Cutting Edge: Inhibition of the Retinoid X Receptor (RXR) Blocks T Helper 2 Differentiation and Prevents Allergic Lung Inflammation

Th cells play a critical role in immune regulation. Th1 cells produce IL-2 and IFN-γ, whereas Th2 cells secrete mostly IL-4, IL-5, and IL-13. Th1 and Th2 cells arise from a common precursor cell. Many factors control Th differentiation, including the cytokine milieu, type of APC, and mode of costimulation (1, 2).

It is becoming increasingly evident that certain nuclear hormone receptors profoundly affect the immune response. The retinoid X receptor (RXR)⁴ is activated by 9-cis-retinoic acid (9-cis-RA), a metabolite of vitamin A. Activation of RXR has been shown to inhibit IL-12 production by macrophages and to promote Th2 differentiation in vitro (3, 4). In vivo, vitamin A-deficient mice are unable to mount an effective Th2-mediated immune response and launch an excessive Th1-response instead (5). RXR serves as a heterodimerization partner for a variety of other nuclear hormone receptors (6). Among these are the retinoic acid receptor (RAR) and the vitamin D receptor (VDR). Ligands for both receptors have been shown to inhibit IL-12 production and to promote Th2 differentiation (3, 7, 8).

The development of allergic asthma is favored by an excessive Th2 response marked by high levels of IL-4 and IL-5, which augment IgE synthesis and eosinophil survival. Because RXR alone and in combination with other nuclear hormone receptors promotes Th2 differentiation, we wondered whether RXR could be exploited to block Th2 differentiation and associated disease. We show in this study that a pan-RXR-specific antagonist efficiently inhibited Th2 differentiation while promoting Th1 differentiation in vitro. Furthermore, the development of allergic lung inflammation was suppressed in vivo. This demonstrates that RXR might be a valuable target for interfering with the development of Th2-driven diseases such as allergic asthma.

CD4⁺ Th cells were isolated from the spleen and lymph nodes of DO11.10 mice using anti-CD4-coupled magnetic beads (Miltenyi Biotec). Purified Th cells were stimulated with 0.3 μM OVA_323–339 peptide and irradiated (2500 rad) BALB/c splenocytes. The pan-RXR antagonist Ro 26-5405 or 9-cis-RA (synthesized at F. Hoffmann-La Roche) were added at the indicated concentrations and were present throughout the culture period. Stock solutions for both compounds were prepared in DMSO. The final concentration of DMSO in the cultures was 0.1% in all conditions. To some cultures, cytokines and Abs were added as indicated. After 72 h, 100 U/ml human rIL-2 was added. The cells were expanded for 1 wk, and growing cells were split with medium containing fresh antagonist and IL-2. In some experiments, the Th cells were restimulated once more under the same conditions. Cytokine production was measured by either ELISA after restimulation with OVA and APC for 48 h or by intracellular cytokine staining after restimulation with PMA and ionomycin for 4 h (9).

Mice were housed under specific pathogen-free conditions, and all experiments were done in accordance with the institutional guidelines for animal care at the DIBIT (Milan, Italy). BALB/c and C57BL/6 ^p40−/− (B6.129S1-IL12b^tm1Jm/J) mice were purchased from Charles River Laboratories and The Jackson Laboratory, respectively. DO11.10 (C.Cg-Tg(DO11.10)10Dlo/J) and BALB/c ^IFNγ−/− (C.129S7(B6)-Ifng^tm1Ts/J) mice were bred in our animal facilities. Female mice aged 8–10 wk were used for all experiments. To induce allergic lung inflammation, mice were sensitized by injecting 10 μg of OVA adsorbed to 200 μl of alum (Serva) i.p. on days 0 and 7. Additionally, some mice were injected i.p. with 2 mg (100 mg/kg) of the RXR antagonist Ro26-5405 suspended in 50 μl of arachis oil on days 0, 1, 7, and 8. Control animals received arachis oil only. On day 14, the mice were placed in a Plexiglas box and exposed to an aerosol of 5% OVA solution in PBS generated by a Pari Star nebulizer (Pari GmbH) for 20 min. Seventy-two hours after this aerosol challenge, bronchoalveolar lavage (BAL) was performed. For differential cell counts, cytospins were prepared and stained with Diff-Quik (Dade Behring). Frozen sections of lung tissue were stained with H&E. OVA-specific serum IgE was measured using a kit (BD Pharmingen) and OVA (0.5 μg/ml) as Ag.

To examine the effect of RXR blockade on Th2 differentiation, we stimulated purified Th cells with APCs and OVA_323–339 in the presence of different doses of the pan-RXR antagonist Ro26-5405 (Fig. 1,g). This compound binds RXRα and γ without activating them. It does not bind to any of the RAR isoforms (Table I). Th cells primed in the absence of Ro26-5405 predominantly produced IL-4 but very little IFN-γ (Fig. 1,a). Treatment with the RXR antagonist reduced IL-4 and enhanced IFN-γ production in a dose-dependent manner. To characterize the effect of RXR blockade on a single-cell level, we performed intracellular cytokine staining. As shown in Fig. 1 b, DMSO-treated Th cells were predominantly of the Th2 phenotype, as judged by the high percentages of IL-4- (54%), IL-5- (57%), IL-10- (80%), and IL-13- (90%) producing cells. Treatment with the RXR antagonist nearly completely inhibited the development of IL-4-, IL-10-, and IL-5-producing cells and reduced the number of IL-13-positive cells to 50%. Instead, the number of IFN-γ-producing cells increased from 7 to 29%. Therefore, inhibition of RXR in vitro impairs Th2 differentiation while promoting Th1 differentiation.

IL-12 drives Th1 differentiation (2), and its expression by monocytes can be inhibited by some nuclear hormone receptor ligands (8, 10), providing a possible explanation for the reduced Th2 differentiation that we observed. However, neutralization of IL-12 did not influence the effect of the RXR antagonist on Th differentiation at all (Fig. 1,c, left panel). In addition to IL-12, IFN-γ is a very potent Th1-inducing cytokine (2). In the presence of neutralizing Abs against IFN-γ, the RXR antagonist only partially inhibited IL-4 and IL-10 production and failed to reduce IL-13 or enhance IFN-γ production. Interestingly, production of IL-5 was still completely blocked (Fig. 1 c, middle panel).

IL-4 is a very potent Th2-driving cytokine. Its reduced expression (Fig. 1,a) might account for the inhibition of other Th2 cytokines. We therefore tested whether addition of IL-4 would overcome the inhibition of Th2 differentiation by RXR blockade. As shown in Fig. 1 c (right panel), stimulation of Th cells under Th2-skewing conditions (IL-4 plus anti-IL-12) further increased the percentage of IL-4-producing cells, compared with nonskewing conditions. The addition of the RXR antagonist did not significantly reduce the production of IL-4, IL-10, and IL-13. However, to our surprise, the development of IL-5-producing Th cells was still completely inhibited by Ro 26-5405.

The inhibition of Th2 differentiation by Ro26-5405 was partly mediated by IFN-γ. In our culture conditions, IFN-γ could be derived from several cell types contained in the splenic APC, but also from the Th cells themselves. To test whether Ro 26-5405 inhibited Th2 differentiation by enhancing Th cell-derived IFN-γ, we stimulated purified Th cells with anti-CD3 and anti-CD28 in the presence or absence of Ro26-5405. As shown in Fig. 1 d, the Th cells developed mainly into Th2 cells, and cytokine production was comparable to cells stimulated with Ag and APC. Addition of Ro26-5405 did not alter the production of IL-4 and IL-13 and only moderately reduced IL-10 production. Therefore, RXR blockade most likely enhances IFN-γ production in non-T cells. In fact, dendritic cells have been demonstrated to produce significant amounts of IFN-γ (11). An early spike in IFN-γ could then induce IFN-γ production by Th cells (2) and thus contribute to the inhibition of Th2 cytokine production. In contrast, the production of IL-5 was completely abolished by RXR blockade even in the absence of APC. This indicated that the RXR antagonist inhibited Th2 differentiation on multiple levels: inhibition of IL-4, −10, and −13 was mediated by APC, whereas IL-5 production was directly inhibited in Th cells.

Th1 and Th2 differentiation are largely controlled by the two transcription factors, T-bet and GATA-3, respectively (2). Both vitamin D and vitamin A have been described to induce GATA-3 (4, 7). Consistent with their Th2-like phenotype, Th cells stimulated under neutral conditions expressed GATA-3 but very little T-bet, regardless of whether the cells had been stimulated with OVA and APC or with anti-CD3 and anti-CD28 (Fig. 1,e, lanes 1 and 3, and Fig. 1 f). Treatment with Ro 26-5405 reduced GATA-3 expression in OVA/APC-stimulated Th cells but not in anti-CD3-stimulated Th cells. This argues against a direct regulation of GATA-3 by RXR and indicates that the lower GATA-3 level was a secondary effect of impaired Th2 differentiation. However, it cannot be excluded that APC might serve as a source for RXR agonists (12) that might be counteracted by Ro26-5405.

Both retinoids and vitamin D have been described to directly enhance IL-4 production by Th cells (4, 7). In contrast, we found that inhibition of most Th2 cytokines by the RXR antagonist required the presence of APC, whereas IL-5 was inhibited directly in Th cells. To exclude the possibility that Ro 26-5405 might inhibit Th2 differentiation via a mechanism not involving RXR or its binding partners, we tested whether activation of RXR would revert all effects of the RXR antagonist. Addition of 9-cis-RA, a ligand for both RAR and RXR, completely restored Th2 differentiation in the APC/OVA system (Fig. 2,a) and IL-5 production in anti-CD3/anti-CD28-stimulated Th cells (Fig. 2 b). This suggests that Ro26-5405 acts on RXR homodimers or on RXR-containing heterodimers.

BALB/c mice were sensitized by two i.p. injections with OVA adsorbed to alum. Additionally, mice received either 2 mg of Ro26-5405 or vehicle alone. Fourteen days after the first immunization, the mice were challenged with an OVA aerosol, and lung inflammation was measured 72 h hours later by BAL. As shown in Fig. 3,a, the BAL of challenged, untreated mice contained high numbers of eosinophils at this time point. Mice that had received the RXR antagonist during sensitization contained significantly lower numbers of eosinophils and lymphocytes in their airways, whereas the numbers of macrophages and neutrophils were unaffected. Lung sections stained with H&E revealed a strongly reduced inflammatory response in treated vs mock-treated mice (Fig. 3,d). Quantitative analysis of the number of eosinophils found in the lung parenchyma near venules or airways reflected the reduction of eosinophil numbers already observed in BAL (Fig. 3,c). Along with eosinophilia, the levels of OVA-specific IgE were reduced significantly in treated mice (Fig. 3,b). The numbers of eosinophils and the level of Ag-specific IgE are strongly dependent on IL-5 and IL-4, respectively. We therefore examined the cytokine production by spleen and peribronchial lymph node (PBLN) cells isolated from treated vs mock-treated animals. Both spleen and LN cells from treated mice produced 3-fold less IL-4, IL-5, and IL-13 after restimulation than cells obtained from control animals. The production of IFN-γ was unchanged (Fig. 3 e).

Our in vitro results showed that blockade of RXR resulted in reduced Th2 and enhanced Th1 differentiation. This effect was independent of IL-12 but partially dependent on IFN-γ. However, in vivo, up-regulation of IL-12 by treatment with Ro26-5405 might still occur and might contribute to the prevention of allergic inflammation. To test whether this was the case, we used IL-12 (p40)-deficient mice in the model of allergic lung inflammation. In IL-12-deficient mice, treatment with the RXR antagonist reduced the eosinophil influx into the airways to the same extent as it did in wild-type animals. In addition, PBLN cells isolated from treated mice produced less IL-5 than those derived from control mice (Fig. 4 a). Therefore, IL-12 did not play a role in the inhibition of allergic inflammation by Ro26-5405.

In vitro, neutralization of IFN-γ partly prevented the switch from Th2 to Th1 differentiation caused by RXR blockade, although it did not restore IL-5 production. Likewise, treatment with Ro26-5405 in vivo might cause increased IFN-γ production and thereby prevent the development of allergic inflammation. However, when we used IFN-γ-deficient mice, treatment with the RXR antagonist still completely prevented lung eosinophilia and IL-5 production by PBLN cells (Fig. 4,b). To rule out the possibility that Ro26-5405 simply inhibits the immune response in general, we examined how treatment with the RXR antagonist would affect a Th1-driven immune response. To this aim, we modified the asthma model such that BALB/c mice were immunized with OVA emulsified in CFA instead of alum. As shown in Fig. 4 c, this protocol induced a Th1 response manifesting in a modest influx of neutrophils instead of eosinophils into the bronchi. The number of neutrophils increased significantly in mice that had been treated with the RXR antagonist during the sensitization phase.

Our data demonstrate that blockade of RXR did not induce a counterregulatory Th1 response, but rather specifically inhibited Th2-mediated inflammation. In fact, others have shown that Th1 cells do not counteract Th2 inflammation in this model, but rather exacerbate disease (13). It might be argued that the RXR antagonist might be toxic for certain cell types and thereby inhibit the immune response. We believe this not to be the case for several reasons. First, as described before, a Th1-driven inflammation was not inhibited. Second, a toxic effect on eosinophils is unlikely because we found the same number of eosinophil-like cells in the bone marrow of treated and untreated mice, and the RXR antagonist did not inhibit IL-5-driven eosinophil differentiation from bone marrow cells in vitro (data not shown). Third, the proliferation of Th cells stimulated with anti-CD3 and anti-CD28 in vitro was not affected by Ro26-5405 (data not shown). However, we did observe somewhat reduced proliferation in the OVA/APC system. This would indicate that Ag presentation or costimulation by APC might be affected. This, in turn, might influence Th differentiation. For example, B7/CD28 interactions appear to be required for Th2 differentiation (14, 15). In contrast, LFA-1/ICAM-1 interactions have been reported to favor Th1 over Th2 differentiation (15, 16, 17), and the signal strength appears to influence T cell differentiation as well (2).

Because RXR serves as a heterodimerization partner for several different nuclear hormone receptors such as VDR, RAR, and peroxisome proliferator-activated receptor, it is possible that treatment with an RXR antagonist might inhibit signaling by these receptors as well. In this context, it is noteworthy that VDR-deficient mice have been reported to be resistant against development of experimentally induced allergic asthma (18).

Taken together, we have shown for the first time that, by blocking the nuclear hormone receptor RXR, the differentiation of Th2 cells can be inhibited in vitro, and that the development of allergic disease is efficiently prevented in vivo. Modulation of nuclear hormone receptor signaling might prove to be a powerful tool to treat Th2-mediated diseases such as allergic asthma or atopic dermatitis.

We thank Dr. U. Wiegand (Roche) for providing 9-cis-RA.

M. Klaus, W. Bollag, F. Sinigaglia, and P. Panina-Bordignon are among the inventors of the patent, “Retinoid Antagonists and Use Thereof” (International Publication Number WO2000/053562).