Short-Distance Intermolecular Correlations of Mono- and Disaccharides in Condensed Solutions: Bulky Character of Trehalose.

Organisms with tolerance to extreme environmental conditions (cryptobiosis) such as desiccation and freezing are known to accumulate stress proteins and/or sugars. Trehalose, a disaccharide, has received considerable attention in the context of cryptobiosis. It has already been shown to have the highest glass-transition temperature and different hydration properties from other mono- and disaccharides. In spite of the importance of understanding cryptobiosis by experimentally clarifying sugar-sugar interactions such as the clustering in concentrated sugar solutions, there is little direct experimental evidence of sugar solution structures formed by intermolecular interactions and/or correlation. Using a wide-angle X-ray scattering method with the real-space resolution from ∼3 to 120 Å, we clarified the characteristics of the structures of sugar solutions (glucose, fructose, mannose, sucrose, and trehalose), over a wide concentration range of 0.05-0.65 g/mL. At low concentrations, the second virial coefficients obtained indicated the repulsive intermolecular interactions for all sugars and also the differences among them depending on the type of sugar. In spite of the presence of such repulsive force, a short-range intermolecular correlation was found to appear at high concentrations for every sugar. The concentration dependence of the observed scattering data and p(r) functions clearly showed that trehalose prefers a more disordered arrangement in solution compared to other sugars, that is, bulky arrangement. The present findings will afford a new insight into the molecular mechanism of the protective functions of the sugars relevant to cryptobiosis, particularly that of trehalose.

Introduction

Sugars and polyols are known as bioprotectants that prevent protein denaturation and enzyme deactivation. They are widely used as nontoxic additives, such as preservatives, humectants, and thickening stabilizers in industrial products, and as cryoprotectants for storing enzymatic reagents, bacteria, nematodes, mammalian embryos, and other biological samples. The protective mechanism of osmolytes such as sugars and polyols is based on their chemical thermodynamics and has been attributed to the specific bindings between the biological components and additives, changes in solvent viscosity/surface tension, and free energy changes upon transfer into additive solutions.[1−6] It is well known that some organisms that show considerable tolerances against extreme environmental conditions produce stress proteins and/or accumulate sugars in their cells. This phenomenon is called cryptobiosis, that is, the ability of an organism to tolerate environmental changes without having to actively adapt to them. In particular, trehalose has attracted attention in the context of cryptobiosis under external stress such as desiccation and freezing. Trehalose is found in animals, plants, and microorganisms.[7−10] It is a natural α-linked non-reducing disaccharide formed by an α,α-1,1-glucoside bond between two α-glucose units. The cryptobiotic activity of trehalose has been explained in terms of the restriction of the intra-and/or-inter-molecular movement by vitrification or the replacement of water molecules by trehalose.[9] Previous results suggest that an understanding of the structure of the sugars and the interactions between sugar and water molecules is a key prerequisite for understanding cryptobiosis.

Early studies using techniques such as nuclear magnetic resonance (NMR)[11] and molecular dynamics (MD) simulation[12,13] suggested that sugars are surrounded by hydration layers in solution. Therefore, sugars have the ability to form intramolecular hydrogen bonds and intermolecular hydrogen bonds with water.[14,15] This property is closely related to the high glass-transition temperature (Tg) of the sugar. In particular, calorimetric measurements and thermodynamics considerations have shown that trehalose has a remarkably high Tg; usually, the addition of small amounts of water does not depress its Tg, as is observed for other sugars.[16−18] This was explained by the fact that much of the absorbed water formed trehalose dihydrate.[19] Recent neutron diffraction studies combined with MD have directly shown the localization of the water molecules surrounding the sugars and the positions of the hydrogen bonds.[20,21] These studies demonstrated that identical chemical groups (hydroxyl group (−OH) and hydroxymethyl group (−CH2OH)) on the sugar molecules can have radically different hydration patterns, depending on their location in a given molecule.[21] These high-spatial-resolution results have afforded relevant insights into the biological effects of sugars, which depend on their molecular structures. Thus, to understand cryptobiosis, it is also important to experimentally clarify sugar–sugar interactions such as the formation of clusters in concentrated sugar solutions.

Recently, by the complementary use of X-ray and neutron scattering techniques, we have provided direct evidence that the protein hydration (solvation) and structural stability against chemical and thermal denaturation significantly depend considerably on the sugar species and glycerol.[22−24] The sugar and glycerol molecules tend to be preferentially or weakly excluded from the protein surface, which preserves the native protein hydration shell; however, the preferential exclusion (preferential hydration) shifts gradually toward the non-preferential solvation (replacement of the hydrated water by sugar or glycerol) as the concentrations of these molecules increase. Owing to the protective action of these molecules on the protein hydration shell, the protein structure is stabilized against chemical (guanidinium chloride) and thermal denaturation. The protective action depends on the sugar species.[24] To understand the above trends and the differences among the sugars in detail, it is important to clarify the characteristics of the solutions containing these additives.

Herein, using the wide-angle X-ray scattering (WAXS) technique over a wide spatial region (∼3–120 Å), we have clarified the structures of the sugar solutions with concentrations ranging from 5 to 52.5% w/w (from 0.05 to 0.65 g/mL) concentrations. The sugars measured in this work were monosaccharides (glucose, fructose, and mannose) and disaccharides (sucrose and trehalose). Based on the resulting wide-spatial-resolution scattering data, we obtained information about the internal structures of the individual sugar molecules and the correlations between them. The average intermolecular distances gradually shortened for all sugar species as their concentrations were increased. The intermolecular interactions between the sugar molecules were essentially exclusive, as evidenced by the presence of a repulsive correlation hole. This trend was clearly weaker for trehalose as compared to the other sugars. The present results show that intermolecular interactions between the sugar molecules differ depending on the type of sugars.

Results and Discussion

Monomeric Structural Properties of the Hydrated Sugar Molecules in Solutions As Observed by WAXS and Comparison with Theoretical Sugar Crystal Structures

Figure depicts the concentration dependence of the WAXS curves of the sugar solutions in 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) at pH 7.0 and at 25 °C. Figure A–E corresponds to the monosaccharides (glucose, fructose, and mannose) and disaccharides (sucrose and trehalose), respectively. The solubility differed depending on the sugar species. The solubility of trehalose was the lowest under the chosen aqueous solvent conditions compared to those of other sugars.[25] The concentrations of the sugars glucose, fructose, mannose, sucrose, and trehalose were varied from 5 to 45% w/w (54.0% g/mL), 5 to 52.5% w/w (64.8% g/mL), 5 to 42.5% w/w (50.5% g/mL), 5 to 52.5% w/w, and 5 to 35% w/w (40.3% g/mL), respectively. The conversion of concentration units from % g/g to % g/mL was carried out using the previously reported excluded volumes of the sugar molecules.[23,24] The measured q-range of the WAXS curves was ∼0.0525 to ∼2.2 Å–1, corresponding to a real-spatial distance (d = 2π/q) range of ∼120 to 3 Å. This q range was sufficiently wide for the analysis of the interparticle interactions between the sugar molecules and the intra-molecular structures based on the molecular sizes of the mono- and disaccharides.[21,22]

Figure 1

Concentration dependence of the WAXS curves of the sugar solutions at 25 °C. (A–E) Monosaccharides (glucose, fructose, and mannose) and disaccharides (sucrose and trehalose), respectively. The concentrations are given in % g/mL. The absolute scattering intensity calibration was performed using water (in 1 mm path length at 25 °C).

In the case of solutions in which the excluded volumes of the solute particles cannot be neglected, the observed scattering intensity of the particles (I(q,c)) in solution is known to obey the following relationship.[26]where Iu(q), Q(q), and O(q) are the spherical-averaged scattering functions of a single particle, pair correlation, and three-body correlation, respectively. Q(q) and O(q) are the functions normalized to 1 at q = 0. K is a factor related to the Thomson factor of an electron and to the average electron densities of the solvent and solute.[27] The first term reflects the intra-molecular structure, and the higher terms correspond to the intermolecular interferences, which are given by the virial coefficients of osmotic pressure from the expansion series. If a multi-body correlation is ignored, eq can be simplified as follows

By extrapolating the plot of c/I(q,c) versus c (called the Zimm plot) to c = 0, we can estimate M, Iu(q), A2 (the second virial coefficient), and A3 (the third virial coefficient), respectively. Figure depicts an example of the Zimm plot of the zero-angle scattering intensities I(0,c). The deviation of the plot from linearity indicates multi-body interactions, especially at high concentrations. The M, A2, and A3 values are listed in Table . The value of A2, which corresponds to pair correlations, was positive for all the sugars, clearly indicating that the sugar solutions were non-ideal solutions that were dominated by the excluded volume effect (repulsive force). Although the values of A3 corresponding to the three-body interactions (attractive force) were several percentage or less than those of A2, it was evident that the contribution of multi-body interactions was enhanced with increasing sugar concentration. The A2 values of the disaccharides (sucrose and trehalose) were smaller than those of the monosaccharides (glucose, fructose, and mannose). The A2 value of trehalose was larger than that of sucrose, indicating that the repulsive pair-correlation between the trehalose molecules was stronger than that between sucrose molecules. These results explain the analysis of the intermolecular correlation between the sugar molecules at high concentration reasonably well (discussed in the following section).

Figure 2

Zimm plot of the observed zero-angle scattering intensities. The dotted lines are the fitting curves considering up to the third virial coefficient.

Table 1

Experimental Values of Rg, I(0), Pmax, Dmax, and Molecular Weight and the Calculated Virial Coefficientsa

	Rg (Å)	I(0)abs (×10–2 cm–1)	I(0) (−)	Pmax (Å)	Dmax (Å)	M (−)	A2 (−)	A3 (−)
From data (5%)
glucose	2.78 ± 0.05	2.039 ± 0.006	1.000 ± 0.003	2.65 ± 0.05	8.0 ± 0.1 (13.6 ± 0.1)
fructose	2.70 ± 0.05	2.259 ± 0.006	1.108 ± 0.003	2.60 ± 0.05	7.0 ± 0.1 (17.2 ± 0.1)
mannose	2.67 ± 0.05	2.017 ± 0.006	0.989 ± 0.003	2.75 ± 0.05	7.6 ± 0.1 (13.9 ± 0.1)
sucrose	3.32 ± 0.02	4.003 ± 0.004	1.963 ± 0.002	3.20 ± 0.05	11.6 ± 0.1 (25.6 ± 0.1)
trehalose	3.68 ± 0.02	4.302 ± 0.004	2.110 ± 0.002	2.90 ± 0.05	13.4 ± 0.1 (30.9 ± 0.1)
By Extrapolation to c = 0%
glucose	2.84 ± 0.04		1.00 ± 0.01	2.70 ± 0.05	8.0 ± 0.1 (14.0 ± 0.1)	1	0.10 ± 0.01	0.0025 ± 0.0002
fructose	2.82 ± 0.04		1.10 ± 0.01	2.70 ± 0.05	9.5 ± 0.1 (17.1 ± 0.1)	1.11	0.100 ± 0.005	0.0022 ± 0.0001
mannose	2.62 ± 0.04		0.96 ± 0.01	2.80 ± 0.05	7.5 ± 0.1 (16.6 ± 0.1)	0.98	0.08 ± 0.01	0.0019 ± 0.0003
sucrose	3.63 ± 0.02		2.126 ± 0.008	4.00 ± 0.05	11.7 ± 0.1 (22.3 ± 0.1)	1.97	0.057 ± 0.004	0.0022 ± 0.0001
trehalose	3.94 ± 0.02		2.216 ± 0.006	4.65 ± 0.05	12.7 ± 0.1 (23.1 ± 0.1)	2.17	0.079 ± 0.002	0.0015 ± 0.0001

Rg: radius of gyration; I(0)abs: absolute zero-angle scattering intensity calibrated by water (1 mm path length at 25 °C; I(0): relative zero-angle scattering intensity normalized using the value of glucose; Pmax: peak position of the distance distribution function; Dmax: maximum diameter of the molecule. The Dmax values in brackets are the values obtained by considering the tailing region in the p(r) functions satisfying the condition p(r) = 0 for r > 0. M, A2, and A3 are the molar mass and the second and third virial coefficients, respectively, obtained using the Zimm plot. I(0) and M values were normalized by those of glucose.

To elucidate the monomeric structures of the sugar molecules in solution, the WAXS curves were obtained by extrapolating the observed WAXS curves in Figure to zero concentration using eq . Figure depicts the WAXS curves and distance distribution functions p(r) at zero concentration. Figure A,B presents the WAXS curves over the entire q region and in the high-q region (q > 0.3 Å–1), respectively; and C illustrates the p(r) functions. The zero-angle scattering intensities of the monomeric sugar molecules were obtained from the scattering curves in the small q region (<∼0.1 Å–1) in Figure A using the Guinier plot. The I(0) values determined by applying the Guinier plot to the extrapolated WAXS curve and to the observed WAXS curve (0.05 g/mL) are summarized in Table . In Table , the values of I(0) and M are normalized by those of glucose.

Figure 3

WAXS curves of sugar solutions obtained by the extrapolating the observed WAXS curves in Figure to zero concentration using the Zimm plot. These curves correspond to the form factors of the sugar molecules. (A,B) WAXS curves over the whole q region and in the high-q region (q > 0.3 Å–1), respectively. C is the distance distribution function, p(r) obtained via the Fourier transform of A.

The molar masses of the monosaccharides (glucose, fructose, and mannose) and disaccharides (sucrose and trehalose) are 180.2 and 342.3 Da, respectively. Although the ratios of I(0) or M between the monosaccharides and the disaccharides are proportional to their molecular weights (1.90), or rather to their numbers of electrons (1.896), the average ratios of I(0) and M in Table were ∼2.04. This indicated that the disaccharide molecules in aqueous solutions accompany more hydrated water than the monosaccharide molecules; the I(0) or M values of fructose were ∼10% larger than those of glucose and mannose. For the disaccharides, the I(0) and M values of trehalose were ∼7% larger than those of sucrose. These results (Table ) directly demonstrate the differences among the sugar structures, including the amounts of hydrated water molecules at the dilution limit in aqueous solutions. Furthermore, these findings were mostly in agreement with the previously reported partial molar volumes suggested by densitometry measurements.[28]

In Table , the Rg values determined by the Guinier plot using the scattering curve in the small-q region (q ≤ 1.3/Rg) clearly show that the difference in the sugar structures depends on both molecular mass and the entire structure, including the hydration shell. The structures of the monomeric sugar molecules in the solutions were characterized in more detail from the profiles of the WAXS curves in the high-q region (q > 0.3 Å–1) in Figure B. The slope of I(q) in the q region ∼0.3–0.7 Å–1 clearly differed depending on the sugar species (monosaccharides or disaccharides). This q region is sensitive to the molecular shape and reflects the differences between the isomeric forms in aqueous solution. The region of the scattering curves above q = ∼1 Å–1 corresponded to the internal structures of the sugar molecules (short-distance correlations including tightly bound water molecules). A broad hump at ∼1.65 Å–1 can be seen in Figure A,B, which indicates the presence of tightly bound water molecules, as explained in the next paragraph. The widths of the peaks of the p(r) functions in Figure C correspond to the average molecular sizes of the sugar molecules. In Table , the distance of the peak position (Pmax) of the p(r) function and the maximum distance (Dmax) of the solute particle are summarized. The Pmax value corresponds to the location of the center of gravity of the scattering density distribution of the solute particle. In the case of the monosaccharides, the p(r) functions had relatively sharp single peak profiles with the Pmax values of ∼2.7 Å. In comparison, the p(r) functions for both disaccharides showed a double peak shape, rather than a single peak shape. This is an intrinsic profile of a disaccharide composed of two units. It was evident that the center of gravity of the p(r) function of trehalose extended to a long-distance region compared with that of sucrose, suggesting that trehalose adopted an expanded structure. In the evaluation of the Dmax values, we used two different conditions. One was an ordinary criterion that Dmax must satisfy the condition p(r) = 0 for r > 0 (considering the tailing part), and the other ignored the tailing part. The differences between the profiles of the peaks of the monosaccharides and the disaccharides reflect their molecular shapes. Thus, the broad asymmetric peak profiles, which deviated from a symmetrical bell-shape, indicated that both the mono- and disaccharide molecules in solutions had non-spherical structures rather a spherical shape; this trend was more clearly seen in the case of the disaccharides. The tailing tendency of the p(r) function in the long-distance region suggested the presence of hydration layers that surrounded the sugar molecules and had electron densities higher than that of bulk water. Because sugars are known to act as a water structure-forming factor (kosmotrope),[29,30] the tailing profile of the p(r) function calculated from the scattering curves obtained by extrapolation was attributed to the intrinsic characteristics of sugar molecules surrounded by water molecules in infinitely dilute solutions. On comparing the peak profiles of the monosaccharides and the disaccharides, it was evident that the peaks of the disaccharides were much broader owing to their larger molecular sizes. For the disaccharides, trehalose was ∼1–2 Å larger than sucrose. Noticeably, the peak profile of trehalose showed a trapezoid-like shape. This reflected the differences in the chemical structure and flexibility between trehalose and sucrose, that is, trehalose is formed by an α,α-1,1-glucoside bond between two α-glucose units, while sucrose is formed by an α,β-1,2-glucoside bond between an α-glucose unit and a β-fructose unit. Such differences agree with the following theoretical simulation based on the sugar crystal data.

We compared the extrapolated experimental scattering curves and corresponding p(r) functions (Figure ) with the theoretical ones calculated from the atomic coordinates of the crystal structures of sugars registered in the Cambridge Structural Database (CSD, UK). Figure depicts the theoretical WAXS curves and p(r) functions of several sugars in aqueous solutions. Most of the sugar crystal data in the database were for anhydrate sugars; however, data for trehalose-dihydrate and α-glucose-monohydrate were also available. The crystal data used for the calculations in Figure were α-glucose-anhydrate (CCDC: 1169299),[31] α-glucose-monohydrate (CCDC: 1169293),[32] β-glucose-anhydrate (CCDC: 1169300),[33] β-fructopyranose-anhydrate (CCDC: 1160295),[34] α-mannopyranose-anhydrate (CCDC: 1488057),[35] β-mannopyranose-anhydrate (CCDC: 1450799),[36] sucrose-anhydrate (CCDC: 1263571),[37] trehalose-anhydrate (CCDC:668079),[38] and trehalose-dihydrate (CCDC:668078).[38]

Figure 4

Theoretical WAXS curves and distance distribution functions, p(r), of some of the sugars (α-glucose-anhydrate, α-glucose-monohydrate, β-glucose-anhydrate, sucrose-anhydrate, trehalose-anhydrate, and trehalose-dihydrate) in solution, calculated from the crystal atomic coordinates of the sugars registered in the Cambridge Structural Database. Panels (A–C) are shown as in Figure . The solid lines correspond to the hydrates and can be compared with the experimental data in Figure .

Figure presents the three-dimensional structures of α-glucose-monohydrate and trehalose-dihydrate oriented such that the position of the tightly bound water can be seen. It is worth noting that the comparison between the simulated and experimental data applies to the core structures of the sugar molecules, including tightly bound water in the crystal state and not to the structures including hydrated water in solution. The theoretical scattering curves were simulated by using the CRYSOL program.[39] This program uses the spherical-harmonics expansion method[40] and accurately reproduces the X-ray scattering curves of biological macromolecules in solutions; it takes the hydration shell into account by estimating the water-accessible surface area of a macromolecule.[39] In solution X-ray scattering, the scattering intensity of a solute particle is proportional to the square of the difference between the electron densities of the solute (sugar) and the solvent (water). In the aforementioned program, the average electron density of the solvent and the difference between the electron densities of the hydration shell (3 Å in width) and the solvent, which is known as hydration shell contrast, were variables. For large molecules such as proteins, the observed scattering curve could be reasonably well reproduced assuming a homogeneous hydration shell with an electron density around ∼1.1 times higher than that of bulk water.[41] However, the presence of such a homogeneous hydration shell is clearly inappropriate because mono- and disaccharides are sufficiently small (<∼8 Å in diameter assuming a spherical particle). For example, the electron density, the excluded volume, and the hydration shell volume of the glucose monomer, calculated by CRYSOL, were 0.4457 e/Å3, 215.4, and 833.1 Å3, respectively. If we use the default value of the hydration-shell electron density (1.1 times that of bulk water) in the CRYSOL program, the Rg value becomes 4.78 Å. This value is too large and overestimated. Therefore, in the use of the CRYSOL program, we set the electron densities of the solvent and the hydration shell as 0.33 and null e/Å3, respectively. The number of spherical harmonics used was up to the 50th order, that is, the maximum value. It should be mentioned that the zero-angle scattering intensity, I(0), calculated by the CRYSOL program gives the microscopic differential scattering cross section for one molecule, expressed in terms of electrons squared for a single molecule as follows

Figure 5

Three-dimensional structures of α-glucose-monohydrate and trehalose-dihydrate displayed using PyMol. The orientation was arranged so that the position of the tightly bound water could be seen. Red, green, and white balls correspond to oxygen, carbon, and hydrogen atoms, respectively. C–C bond distance is 1.5 Å. O–O distance of bound waters in trehalose-dihydrate is 9.7 Å.

Therefore, the square-root of the theoretical I(0) value could be compared with the experimental I(0) value defined by eq . Table summarizes the structural parameters obtained from the simulated WAXS curves based on the crystal structures. The square-root of the zero-angle scattering intensity, I(0)1/2, was normalized using that of glucose. The I(0)1/2 values are simply proportional to the molecular weights, which in turn depend on whether the sugar is a monosaccharide or disaccharide and anhydrate or hydrate. The Rg values in Table were determined using the Guinier plots in the same q range (0.055–0.1 Å–1) used for Table . The p(r) functions were also calculated under the same conditions as those used for the experimental data. The theoretical Rg values in Table are much smaller than the experimental ones in Table . This clearly indicates that the sugar molecules in solution are more hydrated than the anhydrate and/or hydrate forms present in the crystals. This trend is the same as that reported for hydrated proteins in solution.[41]

Table 2

Structural Parameters of Sugars in Solution Calculated from Sugar Crystal Dataa

	Rg (Å)	I(0)1/2 (−)	Pmax (Å)	Dmax (Å)
α-glucose anhydrate	1.464 ± 0.001	1.000 ± 0.001	2.80 ± 0.05	6.97 ± 0.05
β-glucose anhydrate	1.425 ± 0.001	1.000 ± 0.001	2.85 ± 0.05	6.85 ± 0.05
α-glucose monohydrate	1.923 ± 0.001	1.145 ± 0.001	3.00 ± 0.05	8.19 ± 0.05
β-fructose anhydrate	0.967 ± 0.004	1.000 ± 0.001	2.95 ± 0.05	5.82 ± 0.05
α-mannose anhydrate	1.102 ± 0.001	1.000 ± 0.001	3.00 ± 0.05	6.13 ± 0.05
β-mannose anhydrate	1.531 ± 0.001	1.000 ± 0.001	2.70 ± 0.05	7.12 ± 0.05
sucrose anhydrate	2.460 ± 0.001	1.855 ± 0.001	3.50 ± 0.05	8.77 ± 0.05
trehalose anhydrate	2.875 ± 0.001	1.855 ± 0.001	3.45 ± 0.05	10.53 ± 0.05
trehalose dihydrate	3.327 ± 0.001	2.145 ± 0.001	4.95 ± 0.05	11.53 ± 0.05

Rg: radius of gyration; I(0)1/2: square root of the zero-angle scattering intensity normalized by the value of glucose; Pmax: the location of the peak position in the distance distribution function; and p(r): Dmax is the maximum diameter of the molecule. The values were obtained from the theoretical scattering curves in Figure under the same calculation conditions used for the experimental data.

While the slopes of the q region ∼0.3–0.7 Å–1 in Figure A reflect differences among the molecular shapes of the sugars in solutions, the q region above ∼1 Å–1 is very sensitive to the conformational isomer (α- or β-pyranose) of the sugar, whether it is a hydrate or anhydrate, and the number of glycosidic linkages. The presence of tightly bound water molecules increased the scattering intensity at ∼1.65 Å–1. Therefore, the broad hump at ∼1.65 Å–1 in the experimental results in Figure B was attributed to water molecules tightly bound to the sugar molecules. The theoretical maximum diameters (Dmax) of the sugar molecules in Figure C, which were determined from the p(r) function with the condition p(r) = 0 for r > 0, were much smaller than the experimental values in Figure C. Unlike the experimental p(r) functions, the theoretical p(r) functions did not exhibit a tail on the long-distance side because the simulation did not consider hydration layers but instead only one or two tightly bonded water molecules. Noticeably, the theoretical Dmax values of sucrose anhydrate and trehalose anhydrate were very different (8.8 and 10.5 Å, respectively). This originated from their different glycosidic bonds and constituent units; namely, the α,α-1,1-glucoside bond between two α-glucose units or the α,β-1,2-glucoside bond between an α-glucose unit and a β-fructose unit. In addition, the Dmax value of trehalose dihydrate increased to 11.4 Å, and its p(r) function profile changed to a trapezoid-like shape. These characteristic trends agree well with those observed in the experimental p(r) functions in Figure C.

The above experimental and theoretical results indicate that the structures of the sugars in relatively low concentration solutions were more hydrated structures than in the crystalline state. In particular, as reflected in by the asymmetry of its p(r) function, hydrated trehalose had a more anisotropic structure than hydrated sucrose, which was suggested by MD simulations.[20]

Intermolecular Correlation between the Sugar Molecules in Concentrated Solutions

Figure presents the enlarged view of the WAXS curves in the high-q region (q > 0.3 Å–1) of Figure . Upon increasing the sugar concentration, the nearest-neighbor correlation peaks between the sugar molecules, which are indicated by the arrows in Figure , became clearer. These peaks were attributed to the presence of repulsive intermolecular interactions between the sugar molecules. For the monosaccharides, the correlation peak at q = ∼0.83 Å–1 was seen at ∼28–31% g/mL for glucose, ∼25–28% g/mL for fructose, and ∼28–31% g/mL for mannose. These peaks shifted toward the high-q region when the concentration was further increased, namely, to ∼0.94 Å–1 at 54% g/mL for glucose, to ∼0.98 Å–1 at 64.8% g/mL for fructose, and to ∼0.93 Å–1 at 55% g/mL for mannose. These shifts corresponded to the reduction in the intermolecular distance from ∼7.6 to 6.4 Å. For sucrose, the q value of the peak changed from ∼0.59 Å–1 at ∼28–31% g/mL to ∼0.72 Å–1 at 64.8% g/mL, corresponding to a reduction in the intermolecular distance from ∼10.6 to 8.2 Å. In the case of trehalose, this correlation peak did not appear until 34% g/mL, and a rather broad shoulder around q = ∼0.59 Å–1 became visible by ∼37–40% g/mL. Using the dependence of the densities of the sugar solutions on their concentrations,[2,23,24] we tentatively estimated the average distance between the sugar molecules for the simplest case (assuming a uniform dispersion), as followswhere D is the average distance and Na is Avogadro’s constant. When the sugar concentration is changed from 28 to 64.8% g/mL, the calculated average intermolecular distances would decrease from 12.7 to 9.58 Å for the monosaccharides and from 15.7 to 11.9 Å for the disaccharides. However, the observed values in Figure were much shorter. This suggests that the sugar molecules were not uniformly dispersed at high concentrations and instead adopted shorter correlation distances. These shorter distances would be necessary to explain the high solubility of the sugars. Notably, in comparison to the other sugars (glucose, fructose, mannose, and sucrose), this trend was weaker for trehalose. This was attributed to the fact that the interaction between trehalose and water was stronger than those of the other sugar molecules, as suggested in the above section. The above results would relate to the previous finding that trehalose forms dihydrate crystals, thereby shielding the remaining glassy trehalose from the effects of added water.[19]

Figure 6

Enlarged view of the observed WAXS curves in the high-q region (q > 0.3 Å–1) of Figure A–E. Panels (A–E) are as in Figure . The arrows indicate the intermolecular correlation peaks.

The presence of nearest-neighbor correlations between the sugar molecules became clearer in the oscillatory profile in the p(r) functions upon increasing the concentration (Figure ). The observed scattering curves in Figure contained information of both the structure and form factors. The p(r) functions in Figure were obtained by the direct Fourier transform using eq without the removal of the interparticle correlation. In the long-distance region, every p(r) function in Figure smoothly damped to zero, suggesting that there did not exist a so-called truncation effect in the Fourier transformation. The p(r) functions in Figure clearly showed the presence of the interparticle correlation, especially at high concentration. In the case of the repulsive interparticle interactions, the p(r) function is known to show a strong negative value that is called a correlation-hole.[42] As shown in Figure C, the first positive peak originated from the self-convolution of the scattering density distribution of a single sugar molecule.[43] The second peaks indicated by the full arrows in Figure correspond to the first nearest-neighbor intermolecular correlation. The positions of the peaks reflect the average distance to the first layer of nearest-neighbor sugar molecules. At the highest sugar concentration of 64.8% g/mL, these distances were ∼8.3 and ∼10.3 Å for fructose and sucrose, respectively. These values are comparable to the maximum diameters of the sugar hydrates shown in Figure . This meant that the sugar molecules were accessible to each other to their nearest position, depending on the concentration, despite the individual molecules being in a hydrated state. At this concentration, another intermolecular correlation peak corresponding to the second nearest-neighbor layer also appeared, with distances of ∼14.1 and ∼18.9 Å for fructose and sucrose, respectively. The ratio of the distances between the first and second nearest-neighbor correlation peaks was ∼1.7, suggesting a close-packing arrangement of identical particles. The first nearest-neighbor correlation peaks of sucrose and trehalose were compared at the same concentration (40% g/mL); the peak for sucrose was more clearly seen at ∼10.8 Å, whereas trehalose exhibited a rather broad peak at ∼11.9 Å. This suggested that trehalose tended to avoid a close-packing arrangement and preserve its excluded volume in solution.

Figure 7

Distance distribution functions, p(r), for different sugar concentrations. The p(r) functions were obtained by the Fourier transform of the WAXS curves in Figure . Panels (A–E) are as in Figure . The first positive peaks correspond to the self-correlation of the sugar molecules. The second positive peaks, indicated by the full arrows, correspond to the correlation distance of the first layer of nearest-neighbor sugar molecules. The open arrows indicate the correlation holes. At the highest concentration (52.5% w/w in fructose and sucrose), another correlation corresponding to the second layer of nearest-neighbor sugar molecules appears.

Figure presents a comparison of the p(r) functions of sucrose and trehalose at a few concentrations. The highest concentrations of sucrose and trehalose were 64.8 and 40% g/mL, respectively. At a concentration of 40% g/mL, the first nearest-neighbor correlation peak of trehalose at ∼12 Å was much broader than that of sucrose, indicating that trehalose tended to disperse more randomly and/or disordered in solution. In addition, at 64.8% g/mL, the p(r) function of sucrose clearly showed a second nearest-neighbor correlation peak at ∼19 Å.

Figure 8

Comparison of the p(r) functions of sucrose and trehalose at 40% g/mL (the highest concentration of trehalose). The p(r) function of 64.8% g/mL sucrose is also shown. These p(r) functions were reproduced from Figure D,E. The arrows indicate the first and second nearest-neighbor correlation peaks.

Conclusions

Using the WAXS technique, we characterized the structures of several sugars in solution in detail over the concentration range of 5.0–64.8% g/mL and over a wide real-space resolution range from ∼120 to 3 Å. Comparison of the experimental structure parameters at infinite-dilution states with the theoretical ones, which were calculated based on the crystal structures of the sugars, indicated that the sugar monomers (glucose, fructose, mannose, sucrose, and trehalose) in solution were more hydrated than those in the anhydrate and/or hydrate crystals. This result agrees well with those from previous NMR and MD simulation studies.[11−14] In particular, trehalose showed a tendency to form a more expanded and/or bulky structure than sucrose. This trend is consistent with the neutron diffraction and MD simulation results that indicated that trehalose forms an anisotropic hydration structure.[20] In addition, the recent terahertz spectroscopic study[44] showed the presence of a rigid but destructured hydrogen bond network around disaccharides. This destructured water structure was simply a hydration shell with an electron density higher than that of bulk water around the sugar molecules. The previous finding supports the present results.

Although the biological functions of sugars have been well studied, few experimental reports are available on the structures of sugar solutions. Therefore, we studied the structural properties of concentrated sugar solutions in detail. We obtained direct evidence of the presence of a nearest-neighbor correlation between the sugar molecules at high concentrations. A broad hump or peak in the scattering curves corresponding to the nearest-neighbor correlations became visible as the sugar concentration was increased. The determined second virial coefficients (corresponding to two-body interactions) showed that repulsive interactions were dominant for the sugar solutions at low concentrations. However, the observed nearest-neighbor correlation distance between the sugar molecules at high concentrations was shorter than that obtained assuming a homogeneous dispersion of sugar molecules in solution. This meant that the sugar molecules were more accessible to each other, despite the presence of the hydration repulsion force between them. The values of the third virial coefficients (corresponding to three-body interactions, i.e., attractive forces), which were several percentage or less of the second virial coefficients, indicated the same situation. Thus, the contribution of multi-body interactions was enhanced by increasing the sugar concentration. This also implied that an inhomogeneous distribution of sugar molecules was unavoidable in highly concentrated solutions owing to multi-body interactions. For fructose and sucrose, at the highest concentration of 64.8% g/mL, the first and second intermolecular correlation peaks were clearly observed. The ratio of the distances between the first and second nearest-neighbor correlation peaks indicated a close-packing arrangement of identical particles, demonstrating that the structures of the sugar solutions were essentially subject to the entropic gain of the system, including hydrated water molecules. In comparison with the properties of other sugars (glucose, fructose, mannose, and sucrose), trehalose exhibited a tendency to avoid a close-packing arrangement and to preserve a large excluded volume. The correlation distance of trehalose at 40.3% g/mL was ∼1 Å longer than that of sucrose at the same concentration. This was attributed to the bulky structure of trehalose in solution. This finding agrees well with the present results that the intermolecular correlation distance of trehalose was longer than that of sucrose. Further, this trend was consistent with the fact that the greater dynamical slowing down of trehalose was significant compared to that of sucrose, as revealed by the terahertz spectroscopy.[44] In addition, this solution property of trehalose would be closely related to its higher Tg than those of other sugars, as reported previously based on calorimetry and MD simulations.[16−20,45] In addition, we found that trehalose molecules tend to disperse more randomly and/or disordered in solution compared to other sugars. Considering that the anisotropic hydration properties of sugars, which depend on the type of sugar species and their glycosidic bonds,[13,17] greatly affect the orientation and dynamics of water molecules surrounding sugar, the above results are reasonable.

The function of sugars in preventing protein denaturation and enzyme deactivation has been attributed to the specific binding of sugars with proteins and/or the change in the free-energy upon the transfer of proteins into sugar solutions. Sugars function as typical kosmotropes, based on the original definition of a kosmotrope as a substance that stabilizes proteins and hydrophobic aggregates in solution and reduces the solubility of hydrophobic compounds.[46] Recently, using X-ray and neutron scattering methods, we showed that sugars (trehalose and glucose) and a polyol (glycerol) were preferentially or weakly excluded from the protein surface to retain its native hydration shell.[22,23] We also observed the sugar-mediated stabilization of a protein against the addition of a chemical denaturant and temperature elevation, particularly for trehalose.[23] The present results show that compared to other sugars, trehalose tends to adopt a bulky solution structure owing to its anisotropic hydration property. This suggests that the restriction of the dynamics of water surrounding trehalose is quite strong. On the other hand, hydration (solvation) has already been shown to be a key determinant of the structural stability and function of proteins[5,6,47] and to be strongly coupled with protein dynamics.[48,49] Therefore, such a structural property of trehalose solution would relate to its protective action, not only against the chemical and thermal denaturation of proteins but also against the denaturation by desiccation and freezing. We expect that this characteristic of trehalose solution will afford new insights into its functions.

Experimental Section

Materials

The following sugars were used: trehalose (crystalline dihydrate powder, ≥98.0% by HPLC) was obtained from Hayashibara Co. (Okayama, Japan) and sucrose (≥99.0% by HPLC) and fructose (crystalline anhydrate powders, ≥ 99.0% by HPLC) were obtained from Wako Pure Chem. Co. (Osaka, Japan); d-glucose (crystalline anhydrate powder, ≥ 99.5% by GC) was obtained from Sigma-Aldrich; and d-mannose (crystalline monohydrate powder, ≥ 95% by SDS-PAGE) from Chemily Gycoscience, Atlanta, United States. All other chemicals were of analytical grade. The solvent used was 10 mM N-(2-hydroxymethyl) piperazine-N’-(2-ethane-sulfonic acid (HEPES) at pH 7.0. The sugar concentrations were varied from 5 to 45% w/w for glucose, from 5 to 52.5% w/w for fructose, from 5 to 42.5% w/w for mannose, from 5 to 52.5% w/w for sucrose, and from 5 to 35% for trehalose, respectively. The above concentrations were those of the anhydrate sugars.

X-ray Scattering and Data Treatments

Synchrotron radiation WAXS measurements were performed using the BL-10C spectrometer of the photon factory (PF), at the High Energy Accelerator Research Organization (KEK, Tsukuba, Japan). The X-ray wavelength and the sample-to-detector distance were 1.55 Å and 24 cm, respectively. The X-ray scattering intensity was recorded by the PILATUS3 2M detector (253.7 × 288.8 cm2 area, 172 μm pixel-resolution) from Dectris Co. The exposure time was 30 s. The solutions were contained in sample cells with a pair of thin quartz windows (20 μm in thickness) and 1 mm in path length. The temperature of the samples was maintained at 25 °C using a model mK2000 temperature controller (Instec, Inc., USA).

Background correction of the wide-angle scattering data was performed based on a previously reported method.[50,51]where I(q)sol, I(q)solv, and I(q)cell are the observed scattering intensities of the sugar solution, the aqueous solvent, and the blank cell, respectively; Bsol, Bsolv, and Bcell are the respective incident beam intensities; Tsol, Tsolv, and Tcell are the respective transmissions; and c and va are the concentrations of the sugars and their partial specific volumes, respectively. Previously reported va values were used.[23,24,28] The calibration of the observed scattering intensities into those absolute scattering intensities were performed using the scaling factor determined from the observed scattering intensity of water (at 25 °C with 1 mm path length) based on the method reported previously,[27,52] where the absolute scattering intensity (dΣ/dΩ) of water at 25 °C used was 0.01646 cm–1.[52] The distance distribution function, p(r), was calculated by the Fourier inversion of the scattering curve I(q) as follows[53]where q is the scattering vector (q = (4π/λ) sin(θ/2) and θ is the scattering angle and λ is the X-ray wavelength). The maximum diameter of the particle, Dmax, was determined from the p(r) function satisfying the condition p(r) = 0 for r > 0. The radius of gyration, Rg, and the zero-angle scattering intensity, I(0), were determined by applying the Guinier plot (ln I(q) vs q2) to the scattering curve in the small-q region (q ≤ 1.3/Rg).[54]