Deciphering the Influence of Estradiol and Estrogen Receptors on Cognitive Function: A Bibliometric Analysis and Emerging Research Trends

Background

Estrogens are steroid hormones mainly produced by the ovaries and adrenal glands; in addition, the brain can also produce endogenous estrogens from cholesterol [1,2]. Estrogens are known to promote female sexual characteristics and reproductive capacity, and they can cross the blood-brain barrier, affecting the brain [2–4]. Estrogens have 3 forms: estrone (E1), estradiol (E2), and estriol (E3). Because E2 is abundant and potent in the body, the term estrogens mainly refers to E2 [2,5,6]. A marked decrease in estrogens levels occurs during a 6-month period around menopause, most pronounced in E2, and during the next 3 years, the levels of E2 and E1 show an essentially parallel, moderate decline [7], but the binding affinity of E1 with intracellular estrogen receptors (ERs) is lower than that of E2 [8]. Subcutaneous injection of E2 significantly increased BrdU-labeled newborn neurons in the hippocampus dentate gyrus (DG) and activation of new neurons in response to spatial memory retrieval, but the same dose of E1 decreased neurogenesis in the DG [5]. Therefore, different estrogens may have different effects on neurogenesis and cognition in the hippocampus.

ERs are found throughout the brain and are predominantly present in the hippocampus [9], prefrontal cortex (PFC) [9], cerebellum [10], hypothalamus [11], ventral tegmental area [12], and amygdala [12], as well as in the raphe nuclei of the midbrain [12]. They exist in neurons and glial cells [13], and the hippocampus is especially rich in ERs [9,13,14]. ERs mainly include nuclear ERs (nERs) and membrane-embedded ER (mER), and nERs have 2 subtypes: ERα and ERβ [9]. In addition, orphan G protein-coupled receptor (GPCR), which binds E2 with high affinity and initiates rapid signal transduction from the plasma membrane, was first found in SKBR3 breast cancer cells [15]. This receptor was discovered and cloned in 1997 and was called GPR30, and in 2007 it was officially named G protein-coupled estrogen receptor (GPER) by the International Union of Basic and Clinical Pharmacology [16].

Articles [17,18] reported the localization of nERs on the plasma membrane by E2 treatment. The post-translational lipid modification of nERs through palmitoylation is helpful for the binding of nERs to caveolin (CAV), thereby transporting nERs to the cell membrane and activating several cell-signaling cascades [17,19–21]. nERs are also localized to mitochondria depending on the cell type [22,23]. Mitochondrial estrogen receptors (mtERs) have been found in the hippocampus, cortex, and hypothalamic of rats, which could mediate the effect of E2 on mitochondria. The density analysis of mtERs in young adult and aged rats showed that aging did not cause a decrease in the number of mtERs in different brain regions, but mtERs were differentially expressed in different brain regions, indicating that they may be involved in different functions in the brain during aging [24]. In primary rat neurons and HT22 mouse hippocampal cell line, ERβ has been shown to co-localize with mitochondria [25] and regulate transcription and protein activity in this region [26]. The mtERs can bind to the estrogen response element (ERE) located in the mitochondrial DNA, thereby increasing the expression of mitochondrial-encoded genes and the activity of electron transport chains (ETC). In addition, co-immunoprecipitation results showed that ERβ could interact with mitochondrial respiratory complex V [27,28].

Studies have shown that E2 has important effects on higher-order brain functions, including locomotor activity [12], mood [12,29], affect [30], anxiety [31,32], fear [30], and cognition [29,33–35]. Cognitive function is the ability to learn, retain, and recall information; in a broad context, cognition includes all mental abilities and processes related to knowledge, including, but not limited to, attention, memory, reasoning, comprehension, and language production [36]. The decline in E2 levels caused by perimenopause and postmenopausal can affect memory and increase the risk of Alzheimer disease (AD) [37,38]. Moreover, the prevalence of AD and related dementia is higher among women than among men worldwide, and this difference increases with age [39]. Researchers suggest that the level of E2 may be closely related to cognition [40]. This possibility provides a basis for the cognitive protection of estrogens replacement therapy (ERT), but long-term exposure to E2 is the main risk factor for breast cancer progression [41,42]. Researchers have attempted to explore the unique cognitive characteristics mediated by different ERs to improve the efficacy of ERT and reduce its adverse effects [19,43,44]. Therefore, we conducted a knowledge domain visualization (KDviz) study on E2/ERs and cognitive-related research fields for the first time to help research groups review and analyze the trends in this knowledge domain. CiteSpace is a freely available Java application for KDviz and analyzes emerging trends and changes in the scientific literature [45]. The main motivation for this work was to simplify the search for important publications in a knowledge domain’s literature so that people can search for visually salient features, such as hub nodes, in a visualized network [45].

Material and Methods

SEARCH STRATEGY:

The data for this study were obtained from articles and reviews published from 2010 to 2021 in the Web of Science Core Collection (WOSCC) database. The search keywords were established in reference to Medical Subject Heading terms from PubMed. The topic terms used were “estrogen,” “estradiol,” “estrogen receptor,” “ovariectomy,” and “postmenopausal”. This query produced 114 629 records, which comprised Set #1. Another set of topic terms consists of “cognition,” “learning and memory,” “synaptic plasticity,” “neural plasticity,” “brain health,” “hippocampal,” and “neurogenesis”. This query led to 278 454 records, which comprised Set #2. We combined Set #1 and Set #2 and obtained the final dataset, Set #3, which consists of 3502 records (Table 1). All the authors reviewed 3502 publications, including 2842 research articles (81.15%) and 660 review articles (18.85%). Of the 3502 publications, Neurosciences ranked first, with 1579 (45.09%). Endocrinology Metabolism ranked second, with 737 (21.05%). Behavioral Sciences ranked third, with 377 (10.77%). Other disciplines include Psychiatry, Biochemistry Molecular Biology, Pharmacology Pharmacy, Geriatrics Gerontology, Clinical Neurology, Obstetrics Gynecology, and Multidisciplinary Sciences. A total of 3502 references were retrieved and downloaded on September 18, 2021. All cited reference information was downloaded and saved as plain text files for data processing and analysis. Finally, the downloaded data were imported into CiteSpace for deduplication. This search strategy has some limitations. First, the retrieval strategy is limited to the WOSCC database. Secondly, the search words may not be very comprehensive, the time range of the search was limited, and non-English publications were excluded.

DATA ANALYSIS METHOD:

We used CiteSpace 5.7.R5W (http://cluster.cis.drexel.edu/~cchen/citespace/) to conduct statistical analysis of publications [46]. CiteSpace can generate a KDviz map, whose primary goal is to monitor the evolution of a knowledge domain [47]. The parameters of CiteSpace were set as follows: Method (the Log-likelihood ratio algorithm), Time slice (2010–2021), Slice length (1 year), Node type (1 selected at a time), Selection criteria (co-authors, co-cited authors, co-author’s institution, and co-author’s country are the top 20 objects in each slice length, and co-cited references and co-occurring keyword clusters set were the top 50 objects in each slice length), and Pruning (Pathfinder and Pruning the merged network) [48]. All CiteSpace analysis terms are summarized in Table 2. The most prominent feature of this software is that it can be used for collaboration network analysis of co-author, co-author’ institution, co-author’s country, co-cited analysis of the article, author, and journal, and co-occurrence analysis of term, keyword, source, and category.

We used CiteSpace to analyze co-authors, co-author’s country and institution, co-cited authors, co-cited references, and co-occurring keyword clusters. Sigma (∑) was used to detect co-authors and co-cited references. Co-authors were defined as 2 or more authors who co-created an article and were jointly responsible for the knowledge content of the article. The co-citation count is the number of times that 2 published articles are co-cited by later articles, then the 2 articles constitute the relationship of co-cited reference and the authors of the 2 articles constitute co-cited authors [49]. Co-occurring keyword clusters are a graphical representation of the number of keywords that appear [50]. When the modularity (the Q score) [51] is greater than 0.3, it is considered that the divided cluster structure is significant. The lower Q score indicates that a network cannot be simplified as a cluster with clear boundaries. When the silhouette (the S score) [45] exceeds 0.7, the cluster is considered to be highly credible. For these 2 indicators, a score close to + 1 represents the best cluster model.

The centrality of a node indicates the importance of its position in the network [45]. A common centrality index is betweenness centrality, which measures the percentage of the number of shortest paths in the network to which a given node belongs [52]. Nodes with high betweenness centrality tend to appear in the paths connecting different clusters and are considered to be key hubs [53]. Citation burst refers to the importance of nodes in time. Burst detection measures the rate of change [45], which determines whether a given frequency function has statistically significant fluctuations over a short time interval over the entire cycle [50]. Cheng et al introduced the burst detection algorithm [45]. Burst is mainly used to detect articles that attract peer scientists’ attention. Sigma combines centrality as a structural attribute with citation burst as a temporal attribute. It is calculated as (centrality + 1)burst, and the higher these values, the greater the potential influence of the authors or articles [45].

The maps consist of different nodes and links. The size of nodes is related to the number of publications and number of times co-cited. Purple outer rings represent the importance of nodes in terms of betweenness centrality (≥0.1), and the thickness of the ring is proportional to the value of the betweenness centrality. The thickness of the red outer ring of the node represents the burst strength [45]. The number and thickness of links represent the degree of cooperation between the 2 nodes and the correlation between co-cited articles. The color of the link indicates the time of the first connection [47].

Results and Discussion

ANALYSIS OF NETWORKS ACROSS COUNTRIES, INSTITUTIONS, AND AUTHORS:

We used CiteSpace 5.7.R5W software to report the countries and institutions with the most publications and the highest centrality, respectively (Figures 1A, 2A and Tables 3, 4). Specifically, the node circles of the United States, People’s Republic of China, and Canada are larger, indicating that these countries have a large number of publications (Figure 1A). However, Scotland, Switzerland, and Belgium have larger purple rings (high centrality), indicating that they have close cooperation with other countries in this field (Figure 1A). Universities are the main research institutions related to E2/ERs and cognition. The University of Wisconsin is the university with the largest number of publications (77), and the centrality in this research area was Stanford University (0.78) (Figure 2A and Tables 3, 4).

We used CiteSpace 5.7.R5W to generate the co-cited authors network and the co-authorship network (Figures 1B, 2B). Co-cited authors were ranked by number of times cited, centrality, Sigma index, and burst strength (Tables 5–8), and the authors with the most publications were listed (Table 9). Woolley CS is the author with the most citations (594), Frick KM has the highest centrality (1.27) and Sigma index (1261582180303.55), Tuscher JJ has the strongest burst strength (40.66), and Espeland MA has the most publications (38). These KDviz maps enabled us to assess the influence and centrality of some important countries, institutions, and authors.

ANALYSIS OF CO-CITED REFERENCE:

Figure 3 shows the generated map of co-cited references and Figure 4 shows the top 50 co-cited references with the strongest citation bursts sorted by burst time. The top 5 co-cited references with high times, centrality, Sigma index, and burst strength were listed (Tables 10–13). These reviews [1,36,44,54] showed that E2 could improve hippocampal formation (structural and functional plasticity), enhance hippocampus-dependent cognitive function, and provide the required cellular signaling mechanism. In addition, the type and distribution of ERs were described in detail. Luine VN et al [36] discussed the “critical age window period” hypothesis of E2 effect and new strategies for the development of ERT. Three clinical trials [55–57] showed that ERT could not prevent mild cognitive impairment (MCI) for women over 65 years old, and it may also increase the risk of dementia, which had a greater impact on the cognitive ability of women with low cognitive function. Therefore, women over 65 years old are not recommended to use ERT to prevent dementia or MCI. Subsequently, Shao HB et al [58] explored the relationship between the time of using ERT and AD, and found that ERT starting at or more than 5 years after menopause was associated with an increased risk of AD, whereas using ERT in early menopause reduced the risk of AD in women by 30%. Another cohort study [38] showed that premenopausal ovariectomy (OVX) was associated with increased risk of cognitive impairment or dementia. Henderson VW et al [59] reviewed the effects of surgical and natural menopause on cognition. Surgical menopause may be accompanied by cognitive impairment, mainly affecting verbal episodic memory. By contrast, natural menopause transition is not accompanied by substantial changes in cognitive ability. When premenopausal women undergo ERT shortly after OVX, they may gain potential short-term cognitive benefits, but elderly postmenopausal women will not. Studies [60] also found that long-term ERT immediately after OVX had a positive effect on working memory (WM) in female rats, whereas long-term hormone deprivation followed by ERT was ineffective. The above studies have shown that ERT has a neuroprotective effect on the brain of premenopausal, perimenopausal, and early postmenopausal women, suggesting that it has a critical window period to treat cognitive impairment.

Woolley CS [1] found that E2, as a neurosteroid, was produced at a high level in the brain and acutely changed neuronal physiology. Hojo et al [61] showed immunoreactions of dehydroepiandrosterone synthase (P45017a) and P450 aromatase (AROM) in the pyramidal and granular neurons of the adult rat hippocampus. P45017a and AROM transformed pregnenolone in neurons into E2, indicating that E2 can be locally synthesized in adult hippocampal neurons. Rosenfeld CS et al [62] introduced 6 AROM inhibitors – Anastrazole, 1,4,6-androstatriene-3,17-dione, Exemestane, Fadrozole, Letrozole, and Vorozole – among which, Letrozole is the most commonly used. They are mainly used for the treatment of women with breast cancer [63]. AROM inhibitors can cause low levels of brain-derived E2 synthesis and cognitive impairment, and women receiving AROM inhibitors are at higher risk of cognitive impairment [64]. Nonetheless, exercise can be implemented early to reduce this risk [65]. Scientists also reported that a primate model for Letrozole treatment reduced levels of peripheral E2 but unexpectedly increased levels of E2 in the hippocampus [66]. In OVX- and Letrozole-treated female mice, the number of dendritic spine synapses in the hippocampus decreased significantly, but not in the PFC or cerebellar cortex [67]. These results showed that the effect of Letrozole might differ by species and brain regions.

ANALYSIS OF RECENT BURST CO-CITED REFERENCES:

We list 13 recent burst co-cited references (Table 14), indicating that these articles and topics have recently received close attention. Based on these recent burst co-cited references and other current studies, researchers found that GPER upregulates ERK (extracellular signal-regulated kinase) signaling in vivo and in vitro [68–71]. However, another in vivo study [43] found no activation of ERK but did find activation of JNK (c-Jun N-terminal kinase) in the dorsal hippocampus (DH) after administration of G-1 (GPER agonists). DH infusion of G-1 can promote OR (Object Recognition) and SR (Social Recognition) but had no effect on OP (Object Placement). These experiments were completed within 40 min after G-1 infusion. Therefore, GPER can rapidly promote OR and SR in female mice [72], and OP can be mediated by nERs [73], but other studies have found that OP can be promoted by GPER activation [43].

Hadjimarkou MM et al [74] summarized agonists and antagonists for all ERs and receptor variants, and discussed crosstalk between GPER and ERα, which is conducive to future research on the mechanism by which E2/ERs improve cognition. Galea LAM et al [75] suggested that studies assessing sex differences must consider the role of other interference factors, such as reproductive and hormonal status. Luine V, Scharfman HE, and Zárate S [76–78] provided evidence for 3 findings: (1) E2 and BDNF can synergistically enhance cognitive ability; (2) E2 can promote the expression of the BDNF gene by binding to nERs and acting on the estrogen response element (ERE) of the BDNF gene; and (3) BDNF promotes the activation of intracellular signaling pathways, including ERK, phosphatidylinositol 3-kinase (PI3K), and cAMP response element-binding protein (CREB) by tropomyosin receptor kinase B (TrkB).

ANALYSIS OF CO-OCCURRING KEYWORD CLUSTERS:

Figure 5 shows 10 different clusters with high Q and S scores (Q=0.8266; S=0.978) identified in the network of co-occurring keywords indicating highly credible clusters. We divided them into 6 themes: (1) E2/ERs and hippocampus (#0 expression, #7 dendritic spine density, #8 neurogenesis and #9 hippocampal neuron); (2) The Women’s Health Initiative Memory Study (#1 women and #4 health initiative memory); (3) Memory type (#3 WM); (4) Sex specificity (#2 sex); (5) Receptor specificity (#5 receptor); (6) The animal models used in the research (#6 mice).

E2/ERS AND HIPPOCAMPUS:

E2 is involved in the influence of the CNS, including the regulation of synaptic plasticity and physiological plasticity, which is considered the key mechanism and basis for the effect of E2 on cognitive function [54,79]. Acute activation of E2 and ERβ has been shown to rapidly regulate the density of dendritic spines [80,81]. In addition, exogenous GPER expression of primary rat cortical neurons grown for 24 days in vitro has a higher density of dendritic spines than that without GPER expression [82]. In the mHippo-E14 cell line, activation of GPER also upregulates actin polymerization and synaptic proteins [83]. Hippocampal slices of nERs KO mice were bathed in 17β-estradiol-3-benthic ester (EB) and G-1, and GPER was found to regulate LTP [70].

The mTORC2 complex is a multimer kinase composed of mammalian target of rapamycin (mTOR), mlst8, msin1, and rictor [84,85]. E2-treatment induces ERK and mTOR-dependent spinogenesis in hippocampus CA1 and PFC pyramidal neurons [86], and it was recently found that aerobic exercise can also improve dendritic spine plasticity in the hippocampus and PFC of OVX mice through the BDNF/mTOR signaling pathway [87]. When rictor, the core component of mTORC2, was deleted, hippocampal actin polymerization decreased. In addition, steroid receptor coactivator-1 (SRC-1) has been shown to enhance the transcriptional activity of steroid nuclear receptors [88,89]. SRC-1 is highly expressed in the hippocampus, and it has been proved to regulate growth and aging and mediate the effect of E2 on synaptogenesis in the hippocampus [90,91]. SRC-1 is regulated upstream of mTORC2, and ERs inactivation results in a significant decrease in SRC-1 expression [91]. In OVX mice, intraperitoneally injected G-1 reversed spatial memory decline, loss of dendritic spines, and actin depolymerization, and G-15 (GPER antagonists) injected continuously for 7 days blocked the effect of G-1 [83]. Based on the above findings, we believe that ERs, mTOR, rictor, and SRC-1 may provide new targets for the prevention and treatment of E2-related cognitive impairment in elderly people by improving dendritic spine plasticity.

Hippocampus-derived E2 plays a critical role in synaptic plasticity. Lan et al [92] constructed AROM-specific RNA interference AAVs and found a critical role of hippocampus-derived E2 in regulating hippocampal structure and function. In adult female mice, intraperitoneal injection of Letrozole significantly decreased severe synapse density, spatial learning, and memory impairment in OVX mice [35,67], and Letrozole inhibits actin polymerization and synaptic plasticity in primary cultured hippocampal neurons of mice [93]. These results indicated that the content of hippocampus-derived E2 should be considered in the process of ERT, not only focusing on peripheral E2. The plasticity of hippocampal neurons includes adult hippocampus neurogenesis (the proliferation, differentiation, and migration of neural progenitor cells [NPCs] in the adult hippocampus) to replace non-functional cells, and activation of ERβ promotes neurogenesis in DG [94]. DPN (ERβ agonists) regulates proliferation, differentiation, and transcription of several key NPCs and survival regulators, including Insulin-like growth factor 2, acidic fibroblast growth factor, Midkine, Prolactin, and cyclin-dependent kinase [95–100].

THE WOMEN’S HEALTH INITIATIVE MEMORY STUDY (WHIMS):

In the last century, women’s life expectancy increased by 30 years, from 50 to 83 years old; however, the average age of spontaneous menopause remained stable, at 50–51 years [101]. As a result, women now live more years with low estrogen levels [102]. The Women’s Health Initiative (WHI) study is a large, multi-site, population-based study assessing the risks and benefits of ERT for healthy postmenopausal women [103]. WHIMS is an additional study of WHI, which aims to study the effect of postmenopausal ERT on cognition and memory in healthy women aged 65 and above at baseline. WHIMS found that older women using conjugated equine estrogen (CEE) with or without medroxyprogesterone acetate (MPA) have an increased risk of dementia and smaller hippocampal and frontal brain volumes [55,56,104]. In some WHIMS participants who participated in the auxiliary WHI cognitive aging study, compared with placebo, CEE + MPA impaired verbal memory over time [105]. The increase of these risks seems to be related to the acceleration of brain atrophy.

The Young Women’s WHIMS (WHIMSY) study investigated the cognitive function of women aged 50–55 who participated in the WHI-hormone therapy (HT) test. Unlike WHISM, the WHIMSY data showed no difference in the cognitive performance of women receiving HT compared with placebo, and no difference was found in the individual binding of CEE and CEE + MPA [106–108]. Other studies have shown that the benefits of estrogen are preserved only when ERT begins in the postmenopausal years [109–111]. These findings led to the “critical age window period” hypothesis that suggests a precise window of opportunity for beneficial ERT in postmenopausal women.

MEMORY TYPE:

In rodents, both the Morris water maze (MWM) and the radial eight-arm maze (RAM) can be used to measure hippocampal spatial memory [112,113], but the RAM can use the reference memory (RM) to reflect the long-term spatial memory and WM to evaluate short-term spatial memory [112]. Four rewarded arms and 4 no-rewarded arms are used in the RAM. RM is memory consistent across sessions (ie, location of rewarded arms), and WM is memory used within a single session (ie, which arms were already visited) [112]. In such tests, individual animals were repeatedly tested for many days, and their ability to find a platform or maze arm that remained constant was measured [114,115]. The MWM is mainly used in animals who hate the water environment and want to escape to a safe platform, whereas the motivation in the RAM maze is food. In the MWM task, there are almost an infinite number of swim path trajectories that could be used, while in the RAM task, the number of trajectories is limited; moreover, the radial arm maze has a spatial WM component not present in the MWM to reflect short-term memory [116].

WM supports many complex cognitive functions that require temporary storage and manipulation of information over short periods of time in order to make a response [117]. Some evidence shows that E2 increases WM [118–120] but does not improve RM [121,122]. In addition, E2 fluctuations at different stages of the estrous cycle, affecting the performance of tasks used to assess memory [122]. Hampson’s group found for the first time that there was a correlation between ERT and WM performance in postmenopausal women [123] and young women [124], which was later reported by other groups [125]. Recent studies [115] on the virtual MWM and the virtual RAM of young female college students showed that both high and low levels of natural-cycle E2 did not affect the WM and RM of the virtual RAM, but high levels of natural E2 promoted the spatial memory performance of the virtual MWM. This suggests that the effect of E2 on women’s spatial memory may be limited to more complex spatial memory tasks, such as the virtual MWM, which is more ambiguous because it does not have a direct path to the target location and participants can move freely in any direction.

In summary, the inconsistent results of spatial memory research may be related to species, test methods, and fluctuations and sources of E2, which need to be further explored.

SEX SPECIFICITY:

Many neurological and neurodegenerative diseases vary by sex [126]. For example, the incidence of autism spectrum disorders is much higher in boys than in girls [127], and women have a higher risk and faster progression of AD than men [128]. Many studies have shown that the brains of males and females are different at the molecular level [129].

Current studies have shown that the beneficial effects of hippocampal E2 on memory consolidation in males and females are mediated by different molecular mechanisms. In female mice, E2 upregulates the ERK/CREB signaling pathway by ERα/mGluR1 crosstalk, while E2 regulates male memory by phosphorylation of CREB independent of ERK [19,130–132]. Woolley’s group [126,133] showed that E2 rapidly suppresses inhibitory synaptic transmission through ERα/mGluR1 crosstalk and activates the phospholipase C (PLC)/IP3/Ca+ signaling pathway to mobilize endocannabinoids in the hippocampus via a sex-specific manner in females. E2 acts via ERβ to increase presynaptic glutamate release probability and through GPER to increase postsynaptic sensitivity to glutamate in females, whereas E2 plays a role through presynaptic ERα to increase glutamate release probability and postsynaptic ERβ to increase glutamate sensitivity in males [134]. Although the degree of synaptic enhancement induced by acute E2 is the same in both sexes, Ca2+/CAMKII, L-type Ca2+ channels and internally stored Ca2+ release are required for E2-induced information enhancement in males and females. However, protein kinase A (PKA) only needs to initiate enhancement in females [135].

In organotypic entorhinal-hippocampal cultures from mice, GPER mediated the increase of dendritic spine density in the stratum lacunosum-molecular layer of the hippocampus, which was sex-specific only in cultures from female mice [136]. Moreover, Letrozole treatment decreased dendritic spine density, synaptophysin, postsynaptic protein, mitochondrial volume, and LTP in female mice. Interestingly, these phenomena were sex-specific, but not found in male animals [137,138]. Studies [139] found that neurogenin 3 (Ngn3) promoted neuritogenesis in hippocampal neurons with sex differences, which was mainly related to higher expression of Ngn3 in female neurons. Therefore, female neurons have higher dendritic complexity and longer axons than male neurons.

The findings of sex differences suggest that therapeutic targets for cognitive impairment in men and women are sex-specific.

RECEPTOR SPECIFICITY:

GPER is located on the cell membrane and mediates rapid E2 signal transduction, whereas ERα and ERβ were initially proven to be transcription factors that function in a genomic way [82,140]. Unlike nERs, GPER is not a transcription factor, so it cannot directly regulate gene expression, and GPER can indirectly regulate gene expression by activating proteins that change the activity of transcription factors [141,142]. GPER widely exists in various brain regions, and it has a similar distribution pattern in the brain tissues of rats, mice, and humans [143–145]. GPER is also expressed in rats and mice hippocampal CA1–CA3 and DG areas [143,146,147]. Although GPER is localized in the plasma membrane, it can exist in the hippocampus cells’ endoplasmic reticulum or Golgi apparatus. In addition, GPER can be transferred from the plasma membrane to the perinuclear space [148]. In a virtual and biomolecular screening of about 10 000 molecules, a compound called G-1 was developed, which did not significantly bind to ERα or ERβ even at high concentrations [149]. Evidence suggests that G-1 binds explicitly to a 36-kDa splice variant of ERα (ERα36) and induces its activity to mediate non-genomic E2 signaling pathways [141]. Studies have reported that PPT (ERα agonists) also has an agonist effect on GPER at high concentrations [150], which complicates investigation of the effects of G-1 or PPT on GPER and ERα in future studies.

Different ERs may be responsible for mediating different memory tasks. ERα KO mice showed complete SR disorder, whereas ERβ KO mice showed partial SR damage [151], and selective activation of GPER can improve SR [72,152] and OR within 40 min [72], but does not affect OP [72]. Chronic (trial 4 h later) and acute (daily injection for 2 days with trial 48 h later) DPN treatment increased OR and OP performance in OVX rats, whereas acute PPT treatment did not change OR and OP performance [153]. In addition, the injection of WAY-200070 (ERβ agonists) promoted spatial learning and memory, dendritic spine density, and the level of synaptic protein in OVX rats, whereas PPT did not show a promotion effect [154].

ANIMAL MODELS USED IN THE RESEARCH:

Prange-Kiel et al [155] first proved the local synthesis of E2 in the hippocampus, and studies confirm that E2 is produced in the brain [2]. The application of Letrozole, AROM KO mouse, and various in vitro models of the hippocampus cell are valuable tools for validating the effects of neurogenic E2 on the morphology and function of hippocampal neurons.

Four transgenic mouse models that downregulate GPER have been established to assess the role of GPER in physiological function [156–159]. Studies of GPER KO mice have also identified the role of GPER in the CNS [160]. The effective interference sequence of GPER short hairpin RNA (shRNA) in cultured cortical neurons [161] and basolateral amygdala of OVX mice [162] has been reported, which can be used for subsequent studies on GPER. Genetically modified mice lacking Cyp19 gene encoding AROM have been used in many studies to detect the effects of lifelong AROM deficiency on various organs, including the brain [163–165]. A conditional KO, an adeno-associated viral vector (AAV) approach, or other targeted methods can be used to delete ERs or AROM expression in specific brain regions.

Conclusions

Through analysis, we finally draw our conclusions. First, the United States and the University of Wisconsin are the most-published countries and institutions, respectively, whereas Scotland and Stanford University have the most centrality. Woolley CS, Frick KM, Tuscher JJ, and Espeland MA are influential authors in this field. Second, the co-cited analysis shows that, in recent years, researchers focused on the “critical age window period” hypothesis of ERT, hippocampus-derived E2, the mediating role of GPER, and crosstalk between ERs. Third, we introduce through co-occurring keyword cluster analysis the 5 research directions that are most concentrated in this research field (E2 and hippocampus, WHIMS, memory type, sex specificity, receptor specificity) and animal models used in the study. These research results can help research groups determine position new directions for future research in this field and provide some potential targets for E2 to improve cognitive ability.

More and more animal studies have indicated that E2 can be used as a neuromodulator to regulate cognitive function, and the results of WHIMS show that this effect has a “critical age window period”. ERT can be implemented earlier in future research. Recent studies have revealed that the regulation of cognition by E2/ERs is closely related to the spine plasticity of neurons, involving several cell-signaling cascades and genomic effects caused by different ERs. There is increasing evidence that E2 improves cognition with sex specificity, the specificity of ERs activating intracellular signals and their effects on different memory types, ERs can crosstalk with each other, and ERs can interact with other membrane receptors. Researchers should also consider the specific effects of E2 on different cell types, the study of female animals’ cognition should exclude the effect of fluctuations in E2 levels during the estrous cycle, and sex should also be considered as a biological variable. To promote the development of this field, researchers can use more new technologies in the future, such as single-cell sequencing, omics analyses, and gene targeting.