11292
J. Phys. Chem. B 2010, 114, 11292–11302
Adsorption of an Antimicrobial Peptide on Self-Assembled Monolayers by Molecular Dynamics Simulation Wael Soliman,† Subir Bhattacharjee,‡ and Kamaljit Kaur*,† Faculty of Pharmacy and Pharmaceutical Sciences, UniVersity of Alberta, Edmonton, Alberta T6G 2N8, Canada, and Department of Mechanical Engineering, UniVersity of Alberta, Edmonton, Alberta, T6G 2G8 Canada ReceiVed: May 3, 2010; ReVised Manuscript ReceiVed: July 12, 2010
We present results of molecular dynamics simulations of the interaction of a 48 amino acid peptide, carnobacteriocin B2, with model hydrophobic, anionic, and cationic self-assembled monolayers (SAMs). The model monolayers were formed by placing alkanethiols, HS-(CH2)10-X, where the terminal functional group X was chosen to be CH3, COO-, or NH3+. The carboxylate and amine groups were modeled as either fully charged or partially charged. Furthermore, simulations are presented for nanopatterned SAMs consisting of parallel stripes of hydrophobic/anionic and anionic/cationic SAMs. These simulations help elucidate the mechanisms of interaction of the peptide with model surfaces that emulate the chemical heterogeneity of lipid bilayer membranes or peptide nanoarrays. The simulation results depict how the nanoscale chemical heterogeneity of surfaces can simultaneously alter the peptide’s interaction with the surface and its secondary structure. Hydrophobic interactions result in the strongest adsorption of the peptide to the monolayer, simultaneously maintaining the structural integrity of the peptide. Electrostatic interactions, on the other hand, tend to enhance the solvation of the peptide, thereby causing radical changes in the secondary structure. Introduction The interaction of peptides and proteins with surfaces is ubiquitous in nature. These interactions dictate cellular signaling, uptake, and transport of nutrients across cell membranes as well as numerous other functions that are essential for life. Understanding the nature of these interactions is gaining importance in biomedical research and nanotechnology, rendering quantitative measurement and prediction of these interactions at an atomistic detail to be of paramount importance. In this respect, numerous experimental methodologies and engineered platforms for studying such interactions are being developed.1–8 In these systems, the peptides and proteins are immobilized on a substrate, followed by observing their interaction with specific analytes. This approach, which can be generally termed as the protein array (or protein chip) technology, is the cornerstone of proteomic analysis, single cell analysis, and molecular diagnostics.9,10 Immobilization of the proteins or peptides on microarrays is generally done through either noncovalent adsorption or covalent bonding. Such systems attempt to mimic the complex interactions between the functional groups of a protein or peptide, surfaces such as cell membranes, and the milieu in which they exist. The complexity of these interactions arises from the chemical heterogeneity of the interacting entities. We have recently performed MD simulations of small cationic antimicrobial peptides (AMPs) interacting with mixed anioniczwitterionic lipid bilayer membranes.11 This study indicated that the peptide interaction and eventual insertion into the bilayer can be strongly influenced by the functional group chemistries and heterogeneity scales of the lipid bilayer. Over the recent years, single cell detection or similar combinatorial techniques * To whom correspondence should be addressed. Tel.: 780-492-8917. Fax: 780-492-1217. E-mail.
[email protected]. † Faculty of Pharmacy and Pharmaceutical Sciences. ‡ Department of Mechanical Engineering.
of biomedical diagnostics are starting to use spot sizes of peptide arrays that increasingly resemble nano- and microscale surface heterogeneity.5,12 For instance, when individual cells need to be immobilized on a substrate, the dimension of the peptide substrate needs to be at best a few micrometers across.5 Protein nanoarrays for precise positioning of proteins have also been reported.13 More recently, dip-pen nanolithography (DPN) based lithographic techniques have allowed creation of SAM patterns that are barely a few tens of nanometers in feature size.12 Such small dimensions invariably bring the interacting moieties in the vicinity of surfaces that are chemically heterogeneous. As technology progresses, even smaller feature sizes and higher array densities will become available, thus making protein or peptide arrays virtually nanoheterogeneous substrates. In one of our recent studies, we created charge heterogeneous substrates consisting of chemically patterned self-assembled monolayers (SAMs) of alkanethiols on a gold substrate.14 These patterns were in the form of alternate positive-negative or hydrophobic-hydrophilic stripes having widths in the range of 1-3 µm. When rigid colloidal particles of comparable dimension were adsorbed onto these stripes, we observed different deposition probabilities near the regions of chemical heterogeneity (stripe edges). This indicates that even rigid colloidal particles experience nonuniform interaction forces and deposit unevenly on such surfaces.14 In light of this, we ask the question as to what happens to the structures and deposition behavior of more readily deformable proteins and peptides as they are brought in contact with chemically heterogeneous surfaces. Second, if these biomolecules are susceptible to structural modification near such surface chemical heterogeneity, do they retain their effectiveness as scaffolds or binding sites for the analytes? To address the above questions, we have used molecular dynamics (MD) simulations of peptide adsorption on model chemically patterned substrates in this study. Bacteriocins or amphipathic peptides have been used for coating different surfaces to create disinfecting antimicrobial
10.1021/jp104024d 2010 American Chemical Society Published on Web 08/06/2010
Adsorption of an Antimicrobial Peptide
Figure 1. Amino acid sequence of CbnB2. The helical residues (18-39) are underlined. Acidic, basic, and hydrophobic residues are shown in red, blue and yellow, respectively.
surfaces.15,16 It is of interest in this study to determine if adsorption of these AMPs to substrates bearing different surface chemistries and chemical heterogeneity patterns cause variations in their secondary structure, which may consequently adversely affect their activity. The objective of this study is to systematically investigate the adsorption of a well-known cationic AMP, carnobacteriocin B2 (CbnB2),11,17,18 on hydrophobic, positively charged, negatively charged, and heterogeneous surfaces. The modification of the peptide secondary structure due to peptideSAM interaction is also investigated. Eleven molecular dynamics simulations are reported, each conducted using different types of planar SAMs, including five simulations with homogeneous (same terminal functional groups) and six simulations with various combinations of heterogeneous surfaces (mixed terminal functional groups). The heterogeneous SAMs were designed as alternating hydrophobic/anionic and cationic/anionic stripes. The results from the simulations provide a systematic understanding of how nanoscale variations of chemical heterogeneity of surfaces can simultaneously alter the peptide’s interaction with the surface and its secondary structure. The mechanisms of interaction of the peptide with SAMs provide insight into the behavior of peptides when in contact with chemically heterogeneous systems such as cell membranes or peptide nanoarrays.12,19,20 Methods Computational Environment. MD simulations were performed using the GROMACS 3.33 simulation package21,22 and the GROMOS96 force field on a four node cluster as described previously.11,23 GROMACS, as well as the data analysis tools are available for free download from the Internet (www. gromacs.org). Swiss-PdbViewer,24 Discovery Studio ViewerLite 5.025 and VMD26 software were used to visualize and superimpose structures. Peptide Structure. CbnB2, a class IIa bacteriocin, is a 48residue cationic antimicrobial peptide with a net charge of +4 at neutral pH (Figure 1).18 It contains about 25 hydrophobic and 5 charged residues. CbnB2 folds into a well-defined central amphipathic R-helical structure (residues 18-39) and disordered N and C termini forming mainly a coil structure. Like most class IIa bacteriocins, the helical secondary structure of cbnB2 is stabilized in 2,2,2-trifluoroethanol or detergent micelles but becomes coiled (unstructured) in water.17,27 The three-dimensional coordinates based on NMR solution structure of CbnB2 were obtained from the protein data bank, PDB code 1CW5.18 The starting structure of CbnB2 used in this study corresponds to the solution structure of CbnB2 (1CW5) in TFE-d3/H2O (9: 1) ∼ pH 2.8. The peptide was considered to have positively and negatively charged N-terminal NH3+ and C-terminal COOgroups, respectively. Lys, Arg, and Glu residues were considered to be charged, and His residue was kept neutral giving CbnB2 a charge of +4. Alkanethiols for the SAMs. The alkanethiols chosen for the formation of the self-assembled monolayers were of the generic formula HS-(CH2)10-X, with X ) CH3, COO-, and NH3+ for a hydrophobic (uncharged), negative, and positive monolayer, respectively. The optimized PDB coordinates for the alkanethiols were generated employing Accelrys MS Modeling software
J. Phys. Chem. B, Vol. 114, No. 34, 2010 11293 (INSIGHT II 2000, Accelrys Inc., San Diego, CA). These coordinates were used to obtain the topology files for the alkanethiols using PRODRG software.28 Simulation Box Construction. Two sets of simulations were conducted, namely, (i) peptide near a homogeneous monolayer (simulations 1, 2, 2a, 3, and 3a) and (ii) peptide near a mixed (heterogeneous) monolayer (simulations 4, 5, 6, 6a, 7, and 7a). For the simulations with homogeneous monolayers, the simulation box was constructed by first placing the alkanethiols on a rectangular face (6 × 10 nm2) of the simulation box. The sulfur atoms of the thiol molecules were placed on either a regular square lattice with a spacing of 0.6 nm or on a hexagonal closepacked lattice (3 × 3 R30° array with 4.97 Å spacing) representing the typical atomic coordination of thiols on gold surface.29,30 Simulations employing these two types of SAMs indicated that the two types of atom placement have very little influence on the equilibrated distribution of the surface functional groups of the alkanethiols. As shown in Figure S1 (Supporting Information), the mass density of the CH3 SAM and water as well as the peptide radius of gyration remained very similar in the two simulations employing different arrangements of the sulfur atoms. Second, two types of SAM monolayers were simulated, the first was partially constrained31,32 and the second unconstrained (except for the terminal sulfur atom).33 In the partially constrained configuration, coordinate positions of all atoms were fixed except for the terminal functional groups and the subsequent five methylenes, X-(CH2)5, which were allowed to move freely in response to the environment. Once again, no significant difference in terms of peptide adsorption or conformational change onto these two types of SAMs was observed. Finally, all the eleven simulations were conducted with partially constrained SAMs on regular square lattice. Following placement of the thiols, the peptide was placed at a specified distance from the surface of the monolayer. The simulation box was then solvated with explicit SPC water molecules,34 and ions (Na+ and Cl-) were added to yield electroneutrality and a specified ionic strength (25 mM) of the solution. The height of the simulation box was selected to be 10 nm such that periodic conditions in the vertical (along z axis) direction do not affect the peptide-monolayer interaction significantly. To assess the influence of the periodicity and the box size, we conducted simulations with the box height ranging between 7 and 15 nm, and found no significant effect of periodicity on the simulation results as long as the box height is maintained at least 10 nm. Table 1 shows the details of the periodic cell for all the simulations conducted. With charged functional groups, we simulated two different protonated/deprotonated states of the SAMs by placing partial charges on these terminal groups. In simulations 6 and 7, we applied integer charges on every NH3+ and COO- group to emulate full protonation/deprotonation. In another set of simulations 6a and 7a, we only employed a partial charge of (0.05 on each of the terminal groups. For instance, the shift in protonation state can be produced by assigning each COO-terminated SAM an average charge of -0.05 as appropriate for a surface pKa of about 8.7 in a solution with pH 7.4. This surface pKa was obtained from the work of Creager and Clarke35 and has been used previously by Agashe and co-workers.31 In the manuscript, these partially charged surfaces are referred as COOH or NH2 whereas fully charged surfaces are referred as COO- or NH3+. SAM-NH2 was treated in a similar manner by assigning each NH2-terminated SAM an average charge of +0.05. For simulations employing mixed amine/carboxylate monolayers (simulations 6, 6a, 7, and 7a), equal numbers of
11294
J. Phys. Chem. B, Vol. 114, No. 34, 2010
TABLE 1: Details of MD Simulationsa
simulation
no. of atoms
no. of water molecules
1 2 2a 3 3a 4
54654 54639 55004 54842 55182 54642
17223 17029 17211 17030 17204 17126
5
54551
17129
6
55121
17219
6a
55097
17211
7
55106
17214
7a
55103
17213
Soliman et al. Results and Discussion
no. of alkanethiols (HS-(CH2)10-X) 192 (X ) CH3) 192 (X ) COO-) 192 (X ) COOH) 192 (X ) NH3+) 192 (X ) NH2) 96 (X ) COO-) and 96 (X ) CH3) 96 (X ) CH3) and 96 (X ) COO-) 96 (X ) COO-) and 96 (X ) NH3+) 96 (X ) COOH) and 96 (X ) NH2) 96 (X ) NH3+) and 96 (X ) COO-) 96 (X ) NH2) and 96 (X ) COOH)
a 1.4 nm initial spacing (at 0 ns) between the center of masses of CbnB2 and the monolayer surface; box size 10 × 6 × 10 nm3; NaCl concentration 25 mM; peptide concentration 2.7 mM; total simulation time 20 ns. Peptide was always placed in the box with N-terminal on the LHS and C-terminal on the RHS.
SAM-NH3+ and SAM-COO- molecules (96 of each) were combined to form a heterogeneous monolayer (one-half amine and the other half carboxylate) as shown in Figure 2. We also constructed mixed monolayers consisting of 50:50 hydrophobic CH3 and carboxylate terminated thiols (simulations 4 and 5). CbnB2 was placed in two orientations relative to these chemically heterogeneous monolayers. In simulations 4 and 5 respectively, the N-terminal region of the peptide was initially positioned over the COO- terminated SAM and the CH3 terminated SAM. In simulations 6 and 6a, the cationic Nterminal of the peptide was placed on top of the anionic (COO-) half of the monolayer and C-terminal was placed toward the cationic (NH3+) monolayer surface, whereas in simulations 7 and 7a, the cationic N-terminal was placed on top of the cationic half of the monolayer (Figure 2). Simulation Methods and Parameters. All systems were simulated at 300 K using periodic boundary conditions. Weak coupling of the proteins to a solvent bath of constant temperature was maintained using the Berendsen thermostat with a coupling constant τT ) 0.1 ps. The pressure was controlled using the Berendsen algorithm at 1 bar with a coupling constant τP ) 1 ps at 300 K. For all simulations, the neighbor list was updated every 10 steps, with a neighbor list cutoff distance of 1.2 nm. The Lennard-Jones interactions were truncated at a cutoff distance of 1.2 nm. The long-range electrostatic interactions were modeled using the particle mesh Ewald (PME) summation method with a cutoff distance of 1.2 nm for the real space. The integration time step was 2 fs, and the coordinates and velocities were saved every 4 ps. The LINCS algorithm was used to restrain all bond lengths.36 The system was energy minimized before the MD simulation using 200 steps of the steepest descent energy minimization method in order to relax any steric conflicts generated during the setup. The equilibration of the CbnB2SAM system was achieved by performing a 3 ns MD run with positional restraint on the peptide molecule. Following this, a full MD run of 20 ns was performed without any restraints on the peptide. Simulations were analyzed using various GROMACS postprocessing routines.
Peptide Adsorption on Homogeneous SAM Surface. The adsorption of the peptide CbnB2 on SAM surfaces was monitored using the separation distance between the center of mass of the peptide and the monolayer surface. Figure 3 shows the variation of the separation distance of CbnB2 from the homogeneous SAMs containing different terminal functional groups with time. For all the simulations, the initial separation distance was about 1.4 nm. CbnB2 is a cationic amphipathic peptide and consists of several hydrophobic residues (Figure 1). It was attracted toward all the SAM surfaces consisting of different terminal groups, except when the SAM consisted of protonated amine terminal groups (simulation 3). For all other types of SAMs, the final distance of the peptide from the SAM was 0.84 nm ((0.06 nm) after 20 ns. The rate of approach was fastest in the case of CH3 SAM (simulation 1) followed by the uncharged amine terminated SAM (simulation 3a). For the charged carboxylate SAM (simulation 2), the peptide adsorbed onto the surface through a slow and gradual approach. In simulation 3, CbnB2 drifted away from the positively charged ammonium ion containing SAM surface and maintained a distance of 2.9 ( 0.2 nm from the surface (Figure 3). The movement of the peptide toward the substrates containing hydrophobic and negatively charged terminal groups can readily be explained on the basis of hydrophobic and Coulomb attraction. It is also evident that the hydrophobic attraction results in a closer positioning of the peptide center of mass relative to the monolayer (as is evident on all three types of uncharged CH3, NH2, and COOH surfaces). For the peptide interacting with a positively charged amine surface, one can observe that the equilibrium distance attained by the positively charged peptide from the SAM was about 2.9 nm, whereas the box dimension in the z direction was 10 nm. For the 25 mM background electrolyte (NaCl) concentration, the Debye screening length of the electric double layer interactions was calculated to be about 2 nm. Therefore, at a separation distance of ∼3 nm, the electric double layer interactions between the peptide and the substrate is very small. In fact, we suspect that attainment of this separation distance by the peptide in presence of electrostatic repulsion might be similar to the equilibration of a small colloid at the secondary minimum of the DerjaguinLandau-Verwey-Overbeek (DLVO) potential.37 The equilibrium distance of the peptide owing to electrostatic repulsion from the substrate was critical in determining how large the simulation box height should be to attenuate the influence of the electrostatic interactions due to the periodic image of the SAMs (acting from the top of the box). In some preliminary studies, when the simulations were conducted with boxes that were
