Communication between genomic and non-genomic signaling events coordinate steroid hormone actions

Abstract

Steroid hormones are lipophilic molecules produced in one cell that can travel great distances within the body to elicit biological effects in another cell. In the canonical pathway, steroid hormone binding to a nuclear receptor (NR), often in the cytoplasm, causes the receptor to undergo a conformational change and translocate to the nucleus, where it interacts with specific sequences of DNA to regulate transcription. In addition to the classical genomic mechanism of action, alternate mechanisms of steroid activity have emerged that involve rapid, non-genomic signaling. The distinction between these two major mechanisms of action lies in the subcellular location of the initiating steroid hormone action. Importantly, the mechanisms of action are not exclusive, in that each can affect the activity of the other. Here, we describe the different types of genomic and non-genomic steroid hormone signaling mechanisms and how they can influence one another to ultimately regulate biology. Further, we discuss the approaches being used to study the non-genomic signaling events and address important caveats to be considered when designing new experiments. Thus, this minireview can serve as an introduction to the diverse signaling mechanisms of steroid hormones and offers initial, experimental guidance to those entering the field.

Introduction

Steroid hormone nuclear receptors (NRs) are transcription factors that are involved in various cellular activities. Classic steroid hormone NR activity occurs by binding of a steroid hormone ligand to its cognate receptor, which are often held in the cytoplasm in an inactive conformation by heat shock proteins (HSPs). Upon binding hormone, the receptor undergoes a conformational change that results in the dissociation of HSPs, translocation to the nucleus, dimerization, association with various coregulators, and binding to specific sequences of DNA termed hormone response elements (HREs). In some cases, NRs may already be located in the nucleus prior to hormone binding. Regardless, the DNA-bound complex can then regulate the transcription of genes [1]. It is now realized that steroid hormones can also activate signal transduction pathways and physiological changes independent of their actions in the nucleus, via a mechanism designated non-genomic signaling [2]. Since this bypasses the process of gene transcription, non-genomic signaling typically occurs on a faster time frame and is referred to as rapid or extranuclear signaling. Rapid signaling often involves NRs trafficking to the plasma membrane, where they can activate kinase pathways either directly or indirectly [3,4]. Importantly, non-genomic signaling can also regulate genomic pathways and vice versa. Hence, the final cellular effect of steroids is often the result of a convergence of events that began at separate locations. Described below are examples of the diverse mechanisms of action of steroid hormones with a particular focus on the estrogen receptors α and β (ERα and β), progesterone receptor (PR) and androgen receptor (AR), receptors for which some of the strongest evidence for non-genomic signaling exists. Finally, we discuss many of the experimental approaches being used to study these non-classical signaling events, highlighting some of their strengths and weaknesses that should be considered when planning new studies.

Classical NR signaling

Steroid hormones are best known to mediate various physiological cell functions via genomic activity. In this regard, steroid hormones typically interact with their cognate receptor in the cytoplasm for AR, glucocorticoid receptor (GR) and PR, but may also bind receptor in the nucleus as appears to often be the case for ERα and ERβ. This ligand binding results in a conformational change in the cytoplasmic NRs that leads to the dissociation of HSPs, translocation of the ligand-bound receptor to the nucleus (Figure 1). In the nucleus, the ligand-bound receptor dimerizes and then binds to DNA at specific HREs to regulate gene transcription. Several NRs can also interact indirectly with DNA by tethering to other transcription factors [5]. While some steroid hormone-induced nuclear events can occur in minutes [6], typically the genomic effects of steroid hormones take longer, with changes in gene expression occurring on the timescale of hours [7,8].

Figure 1.

Interaction of genomic and non-genomic responses mediated by steroid hormone signaling. Classical steroid hormone signaling occurs when hormone binds nuclear receptors (NR) in the cytoplasm, setting off a chain of genomic events that results in, among other changes, dimerization and translocation to the nucleus where the ligand-bound receptor forms a complex with coregulators to modulate gene transcription through direct interactions with a hormone response element (HRE). This transcription can lead to changes such as an increase of membrane growth factor receptors (GFRs) and calcium channels. Alternatively, steroid hormones may function through non-genomic signaling. This can occur through the plasma membrane localization of NRs by recruitment to the membrane upon palmitoylation and interaction with scaffolding proteins or by hormone-responsive GPCRs like G protein-coupled estrogen receptor (GPER). Non-genomic signaling can mediate physiological changes within the cell often by activating kinase cascades in the cytoplasm or by interaction with growth factor receptors at the plasma membrane. These signaling pathways can in turn increase the transcription of certain genes involved in various cell functions. In this way, non-genomic signaling can regulate genomic events. Conversely, genomic events can lead to the expression of genes involved in rapid signaling. Together, this demonstrates that both the genomic and non-genomic actions of steroid hormones can alter biological processes either independently or by influencing the other pathway.

Non-genomic signaling: localization of receptors

In addition to their classical genomic roles, NRs have been found at the plasma membrane of cells, where they can propagate signal transduction often through kinase pathways. The direct localization of ER, PR and AR to the plasma membrane is facilitated by palmitoylation by a palmitoylacyltransferase (PAT) at a conserved residue of the ligand binding domain of the receptor [3,9,10]. For example, DHHC-7 and -21 have been identified as PATs that may be involved in palmitoylation of ERα at the Golgi [10]. NR plasma membrane localization is also mediated by Hsp27 [11]. The receptor associates with the membrane at caveolae lipid rafts through interactions with various proteins including caveolin-1, Src and striatin [11–15]. From there, it can associate with membrane-associated kinases or activate G proteins to promote downstream cellular events.

Besides the trafficking of classical steroid hormone receptors to the plasma membrane, several other NR variants and even non-NR proteins located at the cell surface can bind to steroid hormones and respond by eliciting rapid signaling events. One example is ERα36, a 36kDa truncated form of ERα that lacks the transcriptional activation domains of the full-length protein. Membrane-localized ERα36 can activate pathways including protein kinase C (PKC) and/or mitogen activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) to promote the progression of various cancers [16–18]. Additionally, G protein-coupled receptor 30 (GPR30), also referred to as G protein-coupled estrogen receptor (GPER), is a membrane-localized receptor that has been observed to respond to estrogen to activate rapid signaling [19]. Another recently reported hormone-responsive G protein coupled receptor is Zip9, which androgens can activate to promote zinc uptake and signaling [20]. Finally, GPRC6A is another G protein-coupled membrane receptor that is responsive to androgen [21]. Mouse models indicate that androgen-mediated non-genomic signaling through this GPCR can modulate male fertility, hormone secretion and prostate cancer progression [22].

Non-genomic signal transduction of steroid hormones

Steroid hormone-mediated, non-genomic signaling can regulate diverse signal transduction pathways. Membrane-localized ER, PR and AR have been reported to modulate the activity of MAPK/ERK, phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt), nitric oxide (NO), PKC, calcium flux and increase inositol triphosphate (IP₃) levels to promote cell processes including autophagy, proliferation, apoptosis, survival, differentiation, and vasodilation [4,15,23–27]. Estrogens have been shown to induce rapid (i.e. seconds) calcium flux via membrane-localized ER (mER) [28,29]. These ER-calcium dynamics lead to activation of kinase pathways such as MAPK/ERK which can result in cellular effects like migration and proliferation [25,30]. Alternatively, 17β-estradiol (E2) has been reported to promote angiogenesis through the activation of GPER-mediated effects on 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (PFKFB3), an important enzyme in glycolysis [31].

Membrane NRs may also mediate rapid signaling through crosstalk with growth factor receptors (GFR). For example, membrane-localized steroid hormone receptors can activate GFR tyrosine kinases to activate diverse pathways [32,33]. This interaction may occur by phosphorylation of the GFR by steroid-activated GPER events [32]. Conversely, there is also evidence that GFRs can phosphorylate hormone receptors. For example, epidermal growth factor receptor (EGFR) can phosphorylate an ER/AR complex in breast and prostate cancer cell lines [34]. In addition, androgens and EGF, via MAPK signaling, can modulate paxillin’s ability to translocate from the cytoplasm to the nucleus and promote transcription mediated by AR and other transcription factors. This indicates that paxillin can function as a liaison between extra-nuclear kinase signaling and nuclear transcription [35,36]. A similar crosstalk occurs between the receptor tyrosine kinase insulin-related growth factor-1 receptor (IGF-IR) and ERα. Not only does IGF-IR activate ERα, but inhibition of IGF-IR downregulates estrogen-mediated ERα activity, suggesting that IGF-IR is essential for maximal ERα signaling. Further, ER activates IGF-IR pathways including MAPK [37]. Additionally, GPER is involved in the transactivation of the EGFR independent of classical ER, and is able to propagate an effect similar to EGF in the presence of E2 [38]. This illustrates the tight interconnection between genomic and non-genomic effects of NRs.

Rapid signaling mediates gene transcription

In addition to the physiological effects activated by direct, non-genomic signal transduction pathways, it is important to emphasize that non-genomic pathways can also lead to genomic effects. As such, many of the reported steroid hormone-mediated genomic effects may actually be in part the indirect result of rapid, non-genomic events (Figure 1). To that end, there are several instances where NR-induced rapid signaling promotes gene expression, suggesting a convergence between classic genomic and rapid, non-genomic signaling pathways. For example, androgen-bound AR associates with the kinase Src at the plasma membrane, activating Src which then leads to a signaling cascade through MAPK/ERK [39]. However, Src can also increase the expression of AR target genes by the ligand-independent transactivation of AR. This can occur by growth factor receptor upregulation of AR coregulators, as well as through MAPK activation of the coregulators or NRs themselves [40,41]. Hence, extranuclear steroid hormone actions can potentially reprogram nuclear NR events. To that end, expression microarray analysis from a transgenic mouse model in which rapid signaling was inhibited revealed that rapid signaling had effects on ER-induced transcriptional changes [42]. Furthermore, this study suggested that estrogen modulated the expression of several genes including endothelial nitric oxide synthase (eNOS) via rapid signaling pathways, an effect dependent on the scaffolding protein striatin that is associated with ER membrane translocation [42]. In another study, it was shown that in the presence of progestin, ERα and PR cooperatively bind to and activate transcription of the PR gene, a classic target gene of ER. Moreover, this process is dependent on rapid signaling via MAPK/ERK and Akt [43]. Additionally, steroid hormones can have epigenetic effects on DNA via non-genomic signaling pathways. For example, E2 can modulate the levels of histone methylation via mER-activated PI3K/Akt signal transduction. These epigenetic changes can then mediate genomic events in uterine tissue and breast cancer cells [10,44].

Non-genomic effects elicited by genomic actions

While much of the discussion of rapid, non-genomic signaling of steroid hormone receptors revolves around their direct downstream effects, it is important to note that many non-genomic signaling events are augmented or initiated following classical genomic NR actions. In this regard, rapid membrane signaling can result from NR-mediated transcription, as expressed genes can in turn activate non-genomic pathways. Growth factor receptors such as IGF-IR, EGFR, and vascular endothelial growth factor receptor (VEGFR) are often downstream transcriptional targets of canonical NR signaling that can promote the progression of tumorigenesis [45,46]. Additionally, since the alteration of calcium homeostasis has a vital role in cancer, genes encoding calcium channels are important targets of steroid hormone receptors that propagate non-genomic signals in the cell. For example, androgen increases the expression of the calcium ion channel transient receptor potential cation channel subfamily M member 8 (TRPM8) via an androgen response element [47]. The TRPM8 channel is implicated in survival and progression of various human cancers; this may be due to its role in modulating intracellular pathways via genomic and non-genomic mechanisms [48]. Calcium flux is also regulated by GPER which upregulates expression of the L-type calcium channel α 1D subunit in endometrial carcinoma [30]. Taken together, hormone-mediated gene activation can result in non-genomic events, further connecting the genomic and non-genomic effects of steroid hormone activity.

Approaches

Often, cellular effects that occur within minutes are assumed to be due to rapid membrane signaling, since transcription of genes typically takes longer. However, it is possible that some genomic effects may begin to occur on this timescale, due to evidence that changes to DNA via chromatin remodeling can occur within minutes as well [6]. This suggests that, perhaps as expected, there can be rapid genomic changes that precede transcription. These early genomic alterations may be the result of rapid, non-genomic effects. Hence, it is important to remember that rapid signaling does not exclude all nuclear events. Therefore, care should be taken when defining non-genomic and genomic (which can include pre-transcriptional and transcriptional changes) events on the basis of time alone. In this regard, further investigation is recommended to confirm true non-genomic/non-nuclear events. There are several approaches commonly used to determine non-genomic events. Since each approach has caveats, a combination of techniques is optimal to demonstrate bona fide non-genomic signaling.

One of the key pieces of evidence needed to prove membrane-associated, non-genomic effects is demonstration of the receptor at the plasma membrane. Typically, this is done using microscopy and/or biochemical fractionation studies. For both approaches, antibody specificity is imperative. This is particularly true for immunofluorescence microscopy where non-specific antibodies can easily lead to the erroneous detection of proteins throughout the cell. To circumvent this problem, many investigators prefer to overexpress tagged versions of their receptor of interest to assess the subcellular location of the tag and therefore protein. However, exogenously expressed proteins may not always be folded in the exact same manner as the endogenous protein. In addition, expression of high, non-physiological levels of any protein may lead to the accumulation of protein in abnormal subcellular locations. Finally, occasionally the tag itself may alter normal subcellular transport. Thus, when possible, assessment of the endogenous protein is preferred. For many steroid hormone receptors, this equates to high levels of nuclear NR levels in the presence of agonist.

Biochemical subcellular fractionation is another commonly used approach that has successfully been used for years to determine the presence of plasma membrane-associated proteins. Crucial to this approach is the obtainment of clean fractions. Purity of fractions must be confirmed with the proper controls. However, it is often difficult to obtain completely pure fractions. As such, what is often evaluated are relative fractions. In addition, the disruption involved in the fractionation process can lead to unexpected membrane structure fusions [49]. Complicating matters is that many of the canonical markers for subcellular fractions can under certain contexts be expressed in other locations; for example, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) is typically used as a cytoplasmic marker, but can occasionally be found at the plasma membrane [50,51].

Another commonly used tool in the field is the use of protein-conjugated steroids. BSA-conjugated steroids have often been used in rapid signaling studies because the large structure prevents hormones from crossing the plasma membrane, thereby ensuring that the hormone exclusively binds to and activates pathways originating from membrane receptors. However, BSA itself may interfere with the biological activity of cell signaling pathways [52]. Further, the conformation of BSA may lead to interference of the lipid rafts at the point of mNR localization [53]. In addition, conjugation of a large protein like BSA to a steroid hormone may limit a hormone’s flexibility or ability to properly move into and out of a binding pocket, while also potentially slowing the process of the hormone binding to its receptor [54]. Importantly, BSA-conjugated estrogen has been shown to activate pathways not normally mediated by estrogen, and it may have its own inherent level of pharmacological activity [54,55]. Also, caveolin-associated BSA has been observed to enter the cell via interaction with caveolin-1 and extracellular BSA [56]. This could potentially provide a mechanism for BSA-conjugated steroid hormones to enter the cell. Hence, BSA alone should at a minimum always be used as a control. Other BSA-conjugated steroids could also be used as another control to confirm the specificity for a particular hormone-mediated event. Alternatively, hormone-linked dendrimers have emerged as a way of circumventing some of the limitations of BSA-conjugated steroids. Dendrimers have greater specificity and binding affinity than BSA and have been conjugated to estradiol using more optimal chemical linkages to study the genomic and non-genomic actions of hormone pathways [57,58]. Currently, both BSA- and dendrimer-conjugated hormones are commonly used for observing membrane-originating signal transduction of steroid hormones [59–61]. It is favorable to confirm the effects of BSA-conjugated steroids with dendrimer-conjugated steroids when possible. In this regard, the development of additional hormone-dendrimer conjugates will undoubtedly benefit the field.

New molecular and genetic tools are emerging that should greatly improve our understanding of the role of NRs at different subcellular sites. To that end, the use of genetically engineered NRs that can be specifically localized to or omitted from various cellular locations (ex. inclusion of plasma membrane localization sequence or deletion of a nuclear localization sequence (NLS)) are now being applied to address the role of membrane NRs in diverse biologies [62–65]. For example, mouse models engineered to express ERα selectively at the plasma membrane (membrane only ER, MOER) or nucleus (nuclear only ER, NOER) were created to study rapid signaling events under more native contexts [64,66]. In MOER mice, estradiol can activate various cellular pathways through rapid signaling independent of nuclear ERα activity. In fact, membrane ER was required for normal cellular function by interaction with the nuclear store of ERα [66]. Additionally, NOER mice demonstrate that nuclear localization of ERα effectively blocks rapid signaling by membrane ERα. Hence, these models have demonstrated the important dependency that one subcellular ER pool can have for the other. So while initial use of these engineered NRs came with some of the same caveats of exogenously overexpressing NRs described above, particularly in cell types that no longer expressed a specific NR and therefore evolved to function without it, new genetic knock-in approaches can provide a much more physiological setting. Such an approach also helps to overcome situations in which the predominant NR species masks the effects of the more limited membrane-localized pool of the receptor.

Conclusions

Rapid signaling via steroid hormones has been extensively studied. It is evident that there are important roles for the non-genomic actions of steroids in cell physiology and disease. In particular, membrane localization is important for rapid signal transduction implicated in cancer, cardiovascular disease, neuronal function and metabolic disease [67–69]. However, there are still unknown aspects of membrane-localized nuclear receptor structure (ex. How do they scaffold?) and function (ex. What specific signaling cascades do they regulate and how context dependent are these events?). While not discussed here, there have also been reports of NRs at other subcellular locations, including the mitochondria and endoplasmic reticulum [70–73]. Therefore, while we have largely focused on NR at the plasma membrane, there are likely principles that can be applied to the NRs at other sites. The process of elucidating the various NR pathways is complicated by the crosstalk between genomic and non-genomic NR signaling as many genomic activities are both regulated by non-genomic signaling and can subsequently influence non-genomic pathways. Importantly, because some steroid hormone chromatin modifications can occur within minutes, it is critical to not assume that rapid signaling equates to only non-genomic events. Likewise, non-genomic events may have sustained effects that resemble classical NR actions. Hence, it is essential that new genomic and non-genomic actions of steroid hormones be validated using orthogonal approaches in an effort to circumvent inherent weaknesses that exist in many experimental techniques. Moving forward, the advent of new genetic engineering approaches will undoubtedly aid in dissecting out the exact contributions of distinct subcellular pools of NRs in a more endogenous setting.