Isolation of a high‐affinity Bet v 1‐specific IgG‐derived ScFv from a subject vaccinated with hypoallergenic Bet v 1 fragments

Abstract Background Recombinant hypoallergenic allergen derivatives have been used in clinical immunotherapy studies, and clinical efficacy seems to be related to the induction of blocking IgG antibodies recognizing the wild‐type allergens. However, so far no treatment‐induced IgG antibodies have been characterized. Objective To clone, express, and characterize IgG antibodies induced by vaccination with two hypoallergenic recombinant fragments of the major birch pollen allergen, Bet v 1 in a nonallergic subject. Methods A phage‐displayed combinatorial single‐chain fragment (ScFv) library was constructed from blood of the immunized subject and screened for Bet v 1‐reactive antibody fragments. ScFvs were tested for specificity and cross‐reactivity to native Bet v 1 and related pollen and food allergens, and epitope mapping was performed. Germline ancestor genes of the antibody were analyzed with the ImMunoGeneTics (IMGT) database. The affinity to Bet v 1 and cross‐reactive allergens was determined by surface plasmon resonance measurements. The ability to inhibit patients’ IgE binding to ELISA plate‐bound allergens and allergen‐induced basophil activation was assessed. Results A combinatorial ScFv library was obtained from the vaccinated donor after three injections with the Bet v 1 fragments. Despite being almost in germline configuration, ScFv (clone H3‐1) reacted with high affinity to native Bet v 1 and homologous allergens, inhibited allergic patients’ polyclonal IgE binding to Bet v 1, and partially suppressed allergen‐induced basophil activation. Conclusion Immunization with unfolded hypoallergenic allergen derivatives induces high‐affinity antibodies even in nonallergic subjects which recognize the folded wild‐type allergens and inhibit polyclonal IgE binding of allergic patients.


| INTRODUCTION
Allergen-specific immunotherapy (AIT) is an effective, disease-modifying, and long-lasting treatment for IgE-associated allergy. [1][2][3] AIT has been conducted originally by vaccination with natural allergen extracts containing allergens in their native structure. 4 To increase the safety of AIT, denatured allergen extracts termed allergoids have been introduced in which the native allergens have been chemically modified to lose their structure and conformational IgE epitopes. 5,6 Despite the loss of the native structure, allergoids seem to induce allergen-specific blocking IgG antibodies [7][8][9][10] which are thought to be responsible for clinical efficacy of AIT besides alterations of cellular and cytokine responses. 3 With the availability of recombinant allergens and DNA as well as peptide-based technologies, recombinant and synthetic allergen derivatives have been engineered for several important allergens which resemble many features of allergoids, in particular reduced IgE reactivity and reduced allergenic activity. 11,12 Several clinical studies performed with such recombinant allergen derivatives and synthetic allergen fragments lacking the folding of natural allergens indicate that the treatment is clinically effective and is associated with the induction of allergen-specific IgG antibodies which compete with allergic patients 0 IgE for binding to the natural, native allergens. [13][14][15][16] Although IgG antibodies induced with recombinant hypoallergenic allergen derivatives and allergen-derived peptides are raised against unfolded allergen derivatives, they seem to induce IgG antibodies in patients which recognize also the folded native allergens. For example, immunization with recombinant, unfolded fragments of the major birch pollen allergen Bet v 1 raised IgG antibodies against Bet v 1 peptides, 17 but serum from the immunized patients also inhibited binding of allergic patients IgE antibodies against conformational epitopes. In this context, an important question remains to be answered: Is the inhibition of allergic patients 0 IgE antibodies recognizing conformational epitopes on the native allergens due to a fraction of IgG antibodies which recognizes conformational epitopes or can the inhibition of IgE antibodies be achieved with IgG antibodies which are primarily induced against peptide epitopes which also recognized the native allergen? To investigate IgG antibodies induced with recombinant unfolded allergen derivatives in detail, we immunized nonallergic subjects with recombinant Bet v 1 fragments and constructed a combinatorial single-chain fragment library from an immunized subject who developed IgG antibodies recognizing the native allergen. We then isolated an IgG-derived Bet v 1-specific ScFv (H3-1) and analyzed the sequence of H3-1 in comparison with ancestor germline sequences, investigated the specificity, cross-reactivity, and affinity of H3-1. Our results demonstrated that IgG antibodies induced with recombinant, unfolded allergen derivatives simultaneously recognize peptide epitopes as well as epitopes on the folded native allergen and thus can act as blocking antibodies.

| Materials
Materials, sera, and reagents are described in the Data S1. IgG to Bet v 1 (2 lg/mL), Bet v 1 fragments (1 lg/mL), or peptides (1 lg/mL) was determined by ELISA. After Bet v 1, fragments or peptides were coated for 5 hours at room temperature to ELISA plates (Nunc Maxi-Sorp, Roskilde, Denmark), plates were blocked with PBST containing 2% BSA (wt/vol) overnight at 4°C. Serum samples diluted 1:100 were incubated overnight at 4°C, and specific IgG levels were detected with a rabbit anti-human IgG antiserum (Jackson Immuno) followed by a horseradish peroxidaselabeled donkey anti-rabbit antiserum (GE Healthcare, Little Chalfont, UK) diluted 1:10 000 and 1:2000, respectively, in PBST containing 0.5% BSA (wt/vol). Color reaction was read at 405 nm, and results are means of duplicates.

| Construction of a combinatorial phagedisplayed ScFv library from PBMCs of a subject immunized with rBet v 1 fragments
Peripheral blood mononuclear cells were isolated 10 days after the third injection when allergen-specific IgG had developed ( Figure 1) by Ficoll Paque (GE Healthcare) density gradient centrifugation for the construction of the combinatorial library (see Data S1).

| Isolation of a Bet v 1-specific ScFv
After library transformation in E. coli TG1, phages were rescued with M13KO7 helper phage and the phage library was subjected to four rounds of solid phase panning. For this purpose, ELISA wells (Nunc, Roskilde, Denmark) were coated with Bet v 1 (1 lg/ well in 1xPBS), blocked, and incubated with the phage library.
Wells were washed with 1xPBS + 0.05% (vol/vol) Tween 20, and bound phages were eluted with 50 lL 100 mmol/L glycine-HCl pH 2.2, neutralized with 3 lL 2 mol/L Tris-HCl pH 8 and used for reinfection of a new TG1 culture that served as a starting point for the next round of panning. To check for enrichment of Bet v 1specific phage pools, we added each eluted phage pool from each round of panning in duplicates to ELISA plate-bound Bet v 1 and BSA as control (10 lg/mL) and detected bound phages with horseradish peroxidase-labeled anti-M13 antiserum (GE Healthcare).
Color reactions were read at 405 nm. Results represent means of duplicate determinations.
After the 3rd and 4th round of panning, an aliquot of transduced TG1 was plated to SOBAG plates for screening purposes (Amersham Biosciences). To this aim, single-phage clones were rescued, incubated with ELISA plate-bound Bet v 1 (10 lg/mL), and detected with an anti-M13 antiserum (GE Healthcare) as described for the panning.
Obtained sequences were aligned with human germline gene sequences from the IMGT database using V-Quest software. 19 Ancestor germline genes were determined, and the positions of mutations were identified. Phagemid DNAs were transformed into the nonsuppressor E. coli strain HB2151 for expression of soluble ScFvs carrying a peptide E tag at the C-terminus. 20 Additionally, ScFvs carrying a C-terminal His tag were constructed as described. 21 Soluble Bet v 1-specific ScFvs were purified by affinity chromatography. 22 The biochemical and structural characterization of H3-1 is described in the Data S1.

| Multiple sequence alignment
The amino acid sequences of Bet v 1 (Swissprot accession number: P15494), Aln g 1 (P38948), Cor a 1 (Q08407), and Mal d 1 (Q9SYW3) that correspond to the Bet v 1-derived peptide 5 were aligned with the multiple sequence alignment tool ClustalW edited and visualized with GeneDoc.

| Inhibition ELISA
ELISA plates (Nunc) were coated with 1 lg/mL of the allergens (Bet were not included in the inhibition calculation. Percentage reductions of patients' IgE binding after pre-incubation with H3-1 compared to the control ScFv were calculated as follows: 100-(OD after pre-incubation with H3-1 * 100/OD after pre-incubation with control ScFv).
All determinations were performed in duplicates, and median values were calculated. Statistically significant differences of IgE binding after pre-incubation with H3-1 versus pre-incubation with control ScFv were determined by Mann-Whitney test using Prism 5.04 (GraphPad Software, La Jolla, CA, USA).

| Effect of H3-1 on Bet v 1-induced basophil activation
The effect of H3-1 on Bet v 1-induced basophil activation was measured by detecting CD203c upregulation as described 26 (Data S1).  Figure 1C) showed that the IgG response induced by the immunization was mainly directed against the Cterminal peptide 5 and peptide 3 which reportedly is part of a region on Bet v 1 containing binding sites for allergic patients IgE. 23 As the IgG response was mainly composed of Bet v 1-specific IgG 1 and Bet v 1-specific IgG 4 (data not shown), preselection for IgG was achieved with specific primers located in the hinge regions of IgG 1 or IgG 4 . VH and VΚ regions were amplified, randomly combined with DNA coding for a ((Gly 4 Ser) 3 ) linker, and cloned into the phagemid vector pCANTAB 5E, leading to a final library size of 9.9 9 10 7 independent clones. We analyzed 13 randomly picked clones to check the diversity of the library. The heavy and light chains of the analyzed clones revealed high differences in their CDR regions and in their V gene family distribution demonstrating that each clone represented an independent antibody (Data S1, Table S1), confirming the nonredundant nature of the library. The VH family distribution was dominated by members of the largest gene subgroup, the VH3 family, followed by VH1 > VH6 > VH2/ VH4 > VH5 (Data S1, Table S1). The VK family distribution was dominated by VK3 followed by VK1/4 > VK7 > VK2/5/6 (Table S1). Bet v 1 in their soluble form. According to sequence analysis, all these 15 clones were identical and therefore designated as one clone, that is, H3-1.

| Construction of a combinatorial phagedisplayed ScFv library from a subject immunized with rBet v 1 fragments and isolation of a Bet v 1-specific ScFv
The H3-1 heavy chain (H3-1_HV) showed 99.65% nucleotide sequence identity with the closest related germline ancestor V gene IGHV1-69*09 ( Figure 2A) and differed only in one nucleotide located in CDR2 and two in CDR3 resulting in two amino acid exchanges (Data S1, Table S2). The D gene of the H3-1_HV was derived from its germline progenitor IGHD2-15*01. The J gene showed 100% sequence identity to the germline IGHJ6*02 (Figure 2A).
The J gene of the light chain seemed to originate from IGKJ1*01 ( Figure 2B).

| Allergen specificity, cross-reactivity, and epitope specificity of H3-1
The H3-1 ScFv was purified by affinity chromatography and analyzed by SDS-PAGE. Coomassie staining revealed a single 28 kDa band without any degradation products ( Figure S2A). The CD spectrum of H3-1 showed a secondary structure dominated by ß-sheet and ß-turn/random coils and no trace of an a-helical signal (Figure S2B).
Next, we studied the cross-reactivity of H3-1 with Bet v 1related allergens. Figure 4A shows that H3-1 reacted best with rBet v 1 but also with the cross-reactive allergens from alder (Aln g 1), hazel (Cor a 1), and apple (Mal d 1), albeit with lower intensity. No reactivity to the control protein BSA was observed ( Figure 4A).

| High-affinity binding of H3-1 to Bet v 1 and homologous allergens
Surface plasmon resonance measurements were conducted for H3-1 Recorded (colored lines) and calculated (black lines) curves, which represent fitting to a 1:1 binding model, were superimposed. Dissociation constants (K D ) as well as association and dissociation rate constants (k a , k d ) are displayed

IgE binding and basophil activation
Next, we tested whether H3-1 is able to inhibit birch pollen-allergic patients' IgE binding to ELISA plate-bound Bet v 1 or to cross-reac- A varying effect of H3-1 on Bet v 1-induced basophil activation was observed ( Figure 5A-D). In three of the four patients examined, H3-1 inhibited Bet v 1-induced in vitro basophil activation.
T A B L E 1 Inhibition of allergic patients 0 IgE binding to Bet v 1, Aln g 1, Cor a 1, and Mal d 1 after pre-incubation with H3-1 compared to a control ScFv

| DISCUSSION
Our study is the first to report the molecular characterization of an allergen-specific IgG antibody induced by immunization with recom- AIT with natural, native allergen-containing extracts. 28,29 In contrast to H3-1, these antibodies showed numerous somatic mutations and signs of affinity maturation.
Nevertheless, we found that H3-1 bound with high affinity to native and folded Bet v 1 as well as to Bet v 1-related allergens.
This finding is remarkable because high-affinity binding was so far considered a feature of highly mutated antibodies. 30