The Single Residue K12 Governs the Exceptional Voltage Sensitivity of Mitochondrial Voltage-Dependent Anion Channel Gating

Abstract

The voltage-dependent anion channel (VDAC) is a β-barrel channel of the mitochondrial outer membrane (MOM) that passively transports ions, metabolites, polypeptides, and single-stranded DNA. VDAC responds to a transmembrane potential by "gating," i.e. transitioning to one of a variety of low-conducting states of unknown structure. The gated state results in nearly complete suppression of multivalent mitochondrial metabolite (such as ATP and ADP) transport, while enhancing calcium transport. Voltage gating is a universal property of β-barrel channels, but VDAC gating is anomalously sensitive to transmembrane potential. Here, we show that a single residue in the pore interior, K12, is responsible for most of VDAC's voltage sensitivity. Using the analysis of over 40 μs of atomistic molecular dynamics (MD) simulations, we explore correlations between motions of charged residues inside the VDAC pore and geometric deformations of the β-barrel. Residue K12 is bistable; its motions between two widely separated positions along the pore axis enhance the fluctuations of the β-barrel and augment the likelihood of gating. Single channel electrophysiology of various K12 mutants reveals a dramatic reduction of the voltage-induced gating transitions. The crystal structure of the K12E mutant at a resolution of 2.6 Å indicates a similar architecture of the K12E mutant to the wild type; however, 60 μs of atomistic MD simulations using the K12E mutant show restricted motion of residue 12, due to enhanced connectivity with neighboring residues, and diminished amplitude of barrel motions. We conclude that β-barrel fluctuations, governed particularly by residue K12, drive VDAC gating transitions.

Figure 1.

Summary of all-atom MD simulations of WT VDAC1. (A) Time records of ionic current for a VDAC1 channel embedded in a DOPC membrane and subjected to +160 mV transmembrane potential for 4 independent simulations. Instantaneous current is filtered with an 8-pole digital Bessel filter (cutoff frequency 6.9 MHz). Histograms are compatible with the existence of conductance substates (horizontal dashed lines); colors correspond to individual traces, while gray is the sum. (B) Free energy landscape of VDAC motions as determined by time-lagged independent correlation analysis. Axes are the mean-subtracted dominant principal components for the set of barrel diameters (D_PC) and the set of axial positions of charged residues (z_charged). (C) Molecular representation of conformational dynamics in the β-barrel and N-terminal domains of VDAC1 channel, with selected residues exhibiting large conformational plasticity labeled. Three representative frames for representative structures from states S1, S2 and S4 of VDAC1 identified in B are shown in gold, magenta, and blue ribbon representations, respectively. (D) Root-mean-square fluctuations by residue for each state identified in (B).

Figure 2.

Summary of BD simulations of WT VDAC. Brownian dynamics simulations of 600 randomly selected structures, 150 from each state. (A) Conductance distributions. Horizontal lines show mean values with the 68% confidence interval of the mean value. (B) Selectivity distributions. (C−D) Color maps of the residue motions that are most strongly correlated to the conductance, as determined by the correlation coefficient between the residue radial (C) or axial (D) distance from the center of mass. The magenta (cyan) color denotes a positive (negative) correlation, i.e. the conductance increases (decreases) with an increase in the corresponding coordinate. K12 is clearly identified as the residue whose motions are most strongly correlated with the channel conductance.

Figure 3.

Gating behavior of VDAC constructs with various mutations at residue 12. (A) Current recordings from single VDACs show that gating is significantly suppressed in mutants relative to the wild type, with the strongest suppression occurring with the charge-reversed mutant K12E. Dashed lines indicate zero current level. Recordings were digitally filtered at 500 Hz using a low pass Bessel (8-pole) filter. (B) Quantification of gating on a multichannel system by the normalized average conductance as a function of the applied voltage, showing the reduction in gating with the charge-reversed residue 12 mutant (left) and the constructs with a neutralized residue 12 (right). Data are means of 14 (WT), 6 (K12E), 5 (K12A), and 4 (K12S) independent experiments ± SD. Error bars are shown every five points for clarity.

Figure 4.

Crystallographic structure of K12E VDAC. (A) Superposed WT VDAC (in blue) and K12E VDAC (in orange) structures (top view). (B) Detail of the superposition of N-terminal helix from WT VDAC1 (in blue) and K12E VDAC1 (in orange). The light dotted lines denote H-bonding with water while the heavy dashes denote intraprotein interactions. The dashes/dots and water molecules, depicted as spheres, are colored the same as the associated protein. The PDB of the K12E mutant is 7TCV.

Figure 5.

Atomistic MD simulations of VDAC-K12E mutant. (A) Current record of 20 μs of simulation. Instantaneous current is filtered with an 8-pole digital Bessel filter (cutoff frequency 6.9 MHz). Histograms and traces are color coded; the gray histogram is the sum of both groups. The MD simulation seeded from new X-ray structures of K12E is denoted with * and shows a similar current level to the simulation seeded from an in silico mutation of the WT crystallographic structure. (B) Free energy landscape of VDAC-K12E motions showing significant alterations from the WT (Figure 1B). (C) 1D Free Energy Profile of residue 12 in the WT (K12) or mutant (K12E) along the Z-axis of the channel. The coordinate z₁₂ denotes the z-coordinates of the nitrogen atom of K12 or the carbon atom of the carboxylate group of E12, which are used to compute free-energy profiles, F = −log(ρ(z₁₂)), where ρ(z₁₂) is the probability density. Unlike the two stable positions (inset graphic) in the WT, which occur over a broad range of z₁₂, the potential minima in the mutant are restricted to z₁₂ < 0.

Figure 6.

Simulated molecular fluctuations. (A) Root mean square fluctuations calculated for each residue in the WT and K12E mutants. The average RMSFs are shown in the inset. (B) Difference in RMSF of each residue between K12E and WT. Residue fluctuations are significantly reduced across much of the protein upon mutation to K12E. The average change in RMSF is ⟨ΔRMSF⟩ = −0.10 ± 0.02 Å (68% CI).

Figure 7.

Interaction networks computed for WT and K12E VDAC. Networks were grouped into interaction communities labeled with numbers. Each community is identified based on time-averaged connectivity among local nodes. Edges were weighted using correlation functions computed between nodes. Graphic produced using VMD.