This is the authors’ final peered reviewed (post print) version of the item
published as:
Hughes, Zak E. and Mancera, Ricardo L. 2013, Molecular dynamics simulations of mixed DOPC–βsitosterol bilayers and their interactions with DMSO, Soft matter, vol. 9, no. 10, pp. 2920-2935.
Available from Deakin Research Online:
http://hdl.handle.net/10536/DRO/DU:30055070
Reproduced with the kind permission of the copyright owner.
Copyright: 2013, Royal Society of Chemistry.
Molecular Dynamics Simulations of Mixed
DOPC-β-Sitosterol Bilayers and their
Interactions with DMSO
Zak E. Hughesa and Ricardo L. Manceraa,b∗
December 6, 2012
a
Western Australian Biomedical Research Institute, Curtin Health Innovation
Research Institute, School of Biomedical Sciences, Curtin University, P.O. Box
U1987, Perth WA, 6845, Australia
b
School of Pharmacy, Curtin University, PO Box U1987, Perth WA 6845,
Australia
Abstract
Cell membrane phospholipid bilayers can be damaged by the large amounts
of dimethyl sulphoxide (DMSO) commonly used in cryopreservation. The interaction of DMSO with model bilayers consisting of 1,2-dioleoyl-sn-glycero3-phosphocholine (DOPC) and β-sitosterol has been studied using molecular
dynamics simulations. Initially the effect of sterol concentration and temperature upon bilayers solvated in pure water was determined, and membranes
containing β-sitosterol were compared with membranes containing cholesterol.
These simulations showed that the presence of sterols has a condensing effect
on the phospholipids, causing a reduction in the area per lipid as the sterol concentration increases, up to a phospholipid-sterol ratio of 2:1. The incorporation
of sterols into the bilayer also increased the thickness and order of the phospho∗ Corresponding
author email: [email protected]
1
lipid acyl tails. DOPC-β-sitosterol bilayers at different relative concentrations
were simulated in solutions of 2.5 and 25.0 mol % DMSO. The interaction of
DMSO with the bilayers caused the bilayers to expand laterally, while thinning
normal to the plane of the bilayer expansion. The same qualitative behaviour
has been shown to occur in pure phosphocholine bilayers. However, the presence
of sterols made the membranes more resistant to the effects of DMSO, to the
extent that the membranes where able to maintain their integrity in 25.0 mol
% DMSO, a concentration that would otherwise cause the destruction of a pure
DOPC bilayer. Increasing the concentration of β-sitosterol within the bilayers
reduced the rate of DMSO diffusion across the bilayer and, if the concentration
was large enough, caused the diffusion mechanism to change. DMSO was observed to disorder the membranes enough to cause an increase in the number of
sterol “flip-flops”. The findings of this work provide a more realistic description
of how DMSO interacts with cell membranes and the role of the composition of
the membrane.
Keywords
Molecular simulation, cryopreservation, cell membranes
2
Introduction
The storage of biological samples by means of cryopreservation1–7 is an increasingly important tool. By storing samples at liquid nitrogen temperatures
(77K) any biological, chemical and physical processes occurring within the sample are slowed to the point that they are effectively suspended. This allows
cryopreserved samples to be stored for decades without any deterioration. Unfortunately, the cryopreservation process itself can cause some damage to cells
but the extent of this damage can be reduced by treating the samples, before
and during cryopreservation, with an aqueous solution containing a number of
cryoprotective agents (CPAs).8–12
There are a number of different CPAs in use such as glycerol, ethylene glycerol, various sugars and DMSO.9, 13, 14 DMSO is one of the most widely used
and important CPAs due to its ability to diffuse through cell membranes15
coupled with its effect on promoting the vitrification of aqueous solutions.16, 17
The formation of a glassy state of water upon rapid cooling of the sample reduces the likelihood of the formation of intracellular ice crystals, which can
cause significant damage to cells. DMSO has been shown to have other interesting properties, such as the ability to induce cell fusion18 and act as an
anti-inflammatory19 and anesthetic.20 As such it is important to determine the
nature of the interactions of DMSO with cell membranes at the molecular level.
The actions of DMSO on cell membranes have been investigated using both
experimental and computational techniques.21–29 The authors recently performed a series of molecular dynamics simulations investigating the interaction
of DMSO with homogeneous 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine
(DPPC) and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) membranes.27
The results of these and other simulations22–24, 26, 28 showed that at low concentrations DMSO causes these model bilayers to expand laterally while thinning
3
normal to the plane of the bilayer. Increasing the concentration of DMSO causes
pores to form in the bilayer and, if the concentration is high enough, the bilayer structure to be destroyed. It was also shown that DOPC bilayers were
more resistant to the effects of DMSO than the DPPC bilayers by being able
to both withstand a higher concentration of DMSO before destruction and having a higher free energy barrier to the diffusion of DMSO molecules through
the bilayer. Simulations have also shown that DMSO displaces water molecules
at the bilayer interface making the membrane dehydrated.24, 27 Experimental
studies have shown that as the concentration of DMSO is increased the spacing
between bilayers is reduced.30–32 This reduction in the repeat spacing of membranes combined with the dehydration of the bilayers are the most likely reason
why the gel-liquid crystal phase transition temperature of phospholipid bilayers
increases with the mole fraction of DMSO present.30–32 Recently, de Ménorval et al. performed simulations of a binary DOPC-cholesterol (at a single 20%
cholesterol concentration) in a range of DMSO concentrations and compared
the simulation results against experiment.33 The simulations showed that the
bilayer behaved in similar manner to pure DOPC bilayers with three regimes
(membrane thinning, pore formation and bilayer destruction) observed. For this
20% cholesterol bilayer it was found that 40 mol % DMSO would cause it to
be destroyed. However, the simulations were relatively short, 45 ns, and it is
possible that longer simulation times would show that the bilayers would also
be unstable at lower DMSO concentrations. The experimental work confirmed
many of the findings of the simulations, with the permeability of the bilayer
increasing as the DMSO thinned the bilayers.
Most previous simulations of DMSO with phospholipid bilayers have investigated bilayers consisting solely of a single phosphocholine (PC) lipid species.
The cell membranes of real biological samples contain a multitude of different
4
lipid species as well as proteins and sterols. The effect of sterols on the properties of cell membranes has been investigated experimentally in some detail.34–42
The presence of sterols makes a membrane more ordered and plays an important role in determining the fluidity and mechanical properties of a membrane.
Differential scanning calorimetry (DSC) of mixed DPPC-sterol bilayers showed
that the presence of sterols caused the transition between the gel and liquid
crystal phases to both broaden and weaken.34, 43 The phase transition temperature of the bilayer changes from ∼313-315 K for pure DPPC to 307-320 K for
a bilayer with a sterol concentration of 27 mol %. This effect increased with
the sterol concentration of the bilayers, leading to the complete elimination of
the phase transition at ∼33% sterol. These results lead to the conclusion that
bilayers containing certain sterol species (including cholesterol and β-sitosterol)
are able to form a liquid-ordered phase, lo , in which the lipids have nearly the
same translational and rotational freedom that they do in the liquid-disordered,
ld , (or liquid-crystalline) phase, but with a degree of conformational order comparable to the solid-ordered, so , (or gel) phase.38–40 More recently Kamal and
Raghunathan determined the phase behaviour of mixed DPPC-sterol bilayers
for a number of different sterols from X-ray diffraction data.42 Their findings
are largely consistent with the DSC studies. In bilayers containing 5-10 mol %
sterol the transition temperature remains roughly constant at ∼310 K but the
gel phase coexists with a sterol-induced modulated phase, Pβ . For bilayers with
sterol concentrations between 12.5 and 20.0 mol % the main transition temperature decreases with the increasing sterol concentration and the gel phase no
longer exists, with the bilayers below the transition temperature being in the
Pβ phase. Increasing the sterol concentration further to more than 25 mol %
leads to the eradication of any phase boundaries, with the system being in the
fluid phase for all temperatures between 283-318 K.
5
Sterols also have an effect on the permeability of bilayers, generally reducing the ability of water, ions and small organic molecules to cross the membrane.34–36, 44–46 The effectiveness of different sterol species at reducing permeability remains an open question and depends strongly on the species of lipids
present in the membrane.35–37 In animal cells cholesterol is the main sterol
species, while in plants β-sitosterol and stigmasterol are much more common
in combination with a higher degree of unsaturation in the acyl chains of the
phospholipids.34–37, 45 These differences can mean that a sterol species may
have little effect on the permeability of a membrane consisting of one lipid
species and yet have a significant effect on a membrane consisting of a different
species.34–37, 45 Sterols also play important roles in determining the elasticity of
a bilayer39 and the formation of lipid rafts within membranes.40, 41 The amount
of sterols that are present in membranes is difficult to determine accurately and
a wide range of values have been given, but a number of studies quote that in
animal cells cholesterol typically makes up 30 to 50% of the bilayer.34, 38, 41 In
plant cells the ratio of sterols to lipids is less well characterised and seems to
vary considerably.36
Several previous simulations43, 47–60 have investigated the effect of a variety
of different sterols on phospholipid bilayers, and two recent reviews provide a
good overview of the subject.61, 62 In nearly all simulations the condensing effect
of the sterols has been observed, with the addition of sterol molecules leading
to a decrease in the area per lipid (APL) of the lipids present in the bilayer. In
addition, the incorporation of sterols into the membrane leads to a thickening
of the bilayer. Due to the large number of variables that such simulations have
to take into account (e.g. amount of sterol, lipid species, sterol species, etc.),
comparisons between different simulations can be difficult. The most popular
sterol to simulate is cholesterol47, 49, 50, 55–58, 61, 63, 64 but other species have been
6
simulated.43, 48, 52, 62, 65 A comparison of cholesterol, ergosterol and lanosterol48
at 11% showed little structural difference in the bilayers of 1,2-dimyristoyl-snglycero-3-phosphocholine (DMPC). In contrast, a simulation52 of the same three
sterols at a concentration of 40% revealed significant differences between the
three sterols, with ergosterol packing closer to the lipids than cholesterol, which
in turn packed more closely than lanosterol. Systems containing plant sterols
have not been as widely investigated but a recent study of DPPC-stigmasterol
and DPPC-β-sitosterol membranes43 showed that plant sterols have the same
condensing effect on lipids as cholesterol. The study also found, in agreement
with experimental results, that β-sitosterol was more effective at ordering the
bilayer than stigmasterol. The chemical structures of cholesterol and β-sitosterol
are shown in Figure 1.
The range of lipids that have been investigated is also quite wide, with simulations having been performed on bilayers containing DMPC, DPPC, 2-oleoyl-1palmitoyl-sn-glycero-3-phosphocholine (POPC), DOPC and sphingomyelin (SM).47, 48, 50, 63, 64
The presence, and location, of an unsaturated double bond within the acyl chain
has been observed to cause significant structural differences in lipid-sterol bilayers,47, 55 with POPC being less able to pack around sterol molecules than DPPC.
This is consistent with conventional wisdom that the addition of unsaturated
bonds to acyl chains increases fluidity and lowers the transition temperature to
a much larger extent than the decrease in length of the acyl chains. A number
of studies49, 51 have observed the formation of small subunits consisting of one
sterol and two lipids or one sterol to one lipid. The ability of sterol-containing
membranes to resist the effect to electroporation showed that cholesterol increases the cohesion of the membrane and that, as the percentage of cholesterol
within the membrane was increased, the electric field needed for the formation of
a pore also increased.60 Although the majority of work has investigated binary
7
mixtures there is an increasing amount of simulations of ternary systems,53, 63, 65
where the preference of sterols for different lipid species can be evaluated. In
one study of mixed DOPC-SM-cholesterol bilayers it was found that cholesterol
tended to arrange itself so that the side of the sterol with the protruding groups
would lie next to DOPC molecules.63
The aims of this work are twofold. Firstly to determine if there are any significant differences between those membranes containing sterols typically found
in animals (cholesterol) with membranes containing sterols that are typically
found in plants (β-sitosterol). Secondly, to determine what effect the presence
of sterols within a bilayer has when the membrane is exposed to DMSO. A recent study evaluating the type and amount of sterols and phospholipids present
in three plant species found that the majority of the phosphocholine lipids had
acyl tails at least eighteen carbon atoms long and that the degree of unsaturation of the lipids was significant.66 The same study also found that in all three
species the most abundant sterol was β-sitosterol. Considering these findings
it was decided that DOPC would be used for the lipid species while cholesterol
and β-sitosterol would be used as the representative animal and plant sterol
species, respectively. As the amount of β-sitosterol within plant bilayers can
vary considerably a range of sterol concentrations were simulated.
Methods
The time- and length-scales required for these systems to be simulated fully
make them unsuitable for simulation with and all atom force-field. Instead
the lipids and sterols molecules were modelled using the united-atom GROMOS
53A6L forcefield,67–69 part of the GROMAS 54A7 force-field.70 While the use of
a united-atom force-field will mean that some detail is lost from the simulation,
8
lipid united-atom force-fields contain enough information that they are able to
provide an accurate description of lipid bilayers. The parameters for the 54A7
force-field have been fitted to allow the phospholipids to reproduce a broad
range of membrane properties.71, 72 The SPC model73 was used for the water
molecules, DMSO was represented using a united-atom with the parameters
taken from the GROMOS 53A6 and 54A7 force-fields.70, 74
The simulations were performed using GROMACS v.3.3.3.75 The equations
of motion were integrated using the leap-frog algorithm and a timestep of 2
fs was used.76 The non-bonded interactions were evaluated using twin-range
cutoffs, with interactions within 0.8 nm calculated every step while interactions
within 1.4 nm and the pair list were updated every five timesteps. The electrostatic interactions were computed using the reaction-field method with the
relative dielectric permittivity constant, εRF =62, suitable for SPC water.73 All
simulations were performed in the NpT ensemble, and the lipids/sterols and
solvent molecules were independently coupled to external temperature baths
with a coupling constant of 0.1 ps using the Berendsen thermostat.77 The pressure of the system was kept constant at 1 bar using the Berendsen barostat77
with a coupling constant, τP , of 1 ps and the isothermal compressibility of the
system was set to 4.6 × 10-5 bar-1 . For the simulations of the preformed bilayers, the box dimensions parallel and perpendicular to the plane of the bilayer
were independently coupled to a pressure bath, while in the case of the simulations starting from randomly distributed lipids/sterols (see below) each box
dimension was independently coupled to a pressure bath.
The initial starting configurations of the pure DOPC bilayers in water were
taken from previous work.27, 69 For the bilayers consisting of a mixture of lipids
and sterols the membranes were generated by randomly replacing a number
of DOPC lipids by sterol molecules (an equal number of sterol molecules were
9
placed in each leaflet). The sterol-lipid interactions were initially “soft” and were
gradually “hardened” to their proper values over a 50 ps run, thus removing any
unfavourable contacts. The systems were then simulated for 300 ns at both
303 and 350 K to ensure they were fully equilibrated. The final configurations
of the lipid-sterol systems in water were used as the starting configurations
for the simulations of the same bilayers in the aqueous DMSO solution. The
total number of solvent molecules was fixed at 5841, representing ∼46 solvent
molecules per lipid and/or sterol. The final 60 ns of each simulation were used
for analysis.
In addition to systems set up in a bilayer formation, a simulation of randomly
distributed lipids and sterols in water was performed in order to determine if
such systems would spontaneously form a bilayer and, if so, that the properties
of such a bilayer would be equivalent to the pre-formed systems. In this case 64
DOPC and 64 β-sitosterol molecules were solvated in 9045 water molecules. The
system was simulated at 303K for 400 ns with the final 60 ns of the simulation
used for analysis.
As mentioned above the primary motivation for this work is to provide a
better understanding of the effect that DMSO has on cell membranes, particularly plant cell membranes, during the cryopreservation process. In order to
make the model bilayers more representative of plant cell membranes66 a sterol
commonly found in plants, β-sitosterol, and an unsaturated lipid, DOPC, have
been used. For these β-sitosterol-containing bilayers, membranes consisting of
10, 33, 40 and 50 % β-sitosterol were simulated in both water and 2.5 mol %
DMSO solutions. The bilayers consisting of 10, 33 and 50% β-sitosterol were
also simulated in 25.0 mol % DMSO solution. As the primary aim of the work
is to investigate the effect of DMSO on lipid bilayers for the purpose of better
understanding the actions of this cryopreservant, and since 25.0 mol % (59 wt
10
%) DMSO is already a significantly more concentrated solution than is typically
used in cryopreservation protocols (5-15 wt %, <5 mol %), concentrations in
excess of 25.0 mol % were not simulated. DMSO is known to cause an increase
of the liquid crystalline-gel phase transition temperature of DPPC and POPE
membranes.30–32 Due to this behaviour many simulations of phospholipid bilayers in the presence of DMSO have been performed at 350 K in order to ensure
the bilayer is in the liquid crystalline phase. While to the authors knowledge
the effect of DMSO upon DOPC membranes has not been investigated experimentally, one would predict to see a similar effect, although the presence of
unsaturated acyl tails will lead to a lower phase transition temperature than
for DPPC. The addition of sterols to the bilayer increases the complexity of
the system further. As a result, in order to allow comparison with previous
simulations the simulations of the mixed DOPC-β-sitosterol with DMSO in this
study were also performed at 350K. The bilayer containing cholesterol was only
simulated in water. The details of the simulation runs are given in Table 1.
Results and Discusstion
Mixed DOPC/sterol bilayers in water
For a bilayer containing only DOPC the APL is predicted to be 0.653 nm2 and
0.683 nm2 at 303 and 350 K, respectively. At 303 K the experimental values
of the APL are in a range of 0.674-0.725 nm2 , showing that the predicted APL
is somewhat below this figure but is still reasonable, and falls within the range
(0.651-0.660 nm2 ) reported for other simulations of DOPC at 303 K.67, 68 The
presence of either cholesterol or β-sitosterol causes the cross-sectional area of
the bilayer to be reduced at both 303 and 350 K. Figure 2 shows the area of
11
the different bilayers over the last 100 ns of the simulations. The area of the
systems remains approximately constant and shows no drift, indicating that the
bilayers are equilibrated. It is apparent that the greater the amount of sterol in
the bilayer the greater the reduction in area. There are a number of different
methods that have been used to determine the area per lipid in a mixed lipidsterol system.49, 50, 53, 54, 64 In this work the method of Hofsäß has been used.50
Here the APL of a lipid is determined from
AL =
2A
NL
1−
NS VS
V − NW VW
(1)
where V is the volume of the system, A is the area of the bilayer in the plane of
the bilayer, NL , NS and NW are the total number of lipids, sterols and water
molecules in the system, respectively, and VS and VW are the volume of a sterol
molecule and a water molecule, respectively. The volume of a water molecule
was determined from simulations of a box of water and is 0.0309 nm3 at 303 K
and 0.0325 nm3 at 350 K. Previous simulations have estimated the volume of
a sterol molecule from the crystal structure; however, in this work a different
approach was taken, with the volume of the sterol molecules being determined
from the simulation of small systems consisting of 80 sterol molecules in 3680
water molecules. The volumes calculated using this approach are 0.715 nm3
and 0.674 nm3 for β-sitosterol and cholesterol, respectively, at 350 K. In a previous work52 the volume of cholesterol, averaged over three published crystal
structures, was given as 0.619 nm3 , which is somewhat lower than the value calculated above but probably not significantly so when the effects of temperature
and the different structures (crystal vs. hydrated sterols) are considered. The
APLs of the different systems are given in Table 2. As in previous simulations,
the presence of the sterols causes the APL to decrease. At 10% no difference
between cholesterol and β-sitosterol is observed, while at 50% cholesterol has
12
a slightly greater condensing effect. For β-sitosterol the APL decreases with
increasing sterol concentration up to the point where a third of the bilayer is
β-sitosterol, and above this point the APL is approximately constant. This behaviour occurs at both 303 and 350 K. The reason for this change in behaviour
is likely due to the fact that at concentrations above one sterol per two lipids,
sterol-sterol interactions start to occur and the phospholipids cannot solvate the
sterols completely, a behaviour also observed experimentally.34, 36
The thickness of the membrane normal to the plane of the bilayer, DHH
(calculated as the distance between the two peaks in the lipid density profile)
is given in Table 3. It is clear that the more sterol is added to the membrane
the greater the increase in the membrane thickness. The difference between the
two sterols was minor.
The density profiles of the bilayers are shown in Figure 3. It is clear that the
presence of sterol molecules in the bilayer makes the membrane more ordered
(this is especially true when the sterol concentration is 50%), the interdigitation
of the acyl chains of the two leaflets of the bilayer is also reduced, causing the
trough in the density profile at the intersection of the two leaflets to become
deeper. The density profiles also show that both cholesterol and β-sitosterol
take up a position in the membrane below that of the DOPC headgroups. The
headgroups of the phospholipids form an “umbrella” over the hydroxyl group
of the sterol, shown in Figure 4. The lipid density profile of the random 50%
β-sitosterol system does not shown any significant differences to that of the 50%
β-sitosterol system starting from a bilayer, but the sterol density profile is more
smoothed out, indicating that the sterol molecules are not quite as well ordered.
The ordering of the acyl tails of the phospholipids can be determined from
the deuterium order parameter, SCD , which provides a measure of the relative orientation of C-D bonds with respect to the bilayer normal. SCD can be
13
calculated from
SCD =
1
< 3 cos2 θ − 1 >
2
(2)
where θ is the angle between a C-D of a methylene in the chain and the bilayer
normal. As the force-field used in this work is a united-atom one, the positions
of the deuteriums are derived from the positions of the neighbouring carbons
and assuming a tetrahedral geometry of the methylenes. The calculated order
parameters for both the sn-1 and sn-2 chains of the DOPC lipids for the different
bilayer compositions are shown in Figure 5. The presence of an unsaturated
bond in the acyl chains of DOPC is responsible for the drop in order at carbons 9
and 10. The addition of sterols into the bilayer forces the acyl chains to become
more ordered. For the bilayers consisting of 10 mol % sterol the increased
ordering is relatively minor but at 50 mol % sterol the acyl tail ordering is
substantial so that the order profile looks rather like that which would be found
for a membrane in the gel-phase. At higher temperatures the ordering of the
lipids is reduced due to the higher thermal motion present in the system. It
might be expected that the bulkier tail group of β-sitosterol would be responsible
for a difference in the ordering at the end of the lipid tails. However, cholesterol
and β-sitosterol show very similar order parameters for carbon atoms at C10 and
above. Instead, for the bilayers which are half sterol in composition the ordering
of the first eight carbon atoms in the DOPC tails is greater for cholesterol. The
order parameters of the S50R system are slightly lower than for the S50 system,
indicating that while the initial random distribution of lipids and sterols leads
to the formation of a bilayer, it is not quite as well ordered as a system that is
initially set up in a bilayer. However, the difference between the two systems is
minor, showing that setting up the systems in a bilayer configuration does not
lead to an unrealistic configuration.
The tilt angle of a sterol is defined as the angle between the bilayer normal,
14
directed away from the bilayer, and the vector formed between two atoms within
the sterol. For the sterol head tilt angle the vector is defined between the carbon
atoms at the top (C3 ) and bottom (C17 ) of the ring structure, and for the tail
tilt angle the vector is defined between the bottom carbon of the ring and the
second to last carbon in the tail of the sterol (C25 ). The distributions of the
tilt angles between the sterol and the bilayer normal are given in Figure 6. For
the systems containing 10% sterol the sterol molecules tend to lie at a slight
angle to the bilayer normal, and no significant difference between cholesterol
and β-sitosterol is observed. As the amount of sterol in the system increases the
distribution is shifted to lower angles, suggesting that the sterol molecules are
aligning themselves nearly parallel to the membrane normal. At 50% sterol the
differences between cholesterol and β-sitosterol become more apparent. While
the peak in the distribution of the head tilt angle is the same for both species,
the peak tail distribution of cholesterol is lower than that of β-sitosterol. This
difference arises due to the bulkier tail of β-sitosterol. A higher temperature
causes the distribution to widen but the position of the peaks does not change.
In the case of S50R the distribution of the tilt angles is slightly shifted to higher
angles when compared to S50, due to the fact that the bilayer is not quite as
well ordered.
It is also possible to calculate the tilt angle between the bilayer normal
and the vector between the P- →N+ dipole moment, ζPN , which indicates the
orientation of the lipid headgroups. The distribution of ζPN is shown in Figure
7. For the systems at 303 K the presence of sterols causes some significant
changes in the distribution. For bilayers that are 33% sterol or less the peak of
the distribution is shifted to slightly lower angles. For bilayers with a 1:1 ratio
of lipids and sterols the peak of the angle distribution is shifted to higher angles
and the distribution narrows slightly. The reason for this variable behaviour
15
is explained in part by Figure 4. As mentioned previously, the headgroups of
DOPC act as “umbrella” for a sterol molecule, resulting in the P- →N+ dipole
moment of the DOPC molecule acting as an umbrella being almost parallel to
the plane of the bilayer. At high sterol concentrations the majority of DOPC
molecules act as umbrellas, resulting in the angle distribution being shifted
towards 90◦ . At lower sterol concentrations only a few lipid molecules will be
acting as umbrellas, but the reduced APL of the lipids result in the headgroups
of the remaining lipids tending to lie more out of the plane of the bilayer, thus
reducing the peak of the angle distribution.
DOPC/β-sitosterol bilayers in DMSO
In the presence of DMSO a pure DOPC bilayer expands laterally, while thinning
normal to the plane of the bilayer. If the concentration of DMSO is high enough
(∼15-20 mol %) then pores spontaneously form in the bilayer structure. These
pores are stable on the timescales reached by molecular simulations.27 However,
at a 25 mol % DMSO the pores become so large that the structure of the
membrane is weakened to such extent that the bilayer falls apart.
In the case of mixed DOPC-β-sitosterol bilayers the presence of DMSO
causes the bilayers to expanded laterally to the plane of the bilayer. The lateral
area of the membrane, the relative increase in the lateral area and the thickness of the membrane are given in Table 4. This matches the behaviour of 20%
cholesterol bilayers.33 At low concentrations (2.5 mol %) the relative increase in
the lateral area of the bilayer appears to be largely independent of the amount
of β-sitosterol in the membrane. The S10 and S33 systems show a similar ∼9%
increase in the area of the membrane compared to that of the pure DOPC bilayer. For the S40 and S50 bilayers the increase is slightly lower and the degree
16
of membrane thinning is likewise reduced slightly. At DMSO concentrations of
25 mol % the effect of the sterol presence in the bilayer is seen more clearly.
As mentioned above, a solution of 25 mol % DMSO will cause a pure DOPC
bilayer to be destroyed. In contrast, the bilayers containing β-sitosterol show
greater resistance to the effects of DMSO. In the case of S10 the bilayer thins
to such extent that a pore is able to form in the bilayer but the system is stable on the 300 ns timescale simulated. The S33 and S50 systems experience
significant thinning and are accompanied by a 39 and 30 % lateral expansion,
respectively, but no pores are formed in the bilayers (for comparison the relative
increase in the lateral area of a pure DOPC bilayer in 12.5 mol % DMSO is 32
%). De Ménorval et al. found that 15.0 mol % DMSO was a sufficiently high
concentration to cause the formation of pores in a 80/20% DOPC/cholesterol
bilayer.33 Due to the different force-fields and temperatures used, it is difficult
to compare that result with our results directly but the findings of both studies
are consistent.
The density profiles of the membranes are shown in Figure 8, and snapshots
of the S10 and S50 systems are shown in Figure 9. For the 2.5 mol % DMSO
solution there is a build-up of DMSO at the surface of the bilayer. The density
profiles of the lipids and sterols are also affected by the presence of DMSO,
with the lipid density profiles becoming less well defined, indicating that the
membrane is becoming less well ordered. The sterol density profile not only
becomes less well defined, with peaks tending to merge together, but also show a
greater asymmetry than the profiles of the same systems in water. The reason for
the greater asymmetry of the profiles is due to the presence of DMSO increasing
the tendency of sterol molecules to move from one leaflet to the other, resulting
in the possibility that the number of sterol molecules in each leaflet may no
longer be equal. While such behaviour is also seen in systems where no DMSO
17
is present, it occurs only rarely, as shown in Table 5, and does not occur at
303 K. The mechanism by which the sterol molecules move between leaflets
involves the tail of the sterol molecule remaining in the centre of the bilayer
while the head moves downwards so that the sterol molecule lies parallel to
the plane of the bilayer, leading to the head diffusing into the other leaflet.
Overall the molecule “flips” from one leaflet to the other, as shown in Figure
10. These flips are generally quite fast taking only a few nanoseconds; however,
a β-sitosterol molecule occasionally becomes trapped for a while between the
two leaflets and may remain there in a horizontal position for some time before
moving into a leaflet. The majority of flips occur between the two leaflets but
in a small minority of cases a sterol molecule will move from a leaflet to the
centre of the bilayer before reinserting itself into the leaflet it came from but
at a different place. This mechanism is in agreement with the results of both
atomistic54, 59 and course-grained simulations of PC-cholesterol bilayers.56–58
However, it is inconsistent with the pathway for sterol flipping determined from
a recent theoretical study which used an implicit description of the (DPPC)
membrane,78 where it was proposed that the rotation and translation of the
sterol molecules occurred in a two-stage process, and that the movement of a
sterol molecule between the leaflets would be perpendicular to the plane of the
bilayer. The explanation of the inconsistency between theory and simulation
could be due to a number of factors, in particular the sterol concentration (the
theoretical work was done for membranes at very low sterol concentrations) and
the degree of unsaturation of the phospholipid. Another coarse-grained study
showed that increasing unsaturation increased the residence time of cholesterol
in the membrane interior during flip-flops.56 For the 25.0 mol % DMSO solution
the weakening of the membrane becomes more apparent. For the S10 system the
formation of a pore is clearly illustrated by the density profile of water becoming
18
non-zero at the centre of the bilayer. In the case of S50, while the membrane
has become significantly thinned and disordered, the characteristic peak and
trough in the lipid density profile is still present. In 25.0 mol % DMSO solution
the rate of sterol flips increases significantly, and allows the observation that as
the sterol concentration in the membrane increases the relative number of flips
decreases. In the case of the S10 system the formation of the pore means that
in addition to the flipping of sterol molecules between the leaflets the DOPC
molecules are also able to flip, this behaviour is consistent with the flipping of
phospholipids in bilayers containing no sterols in which a pore has formed.26
As in the case of pure DOPC bilayers, the lateral expansion of the membrane
reduces the ordering of the lipid tails, as shown in Figure 11. The presence of
the sterol molecules again counteracts the effect of DMSO to some extent. This
is more apparent at 25.0 mol % DMSO where, although the ordering has been
reduced substantially, the sterol molecules still manage to keep the lipid tails
ordered to some extent. Part of the reason for this decrease in the order of the
acyl tails is that the increased undulations of a bilayer solvated in a solution of
DMSO result in the close interactions of the sterol molecules with the DOPC
molecules being disturbed. As discussed above, in water DOPC molecules can
act as umbrellas to the sterol molecules, which imposes an ordering effect on the
acyl tails of the lipids. As DMSO disorders the bilayer, the interaction of the
DOPC with the β-sitosterol is hindered and the umbrella structure is weakened,
which in turn weakens the ability of the β-sitosterol molecules to order the acyl
chains. The disordering of the membrane also affects the sterols. Figure 12
shows the distribution of the tilt angles of the sterols for the bilayers solvated
in DMSO. The distributions of the tilt angles have become noticeably broader
due to the fact that the sterol molecules tend to lie at more of an angle to the
bilayer.
19
In pure DOPC bilayers DMSO molecules are able to diffuse across the bilayer relatively easily. In 2.5 mol % DMSO solution the free energy barrier
for a molecule to diffuse across the bilayer is ≈14 kJ mol-1 .27 For the bilayers
containing β-sitosterol DMSO molecules are still observed crossing the bilayer
(even for a bilayer containing 50% sterol), but the frequency of such crossings
is reduced significantly for sterol concentrations greater than 10%, as shown in
Table 6. In addition to the sterols making it more difficult for DMSO to cross
the bilayer, they also change the mechanism of diffusion of DMSO through the
membrane.
For a pure DOPC bilayer the maximum of the free energy barrier for a DMSO
molecule to cross through the bilayer is located at the centre of the bilayer, where
the hydrophobic tails of the lipids are. This is also true for bilayers with only
a small amount of sterol; however, for bilayers with a high sterol concentration,
DMSO molecules that cross the bilayer can be seen to spend some time in the
centre of the membrane, (as shown in Figure 13) indicating that the centre of
the bilayer can no longer be at the maximum in the free energy profile. The
reason for this is that, as the sterol concentration increases, the interdigitation
between the tails of the lipid and the sterol molecules in the two leaflets is
reduced substantially. This means that the centre of the bilayer becomes a local
minimum between two maxima where the density of the hydrophobic tails is
highest.
In addition to the simulations of single bilayers in solutions of DMSO, simulations of systems containing two bilayers were performed. In these simulations
a DMSO concentration gradient was established by having one of the chambers
(A) separating the bilayers contain a solution of 2.5 mol % DMSO and the other
chamber (B) contain only water. The full details of the setup and findings have
been described in detail previously for pure DOPC bilayers.27 Double bilayer
20
simulations of 2:1 DOPC-β-sitosterol and 1:1 DOPC-β-sitosterol systems have
been performed.
Figure 14 shows the number of DMSO and water molecules present in chamber A over 300 ns. On the timescale simulated none of the systems reached equilibrium but the diffusion of molecules across the bilayers was observed. For all
three systems, pure DOPC, 33% and 50% β-sitosterol, there is a flow of DMSO
molecules from chamber A to chamber B and the flow of water molecules occurs in the opposite direction. The presence of sterol molecules in the bilayer
strongly reduces the rate of the flow of both DMSO and water molecules: the
greater the sterol concentration in the bilayer the greater the reduction in the
rate of flow.
These simulations show that DOPC bilayers containing sterols are better
able to withstand the deleterious effects of DMSO. Experimentally it has been
shown that subjecting plants to cold temperatures can increase the sterol content
of their cell membranes.66 Thus, our work helps to explain how cold-acclimation
may assist in reducing the damage to the tissues caused by cryosolutions during cryopreservation through the changes that occur to the composition of cell
membranes.
Conclusion
Simulations of mixed DOPC-sterol bilayers have been performed at different
sterol concentrations, different temperatures and a variety of DMSO concentrations. Comparison of bilayers containing β-sitosterol with those containing
cholesterol showed that the differences between such systems were minor. In
both cases the findings of previous phospholipid-sterol simulations - that sterols
cause the condensing of the phospholipids - were confirmed. The APL of DOPC-
21
β-sitosterol decreases with increasing sterol concentration, up to the point where
one third of the bilayer is β-sitosterol, with the APL remaining constant above
this point. The presence of sterols in the bilayer also causes the membrane to
thicken and leads to an increase in the ordering of the lipid acyl-chains. While
increasing the temperature caused the APL to increase, DHH to decrease and
SCD to decrease, no phase change was observed for any of the systems simulated. The diffusion of sterol molecules from one leaflet to the other was affected
when the bilayers were simulated at 350 K. At 303 K no sterol molecules were
observed to “flip” between leaflets on the timescales simulated. At 350 K, while
such flipping was still a rare event (∼1 flip in 100 ns), it was observed.
At low DMSO concentrations the mixed bilayers behaved in a similar manner
to pure DOPC bilayers, with the bilayer expanding laterally while concurrently
thinning transversely. The relative increase in bilayer area was approximately
the same for bilayers with ≥ 33% β-sitosterol as for pure DOPC bilayers. Above
this point the relative increase in APL was reduced for the sterol-containing
bilayers compared with the pure DOPC bilayers. The build up of DMSO at
the surface of the bilayer caused the membrane structure to become less well
defined. At high DMSO concentrations (25.0 mol %) the pure DOPC membrane
are destroyed. By contrast, even a small amount of β-sitosterol (10%) provided
enough rigidity such that the bilayer was able to maintain its integrity, although
a pore forms in the bilayer. Increasing the amount of β-sitosterol stopped the
formation of pores (on the 300 ns timescale simulated) and significantly reduced
the deleterious effects of DMSO upon the bilayers.
The lateral expansion of the bilayer causes the acyl-tails of the lipids to
become more disordered and the β-sitosterol molecules to lie at more of an angle
to the normal of the bilayer. The presence of DMSO also caused an increase
in the number of sterol flips. As in the case of pure DOPC bilayers, DMSO
22
molecules are observed to diffuse across the bilayers, but at a much lower rate,
indicating that the energy barrier to diffusion increases with the amount of sterol
present within the bilayer. Despite the same relative increase in the APL for
2:1 DOPC-β-sitosterol bilayers and pure DOPC bilayers, the rate of diffusion of
DMSO was significantly reduced in the former compared to the latter, suggesting
that “looseness” of the bilayer does not play a major role in determining the
height of the energy barrier to the diffusion of molecules. The presence of βsitosterol not only appeared to increase the energy barrier for DMSO molecules
to cross the bilayer but also modifies the mechanism of diffusion of DMSO
across the bilayer. In pure DOPC or bilayers with 10% β-sitosterol the DMSO
molecules cross the hydrophobic tail region of the bilayer (where the maximum
of the energy barrier is) in a single jump. At β-sitosterol concentrations of ≥ 33%
the DMSO molecules diffuse across each leaflet in a separate jump. This change
in behaviour is due to the fact that as the concentration of β-sitosterol increases
the interdigitation of the two leaflets is reduced. Double bilayer simulations
confirmed that the rate of diffusion of both DMSO and water molecules is not
only reduced at equilibrium but also when a concentration gradient is present
in the system.
The findings of this work also reveal how the cold-acclimation of plants
may reduce the damage that DMSO causes to cells during cryopreservation, by
increasing the sterol concentration of cell membranes and thus making these
lipid bilayers more resistant to the effects of DMSO.
Acknowledgements
The authors would like to thank E. Bunn, A. Kaczmarczyk and B. Funnekotter
at the Botanic Gardens and Parks Authority (BGPA) in WA and D. Poger and
23
A.E. Mark at the University of Queensland (UQ) for valuable discussions. This
research was supported by Australian Research Council (ARC) Linkage Grant
LP0884027 and with the further financial support of Alcoa of Australia and
BHP Billiton Worsley Alumina. The computational resources required for the
work were provided by iVEC.
References
[1] Meryman, H. T.; Williams, R. J.; Douglas, M. S. J. Cryobiology 1977, 14,
287–302.
[2] Wusteman, M.; Robinson, M.; Pegg, D. Cryobiology 2004, 48, 179–189.
[3] Engelmann, F. In Vitro Cell Dev. Biol. 2004, 40, 427–433.
[4] Day, J. G.; Harding, K. C.; Nadarajan, J.; Benson, E. E. Mol. Biomethods
Handb. 2008, 917–947.
[5] Kaczmarczyk, A.; Rokka, V.-M.; Keller, E. R. J. Potato Res. 2011, 54,
45–79.
[6] Kaczmarczyk, A.; Turner, S. R.; Bunn, E.; Mancera, R. L.; Dixon, K. W.
In Vitro Cell Dev. Biol. 2011, 47, 17–25.
[7] Kaczmarczyk, A.; Funnekotter, B.; Menon, A.; Phang, P. Y.; AlHanbali, A.; Bunn, E.; Mancera, R. L. In Current Issues in Plant Cryopreservation; Katkov, I. I., Ed.; InTech Open Access Publisher, 2012.
[8] Sakai, A.; Kobayashi, S.; Oiyama, I. Plant Cell Rep. 1990, 9, 30–33.
[9] Kim, H. H.; Cho, E. G.; Baek, H. J.; Kim, C. Y.; Keller, E. R. J.; Engelmann, F. Cryoletters 2004, 25, 59–70.
[10] Benson, E. E.; Harding, K.; Johnston, J. W. Methods In Molecular Biology
2007, 368, 163–183.
[11] Sakai, A.; Engelmann, F. Cryoletters 2007, 28, 151–172.
[12] Benson, E. E. Crit. Rev. Plant Sci. 2008, 27, 141–219.
[13] Lovelock, J. E.; Bishop, M. W. H. Nature 1959, 183, 1394–1395.
[14] Wolfe, J.; Bryant, G. Int J Refrig 2001, 24, 438–450.
[15] Anchordoguy, T. J.; Carpenter, J. F.; Crowe, J. H.; Crowe, L. M. Biochim.
Biophys. Acta 1992, 1104, 117–122.
24
[16] Mandumpal, J. B.; Kreck, C. A.; Mancera, R. L. Phys. Chem. Chem. Phys.
2011, 13, 3839–3842.
[17] Kreck, C. A.; Mandumpal, J. B.; Mancera, R. L. Chem. Phys. Lett. 2011,
501, 273–277.
[18] Ahkong, Q. F.; Fisher, D.; Tampion, W.; Lucy, J. A. Nature 1975, 253,
194–195.
[19] Colucci, M.; Maione, F.; Bonito, M. C.; Piscopo, A.; Giannuario, A. D.;
Pieretti, S. Pharmacol. Res. 2008, 57, 419–425.
[20] Haigler, H. J.; Spring, D. D. Ann. Ny. Acad. Sci. 1983, 411, 19–27.
[21] Sum, A. K.; de Pablo, J. J. Biophys. J. 2003, 85, 3636–3645.
[22] Notman, R.; Noro, M.; O’Malley, B.; Anwar, J. J. Am. Chem. Soc. 2006,
128, 13982–13983.
[23] Gurtovenko, A. A.; Anwar, J. J. Phys. Chem. B 2007, 111, 13379–13382.
[24] Gurtovenko, A. A.; Anwar, J. J. Phys. Chem. B 2007, 111, 10453–10460.
[25] Notman, R.; den Otter, W. K.; Noro, M. G.; Briels, W. J.; Anwar, J.
Biophys. J. 2007, 93, 2056–2068.
[26] Gurtovenko, A. A.; Onike, O. I.; Anwar, J. Langmuir 2008, 24, 9656–9660.
[27] Hughes, Z. E.; Mark, A. E.; Mancera, R. L. J. Phys. Chem. B 2012, 116,
11911–11923.
[28] Lin, J.; Novak, B.; Moldovan, D. J. Phys. Chem. B 2012, 116, 1299–1308.
[29] Notman, R.; Anwar, J. Adv. Drug Deliv. Rev. 2012.
[30] Shashkov, S. N.; Kiselev, M. A.; Tioutiounnikov, S. N.; Kiselev, A. M.;
Lesieur, P. Physica B 1999, 271, 184–191.
[31] Kiselev, M. A.; Lesieur, P.; Kisselev, A. M.; Grabielle-Madelmond, C.;
Ollivon, M. J. Alloy Compd. 1999, 286, 195–202.
[32] Yamashita, Y.; Kinoshita, K.; Yamazaki, M. Biochim. Biophys. Acta,
Biomembr. 2000, 1467, 395–405.
[33] de Menorval, M.-A.; Mir, L. M.; Fernandez, M. L.; Reigada, R. PLoS ONE
2012, 7, e41733.
[34] McKersie, B. D.; Thompson, J. E. Plant Physiol. 1979, 63, 802–805.
[35] Schuler, I.; Duportail, G.; Glasser, N.; Benveniste, P.; Hartmann, M. A.
Biochim. Biophys. Acta 1990, 1028, 82–88.
25
[36] Schuler, I.; Milon, A.; Nakatani, Y.; Ourisson, G.; Albrecht, A. M.; Benveniste, P.; Hartmann, M. A. PNAS 1991, 88, 6926–6930.
[37] Hartmann, M. A. Trends Plant Sci. 1998, 3, 170–175.
[38] Nielsen, M.; Thewalt, J.; Miao, L.; Ipsen, J. H.; Bloom, M.; Zuckermann, M. J.; Mouritsen, O. G. Europhys. Lett. 2000, 52, 368–374.
[39] Mouritsen, O. G.; Zuckermann, M. J. Lipids 2004, 39, 1101–1113.
[40] Simons, K.; Vaz, W. L. C. Annu. Rev. Biophys. Biomol. Struct. 2004, 33,
269–295.
[41] Dufourc, E. J. J. Chem. Biol. 2008, 1, 63–77.
[42] Kamal, M. A.; Raghunathan, V. A. Soft Matter 2012, 8, 8952–8958.
[43] Silva, C.; Aranda, F. J.; Ortiz, A.; Martinez, V.; Carvajal, M.; Teruel, J. A.
J. Colloid Interf. Sci. 2011, 358, 192–201.
[44] Grunwald, C. Plant Physiol. 1974, 54, 624–628.
[45] Krajewski-Bertrand, M. A.; Milon, A.; Hartmann, M. A. Chem. Phys.
Lipids 1992, 63, 235–241.
[46] Haines, T. H. Prog. Lipid Res. 2001, 40, 299–324.
[47] Chiu, S. W.; Jakobsson, E.; Scott, H. L. Biophys. J. 2001, 80, 1104–1114.
[48] Smondyrev, A. M.; Berkowitz, M. L. Biophys. J. 2001, 80, 1649–1658.
[49] Chiu, S. W.; Jakobsson, E.; Mashl, R. J.; Scott, H. L. Biophys. J. 2002,
83, 1842–1853.
[50] Hofsass, C.; Lindahl, E.; Edholm, O. Biophys. J. 2003, 84, 2192–2206.
[51] Pandit, S. A.; Bostick, D.; Berkowitz, M. L. Biophys. J. 2004, 86, 1345–
1356.
[52] Cournia, Z.; Ullmann, G. M.; Smith, J. C. J Phys Chem B 2007, 111,
1786–1801.
[53] Cheng, M. H.; Liu, L. T.; Saladino, A. C.; Xu, Y.; Tang, P. J. Phys. Chem.
B 2007, 111, 14186–14192.
[54] Rog, T.; Stimson, L. M.; Pasenkiewicz-Gierula, M.; Vattulainen, I.; Karttunen, M. J. Phys. Chem. B 2008, 112, 1946–1952.
[55] Martinez-Seara, H.; Rog, T.; Pasenkiewicz-Gierula, M.; Vattulainen, I.;
Karttunen, M.; Reigada, R. Biophys. J. 2008, 95, 3295–3305.
[56] Marrink, S. J.; de Vries, A. H.; Harroun, T. A.; Katsaras, J.; Wassall, S. R.
J. Am. Chem. Soc. 2008, 130, 10–11.
26
[57] Bennett, W. F. D.; MacCallum, J. L.; Hinner, M. J.; Marrink, S. J.; Tieleman, D. P. J. Am. Chem. Soc. 2009, 131, 12714–12720.
[58] Bennett, W. F. D.; MacCallum, J. L.; Tieleman, D. P. J. Am. Chem. Soc.
2009, 131, 1972–1978.
[59] Jo, S.; Rui, H.; Lim, J. B.; Klauda, J. B.; Im, W. J. Phys. Chem. B 2010,
114, 13342–13348.
[60] Fernandez, M. L.; Marshall, G.; Sagues, F.; Reigada, R. J. Phys. Chem. B
2010, 114, 6855–6865.
[61] Rog, T.; Pasenkiewicz-Gierula, M.; Vattulainen, I.; Karttunen, M. Biochim.
Biophys. Acta Biomembr. 2009, 1788, 97–121.
[62] Lyubartsev, A. P.; Rabinovich, A. L. Soft Matter 2011, 7, 25–39.
[63] Pandit, S. A.; Jakobsson, E.; Scott, H. L. Biophys. J. 2004, 87, 3312–3322.
[64] Pandit, S. A.; Vasudevan, S.; Chiu, S. W.; Mashl, R. J.; Jakobsson, E.;
Scott, H. L. Biophys. J. 2004, 87, 1092–1100.
[65] Bennett, W. F. D.; Tieleman, D. P. J. Lipid Res. 2012, 53, 421–429.
[66] Funnekotter, B.; Kaczmarczyk, A.; Turner, S. R.; Bunn, E.; Zhou, W.;
Smith, S.; Flematti, G.; Mancera, R. L. Submitted for publication.
[67] Poger, D.; van Gunsteren, W. F.; Mark, A. E. J. Comput. Chem. 2010,
31, 1117–1125.
[68] Poger, D.; Mark, A. E. J. Chem. Theory Comput. 2010, 6, 325–336.
[69] Malde, A. K.; Zuo, L.; Breeze, M.; Stroet, M.; Poger, D.; Nair, P. C.;
Oostenbrink, C.; Mark, A. E. J. Chem. Theory Comput. 2011, 7, 4026–
4037.
[70] Schmid, N.; Eichenberger, A. P.; Choutko, A.; Riniker, S.; Winger, M.;
Mark, A. E.; van Gunsteren, W. F. Eur. Biophys. J. 2011, 40, 843–856.
[71] Piggot, T. J.; Pineiro, A.; Khalid, S. J. Chem. Theory Comput. 2012, 8,
4593–4609.
[72] Poger, D.; Mark, A. E. J. Chem. Theory Comput. 2012, 8, 4807–4817.
[73] Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J.
In Intermolecular Forces; Dordrecht, Reidel, 1981.
[74] Oostenbrink, C.; Villa, A.; Mark, A. E.; van Gunsteren, W. F. J. Comput.
Chem. 2004, 25, 1656–1676.
[75] van der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.; Berendsen, H. J. C. J. Comput. Chem. 2005, 26, 1701–1718.
27
[76] Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Oxford
University Press, New York, 1989.
[77] Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Dinola, A.;
Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690.
[78] Parisio, G.; Sperotto, M. M.; Ferrarini, A. J. Am. Chem. Soc. 2012, 134,
12198–12208.
28
System
DOPC
S10
S33
S40
S50
S50R
C10
C50
Sterol
None
β-sitosterol
β-sitosterol
β-sitosterol
β-sitosterol
β-sitosterol
cholesterol
cholesterol
% sterol
0
10
33
40
50
50
10
50
# Lipids
128
116
86
78
64
64
116
64
# Sterols
0
12
42
50
64
64
12
64
mol. % DMSO
0, 2.5, 25.0
0, 2.5, 25.0
0, 2.5, 25.0
0, 2.5
0, 2.5, 25.0
0
0
0
Table 1: Details of the different systems simulated.
29
System
DOPC
S10
S33
S40
S50
S50R
C10
C50
Area per lipid / nm2
303K
350K
0.653 ± 0.010 0.683 ± 0.008
0.622 ± 0.008 0.659 ± 0.011
0.587 ± 0.008 0.603 ± 0.009
0.594 ± 0.007 0.596 ± 0.009
0.565 ± 0.007 0.594 ± 0.009
0.566 ± 0.007
0.619 ± 0.014 0.683 ± 0.010
0.554 ± 0.005 0.588 ± 0.011
Table 2: Area per lipid for the different systems at 303K and 350K.
30
System
DOPC
S10
S33
S40
S50
S50R
C10
C50
DHH / nm
303K 350K
3.14
3.11
3.37
3.22
3.83
3.62
3.97
3.82
3.98
3.92
4.00
3.38
3.37
3.99
3.98
Table 3: Membrane thickness, DHH , determined from the distance between the
peaks of the lipid density profile.
31
System
DOPC
S10
S33
S40
S50
2.5 mol % DMSO
Area /nm2 % Area % DHH
47.8 ± 1.0
109
95
44.2 ± 0.8
109
96
36.0 ± 0.6
109
97
33.5 ± 0.4
106
98
31.1 ± 0.4
105
98
25.0 mol % DMSO
Area /nm2 % Area % DHH
Destroyed
Pore
45.9 ± 1.3
139
65
38.4 ± 0.9
130
80
Table 4: Lateral area of and the relative area and thickness of DOPC-β-sitosterol
bilayers in the presence of DMSO compared against the same bilayers in water.
32
System
S10
S33
S40
S50
C10
C50
Water
(303 K)
0
0
0
0
0
0
Water
(350 K)
2
3
4
3
3
3
Number of sterol flips
2.5 mol % DMSO 25.0 mol % DMSO
(350 K)
(350 K)
2
18
8
47
6
4
20
-
Table 5: Number of sterol flips for the different bilayers under the different
simulation conditions.
33
System
DOPC
S10
S33
S40
S50
DMSO concentration /mol %
2.5
25.0
14
Destroyed
14
Pore
7
93
2
3
34
Table 6: Number of DMSO molecules that cross the bilayer in a 60 ns period.
34
Figure Captions
Figure 1:The molecular structure of (a) β-sitosterol and (b) cholesterol.
Figure 2:The area of the bilayers at 303K over the last 100 ns of the simulations.
Figure 3: Density profiles for the different bilayer systems (a) S10, (b) S33, (c)
S50, (d) S50R, (e) C10 and (f) C50.
Figure 4: Snapshot of two DOPC molecules forming an “umbrella” over a
β-sitosterol molecule, taken from the simulation of the S33 bilayer at 350K.
Figure 5: Order parameters of DOPC molecules in the bilayers (a) sn-1 chain
at 303K, (b) sn-2 chain at 303K, (c) sn-1 chain at 350K and (d) sn-2 chain at
350K.
Figure 6: Tilt angles of the sterol molecules with the plane of the bilayer. (a)
sterol head at 303K, (b) sterol head at 350K, (c) sterol tail at 303K and (d)
sterol tail at 350K.
Figure 7: The probability distribution function of the angle between the bilayer
normal, directed away from the middle of the bilayer, and the vector between
the P- →N+ atoms in the lipid headgroup for simulations of the DOPC molecules
at (a) 303K and (b) 350K in water.
Figure 8: Density profiles for the bilayer systems (a) DOPC, (b) S10, (c) S33,
(d) S50 at 350K and 2.5 mol % DMSO, (e) S10 and (f) S50 at 25.0 % mol
DMSO.
Figure 9: Snapshots of mixed lipid-sterol bilayers in DMSO, (a) S10 in 2.5
mol % DMSO, (b) S10 in 25.0 mol % DMSO, (c) S50 in 2.5 mol % DMSO and
(d) S50 in 25.0 mol % DMSO. The DOPC phosphate headgroups, DOPC tails,
β-sitosterol, water and DMSO are coloured red, grey, green, blue and yellow,
respectively.
Figure 10: Snapshots from the S10 system at 350K showing how sterol molecules
flip from one leaflet to the opposite, taken at (a) 91.6, (b) 94.6, (c) 95.3, (d)
96.1, (e) 96.6 and (f) 100.0 ns. The DOPC/β-sitosterol are coloured grey, the
water molecules, phosphate headgroups and β-sitosterol molecule undergoing
the “flip” are coloured blue, purple/red and yellow, respectively.
Figure 11: Order parameters of DOPC molecules in the bilayers (a) sn-1 chain
in 2.5 mol % DMSO, (b) sn-2 chain in 2.5 mol % DMSO, (c) sn-1 chain in 2.5
mol % DMSO and (d) sn-2 chain in 2.5 mol % DMSO.
35
Figure 12: Tilt angles of the sterol molecules with the plane of the bilayer (a)
sterol head in 2.5 mol % DMSO, (b) sterol head in 25.0 mol % DMSO, (c) sterol
tail in 2.5 mol % DMSO and (d) sterol tail in 25.0 mol % DMSO.
Figure 13: The trajectories along the z-axis (normal to the bilayers) of DMSO
molecules that diffuse through the S50 bilayer (shown by the blue and green
lines), taken from 40 ns simulation of a system at a DMSO concentration of 2.5
mol % at 350K. The red lines indicate the positions of the phosphate groups.
Figure 14: The number of (a) DMSO and (b) water molecules in chamber A
(the high DMSO concentration chamber) of the double bilayer simulations as a
function of simulation time.
36
Figure 1: The molecular structure of (a) β-sitosterol and (b) cholesterol.
37
40
Area of bilayer / nm2
38
36
10% β-sitosterol
33% β-sitosterol
40% β-sitosterol
50% β-sitosterol
10% Cholesterol
50% Cholesterol
34
32
30
28
26
200
220
240
260
280
Time / ns
Figure 2: The area of the bilayers at 303K over the last 100 ns of the simulations.
38
300
1200
1200
DOPC
β-sitosterol
Water
Phosphate
1000
800
800
Density / kg m-3
Density / kg m
-3
1000
600
400
600
400
200
0
200
-4
-2
0
2
4
Z coordinate / nm
0
1200
Density / kg m-3
Density / kg m-3
600
400
-4
-2
0
2
4
2
4
2
4
DOPC
β-sitosterol
Water
Phosphate
400
0
-4
-2
0
Z coordinate / nm
1200
DOPC
Cholesterol
Water
Phosphate
1000
DOPC
Cholesterol
Water
Phosphate
1000
800
Density / kg m-3
-3
4
600
1200
Density / kg m
2
200
Z coordinate / nm
600
400
200
0
0
Z coordinate / nm
800
200
-2
1000
800
0
-4
1200
DOPC
β-sitosterol
Water
Phosphate
1000
DOPC
β-sitosterol
Water
Phosphate
800
600
400
200
-4
-2
0
2
4
Z coordinate / nm
0
-4
-2
0
Z coordinate / nm
Figure 3: Density profiles for the different bilayer systems (a) S10, (b) S33, (c)
S50, (d) S50R, (e) C10 and (f) C50.
39
40
Figure 4: Snapshot of two DOPC molecules forming an “umbrella” over a βsitosterol molecule, taken from the simulation of the S33 bilayer at 350K.
0.4
0.4
DOPC
S10
S50
S50R
C10
C50
0.35
0.3
0.35
0.3
0.25
|SCD|
|SCD|
0.25
0.2
0.15
0.15
0.1
0.1
0.05
0.05
0
2
4
6
8
10
12
14
Methylene postion in acyl chain
0.4
18
0
0.3
0.25
0.25
0.2
0.15
0.1
0.1
0.05
0.05
4
6
8
10
12
14
4
6
8
10
12
14
Methylene postion in acyl chain
16
Methylene postion in acyl chain
18
18
16
18
DOPC
S10
S50
C10
C50
0
2
4
6
8
10
12
14
Methylene postion in acyl chain
Figure 5: Order parameters of DOPC molecules in the bilayers (a) sn-1 chain
at 303K, (b) sn-2 chain at 303K, (c) sn-1 chain at 350K and (d) sn-2 chain at
350K.
41
16
0.2
0.15
2
2
0.35
0.3
0
DOPC
S10
S50
S50R
C10
C50
0.4
|SCD|
|SCD|
16
DOPC
S10
S50
C10
C50
0.35
0.2
0.07
0.06
0.03
0.03
0.02
0.01
0.01
0
10
20
30
40
50
60
Angle / degrees
70
80
90
0.03
0.01
0.01
20
30
40
50
60
70
30
40
50
60
Angle / degrees
70
80
90
80
90
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
10% Cholesterol
50% Cholesterol
0.03
0.02
10
20
0.04
0.02
0
10
0.05
0.04
0
0
0.06
Probability
0.05
0
0.07
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
50% β-sitosterol Random
10% Cholesterol
50% Cholesterol
0.06
Probability
0.04
0.02
0.07
0.05
0.04
0
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
10% Cholesterol
50% Cholesterol
0.06
Probability
Probability
0.05
0.07
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
50% β-sitosterol Random
10% Cholesterol
50% Cholesterol
80
Angle / degrees
90
0
0
10
20
30
40
50
60
Angle / degrees
Figure 6: Tilt angles of the sterol molecules with the plane of the bilayer. (a)
sterol head at 303K, (b) sterol head at 350K, (c) sterol tail at 303K and (d)
sterol tail at 350K.
42
70
0.03
DOPC
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
10% Cholesterol
50% Cholesterol
0.025
Probability
0.02
0.015
0.01
0.005
0
0
20
40
60
80
100
120
140
160
180
160
180
Angle / degrees
0.03
DOPC
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
10% Cholesterol
50% Cholesterol
0.025
Probability
0.02
0.015
0.01
0.005
0
0
20
40
60
80
100
120
140
Angle / degrees
Figure 7: The probability distribution function of the angle between the bilayer
normal, directed away from the middle of the bilayer, and the vector between
the P- →N+ atoms in the lipid headgroup for simulations of the DOPC molecules
at (a) 303K and (b) 350K in water.
43
1200
800
-3
800
600
400
600
400
200
0
200
-4
-2
0
1200
0
600
400
-4
-2
0
1200
4
4
DOPC
β-sitosterol
DMSO
Water
Phosphate
600
400
0
-4
-2
0
600
400
200
4
DOPC
β-sitosterol
DMSO
Water
Phosphate
1000
800
2
Z coordinate / nm
1200
Density / kg m-3
-3
2
DOPC
β-sitosterol
DMSO
Water
Phosphate
1000
Density / kg m
2
200
Z coordinate / nm
0
0
Z coordinate / nm
800
200
-2
1000
800
0
-4
1200
Density / kg m-3
-3
4
DOPC
β-sitosterol
DMSO
Water
Phosphate
1000
Density / kg m
2
Z coordinate / nm
DOPC
β-sitosterol
DMSO
Water
Phosphate
1000
Density / kg m
-3
1000
Density / kg m
1200
DOPC
DMSO
Water
Phosphate
800
600
400
200
-4
-2
0
2
4
Z coordinate / nm
0
-4
-2
0
Z coordinate / nm
Figure 8: Density profiles for the bilayer systems (a) DOPC, (b) S10, (c) S33,
(d) S50 at 350K and 2.5 mol % DMSO, (e) S10 and (f) S50 at 25.0 % mol
DMSO.
44
2
4
Figure 9: Snapshots of mixed lipid-sterol bilayers in DMSO, (a) S10 in 2.5 mol
% DMSO, (b) S10 in 25.0 mol % DMSO, (c) S50 in 2.5 mol % DMSO and
(d) S50 in 25.0 mol % DMSO. The DOPC phosphate headgroups, DOPC tails,
β-sitosterol, water and DMSO are coloured red, grey, green, blue and yellow,
respectively.
45
Figure 10: Snapshots from the S10 system at 350K showing how sterol molecules
flip from one leaflet to the opposite, taken at (a) 91.6, (b) 94.6, (c) 95.3, (d)
96.1, (e) 96.6 and (f) 100.0 ns. The DOPC/β-sitosterol are coloured grey, the
water molecules, phosphate headgroups and β-sitosterol molecule undergoing
the “flip” are coloured blue, purple/red and yellow, respectively.
46
0.3
0.3
DOPC
S10
S33
S40
S50
0.25
0.25
0.2
|SCD|
|SCD|
0.2
0.15
0.15
0.1
0.1
0.05
0.05
0
DOPC
S10
S33
S40
S50
2
4
6
8
10
12
14
16
18
0
2
4
Methylene postion in acyl chain
0.25
0.25
0.2
0.2
0.15
0.1
0.05
0.05
2
4
6
8
10
12
14
10
12
14
16
18
16
18
Methylene postion in acyl chain
16
18
S10
S33
S50
0.15
0.1
0
8
0.3
S10
S33
S50
|SCD|
|SCD|
0.3
6
Methylene postion in acyl chain
0
2
4
6
8
10
12
14
Methylene postion in acyl chain
Figure 11: Order parameters of DOPC molecules in the bilayers (a) sn-1 chain
in 2.5 mol % DMSO, (b) sn-2 chain in 2.5 mol % DMSO, (c) sn-1 chain in 2.5
mol % DMSO and (d) sn-2 chain in 2.5 mol % DMSO.
47
0.05
0.04
0.03
0.02
0.01
0
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
0.04
Probability
Probability
0.05
10% β-sitosterol
33% β-sitosterol
40% β-sitosterol
50% β-sitosterol
0.03
0.02
0.01
0
10
20
30
40
50
60
70
80
90
0
0
10
20
Angle / degrees
0.05
0.03
0.02
0
50
60
70
80
90
80
90
10% β-sitosterol
33% β-sitosterol
50% β-sitosterol
0.04
Probability
Probability
0.05
0.01
40
Angle / degrees
10% β-sitosterol
33% β-sitosterol
40% β-sitosterol
50% β-sitosterol
0.04
30
0.03
0.02
0.01
0
10
20
30
40
50
60
70
80
90
Angle / degrees
0
0
10
20
30
40
50
60
70
Angle / degrees
Figure 12: Tilt angles of the sterol molecules with the plane of the bilayer (a)
sterol head in 2.5 mol % DMSO, (b) sterol head in 25.0 mol % DMSO, (c) sterol
tail in 2.5 mol % DMSO and (d) sterol tail in 25.0 mol % DMSO.
48
20
18
16
Z-coordinate / nm
14
12
10
8
6
4
2
0
160
165
170
175
180
185
190
195
Time / ns
Figure 13: The trajectories along the z-axis (normal to the bilayers) of DMSO
molecules that diffuse through the S50 bilayer (shown by the blue and green
lines), taken from 40 ns simulation of a system at a DMSO concentration of 2.5
mol % at 350K. The red lines indicate the positions of the phosphate groups.
49
200
150
DOPC
33% β-sitosterol
50% β-sitosterol
# of DMSO molecules in chamber A
145
140
135
130
125
120
115
110
105
0
50
100
# of Water molecules in chamber A
5750
150
time / ns
200
250
300
DOPC
33% β-sitosterol
50% β-sitosterol
5740
5730
5720
5710
5700
5690
0
50
100
150
time / ns
200
250
300
Figure 14: The number of (a) DMSO and (b) water molecules in chamber A
(the high DMSO concentration chamber) of the double bilayer simulations as a
function of simulation time.
50
Table of Contents Image
Table of Contents Figure: Snapshot of two DOPC molecules forming an “umbrella” over a β-sitosterol molecule (left). A snapshot of a 1:1 DOPC-β-sitosterol
bilayer solvated in 2.5 mol. % DMSO solution (right). The DOPC, β-sitosterol,
DMSO and water molecules are coloured grey, cyan, yellow and blue, respectively.
51