Molecular Insights into the Adsorption Mechanism of Human β

Feb 2, 2016 - In this work, we carried out molecular dynamics (MD) simulations to explore its adsorption mechanism on, and the structural and ...
1 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/Langmuir

Molecular Insights into the Adsorption Mechanism of Human β‑Defensin‑3 on Bacterial Membranes Juho Lee,† Sang Won Jung,† and Art E. Cho* Department of Bioinformatics, Korea University, Sejong 02841, Korea S Supporting Information *

ABSTRACT: Human β-defensin-3 (hBD3) is an endogenous antimicrobial peptide that exhibits broad-spectrum antibacterial activity without eukaryotic cytotoxicity. In this work, we carried out molecular dynamics (MD) simulations to explore its adsorption mechanism on, and the structural and thermodynamic contributions of individual residues to its antibacterial activity with both Gram-negative (GN) and Gram-positive (GP) bacterial membrane. Due to the strong electrostatic interaction of hBD3 with POPG lipids, which are more prevalent on the GP membrane, its adhesion to the GP membrane is stronger than to the GN membrane and stabilized more rapidly. On the surface of both bacterial membranes, the orientation of hBD3 is dominated by an electric dipole. We next analyzed the binding free energy decompositions of the hBD3-membrane complex using the molecular mechanics Poisson−Boltzmann surface area (MM-PBSA) method. The results of both the GN and the GP membrane simulations show that Arg17, Arg36, and Arg38 form both polar and nonpolar interactions and are potentially the key residues for hBD3 antibacterial activity. On the other hand, there was a significant difference in the energy contribution of Arg12 between the GP and GN membrane simulations, suggesting that Arg12 is a key factor in the toxicity of hBD3 to specifically GP bacteria. Our findings shed light on the antibacterial activity of hBD3 on bacterial membranes and yield insights useful for the design of potent antimicrobial peptides targeting multidrug resistant bacteria.

1. INTRODUCTION The treatment of infectious diseases caused by multidrugresistant (MDR) bacteria is a growing challenge worldwide as conventional antibiotics become increasingly ineffective. Conventional antibiotics are restricted to a limited number of specific targets and their mechanisms are subject to severe and growing resistance problems.1,2 Efforts to overcome the resistance problem of conventional antibiotics have highlighted antimicrobial peptides (AMPs) as potential alternatives. A part of the innate immune system, AMPs are host defense peptides that exhibit broad spectrum antimicrobial activity against pathogenic bacteria without incurring antimicrobial resistance.3−5 This is because their mechanism of action is through their selective interference with the integrity of target cell membrane lipids rather than through their binding to specific protein receptors.6 In other words, to develop resistance, bacteria would have to redesign their membrane lipid composition and topology, and that is a “costly” solution. Of the numerous types of AMP, human β defensins (hBDs), a defensin subfamily, are particularly interesting due to their extensive presence in human tissues.7 These hBDs share common structural features such as having less than 45 amino acids, conserved cysteine residues, and the formation of three disulfide bridges in the arrangement Cys1-Cys5, Cys2-Cys4, and Cys3-Cys6. Further understanding their core structural characteristics and modifications is considered an important step toward the pharmaceutical application of AMPs. As a member of the hBD family, human β defensin-3 (hBD3) is known to exhibit many interesting features. hBD3 shares low © 2016 American Chemical Society

sequence similarity with other hBDs and possesses an unusually high positive charge (+11) with most of the charged residues arranged near the C-terminal region.8 Interactions between the residues on the second β-strands of two monomers constitute the basis for the formation of an amphipathic symmetrical dimer, which exhibits a relatively more positively charged surface. These structural features are known to make hBD3 preferentially interact with bacterial membranes, which are generally composed of negatively charged lipids, and result in a broad spectrum of salt-insensitive bactericidal activities against not only gram−negative but also Gram-positive bacteria. Much research has tried to understand the principles underlying the exceptional antibacterial activities of hBD3. Wu et al. (2003) investigated the influence of the formation of the three hBD3 disulfide bridges on bactericidal activity by shuffling their connectivity.9 Although the chemotacticmediated activity of hBD3 is affected by changes in its disulfide bonding, its antimicrobial activity is not. Another approach to investigate the relationship between hBD3 structure and activity has been structural modifications, such as the mutation of hBD310,11 or its fragmentation.12 Although the dimerization of hBD3 is a characteristic feature and is important in the immune system, it has proven to be an insignificant factor for hBD3’s antimicrobial activity.13 Instead, what affects hBD3’s intrinsic antimicrobial activity most is the residual contributions Received: November 8, 2015 Revised: January 10, 2016 Published: February 2, 2016 1782

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir of the monomer unit.13,14 A key to antimicrobial activity is the distribution of positively charged and hydrophobic residues in its monomeric structure.11 Despite the significant progress such research has made, experimental works exploring the contributions of specific residues to the antibacterial mechanism of hBD3 are limited. Here, we demonstrate the molecular mechanism of adsorption and the contributions of specific residues to the antimicrobial activity of monomeric hBD3 using Gram-negative (GN), Gram-positive (GP), and mammalian membrane models, through a series of all-atom molecular dynamics (MD) simulations and the molecular mechanics Poisson− Boltzmann surface area (MM-PBSA) method. MM-PBSA can successfully calculate an approximate binding free energy that includes the effects of thermal averaging with a force field and continuum solvent models. First, to determine a preferential adsorption orientation, the orientation of monomer hBD3 adsorbed on membranes was studied based on different initial configurations. Its electric and hydrophobic dipole moments were described for further understanding of its orientation. Moreover, we investigated the effect of hBD3 on membranes by measuring their thickness and acyl chain bond orientation order. The interactions between the peptide and the membranes were then studied. A decomposition of the binding free energy on a per-residue basis was determined using the MM-PBSA method. Our findings will elucidate some of the principles of the antimicrobial activity of hBD3 on both GN and GP membranes and give insights that will help the design of potentially potent antimicrobial peptides targeting multidrug resistant bacteria.

Figure 1. Atomic structure of hBD3 represented as a new cartoon model and its one-letter sequence. The α-helix is in purple and the βsheet is in yellow. The positively and negatively charged residues are highlighted with blue and red, respectively. conditions were applied in all directions, and the cutoff of shortrange van der Waals interactions was 1.2 nm. The particle mesh Ewald method32 was applied to treat long-range electrostatic interactions, with a 1.2 nm real-space contribution cutoff for Coulombic interactions. The covalent bonds between hydrogen atoms and any heavy atoms were constrained with the LINCS algorithm,33 allowing an integration time step of 2 fs. A temperature of 323 K°, higher than the phase transition temperatures of the POPE/POPG and POPC bilayers, and a pressure of 1 atm was maintained by the Nosé−Hoover thermostat34,35 and semi-isotropic Parrinello−Rahman barostat36 respectively. In all systems, the protonation states of peptides were assigned based on calculations at pH 7 and with 50 mM NaCl, as used in the experiments. The systems were equilibrated for 300 ps under a gradual increase in temperature from 10.0 to 323 K after energy minimization with the steepest descent method and then further equilibrated for 300 ps under an isothermal−isobaric (NPT) ensemble with the Berendsen weak coupling method.37 During the series of equilibration steps, z-axis position restraints were placed on lipid atoms to restrict their motion to the x−y plane. After the equilibration process, all simulations were performed for 300 ns under the NPT ensemble without any restraints. All the simulation systems, with their details and run times, are listed in Table 1. 2.3. Orientation Angle. A previous study has characterized the protein orientation on membrane surfaces based on the orientation angle between the surface normal vector and hydrophobic (θ) or electric (ψ) dipole moments of the protein.38 We also demonstrate the orientation angles here to compare in detail the effects of different types of membranes on hBD3 orientations. The hydrophobic dipole moment of the protein, Hm, is defined as

2. MATERIALS AND METHODS 2.1. System Construction. We modeled different membrane types, including those of bacterial and mammalian cells. Two types of bacterial membranes, GN and GP, are prepared based on a mixture of palmitoyloleoylphosphatidylglycerol (POPG) and palmitoyloleoylphosphatidylethanolamine (POPE). In bacteria, these lipids are the major anionic and zwitterionic species. To model the membranes of GN bacteria, such as Escherichia coli, we simulated a 1:3 ratio of POPG/POPE15,16 and to model the membranes of GP bacteria, such as Bacillus subtilis (Dowhan, 1997; Denning and Beckstein, 2013),15,16 we simulated a 3:1 ratio of POPG/POPE.16,17 For mammalian membranes, of which the lipid composition is simpler, zwitterionic POPC lipids were used.18 The membrane models were generated using the input generator from the Web site of CHARMM-GUI (http://www.charmm-gui.org/).19−22 The coordinates of hBD3 were taken from PDB entry 1KJ6 (Figure 1).14 In order to define appropriate systems, simulations were carried out with various initial poses of hBD3 on the membrane surfaces. Although the initial configurations are important, previous computational studies have not considered them in detail.23 To obtain proper samplings, three different initial configurations of hBD3−membrane complexes were prepared: Configuration A is initially oriented based on the vector from the C-termini to the N-termini, which is perpendicular to the surface normal (Figure 2a). The structures of the proteins in Figure 2b and Figure 2c are obtained by rotating the initial configuration A 90° and 180° around y axis, respectively. In all cases, the hBD3 molecule is placed such that the minimum distance between its atoms and the membrane’s atoms is over 15 Å. 2.2. Molecular Dynamics Simulation. All simulations were performed using the GROMACS 4.6.5 package 24 with the CHARMM36 force field25−28 without cMAP correction, which gives properties of both AMPs and various lipids with good agreement with experimental data.28,29 The TIP3P water model was used to generate explicit solvation conditions30 and Newton’s equations of motion were integrated using the leapfrog algorithm.31 Periodic boundary

N

Hm =

∑ Hisi i=1

(1)

in which Hi is the hydrophobicity of residue i, N is the total number of residues, and si is a unit vector pointing from the α-carbon atom of residue i to the center of mass (COM) of the residue’s side chain.39 In this work, for the hydrophobicity scales the normalized “consensus” scale40 was used. The electric dipole moment of the protein, μ, is calculated using g_dipoles, as follows N

μ=

∑ qixi i=1

(2)

where qi is the partial charge of atom i and xi is the position of atom i from the COM of the molecule. The hydrophobic (θ) and electric (ψ) orientation angles are illustrated in Figure 2a. 2.4. Binding Free Energy Calculation. In order to investigate the binding energy of hBD3 based on the types of bacterial membranes, the molecular mechanics Poisson−Boltzmann surface area (MM-PBSA) method was used using the g_mmpbsa code 1783

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

Figure 2. An illustration of orientation angles within HBD-3 with respect to surface normal (n⃗ ). (a) The black arrow (h⃗) and red arrow (e⃗) indicate the direction of the hydrophobic and electric dipole moments, respectively. The structures of the proteins are obtained by rotating the initial configuration A 90° around the y axis (b) and 180° around the y axis (c), respectively.

Table 1. Table of All Simulations with Detailsa membrane

lipid composition

model

length (ns)

Gram-negative

75% PE, 25% PG 75% PE, 25% PG 75% PE, 25% PG 75% PE, 25% PG 25% PE, 75%PG 25% PE, 75%PG 25% PE, 75%PG 25% PE, 75%PG 100% PC

control A B C control A B C A

200 300 300 500 200 300 300 500 300

Gram-positive

mammalian a

set to 0.05 M, the salt concentration used for the simulation conditions. To calculate the nonpolar solvation free energy, Gnps, the following equation is used (eq 5):

Gnps = γ SASA + b

where SASA is the abbreviation of the solvent accessible surface area calculated using a water probe radius of 1.4 Å. The constants γ and b were set to 0.00542 kcal/mol/Å2 and 0.92 kcal/mol, respectively.44

3. RESULTS AND DISCUSSION 3.1. Selectivity for Bacterial Membrane. In general, hBD3 has broad antibacterial activity with both GN and GP bacteria without eukaryotic cytotoxicity and the primary path of its antibacterial activity is adsorption on the bacterial membrane.8 To address the membrane selectivity of hBD3, we investigated how its adsorption on the surface of mammalian, GN, and GP membranes differed based on various initial configurations. In order to present the adsorption, the distances between the COM of the hBD3 molecule and the COM of the membranes were measured as a function of time. The results are shown in Figure 3. In the mammalian simulation (Figure 3c), the hBD3 did not bind stably to the membrane, and there was a relatively wide fluctuation in the distance between the protein and the membrane. This is because the zwitterionic POPC lipids, which are characterized by a positively charged amino group at their outer head regions, disturb the interaction with the cationic hBD3, consistent with previous studies.23 On the other hand, the hBD3 associates with both GP and GN bacterial membranes regardless of initial configuration (Figure 3a,b). However, the hBD3 molecule is slightly closer to the center of the GP membrane than to the center of the GN membrane. The difference may be due to either varying effects of the hBD3 on the membrane or a variation in membrane thickness caused by differing membrane lipid compositions. To clarify its effect, membrane thicknesses and adsorption depth of hBD3 were measured (Table 2). The membrane thickness measured was the distance between the phosphates of the top and the bottom leaflets. The adsorption depth indicates the adsorption position of hBD3 with respect to the membrane surface by calculating the gap between the phosphate in the top leaflet and the hBD3 molecule. The results show that the equilibrated GN membrane is thicker than the equilibrated GP membrane by about 2.3 Å. Also, the distance from the hBD3 to the top phosphate was found to be about 0.63 Å for GN and 0.3 Å for GP without considering the variation of thickness

All systems include 120 lipids.

distributed by Kumari.41 Although there are more rigorous methods of calculating binding energies, such as free energy perturbation42 and thermodynamic integration,43 they are of limited use for larger systems due to their expensive computational cost. For each hBD3-membrane complex MD simulation, 100 snapshots were extracted from the last 20 ns trajectory. The binding free energy of the hBD3-membrane complex in solvent was estimated following eq 2

ΔG bind = Gcomplex − (Gpeptide + Gmembrane)

(3)

where, Gcomplex is the free energy of the hBD3-membrane complex and Gpeptide and Gmembrane are the free energies of isolated hBD3 or membrane, respectively, in solvent. Since the main purpose of the MM-PBSA analysis is to identify the contribution of each residue to the binding free energies, this study uses the relative binding free energy, which neglects the entropy contributions of the peptide (eq 3):

G = ⟨EMM⟩ + ⟨Gsol ⟩

(4)

where ⟨EMM⟩ is the average molecular mechanics potential energy using a classical molecular force field (CHARMM 36) in a vacuum, and ⟨Gsol⟩ is the free energy of solvation. EMM is the sum of both the bonded and nonbonded interactions. Due to the single trajectory approach, the bonded interaction is taken to be zero. The nonbonded interactions including both electrostatic and van der Waals interactions are calculated using Coulomb and Lennard-Jones (LJ) potential functions, respectively. The solvation free energy, Gsol, is composed of two terms (eq 4). Gsol = Gps + Gnps

(6)

(5)

Gps is the electrostatic term of the solvation free energy estimated by solving the nonlinear Poisson−Boltzmann (PB) equation. The values of the vacuum (vdie), solute (pdie) and solvent (sdie) dielectric constants were set to 1, 7, and 80 respectively. The ionic strength was 1784

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

toward the membrane, for 0 it is parallel to the surface, and for 1 it is parallel to the surface normal, that is, pointing out from the membrane. On the surface of the GN bacterial membrane, orientations of hBD3 varied based on the initial orientation. For configuration A, electric and hydrophobic dipole moments fluctuated back and forth during the whole simulation, meaning that the orientation of hBD3 on the surface was unstable (Figure 4a). For configuration B, although two dipole moments divided into opposite directions, the distributions of the dipoles overlapped even after 280 ns of large fluctuation (Figure 4b). With configuration C (Figure 4c), unlike other initial configurations, both the hydrophobic moment and especially the electric dipole moment were highly stabilized and, in the GN membrane simulation, each dipole moment had a distinct distribution. On the other hand, in the GP membrane simulation, similar binding orientations of the hBD3 on the membrane surface were observed regardless of initial orientation. The electric and hydrophobic moments of hBD3 on the surface of the GP membrane are clearly presented as being in opposite directions with cosine values of about −0.9 and 0.7 respectively, with similar results for the configuration C in the GN membrane simulation (Figure 4c and 5). This specific orientation reflects the hydrophilic region of hBD3 binding mostly to the surface. In contrast, the hydrophobic moment is pointing outward from the membrane, indicating that the hydrophobic region of hBD3, known to be the N-terminus region, does not contribute as much to its adsorption as the hydrophilic one. Our simulation results showed that the binding of hBD3 to the GP membrane is stabilized rapidly regardless of its initial orientations due to strong electrostatic interaction caused by the high concentration of negatively charged POPG lipids in the membrane as compared with the GN membrane. On the other hand, three different binding modes of hBD3-GN membrane complex were resulted from the simulation depending on different initial configurations. Binding energies of these binding modes were very close to one another ranging within only 2.6 kcal/mol (−290.7 kcal/mol, − 292.6 kcal/mol and −293.3 kcal/mol for configurations A, B, and C, respectively). We conclude from these findings that, due to sparseness of POPG lipids, the binding of hBD3 to the GN membrane is weak and unstable. 3.3. Disruption of Bacterial Membrane Induced by hBD3 Monomer. For further description of the activity of hBD3 on bacterial membranes, the average deuterium order parameter, SCD, was calculated using the GROMACS analysis tool g_order. Here, we present the order parameter of sn-1 saturated acyl chains of each membrane leaflet as a function of carbon number. These parameters are calculated by the following equation:

Figure 3. Distance of hBD3 from Gram-negative (a), Gram-positive (b), and mammalian membrane (c) along the bilayer normal (z-axis) as a function of time. The distance was measured between the COM of each molecule. The letters A, B, and C represent the initial orientations described as in Figure 2.

Table 2. Summary of Membrane Thickness and Adsorption Depth membrane

model

membrane thicknessa

adsorption depthb

Gram-negative

control A B C control A B C

4.05 4.03 3.92 4.08 3.83 3.73 3.79 3.82

N/A 0.34 0.82 0.72 N/A 0.40 0.23 0.27

Gram-positive

All units are in nanometers. aMembrane thickness indicates the distance of phosphates between the top and bottom leaflets. b Adsorption depth was measured between the phosphates and hBD3.

induced by hBD3. In short, hBD3 binds to the GP membrane more deeply than to the GN membrane. Our findings suggest that this difference is due to the GP membrane having higher concentrations of POPG lipids which promote a stronger interaction with hBD3. 3.2. Adsorption Orientation Based on Types of Bacterial Membrane. The orientation of hBD3 on the surface of the membrane is known to be critical to preservation and maximization of its functionality.38,45 Considering the amphiphilicity of hBD3, understanding its orientation in terms of both its polar and nonpolar characteristics is required. Here, we describe the orientation of hBD3 by presenting the variation of the cosine of its electrical and hydrophobic dipole orientation angles (Figure 2a). The importance of both of these dipole moments in determining probable protein orientations on charged or hydrophobic surfaces has been highlighted in several studies.38,46−48 For a cosine value of −1, the dipole is antiparallel to the surface normal, that is, pointing

SCD =

3 cos2 θ − 1 2

(7)

where θ is the angle between the C−D bond vector and the bilayer normal. The brackets indicate an average over all equivalent lipid molecules along the MD simulation trajectory. The average data were collected over the last 50 ns trajectories. The smaller value of SCD in the hBD3-membrance complex than the pure membrane system indicates that the peptide disturbed the order of acyl chains. In the case of GN membrane, the disruption induced by the hBD3 was observed between acyl chain carbon numbers from 3 to 10 in only 1785

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

Figure 4. Orientation of the electric dipole (blue) and hydrophobic dipole (red) of hBD3 on the surface of the Gram-negative model membrane. The final snapshots of simulations with pose A (a), B (b), and C (c) are presented. The orange lipid is POPG and the blue is POPE. The orientation distributions were taken only from 280 to 300 ns trajectories.

Figure 5. Orientation of the electric dipole (blue) and hydrophobic dipole (red) of hBD3 on the surface of the Gram-positive model membrane. The final snapshots of simulations with pose A (a), B (b), and C (c) are presented. The orange lipid is POPG and the blue is POPE. The orientation distributions were taken only from 280 to 300 ns trajectories.

configuration C (Figure S1a). This result supports our finding in which the configuration C was determined to be the most stabilized orientation of hBD3 for GN membrane. Similar results of membrane disruption were seen for the hBD3−GP complex as demonstrated by Figure S1b. Although the overall

disruptions of the membrane induced by the hBD3 were observed, there was no clear influence of hBD3 on either GN or GP membrane thickness, of which the variations were only ±0.08 and ±0.05 nm, respectively (Table 2). Using coarsegrained (CG) MD,23 Zhao et al. reported that more than 12 1786

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

suggested in the literature, 50 as those whose energy contribution value is more than 2.5 kcal/mol. First, we analyzed the per-residue polar energy contributions. Figure 6 shows that effective polar interaction contributions

hBD3 monomers are required to affect membrane disruption, explaining why the membrane thinning effect was not clearly observed with a single hBD3 monomer. 3.4. Binding Free Energy Analysis of the hBD3− Bacterial Membrane Complex. To provide further insight into the interactions between hBD3 and the bacterial membranes, the binding free energies of the hBD3−membrane complex were calculated using the MM-PBSA method. Using configuration C, which is determined to be the most stabilized orientation of hBD3 for both GN and GP membranes in our simulation, we calculate the binding free energy of the hBD3− membrane complex to identify detailed features of the interaction. Here, a dielectric constant of 7 was used since it is the average value of the phospholipid head groups,49 to which hBD3 predominantly binds (Table 3). Table 3. Relative Binding Free Energy Components of the hBD3−Membrane Complex energetic components

hBD3-GN complex

hBD3-GP complex

ΔGvdw ΔGnps ΔGelec ΔGps ΔGnonpolara ΔGpolarb ΔGbind

−81.0 −11.8 −654.1 453.6 −92.8 −200.5 −293.3

−132.4 −17.8 −1586.9 614.8 −150.2 −972.1 −1122.4

All energy units are in kcal/mol. aΔGnonpolar = ΔGvdw + ΔGnps, indicating nonpolar energy. bΔGpolar = ΔGelec + ΔGps, indicating polar energy.

Figure 6. Polar binding energy contribution of each residue of hBD3 in the hBD3−GN complex (a), and hBD3-GP complex (b). The residues with the most favorable (< −2.5 kcal/mol) contributions and unfavorable (>2.5 kcal/mol) are labeled.

Table 3 presents the free energy components of the hBD3− GN and −GP complexes. The polar interaction energy (ΔGpolar) is defined as the sum of the difference of electrostatic interaction energy (ΔGelec) and the difference of polar solvation energy (ΔGps). The nonpolar interaction energy (ΔGnonpolar) is defined as the sum of the difference of vdW interaction energy (ΔGvdw) and the difference of nonpolar solvation energy (ΔGnps). The binding free energy is defined as the sum of ΔGpolar and ΔGnonpolar. The results show that both ΔGpolar and ΔGnonpolar favor the adsorption of hBD3 on the surfaces of both bacterial membranes even if ΔGpolar (−200.5 kcal/mol with GN, and −972.1 kcal/mol with GP) is much stronger than ΔGnonpolar (−92.8 kcal/mol with GN, and −150.2 kcal/mol with GP), suggesting that the polar interactions are a key factor in the formation of both complexes. It also shows that the binding energy of the hBD3−GP complex obtained was over 3 times larger than that of the energy for the hBD3−GN complex. This difference derives from the polar interactions, especially the electrostatic ones (93.1%). This is due to the different lipid compositions of the GN and GP bacterial model membranes, of which the net charges are −30 and −90, respectively. Assuming that the strength of the interactions between hBD3 and the membrane is correlated to hBD3’s antibacterial activities, our finding is in accordance with the experimental observation of stronger antibacterial effects of hBD3 against GP bacteria than GN bacteria. We next carried out the free energy decomposition of each residue in hBD3 to identify the residues critical for the interaction of the hBD3−membrane complexes. In order to describe the free energy decomposition in detail, the energy contributions of the polar and the nonpolar energies were analyzed separately. The important residues are defined, as

were obtained from the majority of the charged residues in both complexes. The energy contributions of the negatively charged residues, Glu27 and Glu28 were found to be unfavorable; whereas those of the positively charged residues, including the N-terminus residue, Gly1, were found to be favorable. However, the interaction contributions of Lys8, Arg12, Lys26, Arg36, and Arg38 were lower in the GN membrane simulation (5.3%, 4.7%, 2.1%, 3.1%, and 7.6%, respectively) (Figure 6a) than the GP membrane simulation (9.7%, 9.6%, 9.1%, 8.3%, and 8.2%, respectively) (Figure 6b). Of particular note was the interaction energy of Lys26 in the GN membrane simulation, since it was much lower than 2.5 kcal/mol. These results clearly demonstrate why the truncation of the N-terminus and Cterminus, where most of the positively charged residues are grouped, influences the antibacterial activity of hBD3 against GN bacteria more than against GP bacteria.12 Finally, the nonpolar energy contributions were identified in Figure 7. Although the polar interaction has been identified as key for the interaction in the hBD3−membrane complex, the importance of a nonpolar interaction related to hydrophobicity in selectivity has also been reported.11 In the GN membrane simulation (Figure 7a), the effective nonpolar interactions were provided by the Cys11, Arg17, Val20, Leu24, Arg36 and Arg38 residues. In the case of the GP membrane simulation (Figure 7b), the 13 residues, Arg12, Val13, Arg14, Arg17, Cys18, Val20, Leu21, Leu24, Pro25, Arg36, Arg38, Arg43, and Lys44 had favorable nonpolar interactions. Arg17, Val20, Leu24, Arg36, and Arg38 have effective contributions in both complexes. Moreover, Arg17, Arg36, and Arg38 provide both polar and nonpolar interactions, indicating that these residues are possibly involved in the antibacterial activity of hBD3. On the other 1787

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

Initially, we characterized hBD3 adsorption on the mammalian and bacterial membrane in terms of its displacement from the center of the membrane. In the simulations, although hBD3 never stably adsorbed with the mammalian membrane, it clearly associated with both GN and GP bacterial membranes. The distances of hBD3 from the centers of the GN and the GP membranes differed since hBD3 binds to the GP membrane more deeply than to the GN membrane. This is caused by the strong interaction between hBD3 and POPG lipids, which are more prevalent in the GP membrane. Next, the adsorption orientation of hBD3 on the bacterial membranes was investigated. Here, in order to describe the influence of its amphiphilicity on the orientation in detail, we considered both electric and hydrophobic dipole moments and measured their orientation angles with respect to the membrane surface normal. A stabilized adsorption of hBD3 was only obtained from configuration C in the GN membrane simulation, with the electric moment pointing toward the membrane. This result suggests that the high concentration of POPG lipids in the GP membrane leads to strong electrostatic interaction with hBD3 and stabilize its adsorption rapidly as compared with hBD3-GN membrane complex. For further investigation relative to the influence of hBD3 on the bacterial membrane, order parameter of lipid acyl chain was calculated. The disruption of carbon numbers from 3 to 10 was observed; however, significant disruption to membrane thickness was not observed due to insufficient number of hBD3 peptides. Finally, the binding free energies of the hBD3−membrane complex were calculated using the MM-PBSA method. The binding free energy of the hBD3-GP complex is over 3 times higher than that of the hBD3−GN complex. This difference was mostly derived from the electrostatic interactions (93.1%). Arg17, Arg36, and Arg38 provide not only strong polar but also strong nonpolar interactions, indicating that these residues are central to the antibacterial activity of hBD3. On the other hand, the contribution of Arg12 differed markedly between the GN and the GP membrane simulations, suggesting that it is a key factor in the antibacterial activity of hBD3 against GP bacteria. Our findings shed light on the structural and thermodynamic bases of the antibacterial activity of hBD3 on both GN and GP membranes and provide insights useful for the design of potent antimicrobial peptides that can be used to target multidrug resistant bacteria.

Figure 7. Nonpolar binding energy contribution of each residue of hBD3 in the hBD3−GN complex (a), and hBD3-GP complex (b). The residues with the most favorable (< −2.5 kcal/mol) contributions are labeled.

hand, the contribution of Arg12 is distinctly different for GN and GP. As shown in Figure 6 and 7, both the polar and nonpolar contribution of Arg12 is markedly depressed in the GN membrane simulation compared with the GP membrane simulation. Our finding suggests that the Arg12 residue is a key factor determining the antibacterial activity of hBD3 against GP bacteria as opposed to GN bacteria. Although our findings show significant accordance with previous experimental studies, understanding of the interaction mechanism of hBD3 with the membranes is limited to its preserved wild-type conformation. It has been experimentally shown that loss of the structural characteristics of hBD3 via an engineering process would impede its antibacterial properties. The work of Hoover et al. showed that a peptide with sequence similar to that of hBD3 from residues 8 to 26 did not exhibit antibacterial activity on Gram-positive bacteria even though Arg12 is included in the peptide.51 Another short engineered hBD3 peptide with identical sequence from residues 6 to 22 also did not show antibacterial activities.11 This apparent contradiction to our claim of particular residues being a key factor can be explained by noting that the conformational uniqueness of hBD3 could also be an important influence factor for its antibacterial activities. Interactions of the key residues could be unrealized in different conformations of hBD3.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.5b04113.

■

4. CONCLUSION AND SUMMARY In summary, MD simulation was carried out to explore the adsorption mechanism of hBD3 on both GN and GP bacterial membranes due to its role in the antibacterial activity of hBD3, and to differentiate the contributions of each of its residues to the binding free energy of the adsorption interactions. Through comparing membrane simulations with different initial hBD3 orientations, the adsorption of hBD3 was described in terms of structural and thermodynamic properties.

Order parameters of acyl chains for hBD3 on gramnegative membrane and gram-positive membrane (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions †

These authors contributed equally to this work.

Notes

The authors declare no competing financial interest. 1788

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir

■

enhances permeability. Biochim. Biophys. Acta, Biomembr. 2013, 1828, 1112−21. (19) Jo, S.; Kim, T.; Im, W. Automated Builder and Database of Protein/Membrane Complexes for Molecular Dynamics Simulations. PLoS One 2007, 2, e880. (20) Jo, S.; Kim, T.; Iyer, V. G.; Im, W. CHARMM-GUI: a webbased graphical user interface for CHARMM. J. Comput. Chem. 2008, 29, 1859−65. (21) Jo, S.; Lim, J. B.; Klauda, J. B.; Im, W. CHARMM-GUI Membrane Builder for mixed bilayers and its application to yeast membranes. Biophys. J. 2009, 97, 50−8. (22) Wu, E. L.; Cheng, X.; Jo, S.; Rui, H.; Song, K. C.; DávilaContreras, E. M.; Qi, Y.; Lee, J.; Monje-Galvan, V.; Venable, R. M.; Klauda, J. B.; Im, W. CHARMM-GUI Membrane Builder toward realistic biological membrane simulations. J. Comput. Chem. 2014, 35, 1997−2004. (23) Zhao, X.; Yu, H.; Yang, L.; Li, Q.; Huang, X. Simulating the antimicrobial mechanism of human beta-defensin-3 with coarsegrained molecular dynamics. J. Biomol. Struct. Dyn. 2015, 33, 2522−29. (24) Pronk, S.; Páll, S.; Schulz, R.; Larsson, P.; Bjelkmar, P.; Apostolov, R.; Shirts, M. R.; Smith, J. C.; Kasson, P. M.; van der Spoel, D.; Hess, B.; Lindahl, E. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 2013, 29, 845−854. (25) Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.; Mackerell, A. D., Jr. CHARMM general force field: A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 2010, 31, 671−90. (26) Best, R. B.; Zhu, X.; Shim, J.; Lopes, P. E. M.; Mittal, J.; Feig, M.; MacKerell, A. D. Optimization of the Additive CHARMM AllAtom Protein Force Field Targeting Improved Sampling of the Backbone ϕ, ψ and Side-Chain χ1 and χ2 Dihedral Angles. J. Chem. Theory Comput. 2012, 8, 3257−3273. (27) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; Joseph-McCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wiórkiewicz-Kuczera, J.; Yin, D.; Karplus, M. All-Atom Empirical Potential for Molecular Modeling and Dynamics Studies of Proteins. J. Phys. Chem. B 1998, 102, 3586−3616. (28) Klauda, J. B.; Venable, R. M.; Freites, J. A.; O’Connor, J. W.; Tobias, D. J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A. D.; 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. (29) Li, J.; Lakshminarayanan, R.; Bai, Y.; Liu, S.; Zhou, L.; Pervushin, K.; Verma, C.; Beuerman, R. W. Molecular dynamics simulations of a new branched antimicrobial peptide: A comparison of force fields. J. Chem. Phys. 2012, 137, 215101. (30) 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. (31) Cuendet, M. A.; van Gunsteren, W. F. On the calculation of velocity-dependent properties in molecular dynamics simulations using the leapfrog integration algorithm. J. Chem. Phys. 2007, 127, 184102. (32) Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: An N· log(N) method for Ewald sums in large systems. J. Chem. Phys. 1993, 98, 10089−10092. (33) Hess, B. P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 116−122. (34) Nosé, S. A unified formulation of the constant temperature molecular dynamics methods. J. Chem. Phys. 1984, 81, 511−519. (35) Hoover, W. G. Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A: At., Mol., Opt. Phys. 1985, 31, 1695−1697. (36) Parrinello, M.; Rahman, A. Polymorphic transitions in single crystals: A new molecular dynamics method. J. Appl. Phys. 1981, 52, 7182−7190.

ACKNOWLEDGMENTS This work was supported by NRF Grant s No. 2013R1A2A2A01067638, and 2012M3C1A6035362.

■

REFERENCES

(1) Fjell, C. D.; Hiss, J. A.; Hancock, R. E.; Schneider, G. Designing antimicrobial peptides: form follows function. Nat. Rev. Neurosci. 2012, 11, 37−51. (2) O’Connell, K. M.; Hodgkinson, J. T.; Sore, H. F.; Welch, M.; Salmond, G. P.; Spring, D. R. Combating multidrug-resistant bacteria: current strategies for the discovery of novel antibacterials. Angew. Chem., Int. Ed. 2013, 52, 10706−33. (3) Hancock, R. E.; Sahl, H. G. Antimicrobial and host-defense peptides as new anti-infective therapeutic strategies. Nat. Biotechnol. 2006, 24, 1551−7. (4) Ganz, T. Defensins: antimicrobial peptides of innate immunity. Nat. Rev. Immunol. 2003, 3, 710−20. (5) Nguyen, L. T.; Haney, E. F.; Vogel, H. J. The expanding scope of antimicrobial peptide structures and their modes of action. Trends Biotechnol. 2011, 29, 464−72. (6) Bond, P. J.; Khalid, S. Antimicrobial and cell-penetrating peptides: structure, assembly and mechanisms of membrane lysis via atomistic and coarse-grained molecular dynamics simulations. Protein Pept. Lett. 2010, 17, 1313−27. (7) Pazgier, M.; Hoover, D. M.; Yang, D.; Lu, W.; Lubkowski, J. Human beta-defensins. Cell. Mol. Life Sci. 2006, 63, 1294−313. (8) Dhople, V.; Krukemeyer, A.; Ramamoorthy, A. The human betadefensin-3, an antibacterial peptide with multiple biological functions. Biochim. Biophys. Acta, Biomembr. 2006, 1758, 1499−512. (9) Wu, Z.; Hoover, D. M.; Yang, D.; Boulegue, C.; Santamaria, F.; Oppenheim, J. J.; Lubkowski, J.; Lu, W. Engineering disulfide bridges to dissect antimicrobial and chemotactic activities of human betadefensin 3. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 8880−5. (10) Chandrababu, K. B.; Ho, B.; Yang, D. Structure, dynamics, and activity of an all-cysteine mutated human beta defensin-3 peptide analogue. Biochemistry 2009, 48, 6052−61. (11) Kluver, E.; Schulz-Maronde, S.; Scheid, S.; Meyer, B.; Forssmann, W. G.; Adermann, K. Structure-activity relation of human beta-defensin 3: influence of disulfide bonds and cysteine substitution on antimicrobial activity and cytotoxicity. Biochemistry 2005, 44, 9804−16. (12) Sudheendra, U. S.; Dhople, V.; Datta, A.; Kar, R. K.; Shelburne, C. E.; Bhunia, A.; Ramamoorthy, A. Membrane disruptive antimicrobial activities of human beta-defensin-3 analogs. Eur. J. Med. Chem. 2015, 91, 91−9. (13) Boniotto, M.; Antcheva, N.; Zelezetsky, I.; Tossi, A.; Palumbo, V.; Verga Falzacappa, M. V.; Sgubin, S.; Braida, L.; Amoroso, A.; Crovella, S. A study of host defence peptide beta-defensin 3 in primates. Biochem. J. 2003, 374, 707−14. (14) Schibli, D. J.; Hunter, H. N.; Aseyev, V.; Starner, T. D.; Wiencek, J. M.; McCray, P. B., Jr.; Tack, B. F.; Vogel, H. J. The solution structures of the human beta-defensins lead to a better understanding of the potent bactericidal activity of HBD3 against Staphylococcus aureus. J. Biol. Chem. 2002, 277, 8279−89. (15) Dowhan, W. Molecular basis for membrane phospholipid diversity: why are there so many lipids? Annu. Rev. Biochem. 1997, 66, 199−232. (16) Denning, E. J.; Beckstein, O. Influence of lipids on proteinmediated transmembrane transport. Chem. Phys. Lipids 2013, 169, 57− 71. (17) Epand, R. F.; Savage, P. B.; Epand, R. M. Bacterial lipid composition and the antimicrobial efficacy of cationic steroid compounds (Ceragenins). Biochim. Biophys. Acta, Biomembr. 2007, 1768, 2500−9. (18) Li, J.; Liu, S.; Lakshminarayanan, R.; Bai, Y.; Pervushin, K.; Verma, C.; Beuerman, R. W. Molecular simulations suggest how a branched antimicrobial peptide perturbs a bacterial membrane and 1789

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790

Article

Langmuir (37) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 1984, 81, 3684−3690. (38) Peng, C.; Liu, J.; Zhao, D.; Zhou, J. Adsorption of hydrophobin on different self-assembled monolayers: the role of the hydrophobic dipole and the electric dipole. Langmuir 2014, 30, 11401−11. (39) Eisenberg, D.; Weiss, R. M.; Terwilliger, T. C.; Wilcox, W. Hydrophobic moments and protein structure. Faraday Symp. Chem. Soc. 1982, 17, 109−120. (40) Eisenberg, D.; Schwarz, E.; Komaromy, M.; Wall, R. Analysis of membrane and surface protein sequences with the hydrophobic moment plot. J. Mol. Biol. 1984, 179, 125−42. (41) Kumari, R.; Kumar, R.; Lynn, A. g_mmpbsa–a GROMACS tool for high-throughput MM-PBSA calculations. J. Chem. Inf. Model. 2014, 54, 1951−62. (42) Parenti, M. D.; Rastelli, G. Advances and applications of binding affinity prediction methods in drug discovery. Biotechnol. Adv. 2012, 30, 244−250. (43) Meirovitch, H. Recent developments in methodologies for calculating the entropy and free energy of biological systems by computer simulation. Curr. Opin. Struct. Biol. 2007, 17, 181−186. (44) Sitkoff, D.; Sharp, K. A.; Honig, B. Accurate Calculation of Hydration Free Energies Using Macroscopic Solvent Models. J. Phys. Chem. 1994, 98, 1978−1988. (45) Yu, L.; Zhang, L.; Sun, Y. Protein behavior at surfaces: orientation, conformational transitions and transport. J. Chromatogr. A 2015, 1382, 118−34. (46) Liu, J.; Liao, C.; Zhou, J. Multiscale simulations of protein G B1 adsorbed on charged self-assembled monolayers. Langmuir 2013, 29, 11366−74. (47) Kubiak-Ossowska, K.; Mulheran, P. A. Mechanism of hen egg white lysozyme adsorption on a charged solid surface. Langmuir 2010, 26, 15954−65. (48) Reisser, S.; Strandberg, E.; Steinbrecher, T.; Ulrich, A. S. 3D hydrophobic moment vectors as a tool to characterize the surface polarity of amphiphilic peptides. Biophys. J. 2014, 106, 2385−94. (49) Tanizaki, S.; Feig, M. A generalized Born formalism for heterogeneous dielectric environments: application to the implicit modeling of biological membranes. J. Chem. Phys. 2005, 122, 124706. (50) Liu, F. F.; Dong, X. Y.; He, L.; Middelberg, A. P.; Sun, Y. Molecular insight into conformational transition of amyloid betapeptide 42 inhibited by (−)-epigallocatechin-3-gallate probed by molecular simulations. J. Phys. Chem. B 2011, 115, 11879−87. (51) Hoover, D. M.; Wu, Z.; Tucker, K.; Lu, W.; Lubkowski, J. Antimicrobial Characterization of Human β-Defensin 3 Derivatives. Antimicrob. Agents Chemother. 2003, 47, 2804−2809.

1790

DOI: 10.1021/acs.langmuir.5b04113 Langmuir 2016, 32, 1782−1790