3328
J. Phys. Chem. B 2008, 112, 3328-3332
Use of Water Spin-Spin Relaxation Rate to Probe the Solvation of Cyclodextrins in Aqueous Solutions Edvaldo Sabadini,*,† Fernanda do Carmo Egı´dio,† Fred Yukio Fujiwara,† and Terence Cosgrove‡ Instituto de Quı´mica, UniVersidade Estadual de Campinas, Caixa Postal 6154, CEP 13084-862, Campinas, Brazil, and School of Chemistry, The UniVersity of Bristol, Cantock’s Close, Bristol BS8 1TS, United Kingdom ReceiVed: October 15, 2007; In Final Form: December 6, 2007
1
H spin-spin relaxation rate constant, R2, of water was measured by using the Carr-Purcell-MeiboomGill sequence in aqueous solutions of native cyclodextrins (R, β, and γ-CD) and chemically modified CDs in order to probe the structuring of the water surrounding these cyclic carbohydrate molecules. R2 values for water in solutions containing glucose and dextran were also measured for comparison. A two-site model for bonded and free water molecules was used to fit the results for the dependence of R2 on the solute concentrations. The order of relaxation rates for water in aqueous solution at a fixed specific hydroxyl group concentration is glucose > dextran = CDs. No significant difference was observed for R2 of water in solutions containing native CDs, which indicates that the size and nature of the cavity has a small effect on the spinspin relaxation times of water. The lower relaxation rate for water in CD solutions was attributed to the intramolecular hydrogen bonding formed between the secondary hydroxyl groups that line the rim of the CDs. For comparison, the relaxation rates for water in solutions of two chemically modified CDs were also studied.
Introduction Cyclodextrins (CDs) are cyclic oligosaccharides consisting of six (cyclomaltohexaose, R-CD), seven (cyclomaltoheptaose, β-CD), or eight (cyclomaltooctaose, γ-CD) glucose units linked by 1,4-R-glucosidic bonds. The structural consequence of this bonding mode is the formation of a shallow truncated “cone shape” molecule with n (number of glucose residues) primary hydroxyl groups lining in one rim of the cone and 2n secondary hydroxyl groups lining the other rim. A cavity arises from this structure containing rows of CH groups (C-3 and C-5) and a row of glucosidic oxygens. Therefore, the cavities are nonpolar relative to their outer surface.1,2 The cavities can act as hosts for a great variety of monomer guests3 and also for polymeric chains. In the latter case, supramolecular adducts are produced by the threading of the polymer chain into the cavity of the CD, forming interesting structures such as necklaces.4-9 In spite of the great scientific interest in such supramolecular chemistry, the understanding of the driving forces that promote the guesthost complex formation in water is not understood. In such processes the role of the interaction between the CDs and water molecules is still unclear. The solubilities of R- and γ-CD in water at 298 K are 0.13 and 0.26 (g/g H2O), respectively. However, for β-CD, the intermediary member in the series of CDs, the solubility is only 0.018.10 This intriguing behavior has been attributed to the effect caused by CDs on the water lattice structure. The local water density around the molecules of the three CDs reveals that β-CD induces stronger ordering on the surrounding water in comparison with the others.11 The solubility of the CDs in D2O at 25 °C is much lower; 0.076, 0.011, and 0.20 g/g D2O for R, β, and γ-CD, respectively, * To whom the correspondence may be addressed. E-mail: sabadini@ iqm.unicamp.br. † Universidade Estadual de Campinas. ‡ The University of Bristol.
due to the stronger hydrogen bonding formed between the heavy water molecules compared to light water.12 Nuclear magnetic resonance (NMR) is a method able to probe the solvatation of CDs molecules using the solvent relaxation time. NMR solvent relaxation is a technique widely used to study polymer adsorption because it is an efficient, noninvasive method that allows an indirect study of the polymer layer by measuring the overall dynamics of the solvent in the system.13-16 This technique was also used to probe the structure of interfacial water on organized assemblies like micelles, reverse micelles and water-in-oil microemulsions.17 Halstead et al.18 studied the relaxation time of water in carbohydrate solutions. In dilute aqueous solutions, water self-diffusion and chemical exchange of protons between the water molecules and labile protons of the solute are sufficiently rapid to ensure that the decay of the 1H NMR transverse magnetization (spin-spin relaxation, designed as T2) is a single-exponential characterized by a spinspin relaxation rate constant R2. As a consequence of this exchange, the water protons retain a memory of the various environments visited during the time scale of the relaxation (T2) and their relaxation behavior can indirectly monitor the dynamics and other properties of the solute. In this paper, we reported the study of the spin-spin water relaxations in dilute aqueous solutions of R, β, and γ-CD, chemically modified β-CD, and other carbohydrates. A range of concentrations was investigated to verify the extension of proton exchange between water and the carbohydrate molecules, in order to correlate the observed effects with the molecular arrangements of the glucose units. Theory The isotropic reorientation of the water molecules in the bulk has a characteristic correlation time, τw, which is ca. 2.4 ps.
10.1021/jp710013h CCC: $40.75 © 2008 American Chemical Society Published on Web 02/28/2008
Solvation of CDs in Aqueous Solutions
J. Phys. Chem. B, Vol. 112, No. 11, 2008 3329
Figure 1. NMR spectrum of water containing 1% of R-CD. The inset shows the magnified peaks of R-CD.
Most of the intermolecular dipolar interaction of the water is averaged by rapid, near isotropic rotational diffusion. However, in aqueous carbohydrate solutions, the reorientation of the water molecules bonded to the hydroxyl groups of the carbohydrate is anisotropic with a longer correlation time. The residual dipolar coupling is averaged on a slower time scale involving motions of the carbohydrate molecule.19 In homogeneous aqueous solution, the proton exchange contribution between the protons of water and protons belong to the carbohydrate can be described by a general two-site exchange formalism. This involves terms such as the mean lifetime of protons on the solute (kb), the proton fraction (Pb), and the intrinsic relaxation rate (R2b) of the hydroxyl protons of the carbohydrate and their chemical shift difference (δw) from the bulk water protons.20,21 In diluted solutions (Pb , 1), the transverse relaxation rate R2 is given by the Swift-Connick expression22
[
R2 ) PaR2a + Pbkb
]
R2b2 + R2bkb + (δw)2 (R2b + kb)2 + (δw)2
(1)
where R2a is the intrinsic relaxation rate of the bulk water and Pa ) 1 - Pb. If R2b is large such that (δw)2 is negligible, eq 1 reduces to
R2 ) PaR2a + Pb/(R2b-1 + kb-1)
(2)
It is safe to assume that the relaxation is exchange rate limited, so that kb-1 . R2b-1. Therefore eq 2 reduces to
R2 ) PaR2a + Pbkb
(3)
If only a fraction F of the carbohydrate hydroxyl proton is accessible for exchange, this becomes
R2 ) PaR2a + Fkb(Pb)max
(4)
where (Pb)max is a theoretical value assuming that all hydroxyl protons of the carbohydrate molecules are exchangeable. Therefore, a plot of R2 against (Pb)max should be linear with slope Fkb. Experimental Section R-CD (lot 60T005), β-CD (lot 70P229), and γ-CD (lot 80P200) were supplied by Wacker-Chemical AG. R-CD and γ-CD were purified using five recrystallization steps from
concentrated aqueous solutions by adding small amounts of ethanol (Merck). The solid phases were filtered, washed with cold water, and dried at 80 °C for 24 h. Due to the low solubility of β-CD in water and the low recovery by addition of ethanol, β-CD was purified by recrystallization from a saturated solution (7% w/w). Hydroxypropyl-β-CD (HP-β-CD, Roquette, lot 777049), dimethyl-β-CD (DMe-β-CD Roquette, lot 777049), dextran with molecular weight 77 000 g mol-1 (Sigma lot 66H0839) and β-D-glucose (Calbiochem, lot B36911) were used as received. Aqueous solutions of the carbohydrates were prepared in the range from 0,08 to 1% (w/w), and the samples were kept at 25 °C for 3 days before the measurements. All the experiments were carried out using analytical grade water from a Millipore Milli-Q Gradient filtration system. A 500-MHz NMR Varian, model Inova spectrometer was used to measure the proton spin-spin relaxation time of water, using the CarrPurcell-Meiboom-Gill (CPMG) pulse sequence 90°x[τ 180°(y - τ - echo]n.23,24 The second half of the nth echo was detected and Fourier transformed. The pulse spacing τ value was 1 ms, and a recycle delay of 5 s was used to avoid saturation. Data were averaged over 4 or 16 acquisitions, and 12-20 spectra were used for each measurement. Transverse relaxation times were determined from a nonlinear regression analysis of the exponential decay of the peak areas. A coaxial capillary containing C6D6 with a small amount of tetramethylsilane was used to provide a 2H lock signal and to calibrate chemical shift scale. The measurements were performed keeping the samples and the probe thermostated at 298 K. Results and Discussion Figure 1 presents the 1H NMR spectra of a solution containing 1% of R-CD in H2O. Due to the dilute character of the solutions studied, the 1H NMR signal is mainly due to the abundant water molecules. This is desirable, in order to avoid solute-solute interactions and then the exchange can be diffusion limited. The inset in Figure 1 shows the magnified 1H NMR spectrum of R-CD. The transverse relaxation measurements were determined from the integrated peak, I(t), centered at δ ) 4.8 ppm (eq 5) where the influence of the signal of the solute can be neglected
I(t) ) I0 exp(-tR2)
(5)
R2 reflects the loss of coherence of the spins due to spin-spin interactions. Figure 2 shows the variation of the peak area as a function of the time for pure water at 298 K. The average value obtained
3330 J. Phys. Chem. B, Vol. 112, No. 11, 2008 for R2 is 0.39 s-1. In the same figure a plot is shown for the water peak in a solution containing 1% (w/w) of glucose at the same temperature. The decay is faster in the glucose solution, yielding a value for R2 ) 1.70 s-1. The single-exponential decay of the 1H NMR of the water in the solution containing glucose reflects the rapid water self-diffusion and chemical exchange of the protons between water molecules and labile protons on the solute (mainly the OH groups). Therefore, the protons retain the memory of the various environments visited on the NMR time scale (R2-1), allowing the indirect determination of the dynamics of the solute.18 The influence of the oxygen molecules dissolved in water on R2 was also investigated because O2 is a paramagnetic molecule and could affect the spin-spin relaxation. The oxygen was removed from water by bubbling nitrogen for 24 h. The value of R2 determined for this sample was essentially the same for water without degasification and therefore subsequent measurements were carried out without previous degasification of the samples. The population of the water molecules visiting the OH groups of the carbohydrate is obviously dependent on the concentration of the solute. Thus, the hydration of each molecule of the solute
Sabadini et al.
Figure 2. Variation of the peak integrals in the CPMG experiment as a function of time for the protons of pure water and of an aqueous solution containing 1% of glucose at 298 K.
can be, in principle, probed. As R-, β-, and γ-CDs (Structure 1)25 and dextran (Structure 2) are basically molecules of glucose
Figure 3. Dependence on OH molar fraction, (Pb)max, of the water proton transverse relaxation rate for different carbohydrates in aqueous solutions at 298 K.
TABLE 1: Values for the Slope (Fkb) and the Fraction of Hydroxyls That Rapidly Exchanges Protons with the Water Molecules for the Carbohydrates Studied at 298 K
arranged in cycles and arranged in a random coil chain, respectively, the comparison of R2 for the protons of water can reveal the hydration of the glucose units with different molecular structures. At a specific concentration (w/w) of the studied carbohydrates, the number of glucose units is approximately the same. However, in the glucose molecule, there are five OH groups compared with three OH groups (per glucose unit) in both CD and dextran. These numbers were considered in the calculations of (Pb)max for each carbohydrate. Figure 3 presents the results for the dependence of the spinspin water relaxation rate constant (R2) in solutions of glucose, R, β, and γ-CD, and dextran as a function of the proton fraction, (Pb)max. According to eq 4 the linear coefficient in a plot of R2 against (Pb)max is R2a. For all saccharides solutions the obtained value is 0.39 s-1. As predicted the lines are approximately linear and the slopes give Fkb. The plots present the following order
carbohydrate
total number of HO group per glucose unit
Fkb/s-1
F
glucose dextran R-CD β-CD γ-CD HP-β-CD DMe-β-CD
5 3 3 3 3 3 1
570 149 167 129 177 164 568
1 0.3 0.3 0.3 0.3 0.3 1
for the slopes: glucose > dextran = R-CD = β-CD = γ-CD. The values of Fkb were obtained by linear regression and are shown in Table 1. According to the two-site model, the different behavior can be associated with either the mean lifetime of protons on the solute (kb) or with the fraction of accessible carbohydrate hydroxyl proton (F). Both aspects can be analyzed based on the systematic study of Halstead and coauthors on water relaxation for a large number of carbohydrate molecules.18 They concluded that variations in Pb determined by fitting of experimental data and calculated from the solute concentration can be explained in terms of the unavailability of some protons
Solvation of CDs in Aqueous Solutions
Figure 4. Molecular structure of β-CD optimized using the program Gamess and visualized using the program Molden (see ref 25). Some hydrogen bonds formed between the 2n secondary HO groups are shown (dashed line) in which the distance H-O-H is ca. 1.8 Å.
for exchange or their low rate of exchange, which could be due to the hydroxyl groups being engaged in hydrogen bonds that persist on the time scale of the exchange (milliseconds). For glucose, it is expected that all protons of the hydroxyls groups are accessible to exchange with the protons of water. Therefore, the rate of relaxation is highest for this system. By consideration in this case that F ) 1, the value of kb is 570 s-1. In the case of dextran, a lower relaxation rate is observed. If the same kb obtained for the water relaxation in glucose solution is used for dextran, we can estimate that F ) 0.3, which means that only ca. 30% of the OH groups are accessible to freely exchange protons with water (Table 1). The other hydroxyl groups can be involved in intramolecular hydrogen bonding. A similar result was observed in the case of aqueous solutions of laminarian (unbranched polysaccharide) consisting of less than 20 linked glucose units; all the hydroxyl protons are accessible, because the structure is like a random coil. However, longer chains (more than 49 linked glucose units) adopt a more rigid conformation, probably a triple helix, and in this case, only 30% of the laminarian hydroxyl protons are accessible to exchange. This is explained either because the OH groups are involved in intramolecular hydrogen bonding or because there is a sheath of tightly bound water surrounding the triple helix.19 In the case of CDs, the number of OH groups per unit of glucose and the Fkb values are the same of dextran. For CDs, only the n primary hydroxyl groups along one rim of the cone are accessible for rapid proton exchange. As shown in the optimized molecular structure of β-CD in Figure 4,26 the distances between the n primary hydroxyl groups are larger than ca. 2.9 Å. This is because the OH groups are bonded to alternate carbons atoms. As a consequence, the OH groups are free to exchange proton with water molecules. On the other hand, the distance between the 2n secondary hydroxyl groups along the other rim of the CDs is ca. 1.8 Å allows the formation of intramolecular hydrogen bonds (dashed line in Figure 4). Therefore, only a third of the OH protons of the CDs are freely accessible to exchange protons with water; this is in agreement with the experimental values obtained (F = 0.3) for all native CDs (see Table 1). In spite of the intriguing difference in the solubility of the CDs in water, no significant difference between the Fkb for the CDs was observed in the range of concentration studied (lower than 1%, w/w). The study was extended to higher concentrations for R and γ-CD, up to 6% (w/w), in order to
J. Phys. Chem. B, Vol. 112, No. 11, 2008 3331
Figure 5. Dependence on OH molar fraction, (Pb)max, of the water proton transverse relaxation rate in higher concentrations of aqueous solutions of R and γ-CD (up to 6% w/w) at 298 K.
identify possible differences between both CDs. This range of concentration was not extended for β-CD due to its solubility limit in water. The results, which are shown in Figure 5, reveal only a very slight difference for water relaxation rates for both CDs. However, a deviation from linearity is observed when (Pb)max is higher than 0.01. To investigate the effects of the intramolecular hydrogen bonding formed between the secondary hydroxyl groups of the CDs on R2 of water, two chemically modified β-CDs, hydrox-
ypropyl-β-CD, HP-β-CD (Structure 3)25 and dimethyl-β-CD, DMe-β-CD (Structure 4),25 were also studied. These chemically
modified β-CDs are interesting molecules to investigate the effect of the substitution of OH groups on the exchange rate between the hydroxyl groups and the protons of water. The substitution of the hydroxyl groups of the β-CD disrupts the hydrogen-bonding network formed between the 2n secondary hydroxyl groups along the rim of the native β-CD. As a result, the remained hydroxyl groups interact much more strongly with water which increases the solubility of the modified CD in comparison with the native one. Figure 6 shows a comparative
3332 J. Phys. Chem. B, Vol. 112, No. 11, 2008
Sabadini et al. arrangements of the carbohydrates molecules. The two-site exchange formalism was found to be appropriate to describe R2 as a function of the OH fractions of the carbohydrates. The following order for the product Fkb was observed: glucose > dextran = R-CD = β-CD = γ-CD, which indicates a very small influence of the CD cavity size on the relaxation rate. The studies involving chemically modified β-CD revealed that some of the hydroxyl groups are still not accessible to exchange with the water molecules for hydroxypropyl-β-CD. However, in the case of dimethyl-β-CD, similar rate as that for glucose was obtained, which means that the totality of the hydroxyl groups exchange protons freely.
Figure 6. Dependence on OH molar fraction, (Pb)max, of the water proton transverse relaxation rate for native and two chemically modified β-CD in aqueous solutions at 298 K.
study of the relaxation rate as a function of the (Pb)max for both native and modified β-CD. The spin-spin relaxation rates for the protons of water in solutions containing the three β-CDs are clearly different. In the case of HP-β-CD, the slope is practically parallel in relation to the slope for β-CD, which indicates that Fkb for both CDs is approximately the same (Table 1). HP-β-CD is derived from β-CD by addition of hydroxypropyl to some of the hydroxyl groups of the carbohydrate. The degree of substitution, according to the supplier, is between 5.25-6.65. This means that the sample of HP-β-CD should be composed of a distribution of isomers in which the number of hydroxypropyl (randomly distributed) per molecules of HP-βCD is 5.25-6.65. Therefore, the average number of OH group per chemically modified glucose unit is 3. Two OH groups are bonded in the carbons atoms of the glucose ring, and the other is bonded to the propyl group. We infer that the lower value for Fkb (lower than that expected) should be associated with only partial disruption of the hydrogen bonding of the secondary OH groups of HP-β-CD. However, the slope for the relaxation of water in solutions of DMe-β-CD, which has only one hydroxyl group per unit of glucose, is very high. The value is comparable to that obtained for water in glucose solutions, which means that F ) 1. Conclusions The spin-spin relaxation rates for water in dilute aqueous solutions of the three native CDs (R-, β-, and γ-CD), glucose, and dextran were studied to investigate the accessibility of the hydroxyls groups for proton exchange with water. The arrangement of the glucose units can favor the formation of intramolecular hydrogen-bonding of the OH groups. Therefore, the mean lifetime of the protons on the solute (kb) and especially the effective proton fraction are highly sensitive to the molecular
Acknowledgment. The authors are very grateful to WackerChemical AG for the samples of CDs, to Dr. Leonardo F. Fraceto for the samples of chemically modified CDs, and to Dr. Nelson Morgon for the picture of the optimized molecular structure of β-CD. E.S. would like to thank CNPq-Brazil for funding of this work. F.C.E. would like to thank CNPq-Brazil for the PhD Fellowship. References and Notes (1) Szejtli, J. In ComprehensiVe Supramolecular Chemistry; Atwood, J. L., Davies, J. E. D., Mac Nicol, D. D., Vogtle, F., Eds.; Pergamon: Exeter, 1996; Vol. 3; p 12. (2) Wenz, G. Angew. Chem., Int. Ed. Engl. 1994, 33, 803. (3) Connors, K. A. Chem. ReV. 1997, 97, 1325. (4) Harada, A.; Li, J.; Kamachi, M. Nature 1993, 364, 516. (5) Harada, A. Coord. Chem. ReV. 1996, 148, 115. (6) Shigekawa, H.; Miyake, K.; Sumaoka, J.; Harada, A.; Komiyama, M. J. Am. Chem. Soc. 2000, 122, 5411. (7) Rusa, C. C.; Bullions, T. A.; Fox, J.; Porbeni, F. E.; Wang, X.; Tonelli, A. E. Langmuir 2002, 18, 10016. (8) Sabadini, E.; Cosgrove, T. Langmuir 2003, 19, 9680. (9) Wenz, G.; Han, B. H.; Mu¨ller, A. Chem. ReV. 2006, 106, 782. (10) Jozwiakowski, M. J.; Connors, K. A. Carbohydr. Res. 1985, 143, 51. (11) Naidoo, K. J.; Chen, J. Y.; Jansson, J. L. M.; Wildmalm Maliniak, A. J. Phys. Chem. B 2004, 108, 4236. (12) Sabadini, E.; Cosgrove, T.; Egı´dio. Carbohydr. Res. 2006, 341, 270. (13) Nelson, A.; Jack, K. S.; Cosgrove, T.; Kozak, D. Langmuir 2002, 18, 2750. (14) Mears, S. J.; Cosgrove, T.; Thompson, L.; Howell, I. Langmuir 1998, 14, 997. (15) Cosgrove, T.; Griffiths, P. C.; Lloyd, P. M. Langmuir 1995, 11, 1457. (16) van der Beek, G. P.; Stuart, M. A. C.; Cosgrove, T. Langmuir 1991, 7, 327. (17) El Seoud, O. A. J. Mol. Liq. 1997, 72, 85. (18) Fabri, D.; Williams, M. A. K.; Halstead, T. K. Carbohydr. Res. 2005, 340, 889. (19) Hills, B. P.; Cano, C.; Belton, P. S. Macromolecules 1991, 24, 2944. (20) Carver, J. P.; Richards, R. E. J. Mag. Res. 1969, 6, 89. (21) Hills, B. P.; Takacs, S. F.; Belton, P. S. Mol. Phys. 1989, 67, 903. (22) Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307. (23) Carr, H. Y.; Purcell, E. M. Phys. ReV. 1954, 94, 630. (24) Meiboom, S.; Gill, D. ReV. Sci. Instrum. 1958, 29, 688. (25) The structures of native and chemically modified CDs were adapted from Wenz, G.; Han, B. H.; Mu¨ller, A. Chem. ReV. 2006, 106, 782. (26) The geometry of β-CD was optimised by using the MP2/6-31g(d) of the freeware program Gamess from http://www.msg.ameslab.gov/ GAMESS/GAMESS.html, and the molecular structure was visualized by using the freeware program Molden, from http://www.cmbi.ru.nl/molden/ molden.html.
