Osteopontin Promotes Protective Antigenic Tolerance against Experimental Allergic Airway Disease

Mechanisms of central and peripheral tolerance are crucial for maintaining immune system homeostasis and preventing exaggerated immune responses to intrinsically harmless self- or foreign Ags. Failure of this mechanism could lead to the development of chronic inflammation, such as allergic asthma and autoimmune diseases.

Because the incidence of allergic disease has risen dramatically, much effort has been put into determining the control mechanisms of peripheral tolerance to allergens in an attempt to find a treatment or prevention strategy for allergic disease. Allergic asthma is a disease caused by aberrant T_H2 immune responses to inhaled allergens leading to eosinophilic airway inflammation, mucus hypersecretion, and variable airway obstruction (1). Regulatory T (Treg) cells are important suppressors of dysregulated T_H2 responses to inhaled Ags, because constitutive or induced deficiency of these cells leads to severe asthmatic reactions (2, 3). Likewise, several groups have demonstrated that conventional dendritic cells (cDCs) and plasmacytoid dendritic cells (pDCs) are key regulators of T_H2 responses in allergic airway inflammation (4–7). As in many processes of immunoregulation, cytokines like TGF-β1 and IL-10 are also important regulators of tolerance to inhaled Ags (3, 8, 9).

Osteopontin (Opn) is a cytokine expressed by immune cells, such as activated T cells and dendritic cells (DCs), as well as by nonimmune cells, including tumor cells and stromal cells (10–12). In inflammatory conditions, Opn affects DC function (5, 13–18) and can drive T_H1, T_H2, and T_H17 effector immune responses (5, 11, 13, 14, 19, 20). In contrast, Opn is constitutively expressed by a great variety of cells under noninflammatory conditions (12, 21–23), but its physiological significance is largely unknown. For example, secreted Opn (sOpn) is expressed in the bone marrow (BM) and also upon inflammatory conditions, primarily in the form of thrombin-cleaved fragments (24, 25). Opn fragments have binding motifs for several integrins: the SLAYGLR motif specifically interacts with integrins α4β1, α9β1, and α4β7, whereas the RGD motif interacts with the αvβ3, αvβ5, αvβ1, and α5β1 integrins (12, 26). In addition, the C-terminal half of Opn interacts with certain CD44 variants (12, 27). Thrombin cleavage of Opn reveals the otherwise cryptic SLAYGLR domain, and this modification is vital for its interaction with α9β1 integrin (28).

Recent reports show that Opn is expressed in Foxp3⁺ Treg cells (29, 30), suggesting its possible role in immune regulation. In this study, we test whether Opn affects tolerance induction during intranasal (i.n.) administration of endotoxin-free Ag. Our results unveil a novel role for Opn as a tolerance enhancer against allergic airway disease, setting up an immunoregulatory milieu and potentiating CCR7-expressing pDC recruitment to the draining lymph nodes (dLNs). In addition, we reveal that, in the absence of pathogen-associated molecular patterns, sOpn, and specifically its integrin-binding SLAYGLR motif, induces low levels of IFN-β expression in Ag-primed pDCs. SLAYGLR-treated pDCs are highly efficient at suppressing allergic airway inflammation via type I IFN. Finally, we show that type I IFNs are crucial during antigenic tolerance induction against allergic airway disease.

BALB/c, C57BL/6J (B6), OVA-specific TCR-transgenic (Tg) C.Cg-Tg(DO11.10)10Dlo/J (DO11.10), C.129P2(B6)-IL-10^tm1Cgn/J, B6(Cg)-Il10^tm1.1Karp (Il10^GFP), B6.129-Ifnb1^tm1Lky/J (Ifnb^EYFP), B6.129S2-Ifnar1^tm1Agt/Mmjax (Ifnar1^−/−), and B6-Tg (C-type lectin domain family 4, member C [CLEC4C]–heparin binding EGF like growth factor [HBEGF]) 956Cln/J (pDC-specific type II C-type lectin [BDCA2]-diphtheria toxin receptor–[DTR]) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). B6-Tg Foxp3-DTR/EGFP (depletion of Treg cell [DEREG]) mice were provided by T.S. B6.129S6(Cg)-secreted phosphoprotein 1 (Spp1)^tm1Blh/J (Spp1^−/−) mice were kindly provided by Dr. L. Liaw (Maine Medical Center Research Institute, Scarborough ME). All mice used in this study were 8–10-wk-old females. Mice were housed at the Animal Facility of the Biomedical Research Foundation of the Academy of Athens and at the University Hospital Ghent (Ghent, Belgium). Use of mice in this study was reviewed and approved by the Bioethics Committee of the Biomedical Research Foundation of the Academy of Athens, the Veterinarian Office of Attica, and the Animal Ethics Committee of Ghent University. All procedures were in accordance with the National Institutes of Health Statement of Compliance (Assurance) with Standards for Humane Care and Use of Laboratory Animals (A5736–01) and with the European Union Directive 86/609/EEC for the protection of animals used for experimental purposes.

For tolerance induction, mice received 200 μg of EndoGrade OVA (Hyglos) i.n. in the presence of 2.5 μg recombinant Opn protein (rOpn) (R&D Systems), 72 ng of synthetic secreted Opn_134–153 fragments (frOpn; IVPTVDVPNGRGDSLAYGLR), or PBS (control). The RGD domain (Arg-Gly-Asp) of frOpn1 is scrambled to RAA (Arg-Ala-Ala). The SLAYGLR domain (Ser-Leu-Ala-Tyr-Gly-Leu-Arg) of frOpn2 is scrambled to LRAGLRS (Leu-Arg-Ala-Gly-Leu-Arg-Ser). frOpn3 has the RGD and SLAYGLR domains scrambled to RAA and LRAGLRS, respectively (Caslo Laboratory ApS). Opn oligopeptide fragments have been described previously (31). Myelin oligodendrocyte glycoprotein (MOG)_35–55 peptide (MEVGWYRSPFSRVVHLYRNGK; Caslo Laboratory ApS) was also used for tolerance induction (250 μg per mouse i.n.). Mice were euthanized 36–40 h later and analyzed. In certain experiments, tolerance was induced prior to allergic airway disease induction: mice received 200 μg of EndoGrade OVA (i.n.) on days −2, −1, and 0 in the presence of 2.5 μg of rOpn or 72 ng of frOpn1-3 synthetic fragments (i.n.). Control mice received PBS. Allergic airway disease (asthma) was subsequently induced: on day 10, mice were immunized with chicken OVA Grade V (Sigma-Aldrich; 50 μg) in 0.2 ml of aluminum hydroxide (alum) (Serva), followed by three challenges with 5% aerosolized OVA between days 16 and 18, as described (5, 15). Mice were euthanized 2 d after the last aerosol treatment (on day 20). DEREG mice received 1 μg of diphtheria toxin (DT; Sigma-Aldrich) (32) or PBS i.p. on days −3 and −2 (6 h prior to i.n. OVA administration). For the DT control group, non-Tg littermates were administered DT. For pDC depletion, mice received 225 μg of 120G8 pDC-depleting Ab (IgG2a; Dendritics, Lyon, France) (33) or an IgG2a isotype control Ab i.p. on days −6, −5, −4, and −3. As in Fig. 4A, mice received 200 μg of EndoGrade OVA i.n., along with 2.5 μg of rOpn, on days −2, −1 and 0 and were euthanized 7 d later. For pDC depletion in BDCA2-DTR mice, 120 ng of DT was administered i.p. on days −4 and −3 in the 7-d tolerance-induction protocol (as in Fig. 4A) (34). Efficient pDC depletion (≥95%) from dLNs in both approaches was determined by FACS analysis (data not shown). For neutralization of IFNAR1, mice received 20 μg of a polyclonal affinity-purified neutralizing Ab to mouse IFNAR1 (clone MAR1-5A3; eBioscience) or goat IgG isotype control (R&D Systems) i.p. 2 h before tolerance induction. For allergy induction, after IFNAR1 neutralization and tolerance, mice were immunized with chicken OVA on day 15, and OVA challenges were performed between days 21 and 23.

Lung function was measured in mice 24 h after the final OVA challenge (day 19) by whole body plethysmography (Buxco Technologies) to calculate enhanced pause (Penh). Responses to inhaled methacholine (Sigma-Aldrich) at concentrations of 3–100 mg/ml were measured for 1 min, as previously described (5).

Bronchoalveolar lavage (BAL) harvesting and analysis were performed as previously described (35, 32). For histological analysis, paraffin-embedded (4-μm) lung sections were stained with H&E and quantified as previously described (35). Goblet cells were quantified in Periodic acid–Schiff (PAS)-stained lung sections (5). A semiquantitative scoring system was used to grade the area of lung infiltrated, as previously described (36). Goblet cells were counted on PAS-stained sections using an arbitrary scoring system, as previously described (36).

For pDC generation, BM cells were isolated and cultured with recombinant human Flt3 ligand, as described (37). On day 11, 7AAD⁻CD3⁻CD19⁻CD11c⁺B220⁺PDCA1⁺Siglec-H⁺ pDCs were sorted to ≥98% purity using a FACSAria III, after enrichment with a CD11c MicroBeads Kit (Miltenyi Biotec). Sorted pDCs from BM cultures were conditioned for 16 h with 100 μg/ml EndoGrade OVA or LoTox Dermatophagoides pteronyssinus allergen 1 (Derp1; Indoor Biotechnologies) in the presence of 250 ng/ml rOpn, 18.2 ng/ml frOpn1 or frOpn3, or PBS. After the culture with OVA, pDCs were washed and analyzed or transferred i.v. via the tail vein (10⁶ cells per mouse). Allergic asthma was induced in mice 7 d later. Mice were euthanized 2 d after the last OVA challenge (Figs. 7B, 10A).

Isolated dLN cells (2 × 10⁵–10⁶) were cultured with 125 μg/ml OVA (Sigma-Aldrich) for 48 h. We performed proliferation assays with thymidine incorporation, as previously described (38). For certain experiments, proliferation of cells was measured as the percentage of Edu⁺ cells by FACS, using a Molecular Probes kit. For cytokine measurements, we used ELISA kits for IL-5 and IFN-γ (both from BD Biosciences), IL-4 and IL-13 (both from R&D Systems), and IFN-β (BioLegend).

Freshly isolated live (7AAD⁻; BD Biosciences) dLN cells and in vitro–derived BM cells were stained with combinations of fluorochrome-conjugated Abs to CD4–Pacific Blue or CD4-PE/Cy5 (clone GK 1.5); CD3-PE/Cy7, CD3-PE/Cy5, CD3-PE, or CD3-FITC (17A2); CD11c-PE/Cy7, CD11c-FITC, or CD11c-Orange 605 (N418); CD11b-PE/Cy7 or CD11b-FITC (M1/70); B220-PE or B220-BV 510 (RA3-6B2); CCR7-PE (4B12); Siglec-H–Pacific Blue or Siglec-H–FITC (551); PDCA-1–PE or PDCA-1–FITC (927); CD19-PE/Cy7 or CD19-PE/Cy5 (6D5); CD25-PE (PC61.5) (BioLegend); and T1ST2 (DIH9) (T1ST2-FITC [MD Biosciences] or T1ST2--PE [BioLegend]). For intranuclear staining of Foxp3, a permeabilization kit and Abs (Foxp3–Pacific Blue or Foxp3-PE/CyC5; clone FJK-16s) were used (eBioscience). Flow cytometric measurements were performed using an Attune Acoustic Focusing Cytometer (Applied Biosystems) and a FACSAria III (BD). FACS sorting of pDCs was performed using a FACSAria III. Data analysis was performed with FlowJo software (TreeStar).

Sorted pDCs from lymph nodes (LNs) and spleen of BALB/c mice were treated with rOpn (500 ng/ml) for 18–20 h and assayed for migration in response to chemokine CCL19 or CCL21 (200 ng/ml) (R&D Systems). The lower chambers of Transwell plates (QCM Chemotaxis Cell Migration Assay, 24-well [5 μm], colorimetric; Millipore) were filled with 500 μl of serum-free medium in the presence or absence of chemokines. DCs (10⁵ cells per 200 μl) resuspended in serum-free medium were deposited in the upper chambers of the Transwell plates and allowed to migrate for 4 h at 37°C in 5% CO₂. For each experiment, pooled total cells from spleen and inguinal and mesenteric LNs of mice (n = 8) were used, and pDCs were isolated by sorting.

BM-derived pDCs, pretreated or not with OVA and/or rOpn, were cultured with naive DO11.10 CD4⁺ T cells for 3 d at a 1:5 ratio. T cells were harvested and cultured in the presence of 1 ng/ml recombinant mouse IL-2 (R&D Systems) for an additional 7 d. Suppressive activity was assayed on 10⁵ freshly purified CFSE (Invitrogen)-labeled DO11.10 CD4⁺ T cells stimulated with 10⁴ irradiated BALB/c splenocytes, with 1 μg/ml OVA_323–339 peptide (Caslo Laboratory ApS), in the presence or absence of 10⁵ DC-stimulated T cells. CFSE uptake was assayed 7 d later (Fig. 7A).

Total RNA was extracted from cells isolated from dLNs with anti–mPDCA-1 and CD11c MicroBeads (Miltenyi Biotec) and further purified with FACS sorting. cDNA synthesis was performed as described (31). Primers were designed using the Primer3 program and are shown in Supplemental Table I (MWG Eurofins). Hprt, Foxp3, Il-10, Il27p28, and Spp1 primers were described previously (31). Real-time PCR was performed and analyzed as previously described (31). The reference gene used for real-time PCR analysis was Hprt.

Data were analyzed using Prism 7 software (GraphPad). The two-tailed Student t test was used for statistical analyses of two-group comparisons. Multigroup comparisons were performed using two-way ANOVA, followed by the Bonferroni correction for the multiplicity of tests. All results are presented as mean ± SEM. In all experiments, statistical significance was defined as *p ≤ 0.0332, **p ≤ 0.0021, ***p ≤ 0.0002, and ****p < 0.0001.

To address whether Opn plays a role in tolerance induction, mice received endotoxin-free chicken OVA, together with endotoxin-free rOpn or PBS control i.n., and protection against disease was assessed using a well-established model of allergic asthma (39) (Fig. 1A). OVA-tolerized mice showed a significant decrease in BAL total cells and eosinophils, as well as lung tissue inflammatory scores and mucus secretion (Fig. 1B, 1C), compared with nontolerized mice. In addition, Penh in OVA-tolerized mice was significantly reduced (Fig. 1D). Importantly, in OVA/rOpn-tolerized mice, numbers of eosinophils in BAL (Fig. 1B), airway hyperresponsiveness (AHR) (Fig. 1D), lung leukocytic infiltration, and mucus secretion (Fig. 1C) were further reduced compared with OVA-tolerized mice. Furthermore, OVA/rOpn tolerization resulted in significantly reduced levels of IL-4, IL-13, and IFN-γ in BAL, as well as a reduction in OVA-specific responses in mediastinal lymph node (mLN) cell cultures and in T_H cell proliferation, compared with OVA tolerization (Fig. 1E–G). Studies from our group (5, 15) and other investigators (40–42) have shown that there are high levels of IFN-γ production in allergic airway disease. The above findings indicated that administration of Opn, along with OVA, promotes enhanced tolerance, conferring significant protection from disease development.

We next investigated whether Opn deficiency had any effect on tolerance induction. Tolerance induction in Spp1^−/− mice was not as effective as in Spp1^+/+ mice, because we did not note a significant change in the number of eosinophils or total cell numbers in BAL (Fig. 2A). In accordance, tolerogenic i.n. OVA administration in Spp1^−/− mice could not efficiently dampen the inflammation and mucus secretion in the lung, whereas it was very efficient in Spp1^+/+ mice (Fig. 2B). T_H2 cytokine production by OVA–stimulated mLN cells was also lower in OVA-tolerized Spp1^−/− mice compared with PBS-treated Spp1^−/− mice (Fig. 2C). OVA-stimulated mLN cells from OVA-tolerized Spp1^+/+ mice produced ∼50% lower levels of T_H2 cytokines compared with cells from PBS-treated Spp1^+/+ mice, whereas OVA-tolerized Spp1^−/− mice had a smaller reduction in cytokine levels compared with Spp1^+/+ mice (Fig. 2C). The reduction in IFN-γ levels was similar among groups (Fig. 2C). Finally, OVA tolerization of Spp1^−/− mice resulted in a smaller decrease in the percentages of CD3⁺ T proliferating cells in cultures of mLN cells compared with those from Spp1^+/+ OVA-treated mice (Fig. 2D). The above results strongly support that tolerance induction is more effective in the presence of Opn.

Migratory nonlymphoid tissue DCs transporting Ags to LNs are involved in promoting tolerance to self-antigens in the steady-state. Because pDCs constitute a tolerogenic DC subset (43–47), and Opn has a dual role in the recruitment of DC subsets (5), we analyzed pDC numbers in dLNs of OVA-tolerized mice and PBS-treated controls (Fig. 3A). Numbers of dLN pDCs were increased in the OVA-treated group compared with the PBS control group (Fig. 3B). Mice treated with OVA/rOpn had increased percentages and significantly higher total numbers of pDCs in the dLNs at 36–40 h following treatment compared with OVA treatment (Fig. 3B). In contrast, cDC numbers in the dLNs of OVA/rOpn mice were significantly reduced and percentages were lower compared with OVA treatment alone (Fig. 3B). Therefore, administration of Opn during tolerance induction affects the balance of DC subsets in dLNs by increasing the numbers of pDCs and reducing the numbers of cDCs.

The importance of Opn for DC subset recruitment in tolerance was also demonstrated using Opn-deficient mice. Antigenic tolerance induction in Spp1^−/− mice resulted in a significant reduction in pDC numbers in their dLNs compared with Spp1^+/+ mice (Supplemental Fig. 1). At the same time, cDC numbers were significantly increased in dLNs of Spp1^−/− mice (Supplemental Fig. 1). Conclusively, these results show that tolerogenic administration of OVA in an Opn-efficient microenvironment enhances pDC accumulation in dLNs.

To explore the reason for the increased numbers of pDCs in dLNs of Opn-treated mice, we investigated the effect of Opn on CCR7⁺ pDCs. CCR7 is a chemokine receptor responsible for homing of DCs to dLNs (48–50). We quantified CCR7⁺ DC subsets in peripheral blood 36 h following OVA/rOpn tolerogenic administration (Fig. 3A). The percentages of CCR7⁺ pDCs per total pDC numbers were significantly elevated in the peripheral blood of mice treated with rOpn during tolerance induction compared with PBS-treated mice (Fig. 3C). The percentages of CCR7⁺ cDCs per total cDC numbers originating from the peripheral blood of mice that had been administered OVA/rOpn were significantly decreased compared with OVA-administered mice (Fig. 3C). Our results demonstrate that Opn leads to enhanced migratory CCR7⁺ pDC percentages per total pDCs in the blood.

We also found that OVA/rOpn-administered mice had significantly higher levels of Ccl19 and Ccl21 (encoding CCL19 and CCL21 chemokines that bind to CCR7) expression in their dLNs compared with OVA-administered mice (Fig. 3D), possibly attracting the increased numbers of CCR7⁺ pDCs to dLNs. Indeed, in vitro transmigration assays showed that sorted pDCs from naive mice pretreated with rOpn had a 2-fold increase in CCL19-induced chemotaxis compared with PBS-treated pDCs (Fig. 3E). The above findings demonstrate that, during tolerance induction, Opn can differentially regulate the percentages of CCR7⁺ DC subsets, thereby affecting their chemotactic migration to dLNs in favor of pDCs. Moreover, the observed increased gene expression of the CCR7 ligands CCL19 and CCL21 provide an extra explanation for the rOpn-mediated migration of pDCs to dLNs.

Because Ag administration for tolerance induction leads to generation of Treg cells (51), we investigated whether Opn had an effect early on this process. We administered OVA with rOpn (OVA/rOpn) or PBS i.n. to mice for three consecutive days, and mice were analyzed 1 wk later (Fig. 4A). OVA/rOpn-treated mice showed a nearly 2-fold increase in the percentages, as well as in the total numbers, of CD4⁺Foxp3⁺ Treg cells in dLNs compared with OVA-treated mice (Fig. 4B). Increased numbers of Foxp3⁺ Treg cells in OVA/rOpn-tolerized mice were accompanied by significantly enhanced Foxp3 expression in dLN cells (Fig. 4C). In addition, OVA/rOpn treatment induced a significant increase in the expression levels of immunoregulatory molecules, such as Il10, Ido, Tgfb1, Fgl2, and Il27, in dLNs compared with OVA treatment (Fig. 4C). Conclusively, Opn administration, along with Ag, promotes Foxp3⁺ Treg cell accumulation, resulting in a highly tolerogenic microenvironment in the dLN.

To directly test whether increased numbers of Foxp3⁺ Treg cells mediate the observed rOpn promotion of tolerance, we depleted Foxp3⁺ Treg cells prior to tolerance induction using DEREG mice (52) (Fig. 5A). Induction of tolerance with rOpn could not protect Treg-depleted mice, which exhibited increased allergic responses, as demonstrated by increased eosinophilia and lymphocytosis in the BAL, increased T cell proliferation, elevated lung inflammation with enhanced mucus production, and increased OVA-specific cytokine responses compared with Treg-sufficient mice (Fig. 5B–E).

To address whether the Opn-mediated increase in pDC numbers was responsible for the observed accumulation of Foxp3⁺ Treg cells, we depleted pDCs by administering 120G8 pDC-depleting Ab (33) prior to induction of tolerance and examined the accumulation of Foxp3⁺ Treg cells in dLNs. This depletion led to a significant reduction in CD4⁺Foxp3⁺ Treg cell numbers compared with OVA/rOpn administration in non–pDC-depleted mice (Fig. 6A), resulting in exacerbated allergic airway disease (Fig. 6B). We observed similar effects of rOpn on Treg cellularity in LNs after in vivo depletion of pDCs in BDCA2-DTR Tg mice (34) (Supplemental Fig. 2). Thus, Opn-mediated pDC accumulation is responsible, at least in part, for the higher numbers of Treg cells in dLNs. In addition, dLN cells from OVA/rOpn-treated pDC-depleted mice were significantly less capable of suppressing OVA-specific T cell proliferation (Fig. 6C) compared with dLN cells from OVA/rOpn-treated non–pDC-depleted mice, reaching an even lower suppressive ability than that of OVA-treated pDC-depleted dLN cells. Thus, OVA/rOpn treatment increased the numbers of pDCs, favoring the accumulation of Foxp3⁺ Treg cells that are crucial for tolerance maintenance.

We tested the suppressive function of in vitro OVA/rOpn-treated BM-derived pDCs by coculturing them with CD4⁺ T cells. T cells obtained from OVA/rOpn-treated pDC cultures significantly suppressed the responses of DO11.10 T cells to OVA compared with T cells obtained from OVA-treated pDC cultures (Fig. 7A).

We also adoptively transferred BM-derived pDCs, which were preconditioned in vitro with OVA or OVA/rOpn, into recipient mice prior to induction of allergic airway inflammation (Fig. 7B). Control mice received PBS-treated pDCs. Total and eosinophil cell numbers, as well as lung tissue inflammatory scores and mucus secretion, were lower in BAL of OVA/rOpn-treated pDC recipient mice compared with OVA-treated pDC recipient mice (Fig. 7B, 7C). Furthermore, OVA/rOpn-treated pDC recipients had significantly reduced OVA-specific responses compared with OVA-treated pDC recipients (Fig. 7D). Thus, OVA/rOpn treatment of pDCs rendered them more regulatory, indicating that Opn affects the intrinsic tolerogenic function of pDCs.

RNA expression analysis revealed that, during tolerance induction, mLN pDCs have increased expression of integrins αvβ3, α4β1, and α9β1 (data not shown). Because the SLAYGLR motif of Opn interacts with α4β1, α4β7, and α9β1 integrins (12), and the RGD motif interacts with the αvβ3 integrin (11, 12), we asked which Opn domain is responsible for the observed effects on pDCs during tolerance induction. Thus, synthetic frOpn_134–153, containing the RGD and the SLAYGLR integrin binding motifs, either intact or scrambled (31), were used with OVA to induce tolerance in mice, allowing us to unveil the involvement of the different integrin-binding motifs of Opn in pDC recruitment (Fig. 8A). frOpn_134–153 represents the thrombin cleaved fragment of Opn that reveals the otherwise cryptic domain SLAYGLR (28). OVA/frOpn1, containing an intact SLAYGLR motif but a scrambled RGD motif, induced the accumulation of higher numbers of pDCs in LNs compared with OVA and OVA/rOpn (Fig. 8A). Conversely, frOpn2, which contains an intact RGD, but a scrambled SLAYGLR motif, induced a reduction in pDC numbers in dLNs compared with all other treatments (Fig. 8A). When both motifs were scrambled, as in the case of frOpn3, the numbers and percentages of pDCs were not significantly altered compared with OVA (Fig. 8A, Supplemental Fig. 3A). OVA/frOpn1 induced higher percentages of pDCs in dLNs compared with OVA and OVA/frOpn3 (Supplemental Fig. 3A). These data revealed that the Opn SLAYGLR motif is responsible for pDC accumulation in dLNs. These data also reveal that the RGD motif suppresses this accumulation.

To examine the in vivo tolerogenic potential of frOpn1 in allergic disease, mice were treated with OVA, together with frOpn1 or frOpn3 for control, before the induction of allergic airway inflammation (Fig. 8B). OVA/frOpn1 was more successful than OVA/frOpn3 in promoting tolerance and, thus, protecting mice against allergic airway inflammation, as demonstrated by lower Penh values (Fig. 8C). BAL eosinophilia, as well as inflammatory score and mucus staining, was also dampened after OVA/frOpn1 treatment (Fig. 8D). In addition, treatment with OVA/frOpn1 led to reduced levels of OVA-specific responses (Fig. 8E) and numbers of T1ST2⁺CD4⁺ T cells in dLNs (Fig. 8D). Collectively, this dataset indicates that the Opn SLAYGLR motif significantly boosts tolerance.

We further asked whether OVA/frOpn1 treatment affects in vivo pDC immune profile 36–40 h after tolerance induction (Fig. 3A). In vitro, OVA/frOpn1 conditioning of pDCs promoted a 2-fold increase in Ifnb mRNA expression at 16 h (Fig. 9A). The observed difference in Ifnb expression was also reflected in the amount of IFN-β secreted, which was 5-fold higher in OVA/frOpn1-conditioned pDCs compared with OVA/frOpn3-conditioned pDCs (Fig. 9B). House dust mite endotoxin-free Derp1 was also used to test whether the increased IFN-β response could be observed using another allergen. Indeed, Derp1/frOpn1 in vitro–conditioning of pDCs resulted in higher Ifnb expression compared with Derp1/frOpn3 (Fig. 9C). We also measured higher levels of Ifnb expression in OVA/rOpn-conditioned pDCs compared with OVA/PBS-conditioned pDCs (Supplemental Fig. 3B).

RNA expression analysis using mLN pDCs isolated from mice treated with OVA versus OVA/rOpn showed differential expression of Ifnb compared to mice treated with OVA (data not shown). We found that, indeed, Ifnb expression was upregulated in mLN pDCs isolated from mice treated with OVA/frOpn1 compared with those isolated from mice treated with OVA/frOpn3 (Fig. 9E). To confirm the changes in IFN-β expression in vivo, we used IFN-β^EYFP reporter mice. OVA/frOpn1-tolerized mice had significantly increased numbers of IFN-β^EYFP+ pDCs and higher IFN-β expression per pDC (mean fluorescence intensity [MFI]) compared with OVA/frOpn3-tolerized mice (Fig. 9D). To test whether this effect is relevant to the Ag used, we tolerized IFN-β^EYFP mice with a self-antigen, MOG_35–55, with frOpn1 or frOpn3. MOG_35–55/frOpn1 administration to IFN-β^EYFP mice resulted in significantly increased accumulation of IFN-β^EYFP+ pDCs and IFN-β expression per pDC (MFI) in LNs (Fig. 9F). However, the increase was modest compared with Opn/OVA, possibly because MOG is a peptide. These results suggest that, under tolerogenic conditions, the Opn SLAYGLR motif can boost Ifnb expression in pDCs through a mechanism independent of the Ag used.

pDCs were primed in vitro with endotoxin-free OVA and frOpn1 or frOpn3 and transferred into mice prior to induction of allergic airway inflammation (Fig. 10A). Recipient mice were Ifnar1^+/+ or Ifnar1^−/−. pDCs treated with frOpn1 were significantly more potent at protecting Ifnar1^+/+ mice against allergic airway inflammation, as demonstrated by reduced lung airway inflammation, BAL eosinophilia, and goblet cell hyperplasia, compared with the control and OVA groups (Fig. 10A). Similar enhanced protection was also observed when pDCs were treated with rOpn and adoptively transferred into Ifnar1^+/+ mice (Fig. 7B–D). Actually, type I IFN produced by pDCs was crucial in maintaining this protective effect, because Ifnar1^−/− recipient mice were not protected against allergic asthma (Fig. 10A). These results demonstrate that the integrin-binding SLAYGLR domain of Opn renders pDCs more regulatory through type I IFN production in the context of allergic airway inflammation. In fact, pDCs cannot preserve their regulatory function in recipients that lack expression of type I IFNR. Finally, neutralization of IFNAR1 during the OVA tolerance-induction phase (Fig. 10B) resulted in reduced protection from allergic airway disease concomitant with increased eosinophilia in BAL (Fig. 10C) and increased OVA-specific responses (Fig. 10D). This indicates that type I IFN is absolutely necessary for effective antigenic tolerance induction in this context.

In this article, we unveil a new role for Opn, and particularly its SLAYGLR motif, in setting up a tolerogenic milieu driving antigenic tolerance induction and, thus, leading to protection from allergic airway inflammation.

The respiratory mucosa is constantly being exposed to a myriad of nonpathogenic environmental Ags. To protect against the immunopathological consequences of this constant stimulus, a default low noninflammatory T_H2 mechanism and/or a T cell–mediated tolerance mechanism is activated (53). The mechanism underlying these processes is not fully understood. In this study, we initially demonstrate that Opn induces tolerance by tilting the pDC/cDC balance in favor of anti-inflammatory pDCs, via differentially regulating their CCR7 expression. CCR7-dependent homing of DCs into dLNs is required for the induction of tolerance (50).

Opn promotes T_H2 effector responses when administered during the allergen-sensitization phase (5). In contrast, we have previously shown that Opn inhibits T_H2 responses during allergen challenge, and administration of rOpn during that phase is suppressive for experimental allergic asthma (5). In this study, we explore the role of Opn during a different phase, which is when Ag is administered during tolerance induction, prior to sensitization. Thus, Opn is administered i.n., prior to the sensitization phase, along with the allergen (antigen), to test its effects upon antigenic tolerance induction. Together, our results from this study, as well as from a previous study (5), show that administration of Opn during antigenic tolerance induction and during antigenic challenge is protective and, thus, can be used as a preventive or a therapeutic agent against allergic airway disease.

Regulatory cytokines, such as Tgfb1, Il10, and Fgl2, and molecules, such as Ido (54), were substantially upregulated in our Opn-driven tolerance model, creating an immunoregulatory microenvironment in the dLN. This tolerogenic milieu was accompanied by the accumulation of Foxp3⁺ Treg cells, which are very important for induction of tolerance (51). Mice that had enhanced Treg cell numbers due to rOpn administration showed suppressed effector responses and were protected from allergic disease. When Foxp3⁺ Treg cells were depleted, Opn-mediated tolerance boosting was abolished. In light of its well-established proinflammatory function (10–12), it was unexpected that Opn could be an inducer of Treg cell accumulation in vivo. In the absence of pDCs, rOpn administration was no longer capable of inducing increased numbers of Foxp3⁺ Treg cells in dLNs, demonstrating that rOpn affects Treg cell numbers primarily through its impact on pDCs. In contrast, after tolerance induction without the addition of Opn, Treg cell numbers do not appear to depend on pDCs, because pDC depletion did not reduce their numbers. These results show that rOpn conditions pDCs to enhance Treg cell accumulation. Functional flexibility and fostering of Treg cell responses are typical features of pDC involvement in tolerance (4, 46), as also revealed by our studies.

Thrombin cleavage of Opn occurs during homeostatic and inflammatory conditions (24, 25), and conformational changes after thrombin cleavage result in higher-affinity binding to certain receptors (26). In fact, the cryptic SLAYGLR motif is revealed upon thrombin cleavage of Opn, which is vital for its interaction with α9β1 integrin (28). Because the SLAYGLR motif of Opn protein interacts with α4β1, α4β7, and α9β1 integrins (12), and the RGD motif interacts with the αvβ3 integrin (11, 12), we used synthetic frOpn_134–153, containing both integrin binding motifs (intact or scrambled) (31). Thrombin cleavage of Opn also produces a fragment containing the C-terminal half of Opn, which interacts with certain CD44 variants (12, 27). In our settings, as well as in other settings (31, 55, 56), frOpn_134–153 has a greater efficiency compared with full-length Opn. Thus, it is possible that interaction of Opn with CD44 interferes with certain Opn effects, such as pDC recruitment. Future investigations will elucidate the role of Opn–CD44 interactions in tolerance induction.

Under endotoxin-free conditions, we demonstrated that the SLAYGLR domain of Opn (frOpn1) enhances IFN-β expression in pDCs exposed to protein or peptide Ag. Accordingly, we showed that adoptively transferred OVA/frOpn1-treated pDCs were more efficient at suppressing allergic airway inflammation in recipient mice. This regulatory function was indeed mediated by the observed upregulated production of IFN-β by pDCs, because Ifnar1^−/− recipient mice were not protected against allergic asthma. During the course of allergic airway inflammation, pDCs, as well as their type I IFN production, are suppressive for disease (4, 57–59). Our studies reveal that pDCs expressing higher IFN-β levels due to exposure to Opn are important for protective tolerance prior to disease induction. Importantly, our experiments point to a crucial role for type I IFNs during the induction of efficient antiallergic antigenic tolerance. Accordingly, IFNAR signaling promotes Treg cell development and function under stress conditions (60).

Opn administered during tolerance also resulted in a remarkable upregulation of the ligands of CCR7, CCL19, and CCL21 in dLNs, reminiscent of that observed in tumors (61). Specifically, high CCL21 expression by melanoma in mice was associated with an immunotolerant microenvironment, which included the induction of lymphoid-like reticular stromal networks, an altered cytokine milieu, and, most importantly, the recruitment of regulatory leukocyte populations (61). High Opn expression is prevalent in many types of malignancy (62); although Opn is considered proinflammatory, most of these tumors escape immune surveillance. Thus, it is possible that Opn and its mediated induction of CCL21 could also be involved in suppression of antitumor immunity.

In this article, we demonstrate that the SLAYGLR motif of sOpn enhances regulatory mechanisms when administered together with endotoxin-free Ag in a tolerogenic context. In this respect, Opn has a unique effect on immunity, differing substantially from its effects in the presence of danger signals. The SLAYGLR motif of Opn interacts with integrins (12). Addressing which integrin mediates the tolerogenic effects of the SLAYGLR motif of Opn on Ag-loaded pDCs, and primarily the induction of IFN-β, will assist in the design of therapies targeting tolerance in allergy. Finally, our results point to novel effects of Opn on Foxp3⁺ Treg cells that remain to be explored.

We thank M. Bessa, M. Willart, M. Plantiga, K. Vergote, S. De Prijck, and M. Van Heerswinghel for assisting with experiments and K. Deswarte and A. Apostolidou for flow cytometric sorting of cellular populations. We thank Amgen and Celldex Therapeutics for kindly providing recombinant human Flt3 ligand.

The authors have no financial conflicts of interest.