To cite this article: Blank S, Seismann H, Michel Y, McIntyre M, Cifuentes L, Braren I, Grunwald T, Darsow U, Ring J, Bredehorst R, Ollert M, Spillner E. Api m 10, a genuine A. mellifera venom allergen, is clinically relevant but underrepresented in therapeutic extracts. Allergy 2011; 66: 1322–1329.
Background: Generalized systemic reactions to stinging hymenoptera venom constitute a potentially fatal condition in venom-allergic individuals. Hence, the identification and characterization of all allergens is imperative for improvement of diagnosis and design of effective immunotherapeutic approaches. Our aim was the immunochemical characterization of the carbohydrate-rich protein Api m 10, an Apis mellifera venom component and putative allergen, with focus on the relevance of glycosylation. Furthermore, the presence of Api m 10 in honeybee venom (HBV) and licensed venom immunotherapy preparations was addressed.
Methods: Api m 10 was produced as soluble, aglycosylated protein in Escherichia coli and as differentially glycosylated protein providing a varying degree of fucosylation in insect cells. IgE reactivity and basophil activation of allergic patients were analyzed. For detection of Api m 10 in different venom preparations, a monoclonal human IgE antibody was generated.
Results: Both, the aglycosylated and the glycosylated variant of Api m 10 devoid of cross-reactive carbohydrate determinants (CCD), exhibited IgE reactivity with approximately 50% of HBV-sensitized patients. A corresponding reactivity could be documented for the activation of basophils. Although the detection of the native protein in crude HBV suggested content comparable to other relevant allergens, three therapeutical HBV extracts lacked detectable amounts of this component.
Conclusion: Api m 10 is a genuine allergen of A. mellifera venom with IgE sensitizing potential in a significant fraction of allergic patients independent of CCD reactivity. Thus, Api m 10 could become a key element for component-resolved diagnostic tests and improved immunotherapeutic approaches in hymenoptera venom allergy.
chitin binding domain
cross-reactive carbohydrate determinant
yellow jacket venom
Although venom immunotherapy (VIT) is an effective treatment in the majority of patients, 10–20% of patients were not protected by honeybee venom (HBV) immunotherapy (1, 2). Thus, there is considerable interest in improving diagnosis as well as design, safety, and efficacy of therapy.
The most prominent HBV allergens include phospholipase A2 (Api m 1), hyaluronidase (Api m 2), and the basic peptide melittin (Api m 4) (3), all constituting medium to high abundance proteins of HBV (4). Nevertheless, hymenoptera venoms comprise a more complex cocktail of different compounds all of which may contribute to sensitization, symptoms, and success of VIT. Significant progress, however, has been made in the recent years to identify additional compounds of lower abundance, such as the acid phosphatase Api m 3 (5) and the DPPIV enzyme Api m 5 (6).
Recombinant approaches are imperative for the assessment of allergenicity and clinical relevance of such venom compounds, whereby expression should meet the requirements of proper folding and correct posttranslational modifications being potentially important for the establishment of conformational epitopes (7). In particular, glycan structures can contribute significantly to characteristics of venom proteins and constitute the underlying principle of cross-reactive carbohydrate determinants (CCD), a peculiarity of hymenoptera and plant allergens interfering with diagnosis and design of therapeutic strategies (8–10).
A venom protein of considerable interest within this context is Api m 10, the carbohydrate-rich protein. Peptides of this protein were identified by two independent groups in 2005 (11, 12). Insoluble, nonglycosylated protein obtained from Escherichia coli exhibited IgE reactivity (13) and an inherent molecular lability. However, without established recombinant expression of soluble Api m 10, its relative abundance in whole HBV as well as its relevance in the context of sensitization and VIT remained elusive.
In this study, we comparatively analyzed differentially glycosylated Api m 10 proteins for their IgE reactivity and basophil activation. The obtained data suggest a pivotal role of this protein as sensitizing agent in HBV allergy, thus supporting its status as a major allergen of clinical relevance in HBV. Furthermore, a monoclonal human IgE antibody allowed quantification of Api m 10 in Apis mellifera venom and also demonstrated the absence of this putatively essential component in widely used therapeutic preparations.
Materials and methods
Three groups of sera from hymenoptera venom-sensitized patients were selected: (i) sera with negative specific IgE (sIgE) test to vespid venom (i3 < 0.35 kU/l) but positive test to HBV (i1 > 0.35 kU/l) (n = 17); (ii) sera with negative sIgE test to HBV (i1 < 0.35 kU/l) but positive test to vespid venom (i3 > 0.35 kU/l) (n = 16); and (iii) sera with positive sIgE test to both (i1 and i3 > 0.35 kU/l) (n = 51). Beekeepers were recruited during daily clinical practice. All patients had given their informed written consent to draw an additional serum sample. The characterization of venom-allergic patients is described in detail in the Data S1.
Cloning of cDNA
Total RNA was isolated from the separated stinger with attached venom sack and glands of honeybee (A. mellifera carnica) using peqGold TriFast™ (Peqlab Biotechnologie, Erlangen, Germany). SuperScript III RT (Invitrogen, Karlsruhe, Germany) and the gene-specific primer 5′-TCAAGCAGTTAATACATCTCCTTGG-3′ were used to synthesize cDNA from the isolated total RNA. Api m 10 cDNA was amplified using Pfu DNA polymerase (Fermentas, St. Leon-Roth, Germany) and the primers 5′-TTCCCTGGTGCACACGATGAGG-3′ and 5′-TCAAGCAGTTAATACATCTCCTTGG-3′. Cloning and expression in insect cells and E. coli is described in detail in the Data S1.
Immunoreactivity of patient sera with recombinant proteins
Specific IgE, IgG, IgG1, and IgG4 immunoreactivity of human sera from hymenoptera venom-allergic patients as well as from beekeepers with purified recombinant proteins (20 μg/ml) was assessed by ELISA. A detailed description of the ELISA is given in the Data S1. The lower end functional cutoff indicated as lines was calculated as the mean of the negative controls plus 2 SDs. Reactivities only slightly higher than the cutoff were excluded.
Statistical analysis was performed with Prism 3.0 software (GraphPad Software, San Digo, CA, USA). Correlation coefficients were calculated using Pearson correlation analyses.
Basophil activation test
The basophil activation test was performed as described previously (6, 14). A brief description is given in the Data S1.
SDS–PAGE and Western blotting are described in detail in the Data S1. Molecular biology standard procedures such as PCR, DNA restriction, ligation, transformation, and plasmid isolation were performed according to established protocols (15). The chimeric human monoclonal IgE antibodies were generated essentially as described recently (16).
Recombinant expression and characterization of Api m 10
For recombinant expression of Api m 10, a putative venom allergen with a theoretical mass of 22 kDa (11, 12), the cDNA of splicing variant 2, was amplified from venom gland cDNA (Fig. S1).
Expression in E. coli as aglycosylated protein (17) yielded soluble Api m 10 with an apparent molecular weight of approximately 35 kDa (Fig. 1A, B), suggesting a modified migration behavior because of its low pI. Glycosylated Api m 10 with or devoid of CCDs was produced by baculovirus infection of Trichoplusia ni (HighFive) or Spodoptera frugiperda (Sf9) insect cells. Purification yielded recombinant Api m 10 with an apparent molecular mass of approximately 50–55 kDa (Fig. 1A, B) underlining the contribution of the carbohydrates to the overall characteristics.
In immunoblot, all three proteins were reactive with a monoclonal anti-Api m 10 IgE and a serum pool of HBV-sensitized patients (Fig. 1C, E). The use of an anti-HRP rabbit serum specific for α-1,3-core fucosyl residues verified pronounced CCD-based cross-reactivity for Api m 10 produced in HighFive cells. In contrast, glycosylated, Sf9-produced as well as E. coli-derived Api m 10 did not exhibit any CCD reactivity (Fig. 1D). Comparable results were obtained with serum of a CCD-reactive but not HBV-allergic patient (Fig. 1F). ELISA analyses corroborated the data obtained in immunoblotting regarding protein identity and presence of CCDs (Fig. 2). These data demonstrate that the host defines the state of glycosylation (18) and, thereby, strongly influences the characteristics of the allergens.
Screening of patient sera for IgE reactivity with Api m 10 variants
Sera of 84 randomly selected patients with a clinical history of insect venom allergy (Table S1) were separated into three groups and assayed by ELISA for sIgE antibodies to Api m 10 produced in Sf9 insect cells.
In group I of 51 sera double positive for HBV and yellow jacket venom (YJV), thus, predominantly cross-reactive, 27 sera (52%) exhibited pronounced reactivity with Sf9-derived Api m 10 (Fig. 3A). From group II of 17 sera with negative sIgE to YJV implying sensitization to HBV only without CCD reactivity, 8 sera (47%) reacted with Api m 10 (Fig. 3B). In group III of 16 sera with negative sIgE to HBV but positive sIgE to YJV, none of the sera recognized Api m 10, suggesting the absence of a cross-reactive molecule in YJV (Fig. 3C). As anticipated, the presence of CCDs upon production of Api m 10 in HighFive insect cells resulted in augmented IgE reactivities in group I (data not shown).
In comparative assessment of sIgE binding to Sf9- and E. coli-derived Api m 10, in group I, 18/38 (47%) and 17/38 (44%), respectively, exhibited pronounced reactivity with the protein variants (Fig. 4A), and in group II 5/11 sera (45%) (Fig. 4B). Group III (Fig. 4C) exhibited no reactivity at all. Reactivities of selected sera found positive in ELISA (patients 16, 19, and 24; patient 69 as control) were further verified in immunoblot (Fig. 4E).
Moreover, sera of beekeepers as well as of venom-allergic patients were investigated for Api m 10- and Api m 1-specific IgG1, IgG4, and IgE antibody responses (Fig. S3). Thereby, over 50% of beekeepers showed relevant IgG1 and IgG4 immune responses to Api m 10, a picture that although less pronounced was also evident for the venom-allergic patients.
Activation of basophils from venom-allergic patients
The capability of Api m 10 produced in HighFive insect cells or E. coli for activation of human basophils was assessed and related to that of established major allergens, Api m 1 (or Ves v 5). All blood samples were obtained from consecutively selected patients (n = 15) (see Table S2) with a clinical history of a severe reaction after a stinging event (13 with honeybee as relevant insect and two additional patients with yellow jacket as relevant insect). All patients had a positive intradermal skin test and sIgE test for HBV and/or YJV.
Overall, 8/13 patients with honeybee as relevant insect showed positive basophil activation with Api m 10. With the species-specific major allergen Api m 1 or Ves v 5, 14/15 patients had positive basophil activation. Figure 5A–F depicts representative results for six exemplary patients. Patients 1, 8, and 10 showed comparable basophil activation (Fig. 5A, C, D) for Api m 10 and Api m 1, patient 6 (Fig. 5B) to a varying degree. Patient 11 exhibited activation by Api m 1 exclusively (Fig. 5E). Patient 15 (Fig. 5F) exhibited activation for Api m 10 and strongly reduced for Api m 1, a picture fully reflected in the sIgE titers. Interestingly, the basophil activation was comparable for the aglycosylated and the glycosylated protein.
Together, these results show that Api m 10 is able to induce relevant effector cell activation and thus has to be considered as an important allergen in A. mellifera venom. Moreover, carbohydrates appear to contribute to the overall IgE reactivity to a minor extent only.
Evaluation of native Api m 10 in Apis mellifera venom
To clarify the molecular integrity and concentration of Api m 10 in the native HBV and in preparations used for VIT, a monoclonal human IgE antibody was generated (Fig. S2), the reactivity of which with Api m 10 was verified in ELISA and immunoblot (Figs 1C and 2A). A human monoclonal IgE antibody with specificity for acid phosphatase (Api m 3) was applied analogously (Fig. 6A, lower panel).
Applying this antibody in immunoblots of crude HBV, a major band at 55 kDa was detected (Fig. 6A, upper panel). To address the quantity of Api m 10 in HBV, the monoclonal IgE was applied to immunoblots providing serial dilutions of recombinant Sf9-derived Api m 10 and HBV (Fig. 6A, upper panel). Densitometric quantification suggested a concentration in the range of 8 ± 1 μg per g of crude HBV, corresponding to 0.8% ± 0.1% of dry weight compared with 1.7% ± 0.4% for Api m 3.
Additionally, crude venom and three different HBV preparations for VIT from different allergen extract producers were analyzed (Fig. 6B, left panel). Api m 1 was used as a control for loading equal amounts (Fig. 6B, lower panel). In stark contrast to the crude venom, no reactivities at all were obtained for any of the three preparations. Api m 3 could be readily detected in whole HBV as well as in two of the preparations, although in significantly lesser amounts (Fig. 6B, right panel). These data demonstrate that the concentrations of the putatively labile HBV allergen Api m 10 and the already established and classical allergen Api m 3 are dramatically reduced in therapeutic venom preparations investigated in this study.
In the last decades, much effort has been spent to characterize a plethora of allergens in a variety of sources. However, hymenoptera venoms as the allergenic source causing the highest incidence of anaphylaxis and sometimes even fatal consequences remain inadequately characterized with regard to their molecular composition. Despite the fact that the higher abundance allergens of HBV (Api m 1, Api m 2, and Api m 4) have already been characterized in detail years ago, comparatively little is known about the sensitizing potential, the allergenicity, and the clinical relevance of the lower abundance allergens. One of these putative allergens is Api m 10, a protein of unknown function, peptides of which were recently identified in HBV by two independent groups (11, 12).
Contrary to a previous report (13), we readily obtained soluble recombinant Api m 10 using both the eukaryotic baculovirus expression system or the strategy of chitin binding domain fusion followed by autocatalytic intein-mediated cleavage in E. coli. The difference of molecular weight between the aglycosylated and the insect cell-derived, glycosylated proteins obviously stems from extensive posttranslational modifications, such as glycosylation.
It has increasingly been recognized that cross-reactivities based on glycosylation, namely by alpha-1,3-core fucose residues of N-glycans, represent a major concern for diagnostic approaches in hymenoptera venom allergy. In this regard, the use of Sf9 insect cell lines constitutes a strategy recently reported by us to circumvent the establishment of CCDs (18) under the aegis of an autologous eukaryotic expression. Further support for this concept becomes evident by the augmented reactivities of the HighFive cell-produced Api m 10 displaying CCDs (see Figs 1 and 2).
Using the differentially glycosylated protein variants, up to approximately 50% of HBV-allergic patients showed reactivity in ELISA and basophil activation tests with recombinant Api m 10 beyond CCD reactivity, thus rendering it an important sensitizing component of HBV. This finding was supported by a pronounced IgG4 immune response in over 50% of beekeeper sera. Induction of IgG4 is an intrinsic potential of all allergens known to be good inducers of IgE, as demonstrated for Api m 1 and other relevant allergens (19). Interestingly, the findings obtained with the aglycosylated protein expressed in E. coli matched those with the glycosylated protein. The complete lack of reactivity in the group of YJV-sensitized patients without sensitization to HBV additionally suggests the absence of a homolog in vespid venom rendering Api m 10 a genuine marker for HBV allergy.
As evident from the data of this study, Api m 10 appears to be a crucial but delicate component of HBV preparations. By the use of the monoclonal anti-Api m 10 IgE, we were for the first time able to detect native Api m 10 in unprocessed HBV. It is present as intact protein with concentrations (0.8%) only slightly lower than those of Api m 3 (1.7%) and Api m 2 (2–3%) (20). Thus, it was surprising that Api m 10 is apparently absent or at least vastly underrepresented in three therapeutic preparations of HBV from independent vendors. Obviously, downstream processing of venoms for VIT affects the distribution of venom proteins, resulting in the loss of particular low abundance components.
While the exact role of lower abundance allergens for therapeutic efficacy of VIT with HBV remains to be analyzed, the presence of lower abundance or inherently labile venom components should become imperative as proof of extract quality.
In summary, Api m 10 from A. mellifera venom for the first time could be produced in soluble form in baculovirus-infected insect cells as well as in E. coli. The results obtained with these proteins suggest an important role for Api m 10 as sIgE-sensitizing component in HBV allergy beyond its carbohydrate-based cross-reactivity and emphasize its clinical relevance. This relevance will additionally be verified in future studies on the relevance of different recombinant venom allergens for therapeutic efficacy. Results of such studies might help to elucidate the needs and pitfalls of successful VIT.
Gratefully acknowledged are the technical contributions by Beate Heuser.
SB, HS, YM, and IB performed cloning and expression; SB, HS, IB, MO, and ES involved in manuscript preparation; MM carried out basophil activation tests; CL and DU involved in recruitment of beekeepers; TG, JR, and RB analyzed the data; and MO and ES involved in study design and analysis.
Conflicts of interest
The authors declare that they have no financial, research, organizational, or other interests to disclose that are relevant to the execution of this research or this publication.