Systemic Toll-like receptor stimulation suppresses experimental allergic asthma and autoimmune diabetes in NOD mice

Abstract

Background: Infections may be associated with exacerbation of allergic and autoimmune diseases. Paradoxically, epidemiological and experimental data have shown that some microorganisms can also prevent these pathologies. This observation is at the origin of the hygiene hypothesis according to which the decline of infections in western countries is at the origin of the increased incidence of both Th1-mediated autoimmune diseases and Th2-mediated allergic diseases over the last decades. We have tested whether Toll-like receptor (TLR) stimulation can recapitulate the protective effect of infectious agents on allergy and autoimmunity.

Methods and findings: Here, we performed a systematic study of the disease-modifying effects of a set of natural or synthetic TLR agonists using two experimental models, ovalbumin (OVA)-induced asthma and spontaneous autoimmune diabetes, presenting the same genetic background of the non obese diabetic mouse (NOD) that is highly susceptible to both pathologies. In the same models, we also investigated the effect of probiotics. Additionally, we examined the effect of the genetic invalidation of MyD88 on the development of allergic asthma and spontaneous diabetes. We demonstrate that multiple TLR agonists prevent from both allergy and autoimmunity when administered parenterally. Probiotics which stimulate TLRs also protect from these two diseases. The physiological relevance of these findings is further suggested by the major acceleration of OVA-induced asthma in MyD88 invalidated mice. Our results strongly indicate that the TLR-mediated effects involve immunoregulatory cytokines such as interleukin (IL)-10 and transforming growth factor (TGF)-beta and different subsets of regulatory T cells, notably CD4+CD25+FoxP3+ T cells for TLR4 agonists and NKT cells for TLR3 agonists.

Conclusions/significance: These observations demonstrate that systemic administration of TLR ligands can suppress both allergic and autoimmune responses. They provide a plausible explanation for the hygiene hypothesis. They also open new therapeutic perspectives for the prevention of these pathologies.

Figure 1. Stimulation of TLR pathways prevents allergic inflammation and airway hyper-responsiveness.

NOD mice were treated as described in the Methods section. Briefly, mice immunized with OVA on day 0 were challenged with OVA or NaCl (controls) on days 7, 8 and 9. Each TLR agonist or phosphate-buffered saline (PBS) was administered 24 hrs and 1 hr before the first challenge. OVA-challenged mice were treated with PBS or (A) P40 (200 µg/challenge/mouse), a TLR2 agonist, (B) Poly(I∶C) (100 µg/challenge/mouse), a TLR3 agonist, (C) LPS (100 µg/challenge/mouse), a TLR4 agonist and (D) R848 (100 µg/challenge/mouse), a TLR7 agonist. AHR to Mch was measured 24 hrs after the last challenge and total cell as well as eosinophils in BALF and cytokine and chemokine concentrations in lungs. Mice treated with TLR agonists showed a decreased AHR, eosinophilia and IL-4 and eotaxin production as compared to control mice (* p<0.05; ** p<0.01). These experiments were performed twice using 4 to 6 mice per group. One representative experiment is shown.

Figure 2. Treatment with the TLR7 agonist R848 prevents airway resistance and compliance.

NOD mice immunized with OVA were injected with R848 and challenged with OVA as described in the methods section. Lung resistance (R) and compliance (C) were measured 24 hrs after the last aerosol challenge with OVA or NaCl. Results are representative of one experiment out of 2 performed using 4 to 6 mice per group.

Figure 3. Stimulation of TLR pathways prevents spontaneous autoimmune diabetes in NOD mice.

A. Female NOD mice were injected i.p. with 200 µg P40, 100 µg Poly(I∶C), 5 µg LPS from S. minnesota, 5 µg Lipid A, 10 µg R848 or PBS once a week, starting at 3–4 weeks of age and for 20 consecutive weeks. Mice were monitored weekly for the advent of glycosuria. Significant prevention from diabetes was observed with most TLR agonists tested (* p<0.05; ** p<0.01) but not with Lipid A. Each panel represents one of 3 independent experiments using 8 mice per group. B. A shorter treatment with the TLR2 agonist P40 administered between 4 and 14 weeks also induced significant protection from diabetes (** p<0.01). C. Treatment with Poly(I∶C) was highly protective if started up to 10 weeks of age but not later (at 16 weeks of age).

Figure 4. Histological analysis of pancreas.

A. Representative photomicrographs of distinct patterns observed in pancreas sections. Histological examination of hematoxylin and eosin stained pancreas sections recovered from the various experimental groups was performed (n = 8 per group). Islet infiltration (insulitis) was scored by deducing the proportion of non-infiltrated islets (healthy islets) and of islets showing a non destructive peripheral insulitis (peri-insulitis) or an invasive/destructive insulitis (destructive insulitis). B. The relative degree of islet inflammation in mice treated with P40, Poly(I∶C), LPS or R848 is shown in a cumulative histogram as compared to PBS-treated controls.

Figure 5. MyD88^−/− NOD mice are sensitive to airway allergic inflammation but resistant to the development of autoimmune diabetes.

As compared to MyD88^+/+ NOD mice, MyD88^−/− NOD mice were hypersensitive to experimental allergic asthma. A. The left panel shows the data observed in MyD88^+/+ (n = 25) and MyD88^−/− (n = 35) mice immunized with our conventional protocol, namely immunization with 100 µg of OVA on day 0 and challenge with 50 mg/ml of OVA or NaCl on days 7, 8 and 9. Results are expressed in percentage of survival. The right panel shows the data observed when using an infraoptimal protocol according to which MyD88^+/+ (n = 10) and MyD88^−/− (n = 15) mice were immunized with 50 µg of OVA on day 0 and challenged only once with 20 mg/ml of OVA or NaCl on day 7. Results are also expressed in percentage of survival. B. Results show that, using the infraoptimal immunization and challenge protocol, the eosinophil recruitment in BALF was more important in MyD88^−/− as compared to MyD88^+/+ mice (* p<0.05). C. Monitoring for the cumulative incidence of spontaneous diabetes showed that MyD88^−/− NOD mice (n = 25) were fully protected from disease (littermate female MyD88^+/+ NOD mice (n = 60) showed a normal disease incidence reaching 80% by 30 weeks of age) (** p<0.01). D. Histological examination of hematoxylin and eosin stained pancreas sections recovered from 20-week-old MyD88^+/+ and MyD88^−/− NOD mice (n = 8 per group) showed that a great majority of islets in MyD88^−/−NOD mice were insulitis free.

Figure 6. Stimulation of the TLR/MyD88 pathway modulates immune regulatory cytokines and lymphocyte subsets.

A. In vitro stimulation of C57BL/6 mouse spleen cells with varying doses of different TLR agonists (P40, 1 or 20 µg/ml; Poly(I∶C), 1 or 10 µg/ml; LPS, 0.1 or 1 µg/ml; R848 0.1 or 1 µg/ml) induced the production of cytokines such as IL-10 (at 48 hrs) and TGF-β (at 72 hrs). Study of splenocytes from MyD88^+/+ or MyD88^−/− C57BL/6 mice confirmed that the effect is dependent on the MyD88 pathway. Results are expressed as mean cytokine level ± SD. Results are representative of three independent experiments. B. Circulating levels of IL-10 and TGF-β were detected following in vivo administration of TLR agonists. NOD mice were injected i.p. with 20 µg of P40, 100 µg of Poly(I∶C) or 10 µg of R848 (n = 8 per group). Control mice were injected with saline (PBS). Sera were collected 24 hours after the injection and cytokine levels were measured by ELISA (*p<0.05; ** p<0.01; *** p<0.005). C. The same conventional treatment protocol (described in Figure 3) with the TLR4 agonist LPS (5 µg/week/mouse) and the TLR3 agonist Poly(I∶C), that protected wild type NOD mice from diabetes, was applied to female NOD mice invalidated for CD28 (CD28^−/−), CD1d (CD1d^−/−) and IL-4 (IL-4^−/−). Results obtained showed that the Poly(I∶C)-induced protective effect was maintained in CD28^−/− NOD mice (*p<0.05) but not in CD1d^−/− and IL-4^−/− NOD mice. As a mirror-like image the LPS-induced protective effect was maintained in CD1d^−/− and IL-4^−/− NOD mice (* p<0.05) but not in CD28^−/− NOD mice. One representative experiment out of 2 performed is shown.

Figure 7. Treatment of NOD mice with TLR4 or TLR7 ligands or probiotics induces CD4⁺CD25⁺FoxP3⁺ Tregs.

Mice were injected i.p. with 5 µg of LPS, 10 µg of R848 or treated orally with the VSL#3 probiotic preparation 5 days a week for 2 weeks (n = 5 per group). Twenty four hrs after the end of treatment spleen cells were recovered, stained with labeled antibodies specific for CD4, CD25 and FoxP3. Representative flow cytometry plots representing proportions of CD4⁺CD25⁺ and CD25⁺FoxP3⁺ T cells (examined on gated CD4⁺ cells) are shown. In addition, the corresponding histograms showing the total proportions of FoxP3⁺ T cells (within the CD4⁺CD25⁺ and CD4⁺CD25⁻ compartments) are detailed for each experimental group. For the FoxP3 staining, the isotypic controls showed values ranging 0.02–0.09%.

Figure 8. Probiotic administration prevents from both allergic asthma and autoimmune diabetes: a TLR/MyD88 pathway-dependent effect.

A. NOD mice received 5 days a week for 6 weeks a preparation of probiotics (VSL#3, 5.10⁹ bacteria/mouse) and underwent the conventional OVA immunization/challenge protocol previously described. AHR as well as eosinophil counts in BALF and IL-4 levels in lung homogenates were measured. Results showed that probiotic treatment significantly prevented from experimental allergic asthma (** p<0.01; *** p<0.005). B. The same probiotic preparation (VSL#3; 5.10⁹ bacteria/mouse) was administered orally by gavage to female NOD mice three times a week starting at 4 weeks of age (n = 8 per group). Results obtained demonstrated a very significant disease protection (*** p<0.005). C. In vitro incubation for 24 hrs of peritoneal macrophages from NOD mice with increasing concentrations of the VSL#3 probiotic preparation induced a dose-dependent production of TNF-α and IL-10. The MyD88 dependency of the effect was demonstrated by the lack of effect when macrophages from MyD88^−/− mice were analyzed. D. The probiotic-induced protection from allergic airway inflammation was MyD88-dependent as illustrated by the comparative results obtained in MyD88^+/+ mice. The effect was illustrated here by the data on eosinophil counts in BALF showing that MyD88^−/− mice (immunized and challenged according to the infraoptimal protocol, see figure 5) were completely refractory to the probiotic-treatment effect as compared to MyD88^+/+ mice (** p<0.01). Results are representative of one experiment out of two.

Figure 9. Probiotic treatment and immune regulation.

A. The VSL#3 probiotic preparation was administered orally to female NOD mice 5 days a week for 2 weeks (n = 5 per group). Twenty four hrs after the last administration sera were collected and circulating TGF-β was measured: increased levels were found in mice treated with the active compound as compared to controls (* p<0.05). B. Mice underwent the conventional OVA immunization/challenge protocol previously described and were treated prior to the first OVA challenge with either a neutralizing monoclonal antibody to the IL-10R or an isotype-matched control. Results show that neutralization of IL-10 did not alter allergic inflammation (as assessed by eosinophil recruitment in BALF and IL-4 production). One representative experiment out of two is shown. C. NOD mice received 5 days a week for 6 weeks either the VSL#3 probiotic preparation or PBS and then underwent the conventional OVA immunization/challenge protocol previously described. Results obtained showed that the probiotic-protective effect was completely reversed (both in terms of reduction of AHR (** p<0.01) and of eosinophil recruitment in BALF (* p<0.05, ** p<0.01)) following IL-10 neutralization upon administration of an anti-IL-10 receptor monoclonal antibody prior to the first challenge. One representative experiment out of two is shown. D. Experimental allergic asthma was induced according to the conventional OVA immunization/challenge protocol already described in normal NOD recipient mice transferred with CD4⁺ cells purified from the spleen of probiotic- or control-treated syngeneic mice. Purified CD4⁺ cells were transferred 24 hrs before the first challenge. Results obtained showed that both AHR and eosinophil recruitment in BALF were significantly decreased (** p<0.01 for both parameters) in recipients of CD4⁺ cells recovered from probiotic-treated donors. One representative experiment out of two is shown.