A specific mixture of FOS derived from chicory inulin consisted of short-chain FOS [scFOS: oligofructose, Raftilose P95, degree of polymerization (DP) < 6] and long-chain FOS (lcFOS: long-chain inulin, Raftiline HP, average DP 23 or higher, <1% DP < 5) and was provided by Orafti (Wijchen, The Netherlands). FOS were added to the base recipe of the semi-purified cow’s milk protein-free pelleted AIN-93G diet (scFOS/lcFOS ratio 9:1, 1% w/w) at Ssniff Spezialdiäten GmbH (Soest, Germany) (Table 1). The AIN-93G diet without FOS supplementation was used as control diet. Both diets were similar in color and were kept in sealed packages at 4°C prior to use.

Specific-pathogen-free 6-week-old C3H/HeOuJ female mice were purchased from Charles River Laboratories (Erkrath, Germany). Upon arrival, all mice were randomly allocated to the control and experimental groups: sham-sensitized control group (n = 5/subgroup); sham no IT, sensitized control group (n = 6/subgroup); sens no IT, FOS supplemented group (n = 6/subgroup); FOS no IT, OIT group (n = 6/subgroup); OIT, and the OIT with FOS supplementation group; OIT + FOS (n = 6/subgroup) (Figure 1). Mice were housed in filter-topped macrolon cages (n = 5–6/cage) at the animal facility of Utrecht University, Utrecht, The Netherlands on a 12-h light/dark cycle with unlimited access to food and water. All mice were fed the AIN-93G control diet and were acclimatized for 6 days. Experimental procedures were approved by the Ethical Committee of Animal Research of Utrecht University and complied with the principles of good laboratory animal care following the European Directive for the protection of animals used for scientific purposes.

The timeline of the animal experiment is depicted in Figure 1. At experimental day (D) 0, 7, 14, 21, and 28, mice were sensitized i.g. to the cow’s milk protein whey (DMV International, Veghel, The Netherlands) dissolved in PBS (20 mg whey in 0.5 ml PBS, Lonza, Verviers, Belgium) using cholera toxin (CT) as an adjuvant (15 µg CT in 0.5 ml PBS, List Biological Laboratories Inc., Campbell, CA, USA). After sensitization (D35), the FOS and OIT + FOS groups received the FOS diet until the end of the protocol. The sham-sensitized control groups, whey-sensitized control groups, and OIT groups were fed the control diet throughout the experiment. Starting at D42, OIT was administered five times per week for three consecutive weeks (i.g. 10 mg whey in 0.5 ml PBS) (D42–D59). On D64, all mice were challenged i.d. in both ear pinnae (10 µg whey in 20 µl PBS/ear) to induce an anaphylactic response. The acute allergic skin response (i.e., ear swelling after i.d. injection), drop in body temperature, and severity of clinical symptoms were measured. Subsequently, mice received an i.g. challenge at D70 (50 mg whey in 0.5 ml PBS) and mast cell-derived mucosal mast cell protease-1 (mMCP-1) was measured in serum samples collected via cheek puncture 30 min after challenge. Finally, all mice were challenged intraperitoneally (i.p.) on D77 to stimulate T cell responses prior to organ collection (50 µg whey in 200 µl PBS). At 6 time points throughout the animal experiment (D0, D35, D50, D63, D71, and D78), subgroups of mice (n = 5–6/group) from each control- and experimental group were killed by cervical dislocation, and blood and organs were collected (Figure 1). Additional groups of donor mice were used in a follow-up experiment to perform the adoptive splenocyte transfer experiment and histological analyses in colon tissue.

The magnitude of the acute allergic skin response after i.d. injection of the allergen was measured as Δ ear swelling by subtracting the mean basal ear thickness from the mean ear thickness 1 h post-challenge (in duplicate in both ears). Ear thickness in micrometers was measured using a digital micrometer (Mitutoyo, Veenendaal, The Netherlands). After blinding the cages, all mice were anesthetized using inhalation of isoflurane to perform the i.d. injection and both ear measurements [n = 10–12/group, data represent the mean of two subgroups: mice scheduled for section at D71 and D78 (Figure 1)]. Body temperature was measured 45 min after i.d. challenge to monitor the anaphylaxis-associated drop in temperature using a rectal thermometer and severity of anaphylaxis symptoms was scored according to the method described by Li et al. (20).

At D71, pooled spleen suspensions derived from sensitized control- and OIT + FOS mice (n = 8/group) from a follow-up experiment were adoptively transferred to naïve specific-pathogen-free 6-week-old C3H/HeOuJ female mice (Charles River Laboratories, n = 6/group), which were fed the control diet and were housed under similar conditions as described earlier. After homogenization and red blood cell lysis, whole spleen suspensions and CD25-depleted fractions were intravenously (i.v.) injected into the tail vein of recipient mice (1 × 10⁶ cells in 100 µl PBS) prior to sensitization and i.d. and i.g. challenges as described in previous sections. CD25-depleted CD4+ fractions were obtained using a CD4+ CD25+ purification kit according to the manufacturer’s instructions (Miltenyi Biotec, Leiden, The Netherlands). Flow cytometric analysis of CD25-depleted fractions showed that <1% of the cells were positive for Foxp3. In addition, whole spleen suspensions showed on average 14% Foxp3 positivity. At D35 after transfer, recipient mice were killed by cervical dislocation 30 min after i.g. challenge, and blood and spleens were collected for further analysis.

Blood was collected via cheek puncture at D0, D35, D50, D63, D71, and D78 prior to sectioning of the mice and from one corresponding subgroup (n = 10–12 samples/group) and at D35 in case of the recipient mice (n = 6 samples/group). Blood samples were centrifuged (10,000 rpm for 10 min) and serum was stored at −20°C until analysis of mMCP-1, whey-specific antibodies, and galectin-9 by means of ELISA. Determination of whey-specific antibodies was performed as described previously (21). Concentrations of mMCP-1 in serum collected 30 min after i.g. challenge [D70; n = 10–12 samples/group, data represent the mean of two subgroups: mice scheduled for section at D71 and D78 (Figure 1) and D35 in recipients; n = 6/group] were measured by using a mMCP-1 Sandwich ELISA Kit (Mouse MCPT-1 ELISA Ready-SET-Go kit, eBioscience, Breda, The Netherlands) according to the manufacturer’s instructions. Serum collected from mice killed at D50, D63, and D71 was used to measure galectin-9 concentrations. Overnight incubation (4°C) with 100 µl of coating antibody in coating buffer (0.75 µg/ml, mouse galectin-9 affinity purified polyclonal goat IgG antibody, R&D Systems, Oxon, UK) in 96-wells high-binding plates (Corning Incorporated, Corning, NY, USA) was followed by a washing step. The wells were blocked for 1 h (RT) with 200 µl blocking buffer (PBS with 1% BSA) and washed prior to 2 h (RT) incubation of serum samples diluted (100×) in dilution buffer (PBS with 1% BSA and 0.05% Tween20) and a standard curve with recombinant mouse galectin-9 (serial 1:1 dilution starting with 500,000 pg/ml, R&D Systems). After washing, the plates were incubated with 100 µl capture antibody (0.75 µg/ml, mouse galectin-9 biotinylated affinity purified goat IgG antibody, R&D Systems) for 1 h at RT. Afterward, the plates were washed and incubated with 100 µl Strep-HRP (Sanquin, Amsterdam, The Netherlands) for 1 h at RT in the dark. The color reaction was initiated by adding 100 µl 3,3′,5,5′-tetramethylbenzidine (1-Step Ultra TMB, Thermo Fisher Scientific, Waltham, MA, USA), and the reaction was stopped with 4 M H₂SO₄ (50 µl). Optical density was measured with a Benchmark microplate reader (BioRad, Hercules, CA, USA) at a wavelength of 450 nm.

Spleen and MLN were collected (n = 5–6/group) at D0, D35, D50, D63, D71, and D78 and homogenized using a syringe and a 70-µm cell strainer. Red blood cells were lysed in the splenocyte suspensions using lysis buffer (8.3 g NH₄Cl, 1 g KHC₃O, and 37.2 mg EDTA dissolved in 1 L demi water, filter sterilized). Cell suspensions were either dissolved in RPMI 1640 medium (Lonza) supplemented with 10% fetal bovine serum (FBS) and penicillin (pen, 100 U/ml)/streptomycin (strep, 100 µg/ml, Sigma-Aldrich Chemicals, Zwijndrecht, The Netherlands) and β-mercaptoethanol (20 µM) prior to ex vivo antigen-specific stimulation assays or dissolved in FACS buffer (PBS with 1% BSA) prior to flow cytometry stainings.

At D63 and D71, small intestine LP tissue (n = 4/group) was collected to isolate lymphocytes as follows: fat and Peyer’s patches were removed from small intestine tissue and after washing in Hank’s Balanced Salt Solution (HBSS, Invitrogen, Life Technologies, Carlsbad, CA, USA) with 15 mM Hepes (Gibco, Life Technologies) at pH 7.2, longitudinally opened tissue was cut into small fragments (0.5 cm). After washing in HBSS/Hepes, tissue samples were incubated in HBSS/Hepes buffer supplemented with 10% FBS, pen/strep, 5 mM EDTA, and 1 mM dithiothreitol (DTT) for 20 min at 37°C (2×). Subsequently, tissue samples were washed in RPMI/FBS/DTT to remove EDTA followed by incubation in RPMI/FBS/DTT with collagenase type D (1 mg/ml, Roche Diagnostics Inc., Almere, The Netherlands) and DNAse (20 µg/ml, Sigma) for 45 min on a plate shaker at 37°C (2×). After digestion, remaining tissue fragments were re-suspended using a syringe (10 ml) and a needle (18 G) and suspensions were filtered with a 100-µm cell strainer afterward. LP-derived cell suspensions were washed in HBSS/Hepes and purified using a percoll (GE Healthcare, Uppsala, Sweden) 40–80% mediated separation after centrifugation. Purified cell fractions were washed to remove traces of percoll and cells were taken up in FACS buffer prior to flow cytometry stainings.

8 × 10⁵ cells/well in 200 µl culture medium (RPMI 1640, 10% FBS, pen/strep, β-mercaptoethanol) in 96-wells U-bottom plates (Greiner, Frickenhausen, Germany) were stimulated with either culture medium, anti-CD3 (1 µg/ml, eBioscience) or anti-CD3/anti-CD28 (10 µg/ml anti-CD3 and 1 µg/ml anti-CD28, eBioscience, transfer experiment) or whey (50 µg/ml). Polyclonal stimulation (48 h) and whey stimulation (96 h) were conducted at 37°C and 5% CO₂. Culture supernatant was collected and stored at −20°C until measurements of IL-5, IL-10, IL-13, and IFNγ production by means of ELISA according to the protocol described earlier for galectin-9. Purified rat anti-mouse coating antibodies (1 µg/ml for IL-5 and IFNγ and 2 µg/ml for IL-10 and IL-13), recombinant mouse cytokines for the standard curve, and biotinylated detection antibodies (1 µg/ml for IL-5, IL-10 and IFNγ and 400 ng/ml for IL-13) were purchased at BD Biosciences.

To increase the expression of latency-associated peptide (LAP) on the surface of MLN-derived lymphocytes, cells were incubated in culture medium (RPMI 1640, 10% FBS, pen/strep) and received polyclonal stimulation with anti-CD3/anti-CD28 (10 µg/ml anti-CD3 and 1 µg/ml anti-CD28, eBioscience) for 24 h at 37°C and 5% CO₂ prior to staining. Otherwise spleen, MLN, and LP-derived cell suspensions in FACS buffer were plated in 96-wells U-bottom Falcon plates (BD Biosciences, 1–0.5 × 10⁶ cells/well). The cells were incubated with anti-mouse CD16/CD32 (mouse Fc Block, BD Biosciences) in FACS buffer to block non-specific binding sites (15 min, on ice). Subsequently, cells were stained extracellularly with the following fluorescent antibodies (all purchased at eBioscience unless stated otherwise) in FACS buffer for 30 min on ice in the dark: anti-CD4-PerCpCy5.5 (1:100, clone RM4-5), anti-CD69-APC (1:100, clone H1.2F3), anti-CXCR3-PE (1:50, clone CXCR3-173), anti-T1St2-FITC (1:50, clone DJ8, mdbioproducts, St. Paul, MN, USA), anti-CD25-AlexaFluor 488 (1:100, clone PC61.5), anti-F4/80-APC-eFluor 780 (1:100, clone BM8), anti-CD103-APC (1:100, clone 2E7), anti-CD11b-PE (1:50, clone M1/70), anti-CD11c-PerCpCy5.5 (1:50, clone N418), anti-MHCII-FITC (1:100, clone NIMR-4), anti-CD45-PE-Cy7 (1:100, clone 30-F11), anti-CD4-FITC (1:100, clone GK1.5), and anti-LAP-PerCP-eFluor 710 (1:50, clone TW7-16B4). Cells stained for extracellular markers were fixed using 1% IC fixation buffer (eBioscience), and cells receiving additional intracellular staining for the transcription factor Foxp3 were fixed and permeabilized using the Foxp3 staining buffer set purchased at eBioscience according to the manufacturer’s instructions. Afterward, cells were incubated for 30 min on ice in the dark with anti-Foxp3-APC (1:50, clone FJK-16s) in permeabilization buffer. Live cells were distinguished from dead cells using Fixable Viability Dye eFluor 780 (FVD, 1:2000, eBioscience) and single cells were separated from aggregated cells based on forward/sideward scatter properties. Isotype controls were used for each antibody and cutoff gates for positivity were established using the fluorescence-minus-one technique. Fluorescence was measured on the FACS Canto II (BD Biosciences) and analyzed with Flowlogic software (Inivai Technologies, Mentone, Australia).

The colon was dissected (n = 3–6/group), opened longitudinally, washed in PBS, and Swiss roles were prepared by rolling the tissue from distal to proximal end with the mucosal side down. Tissue roles were fixed in formalin (10% v/v) and embedded in paraffin (Leica IG1150c, Leica Microsystems, Rijswijk, The Netherlands). Tissue sections (5 µm) were cut with a microtome (Leica Microsystems) and mounted on slides prior to deparaffinization and hydration. To stain mast cells in the tissue sections, May-Grunwald and Giemsa solutions were used according to the manufacturer’s instructions (Giemsa Stain, Abcam, Cambridge, UK). Intracellular Foxp3 expression was stained as described previously (22). After dewaxing, the sections were boiled in sodium citrate buffer (0.01 M) for 15 min. Then the sections were incubated with 0.2% Tween20 in PBS for 20 min. After blocking with 5% rabbit serum (Dako, Heverlee, Belgium) in PBS with 1% BSA (PBS/BSA) for 30 min, the sections were incubated overnight (4°C) with rat anti-mouse Foxp3 purified antibody (10 µg/ml, eBioscience). Afterward, sections were incubated with 3% H₂O₂ in PBS for 30 min. Detection of the primary antibody was conducted with a biotinylated rabbit anti-rat IgG (2.5 µg/ml in PBS/BSA, Jackson ImmunoResearch, West Grove, PA, USA) for 1 h (RT). Then sections were incubated with avidin biotin complex (ABC HRP kit, Vector Laboratories, Peterborough, UK) in PBS/BSA for 1 h. The color reaction was developed with 3,3′-diaminobenzidine (Sigma) and counterstaining was performed with hematoxylin. After dehydration, the sections were covered with Pertex mounting medium (Histolab, Göteborg, Sweden) and cover glass. Stained cells were counted per 100 intact cripts in each colon section.

One centimetre of the proximal colon was dissected and stored in RNAlater (Sigma) at −80°C until further processing (n = 5–6/group). After homogenization, RNA extraction was conducted using the Qiagen RNeasy isolation kit and the RNAse-free DNAse Set (Qiagen GmbH, Hilden, Germany) and cDNA was synthesized using the iScript™ cDNA synthesis kit (Biorad) according to the manufacturer’s instructions. RT² qPCR Primer Assays to measure pPIP5k1 (housekeeping gene), Foxp3 and IL-10 were purchased at SA Biosciences (Qiagen, German Town, MD, USA) and quantitative real-time PCR was performed on a CFX96 real-time PCR detection system (Biorad) using iQ SYBR green supermix as described previously (19). Foxp3 and IL-10 mRNA expression data were normalized to pPIPk1 and depicted as the fold change in expression compared to the sham-sensitized control group.

Short-chain fatty acid analysis was conducted as described elsewhere (23). Briefly, cecum content was collected (n = 5–6/group) and frozen at −80°C until further analysis. Samples were defrosted, homogenized by vortexing, and diluted in ice cold PBS (1:10). After centrifugation (13,000 rpm for 10 min), the supernatant was analyzed using a Shimadzu GC2010 gas chromatograph (Shimadzu Corporation, Kyoto, Japan) and acetic, propionic, butyric, iso-butyric, valeric, and iso-valeric acid concentrations were quantitated (based on 2-ethylbutyric acid internal standard).

Data are depicted as mean ± SEM and were statistically analyzed with GraphPad Prism software version 6.00 (GraphPad software, La Jolla, CA, USA) using one-way ANOVA and Bonferroni’s post hoc test to compare pre-selected combinations. Calculated p-values were adjusted for the total number of comparisons made and were considered statistically significant when p < 0.05. In addition, whey-specific antibody data were log transformed prior to testing. The anaphylaxis symptom scores were analyzed using Kruskal–Wallis test for non-parametric data with Dunn’s post hoc test to compare pre-selected combinations.

Sensitization to the cow’s milk protein whey increased the acute allergic skin response measured as ear swelling (p < 0.0001) and caused severe anaphylaxis (p < 0.0001) with a characteristic drop in body temperature (p < 0.0001) upon i.d. challenge (D64) compared to sham-sensitization (Figures 2A–C). An increased serum mMCP-1 concentration, indicative for mast cell degranulation, was observed upon i.g. challenge (D70) in the sensitized control animals compared to sham-sensitized control animals (p < 0.0001) (Figure 2D). OIT + FOS induced protection against these allergic symptoms: the acute allergic skin response was reduced after OIT + FOS compared to FOS (p = 0.0048) or OIT alone (p = 0.023) (Figure 2A), indicating the improved efficacy of the combination strategy. OIT + FOS protected against the anaphylaxis-associated drop in body temperature compared to sensitized control mice (p = 0.0045) and FOS no IT (p = 0.0123) (Figure 2B) and both OIT and OIT + FOS reduced the severity of anaphylaxis symptoms (Figure 2C) compared to the sensitized control (p = 0.0322 and p = 0.0002, respectively). After i.g. challenge (D70), OIT + FOS reduced mast cell degranulation (mMCP-1) compared to the sensitized control (p = 0.0321) (Figure 2D). The reduction in mMCP-1 coincided with reduced local mast cell numbers in colon tissue of OIT + FOS mice, but only compared to FOS supplementation alone (FOS no IT, p = 0.0227) (Figures 2E,F).

Oral sensitization to whey increased the level of whey-specific IgE, IgG1, and IgA in serum compared to sham-sensitized mice (p < 0.0001) (D35, D50, D63, D71, and D78) (Figures 3A–C). As observed after immunotherapy (D63), OIT increased whey-specific IgG1 and IgA levels compared to sensitized controls (p = 0.0026 for IgG1 and p < 0.0001 for IgA) and a similar pattern was observed for IgE (Figures 3A–C). FOS supplementation did not affect the OIT-induced rise in whey-specific antibodies, since no differences were observed between OIT and OIT + FOS. Whey-specific IgG1 and IgA levels decreased despite a series of challenges (D63–D78) in the OIT and OIT + FOS groups (Figures 3B,C). Interestingly, OIT + FOS protected against the rise in whey-specific IgE (D71) observed after i.d. and i.g. challenge, since levels were comparable to the time point preceding the challenges (D63) (Figure 3A).

No differences in Th1 and Th2 cell percentages were shown between OIT and OIT + FOS animals, however, differences were observed at D63 (after OIT) and D78 (after i.p. challenge) between OIT + FOS mice and sensitized controls (Figure 4). At D63, an increase in activated Th1 cells (CXCR3+ of CD4+ CD69+) was observed in the OIT + FOS group compared to the sensitized control group in the MLN (p = 0.0159) (Figure 4B) and compared to the FOS no IT group in the spleen (p = 0.0096) (Figure 4A). At D78, activated Th2 cells (T1St2+ of CD4+ CD69+) were increased in the sensitized control animals compared to sham-sensitized control animals (p = 0.0009) and both FOS (p = 0.0259) and OIT + FOS (p = 0.0419) reduced the percentage of Th2 cells in spleen (Figure 4D). No differences were observed in the MLN at D78, except the elevated percentage of activated Th2 cells in the OIT + FOS group compared to the sensitized control (p = 0.0297) and FOS no IT (p = 0.0121) groups (Figures 4E,F).

As shown in Figure 5A, an early increase in the percentage of CD4+ CD25+ Foxp3+ Tregs was observed halfway immunotherapy (D50) in the MLN of the OIT + FOS mice compared to OIT (p = 0.0457). The higher percentage of Foxp3+ Tregs in MLN coincided with reduced IL-5 concentrations upon antigen-specific stimulation of splenocytes derived from OIT + FOS mice (D50) compared to the sensitized control (p = 0.0365) (Figure 5I). A similar pattern was found for IL-10 (p = 0.0651; not significant) (Figure 5J). No differences in cytokine concentrations were observed after medium and anti-CD3 stimulation of splenocytes (Figures 5I,J). At D63, the percentage of Foxp3+ Tregs in LP-derived lymphocyte fractions suggests an increase in the OIT + FOS mice compared to sensitized control mice, however, these results are not statistically significant (Figure 5B). A tendency toward increased percentages of CD103+ CD11b− DCs in LP was observed in the OIT + FOS group compared to the sensitized control (p = 0.0889; not significant) after challenge (D71), but not after immunotherapy (D63) (Figure 5C). The percentage of LAP+ Tregs was increased in the OIT + FOS group compared to OIT at D71 in MLN (p = 0.0327) (Figure 5D). Levels of Foxp3+ cells and Foxp3 and IL-10 mRNA expression in colon tissue were not statistically different between the groups at D71 (Figures 5E–H). A trend (p = 0.0983) toward an increase in the number of Foxp3+ cells was observed in colon samples of FOS-treated animals compared to sensitized control animals (Figure 5F).

To confirm the involvement of the induced Foxp3+ Tregs in protection against allergic symptoms, pooled whole spleen suspensions and CD25-depleted fractions derived from sensitized control mice and OIT + FOS mice were adoptively transferred to naïve recipients. In contrast to body temperature (Figure 6B), challenge-induced acute allergic skin responses were significantly increased in the whey-sensitized recipient control mice compared to the sham-sensitized recipient control mice (p < 0.0001), indicating that cell transfer was not responsible for the allergen-induced ear swelling response (Figure 6A). OIT + FOS splenocyte transfer did not protect against the increased ear swelling response (Figure 6A). CD25 depletion induced a more severe drop in body temperature in the OIT + FOS recipients compared to the recipients of the whole spleen suspension (p = 0.0426) (Figure 6B). Similarly, challenge-induced mast cell activation was increased in mice which received the CD25-depleted fraction compared to the recipients of the OIT + FOS whole spleen suspension (p = 0.0003) (Figure 6C). In addition, CD25 depletion increased the IL-10 concentration in whey-stimulated splenocyte cultures derived from OIT + FOS recipients (p = 0.0125) (Figure 6E). A similar pattern was observed for IL-5, IL-13, and IFNγ; however, differences were not significant (Figures 6D,F,G). No differences in cytokine concentrations were observed after medium and anti-CD3/anti-CD28 stimulation (Figures 6D–G).

Bacterial fermentation of FOS leads to production of SCFA. After sensitization at D35, the mean total SCFA level was higher in the cecum of sensitized control mice than in sham-sensitized control mice (p = 0.017) (Figure 7A). Considering individual SCFA, a similar pattern was observed for butyric acid; however, differences between the groups were not significant (Figure 7B). The intestinal epithelium-derived factor galectin-9 was measured in serum halfway- (D50) and after immunotherapy (D63) and after challenge (D71) (Figures 7C–E). OIT + FOS increased serum galectin-9 levels at D63 (p = 0.0014 compared to sensitized control, p = 0.001 compared to FOS no IT and p = 0.0161 compared to OIT) (Figure 7D) and at D71 (compared to sensitized control, FOS no IT and OIT, all p < 0.0001) (Figure 7E).