A General Introduction to Glucocorticoid Biology

Abstract

Glucocorticoids (GCs) are steroid hormones widely used for the treatment of inflammation, autoimmune diseases, and cancer. To exert their broad physiological and therapeutic effects, GCs bind to the GC receptor (GR) which belongs to the nuclear receptor superfamily of transcription factors. Despite their success, GCs are hindered by the occurrence of side effects and glucocorticoid resistance (GCR). Increased knowledge on GC and GR biology together with a better understanding of the molecular mechanisms underlying the GC side effects and GCR are necessary for improved GC therapy development. We here provide a general overview on the current insights in GC biology with a focus on GC synthesis, regulation and physiology, role in inflammation inhibition, and on GR function and plasticity. Furthermore, novel and selective therapeutic strategies are proposed based on recently recognized distinct molecular mechanisms of the GR. We will explain the SEDIGRAM concept, which was launched based on our research results.

Discovery of Glucocorticoids and the Glucocorticoid Receptor

The first steps leading to the discovery of glucocorticoids (GCs) took place in the 19th century when the physician Thomas Addison described that patients suffering from (chronic) fatigue, muscular degeneration, weight loss, and a strange darkening of the skin could obtain beneficial effects from adrenal extracts (1). This disease is now known as Addison's disease, which is a form of adrenal insufficiency. In 1946, Edward Calvin Kendall isolated four steroidal compounds from adrenal extracts, which he named compounds A, B, E, and F (2). Compound E, would become known as cortisol and was synthesized later that year by Sarett (3). The therapeutic potential was discovered by rheumatologist Philip Hench in a patient suffering from rheumatoid arthritis (4). Hench and Kendall were awarded the Nobel prize for Medicine and Physiology in 1950 together with Tadeus Reichstein who succeeded in isolating several steroid hormones from the adrenals, eventually leading to the discovery of cortisol. Since the discovery of their anti-inflammatory potential GCs were hailed as wonder drugs to treat various inflammatory diseases and became part of the group of most used and cost-effective anti-inflammatory drugs.

GCs bind the GC receptor (GR), a member of the nuclear receptor (NR) family of intracellular receptors, which also contains the estrogen receptor (ER), progesterone receptor (PR), androgen receptor (AR), and mineralocorticoid receptor (MR) as well as several orphan receptors (with no known ligand) (5, 6). In 1966, the GR was identified as the principal receptor responsible for the physiological and pharmacological effects of GCs (7). It would take almost two more decades for the human GR-coding gene, NR3C1 to be cloned (8, 9). The GR is very closely related to the MR and these receptors exhibit some cross-reactivity, more specifically the MR is activated both by its own ligands, mineralocorticoids (MCs) and by GCs, but GR is activated only by GCs (10). NRs are involved in many aspects of mammalian biology, including various metabolic functions, cardiac function, reproduction and (embryonic) development, and the immune system (11).

Glucocorticoid Synthesis, Regulation and Biological Availability

GCs are steroid hormones that are essential for the daily functioning of mammals. They are involved in several physiological processes, namely in metabolism (12), water and electrolyte balance (13), the immune response (14, 15), growth (16), cardiovascular function (17, 18), mood and cognitive functions (19–21), reproduction (22), and development (23). GCs are mainly synthesized in the cortex of the adrenal gland together with aldosterone (a MC) and dehydro-epi-androsterone (DHEA). The latter is the precursor of testosterone and estrogen. Aldosterone, GCs, and DHEA are synthesized by different steroidogenic enzymes in the mitochondria of, respectively, the zona glomerulosa, the zona fasciculate, and the zona reticularis of the adrenal cortex. They are however all synthesized from the same precursor, namely cholesterol (24). Extra-adrenal GC production in the thymus, vasculature, brain, and epithelial barriers has also been observed (25–30). These locally produced GCs are thought to predominantly exert local effects and contribute only minimally to the systemically circulating pool of GCs allowing a high spatial specificity of steroid actions, which are also independent of the circadian and stress induced regulation of endogenous GCs.

Adrenal GC production is regulated by the hypothalamic-pituitary-adrenal (HPA) axis (Figure 1). Under basal, unstressed conditions GCs are released from the adrenal glands in the bloodstream in a circadian and ultradian rhythm characterized by peak levels during the active phase which is in the morning in humans and in the beginning of nighttime in nocturnal animals such as mice. The activity of the HPA axis is further increased upon physiological (e.g., activated immune response) and emotional stress. When the HPA-axis is stimulated, corticotropin-releasing hormone (CRH), and arginine vasopressin (AVP) are released from the hypothalamic paraventricular nucleus (PVN). Subsequently, CRH and AVP bind their receptor CRH-R1 and V1B in the anterior pituitary inducing the release of adrenocorticotrophic hormone (ACTH) in the circulation. ACTH will in turn stimulate the adrenal gland to synthetize and secrete GC hormones (cortisol) in the circulation (31).

Figure 1.

Hypothalamic-pituitary-adrenal axis. The hypothalamic-pituitary-adrenal (HPA) axis activity is controlled by the circadian rhythm and can be induced by physiological and emotional stress. When activated, corticotrophin-releasing hormone (CRH), and arginine vasopressin (AVP) are released from the hypothalamic paraventricular nucleus (PVN). This induces the release of adrenocorticotrophic hormone (ACTH) from the pituitary gland into the systemic circulation. ACTH will activate cortisol synthesis in the cortex of the adrenal gland. Cortisol negatively regulates the HPA-axis activity, e.g., by repressing the transcription of CRH and POMC by binding to negative glucocorticoid responsive elements (nGRE) or by binding to the transcription factor Nur77 involved in the POMC expression.

The HPA axis is subject to a negative feedback inhibition by GCs, both in a genomic and a non-genomic way. The genomic feedback regulation is mediated through binding of GCs to the GR both at the level of the PVN and the pituitary gland, thereby repressing the CRH, CRH-R1, and the POMC gene (Figure 1). POMC codes for the proopiomelanocortin prohormone which is the precursor of ACTH. CRH, CRH-R1, and POMC gene expression are repressed by the binding of GR to negative glucocorticoid responsive elements (nGREs) (32–34). Next to this, GR is also able to physically interact with the Nur77 protein which also binds in the POMC promoter, thereby preventing it from performing its transcription function (35, 36). Non-genomically, GCs regulate the HPA axis for example via the release of endocannabinoid from CRH neurons thereby suppressing the release of glutamate from presynaptic excitatory synapses (37), or via γ-aminobutyric acid (GABA) release at the inhibitory synapses of CRH neurons (38).

Once secreted in the bloodstream GCs are bound to and transported by plasma proteins which keep the GCs inactive. Corticosteroid-binding globulin (CBG) is the main GC-binding protein in the plasma, with about 80–90% of the GCs bound to it (39). Several proteases target CBG, such as neutrophil elastase at sites of infection (40), causing the release of bound GCs. Approximately 10% of the GCs are bound to albumin that binds GCs with less affinity than CBG (39).

Due to their lipophilic nature, free GCs diffuse through the cell membrane to exert their function. However, the actual bioavailability of GCs in the cytoplasm is regulated by the balance between active and inactive forms of GCs. Two enzymes are responsible for the conversion between inactive cortisone (or 11-dehydrocorticosterone in mice) on the one hand and the active cortisol (or corticosterone, in mice) on the other hand. While 11β-hydroxysteroid dehydrogenase 1 (11β-HSD1) catalyzes the conversion of cortisone to cortisol, 11β-HSD2 carries out the opposite reaction (Figure 2). 11β-HSD2 is highly expressed in tissues with high MR expression, such as the kidneys, to prevent GC-induced MR activation which is known to cause salt and water dyshomeostasis (41, 42). Biologically active GCs will bind their receptor in the cytoplasm which exerts their physiological effects. This mechanism also confers a tight spatial regulation of GC actions, as the levels of these enzymes may be tissue or even cell specifically regulated and will directly determine the balance between the inactive and active form of GCs and thus the strength of the effect.

Figure 2.

Conversion of inactive GCs to active GCs. Inactive Cortisone (human) and 11-dehydrocorticosterone (mouse) are activated to active cortisol and corticosterone by 11β-hydroxysteroid dehydrogenase 1 (11β-HSD1), and inactivated again by 11β-HSD2.

Under physiological conditions the role of endogenous GCs is not simply anti-inflammatory or immunosuppressive and shows more immunomodulation. It has been shown that GCs can also work pro-inflammatory (14). This occurs mainly in conditions of acute stress and is related to the concentration of GCs present (14, 43). Such pro-inflammatory actions were shown to include: elevation of pro-inflammatory cytokine levels (IL-1β) (44) or an exacerbation of the peripheral immune response in delayed type hypersensitivity (45).

Next to the endogenous GCs, various synthetic GCs (e.g., Prednisolone, Methylprednisolone, Fluticasone, Budesonide, and Dexamethasone) have been developed by the pharmaceutical industry that serve as treatments for various diseases. All these synthetic GCs were developed based on the structure of endogenous GCs (cortisol/hydrocortisone) (46). Experiments with structural modifications, mainly replacing side chains, resulted in synthetic GCs with optimized characteristics for medical use (pharmacokinetics, bioavailability, cross-reactivity with the MR). The most obvious differences between synthetic and endogenous GCs are (i) potency, as the synthetic variants are usually much better activators of the receptor than cortisol (4x−80x more) (47). (ii) Specificity, since endogenous GCs activate both GR and MR, but many synthetic GCs (e.g., dexamethasone, methylprednisolone) act (almost) exclusively on the GR. And (iii) synthetic GCs may (prednisolone) or may not (dexamethasone) be subject to processing by 11β-HSD1/2 which has a major impact on their bioavailability, as some synthetic GCs may (not) need to be activated by these enzymes or cannot be changed into an inactive form by them. Also, most synthetic GCs also do not bind the carrier proteins such as CBG (48–50). These facts are important to keep in mind when giving GC treatment or performing research using synthetic GCs.

The Glucocorticoid Receptor

The GR mediates the actions of GCs in cells. It belongs to the nuclear receptor superfamily of transcription factors (TFs) and is a 97 kDa protein that is constitutively and ubiquitously expressed throughout the body (51). Nevertheless, GCs exert cellular and tissue-specific effects due to the existence of different GR isoforms on the one hand and cell- and context-specific allosteric signals influencing GR function on the other hand (52–54). The GR functions by regulating the expression of GC responsive genes in a positive or negative manner. It is estimated that there are between 1,000 and 2,000 genes that are subject to GR mediated regulation, with some studies stating that up to 20% of all genes are responsive to the GR in some way (55).

GR Gene and Protein

The human gene encoding the GR is the “nuclear receptor subfamily 3 group c member 1” (NR3C1) gene localized on chromosome 5 (5q31.3). The mouse Nr3c1 gene is localized on chromosome 18. The hGR gene consists of 9 exons of which exon 1 forms the 5′ untranslated region (UTR) and exons 2–9 encode the GR protein (52).

The 5′ UTR of the hGR is GC-rich, but does not contain TATA or CAT boxes (56). Thus, far 13 hGR exon 1 variants differing in upstream promoter regions have been identified (A1–3, B, C1–3, D–F, H–J) (Figure 3). Differential use of these promoters, located about 5 kb upstream of the transcription start site, causes varying expression levels of GR protein isoforms between cells and tissues (57–60). These promoters contain multiple binding sites for several TFs such as AP-1 (61) and Interferon Regulatory Factor (IRF) (62), but also for GR itself, thereby enabling the regulation of its own expression (63). Furthermore, these exon-1 variants are subject to epigenetic regulation. Several epigenetic modifications, such as DNA methylation and histone acetylation/methylation are known to occur in this region (or in other regions). The presence or absence of such modifications has been related to GR gene expression levels, GC resistance in certain cancers, promotion of cancer development, and mental health (64–69).

Figure 3.

Glucocorticoid receptor gene and protein. (A) Genomic structure of the glucocorticoid receptor (GR) gene. (B) Alternative splice and translation-initiation variants of the GR protein. (C) Structure of the GR protein consisting of an N-terminal domain (NTD), DNA-binding domain (DBD), a hinge region (H), and a ligand-binding domain (LBD), with a focus on the two zinc-fingers of the DBD and the GR^Dim mutation (A458T in human, A465T in mouse). Identified post-translational modifications of the GR are indicated in the black circles. Regions important in GR function are indicated below the protein. AF, Activation function; NES, Nuclear Export Signal; NLS, Nuclear Localization Signal; NRS, Nuclear Retention Signal; P, phosphorylation; S, sumoylation; U, ubiquitination; N, nitrosylation; O, oxidation; A, acetylation.

The hGR protein (Figure 3) is a modular protein that, like other NR family proteins, is built up out of an amino-terminal domain (NTD), a DNA-binding domain (DBD), a hinge region, and a C-terminal ligand-binding domain (LBD) (52). The NTD is encoded by exon 2 and is the least conserved region of the NR family. It is inherently unstructured, vulnerable to proteases and only becomes structured when the protein binds DNA and forms dimers (70). In the NTD the ligand independent activation function 1 (AF1) is located. This AF1 binds cofactors, chromatin modulators, and the transcription machinery (71–73). The GR DBD is encoded by exons 3 and 4 and is important for DNA binding and GR dimerization. It is characterized by two highly conserved subdomains each containing a Cys4-type zinc finger. In the first subdomain the GR's proximal box (P box) is contained which is important for site specific GR DNA binding. The second subdomain contains the distal box (D box) which is important for GR dimerization (74). Exons 5–9 of the NR3C1 gene encode the GR's hinge region and LBD. The former provides both flexibility between the DBD and LBD as well as a regulatory interface. The hinge region can be acetylated (lysine residues) and is a target of CLOCK/BMAL acetylation and the presence of acetyl moieties in this area reduces GR activity. Research has also shown that the interaction between the GR and CLOCK/BMAL can be uncoupled, such as by chronic stress or night shift work, which may cause hypercortisolism related pathologies (75, 76). The latter contains a ligand binding pocket, which is formed by 12 α-helices and 4 β-sheets, and the ligand-dependent AF-2 domain. The LBD has also been found important in GR dimerization (77). Further, nuclear localization (NLS), nuclear export (NES), and nuclear retention signals (NRS) have been identified in the GR protein and these are important for the subcellular distribution of the GR. Two NLS have been identified, one in the DBD and one in the LBD (78). A NES is located between the 2 zinc fingers (79) and a NRS delaying GR nuclear export overlaps with NLS1 (80).

Not a single, but multiple GR protein isoforms are identified. This is the result from alternative splicing and the use of 8 different translation initiation start sites (81). Alternative splicing at exon 9 results in two different GR splice variants, namely the classical 777 AA-long GRα or the 742 AA-containing GRβ (8). Both isoforms are identical up to AA 727, but contain non-homologous AA thereafter. Hence, GRβ has a shortened LBD lacking helix 12 and therefore it cannot bind GCs (82). Despite this, GRβ is constitutively found in the nucleus where performs several functions. It was believed and later also shown to be an antagonist to the GRα isoform. Several mechanisms have been proposed for the dominant negative action of GRβ, such as competing with GRα for GR-binding sites and co-regulators and the formation of inactive GRα/β heterodimers (82–84). The role of the GRβ is more extensive than being a simple antagonist. Other studies have shown that the GRβ regulates gene transcription of non-GRα target genes in an GRα and GC independent manner (85). Furthermore, while GRβ cannot bind endogenous GCs, it was show to bind the GR antagonist RU-468, and is modulated by it (86). Perhaps some synthetic GR agonists could also bind to this isoform. The GRβ isoform plays a role in GC resistance (insensitivity to GC treatment) in patients for several diseases. This resistance can be caused by its GRα antagonism as well as by the transcriptome changes its presence causes. A recent study showed that overexpression of GRβ in colonocytes causes dysregulation of many genes also found back in IBD patients (87). Next to GRα and GRβ, GRγ, GR-A, and GR-B splice variants have also been identified (illustrated in Figure 3). All splice-isoforms show diminished activity compared to GRα (88–90). Besides splicing, GR mRNA is further regulated post-transcriptionally via adenine uridylate-rich elements (ARE) in the 3′ UTR of the GR mRNA which mediate GR destabilization (91). Next to this, GR mRNA stability is also regulated by microRNAs (for example: miR-124) which bind to their binding motifs, mostly in the 3′ UTR (92, 93).

Eight GRα translation initiation variants have been identified (GRα-A, -B, -C1, -C2, -C3, -D1, D2, and D3) which is the result from the existence of 8 highly conserved AUG start codons in exon 2 (Figure 3) (94). The AUG start codons are differently selected due to ribosomal leaky scanning and ribosomal shunting mechanisms (94). Because the same AUG start sites are also present in the GR splice-variants, all the translation-initiation isoforms are expected to occur in each of the splice-variants (95). The GR translation variants all have a similar GC and glucocorticoid responsive element (GRE)-binding affinity, but they differ in the length of their N-termini and their transcriptional activity. They show different subcellular localization, regulate distinct sets of genes and their relative levels vary between and within cells (94). The mechanism of regulation of alternative translation start sites and alternative splicing in response to physiological, pathological, and cell-specific signals is still poorly understood. In vitro work proved that these isoforms do have the capability to regulate distinct transcriptional programs (96). A later study showed that the different isoforms can regulate apoptosis with the GRα-C3 being pro-apoptotic and the GRα-D3 anti-apoptotic (97).

GR Activation and Nuclear Translocation

In the absence of intracellular bioactive GCs, the GR finds itself as a monomer in the cytoplasm where it resides in a multiprotein complex. This chaperone complex is important for GR maturation, ligand binding, nuclear transport, and activation. The composition of the chaperone complex changes during the different GR maturation/activation states (Figure 4) (98). After GR translation the GR is bound by Hsp70, an interaction that is accelerated by the Hsp40 co-chaperone. Once the folding process is complete GR is transferred from Hsp40/Hsp70 to Hsp90, a transfer that is mediated by Hop (99–101). Recruitment of p23 (102) and FKBP51 to the multiprotein complex leads to maturation of GR-chaperone complex into a conformation that has very high affinity for GR ligands. After GC-binding the GR-chaperone complex again reorganizes (FKBP51 is replaced by FKBP52) and a GR conformational change is induced, leading to the exposure of the GR's 2 nuclear localization signals (103). These are subsequently bound by nucleoporin and importins that carry the GR through the nuclear pore complex into the nucleus (104, 105). Initially it was believed that the GR disassociates from the cytoplasmic chaperone complex upon ligand binding. However, recent research has shown that the chaperone complex is required for efficient nuclear translocation of the receptor (106).

Figure 4.

Glucocorticoid receptor chaperone complex and maturation. (A) After glucocorticoid receptor (GR) translation an Hsp70-Hsp40-GR complex is formed in the cytoplasm. (B) A subsequent ADP-dependent Hsp70 change induces the binding of Hop. (C) Hop induces the binding of Hsp90. (D) After Hsp90 binding to Hop, Hsp70, and Hsp40 are released from the chaperone complex and replaced by p23 and FKPB51. The GR has now matured into a high affinity complex. (E) After binding of glucocorticoids FKBP51 is replaced by FKBP52, which is necessary for the transport of the GR to the nucleus.

Once inside the nucleus, the activated GR can go on to exert its function or it can be transported back to the cytoplasm, inhibiting the GR's transcriptional activity. Nuclear export of GR is regulated by exportins and calreticulin (CRT) which binds to the GR NES, thereby disrupting the GR-DNA binding (107, 108).

The balance between nuclear import and export determines the proportion of GR protein in the nucleus and has a direct influence on the strength of GR's transcriptional activities. In the nucleus, the GR acts as a TF that can activate (trans-activation) or inhibit (trans-repression) genes as well as modulate the function of other TFs (tethering). Most of the GR functions are restricted to the nucleus, but some non-nuclear actions of GR are also known.

GR Function

In the nucleus, the GR is able to transcriptionally activate (transactivate (TA)) or transcriptionally repress (transrepress (TR)) gene-expression, both as a monomer and as a dimer, and usually via direct contact with DNA. Recently it was discovered that the GR can also bind to the DNA as a tetramer (Figure 5) (109, 110). The importance of this GR tetramer in transcriptional regulation is not well-understood and needs further investigation.

Figure 5.

Glucocorticoid receptor activation and function. Lipophilic glucocorticoids (GCs) diffuse through the cell membrane and bind the glucocorticoid receptor (GR) in the cytoplasm. This induces a change in the chaperone complex bound to GR, after which it translocates to the nucleus to transactivate (+) or transrepress (-) gene transcription as a monomer or a dimer. The GR can transactivate genes by binding to glucocorticoid responsive elements (GRE) as a dimer, but also as a monomer by binding to other transcription factors (TF) through tethering or by binding to composite-elements. The GR can further transrepress gene-expression by binding to inverted repeat GR-binding sequences (IR-GBS), by tethering, by composite-elements, by competing for DNA binding-sites (BS), by sequestrating TFs and by competing for cofactors with other TFs. GR might also function as a tetramer, but its function is not known.

The GR associates with specific genomic loci and orchestrates the assembly of TF regulatory complexes containing the GR, other TFs and co-regulators that modulate the activity of the RNA polymerase II (RNApolII). Different modes of genomic GR transcriptional regulation are described (Figure 5).

The simplest form of GR-DNA interaction is the binding of GR to genomic glucocorticoid binding sites (GBS) containing a GRE. Classically, the GR exerts its transactivation function by binding to GREs, which are 15 bp long sequence motifs of 2 imperfect inverted palindromic repeats of 6 bp separated by a 3 bp spacer. The generally accepted GRE consensus sequence is AGAACAnnnTGTTCT. However, this may be better represented as a sequence logo (Figure 6), which illustrates that some positions are much more variable than others. The GR binds to the GRE as a homodimer and each GR DBD makes contact with about 3 nucleotides in each of the half site hexamers. The two GR molecules bind the GRE in a head-to-tail fashion and 5 AA within the D box of the second GR zinc finger provide critical protein-protein contacts between the two GR partners important for stabilization of the GR DBD on the DNA. In this D box a hydrogen bond is formed between Ala458 of one dimer partner and Ile483 of the other partner (74, 111). A second interface important for dimerization (Ile628) has been identified in the LBD (77, 112). Recent research proposes that the LBD may have other dimerization interfaces related to another dimer structure (113).

Figure 6.

Sequence logo of the human glucocorticoid responsive element to which the GR binds to. See text for more details.

GREs contain relatively few highly conserved residues and because GREs are rather short, they are abundantly present in the genome. ChIP-seq experiments with antibodies against GR showed however that only a small fraction of GRE sites are in fact occupied by the GR (114). Why this is the case is still a topic of research, but it has been shown that the chromatin structure plays a big role in determining which sites are accessible to GR under certain conditions (115, 116). It has also been shown that many GR binding sites can be found very far from a (known) gene or transcriptionally active sites, indicating that GR often occupies enhancer regions and/or chromatin looping is involved in GR transcriptional regulation (114).

Evidence has been found for a 2nd mode of GR-DNA interaction where GR, as a monomer, binds to half sites with an AGAACA (or the reverse complement TGTTCT) consensus sequence (117). If a binding site for another TF is nearby the GRE-half site, both elements may act as a composite site where there is an interaction (positive or negative) between the GR (monomer) and the other TF (118) (Figure 5). An analysis in mouse liver showed that under endogenous corticosterone levels (i.e., low concentrations) GR binding to half sites as a monomer is more prevalent than binding of full GRE sites by homodimers. In response to exogenous GCs (i.e., high concentration) the GR dimers assemble on full length GRE near known induced genes and this happens in concert with monomer removal of sites near repressed genes (119).

A third class of GR-DNA interactions involves inverted-repeat GBS (Figure 5). Binding to such an element leads to inhibition of gene expression. These IR-nGREs have a consensus CTCC(N)₀₋₂GGAGA sequence and structural analysis showed that at these sites 2 GR monomers bind on the opposite sides of the DNA, in a head-to-tail orientation and with negative co-operativity with each other (120, 121).

Lastly, there are the indirect binding, or tethering, sites where GR is recruited to a TF complex through protein-protein interactions with heterologous DNA-bound TFs (Figure 5). These GBSs lack a GRE, IR-nGRE, or a GRE half site. Several TFs are known to recruit ligand bound GR via tethering including members from the AP1, STAT, and NF-κB families of TFs. These interactions directly alter the capacity of the directly DNA-bound TF to bind DNA, recruit cofactors, and activate/repress gene transcription (122, 123).

The GR can also TR gene-expression by competing with other TFs for binding to overlapping DNA-binding sequences. Indeed, recently GR half-sites were even found embedded in AP-1 response elements (124). Finally, the GR can TR gene-expression by competing with other TFs for the binding of cofactors (125–127) or by sequestrating TFs, thereby obstructing them to bind to the DNA (128) (Figure 5).

GR Plasticity

The GR operates in a cell- and context-specific manner. This is not only due to a different expression of GR protein isoforms but is also the cause of different signals that modulate the GR's activity at specific GBSs. Four signals are described to influence the GR's function.

A first signal that modulates GR activity is the DNA, which acts as an allosteric regulator of the GR. GRE sequences differing by only one single base pair were namely shown to affect GR conformation and regulatory activity (129). Moreover, allosteric changes provoked by one half site can be transduced via the GR lever arm (located between the P and D box, see Figure 3) and the receptor's D box to the dimer partner, affecting the GR's transcriptional activity (130, 131).

A second signal influencing the GR transcriptional output obviously comes from the ligand that binds to the LBD. After ligand-binding helix 12 is exposed and cofactors are recruited to the AF2 in the LBD. Depending on the ligand, the LBD will adopt another conformation and attract other cofactors thereby influencing the GR's transcriptional outcome (132, 133). The latter forms the basis of the research for “Selective GR Agonists and Modulators” (SEGRAM).

Third, the GR is heavily modified by potential post-translational modifications (PTMs). Several phosphorylation (134–140), ubiquitination (141), sumoylation (142), acetylation (76), and nitrosylation sites (143) as depicted in Figure 3 have been identified influencing GR-localization, stability, DNA binding, ligand response, and regulatory activity.

Last, the GR's transcriptional output is influenced by its interaction partners. These include other TFs that bind direct or indirect to GR and cofactors which are recruited to GR and are involved in functions such as chromatin regulation and regulation of the transcriptional machinery function (53, 144). The composition of the cofactor complex recruited to the GR depends on the cell specific expression of cofactors, the cell context and the integration of the previous described signals (DNA, ligand, and PTMs) that influence the GR's conformation (145). This cofactor complex eventually determines the transcriptional output of the GR.

Non-genomic GC and GR Actions

The GR is not only able to function by genomic actions, but also through non-genomic actions. Non-genomic GC/GR actions are fast and do not require transcription or protein synthesis. Limited knowledge is however available on non-genomic GC/GR actions. These include GC-mediated effects on membrane lipids, changing their physicochemical properties (146). Further, GCs have also been seen to act on a membrane-bound GR which is related to the classical GR and probably the result from differential splicing, alternative transcription initiation and PTMs (146, 147). Another membrane receptor, unrelated to the classical GR, probably also binds GCs. This protein is probably a G-coupled receptor that signals through cAMP and that binds endogenous GCs with high affinity. However, it does not bind most GC analogs such as dexamethasone (148). Other non-genomic actions, e.g., modulation of the MAPK signaling cascade, might result from components that are released from the GR chaperone complex upon the binding of GCs to the GR or from membrane bound GR (149, 150).

A final type of non-genomic action of the GR is its effect on mitochondrial function. It was show that the GR can translocate to and reside in mitochondria (151, 152). This mitochondrial GR is capable of regulating gene transcription from the mitochondrial chromosome by binding to GRE like elements alone or in complex with other factors. This was demonstrated in vitro, using a hepatoma cell line and in brain cell of mice and rats (153–155). A recent study showed that a GR isoform, GR?, is located in the mitochondria and plays a role in regulating cell energy metabolism in a ligand independent manner (156).

GC Therapy: Drawbacks and Optimization

GCs are therapeutically mainly used for their anti-inflammatory and immunosuppressive effects. These are a.o. the result of the transcriptional induction of several anti-inflammatory protein-coding genes such as TSC22D3 (coding for glucocorticoid-induced leucine zipper, GILZ) and DUSP1 (coding for Map Kinase Phosphatase 1, MKP1) and from the repression of pro-inflammatory TFs such as NF-κB and AP-1. GCs are used to treat inflammatory disorders such as asthma (157), skin rashes (158), rheumatoid arthritis (RA) (159), multiple sclerosis (160), and systemic lupus erythematosus (SLE) (161). In most cases, synthetic glucocorticoids are used but hydrocortisone is also a popular option.

Despite its strong anti-inflammatory capacity, GC therapy is limited by two major drawbacks. First, GCs are well-known to be associated with adverse effects, particularly when given in high doses for long time periods. Figure 7 graphically presents GC-associated side effects, with osteoporosis, hyperglycemia, cardiovascular diseases, and infections as the four most worrisome adverse effects for clinicians (162). These side effect may be severe enough to affect the therapy or cause an increased risk to other negative effects. A recent study in RA patients showed a clearly increased risk of bone fractures correlated with the administration of GCs (osteoporosis) (163). Second, some patients are refractory to the therapy and are GC resistant (GCR). GCR can either be inherited, mostly via mutations in the NR3C1 gene (52, 164), or acquired (165). The latter can be caused by ligand induced homologous downregulation of the GR, caused chronical GC treatment (166, 167), or by pathophysiological processes accompanying the inflammatory disease states [e.g., chronic obstructive pulmonary disease (COPD) (168), SLE (169)]. The pathophysiological processes provoking GCR are very heterogeneous, e.g., oxidative stress and inflammatory cytokines are known triggers of GCR and have multiple effects on GR biology (170–176). GCR occurs in 4–10% of the asthma patients, 30% of the RA patients and in almost all of the sepsis and COPD patients (177–179).

Figure 7.

Overview of glucocorticoid-associated side effects.

To achieve a positive benefit-to-risk ratio when using GCs, guideline recommendations regarding optimal dosing must be followed and potential adverse effects must be monitored, prevented and managed (180–183). Next to this, much research effort is put in developing innovative GCs or GR ligands that improve the therapeutic balance (184–186).

Currently available GCs in the clinic activate all GR activities. During the past 20 years intensive research for SEGRAMs, which promote a GR conformation favoring TR over TA, has been performed. This search for SEGRAMs is based on the central dogma in GR biology which states that GR monomer-mediated TR is sufficient to counteract inflammation, while GR dimer-mediated TA is responsible for most of the adverse effects of GCs, e.g., by the induction of genes encoding glucose-6-phosphatase (G6P) and phosphoenolpyruvate carboxykinase (PCK1). This long accepted dogma in GR biology originates from initial work with the GR^Dim mutant (187). This GR^Dim mutant carries a A465T mutation in the D-loop of the second zinc finger of the GR-DBD.

This D-loop is one of the primary dimerization interfaces, consequently this mutant shows impaired homodimerization and reduced functionality. Initial observation on the GR^Dim mutant showed a strongly impaired transactivation and retained capability to transcriptionally repress genes, particularly as a monomer (111). Follow-up work on the GR^Dim found that there was still transactivation of certain genes possible by these mutant receptors (129, 131). This raised the question again if the GR^Dim was still capable of some dimerization and or DNA binding. An in vivo imaging study with labeled GR showed that the GR^Dim is still capable of dimerization with endogenous and synthetic GCs, but with a lower efficiency than WT for endogenous GCs (188). The ability of the GR^Dim mutant to bind to the DNA has been a point of controversy since there is evidence against (111, 189) and pro DNA binding (131, 190, 191). Current evidence seems to suggest that the DNA binding capacity of the mutant is at least partially preserved. A second GR mutant was generated with an additional point mutation in the LBD of the receptor. This mutation is believed to disrupt a secondary dimerization interface present in the LBD, leading to even poorer dimerization and function than the GR^Dim mutant (188). In addition, under normal physiological conditions, GR^Dim mice are healthy and show no obvious phenotypes, except that they express interferon genes in their intestinal epithelium (192), It has been shown that under physiological conditions, GR binds to the DNA as a monomer, exerting transcriptional functions related to cell-type-specific functions, and that only after acute stress or injection of GCs, GR dimers are formed leading to binding to full GRE elements (119). Also, elegant, NMR-based work by Watson et al. has shown that, depending on the DNA sequence where GR dimers bind, an intramolecular signal, via a lever arm, provides a dimer- and DNA-binding-stabilizing interaction between two DBD domains, precisely via the amino acid that was mutated in the GRDim version. The absence of this amino acid “weak binding” in the GR^Dim version was enough to cause less robust dimers and DNA binding (131).

It has been stated that the picture about the mechanisms of glucocorticoid actions (transactivation/transrepression) is still far from complete, especially for known GR mutants. In addition, the aforementioned functional PPI interfaces, recent structural biology work shows that the knowledge on GR dimerization and structural conformation may be incomplete based on structural homology and residue conservation between the NR transcription factor family, and new dimer interfaces that remain unexplored so far. In one study researchers have postulated that the conformation of the GR that is generally accepted as the dimeric conformation might not be correct and they propose different configurations (113). The fact most of the structural work so far was done on subdomains of the GR, as the whole protein is very hard to crystalize, may contribute to this limited knowledge of GR structure.

Many studies have investigated steroidal and non-steroidal SEGRAM in the hope to be able to dissociate the GC-induced anti-inflammatory effects from the GC-induced side effects (193–197). Several interesting SEGRAM have been characterized [e.g., Al-438, LGD-5552, ZK216348, Mapracorat and Compound A (CpdA)] and were shown to have dissociative profiles in vivo (198–206). Despite the intensive research, none of the SEGRAM have reached the market today. So far, only Fosdagrocorat (for RA) (207–209) and Mapracorat (for ocular inflammatory diseases and skin inflammation) have reached clinical trials.

To prevent GC-induced side effects, strategies other than shifting the balance between the monomeric and the dimeric GR are also followed (184–186). Some aim at cell-specific targeting of GCs via antibody- or peptide-GC conjugates (210) or via liposomes (211), thereby preventing systemic GC-effects. Other studies investigate the therapeutic use of GC-induced proteins (e.g., GILZ, the protein coded by the TSC22D3 gene) without administrating GCs themselves. By this, steps are undertaken to develop therapies that stimulate only the wanted anti-inflammatory GC-functions without inducing the broad and also the unwanted GC-effects (212). Further, studies also invest in the therapeutic potential of combination therapies, such as the combination of GR and PPAR agonists (213, 214).

GC Therapy in Acute vs. Chronic Inflammation: Sirs and the Sedigram Concept

During the recent years, it has become clear that the old idea in GC-research, that claims that GC anti-inflammatory effects can be separated from GC-induced side effects by simply dissociating GR TR from GR TA, because the former would be mainly monomeric-driven GR functions and the latter GR homodimeric-driven functions. To date it is known that this separation cannot be made that strictly. In addition, GR^Dim mice studies showed that not all GC-induced side effects are GR dimer-driven and that thus also monomeric GR is involved in at least some side effects. Indeed, GR^Dim mice were observed to develop osteoporosis and muscle atrophy, despite their lack of GR dimer-dependent effects (215, 216). Next to this, the GR dimer was found to be indispensable for the GC-mediated protection in models of acute inflammation. GR^Dim mice are strongly sensitized in models of TNF- and LPS-induced Systemic Inflammatory Response Syndrome (SIRS) (217, 218) and these mice could furthermore no longer be protected by a prophylactic Dexamethasone administration (192). Additionally, GR dimer-induced GRE genes were found to be important in the protection against SIRS: this was shown for DUSP1 (217) (encoding MKP-1) and TSC22D3 (212) (encoding GILZ). Finally, skewing the GR toward the monomer by using CpdA sensitized mice for TNF-induced SIRS, suggesting that GR monomers are unable to protect in this model of acute inflammation and that GR monomers should rather be avoided in SIRS (219). Altogether these data illustrate the importance of the GR dimer in the protection against acute-inflammation.

As a consequence of the former observations in GR^Dim mice, the SEGRAM concept needed to be revised. Therefore, recently, it was proposed that chronic inflammatory diseases which require a long-term GC therapy would benefit from “Selective Monomer GR Agonists and Modulators” (SEMOGRAMs), since these SEMOGRAMs would avoid important side effects such as hyperglycemia that are detrimental for the patients. Recently, it was also observed that ligand-induced GR turnover leading to GCR is GR dimer dependent (220). The latter observation thus further supports the need for SEMOGRAMs for the treatment of chronic inflammation. On the other hand, in acute-inflammatory settings such as SIRS, where GR dimers are indispensable, the administration of GCs that increase the GR dimerizing potential, termed “Selective Dimer GR Agonists and Modulators” (SEDIGRAMs), would be the preferred strategy to follow (221).

There has been some doubt about the value of the GR^Dim mouse tool and its inability to form homodimers and bind DNA. Although in vitro experiments (making use of high GC-doses) showed very little effect of the Dim-mutation on GR dimerization and DNA binding (188, 191), in vivo research confirmed that GC-induced transcription is very broadly hampered in GR^Dim vs. GR^WT mice (192, 222). Moreover, the remaining GR^Dim transcription was observed to be especially the result of GR monomer functioning at half-sites (119). Although we are aware that a second interface in the GR LBD is also of relevance for dimerization and that remaining dimerization in the GR^Dim mutant is probably provided through this protein-protein contact, the latter studies confirm the value of the GR^Dim mouse-tool.

Future Perspectives

Certain challenges and (new) questions remain to be answered or further investigated. GCR in patients is still largely an unresolved issue, especially in complex diseases such as sepsis but also in severe asthma. Understanding GC resistance, preventing or reverting it could mean a real breakthrough in current medical practice. Another avenue of research, aside from more selective dimer/monomer ligands, is GR structure and DNA binding conformation as some more recent research suggested that the GR can bind to DNA is a tetramer conformation instead of a dimer. Also, the non-genomic effects of GCs and GR are far from understood and need more research. Finally, a wealth of information has been published using a variety of GR ligands, some being endogenous ligands, others synthetic ligands, all of which may have very different effects on the canonical GR and non-canonical ones (splice variants, shorter proteins) and even different effects in different mammalian species or cell types. It is a big challenge for the community to try to streamline this information in a comprehensive way.

Author Contributions

JS and ST wrote the manuscript. CL reviewed the manuscript.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.