Sulforaphane Suppresses Oligomerization of TLR4 in a Thiol-Dependent Manner

A naturally occurring isothiocyanate derivative, 1-isothiocyanato-4-(methylsulfinyl)-butane (sulforaphane or SFN), is abundant in cruciferous vegetables such as broccoli and cauliflower. SFN has been well known for its chemopreventive activity, which is mediated through the inhibition of phase I enzymes that produce carcinogens and the induction of phase II enzymes that are involved in detoxification of carcinogens (1). In addition, SFN exerts anti-inflammatory effects. SFN reduced the expression of inflammatory proteins such as inducible NO synthase (iNOS), cyclooxygenase-2 (COX-2), and TNF-α in LPS-treated macrophages or neuronal cells (2, 3). SFN inhibited the LPS-induced secretion of high-mobility group box 1, which functions as a cytokine-like molecule and is actively secreted by immunostimulation, in RAW264.7 cells (4). SFN suppressed the expression of the adhesion molecule, intercellular adhesion molecule-1, in LPS-treated endothelial cells (ECs), resulting in the reduction of monocyte adhesion to ECs (5). The diet of dried broccoli, which is a rich source of SFN decreased infiltration of macrophages to blood vessel walls, hearts, and kidneys of spontaneous hypertensive rats (6). Because inflammation is one of the key etiological factors to carcinogenesis, anti-inflammatory activity of SFN may also contribute to its anticancer effects.

Inflammation can be initiated by microbial infection or tissue injury. TLRs are pathogen recognition receptors that detect structural components derived from invading microorganisms or damaged cells/tissues. The activation of TLRs triggers immune and inflammatory responses through the activation of cellular signal transduction and the expression of immune and inflammatory mediators. TLR4 recognizes LPS, a cell wall component of Gram-negative bacteria, which causes systemic inflammatory sepsis. TLR4 can be also activated by endogenous molecules such as heparan sulfate and fibronectin, which are released from an injured tissue site, resulting in aseptic inflammatory responses. The activation of TLR4 by LPS results in oligomerization of TLR4 and subsequently the recruitment of adaptor molecules. In general, TLR4 signaling is dependent on MyD88 and Toll/IL-1R domain containing adaptor-inducing IFN-β (TRIF), the major adaptor molecules interacting with cytoplasmic region of TLR. MyD88 recruits IL-1R–associated kinase (IRAK)-4 and IRAK-1, which further associates with TNFR-associated factor (TRAF)6, leading to the activation of IκB kinase (IKK) α/β/γ complex and MAPKs. TRIF is responsible for the activation of MyD88-independent signaling pathways. TRIF activates receptor interacting protein 1, TANK-binding kinase 1 (TBK1), and IKKε through the interaction with TRAF6 or TRAF3 (7–11).

The activation of TLR4 culminates in the production of proinflammatory proteins including iNOS, COX-2, and TNF-α. Deregulated TLR activity is known to be closely linked to the development of chronic inflammatory diseases as well as immunological disorders (12). TLRs and the downstream signaling components can be good therapeutic targets for many inflammatory diseases (13, 14). Therefore, we examined whether SFN, a well-known anti-inflammatory agent, modulates the activation of TLR4 and downstream signaling pathways and what is the anti-inflammatory target in TLR4 signaling regulated by SFN.

Animal care and the study protocols were carried out in accordance with the guidelines of the Institutional Animal Care and Use Committee of Amorepacific R&D Center or Gwangju Institute of Science and Technology. Mice were purchased from Orient Bio (Seoul, South Korea) and acclimated under specific pathogen-free conditions in an animal facility for at least 1 wk before use. The mice were housed in a temperature (23 ± 3°C) and relative humidity (40–60%)-controlled room. Lighting was adjusted automatically at a cycle of 12 h of light and 12 h of dark.

SFN was purchased from Sigma-Aldrich (St. Louis, MO). Purified LPS was obtained from List Biological Laboratory (Campbell, CA) and dissolved in endotoxin-free water. Ab for IRAK-1 was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Ab for COX-2 was from Cayman Chemical (Ann Arbor, MI). All other reagents were purchased from Sigma-Aldrich unless otherwise described.

RAW264.7 cells (a murine monocytic cell line, ATCC TIB-71) and 293T (human embryonic kidney cells) were cultured in DMEM containing 10% (v/v) heat-inactivated FBS (HyClone, Logan, UT), 100 U/ml penicillin, and 100 μg/ml streptomycin. Ba/F3 cells, an IL-3–dependent murine pro-B cell line, expressing TLR4 (Flag or GFP-tagged), cluster of differentiation (CD)14, lymphocyte Ag 96 (MD-2) (Flag-tagged), and NF-κB luciferase reporter gene were cultured as described previously (15). Embryonic fibroblast cells from 13.5-d embryos of mice with wild-type (WT) and nrf2-disrupted genotypes (16) were maintained in IMDM (Invitrogen, Carlsbad, CA) containing 10% FBS and antibiotics (17). Cells were maintained at 37°C in a 5% CO₂/air environment.

An NF-κB(2×)-luciferase reporter construct was provided by Frank Mercurio (Signal Pharmaceuticals, San Diego, CA). An IFN-β–positive regulatory domains III–I (PRDIII–I)-luciferase reporter plasmid and a WT TBK1 expression plasmid were gifts from K. Fitzgerald (University of Massachusetts Medical School, Worcester, MA). Heat shock protein 70-β–galactosidase reporter plasmid was from R. Modlin (University of California, Los Angeles, CA). A constitutively active chimeric CD4-TLR4 was obtained from C. A. Janeway, Jr. (Yale University, New Haven, CT). A WT TLR4 expression plasmid was obtained from A. Hajjar (University of Washington, Seattle, WA). A WT of MyD88 was provided by J. Tschopp (University of Lausanne, Lausanne, Switzerland). A TRIF expression plasmid was provided by S. Akira (Osaka University, Osaka, Japan). A WT IKKβ was obtained from M. Karin (University of California, San Diego, CA). All DNA constructs were prepared in large scale using EndoFree Plasmid Maxi kit (Qiagen, Chatsworth, CA) for transfection.

RAW264.7 or 293T cells were transfected with a luciferase plasmid and various expression plasmids of signaling components using SuperFect transfection reagent (Qiagen, Valencia, CA), according to the manufacturer’s instructions. Heat shock protein 70-β–galactosidase plasmid was cotransfected as an internal control. The total amount of transfected plasmids was equalized by supplementing with the corresponding empty vector. Luciferase and β-galactosidase enzyme activities were determined using the Luciferase Assay System and β-galactosidase Enzyme System (Promega, Madison, WI), according to the manufacturer’s instructions. Luciferase activity was normalized by β-galactosidase activity to determine relative luciferase activity.

Equal amounts of cell extracts were resolved on SDS-PAGE and electrotransferred to polyvinylidene difluoride membrane. The membranes were blotted with the primary Ab and the secondary Ab conjugated to HRP (Amersham Biosciences, Arlington Heights, IL). The reactive bands were visualized with the enhanced chemiluminescence system (Amersham Biosciences).

Protein extracts were prepared from Ba/F3 cells stably expressing Flag-TLR4, GFP-TLR4, CD14, and MD-2 and from 293T cells transfected with hemagglutinin (HA)-ΔTLR4 and Flag-ΔTLR4, which are truncated form of TLR4 with constitutive activity (18). Supernatants were incubated with monoclonal anti-Flag Ab (Sigma-Aldrich) for Ba/F3 cell lysates and anti-HA (Roche, Mannheim, Germany) for 293T cell lysates for 4 h and further incubated with 70 μl of 50% (v/v) protein A-agarose (Amersham Biosciences) for overnight at 4°C with rocking. Immune complexes were solubilized with Laemmli sample buffer after four times of washing with lysis buffer. The solubilized immune complex was resolved on SDS-PAGE. The membranes were immunoblotted with anti-Flag Ab and rabbit anti-GFP Ab (Molecular Probes, Eugene, OR) for Ba/F3 cell lysates and anti-Flag Ab and anti-HA Ab for 293T cell lysates.

Purified extracellular domain of human TLR4 (aa 27–631) or mouse TLR4 (aa 26–629) was obtained from J.-O. Lee (Korea Advanced Institute of Science and Technology, Daejeon, Korea) (19, 20), and 2 μg of each protein was incubated with SFN (100 μM) for 1 h at 37°C prior to in-solution digestion and micro-liquid chromatography-tandem mass spectrometry (LC–MS/MS) analysis. In-solution digestion was carried out using chymotrypsin and trypsin to generate peptides with various cleavage sites. SFN-modified extracellular domains of TLR4 were digested in a digestion buffer [100 mM Tris-HCl (pH 7.8) and 10 mM CaCl₂] with chymotrypsin (1:50 enzyme to substrate ratio) at 37°C for 12 h. Aliquots of chymotrypsin-treated samples were further digested with trypsin (1:50 enzyme to substrate ratio) at 37°C for 12 h after thermal denaturation of chymotrypsin. Protein digestions were finally quenched by adding 90% formic acid (final concentration of 5%). Either chymotrypsin-treated or chymotrypsin/trypsin-treated samples were subjected to micro-LC–MS/MS analysis. MS/MS spectra were searched using TurboSequest against the homemade protein database containing human and mouse TLR4 sequences. Bioworks version 3.1 was used to filter the search results, and the following Xcorr values and filtering criteria were applied to different charge states of peptides: 1.8 for singly charged peptides, 2.5 for doubly charged peptides, 3.5 for triply charged peptides, and all charge states peptides with ΔCn of 0.08.

Female C57BL/6 mice were orally administered with vehicle (PBS) or SFN (25 mg/kg) at 24 and 1 h before i.p. injection of LPS (7.5 mg/kg). Blood was collected at 2 h after LPS injection by retro-orbital bleeding and allowed to clot at room temperature. Serum was separated by centrifugation and stored at −80°C until analysis. Concentrations of cytokines TNF-α, IL-1β, and IL-6 were determined with the Quantikine ELISA kit specific for each cytokine, according to the manufacturer’s instruction (R&D Systems, Minneapolis, MN).

Male BALB/c mice were orally administered with vehicle (3% DMSO in PBS), SFN (10 and 30 mg/kg), or indomethacin (3 mg/kg) at 24 and 1 h before and 6 h after intradermal injection of LPS (50 μg/25 μl in PBS) on the center of right ear. Twenty-four hours after LPS injection, 6-mm biopsies of both ears were collected and weighed for the evaluation of LPS-induced inflammation and edema. Untreated left ear was used for the normalization of individual variation and the ratio of right over left ear was presented as ear swelling values. After weighing, biopsy tissues were processed and stained with H&E for histological examination by a histologist who was blind to the treatment group.

Immunoblot data were presented with representative figures from more than three independent experiments. Other data were expressed as mean ± SEM. Difference between groups was examined by Student t test (significant when p < 0.05).

To investigate whether SFN modulates TLR4 activation, macrophages (RAW264.7) were stimulated with LPS, a representative TLR4 agonist, in the presence of SFN, and the activation of downstream signaling molecules and the expression of an inflammatory gene were determined.

The activation of TLR4 by ligands leads to the recruitment of MyD88, which forms a complex with IRAK-4 and IRAK-1, resulting in phosphorylation and subsequent degradation of IRAK-1. Therefore, IRAK-1 degradation is one of the readouts for early events of TLR4 activation. SFN blocked IRAK-1 degradation induced by LPS in macrophages (Fig. 1A). SFN also suppressed LPS-induced activation of transcription factors, NF-κB, and IFN regulatory factor 3 (IRF3) as determined by reporter assay using a luciferase reporter gene containing NF-κB binding site or IRF3 binding site (IFN-β PRDIII–I) (Fig. 1B, 1C). In addition, LPS-induced COX-2 expression was decreased by SFN (Fig. 1D). These results showed that SFN inhibited ligand-induced activation of TLR4.

The replacement of extracellular domain of TLR4 with CD4 resulted in the constitutive activity to activate NF-κB without ligand (21), and overexpression of WT TLR4 bypasses the requirement of ligand to activate intracellular signaling. Therefore, we determined whether SFN can inhibit ligand-independent activation of TLR4 by examining the activation of TLR4 signaling stimulated with CD4-TLR4 or TLR4(WT) in the presence of SFN. SFN inhibited NF-κB activation induced by overexpression of constitutively active TLR4 (CD4-TLR4) or TLR4(WT) in 293T cells (Fig. 2A, 2B). IRF3 activation induced by CD4-TLR4 was also reduced by SFN in RAW264.7 cells (Fig. 2C).

These results showed that SFN was able to inhibit both ligand-dependent and -independent activation of TLR4. These further suggest that the inhibitory target of SFN may not be the interaction between ligand and receptor.

To identify an anti-inflammatory target of SFN in TLR4 signaling pathways, we investigated which downstream signaling molecule is affected by SFN. TLR4 signaling pathways are generally composed of MyD88-dependent and TRIF-dependent pathways. The MyD88 pathway of TLR4 leads to the activation of NF-κB through the activation of IKKβ complex, whereas activation of TRIF-pathway of TLR4 culminates in the activation of IRF3 through the activation of TBK1/IKKε. To determine whether SFN affects the activity of MyD88- or TRIF-dependent signaling components, 293T cells were transfected with each signaling component with NF-κB reporter gene for MyD88 pathway and IRF3 reporter gene for TRIF pathway. SFN blocked NF-κB activation induced by the overexpression of MyD88 or IKKβ in 293T cells (Fig. 3A, 3B), suggesting that the target of SFN exists in MyD88 pathway. It has been reported that SFN can directly inhibit the kinase activity of IKKβ and the DNA-binding activity of NF-κB (2). Therefore, these may account for the inhibitory effect of SFN on MyD88- or IKKβ-induced NF-κB activation.

In contrast, SFN failed to suppress IRF3 activation induced by overexpression of TRIF (Fig. 3C). Furthermore, SFN did not inhibit IRF3 activation induced by TBK1, a kinase downstream of TRIF (Fig. 3D). These suggest that the target of SFN is neither TRIF itself nor the signaling components downstream of TRIF. Although SFN did not inhibit TRIF- and TBK1-induced IRF3 activation, SFN was able to suppress TLR4 ligand (LPS)- and TLR4 itself (CD4-TLR4)-induced IRF3 activation (Figs. 1C, 2C). Therefore, these suggest that SFN may have other inhibitory targets present at upstream of TRIF, possibly at the receptor level, in addition to IKKβ and NF-κB.

Receptor oligomerization is one of the initial steps triggering the activation of TLR4 and intracellular signaling (15, 22). Oligomerization of TLR4 provides the cytosolic platform for the association of downstream signaling molecules to transmit extracellular stimuli to intracellular signaling events. Therefore, we investigated whether SFN affected oligomerization of TLR4 by immunoprecipitation study using Ba/F3 cells stably expressing both Flag-TLR4 and GFP-TLR4. GFP-TLR4 was coprecipitated with Flag-TLR4 upon LPS stimulation, indicating the association of Flag-TLR4 and GFP-TLR4 to form oligomers. SFN attenuated LPS-induced association of Flag-TLR4 and GFP-TLR4, demonstrating that SFN inhibited ligand-induced oligomerization of TLR4 (Fig. 4A). The inhibitory effect of SFN on NF-κB activation was also observed in the same Ba/F3 cells (Fig. 4B).

We further determined whether SFN inhibited ligand-independent oligomerization of TLR4 by immunoprecipitation study using HA-tagged or Flag-tagged truncated TLR4 (ΔTLR4), which have constitutive activity and can form oligomers without ligand stimulation (18). SFN prevented TLR4 oligomerization and NF-κB activation induced by ΔTLR4 (Fig. 4C, 4D). These results showed that SFN inhibited both ligand-dependent and -independent oligomerization of TLR4. Furthermore, the results suggest that the attenuation of NF-κB by SFN results from the inhibitory effect on receptor oligomerization in addition to the direct inhibition of NF-κB.

Because SFN is well-known to react with sulfhydryl moiety in proteins (23), we investigated whether SFN could react with cysteines in TLR4 protein. SFN was incubated with purified extracellular domain of human TLR4 (aa 27–631) or mouse TLR4 (aa 26–629), and the formation of adducts was examined by micro-LC–MS/MS analysis. SFN formed adducts with cysteines 88, 192, 246, 281, 340, 506, 542, and 609 of human TLR4 and cysteines 87, 164, 245, 280, 504, 549, and 606 of mouse TLR4 among 16 cysteines present in each extracellular domain (Table I and Supplemental Fig. 1). Among them, cysteines 88, 192, 246, 340, 506, and 542 of human TLR4 and cysteines 87, 164, 245, 504, and 549 of mouse TLR4 are free cysteines and others participate in disulfide bonds. The results demonstrated that SFN was able to bind cysteine residues in extracellular domain of TLR4. We next determined whether the inhibition of TLR4 oligomerization by SFN was dependent on its reactivity to sulfhydryl group. The treatment with thiol donors, DTT or N-acetyl-l-cysteine (NAC), blocked the suppressive effects of SFN on TLR4 oligomerization and NF-κB activation (Fig. 5). These results suggest that the inhibitory effect of SFN on TLR4 activation is attributed to its reactivity to sulfhydryl moiety. Consistently, another thiol-reactive compound, parthenolide (5 and 10 μM) was able to significantly reduce LPS-induced TLR4 oligomerization and NF-κB activation (Fig. 6A, 6B). Ethacrynic acid slightly attenuated LPS-induced TLR4 oligomerization at relatively high concentrations, such as 50 and 100 μM, whereas NF-κB activation was significantly reduced at these concentrations (Fig. 6C, 6D).

SFN is well-known to activate keap-1/Nrf2 pathway, which induces antioxidants and phase II detoxifying enzymes. To investigate whether Nrf2 is involved in inhibitory effects of SFN on TLR4-dependent signaling, we determined the effects of SFN on TLR4-dependent signaling in mouse embryonic fibroblasts (MEFs) derived from WT or Nrf2-knockout (KO) mice. The inhibitory effects of SFN on LPS-induced NF-κB activation and Akt phosphorylation were still observed in Nrf2-KO MEFs as well as WT MEFs to a similar extent (Fig. 7A, 7B). In addition, SFN inhibited constitutively active TLR4-induced NF-κB activation in both WT and Nrf2 KO MEFs to a similar degree (Fig. 7C). These results showed that the KO of Nrf2 did not abolish the inhibitory effects of SFN on TLR4-dependent signaling.

Taken together, our results demonstrated that SFN suppressed oligomerization of TLR4 in a thiol-dependent manner, which contributes to the decrease in activation of downstream signaling pathways and target gene expression.

To investigate whether SFN exerted the similar inhibitory effects on LPS signaling in vivo, mice were challenged with LPS injection with or without SFN treatment, and the serum levels of cytokines were determined. LPS challenge to mice significantly increased serum levels of inflammatory cytokines, such as TNF-α, IL-1β, and IL-6, whereas oral administration of SFN greatly attenuated the increased production of cytokines (Fig. 8). To further confirm the in vivo inhibitory effects of SFN on LPS signaling, we used LPS-induced skin inflammation animal model. Skin inflammation was induced by intradermal injection of LPS on the center of mouse right ear, and 24 h later, ears were collected and examined for the evaluation of LPS-induced inflammation and edema. LPS injection induced ear swelling as shown by increased ear weight. However, oral administration of SFN reduced ear swelling and weight (Fig. 9A). Histological analysis indicated that ears injected with LPS exhibited moderate dermal inflammation and severe edema, whereas ears from the SFN treatment group showed minimal to slight dermal inflammation and minimal edema (Fig. 9B). The protective effects of SFN against LPS-induced ear skin inflammation were almost as potent as those of indomethacin, a well-known nonsteroidal anti-inflammatory drug, which was used as a positive control. Collectively, blockade of TLR4 signaling by SFN resulted in the reduced production of inflammatory cytokines and the decreased dermal inflammation and edema in vivo in experimental inflammatory animal models.

It has been shown that SFN has multiple cellular targets to exert its chemopreventive or anti-inflammatory activities. Anti-inflammatory activity of SFN was partly explained by the inhibitory effects on NF-κB activation. SFN inhibited transcriptional activity of NF-κB, nuclear translocation of p65, and degradation of IκBα and IKKβ kinase activity in prostate cancer cells (24). Reduction in nuclear translocation of NF-κB in heart ventricles and kidney medullas of animals fed with broccoli diet was also observed (6). SFN suppressed both NF-κB translocation and IκBα degradation in LPS-stimulated bovine aortic ECs (5), whereas SFN inhibited DNA binding of NF-κB without interfering nuclear translocation of NF-κB and degradation of IκBα in LPS-stimulated macrophages (2). These showed that the activation of NF-κB is suppressed by SFN, although the mechanism for NF-κB suppression is varied depending on types of cell/tissue and stimuli. Nevertheless, our results consistently showed that NF-κB activation induced by LPS, TLR4, MyD88, or IKKβ was attenuated by SFN. The common downregulation of NF-κB by SFN regardless of the types of stimuli used suggests that the inhibitory effect of SFN on NF-κB activation in TLR4 signaling may be at least partly due to the direct inhibition of IKKβ and NF-κB. Interestingly, our results showed that SFN did not inhibit IRF3 activation induced by downstream molecules, such as TRIF and TBK1, whereas it inhibited LPS- and TLR4-induced IRF3 activation. These suggest that targets of SFN may exist upstream of TRIF in addition to the direct inhibition of IKKβ and NF-κB. This prompted us to investigate whether SFN affects receptor oligomerization. Our results demonstrated that SFN suppressed both ligand-induced and ligand-independent oligomerization of TLR4, suggesting TLR4 oligomerization as a novel anti-inflammatory target of SFN. Receptor oligomerization is one of the initial steps of TLR activation leading to the further activation of downstream molecules (15, 19, 25). TLR4 associates with TLR4 to form homodimers (15, 22). TLR2 forms heterodimers with TLR6 or TLR1, depending on the nature of ligand (26, 27). TLR3 oligomerization is required for the recognition of dsRNA and subsequent activation of downstream signaling (28, 29). Although TLR monomer is still active to induce NF-κB activation to a certain degree, oligomerization maximizes the activation of TLRs (22). Forced oligomerization of chimeric TLRs in which the extracellular domain is replaced with CD4 or integrin chains resulted in ligand-independent activation of downstream transcription factor (21, 22). Therefore, receptor oligomerization is the critical event triggering the maximal activation of TLRs. Our results demonstrated that SFN suppressed TLR4 oligomerization contributing to the reduction in the activation of downstream transcription factors and the expression of inflammatory protein. These results further suggest that receptor oligomerization can be an effective target to modulate TLR activity and to reduce the risk of TLR-related inflammatory diseases. Although regulation of TLR4 oligomerization and its significance need to be further confirmed in vivo in future study, our in vivo studies demonstrated that oral administration of SFN exhibited protective effects against LPS-induced inflammation, suggesting that inhibitory effects of SFN on TLR4 oligomerization can culminate in anti-inflammatory responses in vivo.

SFN possesses sulfhydryl-modifying activity through the reaction of its isothiocyanate moiety with thiols in proteins to form thionoacyl adducts (23). It has been shown that SFN transiently depletes the intracellular glutathione level (5). SFN modifies several cysteine residues in Keap1 protein, resulting in the modulation of Nrf2 activation (23). The inhibitory effect of SFN on TLR4 oligomerization is dependent on its reactivity with thiol group, because our LC-MS/MS analysis showed the adduct formation of SFN with cysteine residues in extracellular domain of TLR4 and thiol donors reversed SFN-induced impairment of receptor oligomerization and NF-κB activation. These results suggest that the reactivity of SFN to sulfhydryl moiety is one of the attributing factors for the inhibition of receptor oligomerization and that the modification of sulfhydryl residues of TLR4 may modulate the activity of TLR4. Consistently, other thiol-reactive compounds, such as parthenolide and ethacrynic acid, were able to suppress LPS-induced TLR4 oligomerization and NF-κB activation. Our previous results also showed that auranofin and cinnamaldehyde, which are reactive with the thiol group, can also suppress TLR4 oligomerization (30, 31). The potency to suppress TLR4 oligomerization seems to be different depending on nature of the compounds. In this study, we newly demonstrated that thiol-reactive compound formed adducts with cysteine residues of TLR4 extracellular domain. These results suggest that thiol modification can be a new strategy for the development of anti-inflammatory therapy by regulating TLR4 oligomerization. It is now well understood that engagement of TLR ligand induces oligomerization of the extracelluar domains and thus initiates adaptor recruitment to the intracellular domain and downstream signal transduction (15, 19, 25, 29). Although complete deletion of extracellular domain of TLR4 resulted in a loss of activity (32), truncated TLR4, which has partial extracellular domain, became constitutively active and formed oligomers spontaneously when overexpressed (18, 33). These suggest the critical role of extracellular domain in the formation of receptor oligomers and the activation of downstream signaling. In addition, when cells are challenged to chemicals, the extracellular domain of membrane receptor should be the first target encountered with chemicals before chemicals become available in cellular cytosolic compartment. Therefore, we have focused to determine whether SFN could modify the extracellular domain of TLR4. Our results demonstrated that SFN formed adducts with cysteine residues in extracellular domain of TLR4. It remains to be further investigated what is the role of these cysteines in oligomer formation process and how modification by SFN can be directly linked to the inhibition of TLR4 oligomerization. In addition, it cannot be still excluded that SFN may also modify other domains of TLR4 or other molecules in receptor complexes and that these events could also contribute to the inhibitory effects of SFN on TLR4 signaling. Regarding the experiments using DTT and NAC, we used the concentrations where these reducing agents themselves do not affect TLR4 oligomerization and signaling (Supplemental Fig. 2). Therefore, it is plausible that these reducing agents blocked inhibitory effects of SFN by reacting with SFN rather than TLR4. Park et al. (19) revealed that oligomerization of TLR4 is mainly dependent on noncovalent interactions, such as hydrogen bond, hydrophobic, and hydrophilic interactions, rather than disulfide bonds. LPS binds to a large hydrophobic pocket in MD-2 and directly mediates dimerization of two TLR4-MD-2 complexes. The hydrophobic R2 lipid chain interacts with two phenyalanines, F440 and F463, and one leucine, L444, of TLR4. The hydrophobic residues, V82, M85, L87, I124, and F126 of MD-2 support this hydrophobic interface. G123, K125, and R90 in MD-2 and the 3-hydroxy group of R2 chain of LPS form hydrogen bonds with S416, N417, E439, and Q436 of TLR4, respectively. Two phosphate groups in LPS also contribute to receptor multimerization by forming ionic interactions with a cluster of positively charged residues in TLR4 and MD-2. In addition, TLR4 makes direct dimerization contact with other TLR4 in the central region of the C-terminal domain in the extracellular domain through the formation of hydrogen bonds and hydrophobic interaction. Because SFN binds to multiple cysteines in TLR4 as demonstrated by LC-MS/MS analysis, modification of TLR4 by SFN may provide bulky interference to the interactions among LPS, MD-2, and TLR4 dimer, resulting in the blockade of oligomerization.

The anti-inflammatory activity of SFN is often explained by the activation of keap-1/Nrf2 pathway. Lin et al. (34) reported that SFN suppressed LPS-induced expression of inflammatory genes in an Nrf2-dependent manner. In this report, Nrf2 KO blocked the inhibitory effects of SFN on mRNA levels of TNF-α, IL-1β, COX-2, and iNOS and protein levels of COX-2 and iNOS. However, inhibitory effects of SFN on production of IL-1β, PGE₂, and nitrites were still observed in Nrf2-KO macrophages, although the magnitude of the inhibitory effects in Nrf2-KO macrophages was smaller than that in WT cells. Our results also showed that inhibitory effects of SFN on TLR4 signaling were still observed in Nrf2-KO MEFs as well as WT MEFs. These suggest that other mechanisms in addition to Nrf2 activation still play a role in anti-inflammatory effects of SFN on TLR4-mediated signaling. Collectively, our results present TLR4 oligomerization as a novel molecular target for anti-inflammatory activity of SFN. TLR4 is closely linked to a variety of inflammatory states including sepsis, atherosclerosis, insulin resistance, obesity, and cancer. Therefore, the modulation of TLR4 activity by SFN may exert the beneficial impacts on the development of chronic diseases.

We thank Dr. J.-O. Lee for purified extracellular domain of human and mouse TLR4 and for helpful discussion.

Disclosures The authors have no financial conflict of interest.