Galectin-1–Driven Tolerogenic Programs Aggravate Yersinia enterocolitica Infection by Repressing Antibacterial Immunity

Yersinia enterocolitica is a Gram-negative enteropathogenic bacterium that causes human gastroenteritis, as well as extraintestinal manifestations, such as reactive arthritis and sepsis (1). During the course of infection, Y. enterocolitica reaches the intestinal tract, invades M cells of Peyer’s patches (PPs), and disseminates to the mesenteric lymph nodes (MLNs), spleen, and liver (1). Although innate immune cells, including NK cells, dendritic cells (DCs), and macrophages, confer early protection during the course of Y. enterocolitica infection, Th1 and Th17 effector cells ultimately provide antibacterial immunity depending on the magnitude of the inoculum, entry route, and engagement of selected pattern recognition receptors (2–8). However, despite activation of a broad spectrum of immune-protective mechanisms, Y. enterocolitica has evolved a complex machinery to subvert antibacterial responses (9).

The pathogenicity of Y. enterocolitica relies on chromosome-encoded and plasmid-encoded virulence factors (9), including a type III secretion system composed of a syringe-like needle structure (injectisome) and effector proteins, termed Yersinia outer proteins (Yops), that are directly injected into host cells (9–11). Translocation of Yops (YopP, YopH, YopE, YopT, YopM, and YopO) into target cells contributes to disarm innate and adaptive immunity through multiple mechanisms. Although YopP/J inhibits NF-κB signaling, suppresses proinflammatory cytokines, modulates Ag uptake, and induces macrophage and DC apoptosis (12–16), YopH counteracts T cell activation (17, 18), and YopE, YopT, and YopO inhibit phagocytosis by disrupting actin cytoskeleton (11). However, in contrast to the intrinsic bacterial factors governing Y. enterocolitica pathogenesis, host factors co-opted by Y. enterocolitica to subvert antibacterial immunity are largely unknown.

Galectin-1 (Gal-1), a member of a family of β-galactoside–binding proteins, has emerged as an immune regulatory checkpoint that influences innate and adaptive immune compartments (19). This endogenous lectin, abundantly expressed in inflamed and mucosal tissues, controls excessive immune responses and promotes resolution of autoimmune pathology (20–22), favors immune escape in cancer (23–26), suppresses allergic reactions (27), and fuels tolerogenic programs during parasitic infection (28). The mechanisms underlying these broad immunoregulatory effects involve selective apoptosis of Th1 and Th17 effector cells (22), promotion of tolerogenic DCs (29), deactivation of M1-type macrophages (30–32), promotion of an IL-10–dependent immunoregulatory signature (20, 33–35), and expansion of Foxp3⁺ regulatory T cells (Tregs) (36). At the molecular level, Gal-1 acts by cross-linking a preferred set of glycosylated receptors, promoting their cell surface retention, and reorganizing and modulating their signaling threshold (37, 38).

Although little is known about the immunoregulatory role of Gal-1 during bacterial infection, emerging studies highlighted the relevance of this lectin during corneal immunopathology induced by Pseudomonas aeruginosa (39), gut inflammation induced by Citrobacter rodentium (40), and airway epithelial pathology induced by pneumococcal neuraminidases (41). Thus, given the broad immune-inhibitory activity of Gal-1 (19), its abundant expression in mucosal tissues (42), and its emerging roles during infection (37, 43, 44), we hypothesized that this endogenous lectin could serve as a homeostatic checkpoint and a mediator of immune evasion during Y. enterocolitica infection. Our results suggest that blocking Gal-1 may help to attenuate Y. enterocolitica pathogenicity and intestinal pathology by reinforcing immune-protective mechanisms.

C57BL/6 Gal-1–knockout (Lgals1^−/−) mice were kindly provided by F. Poirier (Institute Jacques Monod, Paris, France). C57BL/6 wild-type (WT) mice were purchased from the Animal Facilities of the National University of La Plata, La Plata, Argentina. Breeding colonies were established at the Animal Facilities of the National University of San Luis. Mice were kept under specific pathogen–free conditions in a cabinet (Ehret, Emmendingen, Germany) and provided with sterile food and water ad libitum. Male or female mice (6–8 wk-old) were used for all experiments. Studies were performed according to institutional and national guidelines.

Y. enterocolitica serotype 0:8 (pYV⁺, WA-314; kindly provided by G. Kapperud, Oslo, Norway) was used for infection. Y. enterocolitica WA-314 deficient in YopH (pYV⁺, WA-C pYV yopH Nal^r Kan^r) (45) and Y. enterocolitica WA-314 deficient in YopP (pYV⁺, WA-C pYV Nal^r Kan^r) (45) were developed in I.A.’s laboratory. Bacteria were cultured as described earlier (5). Mice were starved for 2 h and infected orogastrically with 2–5 × 10⁸ bacteria in 0.2 ml of sterile PBS (pH 7.4) using a gastric tube. Some groups were inoculated with 6 mg/kg aminoguanidine (AG), an inducible NO synthase (iNOS) inhibitor. Control mice were treated with PBS. The number of inoculated bacteria was controlled by plating serial dilutions of the inoculated suspension on Trypticase soy agar. Survival and weight were determined in infected mice. At 5, 14, and 21 d postinfection, serum samples from each mouse were collected according to standard methods. The spleen, PPs, and MLNs were aseptically removed and homogenized in PBS. Then, duplicates of 0.05 ml of serial dilutions of spleen and MLN homogenates were plated on Trypticase soy agar (46), whereas PP homogenates were plated on MacConkey–Igarsan agar. CFU were counted after incubation at 27°C for 48 h. The limit of detectable CFU was 25 (log₁₀ 25 = 1.4) (47). All experiments were repeated at least three times, revealing comparable results.

Resident peritoneal macrophages were obtained from mice using 5–10 ml of sterile pyrogen-free saline, washed twice by centrifugation at 200 × g for 10 min at 4°C, and resuspended in DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated FBS, 5 mM l-glutamine, 50 μM 2-ME, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml gentamicin. This cellular suspension was then added to each well of a 24-well culture plate to obtain 2 × 10⁶ cells per well. Cells were incubated at 37°C in a 5% CO₂ atmosphere for 24 h and, after washing three times with saline, adherent cells were incubated without (Medium) or with Y. enterocolitica WA-314 (Y. enterocolitica) (multiplicity of infection, 10:1) in the absence or presence of 5 μM ERK1/2 inhibitor (PD98059) or p38 inhibitor (SB203580; both from Calbiochem, San Diego, CA). Macrophages were incubated for 1 h at 37°C in a 5% CO₂ atmosphere, and 5 μl gentamicin was added. Cells were incubated overnight, and culture supernatants were obtained and immediately used for NO determination or stored at −20°C until Gal-1 determination.

Nitrite production, as a stable end product of NO metabolism, was measured in culture supernatants from macrophages that were treated or not for 12 h using the Griess reaction assay. Briefly, 100 μl of culture supernatant was mixed with 100 μl of Griess reagent in a 96-well flat-bottom plate and incubated for 10 min at room temperature. OD₅₅₀ was determined by a plate reader (Bio-Rad, New York, NY). Urea was determined in supernatants of macrophage cultures and in tissue homogenates using the Urea Color 2R kit (Wiener, Rosario, Argentina), following the manufacturer’s instructions.

Macrophages isolated from WT or Lgals1^−/− mice were infected with Y. enterocolitica and incubated for 2 h at 37°C in a 5% CO₂ atmosphere. DMSO was used as a positive control for apoptosis. Cells (1 × 10⁶) were washed twice with PBS and suspended in binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl₂ [pH 7.4]). Macrophages were incubated with Annexin V-FITC (Sigma, San Diego, CA) for 15 min at room temperature in the dark. Cells were washed twice and resuspended in 500 μl of binding buffer. Then, 2.5 μl of propidium iodide was added, and macrophages were analyzed by flow cytometry using a FACSCalibur cytometer (Becton, Dickinson and Company).

Homogenates of PPs and spleen, as well as macrophage lysates, were processed for SDS-PAGE and Western blot. Briefly, tissue homogenates were washed with cold PBS, and cells were prepared with lysis buffer (150 mM NaCl, 50 mM Tris-HCl [pH 7.5], 10 mM EDTA, 1% [v/v] Nonidet P-40) in the presence of a mixture of protease inhibitors (1:100). After 20 min of incubation on ice, lysates were centrifuged for 15 min at 12,000 × g at 4°C. To concentrate proteins, tissue homogenates were precipitated with 4 vol. acetone at −20°C for 6 h. Then, precipitated proteins were centrifuged at 6400 × g for 30 min at 4°C, supernatants were removed, and the pellet was dried. Finally, pellets were dissolved in PBS supplemented with protease inhibitors, and protein concentration was determined using a Bradford assay. SDS-PAGE and Western blot analysis were performed as previously described (23). Equal amounts of protein were resolved using 15% SDS-PAGE. Separated proteins were electroblotted onto PVDF Membranes (Amersham Biosciences, London, U.K.) and probed with rabbit anti–Gal-1 (1:5000 dilution) Ab that was generated as described (48) or goat anti–β-actin (1:3000 dilution) polyclonal Ab. Membranes were incubated with a peroxidase-labeled anti-rabbit IgG or anti-goat IgG and developed using an enhanced chemiluminescence detection kit (Amersham Biosciences) and Kodak BioMax films (Rochester, NY). Films were analyzed using ImageJ software and the intensity of each band was recorded and documented as relative expression.

Cytokines (IL-17, IFN-γ, IL-6, and TNF) were assessed 5 d postinfection in sera and tissue samples of infected and uninfected WT and Lgals1^−/− mice using capture ELISA kits (eBioscience, San Diego, CA), according to the manufacturer’s instructions. Cytokine concentrations were normalized to the protein content. Limits of detection were 4 pg/ml for IL-17 and IL-6, 15 pg/ml for IFN-γ, and 8 pg/ml for TNF.

IFN-γ and IL-17 were neutralized in vivo using a hamster anti-mouse IFN-γ mAb (MAB4851) and a rat anti-mouse IL-17 mAb (MAB421; both from R&D Systems), as described (5). The same doses of isotype-control mAbs were injected as controls. Briefly, Lgals1^−/− mice were treated in vivo with anti–IFN-γ, anti–IL-17, or isotype-control mAb i.p. (3.2 mg/kg) on days 0 and 3 postinfection. The percentage of mice surviving in each group was recorded. On day 5 postinfection, spleen, PPs, and MLNs were obtained, and CFU were counted as described (5).

Recombinant Gal-1 (rGal-1) was produced and purified as outlined previously (31). An anti–Gal-1–neutralizing mAb was produced, evaluated, and validated as described (24, 49–51). Four animal groups were used for in vivo rescue experiments: Lgals1^−/− mice injected i.p. with rGal-1 (3.2 mg/kg) daily for 5 d starting on infection day (day 0), as described (22); Lgals1^−/− mice injected i.p. with vehicle control; WT mice treated i.p. with anti–Gal-1–neutralizing mAb (5 mg/kg) in 0.2 ml of PBS on the day of infection (day 0) and at day 3 postinfection; and WT mice treated with an isotype-control mAb, as described (24, 49). The percentage of surviving mice, as well as animal weight, were recorded. Mice were killed at day 5 postinfection, and CFU were counted in homogenates of spleen, PPs, and MLNs. Levels of IL-17, IFN-γ, IL-6, and TNF were determined in serum or in PPs, as described above.

NF-κB activation was determined in PPs of WT mice treated with anti–Gal-1 or isotype-control mAb and in Lgals1^−/− mice treated with rGal-1 or vehicle control using a nonradioactive, sensitive method for detection of NF-κB DNA-binding activity in nuclear extracts, following the manufacturer’s recommended protocol [NF-κB (p65) Transcription Factor Assay Kit; Cayman Chemical, MI]. Results were expressed as absorbance per 10⁶ cells.

For histopathological examination, intestine was collected at day 5 postinfection from mice belonging to each group, fixed in 10% buffered formalin phosphate, embedded in paraffin sections, and stained with H&E.

Differences between the groups were tested for significance using the Mann–Whitney U test or one-way ANOVA with the Dunnett multiple-comparison test, as appropriate. A p value <0.05 was considered statistically significant. Survival curves were generated using Kaplan–Meier survival analysis. Results are expressed as the mean ± SEM. All statistical analyses were carried out using Prism version 5.0 (GraphPad, La Jolla, CA).

To investigate the relevance of Gal-1 in Y. enterocolitica–induced immunopathogenesis, we first analyzed expression of this lectin during the course of Y. enterocolitica infection in vivo. We observed substantial upregulation of Gal-1 in PPs and spleen of Y. enterocolitica–infected C57BL/6 WT mice compared with noninfected C57BL/6 WT mice (p < 0.01 in spleen at day 5, p < 0.001 in PPs at day 5, p < 0.05 in spleen and PPs at day 7) (Fig. 1A–D). This effect was prevented when mice were infected with Y. enterocolitica lacking the effector protein YopP (p < 0.01) or YopH (p < 0.05) (Fig. 1E). Moreover, in vitro infection of macrophages with Y. enterocolitica led to significant upregulation of Gal-1 expression (p < 0.05); this effect was prevented when ERK1/2 or p38 signaling was interrupted (Fig. 1F, p < 0.05). Thus, bacterial infection induces local and systemic upregulation of Gal-1 through mechanisms involving Yops and MAPK signaling.

To evaluate whether Gal-1 expression influences the clinical outcome of Y. enterocolitica infection, we compared the survival of infected WT and Lgals1^−/− mice. We found a significantly higher survival rate in Lgals1^−/− mice compared with their WT counterparts (Fig. 2A, p < 0.001). Moreover, mice lacking Gal-1 displayed higher weight compared with WT mice, reaching statistical significance at day 5 postinfection (Fig. 2B, p < 0.05). Furthermore, lower numbers of CFU were detected in PPs, but not in MLNs or spleen, of Lgals1^−/− mice versus WT mice at 5 and 14 d postinfection (p < 0.001, p < 0.05, respectively) (Fig. 2C, 2D). However, at 21 d postinfection, differences between Gal-1 mutant and WT mice were evident only in the spleen (p < 0.05) (Fig. 2E), suggesting a spatiotemporal regulation of Gal-1 activity that might act locally (at the mucosal level) during the early stages of infection and shift toward a systemic effect at later time periods. Thus, lack of endogenous Gal-1 confers protection against Y. enterocolitica infection.

To identify possible mechanisms underlying the protective effects of Gal-1 deficiency, we first explored central innate immune pathways involved in Y. enterocolitica clearance (52). Five days after Y. enterocolitica infection, PP homogenates isolated from Lgals1^−/− mice showed increased nuclear translocation of the NF-κB transcription factor (Fig. 3A, p < 0.001) and higher amounts of TNF (Fig. 3B, p < 0.001) compared with PPs from WT mice, suggesting that Gal-1 may act by attenuating mucosal proinflammatory responses. However, no differences were observed in IL-6 production in PPs between WT and mutant strains (Fig. 3C).

Gal-1 modulates l-arginine metabolism in peritoneal macrophages and microglia by skewing the balance from classically activated M1 macrophages (synthesizing high levels of NO) toward alternatively activated M2 macrophages (expressing high levels of arginase) (30, 32). To explore whether modulation of l-arginine metabolism might account for the immunoregulatory activity of this lectin during Y. enterocolitica infection, we analyzed the synthesis of NO and urea (a metabolic product of arginase enzymatic activity) in peritoneal macrophages obtained from Gal-1–deficient and WT mice. Lgals1^−/− macrophages produced considerably higher amounts of NO and lower amounts of urea than did WT mice when infected in vitro with Y. enterocolitica (Fig. 4A, 4B, p < 0.001). This effect was recapitulated in peritoneal macrophages isolated from in vivo–infected Lgals1^−/− and WT mice (data not shown). Consistent with these findings, treatment of Y. enterocolitica–infected Lgals1^−/− mice with AG, a selective inhibitor of iNOS activity, prevented the protective effect triggered by Gal-1 deficiency (Fig. 4C, p < 0.05). Remarkably, inhibition of NO synthesis led to increased bacterial load (p < 0.05) and higher amounts of urea (p < 0.05) in PPs from Y. enterocolitica–infected Lgals1^−/− mice (Fig. 4C, 4D). However, no significant differences in apoptosis were detected between in vitro–infected Lgals1^−/− and WT macrophages (Fig. 4E). Thus, in response to Y. enterocolitica infection, endogenous Gal-1 limits antibacterial responses, at least in part, by restraining mucosal proinflammatory circuits and counteracting the classical l-arginine–NO pathway.

The critical role of Th1 and Th17 cells in controlling Y. enterocolitica infection (4, 5, 53) and the enhanced sensitivity of these proinflammatory cells to the regulatory effects of Gal-1 (22, 35, 54, 55) prompted us to investigate the contribution of IFN-γ and IL-17 to host protection conferred by Gal-1 deficiency. We found significantly higher levels of IFN-γ and IL-17 in sera collected from Lgals1^−/− mice compared with WT mice (p < 0.05, p < 0.001, respectively) 5 d postinfection (Fig. 5A, 5B). These differences were not observed in sera from uninfected mice, regardless of their genotype (Fig. 5A, 5B), suggesting that Gal-1 deficiency may unleash proinflammatory responses following antigenic challenge.

To investigate the relevance of IFN-γ and IL-17 in host protection conferred by Gal-1 deficiency, we treated Y. enterocolitica–infected Lgals-1^−/− mice with anti–IFN-γ– or anti–IL-17–neutralizing mAb and monitored the clinical course of infection. Remarkably, protection against Y. enterocolitica infection was completely abrogated in mice receiving an anti–IFN-γ–neutralizing mAb, regardless of Gal-1 deficiency (100% of Lgals1^−/− infected mice died at day 4 postinfection), whereas Y. enterocolitica–infected Lgals1^−/− mice treated with anti–IL-17 mAb showed only a partial effect (Fig. 5C, p < 0.05). Of note, administration of an isotype-control mAb did not affect mice survival, and it preserved the protective effect observed in the absence of Gal-1 (Fig. 5C, 5D). Furthermore, spleens from Lgals-1^−/− infected mice treated with anti-cytokine mAb showed higher bacterial load compared with those receiving isotype-control mAb (Fig. 5D, p < 0.05). These results demonstrate the different roles of IFN-γ and IL-17 in Y. enterocolitica clearance and host protection conferred by Gal-1 deficiency.

To confirm the direct role of Gal-1 in hindering immunity against Y. enterocolitica infection, we explored whether exogenous rGal-1 could override the protective effect observed in Lgals1^−/− hosts and rescue the mutant phenotype. Compared with Lgals1^−/− mice receiving vehicle control, Y. enterocolitica–infected Lgals1^−/− mice treated with rGal-1 (3.2 mg/kg daily for 5 d starting on the day of infection) showed reduced protection, as revealed by the increased number of CFU in spleen, PPs, and MLNs (Fig 6A, p < 0.05 at day 5 postinfection). Moreover, rGal-1 supplementation in Lgals-1^−/− mice restored NF-κB activation and TNF synthesis in PPs to levels similar to those attained in WT hosts (Fig. 6B, 6C, p < 0.001, p < 0.05, respectively). Furthermore, rGal-1 administration suppressed IL-6 production in PPs (Fig. 6D, p < 0.05). However, administration of the exogenous lectin was not sufficient to restrain heightened IFN-γ and IL-17 responses triggered in the context of Gal-1 deficiency (Fig. 6E, 6F), suggesting that endogenous Gal-1 might be crucial to counterbalance Th1–Th17 dysregulation as has been reported (22). Nevertheless, exogenous rGal-1 was sufficient to thwart the antibacterial protective effect and counteract proinflammatory mechanisms unleashed in the absence of the endogenous lectin.

In the search for a potential adjuvant strategy that could reinforce antibacterial responses, we investigated the effect of Ab-mediated Gal-1 blockade during the course of Y. enterocolitica infection. Remarkably, WT mice treated with anti–Gal-1–neutralizing mAb (5 mg/kg given at days 0 and 3 postinfection) showed lower bacterial load in PPs, MLNs, and spleen compared with WT mice treated with isotype-control mAb (Fig. 7A, p < 0.01 in PPs, p < 0.05 in MLNs and spleen). Moreover, Ab-mediated Gal-1 blockade induced greater NF-κB activation in PPs from Y. enterocolitica–infected WT mice compared with isotype-treated groups (Fig. 7B, p < 0.01), showing comparable levels as those attained in PPs from Y. enterocolitica–infected Lgals1^−/− mice (Fig. 3A). Histopathological assessment revealed a more prominent mononuclear cell infiltrate in PPs from anti–Gal-1–treated Y. enterocolitica–infected WT mice, similar to that observed in Y. enterocolitica–infected Lgals1^−/− animals (Fig. 7C). Finally, higher levels of IFN-γ and IL-17 were detected in WT mice receiving the anti–Gal-1 mAb (Fig. 7D, 7E, p < 0.05) compared with those given isotype control. Collectively, these results validate the therapeutic potential of Gal-1 blockade as an adjuvant mechanism to reinforce host protective immunity during bacterial infection.

Negative-regulatory pathways function as homeostatic signals that control unresolved inflammation and curb tissue damage while limiting protective immune responses (56). These pathways include the so-called “inhibitory checkpoints,” which may be usurped by pathogens to subvert innate and adaptive immune mechanisms (56). Galectins, a family of soluble glycan-binding proteins, have emerged as regulatory checkpoints that are secreted to the extracellular medium through an unconventional pathway and control immune cell fate by interacting with relevant glycosylated receptors, including CD45, CD43, T-cell Ig and mucin-domain containing-3, CTLA-4, and lymphocyte activation-gene-3 (19). Gal-1, a prototype member of this family, is expressed by a wide variety of cell types, including macrophages (57), activated T cells (58), activated B cells (59), and DCs (29), and is highly represented in privileged and mucosal tissue (20, 42). This endogenous lectin contributes to disarm host immune mechanisms by promoting contraction of the Th1/Th17 compartment (22), favoring expansion of Foxp3⁺ and Foxp3⁻ Tregs (20, 28, 35, 36), deactivating macrophages and microglia (30–32), and inducing tolerogenic DCs (29). These regulatory circuits, hierarchically driven by Gal-1, have been investigated in settings of cancer (23, 36, 60), fetal loss (61, 62), autoimmunity (20–22), and parasite infection (28).

In this study, we identified a central role for Gal-1 in Y. enterocolitica immunopathogenesis using loss-of-function and gain-of-function experiments. Oral Y. enterocolitica administration induced considerable upregulation of Gal-1 in PPs and spleen through a YopP/YopH-dependent mechanism. Lack of Gal-1 conferred protection against Y. enterocolitica infection by reinforcing innate and adaptive immune programs, including NF-κB activation, TNF and NO production, and heightened expression of IFN-γ and IL-17. This protective effect was substantiated by supplementation of rGal-1 into Y. enterocolitica–infected Lgals1^−/− mice or by treatment of infected WT mice with a neutralizing anti–Gal-1 mAb. Similarly, in a model of corneal immunopathology induced by P. aeruginosa, local treatment with rGal-1 attenuated corneal inflammation through mechanisms involving reduced Th17 responses and expansion of Th2 cells (39). Moreover, previous studies showed that influenza infection predisposes airway epithelial cells to Streptococcus pneumoniae infection via release of neuraminidases with subsequent exposure of Gal-1–specific glycan epitopes (41), suggesting a broader role for Gal-1–glycan interactions in bacterial infection. In this regard, Gal-1 may also facilitate infection by other pathogens, including HIV (63), dengue virus (64), and Trichomonas vaginalis (65), whereas it confers protection against Nipah virus (66) and can differentially regulate Trypanosoma cruzi infection at different stages of the disease (28, 67). These data highlight the diverse roles of Gal-1 in microbial invasion and pathogenesis through immune- and nonimmune-mediated mechanisms.

Glycosylation controls a variety of signaling programs, including those operating in immune cell signaling and host–pathogen interactions (38). Because bacteria and eukaryotic cells possess O- and N-linked glycans (68, 69), glycan-binding proteins may serve as a bridge that regulates bacterial infection, internalization, and immunity. Whether YopP/YopH associate with endogenous Gal-1 to control Y. enterocolitica pathogenesis should be further explored. Following this direction, future efforts should be aimed at analyzing glycosylation profiles of Yersinia virulence factors and their specific interactions with endogenous galectins. Interestingly, recent studies showed that epithelial cell glycosylation is regulated by gut microbes and by group 3 innate lymphoid cells (70), suggesting the possibility that, during infection, Y. enterocolitica could also alter the glycosylation profile of enterocytes, M cells, macrophages, and/or T cells, favoring exposure of Gal-1–specific glycan epitopes. In this regard, we previously showed that differential glycosylation of Th1, Th17, and Th2 cells may selectively regulate their viability by unmasking or masking Gal-1–specific ligands (22). Accordingly, we found higher levels of IFN-γ and IL-17 in serum from Y. enterocolitica–infected Lgals1^−/− mice, as well as in WT mice treated with a neutralizing anti–Gal-1 mAb; this effect was a critical determinant of host protection, because specific blockade of these proinflammatory cytokines increased local and systemic bacterial load, regardless of the absence of this lectin. However, supplementation of rGal-1 into infected Gal-1–deficient hosts did not rescue systemic Th1/Th17 dysregulation, suggesting that a combination of exogenous and endogenous Gal-1 may be required to achieve full suppression, as has been previously shown in experimental autoimmune settings (20, 22). Although Gal-1 deficiency promoted upregulation of IFN-γ and IL-17 and prolonged mice survival, systemic blockade of IFN-γ completely rescued this phenotype, whereas IL-17 neutralization induced a partial effect (Fig. 5A–C). Interestingly, previous studies reported selective protective roles for IFN-γ (5, 71) and IL-17 (4, 5, 53) during Y. enterocolitica infection, depending on preferential TLR activation and predominant routes of infection (4). Moreover, recent studies showed that Y. pseudotuberculosis supports Th17 expansion, whereas it limits Treg differentiation by directly interfering with TCR signaling (72).

Remarkably, host protection induced by Gal-1 deficiency involved increased NO synthesis and decreased urea production in PPs of Y. enterocolitica–infected mice. These data are in agreement with previous findings on the ability of exogenous Gal-1 to skew the balance from classically activated M1 macrophages/microglia toward an alternatively activated M2 phenotype (30, 32). These results are relevant in the context of Yersinia infection because Y. enterocolitica and Y. pseudotuberculosis control NO production in murine macrophages, with YopJ, YopE, and YopH being responsible for such effects (8, 73, 74). In this regard, although Y. enterocolitica can trigger apoptosis of infected macrophages (11, 12, 15, 16), Gal-1 did not seem to mediate this effect (Fig. 4E). Moreover, lack of Gal-1 promoted greater activation of the NF-κB transcription factor and release of TNF, an NF-κB–regulated cytokine, in mucosal tissue of Y. enterocolitica–infected mice. Supporting these findings, a regulatory loop has been reported by which NF-κB and MAPKs induce Gal-1 expression in T cells; this, in turn, leads to inhibition of NF-κB signaling (58, 75). However, recent studies in a setting of osteoarthritis demonstrated that exogenous Gal-1 could also activate an NF-κB–regulated gene network in chondrocytes (76), suggesting that this lectin may differentially control pro- or anti-inflammatory gene expression in different cell types and physiologic/pathologic settings. In this context, recent studies suggested that TNFR p55 acts as a regulatory switch that limits Y. enterocolitica–induced reactive arthritis by modulating IL-6, NO, and NF-κB pathways and controlling Treg expansion (5, 8, 77). Because these molecular and cellular pathways are preferred targets of the immunoregulatory activity of Gal-1, a possible cross-talk between TNFR p55 signaling and this endogenous lectin might contribute to hierarchically dissect the immunopathogenesis of Y. enterocolitica infection and the underlying tissue damage. Interestingly, although endogenous Gal-1 controlled NF-κB activation and TNF synthesis, it had no apparent role in IL-6 production during the time period analyzed (Fig. 3). However, TNF and IL-6 were downregulated when exogenous Gal-1 was administered to Lgals1^−/− mice (Fig. 6C, 6D). This distinctive effect could be related to different kinetics of cytokine synthesis and regulation in response to endogenous versus exogenous Gal-1.

In conclusion, our study highlights the involvement of host Gal-1 as a critical mediator of the immune inhibitory activity of Y. enterocolitica in vivo. By targeting innate and adaptive immune compartments, Gal-1 limits bacterial clearance and curtails mice survival. Disruption of Gal-1–glycan interactions, using a neutralizing anti–Gal-1 mAb, conferred protection against Y. enterocolitica infection by reprogramming antibacterial immunity. This study may have broad therapeutic implications for strengthening immune responses in a wide range of enteropathogenic microbial infections.

We thank R. Morales and S. Gatto for dedicated animal care and P. Hockl for technical assistance. We also thank M. Shipp and J. Ouyang for sharing the anti–Gal-1 mAb that was generated and validated in collaboration with G.A.R.

The authors have no financial conflicts of interest.