Direct Engagement of TLR4 in Invariant NKT Cells Regulates Immune Diseases by Differential IL-4 and IFN-γ Production in Mice

During interaction with APCs, invariant (i) NKT cells are thought to be indirectly activated by TLR4-dependently activated APCs. However, whether TLR4 directly activates iNKT cells is unknown. Therefore, the expression and function of TLR4 in iNKT cells were investigated. Flow cytometric and confocal microscopic analysis revealed TLR4 expression on the surface and in the endosome of iNKT cells. Upon LPS stimulation, iNKT cells enhanced IFN-γ production, but reduced IL-4 production, in the presence of TCR signals, depending on TLR4, MyD88, TRIF, and the endosome. However, enhanced TLR4-mediated IFN-γ production by iNKT cells did not affect IL-12 production or CD1d expression by DCs. Adoptive transfer of WT, but not TLR4-deficient, iNKT cells promoted antibody-induced arthritis in CD1d−/− mice, suggesting that endogenous TLR4 ligands modulate iNKT cell function in arthritis. Furthermore, LPS-pretreated WT, but not TLR4-deficient, iNKT cells suppressed pulmonary fibrosis, but worsened hypersensitivity pneumonitis more than untreated WT iNKT cells, indicating that exogenous TLR4 ligands regulate iNKT cell functions in pulmonary diseases. Taken together, we propose a novel direct activation pathway of iNKT cells in the presence of TCR signals via endogenous or exogenous ligand-mediated engagement of TLR4 in iNKT cells, which regulates immune diseases by altering IFN-γ and IL-4 production.

Toll-like receptors (TLRs) are pathogen recognition molecules that activate the immune system as part of innate immunity [12]. TLRs expressed in dendritic cells (DCs) link innate and adaptive immunity by regulating the expression of chemokine receptors and adhesion molecules that allow DCs to mature and migrate into the draining lymph nodes [12]. TLR4 is complexed with MD-2 and CD14, and binds lipopolysaccharide (LPS) [13]. Upon ligand binding, TLR4 associates with myeloid differentiation factor 88 (MyD88) and Toll-interleukin-1 receptor domain-containing adaptor inducing IFN-b (TRIF), which is crucial for the recruitment of several proteins that are critical in signal transduction [14,15]. MyD88-deficient mice have a profound defect in the activation of antigen-specific Th1 but not Th2 immune responses, suggesting that TLR signals play a critical role in balancing Th1/Th2 responses [16]. TLR4 specifically promotes production of the Th1-inducing cytokine IL-12 [17] and the chemokine interferon-gamma inducible protein (IP)-10 [18]. In contrast, Dabbagh and colleagues demonstrated that TLR4 is required for optimal Th2 responses against nonpathogenic antigens, which are dependent on DC maturation and cytokine production [19]. These conflicting data suggest that TLR4 plays a complex role in regulating the Th1/Th2 balance, depending on target tissue microenvironment. Upon TLR4 engagement, antigen presenting cells (APCs) are activated and produce IL-12 (cytokinedriven) and/or enhance presentation of endogenous glycolipid by CD1d molecules (self-antigen-driven) to iNKT cells, triggering their activation [20,21]. The constitutive expression of high levels of IL-12 receptor endows iNKT cells with a rapid response to IL-12 and activation [22,23]. Based on these findings, cytokine-and self-antigen-driven indirect pathways have been suggested to be the mechanisms by which TLR4-mediated activation of APCs regulates iNKT cell functions [21]. Therefore, it is generally accepted that iNKT cells are activated indirectly by TLR4dependently activated DCs rather than directly by self-expressed TLR4 expressed.
Although the expression pattern and functions of TLR4 in iNKT cells remain controversial, TLR4 is expressed on the cell surface of naïve CD4 + T cells in mice and activated T cells in humans [24]. Moreover, TLR4 regulates subsequent TCRdependent CD4 + T cell responses in experimental colitis in mice [25,26]. These findings led us to hypothesize that engagement of TLR4 might directly activate iNKT cells, resulting in modulation of immune responses in various diseases. Therefore, to address this hypothesis, we investigated whether (i) TLR4 is expressed in iNKT cells and (ii) direct engagement of TLR4 modulates the function of iNKT cells in vitro and in vivo. Our data demonstrate that direct engagement of TLR4, expressed on the cell surface and localized in the early endosome, enhances IFN-c production and reduces IL-4 production by iNKT cells, which modulates immune responses in various iNKT cell-mediated immune diseases.

Results
iNKT Cells Constitutively Express TLR4 on the Surface and in the Early Endosome To explore whether iNKT cells express TLR4, liver mononuclear cells were obtained from wild type (WT) B6 or TLR4 2/2 mice and TLR4 expression was examined both on the cell surface and in the cytoplasm of iNKT cells, macrophages, and conventional T cells. Flow cytometric and confocal microscopic analysis revealed that TLR4 was significantly expressed both on the cell surface and in the cytoplasm of macrophages and iNKT cells from WT, but not TLR4 2/2 mice, whereas none of T cells from WT and TLR4 2/2 mice expressed TLR4 ( Fig. 1A-C). Moreover, TLR4 was co-localized with the early endosome maker EEA-1 in the cytoplasm of a-GalCer/CD1d tetramer + iNKT cells from WT B6 mice, indicating that TLR4 is localized in the endosomal compartment of the cytoplasm in iNKT cells. Like as TLR4, both F4/80 + macrophages and iNKT cells also expressed CD14 on their surface (Fig. 1D). These findings indicate that a-GalCer/ CD1d tetramer + iNKT cells express TLR4 on the cell surface and in the cytoplasmic endosomal compartment.

Direct Engagement of TLR4 in iNKT Cells Enhances IFN-c Production, but Suppresses IL-4 Production
To investigate the functional roles of TLR4 in iNKT and NK1.1 + TCR-b + NKT cells, the production of IL-4 and IFN-c by sorted iNKT NK1.1 + TCR-b + NKT cells was measured after LPS and/or TCR stimulation. LPS-mediated TLR4 engagement reduced IL-4 production in B6 mouse iNKT and NK1.1 + TCRb + NKT cells activated by anti-CD3+ CD28 mAbs, whereas LPS enhanced IFN-c production ( Fig. 2A and Fig. S1A, B). Moreover, LPS modulated IL-4 and IFN-c production by activated iNKT cells in dose-dependent and TLR2-independent manners (Fig.  S1C, D). Unlike WT NKT cells, LPS did not alter the production of these cytokines by TLR4-deficient iNKT and NK1.1 + TCR-b + NKT cells in the presence of anti-CD3+ CD28 mAbs ( Fig. 2A and Fig. S1B). These results suggest that TLR4 engagement in TCRmediated activated iNKT and NK1.1 + TCR-b + NKT cells enhanced IFN-c production, but reduced IL-4 production.
T-bet and GATA-3 are key transcription factors that regulate the expression of Th1/Th2 cytokine genes, respectively [28,29,30]. Therefore, it has been hypothesized that T-bet and/or GATA-3 might contribute to the differential cytokine production by iNKT cells through TLR4 engagement. To address this hypothesis, we measured the transcriptional levels of GATA-3 and T-bet in gated iNKT cells from B6 or TLR4 2/2 mice stimulated with anti-CD3+ CD28 mAbs and/or LPS (Fig. 2B). iNKT cells from B6 and TLR4 2/2 mice stimulated with anti-CD3+ CD28 mAbs showed an increase in both GATA-3 and T-bet mRNA levels as compared with naive iNKT cells. In contrast, LPS-mediated TLR4 engagement reduced GATA-3 transcriptional levels and enhanced T-bet expression in WT iNKT cells, but not TLR4-deficient iNKT cells in the presence of TCR signals as compared with iNKT cells stimulated with anti-CD3+ CD28 mAbs alone. Although transcriptional levels of GATA-3 and T-bet do not completely reflect protein expression, these findings suggest that TLR4 engagement modulates the expression of GATA-3 and T-bet, which in turn differentially regulates Th1/Th2 cytokine production.
To confirm TLR4-mediated differential regulation of IL-4 and IFN-c production by iNKT cells in vivo, we injected B6, CD1d 2/2 , or TLR4 2/2 mice with a-GalCer and/or LPS and determined serum IFN-c and IL-4 levels. a-GalCer treatment of B6 and TLR4 2/2 mice significantly enhanced the production of IL-4 and IFN-c, whereas CD1d 2/2 mice showed no increases in the production of these cytokines (Fig. 2C). However, injection of a-GalCer and LPS into B6 mice reduced IL-4 production, but enhanced IFN-c production in the sera, whereas the differential IL-4 and IFN-c production by LPS and a-GalCer were not found in sera of TLR4 2/2 mice. These results indicate that TLR4 engagement in iNKT cells enhances IFN-c production and reduces IL-4 production in the presence of TCR stimulation in vivo. Collectively, TLR4 engagement in iNKT cells regulates T-bet and GATA3 expression, which in turn affects the production of IL-4 and IFN-c. Next, to investigate whether this differential cytokine production affect DC function, IL-12 production and CD1d expression on DCs were measured during co-culture of iNKT cells and bone marrowderived DCs (BMDCs) from WT or TLR4 2/2 mice in the presence of LPS and/or a-GalCer. Treatment with a-GalCer + LPS did not alter IL-12 production in culture supernatant of iNKT cells and WT or TLR4-deficient BMDCs as compared with that of a-GalCer. However, IL-12 production in a culture containing irradiated BMDCs from WT or TLR4 2/2 mice and iNKT cells was minimal. Furthermore, LPS and/or a-GalCer did not affect the expression of CD1d by TLR4-deficient BMDCs, whereas WT BMDC displayed enhancement of CD1d expression after LPS stimulation (Fig. 2D). These findings indicate that differential IFN-c and IL-4 production by TLR4-dependently activated iNKT cells might not affect IL-12 production and CD1d expression by DCs during TLR4-mediated interaction.

TLR4 Signals Mediated via MyD88 and TRIF Modulate Cytokine Production by iNKT Cells, which is Dependent on Endocytosis and Endosome Formation
Upon engagement of TLR4, activation signals are delivered into cells via the MyD88-dependent and TRIF-dependent pathways [15]. Therefore, to investigate whether TLR4-mediated signals in iNKT cells depend on MyD88 or TRIF, sorted iNKT cells were stimulated with LPS, anti-CD3+ CD28 mAbs, or LPS + anti-CD3+ CD28 mAbs in the presence of MyD88 or TRIF inhibitors, or a control peptide. MyD88 and TRIF inhibitors suppressed IFN-c production, but enhanced IL-4 production by iNKT cells stimulated with anti-CD3+ CD28 mAbs and LPS as compared with the control peptide (Fig. 3A). These data suggest that signaling via TLR4 in iNKT cells depends on MyD88 and TRIF.
It has been reported that LPS triggers endocytosis of membrane TLR4 via the dynamin-dependent pathway, and TLR4 engages MyD88 and TRIF sequentially at the surface and in the endosome, respectively [31]. Therefore, to explore whether these events occur in TLR4 in NKT cells, iNKT cells were incubated with dynamin or endosome (Baflilomycin A [Balfi] and Chloroquine [Chlo]) inhibitors, and co-cultured with irradiated splenocytes of WT mice in the presence of LPS, a-GalCer, or LPS + a-GalCer. Treatment with dynamin or endosome inhibitors diminished LPS-induced cytokine modulation of IFN-c and IL-4 by WT iNKT cells in the presence of TCR engagement (Fig. 3B, C). Furthermore, bead-bound LPS or blockade of surface TLR4 or CD14 using mAbs neither enhanced IFN-c production and nor reduced IL-4 production by sorted iNKT cells in the presence of TCR signals as compared with unbound LPS or isotype-matched control mAbs, respectively (Fig. 3D, E). Consistent with these findings, CD14-independent (rough) LPS could not alter IL-4 and IFN-c production by activated iNKT cells, suggesting that LPSmediated effect on iNKT cells also depends on CD14 (Fig. S1E). These findings indicate that TLR4-mediated cytokine modulation by iNKT cells in the presence of TCR signals depends on endocytosis of LPS and TLR4, and early endosome formation.

LPS-mediated Engagement of TLR4 in iNKT Cells Aggravates Saccharopolyspora Rectivirgula (SR)-induced Hypersensitivity Pneumonitis, but Attenuates Bleomycininduced Pulmonary Fibrosis
Previously, we reported that IL-4-producing NKT cells play a protective role in SR-induced hypersensitivity pneumonitis (HP) by suppressing the IFN-c-producing neutrophils, thereby decreasing SR-specific IgG levels, while NKT cells attenuate bleomycin-induced pulmonary fibrosis (BIPF) by producing IFNc [7,32]. Therefore, based on TLR4-mediated differential IL-4 and IFN-c production by NKT cells, we hypothesized that TLR4 engagement in iNKT cells reduces BIPF and aggravates SRinduced HP. To confirm this, we examined SR-induced HP and BIPF in WT and TLR4 2/2 mice. However, there was no difference in the phenotypes of SR-induced HP and BIPF between WT and TLR4 2/2 mice ( Fig. 4A and 5A). Moreover, adoptive transfer of TLR4-deficient iNKT cells reduced both serum SRspecific IgG levels in HP and hydroxyproline levels in the lungs of CD1d 2/2 mice in BIPF as much as did WT iNKT cells ( Fig. 4B and 5B). Histological alteration in the lungs from mice with HP was consistent with serum SR-specific IgG levels, although histological alteration was not homogenous in the lungs with HP (Fig. 4E). These findings indicate that TLR4-deficient iNKT cells are functionally similar to WT iNKT cells in terms of their regulatory effects on HP and BIPF. These findings suggest that endogenous TLR4 ligands in iNKT cells are only minimally involved in the regulation of HP and BIPF. Therefore, to examine whether exogenous ligand LPS-mediated TLR4 engagement in iNKT cells modulates immune responses in SR-induced HP and BIPF, sorted WT or TLR4-deficient iNKT cells were preincubated with LPS in vitro and adoptively transferred into CD1d 2/2 mice. LPS pretreatment of WT iNKT cells enhanced the levels of SR-specific IgG in serum and reduced hydroxyproline levels in the lungs of CD1d 2/2 mice as compared with untreated WT iNKT cells ( Fig. 4B and 5B). Real-time PCR revealed that adoptive transfer of LPS-pretreated WT iNKT cells increased IFN-c production and reduced IL-4 production in the lungs of CD1d 2/2 mice as compared with those mice adoptively transferred with untreated iNKT cells in the HP and BIPF models (Fig. 4C, D and 5C). In contrast, pretreatment of TLR4-deficient iNKT cells with LPS did not affect SR-specific IgG levels or hydroxyproline levels in the lungs in SR-induced HP and BIPF as compared with that of untreated TLR4-deficient iNKT cells. These findings suggest that exogenous but not endogenous TLR4 ligands affect iNKT cell function in SR-induced HP and BIPF.

Endogenous Ligand-mediated TLR4 Signals in iNKT Cells Promote Antibody-induced Arthritis
To investigate whether TLR4-mediated differential production of IL-4 and IFN-c by iNKT cells regulates iNKT cell-mediated immune diseases, we compared joint inflammation in CD1d 2/2 mice adoptively transferred with either WT or TLR4-deficient iNKT cells in the K/BxN serum transfer model. In a previous report, we demonstrated that NKT cells promoted antibodyinduced arthritis by producing IL-4 and IFN-c [5]. WT iNKT cells promoted joint inflammation in CD1d 2/2 mice as much as in B6 mice, whereas TLR4-deficient iNKT cells did not (Fig. 6A), which was consistent with histological alteration in the joints (Fig. 6C). Moreover, WT iNKT cells increased production of IL-4 and IFN-c and suppressed TGF-b production in the joint tissues of CD1d 2/2 mice, whereas TLR4-deficient iNKT cells did not alter cytokine production (Fig. 6B). To explain the failure of TLR4deficient iNKT cells to promote arthritis, it is possible that TLR4deficient iNKT cells might be functionally defective with respect to activation in the serum transfer arthritis model, thereby contributing to the failure to promote joint inflammation in CD1d 2/2 mice. However, the ability of TLR4-deficient iNKT cells stimulated with a-GalCer to produce cytokines in vitro and in vivo indicates that TLR4-deficient iNKT cells are functionally intact in terms of activation by a-GalCer/CD1d complexes ( Fig. 2A and  C). Therefore, it is unlikely that the failure to promote antibodyinduced joint inflammation in CD1d 2/2 mice given TLR4- deficient iNKT cells was attributable to functional defects in TLR4-deficient iNKT cells. These data suggest that endogenous ligand-mediated signaling through TLR4 in iNKT cells contributes to their activation, resulting in the promotion of antibodyinduced joint inflammation.

Discussion
Data regarding the expression patterns of TLR4 in NKT cells are contradictory [33,34,35]. Nagarajan et al. suggested that neither TLR4 nor CD14 were expressed by iNKT cells [35]. In contrast, TLR4 expression was demonstrated in sorted NK1.1 + TCR-b + NKT cells by RT-PCR [33] and in the cytoplasm of iNKT cells using flow cytometry [34]. However, cell surface expression and cytosolic localization of TLR4 in iNKT cells was not fully characterized. Our experiments clearly demonstrate that iNKT cells expressed TLR4 on the cell surface and in the early endosome. Several lines of reasoning might explain the contra-dictory observations of TLR4 expression patterns in iNKT cells. TLR4 expression is constitutively detected in naïve murine T cells and down-regulated in activated murine T cells [25,26], indicating that the level of TLR4 expression depends on activation status. However, constitutive expression of TLR4 in iNKT cells was not reduced after TCR-mediated activation (data not shown). Therefore, it is unlikely that the different activation status of iNKT cells in these experiments accounts for the contradictory expression pattern of TLR4 in iNKT cells. Moreover, it is also less likely that positive TLR4 expression pattern in NKT cells of our experiments might be attributable to contamination of other cell types in sorted NKT cells because purity of sorted NKT cells was extremely high (Fig. S1B). Alternatively, the sensitivity of the equipment, such as the flow cytometer, technical limitations, and/ or the use of different reagents might affect detection of TLR4 expression by iNKT cells.
Although TLR4 has been reported to be expressed on the surface of most cell types [12,13,14,16], it is also localized in the  [36,37,38]. Moreover, LPS-induced TLR4 signaling in the cytoplasm of intestinal epithelial cells requires internalization of LPS [36,39,40]. Consistent with these findings, LPS triggers endocytosis of TLR4 on the surface, which is dependent on dynamin and early endosome formation. Furthermore, TLR4 sequentially engages MyD88 in the cell membrane and TRIF in the endosome [31]. In our experiments, TLR4-mediated differential IL-4 and IFN-c production by iNKT cells was attenuated by treatment with anti-TLR4 or CD14 mAb, or inhibitors of dynamin, MyD88, TRIF, or the endosome. These data suggest that LPS binds TLR4 expressed on the cell surface and triggers endocytosis of TLR4, which in turn is re-localized to the early endosome, resulting in modulation of the functions of iNKT cells in MyD88-and TRIF-dependent manners.
Askenase et al. reported that stimulation of iNKT cells with LPS alone induced rapid, preferential, and transient production of IL-4, but not IFN-c in vitro and in vivo [34]. In contrast, our experiments demonstrated that treatment of LPS alone did not alter production of IL-4 and IFN-c by iNKT cells in vitro. Furthermore, LPS-mediated TLR4 signaling in iNKT cells enhanced IFN-c production, but reduced IL-4 production in the presence of TCR signals. Consistent with our in vitro results, two independent studies demonstrated that stimulation of iNKT cells or CD1d-restricted T cell clones with LPS enhanced IFN-c but not IL-4 production in the presence of DCs [41,42]. Furthermore, in our in vivo experiments, LPS-mediated TLR4 signaling in iNKT cells enhanced IFN-c production but reduced IL-4 levels in the lungs of mice with HP or BIPF, resulting in the regulation of these pulmonary diseases. Consistent with these results, iNKT cells have been reported to promote the development of LPS-induced lethal shock syndrome by producing IFN-c [43]. Therefore, these findings suggest that signaling via TLR4 enhances IFN-c production by iNKT cells in the presence of TCR engagement both in vitro and in vivo. Taken together, these data suggest that direct engagement of TLR4 in iNKT cells regulates immune responses by affecting the production of IFN-c and IL-4 in the presence of TCR signaling.
In this study, TLR4 engagement directly activated iNKT cells, triggering differential production of IFN-c and IL-4. Based on these findings, we suggest a novel TLR4-mediated direct activation pathway of iNKT cells during interactions between APCs and iNKT cells, in addition to the indirect pathways suggested by previous studies [20,21]. However, it is unlikely that this differential cytokine production sequentially activates DCs to enhance IL-12 production and CD1d expression during in-teraction between DCs and iNKT cells. Therefore, it is conceivable that DAMPs (damage associated molecular pattern molecules) and PAMPs (pathogen-associated molecular patterns), released from damaged cells during tissue injury and inflammation, activate both DCs and iNKT cells by direct engagement of TLR4 expressed on these cells, and interactions between activated NKT cells and DCs further augments their activation status. Therefore, iNKT cells can be activated by either direct engagement of TLR4 and/or indirectly via IL-12 production (cytokine-driven) and enhanced presentation of endogenous glycolipid (self-antigendriven) by TLR4-dependently activated DCs. This may contribute to the regulation of immune responses in various immune diseases including autoimmunity, transplantation, and tumor surveillance.
Meanwhile, several groups have demonstrated that NKT cells directly recognize cell-wall glycosylceramides of the Gramnegative, LPS-negative Sphingomonas [44,45,46]. Therefore, in the case of bacterial infection, iNKT cells are activated not only via TLR4-mediated iNKT cell activation pathways that occur during With respect to the regulation of HP and BIPF, there was no functional difference between TLR4-deficient and WT iNKT cells in CD1d 2/2 mice in adoptive transfer experiments. However, pretreatment of iNKT cells with LPS suppressed BIPF, but worsened SR-induced HP to a greater degree than did untreated-iNKT cells, indicating that exogenous rather than endogenous TLR4 ligands likely contribute to iNKT cell-mediated immune modulation in HP and BIPF. Unlike HP and BIPF, TLR4deficient iNKT cells could not restore joint inflammation in CD1d 2/2 mice in antibody-induced arthritis, whereas WT iNKT cells did. These findings suggest that endogenous ligands trigger TLR4-mediated signals in iNKT cells, resulting in the modulation of joint inflammation. Therefore, endogenous TLR4 ligandmediated signals in iNKT cells promote inflammatory arthritis, whereas exogenous TLR4 ligands regulate iNKT cell functions in HP and BIPF. Consistent with these findings, endogenous ligandmediated engagement of TLR4 is critical for the induction of rheumatoid arthritis [47,48,49,50]. Taken together, exogenous and/or endogenous ligands directly engage TLR4 on iNKT cells in vivo and trigger TLR4-mediated immune modulation by iNKT cells, depending on the microenvironment of the disease model.
In conclusion, our experiments demonstrate that TLR4 is expressed on the surface and localized in the early endosome of iNKT cells. Direct engagement of TLR4 by either endogenous or exogenous ligands enhances IFN-c production and reduces IL-4 production by iNKT cells. Moreover, TLR4-mediated differential cytokine production by iNKT cells modulates immune responses in various iNKT cell-mediated immune diseases such as SRinduced HP, BIPF, and antibody-induced joint inflammation in mice.

Real-time PCR Analysis
To perform real time PCR, total RNA was isolated using an RNeasy kit (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. RNA was reverse transcribed into cDNA using a MMLV-RT Taq polymerase (Promega, Madison, Korea) prior to PCR. For quantitative real-time PCR, genespecific PCR products were measured using an Applied Biosystems 7500 Sequence Detection System (Perkin-Elmer Biosystem), and results for each cytokine were normalized versus GAPDH expression. The following primers and probe were synthesized by

Preparation of Mononuclear Cells from the Livers
To hepatic mononuclear cells, the livers were homogenized and resuspended in loading buffer (PBS plus 10% FBS and 1 mM EDTA) and overlaid onto lympholyte-M (Cedarlane, Ontario, Canada). After centrifugation for 20 min at 9006g at 15uC, hepatic mononuclear cells were isolated from the interface.

Inhibition of Endocytosis and Endosomal Compartments in iNKT Cells
The irradiated splenocytes (1000 rad) and iNKT cells were cocultured with a-GalCer or a-GalCer+LPS for 24 h. Before coculture, iNKT cells were preincubated with 100 mM/ml of dynamin inhibitor (Dynasore, Merck, Darmstadt, Germany), 100 nM/ml of bafilomycinA1 (Sigma, St. Louis, Mo., USA) for 30 min or 100 mM/ml of chloroquine (Sigma, St. Louis, Mo., USA) for 2 h. a-GalCer was synthesized using the method developed by Kim et al. [27].

Coating Agarose Beads with LPS
To inhibit internalization of LPS into the cells, LPS-coated beads were prepared using the AminoLinkH plus Immobilization kit (Thermo Scientific Inc., Bremen, Germany) according to the manufacturer's instructions. Briefly, LPS (2 mg) was diluted in 1 ml pH 7.2 coupling buffer, and end-over-end mixed with 4% beaded agarose supplied as 50% slurry in 0.02% sodium azide. Final LPS concentrations were determined using a Limulus ameboycte lysate (LAL) assay kit (Lonza, Walkersville, MD, USA).

Preparation of Bone Marrow Dendritic Cells (BMDC) and
Co-culture with iNKT Cells To obtain BM cells, the tibia and femurs of WT and TLR4 2/2 mice were flushed. CD4 + , CD8 + T, B220 + , and I-A b+ cells in BM were depleted using a magnetic bead separation kit (Miltenyi Biotec, GmbH, Bergisch-Gladbach, Germany). BM cells were incubated with 10 ng/ml of IL-4 (eBioscience, San Diego, CA, USA) and 20 ng/ml of GM-CSF (ProSpec-Tany TechnoGene, 51 Rehovot, Israel) for 5-8 d. BMDC differentiation was determined by flow cytometric analysis using a PE-conjugated anti-mouse CD11c mAb (.70% purity). Normal or irradiated BMDCs were co-cultured with iNKT cells stimulated with a-GalCer and/or LPS for 24 h.

ELISA
For measuring cytokine production, sera and culture supernatants of sorted iNKT cells were collected and analyzed by ELISA. All cytokine ELISA kits were obtained from BD Biosciences, and ELISA was performed according to the manufacturer's instructions. Standard curves were generated using known amounts of purified murine recombinant (r) IL-4, rIFN-c, and rIL-12 (BD Biosciences, San Diego, CA, USA). The reaction was stopped with 3 N hydrochloric acid and the absorbance at 450 and 570 nm was read. Blood collected from the lateral tail veins was centrifuged, and SR-specific IgG levels in the serum (dilution 1/10) were determined by ELISA.

Induction of hypersensitivity Pneumonitis (HP)
SR antigen was prepared from an SR strain obtained from the American Type Culture Collection (ATCC), and re-suspended in pyrogen-free saline. The SR antigen contained less than 20 ng/mg of endotoxin, estimated using a limulus amebocyte lysate assay (Sigma Chemical Co., St. Louis, MO). HP was induced by intranasally instilling 150 mg of SR antigen in PBS into mice under light anesthesia. This procedure was performed on three consecutive days per week for three weeks. Mice were sacrificed four days after the final treatment. Bronchoalveolar lavage (BAL) was performed by delivering 0.8 ml of ice-cold PBS into the cannulated trachea and gently aspirating the fluid.

Serum Transfer and Arthritis Scoring
Arthritic adult K/BxN mice were bled and sera were pooled. Recipient mice were administered a 150 ml intraperitoneal injection of pooled K/BxN sera on days zero and two. Ankle thickness was measured with a caliper (Manostat, Switzerland). Joint swellings were monitored and scored as follows: 0, no joint swelling; 1, swelling of one finger joint; 2, mild swelling of wrist or ankle; 3, severe swelling of the wrist or ankle.

Histological Examination
To evaluate histological alterations of the lungs and ankles, whole lungs or joint tissues were fixed in 10% formalin 7 days after first SR treatment or K/BxN serum transfer. Paraffin-embedded tissues were cut and stained with hematoxylin and eosin (H&E).
Adoptive Transfer of iNKT Cells into CD1d 2/2 Mice Sorted iNKT cells (1610 5 cells per mouse) were adoptively transferred into CD1d 2/2 mice by intravenous injection 1 day prior to the first intranasal instillation of SR antigen, bleomycin, or K/BxN serum transfer.

Statistical Analysis
Statistical significance was analyzed using Prism ver. 5.0 (GraphPad Software, Inc., San Diego, CA, USA). A t-test was performed to compare two groups. To compare multiple groups, a one-way analysis of variance (ANOVA) using Turkey's post hoc test was used. For all analyses, P,0.05 was considered statistically significant.