A review of estrogen receptor alpha gene (ESR1) polymorphisms, mood, and cognition.

Estrogen facilitates synaptogenesis, induces growth factor production, protects against oxidative stress, and regulates neurotransmission (e.g., serotonin, norepinephrine and acetylcholine) in brain systems associated with cognition and mood. 1 , 2 , 3 , 4 The regulatory effects of estrogen on choline acetyltransferase (ChAT) and acetylcholinesterase (AChE) are particularly relevant to cognition given the key role of cholinergic transmission in learning, memory, and Alzheimer’s disease. 5 , 6 , 7 Clinical studies demonstrate effects of estrogen supplementation and withdrawal on mood and cognitive outcomes, but the magnitude and direction of these effects appears to be influenced by the dose, timing and formulation of estrogen treatment. 8 , 9 There is notable variability in clinical response to hormone treatments, including hormone therapy for menopausal symptoms 10 and tamoxifen treatment for breast cancer. 11 , 12 Moreover, there are significant individual differences in the extent to which cognitive function and mood change as a function of reproductive stage 13 , 14 and a high degree of inter- and intra-individual variability in the risk and severity of postpartum depression 15 and premenstrual dysphoric disorder. 16 , 17 This review explores the extent to which individual differences in genes associated with estrogen signaling predict cognitive and mood outcomes. Estrogen activity in the brain is mediated through activation of intracellular, transmembrane, and membrane-bound estrogen receptors (ERs) along with non-genomic mechanisms. In addition to the transmembrane estrogen G protein coupled receptor GPR30, there are two subtypes of estrogen receptors, ERα and ERβ. 18 ERα and ERβ receptor subtypes are members of the superfamily of nuclear receptors that regulate transcription in target genes containing estrogen response binding elements. The ER receptor subtypes are encoded by two distinct genes. The ERα gene ( ESR1 ) is located on chromosome 6q25.1, and the ERβ ( ESR2 ) gene is located on chromosome 14q22–24. 19 Both ERα and ERβ receptor subtypes are located in brain regions associated with cognitive function and emotion. The ERα receptor is predominantly expressed in the hypothalamus and amygdala, areas involved in autonomic function, emotional regulation, and associative and emotional memory. The ERβ receptor is predominantly expressed in the hippocampal formation and entorhinal cortex, brain areas involved in declarative memory. 18 The high level of ERβ expression in the hippocampus suggests a role for ERβ in cognitive function; however, mental health studies have mainly focused on the ERα gene. Studies in large populations have reported independent relationships between ESR2 polymorphisms and risk for depression in adolescents 22 andAlzheimer’s disease 20 , 21 . These associations must be considered preliminary because ESR2 studies are few in number. Moreover, subsequent studies have failed to replicate the findings associating ESR2 polymorphisms and mental health outcomes. 23 , 24 , 25 GPR30 polymorphisms have been studied in relation to breast cancer, but not mental health outcomes. 26 For these reasons, the present review focuses on polymorphisms in the ESR1 gene and their role in mental health. Several single nucleotide polymorphisms (SNPs) and variable-number tandem repeat (VNTR) polymorphisms have been identified in ESR1 . 27 Among these identified polymorphisms, only a few have been extensively studied in relation to health outcomes. One well-studied polymorphism is a TA-variable number of tandem repeats (VNTR) in the promoter region that may impact tissue-specific gene expression. 28 This TA-repeat polymorphism is in strong linkage disequilibrium with two of the most widely studied SNPs in the ESR1 gene, PvuII (rs9340799) and Xbal (rs223493), which are located in the first intron of ESR1 and are in strong linkage disequilibrium with each other. 29 See Figure 1 for a schematic figure of the genomic organization of the ESR1 gene. In linkage disequilibrium, some combinations of alleles occur more frequently in a population than would be expected from random assortment. Therefore, because the PvuII and Xbal are in linkage disequilibrium, one can predict the allele at one SNP given the allele at the other SNP with a high level of confidence. The T and C allele of the PvuII SNP are commonly referred to as the p and P allele, respectively. The A and G allele of Xbal are commonly referred to as the x and X allele, respectively. The X and P are the variant alleles that lack the respective XbaI and PvuII restriction enzyme recognition site. Currently, there is no definitive evidence concerning the functionality of these SNPs or the biological pathways they affect; however, recent findings support the hypothesis that these SNPs impact estrogen activity by influencing transcription of the ESR1 gene via altered transcription factor binding. 30 , 31 It is possible that the Xbal and PvuII SNPs as well as the TA-repeat do not affect estrogen signaling directly but may be in linkage disequilibrium with another functional variant in the ESR1 gene. The ESR1 polymorphisms appear to be markers for important clinical outcomes in women such as bone mineral density 29 , 32 , 33 , 34 and cardiovascular health. 30 , 35 , 36 , 37 The polymorphisms have been implicated in estrogen-related diseases including breast cancer, 38 , 39 , 40 osteoporosis, 29 , 33 , 41 , 42 endometriosis 43 , 44 , 45 , 46 and familiar premature ovarian dysfunction. 47 These associations are controversial because of inconsistent findings across studies. 48 , 49 It is difficult to determine which alleles are associated with physiological states that are consistent with optimal estrogen signaling. The majority of studies indicate that the p and x alleles appear to confer risk for breast cancer and osteoporosis, yet reduce risk for endometriosis. The associations between the TA-repeat polymorphism and health outcomes appear to be more consistent. Typically, a lower number of repeats is associated with lower bone mineral density and a higher incidence of osteoporotic fractures. 29 , 33 , 42 Additionally, a lower number of repeats confer a greater risk of endometriosis 43 , 44 , 45 and premature ovarian dysfunction. 47 ESR1 associations are frequently found to be ethnic/race specific. For example, a recent meta-analysis concluded that most studies in Asian populations found an association between the PP and/or xx genotypes with increased risk of osteoporosis, whereas the pp and/or xx genotypes were associated with increased risk of osteoporosis in Caucasian populations. 50 A genetic haplotype is a combination of alleles from different loci along a chromosome that tend to be inherited together. It is reasoned that the specificity of ESR1 relationships by race or ethnicity may reflect the two-fold higher prevalence of the Px haplotype in Asian populations relative to Caucasian populations. 42 Two recent molecular studies have supported a relationship between the p allele and risk of osteoporosis. A functional binding site for the transcription factor B-myb was found to be absent with the p allele of PvuII, which, in turn, may reduce ESR1 transcription rates or produce a functionally different ESR1 isoform. 31 , 35 Higher plasma estradiol levels were found in postmenopausal women carrying the XP haplotype 31 ; however, other studies found no association between serum estradiol levels and ESR1 polymorphisms. 51 , 52 Taken together, the current literature suggests that ESR1 transcription activity and/or gene regulation are affected by ESR1 polymorphisms, but the exact biological mechanism remains to be determined. There is a striking similarity between the regional distribution of ERs and brain areas affected by Alzheimer’s disease (AD) neuropathology including the hippocampus, entorhinal cortex and the basal forebrain. 1 , 53 Decreased levels of ERα have been found in hippocampal neurons in the brains of AD patients. 54 Mice lacking the ERα (i.e., ESR1 knockout mice) show impaired performance on cognitive tasks dependent on the hippocampus, a brain region that is critical for memory formation. 55 , 56 This cognitive impairment is reversed with estradiol administration 56 or with lentiviral delivery of the ESR1 gene to the hippocampus. 57 Such evidence that ERα plays a role in cognitive function has fueled a series of studies examining the relationship between ESR1 genotype and cognitive function. Fewer studies have examined the association between ESR1 genotype and mood outcomes. The present review aims to facilitate future research in relationships between genetic variation in the ESR1 gene and mental health by summarizing the current findings. This review focuses primarily on findings pertaining to the most widely studied polymorphisms, PvuII, Xbal and the TA-repeat; however, findings regarding other ESR1 polymorphisms are reported when they were investigated in conjunction with PvuII, Xbal, and/or the TA-repeat. Additionally, we offer potential explanations for discrepant results, discuss potential clinical implications, and suggest areas of focus for future research and methodology that may be advantageous in further characterizing relationships between ER genes and mental health.

For this systematic review of ESR1 polymorphisms and mental health, we focused on studies relating ESR1 polymorphisms to cognitive and mood outcomes. We carried out a literature search from 1995 to November 2009 through PubMed, http://www.ncbi.nlm.nih.gov/pubmed/ , EMBASE, http://www.embase.com/ and PsychINFO, http://www.apa.org/pubs/databases/psycinfo/index.aspx . We also consulted previous reviews. We selected only publications in the English language and used combinations of the following key words “estrogen receptor polymorphisms” “ ESR1 polymorphisms”, “estrogen receptor alpha”, “rs2234693”, “rs9340799”, “cognition”, “cognitive function”, “memory”, “mood”, “depression”, and “mental disorders.” This resulted in 136 publications for which abstracts were reviewed to identify relevance to the topic of ESR1 polymorphisms and cognition and mood. From this initial list we identified 25 manuscripts that were published in English and summarize investigations of ESR1 in mental health as described above. The studies included were heterogeneous in study design, sample size and the statistical covariance of factors other than genetics that influence estrogen signaling (e.g., reproductive stage, age, use of hormone therapy or psychotropic medication). This heterogeneity among ESR1 studies precluded conducting a formal meta-analysis. We therefore summarize findings by study design in a narrative review. Details of the studies investigating ESR1 polymorphisms and cognitive function are summarized in Tables 1 and 2 . Details of studies investigating ESR1 polymorphisms and mood outcomes are summarized in Table 3 . Figures 2 and 3 present a basic summary of the main findings regarding ESR1 allele-specific relationships with cognitive and mood outcomes.

Twin and family-based studies indicate a significant genetic component to Major Depressive Disorder (MDD), and heritability is highest with severe, recurrent and early-onset forms of the disease. 82 , 83 , 84 , 85 Epidemiologic studies have consistently demonstrated a higher prevalence for MDD in females compared to males. 86 , 87 This sex disparity emerges at puberty. 88 , 89 Multiple lines of evidence have implicated estrogen signaling as contributing to this sex difference. 90 , 91 , 92 Females show an increased risk of MDD during the perimenopausal period, a reproductive stage characterized by fluctuations in estradiol. 93 Estrogen modulates serotonergic and norepinephrinergic function 94 , both of which are central to affective processing. Lastly, of the two estrogen receptor subtypes, ERα is more predominantly expressed in brain areas that mediate affective and motivational processing including the amygdala and hypothalamus. Furthermore, as noted earlier, smaller amygdala and hippocampal volumes were associated with the p and x alleles of the PvuII and Xbal SNPs in older, healthy women. 67 Table 3 summarizes the studies to date examining the link between ESR1 polymorphisms and mood outcomes. Figure 3 illustrates that 60% of studies examining ESR1 polymorphisms in relation to anxiety or depressive symptoms found significant associations. Each of these positive studies reported that the X and/or P alleles of Xbal and PvuII or longer TA repeat alleles were risk factors for anxiety or depression. A population-based cohort study of Caucasians aged 55 and older found that the Xbal and PvuII polymorphisms related to anxiety in women but not depressive symptoms or DSM-IV depressive disorder in women or men. 95 The XP haplotype was associated with an increased risk for anxiety symptoms in women in a dose-dependent manner. 95 A Chinese study involving patients with MDD and healthy controls reported that a 3-fold higher risk of MDD in women who were homozygous for the P allele of the PvuII variant compared to women with Pp or pp genotypes. 96 No such relationships were evident in men. In a sample of adult males treated for substance abuse, the ESR1 TA-repeat polymorphism was significantly associated with anxiety symptoms. 97 There is variable evidence of a relationship between ESR1 polymorphisms and depression or anxiety in female-only samples. There was no evidence of an association between ESR1 SNPs and depressive symptoms as measured by the Center for Epidemiological Studies Depression Scale (CESD) in a multiracial/multiethnic sample of women participating in SWAN. 24 Importantly, women taking antidepressant medication were excluded from these analyses which may represent a form of selection bias if, compared to the study population, there is a higher prevalence of the genetic risk factor among the women excluded due to use of antidepressants. A recent study examined the relationship between premenstrual dysphoric disorder and genetic variation in the ESR1 gene, the ESR2 gene, and the catechol-o-methyltransferase ( COMT ) gene, a gene involved in estrogen metabolism. 98 Four SNPs in intron 4 of ESR1 showed significantly different allele distributions and genotype distributions between affected women and controls. 98 Additional studies have investigated ESR1 polymorphisms and prevalence of a mood disorder and personality traits reflecting dysphoric mood states. PvuII and Xbal variants did not predict odds of developing a childhood-onset mood disorder in a large family-based sample of affected children and their immediate family members. 99 In females, however, there was an association between childhood-onset mood disorders and a haplotype of three other ESR1 SNPs in strong linkage disequilibrium (rs746432, rs2077647, rs532010). 99 No relationship between the PvuII variant and MDD or Bipolar Disorder was found in a Japanese study. 100 No relationship between ESR1 polymorphisms and suicidal behaviors was found in a German population. 101 The paucity of studies examining ESR1 genotype and mood outcomes necessitates caution when interpreting the results; however, the findings offer preliminary evidence that the ESR1 gene may play a role in susceptibility to mood disorders, particularly for females. It becomes especially important to investigate the influence of ER polymorphisms in clinical populations that are prone to mood disorders such as HIV, dementia, chronic inflammatory conditions and cancer patients. Such investigations will bring us closer to the possibility of using genetic markers to identify patients who are at increased risk for mood disorders.

There appears to be a strong relationship between ESR1 polymorphisms and cognitive outcomes. Initial studies investigating ESR1 variants and mood disorders suggest a similar trend. As with the literature on osteoporotic fracture risk, the literature on cognitive risk factors shows some inconsistencies concerning which particular ESR1 polymorphisms conferred risk and the directionality of allele-specific effects. Most, but not all, of the case-control studies indicate that the X and P alleles of XbaI and PvuII confer the risk for cognitive impairment. Conversely, three prospective cohort studies in a healthy, older adult population suggest that the x and p alleles confer the risk. Despite these inconsistencies, two findings appear to be reliable; first that the risk for cognitive impairment is found primarily in women and second that the risk is for AD rather than for other causes of dementia. Lastly, many of the studies of ESR1 polymorphisms and mental health outcomes found that relationships were evident in an exclusively female sample or that females drove such effects in mixed gender samples. Future studies should acquire sufficiently large sample sizes to permit gender-stratified analyses. Additional insights into the discordant results will come from future studies that include haplotype analyses and analyses by ethnic/racial populations. Investigations involving ESR1 haplotypes and epistatic relationships appear to produce more reliable and robust results compared to investigations involving single SNP analyses. The majority of studies using haplotype analyses found stronger gene associations when considering combinations of ESR1 PvuII and Xbal polymorphism genotypes and combinations of ESR1 genotypes with APOE polymorphisms. ESR1 haplotypes and epistatic relationships may relate more strongly to functional outcomes than single SNP analyses, because they provide more information on other genetic variants that influence estrogen activity through different mechanisms. Many studies report ethnicity- or race-specific gene associations, which underscore the importance of considering genetic background and population stratification. There is minor racial and ethnic variation in distributions of the PvuII and Xbal genotypes; 102 however, the degree of linkage disequilibrium between ESR1 polymorphisms may vary between ethnic/race groups. There is a two-fold higher prevalence of the Px haplotype in Asian populations relative to Caucasian populations. 42 Additionally, the number of TA repeats differs by ethnicity, with estimated major peaks at 14 repeats in European populations and 15 repeats in Asian populations. 50 Moreover, studies indicate that there are ethnic disparities in the presentation and severity of menopause symptoms, particularly psychosomatic symptoms and hot flashes 13 , 103 , 104 suggesting that ethnicity/race may be a factor influencing one’s sensitivity to sex hormone fluctuations. The variations in menopausal symptoms and the distribution of ESR1 genotypes by ethnicity/race underscore the need to control for population stratification or heterogeneity in future studies. Two major differences between the case-control and cohort studies allow for further speculation into the disparate results. Across all studies, the average age of participants in the cohort studies was somewhat younger than in the case-control studies. Additionally, the majority of case-control studies involved AD outcomes whereas cohort studies involved cognitive decline outcomes. The case-control findings might therefore represent the impact of the SNPs on the more severe form of cognitive impairment in an elderly population, whereas the cohort findings represent the impact of the SNPs on generalized cognitive aging in a mid-life to elderly population. This pattern might then suggest that the optimal level of estrogen signaling is one level in an intact brain (or earlier in life) and a different level in a diseased brain (or later in life). Such a conclusion would be similar to the “healthy cell bias” of estrogen, 105 which proposes that estrogen can only exert neuroprotective effects in a healthy brain that is free of neurodegeneration. Although the literature on ESR1 and mood is sparse, there is preliminary evidence that mood varies with ESR1 genotype. One study found an association between anxiety symptoms and the XP haplotype in women, 95 while another study demonstrated an increased risk of MDD diagnosis in women homozygous for the P allele. 96 Other ESR1 SNPs have been associated with childhood-onset mood disorders 99 and premenstrual dysphoric disorder. 98 Among males treated for substance abuse, the TA-repeat polymorphism was associated with anxiety symptoms. 97 To date, no studies have examined the interaction between estrogen receptor polymorphisms and serotonin signaling genes such as the serotonin transporter gene ( 5HTT ). Estrogen regulates serotonergic synthesis, degradation and neuron firing rates in the raphe nuclei. 106 Some studies have found that the clinical response to serotonin reuptake inhibitors (SSRIs) is negatively affected by hypoestrogenic states such as menopause. 107 , 108 , 109 A few studies showed improved clinical response to SSRIs in postmenopausal women when the medication regime was supplemented with hormone therapy compared to placebo. 109 , 110 , 111 Given that neurocognitive dysfunction is typically recognized as a core feature of MDD, it is worthwhile to consider the potential interactions among ESR1 polymorphisms, mood outcomes and cognitive impairment. A history of depressive symptoms is common in about 20 to 30% of individuals dementia. 112 Hippocampal volume loss, a characteristic of AD, is also reliably observed in patients with MDD. 113 , 114 , 115 , 116 Hyperactivity of the hypothalamic-pituitary-adrenal (HPA) axis affects hippocampal and amygdala volume and is evident in both depression and AD each stem in part from. 117 Despite these potential links, there is no strong evidence to date that ESR1 effects on cognitive impairment are related to effects on mood. For example, the x and p alleles were risk factors for memory decline in Caucasian and African American postmenopausal women enrolled in SWAN 23 but were not associated with an increased risk of elevated depression scores. 24 Across non-Asian populations, x and p are risk factors for cognitive decline but are associated with decreased risk of depression. There are many more studies linking mental health outcomes to ESR1 SNPs than to ESR2 SNPs. This discrepancy may suggest that ERα receptor is a stronger determinant of mental health outcomes or that there is a publication bias against negative ESR2 findings. Although the ERα and ERβ are similar in structure and function, mechanistic differences between the two receptor subtypes may augment or minimize the effects of polymorphisms on cognition and mood. These differences include the binding and transcription efficacy for non-steroidal ligands 118 , the levels of circulating 17β estradiol needed to activate a particular promoter 119 and cofactors that help to initiate transcription. 120 Although ESR2 is highly polymorphic, there may be fewer functional polymorphisms on the ESR2 versus ESR1 gene. To date, one functional ESR2 polymorphism has been identified in an African American population that influences estrogen binding affinity. 121 Future molecular and genetic association studies with the ESR2 gene will help to elucidate these possibilities. Finally, no studies have investigated mental health outcomes in relation to the GPR30 polymorphisms, though GPR30 is found in the cortex, hippocampus, and locus coeruleus of rodent brain. Lastly, there is a need for studies focusing on the clinical significance of the ESR1 polymorphisms in relation to exogenous and endogenous estrogen levels. Bone mineral density has been shown to be more dependent on estradiol levels in men with the X and P alleles compared to men with the xx or pp genotypes. 122 Additionally, women with the XX or PP genotypes show an augmented response of high-density lipoprotein cholesterol or serum SHBG levels to hormone therapy. 35 These findings raise the possibility that ESR1 genotype might modulate the relationship between endogenous or exogenous estrogen and mental health outcomes. In this way, genetic studies can provide additional insights into inter- and intra-individual differences in the extent to which cognition and mood change with menopause and hormone therapy.

There appears to be strong evidence for a relationship between genetic variants on the ESR1 gene and cognitive outcomes. The relationships between ESR1 and cognitive impairment tend to be specific to or driven by women and restricted to risk for Alzheimer’s disease rather than other dementia causes. Case-control studies suggest that the X and P alleles of the Xbal and PvuII polymorphisms confer risk, whereas cohort studies suggest that the x and p alleles confer risk. Part of this discrepancy appears to be due to the greater prevalence of Asian cohorts among the case-control studies, as demonstrated in a meta-analysis of case-control studies that demonstrated that X and P are risk factors only in Asian cohorts. 66 Future studies that include haplotype, ethnicity-specific, and gender-stratified analyses may provide insights into the discrepant results. Most studies examined either Alzheimer’s disease or MMSE scores as primary outcomes; only two studies examined comprehensive neuropsychological test batteries. Episodic memory impairment has been shown in numerous studies to be the earliest and strongest neurocognitive predictor of Alzheimer’s disease. In the one study to examine ESR1 polymorphisms in relation to episodic memory performance, the x and p alleles were uniquely associated with an increased risk of episodic memory impairment in non-demented postmenopausal women. Future studies are needed to evaluate ESR1 polymorphisms are reliably associated with a lower level of memory performance in non-demented individuals. Fewer studies have examined ESR1 variants in relation to mood outcomes, and these studies offer preliminary evidence for a relationship between ESR1 variants and risk for certain mood disorders including anxiety, depression, childhood-onset mood disorder and premenstrual dysphoric disorder. Future studies are needed to explicitly evaluate whether ESR1 variants may help to explain the significant individual differences in the extent to which mood and cognition change as a function of reproductive stage, menstrual phase, postpartum, and hormone therapy.