| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, March 2002, p. 1429-1444, Vol. 82, No. 3
and
*MEMPHYS, Physics Department, University of Southern
Denmark-Odense, DK-5230 Odense M, Denmark; Department of
Chemistry, Technical University of Denmark, Building 206, DK-2800
Lyngby, Denmark; Centre for the Physics of Materials,
Department of Physics, McGill University, Montreal, Quebec H3A 2T5,
Canada; §Department of Physics, Simon Fraser University,
Burnaby, V5A 1S6 British Columbia, Canada; and ¶Department
of Physics and Astronomy, University of British Columbia, Vancouver,
V6T 1Z3 British Columbia, Canada
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Cholesterol is an important molecular component of the plasma membranes of mammalian cells. Its precursor in the sterol biosynthetic pathway, lanosterol, has been argued by Konrad Bloch (Bloch, K. 1965. Science. 150:19-28; 1983. CRC Crit. Rev. Biochem. 14:47-92; 1994. Blonds in Venetian Paintings, the Nine-Banded Armadillo, and Other Essays in Biochemistry. Yale University Press, New Haven, CT.) to also be a precursor in the molecular evolution of cholesterol. We present a comparative study of the effects of cholesterol and lanosterol on molecular conformational order and phase equilibria of lipid-bilayer membranes. By using deuterium NMR spectroscopy on multilamellar lipid-sterol systems in combination with Monte Carlo simulations of microscopic models of lipid-sterol interactions, we demonstrate that the evolution in the molecular chemistry from lanosterol to cholesterol is manifested in the model lipid-sterol membranes by an increase in the ability of the sterols to promote and stabilize a particular membrane phase, the liquid-ordered phase, and to induce collective order in the acyl-chain conformations of lipid molecules. We also discuss the biological relevance of our results, in particular in the context of membrane domains and rafts.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Cholesterol is the sterol molecule universally present in mammalian cells. It is predominantly distributed in the cell plasma membrane, amounting to 30-50 mol % of the total lipid fraction of the membrane. Given the fact that a plasma membrane consists of >200 lipid species and that the lipid composition varies greatly from one cell type to another, the presence of cholesterol certainly appears strikingly singular. At the level of molecular structure, cholesterol, while being amphiphilic, also differs significantly from the other lipid species. Its hydrophobic part consists chemically of a planar steroid ring and a short hydrocarbon tail (Fig. 1 a). Consequently, this part of the molecule is physically rigid and smooth at atomic scale (Fig. 1 b). This molecular characteristic has important implications for the interactions of cholesterol with other lipid species.
|
A considerable amount of research has been carried out during the last
few decades to elucidate the biological functions of cholesterol in
cells, to understand the physical basis for the biological functions by
investigating the role of cholesterol in modulating physical properties
of artificial and biological membranes, and to unravel the relationship
between the functions and the molecular structure of cholesterol
(Finegold, 1993
; Vance and Van den Bosch, 2000). However, an equally
important and intriguing question, the question of the origin of
cholesterol in the context of cell evolution, has received relatively
less attention, perhaps due to the apparent overwhelming complexity
that must be involved in answering the question.
Nevertheless, a hypothesis, based on the pioneering work of Konrad
Bloch, has been put forward (Bloch, 1965, 1983, 1994; Bloom and
Mouritsen, 1995): cholesterol has been selected over the almost unimaginably long time scale of natural evolution for its ability to
optimize certain physical properties of cell membranes with regard to
biological functions. As a working hypothesis, it underlines the need
to do comparative studies of the physical effects of cholesterol and
its evolutionary precursors on cell membranes.
Although no fossils exist of evolutionary precursors to cholesterol,
living "fossils" have been argued by Bloch to be present in the
contemporary biosynthetic pathways of sterols (Bloch, 1965, 1983,
1994). In other words, the temporal sequence of the biosynthetic pathway can be taken to represent the evolutionary sequence. The evidence comes from a series of studies of sterol biochemistry and
organism evolution (Bloch, 1994) and appear highly convincing. This
concept of living molecular fossils offers the possibility of an
experimental laboratory program to identify the physical properties
that are relevant to evolutionary "optimization."
Inspired by Konrad Bloch's idea of sterol evolution, we have chosen to
study cholesterol in comparison with its precursor, lanosterol.
Lanosterol occupies an important position in the sterol pathways, as it
is a common precursor in both the mammalian and fungal sterol pathways.
A comparative analysis of the molecular structures of the two sterols
(see Fig. 1) clearly shows that lanosterol, with three additional
methyl groups, can be characterized as being structurally less smooth
in its hydrophobic part than cholesterol: the methyl group at position
14, in particular, protrudes from the steroid surface. In fact, the
biochemical process of converting lanosterol to cholesterol is
essentially a process of streamlining the steroid surface by successive
removal of the three methyl groups, where the C-14 methyl group is the
first to be removed (Bloch, 1983, 1994). This structural
differentiation will turn out to have consequences in terms of the
physical roles each of the two sterols plays as a molecular component
in lipid membranes.
Being fully aware of the fact that the issue of sterol evolution in its full biological context is not an easy one to approach by purely physical techniques, we have taken on a somewhat smaller task of posing a well-defined question that can be answered quantitatively by use of physical concepts and techniques. Because artificial bilayer membranes are important model systems for biological membranes, we have studied model membranes composed of a single species of phosphatidylcholine and either one of the two sterols. Due to their amphiphilic nature, both cholesterol and lanosterol easily intercalate in a lipid-bilayer membrane, with their hydroxyl group positioned on the average at the hydrophobic-hydrophilic interface and with the hydrophobic steroid skeleton embedded in the bilayer core. The molecular ordering and the phase equilibria of these lipid-sterol systems provide a specific context within which differential effects of lanosterol and cholesterol can be unambiguously defined and quantitatively assessed through experimental and theoretical investigations.
Several other comparative studies of lanosterol-lipid and
cholesterol-lipid membranes have been reported, each focusing on a
particular aspect of the effects of the sterols on the physical properties of the lipid membranes. In particular, the
"microviscosities" (or more precisely, the conformational order of
the lipid acyl chains) of cholesterol-lipid and lanosterol-lipid
membranes were measured and compared (Dahl et al., 1980), and it was
shown that cholesterol increased the "microviscosity" of the
membranes much more effectively than lanosterol. The differential
effects of lanosterol and cholesterol on the permeability of lipid
membranes to small sugar molecules (Yeagle et al., 1977; Bloch,
1983) were also investigated, and it was again demonstrated
that cholesterol reduced the membrane permeability more effectively
than lanosterol. Moreover, molecular ordering and dynamics of
phosphatidylcholine bilayers in the presence of either cholesterol (or
ergosterol) or lanosterol were investigated experimentally by use of
NMR techniques at a couple of specific temperatures and sterol
concentrations (Yeagle, 1985; Urbina et al., 1995). Recently, the
effects on membrane domain formation of the molecular structures of
several sterols, including both lanosterol and cholesterol, were
studied in both fluorescence-quenching and detergent-solubilization
experiments and compared (Xu and London, 2000). These studies showed
that cholesterol has a stronger ability than lanosterol to induce
conformational order in lipid chains and promote domain formation in
membranes. Even more recently, a theoretical study of lipid-lanosterol
and lipid-cholesterol bilayers by use of molecular dynamics simulation methods was carried out (Smondyrev and Berkowitz, 2001). With simulations performed over microscopic time scales (nanoseconds), the
study focused on investigating differences in physical properties of
the two types of lipid-sterol bilayers, which include lipid-chain ordering, the dynamics, and the location in the bilayer transverse direction of the sterol molecules. At a relatively modest sterol concentration (~10%), subtle differences in sterol location and mobility in the bilayers were identified, which indicate that lanosterol is more mobile than cholesterol as a bilayer constituent. At
a sterol concentration of 50%, cholesterol was observed to have an
overall slightly stronger condensing effect on the bilayer than
lanosterol. The objective of our study is to establish the systematics
of thermodynamic behavior of lanosterol-lipid and cholesterol-lipid
membranes in terms of phase equilibria, and furthermore, to understand
the molecular basis underlying the phase equilibria by combining
theoretical modeling with experimental investigations.
The phase equilibria of many different types of model membranes have
been understood. A bilayer composed of a single lipid species displays
several phase transitions when driven thermally. We will be concerned
with one of them, the main transition. The transition is first-order,
involving two simultaneous macroscopic processes corresponding to the
ordering of the translational and the internal chain conformational
variables that pertain to lipid molecules. The transition takes the
pure lipid bilayer from a solid-ordered (so) phase, where
the bilayer is a two-dimensional crystal of the lipids with their
chains in conformationally ordered states, to a liquid-disordered
(ld) phase, where the bilayer is a two-dimensional liquid of
the lipids with their chains in conformationally disordered states
(Mouritsen, 1991). In other words, the two ordering processes are
strongly coupled in the main transition, even though no fundamental
principles dictate that they should be so.
The main transition is of key importance for understanding the
influence of sterols in membranes. Sterol molecules such as lanosterol
and cholesterol are, in contrast to phospholipids, essentially rigid
and relatively smooth in their hydrophobic parts. Their mode of
interaction with lipid molecules is of dual nature. On the one hand,
they prefer to have next to them acyl chains that are ordered as in the
so phase, thereby inducing chain ordering. On the other
hand, because the sterols have different molecular shapes from a
conformationally ordered lipid chain, they tend to break the lateral
packing order of the so phase. Consequently, these sterols
can uncouple the two types of macroscopic order in the translational
and conformational variables of the lipid molecules. This effect can
lead to the emergence of a physical state of lipid-sterol membranes,
which has characteristics intermediate between the so and
the ld states: the liquid-ordered (lo) state. This phenomenology underlies the first simple theory of the phase equilibria in cholesterol-lipid systems (Ipsen et al., 1987), which
yielded predictions consistent with subsequent experimental observations (Vist and Davis, 1990; Thewalt and Bloom, 1992; Silvius et
al., 1996). There is mounting evidence that the lipid-bilayer component
of many biological membranes needs to be in this liquid-ordered state
to perform its biological functions properly (Bloom et al., 1991).
The nature of the phase equilibria of lipid-cholesterol membranes has
been a particularly elusive problem, and to some researchers the
problem still remains controversial. Specifically, a few other different pictures of the phase equilibria of lipid-cholesterol membranes have been put forward: one picture involving the concept of
molecular complex formation is based on experimental studies of
monolayers formed at air-water interfaces from binary mixtures of
cholesterol and a charged phospholipid (dimyristoylphosphatidylserine) (Radhakrishnan and McConnell, 1999; Radhakrishnan et al., 2000). Another picture is derived from differential scanning calorimetry (DSC)
measurements of bilayer membranes of binary mixtures of cholesterol and
dipalmitoylphosphatidylcholine and relies on a particular
interpretation of observed features in the specific heat (McMullen and
McElhaney, 1995). A theoretical study has also been carried out, which
suggests a microscopic interaction model that involves many-body
interactions between constituent molecules in membranes of
cholesterol-lipid binary mixtures at very high cholesterol
concentrations (Huang and Feigenson, 1999).
Our studies of simple bilayers of binary mixtures of phospholipids and
the sterols consist of two parts: experiments using solid-state
deuterium-NMR spectroscopy and calorimetry, which investigate and
characterize the collective conformational ordering of the lipid
molecules and the thermal behavior of the bilayer membranes,
respectively, and parallel statistical mechanical modeling and computer
simulations, which focus on understanding and relating the observed
macroscopic phenomena to molecular interactions, and in turn, to the
different molecular chemistries of the two sterol molecules. Our most
important conclusion is that cholesterol, with its streamlined
molecular structure, interacts more effectively with lipid chains with
conformational order and stabilizes the liquid-ordered state of the
lipid bilayers more effectively than lanosterol. This main finding has
been briefly reported in a recent letter publication (Nielsen et al.,
2000). In the present full article we will present a complete account
of our studies, which includes additional data and results not yet
reported. We will also discuss the results in connection to the
biological functions of plasma membranes.
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Experimental
Chemicals and sample preparations
1-palmitoyl-2-petroselinoyl-sn-glycero-3-phosphatidylcholine (PPetPC) is cis-unsaturated at position C6-7 of the sn-2 chain, and was obtained from Avanti Polar Lipids Inc. (Birmingham, AL). This lipid differs from POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine)
a lipid species common in biological membranesin one respect: its single unsaturated C-C bond is closer to the hydrocarbon-water interface. The advantage of using PPetPC instead of POPC is that its
main-transition temperature is above the freezing point of water, and
this offers us an advantage in carrying out experiments. For NMR
experiments, PPetPC-d31, of which the palmitoyl
chain is perdeuterated, was obtained by custom synthesis from Avanti. It contained ~15% of the lipid with chains interchanged
(PetPPC-d31) as a byproduct of the synthesis.
Cholesterol, lanosterol, and deuterium-depleted water were obtained
from Sigma Chemical Co. (St. Louis, MO) and used without further
purification. The main impurity in lanosterol is 24-dihydrolanosterol.
To prepare multilamellar lipid dispersions, aliquots of
CHCl3 stock solutions of phospholipid and sterol
were mixed in the appropriate quantities. The
CHCl3 was evaporated under a thin stream of
N2 and the samples were then placed under high
vacuum overnight. The resulting films were dissolved in
benzene/methanol (95:5 vol/vol) and then lyophilized overnight.
Finally, the samples were hydrated at room temperature with a buffer
(20 mM Hepes and 300 mM NaCl in deuterium-depleted water at pH 7.4) and
mixed thoroughly; 50 mg PPetPC-d31 and 600 µl
buffer, along with an appropriate mass of sterol, were used to make the
samples for NMR measurements. In DSC measurements the total sample
volume was 0.7 ml, containing ~20 mg phospholipid, along with an
appropriate amount of sterol.
NMR and DSC measurements
Deuterium NMR spectra were obtained by using the quadrupolar echo technique with a locally built spectrometer operating at 46 MHz. A typical spectrum resulted from 10,000 repetitions of the two-pulse sequence with a 90° pulse length of 4 µs, inter-pulse spacing of
= 40 µs, and dwell time of 2 µs. The delay between acquisitions was 300 ms, and data were collected in quadrature with
Cyclops 8-cycle phase cycling. Occasionally the spectra were symmetrized before subtraction by zeroing the out-of-phase channel to
improve the signal-to-noise ratio. If the quadrupolar echo spectra of
dispersions having different sterol concentrations were not uniformly
affected by relaxation, a correction was performed to ensure correct
weighting of the spectral components (Thewalt et al., 1992).
From the spectral data, the average half first moment,
M1, was calculated as
|
(1) |
) is the signal at a given frequency separation from the central (Larmor) frequency. The non-zero values of
f() are found between the points 
x and
x where the spectrum ends. In practice, x and
x were chosen to be ±125 kHz. The first moment,
M1, is related to the chain-averaged carbon-deuteron order parameter
SCD
as
|
(2) |
Theoretical modeling
As a parallel to the experiments, the theoretical modeling of the phase equilibria of the lipid-sterol systems is used to provide insight into the microscopic physics underlying the experimental observations. In this part, basic microscopic models for lipid-sterol interactions are first proposed and the thermodynamic phase equilibria are then investigated by the statistical mechanical studies of the models by use of Monte Carlo computer simulations.
Model
Recently, a microscopic model was proposed for bilayer membranes of phospholipid-cholesterol binary mixtures to describe the phenomenology of the molecular interactions between cholesterol and lipids, and the equilibrium phase behavior of the membrane systems predicted by the model is entirely consistent with the experimental observations of the same systems (Nielsen et al., 1999). In the present
work, we have extended that theoretical modeling to investigate
systematically the phase equilibria of lipid-sterol bilayers as a
function of the evolutionary chemistry of the sterols.
Our model description of a lipid-sterol membrane consists of three
basic ingredients: a representation of the translational degrees of
freedom, a description of molecular conformational degrees of freedom,
and a minimal model of the interactions between the molecules. A
complete presentation of the model is given in the Appendix.
The physical essence of our modeling of molecular interactions may
deserve some emphasis here. Both the lipid and sterol molecules are
modeled as hard-core particles that interact with each other. The
molecular interactions are modeled by combinations of square-well potentials that are shown schematically in Fig.
2. The model is proposed based on the
established phenomenology of lipid-cholesterol interactions (Ipsen et
al., 1987; Nielsen et al., 1999) and the comparative analysis of the
molecular structures of cholesterol and lanosterol. The figure captures
the dual nature of the microscopic interactions between a sterol
molecule and a lipid molecule, which is at the center of the
microscopic interpretations of the lipid-cholesterol phenomenology; it
also emphasizes the difference in the lanosterol-lipid and
cholesterol-lipid interactions (see Fig. 2 B). Within the model, the depth of the longer-range potential well in Fig. 2 B, Jchol or
Jlan, is a measure of the strength of
cohesive interactions between a sterol and a lipid molecule with
conformationally ordered chains. Thus, our modeling proposes that the
predominant (though not the only) physical effect of varying
molecular structure of a sterol is to modulate the cohesive
interactions, i.e., the packing, between the sterol and a lipid
molecule (Bloch, 1983). In other words, J may be taken as a
principal physical representation of the degree of molecular smoothness
of sterols: from lanosterol to cholesterol there is an increase in the
value of J, as illustrated both qualitatively in Fig. 2
B and quantitatively in Table
1. Hence, the microscopic
physical consequences of the evolution in the sterol chemistry are
described by a single-parameter-tuned potential of sterol-lipid
interactions. In this sense our approach is a minimal-model approach.
Details on how the values of Jchol or
Jlan and the values of other
parameters are chosen are provided in the Appendix.
|
|
Simulation methods and data analysis
The models presented above are studied by using the statistical mechanical method of Monte Carlo simulations (Frenkel and Smit, 1996).
From the simulation data, the phase equilibria of the model systems are
established in terms of both phase diagrams and thermodynamic
characterizations of the phases by molecular ordering and lateral
distributions (structures).
The basic simulation method is that of Monte Carlo importance sampling
(Frenkel and Smit, 1996). The straightforward application of the
sampling method directly uses the microscopic Hamiltonian describing a
system, and each sampling run generates a statistical ensemble of
microscopic states according to the Boltzmann distribution. In our
simulation studies of the lipid-sterol systems this direct sampling
method suffices when we explore regions in the parameter space where no
phase transitions are encountered. To deal with the difficulty of
barrier crossing associated with strongly first-order phase
transitions, however, a modified sampling method is used, where the
"Hamiltonian" directly controlling the sampling process consists of
the physical Hamiltonian and an artificial part designed to reduce the
energy barrier separating two coexisting phases. Consequently, the
statistical ensemble generated by a sampling run does not follow the
Boltzmann distribution corresponding to the physical Hamiltonian.
Efficient sampling, however, is achieved, and the physical Boltzmann
distribution can be constructed from the data collected. This type of
non-Boltzmann sampling methods are discussed in detail elsewhere
(Besold et al., 1999; Nielsen et al., 1999).
In each simulation run, periodic boundary conditions and planar
geometry are imposed on the system, and the following parameters are
fixed: the total number of molecules (lipid plus sterol molecules) N, the temperature T, and a lateral
"pressure" 
, which represents the self-stabilizing force of a
bilayer membrane. In particular, is kept fixed in all
simulations. The chemical composition of the lipid-sterol mixtures is
either fixed absolutely by fixing the respective numbers of the lipid
and sterol molecules or fixed only on average via an effective
"chemical potential." The former case is referred to as the
canonical ensemble and the latter case as the semi-grand canonical
ensemble (Frenkel and Smit, 1996). To derive phase diagrams, many
different simulation runs are performed systematically by varying
T and the chemical composition. At each fixed setting of
T and chemical composition, many simulation runs are carried
out for different values of N, such that phase transitions can be identified through finite-size scaling analysis.
A simulation run consists of preparing the system in a high-temperature
liquid state, then quenching it down to a lower temperature desired,
equilibrating the system to that temperature, and finally taking a
sufficiently large number of "measurements" of relevant quantities
for statistical analysis. In the data analysis that leads to the
determination of phase diagrams several different techniques are used,
including histogram and thermodynamic reweighting techniques and
finite-size scaling analysis (Nielsen et al., 1999).
Various thermodynamic quantities are calculated to explicitly
characterize different physical phases. Among them is 
, the equilibrium conformational order parameter of the lipid chains. This
parameter, which is defined in the Appendix, is within our model the
theoretical counterpart of SCD,
the experimental chain order parameter defined in Eq. 2. Another class
of the calculated quantities are the so-called structure factors,
ST(
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Experimental results
DSC data, NMR spectra, and data analysis
We performed DSC measurements on multilamellar dispersions of PPetPC/cholesterol and PPetPC/lanosterol mixtures as functions of sterol concentration. Representatives of the obtained data are shown in Fig. 3.
|
|
Phase diagrams
In deriving experimental phase diagrams, both the DSC data and the NMR spectra were used. The apparent difference between the main-transition temperature reported by the DSC data (16.8°C) and that yielded by the NMR data (15°C) on the pure PPetPC membrane can easily be explained. The DSC data presented were obtained for non-deuterated PPetPC, while the NMR measurements required the deuterated PPetPC. The pure membranes made of the deuterated PPetPC were shown by DSC scans to have the main-transition temperature at 15°C (data not shown here). When phase diagrams were derived by combining the DSC data together with the NMR data, this difference was taken into account and properly offset by subtracting 1.8°C from the DSC data obtained for non-deuterated PPetPC. The spectral-subtraction method mentioned above has been described thoroughly elsewhere (Vist and Davis, 1990; Thewalt et al., 1992).
Briefly, two spectra corresponding to membrane systems of different
sterol concentrations at a given temperature are used and analyzed
together. For obtaining the solidus endpoint at the given temperature,
a judiciously determined fraction, K, of the
higher-concentration spectrum can be subtracted from the lower-concentration spectrum to yield just a spectrum of the purely so membrane. The sterol concentration corresponding to this
so endpoint spectrum can then be calculated in a
straightforward manner by using the two known sterol concentrations and
the fraction, K. The sterol concentration marking the
liquidus endpoint of the coexistence is obtained in a similar way, by
subtracting a fraction of the lower-concentration spectrum from the
high-concentration spectrum.
In the case of PPetPC/cholesterol membranes, subtractions were
performed by using the spectra from membrane dispersions containing 10, 15, and 20 mol % cholesterol at each temperature in the range from 0 to 12°C, and resulted in nearly vertical solidus and liquidus boundaries at sterol concentrations of 8-9 mol % and 28-30 mol %,
respectively. For PPetPC/lanosterol, subtractions were possible only up
to 9°C (due to the emergence of the ld phase at higher temperatures) and resulted in solidus and liquidus lines that tended
toward higher lanosterol mole fractions as the temperature was lowered
(see Fig. 5).
|
|
Theoretical results
In this section we present the results of the Monte Carlo simulations in terms of the equilibrium phase diagrams, collective conformational ordering of the lipid chains, and structural characterizations of the lipid-sterol membranes.
Phase diagrams
Shown in Fig. 7 are the equilibrium phase diagrams for the lipid-cholesterol and the lipid-lanosterol membranes, which are derived from a large amount of simulation data. The phase diagrams are given in the thermodynamic parameter space spanned by the sterol concentration xsterol (where xsterol = xchol or xlan) and a reduced temperature, T/Tm, where Tm is the main transition temperature for the pure lipid system. In terms of microscopic interaction parameters, the lipid-cholesterol systems are distinguished from the lipid-lanosterol systems by a rather modest (~10%) increase in the strength of the cohesive interactions between a sterol molecule and a lipid chain in its conformationally ordered state (see Table 1).
|
1.0075Tm,
xchol 0.298. The temperature of
the three-phase coexistence is estimated to be T = 0.9977Tm, and the concentrations of
cholesterol in the three coexisting phases is
xchol,so = 0.030, xchol,ld = 0.068, and
xchol,lo = 0.298, respectively. In the
phase diagrams of the lipid-lanosterol systems, no ld-lo
coexistence can be identified; correspondingly, only a metastable
critical point exists and there is no three-phase line.
The systematic development of the stable ld-lo coexistence
as lanosterol "evolves" to cholesterol is a macroscopic signature of an increase in the capacity of the sterols to stabilize the lo phase. Another consistent signature is that with the
sterol "evolution" the lo phase boundary of the
low-temperature so-lo coexistence moves toward lower sterol
concentrations, indicating a broadening of the region of stability of
the lo phase in the cholesterol-lipid membranes.
Conformational ordering of the lipid molecules
To characterize quantitatively the differential effects of the two sterols on the physical properties of the mixture membranes, the lipid-chain order parameter,, as defined in the Appendix, is
calculated and presented in Fig. 8. This
thermal average represents the collective conformational ordering of
the lipid molecules.
|
is shown, for a fixed temperature
T = 1.0129Tm, as a
function of sterol concentration for both types of the lipid-sterol
systems. This temperature is above that corresponding to the critical
point of the ld-lo coexistence region of the
lipid-cholesterol system. At this temperature, neither of the two types
of the lipid-sterol systems undergoes any phase transitions as the
sterol concentration is changed. In Fig. 8 B, is given
for a fixed sterol concentration xsterol = 0.367. Clearly, both
cholesterol and lanosterol have order-inducing effects on the lipid
chains, but cholesterol is much more potent. As shown in Fig. 8
A, for example, cholesterol at
xchol = 0.40 is able to rigidify close
to 55% of the lipid chains, whereas lanosterol at the same
concentration can only rigidify 30%.
Molecular organization
The statistical mechanical modeling of the microscopic models also allows us to analyze quantitatively the lateral distribution of the molecules in the lipid-sterol membranes. The quantities characterizing lateral structures are the structure factors, the definitions of which are given in the Appendix. Fig. 9 contains examples of the calculated structure factors. The data shown are calculated at T = 0.9806Tm and xsterol = 0.367, where the membranes are in an lo state. The figure shows the circular averages (over the directions of the Fourier wave vectors,
0.35·2
/d, where d is the radius of the
hard-disk cross-section of each particle, as defined in the Appendix.
This particular q-value corresponds to a real-space length
scale that covers several molecules, if a quantitative value of 5 Å for d may be used. The signal almost disappears in Slan(
|
|
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
The "thread-like" microdomains
It is useful to compare the last result on the microscopic domains
with the results of a Monte Carlo simulation study by Huang and
Feigenson (1999). Their study focused on understanding the high-concentration solubility limit of cholesterol in lipid
membranes, and it was based on a model which, in addition to the
pairwise interactions, introduced cholesterol multibody interactions.
Furthermore, the model used an underlying lattice representation for
molecular positions. In contrast, our modeling is based on an
off-lattice representation and does not in any way deal with the issue
of solubility. The two-body interactions in Huang and Feigenson's model were nevertheless qualitatively very similar to the microscopic interactions in our model, and produced a molecular distribution pattern on a microscopic scale that resembles the "thread-like" structure observed in our study. The cholesterol multibody interactions were only introduced as a possible mechanism for the removal of cholesterol domains from the membranes, thus accounting for the solubility limit of cholesterol. They also induced some super-lattice structures close to and at the solubility limit, into which the microscopic structures appeared to organize. It is, however, difficult to distinguish unequivocally the super-lattice structures from artifacts, given the underlying lattice representation specifically and
given the lack of clear experimental evidences for super-lattice structures in general.
Naturally, the "thread-like" microstructures found in the present
study are yet to be verified experimentally. They may, however, provide
a microscopic basis for the notion of
"cholesterol-lipid" complexes, which underlies a model for the
thermodynamics of multicomponent lipid-cholesterol systems
(Radhakrishnan and McConnell, 1999; Radhakrishnan et al., 2000).
Lipid-sterol interactions, the lo domains, and the rafts
We have in the previous sections presented both the results of our experimental studies and the results of our theoretical modeling of the phase equilibria of the lipid-cholesterol and the lipid-lanosterol membrane systems. The focus of the studies is to identify and characterize the differential effects of cholesterol and its (evolutionary) precursor, lanosterol, on equilibrium physical properties of the membranes and to relate those effects to the difference in the molecular chemistries of the two sterols.
Clearly, the theoretically calculated phase diagrams (shown in Fig. 7)
agree with those determined experimentally (shown in Fig. 5) for both
types of the lipid-sterol membranes; there is also an agreement between
the theoretical and experimental results for the conformational
ordering of the lipid molecules (see Figs. 6 and 8). Because the
theoretical results are based on our microscopic model of sterol-lipid
interactions with no extensive or specific parameter-fitting, the
agreement provides concrete support for the model which proposes a
physical interpretation of the differential molecular chemistry of
cholesterol and lanosterol. The streamlined molecular structure of
cholesterol (when compared to that of lanosterol) may indeed enable it
to interact more strongly with conformationally ordered lipid chains
(Bloch, 1965; Yeagle et al., 1977; Yeagle, 1985; Slotte, 1995; Urbina
et al., 1995), thereby rendering it more effective than its precursors
in the dual act of inducing high conformational order in the lipid
chains and breaking the lateral packing order of the lipid molecules.
This microscopic effect is, in the membrane systems that we have
studied, macroscopically manifested both as the enhanced stability of
the lo phase with respect to both temperature and sterol
concentrations, as indicated by the phase diagrams, and as an increased
conformational order of the lipid chains, as quantified also in Figs. 6
and 8.
Our physical interpretation of the different molecular chemistry of
cholesterol and lanosterol in terms of a simple model of sterol-lipid
interactions appears to receive additional support from other
experimental studies, where it has been demonstrated that cholesterol
shows a stronger ability than lanosterol to promote and stabilize the
lo structure in membranes formed of lipids other than the
specific lipid (PPetPC) used in our studies. In an earlier comparative
study of the effects of cholesterol, ergosterol, and lanosterol on
molecular order and dynamics of phosphatidylcholine (PC) bilayer
membranes (Urbina et al., 1995), it was shown that at 30 mol % sterol
content and at several temperatures, cholesterol was significantly more
effective than lanosterol at increasing lipid order in both membranes
formed of saturated DMPC and membranes formed of mono-unsaturated POPC.
The difference in the effectiveness of the sterols was suggested to be
caused by the additional steric constraints imposed by the three bulky
methyl groups of lanosterol on its interactions with the PC lipids. In
a more recent study of the effect of sterol structure on domain
formation in membranes containing both a saturated PC (DPPC) and an
unsaturated PC (either a fluorescent analog or DOPC) (Xu and London,
2000), domains with characteristics of the lo structure and
enriched in the saturated lipid were shown to exist in
cholesterol-lipid membranes, while no discernible domain formation was
observed in lanosterol-lipid membranes. These observed effects were
again rationalized in terms of the ability of cholesterol, and the lack
of ability of lanosterol, to participate in strong close-packing with
saturated lipids that are conformationally ordered. However, the work
of Urbina et al. (1995) concerning ergosterol also suggests that the
structure of the steroid rings of a sterol may not be the only
structural factor influencing the packing of the sterol with lipids.
The bulky and stiff C-24 methylated side chain of ergosterol seems to
render the sterol unable to interact effectively with phospholipids with one mono-unsaturated acyl chain.
Regarding our use of PPetPC as the phospholipid in our membrane
systems, there arises a question concerning the difference between
lipids with saturated chains and lipids with mono-unsaturated chains,
and the effect of the position of the double bond along an
monounsaturated chain. In this context there is considerable experimental evidence that the phase behavior observed in
DPPC-cholesterol systems is generic for a large number of
PC-cholesterol systems (Thewalt et al., 1992; Silvius et al., 1996). In
particular, different binary mixtures of cholesterol with saturated
lipids of different chain length display qualitatively similar phase
diagrams. Furthermore, changing the lipid from a saturated species to
one of a number of mono-unsaturated lipids either having a
cis double bond at either 6-7 position or 9-10 position,
or having a trans double bond at 9-10 position preserves
the qualitative topology of the DPPC-cholesterol phase diagram. This
suggests that certain specific details in the molecular chemistry of
lipids are not of primary importance to the qualitative nature of the
macroscopic phase equilibria. Our theoretical model of lipid-sterol
interactions, being minimal, naturally neglects such details. Within
the context of the model it is not the absolute strength of the
lipid-lipid and lipid-sterol interactions, but rather their relative
strength that primarily determines the qualitative features of phase
equilibria of lipid-sterol systems.
In relation to the hypothesis on the molecular evolution of
cholesterol, it is interesting to discuss our results in terms of the
biological relevance of the lo structure in membranes. As
discussed in the previous paragraph, the lo structure is
generic for lipid-cholesterol mixtures in that it appears as a stable equilibrium structure in a large number of different mixture membranes over a wide range of compositions and temperatures (Thewalt et al.,
1992; Almeida et al., 1992; Mateo et al., 1995; Silvius et al., 1996).
Because cholesterol is abundant in plasma membranes of mammalian cells,
and saturated and mono-unsaturated lipids make up a major fraction
among the rest of lipid constituents in the membranes (White, 1973), it
may be expected that the physical properties of the lo
structure play a crucial role in supporting some of the essential
biological functions of the membranes.
Among their many different functions, cell plasma membranes must act as
physical protection to the intracellular organelles and biochemical
processes, and must therefore have the kind of mechanical properties
that ensure their structural coherence, flexibility, and integrity. The
membranes must also act as an effective physical barrier to passive
cross-membrane transport of molecules to secure the active transport
processes that are regulated by molecular pumps, channels, and
carriers. At the same time, the membranes must maintain fluidity, a
physical property essential for the necessary lateral mobility of
lipids and membrane-associated proteins required for many
membrane-related cellular processes (Bloch, 1965; Bloom et al., 1991;
Bloom and Mouritsen, 1995). Although a physical understanding that
would unify these different, unusual, some even seemingly mutually
conflicting, aspects of the cell membranes is still missing, a unifying
picture may be emerging for the simpler systems of model
lipid-cholesterol membranes. In particular, phospholipid bilayers
containing ~20-30% cholesterol are known to be fluid (Evans and
Needham, 1986) and to have elasto-mechanical strengths comparable to
the membrane of red blood cells (Evans and Needham, 1986; Needham and
Nunn, 1990; Düwe et al., 1990; Meleard et al., 1997). It
has also been reported that 50% cholesterol in lipid bilayers almost
completely abolishes the cross-membrane permeation of small molecules
and ions (Bloch, 1983; Bhattacharya and Haldar, 2000), while lanosterol
has hardly any effect on the permeation (Bloch, 1983). Our results
suggest that the notion and the existence of the lo
structure in the lipid-sterol bilayer membranes may be the unifying
basis underlying the various types of observations.
In recent years, cholesterol-rich lo structures have
received much attention within another context of cell biology.
Detergent-resistant membrane fragments (DRMs) have been isolated from
detergent treatment of a variety of eukaryotic cells. These membrane
domains are rich in cholesterol and saturated sphingolipids, and are
suggested to have the physical characteristics of the lo
phase (Ostermeyer et al., 1999). Based on the studies of DRMs, a
hypothesis has been put forward that rafts, which are domain
structures enriched in cholesterol and sphingolipids, exist in cell
membranes (Simons and Ikonen, 1997; Brown and London, 1997).
This hypothesis is entirely consistent with the known facts that
cholesterol strongly favors interacting with lipids with saturated acyl
chains and that sphingolipids have fully saturated acyl chains and
constitute a significant fraction of the membrane lipids. The potential
functional importance of rafts in processes such as intracellular
membrane sorting and signal transduction at the cell surface has been
proposed (Brown and London, 2000). The existence of rafts in cell
membranes is yet to be unambiguously verified. However, rafts have been observed in multicomponent model membranes formed of unsaturated phosphatidylcholines, cholesterol, and sphingomyelin (Dietrich et al.,
2001; Rinia and de Kruijff, 2001). Moreover, in ternary mixtures of
cholesterol with both a saturated lipid and an unsaturated lipid,
domains, which are in the lo phase and are enriched in the
saturated lipid, and which show resistance to detergent solubilization, have been demonstrated to coexist with domains in the ld
phase that are enriched in the unsaturated lipid (Xu and London, 2000). It is also suggested that cholesterol concentration is higher in the
lo domains than in the ld domains. This observed
phenomenon is rationalized by a better structural fitting, or in our
terms, a stronger cohesive interaction, between cholesterol and the
saturated lipid than between the sterol and the unsaturated lipid. The
comparative study of the different lo domains corresponding
to different sterols (including both cholesterol and lanosterol) used
in the mixtures further supports the notion that the structural
smoothness of a sterol is one of the key parameters in domain
formation: the better a sterol can form tight packing with lipids with
saturated acyl chains, the more strongly it promotes the formation of
the detergent-insoluble domains (Xu and London, 2000). This
structure-based rationalization is indeed at the center of our
theoretical model of sterol-lipid interactions. Furthermore, the
experimental work on rafts underlies the crucial role of
cholesterol in promoting and stabilizing the lo-type domains
in membranes, which is also the main conclusion of our study.
| |
APPENDIX |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
APPENDIX
REFERENCES
|
|---|
Theoretical microscopic model
The model proposed in this paper for the systems of lipid-sterol
mixtures is in essence similar to a model earlier proposed by us for
the lipid-cholesterol mixture system (Nielsen et al., 1999). The model
consists of three basic ingredients: a representation of the
translational degrees of freedom, a description of molecular conformational degrees of freedom, and a minimal model of the interactions between the molecules.
In representing the translational degrees of freedom, the model uses an
algorithm that formulates a spatial distribution of molecules in the
two-dimensional space as a triangulation of a plane. By randomly
accessing all forms of triangulations, the algorithm explores all
possible configurations of molecular distribution. It therefore allows
for local density fluctuations and molecular diffusion in the space,
which are the essential elements in a realistic description of
molecular translational degrees of freedom. The details of this
triangulation algorithm are given elsewhere (Nielsen et al., 1996).
The conformational degrees of freedom associated with the lipid chains
are represented by two states (Doniach, 1978). One state, the
"ordered" state, has zero internal energy and is non-degenerate, characteristic of the conformational state of a lipid chain in the
"gel" phase. The other state, the "disordered" state, has a
high internal energy, reflecting that energy is required for conformational excitations, and a large degeneracy, effectively representing the large number of conformational excitations that a
lipid chain can assume. Both cholesterol and lanosterol molecules are
modeled as rigid particles with no conformational degrees of freedom.
All particles are considered as hard-core particles.
The energy function that models molecular interactions in the
lipid-sterol mixture systems is given by
|
(3) |
|
|
|
(4) |
i < j denotes a summation
over nearest neighbors, and i is an index labeling the
particles in the system. 
io,
id, and
is are occupation variables which
are unity when the ith particle is a lipid chain in the
ordered state, a lipid chain in the disordered state, and a sterol
molecule, respectively, and which are zero otherwise. The energy of
interaction between two chains that are both in the disordered state is
approximated to be zero, which sets the reference point for the
interaction energies.
H0 is the energy function describing
the interactions in the pure lipid system. It is formulated as
|
(5) |
|
is, in effect, a lateral
surface pressure stabilizing the system against lateral expansion, and
A is the total area of the system.
All of the microscopic interaction potentials are approximated by a sum
of a hard-core repulsive potential of range d, a short-range square-well potential of range R0,
Vs(R), and a longer-range
attractive square-well potential of range lmax,
Vl(R).
Vs(R) and
Vl(R) are given by
|
(6) |
|
(7) |
Some of the microscopic interactions are sketched in Fig. 2 to illustrate our minimal modeling of the dual molecular mechanism of sterol molecules in lipid bilayers. A comparison between Fig. 2, A and B illustrates the "crystal-breaker" mechanism, as the interactions involved imply that a sterol molecule dissolved in a
-CH3, and 4-
-CH3.
