Assessing the Perturbing Effects of Drugs on Lipid Bilayers Using Gramicidin Channel-Based In Silico and In Vitro Assays

Abstract

Partitioning of bioactive molecules, including drugs, into cell membranes may produce indiscriminate changes in membrane protein function. As a guide to safe drug development, it therefore becomes important to be able to predict the bilayer-perturbing potency of hydrophobic/amphiphilic drugs candidates. Toward this end, we exploited gramicidin channels as molecular force probes and developed in silico and in vitro assays to measure drugs' bilayer-modifying potency. We examined eight drug-like molecules that were found to enhance or suppress gramicidin channel function in a thick 1,2-dierucoyl-sn-glycero-3-phosphocholine (DC_22:1PC) but not in thin 1,2-dioleoyl-sn-glycero-3-phosphocholine (DC_18:1PC) lipid bilayer. The mechanism underlying this difference was attributable to the changes in gramicidin dimerization free energy by drug-induced perturbations of lipid bilayer physical properties and bilayer-gramicidin interactions. The combined in silico and in vitro approaches, which allow for predicting the perturbing effects of drug candidates on membrane protein function, have implications for preclinical drug safety assessment.

Figure 1

Gramicidin channel function. (a) Cation conducting β^6.3-helical gramicidin A (gA) channels form by transmembrane dimerization of two non-conducting gA monomer subunits. (b) When amphiphiles (drugs) are added to the aqueous phase and partition into the bilayer, it will alter physical properties and thereby shift the gramicidin monomer ↔ dimer equilibrium, usually toward the dimer.

Figure 2

(a) PMF profiles for gA monomer ↔ dimer transition in the DC_18:1PC and DC_22:1PC bilayers (orange: AA-DC_22:1PC; red: CG-DC_22:1PC; blue: AA-DC_18:1PC; black: CG-DC_18:1PC). The reaction coordinate is the center-of-mass distance between the two gA monomers. In the AA simulations, the center-of-mass of the gA monomer is defined using all Cα atoms of the monomer; in the CG simulations, the center of mass of the monomer is defined using all backbone beads of the monomer. In the AA-REUS simulations, two structurally different dimers are obtained at gA–gA distances of ∼1.3 and ∼1.5 nm. At d_gA–gA ≈ 1.3 nm, the two subunits can form a maximum number of six hydrogen bonds, while at d_gA–gA ≈ 1.5 nm, the two subunits can only form a maximum number of four hydrogen bonds due to the relative rotation between the two monomers. The two different gA dimer structures are also observed in the CG-REUS simulations, and the derived CG PMF profiles exhibit a free energy minimum at d_gA–gA = 1.3 nm and a kink at d_gA–gA = 1.6 nm. (b) Effects of the tested drugs on the hydrophobic thickness of DC_18:1PC and DC_22:1PC bilayers. The methods used to calculate the bilayer’s hydrophobic thickness are illustrated in Figure S1. For the DC_18:1PC bilayer, the calculated hydrophobic thickness values using the AA and the CG models deviate by ∼0.2 nm. This discrepancy can be attributed to the four-carbon mapping scheme of the CG building blocks, as illustrated in Figure S1. The error bars for AA and CG simulations are ±0.05 and ± 0.03 kcal/mol, respectively.

Figure 3

Effects of drugs on gA monomer ↔ dimer transition. (a) Snapshots for two gA monomers (colored cyan and orange) and the dimeric channel in the DC_22:1PC bilayer doped with capsaicin (colored yellow). Effects of different drugs on the PMF for gA monomer ↔ dimer transition in (b) DC_18:1PC and (c) DC_22:1PC bilayers. All of the CG PMF profiles are well converged (see Supporting Information) and the error bars are within 0.05 kcal/mol.

Figure 4

Effects of drugs on the gramicidin monomer ↔ dimer equilibrium. (a) Schematic description of the stopped-flow fluorescence quench experiments: gramicidin permeable Tl⁺ quenches the LUV encapsulated fluorophore ANTS. (b) Representative fluorescence quench traces in DC_22:1PC LUVs doped with gD. (c) Effects of drugs on the fluorescence quench rates (gramicidin monomer ↔ dimer equilibrium) in DC_18:1PC LUVs doped with gD (light gray), DC_22:1PC LUVs doped with gD (dark gray), and DC_22:1PC LUVs doped with gA (black). The aqueous drug concentrations were 100, 30, 1800, 10, 300, 30, and 30,000 μM for capsaicin, resveratrol, octanol, C12E6, FC12, Triton X-100, and cyclohexane, respectively, and the estimated molar ratios of the drugs in the bilayers were 0.32, 0.02, 0.48, 0.04, 0.21, 0.04, and 0.93, respectively; cholesterol was added at a molar ratio of cholesterol:lipid of 1:10 when preparing the LUVs. Mean ± S.D. n = 3–4.

Figure 5

Decomposition of the PMF profiles into energetic contributions associated with (a) gA–lipid bilayer, (b) gA–gA, (c) gA–drug, and (d) gA–solvent interactions in both DC_22:1PC and DC_18:1PC bilayers. The decomposed PMF profiles do not contain the Jacobian correction term for the free energy, but this correction term is important to reconcile the PMF obtained with different methods (see Figures S10 and S11 for the differences between WHAM, force integration (FI) and decomposed force integration (DFI) approaches).