Release of Major Peanut Allergens from Their Matrix under Various pH and Simulated Saliva Conditions—Ara h2 and Ara h6 Are Readily Bio-Accessible

The oral mucosa is the first immune tissue that encounters allergens upon ingestion of food. We hypothesized that the bio-accessibility of allergens at this stage may be a key determinant for sensitization. Light roasted peanut flour was suspended at various pH in buffers mimicking saliva. Protein concentrations and allergens profiles were determined in the supernatants. Peanut protein solubility was poor in the pH range between 3 and 6, while at a low pH (1.5) and at moderately high pHs (>8), it increased. In the pH range of saliva, between 6.5 and 8.5, the allergens Ara h2 and Ara h6 were readily released, whereas Ara h1 and Ara h3 were poorly released. Increasing the pH from 6.5 to 8.5 slightly increased the release of Ara h1 and Ara h3, but the recovery remained low (approximately 20%) compared to that of Ara h2 and Ara h6 (approximately 100% and 65%, respectively). This remarkable difference in the extraction kinetics suggests that Ara h2 and Ara h6 are the first allergens an individual is exposed to upon ingestion of peanut-containing food. We conclude that the peanut allergens Ara h2 and Ara h6 are quickly bio-accessible in the mouth, potentially explaining their extraordinary allergenicity.


Introduction
Peanuts (Arachis hypogaea L.) are widely consumed in Western countries, Asia, and Africa. The current top three countries in peanut production are China, India, and the USA. The peanut plant thrives in tropical and subtropical climates, preferring well-drained sandy soils, and is cultivated within a wide geographical area [1]. Although peanut is consumed raw or boiled in some cultures,

Screening the pH Effect on the Efficiency of Peanut Protein Extraction
Extraction media covering a wide range of pH were used to screen the effect of the extraction pH on the protein content and protein profile of peanut flour extracts. Figure 1 shows a clear minimum efficiency in the extraction of proteins between pH 3 and pH 6. Both at lower pH values and higher pH values, protein extractability became higher, in particular at a pH corresponding to the typical stomach conditions. Nutrients 2018, 10, x FOR PEER REVIEW 4 of 13

Screening the pH Effect on the Efficiency of Peanut Protein Extraction
Extraction media covering a wide range of pH were used to screen the effect of the extraction pH on the protein content and protein profile of peanut flour extracts. Figure 1 shows a clear minimum efficiency in the extraction of proteins between pH 3 and pH 6. Both at lower pH values and higher pH values, protein extractability became higher, in particular at a pH corresponding to the typical stomach conditions. Such minimal protein extractability has been shown for several other species of the legume plants [20,21] and has been described in detail for soy [22,23]. The iso-electric point of the abundant proteins in legume seeds is in the range from 4 to 5 [24], and, at this pH, a protein has limited charge and therefore low polarity, which leads to poor solubility in aqueous media. In fact, iso-electric precipitation is industrially used to manufacture soy protein concentrates and isolates [25]. While peanut protein concentrates or isolates are not commercially available, the poor solubility of peanut proteins at pH between 3 and 6 has been shown earlier [26]. Using the extraction ratio (1:30 w/v) and taking into account the protein content of the peanut flour (50%), the theoretical maximal protein concentration is 16.7 mg/mL. The highest protein concentration we found was about 8.5 mg/mL at pH 1.5. This concentration corresponds to a recovery of about 50%. At a high pH, we found a protein concentration of about 5 mg/mL, which corresponds to about 30% recovery. Thus, even at an extremely low pH and at moderately high pH, protein extraction is not complete in the chosen extraction conditions (90 min at room temperature). It may be that a higher pH would lead to higher extraction recoveries, but pH > 9 is not physiological and was not tested. We included a pH as low as 1.5 because this can reflect the stomach pH. The fact that not all proteins were recovered in the extract is not surprising, as it is known that (light) roasting decreases peanut protein extractability [27].
Apart from the overall protein extractability, we were interested to determine if the pH during the extraction would impact protein composition. Figure 2 shows the protein profiles of the different extracts. In order to facilitate band comparisons, all extracts were diluted to 1 mg/mL, except for the extracts at pH 3, 4, and 5, whose concentrations were already slightly lower than 1 mg/mL. At extremely low pHs and at moderately high pHs, the typical protein band pattern of peanut extracts was observed. In a recent study, we investigated the protein profile of different peanut types and assigned bands using purified individual peanut proteins, in particular the allergens Ara h1, Ara h2, Ara h3, and Ara h6, which together represent about 90-95% of the protein content of a peanut kernel Such minimal protein extractability has been shown for several other species of the legume plants [20,21] and has been described in detail for soy [22,23]. The iso-electric point of the abundant proteins in legume seeds is in the range from 4 to 5 [24], and, at this pH, a protein has limited charge and therefore low polarity, which leads to poor solubility in aqueous media. In fact, iso-electric precipitation is industrially used to manufacture soy protein concentrates and isolates [25]. While peanut protein concentrates or isolates are not commercially available, the poor solubility of peanut proteins at pH between 3 and 6 has been shown earlier [26]. Using the extraction ratio (1:30 w/v) and taking into account the protein content of the peanut flour (50%), the theoretical maximal protein concentration is 16.7 mg/mL. The highest protein concentration we found was about 8.5 mg/mL at pH 1.5. This concentration corresponds to a recovery of about 50%. At a high pH, we found a protein concentration of about 5 mg/mL, which corresponds to about 30% recovery. Thus, even at an extremely low pH and at moderately high pH, protein extraction is not complete in the chosen extraction conditions (90 min at room temperature). It may be that a higher pH would lead to higher extraction recoveries, but pH > 9 is not physiological and was not tested. We included a pH as low as 1.5 because this can reflect the stomach pH. The fact that not all proteins were recovered in the extract is not surprising, as it is known that (light) roasting decreases peanut protein extractability [27].
Apart from the overall protein extractability, we were interested to determine if the pH during the extraction would impact protein composition. Figure 2 shows the protein profiles of the different extracts. In order to facilitate band comparisons, all extracts were diluted to 1 mg/mL, except for the extracts at pH 3, 4, and 5, whose concentrations were already slightly lower than 1 mg/mL. At extremely low pHs and at moderately high pHs, the typical protein band pattern of peanut extracts was observed. In a recent study, we investigated the protein profile of different peanut types and assigned bands using purified individual peanut proteins, in particular the allergens Ara h1, Ara h2, Ara h3, and Ara h6, which together represent about 90-95% of the protein content of a peanut kernel [16]. All these bands were well visible at extremely low pHs and at moderately high pHs. In contrast, at neutral and slightly acidic pHs (3 to 7.6), the bands corresponding to Ara h3 and, to a lesser extent, Ara h1, were far less clear. The bands corresponding to Ara h2 and Ara h6 remained extractable over a wider pH range, although they were poorly visible at pH 4 and 5.
Nutrients 2018, 10, x FOR PEER REVIEW 5 of 13 [16]. All these bands were well visible at extremely low pHs and at moderately high pHs. In contrast, at neutral and slightly acidic pHs (3 to 7.6), the bands corresponding to Ara h3 and, to a lesser extent, Ara h1, were far less clear. The bands corresponding to Ara h2 and Ara h6 remained extractable over a wider pH range, although they were poorly visible at pH 4 and 5. Interestingly, at pH 6 the protein profile of the extract indicated a strong enrichment in Ara h2 and Ara h6. Thus, at this pH, Ara h2 and Ara h6 were highly bio-accessible, while the other allergens Ara h1 and Ara h3 were not. At pH 1.5, a condition similar to that of the human stomach, the protein extractability was high, and all major allergen bands were present. This suggests that in the stomach, all allergens from light roasted peanut flour become bio-accessible.
Centrifugation was used to separate the soluble parts of the extracts (supernatant) from the insoluble parts (pellet). We cannot exclude that some undissolved material or soluble aggregates appeared inadvertently in the supernatants, but, if any, this would be very limited, given the low recoveries of extraction at pH 3 and 4 ( Figure 1) and the absence of Ara h1 band in extracts at these pHs ( Figure 2).
Poms et al. [28] tested the effect of pH on peanut protein extraction and the reactivity of peanut extracts in commercial enzyme-linked immune-sorbent assays (ELISAs). Interestingly, one ELISA protocol that utilized pH 6.7 as the extraction pH showed poor recovery, while the other ELISA protocols, all using extraction pHs of 7.4 or higher, showed better recoveries. It is not known whether this difference was solely due to the lower extraction pH or whether other factors played a role as well. The specificity of the used ELISA tests was not known at the time of this study, but a recent paper showed that the commercial ELISA kits used by Poms [28] essentially recognize Ara h3 [29], the protein whose extractability appeared low at pHs from 6 to 7.2 in our experiments. A more recent study by Sathe et al. [30] showed that the recovery of proteins by extraction from peanuts was about two times higher at pH 8.45 than at pH 7.2, and that in particular Ara h3 showed a different solubility. Our data are also in line with the work of Walczyk et al. [31], who showed that peanut protein solubility is low at pH 4.75 and increases at pH ≥ 8.5. Interestingly, at pH 6 the protein profile of the extract indicated a strong enrichment in Ara h2 and Ara h6. Thus, at this pH, Ara h2 and Ara h6 were highly bio-accessible, while the other allergens Ara h1 and Ara h3 were not. At pH 1.5, a condition similar to that of the human stomach, the protein extractability was high, and all major allergen bands were present. This suggests that in the stomach, all allergens from light roasted peanut flour become bio-accessible.
Centrifugation was used to separate the soluble parts of the extracts (supernatant) from the insoluble parts (pellet). We cannot exclude that some undissolved material or soluble aggregates appeared inadvertently in the supernatants, but, if any, this would be very limited, given the low recoveries of extraction at pH 3 and 4 ( Figure 1) and the absence of Ara h1 band in extracts at these pHs ( Figure 2).
Poms et al. [28] tested the effect of pH on peanut protein extraction and the reactivity of peanut extracts in commercial enzyme-linked immune-sorbent assays (ELISAs). Interestingly, one ELISA protocol that utilized pH 6.7 as the extraction pH showed poor recovery, while the other ELISA protocols, all using extraction pHs of 7.4 or higher, showed better recoveries. It is not known whether this difference was solely due to the lower extraction pH or whether other factors played a role as well. The specificity of the used ELISA tests was not known at the time of this study, but a recent paper showed that the commercial ELISA kits used by Poms [28] essentially recognize Ara h3 [29], the protein whose extractability appeared low at pHs from 6 to 7.2 in our experiments. A more recent study by Sathe et al. [30] showed that the recovery of proteins by extraction from peanuts was about two times higher at pH 8.45 than at pH 7.2, and that in particular Ara h3 showed a different solubility. Our data are also in line with the work of Walczyk et al. [31], who showed that peanut protein solubility is low at pH 4.75 and increases at pH ≥ 8.5.

Bio-Accessibility of Peanut Allergens in Artificial Saliva
The production, composition, and pH of saliva is influenced by many different factors, including mastication. At resting conditions, the dominant saliva-producing glands are the submandibular and sublingual glands, while upon stimulation such as by chewing, the parotid glands take over the Nutrients 2018, 10, 1281 6 of 13 majority of saliva production [32,33]. Saliva produced by the parotid glands has a higher carbonate concentration than saliva produced by the submandibular and sublingual glands, which induces an increase in pH upon chewing. In resting conditions, saliva pH ranges normally from 6.5 to 7.5, while the pH may rise to above 8 under chewing conditions [33]. On the basis of the literature and experimental data, Crea et al. [18] proposed a composition for artificial saliva. To investigate peanut protein dissolution under saliva conditions, we used the essential elements of this composition and set the pH at 7.8 to mimic the chewing conditions. Figure 3 shows the protein profiles of artificial saliva extracts of peanut flour prepared with 2, 6, and 20 min of extraction at 37 • C. This time course was short to mimic the exposure in the mouth and during transportation through the esophagus to the stomach. In order to meet the first (2 min) time point, the centrifugation step was kept short, and this could potentially result in the presence of some undissolved material or soluble aggregates in the supernatant samples. The observation that, in some conditions (such as at pH 6.5, see Figure 4), hardly any Ara h1 was visible on SDS-PAGE indicated that this effect was limited. Using dissolution in artificial saliva, protein bands of all allergens (Ara h1, Ara h2, Ara h3, and Ara h6) were clearly visible, even though the extraction time was short. In fact, already at 2 min, the earliest possible time point because of the time needed for sampling and centrifuging, intense bands were observed for these allergens. Increasing the extraction time from 2 to 6 to 20 min only moderately increased the intensity of the protein bands. This suggests that peanut allergens are readily available in the mouth very soon after ingestion of peanut flour, when saliva pH is relatively high.

Bio-Accessibility of Peanut Allergens in Artificial Saliva
The production, composition, and pH of saliva is influenced by many different factors, including mastication. At resting conditions, the dominant saliva-producing glands are the submandibular and sublingual glands, while upon stimulation such as by chewing, the parotid glands take over the majority of saliva production [32,33]. Saliva produced by the parotid glands has a higher carbonate concentration than saliva produced by the submandibular and sublingual glands, which induces an increase in pH upon chewing. In resting conditions, saliva pH ranges normally from 6.5 to 7.5, while the pH may rise to above 8 under chewing conditions [33]. On the basis of the literature and experimental data, Crea et al. [18] proposed a composition for artificial saliva. To investigate peanut protein dissolution under saliva conditions, we used the essential elements of this composition and set the pH at 7.8 to mimic the chewing conditions. Figure 3 shows the protein profiles of artificial saliva extracts of peanut flour prepared with 2, 6, and 20 min of extraction at 37 °C. This time course was short to mimic the exposure in the mouth and during transportation through the esophagus to the stomach. In order to meet the first (2 min) time point, the centrifugation step was kept short, and this could potentially result in the presence of some undissolved material or soluble aggregates in the supernatant samples. The observation that, in some conditions (such as at pH 6.5, see Figure 4), hardly any Ara h1 was visible on SDS-PAGE indicated that this effect was limited. Using dissolution in artificial saliva, protein bands of all allergens (Ara h1, Ara h2, Ara h3, and Ara h6) were clearly visible, even though the extraction time was short. In fact, already at 2 min, the earliest possible time point because of the time needed for sampling and centrifuging, intense bands were observed for these allergens. Increasing the extraction time from 2 to 6 to 20 min only moderately increased the intensity of the protein bands. This suggests that peanut allergens are readily available in the mouth very soon after ingestion of peanut flour, when saliva pH is relatively high.   There is limited research published on the release of peanut allergens in saliva. One study showed that Ara h1 could be detected in saliva 5 min after the ingestion of peanut butter [8]. However, in that study the saliva samples were extracted before quantification of Ara h1, and this extraction step may have released Ara h1 from undissolved peanut material in the sample, leading to an overestimation of the Ara h1 solubility. To our knowledge, no other published data exist on the release of peanut or other allergens in saliva.

The pH of the Extraction Medium Can Be Influenced by Peanut Flour
An important difference between the in vivo situation corresponding to human ingestion of peanut flour-containing food products and that realized through our extraction model is that our model is based on a fixed volume of extraction medium per amount of peanut flour. In vivo, the salivary glands will be stimulated to compensate for the saliva absorbed by the food in the mouth. In our extraction model, the peanut flour may change the pH as a consequence of the buffering effect of To investigate to what extent this occurs, the pH of a range of buffers mimicking saliva (pH range from 6.5 to 8.5) was measured after preparing the buffer and after mixing the peanut flour with the buffer (Table 1). For buffers with pH < 7, a small acidifying effect was observed (less than 0.2 pH units). However, for buffers at higher pHs, the acidifying effect was stronger, up to 0.8 pH units. The amounts of hydroxide needed to re-adjust the pH increased as a consequence of the increasing pH of the extraction medium. This observation may have implications for protein dissolution studies in general, as the targeted pH is not maintained after mixing protein powders with the extraction medium and may lead to false conclusions when a change in pH is not acknowledged or not adjusted. Our data showed that the acidifying effect was quick and that, as soon as the pH was readjusted, it remained stable (Table 1). For further experiments described in this paper, the required amount of hydroxide was added to the mix of peanut flour and buffer to adjust the pH to the target value.

Bio-Accessibility of Peanut Allergens at the Normal pH Range of Saliva
Because saliva pH can vary depending on mastication activity, we evaluated the effect of pH (range 6.5 to 8.5) on the dissolution of peanut allergens. Figure 4 shows that, at the higher end of the range, intense bands, corresponding to Ara h1, Ara h2, Ara h3, and Ara h6, were recovered in the extracts. A gradual decrease of band intensity was observed when decreasing the pH from pH 7.5 to 6.5 (Figure 4).  There is limited research published on the release of peanut allergens in saliva. One study showed that Ara h1 could be detected in saliva 5 min after the ingestion of peanut butter [8]. However, in that study the saliva samples were extracted before quantification of Ara h1, and this extraction step may have released Ara h1 from undissolved peanut material in the sample, leading  The bands of Ara h3 were more prone to this pH-dependent decrease than the bands of the other allergens, which is in line with the difference in extractability of Ara h3 between pH 7.2 and 8.45 reported by Sathe et al. [30]. Using densitometry, the amount of each allergen was determined in extracts prepared at the different pH conditions. Figure 5 shows that, at pH 6.5, the concentrations of Ara h1, Ara h2, Ara h3, and Ara h6 were 0.24, 1.52, 1.58, and 0.40 mg/mL, respectively, and these concentrations increased with increasing extraction pH, in particular for Ara h3, but also for Ara h1 and Ara h6. The recoveries for Ara h1 and Ara h3 were low at pH 6.5, i.e., 5% and 6% of the theoretical maximal values, respectively ( Figure 5B). These recovery percentages increased to 18 and 15% for Ara h1 and Ara h3, respectively, when the extraction pH was elevated to 8.5. The recovery of Ara h2 was around 100% at all saliva pHs, and the recovery of Ara h 6 increased from about 35% to 65% with increases in the pH. mix of peanut flour and buffer to adjust the pH to the target value.

Bio-Accessibility of Peanut Allergens at the Normal pH Range of Saliva
Because saliva pH can vary depending on mastication activity, we evaluated the effect of pH (range 6.5 to 8.5) on the dissolution of peanut allergens. Figure 4 shows that, at the higher end of the range, intense bands, corresponding to Ara h1, Ara h2, Ara h3, and Ara h6, were recovered in the extracts. A gradual decrease of band intensity was observed when decreasing the pH from pH 7.5 to 6.5 (Figure 4).
The bands of Ara h3 were more prone to this pH-dependent decrease than the bands of the other allergens, which is in line with the difference in extractability of Ara h3 between pH 7.2 and 8.45 reported by Sathe et al. [30]. Using densitometry, the amount of each allergen was determined in extracts prepared at the different pH conditions. Figure 5 shows that, at pH 6.5, the concentrations of Ara h1, Ara h2, Ara h3, and Ara h6 were 0.24, 1.52, 1.58, and 0.40 mg/mL, respectively, and these concentrations increased with increasing extraction pH, in particular for Ara h3, but also for Ara h1 and Ara h6. The recoveries for Ara h1 and Ara h3 were low at pH 6.5, i.e., 5% and 6% of the theoretical maximal values, respectively ( Figure 5B). These recovery percentages increased to 18 and 15% for Ara h1 and Ara h3, respectively, when the extraction pH was elevated to 8.5. The recovery of Ara h2 was around 100% at all saliva pHs, and the recovery of Ara h 6 increased from about 35% to 65% with increases in the pH. Because Ara h2 and Ara h6 are highly similar in biochemical characteristics, it was expected that they would behave the same in the extraction studies; however, the recoveries for Ara h6 were somewhat lower than those for Ara h2. Nevertheless, it is clear that at the lower end of the saliva pH range, the solubility of Ara h2 and Ara h6 was substantially higher than that of Ara h1 and Ara h3. Because Ara h2 and Ara h6 are highly similar in biochemical characteristics, it was expected that they would behave the same in the extraction studies; however, the recoveries for Ara h6 were somewhat lower than those for Ara h2. Nevertheless, it is clear that at the lower end of the saliva pH range, the solubility of Ara h2 and Ara h6 was substantially higher than that of Ara h1 and Ara h3. At the higher pHs, the recovery of Ara h1 and Ara h3 increased, yet the recovery percentages were still much lower than those of Ara h2 and Ara h6.
In addition to investigating the extracted material, we also investigated the protein profile of the non-dissolved material. By boiling the non-dissolved residues remaining after extraction in buffer containing the strong detergent SDS and a reducing agent to break disulfide bonds, we could recover proteins from the residues left over after the dissolution experiments. Figure 6 shows that, for all extraction pHs, the typical bands of Ara h1 and Ara h3 were in the pelleted fraction, together with some aggregated material visible at the top of the lanes.
In addition to investigating the extracted material, we also investigated the protein profile of the non-dissolved material. By boiling the non-dissolved residues remaining after extraction in buffer containing the strong detergent SDS and a reducing agent to break disulfide bonds, we could recover proteins from the residues left over after the dissolution experiments. Figure 6 shows that, for all extraction pHs, the typical bands of Ara h1 and Ara h3 were in the pelleted fraction, together with some aggregated material visible at the top of the lanes. The presence of intense bands for Ara h1 and Ara h3 in the pellet material is in line with the observation that only a small part of Ara h1 and Ara h3 was extracted (less than 20%; Figure 5B). In the pellet material, no Ara h2 bands were observed, and, for Ara h6, only a vague band was seen at the lowest tested pH conditions ( Figure 6). This is in line with the high recovery percentages of Ara h2 and Ara h6 ( Figure 5B).

Implications for Peanut Allergy Research
We have investigated the release of peanut allergens from peanut flour in various conditions mimicking human saliva. To date, it is not known how peanut allergens behave in the oro-esophageal area, where the first contact between food proteins and the mucosal immune system takes place. Because peanut is an extraordinary potent allergen, with some of its major allergens being more potent than others, we hypothesized that different peanut allergens my possess different bioaccessibility to the first site of contact, i.e., the mouth, thereby giving rise to different sensitization characteristics.
We selected light roasted peanut flour because it is a commonly used food ingredient and it can represent other heat-treated peanut products in terms of roasting degree. Furthermore, it is available as a well-characterized material and has been used in many peanut allergy studies because of its high lot-to-lot consistency [27,34,35].
Our data showed that Ara h2 and Ara h6 were readily released from the matrix at common saliva pH values (6.5 to 8.5), while Ara h1 and Ara h3 were poorly released. Increasing the pH to higher values reflecting the chewing conditions somewhat increased the release of Ara h1 and Ara h3, but still the recovery levels were low compared to those of Ara h2 and Ara h6. At a lower saliva  The presence of intense bands for Ara h1 and Ara h3 in the pellet material is in line with the observation that only a small part of Ara h1 and Ara h3 was extracted (less than 20%; Figure 5B). In the pellet material, no Ara h2 bands were observed, and, for Ara h6, only a vague band was seen at the lowest tested pH conditions ( Figure 6). This is in line with the high recovery percentages of Ara h2 and Ara h6 ( Figure 5B).

Implications for Peanut Allergy Research
We have investigated the release of peanut allergens from peanut flour in various conditions mimicking human saliva. To date, it is not known how peanut allergens behave in the oro-esophageal area, where the first contact between food proteins and the mucosal immune system takes place. Because peanut is an extraordinary potent allergen, with some of its major allergens being more potent than others, we hypothesized that different peanut allergens my possess different bio-accessibility to the first site of contact, i.e., the mouth, thereby giving rise to different sensitization characteristics.
We selected light roasted peanut flour because it is a commonly used food ingredient and it can represent other heat-treated peanut products in terms of roasting degree. Furthermore, it is available as a well-characterized material and has been used in many peanut allergy studies because of its high lot-to-lot consistency [27,34,35].
Our data showed that Ara h2 and Ara h6 were readily released from the matrix at common saliva pH values (6.5 to 8.5), while Ara h1 and Ara h3 were poorly released. Increasing the pH to higher values reflecting the chewing conditions somewhat increased the release of Ara h1 and Ara h3, but still the recovery levels were low compared to those of Ara h2 and Ara h6. At a lower saliva pH (pH 6), in situations of low production of saliva [33], the extracted proteins from peanut were strongly enriched in Ara h2 and Ara h6. Although this pH does not reflect an active chewing situation, it may represent the situation at the first bite during consumption, i.e., the moment where food is indeed exposed to such pH. The fact that the peanut flour has an acidifying effect on the dissolution media suggests that this low pH situation in the mouth may be maintained for a period of time. Thus, Ara h2 and Ara h6 may be the first peanut allergens that individuals are exposed to upon ingesting peanut or peanut-containing foods.
Collectively, our data suggest that Ara h2 and Ara h6 are highly bio-accessible in the mouth and esophagus and may trigger immunological effects in the oro-esophageal mucosa. The difference in solubility between Ara h2 and Ara h6 on the one hand, and Ara h1 and Ara h3 on the other, is not due to different isoelectric points, because all four allergens have isoelectric points between 5 and 6. The solubility difference may be determined by their molecular weight and quaternary structure. Ara h2 and Ara h6 are small (15-20 kDa) monomeric proteins [36,37], while Ara h1 and Ara h3 are larger and form complexes up to 180-700 kDa and 360-380 kDa, respectively [38][39][40]. Upon roasting, Ara h1 and Ara h3 may aggregate, further limiting their solubility, while this is not the case for Ara h2 and Ara h6 [41]. For Ara h 1, denaturing media such as high-molarity urea are needed for optimal solubilization [31]. Indeed, when we suspended the undissolved material in a strongly denaturing buffer (SDS-containing buffer with a reducing agent) for SDS-PAGE analysis, we recovered Ara h1 and Ara h3 bands as well ( Figure 6). These remarkable differences in size and quaternary organization of peanut allergens may partially explain why Ara h2 and Ara h6 are more soluble than Ara h1 and Ara h3.
A limitation of our study is that we have not investigated if the solubility of peanut proteins in the mouth and esophagus indeed leads to enhanced sensitization potency. Investigating this would require in vivo experiments, for example using a mouse model and applying oral sensitization [42], but this was beyond the scope of the current work.
Ara h2 and Ara h6 are the most potent peanut allergens [43,44], and the presence of IgE to these two allergens is the best predictor of clinical peanut allergy [45,46]. The vast majority of peanut-allergic patients have IgE to Ara h2 or Ara h6 or both [34]. In part, the potency of Ara h2 and Ara h6 may be explained by their resistance to digestion in the human gastrointestinal tract. It has been shown that Ara h2 and Ara h6 are more resistant against digestion with pepsin and trypsin than Ara h1 and Ara h3 [47,48] and survive the gastrointestinal conditions relatively unaffected [49]. Our data showing that Ara h2 and Ara h6 are highly bio-accessible in the mouth (and esophagus) suggest that these allergens have the unique opportunity to interact with the mucosal immune system before reaching the stomach, which may lead to different sensitizing potential in comparison to other peanut allergens. This may provide an additional explanation for the extraordinary allergenicity of Ara h2 and Ara h6 compared to other peanut allergens.

Conclusions
We show that allergens Ara h2 and Ara h6 are quickly released from peanut matrix at saliva conditions. This makes them bio-accessible in the mouth and esophagus, and enable them to interact with the oro-esophageal mucosal immune system, potentially explaining their extraordinary allergenicity compared to other peanut allergens.