Intracellular Galectin-9 Enhances Proximal TCR Signaling and Potentiates Autoimmune Diseases

Inflammatory bowel disease (IBD) is a chronic relapsing disorder of the gastrointestinal (GI) tract with pathological characterization of intestinal inflammation and epithelial injury. Patients with IBD experience severe diarrhea, abdominal pain, fatigue, and weight loss. IBD is categorized into two major types, Crohn disease (CD) and ulcerative colitis (UC). CD, a Th1/Th17-mediated inflammatory process, affects any part of the GI tract from the mouth to the anus, whereas UC seems to be a Th2-mediated disease characterized with inflammation confined to the top layers of colon (1–3). The current therapeutic strategies for IBD are limited by low responsive rate, moderate effectiveness, high costs, and/or side effects, underlining the importance of identifying novel molecular markers and/or targets for optimized therapies.

Jostins et al. (4) conducted an imputation-based association analysis by using autosomal genotype-level data from 15 genome-wide association studies of IBD and identified various IBD loci markedly enriched in genes for immune regulation (ADA, CD40, IFNGR2, IL12B, IRF8, STAT1, STAT3, TAP1, TAP2, and TYK2). Interestingly, LGALS9 was identified as an IBD gene for a total 163 IBD loci, suggesting its putative role in modulating autoimmune diseases (4).

Expressed by many organisms, including nematodes and mammals (5), galectins are β-galactoside–binding proteins with conserved carbohydrate-recognition domains (CRDs) that bind to cell surface glycans (6). Currently, 15 galectin members have been identified in mammals with broad tissue distribution (7–9). Galectin-9 is highly expressed in lymphoid organs and plays unique roles in immune regulation (10, 11). Galectin-9 is expressed in thymic epithelial cells and modulates the process of negative selection during T cell development (11). Galectin-9 is also an apoptosis-inducing factor that binds to T cell Ig mucin domain 3 (Tim-3) on Th1 cells (12). Galectin-9 can bind to 4-1BB on T cells and promote their effector functions (13). Current studies show that galectin-9 interacts with CD44 or DR3, which further increases the stability and function of adaptive regulatory T cells (Tregs) (14, 15). Overall, these previous reports provide strong evidence of the diverse roles of galectin-9 in T cell development and function. However, these data have been obtained mainly using an exogenous galectin-9–based approach in which the physiological role of endogenous galectin-9 in the modulation of T cell responses may have been missed and/or masked.

Interestingly, galectins are also localized in the cytosol and nucleus, where they interact with other intracellular regulators to modulate a variety of cellular functions in a glycan-dependent or glycan-independent manner (16). For example, galectin-1 and galectin-3 can serve as splicing factors by directly interacting with Gemin4 during the process of spliceosome formation (17). Moreover, galectin-3 positively regulates macrophage phagocytosis through an intracellular mechanism (18). Other studies have demonstrated that galectin-3 and galectin-8 modulate antibacterial autophagy in opposite ways by binding to different host glycans exposed on damaged phagosomes (19, 20). In T cells, galectin-3 is induced in an activation-dependent manner and is expressed mainly in the cytosol but is barely detectable on the surface of activated T cells (21). Another report has indicated that endogenous galectin-3 interacts with Alix via a glycan-independent manner to trigger TCR internalization and to downregulate proximal TCR signaling (22). One study has reported that galectin-9 can be expressed in the cytosol of Jurkat human leukemia cells (23), which suggests an intracellular role of galectin-9 in T lymphocytes. However, the physiological functions of intracellular galectin-9 in T cells are not completely understood.

Although administration of exogenous galectin-9 or overexpression of galectin-9 plasmids in mice can suppress autoimmune diseases by engaging different cell surface glycoproteins or glycolipids (12, 14, 24–26), the precise effects on these diverse receptor-bearing cells are difficult to identify. Moreover, the potential mechanisms through which endogenous galectin-9 mediates T cell responses may have been overlooked in previous studies that induce autoimmune diseases in conventional Lgals9^−/− mice. To examine the modulatory potential of galectin-9 in T cells, we analyzed the expression kinetics and localization of endogenous galectin-9 in T cells and investigated the modulatory role of endogenous galectin-9 in T cell–mediated autoimmune diseases in Lgals9^−/− mice. We found that galectin-9 was recruited to immune synapses through T cell activation and promoted T cell proliferation and Th17 cell differentiation. These actions may be involved as positive regulators of the pathogenesis of T cell–mediated colitis and experimental autoimmune encephalomyelitis (EAE).

Publicly available microarray and RNA sequence datasets were downloaded from Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) along with appropriate chip annotation data. All analyses were carried out using Excel. GEO GSE57945 and GEO GSE3365 datasets were used for human IBD analysis, GEO GSE27302 dataset for analysis of mouse colitis, and GEO GSE78244 for human multiple sclerosis (MS) analysis. To analyze behaviors of specific genes in different clusters, fraction of cells expressing the gene of interest and average expression values among expressing cells were calculated (z score). The value of expression above the average expression values were showed in red color, and the value of expression below the average expression values were showed in blue color. Correlation analysis was performed on normalized expression values of interest genes in each classification.

C57BL/6 Lgals9^−/− mice were a gift from J. C. Paulson (The Scripps Research Institute, San Diego, CA) and were kindly provided by F.-T. Liu (Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan), and C57BL/6 2D2 transgenic (Tg) and Rag1^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and subsequently bred at the Animal Center of the National Defense Medical Center in Taipei, Taiwan, under specific pathogen-free conditions. Mice were treated in accordance with the Institutional Animal Care and Use Committee of the National Defense Medical Center guidelines for experiments and approved by a committee in the same office.

The cells were isolated from mouse spleen, mesenteric lymph node (MLN), and cervical lymph node (CLN) by mash tissues between two frosted microscope slides. The cells were further treated with RBC lysis buffer to eliminate erythrocytes, washed, and resuspended in RPMI 1640 supplemented with 10% FBS, 2 mM l-glutamine, 100 U/ml penicillin G, 0.1 mg/ml streptomycin, and 10 mM HEPES (Life Technologies, Waltham, MA).

Mouse CD4⁺ T cells were isolated by negative selection using a Mouse CD4 T Lymphocyte Enrichment Set (BD Pharmingen, San Jose, CA). To deplete the Tregs, biotin-conjugated anti-mouse CD25 was added in the Ab mixtures. The purity of isolated cells was >90%.

Lamina propria mononuclear cells were isolated from mouse colons tissues followed by previous report (27). Briefly, the colon was removed and cut into pieces, and the colon pieces were incubated with EDTA and DTT in HBSS, the suspension including epithelial cells, villus cells, subepithelial cells, and intraepithelial lymphocytes were removed. The remaining lamina propria was digested by collagenase type VIII (Sigma-Aldrich, St Louis, MO), and the suspension was subjected to Percoll-gradient separation (GE Healthcare Bio-Sciences, Uppsala, Sweden). The collected cells were stained with Abs and analyze by flow cytometry.

CD4⁺CD25^– T cells were attached to poly-l-lysine–coated coverslips for 15 min at 37°C, then stained with anti-mouse CD4–Alexa Fluor 488 for 1 h at 4°C, and the coverslips were fixed with 2% paraformaldehyde to stop the reaction for 1 h. The cells were then stained with anti-mouse galectin-9–Alexa Fluor 594 (137906; BioLegend) at 4°C overnight and stained with DAPI (422801; BioLegend) before microscope analysis. For the stimulation, CD4⁺CD25^– T cells stimulated with anti-CD3 (1 μg/ml, 145-2C11; BD Pharmingen) and anti-CD28 (1 μg/ml, 37.51; BioLegend, San Diego, CA) for 30 min and then cross-linked with mouse anti-hamster IgG mixture (25 μg/ml, G94-56 and G70-204; BD Pharmingen) for 1 h. After stimulation, cells were attached to poly-l-lysine–coated coverslips for 15 min, and the coverslips were fixed with 2% paraformaldehyde to stop the reaction for 1 h. The cells were then stained with anti-mouse Zap70–Alexa Fluor 594 (693505; BioLegend) and anti-mouse galectin-9–Alexa Fluor 594 (137906; BioLegend) at 4°C overnight and stained with DAPI (422801; BioLegend) before microscope analysis. Images were acquired at room temperature with Zen 2 (black edition) version 2.0 (ZEISS, Oberkochen, Germany) using an LSM880 confocal microscope (ZEISS) equipped with Plan-Apochromat 100× with 1.4 numerical aperture oil objectives (ZEISS). Data were analyzed by Zen 2 (black edition) version 2.0.

To measure cell proliferation, CD4⁺CD25^– T cells were cultured in triplicate wells of 96-well, flat-bottom plates (4–5 × 10⁵ cells/200 μl/well) with the indicated amounts of plate-bound anti-CD3/CD28. After 48 h, the cultured cells were pulsed with 1 μCi [methyl-³H] thymidine (PerkinElmer, Shelton, CT) per well and harvested after 16–18 h. The plates were harvested into a UniFilter-96 GF/C microplate (PerkinElmer), and the incorporated [methyl-³H] thymidine was detected with a Packard TopCount Microplate Scintillation Counter (PerkinElmer). To measure cell division, CD4⁺CD25^– T cells were labeled with CFSE, CellTrace Far Red (CTR), or CellTrace Violet (CTV) Cell Proliferation Kit (all from Life Technologies) by incubating 1 × 10⁷ cells in 1 ml of PBS with 2.5 mM CFSE and 5 μM CTR or 10 μM CTV for 10 min at 37°C. The reaction was quenched by adding 10 ml of cold complete RPMI 1640 medium, and the cells were washed twice. The labeled cells were cultured in 96-well, flat-bottom plates (5 × 10⁵ cells/200 μl/well) and stimulated with plate-bound anti-CD3 (0.5 μg/ml) and anti-CD28 (0.5 μg/ml) for 3 d. Data were collected on an FACSCalibur flow cytometer (BD Biosciences, San Jose, CA). For bone marrow–derived dendritic cell (BMDC)–T cell coculture, purified wild-type (WT) or Lgals9^−/− CTV–labeled CD4⁺CD25^– T cells (2.5 × 10⁵ cells/200 μl/well) were cocultured with BMDCs from C57BL/6 mice (5 × 10⁴ cells/200 μl/well) in the presence of 50 μg of myelin oligodendrocyte glycoprotein (MOG)_35–55 for 3 d. Cells were collected, and cell division was analyzed with an FACSVerse flow cytometer (BD Biosciences).

Bone marrow cells from tibias and femurs of C57BL/6 mice were cultured with 10% FBS (Life Technologies), 5 × 10⁻⁵ M 2-ME (Life Technologies), 5 mg/ml l-glutamine (Sigma-Aldrich), 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies), and 20 ng/ml GM-CSF (Sigma-Aldrich). Myeloid progenitor cells were cultured in a 10-cm dish at 4 × 10⁶ cells. On day 3 of culture, 10 ml of fresh medium containing GM-CSF was added to the culture. On day 6, half of the medium was replaced by fresh medium containing 20 ng/ml GM-CSF. On day 8, the final nonadherent population was BMDCs.

CD4⁺CD25^– T cells were stimulated at 37°C with anti-CD3 (1 μg/ml, 145-2C11; BD Pharmingen) and anti-CD28 (1 μg/ml, 37.51; BioLegend) and then cross-linked with mouse anti-hamster IgG mixture (25 μg/ml, G94-56 and G70-204; BD Pharmingen) for the indicated time periods. After stimulation, cells were immediately resuspended in 2× lysis buffer (50 mM Tris [pH 7.4], 10% glycerol, and 150 mM NaCl) containing protease inhibitor mixture (Sigma-Aldrich) and phosphatase inhibitor mixture (Roche, Mannheim, Germany). Cell lysates were separated by 10% SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and probed with Abs to Zap70 (2705), phospho-Zap70 Tyr319 (2701), Lck (2752), phospho-Lck Tyr394 (2101), phospholipase Cγ1 (PLC-γ1) (2822), phospho–PLC-γ1 Tyr394 (2821) (Cell Signaling Technology, Beverly, MA), and β-actin (AC-15; Sigma-Aldrich).

Naive CD4⁺ T cells were stimulated for 3 d with the indicated amounts of plate-coated anti-CD3 (0.05–10 μg/ml) and soluble anti-CD28 (1 μg/ml) mAbs under conditions that stimulate Th1 cells (10 ng/ml IL-12, 5 ng/ml IL-2, and 5 mg/ml anti–IL-4); Th2 cells (30 ng/ml IL-4, 5 ng/ml IL-2, and 5 mg/ml anti–IFN-γ); Th17 cells (50 ng/ml IL-6, 5 ng/ml TGF-β, 5 ng/ml IL-2, anti–IL-4, and 5 mg/ml anti–IFN-γ); or induced Tregs (iTregs) (5 ng/ml TGF-β; 5 ng/ml IL-2). For intracellular cytokine staining, T cells were stimulated with PMA and ionomycin in the presence of monensin (all from Sigma-Aldrich) for 3.5 h. The cells were analyzed by gating on CD4⁺ T cells. Data were collected on an FACSVerse flow cytometer (BD Biosciences).

The purified CD4⁺CD25^– T cells were stimulated with anti-CD3 (0.5 μg/ml) plus anti-CD28 (0.5 μg/ml) for various lengths of time (days). For recombinant galectin-9 treatment, purified CD4⁺CD25^– T cells from WT or Lgals9^−/− mice were labeled with CTV and stimulated with anti-CD3 and anti-CD28. After 72 h, the cells were purified with Histopaque (Sigma-Aldrich) and then cultured for 18 h at a density of 2 × 10⁵ cells per well in the presence of recombinant galectin-9 or PBS. The cells (1 × 10⁵ cells/100 μl/tube) were collected and stained with annexin V–FITC and 7-aminoactinomycin D in annexin V binding buffer (all from BD Pharmingen) for 15 min at room temperature in the dark. After staining, the cells were analyzed within 1 h on an FACSCalibur or FACSVerse flow cytometer.

Lgals9^−/− or WT mice were i.p. immunized with 100 μg of (4-hydroxy-3-nitrophenyl) acetyl (NP)₃₂–keyhole limpet hemocyanin (KLH) (Biosearch Technologies, Petaluma, CA). Lgals9^−/− or WT mice were immunized with 10 μl of sheep RBCs (SRBCs) in 500 μl of saline by i.p. injection, and serum was collected on the indicated days or at the time of sacrifice on day 7. The frequency of apoptotic germinal center (GC) B cells was determined using an Annexin V Apoptosis Detection Kit (BD Pharmingen). To label cells with BrdU in vivo, immunized mice were injected i.p. with 1 mg of BrdU (Sigma-Aldrich) in PBS 24 h before sacrifice, The splenocytes were isolated, and a BrdU Flow Kit (BD Pharmingen) was used to analyze the frequency of BrdU-incorporated GC B cells.

Culture supernatant fractions from T cell cultures were analyzed for levels of IFN-γ, IL-17, TNF-⍺, galectin-1, galectin-3, and galectin-9. The procedures followed the standard protocols provided by R&D Systems (Minneapolis, MN) and LifeSpan BioSciences systems (Seattle, WA). NP-specific Ab titers were measured in WT or Lgals9^−/− mice as previously described (28). Briefly, the wells of Nunc-Immuno ELISA microplates (Sigma-Aldrich) were coated with 5 μg of either NP₃₂–BSA (to measure the total amount of IgG) or NP₄–BSA (to measure the amount of high-affinity IgG) (N-5050; Biosearch Technologies) for 2 h at 37°C. The plates were washed five times with TBST, blocked with 1% BSA for 30 min, and then washed three times with TBST. Serum samples collected from WT or Lgals9^−/− mice were diluted and added to the plates, and the plates were incubated for 90 min. The plates were washed five times with TBST and then incubated with secondary Ab, HRP-conjugated anti-mouse IgG (1:20,000; Sigma-Aldrich). Finally, the plates were washed five times with TBST, and 3,3′,5,5′-tetramethylbenzidine substrate was added (BD Pharmagen). The absorbance was measured at 450 nm with an ELISA plate reader (SpectraMax M2; Molecular Devices, Sunnyvale, CA). To analyze Ig isotypes, mouse serum was collected and analyzed using a bead-based multiplex assay and the MILLIPLEX MAP Mouse Ig Isotyping Magnetic Bead Panel (MilliporeSigma, Billerica, MA).

Mouse splenic B cells were purified by positive selection using B220 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). The purified B cells (1–2 × 10⁶/ml) were stimulated with goat anti-mouse F(ab′)₂ (10 μg/ml; Jackson ImmunoResearch Laboratories), anti-CD40 (1 μg/ml; BD Biosciences), and IL-21 (200 ng/ml; Life Technologies). The frequency of plasma cells was determined 5 d later. To assess B cell activation, isolated B220⁺ cells were stimulated with goat anti-mouse F(ab′)₂ (10 μg/ml; Jackson ImmunoResearch Laboratories) and harvested 24 h later.

Cells were lysed with TRIzol Reagent (Invitrogen, Carlsbad, CA). Nucleoprotein complexes were then separated through bromochloropropane (Sigma-Aldrich), and the samples were centrifuged at 12,000 rpm at 4°C for 15 min. RNA was collected and precipitated with an equal volume of isopropanol. The samples were incubated at 25°C for 10 min and centrifuged at 12,000 rpm at 4°C for 15 min. The RNA pellets were collected and washed with ethanol by vigorous mixing. RNA was obtained by centrifugation at 7500 rpm at 4°C for 10 min. After air-drying, the RNA pellets were dissolved in RNase-free double-distilled water. Reverse transcription was performed using a Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA) to synthesize cDNA. Quantitative real-time PCR (qPCR) was performed using Power SYBR Green Master Mix (Life Technologies), and cDNA was amplified using a StepOnePlus Real-Time PCR System (Applied Biosystems). Mouse β-actin mRNA was used for internal normalization in all experiments. The fold change for each gene was calculated as the reciprocal of each mRNA normalized to the internal control. The SYBR green primer sequences were Lgals9 forward, 5′-CTTTCTACACCCCCATTCCA-3′ and Lgals9 reverse, 5′-CTCGTAGCATCTGGCAAG-3′ and Actb forward, 5′-CATTGCTGACAGGATGCAGAAGG-3′ and Actb reverse, 5′-TGCTGGAAGGTGGACAGTGAGG-3′.

Cells were resuspended in PBS supplemented with 0.5% BSA and 1 mM EDTA (FACS buffer) and stained with the Abs listed below at 4°C for 30 min. Mouse cell suspensions were preincubated with anti-mouse CD16/32 (Fc Block, clone 93; eBioscience) for 5 min before the primary Abs were added. The anti-mouse Abs used for flow cytometric analysis were the following: anti-CD45R/B220–allophycocyanin/allophycocyanin–Cy7 (RA3-6B2), anti–GL-7–FITC (GL-7), anti-CD95–PE–Cy7 (Jo2), annexin V–allophycocyanin (550474), anti-CD86–PE (GL1), anti-CD4–Alexa Fluor 700/FITC (GK1.5/RM4-5), anti-CD62L–PE (MEL-14), anti-CD44–allophycocyanin (IM7), anti-IgM–PerCP–Cy5.5 (R6-60.2), anti–BP-1–PE (BP-1), anti-CD43–allophycocyanin (S7), and anti-CD5–FITC (53-7.3) (all from BD Pharmingen); anti-CD38–PE–Cy7 (90), anti-BrdU–allophycocyanin (Bu20a), anti-CD69–allophycocyanin (H1.2F3), anti-CD138–allophycocyanin (281-2), anti-IgM–PE (RMM-1), anti-CD21/35–allophycocyanin (7E9), and anti-CD23–PE–Cy7 (B3B4) (from BioLegend).

The purified CD4⁺CD25^– T cells from 6- to 8-wk-old WT or Lgals9^−/− mice were further sorted into CD4⁺CD45RB^hiCD62L^hi T cells using an FACSAria Fusion flow cytometer (BD Biosciences) and then transferred into 6–8-wk-old immunodeficient Rag1^−/− recipients. The body weight of the recipients was monitored twice a week. At 35 d after transfer, the recipients were sacrificed, and the colon length and weight were measured. The lymphocytes were isolated from the MLNs and lamina propria, and the percentages of IFN-γ⁺ and IL-17⁺ T cells were measured.

Mouse 2D2 T cell clone can specifically recognize MOG_35–55 presented by MHC class II molecule H2-IA^b and orchestrate immune responses to destroy myelin sheath. Therefore, CD4 T cells harvested from 2D2 Tg mice can be considered as autoreactive T cells to induce EAE (29, 30). To investigate the potential role of galectin-9 in T cell and its subsequent effect in the pathogenesis of EAE, we generated Lgals9^−/− 2D2 Tg mice. After modified from previous report (29, 30), purified CD4⁺CD25^– T cells from 6- to 8-wk-old WT 2D2 Tg mice or Lgals9^−/− 2D2 Tg mice were transferred into 6–8-wk-old immunodeficient Rag1^−/− recipients, and the recipients were injected i.p. with pertussis toxin twice at days 1 and 2. EAE clinical manifestations were evaluated by daily assignment of scores from 0 to 5 as follows: 0, no clinical manifestation; 0.5, partial weakness of limb tail; 1, complete paralysis of the tail; 1.5, paralysis of the tail and waddling gait; 2, paralysis of one hind limb; 2.5, paralysis of one hind limb and partial paralysis of the other hind limb; 3, paralysis of both hind limbs; 3.5, forelimb weakness; 4, forelimb paralysis; and 5, moribund or dead. At 28 d after transfer, the recipients were sacrificed, and the lymphocytes were isolated from the CNS for analysis of the percentage of IFN-γ⁺ and IL-17⁺ T cells.

Prism v8.00 software (GraphPad Software, San Diego, CA) was used to generate graphs and for statistical analysis. Student unpaired t test was used for statistical analysis of the experiments in this study. All figures are presented as mean ± SEM. A p value <0.05 was defined as significant (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).

Previous report has demonstrated that LGALS9 is a risk gene of IBD (4). However, the modulatory role of LGALS9 in IBD pathogenesis is still unknown. We first sought to interrogate whether LGALS9 expression is increased within the GI tract during intestinal inflammation. Through analyses of published transcriptomic data of ileal biopsies from a pediatric cohort of newly diagnosed IBD patients (31), we found an enrichment of LGALS9 transcripts within inflamed intestine and a positive correlation between its expression level and disease severity (Fig. 1A, 1B). Interestingly, the expression of genes related to T cell activation and cytokines (CD4, CD44, IL17A, IL17F, IL6, IL1A, IL1B, TNF, and IFNG) were also increased in IBD patients, and the patterns were similar to LGALS9 (Fig. 1A, 1B), suggesting that T cells of IBD patients express higher level of LGALS9. To address this issue, we analyzed the correlation between LGALS9 and CD4, IL1R1, and IL6. In our results, LGALS9 expression level was positively correlated with these T cell–related markers (Fig. 1C). We further analyzed other published transcriptomic data of PBMCs from adult IBD patients (32)and revealed that the expression of LGALS9 was moderately increased in PBMCs from IBD patients (Fig. 1D), and its expression was positively correlated with the expression of CD4 (Fig. 1E). However, the expression levels of major cytokine genes (IL17A and IFNG) were indistinguishable between healthy control and IBD patients (Fig. 1D, 1F). These results could be explained by the immunosuppressive treatment in those patients (33–35). To further investigate whether an increased Lgals9 expression is also positively correlated with disease severity in mouse model, we analyzed the published transcriptomic data of colon from T cell–transferred recipients (36). The expressions of Lgals9 and T cell–related markers (Cd5, Cd3d, Cd44, Cd3g, Cd69, Cd4, Il21r, Il2rg, Il17ra, Il2rb, Tnf, Ifng, Il1b, Il2, Il17a, Il1a, and Il6) were also positively correlated with disease severity (Fig. 1G, 1H). Overall, only galectin-9 is significantly increased in intestines and/or PBMCs among all galectins in various transcriptomic databases, suggesting a modulatory role of this molecule in the pathogenesis of colitis.

To evaluate the potential contribution of galectin-9 on the development of T cell–mediated colitis, we transferred WT or Lgals9^−/− CD4⁺CD45RB^hi T cells into Rag1^−/− recipients (Fig. 2A). We found that intestinal inflammation, as shown by weight loss, colon length, colon weight, and histology, was less severe in recipients of Lgals9^−/− T cell transfer (Fig. 2B–H). Although the numbers of MLN cells were similar in WT and Lgals9^−/− recipients, the Lgals9^−/− group showed lower frequencies and numbers of T cells in the lamina propria (Fig. 2I, 2J). Moreover, the percentages of proliferating cells (Ki67⁺) and Th17 cells were lower in MLNs (Fig. 2K, 2L). These results demonstrate a positive role of galectin-9 in the development of colitis. However, previous reports have indicated that recombinant galectin-9 ameliorates autoimmune diseases by engaging different cell surface glycoproteins, including Tim-3, DR3, CD44, and 4-1BB (12, 14, 24–26). This discrepancy may be due to the differential modulatory roles of exogenous and endogenous galectin-9 in T cells or indirect effects by other galectin-9–acted cells such as dendritic cells (DCs), intestinal epithelial cells, or B cells.

Previous report has demonstrated that administration of galectin-9 in mice promotes thymocyte negative selection and T cell apoptosis (9). However, these exogenous and high-dose approaches may not elaborate the physiological roles of galectin-9 in T cell development and homeostasis. To address this issue, we compared thymocyte maturation and T cell development between WT and Lgals9^−/− mice. The size and cellularity of thymi during different stages did not differ between Lgals9^−/− mice and WT mice (Supplemental Fig. 1A–C), which indicated that galectin-9 is dispensable for thymocyte maturation. Moreover, the cell numbers in spleen and lymph nodes and percentages of different lymphocyte populations (CD4⁺, CD8⁺, and CD19⁺) were indistinguishable between Lgals9^−/− and WT mice (Supplemental Fig. 1D, 1E). Although previous reports have indicated that galectin-9 increases the population of adaptive Tregs (14, 15), the percentages of thymus-derived and peripherally derived Tregs in spleen and lymph nodes (Supplemental Fig. 1F) were similar between Lgals9^−/− and their control littermates. These results suggest that endogenous galectin-9 is not critical during the development of thymocytes and peripheral lymphocytes.

To investigate further whether endogenous galectin-9 modulates T cell homeostasis in an age-dependent manner, we analyzed the frequencies of naive (CD44^loCD62L^hi) and memory-like (CD44^hiCD62L^lo) T cells in WT and Lgals9^−/− mice at different ages. The age-dependent increase in the memory-like phenotype was attenuated in Lgals9^−/− T cells (Fig. 3A–C), which demonstrates that galectin-9 positively regulates the homeostasis of memory T cells in peripheral lymphoid organs.

Galectins usually colocalize with glycosylated surface molecules but are also detected in the cytosol and nucleus, where they interact with various regulators to modulate a variety of cell functions (16, 22, 37). However, the expression kinetics and pattern of galectin-9 in T cells are unclear. We analyzed the kinetics of galectin-9 RNA expression in resting and activated T cells. The expression of Lgals9 decreased significantly after anti-CD3/CD28 stimulation (Fig. 3D).

To investigate further whether galectin-9 is expressed inside or on the surface of T cells or is secreted by T cells, we used flow cytometric analysis to identify its surface and intracellular expression and ELISA to quantify its secreted form. Galectin-9 was barely detected on the T cell surface but was abundantly expressed in the cytosol of resting T cells. Similar to the RNA kinetics, the expression level of galectin-9 decreased at different time points after T cell activation (Fig. 3E–G). We also detected the amounts of galectin-1 and galectin-3 in the supernatants, because these two galectins have been reported to be expressed in T cells. In our results, galectin-1 and galectin-3 were abundantly detected in the supernatants. Again, galectin-9 was barely detected in the supernatants (Fig. 3H). These results suggest that endogenously produced galectin-9 can physiologically modulate T cell functions in an intracellular manner.

To evaluate galectin-9 localization in T cells further, we performed confocal microscopic analysis. Galectin-9 expression exhibited a disperse distribution within the cytosol of resting T cells (Fig. 3I, 3J). Interestingly, galectin-9 was recruited to immune synapses after anti-CD3/CD28 stimulation (Fig. 3I, 3J). These results suggest that galectin-9 may play a role in T cell activation.

Our confocal microscopic data led us to hypothesize that galectin-9 modulates TCR signaling and influences T cell functions. We next analyzed the cytokine production by CD4⁺ T cells from WT and Lgals9^−/− mice. CD3/CD28-stimulated Lgals9^−/− T cells secreted less IFN-γ, IL-17, and TNF-α than did WT T cells (Fig. 4A–C). Cell proliferation was significantly lower for Lgals9^−/− T cells than for WT T cells, as determined by ³H-thymidine incorporation assay (Fig. 4D) and CFSE-labeling cell-division assay (Fig. 4E).

To investigate further whether galectin-9 mediates T cell proliferation in an Ag-specific manner, WT or Lgals9^−/− 2D2 T cells were cocultured with BMDCs in the presence of MOG_35–55. Consistent with the results from anti-CD3/CD28 stimulation, Ag-specific T cell proliferation was diminished in Lgals9^−/− 2D2 T cells (Fig. 4F). Interestingly, Lgals9^−/− T cells proliferated similarly to WT T cells when stimulated with PMA and ionomycin (Fig. 4G), which suggests that galectin-9 modulates T cells in the proximal signaling stage. To examine this issue further, we analyzed the phosphorylation status of proximal signaling-related molecules from stimulated T cells. The phosphorylation of TCR-mediated Lck, Zap70, and PLCγ1 was attenuated in Lgals9^−/− T cells (Fig. 4H, Supplemental Fig. 2), and this subsequently impaired T cell proliferation.

Although we found that endogenous galectin-9 facilitates T cell activation, it was unclear whether the modulatory effect of galectin-9 in T cells occurs through an intrinsic or extrinsic manner. To examine this issue, we assessed cell division by coculturing CTR-labeled WT and CFSE-labeled Lgals9^−/− CD4⁺ T cells. The proliferative impairment of Lgals9^−/− T cells could not be rescued in cocultured WT T cells, which suggests that galectin-9 modulates T cell proliferation in an intracellular manner (Fig. 4I). Considering that WT T cells secrete a very low amount of galectin-9 (Fig. 3H), we next added recombinant galectin-9 in this assay. An addition of high level of exogenous galectin-9 increased apoptosis in T cells (Supplemental Fig. 3B), which is correlated with previous reports (12, 38, 39). Interestingly, the proliferation defect in Lgals9^−/− T cells could not be recovered by recombinant galectin-9 (Fig. 4J). This finding supports an intracellular role of galectin-9 in T cell activation and proliferation.

Previous reports have demonstrated that galectin-9 can interact with Tim-3 to trigger cell death (12, 38, 39). We found that Tim-3 was induced after TCR stimulation and that the Tim-3 levels were indistinguishable between WT and Lgals9^−/− T cells (Supplemental Fig. 3A). We also observed that the addition of exogenous galectin-9 increased apoptosis in both WT and Lgals9^−/− T cells (Supplemental Fig. 3B). However, it was unclear whether endogenous galectin-9 also participates in T cell apoptosis. To examine this issue, we analyzed activation-induced cell death without the addition of exogenous galectin-9 and found similar percentages of apoptotic cells in WT and Lgals9^−/− T cells (Supplemental Fig. 3C, 3D). These findings suggest that exogenous galectin-9 can trigger Tim-3–dependent, but endogenous galectin-9–independent, cell death. Endogenous galectin-9 appears to be directly involved in T cell activation and proliferation but may be dispensable for cell death.

In addition to the regulatory role of galectin-9 in T cells, recent reports have also indicated that addition of galectin-9 suppresses B cell signaling by binding to N-glycans of the BCR (40, 41). However, the role of endogenous galectin-9 in B cell development is unclear. The cellularity of B cells in bone marrows or spleens during the different developmental stages was similar in WT and Lgals9^−/− mice (Supplemental Fig. 4A). Interestingly, the IgA and IgG3 levels were significantly lower in Lgals9^−/− than in WT mice (Supplemental Fig. 4B), which suggests an endogenous role of galectin-9 in B cell function.

To investigate whether impairment of Ab production is affected by galectin-9 expression, we analyzed the galectin-9 RNA expression in naive and GC B cells. The expression of Lgals9 was downregulated in GC B cells compared with naive B cells (Fig. 5A). This activation-dependent decrease in galectin-9 expression in B cells is consistent with previous observation in T cells (Fig. 3D). To examine whether galectin-9 modulates the affinity maturation of Abs, we analyzed the production of high-affinity NP-specific IgG in NP-KLH–immunized Lgals9^−/− mice. The amounts of total and high-affinity NP-specific IgG in the serum were lower in Lgals9^−/− mice than in WT controls (Fig. 5B, 5C). Interestingly, the ratio of high-affinity/total NP-specific IgG was lower in Lgals9^−/− mice than in WT controls, especially on day 14 (Fig. 5D). Moreover, the number of GC B cells was lower in NP–KLH-immunized Lgals9^−/− mice than in controls (Fig. 5E). Taken together, these findings suggest a positive role of galectin-9 in affinity maturation.

We also evaluated B cell proliferation and apoptosis after immunization with SRBCs. B cell proliferation decreased after immunization in Lgals9^−/− mice (Fig. 5F). However, the percentage of apoptotic cells was similar between WT and Lgals9^−/− B cells (Fig. 5G). These results suggest that endogenous galectin-9 preferentially affects B cell proliferation, but not apoptosis, which was similar to our findings in T cells (Fig. 4D, Supplemental Fig. 3C, 3D). We next investigated whether impairment of B cell functions in Lgals9^−/− mice is B cell autonomous. Anti-IgM–stimulated B cell activation (Fig. 5H) and plasmablast populations (Fig. 5I) were similar between isolated WT and Lgals9^−/− B cells, which suggests that intrinsic galectin-9 may be dispensable in modulating B cell functions. Therefore, the functional impairment of B cells, in terms of cell division, GC B cell population, Ab production, and affinity maturation observed in vivo may relate to the defects in Lgals9^−/− T cells. This assumption is supported by the finding that the frequencies of CD4⁺ T cells and memory T cells decreased after NP–KLH immunization in Lgals9^−/− mice (Fig. 5J, 5K).

TCR activation and strength differentially modulate Th1, Th2, and Th17 cell differentiation (42–45). To investigate whether galectin-9 regulates the differentiation of Th cells, we polarized Lgals9^−/− T cells into different Th cells. Strikingly, Th17 polarization was significantly impaired in Lgals9^−/− T cells, compared with WT control cells (Fig. 5L), which suggested a positive role of galectin-9 in Th17 differentiation. To investigate further whether the defect in Th17 differentiation could be rescued by strong TCR signaling, we stimulated Lgals9^−/− T cells with higher anti-CD3 concentrations than those used normally in Th17 polarization. The increased TCR signaling rescued the defect of Lgals9^−/− Th17 differentiation (Fig. 5M). This observation supports the idea that Th17 differentiation can be positively sustained by galectin-9 in the presence of TCR signaling below a certain range. However, the percentages of Th1, Th2, and iTreg cells were similar between Lgals9^−/− and WT T cells, which suggests that galectin-9 is dispensable in the differentiation of Th cells with a lower TCR signaling requirement.

Our data showed that galectin-9 positively modulates T cell activation and Th17 differentiation, suggesting that it may also involve in the pathogenesis of Th17-mediated autoimmune diseases such as MS. We analyzed the published transcriptomic data of CD4⁺ T cells isolated from PBMCs of MS patients. Our results revealed that LGALS9 was highly expressed in patients, compared with healthy controls. Moreover, expression of T cell activation markers (CD44, CD5, and CD69), cytokine receptors (IFNGR1, IL13RA1, and IL17RA), and cytokine (IL2) were also increased in patients when compared with healthy controls (Fig. 6A, 6B). Overall, these results support a positive role of galectin-9 in the development of MS. To validate further whether the pathogenesis of EAE is mediated by endogenous galectin-9 in T cells, we transferred naive CD4⁺ T cells isolated from WT or Lgals9^−/− 2D2 TCR Tg mice into Rag1^−/− mice and then injected the mice with pertussis toxin (Fig. 6C). We found that the disease was ameliorated in recipients of Lgals9^−/− T cells (Fig. 6D, 6E). On day 27 after transfer, the mice were sacrificed for cell analysis. Consistent with our data for T cell–mediated colitis, the total T and IL-17⁺ T cell numbers in the inflamed CNS were significantly lower in Lgals9^−/−-transferred mice (Fig. 6F–H). These results indicate that endogenous galectin-9 potentiates T cell activation and differentiation and further involves in the pathogenesis of autoimmune diseases, such as colitis and EAE.

Our finding demonstrates an intracellular role of galectin-9 in T cell activation and proliferation. Moreover, the proliferation defect in Lgals9^−/− T cells could not be recovered by recombinant galectin-9 (Fig. 4J). However, a previous report has demonstrated that exogenous galectin-9 promotes Lck phosphorylation in Jurkat cells (46). There are two possible explanations for this discrepancy in TCR signaling: 1) the Jurkat human leukemia cell has an intrinsic mitogenic effect that may potentiate galectin-9–mediated activation, or 2) the potential effect of exogenous galectin-9 in primary T cells may be masked by strong anti-CD3/CD28 stimulation.

A previous report indicated that the BCR signaling was increased in anti-IgM–stimulated Lgals9^−/− B cells (40), which is different from our finding demonstrated that intrinsic galectin-9 may be dispensable in modulating B cell functions (Fig. 5H). This discrepancy implies that intrinsic galectin-9 may be involved in BCR signaling during early phase stimulation (40) and that stronger signaling may override the potential defect in Lgals9^−/− B cell activation (Fig. 5H).

Previous reports indicated that recombinant galectin-9 suppresses Th17 differentiation (26, 47). However, this exogenously provided dose may not represent the physiological condition and could mask the potential role of endogenous galectin-9 in Th17 differentiation. Recently, Liang et al. (48) used Lgals9^−/− mice to dissect the role of endogenous galectin-9 in mucosal immunity. They found that cholera toxin (CT)–immunized Lgals9^−/− mice developed severe diarrhea and that this could be modulated by impaired mucosal CT-specific IgA responses through a non–B cell–autonomous pathway. Moreover, the difference in mucosal IgA responses between immunized WT and Lgals9^−/− mice disappeared after in vivo IL-17 blockade, which suggests that galectin-9–mediated mucosal immunity is Th17 dependent (48). Our finding that galectin-9 modulates B cell responses through a T cell–autonomous manner (Fig. 5J, 5K) is consistent with the findings of Liang et al. (48). However, galectin-9 is expressed in various cell types, including T cells, B cells, DCs, fibroblasts, and epithelial cells. Whether the decrease in Th17 numbers in CT-immunized Lgals9^−/− mice is modulated by endogenous galectin-9 in T cells and/or the subsequent effects by other galectin-9-bearing cells remain to be elucidated.

Previous studies have reported that galectin-9 is highly expressed by iTregs, and it increases their functions and stability (14, 47). Both studies indicated that the impaired suppressive function of Lgals9^−/− iTregs could be rescued by exogenous galectin-9 treatment, which suggests an extracellular role of galectin-9 in iTreg. We found that galectin-9 was expressed mainly in the cytoplasm of T cells and that its secretion was very limited, which provides evidence of its endogenous role in T cell function. However, a deficiency of galectin-9 did not affect iTreg differentiation (Fig. 5L), which supports the idea that galectin-9 is dispensable for the induction of Tregs. Interestingly, we observed that suboptimal Th17 polarization by low-TCR signaling promoted iTreg differentiation in Lgals9^−/− T cells. This finding implies a dual role of galectin-9 in sustaining Th17 and suppressing iTreg differentiation in a low-TCR signaling–dependent manner.

Previous reports have indicated a divergent role of galectin-9 in Lgals9^−/− EAE models (13, 26). Oomizu et al. (26) reported that the disease was more severe in Lgals9^−/− mice. By contrast, Madireddi et al. (13) reported that Lgals9^−/− mice developed EAE with severity similarly to WT mice. They also found that the action of agonistic anti–4-1BB in suppressing EAE was galectin-9 dependent, because anti–4-1BB treatment was unable to suppress the disease in Lgals9^−/− mice. These controversial results may highlight a divergent role of galectin-9 in regulating various cell functions through different pathways. For example, our preliminary data revealed that Lgals9^−/− DCs have increased ability to promote T cell responses (data not shown). Therefore, the endogenous role of galectin-9 in T cells may be missed or masked in conventional Lgals9^−/− disease models.

Our findings have demonstrated that IFN-γ⁺IL-17⁺, but not IFN-γ^–IL-17⁺, T cells in the inflamed CNS were significantly lower in Lgals9^−/−-transferred mice. Although Th17 cells play major pathogenic role in EAE, several reports have demonstrated that Il17a^−/− or Il17f^−/− mice were only partially resistant to EAE pathogenesis (49, 50). Moreover, Th17 cells were plastic, which have been reported to acquire IFN-γ expression, and were expanded in the CNS during EAE that further induced CNS autoimmunity (51–53). Our results also indicated that these IFN-γ⁺IL-17⁺ T cells were accumulated in CNS, which actively participate the pathogenesis of EAE. Several reports have demonstrated that a transition from the Th17 cells into IFN-γ–producing cells were found in autoimmune disease, and these cells can promote disease pathogenicity (51, 54), suggesting that IFN-γ⁺IL-17⁺ T cells may further transit to Th1-like cells and promote the EAE severity.

Most studies have demonstrated that galectin-9 regulates cell responses through the CRDs and that lactose treatment can diminish these galectin-9–mediated effects (8, 55). A recent report indicated that galectin-9 binds to dectin-1 on macrophages through a glycan-independent pathway (56); this observation provides a striking insight into a divergent role of galectins. Similarly, Chen et al. (22) reported that intracellular galectin-3 interacts with Alix in T cells through short proline-rich motifs. In our results, intracellular galectin-9 was recruited to the immune synapse during T cell activation. This finding suggests that galectin-9 interacts with some activation-related molecules; however, the precise mechanism of action of endogenous galectin-9 in T cells is not clear. Nevertheless, our results demonstrate that the expression of galectin-9 in CD4⁺ T cells isolated from PBMCs is increased in patients and positively correlated with disease severity, providing a critical insight into the development of galectin-9–based biomarker for autoimmune diseases. In conclusion, we found that galectin-9 promotes proximal TCR signaling in T cells and that this affects Th cell differentiation and B cell responses. Endogenous galectin-9 in T cells appears to modulate positively the pathogenesis of some autoimmune diseases. Nevertheless, the possibility of its extracellular functions cannot be excluded. Further studies using conditional Lgals9^−/− mice are warranted to dissect the roles of galectin-9 in various cell types.

We thank James C. Paulson for providing the C57BL/6 Lgals9^−/− mice. We thank Jia-Ling Dong for technical support, and we also acknowledge the technical services provided by the Instrument Center of the National Defense Medical Center.

The authors have no financial conflicts of interest.