Our main finding is that estrogen-mediated suppression of inflammation as seen in the TT-DTH response and in experimental arthritis is mediated via ERα but not ERβ.
The mechanisms underlying estrogen modulation of inflammation are not well understood. Both pro-inflammatory and anti-inflammatory effects have been reported (recently reviewed by Straub [20
]. The dualistic action of estrogens is, among others, explained by the concentration of circulating estrogen, the differential expression of ERα or ERβ in different cell types and the microenvironment involved impacting the class of the immune response. Here, we focused on defining the contribution of ERα or ERβ to modulation of the inflammatory response in experimental rat and mouse models.
Previously, ERβ-mediated suppression of inflammation in Lewis rat AA was reported [25
]. This prompted us to study the effects of EE and our selective ERB-79 in rat AA. Interestingly, our data showed significant suppression of signs and symptoms in rat AA with EE but not with ERB-79, suggesting that in this model the suppression of arthritis is ERα-mediated. Thus, our findings do not confirm the reported effects of an other ERβ agonistic compound (ERB-041) in rat AA [25
]. This discrepancy may be explained by a higher potency and selectivity of ERB-79 (484-fold over ERα). The combined data on ERB-79 selectivity, PharmacoKinetics (PK) and the results of the ERα titration study in vivo
provides evidence that ERB-79 at 3 mg/kg subcutaneous does not demonstrate significant ERα-mediated activity in vivo
, but is very likely to engage ERβ.
Next, we chose to further study the role of the different ERs in mice in vivo
using well-described ERα and ERβ knockout mice (compared with wild type) and two highly selective compounds from our compound libraries, which agonistically engage ERα (ERA-63) or ERβ (ERB-79) for cross comparison. This approach was inspired by the notions that: female mice, as seen in the human situation, show pregnancy-associated protection of joint disease with post-partum flares of arthritis [4
]; and effects of estrogens are best studied in vivo
representing a system with near physiological levels of ERα and ERβ allowing for ER cross regulation and signaling in context.
Suppressive estrogen effects on DTH responses have been observed previously [37
]. Also, blocking of ERs by the antagonist ICI 182,780 significantly increased the DTH response [38
]. In addition, Islander and colleagues [37
] showed that E2 decreased the DTH response in wild type mice whereas this was not seen in ERαβ double knockout mice. These data substantiate the role of estrogens in suppression of the DTH response but do not elucidate the relative roles of ERα or ERβ in this process. Our study is the first to show that treatment with a selective ERα agonist (ERA-63), but not with an ERβ agonist (ERB-79), significantly reduces antigen-specific swelling in the TT-DTH model. This was further confirmed by the use of ERα- and Erβ knockout mice where the ERα agonist ERA-63 decreased the DTH response in both wild type and ERβ-/-
but not in ERα-/-
Previous studies have demonstrated effective treatment of inflammation in models of autoimmune disease using estrogens [8
]. Also, estrogens were effective in suppression of joint inflammation and clinical signs of arthritis in mouse and rat CIA [23
]. Moreover, ER-receptor blockade using the ER antagonist ICI 182,780 triggered an earlier onset and increased severity of CIA [10
]. A number of studies using different selective ER modulators in experimental models of autoimmunity suggest that suppression of inflammation is ERα-mediated rather than ERβ-mediated [41
]. Recently, the study by Yh and colleagues showed that estrogen-mediated modulation of inflammatory symptoms in mouse antigen-induced arthritis was ERα-mediated. An ERβ selective compound (8beta-VE2) had no effect in this model [42
]. In addition, it has been suggested that ERα, in contrast to ERβ, has a major role in bone homeostasis and therefore may protect against inflammation-induced bone loss [43
To confirm that estrogen-mediated suppression of inflammation is ERα-mediated in ongoing arthritis, mice with CIA, having scores ranging between 0.25 and 1.25, were treated with EE and ERA-63. ERA-63 strongly suppressed the ongoing arthritic process as evidenced by both a significant reduction of the AUC and a reduction in joint histopathology scores. Moreover, we observed significantly decreased serum COMP levels in the ERA-63 and EE-treated mice. The reductions in COMP levels were associated with prevention of cartilage destruction as evidenced by histopathological examination.
Experimental and clinical studies have established prominent roles for TNFα, IL-6 and IL-1 inflammatory pathways in arthritis. In CIA, an increase in the arthritis score of the knee joints was associated with an increase in IL-1 mRNA levels [33
]. In addition, suppression of CIA was observed using antibodies against TNFα and IL-1 [32
]. We showed that reduction in symptoms and associated joint pathology by ERA-63 was associated with significantly reduced IL-1β, IL-6, IL-12p40, KC and RANTES protein levels in the synovium. This is in line with previous studies showing estrogen-mediated suppression of nuclear factor (NF) κB activation. It is tempting to speculate that ER cross talk with NFκB may be ligand dependent. Selective ER modulators or ERα-selective ligands may thus have differential effects in different cells. Indeed, E2 was found to suppress NFκB activation whereas the selective ER modulators raloxifene or tamoxifene were inactive in this model system [41
The role of estrogens in inflammation was recently reviewed [20
]. It was proposed, substantiated by numerous studies, that the humoral immune response is stimulated at a broad range of physiologic estrogen concentrations (post-menopausal through to late pregnancy levels) whereas both the innate and the cellular response are suppressed at high physiologic estrogen concentrations (pregnancy levels). This hypothesis would, to a certain extent, explain the higher frequencies of certain autoimmune diseases with a strong B cell component (for example, SLE) in women in the reproductive years. Moreover, it would explain the increase in development of autoimmunity (for example, RA) in menopause when estrogen levels are relatively low.
Our studies unequivocally show that in DTH and in two experimental arthritis models, ERα agonism is needed to suppress the inflammatory response. There is still some controversy around the topic of additional ERs such as GPR30 [44
]. Our current study and the study by Engdahl and colleagues confirm the important role of ERα in arthritis suppression and imply that a role for GPR30 in inflammation is not likely [45
]. Further studies will be needed to elucidate the relative roles of ERα and ERβ in human autoimmune diseases.
Differential effects of ERα and ERβ ligands in EAE have been described [46
]. Moreover, clinical trials with oestrogens in multiple sclerosis have been described showing immune modulatory effects [48
]. Clinical trials involving estrogen suppletion in RA have led to conflicting reports. Early studies, without placebo-controlled treatment groups, demonstrated efficacy of estrogen treatment in RA [50
]. In placebo-controlled trials, however, different outcomes were documented. Studies with clinical efficacy [52
] but also studies with marginal [54
] or no improvement have been reported [55
]. The reasons for the contradictory results on clinical signs in these studies were attributed to selection of the patients, design of the study and the readouts, the power to detect a clinical effect and the use of a combination of estrogens and progestagens, which may obscure effects of estrogen alone [5
]. Importantly, in several trials, changes in bone formation (osteocalcin) and bone resorption (CTXII) markers were in agreement with favorable effects of estrogens on bone mineral density [54
]. Recently, the data from a first proof of concept trial in postmenopausal female RA patients (on concomitant treatment with methotrexate or sulfasalazine) failed to demonstrate efficacy of ERA-63 in spite of good pharmacodynamics [57
]. It is feasible that the length of the treatment period (10 weeks) is too short to modulate clinical disease expression under the cover of concurrent treatment. Alternatively, ERα agonism is not beneficial in this group of RA patients. Clearly, further studies are needed to elucidate the relative roles of ERα and ERβ in human autoimmune disease in order to effectively translate this knowledge to novel targeted therapies.