Estrogens, essentially the three major forms: estrone (E1), estradiol (E2) and estriol (E3), are a group of hormones that are necessary for the development of female characteristics and reproduction (1–4). Estetrol (E4) is also an estrogenic steroid that is exclusively synthesized in the fetal liver during human pregnancy, yet remains with unknown function (5, 6). However, long gone are the days when these hormones were solely considered as “reproductive hormones” since a wealth of data acknowledge estrogens, as well as other reproductive hormones, as essential players in nervous system development and function (7–13). Once secreted, estrogens can be delivered from the periphery into the nervous system via the blood stream. Estrogens can also be synthesized locally within the nervous system and target adjacent cells through paracrine activity, or synthetized and signal within the same cells through autocrine activity (14). The very early exposure of vertebrate embryos to estrogens underscores their fundamental role during development. Indeed, the mammalian embryo grows in a rich estrogenic environment, and estrogen is later provided to embryos maternally through the placenta (15). It is also delivered in the egg yolk of oviparous vertebrates (16). Estradiol, being the major female sex hormone and most effective of the three major estrogens, has been the focus of most estrogenic pathway studies in animals and humans.

In all cases, estrogens mainly exert their function via interaction with specific receptors, called estrogen receptors (ERs). Estrogens mediate their function via classical ERs, or membrane-associated ERs. ERα and ERβ are responsible for genomic estrogen effects, whereby estrogens bind to the ER in the cytoplasm which then dimerizes and translocates to the nucleus, to finally interact with estrogen responsive element (ERE) DNA sequences found in target genes (3). This classical hormone action is defined as slow response mechanism, considering that ERs must shuttle between cytoplasm and nucleus to exert their transcriptional function. However, other studies have reported a very rapid increase in cAMP in response to E2, highlighting a possible interaction with the adenyl cyclase machinery, thus a non-genomic action. This fast non-genomic estrogen activity could be attributed to a specific membrane initiated steroid signal (MISS) on the ER, that allows the latter to translocate to the membrane following posttranslational modifications (3). On the other hand, it was only recently that a 7-transmembrane G protein coupled receptor, GPR30 or GPER (G protein-coupled estrogen receptor), was proposed as a novel non-classical ER that would mediate estrogen rapid signaling (3, 17–21).

The zebrafish is a fantastic vertebrate model to follow highly dynamic activities of neural cells and their interaction with neighboring cells. Its external development makes it an ideal model for genetic manipulation as early as the one-cell stage and provides a vertebrate model for drug screening and signaling analysis. Their ability to absorb drug compounds enables testing of hundreds of molecules in a relatively short time. Furthermore, zebrafish larvae remain transparent throughout the first weeks of development, which enables careful imaging of live cellular and intracellular events, at a level of detail unfeasible in any other vertebrate organism (22–26). Even though zebrafish generation time is similar to rodents, they develop relatively fast when compared to other vertebrate models. Most importantly, they share conserved molecular mechanisms with other organisms, including regulation of neural development (27).

In this review, we will highlight recent findings from zebrafish and rodents that report nuclear and non-genomic activities of ER signaling in embryonic nervous system with a focus on neural development.

Even though some hormones vary between humans and animals in their spatial and temporal expression, it is important to note that so far, every animal organism has contributed to our understanding of hormonal function, sometimes with astonishing and unexpected outcomes.

Regarding estrogens, scientists have made a great progress in understanding ligand/receptor interactions, their downstream effectors and contribution to physiological functions. Moreover, additional progress is expected in the coming years to dissect estrogens, and more specifically ER signaling in neural circuit formation and interaction.

Estrogen receptors are part of the so-called nuclear receptors, known for their transcriptional activity by binding to specific response elements. These receptors present a conserved functional domain organization, with four to five shared domains. Among these are i) the N terminal domain that contains the first of two transactivation domains, and is highly variable; ii) the C domain, which contains the highly conserved DNA-binding domain (DBD); and iii) the E domain, which contains the ligand-binding domain (LBD) and the second transactivation domain, that is also well-conserved and responsible for dimerization (28). Indeed, as mentioned above, there are two types of ERs in rodents, ERα and ERβ. Mouse ERα amino acid sequence shares an overall homology of 88.6% and 97.3% with human and rat ERα sequences respectively, while human ERβ shares 89% identity with rat ERβ and 88% with mouse ERβ (29–31). Moreover, rat ERβ shares more than 95% homology in the DBD domain, and 55% amino acid identity in ligand-binding domain with rat ERα (32). Similar findings were observed in mice ERs, whereby the DBD domain presents a high degree of conversation between the two subtypes (96%) (33). Furthermore, whilst ERα and β can form homodimers of either subtype and interact with their response elements, the two ER subtypes are also able to form DNA-binding heterodimers and potentially diversify estrogen signaling pathways (33).

Two types of estrogen receptors are found in zebrafish, Erα and Erβ, encoded by three distinct genes: erα or esr1, erβ1 or esr2b and erβ2 or esr2a; erβ being duplicated. Initial sequence analysis indicated that zebrafish Erα shares 47.1% identity with human ERα, while Erβ1 and β2 had 46.8% and 51.5% identity, respectively, with human ERβ (34, 35). The characterization of these receptors showed Kd values of 0.74 nM for Esr1, 0.75 nM for Esr2a and 0.42 nM for Esr2b (36). Moreover, all ERs were able to induce a reporter gene activity with an ERE that is estrogen dependent. A link between estrogen activity and estrogen responsive element has also been established through a transcriptomic study. This revealed that estrogens stimulate metabolic pathways during zebrafish development, that liver, pancreas and brain are the most responsive organs to estrogen treatment and that estrogen effects on zebrafish development are stage-specific (37).

Apart from the well characterized estrogen nuclear receptors, it has been shown that a G protein coupled receptor, GPR30 or GPER, is activated by E2 at the cell membrane (18). Weigel and colleagues originally isolated and cloned GPR30 from an estrogen receptor (ER)-positive carcinoma cell line (38). They mapped it to chromosome 7p22 and showed that its transcript encodes a 375 amino acid protein. Using SKBR-3 cells, Dong and colleagues found that estrogen binds to GPR30 with a Kd of 2.7 nmol/l (39).

In 2009, Liu and colleagues cloned a full-length cDNA homologous to the GPER of rodents from the testis of zebrafish. It is located on chromosome 3, contains three exons while human ortholog has two; its protein sequence shares 71.5% identity with human GPER (40) (updated sequence analyses are found in Table 1 , Figure 1 ). Using gper-transfected Cos-7 cell line, they revealed the presence of E2-binding sites in GPER, with a Kd of 2.3 nM.

The nervous system is a heterogeneous structure of different cell types that originally derive from neural stem cells (NSCs), to give rise to neurons and glia. The terms neurogenesis and gliogenesis are used to define the spatially and temporally controlled transformation of NSCs into differentiated neurons and glia, respectively (41). Thus, the incredible diversity of neurons and glia in the nervous system, results from the tight and fine balance between proliferation and differentiation of neural progenitor cells. This is achieved through the coordination of a multitude of signals, combining extrinsic cues with intrinsic signaling pathways, that are both well defined in time and space. Accordingly, any alteration to the diversity and numbers of neurons or glia, will systematically lead to defects in either brain size, such as microcephaly and macrocephaly, or function, through defective wiring or neural network activity (42).

Estrogen receptors are both widely expressed in the developing fetal rodent brain, from as early as E16.5 for ERα, and E10.5 for ERβ. Erα is more localized to the hypothalamus after birth, while Erβ expression remains more dispersed and found in several areas of the brain and within different cell types, including serotonergic neurons, interneurons, microglia and oligodendrocytes (12).

Zebrafish erα is expressed, through different isoforms, from very early stages of development, highlighting a maternal contribution (43). Zygotic expression is also evident, with high levels of expression observed until 96 hours post fertilization (hpf) (latest to be analyzed). The expression of erβ2 is very low during early stages, but then progressively increases following zygotic transcription. erβ1 is highly expressed at early stages, drops down and then increases between 24 and 48 hpf (44). However, the highlighted results from qPCR experiments do not correlate with whole mount in situ hybridization, since no expression of the three different er mRNAs was observed at early stages. esr1 expression was only detected in the liver at 48 hpf and at 14 days post fertilization (dpf) in the forebrain. esr2a and b expression is visible at 32 hpf in the forebrain, followed by an expression in the hypothalamus at 48 hpf (34, 44). Thus, the precise spatiotemporal developmental expression of these receptors is yet to be clarified. Work form Olivier Kah’s group shows that estrogens stimulate the expression of aromatase B, a key enzyme responsible for converting androgens to estrogens, in the presence of estrogen receptors, with a higher activity in the presence of Esr2b and a (44). It is possible that fish aromatase is highly expressed in brain regions where ER are strongly expressed too, and that Esr2a and b might be responsible for aromatase expression in radial glial cells in zebrafish brain. Whether there is a direct correlation between the expression and function of ER and aromatase, is yet to be demonstrated, since no functional genetic studies have addressed this issue so far.

GPER expression is mainly studied in adult brain, showing an expression in multiple areas of the central and peripheral nervous systems, including the hypothalamus, spinal cord and dorsal root ganglia (17, 45, 46). Zebrafish gper, on the other hand, was found to be expressed at very early stages, and is widely distributed in different regions of the developing brain, as early as 18 hpf (47, 48).

Most studies have focused on estrogens or molecules and compounds with estrogenic activity as important players in neuroprotection under pathological conditions. Estrogens, indeed, promote neuronal cell survival by increasing the expression of growth factors and/or anti-apoptotic molecules (49, 50). This estrogen activity might also be related to their capacity to modulate dendritic spines, axonal growth, synaptic signaling and plasticity (12, 51–57).

As mentioned above, ERs are widely expressed in the nervous system, however, only a small number of studies have analyzed the impact of estrogen receptor genetic invalidation on neural development in vivo. The majority of studies have used selective ER agonists or antagonists to study the role of ERs in biological processes and to demonstrate receptor specificity. To evaluate the effects of estrogen on ERα and ERβ, some have utilized the ERα and ERβ antagonist ICI 182,780 in combination with estrogen (44, 58). However, while ICI 182,780 is an antagonist of ERα and β, it has also been shown to act as an agonist of GPER (59). This suggests that some of the positive effects of estrogen may be mediated by GPER. Researchers have also used selective estrogen receptor modulators (SERMs). SERMs are ER ligands that exhibit preferential binding affinity towards one receptor isotype over the other, and can help clarify the specific contributions of each receptor subtype to the biological effects of estrogen (60). Thus, gene invalidation of each of the ERs remains a good strategy to assess their role(s) in neural development in vivo.

Neurogenesis takes place in the two proliferative regions of the mammalian brain, the subventricular zone (SVZ), and the subgranular zone (SGZ) of the dentate gyrus in the hippocampus, where NSCs are abundant (61). Interestingly, both ERα and ERβ, as well as GPER are all expressed in NSCs of rat embryos, highlighting a potential role for these receptors in the behavior of NSCs (8, 62). A wealth of studies shows an important role for estrogens in the proliferation of NSCs. Treating NSCs with E2 enhances the proliferative activity of NSCs, either using human NSCs, or primary cultures of embryonic rat derived NSCs (63). E2 activity, in this case, seems to be predominantly mediated by ERβ (8). Indeed, ERβ ^-/- mouse brains show a significant decrease in the number of neurons in the cortex, and their brain is smaller than those of controls (64). Studies from Gustafsson’s lab propose a role for ERβ in neuronal migration and preventing apoptosis during development (65). Using mouse embryonic stem cells (mESCs), studies from the same group found that proliferation was higher and neurogenesis reduced in ERβ KO mESCs. Data provide evidence that ERβ plays an important role in maintaining stem cell identity by curbing proliferation, and possibly favoring nonneuronal fate (13, 66). It remains hard to reconciliate all these data given: i) the important role of E2 in enhancing proliferation of NSCs and stimulating neuronal differentiation in vivo and in vitro; ii) the smaller brain in ERβ ^-/- mice and increased levels of apoptotic neuronal death, while ERβ is shown to mediate apoptosis in neuronal cells; iii) high proliferation in NPCs derived from ERβ KO mice, with no significant difference in apoptosis between controls and KO mice, and no changes in the expression of neuronal markers. Few studies have addressed the role of ERα in neural development, however, some data provide evidence of an important role for ERα in mediating the differentiating and neuroprotective effects of estrogens in vitro, in PC12 cells, with a focus on neurite outgrowth (67).

The general consensus, even though results might depend on the timing and location of estrogen activity, is that estrogens stimulate the proliferation of neural stem cells in rodents. Thus, one of the striking differences between zebrafish and rodent studies, is the inhibitory effect of estrogens on cell proliferation in the brain of adult zebrafish, as well as the strong expression of aromatase in radial glial cells (RGs). Using ICI 182,780 as inhibitor of ERs activity (although presenting GPR30 agonist properties), Olivier Kah’s group showed a significant increase in the number of PCNA positive cells in different areas of adult zebrafish brain. Moreover, 17β-estradiol treatment led to a significant decrease in PCNA positive cells, suggesting a role for estrogens in inhibiting cell proliferation through their nuclear receptors, at least partially, in adult zebrafish brain (44, 68–70). As for embryonic studies, treating zebrafish embryos with E2, during nervous system development, decreased the number of BrdU positive cells in the thalamus, olfactory bulbs, telencephalon and preoptic areas, while no difference was observed in mediobasal and caudal hypothalamus (71). Even though some of the areas affected differ between zebrafish adults and larvae, a clear inhibitory effect of estradiol on proliferative activity of neural cells is observed in zebrafish (69, 71). Several studies highlighted a potential role for aromatase in RG development, given its high expression in RGs. However, there is no evidence so far of a role of aromatase in the behavior of RGs, or in neurogenesis per se.

While most behavioral studies focused on GPER-selective agonists and antagonists to study the role of GPER in mice behavior (56, 72–74), only few studies assessed its direct role in anxiety and stress responses using GPER KO mice and GPER-deficient rats (75, 76). A potential role for GPER in neural development is yet to be revealed in rodents. A recent study by Pemberton and colleagues has shown, although limited to selective agonist G-1 and E2, a role for GPER activation in neural growth, neural firing activity and intracellular Ca²⁺ rise in primarily cultured E18 rat embryonic neurons (77).

Zebrafish studies have brought more insight into gper function during development. Using a morpholino knockdown approach, Lin H. and colleagues revealed an increase in apoptosis and a significant decrease in the expression of some neuronal markers in gper morphant embryos, such as Zn-12, Znp-1 and Zn-5 (48). Additional studies, including gper KO mutant, are needed for more accurate analysis of gper activity in neural development. The first functional analysis of gper function during development, using a gper KO mutant, is led by Romano and colleagues, in which they show a fundamental role of Gper, centrally, in regulating zebrafish embryonic heart rate, by modulating estrogen and T3 levels in the developing brain (47).

As mentioned above, estrogens can regulate several aspects of neural development, however, it was not clear until recently how ERs might contribute to neural development. As the Notch pathway is critical to neurogenesis, it was reasonable to consider an interaction between ER and Notch signaling. Both neurogenic genes (notch, delta), as well as proneural genes (neurogenin, neuroD), are required for neurogenesis (i.e. neuronal cell fate) in zebrafish and mice. While notch is expressed in proliferative neural stem and progenitor cells regions, neuroD and neurogenin are expressed in postmitotic neurons (78). Thus, the molecular mechanisms driving neuronal development in zebrafish require a similar regulatory cascade to rodents. Moreover, zebrafish present a unique opportunity to analyze the development, behavior and function of not only neurons, but also major glial cell types in the nervous system, from radial glial cells, oligodendrocyte precursor cells, oligodendrocytes, Schwann cells, microglia and the recently identified astrocytes (79–85).

Given the complexity of the nervous system and the diversity of its population during development in vivo, a recent study led by Gustafsson’s group tried to address ER and Notch interactions using embryonic stem cells derived from controls and ERβ KO mice. Using a targeted gene-expression profiling in combination with pluripotency markers, the authors provide evidence of reduced neurogenesis and enhanced oligodendrogliogenesis in ERβ KO stem cells, although, there was no significant difference in the expression of neuronal markers. This correlated with higher proliferation, and no measurable differences in apoptosis. Authors also show a sharp decrease (75%) in the expression of Hes3 transcript in ERβ KO stem cells (66). Indeed, Notch-Hes signaling is a major driver of neural stem cell renewable since it prevents premature differentiation through Notch-Delta lateral inhibition. Hes genes are found highly expressed in neural stem cells and are considered as repressors of neural differentiation. Hence, reducing Hes levels leads to a significant increase in proneural genes’ activity, a premature neurogenesis, as well as rapid depletion of the stem cell pool. This is the first clear demonstration of a role of ERβ in the transcriptional activity of a major signaling player, Notch-Hes, in neurogenesis. This highly defined cell culture system is hence a powerful in vitro tool to assess gene-expression profiling. However, the picture is far from clear when it comes to ER activity during nervous system development in vivo, where intercellular communication between the different players, and varied extrinsic peripheral and local signaling is established. This added to the complexity of Notch-Hes oscillating activities that drive either proliferation or differentiation via other oscillating partners, makes it hard to define a clear role for ER in neurogenesis/gliogenesis (86, 87). Overall, in vitro studies have so far established ERβ as a major modulator of Notch-Hes activity in neural stem cells.

Other studies have linked estradiol to dendritogenesis and Notch. Estradiol, by the inhibition of Notch signaling, increases the expression of the proneural gene neurogenin 3, and regulates neuritogenesis in developing hippocampal neurons; a mechanism that involves, at least partially, GPER (51). Collectively, data point to a major role of estradiol in mediating several aspects of neural development by modulating Notch signaling, and involving classical ERs, as well as GPER.

With regard to estrogen receptor activity in zebrafish, it has been shown that Esr are fully functional during development. esrβ2 is shown to regulate the development of sensory hair cells within neuromasts, part of the lateral line organ that mediates directional water movements, prey capture and predator avoidance. The number of sensory hair cells was significantly reduced in esrβ2 morphants, while supporting cells were present. It is important to note that lateral inhibition is the main mechanism driving zebrafish neuromast differentiation, by imposing a binary fate between hair and supporting cells. Nascent hair cells, expressing Delta protein, inhibit their neighboring cells from adopting hair cell fate, forcing them to become supporting cells; notch1a and notch3 appear to be upregulated in esrβ2 morphants. Two of notch ligands, deltaA and deltaB were also upregulated, a mechanism that might explain, at least partially, the suppression of hair cell differentiation (88). On the other hand, it has been shown that esr1 is required for cell migration within zebrafish posterior lateral line primordium, by repressing chemokine receptor CXCR4 (89). Whether CXCR4 and Notch interact in this particular context is still to be investigated. Moreover, it would be interesting to assess whether this defect is observed in esr1^-/- mutants.

While in vitro studies in rodents established a strong link between ERβ and Notch signaling, there remain many open questions: Do ERs contribute to generating the cell diversity within the nervous system of zebrafish? Do they interact with Notch signaling in vivo?

Estrogen receptors are expressed in both OPCs and oligodendrocytes (OLs), in vitro and in vivo, suggesting that estrogen signaling may play a role in regulating the proliferation, differentiation, and survival of these cells. Indeed, studies have shown that estrogen treatment can increase the number of oligodendrocytes and myelin production, in vitro and in vivo (90). In particular, estrogen receptors have been shown to play a role in promoting the differentiation of OPCs into mature oligodendrocytes. Estradiol–ER axis was found to activate the pAkt/mTOR pathway in oligodendrocytes, a pathway known to regulate and promote oligodendrocyte differentiation (72). Studies have suggested that ERα signaling may be particularly important for promoting oligodendrocyte differentiation, while ERβ signaling may be more involved in promoting oligodendrocyte survival and myelin maintenance.

Additionally, well-established animal models of demyelination have shown a prominent role of these nuclear hormone receptors in myelination, by promoting oligodendrocyte maturation and development. It has been suggested that estrogen signaling may have a protective effect on myelin and oligodendrocytes in various conditions that involve demyelination or damage to oligodendrocytes, such as multiple sclerosis. Studies have shown that estrogen treatment can improve myelin repair and reduce inflammation and demyelination in animal models of multiple sclerosis. Mice lacking ERβ in oligodendrocytes are more prone to myelin damage than WT mice in the experimental autoimmune encephalitis model of multiple sclerosis (91). Nevertheless, it was found that ERs are not necessary for SERMs to exhibit their potent effects on OPC differentiation and remyelination in vivo (92).

Comparative analysis of the transcriptome in the cortex of ERβ knockout male mice (BERKO) and wild type (WT), revealed upregulation of myelin genes in BERKO mice. Qualitative analysis further demonstrated disrupted layering in the motor cortex of BERKO mice, as evidenced by staining for myelin basic protein (MBP). Transmission electron microscopy (TEM) confirmed a significant increase in axonal myelination thickness in the KO cortex, which was surprising. However, it is possible that loss of ERβ promotes oligodendrogliogenesis, but impairs OL functionality (93). Interestingly, microarray data revealed a significant upregulation of oligodendrocyte-specific factors, including Omg (oligodendrocyte-myelin glycoprotein), and the oligodendrocyte fate-specific transcription factor Olig2 (oligodendrocyte transcription factor 2), in BERKO cultures. Overall, findings suggest that loss of ERβ may enhance oligodendrocyte differentiation and proliferation, possibly through the dysregulation of oligodendrocyte-specific genes (66). Whether ERs have distinct functions during the different stages of OL development, in vivo, remains to be clarified.

GPER is expressed in oligodendrocytes within the rat spinal cord and corpus callosum (94). It is also detected throughout the different stages of oligodendrocyte differentiation and promyelinating stages in primary oligodendrocyte cultures. Thus, GPER may play a role in oligodendrocyte development, a function that is yet to be studied.

A recent example of the role of estrogens in the development of zebrafish nervous system comes from Charles Tyler’s lab. In this nicely executed work, authors reveal a new function of estrogens during early brain development. They identify novel estrogen responsive cells, EROB, that play an important role in the development and function of the olfactory sensory system, at least by modulating the intrinsic neuronal activity in the olfactory bulb of developing zebrafish (95). Although, a precise mechanism of estrogen activity within this newly identified glia is still missing, this work identifies a fundamental role of estrogens in the development of the olfactory sensory system. Interestingly, alteration in estrogen activity has also been linked to several neurodevelopmental disorders (96), and estrogenic compounds were able to rescue the nighttime hyperactivity phenotype observed in zebrafish mutant embryos of contactin associated protein-like 2 (cntnap2), an autism-related gene (97). This result might be relevant to understanding the significantly high prevalence of Autism Spectrum Disorder (ASD) in Preterm Infants (98). Indeed, the human fetus is exposed to different levels of estrogens that reach their highest peak during the third trimester, a period characterized with maturation and rapid growth of the brain (99). It is possible that this high prevalence of ASD is related to the reduced hormonal activity, including from estrogens, that preterm infants experience during their development. Studies have shown that increased high risk of ASD is directly linked to loss of placental hormones, particularly in males (100). Even though zebrafish development is substantially different to mammals, notably in the absence of a placenta, it is quite remarkable to observe such a conserved link between hormones, such as estrogens, and autism being established during embryogenesis.

Overall, these studies identify estrogens as modifiers of developmental neural circuits with profound impact on adult behavior. The question remains whether these estrogen related activities signal through ERs.

Neural development describes the process by which neural progenitor cells proliferate, self-renew and generate differentiated cell types in the nervous system, including neurons, oligodendrocytes and astrocytes in a timely manner. Although a wealth of studies provides evidence of a direct role for estrogens in this process, we are just starting to understand the underlying molecular and cellular mechanisms, and the role of different ERs in generating the diversity of neuron/glial cells. Zebrafish offers a unique opportunity to study embryological decisions including neural lineage, the timing of maturation (cells acquiring a certain fate), and the molecular mechanisms of fate decisions. in vivo live imaging makes it possible to track individual cells as they divide and differentiate, as well as analyze symmetric and asymmetric divisions that generate differentiating neurons and glial lineages, while renewing the population of progenitor cells. This imaging capability, combined with recently available ER mutants and transgenics that mirror gene expression in vivo (e.g. oligos, astrocytes, ERE activity, aromatase activity, notch sensors…) ( Table 2 ), will allow us to dissect the role of the different ERs in embryonic neural development and circuit formation. Furthermore, zebrafish mutants will be helpful to address the possible redundancy between the different nuclear and membranous ERs in neural development, an important feature of ER activity that is yet to be tested in vivo.

M-JB, OE-H, CD and MT: writing-original draft. M-JB, CD and MT: Figure and tables. All authors contributed to the article and approved the submitted version.