β-glucan and arabinoxylan samples isolated and purified from the common Norwegian barley variety Tyra were the main basis for our investigations. As shown in Table 1, the constituent sugar analysis combined with ¹H-NMR [23] (spectra not shown) revealed that WUM-BS contained 70% arabinoxylan and 30% mixed-linked β-glucan. Treatment with lichenase (L) and amyloglucosidase (A) efficiently removed most of the remaining mixed-linked β-glucan from this fraction; the enzyme treated fraction WUM-BS-LA contained 96.4% arabinoxylan. Trace amounts of mannose were attributed to the glycoprotein part of the enzyme preparation used. In this fraction, ferulic acid is not present due to alkali treatment during the extraction procedure.

WSM-TP was composed of 90% mixed-linked β-glucan in addition to 10% co-extracted arabinoxylan. Most of the arabinoxylan was removed by enzymatic treatment with xylanase (X). The enzyme treated fraction WSM-TPX was composed of 96.5% glucose and only 3.5% arabinoxylan.

The estimated relative weight average molecular weights (Mw) based on the pullulan series were about 886 and 156 kDa for WSM-TPX and WUM-BS-LA, respectively (Table 1). Molecular weight decreased during the enzymatic treatment of WSM-TP and WUM-BS from about 1090 and 412 kDa, respectively, giving samples with less polydispersities (Mw/Mn).

All previous studies on immunomodulatory activity of mixed-linked β-glucans from barley have been performed on commercially available samples. It should be noticed [20] that choice of isolation method may influence polysaccharide characteristics, such as molecular weight and solubility, and thereby their biological activities. In addition, co-extracted substances or contaminants of an endotoxin nature that may occur during isolation may contribute to significant activities in immunological test systems.

Potential degradation of dietary fiber during food processing has not been taken into account in this study. In addition, the fact that dietary fiber very seldom is eaten alone without subsequent intake of several other food constituents makes the picture quite complex and complicated to explore. In the present study, all fiber fractions had relatively high molecular weight after isolation, and no attempts were made to alter the chain length in either of the samples. This was because we primarily wanted to investigate intact carbohydrate dietary fibers with the presumption that dietary fiber remains undegraded until reaching the microflora in the colon.

To test the inflammatory response of the fiber fractions on gut epithelial cells the modulation of IL-8 (CXCL8) secretion from the human intestinal epithelial cell lines Caco-2 and HT-29 cells was determined. The concentration used of 1 mg/mL is physiologically relevant as a concentration of 1 mg/mL barley fiber in the intestine corresponds to the consumption of approximately 20 g barley, an amount found, for example, in two slices of barley bread (40% barley flour). In addition, the potential toxic effect of the fiber fractions on the Caco-2 and HT-9 cells was determined by measuring the effect of the fiber fractions at different concentrations on cell proliferation using the MTT assay.

We found that the fiber fractions had no significant effect on the cell proliferation of the human intestinal epithelial cell lines Caco-2 and HT-29 cells (Figure 1A, B). It was observed that the HT-29 cell line in general secretes considerably higher levels of IL-8 than Caco-2, but the barley fiber fractions had no significant effect on this secretion either from Caco-2 (Figure 2A) or HT-29 cells (Figure 2B). Only the positive controls, PMII and zymosan increased secretion of IL-8 from both cell lines significantly (p = 0.001) compared to the respective controls. Zymosan is a crude extract from yeast (Saccharomyces cerevisiae) and contains mainly β-glucan but also some mannan [20], protein, fat and chitin [36]. Immunomodulatory activities of β-glucan from yeast and from other sources have been studied extensively, for review on this topic see [37]. PMII is a pectic polysaccharide fraction isolated from Plantago major L. leaves, a plant used in traditional medicine to aid the healing of wounds. PMII has shown immunomodulatory activities both in vitro and in vivo: Increased resistance against bacterial infection in mice, activation of human monocytes and activation of the complement system [35,38,39]. PMII is therefore considered useful as positive control in immunological test systems.

Even though we did not find any direct effect of the barley fiber fractions on the intestinal epithelial cell model system, barley fiber may affect inflammatory processes and immune response by other mechanisms. As outlined in the introduction, barley fiber may be taken up by intestinal macrophages or M-cells and delivered to underlying immune cells where binding of barley β-glucan to the lectin site of CR3 on effector cells has been shown to enhance cytotoxic activity [15,27].

The nuclear transcription factor kappa B (NF-κB) plays a central role in inflammatory response [40]. Thus, to further study the effect of the fiber fractions on the immune response, the ability of the fiber fractions to modulate basal and LPS-induced NF-κB activity was tested using the U937-3κB-LUC monocytic cell line stably transfected with a luciferase reporter containing three NF-κB binding sites. It has been shown that this model system correlates well with in vivo NF-κB activity [41,42] Due to the limitations of the test system, lower concentrations of the samples (0.1, 0.2 and 0.4 mg/mL) were used compared to experiments with the Caco-2 and HT-29 cell lines. However, as the U937-3κB-LUC cell line is quite sensitive, the response is still considered relevant. The activities of the different fiber fractions were compared to the positive control, PMII [39]. Of the different fiber fractions only the highest concentration of WUM-BS had a significant effect on basal NF-κB activity (p = 0.004) (Figure 3A), giving an increase of the activity to 270% compared to control. The apparent dose response from 0.1 mg/mL to 0.4 mg/mL of all fractions of the basal NF-κB activity was statistically not significant compared to the control. However, all concentrations of PMII significantly increased basal NF-κB activity (p < 0.001) in the test system compared to the control. None of the fiber fractions had significant effect on the LPS-induced NF-κB activity, only the highest concentration of PMII increased the LPS-induced NF-κB activity significantly (p = 0.013) (Figure 3B).

Only high concentrations of a fraction containing 70% arabinoxylan and 30% mixed-linked β-glucan (WUM-BS) increased the activity of the pro-inflammatory transcription factor NF-κB in monocytes. This fraction was obtained after alkaline extraction of a water insoluble residue. Neither pure arabinoxylan nor pure mixed-linked β-glucan was active in this test system. Some biological effects of β-glucans are initiated by binding to Dectin-1 on macrophages and dendritic cells. Barley β-glucan has previously been found to activate NF-κB when Dectin-1, Syk, SARD9 and Bcl10 were co-expressed in the cells [33], and it was concluded that Dectin-1 was involved in these activities. However, binding to Dectin-1 requires β-glucans with a minimum of 10- or 11- mer 1,3-linked glucose oligomers [43] which are structural elements not found in barley. Barley β-glucans only contain single 1,3-linked glucose units separating two or three 1,4-linked glucose oligomers [44]. Transcription factor NF-κB can be activated via many different pathways including proinflammatory cytokines, TLR activation, for example, by LPS and by T-cell activation [40]. As shown in Figure 3B, LPS induced NF-κB activity was not significantly altered by any of the barley fractions indicating that the basal activity observed after stimulation of 0.4 mg/mL WUM-BS may be due to contamination by LPS. In any case, the activity found in Figure 3A cannot be due to either arabinoxylan or mixed-linked β-glucan since other fractions containing higher levels of mixed-linked β-glucan (WSM-TP and WSM-TPX) and arabinoxylan (WUM-BS-LA) were inactive in the test system. The positive control PMII increased LPS induced activity but to a lesser extent than measured with PMII alone (Figure 3A). This shows that PMII is active per se, but confirms presence of LPS. Previously, it has been shown that PMII can activate monocytes and induce secretion of TNFα [39]. One might speculate that secreted TNFα in turn activate NF-κB [40], alternatively PMII may bind to NF-κB activating receptors directly.

Purified β-glucan (WSM-TPX) and arabinoxylan (WUM-BS-LA) from Tyra were tested for activity in the complement-fixing test. Both showed lower activities than the positive control, PMII. At 1 mg/mL WSM-TPX was significantly more active than WUM-BS-LA (p = 0.009) (Figure 4).

β-glucans and purified arabinoxylans from other barley varieties [45] were also subjected to this test. As shown in Figure 4, all arabinoxylan fractions (WUM-BS-LA) had relatively low activity compared to the positive control. Arabinoxylan isolated from Tyra, NK96300 and SB94897 had very similar activities; the one from CDC Gainer was almost inactive.

Starch-free mixed-linked β-glucans that had not been subjected to a xylanase treatment (WSM-TP) containing additional small amounts of arabinoxylan, had the highest activity in this test system. Such WSM-TP samples originating from CDC Gainer and NK96300 showed activity at the same level as the positive control, while similar fractions from Tyra and SB94897 were less active (p < 0.035). Figure 4 furthermore shows that a xylanase purification step of WSM-TP into purified mixed-linked β-glucan (WSM-TPX) did not alter the complement-fixing activity significantly.

In general, all mixed-linked β-glucan rich fractions had a significantly higher complement-fixing activity than the arabinoxylan-rich fractions (p < 0.027).

According to Figure 4 the fractions can be listed as follows with regard to decreasing activity in the complement fixing test: PMII = WSM-TP CDC Gainer = WSM-TP NK96300 > WSM-TP Tyra = WSM TPX Tyra = WSM TP SB94897 > WUM BS-LA Tyra = WUM-BS-LA NK96300 = WUM-BS-LA SB94897 > WUM-BS-LA CDC Gainer.

Mixed-linked β-glucans from barley have previously shown to activate the complement system via the alternative pathway [26], the present study demonstrates an effect also on the classical pathway. The complement system provides a first line of protection against potential harmful invaders and is part of the innate immune system. It consists of a group of serum proteins that are activated in a cascade mechanism. Many of these proteins are pro-enzymes that are activated by proteolytical cleavage which in turn activate the next step in the cascade. Activation can be initiated by three pathways; the classical pathway, the alternative pathway or the lectin pathway, and is important for initiating inflammation, activation of leucocytes, lysis of target cells and opsonisation [46,47]. The test system employed has some limitations since it does not distinguish between activation and inhibition of the complement cascade, only a “consumption” of complement activity is registered. From previous studies however, it is established that PMII, the positive control, is an activator of the complement system [35], and it has also shown to protect against bacterial infection in vivo [38].

The mixed-linked β-glucan fractions tested were more active than the arabinoxylan fractions. The reason for the differences in activity of the different β-glucans might be due to differences in their primary structure. The ratio of (1→4)/(1→3) linkages present varies between the different barley varieties tested. NK96300 has the highest ratio (2.76) followed by CDC Gainer (2.59), Tyra (2.48) and SB94897 (2.30) [23]. The varieties with the highest (1→4)/(1→3) ratio have the highest activity in the complement fixing test, but statistical analysis shows no significant correlation between linkage ratio and activity or between molecular weight of the WSM-TP fractions and activity. Mw of WSM-TP fractions from NK96300, CDC Gainer and SB94897 were estimated to 1040, 1130 and 1040 kDa, respectively. The estimated molecular weights of the WSM-TP fractions were significantly higher than the corresponding arabinoxylan (WUM-BS-LA) fractions (p < 0.001). Mw of the WUM-BS-LA fractions from NK96300, CDC Gainer and SB94897 were estimated to 214, 203 and 190 kDa, respectively. The WUM-BS-LA fractions from different barley varieties tested had in general low activities in the complement fixing test. However, when both these arabinoxylan fractions and the more active β-glucan rich fractions were included in the statistical test a significant positive correlation was found between molecular weight and activity (p = 0.002). On the other hand, when these two classes of dietary fiber are evaluated separately there is no correlation between activity and their estimated molecular weights. Contamination of LPS does not affect this test system [48], so activity is not attributed to presence of endotoxin.

The arabinoxylans from different barley varieties tested had in general low activities in the complement fixing test leaving this hemicellulose type of dietary fiber non-responsive in all the test systems in the present study. To our knowledge, high molecular weight arabinoxylans have not been ascribed immunomodulatory activities. On the other hand arabinoxylan oligosaccharides have been studied for prebiotic properties [49] and have been shown to reduce preneoplastic lesions in the colon of rats treated with a carcinogen [50].

Fiber fractions were extracted from four barley varieties; NK96300, Tyra, CDC Gainer and SB94897 basically as previously described [51]. Briefly; milled (0.5 mm) barley samples (48 g) were extracted and washed with boiling ethanol. This removed low molecular weight constituents and is promoting the denaturation of endogenous hydrolytic enzymes such as β-glucanase. Following defatting with hexane, extraction with boiling water gave a water soluble material (WSM) and a residue of water insoluble material (WUM). WSM was furthermore treated with 7 mL amylase (Termamyl 120 L, Type L, Novozymes ) and 75 mg protease (Porcine Pancreatine, SIGMA) filtered and recovered with alcohol precipitation resulting in the starch free fraction designated WSM-TP.

In an attempt to remove small amounts of co-extracted arabinoxylan, WSM-TP was treated with a xylanase. WSM-TP Tyra (1 g) was dissolved in sodium acetate buffer pH 4.5, and 10 μL (21 U) endo β-xylanase (β-xylanase M6, Megazyme) was added at 40 °C and left for 3 h with gentle stirring. Polysaccharide material was precipitated with isopropanol and centrifugated at 1000 × g for 10 min. The pellet was redissolved in water, dialyzed against distilled water using a dialyzing tube with cut off 12,000–14,000 (Medicell Int. Ltd); freeze dried and designated WSM-TPX.

Base soluble material (WUM-BS) was then extracted from the previous water insoluble residue (WUM) with 1 M NaOH added 1% NaBH₄. Co-extracted mixed-linked β-glucans and starch were removed by adding 50 U lichenase (Lichenase EC 3.2.1.73, Megazyme) and 400 μL amyloglucosidase (Amyloglucosidase for Total Dietary Fiber Assay EC 3.2.1.3, SIGMA) as described elsewhere [51] giving fractions designated WUM-BS-LA.

Methanolysis combined with TMS-derivatisation and GC were performed according to the method of Chambers & Clamp [52] with modifications as previously described [53] using 4 M HCl in anhydrous methanol for 24 h at 80 °C.

¹H-NMR spectra of selected samples were obtained on a Varian Mercury 300 system. Approximately 4 mg of freeze dried material was solubilized in 0.7 mL D₂O, transferred to NMR glass tubes and acquired at 80 °C with typically 64 scans. Further details of the method are described in Knutsen & Holtekjolen [23].

GPC-SEC was performed using a DIONEX P680 pump with a Spectraphysics AS3500 auto injector and a Shimadzu RID6A refractive index detector controlled with Chromeleon 6.80 software. Serially connected Shodex OHPack SB-806-HQ and SB 804-HQ columns were connected to a Shodex OHPack SB-LG precolumn and eluted at 40 °C with 50 mM Na₂SO₄ (0.5 mL/min), and samples (1 mg/mL) were injected using a 100 μL loop. Relative molecular weight averages (Mw and Mn) were estimated offline by the software WINGPC −6.2 using pullulan molecular weight standards ranging from Mp 342 to 1,520,000 Da for calibration. Software and standards were obtained from PSS (Polymer Standards Service GmbH, Mainz, Germany).

The Caco-2 cell line (obtained from the American Type Culture Collection (ATCC), and a generous gift from Professor Kirsten Sandvig, Norwegian Radium Hospital) and HT-29 cell line (obtained from ATCC, and a generous gift from Professor Tor Lea, Norwegian University of Life Sciences) were grown in DMEM medium containing 10% fetal calf serum, 1% non-essential amino acids, 100 U/mL penicillin, and 100 mg/mL streptomycin. The U937-3xkB-LUC cell line (a generous gift from Professor Rune Blomhoff, University of Oslo) was grown in RPMI-1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 50 U/mL penicillin, 50 mg/mL streptomycin and 75 μg/mL hygromycin (Sigma-Aldrich, St. Louis, MO). The cells were maintained at 37 °C and 5% CO₂ in a humidified incubator. If not otherwise stated, all solutions were obtained from Invitrogen (Carlstad, CA).

For IL-8 secretion, cells were plated in 12-well plates. For cell proliferation, cells were plated in 96-well plates. Cells were plated at a concentration of 1.0 × 10⁵ cells/mL (Caco-2) and 1.5 × 10⁵ cells/mL (HT-29) and incubated until they reached 80 % confluency (48 h). The fiber fractions were solubilized in water by boiling for 20 min, aliquoted, freeze dried and re-solubilized in growth medium to treatment concentrations of 0.5–3 mg/mL. A yeast derived beta-glucan (Zymosan A Z4250 Sigma) and PMII, a plant polysaccharide fraction from Plantago major L with known immune stimulating activity [35,38,39] were used as positive controls for IL-8 secretion. Cells were incubated 24 h with 1.5 mL (12-well plates) or 100 μL (96-well plates) growth medium or solubilized fiber fractions in duplicate (12-well plates) or triplicate (96-well plates). At the end of the incubation, the plates were processed for measurement of either cell proliferation or IL-8 secretion.

Cell proliferation was determined using the colorimetric MTT assay (Roche Diagnostics GmbH, Mannheim, Germany) that measures the ability of metabolic active cells to cleave tetrazolium sodium salt to purple formazan crystals [54]. The resulting purple precipitate in each cell was dissolved in 100 μL isopropanol containing 0.04 M HCl, and the absorbance measured at 562 nm using Titertek Multiscan plus MK II plate reader (Labsystems, Finland). IL-8 concentrations in the cell culture supernatants were determined using an enzyme linked immunosorbent assay (ELISA). Monoclonal mouse anti-human IL-8 antibody (BD Bioscience Pharmingen, San Diego, CA) suspended in coating buffer (0.1 M Carbonate/Bicarbonate buffer pH 9.6) was added to MaxiSorp^TM ELISA plates (Nunc, Roskilde, Denmark) and incubated over night at 4 °C. Plates were washed three times with PBS containing 0.01% Tween-20 and unspecific binding-sites were blocked by incubating with 5% BSA in PBS for 1 h at room temperature. After washing five times with PBS-Tween, samples and human recombinant IL-8 standards (BD Bioscience Pharmingen,) diluted in working strength high performance ELISA (HPE) buffer from Sanquin (Amsterdam, Netherlands) were added to the plates, which were then incubated for 1.5 h at room temperature followed by washing five times with PBS-Tween. Plates were then incubated for 1 h with biotinylated mouse anti-human IL-8 monoclonal antibody (BD Bioscience Pharmingen) in HPE buffer. After another washing step streptavidin-horseradishperoxidase conjugate (BD Bioscience Pharmingen) in HPE buffer was added and incubated at room temperature for 30 min. Plates were then washed five times with 30 sec between each wash. Color developed after addition of 3,3′,5,5′-tetramethylbenzidine (Sigma-Aldrich) in 0.05 M Phosphate-Citrate-Buffer containing H₂O₂. After 10 min the reaction was stopped by addition of 1 M H₂SO₄, and absorbance was measured at 450 nm using the Titertek Multiscan plus MK II plate reader (Labsystems, Finland). The detection limit of the IL-8 ELISA was 2 pg/mL.

In order to measure NF-κB activity the U937-3xκB-LUC cell line were transferred to RPMI medium with 2 % fetal bovine serum and seeded out in 96 well plates. The fiber fractions extracted from the barley variety Tyra were mixed with highly purified water (Milli-Q, 18.2 MΩ) to a final concentration of 1 mg/mL in Precellys CK14 homogenization tubes (Bertin Technologies, Montigny le Bretonneux, France), solubilized using the Precellys 24 homogenizer (Bertin Technologies) followed by boiling the samples for 5 min, and then freeze dried. The resulting freeze dried fiber fractions were dissolved in medium with 2% serum to a concentration of 4 mg/mL. PMII from Plantago major L. leaves was used as positive control [35,38,39]. This final solution of the fiber fractions was diluted directly in the wells giving the final concentrations of 0.1, 0.2 and 0.4 mg/mL. To measure basal NF-κB activity, cells were incubated with fiber fractions or vehicle control for 6.5 hours. To measure lipopolysaccharide (LPS)-induced NF-κB activity, cells were pre-incubated with fiber fractions or vehicle control for 30 min, then 1 μg/mL lipopolysaccharide isolated from E. coli 0111:B4 (Sigma-Aldrich, St. Louis, MO, U.S.) was added to the cells and the incubation continued for 6 hours. Cell viability for these cells was determined with the use of the CellTiter-Glo Luminiscent Cell Viability Assay (Promega, Madison, WI, U.S.) with cut-off value of 10% non-viable cells. The NF-κB activity was determined by measuring the luciferase activity after addition of Bright-Glo^TM Reagent (Pomega, Madison, WI, U.S.) in accordance to the manufacturer’s instructions. Luminescence was detected for 1 sec using the Glomax96 Microplate Luminometer (Promega, Madison, WI, U.S.).

Human complement proteins were incubated with fiber fractions that might either activate or inhibit activation of the complement proteins. In both situations complement activity is depleted with a negative influence on a balanced hemolysis system involving antibody-sensitized sheep red blood cells and a human serum diluted to give 50% hemolysis. The degree of hemolysis was measured as absorbency at 405 nm. Fiber fractions were tested in triplicates using PMII, a polysaccharide fraction from Plantago major L. as positive control [35,39].

Analysis of significant differences was tested by one-way analysis of variance (ANOVA) with Dunnett’s comparisons with a control and Pearson correlation analysis using Minitab Version 16. Differences were considered significant when p < 0.05.