efectele estrogenilor asupra snc

8/12/2019 Efectele Estrogenilor Asupra SNC

1/29


2/29

brain and brainstem monoaminergic systems, and spinalcord, in view of their widespread involvement with brainfunctions that subserve affective state and movement, as wellas analgesia and nociception. First, however, we will sum-marize the cellular mechanisms of estrogen action in neuraltissue.

II. Mechanisms of Estrogen Action

A. Genomic and nongenomic mechanisms

It has been customary to distinguish between steroid hor-mone actions that are delayed in onset and prolonged induration and are called genomic effects and other steroidhormone actions that are rapid in onset and short in durationand are called nongenomic (12). This is because the dis-covery of intracellular steroid hormone receptors in the early1960s created, for a number of decades, a single-mindedfocus on the long lasting effects of steroids on cell function,even though rapid actions of steroids were known since the1930s from anesthetic effects of progesterone (13).

While rapid and delayed effects of steroids are clearlydistinguishable from each other at their extremes in terms ofmechanism, there is a gray area of uncertainty for actions thathave onset times of minutes, as to whether genomic ornongenomic mechanisms apply. Genomic actions of glu-cocorticoids on lymphocytes were reported with onset la-tencies of 20 min (12). Thus, changes in neural activity re-corded in vivo after systemic administration of steroids couldhave onset latencies of minutes, leading to uncertainty as towhether the lag was due to a delay in steroid reaching thetissue or to an intrinsic delay in the mechanism of action (14).

Still further uncertainty about mechanism of action wasprovided by demonstrations of rapid, but apparently

genomic, actions of steroids affecting neuronal excitabilityand promoting or suppressing long-term potentiation (1519). On the other hand, at least several steroid actions onmembranes involve either a demonstrated coupling to Gproteins or an effect that resulted in the generation of asecond messenger (20), raising the possibility that a mem-brane steroid receptor may regulate gene expression indi-rectly via a second messenger-regulated DNA-binding pro-tein such as a member of the cAMP response element-binding protein (CREB) family (21).

Even more in contradiction to the simple stereotype, it hasbecome apparent that some important steroid actions requirethe coparticipation of certain neurotransmitters, involvinghormone actions on cells that do not appear to have the

genomic steroid receptors inside of them; rather, the effectsmay be transmitted via other steroid-sensitive neurons. Animportant example is the GnRHsystemof the hypothalamus.The activity of GnRH neurons is regulated by the ovariansteroids; yet, in vivo, these cells have not been found toconcentrate estrogen (22) or express the classical ER, ER,or PR, in any species studied (2325). However, distinctpopulationsof adjacent cells, immunoreactive to neurotensin(23), galanin (26), -aminobutyric acid (GABA), or glutamate(24), have been shown to express ER and/or PR protein.Furthermore, it is known that GnRH release is regulated byhypothalamic amino acid transmitter systems (27, 28). Col-

lectively, these findings point to an indirect, transsynapticregulation of GnRH neurons by estrogen and progesteronein vivo, although it should be pointed out that GT17 cells,immortalized mouse GnRH neurons, do express seeminglyfunctional ER(29). A similar, possibly transsynaptic regu-lation by estrogen of synapse formation occurs in hippocam-pus, as will be discussed below in Section III.F. The finding

of estrogen induction of synapses in hippocampus and alsohypothalamus has challenged the notion of a morphologi-cally stable adult brain by showing that steroids alter struc-tures of the adult brain, including remodeling of synapses,changes in dendritic structure, and neurogenesis, as will bereviewed below in Section III.F.

B. Steroid hormone actions on gene expression

The identification and mapping of cells expressing thegenomic steroid receptors by binding, immunocytochemis-try, andin situ hybridization have provided the target sitesfor investigation of hormonal control of gene expression.Nevertheless, it is only a starting point, because the quali-tative nature of hormonal regulation of gene expression can-not be predicted with any certainty from one brain region toanother. For example, vasopressin is an important neuropep-tide system that is subject to gonadal hormone regulation,and vasopressin mRNA levels are induced by androgens inthe bed nucleus of the stria terminalis (30) and are sup-pressed by glucocorticoids in the paraventricular nuclei (31).CRH gene expression is suppressed by glucocorticoids inparaventricular nuclei, but induced in placenta (32, 33); inaddition, there are brain areas that contain both glucocorti-coid receptors and CRH in which there is no apparent glu-cocorticoid regulation of CRH gene expression (32).

Sexual differentiation involves more than sex differences

in structure and wiring of the brain. There are also sex dif-ferences in gene expression, in which the male and femalebrain respond differently to the same hormone (34). Forexample, estradiol has a double roleas an ovarian steroidin females and as the product of the aromatization of tes-tosterone in males. Therefore, it is not surprising that estra-diol can produce somewhat different effects on the male andfemale brain,e.g., inducing prodynorphin mRNA in the an-terioventral periventricular nucleus of female, but not male,rats (35). Moreover, some of these sex differences are knownto be reversed by the hormonal conditions during early lifethat reverse the sex differences in sexual behavior. For ex-ample, in anterioventral periventricular nucleus, male ratsexpress more preproenkephalin mRNA than females,

whereas the reverse is true forprodynorphin mRNA; femalesthat are androgen sterilized at birth show male patterns ofneuropeptide gene expression (35). Alternatively, somegenes appear to be similarly regulated by estradiol in bothsexes, such as the hypothalamic oxytocin receptor (36), theserotonin 2A receptor (37), and the -form of the estrogenreceptor (ER) in some brain regions (38, 39).

C. Subtypes of ERs

The discovery and cloning of the -isoform of the estrogenreceptor (ER) (4042) radically changed our view of estro-

280 MCEWEN AND ALVES Vol. 20, No. 3


3/29

gen action and provided, among other things, a basis forunderstanding how the knockout of ER (ERKO) (43, 44)could result in a viable organism and a continued respon-siveness of at least some tissues to estrogens. Before the fullrecognition of ER, a mapping study was carried out in theERKO brain using [125I]estrogen, and estrogen induction ofprogestin receptor (PR) was also mapped by in situhybrid-

ization of the mRNA (45). A low level of residual estrogenbinding was found in the medial preoptic nucleus, arcuatenucleus, bed nucleus of the stria terminalis, and amygdala,and a significant estrogen-induced up-regulation of PRmRNA was found in the medial preoptic nucleus (45). Sub-sequent attempts to map ERhave confirmed that residualestrogen binding and action in ERKO mice might be due toER, and these studies have also provided some novel sitesfor ER as well as some overlap with ER (46 48). As for PRregulation and the functionality of residual ER (particularlyER) in ERKO mice, a recent immunocytochemical study hasshown estrogen induction of PR immunoreactivity in severalhypothalamic nuclei and in amygdala (49).

A recent study by Krege and colleagues (50) has reportedthe generation of mice lacking functional ER. Interestingly,ER knockout females and males appear to develop nor-mally, exhibit normal sexual behavior, and are reproduc-tively competent, although females do exhibit reduced fer-tility (i.e., fewer and smaller litters), apparently due todecreased ovarian efficiency. This is in sharp contrast toERKO mice, which are sterile and do not display normalsexual behavior (5154). Thus, it appears that ER, more sothan ER, is necessary for the estrogen-mediated regulationof reproductive physiology, including the behavioral com-ponents.

Distributions of ER and ER in the body differ quitemarkedly, with moderate to high expression of ER in pi-

tuitary, kidney, epididymis, and adrenal, moderate to highexpression of ER in prostate, lung, and bladder, and over-lapping high expression in brain, ovary, testis, and uterus(48, 55, 56). It is now known that at least several isoforms ofER are expressed (5760). The best characterized of thesevariants has been termed ER2,vs. the originally identifiedER1, and this isoform appears to have a lower affinity forestrogens (61), presumably due to an 18-amino acid insertionin the ligand-binding domain (58). The ER2 variant wasfound at levels equal to ER1 in ovary, prostate, pituitary,and muscle; in brain, expression was found in cortex, hy-pothalamus, and hippocampus, although at lower levelsthan ER1 (57). Despite diminishedligand binding, ER2canbind at the ERE, and it apparently acts as a negative regulator

of estrogen action, as it was found to suppress ER and1-mediated transcriptional activation in a dose-dependentmanner (58). Moreover, both ER1 and ERare reported tointeract with the estrogen-dependent coactivator, SRC-1,whereas ER2 does not do so and requires 100- to 1000-foldhigher 17-estradiol concentrations to activate a promotorcontaining the estrogen response element (ERE) (62). HumanER isoforms 25, with alterations in the ligand-bindingdomain, have also been identified, and they can form homo-and heterodimers with ER1 and ER (42, 59). Another vari-ant, termed ERcx, has a truncated C terminus, but has 26additional amino acids due to alternative splicing (60). This

form appears to specifically inhibit ER-induced transcrip-tion; however, ERcx has not yet been found in brain (60).The genomic effects of estradiol via intracellular ER are de-picted in the top panel of Fig. 1.

In brain, the distribution of ERis fairly well established,but there is less certainty and more controversy surroundingthe localization of functional ER. The original autoradio-

graphic maps of [3

H]estradiol uptake and retention in brain(63, 64) reflect binding to all forms of the ER, particularly theERand the ER1 isoform, which have similar affinities for17-estradiol (55). In situ hybridization data suggest wide-spread distribution of ERmRNA throughout much of the

FIG. 1. Schematic diagram of intracellular estrogen action via ERandER, as well as possible cell surface effects of putative membraneERs that produce neuroprotection (top) or affect intracellular signal-ing (bottom) via the cAMP and MAP kinase pathways. Top panel,Estradiol exerts its effects intracellularly via two principal receptortypes, ERand ER, and these are characterized by a distinct spec-ificity for 17-estradiol over 17-estradiol. Estrogens also exert neu-roprotective effects in part via a mechanism in which 17-estradiolhas equal or greater potency compared with 17-estradiol. Bottompanel, Estradiol acts either via cell surface receptors or an intracel-lular ER to activate two different second messenger pathways, oneinvolving the MAP kinase cascade and the other involving cAMP.Both pathways result in activation of gene transcription via at leastthree possible response elements: CRE, SRE, and AP-1. Note that inthe case of intracellular second messengers there is some uncertaintyconcerning the involvement of ERand ERin the signaling processvs.the role of other, as yet uncharacterized, receptors (see text). AC,Adenylate cyclase; CREB-P, phosphorylated form of CREB; ras, rasoncogene; MAPK, mitogen-activated protein kinase; MAPKK, mito-gen-activated protein kinase kinase; fos-jun, fos-jun heterodimer.

June, 1999 ESTROGEN ACTIONS IN THE CNS 281


4/29

brain (46, 47) while results from immunocytochemical stud-ies (see below) tend to indicate a more restricted localizationof detectable protein, raising questions about the specificityof thein situhybridization procedure, on the one hand, andthe efficacy of the antibodies used for immunocytochemistry,on the other.

The initial commercially available polyclonal antiserum to

ER (no. 310, Affinity BioReagents, Inc.), raised against ashort fragment of the C terminus of the receptor, has pro-duced a consistent pattern of strong cell nuclear label in themedial amygdala, paraventricular nucleus (PVN), and pre-optic area, and striking cytoplasmic/fiber label in cells of thelateral septum (65 67). Detection of cell nuclear ER labelingin the supraoptic nucleus (SON) may be dependent upon theuse of acrolein in the fixation procedure (65, 67). If the Cterminus antiserum is preabsorbed with the synthetic pep-tide in a 1:1 (wt/vol) ratio, labeling is completely absent inthese brain regions (6567). However, in other brain regionssuch as hippocampus and cortex, immunolabeling is not asconsistent, and such labeling is not always obliterated bypreabsorption of the antiserum (S. E. Alves and B. S.McEwen, unpublished results). This is particularly true forthe N terminus antiserum (no. 311, Affinity BioReagents,Inc., Golden, CO), which produces both cytoplasmic andnuclear labeling. Factors such as fixative (acrolein vs. para-formaldehyde), gonadal state of the animal (gonadectomizedvs.intact), or stage of the estrous cycle, appear to affect ERprotein detection in the brain. Thus, caution must be takenin the interpretation of ER immunoreactivity in brain re-gions other than the hypothalamus and amygdala usingthese antibodies.

Moreover, reports of the neurochemical phenotypes ofER-containing cells in some brain regions, particularly thePVN and SON, have been somewhat conflicting. While all

studies thus far have identified subpopulations of oxytocinneurons in the PVN and/or SON to contain ER (46, 66 68),differences have been reported in the specific cell popula-tions, as well as for other peptide systems. For example, twostudies measuring either ERmRNA (8) or protein (7) havereported colocalization with vasopressin, particularly in theSON; another study has reported that colocalization betweenER mRNA and vasopressin was only seen in scattered cellsof the parvocellular PVN (46). This later study also reportedthat more than half of the CRH neurons of the caudal par-vicellular PVN contain ERmRNA. In contrast, Alves andco-workers (67) observed only few scattered parvicellularCRH-positive neurons to contain ERprotein, using the Cterminus antiserum.

Several viable explanations for discrepancies betweenmessage and protein data exist. Considering the existence ofreceptor variants, one possible explanation is that a form ofER, not recognized by the short C terminus antiserum, isexpressed in these cells, which would result in an underes-timation of ER-expressing cells when using this antiserum.Alternatively, perhaps not all ERmRNA is translated intofunctional protein or it is translated in a transient manner thatdiffers between cell types and developmental stages. On theother hand, it is also possible that the discrepancies in theresults of different laboratories are due to false-positives withthe existing probes to ER.

However, if the reported CNS distribution of ERmRNAis found to reflect the expression of some functional ERprotein in those brain regions, this would certainly help toexplain numerous estrogen actions in brain regions with littleor no ER. This includes areas such as the olfactory bulbs,cerebellum, and cerebral cortex, in which ER mRNA hasbeen abundantly detected (48). It is hoped that future inves-

tigations into the seemingly complex detection and expres-sion of ERwill provide a clearer picture of the distributionand phenotype of cells that contain functional ER.

As noted above, ER and ER1 are similar not only inaffinity for a number of estrogens and estrogen antagonists(55), but also in their ability to regulate genes in which theERE is the primary site of interaction (69) (see top panel of Fig.1). The major differences between ER and ER1 concerntheir ability to regulate transcription via the AP-1 responseelement. For interactions of ER with AP-1, 17-estradiol, aswell as a number of antiestrogens, activated transcription;however, for ER1 interacting with AP-1, 17-estradiolfailed to activate transcription but antiestrogens activatedtranscription (69). As mentioned above, ERand ER1 canform heterodimers when expressed in the same cells, thusgiving rise to additional possible variants of gene regulation(42). Thus far, the endogenous colocalization of ER andERhas been reported recently in the hypothalamicpreoptic area,bed nucleus of the stria terminalis, and medial amygdaloidnucleus (68).

The agonist effects of estrogen antagonists bring to mindearlier studies in which estrogen antagonists produced es-trogen-like effects on some neurochemical endpoints andantagonistic effects on others. The antagonistic effects forCI-628, a tamoxifen-like estrogen antagonist, were seen interms of PR induction and lordosis behavior (70, 71) (see Fig.2), whereas the agonist-like effects of CI-628 were seen for

choline acetyltransferase regulation and monoamine oxidaseA regulation, but not for the regulation of glucose-6-phos-phate dehydrogenase in pituitary and uterus (72) (see Fig. 3).The molecular mechanisms underlying the differences inthese antiestrogen effects remain to be explored, and theymay reflect the operation in some of the cases of a responseelement other than the ERE and perhaps even the operationof heterodimers of ER and , but the diverse effects ofCI-628 indicate that the nonsteroidal antiestrogens do nothave uniform agonist-like or antagonist-like effects in thebrain. This is an important consideration for the therapeuticuse of estrogen antagonists of this type.

D. Steroid hormone actions on putative receptors

on membranes

Membrane ERs have been reported on pituitary, uterine,ovarian granulosa cell, and liver cell membranes, but theyhave been characterized only partially and have not yet beenshown to be linked to signal transduction mechanisms (7380). For membrane fractions from pituitaryand ovarian gran-ulosa cells, the specificity of the binding sites shows equalpotency of 17- and 17-estradiol, estriol, and estrone; forliver and uterine cells, there is a preference for 17- over17-estradiol (73, 75, 76, 79, 80). However, for GH3/B6 pi-tuitary cells, a 17-estradiol binding site was identified using



5/29

an estrogen-BSA conjugate; but, in contrast to the other re-ports suggesting a novel membrane ER (cited above), mono-clonal antibodies to the intracellular ER, H226, and H222,as well as the polyclonal antiserum, ER21, each recognizinga unique epitope on ER, labeled sites on thesecells in or nearthe cell surface (77, 81). A more recent report used transienttransfection of both ERand ERcDNA into Chinese ham-

ster ovarian cells and demonstrated both types of ER ex-pressed in both cell membrane and nuclear fractions; thebinding affinities for estradiol were similar in both mem-brane nuclear fractions, and, in membranes, estradiol acti-vated Gq and Gs proteins in the membrane and rapidlysimulated, respectively, inositol phosphate production andadenylate cyclase activity (82).

These limited findings can be viewed in relation to dataabout other membrane steroid receptors. The best under-stood membrane receptor for a steroid is the GABA-A re-ceptor. Anesthetic effects of progesterone derivatives (13)led, after many years, to the recognition of a unique mem-

brane recognition site on many subunit combinations of theGABA-A-benzodiazepine receptor system (83, 84). A-ring-reduced metabolites of progesterone and deoxycorticoste-rone are among the most active steroids affecting theGABA-A receptor system, and such metabolites are pro-duced in the body, including the brain, from the parentsteroid. The effects of these steroid metabolites include notonly anesthetic effects but also antiepileptic, sedative-hyp-notic, and anxiolytic actions (83). The efficacy of these me-tabolites in normal physiology is suggested by experimentson progesterone facilitation of lordosis in the hamster, inwhich it was found that local application of GABA-A-activederivatives of progesterone to the midbrain ventral tegmen-tal area (VTA) of the estrogen-primed hamster was able to

facilitate lordosis (8587). Moreover, while GABA-A recep-tor-active pregnane steroids, applied to the ventral tegmentalarea, facilitate lordosis behavior, inhibitors of 5-reductase,the first step in A-ring reduction, prevented systemicallyapplied progesterone from facilitating lordosis (20).

Other membrane receptors for steroid hormones are not sowell characterized (20). One exception is the membrane-binding site for 1,25-dihydroxyvitamin D3 in basal-lateralmembranes of chick intestinal epithelium, which demon-strates pharmacological specificity for a receptor modulatingnongenomic transport of calcium (88). Another example is amembrane steroid receptor site for corticosterone in the

FIG. 2. Effect of the estrogen antagonist, CI-628 (nitromiphene ci-trate; -[4-pyrrolidin-oethoxyl] phenyl-4-methoxy--nitrostilbene),on the induction of cytosol PRs in the hypothalamic/preopticregion ofovariectomized female rats by estradiol (E2) or estradiol benzoate(EB). Treatment protocols are referred to as follows: 1 DAY: 5 g E2and 2 mg CI-628 at 0 h, killed at 24 h; 2 DAYS: 2 g EB at 0 h andCI-628 at 0 h and 24 h, killed at 48 h; 3 DAYS: 15 g EB and 2 mgCI-628 at 0, 24, 48 h, sacrificed at 72 h. Cytosolic receptors wereassayed using 3H-labeled R 5020, a synthetic progestin. Black barindicates increase above control levels. It can be seen that estrogentreatment increased PR binding and that CI-628 at least partially

antagonized the effects of estrogen treatment under all conditions. Inthe lower right panel, a Scatchard analysis of PR binding of ratstreated with 2 g EB at 0 h vs. 2 g EB at 0 h plus two injections of2 mg CI-628 at 0 h and 24 h, killed at 48 h (2 DAYS protocol).[Reprinted with permission from E. Roy et al.: Endocrinology 104:13331336, 1979 (70). The Endocrine Society.]

FIG. 3. Effect of the estrogen antagonist, CI-628, on estradiol ben-zoate (EB)-dependent enzyme changes in uterus, pituitary, and brain.Ovariectomized rats were injected for 5 days with sesame oil vehicle,CI-628 (18 mg/kg), or EB (140 g/kg) either alone or in combination.Enzyme activities are reported as mean SEM for choline acetyl-transferase (CAT) in preoptic area, type A monoamine oxidase (MAO)in amygdala, glucose-6-phosphate dehydrogenase (G6PDH) in pitu-itary and uterus. Details of the assay may be found in the originalpublication (72). Differences among groups were tested by Newman-Keuls procedures: *, Different from OVX, P 0.05; **, P 0.01;different from EB CI, P 0.05. For uterus, all groups were sig-nificantly different from one another. [Reproduced with permissionfrom V. Luine and B. S. McEwen: Endocrinology100:903910, 1977(72). The Endocrine Society.] It can be seen that CI-628 exertedagonist-like effects on CAT and MAO activity, whereas it antagonizedestrogen induction of G6PDH activity in pituitary and uterus.



6/29


7/29

trogen antagonist ICI 182780 and which, therefore, appearsto involve an intracellular ER (115117). In a follow-up study,progesterone activation of this pathway was shown to occurby an association of the PR with an N-terminal region of ERand not with c-Src directly (116).

In neuroblastoma SK-N-SH cells and in cortical explants,17-estradiol was reported to activate the MAP kinase and

phosphorylate and activate two of them, ERK-1 and ERK-2(118). In contrast to MCF-7 cells, the response in SK-N-SHneuroblastoma cells and cortical explants was not blocked byICI 182780 or by tamoxifen (118), indicating that a classicalER may not be involved. Furthersupport for this in SK-N-SHcells came from the finding that 17-estradiol conjugated toBSA activates a reporter gene driven by the mouse c-fosprotooncogene, which responds to MAP kinase activation,but does not activate transcription mediated by a promotorcontaining the ER response element, ERE (118). A possiblesequence of signaling events is summarized in the bottompanel of Fig. 1. Estrogen-dependent activation of MAP ki-nases, ERK-1 and ERK-2, has also been studied in embryoniccerebral cortical explants grown in culture (119, 120). Theactivation of ERK was blocked by the MEK-1 inhibitor,PD98059, but not by ER antagonist ICI 182780, again sug-gesting that a conventional ER may not be involved (119).

3. Calcium homeostasis. Calcium ions also constitute an im-portant player in second messenger pathways, and the effectof estrogen on calcium channels and calcium release fromintracellular stores has also emerged as a possible cellularmechanism of estrogen action that is relevant to excitabilityand neuronal vulnerability to damage. There appear to be atleast three distinct pathways for estrogen action on calciumhomeostasis that have been documented in different cellsystems, each involving a different type of membrane ER,

and at least one pathway for affecting calcium homeostasisvia a genomic mechanism.In the first of the nongenomic pathways, 17-estradiol

activates a G-protein-coupled receptor in rat neostriatal neu-rons and, within seconds, suppresses currents mediated byL-type calcium channels (121). 17-Estradiol is considerablyless potent than 17-estradiol, as are other steroids, and theestrogen antagonist, tamoxifen, mimicked estrogen actionand did not block them (121), thus further supporting aunique ER on the cell surface. Interestingly, these estrogeneffects were sex specific, occurring more robustly in neuronsfrom female rats (121). A similar effect of 17-estradiol toinhibit currents mediated by L-type calcium channels wasreported in aortic smooth muscle, providing a basis for the

well known effects of estradiol to regulate vascular tone(122).

In a second nongenomic pathway, as shown for liver anduterine endometrial cell membranes, estradiol binds ste-reospecifically (17 17), and these sites may be respon-sible for estrogenic stimulation of calcium influx into thesecells (7476, 78, 79). In contrast, via a third pathway, inchicken ovarian granulosa cells, 17- and 17-estradiol fa-cilitated release of intracellular calcium stores equipotentlyin a concentration range of 1010 to 106 m(80). Estriol andestrone were also effective in the same range, but progestinsand androgens were ineffective; moreover, estrogen actions

were not blocked by tamoxifen or by RNA and protein syn-thesis inhibitors (80), thus defining an estrogen membranesite of broader specificity than that seen in liver or endome-trial cells or striatal neurons.

There are, however, reported estrogen effects on calciumcurrents that are more consistent with an intracellular,genomic action of estradiol. In a study on GH3 pituitary cells,

17-estradiol treatment increased low voltage-activated cal-cium currents over 24 h by a mechanism requiring proteinsynthesis (123). Moreover, in a hippocampal slice study, invivotreatment with estradiol increased in vitroboth the sus-tained and transient calcium currents, whilein vivoproges-terone acutely amplified the estrogen effects over 4 h; incontrast, potassium currents were not altered by these sametreatments (124). These results appear to be more consistentwith the genomic actions of estradiol that are related tosynaptogenesis and will be discussed below.

G. Neuroprotective effects of estrogens

Estrogens exert protective effects on neuronal cells in cul-ture that may be mediated, at least in part, by their ability toalter free radical production and/or free radical action oncells. However, as was also thecasefor the second messengersystems, the evidence for involvement of intracellular ERs vs.novel membrane receptors is controversial, although tendingto point to a nontraditional receptor mechanism. The dis-tinction between neuroprotective effects of estrogens medi-ated by intracellular receptors and those mediated by puta-tive receptors located in other parts of the cell aresummarized in the top panel of Fig. 1 and is based on thedifferent estrogen structure-activity profile, as will be de-scribed below.

The first neuroprotective actions of estradiol were de-

scribed in relation to the effects of serum deprivation onneuronal survival in cell culture (125129). In one of thesestudies (127), picomolar levels of 17-estradiol enhancedfetal rat hypothalamic neuronal survival in a serum-freemedium in the presence or absence of glial cells by a mech-anism that was blocked by tamoxifen and that, presumably,involves intracellular ERs. A similar example will be givenat the end of this section regarding estrogenic neuroprotec-tion from glutamate toxicity (130).

In serum-free medium, embryonic cortical neurons wereshown to survive better in the presence of nanomolar con-centrations of 17-estradiol; in fact, 17-estradiol facilitatedneurite outgrowth by a process that was blocked by AP5, anN-methyl-d-aspartate (NMDA) receptor antagonist, but not

by ICI 182,780, an ER antagonist, suggesting a possible non-genomic mechanism (131).

In another series of studies (128, 129), neuroblastoma SK-N-SH cells were protected by concentrations of both 17-and17-estradiol in serum-free media. In the first of these studies(128), 17-estradiol concentrations of 2 m enhanced totallive cell number for up to 48 h without increasing thymidineincorporation, indicating an effect on cell survival and notcell division. In the second study in this series (129), 17- and17-estradiol in the range of 0.22 nm protected SK-N-SHcells in culture, and a 10-fold molar excess of tamoxifenantagonized only one-third of the neuroprotective effect.



8/29

Taken together, the absence of stereoselectivity of 17- vs.17-estradiol and the weak antagonism by tamoxifen argueagainst involvement of the classical intracellular ERs in neu-roprotection of SK-N-SH cells.

The other situation in which neuroprotection by estrogenicsteroids has been described is in relation to oxidative dam-age, and, once again, the weight of evidence is against a role

for the classical intracellular ERs. Before discussing estrogeneffects on cell survival, we note a chemical study carried outin the absence of living cells or cellular extracts, in which theaddition of 200 nm 17-, 17-estradiol, or estriol each re-duced the generation of free radicals, whereas other steroidswere ineffective (132). This suggests that features of the es-trogen A ring, involving the 3 hydroxyl group, have theability to interfere with free radical production in the absenceof proteins or other cellular materials.

There have been a number of studies investigating a neu-roprotective role of estrogen from free radical damagein cellsin culture. In the first of these, exposing cloned mouse HT22hippocampal cells to amyloid-, hydrogen peroxide or glu-tamate resulted in oxidative damage and cell loss that wasreduced by preincubating cells for 20 h with 10 m 17-estradiol or 200 umvitamin E,but not by10mprogesterone,aldosterone, corticosterone, or cholesterol; 17-estradiol at aconcentration of 0.1 m was not effective in these studies(133).A follow-upstructure-activitystudy by the same grouprevealed a broad spectrum of estrogen specificity, with 17-and 17-estradiol as well as estriol and estrone all beingeffective and pointing to the C3 -hydroxyl group on thesteroid A ring as being important as well as implying thatintracellular ERs are not involved (134).

Dissociated embryonic hippocampal neurons were alsosensitive to estrogen-mediated protection against amyloid-toxicity, glucose deprivation, glutamate treatment, or FeSO4

toxicity; in this study, the effective steroid concentrationrange was 100 nm to 10 m, and estriol and progesteronewere also effective, whereas corticosterone enhanced neu-rotoxicity across this concentration range (135).

In contrast to these studies using high, supraphysiologicalconcentrations of estrogens, another recent investigationshowed that as little as 0.22 nmestradiol, either 17or 17,protected SK-N-SH cells against -amyloid toxicity (136).This raises the question of how different investigators havefound such different concentrations of estrogens to be effec-tive (136). These discrepancies in effective concentrationranges of estrogens aredifficult to understand.However, onepossible clue points to the concentration of natural reducingagents in the culture medium, i.e., a recent report showed that

the addition of glutathione to HT22 cells in culture, whichlack functional ER, reduced the dose range of estrogen neu-roprotection by 400-fold (137).

In contradistinction to all of the above mentioned neuro-protection studies, a report on glutamate toxicity in primarycortical neurons in culture suggests involvement of an in-tracellular ER; 24-h pretreatment with 1550 nm 17-estra-diol reduced glutamate-induced toxicity, measured by lac-tate dehydrogenase release, an effect that was blocked by theestrogen antagonist, tamoxifen. Thus, this finding suggeststhat there are situations in which estrogen neuroprotectionmay involve the activation of intracellular ER (130).

In a related study, the toxicity of gp120 for hippocampalcells in culture is inhibited over 72 h by 17-estradiol atconcentrations of 1 nmor above (138). Since gp120 exerts itseffects via excitatory amino acids and calcium ions, culmi-nating in free radical-induced damage (138), it is noteworthythat estradiol reduces the free radical accumulation inducedby gp120 (R. Sapolsky, personal communication). However,

there are no experiments with this system to indicatewhether a conventional ER mechanism or a novel type ofreceptor is involved.

Finally, there is another mechanism potentially involvedin neuroprotective estrogen effects, namely, the regulation ofthe Bcl-2 family of genes (139). Some members of this family,such as Bcl-2 and Bcl-XL, suppress programmed cell death,whereas others such as Bax.Bad and Bid act as positive reg-ulators of apoptosis (see Ref. 139). In arcuate nucleus neuronsof female rats, estrogen treatment up-regulated expression ofBcl-2 immunoreactivity (139). Estrogen up-regulation ofBcl-XL was also reported both in vivo and in vitro in hip-pocampal and cortical cells (140). Thus, estradiol may alsoincrease gene expression that inhibits programmed celldeath, although the mechanism by which estrogen regulatesBcl expression is not known at this time; intracellular ERand/or ER may very well be involved in such genomicregulation, as both receptors are found in these brain regions.

H. Summary

Estrogen actions on brain cells occur through at least twotypes of intracellular receptors as well as other mechanismsfor which receptor sites are not yet clearly identified. Indeed,for a number of processes, there are conflicting reports, basedupon structure-activity studies with different estrogens andthe actions of estrogen antagonists, as to whether intracel-

lular receptors are involved. For estrogen actions on someaspects of calcium homeostasis, activation of certain secondmessenger systems, and some features of neuroprotection, anovel receptor mechanism may exist in which stereospeci-ficity for 17- over 17-estradiol is replaced by a broaderspecificity for the 3-hydroxyl group on the A ring.

III. Areas of the Brain Affected Outside of

the Hypothalamus

A. Studies of hypothalamic and extrahypothalamic actions

of estrogens

Until recently, the hypothalamus has been the focus of

much of the attention regarding neural effects of gonadalhormones. Ovarian hormone actions on neurons of the ven-tromedial hypothalamus, which are important for the reg-ulation of sexual behavior in female rats, include regulationof neuropeptide gene expression (141) and second messengersystems (142) and induction of oxytocin receptors, PR, andthe regulation of cyclic synaptogenesis (36, 143145). Thereare also developmentally programmed sex differences in-volving both neuronal wiring as well as programming ofresponses to hormonal activation of gene expression (144,146). All of these actions occur in neurons that express highlevels of ERand PR, in contrast to the hippocampus, mid-



9/29

brain raphe, basal forebrain, brainstem, and spinal cord, inwhich ERand PR, if detected at all, appear to be relativelyscarce and are found in many cases in interneurons or scat-tered principal neurons. As noted in Section II.C, the extra-hypothalamic distribution of ER protein is still unclear, butthe presence of ER mRNA in brain regions such as thecerebellum, hippocampus, cerebral cortex, and olfactory

bulbs (47, 48) suggests that ERmust be regarded as a po-tential mediator of estrogen action in those brain areas.

This is a very important consideration, since gonadal hor-mones, and, in particular, estrogens,have many effects on thenervous system that extend beyond their very importantactions in hormonal regulation of reproductive function.Moreover, many of these estrogen effects differ qualitativelyor quantitatively between the sexes, suggesting that theymay be influenced by the process of sexual differentiationduring early pre- or postnatal development and/or by dif-ferent levels of circulating sex hormones. In consideringthese nonreproductive actions of estrogens, we must alsoconsider other brain regions outside of the hypothalamus.However, this does not mean that the hypothalamus is con-cerned only with reproduction, nor does it mean that theextrahypothalamic brain regions are not contributing to re-productive functions. Indeed, the hypothalamus is con-cerned with many aspects of autonomic and neuroendocrinecontrol, and extrahypothalamic systems such as serotoninand the catecholamines play important supporting roles inreproductive endocrinology. Nevertheless, in addition to re-production there is much more to brain function that isinfluenced by estrogens and by sex differences, and this is theprimary focus of the following discussion.

Sex differences in brain function also include gender dif-ferences in the incidence of psychopathologies such as de-pressive illness, which is more common in women, and sub-

stance abuse and antisocial behaviors, which are morecommon in men, as well as pain sensitivity (11). Sex differ-ences and estrogen effects upon the serotonergic, cholinergic,dopaminergic, and noradrenergic systems all may contributeto many aspects of brain function that are affected by ovarianhormones, including affective state (7), movement disorders(6), and cognitive function (2, 147). Furthermore, the discov-ery of estrogen-induced synapse formation in hippocampusand hypothalamus, which are described below, are relevantto postmenopausal changes in brain function, including de-cline of short-term verbal memory (147), as well as to theoccurrence of dementia, which becomes more prevalent inwomen after the menopause as well as in men as they age(148). Recent epidemiological studies have suggested a pos-

sible protective role for postmenopausal estrogen therapyagainst Alzheimers disease (149, 150). Furthermore, estro-gen treatment trials have shown some benefit to dementedwoman as far as global cognitive function and mood (151,152) as well as to normal women (3, 4, 153) as far as verbalmemory. A recent study also suggests a positive role forestrogen therapy on cognitive function in multiple sclerosis(154). Moreover, estrogen and progestin-induced regulationof synapse formation and excitability may play a role incatamenial epilepsy, which varies in frequency during themenstrual cycle (155). The presence or absence of hormonesalso contributes to aging of the brain,e.g., loss of hippocam-

pal neurons as a result of elevated glucocorticoid activity(156, 157); and consequences of estrogen loss in females mayinclude loss of synaptic connections in hippocampus (158) ordecline in basal forebrain cholinergic function in the absenceof circulating estrogens (159). An additional aspect of estro-gen action is the regulation of neurogenesis in the dentategyrus, which continues to produce new neurons in adult life.

A recent report indicates that female rats have a higher rateof neurogenesis than males and that neurogenesis variesduring the estrous cycle with a peak on the day of proestrus(160).

Sex differences in brain structures and mechanisms areprogrammed early in life by gonadal hormones and are per-manent for the life of the individual. Sex differences occur inbrain regions other than the hypothalamus, such as hip-pocampus, and they appear to be involved in aspects ofcognitive function and other processes that go beyond thereproductive process itself. In this review, we refer to sexdifferences and the process of sexual differentiation butnot to sexual dimorphism, which is a term that refers tononoverlapping differences in phenotype between the sexes.This is because true sexual dimorphisms are very rare, andthe more common pattern of sex differences involves over-lapping, but significantly different, distributions of pheno-typic traits.

Understanding the cellular and molecular basis of sexdifferences and of sex differences in the actions of gonadalhormones is vitally important for assessing how pharma-ceutical agents differentially affect the brains of males andfemales (161), as well as in understanding other male-femaledifferences relevant to health and disease, such as the higherincidence of depression in women and of substance abuse inmales (7). There are also sex differences in the severity ofbrain damage resulting from transient ischemia (162) and sex

differences in the response of the brain to lesions (163) andto severe, chronic stress (164, 165).The diversity of these effects implies that regions of the

brain are involved outside of the hypothalamus. Indeed, aswe have noted in Section II.C above, mapping of intracellularreceptors, which modulate genomic actions, has revealed thepresence of ER and/or PR expression in regions such as theolfactory lobe, hippocampus, cortex, locus ceruleus, mid-brain raphe nuclei, and midbrain central gray and cerebel-lum. Although the density of ERis often lower and morediffuse in many of these brain areas compared with hypo-thalamus and amygdala, the existence of prominent estrogenand progestin effects in many of these brain areas requires acareful examination of the role of the cells that express in-

tracellular receptors in these brain regions. We have noted inSection IIabove that the localization and expression of ERis an important consideration, along with a consideration ofpossible alternative mechanisms of steroid action. We nowconsider a number of the extrahypothalamic brain regionsthat are sensitive to estrogens and progestins.

B. Estrogens and the cholinergic system

The basal forebrain contains cholinergic neurons thatproject to cerebral cortex and hippocampus, where they playan important role in cognitive function. Studies of estrogen



10/29

effects on the expression of cholinergic enzymes were amongthe first that pointed to nonreproductive actions of gonadalsteroids (166). Experiments with ovariectomy and estrogenreplacement therapy revealed an induction of choline acetyl-transferase (ChAT), the rate-limiting enzyme for acetylcho-line formation, within 6 24h in basal forebrain of female rats.In addition, estrogen treatment increased ChAT activity in

projection areas of the basal forebrain 10 days after hormoneinjection, suggesting that estrogen-induced ChAT was trans-ported from cell bodies to nerve endings in the cerebralcortex and hippocampus (166). 17-Estradiol treatment alsoinduced acetylcholinesterase, as well as ChAT activity, im-plying that a general trophic effect on the cholinergic neuronsmight occur (166). The estrogen induction of ChAT wasmimicked by the estrogen antagonist, CI-628 (see Fig. 3),suggesting that a different type of interaction with the ge-nome is involved than the traditional one involving an ERoperating via the ERE (see Section II.C) (72).

A recent investigation of long-term (528 wk) ovariectomyand long-term estrogen replacement in rats revealed a de-cline in high-affinity choline uptake and in ChAT activity infrontal cortex and hippocampus that was at least partiallyprevented by estrogen treatment (167). Estrogen treatmentalso increased the acetylcholine released by potassium de-polarization (168). Estrous cycle variations in ChAT mRNAlevels were also reported in the basal forebrain cholinergicsystem (168). Along with these effects, long-term ovariec-tomy caused a decline in learned performance of activeavoidance behavior that was prevented by estrogen replace-ment therapy (167).

One possible candidate as a regulator of the cholinergicsystem of the basal forebrain is nerve growth factor (NGF),which is produced by the hippocampus and transportedretrogradely to basal forebrain neurons to produce trophic

effects. Although the effects of estrogen treatment on NGFlevels in hippocampus remain to be investigated, ERhavebeen reported to colocalize with low-affinity NGF receptorsin cholinergic neurons of the basal forebrain of the newbornrat (169). Moreover, estrogen replacement in both young andaged female rats increases both trkA (NGF receptor) mRNAand ChAT mRNA expression in basal forebrain (170).

The basal forebrain of male rats failed to show the sameresponse to estrogen treatment as females, and postnatalestrogen treatment of females or blockade of aromatizationin males failed to change this sex difference (166, 171), in-dicating that the sexual differentiation of the cholinergicsystem either occurs earlier in development or does not in-volve the aromatization of testoserone to estradiol. Addi-

tional studies revealed that the basal forebrain cholinergicsystem differs between male and female rats, with femaleshaving smaller and more densely packed cholinergic neu-rons compared with untreated males (172). Moreover, ap-plication of T3 to newborn male and female rats, creatingtransient hyperthyroidism during the first week of postnatallife, revealed further indications of sexual differentiation ofthe basal forebrain cholinergic system in which male ratsresponded to the treatment while females did not (172). Forexample, treatment with T3 increased cholinergic cell densityand induced increased ChAT activity and muscarinic recep-tor binding in the septum/diagonal band region of males.

Females did not respond to T3 in most respects, except inmedial septum where they showed the opposite effect tomales, namely, an increased cholinergic cell body area (172).This finding suggests that there is an interaction between theprenatal effects of testosterone in the development of thecholinergic system of the basal forebrain (173) and the post-natal effects of T3(172).

On the other hand, in another study of sex differences inthe cholinergic system, femalerats showed larger effects thanmales to the cholinergic lesions produced in hippocampus bythe specific cholinergic neurotoxin, AF64A,and females wereparticularly sensitive when the toxin was administered intothe lateral ventricles on the day of proestrus (174). A recentclinical study of estrogen replacement in relation to Alzhei-mers disease revealed that the beneficial effects of tacrine, acholinergic-enhancing drug, were evident in women on es-trogen replacement therapy and not in women who did notreceive estrogen replacement therapy (175).

Taken together, these results point to a sexually differen-tiated organization of the basal forebrain cholinergic systemin the rat, involving a prenatally programmed difference inthe neuroanatomical organization as well as sex differencesin response to estradiol as far as cholinergic enzyme induc-tion in adult life and theeffects ofT3 treatment within the firstweek of postnatal life. These differences may underlie, atleast in part, the sex differences in spatial learning that arediscussed below.

C. Estrogens and the serotonergic system

Serotonin neurons of the midbrain/brainstem raphe nu-clei are among the earliest neuronal phenotype to becomedifferentiated during CNS development, and serotonin isbelieved to act as a regulatory/developmental agent (176,

177). The more rostral nuclei (primarily the dorsal and me-dial raphe) form ascending projections, densely innervatingsuch forebrain regions as the hypothalamus, hippocampus,and cortex. Thus, the serotonergic system is involved in theregulation of such diverse functions as reproduction, mood,sleep, and cognition. While serotonergic activity is regulatedby the ovarian steroids, the mechanisms by which such reg-ulation occurs are not fully understood. Here, we will brieflyreview findings that suggest involvement of both presyn-aptic and postsynaptic actions.

A sex difference in the serotonin system of the rat brain isestablished by the end of the second postnatal week (178).Female rats demonstrate higher serotonin levels and/or syn-thesis measured in whole brain (179), forebrain (180), raphe

(181), frontal cortex (182), hypothalamus (181183), and hip-pocampus (182, 184) compared with the male rat brain. Asimilar sex difference in rat brain serotonin turnover, anindication of serotonergic activity, has also been reported(180, 185). Furthermore, brain serotonin levels and activityare altered during periods of physiological ovarian hormonefluctuation, including the estrous cycle, pregnancy, or thepostpartum period (186190) in the rodent. In addition, es-trogen and/or progesterone treatment of ovariectomizedrats has been shown to positively affect the serotonergicsystem of the female rat brain (51, 191202).

In addition to reporting significant increases in hippocam-



11/29

pal serotonin levels and synthesis rate in females, Haleemand colleagues (184) found that female rats are much moreresponsive to the 5-HT1A receptor-mediated inhibition ofserotonin synthesis. That is, after the administration of the5-HT1A receptor agonist 8-hydroxy-2-(di-n-propylamino)-tetralin, female rats exhibited a decrease in hippocampalserotonin synthesis that was twice that seen in males. This

may be partly explained by the finding that estrogen treat-ment increases the efficiency of the 5-HT1A receptor to in-hibit cAMP formation in isolated membrane fractions in thehippocampus (203).

With regard to the estrogen sensitivity of serotonergicneurons, the direct or indirect nature of hormone action isonly now beginning to emerge at the level of theraphe nuclei.Estrogen-concentrating cells, determined by autoradiogra-phy, have been previously reported in the raphe nucleus inthe male and female lizard,Anolis carolinensis, but it was notdetermined whether the cells were serotonergic (204). Morerecently, in rhesus macaques, Bethea (205) demonstrated thepresence of estrogen-inducible PR in a majority of serotoninneurons, as well as in nonserotonin cells, in the dorsal andventral (medial) raphe of intact and spayed estrogen- andprogesterone-treated macaques. Because progesterone treat-ment of estrogen-primed macaques increases PRL release viaa serotonergic mechanism (206, 207), the finding of PR inserotonin neurons provides a direct means by which theovarian steroids can regulate serotonergic function. In ad-dition, Betheas group has demonstrated that ovarian hor-mones increase the expression of tryptophan hydroxylase(TPH), the key enzyme in serotonin biosynthesis, and sup-press expression of the serotonin transporter (SERT) in themacaque raphe nuclei (208, 209). For a recent review, see(210). While TPH mRNA levels do not appear to be regulatedby estrogen or progesterone in the rat dorsal raphe (S. Alves

and B. McEwen, unpublished results), SERT mRNA has beenreported to be increased by estrogen in the dorsal raphe ofthis rodent species (211). It is not presently known whetherthis difference in SERT regulation between the macaque andthe rat is due to difference(s) in species and/or length ofhormone treatment.

Curiously, the rat does not show localization of ER or PRin serotonergic neurons (38) in spite of the ample evidencefor ovarian hormone regulation of serotonergic function inthe rat brain. However, a number of ERand/or PR immu-noreactive neurons are found within the female and male ratdorsal raphe, adjacent to the serotonin cells (Fig. 4), suggest-ing transsynaptic regulation; females were found to havesignificantly more PR-containing cells, but no sex difference

in the number of ER-labeled cells was observed (38). Recentdata indicate that a subpopulation of these steroid target cellsdemonstrate immunoreactivity to the excitatory amino acids,glutamate and aspartate (212). ERmRNA has been reportedwithin the dorsal raphe of the rat (47), although ER proteinhas not yet been detected (38) (seeSection II.Cfor discussionof possible explanations for this type of discrepancy). Yetthere are estrogen effects in the rat serotonin system that mayeventually be explained by the presence of functional ER.

Estrogen and progesterone treatment alters the expressionof several genes within the rat dorsal raphe nucleus that areinvolved in serotonergic transmission: the postsynaptic

5-HT2A receptor (37, 213) and the presynaptic SERT (211)and vesicular monoamine transporter (VMAT2) (214). Thesedata suggest that ovarian steroids are likely to modulateserotonergic transmission at the dorsal raphe by regulatingboth nonserotonergic and serotonergic cells. Interestingly,recent findings indicate that the former two genes appear tobe similarly regulated by estrogen in females and males (37,

215), which is in agreement with the lack of a gender dif-ference in the number of ER-containing cells within thisnucleus. Ovarian steroid regulation of VMAT2 has been re-ported only in females, and in this studyit was demonstratedthat progesterone, either alone or in combination with es-trogen treatment, decreased VMAT2 mRNA to a similar ex-tent. It would be interesting to investigate whether suchregulation occurs in males, considering the gender differencein PR immunoreactive cells within the dorsal raphe (38).

Thus far, the only conclusion to be drawn is that ovarianhormones may work indirectly in the rat brain throughadjacent neurons that express ER and/or PR, and per-haps both directly and indirectly in the macaque raphe

nuclei, to influence serotonergic function at the midbrainlevel. However, as noted, the demonstration of functionalERin the dorsal raphe could change this interpretation,particularly for the rat. Moreover, the rat may be unusual,in that preliminary evidence from the mouse suggests thatERor PR immunoreactivity occurs in some TPH-labeledneurons, and abundantly in non-TPH cells in the dorsalraphe, suggesting direct steroid regulation of at least asubpopulation of serotonin cells in this rodent species(212). It should be mentioned that while estrogen-inducedPR have been identified in serotonin neurons in the ma-caque (as described above), ERs have not been detectedin the macaque raphe (C. L. Bethea, personal communi-

cation), once again raising the issue of a problem in antigendetection and/or rather that ER may be the functional ERin this species. Recent evidence from the ERKO mouseindicates abundant estrogen binding in the dorsal raphe(216), strengthening the idea that ER may play an im-portant role in this brain region.

Thus, by a multiplicity of pre- and postsynaptic mecha-nisms, ovarian steroids affect serotonergic function in a sex-ually dimorphic fashion, and these actions are relevant to theactions of estrogens on mood and cognition. High doses ofestrogens were reported to have antidepressant effects inhuman subjects (217), and estrogen treatment influences theresponse to antidepressant drugs in animal models (8) and

in clinical studies (10). Moreover, estrogen treatment ofovariectomized rats led to less struggling or immobility, andmore time swimming, in the forced swim test, a measure ofanxiety; and estrogen treatment reduced the number of cellsexpressing the immediate early gene, c-fos, during the forcedswim test (218). Both of these results are consistent with anantianxiety effect of estrogens in the rat and human. Indeed,results from a recent clinical trial of fluoxetine (Prozac, EliLilly, Indianapolis, IN) indicated that women receiving es-trogens as well as Prozac were the most responsive (10).However, in view of the small sample size in that study, thisis a finding that needs to be replicated in a larger study.



12/29

D. Catecholaminergic neurons

1. Noradrenergic system. In addition to the cholinergic andserotonergic systems, catecholaminergic systems respond toestrogens, i.e., brainstem catecholaminergic neurons (A6 andto a lesser extent A5 and A7) contain small numbers of ER(219), and estrogen treatment after gonadectomy exerts com-plex, time-dependent effects on the level of tyrosine hydrox-ylase mRNA (220). Moreover, recent studies of rats (221) andsheep (222) indicate that A1 and A2 noradrenergic neuronsspecifically express the ER and show cyclical and estrogen-

dependent patterns of immediate early gene expression (223,224). In the rat locus ceruleus, galanin is coexpressed in manynoradrenergic neurons, and estrogen treatment increased theexpression of galanin mRNA, leading to the speculation thatestrogen treatment might reduce noradrenergic tone in theabsence of separate effects on tyrosine hydroxylase expres-sion by enhancing the cosecretion of galanin, which reducesnoradrenaline release (225).

2. Dopaminergic systems. Incertohypothalamic dopamine neu-rons are distributed in the rostral, periventricular, caudal,

FIG. 4. Map of ER- and PR-containing neurons in the rat midbrain dorsal raphe region and surrounding periaqueductal gray (PAG). Eachdotrepresents approximatelytwo immunoreactive cells. A, Nuclear ERs (ER) in gonadectomized (GDX) rats. B, Nuclear PRsin GDXrats treatedwith estradiol benzoate for several days. The brain levels depicted are measured in distances from Bregma (B). Note the higher density of cellscontaining ERimmunoreactivity at the more rostral levels of the dorsal raphe nucleus (DRN) but the higher concentration of PR-immuno-reactive cells specifically within the lateral wings (LW) of the DRN, depicted at level B 8.30. In contrast to most other regions of the DRNexamined, this population of neurons maintains rather abundant PR immunoreactivity without E priming. AQ, Cerebral aqueduct; ELi, caudallinear raphe nucleus; V4,fourth ventricle. [Reproduced withpermission fromS. Alveset al.:J Comp Neurol 391:322334,1998 (38). Wiley-Liss,Inc., a subsidiary of John Wiley & Sons, Inc.]



13/29

and dorsomedial regions of the hypothalamus and representan internal source of dopamine innervation for the hypo-thalamus and preoptic region (226). The incertohypotha-lamic dopamine neurons express sex differences in neuronnumber and function (227). Estrogen and PRL have hetero-geneous effects on dopamine turnover, increasing it in dor-somedial nucleus and decreasing it in rostral periventricular,

medial preoptic, and preoptico-suprachiasmatic nuclei (228).In the midbrain dopaminergic projections to the corpus

striatum and nucleus accumbens, there are sexually dimor-phic actions of estrogens and progestins, involving both pro-and antidopaminergic effects that depend on the dose andtime course of estrogen administration and are manifested inboth the nigrostriatal and mesolimbic dopaminergic systems(229, 230). Estrogen facilitates amphetamine- or apomor-phine-stimulated dopamine release and locomotor activity inrats unilaterally lesioned by 6-hydroxydopamine (231233),and this activity is responsive to natural fluctuations in es-tradiol and generally increased during late proestrus andearly estrus (232234). Spontaneous sensorimotor activity isalso influenced by estrogens, e.g., in a tight-rope walkingtask (235). Coordination of locomotor activity may also in-volve estrogen actions in other brain regions, such as cere-bellum, where membrane actions of the steroid are suspectedon the basis of rapidity of effects and absence of knownintracellular receptors (1, 98). In spite of their rapidity, theseeffects must still be considered in terms of the presence ofER mRNA in cerebellum and cerebral cortex (48) eventhough functional ER protein has not yet been demon-strated (seeSection II.C).

In striatum, ovariectomy decreases, and administration ofestradiol potentiates, the depolarization-induced release ofdopamine as well as rotational behavior in a sexually di-morphic pattern (236 240).Male rats show smaller responses

to estrogen than females, and castration of males does notaffect the amphetamine stimulation of rotational behavior orstriatal dopamine release (234, 241, 242).

In females, no classical ER have been identified in striatum(63, 64, 243); nevertheless, intrastriatal application of estra-diol rapidly causes rotational behavior (244) and enhancessensorimotor performance (235). Estrogen directly potenti-ates potassium-stimulated dopamine release from rat nu-cleus accumbens (102), and estrogen pretreatment increasesthe firing rate of neostriatal neurons in response to dopamine(245), possibly via changes in D1 or D2 receptor coupling(246). A variety of estrogen effects on dopamine receptorbinding have been reported (see Refs. 229 and 230 for re-views).

Estrogen actions in the striatum that do not involve theclassical ER have been proposed on the basis of four types ofevidence: 1) the lack of intracellular ER in striatum; 2) therapidity of estrogen effects; 3) the pharmacological profile ofestrogen action, particularly the ineffectiveness of diethyl-stilbestrol; and 4) the ability of estradiol conjugated to BSAto mimic effects of free estradiol (247). One possible expla-nation, already described above, are the actions of estradiolto reduce L-type calcium channel activity in striatal neuronsvia a G-protein-coupled receptor (121).

The dopamine system shows declining function in theaging brain (248), and clinical observations indicate antido-

paminergic effects of moderate to high doses of estrogens.Relatively high levels of estrogens, including oral contracep-tives and estrogen replacement therapy, exacerbate symp-toms of Parkinsons disease (6, 249, 250), pointing to anti-dopaminergic actions that are opposite to the actions ofphysiological levels of estradiol. A similar antagonistic effectof chronic or high-dose estrogen was found in male and

female rats for drug-induced motor activity (251, 252).

E. Spinal cord

The spinal cord contains limited numbers of cells dem-onstrating intracellular ER, and there is also evidence forantinociceptive and analgesic actions of estrogens, with alarge sex difference that may be mediated at the spinal levelor at other levels of the neuraxis (for discussion see Refs. 253and 254). However, the information is rather limited regard-ing possible genomic or nongenomic mechanisms, and func-tional studies do not coincide with the information about ERlocalization.

ERs have been colocalized by immunocytochemistrywith enkephalin in many neurons in the medullary andspinal dorsal horn, particularly in the superficial laminaewhere they could be involved in modulating sensory andnociceptive processing (253, 254). A moderate concentrationof labeled cells expressing ERmRNA has been reported inlamina II of the spinal cord, whereas scattered cells express-ing ERmRNA were found in laminae I and II, the medialportion of laminae VI and VII, and in lamina X near thecentral canal (47).

Pain sensitivity differs strikingly between men andwomen and in women in different reproductive hormonestates (see below and Refs. 253 and 254). Sex differences inanalgesia have been reported in mice along with sex-specific

effects of estrogens. In particular, nonopioid analgesia pro-duced by swim stress was different between male and femaleSwiss-Webster mice and became equalized by ovariectomy;estrogen replacement of ovariectomized females reversedthe effect, but estrogen treatment of intact or castrated maleshad no effect, indicating an insensitivity of this system toestrogens in the male mouse (254). In a follow-up study,quantitative trait locus (QTL) mapping was carried out andled to the identification of a female-specific QTL on chro-mosome 8 (255). This female-specific mechanism, which issensitive to estrogen modulation, is consistent with a genethat is turned off by testosterone exposure during sexualdifferentiation (256). Because it involves nonopioid analge-sia, this form of estrogen-sensitive analgesia is unlikely to be

related to the enkephalin/estrogen colocalization describedabove or to NMDA-receptor mediated analgesia to whichmice are also insensitive; rather, a novel form of nonopioid,non-NMDA analgesia is indicated (254, 255). The role of ERmRNA expression in spinal cord and its relationship to func-tional ER receptors in this structure remain to be estab-lished.

F. Hippocampus

1. Cyclic synaptogenesis on hippocampal neurons. While syn-apses are formed and eliminated during development, syn-



14/29

aptogenesis was, until recently, believed to be more limitedin the adult nervous system. Estrogens regulate synapsedensity in the adult rat hypothalamic ventromedial nucleusthat differs between males and females (145, 257, 258). Thisdiscovery led to the finding that the ovarian cycle regulatescyclic synaptogenesis on excitatory spines in hippocampalCA1 pyramidal neurons in female but not in male rats (259,

260). Synaptogenesis is cyclic, and fluctuations in synapsedensity occur throughout the estrous cycle of the female rat(158). The increase in synapses on dendritic spines afterestrogen treatment is shown in Fig. 5, along with the decreasein spine synapse density that occurs between the days ofproestrus and estrus in cycling female rats. Male rats showmuch less estrogen-induced synapse formation unless theyare treated at birth with an aromatase inhibitor (260). Thissuggests that the developmentally regulated expression ofERs and aromatase activity in hippocampus (261, 262) isinvolved in programming the response of the adult hip-pocampus.

Cyclic synaptic turnover in the hippocampus and hypo-thalamus during the estrous cycle of a female rat shows ahigh degree of specificity. For example, in hypothalamus,neurons of the ventromedial nucleus show cyclic synapto-genesis directed by ovarian steroids. In CA1 pyramidal neu-rons of the hippocampus, estrogen-induced synaptogenesisoccurs on dendritic spines and not on shafts, and there areno estrogen effects on dendritic length or branching; more-over, as far as one can tell, such synaptic plasticity is ex-tremely specific and does not occur on CA3 pyramidal neu-ronsor dentate gyrus granule neurons (263).The discretenessand specificity of this synapse formation imply that molec-ular markers may be very specific or subtle and that themechanism may involve changes in a limited number ofcellular events, including transcription of discrete structural

genes and posttranscriptional events such as translation ofmRNAs for structural proteins. Moreover, local regulation,as via afferent input or interneurons, may be very important.

One of the surprises of the synaptogenesis story is thatestrogen induction of synapses is blocked by NMDA recep-tor antagonist treatment, indicating that excitatory aminoacids and NMDA receptors are involved in synapse forma-tion (264, 265). Progesterone secreted at the time of ovulation

appears to be responsible for down-regulation of estrogen-induced synapses in the CA1 region (266), and the cellularlocation of PRs, as well as of the ERs, is a prime question.

2. Localization of intracellular ERs. The presence of the classicalERand the recently discovered ERcomplicates the storyof estrogen action. ERs have been identified by immuno-

cytochemistry in scattered GABA-ergic interneurons in therat hippocampus (39), and this distribution of ER is in agree-ment with autoradiography of [3H]estradiol uptake (267), sothat one does not need to postulate the existence of anotherhigh-affinity intracellular ER. The localization of ER is sum-marized for the hippocampus and adjacent cerebral cortex inFig. 6. ERexpression has been claimed in pyramidal neu-rons by immunostaining and also mRNA expression (46, 47,65), although, as mentioned previously, our laboratory hasnot seen consistent ER immunostaining in hippocampus(N. Weiland, S. E. Alves, V. Lopez, and K. Bulloch, unpub-lished). Clearly, more studies are needed on this issue. Thereare a number of plant estrogens, genestein and daidzein,with approximately 20-fold higher affinities for ER thanER, which makes them useful to discriminate between thetwo receptor types (55, 268), and these may be useful infurther studies on the role of ER in the hippocampus.

3. Role of the NMDA receptors. Antagonists of NMDA recep-tors blocked estrogen-induced synaptogenesis on dendriticspines in ovariectomized female rats (264, 265). Because es-trogen treatment increases the density of NMDA receptors inthe CA1 region of hippocampus (265, 269, 270), it is possiblethat activation of NMDA receptors by glutamate is the pri-mary action that induces new excitatory synapses to develop.Spines are occupied by asymmetric, excitatory synapses, andthey are sites of Ca ion accumulation and thus ideal sites

for NMDA receptors (271). NMDA receptors are expressedin large amounts in CA1 pyramidal neurons and can beimaged by conventional immunocytochemistry as well as byconfocal imaging (270), in which individual dendrites andspines can be studied for colocalization with other markers(272, 273). NMDA receptor mRNA can also be measured byin situ hybridization, and four different forms show differentregional patterns and developmental regulation (274).

FIG. 5. Depiction of ovarian steroidregulation of the density of excitatoryspine synapses in the CA1 region of thefemale rat hippocampus. Estimateddensity of synapses on dendritic spinesin the stratum radiatum of the CA1 re-gion of the hippocampus, in panel A, ofan ovariectomized (OVX) rat treatedwith oil (O) or estradiol (E) or, in panelB, of intact rats in the proestrus or es-trus phase of the estrous cycle. Valuesrepresent mean SEM obtained usingthe Disector method. Note that in eachcase, higher estradiol levels are corre-lated with a greater density of syn-apses. Data were analyzed with un-paired, two-tailedt tests, n 4 in eachcase. *, P 0.025. [Reproduced withpermission from C. Woolley and B. S.McEwen: J Neurosci 12:25492554,1992 (158).]



15/29

NMDA receptors are implicated in other morphogeneticprocesses in the adult brain such as suppressing neurogen-esis in the dentate gyrus (275), and they are also involved inthe developing nervous system as facilitators of neuronalmigration (276, 277). However, there is a noteworthy para-dox, in that NMDA receptors are implicated during visualsystem development in the reduction of synaptic contact in

the developing retinal axon arbors (278), and NMDA recep-torblockaderesults in rapid acquisition of dendritic spines byvisual thalamic neurons (279). It appears likely that hip-pocampus and visual system neurons respond in oppositeways to NMDA receptors, since a recent report on embryonichippocampal neurons in culture (see Section III.F.5 below)indicates that NMDA receptor blockade prevents estrogen-induced synaptogenesis (280).

4. Genomic vs. nongenomic actions of estrogens on synapse for-mation. The paradoxical estrogen effects on hippocampal py-ramidal neurons that do not appear to have intracellular ERor show uptake and cell nuclear retention of [3H]estradiolmight be explainable if there were cell surface ERs. Rapidestrogen effects on CA1 pyramidal neurons of the hippocam-pushave beendescribed by in vitro electrophysiological stud-ies on slices from this brain region, and these appear to involvenon-NMDA excitatory amino acid receptors (94, 95) that arevery likely to be AMPA (-amino-3-hydroxy-5-methyl-4-isox-azole propionic acid) receptors (96). One approach to rule in orout nongenomic actions of estrogen would be to study ERKOmice lacking intracellular ER (43), and a recent study withmice lacking ERhas shown that estrogen actions on kainate-stimulated ionic currents are still present (103). An ER doubleknockout would be even better, provided there are only twointracellular ER genes.

Another approach to discriminate between classical intra-

cellular ER and membrane ER is to use antiestrogens thatbind to the intracellular ER but which mimic, rather thanblock, the rapid membrane effects, such as was the case forestrogen effects on calcium currents in neurons from thecorpus striatum (121). Antiestrogens also have another use,namely, to discriminate between the response elements thatthe ER uses to activate transcription. As noted above, themajor differences between ER and ER1 concern their abil-ity to regulate transcription via the AP-1 response element.For interactions of ERwith AP-1, 17-estradiol as well as anumber of antiestrogens activated transcription; however,for ER1 interacting with AP-1, 17-estradiol failed to acti-vate transcription but antiestrogens activated transcription(69).

Estrogen antagonists have been very useful in testing al-ternatives to conventional genomic actions of estrogen onhippocampal synapse formation by providing pharmacolog-ical evidence in favor of a particular pathway of hormoneaction and against other possible mechanisms (281). Theantiestrogen, CI-628, has previously been shown to enter thebrain and block estrogen induction PR (see Fig. 3). The samedose of CI-628 that blocked PR induction was also able toblock spine synapse induction by estrogen in the hippocam-pus, and CI-628 did not have any agonist-like activity of itsown (281) (see Fig. 7). An agonist-like action of CI-628 wouldhave been expected had it exerted its action nongenomically

FIG. 6. MapofER immunoreactivity in the hippocampus and cortexof the rat brain. [Drawings are modified from L. W. Swanson, BrainMaps: Computer Graphic Files (version 1.0), and represent coronalsections from three different levels of brain measured from bregma.]Each dot represents one ER-immunoreactive cell (total number is

mean from male and female rats, which do not differ significantlyfrom each other). Note the ERimmunoreactivity cells are interneu-rons, not pyramidal neurons, in agreement with previous autoradio-graphic studies (see text). DGlb, Lateral blade of the dentate gyrus;ENT, entorhinal cortex; fc, fasciola cinerea; fi, fimbria; hf, hilar fis-sure; mo, molecular layer of the dentate gyrus; PAR, parietal cortex;po, polymorph layer (hilus) of the dentate; RSP, retrosplenial cortex;sg, stratum granulosum; so, stratum oriens; sp, stratum pyramidale;sr, stratum radiatum; SUB, subiculum;v3, third ventricle; vip, veluminterpositum; vl, lateral ventricle. [Reproduced with permission fromN. G. Weilandet al.:J Comp Neurol 388:603612, 1997 (39). Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc.]



16/29

via calcium channels, as has been shown for striatal neurons(121). An agonist-like action might also have occurred viaER or -, or a heterodimer, acting via another responseelement than the ERE (see Section II.C). The fact that CI-628blocked, rather than mimicked, estrogen action is inconsis-tent with any known nongenomic effect and is similar to theestrogen induction of PRs that is believed to involve an ERE(282). Moreover, it is consistent with an action of estrogen viathe intracellular ERs that are knownto exist in hippocampalinterneurons, although, again, it should be pointed out thatER could also mediate actions via an ERE and that thepresence of some functional ER in hippocampus is still adistinct possibility given the presence of ERmRNA in thisbrain region (see Section II.C).

5. Synapse formation in cultured hippocampal neurons. Recentstudies on hippocampal neurons in culture have revealed

that estrogen induces spines on dendrites of dissociated hip-pocampal neurons in culture by a process that is blocked byan NMDA receptor blocker and not by an AMPA/kainatereceptor blocker (283). In a subsequent study, estrogen treat-ment was found to increase expression of PCREB and CREB,and a specific antisense to CREB prevented both the forma-tion of dendritic spines and the elevation in PCREB (284).

ERs have been located on glutamic acid carboxylase(GAD)-immunoreactive cellsin vitrothat constitute approx-imately 20% of neurons in the culture (39), and this is con-sistent with in vivo data summarized above. 17-Estradioltreatment of cultured cells caused GAD content and thenumber of neurons expressing GAD to decrease, and mim-icking this decrease with an inhibitor of GABA synthesis,mercaptopropionic acid, caused an up-regulation of den-dritic spine density, simulating the effects of 17-estradiol(285). Figure 8 summarizes the hypothesized interaction be-tween these GABA interneurons and the pyramidal neuronsupon which the synapses are induced by estrogen treatment.Both cell culture data andin vivostudies summarized aboveare consistent with this model.

An additional factor in the formation of dendritic spinesin thein vitro cell culture model is the neurotrophin, brain-derived neurotropic factor (BDNF), which is expressed inGABA interneurons in hippocampal cell cultures (286). Inaddition to down-regulating GABA in these interneurons,estrogen treatment also reduced BDNF by 60% within 24 h(286). Since exogenous BDNF blocked estrogen induction ofdendritic spines and BDNF depletion with an antisense orblockade with BDNF antibodies both mimicked estrogen ininducing spine density, the authors suggest that BDNF is animportant player in the regulation of GABA inhibition,which in turn blocks activity-dependent regulation of den-dritic spines in hippocampal neurons (286). It is interesting

to note that neurotrophins such as BDNF and NT-3 alsoincrease the function of inhibitory and excitatory synapses inhippocampal cell cultures, and BDNF causes an increase inaxonal branching and length of GABA-ergic interneurons(287).

6. Developmentally regulated sex differences in the hippocampus.The hippocampus is one of a number of extrahypothalamicbrain structures that shows subtle sex differences. For ex-ample, there are sex differences in the density of apical den-dritic excrescences and branching of dendrites of CA3 py-ramidal neurons. Treatment with T3during the first week ofpostnatal life enhanced these differences (288). Excrescenceson the proximal region of apical dendrites receive input from

mossy fiber synapses from granule neurons of the dentategyrus. Therefore, the greater density of excrescences in malesis consistent with a report that male rats have a greaternumber of mossy fiber synapses than females (289). Otherstudies have pointed to sex differences in hippocampal mor-phology that are dependent on the rearing environment(290).

The dentate gyrus of mice and rats also shows sex differ-ences. In mice, there are strain-dependent sex differences: instrains with large numbers of granule neurons, males havemore neurons than females, while in strains with fewer gran-ule neurons,the sexes do not differ from each other in neuron

FIG. 7. Effect of the estrogen antagonist, CI-628, on estrogen induc-tion of increased dendritic spine density on CA1 pyramidal neurons.CI-628 (10mg/kg) wasgivenat 0, 24,and 48 h, andestradiol benzoate(EB, 10g/kg) wasgiven at 24 and 48 h to ovariectomized femalerats,which were then killedat 72h. Thesingle-sectionGolgi procedurewascarried out to stain dendrites of CA1 pyramidal neurons for visual-ization of dendritic spines. Data are expressed as the number ofspines/10m length on secondary dendrites that were greater than10 m in length andlocated 150200m awayfromthe cell body. SixCA1 neurons fulfilling the criteria described in the original publica-tion (281) were analyzed for each rat brain. The number of rats pergroup was six for the CI628 EB group and five for each of the othertreatments. The error bars show the SEM; this is based on the meanand variance calculated across animals, with data for the six neuronsof each animal in a treatment compiled into a single average. Sta-tistical analysis of data revealed that there was an overall treatmenteffect,P 0.0001, by one-way ANOVA. A Tukey post hoc comparisonrevealed a clear E induction of spines, in which ovx E was differentfrom each of the other groups (P 0.001). Moreover, CI-628 partiallyblocked the E effect, in that ovx E CI-628 was significantlyelevated comparedwith OVX (P0.01) and significantlyless thatovxE (P0.001). There was no agonisteffect of CI-628 byitself onspinedensity (P 0.92). [Reproduced with permission from B. S. McEwenet al.: Endocrinology 140:10441047, 1999 (281). The EndocrineSociety.]



17/29

number (291). Male rats have a larger and more asymmetricdentate gyrus than females, and neonatal testosterone treat-ment caused the genetically female dentate gyrus to appearmale like (292). Neonatal testosterone treatment in femalerats also improved spatial learning ability in a Morris watermaze (292).

How do these sex differences come about during devel-opment? Like the cerebral cortex, the rat hippocampus ex-presses ERtransiently during perinatal development (261,293). The presence of these receptors in hippocampus coin-cides with the transient expression of the aromatizing en-zyme system that converts testosterone to estradiol (294); asa result, ERin male rats would be exposed to locally gen-erated estradiol, and this could lead to sexual differentiationof hippocampal structure and function. Consistent with thisscenario are data showing that, while neonatal castration ofmale rats produced female-like learning curves in a Morriswater maze, the administration of estradiol to newborn fe-male rats produced a male-like learning curve (295).

It should be noted that the cell culture model describedabove for studying estrogen-induced synaptogenesisin vitrolies right on the interface between developmental actions ofestrogens and the activational effects in mature neurons.That is, the cell cultures are generated from late fetal brain

tissue before the stage of sexual differentiation has beencompleted; however, the fact that the hippocampal cell cul-tures are allowed to mature in vitro, differentiate into exci-tatory and inhibitory neurons, and form synaptic connec-tions makes them more like the mature nervous system. It isinteresting to consider whether application of gonadal ste-roids during the differentiation and formation of synapticconnection might mimic aspects of the sexual differentiationof hippocampal circuits and functions described above andmight lead, for example, to a permanent male-like inabilityof the cultures to show synapse induction in response toestradiol.

7. Comparison with other forms of structural plasticity.The hip-pocampus also undergoes two other forms of plasticity, inwhich circulating hormones and excitatory amino acids act-ing via NMDA receptors are involved. One of these is theongoing neurogenesis in the adult rat dentate gyrus, whichcontinues for at least 1 yr after birth and can be increasedeither by adrenalectomy or by treatment with an NMDAreceptor antagonist (275). Although the male dentate gyrusis larger than that of the female (296), there are data for theprairie vole (297) and rat (298) indicating that estrogens in-crease neurogenesis of granule neurons in the female. Thus,it remains to be established for males and females what thebalance is between neurogenesis and programmed cell deathto account for sex differences in overall neuron number be-tween the sexes.

Dentate gyrus granule neurons innervate the CA3 regionof Ammons horn, and stress causes apical dendrites of CA3pyramidal neurons to undergo atrophy by a process that isdependent in part on circulating adrenal steroids and in parton excitatory amino acids acting via NMDA receptors (299).Stress-induced dendritic atrophy is also reversible (A. M.Magarinos and B. S. McEwen, unpublished), but severe andprolonged social stress (in vervet monkeys) and cold-swimstress (in rats) causes CA3 pyramidal neuron loss in males

that is not evident in females (164, 165). Thus, there is thepossibility that intrinsic sex differences in hippocampal mor-phology or in response to hormones or excitatory aminoacids may have a protective role in the female.

A recent study indicates that female rats are also resistantto the stress-induced atrophy of CA3 pyramidal neurons inhippocampus (300). In addition to the larger dentate gyrusof the male (296), male CA3 neurons have more excrescencesfor mossy fiber contacts, while female CA3 apical dendritesare more extensively branched (288). However, it is not clearhow these differences might contribute to the sex differencesin the effects of stress.

FIG. 8. This model of synaptogenesisinthe hippocampus emphasizes the role ofNMDA receptors and the key role of in-hibitory GABA interneurons. ER is

present in interneurons, and its pres-ence coincides with the distribution ofER-binding sitesfrom in vivo [3H]estra-diol autoradiography. According to thebest evidence to date, based upon im-munocytochemistry of hippocampusand cell culture studies, estrogens sup-press GABA function transiently andlead to disinhibition of a large numberof innervated CA1 neurons resulting inup-regulation of NMDA receptors andsynapse formation. Blocking NMDA re-ceptors prevents estrogen-induced syn-apse formation.



18/29

8. The functional significance of synaptogenesis in the hippocam-pus. The functional significance of synaptogenesis in thehippocampal CA1 region has been shown in electro-physiological studies indicating that estrogen treatment ofovariectomized rats produces a delayed facilitation of syn-aptic transmission in CA1 neurons that is NMDA mediated(95) and leads to an enhancement of voltage-gated Ca

currents (94, 95). This approach has now been taken to a newlevel by Woolley, who has used biocytin injection and im-munostaining after recording from CA1 pyramidal neuronsto visualize estrogen induction of spines; she found thatspine density correlates negatively with input resistance andthat input/output curves show an increased slope underconditions in which NMDA receptor-mediated currents pre-dominate, whereas there is no increased slope where AMPAreceptor currents predominate (265). Moreover, in intact fe-male rats, there is a peak of LTP sensitivity on the afternoonof proestrus in female rats at exactly the time when excitatorysynapse density has reached its peak (301).

Proestrus is also the time of the estrous cycle when seizurethresholds in dorsal hippocampus are the lowest (302). Be-cause activation of NMDA receptors in hippocampus is en-hanced via AMPA receptors in some cases but not in others(303), it remains to be seen how plastic the AMPA receptorsystem is to ovarian steroid manipulations or whether theestrogen-induced synapses are so-called silent synapses orones in which AMPA receptors are induced by LTP. Block-ade of AMPA receptors with 6-nitro-7-sulfamobenzo(f) qui-noxaline-2,3-dione during estrogen treatment failed to blocksynaptogenesis (264), which suggests that AMPA receptorsdo not pl

efectele estrogenilor asupra snc

Documents