Edited by Jan-Åke Gustafsson, Karolinska Institute, Huddinge, Sweden (received for review August 9, 2004)
Inflammation is now recognized as a key component in a number of diseases such as atherosclerosis, rheumatoid arthritis, and inflammatory bowel disease. The transcription factor NF-κB has been shown to be involved in both the early and late stages of the inflammatory-proliferative process. In this report, we describe the identification of the pathway-selective estrogen receptor (ER) ligand, WAY-169916, that inhibits NF-κB transcriptional activity but is devoid of conventional estrogenic activity. This pathway-selective ligand does not promote the classic actions of estrogens such as stimulation of uterine proliferation or ER-mediated gene expression, but is a potent antiinflammatory agent, as demonstrated in the HLA-B27 transgenic rat model of inflammatory bowel disease. Our results indicate the potential utility of pathway-selective ER ligands such as WAY-169916 in the treatment of chronic inflammatory diseases.
Inflammation is now recognized as one of the primary components underlying a variety of diseases, including atherosclerosis, arthritis, asthma, sepsis, and inflammatory bowel disease (IBD) (1). Cellular events associated with inflammation include immune cell adhesion and infiltration, resulting in the amplification and prolongation of an inflammatory response that can be mediated through NF-κB activation (2, 3). NF-κB is a redoxsensitive transcription factor that regulates a multitude of inflammatory genes, including cytokines, chemokines, adhesion molecules, and acute phase proteins. NF-κB is a dimeric transcription factor composed of homodimeric and heterodimeric complexes of the Rel family of proteins, p65 (Rel A), p50/105, c-Rel, p52/100, and Rel B. Binding of IκB to NF-κB masks the NF-κB nuclear localization signal and sequesters NF-κB in a nonactivated form in the cytoplasm. Cell activation by a variety of extracellular signals such as oxidative stress, cytokines, and lipopolysaccharide induces a cascade of events that lead to degradation of IκB and activated NF-κB then translocates to the nucleus where it binds to DNA elements in the promoters of a number of proinflammatory gene families (4).
Although well known for their classic effects on the reproductive tract and action by means of estrogen response elements in gene promoters, nonselective estrogens such as 17β-estradiol (E2) are also known to have antiinflammatory activity (5–8), and this activity is observed in diverse models of disease such as atherosclerosis, sepsis, arthritis, and IBD (9–14). This nonclassical antiinflammatory activity of estrogen has been attributed to interference of NF-κB activity (15, 16), and one that has been reported to occur through multiple mechanisms, including direct protein–protein interactions (15, 16), inhibition of NF-κB DNA binding (17, 18), induction of IκB expression (19), or by means of coactivator sharing, as we and others have demonstrated (20, 21).
Two estrogen receptors (ERs) have been identified (ERα and ERβ), and both are widely distributed throughout numerous organs (22) and cells of the immune system (23, 24). ERα expression predominates in the uterus, pituitary, kidney, and adrenal gland, whereas ERβ expression is highest in the prostate, ovary, bladder, and lung. The two receptors have nearly identical DNA-binding domains and can activate transcription through binding to identical ER response elements (25, 26), and both can antagonize NF-κB functional activity (6, 27).
We set out to identify pathway-selective ER ligands, compounds that would retain the ability to antagonize NF-κB functional activity but be devoid of conventional estrogenic action. We report on the first compound in this class, the nonsteroidal-selective NF-κB inhibitor WAY-169916. It is an orally active molecule that selectively inhibits NF-κB activity by means of either ERα or ERβ. This selectivity of action was confirmed in vivo in a high-fat mouse model in which WAY-169916 inhibited diet-induced hepatic inflammatory gene expression without stimulating uterine wet weight or hepatic ER-mediated gene expression. Finally, the selective inhibition of NF-κB activity by WAY-169916 may have utility in the treatment of chronic inflammatory diseases, as demonstrated in the HLA-B27 transgenic rat model of IBD.
Viral Vectors. The Ad5-IL6.Luc virus was created by cloning the XbaI to SalI ≈1.2 kb of the human IL-6 promoter, luciferase gene, and simian virus 40 polyadenylation signal sequence into an Ad5 ΔE1a vector plasmid containing adenovirus sequences from map units 0–17, with deletion of E1a region between map units 1.4–9.1. The Ad5 recombinant plasmid was then linearized with EcoRI and transfected along with the ClaI fragment of Ad5 with the E3 region deletion (80–88 map units) into human embryonic kidney (HEK) 293 cells. Virus plaques generated by homologous recombination were isolated and amplified and the virus stock was prepared in HEK 293 cells.
The Ad5-wt-ERα virus was created by cloning the MluI–-XhoI fragment containing the CMV promoter and human ERα-coding sequence into the corresponding site in a modified Ad5 E1a vector plasmid, containing simian virus 40 polyadenylation signal sequence downstream to an XbaI site. The resulting adenorecombinant plasmid was then linearized and used for recombinant virus isolation as described above. The Ad5–3x(NFκB).Luc and Ad5-wt-ERβ virus were described (20, 28).
Transfection Experiments. Human aorta endothelial cell 1 (HAECT-1) cells (29) were cultured in EGM (GIBCO) containing 10% FBS, 1% GlutaMAX, and 0.1% antimycotic-antibiotic in Falcon T-175 culture flasks at 34°C in 5% CO2, and infected for 4 h with Ad5-wt-ER virus. After overnight recovery, the cells were next infected for 2 h with the Ad5–3x(NFκB).Luc or Ad5-IL6.Luc virus. Test compound and 2 ng/ml IL-1β were added to the cells for 15–20 h, the media were removed and used for analysis of IL-6 levels by ELISA (R & D Systems). The cell lysate was assayed for luciferase activity and evaluation of creatine kinase (CK) expression (Stanbio Laboratory, Boerne, TX). The values are reported as mean ± SEM with n = 4 from at least three independent experiments.
High-Fat Mouse Model. Experiments were performed as described (30). Briefly, ovariectomized C57BL/6 mice (Taconic Farms) were fed a chow or high-fat diet (15.75% fat, 1.25% cholesterol, and 0.5% sodium cholate) (Purina diet nos. 8927 and 21539, respectively), and 17α-ethynylestradiol (EE; 10 μg/kg) or WAY-169916 (10 mg/kg) was administered daily by gavage in a 0.5% methylcellulose/2% Tween 80 vehicle (0.1 ml per mouse) for 5 weeks. Liver RNA was isolated for gene expression analysis by real-time RT-PCR, and uterine wet weight was recorded. All mRNA levels were normalized for GAPDH expression. Values are reported as the mean ± SEM for each group from two separate experiments with eight mice per group, with the mean expression level in mice fed the chow diet defined as 1. Differences are considered significant if P < 0.05 vs. atherogenic diet with vehicle treatment.
Uterotrophic Assay. The effect of WAY-169916 on increasing the wet weight of the sexually immature rat uterus was evaluated as described (31). Briefly, rats (19 days of age) were treated s.c. for 3 d with test compound, and their uteri were excised and weighed after expressing any luminal fluid. A one-way ANOVA (Dunnett's test with Huber weighting) was used to assess differences between the groups with the vehicle group set as the comparator, and differences are considered significant if P < 0.05.
HLA-B27 Transgenic Rat Model of IBD. Male HLA-B27 transgenic rats were obtained from Taconic Farms, and the details of treatment, assessment of clinical signs, and evaluation of colonic histological lesions of the HLA-B27 transgenic rat have been described (14, 32). Briefly, 22- to 26-week-old rats were placed into groups with four rats per group, and were dosed orally, once per day for 8 days with vehicle (2% Tween 80/0.5% methylcellulose), WAY-169916 (0.5, 0.15, and 0.05 mg/kg), or WAY-169916 (0.5 mg/kg) plus ICI 182,780 (ICI; 25 mg/kg s.c. twice a day). Stool quality was observed daily and graded according to the following scale: diarrhea = 3, soft stool = 2, and normal stool = 1. Histologic results were obtained by systematic analysis of four standard areas of the descending colon from each rat (33) and are presented as mean ± SD. Differences are considered significant if P < 0.05.
Identification of Pathway-Selective ER NF-κB Inhibitors. HAECT-1 cells infected with an ERα expression vector and a NF-κB luciferase reporter were used to demonstrate the in vitro selectivity of WAY-169916. NF-κB luciferase reporter activity was induced by treatment with IL-1β, and its activity can be repressed upon cotreatment with nonselective estrogens such as E2 as observed previously (15, 16, 20). Estrogen treatment of HAECT-1 cells also induces the expression of CK (34). CK is an estrogen-inducible enzyme through an estrogen response element in its promoter and served as an in vitro indicator of conventional estrogenic action.
In ERα-expressing HAECT-1 cells, the concentration of WAY-169916 (Fig. 1A ) that inhibited 50% of IL-1β-stimulated NF-κB reporter activity (IC50) was 93 nM and was 82% as efficacious as E2, which had an IC50 value of 1 nM (Fig. 1B ). The tissue-selective estrogen receptor modulator, raloxifene, was unable to inhibit NF-κB activity, which is consistent with previous observations (20). This inhibition of NF-κB luciferase activity by WAY-169916 correlated with the reduction of IL-6 expressed in the media from these cells (Fig. 1C ). Importantly, WAY-169916 did not stimulate CK expression comparable with E2, which stimulated CK activity ≈1.9-fold with an EC50 value of 2 nM (Fig. 1D ). WAY-169916 treatment resulted in a slight elevation of CK activity (≈20%) at 1 μM.
The in vitro activity of WAY-169916. (A) The structure of WAY-169916. (B) Effect of WAY-169916 on NF-κB-mediated luciferase reporter activity in HAECT-1 cells expressing ERα. The cells were stimulated with IL-1β and cotreated with increasing concentrations of WAY-169916 (▵), E2 (▪), or raloxifene (•). (C) IL-6 expression from the same experiment was determined by ELISA, using the culture media. (D) The remaining lysate from B was used to determine CK activity. (E) The experiment was performed as in B, except the cells were treated with 1 μM WAY-169916, raloxifene, or ICI as indicated. (F) The experiment was performed as in D, except the cells were treated with 2 μM WAY-169916 or 30 nM E2 as indicated. *, P < 0.1 vs. vehicle treatment; **, P < 0.1 vs. E2 treatment.
WAY-169916 had no effect on ERα expression levels, and no inhibition of NF-κB reporter activity was observed at concentrations up to 10 μM in cells that did not express ERα (Fig. 5, which is published as supporting information on the PNAS web site). Moreover, both raloxifene and the pure ER antagonist ICI effectively antagonized WAY-169916's activity (Fig. 1E ). Conversely, WAY-169916 treatment could also antagonize the E2-mediated activation of CK (Fig. 1F ), demonstrating that WAY-169916 functions in an ER-dependent manner.
WAY-169916 Can Function Through both ERα and ERβ. To determine whether WAY-169916 could also function through ERβ, HAECT-1 cell transfection experiments overexpressing ERβ were conducted. WAY-169916 inhibited a NF-κB-driven human IL-6 promoter reporter with an IC50 value of 100 nM with an effect of 145% relative to E2 (Fig. 2A ). In the presence of ERα, IL-6 promoter activity was inhibited by WAY-169916 with an IC50 value of 90 nM and an effect of 100% relative to E2 (Fig. 2B ). As expected, cotreatment with ICI antagonized the WAY-169916-meidated effects, and no significant induction of CK was observed with either ERα or ERβ (data not shown). These results demonstrated that WAY-169916 can inhibit NF-κB transcriptional activity in HAECT-1 cells expressing either ERα or ERβ.
WAY-169916 inhibits IL-6 promoter activity through ERα or ERβ. (A) Effect of WAY-169916 on the human IL-6 promoter luciferase reporter activity in HAECT-1 cells expressing ERβ. The cells were stimulated with IL-1β and cotreated with increasing concentrations of either WAY-169916 (▵) or E2 (▪). (B) Experiments were performed as above, except the cells were expressing ERα instead of ERβ.
The interaction of WAY-169916 with both ERα and ERβ was confirmed in competitive radioligand-binding assays. WAY-169916 displaced [3H]E2 from ERα-ligand-binding domain protein with an IC50 value of 1,300 nM, and from ERβ-ligand-binding domain with an IC50 value 106 nM (Table 3, which is published as supporting information on the PNAS web site). In comparison, raloxifene resulted in IC50 values of 4.1 and 43 nM for ERα and ERβ, respectively. Self-competition with radioinert E2 resulted in IC50 values of 3 nM for both receptors. Therefore, WAY-169916 can function through both ERα and ERβ and can antagonize E2's binding to both ERs.
In Vivo Selectivity of WAY-169916. To demonstrate the selectivity of WAY-169916 in vivo, its activity was determined in a high-fat diet mouse model in which the diet induces NF-κB activity and expression of inflammatory genes in the liver (35). The inflammation-associated genes induced by the diet include vascular cell adhesion molecule 1, TNF-α, and regulated upon activation, normal T cell expressed and secreted factor (30). Daily oral treatment of ovariectomized mice with EE (10 μg/kg) or WAY-169916 (10 mg/kg) significantly inhibited the diet induction of all of these genes (Fig. 3A ). More importantly, unlike EE, which increases uterine wet weight and stimulates liver expression of estrogen target genes, including intestinal trefoil factor and myo-inositol-1-phosphate synthase, WAY-169916 was without effect (Fig. 3B ).
Selective inhibition of diet-induced inflammatory gene expression by WAY-169916. Ovariectomized C57BL/6 mice were fed a chow diet with vehicle treatment (hatched boxes), a high-fat diet with vehicle (black boxes), WAY-169916 (10 mg/kg orally, white boxes), or EE (10 μg/kg orally, gray boxes). Steady-state liver mRNA levels for vascular cell adhesion molecule 1 (VCAM-1), TNFα, and regulated upon activation, normal T cell-expressed and secreted factor (RANTES) (A) or intestinal trefoil factor (ITF) and myo-inositol-1-phosphate synthase (IPS) (B) are shown. Uterine wet weights were also recorded.
The lack of estrogenic activity of WAY-169916 on the uterus was further confirmed in the sexually immature rat by using a highly sensitive estrogenic bioassay (31). The nonselective estrogen EE (0.06 μg per 50-g rat) increased uterine wet weight by 4.6-fold, whereas WAY-169916 administered orally at 2 mg per 50-g rat (≈50 mg/kg) was without effect (Table 1). These data demonstrate the in vivo selectivity of WAY-169916, the ability to inhibit inflammatory gene expression without classic estrogenic effects such as its transcriptional and uterotrophic activity.
WAY-169916 Activity in the HLA-B27 Transgenic Rat Model of IBD. The HLA-B27 transgenic rat expresses two human proteins (HLA-B27 and β2 microglobulin) that over time, provoke a misdirected immune response. One of the first phenotypes the rats manifest is chronic intestinal inflammation, and thus provide a model of IBD (36) in which EE treatment has been shown to be effective when dosed as low as 0.01 mg/kg orally (14). Treatment of male HLA-B27 transgenic rats with WAY-169916 rapidly converts their chronic diarrhea to a normal stool (Fig. 4A ). Efficacy was observed at oral daily dosages as low as 0.05 mg/kg. Importantly, the effects of WAY-169916 are blocked by coadministration of the ER antagonist ICI (Fig. 4B ), indicating that the actions of WAY-169916 in this model are mediated by means of the ER. The results of microscopic examination of the colons from these animals are seen in Table 2. Treatment with WAY-169916 significantly reduced all histologic parameters of intestinal inflammation, including ulceration, inflammatory cell infiltrates, depth of lesion, degree of fibrosis, and total histological score. Representative photomicrographs of colon specimens treated with vehicle or WAY-169916 (0.5 mg/kg) are seen in Fig. 6, which is published as supporting information on the PNAS web site. Overall, the efficacy of WAY-169916 was comparable with that observed with EE (Table 4, which is published as supporting information on the PNAS web site) (14).
Effect of WAY-169916 on stool scores in the HLA-B27 transgenic rat. (A) Rats demonstrating chronic diarrhea were treated with vehicle or WAY-169916 (0.5, 0.15, and 0.05 mg/kg orally) for 8 days. (B) Rats demonstrating chronic diarrhea were treated with vehicle, WAY-169916 (0.5 mg/kg orally, once a day), or WAY-169916 plus ICI (25 mg/kg s.c., twice a day) for 8 days. Stool scoring was as follows: 3, diarrhea; 2, soft stool; 1, normal stool, and was plotted as mean ± SD.
Estrogens exert a diverse array of biological effects throughout the body. The beneficial role of estrogens on bone density, plasma lipid profiles, vasomotor symptoms, and colon cancer have been well documented (reviewed in ref. 37). However, estrogen therapy has also been associated with an increase risk for endometrial cancer, venous thrombosis, and uterine bleeding. In light of these divergent effects, the identification of more selective ligands for the ER have been sought, ones that would retain the beneficial aspects of estrogen therapy, while eliminating the potentially harmful side effects. This goal has been accomplished with the tissue-selective estrogen, raloxifene, currently prescribed for the treatment of osteoporosis and demonstrated preclinically with the synthetic ER ligand, estren, working through a nongenotropic mechanism (38). Here, we describe the identification of a first-in-class pathway-selective ER ligand, specifically selective inflammatory modulators such as WAY-169916 that specifically antagonize NF-κB transcriptional activity.
The inhibition of NF-κB transcriptional activity by E2 bound to ERα or ERβ has been well documented (6, 27). Here, we demonstrate that WAY-169916 retains the ability to inhibit NF-κB activity through both human ERα and ERβ in vitro with little or no induction of CK activity, suggesting that WAY-169916 pharmacologically dissociates the NF-κB-suppressive activity of ER from its transcriptional activity. The ability to dissociate these two activities has been previously demonstrated by using site-directed mutagenesis on ERα. Mutations within both helix 12 and helix 3 of ERα were identified that impaired ER transcriptional activity but retained the ability to antagonize NF-κB transcriptional activity in a ligand-dependent fashion (27). It will be important to determine whether these residues also contribute in the selectivity observed with WAY-169916.
Another means to probe for suppression of NF-κB transcriptional activity by ER is through pharmacological profiling. A variety of selective ER ligands have been described, and each of them displays an unique biological profile (32, 39, 40). Whereas both WAY-169916 and E2 can inhibit NF-κB functional activity, neither the tissue-selective estrogen receptor modulator, raloxifene, nor the pure ER antagonist, ICI, can promote this activity, which is consistent with previous observations (20). Therefore, the nature of the ER ligand imparts the NF-κB-suppressive activity, likely through the conformation the ER adopts upon ligand binding. It is unclear at the present time what critical changes occur in the ER to allow WAY-169916 to selectively inhibit NF-κB transcriptional activity. What is more apparent is why WAY-169916 is nonestrogenic. WAY-169916 was cocrystallized with the ligand-binding domain of human ERα and the 3D structure clearly shows that WAY-169916 places the receptor in an antagonist-like conformation (Z. Xu, unpublished work), and a more detailed analysis of this crystal structure may provide insight into the NF-κB selectivity observed with WAY-169916.
To demonstrate the selectivity of WAY-169916 in vivo, its activity was determined in a high-fat diet mouse model, as previously used to characterize EE's ability to antagonize diet-induced inflammatory gene expression (30). In this model, EE resulted in a global suppression of inflammatory gene expression after 5 weeks. ER was required for this activity because ICI completely blocked the inhibitory activity of EE. WAY-169916 was demonstrated to provide an equal suppression in inflammatory gene expression to that of EE. Importantly, no uterotrophic effect was observed with WAY-169916 treatment, unlike with EE, which resulted in a 4-fold elevation in uterine wet weight. These data provide in vivo evidence of a selective antiinflammatory activity for WAY-169916. WAY-169916 was also tested in a more stringent uterotrophic assay, the immature rat model (41). In this experiment, WAY-169916 was dosed at a concentration of 2 mg per rat, which was approximately a dose of 50 mg/kg. There was no effect on uterine activity observed with WAY-169916, and this dose was 1,000 times higher than its efficacious dose in the HLA-B27 transgenic rat. Therefore, it appears that WAY-169916 retains the antiinflammatory activity associated with estrogen, but is devoid of classic estrogenic activity. This finding was further demonstrated by its lack of activity in a rat model of hot flush or preventing ovariectomy-induced bone loss (data not shown). This approach has also been accomplished with the glucocorticoid receptor in which two nonsteroidal ligands have been identified that are more selective for transrepression than transactivation (42, 43). We believe WAY-169916 will provide a similar antiinflammatory activity without the side effects associated with either glucocorticoid or estrogen therapy.
The antiinflammatory activity of WAY-169916 was demonstrated in the HLA-B27 transgenic rat model of IBD. This model represents a chronic intestinal inflammation induced by the human class I major histocompatibility allele HLA-B27 that is associated with human disease. Treatment of HLA-B27 transgenic rats exhibiting chronic diarrhea with WAY-169916 resulted in a restoration of normal bowel scores and pronounced improvements in the colonic architecture that could be abrogated by ICI coadministration. Both ERα and ERβ are present in the colonic enterocytes and in the resident immune cells (44, 45), and targeting through either receptor appears capable of eliciting this antiinflammatory activity (14, 32). No loss of NF-κB nuclear localization or DNA binding was observed with WAY-169916 treatment (data not shown), suggesting that the ER is targeting NF-κB at its transcriptional level, as previously reported for EE in this model, possibly through regulation of cytokine expression in mast cells (Fig. 7, which is published as supporting information on the PNAS web site). Mast cells are thought to play a pathologic role in this model, and WAY-169916 regulation is consistent with that previously reported for EE (14). In addition, male rats were used in this study, demonstrating that the utility of WAY-169916 is not restricted by gender.
In summary, we detailed the activity of a pathway-selective ER ligand, WAY-169916, that selectively inhibits NF-κB transcriptional activity by means of ERα or ERβ, and imparts significant antiinflammatory activity in vivo. More importantly, no evidence for classic estrogenic activity, such as an increase in uterine weight, has been observed with this compound. These data provide evidence that the nonsteroidal pathway-selective ER ligand, WAY-169916, and other compounds in its class, may be therapeutically useful in the treatment of inflammatory diseases.
We thank all of the members of the Discovery and Development teams that contributed to this program and Marion Kasaian and Agnes Ciarletta (both of Wyeth Research, Cambridge, MA) for their contribution with the mast cell data.
↵ ** To whom correspondence should be addressed. E-mail: harnisd{at}wyeth.com.
↵ † Present address: Life Diagnostics Inc., 909 Old Fern Hill Road, West Chester, PA 19380.
↵ ∥ Present address: ArQule Research, 19 Presidential Way, Woburn, MA 01801.
Author contributions: C.C.C., J.C.K., H.A.H., E.T., R.C.W., S.J.A., R.J.S., and D.C.H. designed research; C.C.C., S.C., E.M., L.B.-M., A.M.E., J.C.K., L.M.A., Y.L., H.A.H., M.A., R.J.S., and D.C.H. performed research; E.M., R.A.B., and R.J.S. contributed new reagents/analytic tools; C.C.C., S.C., E.M., L.B.-M., A.M.E., J.C.K., L.M.A., Y.L., H.A.H., R.J.S., and D.C.H. analyzed data; and D.C.H. wrote the paper.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: ER, estrogen receptor; IBD, inflammatory bowel disease; E2, 17β-estradiol; EE, 17α-ethynylestradiol; CK, creatine kinase; ICI, ICI 182,780; HAECT-1, human aorta endothelial cell 1.
