Sulforaphane (SFN), an isothiocyanate, is part of an important group of naturally occurring small molecules with anti-inflammatory properties. The published reports are best conceivable with an inhibition of T cell function, but the mode of action remains unknown. We therefore analyzed the effect of SFN on T cell–mediated autoimmune disease. Feeding mice with SFN protected from severe experimental autoimmune encephalomyelitis. Disease amelioration was associated with reduced IL-17 and IFN-γ expression in draining lymph nodes. In vitro, SFN treatment of T cells did not directly alter T cell cytokine secretion. In contrast, SFN treatment of dendritic cells (DCs) inhibited TLR4-induced IL-12 and IL-23 production, and severely suppressed Th1 and Th17 development of T cells primed by SFN-treated DCs. SFN regulated the activity of the TLR4-induced transcription factor NF-κB, without affecting the degradation of its inhibitor IκB-α. Instead, SFN treatment of DCs resulted in strong expression of the stress response protein heme oxygenase-1 (HO-1), which interacts with and thereby inhibits NF-κB p65. Consistent with these findings, HO-1 bound to p65 and subsequently inhibited the p65 activity at the IL23a and IL12b promoters. Importantly, SFN suppressed Il23a and Il12b expression in vivo and silenced Th17/Th1 responses within the CNS. Thus, our data show that SFN improves Th17/Th1-mediated autoimmune disease by inducing HO-1 and inhibiting NF-κB p65-regulated IL-23 and IL-12 expression.
A number of important small molecules that naturally occur in plants and their modified derivatives have been established in the therapy of human diseases; for example, vincristine, an alkaloid extracted from Catharanthus roseus, etoposide, a cell toxin derived from the American Mayapple, or the natural taxoid derivatives paclitaxel and cabazitaxel that are used in cancer therapy (1). The immunotherapeutic agent cyclosporin A has been isolated from the fungus Tolypocladium inflatum (2). Other immunomodulating compounds originally derived from natural products are fingolimod, a myricin derivative of the fungus Isaria sinclairii (3), and the fumaric acid esters like dimethylfumarate (DMF), a derivative of fumaric acid that is found in the plant Fumaria officinalis (4). The isothiocyanate 1-isothiocyanato-(4R)-(methylsulfinyl) butane (sulforaphane) that naturally occurs in cruciferous vegetables is becoming part of this important group of natural substances with bioactivity. Isothiocyanates like sulforaphane (SFN) have been reported to exhibit antioxidative, antimicrobial, anti-inflammatory, and antitumoral properties (5, 6). Various mechanisms have been proposed to explain the beneficial effects of SFN on cancer prevention. SFN has been suggested to inhibit tumor cell growth in different human cancer cell lines through cell-cycle arrest and induction of apoptosis (7, 8). However, the antiproliferative and apoptosis-promoting effects of SFN have only been described in vitro and seem to require high doses (10–20 μM SFN) (7, 9). Chemoprevention is another potential pharmacologic effect of SFN. Early studies reported that SFN is a potent activator of phase II detoxifying enzymes (9). Subsequent studies confirmed that SFN is able to induce expression of phase II enzymes in various cell types and tissues in vitro and in vivo (10, 11). In contrast, SFN modifies phase I reactions during detoxification in the liver by inhibiting the activity of several cytochrome P450 enzymes potentially preventing the activation of procarcinogens (12, 13). A major proportion of the effects of SFN seems to be mediated via antioxidant response elements (ARE) and the transcription factor NF-E2–related factor 2 (Nrf2) (5, 14). Nrf2 regulates the transcription of cytoprotective genes, which promote cell protection toward oxidative damage (15) and enhance resistance to carcinogenesis (16). Chemically induced cancer formation is enhanced in Nrf2-deficient mice (17), supporting the protective role of SFN and Nrf2 activation in early carcinogenesis. Yet, the role of Nrf2 in cancer is more complex. In humans, enhanced Nrf2 expression is found in various types of epithelial cancer bearing mutations within the NRF2 gene (18) and also in oncogene-driven cancer tissue, where Nrf2 promotes antioxidant programs (19). SFN directly activates Nrf2 by binding cysteine residues of kelch-like ECH-associated protein 1 (Keap1), which is responsible for Nrf2 ubiquitination (20). SFN blocks the function of Keap1, resulting in Nrf2 stabilization. Therefore, SFN is considered to be an indirect antioxidant. In addition to the described cytoprotective actions of SFN preventing malignant transformation of cells, initial carcinogenesis is influenced by inflammatory signals (21, 22).
Although it was also shown that SFN may have anti-inflammatory effects, the anti-inflammatory mechanisms are less well understood. Considering SFN as a direct Nrf2 activator, this natural compound should affect immune responses in a similar way like Nrf2 itself. Alveolar macrophages expressing Nrf2 seem to be protective in different models of pulmonary inflammation (23). In contrast, Nrf2-deficient mice show decreased induction of antioxidative genes, increased Th2 responses, and enhanced airway inflammation in a mouse model of allergen-driven asthma (24). Similarly, enhanced oxidative stress and cartilage damage were observed in Nrf2-deficent mice during experimental arthritis (25). Enhanced pathology was observed in Nrf2-deficient mice in dextran sulfate sodium-induced colitis (26) and experimental autoimmune encephalomyelitis (EAE) (27). Yet, the exact effects of SFN on immune responses have been mainly studied in the setting of cancer or microbial infections. SFN enhances bacterial clearance by increasing the phagocytic activity of alveolar macrophages (28) and is beneficial against Helicobacter pylori infections (13). In contrast, high doses of SFN seem to decrease the expression of innate cytokines like IL-6, TNF, and IL-1 by macrophages in vitro in response to LPS (29). Similarly, SFN slightly impaired the induction of IL-6 and TNF in vivo in response to LPS (30).
Thus, SFN may not only prevent initial carcinogenesis by its cytoprotective actions but also by its effects on innate cytokines like TNF. Despite its antioxidative and antitumoral actions, little is known about the effects of SFN in regard to T cell–mediated autoimmunity. Molecular events by SFN in specialized immune cells have been widely neglected. To clarify the effects of SFN on inflammatory autoimmune responses, we studied the direct influence of low and nontoxic doses of SFN on T cells and professional APCs, namely, dendritic cells (DCs). In particular, we investigated the possible beneficial role of oral SFN administration in EAE, a prototypic T cell–mediated autoimmune disease, initiated by IL-23 that induces autoreactive Th17 cells and IL-12 that induces autoreactive Th1 cells. Based on these results, we asked whether any of the previously reported antitumoral and antioxidant mechanisms of SFN may interfere with Th17 and Th1 responses. Our first results show that orally administered SFN protects against actively induced EAE in mice. SFN limited Th17 and Th1 differentiation of autoreactive T cells in vivo. In vitro data unraveled that SFN did not directly influence T cell cytokine production but indirectly through the modulation of the DC phenotype. Global gene expression profiling of DCs revealed that SFN induces the expression of Nrf2-dependent antioxidative genes and simultaneously inhibits the expression of Il12b and Il23a. Treatment of DCs with SFN suppressed the protein production of IL-23 and IL-12 after LPS activation and was associated with strong heme oxygenase 1 (HO-1) induction. Studying the underlying molecular mechanisms revealed that nuclear SFN-induced HO-1 interacted with NF-κB p65 and inhibited its binding to the Il23a and the Il12b promoter regions.
Materials and Methods
Female SJL and C57BL/6 mice were purchased from Janvier or Charles River; transgenic OT-II mice were bred and housed in the animal care facilities of the University Medical Center Tübingen under specific pathogen-free conditions. Animal experiments were performed according to the institutional animal care guidelines and were approved by the Regierungspräsidium Tübingen (HT 1/08).
EAE was induced by s.c. immunization of female SJL mice with 37.5 μg proteolipid protein (PLP)139–151 peptide (EMC microcollections) in CFA (Difco) and i.p. injection of 200 ng pertussis toxin (Calbiochem). Mice were fed with food pellets containing 0.003% SFN in DMSO (Calbiochem) or DMSO (vehicle control) during the observation period, starting 10 d before immunization. The clinical EAE score was followed and rated by the following scale: 0 = no disease; 1 = limp tail; 2 = hind-limb weakness or partial paralysis; 3 = complete hind-limb paralysis; 4 = forelimb and hind-limb paralysis; 5 = moribund state.
Bone marrow–derived DCs were isolated and cultivated as described previously (31). DCs were harvested on day 7, stimulated with 0.1–0.5 μM SFN (Sigma-Aldrich) in DMSO or DMSO alone as control, and activated with 100 ng/ml LPS from Escherichia coli O111:B4 (Sigma-Aldrich) for the indicated time points. CD4+ T cells were isolated from spleen and lymph nodes of naive C57BL/6 mice by using CD4 microbeads and MACS purification system (Miltenyi Biotec). Isolated T cells were activated for 3 d with 10 μg/ml plate-bound anti-CD3ε and anti-CD28 Abs (eBioscience) with DMSO or 0.1 μM SFN in DMSO and expanded for 4 d with IL-2 (Chiron Therapeutics). For coculture experiments, DCs were first treated with 0.3 μM SFN in DMSO or DMSO for 1 h, left untreated or activated with 100 ng/ml LPS, and loaded with 100 ng/ml OVA (OVA323–339)-peptide (EMC microcollections) for another 1 h. Then purified CD4+ T cells isolated from OT-II mice were cultured with these DCs. Alternatively, SFN (0.3 μM) was added to the coculture on day 3 after initial activation of T cells by OVA323–339 peptide and LPS-activated DCs. On day 7, cytokine expression by T cells was determined after activation by intracellular staining and flow cytometry using an LSRII flow cytometer (BD Biosciences).
RNA isolation and real-time PCR
For cultured cells or ex vivo isolated cells, total RNA was purified and reverse transcribed into cDNA using commercially available kits (Finnzymes). Relative gene expression levels were determined by quantitative real-time PCR (qRT-PCR) using TaqMan probes (TIB MolBiol) for Il12a, Il12b, Il23a, Il4, Il6, Il17, Ifnγ, and Hmox1 and the LightCycler480 system (Roche). The relative expression of the indicated genes was calculated to the detected internal control (β-actin) using the2ΔΔCt method. For in vitro experiments, DMSO-treated, nonactivated cells were set as 1.0; for ex vivo experiments, values from individual mice were compared with the average DMSO-treated control pool, set as 1.0.
Cytokine analysis and flow cytometry
Commercially available ELISA kits were used for the quantification of the cytokines IL-12p70 (BD), IL-6 (BD, Biosciences), and IL-23p19 (eBioscience) from cell culture supernatants. Intracellular cytokine staining was performed after stimulating cells with PMA (Sigma-Aldrich) and ionomycin (Sigma-Aldrich) in the presence of monensin (BD Biosciences) for 4 h. Cells were fixed with 2% formaldehyde, permeabilized with saponin containing buffer, and stained with fluorochrome-labeled Abs directed against CD4 (Gk1.5; Biolegend) and IFN-γ (XMG1.2; eBioscience), IL-4 (11B11; eBioscience), IL-17A (TC11-18H10; BD Biosciences), IL-2 (JES6-5H4; eBioscience), or TNF-α (MP6-XT22; eBioscience).
For detecting apoptosis, DCs were treated with SFN in DMSO or DMSO alone (control) for 2 h in medium or activated with LPS (100 ng/ml) for 18 h before staining with Annexin VFITC and propidium iodide (BD Biosciences). The expression of cell-surface markers was determined by using fluorochrome-labeled anti-CD80 (1610A1; BD Biosciences), anti-CD86 (GL1; eBioscience), and anti-CD11c (HL3; BD Biosciences) Abs. Cells were analyzed by flow cytometry (FACSCalibur; BD Biosciences) and FCS Express software (De Novo Software).
Microarray data collection and analysis
A total of 100 ng total RNA was processed for Affymetrix Gene Chips using WT Expression Kit (Ambion). Fragmented and labeled cDNA was hybridized onto murine MoGene2.1 ST Gene Chips (Affymetrix). Hybridization, washing, staining, and scanning were performed automatically in a GeneTitan instrument (Affymetrix). Scanned images were subjected to visual inspection to control for hybridization artifacts and proper grid alignment, and analyzed with AGCC 3.0 (Affymetrix) to generate CEL files. All data analysis steps were performed on the software platform R 2.15.1 and Bioconductor 2.16.0 (32). First, the expression data from all chips were background corrected, quantile normalized, and summarized with robust multiarray analysis (33). A linear model, which captures the influence of SFN treatment on gene expression levels, was defined and coefficients were estimated. A nonspecific filter based on overall variance was applied to remove noninformative features before the fitting of the linear model was performed. Empirical Bayes shrinkage of the SEs was used to calculate the moderated F-statistic (34). The resulting p values were corrected for multiple testing with Benjamini-Hochberg (35). The lists of differentially regulated transcripts were analyzed for overrepresentation of gene ontology terms and KEGG pathways, respectively, using hypergeometric tests. Cluster analysis for selected probe sets was performed in R 2.15.1. Signal intensities were scaled and centered, and the distance between two expression profiles was calculated using Euclidian distance measure. Hierarchical cluster analysis was performed with average linkage for both genes and samples. Heat maps were generated with Bioconductor package gplots. The microarray data are available through the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE54980.
NF-κB reporter assay
RAW 264.7 macrophages were transfected with a pNFκB-TA-Luc reporter plasmid (Clontech) using Lipofectamine 2000 (Invitrogen). The reporter plasmid contained four tandem copies of the NF-κB consensus sequence, which are located upstream of the minimal TA promoter. Twenty-four hours posttransfection, cells were pretreated with 0.3 μM SFN in DMSO or DMSO alone for 1 h and activated with 100 ng/ml LPS. After 6 h, cells were lysed with passive lysis buffer (Promega) and luciferase activity was measured using a luminescence microplate reader (Berthold Technologies). Protein concentration of each sample was determined by a protein assay (Roth), and luciferase activity was calculated in relation to the protein content.
DCs were incubated with 0.3 μM SFN in DMSO or DMSO alone and activated with 100 ng/ml LPS for the indicated time points. Cells were washed in ice-cold PBS, lysed, and heated in SDS sample buffer (125 mM Tris-HCl [pH 6.8], 2% w/v SDS, 10% glycerol, 100 mM DTT, and 0.01% w/v bromphenol blue) at 95°C for 5 min. Equal amounts of samples were loaded onto a 12% PAA gel and transferred to a PVDF membrane or to a nitrocellulose membrane. After blocking with blocking buffer for 1 h, the membrane was incubated with anti–IκB-α (Cell Signaling Technology) or anti–HO-1 (Enzo Life Sciences) Abs overnight at 4°C, followed by incubation with either HRP-conjugated secondary Ab (Cell Signaling Technology) or IRDye800CW-conjugated secondary Ab (Licor) for 1 h. Proteins were detected by adding ECL Reagent (Cell Signaling Technology) and exposure to X-ray film or by OdysseySA Infrared Imaging System (Licor). The band size was quantified using the ImageJ software (Wayne Rasband, National Institutes of Health, Bethesda, MD) or the OdysseySA software version 1.1.
Chromatin immunoprecipitation (ChIP) was performed as described previously (36). In brief, DCs were treated with SFN in DMSO (0.3 μM) or DMSO alone for 1 h and activated with 100 ng/ml LPS for 1.5 h. Cells were cross-linked, harvested, and lysed by sonication. Subsequently, DC lysates were immunoprecipitated with anti–HO-1 (Enzo Life Sciences) or anti-acetyl histone H3 (Millipore) Abs. The precipitated and eluted DNA was analyzed by qRT-PCR. Primers were designed to amplify 150- to 250-bp fragments from selected genomic regions. The qRT-PCR was performed in duplicate on each sample and input DNA using LightCycler480 (Roche) and SYBR Green I Master Mix (Roche) according to manufacturer’s instructions. The product specificity was monitored by melting curve analysis. To account for differences in DNA quantity, for every genomic sequence studied, a ΔCt value was calculated for each sample by subtracting the Ct value of the chromatin-immunoprecipitated sample from the Ct value obtained for the input and no Ab ChIP, respectively. Calculating 2ΔΔCt yielded the relative amount of PCR product (relative enrichment). Binding sites for NF-κB p65 within the Il23a, Il12b, and Il6 promoter regions were identified with Matinspector (Genomatix). Primers were designed with Primer3 encompassing the NF-κB p65 site in the promoter regions of Il23a gene (sense: 5′-GGCTCTCCAAAGAGGGAGAT-3′; antisense: 5′-GCACCTCCTTTGGTTCTGAG-3′), the Il12b gene (sense: 5′-GTATCTCTGCCTCCTTCCTT-3′; antisense: 5′-CTGATGGAAACCCAAAGTAG-3′), the Il6 gene (sense: 5′-GACATGCTCAAGTGCTGAGTCAC-3′; antisense: 5′-AGATTGCACAATGTGACGTCG-3′), and the control (App gene) (sense: 5′-TTG CAG AAG ATG TGG GTT CA-3′; antisense: 5′-CTA CTT GGA AAC ATG GAG TC-3′).
DCs were treated with SFN in DMSO (0.3 μM) or DMSO alone for 1 h and activated with 100 ng/ml LPS for 3 h. Nuclei were extracted using nuclear and cytoplasmic extraction reagents (Thermo Scientific). Nuclear proteins were immunoprecipitated with anti–NF-κB p65 Ab or IgG (Abcam) overnight at 4°C. Immune complexes were collected using Dynabeads protein A and eluated in sample buffer. Protein concentration was determined via protein assay (Roth). A total of 20 μg of each sample was separated on SDS-12% PAGE, transferred to nitrocellulose membrane (Millipore), and subjected to HO-1 Western blot analysis.
T cell proliferation assay
CD4+ T cells from OT-II mice were cocultured with OVA-pulsed DCs for 3 d in 96-well plates. For determining proliferation of in vivo–activated T cells during EAE, CD4+ cells were isolated from draining lymph nodes of immunized mice and cocultured with PLP-pulsed DCs for 3 d. [3H]thymidine (0.25 μCi/well; PerkinElmer) was added for 16 h before harvest. Incorporation of radioactivity was determined by scintillation counting. Data are shown as means of triplicate cultures. The SD was <10% of the mean cpm.
Statistical analyses were performed by paired or unpaired t test or by ANOVA, followed by Dunnett’s multiple-comparison test using GraphPad Prism 5 software. The p values <0.05 were considered significant.
SFN protects from severe EAE by inhibiting Th17 responses
Previous reports have shown that SFN exhibits anti-inflammatory effects on macrophages (29, 37). In addition, i.p. administration of SFN ameliorates EAE in C57BL/6 mice actively immunized against MOG35–55 peptide, whereas SFN was less beneficial in a therapeutic setting (38). However, the natural route of SFN uptake is through the gastrointestinal system. We therefore investigated whether oral administration of this nutritional compound is able to specifically influence T cell–mediated autoimmune neuroinflammation in actively induced EAE. We fed SJL mice with SFN containing food pellets or a control diet starting 10 d before immunization for EAE. Control mice immunized against PLP139–151 peptide developed severe encephalomyelitis with clinical symptoms starting on day 10 after immunization (Fig. 1A). In contrast, SFN-fed mice developed either no or only mild symptoms of encephalomyelitis. Thus, in agreement with the beneficial effect of i.p. SFN injection on EAE, oral application of this nutritional compound also is able to ameliorate severe Th17/Th1-dependent EAE in vivo. To explain these results, we first analyzed the Th cell lineage–characterizing cytokines in draining lymph nodes of SFN-treated mice and control mice on day 7 after immunization. By intracellular cytokine staining, we found a significant suppression of IL-17 and IFN-γ production by CD4+ T cells in mice fed with SFN as compared with control mice (Fig. 1B, 1C). In contrast, the expression of IL-2, IL-4, IL-10, or TNF remained unaffected by SFN treatment. To get an independent mode of evaluation of our results, we restimulated equal numbers of isolated CD4+ T cells from draining lymph nodes of immunized mice in vitro with DC and PLP139–151 peptide. CD4+ T cells derived from immunized SFN-fed mice produced significantly less IL-17 and less IFN-γ in response to PLP139–151 peptide than CD4+ T cells from immunized control mice (Fig. 1D). These results show that SFN simultaneously impairs Th17 and Th1 responses in vivo and improves EAE.
SFN does not directly affect T cells
To determine whether SFN directly affects cytokine production by T cells, we next analyzed APC-redundant T cell differentiation and cytokine production in the absence or presence of SFN. T cells were activated by plate-bound anti-CD3 and anti-CD28 Abs in the presence or absence of SFN. After 3 d of stimulation, T cells were analyzed for the production of IFN-γ, IL-4, IL-17, IL-2, and TNF. We found no major differences in T cell cytokine production by SFN, indicating that the in vivo observation in EAE is not mediated by a direct action of SFN on T cell cytokine production (Fig. 2A).
Because SFN has been reported to affect cytokine production in macrophages, we wondered whether SFN influences T cell differentiation by acting on APCs. Therefore, we followed the cytokine production of T cells activated by DCs, the most potent APC. DMSO-treated DCs or SFN-treated DCs were loaded with OVA peptide and cocultured with OVA-specific TCR-transgenic (OT-II) T cells. On day 7 of the coculture, we found some expression of IFN-γ, IL-17, and IL-4, and clear production of IL-2 and TNF. SFN treatment of nonactivated DCs had no major impact on cytokine expression by T cells (Fig. 2B, 2C). Next, we activated OVA-loaded DCs with LPS before starting the coculture with OT-II T cells. As expected, LPS-activated DCs now induced strong production of IFN-γ and IL-17 in OVA-reactive T cells. In contrast, treatment of DCs with SFN and subsequent activation with LPS significantly suppressed the capacity of DCs to induce IFN-γ– and IL-17–producing OT-II T cells (Fig. 2B, 2C). Importantly, addition of SFN 3 d after DC activation by LPS had no significant impact on IFN-γ and IL-17 production by OT-II cells (Fig. 2B, 2C). Because these results indicated that SFN affects T cell responses by modulating the DC phenotype, we decided to further analyze the effects of SFN on DCs.
SFN does not affect DC maturation or T cell activation
First, we decided to exclude enhanced cell death in DCs treated with SFN. We could not find increased apoptosis in SFN-treated DCs as compared with control DCs using SFN concentrations up to 0.5 μM (Fig. 3A). Next, we characterized the maturation status of DCs treated with SFN as determined by the expression of the costimulatory molecules CD80 and CD86. As expected, LPS induced the upregulation of CD80 and CD86 on DCs. SFN treatment did not significantly affect the maturation status, neither in quiescent nor in LPS-stimulated DCs (Fig. 3B). Similarly, SFN did not influence MHC class II expression or CD83 expression (data not shown). Thus, the altered cytokine secretion by T cells primed by SFN-treated DCs could not be explained by impaired maturation of DCs. Likewise, T cell proliferation was not affected by SFN treatment of DCs, arguing against a functional impairment of the T cell–activating potential of these DCs (Fig. 3C). To explore the potential mechanisms by which SFN treatment of DCs inhibited Th1 and Th17 cell differentiation, we analyzed global gene expression in DCs treated with SFN and activated with LPS. We found that 328 genes showed a significant difference in SFN-treated versus control DCs (Fig. 3D). SFN induced the expression of Nrf2-regulated genes like Aox1, Nqo1, Prdx1, and Hmox1 (Fig. 3D, 3E). Interestingly, among the genes significantly suppressed in SFN-treated DCs were Il23a, Il12b, and Il12a, the key cytokines for the induction of Th17 and Th1 immune responses in EAE (39–45). The expression of other IL-17–inducing cytokines like Il6 or Il1b was not significantly affected by SFN treatment (Fig. 3E).
SFN impairs IL-12 and IL-23 production by DCs
To confirm our findings, we treated DCs with SFN or the vehicle DMSO before LPS activation and quantified early mRNA expression of Il12a, Il12b, Il23a, and Il6. In control DCs, LPS induced the expression of Il12a, Il12b, Il23a, and Il6. Treatment with SFN impaired LPS-induced Il12b and Il23a expression, and showed some inhibition on Il12a expression, whereas Il6 expression by DCs was not affected by SFN (Fig. 4A). We then studied the protein levels of the indicated cytokines after 18 h of LPS activation. Again, stimulation of control DCs with LPS induced clear production of IL-12p70, IL-23, and IL-6 protein. SFN treatment of DCs showed a dose-dependent inhibition of IL-12p70 secretion and a profound suppression of IL-23 production even at very low concentrations (Fig. 4B). Thus, SFN seems to be a potent suppressor of IL-23 expression by LPS-activated DCs and also inhibited IL-12 production. IL-6 secretion by DCs was not affected by SFN in vitro.
SFN affects DNA binding of NF-κB
Because SFN has been reported to influence the phenotype of LPS-activated macrophages (29, 37), we next studied the effect of SFN on LPS signaling in DCs. Altered NF-κB activation in DCs could explain the profound differences in IL-23 and IL-12 production by SFN. To directly study the role of SFN on NF-κB activation, we transfected RAW 246.7 macrophages with a NF-κB luciferase reporter plasmid.
SFN treatment significantly inhibited LPS-mediated NF-κB activation in macrophages (Fig. 5A). At steady-state, RAW 246.7 cells showed low reporter activity. This basal reporter activity in nonstimulated macrophages was still suppressed by SFN. LPS activation induced a strong NF-κB reporter activity in macrophages. SFN fully prevented NF-κB reporter activity in LPS-stimulated macrophages (Fig. 5A). Reduced NF-κB activation, as observed in the SFN-treated condition, could be mediated through effects interrupting TLR4 signaling upstream of IκB-α or through effects downstream of IκB-α. To characterize the localization of SFN interaction with the NF-κB signaling pathway, we first analyzed IκB-α degradation. As expected, TLR activation of control DCs resulted in IκB-α degradation. However, we could not find significant differences in LPS-mediated IκB-α degradation between cells treated with SFN or DMSO-treated control cells (Fig. 5B). Thus, a nuclear SFN-mediated interaction with NF-κB downstream of IκB-α should be relevant for the regulation of IL-12 and IL-23 production. To determine this aspect in detail, we analyzed the binding of NF-κB to the Il23a and Il12b promoter regions by ChIP.
As reported previously (46–48), we found binding of the NF-κB family member p65 to the Il23a and Il12b promoter in DCs activated with LPS. This binding was associated with positive transcription and increased histone 3 acetylation (H3Ac) within the Il23a and Il12b promoter regions. In contrast, when DCs were treated with SFN, binding of p65 to the Il23a and the Il12b promoter regions was severely inhibited upon LPS activation (Fig. 5C). Moreover, the Il23a and Il12b promoter regions were less accessible in SFN-treated and LPS-activated DCs due to decreased H3Ac. As control, we studied p65 binding to and histone H3 acetylation of the Il6 promoter. In this study, SFN treatment resulted in increased p65 binding (p = 0.0794) but did not reach significant changes in H3Ac or Il6 expression in vitro (Figs. 3D, 3E, 4A, 4B, 5C). Thus, SFN suppressed IL-23 and IL-12 expression in DCs by inhibiting NF-κB binding and H3Ac of p65 to the Il23a and Il12b promoter sites.
SFN-induced HO-1 interacts with NF-κB p65
Because SFN is unlikely to act as a direct regulator of transcriptional processes, we next addressed the exact mechanism by which SFN mediates decreased p65 binding to the Il23a and Il12b promoter regions. The direct interaction of SFN with the Nrf2/ARE pathway (20) suggested that cellular stress might be responsible for our observation of reduced p65 binding to the Il23a promoter. In DCs, SFN induced the upregulation of typical Nrf2 target genes (Fig. 3E). Among these target genes, we found profound induction of Hmox1, a gene previously linked to the regulation of IL-12 and IL-23 (36). Recently, we demonstrated that N-terminal HO-1, an inducible, Nrf2-dependent stress response protein, is capable to translocate into the nucleus, where it interacts with p65. Interestingly, HO-1 can directly impair p65 binding to the Il23a promoter (36). To decipher the possible involvement of HO-1 in SFN-mediated suppression of IL-23 and IL-12 in DCs, we first investigated the regulation of HO-1 in DCs by SFN. Quantitative analysis showed that SFN induced Hmox1 mRNA expression in DCs, especially after LPS activation (Fig. 6A). We confirmed the transcriptional induction of Hmox1 by SFN by Western blot analysis of HO-1 protein expression (Fig. 6B). In agreement with the reported nuclear translocation of HO-1 (36, 49), we detected HO-1 in nuclear extracts of SFN-treated DCs (Fig. 6C). Because nuclear HO-1 has been reported to inhibit NF-κB function by protein–protein interaction, we coimmunoprecipitated lysates from SFN-treated or control DCs with NF-κB p65 Abs and blotted for HO-1 after protein gel electrophoresis. Although we detected a strong HO-1 signal in p65-precipitated nuclear extracts from SFN-treated DCs after LPS activation, no HO-1 binding was found in DCs after LPS stimulation in the absence of SFN (Fig. 6D). Thus, our data reveal a direct interaction of HO-1 with NF-κB, linking the activation of Nrf2 pathways by SFN to the SFN-dependent inhibition of IL-23 and IL-12 expression in activated DCs.
SFN treatment inhibits Il23a/Il12b expression and Th17/Th1 development in vivo
To test in vivo the biological relevance of our in vitro findings demonstrating that SFN negatively regulates IL-23 and IL-12, we immunized control mice or SFN-fed mice for EAE and analyzed the expression of Il23a, Il12a, Il12b, and Il6 in draining lymph nodes. Similar as in vitro, the suppression of Il23a and Il12b expression in lymph nodes from SFN-treated mice compared with lymph nodes from control mice was most prominent (Fig. 7A; Il23a: p = 0.0048; Il12b: p = 0.0368). SFN treatment also impaired Il12a and Il6 expression, but these effects were less pronounced and not significant (Fig. 7A; Il12a: p = 0.2394; Il6: p = 0.0991). Hence we conclude that SFN predominantly suppresses Il23a and Il12b expression by DCs in vitro and in vivo, and impairs Th17 responses more efficiently than Th1 responses, resulting in the strong protection from IL-23–mediated neuroinflammation. T cell proliferation of in vivo primed T cells in response to PLP139–151 peptide was not affected by SFN treatment (Fig. 7B). These results confirmed our in vitro findings that SFN treatment of DCs does not affect the activation of an Ag-specific T cell response (Fig. 3C). Likewise, we did not find a significant suppression of the number of mononuclear cells within the CNS in immunized mice that received oral SFN treatment (Fig. 7C). Instead, analysis of the cytokine profile of CNS-infiltrating T cells demonstrated a significant suppression of IL-17 and IFN-γ production by SFN treatment, whereas IL-2 and TNF production was not affected (Fig. 7D). Of note, we did not find differences in IL-6 expression in CNS-infiltrating mononuclear cells and no induction of Foxp3 expression. Thus, the induction of HO-1 by SFN and its influence on p65 activity at the Il23a and Il12b promoter sites seem to mediate the immunomodulatory effect of this natural compound.
SFN, which is found at high levels in broccoli, is currently held as one of the most promising natural compounds for clinical implementation (9, 50–52). The anticarcinogenic effects of high doses of SFN are well characterized (5), and first clinical trials investigating the effects of oral SFN in patients with prostate cancer have been conducted. Antioxidative and antimicrobial properties of SFN, like the induction of phase II detoxifying enzymes and the inhibition of H. pylori growth, seem to contribute to its anticarcinogenic properties (6). More recent reports unraveled beneficial effects of SFN in experimental chronic obstructive pulmonary disease (COPD) (28, 53). SFN reduced pulmonary inflammation in COPD, a condition associated with Th17 responses (54, 55). The aim of our study was to elucidate the immunoregulatory capabilities of SFN in terms of affected immune cells and underlying molecular mechanisms during T cell–mediated autoimmune responses.
We show that feeding mice with SFN ameliorates IL-23–mediated encephalomyelitis. Using the MOG model of EAE, another group recently demonstrated that i.p. application of SNF is also effective in protecting from severe EAE (38). EAE is a prototypic T cell–mediated autoimmune disease initiated by IL-17–producing Th17 cells and maintained by Th17 and Th1 cells (56). Our results demonstrate that oral SFN treatment protects mice from severe EAE by silencing Th17 and Th1 immune responses in vivo. First, our in vitro data revealed that SFN influenced DC-dependent T cell differentiation, whereas DC-independent T cell polarization remained unaffected by this phytochemical. SFN impaired the expression of IL-12 and Th1 responses in vitro and in vivo. However, Ag-specific T cell proliferation was not affected by oral SFN. In agreement with others, we did not find a significant induction of Th2 responses or regulatory T cells by SFN (38). One major observation of this study was the remarkable suppression of IL-23 production by SFN in vitro and in vivo. IL-23 is a key mediator in various autoimmune diseases and is crucially involved in EAE pathology, as shown in mice lacking the p40 or p19 unit of IL-23 (57). IL-23 is crucial for generating Th17 cells that do not produce IL-10 and instead develop an encephalitogenic phenotype (41, 58). DCs are main producers of IL-23, and pharmacological inhibition of IL-23 production or its neutralization is not only effective in experimental mouse models of EAE, but also in humans with Th17-dominated autoimmune diseases like psoriasis (36, 59–61). The mechanism underlying SFN-mediated IL-23 suppression through the direct Nrf2 activator SFN was one central question of our study. TLR4 activation by LPS activates MyD88 and downstream NF-κB, resulting in the transcription of inflammatory cytokines like IL-6, IL-12, and IL-23. The presence of IL-6 and IL-23 during priming of myelin-specific Th17 cells is essential for triggering EAE (40, 41, 62, 63). Because TLR/MyD88 signaling activates NF-κB, resulting in IL-23 transcription and subsequently Th17 specification (47), it was likely that SFN controls IL-23 production by altering the NF-κB pathway. According to our assumption, we found suppressed NF-κB reporter activity by SFN without affecting IκB-α degradation. Instead, SFN treatment reduced p65 binding to the Il23a and Il12b promoter, and histone H3 acetylation of both loci. Unlike Il23a and Il12b, we found stronger binding of p65 to the Il6 promoter in SFN-treated DCs, which was not associated with altered histone H3 acetylation or IL-6 expression in vitro. In vivo, we observed a trend toward decreased IL-6 expression by oral SFN in immunized mice. This is in agreement with previous reports on the regulation of IL-6 in LPS-injected Nrf2−/− mice (30) or in LPS-injected mice treated with indirect Nrf2 activators inducing ATF3, a factor negatively regulating IL-6 expression in macrophages (64). Most strikingly, SFN seems to selectively modulate the expression of IL-23 and IL-12 in DCs by a mechanism influencing the DNA binding of NF-κB to the promoter regions of both IL-23 subunits, IL23a and IL12b. In agreement with our findings, SFN has been reported to impair TLR4 signaling and to reduce DNA binding of NF-κB without blocking nuclear translocation of the transcription factor (37, 65). Based on the observation that SFN affects NF-κB–dependent gene transcription, several mechanisms have been suggested. In a human prostate cancer cell line, the NF-κB–inhibitory effect of high-dose SFN (20–30 μM) was associated with the inhibition of IκB kinase β and IκB-α phosphorylation (66). In contrast, Heiss et al. (37) proposed that the inactivation of NF-κB by SFN is likely to be mediated through the interaction with cellular redox regulators like glutathione or thioredoxin. The SFN-mediated events on cellular systems regulating oxidative stress are well documented, and the activation of several cytoprotective genes by SFN is mediated through the Nrf2/ARE signaling pathway (67, 68). Some of the beneficial effects of Nrf2 activators in EAE may also be mediated by cytoprotective effects within the CNS (38, 69) Li et al. (38) demonstrated the upregulation of antioxidative genes in the CNS by SFN, and we found the upregulation of the same genes in DCs. The question arises whether the simultaneous anti-inflammatory and antioxidative response by Nrf2 activators is an epiphenomenon or has direct association. The impaired secretion of inflammatory mediators by SFN has been linked to Nrf2 in LPS-activated macrophages previously (29). Of note, these effects required 20- to 40-fold higher doses in vitro as the effect of SFN on IL-23 expression as described in our study. Recently, we demonstrated that the induction of HO-1, a stress response protein regulated by Nrf2, affects IL-23 production in DCs (36, 49). Nuclear HO-1 influences the activity of NF-κB (36, 49). Overexpression of HO-1 or pharmacological induction of HO-1 by DMF results in a direct inhibition of NF-κB p65 promoter activity within the IL-23 promoter (36). HO-1 is described to be protective in EAE (70) and experimental colitis (71). HO-1–inducing DMF is established for the therapy of Th17/Th1-mediated psoriasis, and recent reports showed efficacy in multiple sclerosis in experimental mice, as well as in humans (36, 72, 73). Similar to DMF, SFN is a potent Nrf2 activator inducing HO-1 in vitro and in vivo (28–30). Our present data demonstrate that HO-1 is readily induced in DCs by SFN treatment, and that nuclear HO-1 fulfills a protein–protein interaction with NF-κB p65 resulting in reduced binding of p65 to the Il23a and Il12b promoter sites. Recently, SFN has been reported to induce HO-1 and other Nrf2-dependent genes in vivo in the setting of EAE (38). The induction of HO-1 and the profound suppression of IL-23 and IL-12 in vitro and in vivo directly link the antioxidative properties of SFN to its anti-inflammatory and protective actions in autoimmunity. Thus, this phytochemical is becoming more and more a promising Nrf2-activating compound (74). Based on our new findings and the tremendous work by other groups, we suggest extending the preclinical and clinical evaluation of oral SFN, which is currently under investigation in the setting of cancer or COPD, to IL-23/IL-12–dependent autoimmune diseases.
This work was supported by the Bundesministerium für Bildung und Forschung (Grant 0315079 to K.G.) and Deutsche Forschungsgemeinschaft Sonderforschungsbereich 685 (to K.G.).
The sequences presented in this article have been submitted to the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE54980.
Abbreviations used in this article:
antioxidant response element
chronic obstructive pulmonary disease
experimental autoimmune encephalomyelitis
histone 3 acetylation
heme oxygenase 1
NF-E2–related factor 2
quantitative real-time PCR
The authors have no financial conflicts of interest.