Introduction

IgE-mediated food allergies are prevalent, affecting approximately 4–8 % of children and 3–4 % of adults. Ingestion of foods triggers activation of allergic effector cells, including mast cells and basophils, by cross-linking IgE bound to the high-affinity FcεRI receptor on the cell surface. Degranulation of these cells is responsible for acute allergic symptoms that can affect the skin, lung, and gastrointestinal tract, and in severe cases may affect the cardiovascular system. For a comprehensive and current review of food allergy, readers are referred to [1•]. There is currently no approved treatment beyond allergen avoidance and acute management of symptoms in the case of accidental exposure. While allergen avoidance is theoretically a very effective management strategy, issues such as cross-contamination of foods during preparation and difficulties in interpreting food labeling make accidental exposures and reactions an unavoidable occurrence for those with food allergies. Furthermore, for children with multiple food allergies, dietary restrictions can have negative consequences on nutritional status. Therefore, treatments are needed that are both safer and more effective than the current standard of care. There are a number of clinical studies completed or in progress assessing the safety and efficacy of different forms of allergen immunotherapy for the treatment of food allergy. To date there are limited reports of long-term follow-up of subjects after oral immunotherapy (OIT), with mixed results on the achievement of long-term tolerance [2, 3]. Most reports agree that a minority of those starting treatment achieve long-term tolerance, and therefore there is a need to develop better therapies for the treatment of food allergy. The focus of this manuscript will be on pre-clinical studies supporting the next wave of human trials.

Immune Mechanisms of IgE-Mediated Food Allergy

Production of allergen-specific IgE is central to the pathogenesis of food allergy. Although non-IgE-mediated food allergies such as food protein-induced enterocolitis syndrome exist, the mechanism of reactions is distinct from the more common IgE-mediated food allergies, and those clinical entities will not be addressed in this manuscript. Presence of allergen-specific IgE is not sufficient to predict clinical reactivity, but increasing levels of IgE are associated with increasing risk of clinical reactivity, and values that are 95 % predictive of reactivity have been established for several foods [4, 5]. IgE binds with high affinity to FcεRI, a receptor composed of an α chain, β chain, and two γ chains on mast cells and basophils. Cross-linking of IgE on basophils and mast cells leads to degranulation of pre-formed granule contents, rapid production of lipid mediators, and a slower synthesis of cytokines. Factors that may contribute to clinical reactivity beyond the amount of IgE include affinity of IgE for the allergen or the ratio of specific to total IgE [6]. Functional assays of effector cell activation, such as skin prick testing, which reflects mast cell activation, or in vitro basophil activation assays have in some cases been shown to have added value when used together with specific IgE levels [7]. However, the gold standard for diagnosis remains the double-blind placebo-controlled food challenge. The discord between sensitization and clinical reactivity has led to the speculation that non-IgE-mediated pathways may contribute to anaphylaxis. In mouse, IgG antibodies can contribute to anaphylaxis through activation of macrophages or basophils [8, 9]. IgG-mediated activation of human neutrophils has been shown to occur with antigen stimulation [10, 11], but a contribution of food-specific IgG to food-induced reactions in humans has not yet been identified. IgG-mediated anaphylaxis has been demonstrated primarily with high-dose intravenous allergen challenge of mice; studies using more moderate doses or oral allergen challenge have shown that mast cells and IgE are the major mechanisms of food-induced anaphylaxis [12–14]. Identifying all contributing mechanisms of anaphylaxis in humans may be important for developing an effective therapy.

Pathways that suppress anaphylaxis may also contribute to the discordance between the presence of food-specific IgE and clinical reactivity. Food-specific IgG4 and IgA antibodies compete with IgE for allergen binding, and aggregation of the FcεRI receptor with the FcγRIIb receptor (by IgE–allergen–IgG1 complexes in human) provides negative signaling to allergic effector cells.

Mediators responsible for symptoms of food-induced systemic anaphylaxis include histamine and platelet-activating factor (PAF), shown by data from both mouse and human [15–17]. In contrast, gastrointestinal manifestations of food allergy in mice are not driven by histamine, but by PAF and serotonin [18].

Production of allergen-specific IgE is supported by the presence of Th2-biased antigen-specific CD4+ T cells. Subjects with IgE-mediated peanut allergy have an increased frequency of peanut-specific T cells compared to healthy controls and produce Th2 cytokines that are absent from peanut-specific T cells from healthy controls or those who have outgrown their peanut allergy [19–21]. An increase in allergen-responsive regulatory T cells has been demonstrated in subjects who have outgrown their non-IgE-mediated milk allergy [22] or are sensitized but tolerant to extensively heated forms of milk [23].

Based on our current knowledge of the immune basis of food allergy, therapeutic strategies have focused on reducing levels of allergen-specific IgE, enhancing levels of allergen-specific IgG or IgA, suppressing Th2 effector cells, or enhancing regulatory T cells through a variety of antigen-specific and antigen non-specific strategies.

Allergen Immunotherapy

Subcutaneous allergen immunotherapy is currently used for desensitization to aeroallergens and bee venom. Administration of peanut immunotherapy by the subcutaneous route was attempted and abandoned because of severe adverse reactions to the therapy [24, 25]. Oral desensitization to foods had been described in early case reports and series and in the past 10 years has been the subject of intense research focus. A search of trials listed at “clinicaltrials.gov” using the search terms “food allergy” or “food hypersensitivity” and “immunotherapy” provides 32 current and completed trials using OIT, 10 using sublingual immunotherapy (SLIT), and 5 using epicutaneous immunotherapy (EPIT). All but three of the trials use allergen immunotherapy without any additional adjuvant or immunomodulatory agents. Results are promising when desensitization, defined as protection from food-induced reactions while receiving therapy, is used as a primary outcome [26–28]. However, concerns about safety and long-term disease-modifying efficacy persist. Adverse reactions to OIT are not uncommon, and a significant number of subjects experience adverse reactions of sufficient severity or persistence to prevent continuation of immunotherapy. Despite the number of trials listed as ongoing or completed, most have a small number of subjects, are not placebo-controlled, and were not designed to look at long-term outcomes with tolerance challenges to assess whether treatment effects persist off therapy. In addition, different foods may respond differently to immunotherapy. Therefore, it has been difficult to obtain an accurate assessment of long-term efficacy of immunotherapy for foods. For a more comprehensive summary of the state of allergen immunotherapy for the treatment of food allergy, readers are referred to recent reviews on the topic [29, 30]. Even the most successful trials report that at least half of patients who begin immunotherapy do not achieve successful long-term tolerance [2, 3, 31•, 32], and therefore despite the pending outcome of long-term follow-up of current immunotherapy trials there remains an urgent need to improve the safety and efficacy of these still-experimental treatments.

The immune basis of tolerance induced by allergen immunotherapy for food allergy is still the subject of intensive research, but immunotherapy is associated with elevations in allergen-specific IgG4 and IgA, and reductions in diversity of epitopes recognized by allergen-specific IgE, skin prick test wheal size, allergen-induced basophil activation, and allergen-induced Th2 cytokine production [2, 33–35]. These parameters are associated with immunotherapy, but so far there have been no biomarkers described that successfully predict tolerance versus desensitization in response to immunotherapy. Tolerant individuals have a more robust IgG4 induction early in therapy compared to those who are desensitized, and a greater suppression of skin prick test wheal size [31•], but there is little definitive information yet on the mechanistic basis of sustained tolerance versus desensitization to foods. The inclusion of mechanistic studies in several current immunotherapy trials with tolerance challenges in the study design will likely yield significant new information in the next few years.

Routes of Immunotherapy

OIT is the farthest ahead of any immunotherapy treatments in showing efficacy for the treatment of food allergy. The default immune response to antigens delivered orally is immune tolerance mediated by regulatory T cells. However, as the oral route is also the route that normally leads to food-induced allergic reactions, safety is a major concern with OIT. SLIT and EPIT were proposed as routes that could have a significantly better safety profile yet retain the ability to induce tolerance. Sublingual antigen-presenting cells induce regulatory T cells similar to those of the intestinal tract [36], and the limited antigen dose applied to this route improves the safety profile [32]. The improvement in safety appears to be at some cost to efficacy [32, 37], although some groups report promising efficacy with SLIT for treatment of peanut allergy [38, 39]. Pre-clinical studies in mice show that EPIT leads to suppression of allergic inflammation in the lung and in the gastrointestinal tract, with a reduction of IgE, enhancement of IgG, and suppression of Th2 effector responses [40, 41]. In mice, application of antigen to non-damaged skin leads to acquisition of antigen by cutaneous dendritic cells that promote the development of regulatory T cells [42]. Data on the efficacy of EPIT for the treatment of peanut allergy are pending, with several trials currently in progress.

Anti-IgE with Immunotherapy

Three trials reported in clinicaltrials.gov include anti-IgE therapy, which is hypothesized not only to enhance the safety of the immunotherapy treatment, but also to improve efficacy by suppressing IgE-facilitated antigen presentation. Anti-IgE therapy increases the threshold of reactivity to peanut in peanut-allergic patients [43]. Two pilot studies have reported rush oral desensitization to milk and peanut in anti-IgE-treated patients [44, 45]. Although the data are clear that anti-IgE therapy increases the threshold of allergen reactivity in vivo and with human blood basophils, the data on the impact of anti-IgE therapy on presentation of antigen to T cells are less compelling [46]. Studies of immunotherapy and anti-IgE therapy for asthma and allergic rhinitis show a significant improvement of symptoms with anti-IgE therapy compared to immunotherapy alone, but the symptom improvement is only observed while anti-IgE therapy is maintained [47]. Results from placebo-controlled studies looking at long-term efficacy of immunotherapy with and without anti-IgE treatment for food allergy are pending.

Modified Allergens

One approach to make allergens safer for immunotherapy is to modify their structure to reduce IgE binding. Some allergens can be modified in this way simply through heating. The majority of children allergic to egg or milk can tolerate these antigens in baked goods or in forms that have been extensively heated [48, 49]. Heating denatures the proteins and destroys conformational epitopes, and there are also matrix effects that influence digestion and absorption of the allergens. Milk- or egg-allergic children enrolled in intervention studies in which they incorporated extensively heated milk or egg into the diet outgrew their unheated egg or milk allergy more quickly than a control group that received standard of care [50, 51], and this inclusion of milk or egg was associated with changes in immune parameters consistent with an immunotherapeutic response (elevation in IgG4, decreases in allergen-specific IgE). It is not known whether these immune changes underlie the generation of tolerance or are a biomarker of antigen exposure. In a murine model of OIT for food-induced anaphylaxis, extensively heated ovomucoid was unable to induce symptoms in mice yet when applied as OIT was as effective as native ovomucoid for the prevention of anaphylaxis [52].

Heating to reduce allergenicity is applicable to egg or milk, but not to antigens such as peanut where high heat increases allergenicity rather than reducing it. Digestion of cashew extracts with pepsin reduces the IgE binding of extracts, and in a mouse model of systemic immunotherapy, performed as well as intact cashew extract in protecting mice from cashew-induced anaphylaxis [53]. In contrast, enzyme hydrolysates of peanut have been shown to retain IgE binding and basophil activation potential [54]. Extract digests are unlikely to be used as subcutaneous immunotherapy because of difficulty in standardization and safety issues, but the pre-clinical study with cashew paves the way for peptide immunotherapy for nut allergy. Peptides that are too short to cross-link IgE but maintain T cell epitopes would have the capacity to generate T cell-mediated immunomodulation. Immuno-dominant peptides in the peanut allergens Ara h 1 [55] and Ara h 2 [56] have recently been identified with the goal of developing peptide immunotherapy. Peptide immunotherapy has been shown in pre-clinical studies as well as preliminary human trials to be effective for the treatment of cat allergy [57]. In addition to digestion and heating, allergens can be modified by chemical modification. Reduction and alkylation of Ara h 2 and Ara h 6, the major allergens of peanut, reduces IgE binding to the allergens [58]. In vivo studies demonstrate that reduced and alkylated Ara h 2 has decreased capacity to trigger anaphylaxis [59]. Experimental immunotherapy with Ara h 2 and Ara h 6 in mice is sufficient to protect against anaphylaxis to whole peanut extract [60], suggesting potential for these modified allergens as a safer form of immunotherapy.

Glycosylation of allergens can significantly modify their immunogenicity and allergenicity. Carbohydrate structures can both promote and suppress allergenicity. There is evidence that exposure of some allergens to high heat can enhance allergenicity through glycation, which allows for recognition of the allergens by pattern recognition receptors on antigen-presenting cells [61, 62]. The peanut allergen Ara h 1 binds to the lectin receptor DC-SIGN on human dendritic cells through carbohydrate residues and has Th2-promoting activity through a carbohydrate-dependent mechanism [63]. But glycosylation can also result in enhanced immune tolerance. The antigen BSA, when heavily conjugated with mannosides, binds to the DC-SIGN homolog SIGN-R1 in mice and promotes the development of immune tolerance through IL-10-producing regulatory T cells [64]. This conjugated antigen has enhanced affinity for dendritic cells of the intestinal lamina propria. Mannosylated ovalbumin has been shown in a murine OIT model to be significantly better than ovalbumin for clinical protection against ovalbumin-induced anaphylaxis [65]. It is not clear whether these effects are due to more efficient targeting to normally tolerogenic dendritic cells or whether binding to pattern recognition receptors such as SIGN-R1 in mice alters the phenotype of the dendritic cell to promote the generation of regulatory T cells. It was recently shown that the intestinal mucin Muc2 enhanced the regulatory phenotype of CD103+ intestinal dendritic cells through upregulation of known tolerogenic pathways (TGF-β, RALDH, IL-10) [66•]. Muc2 bound to a complex of galectin 3, dectin-1, and FcγRIIb on the surface of dendritic cells, resulting in the promotion of a tolerogenic phenotype and expansion of Foxp3+ regulatory T cells. Thus, allergens modified with appropriate carbohydrate structures may have potential as immunotherapeutic agents for the treatment of food allergy.

Adjuvants for Allergen Immunotherapy

Administration of allergen alone may be insufficient to fundamentally change the nature of the adaptive immune response to that allergen. Adjuvants that amplify either a Th1 response or a regulatory response may be necessary to sufficiently suppress the Th2-skewed immunity that drives the allergic response to foods. Many of these adjuvants are of microbial origin and range from whole heat-killed bacteria to co-administered purified microbial products to fusion proteins incorporating allergen and adjuvant in one. By binding to innate pattern recognition receptors on antigen-presenting cells, these adjuvants are thought to drive the T cell response away from a Th2 response. Adjuvants not only modify the nature of the immune response, but amplify the response such that significantly lower doses of allergen may be sufficient for an immunotherapeutic effect.

Subcutaneous administration of heat-killed Listeria monocytogenes together with native or modified peanut protein was shown to suppress peanut-induced anaphylaxis in mice [67] and dogs [68] by driving a strong Th1 response. A next generation approach used E. coli to express Ara h 1–3 allergens that had been modified to remove IgE binding epitopes [69•]. Heat-killed E. coli bearing modified Ara h 1–3 was administered rectally to peanut-allergic mice and resulted in sustained clinical protection of mice against peanut-induced anaphylaxis. This is a unique example of a novel immunotherapy for the treatment of peanut allergy that has been brought to clinical trial [70•]. Surprisingly, rectal administration of the therapeutic resulted in adverse reactions including anaphylactic reactions in two of ten peanut-allergic subjects tested. These results highlight the tremendous difficulty of administering a safe allergen immunotherapy to food-allergic subjects. Even allergens mutated to remove IgE-binding epitopes, and delivered rectally within bacterial encapsulation at microgram to low milligram doses, were sufficient to induce anaphylaxis in 20 % of subjects.

Other pre-clinical adjuvant approaches to the treatment of food allergy have involved synthetic oligonucleotides containing CpG motifs, which when administered together with peanut protein was significantly better than peanut alone in protecting mice from peanut-induced anaphylaxis [71]. CpG motifs activate dendritic cells and B cells via TLR9. A ragweed allergen–CpG conjugate was developed for immunotherapy, but despite promising results in the suppression of allergic symptoms [72], there have been no further published human trials on CpG–allergen conjugates. CpG packaged within virus-like nanoparticles have been tested in asthmatic patients with significant improvements in symptom scores [73], but this approach has not utilized any allergen-specific component. Virus-like particles containing the cat allergen Fel d 1 have been successfully used in pre-clinical studies of cat allergy [74]. These virus-like particles have not yet been tried for food allergy. A flagellin–ovalbumin fusion protein was reported to be highly effective in the prevention of symptoms in a model of gastrointestinal food allergy and modestly effective when administered as a therapeutic [75]. Flagellin binds to TLR5, which is highly expressed on some gastrointestinal dendritic cell subsets. Studies are clearly needed to optimize the choice of adjuvants (including Th1 vs. Treg optimization) as well as to improve safety of administration for immunotherapy with allergen–adjuvant combinations.

Tolerogenic Allergen–Cell Conjugates

One interesting approach to the generation of food allergen-specific tolerance has been the administration of syngeneic apoptotic cells coated with peanut allergen [76•]. Peanut-allergic mice were treated with intravenous syngeneic spleen cells that had been chemically conjugated with peanut. The conjugation procedure has previously been shown to induce apoptosis of the donor cells, which may contribute important tolerogenic signals to the recipient antigen-presenting cells that engulf the cell–allergen conjugate [77]. This approach was shown to be effective in both an asthma model and a food allergy model. The challenge of how to administer such an immunotherapy safely to food-allergic patients remains to be solved, although a similar approach was well tolerated in a phase I trial for multiple sclerosis [78].

Fcγ-Allergen Conjugates

An approach to making allergens safer for immunotherapy has been to prepare fusion proteins between allergen and human Fcγ chains. The hypothesis behind these constructs is that the Fcγ chain will bind to the inhibitory receptor FcγRIIb simultaneously to binding to IgE on the surface of effector cells including basophils and mast cells. Inhibitory signaling via FcγRIIb then shuts off activation of the allergic effector cells. Liu et al. [79] developed an Ara h 2–Fcγ fusion protein and tested its efficacy as immunotherapy in peanut-allergic mice. The fusion protein did not itself induce anaphylaxis in peanut-sensitized mice. The construct was not tried as a conventional immunotherapy, but pre-treatment of mice with the Ara h 2–Fcγ fusion protein modestly suppressed some measures of peanut-induced anaphylaxis.

Allergen Non-specific Therapies

For patients with sensitization to multiple foods, allergen immunotherapy with specialized allergen–adjuvant constructs may be of limited value. Single food immunotherapy has shown some evidence for bystander protection in mouse models [80] and in humans (assessing basophil activation to an unrelated food allergen) [81]. But other approaches have been tried in which allergen administration is not a feature of the therapy. This includes herbal medications that have immunomodulatory effects on basophils, mast cells, and T lymphocytes [82, 83]. Soy isoflavones have been shown to prevent sensitization through actions on dendritic cells, but when given after sensitization has already occurred can also suppress peanut-induced anaphylaxis [84]. Pharmacologic approaches include treatment with Janus kinase inhibitors [85, 86] and mTOR inhibitors, both of which are immunomodulatory and suppress mast cell expansion in the gastrointestinal tract. Anti-IgE treatment, as previously mentioned, increases the threshold of reactivity to peanut [43, 87], but has not been adopted as a therapeutic approach for food allergy outside of its addition to OIT in experimental settings.

There is significant interest in probiotics and the possibility of manipulating the microbiome for therapeutic purposes. Meta-analyses of probiotics for the treatment of allergic disease have shown a protective effect on allergic sensitization, but only when supplementation is started prenatally through maternal ingestion [88]. It was recently found that clostridia strains promote the development of regulatory T cells in the intestine [89], and when a mix of human clostridia strains were administered to mice, they could suppress the development of food allergy [90•]. It remains to be seen whether this type of approach could be used therapeutically, either alone or in the context of allergen immunotherapy.

Conclusions

There are no currently available approved therapies for food allergy. The majority of clinical research focus is currently devoted to allergen immunotherapy by the oral, sublingual, or epicutaneous routes, but improvements in safety and efficacy are needed. Pre-clinical work on food allergy immunotherapy safety has primarily focused on modifications to allergen structure, while improvements in efficacy are focused on adjuvant optimization at the pre-clinical level. Modification of the intestinal microbiome may offer a new approach to therapy for food allergy. There is a need for more pre-clinical studies to provide novel and unique approaches that are practical and translatable for use in human trials.