Chemistry and Physics of Lipids 183 (2014) 208217

Contents lists available at ScienceDirect

Chemistry and Physics of Lipids

journal homepage: www.elsevier.com/locate/chemphyslip

Combined force spectroscopy, AFM and calorimetric studies to reveal

the nanostructural organization of biomimetic membranes
C. Surez-Germ b,c, A. Morros a , M.T. Montero b,c, J. Hernndez-Borrell b,c ,
. Domnech b,c, *
a
b
c

Unitat de Biofsica, Departament de Bioqumica i Biologia Molecular, Facultat de Medicina and Centre d'Estudis en Biofsica (CEB), UAB, Spain
Departament de Fisicoqumica, Facultat de Farmcia, UB, Spain
Institut de Nanocincia i Nanotecnologia IN2UB, Barcelona, Catalonia, Spain

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 3 June 2014
Received in revised form 14 July 2014
Accepted 28 July 2014
Available online 2 August 2014

In this work we studied a binary lipid matrix of1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine

(POPE) and1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(10 -rac-glycerol) (POPG), a composition that
mimics the inner membrane of Escherichia coli. More specically, liposomes with varying fractions of
POPG were analysed by differential scanning calorimetry (DSC) and a binary phase diagram of the system
was created. Additionally, we performed atomic force microscopy (AFM) imaging of supported lipid
bilayers (SLBs) of similar compositions at different temperatures, in order to create a pseudo-binary
phase diagram specic to this membrane model. AFM study of SLBs is of particular interest, as it is
conceived as the most adequate technique not only for studying lipid bilayer systems but also for imaging
and even nanomanipulating inserted membrane proteins. The construction of theabove-mentioned
phase diagram enabled us to grasp better the thermodynamics of the thermal lipid transition from a gellike POPE:POPG phase system to a more uid phase system. Finally, AFM force spectroscopy (FS) was used
to determine the nanomechanics of these two lipid phases at 27  C and at different POPG fractions. The
resulting data correlated with the specic composition of each phase was calculated from the AFM phase
diagram obtained. All the experiments were done in the presence of 10 mM of Ca2+, as this ion is
commonly used when performing AFM with negatively charged phospholipids.
2014 Elsevier Ireland Ltd. All rights reserved.

Keywords:
Biomembranes
Phase diagram
Atomic force microscopy
Force spectroscopy

1. Introduction
In strictly physical terms, the plasma membrane can be seen as
the boundary region that separates the discrete mass of the
cytoplasm from its outer environment. This membrane, consisting
mainly of a phospholipid bilayer and proteins interacting in various
ways, is recognized as a heterogeneous structure that provides the
basis not only for cell compartmentalization but also for specic
metabolic processes to take place, among which are processes
from signal or energy transduction to transport of drugs and
metabolites, viral and bacterial infections, or tissue development
and metastasis. In this respect, characterization of the physicochemical properties of the plasma membrane is crucial to
understanding the molecular aspects behind these processes.
Biological membranes contain a complex mixture of lipid species
that, depending on their molecular structure and physicochemical

* Corresponding author. Tel.: +34 934035985; fax: +34 934035987.

E-mail address: odomenech@ub.edu (. Domnech).
http://dx.doi.org/10.1016/j.chemphyslip.2014.07.009
0009-3084/ 2014 Elsevier Ireland Ltd. All rights reserved.

conditions, such as pH, temperature (T) or ionic strength (I), may

show phase separation and become laterally segregated into
nano- or micro-domains (Bagatolli, 2006). How the physical
properties of phospholipid bilayers (e.g. phase segregation, lipid
curvature, elasticity) are related to those found in natural
biomembranes is, at the least, intriguing. Although, the universality of the phospholipid bilayer can be postulated, to assume that
behaviour observed in model systems occurs in natural membranes remains a matter of debate in the eld of membrane
biophysics. Thus, as phospholipids belong to mesomorphic matter,
they can have several physical states, depending on the physicochemical conditions. In excess water, phospholipids form fully
solvated lipid bilayers, which undergo the known phase transition
from the solid-like gel state (Lb) to the uidliquid-crystalline state
(La) at a denite transition temperature (Tm) that is specic to each
species and phospholipid mixture (Houslay and Stanley, 1982).
Integral membrane proteins have one or more segments that are
embedded in the phospholipid bilayer, where they interact strongly
and selectively (Picas et al., 2010a) with the surrounding lipids.
Indeed, there is evidence indicating that the activity and folding of

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

some integral proteins depend on the physicochemical properties of

neighbouring phospholipids (Vitrac et al., 2013). Furthermore, most
of the integral membrane proteins reside almost exclusively within
the La phase (Picas et al., 2010b) and, though this is sometimes
controversial, there is evidence that some proteins are excluded
from the Lb phase (Lee, 2004). In an early paper from Facci's group,
(Seeger et al., 2009a) such behaviour was conclusively demonstrated through atomic force microscopy (AFM) observations, when the
channel protein KcsA was reconstituted in supported lipid bilayers
(SLBs) of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) andpalmitoyl-2-oleoyl-sn-glycero-3-phospho-(10 rac-glycerol) (POPG) and the temperature was reduced below
the Tm of the phospholipid mixture. Then, KcsA monomers were
observed to be excluded from the solid-like domain (dened there
as Lo) and remained segregated in the uid phase (dened there as
Ld) and at the edge between uid and gel phase boundaries (see
Fig. 4 in Seeger's paper). Besides, all integral membrane proteins are
surrounded by a layer of phospholipids, theso-called annular
region, which provides the adequate lateral pressure (Marsh,
2008) and uidity to seal the membrane during the structural
changes in the protein during transport events. It should be noted
that, as shown in the case of lactose permease (LacY) of
Escherichia coli (E. coli), the recruitment of one of the
phospholipid species by the protein is dependent on the POPE:
POPG molar ratio, which is commonly accepted as the biomimetic
binary lipid mixture for LacY (Picas et al., 2010c). Although, a POPE:
POPG phase diagram constructed from differential scanning
calorimetry (DSC) measurements of liposomes in absence of Ca2+
is available (Pozo Navas et al., 2005), it is important to dispose of a
phase diagram for the same system that enables us to predict, at
least approximately, the composition of each phase in SLBs in
presence of Ca2+. As discussed elsewhere (Reviakine and Brisson,
2000) Ca2+ is required to enhance SLBs formation, particularly,
when using negative charged phospholipids as POPG. AFM of SLBs
provides topographic information on the lipid bilayers enabling the
discrimination between La and Lb phases based on the height of the
different domains, with the additional advantage of working in a
liquid environment (Picas et al., 2012). Furthermore, when working
in force spectroscopy (FS) mode, AFM can throw light on the
nanomechanical properties of the relevant lipid planar systems,
mainly the breakthrough or yield threshold force (Fy) and the
adhesion force (Fadh) (Dufrne and Barger, 1997; Garcia-Manyes
et al., 2007; Oncins et al., 2007). In fact, force spectroscopy has been
tested as a suitable technique to distinguish between SLBs of several
pure and mixed phospholipid compositions through Fy values
(Garcia-Manyes et al., 2010; Redondo-Morata et al., 2012a). The
formation of SLBs of negatively charged systems like POPE:POPG
requires the presence of divalent cations to screen the negative
charge borne by mica (Richter et al., 2005; Reviakine and Brisson,
2000), in order to provide stability to the system (Redondo-Morata
et al., 2012b). The purpose of this research paper is two-fold: rst, to
gain more insight into the nanostructure and nanomechanics of
SLBs of POPE:POPG in presence of Ca2+; and secondly, to build and
compare binary phase diagrams for the POPE:POPG system
obtained either from liposomes through DSC data or from SLBs
through AFM observations.

2. Materials and methods

DSC of multilamellar liposomes and AFM of SLBs were
combined to study the thermotropic properties of the binary
system, POPE:POPG. Sample preparation and experimental
procedures have been published elsewhere (Surez-Germ
et al., 2011) and details are available in the Supporting
information, SI1.

209

Supplementry material related to this article found, in the

online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
Due to the preparation of the systems might be critical to its
properties it is important here to highlight the procedure used for
the AFM samples. Freshly cleaved mica discs (1 cm2) mounted on
round Teon discs were glued to steel discs. Liposome suspensions
were incubated on mica discs for 2 h over the transition
temperature of the mixture used. To prevent sample evaporation,
the steel disc containing the mica and the sample was enclosed in a
small petri dish inside a bigger petri dish with some water at the
bottom used as a reservoir. The big petri dish was then sealed with
Teon ribbon and placed inside an oven (Termaks AS, Bergen,
Norway) with temperature control of 0.2  C from the desired
temperature. After incubation, non-adsorbed liposomes were
removed by gently rinsing samples with buffer and letting them
stabilize overnight. Samples were then directly mounted on the
AFM scanner and stabilised for no less than 30 min. When working
above room temperature, anO-ring was then incorporated with the
aim of enclosing and sealing the sample, in order to avoid further
buffer evaporation during experiments. During the ramp, temperature experiments at minimum of 30 min were needed to stabilize
the sample before proceeding with the scan at each temperature
evaluated. AFM was equipped with an E scanner (10 mm) and
images were acquired in liquid and under intermittent contact
mode at 0 scan angle with a scan rate of 1.5 Hz. The vertical force
was maintained at the minimum value, maximizing the amplitude
set point value while keeping the vibration amplitude as low
as possible. All images were processed by a NanoScope Analysis
software (Bruker AXS Corporation, Santa Barbara, CA).
3. Results and discussion
3.1. Constructing diagrams from DSC data and AFM imaging
The phase behaviour of the POPE:POPG binary system was
investigated by DSC. To build a phase diagram for the POPE:POPG
system, the onset and completion temperatures (Tonset and Toffset,
respectively) from the endotherms of a series of mixtures have to
be determined. Fig. 1 shows the response function of multilamellar
vesicles in presence of 10 mM of Ca2+ of different POPE:POPG
mixtures as a function of temperature. As observed from
normalized endotherms, the Tm, the temperature at the maximum
heat capacity, remains withheld nearly 24  C up to the equimolar of
both phospholipids. Further increase in the molar ratio of POPG
results in a signicant decrease in the Tm. Remarkably, endotherms
are slightly asymmetric, skewed to the low temperature side in all
cases and becoming clearly asymmetric at 1 > xPOPG = 0.50. Indeed,
this behaviour indicates an enhanced non-ideal mixing behaviour
of the system as the POPG proportion increases. Mathematical
adjustments performed on individual endotherms (see Supplementary materials SI) allowed the determination of the thermodynamic parameters of the transition: Tm and the enthalpy
involved in the process of transformation ( D transf H ), from
the Lb to the La phase (Table 1).
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
Fig. 2A shows the binary phase diagram obtained from the
endotherms constructed after connecting the corresponding Tonset
(open circles) and Toffset (full circles). As observed, this pseudophase diagram isS-shaped, which indicates a substantial deviation
from the ideal mixing behaviour at all the molar fractions analysed.
Actually, the shift of Tm towards lower temperatures of those
liposome mixtures containing high POPG proportions (Harlos and
Eibl, 1980) and most specically, the dependence of POPG on the

210

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

Fig. 1. Normalized excess heat capacity proles for large multilamellar liposomes of
POPE:POPG mixtures at the indicated molar ratios. The total phospholipid
concentration was 2.0 mg mL1 and the heating scan rate was 0.44  C min1.

thermal history of the sample (Fleming and Keough, 1983) were

earlier reported. Both facts provided means for the discrepancies
found between the shape of the phase diagram found by PozoNavas et al. (Pozo Navas et al., 2005), (in absence of Ca2+) and,
additionally, to the differences in the nanomechanical magnitudes
found in precedent papers (Picas et al., 2012).
Actually, an interesting feature of the diagram here presented is
the quite horizontal segment of the solidus line from xPOPG > 0.7,
which is consistent with the occurrence of a miscibility gap within
the Lb phase. This behaviour means that, for the present system
and for xPOPG > 0.7, the formation of two gel lipidsub-domains with
different compositions (Lb1 and Lb2) should be considered (Garidel
et al., 1997, 2005). However, to probe the existence of these two gel
phasesX-ray diffraction or other techniques different of DSC should
be applied. In any event, such investigation is out of the scopes of
the present research. It has been shown that SLBs of POPE:POPG

Table 1
Thermodynamic parameters obtained from the thermograms and from tting the
AFM data to the van't Hoff equation.

xPOPG


0.25

0.50

0.75

Tm ( C)

DSC
AFM

23.50  0.15
27.80  0.12

23.70  0.13
24.35  0.15

19.30  0.10
25.85  0.03

DtransfH

DSC

17.7  0.2

18.9  0.2

24.7  0.2

AFM

1560  150

1600  300

980  30

(kJ mol1)

DHAFM/DHDSC

87

85

40

Fig. 2. (A) Phase diagram for POPE:POPG mixtures obtained from the heat capacity
curves shown in Fig. 1. Empty circles correspond to the experimental Tonset and lled
circles to the Toffset of the main phase transition. (B) Phase diagram of POPE:POPG
mixtures obtained from the AFM topography images. Empty squares correspond to
the experimental T where the rst La domain appeared and lled squares
correspond to the experimental T where the last Lb domain vanished. Error bars
correspond to standard deviations in at least three replicate experiments, if not
shown means they are inside the symbol.

prepared either by the vesicle fusion technique (Seeger et al.,

2009b) or LangmuirBlodgget double deposition (Picas et al.,
2010d) display expected La/Lb phase coexistence at room
temperature. In the present study, AFM observations of SLBs with
different mole fractions of POPG have been conducted at several
temperatures, to construct a phase diagram (Fig. 2B) and compare
it to the phase diagram obtained from DSC data (Fig. 2A). Of course
the comparison between both pseudo-phase diagrams should
be taken with caution. Firstly, because the inuence of the
substrate plays a major role in the lipid behaviour. Among other
effects the presence of the negatively charged POPG, may induce
differences in the composition of both leaets of the resulting SLBs.
Mica is the more popularized substrate because it is smooth and
hydrophilic, but it is worth to mention that other substrates, as
fused silica, borosilicate glass or titanium oxide are common. In
such cases the pseudo-phase diagram would present probably
different trends. Besides it is not only the inuence of the substrate
but the ionic strength and type of cation (Seantier and Kasemo,
2009; Garcia-Manyes et al., 2005a) or osmolarity (Hain et al., 2013)
that affect the properties of the phase transition. In any case, the
use of mica and a precise buffer is justied here by the need to
correlate the obtained results with other studies of our group
(Picas et al., 2009, 2010b). Secondly, it should be noted that,
although the temperature was controlled with great accuracy in
our AFM experiments (see Supporting information, SI1), the high

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

211

Fig. 3. AFM images of POPE:POPG phospholipid bilayers with xPOPG = 0.25 (A, D, G), xPOPG = 0.50 (B, E, H) and xPOPG = 0.75 (C, F, I) at 23, 27 and 29  C. White stars correspond to
mica surface, black arrows in image C correspond to La domains and white dotted lines correspond to height proles shown at the bottom of each composition. Z scale bar was
20 nm except for images (C, F, I) that was 10 nm.

sensitivity and accuracy attained with DSC cannot be technically

achieved with the actualtemperature-control system of the AFM,
which anticipates the difculty for nding a direct correspondence
between the phase diagrams obtained from both techniques.
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
Fig. 3 shows a selection of AFM topography images of SLBs
obtained at three different compositions (xPOPG = 0.25, 0.50 and
0.75) and three temperatures (23, 27 and 29  C). Although 37  C
would be desirable because it is considered physiological, at this
temperature only one phase is present. Indeed, E. coli has been
proved to present populations normally living out of warmblooded animals (Ishii and Sadowsky, 2008), and thus, studies at

room temperature might be also relevant and this provides means

for the present study on the phase separation. Assuming that many
factors could affect the sample, e.g. thermal history or sample
preparation (Fleming and Keough, 1983), and considering that
we were not far from the equilibrium allowing the sample to
stabilize after each increase of the temperature (see section 2), we
standardized the procedure resulting in reproducible AFM images
as well as the lipid domain boundaries after repetitive scans of the
same region. At 23  C, SLBs of xPOPG = 0.25 (Fig. 3A) and xPOPG = 0.50
(Fig. 3B) show at and homogeneous surfaces with step height
differences of 3.4 and 5.4 nm from the bare mica surface,
respectively. At this temperature, both images show SLBs in Lb
phase. For the SLB with xPOPG = 0.75 (Fig. 3C), the step height
difference was established at 5.63 nm. In this case, although the

212

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

surface of the SLBs is rather at and homogeneous, some small

patches covering less than 3% of the surface can be observed (see
black arrows in the image). These patches were 0.60 nm shorter
than the extended lipid structure, step height that matches with
the step height between La/Lb coexisting phases (Seeger et al.,
2009b). Indeed, these patches indicate that, for this temperature
and composition, the SLB is close to Tonset, so just starting the Lb to
La phase transition. At 23  C we conrmed the shift of the Tonset
towards higher temperatures than those observed from DSC data.
This behaviour, previously reported for SLBs (Garcia-Manyes et al.,
2005c; Keller et al., 2005), is the result of the potential interaction
of the proximal leaet of the SLBs with the mica substrate. In the
present study, such interaction is enhanced at higher POPG molar
fractions, as a consequence of a bridging effect of Ca2+ between
POPG and the negatively charged mica surface. Actually, the
interaction of the proximal leaet with mica substrate has been
extensively debated (Picas et al., 2010d; Garcia-Manyes et al.,
2005b; Attwood et al., 2013). This interaction would in fact
modulate not only the thermotropic behaviour of SLBs, but also
their topographical features and nanomechanical properties.
At 27  C, all SLBs (Fig. 3D, E and F) have two lipid domains that
show the La/Lb phase coexistence. The step height difference
respect to the bare mica for the lighter (taller) domains in these
images compares well with the value reported at 23  C (Lb phase).
In turn, the darker (shorter) domains are 0.6 to 1.4 nm lower than
those observed at 23  C, which is the expected height difference
between the Lb and La phases (Seeger et al., 2009b). After the
temperature was raised to 29  C (Fig. 3H and I), SLBs showed a
single domain with an average height that falls within the range
expected for La phases. Although residual Lb domains are still
present in Fig. 3G, the La phase is, however, the predominant
phase. Heights and roughness values obtained for each image
shown in Fig. 3 are available in a Table given in the Supporting
information, SII4. After a careful analysis of the topographical
images, the phase diagram was constructed from the Tonset (open
squares) at which the La phase appears and the Toffset (closed
squares), at which the Lb phase vanishes. The pseudo-binary phase
diagram is shown in Fig. 2B; Table 1 shows the values derived from
this data. As can be seen, the Tm determined from AFM imaging
were higher than those obtained from DSC. As stated above, this
behaviour was expected, since it is well known that SLB transition
encompasses a higher and wider range of temperatures than
transitions observed by performing DSC on liposomes with the
same composition. This behaviour, extensively discussed by
several authors (Oncins et al., 2007; Keller et al., 2005;
Tokumasu et al., 2003; Giocondi et al., 2001; Yang and
Appleyard, 2000), has been attributed to different effects: (i) the
presence of the substrate in contact with the proximal leaet of the
bilayer; (ii) the innite radius of curvature of the SLBs, which
decreases the lateral tension between phospholipids in the bilayer;
and (iii) the decoupling between the leaets of the bilayers which
results in a double transition. As concluded from detailed
investigations on POPEPOPG system reported by Facci's group
(Seeger et al., 2009b), the last effect appears to depend on the
experimental conditions followed during SLB preparation. In the
present study, SLBs were obtained by liposome spreading and
incubation of the sample above the Tm of the lipid mixture, under
high ionic strength conditions. This is why we did not observe the
intermediate phase postulated by Seeger et al. (Seeger et al.,
2009b). Indeed, in our experiments, only one transition
demonstrating the coupling between the two leaets of our
SLBs was observed. However, it is intriguing that in the SLB with
xPOPG = 0.5, the difference observed between the Tm obtained from
DSC and from AFM studies is less than 1  C. This observation
suggests that this composition is less affected by the supporting
surface. It has been widely reported that negatively charged

phospholipids adsorbed onto mica are limited to lateral diffusion

because of the interaction between the negative polar head and the
substrate (Reviakine and Brisson, 2000). The liquidus line (closed
squares in Fig. 2B) obtained on SLBs does not signicantly change
on the POPG molar fraction decreasing and is, remarkably, quite
similar to the Tm of pure POPE (xPOPG = 0).
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
Despite it being technically more difcult than DSC, the AFMbasedpseudo-phase diagram provides a fast, approximate evaluation of the composition of La and Lb phases on SLBs and thus, a way
to estimate the composition of the domains observed in the images
shown in Fig. 3. Therefore, by taking a connection line along 27  C
(the temperature selected for the nanomechanical studies) and
applying the lever rule, we can determine the composition of each
phase for the composition xPOPG = 0.25. Thus, we obtain that POPG
is distributed as follows: xPOPGa = 0.11 (molar fraction of POPG in
the La phase) and xPOPGb = 0.14 (molar fraction of POPG in Lb
phase). Consequently, POPE's presence in La and Lb phases
corresponds to xPOPEa = 0.03 and xPOPEb = 0.72, respectively. Thus,
the composition xPOPG = 0.25 shows an enrichment of POPG in La
and an enrichment of POPE in Lb phase. For the sake of clarity an
illustration showing the match between the real (calculated from
AFM ramp temperature images in SII1, SII2 and SII3) and the
expected (from the pseudo-binary phase diagram in Fig. 2B)
percentage of lipids in each phase is provided as Supporting
information, SII5.
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
3.2. Thermodynamics of the POPE:POPG phase transition onto mica
In a system where only two possible states (La and Lb) are
assumed, the phase transition of the SLBs onto a surface may be
described by an equilibrium constant (K) that can be dened as
K

1u

(1)

where u is the fraction of the La phase in the SLBs calculated by

measuring its area after the AFM images (Seeger et al., 2009b;
Keller et al., 2005; Tokumasu et al., 2003). According to Mabrey and
Sturtevant (Mabrey and Sturtevant, 1976), K can be described as a
function of T by the integrated form of the van't Hoff equation



DHvH 1 1

(2)
ln K 
R
T Tm
where R is the gas constant, Tm is the transition temperature and

DHvH is the van't Hoff molar enthalpy. Then, by inserting Eq. (2) in
(1) we obtain the following expression:

1
h

i
1 exp DHRvH 1T  T1m

(3)

which relates u at a given temperature with the DHvH involved in

the transition occurring on the surface. As a rst approach, we
applied the two-states approximation to the different systems
evaluated in this work. u values were determined at 10 different
temperatures and Eq. (3) was tted to the data. For a better
comparison, the results obtained from analysing the thermal
behaviour of the SLBs are summarized in Table 1 along with the
values obtained from DSC experiments. It is thought that, when the
temperature of the system increases and the phospholipids
undergo the phase transition, part of the energy is used to
counterbalance the interactions with the substrate. Therefore, this

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

would explain why the DHvH are higher for SLBs than DH for
liposomes (Richter et al., 2005; Yokoyama et al., 2013). For the SLBs
at xPOPG = 0.25 and 0.5, the magnitudes of the DHvH are quite
similar ( 1600 kJ mol1), but for the SLB at xPOPG = 0.75 it
decreases to 980 kJ mol1. Such a difference can be attributed to
the high negative charge present in the latter composition (see the
zeta potential values for each composition provided as Supporting
information, SII6), which increases the electrostatic repulsion
between the bilayer and the substrate and results in a decrease of
DHvH . The approximate number of lipids experiencing the
transition as a single unit, interpreted as the cooperativity unit
(Gennis, 1989), is given by N = DHAFM/DHDSC. The N values are
87 and 85 for the SLBs with xPOPG = 0.25 and 0.50, respectively, and
40 for the SLB at xPOPG = 0.75. This makes clear that, as the
proportion of negatively charged phospholipid increases in the
SLBs, the cooperativity unit of the transformation decreases. It is
most likely related with the fact that Ca2+ induces phase separation
because of its ability to bind stoichiometrically to negatively
charged phospholipids (e.g. POPG). Thus, it results in a reduction of
their surface charge and area with consequent increase in the
transition temperature and the promotion of a bilayer with a more
gel-like nature than in absence of the bivalent cation (Tokutomi
et al., 1981; Pedersen et al., 2006). This behaviour can explain the
quantitative differences found when comparing with other N
values obtained in the same system in absence of Ca2+ (Seeger
et al., 2009b). The decrease in N on the increase in the POPG molar
fraction might also be related, at least at 23  C, with the postulated
existence of a miscibility gap in the Lb phase when xPOPG > 0.7
(Pozo Navas et al., 2005; Garidel et al., 2005). It should be noted,
however, that the AFM experiments performed on SLBs did not
have enough resolution to provide visual evidence for the
existence of these two different Lb domains.
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
3.3. Nanomechanics of SLBs
Force spectroscopy has been extensively used to probe the
nanomechanical properties of lipid layers (Picas et al., 2010d;

213

Garcia-Manyes and Sanz, 2010; Alessandrini and Facci, 2012). Since

the pioneer works of Dufrne (Dufrne and Barger, 1997; Dufrne
and Boland, 1999), force spectroscopy has become the tool for
exploring the nanomechanical properties of pure phospholipid
SLBs (Garcia-Manyes et al., 2010; Drolle et al., 2012) and, more
importantly, assessing the nanostructural organization of more
complex lipid mixtures (Attwood et al., 2013; Redondo-Morata
et al., 2012c). Although, the biological meaning behind force
spectroscopy measurements is still controversial (Alessandrini
et al., 2011), the potential of this technique for revealing the
nanomechanics of complex systems, from models to natural
membranes (Francius et al., 2008) and living cells (Cross and Jin,
2007; Dufrene, 2008) is, however, widely accepted. The basic
magnitudes extracted from traditional force spectroscopy experiments (Fig. 4) are: (i) the breakthrough or yield threshold force (Fy),
i.e. the force that the bilayer can withstand before being indented;
and (ii) the adhesion force (Fadh), i.e. the pull-off force between the
tip and the bilayer (Dufrne and Barger, 1997; Picas et al., 2010d).
To clarify further, the nature of phase separation in complex
biomimetic systems, e.g. displaying phase separation in binary
lipid mixtures, we performed force spectroscopy measurements
on SLBs of POPE:POPG at different xPOPG. In terms of biological
relevance, we focused the analysis on the composition
xPOPG = 0.25. Force spectroscopy measurements were performed
at 27  C because we wanted to discriminate in force spectroscopy
terms between the Lb and La phases. Distributions of Fy for each
lipid domain at each studied composition are shown in Fig. 5. For
SLBs at xPOPG = 0.25, where the two phases can be spatially
resolved (Fig. 3D), the mean values of Fy obtained were quite
similar for Lb and La, 0.250  0.005 nN (Table 2). On the one hand,
this value compares well with the one we previously obtained for
the La phase in the same lipid mixture (Picas et al., 2009). On the
other hand, the value for the Lb phase is lower than that reported in
the same study.
To rationalize this discrepancy, three related factors should be
taken into account: (i) the higher acquisition temperature (27  C),
which is some degrees above the temperature used in our earlier
study (Picas et al., 2009); (ii) according to the phase diagram given,
whilst at room temperature we are close to the solidus line, at 27  C
we are in the coexistence region (Fig. 2B); and (iii) the possible

Fig. 4. Schematic representation of the experimental procedure used to obtain threshold and adhesion forces. (Right) Topographic images were rst acquired to visualize the
phospholipid domains and thereafter the AFM tip was centered in the domain chosen for analysis. (Left) Real force curve on a lipid domain. First, tip approaches (blue line) to
the surface (a), it touches the surface, it begins to press down the SLB (b) until the force is enough to punch the bilayer (breakthrough force) and the tip continues pressing the
mica surface (c). Afterwards, the tip begins to separate (red line) from the mica surface (d) until the tip is completely free from the sample (adhesion force) and the tip moves
away from the sample (a).

214

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

Fig. 5. Fy distribution for xPOPG = 0.25 (A, B), xPOPG = 0.50 (C, D) and xPOPG = 0.75 (E, F), at 27  C in the La and Lb lipid domains. Fits to the continuum nucleation model (Eq. (4)
are represented as solid lines.

formation of overlayed bilayers. As already considered in our

previous paper (Picas et al., 2009), the formation of double bilayers
(see Supplementary materials SII7) will result in different nanomechanical magnitudes (in a force curve tip must break through
two lipid bilayers before touching the mica surface) than those
presented here. This provides evidence for the differences found in
the present paper. Although far from the scope of this manuscript,
the statement on force spectroscopy magnitudes as ngerprint
for specic phospholipids (Garcia-Manyes et al., 2010) is, in
absence of standardised condition, doubtful. For the equimolar SLB
composition, xPOPG = 0.50 (Fig. 3E), the mean Fy values were
2.114  0.016 nN and 1.046  0.016 nN for the Lb and La domain,
respectively (Table 2). This means that we need 2 times more
force to indent the Lb than the La domain. For SLBs at xPOPG = 0.75
(Fig. 3F), the mean Fy value for the Lb phase was 0.531  0.013 nN,
whilst the La phase had almost twice this value (0.922  0.009 nN).

Table 2
Mean Fy and Fadh values from data presented in Figs. 5 and 6, respectively. Data in
Fig. 5 was tted to the continuum nucleation model and calculated G and S
parameters and data in Fig. 6 was tted to a Gaussian distribution. Errors values are
standard deviation from the mathematical statistics.

xPOPG
0.25

0.50

0.75

Fy (nN)

Lb
La

0.250  0.005
0.250  0.006

2.114  0.016
1.046  0.011

0.531  0.013
0.922  0.009

Fadh (nN)

Lb
La

0.450  0.006
0.175  0.006

1.119  0.010
0.278  0.002

0.54  0.02
1.115  0.019

G (nN)

Lb
La

21.8  1.6
19  3

38  2
29.9  1.1

31.0  1.7
28.0  1.0

S (mN m1)

Lb
La

9.53  0.06
6.3  0.2

6.61  0.11
9.87  0.03

17.8  0.3
7.00  0.09

Actually, the observation is in apparent contradiction with a

general assumption that higher Fy values are quite closely
associated with Lb phases. The results for xPOPG > 0.50 most
probably reect that, for SLBs containing high amounts of
negatively charged lipids, stiffness and electrostatic repulsion
variations should be taken into account. Notice that at this
composition, the nucleation model does not t the experimental
data as well as the other samples. This indicates a different
behaviour and physicochemical properties at high POPG proportions. Actually, the ratio of the two phases present can be found by
using the lever rule. When it is applied in the phase diagram shown
in Fig. 2B, the amount of POPG estimated for the La phase of SLBs at
xPOPG = 0.75 rises to 0.74 (Supplementary material, SII5), which
indeed corresponds to almost the entire amount of the POPG
present in the system.
Supplementry material related to this article found, in the
online
version,
at
http://dx.doi.org/10.1016/j.chemphyslip.
2014.07.009.
As predicted by the lever phase rule, the increase of POPG, in
absolute terms, should be more prominent in the La than in the Lb
phase. This results in smaller domains of the gel-crystalline phase
(Fig. 3DF) and in eventual changes in line tension values (GarcaSez et al., 2007). Thus, Fy values are strongly dependent not only
on negative repulsion between POPG molecules, but also on the
formation of solid clusters within the La phase in presence of Ca2+
(Houslay and Stanley, 1982). As a result, the force needed to indent
La domains highly enriched in POPG is greater, under our
experimental conditions, than the force required to overcome
the potential barrier to make a hole in a zwitterionic phospholipid
(Butt and Franz, 2002; Franz et al., 2002). Inspection of the
retraction part of force curves performed on SLBs with different
xPOPG allows obtaining the corresponding adhesion forces (see left
drawing in Fig. 4). In a seminal work by Dufrne et al. (Dufrne and
Barger, 1997), it was shown that Fadh are directly related to the
strength of the lateral forces between the molecules that provide

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

215

Fig. 6. Fadh distribution for xPOPG = 0.25 (A, B), xPOPG = 0.50 (C, D) and xPOPG = 0.75 (E, F), at 27  C in the La and Lb lipid domains. Fits to a Gaussian distribution are represented
as solid lines.

means for the cohesive forces within the lms. However, the actual
force required to separate the tip from the lipid may be largely
affected by the negative charge borne by POPG and the presence of
the Ca2+ in the environment. Distributions of Fadh for each domain
at different POPG mole fractions are shown in Fig. 6. Additionally,
mean Fadh values are summarized in Table 2. For xPOPG = 0.25, two
different Fadh values, 0.450  0.006 and 0.175  0.006 nN, were
obtained for the Lb and La domains, respectively. Actually, whilst
such behaviour was observed for complex mixtures (Sullan et al.,
2009), the converse situation, higher Fadh for uid than ordered
phases was reported in neutral phospholipids (Leonenko et al.,
2007). Besides, Fadh is strongly dependent on the nature and
geometry of the tip, surface roughness and preparation procedures
(Truno-Sfarghiu et al., 2008) among other conditions (Israelachvili, 2002). Therefore, Fadh may not be considered as an intrinsic
property of the SLBs. Then, by taking into account the composition
of the uid phase it is conceivable that La presents lower adhesive
forces than Lb, which can be related with the clustering of POPG
molecules in presence of Ca2+ (Houslay and Stanley, 1982).
The Fadh for the SLB with xPOPG = 0.50 were 1.119  0.020 and
0.278  0.002 nN for the Lb and La phases, respectively. Note,
however, that for this composition the mean Fadh is 4 times higher
for the Lb than for the La phase. Conversely, the Fadh were
signicantly lower for the Lb phase (0.54  0.02 nN) than for the La
one (1.115  0.019 nN) when xPOPG = 0.75 (Table 2), which is may be
due to a non Gaussian distribution (Fig. 6E). Strikingly, for this
composition the Fy values for the La phase result from a strong
repulsion with the tip (Picas et al., 2009) although the Fadh values
suggest a strong cohesion of the lm. Although the La phase is
enriched in POPG molecules and the repulsion between the
neighbour phospholipids and the AFM tip may occur, there are
enough POPE molecules to stabilize the interactions between
neighbouring lipids.

3.4. Applying the continuum nucleation model

The parameters that determine how the AFM tip can go through
the SLB in a forcedistance curve have been evaluated in both
theoretical and experimental terms (Butt and Franz, 2002; Loi
et al., 2002). A seminal work by Butt and co-workers (Butt and
Franz, 2002; Franz et al., 2002) introduced a theory to calculate the
activation energy required to form a hole in a lipid lm when an
AFM tip indents a planar layer. Although this model reduces the
system by considering the SLB as a liquid structure in a XY plane, it
has been successfully tested on several substrates and conditions
(Garca-Sez et al., 2007; Franz et al., 2002). Notably, this has also
been applied to SLBs, in order to unambiguously assign physical
parameters and provide the basis for discrimination between
phospholipid species (homo- and hetero-acids) with different
headgroups and, even, different compositions (Tokutomi et al.,
1981). As discussed by Butt and co-workers (Butt and Franz, 2002),
at constant velocity of the cantilever the continuum nucleation
model ts the probabilities of distribution of force P(F) according to



Z
V F
c
ln PF 
exp 
(4)
dF
F  F S
kC no F S
with
2 p2 G R
kB T
2

(5)

and
F s 2pRS

(6)

where V, as a rst approximation, can be interpreted as the

resonance frequency of the cantilever, kc is the spring constant of
the cantilever, n0 is the velocity, G is a line tension associated with
the unsaturated bonds of the molecules in the periphery of the

216

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

hole, R is the tip radius, kB is the Boltzmann constant, T is

the temperature and S is the spreading pressure associated with
the energy per unit area gained by the layer when relling the hole
formed after the force curve punch through. Eq. (4) can
be integrated analytically and the yield probability dP/dF can be
expressed as (Garca-Sez et al., 2007)
(
"



dP
V
c
V
c

exp 

F 0  F s
exp  0
dF kc vo
F  F s kc vo
F  Fs



c
cEi x 0
(7)
F  Fs

experiments is not easily relled, which means that the SLB loses
energy in some way during this process. However, no further
defects were observed when performing force spectroscopy
experiments, which might indicate that the kinetics of the process
is faster than the time needed for a typical approach-retract cycle.
The absolute largest and lowest estimated S values correspond to
the Lb phase in the SLBs with xPOPG= 0.75 and to the La phase with
xPOPG = 0.25 (Table 2). The whole set of S values emphasizes how,
the more domains become enriched in POPG, the more difcult is
for the lm to spread into the gap between the tip and the
substrate.

where
Z
Ei x

1
x

ey
dy
y

3.5. Biological relevance

(8)

The robustness of the model has been unambiguously

demonstrated for lipid bilayers of DOTAP (Franz et al., 2002)
and successfully extended to a wide variety of pure phospholipids
and mixtures with cholesterol by Garcia-Manyes et al. (GarciaManyes et al., 2010). To this end, the yield probability dP/dF was
adjusted to the experimental data and the tted parameters are
listed in Table 2. G values obtained for SLBs at xPOPG = 0.25 and
0.75 were quite similar for both phospholipid domains, ranging
from 20 to 30 nN, respectively. It should be noted that the Lb phase
showed higher values of G for both compositions. Since both POPE
and POPG are heteroacid phospholipids with the same acyl chain
composition, line tension should arise from the different interaction between the headgroup moieties, mainly entropic and
repulsion contributions (Garca-Sez et al., 2007), and the different
composition of lipids that integrate the periphery of the hole. To
rationalize these values one should assume that for SLBs at
xPOPG = 0.75, the phospholipids at the periphery of the hole would
probably be more enriched in POPG than in POPE. This assumption
is indirectly supported by previous Frster energy transfer
experiments performed with proteoliposomes with different
POPE:POPG compositions (Picas et al., 2010c). This POPG enrichment results in an electrostatic repulsion between neighbouring
molecules, leading to a relative increase in G values. This
observation corroborates the topographic features observed in
Fig. 3F (xPOPG = 0.75), where a decrease in the domain size most
probably reects the increase in the length of the boundary region
between the two phases. Then, for the SLB at xPOPG = 0.25, such
electrostatic repulsion decreases and thus, phospholipid molecules can be closer to each other, resulting in the decrease of G
values. Consequently, the size of the domains will be the largest
observed (Fig. 3D). When the SLBs are equimolar in composition,
xPOPG = 0.5, the values of G obtained for the high and low domains
were 38  2.0 nN and 29.9  1.1 nN, respectively, which indicates
that the Lb phase should be more likely to withstand higher
indentation forces. In addition, the La phase has a similar G value as
in the case of xPOPG = 0.75. This observation is consistent with the
expected enrichment of POPG within the uid phase, as predicted
by using the lever phase rule in the AFM phase diagram
(xPOPGa = 0.46, xPOPGb = 0.03, xPOPEa = 0.11, xPOPEb = 0.39, when
xPOPG = 0.5). Conversely, POPE should be the main component of
the Lb phase in equimolar composition.
As recently discussed (Garcia-Manyes et al., 2010), the
parameter S is directly related to the nature of the headgroup,
being negative for PE and PG. Thus, the parameter S seems to obey
subtle balances between the acyl chain and headgroup composition. Even if this assumption is theoretical and still not
experimentally determined, this parameter is most likely to be
related to the lateral surface pressure prole (Marsh, 2008; Ollila
et al., 2007; Bezrukov, 2000). Negative values of S indicate that the
hole formed after AFM indentation during force spectroscopy

Phase diagrams for mixtures of lipids mimicking biological

membranes are important in order to understand how phase
properties may affect the membrane physiology and function. It
could be interesting, for instance, to get a rationale for the
development of membrane disruptive antibiotics, directed
specically to the prokaryotic membrane but with non action
against the eukaryotic membrane. Whether the nanostructure of
biomembranes is becoming more complex by the compilation of
new evidences that transient and permanent lipid-protein
associations are present, there is no doubt that the biomembranes
are always in uid phase in order to guarantee their functional
properties; however, some biological mechanisms seem to
depend on the gel state. For instance it is well known that cells
modify their membrane composition under stress to adapt to the
environment (Denich et al., 2003; Sinensky, 1974). Hence, they
respond to changes in temperature, resulting in the increase of
low-transition temperature phospholipids. The reason for this is
that highly unsaturated phospholipids have enthalpy values
similar to those of saturated species. We are investigating
thoroughly the possible inuence of phospholipids on the activity
of LacY of E. coli, for which the composition mimicking the inner
membrane of the bacteria is provided by POPE and POPG at a
3:1 molar ratio. Importantly, for the adaptation of LacY to possible
changes occurring in the environment, the knowledge of the POPE:
POPG phase diagram becomes of relevance. Thus, by measuring
the resonance energy transfer between a single tryptophan
mutant of LacY (single-W151/C154G) and pyrene labelled phospholipids we have shown that the protein may recruit either POPE
or POPG, depending on the molar phospholipid ratio used to
reconstitute the protein (Picas et al., 2010c). Although FRET
experiments in solution (Picas et al., 2010a,c) were carried out
using this non-transporting mutant, the ndings correlate directly
with other observations on the modications experienced by
natural bacterial membranes under stress conditions. Besides,
by assuming the existence of an annular region of phospholipids
around LacY, we also demonstrated that there is a specic
selectivity of LacY for POPE. Similarly, there are studies of
mechanosensitive proteins like KcsA Schmidt et al., 2006 that
clearly demonstrate the requirement of specic headgroup for the
protein activity. Furthermore, a recent study by Weingarth et al.
(Weingarth et al., 2013) provided strong evidence on
non-annularlipidKcsA specic interactions.
Since many structural resolution and nanomanipulation
investigations of membrane proteins are obtained from AFM by
using reconstitution procedures into SLBs (Milhiet et al., 2006), it
becomes of crucial relevance to dispose of phase diagrams of the
lipid mixturesin-plane. This information would be the basis for
understanding the distribution of the transmembrane proteins
between the different lipid phases, lipid-protein association
among other properties observed in membrane models (SurezGerm et al., 2014) and cell membranes.

C. Surez-Germ et al. / Chemistry and Physics of Lipids 183 (2014) 208217

4. Conclusions
The behaviour of POPE:POPG binary system in the presence of
10 mM of Ca2+ has been investigated in this study through the
construction of two phase diagrams: one coming from DSC
analysis of liposomes and another constructed from SLBs imaging
bytemperature-controlled AFM. Specically, obtaining a phase
diagram for SLBs is of great relevance for understanding the phase
separation phenomena when working with this mixture, widely
used as the composition that mimics the inner membrane of E. coli.
The study was completed with the force spectroscopy
nanomechanical analysis of SLBs varying the xPOPG at 27  C. The
obtained results evidenced a strong inuence of the negatively
charged PG in the system, which seems to conrm that Ca2+
interacts directly with the PG headgroup promoting a clustering
effect. Hence, we showed that the presence of divalent ions in
negatively charged bilayers can largely modify the physicochemical behaviour of the system, and therefore, it becomes important to
take it into consideration regarding SLBs formation and also
possible implications of biological relevance as the interaction
with membrane proteins.
Conict of interest
The authors declare that there are no conicts of interest.
Transparency document
The Transparency document associated with this article can be
found in the online version.
Acknowledgements
C.S.G. is recipient of a FPI fellowship from the Ministerio de
Economa y Competitividad of Spain. This work has been
supported by grant CTQ-2008-03922/BQU from Ministerio de
Ciencia e Innovacin of Spain. Authors thank the Universitat
de Barcelona for nancial support. We thank Laura Picas for
valuable insights and comments.
References
Alessandrini, A., Facci, P., 2012. Micron 43, 1212.
Alessandrini, A., Seeger, H.M., Di Cerbo, A., Caramaschi, T., Facci, P., 2011. Soft Matter
7, 7054.
Attwood, S.J., Choi, Y., Leonenko, Z., 2013. Int J. Mol. Sci. 14, 3514.
Bagatolli, L.A., 2006. Biochim. Biophys. Acta 1758, 1556.
Bezrukov, S.M., 2000. Curr. Opin. Colloid Int. 5, 237.
Butt, H., Franz, V., 2002. Phys. Rev. E 66, 031601.
Cross, S.E., Jin, Y., Rao, J., Gimzewski, J.K., 2007. Nat. Nanotechnol. 2, 780.
Denich, T.J., Beaudette, L.A., Lee, H., Trevors, J.T., 2003. J. Microbiol. Meth. 52, 149.
Drolle, E., Gaikwad, R., Leonenko, Z., 2012. Biophys. J. 103, L27.
Dufrne, Y.F., Barger, W.R., Green, J.D., Lee, G.U., 1997. Langmuir 13, 4779.
Dufrne, Y., Boland, F.T., Schneider, J., Barger, W., Lee, G., 1999. Faraday Discuss 111,
79.
Dufrene, Y.F., 2008. Nat Rev. Micro. 6, 674.
Fleming, B.D., Keough, K.M.W., 1983. Can. J. Biochem. Cell Biol. 61, 882.
Francius, G., Domnech, ., Mingeot-Leclercq, M.P., Dufrene, Y., 2008. J. Bacteriol.
190, 7904.
Franz, V., Loi, S., Mller, H., Bamberg, E., Butt, H., 2002. Colloids Surf. B 23, 191.
Garca-Sez, A.J., Chiantia, S., Salgado, J., Schwille, P., 2007. Biophys. J. 93, 103.
Garcia-Manyes, S., Sanz, F., 2010. Biochim. Biophys. Acta 1798, 741.

217

Garcia-Manyes, S., Oncins, G., Sanz, F., 2005a. Biophys. J. 89, 1812.
Garcia-Manyes, S., Oncins, G., Sanz, F., 2005b. Biophys J. 89, 1812.
Garcia-Manyes, S., Oncins, G., Sanz, F., 2005c. Biophys. J. 89, 4261.
Garcia-Manyes, S., Domnech, ., Sanz, F., Montero, M.T., Hernndez-Borrell, J.,
2007. Biochim. Biophys. Acta 1768, 1190.
Garcia-Manyes, S., Redondo-Morata, L., Oncins, G., Sanz, F., 2010. J. Am. Chem. Soc.
132, 12874.
Garidel, P., Johann, C., Mennicke, L., Blume, A., 1997. Eur. Biophys. J. 26, 447.
Garidel, P., Johann, C., Blume, A., 2005. J. Therm. Anal. Calorim. 82, 447.
Gennis, R.B., 1989. Biomembranes: Structure and Function. Springer-Verlag, New
York.
Giocondi, M., Pacheco, L., Milhiet, P.E., Le Grimellec, C., 2001. Ultramicroscopy 86,
151.
Hain, N., Gallego, M., Reviakine, I., 2013. Langmuir 29, 2282.
Harlos, K., Eibl, H., 1980. Biochemistry (N.Y.) 19, 895.
Houslay, M.D., Stanley, K.K., 1982. Dynamics of Biological Membranes. Headington
Hill Hall, Chischester.
Ishii, S., Sadowsky, M.J., 2008. Microbes Environ. 23, 101.
Israelachvili, J., 2002. Electrostatic Forces Between Surfaces in Liquid. Academic
Press, London.
Keller, D., Larsen, N.B., Mller, I.M., Mouritsen, O.G., 2005. Phys Rev. Lett. 94, 025701.
Lee, A.G., 2004. Biochim. Biophys. Acta 1666, 62.
Leonenko, Z., Finot, E., Cramb, D., 2007. In: Dopico, A. (Ed.), Atomic Force
Microscopy: Interaction Forces Measured in Phospholipid Monolayers, Bilayers,
and Cell Membranes. Humana Press, pp. 601609.
Loi, S., Sun, G., Franz, V., Butt, H., 2002. Phys Rev. E 66, 031602.
Mabrey, S., Sturtevant, J.M., 1976. Proc. Natl. Acad. Sci. U.S.A 73, 3862.
Marsh, D., 2008. Biochim. Biophys. Acta 1778, 1545.
Milhiet, P., Gubellini, F., Berquand, A., Dosset, P., Rigaud, J., Le Grimellec, C., Lvy, D.,
2006. Biophys. J. 91, 3268.
Ollila, S., Hyvonen, M.T., Vattulainen, I., 2007. J. Phys. Chem. B 111, 3139.
Oncins, G., Picas, L., Hernndez-Borrell, J., Garcia-Manyes, S., Sanz, F., 2007. J.
Biophys. 93, 2713.
Pedersen, U.R., Leidy, C., Westh, P., Peters, G.H., 2006. Biochim. Biophys. Acta 1758,
573.
Picas, L., Montero, M.T., Morros, A., Cabaas, M.E., Seantier, B., Milhiet, P.E.,
Hernndez-Borrell, J., 2009. J. Phys. Chem. B 113, 4648.
Picas, L., Surez-Germ, C., Montero, M.T., Vzquez-Ibar, J.L., Hernndez-Borrell, J.,
Prieto, M., Loura, L.M.S., 2010a. Biochim. Biophys. Acta 1798, 1707.
Picas, L., Carretero-Genevrier, A., Montero, M.T., Vzquez-Ibar, J.L., Seantier, B.,
Milhiet, P.E., Hernndez-Borrell, J., 2010b. Biochim. Biophys. Acta 1798, 1014.
Picas, L., Montero, M.T., Morros, A., Vzquez-Ibar, J.L., Hernndez-Borrell, J., 2010c.
Biochim. Biophys. Acta 1798, 291.
Picas, L., Surez-Germ, C., Montero, M.T., Hernndez-Borrell, J., 2010d. J Phys.
Chem. B 114, 3543.
Picas, L., Milhiet, P., Hernndez-Borrell, J., 2012. Chem. Phys. Lipids 165, 845.
Pozo Navas, B., Lohner, K., Deutsch, G., Sevcsik, E., Riske, K.A., Dimova, R., Garidel, P.,
Pabst, G., 2005. Biochim. Biophys. Acta 1716, 40.
Redondo-Morata, L., Oncins, G., Sanz, F., 2012a. Biophys. J. 102, 66.
Redondo-Morata, L., Giannotti, M.I., Sanz, F., 2012b. In: Anonymous (Ed.), Atomic
Force Microscopy in Liquid. Wiley-VCH Verlag GmbH & Co. KGaA, pp. 259284.
Redondo-Morata, L., Giannotti, M.I., Sanz, F., 2012c. Langmuir 28, 12851.
Reviakine, I., Brisson, A., 2000. Langmuir 16, 1806.
Richter, R.P., Maury, N., Brisson, A.R., 2005. Langmuir 21, 299.
Schmidt, D., Jiang, Q., MacKinnon, R., 2006. Nature 444, 775.
Seantier, B., Kasemo, B., 2009. Langmuir 25, 5767.
Seeger, H.M., Bortolotti, C.A., Alessandrini, A., Facci, P., 2009a. J. Phys. Chem. B 113,
16654.
Seeger, H.M., Marino, G., Alessandrini, A., Facci, P., 2009b. Biophys. J. 97, 1067.
Sinensky, M., 1974. Proc. Natl. Acad. Sci. U.S.A. 71, 522.
Surez-Germ, C., Montero, M.T., Igns-Mullol, J., Hernndez-Borrell, J., Domnech,
., 2011. J. Phys. Chem. B 115, 12778.
Surez-Germ, C., Domnech, ., Montero, M.T., Hernndez-Borrell, J., 2014.
Biochim. Biophys. Acta 1838, 842.
Sullan, R.M.A., Li, J.K., Zou, S., 2009. Langmuir 25, 7471.
Tokumasu, F., Jin, A.J., Feigenson, G.W., Dvorak, J.A., 2003. Ultramicroscopy 97, 217.
Tokutomi, S., Lew, R., Ohnishi, S., 1981. Biochim. Biophys. Acta 643, 276.
Truno-Sfarghiu, Y., Meurisse, M., Rieu, J., 2008. Langmuir 24, 8765.
Vitrac, H., Bogdanov, M., Dowhan, W., 2013. Proc. Natl. Acad. Sci. U.S.A. 110, 9338.
Weingarth, M., Prokofyev, A., van, d.C., Nand, D., Bonvin, A.M.J.J., Pongs, O., Baldus,
M., 2013. J. Am. Chem. Soc. 135, 3983.
Yang, J., Appleyard, J., 2000. J. Phys. Chem. B 104, 8097.
Yokoyama, H., Ikeda, K., Wakabayashi, M., Ishihama, Y., Nakano, M., 2013. Langmuir
29, 857.