Comparative Study of Small Linear and Branched Î±-Glucans Using

Biomacromolecules 2005, 6, 143-151

143

Comparative Study of Small Linear and Branched r-Glucans Using Size Exclusion Chromatography and Static and Dynamic Light Scattering# Mohammed Saddik Motawia,† Iben Damager,† Carl Erik Olsen,‡ Birger Lindberg Møller,† Søren Balling Engelsen,§ Steen Hansen,*,| Lars Holm Øgendal,| and Rogert Bauer| Plant Biochemistry Laboratory, Department of Plant Biology and Center for Molecular Plant Physiology (PlaCe), The Royal Veterinary and Agricultural University, DK-1871 Frederiksberg C, Denmark, Department of Natural Sciences, The Royal Veterinary and Agricultural University, DK-1871 Frederiksberg C, Denmark, Department of Food Science, The Royal Veterinary and Agricultural University, DK-1871 Frederiksberg C, Denmark, Department of Natural Sciences and Center for Molecular Plant Physiology (PlaCe), The Royal Veterinary and Agricultural University, DK-1871 Frederiksberg C, Denmark Received June 24, 2004; Revised Manuscript Received September 20, 2004

A series of synthesized small linear and branched R-glucans has been studied by dynamic light scattering and combined size exclusion chromatography, refractive index measurement and static light scattering. The R-glucan molecules studied were maltose, maltotriose, maltopentaose, maltohexaose, maltoheptaose, panose, 6′-R-maltosyl-maltotriose, methyl 6′-R-maltosyl-maltotrioside, 6′′-R-maltosyl-maltotetraose, 6′′′-R-maltotriosyl-maltohexaose, and 6,6′′′′-bis(R-maltosyl)-maltohexaose. The R-glucan oligosaccharides appeared to be very flexible molecules having a variety of conformations and self-associating into noncovalent dimers and trimers (referring to the single molecule). The size distributions were narrow (compared to pullulan) indicating that the R-glucan oligosaccharides are relatively compact molecules. The branched oligomers that include one or more flexible R-(1 f 6) linkages exhibit size distributions corresponding to more compact conformations than their linear counterparts. This observation may be explained by intermolecular interactions or water bridges facilitated by the additional flexibility of these molecules. For the branched maltohexaose, a significant noncovalent trimer formation was observed, whereas in all other cases, noncovalent dimers were formed. Model calculations suggest that both the linear and branched oligomers containing 5-10 R-glucose units exist predominantly in a partial or full single turn helix in agreement with the glycosidic linkage preferences derived for these molecules. Introduction Starch and starch derivatives are important food ingredients providing desired stability and mouth-feel and are the dominant energy source in human food. Consequently, it is relevant to study the microstructure and hydration of starch and its consequences for food functionality. Furthermore knowledge about starch would enable in vivo production of starches with desired functional properties. Starch is composed of two main types of high molecular weight macromolecules, amylose and amylopectin, both being homoglucans. However, the complexity, the stochastic nature and the tendency of these molecules to aggregate and form gels present great difficulties with regard to detailed structural characterization.1,2 As a model system for studying more # In commemoration of Professor Rogert Bauer, deceased 15th of April 2004. * To whom correspondence should be addressed. E-mail: [email protected]. † Department of Plant Biology and Center for Molecular Plant Physiology (PlaCe). ‡ Department of Natural Sciences and Center for Molecular Plant Physiology (PlaCe). § Department of Food Science. | Department of Natural Sciences.

simple interactions, we have chemically synthesized a range of linear and branched R-glucans and studied these by dynamic light scattering and combined size exclusion chromatography, refractive index measurement, and static light scattering. The results show clear differences in the nature of conformations between linear and branched maltooligosaccharides and in their tendency to form aggregates. Materials and Methods Materials. Maltose (Mal) (1), maltotriose (MalTri) (2), maltopentaose (MalPen) (3), maltohexaose (MalHex) (4), maltoheptaose (MalHep) (5), panose (6), D-glucose (12), and methyl R-D-glucopyranoside (13) were purchased from Sigma Chemical Co. Pullulan (p5) was purchased from Showa Denko K. K., Tokyo, Japan. The branched maltooligosaccharides 6′-R-maltosyl-maltotriose3 (Mal-MalTri) (7), methyl 6′-R-maltosyl-maltotrioside4 (Me-Mal-MalTri) (8), and 6′′′-R-maltotriosyl-maltohexaose5 (MalTri-MalHex) (10) were chemically synthesized as previously described.3-5 6,6′′′′-Bis(R-maltosyl)-maltohexaose (Mal-Mal-MalHex) (11) was chemically synthesized (unpublished procedure). 6′′-R-

10.1021/bm049634e CCC: $30.25 © 2005 American Chemical Society Published on Web 11/12/2004

144

Biomacromolecules, Vol. 6, No. 1, 2005

Maltosyl-maltotetraose (Mal-MalTet) (9) was also chemically synthesized (see the Supporting Information). Combined Size Exclusion Chromatography and Static Light Scattering. Samples (0.25 to 2.5 mg R-glucans in 50 µL of buffer prefiltered through an 0.22 µm syringe filter) were injected into a SUPERDEX PEPTIDE column (Amersham Bioscience, Hillerød, Denmark) with dimension 300 × 7.8 mm i.d. operated at a flow rate of 0.5 mL/min. The fact that the sample is passed through a column makes the sample preparation far less critical than it is when traditional batch mode light scattering measurements are done since the problem of high molecular weight impurities dominating the light scattering is eliminated as the column separates the small molecules from the impurities. The column was equilibrated in 10 mM NaH2PO4/Na2HPO4 (pH 6.7), 60 mM NaCl. Upon elution from the column, the sample was first passed through the light scattering instrument (Dawn F or Dawn EOS with a K2 cell, Wyatt Technology, Santa Barbara, CA), and then through the refractive index detector (RID10A, Shimadzu, Japan). All flow measurements were carried out at ambient temperature. The intensity of the scattered light is propotional to the product of the weight concentration and the molecular mass. The weight concentration of the eluting species was determined by the refractive index measurements (the signal is proportional to the weight concentration). The light scattering detectors were calibrated using bovine-R-lactalbumin (Sigma) assuming isotropic light scattering (for molecules with molar mass less than 105 g/mol, light scattering is isotropic). Subsequently, the weight average molar mass of oligomers of R-glucans was determined by the ratio of the area of the light scattering peak to the area of the refractive index peak. The average signal of four detectors was used including the 90° detector for mass determination. In a few cases, the molecular mass was determined at 10 times lower concentration. In all cases, this yielded the same molecular mass but with a larger uncertainty. Since there was no sure indication of concentration effects the molar masses found are based on just one concentration. The weight average molar mass of the pullulan sample (p5) was measured at 5700 g/mol. As the refractive index increment for pullulan is equal to that of maltose, 0.146 g/mL, this value was used for all samples. Dynamic Light Scattering. Sample Preparation. Samples were prepared by dissolving the appropriate amount of dry powder in buffer. Due to scarcity of material no filtering of the samples was attempted. Measurements. Several runs of 300 s were done at several concentrations in the approximate range 10-100 mg/mL, allowing extrapolation of individual analysis results to zero concentration. One typical example of an autocorrelation function (maltose, 50 mg/mL) is shown in Figure 1. Instrument. The instrument uses a green (λ ) 532 nm) 150 mW Nd:YAG laser (ADLAS 325) and measures at two different scattering angles (105° and 135°) and an ALV5000 correlator (ALV-Laser Vertriebsgesellscahft m.b.H., Langen, Hessen, Germany). For a detailed description of the instrument see Bauer et al. (1995).6 Data Analysis. The instrument measures the intensity autocorrelation function, g2. A monodisperse solution of

Motawia et al.

Figure 1. Autocorrelation function for maltose 50 mg/mL at scattering angle 105°. Channels 6-190 are shown.

particles gives rise to an exponentially decaying autocorrelation function with relaxation constant γ ) 2Dq2, where D is the diffusion constant and q is the length of the scattering vector. Likewise, if several species are present, the autocorrelation function becomes a squared sum of exponential decays. Therefore, the experimental autocorrelation functions g2(t) were analyzed using the multiexponential fitting procedure that is part of the autocorrelator software, using up to four exponential functions (allowing for up to four different species in the sample). The g2(t) was fitted with a model function g(t) of the form N

g(t) ) (

∑i Aie-γ t)2 + 1 i

(1)

where Ai are amplitude factors, γi ) Diq2 are the relaxation constants, and Di are the diffusion coefficients of the different species. The first five channels were omitted from the fit as they are heavily influenced by detector imperfections at low count rates. At a scattering angle of 105°, the term in the sum (eq 1) corresponding to a hydrodynamic radius of 0.5 nm will have decayed to 10% of its maximum value in channel 17 of the ALV5000 correlator. Although the radii sought for are in the range of 0.5-1 nm, approximately, they are determined by at least 12 channels. Furthermore, they turn out to be surprisingly well defined because their terms carry more than 50% of the total amplitude and because the other terms in the fit correspond to relaxation constants that are more than 100 times larger. We use the Stokes-Einstein relation for the diffusion coefficient of spherical particles and define the hydrodynamic radius Rh pertaining to any of the relaxation constants, γ, through the equation Rh )

kTq2 6πηγ

(2)

where k is the Boltzmann constant, T is the absolute temperature and η is the solvent viscosity. Model Calculations of Hydrodynamic Radii. Hydrodynamic radii were calculated for the purpose of comparing the experimental data to theoretical models. This was done in two different ways: (1) Simple models were constructed from touching spheres. In these calculations, the program HYDRO by Garcıá de la Torre et al.7 was used for

Small Linear and Branched R-Glucans

Biomacromolecules, Vol. 6, No. 1, 2005 145

Figure 2. Chemical structures of the linear and branched maltooligosaccharides used in the study.

calculation of the hydrodynamic radius. (2) The hydrodynamic radius was calculated directly from the atomic coordinates of the relevant structure using the program HYDROPRO, described in Garcıá de la Torre et al.8 In the latter, hydration was included using the atomic element radius (AER), which assigns spheres of variable radii to the atomic elements. Results The linear and branched malto-oligosaccharides 1-11 investigated are shown in Figure 2. Synthesis of the branched malto-oligosaccharides 1, 12, and methyl-R-D-glucopyranoside (14), were used as starting materials for the chemical synthesis of compounds 7-11. The general strategy for synthesis of compounds 7-9 and 11 is outlined in Chart 1. We have previously reported9 the synthesis of 7 using maltose and D-glucose as starting materials. In the described synthesis, maltose was chemically manipulated and transformed into a suitable glycosyl donor and a glycosyl acceptor and coupled together to produce a phenyl protected branched tetrasaccharide phenylthioglycoside derivative,3,9 which was then transformed into the glycosyl donor 20.3 Compounds 18-20 are the key intermediates for the synthesis of 7-9 and 11. We further described the conversion of 12 into the glycosyl acceptor 133 which was used for the synthesis of 7.3 Compound 14 was chemically modified and converted into the glycosyl acceptor 15 according to known procedures.10 Compounds 15 and 20 were used for the synthesis of 8.4 More recently, we have developed11 an efficient strategy for the conversion of 1 into the maltose-derived building blocks 16-19 that were used for the synthesis of

203 and 21 which were further used for the synthesis of 11 (unpublished procedure). The conversion of 21 into 9 is described in the Supporting Information. The chemical synthesis of 10 has been reported5 using a combination of glucose- and maltose-derived building blocks to obtain the maltotriose-derived building blocks 23-255,12 used for the synthesis of 105 as illustrated in Chart 2. Combined Static Light Scattering and Size Exclusion Chromatography. Each maltooligosaccharide was subjected to size exclusion chromatography. Aliquots (50 µL) containing 50 mg/mL maltooligosaccharide were applied to the column (except for 7 where 40 mg/mL was applied). For comparison, the peak position of the different maltooligosaccharide monomers was shifted in elution volume to the peak position of maltose and normalized to a peak intensity of 1 as shown in Figure 3. All linear maltooligosaccharides show roughly the same narrow distribution in size relative to maltotriose (Figure 3), except that for 1, 3, and 5, additional intensities are present to the right of the peak position indicating significant populations of conformations with smaller sizes than maltotriose. The relatively broad size distribution observed for p5 shows that the pullulan sample is a mixture of different molecular masses. This can be quantified by calculating the polydispersity index Mw/Mn, i.e., weight averaged over number averaged molecular mass. The result obtained from the measured chromatograms is in agreement with the manufacturers value of 1.07. The branched molecules 7-11 show an even more pronounced broadening in size distribution toward smaller sizes except for 6 and 8, which has almost the same narrow size distribution as 2. The branched pentamer (7), the nonamer (10), and the decamer (11) display increased populations of

146


Motawia et al.

Figure 3. Relative concentration of the R-glucans as a function of shifted elution volume (peak values for the concentration shifted to that of maltose). The peak values are all normalized to 1. Elution profile of maltotriose (2) is shown as reference (dotted lines). Chart 1. Illustration of the General Strategy to Chemically Synthesize the Branched Maltooligosaccharides 7-9 and 11

molecules with smaller sizes, but the branched hexamer (9) also exhibits conformations with larger sizes.

The formation of noncovalent oligomers of the different maltooligosaccharides was monitored by static light scatter-



Chart 2. Chemical Synthesis of Compound 10 from D-Glucose- and Maltose-Derived Building Blocks

Table 1. Theoretical as Well as Measured Values of Molar Weight and Hydrodynamic Radiusa

a

molecule

Mw (theory) (g/mol)

Mw (meas) (g/mol)

noncovalent oligomer (g/mol)

Rhb (nm)

Mal (1) MalTri (2) MalPen (3) MalHex (4) MalHep (5) Me-Mal-MalTri (8) Mal-MalTet (9) MalTri-MalHex (10) Mal-Mal-MalHex (11) Pullulan (P5)

360.32 504.44 828,73 990.87 1153.02 842.76 990.87 1477.30 1639.45 (5800)

356 ( 36 523 ( 41 872 ( 92 1040 ( 106 1208 ( 112 1180 ( 240 1024 ( 92 2030 ( 500 2250 ( 320 5700 ( 364

nd nd 1887 ( 137 2116 ( 121 2380 ( 191 nd 2778 ( 283 nd nd 43587 ( 3933

0.52 ( 0.03 0.62 ( 0.02 0.80 ( 0.02 0.81 ( 0.03 0.91 ( 0.02 nd nd nd nd 1.87 ( 0.01

The values for the hydrodynamic radii have been obtained by extrapolation to infinite dilution. b Averages of at least 7 measurements.

Figure 4. Relative concentration of the R-glucans. The vertical lines indicate the positions of the two maltohexaose noncovalent dimers.

ing combined with refractive index measurements after size exclusion chromatography (Figure 4 and Table 1). It is noted, that the words “dimer”, “trimer”, and “oligomer” in this text

are being used with reference to the single molecule and not the repeating unit. Compounds 3-5 form noncovalent dimers, whereas the branched maltohexaose (9) forms a noncovalent trimer (Table 1). For 1, 2, and the branched maltodecaose (11), peak positions in Figure 4 indicate contributions from noncovalent dimers. However, the different peak positions indicate that the noncovalent dimers exhibit a broad range of sizes. This is confirmed for the linear hexamer where two peaks (both representing dimers) are observed at -2 and -1.3 mL in shifted elution volume both having the same mass. This is illustrated in Figure 5, which shows that the relative light scattering intensity and refractive index signal intensity are not significantly different at the two peak positions. Also for pullulan a noncovalent oligomer is observed. However, in this case, it is a heptamer ((1 monomer unit, see Table 1). For the branched nonamer (10) and both of the branched pentamers (7) and (8) there is only little presence of noncovalent oligomers. The noncovalent oligomers constitute about 1% or less relative to the monomer content. The position and relative intensity of noncovalent oligomers is independent of the monomer concentration.

148


Figure 5. Light scattering signal and refractive index signal (proportional to concentration) for the linear hexamer (4) as a function of the shifted elution volume scaled such that the two signals are about equal at the two peak positions at -2 and -1.3 mL indicated by the two vertical lines.

Figure 6. Linear fit (shown as a solid line) to the hydrodynamic radius for all linear R-glucans studied as a function of elution volume. The position for p5 is also plotted.

Dynamic Light Scattering. Dynamic light scattering was used to determine the hydrodynamic radius for all of the linear maltooligosaccharides 1-5 as well as for p5 using the buffer, which was used for size exclusion chromatography. For the linear hexamer (5) and p5, the effect of concentration of the R-glucans is small despite changes in the macroscopic viscosity as a function of concentration. We therefore used linear extrapolation to infinite dilution in order to derive the values given in Table 1 for the hydrodynamic radii. In size exclusion chromatography, the position in elution volume, for a specific compound is determined by the size of the compound, which could be represented by its hydrodynamic radius. A linear fit to the R-glucans 1-5 is shown in Figure 6. The constant and the slope fitted are 3.8 ( 0.1 nm and -0.19 ( 0.01 nm/mL, respectively. This fit matches also the hydrodynamic radius for p5 indicating that a linear relation exists between elution volume and hydrodynamic radii. On the basis of this linear correlation, we estimated the hydrodynamic radii for those species where

Motawia et al.

Figure 7. Hydrodynamic radii (from Table 1 and derived from Figure 6) versus number of R-glucose units. Linear R-glucans (1-5 and p5) and branched glucans (7, 9-11) are marked by open circles and ×, respectively. The downward pointing triangles represent noncovalent oligomers formed of the linear R-glucans, and the upward pointing triangle represent noncovalent oligomers formed from branched R-glucans. The oligomers are positioned using the parameters from the line fit in Figure 6. Note that the linear hexamer has two different dimers (see Figures 4 and 5). All data shown, except that for p5, have been fitted with a power law shown as a solid line. The dotted line shows the result of modeling the R-glucan units as a linear structure.

Figure 8. Models for the linear R-glucans and corresponding experimental and calculated values for the hydrodynamic radius. The experimental values are from Table 2 and the theoretical values calculated by joining spheres either as a linear array or as a circular arc (see Materials and Methods) and scaled to the experimental value for maltose. First theoretical value is for the linear model and the second value for the circular model.

no dynamic light scattering data could be obtained, i.e., branched R-glucans and noncovalent oligomers. These are plotted in Figure 7 with a power law fit of the hydrodynamic radii versus numbers of R-glucose units Rh ) KnR

(3)

The value obtained for K was 0.39 ( 0.01 nm and for R ) 0.45 ( 0.02. This can be compared to R ) 0.5 for random coils and R ) 0.333 for compact objects. A linear and a circular model for the linear and branched R-glucans was applied in order to calculate the hydrodynamic radii (see Materials and Methods). The results are shown in Figures 8 and 9. From these figures, it follows that the circular models give good agreement with the experimental values. The linear models however do not fit the experimental values. The values derived for the hydrodynamic radius for the noncovalent oligomers follow the fitted curve in Figure 7 and do therefore, also in these cases, suggest relatively compact conformations. For the linear hexamer (4), two



Figure 12. Models for the dimer of the branched decamer and corresponding experimental and calculated values for the hydrodynamic radius. The experimental values are from Figure 7 and the theoretical values calculated as described in the Material and Method section. Figure 9. Models for the branched R-glucans and corresponding experimental and calculated values for the hydrodynamic radius. The experimental values are from Figure 7 and the theoretical values calculated as described in the Material and Method section.

Table 2. Calculations of Hydrodynamic Radii from Coordinates for Different Values of the Atomic Element Radius to Account for Hydration

Mal (1) MalTri (2) MalPen (3) MalHex (4) MalHep (5) Me-Mal-MalTri (8) Mal-MalTet (9) MalTri-MalHex (10) Mal-Mal-MalHex (11) Figure 10. Models for the dimers of the linear hexamer and corresponding experimental and calculated values for the hydrodynamic radius. The experimental values are from Figure 7 and the theoretical values calculated as described in the Materials and Methods section

0.25 (AER) (nm)

0.30 (AER) (nm)

0.35 (AER) (nm)

expt value (nm)

0.55 0.64 0.79 0.83 0.90 0.76 0.86 1.02 1.03

0.62 0.68 0.85 0.90 0.99 0.85 0.90 1.09 1.10

0.68 0.74 0.89 0.97 1.03 0.91 0.98 1.14 1.14

0.52 0.62 0.80 0.81 0.91 0.81 0.87 1.05 1.13

dimer of the branched decamer, the best agreement is obtained with the linear model. Conformational flexibility around the glycosidic linkage R-(1f4) is described by the two torsional angles: Φ ) O5C1-O1-C4′ and Ψ ) C1-O1-C4′-C5′ We did calculations based on the structure of the glucose unit in maltose choosing these angles such that the hexamer performs one helical turn in a left-handed helix (Φ ) 92° and Ψ ) -142°).13 The results are shown in Table 2. The values were calculated for three different values of the atomic element radius.8 Within the uncertainty, all experimental values fall within the range of theoretical values. Discussion

Figure 11. Models for the trimer of the branched hexamer and corresponding experimental and calculated values for the hydrodynamic radius. The experimental values are from Figure 7 and the theoretical values calculated as described in the Material and Method section.

noncovalent dimers are observed, having different values for the hydrodynamic radius. Two types of models for the association of the monomers into noncovalent dimers or trimers were made: a linear sidewise association and a circular stacking (see Figures 10-12). In case of the linear hexamer where two dimers are observed, the calculations of the hydrodynamic radii gave values slightly less than the experimental values. However, the differences between the two experimental values and the two theoretical values are similar. For the trimer from the branched hexamer and the

It is an intriguing question in starch research how the chain length distribution of R-glucans and the branching ratio affect disordered structure versus ordered helix structure formation of a given starch molecule and how this affects deposition and packing in a semicrystalline form in the starch grain. Both amylose with a low branching ratio and amylopectin with a high branching ratio show elements of disordered stochastic structure a well as ordered helix structure. Linear polysaccharides such as pullulans that are made up of maltotriose repeating units polymerized through R-(1 f 6) linkages exhibit a more random structure. Helical substructures are absent most likely because of the regular presence of flexible R-(1 f 6) linkages in the backbone.14 At the other extreme, small oligomers of R-(1 f 4)-glucans might initiate helix formation when the number of units is sufficient to

150


make a single turn of an amylosidic helix, that is n ) 6 or higher.15 Oligo- and polysaccharides do not usually form single crystals, for which reason atomic resolution structural information is unavailable. Accordingly, it is not known how branching affects the tendency of linear R-(1 f 4)-glucans to form helices. Fiber diffraction combined with molecular modeling may give detailed information about the ordered state of these molecules. However, this approach presents major difficulties due to the large conformational flexibility of the polysaccharides.16 Furthermore, due to the exorbitant size of amylose and amylopectin, ordered structures only occur in domains. Detailed theoretical calculations of R-glucan structure have been performed on small model compounds, e.g., the maltooligosaccharides maltose,17-19 isomaltose,20,21 methyl R-maltotrioside,22 and 6.23 With respect to the structure of R-glucans in aqueous solution, the hydration of maltose and isomaltose has been studied extensively with molecular dynamics simulations.24-27 This study was recently extended to larger amylosidic fragments28,29 in which the water structure around the maltohexaose and maltodecaose units of amylose was deduced from molecular dynamics computer simulations. NMR may provide a number of important structural and physical parameters on carbohydrates in aqueous solution including J-couplings, nOe’s, and diffusion coefficents30 and has been applied to small linear maltooligomers. Most recently, combined molecular modeling and NMR studies of a branched tetrasaccharide and a branched pentasaccharide have enabled detailed studies of the R-(1 f 6) branch point of amylopectin in solution.32 The gross conclusions reached from these studies are that the R-(1 f 4) linkage has essentially two sets of preferred dihedral angles (Φ; Ψ): an extended conformation and a conformation where the formation of a helix structure is favored. The structure appropriate for 6-fold helix formation is predicted from Ramachandran plots of the two major conformational flexibilities of the glycosodic linkage: Φ and Ψ.7 The angle Φ from one monomer to the next is 40° for the conformation which generates the left-handed helix. To test whether this theoretically predicted preference for helix formation actually occurs, we have measured the hydrodynamic radii for small linear and branched maltooligosaccharides in solution. A linear relation between hydrodynamic radius and elution volume was found for a number of linear R-glucans (Figure 6). This made it possible to assign hydrodynamic radii to all molecules including those where only the peak position in elution volume was available. A power law fit to the hydrodynamic radius versus number of R-glucose units showed a square root like behavior between radius and elution volume. When molecules containing two R-glucan chains linked by an R-(1 f 6) linkage were used, the measurements show that these branched maltooligosaccharides do not exist as linear extended molecules (dotted line in Figure 7) but rather as more compact molecules. Whether these represent helical like structures or not will now be discussed separately for the linear and branched R-glucans. Linear r-Glucans. In Figure 8, the measured hydrodynamic radii for the linear R-glucans are compared to two

Motawia et al.

different models for the molecules: a linear array or a circular structure. The circular structure fits the data well, whereas the linear array does not. As the circular structure resembles a first turn of a helix (except for the helical rise), this may indicate that the linear R-glucans initiates a helix conformation when dissolved in water. A narrow size distribution was observed for the linear R-(1 f 4)-glucans relative to the branched R-glucans (Figure 3). Most likely this is due to the restricted conformational flexibility of the R-(1 f 4) linkage compared to the R-(1 f 6) linkage. Compounds 1, 3, 4, and 5 exhibit additional conformational preferences toward lower sizes seen as a skewness in the peak shape to the right (Figure 3). For the linear R-glucans, two noncovalent dimers are present in the case of maltohexaose (Figures 4 and 7). To investigate whether the dimer with the smallest size could represent one turn in a double helix and the dimer with the largest size, a more extended association of two hexamers, we calculated the hydrodynamic radius for two hexamer circles stacked on top of each other and two hexamers associated sidewise. The lowest experimental value for the hydrodynamic radius is higher than the calculated value for the stacking of two rings (Figure 10). This is expected, as the turn in a double helix will be longer than the stacking of two rings by somewhat less than 0.1 nm. The larger dimer has a hydrodynamic radius close to the value calculated for the sidewise association. This may indicate that the linear hexamer can associate in both configurations. Branched r-Glucans. The branched malto-oligosaccharides have rather compact conformations, far from an extended linear conformation (Figure 7). Primarily this is a consequence of the presence of a branch point that enables only four units in a linear array for the branched hexamer instead of six in the case of the linear hexamer. For this reason, we calculated the hydrodynamic radius for a branched oligomer having the units after the R-(1 f 6) linkage stretched linearly and perpendicularly out from the branch point as well as values for models in which ring structures were included. Models and results are illustrated in Figure 9. For all branched oligosaccharides, the experimental data agreed well with the theoretical calculations based on compact circular models. This provides further evidence for initiation of helix formation for both linear and branched oligosaccharides (Figure 9). All branched malto-oligosaccharides have additional conformations compared to linear malto-oligosaccharide because of the presence of the R-(1 f 6)-linkage that enables additional conformational flexibility. When the configuration of the anomeric center is fixed by methyl-glycosidation, a molecule covering the same narrow distribution as the linear oligomers is obtained. This may indicate the involvement of the hydroxyl group at the reducing end in attaining the additional conformations with sizes smaller than the average. The methyl-glycosylated branched pentamer (8) elutes at a position close to that of the hexamer, which may be due to the extra length imposed by the addition of a methyl group that locks the glucose into a ring configuration. Only for the branched hexamer (8) and decamer (10) prominent noncovalent oligomers were ob-


served: a trimer and dimer for the hexamer and decamer, respectively. Both the noncovalent trimer and the dimer for the branched hexamer and the branched decamer were modeled as stacked arrays or circles, as was done with the dimers of the linear hexamer with the branched unit stretched 90° (Figures 11 and 12). In both cases, the linear model has the best fit indicating that branching prevents the stacking of helix like rings for the noncovalent branched oligomers. Conclusion In this work, we have demonstrated that isolated and welldefined linear or branched malto-oligosaccharides may serve as model compounds for predicting the conformation of amylose and amylopectin. The R-glucan oligosaccharides appeared to be very flexible molecules having a variety of conformations and the ability to self-associate into noncovalent dimers and trimers. All molecules studied had conformations far from being extended, linear and all R-glucans with more than four glucose units were able to adopt circular or helix-like structures. Therefore, the results obtained for these molecules may indicate the minimum degree of polymerization necessary for the formation of a helix structure. The branched oligomers that include one or more flexible R-(1 f 6) linkages generally exhibit a size distribution with more compact conformation than their linear counterparts. This may be caused by the additional flexibility of these molecules which allows for intramolecular interactions or water bridges30 that “lock” the molecule into a preferred conformation. Acknowledgment. This work was financially supported by the, Danish National Research Foundation, by the Danish Directorate for Development (Nonfood program), and by National Research the EU FAIR program contract CT950568. We are indebted to Marianne Lund Jensen for excellent help in sample preparations. Supporting Information Available. Synthesis of methyl 6′-R-maltosyl-maltotrioside (Me-Mal-MalTri) 7 and 6′′-Rmaltosyl-maltotetraose (Mal-MalTet) 9. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Roger, P.; Axelos, M. A. V.; Colonna, P. Macromolecules 2000, 33, 2446-2455. (2) Aberle, T.; Burchard, W.; Vorwerg, W.; Radosta, S. Starch/Sta¨rke 1994, 46, 329-335.

Biomacromolecules, Vol. 6, No. 1, 2005 151 (3) Motawia, M. S.; Olsen, C. E.; Enevoldsen, K.; Marcussen, J.; Møller, B. L. Carbohydr. Res. 1995, 277, 109-123. (4) Damager, I.; Motawia, M. S.; Olsen, C. E.; Blennow, A.; Denyer, K.; Møller, B. L. Carbohydr. Res. 2003, 338, 189-197. (5) Damager, I.; Olsen, C. E.; Møller, B. L.; Motawia, M. S. Carbohydr. Res. 1999, 320, 19-30. (6) Bauer, R.; Hansen, M.; Hansen, S.; Øgendal, L.; Lomholt, S.; Qvist, K.; Harne, D. J. Chem. Phys. 1995, 7, 2827-2737. (7) Garcıá de la Torre, J.; Navarro, S.; Lo´pez Martinez, M. C.; Dıáz, F. G.; Lo´pez Cascales, J. J. Biophys. J. 1994, 67, 530-531. (8) Garcıá de la Torre, J.; Huertas, M. L.; Carrasco, B. Biophys. J. 2000, 78, 719-730. (9) Motawia, M. S.; Olsen, C. E.; Møller, B. L.; Marcussen, J. Carbohydr. Res. 1994, 252, 69-84. (10) (a) Bell, D. J.; Lorber, J. J. Chem. Soc. 1940, 453-455. (b) Petit, J. M.; Jacquinet, J. C.; Sinay¨ , P. Carbohydr. Res. 1980, 82, 130-134. (c) Garegg, P. J.; Hultberg, H. Carbohydr. Res. 1981, 93, C10C11. (d) Deninno, M. P.; Etienne, J. B.; Duplantier, K. C. Tetrahedron Lett. 1995, 36, 669-672. (e) Debenham, S. D.; Toone, E. J. Tetrahedron-Asymmetry 2000, 11, 385-387. (f) Elhalabi, J.; Rice, K. G. Carbohydr. Res. 2001, 335, 159-165. (11) Motawia, M. S.; Larsen, K.; Olsen, C. E.; Møller, B. L. Synthesis 2000, 1547-1556. (12) Damager, I.; Olsen, C. E.; Møller, B. L.; Motawia, M. S. Synthesis 2002, 418-426. (13) O’Sullivan, A. C.; Pe´rez, S. Biopolymers 1999, 50, 381-390. (14) Liu, J. H. Y.; Brameld, K. A.; Brant, D. A.; Goddard, W. A. Polymer 2002, 43, 509-516. (15) Imberty, A.; Buleón, A.; Tran, V.; Pe´rez, S. Starch/Sta¨rke 1991, 43, 375-384. (16) Pe´rez, S.; Kouwijzer, M. L. C. E.; Mazeau, K.; Engelsen, S. B. J. Mol. Graph. 1996, 14, 307-321. (17) Takusagawa, F.; Jacobson, R. A. Acta Crystallogr. 1978, B34, 213218. (18) Ha, S. N.; Madsen, L. J.; Brady, J. W. Biopolymers 1988, 27, 19271952. (19) Tran, V.; Buleón, A.; Imberty, A.; Pe´rez, S., Biopolymers 1989, 28, 679-690. (20) Imberty, A.; Pe´rez, S. Int. J. Biol. Macromol. 1989, 11, 177-185. (21) Dowd, M. K.; Reilly, P. J.; French, A. D. Biopolymers 1994, 34, 625-638. (22) Pangborn, W.; Langs, D.; Pe´rez, S. Int. J. Biol. Macromol. 1985, 7, 363-369. (23) Imberty, A.; Pe´rez, S. Carbohydrate Res. 1988, 181, 41-55. (24) Brady, J. W.; Schmidt, R. K. J. Phys. Chem. 1993, 97, 958-966. (25) Ott, K.-H.; Meyer, B. Carbohydrate Res. 1996, 281, 11-34. (26) Best, R. B.; Jackson, G. E.; Naidoo, K. J. J. Phys. Chem. B 2001, 105, 4742-4751. (27) Corzana, F.; Motawia, M. S.; Herve´ du Penhoat, C.; Pe´rez, S.; Tschampel, S. M.; Woods, R. J.; Engelsen, S. B. J. Comput. Chem. 2004, 25, 573-586. (28) Momany, F. A.; Willett, J. L. Biopolymers 2002, 63, 99-110. (29) Naidoo, K. J.; Kuttel, M J. Comput. Chem. 2001, 22, 445-456. (30) Engelsen, S. B.; Monteiro, C.; Herve´ du Penhoat, C.; Pe´rez, S. Biophys. Chem. 2001, 93, 103-127. (31) Best, R. B.; Jackson, G. E.; Naidoo, K. J. J. Phys. Chem. B 2002, 106, 5091-5098. (32) Corzana, F.; Motawia, M. S.; Herve´ du Penhoat, C.; van den Berg, F.; Blennow, A., Perez, S.; Engelsen, S. B. J. Am. Chem. Soc. 2004, 126, 13144-13155.

BM049634E

Comparative Study of Small Linear and Branched Î±-Glucans Using

Recommend Documents