Biomacromolecules 2011, 12, 409–418
409
Structural Analysis of Enzyme-Resistant Isomaltooligosaccharides Reveals the Elongation of r-(1f3)-Linked Branches in Weissella confusa Dextran Ndegwa Henry Maina,*,† Liisa Virkki,† Henna Pynno¨nen,† Hannu Maaheimo,‡ and Maija Tenkanen† Department of Food and Environmental Sciences, P.O. Box 27, FIN-00014 University of Helsinki, Finland, and VTT Technical Research Centre of Finland, P.O. Box 1000, FI-02044 VTT, Finland Received September 25, 2010; Revised Manuscript Received December 7, 2010
Weissella confusa VTT E-90392 is an efficient producer of a dextran that is mainly composed of R-(1f6)-linked D-glucosyl units and very few R-(1f3) branch linkages. A mixture of the Chaetomium erraticum endodextranase and the Aspergillus niger R-glucosidase was used to hydrolyze W. confusa dextran to glucose and a set of enzymeresistant isomaltooligosaccharides. Two of the oligosaccharides (tetra- and hexasaccharide) were isolated in pure form and their structures elucidated. The tetrasaccharide had a nonreducing end terminal R-(1f3)-linked glucosyl unit (R-D-Glcp-(1f3)-R-D-Glcp-(1f6)-R-D-Glcp-(1f6)-R-D-Glc), whereas the hexasaccharide had an R-(1f3)linked isomaltosyl side group (R-D-Glcp-(1f6)[R-D-Glcp-(1f6)-R-D-Glcp-(1f3)]-R-D-Glcp-(1f6)-R-D-Glcp(1f6)-R-D-Glc). A mixture of two isomeric oligosaccharides was also obtained in the pentasaccharide fraction, which were identified as (R-D-Glcp-(1f6)-R-D-Glcp-(1f3)-R-D-Glcp-(1f6)-R-D-Glcp-(1f6)-R-D-Glc) and (RD-Glcp-(1f6)[R-D-Glcp-(1f3)]-R-D-Glcp-(1f6)-R-D-Glcp-(1f6)-R-D-Glc). The structures of the oligosaccharides indicated that W. confusa dextran contains both terminal and elongated R-(1f3)-branches. This is the first report evidencing the presence of elongated branches in W. confusa dextran. The 1H and 13C NMR spectroscopic data on the enzyme-resistant isomaltooligosaccharides with R-(1f3)-linked glucosyl and isomaltosyl groups are published here for the first time.
Introduction Dextrans are R-glucan exopolysaccharides (EPS) produced by lactic acid bacteria (LAB) that belong to Leuconostoc, Lactobacillus, Streptococcus, and Weissella genera.1,2 Dextrans are synthesized extracellularly from sucrose by dextransucrases (EC 2.4.1.5). LAB produce a variety of dextrans that are heterogeneous in size and structure. In some cases, a strain can synthesize two or more dextrans, which differ in the type and portion of glycosidic linkages.3 A survey of dextrans from 96 strains (primarily Leuconostoc mesenteroides) demonstrated that the amount of R-(1f6) linkages in a specific dextran can vary from 50 to 97% of the total glycosidic linkages.4 Based on such studies, dextrans are divided into three classes. Class 1 dextrans have contiguous R-(1f6)-linked D-glucopyranosyl units in the main chain and R-(1f2), R-(1f3), or R-(1f4) branch linkages. Class 2 dextrans (alternans) contain alternating R-(1f3) and R-(1f6) linkages with R-(1f3) branch linkages, while class 3 dextrans (mutans) have consecutive R-(1f3) linkages and R-(1f6) branch linkages.5 LAB have a Generally Recognized As Safe (GRAS) status and are, therefore, safely used in various applications. Due to the heterogeneity of dextrans produced by LAB, their commercial utilization may depend on well-defined chemical and physicochemical properties.3 For instance, in clinical applications, where dextran is used as a blood plasma substitute, dextran produced by L. mesenteroides B 512F (5% R-(1f3) branch linkages) is preferred, due to its low antigenicity, high water * To whom correspondence should be addressed. Tel.: +358-9-19158403. Fax: +358-9-19158475. E-mail:
[email protected]. † University of Helsinki. ‡ VTT Technical Research Centre of Finland.
solubility, and high biological stability in the human bloodstream.1 In food applications, the preferred dextran is not clearly defined. In confectioneries, dextrans are used for viscosity, moisture retention, to inhibit sugar crystallization, and as gelling agents in gum and jelly candies.6 Dextran is also used as a cryoprotectant in ice cream.7 A current consumer trend toward healthy and additive free food has made dextrans an attractive solution for the replacement of commercial hydrocolloids used in food applications, such as bread production. The novelty in this case, is the possibility to introduce the hydrocolloid (dextran) by in situ biosynthesis during fermentation, rather than as a purified additive.8–10 Recent studies have proven that dextran produced in situ during sourdough fermentation improves moisture retention, dough rheology, bread volume, and shelf life, and retards staling.10,11 In situ production is only feasible if the potential strain produces sufficient amounts of dextran in the cereal matrix within the limited sourdough fermentation time. Additionally, the dextran should have a high molecular weight and few branch linkages.12 Weissella species have lately been identified as common members of the sourdough microflora,13,14 soya,15 and kimchi.16 In a recent study, we showed that Weisella confusa E392, isolated from carrot mash, is a highly prospective strain for mild acidification and in situ production of dextran during sourdough fermentation.10 The strain is not only a more efficient dextran producer than the conventional L. mesenteriodes B512F, but also produces a dextran with very few branch linkages (2.7% R-(1f3) branch linkages).2 Bouniax et al.13 also found that dextrans from various Weissella strains have few branch linkages, indicating that this may be a common feature of Weissella species. However, studies on the exact nature of these branch linkages are lacking.
10.1021/bm1011536 2011 American Chemical Society Published on Web 01/05/2011
410
Biomacromolecules, Vol. 12, No. 2, 2011
Generally, physicochemical studies tend to indicate that some of the branch linkages in dextrans are elongated, which results in a compact ramified structure rather than a comblike structure. Even when these extended branches are few, they substantially influence the physicochemical properties of dextrans.17–19 It is proposed that these extended branches explain the Newtonian behavior of dextrans, even at concentrations of 30% wt, for dextrans with molecular weights (MW) as high as 2 × 106 Daltons. These branches may also explain why the dependence of intrinsic viscosity [η] on the molecular weight of dextrans decreases with increasing molecular weight.19 To better understand the functional properties of dextran from W. confusa E392, a more detailed analysis of its R-(1f3)-branches, particularly elongation, was undertaken in this study. In our earlier study, several presumably branched dextran fragments remained after extensive enzyme hydrolysis.10 The shortest enzyme-resistant isomaltooligosaccharides (IMO) were now isolated and their structure determined. Accordingly, the nature of branch linkages in W. confusa E392 dextran is proposed.
Experimental Section Materials. The dextran produced by Weissella confusa VTT E-90392 () DSM 20194, NCDO 1975) was obtained as described previously.2 Monosaccharide composition analysis indicated that the dextran extract was 83.6% glucose.2 Commercial dextran produced by Leuconostoc mesenteroides NRRL B512F (GE Healthcare, Uppsala, Sweden) was studied as a comparison. The enzymes used included endodextranase (EC 3.2.1.11) from Chaetomium erraticum (D0443, Sigma-Aldrich, Steinheim, Germany) and R-glucosidase (EC 2.4.1.24) from Aspergillus niger (E-TRNGL, Megazyme, Wicklow, Ireland). The dextranase randomly hydrolyses R-(1f6)-D-glucoside linkages in dextrans, while the R-glucosidase is reported by the manufacturer to have activity toward different glucosidic disaccharides: maltose (R-(1f4)-D-glucoside linkage), isomaltose (R(1f6)-D-glucoside linkage), nigerose (R-(1f3)-D-glucoside linkage), and kojibiose (R-(1f2)-D-glucoside linkage). The enzymatic activity of the dextranase (195 808 nkat/mL) was determined as previously described.10 The activity of the R-glucosidase (16670 nkat/mL) was provided by the manufacture. Reagents used included: HPLC grade dimethylsulfoxide from LabScan Analytical Sciences (Dublin, Ireland), sodium hydroxide powder, methyliodide, myo-inositol, and maltose from Sigma-Aldrich Chemie (Steinheim, Germany), sodium sulfate, trifluoroacetic acid (TFA), ammonia 25%, glacial acetic acid, 1-methyl-imidazole, sulphuric acid (H2SO4), acetic acid anhydride, and deuterium oxide (D2O, 99.8%) from Merck (Darmstadt, Germany), orcinol from Merck, (Schuchardt, Germany), ethanol from Altia Oy (Rajama¨ki, Finland), dichloromethane from Rathburn Chemicals Ltd. (Walkerburn, Scotland), sodium borohydride from Fluka Chemie, (Buchs, Germany), and isomaltose (IM2) and isomaltotriose (IM3) from TCI Europe (Zwijndrecht, Belgium). Enzymatic Hydrolysis and Fractionation of Oligosaccharides. To study the action of the enzymes, 5 mg/mL of dextran in 0.05 mM sodium citrate buffer pH 5.5 was prepared and hydrolyzed with 10000 nkat/g dextranase and 1000 nkat/g R-glucosidase for 48 h at 30 °C. A parallel sample was also prepared in a similar way, but hydrolyzed with dextranase only. During the hydrolysis, 1 mL samples were collected at various time points (10 min, 0.5, 1, 2, 4, 6, 8, 24 h) for oligosaccharide analysis. To stop the reactions, the enzymes were denatured by keeping the samples in a boiling water bath for 10 min. The total glucose released was quantified using the D-glucose assay kit (K-GLUC, Megazyme, Wicklow, Ireland). For structural analysis of the enzyme-resistant IMO, 0.5 g of W. confusa dextran and 1 g commercial dextran were dissolved in 15 mL buffer and hydrolyzed with both enzymes for 48 h. The enzyme-resistant IMO were fractionated by gel permeation chromatography (GPC) with a Biogel P2 column (5 × 95 cm; Bio-Rad, Hercules, CA, U.S.A.),
Maina et al. using water as the eluent. The hydrolysates were filtered with 0.45 µm membrane filters (Acrodisc 13, Pall Corporation, Ann Arbor, Michican, U.S.A.) and injected into the column. The flow rate was kept at 0.5 mL/min overnight to elute the void volume (600 mL), followed by the collection of 5 mL fractions the following day at a flow rate of 2 mL/ min. The presence of oligosaccharides in the fractions was first detected by spotting a small sample on silica gel 60 plates (Merck). The oligosaccharides were made visible by spraying the plates with 2% (w/v) orcinol in ethanol/H2SO4/H2O (80:10:10), followed by heating at 100 °C for 10 min. The oligosaccharide positive fractions were further analyzed by high performance anion exchange chromatography with pulse amperometric detection (HPAEC-PAD) to access their composition. Similar fractions were pooled and freeze-dried for further analysis. HPAEC-PAD Analysis. The HPAEC-PAD equipment consisted of a Waters 717 autosampler, two Waters 515 HPLC pumps, an analytical CarboPac PA-100 column (250 × 4 mm, i.d, Dionex, Sunnyvale, CA, U.S.A.), and a Decade detector with a gold electrode (Antec Leyden, Zoeterwoude, The Netherlands). The analyses were carried out using the oligosaccharide method with a 1 M NaOAc gradient in 100 mM NaOH, at a flow rate of 1 mL/min, as described by Rantanen et al.20 Glucose, IM2, and IM3 were used as external qualitative standards. The samples were filtered through 0.45 µm membranes (Acrodisc 13, Pall Corporation, Ann Arbor, Michican, U.S.A.), and the injection volume was 10 µL. Methylation Analysis. Methylation analysis was conducted according to the method of Ciucanu and Kerek,21 with some modifications. Briefly, after permethylation, the samples were extracted with dichloromethane. The dichloromethane phase was evaporated and the samples dried under vacuum before hydrolysis with 2 M TFA for 2 h at 120 °C. The partially methylated glucosides were reduced and acetylated as described by Virkki et al.22 The methylation products were identified by gas chromatography with mass spectrometry detection (GC-MS). The GC-MS used was Hewlett-Packed 6890 with a RTX column (60 m × 0.32 mm × 0.10 µm) Agilent Technologies (Foster City, CA, U.S.A.). The oven temperature program was 170 °C (10 min), gradient at 4 °C/min to 200 °C (10 min). Mass Spectrometry. The molar masses of the isolated oligosaccharides were determined by mass spectrometry with electron spray ionization (ESI-MS) in positive mode. The MS equipment was the Agilent 1100 Series LC/MSD with an Ion Trap XCT Plus and Electrospray Ionizator (Agilent Technology). The oligosaccharide fractions (10 µL) were mixed with 400 µL of MeOH/water/formic acid solution (50:49:1). The samples were directly introduced into the electrospray ionizator at a flow rate of 10 µL/min. The nebulizer gas (N2) pressure was 15.0 psi, the drying gas flow rate was 4 L/min, and the temperature was 325 °C. The capillary voltage and the capillary offset value were 3164 and -500 V, respectively. A skimmer potential of 40.0 V and a trap drive value of 67.9 were used. Spectra were recorded using a scan range from 100-2000 m/z. NMR Spectroscopy Analysis. The fine structure of the oligosaccharides was determined by one-dimensional (1D) and two-dimensional (2D) NMR spectroscopy. NMR spectroscopy was carried out with a Varian Unity Inova spectrometer (Varian NMR Systems, Palo Alto, CA, U.S.A.) operating at 500 or 800 MHz for 1H and using 5 mm triple-resonance pulsed-field gradient (PFG) probes. NMR samples (10-15 mg/mL) were exchanged three times with D2O and then placed in NMR tubes (Sigma-Aldrich Chemie). The measurements at 500 MHz were performed at 23 °C (oligosaccharides) and 50 °C (native dextran), while those at 800 MHz were performed at 70 °C. The chemical shifts were referenced to acetone (δH ) 2.225 ppm and δ C ) 31.55 ppm). Acquisition and handling of the data was carried out as described before.2
Results Enzymatic Hydrolysis. W. confusa dextran was hydrolyzed with dextranase only and a mixture of dextranase and R-glucosidase. The action of dextranase on dextran was first followed
Enzyme-Resistant Isomaltooligosaccharides
Biomacromolecules, Vol. 12, No. 2, 2011
411
Figure 1. HPAEC-PAD chromatograms showing the profile of enzyme-resistant IMO remaining after hydrolysis of dextran from W. confusa E392 with dextranase (10000 nkat/g) only, for 10 min, 2, 8, 24, and 48 h (A-E), and hydrolysis with both dextranase and R-glucosidase (1000 nkat/g) for 48 h (F): Glc ) glucose, IM2 ) isomaltose, IM3 ) isomaltotriose, and IM4 ) isomaltotetraose.
by taking samples at various time points during the hydrolysis (Figure 1). In the early stage of the hydrolysis, the random endoaction of dextranase resulted in formation of IM2, IM3, and several IMO eluting between 25 and 35 min (Figure 1 A). As the dextranase hydrolysis proceeded further, longer IMO were transformed to glucose, IM2, and IM3, and dextranaseresistant IMO started to build-up (Figure 1B-E). It should be noted, that some susceptible IMO have almost the same retention time as the dextranase-resistant IMO. The peak eluting at 25 min decreases as the hydrolysis proceeds from 10 min to 2 h then increases henceforth indicating that it is a mixture of the first dextranase-resistant IMO and a susceptible IMO. The slow increase of first peak eluting between 27 and 28 min indicated that it is mainly formed during prolonged hydrolysis. When R-glucosidase was present (Figure 1F), it catalyzed hydrolysis of IM2 and IM3 to glucose and slightly changed the profile of the dextranase-resistant IMO. As shown in Figure 1F, the ratio of the peaks eluting between 27 and 28 min changed, while peaks eluting after 29 min slightly decreased. Therefore, most of the enzyme-resistant IMO after hydrolysis with both enzymes were already end products resulting from the prolonged action of dextranase and represented the structurally complex, presumably R-(1f3)-branched sections in the dextran. The yield of glucose after hydrolysis with both enzymes was 70% of the freeze-dried material, which represented 84% of the glucose analyzed by methanolysis.2 As a comparison, commercial dextran produced by L. mesenteroides B512F was also hydrolyzed with both enzymes. HPAEC-PAD analysis indicated that the profile of its enzyme-resistant IMO (Figure 1S, Supporting Information) was similar to that of W. confusa dextran. Fractionation, MS, and Methylation Analysis of EnzymeResistant IMO. For structure analysis, W. confusa dextran and the commercial dextran were hydrolyzed with dextranase and R-glucosidase, and the resulting enzyme-resistant IMO fractionated by gel permeation chromatography. The R-glucosidase was included to degrade linear IMO, which simplified purification of the oligosaccharides. The fractionation resulted in three major
pools, presumably branched IMO (BIM4, BIM5, and BIM6; Figure 2). BIM4 and BIM6 had one major peak in the HPAECPAD analysis, while BIM5 contained two peaks (Figure 2). ESIMS analysis revealed that BIM4 was a tetrasaccharide (pseudomolecular ion [M + Na]+ at 689 m/z), BIM5 was composed of pentasaccharides (851 m/z), and BIM6 was a hexasaccharide (1013 m/z). As the BIM5 pool contained two peaks in the HPAEC-PAD chromatogram, the MS results indicated that they were two isomeric pentasaccharides. The BIM5 pool was analyzed as a mixture without further purification. The fractions eluting prior BIM6 (not shown) were mixtures of various oligosaccharides (1013, 1175, and 1337 m/z), indicating the presence of hexa-, hepta-, and octasaccharides. These fractions were not analyzed further. Considering that the NMR data on enzyme-resistant IMO from W. confusa and commercial dextran were similar (Figure 3 and Figure 2S, Supporting Information), methylation analysis was performed on oligosaccharides isolated from commercial dextran. All material obtained from W. confusa dextran was use in NMR spectroscopy analysis. A summary of the main glycosidic linkages in BIM4, BIM5, and BIM6 is illustrated in Table 1. BIM4-6 contained 2,3,4,6-tetra-O-methyl- and 2,3,4tri-O-methyl-glucosides originating from the nonreducing terminal, and 6-O-monosubstituted glucosyl residues, respectively. In addition, BIM4 and BIM5 contained 2,4,6 tri-O-methylglucoside, indicating the presence of 3-O-monosubstituted glucosyl residues. BIM5 and BIM6 also contained 2,4-di-Omethyl-glucoside, indicating the presence of 3,6-O-disubsituted glucosyl residues. NMR Spectroscopy Analysis. Tetrasaccharide Pool (BIM4). The 1H NMR spectrum of BIM4 (Figure 3) had four anomeric signals at δH 5.35 (d, J1,2 3.6 Hz), 5.24 (d, J1,2 3.6 Hz), 4.97 and 4.68 (d, J1,2 8.2 Hz) ppm, with an integral ratio of 1:0.4:2:0.6, respectively. Two of the anomeric signals were assigned to the R (δH 5.24 ppm) and β (δH 4.68 ppm) anomers of the reducing end (RR and Rβ, respectively). Thus, the integral ratio further verified that BIM4 was a tetrasaccharide. 2D NMR
412
Biomacromolecules, Vol. 12, No. 2, 2011
Maina et al.
Figure 2. HPAEC-PAD chromatograms of top fractions containing the enzyme-resistant IMO (BIM4-6) isolated after 48 h hydrolysis of W. confusa E392 dextran with dextranase and R-glucosidase. Mass spectra ([M + Na]+) of each fraction are shown as an inset. Note, the retention times of the peaks are different from those in Figure 1 due to analysis at different times.
Figure 3. The 1D 1H spectra of the native dextran and enzyme-resistant IMO (BIM4, BIM5, and BIM6) obtained after hydrolysis of W. confusa E392 dextran with dextranase and R-glucosidase. Glucosyl residues are numbered A, B, and C (main chain) and A1 and B1 (residues in the branches) from the nonreducing end, with the reducing end residue referred to as residue RR/Rβ (see Figure 5). The spectra were measured at 500 MHz in D2O, at 23 (oligosaccharides) and 50 °C (native dextran). The chemical shifts are referenced to internal acetone (δH ) 2.225 ppm).
experiments were used to assign the chemical shift of each spin system. These included double quantum filtered correlation spectroscopy (DQF-COSY), total correlation spectroscopy (TOCSY), heteronuclear single quantum coherence (HSQC), HSQC-TOCSY, and heteronuclear multibond correlation (HMBC)
experiments. From the 2D TOCSY experiments, subspectra (obtained by taking 1D traces starting from each anomeric signal or other well-resolved peaks) were used for assignment. The multiplicity of the signals23 was helpful in the analysis of TOCSY subspectra (Figure 4I-III). For clarity, the residues are
Enzyme-Resistant Isomaltooligosaccharides
Biomacromolecules, Vol. 12, No. 2, 2011
413
Table 1. Methylation Analysis and MS Data of the Enzyme-Resistant IMO Isolated after Hydrolysis of Commercial Dextran with dextranase and R-Glucosidase linkages obtained by methylation analysis oligosaccharide
Glc-(1f
f6)Glc-(1f
f3)Glc-(1f
f3,6)Glc-(1f
ESI-MS [M + Na]+ (m/z)
BIM4 BIM5 BIM6
+ + +
+ + +
+ + -
+ +
689 851 1013
numbered from the nonreducing end (A-C) with the reducing end residue referred to as residue R in all cases (Figure 5). In accordance with the structural-reporter groups developed by van Leeuwen et al.,24 the chemical shifts of the reducing end (R and β anomeric signals, δH 5.24 and 4.68 ppm, respectively) were typical for an R-(1f6)-linked glucosyl unit. This was confirmed by a correlation peak between the H4 of
the reducing end residue (δH 3.51 and 3.52 ppm, R and β forms, respectively) and its own bound C6 (δC 67.0 and 67.1 ppm, respectively) in the HMBC spectrum (data not shown). The chemical shifts (H2 to H5) for the glucosyl residue with an anomeric chemical shift at δH 5.35 ppm were assigned with the TOCSY subspectra (Figure 4II). This residue had an H4 chemical shift at δH 3.45 ppm, which is typical for a nonreducing
Figure 4. TOCSY subspectra starting from the anomeric signals of (I) residue Rβ 4.68 ppm, (II) residue A 5.35 ppm from the 2D TOCSY spectra of the BIM4, and (III) residue B1, 5.32 ppm from the 2D TOCSY spectra of the BIM6. The spectra were measured at 500 MHz in D2O, at 23 °C. The chemical shifts are referenced to internal acetone (δH 2.225 ppm). Glucosyl residues are numbered A (main chain) and B1 (residue in the branch chain) from the nonreducing end with the reducing end residue referred to as residue RR/Rβ (see Figure 5).
Figure 5. Structures of the enzyme-resistant IMO (BIM4, BIM5-1, BIM5-2, and BIM6) obtained from W. confusa E392 dextran. The glucosyl residues are labeled A-D (main chain) and A1 and B1 (residues in the branch chain) from the nonreducing end with the reducing end residue referred to as residue R in all cases.
414
Biomacromolecules, Vol. 12, No. 2, 2011
terminal residue (A).24 The next two glucose residues from the reducing end (B and C) had overlapping anomeric signals at δH 4.97 ppm. H2 chemical shifts for the two residues (δH 3.66 and 3.58 ppm) were easily assigned with the COSY spectrum (not shown). To assign H3 to H6a and b chemical shifts, various 2D correlation peaks were considered. From the HSQC-TOCSY spectrum (not shown), it was clearly observed that one of the two residues had a bound C6 at δC 66.5 ppm, while the other had an unbound C6 at δC 61.5 ppm. Additionally, one of the residues had a carbon chemical shift at δC 81.1 ppm, typical for a bound C3. According to the HMBC spectrum (not shown), the bound C3 was correlated to the anomeric signal of residue A (δH 5.35 ppm) constituting the following building block: R-DGlcp-(1f3)-R-D-Glcp- (residue A-(1f3)-B). From the HSQC spectrum (not shown), H3 of residue B was assigned with the H3fC3 correlation peak at δH 3.85 ppmfδC 81.1 ppm. An HSQC-TOCSY spectrum acquired at 800 MHz, 70 °C (not shown), for better 1H signal dispersion, showed that the residue (B) with a bound C3 had an unbound C6, while residue C had a bound C6. Consequently, correlation peaks between H4 and C6 in the HMBC spectrum could be used to ambiguously assign H4 chemical shifts for residues B and C (δH 3.65, δC 61.5 ppm and δH 3.54, δC 66.5 ppm, respectively). With these C6 chemical shifts, H6a and b were assigned from the HSQC spectrum. Residue B had H6 chemical shifts between δH 3.75 and 3.85 ppm. The actual values could not be assigned because this residue shared the same C6 chemical shift with residue A. H6 chemical shifts for residue C were δH 3.75 and 3.98 ppm for H6a and H6b, respectively. C5 chemical shifts for residue B and C were obtained from H1fC5 cross peaks in the HMBC spectrum (δC 72.9 and 71.4 ppm), which were then used to determine H5 chemical shifts from the HSQC spectrum (δH 3.74 and 3.93 ppm, respectively). With this data, assignment of the H5 (δH 3.93 ppm) and C5 (δC 71.4 ppm) chemical shifts to residue C was confirmed by a COSY cross peak at δH 3.54f3.93 ppm (H4fH5). The data thus confirmed that the structure of BIM4 is R-D-Glcp-(1f3)-R-D-Glcp-(1f6)R-D-Glcp-(1f6)-R-D-Glc (33-R-D-glucosylisomaltotriose, Figure 5 BIM4). A summary of the chemical shifts for residues A-C and R are shown in Table 1S, Supporting Information. Hexasaccharide Pool (BIM6). The 1H NMR spectrum of BIM6 (Figure 3) had six anomeric signals at δH 5.34 (d, J1,2 3.7 Hz), 5.32 (d, J1,2 3.7 Hz), 5.24 (d, J1,2 3.6 Hz), 4.99, 4.96, and 4.68 (d, J1,2 8.1 Hz) ppm, with an integral ratio of 0.2:0.8: 0.4:2.0:1.8:0.7. Similar to BIM4, the signals at δH 5.24 and 4.68 ppm were the R and β anomers of the (1f6)-linked reducing end glucosyl residue (R). From the integral ratios and considering that BIM6 is a hexasaccharide, the signal at 5.34 was considered an impurity. The assigned chemical shifts for all spin systems in the hexasaccharide are depicted in Table 2S (Supporting Information). Based on COSY (not shown) and 2D TOCSY (Figure 6I) assignments, the anomeric protons at δH 4.96 ppm had an H4 at δH 3.42 ppm, which is typical of a nonreducing end unit.24 The integral of this anomeric signal was two, signifying that this was a branched oligosaccharide with two nonreducing terminals (residues A and A1, where A1 represents the terminal residue of the branch chain). On the HSQC-TOCSY spectrum (not shown) A and A1 were the only residues with an unbound C6 δC 61.6 ppm, making it easy to confirm their H6a and b chemical shifts from the HSQC spectrum (not shown, δH 3.77 and 3.85 ppm). According to the HMBC spectrum (Figure 6II), the two anomeric protons at δH 4.96 ppm (A and A1) had interresidue correlation peaks at δC 66.7 ppm (A(H1)fB(C6) and
Maina et al.
A1(H1)fB1(C6), indicating that they were R-(1f6)-linked glucosyl residues. Additionally, H1fC3 and H1fC5 intraresidue correlation peaks in the HMBC spectrum provided the C3 and C5 chemical shifts at δC 74.4 and 73.2 ppm, respectively. These values were used to assign the proton chemical shifts from the HSQC spectrum (δH 3.72 and 3.73 ppm, respectively). The signal at δH 4.99 ppm also represented two anomeric signals (B and C). However, these two residues were not equivalent. This was clearly observed in the COSY spectrum (H2 at δH 3.59 and 3.68 ppm), and in the HMBC (Figure 6 II) intraresidue correlation peaks (H1fC3 at δH 4.99fδC 82.3 and δH 4.99f δC 74.8 ppm). The HSQC-TOCSY further revealed that the two residues had a bound C6 at δC 66.7 ppm. This data was consistent with an internal 6-O-monosubstituted glucosyl residue (H2 at δH 3.59, C3 at δC 74.8 ppm and a bound C6 at δC 66.7 ppm) and a 3,6-O-disubstituted branch point (H2 at δH 3.68, bound C3 and C6 at δC 82.3 and 66.7 ppm, respectively). H4 signals for B and C were assigned using intraresidue H4fC6 HMBC correlation peaks at δH 3.77fδC 66.7 and δH 3.54fδC 66.7 ppm. Using the COSY spectrum, the H4 at δH 3.77 ppm was assigned to the 3,6-O-disubstituted residue using the H3fH4 correlation peak (δH 3.82f3.77 ppm). Furthermore, the H5 of this residue was assigned through a COSY correlation peak (δH 3.77f3.93 ppm, H4fH5). Assignment of the remaining chemical shifts for these residues was hindered by signal overlap. The signal at δH 5.32 ppm was for an elongated R-(1f3) branch point. This was substantiated by a correlation peak from the anomeric signal at δH 5.32 ppm to a bound C3 (δC 82.3 ppm) on the HMBC spectrum and the fact that it had a bound C6 at δC 66.5 ppm. Furthermore, the H5 chemical shift for this residue at δH 4.20 ppm was also unique and has been established to be a structural reporter for a -R-(1f6)-R-D-Glcp-(1f3)unit.25 As shown in Figure 3, a similar signal is also present in the 1D 1H spectrum of the native dextran from W. confusa E392. Complete assignment of the chemical shifts of this residue was easily determined from multiplicity of the signals in the TOCSY subspectra (Figure 4III). Based on this data, the hexasaccharide had an R-(1f3) linked isomaltosyl branch unit (R-(1f6)-RD-Glcp-(1f3)-, A1fB1-). The main chain contained R-(1f6) linkages with one of the residues being 3,6-O-disubstituted (AfBfCfR). Whether the disubstituted residue was in position B or C could not be verified with the current data. However, based on previous reports,26 residue B is most likely disubstituted forming the following structure R-D-Glcp-(1f6)[RD-Glcp-(1f6)-R-D-Glcp-(1f3)]-R-D-Glcp-(1f6)-R-D-Glcp(1f6)-R-D-Glc (33- isomaltosylisomaltotetraose, Figure 5). The anomeric signal at δH 5.34 ppm was considered part of another oligosaccharide. On the TOCSY spectrum (Figure 6I, F1 dimension), the residue clearly had a signal at δH 4.20 ppm, which is typical for a H5 in a -R-(1f6)-R-D-Glcp-(1f3)glucosyl unit. It is possibly an internal residue between two 6-monosubsituted glucose residues. Similar anomeric signals at δH 5.34 ppm have previously been observed.25 Pentasaccharide Pool (BIM5). Based on the assignments of BIM4 and BIM6, it was possible to make conclusions regarding the structure of the isomeric mixture in the pentasaccharide pool. The 1H NMR spectrum of the mixture (Figure 3) showed six anomeric signals at δH 5.33 (d, J1,2 3.7 Hz), 5.32 (d, J1,2 3.7 Hz), 5.24 (d, J1,2 3.8 Hz), 4.98, 4.96 (d, J1,2 3.1 Hz), and 4.68 (d, J1,2 8.1 Hz) ppm. The signals at δH 5.24 and 4.68 ppm were assigned to the reducing end residues RR and Rβ, respectively. As explained for BIM6, the signal at δH 4.96 ppm was assigned to the nonreducing end R-(1f6)-R-D-Glcp terminal residue.
Enzyme-Resistant Isomaltooligosaccharides
Biomacromolecules, Vol. 12, No. 2, 2011
415
Figure 6. 2D TOCSY (I) and 2D HMBC (II) spectra of BIM6 obtained by enzymatic hydrolysis of commercial dextran. The spectra were measured at 500 MHz in D2O at 23 °C. The chemical shifts are referenced to internal acetone (δH 2.225 ppm, δC 31.55 ppm). Glucosyl residues are numbered A-C (main chain) and A1 and B1 (residues in the branch chain) from the nonreducing end, with the reducing end residue referred to as residue RR/Rβ (see Figure 5).
Starting from the H1fH5 correlation peak (δH 5.33f4.20 ppm) in the 2D TOCSY spectrum (Figure 3S, Supporting Information), the chemical shifts for this residue were assigned (δH 5.33 (H1), 3.57 (H2), 3.73 (H3), 3.56 (H4), 4.20 (H5), 3.68 (H6a), and 4.02 (H6b) ppm). The H5 chemical shift of this residue (δH 4.20 ppm) signified that it was an -R-(1f6)-R-D-Glcp(1f3)- unit. In the HSQC-TOCSY spectrum (Figure 4S, Supporting Information), it is clear that the signal at δH 5.32 ppm has an unbound C6 at δC 61.6 ppm, while the signal at δH 5.33 ppm has a bound C6 at δC 66.4 ppm. Along the TOCSY trace starting from the anomeric signals at δH 5.33 and 5.32 ppm, a triplet signal at δH 3.45 ppm was observed, which can be assigned to
H4 of an R-(1f3)-linked terminal residue.24 This signal was assigned to the anomeric signal at δH 5.32 ppm, as the H4 of the signal at δH 5.33 ppm was already assigned to δH 3.56 ppm. In the HMBC spectrum (not shown), both of these anomeric signals had a correlation peak to a bound C3 at 81.7 ppm. It was therefore possible to conclude that the anomeric signal at δH 5.32 ppm is for a R-(1f3)-R-D-Glcp terminal residue, while the signal at δH 5.33 ppm was for an internal R-(1f3)-linked glucosyl residue. The residues with anomeric signals centered at δH 4.98 ppm could not be assigned due to signal overlap. However, on the HSQC-TOCSY, it was clear that one of the signals had a bound C6 and the other had an unbound C6 (δc 66.6 and 61.6 ppm, respectively). On the HMBC spectrum, this
416
Biomacromolecules, Vol. 12, No. 2, 2011
anomeric signal was correlated to two bound C6 (δc 67.0 and 66.6 ppm) that were assigned to one of the anomeric signals at δH 4.98 ppm and to the Rβ anomeric signal at δH 4.68 ppm, indicating a -R-D-Glcp-(1f6)-R-D-Glcp-(1f6)-D-Glcp-(1f6)D-Glc building block. According to this data, the pentasaccharides differed in the position of their R-(1f3)-linked glucosyl residue, one being a terminal branch unit, while the other was an internal residue. According to methylation analysis data, the pentasaccharide mixture had a 3,6-O-disubstituted residue. This was verified by a H4fC6 cross peak (δH 3.77f δC 66.7 ppm) in the HMBC spectrum (not shown). As in BIM6, the H4 at δH 3.77 ppm is typical for a 3,6-O-disubstituted residue. Most likely, the R-(1f3)-linked terminal glucosyl residue is associated with the 3,6-O-disubstituted residue, which explains the similarity of its anomeric signal (δH 5.32 ppm) to that of residue B1 in BIM6. Furthermore, it is different from the anomeric signal (δH 5.35 ppm) of an R-(1f3)-linked glucosyl residue associated with a 3-O-monosubstituted residue in BIM4. We can therefore conclude that the 3-O-monosubstituted residue observed in methylation analysis is associated with the internal R-(1f3)linked glucosyl residue (δH 5.33 ppm). The presence of a -RD-Glcp-(1f6)-R-D-Glcp-(1f6)-D-Glcp-(1f6)-D-Glc building block suggested that the structures of the pentasaccharides are most likely R-D-Glcp-(1f6)[R-D-Glcp-(1f3)]-R-D-Glcp-(1f6)R-D-Glcp-(1f6)-R-D-Glc (33-R-D-glucosylisomaltotetraose), and R-D-Glcp-(1f6)-R-D-Glcp-(1f3)-R-D-Glcp-(1f6)-R-D-Glcp(1f6)-R-D-Glc (33-isomaltosylisomaltotriose), henceforth referred to as BIM5-1 and BIM5-2, respectively (Figure 5).
Discussion Recently, we have reported that dextran from W. confusa E392 has few R-(1f3) branch linkages (2.7%).2 In this study, the nature of the branches was evaluated by structural analysis of the four shortest enzyme-resistant IMO obtained after hydrolysis of the native dextran with C. erraticum dextranase and A. niger R-glucosidase. The dextranase hydrolyzed internal R-(1f6) linkages in dextran, producing glucose, IM2, IM3, and a set of dextranase-resistant IMO. Though the R-glucosidase has activity toward glucosidic disaccharides with different R-linkages, it mainly acted on IM2 and IM3, leaving most of the dextranase-resistant IMO intact (Figure 1E,F). Enzyme hydrolysis was preferred because it is reproducible and it has previously been shown that secondary glycoside linkages in dextrans are inherently more acid labile than R-(1f6)-linkages,27 which results in the loss of the key structural elements. The enzyme-resistant IMO were fractionated and analyzed by mass spectrometry, methylation analysis, and NMR spectroscopy. Structural analysis of the enzyme-resistant IMO revealed that they included a tetrasaccharide (BIM4) with an R-(1f3)-linked glucosyl unit at the nonreducing end (33-R-D-glucosylisomaltotriose), two isomeric pentasaccharides, one with a terminal R-(1f3)-linked branch unit and the other containing an internal R-(1f3)-linked glucosyl residue (33-R-D-glucosylisomaltotetraose, BIM5-1 and 33-isomaltosylisomaltotriose, BIM5-2), and a hexasaccharide (BIM6) containing an R-(1f3)-linked isomaltosyl branch unit (33-isomaltosylisomaltotetraose). The location of the R-(1f3) linkage at the terminal nonreducing end of BIM4 implies that the productive binding to the C. erraticum dextranase requires two 6-O-monosubstituted glucosyl residues on the left side of the hydrolyzed linkage. This mode of action is comparable to that of P. funiculosum dextranase.26 Depending on the initial binding site of the dextranase, the terminal
Maina et al.
R-(1f3)-linked glucosyl units in the BIM4 and BIM5-1 can originate from both single glucosyl residue and elongated side groups in the native W. confusa dextran. However, the R-(1f3)linked isomaltosyl units in BIM5-2 and BIM6 confirm the presence of elongated branches in the native dextran. Taylor et al.26 have previously analyzed the oligosaccharides obtained after enzymatic hydrolysis of different dextrans with P. funiculosum dextranase. According to their investigation, enzymatic hydrolysis of a synthetic dextran V39 and Streptococcus Viridans NRRL B-1351 dextran, which only contain terminal R-(1f3)-linked glucosyl branches, resulted in a homologous series of branched IMO. The smallest member of this series was analogous to the BIM4 (33-R-D-glucosylisomaltotriose) obtained in this study. The other members (e.g., DP5, DP6) consisted of glucosylisomaltooligosaccharides, the terminal glucosyl branch unit always R-(1f3)-linked to the third glucosyl residue from the reducing end of the R-(1f6)-linked main chain. In the same study, hydrolysis of L. mesenteroides B512F dextran that contains both single unit and elongated side chains resulted in two isomeric homologous series of branched IMO. One of the series was similar to the one obtained from V39 and Streptococcus Viridans NRRL B-1351 dextrans, whereas the second one contained isomaltosyl branch units. The smallest member of the latter series was analogous to BIM5-2, with an internal R-(1f3)-linked glucosyl unit (33-isomaltosylisomaltotriose), while the rest of the members were longer 33isomaltosylisomaltooligosaccharides carrying the R-(1f3)linked isomaltosyl side group at the third glucosyl residue from the reducing end of the main chain.26 Thus, the structures of the shortest enzyme-resistant IMO (BIM4, BIM5-1, BIM5-2, BIM6) obtained from W. confusa E392 dextran in this study were similar to those obtained by Taylor et al.26 from L. mesenteroides B512F dextran using P. funiculosum dextranase. Furthermore, the occurrence of an isomeric pentasaccharide mixture (BIM5-1 and BIM5-2) in our analysis is consistent with the previous results.26 As a comparison, we performed the same enzymatic hydrolysis on the commercial L. mesenteroides B512F dextran, which resulted in a similar oligosaccharide profile in the HPAEC-PAD analysis and exactly the same enzyme-resistant IMO (BIM4-6, 1D NMR spectra are shown in Figure 2S, Supporting Information) obtained from the W. confusa E392 dextran. This was expected, as the strains produce structurally similar dextrans that differ only in the content of their R-(1f3)-linkage.2 Taylor et al.26 showed that 33-isomaltosylisomaltotriose (BIM5-2) gradually increased and dominated with prolonged hydrolysis, while 33-glucosylisomaltotetraose (BIM5-1) remains constant. A similar phenomenon occurs with C. erraticum dextranase, as illustrated in Figure 1A-E, suggesting that the first, progressively increasing pentasaccharide peak is BIM5-2, and the second one is BIM5-1. The impurity in the BIM6 pool was not an isomeric hexasaccharide with a single unit glucosyl branch residue (33 glucosylisomatopentose), implying that with extensive hydrolysis, longer 33 glucosylisomaltooligosaccharides are susceptible to C. erraticum dextranase hydrolysis, as previously reported for P. funiculosum endodextranase.26 We can therefore speculate that the longer enzyme-resistant IMO eluting before the BIM6 pool during fractionation (hepta- and octasaccharides) contain either isomaltosyl or longer branches. The detailed NMR spectroscopy data on the dextranaseresistant IMO (BIM4-6) is reported for the first time. As shown in the results, analysis of TOCSY subspectra based on the multiplicity of the proton signals, and a combination
Enzyme-Resistant Isomaltooligosaccharides
of data from various 2D experiments gave the assignment of the spin systems. For example, the H4fC6, H1fC5, and H1fC3 correlation peaks in the HMBC spectrum combined with HSQC data was helpful in assigning the chemical shifts of each glucosyl residue. Unlike 2,6-O-disubstituted residues that have a unique anomeric signal,2 the anomeric signal of 3,6-O-disubstituted and 6-O-monosubstituted residues resonate in a narrow region (δH 4.99-4.97 ppm). The presence of 3,6-O-disubstituted residues can be confirmed by the H4fC6 correlation peaks in the HMBC as it has a unique H4 at δH 3.77 ppm (23 °C 500 MHz). Similarly, the anomeric chemical shifts for R-(1f3)-linkages in different environments cluster in a narrow chemical shift range (δH 5.32-5.35 ppm). Our results indicate that the signal at δH 5.32 ppm (23 °C 500 MHz) is characteristic for an R-(1f3)-linked glucosyl residue attached to a 3,6-O-disubstituted branch point, whether it contains a single glucosyl unit or an elongated side chain. This anomeric signal can be used to substantiate the presence of elongated branches in native dextrans, if its H5 resonates at δH 4.20 ppm. This H5 signal is unique for an R-(1f6)-R-D-Glcp-(1f3)-structural unit.25 As shown in Figure 3, a signal at δH 4.20 ppm is present, in the 1D spectrum of the native W. confusa dextran. Even when it is minor, this signal should not be disregarded in structural analysis of dextrans. The results obtained here are not sufficient to determine whether the branches are longer than the detected disaccharide isomaltosyl unit, forming a compact ramified structure analogous to that of amylopectin. This is, nonetheless, highly possible, as dextranase most likely hydrolyzes long R-(1f6)-linked branches, leaving at most two glucosyl residues to the R-(1f3)-linked branch point. The presence of long branches in dextrans has previously been substantiated. Using sequential chemical removal of terminal D-glucosyl groups, Larm et al.28 concluded that 40% of the side chains in dextran produced by L. mesenteroides B-512F contained one glucosyl residue, 45% were two glucosyl residues long, and the rest (15%) were longer than two residues. Based on physicochemical data, Ioan et al.29 estimated that elongated branches in dextran from L. mesenteroides B-512F can be 179 units long. Additionally, the mechanism for synthesis of branch points in dextrans proposed by Robyt et al.30 suggests that the formation of branches with varying lengths is possible. According to the mechanism, in addition to elongation of dextranyl chains with R-(1f6)linkages, the dextransucrases can participate in secondary transglycosylation reactions involving acceptors such as monosaccharides, oligosaccharides, or dextranyl chains. Secondary transglycosylation is responsible for transfer of glucosyl residues to a dextranyl chain to form single unit R-(1f3)-branch points or the transfer of dextranyl chains of varying lengths to produce dextranyl branched dextrans.30 As we have previously proposed,10 the HPAEC-PAD profile of the dextranase-resistant IMO can be a versatile tool for screening structural variations in dextrans produced by different lactic acid bacteria in varying conditions and environments. However, a feasible method to determine the length and portion of the elongated branch linkages in dextrans is still lacking. Such details would be highly useful for understanding and comparison of the physiochemical properties of different dextrans. To our knowledge, there are no reports on an R-glucanase having specific activity toward terminal or elongated R-(1f3)-branch linkages in class 1 dextrans. The latter type of activity would enable the determination of the length of branches after the enzyme treatment.
Biomacromolecules, Vol. 12, No. 2, 2011
417
Conclusion In this study, C. erraticum dextranase and A. niger R-glucosidase were used to produce structural segments of the native W. confusa dextran. As a comparison, oligosaccharides were also obtained from commercial dextran produced by L. mesenteroides B512F with the same enzymes. Structural evaluation of the enzyme-resistant IMO confirmed the presence of elongated branches in both dextrans. The results also indicated that the action of the C. erraticum dextranase on dextrans is closely related to that of P. funiculosum endodextranase reported previously. The structures of the enzyme-resistant IMO, as well as knowledge of their behavior in the HPAEC-PAD analysis, will facilitate chromatogram interpretation and quantitative assessment of elongated and single unit branches in dextrans. Acknowledgment. The Academy of Finland (project 128898), Glycoscience Graduate School, Finnish Cultural Foundation, and Raisio Research Foundation are gratefully acknowledged for their financial support. Jouni Jokela is thanked for assistance with mass spectrometry analysis. Note Added after ASAP Publication. This article posted ASAP on January 5, 2011. Figure 3 has been revised. The correct version posted on January 12, 2011. Supporting Information Available. Summary tables with H and 13C NMR chemical shifts of the glucosyl residues in BIM4 and BIM6 obtained by enzymatic hydrolysis of dextran produced by W. confusa E392. HPAEC-PAD profile of enzyme resistant IMO from commercial dextran, 1H spectra of BIM4-6 obtained by enzymatic hydrolysis of commercial dextran, and TOCSY and HSQC-TOCSY spectra of the pentasaccharide mixture. This material is available free of charge via the Internet at http://pubs.acs.org.
1
References and Notes (1) Naessens, M.; Cerdobbel, A.; Soetaert, W.; Vandamme, E. J. J. Chem. Technol. Biotechnol. 2005, 80, 845–860. (2) Maina, N. H.; Tenkanen, M.; Maaheimo, H.; Juvonen, R.; Virkki, L. Carbohydr. Res. 2008, 343, 1446–1455. (3) Wilham, C. A.; Alexander, B. H.; Jeanes, A. Arch. Biochem. Biophys. 1955, 59, 61–75. (4) Jeanes, A.; Haynes, W. C.; Wilham, C. A.; Rankin, J. C.; Melvin, E. H.; Austin, M.; Cluskey, J. E.; Fisher, B. E.; Tsuchiya, H. M.; Rist, C. E. J. Am. Chem. Soc. 1954, 76, 5041–5052. (5) Robyt, J. F. Dextran. In Encyclopedia of Polymer Science and Engineering; Kroschwitz, J. I., Ed.; Wiley-VCH: New York, 1986; Vol. 4, pp 752-767. (6) Khan, T.; Park, J.; Kwon, J. Korean J. Chem. Eng. 2007, 24, 816– 826. (7) Adapa, S.; Schmidt, K. A.; Jeon, I. J.; Herald, T. J.; Flores, R. A. Food ReV. Int. 2000, 16, 259. (8) Tieking, M.; Korakli, M.; Ehrmann, M. A.; Ga¨nzle, M. G.; Vogel, R. F. Appl. EnViron. Microbiol. 2003, 69, 945–952. (9) Korakli, M.; Rossmann, A.; Ganzle, M. G.; Vogel, R. F. J. Agric. Food Chem. 2001, 49, 5194–5200. (10) Katina, K.; Maina, N. H.; Juvonen, R.; Flander, L.; Johansson, L.; Virkki, L.; Tenkanen, M.; Laitila, A. Food Microbiol. 2009, 26, 734– 743. (11) Tieking, M.; Ganzle, M. G. Trends Food Sci. Technol. 2005, 16, 79– 84. (12) Lacaze, G.; Wick, M.; Cappelle, S. Food Microbiol. 2007, 24, 155– 160. (13) Bounaix, M.; Gabriel, V.; Morel, S.; Robert, H.; Rabier, P.; RemaudSime´on, M.; Gabriel, B.; Fontagne´-Faucher, C. J. Agric. Food Chem. 2009, 57, 10889–10897. (14) DiCagno, R.; De Angelis, M.; Limitone, A.; Minervini, F.; Carnevali, P.; Corsetti, A.; Gaenzle, M.; Ciati, R.; Gobbetti, M. J. Agric. Food Chem. 2006, 54, 9873–9881.
418
Biomacromolecules, Vol. 12, No. 2, 2011
(15) Malik, A.; Radji, M.; Kralj, S.; Dijkhuizen, L. FEMS Microbiol. Lett. 2009, 300, 131–138. (16) Kim, M.; Chun, J. Int. J. Food Microbiol. 2005, 103, 91–96. (17) Bovey, F. A. J. Polym. Sci. 1959, 35, 167–182. (18) De Belder, A. N. Dextran. In Industrial Gums: Polysaccharides and Their DeriVatiVes; Whistler, R. L., BeMiller, J. N., Eds.; Academic Press: San Diego, 1993; pp399-425. (19) Tirtaatmadja, V.; Dunstan, D. E.; Boger, D. V. J. Non-Newtonian Fluid Mech. 2001, 97, 295–301. (20) Rantanen, H.; Virkki, L.; Tuomainen, P.; Kabel, M.; Schols, H.; Tenkanen, M. Carbohydr. Polym. 2007, 68, 350–359. (21) Ciucanu, I.; Kerek, F. Carbohydr. Res. 1984, 131, 209–217. (22) Virkki, L.; Maina, H. N.; Johansson, L.; Tenkanen, M. Carbohydr. Res. 2008, 343, 521–529. (23) Roslund, M. U.; Ta¨htinen, P.; Niemitz, M.; Sjo¨holm, R. Carbohydr. Res. 2008, 1, 101–112.
Maina et al. (24) van Leeuwen, S. S.; Leeflang, B. R.; Gerwig, G. J.; Kamerling, J. P. Carbohydr. Res. 2008, 343, 1114–1119. (25) van Leeuwen, S. S.; Kralj, S.; van Geel-Schutten, I. H.; Gerwig, G. J.; Dijkhuizen, L.; Kamerling, J. P. Carbohydr. Res. 2008, 343, 1237– 1250. (26) Taylor, C.; Cheetham, N. W. H.; Walker, G. J. Carbohydr. Res. 1985, 137, 1–12. (27) Sidebotham, R. L. AdV. Carbohydr. Chem. Biochem. 1974, 30, 371– 444. (28) Larm, O.; Lindberg, B.; Svensson, S. Carbohydr. Res. 1971, 20, 39– 48. (29) Ioan, C. E.; Aberle, T.; Burchard, W. Macromolecules 2001, 34, 3765– 3771. (30) Robyt, J. F.; Yoon, S. H.; Mukerjea, R. Carbohydr. Res. 2008, 343, 3039–3048.
BM1011536
