QS-21 is a triterpene glycoside that can be purified from the soap bark tree (Quillaja saponaria) (106). It has been tested as a therapeutic vaccine adjuvant in clinical trials against cancer (lymphoma, leukemia, melanoma, breast-, prostate-, ovary-, or lung cancer) and infectious diseases (HIV-1 and hepatitis B) (81, 107).

The mechanisms underlying the adjuvant effects of QS-21 are not extensively studied. We know that induction of immune responses by QS-21 is mediated by a pronounced activation of innate immune cells: upon injection, QS-21 activates subcapsular CD11b⁺CD169⁺ macrophages in draining lymph nodes that in turn recruit neutrophils and DCs (108, 109). In DCs, QS-21 in turn triggers a MyD88-dependent inflammasome activation that involves a partial activation of the high-mobility group protein B1-TLR4 (109) and cathepsin B (shown for QS-21 in AS01), an inflammasome-dependent release of IL-1β, and the activation of both CD4⁺ and CD8⁺ T cells, and production of antibodies ((109) and reviewed in (110)). While QS-21 did not bind to either TLR2 or TLR4 (111), both lysosomal destabilization and spleen tyrosine kinase (Syk) activation were shown to be essential for both QS-21-mediated activation of monocyte-derived DCs (108), antigen cleavage, and subsequent presentation to CD8⁺ T cells (81)

However, there are several drawbacks that limit the application of QS-21, including (I) strong side effects when using high-dosages (e.g. local erythema, induration, and flu-like symptoms); (II) its chemical instability, especially in warm temperatures and pH values higher than 7.4; and (III) limited supplies due to difficulties extracting highly pure QS-21 from its natural source (107).

Therefore, in current clinical usage, QS-21 needs to be co-applied with other adjuvants and is most commonly applied packaged in liposomes in combination with MPLA (a TLR4-ligand, see above) as AS01. A study from Welsby et al. compared the effects of either QS-21 or MPLA on human DCs in culture, which were both packaged independently of each other in liposomes (108). Their results showed, that both QS-21- and MPLA-liposomes could induce higher IL-6, TNF-α, and IL-8 secretion, as well as CD86 expression (108). Interestingly these two liposome-packaged adjuvants showed different kinetics on cytokine production: MPLA-liposomes induced fast DC activation (two to four hours post-stimulation) while for QS-21-liposomes DC activation was observed at later time points (six hours post-stimulation) (108). Using microarray analysis the authors found, that QS-21- and MPLA-liposomes activated similar pathways such as cytokine-, NLR, and G-protein-couple receptor-signaling in human DCs (108). In contrast, the activation of the activator protein 1 (AP-1) and activating transcription factor 2 (ATF2)-transcription factor networks were found to be specific for QS-21-liposomes, while MPLA-liposomes specifically induced TLR- and IL-1-signaling in human DCs (108). The observed DC-activation by QS-21-liposomes was shown to depend on cholesterol-mediated endocytosis followed by lysosomal destabilization (108), which was different from MPLA since it is known to activate TLR4-signaling (60).

Although QS-21 is a strong adjuvant, it is well tolerated in clinical trials with systemic side effects being infrequent: Compared to placebo, QS-21 increased injection site pain and occurrence of diarrhea in two Pfizer-sponsored Alzheimer studies (NCT00479557 and NCT00498602) (meta-analysis in (112)). QS-21-mediated injection site pain was also reported in three phase I trials by Waite and colleagues (113) and upon intra-muscular application of QS-21 mixed with HIV-1 glycoprotein 120 (114). Also Gilewski reported local skin reactions at the injection site (duration 4 to 5 days) and mild flu-like symptoms (duration 1 to 2 days) upon application of a QS-21-adjuvanted vaccine to prevent breast cancer (115).

Although there is currently no study testing QS-21 as an adjuvant for allergy treatment, QS-21 was shown to induce Th1-biased immune responses with high titers of IgG_2a and IgG_2b antibodies (81), making it an attractive candidate for future research.

Glucans are polysaccharides derived from either plants or microorganisms (such as bacteria, algae, or fungi) that are composed of D-glucose units linked by different glycosidic bonds (Figure 4) (116). Glucans encompass several different α-glucans (e.g., dextran, glycogen, pullulan, and starch) and β-glucans (e.g., cellulose, curdlan, laminarin, chrysolaminarin, lentinan, lichenin, pleuran, and zymosan) (116). The different glucan families considerably vary in structure, type of linkage, and length of the macromolecules. The mechanisms by which glucans activate immune cells are not fully understood. However, some glucans were reported to bind to either carbohydrate receptors or TLRs expressed on the surface of APCs, suggesting these molecules to have adjuvant potential (116). For example, Dillon et al. demonstrated that yeast zymosan could bind to both the Dectin-1 receptor and TLR2 on human and murine DCs, which induced higher secretion of IL-12(p70), IL-6, and IL-10 in vitro (117). Moreover, two studies used β-glucans chemically conjugated with either the detoxified diphtheria toxin CRM197 or the cancer-associated mucus protein mucin 1 as novel vaccines against either diphtheria or cancer (118, 119). Their results showed that, as adjuvants, β-glucans can serve as both carriers and immune activators promoting immune responses in vitro and in vivo, which enhanced the delivery of antigens by binding to APCs and enforced antigen-specific antibody production (118, 119).

The potential of glucans for allergen-specific immunotherapy is still controversial. Some researchers demonstrated β-(1,3)-glucans, which can be found in either HDM feces or pollen grains derived from several plant species, to exacerbate allergic inflammation by activating DCs in vitro and enhancing allergen-specific IgE antibody as well as Th2 cytokine production in vivo (120, 121). However, in another in vivo OVA-based food allergy mouse model, the authors applied either the low-molecular-weight β-(1,3)- or a 50–80% branched β-(1,6)-glucan (purified from Aureobasidium pullulans). Here, these two β-glucans were shown to suppress OVA-specific IgE production (122). In a non-industry-sponsored, multicenter open split-body study enrolling 105 patients with atopic dermatitis, regular topical application of a 0.25% β-glucan-based cream (Imunoglukan P4H^® cream) over a period of 24 weeks reduced pruritus and severity of atopic dermatitis while increasing the number of days without exacerbation (123). The β-glucan-containing cream was well tolerated with only mild local side effects (contact dermatitis and other undisclosed skin side effects) that disappeared during regular application (123). In a randomized, double-blind, parallel group, placebo-controlled study without industry sponsorship 26 patients with seasonal AR sensitized to olive pollen were treated orally with 10 mg β-(1,3–1,6)-glucan twice a day for 12 weeks (124). β-glucan treatment significantly reduced levels of Th2 cytokines and eosinophils in nasal lavage fluid while increasing levels of IL-12 (124). Recently, a β-(1,3)-glucan and a β-(1,3-1,6)-glucan both derived from Saccharomyces cerevisiae were investigated in an OVA-based mouse intestinal allergy model: Here, oral treatment with β-(1,3)-glucan (but not β-(1,3-1,6)-glucan) was shown to further promote intestinal inflammation and allergic responses (125), suggesting that this β-(1,3)-glucan might not be suitable for immunotherapy purposes. Gut microbiome analysis in the allergen-sensitized and β-(1,3)-glucan fed group showed an increased ratio of Firmicutes to Bacteroidetes, a potential indication of dysbiosis (125). In a phase IV clinical study (sponsored by the Meiji University of Oriental Medicine, Kyoto, Japan) oral-administration of superfine dispersed β-(1,3)-glucan (particle-size: 0.08 μm), instead of non-dispersed shiitake extract (particle-size: 288 μm), was shown to alleviate allergic symptoms like rhinorrhea, sneezing, nasal congestion, and itchy watery eyes otherwise induced by Japanese cedar pollen in allergic patients (21). Along with the reduction of the aforementioned clinical symptoms, the superfine dispersed β-(1,3)-glucan treatment group also displayed lower pollen-specific IgE antibody levels (21). Based on these findings the authors speculated, that differences in size, structure, and purification method of carbohydrates derived from natural sources might lead to distinct immune responses (21, 122). Therefore, more detailed studies are needed to understand the adjuvant potential and potential side effects of glucans.

Mannan is a β-(1,4)-mannose polysaccharide which can be found in the cell walls of both plant and fungal cells (81). Mannan can modulate immune responses towards both Th1 and Th2 responses (depending on its oxidation status, reduced mannan: Th2, oxidized mannan: Th1) by binding to both mannan- and C-type lectin receptors (DC-SIGN) expressed e.g., by APCs (116), which leads to activation of complement pathways (116), the NLRP3 inflammasome, as well as increased phagocytosis and cytokine secretion in APCs (81, 116). The first study using mannan as a vaccine adjuvant in mice was reported in 1992. Okawa et al. conjugated mannan with the HBV 139-147 peptide (126). The mannan:HBV peptide conjugate was shown to induce higher IgG titers in mice than corresponding dextran:HBV peptide-conjugates (126). So far, either native, oxidized, or reduced forms of mannan have been conjugated to different antigens to be tested as adjuvants for targeting APCs (reviewed in (81)). The results demonstrated both higher antigen uptake and presentation by human monocyte-derived DCs (127) coupled with the induction of distinct antigen-specific Th1- or Th2-responses in mice (128, 129).

There are some studies investigating mannan as an adjuvant for improving allergen immunotherapy. Weinberger and colleagues used oxidized mannan derived from the yeast Saccharomyces cerevisiae to generate reactive aldehyde groups for covalent attachment of OVA via amine-containing amino acid residues (MN–Ova) (130). Compared to OVA alone, MN–Ova was taken up more strongly by mouse DCs both in vitro and in vivo, while also inducing higher IgG antibody production in mice at the injection site (130).

In a second set of studies, the native form of Saccharomyces cerevisae mannan was conjugated to defatted grass pollen grain-based allergoids from Phleum pratense (PM) (127, 131, 132). PM reduced skin prick test reactivity compared to either grass pollen allergens or grass pollen-based allergoids in a phase II trial (sponsored by Inmunotek SL, Madrid Spain)(Figure 5) (127). Immunologically, in vitro stimulation of human monocyte-derived DCs (hmoDCs) with PM induced higher secretion of IL-6 and IL-10 paralleled by a lower production of the Th2 cytokine IL-4. PM additionally promoted a mTOR-dependent increase in hmoDC glycolysis, and was more efficiently taken up compared to either pollen extracts or pollen allergoids (Figure 5) (127, 131, 132). In vivo, PM promoted tolerogenic immune responses characterized by the differentiation of CD4⁺CD25^highFOXP3⁺ T_reg cells and increased pollen-specific IgG_2a/IgE ratios in mice (Figure 5) (127). These immune-modulating properties were strongly dependent on the intact structure of the mannan, as conjugates containing oxidized mannan showed a strongly reduced uptake, IL-10 secretion, programmed death-ligand 1 (PD-L1) expression, and frequency of induced T_regs (127).

In a follow-up pre-clinical study, Benito-Villalvilla et al. could show non-oxidized mannan conjugated to allergoids to induce the differentiation of tolerogenic DCs from monocytes (133). In the presence of IL-4, GM-CSF, and the mannan:allergoid conjugates, monocytes from both non-atopic and allergic donors differentiated into tolerogenic DCs with an increased mitochondrial oxidative phosphorylation (OXPHOS) and epigenetic reprogramming characterized by an open chromatin structure with both increased histone 3 lysine 4 (H3K4) trimethylation at the promotors of IL-10, PD-L1, suppressor of cytokine signaling 3 (SOCS3), and increased histone 3 lysine 27 (H3K27) acetylation at the promotor regions of PD-L1 and SOCS1 (133). Furthermore, monocyte-derived tolerogenic DCs expressed higher amounts of the microRNAs 146 a and b while having reduced levels of miR-155 and histone deacetylases (133). These tolerogenic DCs secreted IL-6 and IL-10 and promoted the differentiation of IL-10-producing CD4⁺CD25⁺FOXP3⁺ T_regs (133). These results suggest mannan, besides its direct immune activating effects, to also trigger extensive epigenetic modifications in the stimulated DCs that may (if also happening in their bone marrow progenitors) potentially result in long-term modification of the respective cells reactivity towards either mannan or the conjugated antigens (so called “trained immune responses”).

In a first multicenter, prospective, randomized, double-blind, double-dummy, placebo-controlled, phase II study encompassing 196 patients with HDM allergic rhinitis with or without asthma (financed by Inmunotek SL, Madrid, Spain), patients and healthy controls were treated with a HDM extract consisting of a mixture of 50% Der p and 50% Der f allergoids conjugated to non-oxidized mannan either sublingually or subcutaneously (22). Here, both s.c. and s.l. application of the mannan:allergoid conjugates increased the percentage of patients showing an improved nasal provocation test up to 50% and reduced combined symptom medication scores by 45 to 70%, depending on the dose and application route (22). Interestingly, s.l. application of the mannan conjugates induced lower levels of HDM-specific IgG₄ antibodies than the s.c. application route (22). Reported side effects were grade I and II reactions (mostly delayed) with all grade II reactions occurring in the 1000 mTU/ml s.c. group (but not in groups receiving higher concentrations) (22). No grade III or IV systemic reaction were reported upon application of the mannan:allergoid conjugates (22). The majority of local reactions (observed in the s.c. treatment group) were mild and occurred after the first injections. However, six delayed reactions were described as severe and led to withdrawal of the respective patients from the trial (22).

Currently, three more phase II clinical trials (EudraCT nos. 2014-005471-88, 2018-002522-23, and 2020-004126-32) as well as two phase III clinical trials for allergoid-mannan conjugates for the treatment of patients with mild to moderate HDM-induced asthma and rhinitis/rhinoconjunctivitis (NCT05400811) or birch pollen-induced allergic rhinitis/rhinoconjunctivitis (EudraCT: 2021-002252-36) are either ongoing or finished. All studies are financed by Inmunotek SL (Madrid, Spain).

In summary, these studies demonstrated the potential of mannan as an adjuvant to improve AIT, confirming its efficacy in a first clinical study with more phase II and III studies currently ongoing.

The TLR5- and NLR family caspase activation and recruitment domain (CARD) containing 4 protein (NLRC4)-ligand flagellin is a bacterial motility protein forming the main body of the bacterial flagellum (135). Vaccines adjuvanted with flagellin were reported to be both safe and well-tolerated in clinical trials (136, 137) where flagellin was demonstrated to be an effective mucosal adjuvant triggering Th1-biased, protective immune responses in mice and monkeys (138–140).

Flagellin has also been analyzed pre-clinically as an adjuvant for improving traditional AIT. The available literature is divided into investigating (I) the mixture of flagellin and different allergens or (II) flagellin fused to allergens as part of fusion proteins. In this context, flagellin-containing fusion proteins were suggested as potential therapeutics combining the adjuvant activity of the TLR5- and NLRC4-ligand and the cargo antigen into a single molecule, allowing for the efficient targeting of antigens to, and simultaneous activation of TLR5⁺ APCs (140–143).

Lee et al. tested either intranasal or intralymphatic injection of Vibrio vulnificus flagellin B (FlaB) mixed with OVA in an OVA-based mouse model of airway inflammation (144, 145) where the OVA/FlaB mixture reduced OVA-induced airway hyper-responsiveness, inflammatory cell infiltration in lung tissue, as well as both systemic (IL-4, IL-5, IL-6, IL-17, and IFN-γ) and local (IL-4 and IL-5) cytokine production; while increasing OVA-specific IgA levels in serum (144, 145). Mechanistically, the ability of the OVA/FlaB mixture to suppress allergic responses was found to be dependent on TLR5 (146). Moreover, FlaB promoted the differentiation of IL-10- and TGF-β-producing tolerogenic DCs (tolDCs), which suppressed both Th1/Th2 responses and enhanced the activation of T_reg cells (146). Besides DCs and T cells, a recent study demonstrated flagellin to stabilize OVA-induced eosinophil activation (releasing major basic protein (MBP) and peroxidase (EPX) in vitro and in vivo) and to reduce oxidative stress in OVA-sensitized eosinophils which was correlated with decreased allergic inflammation (147). Additionally, Zeng et al. found flagellin to diminish allergen-induced oxidative stress in B_regs from both a food allergy (FA) mouse model and patients’ blood samples (148). Here, adding flagellin to OVA improved traditional AIT in an OVA-induced food allergy model in vivo (148). These results showed that flagellin can activate different cell types, changing both their cytokine secretion and metabolic status, which correlated with improved treatment outcomes in AIT in mice.

On the other hand, flagellin:allergen fusion proteins have demonstrated an even higher potential to improve allergy treatment than the mixture of flagellin and allergen. Tan et al. generated a recombinant fusion protein consisting of FlaB fused to the C-terminus of the HDM allergen Der p 2 (rDerp2:FlaB) (149). Compared to the non-fused Der p 2 + FlaB mixture, rDerp2:FlaB more efficiently suppressed airway hyper-responsiveness, serum IgE levels, and secretion of Th2 cytokines into bronchoalveolar lavage fluid in a HDM-induced mouse asthma model (149).

Moreover, Kitzmüller and colleagues also demonstrated that Salmonella FliC (genetically modified to lack the middle hypervariable domain of flagellin described to induce neutralizing anti-flagellin antibody responses) fused to either the N- or C-terminus of Bet v 1 (rFliC:Betv1 or rBetv1:FliC, respectively) could activate monocyte-derived DCs from pollen-allergic patients (150). Immunization of mice with either rFliC:Betv1 or rBetv1:FliC fusion proteins induced lower serum levels of Bet v 1-specific IgE paralleled by a higher production of IgG antibodies (150), suggesting flagellin:allergen fusion proteins to be promising novel treatment options for pollen allergy.

In own previous studies we generated fusion proteins combining flagellin A (FlaA) from Listeria monocytogenes with different allergens (141–143, 151). We demonstrated that prophylactic vaccination with rFlaA:OVA, but not the mixture of both proteins, induced higher production of allergen-specific IgG_2a antibodies paralleled with decreased levels of both IgE and Th2 cytokines (IL-5 and IL-13), resulting in the prevention of clinical symptoms such as softness of faces, body weight loss, and drop in core body temperature in an OVA-induced intestinal allergy mouse model (142, 152).

More in-depth analyses showed, that all generated rFlaA:allergen fusion proteins stimulated mDCs to secrete higher amounts of both pro- (IL-6, IL-1β, TNF-α, and IL-12) and anti-inflammatory (IL-10) cytokines (Figure 6) (141, 143, 151–153). rFlaA:allergen fusion protein-treated mDCs also displayed an activated phenotype with higher surface expression levels of the maturation markers CD40, CD69, CD80, and CD86 (Figure 6) (141, 151). Mechanistically, the induced mDC activation was mediated by intracellular MyD88-, MAPK-, and nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB)-pathways while being largely TLR5-independent (143, 152, 153). The rFlaA:allergen fusion protein-stimulated mDCs efficiently modulated allergen-specific T cell responses, suppressing allergen-induced Th2 cytokine (IL-4, IL-5, and IL-13) production from ex vivo-isolated, Th2-biased CD4⁺ T cells in an IL-10-dependent manner (Figure 6) (142, 143, 151). The flagellin:antigen fusion proteins formed high-molecular aggregates possibly due to self-assembly of flagellin molecules into flagella-like structures, enabling the formation of intermolecular cysteine bridges between the fused allergens (FlaA itself does not contain cysteine residues) (143, 152). These aggregates were both taken up more strongly by mDCs and inhibition of aggregate uptake dose-dependently suppressed fusion protein-induced cytokine secretion (143). Therefore, these results suggest the strong activation of mDCs by the flagellin:allergen fusion proteins, while being mostly TLR5-independent, to be at least in part caused by aggregation-mediated enhanced uptake and subsequent triggering of above-described intracellular signaling cascades.

Interestingly, the tolerogenic phenotype of mDCs stimulated with a flagellin fusion protein incorporating the major birch pollen allergen Bet v 1 (rFlaA:Betv1) was shown to depend on a JNK-MAPK-dependent activation of the mTOR pathway (143, 153). mTOR is a conserved serine/threonine protein kinase that belongs to the phosphatidylinositol 3-kinase-(PI3K) family. mTOR not only integrates various nutritional and environmental stimuli, including levels of growth factors, cellular energy reserves, and stress levels, but is also a master regulator of cellular metabolism that affects innate and adaptive immune responses (154). mTOR activation in rFlaA:Betv1-stimulated mDCs resulted in alterations of mDC metabolism with increased rates of glycolysis, Warburg metabolism, and reduced mitochondrial respiration (155). Inhibition of either glycolysis or mTOR activation suppressed rFlaA:Betv1-induced IL-10 secretion (143, 153, 155). In addition to glycolysis, fatty acid synthesis also significantly contributed to rFlaA:Betv1-mediated cytokine secretion, the production of antimicrobial molecules, and the modulation of T cell responses (155).

In line with the results presented for mDCs, stimulation of murine bone marrow-derived macrophages with rFlaA:Betv1 also triggered a MyD88-dependent, but only partly TLR5-dependent, cytokine production, stronger activation of hypoxia inducible factor 1 alpha (HIF-1a)-, MAPK-, NFκB-, and mTOR-signaling, and increased glucose metabolism (156). Therefore, macrophages also contribute to the strong immune-modulating effects of rFlaA:Betv1 previously observed in vivo and should be considered important target cells contributing to the re-establishment of allergen tolerance in AIT.

Moreover, the murine lung epithelial cell line LA-4 was shown to be activated by rFlaA:Betv1. Here, the fusion protein was shown to be taken up more strongly, triggering a MAPK-, NFκB-, and cyclo oxygenase 2 (COX2)-dependent activation of epithelial cells, characterized by a pronounced secretion of the cytokine IL-6, the myeloid chemo-attractants CCL2 and CCL20, as well as production of prostaglandin E2 (Figure 6) (157). rFlaA:Betv1-stimulated LA-4 cells modulated the activation of mDCs in a COX-2-dependent manner, resulting in lower secretion of the pro-inflammatory cytokines IL-12 and TNF-α, while the levels of the IL-1β and IL-10 remained unchanged (157). Therefore, epithelial cells also contribute to the immune modulating properties of potential allergen therapeutics and should be kept in mind as potential target cells for AIT approaches.

Finally, rFlaA:Betv1 was shown to induce a MyD88- and mTOR-dependent, but thymus-independent activation of ex vivo-isolated B cells, inducing a regulatory B cell phenotype (CD19⁺CD1d⁺CD24⁺CD38⁺IgM⁺) characterized by: (I) the secretion of the anti-inflammatory cytokine IL-10, (II) the production of antigen-binding IgG(1&3) and non-antigen-binding IgA antibodies, and (III) the capacity to suppress Bet v 1-induced Th2-responses (Figure 6) (158). To our knowledge, this is the first report showing the induction of B_regs by a potential therapeutic candidate for the treatment of allergic diseases.

So far Flagellin:antigen fusion proteins have not been investigated clinically for allergy treatment, but they were tested in several phase II clinical trials for the prevention of influenza infection (136, 137, 159–161). In these studies, the vaccines were well tolerated with lower doses (up to 3 µg) resulting in mild side effects (injection site pain, headache, fatigue, and myalgia) paralleled by increased serum CRP levels (136, 137, 159–161). Higher doses (10 µg) increased severity of fatigue, headache, myalgia, joint pain, fever, and gastrointestinal symptoms (nausea, vomiting, diarrhea) (136, 160). Symptoms resolved within 24 h of vaccination (136).

In summary, both flagellin and especially flagellin-containing fusion proteins incorporating different allergens were shown to efficiently activate many types of immune cells and thereby modulate immune responses, making these allergen-adjuvant conjugates interesting candidates for further development in AIT treatment.

Bacterial DNA contains a high frequency of unmethylated CpG motifs that can bind to TLR9 and trigger immune responses (162). Therefore, synthetic immunostimulatory oligodeoxynucleotides containing unmethylated CpG motifs (CpG ODN) have been investigated to either mimic natural bacterial infection or as vaccine adjuvants (162). So far, several different types of CpG ODNs have been tested in clinical phase I/II trials as vaccine adjuvants against either infectious diseases (hepatitis B, malaria, and pneumonia) or cancer (melanoma, breast cancer, sarcoma, ovarian cancer, glioblastoma, and lymphoma) (162). The overall results showed that CpG ODN adjuvants only induced short-term mild-to-moderate adverse effects, while inducing Th1-biased immune responses and promoting CD8⁺ T cell activation (162).

The proposed advantages of using CpG ODN for allergen-specific immunotherapy include: (I) CpG ODN can activate DCs and macrophages via TLR9, promoting especially pDCs, to increase allergen uptake and produce IL-10, TGF-β, and indoleamine 2,3-dioxygenase (IDO) (Figure 7). This in turn drives differentiation and activation of T_reg responses while also inducing DC-derived secretion of IFN, IL-6, IL-12, and IL-18 that promote the differentiation of IFN-γ producing Th1 cells and isotype-switch towards IgG_2a (Figure 7) (162). (II) Since B cells express high levels of TLR9, CpG ODN can also directly activate B cells (Figure 7). An in vivo study that used ragweed pollen-sensitized mice demonstrated that combining CpG ODN with ragweed pollen could induce higher frequencies of IL-10-producing B_reg (B220⁺CD19⁺CD23⁺IgM⁺CD40⁺MHCII^hi), which contribute to the suppression of allergen-induced inflammatory responses (163). However, CpG ODN application was also reported to result in increased serum levels of TNF-α that were associated with toxic shock in mice (Figure 7) (164, 165).

In a ragweed allergen-induced mouse asthma model, CpG ODN was shown to induce IFN-γ-dependent Th1-responses while suppressing allergic lung inflammation (166). As an approach to treat ragweed allergy, the immune-stimulating sequence 1018 (ISS-1018, 5’-TGACTGTGAACGTTCGAGATGA-3’) was chemically conjugated to the major ragweed allergen Amb a 1, resulting in an Amb a 1:CpG conjugate with a mean conjugation rate of 1 to 4 (167). This conjugate was termed TOLAMBA. In BALB/c mice, TOLAMBA induced Th1-biased immune responses characterized by the production of IFN-γ, IgG₁- (also observed in rabbits and cynomolgus monkeys) and IgG_2a-antibodies, while the non-conjugated mixture of Amb a 1 and ISS 1018 predominantly induced Th2-biased immune responses (167). In Amb a 1-sensitized and -challenged BALB/c mice, TOLAMBA reduced the production of IgG₁ and IgE antibodies while boosting production of IgG_2a (167). Similar results were described for Amb a 1-ISS-1080 by both Marshall et al. and Santeliz et al. (168, 169) as well as OVA-CpG ODN (1826) by Shirota et al. (170).

In 2006 Creticos and coworkers reported the results of a randomized, double-blind, placebo-controlled phase II trial testing TOLAMBA in 25 ragweed allergic patients (sponsored by the Johns Hopkins University School of Medicine, Baltimore, USA under participation of Dynavax Technologies, Berkeley, California, USA) (171). Patients were treated with six weekly injections of increasing doses of TOLAMBA (0.06, 0.3, 1.2, 3, 6, and 12 µg) and monitored during the following two ragweed seasons (171). Compared to placebo, TOLAMBA reduced rhinitis visual analogue scores by two thirds in both ragweed seasons, and also Amb a 1-specific IgE levels were lower during the first ragweed season (171). While the authors reported no vaccine-associated systemic adverse events or clinically significant abnormalities, also no effect on the primary endpoint (nasal vascular permeability as assessed by nasal-lavage albumin) was achieved (171).

To follow up on these results, a phase IIb study with TOLAMBA in 738 ragweed allergic patients was conducted (sponsored by Dynavax Technologies). It was initially reported in a conference abstract, that treatment during the first ragweed season was well tolerated in all groups without TOLAMBA-related serious adverse events (172). However, later on, the company reported problems with the study as they observed no measurable disease in any of the study groups during the ragweed season (http://investors.dynavax.com/releasedetail.cfm?releaseid=231013). In May 2008 the company reported, that although TOLAMBA showed a trend towards a reduction of the symptoms of ragweed allergic individuals relative to placebo, no statistical significance was achieved (https://investors.dynavax.com/news-releases/news-release-details/dynavax-tolambatm-chamber-study-misses-primary-endpoint-company). With no original paper published on the study so far, the company referred to “an unexpectedly high degree of variability in the data set possibly due to the subjective nature of symptom scoring used to assess efficacy” and decided to discontinue clinical development of TOLAMBA (https://investors.dynavax.com/news-releases/news-release-details/dynavax-tolambatm-chamber-study-misses-primary-endpoint-company).

As an approach to treat food allergy, Rodriguez et al. treated BALB/c mice sensitized to the major peach allergen Pru p 3 with a mixture of Pru p 3 peptides and CpG ODNs (173). In their experimental model, treatment with Pru p 3 peptides plus CpG ODNs suppressed the temperature drop induced by i.p. challenge with the allergen and strongly reduced Pru p 3-specific IgE levels (173). This CpG ODN-mediated immune modulation was accompanied by an increased production of Pru p 3-specific IgG_2a antibodies as well as an increase in IL-10-producing CD4⁺CD25⁺ T_regs and IFN-γ-producing Th1 cells (173). Interestingly, the tolerance was maintained even after stopping the treatment for three weeks (173).

In a recent review article Montamat and colleagues recapitulated the development of CpG ODN adjuvants for allergy treatment in more detail (174). When analyzing the complex immune-stimulating properties of CpG ODN adjuvants, they concluded that the detrimental inflammation and Th1 responses observed upon application of CpG ODN adjuvants may be either preferentially triggered by low doses of CpG ODNs or be the byproduct of LPS potentially contaminating the used CPG ODN preparations (174). They concluded, that application of higher doses of CpG ODN may reliably induce the desired, pro-tolerogenic T_reg responses (174). They proposed, that “doses between 0.5 and 1.5 mg/kg, dependent on the administration route, should induce the desired immune tolerance toward the allergen with a minimal risk of adverse reactions” (174).

In theory, the immune-activating properties of CpG ODNs are of interest in allergy treatment and initial mouse studies showed promising immune-modulating effects [induction of Th1 responses paralleled by suppression of Th2 responses (167)]. However, currently there are no clinical trials demonstrating a clear benefit of CpG ODNs in allergic patients. As neither an optimal dose nor application route of CpG ODNs are established, further studies improving on the published protocols should investigate the potential of CpG ODNs as adjuvants for advanced AIT treatment.