Minireview: Estrogen Receptor-Initiated Mechanisms Causal to Mammalian Reproductive Behaviors

Sex hormone-dependent behaviors provide new outlets for novel molecular endocrine analyses of brain-behavior relationships.

Abstract

While estrogen-facilitated changes in gene expression constitute some of the best-analyzed biochemical phenomena in the regulation of transcription, there have been at least two aspects of this topic that have led to much experimental work about estrogen actions on brain and behavior. The first task has required parsing various behavioral and neurochemical functions according to whether they depend on estrogen receptor-α or estrogen receptor-β. The second task has been the formulation of how nuclear actions of estrogens comport with membrane-initiated actions. With respect to these issues, applications of molecular endocrine approaches to lordosis behavior came first. Currently, the last in the chain of reproductive behaviors, maternal behavior, and an entire range of neural and cognitive functions even more complex in their determinants, must be analyzed using current molecular techniques. This minireview of estrogen actions on the chain of female reproductive behaviors highlights challenging new questions about estrogen actions on cells in the brain, questions that have important practical applications far beyond traditionally studied sex behaviors.

Formerly considered a ‘boutique field’ in neuroscience, neuroendocrinology has been shown to be the earliest subfield in which detailed chemical mechanisms of transcription could be linked to complete, biologically crucial mammalian behaviors. As the most prominent example, behaviors essential for reproduction by female laboratory animals, to which this minireview will be limited, constitute a chain of behavioral responses beginning with courtship, leading to copulation (lordosis), which in turn ultimately leads to maternal behaviors. Mechanisms for all of these are understood in some detail, but studies of such mechanisms have led to interesting new questions. Several such questions are mentioned at the end of this minireview.

The first set of mechanisms to be explained were those for the simplest sex behavior, lordosis (1, 2). The elucidation of cell nuclear actions of estradiol (E2) in ventromedial hypothalamic (VMH) neurons, the transcriptional consequences of E2 binding in those neurons, and the demonstration of how those neurons then regulate a neural circuit that produces lordosis behavior showed exactly how chemical changes in specific neurons can facilitate a complete vertebrate behavior.

Another relatively simple analysis deals with the regulation of locomotion involved in the courtship behaviors that precede the male's mount and the female's lordosis. Neurons on the lateral side of the preoptic area are required for normal courtship behaviors in female rats (3). The estrogenic facilitation of locomotion depends on expression of the estrogen receptor-α (ERα) gene (4).

With the clearest sets of findings behind us, two important new issues have arisen. First, after the discovery of the ERβ gene by Kuiper et al. (5), it became apparent that at least two genes encode mRNAs that yield nuclear receptors that bind E2. Both are ligand-activated transcription factors. Second, the notion that important cellular effects of E2 could be initiated outside the nerve cell nucleus was revived by the neurophysiological discoveries of Kelly et al. (6), who uncovered E2-induced changes in neuronal electrical activity that were too fast to be explained by gene transcription. We will address each of these two issues in order and use some of our knowledge of maternal behaviors to illustrate them.

Two Nuclear Receptors for Estrogens

Neuroanatomical localizations of ERα and ERβ gene products by in situ hybridization and by immunocytochemistry have revealed that the two receptors have both overlapping and unique distributions in the central nervous system, depending on the brain region examined (7–9). While the functional consequences of activation of one receptor vs. the other are sometimes very clear (e.g., the dependence of lordosis behavior on the expression of ERα in cells in VMH) (10), the full story of E2 action on behavior is subtler. The behavioral consequence of the expression of an ER gene can depend on whether it is being expressed in a male or a female, and the contribution of ERα can actually oppose that of ERβ (for summary, see Ref. 11).

Overall, accomplishments in modern neuroendocrinology can be contrasted to the history of genetics, for which the mantra, established by the Nobel laureates George Beadle and Edward Tatum in 1941, was the ‘one gene, one enzyme’ principle. The literature covering the functional consequences of ERα and ERβ, likely gene duplication products, shows clearly (12) that temporal patterns of gene expression in the brain regulate temporal patterns of behaviors related to reproduction. Temporal patterns of maternal behaviors not only represent the most complex among female reproductive behaviors but also lead to new questions about the two issues highlighted in this minireview.

Maternal behaviors

Maternal behaviors require repetitive cycles of relatively complex activities: nest building, finding and recognizing pups, retrieving pups to the nest, licking, grooming, and nursing (13, 14). The mother needs to be attracted to her pups while also reducing her fear of the pups (15, 16). The hormonal events that permit short-latency maternal behavior in rats include rising levels of placental lactogens, pituitary prolactin, and E2, superimposed on a steep drop in progesterone, all of which occur toward the end of a 22-d pregnancy (17, 18). With respect to ER involvement, E2 implanted directly into the medial preoptic area (MPOA) greatly facilitates maternal behavior in rats (19, 20), and the performance of maternal behaviors by female mice depends on expression of the ERα gene in MPOA (21). Thus, we can say that estrogens circulating during late pregnancy enter preoptic neurons that themselves are essential for maternal behaviors (19, 20), bind to ERα, and subsequently through a variety of mechanisms (14, 22) allow maternal behaviors to occur. Future questions will deal with exactly how E2 does this job. What genes are turned on by E2 in MPOA neurons whose gene products foster maternal behaviors? In turn, what chromatin changes associated with the promoters of those genes allow ligand-activated ERα to bind to estrogen response elements with the effect of increasing transcription?

Oxytocinergic systems

Oxytocinergic systems illustrate how, even though ERα and ERβ may oppose each other in many tissues (23, 24), with respect to maternal and other social behaviors they effectively work together. Oxytocin (OT) working through the OT receptor (OTR) clearly increases maternal behavior (25–28). Notably, these OT effects are dependent upon E2 priming. Research has indicated the MPOA, ventral tegmental area, and nucleus accumbens are likely sites of OT action, and it has been suggested that such OT action may facilitate the ability of MPOA neurons to activate the mesolimbic dopamine (DA) system to increase the parturient female's attraction to young (28). Also, the well-known anxiolytic effects of OT are likely to be important for blocking the disruption of maternal behavior by fear (29–32), and their dependence on E2 makes them especially interesting (33). This well developed story leads to two new questions: What are the signal transduction pathways through which the activation of OTR has these behavioral effects? And what is the relation of these OT effects to the recently detected involvement of vasopressin 1A receptor in maternal behavior control (34)?

With respect to the comparison of ERα and ERβ, E2 treatment increases the synthesis of OT mRNA in paraventricular nucleus (PVN) (35), and to do so requires ERβ (36). In contrast, E2 works through ERα to stimulate OTR gene transcription (37), and mice with a knockout mutation of ERα no longer exhibit an E2-stimulated synthesis of OTR (38). The importance of ERα-OTR interactions in the MPOA for maternal behavior is supported by Champagne et al. (39–41), who showed that individual differences exist between rats in the ability of E2 to induce increases in OT binding within MPOA and these differences are inversely related to the degree of methylation within the promoter region of the ERα gene. Importantly, increased synthesis of ERα within MPOA, and increased E2-induced OT binding, are associated with increased levels of maternal behavior.

A critical question arises: To what extent do effects attributed to E2 action on MPOA neurons actually depend in part on diffusion posterior to the PVN, or anterior to nucleus accumbens? PVN could be involved as a site of OT synthesis, and the accumbens could be involved as a target of ascending DA neurons activated upon pup presentation. Can precise new molecular pharmacological techniques for neuronal manipulation, such as microfluidics, address these technical questions?

These results with maternal behavior are similar in some respects to research on the regulation of social recognition in mice, where ERα and ERβ work together, in a four-gene micronet (42, 43). Under the influence of ERβ, OT synthesis is increased by E2 in PVN, whereas E2 bound to ERα increases transcription of the OTR in the amygdala. Question: does this set of E2 effects that facilitate social recognition also operate during maternal behavior?

Significantly, as noted below, behavioral roles for OT, which have been further elaborated on in other species, suggest important therapeutic uses. Further, there is the possibility of applying to the human condition our new knowledge (42, 43) that oxytocinergic systems are involved in the improvement of social cognition in animal models.

Multiple Hormone-Receptive Mechanisms

During the earliest days of experimentation on steroid hormone actions, until 1962, it was assumed that lipid-soluble steroids would diffuse across cell membranes and act in the cell's cytoplasm. With respect to estrogen action, leaders in the field like Claude Villée and Dwight Hagerman charted effects of estrogens on steroid metabolizing enzymes in placental cells. In 1962, the publication by Elwood Jensen and Herbert Jacobson in Recent Progress in Hormone Research, showing prolonged retention of tritiated E2 in the cell nuclei of uterine cells, coupled with the onset of molecular biological concepts that featured gene transcription, overturned the field of steroid hormone action. Thus, for almost 40 yr, nuclear hormone receptors held center stage. This, despite the fact that Martin Kelly and his colleagues reported data in 1976 (6, 44) that clearly showed rapid electrophysiological effects on E2 on neurons, effects that occurred much too fast to be explained by altered gene transcription. Thus, the results from Kelly and his colleagues (6, 44) argued against the nuclear hormone receptor zeitgeist of that time. During the past few years, in several tissues, the rapid effects of steroid sex hormones have begun to be addressed.

Along these lines, a startling result in the field of hormones and behavior was reported by Stolzenberg et al. (45). Despite a long literature emphasizing the importance of long exposures to high levels of estrogens (coupled with progesterone withdrawal) for the production of short latency maternal behavior, Stolzenberg found that a single injection of E2 at the time of pup presentation could facilitate the onset of maternal behavior. This effect was much more consistent with the idea of rapid membrane-initiated effects of E2 than with nuclear/transcriptional mechanisms. Importantly, in addition to E2 effects on MPOA which stimulate maternal behavior, it has recently been shown that DA acts on MPOA to substitute for E2 to stimulate maternal behavior in rats (46), leading to the proposal that ligand-independent activation of ERs may influence maternal behavior or that E2 and DA influence maternal behavior by affecting similar intracellular signaling pathways (28). The concept of ligand-independent activation of ERs may be relevant to the fact that while E2 is essential for the onset of maternal behavior at parturition in rats, once the behavior becomes established, the continuance or maintenance of maternal motivation during the remainder of the postpartum period is independent of hormonal, including E2, control. Important to ask is whether, when E2 itself is not directly involved in maternal behavior that has been already established, is ligand-independent activation of ERs necessary?

In molecular endocrinology, some workers have chosen to emphasize membrane effects ‘as opposed to’ nuclear effects. However, in some CNS systems studied thus far, it appears that membrane-initiated and cell nuclear/transcriptional steroid effects work together to achieve physiological regulation of neuronal activity and behavior. At the molecular level, this idea of coordinated membrane-initiated and nuclear transcriptional mechanisms has received support from transient transfection studies (reviewed in Ref. 47). At the behavioral level, nuclear and membrane-initiated mechanisms clearly work together in VMH neurons that regulate lordosis behavior. In a long series of experiments (48), it was shown that application of a membrane-impermeable form of E2 to VMH neurons rapidly enhanced long-term actions of E2 on lordosis (previously shown to be dependent on new gene expression and protein synthesis). Simple sex behaviors provide some of the best systems for discovering, in the future, exactly how nuclear and nonnuclear hormone-dependent mechanisms work together to regulate biologically important behaviors.

Perspectives Regarding Funding for Mechanisms Underlying Basic Reproductive Behaviors

Basic research analyzing sex hormone effects on the entire chain of reproductive behaviors is compelled by the following 1) the proven incisiveness of such research; 2) the commonalities in these mechanisms between laboratory animals and humans; 3) among those commonalities, the practical applicability of OT actions; 4) the disastrous effects of abnormalities in these reproductive behavior mechanisms; and 5) the contributions of this field to understanding sexually differentiated maladies of human patients.

First, leading molecular geneticists such as Sydney Brenner have, for years, proclaimed that the major challenges facing modern biologists would constitute the translation of genotype into phenotype. For answering this challenge as it applies to molecular neurobiology, the kinds of hormone-sensitive behaviors considered here have already been proven to be the fastest and most robust way to link detailed investigations of the molecular aspects of nerve cell biology to natural, complete behaviors of undoubted physiological importance. That is, steroid hormone actions on basic reproductive behaviors offer researchers some of the best phenotypic phenomena that can be analyzed, with modern techniques for the chemistry of gene transcription. Moreover, it is clear from current developments in the fields of work reviewed here that the brain can be ‘reprogrammed’ during brain development, for example by early hormone administration or by variations in maternal care (49), with mechanisms of reprogramming ranging from methylation of cytosines in the ERα gene (50) to histone modifications (51). These developments clearly put hormone/behavior relations at the forefront of modern molecular neurobiology.

Second, commonalities abound in reproductive neuroendocrinology between mechanisms in laboratory animals and those in humans. The chemistry and neuroanatomy of steroid hormone receptors are conserved into the human brain, and the nuclear receptor coregulators discovered in animal tissues are also closely related to those in human cells.

Further, understanding the molecular neurobiology of steroid hormone actions on maternal behavior may be particularly relevant to the human condition: Maternal behavior is a characteristic of all mammals and is likely to be regulated in part by evolutionarily conserved core neural circuits and mechanisms. Therefore, research on the neurobiology of maternal behavior in rodents should have broad applicability across mammals, with the strong potential to provide insights into the etiology of abnormal parenting, such as the occurrence of child abuse, child neglect, and postpartum depression (14).

Also of significance is that although all mammals show strong mother-infant bonds, the nature and degree of other aspects of sociality varies among adult mammals. For example, only about 5% of mammals are monogamous, exhibiting strong social bonds after mating (52, 53). Therefore, when sociality proves to be adaptive, the molecular neurobiology and neural circuitry underlying maternal behavior may serve as a foundation upon which other types of social bonds are built (54). With respect to humans, partial support for this proposal comes from research that shows the importance of OT neural systems for human sociality, empathy, cooperation, and altruism (55, 56).

Third, as reviewed above, research on rodents has emphasized the importance of OT for maternal behavior and the complementary roles of ERα and ERβ in regulating OTR and OT synthesis, respectively. We also suggested the possibility of ligand-independent activation of ERs, and this process may be particularly relevant to humans. In support of the proposal that the type of research reviewed here is relevant to humans, we note the following: 1) Plasma OT levels in women predict a variety of behaviors associated with maternal bonding (57–59); 2) Genetic variations in the OTR gene are associated with variations in maternal sensitivity (60); and 3) Strathern et al. (61) classified mothers as either secure or insecure. In this functional MRI study, mothers viewed their infant's faces and it was found that when compared with insecure mothers, secure mothers had greater activation of the ‘reward area’ of the ventral striatum (nucleus accumbens) and also had significantly higher OT responses to contact with their infants. Further support for the proposal that these phenomena are relevant to humans can be found in the findings that children who have experienced early neglect or abuse have lower plasma levels of OT (and vasopressin) than do control children (62).

With respect to psychiatric conditions, roles for OT in maternal behaviors and sociality are significant in at least two respects. First, OT effects have been demonstrated on affiliative behaviors in mothers with mood disorders (63). Furthermore, 2–3 wk of daily treatment with OT has been shown to decrease psychotic symptoms in schizophrenics and to improve their social cognition (64–67). In fact, every serious psychiatric disorder is associated with deficiencies in social behavior. Therefore, if we accept the widely held belief (reviewed in Ref. 54) that some of the roots of social interactions are to be found in mother-infant interactions, then the broad importance of the work reviewed here is easily understood.

Fourth, it is precisely when something goes wrong with mechanisms that regulate ‘instinctive’ behaviors, such as sex and parental behaviors, that problems result which are of considerable importance to medicine and public health. Inadequate parenting will have consequences that last the lifetime of the child, sometimes leading to criminal behavior at a great cost to society. Male-typical violence, likewise, can lead to dangerous and expensive consequences. Of even greater significance is that there is an intergenerational continuity or transmission of maternal behavior in humans and other mammals (reviewed in Refs. 14, 28, 39, 68). That is, human children who have been abused or neglected by their parents are likely to become abusive/neglectful parents themselves, and research has indicated that this intergenerational continuity is attributable, in part, to the early adverse effects of abuse and/or neglect on the offspring's brain development.

Research on rodents has shown that there is a positive relationship between the level of maternal care that a rat dam exhibits toward its offspring and the degree to which the maternal neural circuitry develops in the brains of the affected offspring (68). In comparison to offspring that received lower levels of maternal care, adult offspring that received higher levels of maternal care have more ERα expression and OTRs in MPOA and show higher levels of maternal behavior toward their own offspring, which is correlated with greater activation of the mesolimbic DA system. Some of these effects of maternal care on the offspring's brain development have been shown to involve epigenetic processes: low levels of maternal care result in increased methylation in the promoter region of the ERα gene, which decreases ERα expression in MPOA, which in turn causes a depression of E2-induced OTR expression in MPOA (40).

In particular, the kinds of molecular questions posed below in this minireview deserve special attention, because problems with nuclear receptor coregulators have been prominently associated with various disease states (69). The coregulators focused on by Kumar and O'Malley (69) comprise histone modifying enzymes whose neuranatomically specific actions have already been related to stress (70) and to sexual differentiation (71).

Finally, it seems likely that molecular explorations of hormone/behavior relations are likely to shed light on the documented sex differences in the incidence of several serious mental disorders, depression and chronic fatigue syndrome, for example, being more prevalent in women, while autism and attention deficit disorders are much more common in males.

Overall, keeping these several arguments about the broad importance of hormone/behavior relations in mind, we list some of the exciting new questions that likely will occupy neuroendocrinologists' attention during the next few years.

Challenging Questions

Molecular endocrine questions

Our understanding of ER functions has gained tremendous progress owing to recent advances in genomic technologies that enable the identification of its target gene expression at the whole-genome level (72–76). Through these studies, a large set of ERα target genes were found to be up-regulated as expected, but there are also a significant number of repressed genes uncovered in response to E2 treatment (72–76).

Therefore, it will be interesting to know whether a similar or different set of genes, compared with nonneural cells, are regulated in E2 responsive neurons to control these reproductive behaviors. More recently, by using Genome Run-On coupled deep sequencing (GRO-seq) method, nascent transcripts were isolated and identified upon short estrogen exposure using breast cancer cells. One surprising result from this study is the large number of miRNAs (so-called ‘microRNAs,’ small sequences, about 20–25 bases) are found to be regulated by E2. This finding adds additional complexity to the regulation of ERα target gene expression, as a single miRNA often regulates the expression of a large network of proteins simultaneously through affecting the mRNA stability and protein translation (77–79). Because many activities in neurons are known to be regulated through these posttranscriptional mechanisms, it will be particularly interesting to see whether this E2-regulated miRNA expression may play important roles in controlling reproductive behaviors.

ERα direct target genes have been further identified by combining chromatin immunoprecipitation (ChIP) approach with DNA microarray (ChIP-chip), deep sequencing (ChIP-seq), or paired-end tag sequencing (ChIP-PET) technologies (74–76, 80–82). It was observed, consistently, that ERα binds prevalently to distal sites that are often 10 to 20 kb away from the transcription start site and engages in long-range controls over gene expression through chromosome looping (76, 80). Further analyses revealed that these ERα binding sites are often localized adjacent to the binding site for other transcription factors (e.g., FoxA1) that are essential in directing the binding of ERα (80, 83).

As there is increased evidence supporting the cell type-specific role of these anchoring/pioneer transcription factors in directing not only ERα but also other transcription factors to specific target genes (76, 80, 83, 84), it will be particularly interesting to examine whether FoxA1 or other transcription factor(s) play this role in hypothalamic neuron important for regulating female reproductive behaviors.

The ultimate action of an ER, after binding to its target genes, is to enhance the recruitment and/or function of the general transcription machinery (85–87). In the past decade, a remarkable number of diverse transcription coregulators (coactivators and corepressors) have been discovered to play key roles in this process (86–89). These coregulators include ATP-dependent chromatin remodeling factors, histone-modifying enzymes, as well as intermediary factors that function to bring above coregulators to the target gene promoters and/or to directly interact with the general transcription machinery (86–88, 90–94). Not surprisingly, more and more studies revealed that these coregulators are often differentially expressed in different tissues and cell types and are selectively recruited to target genes' promoters (86, 87, 95, 96). Therefore, it is important to investigate which coregulators are expressed and required for ER-mediated functions in neurons that regulate reproductive behaviors. Given highly tissue-, cell-, and gene-specific in vivo roles of these coregulators (96–99), it is highly likely that new information or paradigms will emanate from these studies.

In turn, how will the emerging stories of coregulator participation in hormone/behavior mechanisms (100, 101) be translated into the detailed biochemistry of histone modifications? That is, we already know that global modifications of histones are involved in stress (70) and E2-dependent sex (102, 103) behaviors. Classically, some of the most striking molecular effects of estrogens are on the expression of specific, interesting genes such as the progesterone receptor (104, 105). Specifically, therefore, using ChIP, what would one find about changes over the progesterone receptor promoter in VMH or medial preoptic neurons? And, more generally, how are the mechanisms we are dealing with in estrogen-sensitive neurons different from estrogen effects on breast cancer cells?

Molecular neurobiological questions

In addition to questions about transcriptional mechanisms that are essentially molecular endocrine questions, two families of questions might be of particular interest to neuroendocrinologists.

First, while some experts in the biology of nuclear receptors have doubted the validity of steroid hormone action outside the cell nucleus, an increasing number of behavioral and electrophysiological effects of estrogens demand an answer to the following question: How exactly do membrane-initiated effects of E2 work together with nuclear mechanisms to produce female reproductive behaviors? Phosphorylation of ERs (106), as would increase their ability to bind to appropriate coregulators, constitutes one set of possibilities, but aren't there others?

Second, for neuroscientists interested in the goal of explaining behaviors, it has been natural that most of our attention would have focused on nerve cells. But during the past few years, it has become increasingly clear that robust hormone effects can be detected in glial cells (107–109). This is important because we now understand that glial cells do much more than provide mechanical and metabolic support for neurons, participating, for example, in the regulation of glutamine and glutamate supply and release at glutamatergic synapses. Therefore, we must ask: How do we integrate our views of mechanisms of E2 action discovered, so far, in nerve cells, with these interesting discoveries of hormone effects on glial cells?

Because this is a minireview concentrating on the chain of behaviors that characterize the female's reproductive mechanisms, there has not been space to give full scope to ER participation in brain functions far beyond reproduction. For example, estrogenic effects in the hippocampus can essentially achieve a remodeling (110), as could underlie E2 effects on learning and memory (111). E2 actions in the hippocampus also support neurogenesis (112, 113), and so the interactions of E2 with the machinery of the cell cycle must now be explored. In a different part of the forebrain, E2 effects on cells in the basal forebrain are so prominent with respect to the improvement of learning capacity (114) that they could be suggested as a potential therapeutic regimen to combat difficulties with memory during aging (115). Not only is ERα an essential participant in E2 mechanisms that protect against extensive ischemic damage (116), but also ERα and ERβ both regulate lipopolysaccharide-induced neuroinflammatory responses (117). Thus, E2, acting through ERα and/or ERβ, exerts a wide range of actions in brain tissue, some of which have importance in medicine and public health well beyond specific reproductive behaviors. Undoubtedly, other applications will be found as National Institutes of Health-funded research proceeds. For these reasons, continued investigations of E2-dependent mechanisms in neurons and glia remain of the utmost importance.