Minireview: Latest Perspectives on Antiinflammatory Actions of Glucocorticoids

Abstract

Taking into consideration that glucocorticoid (GC) hormones have been used clinically for over half a century and that more than 20 yr have passed since the cloning of the GC receptor (GR), it is hard to imagine that novel aspects in the molecular mechanism by which GCs mediate their antiinflammatory actions are still being unveiled today. Partly, this is because almost on a daily basis, novel insights arise from parallel fields, e.g. nuclear receptor cofactor and chromatin regulation and their concomitant impact on gene transcription events, eventually leading to a revisitation or refinement of old hypotheses. On the other hand, it does remain striking and puzzling why GCs use different mechanisms in so many different cell types and on many different target genes to elicit an antiinflammatory effect. Meanwhile, the obvious question for the clinic remains: is the separation of GR functionalities through differential ligand design the strategy of choice to avoid most GC-mediated side effects? This minireview aims to highlight some of the latest findings on aspects of the antiinflammatory working mechanisms of GCs.

A number of recent advances are discussed regarding various anti-inflammatory mechanisms that are mediated by the glucocorticoid receptor.

Most effects of glucocorticoids (GCs) are mediated by the intracellular receptor, GC receptor (GR). This nuclear hormone receptor is involved in the general regulation of homeostasis and controls stress pathways of diverse origin. GR can be found in almost all tissues of the human body. Nevertheless, the levels of GR protein, of which different splice and translation variants have been identified, are regulated in a tissue- and cell cycle-specific manner (1, 2, 3, 4, 5, 6).

Three key elements were originally described that affect the functionality and transcriptional regulation by GCs: first, the availability of ligand; second, the receptor itself; and third, the recruitment of cofactors and other proteins (3). Evidence is growing to support that neither ligand availability nor GR-interacting protein levels suffice to explain the observed tissue-specific gene regulation via GR. Therefore, the unique transcriptional activities and distinct tissue-specific distribution patterns of GRα isoforms may well provide a novel mechanism that helps to explain tissue-specific GC responses. Tissue-selective targeting of various mutants of GR or tissue-selective knockout of GR in mice are momentarily among the most elegant ways to find out more about the mechanism of action of GR in diverse functional programs (7, 8, 9). For new antiinflammatory drug design purposes, however, an important question to resolve is which nonexclusive mechanisms different GR isoforms may use in various cell types to combat inflammation.

Although it may appear from static immunofluorescence analyses that unliganded GR is mainly being kept inactive in the cytoplasm of the cell, complexed by chaperone proteins, such as heat-shock proteins (e.g. Hsp90, Hsp70, and Hsp23) and immunophilins (e.g. FKBP51, FKBP52, Cyp44, and PP5) (10), in fact, it was found instead that a continuous shuttling of the receptor between the two cellular compartments occurs (11). Also, chaperoning proteins are not restricted to the cytosol. Apart from ensuring ligand accessibility to the ligand-binding pocket of GR, Hsp90 interaction with ligand-loaded GR mimics the interaction of GR with transcriptional coactivators (12).

Upon ligand binding, GR undergoes a conformational change (13) causing exposure of a nuclear localization signal, subsequently allowing the receptor to translocate to the nucleus, to recruit regulatory cofactor complexes, and to influence target gene transcription as a genuine transcription factor.

In general, the segregation of nuclear receptors in different subcellular compartments is believed to act as an important regulatory checkpoint. Recently, Carrigan and co-workers (14) have defined a nuclear retention signal in the hinge region of GR. Active nuclear retention of GR was subsequently correlated with a strong inducibility of the transcriptional activity of GR.

GR target genes that are positively regulated through the transactivation mechanism, carry GC response elements (GREs), typically consisting of two conserved six-nucleotide halves separated by three nonconserved bases (5′-GGTACAnnnTGTTCT-3′), onto which GR can directly bind as a homodimer (15). Although the core GR-binding sequences are highly variable, the precise sequence of an individual GRE is highly conserved across mammalian species (16). GR-binding sites are now rather regarded as distinct GR ligands themselves. Indeed, different GR-binding sites exert different cofactor requirements, different receptor domain utilization, different levels of transactivation, and a different conformation of the GR DNA-binding domain. As such, one base pair difference in a GR-binding site can result in distinct transcriptional programs (Meijsing, S., M. Pufall, and K. Yamamoto, personal communication). GR/GRE complexes undergo a continuous assembly/disassembly; this exchange is regulated via the ligand-binding domain of GR, but the dissociation has been found to happen independently of ligand release (17). GR binding further invariably occurs at inducible or constitutive deoxyribonuclease I hypersensitive sites, involving different remodeling complexes (18).

Interestingly, only a small proportion of the directly up-regulated target genes of GR have been identified to carry a conventional GRE within 10 kb of transcribed genes (19, 20), suggesting that the vast majority of genes up-regulated by GCs are subject to other types of regulatory mechanisms. GC-induced gene expression is often enhanced via composite GC-responsive regions, in which binding of additional transcription factors allows an efficient induction of GC-mediated gene expression. Another level of regulation of transactivation is imposed by receptor modifications (21, 22). For example, GR phospho-isoforms have recently been found to selectively occupy promoters of some GR target genes but not of others (23).

Target genes that are negatively regulated by GR, via the transrepression mechanism or so-called tethering mechanism, most often involve the negative interference of GR with the activity of other DNA-bound transcription factors, such as nuclear factor (NF)-κB, cAMP response element-binding protein (CREB), interferon regulatory factor 3 (IRF3), nuclear factor of activated T cells (NFAT), signal transducer and activator of transcription (STAT), T-box expressed in T cells (T-Bet), GATA-3, and activating protein (AP)-1 (see below) (see Fig. 2). Typical target genes include a vast number of inflammatory proteins including IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-12, IL-18, cyclooxygenase (COX)-2, E-selectin, inducible NO synthase (iNOS), interferon (IFN) γ, TNFα, and intercellular adhesion molecule (ICAM), monocyte chemoattractant protein 1 (MCP-1) [chemokine (C-C motif), and ligand 2] vascular cell adhesion molecule (VCAM).

Fig. 2.

Targets of GR for immunomodulation. Hormone-activated GR is able to negatively regulate the activity of various other DNA-bound transcription factors, including among others NF-κB, CREB, IRF3, NFAT, STAT, T-Bet, GATA-3, and AP-1, via the transrepression mechanism or so-called tethering mechanism (factors inside circle). For AP-1, it has been described that GR can either negatively or positively influence its activity, depending on the composition of this dimer (see text for details). The pleiotropic GR is further capable of exerting its immune system modulatory effects through additional mechanisms (events depicted outside the circle). pol II, RNA polymerase II; TCR, T cell receptor. Arrow, Activating signal; blocked arrow, inhibitory signal; round arrow, modulatory signal, which can be either activating or inhibitory depending on the context, e.g. the loss of a coactivator or the recruitment of a corepressor molecule.

Transactivation vs. Transrepression: Separable Entities with a Clinical Benefit?

Exogenous GCs are used in the clinic to treat inflammatory, autoimmune, and allergic disorders; to attenuate organ rejection after transplantation; to treat brain edema, shock, and various blood cancers; and to balance out adrenal cortex insufficiencies. A number of synthetic analogs of the natural human GC, cortisol, have been developed by the pharmaceutical industry and include among others dexamethasone (DEX), betamethasone, triamcinolone, prednisone, prednisolone, and methylprednisolone (15). As stated above, their clinical success as effective antiinflammatory agents is largely attributed to their ability to reduce the expression of proinflammatory genes, via activation of the GR and the concomitant inhibition of the activity of proinflammatory transcription factors, including NF-κB and AP-1, through a mechanism called transrepression (24, 25, 26). Despite this, their use in the clinic is nevertheless compromised by the appearance of a range of side effects, which mainly arise from the ability of the steroid-activated GR to activate target genes involved in the metabolism of sugar, protein, fat, muscle, and bone via a mechanism called transactivation (27). Bona fide functional GR target genes in this respect include tyrosine aminotransferase (involved in amino acid catabolism), glutamine synthetase (involved in muscle catabolism), and glucose-6-phosphatase and phosphoenol pyruvate carboxy kinase (involved in gluconeogenesis) (27, 28, 29).

The therapeutic usage of GCs for the treatment of inflammatory and autoimmune disorders or, rather, the quest for ways how to specifically modulate the respective GR, has not been given up yet, at least not when judging from the plethora of recent work dealing with the characterization of novel selective GR modulators (28, 30, 31, 32, 33, 34, 35, 36, 37, 38). Furthermore, ongoing studies are carefully comparing the nuclear receptor selectivity profiles and benefit to side effect ratio of newer-generation GCs (39) (Fig. 1).

Fig. 1.

What steroid pharmacologists are aiming for, based on GR. Classical GCs, e.g. DEX and triamcinolone, elicit transrepression and transactivation mechanisms equally well. The latter event is deemed responsible for many unwanted effects, the so-called metabolic side effects. Dissociated steroidal ligands have been developed and are still being developed, which should mainly focus on the transrepression mechanism and stimulate the side effect pathway to a lesser extent, at least in specific tissues (smaller picture on the right), e.g. RU24858 and AL-438 (26 31 40 41 42 43 44 ). A newer generation of antiinflammatory drugs includes the nonsteroidal dissociated GR modulators. So far, they have shown promising benefit to side effect ratios, e.g. ZK216348 (27 ) and CpdA (24 29 ). The X means that in this category, some of them do not support transactivation. For example, for the plant-derived GR modulator CpdA, no GRE-dependent gene stimulation has been found so far, in vitro and in vivo (24 29 ). Note that the different ligands are able to impose different receptor conformations, partially explaining their differential effects on gene regulation.

Selective GR modulators are usually also termed as so-called dissociating ligands, e.g. displaying limited transactivation from a simple GRE but still able to transrepress transcription factors, of which AP-1 and/or NF-κB represent predominant targets. The initial belief that such dissociating compounds could be developed came from studies using a GR mutant with a defect in its dimerization capacity (A458T) and also in its subsequent DNA-binding and GRE-mediated transactivation ability (40) yet allowing transrepression to proceed normally. Upon the replacement in mice of wild-type GR with this dimerization-defective GR (creating a so-called knock-in strain called GRdim), it was found that endogenous GRE-dependent TAT and PEPCK expression was hampered, whereas transrepression mechanisms remained largely unaffected (41, 42, 43). In macrophages or T cells from GRdim mice, DEX still reduced the expression of TNF, IL-1β, IL-6, IL-2, and COX-2 (42). These findings, which supported the hypothesis that gene-activating and gene-repressing properties from GR can be uncoupled from one another, spurred the quest for dissociating ligands even more vigorously than before (Fig. 1). However, this viewpoint soon proved to be too optimistic, because GR functionalities, either beneficial or detrimental, do not necessarily display a similar degree of uncoupling. After establishing a dissociating profile, the need for a more extensive profiling of candidate compounds is nicely illustrated with the following example. RU24848 was characterized as a compound fulfilling the abovementioned criteria in vitro and shown to display antiinflammatory activities in vivo (44, 45, 46). However, this compound still elicited losses of weight and bone mass and was able to induce GRE-mediated lipocortin-1 expression in human eosinophils, gravely compromising its earlier described advantageous status over DEX or prednisolone (46, 47). Tanigawa and co-workers (48) proposed that the in vivo potency of the compound might be modulated by additional metabolism pathways and therefore may explain why the in vivo data do not necessarily correlate with in vitro data. Upon feeding GCs to mice that had been inoculated with various GRE-dependent reporter genes in mouse abdominal skin by means of a gene gun, a higher reporter gene activity was noted for RU24848 than for prednisolone (48). Also in liver cells, RU24848 was able to induce TAT gene expression, albeit to a different extent for different hepatoma cells, perhaps due to a differential ability between these cell lines to metabolize this compound (48). By contrast, in osteoblastic cells, this steroidal compound was a poor inducer of receptor-activator of NF-κB ligand (RANKL), of which the gene product is involved in stimulating bone resorption (49). From these examples, it can be concluded that the transactivation vs. transrepression characteristics are highly cell type and gene specific and, therefore, highly context dependent. Caution must therefore be taken when predicting the behavior of these compounds on a whole animal level, and more information should become available on what set of parameters are minimally required to make the most accurate prediction, especially concerning the clinically most important side effects, i.e. diabetes, osteoporosis, and growth retardation in children.

Different GC-inducible genes are in need of different aspects of GR functionalities. The most striking evidence for this assumption came from studying the expression of the phenylethanolamine N-methyltransferase (PNMT) gene, which is involved in adrenalin biosynthesis and which harbors a complex GRE promoter. Strikingly, whereas the dimerization-defective GR is unable to transactivate simple GRE-driven genes, the PNMT gene remains highly GC-inducible (50). The same goes for the expression of the antiinflammatory gene MAPK phosphatase 1 (MKP-1) (51). From a drug development perspective, the fact that a dimerization mutant is still competent for the activation of a subset of GR-dependent genes is an important complication. Again, the study of complex GRE-driven genes may be more indicative when investigating a novel dissociated compound for GR. In support and as already mentioned above, in a study investigating in depth the cell- and gene-specific determinants of transcriptional regulation by GR, the majority of GC-responsive genes was found to contain GRE sites that diverge from the simple GRE palindrome but are leaning more toward so-called composite elements (20).

Recently, more and more studies are questioning the benefit of ending up with a compound that would completely eradicate all GRE-driven gene expression (reviewed in Refs. 24 and 52). The reason for this is that there are actually quite a number of recently characterized genes that are being up-regulated by GR and that have distinct antiinflammatory roles. These include, besides the already known IκBα, the genes coding for MKP-1, lipocortin-1, secretory leukoprotease inhibitor (SLPI), type II IL-1 (decoy), annexin A1, IL-10, and GC-induced leucine zipper (GILZ), also some newly characterized proteins, including docking protein 1 (DOK-1), Dexras, p11/calpactin-binding protein, and tristetraprolin (TTP), which inhibit various stages of cytokine signaling, synthesis, secretion, and activity (24, 52). Recent studies using a phytomodulator ligand of GR, CpdA, exhibiting clear dissociated properties on GR signaling, may shed some light on this concern. It was found that CpdA could mediate transrepression of NF-κB-driven genes but did not support the transactivation properties of GR through GRE-driven gene expression, neither from simple nor from more complex promoters, exemplified by a study of the gene regulation of TAT, glucose-6-phosphatase, and PNMT (28, 33) (our unpublished results) (Fig. 1). Yet, CpdA was able to inhibit the progression of rheumatoid arthritis in a CIA mouse model to a quite reasonable extent as compared with DEX, one of the strongest agonists of GR (28).

Leaving the dogma behind that only molecules with modified steroidal scaffolds can activate GR through binding in the ligand-binding domain pocket, a novel and broad way has been paved for the development of various classes of safer GR modulators. Because a steroidal backbone may still allow binding to other steroid receptors, e.g. mineralocorticoid receptor and progesterone receptor, thereby mediating side effects through the activation of other hormonal pathways, an additional advantage of nonsteroidal ligands may be an increase in target specificity. For example, the recently characterized antiinflammatory compound benzylidene LGD5552, which binds GR ligand-binding domain and displays antagonistic activities on the mineralocorticoid receptor, has no effect on the mean arterial blood pressure in rats (38). A reduced impact on blood pressure in patients would certainly be advantageous over the standard GC treatment.

The quest for novel GR modulators displaying a specific gene expression profile may, however, be complicated by the fact that a number of genes may contribute both to antiinflammatory and side effects, depending on the target tissue, e.g. MKP-1 (or DUSP1), MIF and AnxA1, or even in a single cell type, as is e.g. demonstrated for GILZ. GILZ is often used as a paradigm for GC-induced gene expression, because it displays a very good GC inducibility and because it has well-characterized functional GREs in its promoter sequence (53). Indeed, knockdown experiments demonstrated a contribution of GILZ to the inhibition of IL-8 gene expression in endothelial cells (54), whereas GILZ-induced stimulation of the expression and activity of ENaCα implicates a role for this protein in GC-induced hypertension (55, 56). Likewise, the fact that AnxA1^−/− mice demonstrated impaired responses to GCs in carrageenan-induced edema, antigen-induced arthritis, and zymosan-induced peritonitis supported their antiinflammatory role, but AnxA1 has also been negatively implicated in mediating suppression of the HPA axis, which leads to the undesirable effect of adrenal insufficiency (57). A plethora of evidence supports a clear antiinflammatory role for MKP-1, but recent reports suggest that MKP-1 may also be involved in GC-induced osteoporosis as well as metabolic dysregulation (52). It is noteworthy that in fact all three of these putative antiinflammatory mediators were strongly up-regulated by the dissociated steroidal compound RU24858 (47, 58), putting a large question behind the truly dissociated character, depending on which cell types are studied. As a final example, macrophage migration inhibitory factor (MIF), a GC-inducible proinflammatory cytokine, has been implicated in the pathogenesis of both rheumatoid arthritis and atherosclerosis (59).

An important clue toward novel ligand-screening approaches is given by the work of Coghlan et al. (60). A modified progestin, AL-438, retained its competence for transcriptional repression of NF-κB-driven genes, yet exhibited reduced side effects. AL-438 differentially affects gene expression by reducing the interaction between GR and PPARγ coactivator 1 (PGC-1) but maintaining the interaction between GR and GR interacting protein 1 (GRIP1)/transcription intermediary factor 2 (TIF2).

Consequently, because PGC-1 seems preferentially used in steroid-mediated glucose up-regulation (60) and because GRIP1 is implicated in the suppression of inflammatory gene expression (61), a better side effect profile is being generated. It would be interesting to explore gene expression patterns differentially affected by these two and also other relevant GR-associated cofactors in different tissues.

A number of genes exist that are transrepressed by GC and that reside in the osteoporosis side-effect circuitry, e.g. osteocalcin and osteoprotegerin. Considering the complexity of pathways regulated by GR, including cell proliferation, differentiation, and apoptosis at the cellular level or immune cell homeostasis, metabolism, and responses to stress pathways at the level of the organism, it is clearly too naive to assume that an ideal exogenous GR modulator, only eliciting the beneficial antiinflammatory effects without any trace of side effects, will ever be found. Life is all about balances. Having said that, it is nevertheless highly recommended not to give in and to try and relieve the suffering of many patients dealing with various inflammatory disorders, via solving the puzzle as to how GR mediates its antiinflammatory effects. With this information, better GR ligands may be able to replace the ones we commonly use now.

Mechanisms of Cross Talk between GR and Proinflammatory Transcription Factors

The way a cell responds to GCs is determined not only by the nature and amount of the ligand but also by a modulation of the signaling capacity of GR through interaction with other signaling pathways. This interaction can occur in the cytoplasm by interference with the activity of various signaling proteins (kinases, phosphatases, etc.) or in the nucleus by interfering with the DNA-binding or transactivation capacity, transcription factor-cofactor interactions, or transcription factor-kinase interactions or through inhibiting contacts with the general transcription machinery by the targeted transcription factors.

Transcription factor-transcription factor interactions

The most-studied cross talk mechanisms are the ones between GR and NF-κB or GR and AP-1 because they form a clear basis for the GC-mediated inhibition of various inflammatory cytokines (e.g. IL-6, IL-1β, and TNFα), enzymes (e.g. iNOS, COX-2, and MMPs), and adhesion molecules (e.g. ICAM-1, VCAM, and E-selectin), which all have one or more NF-κB and/or AP-1 elements in their gene promoters (reviewed in Refs. 15 , 62 , and 63). The cross talk mechanism is not restricted to these well known transcription factors, but has in recent years been expanded to other factors as well, including CREB, NFAT, STAT, and T-Bet. Recently, it was found that GCs are able to block the activity of T-Bet, a transcription factor with a role in T-cell differentiation and inflammation and that can drive expression of e.g. the IFNγ target gene. The molecular mechanism involves a direct interaction between T-Bet and GR and a diminished DNA binding of T-Bet as well as a down-regulation of T-Bet mRNA and protein expression in T cells (64).

A simple direct physical protein-protein interaction was indeed also among the first described mechanisms, explaining cross talk between GR and NF-κB (65, 66).

In general, in terms of the mechanism GR is deploying for protein-protein interactions, ample evidence exists that GR would block the activity of DNA-bound transcription factors in its monomeric form rather than as a homodimer and without contacting the DNA itself (28, 41). As described above, antiinflammatory effects have already been ascribed to GILZ in immune cells (67). Recently, it was shown that the effects of GILZ are not restricted to T cells but are also apparent for airway epithelial cells, in which the knockdown of GILZ leads to a desensitization of GC-mediated chemokine repression (54). The mechanism by which GILZ acts is dual; besides binding to NF-κB and AP-1 family members, it can also associate with Raf-1, blocking its potential to activate downstream ERK MAPKs (68, 69) and through this way effectively blocking inflammatory gene transcription. Intriguingly, only the interaction between GILZ and NF-κB requires GILZ homodimerization through their leucine zipper domains (70). Recently, multiple isoforms of GILZ have been described, GILZ1 to GILZ4, explaining previously reported distinct roles of GILZ in cellular proliferation and ion transport mechanisms (71).

Interactions between GR and c-Jun have long been known to modulate AP-1 activity. In most cases, the outcome is a down-regulatory effect (72, 73, 74), exemplified in fibroblasts for the c-jun gene promoter (75). However, in T lymphoblasts, the cross talk between GR and AP-1 rather results in an enhanced transcription of the c-jun gene (76, 77) (Fig. 2). Thus, although the underlying mechanism governing the immunosuppressive effect of activated GR with NF-κB or of activated GR with AP-1 seems highly similar, important differences exist. This is further documented by the fact that a mutation in the first zinc finger of GR can affect NF-κB but not AP-1 inhibition (78). Vice versa, a GR point mutant exists, namely GRR488Q, that is unable to repress NF-κB yet does repress AP-1 (79).

Although it was generally assumed and shown before that GR is able to repress AP-1-dependent transcription independent of the composition of AP-1 subunits (80), the group of Kassel recently presented evidence that only Fos-containing dimers are transrepressed by GR, additionally involving a role for nTrip6, a nuclear isoform of the LIM-domain protein Trip6 (81). The underlying reason for this discrepancy has so far not been studied; one explanation may be that different cell lines were used in the two different studies.

At any rate, cross talk between NF-κB and GR also does not necessarily lead to a mutual inhibition. For some genes, activation of both NF-κB and GR results in a cooperative environment. In fact, gene profiling studies demonstrated both enhancing and suppressive effects of GCs on immune cells (82). For example, the promoter for Toll-like receptor 2 (TLR2) is cooperatively stimulated by GCs and TNFα, through the necessary presence of a functional NF-κB, a 3′-GRE, and a STAT-binding element (83). In terms of functionality, TLR2 is a transmembrane receptor, acting as a so-called pattern recognition receptor for diverse bacteria and thus making part of the innate immunity defense system. It is in this respect important to note that in the complex transcriptional networks stimulated by both TLRs and GCs various other cross talk mechanisms have been identified that cosupport the emerging role of GR in the regulation of innate immunity (84). Transrepression mechanisms by GR in macrophages further displayed a highly signal- and gene-dependent character, allowing the receptor to differentially modulate pathogen-specific gene expression pathways. As such, activated GR could affect TLR4- and TLR9-driven gene activation but not TLR3-dependent gene activation. Other nuclear receptors with antiinflammatory activities, including peroxisome proliferator-activated receptor γ (PPARγ) and liver X receptor (LXR), even demonstrate a synergy with GR to transrepress a specific subset of TLR-dependent genes (85). Finally, very recent data indicate that a cross talk mechanism exists between GR and PPARα, whereby PPARα cooperates with the activated GR for transrepression on NF-κB but was found to block GR-mediated transactivation (our unpublished results). Taken together, these recent findings not only illustrate the combinatorial control mechanisms used by nuclear receptors to restore immune homeostasis but may also hold the key for the development of future therapeutics.

Transcription factor-cofactor interactions

The family of p160 coactivators, including, e.g. steroid receptor coactivator 1 (SRC-1) and GRIP1, has been described to act as adaptor proteins bridging the GR with other cofactors, e.g. p300 and CREB-binding protein (CBP) (which are histone acetyl transferases), and protein arginine methyltransferase 1 and coactivator-associated arginine methyltransferase (which are histone methyl transferases). These cofactors thus modulate the activity of GR and undoubtedly also of its diverse isoforms in a tissue-specific manner (reviewed in Ref. 86).

Because of their newly discovered role in alternative splicing processes, it recently became clear that cofactors not only influence the abundance but also the nature of their products (87).

Furthermore, the spatial and temporal mode in which the process of cofactor recruitment occurs varies not only for different nuclear receptors but also, using the same nuclear receptor, for different promoters (87). With this new information, the picture once again becomes blurry on how exactly coregulators are involved in the regulation of cytokine gene repression mechanisms by GR or how redundant the functionalities of the plethora of GR-interacting cofactors may be.

Before, the differential gene regulation induced by nonsteroidal GR ligands has largely been attributed to selective interaction with a number of coactivator or corepressor molecules. GC-mediated transcriptional repression of critical inflammatory genes has been linked to interaction with corepressors in some cases (88) and coactivators in other cases (61). Also, cofactor competition events between GR and proinflammatory transcription factors have been described (reviewed in Ref. 62). Clearly, in light of the recent findings on cofactor functionality, the research area studying their impact on GR-mediated gene regulation mechanisms will benefit from novel investigations. One important strategy is to define, through peptide array analyses, which cofactors are crucial determinants in governing the behavior of GR as a gene repressor by using dissociated ligands.

Histone deacetylases (HDACs) are another category of molecules capable of modulating GC sensitivity. They form the counterpart of histone acetyltransferases, because they reverse histone acetylation events. Some time ago, HDAC2 was described to be recruited to inflammatory promoters upon GC treatment (89), helping to explain gene inhibitory effects. However, their territory has recently been expanded with their ability to deacetylate nonhistone proteins. Indeed, by targeting acetylated GR, HDAC2 is capable of potentiating the inhibitory effect of GCs (90). Moreover, the deacetylation of GR by HDAC2 seems critical for the interaction between p65 and the receptor (91). Not only (de)acetylation events are important, a complex interplay between histone-modifying HDACs, phosphatases, methylases, and heterochromatin protein 1α culminates at the SP-A gene promoter in response to DEX and contributes to a closure of the chromatin structure, concomitant with the GR-mediated inhibition of gene activation (92).

Another level of control on GR signaling is imposed by the orphan receptor dosage-sensitive sex reversal-adrenal hypoplasia congenita critical region on the X-chromosome, gene 1 (DAX-1) (93), which can function as a corepressor and has been described to negatively regulate GC production. Recently, it became clear that DAX-1 selectively inhibits GR-mediated transactivation but not GR-mediated transrepression through a mechanism involving competition for GRIP1 coactivator binding to the GR (94).

Interference with upstream signaling components

Under physiological conditions, gene induction of, e.g. AP-1 or NF-κB target genes, is the sum of a simultaneous and parallel activation of multiple kinases (e.g. MAPK and IκB kinases), the phosphorylation of more than one transcription factor, and the transmission of signals to a given gene through multiple sequence elements (95). Taking into account that kinases are not the only signal-stimulated and transcription factor-modifying enzymes but that also histone and factor (de)acetylases can play a role, a plethora of possible combinatorial events further adds to the complexity of the studied system of transcriptional initiation. Adding on top of this an extra inhibitory signaling pathway, i.e. through the activated GR, functional interference can take place at many different key points, ultimately resulting in an efficient transcriptional inhibition of target genes.

Inflammatory signaling is propagated via a kinase-activating cascade, including at the almost distal end the MAPKs p38, ERK, and c-Jun N-terminal kinase (JNK). All of the MAPKs have been identified as potential targets for the antiinflammatory actions of GCs through blockage of their activating phosphorylations (Fig. 2). Which MAPKs are preferentially targeted seems to be cell type and GR ligand dependent (52, 96, 97, 98, 99). Still in terms of the immunosuppressive action of GCs, but dependent on novel protein synthesis, it was found for some cell types that GCs inhibit the cytokine-induced phosphorylation and activation of MAPKs p38 and JNK via the up-regulation of MKP-1, a dual-specificity phosphatase, which has been characterized as a negative feedback mechanism restricting inflammatory and innate immune responses (51) (Fig. 2). In MKP-1^−/− murine macrophages, GC treatment could no longer inhibit lipopolysaccharide-induced JNK and p38 activation or repress the synthesis of typical inflammatory mediators COX-2, TNFα, and IL-1β (51). MKP-1 absence was further also associated with an increased lethality in response to endotoxin (100) and with a marked increase in the frequency and severity of the murine collagen-induced arthritis model (101). The precise mechanism by which GCs up-regulate MKP-1 is as yet unknown, because the proximal promoter region lacks obvious classical palindromic GRE sequences. However, from studies employing primary macrophages from GR dimer mutant mice, it was found that endogenous MKP-1 expression still responded to GCs (51), suggesting a mechanism similar as for the PNMT gene, namely that GR monomers can still transactivate on promoters bearing concerted GREs (50). Alternatively, in endothelial cells, the rapid DEX-induced stimulation of both ERK and JNK MAPK (the latter activated through DEX-mediated generation of reactive oxygen species), leading to activation of CREB and AP-1, was suggested to be involved in the DEX-mediated up-regulation of MKP-1 (102).

New kinase targets of GCs are still being discovered. The synthetic GC DEX was reported to inhibit phosphorylation of TNF receptor-associated factor (TRAF) family member-associated NF-κB activator (TANK)-binding kinase 1 (TBK1) and subsequent TBK1 kinase activity. Because TBK1 is required for the activation of IRF3 downstream of stimulation of both TLR3 [responsive to the double-stranded RNA mimic poly(inosine-cytosine)] and TLR4 (activation in response to lipopolysaccharide) (103, 104), these data illustrate a novel level of antiinflammatory regulation by GCs (Fig. 2).

Vice versa, cytokine-activated MAPK signaling can phosphorylate the GR protein itself, thereby modulating its turnover and its transcriptional activity (105, 106), imposing an extra layer of regulation on the functionality of GR. This mutually antagonistic cross talk mechanism may well contribute to the occurrence of steroid insensitivity (107, 108), further illustrating the need for new GR modulators. The detrimental effects of MAPK in GC resistance are proposed to be attributed to an altered phosphorylation status of the receptor, affecting GR ligand binding, Hsp90 interactions, subcellular localization, and transactivation potential of the receptor (91). The phosphorylation status of GR can further also affect the magnitude of repression by GR in a gene-selective manner (109). At the level of diverse pathway integrations, a cross talk between JNK and small ubiquitin-like modifier (SUMO) pathways was recently found to modulate GR transcriptional activity, albeit again in a target-gene-specific manner (110).

Not less important to the inflammatory cascade are second-stage activated cytokines such as IL-6 that signal through the Janus kinase (JAK)/STAT pathways, which in turn are subject to a negative feedback by suppressors of cytokine signaling (SOCS) proteins. The latter proteins target the signaling Janus kinases for degradation (111). SOCS-1 has not only been reported to interfere with TLR signaling, but its mRNA levels are also up-regulated by GCs in hematopoietic cells and immune cell cancers, via a presently unknown mechanism (84) (Fig. 2). A responsiveness of the multiple GRE elements in the promoter region of SOCS-1 remains to be explored. A cross talk between GR and SOCS-1 also manifests itself at an entirely different level: SOCS-1 was recently reported to negatively influence transcription of FKBP5 and MKP-1, two GR-regulated target genes, possibly via a direct interaction between GR and SOCS-1 (112).

In a number of cell types, including immune cells, GCs block the activity of transcriptional activity of NF-κB via the up-regulation of the cytoplasmic inhibitor IκBα, leading to the retention of NF-κB in the cytoplasm (113, 114). For years, the underlying mechanism represented a puzzle, because no GRE could be found in the IκBα promoter. Later on, the group of Archer (115) found that the mechanism involves a higher accessibility and stability of transcription factor binding at the promoter (that already bears an open chromatin structure in absence of hormone), leading to effective gene activation.

Recently, a completely novel aspect within the action mechanism of GC transrepression, specifically for NF-κB-driven gene expression, was unveiled. It was found that GCs are able to modulate the chromatin environment and the functionality of the inflammatory enhanceosome via targeting the nuclear kinase mitogen- and stress-activated protein kinase-1 (MSK1). GC-mediated gene repression was found to involve a loss of MSK1 recruitment at inflammatory gene promoters, causing inhibition of NF-κB transactivation and H3 S10 phosphorylation (116). Earlier reports had identified MSK1 as the targeting kinase of H3 S10 (117, 118, 119). Together with histone acetylation, histone H3 S10 phosphorylation contributes to the conformational change of chromatin from a so-called closed to a more open configuration, allowing transcription factor access and formation of the preinitiation transcription machinery. MSK1, a downstream target of ERK and p38 MAPK, was also characterized before as an essential NF-κB p65 S276 kinase (119). Phosphorylation of this serine residue mediates CBP histone acetylase effects, ensuring an optimal expression of NF-κB target genes (120). MSK1 was further found to associate with GR, and interestingly, a substantial amount of activated MSK1 is observed to be exported to the cytoplasm in a GR- and chromosomal region maintenance 1 (CRM1)-dependent manner, revealing a completely novel aspect within the molecular mechanism of GC-dependent inhibition of NF-κB (116) (Fig. 2). It is still unclear whether this specific fraction of MSK1 is merely exported to the cytoplasm and being targeted for degradation or, alternatively, is subject to a continuous shuttling mechanism, perhaps via a tightly regulated (de)phosphorylation event.

Notwithstanding some exceptions (121), GR activation in general does not block promoter recruitment of the targeted transcription factors NF-κB or AP-1, as demonstrated by chromatin immunoprecipitation analysis and genomic footprinting experiments (33, 121, 122, 123). A couple of years ago, Luecke and Yamamoto (124) found that the IL-8 promoter-bound GR competes with the Cdk9 and cyclinT complex, positive transcription elongation factor b (P-TEFb), for recruitment at the IL-8 promoter, thus impeding the P-TEFb-mediated phosphorylation of the RNA polymerase II C-terminal domain on the S2 residue. The combined data possibly entail a dual mechanism in which GR may inhibit both the transcription-facilitating phosphorylation of H3 S10 and of RNA polymerase II C-terminal domain (CTD) S2, by blocking the recruitment of both MSK1 and P-TEFb, respectively (Fig. 2). These results also still fit in the framework of a protein-protein interference work model in which GC-mediated inhibition of NF-κB-activated genes involves the association of GR and p65 NF-κB.

Other mechanisms

Another level at which GCs contribute in eliciting immunosuppression is by promoting apoptosis of a subset of immune cells. In accordance with the finding that in GRdim mice, thymocyte apoptosis was compromised, GCs are mainly believed to induce proapoptotic genes in thymocytes and T cells, e.g. thioredoxin-interacting protein (Txnip) (125), a mechanism supplemented with GC-mediated repression of antiapoptotic factors or by GC-mediated posttranscriptional mRNA destabilization of positive cell cycle genes, e.g. cyclin D₃ (126, 127).

The GC-mediated immunosuppressive effect on T cell activation is recently explained via an alternative mechanism: after ligand activation of GR, a physical interaction between GR and the T cell receptor (TCR) complex is disturbed, leading to impaired T cell signaling (128) (Fig. 2).

Furthermore, in the rapidly evolving field of nongenomic actions of GCs, Buttgereit and colleagues (129) showed that high concentrations of GCs can intercalate into the plasma membrane of immune cells, thereby interfering with calcium and sodium cycling across the membranes. Although the mechanistic details of these rapid actions are still lacking, steroids do seem to be able to increase several second messengers such as inositol 1,4,5-trisphosphate, cAMP, and Ca²⁺ (130).

Conclusion

This review has aimed to highlight the most recent findings on the molecular mechanisms of GR and has tried to put these findings in context of the ongoing quest for novel selective GR modulators, which will aid in the fight against various inflammatory disorders. We have discussed the complexity and difficulties researchers are facing when developing novel strategies to combat chronic inflammatory disorders, when choosing GR as a target molecule. Simple cellular experimental models investigating the dissociated character of novel GR modulators do not suffice to accurately predict the therapeutic index. Complementing genome-wide gene profiling studies and transcription factor/DNA-binding patterns on various target tissues at once will become an adamant strategy for the future.

NURSA Molecule Pages:

• Coregulators: AR | DAX1 | LXRα | LXRβ | PPARα | PPARγ.