Circulating Ara h 6 as a marker of peanut protein absorption in tolerant and allergic humans following ingestion of peanut‐containing foods

Bioaccessibility of food allergens may be a key determinant of allergic reactions.


| INTRODUC TI ON
in healthy volunteers, reflecting their bioaccessibility. 2 Peanut proteins were detected in serum using a cell-based assay, following both ingestion but also isolated chewing (without ingestion) in non-allergic volunteers, demonstrating the existence of an oral phase to allergen absorption across the buccal mucosa. 3 Food proteins have also been detected at low levels in placental tissue [4][5][6] and in breastmilk. [7][8][9][10][11] The transfer of food antigens may be enhanced in diseases where intestinal barrier integrity is impaired, such as coeliac disease 12 and in subjects with underlying food allergy, [13][14][15][16] although data are limited. Intestinal permeability is not predictive of food allergy, 17 but may play a role in specific conditions such as wheat-dependent exercise-induced anaphylaxis. 13 Whether modified gut permeability might be a contributory factor or consequence of food allergy beyond exercise-induced anaphylaxis to wheat is unclear. However, the assessment of passage of allergens into the bloodstream could yield insights into the pathogenesis of IgE-mediated food allergy and provide information on factors which contribute to symptom severity. 18 The passage of food allergens into the blood may be affected by the form in which they are ingested, which in turn reflects the impact of food processing and of the food matrix on physicochemical properties of allergens. This can be due to the chemical modification, aggregation, intermolecular interactions and/ or denaturation of the allergen itself, as well as interactions of the allergen with matrix components. 19 The inclusion of egg or cow's milk into a baked matrix does not seem to impact on the dose of allergen needed to elicit objective symptoms in allergic patients, 20,21 although fat-rich matrices can impact on both the kinetics of symptoms and potentially their severity, at least for peanut. 22 Higher fat contents may prolong gastric residence time, delaying the onset of symptoms and potentially allowing more allergen to be consumed. 23 The protein content of the food matrix may also affect allergen absorption. 24 Ara h 6, a peanut 2S albumin, is a small and compact protein which is highly resistant to gastrointestinal digestion and is able to cross the intestinal epithelial barrier in a Caco-2 cell model. [25][26][27] It is a major allergenic determinant of peanut allergy, 28,29 which is the most common cause of food-related anaphylaxis in Europe. 30,31 Ara h 6 is therefore a good candidate to evaluate transfer of peanut allergen into the blood compartment. Indeed, a recent study detected low levels of Ara h 6 (but not Ara h 2) in the sera of some but not all healthy volunteers following ingestion of high amounts of peanut (at least 25 g peanut protein, equivalent to >100 peanuts). 32 Allergen detection in blood is difficult, because the allergens may be present at very low concentrations and be modified during digestion, impairing detection by antibody-dependent assays. In addition, immunoassays are subject to interference from the interaction of food proteins or fragments derived thereof with other plasma constituents, such as allergen-specific antibodies: JanssenDuijghuijsen et al recently described the significant interference of IgG antibodies on the immunodetection of Ara h 6 following ingestion of peanut by healthy volunteers. 33 We therefore aimed to develop a simple blood pretreatment to minimize interference from serum proteins such as specific antibodies, irrespective of their isotype and concentration. We combined this approach with an in-house immunoassay previously used for Ara h 6 detection in breastmilk. 10 Using this approach, we assessed the absorption and kinetics of Ara h 6 passage into the blood following ingestion of low amounts of peanut incurred in different food matrices. We also evaluated whether uptake kinetics might differ between peanut-allergic and non-allergic individuals.

| Biological material
For the development and optimization of the pretreatment protocol and associated immunoassay, plasma from peanut-allergic patients was purchased from PlasmaLab International (Everett, WA, USA).
Following assay optimization, serum samples were collected from healthy volunteers and peanut-allergic patients as described below.

Conclusions and Clinical Relevance:
The kinetics and intensity of Ara h 6 passage in bloodstream depend on both individual and food matrix. Peanut-allergic patients appear to demonstrate higher absorption rate, the clinical significance of which warrants further evaluation.

K E Y W O R D S
allergy, Ara h 6, bioaccessibility, food matrix, peanut 2.2 | Analysis of specific humoral response to Ara h 6 in plasma/serum Enzyme immunometric assays were performed in 96-well microtiter plates (Immunoplate Maxisorb®, Nunc) using AutoPlate Washer and Microfill dispenser equipment from BioTek instruments (Avantec).
Specific antibodies against Ara h 6 purified from roasted peanut 34 were analysed by direct enzyme allergosorbant test on allergencoated plates, as previously described. 35,36

| Food matrices used for feeding trials and food challenges
The challenge matrices are described in Tables S1-S3. Roasted peanuts were purchased from KP nuts; peanut butter from Kraft Foods.
Defatted light roasted peanut flour (Golden Peanut Company; 12% fat) was incurred at different levels into a water-continuous dessert base matrix adapted from that developed within the EuroPrevall project (EU-funded FP6) for double-blinded, placebo-controlled food challenges (DBPCFC) and hydrated prior to use. 37,38 The same peanut flour was also included in a baked cookie matrix. The protein content of the defatted peanut flour was determined by Kjeldahl total nitrogen employing a conversion factor of 5.4, while the manufacturer's specified protein content was used for roasted peanut and peanut butter.

| Human subjects and ethics
Subjects were recruited at Imperial College (London, UK). Informed consent was obtained from all volunteers, and the study was approved by the NHS Health Research Authority (reference 15/ LO/0286).

| Healthy volunteer study
Volunteers (n = 6) were fasted for at least four hours prior to study and had avoided peanut consumption for at least 7 days prior. An intravenous cannula was sited and baseline samples taken, following which volunteers ingested increasing amounts of peanut proteins (100-3000 mg peanut protein) incurred in one of four different matrices (water-continuous dessert matrix, cookie, peanut butter and as roasted peanuts). The different matrices were tested by each volunteer on separate occasions, at least one week apart, and blood samples collected as shown in Figure S1. Due to the excessive amount of water-continuous dessert matrix required to provide a 3000 mg dose, the final dose was instead substituted with a 10 g portion of roasted peanuts (~2.5 g peanut protein) to serve as a positive control. Blood samples were allowed to clot at room temperature for 20 minutes, centrifuged and serum aliquoted and snap-frozen at −80°C.

| Oral food challenges in peanut-allergic individuals
Peanut-allergic subjects (n = 14; Table 1) were recruited to a randomized crossover study (TRACE Peanut study, described elsewhere) 39 and had undergone initial double-blind, placebo-controlled food challenge (DBPCFC) to peanut to confirm their allergic status.
Participants then underwent further DBPCFC conducted according international consensus criteria (PRACTALL). 40 We used an identical protocol to that used in the TRACE Peanut study. 39 In brief, DBPCFC was performed on two separate days, at least 14 days apart. On each day, subjects were fasted from at least 4 hours prior to challenge.
A cannula was sited and baseline blood sample collected. Subjects then received increasing doses, every 30 minutes, of peanut protein (or placebo) at the following doses: 3 µg, 30 µg, 300 µg, 3 mg, 30 mg, 100 mg, 300 mg and 1000 mg (incurred in the same water-continuous dessert matrix used above) until stopping criteria were met. 39 The order of DBPCFC challenges was determined by a computergenerated randomization table. Serum samples were collected at 30 and 120 minutes after occurrence of objective allergic reaction and processed as above.

| Production and characterization of monoclonal antibody for detection of Ara h 6
Anti-Ara h6 monoclonal antibodies (mAbs) were generated in mice using a classical fusion procedure, 41  Anaphylaxis at challenge 21% reactivity towards Ara h 6 and Ara h 2, and their specificity fully characterized using mutated Ara h 6, proteolysed isoforms of Ara h 6 and chimeric 2S-albumins, as previously described for human IgEbinding studies. 34,42,43 Anti-Ara h 6 mAbs produced have been involved in various studies. [44][45][46] The two mAbs selected for the current study recognized proteolysed forms of Ara h 6 and were directed against C-terminal and N-terminal part of Ara h 6, respectively.

| Ara h 6 quantification in human plasma/ serum samples
Anti-Ara h 6 mAb was immobilized on 96-well microtiter plates Ellman's reagent was used as enzyme substrate and absorbance was measured at 414 nm using automatic reader plates. Mean blank and SD blank were estimated by measuring 8 to 10 replicates of signals obtained with EIA buffer. The limit of detection (LoD, mean blank + 3xS-D blank ) and of quantification (LoQ, mean blank + 10*SD blank ) of Ara h 6 were 0.006 and 0.008 ng/mL, respectively. No cross-reactivity was observed with other peanut proteins, including Ara h 2 (Table S4).
To quantify Ara h 6 in samples from the volunteers/allergic patients after the different feeding trials or DBPCFC, Whole Peanut Protein Extract (WPPE) was prepared from each of the peanut source incurred in matrices (ie roasted peanut, peanut flour or peanut butter) and used as a standard for Ara h 6 quantification.
One gram of roasted peanut, peanut flour or peanut butter was resuspended in 20 mL of 0.3% sodium borate pH 9.0 including 0. Ara h 6 content in the different WPPEs were assessed by the sandwich ELISA assay described in this work, using Ara h 6 purified from roasted peanut as a reference. 34 Whole peanut protein extracted from roasted peanut, peanut flour or butter contained 3.8%, 4% and 4.4% of Ara h 6, respectively.

| Treatment of blood samples
Two hundred microlitre aliquots of plasma or serum were mixed with

| Endogenous antibodies interfere with Ara h 6 detection in plasma/serum
We first evaluated immunodetection of Ara h 6 in blood samples from healthy volunteers collected prior to peanut ingestion and then spiked with WPPE (0-10 ng/mL, corresponding to 0-0.4 ng/mL of Ara h 6). Ara h 6 was poorly detected in spiked serum ( Figure S2A), and detection was not enhanced when using a mix of mAbs with various specificities, both as capture and revelation antibodies, in the Ara h 6 immunoassays ( Figure S2B). Similar experiment performed on individually spiked sera from 4 other healthy volunteers showed comparable results, with a maximum recovery of 50% depending on the serum tested. These results suggested that factors present in serum were interfering with Ara h 6 detection.
Ara h 6 detection was improved (recovery of 30%) if the plasma had been depleted of IgE, and was increased up to ~70% when upstream partial depletion of immunoglobulins was performed using protein A (removing mainly IgGs), demonstrating that antibodies (such as IgG and IgE) present in plasma were causing significant interference with Ara h 6 immunodetection ( Figure S3). Moreover, Ara h 6 was poorly detected in the flowthrough fraction when plasma was spiked prior to protein A depletion, indicating that Ara h 6 was complexed to immunoglobulins retained on protein A.
Accordingly, Ara h 6-specific IgE, IgG1, IgG2 and IgG4 were detected at significant levels in blood samples from healthy volunteers and peanut-allergic patients (Table S5). Sera from healthy volunteers contained less anti-Ara h 6 IgGs compared to peanut-allergic patients, and no detectable IgE to Ara h 6. Allergic patients exhibited various levels of IgE, IgG1 and IgG4 specific to Ara h 6 (0.6-48 kUA/L, 280-3380 ng/mL and <10-820 ng/mL respectively).

| Development of a pretreatment protocol to suppress interference due to endogenous anti-Ara h 6 antibodies
We then tested various pretreatment protocols on spiked samples with the aim to irreversibly dissociate endogenous plasma protein-Ara h 6 complexes, thus liberating Ara h 6 for immunodetection.
Tested conditions relied on acidic shock, with or without additional heat treatment. We utilized commercially-sourced pooled plasma from peanut-allergic patients due to its anti-Ara h 6 IgE and IgGs levels and the availability of sufficient volume for protocol development. We observed that Ara h 6 recognition by mAb was not drastically modified by the tested treatments, by assaying Ara h 6 in buffer spiked with 20 ng/mL WPPE (0.8 ng/mL of Ara h 6) and then treated ( Figure 1, empty grey bars). No background signal intensity, assessed on non-spiked buffer, was observed after the different treatments.
Ara h 6 immunodetection in spiked buffer was only slightly decreased after acidic shock, and additional heating of sample (30 minutes, 60°C) after acidic shock did not further decrease recovery ( Figure 1 grey bars, conditions C-IV to CVIII).
A similar experiment was then performed using pooled plasma from peanut-allergic patients spiked with 20 ng/mL WPPE. We con- Whereas Ara h 6 was poorly detected in spiked sera without treatment, even at higher concentrations, recovery was high in treated plasma ( Figure S4). Curves obtained in treated plasma were equivalent to those obtained in buffer. The limit of detection for Ara h 6 in spiked and treated sera was 0.2 ng/mL of equivalent peanut protein, close to that obtained in buffer.
Finally, spiking experiments were performed on individual serum samples from healthy volunteers (n = 6) and allergic patients (n = 14) collected before the feeding trials. Each sample was spiked (or not) with 20 ng/mL WPPE (corresponding to 0.8 ng/mL of Ara h 6). Ara h 6 detection was then performed on both spiked and unspiked sera, with or without the acidic shock/heat pretreatment. No significant signal was observed in non-spiked basal serum (with our without treatment). In untreated samples, 22%-50% of Ara h 6 was recovered in sera from healthy volunteers, but <5% in spiked samples from peanut-allergic patients ( Figure 2). However, pretreatment resulted in Ara h 6 recovery of 88%-99% and 72%-98% in individual samples from healthy volunteers and allergic patients, respectively ( Figure 2).

| Detection of Ara h 6 in blood from healthy volunteers following roasted peanut kernel ingestion
Ara h 6 was then assayed in serum samples from six healthy volunteers (median age 28 years, 50% male) collected following ingestion of increasing quantities of peanut as the roasted kernel (protocol shown in Figure S1) and then subjected to pretreatment.

| Impact of the food matrix on Ara h 6 bioaccessibility
We then compared the absorption kinetics of Ara h 6 in the same healthy volunteers after ingestion of peanut incurred in different matrices, that is peanut kernel, cookies, water-continuous dessert base and peanut buffer. Ara h 6 was detected in most of the volunteers 30 minutes after ingestion of 300 mg of peanut protein, irrespective of the matrix (Figure 4). However, there were differences between the matrices tested: Ara h 6 was detected in all samples from 30 minutes following ingestion of 1000 mg peanut protein when incorporated into a cookie or peanut butter, but only after 120 minutes with the water-continuous dessert base. Interestingly, Ara h 6 concentrations in serum were highest following ingestion of peanut as the roasted kernel. We did not otherwise observe statistical differences in absorption/detection kinetics between the different matrices, although these data are limited by the small sample size and inter-individual variability.

| Absorption of Ara h 6 in peanut-allergic compared to healthy volunteers
Finally, we compared minimum ingestion levels of peanut needed to result in detectable Ara h 6 in treated samples, in 8 healthy (six of whom were included in the data presented above) and 14 peanut-allergic subjects, the latter who had undergone repeat DBPCFC to peanut using the same water-continuous dessert base as that eaten by non-allergic volunteers. In allergic subjects, Ara h 6 was detected following consumption amounts as low as 30 mg peanut protein (P < .05, Friedman's test) in contrast to healthy volunteers ingesting the same discrete dose of peanut protein ( Figure 5). No Ara h 6 was detected at placebo challenge.

| D ISCUSS I ON
Immunoassays generally remain the method of choice in detecting food allergens in complex food matrices due to their high sensitivity, specificity and simplicity, and have also been applied to the detection of peanut allergens such as Ara h 6 in human biological samples. 10,11,32,33 However, endogenous proteins, particularly in the blood, can cause significant interference, either by inducing a non-specific signal (eg heterophilic antibodies) or limiting the detection of the target molecule. Immunoassays (including Ara h 6 ELISA) have been used to evaluate peanut absorption in healthy individuals following consumption of high amounts (100 g) of roasted peanut or peanut flour, but Ara h 6 was undetectable in 33%-40% of volunteers. 32,33 Using spiking experiments in combination with immunoglobulin G depletion, JanssenDuijghuijsen et al concluded that allergen-specific IgG/IgG 4 in the blood of healthy volunteers interfered with the detection of Ara h 6 following peanut consumption. 33 Our data are consistent with this and further suggest that Ara h 6-specific IgG (IgG1 and IgG4) but also IgE antibodies may interfere or even totally inhibit Ara h 6 immunodetection, even with a highly sensitive immunoassay. We further demonstrated through spiking experiments that the recovery of Ara h 6 is dramatically lower in serum from peanut-allergic patients compared to non-allergic individuals who presented with lower concentrations of Ara h 6-specific IgG and no IgE.
We therefore developed a specific pretreatment to induce irreversible dissociation of human Ig-Ara h 6 complexes, thus liberating Ara h 6 for detection. Such a treatment had to be compatible F I G U R E 3 Kinetics of Ara h 6 bioaccessibility assessed in six healthy volunteers who ingested increasing amounts of peanut proteins as the roasted kernel. Ara h 6 immunoassay was performed on individual plasma collected at various time-points ( Figure S1). Samples were treated using simultaneous acidic shock (citrate, pH 3) and heating (60°C, 30 minutes). Results are expressed as ng/ mL of whole peanut protein extract (Left Y-axis) and as ng/mL of Ara h 6 (Right Y-axis) F I G U R E 4 Ara h 6 detection expressed as ng/mL of WPPE (Left Y-axis) and as ng/mL of Ara h 6 (Right Y-axis) in treated serum from the six healthy volunteers who ingested increasing amount of peanut proteins incorporated in various food matrices and as roasted peanut kernels. *P < .05 with our immunoassay conditions and not cause structural changes that would prevent recognition of Ara h 6 by the mAb used in the immunoassay. Acidic shock is classically used to disrupt the binding of antigens to antibodies during immunopurification, but the acidic conditions tested were insufficient, probably because they did not irreversibly denature antibodies and prevent further re-association of endogenous antibody to the liberated Ara h 6. Various physical and chemical methods have been previously used to fractionate serum proteins. For example, heating plasma to 60°C at pH 5 induced denaturation and precipitation of proteins other than albumin. 47 Many studies have described the impact of temperature and/ or pH on denaturation, unfolding and aggregation states of immunoglobulins, [48][49][50] thus reducing their capacity to bind antigen. Since Ara h 6 has been shown to be a highly structured thermostable soluble protein, 51 we applied various combinations of acidic and heat pretreatment to dissociate Ara h 6-Ig complexes and then denature the dissociated antibodies. The resulting procedure, combining simultaneous acidic shock at pH 3 and heating at 60°C, allowed almost complete recovery of Ara h 6, irrespective of the concentrations and isotype of endogenous antibodies. Although the procedure may have led to slight structural modifications of Ara h 6, it allowed further recognition of Ara h 6 without compromising sensitivity.
With this strategy, we were successful in detecting Ara h 6 in serum samples collected from both healthy and allergic volunteers following ingestion of much lower amounts of peanut (<1 g) than that previously reported in the literature. Actually, we were able to detect Ara h 6 following consumption of 300 mg peanut protein (equivalent to ~1½ peanuts) by all healthy volunteers, with Ara h 6 detection increasing in a dose-dependent manner. JanssenDuijghuijsen et al estimated that their assay was able to detect 0.0001% of Ara h 6 intake, following consumption of 100 g peanut (~25 g peanut protein). 33 Applying a similar methodology to our data, we were able to measure 0.12-0.36 ng/mL of Ara h 6 (3-9 ng/mL of WPPE) in serum from volunteers, 30 minutes following ingestion of 120 mg of Ara h 6 (3000 mg of peanut proteins from roasted kernels). This equates to ~0.001% of estimated intake, that is ~10× more recovery than that described by JanssenDuijghuijsen et al.
Of note, significantly lower levels of ingestion (just 30 mg peanut protein) were required in peanut-allergic individuals to detect peanut protein in patient sera. In addition, higher levels of Ara h 6 were detected in sera from peanut-allergic individuals compared to healthy controls, for the same dose of peanut exposure in the same matrix (the feeding protocols could not be identical, due to the need to utilize a low dose, incremental challenge protocol for allergic volunteers to ensure patient safety). These data imply that peanut-allergic individuals have different absorption kinetics from non-allergic individuals, potentially due to antibody-mediated facilitated absorption, something that has also been demonstrated via the low affinity-IgE receptor in a laboratory model. 52  Other factors can also modify the bioaccessibility of food proteins.
Food processing may alter structure of allergens, depending on their physicochemical properties. Moreover, the composition of the food matrix and its caloric value alters the digestive process, modifying digestive secretions and other factors such as gastric residence time. These will all impact on apparent allergen bioaccessibility. It is difficult to determine which might be the predominant factors in explaining the differences in Ara h 6 kinetics we observed. In healthy F I G U R E 5 Ara h 6 detection (expressed in equivalents of WPPE) in treated serum samples from 8 healthy volunteers versus 14 allergic volunteers prior to, and following ingestion of discrete doses of peanut protein in a water-continuous dessert base. Samples were collected 30 and 120 minutes after consumption in healthy volunteers, and 30 and 120 minutes following objective reaction to the same amount of peanut protein consumed in allergic individuals. *P < .05, **P < .01 and ***P < .001 volunteers, the bioaccessibility of Ara h 6 was greatest when peanut was ingested as the roasted kernel. Thus, incorporating peanut in complex food matrix or using defatted peanut flour is likely to affect Ara h 6 passage into bloodstream. The fat content of the food matrix can affect onset and severity of symptoms experienced in peanut-allergic individuals, 22,23 which may be due to reduced IgEbinding capacity to peanut proteins for food matrices with higher fat content 22,53 and/or delayed gastric emptying. 23 In conclusion, we have developed a method which allows for the evaluation of peanut allergen passage into the blood following ingestion of small amounts of peanut protein in human subjects, something which hitherto has not been possible in allergic individuals due to the small levels of allergen exposure needed to trigger reactions.
Our study underlines the variability of absorption kinetics between volunteers-and in particular, allergic versus non-allergic subjectsand the impact of the food matrix on bioaccessibility. Further studies are needed to assess how other factors, such as medication, alcohol and exercise, may further impact upon this, and the extent to which levels of bioaccessible allergen may determine clinical reactions.

ACK N OWLED G EM ENTS
We thank Professor Stephen Durham, Dr Robert Boyle and Dr Isabel Skypala for their clinical support. We also thank all the volunteers who provided samples for the analyses presented in this report, including allergic volunteers recruited through the TRACE Peanut study (funded by the UK Food Standards Agency). We are grateful to the TRACE study investigators (CI A Clark) and the Food Standards Agency for their support.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

CLI N I C A L TR I A L R EG I S TR ATI O N
The study was approved by the UK NHS Health Research Authority (reference 15/LO/0286).