Poisson-Nernst-Planck models of nonequilibrium ion electrodiffusion through a protegrin transmembrane pore

Abstract

Protegrin peptides are potent antimicrobial agents believed to act against a variety of pathogens by forming nonselective transmembrane pores in the bacterial cell membrane. We have employed 3D Poisson-Nernst-Planck (PNP) calculations to determine the steady-state ion conduction characteristics of such pores at applied voltages in the range of -100 to +100 mV in 0.1 M KCl bath solutions. We have tested a variety of pore structures extracted from molecular dynamics (MD) simulations based on an experimentally proposed octomeric pore structure. The computed single-channel conductance values were in the range of 290-680 pS. Better agreement with the experimental range of 40-360 pS was obtained using structures from the last 40 ns of the MD simulation, where conductance values range from 280 to 430 pS. We observed no significant variation of the conductance with applied voltage in any of the structures that we tested, suggesting that the voltage dependence observed experimentally is a result of voltage-dependent channel formation rather than an inherent feature of the open pore structure. We have found the pore to be highly selective for anions, with anionic to cationic current ratios (I(Cl-)/I(K+)) on the order of 10(3). This is consistent with the highly cationic nature of the pore but surprisingly in disagreement with the experimental finding of only slight anionic selectivity. We have additionally tested the sensitivity of our PNP model to several parameters and found the ion diffusion coefficients to have a significant influence on conductance characteristics. The best agreement with experimental data was obtained using a diffusion coefficient for each ion set to 10% of the bulk literature value everywhere inside the channel, a scaling used by several other studies employing PNP calculations. Overall, this work presents a useful link between previous work focused on the structure of protegrin pores and experimental efforts aimed at investigating their conductance characteristics.

Figure 1. Pore-forming protegrins.

(A) A single protegrin peptide; note the abundance of positive charges due to arginine side chains (shown in blue). Cysteine-cysteine disulfide bonds are shown in yellow. (B) Eight protegrin peptides aggregate to form a membrane-spanning pore.

Figure 2. Concentration profiles of potassium and chloride.

The values shown are averages over the x and y directions within the ion-accessible region. These profiles correspond to the PNP model applied to a snapshot at 93.5 ns of the NPT segment of the MD simulations of Langham et al., with an applied voltage of −20 mV, 0.1 M KCl bath concentrations, and diffusion coefficient profile D4.

Figure 3. Comparison of I–V curves from the PNP model and experiments .

All data correspond to a snapshot at 93.5 ns of the NPT segment of the simulations of Langham and coworkers , using diffusion coefficient profile D4 (see Methods section). KCl salt bath concentrations were set to 0.1 M on both sides of the membrane in order to match experimental conditions.

Figure 4. Interaction of arginine residues with phosphate groups and chloride ions.

The numbers on the colour scales correspond to the number of phosphate groups/chloride ions within 7.5 Å of the ζ-carbon of the respective arginine residue. Data are averaged over the last 50 ns of the molecular dynamics simulation, as well as over the eight peptides.

Figure 5. Definition of arginine side chain dihedral angles.

Figure 6. Ion diffusion coefficient profiles.

All the profiles tested (D1–D4) are shown for both ions. Profile D3 is the only one that varies with the pore structure; the figure depicts diffusion coefficient profiles corresponding to the structure at 93.5 ns into the NPT segment of the MD simulations.