Although the efficacy of SIT has been well established, there are still concerns on its safety and on the particularly long duration of the treatment [4, 5]. Despite improved conventional protocols of SIT, severe anaphylactic reactions may sometimes occur even in well-trained centers. The duration of the therapy is another factor of discouragement for patients. The possibility to shorten the treatment period, and to decrease the number of injections, together with improved safety, would provide a considerably more satisfying strategy for allergists to treat patients.
There have been so far several attempts to decrease the allergenicity of immunotherapy products, including so-called allergoid preparations [36–38], and more recently T cell peptides, allergen fragments or mutated allergens. T cell epitopes are short enough to stay unrecognized by IgE in most cases, but are well recognized by T cell receptors. They appear to induce tolerance in volunteers injected intradermally with multi-T cell epitope preparations . They did not induce immediate-type reactions but late respiratory symptoms including dyspnea, which occurred a few hours after the injections [39, 40]. Long synthetic peptides overlapping by 5 to 20 amino acids (COPs), as applied in the current study, were initially developed from PLA2, the major bee venom allergen [17, 18]. PLA2 hypersensitive volunteers did not react to intradermal tests with PLA2 derived long synthetic peptides. A phase I clinical trial in bee venom hypersensitive patients was safe and induced only transient late allergic reactions (sensation of heat in two volunteers) and was immunogenic, namely inducing a significant rise in PLA2 specific IgG4 . In the present study, the same approach was applied to Bet v 1, the dominant allergen from birch pollen, a very common respiratory allergy in Northern Europe. Three sets of peptides were obtained, two consisting in three overlapping peptides T1-T2-T3 and T6-T7-T8, another in two overlapping peptides T4-T5. Their sequences were derived from the most representative Bet v 1 isoform, rBet v 1.01A. The interest of COPs relies in part in their capacity to cover all potential T cell epitopes known to be spread all along the Bet v 1 molecule , as previously observed also with phospholipase A2 . In contrast to other hypoallergenic fragments [15, 41], overlapping of COPs prevents the loss of T cell epitopes possibly located at the site of cleavage of the fragments, thus avoiding the risk of any HLA restriction . Furthermore, the purity of the product, compared to pollen extracts, insures the absence of contaminating allergens , endotoxins  and potentially reduces the risk to induce IgE sensitization to uncontrolled components .
Peptide IgE binding capacity was first evaluated in competition ELISA with human sera. The endpoint of the assay was at least a ten-fold lower IgE binding activity of peptides as compared to recombinant Bet v 1, an objective largely fulfilled here. Indeed, in comparison to rBet v 1, we can estimate that AllerT showed at least a 107-fold lower ability to bind IgE. This value is below values previously reported for hypoallergenic Bet v 1 variants, including folded variant , fragments  or trimers .
We examined in parallel the capacity of AllerT to induce anaphylactic reaction in a murine model. BALB/c mice sensitized to rBet v 1 were not affected by the intraperitoneal injection of AllerT (at a five-fold higher concentration than rBet v 1), whereas the i.p. injection of rBet v 1 induced a sharp drop in rectal temperature, a surrogate marker of anaphylaxis. These in vivo data largely supported the results obtained with the rat basophil leukemia cell line in vitro.
Although we cannot fully insure that anaphylactic reaction was dependent on the induction of IgE, anti-rBet v 1 specific murine IgE reached high titers after the fourth subcutaneous injection as measured by ELISA (Figure 3A). In parallel we also observed a strong increase in rBet v 1 specific IgG1 and IgG2a. Massive IgG1 activation has been shown to induce anaphylactic shock in mice in the absence of IgE . In contrast IgG2a, a possible murine equivalent of IgG4 , may antagonize allergen-IgE binding and protect against anaphylaxis. Despite the presence of both IgG1 and IgE against rBet v 1 in sensitized mice, our combination of COPs, AllerT, was unable to induce anaphylaxis in contrast to the recombinant allergen.
The low, if any, IgE binding activity of COPs as compared to rBet v 1 was confirmed by a functional human basophil activation assay as well as in an assay based on huFcϵRI transfected RBL. Using individual peptides as well as their mixes, only background basophil or RBL activation was noted in presence of sera from birch pollen hypersensitive patients. Finally in in vivo assays, twenty volunteers were tested by skin prick tests for reactivity to Bet v 1, to birch pollen extract as well as to the five COPs and their mixes (AllerT, T4-T5 mix). As shown in Table 3, a decrease in skin sensitivity of about 10 to 100 folds was observed. This was true for T1 to T5 COPs individually. Interestingly, in 6 patients, the mix of T4-T5 COPs was associated with a skin response of 1 to 3 mm in diameter, in one patient for COP T5 only and in two patients for COP T1. Together with residual IgE binding activity in ELISA competition assays, these results suggest a partial refolding of T4 and T5, which potentially reconstitute the original IgE epitopes in solution. This has already been observed for other proteins [50, 51]. This example shows the importance to select for peptides with no detectable IgE binding either in competition ELISA or degranulation tests before going to human trial since possible re-association may occur after in vivo injection as exemplified by the skin prick tests reported in this study.