Mechanistic study on uptake of cell penetrating peptides into model lipid bilayers and transient pore formation
Date
2015
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Structural mechanisms and underlying thermodynamic determinants of efficient internalization of charged cationic peptides (cell-penetrating peptides, CPPs) such as TAT, polyarginine, and their variants, into cells, cellular constructs, and model membrane/lipid bilayers (large and giant unilamellar or multilamelar vesicles) continue to garner significant attention. Two widely-held views on the translocation mechanism center on endocytotic and non-endocytotic (diffusive) processes. Using the translocation of short, charged cationic oligo-arginine peptides (mono-, di-, and tri-arginine) from bulk aqueous solution into model DMPC bilayers, we explore the question of the similarity of thermodynamic and structural predictions obtained from molecular dynamics simulations using all-atom and the Martini coarse-grain force fields. Specifically, we estimate potentials of mean force associated with translocation using standard all-atom (CHARMM36 lipid) and polarizable and non-polarizable Martini force fields, as well as a series of modified Martini-based parameter sets. We are able to reproduce qualitative features of potentials of mean force of single amino acid sidechain analogues into model bilayers. Our simulations also demonstrate a remarkable similarity in the structural aspects of the ensemble of configurations generated using the all-atom and coarse-grain force fields. We next explored the dependence of translocation free energetics on peptide structure and conformation via calculation of potentials of mean force along pre-selected reaction paths allowing and preventing membrane deformations that lead to pore formation by using umbrella sampling molecular dynamics simulations with coarse-grained Martini lipids and polarizable coarse-grained water. Within the context of the coarse-grained force fields we employ, we observe significant barriers for Arg 9 translocation from bulk aqueous solution to bilayer center. The pore-forming paths systematically predict lower free energy barriers (ca. 90 kJ/mol lower) than the non pore-forming paths, again consistent with all-atom force field simulations. The current force field suggests no preference for the more compact or covalently cyclic structures upon entering the bilayer. Decomposition of the PMF into system's components indicates that the dominant stabilizing contribution along the pore-forming path originates from the membrane as both layers are deformed due to the formation of a pore. Our analysis revealed that though there is significant entropic stabilization arising from the enhanced configurational entropy exposing more states as the peptide moves through the bilayer, the enthalpic loss (as predicted by the interactions of this coarse-grained model) far outweighs any former stabilization, thus leading to a significant barrier to translocation. We also observe reduction in the translocation free energy barrier for a second Arg 9 entering the bilayer in the presence of an initial peptide restrained at the center, again, in qualitative agreement with all-atom force fields. Furthermore, we explored the roles of cholesterol and anionic lipid in the systems of cationic TAT peptide transferring across a series of DPPC/DPPS model membranes with 0-30 mol% cholesterol. The trend of the translocation free energetics are in agreement with experimental findings, where the TAT peptide translocates through highly negatively charged and cholesterol-depleted membrane at low free energy cost. Structural analysis indicates that cholesterol increases the stiffness of the membrane, and raises the penalty of the lipid reorientation and membrane deformation which is known as the necessary steps of the peptide permeation. On the contrary, the addition of anionic lipid DPPS enhances the peptide association on the membrane surface. The strong peptide-lipid interaction with both leaflets of the bilayer favors the transmembrane pore-formation, and reduces the barrier. Moreover, although the pore formation provides a low free energetic path for CPPs translocation, the initiation and formation of pores are associated with a non-trivial free energy cost. Due to experimental and modeling challenges related to the long timescales of the translocation process, we use umbrella-sampling molecular dynamics simulations with a lipid-density based order parameter to investigate membrane pore-formation free energy employing Martini coarse-grained models. We investigate structure and thermodynamic features of the pore in 18 lipids spanning a range of headgroups, charge states, acyl chain lengths and saturation. We probe the dependence of pore-formation barriers on area per lipid, lipid bilayer thickness, membrane bending rigidities in three different lipid classes. The pore formation free energy in pure bilayers and peptide translocating scenarios are significantly coupled with bilayer thickness. Thicker bilayers require more reversible work to create pores. Pore formation free energy is higher in peptide-lipid systems relative to the peptide-free lipid systems due to penalties to maintain solvation of charged hydrophilic solutes within the membrane environment. Lastly, we estimated the free energetics of a new class of cell-penetrating peptides, SMTPs, that were recently identified using high-throughput, orthogonal screening assays. We consider model bilayers composed of POPC and POPG, the latter providing the anionic component as used in experimental studies. In the case of the SMTP, potentials of mean force are systematically lower relative to the Arg9 case. Though the barriers predicted by the simulations, on the order of 20 kcal/mol, are still rather large to recapitulate the experimental kinetics of internalization, we emphasize that the qualitative trend of reduction of barrier heights is a significant result. At the very least, we seek to find trends demonstrating higher stability (or lower free energy barriers to translocation) for screened and/or designed peptides known to spontaneously partition into cells and bilayers. We note that the binding of the SMTPs is critically dependent on the mix of hydrophilic and hydrophobic residues that constitute the amino acid motif/sequence of these peptides. For the cationic Arg9 which only contains hydrophilic residues, there is no tight binding observed. The specific motif ΦRΦΦR (Φ is hydrophobic residue, and R is arginine) is a potential sequence in drug/peptide design. The SMTPs with this motif are able to translocate into membrane at a significantly lower free energy cost, compared to the negative control peptides. Finally, we compare the different membrane perturbations induced by the presence of the different peptides in the bilayer center. In some cases, hydrophilic pores are observed to form, thus conferring stability to the internalized state. In other cases, SMTPs are associated only with membrane defects such as induced membrane curvature. These latter observations suggest some influence of membrane rigidity as embodied in the full range of membrane undulatory modes in defining pore forming propensities in bilayers.