Anda di halaman 1dari 10

Article

pubs.acs.org/Langmuir

Amphipathic Membrane-Active Peptides Recognize and Stabilize


Ruptured Membrane Pores: Exploring Cause and Effect with Coarse-
Grained Simulations
Delin Sun,† Jan Forsman,‡ and Clifford E. Woodward*,†

School of Physical, Environmental and Mathematical Sciences, University of New South Wales, Canberra ACT 2600, Australia

Theoretical Chemistry, Chemical Centre, Lund University, P.O. Box 124, S-221 00 Lund, Sweden
*
S Supporting Information
Downloaded via INDIAN INST OF TECHNOLOGY BOMBAY on February 19, 2019 at 05:31:52 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Induction of membrane pores has been suggested as


the common molecular action by which a variety of amphipathic
membrane-active peptides cause damage to cells. In this study, we
have performed coarse-grained molecular dynamics simulations to
establish two clear molecular processes that seem critical for the
activity of amphipathic peptides. They are (i) the recognition and (ii)
the stabilization of ruptured membrane pores. By considering 12
structurally different peptide types, we reveal that peptide secondary
structure content, hydrophobicity, and length are important
physicochemical factors that allow amphipathic peptides to aggregate
in and stabilize ruptured membrane pores. The simulated inner
diameters of peptide-stabilized membrane pores are in good
agreement with available experimental data. However, the orienta-
tions of α-helical peptides in the membrane pore were found to be quite dispersed. This supports recent challenges to the
traditional depictions to peptide orientations in the classical toroidal and barrel-stave pore models.

1. INTRODUCTION lipid vesicles loaded with fluorescent dye in order to establish


The damage incurred by cell membranes through induction of the dynamic character of membrane pores induced by
nanoscale pores has been suggested as a common molecular magainin-2 (an antimicrobial peptide) and Baxα5 (a
action for a wide variety of amphipathic membrane-active mitochondria-apoptotic peptide). It was found that the
peptides, including cell-penetrating peptides,1,2 antimicrobial ruptured membrane pores were unstable, with initial diameters
peptides,3,4 amyloidal peptides,5,6 and Baxα5 mitochondria- of some tens of nanometers. However, within minutes the pore
apoptotic peptide.7,8 However, the mechanisms by which diameters decreased to a smaller equilibrium value. This
amphipathic peptides induce membrane pores are still being phenomenon is consistent with the findings of Brochard-Wyart
debated. Experimental studies using model lipid membranes and co-workers, who found that application of mechanical
have provided valuable evidence that membrane-active peptides tension could rupture a giant lipid vesicle to form a large
generate membrane pores via a cooperative kinetic proc- transient pore of micrometer size, which closes within seconds,
ess.1,2,4−7,9−13 Huang and co-workers, using X-ray diffraction once the stress is removed.14 These experimental studies
and oriented circular dichroism (OCD) techniques, established suggest that amphipathic peptides induce and then stabilize
that a minimum concentration of amphipathic peptides was membrane pores, thus allowing persistent leakage of internal-
required for membrane pore formation, suggesting a cooper- ized contents, such as fluorescent dyes.15 Several amphipathic
ative mechanism for these peptides.4,12 According to the “two- peptides, like magainin-2 and Baxα5, were found to exhibit so-
state” model proposed by Huang et al., amphipathic peptides called “all-or-none” release kinetics,10,11 whereby vesicles either
first accumulate on the outer leaflet of the membrane with an remain intact or else completely release their contents. Other
orientation parallel to the membrane surface (the S-state). The pore-forming peptides which show this type of release kinetics
asymmetric distribution of peptides on the lipid membrane are the GALA peptide,16 cecropin A,13 human cathelicidin (LL-
causes surface area expansion coupled with an internal 37),17 human defensins,18 pardaxin,19 α-hemolysin,20 human
membrane tension. Above a threshold concentration of islet amyloid polypeptide (IAPP),6 α-synuclein,21 and cyto-
peptides, the lipid membrane ruptures to form a pore. chrome c.22 The ability to significantly stabilize and maintain
Concomitantly, a fraction of membrane-bound peptides alter
their orientations to become perpendicular (the I-state) as they Received: September 26, 2014
presumably insert themselves into the membrane pore. Tamba Revised: November 27, 2014
et al.9 and Fuertes et al.11 performed kinetic experiments with Published: December 9, 2014

© 2014 American Chemical Society 752 DOI: 10.1021/la5038266


Langmuir 2015, 31, 752−761
Langmuir Article

membrane pores would appear to be a general requirement for model predicts potentials of mean force for amino acid
the activity of these pore-forming peptides. Notwithstanding translocation through zwitterionic bilayers,32 which are in
the significant progress achieved by experimental studies, the reasonable agreement with full atomistic simulations results33 as
process by which these types of amphipathic peptides carry out well as the Wimley−White interfacial scale.34
this action remains unclear. In experiments, large membrane pores are purported to open
Molecular simulations have increasingly been used to study due to the surface tension caused by the asymmetric adsorption
interactions of membrane-active peptides with lipid bilayers. of peptides. Such pores are not likely to be observed in the
Indeed, some of that work has reported that high relatively short time span of our simulations (microseconds).
concentrations of antimicrobial peptides are able to generate To circumvent this problem, we created an initial pore in the
membrane pores in lipid bilayers within a few microseconds or membrane using artificial means. In lipid vesicle experiments,
only a few hundred nanoseconds.23−26 However, these membrane rupturing is usually obtained through application of
simulated pores have generally displayed very small internal a mechanical stress.9 Simulations by Tieleman et al. showed
diameters (∼1−2 nm) which do not appear to correspond to that an electric field and mechanical stress are both able to
the large equilibrium membrane pores (diameter ∼ 3−5 nm) rupture a lipid bilayer.35 In our study, a pore was produced by
reported in experiments.9,12 Furthermore, kinetic experiments an electric field of magnitude of 0.7 V/nm, applied across a
with fluorescent dye-loaded lipid vesicles have consistently lipid membrane (electroporation). For the latter we used a
reported that amphipathic peptides trigger leakage of model of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine
fluorescent dyes on the time scale of seconds or even (POPC), containing 1152 lipids and solvated by 46 000 water
minutes,9,10 which is several orders of magnitude slower than beads. The electric field induced a large pore, approximately 15
the rate of pore formation observed in the aforementioned nm in diameter in the bilayer. To equilibrate the porated
simulation studies.23−26 This then raises doubt as to whether bilayer, 200 ns of MD simulations was run in an isotropic
the simulated mechanisms accurately represent the actual pressure (1 bar) ensemble. This ensemble constrains volume
behavior of these peptides. On the other hand, rapidly formed fluctuations to be uniform in all dimensions and is not generally
small pores may play the role of nucleation sites for the sudden used for bilayer simulations. Instead, it is usual to employ a
rupture of the lipid membrane over a longer time scale.27 semi-isotropic pressure ensemble, which allows independent
Simulations by Leontiadou et al. showed that the presence of a volume fluctuations with a uniform pressure tensor (normal
small pore in a lipid bilayer could dramatically reduce the and parallel to the membrane). However, we found that if the
threshold surface tension required to rupture the membrane.28 semi-isotropic ensemble was used during the pore equilibration
Furthermore, experiments by Lee et al. support the role of simulations, rapid closure of the pore would ensue within 30 ns.
precursor membrane defects for inducing stable membrane There is ample evidence that the MARTINI model tends to
pores.4 Those authors also reported that peptide translocation accelerate pore closure dynamics. For example, simulations by
could occur (presumably via transient pores) before a large and Bennett et al., using the united-atom Berger lipid model, found
stable membrane pore was created.12 that a small pore induced by lipid flip-flop in a lipid bilayer
The mechanisms by which amphipathic membrane-active could survive for hundreds of nanoseconds.36 Nevertheless, the
peptides induce membrane rupture, and the potential role relatively rapid dynamics of the MARTINI model does allow us
played by small transient pores, remain to be established. In the to investigate phenomena, which may not be easily accessible to
simulation work reported here we do not attempt to answer more detailed atomistic simulations. During the 200 ns
this question. Instead, we will focus on the later stages of equilibration simulations, the lipids rapidly adopted a
membrane rupture leading up to the formation of stable pores, configuration consistent with a toroidal pore. We subsequently
as observed in many experiments, e.g., Tamba et al. and Fuertes added membrane-active peptides to the equilibrated system in
et al.9,11 That is, we assume a mechanism whereby amphipathic order to investigate their response to the presence of the pore.
peptides rupture the membrane, presumably through surface 2.2. Adsorption of Peptide to a Stabilized Pore. We
tension (though we do not discount other mechanisms) to studied 12 structurally different peptides. They were the
form a large unstable pore, which is then rapidly stabilized. We antimicrobial peptides magainin-2 (PDB: 2MAG), melittin
focus on the role that amphipathic peptides play in the (PDB: 2MLT), LL-37 (PDB: 2K6O), mastoparan (PDB:
transition of the large ruptured membrane pore to the smaller, 1D7N), human-β-defensin-1 (PDB: 1IJV), protegrin-1 (PDB:
stable equilibrium pore. Our aim is to elucidate the key 1PG1), lactoferricin (PDB 1LFC), indolicidin (PDB: 1G89),
molecular determinants that significantly affect the peptide’s tritrpticin (PDB: 1D6X); the amyloid human IAPP (PDB:
efficacy in this pore transition process. 2L86), the apoptotic Baxα5 peptide (sequence of 108−126 of
the apoptosis-regulated BAX protein, PDB: 1F16), and the cell-
2. SIMULATION METHODS penetrating peptide transportan (PDB: 1SMZ). The terms in
2.1. Model and Pore Creation. We performed molecular parentheses represent the Protein Data Bank (PDB)
dynamics (MD) simulations using the nonpolarizable coarse- nomenclature. The atomistic structures of these peptides
grained MARTINI force field developed by Marrink et al.29,30 were downloaded from the Protein Data Bank. The secondary
Our purpose was to investigate amphipathic peptides in the structures were determined using the DSSP program (http://
presence of a ruptured membrane pore. While there is evidence binf.gmu.edu/software/DSSPCMBI.html), and mapping to the
that the MARTINI force field is inadequate for the simulation coarse-grained topology was finally achieved with the martinize
of membrane defect (small pore) formation per se,31 we use it code (http://md.chem.rug.nl/cgmartini/). Here, we should
here to model the response of amphipathic peptides to a note that in the MARTINI model the predetermined peptide
membrane pore which has already been formed. Thus, our secondary structures are artificially maintained by geometric
study relies on the MARTINI model’s ability to adequately constraints throughout the simulations. Hence, the MARTINI
describe peptide interactions with the lipid bilayer. In this model precludes the study of phenomena involving secondary
regard, it is worth noting that the nonpolarizable MARTINI structure changes of peptides upon adsorption onto lipid
753 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

bilayers. This apparent “defect” of the MARTINI model helical secondary structures and the other where these
somewhat limits our study, as the adsorption of membrane- structural constraints were removed.
active peptides to lipid bilayers is usually accompanied by All simulations were run in the GROMACS 4.5.5 package.38
significant structural changes in the peptides. For example, The temperature was maintained at 310 K using the Berendsen
many α-helical antimicrobial peptides adopt a coiled structure thermostat.39 The system’s pressure was controlled with the
in solution and only fold into an α-helical form upon Berendsen barostat39 at 1 bar with a compressibility of 3.0 ×
adsorption onto cell membranes.37 The mechanism by which 10−5 bar−1. Nonbonded Lennard-Jones (LJ) and electrostatic
the lipid membrane induces these structural changes is beyond interactions were both truncated at a distance rcut = 1.2 nm. To
the scope of this simulation study. Instead, we focus on the role avoid generation of unwanted noise, the standard shift function
played by adsorbed peptides that have already adopted their of GROMACS was used in which both the energy and force
lipid-associated structures. However, in this work, we do probe vanished at the cutoff distance. The LJ potential was shifted
the importance of peptide secondary structure on the peptide’s from 0.9 to 1.2 nm, and the electrostatic potential was shifted
ability to stabilize membrane pore. from 0 to 1.2 nm. All systems were simulated with an
The peptides were gradually introduced with random integration time step of 30 fs. We should note that for the
configurations into the simulation box after the pore was coarse-grained MARTINI model the simulation time is an
generated and equilibrated. In order to obtain a peptide to lipid effective one, which corresponds to being about 4 times longer
ratio (P:L) of 1:144, we initially placed eight peptides (and than “real” time.
neutralizing counterions) at random locations in the water 2.4. Calculating the Inner Diameter of Membrane
phase above one leaflet of the porated POPC bilayer. We then Pore. Two assumptions were made in order to calculate the
equilibrated the system for 500 ns. During this equilibration inner diameters of membrane pores: (i) the inner geometry of
stage, we again used the isotropic pressure coupling method, the pore was viewed as a cylinder, as shown in Scheme 1A; (ii)
which stabilized the membrane pore, while the peptide
distribution reached equilibrium. If a higher P:L ratio was Scheme 1. (A) Inner Geometry of Membrane Pore Can Be
required, eight more peptides and counterions were inserted as Viewed as a Cylinder; (B) Mass Density Distribution Profile
before and the system equilibrated for a further 500 ns. This for D3B and C3A Beads of MARTINI POPC Lipid Bilayera
process was repeated until the desired peptide concentration
was reached. The peptide concentrations we investigated are
given by P:L = 1:144, 1:72, 1:48, and 1:36. With the number of
peptides investigated, this amounted to 48 different systems in
total.
The process above describes equilibration of peptide
configurations in the presence of a pore, which was artificially
stabilized by the chosen ensemble. Hence, we denote it as the
constrained pore adsorption simulation stage (CPAS). The
modeling employed allows us to investigate the response of the
peptides to the generated pore over a time scale much longer
than it takes for the isolated pore to collapse. In experimental a
The distance between the two peaks is chosen to be the height of the
scenarios, if peptides were already strongly adsorbed on the cylinder.
bilayer surface when the membrane was ruptured, they would
already be proximal to the edge of the ruptured pore where
they could quickly act to reduce the line tension. They could
also further stabilize the ruptured pore by the generation of the density of water in the pore is the same as that of bulk water
lateral surface tension in the bilayer. It is indicative that the (1 g/cm3). Similar assumptions were also used in the
lifetime of ruptured pores, in the presence of membrane-active simulation study of Leontiadou et al.28 The central region of
peptides, was reported to be of the order of seconds,9,11 which the pore was inscribed with a circular cylinder of height H. The
is much longer than the 30 ns lifetimes seen in our inner diameter, d (= 2r), of the pore was then calculated (in
unconstrained coarse-grained simulations of isolated pores. nanometers) using the relation
2.3. Relaxation of the Pore in the Presence of NA 4N 72N
Adsorbed Peptide. The final configurations from the CPAS 3
= × 1021, d = 2r = 2
simulations, described above, initiated our next stage of 18 cm πr 2H 602πH
simulations, which aimed at determining the stability of the where H is chosen as the distance between two peaks in the
membrane pore in the presence of the adsorbed peptides. density distribution profile for the D3B and C3A beads of the
During these unconstrained pore stage (UCPS) simulations, we POPC lipid bilayer (see Scheme 1B). NA is Avogadro’s
performed a further 1.5 μs of MD simulations, while the constant, and N is the average number of water beads in the
pressure was allowed to fluctuate according to the semi- cylinder (note: the factor 4 appearing as one MARTINI water
isotropic coupling method. Here, volume fluctuations parallel bead is equivalent to four real waters).
and perpendicular to the bilayer plane were independent, with 2.5. Calculating α-Helical Peptide Tilt Angles. The tilt
the same value of 1 bar for the components of the average angles for α-helical magainin-2 (9 ALA−16 PHE), Baxα5 (6
pressure tensor. This maintains a zero surface tension, averaged LEU−17 ALA), melittin (15 ALA−22 ARG), and LL-37 (10
across the bilayer and (without peptide present) would lead to LYS−21 VAL) were calculated in our simulations as follows.
rapid pore closure within 30 ns. For magainin-2, Baxα5, and The tilt angle is defined as that between the helix axis, as
LL-37, simulations were performed using two different determined by two backbone beads (listed in the parentheses
scenarios: one where the peptides were constrained to α- above) and the bilayer normal (see Scheme 2). In all cases, the
754 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

average tilt angle distributions were obtained over the last 500 amphipathic peptides. While we have initiated membrane
ns of the UCPS simulations. rupture via electroporation, in experiments it appears to be the
excess peptide adsorption to the bilayer that induces pore
Scheme 2. Tilt Angle of α-Helical Peptides in Membrane formation. However, once the pore has formed, our results
Pore show that peptides would rapidly adsorb to it via diffusion. That
is, the pore provides a region of significantly lower chemical
potential for these peptides. We note that, unlike other
simulation studies,42−44 the peptides have not been purposely
inserted into the pores. Thus, we are able to demonstrate that
these peptides preferentially adsorb to the pore compared to
other environments, e.g., an intact membrane surface.
3.2. Nonelectrostatic Interactions Are Important for
Pore Recognition. We also performed modified CPAS
simulations in order to elucidate the mechanisms leading to
the peptide aggregation at membrane pores. The coarse-grained
3. RESULTS AND DISCUSSION POPC lipid model was divided into three structural
3.1. CPAS Simulations Show That Amphipathic components: lipid head (two beads), glycerol region (two
Peptides Recognize the Ruptured Membrane Pore. All beads), and alkyl chain (nine beads). Using melittin as a
the membrane-active peptides studied here displayed the same representative peptide, we began by converting all arginine and
behavior during the course of the CPAS simulations: from the lysine residues to their neutral forms so that the peptides did
bulk water above the porated membrane they selectively not interact electrostatically with the lipid heads. With this
adsorbed to the edges of the ruptured membrane. Figure 1 model of “neutral” melittin, we repeated the CPAS simulations.
We found that, even with electrostatic interactions absent, the
peptides still adsorbed strongly to the membrane pore (Figure
2A). This is somewhat surprising, as it is often suggested that

Figure 2. Side views of the adsorption of melittin peptides to the


artificially stabilized ruptured membrane pore. (A) Electrostatic
interactions between peptides and lipids switched off. (B) Lennard-
Jones interactions between peptides and lipid tails are switched off.
Only lipid heads (blue) and peptides (orange) are shown for clarity.

electrostatic attractions between cationic amphipathic peptides


and lipids play a dominant role in peptide activity.45 For
example, simulation work by Sengupta et al. suggests that
electrostatic interactions are an essential component for the
Figure 1. Top-viewed snapshots illustrating the selective adsorption of formation of small transient pores in membranes26 due to the
amphipathic peptides to the artificially stabilized ruptured membrane creation of electrostatic potentials across membranes. While we
pore. Peptides are colored orange; lipids are colored blue.
cannot rule out the importance of electrostatics for the
formation of pores in membranes, it appears not to play an
shows snapshots of the final configurations of the CPAS essential role in the subsequent adsorption of peptides to pores.
simulations for the different peptides at a concentration of P:L This conclusion may of course be different for the case of
= 1:36. In all cases, the peptides preferred to accumulate at the negatively charged lipids. In this context, we note that melittin
edge and within the constrained membrane pores. At these (+5) has a charge at the higher end of the spectrum of peptides
regions, the amphipathic peptides tend to insert their we studied.
hydrophobic residues into the lipid alkyl chain regions while In a second set of simulations, the electrostatic interactions
their hydrophilic residues are proximal to the lipid head groups. were maintained while attractive hydrophobic interactions
Simulations by Mihajlovic and Lazaridis, using an implicit (modeled by LJ potentials) between the melittin peptides
membrane model, reported similar behavior. They found that and the glycerol and alkyl chain regions were switched off.
models for alamethicin, melittin, magainin-H2, and piscidin-1 Following CPAS simulations with this modified “hydrophilic”
antimicrobial peptides all adsorbed strongly to a membrane melittin model, we found significantly less peptide bound to the
pore.40 Experiments by Rakowska et al. also showed the membrane pore (Figure 2B). Thus, hydrophobic interactions
migration of antimicrobial peptide on a membrane surface to between the peptide and the inner part of the lipids appear to
the edge of pores leading to their expansion (E-state).41 Our play a crucial role in peptide adsorption to pores. The
simulation results suggest that the ability to recognize membrane pore is a region of high negative Gaussian curvature
membrane pores is a common trait of membrane-active and (due to bilayer distortion) is rich in packing defects, which
755 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

expose hydrophobic lipid tails to the water. It is in such regions


of the lipid bilayer that opportunistic binding with the
hydrophobic regions of amphipathic peptides can occur.
Given that the structural features of melittin are common to
other amphipathic peptides, we expect this behavior to be
representative of all the peptides modeled in this study. Indeed,
virtually all amphipathic membrane-active peptides have a high
content of hydrophobic amino acids, allowing them to interact
favorably with the inner part of the bilayer leaflet. The ability to
sense membrane defects is a trait that should be shared by all
proteins, which contain amphipathic α-helices. Specific
examples such as α-synuclein,46 Golgi-associated Arf GTPase-
activating protein 1,47 and Bin/amphiphysin/Rvs (BAR)
domains48 have been shown in experiments to preferentially
bind to highly curved regions of lipid membranes.
We should note that similar findings to ours have been
reported in simulations by Vácha and Frenkel,49 who
investigated a rigid-rod model for peptides and a simplified
model for lipids. Those authors found that the rigid peptides
could form a “double-belt” structure, anchored to the pore Figure 3. Top-viewed snapshots illustrating final configurations at the
surface by hydrophobic interactions while associating with each end of the 1.5 μs unconstrained pore stage (UCPS) simulations.
other along the helical axis. The MARTINI force field used Magainin-2, Baxα5, LL-37, defensin-1, protegrin-1, human IAPP, and
here provides an intermediate level of complexity that falls transportan are able to stabilize the membrane pores whereas
mastoparan, lactoferricin, indolicidin, and tritrpticin cannot stabilize
between such very coarse-grained models and all-atom models. the pores.
3.3. UCPS Simulations Show That Amphipathic
Peptides Can Stabilize Membrane Pores. Experiments by
Peuch et al. have shown that adsorption of amphipathic simulations (an effective time of approximately 6 μs) suggests a
detergents can stabilize membrane pores by reducing their line kinetic barrier to pore closure of several kBT. It is of interest to
tension.50 It is plausible that amphipathic peptides may act in a investigate the molecular factors that determine the amount of
similar fashion. The subsequent stabilization of pores by stability that adsorbed peptides can impart to a pore.
adsorbed peptides should be reflected in the pore lifetimes. We A number of experimental studies have reported that peptide
tested this assertion with the 48 systems described above by length, hydrophobicity, charge, and secondary structure are
subsequently running unconstrained pore stage (UCPS) important molecular factors governing the efficacy of
simulations for a further 1.5 μs on the 48 systems described amphipathic peptides.51,52 In Table 1, we include key structural
above. We recall that in the UCPS simulations the pressure parameters of the 12 peptides in our study. These are the
coupling was switched to the semi-isotropic scheme, wherein number of residues (peptide length), secondary structure
volume fluctuations in the directions parallel and perpendicular content, and net charge. From this table, it can be immediately
to the bilayer are uncoupled. Hence, the isotropic volume inferred that peptide charge seems to have only a small effect
fluctuations that artificially stabilized the pores in the CPAS on pore stabilization in the POPC bilayer. Magainin-2, Baxα5,
simulations were removed. With UCPS simulations, a mastoparan, and indolicidin all carry a net charge of +3.
preformed pore (in the absence of peptides) would rapidly However, magainin-2 and Baxα5 gave open pores at the end of
close within about 30 ns (for the MARTINI model). To the UCPS simulations, whereas for mastoparan and indolicidin
investigate the effect of adsorbed peptide, we initiated the the pores closed. To further investigate if peptide charge is
UCPS simulations with the final configurations of the CPAS necessary to maintain an open pore, we repeated a UCPS
simulations described above. During the UCPS simulations, we simulation using the neutral form of magainin-2 peptide. As
found that with adsorbed peptides the pore diameters did shown in Figure 4A, the neutral magainin-2 peptide is still able
decrease, though generally over a time scale much larger than to maintain an open membrane pore over the length of the
30 ns. The time dependence of the pore diameters is shown in UCPS simulation. We also found little correlation between a
Figure S1 (Supporting Information). In the presence of some larger peptide charge and the size of the stable pores. For
peptide species, the pore remained open over the full length of example, LL-37 and protegrin-1 both have a charge of +6, but
the UCPS simulations and appeared to reach an equilibrium LL-37 forms much larger pores than protegrin-1. Furthermore,
value (Figure S1). Figure 3 shows snapshots of the final lactoferricin with a charge of +8 gave rise to closed pores at the
configurations of the UCPS simulations (with P:L = 1:36). The end of the UCPS simulation.
peptides magainin-2, Baxα5, LL-37, defensin-1, protegrin-1, The secondary structure content of peptides seems to have a
human IAPP, and transportan accumulated strongly in the marked effect on the pore-stabilization property of amphipathic
pores, which remained open for the duration of the UCPS peptides. Indolicidin, tritrpticin, and lactoferricin all have very
simulations. For the peptides mastoparan, lactoferricin, low or no secondary structure content, and we found that they
indolicidin, and tritrpticin the pore eventually closed, and the did not maintain open pores over the length of the UCPS
peptides were dispersed on the bilayer. If an isolated pore is simulations, even at the highest concentration investigated.
unstable or at least weakly metastable, the presence of adsorbed Experimental evidence indirectly supports this result in that,
peptides appears to enhance the pore metastability as is rather than large permanent pores, these three peptides have
reflected in the increased pore lifetimes. The fact that the some been found to generate only transient ion-channel-like
pores remained stable over the at least full length of the UCPS membrane defects, which nevertheless are able to depolarize
756 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

Table 1. Structural Parameters of 12 Amphipathic Peptides and the Calculated Inner Diameters of Peptides Stabilized
Membrane Pores at Varying P:L Ratios (1:144, 1:72, 1:48, and 1:36)
diameters of membrane pores (nm)
peptides number of residues secondary structure contentsa charges 1:144 1:72 1:48 1:36
magainin-2 23 0.87 +3 0 0.8 1.3 3.6
Baxα5 19 0.89 +3 0 0.7 1.6 3.9
melittin 26 0.92 +5 0 0.7 2.4 2.9
LL-37 37 0.82 +6 0 3.8 6.3 6.6
mastoparan 14 0.71 +3 0 0 0 0
defensin-1 36 0.72 +4 0 1.0 2.2 3.5
protegrin-1 18 0.78 +6 0 0 0 2.6
lactoferricin 25 0.40 +8 0 0 0 0
indolicidin 13 0 +3 0 0 0 0
tritrpticin 13 0 +4 0 0 0 0
human IAPP 37 0.77 +2 0 0.7 1.7 2.2
transportan 27 0.74 +4 0 0 0.8 1.8
a
Secondary structure contents were determined with the DSSP program.

From Figure 3, it can be seen that the size of the pore lined by
the longest peptide, LL-37, is noticeably larger than the other
two. The rodlike form of the α-helix promotes peptide−peptide
aggregation, with α-helices tending to associate with their long
axes parallel. The average size of such clusters will tend to
increase for longer peptide length, thus enhancing cooperative
adsorption to the pore. Consistent with this is the behavior of
Figure 4. (A) Stabilization of membrane pore by neutral α-helical mastoparan. Despite its high secondary structure content, it is
magainin-2 peptides. (B−D) Stabilization of membrane pores by the shortest α-helical peptide we have studied and is unable to
coiled forms of magainin-2, Baxα5, and LL-37 peptides. maintain an open membrane pore over the UCPS simulation
time (see Table 1). This is supported by the experiments of
bacterial cells.53−55 It is known that many amphipathic peptides Arbuzova and Schwarz which showed that, while mastoparan
undergo a coil to α-helix transition upon adsorbing onto the can generate pores in POPC lipid vesicles, they have a small
cell membrane.37 While the functional role of the helical average diameter (∼1 nm) and relatively short lifetimes.58
structure is elusive, our simulations imply that it helps to 3.4. Stable Pore Size and Release Kinetics Induced by
stabilize the membrane pore. To further investigate this Peptides. Table 1 lists the inner diameters of the pores at the
conjecture, we repeated the UCPS simulations for α-helical end of the UCPS simulations, calculated with the method
magainin-2, Baxα5, and LL-37. However, this time the model described above. The time evolution of the pore diameters (see
constraints, which maintained the helical structures of the Figure S1) indicates that when the pores persist, they seem to
peptides, were removed. The resulting snapshots of the fluctuate about an equilibrium value over the final one-third of
equilibrium membrane pores are shown in Figure 4B−D. We the simulation time. So we used the last 500 ns in order to
found that the coiled forms of magainin-2 and LL-37 peptides estimate the pore diameters. The results indicate that, for a
were still able to aggregate in the pore and keep it open for the given peptide, the average size of the membrane pore (if it
duration of the simulations, but the pore diameters were persists at the end of the simulation) increases with the peptide
significantly smaller than those with the structured peptides. concentration. This is consistent with several experimental
On the other hand, the coiled form of Baxα5 gave a closed findings.15,59 Furthermore, the calculated diameters are in good
pore. Taken together, these results suggest that the agreement with experimental data, when available at similar
unstructured peptides impart less stabilization to the pore. It peptide concentrations. For example, at P:L = 1:48, the
is possible that a peptide must possess a molecular rigidity simulated inner diameter of the LL-37-stabilized membrane
(imparted by its secondary structure) in order to stabilize pores pore is 6.3 nm, which compares favorably with the
effectively. These results are consistent with experimental experimentally measured value of 6.6 nm.17 At P:L = 1:36,
studies which have reported that peptide rigidity is an the inner diameter of a magainin-2-stabilized pore was
important variable in modulating the ability of peptides to determined to be 3.6 nm, which is close to the experimental
disrupt membranes.56,57 The α-helical forms of the peptides are value of 3.7 nm (albeit for a mixed DMPC/DMPG bilayer).60
also likely to aggregate with each other more readily, as there is The simulated inner diameter of the melittin stabilized pore is
less configurational entropy cost to form aggregated structures 2.4 nm, at a peptide concentration of 1:48, and this result is
of rods, compared with random coils. Furthermore, α-helix and close to the experimentally reported range of 2.5−3.0 nm for
β-sheet secondary structures give greater hydrophobicity to the melittin with a POPC bilayer.61 Finally, at P:L = 1:36, the
peptide. This has been shown for the MARTINI force field in simulated inner diameter of a protegrin-1-stabilized pore is 2.6
particular.29 The peptide length may also be a factor as to why nm, while Mani et al. reported an experimental value of 2.1 nm
random coil magainin-2 and LL-37 are able to stabilize pores, (for a POPE/POPG bilayer).62 Given the approximate form of
but not Baxα5. Both magainin-2 and LL-37 are relatively longer the MARTINI model, it gives surprisingly good predictions for
than Baxα5. However, even in their folded forms, magainin-2, the ultimate diameters of these peptide-stabilized pores. This
Baxα5, and LL-37 display different equilibrium pore diameters. perhaps reflects the fact that the peptide properties that
757 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

determine pore size (and stability) are generic ones, which can
be effectively modeled at the coarse-grained level.
The size and lifetime of pores have been investigated in
several kinetic studies on the release of fluorescent dyes from
the interior of vesicles.63 Wheaten et al.64 have investigated
release kinetics using kinetic Monte Carlo simulations and
established that the character of the dye release kinetics can be
explained by the rates of pore opening and closing, relative to
the rate of dye efflux. In particular, faster rates of pore closure
can lead to so-called graded release, and slower rates of closure
give all-or-none release. In graded release, vesicles are
homogeneous, so at around the half-life of the dye efflux
process most vesicles contain half the original amount of dye.
In all-or-none release, at the same time point, most vesicles Figure 5. Simulation derived prefibrillar IAPP structure. Individual
either contain all the original amount of dye or are empty. peptide is colored differently for clarification.
Four of the peptides studied here have been reported to
exhibit all-or-none kinetics; they are magainin-2, Baxα5, LL-37, secondary structure contents (0.82 for LL-37 and 0.77 for
and defensin-1. These peptides also give rise to large stable human IAPP). On the other hand, LL-37 contains 5 anionic
membrane pores in our simulations. On the other hand, residues and 11 cationic residues (with a net +6 charge),
mastoparan and transportan, which are known to give graded whereas human IAPP contains only 2 cationic residues. The
release kinetics, give either no stable pore (mastoparan) or a reduced electrostatic repulsions between peptides as well as the
pore with only a small radius (transportan) even at the highest weakened affinity of the peptide to the membrane head groups
peptide concentration. It is tempting to use the final pore would lead to human IAPP aggregation at the pore edges, while
diameter of our UCPS simulations as a predictor for the type of LL-37 displays no such structuring. While IAPP peptide
release kinetics induced by a peptide. Those peptides that give aggregation is also possible in the bulk solution, it is the
large, relatively stable pores with long lifetimes allow unfettered higher concentration of peptide at the pore edges that
transport out of vesicles via all-or-none kinetics. The peptides promotes their nucleation. It is well-known that human IAPP
that give small stable pores at high peptide concentrations can is able to form fibrillar structures in its amyloidal form.67
induce graded dye release by transient pores that are large when However, it is the prefibrillar structures which have been
initially nucleated, then either rapidly shrink to a size, too small implicated in their cytotoxic activity, via a mechanism as yet
to allow dye efflux, or else close altogether. In this context, it is undetermined.68 The stalklike α-helical aggregate that we
interesting to note that there is some uncertainty regarding the observe in our simulations is a plausible first member of
release kinetics induced by melittin. Benachir and Lafleur65 prefibrillar structures, which are formed in a cascade of
reported that melittin induced all-or-none release of fluorescent reactions that ultimately lead to amyloidal fibrils via cooperative
dyes whereas Rex and Schwarz66 asserted a graded mechanism. crossed β-sheet formation within the peptide stack. The
Somewhat consistent with these findings is our observation that nucleation of the stalk-like α-helical aggregate by the membrane
melittin gives a membrane pore of intermediate size at the pore may thus explain the association of prefibrils with
highest peptide concentration. cytotoxicity (brought about by long-lived membrane pores).
3.5. Human IAPP: Prefibrils and Membrane Pores. Of 3.6. Orientations of α-Helical Peptides in the
course, we cannot expect that our simulations will always Membrane Pore. While the literature contains much debate
provide reliable predictions for release kinetics induced by on the subject, it is often asserted that the majority of pore-
specific peptides. For example, human IAPP peptide has been forming peptides create toroidal pores, characterized by
reported to exhibit all-or-none release kinetics,6 even though we outwardly arranged lipid heads throughout the pore surface.
predict the final pore diameter, induced by IAPP, to be It is also probably fair to say that prevailing opinion asserts that
generally smaller than that of mellitin. However, IAPP is peptides in pores are oriented (mainly) perpendicularly to the
noteworthy for another reason. We found that this peptide bilayer plane.3 Contradicting this “classic” picture are
exhibits an interesting supramolecular structuring at the pore, simulations by Marrink and co-workers, who reported a so-
which is completely different to that seen for the other peptides called disordered toroidal pore structure.26 This type of pore is
we investigated. As with the other the peptides, human IAPP characterized by α-helical peptides being randomly oriented
readily adsorbed within the pore and at its edge. However, within the pore. Vácha and Frenkel (using a very coarse-grained
when the peptide aggregated at the pore, they also appeared to peptide rod model) also observed a novel “double-belt”
nucleate a stalklike structure of stacked molecules with aligned arrangement for some peptides.49 This structure seems
α-helices that projected away from the pore into the characteristic of longer peptides, such as the 189-residue
surrounding solution (Figure 5). This stacking seems to be apolipoprotein Apo-A1 peptide, which appears to stabilize
caused by the combination of the rigidity and low electrostatic pores by adsorbing onto the inner surface in a parallel
repulsion between peptides. As noted earlier, rigid peptides orientation. Indeed, two rows of peptides in this configuration
(with high α-helical content) pay less entropy cost with were observed (hence the nomenclature “double belt”). Figure
aggregation. Additionally, IAPP has a charge of only +2, which 6 shows typical pore configurations that we obtained for α-
means that its stacking will be dominated by the attractive LJ helical magainin-2, Baxα5, melittin, and LL-37 peptides, with a
interactions. In this context, it is worth noting that the peptide peptide concentration of P:L = 1:36. As discussed earlier, our
LL-37 is very similar to human IAPP, except for its charge. Both simulated membrane pores had diameters which were very
peptides contain 37 residues (14 hydrophobic residues for LL- close to experiments, so these configurations give a realistic
37 and 12 hydrophobic residues for human IAPP) and high representation of the surface area available to the adsorbed
758 DOI: 10.1021/la5038266
Langmuir 2015, 31, 752−761
Langmuir Article

somewhat similar to the longer peptide orientations observed


by Vácha and Frenkel.49 Such a distribution is caused by the
mismatch between the bilayer thickness and the relatively
longer peptide. The parallel orientation means that LL-37
creates an equilibrium pore with a larger diameter than shorter
Figure 6. Side views of α-helical peptide-stabilized membrane pores: peptides.
(A) magainin-2, (B) Baxα5, (C) melittin, and (D) LL-37. Waters are
colored cyan, lipid phosphate groups are colored tan, and peptides are 4. CONCLUSION
colored differently.
Using MARTINI coarse-grained simulations, we have shown
peptides.17,60,61 The average numbers of magainin-2, Baxα5, that amphipathic peptides are able to recognize and stabilize
melittin, and LL-37 peptides that we observed were 19, 27, 22, membrane pores by aggregating with high affinity within and at
and 25, respectively. The number for melittin is much larger the edge of membrane pores. Our conclusions rely on the
than the 1−2 peptides observed in earlier simulations of a much ability of the MARTINI model to adequately describe peptide
smaller disordered toroidal pore.26 OCD measurements of Lee interactions with zwitterionic lipid bilayers. We have already
et al. for melittin-induced pores in a POPC bilayer suggest that noted that this force field is able to predict translocation of
only 4−7 peptides are found on average in a perpendicular amino acid residues through zwitterionic bilayers with
orientation (presumably in the membrane pore). In this case, reasonable accuracy.32 Additionally, we have carried out further
the experimentally determined pore diameter was 4.4 nm at P:L simulations to ascertain if the polarizable MARTINI model
= 1:15.12 This diameter was obtained at a much higher peptide predicts similar results. This force field seems to give a better
concentration than was used in our work (which may account description of bilayer thermodynamics, though it is computa-
for its larger reported size). However, it does appear tionally much more expensive to implement.69 We tested it
inconsistent with the larger number of pore-bound peptides with the magainin-2 peptide, with the result that the
we observed in our simulations. This discrepancy can be largely recognition and stabilization of the ruptured membrane pore
traced to the orientational distribution of peptides in pores. In by the peptide were still captured (see Figure S2).
the classic toroidal pore structure, peptides are usually assumed We found that the secondary structure content, peptide
to lie perpendicular to the plane of the membrane (I-state). In length, and hydrophobic interactions significantly affect the
reality, however, this may not be the case. accumulation of peptide in the pore. On the other hand,
From the final 500 ns of the UCPS simulation, we calculated peptide charge seemed to have little influence, at least for the
the tilt angle distributions of peptides within the pore. In Figure zwitterionic lipids used in this study. The simulated membrane
7, we have plotted the distributions for the α-helical peptides pore sizes were in good agreement with available experimental
magainin-2, Baxα5, melittin, and LL-37. It is clear that the data, which lends some credence to our findings. Our results
orientations of magainin-2, Baxα5, and melittin peptides in the suggest that α-helical peptides in the pore are generally
membrane pore are quite disperse. For example, the tilt angles oriented in a largely disordered fashion, consistent with the
for magainin-2 span the range 22°−85°, with an average value disordered pore model, at least for peptides of moderate length.
of 50.3°. This distribution is very similar to that of a disordered The longer LL-37 peptide adopted a parallel orientation, similar
toroidal pore observed by Marrink et al.26 OCD measurements to the “double-belt” configurations first described by Vácha and
determine an effective fraction of pore-bound peptides in the Frenkel.49 Finally, we note that for the human IAPP peptide the
so-called I-state, i.e., having a perpendicular orientation. This presence of the pore appears to nucleate prefibrillar structures
fraction corresponds to averaging cos2 θ over the tilt angle made up of stacked α-helices, which may explain why
distribution. Hence, the actual number of peptides in a cytotoxicity and prefibrillar structures seem to be associated
membrane pore (as predicted by OCD) will be usually for these peptides.


underestimated if the orientation of all peptides is assumed
to be perpendicular. For example, from our simulations, we ASSOCIATED CONTENT
determine an average value of 7−8 magainin-2 peptides in the
I-state (for a pore with a diameter of 3.6 nm), much smaller *
S Supporting Information

than the actual number of peptides present (approximately 19). Time evolution of the inner diameters of pores in the presence
Interestingly, the tilt angle distribution profile for LL-37 of different peptides during the unconstrained pore stage
peptides in the membrane pore suggests that LL-37 peptides (UCPS) simulations; the recognition and stabilization of
are orientated more parallel to the membrane surface, membrane pore by magainin-2 peptide modeled with the

Figure 7. Tilt angle distribution profiles for α-helical peptides in the membrane pore: (A) magainin-2, (B) Baxα5, (C) melittin, and (D) LL-37.

759 DOI: 10.1021/la5038266


Langmuir 2015, 31, 752−761
Langmuir Article

polarizable MARTINI force field. This material is available free (16) Parente, R. A.; Nir, S.; Szoka, F. C. Mechanism of Leakage of
of charge via the Internet at http://pubs.acs.org. Phospholipid Vesicle Contents Induced by the Peptide GALA.


Biochemistry 1990, 29, 8720−8728.
(17) Lee, C. C.; Sun, Y.; Qian, S.; Huang, H. W. Transmembrane
AUTHOR INFORMATION Pores Formed by Human Antimicrobial Peptide LL-37. Biophys. J.
Corresponding Author 2011, 100, 1688−1696.
*E-mail: c.woodward@adfa.edu.au (C.E.W.). (18) Wimley, W. C.; Selsted, M. E.; White, S. H. Interactions
between Human Defensins and Lipid Bilayers: Evidence for Formation
Notes of Multimeric Pores. Protein Sci. 1994, 3, 1362−1373.
The authors declare no competing financial interest. (19) Rapaport, D.; Peled, R.; Nir, S.; Shai, Y. Reversible Surface


Aggregation in Pore Formation by Pardaxin. Biophys. J. 1996, 70,
2502−2512.
ACKNOWLEDGMENTS (20) Ostolaza, H.; Bartolome, B.; Dezarate, I. O.; Delacruz, F.; Goni,
An allocation time from the Lunarc Computing Center at Lund F. M. Release of Lipid Vesicle Contents by the Bacterial Protein Toxin
University is gratefully acknowledged. J.F. acknowledges Alpha-Haemolysin. Biochim. Biophys. Acta 1993, 1147, 81−88.
financial support from the Swedish Research Council. (21) van Rooijen, B. D.; Claessens, M. M. A. E.; Subramaniam, V.


Membrane Permeabilization by Oligomeric alpha-Synuclein: In Search
of the Mechanism. PLoS One 2010, 5.
REFERENCES (22) Bergstrom, C. L.; Beales, P. A.; Lv, Y.; Vanderlick, T. K.; Groves,
(1) Ciobanasu, C.; Siebrasse, J. P.; Kubitscheck, U. Cell-Penetrating J. T. Cytochrome C Causes Pore Formation in Cardiolipin-Containing
HIV1 TAT Peptides Can Generate Pores in Model Membranes. Membranes. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 6269−6274.
Biophys. J. 2010, 99, 153−162. (23) Leontiadou, H.; Mark, A. E.; Marrink, S. J. Antimicrobial
(2) Islam, M. Z.; Ariyama, H.; Alam, J. M.; Yamazaki, M. Entry of Peptides in Action. J. Am. Chem. Soc. 2006, 128, 12156−12161.
Cell-Penetrating Peptide Transportan 10 into a Single Vesicle by (24) Santo, K. P.; Berkowitz, M. L. Difference between Magainin-2
Translocating Across Lipid Membrane and Its Induced Pores. and Melittin Assemblies in Phosphatidylcholine Bilayers: Results from
Biochemistry 2014, 53, 386−396. Coarse-Grained Simulations. J. Phys. Chem. B 2012, 116, 3021−3030.
(3) Brogden, K. A. Antimicrobial Peptides: Pore Formers or (25) Santo, K. P.; Irudayam, S. J.; Berkowitz, M. L. Melittin Creates
Metabolic Inhibitors in Bacteria? Nat. Rev. Microbiol. 2005, 3, 238− Transient Pores in a Lipid Bilayer: Results from Computer
250. Simulations. J. Phys. Chem. B 2013, 117, 5031−5042.
(4) Lee, M. T.; Huang, W. C.; Chen, F. Y.; Huang, H. W. Mechanism (26) Sengupta, D.; Leontiadou, H.; Mark, A. E.; Marrink, S. J.
and Kinetics of Pore Formation in Membranes by Water-Soluble Toroidal Pores Formed by Antimicrobial Peptides Show Significant
Amphipathic Peptides. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 5087. Disorder. Biochim. Biophys. Acta, Biomembr. 2008, 1778, 2308−2317.
(5) Last, N. B.; Miranker, A. D. Common Mechanism Unites (27) Evans, E.; Heinrich, V.; Ludwig, F.; Rawicz, W. Dynamic
Membrane Poration by Amyloid and Antimicrobial Peptides. Proc. Tension Spectroscopy and Strength of Biomembranes. Biophys. J.
Natl. Acad. Sci. U. S. A. 2013, 110, 6382−6387. 2003, 85, 2342−2350.
(6) Last, N. B.; Rhoades, E.; Miranker, A. D. Islet Amyloid (28) Leontiadou, H.; Mark, A. E.; Marrink, S. J. Molecular Dynamics
Polypeptide Demonstrates a Persistent Capacity to Disrupt Membrane Simulations of Hydrophilic Pores in Lipid Bilayers. Biophys. J. 2004,
Integrity. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 9460−9465. 86, 2156−2164.
(7) García-Sáez, A. J.; Chiantia, S.; Salgado, J.; Schwille, P. Pore (29) Monticelli, L.; Kandasamy, S. K.; Periole, X.; Larson, R. G.;
Formation by a Bax-Derived Peptide: Effect on the Line Tension of Tieleman, D. P.; Marrink, S. J. The MARTINI Coarse-Grained Force
the Membrane Probed by AFM. Biophys. J. 2007, 93, 103−112. Field: Extension to Proteins. J. Chem. Theory Comput. 2008, 4, 819−
(8) Qian, S.; Wang, W. C.; Yang, L.; W, H. H. Structure of 834.
Transmembrane Pore Induced by Bax-Derived Peptide: Evidence for (30) Marrink, S. J.; Risselada, H. J.; Yefimov, S.; Tieleman, D. P.; de
Lipidic Pores. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 17379−17383.
Vries, A. H. The MARTINI Force Field: Coarse Grained Model for
(9) Tamba, Y.; Ariyama, H.; Levadny, V.; Yamazaki, M. Kinetic
Biomolecular Simulations. J. Phys. Chem. B 2007, 111, 7812−7824.
Pathway of Antimicrobial Peptide Magainin 2-Induced Pore
(31) Bennett, W.; Tieleman, D. Water Defect and Pore Formation in
Formation in Lipid Membranes. J. Phys. Chem. B 2010, 114, 12018−
Atomistic and Coarse-Grained Lipid Membranes: Pushing the Limits
12026.
(10) Gregory, S. M.; Pokorny, A.; Almeida, P. F. F. Magainin 2 of Coarse Graining. J. Chem. Theory Comput. 2011, 7, 2981−2988.
Revisited: A Test of the Quantitative Model for the All-or-None (32) de Jong, D. H.; Singh, G.; Bennett, W. F. D.; Arnarez, C.;
Permeabilization of Phospholipid Vesicles. Biophys. J. 2009, 96, 116− Wassenaar, T. A.; Schafer, L. V.; Periole, X.; Tieleman, D. P.; Marrink,
131. S. J. Improved Parameters for the Martini Coarse-Grained Protein
(11) Fuertes, G.; García-Sáez, A. J.; Esteban-Martín, S.; Giménez, D.; Force Field. J. Chem. Theory Comput. 2013, 9, 687−697.
Sánchez-Muñoz, O. L.; Schwille, P.; Salgado, J. Pores Formed by Bax (33) MacCallum, J. L.; Bennett, W. F. D.; Tieleman, D. P.
alpha 5 Relax to a Smaller Size and Keep at Equilibrium. Biophys. J. Distribution of Amino Acids in a Lipid Bilayer from Computer
2010, 99, 2917−2925. Simulations. Biophys. J. 2008, 94, 3393−3404.
(12) Lee, M. T.; Sun, T. L.; Huang, W. C.; Huang, H. W. Process of (34) Wimley, W. C.; White, S. H. Experimentally Determined
Inducing Pores in Membranes by Melittin. Proc. Natl. Acad. Sci. U. S. A. Hydrophobicity Scale for Proteins at Membrane Interfaces. Nat. Struct.
2013, 110, 14243−14248. Biol. 1996, 3, 842−848.
(13) Gregory, S. M.; Cavenaugh, A.; Journigan, V.; Pokorny, A.; (35) Tieleman, D.; Leontiadou, H.; Mark, A. E.; Marrink, S. J.
Almeida, P. F. F. A Quantitative Model for the All-or-None Simulation of Pore Formation in Lipid Bilayers by Mechanical Stress
Permeabilization of Phospholipid Vesicles by the Antimicrobial and Electric Fields. J. Am. Chem. Soc. 2003, 125, 6382−6383.
Peptide Cecropin A. Biophys. J. 2008, 94, 1667−1680. (36) Bennett, W. F. D.; Sapay, N.; Tieleman, D. P. Atomistic
(14) Karatekin, E.; Sandre, O.; Guitouni, H.; Borghi, N.; Puech, P. Simulations of Pore Formation and Closure in Lipid Bilayers. Biophys.
H.; Brochard-Wyart, F. Cascades of Transient Pores in Giant Vesicles: J. 2014, 106, 210−219.
Line Tension and Transport. Biophys. J. 2003, 84, 1734−1749. (37) Sato, H.; Felix, J. B. Peptide-Membrane Interactions and
(15) Huang, H. W. Molecular Mechanism of Antimicrobial Peptides: Mechanisms of Membrane Destruction by Amphipathic Alpha-Helical
The Origin of Cooperativity. Biochim. Biophys. Acta, Biomembr. 2006, Antimicrobial Peptides. Biochim. Biophys. Acta, Biomembr. 2006, 1758,
1758, 1292−1302. 1245−1256.

760 DOI: 10.1021/la5038266


Langmuir 2015, 31, 752−761
Langmuir Article

(38) Pronk, S.; Páll, S.; Schulz, R.; Larsson, P.; Bjelkmar, P.; (57) Liu, L.; Fang, Y.; Huang, Q. S.; Wu, J. H. A Rigidity-Enhanced
Apostolov, R.; Shirts, M. R.; Smith, J. C.; Kasson, P. M.; van der Spoel, Antimicrobial Activity: A Case for Linear Cationic α-Helical Peptide
D.; Hess, B.; Lindahl, E. GROMACS 4.5: A High-Throughput and HP(2-20) and Its Four Analogues. PLoS One 2011, 6, e16441.
Highly Parallel Open Source Molecular Simulation Toolkit. Bio- (58) Arbuzova, A.; Schwarz, G. Pore-Forming Action of Mastoparan
informatics 2013, 29, 845−854. Peptides on Liposomes: A Quantitative Analysis. Biochim. Biophys.
(39) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Acta, Biomembr. 1999, 1420, 139−152.
DiNola, A.; Haak, J. R. Molecular Dynamics with Coupling to An (59) Matsuzaki, K.; Yoneyama, S.; Miyajima, K. Pore Formation and
External Bath. J. Chem. Phys. 1984, 81, 3684. Translocation of Melittin. Biophys. J. 1997, 73, 831−838.
(40) Mihajlovic, M.; Lazaridis, T. Antimicrobial Peptides Bind More (60) Ludtke, S. J.; He, K.; Heller, W. T.; Harroun, T. A.; Yang, L.;
Strongly to Membrane Pores. Biochim. Biophys. Acta, Biomembr. 2010, Huang, H. W. Membrane Pores Induced by Magainin. Biochemistry
1798, 1494−1502. 1996, 35, 13723−13728.
(41) Rakowska, P.; Jiang, H. B.; Ray, S.; Pyne, A.; Lamarre, B.; Carr, (61) Ladokhin, A. S.; Selsted, M. E.; White, S. H. Sizing Membrane
M.; Judge, P. J.; Ravi, J.; Gerling, U. I. M.; Koksch, B.; Martyna, G. J.; Pores in Lipid Vesicles by Leakage of Co-Encapsulated Markers: Pore
Hoogenboom, B. W.; Watts, A.; Crain, J.; Grovenor, C. R. M.; Formation by Melittin. Biophys. J. 1997, 72, 1762−1766.
Ryadnov, M. G. Nanoscale Imaging Reveals Laterally Expanding (62) Mani, R.; Cady, S. D.; Tang, M.; Waring, A. J.; Lehrert, R. I.;
Antimicrobial Pores in Lipid Bilayers. Proc. Natl. Acad. Sci. U. S. A. Hong, M. Membrane-Dependent Oligomeric Structure and Pore
2013, 110, 8918−8923. Formation of Beta-Hairpin Antimicrobial Peptide in Lipid Bilayers
(42) Connelly, L.; Jang, H.; Arce, F. T.; Capone, R.; Kotler, S. A.; from Solid-State NMR. Proc. Natl. Acad. Sci. U. S. A. 2006, 103,
Ramachandran, S.; Kagan, B. L.; Nussinov, R.; Lal, R. Atomic Force 16242−16247.
Microscopy and MD Simulations Reveal Pore-Like Structures of All- (63) Almeida, P. F. Membrane-Active Peptides: Binding, Trans-
D-Enantiomer of Alzheimer’s beta-Amyloid Peptide: Relevance to the location, and Flux in Lipid Vesicles. Biochim. Biophys. Acta, Biomembr.
Ion Channel Mechanism of AD Pathology. J. Phys. Chem. B 2012, 116, 2014, 1838, 2216−2227.
1728−1735. (64) Wheaten, S. A.; Lakshmanan, A.; Almeida, P. F. Statistical
(43) Mihajlovic, M.; Lazaridis, T. Charge Distribution and Imperfect Analysis of Peptide-Induced Graded and All-or-None Fluxes in Giant
Amphipathicity Affect Pore Formation by Antimicrobial Peptides. Vesicles. Biophys. J. 2013, 105, 432−443.
Biochim. Biophys. Acta, Biomembr. 2012, 1818, 1274−1283. (65) Benachir, T.; Lafleur, M. Study of Vesicle Leakage Induced by
(44) Prieto, L.; He, Y.; Lazaridis, T. Protein Arcs May Form Stable Melittin. Biochim. Biophys. Acta, Biomembr. 1995, 1235, 452−460.
Pores in Lipid Membranes. Biophys. J. 2013, 106, 154−161. (66) Rex, S.; Schwarz, G. Quantitative Studies on the Melittin-
(45) Shai, Y. Mechanism of the Binding, Insertion and Destabiliza- Induced Leakage Mechanism of Lipid Vesicles. Biochemistry 1998, 37,
tion of Phospholipid Bilayer Membranes by Alpha-Helical Antimicro- 2336−2345.
bial and Cell Non-Selective Membrane-Lytic Peptides. Biochim. (67) Westermark, P.; Andersson, A.; Westermark, G. T. Islet Amyloid
Biophys. Acta, Biomembr. 1999, 1462, 55−70. Polypeptide, Islet Amyloid, and Diabetes Mellitus. Physiol. Rev. 2011,
(46) Ouberai, M. M.; Wang, J.; Swann, M. J.; Galvagnion, C.; 91, 795−826.
Guilliams, T.; Dobson, C. M.; Welland, M. E. α-Synuclein Senses Lipid (68) Porat, Y.; Kolusheva, S.; Jelinek, R.; Gazit, E. The Human Islet
Packing Defects and Induces Lateral Expansion of Lipids Leading to Amyloid Polypeptide Forms Transient Membrane-Active Prefibrillar
Membrane Remodeling. J. Biol. Chem. 2013, 288, 20883−20895. Assemblies. Biochemistry 2003, 42, 10971−10977.
(47) Bigay, J.; Gounon, P.; Robineau, S.; Antonnu, B. Lipid Packing (69) Marrink, S. J.; Tieleman, D. P. Perspective on the Martini
Model. Chem. Soc. Rev. 2013, 42, 6801−6822.


Sensed by ArfGAP1 Couples COPI Coat Disassembly to Membrane
Bilayer Curvature. Nature 2003, 426, 563−566.
(48) Bhatia, V. K.; Madsen, K. L.; Bolinger, P. Y.; Kunding, A.; NOTE ADDED IN PROOF
Hedegard, P.; Gether, U.; Satamou, D. Amphipathic Motifs in BAR While we cannot discount the possibility that the pores that are
Domains Are Essential for Membrane Curvature Sensing. EMBO J. observed during the UCPS simulations do not always achieve
2009, 28, 3303−3314. equilibrium, many of the peptides we have investigated
(49) Vácha, R.; Frenkel, D. Simulations Suggest Possible Novel significantly prolong the pore lifetimes, indicating their ability
Membrane Pore Structure. Langmuir 2014, 30, 1304−1310.
to increase pore stability.
(50) Puech, P. H.; Borghi, N.; Karatekin, E.; Brochard-Wyart, F. Line
Thermodynamics: Adsorption at a Membrane Edge. Phys. Rev. Lett.
2003, 90, 128304.
(51) Torrent, M.; Andreu, D.; Nogues, V. M.; Boix, E. Connecting
Peptide Physicochemical and Antimicrobial Properties by a Rational
Prediction Model. PLoS One 2011, 6, e16968.
(52) Stromstedt, A.; Ringstad, L.; Schmidtchen, A.; Malmsten, M.
Interaction between Amphiphilic Peptides and Phospholipid Mem-
branes. Curr. Opin. Colloid Interface Sci. 2010, 15, 467−478.
(53) Salay, L. C.; Procopio, J.; Oliveira, E.; Nakaie, C. R.; Schreier, S.
Ion Channel-Like Activity of the Antimicrobial Peptide Tritrpticin in
Planar Lipid Bilayers. FEBS Lett. 2004, 565, 171−175.
(54) Ulvatne, H.; Haukland, H. H.; Olsvik, O.; Vorland, L. H.
Lactoferricin B Causes Depolarization of the Cytoplasmic Membrane
of Escherichia Coli ATCC 25922 and Fusion of Negatively Charged
Liposomes. FEBS Lett. 2001, 492, 62−65.
(55) Falla, T. J.; Karunaratne, D. N.; Hancock, R. E. W. Mode of
Action of the Antimicrobial Peptide Indolicidin. J. Biol. Chem. 1996,
271, 19298−19303.
(56) Lättig-Tünnemann, G.; Prinz, M.; Hoffmann, D.; Behlke, J.;
Palm-Apergi, C.; Morano, I.; Herce, H. D.; Cardoso, M. C. Backbone
Rigidity and Static Presentation of Guanidinium Groups Increases
Cellular Uptake of Arginine-Rich Cell-Penetrating Peptides. Nat.
Commun. 2011, 2, 453.

761 DOI: 10.1021/la5038266


Langmuir 2015, 31, 752−761

Anda mungkin juga menyukai