DOI: 10.1021/la100891xLangmuir 2010, 26(13), 11118–11126Published on Web 06/15/2010
©2010 American Chemical Society
Water Replacement Hypothesis in Atomic Details: Effect of Trehalose
on the Structure of Single Dehydrated POPC Bilayers
E. A. Golovina,†A. Golovin,‡F. A. Hoekstra,†and R. Faller*,§
†Laboratory of Plant Physiology, Wageningen University, Wageningen, The Netherlands,‡Faculty of
Bioengineering and Bioinformatics, Moscow State University, Moscow, Russia, and
Engineering & Materials Science, University of California, Davis, Davis, California 95616
§Department of Chemical
Received March 3, 2010. Revised Manuscript Received June 4, 2010
We present molecular dynamics (MD) simulations to study the plausibility of the water replacement hypothesis
(WRH) from the viewpoint of structural chemistry. A total of 256 2-oleoyl-1-palmitoyl-sn-glycero-3-phosphocholine
(POPC) lipids were modeled for 400 ns at 11.7 or 5.4 waters/lipid. To obtain a single dehydrated bilayer relevant to the
WRH, simulations were performed in the NPxyhzT ensemble with hz>8 nm, allowing interactions between lipids in the
membrane plane and preventing interactions between neighboring membranes via periodic boundary conditions. This
setupresultedin astablesinglebilayerin (ornear)thegel state.Trehalose causedaconcentration-dependent increaseof
the area per lipid (APL) accompanied by fluidizing the bilayer core. This mechanism has been suggested by the WRH.
However, dehydrated bilayers in the presence of trehalose were not structurally identical to fully hydrated bilayers. The
headgroup vector was in a more parallel orientation in dehydrated bilayers with respect to the bilayer plane and main-
tained this orientation in the presence of trehalose in spite of APL increase. The total dipole potential changed sign in
dehydrated bilayers and remained slightly positive in the presence of trehalose. The model of a dehydrated bilayer
It is well establishedthattrehaloseprotects membranes indesic-
cationtolerant organisms.1,2This “lesson fromnature” is used to
protect the content of dry liposomes against leakage in the phar-
blood cells during freeze-drying,5which bears promise for blood
nism of membrane protection by trehalose.1,2This mechanism is
based on replacement of water molecules by sugars in their inter-
actions with polar groups of membrane lipids. These interactions
maintain spacing between lipids and prevent the increase of the
consequence, dry membranes remain in a fluid state at physio-
logical temperatures and avoid a phase transition during rehy-
dration. The transient coexistence of fluid and gel phases in a
membrane during rehydration causes leakage and is detrimental
for living organisms.
The WRH has considerable experimental support (see e.g.
refs 2 and 3 and references therein). However, all experimental
that can support mechanisms described by WRH.6,7Thus, alter-
native hypotheses which explain experimental data by other
mechanisms than interactions of disaccharides with lipid polar
interactions and consider sugar vitrification as the main mechan-
ism of membrane protection by trehalose at low (<20%) water
branes to remain in the phase they were in at the time of vitrifi-
of membranes occurs when membranes are in a fluid state, they
thus preserving their hydration shell and maintaining the neces-
sary level of hydration during osmotic stress. This theory is valid
for osmotic stress but does not relate to severe dehydration. Belton
*To whom correspondence should be addressed.
(1) Crowe, J. H.; Crowe, L. M.; Chapman, D. Preservation of membranes in
anhydrobiotic organisms - The role of trehalose. Science 1984, 223 (4637), 701-
(2) Crowe, J. H.; Hoekstra, F. A.; Crowe, L. M. Anhydrobiosis. Annu. Rev.
Physiol. 1992, 54, 579-599.
(3) Crowe, J. H.; Crowe, L. M.; Oliver, A. E.; Tsvetkova, N.; Wolkers, W.;
Tablin, F. The trehalose myth revisited: Introduction to a symposium on
stabilization of cells in the dry state. Cryobiology 2001, 43 (2), 89-105.
(4) Crowe, J. H.; Crowe, L. M.; Wolkers, W. F.; Oliver, A. E.; Ma, X.; Auh,
J.-H.; Tang, M.; Zhu, S.; Norris, J.; Tablin, F. Stabilization of Dry Mammalian
Cells: Lessons from Nature. Integr. Comp. Biol. 2005, 45, 810-820.
(5) Wolkers, W. F.; Walker, N. J.; Tablin, F.; Crowe, J. H. Human platelets
loaded with trehalose survive freeze-drying. Cryobiology 2001, 42, 79-87.
(6) Lee, C. W. B.; Das Gupta, S. K.; Mattai, J.; Shipley, G. G.; Abdel-Mageed,
O. J.; Makriyannis, A.; Griffin, R. G. Characterization of the L-lambda phase in
trehalose-stabilized dry membranes by solid-state NMR and X-ray diffraction.
Biochemistry 1989, 28, 5000-5009.
(8) Koster, K. L.; Webb, M. S.; Bryant, G.; Lynch, D. V. Interactions between
soluble sugars and POPC (1-palmitoyl-2-oleoylphosphatidylcholine) during dehy-
dration: vitrification of sugars alters the phase behavior of the phospholipid.
Biochim. Biophys. Acta 1994, 1193, 143-150.
(9) Wolfe, J.; Bryant, G. Freezing, drying, and/or vitrification of membrane-
solute-water systems. Cryobiology 1999, 39, 103-129.
(10) Koster, K. L.; Maddocks, K. J.; Bryant, G. Exclusion of maltodextrins
from phosphatidylcholine multilayers during dehydration: effects on membrane-
phase behaviour. Eur. Biophys. J. 2003, 32, 96-105.
(11) Lenn? e,T.;Bryant,G.;Garvey,C.J.;Keiderling,U.;Koster,K.L.Location
ofsugars in multilamellar membranes atlowhydration. PhysicaB2006,385-386,
(12) Arakawa, T.; Timasheff, S. N. Preferential interactions of proteins with
salts in concentrated solutions. Biochemistry 1982, 21 (25), 6545-6552.
of trehalose with hen egg white lysozyme. Biopolymers 1994, 34 (7), 957-961.
Langmuir 2010, 26(13), 11118–11126
Golovina et al. Article
which was preferentially kept at the protein surface during
osmotic stress, remains near protein surface at dehydration due
to entrapment by sugar glasses.
In spite of different alternative hypotheses of membrane
protection at desiccation, WRH remains the most attractive due
toits simplicity.Thefirst attempt toinvestigatetheplausibilityof
WRH using structural chemistrywas based on molecular graphics
Reasonable binding geometries were found for trehalose with
a monolayer excised from the 1,2-dimyristoyl-sn-glycero-3-phos-
phocholine (DMPC) crystal structure.
Nowadays, molecular dynamic simulations open new oppor-
This is mainly due to technical difficulties and long relaxation
times in simulations of dry bilayers.
In molecular dynamics (MD) simulations periodic boundary
conditions (PBC) are usually used to avoid artifacts due to the
smallsizeofthesimulation box.Thiseffectivelyresults insimula-
tions of a infinite stack of bilayers instead of a single bilayer.23
When lipids are fully hydrated (around 30 waters per lipid),
bilayers are separated by a sufficient water layer, which prevents
significant interactions between bilayers. However, when consid-
erable water is removed from the interbilayer space to create
dehydrated bilayers, separation between bilayers is reduced and
avoid self-interaction by simulating multiple layers, we cannot
compare to single dehydrated membranes as occurring in nature.
In a recent study of a dehydrated bilayer stack under periodic
boundary conditions we have shown small interpenetration of
headgroups of neighboring bilayers at 11.7 waters/lipid and
Interpenetration creates self-spacing of bilayers and prevents
has been seen earlier in MD of dry bilayers.25-27We conjectured
that interpenetration becomes possible due to reorientation of
headgroup PN vectors from facing outward to inward which
dehydrated bilayer stacks is relevant to X-ray studies of bilayer
structure. Bilayer stacks at hydration levels below 14.5 waters/
lipid are usedtoobtaintrusted diffractionintensities.28However,
the disappearance of water layer between bilayers and overlap of
headgroups at low water contents cause changes, which do not
allow calculation of bilayer area per lipid and other structural
parameters.28Diffraction patterns of dehydrated lipd/trehalose
angle regions.6The low-angle diffraction spacing index is indica-
the wide-angle diffraction pattern due the presence of trehalose
dehydrated stacks with and without trehalose complete the pic-
ture of interactions between dry bilayers and trehalose, obtained
by X-ray scattering and solid-state NMR.6,7,29
single bilayer, and interactions with adjacent bilayers are not
consideredapriori. This hypothesisrelates tothesituationinvivo,
where cell membranes do not form stacks at dehydration but are
extracellular media. Therefore, the plausibility of the WRH from
the point of structural chemistry needs to be investigated by MD
simulations of a single dry bilayer. Such conditions of single
bilayers obviously occur in dry cells.
and Sum.30The separation of neighboring 1,2-dipalmitoyl-sn-
glycero-3-phosphocholine (DPPC) bilayers was provided by
argon molecules introduced between bilayers. Their results show
that unilamellar bilayers become unstable and disintegrate to non-
bilayer structures at 10 waters/lipid in the absence of sugars and
maintain their stability with sugars. This breakdown was attri-
water and lipids. However, such a model is not directly experi-
mentally relevant. According to the molecular shape concept of
lipid polymorphism, DPPC molecules organize themselves into
bilayers in both the hydrated and dehydrated state.31,32None-
theless, the attempt to simulate a single dry bilayer is a valid
approach to study the mechanisms of membrane protection
against desiccation and needs further development.
Details of the Simulation. All MD simulations were carried
(14) Chandrasekhar, I.; Gaber, B. P. Stabilization of the bio-membrane by small
molecules: interaction of trehalose with the phospholipids bilayer. J. Biomol. Struct.
Dyn. 1988, 5, 1163-1171.
(15) Rudolph, B. R.; Chandrasekhar, I.; Gaber, B. P.; Nagumo, M. Molecular
modelling of saccharide-lipid interactions. Chem. Phys. Lipids 1990, 53 (2-3),
(16) Sum, A. K.; de Pablo, J. J. Molecular simulation study on the influence of
dimethylsulfoxide on the structure of phospholipid bilayers. Biophys. J. 2003, 85
The effect of trehalose and cholesterol on a phospholipid bilayer. J. Phys. Chem. B
2005, 109 (50), 24173-24181.
(18) Pereira, C. S.; Hunenberger, P. H. Effect of trehalose on a phospholipid
membrane under mechanical stress. Biophys. J. 2008, 95 (8), 3525-3534.
(19) Pereira, C. S.; Lins, R. D.; Chandrasekhar, I.; Carlos, L.; Freitas, G.;
Bilayer: A Molecular Dynamics Study. Biophys. J. 2004, 86 (4), 2273-2285.
(20) Sum, A. K.; Faller, R.; de Pablo, J. J. Molecular simulation study of
phospholipid bilayers and insights of the interactions with disaccharides. Biophys.
J. 2003, 85 (5), 2830-2844.
of the Response of Lipid Bilayers and Monolayers to Trehalose. Biophys. J. 2005,
89 (6), 4111-4121.
(22) Villarreal, M. A.; Diaz, S. B.; Disalvo, E. A.; Montich, G. G. Molecular
dynamics simulation study of the interaction of trehalose with lipid membranes.
Langmuir 2004, 20 (18), 7844-7851.
of membranes: Molecular dynamics studies of lipid bilayer systems. Biochim.
Biophys. Acta, Rev. Biomembr. 1997, 1331 (3), 235-270.
(24) Golovina, E. A.; Golovin, A. V.; Hoekstra, F. A.; Faller, R. Water
replacement hypothesis in atomic details - factors determining area per lipid in
dehydrated stack bilayers. Biophys. J. 2009, 97 (2), 490-499.
(25) Essmann, U.; Perera, L.; Berkowitz, M. L. The origin of the hydration
interaction of lipid bilayers from MD simulation of dipalmitoylphosphatidylcholine
membranes in gel and liquid crystalline phases. Langmuir 1995, 11, 4519-4531.
(26) Feller, S. E.; Yin, D.; Pastor, R. W.; MacKerell, A. D., Jr. Molecular
dynamics simulation of unsaturated lipid bilayers at low hydration: parameteri-
zation and comparison with diffraction studies. Biophys. J. 1997, 73 (5), 2269-
(27) Mashl, R. J.; Scott, H. L.; Subramaniam, S.; Jakobsson, E. Molecular
Simulation of Dioleoylphosphatidylcholine Lipid Bilayers at Differing Levels of
Hydration. Biophys. J. 2001, 81 (6), 3005-3015.
(28) Tristram-Nagle, S.; Nagle, J. F. Lipid bilayers: thermodynamics, structure,
fluctuations and interactions. Chem. Phys. Lipids 2004, 127, 3-14.
(29) Quinn, P. J.; Koynova, R. D.; Lis, L. J.; Tenchov, B. G. Lamellar gel-
lamellar liquid crystal phase transition of dipalmitoylphospatidylcholine multi-
layers freez-dried from aqueous trehalose solutions. A real-time X-ray difraction
study. Biochim. Biophys. Acta 1988, 942, 315-323.
(30) Leekumjorn, S.; Sum, A. K. Molecular dynamics study on the stabilization
of dehydrated lipid bilayers with glucose and trehalose. J. Phys. Chem. B 2008,
(31) Cullis, P. R.; de Kruijff, B. Lipid polymorphism and the functional roles of
lipids in biological membranes. Biochim. Biophys. Acta 1979, 559, 399-420.
(32) Epand, R. M. Membrane lipid polymorphism. Relationship to bilayer
properties and protein function. In Methods in Membrane Lipids; Dopico, A. M.,
Ed.; Humana Press: New York, 2007; pp 15-26.
DOI: 10.1021/la100891xLangmuir 2010, 26(13), 11118–11126
Article Golovina et al.
short simulations were done with version 3.3.3,33and long
trajectories were produced with version 4.0.34The authors have
donepreviouslyextensive test simulationstoensurethatdifferent
description for the lipids was used. Parameters for bonded and
nonbonded interactions were taken from a study of DPPC bi-
layers,35electronically available at http://moose.bio.ucalgary.ca/
files/lipid.itp. Partial charges were obtained from Tieleman and
Berendsen36which are based on calculations by Chiu et al.37and
can be found at http://moose.bio.ucalgary.ca/files/popc.itp. The
increase the time of simulation to 400 ns and the size of the
acceptable time steps. But it was shown that using LINCS time
steps of 5 and 2 fs provides similar accuracy in bilayer simula-
tions.40The water geometry (as a rigid molecule) was maintained
with the SETTLE algorithm.41In some cases LINCS cannot
ing both bonds and angles. To solve this issue, Feenstra et al.42
introduced dummy atoms and increased the hydrogen mass. For
our simulations this was not necessary. For Lennard-Jones inter-
actions we used a plain cutoff (without shift function) of 1.2 nm.
step, while interactions beyond this range every 10 time steps.
Long-range electrostatics was handled by means of the particle-
mesh Ewald technique.43Neighbor searching used a twin-range
approach with the cutoff of 1 nm. POPC lipids, trehalose, and
using the Berendsen algorithm44with a coupling constant of
0.1 ps. The box size was maintained constant in the z direction.
Therefore, the simulations were carried out in the NhzpxyT en-
semble. Pressure was separately coupled to 1 atm for x and y
directions with coupling constant of 1 ps using the Berendsen
weak coupling algorithm.44The starting point for all simulations
a fluid state. For the systems with 11.7 and 5.4 waters per lipid,
water was removedfrom the midplane ofthe interlamellar region
between bilayers. To create a bilayer with trehalose, different
numbersofsugarswereplacedrandomlyinthe empty space after
water removal. Initial constraints on the lipids resulted in prefe-
rential location of trehalose and water along the lipid interface
before any changes in area per lipid take place. We used a void
their periodic images. The vapor pressure of water is negligible
To keep the same thickness of this vacuum layer (4 nm) between
bilayers through PBC, the box height was set separately for each
conditions was performed. To avoid artifacts due to the small
xy plane. After multiplicationwe run eachsystem for 400 ns with
256 POPC lipids and a corresponding number of water and
trehalose molecules. The exact compositions and our nomencla-
ture for all systems are presented in Table 1.
Analysis Techniques. All analysis was averaged over the last
50 ns (350-400 ns) in the simulation which means at or at least
near equilibrium conditions. The area per lipid (APL) was
calculated by dividing the xy plane area of the simulation box
APL to the exponential function y=A1exp(-x/t1) þ y0, where
(Figure1).The valuesofAPL, obtainedbyfittingand byaverag-
ing, do not deviate significantly (Table 2), which indicates that
density profiles, the box was divided into 200 slices along the
Table 1. Vacuum Model Composition and Nomenclature
Used in This Work
Figure 1. Timeevolution ofareaper lipid atintermediate(A;11.7
waters/lipid) and low (B; 5.4 waters/lipid) water contents and
different trehalose concentrations (Table 1). For comparison, the
data for the fully hydrated bilayer (h28-00; 28.5 waters/lipid) are
alsoplotted.Graylines represent exponential fits. Insets: averaged
over the last 50 ns area per lipid in dehydrated bilayers at different
trehalose:lipid ratio. Standard deviation (SD) bars are visible if
the fully hydrated bilayer h28-00.
(33) Van Der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.;
Berendsen, H. J. C. GROMACS: Fast, flexible, and free. J. Comput. Chem. 2005,
26 (16), 1701-1718.
(34) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4:
Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simula-
tion. J. Chem. Theory Comput. 2008, 4 (3), 435-447.
(35) Berger, O.; Edholm, O.; Jahnig, F. Molecular dynamics simulations of a
fluid bilayer of dipalmitoylphosphatidylcholine at full hydration, constant pres-
sure, and constant temperature. Biophys. J. 1997, 72 (5), 2002-2013.
(36) Tieleman, D. P.; Berendsen, H. J. C. Molecular dynamics simulations of
fully hydrated dipalmitoylphosphatidylcholine bilayer with different macroscopic
boundary conditions and parameters. J. Chem. Phys. 1996, 105 (11), 4871-4880.
E. Incorporation of Surface Tension into Molecular Dynamics Simulation of an
(38) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J.
Interaction models for water in relation to protein hydration. In Intermolecular
Forces; Pullman, B., Ed.; Reidel: Dordrecht, 1981; pp 331-342.
(39) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. LINCS: A
Linear Constraint Solver for Molecular Simulations. J. Comput. Chem. 1997, 18
(40) Anezo, C.; de Vries, A. H.; Holtje, H.; Tieleman, D. P.; Marrink, S. J.
(41) Miyamoto, S.; Kollman, P. A. Settle: An analytical version of the SHAKE
and RATTLE algorithm for rigid water models. J. Comput. Chem. 1992, 13 (8),
(42) Feenstra, K. A.; Hess, B.; Berendsen, H. J. C. Improving efficiency of large
time-scale molecular dynamics simulations of hydrogen-rich systems. J. Comput.
Chem. 1999, 20 (8), 786-798.
(44) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.;
Haak, J. R. Molecular dynamics with coupling to an external heat bath. J. Chem.
Phys. 1984, 81 (8), 3684-3690.
Langmuir 2010, 26(13), 11118–11126
Golovina et al.Article
bilayer normal, and the number of atoms of the given molecule
(POPC, water, or trehalose) was calculated in each slice. To cal-
was multiplied but its atomic weight. The height of the box was
different in different modes (Table 1); therefore, the thickness of
the slices was also different. To compare density profiles between
the models, we used the same thickness of the slices in all simula-
but beforethe analysisto themaximum value of 17nm(Table 1).
The density profiles were plotted separately for lipids, water,
trehalose, and P atoms. The profiles are symmetrized between
leaflets. Headgroup orientation is characterized by the cosine of
the angle between the PN vector (the dipole between the phos-
phate P atom and the choline N atom) and the bilayer normal.
of the transformation between internal non-Cartesian and exter-
nal Cartesian coordinates.45For characterizations we fit the dis-
tributions with several Gaussians. Electrostatic potentials for
lipids, water, and trehalose are calculated separately. Charge dis-
tion along the bilayer normal and integrated twice. All potentials
are defined as zero in the center of the bilayer.
Effect of Dehydration and Trehalose on the APL of POPC
Bilayer. Figure 1 shows the evolution of the APL during the
lipid) water contents with different amounts of trehalose. Differ-
ent models need different time to attain near-equilibrium condi-
tions ranging from several tens to a few hundred nanoseconds.
The slowest equilibration was in the models with low concentra-
tion of trehalose (less than 1 trehalose per lipid).
for 5.4 waters/lipid. The APL decreases with water content from
(0.515 nm2at 11.7 waters/lipid) or even in the gel state (0.48 nm2
at 5.4 waters/lipid) (Table 2, Figure 1). The value of APL for the
rature, and this discrepancy is discussed in our previous paper.24
increases at low concentration of trehalose (Figure 1 A, inset).
Considerable;almost linear;growth of the APL is observed at
higher trehalose concentrations. At 5.4 waters per lipid the APL
increases linearly over the whole range of trehalose concentrations
lipid as in v11 models (Table 2). However, the APL of the fully
hydratedbilayer isnot reachedineithermodel(dashedredlinein
insets of Figure 1).
Effect of Dehydration and Trehalose on Molecular Order
Parameter. The lipid chain molecular order parameters Smol
(using the usual definition23) were used to quantify the extent of
chain ordering induced by dehydration and trehalose in a single
dry bilayer. Smolwas calculated according to the equation
Smol ¼ ð3=2Þ½cos2θ? -1=2
where θ is the angle between the z-axis of the simulation box and
vector from Cn-1to Cnþ1and therefore can be applied to the
the hydrophobic core46and can be directly compared with the
structural parameters of the bilayer.
The order parameter profiles for the sn-1 palmitoyl chains of
POPC lipids for low (5.4 waters/lipid) and intermediate (11.7
waters/lipid) water contents and different amounts of trehalose
are shown in Figure 2 in comparison with the fully hydrated
than in v11 models. Trehalose causes concentration-dependent
disorderingoftheacyl chainsatbothwater contentsbutdoesnot
The deuterium order parameter profile of sn-2 oleoyl chain is
different from that of sn-1 palmytoil chain due to the bend in the
sn-2 chain after C2and alignment of the double bond almost
paralleltothebilayer normal.47Thisuniqueorientation ofthecis
double bond in membranes causes the characteristic dip at C10.
However, after correction for this geometric factor, molecular
order parameters are identical in both chains.47Therefore, both
and we can use sn-1 to characterize the degree of ordering in the
the integrated order parameter (the sum of all order parameters
Model (SE: Standard Error, SD: Standard Deviation)
fit ( SE
averaged over the
last 50 ns ( SD
APL [nm2] block
averaged: last 50 ns,
5 blocks each
10 ns ( SD
Figure 2. Profiles of the order parameter Smolfrom C3to C15at
different water and trehalose contents: (A) 5.4 waters/lipid, (B)
(45) Fixman, M. Simulation of polymer dynamics: I. General theory. J. Chem.
Phys. 1978, 69 (4), 1527-1537.
(46) Seelig, A.; Seelig, J. Effect of a Single Cis Double-Bond on Structure of a
Phospholipid Bilayer. Biochemistry 1977, 16 (1), 45-50.
(47) Seelig,J.;Waespe-?Sar? cevi? c,N.Molecularorderincisandtransunsaturated
phospholipid bilayers. Biochemistry 1978, 17 (16), 3310-3315.
DOI: 10.1021/la100891x Langmuir 2010, 26(13), 11118–11126
Article Golovina et al.
Surprisingly, water content influences the integrated order para-
meter only in models without trehalose. In the presence of sugar
the degree of disordering is determined by trehalose content
rather than hydration level (Figure 3A).
shows the correlation between APL and the integrated order
parameter of the palmitoyl chain for all 11 models. In a log-log
range,we refrainfromdeterminingascalingexponent. The order
parameter is more sensitive to changes of APL in or near the gel
state rather than in a fluid state of the bilayer.
Mass Density Profiles of POPC. Mass density profiles
change with dehydration (Figure 4, Table 3). The peak-to-peak
separation (PP) is partly determined by the hydrophobic thick-
ness of the bilayer. A considerable increase of PP with dehydra-
tion results from the extension of the acyl chains in the gel state.
Dehydration is also accompanied by the increase of peak density
(Dmax) and peak sharpness originating probably from immobili-
v5-00 than in the fully hydrated bilayer (Figure 4, Table 3),
indicating more highly ordered hydrocarbon chains as compared
to the fluid phase.48The more pronounced methyl trough at low
water contents together with the gradual decrease of order para-
meter toward the end of acyl chains (Figure 2) show that bilayer
dehydration causes acyl chain immobilization without interdigi-
tation. The appearance of a region of constant density within the
hydrophobic part of the bilayer (Figure 4B,C) correlates with the
relatively ordered methylene groups of the acyl chains.49All
structural changes relate to the transition from the fluid to the
gel state of the acyl chains under dehydration.
In the presence of trehalose peak-to-peak separation (PP)
decreases inbothv5andv11models(Table3). Withintermediate
water content (v11) PP reduces to the level of the fully hydra-
ted bilayer at lipid:trehalose = 1:1 and remains there at higher
trehalose concentration (Table 3). At low water content (v5) the
reaching the value for the fully hydrated bilayer (Table 3).
The regions of constant density between headgroups and ter-
minal methylsdisappear indehydratedbilayersinthe presenceof
trehalose (Figure 5). The depth of the terminal methyl trough
decreases and in the model with intermediate water content rea-
ches the value of that of the hydrated bilayer at trehalose/lipid=
1:2 (Table 3). At low water content the depth of the trough
decreases gradually and does not reach the value of the hydrated
bilayer even at the highest trehalose concentration (Table 3).
Overall, thestructuralchanges indicateaconcentrationdepen-
dent fluidizing of dehydrated bilayers by trehalose. However, the
lipid density profiles remain different at the interface (Figure 5).
The distance between the headgroup peak and the outer edge of
the bilayer (Wouter), defined as the distance over which the head-
group density drops from 90% to 10%, decreases with dehydra-
v11-00 and v5-00, respectively (Table 3, Figure 4). Trehalose
causes a concentration-dependent increase of Wouter. It levels off
around the fully hydrated value of 0.7 nm at trehalose:lipid=1:2
for lowwatercontentv5(Table3). Atintermediatewatercontent
(v11) Wouteris wider than in the fully hydrated bilayer at all
trehalose concentrations (Figure 5, Table 3) This is particularly
visible in models v11-10 and v11-14 where the peak positions are
similar to h28-00, but interfaces are considerably extended out-
the fully hydrated bilayer (Figure 5). Therefore, the structure of
bilayer in spite of acyl chain fluidizing.
(P) Atoms. Water densityprofilesina fully hydratedbilayer can
be divided into two zones: the interbilayer bulk water (plateau)
and the lipid hydration shell. The density of water gradually dec-
Figure 3. Integratedorderparameter(sumofSmolfromC3toC15).
(A) Dependence on trehalose content for intermediate v11 (11.7
waters/lipid, black line) and low v5 (5.4 waters/lipid, red line) water
contents; dashed line is integrated order parameter for the fully
integrated order parameter and area per lipid (points represent the
data for all the models).
Figure 4. Mass density profiles of different components (POPC:
black; water: blue; P atoms: red) in fully hydrated bilayer h28-00
(A), at intermediate water content v11-00 (B), and at low water
content v5-00 (C). The mass density profile of water in each leaflet
of dehydrated bilayers can be represented as two Gaussian dis-
tributions (gray lines), which are referred to as outer and inner
water according to the positions within the leaflets. The center of
the bilayer is at z = 0.
(48) McIntosh,T.J.; Simon,S. A.Area perMoleculeandDistributionofWater
in Fully Hydrated Dilauroylphosphtidylethanolamine Bilayers.
1986, 25, 4948-4952.
(49) Nagle, J. F.; Wiener, M. C. Relations for lipid bilayers. Connection of
electron density profiles to other structural quantities.
Biophys. J. 1989, 55,
Langmuir 2010, 26(13), 11118–11126
Golovina et al.Article
the membrane core (Figure 4A). We estimate the depth of water
penetration into the bilayer core as the position from the bilayer
≈0.6 nm, which is slightly deeper than measured in experiment.50
Water density profiles in leaflets of dehydrated bilayers (v11-00
and v5-00 models) can be fitted by two Gaussians (Figure 4B,C).
We identify them as outer and inner water, respectively. Inner
water associates with the positions of P atoms, whereas the outer
water is shifted to the outer edge of the interface. The inner
distribution is wider than the outer in both dehydrated systems
(Figure 4B,C). At intermediate water content (v11) a considera-
ble part of the outer water is outside of the interface (Figure 5B).
The penetrationdepth ofwater toward the chainsdecreases from
≈0.6 nm in h28-00 to ≈1.2 nm and ≈1.5 nm in v11-00 and v5-00,
Trehalose distribution depends on concentration. When the
trehalose is within the interface region (Figure 5). For trehalose:
of the bilayer. Trehalose penetrates as deep into the bilayer as
water. The presence of trehalose changes the water distribution
The outer water distribution moves out of the interface when
excess trehalose forms the layer outside of the lipid boundary
sence of trehalose but its position does not change with concen-
tration. Water molecules penetrate deeper into bilayer in dehy-
drated bilayer with trehalose than in the fully hydrated bilayer.
The depth of penetration gradually increases with the increase of
fully hydrated bilayer (Figure 5).
Lipid Potential and PN Vector Orientation. Lipid electro-
phosphorusandnitrogenatoms (PNvector) arecloselyrelated.24
the distribution of charges in the simulations can change the
electrostatic potential. The dominant contribution comes from
electrostatic potential of the lipid is the orientation of the dipole
between the phosphorus and the choline nitrogen.
Table 3. Structural Parameters of Dehydrated Bilayer with Trehalose Derived from Density Profiles
5.4 waters/lipid11.7 waters/lipid
trehalose/box PP [nm]Wouter[nm]Dmin[kg/m3]Dmax[kg/m3] PP [nm]Wouter[nm]Dmin[kg/m3]Dmax[kg/m3]
Figure 5. Mass density profiles of POPC (black), water (external
dark blue and internal light blue), and trehalose (red) in bilayer
leaflets at intermediate 11.7 waters/lipid (left column) and low
5.4 waters/lipid (right column) water contents and different treha-
lose concentrations (Table 1). For comparison, the density profile
of POPC in a fully hydrated bilayer 28. Five waters/lipid (h28-00)
is presented in all graphs as gray lines. Vertical line is the posi-
tion of the outer edge of the lipid interface defined at 10% of the
maximal lipid density at peak position. The center of the bilayer is
at z = 0.
Figure 6. Lipid (solid line), water (dashed line), and total (dotted
line) potential profiles in hydrated (h28-00, black line) and dehy-
bilayer is at z = 0.
(50) Ku? cerka, N.; Tristram-Nagle, S.; Nagle, J. F. Structure of Fully Hydrated
Fluid Phase Lipid Bilayers with Monounsaturated Chains. J. Membr. Biol. 2005,
DOI: 10.1021/la100891x Langmuir 2010, 26(13), 11118–11126
Article Golovina et al.
In our vacuum model dehydration causes a decrease of lipid
potential proportionally to water loss (Figure 6). The presence of
trehalose does not increase the lipid potential in dehydrated
bilayer. Instead, the lipid potential decreases further and drops
under 4 V at high trehalose contents in both v5 and v11 models
(black solid and dashed lines, respectively, in Figure 7).
The PN vector orientation is characterized by the angle θ with
respect to the bilayer normal. In a fully hydrated POPC bilayer
ones reported for previous simulations25,51,52and found in experi-
ments.53The distribution of cos θ in the dehydrated bilayer in the
vacuummodel ismorecomplex. Inthev11-00bilayer thedistribu-
by four Gaussians (Figure 8). This is probably the result of slow
averaged out within the time of observation (the last 50 ns).
When cos θ>0, the PN vector points outward (projection on
The total lipidpotential is thesum of the z-projections of all orien-
cos θ < 0 will decrease the lipid potential due to increase of the
negative z-projection. We estimate the portion of inward oriented
PN vectors, producing negative z-projection of the PN vector, as
the area under the cumulative curve at -1 < cos θ < 0 (Figure 9,
The data were fitted by a Boltzmann distribution (blue line).
Goodness of fit is calculated as adjusted R2=0.85798.
Trehalose, Water, and Total Potentials. The water poten-
tial decreases with dehydration to a greater extent than the lipid
potential (Figure 6). Thus, the water potential does not compen-
positive in v11-00 models and even more so in v5-00. The water
potential becomes less negative with trehalose, and this effect is
concentration dependent (Figure 7).
Trehalose also causes a negative potential itself (model v11-10
as example in Figure 7 inset) on top of the negative potential
Figure 7. PotentialsofPOPC(black),water(red),trehalose(green),
symbols, dashed lines) and at 5.4 waters/lipid (v5, closed symbols,
lipid (black), water (red), and trehalose (green) and total potential
Figure 8. Distributions of PN vector orientations (cos θ) in dehy-
drated models without trehalose (v11-00 and v5-00, gray lines) in
comparison withthe fully hydrated bilayer(h28-00,blackline).The
distributions were fitted with two (A, v11-00) and four (B, v5-00)
Figure 9. Correlation between population of inward oriented PN
vectors (cos θ < 0) and lipid potential. Data represent all the
water contents (indicated by arrows); red circles are models with
different trehalose concentration at water content of 11.7 waters/
lipid; blue triangles are models with different trehalose concentra-
tion and 5.4 waters/lipid. The data are fitted by Boltzmann func-
tion (blue line). Goodness of fit is calculated as adjusted R2=
0.85798. The population of inward oriented PN vectors is calcu-
lated as area under the cumulative curve of PN distribution over
cos θ. The way of calculation is shown in the inset for v5-10 as an
example. In the inset the distribution of PN vector orientations
with fitted Gaussian functions are shown in color. The cumulative
represents the total number of PN vectors oriented inward the bi-
layer. Because the total lipid potential is the sum of the z-projec-
tions of all orientations of the PN vector, increasing populations
due to increase of the negative z-projection.
(51) Pandit, S. A.; Bostick, D.; Berkowitz, M. L. Molecular dynamics simula-
(52) Mukhopadhyay, P.; Monticelli, L.; Tieleman, D. P. Molecular dynamics
simulation of a palmitoyl-oleoyl phosphatidylserine bilayer with Naþ counterions
and NaCl. Biophys. J. 2004, 86 (3), 1601-1609.
(53) Seelig, J. [2H] Hydrogen and [31P] phosphorus nuclear-magnetic-resonance
and neutron-diffraction studies of membranes. Biochem. Soc. Trans. 1978, 6, 40-42.
Langmuir 2010, 26(13), 11118–11126
Golovina et al.Article
of water. The absolute value of the trehalose potential slightly
and trehalose potentials togethercannotcompensatefor eventhe
decreased lipid potential, and the total potential remains slightly
positive in all cases except v5-14 (Figure 7).
In this paper we present results of MD simulations of a single
POPC bilayer at intermediate (11.7 waters/lipid) and low (5.4
waters/lipid) hydration. The separation of lipid bilayers was
provided by a 4 nm layer of empty space, which we call vacuum.
The necessity of separation of dehydrated lipid bilayers became
evident when we failed to obtain the gel state in MD simulations
of POPC bilayer stacks even at 5.4 waters/lipid,24while experi-
mentally a gel state of dehydrated PC bilayers is well established.
We expect that the inconsistency between MD simulations and
experimental data results largely from periodic boundary condi-
tions. There are, however, a number of secondary effects like the
ation dependent quantities), the force field, and several others.
For a more detailed discussion we refer the reader to a recent
contribution by Poger et al.54
When the water layer in the interbilayer space is absent, the
interfaces of neighboring bilayers come into close contact, which
shared by both bilayers. Second, interfaces of adjacent bilayers
can interpenetrate and cause self-spacing. The first factor would
result in an effective almost doubling of water content, and 5.4
waters/lipids in fact can be considered close to 10.8 waters per
lipid. However, even the further decrease of water content to
shown that interpenetration and self-spacing of two adjacent
bilayers in dehydrated stack bilayers may be one of the main rea-
sons of the absence of a gel state of lipid acyl chains at low water
content.24The headgroups overlap is considered as the main
problem in APL calculations in stack bilayers at water contents
less than 12 waters/lipid in X-ray experiments.28
To prevent overlap, the interfaces have to be separated. There
are two requirements for the separating medium: it should not
cancel or at least significantly weaken the interactions between
bilayer interfaces. We have found that in our case 4 nm (2 nm
from each leaflet) of empty space (vacuum) between bilayers is
enough to exclude interactions; the actual value for the vacuum
layer will depend on the details of the simulations, most notably
the cutoff and the implementation of the electrostatics. The
ensemble NPT was converted to NhzpxyT, where the constant
z-directionthe height of the box was fixed and was big enough to
prevent lipid polar group interactions via PBC. In such a model
the absence of interactions between bilayers along the z-axis and
maintaining interactions in the xy-plane results in a stable single
bilayer (400 ns of simulation) in (or near) a gel state both at 5.4
and 11.7 waters/lipid. The (near) gel state of dehydrated POPC
bilayers has been concluded from low APL (0.485 nm2for v5-00
and 0.515 nm2for v11-00, Table 2) accompanied by an increased
order parameter for all carbon atoms (Figure 2), an increased PP
separation, and more pronounced methyl trough (Table 3,
Our dehydrated single bilayer model is different from the one
proposed by Leekumjorn and Sum,30where dehydrated DPPC
(10 waters/lipid) bilayers lost structural integrity within 30 ns of
simulation. Formation of nonbilayer structures of dehydrated
steric component to prevent this lipid from forming nonbilayer
structures.32The lamellar phase reappears even in PE lipids at
sufficiently low hydration.32The reported formation of nonbi-
layer structures of DPPC at 10 waters/lipid30probably results
from an unusual setup of MD simulations. These authors fixed
the size of the xy area of a dehydrated bilayer to the value of
hydrated bilayers (APL 0.645 nm2) under zero lateral compres-
sibility (NPzAT ensemble). Such conditions prevent lateral inter-
our model, using the NPxyhzT ensemble, we allow lipid interac-
tions within xy-plane in a dehydrated bilayer. As a result, we
obtained a bilayer in a gel state, which was stable within 400 ns.
The aim of this work was to study the water replacement
statements: (1) dehydrated membranes are in a gel state; (2) re-
hydration of the membrane in the gel state causes leakage, which
isdetrimental formembraneintegrity;(3) trehaloseinteractswith
dehydrated lipids and prevents gel phase formation; and (4)
rehydration of membranes in fluid state does not cause leakage.
Using empty space between lipid lamellae, we obtained dehy-
spacing between lipids and promote fluid state of a dehydrated
bilayer. Here, we call a bilayer dehydrated if there is no inter-
the size of the hydration shell (around 12 waters/lipid).
POPC bilayers (at both 11.7 and 5.4 waters/lipid) in a concentra-
tion-dependent manner (Figure 1, Table 2). Increasing APL
correlateswithdecreasingorder parameter (Figure3), PP separa-
tion and depth of the methyl trough (Table 3). These structural
data are consistent with gradual concentration-dependent fluidi-
zation of bilayer core resulted from interactions between lipids
and trehalose.Therefore, our model confirmsthe third statement
of the WRH as well.
However, a dehydrated bilayer in the presence of trehalose is
degree of fluidization of the membrane core (Figure 5, Table 3).
Even at trehalose concentration of 1.4 trehalose/lipid the APL
does not reach the value of the fully hydrated bilayer in both v11
effect on membrane phase transition according to calorimetric
ted bilayer (Figures 2 and 3). Contrary to this, some experiments
show overfluidization of dry DPPC bilayers in the presence of
acyl chains are much more disordered in dehydrated DPPC:
trehalose mixtures (1:2) abovethe phase transitionthan in a fluid
stateofhydratedDPPC bilayer.Theauthors call thisnew typeof
fluid state a λ-phase.7On the other hand, Quinn et al.29did not
find any structural evidence of a new fluid state designated as
λ-phase studying dehydrated trehalose/DPPC (2:1) mixture by a
(54) Poger, D.; Mark, A. E. On the Validation of Molecular Dynamics
Simulations of Saturated and cis-Monounsaturated Phosphatidylcholine Lipid
Bilayers: A Comparison with Experiment. J. Chem. Theory Comput. 2010, 6 (1),
(55) Tsvetkov, T. D.; Tsonev, L. I.; Tsvetkova, N. M.; Koynova, R. D.;
Tenchov, B. G. Effect of trehalose on the phase properties of hydrated and
lyophilized dipalmitoylphosphatidylcholine multilayers. Cryobiology 1989, 26,
DOI: 10.1021/la100891xLangmuir 2010, 26(13), 11118–11126
Article Golovina et al.
1.4 trehalose/lipid at 310 K in comparison with the fully hydra-
ted bilayer, which is in a fluid phase at the same temperature
(Figure 3). This is consistent with the lower APL in comparison
with the fully hydrated bilayer (Table 2). Therefore, our model
showed the trend of increasing APL of dehydrated bilayer by
trehalose, but we failed to obtain the structural characteristics
similar to that of the hydrated bilayer even in the excess of treha-
More structural differences between fully hydrated POPC and
dehydrated POPC/trehalose mixtures are observed in the head-
group region. The average orientation of the PN vector is shifted
paralleltothe membrane interface with dehydration and remains
like that with trehalose (Figures 8 and 9) consistent with NMR
and neutron diffraction experiments. NMR spectra of selectively
results in the choline group aligning more closely with the bilayer
surface.56,57Neutron diffraction indicated a spatial limit for PN
parallel orientation than in the hydrated POPC bilayer without
trehalose.58A similar effect of trehalose on PN orientation might
be expected in a dehydrated bilayer.
The shift of the average PN orientation in dehydrated bilayer
results in the increased proportionofinwardoriented PN vectors
(Figure 9), which decreases the z-component of the lipid dipole
and, consequently, the lipid potential (Figure 6). The absolute
value of water potential decreases with dehydration (Figure 6)
and decreases further with trehalose (Figure 7) probably due to
to trehalose (Figure 5). Although trehalose creates a negative
potential, both negative potentials of trehalose and water cannot
compensatefor the decreased lipid potential,and the total poten-
tial remains slightly positive in dehydrated bilayer at all trehalose
concentrations, while the total potential in hydrated POPC
bilayers (h28-00) is negative, -0.46 V.24
The value of the total potential of hydrated POPC bilayer is
in agreement with the experimental data on dipole potential
(400 mV)59but higher than that in bilayers (220-280 mV).60
The values of the dipole potential obtained in MD are usually
higherthanexperimental values(around 600mV).61Thisislikely
fixed at ε=1. Simulations of dry DOPC bilayers have shown the
decrease of the total potential from 500 mV at 16 waters/lipid to
-300 mV at 11.4 waters/lipid and þ400 mV at 5.4 waters/lipid.27
This is in agreement with our observation of changing the total
potential with dehydration. In the model here the total potential
respectively. Because experimentally the dipole potential can be
obtained for fully hydrated bilayer only, we cannot compare our
results on dehydrated bilayers with experimental data. However,
some support comes from data by Luzardo et al.62where a
monolayers in the presence of trehalose has been observed.
The inversion of the sign of the total dipole potential in dry
bilayers with and without trehalose might provide a new mecha-
nism of membrane protection against dehydration by some
proteins. Late embryogenesis abundant (LEA) proteins have
been proposed to contribute toward desiccation tolerance, but
the actual mechanism of action is unclear.63Yeast HSP 12 was
first identified as a putative heat shock protein but later has been
classified as a LEA-like protein.64Membrane protection by HSP
liposomes and not with either neutral or negatively charged
tial of a bilayer during dehydration from negative to slightly
positive might provide the conditions for interactions with the
LEA-like proteins and thus stabilize dehydrated membranes.
The data presented in this study show that the separation of
dehydrated bilayers by empty space in MD simulations provides
a means to study of the mechanisms of membrane protec-
tion against desiccation by different compounds. This model is
experimentally relevant for two reasons. First, POPC bilayer is
Second, the area per lipid of dry POPC increases when trehalose is
of a dry bilayer with trehalose, the decrease of the phase transition
temperature Tmof dry PC bilayer in the presence of trehalose is
commonly used as an indicator of the increased APL.66
in both stack and vacuum models, so the comparison with the
experiments, which were carried out on multilamellar liposomes,
is still valid for such data.
Our vacuum model partly validates the WRH by showing that
dehydration causes the decrease of APL and gel-state formation,
and trehalose increases APL and fluidizes the core of the dry
bilayer. However, the detailed structure of dry POPC bilayer in
the excess of trehalose is different from a fully hydrated bilayer,
particularly in the headgroup region.
Acknowledgment. This work was partly supported by project
no. 10195 from the Dutch Foundation for Technological Re-
search STW (E.A.G.) and partly by the NATO Science Program
(NATO collaborative linkage grant LST.CLG.980168 (A.V.G.
Computing Center of Moscow State University. The supercom-
puter “Chebyshev” was used for all modeling studies.
(56) Bechinger, B.; Seelig, J. Conformational changes of the phosphatidylcho-
line headgroup due to membrane dehydration. A 2H-NMR study. Chem. Phys.
Lipids 1991, 58, 1-5.
(57) Ulrich, A. S.; Watts, A. Molecular response of the lipid headgroup to
bilayer hydration monitored by 2H-NMR. Biophys. J. 1994, 66, 1441-1449.
(58) Bechinger, B.; Macdonald, P. M.; Seelig, J. Deuterium NMR studies of the
interactions of polyhydroxyl compounds and of glycolipids with lipid model
membranes. Biochim. Biophys. Acta 1988, 943, 381-385.
(59) Brockman, H. Dipole potential of lipid membranes. Chem. Phys. Lipids
1994, 73, 57-79.
for its detection. Adv. Colloid Interface Sci. 2001, 89-90, 263-281.
(61) Berkowitz, M. L.; Bostick, D. L.; Pandit, S. Aqueous solutions next to
phospholipid membrane surfaces: insights from simulations. Chem. Rev. 2006,
(62) Luzardo, M. C.; Amalfa, F.; Nu? nez, A. M.; Diaz, S.; Lopez, A. C. B.;
of lipid bilayers. Biophys. J. 2000, 78, 2452-2458.
(63) Chakrabortee, S.; Boschetti, C.; Walton, L. J.; Sarkar, S.; Rubinsztein,
D. C.; Tunnacliffe, A. Hydrophilic protein associated with desiccation tolerance
exhibits broad protein stabilization function. Proc. Natl. Acad. Sci. U.S.A. 2007,
104 (46), 18073-18078.
(64) Mtwisha, L.; Brandt, W.; McCready, S.; Lindsey, G. G. HSP 12 is a LEA-
like protein in Saccharomyces cerevisiae. Plant Mol. Biol. 1998, 37, 513-521.
(65) Sales, K.; Brandt, W.; Rumbak,E.; Lindsey, G. The LEA-like protein HSP
12 in Saccharomyces cerevisiae has a plasma membrane location and protects
membranes against desiccation and ethanol-induced stress. Biochim. Biophys.
Acta 2000, 1463, 267-278.
Crowe, L. M. Membrane stabilization in the dry state. Comp. Biochem. Physiol.,
Part A: Mol. Integr. Physiol. 1997, 117 (3), 335-341.