Lorenz et al.: IgE-Binding Properties of Recombinant Food Allergens
Internet Symposium on Food Allergens 3(3):1-36  (2001) [http://www.food-allergens.de]


Extracts from allergenic foods are the source of natural food allergens as they are applied in in vivo and in vitro tests for diagnostic and investigational purposes. Alternatively, it has become possible to produce allergens of interest in vitro by cultured, protein-expressing cells such as transformed E.coli. Expressing cells synthesise and express the protein according to an inserted DNA molecule (cDNA) which is a complementary copy of the messenger RNA (mRNA) of the protein of interest.
The main difficulty in producing recombinant food allergens is the identification and selection of the allergen-specific cDNA from the pool of diverse cDNA molecules generated from the cellular mRNA. If the amino acid or nucleotide sequence is (partly) known, the target cDNA can be amplified by PCR strategy with specific primers (Scheurer et al. 1997, Hoffmann-Sommergruber et al. 1999b, Son et al. 1999, Scheurer et al. 2000b). If there is no sequence information on the food allergen, a cDNA library has to be constructed which can be screened, for example, with sera of food-allergic subjects (Burks et al. 1995b, Scheiner et al. 1997, Kleber-Janke et al. 1999).
The expression of a recombinant allergen can be very laborious. Once established, it is possible to produce virtually unlimited amounts of the food allergen of interest. In comparison to natural allergens in extracts, recombinant food allergens are available in defined quantity and controlled quality. They may serve as pure reference materials for the standardisation of allergen extracts. During the last ten years, many recombinant allergens have been produced because they offer new prospects of an improvement  of diagnosis and therapy of allergies as well as molecular allergology such as immunological and structural studies on allergenic proteins. This review summarises the data of more than 40 recombinant food allergens derived from fruits, vegetables, milk, nuts and seafood. Special attention is directed towards the structural differences between natural and recombinant food allergens and their effects for the IgE-binding and IgE-cross-linking capacity of these molecules.


E.coli was the most widely used cell type for the expression of approximately 40 recombinant food allergens, whereas five recombinant food allergens were produced in eukaryotic yeast cells of the species Pichia pastoris (see tables 1 and 2). As E.coli are prokaryotic they, in contrast to eukaryotic cells, do not possess the intracellular machinery for post-translational modifications such as glycosylations or disulfide bridges (Bhatia & Mukhopadhyay 1998). As a consequence, recombinant food allergens can structurally vary from the native analogues as these are produced from eukaryotic plant or animal cells. IgE antibodies bind to linear stretches of amino acids on an allergenic protein or to discontinuous epitopes formed by the secondary or tertiary structure of the protein (Kuby 1994). Therefore, changes of the structure may alter allergenic determinants and accordingly may modify the allergenic potential of a protein. The following paragraphs point out possible structural differences between recombinant food allergens and their natural counterparts.

2.1 Post-translational modifications

The lack of post-translationally added carbohydrates is a characteristic of the bacterial expression system. Lacking glycosylations can have an impact on the IgE-binding capacity due to a possibly altered tertiary structure or the absence of carbohydrate epitopes (Smith et al. 1996, Petersen et al. 1998, van Ree et al. 2000). Yeast cells glycosylate polypeptides, but in a different manner than other cell types. Though yeast cells attach monosaccharides to the polypeptide chain at the correct glycosylation site, they tend to hyperglycosylate and link the monomers in varying patterns (Tschopp et al. 1987, Sudbery 1996, Petersen et al. 1997, Cereghino & Cregg 2000). Thus, the carbohydrate side chains of yeast-expressed proteins can differ in size and composition. Pichia pastoris yeast cells are known to perform N-linked glycosylations resulting in long carbohydrate structures of the high-mannose type without terminal alpha-1,2 mannose (Grinna & Tschopp 1989, Sudbery 1996, Verma et al. 1998). Food allergens from peanut (Ara h 1), avocado (Pers a 1), rapeseed (BnI), wheat (lipid transfer protein [LTP]) and carrot (LTP) were produced in Pichia pastoris cells (Sowka et al. 1998, Maid 1998, Asero et al. 2000b, Villalba et al. 2000), but none of these five recombinant allergens were applied to study glycosylations: Recombinant Pers a 1 (rPers a 1) lacked a potential N-glycosylation site and the proportion of O-linked sugars in natural Pers a 1 (nPers a 1) was very low (Sowka et al. 1998). Recombinant lipid transfer proteins of wheat and carrot were not tested on glycans (Asero et al. 2000b). The glycoprotein Ara h 1 from peanut could not be obtained in a glycosylated recombinant form because the extracellular targeting of the expressed protein failed. Finally, rAra h 1 was isolated from the cytoplasm as a non-glycosylated protein (Maid 1998).

Disulfide bridges are a second post-translational modification with a high impact on the conformation and therefore the IgE-binding activity of a protein. E.coli are not able to introduce disulfide bonds into polypeptide chains intracellularly but in periplasm where sulfhydryl-connecting enzymes are present. Thus, to express proteins with disulfide bonds in E.coli, the protein has to contain special signal sequences targeting the polypeptide to the periplasm. Recombinant milk allergen Bos d 5 was expressed according to this technique and contained at least one of two possible bonds (Chatel et al. 1996).

Further post-translational modifications, not performed in prokaryotic cells, are hydroxylation of prolines or N-terminal acetylations which can influence the IgE-binding activity of proteins (Petersen et al. 1998).

2.2 Tertiary structure

The correct folding of a recombinant protein is also of great importance, especially when the IgE antibodies bind to discontinuous epitopes. The folding machinery of eukaryotic cells is more advanced than that of E.coli. To benefit from the eukaryotic processing pathways and to obtain correctly folded recombinant food allergens, Pichia pastoris was chosen as the expression cell in the case of Pers a 1 from avocado and rapeseed allergen (Sowka et al. 1998, Villalba et al. 2000). During the expression of rPers a 1 in Pichia pastoris, the leader peptide was cleaved and rPers a 1 showed enzymatic activity which was comparable to that of nPers a 1. These results were taken as proof of the correct processing and folding of the protein in the yeast system (Sowka et al. 1998). Likewise, the proper folding of yeast-expressed rapeseed allergen r"BnI" was demonstrated by equivalent spectroscopic features in comparison to the natural analogue (Villalba et al. 2000). In addition, several E.coli-expressed food allergens were accurately folded. For example, the recombinant mustard allergen rSin a 1 showed the same resistance to trypsin digestion as nSin a 1. As trypsin resistance is regarded as a conformation-dependent property, it was concluded that rSin a 1 was properly folded (González De La Peña et al. 1996).

As frequently observed during expression in bacteria, several recombinant food allergens aggregated as insoluble "inclusion bodies" (Batt et al. 1990, González De La Peña et al. 1993, González De La Peña et al. 1996, Chatel et al. 1996, Karamloo et al. 2001). Inclusion bodies had to be solubilised under harsh denaturing conditions, for example with urea and acidic pH (Karamloo et al. 2001), urea and DTT (Chatel et al. 1996) or by guanidine-HCl treatment plus DTT (Batt et al. 1990). Subsequently, the solubilised proteins underwent successful refolding procedures (Chatel et al. 1996, Karamloo et al. 2001) according to special renaturation protocols (Cho et al. 1994, Boehm & Rösch 1997). In contrast, inclusion bodies containing aggregated rSin a 1 could not be solubilised by detergents, organic solvents, urea or guanidinium chloride (González De La Peña et al. 1996).

2.3 Codon usage

Problems in the production of peanut allergen Ara h 1 in bacteria can serve as an example to demonstrate a specific problem with bacterial expression systems. During the biosynthesis of rAra h 1, many fragments occurred and the yield of rAra h 1 was rather low. It was concluded that rare codons used by E.coli were responsible for these phenomena. Two of six possible arginine-coding triplets are scarce in E.coli, but frequent in Ara h 1-coding mRNA and the corresponding tRNA molecules are rare in E.coli. This fact possibly led to the termination of translation (Burks et al. 1995b, Maid 1998). Truncation or low expression rates were also observed during the production of rAra h 2, rAra h 6 and rAra h 7 which are rich in rare arginine codons (Kleber-Janke et al. 1999, Kleber-Janke & Becker 2000). However, high-level expression of the peanut allergens Ara h 1, Ara h 2 and Ara h 6 could be gained by the application of modified E.coli cells. Those expressing cells were transformed with extra tRNA genes specific for rare arginine codons (Kleber-Janke & Becker 2000).

2.4 Technical reasons leading to altered structures

As illustrated above, structural differences between recombinant food allergens and the natural counterparts can be due to the different protein biosynthesis pathways of bacterial and eukaryotic cells. In addition, technical aspects of the expression procedure can cause structural differences.
The majority of recombinant food allergens were cloned with fusion peptides for which specific ligands were available. This enabled the purification of the recombinant protein from the bulk of cellular proteins on an affinity column specific for the fusion peptide. A very common fusion peptide in the production of recombinant food allergens were tags consisting of six or ten histidine (his) residues at the N- or C terminus of the protein (Scheurer et al. 1997, Burks et al. 1999, Rabjohn et al. 1999, King et al. 1999, Beardslee et al. 1999, Hoffmann-Sommergruber et al. 1999b, Son et al. 1999, van Do et al. 1999, Karamloo et al. 2001). In most cases, these his-tagged recombinant proteins were purified by metal-chelate affinity chromatography on an Ni-NTA-matrix column (NTA: nitrilotriacetic acid) (Scheurer et al. 1997, Rabjohn et al. 1999, Hoffmann-Sommergruber et al. 1999b, Son et al. 1999, Karamloo et al. 2001). Several other food allergens were fused to the enzyme glutathione-S-transferase (GST) from Schistosoma japonicum (Leung et al. 1994, González De La Peña et al. 1996, Leung et al. 1996, Teuber et al. 1998, Leung et al. 1998a, Leung et al. 1998b, Teuber et al. 1999). These recombinant GST-fused food allergens were purified from cell lysate by using a glutathione column. Other fusion peptides used with recombinant food allergens were maltose-binding protein (MBP) (Rihs et al. 1999), beta-galactosidase (beta-gal) (Bugajska-Schretter et al. 1998, Bugajska-Schretter et al. 1999) and intein-CBD protein (CBD: chitin-binding domain) (Lüttkopf et al. 1999, Vieths et al. 2000). In some cases, the fusion peptides were cleaved by proteolytic or chemical treatment after purification to obtain the recombinant non-fusion protein (Lüttkopf et al. 2000b). In the case of enzymatic cleavage of his-tags, additional amino acids remained at the terminus of the recombinant non-fusion protein (Scheurer et al. 1997, Son et al. 1999, Karamloo et al. 2001). Moreover, several recombinant food allergens were directly subjected to IgE-binding assays without cleavage of the fusion peptide (see table 2). Fusion peptides such as his-tags with a molecular mass of about 3-4 kDa (Scheurer et al. 1997) were rather small compared to GST which had a size of about 26 kDa. In the recombinant constructs rSin a 1-GST or rJug r 1-GST, the fusion part GST even had a higher mass than the respective allergens of 18.8 kDa or 15-16 kDa (González De La Peña et al. 1996, Teuber et al. 1998).
Fusion tags are not necessarily required for cloning and purification of food allergens. For example, Mal d 1 the major allergen from apple was expressed as his-tagged protein (Son et al. 1999) as well as non-fusion protein which was purified by an anion-exchange column (Vanek-Krebitz et al. 1995). Plant profilins required no linkage to fusion peptides as they bind physiologically to proline-rich sequence stretches and therefore could be purified by poly-L-proline affinity chromatography (Karamloo 1999a, Scheurer et al. 2000).

Heterodimeric allergens such as seed storage proteins consist of two polypeptide chains linked by disulfide bridges. They emerge from a long precursor polypeptide which is enzymatically cleaved into a large and a short chain which are then linked by disulfide bonds. Allergenic seed storage proteins were cloned in E.coli as the unprocessed precursor form. The cDNA coding for the precursor molecule was selected from a cDNA library and used as template for production in E.coli. According to that method, the 2S seed storage proteins Ara h 6, Ara h 7 and Jug r 1 from walnut as well as the 11S seed storage protein Ara h 3 were cloned (Smith et al. 1997, Kleber-Janke et al. 1999). Recombinant 2S albumins from mustard and rapeseed were produced by a similar strategy. The cDNA used for the expression was not obtained from a library but constructed by fusing DNA fragments amplified from genomic DNA, each coding for one subunit of the precursor. The resulting recombinant allergens differed from the original precursor because they carried three additional internal amino acid residues (González De La Peña et al. 1996, Villalba et al. 2000).


As shown in section 2, recombinant food allergens are, in general, not an identical copy of the natural analogue. Therefore, changes in the IgE-binding property of the recombinant protein may occur. Through the loss of linear or conformational IgE-binding epitopes, the allergenic reactivity may be altered or even be eliminated. Thus, it is important that the IgE binding properties of a recombinant food allergen are well characterized before it is used in research, diagnostic or therapeutic applications. The present data collection summarises the recombinant food allergens published up to July 2000 (tables 1 and 2). The criterion for enlisting a recombinant food allergen is that it was tested for IgE-binding activity. The mere in vitro production of a molecule known as a food allergen (such as ovalbumin or bovine lactoglobulin) was not sufficient to be included in the list. The cloning of the postulated celery allergen Api g 3 (rApi g 3) was reported, but as rApi g 3 was not assayed with patients' sera it was not included in the list (Scheiner et al. 1997).

The level of characterization of an allergen depends on the diversity of test systems applied. In immunoblot tests under denaturing conditions, loss of conformational epitopes may occur. In inhibition tests, the inhibitor protein is in fluid phase and thus in correct conformation. Mediator release assays show the biological activity indicating that the recombinant food allergen is able to cross-link cell-bound IgE and elicit degranulation of basophils.
All recombinant food allergens listed in table 2 were at least tested in IgE immunoblotting and showed a considerable IgE-binding capacity. Recombinant Api g 1 from celery, for example, had a similar IgE reactivity to that of the natural analogue. When tested by IgE immunoblotting, both rApi g 1 and nApi g 1 in celery extract were recognized by sera from 11 patients with symptoms after ingestion of celery (Breiteneder et al. 1995). The recombinant apple allergen rMal d 1, like nMal d 1, bound IgE of 13 sera of patients in immunoblot analysis and EAST (enzyme allergosorbent test) (Son et al. 1999), and similarly, both rDau c 1 and nDau c 1 from carrot were IgE-reactive to sera of six carrot-allergic subjects (Hoffmann-Sommergruber et al. 1999b). Several recombinant food allergens underwent a panel of different tests including mediator release assay providing evidence for IgE-binding and cross-linking activity (Scheurer et al. 1997, Scheurer et al. 1999, Karamloo 1999a, Son et al. 1999, Fötisch et al. 1999, Scheurer et al. 2000b, Vieths et al. 2000, Karamloo et al. 2001). Recombinant cherry allergen rPru av 1 for example, caused histamine release in basophils of three cherry-allergic patients as did cherry extract (Scheurer et al. 1997). Three isoforms of rMal d 1 from apple were compared in mediator release assay with nMal d 1. One rMal d 1 isoform provoked release similar to nMal d 1. This isoform displayed the highest EAST classes with sera from 13 apple-allergic patients. The two other rMal d 1 isoforms with lower EAST classes elicited moderate releases (Son et al. 1999). The first and until now the only recombinant food allergen applied in skin prick tests was rApi g 1 which elicited positive skin reactions in Api g 1-sensitive subjects (Hoffmann-Sommergruber et al. 1999a).

An important question is whether structural variations of a recombinant food allergen influence the IgE-binding properties. Information on the impact of carbohydrates is rare. The known glycoprotein allergens Ara h 1, wheat glutenins and wheat gliadins were expressed in E.coli and thus did not contain carbohydrate structures. However, all recombinant forms bound IgE antibodies of patients' sera in immunoblot tests or RAST (Burks et al. 1995a, Burks et al. 1995b, Maruyama et al. 1998, Kleber-Janke et al. 1999). Eight out of 11 patients' sera with IgE to nAra h 1 recognized rAra h 1 in immunoblotting indicating that a considerable number of epitopes are unaffected by carbohydrate components (Burks et al. 1995a).

A correct folding of recombinant food allergens was concluded by some authors from results of circular dichroism (CD) spectroscopy, immunological assays, and tests for enzymatic activity (González De La Peña et al. 1996, Chatel et al. 1996, Villalba et al. 2000). The evidence of IgE-binding activity supported this conclusion. Profilins for example, are regarded as allergens with a highly conformation-dependent IgE reactivity (Valenta et al. 1991, Vrtala et al. 1996, Rihs et al. 1999). Nevertheless, recombinant profilins (rApi g 4, rAra h 5, rGly m 3, rPru av 4, rPyr c 4) bound IgE in EAST tests and two of them even elicited mediator release (Kleber-Janke et al. 1999, Scheurer et al. 2000a, Karamloo 1999a, Rihs et al. 1999, Scheurer et al. 2000b). A correct conformation was assumed for E.coli-expressed rBos d 5 which contained at least one of two possible disulfide bonds. Studies with monoclonal and polyclonal antibodies revealed evidence that natural and recombinant Bos d 5 had a comparable conformation. Correspondingly, the IgE reactivity of nBos d 5 and that of rBos d 5 in five patients' sera were similar (Chatel et al. 1996).

In several cases, recombinant food allergens were extracted from inclusion bodies and successfully refolded by subsequent renaturation procedures. Recombinant pear allergen rPyr c 1, for example, was characterized in various tests including mediator release assay with a large number of sera and was found to have a high biological activity (Scheurer et al. 1999, Vieths et al. 2000). Recombinant Pyr c 1 detected 89% and pear extract 85% of the sera of 61 pear-allergic patients indicating that rPyr c 1 was more active than the extract (Vieths et al. 2000).

Differences in the amino acid sequence can alter the IgE-binding pattern of allergens. Mutation studies revealed that the substitution of one single amino acid drastically reduced the ability to bind IgE (Scheurer et al. 1999, Son et al. 1999). The recombinant mustard allergen rSin a 1 contained three additional amino acids. Inhibition studies with monoclonal antibodies showed slight conformational changes between natural and recombinant Sin a 1 due to the additional residues, but both forms inhibited the binding of IgE to a similar degree (González De La Peña et al. 1996). In some cases, recombinant food allergens with N- or C terminally added fusion peptides were used in IgE assays. There was no indication that the fusion tags had an influence on the IgE-binding capacity of the recombinant allergen. Even the fusion peptide GST with a relatively high molecular mass of about 26 kDa did not affect IgE-binding activity. Tropomyosins from several seafoods with a molecular mass of 34 kDa were expressed as fusion proteins with GST and bound IgE from patients with crustacean allergy (Leung et al. 1998a, Leung et al. 1994, Leung et al. 1996, Leung et al. 1998b). In rJug r 1-GST, the difference of molecular mass was considerable: the Jug r 1 had a molecular mass of 15-16 kDa while the fusion part was estimated at 26 kDa. The fusion protein bound IgE in immunoblotting whereas the binding of IgE to GST was excluded (Teuber et al. 1998). His-tags which had a molecular mass of about 3-4 kDa and were thus smaller than the respective allergen had no apparent effect on the IgE binding activity. Both the recombinant his-tagged Pru av 1 and the non-fusion rPru av 1 reacted with serum IgE antibodies. In inhibition studies, the same results were obtained with fusion and non-fusion forms (Scheurer et al. 1997, Scheurer et al. 1999).


4.1 Diagnostics

Recombinant food allergens offer new perspectives to solve current problems in diagnostics of food allergies. In vivo and in vitro tests are performed with extract preparations of the allergenic food. To date, these commercially available test extracts are standardised neither with respect to content of allergens nor with respect to biological activity (Bruijnzeel-Koomen et al. 1995, AAAAI 1997). There are differences between extracts from various manufacturers and even between batches from one manufacturer (Sampson 1988, Ortolani et al. 1989). A raw food, used as source for the production of extracts, can vary in its allergen content. In the case of fruits, factors such as degree of ripening or strain can influence the amount of allergens (Chapman et al. 1987, Vieths et al. 1993, Vieths et al. 1995, Fernández-Rivas & Cuevas 1999). Apples of the variety "Golden Delicious", for example, were found to contain eleven times more major allergen Mal d 1 than those of the variety "Gloster" (Son et al. 1999). In addition, it is possible that food extracts lose activity during extraction procedure or during storage (Bernhisel-Broadbent et al. 1992b, Norgaard et al. 1992, Norgaard et al. 1995, Vieths et al. 1998). The IgE-binding capacity of allergens in an extract can be altered by enzymatic degradation or by interaction with phenolic components (Björkstén et al. 1980, Ortolani et al. 1989, Deluze et al. 1991, Vieths et al. 1994, Vieths et al. 1995, Rudeschko et al. 1995, Varjonen et al. 1996). It was shown that of four commercially available apple skin test extracts none could verify the existing sensitisation to apple in 72 patients (Vieths et al. 1995). In conclusion, these data show that results of a test performed with an extract may be false-negative.

Pure recombinant food allergens could be used as reference material to standardise extracts for in vitro use and skin tests. Diagnostic results obtained with standardised preparations would be reproducible and more precise. Recombinant allergens may not only improve the quality of extracts, but also could replace extracts in in vitro and in vivo tests. Tests could be carried out with a panel of recombinant food allergens representing the allergenic components of a food. The prerequisite for this concept is that all allergens of a food are known and available in recombinant form that is active and shares the same IgE-binding capacity as the natural analogue. In screening tests, problematic food extracts could be supplemented by recombinant food allergens or mixtures thereof, to make sure that all sensitised patients are identified by the screening procedures.

A new aspect of testing with single recombinant allergens is that it will be possible to compile a patient's individual sensitisation pattern, whereas with extracts, in general, only sensitisation to the whole food is determined. A knowledge of the sensitisation pattern, will make prognostic statements as well as preventive recommendations possible. This aspect bases on the phenomenon that birch pollen-allergic patients reacting to birch pollen allergen Bet v 1, will probably develop a food allergy against a certain spectrum of fruits and vegetables, such as cherry, pear, apple and celery. This is due to allergens of approximately 17-18 kDa in the respective food which belong to the Bet v 1-family, a group of allergens with sequence homology to the Bet v 1 allergen. The symptoms are caused by cross-reactivities between the members of the Bet v 1-family, as shown by studies with recombinant proteins (Hoffmann-Sommergruber et al. 1999a, Scheurer et al. 1999, Hoffmann-Sommergruber et al. 1999b, Karamloo et al. 2001, Kazemi-Shirazi et al. 2000, Lüttkopf et al. 2000a). In these cases, it is favorable to reduce consumption of food containing Bet v 1-related allergens. If the sensitisation pattern reveals reactivity to profilin, avoidance will not be possible because profilin occurs in all eukaryotic cells (Vieths et al. 1996a, Vieths et al. 1996b). Allergenic profilins have been identified in many fruits, vegetables and pollens. Studies have demonstrated cross-reactivities between allergenic profilins from different organisms. Thus, ubiquitous profilin is regarded as a pan-allergen in plants (see table 1) (Valenta et al. 1992).

Initial experience with recombinant food allergens as diagnostic tools is promising. For example, a sample of 43 sera of patients with clinical hazelnut allergy, verified by double-blind placebo-controlled food challenges (DBPCFC), was tested by EAST with recombinant hazelnut allergen rCor a 1.0401 and hazelnut extract. The one recombinant allergen rCor a 1.0401 proved 41 of 43 sera positive while the extract was only recognized by IgE of 32 of 43 sera (Vieths et al. 2000, Lüttkopf et al. 2000b). Pear extract bound 85% of 61 sera of pear-allergic patients while one recombinant pear allergen (rPyr c 1) bound IgE of 89% of the sera (Vieths et al. 2000). In these two examples one single recombinant food allergen was more sensitive in the EAST test than the whole extract and led to a lower number of false-negative results. This fact may be caused by higher amounts of the allergen as recombinant proteins in the tests, whereas the relative amounts of single allergens in the whole extracts were not sufficient to be detected by certain sera. The replacement of food extracts in screening tests by a panel of recombinant food allergens seems to be possible. In a recent study, six recombinant peanut allergens were able to identify all 40 peanut-sensitive individuals. Various individual sensitisation patterns were observed, but ultimately the six recombinant allergens were sufficient to detect all 40 sera (Kleber-Janke et al. 1999). Another example of recombinant allergens resembling the activity of extracts is the pear allergen rPyr c 1 in combination with the minor allergens rPyr c 4 and rPyr c 5 (Vieths et al. 2000).

It is a well-known fact that results of serological assays, skin tests or clinical tests do not always correlate (Bernhisel-Broadbent et al. 1989, Bernhisel-Broadbent et al. 1992a, Norgaard et al. 1992, Eigenmann & Sampson 1998, Majamaa et al. 1999). A group of 24 patients with a clinical history of celery allergy showed different results in three different test systems. All subjects had a positive SPT to crude celery and to celery extract, 20/24 reacted to extract in IgE immunoblotting and only 14/24 had a RAST class 1 and higher. In contrast to these results, in another group of 12 celery-allergic patients 11/12 had a positive SPT to fresh celery but only 3/12 reacted to celery extract in SPT, indicating that the extract applied in the latter group probably was not active (Hoffmann-Sommergruber et al. 1999a). With the help of recombinant food allergens as test material or as reference compound for standardisation it may be possible to highlight whether contradicting results of diagnostic tests depend on test compounds or test systems. In vitro tests, for example, are often false-positive with respect to clinical food allergy (Ebner et al. 1991). They determine sensitisation to IgE epitopes in foods, but do not prove clinical allergy. Thus, the predictive value of routine tests such as RAST or CAP is limited. The golden standard for verifying a food allergy is the DBPCFC, but it has to be performed by well-educated allergists and is very time-consuming and costly (Bock et al. 1988). Therefore, the DBPCFC will not become a routine technique. Clinically false-positive in vitro results may be due to cross-reactive structures in foods such as, for example, IgE-binding carbohydrate structures. Some authors postulate that these cross-reactive carbohydrate determinants (CCD), which are broadly present in foods (Vieths 1997, Aalberse & van Ree 1997), are clinically irrelevant and therefore cause false-positive in vitro tests (van der Veen et al. 1997, Aalberse 1998). Although this statement is not generally accepted (Fötisch et al. 1999), it is reasonable to investigate these reactivities to improve the predictive value of in vitro tests (van Ree et al. 2000). This could be another possible approach for applying recombinant food allergens.

4.2 Therapy

Food allergies may elicit severe acute hypersensitivity reactions and are a main cause of emergency treatment due to life- threatening anaphylaxis (Yunginger et al. 1988, Sampson et al. 1992, Yocum & Khan 1994, Hourihane et al. 1997a). Pumphrey et al. showed that more than 50% of anaphylactic reactions were due to food allergies (Pumphrey & Stanworth 1996). Patients with severe acute reactions to food are instructed to avoid the allergenic food, which can be very difficult to manage. When patients consume food to which they are allergic, it mostly happens accidentally and inadvertently (Sampson et al. 1992, Borelli et al. 1999, Sicherer et al. 1999). For that reason, patients with an anaphylactic history to food ingestion are equipped with emergency medications. In the case of peanut allergy it is virtually impossible to abstain from peanut because peanut proteins are widely used in food products and in the pharmaceutical industry (Moneret-Vautrin et al. 1994, Ewan 1996, Lever 1996). Since only micrograms of peanut protein are sufficient to elicit symptoms, contaminations of foods by hidden peanut proteins are a high risk for peanut-allergic subjects (Hourihane et al. 1997b).

Apart from avoidance of the offending foods, food allergies are treated with symptomatic medication. Current developments of causative therapeutic approaches are focused on DNA vaccination strategies (Roy et al. 1999, Adel-Patient et al. 2000) or oral desensitisation protocols (Patriarca et al. 1984, Patriarca et al. 1998, Bauer et al. 1999).
Up to now, there is no established causally effective immunotherapy for food allergies. In patients with pollen-related food allergy who underwent specific immunotherapy (SIT) by subcutaneous injection of pollen extracts, the food allergy improved in approximately 37% to 84% of the cases (Möller 1989, Henzgen et al. 1991, Henzgen et al. 1994, Kelso et al. 1995, Herrmann et al. 1995, Asero 1998a, Henzgen et al. 1999, Asero 2000). The positive effect of pollen SIT on food allergy reflects cross-reactivity of food and pollen allergens at T cell level. It was shown that birch pollen allergen Bet v 1 stimulated T cells specific for homologous apple allergen Mal d 1 and vice versa. Cross-reactive T cell epitopes could be characterized (Fritsch et al. 1998). Since pollen-related food allergy did not improve in all patients under pollen SIT, development of food allergen-specific SIT to treat unsuccessful cases may be required.

A few years ago, initial attempts were undertaken to develop a SIT for peanut allergy. A rush immunotherapy over a period of one month performed by injecting increasing doses of aqueous peanut extracts to peanut-allergic subjects was effective; at the end of one month patients tolerated higher doses of peanut but all subjects experienced allergic side effects during the therapy (Oppenheimer et al. 1992, Nelson et al. 1997). When the therapy was prolonged up to one year by application of maintenance doses, the positive effects of the rush phase persisted or improved in three of six patients. However, all patients required symptomatic treatment during the immunotherapy (Nelson et al. 1997).
These studies indicate that classical SIT by injection could be a strategy to treat food allergies. However, side effects should be avoided. Recombinant DNA technology could provide the techniques to develop effective and safe therapeutic reagents. First, recombinant isoforms and mutants of allergens with low IgE-binding capacity and retained T cell activity may be candidates. These modified allergens may be safe, because they do not elicit side effects due to cross-linking of cell-bound IgE antibodies. However, to be effective, they have to be T cell-reactive to induce T cell non-responsiveness to allergens. Second, small peptides and fragments containing T cell epitopes may be other candidates for SIT. In two studies, SIT of bee venom- and cat-allergic subjects with T cell epitope-containing peptides were successful (Müller et al. 1998, Norman et al. 1996). Peptides for therapy could be produced as defined synthetic peptides or recombinant allergen fragments.

Hypoallergenic isoforms, mutants, and fragments of food allergens were produced. For example, five recombinant fragments of the cherry allergen Pru av 1 of 59 to 116 amino acids in length were described. One of these fragments bound no IgE antibodies whereas four were IgE antibody-reactive (Scheurer et al. 1999). Three hypoallergenic recombinant fragments of soybean profilin Gly m 3, comprising 50 to 81 amino acids and spanning the whole allergen molecule scarcely inhibited binding of IgE to the entire recombinant allergen (Rihs et al. 1999). Hypoallergenic isoforms as well as mutants of the apple major allergen Mal d 1 were cloned. One of three recombinant Mal d 1 isoforms had a weaker IgE antibody-binding activity by EAST than the others (Son et al. 1999). A mutant of Mal d 1, differing by a single point mutation from the wild-type form, showed a drastically reduced IgE-binding capacity. This mutant bound no IgE antibodies of 13 patients' sera while the wild-type rMal d 1 was recognized by all sera (Son et al. 1999). Three of seven mutants of the cherry allergen Pru av 1 with one or two point mutations showed a strongly reduced or even abolished IgE-binding activity (Scheurer et al. 1997). Hypoallergenic recombinant mutants of peanut allergens Ara h 1, Ara h 2, Ara h 3 (Burks et al. 1999, King et al. 1999, Shin et al. 1999, Burks et al. 2000), Mal d 1 and Api g 1 (Ferreira et al. 2000) and potato allergen Sol t 1 (Alibhai et al. 2000) containing amino acid substitutions in IgE-binding epitopes were reported.
It has to be considered that "hypoallergenicity" of an isoform or mutant is determined by the epitope recognition of the individual patient in serological or skin tests. A mutant of rAra h 2 was tested with 16 sera of peanut-sensitive patients. A strong decrease in IgE binding was shown in 12 of 16 sera, while three sera bound the modified allergen similar to the wild-type allergen, and in one serum the binding was even increased (Burks et al. 1999). Thus, the safety of a modified allergen for a patient has to be tested before any kind of application, and it might be necessary to develop a panel of different immunotherapeutic allergens. The search for this patient-tailored panel of hypoallergenic isoallergens or mutants may be extensive. However, there is no risk of new sensitization to other proteins of the allergenic source, which theoretically exists when whole extracts are applied.
To be applied as therapeutic reagents, the above mentioned isoforms, mutants and fragments have to show that they have (1) low IgE-binding capacity and (2) T cell reactivity and that they are (3) therapeutically effective. So far, the incapacity to bind IgE antibodies was only shown in in vitro tests. Skin tests still have to be performed. Additionally, the T cell reactivity of the hypoallergenic derivatives has to be determined. So far, recombinant food allergens were not used as therapeutic reagents. Only in the case of peanut allergen Ara h 2, was the T cell-stimulatory ability of a hypoallergenic rAra h 2 mutant determined and was comparable to that of wild-type Ara h 2 (Burks et al. 1999). More information exists on recombinant non-food allergens. Several hypoallergenic isoforms, mutants and fragments displayed weak IgE-binding activity not only in in vitro tests but also in skin tests and furthermore induced T cell proliferation (Kauppinen et al. 1999, Takai et al. 1999, Ferreira et al. 1996, Zeiler et al. 1997). In animal model recombinant inhalant allergens proved to be effective therapeutics in SIT (Yasue et al. 1998, Yasue et al. 1999, Vrtala et al. 1995).
These data on recombinant hypoallergenic derivatives of food allergens indicate new approaches to SIT of food allergies, but more studies, especially on T cell-stimulatory capacity of the derivatives, have to be performed.

4.3 Research

Recombinant DNA technology provides tools to investigate the possible links between structure and allergenicity. With recombinant fragments of allergens, for example, one is able to characterize the location of epitopes and whether they are linear or conformational. Allergenic shrimp tropomyosin was mapped for epitopes by recombinant fragments of 13 to 22 amino acids in length and synthetic peptides. The epitopes were located at the C terminus and in the centre of the allergen sequence (Reese et al. 1997, Reese et al. 1999). Five recombinant fragments of the cherry allergen Pru av 1, with a size of 59 to 120 amino acids and spanning the whole molecule, displayed weak IgE-binding activity, with one fragment which bound no IgE antibodies (Scheurer et al. 1999). The results indicated that Pru av 1 epitopes are mainly conformational. Similarly, the IgE-binding capacity of soybean profilin Gly m 3 depends on the structural integrity of the entire molecule, as deduced from tests with recombinant fragments (50-81 amino acids) which were not able to inhibit binding of IgE antibodies to rGly m 3 (Rihs et al. 1999).

Recombinant mutants of food allergens are further tools to reveal strucutural characteristics of epitopes and relevance for IgE-binding capacity. Mutants of three related allergens of birch pollen (Bet v 1), cherry (Pru av 1) and apple (Mal d 1) with regards to the amino acid 112 (or equivalent 111 in Mal d 1) were constructed. The replacement of original serine by proline drastically reduced the IgE-binding ability of all three allergens compared to that of recombinant wild-type form (Scheurer et al. 1999, Son et al. 1999). The results showed that the amino acid at position 112 is a key residue in proteins of the Bet v 1-family with implications for IgE reactivity of the entire molecule. Similar consequences from equivalent mutations have also been reported from homologous allergens Mal d 1, Api g 1 and Bet v 1. Six substitutions in IgE-binding epitopes similarly reduced IgE-binding capacity of the allergens (Ferreira et al. 2000).

Furthermore, involvement of certain structures in allergenicity can be studied on the basis of spectroscopic and immunological data obtained by recombinant proteins. In the cherry allergen Pru av 1 two mutations affecting the P loop region (amino acids 46-52) reduced the IgE binding activity in six of seven sera of patients indicating that the P-loop region possibly displays an IgE-binding epitope or, at least, is involved in forming a conformational IgE binding structure (Scheurer et al. 1999). Recombinant proteins are well-suited test compounds in spectroscopic techniques such as circular dichroism (CD) spectroscopy or heteronuclear NMR spectroscopy because they are available as pure substances in high amounts of controlled quality. Heteronuclear NMR spectroscopy can be performed with allergens which contain special detectable isotopes (13C, 15N, 1H) introduced during the expression of recombinant proteins. Spectroscopic studies were undertaken to investigate whether similar allergenicity of related proteins were based on related structural elements. A highly resolved tertiary structure of cherry allergen Pru av 1 was calculated on the basis of NMR data of double-labelled recombinant protein. Apart from the conclusion that cross-reactivity is based on structural homology, studies may be useful to identify conserved surface patches that may represent the epitope regions (Schweimer et al. 1900, Neudecker et al. 2000b).

Knowledge on structural characteristics of IgE-binding epitopes may serve to develop hypoallergenic processed foods. Peanut major allergen Ara h 1 or LTP from peach are both heat-stable food allergens containing linear epitopes (Koppelman et al. 1999, Brenna et al. 2000). Therefore, other technological processes than heat treatments have to be applied to reduce the allergenicity of foods. In contrast, celery allergen Api g 1, which has conformational epitopes, is heat-labile and heating as well as non-thermal procedures were able to decrease the IgE-binding activity (Jankiewicz et al. 1996, Vieths et al. 1998).
Another aspect of recombinant DNA technology is that nucleotide sequence data of a food allergen gained during cloning procedure can be applied for genetic immunisation with cDNA of an allergen. This is a new approach for an allergen-specific immunotherapy with the potency for treatment as well as for protecting from allergy (Roy et al. 1999, Adel-Patient et al. 2000).


Up to now, more than 40 recombinant food allergens have been expressed in vitro, most of them in E.coli, several in the yeast system (see table 1). The recombinant allergens may differ structurally from the natural counterpart, for various reasons. A most striking feature of E.coli-expressed recombinant proteins is the lack of post-translational modifications such as carbohydrate side chains, disulfide bonds (if expressed intracellularly), hydroxylated prolines or acetylations (Burks et al. 1995b, Kleber-Janke et al. 1999). Recombinant allergens cloned in yeast cells can be glycosylated but the glycans may differ in size and pattern from those of other eukaryotic cell types (Petersen et al. 1997). Some recombinant food allergens have been expressed with fusion peptides which have been applied as specific ligands for purifying expressed protein from the pool of cellular proteins (see table 2). After cleavage of the fusion peptide additional amino acid residues may remain at the recombinant food allergen (Scheurer et al. 1997, Son et al. 1999). Heterodimeric allergens have been cloned as monomeric precursor molecules (González De La Peña et al. 1996, Smith et al. 1997, Kleber-Janke et al. 1999, Villalba et al. 2000). Recombinant proteins may display conformations differing from wild-type proteins, for example, if they have to be solubilised from aggregates (inclusion bodies) (Batt et al. 1990, Chatel et al. 1996, Karamloo et al. 2001). Amino acid sequences as well as secondary and tertiary structures of a protein have an impact on linear and discontinuous IgE-binding epitopes. Therefore, structural differences may alter the IgE binding capacity of a recombinant allergen compared to the natural analogue. We have reviewed data of the recombinant food allergens cloned to date and concluded that they showed a considerable IgE-binding activity in vitro. Several of them even displayed biological activity in mediator release assays or skin tests (see table 2). In comparative IgE binding studies, natural and recombinant food allergens shared similar properties in many cases (Breiteneder et al. 1995, Chatel et al. 1996, Son et al. 1999, Karamloo et al. 2001). There are also examples of recombinant food allergens with an altered IgE reactivity compared to natural counterpart (Burks et al. 1995a, Burks et al. 1995b, Son et al. 1999).

The application of recombinant food allergens could possibly improve diagnostics and therapy of food allergies. They may serve as pure reference material for standardisation of food extracts. Biologically and quantitatively standardised food extracts could provide more precise results in diagnostic tests. Recombinant food allergens could be applied as diagnostic tools in addition to or instead of labile extracts. As a prerequesite, all allergens of a food have to be known and available in recombinant form of comparable IgE reactivity. In initial studies, tests with one, three or six recombinant allergens were more sensitive than the respective food extract (Kleber-Janke et al. 1999, Vieths et al. 2000). A panel of six recombinant minor and major peanut allergens were sufficient to verify 40 peanut-sensitive subjects (Kleber-Janke et al. 1999). With single recombinant food allergens, the patient's individual sensitisation pattern could be compiled with possible effects for therapy and prognostic statements.
So far, a causally effective specific immunotherapy against food allergies does not exist. The only means to handle food allergy are avoidance of the offending food, application of symptomatic medication or oral hyposensitisation strategies resulting in increased tolerance to the respective food (Patriarca et al. 1984, Bauer et al. 1999). In vitro-expressed "hypoallergenic" mutants, isoforms or fragments of food allergens with weak or lacking IgE binding potency are discussed as possible candidates for an immunotherapy. They could be effective if they have T cell reactivity and safe because of inability to bind and cross-link IgE antibodies. Hypoallergenic mutants, isoforms and fragments have been described, but T cell activity still has to be demonstrated (Scheurer et al. 1997, Burks et al. 1999, King et al. 1999, Son et al. 1999, Burks et al. 2000, Shin et al. 1999, Ferreira et al. 2000, Alibhai et al. 2000). Another approach for causative therapy by DNA vaccination depends on nucleotide sequence information obtained during cloning procedure (Roy et al. 1999, Adel-Patient et al. 2000).

Since high amounts of recombinant allergens can be produced and can be labelled with specific isotypes, they are suitable tools in spectroscopic techniques to resolve allergen structures and to investigate the link between conformational features and IgE binding. It has been demonstrated that cross-reactivity in homologous allergens depends on related structures (Schweimer et al. 1900, Neudecker et al. 2000b). Knowledge of the relevance of structure for allergenicity may have implications for the development of food processing techniques to create hypoallergenic foods (Vieths et al. 1998, Brenna et al. 2000).
In summary, recombinant food allergens are suitable tools to gain more insight into the allergenicity of food proteins. They offer prospects to solve current problems in the fields of diagnostics and therapy of food allergies and in molecular allergology.

Table 1: List of recombinant food allergens
Table 2: IgE-binding activity of recombinant food allergens


[Summary] [Article] [Table 1] [Table 2] [Abbreveations]
 copyright © 2001 by matthias besler - ONLINE PUBLISHER