This Article Abstract Full Text (PDF) All Versions of this Article: 28/1/1    most recent Final Manuscript Author Manuscript Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Vasudevan, N. Articles by Pfaff, D. W. Search for Related Content PubMed PubMed Citation Articles by Vasudevan, N. Articles by Pfaff, D. W.

Membrane-Initiated Actions of Estrogens in Neuroendocrinology: Emerging Principles

Department of Biology (N.V.), Pennsylvania State University, University Park, Pennsylvania 16802; and Laboratory of Neurobiology and Behavior (D.W.P.), The Rockefeller University, New York, New York 10021

Correspondence: Address all correspondence and requests for reprints to: Dr. Nandini Vasudevan, 208 Mueller Laboratory, Department of Biology, Pennsylvania State University, University Park, Pennsylvania 16802. E-mail: nuv1{at}psu.edu

	Abstract

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
Hormonal ligands for the nuclear receptor superfamily have at least two interacting mechanisms of action: 1) classical transcriptional regulation of target genes (genomic mechanisms); and 2) nongenomic actions that are initiated at the cell membrane, which could impact transcription. Although transcriptional mechanisms are increasingly well understood, membrane-initiated actions of these ligands are incompletely understood. Historically, this has led to a considerable divergence of thought in the molecular endocrine field.

We have attempted to uncover principles of hormone action that are relevant to membrane-initiated actions of estrogens. There is evidence that the membrane-limited actions of hormones, particularly estrogens, involve the rapid activation of kinases and the release of calcium. Membrane actions of estrogens, which activate these rapid signaling cascades, can also potentiate nuclear transcription. These signaling cascades may occur in parallel or in series but subsequently converge at the level of modification of transcriptionally relevant molecules such as nuclear receptors and/or coactivators. In addition, other hormones or neurotransmitters may also activate cascades to crosstalk with estrogen receptor-mediated transcription. The idea of synergistic coupling between membrane-initiated and genomic actions of hormones fundamentally revises the paradigms of cell signaling in neuroendocrinology.

I. Introduction

II. Estrogens Signal via Both Genomic and Nongenomic Modes

A. The genomic mode of estrogen action

B. Membrane-initiated (nongenomic) actions of estrogens: the use of membrane-limited conjugates as an experimental probe in neural cells

C. Membrane-initiated (nongenomic) actions of estrogens in neural cells

III. Membrane-Initiated Hormone Actions Potentiate Their Transcriptional Actions

A. Coupling of the membrane-initiated actions of estrogen to transcription: the consequence for reproductive behavior

B. Coupling of membrane-initiated actions of estrogens to transcription in nonreproductive functions

C. The coupling of membrane-initiated actions to transcription is studied in vitro by using a two-pulse transcriptional paradigm in cell culture

IV. The Signal Transduction Cascades Initiated by Estrogens at the Membrane May Occur in Parallel or in Series

A. Membrane-initiated actions of estrogens may also occur in parallel or in series in the MCF-7 breast carcinoma cell line

V. Convergence of the Membrane-Initiated Actions of Hormones to Transcription Involves Protein-Protein Interactions, Protein Translocation, and Protein Phosphorylation

A. Coupling to transcription via stabilization or dimerization of molecules relevant to transcription

B. Coupling to transcription via the physical movement of molecules into the nucleus

C. Coupling to transcription via phosphorylation of molecules relevant to transcription

VI. Initiation of Membrane-Initiated Actions at the Membrane Involves Both Classical "Nuclear Receptors" and Other Proteins

A. The membrane estrogen receptor could be a novel protein in some cell types

B. The membrane estrogen receptor is the classical ER in other cell types

C. Do different cell types have different receptors?

VII. Summary and Future Directions

	I. Introduction

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
ESTROGENS ARE IMPORTANT endocrine effectors of reproduction (1, 2, 3). They also play an important and sometimes permissive role in liver and cardiovascular physiology, neuronal growth and differentiation, neuroprotection, cognition, and regulation of mood (4, 5, 6, 7). They are ligands for estrogen receptors (ERs) and ß, members of the nuclear receptor superfamily. These isoforms, have a modular "domain" structure, a characteristic that they share with other members of the nuclear receptor superfamily (8, 9, 10, 11, 12). The molecular mechanisms by which estrogens change the response of target tissues can be broadly divided into two categories: 1) genomic actions, and 2) nongenomic actions. The focus of this review is the interplay between both of these mechanisms with respect to transcription, particularly in the central nervous system (CNS). Each of the principles of estrogen action, stated as a section heading, is elaborated upon in that section.

	II. Estrogens Signal via Both Genomic and Nongenomic Modes

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
A. The genomic mode of estrogen action
Genomic actions are those wherein the ER exerts transcriptional effects via interaction with estrogens (see Ref. 12 and references therein). In the absence of estrogens, the ER in various target tissues is associated with heat-shock proteins in a transcriptionally inactive state (13). Binding of 17ß-estradiol, the natural endogenous ligand, or a similar agonist molecule induces a conformational change in the ER, promoting ER homodimerization and subsequent nuclear translocation. In the nucleus, the ER acts as a ligand-dependent transcription factor, binding estrogen response elements (EREs) regulating gene transcription (9, 10, 11, 14) (12). This mechanism is of importance in pathological states such as hormone-dependent breast cancers (15, 16, 17). Therapy that prevents hormone binding to ER can decrease cell proliferation, whereas the presence of 17ß-estradiol can increase the proliferation rate (14). In the CNS, the focus of this review, several genes are also regulated in this manner by estrogen-bound ER. For example, in ER knockout mice, estrogen cannot induce the progesterone receptor (PR) gene in the hippocampus, indicating that genomic effects on this important estrogen-marker gene required the ER (18). Increased transcription of neurotrophins such as brain-derived neurotrophic factor (BDNF) and their high-affinity receptors (19, 20, 21), the oxytocin-oxytocin receptor genes (22, 23, 24), the preproenkephalin gene (25, 26, 27), the antiapoptotic genes such as Bcl-2 (28, 29), are important in neuroprotection (7), neurodevelopment (30), and reproductive behaviors driven by estrogens (1, 31, 32). Generally, cellular signaling cascades activated by estrogens and other hormones were thought not only to be disparate from classical genomic actions (33, 34, 35, 36) but also to lead to fewer or less important phenotypes.

B. Membrane-initiated (nongenomic) actions of estrogens: the use of membrane-limited conjugates as an experimental probe in neural cells
First reported in 1967 by Szego and Davis (37), acute 17ß-estradiol administration could increase cAMP in the uterus of ovariectomized mice within 15 sec, in stark contrast to the slower genomic actions of estrogens, which take hours for final changes in protein expression to occur, after transcriptional regulation. Nongenomic actions of estrogens can therefore be defined as rapid (milliseconds to minutes) effects, generally initiated at the plasma membrane by 17ß-estradiol, resulting in the activation of signal transduction pathways, including calcium flux, within target cells. The nature of the receptor that binds 17ß-estradiol at the plasma membrane is immaterial to this definition. Studies that attempt to conclusively demonstrate membrane-initiated estrogen effects that are confined to the membrane often utilize membrane-limited estrogen conjugates, such as E2-BSA (17ß-estradiol linked to BSA) (38) or E2-HRP (17ß-estradiol linked to horseradish peroxidase) to separate membrane-initiated actions of these conjugates from effects that arise from the steroid inside the cell. These compounds consist of 17ß-estradiol molecules (32–38 moles of 17ß-estradiol/BSA conjugated to the BSA molecule at a specific site, such as to the sixth carbon via an oxime linkage [17ß-estradiol-6-O-carboxymethyloxime-BSA (E2–6-BSA)], to the 17th carbon [1,3,5, (10)-estratrien-3,17ß-diol 17-hemisuccinate-BSA (E2–17-BSA)] or to the third carbon (39). In the immortalized GT-1 cell line, both 17ß-estradiol and E2–6-BSA increase cAMP (40). Similarly, in primary cultures of neostriatal neurons, both 17ß-estradiol and E2–6-BSA reduce calcium currents (41). Although most studies have used one or sometimes several of these compounds to mimic a membrane-limited action for 17ß-estradiol, several caveats should be noted regarding the use of these conjugates. There has been considerable controversy over the possible release of free 17ß-estradiol from membrane-limited conjugates and the possibility of endocytosis of the conjugate itself with the subsequent release of free 17ß-estradiol into cells. Stevis et al. (42) demonstrated that both E2–6-BSA and E2–17-BSA had free, presumably genomically active, 17ß-estradiol in the preparation as measured by RIA and required filtration to remove the free 17ß-estradiol. This suggests that, in some studies, signal transduction pathways that were thought to have been initiated at the membrane might have been initiated within the cell by free 17ß-estradiol present in the unfiltered E2-BSA preparation. In addition, the activity of E2-BSA conjugates has not always paralleled the activity of free 17ß-estradiol itself. For example, E2-BSA devoid of free 17ß-estradiol (achieved by filtration) did not bind either ER or ERß or to an ERE under in vitro conditions where 17ß-estradiol did bind both ER isoforms and the ERE (42). In the SK-N-SH neuroblastoma cell line, filtered E2-BSA induced MAPK, whereas free 17ß-estradiol did not do so (42). This suggests that the E2-BSA binding conformation to the ER differs from that of 17ß-estradiol with a subsequent change in the ability of the receptor to bind or transactivate genes that are regulated by EREs. This could result in artifactual results, i.e., nonphysiological activation of rapid signal transduction pathways, with E2-BSA if ER or ERß acts as the membrane ER.

Within neurons, studies from Susan Wray’s laboratory also show functional differences between different E2-BSA conjugates in the activation of the protein kinase A (PKA) pathway and calcium flux. Although both 17ß-estradiol and filtered E2–17-BSA increased intracellular Ca²⁺ levels in GnRH-positive cells in nasal explants, the transcription inhibitor, 5,6-dichloro-1-ß-D-ribobenzimidazole (DRB) could abrogate the 17ß-estradiol-mediated increase but not the E2–17-BSA increase in calcium in explants (43). In a primary culture of GnRH neurons, E2–6-BSA did not increase cAMP response element (CRE) binding protein (CREB) phosphorylation although 17ß-estradiol did so (44). This suggests that the mode of conjugation of the 17ß-estradiol moiety to BSA may allow different conjugates to contact different membrane ERs, i.e., different proteins. Alternatively, binding of free 17ß-estradiol and E2-BSA conjugates to the same receptor protein could lead to differences in protein conformation, which presumably lead to differences in calcium flux. All of these studies using filtered E2-BSA conjugates and 17ß-estradiol demonstrate that different signal transduction pathways could be activated by different E2-BSA conjugates or 17ß-estradiol, depending on cell type. One implication is that different cell types possess varied combinations of proteins capable of binding E2-BSA and/or free 17ß-estradiol. Recently, the Katzenellenbogen laboratory has synthesized estrogen dendrimer conjugates that are stable, do not seem to leach free 17ß-estradiol, and do not enter the nucleus of breast carcinoma cells to conduct genomic actions. They do not induce transcription of endogenous genes and can activate the ERK pathway rapidly at 2 min after treatment at levels similar to free 17ß-estradiol (45). These studies underscore the importance of using free 17ß-estradiol (the endogenous ER ligand) as a comparison to the artificial conjugate, E2-BSA in all studies. If the actions of the artificial conjugate, E2-BSA, and 17ß-estradiol are comparable, the E2-BSA preparation should preferably have been filtered and demonstrated to possess no transcriptional activity to conclude a true nongenomic, membrane-initiated effect. Unfiltered E2-BSA could show effects similar to 17ß-estradiol due to the presence of free 17ß-estradiol in the preparation. It should be noted that very few studies, particularly those before 2000, have explicitly shown that the E2-BSA preparation used in the study has no free 17ß-estradiol or no genomic activity.

What, therefore, could be considered a nongenomic effect? If filtered E2-BSA is not used in the study, an acute application of 17ß-estradiol that activates classically nongenomic effects (i.e., kinase activation or calcium flux) in a short time frame could also be considered to be a membrane-initiated action of estrogen. However, calcium flux could also be regulated nongenomically by 17ß-estradiol acting directly on intracellular calcium stores, apart from being initiated at the membrane. Similarly, kinase activation could also be a result of stabilization or localization; these processes may be nongenomically activated by estrogens but may not be plasma membrane-initiated.

C. Membrane-initiated (nongenomic) actions of estrogens in neural cells
In this review, we will discuss nongenomic actions of estrogens in the CNS only when they occur in physiologically relevant contexts, which will also be described to lend perspective to these actions. In the CNS, membrane-initiated actions of estrogens have been observed predominantly in the context of reproduction, neuroprotection, and neurotrophism, all of which are also affected significantly by the genomic actions of estrogens.

1. Nongenomic actions play a dominant role in the estrogen effect on neuronal excitability.
Estrogens have a rapid effect on neuronal excitability; this has been typically studied using neurons from the hypothalamus relevant to reproductive physiology, such as those that produce GnRH or neurons in the ventromedial hypothalamus (VMH) that control reproductive behavior or lordosis. In hypothalamic slice cultures from ovariectomized guinea pigs, 17ß-estradiol could uncouple the µ-opioid receptor in pro-opiomelanocortin (POMC) neurons and the -aminobutyric acid (GABA) (B) receptors from dopaminergic neurons from the small conductance, G protein-coupled inwardly rectifying K⁺ channels (GIRK), resulting in lower efficacy of µ-opioid and GABA agonists activation of these channels and lower hyperpolarization of these cells (46). This increase in 17ß-estradiol-mediated excitation is blocked by a variety of inhibitors to intracellular signaling cascades, such as phospholipase C, PKA, and protein kinase C (PKC) (46, 47). The uncoupling effect can be mimicked by membrane impermeant E2-BSA and the selective ER modulators, raloxifene and 4-hydroxytamoxifen (48). The inhibitor data, along with the blockade of uncoupling by an ER antagonist, ICI 182,780, suggest that the ER is G protein-coupled in hypothalamic neurons (36). An alternate explanation could be that the ER is required intracellularly for this effect, whereas the rapid signaling cascades are activated independently by a membrane ER (mER) coupled to a G protein. Qiu et al. (49) have proposed a model whereby a G_q coupled ER at the membrane initiates this signaling cascade of enzymes, which appear to be arranged in series. Using subtractive hybridization and microarray analysis, Malyala et al. (50, 51) showed that estrogen altered the mRNA levels for a number of neurotransmitter receptors and synaptic vesicle release molecules [e.g., GABA (B) R2 subunit and synaptobrevin] as well as for enzymes in rapid signaling cascades [i.e., phosphotidylinositol-3-kinase (PI3K) p55, gec-1, Rab 11a GTPase]. Hence, these molecules could be regulated at the level of enzyme activation by rapid, membrane-initiated mechanisms and at the level of transcription by genomic mechanisms. However, at least some part of the effect of 17ß-estradiol on neuronal excitability is solely rapid and membrane-initiated, as evidenced by both the short time scales of application and the data that cycloheximide use in hypothalamic slices could not block the uncoupling of µ-opioid receptors to the GIRK channels (52). Contrary to the evidence that a G_q receptor may increase excitability in the arcuate nucleus, in CA1 hippocampal neurons, this could be attributed to a G_s receptor that is activated rapidly by estrogens. 17ß-Estradiol as well as E2-BSA rapidly potentiated kainate-induced currents via PKA activation initiated by a G_s receptor (53). This was preserved in both wild-type and ER knockout mice, suggesting that the classical ER was not required for the increase in excitability in CA1 hippocampal cells in response to kainate (54). The increase in neuronal excitability by estrogens is a scenario where nongenomic effects of estrogens are probably dominant over genomic effects.

2. Effects of estrogen on neuroprotection involve the activation of kinases, calcium flux, and the induction of genes.
Estrogen-mediated neuroprotection against various agents such as ischemia, glutamate, and N-methyl-D-aspartate (NMDA)-mediated excitotoxicity and degenerative diseases such as Parkinson’s and Alzheimer’s disease (5, 55) has attracted intense interest. A majority of this work stems from the finding that intact female rats had lower mortality and better neuronal survival than male rats after middle cerebral artery occlusion (MCAO) but that this effect was eliminated by ovariectomy (56). Administration of 17ß-estradiol, but not progesterone, to ovariectomized female rats (57, 58, 59), aged reproductively senescent rats (60), and male rats (61) reduced infarct volumes after MCAO. In general, to achieve neuroprotection against excitotoxicity and ß-amyloid-mediated toxicity in neurons, estrogens regulate molecules that are involved in programmed cell death and inflammation in a multifaceted manner using both genomic and membrane-initiated modes of action. Central to neuroprotection in hippocampal neurons and in NT2 cells is the up-regulation by 17ß-estradiol of the antiapoptotic genes Bcl-2 and Bcl-XL (28, 62, 63) and the down-regulation of Nip2, an inhibitor of Bcl-2 (64), as well as the activation of PI3K and the downstream substrate, protein kinase B (Akt/PKB) (65, 66). In addition, 17ß-estradiol induction of Bcl-2 mRNA has been shown to require IGF-I receptor activation (67); ER is abundantly coexpressed with IGF-I receptor in female rat brain (68). Because both 17ß-estradiol and ER are needed, presumably ligand-dependent transcription by the ER is needed. However, the requirement for IGF-I receptor activation is less clear; it is possible that the downstream signaling events from the IGF-I receptor, i.e., MAPK, and Akt kinase activation are needed for Bcl-2 induction. Indeed, in PC12 cells, estrogen-mediated Akt/PKB activation induces Bcl-2 mRNA via the phosphorylation of the CREB and the subsequent binding of this protein to the CRE in the Bcl-2 promoter (69). Recently, Wu et al. (29) have reported that calcium influx mediated by rapid regulation of L-type calcium channels by 17ß-estradiol is critical in initiating a src/ERK signaling cascade that also results in enhanced Bcl-2 expression in cultured rat hippocampal neurons. Mendez et al. (70) have proposed a hypothetical model for IGF-ER interaction in the brain that is relevant to neuroprotection. On activation of the IGF-IR by IGF, the ER associates with the receptor at the plasma membrane and activates PI3K via interaction with the p85 subunit of PI3K (70). Regulation of the downstream Akt/PKB target, glycogen synthase 3ß (GSK-3ß), by 17ß-estradiol results in a decrease in phosphorylated Tau protein in the rat hippocampus, providing protection against Alzheimer’s disease (71). However, it is not clear whether the dynamics of this model is similar with ligand-free ER or ligand-bound ER or how the activation of GSK-3ß decreases Tau protein.

Rapid activation of both PKB and MAPK, and to a lesser extent PKC, by estrogens has also been shown to be important in neuroprotection in a number of neuronal cell types and against different types of insults. In the rat retina, rapid activation of the PI3K pathway is important in estrogen-mediated protection against light-induced photoreceptor degeneration (72). 17ß-Estradiol also rapidly activates p44/p42 MAPK in cortical slices (73, 74) and in hippocampal neurites (75). In the rat hippocampus, rapid induction of MAPK by estrogens protects against quinoline-induced toxicity (76, 77), whereas in hippocampal primary culture as well as in SK-N-MC neuroblastoma cells, phosphorylation of the MAPK ERK 1/2 by estrogens leads to enhanced secretion of ß-amyloid precursor protein (sAPP) (78). In primary rat cortical neurons, application of 17ß-estradiol or a membrane-limited E2-BSA conjugate rapidly stimulated the release of sAPP, whereas PKC inhibitors, added in addition to estrogens, block this effect (79). This secreted product decreases the generation of the ß-amyloid protein itself, which is a dominant neurotoxic component in the plaques of Alzheimer’s patients. Exogenous 17ß-estradiol treatment did not confer neuroprotection against MCAO in PKC knockout mice, suggesting that this PKC isoform mediates neuroprotection by 17ß-estradiol against ischemia (80). Although these studies show that both 17ß-estradiol-activated MAPK and Akt pathways appear to be necessary for neuroprotection, alone they are not sufficient. Although Bcl-2 transcription appears dependent on rapid signaling pathways, not all genes involved in neuroprotection may be so dependent, and not all signal transduction cascades may affect transcription. In addition, the relative contribution of membrane-initiated vs. genomic actions in neuroprotection is unclear.

3. Effects of estrogen on dendritic spines involve both nongenomic and genomic actions.
17ß-Estradiol also increases the growth of dendritic spines, modulating synaptic plasticity, in several hormone-responsive brain areas such as the VMH (81), hippocampus, and amygdala (82) via both genomic and membrane-initiated mechanisms. Estrogen-mediated development of spines in rat hippocampal cultures is dependent on rapid NMDA receptor activation (83) via c-src, which in turn may be activated by 17ß-estradiol binding a putative membrane ER (4). Similar to the scenario in midbrain dopaminergic neurons, PKA activation and phospho-CREB (pCREB) increases are also required for 17ß-estradiol-induced spine increase in rat hippocampal primary cultures (84, 85). Estrogen-induced pCREB in the rat hippocampus, which occurs as early as 1 h after addition of 17ß-estradiol, requires calmodulin kinase (CaMK)II and MAPK signaling but is less dependent on PKB (86). However, the longer time frame of inhibitor addition in this study does not allow conclusive confirmation of a rapid or membrane-limited effect of estrogen on these pathways, although it is highly probable, based on the signaling pathways activated. In female mice, administration of 17ß-estradiol for 1 h increased pCREB immunoreactivity in the VMH and medial preoptic area, whereas in male mice, pCREB was increased in the CA1 field of the hippocampus and in the medial septum (87). In cultured rat hippocampal neurons, 17ß-estradiol administration increased MAPK-dependent CREB activation in as little as 3 min (88 ). In differentiated NG108–15 neuroblastoma cells, Akt/PKB activation by 17ß-estradiol increased protein synthesis of PSD-95, a major component of spines, through phosphorylation of the elongation factor, 4E-BP1; this is clearly a nongenomic effect because PSD-95 mRNA does not appear to be regulated by estrogen in this system. Estrogen-induced stabilization and maintenance of spines may also depend on the decrease of GABA inhibition exerted by interneurons on CA1 pyramidal neurons (4, 89) via decrease in glutamate decarboxylase content in interneurons by 17ß-estradiol. Such a GABA decrease could also increase BDNF synthesis by solely genomic mechanisms such as transcriptional up-regulation of BDNF mRNA via a noncanonical ERE (20) in hippocampal pyramidal cells. A more controversial model for 17ß-estradiol-mediated increase in BDNF levels proposes that membrane ER signaling results in BDNF translation in giant mossy fiber boutons, which in turn synapse on CA3 pyramidal neurons (90, 91). In differentiating midbrain dopaminergic neurons, 17ß-estradiol rapidly (within 30 min) activates the PI3K/Akt signaling cascade (92); both this pathway and estrogen-induced PKA are important in neurite outgrowth in these neurons (93). Blockade of PKA or NMDA receptor or BDNF abrogates the effect of estrogens on spines in hippocampal neurons, suggesting that all of these pathways are needed for spine formation and stabilization. Alternatively, these could be part of a single pathway that leads to spine formation; hence, decreasing the levels of one molecule in the pathway could decrease the number of spines dramatically. Because all of these studies, with the exception of studies that involve pCREB and the PKA signaling pathway (86), use application of free 17ß-estradiol on longer time scales than is traditional for membrane-initiated actions, it is impossible to conclusively rule out transcriptional genomic actions of 17ß-estradiol on these molecules. However, the requirement for the activation of several kinases suggests that nongenomic pathways are important in dendritic spine growth, in concert with genomic effects of estrogens.

The data in these studies suggest that the PI3K/PKB pathway may be predominant in neuroprotection, whereas the PKA pathway could be more important in neurotrophism. Both signal transduction pathways, initiated by G protein-coupled membrane receptors, affect neuronal excitability in the hypothalamus and hippocampus.

	III. Membrane-Initiated Hormone Actions Potentiate Their Transcriptional Actions

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
The hypothesis that rapid signaling cascades initiated at the membrane by estrogens can lead to changes in transcription from ERE-containing genes is inspired by the fact that reproductive behavior or lordosis in female mammals is dependent on gene expression in discrete brain areas as well as kinase activation and changes in neuronal excitability by estrogens. Hence, this is a good model system to study the interaction between rapid and slow actions of estrogens.

A. Coupling of the membrane-initiated actions of estrogen to transcription: the consequence for reproductive behavior
The estrogen control of female reproductive behavior or lordosis (32) presents a scenario for the interplay between membrane-initiated signal transduction mechanisms and the classical genomic mode of transcription. In the VMH, estrogen imposes hormonal dependence on neurons that control descending efferents to the muscles in the lower back that allow for the arching of the back, a posture characteristic of lordosis (32, 94, 95). Genes transcriptionally regulated by estrogen that may play a role in approach and sex behaviors in rodents include the PR; the oxytocin receptor and its ligand, oxytocin; the preproenkephalin peptides; and the 1b-adrenergic and muscarinic receptors (96, 97).

In support of the idea that membrane-initiated actions play a role in rodent lordosis behavior, second messengers in kinase cascades can potentiate lordosis. For example, phorbol esters (PKC activators) infused into the midbrain central gray could facilitate lordosis (98), whereas PKC inhibitors could reduce lordosis behavior (95). Dibutryl cAMP and 8-bromo cAMP infused into the VMH could reduce the inhibition mediated by serotonin (99), suggesting that the coupling of the serotonin receptor to PKA inhibition has physiological relevance. Antagonists to protein kinases or to G protein-coupled receptors (GPCRs) reduce lordosis behavior, suggesting that activation of kinase cascades is permissive to lordosis. Infusion of PI3K or MAPK inhibitors into the VMH during estrogen priming of ovariectomized rats attenuates lordosis (100). In addition, IGF receptor antagonists block lordosis induction by estrogen in female rats (100, 101), whereas -opioid receptor agonists infused into the VMH, but not into the medial preoptic area, can increase lordosis in female rats already primed with estrogen and progesterone (102). In the VMH, estrogen increase in neuronal excitability is thought to be relevant to lordosis behavior in rodents (103, 104). Similar to the effects of estrogen on POMC and dopamine neurons (see Section II.C), in VMH slices, acute 17ß-estradiol administration (10 min) can potentiate the neuronal excitation caused by both histamine and NMDA, two neurotransmitters that act via GPCRs and ligand-gated ion channels, respectively (105). This suggests that 17ß-estradiol can influence two different transduction pathways to achieve a final effect, i.e., increase in depolarization of neurons. Because it appears that both genomic and membrane-initiated actions of estrogen are possibly important in lordosis behavior, we initially hypothesized that membrane-initiated actions of estrogen impacting signal transduction cascades could regulate gene transcription, also induced by estrogen (106).

B. Coupling of membrane-initiated actions of estrogens to transcription in nonreproductive functions
Apart from lordosis behavior, the above hypothesis also has roots in neuronal and non-neuronal cell culture studies which show that estrogen-induced gene transcription is indeed influenced by signaling molecules. In the human neuroblastoma cell line, SK-N-SH, a membrane-limited estrogen conjugate (E2-BSA) can induce the c-fos gene via MAPK kinase and MAPK, demonstrating a mechanism by which genes possessing non-ERE promoters might still be induced by estrogen (107). In the same cell line, E2-BSA could induce the neurotensin gene, an effect blocked by PKA inhibitors. E2-BSA alone did not induce transcription from the consensus ERE (108); presumably, there was no free 17ß-estradiol in this preparation. In the MCF-7 breast carcinoma cell line, PR mRNA is induced by both estrogen and PKA activators (109), whereas the estrogen up-regulation of cyclin D1 has been reported to be through the CRE (110). The levels of estrogen-regulated mitogenic proteins, cyclin D1 and c-myc, can be decreased by inhibition of MAPK and PI3K (111). Inhibitors to PI3K, dominant-negative Akt mutants, and a selective ErbB2 inhibitor can decrease the up-regulation of ER mRNA by estrogen (autoregulation) in MCF-7 cells (112, 113). In endothelial cells, inhibitors for PI3K or the use of dominant-negative constructs to PI3K also block estrogen-induced cyclo-oxygenase-2 (114). An inhibitor of PKC could decrease transcriptional potentiation by E2-BSA of ER-mediated transcription from a consensus ERE in MCF-7 cells (115). These studies show that a number of estrogen-induced genes or estrogen-induced promoters can be regulated by second messenger pathways.

C. The coupling of membrane-initiated actions to transcription is studied in vitro by using a two-pulse transcriptional paradigm in cell culture
A "proof of concept" result for the hypothesis that transcription can be influenced by kinase cascades was first demonstrated in a neuroblastoma cell line, SK-N-BE(2)C, which is devoid of endogenous ER or ERß isoforms and has been used previously as a model for estrogen action in the brain upon the transient or stable transfection with ER (116, 117). This cell line has been studied extensively for the effects of both µ- and -opioid (118, 119) receptors, demonstrating that GPCR-mediated receptor effects are possible in this cell line.

The use of a novel transcriptional two-pulse hormonal administration paradigm in cell culture as a tool to study transcriptional potentiation was inspired by earlier studies where discontinuous, pulsatile estrogen administration is effective in achieving cellular division and behavior. In a pioneering study, Harris and Gorski (120) showed that a discontinuous schedule of estrogen administration can induce uterine cell division. Discontinuous schedules of two 1-h pulses of estradiol benzoate (EB) with an interpulse interval of 6 h are as effective in inducing both lordosis and the PR in the VMH of female rats 48 h after the first EB administration as a continuous administration of EB for 48 h (121). It is possible that in the first pulse, estrogen may initiate kinase cascades (rapid membrane-initiated actions) that may potentiate genomic transcriptional actions in the second pulse.

To separate the membrane-initiated actions of estrogen from the genomic slower actions of estrogen in this cell line, estrogens added in the first 20-min pulse can initiate signaling from the membrane, whereas 17ß-estradiol in the second 2-h pulse promotes genomic action on a consensus ERE-linked reporter in neuroblastoma cells. We cannot completely rule out membrane actions in the second pulse because free 17ß-estradiol also has the ability to act at membrane sites. To initiate signaling cascades solely at the membrane, a membrane-impermeable estrogen conjugate, E2-BSA, filtered (42) to remove free 17ß-estradiol, was applied in the first pulse. Despite controversy over the possible leaching of free 17ß-estradiol into cells (122) (see Section II.B), we have shown that continuous administration of E2-BSA for 24 h to transfected SK-N-BE(2)C cells does not increase transcription above basal levels from a consensus ERE-based reporter gene (123). In addition, Taguchi et al. (124) showed that E2-BSA possessed the ability to bind to the ER at membrane sites.

In these initial experiments, nanomolar concentrations of E2-BSA can potentiate transcription from a transfected consensus ERE-driven luciferase gene in neuroblastoma cells induced by a second 2-h pulse of 10^–9 M 17ß-estradiol (E2-BSA-mediated transcriptional potentiation) (106) (Fig. 1). E2-BSA can hence substitute for 17ß-estradiol in the first pulse. A lower level of transcription (albeit higher than control) is seen with vehicle in the first pulse, followed by the addition of 17ß-estradiol in the second longer pulse. A "reverse" pulse paradigm where 17ß-estradiol is given in the first pulse followed by E2-BSA in the second pulse does not potentiate transcription, demonstrating that preliminary membrane effects are both sufficient and essential for later transcriptional effects (106). To pinpoint specific kinases that play a role in E2-BSA transcriptional potentiation, specific inhibitors in the first pulse along with E2-BSA or in the second pulse with 17ß-estradiol were used. This strategy demonstrated that PKA and PKC may play a role in E2-BSA-mediated transcriptional potentiation because the PKC and PKA inhibitors chelerythrine and Sp-cAMPs in the first pulse reduce the level of transcription but do not interfere with potentiation when added in the second pulse. On the other hand, the calcium chelator BAPTA-AM blocks E2-BSA-mediated transcriptional potentiation when added either in the first pulse along with E2-BSA or in the second pulse with 17ß-estradiol, suggesting that calcium is important in both the membrane-initiated and genomic actions of estrogen (106). Using selective inhibitors, we have also provided evidence (125) that the entry of calcium via N-type voltage-gated calcium channels is important in both genomic and membrane-initiated actions of estrogens, whereas the ryanodine receptor, which modulates intracellular stores of calcium, is important in coupling membrane-initiated actions of estrogen initiated at the membrane to transcription.

View larger version (18K): [in this window] [in a new window]   FIG. 1. E2-BSA (10–9 M) potentiates transcription mediated by 17ß-estradiol (10–9 M) in the second pulse. SK-N-BE (2 )C neuroblastoma cells are cotransfected using Effectene reagent (QIAGEN Inc., Valencia, CA) with pGL2-TATA-Inr-Luc (reporter; 3xereluc) (200 ng), pSG-hER (80 ng), pSV-ßgal (80 ng) (Promega, Madison, WI), and pBSKII+ (Stratagene, La Jolla, CA) to a total of 400 ng/well. Twenty-four hours after transfection, a two-pulse regimen is initiated with both pulses separated by a hormone-free 4-h interval. Results (n = at least 4 per treatment group in duplicate experiments) are analyzed using one-way ANOVA followed by Student-Newman-Keuls post hoc test (Prism Software Corp., Irvine, CA) to compare normalized luciferase activity between treatment groups (106 ). E2-BSA (10–9 M) was given for 20 min in the first pulse, and 17ß-estradiol (10–9 M) was given in the second pulse for a duration of 2 h (106 ). Test of "potentiation" hypothesis: E2-BSA given in the first pulse was followed by 17ß-estradiol in the second pulse (far right bar). Results (n = 4 per treatment group from replicate experiments) are represented as mean + SEM. *, P < 0.05 compared with all other treatment groups; #, P < 0.05 compared with the vehicle group (106 ).

	IV. The Signal Transduction Cascades Initiated by Estrogens at the Membrane May Occur in Parallel or in Series

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

A. Membrane-initiated actions of estrogens may also occur in parallel or in series in the MCF-7 breast carcinoma cell line
Based on our data, it is probable that a guiding principle of estrogen action is to activate several membrane-initiated kinase cascades in a single neuronal cell type. Is this borne out by data in other cell types? Addition of estrogens results in proliferation in breast cancer, typified by the response in the MCF-7 cell line. MCF-7 cells also respond to estrogens with rapid activation of the MAPK/ERK pathway (130) via incompletely understood mechanisms. A number of studies point to different modes of MAPK activation by estrogens. For example, addition of an antibody to the heregulin receptor or depletion of heregulin from the conditioned media of MCF-7 attenuates estrogen-mediated ERK activation, suggesting that estrogen induction of growth factors plays an autocrine/paracrine role in ERK activation and hence, cell proliferation (131). In addition, inhibition of PKC and Ras signaling also blocks MAPK activation (131), implying an upstream role for these molecules. In tamoxifen-resistant cells with equivalent levels of ER as wild-type MCF-7 cells, the ERK activation appears to be mediated via TGF and increased levels of epidermal growth factor (EGF) receptors (132). Liganded ER can also activate MAPK in MCF-7 cells via a rise in calcium from intracellular calcium stores (74). Because growth factor signaling to MAPK pathways involves the adaptor protein Shc, a similar pathway was proposed and demonstrated for estrogenic activation of MAPK. Shc binding to liganded ER in the cytosol of breast cancer cells leads to association with the IGF receptor and subsequent activation of MAPK via the Grb2-Sos-Ras-Raf pathway (130, 133). Physical association with the p85 subunit of PI3K (130, 133, 134) as well as with Src (134) has also been demonstrated with the ER in MCF-7 cells; inhibition of PI3K activity using the inhibitor LY294002 abrogates estrogen-induced Akt/PKB and Src activation (134). The transcriptional up-regulation of macrophage inhibitory cytokine-1 by PKB prolongs estrogen-induced phosphorylation of ERK (135). Membrane-limited E2-BSA activates PKC rapidly in MCF-7, independently of ER or ERß, and recurrent tumors have higher levels of PKC (136). Furthermore, in MCF-7 cells, estrogen-induced PI3K activates the atypical PKC, which in turn activates the src/Ras/MAPK signaling cascade (137). Other PKC isoforms have also been associated with proliferation in MCF-7 cells. For example, overexpression of PKC allows tamoxifen-sensitive MCF-7 lines to acquire tamoxifen resistance (138). In contrast, stable transfection of PKC is accompanied by ER down-regulation and elevated basal AP-1 activity (139). An early study also reported that transfection of this cell line with PKA or treatment with 8-bromo-cAMP could increase the agonist activity of tamoxifen on a consensus ERE-based reporter construct to about 20–75% of the level obtained with 17ß-estradiol (140). Conversely, melanin inhibited ER transactivation by inhibiting forskolin-induced and estrogen-induced cAMP increases (141).

How can these data be resolved? First, these data imply that several kinases, i.e., MAPK, PI3K, PKA, and PKC, can be activated rapidly by estrogens, possibly via the ER, in MCF-7 cells. This is similar to the data seen with neuroblastoma cells (see Section III). Hence, it is probable that several kinases coordinately influence gene regulation in several target cell types to achieve a final phenotype.

Finally, how proliferation of the MCF-7 cell line, i.e., the final phenotype, is affected by the membrane-initiated activation of kinases is still unclear. Long-term estrogen-deprived MCF-7 cells (the LTED lines), which mimic tamoxifen resistance, show hyperphosphorylation of ERK, whereas both MAPK/ERK kinase (MEK) and PI3K inhibitors could decrease proliferation (142). However, PI3K and MEK inhibitors do not decrease c-myc and c-fos expression although hyperphosphorylated retinoblastoma and cyclin D1 proteins decrease, suggesting that rapid signaling cascades are permissive to cell cycle progression but not to mitogenesis via the immediate early genes myc and fos (143, 144). These studies strongly imply that estrogen-regulated rapid signal transduction cascades, that is, membrane-initiated actions are relevant to the final outcome in breast cancer cells, i.e., proliferation.

We note that for full potentiation of a transcriptional effect of estrogens by a membrane-initiated action, the data using kinase activators suggest that more than one signal transduction system is sufficient, but the data with kinase inhibitors also imply that more than one signal transduction system is necessary. How can this be? If a particular kinase cascade is "sufficient," how can another one be "necessary"? If several signal transduction systems in parallel maintain a basal state of transcriptional readiness, ablating any one of them lowers that basal state such that a genomic action of 17ß-estradiol is ineffective in potentiating transcription. Conversely, if any one of the transduction systems involved is highly activated, it can substitute for the membrane-initiated action of 17ß-estradiol in potentiating the later, genomic action. This idea, therefore, postulates the existence of a transcription threshold for genes.

	V. Convergence of the Membrane-Initiated Actions of Hormones to Transcription Involves Protein-Protein Interactions, Protein Translocation, and Protein Phosphorylation

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
Most early studies have treated membrane-initiated actions as disparate from the genomic actions of hormones encompassing and culminating at kinase activation or calcium increases, both of which have cytosolic substrates (34, 145, 146). Other studies, have examined the effect of growth factor receptor ligands, notably IGF and EGF, on ligand-independent activation of the ER (Refs. 130 and 147, 148, 149 and references therein). Such activation is particularly important in hormone therapy-resistant cancers; combined targeting to both the ER and growth factor pathways, therefore, has therapeutic significance (150). Another subset of studies has involved the role of the classical ERs in rapid, membrane-initiated actions, particularly in breast cancer (130, 133, 151, 152) and cardiovascular physiology (114, 153, 154, 155, 156, 157, 158, 159). However, how the ligand at the membrane can regulate transcription in the nucleus is unclear if the same ligand, i.e., 17ß-estradiol has the ability to signal at both membrane and nucleus and if membrane-initiated rapid actions by the ligand influence transcription on EREs by ligand-dependent nuclear receptor action. For estrogen-regulated genes, one can mechanistically envision such coupling from membrane to nucleus at various intersecting levels such as 1) stability of the ER (protein-protein interactions); 2) nuclear translocation of the ER; and 3) modification of proteins important for transcription. Some of these aspects, especially for genes that do not contain EREs in their promoters, have been recently reviewed by Bjornstrom and Sjoberg (160).

A. Coupling to transcription via stabilization or dimerization of molecules relevant to transcription
Heat shock proteins (hsp 70/90) are important for proper folding, maturation, and translocation of the steroid hormone receptors (13, 161). There are a few studies that report estrogenic regulation of the heat shock chaperones. 17ß-Estradiol and bisphenol A act via the ER in uterine tissue to increase hsp90 and hsp72 mRNA via PKC (162, 163). In the CNS, estrogens and androgens protect neurons from ß-amyloid toxicity by increasing the levels of hsp70; microinjection of hsp70 blocked the toxicity of added intracellular ß-amyloid peptide (iamyloidß1–42) (164), suggesting a role in neuroprotection. However, it is not clear whether 17ß-estradiol can regulate these proteins to enhance transcription driven by the ER.

The stability of the ER has important consequences for function. The autologous down-regulation of ER and subsequent proteolysis by the 26S proteasome is thought to be required for efficient transactivation from the consensus ERE (165). In MCF-7 cells, PKC can enhance proteasome degradation of ligand bound and ligand free ER, whereas PKA, PI3K, and MAPK impede degradation. On the other hand, in pituitary tumor cells, PKA increases transactivation mediated by 17ß-estradiol-liganded ER from a consensus ERE while protecting the ER from estrogen-induced degradation. This suggests that 17ß-estradiol increases transcription, partly by autologous down-regulation of the ER, whereas PKA could increase transcription by stabilizing the ER (166). The relevance of ER degradation is, however, controversial. For example, proteolysis is not required for ER-mediated induction of the prolactin gene or for proliferation in anterior pituitary lactotrope cells, demonstrating different requirements for the stability of ER, depending on the ERE (167).

B. Coupling to transcription via the physical movement of molecules into the nucleus
17ß-Estradiol can also regulate the translocation of the ER, possibly via kinase cascades. PKC activation, via GSK-3 results in nuclear translocation of the ER with a concomitant increase in transcription (168). In endometrial adenocarcinoma, estrogenic up-regulation of p38MAPK, in turn, increases ER phosphorylation on Thr³¹¹; mutation of this site inhibits nuclear localization as well as the interaction with coactivators (169). In human vascular smooth muscle cells, the MAPK phosphorylation of the ER on Ser¹¹⁸ and nuclear translocation could also be ligand-independent (170).

C. Coupling to transcription via phosphorylation of molecules relevant to transcription
Our inhibitor studies strongly suggest that kinases such as PKA, PKC, PI3K, and MEK are activated during rapid exposure to either membrane-impermeable estrogen (E2-BSA) or T₃ (see Section VII.A) at the membrane in neuroblastoma cells. Phosphorylation of molecules relevant to transcription, for example, the ER itself, by these kinases could provide a simple coupling mechanism between events at the membrane and transcriptional events in the nucleus. Endogenous ER has a basal level of phosphorylation that is enhanced by physiological concentrations of estrogen at amino acid Tyr⁵³⁷ and at serines at positions 104, 106, 118, 154, and 167 in the activation function (AF)-1 N-terminal domain (171, 172). Estrogen-induced phosphorylation occurs via kinase activation. PKA can phosphorylate the ER at serines 236, 305, 338, and S518 (173). In contrast, PKC phosphorylation of the ER may be at Ser¹¹⁸ in the AF-1 domain (174). PI3K-activated PKB has been shown to phosphorylate ER on Ser¹⁶⁷ in vitro (175, 176). MAPK phosphorylation of ER occurs on Ser¹¹⁸ in vivo as well as in vitro (174, 177, 178). How phosphorylation affects ER function is less clear. PKA phosphorylation at Ser²³⁶ in the DNA binding domain increases dimerization of the ER (179). A second PKA phosphorylation site at Ser³⁰⁵ in the ER blocks acetylation of the ER at the proximal Lys³⁰³ site, enhancing transcriptional ability (180). Mutations at the Ser¹¹⁸ and Ser¹⁶⁷ sites decrease transcription in fibroblast and epithelial cells, suggesting that these sites were critical for transcription (174, 178, 181, 182, 183, 184). In vivo, MAPK-mediated phosphorylation (185) of Ser¹⁶⁷ occurs via the ribosomal S6 kinase (184), whereas in vitro, Ser¹⁶⁷ is phosphorylated by both Akt and casein kinase II (113, 176, 186). Phosphorylation on Ser³⁰⁵ also increases transcription from the cyclin D1 gene in breast cancer (187). This is possibly because phosphorylation of nuclear receptors appears to enhance interaction with coactivators (188). In support of this idea, Ser¹¹⁸ phosphorylation in the AF-1 domain facilitates the interaction of the ERß with the steroid receptor coactivator 1 (SRC-1), hence increasing the transcription rate (189). Table 1 presents a list of ER phosphorylation sites and their possible impact on ER function. The stoichiometry and the sites of ER phosphorylation, however, vary depending on the cell type, presumably due to the tissue-specific differences in kinase and phosphatase activities (177). Translocation of the receptor may also be affected by kinase-induced phosphorylation. Blocking MAPK causes loss of phosphorylation on Ser²⁹⁴ and nuclear accumulation of PR-B and prevents ligand-dependent degradation of the receptor. In turn, this decreases transcriptional activation (190).

View this table: [in this window] [in a new window]   TABLE 1. Phosphorylation sites on ER and the impact of phosphorylation on function

As is evident, none of these processes need to be mutually exclusive. Extracellular signals (exemplified in our paradigm by E2-BSA) may initially converge on a molecule, for example, the ER, using multiple molecular mechanisms such as translocation, phosphorylation, and stabilization. This initial convergence may lead to molecular activation, which may, in turn, impact divergent molecular processes within cells to achieve a final differentiated response to the signal. Conceivably, in neurons, the "activated" ER could enhance transcription, excitability due to effects on ion channels, and neurotransmitter release. Alternatively, the extracellular signal may activate several molecules, all of which converge on a single molecular process. For example, in our paradigm, if phosphorylation activated downstream transcription factors such as CREB binding protein (CBP), coactivators such as SRC-1 and SRC-3, and the ER, transcription would be greatly enhanced, and amplification of the original extracellular signal would be achieved.

	VI. Initiation of Membrane-Initiated Actions at the Membrane Involves Both Classical "Nuclear Receptors" and Other Proteins

Top Abstract I. Introduction II. Estrogens Signal via... III. Membrane-Initiated Hormone... IV. The Signal Transduction... V. Convergence of the... VI. Initiation of Membrane... VII. Summary and Future... References

 
A. The membrane estrogen receptor could be a novel protein in some cell types
The elucidation of the mER has generated enormous interest and effort (34, 146, 197, 198, 199) because knowledge of a mER can lead to dissection of membrane-initiated signal transduction pathways activated by estrogens (200). Any protein that is a candidate mER should bind 17ß-estradiol with specificity and saturability and be present at the plasma membrane of the cell to initiate rapid membrane-initiated action. Over the years, there have been conflicting reports on the mER in different cell types, predominantly based on the application of ER antagonists, such as ICI and immunocytochemistry. ICI, a pure ER and ERß antagonist, was unable to block the ability of E2-BSA to activate PKC rapidly in chondrocytes (201) or MAPK in rat hippocampus (77), suggesting a novel membrane receptor for estrogen. In the neocortex, a novel, high-affinity mER, called ER-X, which supports MAPK phosphorylation by estrogens, has been reported (202). A novel G_q protein-coupled mER agonist, STX, can attenuate the weight gain seen in hypoestrogenic animals. This is similar to the attenuation seen with 17ß-estradiol and occurs in double (ER^–/–, ERß^–/–) knockout mice (203), suggesting that a novel protein may be present at the membrane in the CNS. The identity of this receptor and its location at the membrane remains to be elucidated. This group has also previously used an ER/ß antagonist, ICI 164,384, which blocked the ability of 17ß-estradiol to rapidly uncouple the µ-opioid receptor from an inwardly rectifying K⁺ channel (204); however, in this study, ICI 164,384 could not block the same response induced by STX (203). This suggests that the depolarization induced by 17ß-estradiol and by STX has differential requirements for the presence of the classical ER. In pancreatic ß-cells, binding of the membrane limited E2-HRP to the membrane can be competed by dopamine and norepinephrine, suggesting a common catecholaminergic receptor. However, the ability of norepinephrine to induce Ca²⁺ oscillations in these cells was unchanged by the addition of 100 nM 17ß-estradiol, which should saturate the common membrane receptor, if any (33, 205). Hence, the functional relevance of such a receptor is unclear. A novel mER has been suggested on the basis of competition binding in the GT1–7 cell line, wherein the binding of E2-HRP conjugate was competed by 17ß-estradiol but not by ICI 182,780 or 4-hydroxytamoxifen (206). The rapid reduction by 17ß-estradiol of the Ca²⁺ transients induced by acetylcholine in these cells may involve a protein kinase G phosphorylation of the inositol triphosphate receptor that modifies Ca²⁺ release from the endoplasmic reticulum (207). However, the GT1–7 cell line, has been shown by others (40) to possess the classical ER at membrane sites. It is unclear how these conflicting data can be resolved. Is it possible that membrane-initiated signaling by 17ß-estradiol in the same cell line may involve more than one type of mER?

In fish, where membrane-initiated actions of progesterone are required for oocyte maturation, the membrane PR is a novel protein coupling to G_i/G_o proteins (208, 209). Recently, an orphan GPCR, GPR30, has been identified as a membrane ER in SKBR3 breast carcinoma cells based on high-affinity binding, membrane localization, and modest levels of adenylyl cyclase activation upon estrogen addition, showing that the GPR30 can couple to G_s (210). In SKBR3 cells that express no classical ER, GPR30 is absolutely required for 17ß-estradiol activation of adenylyl cyclase. This is the first report of a mER structurally unrelated to the classical ER. In addition, binding of ICI compounds, 17ß-estradiol and 17-estradiol, occurs with equal affinity to the GPR30 (210). On the other hand, Revankar et al. (211) have reported that the GPR30 protein is uniquely localized to the endoplasmic reticulum and may serve as an intracellular ER that increases calcium in response to 17ß-estradiol. However, they have also shown that GPR30-transfected SKBR3 cells can increase PIP2 concentrations, supporting the idea that GPR30 can increase nongenomic signaling in response to 17ß-estradiol (211). However, another laboratory has reported that ERK activation and cAMP accumulation in response to 17ß-estradiol is abrogated in ER^–/GPR30⁺ SKBR3 cells. This may be due to differences in the parameters (fluorescence localization vs. enzymatic activity) (212) measured in these two studies. The localization of GPR30 and its functional relevance in other cell types, perhaps using a selective, recently synthesized agonist, G-1 (213), remains to be elucidated.

B. The membrane estrogen receptor is the classical ER in other cell types
Several reports support the notion that a classical ER may be tethered to the membrane (reviewed in Ref. 154). The E2-BSA activation of MAPK in Rat-2 fibroblasts (214) and endothelial cells (215) and the interaction between ER and G_i in IC-21 macrophages (216) can be blocked by ICI compounds, which antagonize both ER and ERß. ICI 182,780 can also block CaMKII activation in the hippocampus by EB (127). In astrocytes, E2-BSA can cause rapid calcium flux that can be blocked by ICI 182,780 (217). The existence of ER at the membrane has also been inferred from functional signaling studies, combined with inhibitor studies. In immortalized GT-1 cells, picomolar concentrations of 17ß-estradiol could inhibit cAMP levels, whereas nanomolar concentrations of 17ß-estradiol could augment cAMP levels. In addition, these effects could be blocked by ICI 182,780 and pertussis toxin, implying that the ER was coupled to a G_i protein in GT-1 cells (40). Qiu et al. (49) have, on the basis of phospholipase C signaling by 17ß-estradiol in POMC and dopamine neurons, proposed a model whereby a G_q coupled receptor at the plasma membrane may rapidly activate signaling cascades in series (see Section II.C). Signaling in these cells was also blocked by ICI 182,780 (48, 49). Functionally, the lack of ER or ERß in double ER knockout mice abrogated the rapid ERK phosphorylation by estrogens in the medial preoptic nucleus (218), suggesting that the ER is required for rapid, nongenomic actions. Also, endothelial cells from double ER knockout mice, compared with wild-type mice, did not support PI3K, cAMP, and ERK phosphorylation or bind 17ß-estradiol (212). These studies demonstrate that classical ERs that mediate genomic actions may also mediate nongenomic actions, but they do not mean that the ER is at the plasma membrane. Using immunocytochemistry, numerous studies have shown that ER is at the plasma membrane. In Chinese hamster ovary (CHO) cells, overexpressed ER gives rise to a single transcript with functional protein expression in both nuclear and membrane fractions (219). Endothelial cells, a major estrogen target cell with relevance to cardiovascular protection, from the ER^–/–,ERß^–/– mouse do not show membrane ER (220). Anti-ER antibodies on fetal rat hippocampal neurons reveal abundant staining at the neurites by confocal microscopy (221). In rat pituitary tumor cells, antisense oligodeoxynucleotide to ER mRNA reduces the staining obtained with specific, high-affinity antibodies at the membrane of fixed, nonpermeabilized cells (222). In addition, plasma membrane fractions from MCF-7 cells that express the ER show that the protein at the membrane that binds 17ß-estradiol is the classical ER (212).

How do nuclear receptors attach to the plasma membrane despite a lack of standard transmembrane structure (215) and hydrophobicity? ER can physically interact with membrane proteins, allowing anchorage to the membrane or to the vicinity of the membrane. Furthermore, the ER may also be present in specialized plasma membrane structures such as caveolae and lipid rafts that possess several signaling enzymes to form a organized signaling module (157) coupled to G_i (223). Indeed, data obtained using E2-BSA do not rule out the possibility that this compound can enter and bind ER in caveolae. Western blotting and mass spectroscopy show caveolin-1 in isolated oligodendrocyte plasma membrane colocalizing with an ERß-like molecule (224). In endothelial cells, ER associates with the p85 subunit of PI3K in caveolae, mediating PI3K/Akt activation of nitric oxide synthase (NOS) (155, 158). Similarly, ERß in caveolae activates NOS (156), possibly via the rapid activation of MAPK (225). Localization of ER to the membrane appears to be dependent on Ser⁵²²; mutation of this residue abrogates membrane localization in caveolae in CHO cells and adenylyl cyclase activation but supports transcription (226), suggesting that the amino acid residues required for membrane-initiated action and genomic action may be different. In breast cancer and endothelial cells, ER exists at the membrane as dimers, in association with caveolin-1 in caveolae (227), and can support adenylyl cyclase and PI3K signaling. NOS activation in endothelial cells was, however, not dependent on dimerization because an ER deficient in dimerization (ERL511R) could still activate NOS (159). Moreover, despite presence at the plasma membrane, nuclear localization signal mutants could not activate ERK, whereas DNA binding mutants of the ER could not activate ligand-dependent PI3K binding to ER (159). Hence, the functional relevance of membrane localization to signaling in endothelial cells is not clear. In addition, palmitoylation supports ER attachment to the membrane and interaction with caveolin in HeLa and HepG2 cells, whereas addition of ligand decreases interaction of the receptor with caveolin (228), suggesting a classical desensitization mechanism. In MCF-7 cells, ER can physically bind the p85 subunit of PI3K at the membrane (229, 230) as well as the Shc protein via the AF-1 domain and can drive the formation of a Shc-Grb2-SOS signaling complex (130). In addition, a novel scaffold protein called MNAR (modulator of nongenomic action of ER) can interact with c-src using its PXXP motif while interacting with the ER using the LXXXL motifs, demonstrating an "adapter" function (231, 232). Using an adapter protein to anchor at the membrane does not, however, make the ER an intrinsic membrane protein. However, work from Cheryl Watson’s group (151) has implicated ER at the membrane in MCF-7 cells. Using immunopanning with specific ER antibodies, MCF-7 cells were separated by flow cytometry into high mER and low mER populations (151). Immunopanning suggests that the antibody binding site on the ER is available to the antibody at the surface of cells. It is difficult to reconcile this data to the previously stated data on MCF-7 cells, which show tethering of ER to the inner plasma membrane. One possibility is that a very small proportion of cells may have small amounts of ER at the surface, whereas a greater number of cells in a given cell type may have ER tethered to the membrane, although the basis of this differential distribution would be difficult to rationalize. Apart from the idea that a small fraction of endogenous cellular ER is permanently anchored to the membrane, a small pool of nuclear ER may dynamically shuttle to the membrane. This possibility has been demonstrated by an ER artificially tethered to the plasma membrane with a rhodopsin fusion protein that can shuttle to the nucleus to increase transcription in 293 cells (233). In MDA-MB-231 breast cells stably transfected with either wild-type ER or myristoylated ER lacking the nuclear localization signal, the myristoylated ER shows membrane localization but is not down-regulated by 17ß-estradiol, as is the case for wild-type ER. In addition, compared with the WT ER, the myristoylated protein cannot up-regulate the transcription of estrogen reporter genes on exposure to 17ß-estradiol (234).

Functionally, the presence of ER at the membrane may have physiological consequences that need to be studied in more detail. In MCF-7 variants sorted for higher ER localization to the membrane, estrogenic induction of adenylyl cyclase is correlated with lower proliferation (151). In Xenopus laevis oocytes, MNAR contributes to meiotic arrest because reduction of MNAR mRNA increased liganded androgen receptor-driven MAPK activation and oocyte maturation (235). In the CNS, differentiation of midbrain dopaminergic neurons appears to be due to membrane-attached ER in astrocytes (236).

C. Do different cell types have different receptors?
The wealth of studies demonstrating novel proteins as the mER or the classical nuclear receptor, the ER, as a candidate mER suggest that different cell types may possess different mERs. None of the novel proteins that are mER candidates (i.e., ER-X or GPR30) show any homology to the novel membrane PR, cloned from sea trout ovaries (209), indicating that, unlike the classical nuclear receptor superfamily, there may not be a parallel "membrane receptor superfamily". It is also possible that different molecules may "moonlight" as the mER either temporally or for distinct functional roles in a single cell type. Additionally, the domains required for ER function at the membrane could be cell specific, depending on the molecule that the ER may be tethered to in that cell type. These ideas about the nature of membrane receptors for hormones remain to be clarified.

	VII. Summary and Future Directions

Top Abstract I. Introduction II. Estrogens Signal via... II