A general mechanism for drug promiscuity: Studies with amiodarone and other antiarrhythmics

Abstract

Amiodarone is a widely prescribed antiarrhythmic drug used to treat the most prevalent type of arrhythmia, atrial fibrillation (AF). At therapeutic concentrations, amiodarone alters the function of many diverse membrane proteins, which results in complex therapeutic and toxicity profiles. Other antiarrhythmics, such as dronedarone, similarly alter the function of multiple membrane proteins, suggesting that a multipronged mechanism may be beneficial for treating AF, but raising questions about how these antiarrhythmics regulate a diverse range of membrane proteins at similar concentrations. One possible mechanism is that these molecules regulate membrane protein function by altering the common environment provided by the host lipid bilayer. We took advantage of the gramicidin (gA) channels' sensitivity to changes in bilayer properties to determine whether commonly used antiarrhythmics--amiodarone, dronedarone, propranolol, and pindolol, whose pharmacological modes of action range from multi-target to specific--perturb lipid bilayer properties at therapeutic concentrations. Using a gA-based fluorescence assay, we found that amiodarone and dronedarone are potent bilayer modifiers at therapeutic concentrations; propranolol alters bilayer properties only at supratherapeutic concentration, and pindolol has little effect. Using single-channel electrophysiology, we found that amiodarone and dronedarone, but not propranolol or pindolol, increase bilayer elasticity. The overlap between therapeutic and bilayer-altering concentrations, which is observed also using plasma membrane-like lipid mixtures, underscores the need to explore the role of the bilayer in therapeutic as well as toxic effects of antiarrhythmic agents.

Figure 1.

A schematic illustration of how amphiphilic drugs can modulate membrane protein function by a bilayer-mediated mechanism and structures of the antiarrhythmics. (A) Schematic representation of the bilayer-mediated regulation of membrane protein function, which arises because the reversible partitioning of the amphiphiles between the aqueous solution and the bilayer–solution interface alters lipid bilayer properties, including the elasticity (Evans et al., 1995; Zhelev, 1998; Bruno et al., 2013) and thus ΔG_def (and therefore $\Delta G_{\text{bilayer}}^{\text{I}\to \text{II}}$). In the figure, conformations I and II are denoted as “closed” and “open,” respectively. (B) Molecular structures of the antiarrhythmics amiodarone, dronedarone, propranolol, and pindolol.

Figure 2.

Antiarrhythmics alter lipid bilayer properties. (A, left) Fluorescence quench traces showing Tl⁺ quenching of ANTS fluorescence in DC_22:1PC LUVs without gA (−gA; the top two traces) and with gA (+gA; the bottom five traces) in the absence of drug (black, control) or with dronedarone (green), amiodarone (orange), propranolol (cyan), and pindolol (purple). Amiodarone, dronedarone, and propranolol increase the fluorescence signal up to 12% depending on the concentration, but the flux rate measurements were not affected. The results for each drug represent five to eight repeats (dots) and their averages (solid lines). (Right) Single repeats (dots) with stretched exponential fit (solid line). (B) Normalized quench rates determined from the stretched exponential fits at varying antiarrhythmic concentrations. Error bars represent mean ± SD (n = 3 – 5).

Figure 3.

Antiarrhythmics increase gA single-channel activity and decrease the bilayer deformation energy ($\Delta \Delta G_{\text{bilayer}}^{\text{M}\to \text{D}}$). (A) gA single-channel traces without (top row) and with (bottom row) the antiarrhythmics at the indicated concentrations; red and blue dashed lines indicate the average gA⁻(13) and gA(15) single-channel current amplitudes. (B) Changes in $\Delta \Delta G_{\text{bilayer}}^{\text{M}\to \text{D}},$ which were estimated from the ratio of the time-averaged number of gA channels in the presence (τ_drug · f_drug) and absence (τ · f) of the antiarrhythmic (compare Eq. 4). Blue symbols denote results for gA(15) channels, and red symbols denote results for gA⁻(13) channels. Error bars represent mean ± SD, if n ≥ 3; mean ± range/2, if n = 2.

Figure 4.

The antiarrhythmic-induced changes in the single-channel lifetimes of gA⁻(13) channels versus the changes in the lifetimes of gA(15) channels. (A) Natural logarithm of relative changes in τ₁₃ (ln(τ₁₃/τ_13cntrl)) versus the natural logarithm of relative changes in τ₁₅ (ln(τ₁₅/τ_15cntrl)) observed for dronedarone (green), amiodarone (orange circle), propranolol (blue triangle), and pindolol (black square) plotted together with results from Lundbæk et al. (2010b) and Rusinova et al. (2011). The points cluster around a straight line with slope 1.2 ± 0.03 (error bars represent mean ± SE). (B) Distribution of the slopes for the lnτ₁₃ versus lnτ₁₅ relations for the individual compounds in A. The distribution is fit by Gaussian function with a mean ± SD (σ calculated from the fit) of 1.3 ± 0.2, σ = 0.3. Changes in the histogram bin size result in the median slope ranging between 1.2 and 1.3. Inset illustrates an individual linear fit to the (ln(τ₁₃/τ_13cntrl)) versus ln(τ₁₅/τ_15cntrl) in the presence of dronedarone (green symbols). Slopes of the linear fits, such as that in the inset, obtained for each compound were used to construct the distribution in B.

Figure 5.

Survivor single-channel lifetime histograms for gA(15) in DC_18:1PC and DC_18:1PC/bSM/Chol bilayers with single-exponential fits. There is no evidence for the existence of more than one population of channels in either membrane.

Figure 6.

Dronedarone alters gA(15) channel function in bilayers formed from ternary DC_18:1PC/bSM/Chol 1:1:1 mixtures. (A) Relative changes in τ. (B) Relative changes in f. (C) The decrease in $\Delta \Delta G_{\text{bilayer}}^{\text{M}\to \text{D}}.$ Error bars represent mean ± range/2 (n = 2).