Structure and phase transformations of DPPC

lipid bilayers in the presence of nanoparticles:

insights from coarse-grained molecular

dynamics simulations

J. P. Prates Ramalho1, P. Gkeka§ and L. Sarkisov2,*

1
Departamento de Química, Universidade de Évora, R. Romão Ramalho, 59, 7000-

671 Évora,

Portugal

2
School of Engineering, University of Edinburgh, Edinburgh.

email: [email protected]

RECEIVED DATE (to be automatically inserted after your manuscript is

accepted if required according to the journal that you are submitting your paper

to)

Lipid bilayers in the presence of nanoparticles.

* Corresponding author: [email protected].

§
Current address: Biomedical Research Foundation Academy of Athens 4, Soranou
Ephessiou, 115 27 Athens

1
In this article we investigate fluid-gel transformations of a DPPC lipid bilayer in the

presence of nanoparticles, using coarse-grained molecular dynamics. Two types of

nanoparticles are considered, specifically a 3nm hydrophobic nanoparticle located in

the core of the bilayer and a 6nm charged nanoparticle located at the interface

between the bilayer and water phase. Both negatively and positively charged

nanoparticles at the bilayer interface are investigated. We demonstrate that the

presence of all types of nanoparticles induces disorder effects in the structure of the

lipid bilayer. These effects are characterized using computer visualization of the gel

phase in the presence of nanoparticles, radial distribution functions and order

parameters. The 3nm hydrophobic nanoparticle immersed in the bilayer core and the

positively charged nanoparticle at the bilayer surface have no effect on the

temperature of the fluid-gel transformation, compared to the bulk case. Interestingly, a

negatively charged hydrophobic nanoparticle located at the surface of the bilayer

causes slight shift of the fluid-gel transformation to a lower temperature, compared to

the bulk bilayer case.

Key words: Lipid bilayer, nanoparticle, phase transition, coarse-grained simulation,

molecular dynamics, MARTINI.

2
I. Introduction

Recent advances in the available experimental techniques to synthesize

nanoparticles from a variety of starting materials and with a well controlled geometry,

size distribution and surface chemistry opened new and unprecedented opportunities

in using these nanoparticles for drug delivery, imaging and as antimicrobial agents.

Rational design of multifunctional nanoparticles with programmable functionalities

requires fundamental understanding of how they interact with lipid membranes.

Experimental studies in the field report a number of possible mechanisms of

interaction between nanoparticles and lipid membranes. These mechanisms depend on

the morphology of nanoparticles (size, shape), surface chemistry and charge as well as

the characteristics of the environment, including type of the cell membrane, pH, and

interaction with other biological entities present in the system. For example, in an

intriguing study of Verma and co-workers, it was demonstrated that the internalization

mechanisms of spherical nanoparticles in the fibroblast cells strongly depend on the

distribution of hydrophobic and hydrophilic domains on the surface of the

nanoparticles1. The authors observed that uniformly polar nanoparticles can be

internalized by the cell via an endocytotic pathway only. On the contrary,

nanoparticles with the surface featuring ordered hydrophobic and hydrophilic stripes

were able to translocate through the cell membrane via some direct mechanism,

independent of endocytosis. The details, however, of this direct mechanism on a

molecular level and precisely how it is initiated by the regular surface patterns remain

unclear. Interactions of fully hydrophobic silver nanoparticles with

dipalmitoylphosphatidylcholine (DPPC) lipid bilayers were explored by Bothun 2. It

was shown that hydrophobic nanoparticles tend to accumulate inside lipid bilayers,

3
and, if present in sufficient concentrations (more than 15:1 w/w DPPC/nanoparticle

ratios), lead to a lowered melting temperature of the gel phase. In another example,

single component phosphocholine bilayers in the presence of charged nanoparticles

have been investigated by Wang and co-workers3. Their results suggest, that charged

nanoparticles position themselves at the bilayer-water interface with negatively

charged nanoparticles inducing local gelation in the fluid bilayers, whereas positively

charged nanoparticles cause local fluidization (disordering) in the gel phase. The

effect of nanoparticle size on the stability of lipid membranes was investigated by

Roiter et al., using AFM measurements4. It was shown that silica nanoparticles in a

particular size range (between 1.2nm and 22nm) can cause formation of holes and

defects in dimyristoylphosphatidylcholine (DMPC) bilayers 4. These are only a few

examples of the recent studies of the membrane-nanoparticles phenomena, varying

broadly in the type of systems and nanoparticles under investigation, conditions,

methods and observations. It is clear that systematic nanoparticle design requires a

general fundamental framework within which membrane-nanoparticle systems can be

described on a molecular level, disparate experimental observations explained and

rationalized, and predictions on nanoparticle behavior as a function of its morphology

made.

In principle, this framework can be developed using molecular modeling. In order

to construct an appropriate model it is important to consider time and length scales

involved in the phenomena of interest. The nanoscale dimension of the particles

implies nanoscale linear dimensions (nanometers and tens of nanometers) of the

model lipid bilayer patches with which they interact. Depending on the size, these

patches should contain between a hundred and several thousands of lipid molecules.

With a typical hydration regime of about 30-50 water molecules per lipid, the model

4
should thus include between thousands and tens of thousands of water molecules.

Routine exploration of the systems of this size is impracticable within fully atomistic

models of the components. Several theoretical models of colloidal and nanoparticle

interactions with a lipid membrane have been proposed over the years, including

those based on the Helfrich Hamiltonian and mean field theories 5-11. However, these

methods often omit important finer details on the structure of the membrane, solvent

and other properties. Alternatively, coarse-grained (CG) models have recently been

playing an increasingly important role in the studies of biological membranes 12. In

these models several atoms are represented as a single interaction site, and both

implicit and explicit solvent models have been developed. The models have been

applied to investigate the behavior of single and multicomponent lipid bilayers,

vesicles and micelles, as well as interaction of these entities with other species such as

cholesterol, peptides and proteins13,14. Recently, several studies have emerged with a

particular focus on bilayer and vesicle interaction with nanotubes, fullerenes and other

nanoparticles15-23.

The idea of this article is to employ molecular modeling to understand the

structure of lipid bilayers and pathways of gel phase formation in the presence of

nanoparticles. In this preliminary work, we will concentrate on two specific cases

reflecting recent experimental studies: a hydrophobic nanoparticle embedded in the

core of a lipid bilayer and a charged nanoparticle at the surface of the bilayer. To

investigate these phenomena on a molecular level, we adopted a coarse-grained

forcefield developed by Marrink and co-workers that has been employed to

investigate the kinetics of fluid-gel phase transformations 24-27 and, at the same time,

has been also recently applied to study behavior of the DPPC bilayers in the presence

5
of several types of nanoparticles18,21,22, and thus seems to provide a unified framework

for the objectives of this research.

Before we formulate the objectives of this study and the anticipated outcomes,

it is important to briefly review several general concepts associated with fluid-gel

phenomena in lipid bilayers and key observations that have emerged from the studies

of fluid-gel transformations by Marrink and co-workers using an earlier version of the

MARTINI forcefield and single component DPPC bilayers as an exemplary and most

characterized system25. The most physiologically relevant phase of a lipid bilayer is

the disordered fluid phase Lα. Upon cooling, this phase undergoes a phase transition to

a gel phase Lβ, characterized, among several other available properties, by an ordered

(but not crystalline) structure, substantially lower area per lipid compared to the fluid

phase, and at the same time still substantial lateral mobility of the lipids compared to

the proper crystalline phase. Several variants of the gel phase L β have been observed

with either perpendicular average orientation of the lipid molecules to the plane of the

bilayer (Lβ) or oriented at a tilt angle (L β’). For DPPC, the fluid-gel transition

temperature is 315K. In reality, the actual temperature at which transformation from

fluid to gel phase takes place may deviate from the equilibrium phase transition

temperature towards lower values. Development of a new phase, as described within

the nucleation theory approach, requires formation of a gel domain of a critical size.

Within the kinetic theory formalism it is possible to show that the lower the

temperature of the system is (compared to the transition temperature), the smaller the

size of the critical domain. The time required for the formation of the critical domain

(or nucleus) also decreases with the increasing degree of subcooling. In the opposite

process, melting of a gel occurs through the formation and development of defects,

6
which according to some theoretical considerations may also involve formation of an

intermediate hexatic phase28.

As a result, the phase transformation from fluid to gel and from gel to fluid

exhibits temperature hysteresis, which is non-equilibrium, kinetically controlled

phenomena. As these processes do not correspond to the thermodynamic equilibrium,

they depend on a number of parameters of the system, including heating and cooling

regimes, with the system often trapped in various types of metastable states,

characterized by very long relaxation timescales. Similarly, in computer simulations

of the CG model of DPPC bilayer, Marrink and co-workers observed that the

temperature at which phase transformation takes place depends on the system size and

the simulation time25. It is lower than the true phase transition temperature for

protocols where the system is cooled and experiences a fluid-gel transformation; it is

higher than the phase transition temperature for the cases where the system is heated

to undergo melting; but in both cases the transformation temperature approaches the

actual bulk phase coexistence temperature (which is estimated to be 295K for DPPC

within the employed coarse-grained model25) as the system size and simulation time

are increased.

In this study we focus on the impact of different types of nanoparticles on the

fluid-gel transformation rather than attempt to identify the location of the true phase

coexistence. Presence of the nanoparticle may influence formation and development

of the new phase and this is what we explore, along with the structural characteristics

of the fluid and gel phases in the presence of these nanoparticles.

II. Methodology

7
 All species in this study are described using the coarse-grained model of lipids and

peptides recently proposed by Marrink and co-workers (MARTINI) 26,27. Briefly, in

this forcefield every four heavy atoms (i.e. not hydrogens) are represented by one

effective bead of 0.47nm in diameter (with an exception made for the ring structures).

There are four primary types of beads representing different levels of interaction,

specifically polar (P), apolar (C), nonpolar (N), and charged (Q). Within each primary

type, several subtypes are available to describe more accurately the overall chemical

nature of the represented group of atoms and to reflect hydrogen-bonding capability

and different levels of polarity. Within this description, polar particles are attracted to

each other more strongly than to hydrophobic particles and this captures the effect of

hydrophobicity. In addition to these effective interactions (implemented via the

Lennard-Jones potential), Coulombic interactions in the system can be described by

using particles of type Q bearing an appropriate charge. Full details of the forcefield

along with the validation case studies are available in the original publications26, 27.

Using this model we consider a system of 4096 DPPC lipid molecules, about

120000 water molecules (with the exact number depending on the system) and either

one, or no nanoparticles (for the bulk lipid bilayer systems). Two types of

nanoparticles are considered. A fully hydrophobic nanoparticle consists of 271 C1

type beads (in the MARTINI notation representing the lowest level of polarity) placed

on the surface of a 3nm (in diameter) hollow sphere (making the collision diameter of

the nanoparticle about 3.47nm). Details of the nanoparticle construction protocol can

be found in the Supplemental Information file. The rigidity of the nanoparticle and the

arrangement of the beads on the surface are maintained via a system of bonds and

distance restraints between the beads. The second type is a charged nanoparticle made

of 1108 coarse-grained beads placed on the surface of a 6nm (in diameter) hollow

8
sphere (making the collision diameter of the nanoparticle about 6.47nm). The

negatively charged nanoparticle has 995 C1 beads and the rest of the beads are

randomly distributed Q0 beads with -1e charge, which corresponds to the surface

charge density of -1.0 e/nm2 (comparable to the charge density of -0.91 e/nm 2,

considered by Wang and co-workers3). For the positively charged nanoparticle, we

change the charge of Q0 beads to +1e, giving the same surface charge density.

Coarse-grained models of the DPPC lipid, the 3nm nanoparticle and negatively

charged 6nm nanoparticle are shown in Figure 1 to illustrate the relative sizes of these

species. In the systems where the charged nanoparticle is considered the

electroneutrality of the system is maintained via an appropriate number of sodium

ions.

In this work, all simulations are performed with the GROMACS simulation

package, version 3.3.3 and 429, 30. The simulations are carried out under the periodic

boundary conditions at constant temperature and pressure. The temperature of the

system is maintained using the Berendsen thermostat with a relaxation time of 1ps 31.

The pressure of the system is maintained at 1 bar using the Berendsen barostat with a

time constant of 5ps and a compressibility of 4.5∙10−5bar−1 under the anisotropic

conditions31. The non-bonded potential energy functions are cut off and shifted at

12Å, with forces smoothly decaying between 9Å and 12Å for the van der Waals

forces and from 0Å to 12Å for the electrostatic forces. The simulations are performed

using a 25fs integration time step. The effective dynamics appear to be faster in the

coarse-grained model due to smoothed potentials. A factor of four is commonly

applied to relate time elapsed in the coarse-grained model to what would be

equivalent in the fully atomistic simulations (i.e. 1fs in the coarse-grained model

corresponds to 4fs in atomistic simulations)24,25. We do not apply this scaling and all

9
the values presented here for the simulation time (including the reference studies) are

in the original, unscaled units.

Systems containing nanoparticles are constructed using a pre-assembled bulk lipid

bilayer at 323K as a starting point. Studies on the 3nm hydrophobic nanoparticle

consider the nanoparticle located in the core of the bilayer as shown in Figure 2, left

panel. This configuration reflects the experimental studies of Bothun on hydrophobic

silver nanoparticles, embedded in the DPPC lipid bilayers 2. During all the simulations

that followed, the hydrophobic nanoparticle remained in the core of the bilayer. In the

second setup, the 6nm charged nanoparticle is placed in the bulk water phase. During

the preliminary equilibration run the nanoparticle positions itself at the bilayer

interface causing the distortion of the bilayer as shown in Figure 2, right panel.

Similar behavior is also observed for the positively charged nanoparticle (see the

Supplemental Information file). This setup is aimed to reflect the experimental

conditions in the studies of Wang and co-workers3.

Several properties are considered for the structural characterization of the lipid

phases. The second-rank order parameter characterizes the

alignment of lipid molecules with the bilayer normal. Here, θ is the angle between the

direction of the bond formed by two coarse-grained beads and the bilayer normal,

with corresponding to perfect alignment, to perfect anti-alignment

and corresponding to a random orientation of molecules with respect to the

bilayer normal. To characterize the gel phase and to detect its formation and

development, we employ the same criteria as in the work of Marrink et al32.

Specifically, we focus on the hydrophobic beads of the lipid tails in the position

marked by the red rectangle for the DPPC molecule in Figure 1. Counted from the

lipid head, this is the second hydrophobic C-type bead in each lipid tail in the lipid

10
molecule and so for future reference throughout the article we call them C 2 beads. For

each leaflet of the bilayer, positions of these beads are projected on the plane. A bead

is considered to belong to the gel phase if it has exactly six neighbors and at least five

of them are within a 0.75nm distance, corresponding to the first minimum in the pair

distribution functions for these beads in the gel phase. Several clusters of the gel

phase may occur or coexist at the same time. Two beads that have been identified as

belonging to the gel phase also belong to the same cluster if they are within 0.55nm

(0.51nm in the work of Marrink and co-workers 25) distance from each other, slightly

further than the first maximum in the radial pair distribution function. In this article

we explore these properties in the bulk lipid bilayer systems and in the presence of

nanoparticles, using computer visualizations of the lipid phases, two dimensional (2D)

radial distribution functions for C2 beads of the lipid tails and order parameters.

III. Results

For the systems described in the methodology section we perform a series of

simulations in order to investigate the fluid-gel transformation processes. For all

systems we use the following protocol. The starting configuration of the system

corresponds to an equilibrium configuration at T=323K (the fluid phase). The system

is then instantly brought to a temperature of interest. At each temperature, the

simulation is run for 200ns. If the phase transformation is detected during this

simulation (via monitoring of various properties of the system, such as the surface

area per lipid), the simulation continues until the formation of the gel phase is

completed. Figure 3 compares the surface area of the lipid bilayer per lipid molecule

at different temperatures for all systems considered in this study. Let us first focus on

the bulk case shown in Figure 3 as red circles. At 290K and above, the DPPC lipid

11
bilayer remains in the fluid phase, with the surface area per lipid linearly increasing

with temperature. As has been shown in the previous studies, the forcefield of

Marrink and co-workers adequately reproduces many properties of the lipid bilayers

(for example, for DPPC, the surface area per lipid at 323K S=0.64nm 2 obtained from

these simulations reproduces the experimental value of Sexp=0.64nm2 at this

temperature)24,25. At T=288K, transformation from fluid to gel phase takes place.

Within the employed forcefield the fluid-gel coexistence temperature has been

estimated at 295K for DPPC24. In the original study of Marrink and co-workers, the

authors noted that the temperatures of the phase transformations approach the true

transition temperature as the system size and the simulation time increase, with the

largest investigated system of 2048 lipids going from fluid to gel phase at about 285K

under 25ns of simulation time (they also looked at an even larger system of 8192

lipids, but for a specific purpose to investigate gel domains) 25. Here we consider a

system twice the size (4096 lipids), simulated on substantially longer timescales

(200ns) and this seems to bring the phase transformation temperature closer to the

phase coexistence point. The formed gel phase is characterized by the lower surface

area per lipid and higher order of the molecules as observed from the radial

distribution function g(r) for C2 beads and the structure order parameter P2 (Figure 4

and Figure 5).

In Figure 6 we visualize the structure of the fluid and gel phases using the Voronoi

analysis. The cyan color corresponds to the C 2 beads of a single lipid bilayer leaflet

that have exactly six neighbors and are in the gel phase formation, whereas other

colors correspond to C2 beads having five, seven and occasionally other number of

neighbors. The disordered structure of the fluid phase is contrasted by the ordered gel

12
phase, forming a single percolating cluster with only few defects (the two dislocations

spanning the system are clearly visible).

Similar analysis is applied to the behavior of DPPC lipid bilayers in the presence of

nanoparticles (we predominantly focus on the 3nm hydrophobic nanoparticle and 6nm

negatively charged nanoparticle to illustrate the point). We begin with the computer

visualization of the gel phases at 278K. In the case of the hydrophobic nanoparticle,

some local disorder is induced in the direct vicinity of the nanoparticle, and the

remaining structure is similar to that of the bulk bilayer at the same temperature.

Image on the right in Figure 7 shows a computer visualization of the gel phase in the

presence of the negatively charged nanoparticle at the surface of the bilayer. Again,

beads in the gel formation are cyan and disordered arrangements of particles use other

colors, depending on the number of neighbors for each bead. Although, this leaflet is

the one that is further away from the nanoparticle (i.e. not the one the nanoparticle

actually touches), it is clear that the presence of the nanoparticle induces disorder in

its vicinity and throughout the bilayer. Radial distribution functions provide an

alternative way to examine the local disorder effects caused by the presence of

nanoparticles. This analysis is presented in Figure 8. The black lines in both left and

right panels of the figure correspond to the 2D radial distribution function for C 2

beads in the pure lipid bilayer at 278K. The red lines correspond to the same functions

within a 6nm by 6nm patch around (in the case of the embedded nanoparticle) or

under the nanoparticle (if it is at the surface) whereas the blue lines are the radial

distribution functions corresponding to a 10nm by 10nm area in the vicinity of a

nanoparticle. From this figure, the radial distribution functions reflect disorder in the

direct vicinity of a nanoparticle compared to the pure bilayer. With a larger portion of

the system included in the analysis, this function approaches the pure bilayer values.

13
From the first maximum of the blue line, it is also evident that this trend is somewhat

faster for the smaller, embedded nanoparticle.

Next, we examine the behavior of order parameters in the vicinity of nanoparticles.

We focus on just three order parameters, between the choline group and the

phosphate group (NC3-PO4) and for the two last bonds of one of the lipid tail chains

(C2A-C3A and C3A-C4A bonds), averaged over multiple system realizations. This

analysis is shown in Figure 9. In this figure, zero value of x-axis corresponds to the

projection of the center of mass of the nanoparticle on the plane of the bilayer. This

figure shows that right under the nanoparticle or in its direct vicinity the values of the

selected order parameters deviate significantly from the bulk reference values and

approach zero, indicating complete disorder of the structure. The most significant

deviations from the bulk values are observed on the length scale comparable to the

size of the nanoparticle (3-4nm in the figure). Again, these effects seem to be

somewhat more pronounced for the larger nanoparticle at the surface of the bilayer

(right panel in Figure 9), compared to the smaller nanoparticle in the core of the

bilayer (left panel in Figure 9). In both cases, the order parameters for C 2A-C3A and

C3A-C4A never fully reach the bulk values. This is associated with the defects present

in the system.

We also briefly explore the effect of the nanoparticle presence on the fluid-gel

transformation from a kinetic perspective. Figure 10 examines maximum size of the

gel phase cluster as a function of time at 284K. Large oscillations of the cluster size

which can be observed, for example, towards the end of the simulation with the 6nm

negatively charged nanoparticle, are associated with the fluctuations in the system

volume and the percolated cluster being temporarily split by the periodic boundary. If

we focus on the time elapsed between 10 and 50ns, one may notice that the rate of the

14
gel phase growth (the slope of the line) is higher for the bulk lipid bilayer case,

compared to that in the presence of nanoparticles. This effect is rather small for the

3nm hydrophobic nanoparticle or 6nm positively charged nanoparticle, but it is more

pronounced for the negatively charged nanoparticle at the bilayer surface (red line in

Figure 10). Computer visualizations of the various states of the system on its way to

the gel phase formation may also provide some useful insights on the details of this

process. These are shown in Figure 11 for the bulk system, and for the systems with

the 3nm nanoparticle in the bilayer core and 6nm negatively charged nanoparticle at

the bilayer surface. At 14ns, all three systems feature a broad distribution of clusters.

In case of the bulk system, the critical nucleus seems to have already formed (in the

lower right region of the system) and is growing. At 110ns and beyond all three

systems are predominantly in the gel phase forming a percolated cluster with several

defects still present. As has been noted before, complete elimination of these defects

may take an extremely long simulation time 32. Presence of these defects explains

some variation in the surface area values for the gel phase in different systems

considered here (Figure 3). At 70ns one may notice that when the nanoparticles are

present, the defects in the structure are grouped around and emanating from the

nanoparticles.

Finally, we return to Figure 3, which summarizes the behavior of the lipid bilayer

surface area per lipid molecule as a function of temperature for all systems. From this

figure, the presence of a hydrophobic nanoparticle, or a positively charged

nanoparticle at the bilayer surface does not significantly impact the fluid-gel

transformation and it takes place at 288K, the same temperature as for the bulk bilayer

case. The presence of a negatively charged hydrophobic nanoparticle at the surface of

the bilayer, however, delays formation of the gel phase, shifting the location of the

15
fluid-gel transformation to a lower temperature of 284K (under the given protocol).

This is consistent with the slower growth of the gel phase in the presence of the

negatively charged nanoparticle presented in Figure 10. The question arises whether

these observations correspond to a particular outcome of a single realization of the

cooling process, or they reflect some general tendencies. To address this issue, we

consider four additional cooling runs for each system (giving total of five independent

realizations of the process) at 280, 284 and 288K. To present our observations in a

compact way, we estimate the probability of the fluid-gel transformation to happen,

based on the outcome of five independent 200ns runs (i.e. P tr is the number of runs

that led to the formation of the gel phase divided by the total number of runs), with

the additional details available in the Supplemental Information file. This is

summarized in Figure 12. At 280K and 284K all simulations for the bulk bilayer and

in the presence of the 3nm hydrophobic nanoparticle or 6nm positively charged

nanoparticle at the surface, lead to the formation of the gel phase (P tr=1). At 288K,

most of the simulations for these three systems lead to the formation of the gel phase

(with some variation between the systems). Interestingly, the behavior of the DPPC

bilayer in the presence of the negatively charged nanoparticle at the surface is

consistently different. First of all, at 288K none of the runs leads to the formation of

the gel phase (Ptr=0). Even at lower temperatures (284K and 280K), in one out of five

runs the system remains in the fluid phase. This trend is discussed in more detail in

the next section.

IV. Discussion

In this study we investigated how the presence of a nanoparticle affects fluid-gel

transformations in DPPC lipid bilayers. For this we employed molecular dynamics

16
simulations based on the MARTINI coarse-grained forcefield and explored

nanoparticles of two primary architectures: a 3nm hydrophobic nanoparticle located in

the core of the bilayer and 6nm charged hydrophobic nanoparticles located at the

bilayer interface. Our conclusions are as follows. We established that the presence of

the nanoparticle either in the core of the bilayer or at the surface leads to the

disordered local structure of the lipid bilayer. This effect is particularly evident from

the second-rank order parameter analysis. For both 3nm hydrophobic nanoparticle

located in the core of the bilayer and 6nm positively charged nanoparticle at the

surface of the bilayer no significant effect on the location of the fluid-gel phase

transformation was detected compared to the bulk bilayer case. For the hydrophobic

nanoparticle in the bilayer core, this is consistent with the earlier experimental

observations, where lipid bilayers could accommodate a significant amount of

hydrophobic nanoparticles without any substantial impact on their overall properties 2.

(Our system corresponds to about 20:1 w/w DPPC/nanoparticle composition and in

this regime no detectable change in the phase behavior can be observed according to

the experimental data2).

A different picture emerged in the case of the negatively charged nanoparticle at

the bilayer surface. The formation of the gel phase was systematically delayed,

leading to a lower temperature of the fluid-gel phase transformation in the presence of

this nanoparticle. On the contrary, an experimentally observed scenario for a

negatively charged nanoparticle at the bilayer surface suggests stronger interaction

between the particle and the positively charged amine groups of the lipid molecules,

leading to a denser packing and higher order of lipids in the vicinity of the

nanoparticle. This in turn should serve as an initial nucleus for the gel phase and

promote, rather than delay, gel phase formation. It is worthwhile to mention here that

17
in a recent study Li and Gu also investigated charged nanoparticles at the surface of

the bilayer using the same forcefield and observed a significant wrapping of the

bilayer around the nanoparticle with almost complete endosome formation 22. The

density of the bilayer also seems to be higher under the negatively charged particle.

We note, however, that their particles were predominantly composed of polar beads

and featured an average surface charge up to 12 times higher than the one considered

here and those employed in the experiments of Wang et al3.

Initially, we proposed the following explanation for this result. The considered

negatively charged nanoparticle causes partial curving of the bilayer around itself.

The surface charges also do not allow a fully hydrophobic nanoparticle to translocate

into the interior of the bilayer. Within our model, however, the density of the negative

charge is not high enough to initiate ordering of the lipid bilayers. As a result, the key

impact of the nanoparticle on the bilayer would be the distortion of the bilayer

surface. The reported effect of the lipid curvature on the fluid-gel transition would

correlate nicely with an earlier observation of Marrink and co-workers, who showed

that in lipid vesicles (20nm in diameter) fluid-phase transformation is shifted to much

lower temperatures compared to planar systems 32. However, lack of these effects for

the positively charged nanoparticle, where a similar distortion of the bilayer was also

observed, points to a different underlying mechanism.

The nature of this mechanism, through which the negatively charged nanoparticle at

the surface of the bilayer affects the fluid-gel transformation, has not been yet

elucidated. One possible investigation route is associated with the details of our model

of nanoparticles. In this study we assumed that the charged nanoparticle is mostly

hydrophobic with discrete charges, randomly located on the surface of the

nanoparticle. This, however, may not be an accurate representation of the

18
experimental system and alternative designs with the same surface charge are possible

(such as a uniform distribution of charge as in the work of Li and Gu 22). It is possible

that the very disorder in the arrangement of the charges in our model is linked to the

disorder induced by the nanoparticle in the underlying lipid bilayer structure, and thus

to the phase behavior of the system. To further probe this hypothesis, the behavior of

the positively charged nanoparticle must be explained from the same perspective.

This, as well fluid-gel transformations as a function of the charge model and surface

charge density will be investigated in a separate study.

Acknowledgements

J.P.P.R. acknowledges the financial support from the “Fundação para a Ciencia e

Tecnologia” (Lisbon, Portugal) through the SFRH/BSAB/955/2009 fellowship. L.S.

would like to thank Mr. Alex Harrison for diligent proofreading and useful

suggestions. This work has made use of the resources provided by the Edinburgh

Compute and Data Facility (ECDF). (http://www.ecdf.ed.ac.uk/). The ECDF is

partially supported by the eDIKT initiative ( http://www.edikt.org.uk).

Supporting Information Available.

The Supporting Information file contains details on the nanoparticle construction

protocol, examples of the simulation run parameters, additional data on the lipid

surface area per lipid molecule for various realizations of the cooling process,

additional data for the positively charged nanoparticle.

FIGURE CAPTIONS

19
Figure 1: Computer visualization of the coarse-grained DPPC molecule (left,

about 2.5nm in size), fully hydrophobic 3nm nanoparticle (center) and negatively

charged 6nm nanoparticle (right), with negative beads bearing -1e charge shown

in red. The red frame in the visualization of the DPPC molecule shows the

position of C2 beads of the lipid tails used to calculate various structural

parameters of the fluid and gel phases.

Figure 2: Typical configurations of the systems containing nanoparticles

(T=323K). On the left side of the figure, the 3nm hydrophobic nanoparticle is

immersed in the core of the DPPC bilayer (side and top views are provided,

respectively); on the right side of the figure, a system containing the 6nm

negatively charged nanoparticle is shown. The nanoparticle positions itself on the

surface of the bilayer, causing partial wrapping of the bilayer (side and top views

are provided, respectively). Similar behavior is also observed for the positively

charged nanoparticle (see the Supplemental Information file). The dimensions of

the bilayer patch are about 30nm by 40nm, depending on the system and

temperature.

Figure 3: Surface area S(nm2) per lipid molecule as a function of temperature

T(K) for the bulk DPPC bilayer (red circles and solid black line) and in the

presence of the 3nm hydrophobic nanoparticle (white circles), the 6nm

negatively charged nanoparticle (black triangles and dashed line) and the 6nm

positively charged nanoparticle (grey triangles).

20
Figure 4: 2D radial distribution function g(r) calculated for C 2 beads of the lipid

tails as a function of the distance r[nm]. The fluid phase at 323K is shown on the

left, the gel phase at 278K is shown on the right.

Figure 5: Order parameter P2 for the bonds in the lipid molecule. Location of the

various beads within the DPPC molecule is shown on the left. Closed symbols

correspond to the fluid phase at 323K and open symbols correspond to the gel

phase at 278K, respectively.

Figure 6: Computer visualization of C 2 beads in a single lipid bilayer leaflet.

Cyan color corresponds to C2 beads with exactly six neighbors and in the gel

formation, whereas other colors depict various other situations. A typical fluid

phase configuration at 323K is shown on the left, and a gel phase configuration

at 278K is shown on the right. The lipid patch dimensions are about 30nm by

40nm depending on the temperature and phase.

Figure 7: Computer visualization of C 2 beads in a single lipid bilayer leaflet in

the presence of the 3nm hydrophobic nanoparticle (left, the position of the

nanoparticle is self-evident) and the 6nm negatively charged nanoparticle (right,

the position of the nanoparticle is shown by the red circle) at 278K. Cyan color

corresponds to C2 beads in the gel phase, with other colors marking disordered

structure.

21
Figure 8: 2D radial distribution function g(r) calculated for C 2 beads of the lipid

tails as a function of the distance r[nm] in the presence of the 3nm nanoparticle

in the bilayer core (left) and 6nm negatively charged nanoparticle at the bilayer

surface (right). Black lines correspond to the bulk reference case at 278K, red

lines correspond to the 6nm by 6nm area in the vicinity of the nanoparticle, and

blue lines to the 10nm by 10nm area, respectively.

Figure 9: Order parameter P 2 for NC3-PO4 bond (black line), C 2A-C3A bond

(blue line) and C3A-C4A bond (red line) as a function of the distance r[nm] from

the location of the 3nm hydrophobic nanoparticle (left) and 6nm negatively

charged nanoparticle on the surface of the bilayer (right) at 278K. r=0 in this

figure corresponds to the xy position of the center of mass of the nanoparticle.

Bulk values of these order parameters are shown as dashed lines of the

corresponding colors. Vertical black dashed line in the left graph marks the

radius of the 3nm hydrophobic nanoparticle.

Figure 10: Maximum cluster size of C2 beads in the gel phase as a function of

simulation time (log scale) for the bulk system (black line), systems with the 3nm

hydrophobic nanoparticle in the bilayer core (blue line), 6nm negatively charged

nanoparticle (red line) and 6nm positively charged nanoparticle (green line) at

284K.

22
Figure 11: Computer visualizations of the gel phase development in the bulk

system (top three graphs), and in the presence of the 3nm nanoparticle (3 graphs

in the middle) and 6nm negatively charged nanoparticle, captured at 14, 70 and

110ns at 284K. The gel cluster of the largest size is shown in cyan color, with the

clusters of other sizes assigned colors at random (beads in fluid phase are not

shown).

Figure 12: Estimated probability (based on five 200ns runs) P tr of the lipid

bilayer transformation to the gel phase as a function of temperature T(K) for the

bulk DPPC bilayer (red circles and solid black line) and in the presence of the

3nm hydrophobic nanoparticle (white circles and dotted line), 6nm negatively

charged nanoparticle (black triangles and dashed line) and 6nm positively

charged nanoparticle (grey triangles).

References

1. Verma, A.; Uzun, O.; Hu, Y. H.; Hu, Y.; Han, H. S.; Watson, N.; Chen, S. L.;

Irvine, D. J.; Stellacci, F., Surface-structure-regulated cell-membrane penetration by

monolayer-protected nanoparticles. Nature Materials 2008, 7, (7), 588-595.

23
2. Bothun, G., Hydrophobic silver nanoparticles trapped in lipid bilayers: Size

distribution, bilayer phase behavior, and optical properties. Journal of

Nanobiotechnology 2008, 6, (1), 13.

3. Wang, B.; Zhang, L. F.; Bae, S. C.; Granick, S., Nanoparticle-induced surface

reconstruction of phospholipid membranes. Proceedings of the National Academy of

Sciences of the United States of America 2008, 105, (47), 18171-18175.

4. Roiter, Y.; Ornatska, M.; Rammohan, A. R.; Balakrishnan, J.; Heine, D. R.;

Minko, S., Interaction of nanoparticles with lipid membrane. Nano Letters 2008, 8,

(3), 941-944.

5. Lipowsky, R.; Dobereiner, H. G., Vesicles in contact with nanoparticles and

colloids. Europhysics Letters 1998, 43, (2), 219-225.

6. Deserno, M.; Gelbart, W. M., Adhesion and wrapping in colloid-vesicle

complexes. Journal of Physical Chemistry B 2002, 106, (21), 5543-5552.

7. Deserno, M., When do fluid membranes engulf sticky colloids? Journal of

Physics-Condensed Matter 2004, 16, (22), S2061-S2070.

8. Livadaru, L.; Kovalenko, A., Fundamental mechanism of translocation across

liquidlike membranes: Toward control over nanoparticle behavior. Nano Letters 2006,

6, 78-83.

9. Ginzburg, V. V.; Balijepalli, S., Modeling the Thermodynamics of the

Interaction of Nanoparticles with Cell Membranes. Nano Letters 2007, 7, (12), 3716-

3722.

24
10. Pogodin, S.; Baulin, V. A., Coarse-grained models of phospholipid

membranes within the single chain mean field theory. Soft Matter 2010, 6, (10), 2216-

2226.

11. Pogodin, S.; Baulin, V. A., Can a Carbon Nanotube Pierce through a

Phospholipid Bilayer? ACS Nano 2010, 4, (9), 5293-5300.

12. Venturoli, M.; Sperotto, M. M.; Kranenburg, M.; Smit, B., Mesoscopic models

of biological membranes. Physics Reports-Review Section of Physics Letters 2006,

437, (1-2), 1-54.

13. Murtola, T.; Bunker, A.; Vattulainen, I.; Deserno, M.; Karttunen, M.,

Multiscale modeling of emergent materials: biological and soft matter. Physical

Chemistry Chemical Physics 2009, 11, (12), 1869-1892.

14. Marrink, S. J.; de Vries, A. H.; Tieleman, D. P., Lipids on the move:

Simulations of membrane pores, domains, stalks and curves. Biochimica Et

Biophysica Acta-Biomembranes 2009, 1788, (1), 149-168.

15. Noguchi, H.; Takasu, M., Adhesion of nanoparticles to vesicles: A Brownian

dynamics simulation. Biophysical Journal 2002, 83, (1), 299-308.

16. Nielsen, S. O.; Ensing, B.; Ortiz, V.; Moore, P. B.; Klein, M. L., Lipid bilayer

perturbations around a transmembrane nanotube: A coarse grain molecular dynamics

study. Biophysical Journal 2005, 88, (6), 3822-3828.

17. Alexeev, A.; Uspal, W. E.; Balazs, A. C., Harnessing Janus nanoparticles to

create controllable pores in membranes. Acs Nano 2008, 2, (6), 1117-1122.

25
18. Li, Y.; Chen, X.; Gu, N., Computational Investigation of Interaction between

Nanoparticles and Membranes: Hydrophobic/Hydrophilic Effect. Journal of Physical

Chemistry B 2008, 112, (51), 16647-16653.

19. Shi, X. H.; Kong, Y.; Gao, H. J., Coarse grained molecular dynamics and

theoretical studies of carbon nanotubes entering cell membrane. Acta Mechanica

Sinica 2008, 24, (2), 161-169.

20. Wong-Ekkabut, J.; Baoukina, S.; Triampo, W.; Tang, I. M.; Tieleman, D. P.;

Monticelli, L., Computer simulation study of fullerene translocation through lipid

membranes. Nat Nano 2008, 3, (6), 363-368.

21. Lin, X. B.; Li, Y.; Gu, N., Nanoparticle's Size Effect on Its Translocation

Across a Lipid Bilayer: A Molecular Dynamics Simulation. Journal of Computational

and Theoretical Nanoscience 2010, 7, (1), 269-276.

22. Li, Y.; Gu, N., Thermodynamics of Charged Nanoparticle Adsorption on

Charge-Neutral Membranes: A Simulation Study. Journal of Physical Chemistry B

114, (8), 2749-2754.

23. D'Rozario, R. S. G.; Wee, C. L.; Wallace, E. J.; Sansom, M. S. P., The

interaction of C-60 and its derivatives with a lipid bilayer via molecular dynamics

simulations. Nanotechnology 2009, 20, (11).

24. Marrink, S. J.; de Vries, A. H.; Mark, A. E., Coarse grained model for

semiquantitative lipid simulations. Journal of Physical Chemistry B 2004, 108, (2),

750-760.

25. Marrink, S. J.; Mark, A. E., Coarse grained simulation of phase transitions of

lipid membranes. Biophysical Journal 2005, 88, (1), 384A-384A.

26
26. Marrink, S. J.; Risselada, H. J.; Yefimov, S.; Tieleman, D. P.; de Vries, A. H.,

The MARTINI force field: Coarse grained model for biomolecular simulations.

Journal of Physical Chemistry B 2007, 111, (27), 7812-7824.

27. Monticelli, L.; Kandasamy, S. K.; Periole, X.; Larson, R. G.; Tieleman, D. P.;

Marrink, S. J., The MARTINI coarse-grained force field: Extension to proteins.

Journal of Chemical Theory and Computation 2008, 4, (5), 819-834.

28. Kosterlitz, J. M.; Thouless, D. J., ORDERING, METASTABILITY AND

PHASE-TRANSITIONS IN 2 DIMENSIONAL SYSTEMS. Journal of Physics C-

Solid State Physics 1973, 6, (7), 1181-1203.

29. Lindahl, E.; Hess, B.; van der Spoel, D., GROMACS 3.0: a package for

molecular simulation and trajectory analysis. Journal of Molecular Modeling 2001, 7,

(8), 306-317.

30. Van der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.;

Berendsen, H. J. C., GROMACS: Fast, flexible, and free. Journal of Computational

Chemistry 2005, 26, (16), 1701-1718.

31. Berendsen, H. J. C.; Postma, J. P. M.; Vangunsteren, W. F.; Dinola, A.; Haak,

J. R., MOLECULAR-DYNAMICS WITH COUPLING TO AN EXTERNAL BATH.

Journal of Chemical Physics 1984, 81, (8), 3684-3690.

32. Marrink, S. J.; Risselada, J.; Mark, A. E., Simulation of gel phase formation

and melting in lipid bilayers using a coarse grained model. Chemistry and Physics of

Lipids 2005, 135, (2), 223-244.

27
SYNOPSIS TOC

Computer visualization of the DPPC gel phase formation in the presence of a

negatively charged 6nm nanoparticle

28
Figure 1: Computer visualization of the coarse-grained DPPC molecule (left,

about 2.5nm in size), fully hydrophobic 3nm nanoparticle (center) and negatively

charged 6nm nanoparticle (right), with negative beads bearing -1e charge shown

in red. The red frame in the visualization of the DPPC molecule shows the

position of C2 beads of the lipid tails used to calculate various structural

parameters of the fluid and gel phases.

29
Figure 2: Typical configurations of the systems containing nanoparticles

(T=323K). On the left side of the figure, the 3nm hydrophobic nanoparticle is

immersed in the core of the DPPC bilayer (side and top views are provided,

respectively); on the right side of the figure, a system containing the 6nm

negatively charged nanoparticle is shown. The nanoparticle positions itself on the

surface of the bilayer, causing partial wrapping of the bilayer (side and top views

are provided, respectively). Similar behavior is also observed for the positively

charged nanoparticle (see the Supplemental Information file). The dimensions of

the bilayer patch are about 30nm by 40nm, depending on the system and

temperature.

30
Figure 3: Surface area S(nm2) per lipid molecule as a function of temperature

T(K) for the bulk DPPC bilayer (red circles and solid black line) and in the

presence of the 3nm hydrophobic nanoparticle (white circles), the 6nm

negatively charged nanoparticle (black triangles and dashed line) and the 6nm

positively charged nanoparticle (grey triangles).

31
Figure 4: 2D radial distribution function g(r) calculated for C 2 beads of the lipid

tails as a function of the distance r[nm]. The fluid phase at 323K is shown on the

left, the gel phase at 278K is shown on the right.

32
Figure 5: Order parameter P2 for the bonds in the lipid molecule. Location of the

various beads within the DPPC molecule is shown on the left. Closed symbols

correspond to the fluid phase at 323K and open symbols correspond to the gel

phase at 278K, respectively.

33
Figure 6: Computer visualization of C 2 beads in a single lipid bilayer leaflet.

Cyan color corresponds to C2 beads with exactly six neighbors and in the gel

formation, whereas other colors depict various other situations. A typical fluid

phase configuration at 323K is shown on the left, and a gel phase configuration

at 278K is shown on the right. The lipid patch dimensions are about 30nm by

40nm depending on the temperature and phase.

34
Figure 7: Computer visualization of C 2 beads in a single lipid bilayer leaflet in

the presence of the 3nm hydrophobic nanoparticle (left, the position of the

nanoparticle is self-evident) and the 6nm negatively charged nanoparticle (right,

the position of the nanoparticle is shown by the red circle) at 278K. Cyan color

corresponds to C2 beads in the gel phase, with other colors marking disordered

structure.

35
Figure 8: 2D radial distribution function g(r) calculated for C 2 beads of the lipid

tails as a function of the distance r[nm] in the presence of the 3nm nanoparticle

in the bilayer core (left) and 6nm negatively charged nanoparticle at the bilayer

surface (right). Black lines correspond to the bulk reference case at 278K, red

lines correspond to the 6nm by 6nm area in the vicinity of the nanoparticle, and

blue lines to the 10nm by 10nm area, respectively.

36
Figure 9: Order parameter P 2 for NC3-PO4 bond (black line), C 2A-C3A bond

(blue line) and C3A-C4A bond (red line) as a function of the distance r[nm] from

the location of the 3nm hydrophobic nanoparticle (left) and 6nm negatively

charged nanoparticle on the surface of the bilayer (right) at 278K. r=0 in this

figure corresponds to the xy position of the center of mass of the nanoparticle.

Bulk values of these order parameters are shown as dashed lines of the

corresponding colors. Vertical black dashed line in the left graph marks the

radius of the 3nm hydrophobic nanoparticle.

37
Figure 10: Maximum cluster size of C2 beads in the gel phase as a function of

simulation time (log scale) for the bulk system (black line), systems with the 3nm

hydrophobic nanoparticle in the bilayer core (blue line), 6nm negatively charged

nanoparticle (red line) and 6nm positively charged nanoparticle (green line) at

284K.

38
Figure 11: Computer visualizations of the gel phase development in the bulk

system (top three graphs), and in the presence of the 3nm nanoparticle (3 graphs

in the middle) and 6nm negatively charged nanoparticle, captured at 14, 70 and

110ns at 284K. The gel cluster of the largest size is shown in cyan color, with the

clusters of other sizes assigned colors at random (beads in fluid phase are not

shown).

39
Figure 12: Estimated probability (based on five 200ns runs) P tr of the lipid

bilayer transformation to the gel phase as a function of temperature T(K) for the

bulk DPPC bilayer (red circles and solid black line) and in the presence of the

3nm hydrophobic nanoparticle (white circles and dotted line), 6nm negatively

charged nanoparticle (black triangles and dashed line) and 6nm positively

charged nanoparticle (grey triangles).

40