INTRODUCTION

Estrogen (E2) has a wide range of effects in body and brain, and its therapeutic potential for mood, among other physiological and psychological processes, has been recognized for some time. Indeed, one of the initial investigations of how mood may be improved by E2 was published over 100 years ago and details 36 case studies of women who had various neuropsychological conditions and were administered ovarian extracts (Easterbrook, 1900). Despite such a history of interest in this topic, many questions remain about the effects, brain targets, and mechanisms of E2 for its effects on mood/affect. This is a clinically relevant question. People are living longer (especially women), and age of menopause has remained stable, such that many women will be postmenopausal, with low or declining E2 levels, for nearly half their lives. Recent controversies on the clinical utility of E2-based hormone therapies further justify the need to investigate E2's effects, mechanisms, and brain targets (Brunner et al, 2005; Hays et al, 2003; Rapp et al, 2003; Rossouw et al, 2002; Shumaker et al, 2003; Smoller et al, 2003). Studies investigating E2's effects on mood in women and limitations to these approaches will be discussed. Given that animal models may provide a means to address these limitations and characterize E2's effects and actions, this literature will also be reviewed and recent findings from our laboratory will be discussed.

Overview of Clinical Findings of E2's Effects to Alter Anxiety and Depression Among Women

Gender differences

Women may be more vulnerable than men to develop anxiety or depression disorders. There is a greater incidence of most types of anxiety disorders (ie social anxiety, phobias, post-traumatic stress disorder, general anxiety disorder) among women compared to men (Breslau et al, 1995; Kessler et al, 1994; Schneier et al, 1992; Seeman, 1997). Women are twice as likely to experience major depression, particularly unipolar depression, compared to men (Earls, 1987; Kessler et al, 1993; Nolen-Hoeksema, 1987). Among people with depression, depressive episodes are more protracted and recur more frequently in women than in men (Earls, 1987; Nolen-Hoeksema, 1987). Women's increased vulnerability to these mood disorders are especially apparent in major depression with comorbid anxiety disorders (Breslau et al, 1995). Given that anxiety may be accompanied by symptoms of depression and be a precursor for development of depression (Paul, 1988), it is important to consider both anxiety and depression when investigating factors, such as E2, that may alter mood (reviewed by Seeman, 1997; Young, 1998; Young and Korszun, 2002). Women's increased vulnerability to mood disorders occurs postpubertally, with the beginning of cyclical changes in E2 secretion from the ovaries (reviewed by Hayward and Sanborn, 2002; Kessler and Walters, 1998; Lewinsohn et al, 1998). Plasma E2 levels are significantly lower among depressed women (Young et al, 2000). Thus, E2 may precipitate the increased incidence and/or symptomology of mood disorders in women.

Effects of endogenous changes in E2 levels

Changes in endogenous E2 levels may increase women's susceptibility to anxiety and depression disorders. Women are uniquely at risk for mood disorders, such as premenstrual dysphoric disorder and postpartum depression, that typically occur concurrent with changes in endogenous E2 levels (Bebbington et al, 1981; reviewed by Bloch et al, 2003; Jenkins, 1987; reviewed by Rubinow and Schmidt, 1995; Weissman and Klerman, 1977). In one study, nearly half of the women admitted to psychiatric hospitals for a variety of symptoms were admitted immediately before or during menstruation, and the incidence of suicide was greater at these times (Dalton, 1959). Some, but not all, studies report higher suicide incidence among women during the luteal, compared to follicular, phase (as reviewed by Baca-Garcia et al, 2000). Furthermore, symptoms of postpartum depression typically occur within the 1st week after parturition when E2 levels precipitously decline from the high levels during pregnancy (Dean and Kendell, 1981). Although these data suggest that changes in E2 levels may underlie the increased susceptibility of women to develop affective disorders, these changes occur concomitant with alterations in levels of progesterone, androgens, and their metabolites, which may also alter mood (Roca et al, 2003). As such, it is important to consider the effects of E2 administration to women with low E2 levels to establish the role of E2 on mood.

Effects of E2 levels on mood with aging

Dramatic changes in E2 levels, as occur with surgical and/or natural menopause, are associated with changes in incidence and symptomology of anxiety and depression. During perimenopause, there are alterations in ovarian function, such that E2 levels can be variable (Burger et al, 1995; Daly et al, 2003; Santoro et al, 1996). This is followed by the postmenopausal period when E2 levels are declining or at nadir. In one longitudinal study, 29 asymptomatic, regularly cycling, premenopausal women were monitored for an average of 5 years until they were amenorrheic for 6 or more months (Schmidt et al, 2004). There were 11 episodes of new-onset depression, as determined by subjective mood ratings and the Structured Clinical Interview for DSM-IV, in nine of the 29 women. Furthermore, during the late perimenopause period in these women (ie 24 months surrounding their last menstruation), the majority of the episodes (nine) of depression occurred. These data suggest that incidence of depression among some women may be increased during late perimenopause. Incidence of anxiety and depression disorders increases postmenopausally. Generalized anxiety disorder occurs in 5% of the population, but the incidence doubles among older, postmenopausal women (Bebbington et al, 1981; Jenkins, 1987; Weissman and Klerman, 1977; Wittchen and Hoyer, 2001). Incidence of bipolar and major depression are increased among postmenopausal women (Bebbington et al, 1998; reviewed by Goodwin and Jamison, 1990; Weissman and Olfson, 1995). In a cross-sectional study of older postmenopausal women, Beck Depression Inventory scores were inversely related to plasma E2 levels. Those in the lower-half of the distribution had higher scores than those in the top-half (Almeida et al, 2005).

E2 replacement to nondepressed, naturally- or surgically menopausal women can improve mood scores, although not all studies find such improvement (Ditkoff et al, 1991; Heinrich and Wolf, 2005; Miller et al, 2002; Morrison et al, 2004; Rausch and Parry, 1993; Sherwin, 1991; Sherwin and Gelfand, 1985). Recent double-blind, placebo-controlled studies of perimenopausal women with depression demonstrated that E2 therapy may provide some mood benefits among older women (Cohen et al, 2003; Schmidt et al, 2000; Soares et al, 2001). Higher scores on the Geriatric Depression Scale (ie greater depressive symptoms) were found among older, postmenopausal women diagnosed with Alzheimer's Disease, particularly among those who were not on E2 therapy, compared to healthy controls (Carlson et al, 2000). Thus, low endogenous E2 levels with aging may be associated with anxiety and/or depression disorders, and E2 may counter some of these effects.

Effects of E2 administration to women with low E2 levels

E2 administration to young women with low E2 levels may alter mood. Transdermal E2 decreased self-reported negative mood in women with severe premenstrual syndrome (Smith et al, 1995). In women with postpartum depression, administration of sublingual or transdermal E2 sufficiently improves depressive symptoms to meet the definition of clinical recovery (Ahokas et al, 2001; Gregoire et al, 1996). Young, adult women diagnosed with major depressive disorder given E2 treatment had improved affect (Klaiber et al, 1979). These data suggest that E2 can have beneficial effects among women with intact neuroendocrine feedback. Additionally, there are similar effects of E2 administration to improve mood among women with low E2 levels. In support, E2, alone or in combination with the selective serotonin reuptake inhibitor, fluvoxamine, significantly decreased self-rated depression scores among oophorectomized women with depressive symptoms (Nagata et al, 2005). Furthermore, among women without pre-existing neuropsychiatric conditions, gonadotropin-releasing hormone agonists, which decrease E2 levels, increase depressive mood symptoms (as reviewed by Warnock et al, 2000). However, a different pattern of effects is observed in women that have been diagnosed with mood disorders, as discussed in detail below.

Limitations to investigating E2's effects on mood of women

The data discussed above focus on the beneficial effects of E2 administration to individuals with low and/or absent E2 levels; however, not all individuals respond favorably to E2. Among women, psychiatric history and current and prior E2 exposure (eg length of time in an E2-deficient, postmenopausal state before initiation of E2 therapy) may alter the responses to E2 (Klaiber et al, 1997). Some women with anxiety disorders report less anxiety when E2 levels are low and/or stable (Schmidt et al, 1998), rather than rising, suggesting that some individuals may be more sensitive to E2 than are others. Indeed, women with postpartum depression or premenstrual syndrome respond favorably to gonadotropin-releasing hormone agonists, which stabilize E2 levels, unlike women without these disorders (Bloch et al, 2000; Schmidt et al, 1998). Furthermore, in two double-blind, placebo-controlled studies of E2 replacement, baseline E2 levels or those produced by E2 therapy did not predict a favorable response to E2 for mood scores, despite women having similar diagnoses and meeting endocrine criteria for perimenopause (Schmidt et al, 2000; Soares et al, 2001). In an in-patient study of severely depressed pre- and postmenopausal women, oral conjugated E2 therapy, compared to placebo, improved ratings on the Hamilton Scale of Depression; however, the response to E2 varied as a function of depression duration, such that women with a shorter history of depression had a better response to E2 than those with a longer history of depression (Klaiber et al, 1979). Other reports have demonstrated that the E2 regimen utilized contributes to its efficacy (Gregoire et al, 1996; Saletu et al, 1995). Negligible findings for beneficial effects of E2 on mood of older, postmenopausal women in the Women's Health Initiative studies further support the idea that response to E2 may be sensitive to individual differences among women (ie differences in exposure to E2 and length of time without such exposure, psychiatric history, etc) and the regimen of E2 utilized (Brunner et al, 2005; Hays et al, 2003; Smoller et al, 2003). Moreover, E2 levels are lower among women with depression, suggesting that neuropsychiatric diagnoses may influence endocrine function (Young et al, 2000). As it is essential to control and/or evaluate these factors to elucidate role of E2, it may be more beneficial to use animal models.

Overview of Supporting Data from Animal Models for E2's Effects on Anxiety and Depression Behavior

Sex differences

There are sex differences in anxiety and depression behavior of adult rodents that may depend, in part, upon E2 levels (as reviewed by Palanza, 2001). Female rodents have increased anxiety behavior compared to males in some tasks and, when estrous cycle is not considered, opposite effects can be seen (Blanchard et al, 1992; as reviewed by Blanchard et al, 1991; Johnston and File, 1991; Zimmerberg and Farley, 1993). The magnitude of the sex difference may depend upon the motor demands of the task, given that E2 increases motor activity of female rodents (Becker et al, 1987; Frye et al, 2000; Morgan and Pfaff, 2001, 2002). Indeed, sex differences in affective behavior of rodents are more evident when endogenous changes in E2 levels during the estrous cycle are considered.

Estrous cycle changes

Changes in endogenous levels of E2 can alter anxiety and depressive behavior of rodents. Rodents show decreased anxiety and depression behavior during the late proestrous phase of the estrous cycle (ie behavioral estrus), when E2 levels are high and sexually receptive behavior is displayed. Naturally receptive rats have increased open arm time in the plus maze, increased time spent in the center of a brightly lit open field, decreased latencies to emerge from a dark chamber, increased time spent in social interaction with a conspecific, decreased freezing in response to footshock, and increased immobility in the forced swim test, compared to rats with lower endogenous E2 levels (ie nonreceptive rats in diestrus or male conspecifics; Contreras et al, 2000; Diaz-Veliz et al, 1997; Frye et al, 2000; Frye and Walf, 2002; Frye and Wawrzycki, 2003; Marcondes et al, 2001; Marvan et al, 1996, 1997; Mora et al, 1996). Additionally, pregnancy in rats is associated with persistently high E2 levels and decreased anxiety and depression behavior (Frye and Walf, 2004b; Zuluaga et al, 2005). Although these data support a role of natural variations in E2 concentrations to alter anxiety and depression behavior of rodents, these variations occur concomitant with fluctuations in progestins and androgens. Progestins and androgens also alter affective behaviors of female rodents (Frye et al, 2000, 2004; Frye and Lacey, 2001; Frye and Walf, 2002, 2004a, 2004b; Rhodes and Frye, 2001; Walf et al, 2005). As such, it is important to determine whether E2 administration to rodents with low E2 levels is sufficient to increase antianxiety and antidepressant-like behavior.

Effects of age-related decline in endogenous E2

Aged rodents provide a useful, but less studied, model to investigate the effects of decline in endogenous E2 source for anxiety and depression behavior. Female mice with lifelong low levels of E2 due to deletion of the follitropin receptor have increased anxiety behavior at ages typically associated with decline in ovarian function among wild-type mice (+20 months old; Danilovich et al, 2003). We have demonstrated that intact aged female mice have very low central E2 levels (Frye et al, 2005), which can be increased with administration of 10 μg E2 systemically. We have begun to look at the effects of E2 administration on anxiety and depression behaviors in aging by using intact female mice that are 24–28 months old. At 1 h before testing in several tasks of affective behavior, mice were administered E2 (10 μg) or vehicle (sesame oil) and then received a second injection of the selective estrogen receptor (ER) modulator (SERM), raloxifene (3 mg/kg), or vehicle. Compared to vehicle, E2 and/or raloxifene to aged mice decreases anxiety and depression behavior across several tasks (see Table 1). Together, these data suggest that decline in E2 with aging is associated with increased anxiety behavior, and aged rodents can respond favorably to E2 or SERMs.

Table 1 Affective Behavior (Mean±SEM) of Aged Female Mice Administered E2 and/or Raloxifene

Effects of extirpation and E2 administration

Ovariectomy (ovx), removal of the primary source of E2, the ovaries, is utilized as a model of E2 deprivation to assess E2's behavioral effects. Ovx increases anxiety and depression behavior and subcutaneous administration of E2 can reverse these effects in several tasks (Bernardi et al, 1989; Bowman et al, 2002; Diaz-Veliz et al, 1997; Estrada-Camarena et al, 2003; Frye and Walf, 2004a; Frye and Wawrzycki, 2003; Hilakivi-Clarke, 1996; Luine et al, 1998; Marcondes et al, 2001; McCarthy et al, 1995, 1996; Mora et al, 1996; Nomikos and Spyraki, 1988; Okada et al, 1997; Rachman et al, 1998; Slater and Blizard, 1976; Walf and Frye, 2005a, 2005b; Walf et al, 2004). Furthermore, in an animal model of postpartum decline in E2, withdrawal from chronically sustained E2 levels in ovx rats increases depressive behavior (Galea et al, 2001). Together, these data suggest that E2 can increase antianxiety and antidepressant-like behavior. However, as in women, there is evidence for experience-, dose-, and/or duration-dependent effects of E2 on anxiety and depression behavior.

Experience and regimen-dependent effects of estrogen in animal models

The length of E2 exposure and/or decline may alter responses to subsequent E2 exposure. Studies investigating the effects of parity suggest that frequent exposure to chronic E2 can alter affective behavior of rodents. In support, exploration in the open field is lower among female rats that have never experienced high E2 levels during pregnancy compared to those that have repeatedly experienced this (Wartella et al, 2003). Thus, some differences observed among rodents for their response to E2 may be due to prior E2 experience.

E2's effects on anxiety and/or depression behavior of female rodents may depend upon the E2 concentration and/or dosing utilized. Rats that are administered an E2 regimen that produces physiological E2 levels (similar to that observed in naturally receptive rats) have decreased anxiety and depression behavior (Estrada-Camarena et al, 2003; Frye and Walf, 2004a; Frye and Wawrzycki, 2003; Nomikos and Spyraki, 1988; Rachman et al, 1998; Slater and Blizard, 1976; Walf and Frye, 2005a, 2005b). In contrast, very low or high dosages of E2, or regimen that would not be expected to significantly increase circulating E2 concentrations at test time to levels observed in naturally receptive rats, generally show little or no decreases in anxiety and depression behavior (Diaz-Veliz et al, 1997, 2000; Estrada-Camarena et al, 2003; Martinez-Mota et al, 2000; Mora et al, 1996; Stoffel and Craft, 2004; Walf and Frye, 2005a). These data suggest that an E2 regimen that produces concentrations of E2 that are similar to that of naturally receptive rats decreases anxiety and depression behavior of ovx rodents.

In addition, the duration of E2 exposure may alter responses of ovx rodents. Activity in the open field is increased in ovx rats administered E2 for 5, but not 35 days, via silastic capsules (Luine et al, 1998). Subchronic (3–7 days) administration of E2 that produces physiological E2 levels, but not higher dosages, decreases anxiety behavior (Koss et al, 2004; McCarthy et al, 1995; Morgan and Pfaff, 2001, 2002; Nomikos and Spyraki, 1988; Rodriguez-Sierra et al, 1984). Similarly, chronic administration of E2, which likely produces prolonged supraphysiological or physiological plasma E2 levels, increases depressive behavior of ovx rats and mice (Galea et al, 2002; Okada et al, 1997). In another model, 10 μg E2 for 1 week or 2 μg E2 for 2 weeks, but neither higher dosages nor longer treatment, increased choline acetyltransferase immunoreactivity in the basal forebrain (Gibbs, 1997). Using the defensive burial task, in which the duration rats spend burying an electrified prod with shavings following footshock is considered an index of anxiety behavior, we have examined the effects of E2 regimen and length of exposure. We found that physiological E2 regimen to ovx rats (10 μg E2 for 2 days; 42±24 s) decreased the duration spent burying an electrified prod, but behavior was not altered by neither very low (2 μg systemic injections=322±36 s) nor very high (silastic implants filled with E2=250±80 s) E2 concentrations present for 2 days prior to testing, as compared to vehicle administration (203±41 s). If silastic implants with E2 remained in place for 4 weeks, ovx rats demonstrated a modest decrease in time spent burying (168±24 s). These data suggest that both E2 concentration and length of exposure may influence E2's functional effects. Another factor that may alter E2's effects on anxiety and depression behavior is the activity of the hypothalamic–pituitary–adrenal axis (HPA).

The role of the HPA axis in modifying the response to E2

Sex- or hormone-related disparities in anxiety and depression disorders, and individual differences in response to E2, may be related to reactivity of the HPA to stressful or threatening situations. Affective behavior of rodents is altered by differences in HPA reactivity. For instance, increasing reactivity of the HPA with gestational stress alters affective responses of rodents as adults (as reviewed by Weinstock, 2001). There are sex differences in the response of rodents to gestational stress, such that females show increased vulnerability to its negative effects. Adult females that experienced gestational stress have greater HPA reactivity (Koehl et al, 1999; McCormick et al, 1995; Szuran et al, 2000), anxiety and depression behavior (Frye and Wawrzycki, 2003; Sternberg, 1999; Takahashi et al, 1992; Weinstock et al, 1992; Weinstock, 1997) and hippocampal cell loss (Schmitz et al, 2002), than do gestationally stressed males or nonstressed conspecifics. E2's effects on antianxiety, antidepressive, and sexual behavior are attenuated in gestationally stressed compared to nonstressed female rats (Frye and Orecki, 2002a, 2002b; Frye and Wawrzycki, 2003; Walf et al, 2003). Thus, sex and individual differences in affective responses may be altered by stress, but whether there is a relationship between E2's effects on the HPA and affective behavior needs further clarification.

E2 may contribute to the normal functioning of the HPA response. There are sex differences in basal and stress-induced glucocorticoids among people and rodents, such that males typically have lower levels of glucocorticoids than do females (Critchlow et al, 1963; Gallucci et al, 1993; Handa et al, 1994; Jezova et al, 1996; Kitay, 1963). High physiological levels of E2 among women increase basal concentrations of cortisol and adrenocorticotrophic hormone (ACTH; Altemus et al, 2001; Genazzani et al, 1975; Marinari et al, 1976). Following a precipitous decline in the high, sustained levels of E2 and progestins during pregnancy, postpartum women experience greater HPA axis response to stressors (Altemus et al, 2001; reviewed by Carter et al, 2001). Naturally receptive rats have higher basal and stress-induced plasma corticosterone levels compared to rats in other stages of the estrous cycle (Carey et al, 1995; Figueiredo et al, 2002; Frye and Bayon, 1999; Raps et al, 1971; Viau and Meaney, 1991). Although HPA activity is increased with an acute rise in E2 during proestrus, it is dampened with stable physiological E2 levels during lactation (Sibolboro Mezzacappa et al, 2003; Viau and Meaney, 1991). These data suggesting that the HPA is altered by changes in endogenous E2 levels are tempered by evidence for progestins’ clear effects to reduce HPA reactivity (Roca et al, 2003; Patchev et al, 1996). It may be that E2's effects to increase production of neuroactive progestins ultimately increases inhibitory feedback on the HPA response (Cheng and Karavolas, 1973; Frye and Rhodes, 2005; Vongher and Frye, 1999).

Experience and regimen-dependent effects of E2 for HPA response

Just as there are regimen-dependent effects of E2 for affective behavior (ie acute, lower dosages decrease anxiety and depression), there are similar patterns observed for E2's modulation of the HPA response. Administration of E2 to individuals with low endogenous E2 levels alters HPA response depending on prior experience and regimen. For instance, although there is some evidence for stress responses to be enhanced among women with lower E2 levels (postnatural or surgical menopause; De Leo et al, 1998), there is also evidence that E2 therapy to postmenopausal women increases cortisol levels (Duka et al, 2000). Acute E2 administration to ovx rats reduces basal and/or stress-induced corticosterone levels, but administration of higher E2 dosages or more chronic regimen produces the opposite effect (Burgess and Handa, 1992; Carey et al, 1995; Dayas et al, 2000; Kitay, 1963; McCormick et al, 2002; Redei et al, 1994; Viau and Meaney, 1991; Walf and Frye, 2005a; Young et al, 2001). Prior exposure to E2 alters stress responses. Primiparous and multiparous rats have decreased restraint stress-induced c-Fos expression in limbic regions, such as the hippocampus, compared to nulliparous rats (Wartella et al, 2003). Notably, acute stress exposure can increase biosynthesis of E2 (Shors et al, 1999). Thus, there may be an optimal level of E2 to dampen HPA reactivity, and whether this effect occurs concomitant with changes in anxiety and depressive behavior is of interest. Given these regimen-dependent effects of E2 on HPA function, it is essential to manipulate and measure both E2 and corticosterone responses to begin to address this question.

Our laboratory investigated whether the regimen-dependent effects of E2 for anxiety and depression behavior may be modified by changes in HPA reactivity. Ovx rats were administered vehicle, low (2 μg), moderate (5 or 10 μg), or high (20 or 50 μg) doses of E2. Anxiety and depression behavior and E2 and corticosterone levels were measured following two types of HPA manipulation (acute stress or adrenalectomy (ADX) with administration of low or high corticosterone concentration in drinking water or saline; Walf and Frye, 2005a). Antianxiety- and antidepressant-like effects of 5 or 10 μg E2, compared to lower or higher dosages, occurred concomitant with lower corticosterone levels. There was an interaction between E2 dosage and stress exposure for anxiety and depression behavior, such that antianxiety- and antidepressant-like effects of 5 or 10 μg E2, but not lower or higher dosages of E2, were attenuated with acute restraint stress, which increased plasma corticosterone. As well, reduced anxiety and depression behavior due to administration of moderate dosage of E2 was not apparent in ADX rats that were not administered corticosterone or those administered high dosages of corticosterone, which produced stress-like plasma corticosterone levels. Together, these data suggest that negative feedback of the HPA may be important for E2's effects on anxiety and depression. Indeed, E2 can directly act on the adrenal gland and central HPA targets to alter HPA feedback (Figueiredo et al, 2002, 2003). It is likely that E2 is acting via central HPA targets upstream of corticosterone. First, effects of ADX and corticosterone-replacement suggest that an intact HPA-negative feedback mechanism is required for these effects of E2 and corticosterone replacement does not abrogate all effects of ADX. Second, restraint stress, which increases negative feedback, attenuates E2's antianxiety- and antidepressant-like effects. However, the restraint paradigm that we utilized produced only modest increases in plasma corticosterone. Together, these data suggest that E2's regimen-dependent effects on affective behavior that occur concomitant with, and may rely on, HPA responses may underlie some of the individual differences in response to E2. These data further suggest which brain areas that may be targets of E2 for these effects, namely the hippocampus and amygdala.

CNS Sites that are Involved in E2's Effects on Anxiety and Depressive Behavior

Although the brain areas that mediate central actions of E2 for affective behavior are not well defined, the hippocampus and amygdala are putative sites for these effects. Both the amygdala and hippocampus have long been considered important components of the limbic system and regulators of the HPA response (LeDoux, 2000; Walker et al, 2003). Manipulations in these regions of progestins in female rats and androgens in male rats produces robust changes in anxiety/fear and depression behavior (Bitran et al, 1999, 2000; Edinger and Frye, 2004, 2005; Frye and Walf, 2002, 2004a, 2004b; Rhodes and Frye, 2001; Walf and Frye, 2003; Walf et al, 2005), suggesting a role of these brain regions for other steroids’ effects on anxiety and depression.

The role of the hippocampus in E2's effects on anxiety and depression behavior

The hippocampus is a target of E2. Radioactively labeled E2 injected into female rats is concentrated in the hippocampus (Pfaff and Keiner, 1973). E2 administration increases activity in the hippocampus as evidenced by increased immunohistochemical staining for the immediate early gene, c-fos (Rudick and Woolley, 2000). E2 also alters the plasticity of the hippocampus. Levels of brain-derived neurotrophic factor in the hippocampus fluctuate across the estrous cycle and are increased in ovx rats following administration of E2 (Gibbs, 1998, 1999). The density of dendritic spines in the hippocampus are increased in naturally receptive rats, or after E2 administration to ovx rats (Gould et al, 1990; MacLusky et al, 2005; Woolley et al, 1990; Woolley and McEwen, 1993). Notably, typical antidepressant treatments increase neurogenesis in the hippocampus (reviewed by Duman et al, 2001).

Behaviors that rely on hippocampal function are sensitive to E2 treatment. E2 administration enhances performance in several hippocampus-dependent cognitive tasks among young ovx or aged female rodents (Bowman et al, 2002; Frick et al, 2002; Frye et al, 2005; Frye and Rhodes, 2002; Gibbs et al, 2004; Li et al, 2004; Luine et al, 2003; as reviewed by Packard, 1998; Rhodes and Frye, 2004). Thus, the hippocampus is a likely target of E2 for its behavioral effects.

Our laboratory has investigated whether the hippocampus is an intergral brain site for E2's effects for anxiety and depression behavior by directly administering E2 to this region. E2, when administered to the hippocampus or subcutaneously, increased antianxiety- and antidepressant-like behavior compared to vehicle administration. Bilateral application of cannulae inserts filled with 17β-E2 to the dorsal hippocampus increased central entries made in the open field, increased time spent on the open arms of the elevated plus maze, and decreased time spent immobile in the forced swim test (see Figure 1). Importantly, this intrahippocampal E2 regimen produced similar behavioral effects as did the systemic E2 regimen utilized, which produces physiological circulating E2 levels (Walf and Frye, 2005b). The same regimen of E2 to the ventral tegmental area did not alter open field central entries (intra-VTA E2: 2.0±1.0, intra-VTA vehicle: 4.6±1.7) or elevated plus maze open arm time (intra-VTA E2: 1.0±1.0 s, intra-VTA vehicle: 1.0±1.0 s) of ovx rats (n=6–7 per group). Thus, the hippocampus is a target for E2's antianxiety- and antidepressant-like effects.

Figure 1
figure 1

Administration of E2 decreases anxiety and depression behavior. E2 or vehicle alone was administered either subcutaneously or intrahippocampally. E2 increased central open field entries (top panel) and time spent in the open arms of the plus maze (center panel). E2 decreased the duration of immobility in the forced swim test (bottom panel). Error bars indicate one SEM. *p<0.05, E2 vs vehicle.

The role of the amygdala in E2's effects on anxiety and depression behavior

The amygdala is another limbic region that is sensitive to E2 treatment. First, cells in the amygdala have high concentrations of radioactively labeled E2 after systemic injection into female rats (Pfaff and Keiner, 1973). Second, E2 administration increases c-fos immunoreactivity in the medial amygdala (Greco et al, 2003a, 2003b; Insel, 1990). Third, the number of synapses on dendritic shafts in amygdala neurons are increased by E2 treatment (Nishizuka and Arai, 1982) and dendritic spine density in the medial amygdala fluctuates across the estrous cycle of rats (Rasia-Filho et al, 2004). These data suggest that the amygdala is an E2-sensitive CNS site that may play a role in E2's effects on anxiety and depression behavior.

E2 alters behavioral responses of rodents in affective tasks that involve the amygdala. Similar effects of subcutaneous injection and intra-amygdala E2 administration to ovx rats are observed in models of fear and nociception, such that both enhance stress-induced analgesia following acute exposure to a predator odor (Walf and Frye, 2003). Thus, the amygdala is a likely target for E2's functional effects on other emotional responses, such as anxiety/depression.

Our laboratory has investigated the role of the amygdala in E2's modulation of anxiety and depressive behavior. E2 to the medial amygdala, and not missed sites, produces similar antianxiety- and antidepressant-like effects as systemic dosing of E2 that produces physiological plasma E2 levels. Intra-amygdala E2 increased central entries in the open field and time spent on the open arms of the elevated plus maze (Frye and Walf, 2004a). The same regimen of E2 to the amygdala also decreases depressive behavior. E2 to the amygdala decreases time spent immobile (217.6±42.2 s) compared to intra-amygdala vehicle (273.6±26.6 s). Similar effects of intra-amygdala and subcutaneous E2 administration suggest that the amygdala is a target of E2 for its antianxiety- and antidepressant-like effects.

These data suggest E2 acts in the hippocampus and amygdala to decrease anxiety and depression behavior. A question that remains is the mechanism(s) of E2 for these effects.

Putative Mechanisms for E2's Effects on Anxiety and Depression

ERs as putative substrates for E2's effects on anxiety and depressive behavior

E2 may act in the hippocampus and/or amygdala to reduce anxiety and depressive behavior via traditional ligand-dependent actions at intracellular ERs. Specific binding sites, which eventually became known as ERs, for E2 were identified over 40 years ago (Jensen and Jacobsen, 1962). Like other steroid receptors, ERs function as transcription factors and their activity is modulated by E2. E2 binds to intracellular ERs, which may be located in the cytoplasm or nucleus, in a ligand-dependent manner, and, subsequently, ERs bind DNA as homodimers with the E2 response element or the activator protein 1-binding site. This results in transcription and translation of new proteins that carry out the cell's functional response (reviewed by Giguere, 2003; Falkenstein et al, 2000; O’Malley and Means, 1974). Indeed, intracellular ERs have been localized to the amygdala and hippocampus (Shughrue et al, 1997, 1998) and E2's effects at ERs in both regions are of interest.

To address whether E2 is acting at intracellular ERs in the hippocampus or amygdala to reduce anxiety and depressive behavior, our laboratory investigated the effects of blocking these receptors on anxiety and depression behavior of naturally receptive rats. Rats were administered bilateral infusions of a specific ER antagonist, ICI 182,780 (10 μg in 1 μl saline; Frye and Rhodes, 2002) to the dorsal hippocampus or medial amygdala 2 h prior to testing in the open field, elevated plus maze, and forced swim test. Compared to vehicle infusions, infusions of ICI 182,780 to the hippocampus of naturally receptive rats reduced central open field entries and open arm activity, and increased immobility in the forced swim test (see Figure 2). Infusions of ICI 182,780 to the amygdala-produced behavior that was not different from vehicle infusions. These data suggest that intracellular ERs in the hippocampus, but not the amygdala, are critical for the antianxiety- and antidepressant-like effects of E2. However, it is also possible that ICI 182,780's effects in the hippocampus were due to actions at membrane ERs (Gu et al, 1999), which has yet to be thoroughly addressed.

Figure 2
figure 2

Administration of an ER antagonist (ICI 182,780) to the hippocampus, but not the amygdala, increases anxiety and depression behavior. The antagonist applied to hippocampus decreased central open field entries (top panel) and time spent on the open arms of the plus maze (middle panel) but increased time spent immobile in the forced swim test (bottom panel). Error bars indicate one SEM. *p<0.05, antagonist vs vehicle.

The role of ER isoforms, α and β, for E2's effects on anxiety and depressive behavior

The variable effects of E2 on affective behavior may be related to E2's actions at two distinct ERs isoforms. In 1996, a second form of ER was identified (ERβ; Kuiper et al, 1996; Tremblay et al, 1997). ERα and ERβ have distinct N-terminal regions and share similar DNA- and ligand-binding domains (Tremblay et al, 1997), but they are encoded by different genes (Green et al, 1986; Kuiper et al, 1996), differentially alter gene regulation, (Kuiper et al, 1997, 1998; Mitchner et al, 1998; Paech et al, 1997; Tena-Sempere et al, 2004), and have distinct temporal patterns of expression in the body and brain (reviewed by Gustafsson, 2003; Shughrue et al, 1997). Notably, ERα and ERβ have been localized in the hippocampus and amygdala (Greco et al, 2003a, 2003b; Osterlund and Hurd, 1998; Shughrue et al, 1997, 1998), supporting further investigation of the ER-isoform-specific mechanisms of E2 in these regions.

ERβ may be required for the antianxiety- and antidepressant-like effects of E2. Studies investigating the effects of ERβ gene knockout suggest that ERβ is necessary for E2's actions on affective behavior. Female homozygous ERβ knockout mice do not respond to E2 with decreased anxiety (greater open arm activity in the plus maze) or depressive (greater immobility in the forced swim test) behavior as do their wild-type littermates; however, there is some evidence that ERβ knockout mice have increased anxiety behavior irrespective of E2 treatment (Imwalle et al, 2005; Krezel et al, 2001; Rocha et al, 2005). We have recently investigated the effects of 10 μg 17β-E2 subcutaneous injections to adult, intact wild-type and homozygous and heterozygous ERβ knockout mice when administered 48 h prior to testing in the open field and elevated plus maze. E2 administration to wild-type, but not ERβ knockout mice, increase antianxiety behavior in the open field and elevated plus maze compared to vehicle (Figure 3).

Figure 3
figure 3

Administration of E2 to wild-type, but not heterozygous (+/−ERβKO) or homozygous (−/−ERβKO) ERβ knockout mice, increases central entries in the open-field (top) and open arm duration (bottom) compared to vehicle administration. *p<0.05, E2 vs vehicle.

Administration of dietary phytoestrogens with a greater affinity for ERβ than ERα, such as genistein and daidzein, decrease anxiety behavior. Genistein exposure from gestation to adulthood decreases anxiety behavior of male and female rats in the elevated plus maze (Lephart et al, 2002; Lund and Lephart, 2001). Supplements containing daidzein and genistein for 1 or 2 weeks in adulthood increased open arm activity in adult female rats (Patisaul et al, 2005). However, exposure to genistein or daidzein for 18 days reduced time spent interacting with a conspecific and open arm activity in the plus maze, and significantly increased stress-induced corticosterone secretion (Forsling et al, 2003; Hartley et al, 2003). These differences may reflect the effects of exposure duration and/or concentration of phytoestrogens and the resulting effects at ERα and/or ERβ to alter the HPA. Although the ability of stress to alter individuals’ response to E2 may not clearly reflect a direct relationship of the HPA and E2's effects on anxiety and depression, there is evidence of a regulatory role of ERs, in particular ERβ, in the HPA. The paraventricular nucleus of the hypothalamus has high levels of ERβ mRNA and corticotropin-releasing hormone cells coexpress ERβ (Isgor et al, 2003; Suzuki and Handa, 2004). Administration of an ER antagonist, tamoxifen, which has antagonistic actions via ERβ, blocks the ability of systemic E2 to reduce stress-induced ACTH and corticosterone levels of ovx rats (Watanabe et al, 1997; Young et al, 2001). Indeed, actions at ERβ may reduce HPA reactivity more than that observed with ERα. In support, stress-induced corticosterone levels of ovx rats are lower following administration of an ERβ-selective ligand (DPN) compared to 17β-E2 or a ERα-selective ligand (PPT; Lund et al, 2005). Thus, some of the exposure-dependent effects of phytoestrogens with greater activity at ERβ than ERα on anxiety may be related to differential ability of these receptor isoforms to modulate the HPA.

SERMs that have differential binding affinity for ERα and ERβ provide a tool to investigate the importance of ER isoforms for functional effects of E2. We and others have shown that administration of 17β-E2 (which has high affinity for both ERα and ERβ), compared to vehicle or ERα-specific SERMs, similarly increases antianxiety and antidepressant-like effects in several tasks as do SERMs with greater specificity for ERβ than ERα (Lund et al, 2005; Walf et al, 2004; Walf and Frye, 2005b). The potential for a modulatory action of ERβ on ERα and the ability of ERα and ERβ to form functional heterodimers (reviewed by Giguere, 2003; Lindberg et al, 2003) are factors that may underlie E2's actions involving ERβ in the hippocampus for affective behavior and requires further investigation.

Another possibility is that E2 may act by targeting mitochondria and altering cellular energy stores. ERβ have been localized to the mitochondria in peripheral and CNS tissue (Yang et al, 2004). Neurons are dependent almost entirely on mitochondrial ATP for their energy demands. It has been suggested that mitochondria may be important targets underlying E2's beneficial effects (Simpkins et al, 2005). An intriguing question that remains to be investigated is whether actions of E2 at mitochondrial ERβ may underlie its modulatory effects on anxiety and depression behavior.

Other novel mechanisms for E2's effects involving membrane actions

ERs, specifically ERβ, are putative targets of E2 for its functional effects, but there are other possible mechanisms by which E2 can influence anxiety and depression behavior. Evidence that an ER antagonist to the amygdala does not alter anxiety or depressive behavior of naturally receptive rats suggests that E2 may act at substrates other than intracellular ERs.

One target of E2 in the amygdala may be the opioid system. E2 increases proenkephalin messenger ribonucleic acid (mRNA) levels in the ventral medial hypothalamus of female rats, thereby enhancing sexual receptivity (Lauber et al, 1990; Romano et al, 1988). Notably, E2 increases preproenkephalin mRNA in the medial amygdala (Sinchak et al, 2000). Although modulation of the opioid system may underlie some of the effects of E2 in this region, E2 may also target other membrane substrates in the amygdala.

Rapid effects of E2 suggest that E2 may act through a membrane-associated ER mechanism. Effects of E2 through intracellular ERs are expected to have a minimum latency of 10–15 min (Pfaff and McEwen, 1983), implying that rapid effects of E2 occur through membrane ERs. In dissociated hippocampal cells, E2 increases kainate-induced currents in 3 min (Gu and Moss, 1996). Rapid actions of E2 may involve membrane ERs, membrane targets, and/or activation of several signal transduction pathways, such as the mitogen-activating protein kinase, extracellular signal-regulated kinase, phosphatidylinositol-3-kinase, or adenylyl cyclase cascades (Beyer et al, 2002; Kelly and Levin, 2001; Moss and Gu, 1999; Nilsen et al, 2002; Toran-Allerand et al, 2002; Wade et al, 2001; Watters et al, 1997). Thus, it may be that E2's is acting at membrane targets and/or signal transduction pathways in the hippocampus and amygdala to alter affective behavior.

Possible interactions of E2's actions in the hippocampus and amygdala for affective behavior

Another possibility to consider is that there is a convergence of E2's actions through intracellular ERs and rapid, membrane-mediated effects in the hippocampus and amygdala. Studies investigating the circuitry of the stress response have demonstrated that activation of both the hippocampus and amygdala alter the functional response to stressors (as reviewed by Herman et al, 2003, 2004). In vivo and in vitro work demonstrates that E2 may act at membrane ERs that potentiate its effects through nuclear ERs and require activation of protein kinase A or C for its effects (Kow and Pfaff, 2004; Vasudevan et al, 2001). It may be that E2-mediated changes in anxiety and depression behavior involve its binding at membrane receptors and subsequent activation of downstream signaling molecules in the amygdala that may depend on an interaction with intracellular ERs in the hippocampus.

Conclusions

In summary, there is some evidence for E2 to have beneficial effects to improve mood among women and decrease anxiety and depression behavior of female rodents. There are some limitations to these studies related to the regimen- and/or exposure-dependent effects of E2 for anxiety and depression and/or the involvement of HPA responses. Limbic regions, such as the hippocampus and/or amygdala, that have connections to the HPA, may be targets for E2's effects on anxiety and depression. E2's actions at intracellular ERs, in particular ERβ, may be critical for these effects. Furthermore, the effects of E2 at other substrates, such as membrane ERs or rapid actions involving signal transduction molecules, may also be important.

Determining the effects, targets, and mechanisms of E2's actions on affective behavior has great clinical significance. Increasing life expectancy of women, together with a relatively constant age of menopause, has resulted in women now spending about one-third to one-half of their lives with levels of endogenous E2 that are low or at nadir (reviewed by Wise, 2003). It is likely that more women will use E2-based therapies to relieve some symptoms associated with E2 decline. However, recent clinical trials examining effects of E2 therapy among postmenopausal women have failed to support beneficial effects of E2 for cognition or mood (Brunner et al, 2005; Hays et al, 2003; Rapp et al, 2003; Rossouw et al, 2002; Shumaker et al, 2003; Smoller et al, 2003). These findings have led to investigation of antiestrogens and/or SERMs, such as raloxifene and tibolone, as therapeutic agents for menopausal symptoms, such as osteoporosis. However, the effects of these compounds on other symptoms of menopause, such as changes in affect, are not clear. Given the possibility of trophic effects of E2 and the differential distribution of ERs in the body, such that proliferative effects of E2 are likely to occur through actions involving ERα in the breast and/or mammary tissue (as reviewed by Gustafsson, 2003), it is critical to discern the receptor mechanisms that are important for E2's antianxiety and antidepressant-like effects. Further research on E2's mechanisms and brain targets may allow for more informed choices about existing therapies and/or the development of new therapies that have beneficial effects without unwanted proliferative effects.