|Home | About | Journals | Submit | Contact Us | Français|
The human glucocorticoid receptor (GR) gene expresses two splicing isoforms α and β through alternative use of specific exons 9α and 9α. In contrast to the classic receptor GRα, which mediates most of the known actions of glucocorticoids, the functions of GRβ have been largely unexplored. Owing to newly developed methods, such as microarrays and the jellyfish fluorescence proteins, we and others have recently revealed novel functions of GRβ. Indeed, this enigmatic GR isoform influences positively and negatively the transcriptional activity of large subsets of genes, most of which are not responsive to glucocorticoids, in addition to its well-known dominant negative effect against GRα-mediated transcriptional activity. A recent report suggested that the “ligand-binding domain” of GRβ is active, forming a functional ligand-binding pocket, associated with the synthetic compound RU 486. In this review, we discuss the functions of GRβ, its mechanisms of action, and its pathologic implications.
Glucocorticoids, the end-products of the hypothalamic-pituitary adrenal axis, are steroid hormones crucial for the regulation of basal and stress-related homeostasis [1,2]. Glucocorticoids are also essential for the proper functioning of virtually all organs and tissues of the organism, including the central nervous (CNS) and cardiovascular systems, metabolic organs, such as the liver and adipose tissue, as well as the immune/inflammatory response [3,4]. In addition, glucocorticoids at “pharmacologic” or “stress-related” doses are irreplaceable therapeutic means for many allergic, inflammatory, autoimmune, and lymphoproliferative diseases .
The actions of glucocorticoids are mediated by a ubiquitous intracellular receptor protein, the glucocorticoid receptor (GR), which functions as a hormone-activated transcription factor of glucocorticoid target genes [5,6]. The human GR gene located in chromosome 5 and encodes two splicing variants GRα and GRβ from alternative use of a different terminal exon 9α and 9β [5,7]. GRα is the classic receptor, binding to glucocorticoids and mediating most of the known glucocorticoid actions . In contrast, GRβ does not bind glucocorticoids but functions as a dominant negative inhibitor of GRα-induced transactivation of GRE-containing, glucocorticoid-responsive promoters; its physiologic/pathologic roles have not been well elucidated as yet [8,9].
Using the microarray technique, which enabled us to evaluate gene expression en masse, we and others recently found that the GRβ isoform has intrinsic, GRα-independent transcriptional activity, in addition to its well-known dominant negative effect on GRα [10,11]. In this review article, we will summarize known GRβ activities and discuss newly identified actions of this GR isoform.
The GR, also called as nuclear receptor superfamily 3, group C, member 1 (NR3C1), belongs to the steroid/sterol/thyroid/retinoid/orphan nuclear receptor superfamily, which consists of over 130 members preserved from the early metazoans to humans [5,12]. The human GR gene, located in the short arm of chromosome 5 (5q31.3), consists of 9 exons, and its expression is regulated by at least three different promoters (A, B and C) [7,13], with promoter A alternatively used with three unique promoter fragments 1A1, 1A2 and 1A3 . Thus the GR gene can produce five different transcripts from different promoters that encode the same GR proteins. In addition to alternative transcripts using the 5’ different promoters, the GR gene generates two 3’ splicing variant transcripts with alternative use of exon 9α and/or 9β (Figure 1). Thus, the GR gene generates 10 different transcripts that encode two protein molecules GRα and GRβ.
Recently, it became evident that the GRα variant mRNA is translated from at least 8 initiation sites into multiple GRα isoforms termed A through D (A, B, C1–C3 and D1–D3), producing different amino terminal isoforms with distinct specific transcriptional activities on glucocorticoid-responsive genes  (Figure 1). These GR molecules are also differentially expressed in several different cell lines and tissues . Given that GRα and GRβ share a common mRNA domain that contains the same translation initiation sites , it appears that the GRβ variant mRNA is also translated through the same initiation sites to a similar host of 8 β isoforms  (Figure 1).
GRα, the classic glucocorticoid receptor is ubiquitously expressed and mediates most of the known actions of glucocorticoids [3,5]. The human GRα consists of 777 amino acids and has 3 major distinct functional domains, the N-terminal or immunogenic domain (NTD), the DNA-binding domain (DBD) and the ligand-binding domain (LBD)  (Figure 1). The LBD of GRα consists of 12 α-helices and 4 β-sheets, among which helices 3, 4, 11 and 12 form the ligand-binding pocket for binding to glucocorticoids [16–18] (Figure 2). GRα is located primarily in the cytoplasm in the absence of glucocorticoid ligand, as part of hetero-oligomeric complexes containing heat shock proteins (HSPs) 90, 70, 50, 20 and, possibly, other proteins as well [5,6] (Figure 3). After binding to its agonist ligand, GRα undergoes conformational changes, dissociates from the heat shock proteins, homo-dimerizes, and translocates as a monomer or dimer into the nucleus through the nuclear pore, via an active ATP-dependent process mediated by its nuclear localization signals (NL)-1 and -2 [12,19]. NL-1 is located in the junction of DBD and the hinge region, while NL-2 spans the entire LBD  (Figure 1).
Inside the nucleus, the ligand-activated GRα directly interacts as a dimer with specific DNA sequences, the glucocorticoid response elements (GREs), in the promoter regions of target genes, or as a monomer or dimer with other transcription factors via protein-protein interactions, indirectly influencing the activity of the latter on their own target genes [5,12] (Figure 3). GR contains two transactivation domains, activation function (AF)-1 and -2, located at its NTD and LBD, respectively, through which the GR interacts with many proteins and protein complexes, such as the nuclear receptor coactivator [p160, p300/CREB-binding protein (CBP) and p300/CBP-associated factor (p/CAF)] complexes and the SWI/SNF and vitamin D receptor-interacting protein/thyroid hormone receptor-associated protein (DRIP/TRAP) chromatin-remodeling complexes, eventually influencing the activity of RNA polymerase II and its ancillary factors, altering the transcription rates of glucocorticoid-responsive genes [5,6,20] (Figure 1).
GR also interacts with the nuclear receptor corepressor (NCoR) and its homolog silencing mediator of retinoic acid and thyroid hormone receptor (SMRT), which are macromolecular docking platforms for nuclear receptors and many transcription factors, repressing the transcriptional activity of the GR by attracting histone deacetylase/Sin3 complexes . The p160 type coactivators and the NCoR/SMRT type corepressors establish equilibrium in their interaction with the GR to respectively facilitate or block its transcriptional activity . Accumulation of coactivators and corepressors on the promoter-bound GR is dependent on the kind of ligands bound to the GR: agonist glucocorticoids attract the coactivator complexes to the promoter-bound GR, while antagonists, like RU 486, accumulate the corepressor complexes  (Figure 2).
In addition to transactivation of the glucocorticoid-responsive genes explained above, GRα modulates other signal transduction cascades through mutual protein-protein interactions with specific transcription factors, by influencing their ability to stimulate or inhibit the transcription rates of their respective target genes (Figure 3). This activity may be more important than the GRE-mediated one, granted that mice harboring a mutant GRα, which is active in terms of protein-protein interactions but inactive in terms of transactivation via DNA GREs, survive and procreate, in contrast to mice with a deletion of the entire GR gene that die immediately after birth from severe respiratory distress syndrome [23,24]. The former mouse model and additional in vitroresults indicate that GR interacts with and influences other transcription factors primarily as a monomer [23,25].
The protein-protein interactions of GRα with other transcription factors may take place on promoters that do not contain GREs (tethering mechanism), as well as on promoters that have both GRE(s) and responsive element(s) of transcription factors that interact with GRα (“composite promoters”) . Suppression of transactivation of other transcription factors through protein-protein interactions may be particularly important in the suppression of immune function and inflammation by glucocorticoids [23,25]. A substantial part of the effects of glucocorticoids on the immune system may be explained by the interaction between GRα with nuclear factor-κB (NF-κB), activator protein-1 (AP-1) and probably the signal transducers and activators of transcription (STATs) [27–30].
In addition to co-regulators and other transcription factors that modulate GR-induced transcriptional activity, several distinct signaling pathways regulate the transcriptional activity of the GR via post-translational modifications of the receptor protein . These include methylation, acetylation, nitrosylation, sumoylation and ubiquitination, as well as phosphorylation, which has been studied best. Indeed, several kinases, such as the cell-cycle-related kinases, mitogen-activated kinases and the glycogen synthase kinases, phosphorylate specific serine or threonine residues of the GR. Interestingly, the majority of these residues are located in the AF-1 domain of the human GR NTD, thus phosphorylation of some or all of them modulates GR-induced transcriptional activity through alteration of co-regulator attraction to the promoter region of glucocorticoid-responsive genes, possibly by changing their affinity to the AF-1 domain of GR .
Similarly to the classic human GRα, the original human GRβ isoform is also ubiquitously expressed in most tissues. This isoform was identified in both the zebrafish and humans, but not in mice [15,32,33]. The human (h) GRβ contains 742 amino acids and shares the first 727 amino acids from the N-terminus with hGRα [6,15] (Figure 1). hGRβ encodes an additional 15 nonhomologous amino acids in the C-terminus, while hGRα possesses an additional 50 amino acids forming a 777 amino acid protein [6,15] (Figure 1). Therefore, hGRβ shares the same NTD and DBD with hGRα, but has a unique “LBD”. Since the divergence point (amino acid 727) is located at the C-terminal end of the helix 10 in the hGRα LBD, the hGRβ “LBD” does not have the helices 11 and 12 of the hGRα. As these helices are important for forming the ligand-binding pocket and for the creation of the AF-2 surface upon ligand binding  (Figure 2), GRβ cannot form an active ligand-binding pocket, does not bind glucocorticoids, and thus, does not directly regulate GRE-containing, glucocorticoid-responsive gene promoters. In the absence of the hGRβ “LBD”, the truncated hGR consisting of NTD and DBD is transcriptionally active on GRE-containing promoters , thus the hGRβ “LBD” somehow attenuates the transcriptional activity of the other subdomains of the molecule on GRE-driven promoters. Inside the cells, hGRβ can localize both in the cytoplasm and nucleus [9,35].
Similarly to the human GR gene, the zebrafish (z) GR gene consists of 9 exons and produces the zGRα and zGRβ proteins, which respectively contain 746 and 737 amino acids  (Figure 4). zGRα and zGRβ share the N-terminal 697 amino acids, while they have specific C-terminal portions, which contain 47 and 40 amino acids, respectively. In contrast to hGRα and hGRβ, which are produced through alternative use of specific exon 9α and 9β, zGRα and zGRβ are formed as a result of intron retention . zGRα and zGRβ use exon 1 to exon 8 for their common N-terminal 697 amino acids. zGRα uses exon 9 for its specific C-terminal portion, while zGRβ continuously employs the rest of exon 8 and uses a stop codon located at 3’ portion of this exon to express its specific C-terminal peptide  (Figure 4). Protein alignment comparison of hGRβ and zGRβ indicated that these two molecules employ exactly the same divergence point, while their β isoform-specific C-terminal peptides show little sequence homology . These pieces of molecular information indicate that hGRβ and zGRβ evolved independently. Nevertheless, zGRβ demonstrated the same functional properties as those of the hGRβ, such as inability to bind glucocorticoids, a dominant negative activity on zGRα transcriptional activity on GRE-drive promoters, and strikingly similar tissue distribution as hGRβ . Thus, hGRβ and zGRβ are produced through convergent evolution, most likely developed through strong requirement of this type of GR isoform in a physiologic situation.
The presence of nonligand-binding C-terminal variants is not unique to the GR. Similarly to the human and zebrafish GR, several other human steroid and nuclear receptors, such as the estrogen receptor β (ERβ), thyroid hormone receptor α (TRα), vitamin D receptor, constitutive androstane receptor (CAR), dosage-sensitive sex reversal-1 (DAX-1), nuclear receptor related2 (Nurr2), neuron-derived orphan receptor-2 (NOR-2), peroxisome proliferator-activated receptor α (PPARα), and PPARγ, also have C-terminally truncated receptor isoforms defective in binding to cognate ligands with dominant negative activity on their corresponding classic receptors [36–45]. This suggests that evolution has allowed the development and retention of such alternative nuclear receptors, probably because they play useful biologic roles.
In contrast to GRα that has numerous and diverse actions , the functions of GRβ had not been revealed until we reported its dominant negative effect on GRα-induced transcriptional activity almost a decade after the original identification of this receptor isoform . The dominant negative activity of GRβ was first demonstrated in transient transfection-based reporter assays using GRE-driven reporter genes, but was subsequently confirmed on endogenous, glucocorticoid-responsive genes, such as the mitogen-activated protein kinase phosphatase-1 (MPK-1), myocilin and fibronectin [46,47]. Further, GRβ was shown to attenuate glucocorticoid-induced repression of the tumor necrosis (TNF) α and interleukin (IL)-6 genes . We also confirmed this negative effect of GRβ on GRα-mediated transrepression using microarray analyses . Several mechanisms explaining this GRβ function have been reported, including (1) competition for GRE binding through their shared DBD, (2) heterodimerization with GRα and (3) coactivator squelching through the preserved AF-1 domain [8,34,48]. All these different mechanisms of action appear to be functional, depending on the promoters and tissues affected by this GR isoform.
Several clinically oriented investigations suggest that GRβ is responsible for the development of tissue-specific insensitivity to glucocorticoids in various disorders, most of them associated with dysregulation of immune function. They include glucocorticoid-resistant asthma, rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), ankylosing spondylitis, chronic lymphocytic leukemia and nasal polyps [49–55]. In these studies, various immune cells expressed elevated levels of GRβ, which correlated with reduced sensitivity to glucocorticoids. Elevated levels of pro-inflammatory cytokines, such as IL-1, -2, -4, -7, -8 and -18, TNFα, and interferons α and γ, might have been responsible for increased GRβ expression in cells from patients with these pathologic conditions, as these cytokines experimentally stimulated the expression of GRβ in lymphocytes, neutrophils or airway smooth muscle cells [56–61]. Further, presence of a single nucleoside polymorphism in the 3’ untranslated region of the hGRβ mRNA (rs6198G allele), which increases the stability of the mRNA, and thus, causes elevated expression of GRβ protein, was associated with increased incidence of RA, SLE, high blood pressure, ischemic heart diseases and nasal carriage of Staphylococcus aureus [50,62–64], possibly through inhibition of glucocorticoid actions by increased concentrations of GRβ. These pieces of clinical evidence further support that GRβ has dominant negative activity on GRα-induced transcription inside the human body, functioning as a negative regulator of glucocorticoid actions in local tissues.
We and others recently performed transcriptome analyses using microarray techniques in cultured cells overexpressing GRβ, and found that these cells had a distinct mRNA expression profile compared to cells not overexpressing GRβ and those expressing GRα and treated with glucocorticoids [10,11]. In a subsequent real-time PCR analysis, we also confirmed that GRβ regulates mRNA expression positively and negatively in a gene-specific fashion . These results indicate that GRβ has intrinsic transcriptional activities independent of the activity of its isoform GRα. We have compared the microarray results obtained by us and those of others [10,11], and found that the 2 studies share 78 genes modulated by overexpression of GRβ (Table 1). Specifically, 29 out of 78 genes were both down-regulated by GRβ overexpression, while only 8 were up-regulated. Interestingly, 41 genes showed opposite response to GRβ between the two studies, suggesting that GRβ modulates mRNA expression of some of its responsive genes in a cell-, and possibly, a cell culture condition-specific fashion.
Apparently, this intrinsic transcriptional activity of GRβ is not mediated by binding of the isoform to classic GREs, as GRβ does not influence the transcriptional activity of classic GREs-driven promoters, while the promoter region of the genes which we identified to be regulated by GRβ, do not contain GRE sequences . Rather, GRβ directly modulates the transcriptional activity of its responsive genes, which are distinct from those responsive to glucocorticoids, possibly by altering the activity of transcriptional intermediate molecules or other transcription factors through physical protein-protein interactions. Indeed, we previously demonstrated that the AF-1 of the GRβ, which presumably keeps the same protein structure and function as that of the GRα, is transcriptionally active, contributing to its dominant negative activity against GRα-induced transactivation . This transactivation domain of GRα interacts with numerous cofactor molecules, including CBP/p300 and p160-type histone acetyltransferase coactivators, components of the SWI/SNF chromatin modulators, DRIP150 of the DRIP/TRAP complex and the steroid receptor RNA coactivator (SRA) [65–70]. Thus, it is possible that GRβ alters the transcriptional activity of its responsive genes by lodging into the transcriptional complexes formed on their promoter region through its AF-1 (Figure 5). This mechanistic hypothesis is further supported by the recent results from other groups, in which GRβ was shown to repress the transcriptional activity of AP-1 and NFκB, possibly through protein-protein interactions similar to those between GRα and these transcription factors .
GRβ was also reported to suppress the transcriptional activity of the GATA3 transcription factor on its responsive IL-5 and -13 promoters by attracting histone deacetylases . Alternatively, GRβ might bind DNA sequences unique to this isoform through its DBD, regulating transcription through hypothetical “GRβ REs” (Figure 5). Since the subdomains of steroid hormone receptors influence each other’s activity [73,74], the unique GRβ “LBD” might alter the binding specificity of its DBD to DNA and allow it to recognize a set of DNA sequences specific to GRβ and distinct from those of GRα.
The importance and exact roles of this intrinsic transcriptional activity of the GRβ isoform in physiology and pathophysiology have not been elucidated as yet. We have performed a pathway analysis of our microarray results to define the biologic pathways where GRβ might play consistent roles , and found that this GR isoform may be involved in regulation of 43 distinct pathways recorded in the Kyoto Encyclopedia of Genes and Genomes (KEGG) (Table 2). Among the pathways we found in this analysis, GRβ might strongly influence several cellular functions, such as cell communications (#13), focal adhesion (#26), ECM-receptor interaction (#27), expression of cell adhesion molecules (#28) and regulation of actin cytoskeleton (#34), as well as metabolism of some animo acids and other bioactive molecules. Interestingly, GRβ might also play a role in the development/activity/apoptosis of cancer cells, as it also regulates mRNA expression of genes important for colorectal (#40), renal cell (#41), prostate (#42), and small cell lung cancer (#43) and apoptosis (#23). To further verify the biologic pathways, in which GRβ plays important roles, development of the mice conditionally overexpressing human GRβ would be very helpful.
A previous publication demonstrated that only RU 486 among 57 native and synthetic steroids tested bound GRβ weakly at the “ligand-binding” pocket of the GRβ and slowly (over 6 hours for completion) induced its nuclear translocation . The results were supported by a nuclear translocation study using fluorescent protein-fused GRβ by scoring cellular localization of this fusion protein in different cells, by a whole-cell ligand-binding assay followed by the crude fractionation of radiolabeled ligand-associated receptors with a Sephadex column and by computer-based modeling of the GRβ “ligand-binding domain” associated with several steroids . This report also demonstrated that RU 486 modulated GRβ-mediated transcriptional activity in microarray analysis . Although the hypothesis presented in this publication is interesting, there are several points to be resolved. Yet undiscovered endogenous steroids or other related compounds with structures similar to that of RU 486 would be expected to be the endogenous ligands of GRβ. Crystallographic structural analysis of the GRβ “LBD” might help identifying a “ligand-binding pocket” in the GRβ “LBD” and hence its binding to RU 486. The cytoplasmic to nuclear translocation of GRβ demonstrated by the previous report was quite slow compared to that of GRα: in the former, the receptor took 6 hours to complete its translocation, while in the latter it did this within minutes . GRβ and GRα share NL-1, which mediates the rapid nuclear translocation of GRα, while GRβ does not appear to have NL-2, which is dependent on the entire LBD of GRα, and causes slower nuclear translocation of the receptor . Thus, presence of yet unknown regulators specific to GRβ might be involved in the nuclear translocation of this isoform.
We independently performed several experiments addressing the potential activation of GRβ by RU 486, its subcellular localization and cytoplasmic to nuclear translocation. In contrast to the previously reported findings , the green fluorescent protein-fused GRβ was mainly localized in the nucleus in HeLa cells stably expressing this fusion protein, while it was heterogeneously distributed both in the cytoplasm and the nucleus in HCT116 cells that expressed the GRβ fusion protein transiently: some cells mainly expressed GRβ in the nucleus, while others had it in the cytoplasm . Addition of RU 486 did not stimulate the transcriptional activity of glucocorticoid-responsive and GRE-containing mouse mammary tumor virus promoter in transiently GRβ-expressing HCT116 cells, and did not induce cytoplasmic to nuclear translocation of this isoform . The inconsistency of our results with the previous report may have been caused by use of different experimental systems, such as cell lines and plasmids. This discrepancy suggests that the mechanisms of the regulatory actions of GRβ on the transcription of responsive genes inside the cells are quite complex.
In 1995, ten years after the original identification of the human GRβ by R. Evans’ group , we reported that GRβ had a dominant negative effect on GRα-induced transcriptional activity, an effect that was replicated a year later [8,35]. After another decade, a new activity of GRβ, namely an intrinsic, GRα-independent transcriptional activity, was discovered by employing microarray-based transcriptome analyses [10,11]. In spite of a continuous effort spanning over 20 years, the molecular mechanisms of action and the roles of GRβ in physiology are still largely unknown, in contrast to those of the classic, glucocorticoid action-mediating GRα. The β isoform cannot modulate the transcriptional activity of GRE-containing promoters in the absence of GRα, even though it shares a perfect DBD with GRα . Lack of GRβ in rodents stands against the elucidation of its in vivoactivity . We hope that physiologic and pathologic roles of GRβ will be further clarified with future technical progress, such as development of mice conditionally expressing human GRβ, sophisticated transcriptome/promoter/proteome analyses with array techniques, evaluation of GRβ subcellular circulation/localization through fusion with fluorescent proteins and crystallography-based structural analyses.
Literary work of this article was funded partly by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD.