T Cell Ig Domain and Mucin Domain 1 Engagement on Invariant NKT Cells in the Presence of TCR Stimulation Enhances IL-4 Production but Inhibits IFN-γ Production

Natural killer T cells are a distinct subset of T cells that are restricted by the MHC class I-like protein CD1d. NKT cells coexpress TCR complex (TCR/CD3) and NK cell markers, such as NK1.1 and Ly49 (1), possibly reflecting that they have functional activities in common with conventional T and NK cells. NK cells play critical roles in innate immunity, whereas conventional T cells typically act in adaptive immunity with B cells (2). Therefore, NKT cells are in the limelight as a functional link between the innate and adaptive immune responses.

A major subset of murine NKT cells express the semi-invariant TCR Vα14-Jα18, which recognizes glycolipid Ags presented by nonpolymorphic CD1d (3, 4). A marine sponge-derived α-galactosyl ceramide (α-GalCer) is a potent activator of these semi-invariant NKT cells (5). Upon activation, NKT cells rapidly produce large amounts of Th1 and Th2 cytokines, including IFN-γ and IL-4, -10, and -13 (6, 7). These functional activities play essential roles in regulating tumor immunity, atherosclerosis, allergy, pulmonary fibrosis, graft-versus-host disease, hypersensitivity pneumonitis, and various autoimmune diseases (8–17). However, little is known about the NKT cell-regulated production of these various cytokines in vivo.

The family of T cell Ig domain and mucin domain (TIM) proteins consists of a group of type I transmembrane proteins, including eight mouse proteins (TIM1–TIM8) and three human proteins (TIM1, TIM3, and TIM4) (18). TIM1, the first TIM protein to be identified, was discovered as a receptor for hepatitis A virus (19) and is expressed on mast cells and activated Th2 cells (20, 21). The engagement of TIM1 on T and mast cells modulates the immune response by regulating Th1 and Th2 cytokine production (20, 21). TIM4, a natural ligand of TIM1 found on mature APCs, induces T cell proliferation and survival (22, 23), and TIM1 induces phagocytosis by recognizing phosphatidylserine on apoptotic cells (24). Therefore, TIM1 has been suggested to play a critical role in immune regulation by modulating immune responses. However, the functional roles of TIM1 expressed on NKT cells have not been explored. In this study, we demonstrate that TIM1 is constitutively expressed on NKT cells and that engagement of this molecule by any of the anti-TIM1 mAbs or physiological ligands modulates cytokine production by NKT cells in the presence of TCR engagement in vitro and in vivo. As a result of these functions, TIM1 also acts as a regulator for bleomycin-induced pulmonary fibrosis (BIPF).

C57BL/6 male mice (8 wk old) were purchased from the Orient Company (Seoul, Korea). CD1d^−/− and Jα18^−/− (C57BL/6 background) mice were a gift from Dr. M. Taniguchi (Chiba University, Chiba, Japan). Mice were maintained in a specific pathogen-free environment at the Clinical Research Institute of Seoul National University Hospital. The Institutional Animal Care and Use Committee of the American Association for the Accreditation of Laboratory Animal Care-accredited Clinical Research Institute of Seoul National Hospital approved all animal experiments.

The Vα14⁺ CD1d-specific NKT hybridoma cell line DN32.D3 was a gift from Dr. Albeit Bendelac (Chicago University, Chicago, IL). Cells were maintained in DMEM supplemented with 10% FBS (HyClone Laboratories, Logan, UT) and 1% penicillin-streptomycin (Life Technologies, Paisley, U.K.).

PE-conjugated anti-mouse mAb NK1.1; CyChrome-conjugated anti-mouse TCR-β mAb; FITC-conjugated streptavidin; purified Dimer ×1: soluble dimeric mouse CD1d: Ig fusion protein; PE-conjugated anti-mouse IgG₁; purified rabbit IgG; and purified anti-mouse CD16/CD32 Ab (FcγIII/IIR, clone 2-4G2) were from BD Pharmingen (San Diego, CA). Biotin-conjugated anti-mouse TIM1 mAb RMT1-4; biotin-conjugated anti-mouse IgG_2b, κ; purified anti-mouse TIM1 mAbs RMT1-10; purified anti-mouse IgG_2a, κ; and PE-Cy5–conjugated anti-mouse F4/80 were from eBioscience (San Diego, CA). The purified anti-mouse TIM1 mAbs 222414 were from R&D Systems (Wiesbaden, Germany). The purified rabbit polyclonal Ab against mouse TIM4 was from Abcam (Cambridge, U.K.).

For flow cytometric analysis, cells were incubated with appropriate Abs for 30 min at 4°C and then washed twice with PBS. Data were acquired using a BD FACSCalibur instrument (PBS/0.05% Tween-20).

Sorted NKT, conventional T, or DN32.D3 cells (1 × 10⁵ cells/well in 24-well plates) were stimulated with purified anti-CD3+CD28 mAb and/or each anti-TIM1 mAb or α-GalCer (Alexis Biochemicals, San Diego, CA) for 24 h. The cytokine levels were measured using ELISA kits obtained from BD Biosciences, according to the manufacturer’s protocols. Colorimetric reactions were stopped by the addition of 3 N HCl, and the optical absorbance at 450 and 570 nm was determined using a spectrophotometer.

DN32.D3 cells were stimulated with each anti-TIM1 or isotype control mAb in the presence or absence of α-GalCer (200 ng/ml) for 24 h. To investigate the expression level of TIM4 in TIM4-knockdown DN32.D3 cells, cells were transfected with TIM4-specific or control small interfering RNA (siRNA). Proteins were eluted from the cells using extraction reagent (GenDepot, Barker, TX), and Western blot analysis was performed, as described previously (25). The blots were subsequently incubated with rabbit anti-mouse p27 kip1 or anti–β-actin mAb (Cell Signaling Technology, Beverly, MA). In addition, goat anti-mouse TIM4 Ab (R&D Systems) was used for siRNA-transfected cells. Proteins were visualized using an LAS-4000 Mini imaging system (Fujifilm, Tokyo, Japan).

siRNA were synthesized by Bioneer (Seoul, Korea) with the TIM4-specific sequence 5′-GCAGACAGACCAUCUUGAU-3′ and the control sequence 5′-AACGAAGCAACUAAGCUCG-3′. DN32.D3 cells were transfected with siRNA by electroporation using a microporator (Digital Biotechnology, Seoul, Korea), according to the manufacturer’s instructions. After 24 h, cells were stimulated with recombinant natural ligands for TIM (rTIM1 or rTIM4) (5 μg; R&D Systems) in the presence or absence of α-GalCer (200 ng). To biochemically characterize the interaction between TIM1 and its natural ligands, these cells were incubated with 1 mM CaCl₂ (USB, Cleveland, OH) and 2 mM EGTA (Sigma-Aldrich, St. Louis, MO).

For intracellular T-bet and GATA-3 staining, hepatic mononuclear cells (MNCs) isolated from B6 mice were stimulated with anti-TIM1 mAb (1 μg/ml or 20 μg/mice) and/or α-GalCer (200 ng/ml or 1 μg/mouse) for 2 or 24 h and then incubated with GolgiPlug (BD Pharmingen) for 6 h. The cells were stained with PE-conjugated anti-mouse NK1.1 and CyChrome-conjugated anti–TCR-β mAb or α-GalCer/CD1d tetramer and then incubated with 200 μl fixation/permeabilization solution from a Cytofix/Cytoperm kit (BD Pharmingen) for 20 min. After washing with 1 ml BD Perm/Wash buffer from the Cytofix/Cytoperm kit, the cells were stained with anti–T-bet or GATA-3 mAb (Santa Cruz Biotechnology, Santa Cruz, CA) and FITC-conjugated anti-mouse IgG₁ mAb (BD Pharmingen). To evaluate intracellular expression of cytokines, hepatic MNCs were isolated from B6 mice 2 h after injection of anti-TIM1 mAb (20 μg/mice) and/or α-GalCer (1 μg/mice) and stained with PE-conjugated anti-mouse IL-4, -10, and -13 and IFN-γ (BD Pharmingen and eBioscience), in the absence of additional stimulation.

To analyze the kinetics of interaction between anti-TIM1 mAbs and rTIM1, a Biacore 2000 system (Biacore Life Science, GE Healthcare, Piscataway, NJ) was used at a reaction temperature of 25°C. Goat anti-rat IgG was immobilized to the CM5 carboxymethylated dextran sensor chip. One hundred nanomolar RMT1-4, 222414, or RMT1-10 was injected over the chip surface for 3 min at 10 μl/min. Various concentrations (20, 10, 5, 2.5, 1.25, and 0.625 nM) of rTIM1 were introduced to the chip surface using a high-performance injection for 4 min at 30 μl/min. Dissociation of bound protein in the buffer flow was recorded for an additional 20 min. The sensor chip surface was regenerated with a 30-s injection of 20 mM HCl at 60 μl/min, followed by a 3-min stabilization period. Data transformation and sensogram plot overlays were prepared using BIAevaluation software of Biacore 2000; kinetic analysis of association and dissociation rates was performed using Biacore 2000 evaluation software. Association and dissociation kinetics were determined using a LanMour algorithm. For EC₅₀ analysis, Immuno-plates (Greiner USA Scientific, Ocala, FL) were coated with 1 μg/ml mouse rTIM1. After overnight incubation at 4°C, the plates were washed with 0.05% PBST and incubated with assay diluents (150 μl/well) for 1 h at room temperature. Various concentrations of anti-TIM1 mAbs were added, followed by peroxidase-conjugated goat anti-rat IgG Ab (Sigma-Aldrich). The reactions were developed with tetramethylbenzidine substrate (BD Biosciences) and read at 450 nm using spectrophotometer.

NK1.1⁺TCR-β⁺ NKT cells, sorted from the livers of B6 mice as described previously (8), were incubated with or without anti-TIM1 mAb for 30 min at 4°C, washed twice with PBS, and adoptively transferred into Jα18^−/− mice (3 × 10⁵ cells/mouse) via tail vein injection. To isolate α-GalCer/CD1d dimer⁺ NKT cells, splenocytes were incubated with anti-mouse CD16/CD32 mAb (2.4G2) and PE-labeled α-GalCer/CD1d dimer for 30 min at room temperature. After two washes with PBS, 1 × 10⁷ cells were mixed with 10 μl anti-PE MicroBeads suspension (Miltenyi Biotec, Bergisch-Gladbach, Germany). α-GalCer/PE-CD1d dimer⁺ NKT cells were isolated using magnetic beads, according to the manufacturer’s instructions. The purity of the isolated cells was >80%.

T cells were isolated from B6 mouse splenocytes by magnetic bead separation using a Pan T cell Isolation kit (Miltenyi Biotec), according to the manufacturer’s protocol. Briefly, splenocytes were depleted using a biotin-Ab mixture (CD11b, CD45R, DX5, and Ter119), antibiotin microbeads, and an LD magnetic bead column (Miltenyi Biotec). The cells in the negative fraction were collected. The purity of the isolated cells was >90%.

To induce pulmonary fibrosis, mice were anesthetized with 2, 2, 2-tribromomethanol (Sigma-Aldrich) and intranasally instilled with 2 mg/kg bleomycin (Nippon Kayaku, Tokyo, Japan) in 50 μl PBS. These mice were injected peritoneally with rIL-12 (1 μg in 250 μl PBS; R&D Systems) daily for 16 d and indomethacin (1.2 mg/kg; Sigma-Aldrich) for 12 d (days 10–21). The mice were sacrificed 21 d after bleomycin injection, and the lungs were removed, homogenized in 2 ml PBS, and dried for 6 h in a vacuum dryer. Samples were mixed with 1 ml 6 N HCl, incubated overnight at 110°C, and passed through a 45-μm syringe-driven filter (Millipore, Bedford, MA). A fifty-microliter aliquot of each sample was incubated with 50 μl citrate/acetate buffer and 1 ml chloramine T solution at room temperature for 15 min. Then, 1 ml Ehrlich’s solution was added, and the samples were incubated for 15 min at 65°C. The optical absorbance at 550 nm was determined using a spectrophotometer.

For quantitative real-time PCR, total RNA was isolated from whole-lung homogenates using an RNeasy Mini kit (Qiagen, Courtaboeuf, France), according to the manufacturer’s instructions. Then, the cDNA was prepared using 2 μg total RNA with Moloney murine leukemia virus-RT Taq (Koschem, Seoul, Korea). The following TaqMan Pre-Developed Assay Reagents were from Applied Biosystems (Foster City, CA): Mouse GAPDH Endogenous Control (4352339E), IFN-γ (4339850F), T-bet (Mm01299452_g1), and GATA-3 (Mm00484683_m1). The IL-4 Primer Set (GMC0044) and Pre-Developed Assay Reagent FRET Probe (GMC0045) were from Invitrogen (San Diego, CA). For TGF-β1, the primers were 5′-GCAACATGTGGAACTCTACCAGAA-3′ (forward), 5′-GACGTCAAA-AGACAGCCACTCA-3′ (reverse), and FAM-ACCTTGGTAACCGGCTG-CTGACCC-TAM-RA. The gene-specific PCR products were measured on an Applied Biosystems 7500 Sequence Detection System. Expression levels for each cytokine were normalized to that of GAPDH.

Statistical analysis was performed using the Prism 4.0 program (GraphPad, San Diego, CA). Unpaired t tests were used for two-group comparisons. Differences were considered statistically significant at p < 0.05.

To investigate the functional roles of TIM1 in NKT cells, we first examined whether NKT cells express TIM1 on their cell surface. Liver MNCs were isolated from B6 mice and stained with anti-NK1.1, anti–TCR-β, and anti-TIM1 mAbs (RMT1-4 or RMT1-10). Flow cytometric analysis revealed that TIM1 was expressed on freshly isolated hepatic NK1.1⁺TCR-β⁻ NK and NK1.1⁺TCR-β⁺ NKT cells, but the level of TIM1 expression was low compared with that in the isotype-matched IgG controls. TIM1 was not expressed on naive NK1.1⁻TCR-β⁺ T cells (Fig. 1A). TIM1 was also expressed on invariant α-GalCer/CD1d dimer⁺ NKT cells (type I NKT cells) (Fig. 1B) and on NKT hybridoma cells (Fig. 1C). These results demonstrate that TIM1 is constitutively expressed on the surface of NK1.1⁺TCR-β⁺ NKT cells, including invariant type I NKT cells.

A recent study demonstrated that TIM1 binds to TIM4 expressed on APCs (22), suggesting that the expression of TIM4 on NKT cells might affect the function of TIM1 on NKT cells. Flow cytometric analysis revealed that TIM4 was expressed at substantial levels on freshly isolated NK1.1⁺TCR-β⁺ NKT, α-GalCer/CD1d dimer⁺ NKT, and DN32.D3 cells, as well as on LPS-stimulated macrophages, but not on naive NK1.1⁻TCR-β⁺ T cells (Fig. 1D). These findings indicate that NKT cells constitutively express TIM1 and TIM4 on their cell surface.

To evaluate the functional effects of TIM1 on NKT cells, we measured cytokine production by DN32.D3 cells upon engagement of TIM1 by anti-TIM1 mAbs. DN32.D3 cells were incubated with anti-TIM1 mAbs and/or α-GalCer, and the levels of IFN-γ and IL-4, -10, and -13 were measured in the culture supernatant after 24 h (Fig. 2A). The anti-TIM1 mAbs RMT1-4, RMT1-10, and 222414 suppressed IFN-γ production in DN32.D3 cells but did not affect IFN-γ production in nonstimulated DN32.D3 cells.

The pattern of Th2 cytokine production in DN32.D3 cells upon TIM1 engagement in the presence of α-GalCer varied with the particular mAb clone used. Cotreatment of DN32.D3 cells with α-GalCer and mAb RMT1-4, but not 222414 or RMT-10, increased the production of IL-4, -10, and -13 by these cells (Fig. 2A). To confirm the specific functions of TIM1 in NKT cells, we sorted hepatic NK1.1⁺TCR-β⁺ and α-GalCer/CD1d dimer⁺ NKT cells from B6 mice (Fig. 2B, 2C) and incubated them with plate-coated anti-CD3+CD28 mAb and/or anti-TIM1 mAbs for 24 h. The resulting pattern of cytokine production in these cells upon TIM1 engagement by RMT1-4 was similar to that observed in DN32.D3 cells. Sorted NK1.1⁺TCR-β⁺ or α-GalCer/CD1d dimer⁺ NKT cells produced IL-4, -10, and -13 and IFN-γ upon engagement of CD3, whereas isotype-matched control Abs for the anti-TIM1 mAbs did not modulate CD3-mediated cytokine production of NKT cells. RMT1-4, RMT1-10, and 222414 reduced IFN-γ production in anti-CD3 mAb-stimulated NK1.1⁺TCR-β⁺ and α-GalCer/CD1d dimer⁺ NKT cells. In sorted NK1.1⁺TCR-β⁺ NKT cells, RMT1-4 enhanced IL-4, -10, and -13 production in the presence of CD3-mediated activation signals, whereas RMT1-10 slightly increased IL-10 production but did not enhance IL-4 and -13 production (Fig. 2B). However, neither RMT1-4 nor RMT1-10 alone modulated cytokine production in sorted NK1.1⁺TCR-β⁺ NKT cells in the absence of CD3 engagement. In contrast to RMT1-4 and RMT1-10, mAb 222414 enhanced IL-4, -10, and -13 production in sorted NK1.1⁺TCR-β⁺ NKT cells, but only in the absence of CD3-mediated activation of the cells. Consistent with the production of cytokines by NK1.1⁺TCR-β⁺ NKT cells, α-GalCer/CD1d dimer⁺ NKT cells responded to anti-TIM1 mAb-induced engagement of TIM1 by increasing their production of IL-4 and decreasing their production of IFN-γ (Fig. 2C).

To compare the functional effects of TIM1 engagement in NKT cells with those in conventional T cells, we sorted conventional T cells and cultured them with anti-CD3+CD28 mAbs and/or anti-TIM1 mAbs for 24 h. Consistent with the pattern of cytokine production previously reported (20), anti-TIM1 mAb did not alter the production of IFN-γ by T cells in the presence of anti-CD3+CD28 mAbs, but TIM1 engagement enhanced IL-4 production (Fig. 2D). In addition, isotype-matched control Abs did not alter cytokine production in T cells stimulated with anti-CD3 and anti-CD28 mAbs. These findings suggest that mAb engagement of TIM1 provides costimulatory signals to NKT cells, resulting in suppressed IFN-γ production and partially enhanced IL-4, -10, and -13 production, in contrast to the effect of TIM1 on conventional CD4⁺ T cells.

The modulation of NKT cell cytokine production upon anti-TIM1 mAb engagement of TIM1 might be attributable to a blockade of the interaction between TIM1 and its ligands. TIM1 engagement on T cells by the recombinant extracellular domain of TIM1 is known to inhibit the degradation of p27, which is initiated by TCR engagement (26). Therefore, to rule out this possibility, we measured the p27 expression level in NKT hybridoma cells upon TIM1 and TCR engagement. In blotting assays, p27 expression in DN32.D3 cells was reduced by α-GalCer stimulation but not by TIM1 engagement alone. Moreover, TIM1 engagement on DN32.D3 cells by any of a series of anti-TIM1 mAbs restored p27 expression in the presence of α-GalCer stimulation (Fig. 2E). Therefore, these findings suggest that the apparent signaling effects of the anti-TIM1 mAbs used in our experiments were mediated by TIM1 and that the mAbs did not block the interaction between TIM1 and its ligands. To estimate the affinity of anti-TIM1 mAbs for TIM1, we performed EC₅₀ and Biacore experiments using three anti-TIM1 mAbs and rTIM1 (Fig. 3). The EC₅₀ and Biacore analysis revealed that RMT1-4 showed the highest affinity (K_D = 1.64 nM) for TIM1, and RMT1-10 showed the lowest affinity (K_D = 8.29 nM), whereas the affinity of 222414 (K_D = 2.86 nM) was closer to that of RMT1-4 (Fig. 3). However, the K_D values for RMT1-10 differed slightly between our experiments and those of Xiao et al. (27) (8.29 nM versus 5.102 nM), which might be attributed to differences in the analysis process of K_D and the reagents used.

To investigate the physiological functions of TIM1 in NKT cells, we used rTIM1 or rTIM4 to engage TIM1. Recently, TIM1 and TIM4 were reported to act as ligands for themselves (24, 28). Because we observed the expression of TIM1 and TIM4 on the NKT cell surface, we sought to eliminate any possible interaction between TIM1 and TIM4 by using siRNA to generate TIM4-knockdown DN32.D3 cells (Fig. 4A). Incubation with rTIM1 or rTIM4 in the presence of α-GalCer enhanced the production of IL-4 and IFN-γ in wild type (WT) DN32.D3 cells (Fig. 4B, 4D), but it reduced IFN-γ production while enhancing IL-4 production in the TIM4-knockdown DN32.D3 cells (Fig. 4C, 4E). Moreover, the interaction between TIM1 on DN32.D3 cells and its natural ligands were dependent of Ca⁺² (Fig. 4F). These results indicate that TIM1 engagement on NKT cells by rTIM1 or rTIM4 inhibits the production of IFN-γ while increasing the production of IL-4, consistent with the pattern of cytokine production that we observed in NKT cells treated with anti-TIM1 mAbs.

T-bet and GATA-3 are major transcription factors that regulate expression of Th1 and Th2 cytokine genes and are associated with Th1 and Th2 responses (27, 29). Therefore, we hypothesized that T-bet and/or GATA-3 expression might affect Th1 and Th2 cytokine production by NKT cells in response to TIM1 engagement. To address this hypothesis, we evaluated the intracellular expression level of T-bet and GATA-3 in NK1.1⁺TCRβ⁺ NKT cells gated from the liver MNCs of B6 mice. The cells were stimulated with anti-TIM1 mAbs and/or α-GalCer for 24 h before assessment (Fig. 5A, 5B). NK1.1⁺TCR-β⁺ NKT cells exhibited increased cytoplasmic expression of T-bet and GATA-3 after treatment with α-GalCer for 24 h. Moreover, stimulation with α-GalCer and an anti-TIM1 mAb (RMT1-4, RMT1-10, or 222414) significantly reduced T-bet expression compared with α-GalCer treatment alone (Fig. 5A). In contrast, cotreatment of NK1.1⁺TCR-β⁺ NKT cells with RMT1-4, but not RMT1-10 or 222414, in the presence of α-GalCer enhanced GATA-3 expression over that obtained with α-GalCer treatment alone (Fig. 5B). Consistent with this expression pattern of T-bet and GATA-3 in NK1.1⁺TCR-β⁺ NKT cells, we observed increased GATA-3 expression and reduced T-bet expression upon TIM1 engagement on α-GalCer/CD1d dimer⁺ NKT cells using anti-TIM1 mAbs (Fig. 5C, 5D). However, anti-TIM1 mAbs alone minimally affected the expression levels of GATA-3 in α-GalCer/CD1d dimer⁺ NKT cells (Supplemental Fig. 1). These findings were confirmed using real-time PCR analysis, which yielded a pattern of T-bet and GATA-3 expression in TIM1-engaged α-GalCer/CD1d dimer⁺ NKT cells that was similar to that obtained by flow cytometric analysis (Fig. 5E). These results suggest that TIM1-mediated modulation of T-bet and/or GATA-3 expression in NKT cells contributes to the regulation of cytokine production.

To examine the costimulatory effects of TIM1 on NKT cells in vivo, we injected B6 and Jα18^−/− mice with α-GalCer or α-GalCer plus an anti-TIM1 mAb (Fig. 6). After 2 or 24 h, the levels of IL-4, -10, and -13 and IFN-γ in mouse sera were determined using ELISA. Consistent with our in vitro results, the injection of α-GalCer and an anti-TIM1 mAb into B6 mice, but not Jα18^−/− mice, suppressed IFN-γ production relative to that obtained after injection of α-GalCer alone (Fig. 6D). IL-4, -10, and -13 levels were significantly higher in B6 mice injected with RMT1-4 and α-GalCer than in B6 mice injected with α-GalCer alone, but neither RMT1-10 nor 222414 had any effect on α-GalCer–induced Th2 cytokine production (Fig. 6A–C). Interestingly, cotreatment of Jα18^−/− mice with α-GalCer and each anti-TIM1 mAb slightly increased IL-13 production (Fig. 6C). In addition, isotype-matched control Abs did not affect cytokine production in α-GalCer–treated B6 or Jα18^−/− mice. To address the direct effects of TIM1 engagement on NKT cells, we investigated the expression of various cytokines in NKT cells that were obtained from B6 mice administered α-GalCer and anti-TIM1 mAbs and stained without any stimulation ex vivo. The gated NKT cells showed enhancement of cytoplasmic IL-4, -10, and -13, whereas reduced expression of IFN-γ was seen in NKT cells from mice treated with α-GalCer and anti-TIM1 mAb compared with mice injected with α-GalCer alone (Fig. 6E). These findings suggest that anti-TIM1 mAb specifically exerts modulation of cytokines on NKT cells. Combined, these results indicate that TIM1 engagement differentially modulates cytokine production in NKT cells in the presence of TCR engagement in vivo.

IFN-γ–producing NKT cells were shown to inhibit BIPF by regulating the production of TGF-β (10). Therefore, we hypothesized that TIM1 engagement on NKT cells modulates BIPF by suppressing IFN-γ production. To functionally link TIM1-mediated modulation of IFN-γ production by NKT cells with pulmonary fibrosis, we sorted NKT cells, preincubated them with anti-TIM1 mAbs, and adoptively transferred them into Jα18^−/− mice using the BIPF model (Fig. 7). In addition, rIL-12 and indomethacin were used, respectively, for antagonistic and agonistic agents for BIPF of B6 mice as controls. Three weeks after bleomycin injection, the amounts of hydroxyproline (a component of collagen) in the lungs of the injected mice were measured (Fig. 7A). As previously reported (30, 31), rIL-12 reduced the levels of hydroxyproline in the lungs of B6 mice, whereas indomethacin enhanced the production of hydroxyproline upon injection of bleomycin (Supplemental Fig. 2). The adoptive transfer of NKT cells preincubated with any of the anti-TIM1 mAbs increased lung hydroxyproline levels (Fig. 7A), whereas the transfer of untreated (WT) NKT cells significantly reduced lung hydroxyproline levels. The body weights of the mice were monitored as a biological index of BIPF. Consistent with the amounts of hydroxyproline in the lungs, the Jα18^−/− mice weighed significantly less than the B6 mice. Moreover, Jα18^−/− mice that were adoptively transferred with anti-TIM1 mAb-preincubated NKT cells weighed significantly less than did noninjected Jα18^−/− mice (Fig. 7B). In contrast, Jα18^−/− mice adoptively transferred with NKT cells that were not preincubated exhibited increased body weight in the BIPF model. These findings suggest that TIM1 engagement on NKT cells modulates BIPF.

To investigate the mechanism by which TIM1-mediated signaling of NKT cells enhanced BIPF, we evaluated the expression of various cytokines in the lungs of Jα18^−/− mice 1 wk after bleomycin injection. The adoptive transfer of NKT cells preincubated with an anti-TIM1 mAb (RMT1-4, RMT1-10, or 222414) reduced the levels of IFN-γ in the lungs of these mice but enhanced their lung levels of TGF-β. In contrast, transfer of NKT cells pretreated with control IgGs increased IFN-γ production and decreased TGF-β production in the lungs (Fig. 7C). The level of IL-4, unlike that of the other cytokines, was not altered by the adoptive transfer of NKT cells pretreated with an anti-TIM1 mAb or control IgG (Fig. 7C). To address whether NKT cells contribute to TIM1-mediated regulation of BIPF by producing TGF-β, sorted NKT and DN32D3 cells were incubated with anti-TIM1 mAb alone or anti-TIM1, CD3, and CD28 mAbs, and the amounts of TGF-β were measured using ELISA. Three anti-TIM1 mAbs induced the production of TGF-β by NKT cells (Fig. 7D). However, anti-CD3 and CD28 mAb did not enhance TGF-β production by NKT cells stimulated with anti-TIM1 mAb. Among anti-TIM1 mAbs, RMT1-10 induced the greatest production of TGF-β by NKT cells, whereas 222414 induced small amounts of TGF-β production. These findings suggest that the engagement of TIM1 induces TGF-β production by NKT cells. However, BIPF was significantly reduced in mice treated with three anti-TIM1 mAbs compared with untreated mice. Moreover, the reduced amounts of BIPF were similar in mice injected with three anti-TIM1 mAbs. Therefore, these findings indicate that the amounts of BIPF were not directly correlated with those of TIM1-induced TGF-β production by NKT cells. Therefore, TGF-β produced by NKT cells does not seem to affect BIPF, although the engagement of TIM1 induces the production of TGF-β by NKT cells. Taken together, these results suggest that mAb-mediated engagement of TIM1 on NKT cells reduced IFN-γ production in the lungs, promoting BIPF.

Our experiments demonstrated that TIM1 and TIM4 are constitutively expressed on NKT cells and that the engagement of TIM1 on NKT cells by mAbs or physiological ligands, such as TIM1 or TIM4, modulates cytokine production in the presence of TCR signals. Several studies demonstrated that levels of TIM1 are high in Th2 cells but lower in Th1 and Th17 cells (20, 22, 32). Moreover, CD4⁺ T cells treated in vitro with anti-TIM1 (3B3) and anti-CD3 mAbs exhibited higher levels of IL-4, but not IFN-γ, than did cells stimulated with anti-CD3 mAb alone (20). These findings suggest that TIM1 plays a critical role in the modulation of Th2 immune responses in vivo.

In contrast to our in vitro findings, administration of anti-TIM1 mAb and a specific Ag into mice increased the production of IL-4 and IFN-γ in naive T cells, which inhibited the development of respiratory tolerance (20). Unlike conventional T cells, TIM1 engagement on NKT cells consistently suppressed IFN-γ production while enhancing IL-4 production in vitro and in vivo. Taken together, these results suggest that TIM1 engagement on NKT cells enhances Th2 cytokine production and reduces Th1 cytokine production, phenomena that might favor the Th2 immune response over that of Th1. Furthermore, TIM1-mediated differential regulation of cytokine production is intriguing in the context that most costimulatory molecules, such as CD28, ICOS, and glucocorticoid-induced TNFR, provide signals to enhance NKT cell production of IL-4 and IFN-γ (33–35). However, in a study by Hayakawa et al. (33), blockade of CD40/CD154 interaction inhibited α-GalCer–induced IFN-γ production but enhanced IL-4 production, suggesting that CD40-mediated signaling differentially regulates α-GalCer–induced cytokine production in NKT cells, creating a pattern opposite to that of cytokine production in NKT cells upon TIM1 engagement.

To confirm the functional effects of TIM1 engagement in NKT cells, we must exclude the possibility that anti-TIM1 mAbs block the interaction between TIM1 and its ligands rather than providing functional signals through TIM1. In a recent study, the low-avidity Ab RMT1-10 minimally inhibited the TIM1–TIM4 interaction, whereas the high-avidity Ab 3B3 strongly blocked the interaction (22, 36). In our experiments, TIM1 engagement on DN32.D3 cells by anti-TIM1 mAbs increased the expression of p27, a mediator of TIM1-induced signals. Therefore, it is unlikely that the modulation of cytokine production of NKT cells is primarily a result of an anti-TIM1 mAb-mediated blockade of the TIM1–ligand interaction.

The functional effects of anti-TIM1 mAbs on T cells depend on the binding epitope, affinity, and avidity of the particular mAb (27, 37). The anti-TIM1 mAb 4A2.2, which binds the IgV domain of TIM1, is antagonistic to the Th2-type response, whereas another anti-TIM1 mAb, 1H8.2, which binds to the mucin domain of TIM1, transmits agonistic signals for the Th2-type response (37). Xiao et al. (36) found that the high-affinity anti-TIM1 mAb 3B3 enhanced the generation of Ag-specific T cells and the production of IFN-γ and IL-17, whereas the low-affinity mAb RMT1-10 inhibited the induction of Ag-specific T cells and the production of these cytokines, causing strong Th2 responses in an experimental autoimmune encephalomyelitis model. Although 3B3 and RMT-10 recognize very similar epitopes located in the TIM1 IgV domain, the affinity of 3B3 for TIM1 is 17 times that of RMT1-10. Moreover, independent studies of the functions of TIM1 in CD4⁺CD25⁺ regulatory T cells provided conflicting results in two transplantation models, which might be attributable to the different TIM1-binding affinities of the Abs (3B3 or RMT1-10) used in the two experiments (38, 39). Unlike T cells, mast cells respond similarly, in terms of IL-4, -6, and -13 production, to high- and low-affinity mAbs upon TIM1 engagement in the presence of IgE and Ag (21). These findings suggest that TIM1-mediated functional modulation in T cells depends on TIM1 binding affinity/avidity, whereas TIM1-mediated cytokine regulation in mast cells does not. However, whether TIM1 binding affinity affects the modulation of cytokine production in NKT cells is unclear. Our experiments demonstrated that RMT1-4 had the greatest affinity for TIM1, and RMT1-10 showed the lowest affinity among the three anti-TIM1 mAbs, whereas the affinity of 222414 was closer to that of RMT1-4. Moreover, RMT1-4 stimulated NKT cells to produce Th2 cytokines in the presence of TCR stimulation, whereas 222414 and RMT1-10 could not. There are two possible explanations for this discrepancy in TIM1 function on Th2 cytokine production by NKT cells using different anti-TIM1 mAbs. Based on the affinity of anti-TIM1 mAbs, those with high affinity might provide costimulatory signals for Th2 cytokines via TIM1 to NKT cells. Alternatively, different binding epitopes of TIM1 protein by these mAbs might determine different signals of TIM1 into NKT cells rather than the binding affinity of these anti-TIM1 mAbs.

With regard to Th1 cytokine, all of the anti-TIM1 mAbs exerted consistently inhibitory effects on IFN-γ production in NKT cells. These findings suggest that TIM1-mediated modulation of IL-4 production by NKT cells might depend on the TIM1 binding site or affinity, whereas TIM1-mediated modulation of IFN-γ production by NKT cells is minimally affected by these conditions.

Several studies demonstrated that TIM4 interacts with TIM1, as well as with TIM4 itself (24). Moreover, TIM1 was reported to interact with itself by homophilic binding (28). In our experiments, flow cytometric analysis showed that TIM4 was constitutively expressed in freshly isolated α-GalCer/CD1d dimer⁺ cells, NK1.1⁺TCR-β⁺ NKT cells, and DN32.D3 cells. Given the diverse possible combinations of interactions between TIM1 and TIM4 in NKT cells, interpreting the effects of TIM1 engagement mediated by physiological ligands would be complicated. Therefore, experimental results obtained using anti-TIM1 Abs in NKT cell studies may be more readily interpretable than would be those obtained using natural TIM1 ligands; by using anti-TIM1 mAbs, we eliminated possible TIM1–TIM1, TIM1–TIM4, and TIM4–TIM4 interactions on these cells.

Despite the advantages of the use of mAbs in TIM1 studies, mAbs can provide only limited information relevant to the physiological functions of TIM1 because they differ from its natural ligands in their binding epitopes, affinities, and/or signaling effects. To address this issue, we also examined cytokine production using rTIM1 or rTIM4. Moreover, to eliminate possible confounding results arising from TIM4-mediated signals, we generated TIM4-knockdown DN32.D3 cells using siRNA. TIM1 engagement on TIM4-knockdown DN32.D3 cells by rTIM1 or rTIM4 in the presence of α-GalCer stimulation suppressed IFN-γ production but enhanced IL-4 production, consistent with our results obtained using anti-TIM1 mAbs. These results indicate that signaling via TIM1 in the presence of TCR engagement enhanced IL-4 production by NKT cells but inhibited their production of IFN-γ. In contrast, rTIM1- or rTIM4-induced signaling in WT DN32.D3 cells enhanced the production of IL-4 and IFN-γ compared with that observed in cells stimulated with α-GalCer alone. These findings led us to hypothesize that TIM4 engagement on NKT cells enhances the production of IL-4 and IFN-γ in the presence of TCR stimulation.

To understand the physiological roles of TIM1 in vivo, we need to identify the cells capable of providing ligands for TIM1 engagement on NKT cells. TIM4 is reportedly expressed selectively in APCs, being expressed at particularly high levels in CD11b⁺CD11c⁺ mature dendritic cells (DCs) and macrophages but not in T cells, whereas TIM1 is expressed on Th2 cells (22, 23). Our experiments clearly demonstrated that NKT cells constitutively express TIM4, as well as TIM1, on their cell surface, suggesting that APCs, Th2, or NKT cells may provide natural ligands for TIM1 on NKT cells. However, no evidence has been reported that NKT cells directly interact with Th2 cells or NKT cells for activation in vivo. Therefore, the candidate ligands for TIM1 interaction on NKT cells in vivo are the TIM4 molecules found on mature DCs and macrophages because these cells express surface CD1d molecules for the interaction between NKT cells and APCs (40).

We used a murine model of BIPF to evaluate the effects of TIM1 on NKT cells in immune disease. Previously, we reported that IFN-γ–producing NKT cells attenuate BIPF (10). In this study, we further substantiated the TIM1-mediated suppression of IFN-γ production by NKT cells in vitro by showing that TIM1 engagement aggravates BIPF by inhibiting IFN-γ production by NKT cells. The adoptive transfer of NKT cells stimulated with anti-TIM1 mAbs into Jα18 knockout mice reduced IFN-γ expression in the lungs, aggravating pulmonary fibrosis (Fig. 6). Thus, TIM1 engagement on NKT cells regulates pulmonary fibrosis in vivo, suggesting that stimulation of TIM1 on NKT cells might modulate the immune response in various immune-mediated diseases. Several studies demonstrated that NKT cells promote respiratory resistance and inflammation in asthma by producing IL-4 and -13 (7, 41, 42). Therefore, TIM1 engagement on NKT cells may promote the induction and progression of asthma and pulmonary fibrosis, although the immunological effects of anti-TIM1 Abs on the development of asthma were contradictory in two independent experiments (20, 43). In contrast, a recent study demonstrated that IFN-γ produced by NKT cells is essential for the formation of postoperative adhesion, providing a potential strategy for suppressing postoperative adhesions (44). In addition, NKT cells inhibit Saccharopolyspora rectivirgula-induced hypersensitivity pneumonitis by producing IL-4 (15). Engagement of TIM1 on NKT cells could conceivably inhibit postoperative intestinal adhesion and hypersensitivity pneumonitis but not asthma or pulmonary fibrosis. Therefore, agonistic anti-TIM1 mAbs or TIM1 ligands are potential therapeutic candidates for suppressing immune-related diseases via modulation of NKT cell cytokine production.

In conclusion, we found that the TIM1 engagement on NKT cells suppresses IFN-γ production but enhances IL-4 production in the presence of TCR stimulation in vitro and in vivo, thereby aggravating BIPF. These results suggest that TIM1 engagement on NKT cells modulates immune responses in vivo and may provide a useful therapeutic target for candidate therapies for various immunopathologies.

We thank the Department of Experimental Animals, Clinical Research Institute, Seoul National University Hospital for animal management.

Disclosures The authors have no financial conflicts of interest.