Dynamic and 2D NMR Studies on Hydrogen-Bonding Aggregates of

The pulsed field gradient spin−echo (PGSE) 1H and 2H NMR were used, respectively, at 600 ... The R value larger than that in hydrogen-bonding 1-octa...
1 downloads 0 Views 140KB Size
Download PDF
J. Phys. Chem. B 2006, 110, 15205-15211

15205

Dynamic and 2D NMR Studies on Hydrogen-Bonding Aggregates of Cholesterol in Low-Polarity Organic Solvents Cristiano Giordani, Chihiro Wakai, Emiko Okamura, Nobuyuki Matubayasi, and Masaru Nakahara* Institute for Chemical Research, Kyoto UniVersity, Uji, Kyoto 611-0011, Japan ReceiVed: April 28, 2006; In Final Form: June 4, 2006

Self-diffusion coefficients (D) are measured for normal (nondeuterated) and deuterated cholesterol-d6 (C26 and C27 methyl groups deuterated) in 1-octanol, chloroform, and cyclohexane at concentrations of 1-700 mM by varying the impurity water concentration (>2 mM) and temperature (30-50 °C). The pulsed field gradient spin-echo (PGSE) 1H and 2H NMR were used, respectively, at 600 and 92 MHz. At 30 °C, the hydrodynamic radius (R) obtained at 20 mM from the D value and solvent viscosity is 5.09, 7.07, and 6.17 Å, respectively, in 1-octanol, chloroform, and cyclohexane when the impurity water is negligible. The R value in 1-octanol is the smallest and comparable with the average length of the molecular axes for the cholesterol molecule. In 1-octanol, R is invariant against the concentration variation, whereas in chloroform, R is larger and increases almost linearly with cholesterol concentration. At the highest concentration, 700 mM, the R in chloroform is 13.5 and 16.7 Å, respectively, when the impurity water is at negligible and saturated concentrations. The R value larger than that in hydrogen-bonding 1-octanol indicates that cholesterol forms an aggregate through hydrogen bonding. The aggregate structure is confirmed by comparing NOESY spectra in chloroform and 1-octanol. The NOESY analysis reveals the presence of one extra cross peak (C4C19) in chloroform compared to 1-octanol. Because the carbon atoms related to the cross peak are close to the hydroxyl group (C3-OH), cholesterol molecules are considered to be not piled but are found to be OHcentered in the aggregate. This is supported also by larger rotational hydrodynamic radii measured on cholesterol deuterated at positions C2, C3, C4, and C6. This shows that the aggregate formation is driven by the hydrogenbonding between cholesterol molecules.

1. Introduction Cholesterol (Figure 1) is involved in the formation of dynamic microdomains referred to as lipid rafts in the membrane.1,2 Because the inside of a lipid domain is typically nonpolar and aprotic, it is of great interest to examine how cholesterol molecules are dissolved in low-polarity aprotic organic solvents that mimic the hydrophobic core of a phospholipid bilayer. Cholesterol can dissolve in such low-polarity solvents as 1-octanol, chloroform, and cyclohexane. At 20 °C, the dielectric constants (parenthesized numbers) are in the order: 1-octanol (10.3) > chloroform (4.8) > cyclohexane (2.0). The solubility of cholesterol is the highest (∼0.8 M) in the weakly polar CHCl3 and the smallest (∼0.3 M) in the nonpolar cyclohexane. Chloroform has no OH group with small, negative partial charges at chlorine atoms and interacts preferably with the hydrophobic moiety (the surface covered by hydrogens with a weakly positive partial charge) of cholesterol but abhors the OH groups to make them self-associated. One may ask whether cholesterol molecules can form an aggregate or not, depending on the solvent environments and the amount of impurity water. The diffusion measurement can provide us with a clue for getting insight into the state of dissolution in a hydrophobic environment in connection to the molecular mechanism of the raft formation in lipid bilayers surrounded by water. * To whom correspondence should be addressed. E-mail: [email protected]. Telephone: +81-774-38-3070. Fax +81-77438-3076.

Figure 1. Cholesterol molecular structure and the carbon numbering.

As shown in a previous paper,3 the self-diffusion coefficient is useful for elucidating the interactions and dynamics involved in lipid bilayers. To examine the aggregate formation of cholesterol, here we have determined the self-diffusion coefficients for cholesterol in 1-octanol, chloroform, and cyclohexane. The mobility of the cholesterol molecule can be transformed into the hydrodynamic radius in order to detect the aggregate formation. When the hydrodynamic radius is large, cholesterol molecules can be regarded as forming an aggregate; the cholesterol concentration dependence should be also observed because the average between the single-molecular and aggregate states is observed. If an aggregate is formed, some structural information can be obtained from the NOESY spectrum for cholesterol. The NOESY experiment can provide information about which protons are close together in space (through-space). The NOESY spectrum is a clue not only to the aggregate formation but also to the aggregation sites within the cholesterol molecule. Our dynamic and 2D NMR studies show that cholesterol molecules

10.1021/jp062607t CCC: $33.50 © 2006 American Chemical Society Published on Web 07/19/2006

15206 J. Phys. Chem. B, Vol. 110, No. 31, 2006

Giordani et al.

are not piled but OH-centered in the aggregate formation, driven by the hydrogen-bonding interaction. If cholesterol molecules are aggregated by the hydrogen bonding, the rotational dynamics related to the 2H atoms close to the OH group (C3) are also expected to be slower than those in a solvent where cholesterol is not aggregated. Thus the spin-lattice relaxation time T1 can give us further evidence for the aggregation among cholesterol molecules. In Section 2, the experimental part, materials and technical information about the diffusion coefficient, spin-lattice relaxation time T1, and one-dimensional (1D) and two-dimensional (2D) NMR measurements are described. How to dry solvents and solutes and the parameter values utilized are also shown. Results obtained by diffusion, 2D NMR, and T1 experiments are discussed in Section 3. Conclusions are given in Section 4. 2. Experimental Section 2.1. Materials. Deuterated cholesterol was purchased from CDN Isotopes, Canada, and used without further purification. Normal (nondeuterated) 1-octanol, chloroform, and cyclohexane were obtained from Nacalai. These solvents were dried as described below. Deuterated 1-octanol (98 at. %) and chloroform (99.96 at. %) were supplied by Isotec and Euriso-top, respectively, and were used as received. Four types of solvents were used: dry and water-saturated chloroform, dry 1-octanol, and dry cyclohexane. Chloroform solutions of cholesterol were prepared by weight at 1, 10, 20, 100, 400, and 700 mM (M, mol dm-3) in dry and water-saturated chloroform. The other solutions of cholesterol were prepared by weight at 20, 100, and 300 mM in dry 1-octanol and at 20, 100, and 250 mM in dry cyclohexane. Molecular sieves 4A (Nacalai) were used in order to decrease significantly the amount of impurity water in normal 1-octanol, chloroform, and cyclohexane. The amount of impurity water was measured by taking proton NMR spectra for the organic solvents. The water content in dry and water-saturated CHCl3 was estimated to be ∼3 and ∼70 mM, respectively, and ∼2 and 2 mM) and temperature (30-50 °C). The free induction decay signals were accumulated 2 times (700 mM), 10 times (400 mM), 20 times (100, 250, and 300 mM), 400 times (20 mM), and 1400 times (10 mM). The D values were obtained by the least-squares analysis of 20 experimental points of the echo signal attenuation. We spent 1/2 h (100-700 mM), 4 h (20 mM), and 16 h (10 mM) on a single value of D. The ratio of signal-to-noise (S/N) exceeded 10 even at 10 mM. At each temperature, measurements were repeated three times. The D values in Tables 1-4 are the averaged ones from three measurements. The experimental uncertainty of D was estimated to be half the difference between the maximum and minimum values. Analogous estimations were performed for the hydrodynamic radius. The uncertainty of temperature was (0.1 °C. An example of diffusion measurement is shown in Figure 2 for deuterated cholesterol at 100 mM in dry CHCl3 at 30 °C together with a typical stack plot of 2H NMR. As illustrated in Figure 2A, the S/N is good enough and the signal-intensity variation with δ is smooth enough for an accurate determination of D (Tables 1-4 and Figure 3). At concentrations lower than 10 mM, the S/N for 2H becomes smaller than 10. To detect cholesterol at a concentration of 1

Hydrogen-Bonding Aggregates of Cholesterol

J. Phys. Chem. B, Vol. 110, No. 31, 2006 15207

Figure 3. Hydrodynamic radius (R) against concentration (C) for cholesterol in water-saturated CHCl3 (circle), dry CHCl3 (square), dry cyclohexane (triangle down), and dry 1-octanol (triangle up) at 30 °C.

mM, we selected normal (nondeuterated) cholesterol dissolved in dry CDCl3. In this case, the protons attached to site C26 and C27 cannot be used because of the spin-spin coupling. We used the methyl protons at position C18 (0.678 ppm) that appear at the highest magnetic field without spin-spin coupling.6 This peak was preferable for its high intensity and because it is fairy well separated from the others. The free induction decay signals were accumulated 1000 times, and the proton S/N was increased to 22. The number of δ values was 20, g was 78.4 G/cm, and ∆ was set to 20 ms (in order to reduce the effect of eddy currents). We spent ∼41 h on a single value of D. Measurements were repeated two times. The deuteron spin-lattice relaxation times T1 were measured for cholesterol-d6 deuterated at carbon positions C2, C3, C4, and C6 in dry 1-octanol (100 mM) and water-saturated chloroform (700 mM), and for two deuterated cholesterols (cholesterol-d6 deuterated at carbon positions C2, C3, C4, and C6 and cholesterol-d6 deuterated at positions C26 and C27, their mixture with 1:1 molar ratio) in dry CHCl3 (100 mM). The T1 values were determined by the inversion-recovery method with the π-t-π/2 pulse sequence; the number of delay times t was 22. The free induction decays were accumulated 356 times. The resulting S/N in the recovered spectra was 50, 65, and 15 for deuterated cholesterol in water-saturated and dry chloroform and 1-octanol, respectively. The number of sampling points was 4096, and the observed frequency range was 1 kHz, so the digital resolution was 0.337 Hz. 2.3. One-Dimensional and Two-Dimensional spectra. 1H NMR spectrum of cholesterol in dry CDCl3 at 100 mM and 30 °C is shown in Figure 4 with the signal assignments for the protons at site C1, C2, C3, C4, C6, C7, C9, C12, C18, C19, C21, C26, and C27 and the water protons.9 The 1H peak assignments of cholesterol are confirmed by 2D-COSY performed for cholesterol at 100 mM in CDCl3 at 30 °C; Figure 5 shows some cross peaks that have been assigned by 2D-COSY. According to the 1H NMR spectrum of cholesterol in dry CDCl3 at 700 mM and 30 °C, we assigned the proton of the OH group attached to position C3 in the cholesterol molecule. Our cholesterol proton assignments are in good agreement with those in the literature.10, 11 The 2D-NOESY spectra of cholesterol at 100 and 700 mM in CDCl3 and at 100 mM in 1-octanol-d17 were recorded at 30

Figure 4. 1H NMR spectrum of cholesterol in dry CDCl3 at 100 mM and 30 °C. In the zoomed regions are shown the assignments with their respective carbon numbers.

Figure 5. (A) Contour plots of the 2D COSY measurement of cholesterol in dry CDCl3 at 100 mM and 30 °C. The number of scans is 48, and the 90° pulse is 14.6 µs. (B) Expanded region of the COSY spectrum where the assignment of some cross peaks is shown.

°C. Cross- and diagonal-peak intensities of cholesterol in CDCl3 and 1-octanol-d17 were recorded as a function of mixing time ranging from 100 to 900 ms. The mixing time of 600 ms was chosen for the optimization of the cross-peak intensities with minimum distortions. Thus, all experiments were conducted with the mixing time of 600 ms.

15208 J. Phys. Chem. B, Vol. 110, No. 31, 2006

Giordani et al.

TABLE 1: Hydrodynamic Radius (R) and Self-Diffusion Coefficients (D) for Cholesterol in Dry 1-Octanol at Various Concentrations (C) and Temperaturesa R/Å (D/10-6 cm2 s-1) temperature/°C cholesterol C/mM

30

35

40

45

50

20

5.09 ( 0.05 (1.03 ( 0.01) 5.05 ( 0.05 (1.04 ( 0.01) 5.03 ( 0.07a (0.74 ( 0.01)a

5.53 ( 0.05 (1.13 ( 0.01)

5.64 ( 0.05 (1.31 ( 0.01)

5.79 ( 0.06 (1.50 ( 0.02)

5.84 ( 0.06 (1.76 ( 0.02)

100 300

a At 300 mM, the solution is more viscous than at lower concentration, and the ratio between half widths at the lowest and highest concentration is a measure of the increment in the viscosity of solution. Half-width at 300 mM for the most upfield peak is 62 Hz, while it is 44 Hz at 20 mM. The viscosity at 300 mM is thus estimated to be 1.4 times larger than that at 20 mM and 100 mM. In the other conditions, this procedure is not necessary because the water peak was not found to be broadened.

TABLE 2: Hydrodynamic Radius (R) and Self-Diffusion Coefficients (D) for Cholesterol in Dry Chloroform at Various Concentrations (C) and Temperatures R/Å (D/10-6 cm2 s-1) temperature/°C cholesterol C/mM 1 10 20 100 400 700

30

35

40

45

50

6.8 ( 0.3 (9.6 ( 0.4) 6.77 ( 0.06 (9.62 ( 0.09) 7.07 ( 0.07 (9.22 ( 0.07) 7.27 ( 0.05 (8.96 ( 0.06) 10.2 ( 0.09 (6.39 ( 0.05) 13.5 ( 0.11 (4.83 ( 0.04)

7.06 ( 0.05 (9.81 ( 0.06) 9.82 ( 0.08 (7.04 ( 0.05) 12.8 ( 0.10 (5.42 ( 0.04)

6.86 ( 0.05 (10.7 ( 0.07) 9.50 ( 0.08 (7.76 ( 0.06) 12.2 ( 0.10 (6.03 ( 0.05)

6.72 ( 0.04 (11.6 ( 0.08) 8.98 ( 0.07 (8.67 ( 0.07) 11.7 ( 0.09 (6.33 ( 0.05)

6.55 ( 0.04 (12.6 ( 0.08) 8.71 ( 0.07 (9.49 ( 0.08) 11.3 ( 0.09 (7.30 ( 0.06)

3. Results and Discussion 3.1. Cholesterol Self-Diffusion. The self-diffusion coefficient (D) has been measured for deuterated cholesterol-d6 (C26 and C27 methyl groups deuterated) in dry and water-saturated chloroform, cyclohexane, and 1-octanol in order to elucidate the aggregate formation. The D values determined are summarized in Tables 1-4. The self-diffusion coefficient is affected by the solvent viscosity. The mobility can be converted into the hydrodynamic radius (R), which is more informative to determine aggregate formation through the Stokes-Einstein (SE) law12 expressed as:

R)

kBT Dfπη

(2)

where η is the solvent viscosity,13-16 kB the Boltzmann constant, T the temperature, R the hydrodynamic radius for the solute molecule, and f between 4 (slip) and 6 (stick), depending on the boundary conditions. Here, the slip value is used in view of the molecular size of cholesterol.5 At 30 °C, the R value of cholesterol at 100 mM in dry 1-octanol is 5.05 Å, as shown in Table 1. The R value in 1-octanol is the smallest among the solvents examined and comparable with the cholesterol molecular size in the fully extended molecular model.17,18 It is to be noted that R in 1-octanol is independent of the concentration over a wide range (20-300 mM), as seen in Table 1 and Figure 3. Thus, cholesterol can be regarded as being dissolved in the singlemolecular state. The OH group of 1-octanol can make hydrogen bonds with that of cholesterol. Thus, the cholesterol molecule can be strongly solvated by a number of 1-octanol molecules,

making no aggregates. The R in 1-octanol, therefore, represents the hydrodynamic radius of a single cholesterol molecule so that it sets the reference to detect the presence of aggregates in chloroform and cyclohexane. It is well-known that cholesterol concentration in the plasma membrane is in the range of 20-50%.19 The maximum concentration of cholesterol in the lipid bilayer can be estimated to be 700 mM;20 we have covered a wider range of cholesterol concentration from 1 to 700 mM when the solvent is chloroform. A large R of 13.5 Å is obtained at the highest concentration in dry chloroform (Table 2 and Figure 3). This is almost 3 times larger than that (5.05 Å) for a single molecule in 1-octanol. At 400 mM (∼30 mol % in the plasma membrane), the R is 10.2 Å, almost double the single-molecule value. The larger hydrodynamic radius thus obtained can be explained by considering an aggregate formed by cholesterol molecules in the low-polarity aprotic organic solvent. When the concentration is lowered, the R decreases, as shown in Figure 3 and Table 2. Even at the lowest concentration of 1 mM, the R is 6.8 Å, which is still larger than the single-molecule value. Because the limiting value for the single molecule is unavailable, no attempt is made to separate the observed diffusion coefficient into the singlemolecule and aggregate components. The monotonic increase of R is in favor of the aggregate formation. In cyclohexane, the slope is steeper than that in chloroform (see Figure 3). The almost linear dependence of R is found in the concentration range examined. The difference in the slope can be ascribed to the difference in solvation and aggregation between chlorinated and normal hydrocarbon solvents. In both chloroform and cyclohexane, when the temperature is elevated, the R has a weak decreasing tendency (∼10%). This is consistent with a common

Hydrogen-Bonding Aggregates of Cholesterol

J. Phys. Chem. B, Vol. 110, No. 31, 2006 15209

TABLE 3: Hydrodynamic Radius (R) and Self-Diffusion Coefficients (D) for Cholesterol in Water-Saturated Chloroform at Various Concentrations (C) and Temperatures R/Å (D/10-6 cm2 s-1) temperature/°C cholesterol C/mM

30

35

40

45

50

20

8.75 ( 0.06 (7.45 ( 0.05) 9.34 ( 0.07 (6.98 ( 0.05) 13.5 ( 0.11 (4.84 ( 0.04) 16.7 ( 0.17 (3.89 ( 0.04)

9.16 ( 0.06 (7.56 ( 0.05) 12.5 ( 0.10 (5.51 ( 0.04) 16.2 ( 0.17 (4.27 ( 0.05)

9.00 ( 0.06 (8.17 ( 0.06) 13.0 ( 0.11 (5.65 ( 0.05) 14.7 ( 0.15 (5.01 ( 0.05)

8.41 ( 0.06 (9.31 ( 0.07) 12.2 ( 0.10 (6.38 ( 0.05) 13.9 ( 0.15 (5.60 ( 0.06)

8.06 ( 0.06 (10.3 ( 0.07) 11.7 ( 0.10 (7.03 ( 0.06) 13.7 ( 0.14 (6.06 ( 0.06)

100 400 700

TABLE 4: Hydrodynamic Radius (R) and Self-Diffusion Coefficients (D) for Cholesterol in Dry Cyclohexane at Various Concentrations (C) and Temperatures R/Å (D/10-6 cm2 s-1) temperature/°C cholesterol C/mM

30

35

40

45

50

20

6.17 ( 0.06 (6.59 ( 0.06) 7.60 ( 0.05 (5.34 ( 0.04) 12.3 ( 0.09 (3.30 ( 0.02)

7.09 ( 0.05 (6.14 ( 0.04) 11.0 ( 0.08 (3.95 ( 0.03)

6.84 ( 0.05 (6.85 ( 0.05) 10.3 ( 0.07 (4.54 ( 0.03)

6.70 ( 0.05 (7.67 ( 0.05) 9.82 ( 0.07 (5.22 ( 0.04)

6.57 ( 0.05 (8.60 ( 0.06) 9.35 ( 0.07 (6.05 ( 0.04)

100 250

view that the aggregate formation is less effective at a higher temperature, being enthalpy-driven. One may ask why cholesterol molecules in chloroform and cyclohexane can be aggregated in such a wide concentration range. Cholesterol has a chemical structure that includes a flexible hydrophobic tail and a planar and rigid tetracyclic fused ring sterol skeleton with a single hydroxyl group at carbon 3 (see Figure 1). In a hydrophobic solvent like chloroform and cyclohexane, the aggregate formation is driven by the hydrogenbonding interaction among OH groups of cholesterol molecules and is not inhibited by the solvent. When cholesterol molecules form an aggregate, the aggregate surface is covered by the hydrogens with a small positive partial charge. When the solvent is chloroform, the surface can be surrounded by chloroform chlorine atoms with small negative partial charges, and the aggregates are further stabilized by solvation with the hydrogenbonding OH groups hidden. As can be seen from Figure 3, the R value at concentrations above 200 mM is in the order:

strengthened by the analysis of the two-dimensional spectra. The comparison of NOESY in CDCl3 (Figure 6) and 1-octanold17 (Figure 7) is crucial in order to investigate the presence of new cross peaks that show proximity in space. If cholesterol molecules are aggregated in chloroform and not in 1-octanol, then new cross peaks that show intermolecular correlations should be found only in the former. The NOESY spectrum in 1-octanol (Figure 7) has been used as a reference in order to investigate the presence of extra cross peaks in CHCl3 (Figure 6), where diffusion data proved the aggregate formation among cholesterol molecules (Figure 3). At first we have compared the COSY cross peaks (Figure 5)

cyclohexane (dry) > CHCl3 (water-saturated) > CHCl3 (dry) > 1-octanol (dry) The R sequence shows the role played by the solvent hydrophobicity and impurity water. The largest R value is obtained in cyclohexane, where the concentration of water is the lowest, which is ∼2 mM. It is worth pointing out that, in such an aprotic nonpolar solvent that mimics the hydrophobic core of the bilayer membrane, the R increases with the concentration faster than those in the other solvents. Water molecules whose concentration is in the same order of magnitude of cholesterol interact with the OH group of cholesterol by hydrogen bonds in watersaturated CHCl3. Increase in the impurity water concentration from ∼3 to ∼70 mM promotes the aggregate formation among cholesterol molecules bridged by the hydrogen bonding with water (as seen in Figure 3, the hydrodynamic radius is larger at any concentration in water-satured chloroform than that in dry chloroform). 3.2. NOESY. The information about the aggregate formation obtained by the self-diffusion coefficient is confirmed and

Figure 6. (A) Contour plots of the 2D NOESY measurement of cholesterol in dry CDCl3 at 100 mM and 30 °C. The mixing time is 600 ms, number of scans is 64 (corresponding to the total measuring time of ∼45 h), and the 90° pulse is 14.2 µs. (B) Expanded region of the NOESY spectrum showing a strong cross peak between C4 and C19. The cross peak is indicated by the circle and arrow in (A) and by the circle in (B).

15210 J. Phys. Chem. B, Vol. 110, No. 31, 2006

Giordani et al. TABLE 5: Spin-Lattice Relaxation Time T1 (ms), Rotational Correlation Time τ2R (ps), and Radius RRot (Å) Values for Cholesterol Deuterated-d6 at Positions C2, C3, C4, and C6 and at Positions C26 and C27 at 30 °C and Concentrations of 100 and 700 mM solvent/cholesterol concentration 1-octanol 100 mM

cholesterol atom positiona H-2R H-2β H-3R H-4R/4β H-6R H-26 and H-27

Figure 7. (A) Contour plots of the 2D NOESY measurement of cholesterol in dry 1-octanol-d17 at 100 mM and 30 °C. The mixing time is 600 ms, number of scans is 64 (corresponding to the total measuring time of ∼45 h), and the 90° pulse is 13.9 µs. (B) Expanded region of the NOESY spectrum showing no cross peak between C4 and C19.

with those in the NOESY spectrum in CHCl3 (Figure 6), and the space-proximity peaks have been identified. Then we have compared the NOESY spectrum in CHCl3 and that in 1-octanol, and the extra through-space peaks arising from the aggregate are picked up. These extra peaks correspond to intermolecular correlation. The C4-C19 cross peak found for the NOESY in CHCl3 is by far the strongest among the extra peaks. The strong cross peak is for a site pair C4-C19, where their protons are more than three bonds apart (Figure 1). The proton positions concerning the new cross peak are in the proximity of the cholesterol OH group. Position C4 is next to position C3, where the OH group is attached. Position C19 lies above the plane occupied by the sterol skeleton. The intensity of the cross peak C4-C19 is enhanced by a factor of 2 in CDCl3 at 700 mM compared to that at 100 mM, in accordance with the concentration dependence of the extent of aggregate formation. NOESY results indicate the edge-to-edge contact among cholesterol molecules; if those were piled up in the aggregate, it was expected to find many NOESY cross peaks between the sterol rings of cholesterol molecules. 3.3. Spin-Lattice Relaxation Time. T1 measurement for deuterated cholesterol is a powerful tool because it is able to provide information for each deuterated atom, as NOESY does. We have measured the spin-lattice relaxation time, T1, to confirm the aggregate formation detected by the self-diffusion coefficient and NOESY. T1 of selected C-D bonds can be affected by the aggregate that makes the rotation restricted due to the edge-to-edge contact of the rigid sterol rings. The aggregate formation is expected to affect hydrogen sites attached to the carbon atoms close to the hydroxyl group (C3-OH). Labeled sites closer to the OH group have undergone a stronger perturbation.21 The aggregate-perturbed deuterons would have

a

T1 τ2R Rrot T1 τ2R Rrot T1 τ2R Rrot T1 τ2R Rrot T1 τ2R Rrot T1 τ2R Rrot

CHCl3 100 mM 700 mM

12.3 190 3.10 14.5 162 2.97 14.5 162 2.93 12.2 190 3.10 10.6 220 3.26

66.7 35.0 4.09 64.1 36.5 4.15 75.0 31.1 3.93 66.5 35.1 4.09 61.8 37.8 4.20 233 10.0 2.70

33.7 69.3 5.13 32.2 72.6 5.21 36.7 63.7 5.00 33.5 69.7 5.14 31.8 73.4 5.23

Actually, all the hydrogen sites are deuterons.

a smaller T1 value. The T1 value in Table 5 has been converted to the rotational correlation time τ2R through:22,23

( )

1 3 2 e2qQ 2 ) π τ2R T1 2 h

(3)

Here, e2qQ/h is the quadrupole coupling constant (QCC); a value of 170 kHz is taken here for the D nucleus in an aliphatic C-D bond.24 The τ2R values obtained are summarized in Table 5. As in the case of the translational diffusion, the hydrodynamic radius Rrot related to the rotational correlation time τ2R for each labeled site is obtained according to the Stokes-EinsteinDebye (SED) law in the stick boundary condition:

Rrot )

(

)

3kBτ2R T 4π η

1/3

(4)

where η is the solvent viscosity13-16 and Rrot the rotational radius of the solute molecule. Using the SED law, we estimated the Rrot values, as shown in Table 5. In CHCl3, the comparison of τ2R between 100 and 700 mM shows a decrease of τ2R by a factor of 2; the motion of the aggregate-perturbed deuterons at positions C2, C3, C4, and C6 is less restricted at 100 mM (less aggregated) than at 700 mM. At the fixed concentration of 100 mM, the τ2R values related to position C26 and C27 are smaller by a factor of ∼3.5 than those at the other positions mentioned above. This is ascribed to the free motion of two methyl groups located at the end of the cholesterol tail. Comparison of Rrot between chloroform and 1-octanol is crucial for finding out the differences in the molecular rotational motion in the presence and absence of aggregate. The Rrot values in chloroform at 100 mM are ∼30% larger than that in 1-octanol; τ2R in 1-octanol is larger than that in CHCl3 at the same atomic sites merely due to the large difference in the solvent viscosity. A similar amount of increase has been observed also in the translational hydrodynamic radius (R) when R is compared between 1-octanol and chloroform at 100 mM and 30 °C (see Tables 1 and 2). The similarity indicates a tight binding of the constituent cholesterol molecules with their

Hydrogen-Bonding Aggregates of Cholesterol molecular configurations almost frozen within the aggregate. This is consistent with the conclusion on the aggregate formation drawn from the cholesterol diffusion and NOESY measurements. 4. Conclusions The cholesterol molecule has been studied in organic solvents that mimic the membrane bilayer hydrophobic core. Our experiments indicate that cholesterol forms aggregates in such organic solvents as chloroform and cyclohexane but it does not in 1-octanol. In the aggregate, cholesterol molecules are not piled but OH-centered. The aggregate formation between cholesterol molecules is driven by the hydrogen bonding, and it is influenced by the amount of impurity water. For the first time, the aggregate formation among cholesterol molecules was revealed by 1H and 2H NMR solution-state selfdiffusion and 1D and 2D measurements. In our future works, we expect that solution-state NMR spectroscopy for selfdiffusion coefficients and NOESY could be a valuable tool25 for obtaining precious information about dynamics, functionality, and structure of cholesterol in models of a bilayer membrane that includes a “raft system”. Acknowledgment. This work is supported by the grant-inaid for scientific research (nos. 15205004, 17550135, and 18350004) from the Japan Society for the Promotion of Science, by the grant-in-aid for scientific research on priority areas (no. 15076205), and the grant-in-aid for creative scientific research (no. 13NP0201) from the Ministry of Education, Culture, Sports, Science, and Technology. We thank Dr. Tomohiro Kimura for stimulating discussions. References and Notes (1) Simons, K.; Ikonen, E. Nature 1997, 387, 569. (2) Simons, K.; Ehehalt, R. J. Clin. InVest. 2002, 110, 597. (3) Okamura, E.; Wakai, C.; Matubayasi, N.; Sugiura, Y.; Nakahara, M. Phys. ReV. Lett. 2004, 93, 248101. (4) Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1987, 19, 1. (5) Wakai, C.; Nakahara, M. J. Chem. Phys. 1997, 106, 7512. (6) The modulation by spin-spin coupling is observed for diffusion measurements using nondeuterated cholesterol; however, in normal cholesterol, the position 18 is almost not affected by spin-spin coupling as

J. Phys. Chem. B, Vol. 110, No. 31, 2006 15211 we proved by the comparison of the diffusion coefficient in nondeuterated (position C18) and deuterated cholesterol (C26 and C27 methyl groups deuterated). Cholesterol deuterated at position C26 and C27 permits elimination of the spin-spin coupling effect because the two deuterons do not interfere with the others in the molecule and the interaction between them is symmetric. In addition, the number of deuterons gives us a strong NMR signal even at low concentration. (7) Bender, H. J.; Zeidler, M. D. Bunsen-Ges., Phys. Chem. 1971, 75, 236. (8) Yoshida, K.; Wakai, C.; Matubayasi, N.; Nakahara, M. J. Chem. Phys. 2005, 123, 164506. (9) Nakahara, M.; Wakai, C. Chem. Lett. 1992, 809. (10) Wilson, W. K.; Sumpter, R. M.; Warren, J. J.; Rogers, P. S.; Ruan, B.; Schoroepfer, G. J. J. Lipid Res. 1996, 37, 1529. (11) Jeannerat, D. Magn. Reson. Chem. 2000, 38, 415. (12) Tyrell, H. J. V.; Harris, K. R. Diffusion in Liquids; Cambridge University Press: London, 1984. (13) Viswanath, D. S.; Natarajan, G. Data Book on the Viscosity of Liquids; Hemisphere: New York, 1989. (14) Saleh, M. A.; Akhtar, S.; Begum, S.; Shamsuddin, M. A.; Begum, S. K. Phys. Chem. Liq. 2004, 42, 615. (15) Tu, C.; Wang, W.; Liu, C.; Chou, Y. J. Chem. Eng. Data 2000, 45, 450. (16) Hoyuelos, F. J.; Garcia, B. R.; Ibeas, A. S.; Leal, J. M. J. Chem. Soc., Faraday Trans. 1996, 92, 219. (17) The long molecular axis is estimated to be 16 and the short one 3.8 Å according to a molecular model (CS ChemDraw). The average radius is 4.9 Å. (18) Gupta, R. K.; Suresh, K. A. Eur. Phys. J. E 2004, 14, 35. (19) Albert, B.; Bray, D.; Lewis, J.; Raft, M.; Roberts, K.; Watson, J. D. Molecular Biology of the Cell, 3rd ed.; Garland Publishing: New York, 1994. (20) The maximum concentration of 700 mM was estimated by taking account of the cholesterol molar ratio (50%) and the molecular volume of dipalmitoylphosphatidylcholine (DPPC) in a lipid bilayer. When it is assumed that the cross section area of DPPC headgroup is ∼60 Å2 and its length 20 Å, the concentration of DPPC is 1.4 M. (21) The tail of the cholesterol molecule is freer to move than any other part of the molecule. In the presence or absence of aggregation, there will not be any difference in the rotational motion at position C26 and C27 because cholesterol molecules form aggregates that involve atomic positions close to the sterol skeleton, as shown by 2D NMR analysis. As can be seen in Figure 4, the peaks at position C3 and C6 are broader than those at position C26 and C27, where the C-H bond vectors are almost free to rotate. (22) Abragam, A. The Principles of Nuclear Magnetism; Oxford University: Oxford, 1961. (23) McConnell, J. The Theory of Nuclear Magnetic Relaxation in Liquids; Cambridge University: Cambridge, 1987. (24) Morrow, M. R.; Grant, C. W. M. Biophys. J. 2000, 79, 2024. (25) Brand, T.; Cabrita, E. J.; Berger, S. Prog. Nucl. Magn. Reson. Spectrosc. 2005, 46, 159.