Easy assessment of the avidity of polyclonal allergen‐specific serum antibodies

Allergen‐specific IgE‐blocking IgG antibodies contribute to successful allergen immunotherapy (AIT), however, not much is known about their affinity. Since affinity measurements of polyclonal antibodies in serum are technically challenging we evaluated the applicability of acidic disruption of antibody‐allergen complexes by a modified ELISA protocol with monoclonal antibodies (mAbs) specific for the relevant major allergens Betv1 and Mald1. Then, AIT‐induced blocking and non‐blocking Mald1‐specific antibodies in sera from individuals with or without reduced apple allergy were compared.


| INTRODUC TI ON
Allergen immunotherapy (AIT) is effective in treating respiratory IgE-mediated allergy, e.g.moderate to severe allergic rhinitis to tree pollen. 1 It has been shown that clinical success is accompanied by a rise of IgE-blocking allergen-specific IgG antibodies. 2 Using patient-derived human monoclonal antibodies (mAbs) we recently showed that IgE-blocking activity enhances with antibody affinity. 3To better understand the mechanisms of successful AIT it would be interesting to assess the avidity, i.e. the net binding force of all specific antibodies in serum samples, in the course of therapy.
Surface plasmon resonance (SPR) and bio-layer interferometry (BLI) are optical phenomena that are utilized for studying antibody-antigen binding in real time. 4,5Typically, these methods include the immobilization of antibody to a stationary phase and subsequent exposure to different concentrations of antigen.The individual cycles of antibody-antigen complex formation need to be separated by regeneration, i.e. the complete removal of bound antigen from the antibody surface.SPR and BLI are well established to assess the affinity of mAbs, however, the analysis of polyclonal antibodies in serum poses a methodological challenge.This applies particularly to allergen-specific antibodies because their concentration among all antibodies is extremely low, which hinders their immobilization.7][8] However, the latter approach requires the use of purified antibodies as serum contains a plethora of proteins that could mask allergen-specific binding events.The purification of antibodies may reduce their functionality and often results in low yields.
Besides, not all allergens remain fully functional after chemical immobilization or regeneration.In summary, avidity measurements of allergen-specific serum antibodies by these techniques are possible, but involve extensive assay optimization together with expertise in specialized instruments and are thus complex to implement.
We recently estimated the collective binding force of Bet v 1-specific antibodies in the course of AIT with birch pollen using a modified ELISA involving the acidic disruption of their immune complexes with allergen. 8Here, we sought to confirm the accuracy of this approach by using a panel of specific mAbs and compared the results with affinity measurements by SPR.We excluded possible effects of serum proteins and finally applied the modified ELISA to compare the avidity of Mal d 1-specific protective and non-protective IgG1 antibodies in sera from individuals after sublingual administration (SLIT) of recombinant Mal d 1 and Bet v 1, respectively. 9

| Recombinant allergens and allergen-specific mAbs
Recombinant Bet v 1 and Mal d 1 were produced and characterized as described before. 9The Bet v 1-specific murine mAbs BIP1, BIP4 10 and mP16, 11 as well as the Mal d 1-specific human mAbs K1.1 and K1.1-1 were generated and characterized in an earlier publication. 3

G R A P H I C A L A B S T R A C T
Complexes of antibodies with allergens were disrupted by acidic buffers in ELISA.Residual antibodies were expressed as the percentage of antibodies bound at neutral conditions set to 100%.Dissociation curves were created and the avidity index (the pH value that dissociated 50% of antibodies) was calculated.The modified ELISA can be used to estimate the net-binding force of allergen-specific polyclonal IgG antibodies in serum.

Key messages
• Complexes of monoclonal antibodies were dissociated with acidic buffers in ELISA.
• Avidity indexes (pH value that dissociated 50% of antibodies) agreed with SPR affinity measurements.
• Acidic dissociation is applicable to estimate the netbinding force of allergen-specific polyclonal IgG antibodies in serum.

| Human serum samples
Three serum samples from birch pollen-allergic individuals with rhinoconjunctivitis in spring and Bet v 1-specific IgE levels of >0.35 kU A /L (ImmunoCAP, Thermo Fisher Scientific, Waltham, MA, USA) and one serum of a non-allergic individual were used with informed consent and ethical clearance by the local ethics committee (EK1344/2018).Furthermore, samples from individuals who received a daily sublingual dose of 25 μg of Bet v 1 (B1-B9) or of 25 μg of Mal d 1 (M1-M8) for 16 weeks were included. 12Mal d 1-specific IgE, IgG1 and IgG4 levels in these samples are shown in Table S1.

| ELISA protocol to test the pH stability of allergens
Allergens were coated on microplates (Nunc MaxiSorp, Thermo Fisher Scientific) overnight (ON) at 4°C at a concentration of 1 μg/ mL in carbonate buffer, pH 9.6.To test their stability in acidic conditions, McIlvaine 13 phosphate-citrate buffers with pH values of 7.4, 5.4 and 3.4 were added for 2 h at room temperature (RT).
After washing with PBS/0.05%Tween 20 (PBS-T), plates were saturated with 1% human serum albumin (HSA, Merck Millipore, Burlington, MA, USA) in PBS-T for 6 h at RT. Sera of birch pollenallergic individuals were added ON and bound IgE and IgG antibodies were detected with an alkaline phosphatase-labelled mouse anti-human IgE (BD Pharmingen, San Jose, CA, USA) or a horseradish peroxidase (HRP)-labelled goat anti-human IgG (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA), respectively.mAbs were diluted in PBS-T containing 0.5% HSA.Bound murine mAbs were detected with a HRP-labelled goat anti-mouse IgG antibody (Jackson ImmunoResearch Laboratories Inc.) and bound human mAbs with a mouse anti-human IgG1 (Thermo Fisher Scientific) followed by a HRP-labelled anti-mouse IgG antibody from sheep (Cytiva, Pittsburgh, PA, USA).IgG1 antibodies in serum samples were detected using a HRP-coupled mouse anti-human IgG1 (Southern Biotech, Birmingham, AL, USA).Optical density (OD) values were measured on a Tecan infinite F50 reader (Tecan, Männedorf, Switzerland).

| Modified ELISA for avidity measurement
Microplates were coated, washed twice, and saturated as indicated were normalized to 100% and the percentage of remaining antibodies was individually calculated.All measurements with mAbs were done in duplicate and repeated for 2-6 times in independent setups.The avidity index (AI), defined as the logEC50 value of the averaged dissociation curve, was calculated using the command 'log(agonist) versus normalized response -Variable slope' in GraphPad Prism version 9.5.0 (San Diego, CA, USA).

| SPR measurements
All SPR measurements were performed on a Biacore™ 3000 and analysed using the BIAevaluation software 4.1 (Cytiva, Uppsala, Sweden).For epitope binning experiments, the first mAb was immobilized on a CM5 SPR chip using standard amine coupling chemistry.
Next, Bet v 1 was injected until saturation followed by an injection of the second mAb.The order of mAbs was reversed for all studied antibody pairs.A second injection of the first mAb served as negative control.
The affinities of antibody-allergen interactions were measured at 25°C using multicycle kinetics assays.To this end, an anti-mouse IgG antibody was immobilized to flow cell (Fc) 1 and 2 on a CM5 sensor chip (Mouse Antibody Capture Kit, Cytiva).Next, mAbs diluted in HBS-EP running buffer (Cytiva) were captured in Fc2 while Fc1 served as a reference.Bet v 1 diluted in running buffer to 64-0.5 nM in 1:2 steps was injected over both Fcs for 6 min at a flow rate of 30 μL/min.Dissociation was monitored for 15 min.Afterwards, the chip was regenerated as specified in the kit.The binding kinetics were evaluated using referenced and blank-subtracted curves utilizing the 1:1 Langmuir binding model.

| pH stability of allergens
To test their pH stability, we exposed plate-bound Bet v 1 and Mal d 1 to buffers of pH 7.4, 5.4 and 3.4 for 2 h followed by incubation with sera from three birch pollen-allergic individuals and the mAbs BIP1, BIP4, mP16 and K1.1.A comparable percentage of bound mAbs was detected at all pH values in a series of 2-6 independent experiments (Figure 1).Together, the use of polyclonal IgE (Figure 1A), IgG (Figure 1B) and mAbs (Figure 1C) revealed that the selected conditions neither removed the allergens from the solid phase nor considerably harmed their antibody-binding property.

| Affinity measurement of mAbs by SPR
We proceeded with SPR analyses of the mAbs.Epitope binning revealed that BIP1 and mP16 inhibit each other from binding to Bet v 1, indicating that they recognize overlapping epitopes (Figure 2A).Still, their epitopes cannot be identical as BIP1 and BIP4 competed for Bet v 1-binding whereas mP16 and BIP4 did not (Figure 2A).The latter result indicates that mP16 and BIP4 bind distant surface areas of the major birch pollen allergen.The dissociation constants (K D ) were 0.75 nM for BIP1, 2.65 nM for BIP4 and 4.77 nM for mP16 (Figure 2B and Table S2).Next, the binding force of the Bet v 1-specific mAbs was tested repeatedly by challenging their complexes with plate-bound allergen with increasingly acidic buffers with pH values ranging from 6.4-3.4.Residual mAbs were expressed as percentage of those binding at pH 7.4 set to 100% (Figure 2C).The AIs, i.e. the pH value removing 50% of the antibodies, were calculated from the averaged dissociation curves and were 4.54, 4.67 and 4.91 for BIP1, BIP4 and mP16, respectively.

| Avidity analysis in the presence of serum proteins
BIP1, mP16, K1.1 and K1.1-1 were diluted in the serum of a nonallergic individual devoid of detectable allergen-specific IgG antibodies (data not shown).Repeated-modified ELISA experiments revealed very similar dissociation profiles of all mAbs in the absence or presence of serum (Figure 3).

| Avidity comparison of allergen-specific serum antibodies
Next, we sought to compare the collective binding force of polyclonal Mal d 1-specific IgG1 antibodies in sera collected from individuals after 16 weeks of SLIT with either recombinant Mal d 1 or Bet v 1. 9,12 The samples differed in functional IgE-blocking to Mal

| DISCUSS ION
The purpose of this study was to develop a straightforward approach to assess the avidity of allergen-specific antibodies in human sera.We utilized monoclonal and polyclonal antibodies recognizing two important allergens.The major birch pollen allergen Bet v 1 is the prime example of a large group of related respiratory allergens of high clinical relevance. 14The homologous major allergen from apple, Mal d 1, represents a frequent cause of birch pollen-related food allergy. 15Both allergens rapidly lose their IgE-binding activity when their 3-dimensional structure is altered. 16,17Other modified ELISA protocols to assess the avidity of polyclonal antibodies in serum samples involve chaotropic agents for the disruption of antibody-antigen complexes, such as thiocyanate or urea, 18 which can destabilize the native state of proteins by weakening the hydrophobic effect. 19Being aware of the limited stability of Mal d 1, 20 we chose milder chemical conditions for immune complex disruption by applying buffers of gradually decreasing pH values.Because acidic solutions may also denature proteins, 21   registered.Independently repeated experiments for each mAb delivered an inter-assay variability of <4% CV (Table S3).However, we noticed that the overall ranking of all mAbs according to K D values (BIP1 > BIP4 > mP16 > K1.1.1 > K1.1) was discrepant from their order according to AIs (BIP1 > K1.1.1 > BIP4 > mP16 > K1.1).We refer this observation to differences in the coating behaviour of Bet v 1 and Mal d 1.Both allergens were used at the same concentration.
However, approximately 20% of the recombinant apple allergen are existent in dimers whereas Bet v 1 is a largely monomeric protein. 12nsequently, the ELISA protocol should only be applied to compare the binding strength of antibodies to the same but not to different allergens.
After confirming that the measurements of four mAbs with In view of the positive association of affinity and IgE-blocking activity 3 together with the requirement of several mAbs specific for different epitopes to successfully reduce allergic symptoms 7,22,23 we hypothesize that the repertoire of highly affine Mal d 1-specific antibodies induced by Bet v 1-SLIT might be more restricted than the repertoire induced by Mal d 1-SLIT.This may represent one reason for the limited curative success of AIT with birch pollen on the associated apple allergy in some individuals.
Here, we focused on AIT-induced allergen-specific IgG1 antibodies as this isotype was associated with clinical improvement in our study groups, thereby allowing the first avidity comparison of protective and non-protective antibodies specific for the same allergen but induced by different treatments.Still, the assay is also applicable for avidity estimation of allergen-specific IgG4. 8Allergen-specific IgG4 levels are low early in AIT but increase with the duration of treatment and may interfere with IgG1.Nevertheless, we previously observed a relatively constant avidity of IgG1 antibodies despite a constantly increasing ratio of IgG4/IgG1 antibodies in samples collected after 12, 24 and 36 months of AIT. 8 Certainly, a positive association of IgE-blocking activity, avidity and clinical efficacy requires confirmation in larger cohorts of AITtreated individuals.For this purpose, the here introduced approach will be practical.Without the need for elaborate sample preparation and specialized instrumentation, the acidic dissociation ELISA may be implemented in any laboratory participating of multicenter AIT studies, which facilitates data acquisition of large numbers of samples.We do not consider this assay as a substitute for affinity measurements.This is exemplified by the mAbs under investigation as additional buffers in the pH range of 5.4-4.0 should have been included for a more precise calculation of their AIs.For polyclonal antibodies in serum samples AIs relativized differences of low and high affinity antibodies observed in the dissociation curves and estimated their collective binding force.We consider the here introduced assay as appropriate to facilitate the investigation of the role of antibody affinity in AIT.
above.mAbs and serum samples were diluted to reach OD values of approximately 1.0 as determined in preceding ELISAs.After five washings, the immune complexes were incubated with McIlvaine buffers with pH values of 7.4, 6.4, 6.0, 5.4, 5.0, 4.4, 4.0 and 3.4 for 2 h at RT.Then, plates were washed five times and bound antibodies were detected as indicated above.The OD values from buffer controls were subtracted.The OD values after incubation at pH 7.4 Since low K D and AI values indicate high binding force, either technique identified BIP1 as the antibody of highest and mP16 as the antibody of lowest affinity.In addition, we utilized the two Mal d 1-specific human mAbs K1.1 and K1.1-1.The latter has a K D of 8.66 nM and represents an affinity-enhanced descendant of K1.1, which has a K D of 19.2 nM. 3 Acid dissociation of their binding to plate-bound Mal d 1 revealed an AI of 4.53 for K1.1-1 and 4.94 for K1.1 and agreed with affinity measurements by SPR (Figure 2D).

d 1 following
Mal d 1-SLIT but not following Bet v 1-SLIT.The individual dissociation curves and calculated AIs are shown in Figure S1.The mean dissociation profiles of all individuals in either group were compared and indicated significantly more residual antibodies in the Mal d 1-SLIT-treated group at pH 6.4 and 6.0 (Figure 4A).At lower pH values, the curves of the two SLIT groups converged.The comparison of the individual AIs shown in Figure 4B further suggested a higher collective binding force of IgG1 antibodies induced by Mal d 1-SLIT.

1
pH stability of allergens.ELISA plate-bound Bet v 1 and Mal d 1 were exposed to indicated pH values before incubation with specific polyclonal IgE (A), IgG (B) and mAbs (C).The percentage of bound antibodies in relation to pH 7.4 set to 100% is shown as mean ± SD of 2-6 repeated experiments.| 283 STROBL et al.
we first controlled the recognition of both allergens by polyclonal IgE and IgG antibodies recognizing several epitopes and mAbs specific for different single epitopes after exposure to pH 5.4 and 3.4.No obvious differences to pH 7.4 were found.A 2-h incubation at pH 3.4 completely removed all mAbs from the allergens including the highly affine BIP1.We therefore selected buffers of pH values ranging from 6.4-3.4 for acidic dissociation of antibodies from either allergen.The resulting AIs of mAbs agreed with K D measurements obtained by SPR.Notably, relatively small differences of affinity, as observed for the Bet v 1-specific mAbs BIP4 and mP16 and the Mal d 1-specific mAbs K1.1 and K1.1-1, were F I G U R E 3 Avidity of mAbs in PBS and serum.The percentage of bound mAbs diluted in PBS (black circles) or in serum (white circles) after exposure to indicated pH values in relation to pH 7.4 set to 100% is shown.The mean values ± SD of 2-6 experiments are shown.F I G U R E 4 Avidity of AIT-induced Mal d 1-specific IgG1 after the sublingual administration of Bet v 1 or Mal d 1. Sera were collected after SLIT with recombinant Mal d 1 (n = 8) or Bet v 1 (n = 9).The mean + SD of the percentage of Mal d 1-bound IgG1 antibodies after exposure to indicated pH values in relation to pH 7.4 set to 100% is shown.Mann-Whitney U-test, *p ≤ 0.05, **p ≤ 0.01 (A).Corresponding AIs are shown.Lines indicate the median values, ns, not significant (B).

F I G U R E 2
Epitope binning and affinity measurement of mAbs.SPR sensorgrams; immobilized mAbs are shown in italics, arrows indicate the injection of Bet v 1 and mAbs (A); dissociation constants (K D ) were calculated after injection of different concentrations of Bet v 1 (grey lines) by the 1:1 Langmuir binding model (black lines) (B).ELISA; mAbs bound to Bet v 1 (C) and Mal d 1 (D) after exposure to indicated pH values expressed as percentage of pH 7.4 set to 100%, mean values ± SD of 2-6 experiments (black lines) and fitted curves (grey dashed lines) for calculation of the AI are shown.
different affinities were uninfluenced by the presence of human serum, we proceeded with the comparison of Mal d 1-specific serum IgG1 antibodies induced by two different AIT protocols.Individuals who had received a daily sublingual dose of 25 μg of recombinant Mal d 1 for 16 weeks displayed reduced allergic reactions to open food challenges with Mal d 1 and apple as well as IgE-blocking IgG1 antibodies preventing Mal d 1-mediated effector cell activation. 9,12The other group had received recombinant Bet v 1, showed no improvement of apple allergy and did not develop IgG1 antibodies protective against Mal d 1.We observed clear differences in the range of pH 6.4-5.0 which may result from antibodies of lower affinity.In contrast, the residual IgG1-binding to Mal d 1 at lower pH values was similar between Mal d 1-and Bet v 1-SLIT-treated individuals suggesting a comparable proportion of highly affine antibodies.However, we do not know how many different clones contributed to these results in either SLIT group.