For decades, scientists have been trying to pin down the channel in ears’ hair cells that converts mechanical energy from sound waves to an electrical message in the brain. Yet for all the work numerous labs have put toward identifying this so-called mechanotransduction channel, it has remained out of reach. Scientists have now identified another piece of the puzzle, a protein called TMIE that is essential for mechanotransduction and that localizes to just the right spot in the cell. Their results were published in Neuron last week (November 20).
“This new paper locks down that TMIE is involved in mechanotransduction,” said Peter Barr-Gillespie of Oregon Health & Science University who did not participate in the study. What remains unknown, however, is whether TMIE is an accessory to the pore-forming part of the channel or actually part of that elusive conductance pathway.
The basic structure of hair cells includes a finger-like protrusion called a stereocilium. Each stereocilium is connected to its neighbor stereocilium by a “tip link,” an extracellular filament that triggers the activity of the mechanotransduction channel when the tip link moves in response to sound waves coursing through the cochlea. At one end of the tip link is the protein cadherin 23; at the other end—adjacent to the channel—is the protein protocadherin 15.
Several other proteins have been pegged to the mechanotransduction complex. Last year, for instance, Jeffrey Holt’s team at Harvard Medical School showed that TMC1 and TMC2, a pair of transmembrane proteins, are essential for mechanotransduction in mouse hair cells. Ulrich Müller of The Scripps Research Institute in La Jolla, California, and others have also shown that another protein, LHFPL5 (also known as TMHS), is a member of the mechanotransduction complex. But they, like Holt, have been unable to prove that these proteins are components of the pore of the channel, so the search has continued for other players.
In this latest study, Müller and his colleagues performed a screen in yeast to find other transmembrane proteins that interact with protocadherin 15 or LHFPL5. TMIE appeared in both screens.
“We got excited about this because TMIE was already known to be a deafness locus in humans,” Müller said. The researchers went on to show that TMIE is essential for proper mechanotransduction and that it appears to reside at the location of the channel. The authors suspect that TMIE links LHFPL5, protocadherin 15, and the transduction channel.
“I think it is highly significant,” said Tony Ricci, a mechanotransduction researcher at Stanford University who was not involved in the study. “We’re finding there are a lot more proteins up there than maybe initially were first thought.”
Whether any of these mechanotransduction-associated proteins are members of the actual channel pore, however, remains to be seen. Although some may move to the membrane in cell culture, they don’t pull together to form a conductance pathway in vitro. It could be that some proteins involved in assembling the channel are missing or, perhaps, additional subunits are still unidentified. “I think there are other pieces that have yet to be found,” Holt told The Scientist.
Part of what has plagued this field is the striking lack of material scientists have to work with. Holt pointed out that in the human cochlea there are only 16,000 hair cells, each with 50 to 100 transduction complexes. “Biochemical approaches have been quite challenging,” he said.
Müller is optimistic that the field is close to closing in on the identity of the channel. “It’s remarkable. For a long time nobody could get anything, now we have three clear components of the transduction complex.”
B. Zhao, “TMIE is an essential component of the mechanotransduction machinery of cochlear hair cells,” Neuron, doi:10.1016/j.neuron.2014.10.041.
Recent discoveries have shown that both hormonal and brain-derived oestradiol have neuroprotective effects. This Review provides a comprehensive review of the multiple cell types, receptors and signalling cascades that underlie oestradiol-mediated neuroprotection.
Abstract
Hormones regulate homeostasis by communicating through the bloodstream to the body's organs, including the brain. As homeostatic regulators of brain function, some hormones exert neuroprotective actions. This is the case for the ovarian hormone 17β-oestradiol, which signals through oestrogen receptors (ERs) that are widely distributed in the male and female brain. Recent discoveries have shown that oestradiol is not only a reproductive hormone but also a brain-derived neuroprotective factor in males and females and that ERs coordinate multiple signalling mechanisms that protect the brain from neurodegenerative diseases, affective disorders and cognitive decline.
Key points
Oestradiol is both a sex steroid hormone and a neurosteroid that is locally synthesized in the brain. Both hormonal oestradiol and brain-derived oestradiol are neuroprotective.
Oestradiol synthesis in the brain is rapidly regulated in neurons by synaptic activity. In turn, brain-derived oestradiol regulates synaptic plasticity, adult neurogenesis, reproductive behaviour, aggressive behaviour, pain processing, affect and cognition.
Under pathological conditions, the expression of aromatase, the enzyme that produces oestradiol, is enhanced in neurons and induced de novo in astrocytes as an endogenous neuroprotective mechanism. Inhibition or silencing of brain aromatase increases gliosis and neurodegeneration after brain injury.
The neuroprotective actions of oestradiol are mediated by two oestrogen receptors (ERs) located in the cell nucleus, ERα and ERβ, and by ERs located in the membrane, including ERα, ERβ, G protein-coupled ER and Gαq protein-coupled ER.
The ERs coordinate various neuroprotective signalling mechanisms, some of which are complementary and some of which are redundant. These include the regulation of transcription by nuclear ERs and the regulation of the activity of different kinases by membrane ERs.
Membrane and intracellular ERs also contribute to the interaction of oestradiol signalling with the signalling of other neuroprotective factors, such as brain-derived neurotrophic factor, insulin-like growth factor 1, WNT and Notch.
Further studies are necessary to determine the role of neuronal and non-neuronal cells in the coordination of oestradiol-mediated neuroprotective signalling mechanisms. The influence of sex and age on these mechanisms should also be studied.
A promising research direction aims to determine the role of metabolic homeostatic regulation in the protective actions of oestradiol, particularly in chronic neurodegenerative diseases.
Further research should also explore alternatives to oestradiol therapy, including new ligands for ERs and modulators of brain oestradiol synthesis.
Introduction
17β-oestradiol (oestradiol) is a steroid hormone that is synthesized in a series of consecutive enzymatic steps that begins with the conversion of cholesterol into pregnenolone in the mitochondria (Box 1). The final enzymatic step, the conversion of testosterone into oestradiol, is catalysed by the enzyme aromatase (also known as oestrogen synthase and encoded by the CYP19A1 gene; Box 1). Oestradiol, as a hormone, is mainly produced in the ovary. Its levels in the plasma change during development and puberty, fluctuate during the menstrual cycle (or the oestrous cycle in rodents) and suddenly decline with menopause. Oestradiol is also locally synthesized in different tissues, including bone, adipose and nervous tissues, in both males and females. Indeed, an important concept that has progressively developed in recent decades is the idea that the brain is a steroidogenic organ1 that expresses the molecules and enzymes necessary for the conversion of cholesterol into steroids such as progesterone, testosterone and oestradiol (Box 1). Therefore, the brain is a target both for steroids derived from peripheral steroidogenic glands and for neurosteroids (steroids synthesized in neural cells)1. Neurosteroids control various neurobiological processes, including cognition, stress, anxiety, depression, aggressiveness, body temperature, blood pressure, locomotion, feeding behaviour and sexual behaviour2. Brain steroidogenesis is regulated independently of peripheral steroidogenesis, and plasma steroid levels do not directly reflect brain steroid levels3, 4. However, after ovariectomy in female rats and orchidectomy in males, the brain adapts the levels of steroids in a sex- and region-specific manner5, suggesting a compensatory adaptation of brain steroidogenesis in response to gonadal steroid deprivation.
Box 1: Brain steroidogenesis
The gonads, the adrenal glands and the placenta are the classical steroidogenic endocrine glands in the body. However, other organs, including the brain, also have steroidogenic activity. The first step in steroidogenesis is the conversion of cholesterol to pregnenolone in the inner mitochondrial membrane (IMM, see the figure, left panel)143. Several proteins are involved in this step, including steroidogenic acute regulatory protein (StAR), translocator protein of 18 KDa (TSPO) and cytochrome P450 cholesterol side chain cleavage enzyme (P450scc). StAR and TSPO are involved in the transport of cholesterol from the outer mitochondrial membrane (OMM) to the IMM, which is a highly regulated step. In the IMM, cholesterol is transformed into pregnenolone by P450scc, the first steroidogenic enzyme. StAR, TSPO and P450scc form a multimolecular complex, termed the transduceosome, together with other proteins, including the 31-kDa voltage-dependent anion-selective channel (VDAC) in the OMM and the 35-kDa mitochondrial adenine nucleotide translocator (ANT) in the IMM143. Pregnenolone is then converted to other steroids through a series of consecutive steps (see the figure, right panel) in the endoplasmic reticulum.
The rodent brain expresses StAR, TSPO, P450scc and the steroidogenic enzymes necessary to produce oestradiol from cholesterol2. Several of these enzymes have also been identified in the human brain143. The steroids produced in the nervous system, known as neurosteroids1, act through various mechanisms, including the direct modulation of the channel properties of synaptic neurotransmitter receptors, such as the NMDA receptor and the GABAA receptor, by the progesterone metabolite 3α,5α-tetrahydroprogesterone (also known as allopregnanolone)144, 145. Some neurosteroids, such as oestradiol, regulate neural function, behaviour, cognition and affect, and have neuroprotective actions144, 146. The term neuroactive steroid is also currently used in the literature to refer to steroids with activity in the nervous system, independently of whether they are peripheral hormones, neurosteroids or synthetic steroids144.
HSD, hydroxysteroid dehydrogenase.
Oestradiol is both a peripheral hormone and a neurosteroid. Oestradiol is synthesized in the adult male and female brain, where it acts as an autocrine factor or a paracrine factor that, under physiological conditions regulates synaptic plasticity, adult neurogenesis, reproductive behaviour, aggressive behaviour, pain processing, affect and cognition (Box 2). The actions of brain-derived oestradiol contribute to the maintenance of brain homeostasis, and recent findings indicate that it is also a neuroprotective factor, decreasing neural damage in animal models of brain injury and neurodegenerative diseases. In this Review, we describe recent experimental evidence that emphasizes the role of brain oestradiol synthesis as an endogenous neuroprotective mechanism.
Box 2: Role of brain aromatase under physiological conditions
The enzyme aromatase, which converts androgens into oestrogens, is widely expressed in neurons from different brain regions of male and female animals. These include brain areas involved in reproductive control and brain regions related to memory, emotion and cognitive processing. For example, in humans, aromatase immunoreactivity has been detected in, among other brain structures, pyramidal neurons and some interneurons in the cerebral cortex and the hippocampus147. In addition, aromatase has been detected at synapses in birds and mammals148, 149, 150.
In the brains of birds and mammals, aromatase is involved in the regulation of synaptogenesis and synaptic plasticity151, 152, neurogenesis153, 154, reproductive behaviour155, aggressive behaviour155, pain processing156, 157, auditory processing158, social communication158, affect and cognition159, 160, 161.
Recent findings indicate that oestradiol synthesis in the hippocampus is regulated by gonadotropin-releasing hormone162, which may coordinate oestradiol synthesis in the brain and the ovary. Furthermore, aromatase activity, and therefore oestradiol production, is rapidly regulated in the brain by glutamatergic synapses155. The synaptic location of aromatase, together with the rapid regulation of its activity by neurotransmitters, suggests that brain-derived oestradiol acts as a rapid neuromodulator in synaptic circuits155, 163. Indeed, rapid changes in brain aromatase activity are linked to swift changes in synaptic function, neuronal activity and behaviour155, 158. Thus, local oestradiol synthesis in the brain, modulated by rapid changes in aromatase activity, seems to play an important part in the regulation of brain function and behaviour under physiological conditions.
In parallel with the advances in our knowledge of the roles of brain-derived oestradiol in neural function and neuroprotection, considerable recent progress has been made in establishing the underlying molecular mechanisms. In particular, it has been discovered that oestrogen receptors (ERs), either associated with or embedded in the plasma membrane, or located in the cytoplasm or the nucleus, coordinate multiple neuroprotective signalling cascades. These include those activated directly by ERs and those activated by the interaction of ERs with the receptors for other neuroprotective factors. Here, we also review these recent discoveries, which represent an essential advance in our understanding of the role of these receptors in the brain and have opened the possibility of using specific ER ligands to promote neuroprotection.
Hormonal oestradiol is neuroprotective
Studies in female rodents have shown that decreasing oestradiol levels in plasma by ovariectomy or other means enhances brain damage under neurodegenerative conditions6, 7, 8, 9 and decreases brain glucose metabolism and increases amyloid-β oligomers in a mouse model of Alzheimer's disease10. In turn, the replacement of oestradiol levels in the plasma of ovariectomized animals by hormonal therapy reduces damage after brain injury6 and normalizes brain glucose metabolism and decreases amyloid-β oligomers in Alzheimer's disease model mice10. Furthermore, the magnitude of brain damage after an acute neurotoxic or neurodegenerative insult is influenced by the changes in oestradiol levels in plasma during the oestrous cycle. For example, the administration of the neurotoxic amino acid kainic acid in the morning of oestrus, after the peak of oestradiol in plasma in the afternoon of pro-oestrus, results in less neuronal damage in the hippocampus than the administration of the toxin in the morning of pro-oestrus, before the oestradiol peak6.
In humans, the decline in oestradiol levels with menopause is associated with an increased risk of cognitive impairment, affective disorders and even Alzheimer's disease pathology11. This suggests that hormonal oestradiol is also neuroprotective in our species. Hormonal therapy — usually a mixture of oestrogens and progestins — has been used for many years to treat the symptoms of menopause, including hot flushes and depression. Prospective studies that have analysed the cognitive outcomes of hormonal therapy show that it enhances cognitive skills — an outcome that is indicative of neuroprotection — in postmenopausal women12, 13.
Brain oestradiol is neuroprotective
An important indication of the role of brain-derived oestradiol in neuroprotection was provided by the observation that, in rodents and birds, the brain responds to an acute injury by increasing the expression and activity of aromatase. This was first observed in male and female rodents after an excitotoxic injury in the hippocampus and after stab wound injuries in the cerebral cortex, the hippocampus and other brain regions14. Both types of brain injury resulted in increased aromatase activity in the injured tissue and de novo expression of aromatase in astrocytes, which do not constitutively express the enzyme in adult rodents. These observations were confirmed in birds, which were shown to express aromatase in radial glia and astrocytes after brain injury15, and were also extended to other forms of acute brain pathology in rodents, such as experimental stroke16, global ischaemia-reperfusion17 and intracranial pressure raising18. In addition, aromatase expression has been detected in a sexually dimorphic pattern in endothelial cells after brain injury19.
Several experimental approaches have been designed to investigate the function of increased brain aromatase expression and activity after brain injury (Fig. 1). The use of aromatase-knockout (ArKO) mice and the systemic administration of aromatase inhibitors have demonstrated that aromatase is neuroprotective7, 8, 9, 20, 21, 22. However, these models were not able to discriminate between brain aromatase and peripheral aromatase activity. More direct proof of a role for brain-derived oestradiol in neuroprotection was provided by experiments in which brain oestradiol synthesis was reduced through intracerebral administration of aromatase inhibitors. This resulted in decreased brain oestradiol levels and in increased neural damage after the application of different neurodegenerative stimuli in the brains of birds and mammals7, 23. More recently, aromatase antisense oligonucleotides have been used to block aromatase expression and oestradiol synthesis in the rodent hippocampus17. Aromatase silencing with this method enhanced the effect of global cerebral ischaemia on neuronal loss and gliosis in hippocampal area CA1. All these findings suggest that the brain enhances local oestradiol synthesis after injury as an endogenous neuroprotective mechanism.
Figure 1: Neuroprotective actions of brain aromatase
a| Under normal conditions, astrocytes in wild-type (WT) mice do not express the oestradiol-synthesizing enzyme aromatase. b| After an acute brain injury, reactive gliosis (including the conversion of surveillance microglia to reactive microglia) and neuron death occur. However, in the WT mouse brain injury also causes astrocytes to become reactive and express aromatase. Aromatase expression is also observed in neurons and endothelial cells. The increased oestradiol levels resulting from the enhanced expression and activity of aromatase protect neurons and reduce reactive gliosis, as shown on the right of the panel.c| Neuronal degeneration and reactive gliosis, with the subsequent release of pro-inflammatory mediators by astrocytes and microglia, is enhanced after the pharmacological inhibition of brain aromatase or in aromatase-knockout (ArKO) mice.
An important point to consider is that brain-derived oestradiol is neuroprotective in both male and female animals. In males, aromatase may use circulating testosterone as a precursor to generate oestradiol. Interestingly, electron microscope studies have shown aromatase immunoreactivity in astroglial processes in contact with the basal laminae of capillaries in the injured brain14. This suggests that as soon as testosterone crosses the blood–brain barrier it can be converted to oestradiol in the astrocyte end-feet. In addition, both male and female rodent brains express the enzymes for steroidogenesis and can produce testosterone from cholesterol (Box 1). Thus, the actual oestradiol levels in the injured brain will depend on the amount of circulating testosterone (predominantly in males), the amount of circulating oestradiol (predominantly in females, but with cyclical changes) and the steroidogenic activity of the brain, which may fluctuate according to different life conditions and is different in males and females24.
The importance of aromatase as a neuroprotective molecule in humans is suggested by the existence of genetic variants of the enzyme that confer an increased risk for Alzheimer's disease25, 26, 27, 28. These genetic variants of aromatase may result in decreased oestradiol synthesis in the brain, which, together with decreased serum oestradiol levels in postmenopausal women or serum testosterone levels in aged men, may increase the risk for developing neurodegenerative diseases. In this regard, it is of interest that aromatase expression is increased in astrocytes in the human prefrontal cortex in the late stages of Alzheimer's disease, a phenomenon that has been interpreted to be part of a rescue programme29.
Oestradiol-mediated neuroprotective signaling
Role of oestrogen receptors. As in other tissues, oestradiol acts in the brain by activating several complementary signalling mechanisms, the best characterized of which is the regulation of gene transcription through classical intracellular ERs (Fig. 2). Classical ERs are transcription factors that have the peculiarity of being activated by a ligand (oestradiol). At present, two main forms of classical ERs — ERα and ERβ — have been identified and cloned30, 31, 32. Both receptor forms have a similar structure, with a DNA-binding domain and a ligand-binding domain. Oestradiol binds to the ligand-binding domain and induces the activation and the homodimerization or heterodimerization of the ER. Subsequently, the ER binds to oestrogen-responsive elements (EREs) in the promoter region of specific genes through the DNA-binding domain, resulting in the recruitment of transcriptional co-activators and co-repressors33. Classical ERs may also regulate gene transcription by acting as transcriptional partners at non-ERE sites, such as activating protein 1 (AP1) sites34. In addition, classical ERs have two activation domains, which allow them to be regulated by kinases activated by the signalling pathways of several growth factors, such as insulin-like growth factor 1 (IGF1)35, 36.
Oestradiol binds to various oestrogen receptors (ERs). The classical ERs (ERα and ERβ) are transcription factors that, after oestradiol binding, form homodimers or heterodimers and regulate transcription. ERα and ERβ are also associated with the plasma membrane, where they activate different signalling cascades. Oestradiol also activates neuroprotective signalling by binding to other membrane receptors, such as G protein-coupled ER (GPER) and Gαq protein-coupled membrane ER (Gq-mER). In addition, oestradiol promotes neuroprotection by an indirect regulation of the signalling of other receptors, such as insulin-like growth factor 1 (IGF1) receptor (IGF1R), the brain-derived neurotrophic factor (BDNF) receptor TRKB, Notch — a receptor involved in cell-to-cell communication that is activated by membrane ligands in adjacent cells, such as Delta (DLL) and Jagged (JAG) — and the WNT receptor complex, which is composed of Frizzled and lipoprotein receptor-related protein (LRP). The mechanisms by which oestradiol regulates IGF1R, TRKB, Notch and WNT are shown in Fig. 4.
Classical ERs are also associated with plasma membrane lipid rafts37, where they interact with neurotransmitter and growth factor receptor signalling (Fig. 2). Thus, classical ERs are a point of convergence for the signalling of oestradiol and the signalling of other neuroprotective factors. This is an important point to keep in mind when considering the neuroprotective mechanisms of oestradiol and ERs. Finally, oestradiol may bind to membrane-associated non-classical ERs, such as G protein-coupled ER (GPER), a member of the G protein-coupled receptor superfamily, which regulates the activity of extracellular signal-regulated kinases (ERKs) and the phosphoinositide 3-kinase (PI3K) signalling pathway, also allowing the interaction with the signalling of other neuroprotective molecules38. Another membrane-associated non-classical ER is Gαq protein-coupled membrane ER (Gq-mER), which was originally identified in hypothalamic neurons, in which it modulates μ-opioid and GABA neurotransmission39.
Both classical and non-classical ERs participate in the neuroprotective actions of oestradiol. Thus, it has been shown that the neuroprotective actions of oestradiol in vivo and in vitro are imitated by selective agonists of ERα, ERβ, GPER and Gq-mER40, 41, 42, 43, 44, 45 and are blocked by ER antagonists, ER silencing and ER knockdown45, 46, 47. It is unclear whether all the ERs participate in the protective actions of the molecule in all models of neuronal degeneration. Depending on the pathological model, some receptors seem to have more relevance than others48. For example, ERα, but not ERβ, was shown to be involved in the neuroprotective action of oestradiol in a focal ischaemia model, after middle cerebral artery occlusion in mice49. However, both ERα and ERβ participate in the induction of neurogenesis by oestradiol after focal ischaemia50 and in the neuroprotective action of the hormone after global cerebral ischaemia in rodents40, 51. Also, different receptors seem to activate complementary and often redundant protective signalling pathways to produce the final neuroprotective effect (Fig. 3).
Figure 3: Redundant neuroprotective signalling elicited by oestrogen receptors.
Oestradiol enhances the expression of anti-apoptotic genes and neuroprotective growth factors and represses the expression of pro-apoptotic genes and pro-inflammatory molecules in the brain. This action of the hormone is mediated by two redundant mechanisms: the binding to the intracellular oestrogen receptors (ERs) ERα and ERβ and subsequent regulation of ER-mediated transcription, and the binding to membrane receptors (ERα, ERβ and G protein-coupled ER (GPER)) and subsequent regulation of kinase-activated transcription factors by the activation of phosphoinositide 3-kinase (PI3K)–AKT, extracellular signal-regulated kinase 1 (ERK1)–ERK2, and Janus kinase (JAK)–signal transducer and activator of transcription 3 (STAT3) signaling.
The role of ERs in the activation of neuroprotective mechanisms has led researchers to assess the neuroprotective potency of different ER ligands, such as selective ER modulators (SERMs)52, 53, 54. SERMs are clinically used synthetic ER ligands that have some advantages over oestradiol as neuroprotective molecules. The use of oestradiol as a neuroprotectant is limited by its feminizing effects and by its possible role in ER-positive cancers. Some SERMs are free of these limitations and are promising drugs for neuroprotection (Box 3).
Box 3: Selective oestrogen receptor modulators
Selective oestrogen receptor (ER) modulators (SERMs) bind to the ligand-binding domain of classical ERs and may act as ER agonists or ER antagonists, depending on the tissue. Thus, SERMs such as tamoxifen and raloxifene are used as ER antagonists for the treatment of ER-positive breast cancer, whereas raloxifene and bazedoxifene are used as ER agonists for the prevention of osteoporosis in postmenopausal women and for the treatment of menopausal symptoms, respectively164. Different SERMs confer a different ER three-dimensional structure, allowing or impeding the recruitment of transcriptional co-activators and co-repressors. This, together with the specific patterns of expression of such transcriptional co-regulators in different cell types, is the cause of the tissue-dependent agonistic or antagonistic activity of SERMs164. In addition, some SERMs, such as tamoxifen and raloxifene, are agonists of G protein-coupled ER (GPER)165.
The discovery of the role of ERs as coordinators of neuroprotective signalling mechanisms has prompted the investigation of the potential neuroprotective activity of SERMs in various experimental models of brain pathology, including cognitive decline166, affective disorders167, 168, traumatic brain and spinal cord injury169, 170, 171, subarachnoid haemorrhage172, stroke173, 174, 175, neuroinflammation176 multiple sclerosis177, 178, Alzheimer's disease179, 180 and Parkinson's disease181. These studies, together with some clinical findings182, support the possible use of SERMs as new drugs for the treatment of acute and chronic neurodegenerative diseases, affective disorders and cognitive decline.
Cell types involved. Oestradiol-mediated neuroprotective signalling mechanisms are activated in various cell types in the brain. Many in vitro studies have shown that oestradiol promotes neuronal survival in primary neurons in the absence of glial cells and that oestradiol directly activates survival mechanisms in neurons55, 56, 57. Neurons constitutively express aromatase and therefore are a source of neuroprotective oestradiol. Indeed, oestradiol produced by neurons in vitro is neuroprotective58. In addition, some studies suggest that neuronal production of oestradiol may increase under neurodegenerative conditions. For example, it is known that excitotoxic brain injury increases the expression of steroidogenic acute regulatory protein (StAR) in hippocampal neurons, suggesting increased neuronal steroidogenesis under neurodegenerative conditions59 (see Box 1 for the role of StAR in steroidogenesis). In addition, sciatic nerve chronic constriction injury increases aromatase activity and oestradiol formation in dorsal root ganglion neurons60 but not in satellite glial cells. Furthermore, increased aromatase expression has been detected in hippocampal neurons of spontaneously hypertensive rats61. Interestingly, oestradiol increases aromatase expression in hippocampal neurons of hypertensive rats61. This may represent a positive-feedback mechanism in which the production of oestradiol by neurons further enhances the expression of aromatase and the synthesis of oestradiol in this cell population. Neuronal oestradiol may act as an autocrine or paracrine factor to increase neuronal survival acting on neurons expressing ERs. In this regard, it is of interest that ERα is upregulated in neurons after brain ischaemia and that oestradiol potentiates this upregulation46. In addition, aromatase inhibition results in the downregulation of ERα and the upregulation of ERβ in hippocampal neurons62. Thus, we may envisage a mechanism by which oestradiol, generated by neurons in response to brain injury, enhances further oestradiol production in these cells and at the same time increases oestradiol-mediated neuroprotective signalling by upregulating ERα in neurons.
Although oestradiol may directly act on neurons to promote neuroprotection in vitro, the participation of other cell types is also necessary to maintain global tissue homeostasis in vivo43, 63. Thus, oestradiol acts on glial and endothelial cells to maintain the function of the neurovascular unit, regulates the inflammatory response of astrocytes and microglia to control neuroinflammation and acts on neurons, astrocytes and oligodendrocytes to maintain the function and propagating properties of neuronal circuits. As mentioned above, the main function of oestradiol as a hormonal signal is to maintain the homeostatic equilibrium of the organism, and this function of oestradiol is also reflected under pathological conditions at the tissue level.
The role of non-neuronal cells in the neuroprotective properties of oestradiol has been recently reviewed63, 64, 65 and is not examined here in detail. Our main focus in this Review is the molecular mechanisms elicited by oestradiol in the brain in vivo, in which the cell types involved have not always been identified. However, it should be noted that glial cells express ERs, including ERα, ERβ and GPER64, 66, 67, and that brain injury induces both the synthesis of oestradiol in reactive astrocytes and the expression of ERs in these cells68. This suggests that astrocytes may play an important part in the neuroprotective actions of oestradiol. Indeed, recent studies using conditional knockout mice deficient in either ERα or ERβ have shown that, in an experimental model of multiple sclerosis, the protective action of oestradiol is mediated by ERα expressed in astrocytes but not by ERα expressed in neurons or ERβ expressed in astrocytes or neurons43.
ERs in glial cells activate several neuroprotective mechanisms in response to oestradiol, including the release of factors that have trophic effects on neurons and other cell types and the control of neuroinflammation, oedema and extracellular glutamate levels. Classical ERs associated with the plasma membrane of astrocytes are involved in the oestradiol-induced release of transforming growth factor-β (TGFβ) through the activation of the PI3K–AKT signalling pathway69, 70. In addition, ERα and ERβ are involved in the anti-inflammatory actions of oestradiol on microglia and astrocytes71, 72. The anti-inflammatory action of ERα is also exerted through the activation of the PI3K pathway, which in turns blocks nuclear factor-κB (NFκB) activation and translocation to the cell nucleus73. ERβ also has an essential role in the regulation of the neuroinflammatory response of astrocytes. This effect is in part mediated by the upregulation of neuroglobin74, a haemoprotein that shares partial sequence identity with vertebrate haemoglobin and myoglobin, which protects neurons from a variety of insults, such as hypoxia, glucose deprivation, oxidative stress, β-amyloid toxicity and experimental stroke.
Main neuroprotective signalling cascades. ERα, ERβ and GPER trigger parallel neuroprotective mechanisms in the brain (Fig. 3), including the activation of extracellular signal-regulated kinase 1 (ERK1)–ERK2 (also known as MAPK3–MAPK1) and PI3K neuroprotective signalling cascades as well as the inhibition of pro-apoptotic JUN amino-terminal kinase (JNK) signalling41, 45, 75. Studies on primary cortical neurons have shown that oestradiol activates ERK1–ERK2 and PI3K neuroprotective signalling in parallel in the same neurons76. The inhibition of either ERK1–ERK2 or PI3K signalling abolishes the neuroprotective actions of oestradiol in various neurodegenerative models, including global cerebral ischaemia and experimental stroke models75, 77.
An important component of the neuroprotective mechanisms of oestradiol is the regulation of B cell lymphoma 2 (BCL-2) family members that are involved in the control of apoptosis (Fig. 3). In the brain, oestradiol upregulates the expression of anti-apoptotic BCL-2 family members, such as BCL-2 (Refs 78,79,80,81,82), BCL-XL83, 84, 85 and BCL-W86, and downregulates the expression of pro-apoptotic BCL-2 family members, such as BCL-associated death promoter (BAD)87 and BCL-2-interacting mediator of cell death (BIM)86. Both ERα and ERβ mediate the oestradiol-induced increase in the expression of BCL-2 in hippocampal neurons88. The activation of either the ERK1–ERK2 or the PI3K signalling pathways by ERs results in the decreased expression of BAD and in the upregulation of BCL-2 (Refs 89,90,91,92). ER activation also induces the transcription of BCL2 and another anti-apoptotic gene, survivin, through signal transducer and activator of transcription 3 (STAT3), a transcription factor that mediates neuroprotective actions of the hormone in experimental cerebral ischaemia93, 94, 95. In addition, some members of the BCL-2 family, such as BCL-W and BIM are regulated by oestradiol through the inhibition of JNK86. Thus, several redundant mechanisms are elicited by oestradiol to control apoptosis in the brain.
Glycogen synthase kinase 3β (GSK3β) has been proposed to be a potential target for the treatment of neurodegenerative diseases, given its role in neurodegeneration and regeneration processes96. One of the effects of the abnormal activation of GSK3β in neurodegenerative diseases is the hyperphosphorylation of the microtubule associated protein Tau, which is the main cause of the dysfunction of this protein in Alzheimer's disease and other tauopathies97. The activity of GSK3β is regulated by phosphorylation at different residues. Its activity is decreased by phosphorylation at Ser9 by AKT. The inhibition of GSK3β activity is a common mechanism of neuroprotection by several factors, such as WNT, IGF1 and oestradiol. Oestradiol, via ERα and GPER, activates PI3K in the brain and in primary neurons. PI3K, in turn, phosphorylates and activates AKT38, 98, 99, which phosphorylates and inhibits GSK3β and reduces Tau phosphorylation90, 98 (Fig. 4).
Figure 4: Oestradiol mediates indirect regulation of other neuroprotective signalling pathways.
a| Oestradiol (E2) activates insulin-like growth factor 1 (IGF1) receptor (IGF1R)-mediated neuroprotective signalling through the transcriptional upregulation of IGF1 and by inducing the interaction of oestrogen receptor-α (ERα) with the p85 catalytic subunit of phosphoinositide 3-kinase (PI3K, which is composed of p85 and p110 subunits). This allows the formation of a multimolecular complex made up of ERα, IGF1R and components of the IGF1R signalling pathway, such as insulin receptor substrate 1 (IRS1), PI3K, AKT and glycogen synthase kinase 3β (GSK3β). Through this pathway, oestradiol downregulates Tau phosphorylation and induces the stabilization of β-catenin. In addition, in the presence of oestradiol, IGF1 downregulates the transcriptional activity of ERα through the PI3K–AKT–GSK3β–β-catenin signalling pathway. b| Oestradiol activates WNT signalling by inhibiting of the expression of Dickkopf 1 (DKK1), a molecule that binds to lipoprotein receptor-related protein (LRP) and prevents the formation of the signalling complex of WNT with Frizzled and LRP. The inhibition of the expression of DKK1 by oestradiol is mediated by the attenuation of JUN amino-terminal protein kinase (JNK)–JUN signalling through a mechanism involving membrane ERs. The activation of WNT signalling by oestradiol results in the inhibition of GSK3β and the translocation of β-catenin to the cell nucleus, where it regulates transcriptional activity through T cell factor (TCF) and lymphoid enhancer-binding factor (LEF), promoting the expression of pro-survival molecules. c| Oestradiol increases the transcription of brain-derived neurotrophic factor (BDNF), which activates TRKB–PI3K-mediated neuroprotective signalling. d| Oestradiol, through a mechanism involving G protein-coupled ER (GPER) and PI3K, inhibits Notch signalling to the nucleus and promotes neuritogenesis. Note that several signalling pathways activate PI3K-mediated neuroprotective signalling. DLL, Delta; JAG, Jagged.
In addition to reducing Tau phosphorylation, the neuroprotective actions of the inhibition of GSK3β also involve the regulation of β-catenin. The levels of β-catenin are regulated by oestradiol through the ERα–PI3K–AKT–GSK3β signalling pathway98, 100. Studies in neuroblastoma cells have shown that the activation of the PI3K–AKT–GSK3β pathway induces the translocation of β-catenin to the cell nucleus, which inhibits the transcriptional activity of ERα36 and regulates transcriptional activity through T cell factor (TCF) and lymphoid enhancer-binding factor (LEF) proteins101 (Fig. 4). The set of genes regulated by oestradiol through this mechanism is similar but not identical to the set regulated by the canonical WNT signalling pathway101 (see below).
Interaction with other factors
As indicated above, the neuroprotective mechanisms of ERs also involve interactions with the protective pathways induced by other neuroprotective factors (Fig. 4).
Brain-derived neurotrophic factor. Oestradiol increases the phosphorylation of cyclic AMP-responsive element-binding (CREB) protein, the expression of its target gene brain-derived neurotrophic factor (BDNF) and the phosphorylation of the BDNF receptor TRKB (also known as NTRK2) in the CA1 region of the hippocampus and in other brain regions. This action of oestradiol is mediated, at least in part, through the ERK1–ERK2 and PI3K signalling pathways75, 82, 91, 102, 103, 104. BDNF is a neuroprotective factor in neurodegenerative and affective disorders and contributes to the homeostatic and neuroprotective actions of oestradiol in the hippocampus, cooperating with the actions of the hormone on synaptic plasticity, neurogenesis and cognition104, 105, 106, 107.
Insulin-like growth factor 1. Like oestradiol, IGF1 is a neuroprotective hormone and a local factor that contributes to maintain body and tissue homeostasis, including the regulation of brain plasticity and function108. The neuroprotective actions of oestradiol and IGF1 have many points in common, and both factors interact in the regulation of neuronal survival, neuritogenesis, adult neurogenesis, brain cholesterol homeostasis, glucose metabolism, neuroprotection and cognition82, 109, 110, 111, 112, 113, 114, 115.
Oestradiol signalling and IGF1 receptor (IGF1R) signalling interact through ERα (Fig. 4). Oestradiol induces the interaction of ERα with the p85 catalytic subunit of PI3K, probably in specific plasma membrane domains, such as the lipid rafts116. This allows the formation of a multimolecular complex composed of ERα, IGF1R and components of the IGF1R signalling pathway, such as insulin receptor substrate 1 (IRS1), PI3K, AKT and GSK3β98, 117. The interaction of ERα with IGF1R signalling allows the PI3K–AKT–GSK3β–β-catenin signalling pathway to be regulated by oestradiol90, 98. In addition, IGF1 regulates the transcriptional activity of ERα. In the presence of oestradiol, IGF1 downregulates the transcriptional activity of ERα through the PI3K–AKT–GSK3β–β-catenin signalling pathway36 (Fig. 4). By contrast, in the absence of oestradiol, IGF1 is known to enhance the transcriptional activity of ERα in neuroblastoma cells35. This suggests that the interaction of ERα and IGF1R constitutes a coincidence signal detector mechanism that allows the biological response to be adapted in response to changing levels of oestradiol and IGF1 under physiological and pathological conditions. This mechanism may be highly relevant under neurodegenerative conditions, as brain injury is associated with increased local expression of ERs and IGF1Rs, as well as increased levels of local oestradiol and IGF1 (Refs 14,40,46,68,112). In addition, plasma levels of oestradiol, the oestradiol precursors testosterone and dehydroepiandrosterone and IGF1 decrease with ageing in men and women, which may affect the function of the coincidence signal detector mechanism. This may explain why oestradiol is protective against experimental stroke in young female mice but increases brain damage after stroke in aged female mice, in which levels of IGF1 are decreased112. Indeed, oestradiol recovers its neuroprotective actions in aged female mice treated with IGF1 (Ref. 112).
WNT. As mentioned above, oestradiol activates β-catenin-mediated transcription through the PI3K–AKT–GSK3β signalling pathway. In addition, oestradiol activates β-catenin-mediated transcription through the canonical WNT signalling pathway, which is also known to be neuroprotective118, 119 (Fig. 4). The canonical WNT signalling pathway is activated when WNT binds to its co-receptors, low-density lipoprotein-related protein 5 (LRP5), LRP6 and Frizzled. The activation of canonical WNT signalling results in the stabilization of cytosolic β-catenin and its translocation to the cell nucleus, where it regulates transcriptional activity through TCF–LEF. Oestradiol upregulates WNT signalling by inhibiting the expression of Dickkopf 1 (DKK1)120, 121, a molecule that binds to LRP5–LRP6 and prevents the formation of the signalling complex of WNT with Frizzled and LRP5–LRP6. Inhibition of canonical WNT signalling by DKK1 results in the downregulation of BCL-2 and the upregulation of the pro-apoptotic molecule BCL2-associated protein X (BAX) and the hyperphosphorylation of Tau122. DKK1 expression is enhanced in neurodegenerative conditions, such as global cerebral ischaemia120, 121. Experiments have shown that oestradiol prevents the upregulation of DKK1 under these conditions and that the inhibition of DKK1 expression by oestradiol promotes neuronal survival, reduces Tau hyperphosphorylation and reduces ischaemia-induced neuronal damage120, 121. The inhibition of the expression of DKK1 by oestradiol is mediated by the attenuation of JNK–JUN signalling120, which is enhanced by global cerebral ischaemia120 and is downregulated by oestradiol through AKT123.
Notch. Notch signalling is a conserved mechanism involved in direct cell-to-cell communication. In neurons, the inhibition of Notch signalling promotes neuronal differentiation124. Notch upregulates the expression of the transcription factor Hairy and enhancer of Split 1 (HES1), which represses the transcription of the neuritogenic factor neurogenin 3 (NGN3; also known as NEUROG3)124. Recent studies have shown that oestradiol inhibits Notch signalling in primary neurons, decreasing the expression of HES1 and therefore increasing the expression of NGN3 and neuritogenesis38, 124. The oestrogenic regulation of Notch signalling and NGN3 expression is mediated by GPER and the activation of the PI3K–AKT signalling pathway38, 124 (Fig. 4). The regulation of Notch signalling in neurons may be involved in the regenerative actions of the hormone by acting to promote neuritic growth.
Conclusions and future directions
The available evidence, reviewed here, suggests an integrated view of oestradiol-mediated neuroprotective signalling. Oestradiol synthesis and ER expression are induced after brain injury, and the newly-synthesized oestradiol activates multiple neuroprotective signalling mechanisms, limiting neural damage by preventing secondary neurodegeneration23. The expression of aromatase and the synthesis of oestradiol are also enhanced in the brain under various neurodegenerative conditions. De novo induction of aromatase in reactive astrocytes has been detected in conditions of acute neurodegeneration, such as physical brain injury14, 15, excitotoxicity14, global cerebral ischaemia20 and stroke16, 22. By contrast, in the two models of chronic neurodegeneration in which brain aromatase has been examined — hypertension61 and sciatic nerve chronic constriction injury60 — the expression of the enzyme is enhanced in neurons and not in astrocytes. However, aromatase expression is increased in astrocytes in the human prefrontal cortex in the late stages of Alzheimer's disease29. Thus, at least in some cases, aromatase expression is induced in astrocytes under chronic neurodegenerative conditions. The factors that induce the expression of aromatase in neurons and astrocytes after brain injury remain to be identified125.
Neurodegenerative conditions also result in an enhanced expression of ERα in neurons and in increased expression of ERα and ERβ in astrocytes. Thus, oestradiol synthesized by brain aromatase may activate ERα-associated neuroprotective signalling pathways in neurons and activate ERα and ERβ signalling in astrocytes to modulate neuroinflammation, oedema and the release of neuroprotective factors. In addition, locally produced oestradiol may also target ERs in other cell types, such as microglia, oligodendrocytes or endothelial cells. Further studies should determine whether neurodegenerative conditions regulate other ERs that mediate neuroprotective actions of oestradiol, such as GPER or Gq-mER.
Interestingly, oestradiol exerts a positive-feedback action on the expression of brain aromatase and ERα in neurons. Thus, aromatase induction after brain injury seems to initiate a cycle that potentiates brain oestradiol synthesis and oestradiol-mediated neuroprotective signalling, reducing neurodegeneration. Furthermore, recent studies suggest that brain oestradiol synthesis not only is an endogenous mechanism of neuroprotection but is also essential for the neuroprotective action of hormonal oestradiol or oestradiol therapies. Thus, in H19-7 hippocampal cells, the inhibition of aromatase prevents the neuroprotective action of exogenous oestradiol126. In addition, the neuroprotective action of circulating oestradiol in a mouse model of Alzheimer's disease is significantly reduced in ArKO mice that have undetectable levels of brain oestradiol127. Further studies are necessary to clarify the role of brain oestradiol synthesis in the neuroprotective outcome of hormone therapy.
The studies reviewed here suggest that brain levels of oestradiol are highly critical for neuroprotection. Unfortunately, our knowledge on brain oestradiol levels — which can only be adequately measured in tissue samples — in humans is very poor. New positron emission tomography imaging techniques for the localization of aromatase in the human living brain128 may help us to better understand the role of the enzyme and that of brain-derived oestradiol in neurologic, affective and cognitive alterations in men and women.
In this Review, we emphasized the multiple and parallel neuroprotective signalling cascades that are activated by oestradiol in the brain. This neuroprotective signalling is initiated, in part, by several membrane-associated ERs and involves the regulation of the activity of different kinases that finally regulate transcriptional activity. In parallel, intracellular ERs directly or indirectly regulate gene expression, contributing to neuroprotection. There is partial redundancy in the neuroprotective actions activated by different ERs, as some of them activate similar protective mechanisms, modulating the expression of the same neuroprotective growth factors, inflammatory mediators and apoptosis regulators. In addition, both membrane-associated and intracellular ERs contribute to the interaction of oestradiol with the neuroprotective signalling of other factors, such as BDNF75, IGF1 (Ref. 36), WNT121 and Notch124. There is also partial redundancy but not complete overlap in the neuroprotective signalling that is directly activated by ERs and the neuroprotective signalling that is activated by ERs through other factors.
Although oestradiol activates multiple neuroprotective signalling pathways, several studies have shown that the inhibition of individual pathways can dramatically reduce or abolish neuroprotection. This may indicate that certain signalling pathways are more relevant than others depending on the type of neurodegenerative condition. Indeed, the signalling of some ERs seems to be more relevant than that of other ERs, depending on the pathological model and the readout examined. However, we favour the idea that the global homeostatic neuroprotective action of oestradiol may be achieved only by the coordinated action of multiple mechanisms, all of them indispensable. Thus, inhibition of an individual mechanism may compromise the final neuroprotective outcome. In this regard, an important gap in our knowledge is how these multiple mechanisms are coordinated in time and among the different brain cells. The limited evidence available indicates that both neurons and glial cells participate in the protective actions of the hormone in vivo. However, it is unclear how neuron-initiated and glia-initiated neuroprotective signalling are coordinated. The potential role of endothelial cells in the neuroprotective actions of oestradiol, in particular after stroke19, also needs further investigation.
Another major gap in our knowledge on the neuroprotective signalling mechanisms of oestradiol is the role of sex differences. Neuroprotective oestradiol signalling has been analysed in male and female animals, but a systematic approach has been rarely performed. Thus, sex is an important factor that should be considered in future studies, given the reported sex differences in the onset, manifestation and outcome of different neurodegenerative diseases and affective disorders, as well as in the response of neural tissue to oestradiol129, 130, 131.
Age is also an important factor to take into consideration in future research on neuroprotective oestradiol signalling, as ageing per se, or age after menopause, is known to affect the response of the brain to the hormone. Animal studies121, 132, 133, 134 and clinical observations11, 135, 136 suggest that hormonal therapy has a window of opportunity in the perimenopausal period during which it may reduce cognitive decline. Several studies have addressed the question on whether age or prolonged ovarian hormone deprivation affect the neuroprotective actions of oestradiol. For example, the neuroprotective action of the hormone on experimental stroke in young female mice is impaired in aged female mice, probably because the decrease in IGF1 plasma levels with ageing alters the cross-talk between ERα and IGF1R112. However, high physiological doses of oestradiol are able to protect the brain of aged rats from global cerebral ischaemia137. Further studies are therefore necessary to determine in more detail how ageing affects the neuroprotective signalling mechanisms elicited by oestradiol.
Most of the neuroprotective oestrogen signalling mechanisms reviewed here have been characterized in acute models of neurodegeneration, such as experimental stroke, global cerebral ischaemia or acute brain injury. The information obtained with these models may not be directly applicable to chronic neurodegenerative diseases, in which neuronal degeneration and death is a slow process accompanied by chronic neuroinflammation. Further studies are necessary to determine the long-term protective signalling mechanisms of oestradiol in chronic neurodegenerative diseases and to analyse the neuroreparative potency of oestradiol, which has been poorly explored. One promising research direction is to determine the role of metabolic control on the neuroprotective actions of oestradiol, as the hormone may exert sustained protective and reparative actions by maintaining metabolic homeostasis in the brain115. Acting on mitochondria, oestradiol regulates glucose metabolism, glycolysis and the tricarboxylic acid cycle-coupled oxidative phosphorylation and ATP generation in neurons115. In addition, the hormone regulates whole-body metabolism by acting on adipose tissue, the pancreas and the hypothalamus115, 138. Substantial evidence supports an influence of metabolic alterations on chronic neurodegenerative diseases and cognitive impairment. Therefore, further studies should determine whether the metabolic control exerted by oestradiol is a principal component of its long-term neuroprotective actions.
Further research should also explore alternatives to oestradiol therapy. In addition to new SERMs that may activate the multiple neuroprotective mechanisms of action of ERs, another promising therapeutic avenue for brain diseases is the development of selective neuroprotective ERα, ERβ and GPER ligands139, 140. In addition, as local oestradiol synthesis in the brain is neuroprotective, studies should be directed to understand the mechanisms that regulate aromatase expression in the brain. The human aromatase gene promoter region is highly complex, with promoters that are used with tissue specificity141. Thus, aromatase tissue-specific regulation by natural or synthetic compounds opens the possibility for selective modulation of oestrogen biosynthesis within the human brain142.
Acknowledgements
The authors thank M. Garcia-Diaz, Stony Brook University School of Medicine, New York, USA, for critical reading of the manuscript. The authors acknowledge support from Ministerio de Economía y Competitividad, Spain (BFU2011-30217-C03-01).
Maria-Angeles Arevalo obtained her Ph.D. in chemistry from the Autonomous University of Madrid, Spain. She worked as a postdoctoral fellow at the Centro de Investigaciones Biológicas, Consejo Superior de Investigaciones Científicas (CSIC), Madrid. She is a staff scientist at the Cajal Institute in Madrid, where she co-chairs the Laboratory of Neuroactive Steroids. Her research focuses on the molecular mechanisms mediating the actions of oestradiol on neurons and glial cells. Maria-Angeles Arevalo's homepage.
Iñigo Azcoitia
Iñigo Azcoitia obtained his Ph.D. in biology from the Complutense University of Madrid. He is a professor in the Department of Cell Biology at the Complutense University School of Biology. His research focuses on the neuroprotective mechanisms of sex steroids and on their role in the regulation of adult neurogenesis.
Luis M. Garcia-Segura
Luis M. Garcia-Segura obtained his Ph.D. in biology from the Complutense University of Madrid. He worked at Geneva University School of Medicine, Switzerland, Yale University School of Medicine, New Haven, Connecticut, USA, and The Babraham Institute, Cambridge, UK. He is a research professor at the Cajal Institute, where he co-chairs the Laboratory of Neuroactive Steroids. His research focuses on the neuroprotective mechanisms of neuroactive steroids. Luis M. Garcia-Segura's homepage.
