Regulatory Roles for NKT Cell Ligands in Environmentally Induced Autoimmunity¹ Free

Autoimmune diseases can arise because of complex interactions between genetic and environmental factors. The heavy metal mercury, in addition to being a common environmental pollutant, is found in a wide range of commercial and medical products, allowing for accidental or occupational exposure. In humans, chronic exposure to low doses of inorganic mercury, a potent immunomodulator (1), can induce glomerulonephritis and proteinuria (2, 3), excessive T cell activation, increased levels of serum IgE, and anti-nuclear Abs in the serum (4, 5). A similar immune dysfunction can be induced in mice wherein administration of subtoxic doses of HgCl₂ to susceptible animals induces an autoimmune syndrome characterized by increases in serum Igs and anti-nucleolar autoantibody (ANoA)³ production (6, 7). Mercury-induced ANoA production depends on the MHC haplotype, with the H-2^s and H-2^b gene products conferring susceptibility and resistance, respectively (8). In addition, susceptibility is influenced by unidentified background genes, with varying susceptibility between different mouse strains. A.SW mice (H-2^s) are very susceptible; C57BL6-SJL (H-2^s) mice are less so, whereas other strains like DBA/2 (H-2^d) are completely resistant (9, 10, 11). The mechanisms by which a heavy metal disrupts immune tolerance and induces autoimmunity remain unknown.

NKT cells are evolutionarily conserved lymphocytes expressing NK and T cell lineage markers. Classical or type 1 NKT cells bear a semiinvariant TCR that recognizes a variety of glycolipid Ags presented by CD1d, a nonpolymorphic MHC class I-like Ag-presenting molecule (12, 13, 14). These cells bear the hallmarks of memory T cells, secreting a burst of cytokines (including IFN-γ, IL-4, IL-13, IL-10, and TGF-β) within hours of encountering their Ags (14, 15). This, combined with the up-regulation of costimulatory molecules like CD40L (12, 15), results in rapid activation of the immune system, making this cell type a very potent modulator of an immune response. The ligand most commonly used to activate these cells is a marine sponge-derived glycolipid, α-galactosyl ceramide (α-GalCer) (12). We used two recently developed synthetic variants of α-GalCer, PBS 57 (16) and 4-deoxy α-GalCer (17), to study the effects of NKT cell activation on mercury-induced autoimmunity.

Recent studies have demonstrated that NKT cells form an important part of the innate immune defense against a variety of microbes. The most potent microbial NKT cell ligands were identified on Gram-negative, LPS-negative members of the Sphingomonas species, indicating that this cell subset is a major innate recognition pathway for LPS-negative bacteria (18). The development or exacerbation of autoimmunity has been linked separately to exposure to infectious agents (6) or chemicals (6, 19, 20). We investigated the combined effect of a chemical, mercury, and a NKT cell ligand-bearing bacterium of the Sphingomonas species, Sphingomonas capsulata (18), on the development of autoimmunity. Susceptible mice can be tolerized to mercury by pre-exposing them to low levels of the heavy metal under nonactivating conditions (our unpublished observations). We examined whether NKT cell activation had regulatory roles in the induction or maintenance of tolerance to mercury. We demonstrate that, overall, NKT cell activation can exacerbate the mercury-induced syndrome. Bacteria bearing both NKT and TLR ligands strongly exacerbated the heavy metal-induced autoimmunity. However, the two synthetic NKT cell activators, when combined with two other immunomodulatory agents, mercury and a TLR9 agonist, exerted profoundly different effects. Only one of the ligands prevented induction of tolerance to mercury, and the two ligands either strongly potentiated or inhibited TLR9-mediated breakage of tolerance. The effects may be related to the differential modulation of cytokine profiles and CD4⁺CD25⁺Foxp3⁺ regulatory T cell (T_reg) populations that arose only when NKT ligands were given in combination with mercury.

A.SW (H-2^s) mice were obtained from The Jackson Laboratory and maintained in our animal facilities. Congenic C57BL/6.SJL mice (H-2^s) were originally obtained from The Jackson Laboratory and are bred and maintained in our animal facility. All of the mice used in our experiments were at least 2 mo old.

Mercury-induced autoimmunity was induced according to a standard protocol (21) by three s.c. injections of 30 μg of HgCl₂ in 100 μl of sterile PBS at days 0, 2, and 4.

Two synthetic variants of α-GalCer, PBS 57 and 4-deoxy α-GalCer (Fig. 1), were used in this study. Their syntheses have been described elsewhere (16, 17). Stock solutions were originally prepared in 100% DMSO at a concentration of 1 mg/ml and were diluted in PBS (PBS 57) or PBS 0.5% Tween 20 (4-deoxy α-GalCer) just before i.p. injection into mice. Control groups received equivalent volumes of PBS/2% DMSO or PBS/2% DMSO/0.5% Tween 20.

S. capsulata (American Type Culture Collection 14666) was heat killed for 2 h at 74°C. Cells were washed after the procedure by centrifugation and taken up in PBS. A total of 0.1 μl of bacterial suspension (equivalent to 10⁷ bacterial CFUs) was injected i.p. at indicated timepoints.

Serum ANoA titers were determined by indirect immunofluorescence. Sera diluted in PBN (PBS containing 1% BSA and 0.02% sodium azide) were incubated with HEp-2 slides (Antibodies, Inc.) for 30 min, and ANoA were detected with FITC-conjugated goat anti-mouse IgG1 or IgG2a Abs (Southern Biotechnology Associates). The inverse of the highest serum dilution at which nucleolar fluorescence could be detected was defined as the ANoA titer (21).

Total serum IgE and IgG2a levels were determined using a sandwich ELISA, as previously described (21).

Immunostimulatory ODN CpG 1826 (TCCATGACGTTCCTGACGTT) on a phosphothiorate background was obtained from Oligos Etc. A.SW mice were injected s.c. in the flank with 100 μg of CpG 1826 or vehicle (PBS) at the indicated timepoint.

Biotinylated, rCD1d molecules were incubated with PBS 57 (at a 2:1 ratio) in PBS for 2 h at 37°C. Free PBS 57 was removed by centrifugation dialysis in a Microcon YM-30 tube (Millipore). Tetramers were generated by mixing PBS 57-loaded monomers with fluorochrome-labeled streptavidin (steptavidin-allophycocyanin; BD Biosciences) at a 5:1 ratio. The resulting solution was incubated for 30 min at 37°C (16). The CD1d molecules were a gift from A. Bendelac (University of Chicago, Chicago, IL) or were obtained from the National Institutes of Health tetramer facility (Emory University).

Nonspecific staining was blocked using 10 μg/ml rat anti-mouse CD16/32 (2.4G2) (BD Biosciences). For staining the NKT cell population, splenocytes were incubated with allophycocyanin-conjugated PBS 57-loaded CD1d tetramers (45 min at 37°C) and then stained for 30 min at 4°C with PE-labeled anti-mouse B220 (RA3-6B2; BD Biosciences) to negatively gate B cells. For staining of regulatory T cells, splenocytes were incubated with FITC-conjugated anti-mouse CD4 (GK1.5; BD Biosciences) and PE-conjugated anti-mouse CD25 (3C7; BD Biosciences) at 4°C, followed by overnight permeabilization using a Cytoperm/Cytofix kit (eBioscience), followed by staining with allophycocyanin-conjugated anti-mouse Foxp3 (FJK-16s) (eBioscience). Regulatory T cell and NKT cell population data analyses were conducted using a dual laser FACSCalibur (BD Biosciences) and a FACSAria (BD Biosciences) respectively. Flow cytometry data were analyzed using FlowJo software (Tree Star).

Splenocytes obtained 8 days after start of indicated treatments were suspended in 6-well plates and stimulated with PMA (20 ng/ml; BIOMOL) and ionomycin (500 ng/ml; Tocris Cookson). In some experiments, splenocytes were suspended in 24-well plates and stimulated with plate-bound anti-CD3 mAb (coated in PBS at 3 μg/ml) and anti-CD28 mAb (coated in PBS at 2 μg/ml). Anti-CD3 and anti-CD28 mAbs were obtained from BioXCell (formerly Bio-express). Cell culture supernatants were assayed at indicated timepoints for cytokine release by sandwich ELISA.

Total serum IL-4, IL-10, and IFN-γ were measured by sandwich ELISA using matched capture and detection Abs (BD Biosciences).

The dependent variables (flow cytometry counts, cytokine levels, etc.) were treated as continuous variables for all analyses. Means, SDs, and number of observations were presented for each variable. The experiments used a two-factor randomized design (group, time period). The null hypothesis was that there would be no difference between groups or time periods. Before analysis, all data were tested for normality using the Shapiro-Wilk test (22). The data were significantly nonnormal for all variables. To apply ANOVA methods, a normalized rank transformation was applied to the data (23, 24). The rank-transformed data were analyzed using a generalized linear model ANOVA, followed by multiple comparisons to detect significant mean differences between groups and periods. Differences between means (rejection of the null hypothesis) were considered significant if the probability of chance occurrence was ≤0.05 using two-tailed tests. Statistical analyses were conducted using SAS v9.1 software (SAS Institute).

The C57BL/6.SJL mouse strain is moderately mercury susceptible and requires several weeks of mercury administration to induce autoimmunity. We examined whether NKT cell activation could accelerate autoimmunity in this strain. We used two synthetic derivatives of α-Galcer, PBS 57 and 4-deoxy α-GalCer, to activate NKT cells. Briefly, mice received only three injections of HgCl₂ with or without 4-deoxy α-GalCer (Fig. 2,A) or PBS 57 (Fig. 2,B). Increase in serum Igs in mercury-administered mice was negligible and was not further enhanced by NKT cell activation (data not shown). The group receiving mercury alone showed little or no autoantibody production (Fig. 2). However, NKT cell activation by 4-deoxy α-GalCer significantly potentiated ANoA production of both the IgG1 (p < 0.001) and the IgG2a (p < 0.0001) isotypes (Fig. 2,A), whereas NKT cell activation by PBS 57 strongly potentiated IgG2a autoantibodies (p < 0.0001), while having a modest effect on IgG1 autoantibodies (Fig. 2 B). NKT cell activation thus increased susceptibility to mercury-induced autoimmunity in this mouse strain.

We next assessed the role of NKT cells in A.SW mice, a strain highly susceptible to mercury-induced autoimmunity. Briefly, mice received three injections of HgCl₂ with or without the NKT cell ligand. A.SW mice in response to mercury produced ANoA and also showed an increase in serum IgE (Fig. 3). Administration of NKT cell ligands did not induce any manifestations of autoimmunity (data not shown). Activation of NKT cells modulated mercury-induced autoimmunity, but surprisingly, the two ligands had strikingly different effects. NKT cell activation by 4-deoxy α-GalCer, but not PBS 57 (Fig. 3), significantly increased IgG2a autoantibody production compared with mercury-treated controls. Both ligands dramatically increased IgE levels relative to the mercury-treated controls, although this effect was more pronounced with PBS 57 (Fig. 3).

We examined the effect of dual administration of mercury and glycosphingolipids on the NKT cell population. A.SW mice received HgCl₂ with or without NKT cell ligands, or NKT cell ligands alone. We first examined the effect of PBS 57. In mice that received mercury only, NKT cells represented 0.37% of the splenocytes (10.48 × 10⁴ NKT cells/spleen), similar to values in untreated mice (data not shown). As expected with a NKT cell ligand, PBS 57 significantly (p < 0.05) expanded the NKT cell population, increasing the percentage to 1.23% (73 × 10⁴ NKT cells/spleen). Interestingly, coadministration of mercury significantly (p < 0.001) potentiated the PBS 57-induced expansion of NKT cells, further increasing it to 2.33% (205 × 10⁴ NKT cells/spleen) (Fig. 4 A).

We next examined the effect of 4-deoxy α-GalCer in combination with mercury. The 4-deoxy α-GalCer also expanded (p < 0.05) the NKT cell population, increasing it to 1.34% (144.4 × 10⁴ NKT cells/spleen). However, the NKT cell population in mice receiving 4-deoxy α-GalCer and mercury was not significantly different from that in mice receiving 4-deoxy α-GalCer alone (Fig. 4 B). Thus, unlike PBS 57, the 4-deoxy α-GalCer-induced NKT cell expansion was not further potentiated by coadministration of mercury.

Studies in our laboratory demonstrated that genetically susceptible mice can be rendered tolerant to mercury by a single preinjection of a lower dose of HgCl₂ (our unpublished observations). We investigated whether NKT cell activation can prevent the induction of tolerance. A.SW mice received PBS 57, 4-deoxy αGalCer, or vehicle 4 h before administration of a low dose (3 μg) of HgCl₂. One week later, all mice received three injections of the regular dose (30 μg) of HgCl₂. Following challenge with the standard course of mercury, tolerized mice showed low titers of autoantibodies and low serum IgE (Fig. 5) levels. Preactivation of NKT cells with PBS 57 prevented tolerance establishment and significantly increased IgG1 (p < 0.05) and IgG2a (p < 0.0001) autoantibodies and serum IgE (p < 0.05) levels (Fig. 5). In contrast, preactivation of NKT cells by 4-deoxy α-GalCer was not effective at preventing tolerance induction; mice treated with this ligand showed no increase in serum IgE levels or IgG1 autoantibodies and only a mild increase in IgG2a autoantibodies at week 3 (Fig. 5).

Our results demonstrate that NKT cell activation can prevent tolerance establishment. We next investigated whether NKT cell activation could break established tolerance. In vivo exposure to microbial NKT ligands is likely to involve coexposure to various TLR ligands. To study the interplay between NKT cell ligands and TLR ligands in this system, we challenged tolerized mice with the NKT cell ligands given alone or in combination with a TLR9 ligand, a stimulatory class B ODN CpG 1826. Briefly, ASW mice received a low dose of mercury i.p. One week later, all mice received the standard regimen of mercury with or without the NKT cell ligands, CpG 1826, or combinations of these, at the indicated timepoints (Fig. 6).

NKT cell activation by either ligand had a mild effect on autoantibody production (Fig. 6), although 4-deoxy α-GalCer administration resulted in higher IgG2a titers than PBS 57. In contrast to this, TLR9 activation was more effective at breaking tolerance. It had a stronger effect on the autoantibody production, especially of the IgG2a isotype (Fig. 6). When both a NKT cell ligand and a TLR9 ligand were administered, 4-deoxy α-GalCer synergized with CpG 1826 and potentiated IgG1 (p < 0.05) autoantibody production (Fig. 6). Interestingly, PBS 57 had the opposite effect, significantly (p < 0.05) inhibiting TLR9-induced restoration of IgG1 and IgG2a autoantibodies (Fig. 6). Thus, NKT cell activation can either synergize with or antagonize TLR9-induced autoantibody production, depending on the activating NKT ligand. In contrast, when we examined serum IgE levels, we did not observe antagonism between PBS 57 and the TLR9 ligand. Both NKT cell ligands behaved similarly and synergized with CpG 1826 to increase serum IgE levels (Fig. 6). Therefore, PBS 57 (administered without mercury) prevents induction of tolerance, but when given with regular dose of mercury, antagonizes TLR9-induced breaking of tolerance.

T_reg are required to induce tolerance in our model (our unpublished observations). Previous studies have demonstrated that NKT cell activation can modulate the T_reg population (25, 26). We examined whether mercury could affect the cross-talk between NKT cells and T_regs. Briefly, A.SW mice received either NKT cell ligand alone or in combination with mercury in the same schedule as in Fig. 2. We observed that, as previously described (26), NKT cell activation moderately increased T_reg numbers in the spleen (Fig. 7,A). Interestingly, mercury very strongly potentiated this effect, increasing T_reg numbers in mice treated with either NKT ligand (Fig. 7,A), although this effect was significantly (p < 0.05) stronger in PBS 57-treated compared with 4-deoxy α-GalCer-treated animals. However, the NKT cell ligands and mercury strongly synergize to increase overall splenocyte numbers. Hence, the increase in T_reg numbers could be a reflection of the increase in the total splenocyte number. When the percentage of T_regs was examined, we found that treatment with mercury, NKT cell ligands, or the combination of the two decreased the T_reg frequency in the spleen (Fig. 7,A). There was no difference among mice receiving NKT cell ligands or mercury alone. However, the percentage of T_regs was significantly (p < 0.05) higher in mice that received PBS 57 with mercury, compared with mercury-treated animals. In contrast, the mice receiving 4-deoxy α-GalCer with mercury had a significantly lower frequency of T_regs than the group that received PBS 57 with mercury (p < 0.001) or the mice that received mercury alone (p < 0.05) (Fig. 7 A). Thus, in mercury-administered animals, PBS 57 increases T_reg frequency, whereas 4-deoxy α-GalCer lowers it.

We also examined cytokine production by splenocytes from the same experimental groups. Briefly, splenocytes collected at day 8 after start of treatments were stimulated in vitro with PMA/ionomycin. Supernatants were analyzed for IFN-γ, IL-4, and IL-10 levels. There were no significant differences in cytokine production among mice receiving either NKT cell ligand (Fig. 7,B). However, administration of mercury significantly (p < 0.05) increased IL-4 and IL-10 production in PBS 57-treated, but not 4-deoxy α-GalCer-treated animals (Fig. 7,B). In contrast, IFN-γ levels were similar between mice given either NKT ligand, with or without mercury (Fig. 7 B). These results suggest that the PBS 57-induced (but not 4-deoxy α-GalCer-induced) cytokine secretion is altered by mercury, with increased skewing toward Th2 cytokine production. Collectively, these results indicate that, in the presence of mercury, 4-deoxy α-GalCer-induced NKT cell activation induces fewer T_regs and decreased Th2 cytokine production when compared with NKT cell activation by PBS 57.

NKT cell ligands when administered with CpG 1826 induce opposing effects on the autoantibody production (Fig. 6). Because T_regs down-regulate autoantibody production in tolerized animals (our unpublished obervations), we examined the T_reg population in tolerized mice challenged with NKT cell ligands and CpG 1826 following the protocol used for the experiments described in Fig. 6. Mice were sacrificed on day 8, the timepoint at which they become positive for anti-nucleolar reactivity. We noted that the absolute number of splenocytes was reduced in mice receiving PBS 57 with CpG 1826 (247.3 × 10⁶) compared with mice receiving CpG 1826 alone (350.6 × 10⁶) or CpG 1826 with 4-deoxy α-GalCer (340 × 10⁶) (Fig. 8,A). Furthermore, the total number of CD4⁺CD25⁺ cells was decreased in mice receiving PBS 57 with CpG 1826, compared with mice receiving CpG 1826 alone or with 4-deoxy α-GalCer (Fig. 8,B). Although the frequency of T_regs (CD4⁺CD25⁺Foxp3⁺ cells) in these three groups was similar (Fig. 8,C), the frequency of T_effectors (CD4⁺CD25⁺ Foxp3⁻ cells) was greatly reduced when mice received PBS 57 with CpG 1826 (Fig. 8,D). The ratio of T_regs to T_effectors is shown in Fig. 8 E. Interestingly, treatment with CpG 1826 significantly (p < 0.05) decreased this ratio compared with control-tolerized or untreated animals (3.61 vs 8.77 or 8.44). Coadministration of 4-deoxy α-GalCer (3.37) maintained this decrease, but coadministration of PBS 57 restored the ratio to 9.33, similar to the value in control-tolerized mice (8.77).

We also examined cytokine production from splenocytes obtained from the same experimental groups. Briefly, splenocytes collected at day 8 were stimulated in vitro with PMA/ionomycin or with plate-bound anti-CD3 and anti-CD28 mAbs. Supernatants were analyzed for IFN-γ, IL-4, and IL-10 levels. As expected, splenocytes from CpG 1826-treated mice produced less IL-4 and IL-10 compared with tolerized controls. Surprisingly, cotreatment with 4-deoxy α-GalCer increased IL-4 (p < 0.05) levels relative to the CpG 1826-treated group, although differences in IL-10 levels did not reach statistical significance (p = 0.105). However, administration of PBS 57 with CpG 1826 significantly (p < 0.05) increased both IL-4 and IL-10 production (Fig. 9).

We observed robust IFN-γ production from splenocytes from untreated or control-tolerized mice. Surprisingly, splenocytes from tolerized mice treated in vivo with CpG 1826 made very little IFN-γ following anti-CD3 and anti-CD28 stimulation (Fig. 9). The implications of this finding will have to be further investigated. Interestingly, 4-deoxy α-GalCer, but not PBS 57, significantly (p < 0.05) increased IFN-γ levels compared with animals receiving CpG 1826 alone (Fig. 9).

Recent work has identified a number of microbes expressing cell wall lipids capable of activating NKT cells (19, 27, 28, 29, 30). Among these are Gram-negative, LPS-negative α-proteobacteria of the Sphingomonas strain that express glycosylceramides recognized by both human and mouse NKT cells (19). To test whether exposure to such microbes had an effect on autoimmune disease induced by mercury, groups of A.SW mice received a heat-killed suspension of S. capsulata, with or without HgCl₂ injections. Administration of bacteria alone did not induce ANoA; however, the group receiving both bacteria and mercury showed significantly increased autoantibody titers and serum IgG2a production (Fig. 10) while showing moderately increased IgE levels (Fig. 10). Thus, when compared with animals receiving mercury only, the heat-killed bacteria potentiated the autoimmunity-inducing effect of mercury.

We examined in this study the effects of NKT cell activation on mercury-induced autoimmunity. The two synthetic NKT cell ligands, PBS 57 and 4-deoxy α-GalCer, differentially modulated mercury-induced autoimmunity. When tested in a model of tolerance induction, NKT cell activation prevented tolerance establishment, but was by itself insufficient to break established tolerance. To disrupt tolerance, NKT cells required an additional signal, which was in this case provided by a TLR9 agonist. In addition, we observed that different NKT ligands can either synergize with or antagonize TLR9-induced breakage of tolerance. Lastly, bacteria-bearing NKT cell ligands have an exacerbating effect on our model of chemically induced autoimmunity.

The outcome of NKT cell activation in autoimmunity depends on a variety of factors, including the mouse strain studied (31, 32), the timing of administration of NKT cell-activating ligand (33), and the NKT cell-activating ligand itself (34, 35). Our results emphasize the importance of the mouse strain and the activating ligand. In the C57BL/6.SJL strain, both NKT cell ligands potentiated development of autoimmunity. However, in the A.SW strain of mice, NKT cell activation has profoundly different results depending on the glycolipid used to stimulate them. The 4-deoxy α-GalCer consistently exacerbated mercury-induced autoantibody production, whereas PBS 57 either had no effect on or down-regulated autoantibody levels. Thus, NKT cells are able to modulate heavy metal-induced autoimmunity differently depending on the ligand that activates them. These results are surprising because both ligands are derivatives of the widely used NKT cell ligand α-GalCer, with earlier studies indicating that their immunostimulatory properties are comparable to the parental ligand (16, 17). We expanded our study by comparing the two ligands in a model of tolerance induction. Because microbial infections can expose the immune system simultaneously to both NKT cell ligands and TLR ligands, we examined interplay between two NKT cell ligands and a TLR9 ligand, CpG 1826, in the breakage of established tolerance. The 4-deoxy α-GalCer-induced NKT cell activation strongly synergized with TLR9 stimulation to break established tolerance. This result is in agreement with a recent study wherein coexposure to a NKT cell ligand, α-GalCer, and various TLR ligands resulted in enhancement of dendritic cell (DC) maturation with increased T cell and B cell responses, suggesting that synergy between NKT ligands and different TLR ligands is a generalized phenomenon (36). Surprisingly, our results showed that NKT cell activation can also antagonize TLR signals. NKT cell activation by PBS 57 antagonized the effects of TLR9 stimulation, reducing TLR9-induced autoantibody production.

Because the two NKT ligands had opposite effects on autoantibody production when administered with mercury, we examined the in vivo effects of dual administration of NKT ligands and mercury. In mercury-exposed animals, PBS-57-induced NKT cell activation increased the frequency of T_regs, whereas 4-deoxy α-GalCer-induced NKT cell activation had the opposite effect. Moreover, in the tolerance model, PBS 57, but not 4-deoxy α-GalCer, restored the balance between T_regs and T_effectors, which had been disrupted by TLR9 stimulation. The importance of this balance has been highlighted by a recent study that demonstrated that a decrease in the ratio of T_regs to T_effectors correlated with progression of disease in NOD mice (37). Similarly, our study demonstrates that a treatment that disrupts tolerance decreases this ratio and suggests that this balance can be modulated by interplay between NKT cell agonists and a TLR9 ligand.

PBS 57 increased Th2 skewing in mercury-treated mice, whereas 4-deoxy α-GalCer increased Th1 skewing in tolerized mice challenged with CpG 1826 and mercury. Because autoantibody production in the mercury model is dependent on IFN-γ, but not IL-4 (8), these results suggest an explanation as to why PBS 57 negatively regulates mercury-induced autoantibody production and why 4-deoxy α-GalCer and PBS 57 have opposing effects on CpG 1826-induced autoantibody production in tolerized mice. Strikingly, because mercury increased Th2 skewing in PBS 57-treated, but not 4-deoxy α-GalCer-treated animals, these results also suggest that mercury (a common environmental pollutant) can modify the outcome of NKT cell activation by certain agonists.

We also examined effects of the NKT cell activation in a model of tolerance induction. Current models of tolerance induction hypothesize that Ag presentation in the steady state by DCs induces peripheral tolerance (38, 39). The nature of these tolerogenic DCs is, however, unclear. Certain studies suggest that tolerogenic DCs are phenotypically immature, whereas maturation imparts the ability to be immunogenic (35, 40, 41). However, in certain cases, phenotypically mature DCs can be tolerogenic (42, 43, 44, 45). Menges et al. (43) showed that DCs matured with repeated injections of TNF-α induce Ag-specific protection of mice against expermental autoimmune encephalomyelitis. Activation of NKT cells induces phenotypic maturation of DCs, causing up-regulation of CD40, CD80, and CD86 costimulatory molecules and MHC class II Ag-presenting molecules (46). This study (45) and others (38, 39) suggest that NKT cell activation could produce immunogenic DCs. In contrast, studies that examined the role of NKT cells in two models of tolerance have shown that this cell population is necessary for tolerance establishment (47, 48). In a model of tolerance establishment following oral administration of the contact allergen nickel, CD4⁺ NKT cells producing IL-4 and IL-10, but not IFN-γ, were required for the induction of tolerogenic APCs (47). Somewhat in contrast to these findings, we demonstrate that NKT cell activation was sufficient to prevent establishment of tolerance to mercury. In our experiment, NKT cell activation by PBS 57, but not 4-deoxy α-GalCer, prevented tolerance establishment. PBS 57 is also a stronger NKT cell agonist than 4-deoxy α-GalCer. It is a more potent inducer of several cytokines, including IL-17 (our unpublished observations). A recent study has demonstrated that IL-17 can block establishment of oral tolerance (49). Further studies are required to elucidate the mechanisms involved in PBS 57-induced prevention of tolerance.

A principal evolutionary role for NKT cells has been identified as antimicrobial defense (30); this cell type is required for clearance of a large number of microbes (18, 27, 28, 29, 30). Bacterial, viral, and parasitic infections have been implicated in the development and exacerbation of autoimmune diseases (6). A number of other studies have shown that exposure to chemicals (drugs or heavy metals) can also trigger or exacerbate autoimmune disease (19, 20, 49, 50). However, the effects of infections and chemicals on autoimmune disease have for the most part been studied separately, whereas human patients are likely to be exposed to both factors. Hence, in this study, we tested the effect of a commonly dispersed chemical and an infectious agent on autoimmunity. NKT cell ligand-bearing bacteria of the Sphingomonas strain are abundant soil microbes, and have been detected in the stools of 25% of healthy individuals (51). Although contact levels vary among individuals, mercury exposure is virtually universal because of its natural release in the environment, abundance as a pollutant, and presence in dental amalgams, cosmetics, preservatives, fumigants, and vaccine preparations (7, 52). As previously demonstrated, normally maintained immune tolerance is broken by exposure to mercury. Administration of both mercury and bacteria induced pronounced anti-nucleolar reactivity and exacerbated the heavy metal-induced autoimmunity. Unlike many bacterial species that primarily activate innate immunity via TLR signaling, S. capsulata bears triggers for both NKT cells and TLRs. These dual stimuli may have contributed to exacerbation of autoimmune manifestations.

Certain synthetic NKT ligands, such as OCH or the C-glycoside analog of α-GalCer, induce strongly different effects, owing to structural differences that result in highly polarized cytokine production. In contrast, PBS 57 and 4-deoxy α-GalCer have both been described as similar to α-GalCer. However, our study demonstrates that these NKT agonists, in combination with immunomodulator mercury, yield strikingly different effects. There could be several explanations for this. PBS 57 is more soluble due to the amide and the double bond in the acyl chain. Differential solubility can affect loading of lipids onto CD1d, which in turn can modulate uptake and presentation by CD1d-expressing APCs (53). Additionally, decrease in acyl chain length (54) and increased unsaturation (55) increase Th2 cytokine production. The latter findings may represent explanations as to why more IL-4 is detected with PBS 57. Mercury is an immunomodulatory agent that increases skewing toward Th2 (8). It is possible that the combination of the two agents results in increased Th2 skewing, and this could be, in part, contributing to the regulatory effects of PBS 57 in mercury-administered A.SW animals. Our study demonstrates that an environmental agent with immunomodulatory capacity can strongly influence the effect of NKT cell activation. This observation serves as a cautionary note when considering NKT cell activators as therapeutics in patients who may be simultaneously exposed to various environmental agents.

We thank Dr. Albert Bendelac for support.

The authors have no financial conflict of interest.