Mechanisms of gender-linked ischemic brain injury

Abstract

Biological sex is an important determinant of stroke risk and outcome. Women are protected from cerebrovascular disease relative to men, an observation commonly attributed to the protective effect of female sex hormones, estrogen and progesterone. However, sex differences in brain injury persist well beyond the menopause and can be found in the pediatric population, suggesting that the effects of reproductive steroids may not completely explain sexual dimorphism in stroke. We review recent advances in our understanding of sex steroids (estradiol, progesterone and testosterone) in the context of ischemic cell death and neuroprotection. Understanding the molecular and cell-based mechanisms underlying sex differences in ischemic brain injury will lead to a better understanding of basic mechanisms of brain cell death and is an important step toward designing more effective therapeutic interventions in stroke.

1. Introduction

Clinical and epidemiological evidence suggests that stroke risk and outcome are sexually dimorphic, and emerging experimental data suggest that mechanisms of ischemic cell death may play differential roles in male vs. female cell death (Vagnerova et al., 2008). It is well established that women during their reproductive years enjoy protection from cardiovascular disease, including stroke, relative to age-matched men. This difference has been observed across cultures, ethnic groups and socio-economic status, suggesting that sex is an independent and strong predictor of stroke risk and outcome (Hurn & Macrae, 2000). Sex differences in ischemic brain injury has been historically attributed to the protective effect of female sex hormones, estrogen and progesterone (Hurn & Brass, 2003). However, sex differences in stroke risk persist well beyond the menopause (Giroud et al., 1991; Sacco et al., 1998), and female sex is associated with favorable outcome after brain injury in newborn children (Donders & Hoffman, 2002; Ingemarsson, 2003; Lauterbach et al., 2001), suggesting that sex differences in brain injury may not be entirely related to the influence of ovarian sex hormones.

To explore the basis of gender-linked ischemic brain injury, we have evaluated animal models that mimic the pathology of human stroke. We observed very early that the amount of brain tissue damage after experimental stroke is greater in male vs female rats (Alkayed et al., 1998), suggesting that sensitivity to cerebral ischemia, i.e. neural tissue damage once an ischemic event has occurred, is sex-specific (Fig. 1). We have replicated this initial observation in animal strains with pathological conditions known to be risk factors for stroke in humans, such as animals with a genetic form of insulin-dependent diabetes (Sieber et al., 2001; Toung et al., 2000), non-insulin dependent diabetes (Vannucci et al., 2001), and hypertension (Alkayed et al., 1998). In each genetic strain and despite deleterious complications from diabetes or hypertension, females were less sensitive to cerebral ischemia relative to their age-matched male counterparts (Fig. 2). This observation was also confirmed by multiple groups using a variety of animal models and experimental conditions (Hall et al., 1991; Li et al., 1996; Payan & Conrad, 1977; Wise et al., 2001; Zhang et al., 1998). In addition to studies showing that ischemic outcome is sexually dimorphic, the incidence of spontaneous strokes in a genetic strain of stroke-prone spontaneously hypertensive rats (SHRSP) is also lower in females than males (Yamori et al., 1976).

Fig. 1.

Sex Differences in Ischemic Brain Injury. Ischemia was induced by MCA occlusion for 2 hours in age-matched male (M), female (F) and ovariectomized (O) female rats. Infarct size was measured in brain sections by TTC staining at 22 hours of reperfusion and expressed as a percentage of ipsilateral cerebral cortex (CTX) or caudoputamen (CP). * Different from M and O (P < 0.05) (Alkayed et al., 1998).

Fig. 2.

Sex differences in Ischemic Brain Injury are Unaffected by Diabetes and Hypertension. Hypoxia-ischemia (HI) was induced in male and female db/db mice, a genetic model of Type II diabetes, by combined common carotid artery ligation and hypoxia, and tissue damage assessed by hematoxylin and eosin staining. Focal cerebral ischemia was induced in age-matched male and female stroke prone spontaneously hypertensive rats (SHR-SP) rats by MCA occlusion for 2 hours, and infarct size was measured at 24 hours by triphenyl-tetrazolium chloride (TTC).

In summary, these studies suggest that male animals, unlike females, must cope with increased sensitivity to ischemic stress. Below we discuss the basis of this sex difference in sensitivity to ischemia and its potential mechanisms and cellular substrates.

2. Estrogen and the estrous cycle

In female animals, brain injury is influenced by the stage of the estrous cycle. Proestrus, the part of the cycle with high endogenous estrogen production, is associated with smaller infarct size after experimental stroke induced by middle cerebral artery occlusion (MCAO). In contrast, metestrus with low estrogen production is associated with large damage (Carswell et al., 2000). It is therefore apparent that both gender and stage in the estrous cycle influence the outcome of cerebral ischemia in the rat. These observations form the basis of the concept that female sex hormones influence stroke sensitivity and contribute to sex differences in outcome from cerebral ischemia.

To test the hypothesis that greater neuroprotection in females versus males is due to female sex hormones,we have evaluated low-hormone status by surgical ovariectomy in sexually mature young adult female animals. We found that ovariectomized females sustained larger ischemic damage relative to gonad-intact females, and infarct size in ovariectomized females was not different from males (Fig. 1) (Alkayed et al., 1998). We also examined sex-linked differences in experimental stroke outcomes in reproductively senescent animals, when ovarian function in females has naturally abated (Alkayed et al., 2000). We found that reproductive senescence in females was associated with larger infarct after MCAO, and that sex differences in infarct size observed in young animals disappeared in reproductively senescent rats, suggesting that female gonadal steroids are protective against ischemic brain injury.

To determine if hormone replacement would restore protection against cerebral ischemia in ovariectomized and reproductively senescent females, we implanted rats subcutaneously with pellets containing 25 μg 17β-estradiol (E2) for 7 days, which restores physiological plasma levels of E2. E2 is the principal and most potent mammalian estrogen and accordingly has been the best studied of the estrogens. Seven days after implantation, rats were subjected to 2 hours of MCAO, and infarct size was measured after 24 hours of reper-fusion. Infarct size was significantly reduced in both ovariectomized (Rusa et al., 1999) and reproductively senescent female rats (Alkayed et al., 2000) replaced with 17β-E2. The neuroprotective effect of acute and chronic exogenous 17β-E2 replacement has been extensively studied across different laboratories, injury models, animal breeders and genetic strains (Hurn & Macrae, 2000). The steroid is protective in both sexes in vivo and reduces neuronal cell death in vitro (Alkayed et al, 2001a). It should be noted, however, that in some studies where the MCA was permanently occluded, estrogen has been shown to exacerbate ischemic damage (Macrae & Carswell, 2006).

3. Role of estrogen receptors: Classical and non-classical

3.1. Classical estrogen receptors

E2 is a pleiotropic hormone with multiple effects and mechanisms of action. We demonstrated that pharmacological blockade of estrogen receptor (ER) using ICI182,780 exacerbates ischemic brain injury in female mice (Sawada et al., 2000), suggesting that the neuroprotective effect of E2 is mediated in part via its nuclear receptors (Fig. 3). At present, there are two known classical ER subtypes, α and β (Kuiper et al., 1996; White et al., 1987), which share homology in DNA-binding domains and potentially activate the same transcriptional elements. Anatomical distributions of both receptors overlap in some brain regions but are quite distinct in others. The highest concentrations of receptors are in areas involved in reproduction, such as the hypothalamus; however, mRNA encoding ERs have been identified in a wide variety of other brain regions, including the cerebral cortex. ERβ in neocortex is constitutively expressed and can be detected throughout life, while neocortical ERα expression is developmentally regulated. ERα is expressed in the cerebral cortex at high levels during developmental periods associated with cortical differentiation (Shughrue et al., 1997) and is inducible in adult cerebral cortex by cerebral ischemia (Dubal et al., 1999).

Fig. 3.

E2 induces neuroprotective genes after ischemia by classical and non-classical mechanisms. In the classical mechanism, E2 binds to either ERα or ERβ receptor and acts as a transcription factor that directly binds to estrogen response elements (EREs) in neuroprotective gene promoters. In the non-classical mechanism, E2 binds to a putative membrane estrogen receptor, resulting in rapid activation of signaling cascades (shown in pink). Activation of signaling cascades leads to the phosphorylation and activation of transcription factors, including STAT3 or Creb (shown in blue), which bind to their cognate response elements (non-ERE sites) in neuroprotective gene promoters.

To determine which receptor subtype may mediate the neuroprotective effect of E2, we subjected mice with ERα gene deletion (ERα Knockout, ERαKO) to transient MCAO and compared infarct size to mice with intact ERα (wild type, WT). We found that ERα gene deletion was not associated with increased cortical infarct size in either male or female mice. In female ERKO mice, ERα gene deletion was paradoxically associated with decreased infarct size, which may be related to enhanced brain tissue perfusion in ERαKO mice (Sampei et al., 2000a). In permanent cerebral ischemia, E2 reduced cortical infarct size in WT and ERβ KO mice, but not in ERαKO mice, suggesting that E2 may signal through ERα to reduce stroke damage in this model (Dubal et al., 2001). In a transient global ischemia model, a selective ERβ but not ERα agonist reduced hippocampal ischemic damage in ovariectomized mice (Carswell et al., 2004), suggesting that different receptor subtypes mediate protective effects in response to different types of injury as well as different brain regions.

However, 17β-E2 may exhibit neuroprotective effects without engaging its nuclear receptors. For example, non-feminizing estrogen-like compounds, with reduced or no binding to either ERα or ERβ, are protective against ischemia in vitro and in vivo, presumably via their antioxidant properties (Simpkins et al., 2004). This effect, however, is expected to play a role at high, supraphysiological levels of plasma E2. Micro-molar concentrations of 17β-E2 reduces reactive oxygen species (ROS) and the associated neuronal damage induced by FeSO4 in the presence of the ER antagonist, tamoxifen (Culmsee et al., 1999). Similarly, E2 prevents intracellular peroxide accumulation and lipid peroxidation by inactivating ROS that could otherwise oxidize lipoproteins (Keller et al., 1997; Vedder et al., 1999). Furthermore, there is evidence for a neuroprotective action of 17β- and 17α-E2 both in vitro (Behl & Manthey, 2000) and in vivo (Simpkins et al., 1997), suggesting that at these doses, E2 may exert neuroprotective effects independently of its receptors.

4. Non-classical estrogen receptors

Recent suggest that E2 may exert its neuroprotective action via so- called rapid effects, mediated through a currently unidentified, membrane-associated receptor (Fig. 3) (Kelly et al., 2003; Toran-Allerand et al., 1999; Toran-Allerand et al., 2002). Indeed, ultrastructural studies reveal ERα and ERβ localization to extra-nuclear regions (McEwen et al., 2001). In addition, a membrane estrogen receptor, GPR30, has now been identified. GPR30 is a G-protein coupled membrane-associated receptor that rapidly activates signal transduction cascades in response to E2 (Funakoshi et al., 2006). GPR30 is expressed in brain regions sensitive to ischemia including the hippocampus, cortex and striatum (Carmeci et al., 1997; Filardo & Thomas, 2005; O'Dowd et al., 1998). However the role of GPR30 in neuroprotection has yet to be determined.

The neuroprotective effect of E2 is associated with the rapid activation of ERK and AKT in vivo (Choi et al., 2004; Honda et al., 2001; Singh, 2001; Wang et al., 2006) and activation of src-family tyrosine kinases, p21(ras)-guanine nucleotide activating protein and ERK, in vitro. The protection afforded by E2 is dependent on ERK phosphorylation as E2-induced neuroprotection is abolished by the ERK (MEK) inhibititor, PD98059 (Singer et al., 1999; Wu et al., 2005). E2 is also associated with rapid phosphorylation of cAMP response element-binding protein (CREB) (Wade & Dorsa, 2003), which can be blocked by ER antagonist, ICI 182,780, and by the mitogen-activated protein kinase (MAPK) /ERK kinase-1 inhibitor PD98059. These data suggest that E2 activation of CREB requires ERK/MAPK signaling (Mize et al., 2003). Interestingly, signaling involving phosphorylation of CREB and MAPK can be elicited in the absence of either one of the classical receptors, but not in ER double-knockouts (Abraham et al., 2004).

In support of a membrane-associated receptor mediated mechanism, we demonstrated that transfected ERα localizes to neurites in cultured cortical neurons and that neurite-localized ERα activates MAPK in response to estradiol (Xu et al., 2003). We also determined that membrane associated receptors are capable of direct activation of transcription via estrogen response element (ERE) (Xu et al., 2004).

5. Mechanism of estradiol-mediated neuroprotection

E2 uses a variety of genomic and non-genomic mechanisms to exert its protective action. Non-genomic effects are exemplified by a variety of cytoplasmic signaling cascades, such as the activation of protein kinase C, phosphatidylinositol-3-OH kinase, protein kinase A/cAMP (Gu & Moss, 1996; Kelly et al., 1999) steroid receptor coactivator (src) and intracellular signaling that activates ion channels, neurotransmitter receptors, and enzymes. These mechanisms may be critical to E2's protection in experimental stroke (Segars & Driggers, 2002).

The genomic effects of E2 are brought about either via direct ER binding to a consensus DNA sequence, the ERE (Klinge, 2000), or indirectly by interacting with and influencing DNA binding affinity of other transcription factors, such as nuclear factor-kappa B (NF-κB) and activator protein 1 (AP1) (McKay & Cidlowski, 1999; Paech et al., 1997) SP1 (Castro-Rivera et al., 2001) and Jun/ATF-2 (Sabbah et al., 1999). Functional EREs have been identified in genes implicated in normal brain function and pathology (Garcia-Segura et al., 2001), including those encoding choline acetyltransferase (Miller et al., 1999), the GABA transporter GAT-1 (Herbison et al., 1995), α-adrenergic receptor (Lee et al., 1998), the opioid precursor preproenkephalin (PPE) (Zhu & Pfaff, 1995), the neuropeptides oxytocin and vasopressin (Harlan, 1988), somatostatin (Xu et al., 1998), galanin (Howard et al., 1997), glial fibrillary acidic protein (GFAP) (Stone et al., 1998b), brain-derived neurotrophic factor (BDNF) (Sohrabji et al., 1995), transforming growth factor-alpha (TGF-α, (El Ashry et al., 1996), cyclin D1 (Sabbah et al., 1999), vascular endothelial growth factor (VEGF) (Mueller et al., 2000) and 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMG CoA) gene, which is regulated by estrogen via a variant of ERE with base pair mismatches (Croce et al., 1999).

Non-genomic and genomic effects of estradiol are not mutually exclusive. Supporting this idea, E2 induces many genes which do not contain EREs (O'Lone et al., 2004). Therefore E2 induces non-genomic or rapid activation of signaling cascades that result in the activation of other transcription factors, which may induce neuroprotective genes. For example, we have recently reported that E2 replacement increases the phosphorylation of the transcription factor, signal transducer and activator of transcription 3 (STAT3) after transient focal ischemia relative to ovariectomized female rats. The effect was observed in the cytosolic fractions where the total amount of STAT3 was unaltered, suggesting rapid signaling events after an ischemic event. The protective effect of E2 to reduce infarct volume appeared to be mediated via phospho-STAT3 because pharmacological inhibition of P-STAT3 abolished the protective effect of estradiol on infarct size (Dziennis et al., 2007). Furthermore, P-STAT3 was observed in cells that were also immunoreactive for bcl-2 or MAP-2 suggesting that P-STAT3 is expressed in surviving neurons.

To this end, we demonstrated that E2 upregulates two neuroprotective genes which lack promoter estrogen response elements; bcl-2 and CART. Bcl-2 is well established for protective effects in brain injury of all types (for review see (Graham et al., 2000; Mattson et al., 2000). E2 up-regulates bcl-2 expression in several brain regions of cycling female rats (Garcia-Segura et al., 1998). E2 increases the expression of bcl-2 mRNA and protein in the peri-infarct region after transient MCA occlusion in rats, accompanied by a decrease in infarct size (Alkayed et al., 1998; Alkayed et al., 2001b; Dubal et al., 1999). Moreover, bcl-2 overexpressing mice are protected from the exacerbation of cerebral ischemic injury ordinarily caused by ovariectomy in wild type female mice (Alkayed et al., 2001b). In the setting of ischemia, the ability of E2 to enhance bcl-2 expression may be through the multiple putative STAT binding sites and/or the CREB binding site in the absence of EREs (Teixeira et al., 1995). Interestingly Meller and colleagues found that CREB binding protein (CBP) rather than CREB, bound to the CRE in an in vitro model of ischemic preconditioning (Meller et al., 2005).

Using DNA microarrays, we identified a gene product, cocaine- and amphetamine-regulated transcript (CART), that is highly induced by E2 in the ischemic cerebral cortex and exhibits potent neuroprotective effects in models of neural injury both in vitro and in vivo. The cart gene encodes a neuropeptide that is implicated in food intake, drug addiction, and the neuroendocrine response to stress (Jaworski & Jones, 2006; Koylu et al., 2006; Kuhar et al., 2002; K. G. Murphy, 2005). We confirmed that E2 increased CART mRNA and protein in cortical neuronal cultures after OGD in the cortex after MCAO. CART co-localized with the neuronal marker NeuN but not the glial marker GFAP after MCAO, suggesting a role for CART in neuroprotection. To test the hypothesis that CART is neuroprotective, CART was administered to primary cortical cultures prior to OGD. CART administration significantly reduced OGD-induced DNA damage and cell death in primary cultured cortical neurons (Xu et al., 2006). CART peptide administration also significantly reduced cortical infarct size after MCAO, demonstrating a protective effect of CART against cerebral ischemia, in vivo. CART, like bcl-2, lacks an ERE, yet contains a STAT3 and a CREB binding site (Lakatos et al., 2002). We found CREB, but not STAT3 bound to the cart promoter after MCAO (Xu et al., 2006).

The mechanism of protection by E2 is multi-faceted and involves multiple brain cell types, including neurons, astrocytes, microglia and oligodendrocytes, as well as vascular and blood-borne cells (Azcoitia et al., 1999a; Gudino-Cabrera & Nieto-Sampedro, 1999; Mor et al., 1999). Depending on cell and injury type, E2's protection may involve any number of mechanisms triggered by E2. E2's reduction of tissue injury is in part related to the steroid's vasodilator actions and ability to promote optimal endothelial function (Mendelsohn & Karas, 1994; White et al., 1995). Cerebral blood flow is higher in premenopausal women than in men or in older women (Davis et al., 1983; Shaw et al., 1984), and estrogen influences blood flow during ischemic stress and reperfusion (Alkayed et al., 1998; Hurn et al., 1995; Pelligrino et al., 1998).

Sex-specific differences in vascular reactivity have been described in the periphery (Li & Duckles, 1994; Sakuma et al., 2002) and in the cerebral circulation (Geary et al., 1998). The effects of estrogen on vasodilation are mediated through changes in endothelium-dependent nitric oxide (NO), prostanoids and endothelial-derived hyperpolarizing factor (EDHF). In addition, estrogen deficiency, either as part of the estrus cycle or long-term after ovariectomy, alters endothelium-dependent nitric oxide (NO)-and prostacyclin-mediated responses of mesenteric arteries in female rats (Liu et al., 2001). Although the identity of EDHF is not known, we have demonstrated that estrogen suppresses soluble epoxide hydrolase, an enzyme involved in the metabolism of an endothelium-dependent hyperpolarizing factor (EDHF) candidate, namely epoxyeicosatrienoic acid (EET) (Koerner et al., 2008). Suppression of soluble epoxide hydrolase would increase EETs, leading to enhanced vasodilator capacity in response to vascular occlusion (Zhang et al., 2008), and may account for the increase in blood flow during MCAO in female rats (Alkayed et al., 1998).

Others have shown that in females, endothelium-dependent hyperpolarizing factor (EDHF)-mediated dilations of the middle cerebral artery (MCA) are significantly attenuated as compared with that of the male (Golding & Kepler, 2001). Dilations in ovariectomized females are enhanced to similar levels observed in the intact males and subsequently lost after chronic E2 replacement, suggesting that this effect is mediated by E2. In contrast to findings in cerebral vessels, estrogen appears to potentiate EDHF-mediated dilations in peripheral vessels (McCulloch & Randall, 1998; White et al., 2000), in agreement with other reports that the mechanism for the EDHF response is distinct in the periphery vs cerebrovasculature (Golding et al., 2002).

Other proposed mechanisms of protection by E2 include the suppression of excitotoxicity (Singer et al., 1996; Smith et al., 1987; Weaver et al., 1997), apoptosis (Alkayed et al., 2000; Alkayed et al., 2001), β amyloid toxicity (Goodman et al., 1996; Shi et al., 1998), edema formation (Roof et al., 1993; Roof et al., 1994), antioxidant production (Ayres et al., 1998; Bruce-Keller et al., 2000; Culmsee et al., 1999; Sawada et al., 1998) and regulation of growth factors (Gollapudi & Oblinger, 1999; Toran-Allerand, 1996). Finally, E2 is anti-inflammatory (Hunt et al., 1997; Ray et al., 1997; Salem et al., 2000; St Clair, 1997; Vegeto et al., 2001) in cerebral ischemia, reducing the number of active microglia (Lei et al., 2003), decreasing intravascular leukocyte adhesion and migration into brain (Santizo et al., 2000), and suppressing endothelial expression of adhesion molecules (Mori et al., 2004; Nathan et al., 1999). Moreover, E2 can suppress reactive gliosis and expression of the inflammation promoting transcription factor NF-kB after cerebral injury (Wen et al., 2004). Taken together, it appears that E2 can restrict cellular damage in the pro-inflammatory milieu of ischemia/reperfusion.

6. E2, neuroregeneration and recovery

E2 promotes recovery from CNS injury by stimulating neural cell proliferation and synaptogenesis and via modulation of synaptic connectivity, leading to axonal sprouting and regeneration (Garcia-Segura et al., 2001; Jones, 1993). These effects are, at least in part, mediated via E2's effects on expression of growth factors and neurotrophins, such as NGF, BDNF and NT-4 (Jezierski & Sohrabji, 2000; Wang et al., 2000), resulting in changes in expression of growth components, such as structural proteins and adhesion molecules. E2 may also affect cellular proliferation and neurite elongation and differentiation by directly regulating the transcription of genes required for growth; such as regulation by estrogen of the expression of tau microtubule-associated protein (Ferreira & Caceres, 1991), the growth-associated protein 43 (GAP-43), neuromodulin (Shughrue & Dorsa, 1993), neurofilament proteins (Scoville et al., 1997), synaptic proteins, such as synaptophysin, syntaxin, and spinophilin (Teter et al., 1999), and structural lipoproteins, such as apolipoprotein E (apoE) (Stone et al., 1998a). Finally, the signaling pathways for E2 and neurotrophins converge on the MAPK cascade, which includes activation of B-Raf and extracellular signal-regulated kinase (ERK) (Singh et al., 2000). This common pathway may explain the ability of E2 and neurotrophins to regulate the same broad array of cytoskeletal and growth-associated genes involved in neurite growth and differentiation. The interaction of E2 with growth factors may also prevent cell death by preventing initiation of apoptosis (Gollapudi & Oblinger, 1999). However, as mentioned above, E2 can also prevent neuronal cell death by directly regulating expression of effectors of cell death, such as members of the bcl-2 family of cell death-associated proteins (Alkayed et al., 2001b).

In adult brain as well as during brain development, ER s co-localize with, and reciprocally regulate expression of, neurotrophins and their cognate receptors, such as the high- (trkA) and low-affinity (p75) receptors for nerve growth factor (NGF). The net effect is to increase the sensitivity of these cells to neurotrophins (Sohrabji et al., 1994; Toran-Allerand et al., 1999). E2's effect on BDNF expression is region-specific, as E2 increases BDNF in cortex (Gibbs, 1999), while decreasing BDNF in hippocampus (Woolley, 1999). The gene encoding BDNF contains a sequence similar to the canonical ERE (Sohrabji et al., 1995).

E2's neuroregenerative effects may be mediated via interaction with other growth factors, such as insulin-like growth factor-1 (IGF-1) (Garcia-Segura et al., 2006). The trophic effects of IGF-I are mediated by IGF-I receptor, a member of the growth factor tyrosine kinase receptor family that signals through the phosphoinositol-3 kinase pathway and the MAPK cascade. E2 and IGF-I co-operate in hippocampus to protect neurons from kainic acid-induced cell death (Azcoitia et al., 1999b). IGF-I receptor blockade prevents E2-induced neuroprotection while IGF-I induced protection is blocked by ICI 182,780. E2 can also protect neurons via growth factors indirectly, via its interaction with growth factor-enhancing proteins, such as activin, a member of the transforming growth factor β (TGFβ) superfamily (Trudeau et al., 1996), which is important for the neuroprotective action of basic fibroblast growth factor (bFGF) (Tretter et al., 2000). Similarly, E2 increases IGF-1 levels by upregulating expression of IGF binding protein-2 (IGFBP-2) (Duenas et al., 1994).

7. Local estrogen production: Role of the P450 aromatase

Local brain E2 may also be relevant to neuroprotection. E2 levels in brain vary between males and females as early as the first day of life (Amateau et al., 2004). Mammalian E2 is naturally synthesized from testosterone by aromatization, a process that occurs not only in gonads, placenta and fat, but also in brain (Azcoitia et al., 2003; Naftolin, 1994). The role of local brain E2 formation on ischemic brain injury has been examined in mice with targeted deletion of P450 aromatase (CYP19A1), which are incapable of converting testosterone to E2. Mice with targeted disruption of exon 9 of the CYP 19A1 gene (aromatase knockout mice) sustained increased brain injury after experimental stroke compared to wild-type mice with intact P450 aromatase gene, suggesting an important role for P450 aromatase in protection from ischemic brain injury (Liu, et al., 2004b; McCullough et al., 2003). In addition to synthesis, local estrogen levels are also determined by metabolism. Estrogens are metabolized into catechol estrogens, 2- and 4-hydroxy estrogens, via the actions of P450 lA and lB. Both genes are expressed in brain (Muskhelishvili et al., 2001), and catechol estrogens have been detected in normal brain (Weisz & Crowley, 1986) and implicated in neuronal cell death (Desjardins et al., 1992).

Estradiol is not the sole hormone active in mediating outcome from ischemic brain injury. Progesterone and testosterone, as well as non-hormonal factors, have been shown to contribute importantly to sex differences in brain injury and recovery. Below, we discuss the contributions of these factors to the differences in ischemic brain injury between males and females.

8. Progesterone and cerebral ischemia

Progesterone has been less well studied in cerebral ischemia, but our studies suggest that the steroid is equipotent to E2 when administered in the correct therapeutic window (Fig. 4). We previously demonstrated that chronic progesterone administration reduces cortical infarct in middle-aged, reproductively senescent females, and protection is not related to effects of progesterone on cerebral blood flow (Azcoitia et al., 2003). Similarly, progesterone administration during reperfusion (Murphy et al., 2002), but not before the ischemia (Murphy et al., 2000), reduces ischemic injury in ovariectomized, young adult female rats. Finally, post-ischemic progesterone has also been shown to reduce lesion size and enhance functional recovery after transient MCAO in normotensive (Gibson & Murphy, 2004) and hypertensive male mice (Kumon et al., 2000) and rats (Chen et al., 1999; Jiang et al., 1996). While the neuroprotective mechanism is unclear, the steroid's membrane-stabilizing antioxidant actions may be relevant (Duenas et al., 1994). Progesterone is also known to modulate y-aminobutyric acid (GABA) receptor channel activity and expression (El Ashry et al., 1996) and attenuate excitatory neuronal responses (Eliasson et al., 1997), which may underlie progesterone's anxiolytic (Ferreira & Caceres, 1991) and antiepileptic (Frye & McCormick, 2000) properties.

Fig. 4.

Female Sex Hormones 17β-Estradiol and Progesterone are Protective against Ischemic Brain Injury. Ischemia was induced in ovariectomized female rats with and without hormone replacement for two hours by MCA occlusion, and infarct size was measured in the cerebral cortex at 24 hours by triphenyl tetrazolium chloride (TTC) and expressed as a percentage of contralateral cortex to account for edema.

9. Role of male sex hormones: Testosterone

Male sex is an acknowledged risk factor for stroke and cerebrovascular disease. In men, normal circulating testosterone levels range from 10 to 30 nM (Winters, 1999), whereas much lower levels (0.6 to 2.5 nM) are found in women (Burger, 2002). Although the adrenal gland can synthesize small quantities of testosterone, testicular Leydig cells are the primary source of androgens in men (Winters, 1999). Ordinarily, plasma luteinizing hormone (LH) interacts with Leydig cell surface receptors to regulate pulsatile testosterone release in a diurnal pattern, with peak levels occurring in the morning. However, other hormones, including prolactin, cortisol, insulin, insulin-like growth factor and E2, are known to influence circulating testosterone (Winters, 1999). In women, peripheral sites such as liver, skin, and adipose tissue provide approximately half of the circulating testosterone, whereas adrenal (25%) and ovarian sources (25%) produce the remainder (Burger, 2002). As members of the steroid superfamily, androgens originate from the cholesterol-derived precursor, pregnenolone. Both E2 and testosterone are derived directly from pregnenolone's by-product, androstenedione. In addition, androstenedione can be formed indirectly from the progesterone degradation product 17β-hydroxyprogesterone. The enzyme 17β- hydroxysteroid dehydrogenase (17 β-HSD) catalyzes this metabolism to testosterone and is also requisite for 17 β-E2 production. Testosterone can be converted to 17 β-estradiol via the aromatase, as described above, or metabolized to dihydrotestosterone (DHT) via the enzyme 5α-reductase, a step that then does not permit aromatization to estradiol. Testosterone interacts with a single known androgen receptor (AR) to initiate gene transcription. However, testosterone has also been shown to produce rapid vascular effects that clearly do not involve genomic mechanisms.

Data in animal models of cerebral ischemia testing effects of androgens are surprisingly few and contradictory. In male rat, castration and subsequent low testosterone is associated with decreased or no difference in histological damage after MCAO (Hawk et al., 1998; Toung et al., 1998). Testosterone administration increases infarct size after MCAO in male rats (Hawk et al., 1998), in part via exacerbation of glutamate neurotoxicity (Yang et al., 2002). In accordance with a neurotoxic role, Yang et al demonstrated that brief anesthetic exposure increases ischemic tolerance by suppressing testosterone production (Yang et al., 2005). In contrast, testosterone or DHT treatment reduces dentate gyrus neuronal loss after adrenalectomy (Frye & McCormick, 2000). In neuronal culture, testosterone appears in some studies to protect cells from oxidative stress, β-amyloid toxicity, and serum deprivation through an AR-dependent mechanism (Ahlbom et al., 1999; Ahlbom et al., 2001; Hammond et al., 2001; Pike, 2001). Further study is required to understand how androgens impact on sex differences in ischemic injury.

10. Hormone-independent mechanisms of protection and injury

New data from cultured cells in which sex steroids are removed from the media support the concept that cell death mechanisms can be sex-specific. Some molecular pathways of cell death or survival diverge, depending on the genetic sex of the tissue (defined by chromosomal gene content as female XX or male XY). For example, female dopaminergic neurons cultured from 14 day old embryos (E14) tolerate exposure to toxic dopamine concentrations at the LD50 level and survive twofold relative to male cells (Lieb et al., 1995). Similarly, female neurons from cortical plate or ventricular zone have greater longevity in culture than do male cells, and differentially express higher levels of phosphorylated kinases such as Akt (Zhang et al., 2003). Sensitivity to glutamate, peroxynitrate (ONOO⁻) and staurosporine in neuronal culture (E17) is sex specific, with male neurons being more susceptible to glutamate and ONOO⁻ than females. In contrast, response to oxidants such as H₂O₂ is not dependent on the sex of the cells (Du et al., 2004). We have recently found that sex specific sensitivity to neurotoxins is also evident in astrocytes. Astrocytes are important players in normal brain functioning (Harder et al., 1998), as well as the response to ischemic brain injury (Chen & Swanson, 2003; Swanson et al., 2004; Takuma et al., 2004). These cells are potentially important players in the regenerative and neuroprotective effects of sex steroids and in determining sex differences in synaptic connectivity and plasticity (Garcia-Segura et al., 1999). We have evaluated cell death resulting from oxygen-glucose deprivation (OGD) in sex-specific astrocyte cultures from rat pups and observed that female astrocytes are more resistant to OGD compared with male astrocytes, but sustain greater cell death when inflammatory mediators are combined with OGD compared to OGD alone. The aromatase inhibitor, Arimidex, abolishes sex differences in OGD-induced cell death (Fig. 5), suggesting that astrocytes isolated from neonatal cortex exhibit marked sex differences in sensitivity to OGD, in part due to enhanced aromatization and estradiol formation in female cells (Liu et al., 2004a; Liu et al., 2007). To verify sex-specific differences in P450 aromatase function in astrocyte cell death following OGD, we developed a novel method to establish sex-specific and genotype-specific single pup primary astrocyte cultures from wild-type (WT) and aromatase knockout (ArKO) mice (Liu et al., 2008). Consistent with data generated in sex-specific rat astrocytes (Liu et al., 2007), we found that female WT mouse astrocytes are more resistant to OGD than male WT mouse astrocytes. It is important to note that male sensitivity and female resistance to OGD are observed across species. Using the new technique developed for this study, we demonstrated that P450 aromatase gene deletion significantly increased cell death after OGD in female astrocytes and abolished the sex differences in sensitivity to OGD seen in WT female and male astrocytes (Fig. 6). These findings confirm that P450 aromatase is instrumental in mediating astrocyte survival following OGD and suggest that P450 aromatase is cytoprotective against ischemic brain injury. This is consistent with an earlier in vivo study of ischemia in which brain damage was greater in female ArKO mice compared to female WT mice (McCullough et al., 2003). These studies suggest that male and female brains may have inherently different susceptibility to ischemic stress. This is consistent with recent evidence suggesting that the brain begins to develop differently in males and females before sex hormones come into play, and that the brains of males and females differ not only in regions specialized for reproduction, but in other brain regions as well (Arnold et al., 2003).

Fig. 5.

Sex-specific astrocytic cell death. A. Female astrocytes are more resistant to cell death induced by oxygen-glucose deprivation. LPS, interleukin-1β (IL-1β), tissue necrosis factor (TNF)-1α and Arimidex exacerbate OGD-induced cell death in female astrocytes compared to OGD alone. * p < 0.05. B. Male astrocytes are more susceptible to cell death induced by oxygen-glucose deprivation. LPS, interleukin-1β (IL-1β), tissue necrosis factor (TNF)-1α and Arimidex did not alter OGD-induced cell death in male astrocytes. Adapted from Liu et al., 2007.

Fig. 6.

Sex differences in OGD-induced cell death in mouse astrocytes. WT female astrocytes are less susceptible to cell death compared to WT male astrocytes after OGD (^#p < 0.05). Cell death markedly increased in ArKO female astrocytes compared to WT female cells (*p < 0.05), but there was no significant difference in cell death between ArKO male astrocytes compared to WT male cells. Sex-specific responses to OGD were abolished in ArKO astrocyte cultures. Adapted from Liu et al., 2008.

11. Sex-specific mechanisms of ischemic cell death

Data from genetically engineered mice also suggest that molecular mechanisms of cell injury are not necessarily identical in male and female brain. When both sexes are studied, ischemic outcome in transgenic mice can be overtly gender-dependent, even when the gene of interest is not linked to sexual development (Loihl et al., 1999; McCullough et al., 2005; Sampei et al., 2000b). For example, it is well known that neuronal nitric oxide synthase (nNOS) plays an important role in intiating ischemic cell death. Nitric oxide cytoxicity through its rapid reaction with superoxide anion results in peroxynitrite formation and protein nitration (Dawson & Dawson, 1998). Genetic deletion or pharmacological inhibition of nNOS is neuroprotection in male animals (Huang et al., 1994), presumably by reduction of NO. However, we have recently observed that loss of nNOS activity in female knockouts or through pharmacological enzyme inhibition paradoxically increases histological infarction after MCA occlusion (McCullough et al., 2005). Another example of a potentially sex-specific cell death mechanism is the activation of the DNA repair enzyme, poly-ADP ribose polymerase (PARP). PARP is hyper-activated in the presence of ischemic DNA damage and enhances neuronal death after exitotoxic or ischemic insults. However, these data all arise from male mice or from cell culture of mixed sex origin (Eliasson et al., 1997; Goto et al., 2002). In females, loss of PARP activity appears to be deleterious. Female PARP knockouts or females treated with PARP inhibitors sustain remarkable exacerbation of ischemic damage after MCA occlusion, unlike their male counterparts (McCullough et al., 2005).

12. Sex differences in response to therapy

The presence of gender specific mechanisms of ischemic brain injury suggests that responses to neuroprotective therapy could be different in males and females. For example, gender differences were observed regarding the efficacy of the anti-oxidant tirilizad in the treatment of subarachnoid hemorrhage (Kassell et al., 1996), with males displaying greater response to therapy than females. Similarly, posttraumatic hypothermia attenuates traumatic brain injury and subsequent neuronal loss in males, but not in females (Suzuki et al., 2003). This effect may be related to the observation that hypothermia reduces cerebral spinal fluid (CSF) markers of excitotoxicity and oxidative damage after brain injury to a greater degree in males than females, suggesting that hypothermia may be more beneficial in males than in females (Wagner et al., 2004). These findings underscore the importance of stratifying by sexing pre-clinical data, with the hope of predicting differing responses to therapeutic agents and intervention strategies in men and women.

13. Conclusion

Stroke is a sexually dimorphic disease. Sex differences in stroke risk and outcome have been observed in humans of all ages and are present in experimental models of cerebral ischemia at the whole animal and cellular levels. Male and female sex steroids, originating from gonads or locally produced within brain, account for some, but not necessarily all, of these differences. New observations suggest that fundamental mechanisms of ischemic cell death play differential roles in male vs. female brain. Stroke remains a major cause of death and disability in both men and women, however results from major clinical trials testing available neuroprotective compounds against stroke injury remain disappointing. Understanding sex-specific mechanisms of ischemic brain damage is warranted if we are to develop more effective therapies for human stroke.