The estrogens regulate a plethora of physiological processes in mammals, including reproduction, cardiovascular protection, bone integrity, cellular homeostasis, and behavior (Simpson et al., 2005). Estrogens act by binding to high-affinity receptors on target cells. Since the discovery of estrogen receptors (ERs), a variety of mechanisms underlying estrogen's diverse physiological actions have been identified.
Traditionally, estrogens, like other steroid hormones, act as nuclear transcription factors, by modulating target genes through complex interactions with coactivator or corepressor proteins, histone-modifying enzymes, and proteins comprising basal transcriptional machinery. Over the years, investigations have provided numerous critical aspects of estrogen signaling at the transcriptional level (Nilsson et al., 2001). In addition to these classic genomic mechanism of estrogen action mediated by intracellular receptors (ERα and ERβ), there is now evidence that estrogen could exert rapid, nongenomic steroid actions presumably initiated at the cell surface. Initially these rapid estrogen actions generated at the plasma membrane were identified almost two decades ago (Pietras and Szego, 1977). This membrane-associated estrogen-binding activity was not the focus of many research groups, and so research on this aspect lagged behind. Emerging data overwhelmingly support the notion that 17β-estradiol, a potent mitogen, generates a series of extranuclear or nongenomic responses. For instance, 17β-estradiol rapidly increases intracellular second messengers, such as calcium and cAMP, and activates mitogen-activated protein kinase (MAPK) and phospholipase C (Morley et al., 1992; Gu and Moss, 1996; Migliaccio et al., 1996; Le Mallay et al., 1997; Ho and Liao, 2002). These intracellular responses are important in physiological processes such as cardiovascular protection, bone preservation, cancer cell proliferation, neuroprotection, spermatogenesis, and keratinocyte proliferation.
Estrogen mediates its cellular actions through ERα and ERβ. The ERs are members of the super-family of ligand-dependent transcriptional activators. They consist of a C-terminal steroid ligand-binding domain, a centrally located DNA-binding domain, and an N-terminal domain that is not as well characterized as the other two domains. They also contain two autonomous transcriptional activation function (AF) domains: AF-1, located in the N terminus, and AF-2, located in the ligand-binding domain. A detailed review of the structure and function of classical steroid receptors has been published elsewhere (Nilsson et al., 2001).
In addition to these classical ERs, a novel transmembrane intracellular estrogen-binding protein, G-protein-coupled receptor (GPR-30) has been identified (Filardo, 2002; Revankar et al., 2005; Thomas et al., 2005) that has been demonstrated as a novel mediator of estrogen action. Discovery of this estrogen-binding receptor, a seven transmembrane domains containing protein, has generated lot of interest in identifying many unknown functions and mechanisms played by estrogen outside the nucleus. From this point, the potential biological effects of estrogen will considerably broaden. In view of both, the introduction of this transmembrane protein such as GPR-30 into the estrogen's action, and keeping in mind the previously provided nongenomic mechanisms of estrogen through classical receptors, in this review, we describe the various actions of estrogen in several physiological systems.
ESTROGEN'S NONGENOMIC SIGNALING
Steroid genomic actions generally involve the entry of a free steroid into a target cell by passive diffusion through the plasma membrane, following which the steroid binds with high affinity to its receptor. The binding of estrogen to its receptor triggers conformational changes (in the tertiary and quaternary structures) in the receptor, which in turn leads to the formation of an activate ligand receptor complex. This complex interacts with specific DNA sequences referred to as hormone response elements, which are located upstream of steroid-responsive genes. Such regulatory interactions result in the synthesis of specific mRNAs and, in turn, the synthesis of proteins responsible for the hormone's cellular effects (O'Lone et al., 2004). In general, these genomic actions induce biological responses more slowly than the nongenomic actions.
In addition to its nuclear responses, estrogen (steroid) generates rapid cellular responses which cannot be explained by its nuclear action alone and thus suggest the existence of alternative mechanisms involving short-term rapid cytoplasmic signaling. These responses possibly explained by the existence of signals generated from cytoplasmic and cell surface steroid receptors, are generally known as nonclassical, or nongenomic steroid signals (Losel and Wehling, 2003). Nongenomic steroid signaling has the following features: responses are very rapid (taking only a few seconds or minutes), insensitive to inhibitors of mRNA and protein synthesis, found in highly specialized cells (e.g., spermatozoa) that do not accomplish mRNA and protein synthesis and in cells lacking steroid nuclear receptors and they can be activated by steroids coupled with high-molecular-weight substances such as estrogen-bovine serum albumin conjugates that do not pass across the plasma membrane. Nongenomic signaling of estrogen involves a series of events depending on the cell type that include mobilization of second messengers, interaction with membrane receptors such as insulin like growth factor-1-receptor (IGF-1R), and epidermal growth factor receptor (EGFR), and stimulation of effector molecules such as Src and PI3 kinase (PI3K), serine/threonine protein kinase (AKT), and MAPK in various cell types (Ho and Liao, 2002).
MECHANISMS OF ESTROGEN NONGENOMIC SIGNALING
The rapid action of estrogen can be divided into two major categories: classical receptor-mediated responses, which are mediated through ERs, and nonclassical, non-receptor-mediated responses, which are mediated through proteins other than ERs, such as GPR30. Although the estrogen actions elicited at the plasma membrane were identified almost two decades ago (Pietras and Szego, 1977), research into these actions has lagged behind since then. However, the existence of plasma membrane ERs and the significance of cytoplasmic ER effects have been increasingly accepted because of the accumulation of strong supporting experimental evidences (Kelly and Levin, 2001).
Membrane translocation of estrogen receptor through post-translational modifications
Despite the lack of proof of a membrane-associated ER, the results of several biochemical and microscopic analyses have suggested the existence of different pools of ER in the cellular environment, including the plasma membrane (Razandi et al., 1999), the mitochondria (Chen et al., 2004), and the endoplasmic reticulum (Govind and Thampan, 2003). In general, about 80% of ERs localize in the nucleus in the absence of estrogen. Ligand stimulation enhances this nuclear accumulation (Ylikomi et al., 1992). Further subcellular distribution of ER content varies from cell to cell. Since many estrogen-stimulated nongenomic pathways are thought to be initiated at the plasma membrane, researchers have made several attempts to identify a membrane-associated ER.
The membrane-binding activity of estrogen was first identified by Pietras and Szego (1977), but the precise nature of the receptors at those sites are still unknown, and direct evidence supporting their existence has not been found. However, circumstantial evidence has come both from immunohistochemical and other studies with membrane-impermeable ligands and from studies of non-physiological systems involving the overexpression of nuclear receptors (Pappas et al., 1995; Norfleet et al., 1999). The plasma membrane ER appears to be identical to the nuclear ER, according to results of immunochemistry analysis with a panel of antibodies against multiple epitopes of the nuclear ER (Ropero et al., 2002; Watson et al., 2002).
Biochemical and microscopic evidence of a membrane ERα has also proved controversial because classical ERs contain no stretches of hydrophobic transmembrane, have no intrinsic kinase or phosphatase activity, and lack myristoylation sites to recruit into the membrane. Nevertheless, emerging data have revealed several post-translational ERα modifications that in principle could contribute to membrane localization of ERα. For example, Razandi et al. (2003) found that the serine 522 residue within the estrogen binding domain of ERα was critical in linking the receptor to the cell membrane through interaction with caveolin-1. However, these researchers also found that expression of an ERα serine-522-alanine mutant in Chinese hamster ovary cells failed to completely abolish ERα membrane localization or MAPK stimulation, suggesting that modification of serine 522 is not the sole basis for ERα membrane localization and that it may not affect the activation of MAPK stimulation (Razandi et al., 2003). These findings were later challenged by the discovery that ERα is also modified by palmitoylation at cysteine 447 and that palmitoylation modification is an important determinant of ERα localization at the plasma membrane, and that such modifications are responsible for the ligand-induced MAPK and PI3K/AKT pathways (Acconcia et al., 2005).
Membrane translocation of estrogen receptor through adaptor molecules
In addition to post-translational modifications of ERα, other mechanisms have been proposed to explain the translocation of ERα to the cell membrane in estrogen-stimulated cells. As we discussed earlier that unlike many other membrane receptors, ERα has no intrinsic transmembrane domain, indicating that ERα is not a membrane protein postulated to be required a third party protein to link ERα to the cell membrane and that requires for activation of MAPK. Since Shc association with membrane-bound receptors is required for MAPK activation (Ravichandran, 2001), Shc has been proposed as an adaptor that supports MAPK stimulation. Indeed estrogen rapidly induces Shc phosphorylation as well as the interaction between ERα and Shc. Interestingly, these interactions occur in a ligand-dependent manner which requires PTB/SH2 domains of Shc and N-terminal region of ERα (Song et al., 2002). Consistent with this result, ICI182780, an anti-estrogen, blocked estradiol-induced ERα membrane translocation, Shc activation and MAP kinase phosphorylation indicating the direct involvement of the ER. Since Shc is known to be an integral part of EGFR signaling, these findings support the idea of cross talk between ERα and EGFR.
Kahlert et al. (2000) showed that estrogen can rapidly induce phosphorylation of IGF-1R and activation of MAPK. Song et al. (2004) extended this idea by showing Shc's role in rapid estrogen signaling. They demonstrated the existence of a ternary complex composed of ERα /Shc/IGFR that leads to ERα membrane association leading to MAPK stimulation. It was further supported by siRNA knockdown analysis using either Shc or IGF-1R specific siRNAs or the use of an anti-estrogen which disrupted the formation of the ternary complex and estrogen's ability to promote ERα membrane translocation clearly indicating the existence of an adaptor molecule in estrogen mediated signaling.
In a study to determine how ovarian hormones stimulate breast tumor growth, Migliaccio et al. (1996) described a mechanism of ligand dependent rapid stimulation of MAPK in MCF-7 breast cancer cells. Estradiol treatment of human mammary cancer MCF-7 cells triggers rapid and transient activation of the mitogen-activated (MAP) kinases, erk-1 and erk-2, increases the active form of p21ras, tyrosine phosphorylation of Shc and p190 protein and induces association of p190 to p21ras-GAP. Since Shc and p190 are substrates of activated Src and once phosphorylated, they interact with other proteins and upregulate p21ras. These initial observations suggested a possible role of c-Src as a “signal integrator” in estrogen-stimulated breast cancer cells since estrogen stimulates MAPK activation through the interaction of c-Src with ERα (Migliaccio et al., 1996).
Although estrogen-dependent interactions of ERα with Src have been reported, it is still unknown whether such regulatory interactions can stimulate kinase activities in vivo or whether other factors are also required. ERα has been shown to interact with the SH2 domain of Src (Martin, 2001; Boonyaratanakornkit et al., 2001), but such interaction is independent of Src activation (Boonyaratanakornkit et al., 2001). To resolve this discrepancy, Wong et al. (2002) identified a protein called modulator of nongenomic activity of estrogen receptor (MNAR) using proteomic approach. At the level of function and amino acid sequence, MNAR is identical to PELP1, which was initially identified by Vadlamudi et al. (2001) as an ERα co-activator. Wong et al. (2002) proposed that this protein might act as an adaptor for connecting ERα with c-Src. Thus, they provided a possible mechanistic basis for ERα's regulation of c-Src stimulation through the stabilization of a ternary complex composed of PELP1/MNAR, ER, and c-Src via the LXXLL and PXXP motifs. This was supported by the fact that the activation of Src kinase depends on the status of phosphorylation of tyrosine at positions 419 and 517 thus releasing internal SH2 and SH3 interactions (Martin, 2001). PELP1/MNAR facilitates estrogen- and ER-dependent activation of c-Src through its PXXP motif, which displaces the intramolecular occupation of SH3 in Src and promote the interaction with ERα. Interaction of ER with the SH2 domain of c-Src may help stabilize Src activation. The recent data indicate that a subpopulation of ERα may be localized in close proximity to the plasma membrane, which potentially can explain the ER–PELP1/MNAR–Src complex formation (Vadlamudi et al., 2005). Since the PELP1/MNAR is localized both in nucleus and cytoplasm, it is possible that it can involve in dual role for ER's actions. Besides activating c-Src by PELP1/MNAR in estrogen-stimulated breast cancer cells, PELP1/MNAR found to contain potential PI3K interacting domains and thus it also has been shown to activate AKT/MAPK pathway (Vadlamudi et al., 2005). In fact this pathway additionally providing the mechanism for tamoxifen resistance in breast cancer cells (Vadlamudi et al., 2005). Since it plays dual role, PELP1/MNAR is probably the first example of a protein that is a coactivator for transcription regulation and add an account for nongenomic activities of ER.
In addition to using Shc and PELP1/MNAR as adaptors for cytoplasmic signaling, the ER also utilizes p130Cas an adaptor molecule, which links the actin cytoskeleton to the extracellular matrix during cell migration, cell invasion, and cell transformation (O'Neill et al., 2000), for rapid actions of estrogen (Cabodi et al., 2004). In human T47D breast cancer cells, it is evident that ligand binding to its receptor promotes the association of the receptor with p130Cas adaptor protein, c-Src, and the p85 subunit of PI3K. Consistent with this result, overexpression of p130Cas accelerates ERα-dependent stimulation of c-Src kinase, Erk1/2, and cyclin D1 expression, indicating that p130Cas is a dynamic component of the ERα signaling complex and plays a role in the early steps of estrogen-dependent nongenomic signaling (Cabodi et al., 2004).
ER–PI3K—growth factor receptor complex
Emerging evidence suggest that the ER signaling is complex, involving multiple coregulatory proteins (Barnes et al., 2004) and cross talk with a number of cellular pathways (Driggers and Segars, 2002; Levin, 2002). PI3K plays an important role in cell growth and the prevention of apoptosis (Leevers et al., 1999). PI3K mediates the cellular effects of platelet-derived growth factor (PDGF), insulin, and vascular endothelial growth factor (VEGF). The predominant form of PI3K comprises p85α, an adaptor/regulatory subunit of relative molecular mass 85,000 (Mr 85K), and p110, a catalytic subunit. Simoncini et al. (2000) provided the evidence that ERα directly associates with p85α in a ligand-dependent manner in endothelial cells. This ligand-dependent interaction was blocked by ICI 182,780 and was absent in p85α−/− fibroblasts transfected with ERα cDNA alone. However, the ER isoform ERβ, which is thought to mediate some of the cardiovascular effects of estrogen, did not interact with p85α or activate PI(3)K after E2 stimulation. This interaction, however, does not involve the src-homology SH2/SH3 domains of p85α. In human vascular endothelial cells, physiological concentrations of 17β-oestradiol increased eNOS activity in a biphasic manner. The initial increase was mediated by mitogen-activated protein (MAP) kinases; the second increase was completely blocked by the PI(3)K inhibitor, wortmannin. And also the increase in eNOS activity was blocked by the ERα antagonist ICI 182,780 while the inactive E2 stereoisomer 17β-oestradiol had no effect indicating the involvement of ER. This interaction stimulates eNOS activity and promotes cell survival. This pathway of eNOS activation through PI3K/AKT has direct implication in the rescue of ischemia-reperfusion-triggered injury.
PI3K is associated with various growth factor signaling pathways and also it has been reported that PI3K interacts with EGFR and IGF-1R (Yamamoto et al., 1992; Zhang et al., 2004). So it is possible that a multiprotein complex like ERα–PI3K-growth factor receptor may exist and such a complex may provide a cross talk between ERα and growth factors. In fact, we have provided an evidence that cytoplasmic localization of PELP1/MNAR, a coactivator of ERα, also interacts with EGFR and PI3K (85 kDa subunit) leading to the activation of AKT/MAPK pathway in breast cancer cells (Vadlamudi et al., 2005). It is more likely that estrogen activates AKT through this cross talk, since the AKT is a downstream target of PI3K.
In addition, Fernando and Wimalasena (2004) recently have shown that estrogen induces BAD phosphorylation through both the Ras/PI3K/Akt and the Ras/ERK/p90RSK1 pathways and suggested that functional activation of the PI3K/AKT pathway may be required for estrogen to block apoptosis induced by tumor necrosis factor, hydroxyperoxide, and serum withdrawal. This model suggest the anti-apoptic activity of estrogen's nongenomic rapid action.
Besides the above described ER interacting proteins, ligand-bound ER also activates other cytoplasmic signaling components, such as Raf-1 (Pratt et al., 1998), protein kinase A (Farhat et al., 1996), protein kinase C (Keshamouni et al., 2002), and SphK (Sukocheva et al., 2003). Interestingly, Sukocheva et al. (2003) found that estrogen-increased spingosine kinase (SphK) activity coincided with enhanced cell growth, ligand-dependent activation of MAPK, and intracellular Ca2+ mobilization in human breast cancer cells. This results suggested that SphK activation is also an important cytoplasmic signaling to transduce estrogen-dependent mitogenic and carcinogenic action in human breast cancer cells.
As is known, upon ligand binding, a majority of ER pool localizes into the nucleus and performs its nuclear functions. However several biochemical and cytochemical analysis suggest that a portion of the receptor is localized in the cytoplasm at different locations including plasma membrane. This raised the question of how the receptor is retained in the cytoplasm and how it brings all the cytoplasmic signaling. It is known that upregulation of metastatic tumour antigen 1 (MTA1) is associated with the invasiveness and metastasis potential of several human cancers acting as a co-repressor for nuclear ERα transactivation and also that it transcriptionally represses ERα (Mazumdar et al., 2001). We have provided a novel mechanism of this ERα cytoplasmic localization through sequestration by a novel ERα interacting protein called MTA1s, of MTA1 family (Kumar et al., 2002). MTA1s, a naturally occurring short form of MTA1, contains a previously unknown sequence of 33 amino acids with an ERα-binding motif, Leu-Arg-Ile-Leu-Leu (LRILL). Interestingly, MTA1s localizes in the cytoplasm, sequesters ER in the cytoplasm, and thereby enhances non-genomic responses of ERα. Further, we have demonstrated that deleting the LRILL motif in MTA1s abolishes its co-repressor function and its interaction with ERα, and restores nuclear localization of ERα. In breast cancer cells, overexpression of MTA1s potentially activates MAPK and prevents ligand-induced nuclear translocation of ERα. Dysregulation of human epidermal growth factor receptor-2 in breast cancer cells enhances the expression of MTA1s and the cytoplasmic sequestration of ERα. The regulation of the cellular localization of ERα by MTA1s represents a novel mechanism for redirecting nuclear receptor signaling by nuclear exclusion. It could be possible that ERα is retained in the cytoplasmic compartment by its sequestration by MTA1s, thus promoting the activation of cytoplasmic ERα functions such as Grb2/Shc interaction and MAPK stimulation. So ERα sequestration by MTA1s might be responsible for augmenting the membrane ERα pool in cells with active growth factor signaling. Since this type of mechanism involves suppression of ERα genomic responses because of the redistribution of ERα to the cytoplasm, it is more likely that cytoplasmic localization of the receptor leads to the activation of nongenomic activities of estrogen.
Caveolae in rapid estrogen signaling
Several reports suggest the importance of rapid actions of estrogen in the vasculature and its atheroprotective properties (Farhat et al., 1996; Mendelsohn, 2002). A decade ago, it was reported that the rapid and short-term action of ethinyl–estrogen causes an immediate decrease in coronary vasomotor tone within 15 min of exposure (Reis et al., 1994). Later on, it become evident that short-term exposure of the vasculature to estrogen leads to nitric oxide (NO)-dependent vasodilation (Denninger and Marletta, 1999). NO is a potent regulator of blood pressure, platelet aggregation, leukocyte adhesion, and vascular smooth muscle mitogenesis in response to endothelial nitric oxide synthase (eNOS) production in the vascular wall (Moncada and Higgs, 1993). In cultured endothelial cells, estrogen enhances the release of NO within minutes without altering the expression of eNOS (Schlegel et al., 1999). Using animal models, Guo et al. (2005), found that rapid actions of estrogen are required for nitric (NO) oxide to be produced and that this NO production eventually leads to vasodilation, presumably through membrane-associated ER signaling. Several evidences suggests that eNOS is targeted to the endothelial plasma membrane, particularly to caveolae, which are specialized, cholesterol-rich domains docked on the plasma membrane that are involved in compartmentalizing signal transduction pathways (Okamoto et al., 1998). These structures are abundant in simple squamous epithelia, fibroblasts, smooth muscle cells, and adipocytes. An important structural component of caveolae is caveolin, a 22-kDa transmembrane phosphoprotein. Caveolin forms a scaffold with several types of signaling molecules to target preassembled signaling complexes (Okamoto et al., 1998). Interestingly, ERα has also been found in caveolae, where it interacts with caveolin in a ligand-dependent manner and activates eNOS (Schlegel et al., 1999).
Vascular endothelial cells represent another excellent model system for studying the nongenomic effects of estrogen on the vasculature. In endothelial cells, estradiol stimulates phosphorylation of Akt at Ser473, which in turn phosphorylates eNOS at Ser1117, a critical residue for eNOS activation and enhanced sensitivity to the resting levels of Ca2+ (Dimmeler et al., 1999). Akt is a downstream substrate of PI3K, further suggesting that PI3K plays a role in estrogen responses. Chen et al. (1999) showed that estrogen-mediated activation of eNOS was abolished by estrogen antagonists and required the hormone-binding domain of ERα. Simoncini et al. (2000), also found that an ERα mutant lacking this hormone-binding domain was not able to activate eNOS, thus suggesting an essential role of this domain in regulating eNOS activity.
In a study to determine the basis of nongenomic eNOS stimulation by estrogen, Chambliss et al. (2005) found that dimerization is not required for ERα coupling to eNOS, however, ERα mutants lacking the nuclear localization signals (NLS),(ERα Δ250–274) or the DNA binding domain (ERα Δ185–251), which targeted normally to plasma membrane and caveolae/rafts, were incapable of activating eNOS suggesting the importance of NLS in Src activation, and DNA binding region in the dynamic interaction between ERα and PI3 kinase.
Recently, Lu et al. (2004) provided a possible mechanism by which estrogen activates the PI3K–Akt kinases and eNOS pathways in endothelial cells. They found that Striatin, a calmodulin-binding member of the WD-repeat family of proteins (Castets et al., 1996), interacts with caveolin. Interestingly, Striatin acts as a molecular anchor to target ERα to the membrane and then organizes the ERα–eNOS signaling complex. They also found that Striatin directly binds to amino acids 183–253 of ERα and serves as a scaffold for the formation of an ER–Gαi complex. Disruption of ERα–Striatin complex formation blocks estrogen-induced rapid activation of MAPK, Akt, and eNOS but has no effect on ERα transactivation functions. These findings suggest that Striatin serves as a molecular scaffold required for rapid, nongenomic estrogen-mediated activation of downstream signaling pathways and, thus differentially influence nongenomic, but not genomic ERα responses.
Most of the estrogen nongenomic signaling mechanisms, provided so far, occur through membrane localization or cytoplasmic retention of ER via receptor modifications, adaptor proteins, binding through membrane associated growth factor receptors or nonreceptor kinases and appears that the rapid actions of estrogen occur via formation of multi-protein complexes with ligand bound receptor (Fig. 1).
Calcium flux in response to rapid actions of estrogen
Calcium ions constitute an important component of the second messenger pathways including estrogen action. Estrogen has been shown to elevate both the intracellular concentration of calcium within 5 sec by mobilizing calcium from the endoplasmic reticulum and the formation of inositol 1,4,5-trisphosphate and diacylglycerol by activating phospholipase C (Lieberherr et al., 1993). The activation of phospholipase C is an early event in the signal transduction pathways that lead to a variety of cellular responses, including metabolism, proliferation, secretion, and motility. Estrogen can also rapidly stimulate the entry of Ca2+ into isolated duodenal enterocytes through a phospholipase C-dependent mechanism (Picotto et al., 1999).
The regulation of intracellular Ca2+ flux via membrane ERα may contribute to estrogen's rapid stimulation of NO formation in monocytes (Stefano et al., 1999). Doolan et al. (2000) found that in female rat colon cells, estrogen's increase of intracellular Ca2+ depended on the activation of L-type, a voltage-gated Ca2+ channel opening. This activation led to the rapid activation of cAMP-dependent protein kinase A. In addition, estrogen activates protein kinase A and protein kinase C cascades in hypothalamic neurons altering synaptic transmission in those cells (Kelly and Wagner, 1999). In an another study, estrogen stimulated the production of cyclic nucleotides in pancreatic B cells and augmented a glucose-induced increase in intracellular calcium and cyclic guanosine monophosphates (Ropero et al., 2002).
In human sperm membranes, rapid activation of intracellular Ca2+ and protein tyrosine phosphorylation occurs in response to estrogen (Baldi et al., 2000). Estrogen signaling can result in the inhibition of the acrosome reaction induced by progesterone (Baldi et al., 2000). Since this type of reaction could be blocked by an antibody to the ligand-binding domain of ER (H222), the investigators concluded that a 29-kDa estrogen-responsive ER was present on the membrane. In another study, Loomis and Thomas (2000) found that estrogen and bovine serum albumin-conjugated estrogen inhibited gonadotropin-induced androgen secretion from fish testicular fragments in less than 5 min. As a result, they identified the presence of a high-affinity saturable membrane ER.
In hypothalamic neurons, estrogen activates a membrane-associated ER coupled with Gαq that leads to the activation of phospholipase C and in turn catalyzes the hydrolysis of membrane-bound phosphatidylinositol 4,5-biphosphate to inositol 1,4,5 triphosphate (IP3) and diacylglycerol (DAG). Calcium, released from intracellular stores (endoplasmic reticulum) by IP3 and DAG, activates PKCδ, which in turn activates adenylate cyclase resulting the production of cAMP. The generation of cAMP activates PKA, which can rapidly uncouple GABAB and µ-opioid (µ) receptors from their effector system through phosphorylation of a downstream effector molecule such as the inwardly rectifying K+ channel (Kelly and Wagner, 1999). The rapid effects of estrogen induce ERα-regulated signaling kinases, which in turn either reduce the ability of neuromodulators such as gamma aminobutyric acid and endorphins to inhibit hypothalamic neuronal excitability or augment the ability of neurotransmitters such as glutamate to increase neuronal excitability (Qiu et al., 2003). All these signaling events triggered by estrogen via nongenomic signaling impacts on neuronal cell survival.
Both estrogen and EGF are potent mitogens for cells from mammary epithelia and uterine endometrium (Dickson and Lippmann, 1998). However, the signaling mechanisms used by these two mitogens are quite different. For example, the proliferative effects of estrogen are primarily mediated by ER and have been linked to ER's ability to induce gene transcription. In contrast, the biological responses of EGF are transmitted through the transmembrane EGF receptor, which signals by recruiting intracellular signaling cascades. Since estrogen has been shown to activate intracellular signaling events through PI3K/Src-MAPK, Shc-IGFR-MAPK, PI3K-eNOS pathways, similar to those activated by EGF suggests that these divergent signaling mechanisms may cross-talk. Initially, the evidence of a relationship between the EGFR and estrogen was provided by data showing that neutralizing antibodies against EGF inhibited estrogen-mediated proliferation in the uterus (Nelson et al., 1991). Studies from the Lieberherr and Grosse (1994) group suggested the increased intracellular concentration of calcium and as well as inositol trisphosphate (IP3) and diacylglycerol (DAG) in response to estradiol are indicative of involvement of membrane receptors coupled to phospholipase through a pertussis toxin-sensitive G protein. Further studies on this line showed that inhibitors of G-protein signaling blocks second messenger signaling by estrogen (Le Mallay et al., 1997). Investigations of nonclassical estrogen signaling suggest nuclear ER or ER-like proteins are likely candidates for the membrane ERs (mERs) mediating these estrogen actions in a variety of target cells, including endothelial, neuronal, and pituitary cells (Chen et al., 1999; Razandi et al., 1999; Song et al., 2002). However, evidence has also been obtained for the involvement of novel membrane ERs unrelated to nuclear ERs (nERs) in nonclassical estrogen actions in several other cell types, many of which are associated with G proteins (Gu et al., 1999; Nadal et al., 2000; Qiu et al., 2003). Another recent discovery of a hitherto unknown family of membrane progesterone receptors (mPRs), unrelated to nuclear steroid receptors, but instead with characteristics of G protein-coupled receptors (GPCRs) (Zhu et al., 2003), suggest the existence of other GPCRs with characteristics of steroid membrane receptors.
G protein-coupled receptors (GPCRs) represent the largest class of signaling molecules containing heptahelical transmembrane proteins. GPCRs transduce their signals via G-protein heterotrimers (αβγ) that dissociate into free Gα-subunit protein and Gβγ-subunit protein complexes following ligand stimulation (Gether, 2000). G protein-coupled receptor, GPR30 has been anticipated as the right candidate for nongenomic signaling of estrogen based on the observations that GPR30 is preferentially expressed in ER-positive relative to ER-negative breast tumor cell lines (Carmeci et al., 1997). In consistence with this idea, Filardo (2002) found that estrogen-induced ERK activation occurs in human SKBR3 breast cancer cells that fail to express either ERα or ERβ. Further, the introduction of GPR30 cDNA into other ER-negative low GPR30-expressing MDA-MB-231 cells enables ERK activation in response to estrogen. Interestingly, GPR30-dependent ERK activation may also be induced by ER antagonists, ICI 182,780, but not by estrogen, suggesting that these estrogen effects are ER independent.
Two independent studies recently confirmed the role of GPR30 in nongenomic estrogen actions that are independent of ERs. In the first, Thomas et al. (2005) reported the presence of high-affinity estradiol bound to the membranes of breast cancer cells that lack ERs but express enough GPR30. Interestingly, membranes fractions from human embryonic kidney cells, which lack all three molecules, acquire estradiol-binding activity when induced to express GPR30. GPR30 binding is selective for estrogen and the ER antagonists tamoxifen and ICI 182,780. In contrast, weak estrogens have very low affinities for GPR30. Antiestrogens such as ICI 182,780 and tamoxifen activate GPR30-mediated responses, yet these agents are antagonists of ER-mediated responses. In addition, it was demonstrated that GPR30 can act as a mER in transfected cells to transduce the signals of estrogenic compounds with high RBAs (receptor binding affinity) for the receptor, resulting in activation of a stimulatory G protein and upregulation of adenylyl cyclase activity, whereas E3 and E1 (estrogen derivatives), which have low RBAs for the receptor, were inactive. Finally, the decrease in mER binding in transfected cell membranes observed after treatment with agents causing uncoupling of G proteins from GPCRs, GTPγ-S, and CTX (Liu and Dillon, 2002), indicates the mER is directly coupled to G protein and is a GPCR, consistent with its identity as GPR30.
In the second study, Revankar et al. (2005) discovered that GPR30 was localized in the endoplasmic reticulum membranes in COS7 cells rather than on the plasma membrane as shown by Thomas et al. (2005). Since estrogen-mediated rapid signaling events are sensitive to pertussis toxin, implies the involvement of Gi/o proteins. Further, they verified the ERα and GPR30 stimulated calcium mobilization and found that ERα- and GPR30-initiated calcium mobilization are mediated by divergent signaling pathways. They also used PI3K signaling pathway as a model pathway for estrogen action in SKBR3 cells, an ERα negative breast cancer cells, and found that PIP3 generation by GPR30, but not by ERα, requires EGFR activation indicating that EGFR transactivation is similarly involved in calcium mobilization and PI3K activation for both ERα and GPR30. These data demonstrate that GPR30 represents the sole estrogen-binding and functionally responsive ER in ERα negative cells. However, these researchers did not determine whether estrogen-GPR30 interactions contribute to overall estrogen signaling at the plasma membrane or whether these effects are downstream physiological effects of estrogen in breast cancer cells.
It is clear that these two studies demonstrate the membrane ER as a protein structurally unrelated to nuclear ERs. The steroid binding characteristics of the recombinant GPR30, like those of the wild-type receptor, fulfill all the criteria for its designation as a membrane ER. Both forms of GPR30 display high affinity and saturable estrogen binding with Kds of approximately 3.0 nM similar to the affinities of other membrane ERs (Thomas et al., 2005). This mechanism explains the rapid actions of estrogen at the cell surface and act in certain nuclear ER-negative target cells. The existence of this novel membrane ER-mediated signaling pathway will explain some of the pleiotropic actions of estrogens in breast and other estrogen target tissues in the near future.
SYNERGY BETWEEN NONGENOMIC AND GENOMIC ACTIONS
In general, signal transduction processes are integrated into other processes that control gene transcription. In this context, activation by both EGF and IGF-1 provides the best example to support the notion that nonnuclear estrogen-independent stimulation modulates ER-nuclear activity. For example, membrane estrogen has been shown to transactivate EGFR and IGFR (Filardo, 2002; Kahlert et al., 2000; Keshamouni et al., 2002), leading to activation of the PI3K, AKT, and MAPK pathways. Growth factor-induced stimulation of MAPKs results in the direct phosphorylation of ERα on Ser118 (Kato et al., 1995; Bunone et al., 1996; Endoh et al., 1999). The phosphorylation of ERα by growth factors enhances ERα binding to the p68 RNA helicase (Endoh et al., 1999) and accounts for enhanced AF-1 transcriptional activity in uterine and ovarian adenocarcinoma cells (Ignar-Trowbridge et al., 1996). In addition, pathways such as PI3K/Akt, which are downstream of EGFR or IGFR, can also activate estrogen-responsive target genes in the absence of estrogen stimulation (Martin et al., 2000). Similarly, activation of p21-activated kinase is sufficient to phosphorylate ERα on serine 305, leading to enhanced transcription of cyclin D1 in breast cancer cells (Wang et al., 2002; Balasenthil et al., 2004). Since the same serine residue in ERα can also be phosphorylated by protein kinase A (Michalides et al., 2004), growth factor signaling could stimulate ERα transactivation in a ligand-independent manner and thus contribute to tamoxifen resistance in women.
In addition to direct phosphorylation of the receptor, EGF can modulate the coactivator phosphorylation state. Steroid receptor coactivator-1, a member of the p160 family of adaptor molecules that recruit other proteins to the coactivator complex, contains potential phosphorylation sites for ERK-1/2. EGF stimulation results in ERK-1/2-mediated phosphorylation of steroid receptor coactivator-1, which potentiates ER transcriptional activity (Rowan et al., 2000). PELP1/MNAR, a coactivator of ER in estrogen genomic signaling and also coactivator of STAT3 in growth factor signaling (Manavathi et al., 2005) has been shown to be tyrosine phosphorylated in response to either growth factors or estrogen and potentially modulate ERα transactivation functions when this coactivator is localized in the cytoplasm (Vadlamudi et al., 2005) and promotes the tamoxifen resistance in breast cancer cells. In addition, it is known that the NF-κB transcription factor complex is a target for phosphorylation by the Akt kinase (protein kinase B) (Romashkova and Makarov, 1999) and that activation of the PI3K/Akt signaling pathway by 17β-estradiol leads to expression of genes that contain NF-κB binding sites (Kawagoe et al., 2003). One such gene targeted by NF-κB is an inducible gene for rapid E2 signaling (Pedram et al., 2002).
The transcription factors Elk-1 (Cruzalegui et al., 1999), C/EBPβ (Buck et al., 1999), and cAMP response element binding protein (CREB) (Bonni et al., 1999) are all targets for phosphorylation by the MAPK signaling pathway. Activation of serum response elements induced by 17β-estradiol requires intact MAPK activity and the phosphorylation of Elk-1 (Duan et al., 2001). Kousteni et al. (2003) reported that the rapid actions of steroids alter the activity of Elk-1, C/EBPβ, and CREB or c-Jun/c-Fos by the actions of extranuclear ER or AR, resulting in activation of the Src/Shc/ERK pathway or downregulation of c-Jun N-terminal kinase. Another important transcriptional factor, AP1, is also regulated through MAPK (Karin, 1995). The activation of the MAPK signaling pathway by 17β-estradiol results in enhanced AP-1 DNA binding activity and transcriptional activation (Björnström and Sjöberg, 2004). In white adipocytes of rats, Garcia et al. (2000) showed that estrogens rapidly increase mRNA and protein expression of the two major components of AP-1, c-fos, and c-jun, and that they also enhance AP-1 DNA binding activity. It appears that this activation occurs, at least in part, through nongenomic activation of the MAPK cascade (Dos Santos et al., 2002).
PHYSIOLOGICAL OUTCOMES OF ESTROGEN NONGENOMIC SIGNALING
Estrogen regulates a wide range of neuronal functions, including gene expression, survival, and differentiation (Garcia-Segura et al., 2001). Neuroprotective effects of estradiol are at least partly mediated through ERs that are expressed in the brain and the rapid actions of estradiol in neurons are believed to be associated with the membranous ER or with a transient association of ER with specific membrane compartments (Toran-Allerand et al., 2002). Recent studies have suggested the involvement of IGF-IR in estrogen regulation of neuronal differentiation (Duenas et al., 1996), synaptic plasticity (Cardona-Gomez et al., 2002), neuronal survival after injury, gonadotrophin secretion, and reproductive behavior (Quesada and Etgen, 2002). Mechanistically, this regulation may be related to the fact that ERα, but not ERβ, interacts with IGF-IR and with the p85 subunit of PI3K in the brain (Mendez et al., 2003), and such interactions could also lead to Akt activation (Cardona-Gomez et al., 2002). Since glycogen synthase kinase- 3 is a downstream target of AKT, the reported neuroprotective effects of estrogen may be mediated by glycogen synthase kinase 3 (Cardona-Gomez et al., 2004).
Estrogen reportedly helps to prevent the primary development of cardiovascular disease in women. Estrogen increases arterial vasodilation, inhibits cellular response to vascular injury, and prevents atherosclerosis. Arterial vasodilation occurs primarily through rapid, nongenomic pathways, which are independent of changes in gene expression (Mendelsohn, 2002). Physiologic concentrations of estrogen can also rapidly stimulate the opening of calcium-activated potassium ion channels in the cell membrane of vascular smooth muscle cells and it may contribute to relaxation of the vascular smooth muscle (Mendelsohn, 2002). Of particular relevance to the vascular system is a membrane receptor in endothelial cells that binds rapidly to estrogen or estrogen-bovine serum albumin conjugate, selectively activates antiapoptotic p38β MAPK, and inhibits proapoptotic p38α, leading to the upregulation of Elk-2 and the phosphorylation of heat shock protein hsp27 (Razandi et al., 2000). The resulting downstream effects include preservation of stress fiber formation and membrane integrity, prevention of hypoxia-induced apoptosis, induction of endothelial cell migration, and formation of primitive capillary tubes. Thus, estrogen may exploit pathways that preserve the actin cytoskeleton during ischemia, prevent cell death, and enhance angiogenesis after injury. In brief, estrogen signaling from the membrane helps to preserve endothelial cell structure and function.
Rapid estrogen actions in osteocytes
Sex steroids play a key role in the development and maintenance of the skeleton in virtually all species (Riggs et al., 2002). Estrogen affects the skeletal system in many ways, including inhibiting bone remodeling, suppressing bone resorption, and presumptively stimulating bone formation. Estrogen deficiency results in an imbalance between bone resorption and formation that does not occur when there is enough estrogen (Syed and Khosla, 2005). ER-alpha and ER-beta are present in osteoblasts, osteoclasts and in their progenitor cells (Eriksen et al., 1988). Estrogens are the bone-protective steroid hormones present in both adult male and female skeletons. Considerable evidence exists at the cellular level to support the notion that estrogen prolongs the lifespan of osteoblasts by inhibiting osteoblast apoptosis (Kousteni et al., 2001). This apoptosis-inhibiting activity in osteoblasts appears to be mediated by the activation of the Src/Shc/ERK signaling pathway. Estrogen rapidly and transiently increases the phosphorylation of ERKs in MLO-Y4 cells with the greatest effect seen at 5 min and a return to baseline by 15 min (Kousteni et al., 2001). Moreover, both pharmacological inhibitors of ERK and as well as Src kinase abrogate the anti-apoptotic effects of estrogen, suggesting that ERK phosphorylation and Src kinase activity are required for the antiapoptotic effects of steroid hormones. These rapid actions are mediated through the ligand-binding domain and eliminated by nuclear targeting of the receptor protein. The antiapoptotic effects of estrogen in bone cells may be linked to downregulation of c-Jun NH2-terminal kinase, leading to alterations, in the activity of key transcription factors, including Elk-1, CCAAT enhancer-binding protein-β (C/EBPβ), cAMP-response element-binding protein (CREB), and c-Jun/c-Fos. It has also been shown that 4-estren-3α,17β-diol, a synthetic ligand for ER or AR, which does not affect classical transcription, may protect bone loss after ovariectomy in mice by stimulating nongenotropic pathways (Kousteni et al., 2002). Taken together, these lines of evidence strongly suggest that, in sex steroid deficiency, loss of transcriptional effects may be responsible for the increased osteoclastogenesis and osteoblastogenesis and thereby the increased rate of bone remodeling. Loss of nongenotropic anti-apoptotic effects on mature osteoblasts and osteocytes, in combination with an opposite effect on the lifespan of mature osteoclasts, may be responsible for the imbalance between formation and resorption and the progressive loss of bone mass and strength. So that is why the elucidation of the dual function of sex steroid receptors has important pathophysiologic and pharmacologic implications.
Estrogen signaling in skin, oviduct, and other tissues
Skin is another nonreproductive tissue that appears to be influenced by estrogen. Estrogen improves collagen content and quality, increases skin thickness, and enhances vascularization (Thornton, 2002). However, little is known about ERs in skin, and the pathways activated by estrogen in skin-derived cells are also still poorly understood. Recent findings by Verdier-Sevrain et al. (2004) indicated that estradiol's effect on keratinocyte proliferation is mediated likely by both the nongenomic activation of Erk1/2 and the genomic signaling pathway through the activation of nuclear receptors.
Evidence also exists that estrogen's nongenomic action may be mediated by the cAMP and protein kinase-A signaling pathway in the mammalian oviduct (Orihuela et al., 2003). Although protein kinase-C activation is necessary for estrogen-induced phosphorylation of oviductal protein, inhibition of this enzyme does not block the effect of estrogen on egg transport. In conclusion, the study by Orihuela et al. (2003) showed that in the rat oviduct, ERα participates in nongenomic estrogen action.
EMERGING ROLE FOR ERβ IN NONGENOMIC SIGNALING OF ESTROGEN
Although the nongenomic potential of ERα has received considerable attention, the nongenomic potential of endogenous ERβ is poorly understood. Chambliss et al. (2002) evaluated eNOS activation in response to short-term exposure to estrogen in cultured endothelial cells, which expresses ERβ, to determine whether the ERβ isoform could exert nongenomic action and also to identify the subcellular localization of its putative function. They found that a subpopulation of ERβ was localized to the endothelial cell plasma membrane, overexpression of ERβ enhanced rapid eNOS stimulation by E2, and the response to endogenous ER activation was inhibited by the ERβ-selective antagonist RR-tetrahydrochrysene (THC). Abraham et al. (2003) demonstrated a nongenomic role for ERβ in gonadotropin-releasing hormone neurons by using the phosphorylation of cAMP response element-binding protein as an index of changes in intracellular signaling. They showed that estrogen acts rapidly and directly on the gonadotropin-releasing hormone neuronal phenotype and that this action requires estrogen to pass through the cell membrane and interact with ERβ. These studies provided the first evidence for a functional ERβ in the gonadotropin-releasing hormone neuron and for ERβ's involvement in nongenomic estrogen signaling in the brain. Moro et al. (2005) recently found that the nongenomic effects of estrogen are mediated through membrane-associated ERβ in enucleated platelet cells. In these cells, 17β-estradiol causes the rapid and transient tyrosine phosphorylation of Src and the formation of a membrane-associated, Src-dependent signaling complex, which includes ERβ, Src, Pyk2, and PI3-K in response to 17β-estradiol. These effects directly implicates ERβ in platelet aggregation. In addition to these effects, there may be lot of unknown mechanisms and physiology for the nongenomic potential of ERβ to be discovered.
FUTURE DIRECTIONS AND CHALLENGES
Mechanisms of estrogen-mediated cellular responses have become increasingly complex. In particular, the identification of a novel membrane ER indicates that estrogen and antiestrogen signaling in human breast cancer is more complex than previously anticipated. Several characteristics of the receptor have important implications for the development of the disease. Currently available data has convincingly established the existence of the physiological relevant nongenomic ER signaling and prevalence of ER components in the proximity of the plasma membrane. However there is still a clear need of experimental evidence to firmly establish the presence of membrane ER in physiologic cellular context. Since several mechanistic actions of estrogen are continued to be identified, the challenges would be to define the key mechanism of nongenomic actions used by estrogen and how exactly such events influences the cellular responses. A major challenge to scientists in academics and industry continues to be the development of ER ligands that retain the beneficial effects of estrogen in the targeted tissue, for example, bone, brain, and cardiovascular tissues, but lack the mitogenic and perhaps carcinogenic actions in the breast and uterus. Since the cross talk between ER and growth factor signals may augment the agonist activity of selective estrogen-receptor modulators, it is increasing clear that studies are needed for exploring how to inhibit growth factor signaling components to enhance the therapeutic efficacy of selective estrogen-receptor modulators. The discovery of novel direct mediator for estrogens, such as GPR30, and new nongenomic modulators such as PELP1/MNAR and MTA1 family members adds another layer of complexity to this issue but offers new opportunities for investigation.
We thank the members of the Kumar's laboratory for useful discussions and Dr. Joseph Mascarenhas for careful reading of this manuscript. We apologize to several of our colleagues for not citing their primary references due to space limitation.