transcription factors such as CREB that act on cell cycle proteins such as cyclin D1. These effects may be reversible as GJIC returns to pretreatment levels over time. A 1 hr GC exposure may also activate classical GR signaling that upregulates CDIs such as p21. The combination of classical and non-classical signaling may have a more lasting impact on cell proliferation than inhibition of GJIC alone.

Clinical Implications

The relevance of dual classical and non-classical signaling in GC action may be especially important in the clinical context. In many instances where GCs are administered clinically, classical GR transcriptional activity is cited as underlying the neurodevelopmental effects of GCs. However, our findings support the notion that GC hormones may alter NPC proliferation even following limited exposure to GC by activating a MAPK-dependent non-classical signaling mechanism.^40,56-58

Interestingly, the discovery of rapid non-classical effects of GCs on NPC proliferation adheres to an important general principle that has been made in other contexts, where rapid non-classical signaling by GR has been observed. In particular, it appears that NPCs utilize non-classical signaling to rapidly initiate a program to reduce cell proliferation, which, over a longer time course, also contains contributions from classical/genomic mechanisms. This is akin to non-classical GC signaling events in the basolateral amygdala and hypothalamus, where rapid effects occurring in a non-classical and non-genomic manner appear to be a precursor to similar effects over a longer time course that are mediated by classical GR signaling.¹⁷

The most straightforward clinical implication of these findings is that exposing the fetal brain to GC hormones even for a short period can activate signaling cascades that may be sufficient to have negative neurodevelopmental consequences. According to this interpretation, the clinical use of GCs should be guided by the need to balance the potential benefits of hormone on lung and heart development vs. the certain negative consequences on neurodevelopment. However, these data may also suggest an alternative clinical interpretation that is, to some extent, supported by the available clinical data. Perhaps there is a time window during which the effects of GC exposure are largely or almost entirely beneficial, but that longer term or more prolonged exposure tips this balance in favor of negative effects of GC exposure. In light of the data in our studies, this interpretation suggests that perhaps the rapid non-classical effects of GC exposure are reversible and temporary, and that the more deleterious consequences occur when classical/genomic programs are activated by more prolonged exposure to hormone. The examination of proliferation from our own studies lends some credence to this idea, since inhibition of GJIC alone reduced the number of NPCs actively in S-phase of the cell cycle but did not actually force NPCs to exit the cell cycle entirely. Presumably, cell cycle exit is largely a classical GR effect, whereas decreased S-phase entry is a non-classical effect and may be reversible as long as classical GR signaling events are not activated.

An examination of the clinical literature also suggests that long-term exposure to hormone may at least partly underlie some of the more serious negative consequences of pre- and postnatal GC exposure. For example, while DEX has historically been the major GC administered to preterm infants, a growing body of evidence suggests that hydrocortisone may be a superior clinical alternative due to fewer side effects, including neurodevelopmental side effects.^59,60 Although both DEX and hydrocortisone can cross the blood brain barrier, one of the critical differences between these two hormones is that hydrocortisone can be inactivated by 11β-HSD2, whereas DEX cannot. This enzyme is highly expressed in the placenta as well as in the brain for the majority of gestation.⁶¹ Consequently, hydrocortisone administration may primarily activate non-classical signaling mechanisms in the brain before it is inactivated by 11β-HSD2. In contrast, continuous DEX treatment will almost certainly activate classical GR signaling pathways, since DEX cannot be inactivated by 11β-HSD2.

The benefits of shorter hormone exposure have also been demonstrated by the treatment of premature infants with pulsatile DEX therapy instead of continuous DEX. This was shown to be clinically effective in decreasing chronic lung disease and the need for oxygen supplementation and was associated with a reduction in side effects. In this study, infants were given two divided doses of DEX per day for three days, instead of a continuous treatment. While neurodevelopment was not monitored, other common side effects from continuous DEX, such as a significant decrease in weight gain and significant increases in mean arterial blood pressure were not observed.⁶² While even a pulse of DEX may activate classical GR signaling mechanisms, pulse therapy presumably leads to a lower fetal DEX concentration between pulses than a continuous infusion and may, therefore, primarily activate non-classical GR signaling pathways. These findings are therefore important in two regards: first, they indicate that pulsed DEX dosing can be clinically effective; second, even though neurodevelopment was not assessed, they indicate that certain side effects are lower from a pulse treatment. These results warrant a more thorough examination of potential side effects, including on neurodevelopment, from pulse therapy.

In summary, the clinical literature on prenatal and postnatal GC therapy suggests that while hormone treatment has an important, and in many contexts indispensable, therapeutic role, it is also associated with negative side effects. Importantly, the negative effects vary depending on the particular GC used. For example side effects from DEX tend to be more severe than with the use of the natural hormone, hydrocortisone. Combining our results with other findings in the field, we can speculate that the negative outcome is partly a result of the activation of classical GR signaling, which may irreversibly inhibit proliferation and/or alter the differentiation of developing NPCs. Therefore, the selective activation of non-classical signaling pathways may preserve the clinical benefits of pre- and postnatal GC therapy while avoiding some of the negative side effects. This may be achieved by using natural GCs, pulse therapy of synthetic GCs or membrane impermeable versions of synthetic or natural GCs that selectively activate non-classical signals. The search for these ligands provides an exciting course of investigation for the future.

Glossary

Abbreviations:

bFGF

basic fibroblast growth factor

CNS

central nervous system

CAH

congenital adrenal hyperplasia

Cx

connexin

cAMP

cyclic adenosine monophosphate

CDKs

cyclin-dependent kinases

CDIs

cyclin-dependent kinase inhibitors

DG

dentate gyrus

Dex

dexamethasone

ECs

endothelial cells

E2

17β-estradiol

ERK-1/2

extracellular signal-regulated kinase-1/2

ER

estrogen receptor

GJIC

gap junction intercellular communication

GC

glucocorticoid

GR

glucocorticoid receptor

GRE

glucocorticoid response element

HPA

hypothalamic-pituitary-adrenal

MR

mineralocorticoid receptor

mEPSCs

miniature excitatory postsynaptic currents

MAPKs

mitogen-activated protein kinases

NPC

neural progenitor cells

VZ

ventricular zone