NMR Studies on Water and Polvmer Diffusion in Dextran Gels.

Influence of Potassium Ions on Microstructure Formation and Gelation Mechanism

J

Tokuko Watanabe, Akihiro Ohtsuka, Norio Murase, Peter Barth, Klaus Gersonde
At room temperature aqueous solutions of dextrans with concentrations >25% (w/w) exhibit a sol-gel transition in the presence of >1.0 M potassium chloride. In dextrans the gelation was unexpected due to missing anionic groups that usually provide the binding sites for cations. The quantitative investigation of the gel formation is based on changes of the diffusibility of water and dextran chains. The apparent diffusion coefficients of bulk water (in the order of cm%) and of water trapped in the junction zones as well as of polymer chains (in the order of lop7 to cm2/s) are determined by employing pulsed field gradient stimulated echo (PFGSTE) NMR. The restricted diffusion of bulk water in viscous sols and in soft and rigid gels has been quantitatively analyzed providing data for interbarrier distances (pore size), permeabilitiesof the diffusion barriers (density of junction zones) and interbarrier diffusion coefficients of water. Based on already published x-ray structure data and in accordance with the diffusion data presented in this paper "potassium-bonding" is assumed to be the most important interaction for the formation of a microstructure and for the stabilization of crosslinks. The ionic radius of the potassium ion perfectly fits to the cage established by six oxygen atoms of glucose units of three polymer chains. Other cations, such as Lif, Na+, Rb+ and Cs+, according to their nonfitting ionic radii, do not provoke dextran gelation under these conditions. The mechanism of the transitions from sol to soft gel and further to rigid gel is discussed on the basis of restricted diffusion and x-ray structure data.

INTRODUCTION

Polysaccharides are important components in food materials (1).Nowadays, the polysaccharides become more and more important due to their tremendous contribution to biochemical reactions and biochemical in vivo functions of biological systems (2). Bound to other biopolymers, such as proteins and lipids, polysaccharides stabilize the biopolymers in their structural orders (i.e., conformations), to enable very specific functions of

MRM 35697-705 (1996) From the Laboratory of Chemistry, Tokyo University of Marine Science, Minato-ku, Tokyo (T.W., A.O.), Tokyo Denki University, Saitama (N.M.), Japan; and Fraunhofer-lnstitut fur Biomedizinische Technik. St. lngbert (P.B., K.G.) and Fachrichtung Medizintechnik, Universitat des Saarlandes, Homburg/Saar (P.B., K.G ), Germany. Address correspondence to: Tokuko Watanabe, Ph.D., Laboratory of Chemistry. Tokyo University of Marine Science, 4-5-7, Konan, Minato-ku, 108, Tokyo, Japan. Received March 29,1995; revised November 2,1995; accepted December 8, 1995. This work was supported by the Fonds der Chemischen lndustrie (K.G.) and by the cooperative research program with the National Institute of Physiological Sciences (T.W., A.O.). 0740-3194/96 $3.00 Copyright 0 1996 by Williams & Wilkins All rights of reproduction in any form reserved. 697

the in vivo system (3,4). It is the interaction between water and polysaccharides that plays an important role for the stabilization of the protein conformations. Therefore, the contribution of polysaccharides to biological functions is strongly linked to the hydration of these macromolecules. Sugar chains exposed on the surface of cells can be considered as antenna for the recognition by other cells (5-8). They also prevent cells from aggregation. Tumor cells, for example, aggregate because of the lack of such sugar chains on the cell surface (9). Hence, polysaccharides carry the information to specify cells and to contribute to the cell recognition in in vivo systems via their structural changes. The elucidation of structural changes of polysaccharides is therefore necessary for a better understanding of in vivo functions of polysaccharides. In many natural polysaccharides a reversible sol-gel phase transition can be observed, which reflects changes of the physicochemical conditions, such as temperature, ionic strength, pH and/or interactions with other polysaccharides. In the gel state a three-dimensional network of cross-links can be observed (10). Several phenomena, such as the partial formation of micro-crystalline regions, hydrogen-bonding, ionic interactions, and helix formation, accompany the gelation of polysaccharides. It is already well known, that Ca2+ and K+ ions enhance the process of cross-linking in polymers of alginic acid and K-carrageenan, respectively (11-14). Models that explain the structure changes and the role of the above mentioned cations in the sol-gel transition have been proposed. The mechanism of the sol-gel transition in alginic acid is based on the assumption that the block chains form so-called egg box junctions due to the intercalation of CaZ+ between the chains (11).In K-carrageenan single or double helices easily aggregate in the presence of K' , which is intercalated between the helices and forms junction zones (12-14). In both cases anionic groups, such as carboxyl and sulfate groups of the glucose units, play an important role for the formation of cross-links and junctions via ion-bonds. Dextran forms linear chains with a-l,6-linked D-glucose units. More rarely the side chain formation is based on a-1,3 and a-1,4 interconnections. Electron microscopy data prove that soluble and slimy polysaccharides in water, such as dextran, A-carrageenan, succinoglycan, and xanthan gum, cannot form gels (15). Gel formation in aqueous dextran solution in the presence of K' , however, was found by Kajiwara et al. on the basis of differential calorimetry measurements and in particular by observing turbidity and elasticity phenomena (16). Dextran does not provide anionic groups for the binding of

698

Wataiiabe et al. stored at room temperature for 1 or 2 days to obtain the stationary state of the gel. Diffusion measurements were performed in the stationary state of the gel at 25C.

cations. The mechanism of gelation of dextrans in the presence of K + is still unknown. It is the purpose of the present work to elucidate the mechanism of the dextran gelation induced by potassium chloride. Furthermore, structural changes occurring during the sol-gel transition, i.e., the transition from random coils of dextrans to a three-dimensional network of crosslinked dextrans, will be investigated as reflected by the diffusibility of water and dextran chains in aqueous solutions in the absence and in the presence of potassium chloride. Dynamic light-scattering data have strengthened the evidence that the fluctuation and/or the translation of the gel network can be interpreted on the basis of the diffusion equation (17). Employing pulsed field gradient NMR (PFG-NMR) with strong field gradients slow diffusion processes can be investigated, resulting in diffusion cocmZ/s.The analysis efficients of the order of lop7 to of the diffusion of polymer chains allows a deeper and a more direct insight into the structural flexibility of sugar chains and into the interactions between sugar chains and water. The diffusional motion of water is mostly influenced by the structure of the gel. Therefore, this parameter reflects properties of the gel structure. A guide for prospective experimental studies of the diffusion in gels has been recently published comparing and reviewing methods for diffusion measurements. This comparison reveals that PFG-NMR is a very accurate and precise method for diffusion measurements (18). In a previous paper diffusion coefficients of water in dextran solutions as well as in the covalently cross-linked dextran gels (sephadex G) were presented indicating that the diffusion of water in the gel network is strongly restricted, while that in dilute dextran solutions seems to be rather free (19). The difference of the diffusional behavior of bulk water in sols and in gels could be an indicator for the transition between sol and gel state and could define the structure of gels.

f Diffusion Coefficients Determination o

The determination of the diffusion coefficients of water and polymer chains was performed employing the pulsed field gradient stimulated echo NMR method (PFGSTE-NMR) (see Fig. 1). The determination of the diffusion coefficient within an infinitely large, homogeneous system is based on the following Eq. [ I ] ( 2 0 , 21):
$ ( t = T~ + T ~ = ) exp{-y2D[6'(A
-

6/3)$

+T

~ ~ ( ~ T~ ~ / 3 ) g , 2 - 6 ( t ~ ' + t ~ ~ + 6 ( t , + tI11 ~)

+2zi2/3-2

T1T2)gg0])

MATERIALS AND METHODS

Preparation of Dextran Gels Dextran powders (products of Pharmacia, Stockholm), 10,000 Dalton) and T70 (Mw such as TI0 (Mw 70,000 Dalton), were dissolved at room temperature in deionized, distilled water to prepare a dextran sol. Polymer gels were prepared by mixing aqueous potassium chloride solutions with the dextran sol at 80C under stirring. The final concentrations were in the range of 0.5 to 2.0 Mfor potassium chloride and 10,20,30,and 34.4% (weight of dry dextran per total weight of gel) for dextran. The pH value of the K'-free dextran solution was 6.1 at 23C and increased to 6.6 in the presence of 1.5 M potassium chloride concentration. At 80C the dextran/potassium chloride mixture is in the sol state. Cooling down to room temperature leads to the formation of gels. The time of gelation depends on the concentrations of dextran and potassium ion. For example, at room temperature T70 (30'1/0)in the presence of 1.5 M KCI forms a gel within 1 day, whereas T70 (34.4%) in the presence of 1.5 M KC1 already forms a gel within a few hours. The sample was

The time parameters t,, t,, T,, r 2 ,6, and A are defined in Fig. 1. The diffusion time A is the time window during which the diffusion is observed. The symbols g and g, stand for the magnitude of the field gradient and of the static field, respectively. The amplitudes of the water and polymer chain proton signals were determined by varying the x gradient at constant A. Fourier transformation of the stimulated echo resulted in two resonances, i.e., for water and for chain protons. The apparent diffusion coefficient D,, was determined from the signal attenuations of both types of protons employing least square fits and Eq. 1 1 1 . The diffusion coefficients were measured following two experimental regimes: (i) Optimal parameter setting for the investigation of the diffusion of bulk water was as follows: A = 60 ms; 6 = 4 ms; g is in the range of 10 to 150 mT/m. (ii) Optimal parameter setting for the investigation of the diffusion of hydration water and polymer chains was as follows: A = 410 ms; 6 = 4 ms; g is in the range of 100 to 400 mT/m. Under both experimental regimes the static gradient field go = 0.01 mT. The diffusion measurements were performed in a 9.4 Tesla NMR spectrometer (MSL400, Bruker Analytische Messtechnik, Karlsruhe, Germany). The temperature was control led at 25 ? 0.5"c. At 400 MHz the water and poly merchain proton signals are well separated. Figure 2 exhibits a set of NMR spectra for the dextran sample T70 (30%) according to varying field gradients ranging from 25 to 400 mT/m at A = 210 ms. The proton resonances at 0 ppm and -1.2 ppm were assigned to water and polymer chain protons, respectively (22). Spectra obtained with smaller field gradients than 100 mT/m exhibited a large water proton signal and a relatively small polymer
x/2

.+ ;*.
'6a

_ .

)ii . 3 . *........ *
.
tl

. go

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . tz 7, I . . ! . . . . . . . . . . . . z* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 +zz . )i

* ,

*:

i 2-T

FIG. 1. Pulsed field gradient stimulated echo sequence for study-

ing diffusion processes (for symbols compare Eq. [ l ] ) .

Water Diffusion in Dextran Gels Formed by K+-Binding

699

-5
16

13
11

11

2
10-6. b Value [s/crn21

-5

-10

-15

Chemical Shift [pprn]

FIG, 2. 400 MHz proton NMR spectra o f dextran T70 (30%) at 25C employing magnetic field gradients with steps o f 25 mT/m. Spectrum 1 correspondsto g = 25 mT/m and spectrum 16 to g = 400 rnT/m. A = 210 rns; 6 = 4 ms; T~ = 10 ms; T~ = 210 ms; go = 1 mT/m. The insertion indicates t h e amplified spectrum 14 (g = 350 mT/m).

FIG. 3. Semilogarithmic plot of the proton signal attenuation of bulk and trapped water in dextran T70 (30%)versus the b value (b = YS'(A - 6/3)g2)in the absence o f K+.The apparent diffusion coefficients of both water components are represented by the two slopes o f t h e plot. For experimental conditions see legend o f Fig. 2.

chain proton signal. In this case the resonance line of the chain protons was largely overlapped by the water proton resonance. If the field gradient, however, was larger than 100 mT/m (in this case the water signal intensity decreased due to the diffusion), the separation of water proton (0ppm) and chain proton resonances (-1.2 ppm) was good enough to allow an accurate determination of the signal amplitudes of water and polymer chain protons (see insertion of Fig. 2). We used diffusion times of 410 ms to reduce the signal intensity of water, which is rapidly diffusing, and to observe now the signal attenuation of water, which is slowly diffusing. Furthermore, to measure the amplitudes of the relatively small signals of the chain protons and of the slowly diffusing water with higher sensitivity. we recorded only spectra obtained at field gradients with g > 100 mT/m. With this experimental strategy it was possible to measure the rapid diffusion of bulk water as well as the slow diffusion of the trapped water (for explanation see Discussion) and polymer chains.

RESULTS
Diffusion Coefficients o f Bulk and Trapped Water in Dextran Sols and Gels

Figure 3 exhibits the NMR signal attenuation of the water proton signal in dextran sol T70 (30%) in the absence of potassium chloride for the diffusion time of 210 ms. The water signal intensity decreases with the increasing field gradient strength. The decrease of the water signal intensity in the range of g = 0 to 400 mT/m reflects two phases, indicating two types of water that differ in their diffusion coefficients. The two slopes of the semilogarithmic plot shown in Fig. 3 represent the apparent dif-

fusion coefficients (7.55 5 0.07) x lo-" and (6.7 2 0.5) x cm2/s. In all sols and gels of dextran, diffusion coefficients for two types of water are observed, which differ by about two orders of magnitude. The rapidly diffusing water exhibits diffusion coefficients in the order of 10 cm"/s assigned to bulk water. The slowly diffusing water shows diffusion coefficients in the range of to lo-" cm"/s and has to be attributed to trapped water (primary hydration water) (23). The residence time of the primary hydration water (which includes the trapped water] should be, although exibiting a very slow motional exchange with bulk water, in the order of 1 s (23). The chemical exchange between protons of trapped water and hydroxyl protons of the polymer is also slow (the residence time in this case is at maximum 50 ms) (24). If in diffusion experiments one employs long diffusion times, e.g., 410 ms, the exchange between bulk and trapped water is unlikely, whereas the exchange between protons of trapped water and polymer hydroxyls could have little influence on the diffusion of trapped water. This should then result in a lowering of the signal intensity of trapped water. In this case D,, of trapped water appears to be larger than it should be. The evidence for a very small influence of the above mentioned exchange processes is manifested by the fact that D,, of trapped water is a little bit larger than that of polymer CH, protons, which do not exchange with other protons at all (see Fig. 5B).
+

Diffusion Coefficients of Polymer Chains in Dextran Sols and Gels

The semilogarithmic plot of the attenuation of the NMR signal at -1.2 pprn attributed to CH, protons of the polymer chains reflects the contribution of only one type of protons. The diffusion of polymer chains in T70 (30%) in the absence of potassium chloride (sol state] is dem-

700

Watanabe et al.

.n

14.2-

g C

e . .14.1d

0 . 0 0 5

2
14.0.

2
1
10-6

0.003

1:s

d
. b Value
[s/cm2]

Potassium Chloride Concentration [MI

FIG. 4. Semilogarithmic plot of t h e chain proton signal attenuaf Fig. 3) tion in dextran T70 (30%)versus the b value (see legend o in the absence o f K ' . The apparent diffusion coefficient of the polymer chains is represented by the slope of the plot. A = 410 ms; 6 = 4 ms; T, = 10 ms; 72 = 410 ms; go = 1 mT/m. For further experimental conditions see legend of Fig. 2.

FIG. 5. Influence o f the potassium chloride concentration at 25C on the apparent diffusion coefficient in T70 (34.4%).(A) Bulk water (0, 0); (B) Trapped water (0, B)and polymer chains (A, A).Open symbols indicate t h e sol state, full symbols the gel state; The arrow assigns the half-transition point o f the sol-gel transition at s marked by shadowing. 0.6 M potassium chloride; The gel phase i The standard deviation in (A) is C 1'YO and in (B) as shown by bars.

onstrated in Fig. 4. In this sol the apparent diffusion coefficient of the polymer chains is (4.5 2 0.5) x cm'/s (A = 410 ms). 4 potassium chloride, the diffuIn the presence of 1.5 A sion coefficient of polymer chains in T70 (30%) (gel state) is (3.0 f 0.5) X lop8 cmz/s (A = 410 ms). Although this apparent diffusion coefficient is significantly smaller than that in the sol, both Dilpvalues are in the same order of magnitude. The D,, values of polymer chains differ, however, by two orders of magnitude from that of bulk water in the sol as well as in the gel state, for example in T70.

The Dar, values of trapped water and polymer chains in dextran T70 (34.4%) (Fig. 5B) also reflect the cffect of increasing amounts of potassium ions on the sol-gel transition. In the gel state a high potassium ion concentration does not affect the apparent diffusion coefficient of the chains. In this state the chains are mostly immobilized.

Influence of the Dextran Concentration on the Diffusion f Water and Polymer Chains Coefficients o
In Figs. 6A and 6B the diffusional behavior of bulk water at varying concentrations of T70 and TI0 dextrans is represented. With increasing dextran concentration the diffusion coefficient of bulk water decreases remarkably because of the increasing viscosity and the obstraction effect of the polymer chains. In the absence of potassium ions and at identical polymer concentrations, the diffusional behavior of bulk water appears to be identical in both polymer solutions. T70 dextran in the concentration range of 10-34.4% exhibits diffusion coefficients that reflect a strong influence of 1.5 Mpotassium ions (Fig. 6A). Despite this effect of 1.5 Mpotassium ions, below 25% T70 dextran remains in the sol state and at 25% in a soft gel state. Binding of potassium ions results in the formation of dextran aggregates. Consequently, aggregation leads to an increase of the void volume for water as reflected by an increase of the apparent diffusion coefficient. At high T70 concentrations, for example, at 30% and 34.4% (w/w), the soft gel network is transformed into a rigid gel network via potassium ions. This transition is indicated by a sharp decrease of the apparent diffusion coefficient, which levels out with a final Dap value of 8 X lo-" cmZ/s at high T70 dextran concentrations. TI0 dextran in the range of 10-30% shows a much smaller effect of 1.5 M potassium ions on the diffusion coefficient of bulk water than T70 dextran (Fig. SB),

Influence of Potassium Ion Concentration on the Diffusion Coefficients of Water and Polymer Chains in Sols and Gels
In Fig. 5 the influence of varying potassium chloride concentrations on the apparent diffusion coefficients of water and polymer chains is demonstrated for dextran T7O (34.4%)at 25OC. The transition from the sol to the gel state is characterized by a so-called half-transition point at 0.6 Mpotassium chloride concentration (marked by an arrow in Fig. 5A). Dextrans T70 (30%) and T70 (34.4%) exhibit identical apparent diffusion coefficients of bulk water in the sol state as well as in the gel state and at their half-transition point (data for T70 (30%) are not shown). The concentration of potassium chloride, which is necessary for gel formation, must be in the molar range as during gel formation nearly half of the amount of potassium ions is trapped in the polymer (see also Discussion). At very high potassium chloride concentrations (then the gel becomes very rigid), the apparent diffusion coefficient slightly increases again (Fig. 5A), indicating that the polymer chains aggregate and the void volume increases (formation of larger pores).

Water Diffusion in Dextran Gels Formed by K+-Binding

701

radius of gyration (9) is a characteristic parameter of the polymer chain in the solution state (26, 27). The apparent diffusion coefficient of the polymer chain (DPJ is related according to the spherical random coil to the value of (9) model for flexible polymers and is inversely proportional to the square root of (S)(26,28).Furthermore, the square root of (S2) is empirically known to be proportional to M in a wide range of the molecular weights M of the polymer. The empirical values of E are in the range from 0.5-0.6 (26). Consequently, Dpol is related to M- as follows:

io

ZO

i0 10 i0 Dextran Concentration [w/w%I

i0

FIG. 6. Influence of t h e dextran concentration on t h e apparent diffusion coefficient at 25C. (A) and (B) refer to T70 and T10, respectively, exhibiting D ,, of bulk water (0, 0). (C) exhibits D ,, o f in T70 (lower curves) and T10 (upper curves). trapped water (0, I) (D) exhibits D ,, of polymer chains (A, A) in T70 (lower curves) and in T10 (upper curves). Open symbols indicate t h e absence and full symbols the presence o f 1.5 M KCI. The standard deviations are 2 1 % for bulk water and 5 5 % for trapped water and polymer

In the case of TI0 and T70 dextran solutions, the ratio D,, (TlO)/D,, (T70) of the apparent diffusion coefficients of both dextrans is observed according to Eq. [ Z ] in the range of 3.2-2.6, assuming 10,000 Dalton for the molecular weight of T I 0 and 70,000 Dalton for that of T70. The apparent diffusion coefficient of polymer chains in TI0 dextran solution is three to four times larger than that observed in T70 dextran solution (Fig. 6D). The diffusion coefficients of polymer chains measured in the solution state of TI0 and T70 dextrans provide firm evidence for the above mentioned model. Hence, one must conclude that the dextrans TI0 and T70 form spherical random coils in aqueous solution. Microstructure of Cross-Links in Gels Formed by DextranANaterlPotassium Chloride Systems The sol-gel transition in dextran/water/potassiuni chloride systems can be considered as a structural transition from small-size clusters (random coils of polymers) to infinite-sized clusters (infinite network). Dextran does not contain ionic groups. Hence, hydrogen-bonding between hydroxyl groups and/or between hydroxyl groups and water molecules leads to interactions that are important for the excluded volume effect of the random coil. The results obtained by the diffusion measurements provide firm evidence, that potassium ions stabilize the network due to an increased cross-linking between the chains. Potassium-bonding leads to additional interactions and to further cross-linking. For a better understanding of the microstructure of cross-links in gels, the crystal structure of a synthetic hydrated dextran was taken as an analogue for the natural dextran. The electron and x-ray diffraction patterns of the synthetic dextran indicate that three chains with six glucose residues and four water molecules constitute one asymmetric unit (29). The water molecules are trapped between the three chains. Each water molecule is hydrogen-bonded to six oxygen atoms belonging to five hydroxyl groups and one glucose ring. The bond length between the oxygen atom of the water molecule and the oxygen atoms in the polymer chains varies be.>tween 0.233 and 0.309 nm, the average bond length is 0.268 nm
(29).

chains. providing clear evidence that the transformation from a soft to a rigid gel does not occur under these conditions. In solution of dextrans the diffusion coefficient of trapped water (Fig. 6C) and polymer chains (Fig. 6D) also decrease with increasing polymer concentration, i.e., with increasing density of the polymer matrix. The D,, values of the TI0 dextran solution are three to four times larger than those of the T70 dextran solution. The length of the chains and the moleculer weight of the polymers are of influence on the diffusional motion of the chains and hence also on that of trapped water. The apparent diffusion coefficients of trapped water and polymer chains decrease to a further extent in the presence of potassium ions in both dextrans and in both states (gel and sol). Potassium ions affect the apparent diffusion coefficient of bulk water and polymer chains, but in an inverse manner.
DISCUSSION Structure of the Dextran Polymer in Aqueous Solution

The physicochemical properties of an aqueous solution of a polymer are mainly influenced by the average molecular weight of the polymer and by the particular macromolecular structure. The high degree of freedom of flexible polymers allows the formation of spherical random coils in the solution state, due to the free internal rotation of each polymer unit (25). The mean square

It is well known that potassium ions form a clathrate complex with crownether (18-crown-6 complex). In this complex 1 2 lone-pair electrons of 6 oxygen atoms of the crownether coordinate to K located in the center of the

702

Watanabe et al.

crown. The bond length between K t and 0 is in the range of 0.277 to 0.283 nm (30).The sum of the Van der Waalsradii of K' and 0 equals 0.278 nm. Therefore the average 0-0 bond length determined by x-ray structure analysis for crystals of hydrated dextran seems to be shorter (0.01 bond. However, in the gel system nm) for the K+-0 enriched with water, the space for the uptake of the potassium ions should be larger. Then the K'-0 bond length could be also larger because of the larger amount of hydration water that separates the polymer chains. Hence, one can assume that the potassium ion replaces the water in its position. As a result a more stable crosslink between polymer chains is formed and thus the micro-crystalline clusters in the gel structure are stabilized. In Fig. 7 the microstructure of the gel, due to the binding of potassium ions, is depicted on the basis of the above mentioned crystal data. In each cross-link six oxygen atoms of three dextran chains form a pocket into which the potassium cation perfectly fits, resulting in "potassium-bonding" analogous to hydrogen-bonding. A zip of cross-links tightened by "potassium-bonding'' forms the crystalline junction zones. This mechanism is the reason for the need of at least 1 M potassium ion concentration for gel formation. The molar concentration of the glucose unit in a 30% dextran solution is about 2 M. Further aggregation of dextran chains leads to the formation of a rigid gel with larger void volume. Generally, the potassium ion carries eight hydration water molecules (31, 32). Therefore each K ' in the dextran pocket should carry at least two hydration water molecules. These water molecules are then trapped, although slowly exchangeable with bulk water. Sodium ions do not induce the formation of dextran gels (15, 16).The Van der Waals-radius of Na' is smaller than that of K+. The bond length between Nat and oxygen atoms in crownether complexes is in the range of 0.235-0.243 nm for the benzo(l5-crown-5) complex (33). The distance between hydration water and Na+ in the crownether complex is in the range of 0.227-0.231 nm (33). The sum of Van der Waals-radii of 0 and Na+ is 0.242 nm. If we compare these bond lengths with the 0-0 bond length (average 0.268 nm) mentioned above, it becomes obvious that the size of the pocket formed by the dextran chains is too large to bind and to trap Nat ions.

Principally, ionic strength and osmotic pressure are important factors in the gelation process (17). On the basis of our experimental results, the effect of ionic strength on dextran gelation seems to be negligible, as a 1.5 M concentration of other cations, such as Li', Na ', Rb', and Cs', does not result in gel formation.

Diffusion Components in DextranNVaterlPotassium Chloride Systems In both states, i.e., in sols and in gels, one observes three diffusion components that can be attributed to bulk water, to water trapped in cross-links and junction zones, and to dextran chains. In experiments employing long diffusion times and strong field gradients, only the diffusion coefficients of trapped water and polymer chains are determined. The appearance of the two resonances, i.e., for water and for CH,-protons, provides firm evidence lor the existence of a small amount of water that is trapped in the polymer matrix. This water experiences slow motional exchange with the bulk water and slow chemical exchange with the hydroxyl groups of the glucose units. Another strong argument for the correct assignment of this diffusion component to slowly diffusible trapped water is based on the close similarity of the diffusion coefficients of polymer chains and trapped water. That means both diffusion components diffuse as a couple and therefore indicate identical diffusibilities. As already mentioned, dextran has the intrinsic: feature to bind additional hydration water (this water is then trapped) in between the chains. According to the crystal structure data (33), the molar ratio of glucosdhydration water is 6 to 4. Therefore the amount of thc trapped water, or example, in the "70 (30%) solution, is at maximum 3% of the total amount of water. If one replaces all hydration water molecules positioned in between the ' binds two water dextran chains by K' and if each K molecules, the amount of trapped water will be about 6%. Dextran polymers in the gel state will additionally bind hydration water (which is not trapped) outside the junction zones. This hydration water, however, will exhibit a much faster motional exchange with the bulk water. Indeed, the signal amplitude assigned to the trapped water does not reflect the exact amount of the trapped water. However, it allows a rough estimation of its fraction (1-2%) of the total water in both sol and gel states. It seems to be likely that in the sol state random coils interact and temporarily form microstructures. This means that water molecules in the sol are also temporarily trapped. These results further indicate that in the sol state the number of microstructure units must be very small. Although in the gel state binding of K' removes hydration water, which now fills up the void volume, a fraction of hydration water remains still trapped. The apparent diffusion coefficients of polymer chains and of trapped water in the absence of K ' are larger than in the presence of K+ (Figs. 6C and 6D). In the presence of K ', the chains are cross-linked and aggregated. As a result both chains and trapped water becomes immobilized.

FIG. 7. Three-dimensional structure of cross-links in a dextran are positioned at their binding sites gel. The potassium ions (0) without constraints. Bond distances and coordinates were taken from the x-ray structure analysis of a synthetic dextran (29). The six oxygen atoms (0)coordinated to each potassium ion are provided from glucose units of three polymer chains.

Water Diffusion in Dextrun Gels Formed by K+-Binding

703

Restricted Diffusion of Bulk Water in DextranNVaterlPotassium Chloride Systems

-.= I
0.9

In a very dilute aqueous solution of dextran (solution state), the diffusion of water is almost that of free water (19). The apparent diffusion coefficient of bulk water in dextran sol (i.e., higher dextran concentration) is by a half to one order of magnitude smaller than in pure water. In this case, however, the diffusion of bulk water is slightly restricted. ' to dilute T70 dextran solutions (sol The addition of K state) results in an increase of the diffusibility of bulk water (Fig. 6A). The presence of K t in the system causes increasing aggregation of polymers (soft gel), as mentioned above, and the break of hydrogen bonds in water (34).In rigid, slightly opaque gels, such as T70 (30%) and the apparT70 (34.4%) containing more than 1.5 M K+, ent diffusion coefficient of bulk water becomes much smaller than in the sol state. If the aggregation increases to a larger extent in the gel state (rigid gel), the apparent diffusion coefficient of bulk water becomes larger in the gel state than in the sol state. This phenomenon has its origin in the increase of the void volume. The decrease of Dal, of bulk water in the rigid gel is indicative for the occurrence of restricted diffusion. As bulk water is included in pores of the network formed by the cross-links, the degree of cross-linking in dextran gels and in viscous sols has been characterized by investigating the restricted diffusion of bulk water. In Fig. 8 plots of D,,/Do versus the diffusion time A are shown for T70 (3O0h)dextran (for sols and gels due to the changes in K' concentration). The parameter D,, is defined as the actual diffusion coefficient of bulk water in the pores (see Eq. [3] and its explanation). The least squares fits to the experimental data (solid lines in Fig. 8) based on a suitable model for the diffusion of water in a three-dimensional network provide information about the structure of the dextran network. The dextran forms permeable barriers for the diffusion of water. Between these barriers the water molecules exhibit free diffusion. Therefore, a likely model for the restricted diffusion is based on a permeable barrier model. The analysis of parameters related to the restricted diffusion is based on Eq. [3] published by von Meerwall and Ferguson (35):

-6,
n

.
P

0 . 6 .

1 - 0
I

sn

560

1000

Diffusion Time A [ms] FIG. 8. Analysis of t h e restricted diffusion of bulk water in T70 (30%)at 25C in the presence o f 0.5 M (U), 1.OM (A), 1.5M (0) and 2.0 M (V)KCI. The curves are fits to the experimental data employing Eq. [3].The standard deviation is less than 55%.

These parameters are the radius of pores a (available space for the diffusion), the interbarrier diffusion coefficient (equal to D,,) and the permeability p of barriers (which form the wall of the pores). The ratio of signal intensities A(A) observed at the diffusion time A and A(0)

observed without field gradient is represented by R(A) = A(A)/A(O). The angle between barrier and the direction of the principal axis of the field gradient is defined as a. The coefficients A and B correspond to A = cos'a/(l + 1/P) and B = 1/(1+ P), respectively. The reduced permeability is expressed by P = aplD,, where p [c:m/s] is the permeability of the barrier. Furthermore, if = yg6a and d =- ( W T ) cos a . From the fit curves shown in Fig. 8 Do, a and p are determined and then plotted versus the potassium ion concentration (Fig. 9). The interbarrier diffusion coefficient D,, (Fig. 9A) and the interbarrier distance a (Fig. 9B) increase with increasing potassium ion concentration. The interbarrier distance exhibits a sudden jump between 0.5 M K+ (sol state) and 1.0 M K + (soft go1 state) and remains practically constant at cation concentrations being larger than 1.0 M , indicating that the structure of the network is already formed in the soft gel. With increasing potassium ion concentration, the network becomes more and more rigid due to further aggregation (rigid gel). In parallel, polymer chains are extracted from the bulk water during dextran aggregation (phase separation) resulting in a lower viscosity of the bulk water in the pores. These properties are clearly reflected by D,, and p (Figs. 9A and 9C). This means that the interbarrier diffusion coefficient of bulk water increases and the permeability of the barrier decreased, if the polymer matrix changes from the sol state to soft gel state and further to the rigid gel state with increasing potassium ion concentrations. In the pores of rigid gel Do = (1.2 2 0.03) X 10 cm"/s. It is larger than that in pores of sols with Do = (1.12 0.03) X cm2/s. The D,, values in the rigid gel and sol are much smaller than the self-diffusion coefficient of free water (2.3 X lo-' cm2/s). A comparison with free water provides further firm evidence for the obstruction effect. Polymer chains are still present in the water of the pores. Notice that the apparent diffusion coefficient of bulk water is smaller in gel than in sol, but the interbarrier diffusion coefficient (i.e., the actual diffusion coefficient) of bulk water is smaller in sol than in gel. Hence, the bulk water of a rigid gel diffuses inside the pores more rapidly than the solvent water in the sol.

704

Watanabe et al.

125

I
0.8 -

%
, u

.
Y)

0.6-

b
3, 0.4n

0.2-

0.5

1.5

FIG. 9. Influence of potassium chloride in T70 (30%) at 25C on the interbarrier diffusion coefficient Do (A), the pore size (B), and the permeability of the barrier p (C). Open symbols indicate the sol state, full symbols the gel state. The standard deviations are indicated in by bars.

Potassium Chloride Concentration [MI

CONCLUSION

The structure of dextran in sols and gels is schematically depicted in Fig. 10. In solutions (Fig. 10A) the dextran chains are homogeneously distributed as random coils. The sol state (Fig. 10C) is characterized by small-sized water compartments (as indicated by short interbarrier distances). These compartments are separated by barriers with larger permeability in sol than in gel. This observation provides evidence that homogeneously distributed random coils in solution and in sol behave as pseudobarriers against the free diffusion of water. In the gel state (Figs. 10B and IOD), potassium ions induce the formation of cross-links and the aggregation of chains. As a result stacked junction zones and pores are formed in the network. An increasing K+ concentration leads to the formation of further stacked junctions and to a thickening of the already formed stacked junctions by aggregation. Then the pore size or the void volume in the gel state becomes larger than the average void volume in the sol state.

The gelation of K-carrageenan has been discussed on the basis of two competitive models, the coil-domain transition model and the single helix model (12-14). In both models the gelation is accompanied by a structural transition of the polymer from random coil to helix domains, which aggregate to form junction zones induced by Kt. On the basis of our results we conclude that in dextrans three polymer chains join to form a gel network stabilized by potassium-bonding and hydrogen-bonding. Binding of potassium ions induces the gelation via formation of stacked junction zones and in addition leads to the binding of water molecules. The water picked up and trapped by the stacked junctions shows very little exchange with the bulk water. The energy of potassium-bonding is too small to stabilize the gel structure at high temperatures. Therefore, sol-to-gel and gel-to-sol transitions are reversible and can be induced by temperature changes.
REFERENCES

..._
KCI

KCI

FIG. 10. Sol-to-gel transition and structure of dextran. (A) Low dextran concentration (solution state with equally distributed random coils). (B) High dextran concentration (sol state with a temporary network of the polymer). (C) Low dextran concentration in the presence of KCI (soft gel state with condensed random coils). (D) High dextran concentration in the presence of KCI (rigid gel state with highly aggregated polymers). Addition of potassium ions (0) to the solution and sol (A and B) induces the transition to soft and rigid gels (C and D).

1. J. M. V. Blanshard, J. R. Mitter, Polysaccharides in Foods, Butterworths, London, 1979. 2. M. Takeuchi, N. Inoue, T. W. Strickland, M. Kubota, M. Wada, R. Shimizu, S. Hoshi, H. Kozutsumi, S. Takasaki, A. Kobata, Relationship between sugar chain structure and biological activity of recombinant human erythroprotein produced in Chinese hamster ovary cells. Proc. iVatl. Acod. Sci. U S A 86, 7819-7828 (1989). 3. S. I. Chavin, S. M. Weidner, Blood clotting factor IX. Loss of activity after cleavage of sialic acid residues. J. Biol. Chern. 259, 3387-3390 (1984). 4. R. E. Feeney, Biological Antifreeze. Am. Sci. 62, 712-719 (1974). 5. V. Ginsburg, P. W. Robbins, Biology of Carbohydrates, Vol. 1, John Wiley & Sons, New York, 1981. 6 . V. Ginsburg, P. W. Robbins, Biology of Carbohydrates, Vol. 2, John Wiley & Sons, New York, 1984. 7. G. Ashwell, J. Harford, Carbohydrate-specific receptors of the liver. Annu. Rev. Biochern. 51, 531-554 (1982). 8. P. B. Ahrens, H. Ankel, The role of asparaginelinked carbohydrate in natural killer cell-mediated cytolysis. J. Biol. Chern. 262, 7575-7579 (1987). 9. A. Kobata, in Development and Recognition of the Transformed Cell (M. I. Greene, T. Hamaoka, Eds.), p. 385, Plenum Publishing Company, New York, 1987. 10. K. Almdal, J. Dyre, S. Hvidt, 0. Kramer, Towards a phenom-

Wafer Diffusion in Dextran Gels Formed by K+-Binding

eno-logical definition of the term gel.Polymer Gel Network
1, 5-17 (1993).

705 24. T. Watanabe, N. Murase, M. Staemmler, K. Gersonde, Mul-

11. G. T. Grant, E. R. Morris, D. A. Rees, P. J. C. Smith, D. Thom,

tiexponential proton relaxation processes of compartmentalized water in gels. Magn. Reson. Med. 27, 118-134
(1992). 25. G. S. Y. Yeh, Order in amorphous polystyrenes as revealed

Biological interactions between polysaccharides and divalent cations. Egg-box model. FEBS Lett. 32, 195-198 (1973). 12. E. R. Morris, D. A. Rees, G. Robinson, Cation-specific aggregation of carrageenan helices: domain model of polymer gel structure. J . Mol. Biol. 138, 349-362 (1980). 13. D. M.Goodall, I. T. Norton, Polysaccharide conformations and kinetics. Acc. Chem. Res. 20, 59-65 (1987). 14. 0. Smidsroed, H.Grasdalen, Some physical properties of carrageenan in solution and gel state. Carbohydr. Polym. 2,
270-272 (1982). 15. Y. Kanzawa, A. Koreeda, A. Harada, T. Harada, Electron

microscopy of the gel-forming ability of polysaccharide food additives. Agric. Riol. Chem. 53, 979-986 (1989). 16. K. Kajiwara, N. Murase, K. Gonda, Freeze-thawing behavior of saturated saline aqueous solutions in the presence of Dextran. Jpn. I. Freezing Drying 33, 27-31 (1987). 17. T. Tanaka, L. 0. Hocker, G. B. Benedek, Spectrum of light scattered from a viscoelestic gel. J. Chem. Phys. 59, 51515159 (1973). 18. B. A. Westrin, A. Axelsson, G. Zacchi, Diffusion measurement in gels. J. Controlled Release 30, 189-199 (1994). 19. T.Watanabe, A.Ohtsuka, in Proc., International Workshop

by electron diffraction and diffraction microscopy. J. Macrornol. Sci. Phys. 6, 451-463 (1972). 26. P. J.Flory, Principles of Polymer Chemistry, Cornell University Press, Ithaca, 1953. 27. M. Doi, S.F. Edwards, The Theory of Polymer Dyiiamics, Oxford Science Publishers, Oxford, 1986. 28. H. Yamakawa, Modern Theory of Polymer Solution, Harper & Row, New York, 1971. 29. C.Guizard, H. Chanzy, A. Sarko, The molecular and crystal structure of dextrans: a combined electron and x-ray diffraction study. 11. A low temperature hydrated polyniorph. J. Mol. Biol. 183, 397-408 (1985). 30. P. Seller, M. Dobler, J. D. Dunitz, Potassium thiocyanate complex of 1 4,7,10,13,16-hexaoxacyclooctadecane. Acta Crysfallogr. B30, 2744-2745 (1974). 31. N. Ohtomo, K. Arakawa, Neutron diffraction study of aqueous ionic solutions. I. Aqueous solutions of lithium chloride and cesium chloride. Bull. Chem. SOC.Jpn. 52, 2755-2789
(1979). 32. N. Ohtomo, K. Arakawa, Neutron diffraction study of aque-

on MRI and MRS on Molecular Environment of Intra- and Extra-cellular Sodium, Okazaki, 1993,p. 77. 20. 1. Karger, H. Pfeifer, W. Heink, Principles and application of self-diffusion measurements by nuclear magnetic resonance. Adv. Mogn. Reson. 12, 1-89 (1988). 21. 1. E. Tanner, Use of the stimulated echo in NMR diffusion studies. J. Cheni. Phys. 52, 2523-2526 (1970). 22. 1. M. Harvey, M. C. R. Symons, R. J. Naftalin, Proton magnetic resonance study of the hydration of glucose. Nature
261, 435-436 (Z976). 23. H. J.C.Berendsen, in Water a Comprehensive Treatise, (F. Franks, Ed.), Vol. 5. pp. 193-330, Plenum Press, London, 1975.

ous ionic solutions. IT. Aqueous solutions of sodium chloride and potassium chloride. Bull. Chem. Soc. [pn. 53,
1789-1794 (1980). 33. M. A. Bush, M. R. Truter, Crystal structures of complexes between alkali metal salts and cyclic polyethers, 111. Aquo(2, 3-beiizo-I, 4, 7, 10, 13-pentaoxacyclopentadec-2-ene) sodium i0dide.J. Chem. Soc. Perkin Trans. 2, 341-344 (1972). 34. K. A. Valiev, M. I. Emelyanov, Spontaneous diffusion of

water molecules in aqueous solutions of electrolytes. I. Metal chlorides. Zh. Strukt. Khim. 5, 674-680 (19Ci4). 35. E. von Meerwall, R. D. Ferguson, Interpreting pulsed-gradient spin-echo diffusion experiments with permeable membranes. J. Chem. Phys. 74,6956-6959 (1981).