26 Multi-site tlierapeutic modaiities for infiammatory bowel diseases - meclianisms of action

(1)

26 Multi-site tlierapeutic modaiities for infiammatory bowel diseases - meclianisms of action

GERHARD ROGLER

Introduction

Genetic susceptibilities such as the recently identified polymorphisms in the N0D2 gene, as well as envir- onmental factors such as certain bacteria, play a role in the etiology of inflammatory bowel diseases (IBD). Over the years many factors have been identified that contribute to the pathogenesis and the process of inflammation in Crohn's disease (CD) and ulcerative colitis (UC), leading to new therapeutic concepts. Among these contributing factors are cytokines and chemokines, as well as adhesion molecules, which are relevant for the emi- gration of immune cells into the intestinal mucosa.

Mucosal and systemic concentrations of many pro- inflammatory cytokines are elevated in IBD. An inadequate, and/or prolonged activation of the intestinal immune system plays an important role in the pathophysiology of chronic mucosal inflammation. An 'imbalance' between proinflammatory and anti-inflammatory cytokines has been described in the inflamed mucosa of patients with CD and UC.

Many, if not all, of the involved cytokine-mediated pathways have back-up systems. Therefore a therapeutic intervention at a particular, singular, very specific point in the complex network of cytokine and chemokine interactions with each other or their receptors is frequently less likely to be successful than a multi-site targeted anti-inflammatory strategy [1].

Even when the etiology of IBD is completely elucidated a causative therapy might not be possible, and the multi-site anti-inflammatory strategy could still be favorable. Therefore the improvement of 'classic' multi-site targeted anti-inflammatory therapies, as well as the development of new concepts for this approach, is of great importance for the future management of patients with IBD. The improvement of 'classic' concepts and therapies is necessary, as so far no therapeutic strategy has proved successful in

all patients. Clinically we observe a 'resistance' of a certain percentage of patients to any particular therapy. An important goal for the future must be a better understanding of mechanisms leading to a relative resistance to classic multi-site anti-inflammatory strategies. This would allow the early identi- fication of patients who are not likely to respond to a treatment modality, would avoid frustrating treat- ments for the patient and the physician, and could allow more specific concepts for the individual patient. To understand the mechanisms of resistance to a particular multi-site targeted anti-inflammatory therapy we first have to understand the molecular and cellular mechanisms involved in an effective therapy. Important insights into a number of those mechanisms of drug action have been made in the last two decades.

A classic example for a multi-site targeting anti- inflammatory treatment is therapy with glucocorticoids. Therefore the principles of glucocorticoid therapy, the structure of the effect-mediating receptor, the molecular mechanisms of action and the effects on the cellular levels will be highlighted first in this chapter. The glucocorticoid receptor (GR) is a member of a family of receptors, the so-called nuclear receptor superfamily, which shares struc- tural and functional similarities. Other members of this family are the peroxisome proliferation-activated receptors (PPAR). Ligands to PPARy have recently been shown to be effective in animal models of IBD whereas potential ligands of PPAR-oc were beneficial in a clinical study in active IBD. The inhibition of the proinflammatory transcription factor nuclear factor kappa B ( N F - K B ) is likely to be the most important target of glucocorticoid therapy as well as PPAR-mediated effects. Therefore, finally the mechanisms and pathways of N F - K B activation, as well as the consequences of N F - K B inhibition, will be explained.

Stephan R. Targan, Fergus Shanahan andLoren C. Karp (eds.J, Inflammatory Bowel Disease: From Bench to Bedside, 2nd Edition, 523-551.

(2)

Glucocorticoids

Glucocorticoids are used for the suppression or reduction of inflammation in a wide variety of diseases such as rheumatoid diseases, allergic diseases, IBD and in general autoimmune diseases [2-8]. In many of these cases they are still the standard or first-line therapy due to their high efficiency;

however, their use is limited by systemic side-effects.

The understanding of the mechanisms by which glucocorticoids suppress or reduce inflammation has increased dramatically during recent years [9-14].

Glucocorticoids have been proven to be the first choice in the treatment of acute flares of IBD in several major studies [15-23]. The systemic administration of glucocorticoids (orally or intravenously) during the acute exacerbation of CD or UC is followed by a multitude of different effects in different body cells. One of the intended eff'ects is the down-regulation of proinflammatory cytokines [24], This mechanism is part of the feedback system between inflammation-derived cytokines and the central nervous system-adrenal axis regulating corticosteroid synthesis with the physiological rele- vance to balance host defense and anti-inflammatory systems of the body [25, 26]. Among the molecules down-regulated by GR action are multiple cytokines and their receptors, chemokines and their receptors, kinins and their receptors, adhesion molecules and inflammation-associated enzymes such as inducible nitric oxide synthase (iNOS) and the inducible cyclooxygenase (COX-2) [3, 27].

Principal meclianisms of giucocorticoid action Most, if not all, effects on cells of naturally occurring glucocorticoids such as Cortisol or synthetic cortico- steroids such as prednisolone and its methylated or acetylated derivates (triamcinolone, dexamethasone or beclomethasone) are mediated by binding to cytosolic GR. GR are present in almost all body cells in concentrations between 2000 and 30 000 binding sites/cell [28].

The GR is a member of the nuclear receptor superfamily, which also includes PPARoc, PPARy and PPAR5 (see below). GR consists of 777 amino acids (AA) and was cloned in 1985 [29-31]. There is only a single GR binding glucocorticoids, with no evidence for subtypes of differing affinity in different tissues [3]. A splice variant of GRoc, termed GRP, has been identified that is not able to bind glucocorticoids (see below).

Normally the interaction of glucocorticoids with GRa is followed by an activation and dissociation of GR from its inhibitory protein complex and by a translocation of the receptor into the nucleus, where the complex of steroid and receptor interacts with promoter regions of different genes, finally leading to an increase or decrease of gene transcription (Fig. 1).

This process is very similar within the family of nuclear receptors, which show a high degree of genetic similarity with 40-90% of identical AA sequences [32-34].

An important problem for the therapy of patients with IBD and in general chronic inflammatory diseases is the occurrence of glucocorticoid responders and non-responders. Non-response to glucocorticoids may be mediated by mutations of the receptor, a reduced number of GR or a down-regulation of the receptor [35]. Studies demonstrated that primary (hereditary) abnormalities in the GR gene make only 2.3% of patients with asthma relatively 'resistant'.

'Resistance' to the beneficial clinical effects of glucocorticoid therapy in patients with IBD, therefore, is probably rarely related to generalized primary (hereditary) glucocorticoid resistance. In the majority of patients with rheumatoid arthritis or asthma the glucocorticoid resistance seems to be acquired and localized to the sites of inflammation, where it reflects high local cytokine production, which inter- feres with glucocorticoid action [36]. One of the basic mechanisms that could be responsible for glucocorticoid resistance, the competition for co-factors, will be explained in detail later.

Glucocorticoid levels (^max) and binding affinities (A^d) vary among patients and have been correlated to patient response. A certain threshold level of GR is necessary for glucocorticoid responsiveness [37]. In patients with rheumatoid arthritis a decrease of systemic GR in leukocytes has been found [38].

However, the GR density did not correlate with inflammatory disease activity.

Ttie molecular structure of glucocorticoid receptor

The mechanisms mediating the glucocorticoid response via GR are complex. Regulation of gene expression and GR actions occurs at several levels in the body and the single cell. However, the glucocorticoid-mediated anti-inflammatory mechanisms are a classic example for the transmission of basic research to the bedside. It is important to first understand the structure of GR and then the molecular

(3)

i l l l l i i i l l l cell membrane

nuclear membrane

DNA

^ translation

iiiiiiiilii

new protein,

altered cell function

Figure 1. Cellular mechanisms of glucocorticoid receptor action. After binding of glucocorticoid(s) to its receptor hsp90 is released from the complex. This is followed by an exposure of the nuclear localization signal, homodimerlzation and a rapid translocation of the activated GR/glucocorticoid complex to the nucleus where it can bind its response elements (glucocorticoid response elements, GRE) and mediate fra/7s-activation or frans-repression of gene transcription.

mechanisms of glucocorticoid action. G R a normally is a phosphorylated 92 kDa protein, which in its inactivated form is bound into a complex with other proteins. Like most members of the nuclear receptor superfamily GR contains a DNA-binding site (DBD) consisting of two zinc-finger domains [10]. This area contains an invariant pattern of eight cysteins arranged in two groups of four, so as to coordinate the binding of two zinc atoms [39-41]. The DBD is located in the middle of the GR molecule (Fig. 2).

The two zinc-finger motifs in the DBD induce the formation of a tertiary structure containing helices that interact with specific DNA sequences organized in 'GR response elements' [42-47]. Between the DBD and the carboxyl-terminal ligand-binding domain (LBD) is a so-called 'hinge region'. This region of the GR protein contains a nuclear localization signal, a binding region for heat-shock protein 90 (hsp90) and a second ^ra^^-activating domain (activation function 2, AF2) [3].

The carboxyl-terminal LBD not only represents the specific binding sites for glucocorticoids but also serves as a h o m o d i m e r l z a t i o n d o m a i n [10].

Additionally it interacts with other proteins regulating GR activity.

The N-terminal part of GR contains another trans-2iCii\2iiion domain (activation function 1, A F l ) that plays an important role in gene regulation [48-50]. This structure may be especially important during trans-a.ctiv3.tion but not ^ra^^'-repression activity [51-54]. It is not yet clear how A F l interacts with other proteins to induce transcription. AF-1 has been reported to interact with the basal transcriptional machinery [55, 56] but also with other adaptors or co-activators [57-60].

The inactivated GR is bound to a protein complex of approximately 300 kDa. It includes two molecules of hsp90, a 59 kDa immunophilin protein and other inhibitory proteins [3] (Fig. 2). The hsp90 proteins cover the nuclear localization site, preventing the unliganded GR from localizing to the nuclear compartment. After binding of the specific ligand to the LBD the tertiary structure of the molecule changes, followed by a release of hsp90 from the complex. This leads to an exposure of the nuclear localization signals and a rapid translocation of the activated GR-glucocorticoid complex to the nucleus where it can bind its response elements (glucocorticoid response elements, GRE).

(4)

COOH

gene activating domain

inhibitory

protein hormone binding site

hinge region

steroid

hormone

I

DNA-binding domian

.i

_{/ "}

DNA'binding site expos^

1

COOH

Steroid hormone

Figure 2. Molecular structure of the glucocorticoid receptor. The N-terminal part of GR contains a fra/7s-activation domain that plays an Important role in gene regulation. The DNA binding site (DBD) consists of two zinc-finger domains and contains an invariant pattern of eight cysteins arranged in two groups of four. Between the DBD and the carboxyl-termlnal ligand binding domain (LBD) is a so-called 'hinge region'. This region of the GR protein contains a nuclear localization signal, a binding region for heat-shock protein 90 (hsp90) and a second frans-activating domain. The C-terminal LBD not only represents the specific binding sites for glucocorticoids but also serves as a homodimerlzatlon domain. Additionally it Interacts with other proteins regulating GR activity.

The inactivated GR is bound to a protein complex of approximately 300 kDa Including two molecules of hspQO. The hspQO proteins cover the nuclear localization site preventing the unliganded GR from localizing to the nuclear compartment. After binding of the specific ligand to the LBD the tertiary structure of the molecule changes, followed by a release of hspQO from the complex.

Molecular mechanisms of glucocorticoid receptor action

Activated and nuclear-translocated GR can act in two principal ways. It can mediate /ra^i^-activation of gene transcription leading to an increased mRNA expression or /ra«^-repression followed by a down- regulation of shut-off of a certain gene product.

Transactivation

In the nucleus GR can bind to classical GRE to activate transcription of the response gene [61-63]

(Fig. 3A). The GRE has a palindromic motif (consensus sequence: GGT ACA N N N TGT TCT).

Usually GR binds this or similar DNA sequences cooperatively as a homodimer (Fig. 3A). For homo- dimerization interaction of a group of five AA known as the dimerization or D loop is needed. They are located within the DBD of the GR molecule and are essential for dimerization and transcriptional activation.

A direct /mA?.v-activation of gene transcription by GR has been described for IL-I type II receptor that binds IL-1 without induction of signal transduction, thus preventing cells from activation and inflammatory reaction [64-67] (Fig. 3A). Another important protein whose transcription is /ra^^-activated by GR is iKBa, the inhibitory protein for the proinflammatory transcription factor N F - K B (see below).

For other genes such as serine protease inhibitor 3 or arginase a cooperative /ra«^-activation involving GR and another transcription factor such as C/EBP or AP-1 has been described (Fig. 3B). This means that binding of activated and translocated GR is necessary but not sufficient for the increased transcription of these genes. However, as soon as both, GR and the cooperating transcription factor (C/EBP or AP-1) are bound to the promoter, increased gene- transcription takes place.

(5)

Figure 3. Frans-activation of gene transcription by glucocorticoid receptor. A: GR binds as a homodimer to classical GRE to activate transcription of response genes. The GRE has a palindromic motif (consensus sequence: GGT ACA NNN TGT TCT). A direct trans- activation of gene transcription by GR has been described for IL-1 type II, I K B , IL-1 RA or lipocortin I. B: For other genes such as serine protease inhibitor 3 or arginase a cooperative fra/]s-activation involving GR and other transcription factors like C/EBP has been described.

GR-mediated rra/t^-activation of genes is also involved in the induction of apoptosis in T cells by dexamethasone [68-78]. The detailed mechanisms mediating the induction of apoptosis are not yet elucidated.

In addition to these trans-?iCi\\diimg actions described, which usually take several hours, GR have more rapid ^ra^i^-repressing activities.

7A5/7S-repression via DNA binding

After characterization of positive transcriptional regulating GRE the existence of negative GRE sites (nGRE) mediating a negative regulation of transcription (/raw^-repression) via glucocorticoids was postulated [63]. However, the concept of a nGRE site is still a matter of discussion, as the consensus binding site is variable and described for only a few genes [13]. A binding to a promoter sequence and subsequent ^ra^^'-repression by GR has been shown for the prO'Opiomelanocortin (POMC) gene, an ACTH precursor allowing a negative feedback circle [79, 80] (Fig. 4A). Another promoter containing a negative GR binding site (at -278 to -249 of the

promoter) is the corticotropin-releasing hormone (CRH) promoter [81, 82]. Thus it seems that direct

^ra«^-repression activity is reserved for the negative feedback circle in the hypothalamic-pituitary-adrenal axis.

GR and /raw^-activating transcription factors may also compete for binding sites in promoters [83]. In this case the presence of GR blocks ^ra^^-'activation by another factor and does not actively rra^^-repress transcription itself (Fig. 4B). In the osteocalcin pro- moter GR overlaps the TATA box. Activation and nuclear translocation of GR can therefore prevent the binding of the basal transcription factor, TATA- binding protein (TBP), which is necessary for the recruitment of RNA polymerase II and initiation of transcription [84]. This may be one of the reasons for the occurrence of osteoporosis during long-term glucocorticoid therapy.

7?'5/?s-repression without DNA binding

Besides the competition for binding sites in promoters a rra^^-repression mechanism for GR has been described that does not involve DNA binding (Fig.

(6)

B

Figure 4. frans-represslon of gene transcription by glucocorticoid receptor. A: A binding to a negative glucocorticoid response element (nGRE) and subsequent fra/7S-repression by GR has been shown for the pro-opiomelanocortin (POMC) gene, an ACTH precursor and CRH both allowing a negative feedback circle. B: GR and frans-activating transcription factors can compete for binding sites in promoters. In the osteocalcin promoter GR overlaps the TATA box. Activation and nuclear translocation of GR can therefore prevent the binding of the basal transcription factor, TATA binding protein (TBP), which is necessary for the recruitment of RNA polymerase II and initiation of transcription. C: For the p65 subunit of NF-KB and GR a direct physical interaction has been demonstrated, leading to a frans-repression of transcription of the N F - K B dependent gene without DNA binding of GR. This interaction effect does not impair the DNA binding ability of NF-KB.

4C). The promoters of most proinflammatory genes contain binding sites for the proinflammatory transcription factor N F - K B or AP-1. In the genes regulated by both factors nGRE are not found. However, glucocorticoids are able to rapidly /ra«^-repress AP- 1 or NF-KB-induced transcription of these genes.

This might in fact be the most important mechanisms triggering their therapeutic efficacy in the treatment of inflammatory diseases.

It is not completely clear whether direct protein- protein interaction between GR and AP-1 or N F - K B are sufficient for /ra/i^-repression [85-87] or whether other mechanisms are involved. However, GR asso- ciation to AP-1 does not reduce AP-1-DNA-binding, as demonstrated by electrophoretic mobility shift assays (EMSA) [86, 88].

Similar observations were made for the interaction of N F - K B and GR. For the p65 subunit of N F - K B and GR a direct or semi-direct physical interaction has been demonstrated [14, 89-94]. This interaction

occurs without impairing the DNA-binding ability of both transcription factors [90-92, 95-97]. These data indicate that co-activating or co-repressing proteins may play a major role for that function.

Proinflammatory transcription factors such as AP-1 and N F - K B on one hand, and GR on the other hand, could compete for a coactivator molecule called CREB-binding protein (CBP) (for more details see ref. 93). CBP plays an essential role in the activation of transcription by numerous transcription factors/transcriptional activators [12 ,93, 9 8 -

103]. CBP binds and is necessary for coactivation of CREB, AP-1, signal transducer and activator of transcription (STAT) proteins and N F - K B , as well as nuclear receptors such as as GR, PPAR, progester- one receptor and retinoid receptors [12, 103]. It is likely that CBP and the related p300 are not acting as single coactivator proteins but are present in the nucleus in a complex of proteins [12, 103]. CPB and p300 acetylate free histones in vitro, which is neces-

(7)

sary for successful transcription [12, 103]. Other coactivators are pl60/SRC, the TRAP/DRIP/ARC complex and the P/CAF complex (for a review see ref. 12).

Factors determining fra^s-activation or fr^^s-repression The factors determining whether glucocorticoid binding to its receptor induces more rra^^-activation or more ^raw^-repression activity on genes are not completely understood. Clearly specific DNA-binding sites play a role. However, the question still arises as to why some ligands are able to induce mainly /ra^^-repression actions while others may mainly induce ^ra/t^'-activation.

In experiments modeling the GR ligand-binding domain it could be shown that tyrosine 735 may interact with the D ring of dexamethasone and that the substitution of D ring functional groups results in partial agonist steroids with reduced ability to direct

^ra«^-activation [104]. A substitution of Tyr735 by phenylalanine (Tyr735Phe) did not reduce ligand- binding affinity and did not alter ^ra/i^'-repression of

N F - K B , but reduced ^raw^'-activation. These data suggest that tyrosine 735 is important for ligand interpretation and ^ra«^-activation [104]. In addition, recent data suggest that GR does not necessa- rily have to form homodimers to modulate gene expression. Negative recognition elements (nGRE) have been identified in keratin gene promoters that bind GR as four monomers [105]. Thereby a specific set of co-repressors is bound, including histone acetyltransferase and CBP but not SRC- 1 and GRIP-1.

In the regulation of GR action so-called glucocorticoid-modulating elements (GME) in promoters and proteins binding to them (GMEB) have gained attention in recent years. A GME in the gene tyrosine aminotransferase (TAT) was identified in 1992 [106].

The GME was identified as a 21-basepair sequence of the rat TAT gene; it is located at -3.6 kb and 1 kb upstream of the GRE [107]. It modulates both the dose-response curve of agonists bound to the GR and the residual agonist activity of GR-bound anti- steroids. The expression of GME activity involves the binding of two novel proteins (GMEB-1 and GMEB-2) [107-112].

Taken together it seems that four important factors determine the dose-response curve and the preference of ^ran^-activation or rm«5'-repression of glucocorticoids: the ligand structure, the GR con- centration, the co-activator or co-repressor concen- tration/availability and the GME. Important effects

from the elucidation of these basic mechanisms for the steroid treatment of patients with IBD can be expected.

Regulation of glucocorticoid receptor expression To further complicate the system of ^ra«5'-activation,

^mn^'-repression with DNA-binding and trans- repression without DNA-binding involving co-activators and co-repressor the expression of the GR gene itself is regulated by activated GR. The GR (GRoc) represses its own synthesis in a hormone- dependent manner [113]. The reduction in cellular receptor levels is followed by insensitivity of the cells to glucocorticoids. It is dose- and time-dependent and reversible upon hormone withdrawal. For this down-regulation of GR expression by GR itself, DNA-binding seems to be crucial [114-118].

However, this is not the end of the story. The phosphorylation status of GR may be important for this feedback function. Deletion of phosphorylation sites from the GR protein by site-directed mutagen- esis resulted in a GR that cannot repress its own transcription upon binding of glucocorticoids [119, 120].

glucocorticoid receptor p

In contrast to G R a the shorter GRP isoform that is generated by alternative splicing lacks a LBD. GRp is identical to G R a through the first 727 AA but differs in the carboxyl-terminus, where it lacks the last 50 AA found in G R a but contains additional 15 non-homologous AA [14]. As GRa, GRp is widely expressed in adult human tissues but is primarily localized to the cell nucleus independent of the presence of ligand [14]. GRp still binds to DNA and may therefore potentially interfere with the action of G R a [121-123]. It has been speculated that GRp might be an antagonist of G R a action as it blocks DNA-binding sites without suppressing gene transcription [30, 121-127]. However, this inhibition seems to require increased GRp expression relative to GRa. It has been studied mainly in transfection assays with overexpression the P-isoform. A clear functional role has not yet been confirmed [13].

Glucocorticoid resistance in asthma patients has been described to be associated with elevated in-vivo expression of the GR P-isoform [128-130]. However, conflicting data have also been obtained showing no overexpression of GRp in resistant patients [131- 133].

(8)

Regulation of glucocorticoid receptor in IBD The regulation of GR action and the mechanisms involved in its function are complex. Investigations on possible mechanisms of steroid-refractory IBD are still at early stages. The few studies raise more questions than they answer. Hsp90 has some features of an inhibitory protein of GR and is bound to its inactive form (see above). The expression of human hsp90 in patients with CD and UC was studied;

however, no differences between patients and controls were found, making a role of hsp90 in glucocorticoid-refractory IBD unlikely [134].

The number and dissociation constant (A^j) of GR in peripheral blood mononuclear cells of six non- responders of glucocorticoid treatment with UC, five responders and 10 healthy controls was determined in another study [135]. A significant increase in the number of binding sites and the dissociation constant in non-responders compared to responders was found. Surprisingly the number of binding sites was highest in non-responders.

When GR levels were determined via dexamethasone binding only in the cytosolic fraction to ensure that only free receptor and not already steroid- associated, translocated receptor molecules could bind [ H]dexamethasone, the situation seemed to be different. The dexamethasone binding in cytosol isolated from peripheral blood mononuclear cells (PBMNC) of corticosteroid-treated IBD patients was significantly lower compared to controls and IBD patients not treated with glucocorticoid [136].

Systemic GR levels in untreated IBD patients did not differ significantly from controls [136]. There was no difference in the binding affinity of patients and control, with an obvious lower binding maximum indicating a reduced receptor number in the steroid- treated patients group, which is in contrast to the results reported by Shimada et al. In contrast to the findings in PBMNC mucosal GR levels of IBD patients were significantly decreased in both steroid- treated and untreated patients compared to controls [136].

The reduced binding of [^H]dexamethasone in cytosol from steroid-treated IBD patients is most likely due to a feedback regulation of GR in the cells by the ligand. This assumption is supported by a number of studies [137-139]. The data indicate that, in contrast to patients with rheumatoid arthritis or other connective tissue diseases in IBD, there is no difference in systemic GR levels between patients not treated with glucocorticoids and controls. This could mean that a localized inflammation in the intestinal

mucosa is not followed by a systemic depression of GR levels in leukocytes.

Honda and co-workers studied the expression of G R a and GRp in PBMNC of patients with UC and controls [140]. They found expression of GR(3 in only 9.1% of patients with steroid-sensitive disease whereas it was present in 83.3% of steroid-resistant patients as detected by polymerase chain reaction (PCR). The authors conclude that the determination of GRp expression could provide a tool to predict steroid responsiveness of UC patients [140]. How- ever, other laboratories have reported that GRP transcripts could be amplified from all control patients investigated, making a role of GRp for glucocorticoid refractivity unlikely (personal com- munication).

Despite these studies most of the questions regard- ing the mechanisms of steroid-refractory disease are still unanswered [141]. It is not clear why some patients express GRP and others do not. It is not clear whether changes in glucocorticoid binding are just an epiphenomenon or a cause of different disease courses. It is not clear whether GRp expression or decreased GRa levels are really crucial for treatment success.

Future studies on GR expression in IBD need to answer several important questions: are GR levels at the onset of the disease predictive for the success of glucocorticoid therapy? Are low levels correlated with the development of a steroid-refractory disease?

The question of whether measurement of GR in the mucosa can be predictive for therapy success needs to be answered for the future management of patients with IBD. Patients with low GR levels could then be primarily treated with other drugs such as azathiopr- ine. Prospective studies investigating GRoc and GRP transcripts, as well as glucocorticoid binding in PMNC and in the mucosa at the onset of disease before any treatment, will need to be performed to clarify whether both mechanisms are related to steroid-refractory IBD.

Cellular mechanisms of glucocorticoid action in IBD

Glucocorticoid receptors that can be activated by ligand binding mediate glucocorticoid action and induce ?ra«5'-activation or /ra/t^-repression of genes as well as an inhibition of proinflammatory transcription factors such as N F - K B and AP-1. The question arises as to which effects the ligand-induced

(9)

Glycocortlcoids/Glucocorticold-Receptor

translation - — ! • new protein.

vrelevant for apoptosis

r M

antagonism to NF-KB

mediated transactivation

of pro-inflammatory genes^ translation

Induction of apoptosis Inhibition of N F - K B

mediated gene induction

Figure 5. Glucocorticold-induced anti-inflammatory mechanisms. Two major principles mediate the anti-inflammatory effect of glucocorticoids. After penetrating the cell membrane and binding to the GR nuclear translocation occurs. The complex can then fraA7s-activate genes that are involved in the induction of apoptosis. The induction of apoptosis in activated lymphocytes, eosinophils, basophils and other cell types reduces the overall amount of circulating cytokines and inflammatory mediators and allows reconstitution of the intestinal mucosa. On the other hand the heterodimer of GR interacts with translocated NF-KB and prevents transcriptional activation by this proinflammatory transcription factor As NF-KB activation has anti-apoptotic effects in a number of cell types antagonism to its action also has pro-apoptotic effects, indicating the interconnection of both mechanisms of glucocorticoid action.

activation of GR have at the cellular level, or more precisely which eflfects can be observed in different cell types? A simplified view could focus on two major principles (Fig. 5):

1. The cellular effects induced by a down-regulation of proinflammatory transcription factors. As mentioned this is facilitated by an incompletely understood antagonism to transcription factors such as N F - K B and AP-1 (Fig. 5).

2. The induction of apoptosis of activated immune cells, which limits the immune response and as a consequence is also followed by reduced levels of circulating proinflammatory factors (Fig. 5).

In this chapter only a few of the effects on some of the relevant cell populations in the intestinal immune system can be highlighted.

Lymphocytes

Glucocorticoids affect the growth, differentiation and function of lymphocytes, the distribution of cellular subsets, and the production of cytokines [142]. In the chronic inflamed mucosa the number of lymphocytes is greatly increased. Most of these cells are activated T-helper cells.

Glucocorticoids reduce lymphocyte proHferation and induce apoptosis in these cells [143-146]. Studies indicate that mainly transcriptional ^ra«^-activation functions are required for this glucocorticoid- mediated apoptosis [147]. Interestingly intestinal intraepithelial lymphocytes (lEL) may be resistant

(10)

to steroid-induced apoptosis, which could be due to the expression of high levels of the anti-apoptotic protein Bcl-2 and Bcl-x [148].

Recent data indicate that glucocorticoid-mediated inhibition of cytokine secretion in inflammatory diseases may be mediated not only by direct action on lymphocyte and monocyte/macrophage N F - K B but in addition indirectly through promotion of aT- helper cell type 2 (Th2) induction [143] with increased levels of Th2 cytokine (IL-4) and reduced levels of Thl cytokine (IL-12) secretion [149]. In addition dexamethasone can inhibit IL-12-induced phosphorylation of STAT-4 without altering IL-4- induced STAT-6 phosphorylation [149]. Both would result in a blockade of proinflammatory T-helper cell type 1 (Thl) cytokine expression.

A further example of the multi-site mechanisms involved in glucocorticoid action is given by another monocyte/macrophage-lymphocyte interaction.

Glucocorticoids have been shown to down-regulate T-cell co-stimulatory molecules B7-1 and B7-2 on macrophages which are essential for clonal T-cell expansion in reaction to antigen-presenting cells [150]. On the other hand co-stimulatory molecules can prevent cells from glucocorticoid-induced apoptosis [151 ]. It is interesting to note in this context that normal intestinal macrophages express no co-stimulatory molecules, whereas there is a clear up-regula- tioninIBD[152, 153].

Macrophages

Macrophages are known to play an important role during inflammation in many diff'erent tissues [154].

Intestinal macrophages represent one of the largest compartments of the mononuclear phagocyte system in the body [155]. They are localized preferentially in the subepithelial region and constitute 10-20% of mononuclear cells in the intestinal lamina propria [156]. Macrophages are able to secrete proinflammatory cytokines which are known to be regulated by

N F - K B such as IL-ip,TNF-oc, IL-6, IL-8, MCP-1.

In normal mucosa only very few macrophages express activation-associated markers such as CD14, CD16, HLA-DR, GDI lb, and G D l l c [152, 153, 157], supporting a concept of anergy in the normal mucosa. Several findings indicate a pheno- typic change of the intestinal macrophage popula- tion in IBD. Mahida etal. demonstrated the presence of GDI6 (Fcylll receptor) in IBD by immuno- histochemical methods and in isolated cells [158].

GD54 (IGAMl) expression increased from 7% to 70% in UG and to 46% in GD [159]. Inflammation-

associated intestinal macrophages express LPS receptors, Fc receptors and co-stimulatory molecules [152, 153, 157], which enable them to stimulate T cells and partially prevent them from glucocorticoid-induced apoptosis (see above).

The number of macrophages is clearly increased in both GD and UG mucosa [160-162]. The activated macrophage population secretes a multitude of inflammatory mediators such as prostaglandins;

leukotrienes; cytokines such as TNF, IL-1, IL-12;

chemokines such as MGP-1 or IL-8 as well as tissue- damaging reactive radicals and tissue-degrading enzymes [163-176]. The transcription, translation and secretion of almost all molecules mentioned above can be reduced or inhibited by the administration of glucocorticoids. In the promoters of most of those gene N F - K B binding sites can be found [177- 187], indicating that the glucocorticoid-mediated inhibition of N F - K B activation may be the most important mechanism for their anti-inflammatory potential.

Intestinal epithelial cells

Intestinal epithelial cells (lEG) play an active role in the intestinal immune system [188-190]. After stimulation with IFN-y they express MHG molecules [191-195]. In addition there is evidence that epithelial cells are able to respond to damage or bacterial invasion by secreting cytokines and chemokines [196-205]. Among the cytokines secreted by intestinal epithelial cells are inflammatory mediators such as IL-6 and IL-8.

As already discussed for the other cell types, it could be shown that, for example, the induction of IL-6 production in human intestinal epithelial cells following stimulation with IL-1P was associated with activation of the transcription factor N F - K B [206].

Jobin and co-workers demonstrated that adeno- virus-mediated transfection of HT-29 and Gaco-2 cells with a N F - K B superrepressor caused a reduced induction of inducible nitric oxide synthase (iNOS), IL-1 (3 and IL-8 by IL-1 or TNF, indicating involve- ment of the N F - K B system in regulation of these genes [207]. In the inflamed mucosa activation of

N F - K B has been demonstrated in intestinal epithelial cells, whereas activation was absent in non-inflamed mucosa [208, 209].

Glucocorticoids down-regulate the expression of IL-6, IL-8, iNOS and class II molecules in intestinal epithelial cells and restore epithelial cell physiology [210, 211]. This is again likely to be mediated by the inhibition of N F - K B activation.

(11)

Other cell types

As GR are expressed in most if not all cell types, effects of glucocorticoid therapy on all cell types involved in intestinal inflammation have been found.

During the course of IBD increased numbers of eosinophils and mast cells are present in the intestinal mucosa [212-216]. These cells are activated and secrete different tissue-damaging proteins as well as cytokines and chemokines [217-224]. Again N F - K B has been found to be a major factor in activation of eosinophils [225], and glucocorticoids have been shown to down-regulate the production and secretion of the proinflammatory proteins [226, 227]. In addition the number of mucosal and circulating eosinophils is decreased by glucocorticoids. This is probably mediated by an induction of apoptosis in these cells [228-233].

As well as the number of eosinophils and mast cells, the number of basophils is increased in IBD mucosa [212]. Glucocorticoids reduce the number of basophils similar to the reduction described for eosinophils.

Neutrophils are a major component in active lesions in UC and to a lesser extent in CD [234-240].

They are recruited under the influence of the neu- trophil chemoattractant IL-8 [237]. An important feature of these neutrophils is their ability to induce the so-called 'oxidative burst' reaction involving the NADPH oxidase system leading to a secretion of oxygen radicals that not only kill surrounding bacteria but also damage surrounding tissue [238].

Glucocorticoids act in different ways in neutrophils.

They down-regulate the oxidative burst reaction and reduce IL-8 secretion, leading to a reduced immigra- tion of neutrophils into the mucosa. On the other hand an increased release from the bone marrow, followed by a leukocytosis, is induced [241].

Another important feature of the treatment with glucocorticoids is down-regulation of the expression of adhesion molecules on endothelial cells, mononuclear cells and epithelial cells [89, 241-247], which again may be mainly mediated by antagonistic effects to N F - K B .

Glucocorticoid ttierapy in combination witti other anti-inflammatory drugs in IBD

Due to this multitude of desired effects the standard therapy for acute flares of IBD is still the systemic application of glucocorticoids. The effectiveness of a glucocorticoid regimen has been shown in numerous

multicenter trials [23, 248]. Initial remission rates in patients with acute flares of CD under a standard therapy vary from 60% to 80%, which is higher than under treatment with sulfasalazine or 5-amino- salicylic acid (40-50%). The combination of glucocorticoids and anti-inflammatory drugs, e.g. sulfasalazine, shows no additive effect and no higher remission rates than therapy with prednisolone alone [23]. This may indicate that glucocorticoids and drugs such as sulfasalazine may both act in a common signal transduction cascade. Recently evidence has been found that some of the anti- inflammatory effects of 5-ASA and sulfasalazine in IBD patients may be mediated by the inhibition of the proinflammatory transcription factor N F - K B . Acetylsalicylic acid, 5-ASA and sodium salicylate inhibit activation of N F - K B by blocking iKB-kinases (IKK), which are key factors in N F - K B activation [249-251]. Similar results were found for sulfasalazine [252,253].The inhibition of bothlKKocandlKKp has been shown [254].

The lack of additional effects of a combination of glucocorticoid and salicylate therapy indicates that the glucocorticoid effect on N F - K B inhibition may be superior. However, long-term studies show that, despite the high initial response rates, only 44% of patients initially treated with glucocorticoids show long-term remission; 25-35% of patients become 'steroid dependent', indicating that steroid treatment cannot be completely tapered and omitted. About 20% of glucocorticoid-treated patients prove to be primarily 'glucocorticoid resistant' [255, 256].

Peroxisome proliferator-activated receptors (PPAR)

The GR that mediates glucocorticoid effects in the treatment of IBD is a member of the so-called steroid or nuclear receptor superfamily. Other members of this family have gained increasing attention in the recent years. Particularly PPARy and PPARa have been shown to have NF-KB-inhibiting activities, and could be possible tools for the treatment of IBD in the future. They are further examples for a multisite treatment approach.

PPARs - like GR - are ligand-activated receptors.

They have been discovered to be regulators of lipid and lipoprotein metabolism [257]. However, in recent years it was shown that they also regulate cellular proliferation, differentiation and apoptosis [257], features that may be very important during

(12)

repair processes after or during intestinal inflammation. Three family members are known. Besides PPARoc and PPARy, PPAR5 is encoded by a sepa- rate gene and expressed in most tissues. PPARs form heterodimers with the retinoid X receptor (RXR) and bind to PPAR response elements (PPRE) in the promoters of target genes.

PPARy

As mentioned, PPARy is a member of the nuclear hormone receptor family and a ligand-activated transcription factor. High expression is found in adipose tissue, the adrenal gland, the spleen and, interestingly, in the colon [257-262]. PPARy is involved in the induction of adipocyte differentiation and glucose-homeostasis [262]. PPARy is also expressed in diff'erentiated macrophages, whereas there is no expression in monocytes [263].

Thiazolidinediones were identified as synthetic l i g a n d s of PPARy [262, 2 6 4 - 2 7 3 ] . Besides pharmacologically developed ligands natural ligands such as free fatty acids and the prostaglandin D2 metabolite 15-deoxy-A12,14 prostaglandin J2 bind to the PPARy protein and stimulate the transcription of target genes [274-279]. Prostaglandin D2 metabo- lites cannot be detected in adipose tissue; however, they are important intermediate products of ara- chidonic acid metabolism in macrophages and antigen-presenting cells [280].

In activated macrophages a significant-upregula- tion of PPARy was found [281-283]. An activation of PPARy by its ligands 15A-PGJ2 or thiazolidine- diones inhibited iNOS, gelatinase B and scavenger receptor A genes. This inhibition is partially mediated by an antagonism to the transcription factors AP-1, STAT-1 and N F - K B [283]. Furthermore it could be demonstrated that PPARy inhibits the expression of a number of proinflammatory cytokines in monocytes [283, 284].

Besides macrophages and adipocytes, colonic epithelial cells express high levels of PPARy mRNA and protein [258, 285-290]. The physiological function of PPARy in intestinal epithelial cells is not well understood. A recent study by Su and co-workers showed that the expression of proinflammatory cytokines was reduced by incubation with PPARy ligands, which was mediated by inhibition of N F - K B activation [291]. The administration of thiazolidinediones during the recovery phase in an acute model of DSS (dextran sulphate sodium)-induced colitis in mice was followed by a dramatic improvement in

histological signs of inflammation [291]. Clinical trials to test the efficacy of thiazolidinediones in IBD are under way.

When activated, PPARy molecules form heterodimers with another transcription factor of the nuclear receptor superfamily, the retinoid-X-receptor (RXR). Heterozygous PPARy and RXRot knock- out mice display a significantly enhanced susceptibility to 2,4,6-trinitrobenzene sulfonic acid (TNBS)- induced colitis compared with their wild-type litter- mates, indicating a role for the RXR/PPARy heterodimer in protection against colon inffammation [292]. The administration of both PPARy and RXR agonists also reduced TNBS-induced colitis, reflected by a decrease in T N F and IL-ip mRNA levels and a reduction of N F - K B DNA-binding activity [292]. A synergistic eff'ect of PPARy and RXR ligands was observed.

In addition to their N F - K B antagonistic properties ligands for PPARy, similar to GR have been proven to induce apoptosis in a number of different cell types and cell lines [257, 293-303]. However, it is not clear whether the induction of apoptosis is simply mediated by the inhibition of the antiapoptotic N F - KB or by a yet-incompletely elucidated specific mechanism. In addition it is necessary to clarify whether the induction of apoptosis by PPARy ligands plays a role in the regulation of the human immune system. It is important to note that PPARy activators have also been discussed as inductors of colon polyp formation [289], which could be a major problem in the therapy of IBD patients, increasing their risk of developing colon cancer. Future studies are necessary to clarify this problem.

PPARa

PPARa is another member of the PPAR family.

Recently it has been shown that dehydroepiandro- sterone (DHEA) could be a natural ligand for PPARa.

DHEA and its sulphated metabolite (DHEAS) are the most abundant steroid hormones in the body.

They are predominantly synthesized in the adrenal glands in response to adrenocorticotrophin (ACTH) from its precursors cholesterol and pregnenolone, and can be further metabolized via androstenedione into androsterone and via testosterone into 17P- estradiol [304]. DHEA inhibits activation of N F - K B

and secretion of IL-6 and IL-12 via the activation of PPARa [305-307]. Furthermore DHEA stimulates the production of IL-10 in murine spleen cells [308].

(13)

Treatment with 50-200 mg/day DHEA was shown to be eflfective in patients with lupus erythematosus [309-312]. On the other hand DHEAS concentrations are decreased in patients with IBD [313, 314].

In a pilot trial treatment with DHEA was safe and effective in patients with refractory active CD or UC [315]. In this study a dose of 200 mg/day DHEA in all patients was used.

Ligands for PPARoc are at present being developed by several major pharmaceutical companies. Future studies are necessary to clarify whether these ligands are favorable compared to the endogenous ligand DHEA, and whether PPARa ligation is followed by antiapoptotic or proapoptotic stimuli [316].

The data for both members of the PPAR family are so far incomplete and controlled randomized clinical trials are needed. However, they show clearly that possibilities for multi-site anti-inflammatory therapies are not limited to GR. PPAR ligands could prove to have great potential for the treatment of IBD.

Inhibition of NF-KB-mediated

fra/75-activation as a centrai target of IBD tiierapy

The discussion of GR functions showed that one of the most important anti-inflammatory mechanisms mediated by glucocorticoid therapy is the inhibition of the proinflammatory transcription factor N F - K B . Furthermore, the first genetic mutations associated with susceptibility to CD were found in NOD2 [317, 318], which is a NF-KB-activating protein [319-321].

Therefore the N F - K B system needs more detailed consideration.

Molecular mechanisms of N F - K B activation Transcription factors of the N F - K B / R C I family form dimeric complexes which control the expression of a variety of inducible genes involved in inflammation and proUferation [322, 323]. The prototypic hetero- dimeric complex N F - K B consists of the subunits p50 and p65 (RelA) [324, 325]. The inactive N F - K B dimer is present in the cytosol bound to inhibitory proteins, termed IKB [326, 327] (Fig. 6). iKBa, I K B P and IKBS have similar functions in inhibiting the translocation of the N F - K B dimer [328]. The activation of N F - K B by inflammatory cytokines and microorganisms requires the release of IKB from the complex [324- 327, 329]. The release of I K B is induced by

phosphorylation of IKB at two conserved amino- terminal serine residues by a multiprotein I K B - kinase complex containing iKB-kinase alpha (IKKoc), beta (IKKP) and gamma (IKKy) (Fig. 6) [330-332].

This is followed by a poly-ubiquination of the IKB proteins by a specialized E3 ubiquitin ligase complex

( E 3 I K B ) [333] which makes them accessible to proteolytic degradation by the 26S proteasome (Fig.

6) [328]. The removal of IKB proteins exposes nuclear localization signals (NLS) followed by translocation of the activated N F - K B into the nucleus [330]. There, the activated N F - K B dimer interacts with regulatory N F - K B elements in promoters and enhancers [324, 325, 329, 330] (Fig. 6). Among the genes trans- activated by the N F - K B dimer, interestingly, is IKBOC [328]. This usually limits N F - K B action. If IKB does not find a binding partner in the cytosol it trans- locates to the nucleus, binds activated N F - K B and mediates re-shutthng of the complex to the cytosol.

N F - K B activation in IBD

There is evidence that N F - K B transcription factors might play an important role in the inflammatory process of IBD. It was shown that the administration of antisense phosphothioate oligonucleotides to the p65 subunit of N F - K B abrogates colitis in IL-10- deficient mice and in the TNBS-colitis model [334].

Inhibition of the proteasome complex which is responsible for the rapid degradation of I K B prevented N F - K B activation and attenuated the colonic and splenic injury and inflammation in the PG/PS model [335]. In the DSS model of chronic colitis the DSS-induced intestinal inflammation was characterized by an increase of N F - K B activity [336].

Blocking N F - K B activation by administering gliotoxin, a fungal product, was accompanied by a significant suppression of intestinal inflammation and mRNA expression of TNF-oc and IL-la in vivo [336].

N F - K B activation was demonstrated in cultured rat intestinal epithelial cefls (IEC-6) by IL-ip and TNF-a [207, 337]. Dexamethasone prevented epithelial cells from being activated by these mediators [207, 337]. In mucosal biopsies from patients suffer- ing from IBD, activation of N F - K B was demonstrated to be mainly localized in two cell types in the intestinal mucosa by double-labeling techniques: (1) in lamina propria macrophages and (2) in epithelial cells [208].

(14)

cell membrane

TNF-RI

activator (e.g. TNF)

nuclear membrane pro raptor DNA

p65

' { R « ' A )

i

transcription

(RelA)

inslocation c ^z^-"^'^

NF-KB translocation licB-degradation

after ubiquinatlon the 26S-proteasome

translation

mm new protein, Mini altered cell function

Figure 6. Principles of NF-KB activation: In the cytoplasm the NF-KB complex is associated with the Inhibitory protein IKB.

Activation of NF-KB IS Induced by a number of different signals, as for example IL-1 and TNF-a. These signals activate an IKB- klnase complex that contains the kB-kinases IKK-a (IKK1) and IKK-jJ (IKK2). IKK-a and IKK-P are phosphorylated, leading to a recruitment of IKB and phosphorylation of IKB at serine32 and serine36. This Is followed by release of IKB from the complex.

Consecutively a rapid proteolytic degradation of IKB and a translocation of the activated NF-KB Into the nucleus takes place. In the nucleus the activated NF-KB dimer Interacts with regulatory NF-KB elements In promoters and enhancers, leading to alteration in transcription rates and altered cell function. When the degradation of IKB is blocked by Inhibitors of the proteasome complex IKB reassociates with the NF-KB dimer and NF-KB activation is inhibited.

Furthermore, mutations in N0D2, a gene which is mainly expressed in monocytes/macrophages and activates N F - K B , has been shown to be associated with susceptibiHty to CD [317, 318].

A variety of genes are induced in the inflamed mucosa which have been shown to be regulated by N F - K B , including the genes encoding TNF-<x [171, 329, 338-341], IL-lp [171, 184, 339, 341-345], IL-6 [346, 347], IL-8 [348, 349], macrophage colony-stimu- lating factor (M-CSF) [329], macrophage granulocyte colony-stimulating factor (GM-CSF) [329], monocyte chemotactic protein-1 (MCP-1) [325, 350], vascular cell adhesion molecule-1 (VCAMl) [351, 352] and intercellular adhesion molecule-1 (ICAMl) [325].

Some of these gene products such as TNF-oc and IL- 1, are also able to activate N F - K B [324, 325], leading to a positive autoregulatory loop [184].

NF-KB and co-factors

As mentioned above, one of the most important eff'ects of glucocorticoids (and ligands for other nuclear receptors) in the treatment of inflammatory diseases is the suppression of the /raw^'-activating ability of N F - K B . Nuclear receptors (GR and PPAR), as well as N F - K B depend on the co-activators CREP-binding protein (CBP) and steroid receptor coactivator-1 (SRC-1) for their transcriptional activity [99, 100]. It has been suggested that some of the inhibitory affect of GR action is mediated by a nuclear competition for limited amounts of the coactivators CBP and SRC-1 [100]. This means that the presence of activated GR would induce binding of CBP and/or SRC-1 and that as a consequence activated N F - K B (p65) does not find a sufficient