Abstract
In mammals, 17β-estradiol (E2), the primary endogenous estrogen, maintains normal central nervous system (CNS) function throughout life and influences brain responses to injury and disease. Estradiol-induced cellular changes are mediated through the activation of nuclear and extranuclear estrogen receptors (ERs), which include ERα, ERβ, and the G-protein coupled receptor, GPER1. ERs are widely expressed throughout the brain, acting as transcriptional effectors or rapidly initiating membrane and cytoplasmic signaling cascades in practically all brain cells including microglia, the resident immune cells of the CNS. Activation of ERs by E2 exerts potent anti-inflammatory effects through mechanisms involving the modification of microglial cell responses to acute or chronic brain injury. Recent studies suggest that microglial maturation is influenced by the internal gonadal hormone milieu and that their functions in the normal and diseased brain are sex specific. However, the role that each ER subtype plays in microglial development or in determining microglial function as the primary cellular defense mechanism against pathogens and injury remains unclear. This is partly due to the fact that most studies investigating the mechanisms by which E2-ER signaling modifies microglial cellular phenotypes have been restricted to one sex or age. This review examines the different in vivo and in vitro models used to study the direct and indirect regulation of microglial cell function by E2 through ERs. Ischemic stroke will be used as an example of a neurologic disease in which activation of ERs shapes microglial phenotypes in response to injury in a sex- and age-specific fashion.
Significance Statement As the primary immune sensors of central nervous system damage, microglia are important potential therapeutic targets. Estrogen receptor signaling modulates microglial responses to brain injury and disease in a sex- and age-specific fashion. Hence, investigating the molecular mechanisms by which estrogen receptors regulate and shape microglial functions throughout life may result in novel and effective therapeutic opportunities that are tailored for each sex and age.
Introduction
Microglial cells are responsible for the innate immune defense of the central nervous system (CNS). Upon sensing foreign stimuli, they undergo phenotypic changes that allow them to act as macrophages, phagocytizing neurotoxins, dying cells, and debris, and to initiate the neuroinflammatory signaling that protects and repairs damage (Salter and Stevens, 2017). Beyond surveillance, microglia are also key partners of neurons in maintaining normal CNS homeostasis. This is exemplified by their role in the activity-dependent refinement of neural circuits via synaptic pruning in both the developing and adult CNS (Paolicelli et al., 2011; Ji et al., 2013). The exquisite sensitivity of microglia to changes in the internal and external environments allows them to make essential contributions to the normal expression of a growing list of CNS functions, including learning and memory (Morris et al., 2013), motor function (Kana et al., 2019), and social (Smith and Bilbo, 2019) and sex behaviors (Lenz et al., 2013). Recent findings showed that microglia adapt to the changing demands of the developing and adult CNS in a sex- and age-specific fashion (Hanamsagar and Bilbo, 2016). Hence, impairment of microglial function will have different impacts on brain responses to injury and on the risk and manifestation of psychiatric and neurologic disorders in men versus women. In this regard, microglial responses to traumatic brain injury (TBI), ischemic stroke, and nociceptive stimuli, as well as the phenotypic changes they undergo during the course of neurologic disorders such as Alzheimer’s disease (AD) and Parkinson’s disease (PD), are age and sex specific (Hanamsagar and Bilbo, 2016; Kerr et al., 2019). The gonadal steroid hormone estrogen is a major signal responsible for shaping developmental and adult sex-specific brain function in a manner that involves modulation of microglial phenotype.
The brain is a target of peripheral and locally produced 17β-estradiol (E2), the primary endogenous estrogen. E2 plays a key role in shaping brain architecture during development and is the principal driver of sexual differentiation in the rodent brain (McCarthy et al., 2018). Testicular production of androgens during embryogenesis (embryonic day; E 16–18) and shortly after birth demarks critical time windows when the conversion of testosterone to E2 by brain aromatase allows the latter to exert permanent organizational effects. In contrast, embryonic and neonatal ovaries are largely inactive, and a lack of exposure to E2 is thought to drive normal female brain development. Hence, male and female brains have very different internal hormonal states during neurogenesis, neuronal migration, and synaptogenesis and importantly, during microglial cell maturation. Starting at E 8.5, yolk sac progenitor microglial cells travel through the bloodstream to the brain until the blood-brain barrier (BBB) is formed. Microglia then follow a gradual and precise developmental program in which colonization and proliferation are accompanied by time-specific changes in their morphology (from amoeboid in the embryonic brain to ramified in the adult brain) and in the ability to regulate or support specific processes such as axon guidance, synaptic patterning, cell genesis, phagocytosis of dead cells, and myelination (Lenz and Nelson, 2018). Thus, beginning in utero, and then postnatally, E2 has the potential to shape microglial development, which might contribute to permanent sex-specific manifestations of normal or impaired brain functions.
E2 also modulates adult microglial phenotypes. The best documented example of this is attenuation of microglial proinflammatory responses to brain injury (Johann and Beyer, 2013; Azcoitia et al., 2019). Although microglia play a vital role in protecting neurons, their prolonged hyperactivation results in the production of reactive oxygen species, cytokines, and proteases that kill neighboring cells and damage the BBB (Nissen, 2017). These proinflammatory responses are stronger in the aging brain, where microglia are already primed into a more reactive, proinflammatory phenotype (Lourbopoulos et al., 2015; Mangold et al., 2017a,b; Zoller et al., 2018). An activated proinflammatory microglial phenotype has also been linked to the development of chronic neurodegenerative diseases such as AD, multiple sclerosis (MS), PD, and amyotrophic lateral sclerosis (Hanamsagar and Bilbo, 2016; Lenz and Nelson, 2018). E2 treatment prevents or decreases expression of the microglial reactive (M1-like) state and promotes the more reparative (M2-like) state in several models of brain injury and neurodegenerative diseases (Tables 1 and 2). Among the functional reactions of microglial cells to E2 are the complete inhibition or dampening of responses such as nitric oxide (NO) production, release of proinflammatory cytokines, cell migration, and phagocytosis, as well as a reduction in the expression of proinflammatory genes. Central to the mechanisms by which E2 exerts these microglial cellular changes are the proteins to which it binds, the estrogen receptors (ERs).
Examples of E2 anti-inflammatory effects in in vivo models of CNS injury and disease
Studies reported the reactive state of microglia utilizing morphologic assessments, expression of M1/M2 markers, or expression of cytokines and chemokines in the affected brain regions. Results might also reflect changes in the number of infiltrating macrophages to lesioned sites. Whether a physiologic or pharmacological dose of E2 was administered is also indicated; dose description is based on measurements of circulating E2 levels reported by the studies or cited literature.
Examples of direct anti-inflammatory effects of E2 in microglia in vitro models
E2 neuroprotective and anti-inflammatory actions are mediated by the activation of three ER subtypes: the classic ERs, ERα and ERβ, and the membrane ER, GPER1 (also known as GPR30). Transgenic mouse models with global or cell-specific ER knockouts, as well as pharmacological tools such as selective ER agonist and antagonists, have been used to investigate the role that each ER subtype has in E2 modulation of microglial phenotypes. However, the contribution of specific ER subtypes and the cellular mechanisms they employ to mediate the anti-inflammatory and hence neuroprotective effects of E2 are still unclear. Even less is known about the role that each ER subtype might play in microglial development and maturation and hence in determining the sex-specific functions of these cells. The present review examines the different experimental models used to study E2 effects on microglia and the specific ER subtypes involved. It also appraises the research conducted on microglial responses to ischemic stroke, a sexually dimorphic neurologic disease in which E2-ER neuroprotective and anti-inflammatory actions are determined by both biologic sex and age.
Estrogen Receptors at a Glance
The majority of studies investigating the effects of E2 on microglial immunologic responses focus on activation of the classic ERs, ERα and ERβ. Classic ERs are members of the nuclear hormone receptor superfamily, and although they are encoded by distinct genes (ESR1 and ESR2, respectively) they share a multidomain protein structure that confers similar mechanisms of hormone response (Hewitt and Korach, 2018). Their ligand binding domain confers affinity and specificity to E2 whereas their DNA binding domain confers binding to estrogen-responsive elements (EREs) in the DNA. In the classic model of ER signaling, ERs form homodimers or heterodimers in the presence of E2, translocate to the nucleus, and trigger transcription by binding to EREs. Alternatively, ERs can interact with other transcriptional proteins such as activating protein 1, signal transducer and activator of transcription (STAT) 3, or nuclear factor kappa B (NF-κB) to regulate transcription at non-ERE sites (Hamilton et al., 2017). ERα and ERβ can also work in a nonclassic membrane-initiated manner and in a ligand-independent fashion. Plasma membrane–initiated signaling is triggered by binding of E2 to either classic ER and leads to diverse and rapid cytoplasmic effects, such as the activation of the mitogen-activated protein kinase (MAPK)/ERK and the phosphoinositide-3-kinase (PI3K)–Akt signaling pathways (Hamilton et al., 2017; Hewitt and Korach, 2018). In addition, growth factors such as insulin-like growth factor 1 may bring about ligand-independent ER activation by inducing the phosphorylation of ERs; phosphorylated ERs may then bind to DNA and stimulate transcription (Hewitt et al., 2017).
Estradiol’s rapid nongenomic actions in the CNS can also be mediated by the activation of nonclassic membrane ERs. Of these, the G protein coupled receptor, GPER1, is the best characterized (Alexander et al., 2017). GPER1 signals via the Gαs protein to activate the protein kinase A and ERK signaling pathways. GPER1 also couples to the pertussis toxin–sensitive Gαi/o protein, which can activate the PI3K or the MAPK/ERK signaling pathways via the transactivation of the epidermal growth factor receptor. In addition, GPER1 activation elevates intracellular Ca2+ levels through phospholipase C and inositol (1,4,5)-triphosphate receptors.
Both the classic ERs and GPER1 are distributed throughout the CNS, with ERα and ERβ showing a sexually dimorphic, brain region–specific distribution (Kelly et al., 2013; Zuloaga et al., 2014). These receptors are highly expressed in neurons (Hazell et al., 2009; Mayer et al., 2010; Zuloaga et al., 2014), astrocytes (Platania et al., 2003; Spence et al., 2011; Morgan and Finch, 2015), and oligodendrocytes (Platania et al., 2003; Khalaj et al., 2013), and ER signaling in these brain cells is diverse. Upregulation of neurotrophic and neuroprotective molecules such as brain-derived neurotrophic factor (BDNF) and insulin-like growth factor 1, activation of cell survival pathways such as PI3K-AKT signaling, and suppression of apoptotic signaling are all associated with membrane and classic ER-mediated neuroprotective actions of E2 in brain cells (Azcoitia et al., 2019). The ER subtype activated by E2 as well as the molecular mechanisms involved depend on the disease model and the brain region under investigation.
E2 also maintains normal CNS homeostasis and protects the brain against injury and disease through its potent anti-inflammatory effects in microglial cells. E2 treatment transforms microglial cell phenotype from a reactive M1-like to a less reactive M2-like state during responses to various insults, such as lipopolysaccharides (LPSs) or hypoxia (Johann and Beyer, 2013; Tables 1 and 2). E2 is likely to modulate microglial cell function through multiple mechanisms. As with other brain cells, the specific ER signaling pathway that mediates E2 effects on microglia depends on the disease model under investigation, brain region, gonadal hormone status, sex, and age.
Although the anti-inflammatory and neuroprotective actions of E2-ER signaling are undisputed, we do not have a clear understanding of how E2-ER signaling varies with age. This is an important gap in the field. A reduction in ER expression or function might explain, at least in part, the increased vulnerability of the brain to damage and the decreased E2 neuroprotective effectiveness with age. Indeed, a decrease in the ratio of ERα:ERβ expression in hippocampal and cortical neurons with age correlated with the inability of E2 to facilitate memory in aged rats (reviewed in Foster (2012). In contrast, the ERα:ERβ ratio in rat astrocytes increased with cessation of estrous cyclicity and with age in females and correlated with the loss of E2-dependent neurotrophic activity (Morgan and Finch, 2015). Less is known about changes in ER expression in the aging mouse brain. Some studies report a decrease in ERβ protein (Sharma and Thakur, 2006) or mRNA (Thakur and Sharma, 2007) in the absence of changes in ERα expression in the cortex of aged female mice (Thakur and Sharma, 2007). On the other hand, a recent study reported a reduction in ERα and ERβ protein expression in the hippocampus of old (20 months) versus young female (3 months) mice (Zhao et al., 2017). As discussed below, aging and decreased gonadal hormones are associated with impaired microglial neuroprotective functions. Because the decrease in sex hormones after menopause is correlated with an increase prevalence of neurologic disorders in women (Brinton et al., 2015), it will be important to examine whether changes in E2-ER signaling contributes to microglial dysfunction.
E2 Anti-Inflammatory Effects in the CNS
Removal of the main source of E2 production by ovariectomy (OVX) has been extensively used to study acute and long-term effects of this hormone on brain function and to examine the contribution of gonadal hormones to the sex differences observed in microglial phenotypes. Chronic (several weeks) as opposed to acute (5–14 days) OVX is also used to mimic the absence of endogenous hormone production experienced by women after menopause. In rats and mice, chronic OVX promoted the appearance of highly reactive microglia (Lei et al., 2003; Vegeto et al., 2006; Cordeau et al., 2016), exacerbated the induction of their M1-like proinflammatory state by immune challenges such as LPS (Vegeto et al., 2006; Wu et al., 2016), and accelerated the phenotypic pro-M1 changes associated with aging in mice (Wu et al., 2016). Most microglial phenotypic changes in response to LPS or injury are prevented by the chronic administration of E2 when the hormone treatment starts at the time OVX or shortly thereafter (Table 1). For example, in young OVX rats, chronic administration of E2 decreases the LPS-induced increase in the number of cortical and hippocampal microglial cells expressing matrix metalloproteinase 9 (MMP-9), a known contributor to neuronal injury (Vegeto et al., 2003). Similarly, the LPS-induced expression of chemokines (e.g., monocyte chemoattractant protein 1 and macrophage inflammatory protein 2) and proinflammatory cytokines [tumor necrosis factor alpha (TNF-α)] was reduced by chronic E2 treatment in OVX mice or rats (Vegeto et al., 2006).
The anti-inflammatory actions of E2 are also observed in animal models of injury or of neurologic conditions characterized by cognitive and/or motor dysfunction such as stroke, TBI, spinal cord injury, PD, or AD (Table 1). For example, E2 treatment diminishes the activation, proliferation, and migration of microglia toward sites of injury. This was demonstrated in the oculomotor nucleus after peripheral axotomy in OVX mice (Gyenes et al., 2010) and in the brain or spinal cord after TBI in both sexes (Sribnick et al., 2005; Barreto et al., 2007, 2014). In the APP23 mouse model of AD, E2 decreased the number of plaque-bearing activated microglia (Vegeto et al., 2006), and in a model of cuprizone-induced demyelination, E2 delayed microglial accumulation and reduced mRNA expression of TNF-α (Taylor et al., 2010). However, E2 treatment does not always fully restore the changes in microglial phenotypes induced by damage. For example, in a study using the toxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) to model PD, E2 treatment prevented microglial cell proliferation without affecting morphologic changes in microglia induced by MPTP (Tripanichkul et al., 2006). Varied effects of E2 might be due to differences in the route, dosage, and/or timing of hormone administration. Most studies used E2 treatments that result in physiologic circulating levels (0.3–145 pg/ml, (Nilsson et al., 2015), while others used treatments that produce supraphysiological E2 levels (Tables 1 and 2). This is important, as the levels of E2 can affect the regulation of ER subtype expression in a brain region– and cell type–dependent fashion. For example, chronic E2 deprivation in rats decreased hippocampal ERα expression (Zhang et al., 2011), whereas treatments producing supraphysiological plasma E2 levels in mice can decrease hippocampal ERα expression (Iivonen et al., 2006). We lack information about how changes in gonadal hormone levels in young and old animals affect ER subtype expression in many cell types, including microglia. Therefore, selection of the appropriate route, dosage, and timing of E2 administration to achieve physiologic levels is critical to generate relevant and reproducible data.
Involvement of Classic and Membrane ERs in E2 Anti-Inflammatory Effects in the CNS
Few studies have focused on identifying which ER(s) is responsible for E2 anti-inflammatory actions and its effects on microglia. However, activation of ERα signaling may play a key role. First, the anti-inflammatory actions of E2 against LPS treatment were lost in mice carrying a global deletion of ERα but not ERβ (Vegeto et al., 2003). Moreover, in transient transfection assays using a reporter for the MMP-9 gene promoter, E2 activation of ERα, but not ERβ, prevented MMP-9 promoter induction, suggesting that ERα transcriptional mechanisms play a role in E2 anti-inflammatory effects. Subsequent studies using primary neonatal microglial cultures showed that E2 inhibits LPS activation of NF-κB signaling through rapid nongenomic actions mediated through the PI3K signaling pathway (Ghisletti et al., 2005). These E2 effects were absent in ERα-null microglia but not in wild-type or ERβ-null microglial cells.
Selective estrogen receptor modulators (SERMs) such as tamoxifen and raloxifene, which bind to both ERα and ERβ, show that SERMs have neuroprotective actions that are often accompanied by a reduction in microglial reactivity (Baez-Jurado et al., 2019). For example, chronic E2 or raloxifene treatment lowered the number of macrophage antigen (Mac)-1 positive microglial cells in the dentate gyrus and CA1 hippocampal regions of aged (20–24 months) OVX mice (Lei et al., 2003). In addition, raloxifene or tamoxifen decreased expression of the major histocompatibility complex II (MHC-II), a marker of antigen-presenting activity, in young and aged OVX rats in a model of TBI (Barreto et al., 2014). In other contexts, such as in vitro studies using microglial cells, these two SERMs were used to block the activation of ERα and ERβ by E2. Raloxifene and tamoxifen can exert anti-inflammatory and neuroprotective effects through the activation of GPER1 (Alexander et al., 2017). Therefore, caution should be taken in interpreting the protective and anti-inflammatory actions of SERMS, as they might involve the activation of both classic and nonclassic ERs (Table 1).
Selective ligands for ERβ or GPER1 have been investigated as alternatives to E2 treatment for the improvement or prevention of microglial hyperactivation and other negative sequelae caused by brain injury or neurodegenerative diseases. In the experimental autoimmune encephalomyelitis (EAE) model of MS, treatment of affected animals with ERβ-selective ligands, such as the steroid hormone derivative 5-androstenediol (Saijo et al., 2011) or the benzopyran LY3201 (Wu et al., 2013), reduced EAE clinical severity scores and mortality and decreased the macrophage/microglia activated state, as evidenced by downregulation of NF-κB and inducible nitric oxide synthase (iNOS) expression in microglia. In a model of TBI, treatment of male rats with the GPER1 agonist G1 improved cognitive function and exerted neuroprotective effects (Pan et al., 2018). The beneficial effects of G1 in this TBI model were explained in part by a G1-induced decrease in neuronal apoptotic signaling, evidenced by a reduction in active caspase-3 positive hippocampal neurons and by the promotion of a more M2-like anti-inflammatory microglial phenotype.
E2 Regulation of Microglial Phenotypes through ERs
To study the mechanisms underlying modulation of microglial phenotypes by E2, immortalized microglial cell lines, primary cultures of neonatal rat and mouse microglia, and primary cultures of adult rodent microglia have been employed. These in vitro systems offer a controlled environment in which to determine the role that each ER subtype has in the E2 modulation of microglial responses to challenges, including LPS, hypoxia, amyloid β, and bacteria (Table 2).
Immortalized Microglial Cell Lines.
BV-2 and N9 cells are widely used surrogates of primary microglia for in vitro studies. They were created by the retroviral transduction of oncogenes into mouse neonatal (BV-2) or mouse embryonic (N9) primary microglia (Stansley et al., 2012; Timmerman et al., 2018). Among the phenotypic characteristics they share with primary microglial cells are the expression of microglial markers such as MAC-1 and MAC-2, as well as LPS-induced production of nitric oxide and of the cytokines interleukin (IL) 6, TNF-α, and IL-1, although the latter responses are relatively lower in the immortalized cells (Stansley et al., 2012). Pre- or cotreatment of these cells with E2 suppressed the LPS-induced activated and proinflammatory M1 phenotype and enhanced their anti-inflammatory M2 characteristics in a dose-dependent manner (Benedek et al., 2017; Thakkar et al., 2018). Moreover, treatment with the nonselective ERα/β antagonist ICI 182 780 blocks E2 effects in both cell systems, suggesting the involvement of the classic ERs (Bruce-Keller et al., 2000). In fact, classic ERs are differentially expressed in BV-2 and N9 cells, with most studies showing that BV-2 cells exclusively express ERβ, whereas N9 cells express both ERα and ERβ mRNA and protein (Table 3). Studies investigating the effects of selective ER agonists have taken advantage of the selective expression of ERβ in BV-2 cells. Treatment of the cells with the ERβ-selective ligand diarylpropionitrile (DPN) reduced LPS-stimulated iNOS and cyclooxygenase-2 (Cox-2) expression, whereas the ERα-selective ligand propyl pyrazole triol (PPT) was without effect (Baker et al., 2004). More recently, opposing actions of ERβ and GPER1 activation on the phagocytic activity of BV-2 cells were described; DPN enhanced but the GPER1 agonist G1 attenuated the ability of microglia to phagocytose apoptotic PC12 cells (Loiola et al., 2019).
Co-expression of estrogen receptor subtypes in different microglial cell systems
In N9 cells, E2 treatment attenuated LPS-induced superoxide release and phagocytic activity in a dose-dependent manner, and these effects were blocked by the MAPK antagonist PD98059 or the ERα/β antagonist ICI 182 780 (Bruce-Keller et al., 2000). Other studies showed that pretreatment of N9 cells with E2 also increased the release of the anti-inflammatory cytokine IL-10 and decreased release of the proinflammatory molecules TNF-α and interferon gamma (Dimayuga et al., 2005).
Even though BV-2 and N9 cells are useful models to elucidate molecular mechanisms mediating E2 effects, these immortalized cells are not identical to primary microglia. For example, they show a more adherent and proliferative phenotype as well as a different transcriptome profile (Stansley et al., 2012; He et al., 2018). Moreover, upon examining the expression of the Y chromosome–specific gene SRY, Crain and Watters (2010) found that BV-2 cells are female and N9 cells are male. Previous studies had not considered the sex of these cells. Hence, it should not be assumed that the results obtained using these cells can be generalized to other injury models in which microglia participate.
Primary Microglial Cultures.
The immunoregulatory effects of E2 in models utilizing primary microglial cells harvested from neonatal or adult rodents are diverse (Table 2). As with their immortalized counterparts, E2 treatment of primary microglia prevents the morphologic changes and production of proinflammatory signals induced by diverse toxic challenges such as LPS and hypoxia. However, there are contradictory reports about the roles ER subtypes play in the direct actions of E2 on microglia in vitro. Some studies reported low but detectable mRNA expression of ERα but not ERβ mRNA in primary cultures of neonatal mouse microglia (Sierra et al., 2008; Crain et al., 2013); Table 3). mRNA and protein expression of ERβ and GPER1 but not ERα was reported in primary microglial cell cultures from neonatal rats (Habib et al., 2014); yet other investigators detected expression of both classic ERs in this system (Vegeto et al., 2001; Liu et al., 2005). Similar discrepancies are found when reviewing studies using primary cultures of adult rat or mouse microglia (Table 3). However, until recently it was not considered that microglia might show sex-specific characteristics in vitro or that neonatal microglia might behave differently.
Also influencing microglia-specific ER expression are the isolation method used and whether the cells are harvested from whole brains or from a specific brain region. For example, it was reported that the gene expression in cultured microglial cells is significantly different from that of microglia freshly isolated from animals of the same age (Crain et al., 2013). In this regard, ERα mRNA levels in cultured primary microglial cells were significantly lower than in freshly isolated male and female microglia of any age. Follow-up studies showed that microglial gene expression of neuroprotective and proinflammatory genes varied by sex, age, and the CNS region from which the microglia were obtained (Crain and Watters, 2015). In male mice of all ages examined, expression of ERα mRNA was significantly higher in microglia isolated from the cortex than the brainstem/spinal cord; in females, the difference in ERα mRNA was significant only at 12 months of age. Therefore, differences in ER subtype expression that occur as a result of age, sex, tissue source, or method of microglial cell isolation may contribute to the diverse and often conflicting findings about the modulation of microglial function by E2-ER signaling. In support of this idea, sex-specific effects on IL-1β mRNA expression after LPS stimulation and/or E2 treatment were reported in both neonatal and adult rat primary microglial cultures (Loram et al., 2012). When using neonatal microglia, the males had a greater IL-1β response to LPS compared with females. Coadministration with E2 suppressed IL-1β mRNA in males, whereas in females, IL-1β mRNA was potentiated by E2. The effects of LPS and of E2 on IL-1β mRNA was also sex specific in microglia isolated from adult males or OVX female rats. In this case, the LPS induction of IL-1β mRNA was greater in males than in females, and coadministration of E2 suppressed IL-1β mRNA in female but not male microglia. In contrast, E2 treatment in vitro potentiated IL-1β mRNA response to LPS in microglia isolated from OVX females that were supplemented with E2 in vivo. Therefore, maturity of microglia at the time of isolation influences proinflammatory responses as well as E2 effects on microglial phenotypes. Because the vast majority of studies have used microglial preparations of mixed sex and/or neonatal microglial cultures, our understanding of ER subtype–specific actions on microglia remains incomplete.
Discrepancies in microglial specific expression of ER subtypes are also encountered in data from various in vivo experiments (Table 3). E2 treatment in vivo alters microglial phenotypes in brain areas such as the cortex and hippocampus where expression of ERα and/or ERβ is low. In a study using electron microscopic immunohistochemistry, sparse cytoplasmic ERα protein expression was reported in adult mouse hippocampal microglia (Sierra et al., 2008). In the rat cerebellum, punctiform ERα but not ERβ immunostaining was observed in the soma and cell processes of microglia, and treatment with LPS, E2, or tamoxifen increased ERα punctiform immunostaining in the microglial perikaryon (Tapia-Gonzalez et al., 2008). However, other studies detected ERβ but not ERα protein expression in microglia or found that microglia express both classic ERs, albeit at very low levels (Table 3). As with studies investigating microglial cell-specific ERα/β mRNA, regional heterogeneity in CNS microglial ER expression, sex differences, and low ERα/β coexpression might explain these discrepancies. Interestingly, GPER1 recently emerged as an important mediator of E2-induced anti-inflammatory effects on microglia. GPER1 is expressed in primary microglial cultures from neonatal rats and mice (Zhao et al., 2016; Zhang et al., 2018) and in adult murine primary microglial cultures (Loiola et al., 2019), and GPER1 protein has been detected in cortical microglia of female rats and mice (Zhao et al., 2016; Zhang et al., 2018). However, whether microglial expression of GPER1 is sexually dimorphic is not known.
These findings suggest that regulation of microglial cell functions by E2-ER signaling cannot be excluded. Nonetheless, ER signaling in neurons and astrocytes likely participates in E2 regulation of microglial reactivity to injury and disease, as the reciprocal communication among the three cells types is well established (Szepesi et al., 2018; Jha et al., 2019). Neurons release immune-related soluble factors that bind to their putative receptors on microglia and promote specific microglial changes (Qin et al., 2019). Neurotrophins, neuropeptides, neurotransmitters, anti-inflammatory cytokines, and chemokines are among the many factors whose expression and release by neurons can be modified by estrogens through direct activation of ERs. Neurotrophins such as neurotrophin-3, BDNF, and nerve growth factor, released by activated neurons, regulate microglial MHC-II expression and induction of proinflammatory molecules (Neumann, 2001). Receptors for these neurotrophic factors colocalized with ERs, and E2 regulates not only the sensitivity of neurons to neurotrophins but also neurotrophin expression and release (Chan and Ye, 2017). Similarly, activation of ERs such as ERα in astrocytes could modulate the release of molecules that in turn regulate microglial phenotypes and functions such as motility and phagocytosis. Therefore, because the activity of one cell type influences the others, E2-ER signaling in neurons or astrocytes might influence microglial phenotypes.
Role of ER Signaling in Sex-Specific Microglial Responses to Ischemic Brain Injury
Ischemic stroke, caused by a loss of blood flow to the brain, accounts for more than 87% of all strokes (Benjamin et al., 2017). Recent analysis of temporal trends suggests that stroke incidence may be declining to a greater extent in men than in women and that the overall decline in stroke incidence observed over time is being driven by a decrease in men (Madsen et al., 2017). The factors underlying this epidemiologic phenomenon are unclear. However, sex and age are important nonmodifiable variables affecting risk, pathophysiology, treatment responses, and outcomes in stroke (Bushnell et al., 2018; Roy-O’Reilly and McCullough, 2018). Fluctuations in the levels of sex hormones such as E2 that occur throughout a women’s lifespan contribute to sex differences in stroke. For instance, menopause increases the risk of stroke in women, and early menopause is associated with a 2-fold increase in risk for ischemic stroke (Rocca et al., 2012). Even though our understanding of E2 neuroprotection in stroke has grown considerably, it is not possible to incorporate the use of this or other sex hormones in the clinic to prevent or treat stroke. Concerns about increasing the risk for certain gynecologic cancers (Collaborative Group on Hormonal Factors in Breast Cancer, 2019) and the potential of E2 to exacerbate stroke incidence and severity if given to elderly women (Wassertheil-Smoller et al., 2003) partly explain the lack of E2 utilization as a therapeutic strategy for stroke. Moreover, in preclinical and clinical studies, promising therapeutic targets for stroke are often effective in males but not females (Hagberg et al., 2004; Amiri-Nikpour et al., 2015).
Activation of microglial cells is a key component of the postischemic inflammatory response that contributes to the extent of brain injury (Qin et al., 2019). Once ischemia occurs, microglia are activated to produce both detrimental and neuroprotective mediators, and the balance of the two counteracting mechanisms determines the fate of injured neurons (Benakis et al., 2015). Therefore, treatments that minimize the detrimental effects and/or maximize the protective actions of microglia have enormous clinical potential. In this regard, the antibiotic drug minocycline, which is commonly used to inhibit microglial activation (Lenz et al., 2013), has been touted as a promising neuroprotective agent in patients with acute stroke. However, in a clinical study of minocycline neuroprotection in ischemic stroke, male but not female patients with stroke who received oral minocycline daily for 5 days had significantly better neurologic outcomes (Amiri-Nikpour et al., 2015). The reasons for the lack of effectiveness of minocycline and other drugs in women are not clear. However, cumulative evidence suggests that microglial responses to stroke injury are sex-specific (McCullough et al., 2016), developmentally determined (Villa et al., 2018), modified by age (Yan et al., 2014), and influenced by reproductive experiences such as pregnancy and parity that are unique to women (Ritzel et al., 2017). Because the expression and function of ERs is also developmentally regulated (Platania et al., 2003; Prewitt and Wilson, 2007; Wilson et al., 2011; Perez-Alvarez et al., 2012) and influenced by age (Zuloaga et al., 2014; Morgan and Finch, 2015; Ianov et al., 2017) and sex (Wilson et al., 2011; Kelly et al., 2013; Zuloaga et al., 2014; Waters et al., 2015) it is reasonable to propose that classic and membrane ERs participate in the manifestation of microglial responses to ischemic injury in a sex- and age-specific fashion.
Developmental Programming of Sex-Specific Microglial Responses to Ischemic Injury.
That the sex of microglia influences the progression of ischemic stroke was recently demonstrated by Villa et al. (2018) in a study in which the progression of brain damage induced by permanent middle cerebral artery occlusion (MCAo) was monitored in male brains transplanted with purified male or female microglia (Villa et al., 2018). The progression of ischemic stroke was found to be more damaging in the male group than in the female group. The transplanted microglia accumulated in the vicinity of the infarct site, and the expression of the microglial anti-inflammatory marker Ym1 was higher in male brains transplanted with female microglia. Significant differences in the transcriptomes of microglia isolated from the brains of healthy adult male and female mice suggest a neuroprotective phenotype of female microglia. Most of the genes that were highly expressed in male microglia are associated with inflammatory processes, such as regulation of cell migration and cytokine production. Moreover, whole-genome molecular signature analysis of transcription factors identified NF-κB as the transcription factor most involved in the regulation of the differentially expressed genes preferentially expressed in male microglia. In contrast, microglial genes expressed in female-derived cells were linked to transcription factors involved in the inhibition of inflammatory responses and promotion of repair mechanisms, including ERα-regulated pathways. The sex differences in adult microglial gene expression were proposed to originate from perinatal exposure to sex hormones, as treatment of female pups with E2 eliminated the sex differences in genes highly expressed in male microglia. Importantly, quantitative analysis of brain ERE-Luc expression (a surrogate of ERα activity) showed that perinatal activation of ERα is restricted to males. Hence, the sex of microglia might be determined early in development through a mechanism that involves ERα signaling.
Similarly, Thion et al. (2018) recently reported that microglia start expressing gene products that sense ligands and pathogens, also known as “sensome” genes, in utero, acquiring a sexually dimorphic transcriptomic profile during the postnatal period. However, in contrast to the results of Villa et al., female microglia were found to have a higher expression of genes associated with inflammatory and apoptotic responses than male microglia. In addition, using germ-free mice, this group found that the maternal microbiota impacted the microglial transcriptome in a temporal and sex-specific fashion. For example, absence of the maternal microbiome most severely affected embryonic microglia from males, whereas in females the most marked perturbations were observed in microglia from adults. Recent studies suggest that perturbation of the gut microbiota or dysbiosis influences the neuroinflammatory responses to ischemic injury (Singh et al., 2016). Whether perturbations in microbiota also influence microglial responses to ischemia in a sex-specific manner awaits examination.
ER Signaling and Microglial Responses to Ischemia in the Aging Brain.
A large number of preclinical studies indicate that E2 is neuroprotective in models of ischemic injury such as permanent or transient MCAo (Dubal and Wise, 2001) and global cerebral ischemia (GCI; De Butte-Smith et al., 2009; Traub et al., 2009). For instance, female rodents were less susceptible to ischemic stress than males (Liu et al., 2009; Dang et al., 2011; Manwani et al., 2013). OVX prior to stroke induction abolishes this sex difference, and treatment of males or OVX females with physiologic doses of E2 restores neuroprotection (Gibson et al., 2006; Petrone et al., 2014). The effectiveness of neuroprotection and the mechanisms used by E2 to protect against ischemic brain injury varies with age, and in the case of females whether E2 is neuroprotective or neurotoxic is also determined by how long the brain has been deprived of E2 (Dubal and Wise, 2001; Zhang et al., 2009; Selvamani and Sohrabji, 2010; Lewis et al., 2012; Liu et al., 2012). This age dependence of E2 actions is significant in view of the increased vulnerability to stroke experienced by the elderly, especially elderly women (Scott et al., 2012).
The decreased of E2 in protecting the brain with age is correlated with changes in ER expression. Expression of ERα splice variants that show dominant-negative regulation of ERα increased in the brain of elderly women (Ishunina and Swaab, 2008, 2009). Similarly, aging and/or extended E2 deprivation caused a reduction in ERα, ERβ, and GPER1 in the hippocampus of female rats (Zhang et al., 2011; Ianov et al., 2017; Wang et al., 2019). Intriguingly, G1 was reported to reverse the reduction of ERα, ERβ, and GPER1 expression within the hippocampus of aged female rats without affecting plasma E2 levels (Wang et al., 2019). Studies of changes in cortical ERβ and ERα expression with age in mice are scant, with one study reporting a decrease in ERα mRNA but not protein expression in the cortex of aged mice (Dietrich et al., 2015). Another study reported a significant decrease in ERβ but not ERα protein in the mouse cortex with aging (Sharma and Thakur, 2006). Although it is clear that important changes in ER subtype expression occur with age, studies analyzing ER expression with cellular and neuroanatomical resolution are missing. Experiments using validated antibodies for each ER subtypes and of microglial cell-specific markers in combination with novel reporter mouse models are needed to fill this gap.
Supporting the idea that loss of ER expression or function may underlie alterations in immune responses and increased vulnerability to stroke, Cordeau et al. (2016) showed that chronic OVX of ERα-null mice significantly altered the microglial activation profile and the innate immune responses after cerebral ischemia. This was shown by a significant attenuation of signals produced by the activation of the scavenger receptor Toll-like receptor (TLR) 2 signaling in OVX ERα knockout animals after stroke. In contrast, chronic OVX of wild-type or OVX ERβ-null mice produced the expected increase in TLR2 signaling before and after stroke. Further analysis associated the attenuation of the TLR2 signaling in OVX ERα-null females with markedly altered profiles of activated microglial cells, selective overexpression of IL-6, and overactivation of the Janus kinase/STAT3 pathway as well as significantly larger ischemic lesions.
Little is known about the role of microglial cell-specific ERα signaling after ischemia. In one study, the role of neuronal versus microglial ERα in E2 neuroprotective effects was investigated using mice with a neuron or myeloid cell–specific ERα deletion subjected to permanent MCAo (Elzer et al., 2010). The infarct volume of E2-treated females lacking ERα in microglia/macrophages was similar to that of E2-treated wild-type mice; in contrast, both male and female neuron-specific ERα-null mice had larger infarcts than control mice after E2 treatment. It was concluded that neuronal but not microglial ERα signaling in mice of both sexes mediates neuroprotective effects of E2 in this model of stroke. In this model, E2 neuroprotective actions were tested using young males and young OVX females that received E2 replacement at the time of surgery. It is unknown whether loss of microglial ERα or other ERs leads to cellular dysfunction and a higher vulnerability to ischemic injury in aged animals. This question is relevant and significant because sex- and/or aging-dependent changes in microglial activation profiles may be responsible for the observed increase in vulnerability to brain ischemia in elderly women.
ERβ and GPER1 Ligands: Alternatives to Decrease Postischemic Reactive Microglial Responses.
According to the “critical period hypothesis,” E2 therapy is beneficial for stroke and other brain diseases when taken by young women or older women during the perimenopausal or early postmenopausal periods but deleterious when taken by women who have been in menopause for many years (Scott et al., 2012). Moreover, due to concerns about the increased risk of breast and uterine cancers caused by E2 treatment (Collaborative Group on Hormonal Factors in Breast Cancer, 2019), ERβ agonists with minor effects in reproductive tissues, or GPER1 ligands that exert nongenomic effects, have been investigated in the context of ischemic stroke and other CNS disorders.
ERβ-selective ligands such as DPN were neuroprotective in various stroke models, such as GCI (Ma et al., 2016) and transient focal ischemia (Shin et al., 2013). DPN neuroprotective effects were associated with a decrease in astrocyte reactivity in OVX mice subjected to GCI, whereas in the focal stroke model DPN effects included a reduction in BBB breakdown. A recent study found that DPN treatment of 8-month-old OVX mice alleviated ischemic brain injury after MCAo and reperfusion and inhibited the activation of microglia and astrocytes in the ischemic penumbra (Guo et al., 2020. DPN treatment suppressed the protein levels of NF-κB and of the proinflammatory cytokines, TNF-α, IL-1β, and IL-6. Moreover, pretreatment of N9 cells with DPN increased cell viability and decreased NF-κB and proinflammatory cytokine expression after oxygen-glucose deprivation and reperfusion (OGD/R), an in vitro model of global ischemia. These studies suggest that ERβ-mediated neuroprotection against ischemic injury is mediated via multiple cell mechanisms, including a reduction in microglial reactivity and inflammation.
GPER1 may also participate in the neuroprotective actions of E2 in focal and global ischemia. For instance, G1 protected against hippocampal neuronal loss induced by ischemia in middle-aged female rats (Lebesgue et al., 2010). In a model of focal ischemia induced by MCAo, treatment of young OVX mice with G1 or E2 reduced infarct volume and improved neurologic deficits (Zhang et al., 2018). The effects of G1 included a reduction in penumbral production of TNF-α, IL-1β, and IL-6, reduced coexpression of TLR4 protein in microglia, and inhibition of NF-κB activity. The effects of E2 were not altered by cotreatment with the ERα/β antagonist, ICI 182 780, suggesting that GPER1 plays a key neuroprotective role in this model. The anti-inflammatory effects induced by GPER1 activation were also observed in primary microglial cultures subjected to OGD/R. In this in vitro model, a reduced release of TNF-α, IL-1β, and IL-6, decreased in TLR4 mRNA levels, and inhibition of NF-κB activity by G1 were reported.
In summary, results from various experimental models of ischemic stroke suggest that E2-ER signaling modifies the microglial proreactive state to suppress neuroinflammation and protect the brain. Anti-inflammatory E2 effects are often accompanied by morphologic changes in and decrease of microglia. Both in vivo and in vitro models show that activation of all major ERs ameliorates microglial cell responses to ischemic injury. Inhibition of the NF-κB pathway and the concomitant reduction in microglial proinflammatory cytokine production play a major role in ER-mediated effects. E2-ER signaling also modulates the expression of TLR2 and TLR4 receptors, which could impact microglial phagocytosis.
Concluding Remarks
Increasing evidence suggests that the role of each ER subtype in shaping microglial phenotypes is influenced by the internal gonadal hormone milieu during development, which in turns determines sex- and age-specific responses to injury in the adult brain. However, to further advance our understanding of E2-ER signaling in microglial cells, we need to 1) use novel genetic tools that can enable a comprehensive characterization of ER expression and function in microglial cells in vivo, 2) expand our investigations of microglial cellular responses to E2 beyond that of their inflammatory phenotypes, and 3) employ in vivo and in vitro models that reflect the physiologic changes in peripheral and local E2 levels that occur with age.
One obstacle impeding the investigation of specific functions of microglia in CNS diseases has been the lack of markers that can distinguish resident microglia from infiltrating myeloid cells. The identification of the transmembrane protein 119 as a specific marker of human and mouse microglia (Bennett et al., 2016) has enabled the generation of a specific antibody (Bennett et al., 2016) and a reporter mouse line (Ruan et al., 2020) that can be used to assess ER subtype expression in microglia with greater specificity. Discrimination between microglia and infiltrating macrophages could be achieved through double immunofluorescence, utilizing ionized calcium-binding adapter molecule 1 (Iba1; a microglia/macrophage marker) in addition to transmembrane protein 119 antibody (Gonzalez Ibanez et al., 2019). Moreover, the research community has yet to use validated Cre mouse lines that specifically target microglia to study ER subtype function. The Cre models used up to now not only lack microglial cell specificity but also may have had a low recombination efficiency (Wieghofer et al., 2015). Complementary in vitro studies using primary cultures and coculture systems may reveal novel mechanisms underlying E2-ER effects on the microglial M1-like and M2-like state. As one example, changes in mitochondrial function in microglia in response to in vitro challenges such as OGD/R could give insight into the mechanisms that control E2-ER–mediated changes in microglia anti-inflammatory or proinflammatory phenotypes. This is important, as strategies that drive microglia toward a neuroprotective phenotype are being sought to treat injury and neurodegenerative diseases. The value of such findings will depend on how well the models reflect physiologic changes in E2 levels that occur throughout life in both sexes.
That microglia are key targets of E2 neuroprotective actions is undeniable. However, we are far from understanding how ER signaling in microglia varies with age and hormonal cycles in females, and these variables are known to impact microglial responses to injury and disease. Examination of the literature suggests that ERα signaling has a key role in determining the sex-specific microglial transcriptome. In addition, judging by studies using global knockout mice, ERα is also important in mediating E2 anti-inflammatory effects and hence neuroprotection against ischemic injury and other neurologic diseases. On the other hand, activation of ERβ and GPER1 recruits alternative pathways by which E2 protects the brain by decreasing microglial hyperreactivity, which if unchecked is neurotoxic. Recent studies suggest that membrane and classic ERs interact with one another to enhance E2 actions. For example, GPER1 required ERα/β to increase BDNF levels and to protect dopaminergic neurons in a mouse model of PD (Bourque et al., 2015). Adding to the complexity of ER-mediated pathways is the fact that changes in ER expression or function with age vary by cell type and hormonal status. It is unknown whether microglia themselves experience changes in the expression or function of the different ERs with age or if altered ER expression in other cell types impacts microglial function. Novel genetic and pharmacological tools to monitor and manipulate the expression of ER subtypes in microglial cells are needed to answer these questions.
According to recent population projections, the number of people over 60 years of age will more than double by 2050, and the number of people over 80 years old is expected to triple over this period (United Nations, 2019). Because of the close relationship between age and the incidence of neurologic diseases such as stroke, the increase in elderly people will be accompanied by a rise in the number of patients affected by these brain diseases (Bejot et al., 2019). To better prepare for this challenge, it will be important to consider additional clinically relevant factors in the design of preclinical models such as those for ischemic stroke. For example, age-comorbid human conditions such as impaired metabolic control should be incorporated in experimental models of stroke and other brain diseases. In this respect, obesity and metabolic syndrome are suspected to play a role in a recent surge in midlife stroke incidence among women (Peters et al., 2014). Therefore, understanding the mechanisms that regulate ER signaling with age in the presence of these and other comorbid conditions is critical to identify effective therapeutics targeting microglia.
Acknowledgments
I would like to thank Dr. Anne M. Etgen, Dr. Helen Fox, Dr. Lisa Ballou, and Dr. Ariel L. Negrón for their invaluable comments and editorial support.
Authorship Contributions
Participated in research design: Acosta-Martínez.
Wrote or contributed to the writing of the manuscript: Acosta-Martínez.
Footnotes
- Received December 12, 2019.
- Accepted May 5, 2020.
Abbreviations
- AD
- Alzheimer’s disease
- BBB
- blood-brain barrier
- BDNF
- brain-derived neurotrophic factor
- CNS
- central nervous system
- Cox-2
- cyclooxygenase 2
- DPN
- diarylpropionitrile
- E2
- estradiol
- EAE
- experimental autoimmune encephalomyelitis
- ER
- estrogen receptor
- ERE
- estrogen-responsive element
- GCI
- global cerebral ischemia
- GPR30
- G protein–coupled receptor 30
- Iba1
- ionized calcium-binding adapter molecule 1
- IL
- interleukin
- iNOS
- inducible nitric oxide synthase
- LPS
- lipopolysaccharide
- MAC
- macrophage antigen
- MAPK
- mitogen-activated protein kinase
- MCAo
- middle cerebral artery occlusion
- MHC-II
- major histocompatibility complex II
- MMP-9
- matrix metalloprotease 9
- MPTP
- 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
- MS
- multiple sclerosis
- NF-κB
- nuclear factor kappa B
- NO
- nitric oxide
- OGD/R
- oxygen glucose deprivation and reperfusion
- OVX
- ovariectomy
- PD
- Parkinson’s disease
- PI3K
- phosphoinositide-3-kinase
- PPT
- propyl pyrazole triol
- SERM
- selective estrogen receptor modulator
- STAT
- transducer and activator of transcription
- TBI
- traumatic brain injury
- TLR
- Toll-like receptor
- TNF-α
- tumor necrosis factor alpha
- Copyright © 2020 by The American Society for Pharmacology and Experimental Therapeutics