Aggregation and gelation of oat β-glucan in aqueous solution probed by NMR relaxometry

Carbohydrate Polymers 163 (2017) 170–180

Contents lists available at ScienceDirect

Carbohydrate Polymers journal homepage: www.elsevier.com/locate/carbpol

Aggregation and gelation of oat ˇ-glucan in aqueous solution probed by NMR relaxometry Jia Wu ∗ , Lan Zhao, Jian Li, Shanshan Jin, Shanshan Wu College of Biological Science and Engineering, Fuzhou University, Fuzhou, Fujian 350116, China

a r t i c l e

i n f o

Article history: Received 15 August 2016 Received in revised form 8 January 2017 Accepted 18 January 2017 Available online 21 January 2017 Keywords: Oat ˇ-glucan Structure Aggregation Gelation NMR Transverse relaxation

a b s t r a c t Nuclear magnetic resonance was used to study water proton transverse relaxation in oat ˇ-glucan aqueous solutions during storage. The molecular weights of the samples ranged from 112 to 494 kDa. The polysaccharide structure was identified by IR, GC–MS, NMR spectroscopy, and high performance anion exchange chromatography. In 1% solutions, 494 kDa ˇ-glucan formed stable solution phase. Aggregates were generated in the solution of low molecular weight polysaccharide. The transverse relaxation time of water associated with the aggregate microphase was about 15 ms. The majority of the solution was bulk water, with a relaxation time of 800–1600 ms. The proton exchange rate was about 103 s−1 , calculated from two-site exchange model. The proportion of ˇ-glucan in the aggregate microphase increased from 0 to over 80% of the total ˇ-glucan during storage of 112 kDa samples. In the 4% solutions, the evolution of water proton relaxation associated with the gel network can be explained by the fractal structure theory. The relaxation behavior suggested the growth and syneresis of the gel network during the gelation of the low molecular weight ˇ-glucan solutions. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction Oat ˇ-glucan is a linear polysaccharide of d-glucopyranosyl residues connected with ˇ-(1 → 3) and ˇ-(1 → 4) glycosidic bonds (Wood, Weisz, & Blackwell, 1994). The reported molecular weight of oat ˇ-glucan ranged from 60 to 3000 kDa (Lazaridou & Biliaderis, 2007). The oligosaccharide fragments released from oat ˇ-glucan by lichenase (EC 3.2.1.73) are often considered as building blocks of the intact ˇ-glucan chain. The hydrolysate contains two major oligosaccharide segments, 3-O-ˇ-cellobiosyl-d-glucose (DP3) and 3-O-ˇ-cellotriosyl-d-glucose (DP4). As a soluble dietary fiber, oat ˇglucan has attracted considerable interest due to its beneficial role in human nutrition (Tosh, 2013; Whitehead, Beck, Tosh, & Wolever, 2014). The health benefits of oat ˇ-glucan are dependent on the polysaccharide structure, solution viscosity, and gelation behavior. It is of great importance to gain deeper insight into the state of the polysaccharide in aqueous system. Oat ˇ-glucan tends to form aggregate in aqueous solution. The solution may convert into a gel if the concentration is high enough. The aggregate structure of oat ˇ-glucan in solution and gel has been studied with static and dynamic light scattering (Li, Cui, Wang, &

∗ Corresponding author. E-mail addresses: [email protected], [email protected] (J. Wu). http://dx.doi.org/10.1016/j.carbpol.2017.01.070 0144-8617/© 2017 Elsevier Ltd. All rights reserved.

Yada, 2011; Varum, Smidsrod, & Brant, 1992), rheometry (Doublier & Wood, 1995; Gómez, Navarro, Gamier, Horta, & Carbonell, 1997), atomic force microscopy (Wu et al., 2006), and confocal microscopy (Moschakis, Lazaridou, & Biliaderis, 2014). The molecular weight, structure, and concentration influence the aggregation state of oat ˇ-glucan in aqueous solutions. Low molecular weight oat ˇ-glucans are more prone to form aggregates and gel than high molecular weight ones (Böhm & Kulicke, 1999). Consecutive DP3 units in oat ˇ-glucan chain are involved in the formation of stable junction zones leading to the aggregation and gelation of the polysaccharide solution (Tosh, Brummer, Wood, Wang, & Weisz, 2004). The ratio of DP3 to DP4 reflects the abundance of DP3. An increase in DP3/DP4 ratio means the rise in the amount of consecutive DP3 units. In the concentrated solution, the gelation rate of ˇ-glucan increases with the concentration of polysaccharide (Papageorgiou, Lakhdara, Lazaridou, Biliaderis, & Izydorczyk, 2005). Time-domain 1 H NMR has been extensively used in the investigation of polysaccharide solution and gel (Okada, Matsukawa, & Watanabe, 2002; Paradossi, Cavalieri, & Crescenzi, 1997; Wu et al., 2016). The transverse relaxation time, T2 , can provide useful information on dynamics and aggregate structure of polymers (Zhang, Matsukawa, & Watanabe, 2004). Addition of polysaccharide into water usually increases aqueous proton transverse relaxation rate, 1/T2 . Rapid proton exchange between water and polysaccharide hydroxyl groups is an important transverse relaxation mechanism

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

in polysaccharide aqueous solution and gel. The water proton transverse relaxation time can be calculated using a two-site exchange model (Carver & Richards, 1972; Hills, Wright, & Belton, 1989). If water protons are in fast diffusional exchange between all the microenvironments on the NMR time scale, an averaged transverse relaxation time will be observed. When the diffusional exchange is slow, water existing in different environments will give rise to a multiexponential relaxation (Hills, Cano, & Belton, 1991). In a previous study, we have used proton NMR relaxation to probe the cryogelation of concentrated oat ˇ-glucan solutions (Wu et al., 2016). Three groups of water in different environments were identified. The evolution of transverse relaxation time indicated the formation of large pores left by ice crystals after freezing and thawing cycles. To the best of our knowledge, the gelation of oat ˇ-glucan above freezing point has never been studied by NMR relaxation, though it is a more common gelation phenomenon. It is necessary to probe the formation of oat ˇ-glucan aggregates and the growth of gel network in the ˇ-glucan aqueous system by NMR relaxometry. The dependence on concentration and molecular weight will be discussed through proton NMR relaxation based on a two-site exchange model and water compartment concept. 2. Materials and methods 2.1. Isolation and purification of oat ˇ-glucan Oat ˇ-glucan was extracted from Weiduyou 1 oat cultivar using the method described by Lazaridou, Biliaderis, Micha-Screttas, and Steele (2004), with minor modification. Briefly, the oat bran was treated with hot 80% ethanol solution, then washed with absolute ethanol. After air-dried, the bran was used for ˇ-glucan extraction with 52 ◦ C water, involving a thermostable ␣-amylase and a pancreatin digestion. The ˇ-glucan was precipitated with ethanol from the solution. Then the precipitate was solubilized with water and lyophilized. 2.2. Partial hydrolysis with acid Mild acid hydrolysis was used to obtain low molecular weight polysaccharide, as described by Tosh, Wood, Wang and Weisz (2004). Two grams of oat ˇ-glucan was dissolved in 200 mL deionized water. The solution was heated to 70 ◦ C with magnetic stirring. Concentrated HCl solution was added to bring acid concentration to 0.1 M. Samples were hydrolyzed at 70 ◦ C for 15–90 min, quickly cooled to room temperature and neutralized with NaOH solution. Two volumes of ethanol was added to precipitate partially hydrolyzed oat ˇ-glucan. The precipitates were solubilized in water, then dialyzed in a membrane tubing with a molecularweight cutoff 10 kDa to remove salt. The dialyzed solutions were lyophilized.

171

The methylated polysaccharides were hydrolyzed with 2 M trifluoroacetic acid for 90 min at 121 ◦ C in a fan-forced oven, the hydrolysate was reduced with sodium borohydride and acetylated with acetic anhydride at 100 ◦ C for 2.5 h. The partially methylated alditol acetates were qualitatively analyzed by a 7890B/5977A GC–MS (Agilent, USA) equipped with a HP-5MS capillary column (30 m × 0.25 mm × 0.25 ␮m). The oven was set to an initial temperature of 140 ◦ C, hold for 2 min and then ramp at 6 ◦ C/min to 320 ◦ C and hold for 3 min. After identification of the partially methylated alditol acetates by GC–MS, the components were quantified by gas chromatography equipped with a FID detector using the same chromatographic conditions. The effective response factors, as given by Sweet, Shapiro and Albersheim (1975), were used for calculation of the molar ratio of the products determined by gas chromatography. 2.3.4. NMR spectroscopy NMR spectroscopy was performed on a Bruker AVANCE III 500 NMR spectrometer (Bruker, Germany) operating at 27 ◦ C. A 5 mm BBO probe was used. About 10 mg of the samples were dissolved in 1.0 mL D2 O. The signals of residual water protons were suppressed using pre-saturation pulse when 1 H spectra were collected. Two dimensional 1 H–13 C heteronuclear single quantum coherence (HSQC) spectra were acquired using the standard pulse sequence. The chemical shifts were referenced to acetone-d6 at 2.225 ppm for 1 H and 31.00 ppm for 13 C. 2.3.5. DP3/DP4 determination The oat ˇ-glucan sample (10 mg) was suspended in 5 mL of PBS (20 mM, pH 6.5). The solution was heated at 85 ◦ C for 3 h with magnetic stirring. Then it was cooled to 50 ◦ C. After adding 5 units of lichenase, the solution was incubated for 90 min. The hydrolysate was filtered through a 0.45 ␮m membrane and analyzed using a Dionex ICS-5000 HPAEC system (Dionex, USA). A 4 × 250 mm Carbopac PA1 column was used to separate oligosaccharide. The mobile phase and the gradient were as described by Tosh, Brummer et al. (2004). Data analysis was performed using Chromeleon 6.8 software (Dionex, USA). 2.3.6. Molecular weight determination The molecular weights of the samples were determined using the same method as described by Wu et al. (2016). The SECMALLS-RI system was equipped with a 1500 HPLC pump (Scientific Systems, USA), an SB-806 HQ and an SB-804 HQ column (Showa Denko, Japan), a DAWN HELEOS-II multiangle laser light scattering detector, and an Optilab T-rEX refractive index detector (Wyatt, USA). Oat ˇ-glucan solution with a concentration of 1 mg/mL was injected into the system at a volume of 200 ␮L. The solution of 0.1 M NaNO3 was used as mobile phase at a flow rate of 0.6 mL/min. Data were analyzed with ASTRA 6.1 software and dn/dc = 0.146 mL/g was set for oat ˇ-glucan.

2.3. Characterization of oat ˇ-glucan samples 2.3.1. ˇ-glucan content measurement The ˇ-glucan content of the samples was measured with ˇglucan enzymatic assay kit (Megazyme, Ireland). 2.3.2. FT-IR spectroscopy The ˇ-glucan samples were ground with KBr under an infrared lamp and the mixture was pressed into a pellet. The FT-IR spectra were collected on a Nicolet 380 FT-IR spectrometer (Thermo Scientific, USA) with a resolution of 4 cm−1 . 2.3.3. Methylation analysis The oat ˇ-glucan samples were methylated according to the method of Pettolino, Walsh, Fincher, and Bacic (2012) with CH3 I.

2.3.7. Intrinsic viscosity measurement A Ubbelohde viscometer was used to determine the flow time of oat ˇ-glucan solutions at 25.0 ± 0.1 ◦ C. The intrinsic viscosity, [], was obtained by Huggins plot. 2.4. Preparation and storage of oat ˇ-glucan solutions Oat ˇ-glucan samples were dispersed in pure water at a concentration of 1% and 4% (w/w). The solutions were stirred at 85 ◦ C for 3 h, then cooled to room temperature. About 2 g of the polysaccharide solution was transferred into a NMR tube with a diameter of 15 mm and sealed tightly with a plug. The samples in the tubes were allowed to stand at 25 ◦ C in an incubator for 15 days.

172

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

Table 1 Molecular characteristics of oat ˇ-glucan samples. Sample

Hydrolysis time (min)

␤-glucan content (%)

DP3/DP4a

Mw b (kDa)

Mw /Mn c

[]d (mL/g)

OBG494 OBG332 OBG250 OBG178 OBG155 OBG112

0 15 30 45 60 90

88.41 ± 1.26 90.13 ± 1.18 89.73 ± 1.04 89.64 ± 1.23 90.72 ± 1.58 91.80 ± 0.97

2.46 ± 0.08 2.51 ± 0.12 2.42 ± 0.09 2.39 ± 0.07 2.56 ± 0.15 2.49 ± 0.11

494 ± 21 332 ± 16 250 ± 13 178 ± 10 155 ± 8 112 ± 5

1.13 ± 0.05 1.08 ± 0.04 1.25 ± 0.03 1.13 ± 0.03 1.40 ± 0.06 1.32 ± 0.03

514 ± 0.53 348 ± 0.20 294 ± 0.18 218 ± 0.15 197 ± 0.13 158 ± 0.12

a b c d

Molar ratio, calculated from peak area of DP3/DP4 × 1.321. Weight average molecular weight. Polydispersity index. Intrinsic viscosity.

2.5. Proton NMR relaxation measurements Proton transverse relaxation was measured using a MiniMR NMR spectrometer (Niumag, China) operating at a 1 H resonance frequency of 23 MHz. The transverse relaxation time T2 was obtained using a Carr–Purcell–Meiboom–Gill (CPMG) pulse sequence. The 90◦ pulse was 18 ␮s, the interval between 90◦ and 180◦ pulse was 100 ␮s, and 18000 echoes were collected. Eight scans were performed with a delay time of 10 s. Niumag NMR analysis software (Niumag, China) was used to collect and analyze transverse magnetization decay signal. The signal curves were fitted to exponential decays using the equation A (t) =

 i



Ai exp −

t T2i



+ A0

(1)

where A(t) is the echo amplitude at time t, Ai and T2i are the amplitude and transverse relaxation time of the component i, and A0 is the noise of the curves.

2.6. Apparent viscosity measurements The apparent viscosity of oat ˇ-glucan solutions was measured with a Physica MCR-301 rheometer (Anton Paar, Austria) at 25 ◦ C using the cone plate geometry with a 50 mm diameter plate (CP501). The shear rate increased from 0.01 to 100 s−1 .

2.7. Statistical analysis All the measurements were performed in triplicates. The statistical analysis was carried out using SPSS 19.0 software. The difference between the measurements was evaluated using Oneway ANOVA with a significance level of p < 0.05.

3. Results and discussion 3.1. Purity of oat ˇ-glucan samples For convenience, the original and partially hydrolyzed samples are termed as OBG494 to OBG112 according to their molecular weights. The ˇ-glucan content of the samples was near 90% (Table 1), there was no significant difference (P > 0.05) between the samples. That is to say, the purity was not affected by partial acid hydrolysis (Tosh, Wood et al., 2004). The major impurity was protein (about 2.5%) and starch (less than 2%) (Wu et al., 2016). The arabinose and xylose were detected by GC–MS and they were about 0.6% of the polysaccharide (Supplementary material Fig. S1), indicating the existence of arabinoxylan in the ˇ-glucan samples (Izydorczyk, Macri, & MacGregor, 1998).

Fig. 1. FT-IR spectra of oat ˇ-glucan samples.

3.2. Structure of oat ˇ-glucan 3.2.1. FT-IR All the samples had similar IR bands (Fig. 1), indicating the similar groups in these polysaccharide samples. The broad bands at 3600–3000 cm−1 were assigned to O H stretching. The bands at 2922 cm−1 were the stretching of C H. The bands at 1638 cm−1 derived from C O stretching and N H bending of residual proteins as well as O H bending of water molecules. The strongest bands at 1200–1000 cm−1 were assigned to C O C vibrations of glucose ring and glycosidic bond, and the stretching of C OH (Virkki, Johansson, Ylinen, Maunu, & Ekholm, 2005). The sharp absorption at 896 cm−1 indicated the ˇ-linkages of the polysaccharide. 3.2.2. Linkage analysis The composition of methylated polysaccharide was identified by MS (Table 2). The ˇ-glucan samples contained mainly ˇ-(1 → 3) and ˇ-(1 → 4) linkages, with a small amount of terminal glucose units (Supplementary material Fig. S2). The MS spectra of partially methylated alditol acetates and the fragmentation patterns were presented in Supplementary material Fig. S3. The calculated ratio of ˇ-(1 → 4)/ˇ-(1 → 3) linkages was between 2.54 to 2.69. The results showed that all the samples had the similar chemical structure. 3.2.3. NMR The 13 C NMR spectra of oat ˇ-glucan samples (Fig. 2) confirmed the structure of mixed-linkage (1 → 3),(1 → 4)-ˇ-d-glucan as assigned by Dais and Perlin (1982). The assignments are shown in Table 3. The C1 of the residue 4G3 had a resonance at 102.8 ppm due to the influence of (1 → 3)-ˇ-linkage. The resonance at 102.6 ppm was assigned to C1 involved in the formation of (1 → 4)-ˇ-linkage in 3G4 and 4G4 residues. The 84.1 ppm singlets of C3 in 3G4 residues suggested the single occurrence of the (1 → 3)-ˇ-linkage. The res-

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

173

Table 2 Glycosidic linkage (mol%) of oat ˇ-glucan samples. Glycosidic linkage

Linkage composition (mol%) OBG494

OBG332

OBG250

OBG178

OBG155

OBG112

(Glcp)1→ →3(Glcp)1→ →4(Glcp)1→ (1 → 4)/(1 → 3)

1.60 27.10 71.31 2.63

1.51 26.67 71.83 2.69

1.24 27.56 71.20 2.58

0.92 27.06 72.01 2.66

1.43 27.07 71.50 2.64

2.15 27.65 70.20 2.54

Fig. 2.

13

C NMR spectra of oat ˇ-glucan samples (in D2 O, 500 MHz, 27 ◦ C).

onances of C4 in 4-O-substituted residues shifted to lower field comparing with that in 3-O-substituted residues. The C6 in 3-Osubstituted and 4-O-substituted residues had a resonance at 60.8 and 60.2 ppm, respectively. These assignments are in good accordance with the pervious literature that the samples were dissolved in D2 O (Colleoni-Sirghie, Fulton, & White, 2003; Ghotra, Vasanthan, & Temelli, 2008; Roubroeks, Mastromauro, Andersson, Christensen, & Åman, 2000). The ratio of (1 → 4) to (1 → 3)-linkages could be calculated by integration of the anomeric signals of 4-O-substituted and 3-O-substituted residues. The ratio was between 2.54 to 2.69 for all the samples, similar to the ratio (2.3) estimated from 13 C NMR signals by Dais and Perlin (1982).

Table 3 Assignment of 13 C NMR spectra of oat ˇ-glucan samples. Sugar residue

→4)-ˇ-Glcp(1 → 3) →3)-ˇ-Glcp(1 → 4) →4)-ˇ-Glcp(1 → 4)

Chemical shift, ppm C1

C2

C3

C4

C5

C6

102.8 102.6 102.6

73.5 73.2 73.3

74.3 84.1 74.3

78.7 68.2 78.7

75.1 75.8 75.1

60.2 60.8 60.2

The 1 H spectra of the samples were collected at 27 ◦ C, the negative signals near 4.70 ppm were due to the water suppression (Fig. 3). The 1 H NMR signals were assigned by referring to the reported data (Colleoni-Sirghie et al., 2003; Roubroeks, Andersson, & Åman, 2000). At 27 ◦ C, The H1 of 4G3 residue was overlapped by water at 4.70 ppm. The group of overlapped signals at 4.45 ppm (27 ◦ C) corresponded to H1 in 3G4 and 4G4 residues. When the temperature was raised to 50 ◦ C, the water signal moved to 4.40 ppm as a negative peak due to water suppression in the spectrum of OBG178HT, the H1 of 4G3 residue appeared at 4.70 ppm as a doublet, the H1 signals of 3G4 and 4G4 at 4.45 ppm were greatly reduced as a result of water suppression. Collecting proton signals at 50 ◦ C resulted in higher resolution of 1 H spectrum (OBG178HT in Fig. 3) compared with that obtained at 27 ◦ C (OBG178 in Fig. 3). Because the NMR of cereal ˇ-glucan was often performed above 60 ◦ C, the 1 H spectra in the literature had better resolution than ours (Colleoni-Sirghie et al., 2003; Roubroeks, Andersson et al., 2000; Roubroeks, Andersson, Mastromauro, Christensen, & Åman, 2001). Another advantage of collecting proton signals above 60 ◦ C was the avoidance of overlapped anomeric proton and water proton. It should be noted that the 1 H spectrum of OBG178HT (50 ◦ C) was aligned to the other spectra (27 ◦ C) by shifting to the lower field.

174

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

Fig. 3.

1

H NMR spectra of oat ˇ-glucan samples (in D2 O, 500 MHz, 27 ◦ C). The spectrum of OBG178HT was collected at 50 ◦ C for the sample OBG178.

The group of signals between 3.91 and 3.68 ppm corresponded to H6 of the glucose residues. H3, H4 and H5 signals were overlapped at 3.66, 3.58, 3.54 and 3.43 ppm. The group of signals between 3.32 and 3.26 ppm corresponded to H2. The signal intensity of the 1 H spectra increased with the decrease of ˇ-glucan molecular weight. With the hydrolysis of the polysaccharide, the weak signals corresponding to the reducing ends anomeric ˇ-configuration appeared at 4.58 ppm, and the anomeric ␣-signals appeared at 5.14 ppm, those weak signals were enlarged for clarity, as shown in the spectrum of OBG178HT (Fig. 3). The presence of the terminal anomeric signals reflected the partial hydrolysis of the polysaccharide (Colleoni-Sirghie et al., 2003; Roubroeks et al., 2001). But the intensity of these terminal signals was so weak, indicating a very small amount of oligosaccharide fragments in the samples. It was probable that the oligosaccharide fragments released by acid hydrolysis cannot be precipitated by ethanol or they lost during dialysis. So our polysaccharide samples lacked oligosaccharide fragments compared to the partially hydrolyzed ˇ-glucans used by other researchers. The anomeric 1 H signals of our samples were more similar to intact cereal ˇ-glucans rather than hydrolyzed ˇ-glucans with small fragments (ColleoniSirghie et al., 2003; Roubroeks, Andersson et al., 2000; Roubroeks et al., 2001). The typical 1 H–13 C HSQC spectrum of the oat ˇ-glucan sample (Fig. 4) showed two cross peaks to the H1 of 4G3 (4.71 ppm), 3G4 and 4G4 (4.46 ppm), excluding the effect of residual water. The correlated carbons and protons were annotated for cross peaks in Fig. 4. The other samples had the similar HSQC spectra, indicating these samples had the similar structure, in spite of the difference in their molecular weight. The HSQC spectra of oat ˇ-glucan samples were similar to those of cereal ˇ-glucans from rye and oat (Colleoni-Sirghie et al., 2003; Roubroeks, Andersson et al., 2000).

3.2.4. DP3/DP4 Partial acid hydrolysis caused a reduction in polysaccharide molecular weight, but the molar ratio of DP3/DP4 was not affected. There was no significant difference in DP3/DP4 ratio (P > 0.05) for all the ˇ-glucan samples (Table 1). The lack of change in the DP3/DP4 ratio of partially acid hydrolyzed oat ˇ-glucan was also reported by other researchers (Agbenorhevi, Kontogiorgos, Kirby, Morris, & Tosh, 2011; Brummer et al., 2014).

3.3. Molecular weight distribution of the samples The weight average molecular weight (Mw ) and intrinsic viscosity ([]) of the samples decreased significantly (P < 0.05) with the hydrolysis time (Table 1). Although it was reported that the polydispersity index (Mw /Mn ) of the ˇ-glucan decreased with hydrolysis time (Roubroeks et al., 2001; Tosh, Wood et al., 2004), there was no such trend in our results. The discrepancy is probably due to the additional dialysis step after partial hydrolysis of the samples. Some low molecular weight polysaccharide would be separated from the hydrolysate by dialysis. The detailed information on molecular weight distribution is shown in Fig. 5a. Actually, a wider distribution of the molecular weight is presented in the RI elution profiles of low molecular weight samples. The molecular weight distribution curves from MALLS are overlapped on a line when log (molecular weight) is used, which is expected for a homologous series only differing in the molecular weight distribution. A small upward curvature in the high molecular weight part of each curve suggests aggregates eluting early in each peak (Roubroeks, Mastromauro et al., 2000). As shown in Fig. 5b, there is a linear increment in 1/Mw with the acid hydrolysis time, indicating a random degradation of the polysaccharide at a constant rate (Roubroeks et al., 2001).

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

Fig. 4.

1

175

H–13 C HSQC of OBG178 (in D2 O, 500 MHz, 27 ◦ C).

Fig. 6. T2 distribution of 1% OBG494 and OBG112 solutions, fresh and stored for 15 days.

3.4. Transverse relaxation in dilute solution

Fig. 5. (a) RI elution profiles of the ˇ-glucan samples and molecular weight distribution for each fraction determined by SEC-MALLS-RI system. (b) The relationship between 1/Mw of partially acid hydrolyzed ˇ-glucan and hydrolysis time.

The NMR relaxation measurements were made on 1% oat ˇglucan solutions. The concentration was below the critical overlap concentration, 4/[], except for OBG494 (Robinson, Ross-Murphy, & Morris, 1982). So these solutions were in the dilute solution regime or near the regime. The transverse relaxation time, T2 , of the polysaccharide solution was measured for four samples, OBG494, OBG250, OBG155, and OBG112. OBG494 solution presented one transverse relaxation time component in both fresh and stored solutions (Fig. 6), and the T2 distribution was nearly overlapping, indicating that the solution structure of OBG494 experienced no change during the storage. The T2 of OBG494 was 1.072 s, corresponding to the water in bulk phase (Hills et al., 1991; Wu

176

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

et al., 2016). The monoexponential relaxation behavior suggested the protons were in fast exchange between water and ˇ-glucan hydroxyl groups. In the limit of very short 90◦ –180◦ pulse spacing (100 ␮s for our instrument), the observed transverse relaxation time, T2obs , is given by two-site exchange expression (Hills et al., 1989)



1/T2obs = Pa /T2a + Pb / T2b + kb−1



(2)

where Pa and Pb are the fraction of water proton and labile proton on polysaccharide. T2a and T2b are the intrinsic transverse relaxation time for water proton and polysaccharide labile proton, kb is the proton exchange rate. In 1% OBG494, T2obs is 1.072 s, Pa is approximately 1, T2a is usually assumed to be close to that of bulk water (about 2 s). Considering the number of exchangeable protons per glucose residue (3/162) and that of water (2/18), 1% (w/w) ˇ-glucan solution gives a Pb value of 1.664 × 10−3 . T2b can be obtained indirectly by assuming that it is the same as T2 of the nonexchanging CH protons of the ˇ-glucan. The T2 of CH protons was determined in 1% OBG494 D2 O solution using CPMG pulse sequence, with 50 echoes and 128 scans due to short relaxation time and low concentration of CH protons. The obtained T2 of CH protons was 1.02 ms. Then, the exchange rate, kb , can be estimated as 354 s−1 using Eq. (2). Since the diffusive exchange rate of bound and free water is many orders of magnitude larger than kb , the two-site exchange model takes no account of possible bound water contributions (Hills et al., 1991). The fresh 1% OBG112 also shows a single T2 component (Fig. 6), so it is possible to estimate kb using the same method. T2obs is 0.811 s, T2b is 1.48 ms determined from OBG112 D2 O solution. The calculated kb is 1266 s−1 for fresh OBG112. The values of kb are within the range expected for carbohydrates (Hills et al., 1991, 1989; Kerr & Wicker, 2000). The estimated exchange rate of fresh OBG112 is larger than that of OBG494, the exact reason is not clear (Fabri, Williams, & Halstead, 2005), but might be postulated to be due to difference in structure or degree of aggregation (Kerr & Wicker, 2000). Two transverse relaxation time components appeared in the hydrolyzed ˇ-glucan solutions stored for 1–15 days. As a representative, the T2 distribution of 1% OBG112 stored for 15 days is shown in Fig. 6. The slow relaxation component (1.632 s) of OBG112 is near the T2 of OBG494, associated with bulk water relaxation. The fast relaxation component (12.93 ms, Figs. 6 and 7a) is not corresponding to the CH protons, because the T2 of CH protons determined in D2 O is near 1 ms. The proportion of this water component is over 1% of the total water in OBG 112 solution, as reflected by the spin density A21 (Fig. 7b), which is higher than the expected proportion of CH protons (about 0.4%) from chemical structure. Oat ˇ-glucans tend to form aggregates in aqueous solution. The construction of the aggregates is fast and they can grow into larger clusters (Li et al., 2011). Based on this fact, the fast relaxation component is assigned to water associated with the ˇ-glucan aggregates. The transverse relaxation component corresponding to water in ˇ-glucan aggregates cannot be detected in the freshly prepared ˇ-glucan solutions, just as shown in Fig. 7a, there is no T21 component for all the samples at day 0. With the growth of the aggregates, the T21 water can be detected after stored for one day. The gradual decrease of T21 in the initial days is mainly ascribed to the increase of ˇ-glucan concentration in the aggregates. As mentioned above, the slow relaxation component, T22 , is corresponding to bulk water. The T22 of OBG494 kept stable during storage. The partially hydrolyzed samples had an ascending trend in T22 until stored for 9 days, then the T22 leveled off in the rest days (Fig. 7c). It should be noted that the proportion of this water component was always nearly 100% for all the samples (Fig. 7d), indicating most water was in the solution phase. Just like the estimation of relaxation parameters for OBG494, it is possible to give a similar analysis for partially hydrolyzed samples. OBG112 is selected as an example. Two water components exist in 1% OBG112 solution.

Fig. 7. Evolution of transverse relaxation time (a, c) and spin density (b, d) with time in 1% oat ˇ-glucan solutions varying in molecular weight.

We first assume that the two relaxation times T2b (1) and T2b (2) are associated with oat ˇ-glucan hydroxyl protons in aggregates and solution. The two separated microphases are labeled (1) and (2). Then the observed relaxation times, T21 and T22 , of 1% OBG112 can be wrote as (Hills et al., 1991) (1)

(1) Pb P 1 = a + (1) T21 T2a T2b + kb−1

(3)

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

177

3.5. Transverse relaxation in concentrated solution and gel

Fig. 8. Variation of ˇ-glucan mass in aggregate phase (m1 ) with time for 1% OBG112, assuming water is 100 g in the aqueous system.

(2)

(2) Pb 1 P = a + (2) T22 T2a T2b + kb−1

(4)

In Eqs. (3) and (4), Pa (1) and Pb (1) are the fraction of water protons and ˇ-glucan hydroxyl protons in the aggregate microphase, Pa (2) and Pb (2) are the corresponding proton fractions in the solution microphase. It is assumed that the exchange rate, kb , is the same in the two microphases. In 1% OBG112 stored for one day, T22 is 0.977 s, Pa (2) is approximately 1, T2a of bulk water is 2 s. T2b (2) is the same as T2b in fresh solution, that is 1.48 ms determined in D2 O. The exchange rate is supposed to be equal to that of fresh 1% OBG112, it is 1266 s−1 . The fraction of ˇ-glucan hydroxyl protons in solution microphase, Pb (2) , is estimated to be 1.19 × 10−3 through Eq. (4). The Pb (2) is defined as (2)

Pb

=

3 162 3 162

× m2

× m2 +

2 18

× 100

(5)

where m2 is the mass of ˇ-glucan in solution phase, 100 g is the mass of water. The calculated m2 is 0.713 g. Because the total ˇglucan mass is 1 g in the aqueous system, the mass of ˇ-glucan in the aggregate microphase, m1 , is 0.287 g. The water associated with aggregates is estimated to be 1.593 g from its proportion A21 (Fig. 7b). The fraction of ˇ-glucan hydroxyl protons in aggregate microphase, Pb (1) , is 0.029 according to its definition. The fraction of water protons in the aggregate microphase, Pa (1) = 1 − Pb (1) = 0.971. The observed transverse relaxation time, T21 is 0.036 ms. Using Eq. (3), The transverse relaxation time of ˇ-glucan hydroxyl protons in aggregates, T2b (1) is calculated to be 0.278 ms, which is shorter than that of ˇ-glucan hydroxyl protons in solution phase, T2b (2) (1.48 ms). The shorter transverse relaxation time suggests the decreased mobility of the protons on polysaccharide chain (Hills et al., 1991). In oat ˇ-glucan aggregates, the polysaccharide chains associate with each other through hydrogen bonding, resulting in decreased mobility (Li et al., 2011; Wu et al., 2006). The mass of ˇglucan in solution phase, m2 , can be derived from Eqs. (4) and (5). Then the mass of ˇ-glucan in aggregate phase, m1 , can be obtained using Eq. (3) and the similar definition as Eq. (5). The variation of m1 with time is shown in Fig. 8 for 1% OBG112. It is clear that the ˇglucan gradually transferred to the aggregate microphase, leading to decreased polysaccharide concentration in solution phase and a rise in T22 of bulk water (Fig. 7c). The monoexponential relaxation behavior of 1% OBG494 indicated that the OBG494 molecules were mainly in solution phase. The hydrolyzed samples had two relaxation components due to the formation of ˇ-glucan aggregates. This is in accordance with the previous results that the degree of aggregation decreased for higher molecular weight ˇ-glucans due to lower diffusion rate (Li et al., 2011).

The 4% ˇ-glucan solutions were above the critical overlap concentration, 4/[], belonging to the concentrated solution regime (Robinson et al., 1982). In concentrated solutions, ˇ-glucan chains overlap and are entangled with each other, offering increased possibility for the formation of aggregates. As shown in Fig. 9a, the 4% OBG494 presented two relaxation components, the fast relaxation component was corresponding to water in the aggregate microphase. This group of water had a T21 at about 30 ms in OBG494 and 10 ms in hydrolyzed samples. The discrepancy was mainly ascribed to lower ˇ-glucan concentration in the aggregate microphase formed in OBG494 solution. Oat ˇ-glucans form aggregates mainly through physical cross-links between consecutive cellotriosyl segments (Tosh, Brummer et al., 2004). There is more possibility for low molecular weight ˇ-glucans to collide into each other and form aggregates (Li et al., 2011), leading to increased ˇ-glucan concentration and decreased T21 in the aggregate microphase. The spin density, A21 , reflects the proportion of water in the aggregate microphase. All the samples had an A21 approaching 2% (Fig. 9b), slightly higher than that in 1% solutions (about 1%). It seemed that the amount of water associated with the aggregate microphase did not increase in proportion to the ˇ-glucan concentration, probably due to the diffusive exchange between water in the aggregate and solution microphase. The emergence of a new relaxation component, T2g , after 3 days aging (Fig. 9c) made us realize that it was associated with the formation of gel network. Oat ˇ-glucan can form a gel when stored at room temperature (Tosh, Brummer et al., 2004). It usually takes several days for the formation of oat ˇ-glucan gel. The relaxation component corresponding to the gel network appeared in OBG112 and OBG155, but there was no such component in OBG494 and OBG250. These samples had the same concentration and DP3/DP4 ratio, but different molecular weight. The effect of molecular weight on the transverse relaxation time of ˇ-glucan gelation is consistent with the results using rheological methods. It is reported that the gelation rate increased with decreasing molecular weight of ˇglucan (Böhm & Kulicke, 1999). In another research, two of the four replicates of 100 kDa ˇ-glucans gelled after 4 days aging, 200 kDa oat ˇ-glucans remained viscous liquids after 7 days when stored at 5 ◦ C (Tosh, Wood et al., 2004). The T2g increased from about 20 to 100 ms, it came to a decrease after 9 days aging. The application of the fractal structure concept seems to be able to explain the change of T2g . Low molecular weight ˇ-glucan gels gradually form an open network, the structure is prone to rearrangements due to intra- and inter-cluster interactions. The rearrangements lead to collapse of the structure, and finally a consequence of syneresis (Kontogiorgos, Vaikousi, Lazaridou, & Biliaderis, 2006). The more opened network structure led to the rise of T2g , and the syneresis resulted in the decrease of T2g after 9 days aging. The growth and syneresis of the ˇ-glucan gel network is also reflected by the evolution of the amount of water in the gel network (Fig. 9d). Although there were not much change in the spin density of water in the aggregates, A21 , the spin density of water in the gel network, A2g , increased to 20% and 14% in 4% OBG112 and OBG155. We can infer that the amount of aggregate may not change much, these aggregates associate together to form clusters, leading to the formation of gel network. The amount of water entrapped in the gel network was much more than that in the aggregates. The transverse relaxation time, T22 , of water protons in the solution phase is shown in Fig. 9e. OBG494 had a higher T22 than the other solutions in the initial days, due to slow proton exchange between ˇ-glucan hydroxyl groups and water, the calculated exchange rate is 354 s−1 in OBG494, much lower than that in OBG112 (1266 s−1 ). There was no significant change in T22 of OBG494 and OBG250 during storage, indicating the absence of gel

178

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

Fig. 9. Evolution of transverse relaxation time (a, c, e) and spin density (b, d, f) with time in 4% oat ˇ-glucan solutions varying in molecular weight.

network in the solutions. For OBG155 and OBG112, T22 presented a sharp rise after 6 days aging and then leveled off after 9 days. The increase of T22 suggested the transfer of oat ˇ-glucan from the solution microphase to the gel microphase. The T22 were about 810 and 670 ms for 4% OBG 112 and OBG155 after 15 days aging (Fig. 9e), the corresponding T22 for 1% solutions were about 1600 and 1500 ms. The decreased T22 in 4% ˇ-glucan was mainly ascribed to increased ˇ-glucan concentration in the solution microphase. The spin density, A22 , of OBG494 and OBG250 was close to 100% (Fig. 9f), indicating most of the water was in the solution phase. The formation of gel microphase in OBG155 and OBG112 resulted in the decrease of water in the solution microphase to about 80% (Fig. 9f). It seemed that there were more water in the gel network microphase of OBG112 than that of OBG155. The increased water in the gel microphase suggested increased amount of gel network in OBG112, which was due to higher mobility of lower molecular weight polysaccharide. Concentrated oat ˇ-glucan is a heterogeneous system with multi-microphase. If the dimension a characterizing the scale of heterogeneity is large enough, the ratio (a2 /D) |1/T2 (1) − 1/T2 (2) | > 1, multi-exponential relaxation is predicted (Hills et al., 1991). D is the diffusion coefficient of water, it can be estimated as

2 × 10−11 m2 s−1 (Ibbett, Wortmann, Varga, & Schuster, 2014). T2 (1) and T2 (2) are the transverse relaxation time in different environments. In 4% OBG494 and OBG250, the estimated a is 0.8 and 0.45 ␮m. The aggregates had been observed by confocal microscopy in oat ˇ-glucan solutions, the size of these aggregates was above 1 ␮m (Moschakis et al., 2014). When the 4% OBG112 was aged for 9 days, the dimension a was calculated to be 1.5 ␮m using T2g = 0.1 s and T22 = 0.77 s, the observed size of the network structures reached over 20 ␮m (Moschakis et al., 2014). The inhomogeneous gel structure was also observed by light microscope (Tosh, Brummer et al., 2004). Based on the above analysis, a gel network model is proposed in Fig. 10 using the fractal structure concept (Kontogiorgos et al., 2006). Oat ˇ-glucan chains tend to form aggregates in the aqueous solution, the water associated with the aggregates has a transverse relaxation time, T21 . During the aging of solutions, the aggregates interact with each other forming clusters, the cluster is indicated as fractal floc in the dotted circles (Fig. 10). Then these clusters are connected to each other, growing into gel network. The water entrapped in the meshes of the network has a transverse relaxation time, T2g. Water in solution microphase has a long transverse relaxation time, T22 , approaching that of pure water.

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

179

entangled network structure in the concentrated solution (Morris, Cutler, Ross-Murphy, Rees, & Price, 1981). The aggregates in 4% OBG494 solution are mainly the entangled structures, with minor junction zones formed by consecutive ˇ-(1 → 3)-linked cellotriosyl units. In 1% solutions of OBG250, OBG155, and OBG112, the flow curves exhibited strong shear thinning behavior at low shear rates (Fig. 11a), implying the formation of aggregates. The consecutive cellotriosyl units are responsible for the interchain associations (Doublier & Wood, 1995; Vaikousi, Biliaderis, & Izydorczyk, 2004). But the phenomena were not obvious in the concentrated polysaccharide solutions. The 4% OBG112 and OBG155 formed gel during aging, meanwhile, OBG250 and OBG494 remained viscous liquids after 15 days. 4. Conclusions

Fig. 10. Schematic illustration of the aggregate, the gel network, and the solution microphase of oat ˇ-glucan aqueous system. Water distribution in different environment is indicated. The model is adopted from fractal structure of barley ˇ-glucan gel network with modification (Kontogiorgos et al., 2006).

The oat ˇ-glucan samples with different molecular weight were produced by partial acid hydrolysis. The similar structure of the samples was identified by FT-IR, GC–MS, NMR spectroscopy, and high performance anion exchange chromatography. NMR transverse relaxation can provide useful information to monitor the microstructural evolution of ˇ-glucan solutions with different concentration and molecular weight. In dilute solution regime, high molecular weight ˇ-glucans form stable solutions, aggregates of low molecular weight ˇ-glucans are generated upon storage. The ˇ-glucan chains gradually moved from solution microphase to the aggregate microphase. In concentrated solutions, the formation of gel network structure was detected in low molecular weight ˇ-glucan solutions. The growth and syneresis of the gel network can be explained by the fractal structure model. The dependence of gelation kinetics on molecular weight was characterized by transverse relaxation. The predicted size of the aggregate and gel network microphase using the NMR relaxometry was in accordance with the observed size of the structures in the previous literature. NMR relaxometry was proved to be a valuable tool in the study of polysaccharide solution and gel. Acknowledgment This work was supported by the National Nature Science Foundation of China (31101224). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.carbpol.2017.01. 070. References

Fig. 11. Viscosity dependence on shear rate for oat ˇ-glucans differing in molecular weight at a concentration of 1% (a) and 4% (b).

3.6. Apparent viscosity of the polysaccharide solution The apparent viscosity measurement was performed the day after the preparation of fresh oat ˇ-glucan solutions. The flow curve of 1% OBG494 exhibited typical behavior of random coil type polysaccharide. There was a Newtonian plateau at low shear rate, and the apparent viscosity decreased when the shear rate increased above 1 s−1 (Fig. 11a). But the shear rate at which the apparent viscosity began to decrease moved to about 0.1 s−1 in 4% OBG 494 (Fig. 11b). The decrease of the critical shear rate with the increase of the polysaccharide concentration can be illustrated by a dynamic

Agbenorhevi, J. K., Kontogiorgos, V., Kirby, A. R., Morris, V. J., & Tosh, S. M. (2011). Rheological and microstructural investigation of oat ␤-glucan isolates varying in molecular weight. International Journal of Biological Macromolecules, 49(3), 369–377. Böhm, N., & Kulicke, W.-M. (1999). Rheological studies of barley (1 → 3)(1 → 4)-␤-glucan in concentrated solution: Mechanistic and kinetic investigation of the gel formation. Carbohydrate Research, 315(3–4), 302–311. Brummer, Y., Defelice, C., Wu, Y., Kwong, M., Wood, P. J., & Tosh, S. M. (2014). Textural and rheological properties of oat beta-glucan gels with varying molecular weight composition. Journal of Agricultural and Food Chemistry, 62(14), 3160–3167. Carver, J. P., & Richards, R. E. (1972). A general two-site solution for the chemical exchange produced dependence of T2 upon the carr-Purcell pulse separation. Journal of Magnetic Resonance (1969), 6(1), 89–105. Colleoni-Sirghie, M., Fulton, D. B., & White, P. J. (2003). Structural features of water soluble (1, 3) (1, 4)-␤-d-glucans from high-␤-glucan and traditional oat lines. Carbohydrate Polymers, 54(2), 237–249. Dais, P., & Perlin, A. S. (1982). High-field, 13 C-N.M.R. spectroscopy of ␤-d-glucans, amylopectin, and glycogen. Carbohydrate Research, 100(1), 103–116.

180

J. Wu et al. / Carbohydrate Polymers 163 (2017) 170–180

Doublier, J. L., & Wood, P. J. (1995). Rheological properties of aqueous solutions of (1 → 3)(1 → 4)-␤-d-glucan from oats (Avena sativa L.). Cereal Chemistry, 72(4), 335–340. Fabri, D., Williams, M. A. K., & Halstead, T. K. (2005). Water T2 relaxation in sugar solutions. Carbohydrate Research, 340(5), 889–905. Gómez, C., Navarro, A., Gamier, C., Horta, A., & Carbonell, J. V. (1997). Physical and structural properties of barley (1→3),(1→4)-␤-d-glucan—III. Formation of aggregates analysed through its viscoelastic and flow behaviour. Carbohydrate Polymers, 34(3), 141–148. Ghotra, B. S., Vasanthan, T., & Temelli, F. (2008). Structural characterization of barley ␤-glucan extracted using a novel fractionation technique. Food Research International, 41(10), 957–963. Hills, B. P., Wright, K. M., & Belton, P. S. (1989). Proton N.M.R. studies of chemical and diffusive exchange in carbohydrate systems. Molecular Physics, 67(6), 1309–1326. Hills, B. P., Cano, C., & Belton, P. S. (1991). Proton NMR relaxation studies of aqueous polysaccharide systems. Macromolecules, 24(10), 2944–2950. Ibbett, R., Wortmann, F., Varga, K., & Schuster, K. C. (2014). A morphological interpretation of water chemical exchange and mobility in cellulose materials derived from proton NMR T2 relaxation. Cellulose, 21(1), 139–152. Izydorczyk, M. S., Macri, L. J., & MacGregor, A. W. (1998). Structure and physicochemical properties of barley non-starch polysaccharides—I. Water-extractable ␤-glucans and arabinoxylans. Carbohydrate Polymers, 35(3), 249–258. Kerr, W. L., & Wicker, L. (2000). NMR proton relaxation measurements of water associated with high methoxy and low methoxy pectins. Carbohydrate Polymers, 42(2), 133–141. Kontogiorgos, V., Vaikousi, H., Lazaridou, A., & Biliaderis, C. G. (2006). A fractal analysis approach to viscoelasticity of physically cross-linked barley ␤-glucan gel networks. Colloids and Surfaces B: Biointerfaces, 49(2), 145–152. Lazaridou, A., & Biliaderis, C. G. (2007). Molecular aspects of cereal ␤-glucan functionality: Physical properties, technological applications and physiological effects. Journal of Cereal Science, 46(2), 101–118. Lazaridou, A., Biliaderis, C. G., Micha-Screttas, M., & Steele, B. R. (2004). A comparative study on structure–function relations of mixed-linkage (1 → 3), (1 → 4) linear ␤-d-glucans. Food Hydrocolloids, 18(5), 837–855. Li, W., Cui, S. W., Wang, Q., & Yada, R. Y. (2011). Studies of aggregation behaviours of cereal ␤-glucans in dilute aqueous solutions by light scattering: Part I. Structure effects. Food Hydrocolloids, 25(2), 189–195. Morris, E. R., Cutler, A. N., Ross-Murphy, S. B., Rees, D. A., & Price, J. (1981). Concentration and shear rate dependence of viscosity in random coil polysaccharide solutions. Carbohydrate Polymers, 1(1), 5–21. Moschakis, T., Lazaridou, A., & Biliaderis, C. G. (2014). A micro- and macro-scale approach to probe the dynamics of sol–gel transition in cereal ␤-glucan solutions varying in molecular characteristics. Food Hydrocolloids, 42(Part 1), 81–91. Okada, R., Matsukawa, S., & Watanabe, T. (2002). Hydration structure and dynamics in pullulan aqueous solution based on 1 H NMR relaxation time. Journal of Molecular Structure, 602–603(0), 473–483. Papageorgiou, M., Lakhdara, N., Lazaridou, A., Biliaderis, C. G., & Izydorczyk, M. S. (2005). Water extractable (1 → 3, 1 → 4)-␤-d-glucans from barley and oats: An intervarietal study on their structural features and rheological behaviour. Journal of Cereal Science, 42(2), 213–224. Paradossi, G., Cavalieri, F., & Crescenzi, V. (1997). 1 H NMR relaxation study of a chitosan-cyclodextrin network. Carbohydrate Research, 300(1), 77–84. Pettolino, F. A., Walsh, C., Fincher, G. B., & Bacic, A. (2012). Determining the polysaccharide composition of plant cell walls. Nature Protocols, 7(9), 1590–1607.

Robinson, G., Ross-Murphy, S. B., & Morris, E. R. (1982). Viscosity-molecular weight relationships, intrinsic chain flexibility, and dynamic solution properties of guar galactomannan. Carbohydrate Research, 107(1), 17–32. Roubroeks, J. P., Andersson, R., Mastromauro, D. I., Christensen, B. E., & Åman, P. (2001). Molecular weight, structure and shape of oat (1 → 3), (1 → 4)-␤-d-glucan fractions obtained by enzymatic degradation with (1 → 4)-␤-d-glucan 4-glucanohydrolase from Trichoderma reesei. Carbohydrate Polymers, 46(3), 275–285. Roubroeks, J. P., Andersson, R., & Åman, P. (2000). Structural features of (1 → 3), (1 → 4)-␤-d-glucan and arabinoxylan fractions isolated from rye bran. Carbohydrate Polymers, 42(1), 3–11. Roubroeks, J. P., Mastromauro, D. I., Andersson, R., Christensen, B. E., & Åman, P. (2000). Molecular weight, structure, and shape of oat (1 → 3), (1 → 4)-␤-d-glucan fractions obtained by enzymatic degradation with lichenase. Biomacromolecules, 1(4), 584–591. Sweet, D. P., Shapiro, R. H., & Albersheim, P. (1975). Quantitative analysis by various g.l.c. response-factor theories for partially methylated and partially ethylated alditol acetates. Carbohydrate Research, 40(2), 217–225. Tosh, S. M. (2013). Review of human studies investigating the post-prandial blood-glucose lowering ability of oat and barley food products. European Journal of Clinical Nutrition, 67(4), 310–317. Tosh, S. M., Brummer, Y., Wood, P. J., Wang, Q., & Weisz, J. (2004). Evaluation of structure in the formation of gels by structurally diverse (1 → 3)(1 → 4)-␤-d-glucans from four cereal and one lichen species. Carbohydrate Polymers, 57(3), 249–259. Tosh, S. M., Wood, P. J., Wang, Q., & Weisz, J. (2004). Structural characteristics and rheological properties of partially hydrolyzed oat ␤-glucan: The effects of molecular weight and hydrolysis method. Carbohydrate Polymers, 55(4), 425–436. Vaikousi, H., Biliaderis, C. G., & Izydorczyk, M. S. (2004). Solution flow behavior and gelling properties of water-soluble barley (1 → 3, 1 → 4)-␤-glucans varying in molecular size. Journal of Cereal Science, 39(1), 119–137. Varum, K. M., Smidsrod, O., & Brant, D. A. (1992). Light scattering reveals micelle-like aggregation in the (1 → 3), (1 → 4)-␤-d-glucans from oat aleurone. Food Hydrocolloids, 5(6), 497–511. Virkki, L., Johansson, L., Ylinen, M., Maunu, S., & Ekholm, P. (2005). Structural characterization of water-insoluble nonstarchy polysaccharides of oats and barley. Carbohydrate Polymers, 59(3), 357–366. Whitehead, A., Beck, E. J., Tosh, S., & Wolever, T. M. (2014). Cholesterol-lowering effects of oat ␤-glucan: A meta-analysis of randomized controlled trials. The American Journal of Clinical Nutrition, 100(6), 1413–1421. Wood, P. J., Weisz, J., & Blackwell, B. A. (1994). Structural studies of (1 → 3), (1 → 4)-␤-d-glucans by 13 C-nuclear magnetic resonance spectroscopy and by rapid analysis of cellulose-like regions using high-performance anion-exchange chromatography of oligosaccharides released by lichenase. Cereal Chemistry, 71(3), 301–307. Wu, J., Zhang, Y., Wang, L., Xie, B. J., Wang, H. B., & Deng, S. P. (2006). Visualization of single and aggregated hulless oat (Avena nuda L.) (1 → 3), (1 → 4)-␤-d-glucan molecules by atomic force microscopy and confocal scanning laser microscopy. Journal of Agricultural and Food Chemistry, 54(3), 925–934. Wu, J., Li, L. L., Wu, X. Y., Dai, Q. L., Zhang, R., & Zhang, Y. (2016). Characterization of oat (Avena nuda L.) ␤-glucan cryogelation process by low-field NMR. Journal of Agricultural and Food Chemistry, 64(1), 310–319. Zhang, Q., Matsukawa, S., & Watanabe, T. (2004). Theoretical analysis of water 1 H T2 based on chemical exchange and polysaccharide mobility during gelation. Food Hydrocolloids, 18(3), 441–449.