Brain Sex Matters: estrogen in cognition and Alzheimer’s disease

Brain estrogen synthesis

Estrogens are the primary female sex hormones and play important roles in female sexual development and regulation of the menstrual cycle. In addition to the well-established functions of estrogens in female reproduction, studies demonstrated significant non-reproductive functions of estrogens in the regulation of lipid and carbohydrate metabolism, skeletal homeostasis, the cardiovascular system, and the central nervous system (CNS) in both males and females [23; 24; 25; 26; 27; 28; 29].

Estrogen can be synthesized locally in the brain, which provides the unique importance of brain estrogen function. Studies showed that the brain is able to synthesize estrogen de novo from cholesterol [30]. Indeed, all enzymes required for the synthesis and metabolism of estrogen, and critical intermediate metabolites, were reported to exist in various regions of the human brain. In particular, aromatase, a key enzyme for the last step in the synthesis process, is widely expressed in the brain with a regional-specific manner in humans, rodents and primates [31; 32; 33; 34; 35]. For example, aromatase mRNA was first time reported in the rat hippocampus by Abdelgadir in 1994 [36]. Since then, more studies have showed that this enzyme is in fact released from hippocampal neurons and is functional in this part of the brain in both human and animals [37; 38; 39; 40; 41; 42]. Recent studies using positron emission tomography (PET) imaging with radiolabeled aromatase inhibitors have allowed a general map of the human brain aromatase distribution to be drawn. The highest level of aromatase was found in the thalamus, followed by the amygdala, preoptic area and medulla oblongata. Significant levels of aromatase were also observed in the temporal and occipital cortices, hippocampus, basal ganglia, cerebellum, pons and white matter [43; 44; 45; 46; 47; 48]. With respect to brain cell type specificity, aromatase in the human brain is mainly produced by neurons, but there is also an astrocyte subpopulation that constitutively expresses the enzyme [49; 50; 51]. As shown in figure 1, estradiol can either be produced from circulating testosterone by the local estrogen synthase aromatase, or be synthesized de novo from cholesterol, by neurons or astrocytes, but not by microglia and oligodendrocytes [41; 51; 52; 53; 54; 55; 56; 57]. It is important to notice that the neuron is the major site for brain estradiol synthesis. In the normal physiological condition, only a few astrocytes in the brain express aromatase while aromatase expression is increased in reactive astrocytes due to various brain injuries [58]. Although tissue-specific promoters as well as first exons have been reported in human aromatase gene [59; 60], it is unknown whether the usage of specific promoters of the aromatase gene can be translated into the level of brain region-specific expression. Such knowledge is important to develop selective aromatase modulators to regulate brain region-specific estrogen synthesis including in the human brain. Studies have shown that the cell type and brain region-specific aromatase expression may directly alter the local estradiol production and specific brain functions [61]. For example, estrogen locally produced in the synapse is reported to regulate synaptogenesis [40], neurotransmission [62] and synaptic plasticity [63; 64]. Furthermore, the synthesis, physiological roles, and regulations of brain estradiol are more or less distinct from ovarian estrogen. For example, mice with depletion of aromatase showed increased brain damage in an ischemic model compared to the wild type littermates or ovariectomized mice [65]. Depletion of aromatase in an animal model for AD caused early and more severe neuropathology than the ovariectomized control mice [5] and responded better to estrogen replacement treatment than in ovariectomized control mice [66]. In addition, studies showed that the fast effect of estradiol on neuronal spin density and synaptic transmission is more mediated through local synthesized estradiol in the developing and adult rat hippocampus [67].

Taken together, the brain produces its own estrogen while circulating estrogen and C19 steroid precursors (substrate for estrogen synthesis) can also penetrate through the blood-brain barrier into the brain and provide the essential substrates for estrogen synthesis in the CNS [68]. In the case of the human brain, three C19 androgen precursors, such as 16α-OH-DHEA, androstenedione and testosterone, have been identified [69; 70; 71]. Therefore, the level of brain estrogen might not be the same level as the blood estrogen which is often used as diagnostic marker in clinic [72; 73; 74; 75]. It is worth to point out that brain estrogen local synthesis and activity becomes more independent from circulating estrogen in non-reproductive females, such as surgical or natural menopausal women. For example, in an earlier study we showed that women with AD express lower activity of aromatase in the brain than that in age-matched women without AD [5]. Therefore, maintaining brain estrogen synthesis during ageing might be critically important for normal cognitive function, as well as for reduction of risk for AD in females

Brain estrogen biology and cognition

A lot of evidence has documented profound effects of estrogens on learning, memory, mood as well as neurodevelopmental processes [12; 76; 77; 78]. Although estrogen is a female hormone, increasing studies have shown that estrogen also plays important roles in the male brain partially due to the converting of testosterone to estrogen by brain aromatase [45; 46; 47; 79]. However, it is well documented that men and women have different types of cognitive function, and the sex differences in cognition are associated with brain estrogen and testosterone in the regions important for cognition, memory and mood such as the cortex, hippocampus and amygdala [80; 81; 82]. In addition, as reported by extensive studies, most adulthood mental disorders begin in childhood and adolescence [83]. Therefore, understanding the effect of sex hormones on early neurological development, especially on the cognition related brain regional development, would be important for developing further strategies for preventing sex-related neurodegenerative disease, such as AD.

As early as prenatal period development, once the Y chromosome (SRY) gene forms the fetal testes, sexual differentiation of the brain is a sex hormone dependent process [84; 85]. Studies in rats showed that circulating levels of sex hormones (mainly testosterone) during a short critical developmental window (between embryonic day 18 and 10 days after birth) caused permanent sex-dependent changes in brain morphology and functions [86]. Interestingly, the roles of testosterone in sex differentiation during brain development might be mediated by its metabolite estradiol, a conversion process catalyzed by neuronal aromatase [87]. This suggests that the vertebrate brain is organized in a sex-dependent fashion under the control of perinatal gonadal steroid hormones, such as estrogen or testosterone [88]. While sex hormones are present during early development and are well known to have permanent effects on sex-related behaviors in human and various animals, recent work has renewed interest in puberty as another organizational period for the effects of sex hormones. For example, late maturing boys and girls had higher spatial abilities than their early maturing counterparts [89]. Furthermore, studies reported that a longer reproductive period was associated with better cognition in later life, including verbal fluency in females [90]. The sex hormone-dependency of brain regional development has been further supported by the study in girls with partial or complete loss of their X-chromosome, which showed disproportionately reduced hippocampal volume, and increased amygdala volume relative to age-matched controls [91]. The estrogen-related brain regional volume reduction in females with endogenous estrogen deficiency is further evidenced by the impairment of executive performance on the task as reported by other groups of researchers [92]. In animal studies, female mice exposed to stress in early adolescence showed more disturbed sexual behaviors than those exposed to stress in late adolescence, emphasizing the importance of timing of exposure even within adolescence [93]. Finally, gender differences in cognitive function have been well demonstrated in adulthood as well as ageing. For example, men demonstrated larger amygdala and thalamus volumes compared to women [94; 95; 96], whereas the size of the hippocampus is larger in females compared to males [94; 97]. It is also worth noticing that there are a relatively higher number of androgen receptors in the amygdala [98] and a relatively higher number of estrogen receptors in the hippocampus [99].

The sex differences in regional brain structure might be responsible for the sex differences in specific cognitive and behavioral tasks [80; 81; 100]. For example, males usually outperform females on visuospatial, quantitative, and targeted motor tasks, whereas females perform better on verbal and perceptual tasks [101; 102]. Furthermore, both human and animal studies showed that administration of androgens to females may induce male-typical cognition and behavior, and the male-type cognition disappearing when the treatment was withdrawn [80; 103]. For example, one recent cross-sectional human study showed that women with polycystic ovary syndrome characterized by elevated endogenous testosterone performed significantly better at three-dimensional mental rotation, a male-favored cognitive behavior, than a female control group [104]. Another study also demonstrated that a single administration of testosterone improves 3D mental rotation abilities in young women [105]. The sex hormone-specific action on cognition is also demonstrated by animal studies, such as testosterone treatment in aged females mice, which enhanced spatial retention in water mazes (male-typical cognition). However, there was no significant improvement in object recognition (female-typical cognition) compared to the placebo treated group [106].

In congruence with brain structure studies, the estrogen action in cognition is also related to the brain’s regional activities. Recent animal studies showed that endogenous estrogen levels changed by reproductive experience in females are associated with enhanced hippocampus-dependent memory [107; 108]. Furthermore, sex steroids have been linked to many of the mechanisms thought to be associated with cognitive decline and several types of dementia. A recent review by Hogervorst [109] reported that women who underwent surgical menopause or had menopause before 47 years old without hormone treatments had an increased risk for global cognitive impairment and dementia in later life, suggesting that earlier menopause is associated with a higher risk for cognitive impairment. Moreover, a new study reported that 17β-estradiol increased neurogenesis in the hippocampus and increased activation of new neurons in response to spatial memory retrieval, while estrone decreased cell survival and had no significant effect on the activation of new neurons [110]. It is known that hippocampal neurogenesis is linked to spatial learning and memory. Estrogen not only modulates neurogenesis in the hippocampus, but also modulates hippocampus-dependent learning and memory across a variety of species from rodents to primates [111]. For instance, studies of object-in-place tasks combined with object recognition and object location showed that the ovariectomized female rats treated with estradiol and progesterone were able to discriminate between moved and unmoved objects while ovariectomized vehicle-treated females and gonadally intact males did not, suggesting that female rats outperform males only when levels of estrogen and progesterone are elevated [112]. However, the interaction between sex hormones and brain regional-driven cognition might be more complicated than previously thought. A longitudinal human study of 10 males and 7 females for 8 weeks showed that both genders showed significant sex hormone related cycle effects in mental rotation performance only at the beginning of the study, while at the end of the study, none of the hormones were significantly related to performance. Thus, the relationship between hormones and certain cognitive activity, such as mental rotation performance, disappeared with repeated testing [113]. Furthermore, a cross-sectional analysis of the association between sex hormones, metabolic parameters, and psychiatric diagnoses with verbal memory in healthy aged men showed that higher levels of serum sex hormone binding globulin (SHBG) were associated with a worse verbal memory [114], suggesting that levels of free testosterone influence male verbal memory.

Indeed, the concept that brain estrogen regulates cognitive function independent from circulating estrogen levels is also supported by more evidence. For instance, serum concentrations of estradiol are much lower than estradiol concentrations in the male rat hippocampus [46], suggesting a locally synthesized estradiol may contribute even stronger to the maintenance of structural synaptic plasticity than estradiol originating from the gonads [40]. The brain estrogen theory is also supported by cell culture studies which showed that axonal outgrowth in hippocampal neurons only occurs when estradiol concentrations are higher than physiological serum levels [42].

Brain estrogen and AD

AD is the most common cause of dementia in the elderly, with women being at a higher risk for AD even after controlling for an increased life span [115; 116]. Women not only have a higher risk of developing AD than age-matched men, supported by higher numbers of female AD patients than male patient with AD [4; 117; 118; 119], but also showed significantly age-related faster decline and greater deterioration of cognition than elderly male [120; 121; 122].

Several studies showed that AD patients face more problems in anterograde episodic memory than frontotemporal lobar degeneration (FTLD) patients [123; 124; 125; 126]. Others reported that both AD and FTLD dementia groups did not differ in their verbal and forgetting rates, besides a potentially worse visual memory in the AD group compared to the FTLD group [127]. Mild cognitive impairment (MCI) is a clinical condition often associated with early stages of AD. A study of 425 MCI patients showed that early onset MCI patients suffer more visuospatial memory impairment, while individuals with late onsets of MCI showed more verbal memory problems [128]. Moreover, other studies also reported no significant differences in the demographic characteristics of patients with mixed dementia and AD, where patients with mixed dementia were significantly more impaired than AD patients in global cognitive composite, attention and visuoconstruction [129]. Other cognitive measurements have also been used in MCI and AD. For example, an antisaccade task is a measurement for executive functions and is correlated with AD-sensitive cortical regional thickness in MCI subjects, but not in normal elderly patients as reported by Heuer [130]. A study using a new cognitive test in 97 mild AD/MCI and 200 control subjects demonstrated a significant detection of mild AD/MCI who “pass” the MMSE from controls [131]. Although females have a high prevalence of AD, there are no clear evidence showing that whether there are gender differences in MCI or FTLD. For example, one study showed that more men develop MCI than women [132], while other studies showed that the incidence or pathology of MCI did not differ by sex or education [133; 134]. While FTLD has a strong genetic basis, studies found equal heritability for both sexes regardless of parental gender [135], as well as no differences in sex distribution in FTLD in different countries [136; 137].

There are two major pathological lesions in AD brain: intracellular inclusions of tau protein in the form of neurofibrillary tangles, and extracellular plaque formation by accumulations of amyloid beta peptide (Aβ) derived from the β-amyloid precursor protein (APP). Accumulating Aβ, a proteolytic byproduct of amyloid precursor protein (APP) metabolism, is an important aspect of AD pathology. APP is processed by two competing pathways, the amyloidogenic pathway through β-secretase (BACE1) and γ-secretase, which produce β-APPs and Aβ40/Aβ42 peptides, and the predominant non-amyloidogenic pathway via α-secretase, which produces neuroprotective α-APPs and several non-amyloidogenic peptides [138]. The sex-specific neuropathology in AD has been reported by neuroimaging and postmortem human studies, such as men with AD have more pronounced pathology in the right hemisphere, whereas women with AD often have more manifested pathology than male AD patients [139; 140].

Compared to age-matched controls, reduced circulating levels of 17β-estradiol and testosterone are observed in female and male patients with AD, respectively [116; 141], suggesting a potential important role of sex hormones in the epidemiology and pathology of AD. With an increase in human longevity, women spend an average of one third of life without sufficient endogenous estrogens due to the significant reduction of endogenous estrogen after menopause. Estrogen is known to have a protective effect on the brain, and loss of estrogen during menopause could, in part, lead to the deficits seen in brain metabolism (mitochondrial impairment) in AD [142]. Estrogen also plays critical roles in neurogenesis. As reviewed by Galea [111], E2 can increase neurogenesis in various brain regions such as dentate gyrus of hippocampus, and these newly generated neurons in the hippocampus contribute to region-specific learning and memory. Studies showed hippocampal atrophy in females with MCI, suggesting an important role for estrogen in the association between cognitive function and hippocampus [143]. The brain region-specific effects of estrogen have also been confirmed by recent studies, which further highlighted the potent effects of estrogens on neuronal morphology and plasticity in area CA1 of hippocampus [63; 67; 144; 145; 146; 147]. As reviewed by Brinton ([148] and references therein), the beneficial effects of estrogen on neural plasticity occur at cellular, morphological, and synaptic levels. E2 can increase neurogenesis in various brain regions such as dentate gyrus of hippocampus, and these newly generated neurons in the hippocampus contribute to region-specific learning and memory. In addition, E2 can rapidly increase dendritic spine numbers or dendritic spine contacts in the hippocampus, the medial amygdala and hypothalamus, thus enhancing hippocampal-dependent memory in monkeys. E2 is also a potent and efficacious mediator of synaptic transmission in hippocampal system. Together, E2 promotes neurogenesis and neuronal plasticity to maintain healthy cognitive function and protect against cognitive decline in females during aging.

Another important neuroprotective role of estrogens in AD is that estrogen could reduce Aβ levels, or prevent them from rising, in the presence of pathological triggers [149]. Estrogens can also reduce Aβ production by favoring the non-amyloidogenic pathway through MARK/ERK activation, reducing BACE1 levels, and promoting Aβ clearance by stimulating microglial phagocytosis and degradation and regulating the levels of major enzymes involved in Aβ degradation [150]. Estrogens can also be protective by regulating the Bcl-2 protein family, increasing the expression of antiapoptotic Bcl-xL and Bcl-w and suppressing expression of proapoptotic Bim to prevent neuronal loss from Aβ-mediated toxicity [151]. Decreased levels of hyperphosphorylated tau (a major component of neurofibrillary tangles) can also be mediated by estrogens through kinases and phosphatases, such as the GSK-3β, Wnt, or PKA pathways [152].

Although the precipitous loss of ovarian estrogens at menopause has been presumed to account for the increased female susceptibility to AD, the role of this peripheral estrogen pool is still disputed [153; 154; 155; 156]. The brain is a sex hormone-responsive tissue and is affected by age related loss of estrogen [5; 66]. As shown in table 1, studies of brain aromatase expression and polymorphisms as well as direct measurement of estradiol levels in the brain of AD suggest that brain estrogen deficiency may increase risk of AD. The hypothesis of brain estrogen deficiency increases risk of AD is further supported by animals studies which showed that genetic eliminating aromatase in the female AD transgenic mice caused early and more severe AD pathology then control mice [5]. Based on analyses of brain estrogens, particularly E2, in the postmortem brain samples from female AD patients and age-matched healthy subjects, studies demonstrated a great reduction of estrogen levels and estrogen biosynthesis in AD brains while same serum estrogen levels was found in both AD and controls, suggesting that brain estrogen deficiency, not circulating E2, is associated with AD pathology [5]. Furthermore, one study also reported a reduction of estradiol in the CSF from AD patients and provided further confirmation of important role of brain estrogen in AD [157]. The differences in brain estrogen level between female AD and health individuals provide some level of explanation on why only 13–15% aged females developed AD, even though every woman lost sufficient endogenous estrogen after menopause in advanced age. In accordance with the downregulation of brain estrogen level, the changes of sex steroid (i.e. estrogens, androgens and progesterone) biosynthesis pathways in human prefrontal cortex have also been reported during the course of AD. For example, brain aromatase is expressed in AD-relevant regions and downregulated in AD [50; 158; 159]. The brain estrogen deficiency in AD is also confirmed by several studies which showed a downregulation of brain aromatase expression in the AD hippocampus [158; 159; 160]. However, in the prefrontal or temporal cortex of AD patients, the expression of aromatase mRNA was found elevated in the astrocyte [50] and no changes in the neurons [161], suggesting a cell type and regional specificity of brain estrogen synthesis. A reduction of brain aromatase immunoreactivity is also identified in the hippocampus of AD patients [159], but not in patients affected by epilepsy [57]. In contrast, during normal ageing, a clear increase in aromatase immunoreactivity has been reported in the basal forebrain [158], suggesting an age-related elevation of brain specific estrogen synthesis as a compensation of circulating estrogen reduction in normal postmenopausal women. In concert to the theory that maintaining healthy brain estrogen level is critical for preventing AD in females, studies reported that APP transgenic mice with low brain estrogen developed earlier, and with more severe AD pathology than ovariectomized APP mice [5]. Furthermore, a greater and better neuroprotective response from estrogen treatment was found in APP mice with brain estrogen deficiency compared to in the ovariectomized APP mice [66]. Together, evidences suggest an important role of brain estrogen in female risk of developing AD.

The concept of brain estrogen regulating cognitive function is also partially supported by clinical studies. For example, aromatase inhibitors (AIs) such as exemestane, anastrozole or letrozole, have been used to treat breast cancer. Studies showed that anastrozole treatment impaired processing speed and verbal memory in postmenopausal women with breast cancer in comparison with healthy women without the anastrozole treatment, or breast cancer patients treated with non AIs [162; 163]. However, AI users showed no significant decline of cognitive function compared to non-AI users in breast cancer patients were reported from a number of other studies, including a later study of anastrozole showing no impairment to cognitive performance [164; 165]. It is worth to note that selective estrogen receptor modulators (SERMs) are also frequently used as an endocrinal therapy to suppress estrogen for breast cancer patients. While the clinical data on the roles of SERMs in cognitive function are not conclusive, some studies suggest that tamoxifen administration may form a risk for cognitive functioning particularly in older women [166]. Since SERMs interfere with ER signaling, one could hypothesize that estrogen receptors are potentially involved in the SERMs-induced cognitive impairment.