Intracellular Glucocorticoid Signaling: A Formerly Simple System Turns Stochastic

See allHide authors and affiliations

Science's STKE  04 Oct 2005:
Vol. 2005, Issue 304, pp. pe48
DOI: 10.1126/stke.3042005pe48


Glucocorticoids contribute fundamentally to the maintenance of basal and stress-related homeostasis in all higher organisms. The major roles of these steroids in physiology are amply matched by their remarkable contributions to pathology. Glucocorticoids influence about 20% of the expressed human genome, and their effects spare almost no organs or tissues. For many years we thought that the numerous actions of glucocorticoids were mediated by a single receptor molecule: the classic glucocorticoid receptor (GR) isoform α, a complex, multifunctional domain protein, operating as a ligand-dependent transcription factor. The GR gene, however, encodes two 3′ splicing variants, GRα and GRβ, from alternative use of two distinct terminal exons (9α and 9β), and each variant mRNA is translated from at least eight initiation sites into multiple GRα and possibly GRβ isoforms, amounting to a minimum of 16 GR monomers and 256 different homo- or heterodimers. The translational GRα isoforms may be produced variably in target tissues, have varying intrinsic transcriptional activities, and influence different complements of glucocorticoid-responsive genes. It is likely that expression and functional differences might also be present between the putative GRβ translational isoforms. The presence of multiple GR monomers and dimers in different quantities with quantitatively and qualitatively different transcriptional activities suggests that the glucocorticoid signaling system is highly stochastic.

Glucocorticoids are secreted by the adrenal glands in a circadian and stress-related fashion (1). These hormones affect nearly every organ and tissue in the body and have diverse life-sustaining effects throughout the life span. They influence the activity and direction of the reactions underlying intermediary metabolism, the maintenance of proper cardiovascular tone, the activity and character of immune and inflammatory reactions, and many functions of the central nervous system, including arousal, cognition, mood, and sleep. Physiologic amounts of glucocorticoids are also essential for normal renal tubular function and thus for water and electrolyte homeostasis.

Glucocorticoids readily penetrate the cell membrane and interact with ubiquitous cytoplasmic and nuclear glucocorticoid receptors (GRs), through which they exert their effects (2). Using DNA microarray analysis, we found that expression of about 20% of the expressed human leukocyte genome was positively or negatively affected by glucocorticoids (3). This percentage is many times the fraction of genes whose expression changes during the transformation of a normal cell into a tumor cell. Even in a single cell type, the expression of a wide array of genes, which served diverse functions, changed. Glucocorticoids are involved in nearly every molecular, cellular, and physiologic network of the organism and play a pivotal role in the unfolding of critical biological processes that make use of several such networks synchronously or in tandem (1, 4). Such vital processes include the behavioral and physiological response to stress, the immune and inflammatory reaction, the sickness syndrome ("nonspecific" responses to disease), and the process of sleep, as well as basic functions such as growth and reproduction (4, 5).

The ubiquitous presence of glucocorticoids, the rapid advances in our general knowledge of the human genome, and our ever-increasing understanding of cell biology (especially cell signaling systems) add to the complexity of their signal transduction and biologic effects and also allow us to view the actions of these hormones and their involvement in human physiology and pathophysiology in a more integrated way. A recent provocative study by Lu and Cidlowski (6) has revealed cell-specific expression of multiple translational GR isoforms, each with quantitatively and qualitatively distinct genomic effects. This new information has further compounded the complexity of our understanding of the regulation of glucocorticoid actions in their target tissues and, hence, of the involvement of these hormones in physiology and pathophysiology.

The Glucocorticoid Receptor Gene, Its Products, and Their Actions

For a little less than half a century, we thought that the diverse actions of glucocorticoids were mediated by a single receptor protein (7). Then we realized that the human GR gene, located on chromosome 5, is regulated by several promoters and encodes at least two 3′ splicing variants, GRα and GRβ, through the alternative use of two distinct terminal exons, 9α and 9β (8, 9) (Fig. 1, top). Lu and Cidlowski showed that the GRα variant mRNA was translated from at least eight initiation sites into multiple GRα isoforms termed GRα-A through GRα-D (A, B, C1 to C3, and D1 to D3) (6) (Fig. 1, middle). Given that GRα and GRβ share a common mRNA domain that contains the same translation initiation sites (7), we hypothesize that the GRβ variant mRNA is translated through these same initiation sites to give rise to a similar host of β isoforms (Fig. 1, middle). The classic GR (now GRα-A) encodes a 777–amino acid protein, whereas the corresponding GR isoform β (now GRβ-A) contains 742 amino acids. The first 727 amino acids from the N terminus are identical in the two A isoforms (10). GRα-A possesses an additional 50 amino acids, whereas GRβ-A encodes an additional 15 nonhomologous amino acids in the C terminus (10). Neutralizing mutations of the GR gene that result in functional alterations of GRα transcriptional activity have been associated with the genetic syndrome of familial or sporadic glucocorticoid resistance (11).

Fig. 1.

Genomic and complementary DNA, protein structures, and functional domains of human GR isoforms. The human GR gene consists of 10 exons. Exon 1 is an untranslated region, exon 2 encodes the N-terminal "immunogenic" domain, exons 3 and 4 encode the DNA binding domain, and exons 5 through 9 encode the hinge region and the ligand-binding domain. The GR gene contains two terminal exon 9s (9α and 9β), which are alternatively spliced to produce the classic GRα (GRα-A) and the non–ligand-binding GRβ-A, which exerts dominant negative effects upon GRα (GRα-A). C-terminal domains colored as light green and yellow in GRαs and GRβs show unique portions of their amino acid sequences. GRα N-terminal translational isoforms expressed from a single GRα transcript are shown in the middle of the figure. The GRβ transcript may also produce similar N-terminal isoforms from the same start sites as GRα. AF-1 and -2, activation function 1 and 2; DBD, DNA binding domain; HR, hinge region; LBD, ligand-binding domain; NL1 and 2, nuclear translocation signal 1 and 2. [Data from (1, 6, 9, 16, 17, 1921, 32)]

For more than two decades, we have known that the classic receptor has three major distinct functional domains—the N-terminal or immunogenic domain, the DNA binding domain (DBD), and the ligand-binding domain (LBD)—as well as several subdomains with specific roles (Fig. 1, bottom). In its unliganded (but ligand-friendly) state, GRα is located primarily in the cytoplasm as part of a hetero-oligomeric complex that contains heat shock proteins (HSPs) 90, 70, 50, and 20 and, possibly, other proteins as well (11) (Fig. 2). After binding to an agonist ligand, GR undergoes conformational changes, dissociates from the HSPs, partly homodimerizes, and translocates into the nucleus through a nuclear pore as a monomer or dimer by means of an active adenosine triphosphate–dependent process mediated by its nuclear localization signals NL1 and NL2 (11). Once in the nucleus, ligand-activated GRα dimers interact directly with specific DNA sequences in the promoter regions of target genes [the glucocorticoid response elements (GREs)]. Ligand-activated GRα monomers, however, interact with other transcription factors [nuclear factor κB (NF-κB), activating protein 1 (AP-1), p53, CRE-binding protein (CREB), signal transducer and activator of transcription 5 (STAT5), and others] through protein-protein interactions, indirectly influencing the activity of the latter on their own target genes (11). GR contains two transactivation domains [activation function 1 (AF-1) and AF-2] at its N-terminal and ligand-binding domain, respectively. GR interacts through AF-1 and AF-2 with various proteins and protein complexes, including the nuclear receptor coactivator complexes [p160, p300/CREB-binding protein (CBP), and p300/CBP-associated factor (p/CAF)] 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 and thereby altering the transcription rates of glucocorticoid-responsive genes (11) (Fig. 2B).

Fig. 2.

A current view of the basic glucocorticoid signaling system. (A) Intracellular GR activation. 11β-HSD1/2 regulates the conversion of the inactive cortisone (white circles) to the active GR ligand cortisol (black circles). After an agonist such as cortisol binds to the cytoplasmic HSP-bound GR, GR (blue rectangles, GRα N-terminal isoforms; beige rectangles, GRβ N-terminal isoforms (GRβ N-terminal isoforms do not bind glucocorticoids) is released from the HSP and translocates into the nucleus. There, as a dimer, GR interacts with DNA target sequences and glucocorticoid response elements (GREs), as well as with nuclear receptor coactivator and other chromatin-remodeling complexes and components of the general transcriptional machinery [see (B)]. Alternatively, GR monomers interact with and alter the activity of a host of transcription factors (dark green oval) such as NF-κB, AP1, and STAT5. GR nucleocytoplasmic shuttling and transcriptional activity are regulated through phosphorylation by p38 MAPK, JNK, ERK, and CDKs as well as through interactions with 14-3-3 proteins and HSPs. Both nuclear and cytoplasmic GR are degraded through the ubiquitin-proteasomal pathway. We hypothesize that there are at least 16 GRα and GRβ monomers and 256 homo- or heterodimers that may participate in intracellular glucocorticoid signaling. [Data from (1, 6, 11, 1618, 20)] B GR transcriptional effects. Dynamic, rapid (subsecond) intranuclear protein-protein and protein-DNA interactions of ligand-bound GRα isoforms with target GREs, nuclear receptor coactivator complexes [p160, p300 (also known as CREB-binding protein or CBP), and p/CAF (p300/CBP-associated factor)], and other chromatin-remodeling complexes (SWI-SNF and DRIP-TRAP), and components of the transcriptional machinery (RNA polymerase II and its ancillary factors) and the proteasome. These interactants define the chromatin residence time of the receptor dimer, and hence its transcriptional activity. Heterodimers containing GRβ isoforms have diminished transcriptional activity, possibly because the GRβ "LBD" does not interact with p160 coactivators and therefore has no GR activating function 2. [Data from (1, 18, 33)]

Although the scheme of GR action has remained basically the same, adjustments need to be made to include the GR isoform β and all of the translational α or β isoforms suggested by Lu and Cidlowski (Fig. 2). GRβ-A does not itself bind glucocorticoids, but rather exerts dominant negative effects on GRα through several mechanisms, such as heterodimerization and competition with GRα for GREs or transcriptional nuclear receptor coactivators (or both) (913). Lu and Cidlowski found that GRα translational isoforms were differentially expressed in various cell lines (6). They were produced by ribosomal leaky scanning or ribosomal shunting (or both) from alternative translation initiation sites located at amino acids 27 (GRα-B), 86 (GRα-C1), 90 (GRα-C2), 98 (GRα-C3), 316 (GRα-D1), 331 (GRα-D2), and 336 (GRα-D3), all located closer to the C terminus than to the classic translation start site (amino acid 1 for GRα-A) (6) (Fig. 1). Thus, they have N-terminal domains of different length but the same DBDs and LBDs. Relative to GRα-A, GRα-C2 and GRα-C3 had greater transcriptional activities on a synthetic GRE-driven promoter, whereas GRα-D1, GRα-D2, and GRα-D3 had weaker activities (6). GRα-B and GRα-C1, however, had transcriptional activities similar to that of GRα-A (6). All of the GRα isoforms translocated into the nucleus in response to ligand, whereas they were differentially distributed in the cytoplasm and the nucleus in the absence of ligand. These isoforms displayed distinct transactivation or transrepression patterns on global gene expression examined by cDNA microarray analyses using stable cell lines that expressed different GRα isoforms (6). Thus, these N-terminal GR isoforms may differentially transduce glucocorticoid hormone signals, depending on their inherent activities and relative expression in different tissues.

It is likely that differential cell-specific production and functional differences similar to those of the GRα translational isoforms could be present among the putative GRβ translational isoforms as well. Recent cDNA microarray studies showed that stably transfected GRβ-A altered the expression of a unique complement of genes with little overlap with those altered by GRα-A (14).

Further adjustments to depictions of the GR signaling cascade in target tissues are required to show the further complexities of the system (Fig. 2). For example, ligand availability is regulated both by local 11β-hydroxysteroid dehydrogenase 1 and 2 (11β-HSD1/2), which respectively convert inactive cortisone to active cortisol or vice versa, and by extrusion of ligand from the cell through the actions of an active export pump (1, 15). Phosphorylation of GR by serine kinases, and interaction of the receptor with 14-3-3 proteins, alter nucleocytoplasmic shuttling and thereby the transcriptional activity of the receptor (16, 17). Finally, the ubiquitin-proteasome–mediated degradation pathway regulates the glucocorticoid signaling system by controlling the degradation rates of GR isoforms and other molecules in the cascade (18). This is particularly pertinent for the regulation of the transcriptional activity of activated GRs at the level of the transcriptosome.

In conclusion, it appears that there are multiple mechanisms through which target cells alter the sensitivity and specificity of the response to glucocorticoids. These take place at the levels of GR gene transcription, mRNA splicing, and mRNA translation, as well as through posttranslational modifications and the inherent functional activity of the expressed isoform monomer(s) or dimer(s) on responsive genes. The data of Lu and Cidlowski suggest that the N-terminal domain of GR substantially contributes to this regulatory diversity.

The N-Terminal Domain of GR: The Largest Domain Has the Fewest Known Functions

AF-1 is located in the portion of the N-terminal domain of GR enclosed by amino acids 77 and 262 (10) (Fig. 1, bottom). Thus, GRα-B contains the entire AF-1, whereas GRα-C1, GRα-C2, and GRα-C3 only contain the C-terminal parts of this subdomain (6) and GRα-D1, GRα-D2 and GRα-D3 do not have an AF-1 (6). Thus, the reduced transactivational activity of the GRα-Ds may result from their lack of the entire AF-1 domain. Because these GRα N-terminal isoforms also exhibit different transrepressive as well as transactivation activities (6), it is likely that the N-terminal domain of GRα has other important functions in addition to its AF-1 activity. The N-terminal domain contains a number of phosphorylation sites of potential functional importance and a novel site that interacts with the β component of the heterotrimeric guanine nucleotide-binding protein (G protein) complex (Fig. 1, middle and bottom).

The extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38 mitogen-activated protein kinase (p38 MAPK) all phosphorylate serines located at amino acids 211 and 226, whereas several cyclin-dependent kinases (CDKs) phosphorylate serines at amino acids 203 and 211, thereby enhancing or decreasing GR transcriptional activity (16, 1921).

Yeast two-hybrid screening experiments using different portions of the GR N-terminal domain as bait revealed that this region of GR, particularly amino acids 263 to 419, physically interacts with the β component of the G protein complex (22) (Fig. 1). This complex propagates cell surface, G protein–coupled receptor (GPCR)–generated signals into the cell cytoplasm (23). The G protein β subunit binds to GR and directly regulates GR-mediated transcriptional activity in the nucleus (22). Thus, extracellular hormones and other compounds that activate GPCRs may also influence GR transcriptional activity. Because GR is transiently attached through the G protein β subunit to the inner surface of the plasma membrane (22), this interaction may explain some nongenomic actions of glucocorticoids observed at the plasma membrane. Similarly to GR, the N-terminal domain of the progesterone receptor (PR) mediates nongenomic actions of this receptor by activating c-Src family tyrosine kinases through its interaction with SH3 domain–containing molecules (24).

Thus, it appears that the N-terminal GR isoforms may differentially preserve some of the transactivational and G protein–related activities of their N-terminal domain and may also contribute to the nongenomic effects of this nuclear receptor.

Stochastic Aspects of Glucocorticoid Signaling

Human cells probably express at least 16 N-terminal GRα and GRβ isoforms, which are produced and catabolized at different rates and act as 16 distinct monomers. These 16 monomers can form 256 homo- or heterodimers, which will display discrete interactions with other molecules that participate in glucocorticoid signaling and will have quantitatively and quantitatively different transcriptional activities on the glucocorticoid-responsive genome and different interactions with the glucocorticoid-responsive proteome (Fig. 3). This indicates that glucocorticoid action is the integrated result of myriads of chemical reactions that determine the overall glucocorticoid action in specific cells, organs, and tissues in a highly stochastic fashion (4, 6).

Fig. 3.

The stochastic aspects of glucocorticoid signaling: Summary of intracellular molecules that may influence the ultimate transcriptional activity of the GR. Red oval: Possibly 16 GRα or GRβ isoform monomers and 256 or more combinations of GRα and GRβ homo- and heterodimers have distinct quantitative and qualitative transcriptional activities on glucocorticoid-responsive genes, which represent about 20% of the expressed human genome. Orange oval: Posttranslational modifications of GR isoforms and GR-related proteins in the nucleus that alter their expression and half-life. These include methylation, acetylation, nitrosylation, sumoylation, and ubiquitination. Some of these modifications also affect GR target chromatin structure and regulate GR transcriptional activity indirectly. Yellow oval: Transcriptional nuclear hormone coactivators, chromatin modifying complexes, other transcription factors, and a host of other regulatory proteins involved in glucocorticoid-mediated alterations in gene transcription. Green oval: Protein-protein interactions and posttranslational modifications of GR isoforms in the cytoplasm that determine their ligand-friendliness, subcellular localization, nucleocytoplasmic shuttling, transcriptional activity, and turnover. [Data from (1, 6, 11, 32, 34)]

Physiologic and Pathologic Implications

As with many other homeostatic functions, too much—as well as too little—glucocorticoid activity is associated with pathology (for instance, Cushing syndrome and Addison disease, respectively) (2). Because the responsiveness of glucocorticoid target tissues is crucial for the magnitude of the end effects of these hormones, Cushing- or Addison-type pathological manifestations may result from tissue glucocorticoid hypersensitivity or resistance due to, respectively, augmented or reduced biologic effects of cortisol in the organs and tissues indicated in Table 1 (10, 11). However, the brain and the pituitary, which regulate the production of glucocorticoid secretion through negative feedback by sensing time-integrated tissue exposure to glucocorticoids, are also targets of glucocorticoids. Thus, any generalized change in the glucocorticoid signaling system will be followed by corrective, compensatory, and generally protective changes in the activity of the hypothalamic-pituitary-adrenal (HPA) axis by adjusting target tissue effects to optimal levels (10, 25). Absence of complete compensation—be it excessive or deficient—could result in chronically altered homeostasis or allostasis and target tissue pathology, as occurs in chronically stressed or depressed individuals who have mild but persistent hyper- or hyposecretion of cortisol (11, 25).

Table 1

Clinical manifestations of tissue hypersensitivity or resistance to glucocorticoids.

The suprahypothalamic, hypothalamic, and pituitary glucocorticoid-sensing network, however, differs from the networks of glucocorticoid signaling systems in the arousal, associative, reward, metabolic, cardiovascular, and immune systems. Any change in one or more molecules that participate in the glucocorticoid signaling system could potentially produce a discrepancy in the sensitivity of target tissues to glucocorticoids and result in pathology (4, 25). The findings of Lu and Cidlowski are particularly relevant and revealing in this respect, as they suggest that each type of GR-expressing cell and each expressed glucocorticoid-responsive gene are different and that the end result involves a stochastic process that may be affected differently by mutations or polymorphisms of glucocorticoid signaling pathway–related molecules. This would result in complex diseases characterized by dysfunction of this system, as suggested in Table 1.

Several autoimmune inflammatory or allergic disorders, including systemic lupus erythematosus, rheumatoid arthritis, and glucocorticoid-resistant asthma, as well as systemic inflammation and acute respiratory distress syndrome, are associated with reduced glucocorticoid action in relevant tissues (11, 25). On the other hand, increased glucocorticoid activity in adipose tissue, blood vessels, or both are evident in visceral obesity-related insulin resistance and diabetes mellitus type II as well as essential hypertension (10, 11, 25). Moreover, tissue sensitivity to glucocorticoids is affected by interactions of the host with viruses, such as human immunodeficiency virus type 1, which increases glucocorticoid sensitivity of host tissues, and the adenovirus, which decreases it (26, 27).

Introduction of an acquired disruption of GR signaling into the mouse forebrain leads to development of a number of physiological and behavioral abnormalities that mimic major depressive disorder in humans, including hyperactivity of the HPA axis, impaired negative glucocorticoid feedback, and depression-like behavioral manifestations (28). Furthermore, a convincing association was recently made between the ER22/23EK polymorphism of the human GR gene and increased human longevity secondary to a healthier metabolic profile (29). These polymorphisms were previously found to be associated with subtle glucocorticoid resistance (30). An even more recent study showed that when the ER22/23EK polymorphism was present, about 15% more GRα-A protein was expressed than when it was absent, whereas total GR levels (GRα-A plus GRα-B) were not affected (31). These results suggested that transcriptional activity in GRα (ER22/23EK) carriers was decreased because more of the less transcriptionally active GRα-A isoform was formed. The authors proposed that this was likely caused by an altered secondary mRNA structure (31).

Beyond the Glucocorticoid Signaling System

Lu and Cidlowski showed that the production of N-terminal isoforms conferred an additional important mechanism for regulation of GR actions (6). The mineralocorticoid (MR), estrogen (ER) α, and progesterone (PR) receptors also contain potential alternative translation initiation sites in their N-terminal domains. Thus, like the GR gene, the genes encoding these receptors might produce N-terminal isoforms (6). Therefore, tissue-specific and regulated variable N-terminal isoform production may be a general mechanism that defines target tissue sensitivity to steroid hormones, further adding to the complexity of their own signal transduction systems.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
View Abstract

Navigate This Article