Evan T. Keller^1,2,3, Jon Wanagat¹, W.B. Ershler^1,3

¹Glennan Center for Geriatrics and Gerontology, Departments of ²Pathology and ³Internal Medicine, Eastern Virginia Medical School.

Received 10/21/96; Accepted 11/01/96; On-line //96

4. Steroids and regulation of interleukin-6 expression

4.1. Glucocorticoid and interleukin-6 expression

Glucocorticoids repress expression of a variety of genes including proliferin, pro-opiomelanocortin, prolactin, and the a-subunit of glycoprotein hormone. Similarly, glucocorticoids inhibit IL-6 expression. During times of stress or inflammation IL-6 levels are increased. IL-6, in turn, can induce release of corticotrophin-releasing factor (124, 125), which results in elevated systemic levels of corticosteroids. These findings along with the observations that natural and synthetic corticosteroids inhibit IL-6 production from a variety of tissues (126-129), provide a mechanism for a negative-feedback loop. It was these observations along with the availability of the IL-6 promoter which were the impetus to analyze how glucocorticoid mediates repression of IL-6 expression at the molecular level. The initial studies demonstrated that dexamethasone could inhibit IL-1-induced transcriptional activation of the proximal 225 bp of the IL-6 promoter (130). Additionally, it was observed that dexamethasone abrogated the activity of the thymidine kinase (TK) minimal promoter fused downstream of either MRE I and MRE II when induced by IL-1, phorbol ester, or forskolin. These findings taken together with the observation that the glucocorticoid receptor (GR) could inhibit pseudorabies virus-induced activation of the proximal 110 bp fragment of the IL-6 promoter, in which the MRE had been deleted, prompted the investigators to examine for interaction of GR and the IL-6 promoter by DNAse I footprinting (130). They found that GR protected the MRE, the TATA box, the major transcription initiation site (similar to the initiator (Inr) (131)), and an as yet functionally uncharacterized region between -201 to -210. These findings were further supported by results from a DNA-binding immunoprecipitation assay which showed that cell extract from HeLa cells transfected with wildtype GR cDNA was capable of binding to the -225 fragment of the IL-6 promoter (albeit not as strongly as to the GRE in the MMTV promoter) (119). Additionally, mutation of the DNA binding domain resulted in loss of GR's ability to repress transcriptional activation (119). These data are compatible with a model in which corticosteroid activates GR which then occludes the IL-6 promoter at these activation sites, thus blocking the binding of positive-acting and basal transcription factors to the IL-6 promoter.

Because the above studies demonstrated that GR bound weakly to the IL-6 promoter and it had been previously documented that GR was capable of protein:protein interactions with c-Jun (132, 133), Ray et al. examined the possibility that GR interacted with NF-kB and NF-IL6; transcription factors known to stimulate the IL-6 promoter (120). Using murine F9 embryonal carcinoma cells, which are devoid of endogenous NF-IL6, AP-1, and Rel-like activities, Ray et al. demonstrated that expression plasmids encoding NF-IL6, or p65 alone could not stimulate the IL-6 promoter, whereas when used together, the IL-6 promoter was stimulated. Furthermore, dexamethasone could inhibit this activation (120). In contrast, transfection of HeLa cells with either plasmid alone, resulted in activation of the IL-6 promoter, suggesting that the transgenic protein interacted with the endogenous co-activating protein. Regardless, in HeLa cells, dexamethasone was capable of inhibiting NF-IL6 and p65-induced IL-6 promoter activity (120). Finally, in cross precipitation assays, it was demonstrated that GR bound to p65, but not NF-IL6. These results suggest that GR mediates inhibition of p65-induced activation of the IL-6 promoter through protein:protein interactions. This mechanism may occur in combination with the promoter occlusion mechanism described earlier.

Yet, additional clues on the action of GR on the IL-6 promoter may be gleaned from studies by Scheinman et. al. and Auphan et al. (134, 135). Though not evaluated on the IL-6 promoter itself, these groups demonstrated that dexamethasone induces IkappaBalpha protein and mRNA expression. Auphan et al. further demonstrated that dexamethasone could inhibit TNF-alpha-stimulated nuclear translocation of p65 (135). These data suggest that GR induces IkappaBalpha protein synthesis which results in cytoplasmic sequestration of NFkappaB culminating in decreased activation of the target promoter. This mechanism does not preclude the previously described mechanisms of promoter occlusion and GR:p65 protein interactions.

4.2. Estrogen and interleukin-6 expression

Estrogen's ability to repress IL-6 expression was first recognized in human endometrial stromal cells (23). Additional clues came from the observations that menopause or ovariectomy resulted in increased IL-6 serum levels (136) , increased IL-6 mRNA levels in bone cells (137), and increased IL-6 secretion by mononuclear cells (75, 138, 139). Further evidence for estrogen's ability to repress IL-6 expression is derived from studies which demonstrated that estradiol inhibits bone marrow stromal cell and osteoblastic cell IL-6 protein and mRNA production in vitro (18, 140) and that estradiol was as effective as neutralizing antibody to IL-6 to suppress osteoclast development in murine bone cell cultures (18) or in ovariectomized mice (19). Taken together, these data provide strong evidence for the occurrence of estrogen-mediated repression of IL-6 expression.

To explore estrogen's effect on the IL-6 promoter, Pottratz et al. (117) and Ray et al. (116) performed transient transfection assays using either a 1.2 Kb fragment of the promoter or a 225 bp fragment of the promoter. They found that basal IL-6 promoter activity was very low when used to drive a chloramphenicol acetyltransferase (CAT) gene in both HeLa, which does not express the estrogen receptor (ER), and the murine bone marrow stromal cell line MBA 13.2, which constitutively expresses the ER. However, phorbol-13-myristate acetate (PMA), IL-1, or TNF stimulated the promoter and 17ß-estradiol inhibited this activity in both cell lines. Transfection of the HeLa cells with ER was required to observe the suppression. These results suggest that 17ß-estradiol inhibits IL-6 gene transcriptional activation by an ER-dependent mechanism.

To investigate whether the ER-mediated repression was due to direct interaction between the ER and the IL-6 promoter, Pottratz et al. performed competition assays which assessed for the 225 bp IL-6 promoter fragment's ability to compete for binding of ER to a labeled estrogen response element (ERE) (117). However, even though ER bound to the labeled ERE, the 225 bp fragment did not compete with the ERE. Additionally, upon electrophoretic mobility shift assay (EMSA) the ERE could not compete shifted complexes formed from incubation of HeLa or MBA 13.2 nuclear extracts with the labeled 225 bp fragment and this fragment could not bind to ER (116, 117). In summary, these results suggested that ER does not bind to the 225 bp IL-6 promoter fragment even though it inhibits its activity. These results were not entirely surprising based on the observation that there was no ERE within the 225 bp IL-6 promoter fragment. However, they led to the hypothesis that ER was working through inhibition of positive acting transcription factors by protein:protein interactions.

At nearly the same time, Ray et al. reported that wildtype ER, but not ER with mutated or deleted DNA-binding domain (DBD), could mediate repression of IL-1-induced IL-6 promoter activity, yet if the ER DBD was replaced with a GR DBD, the resulting chimeric receptor was capable of mediating repression (116). These results were also observed if the IL-6 promoter activity was induced by co-transfection of HeLa cells with NF-IL6 and NFkappaB p65 subunit. However, the chimeric receptor, which could mediate repression, could not stimulate a ERE-reporter construct, thus suggesting that repression was not dependent on direct binding to the IL-6 promoter. Furthermore, overexpression of NFkappaB p65 by transient transfection inhibited ER's ability to transactivate an ERE-reporter construct (116). This result provided evidence for interaction between NFkappaB p65 and ER.

These studies were extended into the U2-OS human osteoblast and MCF-7 breast carcinoma cell lines by Stein and Yang (141). Similar to the observations in HeLa cells described above, the IL-6 promoter, even when deleted to 109 bp, was stimulated by IL-1alpha and this activation was repressed by 17ß-estradiol in the presence of either co-transfected ER (U2-OS cells) or native ER (MCF-7 cells). Further deletion of the promoter to 49 bp, in which both NF-IL6 and NFkappaB response elements are deleted, resulted in loss of promoter induction by IL-1b. Based on these data, Stein and Yang concluded that the ER target is between -109 and -49 (141). However, since the 49 bp promoter region was not stimulated, they could not observe ER-mediated repression if it was present, hence this conclusion may be premature.

To deduce which regions of the ER were necessary for repression, Stein and Yang performed a series of transient co-transfection experiments using mutated ER constructs and the IL-6 promoter (141). Deletion of the amino-terminus including the transcription accessory factor (TAF)-1 (delta1-179) domain still allowed for ER-mediated repression. Extending this deletion to include the DBD (delta1-281) resulted in loss of repression as did isolated deletion of the DBD (delta185-251). Additionally, deletion of the carboxy-terminus (delta271-595) including TAF-2 domain and the LBD resulted in loss of repression. Based on these data, the author's concluded that the DBD contributed to transrepression.

Based on the previous data that ER does not appear to bind to the IL-6 promoter (116, 117), yet can mediate transrepression of the IL-6 promoter, Stein and Yang explored for direct interaction between ER and NFkappaB p65, NFkappaB p50, or NF-IL6 (141). They found that all these in vitro translated proteins bound with bacterially expressed ER. Intriguingly, this interaction was not dependent on estrogen and deletion of the DBD did not effect the interaction. Furthermore, they demonstrated that ER and NFkappaB p65 or NF-IL-6 mutually represses each others transactivation abilities through a mechanism which does not induce IkappaBalpha. Based on these data, Stein and Yang concluded that binding of NFkappaB p65 or NF-IL-6 is the driving force which mediates transrepression. However, when considered with the observation described above that isolated deletion of the ER DBD results in loss of transrepression, these data suggest that ER's binding capability for these transcription factors and its ability to mediate transrepression are in fact on different domains of the ER and are mediated by some mechanism which involves more than just binding of these transcription factors. Further studies are needed to resolve these issues.

4.3. Androgen and interleukin-6 expression

Androgens can repress expression of a variety of gene products (142-157). The first demonstration of androgen's ability to repress IL-6 expression was made in +/+LDA11 murine bone marrow stromal cell line which had been stimulated with IL-1 and TNF (18). In this study, 10 nM T was able to repress bioactive IL-6 expression by approximately 20% (as opposed to approximately 60% for 10 nM 17ßE₂). Curiously, when HeLa cells were co-transfected with ER and a CAT-reporter plasmid driven by a 225 bp fragment of the IL-6 promoter, 10 nM DHT inhibited PMA-induced activation by approximately 50% (as opposed to approximately 90% for 10 nM 17ß-estradiol (17ßE₂) (117). The authors accounted for this effect as due to T's affinity for the ER (117). However, this is unlikely as it has been previously demonstrated that T and DHT inhibits PMA-induced IL-6 promoter activation in HeLa cells transfected with AR, but not ER (158). This supports an earlier study in which it was reported that DHT antagonizes estrogen's effect in the uterus by decreasing estrogen-induced RNA transcription at a point subsequent to estrogen receptor binding (159). Several adrenal androgens which are not known to bind the AR (i.e., androstenedione, androstenediol, and dehydroepiandrosterone sulfate) mediate repression of the IL-6 promoter in HeLa cells transfected with AR. These experiments suggest that androgens are capable of mediating transrepression of IL-6 promoter activation.

We have further demonstrated that DHT requires the AR to mediated DHT's repressive effect in a transient transfection assay system (118). In our system, DHT inhibited PMA-induced activation of the IL-6 promoter by inhibiting translocation of NFkappaB. This was achieved through maintenance of IkappaBalpha levels even in the presence of PMA. Currently, it is unknown how DHT maintains IkappaBaalpha levels, but decreased phosphorylation or increased protein production are two possibilities.

The in vivo relevance of the above observations was demonstrated by Bellido et al. in an orchiectomized mouse model (158). Though they did not report serum or bone marrow IL-6 levels, they found that orchiectomy resulted in increased replication of bone marrow osteoclast progenitors and found that this could be prevented by administration of IL-6 neutralizing antibody or implantation of a slow release form of T. However, because T is converted to 17ßE₂, these results are not conclusive evidence of androgen's action. In fact, the observation of decreased bone density observed in a male patient whom had normal androgen levels, but a mutation resulting in a non-functional ER, suggests that estrogen's effects on bone in men are as important as androgen's (160). This hypothesis is supported by the observation that 17ßE₂ can inhibit bone loss observed in men treated by orchiectomy for prostate cancer (161). However, estrogen's ability to inhibit bone loss may be mediated through its transrepression of the IL-6 promoter, and thus these observations are still consistent with androgen loss resulting in increased IL-6 activity. Finally, that orchiectomy of mice without the IL-6 gene (generated by knockout technology) do not demonstrate the increased osteoclast proliferative effects, strongly supports that loss of androgen results in increased IL-6 levels (158).