Construction of a combinatorial IgE library from an allergic patient. Isolation and characterization of human IgE Fabs with specificity for the major timothy grass pollen allergen, Phl p 5.

To characterize human IgE antibodies with specificity for a major allergen at the molecular level, we have constructed an IgE combinatorial library from a grass pollen allergic patient. cDNAs coding for IgE heavy chain fragments and for light chains were reverse-transcribed and polymerase chain reaction-amplified from RNA of peripheral blood lymphocytes and randomly combined in plasmid pComb3H to yield a combinatorial library of 5 x 10(7) primary clones. IgE Fabs with specificity for Phl p 5, a major timothy grass pollen allergen, were isolated by panning. Sequence analysis showed that the 4 of the Fabs used the same heavy chain fragments which had combined with different kappa light chains. Soluble recombinant IgE Fabs were purified by affinity chromatography to Phl p 5 and, like natural IgE antibodies, cross-reacted with group 5 allergens from different grass species. The described approach should facilitate studies on the molecular interaction between IgE antibodies and allergens and encourages the consideration of specific IgE Fabs that are capable of interfering with allergen-IgE binding as potential therapeutic tools.

To characterize human IgE antibodies with specificity for a major allergen at the molecular level, we have constructed an IgE combinatorial library from a grass pollen allergic patient. cDNAs coding for IgE heavy chain fragments and for light chains were reverse-transcribed and polymerase chain reaction-amplified from RNA of peripheral blood lymphocytes and randomly combined in plasmid pComb3H to yield a combinatorial library of 5 ؋ 10 7 primary clones. IgE Fabs with specificity for Phl p 5, a major timothy grass pollen allergen, were isolated by panning. Sequence analysis showed that the 4 of the Fabs used the same heavy chain fragments which had combined with different kappa light chains. Soluble recombinant IgE Fabs were purified by affinity chromatography to Phl p 5 and, like natural IgE antibodies, cross-reacted with group 5 allergens from different grass species. The described approach should facilitate studies on the molecular interaction between IgE antibodies and allergens and encourages the consideration of specific IgE Fabs that are capable of interfering with allergen-IgE binding as potential therapeutic tools.
More than 20% of the population suffers from type I allergic reactions (allergic rhinitis, conjunctivitis, and bronchial asthma). The symptoms of type I allergy are due to release of mediators (e.g. histamine) resulting from the cross-linking of specific IgE antibodies, which are bound to allergic effector cells (mast cells and basophils). Studies on the primary structure of immunoglobulin E were initially hampered by the extremely low concentration of IgE (10 -400 ng/ml) in the serum. Due to the availability of IgE-secreting myeloma cells, it was, however, possible to characterize IgE antibodies by immunochemical, protein chemical, and finally molecular biological techniques (Bennich et al., 1968(Bennich et al., , 1973Ishizaka and Ishizaka, 1970;Terry et al., 1970;Kochwa et al., 1971;Flanagan and Rabbitts, 1982;Kurokawa et al., 1983;Seno et al., 1983). The cDNA sequence of human C⑀ could be determined (Flanagan and Rabbitts, 1982;Kurokawa et al., 1983;Seno et al., 1983), and those portions of C⑀ that interact with the high affinity Fc⑀ receptor were characterized as possible targets for a therapy of Type I allergic diseases (Helm et al., 1988;Nissim and Eshar, 1992;Presta et al., 1994).
To investigate the molecular interaction of IgE antibodies and allergens, studies on the V regions of specific IgE antibodies would be needed. Because of the low number of IgE-secreting B-cells in the peripheral blood of allergic patients (McKenzie and Dosch, 1989), a detailed study of allergen-specific IgE antibodies and in particular of their V regions has proven to be extremely difficult. In addition, it has been so far impossible to immortalize B-lymphocytes that were switched to specific IgE production in vivo.
Using PCR 1 techniques, nucleotide sequences of epsilon VH 5 transcripts from peripheral blood B-cells of atopic patients were analyzed, suggesting that the molecular characteristics of the ⑀V H regions argue for a selection process due to recurrent or chronic stimulation of the immune system by antigens (e.g.. allergens), but nothing is known about their specificities (van der Stoep et al., 1993).
In the present study, the isolation and characterization of human IgE Fabs with specificity for a major allergen is reported. For this purpose we have constructed an IgE combinatorial library from blood lymphocytes of a grass pollen allergic patient by reverse transcription and PCR amplification of cDNAs coding for IgE-Fd and L chains. The cDNAs were randomly combined in the pComb3H vector (Barbas et al., 1991;Kang et al., 1991a) and expressed on the surface of filamentous phage to allow the selection of IgE Fab-expressing phage clones by panning to given allergens. Purified recombinant timothy grass pollen allergens Phl p 1 , Phl p 2 (Dolecek et al., 1993), and Phl p 5 (Vrtala et al., 1993a) were used to determine the IgE specificities of the allergic patient.
Recombinant human IgE Fabs with specificity for Phl p 5, a major timothy grass pollen allergen (Vrtala et al., 1993a), were isolated by panning and analyzed.
The present approach may contribute to the molecular analysis of allergen-IgE interactions and may perhaps be useful to define recombinant Fabs, which, due to the lack of the Fc⑀ receptor binding site, may be envisaged as potential therapeutic tools that can compete with natural IgE antibodies for the allergen binding.

EXPERIMENTAL PROCEDURES
Characterization of the Grass Pollen Allergic Patient-For the construction of the IgE combinatorial library, peripheral blood mononuclear cells were obtained during the grass pollen season from a grass pollen allergic patient after informed consent was given. The patient suffered from allergic conjunctivitis and rhinitis during the grass pollen season, was skin test-positive for timothy grass, and had received no hyposensitization treatment. Total serum IgE was determined by radioimmunoassay with a RIACT kit (Pharmacia, Uppsala, Sweden) to be 500,000 units/liter and a RAST class of Ͼ4 for timothy grass pollen was measured. The patient was further tested for IgE reactivity to purified recombinant birch (Betv 1 (Breiteneder et al., 1989) and Bet v 2 (Valenta et al., 1991)) and timothy grass pollen allergens (Phl p 1, Phl p 2, and Phl p 5) to determine his allergogram Dolecek et al., 1993;Vrtala et al., 1993a).
Preparation of RNA and PCR Amplification of cDNAs Coding for IgE Heavy Chain Fragments and Light Chains from Peripheral Blood Mononuclear Cells of the Grass Pollen Allergic Patient-150 ml of heparinized blood were obtained from the allergic donor during the grass pollen season (with informed consent). Peripheral blood mononuclear cells were prepared by Ficoll-Paque density gradient centrifugation (Pharmacia) (Steinberger et al., 1995). RNA was prepared by a guanidinium isothiocyanate method (Davis et al., 1986). Several independent cDNA synthesis and PCR amplification reactions were carried out using a RNA PCR kit (Perkin-Elmer). In brief, total RNA (20 -60 g) was mixed with 10 -20 pmol of oligonucleotide primers specific for the constant region of the epsilon chains ( V1, 5Ј-AAT TTT GAG CTC ACT CAG CCC CAC; V3, 5Ј-TCT GTG GAG CTC CAG CCG CCC TCA GTG) were then used in a 100-l hot start PCR amplification at the following conditions: 1 cycle of 5 min at 95°C for denaturation, 50 s annealing at 54°C, and 50 s elongation at 72°C followed by 40 cycles: 1 min denaturation at 92°C, 50 s annealing at 54°C, and 50 s elongation at 72°C. PCR reactions were done using a combination of each constant region and variable region primer and pooled for the construction of the library. The sequences of oligonucleotide primers of the epsilon chains and light chains were synthesized according to (Kabat et al., 1987). Oligonucleotide primers specific for the variable region of the heavy chains and variable and constant regions of theand -chains were synthesized according to Persson et al. (1991) and Kang et al. (1991b).
Construction of the IgE Combinatorial Library in Plasmid pComb3H-The PCR products coding for IgE Fds and light chains were ethanol-precipitated, gel-purified, and cut with SpeI/XhoI and SacI/ XbaI, respectively (Boehringer Mannheim). The digested PCR products were again ethanol-precipitated and gel-purified. For the construction of the IgE combinatorial library, the light chains were first ligated into the SacI/XbaI site of pComb3H and transformed into Escherichia coli XL-1Blue to yield a light chain library of 3 ϫ 10 7 independent clones. The plasmid DNA containing the light chain library was then isolated, cut with SpeI/XhoI to release the heavy chain stuffer, and gel-purified. The ligation of the cDNAs coding for the IgE Fds into the light chain plasmid yielded a library of 5 ϫ 10 7 independent primary clones. All molecular biological manipulations used for the construction of the IgE combinatorial library followed the protocols of the Cold Spring Harbor Course on Monoclonal Antibodies from Combinatorial Libraries by Carlos F. Barbas and Dennis R. Burton.

Isolation of Phage Clones Expressing Fab Fragments with Specificity for the Major Timothy Grass Pollen Allergen Phl p 5 by Panning-
ELISA plates (Costar 3690, Cambridge, MA) were coated with purified recombinant timothy grass pollen allergens rPhl p 1, rPhl p 2, and rPhl p 5, respectively (0.2 g/well). The wells were blocked with phosphatebuffered saline containing 3% (w/v) bovine serum albumin. Freshly prepared phage suspension (approximately 10 12 plaque-forming units) was added to each well and incubated at room temperature for 2 h. The phage were then removed, and the wells were washed with Tris-buffered saline containing 0.05% (v/v) Tween 20 once. Phage were eluted with 0.1 M glycine-HCl, pH 2.2, containing 1 mg/ml bovine serum albumin, and the eluent was neutralized with 2 M Tris. Freshly grown E. coli XL-1Blue were then infected with the eluted phage. An aliquot was used to determine the titer of infected E. coli. The culture was grown in SB medium containing 50 g/ml ampicillin and 10 g/ml tetracyclin. By infection with helper phage VCS M13, filamentous phage were produced for the next round of panning as described (Barbas et al., 1991). The panning was repeated four times. During the subsequent pannings, additional washing of the wells was done and individual clones were then analyzed for the production of Phl p 5-specific Fabs by ELISA.
Sequence Analysis of the cDNAs Coding for IgE Fds and Light Chains-Clones 5, 14, 28, and 31 were checked for the production of Phl p 5-specific Fabs by ELISA and for the correct insertion of cDNAs coding for heavy chain fragments and light chains by restriction analysis before sequencing. Plasmid DNA was prepared from recombi-nantE. coli XL-1 Blue using Qiagen tips (Hilden, Germany). Both DNA strands were sequenced using 35 S-dCTP (DuPont NEN) and a T7 polymerase sequencing kit (Pharmacia) by primer walking (Sanger et al., 1977). Sequencing primers were obtained from Pharmacia. The DNA and deduced amino acid sequence of the heavy and light chains were compared with the GenBank and Swissprot library.
Production of Soluble Recombinant Fab Fragments with Specificity for Phl p 5-For the production of soluble Fab fragments, DNA was isolated from several independent clones after the fifth round of panning. The plasmid DNA was digested with SpeI and NheI, recovered from a 1% agarose gel, self-ligated, and retransformed into E. coli XL-1 Blue. E. coli containing the correctly religated plasmid were then used to produce soluble Fab fragments. In brief, single colonies were inoculated into SB medium containing 20 mM MgCl 2 and 50 g/ml carbenicillin. The cultures were grown at 37°C for 6 h and then induced by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 4 mM. Induced E. coli were then grown at 30°C overnight, and cells were harvested by centrifugation at 3000 ϫ g for 10 min at 4°C. The E. coli supernatants were used for ELISA assays, immunoblotting, and for the affinity purification of Phl p 5-specific Fabs.
Purification of Phl p 5-specific IgE Fabs by Affinity to Purified Recombinant Phl p 5-2.5 mg of purified recombinant Phl p 5 was coupled to an AminoLink™ column (Pierce) according to the manufacturer's advice. Approximately 200 ml of E. coli supernant containing Phl p 5-specific Fabs were centrifuged at 20,000 ϫ g and subsequently filtered through folded filters (Macherey-Nagel, Dü ren, Germany) to remove debris from the solution. The supernatants were applied to the column at 4°C, and the column was then washed extensively with phosphatebuffered saline until no protein could be detected by photometry at 280 nm in the wash fractions. Bound Phl p 5-specific Fabs were eluted with 100 mM glycine-HCl, pH 2.7, and neutralized in 3 M Tris, pH 9.
Immunoblotting: IgE Competition Studies-Natural grass pollen extracts were prepared from rye grass (Lolium perenne), rye (Secale cereale), Kentucky Bluegrass (Poa pratensis), timothy grass (Phleum pratense), and birch (Betula verrucos) pollen purchased from Allergon, Vä linge, Sweden (Vrtala et al., 1993b). Recombinant timothy grass pollen allergens were purified as described (Vrtala et al., 1996). Approximately 100 g of natural pollen extract/cm of preparative 12% SDS-PAGE (Fling and Gregerson, 1986) were applied, whereas 1 g/cm gel of the purified recombinant allergens was separated. Proteins were electroblotted to nitrocellulose (Towbin et al., 1979), and nitrocellulose strips were incubated with 1:10 diluted sera from allergic patients followed by 125 I-labeled anti-human IgE antibodies (Pharmacia). For IgE inhibition studies, nitrocellulose-blotted recombinant Phl p 5 was preincubated with the Phl p 5-specific Fabs. Fab preparations with specificity for a different allergen (Bet v 1) were included as negative controls. IgE binding was performed as described above and visualized by autoradiography.
The detection of nitrocellulose-blotted or ELISA plate-coupled allergens with the IgE Fabs was done using an alkaline phosphatase coupled goat anti-human Fab antiserum (Pierce).

Determination of the Patients' IgE Reactivity Profile to Grass
Pollen Allergens-Serum from the donor who was used for the construction of the IgE combinatorial library was tested for the presence of grass pollen-specific IgE. Fig. 1 shows that the patient displayed IgE cross-reactivity to natural grass pollen extracts from rye grass, Kentucky Bluegrass, rye, and timothy grass. Most of the grass pollen-specific IgE bound to proteins of approximately 30 kDa, which represent group 1 and group 5 allergens Vrtala et al., 1993a). The testing with recombinant Phl p 1 (lane 5) and Phl p 5 (lane 7) indicated that most of this binding was due to the reactivity to group 5 allergens.

PCR Amplification of cDNAs Coding for IgE Fds and Light Chains from Peripheral Blood Lymphocytes of a Grass Pollen
Allergic Individual-Starting from RNA of peripheral blood lymphocytes of the grass pollen allergic donor, cDNAs coding for IgE Fds could be amplified. Fig. 2 shows the successful PCR amplification of IgE Fds using primers specific for different V H families (V H 1-V H 6) and a primer located in the first constant region of human IgE. RNA was isolated at different times of the year during 2 years, and it is noteworthy that the efficacy of the PCR amplification was best during the grass pollen season. Since the patient did not receive hyposensitization therapy, it was assumed that the increased IgE production during the grass pollen season was mostly due to stimulation by allergen contact. The obtained PCR product therefore may contain a high proportion of allergen-specific IgE Fds. The PCR products were confirmed to represent IgE-Fd encoding fragments by differential hybridization using synthetic oligonucleotides specific for IgE and IgG as described (Steinberger et al., 1995).
Construction and Characterization of an IgE Combinatorial Library from a Grass Pollen Allergic Patient-Subcloning of light chain cDNAs into plasmid pComb3H yielded a plasmid library consisting of 3 ϫ 10 7 primary clones. The subsequent ligation of the IgE-Fd cDNAs led to the construction of a combinatorial library of approximately 5 ϫ 10 7 primary clones which had contained heavy and light chain cDNAs. Fig. 3 shows the restriction analysis of 20 randomly isolated clones of the IgE combinatorial library. Only seven clones did not contain both cDNAs coding for the heavy chain fragment and light chain, indicating that 65% of the clones had combined correctly both cDNAs.
Isolation and Characterization of Recombinant Human IgE Fabs with Specificity for the Major Timothy Grass Pollen Al-lergen Phl p 5-Filamentous phage expressing IgE Fabs on their surface were panned five times using ELISA plate-immobilized recombinant Phl p 5, Phl p 1, Phl p 2, and bovine serum albumin as a control. After the last round of panning, clones were converted by restriction with SpeI and NheI to produce soluble Fabs. Supernatants from 20 clones were then tested by ELISA for the presence of Phl p 5-specific Fabs, among which four Phl p 5-specific clones designated clone 5, 14, 28, and 31 were characterized in detail. The specificity of the supernatants for Phl p 5 was confirmed by testing them for binding to recombinant Phl p 1, Phl p 2, Phl p 5, timothy grass pollen extract, and birch pollen extract. Supernatants from these clones bound to recombinant Phl p 5 and timothy grass pollen extract but not to any of the other proteins tested. No differences in the binding to native ELISA plate-immobilized, nitrocellulose-blotted Phl p 5 were observed.
Both strands of the DNA sequences coding for the Fd fragments and light chains of the clones were determined according to Sanger by primer walking, and the amino acid sequence was deduced. Fig. 4 shows the DNA and deduced amino acid sequence of the IgE heavy chain fragment, which was utilized by all four clones. It is noteworthy that identical heavy chain fragments were obtained by using two different PCR primers for the c⑀ constant region, indicating a positive selection for these particular IgE Fds during the panning process. The parts of the C⑀ domain were completely identical with known human IgE sequences (Flanagan and Rabbitts, 1982;Kurokawa et al., 1983). A molecular mass of 24.2 kDa could be predicted for the Fd-fragments of clones 5 and 28, whereas a molecular mass of 22.5 kDa could be deduced for clones 14 and 31, which were generated by a constant region primer located closer to the variable region (Fig. 4).
As can be seen in Fig. 5 (A and B), different light chains were used by the four clones. Most of the differences in the nucleotide sequences of the framework regions were silent. Regarding the CDRs of the four light chains, most differences were found in the CDR3 and CDR1. In conclusion, the panning procedure with recombinant Phl p 5 had enriched different IgE Fabs, which used identical heavy chain fragments that had combined with different light chains.
Recombinant Human IgE Fabs Specific for Phl p 5 Crossreact with Group 5 Allergens from Three Different Grass Species-More than 80% of grass pollen allergic patients display IgE cross-reactivity to group 5 allergens (Vrtala et al., 1993a). cDNAs coding for functional recombinant group 5 allergens were isolated from these species (Singh et al., 1991;Silvanovic et al., 1991;Vrtala et al., 1993a) and shown to be highly homologous. In order to investigate whether the recombinant human IgE Fabs that were isolated by panning to recombinant Phl p 5 cross-react with natural group 5 allergens from different grass species, grass pollen extracts from rye grass, Kentucky Bluegrass, rye, and timothy grass were probed. Fig. 6 shows that supernatants containing soluble Fabs from clones 5 and 28 reacted with nitrocellulose blotted rye grass (L. perenne), Kentucky Bluegrass (P. pratensis), rye (S. cereale), and timothy grass (P. pratense) pollen extract at approximately 30 kDa, which corresponds to the molecular mass of group 5 allergens. Additional weak binding to a 17-20 kDa component was observed in Kentucky Bluegrass and rye. Group 6 allergens, which share a high degree of sequence homology with group 5 allergens, were described to migrate at that molecular mass; however, the question whether group 6 allergens represent cleavage products of group 5 allergens or cross-reactive allergens is not yet clear (Matthiesen et al., 1993).
No reactivity of the Phl p 5-specific IgE Fabs was observed with birch pollen extract, which does not contain group 5 allergens. A recombinant IgG Fab (Co) with specificity for the major birch pollen allergen Bet v 1 was included as control and showed no reactivity with grass pollen extracts, whereas it bound to Bet v 1 at 17 kDa in birch pollen extract.
Recombinant human IgE Fabs, which were isolated by panning to purified recombinant Phl p 5, cross-reacted with natural group 5 allergens from different grass species as is known for natural IgE antibodies from grass pollen allergic patients.

Purification of Soluble Recombinant IgE Fabs Specific for
Phl p 5-For the purification of soluble IgE Fabs with specificity for Phl p 5, we used a single-step affinity purification procedure starting from E. coli supernatants that contained soluble Fabs. Purified recombinant Phl p 5 was coupled to an AminoLink™ column (Pierce), and E. coli supernatants containing soluble Phl p 5-specific Fabs were applied to the column. Fig. 7A shows a Coomassie Blue-stained SDS-PAGE of purified Fab preparations from clone 5 and 31 under reducing conditions. Heavy and light chains comigrated at approximately 25 kDa, and two weak bands were observed at approximately 15 and 16 kDa. The slight difference in the molecular masses of clones 5 and 31 is due to the fact that the cDNAs coding for the Fds were generated with different PCR primers. The Western blot in Fig. 7B containing the purified recombinant Fabs was probed with a goat anti-human Fab antiserum. The supernatant fraction contained low amounts of Fab, and no Fab was detected in the column wash fraction. In the elution fractions, a prominent 25-kDa band and bands at 15 and 16 kDa were bound by the anti-Fab antiserum, indicating that the lower molecular mass components observed in the Coomassie gel in Fig. 7A represented cleavage products of the Fab. The purified IgE Fab preparation contained approximately 0.1 mg/ml purified Fab and could be used to detect nitrocellulose FIG. 4. cDNA and deduced amino acid sequence of the heavy chain fragment of the Fabs with specificity for the major timothy grass pollen allergens Phl p 5. The cDNA and deduced amino acid sequence corresponding to the C⑀1 and C⑀2 portion, the framework regions (FR), and the complementarity determining regions (CDR1-CDR3) are indicated. The SpeI and XhoI sites are printed in italics, and the regions corresponding to the constant region primers are underlined. The cDNA sequences of the heavy chain fragments from clones 5, 14, 28, and 31 were found to be identical, although they were generated with two different primers. Recombinant IgE Fabs Specific for Phl p 5 Compete with Allergic Patients IgE Binding-Using competition experiments, it was investigated whether recombinant Phl p 5-specific IgE Fabs might be able to compete with grass pollen allergic patient IgE antibodies. Pairs of nitrocellulose strips containing equal amounts of blotted recombinant Phl p 5 were preincubated with supernatants from two Phl p 5-specific Fab clones (Fig. 8, 5 and 31) or with supernatants from E. coli expressing Bet v 1-specific Fabs (Fig. 8, Co). The nitrocellulose strips were then incubated with serum IgE from two patients (A and B), and bound IgE was detected. Preincubation of Phl p 5 with the Phl p 5-specific IgE Fabs led to a weak but clearly visible reduction of IgE binding compared to the control. The weak competition was not unexpected in view of the fact that Phl p 5 harbors multiple IgE epitopes (Bufe et al., 1994).

DISCUSSION
The cross-linking of effector cell-bound IgE antibodies by allergens has been recognized as the key event leading to Type I allergic reactions. Although IgE antibodies are present at extremely low levels in serum (10 -400 ng/ml), the release of mediators triggered by the cross-linking event causes severe allergic reactions (rhinitis, conjunctivitis, allergic asthma, and anaphylaxis). For this reason immunoglobulin E has been characterized extensively by protein chemical and molecular biological techniques (Terry et al., 1970;Kochwa et al., 1971;Bennich et al., 1973;Flanagan and Rabbitts, 1982;Kurokawa et al., 1983;Seno et al., 1983). Whereas considerable progress was achieved regarding the characterization of the constant regions of IgE, in particular the binding site for the high affinity receptor (Helm et al., 1988;Nissim and Eshhar, 1992;Presta et al., 1994), nothing was known about the V regions of IgE antibodies with specifities for allergens. In the present study we have used the filamentous phage display system to isolate human allergen-specific recombinant IgE Fab fragments. To achieve this goal, a highly sensitive PCR technique was established to allow the amplification of IgE Fd from the peripheral blood of allergic patients (Steinberger et al., 1995). An IgE combinatorial library was constructed in the pComb3H plasmid, starting from peripheral blood lymphocytes from a grass pollen allergic patient. Using purified recombinant timothy grass pollen allergens for the panning procedure, human IgE Fabs with specificity for the major timothy grass pollen aller-gen, Phl p 5 (Vrtala et al., 1993a), could be isolated. Phl p 5 was used as a model allergen, because it represents a major allergen for more than 80% of grass pollen allergic individuals and cross-reacts with group 5 allergens from most grass species (van Ree et al., 1992). Another reason for selecting Phl p 5 was the fact that a high percentage of grass pollen-specific IgE antibodies are directed against this allergen in most patients (Vrtala et al., 1993a(Vrtala et al., , 1996. Serum from the patient who was used for the construction of the combinatorial library contained high levels of Phl p 5-specific IgE compared to Phl p 1-specific IgE (Fig. 1). In fact, many more phage clones with specificity for Phl p 5 could be recovered from the combinatorial library after five rounds of panning than clones that reacted with Phl p 1. 2 It appeared hence that the repertoire represented in the IgE combinatorial library closely reflected the natural IgE antibody 2 P. Steinberger, D. Kraft, and R. Valenta, unpublished data.
FIG. 6. Cross-reactivity of Phl p 5-specific recombinant IgE Fabs with natural group 5 allergens from different grasses. Two Phl p 5-specific IgE Fabs (clone 5 and 28) and a recombinant IgG Fab with specificity for the major birch pollen allergen Bet v 1 (Co, negative control) were tested for reactivity with nitrocellulose-blotted grass pollen extracts (rye grass, L. perenne; Kentucky Bluegrass, P. pratense; rye, S. cereale; timothy grass, P. pratense) and birch pollen extract (B. verrucosa). Fig. 7A shows purified recombinant Phl p 5-specific Fabs from clone 5 and 31 separated under reducing conditions. A gel containing samples of the total E. coli supernatant: Fabs (SN), the wash fraction (lane 1), and the elution fractions (lanes 2-5) of the Phl p 5 affinity column, was blotted onto nitrocellulose and Fabs were detected with a goat anti-human Fab antiserum (Fig. 7B).

FIG. 7. Purification of a recombinant Phl p 5-specific IgE Fab by affinity chromatography to immobilized recombinant Phl p 5. The Coomassie Blue-stained SDS-PAGE in
FIG. 8. Influence of the preincubation of recombinant Phl p 5 with a Phl p 5-specific IgE Fab on the IgE binding of grass pollen allergic patients. Nitrocellulose strips containing blotted recombinant Phl p 5 (duplicates) were preincubated with Phl p 5-specific IgE Fabs (clones 5 and 31) or with Bet v 1 IgG Fabs (negative control, Co) before serum IgE from two grass pollen allergic patients (A and B) was applied. Bound IgE was detected with 125 I-labeled anti-human IgE monoclonal antibodies.
response of the patient.
The sequence analysis of four independent Phl p 5-specific IgE Fabs revealed that all four clones used the same type of heavy chain fragment originating from different PCR reactions, which had recombined with different kappa light chains. The finding that different PCR products of the same IgE Fd and similar light chains had combined to form the Fabs that were selected by the panning procedure indicated that the recombinant Fabs might closely reflect the structure of Phl p 5-specific natural IgE antibodies. The fact that, during 2 years of continuous blood sampling and PCR amplifications of IgE Fds, PCR products were most efficiently obtained during the grass pollen season indicated that the IgE Fabs represented in the combinatorial library most likely were produced in response to repeated allergen stimulation.
Like the natural IgE antibodies, the Phl p 5-specific IgE Fabs cross-reacted with group 5 allergens from rye grass, Kentucky Bluegrass, and rye. Using immobilized recombinant Phl p 5, soluble human IgE Fabs could be purified to homogeneity up to milligram amounts. Due to a lack of the c⑀2-c⑀4 domain, the recombinant IgE Fabs were ineffective to trigger basophil degranulation in combination with purified Phl p 5 (data not shown). In addition it could be shown that the IgE Fabs were able to compete with the IgE binding to Phl p 5 using sera from grass pollen allergic individuals. The inhibitory effect was, however, very weak, which was not surprising in view of the fact that Phl p 5 bears a number of different IgE epitopes (Bufe et al., 1994) and on the basis that the Phl p 5-specific IgEresponse in patients is polyclonal. Despite this, we believe that the use of the combinatorial approach to define recombinant Fabs with specificity for major allergens may have possible therapeutic implications, as were discussed for antibodies directed against tumor or viral antigens (Waldmann, 1991). In the case of the major birch pollen allergen, Bet v 1, human and mouse monoclonal antibodies could be defined that strongly inhibited the binding of patients IgE to the allergen 3,4 so that a local application of such blocking antibodies for a passive therapy in the allergic effector organs (nose, eyes, and lung) might be envisaged . Using the cDNAs coding for allergen-specific Fabs, procedures such as in vitro affinity maturation might be used to "improve" the antibodies for such therapeutic applications (Barbas et al., 1994). Apart from the possible therapeutic implications, we believe that the combinatorial approach will allow the study of the interaction of human IgE antibodies and allergens at the molecular level by using purified recombinant allergens and human IgE Fabs for structural analysis (x-ray crystallography and NMR).