Central aromatization: a dramatic and responsive defense against threat and trauma to the vertebrate brain.

Abstract

Aromatase is the requisite and limiting enzyme in the production of estrogens from androgens. Estrogens synthesized centrally have also emerged as a potent neuroprotectant in the vertebrate brain. Studies in rodents and songbirds have identified key mechanisms that underlie both; the injury-dependent induction of central aromatization, and the protective effects of centrally synthesized estrogens. Injury-induced aromatase expression in astrocytes occurs following a broad range of traumatic brain damage including excitotoxic, penetrating, and concussive injury. Responses to neural insult such as edema and inflammation involve signaling pathways the components of which are excellent candidates as inducers of this astrocytic response. Finally, estradiol from astrocytes exerts a paracrine neuroprotective influence via the potent inhibition of inflammatory pathways. Taken together, these data suggest a novel role for neural aromatization as a protective mechanism against the threat of inflammation and suggests that central estrogen provision is a wide-ranging neuroprotectant in the vertebrate brain.

Introduction

Estrogens have long been appreciated for their potent organizational and activational effects on sex-specific neural circuits. Recent evidence however, has also implicated 17β-estradiol (E₂) in neuroprotection (Maggi et al., 2004; McEwen et al., 2001; Wise, 2003). In several species, elevations in peripheral E₂ are associated with dramatic decreases in markers of cell death and the numbers of necrotic or apoptotic cells in the CNS (Belcredito et al., 2001; Bryant et al., 2006; Roselli, 2007; Simpkins et al., 2005; Simpkins and Dykens, 2008). Elevations in circulating E₂ are likely the consequence of increases in the expression and activity of aromatase (estrogen-synthase), the rate-limiting enzyme in the synthesis of estrogens (Simpson et al., 1994). Indeed, aromatase expression has been documented in multiple peripheral tissues including the ovary, testes, placenta, bone, adipose, and adrenals (Vanselow et al., 1999).

The vertebrate brain is also a site of estrogen synthesis (Balthazart et al., 1990; Naftolin et al., 1996; Negri-Cesi et al., 2001; Peterson et al., 2005; Roselli et al., 2004; Roselli and Resko, 2001; Saldanha et al., 2011, 2000). In mammals, we have known about the critical role of hypothalamic aromatization in the development and activation of male-specific neural circuits and behaviors for some time; a phenomenon generalizable to multiple rodent species (MacLusky and Naftolin, 1981). While aromatase has been documented at extra-hypothalamic, forebrain sites of some other mammals, its physiological role remains somewhat equivocal (Naftolin et al., 1996; Roselli et al., 2004; Yague et al., 2008b; Zhong et al., 2017). Across species however, the constitutive expression of aromatase in the mammalian brain appears to be neuronal, suggested and documented with studies that have used in-situ hybridization and immunocytochemistry (Gottfried-Blackmore et al., 2008; Roselli, 2007; Yague et al., 2008a).

In birds, the role of neural aromatase in development and adulthood is perhaps better understood, due in part to pioneering work in Passeriformes (Gurney and Konishi, 1980), Columbiformes (Steimer and Hutchison, 1981, 1980), and Galliformes (Schumacher and Balthazart, 1984). Passerines (songbirds) distinguish themselves from other bird species since the expression of aromatase is particularly widespread and abundant (Metzdorf et al., 1999; Saldanha et al., 2000; Schlinger and Arnold, 1991). Among songbirds, the zebra finch (Taeniopygia guttata), appears particularly interesting because aromatase appears to be among the highest measured in the CNS of any homeotherm, and occurs at levels sufficient to enrich plasma levels of E₂ in both sexes, but especially in males (Adkins-Regan and Ascenzi, 1990; Schlinger and Arnold, 1993, 1991). Indeed, the neural estrogen synthetic capability in this species has been confirmed via multiple techniques including enzyme activity (Schlinger and Arnold, 1992, 1991), in situ hybridization (Saldanha et al., 1998; Schlinger et al., 1994; Shen et al., 1995; Soma et al., 2003) and immunocytochemistry (Peterson et al., 2005; Saldanha et al., 2000). While more abundant in the songbird relative to mammals, the constitutive expression of aromatase nonetheless appears to be restricted to neurons in every avian species studied (Peterson et al., 2005; Saldanha et al., 2011, 2000; Shen et al., 1995).

Despite the assertions above, aromatization is neither exclusively constitutive nor neuronal in the mammalian and passerine brain. Schlinger et al., first presented intriguing data suggesting that primary dissociated cell cultures of hatchling songbird telencephalon, expressed aromatase in glial cells (Schlinger, 1994). This was later confirmed in vivo by several studies which documented the expression of aromatase in astrocytes following excitotoxicity or mechanical injury in rodent and songbirds (Azcoitia et al., 2010; Garcia-Segura et al., 1999; Peterson et al., 2001; Saldanha et al., 2005). Interestingly, in some cases the expression of injury-induced aromatase is also seen in radial glia suggesting that both types of astroglia are capable of induced estrogen synthesis in songbirds (Peterson et al., 2004).

Here we review literature on the various types of challenges that appear to reliably induce aromatase in multiple species and introduce the suggestion that neural, astrocytic aromatization appears to be a potent neuroprotectant regardless of the type of insult in some vertebrates. We then present what is known about the induction of aromatase, go on to discuss the regulation and consequences of injury-induced aromatase following various forms of central perturbation in songbirds and mammals. In closing we present data supporting a novel role for neural aromatase in neuroprotection from the consequences of peripheral inflammation, and end with the hypothesis that induced aromatase appears to have evolved in multiple species as a neuroprotectant against a broad range of threat to the CNS, with intimate ties to the innate immune system.

A broad range of neural damage induces aromatase expression in astroglia

Primary evidence suggesting an association between neural degeneration and aromatase expression in astrocytes was presented by Garcia-Segura (Garcia-Segura et al., 1999). In these studies, neurodegeneration was induced in rats and mice, either with an excitotoxic dose of kainic acid or a penetrating mechanical injury. Eight or ten days later respectively, subjects showed increased aromatase activity and expression in the hippocampus, the brain area where degeneration was most apparent. Notably, immunoreactive aromatase colocalized with glial fibrillary acidic protein (GFAP) revealing the astrocytic nature of the cells with induced aromatase expression. Importantly, no degeneration or astrocytic aromatase was detected in control animals, establishing a correlation between neurodegeneration and the induction of aromatase in astrocytes. Peterson et al., echoed these findings using adult zebra finches, and added that the induction of aromatase following penetrating damage was detectable as soon as 24 hours post-injury in the songbird (later work showed increases even more rapidly) (Peterson et al., 2001). Together, these studies opened a new area of investigation on the spatial and temporal specificity of E₂ provision in the vertebrate brain, and accomplished two significant ends. First, they provided a context for earlier studies that reported astrocytic aromatase and an increase in aromatase activity in primary dissociated cell cultures of hatchling songbirds following the death of neurons in vitro (Schlinger et al., 1994). Perhaps more importantly, they strongly suggested an obligatory association between damage to the neuropil and central aromatization. We have learned that this association spans a wide range of neurotrauma including experimental models of stroke and concussive brain injury.

Ischemia and central aromatization.

Circulating E₂ may be neuroprotective in experimental stroke models using rodents. Ischemic damage caused by medial carotid artery occlusion (MCAO) is lowest when it is performed on female rats in proestrus relative to the identical manipulation in diestrus or metestrus animals (Carswell et al., 2000). Correspondingly, experimentally induced decreases of circulating estrogens by ovariectomy, pharmacological inhibition, or normal reproductive senescence result in higher lesion volumes following MCAO, suggesting a role for aromatization from peripheral sources on neural injury and damage (Alkayed et al., 1998; Rusa et al., 1999; Sawada et al., 2000).

More recent work points to an extragonadal, and likely neural source for neuroprotective E₂ following ischemia. McCullough et al., compared infarct volume following MCAO in female aromatase knock-out (ARKO) mice with ovariectomized wild-type (OVXWT) littermates and documented less damage in the latter, suggesting the possibility of an extragonadal neuroprotectant (McCullough et al., 2003). E₂ via aromatization is the likely neuroprotectant since infarct size in OVXWT mice treated chronically with an aromatase inhibitor had comparable damage to that of ARKO mice and E₂ treatment of ARKO mice completely ameliorated the increased susceptibility to MCAO-induced neural damage (McCullough et al., 2003). Indeed, the key site of aromatization in this neuroprotective effect may be the area around the lesion itself. In spontaneously hypertensive rats, MCAO increases the expression of aromatase in astrocytic processes in the penumbra of the neural infarct (Carswell et al., 2005). This upregulation is detectable at 24 hours and 8 days following treatment but is not 2 hours or 30 days after MCAO. Taken together these data strongly suggest that disruption of the neuropil due to experimental stroke can alter E₂ provision in the rodent brain via the induction of aromatase expression in a manner seemingly similar to that of other types of neurotrauma.

Concussion and central aromatization.

Our recent unpublished work has further extended the types of neural insult capable up altering the neural expression of aromatase. In the zebra finch and in the mouse, multiple mild concussive treatments using the controlled weight-drop technique reliably result in the upregulation or aromatase around the focus of concussive damage (see Figure 1). This upregulation is detectable at both the transcript and protein levels (in different animal models) and represents yet another type of neurotraumatic event that can upregulate aromatase in the homeotherm brain.

Figure 1.

Multiple mild concussive traumatic brain injury (mmTBI) were created by briefly anesthetizing subjects with isofluorane. Subjects were then placed on a modified weight drop apparatus (Laboratory weight drop device, Northeast Biomedical) where a 293.19 g rod was raised 1 cm above the head of the subject and dropped. The above procedure was repeated 4 times over 7 days and subjects were collected 10 days after the first injury. In zebra finches (A) mmTBI upregulates glial aromatase in astrocytes (purple) around the site of injury. Following mmTBI in mice (N=3/group), aromatase transcript was significantly upregulated in injured brains when compared to sham tissue following one-way ANOVA analysis. * denotes a significant difference at P <0.05 (one-way ANOVA)

The obvious question to be asked is: what signals, perhaps shared by these multiple types of neurotraumatic events, could be candidates as regulators of aromatase transcription in astrocytes? Several possibilities come immediately to mind, including cell death, changes in pressure and perhaps inflammatory signals.

Possible regulators of astrocytic aromatase expression

Cell Death.

Damage to the brain can be broadly separated into two phases, the initial primary physical trauma that results from the physical displacements of brain tissue, and the second injury that is the brain’s response to the trauma. The secondary injury includes the long-term effects of the initial trauma (necrosis or apoptosis), as well as the production of free radicals, increased excitotoxicity, and increased inflammation and edema (Bramlett and Dietrich, 2007; Lenzlinger et al., 2001; Maas et al., 2008; Marciano et al., 2002; Stein, 2008). Changes in signals associated with cell death are less compelling as a regulator of glial aromatase as multiple cells undergo pyknosis for various reasons at all times in the adult CNS. Furthermore, aromatase expression in the constitutive rodent and songbird brain appears to be neuronal, lessening the likelihood that signals associated with cell death could serve as triggers of astrocytic aromatase transcription. While one could argue that low levels of pyknosis may not be enough to reach a hypothetical threshold for the upregulation of aromatase in astrocytes, it is noteworthy that massive amounts of cell death occur every autumn within the nuclei of the song circuit, an interconnected set of brain areas necessary for singing behavior, in temperate-zone songbirds in the absence of observed glial aromatase expression, further lessening the likelihood of these signals in the regulation of astrocytic aromatase. Furthermore, in some songbirds like zebra finches, both male and females undergo massive neuronal proliferation during development, however females also undergo a period of neuronal pruning and regression that results in the sex differences observed in song nuclei number and size (Burek et al., 1997; Kirn and DeVoogd, 1989; Nordeen and Nordeen, 1988). Interestingly, this neuronal pruning or naturally occurring cell death does not result in glial aromatase expression either.

Edema.

Cerebral edema has a crucial impact on morbidity following TBI as it increases cranial pressure, impairs cerebral perfusion and oxygenation, and contributes to the overall degeneration of the brain (Sorby-Adams et al., 2017; Stokum et al., 2015). Astrocytes are key participants in cerebral edema by virtue of their relationship with the cerebral vasculature, their unique compliment of solute and water transport proteins, and their general role in brain volume homeostasis (Stokum et al., 2015). Specifically, one of the effects of even a single concussive event is increased edema over controls (unpublished Figure 2). Estradiol reduces edema formation and infarct volume in vivo (O’Donnell et al., 2006). Changes in pressure are fairly ubiquitous in multiple forms of neurotrauma including stroke, penetrating, and concussive injuries, however to the best of our knowledge these data have not been tested in vivo (Sorby-Adams et al., 2017). Intriguing evidence suggests that astrocytes may be capable of sensing changes in pressure. Gatson et al., exposed astrocytic cultures to various pressures and reported increases in aromatase message relative to GAPDH relative to unmanipulated controls (Gatson et al., 2011). Furthermore, perihematomal cerebral edema specifically up-regulates thrombin that further activates secretion of the cytokines: tumor necrosis factor (TNF) and IL-1β from glial cells (Stokum et al., 2015). These data again point to inflammatory cascades as the likely candidate for the potent induction of injury-associated aromatase expression, however the specific mechanism(s) of this influence have only recently been revealed.

Figure 2.

A single concussive injury was directed to the brain of an anesthetized adult zebra finches as described in Figure 1. 6 hours following injury, a 4mm diameter brain sample was collected and immediately weighed for the wet weight (WW). Brain samples were then placed at 110C for 24 hours and weighed again for the dry weight (DW). Brain water content or edema was calculated as %H2O = (WW - DW) × 100/WW. Edema (as measured by brain water content) is significantly higher in birds that underwent concussive traumatic brain injury (cTBI) relative to shams as measured using a one tailed Mann-Whitney U. * denotes a significant difference at P <0.05

Inflammation.

Since increased inflammation is an important component of the secondary response to damage, we sought to examine the role of inflammation alone in the regulation of astrocytic aromatase expression. In adult zebra finches, we unilaterally exposed the neuropil to either the inflammagen phytohaemagglutinin (PHA) or vehicle without causing cell death or tissue damage (Duncan and Saldanha, 2011), thus we were able to determine the role of inflammation on the brain independent of the cell death that is generally associated with brain damage. Treatment with PHA increased inflammation as confirmed by the upregulation of the expression of two cytokines (interleukin 1b and 6 (IL-1b and IL-6)). Importantly, tissue exposed to PHA had significantly higher expression of aromatase and this upregulation was in astrocytes as confirmed by colocalization of aromatase with the astrocytic marker vimentin. Furthermore, this was done without causing cell damage or death as we were unable to detect apoptosis or pyknosis in the tissue. Thus, we were able to determine that neuroinflammation and not cell death is the necessary initiation factor for the induction of glial aromatase (Duncan and Saldanha, 2011). This hypothesis is consistent with previous data. Injury induced inflammation is mediated by a number of factors, specifically cytokines (Duncan and Saldanha, 2011). Cytokines dramatically regulate aromatase gene expression in several mammalian cells, including normal and malignant breast tissue (Honma et al., 2002; Morioka et al., 2000; Purohit et al., 2002). Taken together, the data strongly suggest that inflammatory signals, such as those resulting from cell damage, may be a potent regulator of astrocytic aromatase expression. A description of the direct tests of this hypothesis and the mechanism of this influence is discussed below.

Inflammation initiates astrocytic aromatase

Inflammation is part of the body’s immune system response to an irritant (pathogen, injury, chemical, or radiation). For many years, it was viewed mainly as a necessary, even beneficial, response to illness or injury. But now research has indicated that inflammation can be more harmful than helpful. While inflammation is necessary for the activation of a number of downstream pathways involved in neuronal repair, if this inflammation becomes chronic or goes unchecked secondary damage can and will occur (“A current view on inflammation,” 2017). Below we will highlight the genes that regulate both the induction and progress of the inflammatory process, specifically those genes and regulators that also mediate glial aromatase.

Prostaglandin E2 (PGE₂).

In rats, administration of PGE₂ during a restricted neonatal period induces E₂ synthesis via neuronal aromatization (Amateau and McCarthy, 2004; MacLusky and Naftolin, 1981). PGE₂ is a proinflammatory signal and is also induced following injury via the enzyme cyclooxygenase (COX)-2 (Chen, 2010; Davidson et al., 2001; Kalinski, 2012; Ricciotti and FitzGerald, 2011). Administration of the COX-1/2 inhibitor indomethacin during development reduced cerebellar aromatase and E₂ and had dramatic effects on dendritic morphology and neurophysiology (Dean et al., 2012a, 2012b). Thus, local COX activity and consequent PGE₂ synthesis regulate aromatase activity in the mammalian brain. Blocking PGE₂ synthesis with indomethacin in the zebra finch brain, was the first identified method for blocking the well-known increase in aromatase and E₂ in the adult zebra finch brain following injury (Pedersen et al., 2018; Pedersen and Saldanha, 2017). Interestingly, indomethacin reduced aromatase expression and E₂ content at 6 hours, but not 24 hours following injury in females. However, in males, the inhibitory effect of indomethacin on aromatase and E₂ was apparent at 24, but not 6 hours after treatment. These data suggest that COX activity, perhaps via consequent prostaglandin secretion, may induce aromatase expression and central E₂, an effect that is detectable in temporally distinct patterns between sexes (Pedersen and Saldanha, 2017).

PGE₂ has a high affinity for four known E-prostanoid (EP) receptors: EP 1-4 (Singh et al., 1997). Binding of PGE₂ to these receptors can regulate aromatase and E₂ via modulation of downstream signaling pathways in other systems (Duncan and Saldanha, 2011; Phillips et al., 1978; Reed et al., 1993). Surprisingly, EP-1 and EP-2 receptors were not represented in the zebra finch genome, however they may play a greater role in non-avian species for the purpose of this review we are going to focus on EP-3 and EP-4 (Warren et al., 2010). The expression of both EP-3 and EP-4 changed in a temporally distinct manner following brain injury, with expression of EP-4 being higher in both sexes, but EP-3 only being significantly higher in females at earlier time-points (Pedersen and Saldanha, 2017). Antagonism of prostanoid receptor(s) did prevent injury-induced E₂ but in a sex-specific manner. Antagonism of EP-3 in males prevented E₂ induction at 24 h, and antagonism of EP-4 prevented the induction in females at 6 h post-injury. Given this data, we believe that PGE₂ may bind to EP-3 in males and EP-4 in females to achieve the robust induction of E₂ that has been well-documented following penetrating brain injury (Pedersen et al., 2018; Pedersen and Saldanha, 2017).

Nuclear factor Kappa B (NFκB).

NFκB is a family of inducible transcription factors and gene regulators that play a variety of evolutionarily conserved roles in the immune system, specifically in the control of inflammation (Hayden and Ghosh, 2004). In its inactive form, NFκB is located in the cytoplasm and is natively bound to its inhibitor, IκB (inhibitor of κB) (Yamamoto and Gaynor, 2004). In response to TBI and concomitant TNF-α release (Hayden and Ghosh, 2004; Xu et al., 1998), the IκB kinase complex activates, which leads to the phosphorylation of IκB-α and IκB-β (DiDonato et al., 1997; Mercurio et al., 1997; Regnier et al., 1997; Shih et al., 2015; Zandi et al., 1997). Once activated, NFκB binds to the κB site of the promoter, which allows transcription of genes involved in the inflammatory response. NFκB-mediated neuroinflammation is the driving force behind the secondary wave of degeneration, and thus inhibition of this pathway has been a target for decreasing damage to the brain. In the avian brain, expression of NFκB is upregulated when compared to non-injury controls at 2 and 24 hours following injury. Thus, NFκB expression following injury occurs prior to glial aromatase expression and thus is in the right time and place to be involved in the activation of glial aromatase (Cook et al., 2018). Future research is ongoing to determine if NFκB is both necessary and sufficient to upregulate glial aromatase alone.

Cytokines.

Ironically, there appears to be a chicken and egg approach to cytokines following TBI. Cytokines such as LL-1β and TNF-α are inductive cues for many of the inflammatory agents that later induce further production of neuroinflammatory products. Thus it is possible that the inflammatory process is driven at least partly by cells crossing the impaired blood–brain barrier in TBI and spinal cord injury alone (Chodobski et al., 2011; Sundholm-Peters et al., 2004; Thelin et al., 2018). However, the process is further modified by local cytokine generation by glial cells, CNS macrophages, and neuronal cells (Freidin et al., 1992; Helmy et al., 2011; Sébire et al., 1993; Thelin et al., 2018). As previously reviewed, following injury both astrocytes and microglial are both upregulated and they are steroid sensitive. Furthermore, cytokines, such as IL-6 and TNF-α and PGE₂, can all stimulate aromatase activity (Macdiarmid et al., 1994; Purohit et al., 1999; Zhao et al., 1996). Thus, we sought to investigate the consequences of specific cytokines that are upregulated following gliosis —namely, IL-1β and TNF-α on induced glial aromatase expression following TBI.

To determine if IL-1β and TNF-α signaling are necessary for the induction of aromatase, we tested the influence of a chronic penetrating injury on the expression of aromatase in wild type (WT) mice and knockout mice deficient in IL-1β receptors (IL1R KO) or TNF-α (TNF-α KO; unpublished Fig. 3). Aromatase mRNA was significantly higher in injured cortical samples compared to sham cortical samples for both WT and IL1R-KO mice, this effect was not apparent in the injured cortex of TNF-α KO animals, despite not differing in initial aromatase expression. These data suggest that TNF-α, but not IL-Ιβ signaling is necessary for the induction of aromatase following brain injury in mice (Figure 3). The mechanisms directly underlying the upregulation of aromatase by TNF-α are still understudied, however TNF-α is a key driver of aromatase promoter I.4-mediated expression in adipose tissue and this could be a possible site for regulation in the brain (Martínez-Chacón et al., 2018; Zhao et al., 1996). Future studies are necessary and ongoing to determine the molecular mechanisms underlying TNF-α induced glial aromatase expression and subsequent biosynthesis of estradiol following injury.

Figure 3.

Adult wild type C57BL/6 (WT) mice and either IL-1 receptor null (IL-1R KO) or TNFα (TNFα KO) were implanted with in-dwelling cannula directed towards the hippocampus. 10 days post-surgery, animals were sacrificed and total RNA was extracted from microdissections immediately adjacent to the cortical needle and the corresponding location in sham animals. Expression of aromatase relative to GAPDH was measured using quantitative PCR using primers specific for the aromatase mRNA sequence. Fold change in aromatase expression 10 days after (A) a unilateral penetrating injury to the hippocampus of male wild type adult mice, and (B) aromatase expression following the same injury in the hippocampus of wild-type, interleukin 1 receptor knockouts (IL-1RKO) and TNFαKO knockouts. Aromatase expression is increased in the hippocampus and also in the distal cerebellum (A). In the hippocampus itself (B) aromatase is upregulated in WT and IL-1RKO, but not in TNFαKO mice. * denotes a significant difference at P <0.05

Given the obvious differences in circulating hormone levels between males and females, and also the changes in hormone levels during menopause, it is not surprising that there are sex differences in the inflammatory response following injury. In fact, the neuroprotective effects of estrogen likely contribute to sex differences in neurodegenerative disease, ischemic stroke and traumatic brain injury. Various groups have shown sex differences in the induction of glial aromatase following injury. In songbirds and in mammals, inhibition of aromatase using fadrozole increases damage after traumatic brain injury in males, an effect that can be prevented with estrogen replacement (Azcoitia et al., 2001; Wynne et al., 2008). In the zebra finch, there is a greater upregulation of aromatase mRNA expression that is more rapid in female than in males after damage to the cerebellum (Mirzatoni et al., 2010). A similar female-biased sex difference in glial aromatase induction was also observed after injury to the zebra finch entopallium (Saldanha et al., 2013). There are also sex differences in aromatase induction specifically in astrocytes. Astrocytes isolated from female neonatal rat cortex have greater aromatase activity and expression than astrocytes from males. Female astrocytes are protected from cell death in response to oxygen glucose deprivation compared to males, and this sex difference is abolished by aromatase inhibition (Liu et al., 2007).

The neuroprotection in females is conferred in part by estrogens modulation of the immune response (Czlonkowska et al., 2006). In astrocytes specifically, estrogen and selective estrogen receptor modulators influence cytokine levels (Cerciat et al., 2010; Dodel et al., 1999). While not many studies have looked at sex differences in the timeline of cytokine induction after injury, in songbirds females have a greater up upregulation of IL-1β, but not IL-6, that occurs prior to the robust upregulation of aromatase that also has a female bias (Saldanha et al., 2013). Together paired with what we know about COX, PGE₂, and EPs, these data suggest that males and females differ in their overall response to injury.

Aromatization-dependent effects on inflammation

There is still an unresolved paradox with respect to the immunomodulating role of estrogens. On one side, we recognize inhibition and suppression of inflammation in several animal models of chronic inflammatory diseases. On the other hand, we realize the immunosupportive role of estrogens in trauma/sepsis and the proinflammatory effects in some chronic autoimmune diseases in humans (Straub, 2007).

Estrogen Receptors.

In animals subjected to trauma and hemorrhage, a general inflammatory condition, Estrogen Receptor beta (ERβ) mRNA expression was increased, whereas Estrogen Receptor (ERα) expression was decreased (Schneider et al., 2000). These studies suggest inflammation-dependent up-regulation of ERβ relative to ERα in mammals (Straub, 2007). In zebra finches, both males and females had elevated ERα and ERβ at multiple time points following injury, but no change in G protein-coupled estrogen receptor (GPER1) was detected at any time point (Pedersen and Saldanha, 2017). Using pharmacological antagonists, the role for ERα vs ERβ was examined on neuroinflammation and response to injury. Antagonism of ERα, but not ERβ resulted in exaggerated neural PGE₂ levels, suggesting that ERα mediates the anti-inflammatory action of estradiol (Pedersen and Saldanha, 2017).

Nuclear factor κB.

As stated previously, NF-κB is an important factor in proinflammatory signaling, however it can also interact with ER pathways (McKay and Cidlowski, 1999; Straub, 2007). In rodents, high levels of E₂ block LPS-induced DNA binding and transcriptional activity of the p65 subunit of NF-κB by preventing its nuclear translocation (Ghisletti et al., 2005). Furthermore, ERα impairs TNF-induction of IL-6 by preventing binding of c-rel and, to a lesser extent, RelA proteins to the NF-κB site on the IL-6 promoter (Galien and Garcia, 1997; Kurebayashi et al., 1997; Straub, 2007). Following damage to the brain, transient middle cerebral artery occlusion in vivo, substantial apoptosis and inflammatory responses were observed, including IκB phosphorylation, NF-κB activation, and iNOS overexpression (Wen et al., 2004). In this model, E₂ treatment (short-term pregnancy levels) produced strong protective effects by reducing infarct volume and neuronal apoptosis by inhibiting NF-κB activation and iNOS overexpression (Straub, 2007; Wen et al., 2004). Thus, high levels of E2 produced following injury inhibit NF-κB activation, reducing, therefore, the proinflammatory signaling that is activated immediately after injury.

Cytokines.

Aromatase decreases expression of TNF-α and IL-1β following injury in an E2 dependent manner (Pedersen et al., 2016). Most effects were apparent in both sexes except that E2 replacement decreased PGE₂ levels in females, but its effect in males was not significant (Pedersen et al., 2016). Taken together, the data strongly suggest that injury-induced E₂ synthesis is a potent modulator of neuroinflammation associated with brain damage from a penetrating brain injury in the finch brain. Interestingly, although similar in general, we found some intriguing differences in the anti-inflammatory effects of E₂ across sexes. When aromatase is inhibited, females have a prolonged elevation of TNF-α and IL-1β, whereas only TNF-α remains high in males. In partial agreement, E₂ administration lowers TNF-α in males and IL-1β in females but not vice versa (Pedersen et al., 2018).

Aromatization-dependent effects on degeneration.

It is very likely that injury induced astrocytic aromatization increases local levels of E₂ around the injury site (Pedersen et al., 2016; 2018). We have excellent evidence supporting multiple modes of neuroprotection associated with peripheral changes in E₂ including, but not limited to: apoptotic genes (D’Astous et al., 2006; Lisztwan et al., 2008; Singer et al., 1998), second messenger cascades (Mannella and Brinton, 2006; Quesada et al., 2008), neurotrophic factors (Buchanan et al., 2000; Fernandez–Galaz et al., 1997; McCarthy et al., 2002; Quesada and Micevych, 2004), mitochondrial function (Green et al., 2000; Prokai and Simpkins, 2007; Simpkins et al., 2005; Simpkins and Dykens, 2008), and inflammatory pathways (Arevalo et al., 2012; Cerciat et al., 2010). While specific evidence implicating these pathways in the role of injury-induced astrocytic provision of E₂ in neuroprotection is less complete, we have good reason to believe the multiple aspects of cell-turnover are affected as consequences of astrocytic aromatization.

The majority of this work has been accomplished in songbirds and rats. In zebra finches, inhibition of injury-induced aromatase increases apoptosis relative to controls (Wynne and Saldanha, 2004). This increase in apoptosis is correlated with increased gliosis (Wynne et al., 2008) and collectively contributes to larger infarct volumes following aromatase inhibition (Saldanha et al., 2005; Wynne et al., 2008; Wynne and Saldanha, 2004). This pattern is matched well by more recent studies using global cerebral ischemia in rats where inhibition of injury induced aromatase via the administration of antisense oligonucleotides increased neuronal death relative to mis-sense probes (Zhang et al., 2014).

These exacerbations in neural damage are effectively ameliorated by concomitant E₂ provision. Correspondingly aromatase inhibition with co-administered E₂ decreases apoptosis, gliosis and infarct volume in zebra finches (Saldanha et al., 2005; Wynne et al., 2008). In association, E2 also increases indices of regeneration including the number of new cells around neural injury in songbirds (Walters et al., 2011). Collectively, it appears that injury-induced E2 provision promotes various indices of cellular preservation in songbirds and rodents and songbirds and contribute to the preservation and perhaps repair of damaged tissue.

A novel role for brain aromatase as a central protectant from peripheral threat

The last two decades have redirected interest in E₂-dependent physiology into areas of neuroprotection and neural plasticity. As described above, multiple investigations using different species and several different models of neurotrauma have consistently reported an upregulation of aromatase, perhaps fueled by inflammatory cascades, whereby locally derived E₂ is then available to modulate cellular turnover and promote neuroprotective mechanisms. We end however, with some very recent observations that perhaps even further widen our appreciation of neural aromatization as a watchful and rapidly responsive system that protects fragile neural tissue from potential damage due to signals from the periphery.

Given the potent inductive role of inflammatory cascades on neural aromatization (Duncan and Saldanha, 2011; Pedersen and Saldanha, 2017), we tested the possibility that peripheral increases in inflammation may likely modulate neural aromatase expression. Indeed, two hours post intra-peritoneal LPS administration, subjects showed decreased locomotion and elevations in the neural expression of TNFα and IL-1β, and had lost more weight as measured 24hr later. At the latter time point however, neural cytokine expression had returned to baseline, LPS-subjects were indistinguishable from controls in their locomotor activity, and peripheral injections of LPS increased aromatase enzyme activity in the brain relative to controls (Pedersen et al., 2018). These data suggest the possibility that the upregulation of aromatase observed 24hr post-LPS may be responsible for the return of neural cytokine expression to baseline and the observed decrease in sickness-like behavior, and this hypothesis is currently under investigation in our laboratory.

Of particular interest was the observation that this increase was not due to glial aromatase, but rather an increase in the number of aromatase-expressing neurons. This observation suggests the possibility that neurotrauma and peripheral infections may have cell-specific effects on aromatase expression. The nature of this difference remains to be understood. Remarkably, similar increases in neural aromatase have also been reported following inflammation during human development. Children between the years of 1 and 9 years and who suffered an inflammatory event prior to death have higher expression of aromatase in the cerebellum relative to those who did not experience an inflammatory event (Wright et al., 2019). Also in keeping with the data presented on brain injury in songbirds, the authors report a positive correlation between PGE₂ and aromatase expression in this postmortem tissue. While the cellular source of these findings remains mysterious, the results point towards a tight association between inflammatory signals, both peripheral and central, in the regulation of neural aromatization. This neural aromatization appears to have evolved as a strong protectant across vertebrates.

Future Directions

The pluripotent role for central estrogens appears to now include an obligatory interaction with the innate immune system. This interaction seems to involve the normal CNS during development (Amateau and McCarthy, 2004; Garcia-Segura and McCarthy, 2004; McCarthy et al., 2002; Wright et al., 2019), but also following a range of physical insult to the brain. More broadly however, inflammatory signals can upregulate the expression of aromatase with little or no trauma to the CNS. Specifically, increases in inflammation caused by the systemic or central administration of inflammagens results in a dramatic upregulation in aromatase expression in the songbird brain (Duncan & Saldanha, 2011; Pedersen et al., 2018). Interestingly, this upregulation apparently occurs in astrocytes following central administration of PHA (Duncan & Saldanha, 2011) but is neuronal following peripheral LPS administration (Pedersen et al., 2018). This differential pattern suggests at least two novel aspects of the interaction between neuroinflammation and neurosteroidogenesis. Firstly, the cell-specific signals that result in neuronal and glial aromatase expression appear to be different. Secondly, the observed increase in neural aromatization following a peripheral threat suggests a novel sentinel and rapidly responsive aspect of neurosteroidogenesis. We hypothesize that neural aromatization may function to protect vulnerable neural circuits from impending, in addition to actual threat. In other words, the upregulation of neural aromatase following peripheral inflammation may reflect a protective mechanism where the vulnerability of the CNS is preemptively mitigated via estrogen synthesis upon detection of the peripheral threat. Neuroinflammation following peripheral activation of the immune system may provide one mechanism whereby the anti-inflammatory functions of central E₂ synthesis can protect against the potential compromise of neural circuits following peripheral infection (Pedersen et al., 2018; Pedersen & Saldanha, 2018). While much work remains to be done toward understanding this complex interaction more fully, there remains little doubt that neurosteroidogenesis appears to be a powerful mechanism of neuroprotection in the vertebrate brain.

Highlights.

Estrogens are established effectors of sex-specific neural circuits.

Astrocytic aromatization is upregulated in response to a broad range of brain damage.

Estrogens synthesized by astrocytic aromatization potently inhibit neuroinflammation.

Neural aromatization may also protect the brain against peripheral inflammation and infection.