Correspondence: Jerry M. Wells, Host-Microbe Interactomics, Animal Sciences, PO box 338, 6700 AH Wageningen, The Netherlands. Tel.: +31 317 484 509; fax: +31 317 487 630; e-mail: firstname.lastname@example.org
Peanut allergy accounts for the majority of severe food-related allergic reactions and there is a need for new prevention and treatment strategies. Probiotics may be considered for treatment on the basis of their immunomodulating properties. Cytokine profiles of probiotic strains were determined by in vitro co-culture with human PBMCs. Three strains were selected to investigate their prophylactic potential in a peanut sensitization model by analysing peanut-specific antibodies, mast cell degranulation and ex vivo cytokine production by splenocytes. The probiotic strains induced highly variable cytokine profiles in PBMCs. L. salivarius HMI001, L. casei Shirota (LCS) and L. plantarum WCFS1 were selected for further investigation owing to their distinct cytokine patterns. Prophylactic treatment with both HMI001 and LCS attenuated the Th2 phenotype (reduced mast cell responses and ex vivo IL-4 and/or IL-5 production). In contrast, WCFS1 augmented the Th2 phenotype (increased mast cell and antibody responses and ex vivo IL-4 production). In vitro PBMC screening was useful in selecting strains with anti-inflammatory and Th1 skewing properties. In case of HMI001 (high IL-10/IL-12 ratio) and LCS (high interferon-γ and IL-12), partial protection was seen in a mouse peanut allergy model. Strikingly, certain strains may worsen the allergic reaction as shown in the case of WCFS1.
Oral administration of certain probiotic lactobacilli has shown promising results in human clinical trials and experimental animal models of inflammatory bowel disease, irritable bowel syndrome and allergy (Kalliomaki et al., 2010, Meijerink & Wells, 2010; Rijkers et al., 2010). The proposed mechanisms involve immunomodulation and enhancement of the epithelial barrier function (Wells et al., 2009) and differ among species (Christensen et al., 2002) and even between strains of the same species (Medina et al., 2007; Meijerink et al., 2010; van Hemert et al., 2010). Suppression of inflammatory reactions in allergy and colitis can be due to the modulation of dendritic cell (DC) function and subsequent expansion or induction of regulatory T cells (Tregs) producing anti-inflammatory cytokines such as transforming growth factor (TGF-β) and interleukin (IL)-10 (Delcenserie et al., 2008, Kwon et al., 2010). Additionally, certain strains of Lactobacillus can skew T-cell responses towards Th1 leading to an inhibition allergen-specific IgE production and an increase in Th1 cytokines (Matsuzaki et al., 1998; Pochard et al., 2002; Ivory et al., 2008). IL-12 promotes the differentiation of naïve T cells into Th1 cells and augments natural killer (NK) cell activity (Papamichail et al., 2004) leading to increased interferon (IFN)-γ and further skewing the immune response from a Th2 to a Th1 response. Probiotics may also stimulate intestinal epithelial cells to secrete thymic stromal lymphopoietin and TGF-β that promotes the differentiation of immature DCs into tolerogenic DCs. This may enhance oral tolerance through the induction of regulatory Foxp3+ T cells expressing gut-homing receptors (Kwon et al., 2010; Wells et al., 2010). Food allergy represents a failure to attain oral tolerance leading to elevated levels of Th2 cytokines such as IL-4, IL-5, higher levels of allergen-specific IgE and increased mast cell degranulation. There is evidence that differences in Treg activity may play a role in the development and resolution of food allergies (Karlsson et al., 2004; Smith et al., 2008; Shreffler et al., 2009). Peanut allergy deserves special attention as it is mostly frequently associated with anaphylaxis and fatal outcomes (Sampson et al., 1992). Furthermore, peanut allergy usually does not resolve itself and therefore has a high impact on the quality of life of the individual (Le et al., 2008). The prevalence of peanut allergic sensitization varies between 1% and 2% (US and Canada) to 10% in the UK, of which 2% have clinical allergy to peanut (Sicherer et al., 2010). Currently, allergen avoidance is the only way to prevent allergic responses (Lee & Burks, 2009) but in the case of peanut allergy, accidental ingestion is common owing to the trace amount of peanut in many food products and cross-contamination during food processing (Clark & Ewan, 2008).
As there are only a few studies comparing potential probiotic strains in vitro and in vivo, our aim was to investigate the effects of three well-characterized probiotic strains, which differ in their immunomodulatory properties, on the prevention of peanut allergy. Twenty-eight strains of Lactobacillus and Bifidobacterium isolated from different commercially available products were screened for their immunomodulatory properties in a co-culture assay with human peripheral blood mononuclear cells (PBMCs). Based on these results, L. plantarum WCFS1, L. salivarius HMI001 and L. casei Shirota (LCS) were tested for their prophylactic effect in a mouse model of allergic sensitization to peanut allergen.
Materials and methods
Twenty-eight different strains comprising 12 species of probiotics or potential probiotic strains from commercially available products were identified by 16S cDNA sequencing and tested in PBMC co-culture assays (Table 1). L. plantarum WCFS1 is a single colony isolate from L. plantarum NCIMB8826, which was originally isolated from human saliva (National Collection of Industrial and Marine Bacteria, Aberdeen, UK; Hayward & Davis, 1956). L. salivarius HMI001 was isolated from human breast milk. L. casei Shirota was isolated from a commercially available LcS-containing drink (Yakult) and was originally isolated from the human intestine. Bacteria were grown overnight to stationary phase in anaerobic conditions at 37 °C in the recommended medium, pelleted by centrifugation, washed twice in phosphate buffered saline (PBS, pH = 7.4) and stored in aliquots at −80 °C at approximately 1 × 108 colony-forming units (CFU) mL−1 in PBS containing 20% glycerol. The exact number of bacterial CFU in a thawed aliquot was determined by plating serial dilutions of the cultures on MRS agar or MRS plus cysteine for the Bifidobacterium sp.
Table 1. Strains used in study
DN 173 010
Urex Biotec Inc
Urex Biotec Inc
Human PBMC co-culture assays
The study was approved by the Wageningen University Ethical Committee and was performed according to the principles of the Declaration of Helsinki. Buffy coats from four blood donors (with informed consent) were obtained from the Sanquin Blood bank Nijmegen, Netherlands. PBMC co-culture assays and cytokine measurements were performed as previously described (van Hemert et al., 2010). The cells were left unstimulated or were stimulated with lipopolysaccharide (1 μg mL−1), Concanavalin A (ConA, 5 μg mL−1) or with bacteria (listed in Table 1) in a 1 : 1 ratio with mononuclear cells.
Female C3H/HeOuJ mice, specific pathogen-free, were purchased from Charles River (Lyon, France) and were maintained under barrier conditions in macrolon type III cages with wood chips bedding, at mean temperature of 22 ± 2 °C, a relative humidity of at least 40% and not exceeding 70% and a 12-h light/dark cycle. Drinking water and standard laboratory food pellets were provided ad libitum. The welfare of the animals was maintained in accordance with the general principles governing the use of animals in experiments of the European Communities (Directive 86/609/EEC) and Dutch legislation (The Experiments on Animals Act, 1997).
Preparation of peanut extract (PE)
Peanut was kindly donated by Imco Nut Products, the Nut Company (Doetinchem, Netherlands) and was supplied by Golden Peanut, plant at Alpharetta (Georgia). Peanut protein extract was made by blending 500-g peanut with 500 mL 20 mM Tris buffer (pH 7.2). PE contained proteins migrating at the same position as Ara h1 (14%), Ara h2 (6.5%), Ara h3 (50%) and Ara h6 (4.5%) in SDS–PAGE. Densitometry was used to estimate their abundance as percentage of total protein (as indicated). After 2-h blending at intervals of 20 min for 1 min at room temperature, the aqueous fraction (PE) was collected by subsequent centrifugation (10 000 g) and stored at −70 °C in aliquots.
Groups of eight mice were orally exposed to PBS (control) or PE plus cholera toxin (CT; Biological Laboratories, Inc, CA) (allergic sensitization) by intragastric dosing of PBS or 6-mg PE plus 10-μg CT in 200 μL. Oral exposure was performed on days 0, 1, 2, 10, 17 and 24. Mice (6 weeks of age) were intragastrically administered 1 × 109 CFU of the different lactobacilli strains, diluted in 300 μL 0.2 M NaHCO3, three times per week, starting 14 days prior to the sensitization phase (day 14) until scheduled necropsy (day 31). The effect of LCS was tested in a separate animal experiment to HMI001 and WCFS1. To compare the results of the different experiments, the measured parameters (i.e. mMCP-1 level, antibody levels and cytokines levels) were normalized to the vehicle-treated PE-sensitized mice. In PE-sensitized mice, a nonprobiotic-treated group was used as a control for the probiotic-treated groups. As a control for the PE-sensitized mice, nonsensitized mice were used; these mice only received PBS instead of CT and PE.
Measurement of serum mouse mast cell protease-1 (mMCP-I)
On day 30, mice were orally challenged with PE by intra-gastric gavage (0.4 mL of 30 mg mL−1) and blood collected after 30 min to determine the levels of mMCP-I by ELISA according to the manufacturer's instructions (Moredun Scientific Ltd, Midlothian, UK).
Measurement of serum IgG1, IgG2a and IgE antibodies
Levels of PE-specific IgG1 and IgG2a in serum were determined by ELISA and levels of PE-specific IgE by sandwich ELISA. Plates (NUNC Immuno Maxisorp plate, Roskilde, Denmark) were coated overnight with 10 μg mL−1 PE (for IgG1 and IgG2a detection) or with 1 μg mL−1 purified rat anti-mouse IgE (BD Pharmingen) in PBS, followed by 1.5-h blocking with ELISA buffer (50 mM TRIS, 136.5 mM NaCl, 2 mM EDTA, 0.05 Tween-20, 0.5% BSA; 7.2 pH). Each test serum was incubated for 2–2.5 h. As a reference serum for PE-specific antibody levels, pooled serum from PE-sensitized mice (intraperitoneal injection of PE plus alum weekly for 3 weeks) was used to generate a standard curve [undiluted reference serum contained 1000 arbitrary units (AU)]. For detection of PE-specific IgG1 and IgG2a, alkaline phosphatase-conjugated anti-IgG1 or anti-IgG2a was added (1 h at RT). Subsequently, alkaline phosphatase buffer (Sigma) was used for the colour reaction, which was stopped with a 3 M NaOH and absorbance was measured at 405 nm using a Benchmark Plus Microplate Spectrophotometer (Bio-Rad Laboratories, Richmond, CA).
To measure PE-specific IgE antibodies, a PE-digoxigenin (DIG) conjugate solution (diluted in High Performance ELISA Buffer, from Sanquin, San Diego, CA) was added (1 h at RT). The coupling of DIG to PE was performed according to the manufacturer's instructions (Boehringer Mannheim, Mannheim, Germany). The coupled proteins were separated on a sephadex G-25 column and the concentration of coupled protein was determined spectrophotometrically at 280 nm. After incubation (1.5 h at RT) with peroxidase-conjugated anti-DIG fragments (Roche Diagnostics, Mannheim, Germany), tetramethylbenzidine substrate solution was added and the colour reaction was stopped with 1 M H2SO4. Absorbance was measured at 450 nm. Blank wells and negative control serum (serum from unsensitized mice) were included as controls for each plate.
Cell culture and cytokine measurement
On the day of necropsy (day 31), mouse splenocytes (5 × 106 cells mL−1) were cultured in RPMI-1640 medium with 10% heat-inactivated FCS and 1% penicillin/streptomycin in the presence or absence of PE (200 μg mL−1) for 96 h at 37 °C and 5% CO2. Culture supernatants were stored in −80 °C for cytokine measurements by multiplex analyses (BenderMed Systems; mouse Th1/Th2 plex).
Data were analysed using GraphPad Prism 4 software. The significance of differences between group means was determined using one-way anova with the Bonferroni post hoc test. Statistical analyses were performed following logarithmic transformation (to achieve normal distribution). P < 0.05 was considered to be statistically significant.
Co-culture of hPBMCs with lactobacilli and bifidobacteria modulates production of IL-10, IL-12 and IFN-γ
PBMCs from four different healthy donors were cultured in the absence or presence of 28 different bacterial strains comprising 12 species. As controls, the PBMCs were stimulated with LPS or ConA, a polyclonal mitogen. The bacterial strains differed considerably in their ability to modulate PBMC immune responses (Fig. 1) as shown by the large range of cytokine responses: IL-10 (6–3789 pg mL−1: 632-fold), IL-12p70 (2–927 pg mL−1: 464-fold), IFN-γ (1.8 pg mL−1–7.3 ng mL−1: 4000-fold). The ranking of the strains according to the levels of induced cytokines was highly consistent between the donors.
For some species, the level of induced cytokines was similar for different isolates (strains) tested (e.g. L. salivarius), whereas for other species the strain variation was considerable, especially for the L. casei strains where IL-10 and IL-12p70 differed by three-fold and 35-fold, respectively. Interestingly, the B. animalis strain induced substantially lower levels of IL-10, IL-12p70 and IFN-γ compared to the other bacterial strains. Three strains with distinct immune profiles were tested for their prophylactic effect in a mouse model of allergic sensitization to peanut allergen. The L. salivarius strain HMI001 has a relatively high IL-10/IL-12 ratio compared to other strains and was predicted to reduce the development of allergy by enhancing regulatory mechanisms to counteract allergic sensitization. The LCS strain was predicted to attenuate exaggerated Th2 responses owing to the relatively high induction of IFN-γ which inhibits proliferation of Th2 cells. The L. plantarum strain WCFS1 was selected because it showed less pronounced cytokine profile in the PBMC assay (mixture of Th1 and Th2 promoting cytokines) than the other strains HMI001 and LCS.
PE-specific antibody responses are modulated by treatment with lactobacilli
To investigate if prophylactic treatment with probiotics modulates the PE-specific immune responses, mice were treated three times a week with 1 × 109 CFU of HMI001, WCFS1 or LCS starting 2 weeks before the PE sensitization phase and ending on day 31.
All PE-sensitized mice developed PE-specific IgG1, IgG2a and IgE antibody responses but not the nonsensitized control mice (Fig. 2). Administration of WCFS1 caused a significant increase in the PE-specific IgG1, IgG2a and IgE antibody levels, whereas administration of HMI001 led to a significant decrease in IgE levels (Fig. 2). In contrast, administration of the LCS strain had no significant effect on the PE-specific antibody titres.
In PE-sensitized mice, mast cell degranulation is modulated by treatment with lactobacilli
After a 6-week probiotic treatment and a 4-week oral exposure regime to PBS or PE with CT, all mice received an oral challenge with PE. Thirty minutes after PE challenge, the serum concentration of mMCP-I was measured as a read-out for the type I hypersensitivity response leading to mast cell degranulation. Control mice sensitized to PE had significantly higher levels of mMCP-1 in the serum after peanut allergen challenge compared to nonsensitized mice (Fig 2). Mice administered WCFS1 had significantly higher levels of serum mMCP-1 compared to the control, whereas serum mMCP-1 was lower in the mice given LCS or HMI001 (Fig. 2).
Treatment with lactobacilli modulated the ex vivo cytokine response to peanut allergen
Splenocytes from PE-sensitized and nonsensitized mice were stimulated ex vivo in the presence or absence of PE. Unstimulated splenocytes secreted low amounts of cytokines (data not shown). PE-stimulated splenocytes from the vehicle control group of peanut-sensitized mice produced increased amounts of IL-4, IL-5 and IL-10 compared to splenocytes from nonsensitized mice (Fig. 3). Prophylactic treatment with HMI001 reduced IL-5 production compared to PE-sensitized control mice, but this was not significant. A reduction in levels of induced Th2 cytokines IL-4 and IL-5 was also observed by treatment of PE-sensitized mice with LCS. In contrast, WCFS1 augmented the Th2 phenotype as shown by an increase in IL-4 production compared to the vehicle control (not significant) (Fig. 3). IL-10 and IFN-γ levels were not affected by any of the lactobacilli.
The immunomodulatory properties of 28 different bacterial strains, isolated from commercially available products, were evaluated in a co-culture assay with human PBMCs. The strain differences in IL-10 (632-fold) and IL-12 (464-fold) induction were much higher than reported previously (Miettinen et al., 1996, 1998; Niers et al., 2005; Foligne et al., 2007; Medina et al., 2007). These results are in line with other studies describing very different and even opposing effects of different species of Lactobacillus on DC activation (Christensen et al., 2002; Foligne et al., 2007; Medina et al., 2007). Different immunomodulatory mechanisms may play a role in the prevention of allergic sensitization. Three different potential probiotic strains were selected based on their PBMC immune profiles to investigate their immunomodulatory properties in an established mouse model of peanut allergy. Prophylactic treatment with L. salivarius HMI001 that induced the highest IL-10 to IL-12 ratio in vitro (Fig. 1) led to a significant decrease in IgE titres and a significant diminution in mMCP-1 (seven-fold lower) upon challenge with PE. The ex vivo Th2-associated IL-5 production was also reduced in HMI001-treated mice, although this was not significant (P = 0.15). A recent study showed that probiotics selected on the ability to induce a high IL-10/IL-12 ratio can suppress experimental immune disorders in mice, such as inflammatory bowel disease, atopic dermatitis and rheumatoid arthritis, by inducing Tregs (Kwon et al., 2010). This suggests a possible mechanism for the prophylactic effects of HMI001 in the peanut allergy model. There were no differences in the percentage of Tregs (CD4+ Foxp3+ CD25+ or CD25−) in HMI001-treated mice (data not shown), but their suppressive activity on activated T cells may have increased as recently described (Kwon et al., 2010). Naturally occurring CD4+CD25+ Tregs are important for the modulation of the PE allergic response (van Wijk et al., 2007a, 2007b), but it is possible that other regulatory cells such as Tr1 (IL-10 secreting) or Th3 (TGF-β secreting) cells play a role in probiotic mechanisms (Allez & Mayer, 2004).
LCS induced high amounts of IFN-γ and IL-12p70, but low amounts of IL-10 in co-culture with PBMCs (Fig. 1), and was selected on the basis that it might counterbalance the Th2 cell-induced hyper responsiveness, by promoting Th1 cell development via IFN-γ and IL-12. Although prophylactic treatment with LCS did not significantly alter the PE-specific antibody responses compared to vehicle-treated PE-sensitized mice, there was a reduction in the Th2 cytokine production (IL-4 and/or IL-5) in ex vivo stimulated splenocytes. This is compatible with a skewing of the immune response towards Th1 and the immunomodulatory properties of the strain in vitro. Furthermore, LCS treatment led to a significant decrease in mast cell degranulation (seven-fold lower) upon oral challenge with peanut allergen. LCS treatment was previously shown to reduce ovalbumin-specific IgE and IgG1 responses and diminished systemic anaphylaxis by promoting a dominant Th1-type response mediated by IL-12 induction in a mouse model for food allergy (Shida et al., 1998; Allakhverdi et al., 2007). Furthermore, LCS administration has been shown to attenuate seasonal allergic rhinitis in adults (Ivory et al., 2008).
The L. plantarum strain WCFS1 was selected because it showed to have a less pronounced phenotype (mixture of Th1 and Th2 promoting cytokines) in the PBMC assay compared to the other strains HMI001 and LCS. This mixed phenotype made it more difficult to predict its effect in vivo. In the PE sensitization model, treatment with L. plantarum strain WCFS1 increased PE-specific antibody levels fivefold compared to vehicle-treated peanut-sensitized mice, which was accompanied by a marked increase in mast cell degranulation after challenge. Additionally, WCFS1 treatment augmented the ex vivo production of the Th2 cytokine, IL-4 in PE-stimulated splenocytes. Taken together, these data indicate that treatment with WCFS1 increased PE-specific humoral response and the development of type I hypersensitivity to peanut allergen. Based on these results, we suggest that WCFS1 may be a useful strain to enhance immunity during infection or immune responses to vaccination, but not in conditions of allergy.
The mouse peanut allergy model used in this study mimics the clinical and immunologic characteristics of peanut allergy in human subjects (Li et al., 2000; van Wijk et al., 2007a, 2007b). In this mouse model, the anaphylaxis induced by intragastric sensitization and challenge with PE closely reflects the clinical characteristics in human subjects. This is based on the model described by Li et al. (2000) showing that peanut-induced anaphylaxis was IgE mediated, and mast cell degranulation and histamine release were associated with the anaphylactic symptoms. This implies that these results are relevant to human studies.
To date, randomized clinical trials of probiotics in allergic diseases have mostly focused on prevention or treatment of children with eczema and atopic eczema (recently reviewed; Kalliomaki et al., 2010). The first studies with the LGG probiotic suggested a therapeutic effect both in eczema and atopic eczema (Majamaa & Isolauri, 1997; Isolauri et al., 2000), whereas the most recent reports show an effect only in patients suffering from atopic eczema (Viljanen et al., 2005) or no effect at all (Brouwer et al., 2006; Folster-Holst et al., 2006; Gruber et al., 2007). To our knowledge, there are at least 12 random controlled trials on probiotics and allergic rhinitis, nine of which showed an improvement in quality of life, or reduced medication or symptoms owing to the use of probiotics compared to placebo (Vliagoftis et al., 2008; Kalliomaki et al., 2010). In the case of seasonal allergic rhinitis, five of the random controlled trials suggested an improvement in clinical outcomes. In contrast trials on the effect of probiotic administration on the treatment of asthma showed no positive effects. Although there is a general consensus that probiotic effects have to be demonstrated in human trials conducted in the final target population, research in the allergy area may still benefit from preclinical model studies like ours. It may help select the most suitable strains for different allergic manifestations and provide insights into their therapeutic or preventative potential in the different periods of life. When the probiotic mechanisms are understood in more detail, it might well be possible to prevent sensitization to allergens and to use combinations of probiotics with different immunomodulating activities to enhance their effects.
In vitro PBMC screening was useful in selecting strains with anti-inflammatory and Th1 skewing properties, and in case of HMI001 (high IL-10/IL-12 ratio) and LCS (high IFN-γ and IL-12), partial protection was seen in a mouse peanut allergy model. Even though there might be some discrepancy between in vitro PBMC and in vivo (in the case of WCFS1), our data clearly indicate that PBMC cytokine profiling might be useful to select the most promising candidate strains.
This work was carried out within the research program of the Netherlands Consortium for Systems Biology (NCSB), which is part of the Netherlands Genomics Initiative/Netherlands Organization for Scientific Research. Reference serum for IgG1, IgG2a and IgE ELISA and PE-digoxigenin were a kind gift to TNO from IRAS. M.M. gratefully acknowledges financial support for her PhD fellowship from TIFN, The Netherlands. The authors have declared that no competing interests exist.
M.M., J.M.W., K.V. and J.v.B. conceived and designed the experiments. M.M., N.T., M.Z.B. and B.H. performed the experiments. M.M., J.M.W., K.V. and J.v.B. analysed the data. J.v.B., M.Z.B., B.H. and K.V. contributed reagents/materials/analysis tools. M.M., J.M.W., K.V. and J.v.B wrote the paper. M.M., J.M.W., N.T., M.Z.B., B.H., K.V. and J.v.B. contributed to final manuscript.