The structural role of cholesterol in biological membranes

Full text of DOI:10.1021/ja016199c

J. Am. Chem. Soc. 2001, 123, 7939-7940
7939
The Structural Role of Cholesterol in Biological
Membranes
Exchangeable forms of cholesterol and phospholipids that were
selected for these NNR studies were 1a, 1b, 1c, 2, 3a, 3b, and
c. The design of 2 was based on two considerations. A carbamate
3
moiety was introduced as a headgroup to provide a means for
attaching a pendant thiol group while maintaining a hydrogen-
bonding element off of the C-3 (â) position. In addition, having
the thiol moiety distal from the A-ring was expected to avoid
conformational strain within dimers 2, 3a, 3b, and 3c. With these
ideas in mind, cholesteryl chloroformate was condensed with
cystamine to give homodimer, 2. The corresponding heterodimers,
3a, 3b, and 3c were synthesized by condensing cholesteryl
chloroformate with 2-amino-1-ethyl-2′-pyridyl disulfide, followed
by displacement with the thiol monomer of 1a, 1b, or 1c. The
Michihiro Sugahara, Maki Uragami, Xun Yan, and
Steven L. Regen*
Department of Chemistry and Zettlemoyer
Center for Surface Studies, Lehigh UniVersity
Bethlehem, PennsylVania 18015
ReceiVed May 15, 2001
ReVised Manuscript ReceiVed June 26, 2001
Cholesterol is a major component of mammalian cell mem-
branes. In erythrocytes and in myelin, it is present in concentra-
tions that approach those of phospholipids.1-3 Despite numerous
investigations involving monolayers and bilayers derived from
cholesterol and phospholipids, the structural role that this sterol
plays in producing condensed, fluid membranes has remained a
mystery. In particular, the mechanism by which cholesterol uncoils
6
syntheses of 1a, 1b, and 1c have previously been described.
4
phospholipids has yet to be elucidated. Here we report the use
of the nearest-neighbor recognition (NNR) methodsa chemical
technique that takes “molecular-level snapshots” of membrane
organizationsto clarify this long-standing issue.5
In essence, NNR experiments detect and measure the thermo-
dynamic tendency of two lipids to become nearest-neighbors in
5
the bilayer state. Thus, two lipids of interest are converted into
exchangeable dimers, which are allowed to undergo monomer
interchange via thiolate-disulfide displacement. The equilibrium
dimer distribution that is produced is then analyzed as formal,
noncovalent bonds between pairs of adjacent lipids. Specifically,
a membrane that is composed of A and B monomers may be
treated as an equilibrium mixture of homodimers and heterodimers
according to eqs 1 and 2. Here, K is the equilibrium constant for
the AA homodimer, BB homodimer, and the AB heterodimer.
When phase separation occurs, however, K represents an apparent
equilibrium constant. If a membrane is made from an equimolar
quantity of AA and BB, and if A and B are randomly distributed
after equilibrium is reached, the observed dimer distribution would
be statistical. In other words, the mole ratio of AA/AB/BB would
be 1/2/1, and the equilibrium constant would be equal to 4. If a
thermodynamic preference for hetero-associations existed, how-
ever, this would be reflected by a value of K that is greater than
To ensure the appropriateness of these lipids in model studies,
we compared the monolayer properties of 1a and 2 with those of
1,2-dimyristoyl-sn-glycero-3-phosphatidylglycerol (DMPG) and
cholesterol at the air/water interface (Figure 1). It should be noted
that both 1a and DMPG have identical acyl chains, very similar
melting behavior, and negatively charged phosphate headgroups.
Similar to literature reports, cholesterol yielded a surface pres-
sure-area isotherm having low compressibility and a limiting
6
2
-1 7
area of 0.395 nm ‚molecule . Examination of 2, under identical
conditions, showed slightly greater compressibility in the region
-
1
2
-1
of 0 to 15 mN‚m and a limiting area of 0.809 nm ‚molecule .
The greater compressibility of this dimer is a likely consequence
of the bridging disulfide unit, which imparts greater flexibility to
the surfactant. The fact that the limiting area of 2 is close to twice
that of cholesterol indicates a strong similarity in their packing
behavior. Additionally, the fact that 1a has a limiting area (1.58
4. In contrast, favored homo-associations are indicated by a value
of K that is less than 4.
K
AA + BBy
\
z2AB
(1)
(2)
2
-1
2
2
nm ‚molecule ), which is twice that of DMPG (0.80 nm ‚mole-
1
K ) [AB] /[AA][BB]
-
cule ), also indicates similar packing behavior. Finally, the
suitability of 2 as an exchangeable form of cholesterol was
confirmed by comparing its condensing effect on 1a with the
condensing effect of cholesterol on DMPG. Thus, a series of
surface pressure-area isotherms were recorded using varying
mixtures of 1a plus 2, and also DMPG plus cholesterol (not
shown). Plots of average molecular areas as a function of the
Two features of the NNR method are noteworthy. First, the
NNR method is a highly sensitive technique that can detect
differences in nearest-neighbor interactions that are as low as ca.
50 cal/mol. It is well-suited, therefore, for probing sterol-
phospholipid interactions, where differences in nearest-neighbor
interactions are expected to be small. Second, although this
method involves the use of exchangeable dimers, it provides
thermodynamic information that relates to nearest-neighbor
interactions between indiVidual lipid monomers; that is, the
-
1
mole fraction of phospholipid at 25 mN‚m are also given in
Figure 1. As is evident from this figure, the condensing effect of
cholesterol on DMPG is very similar to the condensing effect of
2
on 1a.
8
disulfide bridge has a negligible influence on K values.
Lipid bilayers, which were used for NNR experiments, were
(
1) Van Deenen, L. L. M.; DeGier, J. The Red Blood Cell; Academic
Press: New York, 1964.
2) Asnell, G. B.; Hawthorne, J. N. Phospholipids; Plenum Press: Am-
prepared in the form of large unilamellar vesicles via reverse phase
(
(6) Krisovitch, S. M.; Regen, S. L. J. Am. Chem. Soc. 1992, 114, 9828-
sterdam, 1964.
9835.
(
3) Gennis, R. B. Biomembranes: Molecular Structure and Function;
(7) Mingotaud, A. F. Handbook of Monolayers; Academic Press: San
Springer-Verlag: New York, 1989.
Diego, 1993.
(4) Vist, M. R.; Davis, J. H. Biochemistry 1990, 29, 451-464.
(8) Vigmond, S.; Dewa, T.; Regen, S. L. J. Am. Chem. Soc. 1995, 117,
7838-7839.
(
5) Davidson, S. K.; Regen, S. L. Chem. ReV. 1997, 97, 1269-1279.
1
0.1021/ja016199c CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/20/2001

7940 J. Am. Chem. Soc., Vol. 123, No. 32, 2001
Communications to the Editor
temperature to 60 °C gave similar results (not shown). For
analogous membranes made from 1b and 2, however, a departure
from randomness was evident at sterol concentrations that were
in excess of ca. 20 mol %. In this case, the sterol and the
phospholipid were now favored as nearest-neighbors. A similar
trend was observed in the case of membranes made from 1c,
except that the affinity between the sterol and the phospholipid
was even greater at the higher sterol concentrations. Despite some
scatter in these data, it is clear that K increases, significantly,
above 4 at high sterol concentrations for membranes derived from
1b and 1c. In the presence of 40 mol % of sterol, monomers of
Figure 1. Surface pressure-area isotherms for (A) DMPG and choles-
terol, and (B) 1a and 2 over an aqueous solution that was 0.10 M in
NaCl at 25 °C. Isotherms for 1a and 2 (‚‚‚) are also shown in (A) as
area/monomer for comparison. The insets show average molecular areas
at 25 mN‚m as a function of mole fraction of phospholipid; isotherms
for each mixture are not shown.
1c and 2 are favored as nearest-neighbors by a free energy
difference of ca. 380 cal/mol.
Additional evidence that sterol-phospholipid affinity increases,
as the length of the phospholipid increases, was obtained by
carrying out a competition experiment in which the exchangeable
sterol produced 3a and 3c in the same membrane. Thus, when
mixed membranes were prepared using an equimolar quantity of
-
1
1a and 1c plus an equimolar quantity of 3a and 3c, such that the
total sterol content was 40 mol %, the equilibrated product mixture
showed a molar ratio of 3c/3a of 1.1 at 60 °C; that is, the
exchangeable sterol favored association with the longer phos-
pholipid.
The main conclusion that emerges from this study is that
significant affinity exists between the palmitoyl (C16)-, and also
the stearoyl (C18)-, based phospholipids and the exchangeable
form of cholesterol, when the sterol concentration reaches
biologically relevant levels. This affinity readily accounts for the
long-known, and poorly understood, condensing effect of cho-
lesterol; that is, the uncoiling of the phospholipids is driven by
hydrophobic interactions between their acyl chains and the rigid
hydrophobic framework of neighboring sterols. At the same time,
these findings support the proposal that cholesterol and phos-
pholipids form “complexes” in fluid bilayers.1
0-12
Our results
indicate, however, that such complexes are likely to form only
when those phospholipids are of an appropriate chain length, and
only when there is a sufficient sterol concentration present in the
12
membrane. The chain length dependency on phospholipid-sterol
affinity that has been observed is best accounted for in terms of
attractive hydrophobic interactions between (i) the sterol nucleus
and the acyl chains of the phospholipids (the fatty acyl chain is
flexible and is able to complement, perfectly, the shape of
cholesterol such that the number of hydrophobic contacts is high
and the packing is tight) and (ii) the acyl chains of two or more
phospholipids that become part of the complexsthe longer the
acyl chains, the greater the hydrophobic contributions to the
stability of the complex. The discontinuity that is apparent in
Figure 2, B and C, is a likely consequence of a transition from
the liquid-disordered state to a liquid-disordered/liquid-ordered
coexistence region, where higher sterol concentrations result in
a greater percentage of the membrane being placed in the liquid-
ordered phase and where tighter packing leads to a higher value
of K.
Figure 2. Plot of K versus mol % of exchangeable sterol present in
bilayers made from (A) 1a, (B) 1b, and (C) 1c; equilibration temperatures
were 40, 60, and 68 °C, respectively. All thiolate-disulfide interchange
reactions were carried out using vesicles made from varying molar ratios
of phospholipid homodimer/heterodimer.
evaporation methods. Varying concentrations of sterol were
included in each membrane type, using varying percentages of a
given phospholipid homodimer and the corresponding het-
erodimer. In all cases, dispersions were made using an aqueous
3
solution that was 140 mM in NaCl, 2 mM in NaN and 10 mM
in borate buffer (pH 7.4). Prior to monomer exchange, each
dispersion was taken to a temperature that was ca. 20 °C above
the gel to liquid-crystalline phase-transition temperature of the
6
phospholipid homodimer. Thus, vesicles that were made using
1
a, 1b, and 1c were maintained at 40, 60, and 68 °C, respectively.
Acknowledgment. We are grateful to the National Institutes of Health
PHS Grant GM56149) for support of this research.
Experimental protocols that were used in forming large unila-
mellar vesicles, promoting monomer interchange via thiolate-
disulfide displacement [after partial reduction (less than 10%) with
(
Supporting Information Available: Synthesis of 2, 3a, 3b, and 3c
PDF). This material is available free of charge via the Internet at
http://pubs.acs.org.
dithiothreitol], and analyzing dimer distributions by HPLC were
(
similar to those that have been reported.5
,8,9
In all cases,
equilibrium was reached within 6 h, as indicated by dimer
distributions that became constant with time. Specific values of
K, the apparent equilibrium constant, which were calculated from
observed dimer distributions, were then plotted as a function of
the mol % of the sterol monomer present in the membrane (Figure
JA016199C
(
9) Shibakami, M.; Inagaki, M.; Regen, S. L. J. Am. Chem. Soc. 1998,
20, 3758-3761.
10) Radhakrishnan, A.; McConnell, H. M. J. Am. Chem. Soc. 1999, 121,
1
(
2).
486-487.
(
11) Radhakrishnan, A.; Li, X.; Brown, R. E.; McConnell, H. M. Biochim.
For bilayers that were made using 1a and 2, a random
distribution of dimers was found with sterol concentrations that
ranged from 9 to 41 mol % at 40 °C; increasing the equilibration
Biophys. Acta 2001, 1511, 1-6.
(
12) Keller, S. L.; Radhakrishnan, A.; McConnell, H. M. J. Phys. Chem.
B. 2000, 104, 7522.