Treg-mediated Suppression of Inflammation Induced by DR3 Signaling is Dependent on Galectin-9

Abstract

Stimulation of several TNF receptor family proteins has been shown to dampen inflammatory disease in murine models through augmenting the number and/or activity of regulatory T cells (Treg). We recently found that one molecule, 4-1BB, utilized binding to Galectin-9 to exert its immunosuppressive effects and drive expansion of CD8⁺Foxp3⁻ Treg. We now show that ligation of another TNFR family molecule, DR3, which has previously been found to strongly expand CD4⁺Foxp3⁺ Treg and suppress inflammation, also requires Galectin-9. We found that the extracellular region of DR3 directly binds to Galectin-9, and that Galectin-9 associates with DR3 in Treg. From studies in vitro with Galectin-9^−/− CD4⁺ T cells and Treg, we found that stimulatory activity induced by ligating DR3 was in part dependent on Galectin-9. In vivo, in a model of EAE we show that an agonist of DR3 suppressed disease, correlating with expansion of CD4⁺Foxp3⁺ Treg cells, and this protective effect was lost in Galectin-9^−/− mice. Similar results were seen in an allergic lung inflammation model. Thus, we demonstrate a novel function of Galectin-9 in facilitating activity of DR3 related to Treg-mediated suppression.

INTRODUCTION

Molecules in the TNF/TNFR superfamily (SF) are of great interest to suppression of inflammatory and autoimmune disease as well as for promoting immune responses against infectious pathogens and cancer cells. General concepts that have emerged over the past 10–15 years are that neutralizing the interaction of a number of TNF family ligands with their receptors may prevent or reduce inflammatory T cell-mediated immune responses, whereas agonist stimulation of some of the TNF family receptors, including OX40 (TNFRSF4), 4-1BB (TNFRSF9), and GITR (TNFRSF18), can expand effector T cell populations that can be protective against viruses and growth of tumors (1, 2). Interestingly, stimulation of the aforementioned receptors, as well as TNFR2 (TNFRSF1b) and DR3 (TNFRSF25) (3, 4), has also been shown in a number of inflammatory models to lead to suppression of disease symptoms, apparently at odds with the latter activity. However, these phenomena have been explained by the fact that regulatory T cells are also capable of expressing the TNFR molecules, and the immunosuppression that can result in certain scenarios when these molecules are engaged can be attributed to driving the expansion or activity of different populations of regulatory T cells (Treg) in preference to promoting the activity of effector T cells. For example, agonist antibodies to 4-1BB can expand CD8⁺ effector CTL that can be protective against viruses and a variety of tumors, but by inducing the expansion of CD8⁺ Treg they can also suppress clinical symptoms in collagen-induced arthritis (CIA), experimental autoimmune encephalomyelitis (EAE), and other disease models (5, 6). The requirements for augmenting Treg activity are largely thought to reflect conventional agonist signaling through the TNF family receptors, although it is possible that additional factors may be required for exerting the effect on Treg. In this regard, we recently showed that Galectin-9, a member of the beta-galactoside binding family of lectins, was critical for the ability of agonist antibodies to 4-1BB to augment the accumulation in vivo of the CD8⁺ Treg that suppress inflammatory disease (7). We found that Galectin-9 bound, in a carbohydrate-dependent manner, the extracellular region of 4-1BB, and we proposed that Galectin-9 aggregated or cross-linked 4-1BB monomers on the surface of cells and facilitated the ability of these molecules to signal and to drive Treg activity. Whether this mechanism might be operational in allowing other TNF receptor family molecules to promote Treg-mediated suppression is not known.

Here we show that the extracellular region of DR3 also binds Galectin-9 and stimulatory reagents against DR3 are reliant on Galectin-9 for suppressing inflammatory disease in vivo. Agonists to DR3 can promote the expansion of CD4⁺ Foxp3⁺ Treg and have been found to lead to suppression of allergic lung inflammation, transplant rejection, and virus-induced keratitis (8–11). We now demonstrate that stimulating DR3 can additionally lead to resolution of EAE, and that Galectin-9 plays a role in the expansion and activity of Foxp3⁺ Treg, such that a deficiency in Galectin-9 prevents an agonist antibody to DR3 from limiting EAE as well as inflammation in a model of asthma. The interaction of Galectin-9 with DR3 is then a previously unknown immunoregulatory checkpoint of significance to immune activity and immune disease.

Methods

Mice

Gal-9^−/− mice, backcrossed onto the C57BL/6 background (n > 9), were originally provided by GalPharma (12). Mice bred at the La Jolla Institute for Allergy and Immunology and WT mice were used as before (7). All experiments were in compliance with the regulations of the La Jolla Institute for Allergy and Immunology animal care committee.

Antibodies and Flow Cytometry

Agonist anti-DR3 (4C12) and TL1A.Ig were previously described (8, 11). Armenian Hamster IgG was purchased from Biolegend. Antibodies used for flow cytometry were: PE-conjugated anti-Gal-9 (108A2), pacific blue-conjugated anti-CD3 (145-2C11) and anti-CD4 (GK1.5), and PE-conjugated anti-Siglec-F (E5-2440), all from BD Biosciences, and PE-conjugated anti-DR3 from Biolegend. In some cases, cells were preincubated with anti-mouse CD16/CD32 (2.4G2; 10 μg/ml) to block FcγR and stained with pacific blue anti-CD4, fixed with Cytofix/Cytoperm (BD Biosciences), and then stained with PerCP-Cy5.5 anti-IL-17A, PE-Cy7 anti-IFN-γ, PE-Cy7 anti-TNF, PE anti-IL-10, PerCP-Cy5.5 anti-Foxp3 (BD Biosciences) or Alexa Fluor anti-IDO (Biolegend). Samples were analyzed after gating on CD4⁺ T cells on a LSR-II flow cytometer (BD Biosciences) with FlowJo software (TreeStar).

Fc fusion proteins, TL1A, and Galectin-9

4-1BB.Fc was produced as described previously (7) and mouse and human DR3.Fc were from either Enzo Life Sciences or R&D Systems. Fc fusion proteins made in-house were expressed using Freestyle HEK 293F expression system (Invitrogen) according to the supplier’s protocol and purified by protein G affinity chromatography. For carbohydrate-mediated binding studies, DR3.Fc was treated with PNGase F enzyme (New England BioLabs) as before (7). Recombinant human and mouse Galectin-9 were from Galpharma Co. Ltd., (Kagawa, Japan) (7). Human and mouse Galectin-9 have observed molecular masses of approximately 36 kDa. Human and mouse TL1A were from R&D Systems.

Binding Assays and Immunoprecipitation

For precipitation binding assays, saturating amounts of human and mouse 4-1BB.Fc or DR3.Fc were coated on to protein G beads and incubated with human or mouse Galectin-9. Co-precipitation was detected by SDS-PAGE (reducing condition) by eluting the proteins from the beads. For competition experiments, Galectin-9 or TL1A were coated onto ELISA plates and binding of DR3.Fc was detected using Biotin anti-DR3 and Streptavidin-HRP similar to previous studies (7). Competition was assessed with DR3.Fc first incubated with TL1A or Galectin-9.

For binding affinity measurements, a Biacore3000 instrument (GE Healthcare) was used as before (7), with anti-human IgG (human antibody capture kit; GE Healthcare) amine coupled to a CM5 sensor chip. A human IgG (Fc fragment) or irrelevant Fc fusion protein was used as a negative control for non-specific binding. Kinetic parameters were calculated using a simple Langmuir 1:1 model with the BIA evaluation software version 4.1.

For cell immunoprecipitation, Treg were purified by cell sorting as CD4⁺CD25⁺ (>96% Foxp3⁺). Lysates were incubated with TL1A.Ig fusion protein bound to protein G Dynabeads (Invitrogen) to precipitate DR3 as performed previously (7). After SDS-PAGE, the proteins were visualized by western blotting with anti-mouse DR3 (R&D Systems) or anti-mouse Galectin-9 (BioLegend) followed by anti-Rat IgG light chain-specific-HRP (Jackson ImmunoResearch).

In vitro T cell cultures

Naïve CD4⁺ splenic T cells or CD4⁺ Treg were purified by cell sorting by gating cells as CD25⁻CD44^loCD62L^hi or CD4⁺CD25⁺, respectively. Naïve T cells (5 x 10⁴/well) were pre-activated with plate-bound anti-CD3 (2C11; 2 μg/ml) and soluble anti-CD28 (37N5; 0.5 μg/ml), with anti-DR3 or TL1A.Ig (10 or 1 μg/ml) or control rat or human IgG (10 or 1 μg/ml) for 48 h. To determine secondary responsiveness, the activated T cells were sorted for identical expression of DR3 and re-cultured with plate-bound anti-CD3 (0.5 μg/ml) in the presence of IgG (1 μg/ml) or TL1A.Ig (1 μg/ml) for another 24–48 h. Culture supernatants were assayed by ELISA with antibodies to IL-2 (JES6-1A12 and biotin-JES6-5H4) and IFN-γ (R46A2 and biotin-XMG1.2; all from BD Biosciences). Treg were stimulated at 5 ×10⁴/well with anti-CD3 (2 μg/ml), IL-2 (10 U/ml) and anti-DR3 or TL1A.Ig (10 or 1 μg/ml) and proliferation ([³H]-thymidine incorporation) or intracellular cytokines assessed after 72 or 48h.

EAE

Mice were immunized as described previously (7) by s.c. injection at the base of the tail with 100 μg of MOG_33–55 peptide (AnaSpec) in CFA (Life Technologies, Gaithersburg, MD) containing 2 mg/ml Mycobacterium tuberculosis H37 RA (Difco, Detroit, MI). The mice also received i.v. injections of 200 ng pertussis toxin (List Biological Laboratories, Campbell, CA) on days 0 and 2 and were administered i.p. 10 μg anti-DR3 or control antibody (rat IgG) on day 0. Individual animals were scored daily for clinical signs of EAE using the following criteria: 0, no detectable signs of disease; 0.5, distal limp tail; 1, complete limp tail; 1.5, limp tail and hind limb weakness; 2, unilateral partial hind paralysis; 2.5, bilateral partial hind limb paralysis; 3, complete bilateral hind limb paralysis; 3.5, complete bilateral hind limb paralysis and unilateral forelimb; 4, total paralysis of fore and hind limbs; and 5, death. Intracellular staining of T cells was performed as before (7) after stimulation with MOG_33–55 peptide (200 μg/ml) for 72h.

Allergic Lung Inflammation

Mice were immunized i.p. with OVA (20 μg; Sigma-Aldrich) in Alum (Fischer Scientific International) on days 1 and 14, followed by 20 μg OVA in PBS given intranasally on days 14–16 (7). Anti-DR3 (10 μg) or control antibody was injected i.p. 1 day before the initial OVA injection. For collection of BAL fluid, the lungs were lavaged 6 times with 800 μl PBS (2% BSA). Eosinophils were determined using flow cytometry with PE-conjugated Siglec-F and FITC-conjugated CD11c (eosinophils were identified as Siglec-F⁺ CD11c⁻). IL-5 was determined by ELISA (BD Biosciences).

Histological analysis

Lungs from asthma-induced mice, perfused with 4% paraformaldehyde (PFA), were dissected, fixed in 4% PFA, and paraffin embedded, and sections (6 μm) were stained with hematoxylin and eosin (H&E). For inflammation scoring in lungs, peribronchial regions (6–8 per mouse) were evaluated at 200× and inflammatory infiltrates around the airways were graded for severity (0, normal; 1, <3 cell diameter thick; 2, 3–10 cells thick; 3, >10 cells thick) and extent (0, normal; 1, <10% of sample; 2, 10–25%; 3, >25%). Scores were calculated by multiplying severity by extent (max 9).

Statistics

Two-tailed Student’s t test was used to determine the statistical significance of differences between groups. P values of less than 0.05 are marked with an asterisk and were considered statistically significant. n.s. non-significant.

RESULTS

Galectin-9 is a binding partner for DR3

We previously found that the extracellular region of one member of the TNFR family, 4-1BB, bound Galectin-9 in a carbohydrate dependent manner, and that several activities induced by ligating 4-1BB were dependent on Galectin-9 (7). We suggested that Galectin-9 acted like a type of co-receptor for 4-1BB monomers, enhancing their ability to cluster or aggregate together to maximize the signal induced by 4-1BB. In a search for other TNFR family glycoproteins that might similarly be regulated by Galectin-9, we discovered an interaction with DR3 (TNFRSF25), whereas several other molecules tested did not show any binding activity (OX40, GITR, HVEM; data not shown). The extracellular portion of both human and mouse DR3 comprises four cysteine-rich domains, and it has a potential N-linked glycosylation site in each of the two most membrane distal domains (CRD1 and CRD2) that could be amenable for binding by Galectin-9. To initially show binding, we used an assay where recombinant human or mouse Galectin-9 was immunoprecipitated with a fusion protein containing the extracellular region of human DR3 (DR3.Fc). Another member of the Galectin family (Galectin-3) and human 4-1BB.Fc (positive control) or Fc protein (negative control) were included for specificity (Fig. 1a).

Figure 1.

Galectin-9 binds to DR3. (a) Immunoprecipitation of recombinant human or mouse Galectin-9 with human DR3.Fc or 4-1BB.Fc. Control, human Galectin-3. (b) Immunoprecipitation of recombinant human or mouse Galectin-9 with mouse DR3.Fc. Observed molecular masses were h/m4-1BB.Fc or DR3.Fc – 50–58 kDa, m/hGalectin-9 – 36 kDa. Data in (a–b) are representative of at least three different experiments each. (c–f) Binding response of increasing concentrations (0.48–2000 nM) of hGalectin-9 (c, d) or mGalectin-9 (e, f) to immobilized hDR3.Fc (c, e) or mDR3.Fc (d, f), measured by SPR. Values represent average of three independent measurements. The response shown is reference-subtracted (unrelated Fc protein). (g–h) Human or mouse Galectin-9 or TL1A were coated onto ELISA plates and binding of WT or de-glycosylated DR3.Fc was detected using biotin anti-DR3 and streptavidin HRP. Competition was assessed with DR3.Fc first incubated with either TL1A or Galectin-9. Data are means ± s.e.m binding from triplicate cultures and representative of three different experiments.

To determine whether this interaction is conserved across species, we performed co-immunoprecipitation experiments with mouse DR3.Fc, and found it also was able to precipitate both human and mouse Galectin-9 (Fig. 1b). We then assessed the dynamics of binding of human and mouse DR3 with human and mouse Galectin-9 using surface plasmon resonance (SPR). The human and mouse Galectin-9 interactions were found to exhibit equilibrium binding constants (K_D) of ~430 nM and ~440 nM with human DR3, respectively (Fig. 1c, e), and ~180 nM and ~360 nM for mouse DR3, respectively (Fig. 1d, f).

In line with Galectin-9 binding being carbohydrate-dependent, PNGase F treatment of human and mouse DR3.Fc to remove N-linked sugars resulted in greatly reduced binding activity to both human and mouse Gal-9 (Fig. 1g, h). In contrast, TL1A, the TNF family ligand of DR3, equivalently bound to DR3 regardless of the presence of N-linked sugars. We lastly tested whether TL1A and Galectin-9 competed for binding to DR3. Human or mouse DR3.Fc were pre-incubated with saturating concentrations of either human or mouse TL1A, or Galectin-9, before assessing binding to Galectin-9 or TL1A, respectively (Fig. 1g, h). In either case, DR3 binding to the alternate ligand was observed, suggesting the two molecules do not compete for the same binding site.

DR3 activity in T cells and Treg cells is impaired in the absence of Galectin-9

DR3 is expressed in conventional CD4⁺ T cells and CD4⁺Foxp3⁺ Treg cells and acts as a stimulatory receptor when ligated. We previously showed that Galectin-9 was expressed at low levels by conventional T cells (7) and so could be available to bind DR3. To then initially assess whether the interaction between DR3 and Galectin-9 was important functionally, we stimulated WT and Galectin-9^−/− naïve CD4 T cells with anti-CD3 in the presence of agonist TL1A.Ig that can trigger DR3 signaling. Production of both IL-2 and IFN-γ induced by DR3 activity were significantly impaired in the absence of Galectin-9, although IL-2 was affected to a greater extent (Fig. 2a). Similarly, production of these cytokines was impaired when Galectin-9^−/− T cells were pre-activated for several days with anti-CD3 before being stimulated with an agonist antibody to DR3 (Fig. 2b). Galectin-9^−/− T cells were not intrinsically impaired, as we demonstrated previously (7), in that the response to anti-CD3 in the absence of TL1A.Ig or anti-DR3 was normal (Fig. 2a/b). Interestingly, and in line with our prior observations of lower 4-1BB expression in the absence of Galectin-9 (7), the level of DR3 expressed in Galectin-9^−/− T cells was reduced compared to WT T cells, suggesting that Galectin-9 can impact the stability and turnover of this receptor on the membrane (Fig. 2c). The expression of several other TNFRSFs (OX40, GITR) as well as non-TNFRSFs (CTLA-4) was not affected in the absence of Gal-9 (Supp. Fig. 1 and see ref (7)), which at least in the case of these TNFRSFs correlated with a lack of binding to Galectin-9 (unpublished data). To determine if the reduced expression of DR3 accounted for the impaired function of DR3, T cells were pre-activated and then sorted based on equivalent levels of DR3. This again revealed an impaired responsiveness to TL1A.Ig in the absence of Galectin-9 (Fig. 2c). These data suggest that Galectin-9 not only controls membrane expression of DR3 but also facilitates the activity of DR3 molecules when they are ligated, similar to our prior results with 4-1BB.

Figure 2.

Defective DR3 activity in T cells from Galectin-9^−/− mice. (a) Naïve CD4⁺ T cells from WT or Galectin-9^−/− mice were activated in vitro with anti-CD3 and anti-CD28 in the presence of control IgG and agonist TL1A.Ig. IL-2 and IFN-γ production were assessed after 48 h. (b) CD4⁺ T cells from WT and Galectin-9^−/− mice were pre-activated with anti-CD3 and anti-CD28, and then after 48 h were restimulated with anti-CD3 in the presence of control IgG or agonist anti-DR3. IL-2 and IFN-γ production were assessed 48 h later. (c) CD4⁺ T cells from WT (solid line) and Galectin-9−/− (dotted line) mice were pre-activated as in (b) and stained for DR3 expression (pre-sort, left). Isotype control staining, shaded. Sorted cells were then restimulated with anti-CD3 in the presence of control IgG or agonist TL1A.Ig. IL-2 production was assessed at 24 h. All data are means ± s.e.m from triplicate cultures and representative of three different experiments. *p<0.05

To extend these results we assessed the activity of DR3 in CD4+Foxp3+ Treg. Intracellular flow cytometry showed that Galectin-9 was strongly expressed in these cells, and DR3 was visualized on the surface of Treg (Fig. 3a). Similar to our prior observations of other T cells (7), we did not detect Galectin-9 on the Treg cell surface by flow cytometry. This may reflect that Galectin-9 is bound to cell surface proteins and cannot be seen by the flow antibodies. In line with this, our prior results showed that membrane localized Galectin-9 could be visualized by confocal microscopy (7). Importantly, showing that Galectin-9 and DR3 could associate together in Treg, immunoprecipitation of DR3 with TL1A.Ig resulted in the pull-down of Galectin-9 (Fig. 3b). To preliminarily assess functionality, Treg were stimulated with anti-CD3 and IL-2 in the presence or absence of TL1A.Ig. Treg from Galectin-9^−/− mice were not intrinsically impaired as they proliferated normally to anti-CD3 and IL-2. However, ligation of DR3 enhanced proliferation of WT Treg whereas this activity was significantly lower in Galectin-9^−/− Treg (Fig. 3c). We then assessed whether the stimulated Treg could suppress naïve T cells in vitro in co-cultures, and found no defect comparing WT cells to those deficient in Galectin-9, regardless of stimulation via DR3 (not shown). As in vitro suppression assays are well known to be highly variable, stimulation condition-dependent, and not necessarily reflective of all the suppressive capabilities of Treg, we assessed production of several inhibitory molecules. Interestingly, intracellular flow cytometry revealed that ligation of DR3 could upregulate IDO and IL-10 in WT Treg, but these activities were reduced in Galectin-9^−/− Treg (Fig. 3d). These data suggest that Galectin-9 is at least partially required for several activities of DR3 in T cells and Treg.

Figure 3.

Treg from Galectin-9^−/− mice are refractory to stimulation through DR3. (a) CD4⁺Foxp3⁺ Treg cells from WT mice were stained for surface DR3 and intracellular Galectin-9. Shaded, isotype control. Dotted line, Treg from Galectin-9^−/− mice. (b) Immunoprecipitation with TL1A.Ig or control Fc was performed on lysates from WT Treg cells. Immunoblotting was carried out with anti-DR3 (top) or anti-Galectin-9 (bottom). MW indicated. Data are representative of three experiments. (c) Treg cells from WT or Galectin-9^−/− mice were stimulated with anti-CD3 and IL-2 in the presence of control IgG or agonist TL1A.Ig. Proliferation was assessed at 72 h. Data are means ± s.e.m from triplicate cultures and representative of three experiments. *Significance TL1A.Ig WT vs. Galectin-9^−/−. (d) WT (solid line) and Galectin-9^−/− (dotted line) Treg cells were stimulated with anti-CD3 and IL-2 in the presence of control IgG or agonist anti-DR3. After 48 h, cells were stained for intracellular IDO and IL-10. MFI indicated. Isotype controls, shaded. Data are representative of three different experiments. *p<0.05

DR3 triggering is unable to ameliorate inflammatory disease in Gal-9^−/− mice

To further these results and focus on an in vivo activity of DR3, we assessed whether Galectin-9 was required for the immunosuppressive effect of agonist reagents to DR3 in inflammatory disease. The Th17 response that underlies the neuro-inflammatory disease in EAE can be highly controlled by Treg, with these cells primarily promoting resolution of disease (13, 14). As agonist anti-DR3 has been shown to transiently but strongly enhance the accumulation of CD4⁺Foxp3⁺ Treg when injected in vivo (8), we asked whether it could suppress demyelinating disease when injected at the time of immunization with MOG peptide. As reported previously, anti-DR3 strongly enhanced the percentage of Treg in the blood, and in line with our in vitro data above, approximately 50% fewer Treg were expanded in mice lacking Galectin-9 (Fig. 4a). Most impressively, anti-DR3 promoted resolution of the disease in WT mice, whereas essentially no effect on clinical symptoms was seen in Galectin-9^−/− mice (Fig. 4b). The knockout animals developed EAE similarly to WT mice when injected with control IgG showing there was no intrinsic defect in mounting this response when Galectin-9 was not expressed. Intracellular staining for IL-17A and TNF in CD4 T cells from draining LNs also revealed a suppressed T cell response in WT mice that received anti-DR3, but not in Galectin-9^−/− mice (Fig. 4c). Lastly, it has been suggested that IL-10 produced by Treg can be important for suppression of EAE (15–17). We observed greater percentages of IL-10⁺Foxp3⁺ Treg as well as IL-10^-Foxp3⁺ Treg in the brains of WT mice injected with anti-DR3 than in the brains of Galectin-9^−/− mice (Fig. 4d). This lack of an effect of anti-DR3 on Treg in Galectin-9^−/− mice correlated with enhanced percentages of pathogenic IL-17⁺ and IFN-γ⁺ CD4 T cells being visualized compared to WT mice receiving anti-DR3 (Fig. 4e), and was in line with the inability of anti-DR3 to suppress disease in the Galectin-9 deficient animals.

Figure 4.

Galectin-9 is required for suppression of EAE by anti-DR3. WT or Galectin-9^−/− mice were immunized with MOG_35-55 peptide and injected with either control IgG or agonist anti-DR3. (a) Percent Foxp3⁺CD4⁺ T cells in peripheral blood 5 days after the last injection of anti-DR3. (b) EAE clinical scores over time. (c) Frequencies of IL-17A⁺ CD4 T cells (left) and TNF⁺ CD4 T cells (middle and right) in draining lymph nodes on days 17 and 30, respectively. (d) Proportion of gated CD4⁺ cells expressing IL-10 and Foxp3 (left) and total numbers of Foxp3⁺IL-10⁺ CD4 T cells (right) in brains of anti-DR3-treated animals at day 17. (e) Proportion of gated CD4⁺ cells expressing IL-17 or IFN-γ (left) and total numbers of IL-17⁺ CD4 T cells (right) in brains of anti-DR3-treated animals at day 17. All data are either representative or means ± sem from five mice per group. Similar results in three different experiments. *p<0.05

We lastly investigated the suppressive ability of anti-DR3 in an allergic asthma model and whether Galectin-9 played a role. Administration of anti-DR3 at the time of immunization with antigen downregulated airway inflammation in WT mice, as shown by histologic evaluation of lung sections (Fig. 5a), and reduced eosinophilia and IL-5 production in the bronchoalveolar lavage (BAL) fluid (Fig. 5b). In contrast, anti-DR3 did not suppress inflammation, eosinophilia, or IL-5 below levels seen in Galectin-9^−/− mice injected with control IgG (Fig. 5a, b). Unexpectedly, the agonist of DR3 exacerbated eosinophilia and IL-5 to an extent in Galectin-9^−/− mice, however the explanation for this is not obvious. In line with suppression of disease in WT mice but not in Galectin-9^−/− mice, anti-DR3 only led to enhanced accumulation of CD4⁺Foxp3⁺ Treg in the BAL and draining lymph nodes of WT mice whereas no effect on Treg was seen in the absence of Galectin-9 (Fig. 5c). Therefore, collectively these data show that Galectin-9 plays an essential role in the immunosuppressive Treg response that is elicited when DR3 is engaged with agonist reagents.

Figure 5.

Galectin-9 is required for suppression of allergic asthma by anti-DR3. WT and Galectin-9^−/− mice were immunized with OVA to induce lung inflammation, and injected with IgG or agonist anti-DR3. (a) Representative H&E staining of lung sections (left), and mean inflammation score (right). (b) Eosinophil numbers and IL-5 expression in BAL. (c) Proportion of CD4⁺Foxp3⁺ T cells in BAL (left) and draining lymph nodes (right). All results are means ± s.e.m from five mice per group, and representative of three independent experiments. *p<0.05

DISCUSSION

In summary we describe a new function for Galectin-9 in binding and facilitating the activity of DR3 with respect to promoting Treg function that limits inflammatory disease. These results highlight a new theme within the TNFR superfamily that first emerged with our prior studies of 4-1BB (7). They suggest that Galectin-9 can play a key role in shaping the maximal activity of some TNF protein receptors in Tregs, as well as conventional T cells, that might be exploited at some point in the future for both positively and negatively manipulating the immune response.

Galectins are carbohydrate-binding proteins, and through their carbohydrate recognition domains they interact with oligosaccharide side chains of glycoproteins in a relatively uncharacterized manner that has however been postulated to involve strong specificity for certain branched glycans and particular motifs displayed on these glycans (18, 19). This implies that the individual Galectin molecules will not bind to every glycoprotein, but which glycoproteins each molecule can bind, especially in the immune system, and which immune glycoproteins are modulated by Galectin interactions, is still largely unappreciated. We tested several other recombinant TNFRSF members (OX40, GITR, HVEM) and did not find an interaction with Galectin-9. Thus, although our data now show that Galectin-9 can bind to 4-1BB and DR3 in the TNFR superfamily, it remains to be determined how many other similar TNF receptors may complex with Galectin-9. Binding is likely highly complex, involving unique arrangements of branched sugars, therefore without knowledge of the exact specification of the sugars required to interact with Galectin-9 and the sugars displayed on TNFRSF or other molecules, it is not predictable which proteins might form complexes with Galectin-9.

For some time, the primary molecule in the immune system thought to bind, and be controlled by, Galectin-9 was the inhibitory receptor Tim-3 whose expression is quite restricted in T cells, largely being found on Th1 cells (20). However, Galectin-9 has also been reported to bind CD44 that is broadly expressed on most activated or memory T cells including Treg (21), in addition to 4-1BB and now DR3 that are similarly found on activated conventional T cells, Treg, as well as on other cells such as NK cells and some dendritic cells. In the case of Tim-3, it has been widely discussed that Galectin-9 is a ligand, in the sense of how a soluble cytokine or a molecule expressed in trans on the membrane of a neighboring cell can be a ligand for a receptor. But Galectin-9 is a tandem-repeat molecule possessing two carbohydrate recognition domains, termed N- and C-, that have 39% amino acid identity to each other (22), and each domain has been suggested to possess high affinity for both specific overlapping as well as unique branched bi-, tri-, and tetra-antennary N-linked glycans with N-acetyllactosamine motifs (23, 24). This suggests that one molecule of Galectin-9 has the potential to interact with a single binding partner or alternate partners in a divalent fashion, implying its true function is likely not as a conventional ligand for a receptor. In addition, the C-terminal domain of Galectin-9 has the ability to self-associate (25). This further implies that one or two molecules of Galectin-9 could cross-link several identical, or different, cell surface receptors in a lattice-like manner, as has been shown for other Galectins such as Galectin-3 (18, 26). Our prior data on 4-1BB (7), and new data here on DR3, further point to the ability to link cell surface proteins, and in some manner aggregate them or stabilize their expression, as being a primary function of Galectin-9.

In line with this notion, we found reduced levels of DR3 on T cells that lacked Galectin-9, highly similar to our observation of reduced expression of 4-1BB in the absence of endogenous production of Galectin-9 by T cells. Thus, at one level the function of Galectin-9 in the biology of these two TNFR family molecules may be in slowing the endocytosis or turnover of these proteins from the surface of T cells. Consistent with this type of activity, Galectin-9 has also been reported to co-localize with glucose transporter 2 in pancreatic β cells with the suggestion that it influences the rate of Glut2 endocytosis (27). Galectin-9 was furthermore suggested to complex with protein disulfide isomerase increasing the surface expression of this molecule on cells (28). However, our data here with DR3 and previously with 4-1BB, also point to an additional activity. When T cells were assessed that had equivalent levels of DR3, we found that the absence of Galectin-9 still resulted in defective functional outcomes when DR3 was ligated. This furthers the notion previously put forward for 4-1BB (7) that the second important function of Galectin-9 is to organize molecules on the cell surface in a manner in which they can signal effectively. Whether this means Galectin-9 facilitates dimerization, trimerization, or oligomerization of DR3 monomers, and how this will relate to the quality of the intracellular signal transmitted through DR3, is not clear at present, and will need to be investigated in the future.

An outstanding question is also whether endogenously produced Galectin-9 is normally required for the function of DR3 (or 4-1BB) or if its activity is only apparent under particular situations, for example when these receptors are engaged with agonist antibodies. While we do effectively show that the activity of DR3, or in our prior report 4-1BB, was impaired in T cells when ligated by their natural ligands in vitro, it is possible that in vivo the natural ligands are presented in a way, such as clustered on an APCs cell surface, that they effectively aggregate or cross-link the receptors in a manner that makes Galectin-9 unimportant. This conclusion is supported by a prior report of DR3-deficient mice (29) that found strongly impaired lung inflammation and EAE in models similar to those used in our experiments whereas our data show that unmanipulated Galectin-9-deficient animals mounted respectable responses in these models. Thus, the role of Galectin-9 may be primarily highlighted with agonist TNFRSF reagents which do not mimic the extent of aggregation or cross-linking that may occur on cells expressing the natural ligands.

Our results also have implications for prior literature that relate to the activity of recombinant Galectin-9. Endogenously produced Galectin-9 is thought to act primarily as a cell surface associated protein or intracellularly, and circulating levels of soluble Galectin-9 are generally low unless an inflammatory response is occurring. Nevertheless, studies of soluble recombinant Galectin-9 injected into animals have shown it is highly inhibitory and may have clinical potential to block inflammation (30). Soluble recombinant Galectin-9 can suppress inflammatory disease in models of EAE, arthritis, and diabetes (12, 20, 31–33), and it has been suggested to do this by either promoting the accumulation and activity of Tregs (12, 34–40) or leading to apoptosis or death in some effector T cells (36, 41–43). A major question is what are the glycoprotein receptors that recombinant Galectin-9 engages to exert these actions. It was initially thought that Tim-3 could be the primary target as it is an inhibitory molecule, but Tim-3 independent suppression has been reported. A more recent paper suggested that recombinant Galectin-9 could promote the differentiation of naïve T cells into CD4⁺Foxp3⁺ Treg through interaction with CD44, implying that CD44 could also be a target of recombinant Galectin-9. However, no in vivo relevance of an interaction with CD44 was demonstrated (40). Our data here with DR3, as well as prior data with 4-1BB, now raise the possibility that these molecules could additionally be engaged by recombinant Galectin-9 on mature Treg (peripheral or thymic) and that they might be instrumental in the immunosuppressive activity of recombinant Galectin-9. Interestingly, one paper administered recombinant Galectin-9 along with an agonist antibody to DR3 and found together they enhanced accumulation of CD4⁺Foxp3⁺ Treg and afforded greater protection of mice against HSV-mediated stromal keratitis compared to either reagent alone (44). This study interpreted their results as two separate activities of anti-DR3 and Galectin-9, whereas our data lead to the hypothesis that Galectin-9 may have allowed greater activity of anti-DR3 by promoting more DR3 monomers to aggregate and enhancing the signal transmitted by DR3 to promote Treg activity. Future studies in this area will be needed to determine how many glycoproteins are indeed engaged by soluble recombinant Galectin-9, whether extensive cross-talk occurs between the currently known binding partners of Galectin-9, and whether activity in Treg is substantially similar or dissimilar to any activities elicited in conventional T cells.

In conclusion, we show that Galectin-9 is a new DR3-associating protein and a regulator of the ability of DR3 to be functional in Treg cells as well as conventional T cells. Conserved binding activity between mouse and human proteins suggests this will be an important interaction that regulates immune function. Given the potential impact of targeting DR3 and Galectin-9 therapeutically, a greater understanding of this interaction may be useful for manipulating immune function-related diseases.