This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

Article

Cite This: ACS Omega 2018, 3, 6056−6065

Insights into Membrane Translocation of Protegrin Antimicrobial

Peptides by Multistep Molecular Dynamics Simulations
Pin-Kuang Lai* and Yiannis N. Kaznessis*,†
Department of Chemical Engineering and Materials Science, University of Minnesota, 421 Washington Avenue SE, Minneapolis,
Minnesota 55455, United States

ABSTRACT: Protegrin-1 (PG-1) is a cationic arginine-rich antimicro-

bial peptide. It is widely accepted that PG-1 induces membrane
disruption by forming pores that lead to cell death. However, the
insertion mechanism for these highly cationic peptides into the
hydrophobic membrane environment is still poorly understood at the
molecular scale. It has previously been determined that the association of
arginine guanidinium and lipid phosphate groups results in strong
bidentate bonds that stabilize peptide−lipid complexes. It has also been
suggested that arginine residues are able to drag phosphate groups as they
insert inside the membrane to form a toroidal pore. However, whether
bidentate bonds play a significant role in inducing a pore formation
remains unclear. To investigate the role of bidentate complexes in PG-1 translocation, we conducted molecular dynamics
simulations. Two computational electroporation methods were implemented to examine the translocation process. We found
that PG-1 could insert into the membrane, provided the external electric potential is large enough to first induce a water column
or a pore within the lipid bilayer membrane. We also found that the highly charged PG-1 is capable in itself of inducing molecular
electroporation. Substitution of arginines with charge-equivalent lysines showed a markedly reduced tendency for insertion. This
indicates that the guanidinium group likely facilitates PG-1 translocation. Potential of mean force calculations suggests that
peptide insertion inside the hydrophobic environment of the membrane core is not favored. We found that formation of a water
column or a pore might be a prerequisite for PG-1 translocation. We also found that PG-1 can stabilize the pore after insertion.
We suggest that PG-1 could be a pore inducer and stabilizer. This work sheds some light on PG-1 translocation mechanisms at
the molecular level. Methods presented in this study may be extended to other arginine-rich antimicrobial and cell-penetrating
peptides.

1. INTRODUCTION The majority of previous studies focused on certain stages of

Antimicrobial peptides (AMPs) are attracting widespread
interest as an alternative to traditional antibiotic treatments
and a potential solution for the emergence of resistance to
antibiotics.1,2 There are over 2000 AMPs discovered in nature
and documented in publicly available databases.3−5 Protegrin-1
(PG-1 RGGRLCYCRRRFCVCVGR-NH2) is a remarkably
potent, naturally occurring cationic AMP, which has 18
amino acid residues and forms a β-hairpin structure (Figure
1). PG-1 was first isolated from porcine leukocytes.6 It has two
intramolecular disulfide bonds and is rich in arginine (Arg).
The atomic structure of PG-1 was determined from NMR
experiments, in monomeric and dimeric forms and then in
octameric pores inside lipid bilayers.7−9
The antimicrobial mechanism of PG-1 is proposed to be
based on pore formation. It is believed that PG-1 binds onto
membranes and inserts inside the hydrophobic core of lipid
bilayer membranes, where it oligomerizes. Oligomers of PG-1
form pores, through which there is fast, uncontrollable ion
transport. The transmembrane (TM) potential collapses rapidly
because of this transport, and the ensuing osmotic lysis results
in quick bacterial death.10−12
There is a plethora of molecular dynamics (MD) simulations
conducted to investigate the mechanism of action of PG-1.10−16
the pore formation process, such as early membrane surface
binding and dimerization, or on the final pore structure.
However, the process of PG-1 membrane translocation remains
enigmatic from a molecular point of view because PG-1 has a
+7 net charge at physiological pH and is thus not expected to
interact favorably with the hydrophobic core of a lipid bilayer.
In the experiment, the 13C−31P distances for PG-1 in the
lipid membrane were found to be very short, which suggested
the formation of bidentate complexes between guanidinium
and lipid phosphate groups (see Figure 2).17−20 This complex
is also commonly found in other Arg-rich AMPs and cell-
penetrating peptides (CPPs).21 Therefore, it was suggested that
the cationic Arg residues could drag the anionic phosphate
groups as they insert into the hydrophobic part of the
membrane, thereby facilitating a toroidal pore formation.17
However, it is still uncertain whether the association energy of a
bidentate complex suffices to overcome the energy barrier for
membrane insertion. In addition, toroidal pores have also been
found in peptide-free systems.22,23 The presence of a bidentate
Received: March 14, 2018
Accepted: May 8, 2018
Published: June 5, 2018

© 2018 American Chemical Society 6056 DOI: 10.1021/acsomega.8b00483

ACS Omega 2018, 3, 6056−6065
ACS Omega
Figure 1. Structure of PG-1. The arrows represent a β-hairpin
structure pointing from the N-terminal to the C-terminal. Arg side
chains are in magenta. Disulfide bonds are in yellow.
Figure 2. Schematic of the structure formed between the guanidinium
and phosphate groups through combined electrostatic forces and
bidentate hydrogen bonding.
complex might not then be necessary or sufficient for pore
formation. Nevertheless, the experimental distance measure-
ments showed only a result in the octameric pore state. To
better understand the membrane insertion mechanism,
intermediate structures and interactions must be investigated.21
There are a host of studies focusing on the translocation of
peptides across the membrane. Fuertes et al.24 suggested that
membrane-active peptides can serve as inducers and stabilizers
of pores, the formation of which may also be induced by
mechanical or electrical tensions. Sun et al.25 studied the Arg-
rich CPPs of octa-arginine peptides. They applied external
tension on the lipids to induce a pore for peptide translocation.
They concluded that octa-arginine peptides can slow down the
kinetic rate for pore closure of naturally occurring pores. Miteva
et al.26 studied NK-lysin associated to a membrane and
6057
Article
proposed that, provided the peptide is sufficiently charged, the
bound peptides on the membrane suffice to trigger electro-
poration.
Herein, we describe a series of MD simulations performed
using different techniques. The adsorption of PG-1 on the
membrane surface was simulated to confirm the presence of
bidentate complexes. In addition, electroporation simulations
were conducted to study the translocation process. This
protocol has been applied before to study pore formation
and transport of small molecules.27−29 Potential of mean force
(PMF) calculations were finally conducted to examine the
effect of bidentate complexes on the insertion energy barrier.
2. RESULTS AND DISCUSSION
2.1. Binding of PG-1 onto the POPE/POPG Bilayers. To
investigate the interactions in the PG-1_POPE:POPG system,
we performed three independent MD simulations (Sim 1, Sim
2, and Sim 3) by placing PG-1 2.0 nm above the bilayer surface
initially. The simulations were carried out for 50 ns. Figure 3
Figure 3. Distance along the membrane normal (z-axis) between the
PG-1 center of mass and the average location of phosphorus atoms in
the upper leaflet. Distance calculations are performed using all three
sets of simulations for the PG-1 and POPE/POPG system.
illustrates the time trajectory of the center-of-mass distance
between PG-1 and the average position of phosphorus atoms
on the top leaflet. It can be observed that PG-1 rapidly binds
onto the membrane surface in all simulations.
Representative snapshots for Sim 2 are provided in Figure 4.
During the simulations, PG-1 freely diffuses in the water
subphase. We observed that the binding process has several
stages. In the beginning (0−5 ns), PG-1 rotates its axis and
moves toward the lipids rapidly. It interacts with the bilayer
with either the two-terminal side or the β-turn side. Both sides
include a cluster of Arg residues, which indicates that the
Figure 4. Characteristic snapshots of one PG-1 with POPE/POPG
simulation. The Arg residues are represented in sticks. The
phosphorus atoms of POPE and POPG lipids are colored in gray
and in purple, respectively. Phosphorus atoms within 0.3 nm are
highlighted with orange color and their lipid tails are shown in sticks.
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega Article

positively charged residues are essential for the initial contact because of the hydrophobic environment of the lipids. This
through electrostatic interactions. In the next stage (5−20 ns), process is well-described by the solubility−diffusion mecha-
the remaining part of the molecule fluctuates in water until nism.30 On the contrary, for hydrophilic molecules and ions, it
additional Arg residues interact with the lipids. In the last stage appears that the formation of a water column or a pore (we use
(20−50 ns), PG-1 has barely any conformational change and these two terms for contiguous water structures that span the
inserts fully into the lipid head groups. lipid bilayer) can facilitate lipid flip-flop and subsequent
In line with the aforementioned results, the lipid contact ratio transport of polar molecules.31 However, in conventional
presented in Figure 5 provides clear evidence that Arg residues equilibrium MD simulations, it is challenging to observe pore
formation because of the very long simulation time needed.
There have been some studies where pore formation was
observed, but these studies utilized relatively old force fields
found to underestimate the free energy barrier for lipid flip-flop
across the membrane.32−34 More recent atomistic simulations
using CHARMM36 force fields estimate a much larger energy
cost for lipid flip-flop. For example, in recent μs-long
simulations, AMPs did not disrupt the lipid bilayer, and no
pores were observed.35
It has been suggested that electroporation (i.e., the
establishment of a potential across the membrane along with
subsequent phenomena) is a reasonable method to induce pore
formation. Moreover, there have been reports on the transport
of molecules under TM potential using MD simulations. In this
Figure 5. Lipid contact ratio of each PG-1 residue in the simulation. study, we apply electroporation to investigate the PG-1
This ratio is defined as the number of nonhydrogen PG-1 atoms translocation and attempt to link simulations to experimental
within 0.3 nm of POPE or POPG lipids divided by the total number of results. We have carried out a set of simulations listed in Table
nonhydrogen atoms for each residue. Results are averaged over the last 1 to investigate the effect of charge imbalanced methods, TM
20 ns of all simulations.
Table 1. List of Translocation Simulations Performed in
on both sides of the β-hairpin have the largest number of atoms This Worka
in contact with the lipids. The average lipid contact is defined as
the number of nonhydrogen PG-1 atoms within 0.3 nm of charged peptides
case residues (P/L ratio) ΔQ(e) methods
lipids divided by the total number of nonhydrogen atoms. It is
noticed that the N-terminal Arg residue has the highest contact. Arg6_10e_I Arg 6 (6:160) 10 ions
This might be due to the longer length of the N-terminal side Arg6_7e_I Arg 6 (6:160) 7 ions
compared to the C-terminal side (see Figure 1), which allows Arg6_7e_P Arg 6 (6:160) 7 peptides
the N-terminal Arg residue to insert more deeply into the Arg6_14e_I Arg 6 (6:160) 14 ions
bilayer. Arg6_14e_P Arg 6 (6:160) 14 peptides
To further examine peptide−lipid interactions, Figure 6 Arg0_10e_I Arg 0 (0:160) 10 ions
presents the H-bonds formed between Arg residues and lipids. Arg1_10e_I Arg 1 (1:160) 10 ions
Lys6_10e_I Lys 6 (6:160) 10 ions
a
Arg represents the wild-type PG-1, while Lys indicates the charge-
equivalent mutant. There are three PG-1 dimers in high P/L ratio
simulation for a total of six peptides. The three dimers were placed
parallel to the membrane surface initially. The charge imbalance can be
generated by different numbers of ions (I) or peptides (P) in the two
double bilayer systems. The charge imbalance for the peptide method
(P) is generated by keeping three dimers in one system and
subsequently removing dimers in another systems.

potential, P/L ratio, and bidentate bonds. Each system was

Figure 6. Number of hydrogen bonds formed between the Arg
residues and lipids. Results are averaged for all three simulations.
For the three simulations, the mean value of H-bonds in the last
20 ns is approximately 12. On average, each Arg residue forms
two H-bonds through the bidentate hydrogen bonding (see
Figure 2).
2.2. MD Simulations with TM Potential. Membrane
transport of small nonpolar molecules is relatively favorable
6058
simulated three separate times, starting from the same positions
but using different initial velocities in the beginning of MD.
2.2.1. PG-1 Translocation under TM Potential. The
Arg6_10e_I case displayed PG-1 translocation (Figure 7A)
under ΔQ = 10e charge imbalance. At 1.0 ns, a water pore was
formed with some lipids reorienting likely to stabilize the pore.
The peptide started to translocate and at 2.7 ns, the Arg residue
at position 4 (Arg4) reached the bilayer center, followed by the
translocation of the N-terminal Arg residue (Arg1). This
inserted state was also stable until the end of the 10 ns
simulation. In this case, peptide translocation within the
membrane was only partial. Furthermore, we observed that a
strong bidentate hydrogen bond formed between the inserted
Arg residue and lipid phosphate groups during the simulation.
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
Figure 7. Representative snapshots of the electroporation simulations for different cases. (A) Arg6_10e_I. (B) Arg6_14e_I. (C) Arg6_14e_P. (D)
Arg0_10e_I. (E) Arg1_10e_I. (F) Lys6_10e_I. The description of each case is in Table 1.
This indicates that the bidentate complex might be crucial for
Arg insertion, which is consistent with experimental findings.
The TM potential for the Arg6_10e_I allows only one or
two Arg residues to translocate at a time. In the following
analysis, the Arg6_10e_I case is regarded as a baseline for all
other tests.
2.2.2. Comparison of Charge Imbalance Methods. For the
Arg6_7e_I and Arg6_7e_P cases, there was no pore formation
during 10 ns simulations, indicating that the TM potential
generated by ΔQ = 7e was too small. Thereby, no other results
are reported here. Generally, if the TM potential is above a
threshold value, the pore formation can be observed in a few
hundred picoseconds.
Figure 7B,C illustrates PG-1 translocation with two different
charge imbalance methods, Arg6_14e_I and Arg6_14e_P,
respectively. Results were similar with both methods. In both
cases, we observe a dimer translocating into the water pore
from the membrane surface under ΔQ = 14e charge imbalance.
For the Arg6_14e_I case, the TM potential is generated by ions
in the water subphase. In contrast, for the Arg6_14e_P case,
the TM potential is generated by the charges on the PG-1 itself,
suggesting molecular electroporation.26 In other words, the
highly charged PG-1 can generate sufficient TM potential to
induce pore formation and facilitate subsequent translocation as
if an external electric field is applied. This also provides
evidence that PG-1 may serve as a pore inducer.24
Removing one peptide dimer from a bath results in a change
of ΔQ = 7e per bilayer. From the Arg6_7e_P and Arg6_14e_P
6059
Article
cases, the system changes rapidly from no pore formation to
one translocated dimer, as shown in Figure 7C. To have a finer
control of the ΔQ, we focused on the charge imbalance method
from ions only.
2.2.3. Effect of TM Potential. Figure 7A,B illustrates the
results for Arg6_10e_I and Arg6_14e_I systems, respectively.
For the Arg6_14e_I case, a large water pore was formed in the
bilayer within several ps, and a few lipid flip-flops occurred. At
0.5 ns, one Arg residue side chain started to protrude into the
water pore, and it reached the bilayer center by 1.0 ns. In
addition, one PG-1 dimer initially parallel to the membrane
near the pore opening rotated to align along the membrane
normal. Shortly after the insertion of the Arg residue, the whole
dimer translocated into the bilayer and spanned the entire pore.
The inserted state remained stable until the end of the 10 ns
simulation. In contrast, the Arg6_10e_I case only shows partial
Arg insertions for PG-1, and the Arg6_7e_I case shows no pore
formation and peptide translocation at all. Clearly, we can
observe a TM voltage dependence on the pore formation and
PG-1 translocation.
2.2.4. Effect of the P/L Ratio. Previous studies have shown
that AMPs adopt two distinct orientations depending on the
peptide/lipid ratio. In one, the orientation is parallel (at low
ratio) and in another perpendicular (at high ratio) to the bilayer
surface.36 The experimentally determined threshold P/L value
for PG-1 is 1:30.37 The underlying molecular reasons for this
are unclear. In this work, we performed simulations to examine
the concentration dependence of PG-1 translocation at low P/
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
Figure 8. Change of the charge difference ΔQ across each bilayer in the 10 ns simulations. Charge transfers due to PG-1 translocation are
highlighted. The N-terminal Arg is denoted as Arg1, which has +2 charge. The Arg residue at position 4 is denoted as Arg4. All others are due to
POPG flip-flop. Results are shown for the Arg6_10e_I, Arg1_10e_I and Lys6_10e_I systems. The points are added every 20 ps.
L (1:160) and at high P/L ratios (6:160). The high P/L ratio is
above the threshold value indicated from the experiment.
Figure 7D shows snapshots for the Arg0_10e_I case (pure
membrane). With a TM potential, the pure membrane system
can also form water pores or columns. We noticed that the pore
size shrinks during the time interval between 3.0 and 10.0 ns.
The stability of the pore will be discussed later.
Figure 7E shows snapshots for the Arg1_10e_I case. At 1.0
ns, small water pores started to form because of the TM
potential. At 3.0 ns, lipid head groups redistributed toward the
membrane interior. By 10.0 ns, a toroidal-shaped pore
developed. However, during these simulations, PG-1 remained
bound in parallel to the membrane surface. This is in contrast
to the case at the high P/L ratio where PG-1 molecules adopted
a perpendicular orientation to the bilayer in an inserted state
(Figure 7A,B).
PG-1 stayed on the membrane surface for a low P/L ratio,
likely because of the fact that the pore was formed relatively far
from the peptide. This might be attributed to the stiffness of the
bilayer in different domains. As previously shown, PG-1 forms
extensive hydrogen bonds with lipids, which could increase the
stability of this domain. It may be less likely for lipids in the
vicinity of PG-1 to form a water defect, which may later grow
into pores. Similar results have been reported by Reigada38 who
simulated the electroporation process for heterogeneous
bilayers composed of liquid-ordered and liquid-disordered
regions. These authors found that pore formation occurs
preferably in the disordered region. On the contrary, as the P/L
ratio increases, the domain for the peptide-free lipid region
shrinks. Under TM potential, pores may form around PG-1,
and the ensuing lipid movement may trigger PG-1 trans-
location.
2.2.5. Effect of Substitution of Arg with Lys. Numerous
investigations have been conducted to examine Arg-rich
peptides. One particular analysis involved substituting Arg
residues with charge-equivalent Lys residues. In our work, we
substituted Arg residues in the Arg6_10e_I case with Lys
residues in the Lys6_10e_I case.
Figure 7F presents representative snapshots of the
Lys6_10e_I system. Compared to wild-type PG-1, a striking
difference was observed, in that the Lys mutant has a much
reduced tendency for insertion. This finding provides
compelling evidence that the Arg guanidinium group is key
for PG-1 translocation. It lends support to the assumption that
bidentate bonds in Arg−phosphate complexes play a significant
role in aiding PG-1 insertion into the water pore. Indeed, NMR
experiments have shown that Lys−phosphate interactions are
less stable than Arg−phosphate interactions.21 This also
indicates that PG-1 translocation is not simply driven by the
force due to TM potential.
6060
Article
It has been shown in the experiment that Arg and Lys
contribute distinctively to cell uptake for CPPs and
antimicrobial potency for AMPs. For example, Mitchell et al.
has shown that a synthetic CPP, polyarginine (Arg)n, enters the
cell more efficiently than polylysine (Lys)n.39 In addition,
Schmidt et al. reported that cryptdin-4 (Crp4), a potent
bactericidal peptide, showed severely reduced antimicrobial
activity for complete Lys-for-Arg substitution.40 Consistent
with these results, our MD simulations provide molecular
insight, demonstrating that the difference might be attributed to
the impact of bidentate bonds on the propensity of peptide
insertion in a lipid bilayer.
2.2.6. Discharging of TM Potential. As described in the
Methods section, one drawback of electroporation by the
charge imbalance method is that the imbalance is not reset
during the simulation. The TM potential drops rapidly after the
formation of a water pore and subsequent ion transfer. To focus
on the interactions between PG-1 and lipids without a
collapsing potential, we have added weak constraints to prevent
ions from passing through the water pore. However, lipid flip-
flop of anionic palmitoyloleoylphosphatidylglycerol (POPG) or
PG-1 translocation (i.e., Arg transfer) can also result in the TM
potential being discharged. Because the focus of the study is the
translocation of PG-1, along with lipid flip-flop, we did not
restrict these motions with added constraints.
It is instructive to show the change of ΔQ in different cases,
as illustrated in Figure 8. One charge transfer will alter ΔQ by
2e. POPG flip-flop occurred within a few ns, and all observed
flip-flop events were directed from Bath 1 (with excess negative
charge) to Bath 2 (with excess positive charge), (see Figure
12.). Conversely, Arg residues always inserted in the reverse
direction. The region of each bath was divided by the bilayer
center. Phosphorus and Cζ atoms were used to determine the
location of lipids and Arg residues in the baths.
As can be observed in Figure 8, the discharging rate was
much faster for the Arg6_10e_I system. It took less than 4 ns
for the system to be fully discharged. For the Lys6_10e_I
system, it took about 6 ns to discharge. The Arg1_10e_I
system discharged the TM potential in more than 10 ns.
Evidently, the discharging rate is faster for systems with the
higher peptide concentration.
It is also worth mentioning that the ion constraints
implemented in this work could be relaxed and the ion-water
swap techniques may be applied to maintain the constant
voltage across the membrane. However, care must be taken
because all transfers of ions, POPG lipids, or charged residues
result in changes of ΔQ. The user could, in principle, reset the
charge imbalance for all of these events. In practice, one must
always be mindful of the artificial nature of the simulated
systems and processes.
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
2.3. PMF Calculation. It has been proposed that
interactions between charged protein residues and lipid
phosphate groups or water can provide the energy needed to
overcome the insertion energy barrier for charged residues to
insert inside the membrane.21 It is therefore interesting to study
the thermodynamic driving forces of PG-1 insertion.
A total of 3 μs umbrella sampling simulations were
conducted to mimic the insertion processes of lipid, PG-1,
and the lipid/PG-1 complex (Figure 9). The PMF profiles for
these processes were obtained and compared (Figure 10).
Figure 9. Illustration of the pulling groups of (A) lipid, (B) PG-1, and
(C) lipid/PG-1 complex at the bilayer center. The green spheres
highlight the pulling atoms. For (A), the pulling atom is the
phosphorus atom. For (B), the pulling atom is the Cζ atom of the Arg.
For (C), the pulling atom is the phosphorus atom, while the bidentate
hydrogen bonds are restrained by harmonic forces shown in yellow
dashes.
Figure 10. PMFs for moving the pulling atom into the center of a
POPE/POPG bilayer for the lipid, peptide, and bidentate complex.
The PMFs were obtained from umbrella sampling calculations, while
the pulling atom was restrained at a varying distance from the bilayer
center of mass, in the direction normal to the plane of the bilayer. The
error bars were determined by bootstrap analysis.
For the pure palmitoyloleoylphosphatidylethanolamine
(POPE)/POPG bilayer, we obtained a relative free energy
barrier for a single POPE flip-flop of 81.1 kJ/mol. The relative
free energy barrier for a PG-1 Arg residue insertion was 94.9
kJ/mol. Figure 9A,B illustrates the lipid head and the Arg
residue at bilayer center, respectively. We notice that there was
no spontaneous pore formation. Only a water-filled membrane
defect (water column from one side) occurred.
The relative free energy barrier for the lipid/PG-1 complex
was 124.8 kJ/mol. This free energy is still higher than that for a
single lipid flip-flop. The actual timescale required for lipid flip-
flops ranges between minutes and hours.23 With a higher free
energy barrier, it is expected that the timescale for the bidentate
complex translocation may take even longer. Consequently, the
argument that the bidentate complex facilitates pore formation
by dragging phosphate groups as Arg residue inserts into the
hydrophobic part of the membrane appears less convincing.17
6061
Article
It is proposed that formation of transient water pores is the
rate-limiting step in the process of TM lipid flip-flop.41,42 Once
a pore has been formed, the ensuing lipid translocation takes
place on timescales of nanoseconds as observed from our
electroporation simulations. We reason that water pore
formation can not only increase lipid flip-flop rates but also
facilitate the PG-1 translocation.
It should be stressed that the insertion processes used for
PMF calculations are not directly comparable to those of the
electroporation processes. In the latter, a water pore forms first,
and lipids reorient themselves before PG-1 insertion takes
place. In the former, a water defect is formed as the forced
insertion process takes place.
It should also be stressed that the presented PMF
calculations are of artificially built systems. The definition of
the PMF zero level is entirely operational and only pertains to
the presented cases. Consequently, these PMFs alone, without
reactive trajectory sampling, can offer only very limited
mechanistic interpretations. In other words, in the absence of
simulations that accurately capture the vast phase space during
the entire process of peptides moving from the bulk water
subphase onto membranes and then inside the lipid core, any
proposal of molecular mechanisms underlying biological
phenomena must be considered tentative. Unfortunately,
despite amazing improvements in both simulation software
and hardware, such simulations are currently implausible.
Nonetheless, MD simulations, for all their limitations, are often
the only scientifically agreeable approach to capturing the vast
complexity of biomolecular interactions in a way fit for forming
sound hypotheses.
2.4. Stability of Water Pores with PG-1 Inserted. It is
reported from the experiment that membrane pores induced by
membrane-active peptides have a much longer lifetime than
transient pores induced by thermal energy. Computer
simulations also showed that many membrane-active peptides
can stabilize membrane pores. This motivates us to study the
role of PG-1 in pore stability.
Figure 11A shows the dynamics of water pores for the pure
membrane system (Arg0_10e_I) after we release the
Figure 11. Representative snapshots of the MD simulations for (A)
pure membrane pores Arg0_10e_I and (B) membrane pores with PG-
1 inserted from the Arg6_10e_I system.
constraints on ions. It is observed that the transient water
pore was resealed within 30 ns. In contrast, if there is a PG-1
inserted in a water pore (Arg6_10e_I), the pore remained open
for the subsequent 200 ns, as shown in Figure 11B. This is
similar to the kinetic stabilization of octa-arginine peptides
reported by Sun et al.25 Therefore, we reason that PG-1 is able
to stabilize and slow down the kinetic rate of water pore
closure. This also provides evidence that PG-1 can serve as a
pore stabilizer after insertion.24
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
2.5. Proposed Mechanism for PG-1 Translocation. We
have applied a computational electroporation method to study
PG-1 translocation. The highly charged PG-1 generates
molecular electroporation as if an external electric field is
applied. Under conditions of a sufficient TM potential, a water
pore or column will form across the membrane.
There have been extensive studies on the collective motions
of pore formation and lipid flip-flop.41,42 It has been proposed
that formation of pores unavoidably results in lipid flip-flop.
This is called the pore-mediated lipid flip-flop mechanism. In
this study, we found that with the strong interactions of
bidentate bonds in the guanidinium−phosphate complex, it is
likely that the Arg residues might be dragged by the lipids
during the diffusive TM transport through the water pores.
This in turn may lead to PG-1 insertion and translocation. The
charge-equivalent lysine mutant showed reduced permeability.
This shows that PG-1 insertion is not simply driven by strong
TM potential, and the bidentate complex may play an
important role in translocation.
Although we applied electroporation by imposing a charge
imbalance to induce pore formation, it should be emphasized
that transient pores may form under different circumstances.
For example, the mechanical stress created by peptide
aggregation is also considered to produce pores to relieve
membrane line tension.25,43
In addition, we also showed that PG-1 insertion can stabilize
water pores and slow down the kinetic rate of pore closure. We
therefore propose that PG-1 may be capable of inducing pore
formation by electrical or mechanical tension. The pore-
mediated, lipid flip-flop driven PG-1 insertion process facilitates
PG-1 translocation. The inserted PG-1 can further increase the
stability of the water pores. The whole process from PG-1
inducing to stabilizing water pores may form a feedback loop,
thereby highlighting a great example of collective motions in
biological membranes.
3. CONCLUSIONS
We have applied various MD protocols to study the
interactions between PG-1 and lipids that underlie the peptide
translocation process. Arg residues are crucial for PG-1 to bind
on the membrane surface. An extensive network of bidentate
bonds is formed between Arg guanidinium and lipid phosphate
groups. Electroporation simulations illustrate that a strong TM
potential results in water pore formation and the disruption of
the membrane. The charge on the PG-1 is large enough to
create TM potential to induce water pores. Lipid flip-flops and
PG-1 translocation then take place in a few ns near the water
pore. In addition, a threshold P/L appears to be necessary for
the water pores to form in the vicinity of PG-1. Furthermore,
the Arg to Lys substitution significantly reduces the insertion
propensity, which strongly indicates that the bidentate
hydrogen bonds facilitate PG-1 translocation. Moreover, the
inserted PG-1 can stabilize water pores. PMF calculations show
that the barrier for the bidentate complex is still high for
insertion into the hydrophobic environment. We reason that
PG-1 might be a pore inducer to facilitate water pore formation,
which is a rate-limiting step for lipid flip-flop and for PG-1
translocation. After pore formation, lipids that flip-flop may
drag PG-1 by virtue of the bidentate bond strength. The
inserted PG-1 serves as a pore stabilizer after translocation.
6062
Article
4. METHODS AND THEORY
4.1. PG-1 Simulation System Preparation. The solvated
peptide−membrane systems for various simulations were
constructed using CHARMM-GUI.44−47 The initial structures
of the PG-1 monomer and dimer were taken from the protein
data bank (PDB IDs: 1PG1 and 1ZY6, respectively.7,8) The
bilayer was modeled using the 3:1 mixture of POPE and POPG
lipids, mimicking the components of the inner membrane of
Gram-negative bacteria.48 There were 120 POPE molecules
and 40 POPG molecules in the system. POPE is neutral, while
the anionic POPG has a net (−1) charge. TIP3P waters49 were
added on each side of the membrane with a thickness of
approximately 3 nm. The systems were neutralized by adding
potassium and chloride as counter ions. There were 40
potassium atoms, and the number of chloride atoms depended
on the number of simulated peptides.
To investigate the interactions of PG-1 with the POPE and
POPG lipid bilayers, three sets of 50 ns simulations were
performed. The PG-1 monomer was initially placed in the
water subphase, approximately 2.0 nm away from the
phosphorus atoms of the upper membrane leaflet. This system
is denoted as PG-1_POPE:POPG.
4.2. Computational Electroporation Setup. The electro-
poration simulations were carried out to mimic a TM potential
and to facilitate PG-1 translocation. The system was set up with
a double bilayer configuration by duplicating a second bilayer
on top of the original bilayer (Figure 12). PG-1 molecules were
initially placed at the equilibrium position on the membrane
(approximately 2.2 nm from the bilayer center) as determined
by PG-1 membrane simulations described in section 2.1. A TM
potential was generated by two different methods. The first one
is by swapping ions and waters in Bath 1 and in Bath 2, creating
a charge imbalance.50 The second one is to keep PG-1 in Bath 1
and partially remove PG-1 in Bath 2 to create a charge
imbalance. The number of chloride is adjusted to maintain a
neutral system. The number of ions remain the same for both
baths. The former is considered as applying an external electric
field, while the latter is regarded as an intrinsic electric field
generated by the charges on the PG-1 itself. The charge
difference is denoted as ΔQ. The whole system maintained zero
net charge.
Nine test cases were considered (listed in Table 1) to
investigate the effect of four different factors on the PG-1
translocation: (1) the comparison of the two charge imbalance
methods. (2) The influence of TM potential was studied by
varying the charge difference ΔQ in the double bilayer system.
(3) The concentration dependence of peptides was examined
by changing the peptide to lipid (P/L) ratio. (4) The
importance of the guanidinium group was investigated by
substituting Arg with charge-equivalent Lys residues.
One important drawback of the charge imbalance method is
that the imbalance is not reset during the simulation. Once a
charged particle transfers to other sides of the membrane, the
system ΔQ changes. A protocol using a swapping method has
been proposed to maintain a constant charge imbalance.51 In
this procedure, the number of ions in the two solution baths is
counted frequently and, if the number differs from the initial
setup, a swapping action is carried out: an ion of one solution is
exchanged by a water molecule of the other solution bath.
However, this method only accounts for the change of charge
imbalance due to ion flow. This approach has some limitations
in our system. In addition to ions, there are anionic lipids,
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
Figure 12. Schematic representation of the double bilayer system with
a charge imbalance ΔQ across each bilayer. The classical 3D periodic
boundary condition is implemented. POPG lipids are shown in purple.
POPE lipids are shown in gray. Potassium ions are shown in orange.
Chloride ions are shown in green. Peptides on the membrane surface
are shown in spherical beads. Waters are represented with a
semitransparent surface.
POPG and cationic PG-1 molecules, in the system. Any POPG
lipid flip-flop or peptide transport through the membrane alters
the imbalance as well. It is not clear how to best deal with some
of these events. In this study, to avoid such situations, we added
a weak one-dimensional constraint (force constant 100 kJ
mol−1 nm−2) on the ion positions along the membrane normal
(z-direction). This constraint still allows ions to diffuse in their
original baths but prevents them from passing across the
membrane. Although the system could still discharge because of
translocation of lipids or peptides, time scales for such events
are slow enough for the dynamic events of interest to take place
well before the system is fully discharged, as shown in the
Results and Discussion section.
4.3. PMF Calculation. To calculate the free energy of PG-1
translocation and to shed light on the importance of the
bidentate complex, we performed umbrella sampling simu-
lations for three systems. In system A, the lipid flip-flop free
energy was obtained by pulling a phosphorus atom of a single
lipid into the bilayer center. In system B, the energy cost of PG-
1 insertion was calculated by pulling a Cζ atom on an Arg
residue into the bilayer center. In system C, the free energy of
the combined lipid and PG-1 system was measured by pulling
into the bilayer center a phosphorus atom on the lipid that has
harmonic constraints on the bidentate hydrogen bonds.
The reaction coordinate of the PMF was defined as the
distance to the bilayer center along the bilayer normal. The
6063
Article
window size was 0.1 nm. An umbrella potential of a force
constant 3000 kJ mol−1 nm−2 was used for each window to hold
the pulling groups together. A production run of 50 ns was
performed for each window. The PMFs were evaluated using
the g_wham tool in GROMACS.52
4.4. MD Simulation Details. All simulations in this work
were conducted using the GROMACS 5.1 package.53 All-atom
simulations were run at a temperature of 310.15 K using a
Nosé−Hoover thermostat54,55 with a coupling time constant of
1.0 ps. The semi-isotropic pressure coupling method was used
with both lateral and perpendicular pressures (both 1 atm)
independently coupled to the Parrinello−Rahman barostat56
with a coupling time constant of 5 ps and a compressibility of
4.5 × 10−5. Periodic boundary conditions were employed. The
simulation time step was 2 fs. The CHARMM36 force field57
was used for proteins and lipids, and the TIP3P model49 was
used for waters. The cutoffs for the van der Waals interactions
and electrostatic interactions calculated using the particle-mesh
Ewald method58 were both 1.2 nm. The van der Waals
interactions were smoothly turned off between 1.0 and 1.2 nm
with the force-switching method. The LINCS algorithm59 was
used to constrain all of the hydrogen bonds.
■ AUTHOR INFORMATION
Corresponding Authors
*E-mail: laixx266@umn.edu (P.-K.L.).
*E-mail: yiannis@umn.edu. Phone: 651-503-2696. Fax: 612-
626-7246 (Y.N.K.).
ORCID
Pin-Kuang Lai: 0000-0003-2894-3900
Yiannis N. Kaznessis: 0000-0002-5088-1104
Present Address
†
General Probiotics Inc., 1000 Westgate Dr, Ste 122, St. Paul,
MN 55114 (Y.N.K.).
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
This work was supported by grants from the National Institutes
of Health (grant no. GM111358) and by a grant from the
National Science Foundation (grant no. CBET-1412283). We
acknowledge computational support from the Minnesota
Supercomputing Institute (MSI) and from the Extreme Science
and Engineering Discovery Environment (XSEDE), which is
supported by National Science Foundation grant no. ACI-
10535753. Support from the University of Minnesota Digital
Technology Center and the University of Minnesota Institute
for Engineering in Medicine is also acknowledged.
■
REFERENCES
(1) Zasloff, M. Antimicrobial peptides of multicellular organisms.
Nature 2002, 415, 389−395.
(2) Brogden, K. A. Antimicrobial peptides: pore formers or metabolic
inhibitors in bacteria? Nat. Rev. Microbiol. 2005, 3, 238−250.
(3) Wang, Z.; Wang, G. APD: the antimicrobial peptide database.
Nucleic Acids Res. 2004, 32, D590−D592.
(4) Wang, G.; Li, X.; Wang, Z. APD2: the updated antimicrobial
peptide database and its application in peptide design. Nucleic Acids
Res. 2008, 37, D933−D937.
(5) Wang, G.; Li, X.; Wang, Z. APD3: the antimicrobial peptide
database as a tool for research and education. Nucleic Acids Res. 2016,
44, D1087−D1093.
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega
(6) Steinberg, D. A.; Hurst, M. A.; Fujii, C. A.; Kung, A. H.; Ho, J. F.;
Cheng, F. C.; Loury, D. J.; Fiddes, J. C. Protegrin-1: a broad-spectrum,
rapidly microbicidal peptide with in vivo activity. Antimicrob. Agents
Chemother. 1997, 41, 1738−1742.
(7) Fahrner, R. L.; Dieckmann, T.; Harwig, S. S. L.; Lehrer, R. I.;
Eisenberg, D.; Feigon, J. Solution structure of protegrin-1, a broad-
spectrum antimicrobial peptide from porcine leukocytes. Chem. Biol.
1996, 3, 543−550.
(8) Mani, R.; Tang, M.; Wu, X.; Buffy, J. J.; Waring, A. J.; Sherman,
M. A.; Hong, M. Membrane-bound dimer structure of a β-hairpin
antimicrobial peptide from rotational-echo double-resonance solid-
state NMR. Biochemistry 2006, 45, 8341−8349.
(9) Mani, R.; Cady, S. D.; Tang, M.; Waring, A. J.; Lehrer, R. I.;
Hong, M. Membrane-dependent oligomeric structure and pore
formation of a beta-hairpin antimicrobial peptide in lipid bilayers
from solid-state NMR. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 16242−
16247.
(10) Langham, A. A.; Ahmad, A. S.; Kaznessis, Y. N. On the nature of
antimicrobial activity: a model for protegrin-1 pores. J. Am. Chem. Soc.
2008, 130, 4338−4346.
(11) Bolintineanu, D.; Hazrati, E.; Davis, H. T.; Lehrer, R. I.;
Kaznessis, Y. N. Antimicrobial mechanism of pore-forming protegrin
peptides: 100 pores to kill E. coli. Peptides 2010, 31, 1−8.
(12) Bolintineanu, D. S.; Kaznessis, Y. N. Computational studies of
protegrin antimicrobial peptides: a review. Peptides 2011, 32, 188−201.
(13) Lipkin, R.; Lazaridis, T. Computational prediction of the
optimal oligomeric state for membrane-inserted β-barrels of protegrin-
1 and related mutants. J. Pept. Sci. 2017, 23, 334−345.
(14) Lipkin, R.; Lazaridis, T. Computational studies of peptide-
induced membrane pore formation. Philos. Trans. R. Soc., B 2017, 372,
20160219.
(15) Lipkin, R.; Pino-Angeles, A.; Lazaridis, T. Transmembrane Pore
Structures of β-Hairpin Antimicrobial Peptides by All-Atom
Simulations. J. Phys. Chem. B 2017, 121, 9126−9140.
(16) Lipkin, R. B.; Lazaridis, T. Implicit membrane investigation of
the stability of antimicrobial peptide β-barrels and arcs. J. Membr. Biol.
2015, 248, 469−486.
(17) Tang, M.; Waring, A. J.; Hong, M. Phosphate-mediated arginine
insertion into lipid membranes and pore formation by a cationic
membrane peptide from solid-state NMR. J. Am. Chem. Soc. 2007, 129,
11438−11446.
(18) Rothbard, J. B.; Jessop, T. C.; Lewis, R. S.; Murray, B. A.;
Wender, P. A. Role of membrane potential and hydrogen bonding in
the mechanism of translocation of guanidinium-rich peptides into cells.
J. Am. Chem. Soc. 2004, 126, 9506−9507.
(19) Rothbard, J.; Jessop, T.; Wender, P. Adaptive translocation: the
role of hydrogen bonding and membrane potential in the uptake of
guanidinium-rich transporters into cells. Adv. Drug Delivery Rev. 2005,
57, 495−504.
(20) Pantos, A.; Tsogas, I.; Paleos, C. M. Guanidinium group: a
versatile moiety inducing transport and multicompartmentalization in
complementary membranes. Biochim. Biophys. Acta, Biomembr. 2008,
1778, 811−823.
(21) Su, Y.; Li, S.; Hong, M. Cationic membrane peptides: atomic-
level insight of structure-activity relationships from solid-state NMR.
Amino Acids 2013, 44, 821−833.
(22) Gurtovenko, A. A.; Vattulainen, I. Ion Leakage through
Transient Water Pores in Protein-Free Lipid Membranes Driven by
Transmembrane Ionic Charge Imbalance. Biophys. J. 2007, 92, 1878−
1890.
(23) Gurtovenko, A.; Vattulainen, I. In Biomembrane Frontiers,
Collective Dynamics in Lipid Membranes: From Pore Formation to Flip-
Flops; Jue, T., Risbud, S., Longo, M., Faller, R., Eds.; Humana Press:
New York, 2009; Chapter 5, pp 121−139.
(24) Fuertes, G.; Giménez, D.; Esteban-Martín, S.; Sánchez-Munoz,
O. L.; Salgado, J. A lipocentric view of peptide-induced pores. Eur.
Biophys. J. 2011, 40, 399−415.
6064
Article
(25) Sun, D.; Forsman, J.; Woodward, C. E. Atomistic molecular
simulations suggest a kinetic model for membrane translocation by
arginine-rich peptides. J. Phys. Chem. B 2015, 119, 14413−14420.
(26) Miteva, M.; Andersson, M.; Karshikoff, A.; Otting, G. Molecular
electroporation: a unifying concept for the description of membrane
pore formation by antibacterial peptides, exemplified with NK-lysin.
FEBS Lett. 1999, 462, 155−158.
(27) Breton, M.; Delemotte, L.; Silve, A.; Mir, L. M.; Tarek, M.
Transport of siRNA through lipid membranes driven by nanosecond
electric pulses: an experimental and computational study. J. Am. Chem.
Soc. 2012, 134, 13938−13941.
(28) Salomone, F.; Breton, M.; Leray, I.; Cardarelli, F.; Boccardi, C.;
Bonhenry, D.; Tarek, M.; Mir, L. M.; Beltram, F. High-yield nontoxic
gene transfer through conjugation of the CM18-Tat11 chimeric
peptide with nanosecond electric pulses. Mol. Pharmaceutics 2014, 11,
2466−2474.
(29) Casciola, M.; Tarek, M. A molecular insight into the electro-
transfer of small molecules through electropores driven by electric
fields. Biochim. Biophys. Acta, Biomembr. 2016, 1858, 2278−2289.
(30) Missner, A.; Pohl, P. 110 years of the Meyer−Overton rule:
predicting membrane permeability of gases and other small
compounds. ChemPhysChem 2009, 10, 1405−1414.
(31) Gurtovenko, A. A.; Anwar, J.; Vattulainen, I. Defect-mediated
trafficking across cell membranes: insights from in silico modeling.
Chem. Rev. 2010, 110, 6077−6103.
(32) Leontiadou, H.; Mark, A. E.; Marrink, S. J. Antimicrobial
peptides in action. J. Am. Chem. Soc. 2006, 128, 12156−12161.
(33) Herce, H. D.; Garcia, A. E. Molecular dynamics simulations
suggest a mechanism for translocation of the HIV-1 TAT peptide
across lipid membranes. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,
20805−20810.
(34) Sengupta, D.; Leontiadou, H.; Mark, A. E.; Marrink, S.-J.
Toroidal pores formed by antimicrobial peptides show significant
disorder. Biochim. Biophys. Acta, Biomembr. 2008, 1778, 2308−2317.
(35) Bennett, W. F. D.; Hong, C. K.; Wang, Y.; Tieleman, D. P.
Antimicrobial peptide simulations and the influence of force field on
the free energy for pore formation in lipid bilayers. J. Chem. Theory
Comput. 2016, 12, 4524−4533.
(36) Huang, H. W. Action of antimicrobial peptides: two-state
model. Biochemistry 2000, 39, 8347−8352.
(37) Yamaguchi, S.; Hong, T.; Waring, A.; Lehrer, R. I.; Hong, M.
Solid-State NMR Investigations of Peptide- Lipid Interaction and
Orientation of a β-Sheet Antimicrobial Peptide, Protegrin. Biochemistry
2002, 41, 9852−9862.
(38) Reigada, R. Electroporation of heterogeneous lipid membranes.
Biochim. Biophys. Acta, Biomembr. 2014, 1838, 814−821.
(39) Mitchell, D. J.; Steinman, L.; Kim, D. T.; Fathman, C. G.;
Rothbard, J. B. Polyarginine enters cells more efficiently than other
polycationic homopolymers. J. Pept. Res. 2000, 56, 318−325.
(40) Schmidt, N. W.; Tai, K. P.; Kamdar, K.; Mishra, A.; Lai, G. H.;
Zhao, K.; Ouellette, A. J.; Wong, G. C. L. Arginine in α-defensins:
differential effects on bactericidal activity correspond to geometry of
membrane curvature generation and peptide-lipid phase behavior. J.
Biol. Chem. 2012, 287, 21866−21872.
(41) Gurtovenko, A. A.; Vattulainen, I. Molecular mechanism for
lipid flip-flops. J. Phys. Chem. B 2007, 111, 13554−13559.
(42) Tieleman, D. P.; Marrink, S.-J. Lipids out of equilibrium:
Energetics of desorption and pore mediated flip-flop. J. Am. Chem. Soc.
2006, 128, 12462−12467.
(43) Henderson, J. M.; Waring, A. J.; Separovic, F.; Lee, K. Y. C.
Antimicrobial peptides share a common interaction driven by
membrane line tension reduction. Biophys. J. 2016, 111, 2176−2189.
(44) Jo, S.; Kim, T.; Iyer, V. G.; Im, W. CHARMM-GUI: a web-
based graphical user interface for CHARMM. J. Comput. Chem. 2008,
29, 1859−1865.
(45) Brooks, B. R.; Brooks, C. L.; MacKerell, A. D.; Nilsson, L.;
Petrella, R. J.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.; Boresch, S.
CHARMM: the biomolecular simulation program. J. Comput. Chem.
2009, 30, 1545−1614.
DOI: 10.1021/acsomega.8b00483
ACS Omega 2018, 3, 6056−6065
ACS Omega Article

(46) Wu, E. L.; Cheng, X.; Jo, S.; Rui, H.; Song, K. C.; Dávila-
Contreras, E. M.; Qi, Y.; Lee, J.; Monje-Galvan, V.; Venable, R. M.
CHARMM-GUI Membrane Builder toward realistic biological
membrane simulations. J. Comput. Chem. 2014, 35, 1997−2004.
(47) Lee, J.; Cheng, X.; Swails, J. M.; Yeom, M. S.; Eastman, P. K.;
Lemkul, J. A.; Wei, S.; Buckner, J.; Jeong, J. C.; Qi, Y. CHARMM-GUI
Input Generator for NAMD, GROMACS, AMBER, OpenMM, and
CHARMM/OpenMM Simulations Using the CHARMM36 Additive
Force Field. J. Chem. Theory Comput. 2015, 12, 405−413.
(48) Murzyn, K.; Róg, T.; Pasenkiewicz-Gierula, M. Phosphatidyle-
thanolamine phosphatidylglycerol bilayer as a model of the inner
bacterial membrane. Biophys. J. 2005, 88, 1091−1103.
(49) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.
W.; Klein, M. L. Comparison of simple potential functions for
simulating liquid water. J. Chem. Phys. 1983, 79, 926−935.
(50) Sachs, J. N.; Crozier, P. S.; Woolf, T. B. Atomistic simulations of
biologically realistic transmembrane potential gradients. J. Chem. Phys.
2004, 121, 10847−10851.
(51) Kutzner, C.; Grubmüller, H.; de Groot, B. L.; Zachariae, U.
Computational electrophysiology: the molecular dynamics of ion
channel permeation and selectivity in atomistic detail. Biophys. J. 2011,
101, 809−817.
(52) Hub, J. S.; de Groot, B. L.; van der Spoel, D. g_wham - A Free
Weighted Histogram Analysis Implementation Including Robust Error
and Autocorrelation Estimates. J. Chem. Theory Comput. 2010, 6,
3713−3720.
(53) Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.;
Hess, B.; Lindahl, E. GROMACS: High performance molecular
simulations through multi-level parallelism from laptops to super-
computers. SoftwareX 2015, 1, 19−25.
(54) Hoover, W. G.; Ladd, A. J. C.; Moran, B. High-strain-rate plastic
flow studied via nonequilibrium molecular dynamics. Phys. Rev. Lett.
1982, 48, 1818.
(55) Nosé, S. A molecular dynamics method for simulations in the
canonical ensemble. Mol. Phys. 1984, 52, 255−268.
(56) Parrinello, M.; Rahman, A. Polymorphic transitions in single
crystals: A new molecular dynamics method. J. Appl. Phys. 1981, 52,
7182−7190.
(57) Klauda, J. B.; Venable, R. M.; Freites, J. A.; O’Connor, J. W.;
Tobias, D. J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A. D.,
Jr.; Pastor, R. W. Update of the CHARMM all-atom additive force
field for lipids: validation on six lipid types. J. Phys. Chem. B 2010, 114,
7830−7843.
(58) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.;
Pedersen, L. G. A smooth particle mesh Ewald method. J. Chem. Phys.
1995, 103, 8577−8593.
(59) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M.
LINCS: a linear constraint solver for molecular simulations. J. Comput.
Chem. 1997, 18, 1463−1472.

6065 DOI: 10.1021/acsomega.8b00483

ACS Omega 2018, 3, 6056−6065