Influence of Chain Length and Polymer Concentration on the Gelation

Sep 10, 2005 - Structural Characterization and Solution Properties of a Galacturonate Polysaccharide Derived from Aloe vera Capable of in Situ Gelatio...
0 downloads 0 Views 131KB Size
Download PDF
Biomacromolecules 2005, 6, 2954-2960

2954

Influence of Chain Length and Polymer Concentration on the Gelation of (Amidated) Low-Methoxyl Pectin Induced by Calcium Franc¸ ois Capel, Taco Nicolai, and Dominique Durand* Polyme` res, Colloı¨des, Interfaces, UMR CNRS, Universite´ du Maine, 72085 Le Mans, Cedex 9, France

Patrick Boulenguer and Virginie Langendorff Degussa Texturant Systems France SAS Research Center, Baupte 50500, France Received March 10, 2005; Revised Manuscript Received August 1, 2005

The gelation of low-methoxyl pectin (LMP) induced by addition of Ca2+ was studied by measuring the storage modulus as a function of temperature during cooling. Samples with different molar masses were prepared by mechanical degradation. The effect of the molar mass and the pectin concentration on the gelation properties was investigated. The effect of partial amidation was studied by comparing LMP and partially amidated LMP with the same molar mass and degree of methylation. The results are compared to those from a model developed for Ca2+-induced pectin gelation, and good agreement is found except at low concentrations and low molar masses where the gels are weaker than predicted. At low concentrations intrachain bonding weakens the gel, while the presence of small pectin chains weakens the gel because it neutralizes binding sites on larger chains. Introduction Pectin is a polysaccharide consisting essentially of linear chains of R-D-galacturonic residues with a small fraction of rhamnose and small side chains formed by other sugars.1-4 Pectin is usually characterized by its degree of esterification (DE) of the carboxyl groups. The DE of natural highmethoxy pectin can be reduced by replacing the methyl groups with hydroxyl groups. Alternatively, the methyl groups can be replaced by amide groups, which modifies the viscoelastic properties of pectin gels and is useful for certain applications. At pH > 4.5 the galacturonic residues are fully charged. If the degree of methylation of the residues is low, sequences of these residues associate by forming a complex with certain divalent cations such as Ca2+. The association occurs below a critical temperature that increases with increasing Ca2+ concentration [Ca2+] and can lead to gelation. Lowering the temperature below a critical value (Tc) gives an immediate increase of the storage modulus followed by aging which continues for at least 24 h. Aging results in a relatively moderate increase of the shear modulus except close to Tc. Earlier5 we reported that partial amidation of the galacturonic residues does not modify significantly the sensitivity to Ca2+ in 0.1 M salt. It was found that the temperature dependence of the storage (G′) and loss (G′′) shear moduli are universal functions of the T - Tc, where Tc is the temperature at which G′ and G′′ cross at 1 Hz. Tc increases with increasing Ca2+ concentration. However, above a certain high Ca2+ concentration a different association process occurs * Corresponding author. E-mail: [email protected].

which is athermal and leads to formation of heterogeneous gels or precipitates.6 We have proposed a model5 that can explain the universality of the gelation as a function of T - Tc for a wide range of [Ca2+]. It also explains the temperature dependence of G′ for T < Tc - 10. The aim of the present work is to investigate the influence of the molar mass and the concentration of low-methoxyl pectin (LMP) and partially amidated low-methoxyl pectin (ALMP).7 We will compare the results with the model, which will be reviewed in the next section. We are concerned here only with the immediate response to cooling. Gelation Model. The main assumption of the model is that the concentration of binding sites (n) decreases exponentially with (absolute) bond energy (E): n(E) ) A′ exp(-B′E). This assumption, in opposition to the classical approach which assigns the same binding energy for all the cross-links,8,9 needs to be justified a posteriori. It can be rationalized if the distribution of nonmethylated galacturonic residues is random and E is a linear function of the length of a sequence of unmethylated galacturonic residues. The probability that a bond with energy E is formed is 1- exp(-E/kT). It follows that the molar concentration of bonds formed at a given temperature decreases exponentially with increasing T: Nc ) A exp(-BT). Gelation occurs at a critical concentration of bonds: Nc* ) A exp(-BTg), where Tg is the gelation temperature. The variation of the N with respect to Tg is thus given by: Nc/Nc* ) exp(B(Tg - T)). If it is furthermore assumed that the increase of the bond energy with decreasing temperature is the same for all binding sites, i.e., B is independent of [Ca2+], then Nc/Nc* is a universal function of T - Tg, independent of the Ca2+ concentration.

10.1021/bm0501858 CCC: $30.25 © 2005 American Chemical Society Published on Web 09/10/2005

Biomacromolecules, Vol. 6, No. 6, 2005 2955

Gelation of LMP Induced by Calcium

Figure 1. Molar mass distributions of the LMP (a) and the ALMP (b) samples used in this study obtained by size exclusion chromatography.

The influence of the Ca2+ concentration is expressed by the parameter A. At a given temperature more bonds are formed more easily if more Ca2+ is present and thus A is larger. If the origin of the elasticity is purely entropy loss of chain sections between cross-links then the elastic modulus (Ge) is given by: Ge ) aνRT, where ν is the molar concentration of elastically active network chains and R is the gas constant. The term a is a constant that depends on the functionality and the mobility of the junctions and is between 0.5 and 1 for tetra-functional cross-links.10 The relation between Nc and ν is difficult to quantify because the network contains dangling ends and elastically inactive loops. These effects can to some extent be accounted for using mean field theory or, very close to the gel point, by the percolation model. However, the pectin chains are polydisperse both in size and in functionality, which renders a more sophisticated approach unfeasible. To obtain nevertheless a first estimate, we will ignore these defects and take into account only the effect of free chain ends. With this approximation ν ) 2Nc - C/Mn, because we assume that the junctions are tetra-functional and C/Mn is the molar concentration of chain ends. With the use of the relation between Nc and T - Tg derived above one obtains the following: Ge C ) [exp[-B(T - Tg)] - 1] aRT Mn

(1)

where Mn is the number average molar mass; we have used the fact that at the gel point ν f 0 implying that Nc* ) 0.5C/ Mn. We cannot exclude the possibility of higher functionality of the cross-links that might occur if the pectin-Ca2+ complex involves more than two chains. However, this would only modify the prefactor a. The model thus predicts that MnGe/(CRT) is a universal function of T - Tg. If the enthalpy contribution is significant, then eq 1 can still be applied if Ge is proportional to the number of strained cross-links. This would lead to a larger prefactor a than if the entropy of the elastically active network chains dominates. The number of bonds per chain at the gel point is independent of the pectin concentration, implying that Tg is independent of C. This means that Ge/(CRT) is a universal

function of T - Tg because a and B do not depend on C. On the other hand, Tg should depend on the molar mass which can be understood as follows. If each chain contains many cross-links the effect of chain ends on Ge is negligible (exp[ - B(T - Tg)] . 1), and Ge should not depend on Mn at a given temperature for T , Tg. It follows that exp(BTg) ∝ Mn because the constants a and B do not depend on Mn so that: Tg ∝ B-1 ln(Mn)

(2)

Experimental Section Materials. The two pectin samples used for this study were obtained from Degussa Texturant Systems. One sample (LMP) had a degree of methylation of the galacturonic residues of 27%, while the other (ALMP) was methylated to a degree of 28.7%, and in addition a further 18% of the residues were amidated. ALMP and LMP are the same as those used in ref 5 where they were designated as S1 and S2, respectively. Lower molar mass samples were obtained by mechanical degradation using a vibrating mill Retsch (MM2 type) as described by Van Deventer-Schriemer and Pilnik.11 The samples were characterized by size exclusion chromatography using a TSK PW 500 column of 30 cm and a TSK PW 600 column of 60 cm in series and a refractive index detector. The elution solvent was 0.1 M NaNO3 at pH 7. Mass distributions were obtained in terms of the molar mass equivalent to pullulan standards. The intrinsic viscosity of the pectin samples was measured in 0.1 M NaCl at pH 7 using a Contraves low-shear 40 rheometer and compared to that of pullulan with the same average elution volume. The true pectin masses were derived using the universal calibration method which assumes that M[η] is the same for the sample and the standard at a given elution volume. The ratio of [η]pect to [η]pull at the same number average molar mass is found to be equal to 0.43 ( 0.07 for all pectin samples used here. Figure 1 shows the molar mass distributions of the LMP and ALMP samples. The number (Mn) and weight (Mw) average molar masses and the intrinsic viscosities of the samples are given in Table 1.

2956

Biomacromolecules, Vol. 6, No. 6, 2005

Capel et al.

Table 1. Number (Mn) and Weight (Mw) Average Molar Masses and Intrinsic Viscosities of the Samples LMP

ALMP

Mn (kg/mol)

Mw (kg/mol)

[η] (L/g)

Mn (kg/mol)

Mw (kg/mol)

[η] (L/g)

66 42 6

194 95 10

0.28 0.13 0.04

70 47 22

206 112 30

0.30 0.17 0.08

Pectin solutions with C ) 20 g/l were prepared by slowly adding pectin powder to Millipore deionized water and stirring overnight. The solutions were filtered through 0.45 µm pore size filters and were extensively dialyzed against deionized water. The concentration of the solutions was determined by measuring the refractive index using dn/dC ) 0.14 g/l.12 The pH of the pectin solutions was 5. The calcium content of the solutions was determined using flame spectroscopy and was 0.62 mM for ALMP and 0.33 mM LMP. In one case a solution was extensively dialyzed against EDTA, which reduced the calcium content by a factor 2. However, this reduction had no influence on the experimental results shown here. Methods. The shear modulus was measured with a stresscontrolled rheometer (AR1000, TA Instruments) with coneplane geometry (60 mm diameter, 1°). The temperature was controlled by a Peltier system, and the sample was covered with mineral oil to avoid evaporation. A constant stress of 1Pa was chosen in order to break weak links between pectin chains that are formed even in the absence of Ca2+ so that only calcium-induced gelation is probed. Solutions of 20 g/l pectin were preheated to about 80 °C and rapidly mixed with preheated CaCl2 solutions to give pectin solutions with the desired pectin and Ca2+ concentrations. All solutions contained 0.1 M NaCl in order to screen electrostatic interactions. The solutions were immediately loaded onto the rheometer with preheated cone-plane geometry. Subsequently, the temperature was decreased while monitoring the loss (G′′) and storage (G′) shear modulus.

Results Ge is the storage modulus at zero frequency, but experimentally one determines G′ at finite frequencies. The frequency dependence of G′ depends on the relaxation processes of the system. The data shown here were determined at 1 Hz in order to have rapid acquisition, but we have verified that the results are almost the same at 0.1 Hz and 0.01 Hz. In ref 5, we showed that the temperature dependence of G′ determined at different Ca2+ concentrations ([Ca2+]) is a universal function of T - Tc. This universality justifies the main assumptions of the model described in the previous section. The range of [Ca2+] that can be explored is limited by the formation of an athermal gel at large values as mentioned in the Introduction. For simplicity, we have defined the reference temperature Tc as the temperature where G′ and G′′ cross, but any definition of Tc gives the same master curves. In fact, since the variation of Tc is relatively small in absolute terms, master curves can be formed both for G′ and G′/(RT). It is therefore not possible to distinguish between a purely entropic origin of the elasticity and a temperature-independent enthalpic contribution. Concentration Dependence. We have measured the temperature dependence of G′ during cooling for different concentrations between 1 and 20 g/L of low-methoxyl pectin (LMP) and partially amidated low-methoxyl pectin (ALMP) for a range of Ca2+ concentrations. Master curves could be obtained at each pectin concentration, by plotting the data obtained at different Ca2+ concentrations as a function of T - Tc. These master curves cover a broader range of temperatures than could be obtained at a single Ca2+ concentration. The model predicts that master curves obtained at different concentrations superimpose if G′ is normalized by C. Figure 2 shows that this is born out by the experimental results for C > 2.5 g/L. The results obtained at 1 g/L are rather noisy because the gels become weak. Within the noise universal behavior is still observed as a function of T - Tc, but the absolute values of Ge/(RTC) are somewhat smaller. With decreasing temperature the storage modulus increases

Figure 2. Dependence of the normalized storage modulus at 1 Hz on T - Tc for undegraded LMP (a) and ALMP (b) at different pectin concentrations indicated in the figure. For each pectin concentration, measurements were done over a range of Ca2+ concentrations.

Gelation of LMP Induced by Calcium

Biomacromolecules, Vol. 6, No. 6, 2005 2957

Figure 3. Dependence of Tc on the pectin concentration at different Ca2+ concentrations indicated in the figure for undegraded LMP (a) and ALMP (b). The solid lines are guides to the eye.

Figure 4. Dependence of the normalized storage modulus at 1 Hz on T - Tc for the two weakly degraded samples: LMP 42 kg/mol (a) and ALMP 47 kg/mol (b) at C ) 10 g/L. For each system, measurements were made over a range of Ca2+ concentrations.

rapidly just below the critical temperature followed by a weak increase for T < Tc - 20. Figure 3 shows the pectin concentration dependence of Tc at different Ca2+ concentrations. For C > 2.5 g/L Tc is independent of the pectin concentration as predicted by the model, but Tc is significantly lower at C ) 1 g/L. The gels formed for C < 1 g/L are too weak to characterize by shear measurement, but light scattering measurements showed that Tc decreases with decreasing pectin concentration.6 The plateau value of Tc at higher pectin concentrations increases with increasing Ca2+ concentration. Molar Mass Dependence. We have measured the temperature dependence of G′ during cooling of degraded LMP and ALMP at 10 g/L for a range of Ca2+ concentrations. The samples with the lowest molar mass did not gel for T > 5 °C. Instead, precipitation was observed if large amounts of Ca2+ were added. Master curves were again obtained for the weakly degraded samples when G′/(RT) is plotted as a function of T - Tc, see Figure 4. The model predicts that G′/(RT) at a given value of T - Tc should be inversely

proportional to Mn, because B depends only on the configuration of the pectin chains and not their molar mass. Instead, we find that G′/(RT) is smaller for the lower molar mass samples. Nevertheless, all master curves can be superimposed by vertical and horizontal shifts, see Figure 5. The implication is that B is the same within the experimental error for ALMP and LMP at both molar masses. The solid line in Figure 5 represents the model predictions with B ) 0.054 K considering only data at T - Tc < -10, because close to Tc the model is not valid, see below. With the use of the values of Mn obtained by chromatography, the remaining model parameters are a and ∆T ) Tc - Tg and are different for each sample. They have been obtained from the fit and the shift factors and are summarized in Table 2. In Figure 6 the dependence of Tc on the Ca2+ concentration is compared for the two different molar masses. Tc increases monotonically with increasing Ca2+ concentration for each sample, but the increase is somewhat weaker for amidated pectin. The model predicts that Tg and thus Tc decrease with decreasing molar mass and that the difference should be

2958

Biomacromolecules, Vol. 6, No. 6, 2005

Capel et al.

Figure 5. Superposition of the master curves obtained for LMP and ALMP shown in Figure 4. The horizontal shift factors are ∆T and the vertical shift factors are Mn/(aC). The solid line represents the model prediction using B ) 0.054K-1. The values of a, Mn, and ∆T are given in Table 2. Table 2. Values of Prefactor a and Shift Factor ∆T ) Tc - Tg Used in the Model for the Samples LMP and ALMP of Molar Masses Given in the Table LMP

Mn (kg/mol) a ∆T

66 0.134 5

ALMP 42 0.028 1.5

70 0.370 10

47 0.082 7

independent of the Ca2+ concentration. A constant difference of about 20 °C for LMP and 15 °C for ALMP is indeed observed. However, these differences are larger than those predicted by eq 2. Discussion The main assumption of the model is that pectin contains a large amount of binding sites with a density that decreases exponentially with increasing bond energy, which can be rationalized if the distribution of nonmethylated galacturonic residues is random, as was discussed in more detail in ref 5. It is this assumption together with the proportionality between G′ and the number of strained cross-links that explains the exponential dependence of G′ on (T - Tc) for T , Tc. Deviation is expected at low temperatures if the number of free binding sites becomes small. Therefore, the model fails if the molar mass between cross-links becomes small. For T , Tc the average molar mass of chain segments between cross-links is approximately C/ν and can be estimated from G′/(RTC) assuming that the prefactor a is unity. We find that even for the most densely cross-linked samples the molar mass between cross-links is on average still higher than 10 kg/mol. This means that the system is probably still far from saturating all potential binding sites even for the most densely cross-linked sample. Deviation from the power law dependence is also expected at high temperatures when finite size effects become significant. The effect of free chain-ends is accounted for in

the model, but deviation from the model predictions occurs for T - Tc > 10 for a number of reasons: (1) G′ was determined at a finite frequency which is not equal to the elastic modulus close to the gel point. In addition, G′ and G′′ do not necessarily cross at the gel point. (2) An increasing fraction of cross-links is situated in dangling ends and elastically inactive loops when T f Tg. (3) The pectin chains are polydisperse both in molar mass and functionality. The influence of polydispersity is more important when the number of cross-links per chain is small, i.e., close to Tc. A difference between Tg and Tc is therefore expected depending on the composition of the samples. The parameters intrinsic to the model are A and B which determine how the number of cross-links vary with [Ca2+] and temperature, respectively: Nc ) A exp(-BT). B is independent of [Ca2+] and determines how the fraction of bonds that are formed varies with temperatures: d ln(Nc) ) -B dT. A determines how the bond energy depends on the Ca2+ concentration. The higher is the calcium concentration the lower is the bond energy, as might be expected when one considers the loss of entropy by binding Ca2+ in the junctions. It appears that the important parameter is the activity of the calcium ions and not the number of calcium ions per nonmethylated galacturonic residue. A is related to Tg as A ≈ 0.5CMn-1 exp(BTg), because approximately two cross-links per chain are formed at Tg. Both A and B depend on the degree of methylation of the pectin chains and the relation between the bond energy and the length of a sequence of unmethylated galacturonic residues. The degree of methylation and the size distribution of LMP and ALMP are almost the same so that any difference between the two samples is caused by the influence of amidation on the bond energy. ALMP shows a weaker dependence of Tg on [Ca2+], but the difference with LMP is not large. For fully charged pectin, i.e., pH > 4.5, the effect of amidation is small, in contrast with the strong effect observed at lower pH.6 The influence of varying the pectin concentration is in agreement with the model for C > 2.5 g/L. The reason for the deviation at lower concentrations is intrachain bonding that leads to the formation of elastically inactive loops. The probability of intrachain bonding becomes increasingly important below the overlap concentration that can be estimated as [η]-1 and is about 3.5 g/L both for LMP and ALMP. The influence of the molar mass is not in agreement with the model. G′ as a function of T - Tg is smaller for the lower molar mass samples in contradiction with eq 1. In addition, the observed decrease of Tg with decreasing molar mass is larger than expected using eq 2, which gives about 8 °C for both systems. Both effects can be attributed to finite chain size effects. In the model we have assumed that all pectin chains are sufficiently large to contain many binding sites. However, if pectin chains are small they do not contain sufficient binding sites to form a gel, which follows from the observation that the LMP and ALMP samples with the lowest molar mass did not gel for T > 5 °C, although a precipitate is formed if a high concentration of Ca2+ (30 mM) is added.

Gelation of LMP Induced by Calcium

Biomacromolecules, Vol. 6, No. 6, 2005 2959

Figure 6. Comparison of the dependence of Tc on the Ca2+ concentration for two molar masses of LMP (a) and ALMP (b) at C ) 10 g/L. The solid lines are guides to the eye.

Figure 7. Temperature dependence of the storage modulus at 1 Hz for undegraded ALMP (70 kg/mol) at C ) 5 g/L (circle) and a mixture of undegraded ALMP (70 kg/mol) at C ) 5 g/L with degraded ALMP (22 kg/mol) at C ) 5 g/L (square). For the two systems, the Ca2+ concentration was 4 mM.

The presence of small chains not only reduces the concentration of pectin that contributes to the gel but may even neutralize binding sites on the larger chains.13 This phenomenon is illustrated in Figure 7 where we compare the temperature dependence of G′ during cooling for ALMP (70 kg/mol) at 5 g/L with a mixture of ALMP at 5 g/L and strongly degraded ALMP (22 kg/mol) at 5 g/L. If the smaller sample was simply inactive we would expect that the mixture would yield the same results as the undegraded sample at the same concentration. In fact, we find that the gelation properties of the mixture are even weaker, i.e., lower Tc and smaller G′. The implication is that small chains neutralize binding sites on larger chains so that Nc needs to be larger in order to obtain the same modulus. Interestingly, Tc and the temperature dependence of G′ of weakly degraded ALMP (47 kg/mol) are close to that of the mixture, see Figures 4

and 6. The reason is that the average molar mass and even the size distribution of this sample are close to that of the mixture. Gelation properties of pectin can thus be weakened by addition of small chains. Inversely, gelation properties can be strengthened by removing small chains from the undegraded sample. As mentioned above, a is between 0.5 and 1 if the theory of rubber elasticity can be applied, i.e., if only entropy loss of the elastically active network chains is considered. If enthalpic contributions caused by stress on the junction zones themselves contributes, a is larger. However, we find smaller values of a, which can be explained by the combined effects of network defects and the presence of small chains that do not contribute to the elasticity that are contained in this parameter. The fraction of small chains increases when the samples are degraded, which explains the lower values found for a for the samples with lower molar mass. We cannot exclude that enthalpy contributes to the elasticity, but its contribution cannot be very large in view of the small values of a. Conclusion The temperature dependence of the elastic modulus of Ca2+-induced pectin gels can be interpreted in terms of a model that assumes a random distribution of binding sites on the pectin chains with a frequency that decreases exponentially with increasing binding energy. The model describes the Ca2+ concentration and the temperature dependence quite well as long as the pectin concentration is above the chain overlap concentration and the molar mass is high. The elastic modulus is proportional to the polymer concentration in the semidilute regime. At lower concentrations the gelation temperature and the storage modulus decrease because intrachain bonding becomes important. Reducing the molar mass of pectin by less than a factor of 2 already leads to an important decrease of the gelation

2960

Biomacromolecules, Vol. 6, No. 6, 2005

temperature and the elastic modulus. More strongly degraded material does not gel. The presence of small pectin chains neutralizes binding sites on larger chains and thus reduces the gelation properties. References and Notes (1) Voragen, A. G. J.; Pilnik, W.; Thibault, J.; Axelos, M. A. V.; Renard, M. G. C. Pectins. In Food Polysaccharides and Their Applications; Stephen, A. M., Ed.; Marcel Dekker: New York, 1995; pp 287338. (2) Thakur, B. R.; Singh, R. K.; Handa, A. K. Crit. ReV. Food Sci. Nutr. 1997, 37, 47. (3) Ralet, M. C.; Bonnin, E.; Thibault, J. F. Pectins. In Biopolymers; Steinbu¨chel, A., Ed.; Polysaccharides, Vol. 6; WILEY-VCH, 2002; pp 345-380. (4) Axelos, M. A. V.; Thibault, J. F. The chemistry of low-methoxyl pectin gelation. In The Chemistry and Technology of Pectins; Walter,

Capel et al.

(5) (6) (7) (8) (9) (10) (11) (12) (13)

R. H., Ed.; Academic Press, Inc.: San Diego, New York, Boston, London, Sydney, Tokyo, Toronto, 1991; pp 109-118. Lootens, D.; Capel, F.; Nicolai, T.; Durand, D.; Boulenguer, P.; Langendorff, V. Food Hydrocolloids 2003, 17, 237. Capel, F.; Nicolai, T.; Durand, D.; Boulenguer, P.; Langendorff, V. Food Hydrocolloids, in press. Racape, E.; Thibault, J. F. Biopolymers 1989, 28, 1435. Clark, A. H.; Evans, K. T.; Farrer, D. B. Int. J. Biol. Macromol. 1994, 15, 125. Clark, A. H.; Farrer, D. B. Food Hydrocolloids 1996, 10, 31. Mark, J. E.; Erman, B. Rubberlike Elasticity a Molecular Primer; John Wiley and Sons: New York, 1988. Van Deventer-Schriemer, W. H.; Pilnik, W. Acta Aliment. 1987, 16, 143. Hourdet, D.; Muller, G. Carbohydr. Polym. 1991, 16, 113. Powell, D. A.; Morris, E. R.; Gidley, M. J.; Rees, D. A. J. Mol. Biol. 1982, 155, 517.

BM0501858