Cannabidiol potentiates hyperpolarization-activated cyclic nucleotide–gated (HCN4) channels

Cannabidiol (CBD) is the main non-psychotropic cannabinoid produced by the Cannabis sativa plant (Lerner, 1963). CBD has many purported effects including being antiarrhythmic, anti-inflammatory, and anticonvulsant (Hill et al., 2012; Walsh et al., 2010). As such, CBD has generated a large amount of pharmacological interest and is approved by the US Food and Drug Administration for the treatment of patients with Dravet and Lennox-Gastaux syndromes (Devinsky et al., 2017, 2018; Scheffer et al., 2021). At micromolar concentrations, CBD inhibits a variety of cardiac ion channels, including Na_V1.5 channels (Ghovanloo et al., 2018, 2021; Fouda et al., 2020, 2022; Fouda and Ruben, 2021; Sait et al., 2020), L-type calcium channels (Ali et al., 2015), and hERG channels (Orvos et al., 2020). The resultant changes in action potential durations have been observed in various cardiomyocyte systems (Ali et al., 2015; Orvos et al., 2020; Le Marois et al., 2020; Topal et al., 2021).

Hyperpolarization-activated cyclic-nucleotide gated–channels (HCN) are important to heartbeat initiation and the diastolic depolarization phase of the action potential in the sinoatrial node of the heart (Baruscotti et al., 2011; Wahl-Schott and Biel, 2009; Stieber et al., 2003; Fenske et al., 2020; Moosmang et al., 2001). The primary cardiac isoform of HCN channels is known as HCN4 (Brioschi et al., 2009; Herrmann et al., 2011). HCN4 channels are tetramers. Each monomer contains six transmembrane helices (S1–S6), where S1–S4 make up a distinct voltage-sensing domain (VSD), and S5–S6 of each monomer combine to form the pore domain (Lee and MacKinnon, 2017, 2019; Saponaro et al., 2021). Additionally, HCN4 channels have two cytoplasmic domains, the C-terminal cyclic-nucleotide binding domain (CNBD) (Zagotta et al., 2003) and the N-terminal HCN domain (Lee and MacKinnon, 2017; Porro et al., 2019). These domains contribute to the potentiation of HCN4 channels by cAMP (Wainger et al., 2001; Zagotta et al., 2003; Xu et al., 2010; Akimoto et al., 2014; Porro et al., 2019; Page et al., 2020), increasing heart rate (HR) in response to the β-adrenergic response (Brown et al., 1979; DiFrancesco and Tortora, 1991; Ono et al., 1993; Schweizer et al., 2010).

CBD has been anecdotally reported to alter HR and, in other cases, have no effect on HR. In an effort to determine the molecular basis for the potential effects of CBD on HR, we measured the biophysical properties of HCN4, a voltage-gated ion channel known to play a major role in regulating HR, in the presence of CBD. We used HEK 293T cells as a heterologous expression system for HCN4 channels to test the effects of CBD application. First, we examined increasing micromolar concentrations of CBD on holo-HCN4 channels. Next, we examined whether the effect of CBD is competitive or additive with known HCN potentiators cAMP and PI(4,5)P_2. Lastly, we used computational docking to predict potential binding sites of CBD on an HCN4 structure and targeted mutagenesis to validate these predictions. We propose that CBD potentiates HCN4 channels via novel biophysical mechanisms accomplished by binding to the C-terminal region of the VSD.

HEK 293T cells were grown at pH 7.4 in sterile-filtered DMEM + Glutamax-I (Gibco) supplemented with 10% FBS at 37°C (5% CO₂). Cells were cotransfected with mouse HCN4 (mHCN4) and eGFP using DNAfectin 2100 (ABM). Cells were plated on sterile glass coverslips 24–48 h after transfection, and experiments were conducted at least 3 h later.

Whole-cell patch clamp experiments were performed in extracellular solution containing (in mM) 110 NaCl, 30 KCl, 0.5 MgCl₂, 1.8 CaCl₂, and 5 HEPES titrated to pH 7.4 with NaOH. Pipettes were fabricated with a P-1000 puller (Sutter) using borosilicate glass (1.5 mm O.D.) thermally polished to a resistance of 1.5–2.5 MΩ. Pipettes were filled with intracellular solution containing (in mM) 10 NaCl, 130 KCl, 0.5 MgCl₂, 1 EGTA, and 5 HEPES, titrated to pH 7.4 with KOH. CBD was dissolved in 100% DMSO to create stock solutions of 25 mM. Cyclic AMP (cAMP) (Cayman Chemical) was dissolved in PBS (pH 7.2) to create stock solutions of 10 mM. cAMP stock solution was added to the internal solution to a final concentration of 30 µM for all holo experiments. 08:0 PI(4,5)P₂, also known as diC8 PI(4,5)P₂, was dissolved in double-distilled H₂O to create stock solutions of 0.5 mM.

Recordings were conducted at 22°C with an EPC9 amplifier using Patchmaster software (HEKA Electronic). Data were lowpass filtered at 1 kHz and digitized at 20 kHz. Leak subtraction was performed online using a P/2 procedure. Gigaohm seals were allowed to stabilize in the on-cell configuration for 1 min prior to entering the whole-cell configuration. Series resistance was <5 MΩ for all recordings. All data were acquired 5 min after entering the whole-cell configuration to allow for completion of run-down and cells were allowed to incubate for at least 5 min after drug or vehicle treatment before data collection. Data were collected as matched pairs, with control or “pre-condition” data collection first, followed by perfusion and data collection for a single CBD concentration or vehicle. Due to a small volume of diC8 PI(4,5)P₂ stock solution, pre-conditions and perfusion were not performed, opting instead for a standard bath solution of extracellular solution as described above with both 10 µM diC8 PI(4,5)P₂ and 5 µM CBD.

Use dependence was determined using a 1/3 Hz protocol (20 sweeps) that stepped from the holding potential of −40 mV to −120 mV for 2 s, followed by tail-current at −100 mV. Membrane potential was stepped to 0 mV after tail current for 0.5 s. To examine the activation of mHCN4 Q401R, we stepped from the holding potential of −40 mV to −120 mV for 4 and 8 s followed by a step to 0 mV to close the channel for 0.5 s.

All statistical analyses were computed using Igor Pro software (WaveMetrics). Statistical analysis of matched pair conditions was done with a paired Student’s t test. Statistical analysis across different conditions consisted of either one-way ANOVA followed by post-hoc Tukey’s test or unpaired Student’s t tests. For all analyses, P < 0.05 was considered statistically significant. Values are written as mean ± SEM, where n ≥ 4.

Computational docking was performed using AutoDock Vina (Trott and Olson, 2010; Eberhardt et al., 2021) with the published cryoEM structure of rabbit HCN4 (PDB ID 7NP4) and CBD (DrugBank access DB09061). Two different search space volumes were used, both >27,000 ų_, large enough to include the entirety of the transmembrane region of either a single monomer or the entire tetramer. The entire tetramer was used to examine potential binding sites at interfaces between the monomers. To account for the large search space volume, the exhaustiveness parameter was increased from the default of 8 to 100 for the monomer and 1,000 for the tetramer. The output was the nine highest affinity binding sites calculated, with multiple general locations having multiple predicted binding conformations. In the tetramer, sites were also repeated across the different monomers of the channel (Fig. S3).

The Q401R mutation was introduced into the pCDNA3.1 mHCN4 plasmid (provided by the laboratory of Dr. C. Proenza, University of Colorado School of Medicine, Aurora, CO, USA) using the QuikChange Lightning Mutagenesis Kit (Agilent). The presence of the mutation was verified with Sanger sequencing (Genewiz), and successfully mutated plasmids were purified using CompactPrep Midi Prep Kit (Qiagen).

Fig. S1 shows verification of the effect of cAMP on mHCN4 channels. Fig. S2 details a comparison of fitting single versus double exponential equations to holo-HCN4 channels. Fig. S3 shows alternative outputs from AutoDock Vina computation docking experiments. Fig. S4 shows additional examples of the HCN4 Q401R mutant activation with before and after CBD application superimposed. Fig. S5 shows computational docking results between HCN4 and cholesterol.

We measured the activation voltage dependence of holo mHCN4 channels (i.e., in the presence of 30 µM cAMP) (Fig. 1 A) before (black) and after perfusion with 5 µM CBD (green). The activation voltage dependence was calculated using the Boltzmann function as described in the Materials and methods section. The average V_1/2 for mHCN4 was −102.1 ± 1.1 mV (n = 9) and was shifted in the depolarized direction after perfusion with 5 µM CBD to −87.5 ± 3.0 mV (n = 9) (Fig. 1 B). CBD did not change the slope of the activation curve (Table 1). In other words, CBD potentiated mHCN4 channels, with channels activating at weaker driving forces.

The ΔV_1/2 was calculated with different concentrations of CBD (1–20 µM) or a vehicle (Fig. 1 C and Table 1). The ΔV_1/2 for 1 µM CBD treatment (−3.65 ± 2.22 mV, n = 6) was not significantly different from that of the vehicle (1.80 ± 3.71 mV, n = 7). However, the ΔV_1/2 after perfusion with 5 µM (−14.5 ± 2.2 mV, n = 9) was significantly different than the vehicle (P < 0.0001) and 1 µM CBD (P < 0.03). Treatment with 5 µM CBD was sufficient to exert maximal shift in ΔV_1/2; we observed no significant increase in ΔV_1/2 with larger concentrations of CBD (10 and 20 µM) (Table 1). Using the Hill equation (Fig. 1 D), the EC₅₀ of CBD was calculated to be 1.59 ± 0.21 µM, suggesting this potentiation could occur at therapeutically relevant concentrations. Additionally, the Hill slope was 2.50 ± 0.63, suggesting positive binding cooperativity between multiple binding sites.

Previous studies found that CBD increased the current density of mHCN1 channels in intact Xenopus laevis oocytes (Mayar et al., 2022). To test if this was true in mHCN4 channels, we examined the relative peak current after perfusion of CBD. We compared peak current density at −120 mV, measured at the end of the activation epoch, before and after perfusion with CBD or vehicle (Fig. 1 E). We found no significant difference in the peak current density for all concentrations of CBD tested (1–20 µM) compared with the vehicle, so we concluded that CBD did not increase HCN4 current density.

We next examined how CBD affected the activation kinetics of mHCN4 channels. We fit the concave-down portions of the activation sweeps with a single exponential equation, allowing us to calculate τ_act (Fig. 2 A). All concentrations of CBD tested significantly sped (P < 0.04) the activation of mHCN4 channels (Fig. 2, A and B) between −100 and −140 mV. For example, at −120 mV (Fig. 2 A), τ_act was 1,270 ± 160 ms (n = 6) after treatment with the vehicle, while τ_act was 404 ± 59 (n = 8) after treatment with 5 µM CBD, a threefold decrease in τ_act (Table 2). This speeding of activation is not fully accounted for by the shift in activation V_1/2 (∼15 mV). To account for ΔV_1/2, we compared τ_act at similar driving forces, i.e., −140 mV after treatment with vehicle and −120 mV after treatment with 5 µM CBD. At similar driving forces, 5 µM CBD still significantly (unpaired Student’s t test P < 0.02) decreased τ_act by 1.5-fold. This suggests that CBD stabilizes the transition to the activated state in addition to its effect on voltage dependence.

We studied the state dependence of CBD’s action on HCN4 channels by using a 1/3 Hz repetitive activation protocol to measure τ_act over time. The protocol was 20 sweeps, for a total of 60 s using −120 mV as the activation voltage. If CBD was bound to the open state, we would expect a decrease in τ_act over time as more CBD molecules bind to the open state. However, if CBD bound the closed state, we would expect a decrease in τ_act immediately. We found that in sweep 1, 5 µM CBD significantly decreased τ_act (P < 0.02): 940 ± 190 ms (n = 7) before treatment compared with 510 ± 51 (n = 7) after treatment (Fig. 2, C and D). This is consistent with our previous results of the effect of CBD on τ_act (Fig. 2 B and Table 2). The speeding of the activation did not significantly change with increasing activation sweeps (Fig. 2, C and D). After treatment with 5 µM CBD, τ_act for sweep 20 (435 ± 64 ms, n = 6) was not significantly different from τ_act on sweep 1 (510 ± 51 ms, n = 7). Combined these results suggest that CBD binds to the closed state of mHCN4.

How does CBD potentiate HCN4 channels? We examined whether CBD was acting cooperatively with cAMP to potentiate mHCN4 channels. Similar to CBD, cAMP speeds activation and shifts the V_1/2 of mHCN4 channels by approximately +10 mV (Fig. S1). We measured ΔV_1/2 after treatment with 5 µM CBD in apo channels (i.e., no cAMP was present in the internal solution) (Fig. 3 A). In the absence of cAMP, the V_1/2 was −113 ± 2.2 mV (n = 7), and after CBD treatment, V_1/2 was −97.6 ± 2.5 mV (n = 7), a ΔV_1/2 of −15.5 ± 2.0 (n = 7) (Fig. 3 B). A comparison to results from Fig. 1 B show that in the presence of CBD, cAMP still shifts the V_1/2 by approximately +10 mV. Additionally, CBD treatment still decreased τ_act by at least 1.5-fold at −130 and −150 mV (Table 2) in apo-HCN4 channels. Combined, this suggests that the biophysical mechanisms by which CBD and cAMP potentiate HCN4 channels are separate and independent of one another.

We next examined whether CBD was acting through a similar mechanism as the lipid PI(4,5)P₂. PI(4,5)P₂ is known to shift V_1/2 of HCN1, HCN2, and HCN4 channels by 15–20 mV (Zolles et al., 2006; Pian et al., 2006, 2007). Our previous experiments were performed after the rundown was complete, so in the absence of PI(4,5)P₂. Therefore, we examined the effect of CBD in the presence of an exogenous PI(4,5)P₂ analog diC8 PI(4,5)P₂ (Fig. 3 C). diC8 PI(4,5)P₂ is faster-acting and more soluble than PI(4,5)P₂ and shifts V_1/2 of HCN2 by 10 mV (Pian et al., 2006). The average V_1/2 after the addition of 10 µM diC8 PI(4,5)P₂ and 5 µM CBD was −76.5 ± 3.4 (n = 5) (Fig. 3 D), a further depolarizing shift of ∼10 mV compared with V_1/2 after treatment with 5 µM CBD (Fig. 1 B and Table 1). This suggests that although CBD and PI(4,5)P₂ are both lipophilic small molecules, they use unique biophysical mechanisms to potentiate HCN4 channels.

To give us further insight into the mechanism CBD utilizes to potentiate HCN4, we performed computational docking of CBD with the cryoEM structure of HCN4. Using AutoDock Vina, we identified nine of the highest affinity binding sites for both a single HCN4 monomer and the full tetramer (highest affinity binding for tetramer shown in Fig. 4, A–D). Cytosolic domains were excluded due to the lipophilic nature of CBD (ClogP = 6.33 [Chemaxon]). Of particular interest was a binding pocket in the center of the VSDs (Fig. 4, A and B), where CBD binding events clustered in the experiments for the monomer and tetramer. This pocket was previously identified as a lipid-binding pocket of unknown effect (Saponaro et al., 2021). CBD binding here would position it in a prime location to modulate voltage sensor movement (Fig. 4, C and D), agreeing with our functional data. To test if CBD binds here, we introduced an arginine at the Q401 position (Fig. 4, D and E). The voltage dependence of the HCN4 Q401R mutant was drastically shifted to hyperpolarized potentials and activation was surprisingly slow, resulting in a linear curve even when activation time was increased up to 8 s (Fig. S4). As such, we were unable to measure activation V_1/2 or calculate τ_Act. Application of 5 µM CBD had no visual effect on the activation of HCN4 Q401R (Fig. 4 E and Fig. S4), suggesting that the activity of CBD was disrupted. This supports the hypothesis that CBD affects HCN4 channels by binding to the binding pocket identified via computational docking.

CBD is a highly promiscuous small molecule, interacting with a variety of transmembrane proteins and ion channels. In general, CBD blocks many of the ion channels it interacts with, including all subtypes of Na_V channels, K_V1.2, KCNQ1/KCNE1 channels, L-type calcium channels, and the HCN relative, the hERG channel (Ali et al., 2015; Ghovanloo et al., 2018; Orvos et al., 2020; Isaev et al., 2022; Topal et al., 2021). Recently, CBD has been shown to potentate some neuronal channels: increasing the current density of HCN1 channels (Mayar et al., 2022) and activating K_V7 channels by shifting the voltage dependence at submicromolar concentrations (Zhang et al., 2022). Our results show that CBD is the first FDA-approved drug to potentiate cardiac HCN channels by modulating HCN4 voltage dependence and activation kinetics.

We propose that CBD binds to HCN4 channels (Fig. 5) resulting in both depolarization of the V_1/2 of activation and speeding of activation kinetics. Thus, CBD allows HCN4 channels to open at weak driving forces. We calculated the V_1/2 from a fixed activation epoch time for all voltages, which is not a true steady state. Steady-state V_1/2 values may be more depolarized than the isochronal V_1/2 values found here. However, the effect of CBD on V_1/2 is expected to be preserved at true steady state as well as its effect on activation kinetics. We found that the shift in V_1/2 after CBD application did not result in a significant increase in the current density of HCN4 channels. In contrast, Mayer et al. (2022) found that in intact oocytes, CBD increased the current density of HCN1 channels without shifting the voltage dependence of activation. This difference could be due to the use of intact oocytes versus the HEK 293T cells used in this study. Alternatively, the distinct responses to CBD could be due to isoform-specific differences in HCN1 and HCN4 channels. HCN1 and HCN4 differ in both activation kinetics, thought to be due to sequence differences in the S2 helix, as well as the response to cAMP (Moosmang et al., 2001; Chen et al., 2001; Wang et al., 2001; Lolicato et al., 2011). Combined, the result is HCN1 channels activating at more depolarized potentials than HCN4 channels. Thus, the distinct responses to CBD could be due to these inherent differences in HCN isoforms.

The EC₅₀ for ΔV_1/2 was 1.59 µM and, although the EC₅₀ for the decrease in τ_act was unable to be calculated with the concentrations of CBD used, we expect it to be <1 µM. Thus, we expect that CBD can potentiate HCN4 channels at therapeutically relevant concentrations. The calculated Hill slope for ΔV_1/2 was >1, indicating multiple binding sites. Since HCN4 channels are tetramers, logic suggests that there would be up to four binding sites present in homomeric channels. Based on the Hill slope, we suggest that at least two CBD molecules bind to potentiate HCN4 channels. Future work could examine the exact number of CBD molecules required and whether increasing numbers of CBD molecules bound increases the potentiation effect.

Activation kinetics were calculated using a single exponential fit. Residuals of single and double exponential fits were compared at the three voltages (Fig. S2); single exponentials were sufficient to accurately model the activation kinetics. It must be noted that at more depolarized potentials, the accuracy of the τ_Act will be decreased. As such, τ_Act values described in this work from −100 mV may be faster than the true speed. While our results from the repetitive pulse protocol suggest that CBD binds to the closed state, we have not excluded the idea that CBD doesn’t have state dependence, binding to closed- or open-state with no change in affinity, or that the CBD on-rate is too fast to be accurately measured by this analysis. Future work could be done to determine whether the effects of CBD are caused by binding to and destabilizing the HCN4 closed state.

CBD’s potentiation was additive to the activities of both cAMP and PI(4,5)P₂, suggesting a unique biophysical mechanism from these two small molecules. cAMP relieves the autoinhibition of HCN4 channels. We tested if CBD acts using a common mechanism as cAMP but found that CBD potentiation was distinct from cAMP. However, since CBD was not tested on a channel incapable of autoinhibition (i.e., without the CNBD), we cannot definitively conclude that CBD does not also relieve autoinhibition in a non-competitive way to cAMP. Claveras Cabezudo et al. (2022) performed computational modeling to predict that phosphoinositides bind to the surface of the HCN domain along with S2 and S3 helices. Our mutational and docking studies do not predict that CBD binds to similar locations (Fig. S3) (Claveras Cabezudo et al., 2022). Thus, our findings that CBD and PI(4,5)P₂ have unique biological functions agree with the predicted binding sites of these molecules.

The highest affinity binding site computed for CBD is located in a lipid binding pocket at the C-terminus of the VSD. Binding here places CBD close to the charged residues of the voltage sensor (Fig. 5), a prime position from which to regulate HCN4 activation voltage dependence. This pocket was bound to a lipid in the cryoEM structure (Saponaro et al., 2021), and with computational docking we found that cholesterol was also predicted to bind in this pocket (Fig. S5). Combined, this suggests that this pocket is a general lipid binding site, where lipids or highly lipophilic molecules like CBD bind primarily using van der Waals and hydrophobic interactions with the S2–S3 linker.

The results of the computational binding analysis also suggested alternative binding sites, some repeated in multiple monomers of the channels (Fig. S3). The most common alternative site was on the surface of the channel, against the C-terminus of the S1 helix. While the N-terminus of the HCN4 S1 helix is implicated in channel gating via interactions with the S4 helix (Ishii et al., 2007; Ramentol et al., 2020), the C-terminal S1 residues have not been implicated in gating. As such, CBD binding here is less likely to produce the observed potentiation than the proposed lipid-binding pocket (Fig. 4).

Mutagenesis to examine the binding site of CBD produced a channel, HCN4 Q401R, with drastically shifted activation voltage dependence and slowed activation kinetics. Increasing the time of the activation epoch (2 s) by two- and fourfold still resulted in a linear trace that was unable to be fit by a single exponential equation (Fig. S4). As such, we chose to qualitatively observe if CBD sped activation kinetics via superimposition of activation traces. Application of 5 µM CBD did not have a qualitative effect on activation, with no substantial change in the linear curve being observed (Fig. 4 E and Fig. S4) across eight different cells. We concluded from this that the Q401R mutation disrupted the binding of CBD, possibly by increasing the polarity of the binding site. This result provides secondary evidence that CBD binds near Q401, in the proposed binding pocket at the C-terminus of the VSD.

The classical gating charge transfer centers were originally documented in the crystal structure of a K_V1.2–K_V2.1 chimera and are conserved in HCN4 channels (Tao et al., 2010; Saponaro et al., 2021). Gating charge transfer centers are composed of three residues: a phenylalanine and two negatively charged residues. The phenylalanine acts in two ways: (1) to block or “cap” the gating pathway, and (2) to stabilize the gating charges via cation–π interactions. These cation–π interactions preferentially bind to lysine over arginine residues, promoting inward movement of the voltage sensor. CBD contains a benzene ring located near gating charges when bound to the proposed binding pocket. Similar to the phenylalanine of the gating charge transfer center, cation–π interactions between gating charges and CBD could promote the inward movement of the voltage sensor. Specifically, in the closed state, the glutamine at the C-terminal end of S4 is closest to CBD. We propose that the benzene ring would preferentially bind to the arginine one helical turn away, resulting in the observed potentiation and promoting the inward motion of the S4 helix.

HCN channels have long been targets for drug development. Ivabradine, a common treatment for chronic heart failure, is an HCN-specific channel blocker that binds to the open state of the channel (Bucchi et al., 2013; Postea and Biel, 2011). Additional blockers have also been developed, although not approved for use in patients, including other imidazolines like alinidine and zatebradine, and the piridinium derivative ZD7288 (Postea and Biel, 2011; Novella Romanelli et al., 2016). Unfortunately, none of these drugs were found to be specific to HCN channels. Unlike these drugs, CBD does not block HCN channels, and as such represents a novel starting point for future drug development.

CBD is not the first natural product that has been found to potentiate HCN channels. The flavonoid fisetin and the phytoestrogen isoflavinoid genistein, both plant metabolites, have been shown to potentiate the related HCN2 channels (Carlson et al., 2013; Rozario et al., 2009). These compounds competed with cAMP via binding to the cAMP binding pocket, while the action of CBD was unaffected by the presence or absence of cAMP. Tanshinone IIA is a lipophilic constituent of the Chinese medicinal herb Salvia miltiorriza, also known as Danshen, that is used in traditional medicine to treat cardiovascular disease. Similar to CBD, tanshinone IIA possesses a myriad of activities, including being cardioprotective, neuroprotective, and anti-inflammatory. Tanshinone IIA has a mixed effect on HCN1 and 2, increasing channel open probability while slowing activation and deactivation (Liang et al., 2009). Since CBD potentiates HCN4 channels by shifting voltage dependence and speeding activation, it must contain novel features or functional groups compared with these three plant metabolites.

HCN channels have been found to be regulated by a variety of human-derived lipids, including phosphoinositides (Pian et al, 2006, 2007; Zolles et al, 2006; Flynn and Zagotta, 2011). While we found that CBD acts in a distinct mechanism from PI(4,5)P₂, two other acidic phospholipids present in human cells, phosphatidic acid (PA) and arachidonic acid (AA), directly modulate HCN1 and HCN2 channels (Fogle et al., 2007). PA and AA both depolarize the V_1/2, like CBD; however, AA also decreases the open probability of the channels, which CBD does not. Future work could be done to examine if CBD utilizes a common biophysical mechanism with PA and AA. Another lipid that regulates HCN channels is cholesterol (Fürst and D’Avanzo, 2015). Cholesterol regulates the main HCN channel isoforms (HCN1, HCN2, and HCN4) in distinct ways. With HCN4 channels, cholesterol facilitates mode-shift, a mechanism where channels display hysteresis in their voltage dependence, having a more stable open state if they have been exposed to activating voltage previously. While cholesterol may bind in a similar location as CBD (Fig. S5), clearly the structural features of these two molecules elicit very different effects on HCN4 channels.

In humans, use of cannabis results in a short-term increase in heart rate by up to 50 beats per minute (Beaconsfield et al., 1972; Jones, 2002). HCN4 channels play a prominent role in the initiation of the diastolic depolarization phase of action potentials in the SA node (Brown et al., 1979; DiFrancesco and Tortora, 1991). Thus, activation of HCN4 channels by CBD would accelerate the heart rate, partially explaining this phenomenon. Previous studies on CBD effects on action potential duration (APD) have noted both shortening and lengthening of APDs depending on the organism used (Orvos et al., 2020; Le Marois et al., 2020; Isaev et al., 2022). However, no increase in QT intervals was reported in clinical trials of CBD in humans (Stanley et al., 2013; Iffland and Grotenhermen, 2017). CBD’s potentiation of HCN4 channels could play a role in this as HCN isoforms including HCN4 are expressed in ventricular cardiomyocytes (Shi et al., 1999; Zhang et al., 2009). Potentiation of HCN4 would result in outward current during the plateau phase of the ventricular action potential counteracting the inhibition of other cardiac potassium channels like I_Kr (hERG) and I_Ks (Ueda et al., 2009; Orvos et al., 2020; Topal et al., 2021). Various heterozygous HCN4 channelopathies in humans have been identified with the phenotype for most patients including bradycardia (Postea and Biel, 2011). Additionally, common medications like β-blockers (atenolol), verapamil, and digitalis can induce sinus node dysfunction (Margolis et al., 1975; Strauss et al., 1976; Breithardt et al., 1978; Manne, 2018). Our results suggest that CBD could be a novel platform for drug development to protect against bradycardia and sinus dysfunction resulting from familial mutations or drug-induced.

The data used to generate tables and figures are openly available in the publicly accessible database: Summit Research Repository at https://summit.sfu.ca/item/36669.

Jeanne M. Nerbonne served as editor.

We thank Dr. C. Proenza for the generous donation of the mHCN4 construct.

This work was supported by the Natural Sciences and Engineering Research Council of Canada to P.C. Ruben (RPGIN03920) and by a MITACS Elevate grant in partnership with Akseera Pharma, Inc. to D.A. Page (IT21730).

We acknowledge that our MITACS partner is a pharmaceutical company interested in cannabis, but this fact did not affect our findings and the authors have no financial interest in the partner company. Akseera Pharma, Inc. was not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.

Author contributions: D.A. Page and P.C. Ruben conceptualized the experiments. D.A. Page collected, analyzed, interpreted, and visualized the data, and wrote the first draft of the manuscript. P.C. Ruben provided project administration and resources, as well as critical review and revision of the manuscript.