A Highly Sensitive Method for the Detection of Hydrolyzed Gluten in Beer Samples Using LFIA

Most gluten analysis methods have been developed to detect intact gluten, but they have shown limitations in certain foods and beverages in which gluten proteins are hydrolyzed. Methods based on G12/A1 moAbs detect the sequences of gluten immunogenic peptides (GIP), which are the main contributors to the immune response of celiac disease (CD). Immunogenic sequences with tandem epitopes for G12/A1 have been found in beers with <20 mg/kg gluten, which could be consumed by CD patients according to the Codex Alimentarius. Therefore, an accurate method for the estimation of the immunogenicity of a beer is to use two moAbs that can recognize celiac T cell epitopes comprising most of the immunogenic response. Here, a specific and sensitive method based on G12/A1 LFIA was developed to detect GIP in beers labeled gluten-free or with low gluten content, with an LOD of 0.5 mg/kg. A total of 107 beers were analyzed, of those 6.5% showed levels higher than 20 mg/kg gluten and 29% showed levels above the LOD. In addition, G12/A1 LFIA detected gluten in 15 more beer samples than competitive ELISA with another antibody. Despite their labeling, these beers contained GIP which may cause symptoms and/or intestinal damage in CD patients.


Introduction
Gluten refers to a complex set of storage proteins present in the endosperm of cereals, such as wheat, rye, barley, oat, and their derivatives. It comprises two protein fractions: prolamins (monomeric and alcohol-soluble proteins) and glutelins (polymeric and alcohol-insoluble proteins) [1,2]. Prolamins and glutelins are responsible for the functional properties of cereal dough and are characterized by a high content of proline and which makes them resistant to gastrointestinal digestion and encouraging their deamidation by tissue transglutaminase. Consequently, the consumption of these proteins causes adverse reactions in people suffering from gluten-related disorders (GRDs), such as wheat allergy, non-celiac gluten sensitivity (NCGS), dermatitis herpetiformis, gluten ataxia, and celiac disease (CD) [3][4][5][6].
Among GRDs, CD has been extensively studied, and the role of gluten in its pathogenesis has been clearly identified. It is a systemic process that produces chronic enteropathy of the small intestine, and it develops through an inadequate immune response to gluten in genetically predisposed individuals [7][8][9][10]. Currently, the only effective treatment is a gluten-free diet (GFD) that relies on the consumption of naturally gluten-free foods, such as animal-based products, fruits, vegetables, legumes, and nuts, as well as gluten-free dietary products that do not contain more than 20 mg/kg of gluten according to the Codex Alimentarius [11], including beer and certain other beverages [12][13][14].

Spiked Samples
Different beer samples with gluten concentrations below the LOD were spiked with PTW-Cubus (0.5 mg/kg PTW-Cubus) to investigate a possible matrix effect and evaluate the recovery.

Techniques Employed 2.4.1. Development of a New Method Lateral Flow Immunochromatographic Assays (LFIAs)
A G12/A1 moAbs strip was used to develop an analytical method based on the reaction of gluten peptides present in beer samples with colored G12 and A1 moAbs already fixed in the absorption zone. Gluten peptides present in beer samples were bound to the antibody labeled with the detection signal during membrane migration to form a labeled antigen-antibody complex. This complex flowed along the membrane to form a sandwich-like immune complex with the captured moAb immobilized on the test line via the free epitope of the analyte. The result was positive, if a red/pink line appeared in the result zone of the membrane, providing a signal. The absence of this line and therefore a signal below the established limit suggested a negative result. The blue control line was always used as a test control.

Analysis of Samples and Data
The beer samples and reagents were kept at room temperature (15-25 • C). The samples were then diluted at 1:50, using 20 µL of the beer sample and 980 µL of the dilution solution. Subsequently, 100 µL of the diluted sample were added to the shells of the strips and incubated for 30 min. Finally, the strips were ready for visual analysis. The peak area of the immunochromatographic strip was determined using a GlutenTox ® Reader (Hygiena, Seville, Spain) ( Figure 1). GlutenTox ® Reader (Hygiena, Seville, Spain) integrates electronics and LFIA Studio software (Hygiena, Seville, Spain) for data processing [44]. The peak area values for each sample and the control were obtained using this software. Spreadsheets were used to compile raw data, and the results were considered positive when the reading of the test line was above the negative sample value plus three times the standard deviation. If this criterion was not met, the test would yield a negative result. The values obtained using the G12/A1 LFIA performed in duplicate for each sample. The results were semiquantitative and expressed in the unit of mg/kg according to the minimum quantification dilution detected.  GlutenTox ® Reader (Hygiena, Seville, Spain) integrates electronics and LFIA Stud software (Hygiena, Seville, Spain) for data processing [44]. The peak area values for ea sample and the control were obtained using this software. Spreadsheets were used to co pile raw data, and the results were considered positive when the reading of the test li was above the negative sample value plus three times the standard deviation. If this c terion was not met, the test would yield a negative result. The values obtained using t G12/A1 LFIA performed in duplicate for each sample. The results were semi-quantitati and expressed in the unit of mg/kg according to the minimum quantification dilution d tected.

Enzyme-Linked Immunosorbent Assay (ELISA)
Gluten content in beer was determined using the ELISA RIDASCREEN ® Gliad competitive R5 moAb (R7021, R-Biopharm, Darmstadt, Germany). All the samples we analyzed in duplicates according to the manufacturer's instructions. The obtained da were expressed as the mean and standard deviation.

Statistical Anlysis
Relative affinity curves were obtained by plotting the percent maximum absorban against the reference standard concentration (ng/mL). The Sigma Plot, version 12.0, Sys Software, Inc., software package (San José, CA, USA) was used to calculate the IC50 a cross-reactivity (CR) for each standard. The IC50 was defined as the concentration of t line that reduced the peak absorbance by 50% in the assay. Cross-reactivity (CR) was c culated as (IC50 of the standard with the highest affinity for the antibody/IC50 of each test standard) × 100.

Validation of the New Analytical Method
The proposed method was evaluated according to the Eurachem guidelines based studies of sensitivity, specificity, suitability, precision, and robustness [43]. Similar te have already been validated for the detection of gluten in food and beverages (with no hydrolyzed gluten) with different compositions and levels of processing, from raw ma rials to processed food [39]. In this study, the method was validated for beer hydrolysat 3.1.1. Immunochromatographic Strip Sensitivity and the LOD The sensitivity was determined based on the characteristics of the immunochrom tographic strips and the batch-to-batch effect. It was calculated using spikes from 0 to Gluten content in beer was determined using the ELISA RIDASCREEN ® Gliadin competitive R5 moAb (R7021, R-Biopharm, Darmstadt, Germany). All the samples were analyzed in duplicates according to the manufacturer's instructions. The obtained data were expressed as the mean and standard deviation.

Statistical Anlysis
Relative affinity curves were obtained by plotting the percent maximum absorbance against the reference standard concentration (ng/mL). The Sigma Plot, version 12.0, Systat Software, Inc., software package (San José, CA, USA) was used to calculate the IC 50 and cross-reactivity (CR) for each standard. The IC 50 was defined as the concentration of the line that reduced the peak absorbance by 50% in the assay. Cross-reactivity (CR) was calculated as (IC 50 of the standard with the highest affinity for the antibody/IC 50 of each tested standard) × 100.

Validation of the New Analytical Method
The proposed method was evaluated according to the Eurachem guidelines based on studies of sensitivity, specificity, suitability, precision, and robustness [43]. Similar tests have already been validated for the detection of gluten in food and beverages (with nonhydrolyzed gluten) with different compositions and levels of processing, from raw materials to processed food [39]. In this study, the method was validated for beer hydrolysates.

Immunochromatographic Strip Sensitivity and the LOD
The sensitivity was determined based on the characteristics of the immunochromatographic strips and the batch-to-batch effect. It was calculated using spikes from 0 to 25 ng/mL PTW-Cubus in a dilution solution with visual measurements after 30 min. These assays were performed in triplicate (analysis I-II-III). All three assays with spikes of 5-25 ng/mL of PTW-Cubus were visually positive; however, only one of the three assays showed positive results with spikes of 2.5 ng/mL PTW-Cubus, and all the assays showed negative results when containing <2.5 ng/mL PTW-Cubus (Table S1).
Based on the sensitivity of PTW-Cubus, the LOD of the method was determined as gluten (mg/kg) = [S × FE × DF × 2]/1000, where S is the calculated sensitivity; DF is the dilution factor; FE is the extraction factor equaling to 1; gliadin-to-gluten ratio is 2. A sensitivity of 5 ng/mL PTW-Cubus was established for reading at 30 min and test dilution of 1:50, resulting in an LOD of 0.5 mg/kg for gluten.

Strip Reader Detection Capacity
To determine the detection capacity of the reader used in the test, the measurements obtained in the analysis of the beer samples (<LOD) were compared with the measurements of the same beer samples spiked with 0.5 mg/kg PTW-Cubus.
First, the average peak area value offered by the reader was established, when five beer samples with the gluten content lower than the LOD were analyzed five times each. In most of the validation protocols reported in the literature, the verification of the method [45] was calculated by executing at least five determinations per concentration and a deviation below 15% or 20% of the expected value [46] as the acceptance criterion (excluding negative samples) [47]. The results showed a negative visual interpretation in all strips. However, the reader detected a signal in 8 of the 25 samples analyzed, with a mean peak area of 35.1 ± 9.7 (Table S2). Based on these results, a minimum peak area of approximately 70 was established to consider a sample as positive. This was calculated by adding three times the deviation to the value of the mean peak area (35.1) [43,48].
Furthermore, the validity of this gluten-detection method was determined using 0.5 mg/kg of PTW-Cubus spiked in four samples (<LOD) five times each. All the strips were positive on visual interpretation, and the results provided by the strip reader established a mean peak area of 256.8 ± 26.5 in these samples (Table 1). Once the average results of the peak area and standard deviations were obtained using the strip reader, the range of reading variation was calculated as a function of the coefficient of variation (σ). Furthermore, confidence intervals around the mean were determined, with confidence levels of 99.7% (peak area: 283.3-230.3), 95% (peak area: 309.8-203.9), and 68% (peak area: 336.2-177.7). Finally, the range ± 2σ was established as the criterion for considering a result as valid, since a 95% coverage was observed in that range ( Figure 2) [49]. coefficient of variation (σ). Furthermore, confidence intervals around the mean were determined, with confidence levels of 99.7% (peak area: 283.3-230.3), 95% (peak area: 309.8-203.9), and 68% (peak area: 336.2-177.7). Finally, the range ± 2σ was established as the criterion for considering a result as valid, since a 95% coverage was observed in that range ( Figure 2) [49].

Accuracy and Precision
According to the Eurachem guide, accuracy and precision indicate the ability of the method to obtain results within the established range of interest. According to these guidelines, the results must be obtained from an average of 6 to 15 replicates of each material, with the same equipment, analyst, and laboratory and in a short period [43].
In this study, accuracy was calculated by taking 18 measurements of 2 beer samples (<LOD) spiked with 0.5 mg/kg PTW-Cubus. These 18 measurements were performed on the same day and from 3 different extractions, each analyzed in triplicates ( Table 2). Meanwhile, precision was achieved by performing 36 measurements of 2 beer samples (<LOD) spiked with 0.5 mg/kg PTW-Cubus. These 36 measurements were carried out on 2 different days, and each measurement was analyzed in triplicates.
All the strips were positive on visual interpretation. Of the 18 replicates analyzed to calculate the accuracy, 16 were within the range (peak area: 203.9-309.8), indicating an accuracy of 88.9% on day 1 and 100% on day 2. In addition, 34 of the 36 replicates used to calculate the precision were within the established range (peak area: 203.9-309.8), indicating a precision of 94.5% (Table 2).

Accuracy and Precision
According to the Eurachem guide, accuracy and precision indicate the ability of the method to obtain results within the established range of interest. According to these guidelines, the results must be obtained from an average of 6 to 15 replicates of each material, with the same equipment, analyst, and laboratory and in a short period [43].
In this study, accuracy was calculated by taking 18 measurements of 2 beer samples (<LOD) spiked with 0.5 mg/kg PTW-Cubus. These 18 measurements were performed on the same day and from 3 different extractions, each analyzed in triplicates (Table 2). Meanwhile, precision was achieved by performing 36 measurements of 2 beer samples (<LOD) spiked with 0.5 mg/kg PTW-Cubus. These 36 measurements were carried out on 2 different days, and each measurement was analyzed in triplicates. All the strips were positive on visual interpretation. Of the 18 replicates analyzed to calculate the accuracy, 16 were within the range (peak area: 203.9-309.8), indicating an accuracy of 88.9% on day 1 and 100% on day 2. In addition, 34 of the 36 replicates used to calculate the precision were within the established range (peak area: 203.9-309.8), indicating a precision of 94.5% (Table 2).

Determination of the Relative Affinity from Different Reference Standards
To confirm the detection capacity of G12/A1 LFIA against the most widely used reference standards in immunological assays, we tested 33-mer peptide, PWG-gliadin, and PTW-Cubus (hydrolyzed wheat prolamin). The affinity of moAbs for the antigens was quantified by calculating the concentration of the antigen, resulting in a 50% reduction in the peak signal in the LFIA (IC 50 ) and CR. The results were compared with those obtained from R5 competitive ELISA.
The results obtained using the R5 competitive ELISA demonstrated that this antibody was able to detect PWG-gliadin and gliadin PTW-Cubus, although with lower affinity than G12/A1 moAb, and the 33-mer peptide was not detected using moAb R5 ( Figure 3B). Different research groups have shown that the α-gliadin 33-mer is poorly identified with R5 competitive ELISA [23,50,51]. By contrast, analysis of T-cell reactivity and enzyme protein detoxification showed that the signals of G12 and A1 moAbs correlate with the potential toxicity of the sample for patients with CD [23,32,[52][53][54][55]. The recognition profiles of G12 and A1 moAbs have been extensively characterized [23,52,53]. Moreno et al. [56] showed that most GIP (responsible for 80-95% of celiac T-cell immunoreactivity) react with G12 moAb. Different research groups have shown that the α-gliadin 33-mer is poorly identified with R5 competitive ELISA [23,50,51]. By contrast, analysis of T-cell reactivity and enzyme protein detoxification showed that the signals of G12 and A1 moAbs correlate with the potential toxicity of the sample for patients with CD [23,32,[52][53][54][55]. The recognition profiles of G12 and A1 moAbs have been extensively characterized [23,52,53]. Moreno et al. [56] showed that most GIP (responsible for 80-95% of celiac T-cell immunoreactivity) react with G12 moAb.

Gluten Analysis in Beer
Gluten analysis was carried out in 107 beers, with 88 labeled gluten-free, 12 low in gluten or with traces of gluten (non-specific labeled), and 7 not labeled for gluten or artisanal. Additionally, we tested two high-gluten beer samples as positive controls. The beer samples were grouped according to their cereal content in their labeling; based on this classification, 65% (70/107) contained cereals with gluten, 16.8% (18/107) contained a mixture of cereals with and without gluten, and only 4.7% (5/107) of the beer samples contained exclusively gluten-free cereals (maize, millet, buckwheat, and rice). According to the type of yeast used in the preparation, 56.8% (49/107) were made with lager yeast, and 43% (46/107) were prepared with ale yeast (Table 3).  50 and CR were obtained for G12/A1 LFIA and R5 competitive ELISA. The IC 50 is defined as the concentration of the line that reduces the peak absorbance by 50% in the assay. These assays were performed in duplicates. In the figure the gray points correspond to PWG-gliadin; the green points, PTW-Cubus; and the red dots, 33-mer peptide. LFIA, lateral flow immunochromatographic assays; N.A., not applicable.

Gluten Analysis in Beer
Gluten analysis was carried out in 107 beers, with 88 labeled gluten-free, 12 low in gluten or with traces of gluten (non-specific labeled), and 7 not labeled for gluten or artisanal. Additionally, we tested two high-gluten beer samples as positive controls. The beer samples were grouped according to their cereal content in their labeling; based on this classification, 65% (70/107) contained cereals with gluten, 16.8% (18/107) contained a mixture of cereals with and without gluten, and only 4.7% (5/107) of the beer samples contained exclusively gluten-free cereals (maize, millet, buckwheat, and rice). According to the type of yeast used in the preparation, 56.8% (49/107) were made with lager yeast, and 43% (46/107) were prepared with ale yeast (Table 3).   The gluten content was below the LOD (0.5 mg/kg gluten) in 71% (76/107) of the samples analyzed using G12/A1 LFIA. Our results showed that 29% (31/107) of the analyzed samples were positive in the range of 0.5-256 mg/kg of gluten (Figure 4), although they were labeled as gluten-free or low-gluten. In addition, 6.5% (7/107) contained levels higher than 20 mg/kg of gluten. This could be related to the brewing process, ingredients, the yeast used, and other factors. The addition or absence of malted cereal and the use of clarification processes, such as filtration, centrifugation, or the cold step, can influence the gluten content in beer [19]. Therefore, the inclusion of additional steps in the production process and/or the use of enzymes to hydrolyze gluten (e.g., PEP) could modify the amount of gluten present [34]. It should be noted that 74.2% (23/31) of the analyzed beer samples contained gluten-containing cereals on their labels. Furthermore, the number of beers with a positive result for gluten and ale yeast was higher than the number of those containing lager yeast (51.6% (16/31) and 32.3% (10/31), respectively). The influence of yeast fermentation on gluten content in the final product was previously noted by Tanner et al. [57] and Fernandez-Gil et al. [30], who highlighted the importance of selecting lager yeasts or excluding cereals with gluten in the brewing process to reduce the gluten content of the final product.
Additionally, the results obtained with G12/A1 LFIA were compared with those of R5 competitive ELISA, which revealed that 85% (91/107) of the samples were below the limit of quantification (LOQ) (<10 mg/kg of gluten) and 15% (16/107) showed positive results ranging from 10.8 to 63.1 mg/kg of gluten ( Figure 4). This was a predictable result, as G12/A1 LFIA had a lower LOD than R5 competitive ELISA. Therefore, G12/A1 LFIA detected gluten in 15 more beer samples (beers Q to EE), 73% (11/15) of which contained gluten-containing cereals on their labels. R5 competitive ELISA, which revealed that 85% (91/107) of the samples were below the limit of quantification (LOQ) (<10 mg/kg of gluten) and 15% (16/107) showed positive results ranging from 10.8 to 63.1 mg/kg of gluten ( Figure 4). This was a predictable result, as G12/A1 LFIA had a lower LOD than R5 competitive ELISA. Therefore, G12/A1 LFIA detected gluten in 15 more beer samples (beers Q to EE), 73% (11/15) of which contained gluten-containing cereals on their labels. Notably, nine beers showed values from 10 to 16 mg/kg gluten using R5 competitive ELISA, four of which had a gluten content ranging from 20 to 128 mg/kg according to G12/A1 LFIA; therefore, these beers should not be consumed by patients with CD. Several studies have indicated that R5-based methods can overestimate or underestimate the net immunotoxicity of beer for patients with CD [56,58,59]. Moreover, in August 2020, the US FDA issued a dossier (FDA-2014-N-1021-0560), stating that it does not know "any scientifically valid analytical method that is effective in accurately detecting and quantifying the protein content of gluten in fermented or hydrolyzed foods in terms of equivalent amounts of intact gluten proteins", this includes R5 competitive ELISA. Additionally, this release emphasized the lack of detection of gluten using R5 competitive ELISA, this antibody does not identify all immunogenic amino acid sequences and gluten may be present in a form not detectable with R5 competitive ELISA [60]. Furthermore, a claim stated in technical and marketing documents of R5 immunomethods is their reactivity to the 33- Notably, nine beers showed values from 10 to 16 mg/kg gluten using R5 competitive ELISA, four of which had a gluten content ranging from 20 to 128 mg/kg according to G12/A1 LFIA; therefore, these beers should not be consumed by patients with CD. Several studies have indicated that R5-based methods can overestimate or underestimate the net immunotoxicity of beer for patients with CD [56,58,59]. Moreover, in August 2020, the US FDA issued a dossier (FDA-2014-N-1021-0560), stating that it does not know "any scientifically valid analytical method that is effective in accurately detecting and quantifying the protein content of gluten in fermented or hydrolyzed foods in terms of equivalent amounts of intact gluten proteins", this includes R5 competitive ELISA. Additionally, this release emphasized the lack of detection of gluten using R5 competitive ELISA, this antibody does not identify all immunogenic amino acid sequences and gluten may be present in a form not detectable with R5 competitive ELISA [60]. Furthermore, a claim stated in technical and marketing documents of R5 immunomethods is their reactivity to the 33-mer α-gliadin. However, in this study, the results revealed that R5 did not react with this peptide (Figure 3).
Several patients with CD experience adverse symptoms after ingesting gluten-free labeled beers, possibly because gluten proteins in beer are hydrolyzed, making GIP more readily bioavailable to trigger the disease to a much greater extent than undigested gluten [24]. These traces of gluten may not be detected using certain analytical methods, but they pose a risk to patients with CD because they preserve their toxicity. If a sufficient amount is ingested, which is likely to be in a drink, it can induce an immune response in the intestinal epithelium of celiac patients. Therefore, it is necessary to review the current gluten-free legislation of the European Union (EU), published in 2009 and regulated in 2012, which specifies two subgroups: gluten-free (≤20 mg/kg) and low-gluten content (21-100 mg/kg) [61]. In addition, lower local limits of gluten have already been established in many countries, ranging from 20 mg/kg in Spain, Italy, the UK [11], Canada, and the USA to 10 mg/kg of gluten in Argentina and 3 mg/kg of gluten in Australia, New Zealand, and Chile [62,63]. Consequently, this more sensitive method was developed to detect gluten concentrations of up to 0.5 mg/kg in beer, so that it can be safely consumed by patients with CD without causing symptoms or intestinal damage.

Conclusions
In this study, a highly sensitive and easy-to-use method based on LFIA was developed that requires no extraction and has an LOD of 0.5 mg/kg of gluten. Furthermore, it can measure the level of bioavailable GIP in hydrolyzed foods. Our results demonstrated that, although different regulations allow beer products to contain less than 20 mg/kg of gluten to be considered gluten-free, they still contain GIP, which should not be consumed by patients with CD. Consequently, we recommend reducing the limit of gluten-free labeling of beers according to the availability of sensitive methods, such as the one described in this study, that can detect bioavailable gluten.