Orientation of fluorinated cholesterol in lipid bilayers analyzed by 19F tensor calculation and solid-state NMR

Article abstract of DOI:10.1021/ja077580l

6-F-cholesterol was reported to exhibit biological and interfacial properties similar to unmodified cholesterol. We have also found that 6-F-cholesterol mimicked the cholesterol activity observed in the systems of amphotericin B and lipid rafts. However, to use 6-F-cholesterol as a molecular probe to explore molecular recognition in membranes, it is indispensable to have detailed knowledge of the dynamic and orientation properties of the molecule in membrane environments. In this paper, we present the molecular orientation of 6-F-cholesterol (30 mol %) in dimyristoylphosphatidylcholine (DMPC) bilayers revealed by combined use of ¹⁹F chemical shift anisotropy (CSA), ²H NMR, and C-F rotational echo double resonance (REDOR) experiments. The axis of rotation of 6-F-cholesterol was shown to be in a similar direction to that of cholesterol in DMPC bilayers, which is almost parallel to the long axis of the molecular frame. The molecular order parameter of 6-F-cholesterol was determined to be ca. 0.85, which is within the range of reported values of cholesterol. These findings suggest that the dynamic properties of 6-F-cholesterol in DMPC are quite similar to those of unmodified cholesterol; therefore, the introduction of a fluorine atom at C6 has virtually no effect on cholesterol dynamics in membranes. In addition, this study demonstrates the practical utility of theoretical calculations for determining the ¹⁹F CSA principal axes, which would be extremely difficult to obtain experimentally. The combined use of quantum calculations and solid-state ¹⁹F NMR will make it possible to apply the orientation information of ¹⁹F CSA tensors to membrane systems.

Full text of DOI:10.1021/ja077580l

Published on Web 03/15/2008
Orientation of Fluorinated Cholesterol in Lipid Bilayers
19
Analyzed by F Tensor Calculation and Solid-State NMR
Nobuaki Matsumori,*,† Yusuke Kasai, Tohru Oishi, Michio Murata, and
†
†
†
‡
Kaoru Nomura
Department of Chemistry, Graduate School of Science, Osaka UniVersity, 1-1 Machikaneyama,
Toyonaka, Osaka 560-0043, Japan, and Suntory Institute for Bioorganic Research,
1
-1-1 Wakayamadai, Shimamoto-Cho, Mishima-Gun, Osaka 618-8503, Japan
Received October 9, 2007; E-mail: matumori@ch.wani.osaka-u.ac.jp
Abstract: 6-F-cholesterol was reported to exhibit biological and interfacial properties similar to unmodified
cholesterol. We have also found that 6-F-cholesterol mimicked the cholesterol activity observed in the
systems of amphotericin B and lipid rafts. However, to use 6-F-cholesterol as a molecular probe to explore
molecular recognition in membranes, it is indispensable to have detailed knowledge of the dynamic and
orientation properties of the molecule in membrane environments. In this paper, we present the molecular
orientation of 6-F-cholesterol (30 mol %) in dimyristoylphosphatidylcholine (DMPC) bilayers revealed by
19
2
combined use of F chemical shift anisotropy (CSA), H NMR, and C-F rotational echo double resonance
REDOR) experiments. The axis of rotation of 6-F-cholesterol was shown to be in a similar direction to that
(
of cholesterol in DMPC bilayers, which is almost parallel to the long axis of the molecular frame. The
molecular order parameter of 6-F-cholesterol was determined to be ca. 0.85, which is within the range of
reported values of cholesterol. These findings suggest that the dynamic properties of 6-F-cholesterol in
DMPC are quite similar to those of unmodified cholesterol; therefore, the introduction of a fluorine atom at
C6 has virtually no effect on cholesterol dynamics in membranes. In addition, this study demonstrates the
practical utility of theoretical calculations for determining the 19F CSA principal axes, which would be
1
9
extremely difficult to obtain experimentally. The combined use of quantum calculations and solid-state
F
NMR will make it possible to apply the orientation information of 19F CSA tensors to membrane systems.
5
Introduction
activity comparable with cholesterol. Then, Kauffmann et al.
studied the air-water interfacial properties of cholesterol and
its derivatives and showed the similarity of 6-F-cholesterol to
Use of 19F NMR has been of interest for many years in
investigating biological systems, due in large part to the
experimentally attractive properties of 19F, including 100%
natural abundance, spin I ) 1/2, large magnetogyric ratio, and
low background signals in biological samples. In particular,
6
unmodified cholesterol. In our recent experiments, 6-F-
cholesterol was shown to reproduce cholesterol activity in the
4
systems of amphotericin B and lipid rafts (to be published in
due course). However, to use 6-F-cholesterol as a molecular
probe in examining intermolecular recognition in membranes,
it is indispensable to have detailed knowledge of its dynamic
and orientation properties in membrane environments.
Due to its large anisotropic effect,1 19F chemical shift
anisotropy (CSA) seems to be the best choice for obtaining the
orientational information of 6-F-cholesterol in membranes. The
1
9
solid-state F NMR has become an indispensable tool for
characterizing the orientation and dynamics of membrane-
1
associated peptides. Subsequently, the need for fluorinated
compounds suitable for 19F NMR measurements has also
increased. In the course of our studies on amphotericin B, which
is a channel-forming antibiotic, and lipid rafts, we used
fluorinated derivatives to investigate molecular recognition in
bilayer membranes2 and found 6-F-cholesterol 1 to be a
successful molecular mimic of cholesterol for investigating
intermolecular interactions involving cholesterol. 6-F-cholesterol
was originally reported as a growth factor of yeast that has the
-4
CSA tensor is defined by three principal values (δ , δ , and
11
22
δ33) and three corresponding orthogonal axes, called principal
7
axes. Thus, motion and orientation information with respect
to the CSA tensor axes can be obtained by chemical shift
13
15
measurements. In particular, C and N CSA tensors in peptide
bonds are frequently used to assign the alignments of helical
†
Osaka University.
Suntory Institute for Bioorganic Research.
‡
(
1) Ulrich, A. S. Prog. Nucl. Magn. Reson. Spectrosc. 2005, 46, 1-21.
2) Matsumori, N.; Umegawa, Y.; Oishi, T.; Murata, M. Bioorg. Med. Chem.
Lett. 2005, 15, 3565-3567.
(
(5) Harte, R. A.; Yeaman, S. J.; McElhinney, J.; Suckling, C. J.; Jackson, B.;
Suckling, K. E. Chem. Phys. Lipids 1996, 83, 45-59.
(6) Kauffman, J. M.; Westerman, P. W.; Carey, M. C. J. Lipid Res. 2000, 41,
991-1003.
(7) Duer, M. J. Introduction to Solid-State NMR Spectroscopy; Blackwell
Publishing: Oxford, UK, 2004.
(
3) Tsuchikawa, H.; Matsushita, N.; Matsumori, N.; Murata, M.; Oishi, T.
Tetrahedron Lett. 2006, 47, 6187-6191.
(4) Kasai, Y.; Matsumori, N.; Umegawa, Y.; Matsuoka, S.; Ueno, H.; Ikeuchi,
H.; Oishi, T.; Murata, M. Chemistry 2008, 14, 1178-1185.
10.1021/ja077580l CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 4757-4766
9
4757

A R T I C L E S
Matsumori et al.
8
-11
However, unlike 13
C
membrane peptides in lipid bilayers.
1
5
and N CSA in peptide bonds that are extensively studied, the
F CSA tensor of each compound must be determined prior to
1
9
utilizing it. While the principal values of the CSA tensor can
be conveniently measured with solid-state NMR techniques, the
determination of the CSA principal axis orientation in a
molecular segment requires more involved methods, such as
NMR analysis of a single crystal and comparison with its
corresponding X-ray structure. Therefore, CSA tensors of nuclei
1
3
15
other than C and N in peptide bonds are seldom used for
NMR analyses in biological systems.
Alternatively, a number of papers have recently appeared
concerning the calculation of CSA tensors by quantum chemistry
19
12,13
methods. For calculation of F CSA tensors,
Oldfield et
at. reported that the gauge including atomic orbitals (GIAO)
method combined with Hartree-Fock (HF) calculations cor-
related well with experimental values in a series of fluoroben-
zenes.13 Hence, in this study, we have adopted theoretical
1
9
calculation as a practical method of determining the F CSA
tensor orientation of 6-F-cholesterol. Simultaneously, we have
measured static solid-state 19F NMR of 6-F-cholesterol in
dimyristoylphosphatidylcholine (DMPC) bilayers and analyzed
the molecular rotation axis in the membrane according to the
1
9
calculated F CSA tensor. Besides, because the orientation
19
information derived solely from F CSA was not enough to
unambiguously specify the rotational axis and the molecular
order parameter of 6-F-cholesterol in the membrane, we also
measured the intramolecular C-F dipole coupling between 6-F
2
2
and C5 and the H NMR of [3- H,6-F]cholesterol. On the basis
of these data, we report the detailed dynamics and orientation
properties of 6-F-cholesterol in DMPC bilayers.
Figure 1. Static solid-state 19F NMR of 6-F-cholesterol in the powder state
(
A) and in DMPC bilayers at 30 mol % (B). Spectra recorded at 30 °C
19
using the DEPTH method to reduce the background F signals of the NMR
probe.14
Results and Discussion
Next, to obtain the 19F CSA principal axis system with respect
2
19
Preparation of 6-F-cholesterol (1) and [3- H,6-F]Choles-
to the molecular frame, we calculated the F CSA tensor using
1
5
terol (2). 6-F-cholesterol 1 was prepared using the procedure
reported by Harte et al.5 (Scheme 1). In brief, acetylated
cholesterol 4 was subjected to hydroboration and H2O2 oxidation
to give alcohol 5, which was further oxidized to ketone 6.
Fluorination of 6 with diethylaminoethyl sulfur trifluoride
the GIAO method. Upon calculation, we used the truncated
structure shown in Figure 2 as the input, because 3-OH and the
CD rings are considered to have minimal or virtually no effect
on 6-F atom, in terms of its electronic state and conformation.
Prior to the CSA tensor calculation, we optimized the structure
1
6
(DAST) and subsequent removal of the acetyl group produced
with Becke’s three-parameter hybrid functional and the Lee,
2
17
6
2
-F-cholesterol 1 with a moderate yield. [3- H,6-F]Cholesterol
was prepared from 6-F-cholesterol 1 by oxidation of 3-OH,
Yang, and Parr correlation functional (B3LYP) method, using
1
9
the 6-311G(d) basis set. For F CSA tensor calculations using
the GIAO method, Oldfield et al. compared the HF methods
with Møeller-Plesset (MP) and density-functional theory (DFT)
calculations and reported that the results of HF methods
and subsequent reduction with NaBD4.
1
9
F CSA of 6-F-cholesterol. We first measured the static
19
solid-state F NMR of 6-F-cholesterol in powder form to obtain
F CSA tensor values (Figure 1A). The spectrum was recorded
without spinning under H decoupling. To reduce background
F signals originating from the Teflon capacitor and other
materials in the NMR probe, we applied the DEPTH method.
1
9
13
correlated best with experimental NMR data. This implies that
1
introduction of the electron correlation does not always enhance
1
9
19
19
F
the accuracy in F CSA tensor calculations. Hence, we adopted
1
4
the HF method with a 6-311G++(2d,2p) basis for the GIAO
19
The obtained spectrum showed a typical powder pattern, from
which we extracted the 19F CSA principal values, as listed in
Table 1.
calculation of F CSA. Surprisingly enough, calculated principal
19
values of F CSA were found to be in excellent agreement
with the observed onesswithin 1 ppm (Table 1). This demon-
strated the validity of the input structure and the calculation
1
9
(
8) Davis, J. H.; Auger, M. Prog. Nucl. Magn. Reson. Spectrosc. 1999, 35,
-84.
9) Bechinger, B.; Sizun, C. Concepts Magn. Reson. A 2003, 18A, 130-145.
method. Figure 2 shows the calculated F CSA principal axis
directions. The consistency between the experimental and
calculated principal values (Table 1) suggests that the theoretical
principal axis directions are also highly reliable. In the calculated
1
(
(
10) Yamaguchi, S.; Hong, T.; Waring, A.; Lehrer, R. I.; Hong, M. Biochemistry
2
002, 41, 9852-9862.
(
(
11) Toraya, S.; Nishimura, K.; Naito, A. Biophys. J. 2004, 87, 3323-3335.
12) Wiberg, K. B.; Kurt W. Zilm, K. W. J. Org. Chem. 2001, 66, 2809-2817.
Chan, J. C. C.; Echert, H. THEOCHEM 2001, 535, 1-8.
13) Sanders, L. K.; Oldfield, E. J. Phys. Chem. A 2001, 105, 8098-8104.
14) Cory; D. G.; Ritchey, W. M. J. Magn. Reson. 1988, 80, 128-132.
(15) Ditchfield, R. Mol. Phys. 1974, 27, 789-807.
(16) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(17) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(
(
4758 J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008

Fluorinated Cholesterol in Lipid Bilayers
Scheme 1. Preparation of 6-F-cholesterol (1) and [3- H,6-F]Cholesterol (2)
A R T I C L E S
2
Table 1. Experimental and Theoretical 19F CSA Principal Values
ppm) of 6-F-cholesterol
19
(polar angle) connecting the rotation axis and the F CSA tensor
axes (Figure 3):
(
9
δ
11
δ
22
δ
33
δ
iso
2
2
2
2
2
experimental
theoretical
-40
-39.8
-123
-123.9
-167
-166.3
-110
-110
δ
|
) δ11 cos â + δ22 sin R sin â + δ33 sin â cos R
(1)
a
and
a
Data were generated using the value for experimental isotropic chemical
shift (-110 ppm). Absolute shielding tensors are presented in the Supporting
Information.
2
2
2
δ ) [δ sin â + δ (1 - sin R sin â) +
⊥
11
22
2
2
δ (1 - sin â cos R)]/2 (2)
3
3
Therefore, the anisotropic parameter, ∆δ ) δ| - δ⊥ , is given
by
2
∆
δ ) δ - δ ) [δ (3 cos â - 1) +
|
⊥
11
2
2
2
2
δ (3 sin R sin â - 1) + δ (3 sin â cos R - 1)]/2 (3)
2
2
33
When introducing the molecular order parameter (Smol) to
account for the wobbling of the rotation axis with respect to
1
,18
the bilayer normal,
the observed value of ∆δ is given by
∆δobs ) Smol × ∆δ (4)
Figure 2. Input structure for CSA tensor calculation and calculated 19F
CSA principal axis directions. The structure was optimized with B3LYP
The Smol parameter varies between 1, wherein there are no
molecular oscillations with respect to the rotation axis, and 0,
wherein all orientations are allowed, as in a liquid phase. In
the following analysis, we assume that the local order parameter,
Sloc, is 1, i.e., the internal mobility within the four rings of the
sterol is negligible. In effect, Marsan et al. reported that the
internal mobility does not need to be included in the calculation
of cholesterol. Considering the similarity in the molecular
structure of cholesterol and 6-F-cholesterol, it is reasonable to
assume the internal mobility is also negligible in 6-F-cholesterol.
Figure 3 shows a contour plot of ∆δ values when varying R
and â angles at a Smol of 1. For calculation of the contour plot,
the experimental values listed in Table 1 were used for δ11,
δ22, and δ33 in eq 3. The red lines represent the combinations
6
-311G(d), and the CSA tensor was calculated with the HF-GIAO
method, using the 6-311G++(2d,2p) basis set. The optimized structure is
shown.
principal axis system, the δ11 axis is almost parallel to the C5-
C6 bond direction and the δ22 axis has a direction similar to
the C6-F bond.
1
9
19F NMR of 6-F-cholesterol in DMPC Bilayers. With the
1
9
F CSA principal values and axes available, we measured the
1
9
static F NMR of 30 mol % 6-F-cholesterol in DMPC bilayers
Figure 1B). The spectrum shows a typical axially symmetric
(
pattern with the two tensor values of -120 and -90 ppm,
corresponding to δ⊥ and δ|, respectively. This means that the
molecule undergoes a fast axial rotation in the membrane, as
in the case of cholesterol. The δ| and δ⊥ values are given as
functions of the polar coordinates R (azimuthal angle) and â
(18) Aussenac, F.; Tavares, M.; Dufourc, E. J. Biochemistry 2003, 42, 1383-
1
390.
(
19) Marsan, M. P.; Muller, I.; Ramos, C.; Rodriguez, F.; Dufourc, E. J.;
Czaplicki, J.; Milon, A. Biophys. J. 1999, 76, 351-359.
J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008 4759

A R T I C L E S
Matsumori et al.
Figure 3. Simulated ∆δ values (ppm) of 19F in 6-F-cholesterol as function of azimuthal angle, R, and polar angle, â, that relate 19F CSA tensor orientation
with molecular rotation axis. Contour plot drawn for Smol ) 1. Red lines represent an experimental value of 30 ppm.
2
solid powder [3- H]cholesterol (Figure S3 in the Supporting
Information). Smol is the molecular order parameter, as described
2
above, and θ is the angle formed by the C3- H bond and the
2
axis of motion. Given the angles (l, m, n) of the C3- H bond
1
9
with respect to the F principal axis system (δ11, δ22, δ33) of
2
1
6
-F-cholesterol (Figure 5), θ is defined by
cos θ )
cos l cos â + cos m sin R sin â + cos n cos R sin â (6)
where R and â are the polar coordinates of the rotation axis in
1
9
the F principal axis frame (Figure 5). In this approach, we
added 3-OH to the structure of Figure 2 and extracted (l, m, n)
angles as 116.8°, 88.3°, and 153.1°, respectively. Using eqs 5
and 6, quadrupolar splittings ∆V can be related to R and â.
Figure 5 shows the calculated quadrupolar splittings ∆V as a
function of R and â at a Smol of 1. The red and blue lines
reproduce the experimental value of 47 kHz. Note that although
the sign of the experimental quadrupolar splitting is unknown,
2
2
Figure 4. H NMR spectrum of 30 mol % [3- H,6-F]cholesterol (2) in
DMPC. Spectrum recorded at 30 °C, using quadrupolar echo sequence. The
quadrupolar splitting ∆V is 47 kHz.
2
the term (3 cos θ - 1) in eq 5 is negative when θ is larger
than the magic angle (54.7°) and vice versa. Assuming that the
of R and â angles that reproduce the experimental ∆δ value of
30 ppm. From the figure, it is clear that while R can take all
values, â is confined between 39° and 51° or between 129°
and 141°.
+
rotation axis is nearly along the molecular long axis, as in the
2
case of cholesterol, the angle θ between the C3- H bond and
the rotation axis should be larger than the magic angle, and
therefore, ∆V should be negative. Thus, we deduced that the
red contour lines representing ∆V of -47 kHz are more
reasonable.
2
2
H NMR of [3- H,6-F]Cholesterol in DMPC Bilayers. To
gain further information on the rotation axis of 6-F-cholesterol,
we measured solid-state H NMR using the [3- H,6-F]choles-
terol 2 prepared above. Quadrupolar splittings, obtained from
H NMR spectra, allow for the determination of the tilt angle
of C- H bond with respect to the rotation axis. Figure 4 shows
the H NMR spectrum of 2 in DMPC bilayers. The observed
quadrupolar splitting, ∆V, was 47 kHz, which is significantly
reduced from the rigid-limit quadrupolar splitting. This also
suggests the rotational averaging in the membrane.
2
2
C-F Dipole Coupling of 6-F-Cholesterol in DMPC Bi-
layers. To further specify the rotational axis direction and the
molecular order parameter Smol of 6-F-cholesterol in DMPC
bilayers, we measured intramolecular C-F dipole coupling,
using the C-F rotational echo double resonance (REDOR)
2
2
2
22
method. For this purpose, we chose the dipole interaction
between C5 and 6-F, because carbon signals other than for C5
and C6 are overlapped in the REDOR spectra and because the
dipole coupling of directly bonded C6-F is too large to detect
with the REDOR method.
2
When the motion of the C3– H segment is axially symmetric,
the quadrupolar splitting ∆V of H NMR is given by
2
18,20
2
∆
V ) (3/4)AQSmol(3 cos θ - 1)/2
(5)
(
20) Shaikh, S. R.; Cherezov, V.; Caffrey, M.; Soni, S. P.; LoCascio, D.;
Stillwell, W.; Wassall, S. R. J. Am. Chem. Soc. 2006, 128, 5375-5383.
21) Dufourc. E. J.; Parish, E. J.; Chitrakorn, S.; Smith, C. P. Biochemistry 1984,
(
where (3/4)AQ is the rigid-limit quadrupolar splitting, and we
used 114 kHz for (3/4)AQ based on our experimental data of
2
3, 6062-6071.
(22) Gullion, T.; Schaefer, J. J. Magn. Reson. 1989, 81, 196-200.
4760 J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008

Fluorinated Cholesterol in Lipid Bilayers
A R T I C L E S
Figure 5. Simulated ∆V values (kHz) of H NMR in [3- H,6-F]cholesterol (2) as a function of azimuthal angle R and polar angle â that relate 19F CSA
2
2
tensor orientation with molecular rotation axis. Angles l, m, and n are 116.8°, 88.3°, and 153.1°, respectively. Contour plot drawn for Smol ) 1. Although
the red and blue contour lines agree with the experimental value (47 kHz), red lines are more probable, as deduced from the angle between the C3- H bond
2
and the rotation axis (see the text).
F, which was calculated to be 2216 Hz based on the C5-F
internuclear distance of 2.341 Å, as determined from the
optimized structure shown in Figure 2. As in eq 6, angle φ is
related to R and â as follows:
cos φ )
cos p cos â + cos q sin R sin â + cos r cos R sin â (8)
In this equation, p, q, and r represent the angles of C5-F with
1
9
respect to the F CSA principal axes (δ11, δ22, δ33) (Figure 7),
which were extracted from Figure 2 as 138.2°, 48.5°, and 94.4°,
respectively. Using eqs 7 and 8, we calculated DC-F values for
various R and â angles (Figure 7). The red and blue contour
lines in Figure 7 reproduce the experimental value of 850 Hz.
Determination of R and â Angles and Molecular Order
Parameter Smol. On the basis of the above results, we can
specify R and â angles, and the molecular order parameter Smol
of 6-F-cholesterol in DMPC bilayers. However, as shown in
Figure 8a, a simple superposition of the contour lines agreeing
with the experimental values does not give a single intersection
point. This is because the contour plots were calculated without
accounting for the molecular order parameter Smol that represents
wobbling of the rotation axis with respect to the membrane
normal. In other words, the Smol is not unity in this system. In
Figure 6. Experimental REDOR dephasing values (∆S/S0) for C5-F (solid
circle) of 30 mol % 6-F-cholesterol in DMPC bilayers. Theoretical dephasing
curves of 820 Hz, 850 Hz (bold line), and 880 Hz are depicted to determine
the C5-F dipole coupling constant.
Figure 6 plots the experimental REDOR dephasing values
(∆S/S0) between C5 and 6-F of 1 in DMPC bilayers. As shown
in Figure 6, only two dephasing times could be measured,
because the buildup of dephasing was steep due to the large
C-F dipolar coupling. Nevertheless, the dipole coupling value
was estimated to be 850 Hz by comparison with theoretical
dephasing curves, as depicted in Figure 6.
fact, reported Smol values range from 0.79 to 0.96 for unmodified
19,21,23
cholesterol in DMPC bilayers.
Hence, we varied Smol
values such that the contour lines derived from the three
experiments meet at a single point. Note that the molecular order
parameter Smol in this system must be greater than 0.82, which
corresponds to the ratio between the maximum 2H NMR
quadrupolar coupling of 57 kHz (one-half of the static quadru-
polar coupling 114 kHz) and the observed value of 47 kHz. If
Smol was less than 0.82, quadrupolar couplings before being
scaled by Smol would exceed the maximum value of 57 kHz.
By changing Smol values from 0.82 to 1.0, two solutions of R,
â, and Smol could be found: R ) 128°, â ) 42°, Smol ) 0.83
Dipolar couplings depend on internuclear distance as well
as molecular motion. For a fixed internuclear distance, such as
C5-F in 1, reduction in the coupling strength from its rigid-
lattice value is caused only by motional averaging. When the
C5-F vector is rotating around an axis, the measured dipolar
1
coupling constant is given by the following equation
2
DC-F ) DstaticSmol(3 cos φ - 1)/2
(7)
(Figure 8b) and R ) 32°, â ) 140°, Smol ) 0.85 (Figure 8c).
where φ is the angle between the C5-F vector and rotation
axis. Dstatic denotes the dipolar coupling constant for static C5-
(
23) Weisz, K.; Grobner, G.; Mayer, C.; Stohrer, J.; Kothe. G. Biochemistry
1992, 31,1100-1112.
J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008 4761

A R T I C L E S
Matsumori et al.
Figure 7. Simulated C-F dipole coupling values (Hz) between C5 and F of 6-F-cholesterol as a function of azimuthal angle R and polar angle â that relate
1
9
F CSA tensor orientation with molecular rotation axis. Angles p, q, and r are 138.2°, 48.5°, and 94.4°, respectively. Contour plot drawn for Smol ) 1. The
red and blue contour lines agree with the experimental value (850 Hz).
Figure 9 shows the orientations of the rotational axes obtained
from Figure 8. It can be postulated that the rotational axis of
change in the CSA tensor orientation and values when in the
1
9
membrane environment. It is known that F chemical shifts
are influenced by dielectric constants. In fact, the dielectric
constant in the polar region of lipids is thought to be around
6-F-cholesterol should be approximately along the long axis of
the molecular frame. As seen in Figure 9, the rotation axis
derived from Figure 8c is not directed along the molecular long
axis, while that derived from Figure 8b appears to be almost
parallel to the long axis. Therefore, we concluded that the axis
from Figure 8b (R ) 128°, â ) 42°, Smol ) 0.83) is the rotational
axis of 6-F-cholesterol in DMPC bilayers.
2
4
19
10, which might induce the F chemical shift change. The
quantum calculation of the effect of dielectric constant on the
1
9
F CSA is beyond the scope of this paper; however, according
to Abraham et al., the shift of F CS induced by the dielectric
constant of 10 is at most 2 ppm. As mentioned above, we already
25
19
19
Now, we address possible errors in the determination of
rotational orientation and the molecular order parameter. The
evaluation of the C5-F dipole coupling constant seems to cause
experimental error, because, as shown in Figure 6, the value
was elucidated from only two points of the REDOR measure-
ments. However, the experimental dephasing values fall com-
pletely between the theoretical curves of 820 and 880 Hz;
therefore, the maximum possible deviation of the C5-F dipole
coupling constant is estimated to be (30 Hz. We also evaluated
the uncertainties in the other data as follows: ∆δ ) +30 ( 2
ppm and ∆V ) -47 ( 1 kHz. We then repeated the above
analysis to include these errors. Figure 10 shows the superposi-
tion of three contour maps for Smol ) 0.89 and 0.81 (Smol values
less than 0.81 do not give the blue contours corresponding to
quadrupolar coupling). Between these Smol values, the three
contours can have intersections. For Smol ) 0.89 (Figure 10a),
the intersection occurs at R ) 140° and â ) 41° and for Smol )
took 2 ppm as the experimental uncertainty of F CSA
1
9
measurements. In addition, the isotropic F chemical shift of
6-F-cholesterol in the DMPC membrane can be calculated to
be δiso ) (2δ⊥ + δ|)/3 ) -110 ppm, where δ⊥ ) -120 ppm
and δ| ) -90 ppm (Figure 1B). This value completely coincides
with that in powder form (Table 1), which also suggests that
the change in the principal values and the CSA tensor orientation
caused by interactions of the molecule with the membrane is,
if present, within the experimental error.
Comparison with Unmodified Cholesterol in DMPC
Bilayers. There are two reports on the rotational axis of 30 mol
1
9,21
% cholesterol in DMPC bilayers at 30 °C.
Figure 11 shows
an overlay of the reported rotation axes of unmodified choles-
terol and the result of 6-F-cholesterol obtained in this study.
The orientation of 6-F-cholesterol falls between the two reported
rotation axes of unmodified cholesterol. The difference between
21
the angle obtained in this study and that by Dufourc et al. is
1
9
0
.81 at R ) 116° and â ) 45° (Figure 10b). As seen from the
5°, and that obtained by Marsan et al. is 20°. This suggests
that the rotational orientation of 6-F-cholesterol is almost parallel
to that of unmodified cholesterol.
figure, â is relatively tolerant with respect to experimental errors,
while R is more sensitive. On the basis of this consideration,
the R, â, and Smol values are rewritten with uncertainties as
follows: R ) 128° ( 12°, â ) 42° ( 3°, and Smol ) 0.85 (
Next, we compared the molecular order parameter Smol of
6
-F-cholesterol with that of unmodified cholesterol in DMPC
0
.04.
bilayers. The molecular order parameter Smol is an accurate
descriptor of the wobbling in a cone of rigid moieties. Dufourc
et al. reported Smol ) 0.79 for 30 mol % cholesterol in DMPC
Another possible source of error in the analysis might be the
19
inaccuracy of the calculated F CSA tensor orientations.
However, judging from the good coincidence between the
experimental and calculated principal values, as seen in Table
21
19
at 30 °C, while Marsan reported 0.96 for the same condition.
1
, the error in the calculated tensor orientation would be within
(24) Sengupta, D.; Meinhold, L.; Langosch, D.; Ullmann, G. M.; Smith J. C.
Proteins 2005, 58, 913-922.
the error generated during measurement. In relation to the
calculation of 19F CSA tensor, we have to refer to possible
(
25) Abraham, R. J.; Wileman, D. F. J. Chem. Soc., Perkin Trans. 2 1973, 1521-
1526.
4762 J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008

Fluorinated Cholesterol in Lipid Bilayers
A R T I C L E S
Figure 9. Two rotational axes obtained from Figure 8b,c.
Figure 8. Superpositions of contour lines derived from 19F CSA (red), H
NMR (blue), and C-F dipole coupling (green) for Smol ) 1.0 (a), 0.83 (b),
and 0.85 (c). In parts b and c, the three contour lines meet at a single
intersection point, giving two solutions of R, â, and Smol: R ) 128°, â )
2
Figure 10. Superpositions of contour maps (after including experimental
errors) derived from 19F CSA (red), 2H NMR (blue), and C-F dipole
coupling (green) for (a) Smol ) 0.89 and (b) Smol ) 0.81.
4
2°, Smol ) 0.83 (b) and R ) 32°, â ) 140°, Smol ) 0.85 (c).
Weitz et al. reported 0.85 for 40 mol % cholesterol in DMPC
at ambient temperature.23 In spite of some disagreement in the
reported molecular order parameters for cholesterol in DMPC,
all these values suggest that the amplitude of cholesterol
wobbling in the membrane is small. With respect to 6-F-
cholesterol, the molecular order parameter Smol (ca. 0.85) falls
within the range of the reported values for unmodified choles-
J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008 4763

A R T I C L E S
Matsumori et al.
Figure 11. Rotational axis orientations for 30 mol % unmodified cholesterol
and 6-F-cholesterol in DMPC bilayers at 30 °C. Data for cholesterol were
2
1
19
reported by Dufourc et al. and Marsan et al.
terol (Smol ) 0.79-0.96), thus suggesting a small wobbling in
DMPC bilayers. In fact, a Smol value of 0.85 leads to a cone
semiangle of 13° for the oscillations, when assuming a Gaussian
distribution of the wobbling angle.20,26
Note that the above analysis using the order parameter Smol
assumes an axial symmetry of the wobbling motion of 6-F-
cholesterol in the membrane. If not, a more strict analysis, Saupe
order tensor analysis, would be needed.27 In the case of
cholesterol in DMPC bilayers, the wobbling motion is reported
Figure 12. Representation of rotational axis orientation and its wobbling
for 6-F-cholesterol in DMPC bilayers. *The side chain is omitted.
19,27
to be nearly axial symmetric;
therefore, the Smol analysis
can be safely applied.19 Then, to see the similarity in wobbling
motion of cholesterol and 6-F-cholesterol, we have measured
motional averaging lies along the long axis of the molecule,
and the molecular order parameter of ca. 0.85 indicates a small
wobbling of the axis; the cone semiangle of the wobbling is
2
2
the H NMR of [3- H]cholesterol in DMPC bilayers (Figure
S3 in the Supporting Information) and compared the quadrupolar
2
2
splitting of [3- H]cholesterol (46 kHz) with that of [3- H,6-F]-
cholesterol (47 kHz, Figure 4). The result strongly suggests that
the wobbling motion of 6-F-cholesterol is virtually identical to
that of cholesterol, thus ensuring the application of the Smol
analysis to 6-F-cholesterol.
1
3°.
Here we will discuss the ordering effect of 6-F-cholesterol
on membranes. As described in the introduction, Kauffmann et
al. measured the air-water interfacial properties of POPC
membranes containing cholesterol or 6-F-cholesterol using a
Langmuir-Pockels surface balance and showed that 6-F-
cholesterol preserves the interfacial properties of cholesterol,
including molecular size, orientation, intercalation into and
The above findings show that the motional properties of 6-F-
cholesterol in DMPC bilayers are quite similar to those of
unmodified cholesterol. In other words, the introduction of a
fluorine atom at C6 of cholesterol has minimal or virtually no
effect on the dynamics of cholesterol in membranes. This is
probably due to the small atomic size of the fluorine atom next
to the hydrogen atom.
Although at present we have no direct data on the immersion
depth of 6-F-cholesterol in the DMPC membrane, previous
reports showed that C6 of unmodified cholesterol to be in the
polar region of the DMPC bilayers.28 Considering the well-
6
condensation of POPC membranes. Recently, we have also
measured the fluorescent anisotropy of sphingomyelin mem-
branes containing 6-F-cholesterol or unmodified cholesterol and
found that both sterols have virtually identical ordering effect
on the sphingomyelin membrane (to be published in due course).
These data are not for the DMPC membranes, but they strongly
support that 6-F-cholesterol has the same ordering effect on
membranes as cholesterol.
known nonmiscibility of fluorine in hydrocarbon chains, it seems
reasonable to assume that the fluorine of 6-F-cholesteol is also
in the polar region. In effect, it was reported that the fluorine
atom of 6-F-cholesterol in a membrane was strongly affected
As mentioned in the Introduction, fluorinated compounds are
expected to be versatile tools in investigating biological systems
1
9
using F NMR. This study shows the potential utility of 6-F-
cholesterol as a molecular probe for examining intermolecular
recognitions relating to cholesterol. In fact, we have found that
sphingomyelin and 6-F-cholesterol, like cholesterol, form a
liquid-ordered membrane (manuscript in preparation), which is
considered closely related to lipid raft systems in biological
2
9
by paramagnetic oxygen existing outside the membrane,
supporting the vicinity of the fluorine to the membrane surface.
Figure 12 represents the dynamics of 6-F-cholesterol in
DMPC bilayers revealed by current data, together with the
estimated immersion depth in the membrane. The axis of
30
membranes. We are currently investigating not only dynamics
and orientation of 6-F-cholesterol in sphingomyelin but also
intermolecular interactions between 6-F-cholesterol and sphin-
gomyelin.
(
26) Oldfield, E.; Meadows, M.; Rice, D.; Jacobs, R. Biochemistry 1978, 17,
2
727-2740.
(
27) Sternberg, U.; Witter, R.; Ulrich, A. S. J. Biomol. NMR 2007, 38, 23-39.
28) Leonard, A.; Escrive, C.; Laguerre, M.; Pebay-Peyroula, E.; Neri, W.; Pott,
T.; Katsaras, J.; Dufourc, E. J. Langmuir 2001, 17, 2019-2030. Villala ´ı n
J. Eur. J. Biochem. 1996, 241, 586-593.
(
(
29) Prosser, R. S.; Luchette, P. A.; Westerman, P. W.; Rozek, A.; Hancock,
R. E. W. Biophys. J. 2001, 80, 1406-1416.
(30) Simons, K.; Ikonen, E. Nature 1997, 387, 569-572.
4764 J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008

Fluorinated Cholesterol in Lipid Bilayers
A R T I C L E S
Conclusion
NMR (500 MHz, CDCl
3
) δ 3.02 (1H, dd, J ) 13, 2.5 Hz), 2.08 (1H,
m), 2.00 (1H, m), 1.70-1.95 (5H, m), 0.95-1.65 (20H, m), 0.98 (3H,
s, 3H), 0.89 (3H, d, J ) 6 Hz), 0.84 (6H, d, J ) 6.5 Hz), 0.70 (3H, s);
In this study, we succeeded in determining the molecular
orientation of 6-F-cholesterol in DMPC bilayers by the com-
+
+
HRMS (ESI-TOF) calcd for C27
H44DOFNa [(M+Na) ] 428.3408,
19
2
bined use of F CSA, H NMR, and C-F REDOR experiments.
The rotational angle of 6-F-cholesterol was shown to be close
to that of cholesterol in DMPC bilayers. In addition, the
molecular order parameter of 6-F-cholesterol is within the range
of reported values of cholesterol. These facts suggest that the
dynamic properties of 6-F-cholesterol are quite similar to those
of unmodified cholesterol. In other words, the introduction of
fluorine at C6 has virtually no effect on cholesterol dynamics
in membranes, possibly due to the small van der Waals radius
of the fluorine atom. Together with our recent finding that 6-F-
cholesterol and sphingomyelin form lipid rafts (manuscript in
preparation), this study shows the potential utility of 6-F-
cholesterol as a versatile tool to probe molecular recognition in
lipid rafts and other membrane systems.
found 428.3418.
Computational Method. All calculations were carried out using
Gaussian 03W31 (Gaussian Inc.) on a Windows XP computer. For the
geometry optimization, we utilized Becke’s three-parameter hybrid
16
functional and the Lee, Yang, and Parr correlation functional
B3LYP)17 method with a uniform 6-311G(d) basis set. Subsequent
(
shielding calculation was done with Hartree-Fock gauge including
atomic orbitals (HF-GIAO) method using a uniform 6-311++G(2d,-
2
p) basis set. The input structure shown in Figure 2 was generated in
GaussView3.0 software (Gaussian Inc.), and the output structure was
visualized in GaussView or Chem3D (CambridgeSoft).
Sample Preparation for Solid-State NMR. For membrane samples,
19
1
(
5.2 µmol of 1 (for static F NMR and REDOR measurements) or 2
2
for H NMR) and 35.4 µmol of DMPC (molar ratio 3:7) were dissolved
in CHCl , and the solvent was removed in vacuo for 12 h. The dried
3
In addition to determining 6-F-cholesterol dynamics in a
membrane, we devised a method that uses the F CSA tensor
to obtain orientation information and demonstrated the validity
of the method. In particular, theoretical calculation was shown
to be a powerful method of elucidating 19F CSA tensor
membrane film was hydrated with 1 mL of water. After a few minutes
of sonication, the suspension was freeze-thawed three times and
vortexed to make multilamellar vesicles. The vesicle solution was
19
19
lyophilized, rehydrated with 30.1 µL of deuterium water (for F NMR
and REDOR) or of deuterium-depleted water (for 2H NMR), and
freeze-thawed three times. The samples were transferred into a 5-mm
directions, which would otherwise be extremely difficult to
determine experimentally. The combined use of quantum
2
glass tube (Wilmad) for H NMR measurement or into 3.2-mm glass
19
tubes (Wilmad) for F NMR and REDOR measurements. All the glass
19
calculations and solid-state F NMR will make it possible to
utilize 19F CSA tensors of a wide variety of fluorinated
compounds for dynamic and orientation studies in membrane
environments.
tubes were sealed with epoxy glue. The 3.2-mm glass tube for REDOR
measurements was inserted into a 5-mm MAS rotor to measure
REDOR.
For measuring the static 19F NMR of the solid powder of 1, 15 mg
of 1 was packed into a 4-mm MAS rotor, sandwiched by Celite instead
of Teflon spacers, and measured without rotation.
Experimental Section
Materials. Cholesterol, pyridinium dichromate (PDC), and Celite
were purchased from Nakarai Tesque (Kyoto, Japan). 1,2-Dimyristoyl-
sn-glycero-3-phosphocholine (DMPC) was purchased from Avanti Polar
Solid-State NMR Measurements. All the solid-state NMR spectra
were recorded on 300-MHz CMX300 spectrometers (Chemagnetics,
Varian, Palo Alto, CA). Static 19F NMR spectra were acquired at 30
1
19
Lipid (Alabaster, AL), and NaBD
4
was from Cambridge Isotope
°C using a 4-mm double-resonance ( H/ F/) MAS probe without
rotation. The DEPTH pulse sequence14 was applied to reduce
19
F
Laboratory (Andover, MA). Deuterium water was purchased from
Euriso-Top, and deuterium-depleted water was from Isotec Inc.
background signals from Teflon capacitor and other 19F materials used
in the probe. The 90° pulse width was 2.4 µs, the sweep width was
200 kHz, and relaxation delay was 5 s. A total of 51200 transients
were accumulated for the membrane sample and 2048 for the solid
(Miamisburg, OH). All other chemicals were obtained from standard
venders. Thin-layer chromatography (TLC) was performed on a glass
plate precoated with silica gel (Merck Kieselgel 60 F254). Column
chromatography was performed with silica gel 60 (Merck, particle size
1
powder. Continuous wave H decoupling was applied during acquisition
with a decoupling power of 80 kHz. The 19F chemical shifts were
0.063-0.200 mm, 60-230 mesh). Solution NMR spectra were recorded
on a GSX-500 spectrometer (JEOL), and high-resolution MS was
measured on a QSTAR Elite (Applied Biosystems).
Preparation of 3-Keto-6-F-5-cholestene (8). 6-F-cholesterol (1) was
prepared from cholesterol by the procedure reported by Harte et al.5
Pyridinium dichromate (67.0 mg, 178 µmol) was added to a suspension
of 6-F-cholesterol (1) (59.3 mg, 147 µmol) and Celite (100 mg) in dry
referenced externally to CFCl
3
(0 ppm).
2
2
H spectrum was recorded at 30 °C with a 5-mm H static probe
using a quadrupolar echo sequence.32 The 90° pulse width was 2 µs,
interpulse delay was 30 µs, and repetition rate was 0.5 s. The sweep
width was 200 kHz, and the number of scans was 80 000.
1
3
19
C{ F} REDOR spectra were acquired with a 5-mm triple
1
19
13
CH
was filtered through Celite, and the filtrate was concentrated by
evaporation. The residue was purified by SiO column chromatography
hexane/ethyl acetate ) 10/1) to afford ketone 8 as a white powder
13.7 mg, 34.1 µmol, 23%): R ) 0.71 (hexane/ethyl acetate ) 3/1);
) δ 3.45 (1H, d, J ) 18.9 Hz), 2.85 (1H,
2
Cl
2
(10 mL). After being stirred at 25 °C for 16 h, the mixture
resonance MAS probe for H, F, and C, using xy-4 phase cycling
for F irradiations. MAS frequency was 7000 Hz, and rotor
19
33
1
2
temperature was kept at 30 °C. The H 90° pulse width was 3.8 µs,
and 180° pulse widths for 13C and F were 9 and 12 µs, respectively.
Contact time for cross-polarization transfer was set to 2.5 ms. The
recycle delay was 2.5 s, the sweep width was 30 kHz, and the number
19
(
(
f
1
H NMR (500 MHz, CDCl
3
1
34
m), 2,45 (1H, m), 2.31 (1H, m), 2.15 (1H, m), 2.05 (2H, m), 1.85 (2H,
m), 1.68 (1H, m), 0.96-1.62 (16H, m), 1.14 (3H, s), 0.91 (3H, d, J )
of scans was 16 000. The TPPM H decoupling was applied during
acquisition with a decoupling power of 65 kHz. The REDOR dephasing
times were 0.571 and 1.143 ms.
6
Hz), 0.85 (6H, d, J ) 6.5 Hz), 0.69 (3H, s).
Preparation of [3- H,6-F]Cholesterol 2. 3-Keto-6-F-5-cholestene
2
(
8) (13.7 mg, 34.1 µmol) was dissolved in 1.5 mL of MeOH-CHCl
3
(31) Frisch, M. J.; et al. Gaussian 03, revision B.05; Gaussian, Inc.: Pittsburgh,
PA, 2003.
(2:1), and to the solution was added NaBD
4
(5.0 mg, 120 µmol). After
(
32) Davis, J. H.; Jeffrey, K. R.; Bloom, M.; Valic, M. I.; Higgs, T. P. Chem.
being stirred for 2 h at room temperature, the solvent was removed by
evaporation, and the residue was purified by SiO column chromatog-
raphy (hexane/ethyl acetate ) 10/1) to afford 2 (4.6 mg, 11.4 µmol,
Phys. Lett. 1976, 42, 390-394.
(33) Gullion, T.; Baker, D. B.; Conradi, M. S. J. Magn. Reson. 1990, 89, 479-
2
484.
(34) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G.
1
3
4%) as a white powder: R
f
) 0.45 (hexane/ethyl acetate ) 3/1); H
J. Chem. Phys. 1995, 103, 6951-6958.
J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008 4765

A R T I C L E S
Matsumori et al.
Acknowledgment. We are grateful to Dr. Takashi Iwashita
of Suntory Institute for Bioorganic Research and Mr. Mototsugu
Doi in our department for advice on solid-state NMR measure-
ments. This work was supported by Grant-In-Aids for Young
Scientists (A) (No. 17681027), for Scientific Research (A) (No.
Supporting Information Available: Complete ref 31, tables
of Cartesian coordinates and 19F CSA tensor of the structure
shown in Figure 2, REDOR spectra of 6-F-cholesterol in DMPC
bilayers, and H NMR spectra of [3- H]cholesterol in powder
state and in DMPC. This material is available free of charge
via the Internet at http://pubs.acs.org.
2
2
1
5201048) and (S) (No. 18101010) from MEXT, Japan, and
by a grant from the CREST, Japan Science and Technology
Corporation.
JA077580L
4766 J. AM. CHEM. SOC.
9
VOL. 130, NO. 14, 2008