Animal models have demonstrated the critical need for an intact HPA axis and glucocorticoid response for survival from a number of insults including bacterial and viral infections and toxins. Removal of endogenous glucocorticoids by adrenalectomy, the glucocorticoid receptor (GR) antagonist RU486, or interruption of the HPA axis by hypophysectomy, significantly enhances mortality from endotoxin or LPS, Shiga toxin, and normally non-lethal doses of the bacterial superantigen Staphylococcus aureus enterotoxin B (SEB) [11,12,13,14,15,16,17,18,19]. Removal of endogenous glucocorticoid responses by RU486 or adrenalectomy also resulted in enhanced Clostridium difficile toxin A-induced fluid secretion and inflammation [20,21]. These effects of loss of HPA axis or GR function could be reversed by exogenous replacement of glucocorticoids. A physiological dose of corticosterone resulted in an inflammatory response following Clostridium difficile toxin A that was equivalent to sham-operated animals, whilst replacement with a high pharmacological corticosterone dose resulted in a reduction of the inflammatory response [20]. Survival rates of BALB/c mice from Shiga toxin 2 were enhanced by 18 hour pre-treatment of either LPS or dexamethasone whereas only one hour of LPS pre-treatment decreased survival rates. This enhanced mortality with one hour pre-treatment of LPS correlated with increased pro-inflammatory mediators, such as TNFα. In fact, pre-treatment with TNFα also decreased survival to Shiga toxin 2. The protection afforded by the 18 hour LPS pre-treatment condition was shown to be due to the increased endogenous corticosterone production secondary to LPS-induced IL-1β activation of the HPA axis [22]. Furthermore, dexamethasone treatment reversed the increased Shiga toxin-induced mortality in adrenalectomized animals [17]. Likewise, administration of exogenous dexamethasone protected adrenalectomized BALB/c mice from bacterial superantigen SEB lethality [18]. Administration of dexamethasone to F344/N rats treated with RU486 similarly prevented mortality from streptococcal bacterial cell walls [19]. Dexamethasone, but not the natural glucocorticoids, corticosterone and deoxycorticosterone, reversed LPS-induced mortality in adrenalectomized animals, suggesting that synthetic glucocorticoids are more effective than endogenous glucocorticoids in protecting against endotoxin/LPS lethality [13,16]. Increased cytokine production, particularly TNFα is the most likely cause of enhanced LPS/endotoxin-induced mortality following removal of endogenous glucocorticoids or HPA axis blockade [13,23]. Increases in cytokine levels (TNFα and IL-6) following LPS/endotoxin administration are enhanced further by HPA axis blockade (adrenalectomy or RU486) and can be reversed by glucocorticoid treatment [12,24]. Finally, the requirement for an intact glucocorticoid response for survival from endotoxin is further demonstrated by the fact that GR over-expression in mice renders them resistant to LPS-induced endotoxic shock [25].

In agreement with the above studies which support the role for an intact HPA axis and glucocorticoid response in survival from a toxic insult, we have shown that adrenalectomy increases lethality to anthrax lethal toxin (LeTx) in BALB/cJ, C57BL/6J and the normally LeTx resistant DBA/2J mice [26]. Likewise, RU486 exacerbated lethality in Balb/cJ mice. However, this could not be reversed by dexamethasone or aldosterone administration [26] suggesting that a careful balance of the HPA axis and glucocorticoid response is required for survival from LeTx.

Glucocorticoids exert their many effects through a cytosolic receptor, GR, a member of the nuclear hormone receptor superfamily, which also includes the thyroid hormone, mineralocorticoid (MR), estrogen (ER) and progesterone receptor (PR) [27]. In the absence of ligand, GR is located in the cytoplasm in a protein complex that includes Hsp90 and Hsp70. Upon ligand activation, GR is released from the protein complex, dimerizes, and translocates to the nucleus where it binds to specific DNA sequences called glucocorticoid response elements (GRE) (Figure 2). Thus, GR functions as a ligand-dependent transcription factor [28]. GR is able to upregulate gene expression through direct DNA binding, for example the gluconeogenic enzyme tyrosine aminotransferase (TAT) whose promoter contains a consensus GRE sequence [29]. GR can also bind to negative GREs (nGRE) to repress gene activation, such as for the proopiomelanocortin (POMC) gene [30]. However, GR primarily represses gene transcription by interfering with the action of other signaling pathways, such as nuclear factor kappa B (NFκB) and activator protein 1 (AP-1) (Figure 2), and it is through this mechanism that glucocorticoids exert many of their anti-inflammatory actions [31,32]. GR is essential for life. Mice lacking GR die shortly after birth due to a defect in lung maturation [33]. However, it appears that the anti-inflammatory actions of GR associated with its ability to interfere with other signaling mechanisms may be the most critical for survival. Dimerization knockout mice (GR^dim/dim) are viable [34]. In these mice GRE-mediated gene activation, which is entirely dependent on GR dimerization, is removed but GR interactions with NFκB and AP-1, which are independent of dimerization, are still possible.

GR mutations exist and play a role in glucocorticoid resistance [35]. However, there are multiple steps in the GR signaling pathway that if defective could also cause glucocorticoid resistance/insensitivity. Included in these possible mechanisms are the disruption of GR signaling, reduced GR numbers [36], abnormal expression of Hsp90 [37,38,39,40], enhanced expression of the dominant negative splice variant of GR, GRβ [41,42,43,44,45], dysregulation of 11β-hydroxysteroid dehydrogenase (11β-HSD) [46], defective GR nuclear translocation [47,48,49], cofactor defects [50,51], increased multidrug resistance (MDR) protein expression [52,53,54,55,56,57,58], reduced histone deacetylase (HDAC) activity [59,60], and p38 phosphorylation of GR [47]. However the role of viral and bacterial infections and toxins in glucocorticoid resistance has been largely neglected. We have recently shown that the anthrax LeTx represses GR function. In addition, it has long been known that bacterial endotoxin or LPS also affects GR function.

Bacillus anthracis produces three proteins – protective antigen (PA), lethal factor (LF) and edema factor (EF), which constitute two toxins. LF and PA combined constitute LeTx and EF and PA the edema toxin. We have shown that LeTx is able to repress the GR and other nuclear hormone receptors. LeTx represses glucocorticoid induction of a GR-responsive promoter in Cos7 cells and glucocorticoid induction of the GR-regulated liver enzyme, TAT in a hepatoma cell line and in an animal model [26,66,67] but does not affect GR-mediated gene repression [67]. LeTx also represses ERα, PR, MR and androgen receptor (AR) in a promoter-specific context [66,67]. LeTx is a metalloprotease that is known to cleave and inactivate mitogen-activated protein kinases (MAPKs) [88,89,90,91,92]. A protease deficient mutant of LeTx did not repress GR-mediated gene activation suggesting that the protease activity was required for the repressive effects on GR [66]. However, LeTx did not alter GR protein levels suggesting that GR itself is not a direct target for LeTx-mediated proteolysis [67]. LeTx acts as a non-competitive inhibitor of GR and has no affect on GR-ligand binding [66]. It does not affect nuclear translocation but does prevent GR-DNA binding [67]. Recently we have shown that LeTx also represses induction of the GR-responsive MMTV promoter by other transcription factors including HNF3, Oct1 and AP-1. This repression was not observed with the protease deficient LeTx mutant and could be prevented by inhibitors of LeTx protease activity. Unlike the effects on GR, LeTx induced proteolysis of these transcription factors but at a much later stage than the well-documented LeTx-mediated proteolysis of MAPKs [93]. These data suggest that LeTx represses multiple transcription factors including GR through different mechanisms.

LPS or endotoxin is the principal component of the outer membrane of Gram-negative bacteria. LPS signals through the Toll-like receptor 4 (TLR4) to activate the MAPK pathways and the NFκB pathway leading to induction of many inflammatory genes [94]. An excessive inflammatory response to LPS can lead to sepsis, septic shock or systemic inflammatory response syndrome.

It has long been known that endotoxin or LPS alters GR-regulated liver enzymes. Since the early 1980s there have been reports of decreased glucocorticoid induction of liver enzymes, including, glucose-6-phosphatase, fructose-1,6-diphosphatase, phosphenolpyruvate carboxykinase (PEPCK), tryptophan oxygenase (TO) and TAT, by endotoxin [69,70,71,72]. Several studies have shown that endotoxin decreases steroid binding sites in liver cytosol [11,70,72,73]. These effects were not only observed in liver but also in other tissues including murine macrophages [74], kidney, skeletal muscle, spleen, lung, heart tissue [70], canine leukocytes [75] and sheep lungs [76]. Despite the number of studies that found an effect of LPS/endotoxin on GR ligand binding, other studies could find no effect of endotoxin on number or affinity of hepatic GR suggesting that down-regulation of receptors is not involved in endotoxin inhibition of glucocorticoid-induced hepatic genes but acts at a stage downstream of ligand binding [82]. In addition, GR numbers were shown to increase after LPS treatment in murine Raw 264.7 and peritoneal macrophages [81] and in a bronchial epithelial cell line [77]. In cell culture, LPS also inhibited glucocorticoid induction of the mouse mammary tumor virus (MMTV) promoter in the fibroblast LMCAT cell line [78,79,80]. Although there are some discrepancies, the majority of the data support the hypothesis that endotoxin/LPS represses GR function. Whether this is at the level of ligand binding or further downstream is debated.

It is not entirely clear if the effects of LPS on GR are mediated directly by LPS or through an intermediate factor. Early studies showed that endotoxin-induced downregulation of hepatic GR was mediated by plasma factors [73]. Berry and colleagues described a glucocorticoid-antagonizing factor (GAF) which was released by macrophages following endotoxin challenge [95]. GAF was shown to reduce liver glycogen levels [96] and inhibit PEPCK activity [95,97]. It was described as a 90 kDa glycoprotein [98] but its exact components have never been fully identified and there has been no mention of it in the literature since 1990. However, it should be noted that the glucocorticoid inhibitory properties of GAF are remarkably similar to cytokines such as TNFα and IL-1 and to macrophage migration inhibitory factor (MIF) that are also released from macrophages following endotoxin challenge [99,100,101].

Pro-inflammatory cytokines have also been shown to modulate GR function. IL-1 decreases steroid binding in liver cytosol [102] and in hepatoma cells and also inhibits glucocorticoid induction of PEPCK [103]. A combination of IL-2 and IL-4, or IL-1β and IL-13 reduces GR affinity [77,104,105]. IL-1α, IL-1β, IL-6, IFNγ and TNFα increased GR numbers [77,106,107,108]. IL-1α inhibited dexamethasone induced GR nuclear translocation and GR–mediated gene transcription [106]. IL-13 prevented glucocorticoid suppression of LPS-induced IL-6 [105]. In addition, the effects of LPS on GR could be mediated through induction of other signaling pathways such as NFκB, AP-1 and MAPK, all of which are known to crosstalk with GR. A mutual antagonism exists between the GR signaling pathway and the AP-1 pathway [109,110,111], as well as the NFκB [112] and MAPK pathways. LPS-induced inhibition of corticosterone induction of CAT activity in LMCAT cells could be reversed by p38 MAPK inhibitors [79] suggesting that the LPS effect was mediated though p38 MAPK. Thus, the effects of endotoxin/LPS may not be mediated directly by LPS, but may be a result of the cytokine production following LPS challenge or induction of other signaling pathways.

Shiga toxins are a family of related toxins with two major groups – Stx1 and Stx2. There are few studies investigating the effect of this toxin on GR. One study did note that Stx2 caused an increase in GR numbers in circulating neutrophils [17]. The reason for, and the consequence of, this increase is unclear.

Bacterial superantigens are a class of antigens which cause non-specific T-cell activation. The bacterial superantigens, SEB, toxic shock syndrome toxin 1 (TSST-1) and Staphylococcus aureaus enterotoxin E (SEE) induced glucocorticoid resistance (as determined by the anti-proliferative effects of glucocorticoids) in PBMCs [83,84,85]. SEB impairs GR nuclear translocation in PBMCs [84] and also induces expression of GRβ [83,86,87]. In another study TSST-1 was shown to reduce glucocorticoid induction of FKBP51 mRNA, a known GR regulated gene, through a mechanism involving Jun N-terminal kinase (JNK) [85]. These data suggest that bacterial superantigens interfere with GR signaling.

The lethal toxin from Clostridium sordellii (TcsL) and toxin A (TcdA) and toxin B (TcdB) from Clostridium difficile repress GR-mediated gene activation. TcsL also prevents dexamethasone inhibition of LPS-induced TNFα production in splenocytes. This effect is suggested to occur through inhibition of p38 MAPK as these toxins all prevent phosphorylation of p38 MAPK [68].

Since the early 1990s, heavy metals such as arsenic, cadmium, zinc, mercury, chromium, selenium, lead and beryllium have been reported to affect GR function. These are reviewed below.

Arsenic. Arsenic is a well known poisonous metalloid, which together with its compounds arsenide and arsenate are commonly found in pesticides, herbicides and alloys. Arsenic can be found in groundwater and has been associated with increased cancer rates in those areas [144]. Arsenic has a biphasic effect on GR function. Extremely low doses enhance glucocorticoid induction of the endogenous GR regulated genes TAT and PEPCK [117,118,119]. Whereas low doses decrease GR-mediated gene activation of a transiently transfected promoter and of endogenous TAT and PEPCK [116,117,118,119,120]. Arsenic does not affect GR-mediated gene repression [118]. This inhibitory effect on GR-induced transcription does not alter GR nuclear translocation [120], total GR protein levels [117], and does not require dimerization [117]. It does require the GR DNA binding domain and can be abolished by mutations in that region [118]. Arsenic reduces the “open” structure of the nucleosomes on the MMTV promoter in response to dexamethasone and causes changes in post translational modifications of histones [116]. Reduced binding of coactivator-associated arginine methyltransferase (CARM1), but not GRIP1, was seen in cells treated with arsenic and over-expression of CARM1 reversed the arsenic repression of GR-induced genes [116]. These data suggest that arsenic represses GR receptor function by interfering with CARM1, a coregulator involved in GR-mediated gene activation. In addition, arsenite, an arsenic oxoanion, inhibits GR ligand binding by interacting with the vincinal thiols in the ligand binding region of GR and thereby preventing ligand binding [124,125].

Cadmium. Cadmium is a highly toxic metal which until recently was routinely used either as a pigment or in the steel industry. Due to the associated health and environmental concerns its use is declining. Like arsenic, cadmium exhibits a biphasic effect on GR function. Low doses reduce GR ligand binding capacity and inhibit GR-induction of the GR responsive MMTV promoter and the endogenous GR regulated gene TAT in rat liver [121,128]. However, higher doses enhance glucocorticoid activation of TAT [121]. Cadmium (administered in vivo) reduces GR ligand and DNA binding in rat liver [121]. Interestingly, the same investigators also reported that cadmium reduced GR ligand binding in liver cytosol only in vitro and not in vivo and that the lack of an effect of the in vivo experiments was due to over-expression of Hsp90 [122]. The reason for this discrepancy is unknown. Other investigators have also shown that Cadmium (II) can inhibit steroid binding to GR [123,124]. The effects of this cadmium ion on GR appear to act through the redox state of the receptor as they can be reversed by the reducing agent dithiothreitol (DTT) [124,128]. Cadmium, like arsenite, binds to the vicinal dithiols in the ligand binding region of GR, thereby preventing ligand binding [124]. These data suggest that cadmium affects GR function through interference of GR ligand and DNA binding, possibly due to changes in the redox state of the receptor.

Zinc. GR is a zinc-finger protein which contains two zinc molecules. Zinc is an essential mineral and commonly found in many biological enzymes and transcription factors. However, excessive zinc can result in ataxia, lethargy and copper deficiency. Zinc administration reduces glucocorticoid ligand binding in liver cytosols. As with cadmium, this could be inhibited by the reducing agent DTT, suggesting the involvement of dithiols in the ligand binding region [136]. In 2305 cells, zinc increases dexamethasone induction of the GR-responsive MMTV promoter possibly through a metallothionein‑mediated pathway [134]. Although not well described, zinc may have effects on GR function.

Mercury. Administration of mercury reduces glucocorticoid induction of the endogenous GR regulated gene TAT in rat livers [131]. Mercury decreases GR ligand binding in liver and kidney which could be reversed by DTT, suggesting the involvement of thiol groups [132]. Mercury also increases the interaction between the GR apo-receptor and Hsp70 and Hsp90 [133]. As for Zinc, mercury increases dexamethasone induction of the GR-responsive MMTV promoter in 2305 cells [134].

Other metals. There are a few indications that other heavy metals may also affect GR function. Extremely low levels of chromium enhance dexamethasone induction whereas higher levels repress dexamethasone induction of PEPCK [119,129]. Selenite, a selenium-containing ion, inhibits GR ligand binding and can be reversed by DTT [123,135]. Lead inhibits glucocorticoid induction of TAT in liver hepatoma cells [130] and low concentrations of beryllium inhibit glucocorticoid induction of TAT and ornithine decarboxylase [126,127].

These data suggest that heavy metals can affect GR function. In some cases (arsenite, cadmium (II), zinc, and selenite) the mechanism is through effects on the thiol groups in the ligand binding pocket of GR. The other effects, with the exception of arsenic, have not been well elucidated.

The effects of smoking on GR function have recently been described. Cigarette smoke contains components such as tar, ammonia, formaldehyde, cadmium, arsenic, and nicotine. As such it could be considered an environmental toxin and is worthy of review here. Differences in GR numbers/affinity and isoforms have been shown between smokers and non-smokers. Human bronchial epithelial cells (HBEC) from smokers contained GRs with a lower ligand binding affinity than non-smokers but with no changes in GR numbers [137]. In one study GR mRNA levels were decreased in COPD patients but there was no difference between smokers and non-smokers and no difference in the GRα/β mRNA ratio in bronchial epithelial cells [138]. However, another study showed that smoking reduced the GRα/β protein ratios in PBMCs both in normal healthy volunteers and in asthmatics [140]. In mice exposed to cigarette smoke a decrease in GRα protein was observed in the lungs [139]. Thus, there seems to be reduced GR protein levels in smokers and an increase in the presence of the dominant negative GRβ isoform. In addition, there are studies suggesting that GR-mediated gene regulation is altered with smoking. Smokers with respiratory disease had a lower amount of CYP3A5, a GR regulated gene, in their alveolar macrophages [141]. Cigarette smoke condensate also inhibits dexamethasone induction of ENaC mRNA in HAE cells [142] and cigarette smoke inhibits dexamethasone repression of IL-1β-induced TNFα and IL-8 in BAL macrophages [143]. This effect on GR suppression of cytokines has been suggested to involve histone deacetylase 2 (HDAC2). Cigarette smoke reduces expression of HDAC2 and HDAC activity, which correlates with the reduced suppression of IL-1β-induced cytokines [143]. This smoking-induced glucocorticoid insensitivity could be mimicked by HDAC inhibitors and hydrogen peroxide [143] and reversed by inhibition of PI3Kδ [139]. This suggests that smoking, through PI3Kδ, reduces the levels and activity of HDAC2, which, in turn inhibits GR-mediated gene repression.