*Enrique Méndez deceased. Received 28 June 2007 Accepted 18 September 2007

Transcription

1 Original article 545 Measurement of wheat gluten and barley hordeins in contaminated oats from Europe, the United States and Canada by Sandwich R5 ELISA Alberto Hernando, Jorge R. Mujico, María C. Mena, Manuel Lombardía and Enrique Méndez * Objectives We have investigated the extent of contamination with wheat, barley, rye or a mixture of these cereals in a large number of grains and commercial oats. We have also attempted to identify the type of cereal contaminant. Methods Sandwich R5 ELISA (using either gliadins or hordeins as standards), western blot, matrix-assisted laser desorption/ionization time-of-flight mass spectrometric and quantitative real-time PCR (Q-PCR) techniques have been used to analyze a total of 134 oats, comprising grains and commercial oat products collected from Europe, the United States and Canada. Results Twenty-five of the 134 pure, uncontaminated oat varieties were found to have undetectable levels of gluten, whereas most of the 109 grains and commercial oat products were mainly contaminated with mixtures of wheat, barley and rye, barley being the predominant contaminant. The percentages of these cereals in the oat samples have been calculated by specific wheat, barley and rye Q-PCR systems. The oat samples were grouped according to the avenin spectra determined by the mass spectrometric technique. The data confirmed that contaminated oat foods, based on the same variety, could have different levels of wheat, barley and rye contamination. Conclusion This study has verified that contamination with wheat gliadins or barley hordeins in oat samples can be measured by the Sandwich R5 ELISA, using either gliadins or hordeins as standards, and also the importance of using confirmatory techniques (such as western blot, Q-PCR and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry) to confirm that most oats are contaminated with mixtures of wheat, barley and rye. Eur J Gastroenterol Hepatol 20: c 2008 Wolters Kluwer Health Lippincott Williams & Wilkins. European Journal of Gastroenterology & Hepatology 2008, 20: Keywords: coeliac disease, gluten, oats, Sandwich R5 ELISA National Center of Biotechnology (Gluten Unit), Madrid, Spain Correspondence to Alberto Hernando, National Center of Biotechnology (Gluten Unit), C/Darwin 3, Cantoblanco, 28049, Madrid, Spain Tel: ; fax: ; ahernand@cnb.uam.es *Enrique Méndez deceased. Received 28 June 2007 Accepted 18 September 2007 Introduction Most in-vivo clinical studies have confirmed the safety of pure, uncontaminated oats in moderate amounts for persons with coeliac disease [1 11] even though some studies support that oats could be potentially harmful [12 14]. In this way, the Codex Alimentarius still considers oats as coeliac toxic. Evidence of gluten contamination in oats-based foods, such as rolled oats, was reported long ago [15]. Since the recent availability of the Sandwich R5 ELISA, which quantifies prolamins from wheat (gliadins), barley (hordeins) and rye (secalins), a large number of studies confirming gluten contamination in most commercial oats [16 18] have been reported. R5 antibody, which recognizes hordeins, formerly undetectable by the conventional barley-insensitive o-elisa [19], is becoming the most accepted method of measuring contamination of gluten in commercial oats-based foods [16 18], as well as starch-based foods [20] and pure, uncontaminated oats X c 2008 Wolters Kluwer Health Lippincott Williams & Wilkins [10,11] for clinical and follow-up studies. Sandwich R5 ELISA represents an important tool in gluten analysis, in particular to control gluten in oats highly contaminated with barley, in countries where oats have been permitted for persons with coeliac disease and where oats-based foods are still being analyzed with the former barleyinsensitive o-elisa. The Sandwich R5 ELISA was validated in a collaborative trial conducted by the Prolamin Working Group [21] and has been recently endorsed as a type I method at the 27th session of the Codex Committee on Methods of Analysis and Sampling [22]. In addition, the US Food and Drug Administration in its proposed ruling on the gluten-free labelling of foods is tentatively considering using Sandwich R5 ELISA [23]. Recently, however, it has been described that Sandwich R5 ELISA overestimates hordeins in barley-contaminated oats, unless a hordein standard is used [24].

2 546 European Journal of Gastroenterology & Hepatology 2008, Vol 20 No 6 The aims of this study are to quantify by Sandwich R5 ELISA (using either gliadins or hordeins as standards) the amount of wheat gluten and barley hordeins in a large number of oat foods from Europe, the United States and Canada, to determine if the R5 antibody overestimates barley contamination, and to confirm the type of cereal contaminant by using complementary techniques such as western blot, quantitative real-time PCR (Q-PCR) and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. Methods Reagents Acetonitrile and trifluoroacetic acid were purchased from Merck (Darmstadt, Germany); ethanol came from Scharlau (Barcelona, Spain); sinapinic acid (trans-3,5- dimethoxy-4-hydroxycinnamic) and octyl-b-d-glucopyranoside from Fluka (Buchs, Switzerland) and acetic acid glacial from Panreac Química S.A. (Barcelona, Spain). Ultrapure water from a Milli-Q purification system (Millipore, Bedford, USA) was used in the preparation of all solutions. Oat samples Commercial oat foods (rolled oats, oat flakes, oatmeals, oat flours, oat brans and oat grains) were collected from different countries in Europe, the United States and Canada: the Paediatric Research Centre, University of Tampere (Finland); Arla Foods (Sweden); the Department of Food Technology, Tampere (Finland); Leipzig University Hospital (Germany); the Department of Medicine, Rikshospitalet, Oslo (Norway); the Food Inspection Service (Holland); St James s Hospital (Ireland); the Food Research and Development Centre (Canada); Peter Kolln Company (Germany); Calgary University (Canada) and Agriculture and Agri-Food (Canada). Twenty-five known pure oat varieties (AC1, Cannele, Kantora, Adamo, Brawi, Mojacar, Kankan, Orblanche, Cobeña, Kazmina, Fuwi, Fringante, Prevision, Acebeda, Araceli, Patones, Anchuela, Aintree, Hamel, Norlys, Karmela, Condor, Cory, Caleche and Mirabel) were collected from Agrusa Company and the University of Alcalá de Henares (Spain). Oat sample preparation About 50 g of dried oat food sample were milled in a coffee mill for around min to obtain a fine homogenized powder. Aqueous ethanol and cocktail extraction Milled, dried oat sample (0.25 g) was weighed, transferred to a propylene tube and then extracted with 10 ml of 60% aqueous ethanol or cocktail solution (based on reducing and disaggregating reagents in phosphatebuffered saline as previously described [25]). Preparation of the hordein standard Flours (1 g) from four barley cultivars (Tabaiba, Clarine, Icare and Cameo) was incubated with 40 ml of 0.5 mol/l NaCl. The soluble albumins and globulins were first removed by centrifugation and the insoluble hordein fraction was extracted with the cocktail solution. The protein contents of these extracts were 7.9, 7.4, 7.1 and 5.7%, respectively, as determined by an assay based on the Lowry method [26] using gliadins as the standard. The hordein standard was prepared by mixing the same volume of the four hordein extracts, previously adjusted to 0.5 mg/ml. Maize flours spiked with barley Samples contaminated with a known amount of barley were prepared by mixing 10 mg of barley flours (cultivars Tabaiba, Clarine, Icare and Cameo) with 5.0 g of an uncontaminated maize flour. The estimated hordein contents of these samples were 158, 147, 141 and 114 ppm, respectively (taking into account the hordein contents of the barley flours, as determined by the Lowry method). Sandwich R5 ELISA We used a homemade Sandwich R5 ELISA based on the unique monoclonal antibody R5 [16,21]. This was used both as a coating antibody and conjugated to horseradish peroxidase for the analysis of prolamins in oat food samples, with a gliadin quantification limit of 1.5 ppm. In this study, we have used the Working Group on Prolamin Analysis and Toxicity (PWG) gliadins [27] as well as a pool of hordeins as standards. Western blot After one-dimensional sodium dodecyl sulphate polyacrylamide gel electrophoresis, proteins were electrotransferred onto polyvinylidene difluoride membranes, incubated directly with R5-HRP and developed by enhanced chemiluminescence immunodetection (Amersham Pharmacia, Buckinghamshire, UK) as previously described [16]. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry To 10 ml of the avenin aqueous ethanol extracts were added 8 ml of the calibrate solution [2 ml of a-sarcin (2 mg/ml) in 50 ml of bovine serum albumin (100 pmol/ml)] and 25 ml of saturated sinapinic acid [3,5-dimethoxy- 4-hydroxycinnamic acid] in 30% (v/v) aqueous acetonitrile containing 0.1% (v/v) trifluoroacetic acid used as a matrix solution. A 1.5 ml volume of sample matrix mixture was placed onto a 100 sample stainless steel plate and allowed to dry at room temperature for 5 min.

3 Measurement of wheat and barley contamination in oats by Sandwich R5 ELISA Hernando et al. 547 Samples were measured on a PE Biosystems (Foster City, California, USA) MALDI-TOF Voyager DE-PRO in standard configuration. Mass spectra were recorded in the linear positive mode at 25-kV acceleration voltage and a grid voltage of 93%, 0.25% of guide wire and 700 ns of delay time by accumulating 100 spectra of single laser shots under threshold irradiance [28]. Only highly intense, well-resolved mass signals arising from 3 5 selected target spots were considered. The equipment was internally calibrated using the single charged signal of a-sarcin and doubly charged signal from bovine serum albumin with molecular masses of and Da, respectively. DNA extractions DNA from oat foods was extracted by a modified sodium dodecyl sulphate/proteinase K/guanidinium hydrochloride method as described [29]. Quantitative real-time PCR Q-PCR systems for either individual or simultaneous detection of wheat, barley and rye DNA, and the analysis of the amplification products by agarose gel electrophoresis, were used as described [29 31]. Results Analysis of maize flours spiked with barley by Sandwich R5 ELISA, using gliadins or hordeins as standards Table 1 shows the results of the analysis by Sandwich R5 ELISA (using the PWG gliadins or hordeins as standards) of maize flours spiked with four different barley varieties. The cocktail solution extracts the total hordeins. When these are estimated using hordeins as the standard, the expected and calculated results are nearly identical. When PWG gliadins are used as the standard, however, the calculated results appear to be overestimated by a factor of around 2.4. Wheat gluten is composed of the monomeric a, b/g and o gliadins plus the polymeric high and low molecular weight glutenins fractions. The gliadin and glutenin fractions were traditionally considered to have a relative ratio Table 1 Levels of hordeins and gluten (in ppm) of maize flours spiked with four barley varieties and analyzed by the Sandwich R5 ELISA, using either hordeins or PWG gliadins as standards Barley varieties Expected hordeins Hordeins a Gluten b Tabalba (0.7) c 258 (1.6) c Clarine (1.2) 442 (3.0) Icare (1.1) 352 (2.5) Cameo (1.0) 282 (2.5) Mean (1.0) (2.4) a Monomeric C and g bordeins and polymeric B hordeins, determined using hordeins as standard. b Monomeric C and g hordeins and polymeric B hordeins, determined using PWG gliadins as standard and multiplied by two. c Ratio: calculated/expected. of 1 : 1 [32] but this is controversial [33]. Although the cocktail solution extracts whole gluten proteins, glutenins are either slightly or not recognized by the R5 antibody, probably because the number of epitopes in glutenins necessary to react with the R5 antibody is very small as compared with gliadins [25]. Therefore, the gliadin values were multiplied by two, to calculate the gluten values. Nevertheless, in the case of barley, the total monomeric (C and g) as well as the polymeric (B) hordeins are recognized by the R5 antibody. As these proteins constitute around 97% of barley gluten [34], the values obtained by the Sandwich R5 ELISA do not need to be multiplied by two and the overestimation factor would be actually around 1.2, which is near to the experimental error of the ELISA technique. These results demonstrate that there are no differences in the measurement of total hordeins in samples contaminated with barley, extracted with cocktail solution and analyzed by the Sandwich R5 ELISA, using either gliadins or hordeins as standards. Analysis by Sandwich R5 ELISA of uncontaminated oat grains from different varieties The analyses of the 25 pure oats of known varieties by the Sandwich R5 ELISA (using the PWG gliadins as standard) confirm the fact that they have undetectable levels of wheat, barley and rye prolamins, below the quantification limit of the technique ( < 1.5 ppm). Analysis of grains and commercial oat foods by Sandwich R5 ELISA Table 2 show the analyses of 109 commercial oat foods, extracted with cocktail solution and analyzed by Sandwich R5 ELISA, using either gliadins or hordeins as standards, and exhibiting a massive contamination of either gluten or total hordeins. Taking into account that the ELISA technique cannot discriminate the type of cereal contaminant, it is impossible to know before testing if the oat samples are contaminated with wheat, barley, rye or mixtures of these cereals. To solve this problem, most oat samples were analyzed by different Q-PCR systems, specific for wheat, barley and rye DNA. The results indicated that most oat samples were contaminated with barley, either alone or mixed with wheat in different percentages. Only a few samples contained mixtures of wheat, barley and rye. These results demonstrate the necessity of using alternative techniques, such as Q-PCR, to resolve these situations. In accordance with the results obtained by the Sandwich R5 ELISA, using the PWG gliadin standard, 25% of oats

4 548 European Journal of Gastroenterology & Hepatology 2008, Vol 20 No 6 Table 2 Levels of gluten and barley total hordeins (in ppm) of oat food samples analyzed by the Sandwich R5 ELISA, using either PWG gliadins or hordeins as standards N Gluten a Total hordeins b Barley DNA (%) c V 1 1 < 3 < < 3 < * * * * V 2 25 <3 < <3 < < 3 < V 3 32 <3 < <3 < <3 <1.5 V * V * d <3 < V 6 48 <3 <1.5 0* 49 <3 <1.5 0* 50 <3 <1.5 0* V V 8 53 <3 < <3 < <3 < d <3 < d <3 <1.5 V 9 58 <3 <1.5 0 V <3 <1.5 0* V <3 < V <3 < <3 < < * * Table 2 (continued ) N Gluten a Total hordeins b Barley DNA (%) c V V V V < 3 < * V < 3 < V < 3 < 1.5 0* 92 < 3 < V V < 3 < V V V V < 3 < < 3 < 1.5 V < 3 < 1.5 a Gluten: determined using PWG gliadins as standard and multiplied by two. b Total hordeins: determined using hordeins as standard. c The percentages are barley DNA with respect to wheat/barley DNA (excluding the samples marked with asterisks, which are with respect to wheat/barley/rye DNA). d Grains washed with 60% aqueous ethanol. showed levels of gluten < 3 ppm, whereas 4% showed levels between 3 and 20 ppm, 12% between 20 and 200 ppm and 59% exceeded the proposed Codex level of 200 ppm [35] (Table 2). On the other hand, using the hordein standard, the percentages were as follows: 27% showed levels of hordeins < 1.5 ppm, whereas 6% showed levels between 1.5 and 20 ppm, 26% between 20 and 200 ppm and 41% exceeded the proposed Codex level of 200 ppm [35] (Table 2). It is interesting to mention that in some oat grains, the cereal contamination seems to be located in powder form on the surface of the grains, as such contamination can be removed simply by washing the grains with 60% aqueous ethanol as in the case of contaminated samples 45, 56 and 57, with levels of gluten of 218, 574 and 1592 ppm, respectively. After washing with 60% ethanol, the samples (now called 45 d,56 d and 57 d ) showed undetectable levels of contamination (Table 2).

5 Measurement of wheat and barley contamination in oats by Sandwich R5 ELISA Hernando et al. 549 Fig V V Percentage of intensity V 16 V Mass(m/z) Mass(m/z) Avenin spectra determined by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of the 24 oat food samples included in V 1, 11 in V 12, 6 in V 16 and 6 in V 19. The molecular weights of selected avenin mass signals are indicated, between the range 19 and 31 kda. Identification of avenin varieties in oat samples by mass spectrometry To identify the type of oat varieties, all oat samples were analyzed by MALDI-TOF mass spectrometry [36]. The 25 oat uncontaminated grains of known varieties displayed different avenin mass profiles, ranging from 20 to 30 kda. The mass spectra of the 109 commercial oat foods also displayed the typical avenin profile, in the range between 20 and 30 kda (data not shown). A total of 37 different avenin mass signals were unequivocally identified, and we have tentatively called these varieties V 1 to V 37. Figure 1 shows the mass profiles of the varieties V 1,V 12,V 16 and V 19. The avenin mass spectrum of the pure oat variety AC1 was identical to the spectra of oat foods from the V 2 group, Orblanche to V 5, Cannele to V 7, Prevision to V 8, Cobeña tov 9, Kazmina to V 11, Fuwi to V 14 and Fringante to V 22. Oat foods belonging to V 1,V 4,V 6,V 17,V 21 and V 23 came from Finland; V 2 from Norway; V 3 from Germany; V 5 from Finland and Spain; V 7,V 15 and V 20 from Ireland; V 8,V 9, V 18 and V 22 from Spain; V 10 from Sweden; V 11 from Ireland and Spain; V 12 from the United States; V 13,V 24 and V 25 from Canada; V 14 from Ireland and Holland; V 16 from Ireland and Germany and, finally, V 19 came from Finland, Holland and Germany. Western blot analyses of oat samples Figures 2 and 3 show the western blots of 82 selected oat food samples and 17 pure, uncontaminated oats of known varieties. Western blot analyses demonstrate that in all contaminated oat foods (as determined by Sandwich R5 ELISA) the immunoreactive bands appear only in the region from 30 to 50 kda, which corresponds to gliadins, hordeins and secalins (Fig. 2). Similarly, no immunoreactive bands were detected in the avenin region, from 20 to 30 kda, either in the pure, uncontaminated oat grains (Fig. 3) or in the commercial oat food samples with undetectable levels of gluten (Fig. 2). Thus, crossreactivity of the R5 antibody with the oat avenins can be ruled out. It is interesting that in oat samples based on the same variety, as in V 1,V 2,V 4,V 5,V 12,V 14,V 16 and V 19 (Fig. 2), the immunoreactive band patterns in the gliadin, hordein and secalin region are very different, indicating that these oat products must be contaminated with either a single type of grain or mixtures of wheat, barley and rye. Only oat foods in the V 13 showed identical immunoreactive bands, as they came from batches of the same oat flour (Fig. 2). Western blot demonstrated the presence and absence of immunoreactive bands before and after washing the grains of samples 45 and 57 with 60% aqueous ethanol (Fig. 2). Analysis of oat samples by quantitative real-time PCR To confirm the cereal type contaminant, some of the oat food samples were analyzed by Q-PCR, for individual or simultaneous quantification of wheat, barley and rye DNA. Quantitative data of the analyses with the three individual Q-PCR systems, for detecting independently wheat, barley and rye DNA, demonstrated that barley was

6 550 European Journal of Gastroenterology & Hepatology 2008, Vol 20 No 6 Fig. 2 kda 50 Gliadins V 1 Hordeins 30 Secalins 20 Avenins Gliadins Hordeins Secalins V 2 V V 4 V 5 V 6 V 7 V 8 V 9 V 10 V V 10 V 11 V 12 V 13 V 14 V 15 V V 17 V 18 V 19 V 20 V 21 V 22 V 23 V 24 V 25 Western blots developed with R5 monoclonal antibody of 82 selected grains and commercial oat samples, comprising 25 varieties. The locations of gliadin, hordein and secalin immunoreactive bands in these oat samples are indicated in boxes. Numbers correspond to oat samples from Table 2. *The immunoreactive bands of wheat gliadins from Jabato, barley hordeins from Volga and rye secalins from Petkus are displayed. w Washed with 60% aqueous ethanol. Fig. 3 kda 50 Gliadins Hordeins 30 Secalins 20 Avenins Gliadins Hordeins Secalins V 5 V 7 V 9 V 11 V 14 V 22 V 26 Orblanche Cannele Cobena Kazmina Fuwi Fringante Kantora V 27 V 28 V 29 V 30 V 31 V 32 V 33 V 34 V 35 V 36 Adamo Brawi Mojacar Kankan Acebeda Araceli Patones Anchuela Aintree Hamel Western blots developed with R5 monoclonal antibody of 17 pure uncontaminated oat varieties. The locations of gliadin, hordein and secalin immunoreactive bands in these oat samples are indicated in boxes. *The immunoreactive bands of wheat gliadins from Jabato, barley hordeins from Volga and rye secalins from Petkus are displayed. the main contaminant in most of the oat food samples, whereas wheat and rye were also present in lower amounts (Table 2). Furthermore, the analysis by agarose gels of the DNA amplification products from the individual or simultaneous Q-PCR systems indicated that 36 of the 50 selected oat food samples were contaminated with either wheat, barley, rye or mixtures of these cereals (Fig. 4), whereas in 14 oat food samples and 12 oat grains of known varieties no DNA from these cereals was detected (Fig. 4, bottom right).

7 Measurement of wheat and barley contamination in oats by Sandwich R5 ELISA Hernando et al. 551 Fig. 4 MWW Controls V 4 V 23 V B R NTC W B W B R W B R W B W B R W B R B W B W B W B R W B R W B W B V 12 V 16 V 19 MW V 2 V 3 V 5 V 7 V 8 V 6 V 9 V 10 V 11 V 13 # NTC MW # V 14 V 15 V 17 V 18 V 20 V 21 V 22 V 24 V 25 NTC MW # NTC V 2 V 5 V 8 V 9 V 11 V 14 V 26 V 27 V 28 V 29 V 30 V 37 AC1 Orblanche Prevision Cobefia Kazmina Fuwi Kantora Adamo Brawi Mojacar Kankan Agarose gels of wheat, barley and rye quantitative real-time PCR amplification products of 50 oat food samples and 12 pure oat grains of oat varieties. Numbers represent selected oat samples from Table 2. MW: DNA molecular weight marker. Positive controls: 1 ng of standard DNA. NTC: no template control (only reaction buffer) as a negative control. w Washed with 60% aqueous ethanol. Saia 6 In addition, the different mobility of the wheat, barley and rye DNA amplification products, obtained from the three individual Q-PCR systems, enabled us to identify the specific cereal contaminant (wheat, barley or rye) in the selected samples. Accordingly, sample 66 is contaminated only with barley (Fig. 4); samples 10, 38, 69, 70, 98, 100 and 106 are contaminated with a mixture of wheat and barley (Fig. 4) and samples 13, 39, 42, 88 and 89 are contaminated with mixtures of wheat, barley and rye (Fig. 4). Similar analyses have been carried out using the Q-PCR technique with other contaminated oat samples not included in this study, confirming barley as the main contaminant in comparison with wheat or rye (data not shown). Direct identification of barley in the oat sample 106 by Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry Figure 5 illustrates the presence of barley in oat sample 106. The aqueous ethanol extract from oat sample 106 was analyzed by MALDI-TOF mass spectrometry. Direct Fig. 5 Percentage of intensity Avenins Hordeins Mass (m/z) Oat 106 Barley (variety) Matrix-assisted laser desorption/ionization time-of-flight mass spectrum of the contaminated oat sample 106. The avenin (top left) and the hordein (top right) mass signals from the oat sample are displayed. Common hordein mass signals from the contaminated oat sample and the barley variety Astrid are in the box. Molecular weights of selected mass signals are indicated.

8 552 European Journal of Gastroenterology & Hepatology 2008, Vol 20 No 6 observation of the mass spectrum displayed both the typical avenin mass profile in the range from 20 to 30 kda (Fig. 5, top left) as well as hordein mass signals from 40 to 45 kda (Fig. 5, top right). In comparison, a hordein mass profile from an aqueous ethanol extract of a barley variety is also displayed (Fig. 5, bottom). Discussion The results obtained in this study show that most of the 109 oat food samples from Europe, the United States and Canada are highly contaminated with wheat, barley, rye or mixtures of these cereals, some of them exceeding the proposed Codex level of 200 ppm of gluten. The absence of gluten in the pure varieties is in agreement with the fact that avenins lack the QQPFP pentapeptide, which is the specific potential toxic epitope for the R5 monoclonal antibody [37,38]. Despite the fact that Sandwich R5 ELISA detects gliadins, hordeins and secalins in food samples, the technique itself cannot distinguish whether the contamination could be owing to wheat, barley, rye or a mixture of these cereals. The availability of complementary and confirmatory techniques, such as western blot, Q-PCR and MALDI-TOF mass spectrometry, has made it possible to identify the specific types of the toxic cereals of wheat, barley or rye in the contaminated oat samples and enabled us to confirm the Sandwich R5 ELISA data. The presence in contaminated oat samples of immunoreactive bands exclusively in the gliadin, hordein and secalin regions, detected with the western blot technique, demonstrated that they are contaminated with wheat, barley, rye or mixtures of these cereals. On the other hand, the absence of immunoreactive bands in the avenin region in both contaminated and noncontaminated oat products rules out the cross-reactivity of R5 antibody with oat avenins. The western blot technique also showed that contaminated oat products from the same oat variety could contain distinct types of wheat, barley and rye prolamins. The MALDI-TOF mass spectrometric technique has demonstrated its applicability in gluten analysis as a nonimmunological method, not only for the direct identification of gliadins, hordeins or secalins in foods [28], but also for its potential in determining wheat varieties [39,40], which has been essential in the case of avenins, as demonstrated in this study. Based on the characteristic avenin mass profile, the MALDI-TOF mass spectrometric technique has enabled the discrimination of 25 oat varieties among the 109 oat samples, as well as 25 pure, uncontaminated oat grains of known varieties. The analysis of the DNA amplification products by agarose gels has confirmed that some contaminated oat samples could contain either only barley, or wheat and barley or mixtures of wheat, barley and rye, with barley being the main contaminant. False positives were not detected by the Q-PCR technique in uncontaminated oat foods. These findings demonstrate the importance of using the complementary techniques of western blot, MALDI- TOF mass spectrometry and Q-PCR to corroborate the Sandwich R5 ELISA data. In recent years, the barley-sensitive Sandwich R5 ELISA has been the technique used most, not only for clinical follow-up studies [10,11,20] but also in industry and research to analyze gluten in oat foods [16 18]. In fact, oat samples 48, 49 and 50 were used for a clinical study, after an earlier confirmation of undetectable levels of gluten by Sandwich R5 ELISA [12]. Oat samples in V 13 were, however, discarded for another clinical study because all of them were contaminated, supporting the importance of using a reliable technique such as Sandwich R5 ELISA for testing gluten in oat foods. The cocktail solution was used in this study as the extraction procedure as many of the oat food samples analyzed, such as rolled oats, are partially heated. This extraction procedure guarantees the complete recovery of gluten, as has been previously demonstrated in rolled oats. It yielded 2 4-fold gluten as compared with extractions performed with conventional 60% aqueous ethanol [25]. According to the Codex standard for gluten-free products, oats is still considered as coeliac toxic. Owing to recent clinical studies, however, most coeliac disease associations, scientists, dieticians and physicians agree with the fact that oats are not toxic for persons with coeliac disease and a consensus on whether oats belong to the glutenfree diet may be near. It may, however, take some time before gluten-free, oats-based foods are totally accepted and oats are removed from the list of toxic cereals. Oats are a good source of fibre and iron, and its incorporation in a gluten-free diet could result in an increase in fibre intake among persons with coeliac disease, with corresponding health benefits. In the light of our results, adequate control of possible gluten contamination in oat foods would be much more rational than excluding oats altogether from the gluten-free diet. It is interesting to note that some oat grains are contaminated with powder containing gluten, which can be removed by washing the oat grains with 60% aqueous ethanol. To avoid gluten contamination in oat products, efforts have to be made by the food industry to produce oats free of significant gluten contamination through carefully monitored growing, harvesting and processing procedures.

9 Measurement of wheat and barley contamination in oats by Sandwich R5 ELISA Hernando et al. 553 Measurement of wheat, barley and rye contaminations in oats can be problematic not only owing to the different extractability of the prolamin subunits with the cocktail extraction, but also when oat foods are contaminated with a mixture of these cereals. Traditionally, gluten proteins have been divided into roughly equal fractions according to their solubility in alcohol water solutions: the soluble gliadins and the insoluble glutenins. Nevertheless, recent studies have reported slight differences in the ratio between gliadins and glutenins, suggesting a factor around 65:35 [41]. In this study, the factor observed between gluten and total hordeins in the analyzed oat food samples (Table 2) is incorrect as the total hordein values do not need to be multiplied by two. The correct factor would be , which is close to the experimental error of the ELISA technique. Therefore, oats contaminated with wheat or barley can be reliably measured by Sandwich R5 ELISA, combined with the cocktail extraction and the PWG gliadins, avoiding the use of two different standards (PWG gliadins or hordeins). In conclusion, this study has demonstrated the importance of using the Sandwich R5 ELISA for measurement of gluten in oat foods. The fact that pure uncontaminated oats have levels of gluten below 3 ppm support the in-vivo research indicating that oats could be safe for persons with coeliac disease. On the contrary, based on the massive contamination of gluten detected in most grains and commercial oat products, it has to be concluded that these samples could represent a health risk for the population with coeliac disease, in particular in countries where oat consumption by persons with coeliac disease is allowed and where the traditional barley-insensitive o-elisa is still the official analytical method to detect gluten. This situation is different in most European countries, in the United States and Canada, where oats has not been endorsed for consumption by persons with coeliac disease. Ideally, persons with coeliac disease should consume only those oat samples that have been tested and found to be free of gluten, thus confirming the importance of using the barley-sensitive Sandwich R5 ELISA when determining the gluten-free status of oats, because barley is the most prevalent cereal contaminant in this type of food samples. To be regarded as gluten-free and safe for persons with coeliac disease, oats should be tested for gluten contamination not only with the Sandwich R5 ELISA, but also with the complementary techniques of western blot, Q-PCR and MALDI-TOF mass spectrometry. Acknowledgements This work was supported by grants from the Ministerio de Ciencia y Tecnología, Plan Nacional (AGL / ALI and AGL ). Conflict of interest: none declared. References 1 Janatuinen EK, Pikkarainen PH, Kemppainen TA, Kosma VM, Jarvinen RM, Uusitupa MI, et al. A comparison of diets with and without oats in adults with coeliac disease. New Engl J Med 1995; 333: Srinivasan U, Leonard N, Jones E, Kasarda DD, Weir DG, O Farrelly C, et al. Absence of oats toxicity in adult coeliac disease. BMJ 1996; 313: Hardman CM. Absence of toxicity of oats in patients with Dermatitis herpetiformis. New Engl J Med 1997; 337: Reunala T, Collin P, Holm K, Pikkarainen P, Miettinen A, Vuolteenaho N, et al. Tolerance to oats in Dermatitis herpetiformis. Gut 1998; 43: Hoffenberg EJ, Haas J, Drescher A, Barnhurst R, Osberg I, Bao F, et al. A trial of oats in children with newly diagnosed coeliac disease. J Pediatr 2000; 137: Janatuinen EK, Kemppainen TA, Pikkarainen PH, Holm KH, Kosma VM, Uusitupa MI, et al. Lack of cellular and humoral immunological responses to oats in adults with coeliac disease. Gut 2000; 46: Janatuinen EK, Kemppainen TA, Julkunen RJK, Kosma VM, Mäki M, Heikkinen M, et al. No harm from five-year ingestion of oats in coeliac disease. Gut 2002; 50: Storsrud S, Olsson M, Lenner RA, Nilsson LA, Nilsson O, Kilander A. Adult coeliac patients do tolerate large amounts of oats. Eur J Clin Nutr 2003; 57: Hogberg L, Laurin P, Falth-Magnusson K, Grant C, Grodzinsky E, Jansson G, et al. Oats to children with newly diagnosed coeliac disease: a randomised double blind study. Gut 2004; 53: Srinivasan U, Jones E, Carolan J, Feighery C. Immunohistochemical analysis of coeliac mucosa following ingestion of oats. Clin Exp Immunol 2006; 144: Holm K, Maki M, Vuolteenaho N, Mustalahti K, Ashorn M, Ruuska T, et al. Oats in the treatment of childhood coeliac disease: a 2-year controlled trial and a long-term clinical follow-up study. Aliment Pharmacol Ther 2006; 23: Lundin KE, Nilsen EM, Scott HG, Løberg EM, Gjøen A, Bratlie J, et al. Oats induced villous atrophy in coeliac disease. Gut 2003; 52: Arentz-Hansen H, Fleckenstein B, Molberg O, Scott H, Koning F, Jung G, et al. The molecular basis for oat intolerance in celiac disease patients. PLOS Med 2004; 1: Silano M, Dessi M, De Vincenzi M, Cornell H. In vitro tests indicate that certain varieties of oats may be harmful to patients with coeliac disease. J Gastroenterol Hepatol 2007; 22: Janssen FW, Hagele GH, van Triest M, de Baaij JA. Contamination of rolled oats with wheat, rye and barley. In: Stern M, editor. Proceeding of the 13th Meeting Working Group on Prolamin Analysis and Toxicity. 5 8 November 1998, Barcelona, Spain. Zwickau: Verlag Wissenschaftliche Scripten pp Valdés I, García E, Llorente M, Méndez E. Innovative approach to low level gluten determination in foods using a novel sandwich ELISA protocol. Eur J Gastroenterol Hepatol 2003; 15: Thompson T. Gluten contamination of commercial oat products in the United States. New Engl J Med 2004; 351: Hernando A, Mujico JR, Juanas D, Mendez E. Confirmation of the cereal type in oat products highly contaminated with gluten. J Am Diet Assoc 2006; 106: Skerritt JH, Hill AS. Enzyme-immunoassay for determination of gluten in food; collaborative study. Assoc Off Anal Chem 1991; 74: Collin P, Thorell L, Kaukinen K, Maki M. The safe threshold for gluten contamination in gluten-free products. Can trace amounts be accepted in the treatment of coeliac disease? Aliment Pharmacol Ther 2004; 19: Méndez E, Vela C, Immer U, Janssen FW. Report of a collaborative trial to investigate the performance of the R5 enzyme linked immunoassay to determine gliadin in gluten-free food. Eur J Gastroenterol Hepatol 2005; 17: CCMAS. Report of the twenty-seventh session of the Codex Committee on Methods of Analysis and Sampling, Budapest, Hungary, May 2006, 12 ALINORM 06/29/23.

10 554 European Journal of Gastroenterology & Hepatology 2008, Vol 20 No 6 23 Department of Health and Human Services. Food and Drug Administation. 21 CFR Part 101. Docket No. 2005N Food Labeling;Gluten-Free Labeling of Foods. Proposed rule. 24 Kanerva PM, Sontag-Strohm TS, Ryoppy PH, Alho-Lehto P, Salovaara HO. Analysis of barley contamination in oats using R5 and o-gliadin antibodies. J Cereal Sci 2006; 44: García E, Llorente M, Hernando A, Kieffer R, Wieser H, Méndez E. Development of a general procedure for complete extraction of gliadins for heat processed and unheated foods. Eur J Gastroenterol Hepatol 2005; 17: Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the folin phenol reagent. J Biol Chem 1951; 193: Van Eckert R, Berghofer E, Ciclitira PJ, Chirdo F, Denery-Papini S, Ellis HJ, et al. Towards a new gliadin reference material isolation and characterisation. J Cereal Sci 2006; 43: Hernando A, Valdes I, Mendez E. New strategy for the determination of gliadins in maize- or rice-based foods matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry: fractionation of gliadins from maize or rice prolamins by acidic treatment. J Mass Spectrom 2003; 38: Mujico JR, Lombardía M,Méndez E. Detection of wheat contamination in gluten-free foods by a quantitative real-time PCR system. In: Stern M, editor. Proceeding of the 18th Meeting Working Group on Prolamin Analysis and Toxicity. 2 5 October 2003, Stockholm, Sweden; Zwickau: Verlag Wissenschaftliche Scripten; pp Mujico JR, Hernando A, Lombardia M, Benavides A, Silio V, Maki M, et al. Quantification of wheat, barley and rye contamination in oats samples by real-time PCR. In: Stern M, editor. Proceeding of the 19th Meeting Working Group on Prolamin Analysis and Toxicity. 30 September 3 October 2004, Prague, Czech Republic. Zwickau: Verlag Wissenschaftliche Scripten pp Mujico JR, Mendez E. Simultaneous detection/quantification of wheat, barley and rye DNA by a new quantitative real-time PCR system. In: Stern M, editor. Proceeding of the 20th Meeting Working Group on Prolamin Analysis and Toxicity September 2005, Maikammer, Germany. Zwickau: Verlag Wissenschaftliche Scripten pp Codex Alimentarius Commission. Draft revised standard for gluten free foods (CX/NFSDU 98/4). In: Codex Committee on Nutrition and Foods for Special Dietary Uses, 22nd session. Berlin; Wieser H, Seilmeier W. Determination of gliadin and gluten in wheat starch by means of alcohol extraction and gel permeation chromatography. In: Stern M, editor. Proceedings of the 17th Meeting of the Working Group on Prolamin Analysis and Toxicity. 3 6 October 2002, London, UK, England. Zwickau: Verlag Wissenschaftliche Scripten; pp Qi Jc, Zhang GP, Zhou MX. Protein and hordein content in barley seeds as affected by nitrogen level and their relationship to Beta-amylase activity. J Cereal Sci 2006; 43: Note of the final report of the 22nd session of the Codex Alimentarius (Codex Committee on Nutrition and Foods for Special Dietary Uses), Berlin 2000, ALINORM 01/26, pp Camafeita E, Mendez E. Screening of gluten avenins in foods by matrixassisted laser desorption/ionization time-of-flight mass spectrometry. J Mass Spectrom 1998; 33: Osman AA, Uhlig HH, Valdes I, Amin M, Mendez E, Mothes T. A monoclonal antibody that recognizes a potential coeliac-toxic repetitive pentapeptide epitope in gliadins. Eur J Gastroenterol Hepatol 2001; 13: Kahlenberg F, Sanchez D, Lachmann I, Tuckova L, Tlaskalova H, Mendez E, et al. Monoclonal antibody R5 for detection of putatively coeliac-toxic gliadin peptides. Eur Food Res Technol 2006; 222: Bloch HA, Kesmir C, Petersen M, Jacobsen S, Sondergaard I. Identification of wheat varieties using matrix-assisted laser desorption/ionisation time-offlight mass spectrometry and an artificial neural network. Rapid Commun Mass Spectrom 1999; 13: Sorensen HA, Petersen MK, Jacobsen S, Sondergaard I. Mass spectrometry and partial least-squares regression: a tool for identification of wheat variety and end-use quality. J Mass Spectrom 2004; 39: Wieser H. Chemistry of gluten proteins. Food Microbiol 2007; 24: