Cannabidiol inhibits the skeletal muscle Nav1.4 by blocking its pore and by altering membrane elasticity

Abstract

Cannabidiol (CBD) is the primary nonpsychotropic phytocannabinoid found in Cannabis sativa, which has been proposed to be therapeutic against many conditions, including muscle spasms. Among its putative targets are voltage-gated sodium channels (Navs), which have been implicated in many conditions. We investigated the effects of CBD on Nav1.4, the skeletal muscle Nav subtype. We explored direct effects, involving physical block of the Nav pore, as well as indirect effects, involving modulation of membrane elasticity that contributes to Nav inhibition. MD simulations revealed CBD's localization inside the membrane and effects on bilayer properties. Nuclear magnetic resonance (NMR) confirmed these results, showing CBD localizing below membrane headgroups. To determine the functional implications of these findings, we used a gramicidin-based fluorescence assay to show that CBD alters membrane elasticity or thickness, which could alter Nav function through bilayer-mediated regulation. Site-directed mutagenesis in the vicinity of the Nav1.4 pore revealed that removing the local anesthetic binding site with F1586A reduces the block of INa by CBD. Altering the fenestrations in the bilayer-spanning domain with Nav1.4-WWWW blocked CBD access from the membrane into the Nav1.4 pore (as judged by MD). The stabilization of inactivation, however, persisted in WWWW, which we ascribe to CBD-induced changes in membrane elasticity. To investigate the potential therapeutic value of CBD against Nav1.4 channelopathies, we used a pathogenic Nav1.4 variant, P1158S, which causes myotonia and periodic paralysis. CBD reduces excitability in both wild-type and the P1158S variant. Our in vitro and in silico results suggest that CBD may have therapeutic value against Nav1.4 hyperexcitability.

Figure 1.

Effects of CBD on POPC membrane via MD simulations and H² NMR. (a and b) The effects of CBD on POPC membrane area per lipid and lipid diffusion. System 0 is the control; system 1 is two CBD molecules in symmetry (one in each leaflet); system 2 is three CBD molecules in asymmetry (only in a single leaflet); and system 3 is six CBD molecules in symmetry (three in each leaflet). Area per lipid (a) and mean square displacement as a function of time (b) are not affected by CBD. (c) Distribution of CBD into the membrane across a range of conditions. The distribution of phosphate groups is shown as solid lines and the distribution of CBD is shown as dotted lines. The bilayer thickness remains ∼4 nm in the presence and absence of CBD. (d) Order parameter of lipid acyl chains estimated from the MD simulations. (e) Snapshot of a CBD molecule in the POPC leaflet extracted from the MD simulations (see Video 1). The zoomed-in image shows localization of CBD molecule below the leaflet headgroup. (f) NMR spectra collected at 20°C. (g) The spectra in f have been dePaked, showing doublets that correspond to individual palmitoyl methylene and methyl groups. The frequency separation of a given doublet is directly proportional to its order parameter. (h) The effect of CBD on the NMR order parameter profile of POPC-d31’s palmitoyl chain.

Figure S1.

²H NMR at different temperatures and further characterization of F1586A. (a–c) Order parameters associated with POPC membranes at 20°C (a), 30°C (b), and 40°C (c).

Figure 2.

CBD alters lipid bilayer properties in GFA. (a) Cartoon representation of gramicidin monomers in each leaflet coming together (dimerizing) to form cationic channels. The dimerization of the gramicidin channels is directly related to membrane elasticity. These properties are used to assay compound (e.g., CBD) effects on membrane elasticity. (b) Chemical structures of CBD and Triton X-100. (c) Fluorescence quench traces showing Tl⁺ quench of ANTS fluorescence in gramicidin-containing DC_22:1PC LUVs with no drug (control, black) and incubated with CBD for 10 min at the noted concentrations. The results for each drug represent five to eight repeats (dots) and their averages (solid white lines). (d) Single repeats (dots) with stretched exponential fits (red solid lines). (e) Fluorescence quench rates determined from the stretched exponential fits at varying concentrations of CBD (red) and Triton X-100 (purple, from Ingólfsson et al., 2010) normalized to quench rates in the absence of drug. Mean ± SD, n = 2 (for CBD).

Figure 3.

Inhibition of Nav1.4 pore by CBD, F1586A reduces inhibition. (a) Side view of CBD docked into the pore of the human Nav1.4 structure. The structure is colored by domain. DIV is colored in deep blue. (b) Zoomed-in side view in which F1586 is colored yellow. (c and d) Representative families of macroscopic current traces from WT-Nav1.4 and F1586A. (e) Voltage dependence of activation as normalized conductance plotted against membrane potential (Nav1.4: V_1/2 = −19.9 ± 2.7 mV, z = 2.8 ± 0.3, n = 5; F1586A: V_1/2 = −22.4 ± 2.2 mV, z = 3.0 ± 0.3, n = 7; P > 0.05 for both V_1/2 and z). (f) Voltage dependence of SSFI as normalized current plotted against membrane potential. Channels were held at −130 mV for 200 ms (Nav1.4: V_1/2 = −64.1 ± 2.4 mV, z = −2.7 ± 0.3, n = 8; F1586A: V_1/2 = −63.3 ± 3.0 mV, z = −3.5 ± 0.3, n = 8; P > 0.05 for both V_1/2 and z). (g and h) Lidocaine/CBD inhibition of Nav1.4 and F1586A from −110 mV (rest) at 1 Hz with a 20-ms depolarizing pulse (lidocaine-Nav1.4: mean block = 60.6 ± 2.3%, n = 3; lidocaine-F1586A: mean block = 24.6 ± 9.3%, n = 3, #, P = 0.020; CBD-Nav1.4: mean block = 47.3 ± 3.7%, n = 5; CBD-F1586A: mean block = 25.3 ± 4.8%, n = 3, *, P = 0.037). Sample traces before and after compound perfusion are shown. All values in e–h are reported as mean ± SEM.

Figure S2.

Nav1.4 WT interactions with CBD. (a–d) CBD posed in the human Nav1.4 structure using molecular docking.

Figure S3.

Nav1.4 F1586A interactions with CBD. (a–c) CBD posed in the human Nav1.4 structure using molecular docking.

Figure S4.

Further characterization of F1586A. (a and b) Normalized activating currents as a function of potential, recovery from fast inactivation (Nav1.4: τ_Fast = 0.0025 ± 0.00069 s, τ_Slow = 0.224 ± 0.046 s, n = 7; F1586A: τ_Fast = 0.0021 ± 0.00043 s; τ_Slow = 0.093718 ± 0.03673 s, n = 6; P > 0.05 for τ_Fast and P = 0.0250 for τ_Slow).

Figure S5.

Sample normalized time dependence. (a–d) Time dependence of CBD (70.5 ± 6.9 s), flecainide (4.8 ± 0.7 s), lidocaine (4.7 ± 0.2 s), and negative control (no compound). Error bars are SE in mean.

Figure S6.

CBD interactions with DIV-S6, using ITC. (a) Representative ITC traces for titration of 100 mM lidocaine into 1 mM peptide or blank buffer. The heat signal, once the binding is saturated is the same as the blank if the blank was only measuring the interaction if lidocaine with the solution. The blank measures 3 interactions, interactions between solute molecules, solute and lidocaine and lidocaine and lidocaine. As more lidocaine is added with each injection the solution is changed. This makes the heat of interaction with in-blank trace different from beginning to end. This change is due to a change in amount of lidocaine in the solution. Because this change is progressive with each injection and that the injections into the peptide are the same volume, subtraction was used as a means to quantify lidocaine peptide interaction. (b) Representative ITC traces for titration of 40 mM CBD into 1 mM peptide or blank buffer. (c and d) The blank condition subtracted heat of titration in protein condition is shown for lidocaine (c) and CBD (d). A peak heat of 968.0 ± 23.4 kcal mol⁻¹ was seen for lidocaine (n = 3) titration, and a peak heat of 1,022.2 ± 160.6 kcal mol⁻¹ was seen for the CBD (n = 4) titration. Error bars are SE in mean.

Figure 4.

CBD interactions with and through Nav fenestrations. (a) Side view of CBD docked into the human Nav1.4 structure. The structure is colored by domain (matched color to domain is shown in f). CBD is represented in purple. (b) Side view of all four sides of human Nav1.4 (colored by domain). Nav1.4 fenestrations are highlighted in red, along with the position of respective residues that were mutated into tryptophans (W). (c) Computationally predicted mutagenesis of fenestrations results two full (pose 1) and two partial (pose 2) occlusions/alterations (paralleled domains). (d) Lidocaine (1.1 mM) inhibition of Nav1.4 and WWWW from −110 mV (rest) at 1 Hz 20 ms depolarizing pulse (Nav1.4: mean block = 60.6 ± 2.3%, n = 3; WWWW: mean block = 53.6 ± 11.7%, n = 3; P > 0.05), flecainide (350 µM) inhibition (Nav1.4: mean block = 64.6 ± 6.0%, n = 3; WWWW: mean block = 76.4 ± 11.3%, n = 3; P > 0.05), and CBD (10 µM) inhibition (Nav1.4: mean block = 47.3 ± 3.7%, n = 5; WWWW: mean block = 6.4 ± 1.3%, n = 5; *, P = 0.0001). Traces before and after compound perfusion are shown. Values are reported as mean ± SEM. (e) CBD pathway through the Nav1.4 fenestration from side view, as predicted by MD simulations. Red and blue correlate with CBD being inside and outside the fenestration, respectively (see Videos 2 and 3). (f) CBD pathway from top view of the channel. (g) Progressive snapshots of the movement of CBD over time from inside to outside the channel.

Figure S7.

Structural integrity MD simulation of CBD pathway through the Nav1.4 fenestration. Root mean square deviation (RMSD) of the fenestration residues as a function of time in the absence (black) and the presence of CBD passing through the fenestration (red and green, two different simulation parameter sets). Apo means unbound in the absence of CBD. The similar RMSD profiles show that CBD’s passage does not distort the structural integrity of the fenestration.

Figure 5.

Effects of CBD on Nav1.4 gating. (a and b) Voltage dependence of activation as normalized conductance plotted against membrane potential in 1 µM CBD (control: V_1/2 = −19.9 ± 4.2 mV, z = 2.8 ± 0.3, n = 5; CBD: V_1/2 = −14.3 ± 4.2 mV, z = 2.8 ± 0.3, n = 5; P > 0.05 for both V_1/2 and z) and normalized activating currents as a function of potential. (c) Voltage dependence of 200 ms (channels were held at −130 mV for 200 ms) F-I curve plotted against membrane potential in 1 µM CBD (control: V_1/2 = −64.1 ± 2.4 mV, z = −2.7 ± 0.3, n = 8; CBD: V_1/2 = −72.7 ± 3.0 mV, z = −2.8 ± 0.4, n = 5; P = 0.0281 for V_1/2 and P > 0.05 for z). (d) Recovery from fast inactivation in 1 µM CBD at 500 ms (control: τ_Fast = 0.0025 ± 0.00069 s, τ_Slow = 0.224 ± 0.046 s; n = 7; CBD: τ_Fast = 0.0048 ± 0.00081 s; τ_Slow = 0.677 ± 0.054 s; n = 5; P = 0.0330 for τ_Fast and P < 0.0001 for τ_Slow). (e) The slow components of recovery from inactivation in control and CBD (1 µM) at 500 ms are shown on the left y axis on a logarithmic scale, and the fraction of slow to fast component of recovery from inactivation is shown on the right y axis. (f) Use-dependent inactivation in control and 2 µM CBD. Normalized current decay plotted as a function of time fitted with an exponential curve (control: τ = 0.14 ± 0.086 s, n = 6; CBD: τ = 0.018 ± 0.0028 s, n = 3; P < 0.05). (g) State-dependent block of peak Nav1.4 current at 10 µM (−110 mV: mean block = 47.3 ± 3.7%, n = 5; %; −70 mV: mean block = 83.8 ± 3.6%, n = 3; *, P = 0.0003). (h) Pulse protocol used for state dependence experiments. Recordings were performed at 1 Hz. Error bars are SE in mean.

Figure S8.

CBD concentration dependence and varying pulse duration F-I curve measurement. (a and b) Show voltage dependence of activation (control: −19.9 ± 4.2 mV, z = 2.8 ± 0.3, n = 5; 1 µM: V_1/2 = −14.3 ± 4.2 mV, z = 2.8 ± 0.3, n = 5; 2 µM: V_1/2 = −21.2 ± 8.1 mV, z = 3.2 ± 0.4, n = 4; P > 0.05 for both V_1/2 and z for both 1 and 2 µM CBD) and 200-ms F-I curve (control: V_1/2 = −64.1 ± 2.4 mV, z = −2.7 ± 0.3, n = 8; 1 µM: V_1/2 = −72.7 ± 3.0 mV, z = −2.8 ± 0.4, n = 5; 2 µM: V_1/2 = −75.5 ± 0.8 mV, z = −2.2 ± 0.5, n = 3; P = 0.0281 for V_1/2 and P > 0.05 for z [1 µM] and P = 0.0073 for V_1/2 and P > 0.05 for z [2 µM]) at 0, 1, and 2 µM CBD. Statistical tests were performed for either CBD concentrations against control/0 µM CBD. (c and d) Voltage dependence of F-I curve from 200 ms (control: V_1/2 = −64.1 ± 2.4 mV, z = −2.7 ± 0.3, n = 8; CBD 1 µM: V_1/2 = −72.7 ± 3.0 mV, z = −2.8 ± 0.4, n = 5; P = 0.0281 for V_1/2 and P > 0.05 for z) and 800 ms (control: V_1/2 = −64.9 ± 1.2 mV, z = −2.1 ± 0.2, n = 4; CBD 1 µM: V_1/2 = −70.9 ± 1.3 mV, z = −1.9 ± 0.2, n = 4; P < 0.05 for V_1/2 and P > 0.05 for z).

Figure S9.

WWWW characterization; CBD stabilizes inactivation in the fenestration-altered construct. (a and b) Conductance voltage (GV) of the construct compared with WT-Nav1.4 (Nav1.4: V_1/2 = −19.9 ± 2.7 mV, z = 2.8 ± 0.3, n = 5; WWWW: V_1/2 = −11.4 ± 0.4 mV, z = 2.0 ± 0.1; P = 0.0113 for V_1/2 and P > 0.05 for z, n = 7) and 200-ms F-I curve (Nav1.4: V_1/2 = −64.1 ± 2.4 mV, z = −2.7 ± 0.3, n = 8; WWWW: V_1/2 = −47.6 ± 0.5 mV, z = 1.7 ± 0.04, n = 5; P < 0.05 for both V_1/2 and z). Both channels are full availability at −110 mV. (c and d) Voltage dependence of 200-ms F-I curve before and after control (c; extracellular solution [ECS]) and CBD (d; 10 µM) in WWWW construct (before control: V_1/2 = −54.7 ± 5.1 mV, n = 6; after control: V_1/2 = −54.2 ± 5.4 mV, n = 6; before CBD: V_1/2 = −48.8 ± 8.8 mV, n = 3; after CBD: V_1/2 = −72.7 ± 5.7 mV, n = 3, P = 0.0068 for CBD in matched-pair analysis). The ECS experiment was performed to ensure that hyperpolarization shifts in the CBD condition are not due to possible confounding effects associated with fluoride in the internal (CsF) solutions. (e and f) Representative families of inactivating currents before and after perfusion. CBD does not block peak currents but shifts the F-I curve to the left. (g) Averaged shift in the midpoint of F-I curve before and after perfusion. *, P < 0.05. Error bars are SE in mean.

Figure 6.

Effects of CBD (1 µM) on gating of a myotonia/hypoPP variant, P1158S. (a and b) Voltage dependence of activation as normalized conductance plotted against membrane potential at pH 7.4 (control: V_1/2 = −30.0 ± 3.3 mV, z = 3.1 ± 0.2, n = 8; CBD: V_1/2 = −32.7 ± 3.6 mV, z = 2.9 ± 0.2, n = 7; P > 0.05 for both V_1/2 and z) and pH 6.4 (control: V_1/2 = −23.0 ± 3.3 mV, z = 2.9 ± 0.2, n = 8; CBD: V_1/2 = −21.1 ± 3.3 mV, z = 2.5 ± 0.2, n = 8; P > 0.05 for both V_1/2 and z). (c and d) Voltage dependence of 200-ms F-I curve plotted against membrane potential (channels were held at −130 mV for 200 ms) at pH 7.4 (control: V_1/2 = −73.2 ± 2.6 mV, z = 2.9 ± 0.2, n = 7; CBD: V_1/2 = −83.0 ± 2.6 mV, z = 3.0 ± 0.3; P = 0.0260 for V_1/2 and P > 0.05 for z) and pH 6.4 (control: V_1/2 = −68.4 ± 3.0 mV, z = 2.7 ± 0.4, n = 5; CBD: V_1/2 = −81.7 ± 2.3 mV, z = 2.7 ± 0.3, n = 9; P = 0.0010 for V_1/2 and P > 0.05 for z). (e and f) Recovery from fast inactivation at 500 ms at pH 7.4 (control: τ_Fast = 0.0018 ± 0.006 s, τ_Slow = 0.15 ± 0.6 s, n = 7; CBD: τ_Fast = 0.24 ± 0.07s; τ_Slow = 2.5 ± 0.6 s, n = 6; P = 0.0347 for τ_Fast and P = 0.0245 for τ_Slow) and pH 6.4 (control: τ_Fast = 0.065 ± 0.04 s, τ_Slow = 0.75 ± 0.4 s, n = 7; CBD: τ_Fast = 0.13 ± 0.07 s; τ_Slow = 0.62 ± 0.1 s, n = 4; P < 0.05 for τ_Fast and P > 0.05 for τ_Slow). (g and h) Persistent currents measured from a 200-ms depolarizing pulse to 0 mV from a holding potential of −130 mV at pH 7.4 (control: percentage = 4.4 ± 1.2%, n = 4; CBD: percentage = 1.0 ± 0.2%, n = 4; *, P = 0.0339; n = 4) and pH 6.4 (control: percentage = 4.4 ± 2.1%, n = 6; CBD: percentage = 5.4 ± 1.2%, n = 5; P > 0.05). Error bars are SE in mean.

Figure 7.

AP simulations of skeletal muscle APs in the presence and absence of CBD, based on voltage-clamp data. Top of the figure shows the pulse protocol used for simulations, and a cartoon representation of P1158S-pH in vitro/in silico assay, where pH can be used to control the P1158S phenotype. (a and b) Simulations in WT-Nav1.4 in the presence and absence of CBD. (c and d) Simulations of P1158S at pH6.4. (e and f) Results from pH 7.4.

Figure 8.

Effects of CBD on rat diaphragm contraction. (a) Image of dissected rat diaphragm muscle. (b) Image of rat diaphragm cut into a hemidiaphragm, which was placed between electric plates that were used for electric stimulation. The subsequent muscle contractions were measured using a force transducer. (c) Normalized quantification of muscle contractions in CBD and TTX (percentage of normalized contraction: control = 100 ± 5.3%, n = 9; CBD = 60.6 ± 3.5%, n = 8; TTX = 28.9 ± 5.3%, n = 6; * indicates P = 0.0006 for CBD; # indicates P < 0.0001 for TTX). Error bars are SE in mean. (d–f) Sample contraction traces across all three conditions.

Figure S10.

Comparison between some of the relevant physicochemical properties of the compounds used in this study. (a) Chemical structures of the compounds used in this study. (b) 3-D structures of the compounds. (c) Volume (ų) and area (Ų) for each compound were calculated using University of California, San Francisco Chimera. LogP values are obtained from the ChEMBL database. CBD, lidocaine, and flecainide all interact with the LA site inside the Nav pore. TTX interacts with the outer selectivity filter of the Nav pore. CBD is several times more hydrophobic than the other compounds. CBD is larger than lidocaine and slightly smaller than flecainide.

Figure 9.

Pathway of skeletal muscle inhibition via Nav1.4. This is a cartoon representation of the mechanism and pathway through which CBD inhibits Nav1.4. Once CBD is exposed to the skeletal muscle, given its high lipophilicity, the majority of it gets inside the sarcolemma. Upon entering the sarcolemma, it localizes in the middle regions of the leaflet and travels through the Nav1.4 fenestrations into the pore. Inside the pore mutation of the LA, F1586A reduces CBD inhibition. CBD also alters the membrane elasticity, which promotes the inactivated state of the Nav channel, which adds to the overall CBD inhibitory effects. The net result is a reduced electrical excitability of the skeletal muscle, which, at least in part, contributes to a reduction in muscle contraction.