Characterization of cereal β-glucan extracts from oat and barley and quantification of proteinaceous matter

An extraction method for mixed-linkage β-glucan from oat and barley was developed in order to minimize the effect of extraction on the β-glucan structure. β-Glucan were characterized in terms of molecular size and molar mass distributions using asymmetric flow field-flow fractionation (AF4) coupled to multiangle light scattering (MALS), differential refractive index (dRI) and fluorescence (FL) detection. The carbohydrate composition of the extracts was analysed using polysaccharide analysis by carbohydrate gel electrophoresis (PACE) and high-performance anion-exchange chromatography (HPAEC). Whether there were any proteinaceous moieties linked to β-glucan was also examined. Purified extracts contained 65% and 53% β-glucan for oats and barley, respectively. The main impurities were degradation products of starch. The extracts contained high molecular weight β-glucan (105–108 g/mol) and large sizes (root-mean-square radii from 20 to 140 nm). No proteins covalently bound to β-glucan were detected; therefore, any suggested functionality of proteins regarding the health benefits of β-glucan can be discounted.


Introduction
The water soluble β-glucan from cereals, such as oat and barley, is a linear β-1,3 and β-1,4 linked polysaccharide formed solely of β-D-glucopyranosyl units. Several health effects of mixed-linkage β-glucan, acting as dietary fibre, have been approved by both, the U.S. Food and Drug Administration (FDA) [1,2] and the scientific panel of the EU (European Food Safety Authority (EFSA)) [3,4], including the reduction of serum cholesterol in the blood as well as reduced glycaemic and insulin responses [4]. Therefore, β-glucan is considered to be a functional and bioactive food ingredient which can reduce the risk of a number of chronic diseases with increased daily intake. Whilst these claims have been accepted, there is no understanding of the mechanism of action of β-glucan or potential ways to improve these health benefits in food for health purposes. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 support in the form of salaries for author MLA], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: Marie-Louise Ainalem is employed by the European Spallation Source ESS AB. There are no patents, products in development or marketed products to declare. This does not alter our adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
The present study describes an extraction procedure for analysis of non-modified β-glucan from oat and barley. Mixed-linkage β-glucan was extracted from oat and barley flour of relatively high purity, considering the conditions of the method applied. The aim of the extraction method used, was to preserve the native properties of the β-glucan polymer such as possible aggregates or protein/ β-glucan links. No proteases were used during the extraction procedure in order to maintain any possible proteinaceous moieties. Polysaccharide analysis by carbohydrate gel electrophoresis (PACE) and high-performance anion-exchange chromatography (HPAEC) analysis of monosaccharide sugar composition were used to assess the purity of the β-glucan extracts. β-Glucan extracts were characterized using AF4 with multiangle light scattering (MALS) and differential refractive index (dRI) detection. Additional AF4 experiments were performed using in-line post-channel Calcofluor labelling for semi-quantitative fluorescence (FL) detection of β-glucan in the extracts. AF4-MALS-FL was also used for determination of the presence of proteinaceous moieties by pre-separation fluorescent labelling of peptide bonds.

Extraction of β-glucan
In order to avoid proteases and harsh conditions, such as elevated pressure and alkaline treatments, the following procedure was used. The first extraction step involved defatting 2 g of flour with 75 mL acetone, stirring the suspension at RT for 2 h, followed by centrifugation for 20 min at 1,000 x g. The supernatant was discarded and the pellet dried overnight at RT. 1.5 g (~45 000 units) heat-stable α-amylase was re-suspended in 50 mL Milli-Q water and incubated at 100˚C for 30 min to terminate any non α-amylase related exogenous activities [27]. The pre-treated α-amylase was mixed with the defatted flour in 100 mL Milli-Q water. The suspension was boiled for 1 h and stirred until the slurry had cooled to RT. 100 mL ethanol was added to precipitate the water soluble β-glucan and the sample stirred at RT for 1 h. After centrifugation for 30 min (2,000 x g), the precipitation step was repeated using the supernatant from the previous step. The pellets from both precipitations were re-dissolved in 50 mL of Milli-Q water, combined and heated to 65˚C for 40 min. After cooling to RT, the viscous liquid was centrifuged for 1 h at 1,000 x g and the supernatant was ovendried at 80˚C for 15 h. The final extract was re-purified by repeating the whole procedure, from α-amylase digestion, a further two times. The amounts of α-amylase and solvents used were adjusted accordingly.
The β-glucan content of the extracts was analysed using the Mixed-Linkage β-Glucan Assay Kit from Megazyme International Ltd., Bray, Co. Wicklow, Ireland, with some modification in order to analyse the high concentration of β-glucan in the preparations produced. The first incubation step at 100˚C was prolonged to 2 h. The amount of lichenase used in the next step was doubled, and the mixture incubated at 50˚C for 2 h. The reaction solution was mixed with 100 mL of sodium acetate buffer (100 mM with a pH of 4) and stirred for 15 min before being centrifuged. The rest of the procedure was unchanged from that described in the kit. The developed procedure was verified by processing and measuring various samples with known high content of β-glucan. The final extracts used in this study contain 65% (w/w) β-glucan in case of oat and 53% (w/w) β-glucan in case of barley flour extract. The starch content of the extracts was analysed using the Total Starch Assay Procedure Kit, available as well from Megazyme.

PACE (polysaccharide analysis by carbohydrate gel electrophoresis)
To check the quality of the extracted β-glucan and to identify possible impurities, PACE was performed as described elsewhere [28]. This method is used, among other applications, for elucidation of structure and characterization of carbohydrates. PACE relies on the migration of fluorophore-labelled mono-and oligosaccharides produced following digestion of a sample with specific glycosyl-hydrolases. The labelled sugars are separated based upon their size and structure. The fact that multiple samples can be run alongside each other on the polyacrylamide gels means that samples are exposed to exactly the same conditions during analysis.
Briefly, 250 μg aliquots of oat and barley extracts were prepared by drying stock solutions in vacuo (Savant Inc., Midland, Michigan, USA), they were then re-dissolved in water and digested with either xylanase 11 (4 μl % 21.92 μg) or lichenase (1 μl % 0.35 U) for 16 h at 40˚C. The same amount of material was digested with α-amylase (2 μL % 35.9 U) for 30 minutes at 50˚C. The total reaction volume for all digestions was 500 μL. The digestion was terminated by boiling the sample for 30 min and samples dried in vacuo. Oat and barley fractions were labelled overnight with ANTS as described [28], as were xylo-oligosaccharides (Xyl) 1-6 and αamylase digested soluble starch. Following labelling, the derivatized sugars were dried in vacuo and reconstituted in 200 μL of 3 M urea before running on PACE gels. Digested soluble starch samples were diluted fourfold prior to loading on gel. PACE gels were visualized under UV light using a GelDoc-It TS2 imager (UVP, Jena, Germany) equipped with a GFP emission filter (513-557 nm).

HPAEC-PAD analysis of monosaccharide sugar composition
HPAEC-PAD (high-performance anion-exchange chromatography equipped with pulseamperometric detector) enables to study the monosaccharide composition of carbohydrates following acid hydrolysis. The monosaccharide analysis of β-glucan extracts was performed as described by Goubet et al. [29]. Briefly, 50 μg of either oat or barley β-glucan extract was incubated for 1 h at 121˚C in 400 μl of 2 M TFA. The hydrolysates were lyophilised, re-suspended in water and analysed using a Dionex ICS3000 HPLC equipped with CarboPac PA 20 column (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA).

AF4 analysis equipment and separation parameters
The AF4 instrument used was an Eclipse 3+ Separation System (Wyatt Technology Europe, Dernbach, Germany). The system was connected to a Dawn Heleos II multiangle light scattering (MALS) detector operating at 658 nm and an Optilab T-rEX differential refractive index (dRI) detector (both Wyatt Technology Europe) operating at the same wavelength. The carrier flow was delivered through an Agilent 1100 series isocratic pump with an in-line vacuum degasser and sample injection was via an Agilent 1100 series autosampler (Agilent Technologies, Waldbronn, Germany). To ensure that only particle free carrier liquid entered the system, a filter-holder with a 100 nm pore-size polyvinylidene fluoride (PVDF) membrane (Millipore Corp.) was placed between the pump and the channel inlet. The AF4 channel was a Wyatt long channel with a tip-to-tip length of 27.5 cm and a nominal thickness of 190 μm. The actual thickness was determined to be 156 μm by calibration with ferritin as described previously [30]. The ultra-filtration membrane forming the accumulation wall was made of hydrophilized polyethersulphone with a cut-off of 10 kDa (Microdyn-Nadir GmbH, Wiesbaden. Germany). 10 mM NaNO 3 and 0.02% (w/v) NaN 3 in Milli-Q water were used as carrier liquid. The samples were prepared at a concentration of 4 mg/ mL by dissolving the extracted β-glucan in the carrier liquid used for elution. The suspension was boiled for 30 min allowing the β-glucan to dissolve. The samples were cooled to RT and filtered through a 22 mm cellulose acetate membrane syringe filter with a 0.45 μm pore size (VWR International) prior to injection onto the channel. The sample was injected at 0.2 mL/ min for 4 minutes. The injected volume was 40 μl for oat and 80 μl for barley, corresponding to an injected mass of 160 μg and 320 μg, respectively. A focusing/ relaxation step of 5 min was performed prior to elution with a focusing flow rate of 0.5 mL/ min resulting in a void time (t 0 ) of 9 min. The initial cross-flow rate during elution was 1 mL/ min followed by an exponential decay with a half-life of 4.5 min in order to avoid excessive retention and elution times. After elution, the AF4 channel was flushed without any cross-flow for 5 min to ensure a clean channel before analysis of the next sample. During the whole measurement the detector flow rate was kept constant at 0.5 mL/ min.

Fluorescence measurements
The use of fluorescence detector enabled detection of both β-glucan and possible peptide residues through specific labelling of the purified β-glucan samples. A Jasco FP-1520 fluorescence detector (Jasco Inc., Easton, Maryland, USA) operating an Ushio Xenon Short Arc lamp (Ushio Inc., Tokyo, Japan) was used. Extracts were labelled in-line, post-channel with Calcofluor to detect β-glucan or pre-injection with EDAC to label peptide bonds, and analysed using the aforementioned AF4-MALS-FL set-up. For Calcofluor labelling of β-glucan, the method described by Ulmius, Ö nning, & Nilsson [25] was adopted, using a 25 mg/ L Calcofluor solution in 0.1 M tris(hydroxymethyl)-aminomethane buffer (pH 8), measuring the emission at λ em = 445 nm (λ ex = 415 nm). For labelling of possible proteinaceous matter in βglucan extracts, the procedure described in [31] was adopted, with minor modification in that the labelling agent was diluted 1 in 10. Thus, 1 mM solution of EDAC in Milli-Q water was mixed for 3 hours with 1 mM solution of 7-methoxycoumarin-3-carboxylic acid in two parts of Milli-Q water and one part of DMSO to enhance the solubility. After mixing, the labelling solution was added to an already dissolved β-glucan sample in 1:1 ratio (v/v), stirred for 1 h and immediately injected onto the AF4 channel. The wavelengths used for detection of possible proteinaceous matter were λ ex = 336 nm and λ em = 402 nm.

Elemental analysis
To measure the total protein content and other peptides that are possibly not associated with β-glucan, the extracts were analysed using Flash EA1112 elemental analyzer from Thermo Fisher Scientific, Delft, the Netherlands. The system uses the Flash Dynamic Combustion method, where complete combustion of the sample is reached through heating to 1000˚C, and subsequently followed by determination of the produced elemental gases, in this case nitrogen from proteinaceous matter. Aspartic acid at three different concentrations was used to calibrate the instrument. The percentage of protein per sample (w/w) was estimated by multiplying the nitrogen content by a 6.25 nitrogen-to-protein conversion factor. The elemental analyzer has sensitivity down to 100 PPM.

Data processing
Astra software in version 5.3.4.14 (Wyatt Technology Europe, Dernbach, Germany) was used to process the data obtained from the FL, MALS and dRI detectors following AF4 separation. M and root-mean-square radii (r rms ) were calculated using the Berry method [32,33] performing a 1 st order fitting with the data obtained from the scattering detectors 9-14 (respective scattering angles: 69.3˚-121.2˚). Lower and higher scattering angles were excluded as they did not correspond to the plot. The RI increment (dn/dc) used was 0.146 mL/ g as defined for β-glucan in aqueous solutions [34] whereas the second virial coefficient A 2 was neglected.

Results and discussion
The β-glucan extracts were analysed for chemical composition using a commercial kit (enzymatic β-glucan determination), PACE and HPAEC-PAD. AF4-MALS-dRI was used to obtain fundamental information regarding β-glucan structure such as M and r rms distributions. Additional AF4-MALS-FL experiments were performed on the extracts to identify β-glucan, any possible polysaccharide-protein interactions and protein contaminations in the β-glucan fractograms.
An oat β-glucan extract of 65% (w/w) purity and barley β-glucan extract of 53% (w/w) purity were prepared. These obtained purities should be recognized as good bearing in mind the mild extraction conditions and methodology designed to retain the natural structure of the extracted polymers. Extraction methods using more extreme conditions, for example protease and alkali treatments, resulted in 57% (w/w) purity for oats [35], and up to 90% (w/w) purity [36] for barley extracts using acidic conditions (pH 4). Immerstrand et al. [35] also mentioned a potential influence of the geographical origin and growing conditions of the starting material on the gained extraction purities and yields when using the same extraction method. Extraction yields for similar conditions as in the present paper could not be found in literature.
The amount of starting material was 2 g of oat or barley flour of which 28% and 24% is known to be β-glucan. The theoretical maximum yield would therefore be 560 mg and 480 mg, respectively. Our yields were around 200 mg for oat with purity of 65% which leads to 130 mg of β-glucan, and 180 mg for barley with purity of 53%, leading to 95 mg of β-glucan.
Therefore, in this study, the extraction efficiency is around 23% of total available β-glucan for oat and 20% for barley.
The PACE gels of enzymatically digested oat and barley extracts with specific glycosylhydrolases are shown in Fig 1. Xylose to xylo-hexose standards (Xyl) 1-6 (labelled as 'std') and α-amylase digested soluble starch (labelled as 'S') were run alongside other digests to identify the main impurities present in oat and barley β-glucan extracts. 'b' are blanks; these samples were incubated with no addition of glycosyl-hydrolases. To obtain additional information of the extracts' purity, HPAEC-PAD analysis of monosaccharide sugar composition was performed. The results are summarized in Table 1. On a dry weight basis, 92.16% (w/w) of the hydrolyzed oat extract is glucose while for the hydrolyzed barley extract the glucose content is 72.07% (w/w). In order to compare these values with the β-glucan content obtained using the commercially available β-glucan assay kit, from which values of 65% and 53% (w/w) for oat and barley, respectively, were obtained, the values need to be recalculated to correct for the difference in M (i.e. glucose: 180 g/ mol and glucose pyranosyl unit in a β-glucan chain: 162 g/ mol). Thus, a correction factor (162 [g/ mol]/180 [g/ mol]) needs to be multiplied with the glucose content in order to estimate how much of that glucose originates from β-glucan. The results show that approximately 17% (oat) and 12% (barley) of glucose does not come from β-glucan. The starch content of the samples was determined to be 11% (w/w) each in both extracts using a commercial starch assay kit (Megazyme).
The origin of the remaining glucose is most likely related to various dextrins originating from degradation of starch during the extraction. Hydrolysis of starch by α-amylase can result in various oligosaccharides and dextrins of which the composition depends on the substrate and the specific α-amylase utilized [37]. The solubility of these enzymatic products is relatively low in ethanol [38,39] and they may precipitate with the β-glucan during the extraction. The results from PACE analysis also indicate this. Other non-glucose monosaccharides were found in very small quantities in the hydrolyzed oat and barley extracts (e.g. xylose, arabinose, mannose and galactose). Rhamnose and galacturonic acid were not detected in any sample. Hence, the main impurities of the extracted samples, not being starch, are other water soluble polysaccharides or sugar residues, coming from other cereal polysaccharides and hemicelluloses e. g. glucomannans, arabinoxylans and starch dextrins in the endosperm or the cell walls of the oat and barley grains.
For further characterization of the β-glucan structure, AF4-MALS-dRI was utilized. Prior to the injection of the sample onto the AF4-channel, the prepared samples were filtered through a 0.45 μm pore size syringe filter with a cellulose acetate membrane. Sample filtering did not have any significant influence on the results but improved the signal-to-noise ratio as shown in Fig 2. These measurements were performed as pre-experiments and the flow conditions were different, resulting in different elution times compared to subsequent experiments reported in this study. The fractogram shows Rayleigh ratio signal (intensity of the scattered light) from the MALS detector and the dRI signal representing the concentration over the elution time as well as the calculated distribution for r rms . The MALS and dRI signals were normalized against the peak maxima. The results (Fig 2) show that the signals and size distributions remain similar with a mass recovery from the AF4 channel of 99% for the non-filtered and 96% for the filtered sample. It is also evident that the noise in the determined r rms distribution is considerably lower in the filtered sample, especially for sizes below 60 nm. The higher noise in the non-filtered sample is most likely originating from very low amounts of very large analytes which co-elute in the steric/ hyperlayer mode of AF4 [40]. The presence of even very low amounts of large species may give rise to noisy MALS signals impairing the size determination [41,42].
The plots in Figs 3 and 4 display fractionation of the extracts as described for Fig 2, including the additional calculated distribution for the weight-average molar mass, M w . It can be seen that the extracts are polydisperse samples with high M fractions. For oat (Fig 3), M w ranges between 10 5 -10 8 g/mol, the r rms ranges between 50 and 120 nm. For barley (Fig 4) the values are in the same range for M w but have a rather wider range of r rms (20 to 140 nm). Mean values for M w and r rms for both extracts as well as their respective mass recoveries from the AF4 separation channel are summarized in Table 1. These results show that oat β-glucan is larger than the barley β-glucan which is in agreement with earlier findings i.e. by Beer, Wood, & Weisz [43] and Ahmad, Anjum, Zahoor, Nawaz, & Dilshad [44]. The differential M distribution of   Utilizing AF4 in combination with FL detection, β-glucan could be successfully labelled and made visible in the fractograms. The results of combining the AF4-MALS with selective FL labelling of the β-glucan can be seen in Fig 6 for oat and Fig 7 for barley. The signals were normalized against the peak maxima. A short delay in the elution times compared to the AF4-MALS-dRI was observed owing to different tubing configuration to allow the in-line labelling procedure. The β-glucan in the extracts have been selectively labelled with Calcofluor and detected at a wavelength of λ em = 445 nm.
The results show the presence of β-glucan over the entire size distribution for oat β-glucan (Fig 6) while the late eluting analytes (retention time >25 min) in the barley β-glucan (Fig 7) sample appear to consist of low amounts of β-glucan or no β-glucan at all. The interpretation of Calcofluor labelled β-glucan data has some limitations as low M β-glucan (M w approx. 4Á10 4 g/ mol) tends to yield a lower signal intensity which for the low M analytes is also dependent on the ionic strength [45]. This might lead to underestimation of the content of low M β-glucan in the sample. For the late eluting analytes the low or non-existent FL-signal also coincides with a low dRI-signal (i.e. low concentration of analytes) in Figs 3 and 4 making it unclear whether the β-glucan content is lower or it is just a consequence of the very low analyte concentration. HPAEC-PAD showed xylose and arabinose, which are products of arabinoxylan hydrolysis, in small amounts (Table 1) when compared to data available in literature [46]. It is possible that the late eluting analytes (>25 min) in the barley extract are traces of very large size arabinoxylan species present at very low concentration and with no FL-signal but with Extraction and characterization of cereal β-glucan distinct MALS-signal (Figs 3 and 7). These are also apparent in xylanase 11 digests of barley extract run on PACE gels (Fig 1). However, it is not possible to be conclusive that this is the case based on the results obtained.
Since a presence of protein in the β-glucan system would play an important role in formation of aggregates and could contribute to the claimed health benefits of β-glucan both extracts were analysed for protein content using two different approaches: AF4-MALS-FL (with preinjection fluorescent labelling) and total N content determination through elemental analysis. AF4-MALS-FL did not show the presence of any proteinaceous moieties in either extracts (data not shown). Similarly, the elemental analysis showed no detectable amounts of nitrogen/ protein (detection limit approx. 100 PPM). Although using an extraction procedure with mild conditions we found no evidence supporting the hypothesis of proteinaceous matter bound or creating a complex or aggregate with oat or barley β-glucan.

Conclusion
Mild extraction conditions were used to isolate β-glucan from oat and barley flour without modifying their original structure. The method gave β-glucan of good yield and purity. The main impurities are starch and possible starch degradation products and trace amounts of arabinose, xylose, mannose, galactose and fucose. The extracts contained β-glucan of high M w (10 5 -10 8 g/ mol) and large size (r rms from 20 to 120 nm for oat, up to 140 nm for barley). Furthermore, the extracts were analysed for possible proteinaceous moieties in the β-glucan. There were no covalently bound proteins found in the β-glucan extracts produced by the method described and therefore this study does not strengthen the hypothesis of covalently bound proteins to β-glucan. Hence, the suggested importance of proteinaceous moieties for the functionality of β-glucan and their beneficial health effects cannot be supported.